Tools for Identifying Biodiversity:

Progress and Problems

Pier Luigi Nimis and Régine Vignes Lebbe (eds.)
 9 788883 032950

Tools for Identifying Biodiversity:

Progress and Problems
Pier Luigi Nimis and Régine Vignes Lebbe (eds.)

Proceedings of the International Congress

Paris, September 20-22, 2010
Muséum national d’Histoire naturelle – Grand Amphithéâtre
 All content is copyrighted by the individual authors of the contributions and licensed
under the Creative Commons Attribution-Share-Alike License (CC by-sa 3.0).
EUT 2010.
Information in this book does not represent the opinion of the European Community and
the European Community is not responsible for any use that might be made of it.

ISBN 978-88-8303-295-0

EUT - Edizioni Università di Trieste

Via E. Weiss, 21 – 34128 Trieste
http://eut.units.it

Cover and layout: Rodolfo Riccamboni, Divulgando S.r.l. – www.divulgando.eu

Printed in Italy by Studio Pixart S.r.l., Via I Maggio, 8, I – 30020 Quarto d’Altino, VENEZIA.
 Scientific Committee
Nicolas Bailly (Philippines)
Daniel Barthélémy (France)
Frank Bisby (United Kingdom)
Walter Berendsohn (Germany)
Gudmundur Gudmundsson (Iceland)
Gregor Hagedorn (Germany)
Bill Hominick (United Kingdom)
Christian Kittl (Austria)
Maurizio Casiraghi (Italy)
Noel Conruyt (La Réunion, France)
Martin Grube (Austria)
Massimo Labra (Italy)
Karol Marhold (Slovak Republic)
Anastasia Miliou (Greece)
Andrea M. Mulrenin (Austria)
Pier Luigi Nimis (Italy)
Bob Press (United Kingdom)
Dave Roberts (United Kingdom)
Peter Schalk (Netherlands)
Edwin van Spronsen (Netherlands)
Régine Vignes Lebbe (France)

Organising Committee
Coordinators:
Pier Luigi Nimis
Régine Vignes Lebbe

Members:
Léa Bled
Florian Causse
Vanessa Demanoff
Zoulika Labghiel
Visotheary Rivière-Ung
Stefano Martellos
Rodolfo Riccamboni
Maxime Venin

i
 Foreword

The correct identification of organisms is fundamental not only for the

assessment and the conservation of biodiversity, but also in agriculture, forestry,
the food and pharmaceutical industries, forensic biology, and in the broad field
of formal and informal education at all levels. However, since the first Meeting
of the Systematics Association on «Biological identification with computers», in
1973, few scientific events have been dedicated to this subject. Furthermore,
taxonomists, workers in biodiversity informatics, and the large community of
users are rarely all gathered together.
Since the 1990s, the number of projects developing information repositories
has greatly increased: Fishbase, GBIF, Species 2000, OBIS, EuroMed-
PlantBase, Fauna Europaea, EoL etc. to cite only some of them. Until now,
identification tools were poorly represented in such systems. This is already
changing, and Fishbase is a good example illustrating the need to include
identification facilities with biodiversity databases, and to adapt the keys to
different types of users. International conferences on biodiversity research,
tools and methods using ICT, are becoming more and more numerous. In the
last decades, important advances have taken place in the ways identification is
carried out, from molecular and biochemical methods of rapid identification to the
development of interactive identification systems based on morpho-anatomical
data. The effort to propose and to popularize identification tools using all types
of biological characters (sequences, morphology, images, sounds etc.) must be
continued.
The event «Tools for identifying biodiversity: progress and problems»
offers an opportunity to provide an overview of recent advances in this field.
It aims at stimulating integration of existing methods and systems, fostering
communication amongst different research groups, and laying the foundations
for integrated projects in the next decade. The congress was organised jointly
by three large European projects dedicated to biodiversity and/or biological
identification: KeyToNature, EDIT (European Distributed Institute of Taxonomy),
and STERNA (Semantic Web-based Thematic European Reference Network
Application).

iii
 The scientific program of the congress was subdivided into four sessions:

• Interactive identification tools based on morpho-anatomical data

• Molecular and biochemical methods for the identification of organisms
• Identification and education
• Industrial and practical applications of the new identification tools: case-
studies and markets

In this book, the reader will find short presentations of current and upcoming
projects (EDIT, KeyToNature, STERNA, Species 2000, Fishbase, BHL, ViBRANT,
etc.), plus a large panel of short articles on software, taxonomic applications,
use of e-keys in the educational field, and practical applications. Single-access
keys are now available on most recent electronic devices; the collaborative
and semantic web opens new ways to develop and to share applications; the
automatic processing of molecular data and images is now based on validated
systems; identification tools appear as an efficient support for environmental
education and training; the monitoring of invasive and protected species and
the study of climate change require intensive identifications of specimens, which
opens new markets for identification research.

Pier Luigi Nimis, Régine Vignes Lebbe

Trieste – Paris, September 2010

v
 Table of Contents
Devising the EDIT Platform for Cybertaxonomy................................................ 1
Walter G. Berendsohn

Descriptive Data in the EDIT Platform for Cybertaxonomy............................... 7

Maxime Venin, Agnes Kirchhoff, Hélène Fradin, Anton Güntsch, Niels Hoffmann,
Andreas Kohlbecker, Elise Kuntzelmann,Ôna Maiocco, Andreas Müller,
Régine Vignes Lebbe, Walter G. Berendsohn

An online authoring and publishing platform for field guides

and identification tools..................................................................................... 13
Gregor Hagedorn, Gisela Weber, Andreas Plank, Mircea Giurgiu, Andrei Homodi,
Cornelia Veja, Gerd Schmidt, Pencho Mihnev, Manol Roujinov, Dagmar Triebel,
Robert A. Morris, Bernhard Zelazny, Edwin van Spronsen, Peter Schalk,
Christian Kittl, Robert Brandner, Stefano Martellos, Pier Luigi Nimis

A search tool for the digital biodiversity resources of KeyToNature................ 19

Mircea Giurgiu, Andrei Homodi, Cornelia Veja, Gregor Hagedorn, Pier Luigi Nimis

Developing Web-based Search Portals on Birds

for Different Target Groups.............................................................................. 25
Renate Steinmann, Andreas Strasser, Andrea Mulrenin, Amy Trayler, Sander Pieterse,
Ivan Teage, Michael De Giovanni, John J. Borg, Noel Zammit

Simple Identification Tools in FishBase........................................................... 31

Nicolas Bailly, Rodolfo Reyes Jr., Rachel Atanacio, Rainer Froese

The Catalogue of Life: towards an integrative taxonomic backbone

for biodiversity................................................................................................. 37
Frank A. Bisby, Yuri R. Roskov

BHL-EUROPE: Biodiversity Heritage Library for Europe................................ 43

Jana Hoffmann, Henning Scholz

A Pan-European Species-directories Infrastructure (PESI)............................. 49

Yde de Jong

ViBRANT–Virtual Biodiversity Research and Access Network

for Taxonomy................................................................................................... 53
Dave Roberts, Vince Smith

Identifications in BioPortals™ . ....................................................................... 55

Wouter Addink, Edwin van Spronsen, Peter H. Schalk

Types of identification keys............................................................................. 59

Gregor Hagedorn, Gerhard Rambold, Stefano Martellos

Learning, Identifying, Sharing.......................................................................... 65

Philippe A. Martin, Noël Conruyt, David Grosser

vii
Identification with iterative nearest neighbors using domain knowledge......... 71
David Grosser, Noël Conruyt, Henri Ralambondrainy

A MediaWiki implementation of single-access keys........................................ 77

Gregor Hagedorn, Bob Press, Sonia Hetzner, Andreas Plank, Gisela Weber,
Sabine von Mering, Stefano Martellos, Pier Luigi Nimis

Simple matrix keys from Excel spreadsheets.................................................. 83

Gregor Hagedorn, Mircea Giurgiu, Andrei Homodi

Wiki keys on mobile devices........................................................................... 89

Gisela Weber, Gregor Hagedorn

A Wiki-based Key to Garden and Village Birds............................................... 95

Tomi Trilar

Wiki-keys for the ferns of the Flora of Equatorial Guinea................................ 99

Francisco Cabezas, Carlos Aedo, Patricia Barberá, Manuel De la Estrella, Maximiliano Fero,
Mauricio Velayos

MyKey: a server-side software to create customized decision trees............. 107

David Gérard, Régine Vignes Lebbe

Xper²: managing descriptive data from their collection to e-monographs..... 113

Visotheary Ung, Florian Causse, Régine Vignes Lebbe

FRIDA 3.0 Multi-authored digital identification keys in the Web.................... 115

Stefano Martellos

Flora Bellissima, an expert software to discover botany

and identify plants......................................................................................... 121
Thierry Pernot, Daniel Mathieu

Modifiable digital identification keys ............................................................. 127

Edwin van Spronsen, Stefano Martellos, Dennis Seijts, Peter Schalk, Pier Luigi Nimis

The Open Key Player: A new approach for online interaction

and user-tracking in identification keys......................................................... 133
Mircea Giurgiu, Andrei Homodi, Edwin van Spronsen, Stefano Martellos, Pier Luigi Nimis

Improvement of identification keys by user-tracking..................................... 137

Gerd Schmidt, Mircea Giurgiu, Sónia Hetzner, Fred Neumann

ARIES: an expert system supporting legislative tasks.

Identifying animal materials using the Linnaeus II software ......................... 145
Leo W.D. van Raamsdonk

An integrated system for producing user-specific keys on demand:

an application to Italian lichens..................................................................... 151
Juri Nascimbene, Stefano Martellos, Pier Luigi Nimis

“Flora Italiana Digitale”:

an interactive identification tool for the Flora of Italy..................................... 157
Riccardo Guarino, Sabina Addamiano, Marco La Rosa,
Sandro Pignatti

viii
eFlora and DialGraph, tools for enhancing identification processes
in plants......................................................................................................... 163
Fernando Sánchez Laulhé, Cecilio Cano Calonge, Antonio Jiménez Montaño

A catalogue of bird bones: an exercise in semantic web practice................. 171

Gudmundur Gudmundsson, Seth D. Brewington, Thomas H. McGovern, Aevar Petersen

Anthos.es: 10 years showing Spanish plant diversity information

in the Internet................................................................................................ 177
Leopoldo Medina, Carlos Aedo

An interactive tool for the identification of airborne and food fungi................ 183
Giovanna Cristina Varese, Antonella Anastasi, Samuele Voyron, Valeria Filipello Marchisio

The Estonian eFlora...................................................................................... 189

Tiina Randlane, Malle Leht, Andres Saag

Keys to plants and lichens on smartphones: Estonian examples................. 195

Andres Saag, Tiina Randlane, Malle Leht

IIKC: An Interactive Identification Key for female Culicoides

(Diptera: Ceratopogonidae) from the West Palearctic region........................ 201
Bruno Mathieu, Catherine Cêtre-Sossah, Claire Garros, David Chavernac,
Thomas Balenghien, Régine Vignes Lebbe, Visotheary Ung, Ermanno Candolfi,
Jean-Claude Delécolle

Indochinese bamboos: biodiversity informatics to assist the identification

of “vernacular taxa”....................................................................................... 207
My Hanh Diep Thi, Régine Vignes Lebbe, Ha Phuong Nguyen, Bich Loan Nguyen Thi

Identification tools as part of Feedsafety research: the case of ragwort....... 213

Leo W.D. van Raamsdonk, Patrick Mulder, Michel Uiterwijk

Two identification tools applied on Mascarene’s corals genera (Xper2)

and species (IKBS)........................................................................................ 217
Yannick Geynet, Noël Conruyt, David Grosser, Gérard Faure, David Caron

Interactive, illustrated, plant identification keys: an example

for the Portuguese flora................................................................................. 219
Maria Helena Abreu Silva, Rosa Maria Ferreira Pinho, Lísia Graciete, Martins Pereira Lopes,
Paulo Cardoso da Silveira

The ORCHIS software used to identify 100 orchids species of Lao PDR..... 221
Pierre Bonnet, André Schuiteman, Boukhaykhone Svengsuksa, Daniel Barthélémy,
Vichith Lamxay, Soulivanh Lanorsavanh, Khamfa Chanthavongsa, Pierre Grard

A collaborative and distributed identification tool for plants .......................... 223

Philippe Laroche

Alternative 2D and 3D Form Characterization Approaches

to the Automated Identification of Biological Species.................................... 225
Norman MacLeod

ix
VeSTIS: A Versatile Semi-Automatic Taxon Identification System
from Digital Images....................................................................................... 231
Nikos Nikolaou, Pantelis Sampaziotis, Marilena Aplikioti, Andreas Drakos,
Ioannis Kirmitzoglou, Marina Argyrou, Nikos Papamarkos, Vasilis J. Promponas

Iterative Search with Local Visual Features for Computer Assisted Plant
Identification.................................................................................................. 237
Wajih Ouertani, Pierre Bonnet, Michel Crucianu, Nozha Boujemaa, Daniel Barthélémy

Image data banks and geometric morphometrics ........................................ 243

Anna Loy, Dennis E. Slice

Outline analysis for identifying Limodorum species from seeds.................... 249

Sara Magrini, Sergio Buono, Emanuele Gransinigh, Massimiliano Rempicci, Silvano Onofri,
Anna Scoppola

Geometric morphometrics as a tool to resolve taxonomic problems:

the case of Ophioglossum species (ferns).................................................... 251
Sara Magrini, Anna Scoppola

Geometric morphometric analysis as a tool to explore covariation

between shape and other quantitative leaf traits in European white oaks.... 257
Vincenzo Viscosi, Anna Loy, Paola Fortini

Landmark based morphometric variation in Common dolphin

(Delphinus delphis L.,1758)........................................................................... 263
Paola Nicolosi, Anna Loy

DNA barcoding: theoretical aspects and practical applications..................... 269

Maurizio Casiraghi, Massimo Labra, Emanuele Ferri, Andrea Galimberti, Fabrizio De Mattia

Strength and Limitations of DNA Barcode

under the Multidimensional Species Perspective.......................................... 275
Valerio Sbordoni

DNA Barcoding and Phylogeny of Patellids from Asturias

(Northern Spain)............................................................................................ 281
Yaisel Juan Borrell, Fernando Romano, Emilia Vázquez, Gloria Blanco, Jose Antonio
Sánchez Prado

Molecular Identification of Italian Mouse-eared Bats

(genus Myotis)............................................................................................... 289
Andrea Galimberti, Adriano Martinoli, Danilo Russo, Mauro Mucedda, Maurizio Casiraghi

Identifying algal symbionts in lichen symbioses............................................ 295

Martin Grube, Lucia Muggia

Identification of polymorphic species within groups of morphologically

conservative taxa: combining morphological and molecular techniques....... 301
Kim Larsen, Elsa Froufe

Coffee species and varietal identification...................................................... 307

Patrizia Tornincasa, Michela Furlan, Alberto Pallavicini, Giorgio Graziosi

x
Mislabelling in megrims: implications for conservation.................................. 315
Victor Crego-Prieto, Daniel Campo, Juliana Perez, Eva Garcia-Vazquez

Seeds in subtribe Orchidinae (Orchidaceae):

the best morphological tool to support molecular analyses........................... 323
Roberto Gamarra, Emma Ortúñez, Ernesto Sanz, Iris Esparza, Pablo Galán

Lentils biodiversity: the characterization of two local landraces.................... 327

Vincenzo Viscosi, Manuela Ialicicco, Mariapina Rocco, Dalila Trupiano, Simona Arena,
Donato Chiatante, Andrea Scaloni, Gabriella Stefania Scippa

A model study for tardigrade identification.................................................... 333

Roberto Bertolani, Lorena Rebecchi, Michele Cesari

DNA Barcoding of Philippine plants............................................................... 341

Esperanza Maribel G. Agoo

Molecular and ecophysiological characterisation of the Tunisian bee:

Apis mellifera intermissa......................................................................... 343
Mohamed Chouchene, Naima Barbouche, Lionel Garnery, Michel Baylac

Biological identifications through mitochondrial and nuclear molecular

markers: the case of commercially important crabs from Indian EEZ........... 345
Sherine Sonia Cubelio, K. K. Bineesh, K. Raj, Suraj Tewari, Achamveettil Gopalakrishnan,
Valaparambil Saidumohammad Basheer, Wazir Singh Lakra

Barcoding Fauna Bavarica – Capturing Central European

Animal Diversity............................................................................................. 347
Lars Hendrich, Michael Balke, Gerhard Haszprunar, Axel Hausmann, Paul Hebert,
Stefan Schmidt

Molecular techniques for identifying North Sea fauna................................... 349

Thomas Knebelsberger, Sandra Ditzler, Silke Laakmann, Inga Mohrbeck,
Michael J. Raupach

DNA Bank Network – connecting biological collections and sequence

databases by long-term DNA storage with online accession........................ 351
Matthias Geiger, Nicolas Straube

Mitochondrial DNA sequences for forensic identification of the endangered

whale shark, Rhincodon typus (Smith, 1828): A Case study......................... 353
Kavungal Abdulkhadar Sajeela, Chandran Rakhee, Janardanan Nair Rekha,
Achamveettil Gopalakrishnan, Valaparambil Saidumohammad Basheer, Joe Kizhakkudan
Shoba, Kizhakkudan Joe, Wazir Singh Lakra

An assignment-based e‑learning course on the use

of KeyToNature e-keys.................................................................................. 355
Pencho Mihnev, Nadezhda Raycheva

User needs for interactive identification tools to organisms

employed in the EU-Project KeyToNature . .................................................. 361
Astrid Tarkus, Emanuel Maxl, Christian Kittl

Teaching biodiversity with online identification tools from KeyToNature:

a comparative study...................................................................................... 367
Felicia Boar, Adelhaida Kerekes

xi
Digital Tools in the Botanical Garden of Madrid............................................. 373
Marina Ferrer, Esther García

Use of KeyToNature Identification Tools in the Schools of Slovenia............. 379

Irena Kodele Krašna

New key-tools for pollen identification in research and education................. 383

Jade Dupont, Nathalie Combourieu Nebout, Jean-Pierre Cazet, Florian Causse,
Régine Vignes Lebbe

The UK urban tree survey............................................................................. 389

Bob Press

Tree School – A new Innovation for Science and Education......................... 395

Della Hopkins, Karen James

Engaging Schools in Cutting Edge Science:

From the Educator’s Perspective.................................................................. 401
Adrian Richardson, Della Hopkins

Educational or emotional languages?

An interactive experiment with the Lucanian flora (S-Italy)........................... 405
Riccardo Guarino, Patrizia Menegoni, Sandro Pignatti

Online sharing educational content on biodiversity topics: a case study

from organic agriculture and agroecology..................................................... 411
Nikos Palavitsinis, Nikos Manouselis, Kostas Kastrantas, John Stoitsis, Xenofon Tsilibaris

JSTOR Plant Science.................................................................................... 417

Michael Sean Gallagher

ecoBalade: Towards a workflow for Citizen Science Nature Trails . ............. 419
Julie Chabalier, Khaled Talbi, Patrick Peters, Amandine Sahl, Olivier Coullet,
Olivier Assunçao, Olivier Rovellotti

Electronic data recording tools and identifying species

in the field...................................................................................................... 421
Alexander Kroupa, Anke Hoffmann, Juan Carlos Monje, Christoph L. Häuser

Cost Assessment of the Field Measurement of Biodiversity:

a Farm-scale Case Study.............................................................................. 423
Stefano Targetti, Davide Viaggi, David Cuming

Markets for biodiversity information products: real or imaginary?................. 429

Bill Hominick, Peter Schalk

A Basic Business Model for Commercial Application

of Identification Tools..................................................................................... 437
Christian Kittl, Peter Schalk, Nicola Dorigo Salamon, Stefano Martellos

Keys to Nature: A test on the iPhone market................................................. 445

Rodolfo Riccamboni, Alessio Mereu, Chiara Boscarol

Author Index.................................................................................................. 451

xii
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 1-6.
ISBN 978-88-8303-295-0. EUT, 2010.

Devising the EDIT

Platform for Cybertaxonomy
Walter G. Berendsohn

Abstract — This contribution describes the original ideas and preparatory

work that led to the implementation of the EDIT Platform for Cybertaxonomy,
a computing environment supporting the entire taxonomic workflow. It also
briefly describes the current state of development of the project, which will end
its EU-funded period in February, 2011.

Index Terms — biodiversity informatics, cybertaxonomy, taxonomic

computing, taxonomy.

—————————— u ——————————

1 Introduction

T
axonomic research is traditionally a highly collaborative endeavour. The
EU project EDIT (European Distributed Institute of Taxonomy) brings
together a consortium including most of the largest European natural
history museums. EDIT aims at integrating taxonomic research at multiple
levels: research policies, collection management, training, outreach and public
relations, and research infrastructure.
Natural history museums are information and knowledge institutions. In EDIT,
the relatively new area of information technologies was seen as a major chance
to integrate the activities of the project partners. Therefore, a third of the project
funding was dedicated to create the “Internet Platform for Cybertaxonomy”.
————————————————
W. G. Berendsohn is with the Botanic Garden and Botanical Museum Berlin-Dahlem, Freie Univer-
sität Berlin.
Instead of filling this page with co-authors, the author here lists the collaborators in the EDIT work-
package 5 “Internet Platform for Cybertaxonomy” in a comprehensive form:
The task leaders in the work package were Wieslaw Bogdanowicz, Museum of Invertebrate Zool-
ogy, Polish Academy of Sciences (MIZPAN); Andras Gubanyi, Hungarian Natural History Museum,
Budapest (HNHM); Anton Güntsch, BGBM; Christoph Häuser, State Museum for Natural History,
Stuttgart (SMNS) (now at MfN); Mark Jackson, Royal Botanic Gardens, Kew (RBGK); Jorge Lobo,
Museo Nacional de Ciencias Naturales, Madrid (CSIC); Karol Marhold, Institute of Botany, Slovak
Academy of Sciences (IBSAS); Patricia Mergen, Royal Museum for Central Afrika, Tervuren
(RMCA); Martin Pullan, Royal Botanic Garden, Edinburgh (RBGE); Henning Scholz, Museum
für Naturkunde, Berlin (MfN); Jane Smith, Natural History Museum, London (NHML); Eduard
Stloukal (CUB) and Régine Vignes, Université Pierre et Marie Curie Paris 6 (UPMC). The EDIT
development team was first led by Markus Döring and from the second year on jointly by Andreas
Kohlbecker and Andreas Müller (BGBM Berlin). Team members were (alphabetically; independ-
ent of the time span they worked for EDIT): Anahit Badadshanjan (BGBM), Elek Bozóky-Szeszich
(HNHM), Garin Cael (RMCA),Pepe Ciardelli (BGBM), Ben Clark (RBGK), Nils Clark-Bernhard
(independent), James Davy (RMCA), Marco Figuidero (NHML), Helene Fradin (UPMC), Giovanni

1
Thirteen institutions from 8 countries directly participated in the workpackage
elaborating the Platform, 7 institutions were involved in software development
(programming), with a total of 25 developers (12 concurrent) busy forging the
code.

2 Arriving at the Specification

At the starting point of the project in March 2005 we knew that we had a well-
resourced project, but we also knew that software development may exceed all
cost expectations, especially when carried out in large cooperative projects and
(worse) when relatively new technologies were to be used. We were determined
not to re-invent the wheel and reuse existing software as much as possible.
The first project year was spent analysing existing software, standards,
infrastructures at partner institutions, and requirements from taxonomists,
especially from those involved in the EDIT “exemplar group” treatments.
We had realised that we would have to cope with a heterogeneous
institutional landscape, with widely differing levels of IT capacity. Nevertheless,
results from the analysis of institutional IST infrastructures were somewhat
frustrating. In spite of the institutions’ central role in taxonomic information
provision, appropriate infrastructures were often completely lacking, and where
in existence, they were rarely outward-looking. Intra-institutional coordination
mechanisms were complex, if existing at all. IT developments were mostly
depending on soft money funding, with all the consequences for personnel,
scope and sustainability. In consequence, we knew that instigating institutional
collaboration (the overarching aim of a Network of Excellence project) would
present a long-term challenge. It also became clear that we needed a solution
independent of database management and operation systems.
We knew that the taxonomic data domain was well analysed and to a large
extend covered by existing information models and data standards. What was
missing was an information model incorporating all these existing, partially
overlapping schemes.
It also became clear that good software applications were available for
taxonomists, for example for descriptive data and the generation of identification
keys. Good technologies and an open source environment were available
for geographical applications. Handling of bibliographic references was well
covered by existing scientific software, including access to citation databases.
————————————————
Gaias (independent), Marcin Gąsior (MIZPAN), Marc Geoffroy (BGBM), Niels Hoffmann (BGBM),
Patricia Kelbert (BGBM), Alexander Kroupa (SMNS, MfN), Dilan Latif (SMNS), Eun-Mok Lee
(BGBM), Katja Luther (BGBM), Ôna Maiocco (UPMC), Bart Meganck (RMCA), Dominik Mikiewicz
(MIZPAN), Maciej Posluszny (MIZPAN), Francisco Revilla (BGBM), Pere Roca Ristol (CSIC),
Pablo Sastre Olmos (CSIC), Bernard Scaife (NHML), Dusan Senko (IBSAS), Lutz Suhrbier (Freie
Universität Berlin), Franck Theeten (RMCA), Maxime Venin (UPMC), and Julius Welby (NHML).
Exploratory work, information gathering, modelling, and software testing was carried out by Lisa
Banfield (RBGE), Franck Dorkelt (INRA), Charles Hussey (NHML), Imre Kilian (HNHM), Boris
Jacob (BGBM), Lellani Farina-Crespo (RMCA), Naomi Korn (NHML), Wolf-Henning Kusber
(BGBM), Elise Kuntzelmann (UPMC), Barbora Šingliarová (SAVBA), Stanislav Spaniel (SAVBA),
David Taylor (RBGK), Dorottya Varsányi (HNHM), Elke Zippel (BGBM), and Magda Zytomska
(MIZ PAN). Workpackage coordination was effected by Malte Ebach, Anke Hoffmann, and Agnes
Kirchhoff (in that sequence) at the BGBM.

2
The two most important data access needs of taxonomists were tackled by
large-scale international initiatives: access to specimen information by the
Global Biodiversity Information Facility (GBIF) and access to digitised taxonomic
literature by the Biodiversity Heritage Library initiative (BHL). However, an overall
integration to cover the needs of taxonomists was lacking. Some of the existing
solutions would be difficult to fully integrate, because they depended on specific
database or operating systems. Very few solutions existed that supported the
full complexity of nomenclatural rules and taxonomic data relations. None was
encompassing the full range of data.
Being faced with the unique chance the EDIT project offered, we took the
decision to devise, implement, provide and propagate a comprehensive solution
for taxonomic computing, the EDIT Platform for Cybertaxonomy [1]. The primary
objective was to support, enhance and increase the efficiency of the taxonomic
work process, for individuals and teams of taxonomists. An explicit aim was to
hide the complexity of taxonomic information processing as far as possible, so
that it was not inhibiting the workflow, as traditional software applications often
did. We knew that new software technologies now offered solutions for some of
the problems that had been in the way of creating user-friendly software earlier
on. At the same time, the underlying framework had to ensure reusability of the
data, seen as the key to future acceleration of taxonomic work processes. On
the technical side, hard- and software platform independence had to be ensured
to guarantee broad acceptance; at least the newly developed solutions had to
be freely available and open source; and for developers wanting to use it for their
software projects the solution should provide an API (Application Programming
Interface) as well as web services.
In order to achieve these aims, we had to strive to professionalise taxonomic
software development. Such a comprehensive solution needed adherence to a
strict technological framework. Searching for this framework for development,
we looked at content management systems, particularly because using this was
a decision taken early-on by another EDIT workpackage -- the “Scratchpads”
approach [2]. We saw and see the virtue of this approach for group
communication, information dissemination, web publication and aggregation,
but we continue to posit that this is not a viable solution for the kind of in-depth
treatment of complex data that taxonomists require in their work process. Our
aim was principally to support the actual generation of taxonomic data. After
weighing several options, Java software development was accepted to provide
the most acceptable general framework for Platform application development.
Web publication for the Platform can still be realised using content management
systems, taking advantage of the Platform’s web services (as demonstrated by
the EDIT Data Portal implementations).

3 The Results
Space restrictions allow for only a brief summary of the results achieved so-
far. Ciardelli & al. [3] provide a more extensive overview; for full information
please refer to the Platform website [4].
The EDIT Common Data Model (CDM) now fully covers the data that are

3
used for systematic treatments resulting from the taxonomic work process
(monographs, flora and fauna treatments, and taxonomic checklists).
This includes the full complexity of nomenclatural information (botany and
zoology), the entire range of taxonomic relationships (including multiple
taxonomic hierarchies, synonymies, concept relationships etc.), structured and
unstructured descriptive data, geographic information, literature, and specimen
data. The CDM is based on existing information models (e.g. the Berlin Model
for taxonomic information [5] or the BioCISE model for natural history collections
[6]) as well as the standardisation efforts of “Biodiversity Information Standards
(TDWG)” -- formerly known as Taxonomic Databases Working Group. Important
TDWG standards in this context were the Taxonomic Concept Schema [7], SDD
(Structured Descriptive Data) [8], and Access to Biological Collection data [9].
The CDM forms the base for the programming code implemented and made
available as the CDM Programming Library. The application programming
interface or the web services based on the CDM library can be used by
programmers to create applications for taxonomists. New functionality created
becomes part of the CDM Library after in-depth testing.
As a first step in a user project, a Community Data Store is created, i.e. a
database that offers the entire scope of information that is covered by the CDM.
This can be installed on an individual’s computer, on a server in an institutional
network, or on servers accessible through the Internet.
Three years of development within EDIT has resulted in a number of CDM-
based applications, the two most important of which are the EDITor and the
CDM Data Portal.
For data input, the EDIT Taxonomic Editor (or EDITor) was developed [10].
It combines an innovative user interface (e.g. allowing full text entry in place
of the traditional form-based approach) with the possibility to edit every detail
of the database content. The project database can be configured, e.g. by
determining which kind of factual data is going to be available for data input
(e.g. distributions, threat category, etc.) and which standard terms (if any) are
allowed (e.g. TDWG area codes, IUCN threat categories). The taxonomic tree
can be displayed and used for navigation and for restructuring by drag and
drop. Apart from the taxon-centric standard interface, a “power user interface”
presents the data in spreadsheet-like fashion and allows bulk editing and data
cleaning. Import and export functionality with several pre-defined formats and
standards is implemented. Users can install the EDITor locally on their computer
for individual work or access to an institutional Community Data Store, or use it
remotely.
The CDM Data Portal is a Drupal-based website used to publish the data
in the Community Data Store. It is highly configurable as to displayed content
and design. It also offers a taxonomic tree for navigation as well as simple and
advanced search functions. The displayed taxon is linked to external resources
such as GBIF, BioCASE, BHL, Tropicos, NCBI, Google Images etc. to offer
integration with the existing biodiversity information infrastructure. The individual
taxon page shows the standard taxonomic data (if the user has configured it
that way), i.e. description and factual information. The distribution is visualised
through the integrated map viewer (an application of the EDIT Geo-Platform).

4
All content can be bibliographically referenced. Synonyms can be displayed as
homotypic groups, followed by the respective type information. Nomenclatural
references are linked to the protologue record (scanned file or web link, where
available). An unlimited number of images can be linked and the image gallery
offers display in different resolutions and features the image metadata (artist,
copyright etc.). CDM Data Portals are in productive use, examples include the
EDIT exemplar group sites, for example that for the International Cichorieae
Network [11].
Software bundles with the EDITor and Data Portal can be downloaded from
the CDM Setup site at http://wp5.e-taxonomy.eu/cdm-setups [12].
Apart from on-line output, functions for pre-formatted print output are being
implemented. Out of the (EDIT) box there will be ready made stylesheets for
a botanical monograph, a zoological monograph, botanical and zoological
checklist, and for the publication of new names in specific journals. Institutional
developers will be able to create custom stylesheets conforming to the editorial
rules of their in-house publication series.
EDIT has also developed a number of software applications that are not
directly CDM based, of which three should at least be mentioned here: (i) The
EDIT Geo-Platform [13], [14]; (ii) ViTaL, the Virtual Taxonomic Library, which
(in close collaboration with the Biodiversity Heritage Library Europe project)
provides an integrated index to taxonomic literature, and (iii) the observation
databases and data input tools for the All Species Inventories and Monitoring
sites of EDIT workpackage 7.

4 Conclusion
For more than 2 decades there are efforts in joint modelling, standard-
building and application development that provide us with excellent knowledge
of the taxonomic domain’s information structures and business rules. The
EDIT Platform is the attempt by European institutions to create a sustainable,
collaborative, and comprehensive software solution to increase the efficiency of
the taxonomic work process

Acknowledgements

Apart from the EDIT collaborators mentioned in the title page footnote, we would also
like to thank numerous taxonomists for their input, in particular those involved with the
EDIT exemplar groups: Irina Brake (NHML), Bill Baker, Simon Mayo and Soraya Villalba
(RBGK), and Norbert Kilian, Ralf Hand and Eckhard von Raab Straube (BGBM). Gregor
Hagedorn gave most valuable advice especially with regard to descriptive data modelling.
This work was supported by the European Commission’s 6th Framework Programme
(Contract No.: 018340).

References
[1] M. Döring and W. G. Berendsohn, “A general concept for the design of the EDIT Platform for
Cybertaxonomy”, EDIT newsletter, vol. 3, pp. 13-15, 2007.
[2] V. S. Smith, S. D. Rycroft, K. T. Harman, B. Scott and D. Roberts. “Scratchpads: a data-
publishing framework to build, share and manage information on the diversity of life”, BMC

5
 Bioinformatics, vol. 10 (Suppl 14): S6doi:10.1186/1471-2105-10-S14-S6, 2009.
[3] P. Ciardelli, P. Kelbert, A. Kohlbecker, N. Hoffmann, A. Güntsch and W. G. Berendsohn,
“The EDIT Platform for Cybertaxonomy and the taxonomic workflow: selected Components”,
Lecture Notes in Informatics (LNI), vol. 154, pp. 625-638, 2009.
[4] Anonymous, “EDIT Platform for Cybertaxonomy, “http://wp5.e-taxonomy.eu, 2010.
[5] W. G. Berendsohn, M. Döring, M. Geoffroy, K. Glück, A. Güntsch, A. Hahn, W.-H. Kusber,
J. -J. Li, D. Röpert and F. Specht, “The Berlin Taxonomic Information Model”, Schriftenreihe
Vegetationsk., vol. 39, pp. 15-42, 2003.
[6] W. G. Berendsohn, A. Anagnostopoulos, G. Hagedorn, J. Jakupovic, P. L. Nimis, B. Valdés,
A. Güntsch, R. Pankhurst and R. J. White, “A comprehensive reference model for biological
collections and surveys”, Taxon, vol. 48, pp. 511-562, 1999. (Preprint: http://www.bgbm.org/
biodivinf/docs/CollectionModel/, accessed 2010).
[7] R. Hyam (Ed.), “Taxonomic Concept Schema – User Guide”, Biodiversity Information
Standards (TDWG), http://www.tdwg.org/fileadmin/subgroups/tnc/User_Guide.pdf, 2008.
[8] G. Hagedorn, K. Thiele, R. Morris and P. B. Heidorn, “The Structured Descriptive Data (SDD)
w3c-xml-schema, version 1.0”, Biodiversity Information Standards (TDWG), http://www.tdwg.
org/standards/116/, 2005 (accessed 2010).
[9] W. G. Berendsohn (ed.), “Access to Biological Collection Data”, Biodiversity Information
Standards (TDWG), http://wiki.tdwg.org/ABCD/, 2010.
[10] P. Ciardelli, A. Müller, A. Güntsch and W. G. Berendsohn, “Introducing the EDIT Desktop
Taxonomic Editor”. In: A. L. Weitzman and L. Belbin (eds.), Proceedings of TDWG 2008,
Fremantle, Australia, http://www.tdwg.org/proceedings/article/view/325, 2008.
[11] R. Hand, N. Kilian and E. von Raab-Straube (eds.), International Cichorieae Network:
Cichorieae Portal, http://wp6-cichorieae.e-taxonomy.eu/portal/, 2009+ (continuously updated).
[12] A. Kirchhoff, A. Kohlbecker, N. Hoffmann and A. Güntsch, “CDM setups site - How to install
the software modules of the EDIT Platform for Cybertaxonomy”, EDIT Newsletter, vol. 21, pp.
6-7, 2010.
[13] P. Sastre, P. Roca, J. M. Lobo and EDIT co-workers: “A Geoplatform for improving accessibility
to environmental cartography”, J. Biogeogr., vol. 36, p. 568, 2009.
[14] P. Mergen and B. Meganck, “Geospatial components for EDIT”, EDIT Newsletter, vol. 5, pp.
14-17, 2007.

6
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 7-11.
ISBN 978-88-8303-295-0. EUT, 2010.

Descriptive Data in the EDIT

Platform for Cybertaxonomy
Maxime Venin, Agnes Kirchhoff, Hélène Fradin, Anton Güntsch,
Niels Hoffmann, Andreas Kohlbecker, Elise Kuntzelmann,
Ôna Maiocco, Andreas Müller, Régine Vignes Lebbe,
Walter G. Berendsohn

Abstract — This paper describes the integration of structured descriptive data

in the EDIT platform for Cybertaxonomy. The platform is composed of several
software modules supporting the taxonomic workflow from data capture
and storage to publication. Descriptive data play an important role within
the taxonomic work process. The integration of these data via import/export
modules to and from the platform and the publication as natural language
output or as keys are explained.

Index Terms —platform, software, taxonomy, description, key, natural

language, structured descriptive data, SDD, Common Data Model, EDIT.

—————————— u ——————————

1 Introduction

O
ne of the achievements of the European Distributed Institute of
Taxonomy (EDIT) [1] is the Internet Platform for Cybertaxonomy, which
provides software tools supporting and accelerating the taxonomic
workflow (Fig. 1). “A main goal of the Platform is to provide an open architecture
to allow connection and integration of existing applications and to provide new
developments where necessary” [2]. The Platform is based on the Common Data
Model (CDM), which is essentially a description of all data that can be used and
edited in the Platform, such as taxon names and concepts, literature references,
specimens, distributions, and structured and unstructured descriptive data. All
data are stored in a repository known as the CDM Community Store. Different
communities can set up their own Store, e.g. to work on a specific monograph,
checklist or Flora/Fauna treatment.
The various Platform components are linked by interfaces to the Community
Store, for example the Taxonomic Editor (EDITor) for data entry and the EDIT
————————————————
M. Venin, H. Fradin, E. Kuntzelmann, Ô. Maiocco, and R. Vignes Lebbe are with the Muséum Na-
tional d’Histoire Naturelle (UPMC-MNHN), CP48, 57 rue Cuvier, 75231 Paris Cedex O5, France,
E-mail: [email protected].
A. Kirchhoff, A. Güntsch, N. Hoffmann, A. Kohlbecker, A. Müller, W. G. Berendsohn are with the
Botanic Garden and Botanical Museum Berlin-Dahlem, Freie Universtität Berlin, Königin Luise Str.
6-8, 14195 Berlin, Germany, Email: [email protected].

7
Data Portal for data publication (see Berendsohn, this volume).
The CDM code Library forms the heart of the Platform software. It enables
the individual Platform components to interact. Software developers can use
the Library to implement taxonomic software without having to re-create the
functionality already developed.

  1 – Overview over the software modules and functions of the EDIT Platform for
Fig.
Cybertaxonomy (EDIT Cybergate).

The EDIT Platform tools are designed to assist the taxonomist from fieldwork
to publication of results, including the management of descriptive data, which
play a key role in the taxonomic revision process.
Descriptive data are one of the most important categories of information
produced by taxonomists when describing new species or performing taxonomic
revisions. Traditionally, taxonomic descriptions were handled as text. However,
storing and handling of descriptive data in a highly structured form has strong
advantages: data exchange and integration is facilitated, and identification keys
(both for printed output and interactive) as well as “natural language descriptions”
can be generated automatically and in multiple languages.
There are several established software tools to manage and analyse descriptive
data, some of them already existing for decades (e.g. DELTA [3]). Consequently,

8
it was decided at the outset of the EDIT project not to develop another application
but to integrate existing descriptive tools into the EDIT Platform. The key for this
is that the CDM complies with the SDD (Structured Descriptive Data) standard
[4]. SDD is the current TDWG (Biodiversity Information Standards) standard
for descriptive data. Many of the existing descriptive data managing tools (e.g.
Lucid [5], Xper² [6], and DiversityDescriptions [7]) do support import and export
of SDD conformant data, allowing their users to exchange descriptive data.

2 Structured descriptive data connected to the EDIT platform

2.1 The SDD-CDM Import and Export Module

An SDD-CDM import/export module was developed to integrate descriptive

data with the EDIT Platform. Once in the CDM Community Store, the data can
be published together with the other information on a specific taxon on the
Internet, e.g. by means of the EDIT Data Portal software.
The SDD export from the CDM provides the possibility to use the different
software tools and benefit from their specific analytical or output capabilities
(interactive identification, comparison of taxa, statistics, etc.) (Fig. 2).
From the technical point of view the general idea behind the import of SDD
elements is to create corresponding elements in the CDM in order to allow
seamless export/import roundtrips.

  2 – Data exchange between descriptive software tools and the EDIT platform for
Fig.
Cybertaxonomy.

9
2.2 Display of descriptive data as natural language descriptions

The CDM Code Library now includes a feature to generate clear and easy to
read output of the descriptive CDM data. The structure of the output can be pre-
defined, which allows the scientists to keep a constant scheme, a very helpful
feature when preparing output that has to adhere to a defined editorial standard.
The natural language description output can be used for publications on the web
or for print publications, or simply as a readable preview to control the content
of the database.

2.3 Display of keys

A next step in the processing of structured descriptive data is the possibility

to automatically generate identification keys from the CDM data. The CDM
Code Library now supports output in the form of dichotomous or polytomous
identification keys that can be shown on taxon pages of higher taxa in the CDM
data portals. Clickable links lead to other key entries or the identified taxa.
The integration of interactive keys will not be possible within the current EDIT
project period, for this Platform users have to resort to existing tools. An example
of such an integration is given by the CATE project [8], where the LUCID Player
is used to provide interactive key functionality on the website.

3 Future developments
The EDIT Taxonomic Editor (EDITor) is the main data entry tool of the EDIT
Platform. It allows the editing and presentation of taxonomic information such
as classifications, synonyms, taxonomic concepts, descriptions, distributions,
specimens and literature references. As any other data in the EDIT Platform this
kind of information is stored in the CDM Community Store.
As mentioned above, the EDIT Platform allows choosing among several
software tools for the management of descriptive data. One of those is Xper2,
“a management system for storage, editing, analysis and online distribution
of descriptive data” [9], which also dynamically creates interactive keys for
identifying specimens. This software was chosen as a way forward for the
integration of descriptive data into the EDIT platform, because it is Java-based,
non-commercial, has been created by an EDIT partner, and it can be integrated
with the EDIT Taxonomic Editor.
In the long term, full integration of Xper² with the Taxonomic Editor is the aim.
A shorter term solution will be to enable Xper² to directly work with the data in a
CDM Community Store. Xper² could then be opened via the EDITor, running as
a separate application, but using the same data.

4 Conclusion
With respect to structured descriptive data, the current state of software
development for the EDIT Platform for Cybertaxonomy can be summarised as
follows:

10
 With the SDD-CDM import/export module the integration of descriptive data
into the Common Data Model has been completed.
The natural language module in the CDM library allows users to easily and
rapidly generate output describing taxa and specimens. Thanks to the integration
with other CDM objects and functions in the CDM Code Library, developers
have a very broad range of possibilities to provide users with functions to create,
use and publish natural language descriptions.
Generating simple keys is possible with the CDM library. It is an entirely
automatic process based on the CDM Community Store. Once the descriptive
data have been imported, a taxonomist can directly use this functionality without
any extra work.

Acknowledgement

The authors gratefully acknowledge the support of: the EU 6th Network of Excellence
Project EDIT (European Distributed Institute of Taxonomy, contract No 018340 - GOCE).

References
[1] N. N., “EDIT - European Distributed Institute of Taxonomy”, http://www.e-taxonomy.eu, 2010.
[2] P. Ciardelli, P. Kelbert, A. Kohlbecker, N. Hoffmann, A. Güntsch and W. G. Berendsohn,
“The EDIT Platform for Cybertaxonomy and the taxonomic workflow: selected Components”,
Lecture Notes in Informatics (LNI), vol. 154, pp. 625-638, 2009.
[3] M. J. Dallwitz, “A flexible computer program for generating identification keys”, Syst. Zool., vol.
23, pp. 50-57, 1974.
[4] G. Hagedorn et al., “The Structured Descriptive Data (SDD) w3c-xml-schema, version 1.1.”,
TDWG, http://wiki.tdwg.org/twiki/bin/view/SDD/Version1dot1, 2006.
[5] N. N., “Lucidcentral”, http://www.lucidcentral.org/. Centre for Biological Information Technology,
The University of Queensland, Brisbane, 2010.
[6] N. N., “Xper2”, http://lis-upmc.snv.jussieu.fr/lis/?q=en/resources/software/xper2. Laboratoire
Informatique & Systématique, Paris, 2010.
[7] N. N., “DiversityDescriptions”, http://www.diversityworkbench.net/Portal/DiversityDescriptions,
2008.
[8] N. N., “The CATE Project”, http://www.cate-project.org/, 2010.
[9] V. Ung, G. Dubus, R. Zaragüeta-Bagils and R. Vignes Lebbe, “Xper²: introducing e-Taxonomy”,
Bioinformatics, vol. 26, no. 5, pp. 703-704, available at http://bioinformatics.oxfordjournals.org/
cgi/reprint/btp715v1.pdf, Jan. 2010.

11
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 13-18.
ISBN 978-88-8303-295-0. EUT, 2010.

An online authoring and

publishing platform for field
guides and identification tools
Gregor Hagedorn, Gisela Weber, Andreas Plank, Mircea Giurgiu,
Andrei Homodi, Cornelia Veja, Gerd Schmidt, Pencho Mihnev,
Manol Roujinov, Dagmar Triebel, Robert A. Morris,
Bernhard Zelazny, Edwin van Spronsen, Peter Schalk,
Christian Kittl, Robert Brandner, Stefano Martellos, Pier Luigi Nimis

Abstract — Various implementation approaches are available for digital field

guides and identification tools that are created for the web and mobile devices.
The architecture of the “biowikifarm” publishing platform and some technical
and social advantages of a document- and author-centric approach based
on the MediaWiki open source software over custom-developed, database
driven software are presented.

Index Terms — field guides, flora, fauna, identification tools, social software,
DELTA, SDD, MediaWiki, agile development.

—————————— u ——————————

1 Introduction

D
igital identification tools may be simple picture guides, printable tabular
tools, or interactive tools (single-access, multi-entry, or multi-access
keys). A mixture of tools and richly illustrated species pages or glossary
definitions is often required. The EU-funded KeyToNature project provides a wide
spectrum of such tools: together with the “biowikifarm.net”, it integrates both the
tools and their content. We describe here the architecture and components or
this internet-based collaborative authoring and publishing platform.

————————————————
G. Hagedorn, G. Weber, A. Plank are with the Julius Kühn-Institute, Federal Research Centre
for Cultiva­ted Plants, Inst. for Epidemiology and Pathogen Diagnostics, Königin-Luise-Str. 19,
D-14195 Ber­lin, E-mail: [email protected] – M. Giurgiu, A. Homodi, C. Veja are with
the Telecomm. Dep., Tech­nical Univ. of Cluj-Napoca, Cluj 400027, Romania – G. Schmidt is
with the Institut f. Lern-Inno­vat­ion, Univ. Erlangen-Nürnberg, D-91052 Erlangen – P. Mihnev, M.
Roujinov are with BIKAM Ltd., Sofia 1505, Bulgaria – D. Triebel is with the Center of the Bavarian
Natural History Collections, Men­zinger Str. 67, D-80638 Munich – R. A. Morris is with the Univ.
of Mas­sa­chusetts, USA – B. Zelaz­ny is with the Internat. Soc. for Pest Information (ISPI) – E. v.
Spron­sen and P. Schalk are with ETI Bioinformat­ics, Amsterdam, The Netherlands – C. Kittl, R.
Brandner are with evolaris next level GmbH, A-8010 Graz – S. Martellos and P. L. Nimis are with
the Department of Life Sciences, Univ. Trieste, I-34127.

13
2 The MediaWiki Software Architecture
The architecture of the biowikifarm publishing platform is based on the
“MediaWiki” open source authoring system [1] that is also used by pro­jects of
the Wikimedia Foundation (e.  g., the Wikipedias, Wikispecies, Wiki­source, or
the Commons Media Repository [2]). MediaWiki provides an object oriented
document storage model of medium granularity (titled chapters called “pages”,
rather than whole works). The storage model is akin in many aspects to the
currently developed “nosql” database mana­ge­ment sys­tems [3] (pre­dating
these developments, however, MediaWiki typically uses mysql). Namespaces
provided by the storage model allow to re-use the basic model for 1st-class
content objects as well as for build­ing objects used in hypertext inclusion.
Examples for the latter are media items (binary plus metadata) in the “File”
namespace or pro­gram­ming blocks and rich text fragments in the “Template”
namespace [4].
The template model provides for flexible schema development. Each template
defines a class with freely definable attributes (equivalent to an “entity type”),
instances of which can be freely embedded into other ob­jects. Template
instances can be hierarchically nested.
The MediaWiki platform is a strong open content and social networking
platform. Essential features are the support of the requirements of creative
commons licenses (perpetuating licences, tracking contributions and attri­buting
all authors of text and media), a version management and compari­son system
making changes in a large community transparent to the end user, and a layered
development system empowering the community to participate in the functional
development of the system.
The latter aspect helps to overcome the discrepancy between user needs and
developer actions. Traditional software development requires cycles of planning,
use-case and information modelling, piloting, imple­men­tation, testing, and rollout,
often resulting in slow and inflexible devel­opment. Although MediaWiki uses an
agile variant of this cycle (in­volving continuous code integration and live alpha
version testing), the php-based core code still suffers from slow development.
However, the domain of slow development has been minimized. An event driven
exten­sion system provides for an ecosystem of independently developed and
tested php-based extensions. Furthermore and highly relevant to the success
of Media­Wiki projects, the domains limited to developers and server owners are
supplemented by further layers (templates, CSS, and JavaScript) that are under
the control of the content-editing community:
The templating system enables authors to define and render their own data
storage and functional schemata. An unlimited number of templates can be
defined, and instances conforming to these schemata (typed and semantically
defined fields) can then be inserted in many content objects. Templates are central
to the ability of MediaWiki to empower the experts in a given knowledge domain
to experiment and achieve information sche­mata satisfying to their needs. For
example, KeyToNature defined sche­mata for media metadata and identification
keys. The functionality of tem­plates is limited to prevent detrimental influence
on the server, limiting possible malfunctions to those objects that include them.

14
As a negative point, the templating language has arisen as a unique ad-hoc
develop­ment, may be difficult to learn, and has no debugging support. Interest­
ingly, this may be a result of social engineering to limit the number of users
creating new templates on Wikipedia.
Further layers are the CSS and JavaScript integration. Like templates, these
layers are stored as normal MediaWiki objects, profiting from the version control
and comparison functionality. Since CSS and JavaScript involve potential
security concerns, editing of these layers is limited to content administrators.
The community focus of these layers was very positive in the KeyToNature
project and supported multiple more or less successful approaches to field
guides and identification tools.

3 The Biowikifarm
The virtual server is designed as a multi-project platform, enabling the joint
administration of a large number of separate wikis. Each wiki can be maintained
under its own domain name (owned by partners). Whereas the content
administration of each wiki is independent, significant synergies are created by
managing multiple MediaWikis on a single “wiki farm”.
Presently, the biowikifarm hosts the main KeyToNature portal, national
KeyToNature portals (pedagogical handbook, Offene Naturführer), the
International Society for Pest Information Wiki, LIAS glossary, Diversity
Workbench, and the Deutsche Phytomedizinische Gesellschaft Wiki.

4 Platform Customization Components

4.1 Media management

The biowikifarm maintains two local media repositories for sharing media
between all wikis on the platform. The “OpenMedia” repository is the pri­mary
repository for Creative Commons-licensed media. It is supplemented by a
“SpecialMedia” repository for media that cannot be openly licensed and are
available only under bilateral agreements.
Furthermore, the “Commons” repository with over 7 million images is di­rectly
integrated through a web service API. All items from Commons are directly
usable as if they were available locally. One problem initially en­countered
was that the Commons servers may occasionally drop web ser­vice requests if
overloaded. This could be solved by implementing a li­cense-compatible delayed
caching solution (every 10 min. in background).
MediaWiki guarantees the attribution requirements of most Creative Commons
licenses by linking media usage to a metadata page con­taining creators and
license information. This page also shows images in a higher resolution.
However, displaying this information forces the user to navigate away from
the present page. Our own usability studies have shown that users expect an
enlarged version of the image without leaving the page context and are confused
by the default functionality. A JavaScript based image zooming facility was

15
therefore added to biowikifarm. The first click on an image will enlarge it in an
over­lay to the page context, to the maximum extent supported by source image
and device resolution. The licensing requirements are fulfilled by presenting a
link to creator, copyright, and license information as part of this overlay.

4.2 Metadata and information management

Metadata stored in MediaWiki templates are supported by a customized

method. A MediaWiki extension harvests all first level templates (on Wiki pages
or inside text files uploaded as attachments) and stores the field-value pairs in
a MySQL database for fast access. A web service then pro­vides for queries or
recent changes, exposing the data as xml for down­stream processing (Fedora
Commons, GSearch).
By enabling Semantic MediaWiki (SMW), syntactically defined template
schemata are semantically annotated using standard ontologies (Dublin Core,
FOAF, SIOC). This allows direct semantic metadata search and inference as
well as exposure in the OWL/RDF format. Semantic que­ries can be embedded,
creating dynamical content in wiki pages (out­side of metadata, SMW is presently
further extensively tested by ISPI).
An embedded Flash application, the MediaIBIS search tool, searches the
metadata objects stored in the KeyToNature online repository. It has a user-
friendly multilingual interface and supports both simple and ad­vanced queries.
Details are presented in a separate contribution [5].

4.3 Embeddable identification tools

The platform provides several embeddable identification tools. DELTA datasets

can be embedded through NaviKey [6], SDD data through IBIS-ID [7] or Xper2
[8]. The embedding is achieved through a simple custom “IdentificationTool”
exten­sion. To embed a tool, users only need to write a simple statement like:
<IdentificationTool>tool=NaviKey5 config=… NavikeyConfig.xml</Identifi­ca­
tionTool>. DELTA and SDD data may be placed on wiki pages (rather than in
binary attachments) and remain directly editable in plain text mode.

4.4 Wiki-editable single-access identification tools

Single-access keys (i.e. tools with a fixed dicho-/polytomous structure) are

implemented in a more direct manner than the multi-access keys. They are based
on a combination of templates, CSS and JavaScript, all of which are directly
editable by administrators of the wiki (no intervention of server administrators is
necessary). On any given wiki page one or sev­eral single-access keys can be
freely embedded as a structural element in a rich-text layout. The keys provide
both a tabular, printable view and an inter­active (step-by-step) mode [9]. Details
about the wiki key imple­men­ta­tion are presented in this volume [10].

16
5 Sustainability and Scalability
Maximizing sustainability in the face of continuous hardware- and software
evolution was a major design priority. Hardware independence can be relatively
easily achieved by means of server virtualization, making entire servers easily
portable from one physical machine to another. Service is assured by follow-up
projects (until 2013) plus a longer-term maintenance pledge of the SNSB IT
Center (the SNSB is the government agency for the natural history collections
of Bavaria).
Software sustainability is more difficult to achieve. The model of iso­lated
systems maintained in stasis for long periods is not applicable to web software
that is dependent on a complex software environment and under permanent
threat of malicious attacks. Whereas major publishers achieve permanent
redevelopment for their in-house-developments, even mid-sized publishers and
software developers have often failed to find the necessary resources. Perhaps
the majority of internet offers in biodiversity that were backed by scientific
institutions or individuals have therefore ceased to exist. A possible solution is
built on three pillars: a) building on a carefully chosen open source software that
is supported by a large com­munity with a long-term perspective; b) minimizing
project-specific custom developments and partitioning them into small, well
documented modules (reducing complexity and the steepness of the learning
curve for new developers); c) building the platform to the needs of multiple
projects, aggregating available resources and achieving synergies.
We consider the long-term sustainability perspective of MediaWiki to be
optimal. It is actively developed, the content of the Wikimedia foundation projects
tied to the software makes it highly unlikely that it is abandoned in favour of
another project, and version upgrades are always fully automatic (in contrast to
some other content management systems that require con­siderable resources
to move from one version to another).
Our own developments are designed to be as modular and layered as possible.
They involve small php extensions, a set of templates that can be maintained
independent of newer developments, and CSS and Java­Script development.
Except for the php extensions, the components are di­rect­ly editable over the
web and can be maintained by a community of users and developers.
An attractive feature of the combination of templates and JavaScripts is
their locality to specific documents. The system offers the option to run older
identification tools in parallel with newer developments. While this may lessen
user experience uniformity, it reduces the analysis and testing re­quirements for
new ideas, enabling agile developments in the future.
Finally, and of great importance to scientific publishing, the principle of locality
also applies to content. Scientific knowledge is a stage in a deve­lopment, no
final truth. Opinion may often (yet) matter. Unlike typical data­bases, the platform
assumes no homogenous single truth. Dissenting opi­nion may be published
and outdated knowledge may be retained (adding pointers to updates, etc.).
Conventional databases may support dissent (e.  g., alternative taxonomic
hierarchies), but these expensive solutions are typically limited to a specific
aspect. On a wiki platform, any update requires no analysis whether it would

17
corrupt relational assumptions of older publications – contributing greatly to
scalability and sustainability.

6 Conclusion
The MediaWiki-based platform is suitable for the development of collaboratively
edited flora and fauna projects. It is powerful, extensible and long-term
sustainable. We have successfully implemented a set of native or embedded
components. Molecu­lar identification extensions are, however, yet missing. The
present platform can be adapted to other purposes in order to create an open
source online community of such tools and the scientific interests around them.
We welcome further partners to share the platform’s use and manage­ment.

Acknowledgement

This work was supported by the KeyToNature Project, ECP-2006-EDU-410019, in the

eContentplus Programme.

References
[1] MediaWiki software, http://www.mediawiki.org/wiki/MediaWiki, 2010-07.
[2] Wikimedia Foundation Projects: http://wikimediafoundation.org/wiki/Our_projects 2010-07.
[3] MediaWiki Templates, http://www.mediawiki.org/wiki/Templates, 2010-07.
[4] NoSQL databases (overview). http://nosql-database.org/, 2010-07.
[5] M. Giurgiu, A. Homodi, C. Veja, G. Hagedorn and P. L. Nimis, “A search tool for the digital
biodiversity resources of KeyToNature”. In: P. L. Nimis and R. Vignes Lebbe (eds.), Tools for
Identifying Biodiversity: Progress and Problems, pp. 19-24, 2010.
[6] D. Neubacher, and G. Rambold. NaviKey, a Java applet and application for accessing
descriptive data coded in DELTA format. http://www.navikey.net. 2005 (onwards), 2010-07.
[7] M. Giurgiu, G. Hagedorn, and A. Homodi, “IBIS-ID, an Adobe FLEX based identification tool
for SDD-encoded multi-access keys”. Proc. of TDWG 2009, 9-13 Nov. 2009, Montpellier, p.
90, 2009.
[8] V. Ung, G. Dubus, R. Zaragüeta-Bagils and R. Vignes Lebbe, “Xper2: introducing e-taxonomy”.
Bioinformatics, vol. 26 (5), pp. 703-704; see also http://lis-upmc.snv.jussieu.fr/lis/?q=en/
resources/software/xper2, 2010.
[9] S. Opitz and G. Hagedorn, “The jKey wiki key player and builder2. Proc. of TDWG 2009, 9-13
Nov. 2009, Montpellier, 2009.
[10] G. Hagedorn, B. Press, S. Hetzner, A. Plank, G. Weber, S. von Mering, S. Martellos and P.
L. Nimis, “A MediaWiki implementation of single-access keys”. In: P. L. Nimis and R. Vignes
Lebbe (eds.), Tools for Identifying Biodiversity: Progress and Problems, pp. 77-82, 2010.

18
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 19-24.
ISBN 978-88-8303-295-0. EUT, 2010.

A search tool for the digital

biodiversity resources
of KeyToNature
Mircea Giurgiu, Andrei Homodi, Cornelia Veja,
Gregor Hagedorn, Pier Luigi Nimis

Abstract — The European KeyToNature project has created a framework to

gather large amounts of biodiversity-related digital data and metadata from
cooperating providers and to make them publicly available via an advanced
search engine tool. The novelty of the solution consists in the creation
of specific queries for the search web service in the KeyToNature digital
repository, in the implementation of a communication protocol between the
client application and the repository server, and in other original solutions for
metadata presentation.

Index Terms — online digital repository, metadata search engine, Rich

Internet Application.

—————————— u ——————————

1 Introduction

K
eyToNature (www.keytonature.eu) is an EU-funded project focus­ing on
interactive educational tools for the identification of organ­isms. It aims at
enhancing the knowledge of biodiversity at all educa­tional levels across
Europe. Some project partners are data providers for an on­line repository for
metadata of media resources which can be used in the creation of interactive,
computer-aided identification keys. These digi­tal objects should become online
searchable and accessible. The solution was to create an online digital object
repository that stores only the meta­data associated with the digital resources.
The associated search tools, based on the assessment of user needs, are
described here.
The repository is now searchable via web services, it can interact with other
web services which support the management and access of the digital repository
[1]. The most important web service is GSearch, which indexes the Fedora
————————————————
M. Giurgiu, A. Homodi, C. Veja are with the Telecommunications Department, Technical University
of Cluj-Napoca, Cluj 400027, Romania, E-mail: [email protected].
G. Hagedorn is with the Julius Kühn-Institute, Fed. Research Centre for Cultivated Plants, Königin-
Luise-Str. 19, D-14195 Berlin, Germany, E-mail: [email protected].
P. L. Nimis is with the Department of Life Sciences, University of Trieste, I-34127, Italy, E-mail:
[email protected].

19
digital repository FOXML objects (Fedora Object eXtended Markup Language)
and of supports searching this index.

2 Data management and structure

A metadata exchange agreement has been created within KeyToNature, so
that all metadata in the repository have standardized semantics and support a
range of syntactical options for data transfer. The metadata for each resource
include information on, e. g., title, creators, keywords, taxa, geographic location,
copyright, license, format, access type (printed, off­line, online, free, login),
URIs, etc. [2]. The actual aggregation of these biodiversity metadata occurs
by a “metadata harvesting” process, implemented using specialized tools and
methods [3].

Fig. 1 – The communication between the search client and the digital biodiversity
repository of KeyToNature.

3 Application design
The chosen framework of Fedora Commons [1] and GSearch is a back­end
framework (Fig. 1). Although generic search interfaces exists, these are primarily
geared towards developers and totally unsuitable for end-users. Furthermore,
a goal of the project was to allow search endpoints in various web presences,
from the KeyToNature portal to various eLearning environments.
The search application was implemented in Adobe Flex [4, 5], which is
embeddable in a large variety of web environments. After defining and
implementing the communication between the Flex-based client and the Fedora-
based digital repository (Fig. 1), the most important step was the creation of the
user interface. The interface exposes the methods and mechanisms that the
search tool will use in order to transmit the user input (the request or query) to
the repository and to present the result for various types of users (beginner to
advanced).

20
3.1 Simple Search

The simple search interface (Fig. 2) consists of a single text input control and
additional drop-down menus. The menus allows users to add addi­tional search
criteria for narrowing or filtering results. Examples are selec­tions according to
the resource type they wish to find (images, identifica­tion keys, etc.) or according
to availability (online, free, printed-only, etc.).
The exhaustive search for organism names uses a thesaurus of syn­
onyms. This is a complex mechanism that helps users find more resour­ces
by extending their search criteria with added synonyms, scientific names and
common names. Despite the underlying complexity, the fea­ture is implemented
and communicated in the simplest possible way. When the results come back
from the repository, users are informed about the extra search terms that were
extracted from the thesaurus reply and used in the query.
The simple search interface automatically chooses the best display mode
(tabular or matrix image gallery) based on the resource type of the media
retrieved.

Fig. 2 – The main user interface for simple search.

3.2 Advanced Search

The advanced search interface (Fig.  3) allows users to interactively create

complex queries, including logical operators. The interface has three sec­tions:
1) Search Conditions: select the group of searched metadata fields; 2) Sorting
Mode: ascending/descending by multiple user-selectable fields; 3) Display
mode: a) gallery mode (Fig. 4) with metadata details accessible via the icons
situated on the upper right side of the thumbnail image (Fig. 5), or b) table mode.

21
 Fig. 3 – The user interface for advanced search.

Fig. 4 – The search results of digital objects displayed in the “gallery” view mode.

The advanced user interface is based on XML (eXtendable Markup Language)

storing the user’s selections and inputs [6], [7]. After a search has been
performed, the created query may be easily revised by the user, without having
to re-compose a query. When browsing through the dis­played results, users
have access to three options which are available for both simple and advanced
search modes: “New Query”, “Revise Query” and “Switch gallery/table mode”.

3.3 Parameters

The KeyToNature search tool currently has several external parameters by

means of which it can be preconfigured when embedded into a web interface:
1) language selection (nine languages are presently supported), 2) preset to
search identification keys only, 3) preset for searching freely available online

22
resources only, 5) preset for searching only online resour­ces which are under a
“Creative Commons” license.

Fig. 5 – Metadata details on a biodiversity-related digital object (in this case, a picture).

3.4 Multilingual interfaces

The KeyToNature metadata search engine is currently available in nine

languages (Bulgarian, Dutch, English, Estonian, German, Italian, Romani­an,
Slovenian, Spanish), based on two XML external configuration files, which are
easily customizable and extendable to additional languages.

3.5 Testing

The search tool has been carefully tested by project partners as well as experts
in software usability. Dedicated wiki [8] pages were created in the KeyToNature
portal for bug-reporting and suggestions. The reported prob­lems are fixed and
suggestions are being analyzed and implemented in an ongoing process.

4 Conclusion
The search tool presented in this paper was implemented as a client appli­
cation in Adobe Flex. It communicates via specific web services and protocols
with the KeyToNature online repository of biodiversity-related digital resources.
The selection of Adobe Flex has proved a successful decision for implementing
the user interface for search. It is an excellent tool for processing the large
amount of XML returned by the digital reposi­tory. The search application is
largely platform‑independent, due to the wide availability and distribution
of the Flash player. The KeyToNature search engine proved to be a robust,
fast application, which can be easily integrated into various portals. It could
be a model for implementing simi­lar applications interacting with online digital
repositories.

23
Acknowledgement

This work was supported by the KeyToNature Project, ECP-2006-EDU-410019, funded

in the frame of the eContentplus Programme.

References
[1] D. Davis and C. Wilper, “Fedora Commons Web Service Interfaces”, http://www.fedora-
commons.org/confluence/display/FCR30/Web+Service+Interfaces, July 2010.
[2] G. Hagedorn, P. L. Nimis, et al., “Resource Metadata Exchange Agreement”, http://www.
keytonature.eu/wiki/Metadata_agreement, July 2010.
[3] C. Veja, M. Giurgiu, G. Weber, and G. Hagedorn, “MediaWiki Interoperability Framework for
Multimedia Digital Resources”, Proc. Of Int. Conf. on Intelligent Computer Communication and
Processing, 26-28 August 2010, Cluj (Pending Publication).
[4] J. D. Herrington, and E. Kim, Getting started with Flex 3, O’Reilly Media Inc., 2008.
[5] C. E. Brown, The Essential Guide to Flex 3, Apress, 2008.
[6] E. R. Harold, and W. S. Means, XML in a nutshell, O’Reilly Media Inc., 2004.
[7] E. T. Ray, Learning XML, O’Reilly Media Inc., 2003.
[8] D. J. Barrett, MediaWiki, O’Reilly Media Inc., 2008.

24
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 25-30.
ISBN 978-88-8303-295-0. EUT, 2010.

Developing Web-based Search

Portals on Birds for Different
Target Groups
Renate Steinmann, Andreas Strasser, Andrea Mulrenin,
Amy Trayler, Sander Pieterse, Ivan Teage, Michael De Giovanni,
John J. Borg, Noel Zammit

Abstract — This paper presents the experiences and interim results from
the ongoing iterative development and testing of four distinctive search
portals on birds. The search portals are developed within the EU STERNA
project and address different target user groups. Based upon specific use
case scenarios the search portals are tested and validated in four specific
phases, applying three different testing methods: WAMMI online evaluation,
focus group evaluation and task-based usability tests. The paper introduces
the four search portals, depicts the testing methodology and presents the first
results from the ongoing user validation process.

Index Terms — digital library, web based search portals, iterative testing and
development.

—————————— u ——————————

1 Introduction

T
his paper presents the ongoing iterative development and testing of four
search portals on birds, each addressing a particular target user group.
The search portals have been developed as part of the EU funded
eContentplus project STERNA. STERNA is a best practice network and stands
for Semantic Web-based Thematic European Reference Network Application
(http://www.sterna-net.eu). STERNA comprises 13 organisations and research
institutes in the fields of natural history, wildlife and biodiversity. The project
started in June 2008 and will finish in November 2010.
————————————————
R. Steinmann, A. Strasser and A. Mulrenin are with the Salzburg Research Forschungsgesellschaft
mbH, Jakob-Haringer-Strasse 5/3, 5020 Salzburg, Austria. E-mail: (renate.steinmann, andreas.
strasser, andrea.mulrenin)@salzburgresearch.at.
S. Pieterse is with The Netherlands Centre for Biodiversity Naturalis, PO Box 9517 2300 RA Lei-
den. E-mail: [email protected].
A.Trayler is with Archipelagos, Institute of Marine Conservation, PO Box 1 Rahes 83301, Ikaria,
Greece. E-mail: [email protected].
I. Teage is with Wildscreen, Bristol BS14HJ, UK. E-Mail: [email protected].
J. J. Borg, M. De Giovanni and N. Zammit are with Heritage Malta, Valletta VLT03, Malta. E-mail:
(john.j.borg, michael.de-giovanni, noel.zammit)@gov.mt.

25
 Following the goals of the European Digital Library Initiative, STERNA
seeks to create a distributed and networked information space on nature and
wildlife. The main architecture of STERNA utilizes semantic web technologies
and standards which allow distributed querying of content based on metadata
represented in the RDF (Resource Description Framework) format as well as
reference structures represented in the SKOS (Simple Knowledge Organisation
System) format.
Related work to this paper includes a short paper depicting the user validation
process of STERNA which we have submitted to the Euromed 2010 Conference
in Cyprus.
The following chapters describe the four search portals (Chapter 2), the
ongoing user validation process and methodology (Chapter 3) and present
results that are available after finishing the first two phases of testing (Chapter
4).

2 Four STERNA Search Portals on Birds

Birds and bird-related information are the main focus of the STERNA information
space. STERNA addresses a wide range of target users that are interested in
birds, ranging from dedicated bird watchers to wildlife enthusiasts in general.
Against this background, we wanted to develop search portals that address
different user communities and that meet their specific requirements. To do so,
we decided to adopt an iterative design and development process, in which we
continuously develop, test and refine different search portals.
The search portals are based on four use case scenarios which were developed
by STERNA partners.
The use case of the Netherlands Centre for Biodiversity Naturalis (NCB
Naturalis) is targeted at bird watchers, a dedicated group of users that ranges from
the more casual to the professional bird watcher. Wildscreen/ARKive, a UK based
organisation promoting the appreciation of wildlife, has developed a use case
that targets a young and digitally savvy audience. The use case developed by
Archipelagos, a non-profit environmental organisation that focuses on the marine
and terrestrial biodiversity of the Aegean Sea, addresses boaters and tourists at
sea; the fourth use case was developed by Heritage Malta and is targeted at the
“humble rambler”, a more general audience that is interested in birds.
All four use cases were developed based on a common template and included
one or more user scenarios, i.e. typical situations in which users would make use
of the STERNA information space to search for bird-related information.
Based on these use cases we developed prototypes of four distinctive search
portals. Each search portal addresses a specific target group and offers particular
search functionalities: While the Wildscreen/ARKive and Archipelagos search
portals provide text based search functionalities, the NCB Naturalis search portal,
which is aimed at the more experienced bird watchers, allows users to search for
bird related content in different ways and to further refine search results (faceted
navigation). In addition to text based search the Heritage Malta search portal also
offers a silhouette driven search functionality to better identify and learn about
birds.

26
 Fig. 1 – Homepage of the Wildscreen/ARKive search portal for young, digitally savvy
users that was evaluated with WAMMI.

3 Iterative Development and Testing of STERNA Search Portals

After building prototypes of the four search portals we started iteratively testing
and improving them against our target users.
Testing and development in STERNA follows an iterative and user-centred
design approach for developing interactive systems. The major principles of
this approach are a strong focus on the actual end users—which means that
a system should suit the user rather than making users suit the system—and
iterative design—which means that the search portals are designed, tested
and modified repeatedly. Users are involved actively and early on in the design
process: this helps us to gain a deeper understanding of user problems and to
avoid taking wrong design paths [1].
The iterative testing and development process in STERNA encompasses four
phases: A first online WAMMI evaluation; focus group evaluation; task-based
usability tests; and a second online WAMMI evaluation. Due to limited available
resources, we apply the WAMMI evaluation only for the NCB Naturalis and the
Wildscreen/ARKive search portals. The more in-depth focus groups and task-
based usability tests are applied for all four search portals.
The focus of testing in all these phases is both on the usability of the user
interfaces as well as the presentation of content and search results.

3.1 WAMMI evaluation

WAMMI stands for Website Analysis and Measurement Inventory and is a web
based analysis tool for testing and measuring the user satisfaction of a website
or a web based solution. User satisfaction is measured in terms of attractiveness,
controllability, efficiency, helpfulness, learnability and the overall global usability.
WAMMI requests users to fill in an online questionnaire and to assess a web
site or solution. It then compares the user reactions with values generated from

27
a comprehensive reference database of other tested sites and solutions, thus
giving a better understanding of the quality of the tested solution(s).
We developed an online WAMMI questionnaire for evaluating the STERNA
search portals which includes the standard 20 WAMMI statements as well as
additional questions. It also invited users to comment on the ease-of-use of the
search portals and to provide suggestions for how to improve them (see: www.
wammi.com).

3.2 Focus Group Evaluation

The main objective of the STERNA focus groups is to identify shortcomings

and problems of the four search portals as well as develop ideas and suggestions
of how to improve them.
Focus groups are conducted with a small number of representative users.
A moderator steers the focus group discussion without discouraging the
participants from expressing their thoughts. With focus groups we can identify
opinions, attitudes and preferences from participants and gather insights that
are sparked by group interaction [1], [2].

3.3 Task-based Usability Tests

With usability tests representative users are requested to perform real-life

tasks and thereby to evaluate the usability of the search portals.
While performing the tasks users are requested to comment on any usability
problems they encounter (“thinking out loud”). Throughout the tests, users are
video-taped for documentation and later analysis (with the documentation kept
anonymous and confidential). After the test, the moderator and test users revisit
the video documentation of the test in the form of a semi-structured interview,
which allows the user to reflect on the test and to provide further feedback [2],
[3], [4].

4 First Results of User Testing and Development

4.1 Status of user testing and development

User validation of the Wildscreen/ARKive and the NCB Naturalis search portal
started in October and November 2009 respectively with the first round of WAMMI
evaluation. Testing continued until early February 2010 when we received WAMMI
evaluation reports and content analyses of the user comments provided.
Based on the findings from the WAMMI evaluation, both search portals were
improved and then—together with the search portals from Heritage Malta and
Archipelagos—tested in focus group evaluations. These took place in June/July
2010. We are currently integrating feedback from focus group evaluations to
further improve our search portals. In late July/August, we will conduct the task-
based usability tests, to be followed by the second round of WAMMI testing in
September/October 2010.

28
4.2 Findings from WAMMI 1 and focus group evaluations

64 users filled in the online WAMMI questionnaire for evaluating the NCB
Naturalis search portal, and 94 users filled in the questionnaire on the Wildscreen/
ARKive search portal. Both search portals were rated below average in relation
to the WAMMI reference database (i.e. the web sites and solutions that were
previously tested), with the Wildscreen/ARKive search portal being rated
considerably better than the NCB Naturalis search portal.
The NCB Naturalis search portal received a mean global usability score (GUS)
of 21.8, the Wildscreen/ARKive search portal (targeting a young, digitally savvy
audience) a mean GUS of 41.4 (on a scale from 1 to 100, where one is lowest
and 100 highest; 50 represents the average of the reference database of tested
web sites and solutions). For both search portals we received a considerable
amount of positive user feedback, which helped us in specifying the main
usability problems of the search portals, as well as providing us with valuable
suggestions of how to improve them.
The four focus group evaluations conducted in June/July 2010 helped
us in further specifying problems of our search portals. After discussing and
identifying the main problems of the four search portals, participants provided
us with concrete ideas and suggestions for how to tackle these problems and
improve the search portals further.
While user feedback from the WAMMI and focus group evaluations was
distinctive for each search portal assessed, it also showed us some common
problems of our search portals that we need to address.
The visual interface design of the search portals was often not regarded as
very attractive and also the search results presented should generally be more
visual. Users often remarked that they would like to get more images, video
or audio recordings while they were, usually, less interested in metadata lists.
Users also noted that the search functionalities and filter mechanisms need to
be improved in order to deliver more fitting results to target users. Navigating
through the search portal could also be difficult at times, and some users also
remarked that they were unsure about the purpose of the search portals.
The search portals thus have to be more intuitive and visually appealing,
deliver more fitting results, and their meaning and functionality need to be more
apparent for users.

5 Conclusions
The iterative design and testing approach that we applied has helped us in
identifying usability problems of our search portals early on in the development
process and hence to make the design and development process as resource
and cost effective as possible. It has also helped us to better meet the needs
and requirements of our respective target user groups. With the next two phases
of user testing we expect to further improve our search portals and make them
more user-friendly (however, since they are developed as part of a best practice
network project, the final search portals will not be “market-ready products”, but
advanced prototypes).

29
References
[1] K. Baxter and C. Courage, Understanding Your Users: A Practical Guide to User Requirements.
Methods, Tools, and Techniques. Amsterdam, Boston, London, New York, Morgan Kaufman
Publishers, 2005.
[2] D. Chisnell and J. Rubin, Handbook of Usability Testing: How to Plan, Design, and Conduct
Effective Tests. 2nd ed. Indianapolis, Wiley Publishing, Inc., 2008.
[3] B. Albert and T. Tullis, Measuring the User Experience: Collecting, Analyzing and Presenting
Usability Metrics. Amsterdam, Boston, London, New York, Morgan Kaufman Publishers, 2008.
[4] J. S. Dumas and J. C. Redish, A Practical Guide to Usability Testing. 2nd ed. Exeter, Portland,
Intellect Ltd., 1999.

30
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 31-36.
ISBN 978-88-8303-295-0. EUT, 2010.

Simple Identification Tools

in FishBase
Nicolas Bailly, Rodolfo Reyes Jr., Rachel Atanacio, Rainer Froese

Abstract — Simple identification tools for fish species were included in the
FishBase information system from its inception. Early tools made use of
the relational model and characters like fin ray meristics. Soon pictures and
drawings were added as a further help, similar to a field guide. Later came
the computerization of existing dichotomous keys, again in combination with
pictures and other information, and the ability to restrict possible species
by country, area, or taxonomic group. Today, www.FishBase.org offers four
different ways to identify species. This paper describes these tools with their
advantages and disadvantages, and suggests various options for further
development. It explores the possibility of a holistic and integrated computer-
aided strategy.

Index Terms — databases, identification strategy, simple identification tools,

web integration.

—————————— u ——————————

1 Introduction

F
ishBase [1] is a Global Species Database (GSD) and a Biodiversity
Information System (BIS) on all extant fish species of the world, with about
32,000 valid species currently recognized as valid. It contains a wide range
of information and data on biology, ecology, chorology, taxonomy, physiology,
human uses, illustrations, etc. and aims at being the web global encyclopaedia
on fishes. Four different types of identification tools are made available on the
FishBase website (www.fishbase.org), which opens directly the search page:
• ‘Eye-balling’ drawings and key features by decreasing taxonomic level
from class downward;
• Display of all pictures available for a given geographic area or a given
family with possible restriction on fin ray meristics;
• Classic dichotomous keys; and
• Uses of simple morphometric ratios.
This paper gives a short description of the tools, how they work and can be
used, where the users can find them, their advantages, disadvantages and
————————————————
N. Bailly, R. Reyes Jr. and R. Atanacio are with the Aquatic Biodiversity Informatics Office, The
WorldFish Center and the FishBase Information Research Group, Inc., Los Baños, 4031 Laguna,
Philippines. E-mail: [email protected].
R. Froese is with the Leibniz Institute of Marine Sciences, IFM-GEOMAR, West Shore Campus,
Düsternbrooker Weg 20, D-24105 Kiel, Germany. E-mail: [email protected].

31
limitations, and their possible improvements in the next future. Tools not yet in
FishBase are also discussed. The need to integrate such tools in a common
identification strategy is stressed.

2 Tools

2.1 Quick Identification Tool with outlines

Description, how to use it, and where to find it. - The user is given the choice
between boxes displaying simple outline drawings of species representing a
given group, together with a short account of key characters. The user clicks on
the box, and the boxes for the next level are displayed. For fishes, we start with
6 classes from the most likely ray-finned fishes to the less likely lampreys and
hagfishes. Each class leads to its orders, and then each order to its families.
Clicking on a family box leads to the “Identification through Pictures” pages
described in the next section. It is possible to restrict the identification process
to a large geographic area or to a country. This tool is in the top menu in the
search page.
Advantages, disadvantages and limitations.  -  The main strength of the
tool is that it is visual. The user is not obliged to read textual accounts. It is
quite simple and useful to search for the family. It may be tricky to identify, e.g.,
eel-like shape groups that can be found in several classes and orders; here,
reading the accounts may help. The number of typical species depicted in a box
is limited in large groups, and sometimes well-known outlines but rare shapes
are not included, such as coelacanths for the lobe-finned fishes box.
Further developments.  -  Below the family level, the outline drawings are
generally not that useful, because species often have the same shape in a
given family. There is a possible improvement using the subfamilies for the most
species-rich families (Cyprinidae, Characidae, …).

2.2 Identification through pictures

Description, how to use it, and where to find it. - The principle is to display
one typical picture per species by area or taxonomic group. Clicking on a picture
opens the corresponding species account with more information.
The tool is accessible from three sections in the search page as “Identification
by pictures” under “Information by Family”, “Information by Country / Island”,
and “Information by Ecosystem”. For the first section, species within a family
are displayed by alphabetical order of scientific names, which shows closely
related species next to each other. From that page, it is possible to restrict the
search by large area, and/or by the number of dorsal and anal spines. The
broad distribution and maximum length of the species are listed as additional
help. For the two last sections, the typical pictures of all species reported in
the given geographic area are displayed using a traditional sorting of the fish
classification, from hagfishes to coelacanths.
Advantages, disadvantages and limitations.  -  The main strength of this

32
tool is that it is visual. The user is not obliged to read texts. It is quite simple
and useful to search within a family up to 100 species. But identification with
pictures only may be difficult or, worse, misleading. As usual, users should read
the species account to verify their identification, which is possible by clicking
the picture. For groups with over 100 species the tool becomes problematic.
Beyond 500 species, the answer time of the server may be prohibitive. Also,
only half of the species have at least one usable illustration for identification in
FishBase.
Further developments. - A major FishBase goal is to get at least one picture
for every species. This is not easily achieved, because many species are only
known from a few museum specimens, if any. The tool could be extended to
subfamily and genus ranks for the most specie-rich families, and to orders and
classes for those with a few species only. For the pages with geographic areas,
there should be a taxonomic table of contents/menu at the top of the page,
allowing the user to jump directly to the class, order or family (maybe genus)
when he knows it.

2.3 Computer-aided dichotomous keys

Description, how to use it, and where to find it. - These are dichotomous
keys digitized as they were published in FAO catalogues, revisions, field guides,
and major ichthyofaunas. We have developed our own simple database and
webpage format for these keys, but we also use LucId Phoenix for an enhanced
interface (see [2] for a description). For the FishBase format, couplet numbers,
character text, and number of the next couplet, number of the previous couplet,
illustration of the character or of the species (+name) stand in 5 columns in a
row of an html table (it corresponds more or less to the database table format).
Clicking on the number of the next couplet leads to the corresponding row in
the table, clicking on the species name leads to the species account. The tool is
accessible from the search page, and from the species summary pages under
the section “Tools / Identification keys”. Coming from the search page, it is
possible to restrict the key selection by large geographic area, Order, or Family,
or to enter the key id when this is known. It is possible to specify whether the list
should show only keys available with the LucId interface.
Advantages, disadvantages and limitations. - Advantages are the inclusion
of pictures in keys that had none, and the ability to easily step forward and
backward during the identification process. Also, the species account is only
one click away when reaching a possible identification, so that it can be verified
with additional information.
Further developments. - The use of the LucId interface is a good development
of our simple format. More species pictures and character illustrations are
needed. One internal improvement is to be able to give at the same time the
name as used in the publication and the current accepted name, highlighting
when the name in the publication is now considered as a synonym.

33
2.4 Identification by Morphometrics Tool

Description, how to use it, and where to find it. - This tool uses measurements
that are easy to obtain from specimens or pictures, and computed standard
ratios. The user needs to take measurements of Total Length (TL), Head Length
(HL), Eye Diameter (ED) and Body Depth (BD). The tool accepts measurements
made in centimeters or inches. In the case of pictures, measurements in pixels
are accepted. Ratios of the head length, eye diameter and body depth to
the total length are computed and compared with values stored in FishBase.
Providing the FAO Area, from where the specimen was collected, the Class, and
the Family (optional) significantly reduces the number or possible species. In
addition, the Total Length (TL), by eliminating species that do not grow as big or
larger than the unknown specimen, further focuses the search to a few species
in many cases. The tool returns a list of possible species. For each species, a
short description, fin counts, a picture and a link to the species summary page
in FishBase is included. The “Identification by Morphometrics” can be found in
the “Tools” section of the search page.
Advantages, disadvantages and limitations. - This tool can be used when
the user does really not know what he has in hands. As far as it was possible,
commercial species were covered first. However, such measurements are
available “only” for a third of species, limited by the number of suitable pictures
from which the reference measurements are taken. It must be also understood
that the standard ratios are usually computed on one picture only, so there is no
statistical range: the possible range during the matching is predefined.
Further developments.  -  An advanced interface which includes other
measurements such as preorbital length, predorsal length and preanal length
is under development. It is expected that including these measurements will
shorten the list of possible species most often to 10 species or less.

3 Discussion
Missing tools: interactive keys, image analysis, Barcode. - Since the late
1960s, computer-aided identification softwares have been developed, using
a matrix of taxa / character states as the basis of the tool (Delta [3], XPER
[4], LucId, etc.). In FishBase, we store data on morphological description of
species, but unfortunately in a format that is primarily incompatible with these
softwares. A first promising attempt was made to transform the data in a correct
format to be included in XPER, which opens the way to the suitable compliance
with the TDWG SDD standard (Structured Descriptive Data [5]). But we face
the usual content issue that is recurrent for large groups: how to describe a
seahorse, a tuna, a hagfish, and a turbot with the same characters? So the
limitations are now on the structure and the standardization of descriptions
more than a technological issue. Image analysis is another tool that we did
not implement. After some tries with both public and commercial products, the
result did not seem efficient enough in order to be incorporated in FishBase:
at the present stage of these tools, it could help to restrict possibilities only as
long as our reference picture collection is more complete than currently; also,

34
these reference pictures must be prepared “manually” to remove noise (e.g.,
focusing on the individual and not the background) if we want to increase the
true positive matching. The other issue is to find solutions to use at the same
time live underwater pictures with different orientations of the individual and
dead specimens with different colours, not to speak about growth allometries,
sexual dimorphism, and intermediary individuals when their sex changes during
their life cycle like in some families. A new approach is the identification offered
by the Barcode of Life (BoLD website [6]), if the user has access to a respective
test kit for the genetic analyses. It is not clear whether this identification tool
can be more integrated in other webpages like FishBase (the two websites are
already cross-linked), or if it is best to use the tool from the BoLD website.
Identification strategy. - At the moment, identification tools in FishBase are
independent, except that the Quick Identification Tool links in the end with the
Identification Through Pictures Tool beyond the family level. The idea is that all
the tools should be integrated into one, and that the user could choose to jump
from one to another anytime during an identification session, or even better, that
the system could guide the user in that choice, according to the declared skill
level of the user. Interactive keys are obviously the start of such developments.
XPER proposes already some modules that can build the identification pathway
according to some constraints (e.g., sort out species first that are the most
abundant, the easiest to identify, or have the most striking forms): someone
has to enter the relevant data. But we are in need of both the design and
the technology to jump from outline drawings to identification key and to real
pictures, from morphometrics/meristics to image analysis, restrict to a size or a
geographical area. The final vision is that the system would guide the user and
suggest how to start and which pathway to use across all tools.

4 Conclusion
FishBase has deliberately favoured the simplicity over more elaborated
identification tools that are costly to develop and maintain, including in terms of
data. Some of these simple tools are really easy to deploy on the web such as the
Quick Identification Tool with simple outline drawings, and the computerization
of the printed dichotomous keys under a simple database format and web
layout. Colleagues could move forward quickly to these simple solutions for
other taxa. However, each of the tools existing or not in FishBase is interesting
in a given context. A long-term goal for the computer-aided identification domain
could be to gather all tools in a unique strategy and interface for the user. But
this still requires research and technological development, such as the work
being done under the European project KeyToNature [7]. The last important
point is that illustrations, including correctly identified pictures, must be made
publicly available. Images are used in 3 of our 4 tools, and we could design
the morphometrics tool only because we had a significant number of pictures
available. Homo sapiens is a species that uses the visual sense to a high degree,
and visual identification is still and may remain for a long time its preferred
method.

35
Acknowledgement

The various tools in FishBase were developed in the last 20 years during various projects
mainly funded by the European Commission. Eli Agbayani, Josephine Barile, Elijah
Laxamana, Christian Elloran and Stacy Militante were the successive programmers who
developed them.

References
[1] R. Froese and D. Pauly (eds.), FishBase 2000: Concepts, design and data sources. Los
Baños, Philippines, ICLARM, xvii+344 pp., 2000.
[2] LucId Central, “LucId Phoenix”,
http://www.lucidcentral.org/Software/LucidPhoenix/tabid/152/Default.aspx, 2010.
[3] M. J. Dallwitz, “Overview of the DELTA System”,
http://delta-intkey.com/www/overview.htm, 2009.
[4] J. Lebbe and R. Vignes Lebbe, “Xper2”.
http://lis-upmc.snv.jussieu.fr/lis/?q=ressources/logiciels/xper2. 2010.
[5] G. Hagedorn, K. Thiele, R. Morris and P. B. Heidorn, The Structured Descriptive Data (SDD)
w3c-xml-schema, version 1.0. http://www.tdwg.org/standards/116/. [Last retrieved 05-May-
2007], 2005.
[6] BoLD, “Barcode of Life Data Systems”
http://www.boldsystems.org/views/login.php, 2010.
[7] KeyToNature, “KeyToNature: a new e-way to discover biodiversity”, http://www.keytonature.
eu/wiki/, 2010.

36
Nimis P. L., Vignes Lebbe R. (eds.)
ù Biodiversity: Progress and Problems – pp. 37-42.
ISBN 978-88-8303-295-0. EUT, 2010.

The Catalogue of Life: towards

an integrative taxonomic
backbone for biodiversity
Frank A. Bisby, Yuri R. Roskov

Abstract — The Catalogue of Life Programme is addresseing the need for

a comprehensive catalogue of the world’s presently known animals, plants,
fungi and micro-organisms. The need is for an electronic catalogue that can
be used as a taxonomic back-bone. In a wide variety of programmes covering
species and documenting many types of biotic materials and records. The First
Phase of the programme has used an architecture based on an array of global
species databases to reach coverage of about two-thirds of known species.
In the Second Phase of the Programme there will be a new architecture, a
new array of services, and a ring of partnerships with global programmes.

Index Terms — Catalogue of Life, global taxonomic framework, species

checklist, synonymic indexing, taxonomic hierarchy.
—————————— u ——————————

1 Introduction

D
espite 250 years of effort in the taxonomic profession, there is still, in
2010, no complete catalogue of all presently known animals, plants,
fungi and micro-organisms of the world. This is a critical problem for
the scientific community, and for national, regional and global organisations
that organise and regulate the exchange of biotic information and materials
worldwide. The set of organisms known to science is a key dimension of human
knowledge concerning global biodiversity, evolution, ecology, natural resources,
and biotic response to climate change. It supplies a vital set of index terms
needed to access most biodiversity knowledge. There is increasing public need
and expectation, focussed through the UN Convention on Biological Diversity
(CBD), to complete such a catalogue of all known organisms for international
uses. Many commentators are surprised that a complete catalogue does not
already exist. In fact it is a non-trivial task that is too large for the individual
————————————————
F. A. Bisby is with the Species 2000 Secretariat, Centre for Plant Diversity & Systematics, School
of Biological Sciences, University of Reading, READING, RG6 6AS, UK.
E-mail: [email protected].
Y. R. Roskov is with the Species 2000 Secretariat, Centre for Plant Diversity & Systematics,
School of Biological Sciences, University of Reading, READING, RG6 6AS, UK. E-mail: y.roskov@
reading.ac.uk.

37
capabilities of even the largest taxonomic institutions, due to the distributed
nature of the knowledge.
The Species 2000 programme, working in partnership with ITIS in N. America,
has made substantial progress with resolving this problem. It has created,
maintained and enlarged its Catalogue of Life to the point where it now covers
1.25 million species of plants, animals, fungi and micro-organisms, more
than two-thirds of the anticipated total of 1.9 million presently known species
worldwide. It has done this by employing a radical architecture of federating
global sectors of taxonomic expert knowledge from a growing array of supplier
databases, and integrating these into a single taxonomic hierarchy and species
checklist. The distributed system harvests taxonomic knowledge provided
and maintained by a community of supplier organisations in the taxonomic
profession, combining work by the major taxonomic institutions with that of
smaller networks and individuals. This process was brought to production scale
by the EC EuroCat project funded as a scientific infrastructure under FP 5 (2003
– 2006) and further developed since then with funding from other sources.
Over the last two years the programme has concentrated on extending
and improving the scientific content of the Catalogue of Life, which is now a
unique and scientifically valuable resource. However, it has come as a bonus
to see the rising and now substantial public usage in Europe and all over the
world, including by GBIF and the Encyclopedia of Life, of what is presently an
incomplete service. The 4D4Life Project provides us with a timely opportunity to
develop a parallel focus on services. It will enable us to enrich the variety and
technical sophistication of taxonomic services that are undoubtedly possible,
exploiting the taxonomic resource that we are already building. The utility of
these services will secure the sustainability of the whole programme into the
future.

2 The present concept

The Species 2000 & ITIS Catalogue of Life (henceforward ‘the Catalogue’)
has a single purpose, to enable users throughout biological and biodiversity
sciences, and across the many scientific and non-scientific disciplines that use
organism information, to access data about all organisms by means of a species
checklist and a taxonomic hierarchy. It is already used to access data such as
organism relationships, ecology, DNA sequences, protection status, invasive
properties or information in any one of a myriad of other data domains. Such a
Catalogue needs to be:
i) comprehensive: covering all known organisms in all groups;
ii) global: organisms of the whole world, in terrestrial, freshwater and marine
environments;
iii) validated: a responsible, modern and professional globalised taxonomic
view of the classification, supported by and embedded in the profession’s
activities;
iv) accurate: reflecting as accurately as is practical the detail of diversity of
living organisms;
v) accessible to all: a clear view of the taxonomy, eventually in multi-lingual

38
presentation;
vi) available to all: widely and freely available in a variety of forms; and
vii) dynamic: updated for taxonomic changes though time, either continuously
or annually.
To be effective in the many applications in which it is used, the classification
and the naming of species and higher taxa must be as close to ‘agreed and
correct’ as is possible in taxonomy. This means for each taxon either using
a consensus system, or selecting by peer review and using consistently one
of the competing classifications where alternatives are in wide use. Because
alternative classifications have been used both today and in the past, users
must be able to locate species known by other names (or concepts) in the
Catalogue, and discover alternative names under which to access data on the
internet or in other resources. Consequently synonymy and common names
must be included for each species. As much as possible should be ‘concept-
based’, a precision provided by some of the supplier databases.
The dream is simple - to create a Catalogue that contains an accurately
maintained synonymic species checklist covering all known groups, connected
in a validated taxonomic hierarchy.

3 The existing (Phase 1) Programme

The present Catalogue of Life Programme, led by the global Species 2000
organisation based in Reading, and working with the N. American organisation
ITIS, was set up as an international programme at a UK-funded (BBSRC)
workshop in 2001. Bringing the programme up to production scale was funded
by the EC as one of its scientific infrastructures (EuroCat), with further funding
by the Japanese Government, the US Government (through ITIS) and GBIF.
Output is via the Catalogue of Life Annual Checklists on the web [1], and on free
DVD [2], and the Dynamic Checklist on the web, both also available as web-
services for electronic use.
In March 2007 an EC-funded ‘Million Species Day’ symposium was held
to celebrate reaching one million species. The 2010 Annual Checklist now
provides a quality species checklist of 1,257,735 species with unique identifiers
and a hierarchy for all organisms (animals, plants, fungi, chromista, protozoa,
bacteria, archaea, viruses). The estimate for the number of known extant
species is currently 1.9 million [3].
The present Catalogue benefits from simplicity of structure incorporating
minimal but standardised data for each species. These contribute to its success
in providing a universal baseline needed by all biologists, and in making the
project practicable. It consists of two knowledge structures, and software that
enables the user to search or traverse them, and to toggle between them. i)
The Species Checklist is a series of Species pages that are located by name
searches, with automated synonymic indexing. Each page gives the Standard
data for a Species, including common names, the higher taxa it belongs to in
the hierarchy, and geographical distribution. ii) The Taxonomic Hierarchy is an
expansible tree that can be followed down through the classification to the 1.25
million individual species. Or it can be used to navigate upwards to the higher

39
taxa containing the one that is viewed. By clicking on a higher taxon listed on
a species page, the user can transfer to the tree for that taxon, and see all its
daughters. Conversely, by clicking on a species at a twig in the tree, the user
can visit the relevant Species page in the Checklist.
A comprehensive checklist cannot be made simply by adding together regional
or single-country lists. Different classification and naming schemes mean that a
simple additive list would be massively duplicative and of little use. The current
system is a successful development of the original BBSRC SPICE project. It
federates the taxonomic sector checklists provided by a distributed array of
global species databases (GSDs), which are globalised checklists of a whole
taxon, harvested across the Internet, and fitted together ‘end-to-end’ within a
single overall classification. When enough sectors are fitted this process can
eventually create a complete list. The number of GSDs contributing one or more
taxonomic sectors to the Catalogue reached 77 for the 2010 edition, including
47 based in Europe, 18 in the USA, 5 in Brazil, and 7 in New Zealand, Russia,
Japan, Taiwan, Australia and the Philippines. The model ensures that sectors
are enhanced taxonomically by the supplier databases, and ca. 3,000 experts
globally contribute to these databases. The whole programme depends on the
integration and aggregation of expert knowledge from these key suppliers.
Each GSD sector is attached at its ‘top point’ (its highest ranking taxon) in the
hierarchy, and in addition to harvesting the checklist, the system also harvests
branches of the tree beneath this top point for the hierarchy leading down to the
species in that sector of checklist. The checklist and hierarchy created from a
growing array of GSDs in this present-day architecture (‘Architecture 1’) referred
to as the ‘Global Hub’.
Despite the evident success of Architecture 1 in permitting the rapid build up
of the Catalogue to its present point, its limitations have been evident for some
time. The difficulty is simply that no-one anywhere in the world is creating global
species databases for some of the least known taxonomic groups, so by this
model these would be destined always to remain as gaps in the Catalogue. In
the EC EuroCat project (2003 – 2006), we additionally experimented with making
a Regional or ‘Euro-hub’ with a further set of European regional databases,
and versions of SPICE that could handle multiple hubs, and the first steps
towards integrating their contents using the LITCHI 2 taxonomically intelligent
integrity tracking. We then started to plan an ‘Architecture 2’, in which an array
of Regional Hubs might be connected to the Global Hub, this providing linkage
to regional databases from many parts of the world, but also the potential for
the Global Hub to harvest data or checklist sectors from Regional treatments
for the species groups that were missing from the Global Hub. Good progress
is being made with initiating these Regional hubs now, and plans in the 4D4Life
Project are to develop a unified concept and specification for this Multi-Hub
Network working with the designated centres for China, New Zealand, Brazil,
and Australia.

40
4 The Phase 2 Programme
In June 2009 Species 2000 and ITIS launched the Phase 2 programme of
the Catalogue of Life with a fresh funding initiative and extended partnerships
around the world planned for the 5-year period 2009 – 2014. In outline Phase
2 involves:
1. A new array of electronic and other services
2. A new service-based cyber-infrastructure: an ecosystem of services
3. A strategy for completing taxonomic coverage of the Catalogue
4. A world-wide multi-hub network of regional hubs
5. A 2nd Edition Catalogue of Life Management Hierarchy
6. A ring of partnerships with global biodiversity programmes

5 4D4LIFE Project
The 4D4Life Project in the EC e-Infrastructure programme has now taken
responsibility for the array of new services, the new cyber-infrastructure, and
designing the world-wide multi-hub network. Sara Oldfield at Botanic Gardens
Conservation International is co-ordinating the Services Team, and Alex Hardisty
at Cardiff University is co-ordinating the System Design Team.

6 I4LIFE Project
The i4Life Project in the EC e-Infrastructures Programme will shortly take
responsibility for the ring of partnerships with global biodiversity programmes
intended to harmonise and integrate between the taxonomic catalogues.

7 Conclusion
Substantial progress has been made with developing a comprehensive
Catalogue of Life. However, there remains much to be done in the ambitious
Species 2000 & ITIS Catalogue of Life Programme. The Catalogue is still far
from complete in terms of taxonomic groups and known species; there is much
to be done in improving both quality and fill of the Standard data set across all
taxa; the new public services need to be fully tested and rolled out, and the
programme needs to make progress with becoming sustainable as a scientific
infrastructure for use around the world.

Acknowledgement

This work was supported in part by the EC DG INFSO FP7 e-Infrastructures Programme
under the 4D4Life Project (Grant 238988).

41
References
[1] F. A. Bisby, Y. R. Roskov, T. M. Orrell, D. Nicolson, L. E. Paglinawan, N. Bailly, P. M. Kirk, T.
Bourgoin and G. Baillargeon, Species 2000 & ITIS Catalogue of Life: 2010 Annual Checklist,
www.catalogueoflife/annual-checklist/2010, Species 2000, Reading, 2010.
[2] F. A. Bisby, Y. R. Roskov, T. M. Orrell, D. Nicolson, L. E. Paglinawan, N. Bailly, P. M. Kirk, T.
Bourgoin and G. Baillargeon, Species 2000 & ITIS Catalogue of Life: 2010 Annual Checklist,
DVD, Species 2000, Reading, 2010.
[3] A. D. Chapman, Numbers of Living Species in Australia and the World, 2nd Edition. Australian
Biological Resources Study, Australian Government, Canberra, 2009.

42
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 43-48.
ISBN 978-88-8303-295-0. EUT, 2010.

BHL-EUROPE: Biodiversity
Heritage Library for Europe
Jana Hoffmann, Henning Scholz

Abstract — The Biodiversity Heritage Library for Europe (BHL-Europe) is

an EU funded project making available biodiversity literature over various
platforms, e.g. a multilingual portal for search and retrieval, the Global
References Index to Biodiversity (GRIB), and Europeana. BHL-Europe brings
together European digital biodiversity content already available and will assist
in future scanning activities. BHL-Europe is asking the scientific community
to contribute to the improvement of the BHL-Europe portal functionalities by
giving feedback and participating in BHL-Europe workshops or to provide
information about unexplored repositories of digital biodiversity content.

Index Terms — Biodiversity Heritage Library for Europe (BHL-Europe),

digital library, biodiversity literature, taxonomy, taxonomic intelligence, optical
character recognition (OCR).

—————————— u ——————————

1 Introduction

T
he lack of access to the published biodiversity literature is still a challenge
in the day-to-day business of taxonomists or researchers dealing with
biodiversity-related questions. In the past, only libraries of large and
renowned institutions such as universities, natural history museums or botanical
gardens housed specific literature indispensable for taxonomic work. Collecting
relevant literature on a certain group of organisms was time consuming, cost-
intensive and required loans of books or even a visit to the respective institution.
Today, quick and easy access to digital literature is more and more important
to facilitate scientific work. However, digitisation of literature is expensive and
requires a lot of additional work on making the content available for extensive
search and retrieval. Furthermore, there are still major problems with right
holders, thus limiting the range of content freely available on the internet. For
scientists it is of high importance to have a sustainable infrastructure they
can rely on with a simple and quick mechanism to search for bibliographic
information and free access to digital content of high quality. This is especially
true for scientists working in developing countries with limited access to literature
in general. As taxonomy is an ‘accumulative’ science it relies more than other
————————————————
The authors are with the Museum für Naturkunde, Leibniz Institute for Research on Evolution and
Biodiversity at the Humboldt University Berlin, Invalidenstr. 43, 10115 Berlin, Germany. E-mail:
[email protected], [email protected].

43
disciplines on a complete record of literature on a group of organisms of interest
and has a stronger focus on historic publications. Moreover global availability
of digital content of biodiversity literature is also important for training students
and early career scientists and helps promoting the importance of taxonomy as
a discipline. Additionally, enhancement of availability of biodiversity literature
for a wider public raises awareness of the importance of protecting our planets
biodiversity.
Over the last few years a large number of library resources for taxonomists
have been made available online - including virtual libraries and search engines
as well as digital libraries. Since 2007, numerous libraries in the UK and USA
are digitising their holdings of biodiversity literature and making them available
on the internet. Today, BHL - the Biodiversity Heritage Library (http://www.
biodiversitylibrary.org) is certainly the largest digital library for taxonomists
offering free access to more than 30 million pages of historical biodiversity
literature (as of July 2010) via the internet originating from 12 major natural
history museum libraries.

2 Overview of BHL-Europe
In 2009 the European Commission launched a new project ‘BHL-Europe -
Biodiversity Heritage Library for Europe’ (http://www.bhl-europe.eu) within the
framework of the eContentplus program. This project will run for 36 months through
April 2012. The BHL-Europe consortium consists of 28 partner institutions (natural
history museums, botanical gardens, libraries, right holders and companies)
including 26 European institutions and two American institutions representing BHL
(US). BHL-Europe aims among others at (1) supporting existing digitisation initiatives
with best practice guides, for example, (2) facilitate and enable the initiation of new
scanning initiatives, and (3) bringing together existing digital content scattered all
over Europe in a number of libraries and natural history institutions. Currently, 18
out of the 28 consortium partners of BHL-Europe are active contributors to the
corpus of digital resources. This corpus of more than 100,000 monographs and
serial volumes in April 2012 will eventually be available on three platforms (Fig. 1)
(1) a multilingual BHL-Europe-Portal for search and retrieval for scientists and public
users, (2) the Global References Index to Biodiversity (GRIB), and (3) Europeana.
The technical architecture of BHL-Europe is based on the Open Archival
Information System (OAIS) reference model. It is the backend of the multilingual
portal for managing content ingestion, archival and delivery of the digital objects
(Fig. 2). A prototype of the new portal will be available in fall 2010, but the final
system is expected for the end of the project in April 2012.
BHL-Europe and EDIT (http://wp5.e-taxonomy.eu/) are building the Global
References Index to Biodiversity (GRIB), a database generated from the partner
libraries catalogues and completed with content management and deduplication
functionalities, that eventually refers to all of the worlds published biodiversity
literature. This will enhance the possibilities of search and retrieval of digital
literature for taxonomists significantly. It will also assist librarians in the process
of scanning planning. A GRIB prototype is working already and the final system is
expected to be finished in spring 2011.

44
 Fig. 1 – The BHL-Europe users (taxonomists, general public) will mostly access the
content either through the BHL-Europe / BHL Portal or Europeana (ESE = Europeana
Semantic Elements). The major access route for the librarians managing the scanning
process is the Global References Index to Biodiversity. It is composed of the catalogue
records of the physical library collections.

Fig. 2 – High-level overview of the OAIS components (grey box) with BHL-Europe Pre
Ingest and Portal. The OAIS reference model differentiates between three kinds of
information objects. The SIP, Submission Information Package, is being sent in by the
data producers (content providers), the AIP, Archive Information Package, is preserved
in the Archival Storage, and the DIP, Dissemination Information Package, is provided to
the consumers.

45
 Since June 2010 BHL-Europe content is made accessible for a wider public
via Europeana (http://www.europeana.eu), the virtual European library. More
than 80,000 books are currently accessible in Europeana and this number will
increase continuously while BHL-Europe is harvesting digital literature from its
content providers.

3 BHL-Europe for taxonomists

3.1 Taxonomists as portal users

A major goal of BHL-Europe is providing and facilitating open access to

taxonomic literature for a number of target users (scientists, hobby scientists,
students, teachers, environmental and conservation agencies, etc.). This will
result in a multilingual access point for search and retrieval of biodiversity content
through a robust biodiversity community portal with an open and distributed
architecture and specific functionalities, as described above. Integrated web
tools like taxonomic intelligence (TI) will facilitate search for taxon-specific
biodiversity information and thus improve efficiency of research in biology and
access to information for a wider public. The key components for achieving
this goal are excellent portal tools and the improvement of optical character
recognition (OCR). OCR processing of scanned pages is the base for extracting
specific terms and taxon names out of the pages. However, this is still a major
challenge as automatic OCR accuracy is still not sufficient for our purpose,
especially for historical literature and texts in multiple languages.
BHL-Europe aims at a deep level of language integration in the indexing
system (multilingual indexing). For taxonomists key data of searchable metadata
are names of organisms, biological groups and provenance. However, names of
taxa and groups of organisms are inconsistent, represent different and changing
scientific concepts or are vernacular names. Thus it is a major task for BHL-
Europe to improve existing taxon-recognition tools (TI) for an advanced search
of taxonomic information.

3.2 Taxonomists as partners

BHL-Europe is building the portal and all associated services for the users
to meet their needs and requirements. BHL-Europe has to understand and
evaluate the requirements of the users and how they are going to use the results
of the project. Therefore, a very close cooperation between the users and the
project is essential to make the project a success.
BHL-Europe is targeting a large number of different users ranging from
libraries over different types of scientists to the general public. A number of
instruments are currently used or will be used for the user interaction to prioritise
the technical and collection development plan:
(1) Web analytics will be used to quantify the use of the portal (visits, unique
visitors, page views, referring sites, country coverage).

46
(2) Users are encouraged to drop feedback messages either using the BHL
online discussion forum or using the online contact form. BHL also has an issue
tracking system (Gemini) in place to collect user feedback.
(3) Face-to-face and virtual interactions between the BHL-Europe members
are helpful to get important input, as the project includes a number of key
users from different user groups (e.g. libraries, taxonomists). These users from
within the project work together in Use Case Workgroups. Their major task is
the development of use cases for the portal prototype and testing of the portal
functionalities.
(4) Suggestions of actors and users of large international projects like EOL or
Europeana that are setting priorities based on their experiences are taken into
account.
(5) BHL-Europe considers developments in biodiversity informatics and
networked scientific communication like TDWG developments or PLoS
Biodiversity Hubs.
(6) Specific user evaluations will be carried out twice during the project. The
results of the first online user survey analysing the demand and service elements
of the project will be publicly available soon and will be fed into the BHL-Europe
IT development plan.
(7) BHL-Europe offers training opportunities on how to use the portal and its
functionalities as well as other BHL-Europe products, e.g. the GRIB, and will ask
for feedback on possible improvements. A first workshop will be held during the
BioSystematics 2011 in Berlin (http://www.biosyst-berlin-2011.de/).
(8) BHL-Europe is also present with talks and posters in numerous scientific
conferences to personally discuss with the scientists and to attract new users.
All information collected from the user’s side this way will form the basis for
developing a comprehensive set of use cases for the BHL-Europe portal and
leads to further improvements of the system infrastructure.
In the past, individual scientist or the scientific community could not influence
the choice of biodiversity literature for major scanning activities. BHL-Europe
is implementing a mechanism that will enable users/ scientists placing a scan
request for a specific volume using the GRIB infrastructure. This will allow
libraries/ content provider to set up a priority list for their scanning activities and
making highly demanded literature available first. As an intermediate solution,
BHL has implemented a scanning request form in their feedback system.
Another goal of BHL-Europe is to seek for new partners (content provider,
right holders) that can potentially contribute open access digitised biodiversity
literature to the overall BHL repository. Therefore, support from the scientific
community is highly welcome in naming additional repository of digital content
and bibliographic data or in discussing with rights holder to make their content
freely available.

4 Conclusion
The Chinese Academy of Science and the Atlas of Living Australia have been
joining BHL already and negotiations with organisations in other countries
are underway to further extend the BHL network. All these projects will work

47
together sharing content, protocols, services and digital preservation practices
and promote the idea of a Global Biodiversity Heritage Library.

Acknowledgement

BHL-Europe is co-funded by the Community Programme eContentplus, which is

gratefully acknowledged.

48
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 49-51.
ISBN 978-88-8303-295-0. EUT, 2010.

A Pan-European Species-
directories Infrastructure (PESI)
Yde de Jong

Abstract — This paper introduces the rationale and aims of the Europe-
wide biodiversity informatics PESI [1] project. PESI defines and coordinates
strategies to enhance the quality and reliability of European biodiversity
information by integrating the infrastructural components of four major
community networks on taxonomic indexing, namely those of marine life,
terrestrial plants, fungi and animals, into a joint work programme. This will
include functional knowledge networks of both taxonomic experts and regional
focal points, which will collaborate on the establishment of standardised
and authoritative taxonomic (meta-) data. In addition PESI will coordinate
the integration and synchronisation of the European taxonomic information
systems into a joint e-infrastructure and the creation of a common user-
interface disseminating the pan-European checklists and associated user-
services results.

Index Terms —biodiversity, infrastructure, taxonomy, nomenclature,

standards, Europe, checklists, bio-informatics, cybertaxonomy.

—————————— u ——————————

1 Introduction

T
he correct use of names and their relationships is essential for
biodiversity management; therefore the availability of taxonomically
validated, standardised nomenclatures is fundamental for biological
e-infrastructures. PESI is the next step in integrating and securing taxonomically
authoritative species name registers, serving to underpin the management of
biodiversity in Europe.
PESI is a joint initiative of two Networks of Excellence: EDIT (European
Distributed Institute of Taxonomy) [2] and MarBEF (Marine Biodiversity and
Ecosystem Functioning) [3], funded by the European Commission under the
Seventh Framework Capacities Work Programme - Research Infrastructures
- and is led by the University of Amsterdam. It was started in May 2008 and
will last three years, involving 40 partner organisations from 26 countries and
several non-contracted associated partners.
————————————————
The author is with the Zoological Museum Amsterdam, Faculty of Science - University of Amster-
dam, Amsterdam, P.O. Box 94766, NL-1090 GT Amsterdam, The Netherlands. E-mail: yjong@
science.uva.nl.

49
2 Integrating Infrastructures

2.1 Rationale

PESI defines and coordinates strategies to integrate the infrastructural

components of four major community networks on taxonomic indexing and their
respective knowledge (social and technical) infrastructures; those of marine life,
terrestrial plants, fungi and animals, into a joint work programme. These include
the three main all-taxon registers in Europe, namely the European Register
of Marine Species [4], Fauna Europaea [5], and Euro+Med PlantBase [6] in
coordination with EU-based nomenclators, i.e. Index Fungorum [7], IPNI [8],
and AlgaeBase [9], plus the network of EU-based Global Species Databases
(GSDs).

2.2 Coordination and integration of European expert networks

The integration of the social expertise networks will result in functional

knowledge systems of taxonomic experts and regional focal points, which will
collaborate on the establishment of standardised and authoritative taxonomic
(meta-)data and the development of approaches for long-term data government.
The sustainability of these taxonomic expert networks is considered as the most
threatening issue to PESI’s success, since Europe is experiencing a decline in
its number of professional taxonomists.
PESI is addressing this concern by advancing the abilities of the Society for
the Management of Electronic Biodiversity Data (SMEBD) [10], by collaborating
with the European Distributes Institute on Taxonomy (EDIT) project and the
Consortium of European Taxonomic Facilities (CETAF) [11], as well as reaching
out to non-professional taxonomists and taxonomic societies in a hope to revive
this vital science.

2.3 Coordination and integration of information e-infrastructures

The technical integration of these checklists into a joint ‘European Taxonomic

Backbone’ relies on the Common Data Model (CDM) [12], ensuring the
conceptual mapping of taxonomic databases. This is hosted in the CDM store as
a denormalised relational database management system (the so-called ‘PESI
data warehouse’). The CDM represents a component of EDIT’s Cybertaxonomy
Platform [13].
PESI is also involved in supporting international efforts on the development
of the ‘Global Names Architecture’ (GNA) [14] by building a common intelligent
name-matching device in consultation with principal initiatives like GBIF [15]
and LifeWatch [16]. This provides a unified cross-reference system to all
stakeholders optimising their taxonomic meta-data service functioning.

50
2.4 Integrated e-Services for users and dissemination

PESI will build an interactive, multilingual web portal [17] to carry out the
dissemination of the developed species names service and to support the use
of the pan-European species data in the e-science domain. This will include
relevant supplementary data, like occurrence details by applying dynamic links
to pertinent e-data services.

Acknowledgement

The authors wish to thank all PESI partners, especially PESI work-package leaders
and managers, for their contributions.

References
[1] PESI (http://www.eu-nomen.eu/pesi)
[2] EDIT (http://www.e-taxonomy.eu)
[3] MarBEF (http://www.marbef.org)
[4] ERMS (http://www.marbef.org/data/erms.php)
[5] Fauna Europaea (http://www.faunaeur.org)
[6] Euro+Med PlantBase (http://www.emplantbase.org/home.html)
[7] Index Fungorum (http://www.indexfungorum.org) also (http://pesi.indexfungorum.org)
[8] IPNI (http://www.ipni.org)
[9] AlgaeBase (http://www.algaebase.org)
[10] SMEBD (http://www.smebd.eu)
[11] CETAF (http://www.cetaf.org)
[12] CDM (http://dev.e-taxonomy.eu/trac/wiki/CommonDataModel)
[13] Cybertaxonomy Platform (http://wp5.e-taxonomy.eu)
[14] GNA (http://www.gbif.org/informatics/name-services/global-names-architecture)
[15] GBIF (http://www.gbif.org)
[16] LifeWatch (http://www.lifewatch.eu)
[17] PESI portal (http://www.eu-nomen.eu/porta)

51
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – p. 53.
ISBN 978-88-8303-295-0. EUT, 2010.

ViBRANT—Virtual Biodiversity
Research and Access Network
for Taxonomy
Dave Roberts, Vince Smith

Abstract — Biodiversity science brings information science and

technologies to bear on the data and information generated by the
study of organisms, their genes, and their interactions. ViBRANT will
help focus the collective output of biodiversity science, making it more
transparent, accountable, and accessible. Mobilising these data will
address global environmental challenges, contribute to sustainable
development, and promote the conservation of biological diversity.
Through a platform of web based informatics tools and services we
have built a successful data-publishing framework (Scratchpads)
that allows distributed groups of scientists to create their own
virtual research communities supporting biodiversity science. The
infrastructure is highly user-oriented, focusing on the needs of
research networks through a flexible and scalable system architecture,
offering adaptable user interfaces for the development of various
services. In just 28 months the Scratchpads have been adopted by
over 120 communities in more than 60 countries, embracing over
1,500 users. ViBRANT will distribute the management, hardware
infrastructure and software development of this system and connect
with the broader landscape of biodiversity initiatives including PESI,
Biodiversity Heritage Library (Europe), GBIF and EoL. The system
will also inform the design of the LifeWatch Service Centre and is
aligned with the ELIXIR and EMBRC objectives, all part of the ESFRI
roadmap. ViBRANT will extend the userbase, reaching out to new
multidisciplinary communities including citizen scientists by offering
an enhanced suite of services and functionality.

————————————————
The authors are with the Natural History Museum,Cromwell Road, London, SW7 5BD, UK. E-mail:
[email protected], [email protected].

53
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 55-57.
ISBN 978-88-8303-295-0. EUT, 2010.

Identifications in BioPortals™
Wouter Addink, Edwin van Spronsen, Peter H. Schalk

Abstract — BioPortals are a ‘Google-like’ webportal solution tailored for

national or thematic biological diversity information needs. This solution allows
for an efficient route to retrieve information from heterogeneous biological
information sources after identification with integrated identification systems.
BioPortals can be used in combination with mobile devices and provide
options to share biological observation data after identification in the field.

Index Terms — BioPortals, species information, identification, webportal,

mobile platforms.

—————————— u ——————————

1 Introduction

D
igital information on biological diversity (i.e. on identification, species,
taxonomy, ecology, genetics, conservation, legislation) is often compiled
for a specific purpose and stored in custom-made, geographically
distributed software using different formats. Mining this information can therefore
be time consuming, and recombination of data can be cumbersome because of
incompatibilities. Furthermore, a proper identification of the species is required
in order to be able to retrieve the right biological information.
International initiatives, such as the Global Biodiversity Information Facility
(GBIF), the Encyclopedia of Life (EoL), and the Consortium for the Barcode
of Life (CBOL), are paving the way to make data in a specific domain globally
accessible. However, much of the demand for information is on a national or
thematic level, driven by defined groups of users with specific questions or
problems. These call for a custom-made answer to their information needs.
ETI developed a ‘Google-like’ webportal solution that provides a single access
point to a large array of heterogeneous biodiversity information sources,
combining them with identification keys. The so-called BioPortal can be
customized to specific information needs and user levels: from scientists to
conservationists, from governments to schools.

2 Identification Keys
Keys are a method to identify by asking a series of questions to the user.
————————————————
W. Addink is Head of Informatics, ETI BioInformatics, Amsterdam. [email protected].
E. van Spronsen is Head of Information, ETI BioInformatics, Amsterdam. [email protected].
P. Schalk is Director of ETI BioInformatics, Amsterdam. [email protected].

55
Each answer excludes some names, until the most likely name for the species
remains. This name can then be compared with the description of the species
to confirm the identification. This method is more efficient than going through
a series of species descriptions until a description has been found that fully
matches the observed species.

Fig. 1 – A picture key in the Tanzanian national biodiversity portal, built with the
BioPortal toolkit.

Traditional keys published on paper are usually dichotomous, text-based

keys, where the user is given a sequence of choices, each choice leading to
a new choice and excluding some names until only one name remains. The
BioPortal toolkit supports such traditional keys, but also other types of keys that
allow for more interaction and easier identification. In such keys the user can
select only the characters that are visible, for example only leaf characters to
identify a plant that is not flowering, or by using suggestions for the characters
that are the most separating ones. Keys that use illustrations or sounds instead
of text, or combinations of these are also supported.

3 From Identification to Information

The BioPortal toolkit includes the Linnaeus II Species Bank compiler that
allows the creation of targeted (e.g., national) species information systems with
identification tools for the Internet (e.g. www.soortenbank.nl), accessible with
web browsers on computers or on mobile devices such as iPad, PDA or iPhone.
The Linnaeus II compiler is compatible with the systems and exchange formats
developed in the KeyToNature project. Identification keys that have been made
available in KeyToNature can therefore be used in BioPortals. This allows users

56
of BioPortals to go directly from an identification made with a KeyToNature
identification key to information about the identified species from a range of
online information sources.

4 Sharing Information after Identification

With mobile identification tools like the Netherlands e-Flora application for
iPhone, identifications can be made on spot. When a species is identified, the
species observation (identified species name, photo, date, long-lat location,
observer) can be uploaded directly to a central server and, after vetting, displayed
in a BioPortal. This allows BioPortal users to see the identified observations in
the portal together with related information from other sources.

5 BioPortal Design and Implementation

The BioPortal toolkit has a modular design that can be tuned to user
needs, including customizing the web-interface and providing support for
multiple languages. The toolkit combines content management system (CMS)
functionality, such as news items and static pages, event tracking, and forum
modules, with functions to access the scientific core data. These are biological
collections, observations, ecological relations, molecular data, and image
libraries compiled on a national or local level, combined with external sources
such as GBIF or EMBnet (European Molecular Biology Network). Sophisticated
technology allows for simultaneous server-side asynchronous searches in
several distributed data sources on the web. The Catalogue of Life is used as a
validated taxonomic index to match searches to connected data sources.
ETI’s BioPortal toolkit was used to build the Netherlands’ national GBIF
portal (NLBIF), http://www.nlbif.nl and the Tanzanian national biodiversity portal
(TanBIF: http://www.tanbif.or.tz). It is also used to build the global pollinators
portal.

Acknowledgement

The authors wish to thank all the persons involved in KeyToNature throughout Europe.
Their efforts and input gave use new ideas and energy to develop them. This paper was
produced in the framework of the the project KeyToNature (www.keytonature.eu, ECP-
2006-EDU-410019), funded in the eContentplus Programme.

57
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 59-64.
ISBN 978-88-8303-295-0. EUT, 2010.

Types of identification keys

Gregor Hagedorn, Gerhard Rambold, Stefano Martellos

Abstract — A number of terms related to identification tools are introduced

and the advantages of selected types of identification keys are compared.

Index Terms — Identification tools, single-access key, dichotomous key,

polytomous key, lead, couplet, free-access key, multi-access key, matrix key,
multi-entry key.

—————————— u ——————————

1 Introduction

T
he generalization of individuals (things, events, etc.) into classes is essen­
tial to transfer knowledge across individual incidents. When learn­ing a
language, we learn the defining features of classes like “table”, “chair”,
“shrub”, etc. Similarly, biology defines formal classes for living things (called
“taxa”) together with class names (“taxon names”) and defining descriptions.
The assignment of an unknown object to a taxon is called “identification” or
“determination”. To non-biologists this may be confusing, the term “identification”
being more commonly associ­ated with the naming of individuals (as in “ID card”
or “record identifier”).
The number of taxa in biology is very large. For example, currently about
900 000 insect taxa alone are recognized. Compared to the aver­age vocabulary
of an educated English native speaker of roughly 25  000 words, it is clear
that teaching the vast “taxon vocabulary” to biology students was always
problematic. Although comparing a collected spe­cimen sequentially with
published descriptions or re­presentative speci­mens is an essential identifica­
tion method, any “linear search” method comparing one specimen after another
soon becomes impractical.
Biologists have therefore developed various forms of “identification keys”
to “unlock” the knowledge that would otherwise remain inaccessible. These
are essentially “divide and conquer” search algorithms that re­duce the result
set recursively until the remainder is small enough to be sol­ved by direct
comparison. The fastest algorithms are those that pro­vide a division into equally
sized partitions (leading to search algorithms that scale logarithmically with

————————————————
G. Hagedorn is with the Julius Kühn-Institute, Federal Research Centre for Cultivated Plants,
Institute for Epidemiology and Pathogen Diagnostics, Königin-Luise-Straße 19, D-14195 Berlin, E-
mail: [email protected]. – G. Rambold is with the Department of Mycology, University
of Bay­reuth, D-95440 Bayreuth – S. Martellos is with the Dept. of Life Sciences, Univ. of Trieste,
I-34127, Trieste, Italy. E-mail:[email protected].

59
the number of taxa). Biological keys don’t al­ways provide this, because other
factors (character observation reliability, convenience, cost, etc) conflict with the
desire to provide fastest progress. The authors of biological identification keys,
however, typically realize that evenly splitting choices are desirable.

2 Single-access, multi- or free-access, and multi-entry keys

The most traditional biological identification keys are the easily printable
dichoto­mous (every choice has two alternatives) or polytomous (two or more
options at each choice) forms. The structure of these keys typically consists
of a series of alternative statements, called “leads”. All leads that need to be
evaluated for a single decision form a “couplet”.
Clearly, dichotomous keys are a special case of polytomous keys. For simple
choices involving a single characteristic (a “character” or “feature”), such as
“wings: 1. present 2. absent” or “flower colour: 1. red 2. blue 3. pink 4. yellow”,
the number of options matters little. However, to achieve reliable identification
in the face of natural variability and continuous varia­tion, it may be desirable to
use complex Boolean statements involving multiple characters. For example,
“leaves hairy and flowers red” versus “leaves glabrous or flowers not red” may
be used in a case where the alternative may include glabrous plants with pinkish-
red flow­ers, or plants with mixed hairiness but other flower colours. Although
it is theo­re­tically possible to construct a polytomous key with Boolean lead
statements, practice has shown that the result is often akin to a logical riddle.
Many editors therefore recommend or require the use of di­cho­tomous keys.
However, a key may be a mixture of simple polytomous and complex di­
chotomous choices. The generalizing term “single-access key” [1] is there­fore
used in the present paper to include both dichotomous and poly­to­mous keys.
The equivalent term in computer science is “decision tree”.
Question 1
Question 2 Question 1 Question 1 Question 1
Question 3 ... Question 2 Question 2 Question 2
Question 4 ... Question 3 Question 3 Question 3 ...
.
Question 5 Question 4 Question 4 Question 4
Question 6 ... Question 5 Question 5 Question 5
Question 7 ...

Fig. 1 – User interaction steps in a single-access key (left, the sequence of steps follow
the data structure) and a free-access key (right, the sequence is determined by the
user). From [2].

An alternative to a single-access key is the free-access key (also known

as multi-access key, matrix key, or, incorrectly, “synoptic key”1). Whereas
in a single-access key a fixed sequence of choices (decisions) is defined by
the author (providing a single path to each result), in a free-access key the
sequence of choices is up to the user. In every step, the user can se­lect from
the list of characters offered, and choose a matching state or value. Thus a free-
access key is the set of all possible single-access keys that arise by permutating
the order of characters (Fig. 1). Al­though print­able free-access keys exist, they

————————————————
1
The term „synoptic key“ has traditionally been used for single-access keys that reflect the
taxonomic hierarchy. Its use for multi-access keys (especially printable ones) should be avoided.

60
are most suitable for computer-aided identification tools, and have a long de­
velopment history [3]. Examples are DELTA-IntKey [4], Lucid [5], NaviKey [6],
Xper2 [7]. The Flash-based IBIS-ID [8] was newly developed in KeyToNature.
In a free-access key, the choice of characters is repeated at every step. A
related form, the multi-entry key, allows free choice of characters in a first step
(a “multi-character-query-form”), followed either by a field-guide-like listing of
remaining taxa, or by a dynamically generated (filtered) single-access key (as in
the FRIDA/Dryades keys [9]).

Single-Access Free-Access Multi-Entry

Variable (none if
Information None (complete informa­ all characters are
High
reduction tion is opti­mal) avail­able in the initial
step)

Average De­pends on user’s back­ Variable between

Depends on the
identification ground knowledge; may single and multi-
creators of the key
speed ex­ceed average access

Complex Yes (not “No” in entry-form,

statements recommen­ded for No “yes” in following
(and, or, etc.) poly­tomous keys) single-access key

(Implicit in charac­
Question- Possible for simple (Implicit in char­acter
ter state or value
answer style statements state or value choice)
choice)

Difficulty of Variable; depen­ds

choosing next None Often high for beginners on completeness of
decision initial entry form

Difficult; all
Skipping un­ Easy in entry-form,
alternative paths
answerable Easy difficult in an optional
must be follow­ed to
choices single-access part
the end

Resources Low for first draft. High for data matrix,

High investment until first
required for Good keys require but the size of the
version can be tested
construction high expertise matrix is variable

Tab. 1 – Comparison of some identification methods. The comparison is aimed at

manually created single-access keys (those generated from a data matrix are not
considered here).

An evenly splitting single-access key requires fewer decisions from be­ginners

than a multi-access key. The latter may in fact generate faster progress in terms
of “steps”, but requires additional decisions as to which character to use next.
Even if the character list is ordered by character suit­ability and fastest progress,
beginners will be tempted to use execute their free character choice. This is
problematic if many characters are not yet understood. In contrast, the fixed

61
identification path of a single-access key also fixes which terms and concepts
must be learned first. A disad­van­tage of single-access keys is that identification
may be impossi­ble if a choice cannot be decided at all. This may occur because
a charac­ter can­not be observed (e. g., a developmental stage is not present in
the spe­ci­men), or because the options are not communicated clearly enough.
The resulting frustration can be high, especially for beginners.
Both free-access and multi-entry keys truly excel in their performance when
used by experts. For these, character selection is intuitive and fast. By choosing
characters, for which a rare state is present in the specimen, identification
progress can then be on an order of magnitude faster than using a single-
access key. This is already possible with moderate experi­ence, since states that
were never observed by a user before are, by defi­ni­tion, rare. Tab. 1 gives an
overview of some differentiating features.
From an author’s perspective, matrix-based keys require a high initial in­vest­
ment to research and fill a large character × taxon matrix. In con­trast, single-
access keys require less formal investment. Due to the inher­ent information
reduction (most characters apply only to a rela­tively small sub­set of taxa), a
reviewable key is faster to produce and proof-reading is less time-consuming
than the creation of an equivalent data matrix con­tain­ing all characters for the
same group of taxa. However, a success­ful single-access key depends strongly
on the expertise of the author to chose characters that are convenient, cost-
effective, reliable across all taxa in the subtree, and avail­able throughout a large
period of the de­vel­op­mental cycle of the organ­ism. Single-access keys may
therefore require several cycles of testing until initially overlooked problems
have been fixed; their production can be akin to the “debugging” of software
code.
Furthermore, the creation of matrix-based keys generally requires learn­ing a
special-purpose application like DELTA or Lucid, whereas single-access keys
may be created in a text-processing application. There­fore, although newly
created single-access keys may occasion­ally be problematic to use, they offer
con­siderable benefits to both pro­du­cers and consum­ers.
Single-access keys, until re­cent­ly, have been developed only rarely as com­
puter-aided, interactive tools. Notewor­thy developments in this direc­tion are the
commercial Lucid Phoenix applica­tion [10], the FRIDA/Dry­a­des software [9],
[11], the KeyToNature Open Key Editor” [12], and the open source WikiKeys and
jKey [13] application on biowikifarm [14].

3 Structural variants of single-access keys

Two additional structural variants of single access keys are relevant when
building information models [2]: 1. Couplets may consist of a ques­tion, with
the leads providing contrasting answers. This question-answer-style is of­ten
appealing to beginners. However, complex statements (involving more than one
character and Boolean expressions) are not possible. Where­as a mixture of
simple polytomous and complex dicho­to­mous couplets in a single key is quite
intelligible, a mixture of simple quest­ion-answer-style with question-less complex
dichotomous couplets is not. 2. The desire for fast identification progress using

62
conven­ient char­ac­ters often conflicts with character variability in a subset of
organisms. As long as the character is reliable for the majority of taxa, a frequent
solution to this problem is to key out taxa with variable character expression
multi­ple times. This may affect only the terminal taxa, or entire branches of the
keys. Whereas the first case will often simply be handled by true duplica­tion,
multiple references to entire branches of a decision tree turn a “tree” structure
into a directed (and generally acyclic) graph (DAG) and requires careful attention
when modelling information models or software. In biol­ogy a DAG is sometimes
called a “reticulated” identifica­tion key.

Linked Key Style (also called “parallel”, “juxtaposition” or “bracketed” style):

1. Ovule solitary, basal ..........................................................................................................2

– Ovules numerous, axile or free-central .............................................................................3
2. Perianth green, membranous or absent; filaments free .........................Chenopodiaceae
– Perianth translucent and papery; filaments often united below ...............Amaranthaceae
3. Placentation axile; leaves alternate ............................................................Saxifragaceae
– Placentation basal or free-central; leaves usually opposite ............................................. 4

Nested Key Style (also called “yoked” or “indented” style):

1. Ovule solitary, basal

2. Perianth green, membranous or absent; filaments free .....................Chenopodiaceae
2. Perianth translucent and papery; filaments often united below ...........Amaranthaceae
1. Ovules numerous, axile or free-central
3. Placentation axile; leaves alternate ........................................................Saxifragaceae
3. Placentation basal or free-central; leaves usually opposite

Fig. 2 – Examples of the linked and nested styles of branching keys in lead style; see [2]
for derivation.

4 Presentational variants of single-access keys

The dominant presentation styles of single-access keys are shown in Fig. 2. In
“linked” keys the connection between couplets is achieved by a link­ing reference
(at the right side) to a couplet ID (left). In nested keys direct nesting of couplets
replaces the explicit linking. Nested keys are more commonly known as “in­den­
ted”, but unfortunately this refers to an acci­den­tal (albeit fre­quent) rather than
essential quality [3]. Nested keys may be printed without indentation to preserve
space (relying solely on corres­pon­ding lead symbols) and linked keys may be
indented to en­hance the visibility of the couplet structure. Further presentational
(“solid keys”, “graphical style”) and semantic (“arti­ficial” or “diagnostic” versus
“natural”, “synoptic”, or “phylogenetic”) variants exist; see [2].

5 Summary
The order of couplets (choices) in an identification tool may be defined by the
creator (single-access key), or may be freely selectable by the user (free-access
key). A multi-entry key is an intermediate form that may com­bine advantages of
both forms if only a small character subset is included in the multi-entry phase.

63
Structural criteria for single-access keys are: a) whether the leads in a couplet
are limited to two (dichotomous) or not (polytomous key); b) whether couplets
are limited to a single character or combinations of multiple characters, involving
Boolean operators such as ‘and’, ‘or’, or ‘not’, are supported; c) whether taxa may
be keyed out in mul­tiple places, and whether redirections into entire sections (or
“bran­ches”) of the key are supported (“reticulated key”); and d) whether leads in
couplets are complete statements or split into a question with the couplet and
leads providing the answers. Certain presentational forms (nested key versus
linked keys) are not structurally relevant.

Acknowledgement

This work was supported by the KeyToNature Project, ECP-2006-EDU-410019, in the

eContentplus Programme. We thank P. L. Nimis and B. Press for review and important
suggestions.

References
[1] J. Winston, Describing Species. Columbia University Press, 1999.
[2] G. Hagedorn, Structuring Descriptive Data of Organisms — Requirement Analysis and
Information Models. Ph. D. Thesis, Universität Bayreuth, 2007.
[3] R. J. Pankhurst, Practical Taxonomic Computing, 1991.
[4] DELTA – DEscription Language for TAxonomy http://delta-intkey.com/, 2010-07.
[5] Lucidcentral.org http://www.lucidcentral.com, 2010-07.
[6] D. Neubacher and G. Rambold, NaviKey – a Java applet and application for access­ing
descriptive data coded in DELTA format, 2005 (onwards). http://www.navikey.net, 2010-07.
[7] V. Ung, G. Dubus, R. Zaragüeta-Bagils and R. Vignes Lebbe, “Xper2: introducing e-taxonomy”.
Bioinformatics, 26 (5): 703-704, 2010; doi: 10.1093/ bioinformatics/btp715; see also http://lis-
upmc.snv.jussieu.fr/lis/?q=en/resources/software/xper2.
[8] M. Giurgiu, G. Hagedorn and A. Homodi, “IBIS-ID, an Adobe FLEX based identification tool
for SDD-encoded multi-access keys”. Proc. of TDWG 2009, 9-13 Nov. 2009, Montpellier, p.
90, 2009.
[9] S. Martellos and P.  L. Nimis, “KeyToNature: Teaching and Learning Biodiversity. Dryades,
the Italian Experience”. In: M. Muñoz, I. Jelìnek, F. Ferreira (eds.), Proceedings of the IASK
Interna­tional Conference Teaching and Learning 2008, pp. 863-868, 2008.
[10] Lucid Phoenix (http://www.lucidcentral.org/LinkClick.aspx?link=152), (2010-07).
[11] S. Martellos, “Multi-authored interactive identification keys: The FRIDA (FRiendly IDentificAtion)
package”, Taxon, vol. 59 (3), pp. 922-929, 2010.
[12] S. Martellos, E. v. Spronsen, D. Seijts, N. Torrescasana Aloy, P. Schalk and P. L. Nimis, “User-
generated content in the digital identification of organisms: the KeyToNature approach”, Int. J.
Information and Operations Management Education, vol. 3, 3, pp. 272-83, 2010.
[13] S. Opitz, G. Hagedorn, “The jKey wiki key player and builder”. Proc. of TDWG 2009, 9-13 Nov.
2009, Montpellier, 2009.
[14] G. Hagedorn, G. Weber, A. Plank, M. Giurgiu, A. Homodi, C. Veja, G. Schmidt, P. Mihnev,
M. Roujinov, D. Triebel, R. A. Morris, B. Zelazny, E. van Spronsen, P. Schalk, C. Kittl, R.
Brandner, S. Martellos and P. L. Nimis, “An online authoring and publishing platform for field
guides and identifi­cation tools”. In: P. L. Nimis and R. Vignes Lebbe (eds.), Tools for Identifying
Biodiversity: Progress and Problems, pp. 13-18, 2010.

64
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 65-70.
ISBN 978-88-8303-295-0. EUT, 2010.

Learning, Identifying, Sharing

Philippe A. Martin, Noël Conruyt, David Grosser

Abstract — This article argues that a cooperatively-built, well-organized,

shared knowledge base is a new – and, from certain viewpoints, optimal –
kind of support (refining and integrating other kinds of supports) for three
complementary tasks: learning about living entities (and how to identify them),
supporting their identification, and sharing knowledge about them. This article
gives the ideas behind our prototype, and argues that knowledge providers
can be not solely specialists, but also amateurs. In essence, for these three
tasks, it argues for the (re-)use of much more semantically organized and
interconnected versions of semantic wikis or scratchpads.

Index Terms — identifying, knowledge sharing, learning, ontologies, semantic

wikis.

—————————— u ——————————

1 Introduction

C
urrent supports for learning about – and identifying – living entities, e.g.,
the supports listed by the KeyToNature project (www.keytonature.eu),
are mostly static files (texts, images, …) and tools based on a formal1
knowledge base (KB). Few tools allow their users to contribute annotations
or other information to their formal or informal KB, let alone use them for i)
helping identification or learning, and ii) publishing them in a way usable by
other tools. Scratchpads [1] and, more generally, semantic wikis2, allow the
cooperative edition and semantic linking of information by any web user, but
not in an organized or formal enough way to be re-used by an identification
tool or a problem-solving tool, nor to permit the automatic detection of partially
redundant/inconsistent information within or between wikis. This automatic
detection is essential to permit the semi-automatic and cooperative organization
of knowledge into a unique semantic network and thus permit i) scalable
information retrieval, comparison, sharing and exploitation, and hence ii)
an easier understanding or learning (by amateurs or specialists) of the
stored information and viewpoints of their authors. Section 2 quickly compares
the various current kinds of supports for the learning and sharing of information
————————————————
The authors are with the IREMIA laboratory, University of La Réunion. E-mail: (Philippe.Martin,
Noel.Conruyt, David.Grosser)@univ-reunion.fr.
1
In this article, “formal” means machine processable and logic-based, while “semantic” means
formal and organized by semantic relations, e.g., “subtype of”, “physical_part of”, “agent of” and
“duration of”.
2
Semantic wikis are collaboratively-built documents with some parts indexed by semantic categories
or interconnected by semantic relations. See http://semwiki.org for more details.

65
about living entities and hence for helping their identification.
Section 3 introduces elements required to support an approach leading to
a global KB composed of collaboratively-built KBs that have no implicit3
“automatically detectable partial redundancies or inconsistencies” neither
within nor between the KBs. As suggested in Section 2, such a global KB – and
hence this approach (which is complementary to the other approaches) – is the
most useful one from a knowledge-sharing, retrieval and learning viewpoint,
but its disadvantages are that i) it requires the users to learn how to read a
textual or graphic notation for representing or interconnecting knowledge,
and ii) for each domain that has not yet been well represented in the shared
KB, the first knowledge providers have a lot of work to do for organizing the
information resulting from the use of other approaches. However, this can be
done incrementally, whenever the benefits finally becomes clearer than the
costs. The elements of this approach are fully or partially implemented in our
knowledge server WebKB-2 [2] (webkb.org).

2 Quick Comparison of Approaches

The smaller the sources of information used for knowledge sharing – i.e. the
less objects of information (e.g., statements or images) these resources contain
– and the less contextual (hence more explicit, precise and formal) these
objects are, the easier it is to automatically index these resources precisely, to
filter out the redundancies and to relate these resources via semantic relations,
e.g., to organize them into a specialization hierarchy4. Then, the easier it is to
retrieve these resources (by querying or browsing)5, compare them (hence,
understand and memorize them), combine them and, more generally, exploit
them for various purposes, e.g., guiding identification. As illustrated in the
following paragraphs, these rather obvious ideas are generally well accepted,
but their ultimate conclusion is socially and technically difficult to bring about
and hence not directly studied. The conclusion is: there should ideally be one
and only one global semantic network (i.e., each index or symbolic resource
should contain only one statement or one formal term; in other words, there
should be no difference between should symbolic data and meta-data) and,
in this network, all manually or automatically detected partial redundancies or
inconsistencies are made explicit via semantic relations. In this article, such a
global semantic network is called a global cbwoKB (cooperatively-built well-
organized KB).
The Learning object (LO) related community and standards (e.g., IEEE LTSC)

————————————————
3
In this article, implicit means “not made explicit via a semantic relation”.
4
Related small individual statements can often be organized into a specialization hierarchy or an
inclusion hierarchy but sets of related statements rarely can (the bigger the sets, the less likely).
5
For example, if the query is of the kind “what are the resources/tools/methods to do ...”, the answer
can be a part/subtask/specialization hierarchy (with associated argumentation structures). Such
semantically structured answers allow a user to find and compare all relevant objects instead of
getting a long list of partially redundant objects or files where original/precise ones are hidden
among/behind objects that are more general, more mainstream or from big organizations.

66
advocate the use of small non-contextual LOs but still only considers the use of
static informal documents indexed by keywords. Semantic LO repositories [3]
use formal terms or statements for indices. This is also the approach used by
STERNA [4].
As highlighted in [5] and [6], the Semantic Web (SW) community currently
essentially focuses on inference mechanisms, KB editors, semantic wikis,
social networks, workflow-based cooperation, and the semi-automatic partial
interconnection of the content of (semi-)independently created KBs or formal
files. Tools created by this community do not directly support the creation of
a cbwoKB (global or local) and, in a sense, they participate to the problems
they are trying to solve since their outputs create new files that are partially
redundant or inconsistent with their input files and without semantic relations
to make this explicit. The current focus of the SW community is to work with
approaches hiding the knowledge representations from the users as much as
possible. The problem is then that the semantic network cannot be completed in
a meaningful way by the users (only low quality knowledge can be automatically
extracted and exploited) nor even browsed to find information. As an example,
semantic wikis are still mainly poorly organized informal documents. Instead, in
WebKB-2 the semantic network can be edited by all Web users via cooperation
protocols and can be viewed in a more or less structured way via various
relatively intuitive syntaxes [7]: Formalized-English, For-Links, etc. However,
reading these syntaxes requires a short training and writing knowledge requires
the following of some given conventions or “best practices”.
Scratchpads are kinds of semantic wikis which, according to some of their
documentation [8], are “independent and unconnected, allowing communities to
create distinct customized sites tailored to their needs”. This strongly reduces
the possibilities of (semi-)automatically comparing and integrating the content of
different scratchpads, and hence works against the goals of identification-related
projects like ViBRANT [9] which is based on the use of scratchpads. With a
cbwoKB, tailoring can be done by each user using filters and presentation rules.
Many identification-related projects use databases, e.g., FishBase (fishbase.
org) and Pl@ntNet (plantnet-project.org). They have a regular structure but a
rather flat one and users cannot directly contribute to the database: annotations,
new objects, new tables (classes of objects), new attributes (relations from/
to objects), etc. Finally, the semantics of the objects of these databases is
unknown unless their semantic relations to other objects from the Semantic
Web are described in a formal file.
Except for WebKB-2, current KB servers/editors (e.g., Ontolingua, OntoWeb,
Ontosaurus, Freebase, CYC and semantic wiki servers) have no shared KB
editing protocols and hence either i) let every authorized user modify what other
ones have entered (this discourages information entering or leads to edit wars),
or ii) require all/some users to approve or not changes made in the KB, possibly
via a workflow system (this is bothersome for the evaluators, may force them
to make arbitrary selections, and this is a bottleneck to information sharing that
often discourages information providers). To complement the generic “knowledge
sharing” features of WebKB-2 with identification features, its integration with IKBS
[10], a KB based identification tool, has begun.

67
3 Underlying Ideas of Solutions for the Proposed Approach
To be a generic “knowledge sharing” support, the shared KB of WebKB-2
has been initialized via a loss-less merge of many ontologies (sets of formal
terms with their associated definitions/constraints/inter-relations): top-level
ones (including methodological ones such as DOLCE) and a lexical one (an
extension and correction of WordNet) [11]. Knowledge normalization rules
have been collected and extended; simultaneously, various complementary,
expressive and relatively intuitive notations enforcing these rules have
been designed [7]. Finally, knowledge sharing protocols have been designed
[2]. The protocols for the collaborative edition of a shared cbwoKB have
been implemented and are introduced in the second next paragraph. This is
not yet the case for the protocols permitting to create a global cbwoKB
composed of several cbwoKB servers. Their underlying idea is that each of
these servers must i) publish its commitment to be a “nexus” for one or several
formal terms, that is, to store all information directly related to these terms, and
ii) point to other nexus for terms it is not the nexus of. In this way, via redirections
of queries and replications of knowledge between servers, it does not matter
which server a user updates or queries first, and the advantages of distribution
and centralization are thus combined.
WebKB-2 has an expressive language model (1st-order logic, n-order types,
meta-statements and collections) but has a simple data model since it is built
on top of an object-oriented DBMS with only three tables: Term, Relation and
Source. Every object of the KB is either a formal/informal term or a formal/semi-
formal/informal statement (e.g., a relation between two quantified terms, and a
relation on a relation in order to represent some spatial and temporal context).
Every object has one or several associated sources: i) the user who created the
object, ii) the original resource (e.g., a person, a language, a document) from
which the user read/heard/took the object and hence interpreted it, and iii) other
users who also believe in that object (if it is a statement). Lexical conflicts
are avoided by prefixing formal terms with the identifier of their creators, e.g.,
wn#bird refers to the most common concept (i.e., meaning) proposed by
WordNet for the word “bird”.
The next sentences introduce the most important basic ideas behind the
shared KB editing protocols of WebKB-2 and hence behind the ways semantic
conflicts are avoided and the KB kept “well organized”. A user can re-use
any object (term or statement) but can only modify or remove an object that he
has created. Adding, modifying or removing a term is done by adding, modifying
or removing at least one statement (generally, one relation) that uses this term. A
new term can only be added by specializing another term. Each object must be
connected to at least another object via relations of specialization/generalization,
identity and/or argumentation (and as many as possible of such relations should
be used). If a user adds, modifies or removes a statement (definition or belief)
and this creates a detected conflict (redundancy and inconsistency) with another
of his statements, the action is rejected. If adding, modifying or removing a
(definition of) a term introduces a conflict with statements of other users, this
conflict highlights an over-interpretation of the term by these other users and

68
this is automatically solved by “cloning” the term, i.e., creating a slightly more
general copy of this term for these other users to repair the over-interpretation.
If adding, modifying or removing a belief introduces a detected potential conflict
(partial/total inconsistency or redundancy) involving beliefs created by other
creators, it is rejected. However, a user may still represent his belief (say, b1) –
and thus “loss-less correct” another user’s belief that he does not believe in (say,
b2) – by connecting b1 to b2 via a corrective relation. E.g., here is a Formalized-
English statement by u2 which corrects a statement made earlier by u1:
u2#` u1#`every bird is agent of a flight´ has for corrective_restriction u2#`most healthy flying_
bird are able to be agent of a flight´.
This statement means: “according to u2, u1’s belief that ‘every bird flies’ is
false and a more precise statement is ‘most healthy flying birds (the carinates)
are able to fly”. This way the KB is kept organized and then, if necessary, an
inference engine can choose between such statements according to the
constraints of a particular application, e.g., it can always choose the most
precise version or it can choose the one authored by someone represented as
an expert in a certain domain. Similarly, in the same way he creates queries,
a user can create filters on the content, authors, …, and popularity of
statements in order to see only what he wants to see when browsing the KB.
With this approach, every author can represent his beliefs, no selection
committee is required, and knowledge integration is loss-less (the sources
can be regenerated). This approach also avoids the problems related to
version control or truth-maintenance.

4 Conclusion
This article compared various knowledge sharing approaches and introduced
elements necessary to support the most precision-oriented and end-user-
controlled approach and the one that combines the advantages of the
centralization and distribution. Thus, it is the approach that most permits to i)
retrieve and compare knowledge about a living entity and hence learn about it,
ii) integrate knowledge from everyone (specialists and amateurs), and iii) leads
to create knowledge that directly or indirectly can be re-used by tools to guide
identification. Most of these elements are implemented in WebKB-2. It will soon
be used to enable Web users to extend the content of FishBase and Pl@ntNet.

References
[1] V. S. Smith, S. D. Rycroft, K. T. Harman, B. Scott and D. Roberts, ‘’Scratchpads: a data-
publishing framework to build, share and manage information on the diversity of life,’’ BMC
Bioinformatics 2009, 10 (suppl. 14). See also http://scratchpads.eu, 2010.
[2] P. Martin, “Protocols for Governance-free Loss-less Well-organized Knowledge Sharing”,
ECAI 2010 workshop on Intelligent Engineering Techniques for Knowledge Bases (I-KBET
2010), Lisbon, Portugal, 17 August 2010.
[3] J. S. Carrion, E. G. Gordo and S Sanchez-Alonso, “Semantic learning object repositories”,
International Journal of Continuing Engineering Education and Life Long Learning, vol. 17, 6,
pp. 432-446, 2007.
[4] STERNA, “Semantic Web-based Thematic European Reference Network Application”, http://
www.sterna-net.eu, 2010.

69
[5] N. Shadbolt, T. Berners-Lee and W. Hall, “The semantic web revisited”, IEEE Intelligent
Systems, 21, vol. 3, pp. 96-101, May/June 2006.
[6] R. Palma, P. Haase, Y. Wang and R. d’Aquin, “Propagation models and strategies”, Deliverable
1.3.1 of NeOn - Lifecycle Support for Networked Ontologies; NEON EU-IST-2005-027595,
2006.
[7] P. Martin, “Knowledge representation in CGLF, CGIF, KIF, Frame-CG and Formalized-English”,
Proc. of ICCS 2002, Springer LNAI 2393, pp. 77-91, 2002.
[8] P. Martin, “Protocols for Governance-free Loss-less Well-organized Knowledge Sharing”,
Proc. ECAI 2010 workshop on Intelligent Engineering Techniques for Knowledge Bases
(I-KBET 2010), Lisbon, Portugal, 17 August 2010.
[9] ViBRANT, “Virtual Biodiversity Research and Access Network for Taxonomy”, E.U. FP6
project, http://vbrant.org, 2010.
[10] N. Conruyt and D. Grosser, “Knowledge management in environmental sciences with IKBS:
application to Systematics of Corals of the Mascarene Archipelago”, Selected Contributions in
Data Analysis and Classification, Springer Series: Studies in Classification, Data Analysis and
Knowledge Organization, pp. 333-344, 2007.
[11] P. Martin, “Correction and extension of WordNet 1.7”, Proc. of ICCS, Springer LNAI 2746, pp.
160-173, 2003.

70
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 71-76.
ISBN 978-88-8303-295-0. EUT, 2010.

Identification with iterative

nearest neighbors
using domain knowledge
David Grosser, Noël Conruyt, Henri Ralambondrainy

Abstract — A new iterative and interactive algorithm called CSN (Classification

by Successive Neighborhood) to be used in a complex descriptive objects
identification approach is presented. Complex objects are those designed
by experts within a knowledge base to describe taxa (monography species)
and also real organisms (collection specimens). The algorithm consists of
neighborhoods computations from an incremental basis of characters using
a dissimilarity function which takes into account structures and values of the
objects. A discriminant power function is combined with domain knowledge on
the features set at each iteration. It is shown that CSN consistently outperforms
methods such as identification trees and simplifies interactive classification
processes comparatively to search for K-Nearest-Neighbors method.

Index Terms — identification, Similarity, K-Nearest-Neighbors, Decision

Trees, structured data, knowledge base, life science.

—————————— u ——————————

1 Introduction

I
n the frame of environmental sciences, for helping to preserve rich
ecosystems from biodiversity loss, the acquisition and production of
knowledge on biological specimens and taxa is an essential part of the
work of systematicians [1]. Indeed, being able to describe, classify and identify a
specimen from morphological characters is a first step for monitoring biodiversity,
because it gives access to information relative to its species name (Biology,
Geography, Ecology, Taxonomy, bibliography, photography). These tasks can
be assisted in biodiversity informatics by databases for storing information
and computer science decision support tools for description, classification
and identification purpose with knowledge bases. In return, these complex
representations deliver interesting models and processing problems to deal
both with domain knowledge and specimen descriptions.
In many fields of real world applications, we can capture a given aspect of
the descriptive domain knowledge by associating attributes of the problem
————————————————
The authors are with the Computer Science and Mathematics laboratory (LIM-IREMIA) of Reunion
University – 97400 Saint-Denis, France. E-mail: [email protected].

71
structure with objects linked by composition and/or specialization relationships.
We can also structure the domain definition of nominal attributes by a hierarchy
of values. These techniques enable the algorithms to take into account mutual
dependencies between attributes and values and to compare case properties
with more accuracy.
For instance, for the knowledge base on Corals of the Mascarene Archipelago
(http://coraux.univ-reunion.fr/), the descriptions of specimens are often highly
structured (composite objects, taxonomic attributes), highly noisy (erroneous
or unknown data) and polymorphous (variable, i.e. simultaneous presence
of states or imprecise data). To take into account this complexity, we need to
define a descriptive model (or Ontology) that includes information about objects’
relationships, attribute types and other semantic aspects: scope of the values,
meaning of special values (defaults, exceptions), observation cost of characters.

 
Fig. 1 – Part of a specimen description made with IKBS. Characters (attributes) are
attached to objects (eventually missing or absent) that are organized with composition
relationships.
For engineering Systematics, we have developed a type of knowledge base
system that supports both taxa and specimens descriptions. IKBS (Iterative
Knowledge Base System) is a knowledge management system available on
the Internet (http://ikbs.sourceforge.net) that helps to define descriptive models,
describe instances of these models (see Fig. 1) and then identify new specimen
descriptions with different identification methods: an Identification Tree based
method (monothetic) and a K-Nearest Neighbors method (polythetic) that uses a
dissimilarity function designed to deal with such complex objects representations
[2].2 Classification by Successive Neighborhood (CSN)

72
2 Classification by Successive Neighborhood (CSN)
CSN is a new iterative and interactive method that uses a similarity measure
and a discriminant character selection to identify complex objects [3]. Starting
from a partially described unlabeled object, the method consists in selecting
at each step an objects’ neighborhood in regards to a similarity measure. A
set of candidate classes is computed from the neighbor set considering class
frequencies. A list of discriminant characters is built from the neighborhood and
the best is chosen among that list. The value is obtained interactively from users
(or another data source). A new neighborhood is computed on the basis of the
new partial description of the object. The process iterates until the candidate
classes set is homogeneous.
The iterative process to identify an unlabeled description called e is made of
the following functions:

2.1 Building neighbors set

The neighbors of e at iteration m is the set Vm of objects inside of the sphere of

radius Dm centered at e:
Vm = {o ∈ O | d(e,o) < Dm }, d is a dissimilarity function.
The radius value is determined from the maximum distance, the max
dissimilarity value between e and the set Vm-1:
Dm = max (d(e, oi)), oi ∈ Vm-1.
Then {Dm} is a decreasing sequence.

2.2 Selecting discriminant characters

An ordered list of informative variables is computed at each step of the

classification process from V. The first element is exposed as a question to
the user who can choose an alternative variable from the list. The list is built in
function of three criteria:
• Discriminant power. Choice of different classical criteria computing the
information gain used in machine learning such as Shannon entropy
measure or Gini index. Straightforwardly, this type of criterion minimizes the
number of questions.
• Selectable characters. The method considers only at each step characters
that may be indicated for choice. Relative questions about presence or
absence of components are also considered as selectable attributes.
• Using attributes weighting in the data model reflecting observation cost or
other strategic knowledge about characters.

3 Experiments
In the following experiments, we have extracted some descriptions from
the Fungiidae Knowledge base on Corals of the Mascarene Archipelago that
counts approximately 150 classes and 800 complex objects. We follow a double

73
objective. Firstly, we aim to illustrate the execution of the CSN algorithm in the
IKBS software. Secondly, we want to compare the classification (identification)
accuracy of the CSN method in regards to an identification tree (IT) based
method and a simple K-Nearest-Neighbors (KNN). Both methods already exist
in IKBS and use respectively the same discriminant character selection method
and the same dissimilarity function.

3.1 Execution of CSN algorithm

The example in Tab. 1 illustrates the identification process of an unlabeled

description e initially empty. The Fungiidae Knowledge base is made of 63
cases with 94 characters and 15 species names. To simulate user interactions,
the data source of the description e to identify is set to er corresponding to
a complete and referenced description of a specimen pertaining to species
Fungia concinna (case number 8 among 63). For the correct identification, er
class (species name) is compared with e class at the end of the process. The
criterion used for the character selection function is the classical Shannon’s
information gain measure.

 
Tab. 1 – Example of identification process by successive iterations (Num) of e.

Tab. 1 shows a selected subset of iterations that conducted to a good

detection of e. 21 iterations (and so 21 character values) were necessary. For
each line, the selected character, the corresponding value and the information
gain associated are showed. For each neighborhood, information about the 3
first objects in V is shown: cases indexes (in the case base), attaching classes,

74
values for the selected character and the dissimilarity values. For convenience
needs, the stopping criterion used is the exact matching with the class of the first
case (in bold in the table).
The most interesting information to observe is the progression of V. Variations
of positions show how supplying information to e modifies distances and
consequently the order of cases in V. Thus, for instance, between iteration 8

Tab. 2   – Comparison of IT, KNN and CSN on classification accuracy of 6 knowledge

bases.
and 9, the case 60 goes up to first position because the value of the character
profile of the Skeleton (noted profile[Skeleton]) corresponds to the reference
value, but not cases 62 and 46. At iteration 19 appears the case 39 in position 2,
labeled with the “good class”. To finish, at iteration 21, the case 39 reached the
first position in front of the case 46 and the process stop with a “good matching”.

3.2 Classification accuracy

In this second experiment, we evaluate relative performances of CSN

comparatively to IT and to KNN methods already implemented in IKBS. The first
IT method [4] is an extension of the supervised classification algorithm C4.5 [5]
adapted to the use of a structured descriptive model.
The second KNN method uses the same dissimilarity measure as CSN for
its neighbors set computation. The validation method used is a “leave-one-out”
cross validation process [6] that consists to classify each case of the base using
the others as training set. This method is applied for the three algorithms with
similar conditions. For convenience needs, K is set to 1 in the KNN method.
Tab. 2 gives results of the validation process on six family knowledge bases. For
each base, we show the number of cases and characters, and for each method,
identification accuracy (score).
It demonstrates that IT method presents a low accuracy rate comparatively
to CSN and KNN. Identification errors are frequents, from 12.5% for Mussidae
family to 29.35% for Faviinae. The best method is KNN that may be intuitively
explained by the fact that it uses the overall information of the objects: the
reference case is fully described. CSN offers an intermediate score, near KNN
and often ouperforms IT. We may observe for instance results of the Faviinae
base that show a difference among 20% of good identification.

75
4 Conclusion
To identify a biological object and to associate a class to it, experts usually
proceed with two phases. The synthetic phase, by global observation of the most
visible characters reduces the field of investigation. The analytical phase, by
precise observation of discriminating attributes refines research until obtaining
the result. Even if the k-nearest-neighbors approach gives a good classification
rate, it is difficult to use in real conditions without background knowledge of the
domain. In fact, it is very useful to dispose of an interactive process to design
features selection such in decision tree approaches.
The classification by successive neighborhood (CSN) method that we
proposed deals with structured and partial objects descriptions. It presents the
interest to correspond to the reasoning followed by biologists. Starting from a
partial description generally containing the most visible or easy to observe and
describe features, the method suggests relevant informa- tion necessary to
supplement to determine the most probable class.
We expect that the CSN method is generic and applicable on any fields where
structured or semi structured data are considered, such as XML data format or
RDF and OWL graph structures. It’s enough to lay out a similarity index and a
discriminant power function adapted to the considered data.

References
[1] J. E. Winston, Describing Species: Practical Taxonomic Procedure for Biologists. New York.
Columbia University Press, 1999.
[2] D. Grosser, J. Diatta and N. Conruyt, “Improving dissimilarity functions with domain knowledge”.
Proc of the 4th European Conference on Principles of Data Mining and Knowledge Discovery
(PKDD’2000), pp. 409-415, 2000.
[3] D. Grosser, H. Ralambondrainy and N. Conruyt, “Classification by successive neighborhood”.
In KDIR 2009, International Conference on Knowledge Discovery and Information Retrieval.
INSTICC Press, 2009.
[4] N. Conruyt and D. Grosser, “Knowledge engineering in environmental sciences with ikbs”. AI
Communications, The European Journal on Artificial Intelligence, 16(3), pp. 267-278, 2003.
[5] J. R. Quinlan, C4.5: Programs for Machine Learning. Morgan Kaufmann Series in Machine
Learning, 1993.
[6] R. Kohavi, “A study of cross-validation and bootstrap for accuracy estimation and model
selection”. Proceedings of the Fourteenth International Joint Conference on Artificial
Intelligence (Morgan Kaufmann, San Mateo), 2 (12), pp. 1137–1143, 1995.

76
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 77-82.
ISBN 978-88-8303-295-0. EUT, 2010.

A MediaWiki implementation of
single-access keys
Gregor Hagedorn, Bob Press, Sonia Hetzner,
Andreas Plank, Gisela Weber, Sabine von Mering,
Stefano Martellos, Pier Luigi Nimis

Abstract — Design principles, features, and data exchange options of an

open source implementation of single access keys for the collaborative
MediaWiki platform are presented.

Index Terms — field guides, flora, fauna, identification tools, KeyToNature,

social software, SDD, MediaWiki, single-access key, dichotomous key,
polytomous key.

—————————— u ——————————

1 Introduction

A
mong the various forms of computer-aided identification keys (compare
[1], [2]), single-access keys have long been neglected, being studied pri­
marily as a printable output from character × taxon matrices. However,
single-access keys may also be used interactively. Examples are Lucid Phoenix
[3] (commercial), Frida/Dryades [4] (closed source), and the two KeyToNature
open source projects: Open Key Editor [5] and biowikifarm [6] “Wiki keys”. The
latter implementation, based on JavaScript enhanced Media­Wiki [7] templates,
is presented here in detail. Its strength is the integration into the collaborative
MediaWiki software with much broader applicabi­lity for developing floras, faunas
or field guides.

2 Design principles
Much of the strength of single access keys derives from information re­duc­tion.
The information which is actually used in the key (and which must be understood
by the user) is only a fraction of the total information pre­sent in the descriptions
of the taxa.

————————————————
G. Hagedorn, A. Plank, G. Weber, S. v. Mering are with the Julius Kühn-Institute, Federal Re-
search Centre for Cultivated Plants, Inst. for Epidemiology and Pathogen Diagnostics, Königin-Lu-
ise-Str. 19, D-14195 Berlin, E-mail: [email protected]. – Bob Press is with the Natural
History Museum London – S. Hetzner is with the Inst. f. Lern-Innovation, Universität Erlangen-
Nürnberg, D-91052 Erlan­gen – S. Martellos, P. L. Nimis are with the Dept. of Life Sciences,
University Trieste, I-34127, Trieste, Italy. E-mail: [email protected], [email protected].

77
 In a perfect world with an unlimited number of characters that are con­venient
and reliable to observe, monomorphic (not variable within spe­cies), available
for observation at all times, and splitting the remaining taxa into evenly sized
partitions, the most reduced identification keys would be the best. In the real
world, however, the various imperatives for character selection are in conflict.
The resulting keys often use character combinations instead of single characters
and may provide verification characters that are not strictly necessary. Similarly,
several suboptimal illustrations may be necessary to understand character
variability.

Fig. 1 – A Wiki key [8] as part of a wiki page, presented in overview mode (“Step-by-
step identification” starts the interactive mode), and with part of the information hidden
(“more…” will display the hidden information). The key metadata (Geographic Scope
following) are initially displayed, but hideable.

Fig. 2 – A variant wiki template generates a horizontal (side-by-side) layout. Both

the horizontal and vertical layout can be used interactively and both provide for sub-
headings (“Tail yellow-white”, etc.).

78
 Fig. 3 – Extra information is initially hidden (top). It will be displayed (bottom) after
clicking on “more…”.

In the MediaWiki single-access keys, initially only terse lead statements

(which should be limited to information strictly complementary within a couplet)
and a subset of images are shown (in standard layout, Fig. 1, small previews in
a margin plus up to 2 larger images below each lead; in horizontal layout, Fig. 2,
up to 8 images above the lead statement).
As a form of “information reduction management”, further information is shown
only on request: general remarks, a gallery of up to 8 secondary illus­trations,
and – for taxa – synonyms, brief diagnosis, occurrence, distri­bu­tion, and habitat
information (Fig. 3).
All images are displayed at a reduced size, sufficient only to enable the user to
decide if enlarging is worthwhile. The zoom functionality acts in the page context
(Fig. 4). It further provides a link to creator, copyright, and license information,
required to be directly accessible by many licenses.
Information hiding is implemented through JavaScript in a graceful de­gra­
dation design (i. e. without JavaScript, no information is hidden).

Fig. 4 – Enlarged image

as an overlay to the page.
At the bottom an image
caption and a link to the IPR
infor­mation (legally required
by the Creative Commons
licen­ses) is provided.

79
 Fig. 5 – Wiki key in JavaScript-based interactive mode. The history of previous
decisions is displayed at the bottom, with decision 3 having been marked as uncertain.

Fig. 6 – Steps in the history or previous decisions are revisable (here step 4 is being
revised). Later steps are confirmable and discarded only if a conflicting decision is
taken.

3 Interactive step-by-step identification

The printable overview can be switched to an interactive mode. The open

source jKey player [9] then displays only a single choice at a time (Fig. 5),
helping esp. students to concentrate. To maintain context, a history of pre­vious
decisions is provided. Steps are revisable, maintaining pro­gress as long as
possible (Fig. 6). The history also helps to verify identifi­cations, since it can be
read as a description of the object that is being identified.

80
 Fig. 7 – Display of a glossary definition in an overlay on the web page (not in a
separate window).

Fig. 8 – After selecting the button: “Undecided: try all alternatives”, multiple alternative
paths may be followed.

4 Data exchange
The KeyToNature-Dryades/Frida system provides a special export format that
directly creates text formatted to be pasted into wiki pages as ready wiki text.
Wiki keys can further be converted to SDD xml data by a con­verter created at
the Natural History Museum in London.

81
5 Pedagogical features
The single-access keys are supported by several pedago­gi­cally relevant
features:
1. Illustrated concept definitions and help pages, stored as editable wiki pages,
may be accessed from any point in the identification key, providing context-
sensitive help. When the user hovers the mouse over a term, the defini­tion
opens in a pop-up layer (Fig. 7). From there a new window can be opened if
needed.
2. The history allows direct access to and revision of any previous deci­
sion. This may occur in a dialogue with the teacher, who can help in re­view­ing
misinterpretations.
3. Users can flag particular decisions as “uncertain” (Fig. 5). Although this does
little more than marking a step in the history, it can greatly en­hance the student-
teacher communication. It allows students to ac­tively seek teacher assistance at
a time when he or she is available, while con­tinuing their work in the meantime.
4. The interactive mode offers an option to not take a decision at a given
step, allowing users to explore the key in multiple directions. After selecting
the “Undecided: try all alternatives” button, the player will con­tinue with the first
alternative. However, the history allows the user to switch between alternative
branches, recording in all branches (Fig. 8).

Acknowledgement

This work was supported by the KeyToNature Project, ECP-2006-EDU-410019, in the

eContentplus Programme.

References
[1] R. J. Pankhurst, Practical Taxonomic Computing, 1991.
[2] G. Hagedorn, G. Rambold and S. Martellos 2010, “Types of identification keys”. In: P. L. Nimis
and R. Vignes Lebbe (eds.), Tools for Identifying Biodiversity: Progress and Problems, pp.
59-64, 2010.
[3] Lucid Phoenix (http://www.lucidcentral.org/LinkClick.aspx?link=152), 2010-07.
[4] S. Martellos and P.  L. Nimis, “KeyToNature: Teaching and Learning Biodiversity. Dryades,
the Italian Experience.” In: M. Muñoz, I. Jelìnek, F. Ferreira (eds.), Proceedings of the IASK
Interna­tional Conference Teaching and Learning 2008, pp. 863-868, 2008.
[5] S. Martellos, E. v. Spronsen, D. Seijts, N. Torrescasana Aloy, P. Schalk and P. L. Nimis, “User-
generated content in the digital identification of organisms: the KeyToNature approach” Int. J.
Information and Operations Management Education, vol. 3, 3, pp. 272-83, 2010.
[6] MediaWiki software, http://www.mediawiki.org/wiki/MediaWiki, 2010-07.
[7] G. Hagedorn, G. Weber, A. Plank, M. Giurgiu, A. Homodi, C. Veja, G. Schmidt, P. Mihnev, M.
Roujinov, D. Triebel, R. A. Morris, B. Zelazny, E. van Spronsen, P. Schalk, C. Kittl, R. Brandner,
S. Martellos and P. L. Nimis, “An online authoring and publishing platform for field guides and
identifi­ca­tion tools”. In: Nimis P. L. Nimis and R. Vignes Lebbe (eds.), Tools for Identifying
Biodiversity: Progress and Problems, pp. 13-18, 2010.
[8] B. Press, Key to common UK street trees. http://www.keytonature.eu/wiki/Key_to_common_
UK_street_trees, 2010-07.
[9] S. Opitz and G. Hagedorn, “The jKey wiki key player and builder”. Proc. of TDWG 2009, pp.
9-13 Nov. 2009, Montpellier, 2009.

82
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 83-88.
ISBN 978-88-8303-295-0. EUT, 2010.

Simple matrix keys from Excel

spreadsheets
Gregor Hagedorn, Mircea Giurgiu, Andrei Homodi

Abstract — An innovative workflow is presented, leading from a simple

character and taxon matrix, prepared in an Excel spreadsheet, to a functioning
free-access matrix key, stored in SDD (Structured Descriptive Data) format
and presented in the flash-based free-access key player IBIS-ID directly on
the web. The subset of matrix key functionality supported by this workflow
include categorical and quantitative characters, multiple character states for
a given taxon, taxon, character, and state illustrations, and a single character
grouping level. Advanced features such as complex character or taxon trees,
character dependency, or taxon-specific state images are not supported.
The workflow aims to attract new contributors for which the learning curve of
special purpose matrix-key editing software is too steep.

Index Terms — identification tools, free-access key player IBIS-ID, DELTA,

SDD.

—————————— u ——————————

1 Introduction

I
dentification tools such as free-access (= multi-access) or multi-entry keys
that are based on a character × taxon data matrix (i. e. a table with the taxa
in one dimension and characters in the other) have certain advan­tages over
single-access keys [1, 2, 3, 4]. However, creating a computer-aided matrix key
typically requires learning a special purpose application. This limits the number
of matrix-based keys produced by biolo­gists, who tend to produce keys similar
to the single-access keys typically encountered in the printed literature.
Many biologists are acquainted with spreadsheet applications, espe­cially
Microsoft Excel, to edit tabular data. Unfortunately, the visualization of a simple
character × taxon table – for which spreadsheet applications are ideal – is a
simplified idealization of a more complex data model. All of taxa, characters,
states, and the descriptive matrix cells may have further structure:
1. Taxa may require common and scientific names, web links to taxon pages,
images, brief diagnostic text, etc.
2. Characters may require a data type, a list of supported states (i.  e.
————————————————
G. Hagedorn is with the Julius Kühn-Institute, Federal Research Centre for Cultiva­ted Plants, Insti-
tute for Epidemiology and Pathogen Diagnostics, Königin-Luise-Str. 19, D-14195 Berlin, Germany,
E-mail: [email protected] – M. Giurgiu and A. Homodi are with the Telecommunica-
tions Depart­ment, Tech­nical University of Cluj-Napoca, Cluj 400027, Romania. E-mail: Mircea.
[email protected].

83
 constraining the vocabulary), illustrations, explanatory notes.
3. States may provide illustrations or explanatory notes.
The cells of the matrix may contain multiple values, modifiers, notes, and
taxon-specific state or character illustrations. For a character: “flower colour”
the cell content may be: “usually pink, sometimes red, or blue (immedi­ately after
opening)”; for a character: “stem hairiness” it may be “long (2-5 mm) or medium
long (1-2 mm) hairs”. A character having more than one state in a taxa is called
“polymorphic” in biol­ogy. It may occur as a result of a true genetic polymorphism
in a popu­lation, environmentally induced phenotypic variation (e.  g., occur­
ring within the set of flowers on a single plant), or relatively minor quan­titative
variation that happens – in the present taxon – to cross the artificially drawn
borders of a continuously varying character (such as hairiness).
However, when relatively simple rules are followed it is possible to support
a subset of the potential complexity of matrix keys within spreadsheets
nevertheless.

Fig. 1 – The workflow from a Microsoft Excel spreadsheet to SDD conversion and
presentation with the free-access key player IBIS-ID inside and automatically published
inside a MediaWiki web page.

2 The spreadsheet
The workflow (Fig. 1) starts with the creation of a character × taxon matrix
by the biologist, following either instructions on the web [5] or supported by a
downloadable template. The simplest layout is indeed one with char­acters being
named in the first row, taxa in the first column and the remain­der filled with the
taxon × character data (Fig. 2).
For categorical characters (ordinal or nominal scale) the categorical value or
state is expressed directly using its label or “name” (e.  g., “red”) rather than
using a code. In contrast, the DELTA [6] or SDD formats use numeric character
and state codes to enforce higher consistency. Multiple states are supported
by separating the state names with a semicolon, slash or ampersand (exam­
ple: “red; blue“). The semicolon is provided as an intui­tive delimiter for most

84
 Fig. 2 – A Microsoft Excel spreadsheet with a simple data matrix. Visible are an extra
column for scientific taxon names (“Wissenschaftlicher Name” in German), the addition
of a measurement unit (“[cm]”) in brackets after quantitative characters, state images in
the column “Blattrand”, and the metadata for the entire dataset or identification tool at
the bottom.

biologists; the “/” and “&” to help those who also use DELTA tools.
The drawback of the direct use of state labels is that the vocabulary of avail­
able states is not controlled. This is a purposeful design decision. While it is
possible to devise spreadsheet layouts that include separate state listings, we
have noticed that our test users found all options to be too diffi­cult and confusing
and were unable to create them autonomously. The vocabulary control is
therefore postponed to the publication of a first draft of the iden­tification tool
in the IBIS-ID player. In the implemented work­flow, the IBIS-ID key player will
make undesirable entries (com­bina­tions of states with modifiers or spelling
variants of states) trans­pa­rent and users can modify their data for the next
revision. “Normalizing” the state labels is well supported by the typical search-
and-replace func­tionality of spread­sheet software. While careful planning and
control is essential for large matrix projects covering hundreds of characters and
taxa, the work­flow presented here aims at smaller datasets, where a post-data-
entry-validation workflow may result in more agile contributions than a plan-
ahead workflow.

85
Fig. 3 – The resulting interactive matrix key running under IBIS-ID (here in stand-alone
mode, not embedded in a web page).

For quantitative characters a DELTA-like encoding is supported. Vari­ous

combinations of minimum-maximum (in parentheses), “typical range” and a
mean are possible (example: “(1-) 3-6.4-8.2 (-15)”).
The characters themselves may be grouped into character groupings by
adding the group name surrounded by “{{…}}” after the character label (i.e. in
the first row). Presently only a single grouping level is supported.
Matrix cells further support modifiers (“usually”, “rarely”, “about”, “weak­ly”,
etc.) and free-form text comments, if they are enclosed in DELTA-like “<…>”
markup.
The spreadsheet rules further provide for the inclusion of multiple taxon,
character, and state-specific images. Character and state images must be
included in double square brackets after the respective label; in contrast taxon
images may be placed in an extra column. Not supported are taxon-×-character
and taxon-×-state-specific images.
For taxa, several additional column for images, common and scientific names,
and web links can be provided. Finally, metadata for the entire key like creators,
title, copyright, license, source may be added.

86
 Fig. 4 – Details of IBIS-ID key player, showing character grouping (left) and state
images (right).

3 The converter
The converter is presently a downloadable Microsoft .NET for standalone
applications (a web-based version is planned). The converter takes the
spreadsheet in Microsoft Excel (XLS) format and converts it into SDD.
The converter supports both wiki-style and direct web image refer­en­ces.
Uploading images to the wiki allows users to manage their images for both
matrix keys, single-access keys and species pages. The simple wiki-style links
are automatically trans­lated by the converter into general web links as they are
supported by IBIS-ID key player.
If the converter finds unexpected content, it will report this either as warnings
(e.  g., “name contains opening double brackets (‘[[‘) but no clos­ing ones, a
malformed image may be present”) or as errors. Error han­dling is considered
important and efforts have been made to help biologi­cal users to understand
minor errors. If no errors are encountered, the resulting SDD file will be uploaded
to a web repository on the MediaWiki based biowikifarm [7].
Furthermore, to enrich the user experience, a wiki page containing the
necessary statement to embed the IBIS-ID player [8] (Fig. 3 and 4) inside a wiki
page is also generated.

4 Conclusions
It is possible to replace some features that Excel is missing to directly support
matrix keys with rules that rely on simple text delimiters. The method is similar
to that used by DELTA the special purpose Win­dows DELTA editor software.

87
However, the point of the workflow presented is to provide a simple functionality
in an environment well known to most biolo­gist, in order to attract new biologists
and educators and increase the pro­duction of matrix-based identification tools.
Although advanced rules may require some learning effort, it is possible to
create useful matrix keys not using these features.

Acknowledgement

This work was supported by the KeyToNature Project, ECP-2006-EDU-410019, in the

eContentplus Programme.

References
[1] R. J. Pankhurst, Practical Taxonomic Computing, 1991.
[2] J. Winston, Describing Species. Columbia University Press,1991.
[3] G. Hagedorn, Structuring Descriptive Data of Organisms - Requirement Analysis and
Information Models. Ph. D. Thesis, Universität Bayreuth, 2007.
[4] G. Hagedorn, G. Rambold and S. Martellos, “Types of identification keys”. In: P. L. Nimis and
R. Vignes Lebbe (eds.), Tools for Identifying Biodiversity: Progress and Problems, pp. 59-64,
2010.
[5] G. Hagedorn et al., The Excel to SDD converter, 2010. http://www.keytonature.eu/wiki/Excel_
to_SDD_converter, 2010-07.
[6] DELTA – DEscription Language for TAxonomy http://delta-intkey.com/, 2010-07.
[7] G. Hagedorn, G. Weber, A. Plank, M. Giurgiu, A. Homodi, C. Veja, G. Schmidt, P. Mihnev,
M. Roujinov, D. Triebel, R. A. Morris, B. Zelazny, E. van Spronsen, P. Schalk, C. Kittl, R.
Brandner, S. Martellos and P. L. Nimis, “An online authoring and publishing platform for field
guides and identification tools”. In: P. L. Nimis and R. Vignes Lebbe (eds.), Tools for Identifying
Biodiversity: Progress and Problems, pp. 13-18, 2010.
[8] M. Giurgiu, G. Hagedorn and A. Homodi, “IBIS-ID, an Adobe FLEX based identification tool
for SDD-encoded multi-access keys”. Proc. of TDWG 2009, 9-13 Nov. 2009, Montpellier, p.
90, 2009.

88
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 89-93.
ISBN 978-88-8303-295-0. EUT, 2010.

Wiki keys on mobile devices

Gisela Weber, Gregor Hagedorn

Abstract — The development of increasingly powerful mobile devices like

PDAs (Personal Digital Assistants) and Smartphones, with larger displays
and greater resolution makes them increasingly suitable for identification
tools available directly “in the field”. One of several approaches towards this
aim in the KeyToNature project is based on wiki-stored documents. Important
features of wiki-based keys, such as hidden text and media information as
well as links to glossary entries are supported. The illustrated keys can be
used online or downloaded in a zip file. An extension to support the application
stores of various mobile platforms (Android, Apple iPhones, etc.) is under
development.

Index Terms — mobile devices, smartphones, MediaWiki, identification

software Tools.

—————————— u ——————————

1 Introduction

D
igital identification keys traditionally are built to be used on CD-ROM
or over the internet. They have some advantage over printed books
in the ease of including a rich selection of colour illustrations, in the
speed of updating, and in certain advantages of interactive use, which are espe­
cially relevant in pedagogical scenarios. However, they are suitable prima­rily for
indoor workplaces, but usually not practical in the field. With the development
of more powerful mobile devices with higher resolution it becomes worthwhile
to create digital keys for mobile devices. Such keys must take into account
the specific requirements of a small display screen and cumbersome typing to
achieve a user-friendly design. In the KeyToNature project, several approaches
towards this aim have been realized in order to compare different approaches
(e.g. [1], [2], [3]).
The application described here is based on the key authoring abilities
of the MediaWiki platform described separately [4]. The Wiki is a docu­ment
storage and authoring platform that allows to embed structured infor­mation
inside unstructured documents (called “templates”). Based on these structured
elements, and taken the freedom of authors to develop new solutions into
account, the web-based identification tools are trans­formed into mobile keys.
These can be used online or downloaded pack­aged into a zip file that can be
transferred to mobile devices for offline use.
————————————————
The authors are with the Julius Kühn-Institute, Federal Research Centre for Cultivated Plants,
Institute for Epidemiology and Pathogen Diagnostics, Königin-Luise-Straße 19, D-14195 Berlin.
E-mail: [email protected].

89
 Fig. 1 – Wiki key to common UK street trees, printable overview with interactive mode
(chosen by upper right “Step-by-step identification” link) shown as overlay in the bottom
right.

2 Wiki keys
In the KeyToNature project, one approach to enable users to create and edit
their own identification keys is the MediaWiki platform. Users can create online
single access keys (i.e. dichotomous or polytomous keys) which include images
and additional information [5]. These keys can be viewed in a printable overview
mode and also interactively in a one-couplet-at-a-time-mode (Fig. 1). Editing
of keys occurs online. Special features of the wiki keys are up to 5 images per
lead in the right-hand side bar, an extra 2 images below the lead statement,
and 6 further images, plus extra text (description, remarks, occurrence) which
is initially hidden and requires user interaction to be shown. This principle of
showing sec­on­dary information only on demand is also used in the display of
illustrated term definition (glossary) directly where they are used in the lead
statements, and in providing additional information, including legally required
IPR and licensing information on the images. All images are zoom­able to the
maximum possible extent of the source image and dis­play device.

3 Wiki keys on mobile devices

For the transformation of the identifi­ca­tion keys from Wiki pages to down­load­
able HTML-pages optimized for mobile devices we have written a soft­ware that
is installable as an ex­ten­s­ion for MediaWiki. This soft­ware, called “MobileKey”
requires another extension “Template­Parame­ter­Index”, also written in the
KeyToNature pro­ject, which harvests name-value pairs from template calls
on Wiki pages and stores them in a database table. The MobileKey extension
uses this index to directly access data behind the Wiki keys. Both extensions of

90
 Fig 2 – Couplet with two alternatives (a key to birds based on their sounds).
MediaWiki will be made available as Open Source.
The mobile key extension adds a “Special Page” that allows to create a
mobile key that starts at any selected Wiki page with an identification key. The
extension:
• recognizes a key on the page and formats the metadata of the key into a
start screen
• aggregates the leads that belong to the same couplet;
• splits the key into couplets (= decision), with each couplet being ren­dered
on its own HTML-page in a layout suitable for the small display of mobile
devices (Fig. 2);
• puts the additional information on extra HTML pages (Fig. 3, 4);
• puts glossary text on extra HTML pages (Fig. 5);
• stores these files that have been optimized for mobile-devices on the server;
• stores the images on the server;
• replaces the existing links on the Wiki page with the appropriate local links;
• packs all pages and images into a zip file to be downloaded.

Fig. 3 – Page with additional information and two images.

The HTML is designed to adjust to some extent to the screen size and land­
scape versus portrait orientation (Fig. 3, 4). As a mechanism to as­sess the

91
display on various devices, the MobileKey Special Page provides two iFrames
with different sizes (240 x 320px and 480 x 320px) in which the mobile key
can be viewed. Images can be displayed side by side or one below the other
according to the display width. Importance is given to good readability of the
texts and clear structuring of the displayed page.
The information for a single decision is often longer than the viewport of the
mobile device, requiring the user to scroll. At the top of each page, information
on how many alternatives are available in the couplet is given. Also the links to
go back to the previous couplet or to return from any couplet to the start of the
key or a subkey (e.g. for species of a genus) are given there. This information
is repeated at the bottom of the page so that the user does not have to scroll all
the way up again.

Fig. 4 – Same as Fig. 3, but in landscape orientation.

The bars which contain that information above and below the text are given
different colours, supporting the users intuition as to whether a key couplet,
a page with additional information or a glossary page is dis­played. Only the
couplet pages allow navigation within the key or to a sub­key, whereas the extra
information and glossary pages only offer a link back to the page from which
they were called

Fig. 5 – Page with glossary links (left, “Unterlippe”, “Schlund”, coloured) and glossary page (right).

92
4 Outlook
The application for mobile keys is still under development. At the moment, one
still has to manually download the zip file to a PC, unzip it, copy the folders to
the mobile device’s SD card using an USB cable, and manually point the mobile
browser to it. It is clear that on mobile devices that sup­port this (especially
Android and Apple iPhone), it would be desirable to wrap the identification tools
into downloadable mobile apps. In fact, this is the only option iPhones provide.

5 Conclusion
The challenge of developing identification keys for mobile devices is be­coming
more and more promising with the evolution of better devices. The MediaWiki
technology appears to be a good platform to combine user input to create and
edit keys with the possibility to make existing keys usable on mobile devices.

Acknowledgement

This work has been supported by the KeyToNature project, ECP-2006-EDU-410019,

in the frame of the eContentplus Programme.

References
[1] P. L. Nimis and S. Martellos, “Progetto Dryades”, http://www.dryades.eu/home1.html, 2008.
[2] S. Martellos, E. v. Spronsen, D. Seijts, N. Torrescasana Aloy, P. Schalk and P. L. Nimis,
“User-Generated Content in the Digital Identification of Organisms: the KeyToNature Approach”.
International Journal of Information and Operations Management Education (IJIOME) vol. 3,
3, pp. 272–283, 2010.
[3] E. v. Spronsen, S. Martellos, D. Seijts, P. Schalk and P.L. Nimis, “Modifiable digital identification
keys”. In: P. L. Nimis and R. Vignes Lebbe (eds.), Tools for Identifying Biodiversity: Progress
and Problems, pp. 127-131, 2010.
[4] G. Hagedorn, G. Weber, A. Plank, M. Giurgiu, C. Veja, G. Schmidt, P. Mihnev, M. Roujinov, D.
Triebel, B. Zelazny, E. v. Spronsen, P. Schalk, C. Kittl, R. Brandner, S. Martellos, P. L. Nimis,
“An online authoring and publishing platform for field guides and identification tools”. In: P. L.
Nimis and R. Vignes Lebbe (eds.), Tools for Identifying Biodiversity: Progress and Problems,
pp. 13-18, 2010.
[5] G. Hagedorn, B. Press, S. Hetzner, A. Plank, G. Weber, S. v. Mering, S. Martellos, P. L. Nimis,
“A MediaWiki implementation of single-access keys”. In: P. L. Nimis and R. Vignes Lebbe
(eds.), Tools for Identifying Biodiversity: Progress and Problems, pp. 77-82, 2010.

93
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 95-98.
ISBN 978-88-8303-295-0. EUT, 2010.

A Wiki-based Key to Garden

and Village Birds
Tomi Trilar

Abstract — A Wiki-based Key to Garden and Village Birds is available in two

versions: a dichotomous, hyperlinked and printable version, and as step-by-
step identification version. It is supported by jKey Player in English, Slovenian,
Spanish, Romanian and German.

Index Terms — birds, Aves, key, identification, wiki, jKey Player.

—————————— u ——————————

1 Introduction

O
one of the first steps in discovering and understanding biodiversity is
to identify the organisms around us. The study of biodiversity, and in
particular the identification of organisms, is becoming part of educational
curricula in primary and secondary schools and in many University courses
across Europe.
After Gutenberg, information useful for identifying organisms was printed
on paper. The constraints of a paper-printed text have forced most authors
to organise information according to the hierarchical scheme of biological
classification. Computer-assisted keys, on the contrary, allow us to identify
organisms without necessarily using the characters of the classification in
the biological system. Classification and identification belong to two different
operational processes. Classification is the job of taxonomists, identification can
be fun for anybody [1].
There are many identification tools available online, based on different
platforms. One of the possible platforms is the Wiki. The advantage of wiki-
based keys is that a community of users can work together in constructing,
improving and enriching them in a collaborative way.
In the European project KeyToNature, two tools for creating and running wiki-
based keys were developed: the jKey Editor and the jKey Player. The JavaScript
based jKey Player is supporting interactive use, tracking and annotating previous
decisions, and allows all decision steps to be easily revisable. The revision
capability is especially useful in classroom use, allowing the quick identification
of erroneous decisions, discussing the reasons for these, and continuing without

————————————————
The author is with the Slovenian Museum of Natural History, Prešernova 20, P.O.Box 290, 1001
Ljubljana, Slovenia. E-mail: [email protected].

95
restarting the entire identification process. Additional functionalities like toggling
the display of secondary text and images, and enlarging images on click, make
the keys more user-friendly. The complementary web-based editing tool jKey
Editor allows a form-based editing of the identification keys, which simplifies
creating, modifying, pedagogically adapting, or translating existing keys [2].

2 Key to Garden and Village Birds

The Key to Garden and Village Birds is a wiki-based identification tool,
which includes text, pictures, drawings and sounds [4]. Original recordings of
bird sounds are also displayed, which were stored in the Slovenian Wildlife
Sound Archive, housed in the Slovenian Museum of Natural History (PMSL) in
Ljubljana, Slovenia [3].
The key can be used both in a dichotomous hyperlinked printable version
(Fig. 1) or in a step-by-step identification version (Fig. 2) supported by jKey
Player. The target audience are pupils in primary and secondary schools, with
the pedagogical goal to observe the birds and listen their singing and calling.
Important was the contribution of the wiki-community, who translated the key
into English [4], Slovenian [5], Spanish [6], Romanian [7] and German [8].

Fig. 1 – Dichotomous hyperlinked printable version of the wiki-based Key to Garden

and Village Birds [4].

96
Fig. 2 – Step-by-step identification version of the wiki-based Key to Garden and Village
Birds with an annotated step of uncertainty [4].

Acknowledgement

I would like to thank Gregor Hagedorn for encouraging me to create the key and for his
help in formatting and dapting it, and Bob Press for style and language improvements. I
would like to also thank Marina Ferrer Canal, Mircea Giurgiu, Gregor Hagedorn and Irena
Kodele Krašna, who organised and helped with the translation of the key into Spanish,
Romanian, Slovenian and German. This work was supported by the KeyToNature
Project, ECP-2006-EDU-410019, in the eContentplus Programme.

References
[1] P. L. Nimis, Keys to the Lichens of Italy. Ed. Goliardiche, Trieste, 341 pp., 2004.
[2] G. Hagedorn and S. Opitz, “JKey Player”, KeyToNature. http://www.keytonature.eu/wiki/JKey_
Player, 2010.
[3] “Slovenian Wildlife Sound Archive”, Slovenian Museum of Natural History. http://www2.pms-lj.
si/staff/bioacoustics/bioacoustics.html, 2010.
[4] “Key to Garden and Village Birds”, KeyToNature. http://www.keytonature.eu/wiki/Key_to_
Garden_and_Village_Birds, 2010.
[5] “Ključ za določanje vrtnih ptic”, KeyToNature. http://www.keytonature.eu/wiki/Ključ_za_
določanje_vrtnih_ptic, 2010.

97
[6] “Clave de Aves Comunes de Jardines y Areas Rurales de España”, KeyToNature. http://www.
keytonature.eu/wiki/Clave_de_Aves_Comunes_de_Jardines_y_ Areas_Rurales_de_España,
2010.
[7] “Cheie de identificare a pasarilor comune care traiesc in zone rurale din Europa”, KeyToNature.
http://www.keytonature.eu/wiki/Cheie_de_identificare_a_ pasarilor_comune_care_traiesc_in_
zone_rurale_din_Europa, 2010.
[8] “Häufige Vögel in Gärten und Siedlungen”, Offene-Naturführer. http://www.offene-naturfuehrer.
de/wiki/Häufige_Vögel_in_Gärten_und_Siedlungen, 2010

98
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 99-105.
ISBN 978-88-8303-295-0. EUT, 2010.

Wiki-keys for the ferns of the

Flora of Equatorial Guinea
Francisco Cabezas, Carlos Aedo, Patricia Barberá,
Manuel De la Estrella, Maximiliano Fero, Mauricio Velayos

Abstract — Flora of Equatorial Guinea is a research project coordinated by

the Real Jardín Botánico of Madrid with the aim to produce a modern Flora for
this almost unknown territory. One of our goals is to develop the website www.
floradeguinea.com. A system of relational databases with an implemented
Thesaurus - to minimize typographical mistakes - allows the on-line managing
and updating of nomenclatural information, herbarium specimens, literature
records, vernacular names and bibliography. Images and maps are also
linked to accepted names. We are now implementing interactive keys for
identification. In our wiki-based system, Scans of herbarium specimens from
all species growing in the country are also being uploaded. Species from
neighboring countries are included as well, since some of them could appear
in Equatorial Guinea.

Index Terms — Africa, Equatorial Guinea, e-Flora, ferns, online-databases,

Wiki key.

—————————— u ——————————

1 Introduction

F
lora of Equatorial Guinea is a research project undertaken by the Real
Jardín Botánico de Madrid-CSIC, with the collaboration of the Spanish
Universities of Salamanca and Córdoba, the Kew Royal Botanic Gardens,
the Nationaal Herbarium Nederland and the Université Libre de Bruxelles. The
project is currently funded by the Spanish administration (reference project
CGL2009-07405). The final aim of our project is to produce a modern flora of
this almost unknown territory, a historic goal of the Spanish botany, rooted in
former times, when these tropical regions were part of the overseas provinces
of Spain, more than sixty years ago.

The interest on this region also derives from its important biodiversity. Tab. 1
shows the surface of humid rain-forests in some countries of Central Africa: only
Gabon has a higher percentage of the territory covered by humid, undisturbed
————————————————
F. Cabezas is with the Department of Botany, Faculty of Biology, University of Salamanca. Méndez
Nieto Av. s/n, 37007 Salamanca, Spain. E-mail: [email protected].
C. Aedo, P. Barberá, M. Fero and M. Velayos are with the Real Jardín Botánico de Madrid, CSIC.
Murillo Sq. 2 28014, Madrid, Spain.
M. Estrella is with the Botany, Ecology and Plant Physiology Department, C-4, Celestino Mutis,
Campus de Rabanales, 14071, Córdoba, Spain.

99
rain-forests in the Guineo-Congolian region, the most biodiverse area in the
mainland of tropical Africa.

Original tropical Tropical rainforest % of tropical rainforest

rainforest surface surface nowdays preserved
(km2) (km2)

Cameroon 376.900 155.330 41,2

Equatorial
26.000 17.004 65,4
Guinea

Central African
324.500 52.236 16,1
Republic

Rwanda 9.400 1.554 16,5

Nigeria 421.000 38.620 9,2

Gabon 258.000 227.500 88,2

Tab. 1 – Surface of humid rain-forest in some countries of Central Africa.

Our interest was also increased by the dissimilar floristic knowledge of the
different regions of the country, reflecting once more the complicated history of
the territories of Equatorial Guinea and Spain since Emilio Guinea’s first trip,
including questions beyond science, for example, such as the independence of
Equatorial Guinea from Spain in 1968.
One of the goals of the planned modern Flora of Equatorial Guinea is the
development of our website, www.floradeguinea.com, where new identifications
are updated immediately. Any specialist working on the flora of Africa can freely
check our results.
Presently, we are determined to go one step forward, implementing in our site
a on-line interactive system with identification keys. In our wiki-based system,
scans of herbarium specimens from all species growing in the country are also
uploaded. Species from Gabon, Cameroon and S. Tomé & Príncipe are included
as well: some of them could appear in Equatorial Guinea, although they have
not been collected yet.

2 Materials and methods

We have planned our task in 4 different steps: 1) compilation of all literature
records, 2) identification of all the available specimens, 3) compilation of
checklists of taxa and specimens growing in the country, 4) compilation of a
Flora, with keys for identification, descriptions, images and maps.
In order to reach the first two goals, we have designed a system of relational
databases with several Thesauri, implemented to minimize typographical
mistakes. Nomenclatural information, specimens data, literature records,
vernacular names and bibliography can be managed and updated on-line.
Digital images and maps with the distribution in Equatorial Guinea are also

100
linked to accepted names.
The second two goals (checklist and Flora) demanded a re-designing of
the whole database structure. We used a wiki-template designed by Gregor
Hagedorn and used by KeyToNature and other projects in the Real Jardín
Botánico-CSIC, in order to translate our printed keys into an interactive system
where updates done by users or editors can be automatically implemented. This
also permits to include more information in the e-keys, such as e.g. scans of
herbarium specimens.
The final output of the keys will be linked to the www.floradeguinea.com
site, where, following the taxon name, the user can find revised nomenclatural
information, a description, a list of identified herbarium specimens, a list of
literature reports, digital images of herbarium specimens, links to pictures of
living plants, and a skecth map where the distribution of the taxon in Equatorial
Guinea is presented.

3 Results and Discussion

3.1 Results of the Flora of Equatorial Guinea

The gathering of literature produced a great stepping stone with the publication
of the documentary databases for the Flora of Equatorial Guinea [1]. In this
book, all the reports of mosses, fungi, ferns, mono- and dicotyledons from
Equatorial Guinea were compiled and databased, reflecting the information as
it was originally published. Widely distributed taxa from neighboring countries
were also included, since they will appear in the keys for identification. Currently,
this database is still growing and is managed completely on-line, since it needs
to be updated with any new publication on floristics for any of the territories of
Equatorial Guinea. We also go on compiling and including the names published
in São Tomé and Príncipe Island, Cameroon, and Gabon. Today, 52,301 records
of vascular plants are included.
The second task (collection effort), has produced as a main result a set of
more than 15,000 collection numbers from Equatorial Guinea in the Madrid
herbarium, with an average of 4 duplicates: this is now the main collection for
the country. In our database we also include the vouchers data, especially from
historical collections kept at K, BM, or in the Netherlands in the WAG herbarium,
as well as those kept in Equatorial Guinea. For comparative purposes, some
collections from neighboring countries were also studied. In this aspect,
Missouri and Portuguese herbaria were essential, and are now databased as
well. Until now, 16,615 herbarium specimens from Equatorial Guinea have been
databased. Among them, about 2000 were studied carefully and assigned to a
correct name.
The lack of floristic knowledge in Africa and, of course, in Equatorial Guinea,
brought us to publish critical checklists before the Flora, splitting the original
idea of Emilio Guinea. The main reason is clear: if we’ll have waited until the
information is complete to publish the Flora, and the identification keys and
descriptions are made, most of the species included in the work could be extinct.

101
These checklists, on the contrary, provide a useful tool to start with conservation
programs and strategies. The latter are especially necessary nowadays,
considering that 13 million ha of primary forests are destroyed every year. This
step was the one with more results in the last years.

FWTA Fl. GUI Increase Nrec Country

(%)

Pteridophytes 163 226 38 8

Cyperaceae 28 96 231 22

Marantaceae 4 26 271 8

Piperaceae 9 13 44 _

Mimosoideae 9 40 344 14

Caesalpinioideae 23 124 525 45

Papilionoideae 121 157 30 48

Ebenaceae 1 28 2700 12

Melastomataceae 18 57 216 26

Commelinaceae 24 45 114 11

Tab. 2 – Numerical summary of families with checklist published since 2001. FWTA=
Number of species mentioned in Equatorial Guinea in the Flora of West Tropical Africa.
Fl.GUI=Number of species found in our study. Increase = percentage increase of our
catalogue compared with the data of FWTA. Nrec Country = Number of species found
in our catalogues not previously reported from Equatorial Guinea. Families in boldface
include new names or species described from material collected in Equatorial Guinea.

Tab. 2 shows the numerical summary of families where checklists were

published since 2001. We have published the accounts of 24 families of vascular
plants and all Pteridophytes. Our main result was that floristic knowledge of
the country increased by about 120 %. Applying this value to the estimation of
Davis [2] of 3,250 species growing in Equatorial Guinea, using data relative to
the Flora of West tropical Africa, the number of species estimated to live in the
country jumps to c. 7,100, a figure similar to those of Cameroon and Gabon.
The final task is that of writing and editing the Flora. Until now, we have
published two volumes of the flora, Pteridophytes and Fabaceae, with 227
and 320 species, respectively. These figures include also citations of species
without herbarium support, whose distribution range and habitat render their
presence in Equatorial Guinea quite possible. Moreover, several monographic
treatments are being undertaken on some genera, which will probably result in
the reinterpretation of several taxa. Once the first reference is published, it is
easier to compare any new collection, so that we would be not surprised if these
numbers will be changing soon.

102
3.2 Results in the website

The development of a new system of databases, based on MySQL allows

the on-line handling of information related to herbarium specimens, literature
reports, and also of those data related to accepted names, bibliography, type
information or descriptions. We have included in the website all the information
that we have produced in the last years, so that now all papers produced by us
or by other authors but related to the country are freely downloadable [3-18].
The next step is to produce and update all the stages of the editorial process
on-line. We have implemented the printed version of the keys, and now we are
presenting interactive keys as well [http://www.keytonature.eu/wiki/Clave_de_
familias_de_Pteridophyta_de_la_Flora_de_Guinea_Ecuatorial_%28RJB%29].
These keys are based on the wiki-structure design developed by Gregor
Hagedorn within the European project KeyToNature, which have been used
also by other projects in the Real Jardín Botánico, as the Gymnosperms key
[“Clave de Gimnospermas (RJB)”].
The printed keys of the published volumes of the Flora are being translated to
this wiki system in a relatively easy way. Until now, we have migrated all the keys
for identification of ferns. In this first step the keys are in Spanish, the mother
language in Equatorial Guinea, but English versions will be also prepared. We
have material for starting with the translation of the key for families of vascular
plants and of all the keys produced for the volume of legumes. At the end all the
keys included in the Flora of Equatorial Guinea will be produced, updated and
corrected on-line, and in the future from the on-line version a printed version will
be published. This system allows to advance the complicated editorial process
of a work like a Flora without the constraints of printed texts.
In our keys, all species growing in Equatorial Guinea are included, and a
digital picture of an herbarium specimen is included close to each name.
Each result in the key (family, genus or species name) is linked to a taxon
page at the www.floradeguinea.com site, which has information related to
nomenclature, specimens, literature records and bibliography or vernacular
names. Furthermore, each accepted name is linked to several sites where
more information can be consulted, such as the African Plants Database, by the
Conservatoire et Jardin Botanique of Geneva (http://www.ville-ge.ch/musinfo/
bd/cjb/africa/recherche.php), where accepted names for the most important
african floras are included and critically presented, or the West African Plants
Database developed by the Senckenberg Forschungsinstitut und Naturmuseum
and by the Frankfurt University (http://www.westafricanplants.senckenberg.de/
root/index.php), where, among other information, pictures of living plants can
be examined.

4 Conclusions
High speed degradation of natural resources and the need of urgent
conservation decisions have increased the value of Floras as the base to study,
understand and preserve plant biodiversity.
Despite their high relevance, many Floras remain incomplete and progress

103
slowly, due to the large number of species involved, the highly distributed
and dissimilar data, and the lack of tools promoting effective collaboration.
Thus, is common to find both personal and duplicated efforts between remote
researchers.
Nowdays, in order to make the information contained in old floristic works
accesible, Floras only available as hard copies are increasingly being digitized.
Most of the initiatives have produced scans or fixed images of the original printed
version. With the use of new technologies, printed versions are not the unique
result of a flora, they appear to be just one of the possibilities.
This can be overcome by making full use of current information technology to
draw the highly distributed data together and to allow the taxonomic research
community to communicate efficiently. The development of an e-way of handling
Floras will change traditional work-flow processes by fostering a collaborative
setting, strengthening existing research networks, and making plant biodiversity
information rapidly and widely accessible in a re-usable format.
Floras provide keys for the identification of plant species and additional
information on each species such as synonymy, economic uses, geographical
distribution and ecology. With the possibility of e-handling, floristic research will
be optimized by the producers of Floras, and the consumers of information will
increase. Potential users will range from traditional readership to ecologists
interested in morphological traits, climate modellers or policy makers interested
in distribution data.
The benefits of the wiki-keys will be:
1. More efficient production of floras, especially the keys for identification.
Keys will be more interactive and easily updated, making a more dynamic and
updated floristic resource in a collaborative system.
2. The output information of the work is higher and easier to access. The
impact of floristic data will be greatly increased. This is providing novel uses of
floristic data. The availability of digital floristic data opens opportunities for uses
as i.e. datasets for modelling and web services connecting to other websites
and databases.
3. The wiki-key system also promotes new collaboration mechanisms for
taxonomists in order to significantly improve local and remote co-working and
to eliminate redundancy of work within this scattered community. Collaborative
research will foster the transfering of knowledge to the benefit of early-stage
researchers, particularly in tropical and developing countries.

Acknowledgement

The authors wish to thank to the Ministry of Science and Innovation of Spain for the
support in the next years. We are also indebted to the Spanish Superior Research
Council, especially the Department of Publications. The authorities of Equatorial Guinea,
the people responsible of BATA herbarium and the National University of Equatorial
Guinea (UNGE) deserve special mention. The vouchers scanned and used are mainly
from our collection in MA. Nevertheless, as can be inferred from some headings, some
species were obtained from abroad institutions as: Botanischer Garten und Botanisches
Museum of Berlín (B), Botanische Staatssammlung München (M), Université Libre de
Bruxelles (BRLU), Royal Botanic Gardens, Kew (K), Natural History Museum of London

104
(BM), National Botanic Garden of Belgium (BR), Muséum National d’Histoire Naturelle,
Paris (P), Wageningen University (WAG) y Cameroon National Herbarium, Yaounde
(YA). All of them are thanked for the permissions to use their herbarium scans.

References

[1] C. Aedo, M. Velayos and M. T. Tellería (eds.), Bases documentales para la Flora de Guinea
Ecuatorial. Plantas vasculares y hongos. Madrid, Consejo Superior de Investigaciones
Científicas & Agencia Española de Cooperación Internacional, 414 pp., 1999.
[2] S. D. Davis, V. H. Heywood and A. C. Hamilton (eds.), Centres of Plant Diversity, A Guide
and Strategy for their Conservation, Europe, Africa, South West Asia and the Middle East.
Cambridge: World Wide Fund for Nature (WWF) and The World Conservation Union (IUCN),
vol. 1, 1994.
[3] M. Fero, F. Cabezas, C. Aedo and M. Velayos, “Checklist of the Piperaceae of Equatorial
Guinea”, Anales Jard. Bot. Madrid, vol. 60(1), pp. 45-60, 2003.
[4] I. Parmentier and D. Geerinck, “Checklist of the Melastomataceae of Equatorial Guinea”,
Anales Jard. Bot. Madrid, vol. 60(2), pp. 331-346, 2003.
[5] F. Cabezas, C. Aedo and M. Velayos, “Checklist of the Cyperaceae of Equatorial Guinea
(Annobón, Bioko and Río Muni), Belg. J. Bot., vol. 137(1), pp. 3-26, 2004.
[6] F. Cabezas, M. Estrella, C. Aedo and M. Velayos, “Marantaceae of Equatorial Guinea”, Ann.
Bot. Fennici, vol. 42(3), pp. 173-184, 2005.
[7] B. Senterre, “Checklist of the Ebenaceae of Equatorial Guinea”, Anales Jard. Bot. Madrid, vol.
62(1), pp. 53-63, 2005.
[8] M. Estrella, F. Cabezas, C. Aedo and M. Velayos, “Checklist of the Mimosoideae of Equatorial
Guinea”, Belg. J. Bot, vol. 138(1), pp. 11-23, 2005.
[9] M. Estrella, F. Cabezas, C. Aedo and M. Velayos, “Checklist of the Caesalpinioideae
(Leguminosae) of Equatorial Guinea (Annobón, Bioko and Río Muni)”, Bot. J. Linn. Soc., vol.
151, pp. 541-562, 2006.
[10] A. P. Davis and E. Figueiredo, “A checklist of the Rubiaceae (coffee family) of Bioko and
Annobon (Equatorial Guinea, Gulf of Guinea)”, Syst. Biodivers., vol. 5(2), pp. 159-186, 2007.
[11] F. Cabezas, M. Estrella, C. Aedo and M. Velayos, “Checklist of the Commelinaceae of
Equatorial Guinea (Annobón, Bioko and Río Muni)”, Bot. J. Linn. Soc., vol. 159, pp. 106-122,
2009.
[12] P. Jiménez-Mejías and F. Cabezas, “Schoenoplectus heptangularis Cabezas & Jiménez
Mejías (Cyperaceae), a new species from Equatorial Guinea”, Candollea, vol. 64, pp. 101-
115, 2009.
[13] M. Estrella, F. Cabezas, C. Aedo and M. Velayos, “The Papilionoideae (Leguminosae) of
Equatorial Guinea (Annobón, Bioko and Río Muni)”, Folia Geobot., vol. 45, pp. 1-57, 2010.
[14] M. E. Leal, “Novitates Rio Munis 1. A new endemic Scaphopetalum (Malvaceae) from Mount
Mitra, Equatorial Guinea”, Blumea, vol. 52, pp. 137-138, 2007.
[15] E. Figueiredo, A. Gascoigne and J. P. Roux, “New records of Pteridophytes from Annobón
Island”, Bothalia, vol. 39,2, pp. 213-216, 2009.
[16] M. S. M. Sosef and N.S. Nguema Miyono, “Novitates Rio Munis 2. A new species of Begonia
section Loasibegonia (Begoniaceae) from the Monte Alen region, Equatorial Guinea”, Blumea,
vol. 55, pp. 91-93, 2010.
[17] M. Estrella, C. Aedo, B. Mackinder and M. Velayos, “Taxonomic Revision of Daniellia
(Leguminosae: Caesalpinioideae)”, Syst. Bot., vol. 35(2), pp. 296-324, 2010.
[18] T. Stévart, V. Cawoy, T. Damen and V. Droissart, “Taxonomy of Atlantic Central African Orchids
1. A New Species of Angraecum sect. Pectinaria (Orchidaceae) from Gabon and Equatorial
Guinea”, Syst. Bot., vol. 35(2), pp. 252-256, 2010.

105
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 107-112.
ISBN 978-88-8303-295-0. EUT, 2010.

MyKey: a server-side software

to create customized decision
trees
David Gérard, Régine Vignes Lebbe

Abstract — To facilitate the identification of specimens, biodiversity informatics

has developed numerous new computer-aided tools which compete with the
old, printed single access keys. Free access keys are accessible for many
different taxonomic groups, but key-generating software are also helpful to
construct single access keys. This paper presents a midfield solution to create
customized decision trees (single access keys) through an interactive web-
based interface. This solution offers an online service to create keys according
to the parameters and context chosen by the final users themselves. It is also
useful for the administrator of this online system, because of low-maintenance
needs, limited to configuration files when new knowledge bases are added on
the server side. Presently, the software is available with a French interface at
the following URL: http://baron.snv.jussieu.fr/cgi-bin/david/MyKey.cgi.

Index Terms — polytomous key, customized key, decision tree, web-based

interface.

—————————— u ——————————

1 Introduction

A
ccessing relevant and critical taxonomic information is often a privilege
for the specialists [1], who can take profit of natural history collections,
taxonomic monographs or low-circulation journals. Lawyers, border
guards, epidemiologists, as well as ecologists or any other biologists, may
also have identification requirements. For their needs, printed dichotomous or
polytomous keys are often included in monographs, floras and faunas, and in
practical field guides. A key has a graph structure, comparable to a decision
tree of Artificial Intelligence [2] (in this paper we will use as synonyms the terms
single access key, and decision tree).
A negative property of classical keys is their static nature: if you cannot answer
to a question (for example if you have no flower and characters of the flowers
are frequently used in botanical keys), the key is useless. Moreover, to create a
————————————————
D. Gérard was student in the Laboratoire Informatique et Systématique, University of Paris 6,
UMR7207 (MNHN, CNRS, UPMC), CP 48, 57 rue Cuvier 75005 Paris, France. E-mail: dagerard@
gmail.com.
R. Vignes Lebbe is with the University of Paris 6, UMR7207 (MNHN, CNRS, UPMC), CP 48, 57
rue Cuvier 75005 Paris, France. E-mail: [email protected].

107
key is a time-consuming task and each taxonomist adapts his/her key to a given
context. But it could be necessary to offer different keys to different user groups
(e.g. an autumn key to trees based only on trunk and leaves characters, a key
based on fruits, a key limited to a geographic area, a key taking into account
immature stages, etc.).
In the late 1960s, biologists [3], [4] began to use computers to produce more
flexible free access keys or computer-aided-identification (CAI) systems [5].
Since the 1980s, knowledge bases formats for structuring descriptive data
appeared, like the DELTA format [6], and user-friendly software (e.g. IntKey [7]
and XPER [8]) were implemented for creating knowledge bases and enabling
CAI. Storage of data became easier, as well as the retrieving of specific
information in a pool of data [9]. The reader can find a good report comparing
these tools in [10], [11], on the DELTA website, and on the BD tracker of the
European project EDIT [http://www.e-taxonomy.eu/].
Causse & Lebbe [12] have demonstrated the strong similarity between CAI
and single access keys, and their common elimination procedure. These
authors introduce the idea of a unique system able to propose identification from
a free access key to a single access key by continuously improving the strategy
advertisement expressed by the taxonomist.
To adapt single access keys to the users, one finds different proposals (see
for example [18] and [19] in this book). This paper offers another solution: it
combines a program to compute automatically single access keys and a web
interface for the final user to define himself the input parameters of the key
constructor. This original prototype, MyKey, is a server-based program. It uses
knowledge bases stored with the XPER system and the key generator MAKEY
[13]. Running on a server of the Laboratoire d’Informatique et Systématique
of the University Paris 6, it is available at the following URL (http://baron.snv.
jussieu.fr/cgi-bin/david/MyKey.cgi).

2 The tools XPER and MAKEY

XPER and its current version Xper² 2.1 is a complete software package for
managing knowledge bases [14]. It provides tools for structuring and using
taxonomic descriptions and for identifying specimens. The basic program
allows to save structured descriptions, comparable with the DELTA format and
consisting of three main elements: taxa, descriptors (characters for DELTA) and
states of descriptors. An import/export in SDD XML schema the new standard
proposition of TDWG, is also available (SDD= Structured Descriptive Data.
See : http://wiki.tdwg.org/SDD).
MAKEY [13], [15] is a key generator software. It selects step by step the best
question (descriptor) to create a node. By default, the choice is based on the
discriminating power (ability to split the pool of taxa in equal disjointed classes,
two or more classes) and so MAKEY tries to generate short and well-balanced
polytomous keys [16].
MAKEY creates keys discriminating the different taxa of the input knowledge
base. Other versions create a key to discriminate groups or classes of the taxa;
these classes are sets of taxa defined by different character states of a selected

108
descriptor. For example, if a descriptor is the toxicity of mushrooms, MAKEY can
create a key to identify the toxicity of a mushroom even if the specimen is not
identified at the species level; in the same manner it is possible to create keys
to identify genera within a knowledge base describing species if a descriptor
associates each species to its genus. The manual to use MAKEY is accessible
on line.

3 Description of MyKey
The final user of a key is the best person to know his observation constraints.
So, the concept of the MyKey service was to offer to the final user the possibility
to create identification keys customized to his needs.
A web interface gives an access to the different input parameters of the Makey
software. We classify these parameters in four categories:
- parameters are related to the data coverage of the key; for example one can
generate a key to all species of the knowledge base or a key restricted to a
given geographical area or to a genus etc.
- parameters are related to the taxonomic domain (importance and easiness
to observe a character),
- parameters have consequences on the topology of the key, like the criterion
to select characters (by default it minimizes the mean number of questions to
achieve an identification),
- parameters concern the format of the result: indentented or bracketed key
(see [21] in this book), text or HTML format etc.
So the interface is divided in four parts according to these categories.

3.1 Goal or terminal nodes of the key

The user selects the goal of the discrimination, it means the terminal nodes of
the key. The key can identify all or just a set of taxa, or any group of taxa defined
by character state. So, considering a knowledge base describing species and
a character “genus”, the user can then create a key discriminating the different
genera and then keys to recognize species within each genus. In the same
manner we can compute a key to identify the toxicity of mushrooms and not the
species themselves.
Considering a knowledge base that covers a world distributed taxon, the
user may need to consider only a subarea (hereafter called “sub-base”). The
sub-base will then only include the taxa specified by the user. If the user can
fill in a background (a specific region or country, a maximal bathymetric range
etc.), a sub-base will be extracted, excluding taxa not compatible with the given
conditions. The decision tree generated by MAKEY is then shorter than the key
including all taxa, and so it minimizes the probability of error. Indeed, if two taxa
are quite similar but are not present in the same altitude/country, using a key
built on a sub-base reduces the risk of misidentification.

109
3.2 Background knowledge

Weights (or ponderation values), one for each descriptor, define a pre-order on
the descriptor set (by default an equal weight is associated to all the descriptors
or characters). MAKEY will respect this pre-order to select the character at
each node. The characters can be ordered by the final user himself to force
their choice in the key. So if flowers are absent the weight of all the flower’s
characters can be minimized or put to zero. At the contrary if some characters
are easy to observe for the user he can associate to these characters a higher
weight.

3.3 Topology of the identification graph

The topology section lets the user to choose some criteria to be used during
the key construction (minimal number of branches at each node; to merge
branches; to eliminate first some taxa etc.). Some statistics measurements help
to compare the topology of the keys with different parameters and to choose the
best decision tree.

3.4 Output format

The user can define the parameters to display the key: nested key (also called
“yoked” or “indented”) or parallel key (also called “bracketed” or “linked” key).
Additional characters and states may be added if they are deduced at a step of
the key.
The generated key is available in HTML format (including an option for a special
layout for handheld devices) or in PDF for printing.

4 Architecture
Mykey is a server-side software implemented as a CGI script written in
PYTHON; the system is easy to maintain and to upgrade, and it is compatible
with any operating system.
According to the user selected parameters, (a) Mykey extracts a sub-base
if necessary, (b) Mykey creates or modifies the file of character weights, (c)
Mykey calls the software MAKEY which is then executed on the server with the
selected parameters and (d) Mykey formats the MAKEY output and the result is
sent to the client browser in the selected design. The key can also be saved on
the server (in fact only the parameters will be saved), to restore it when needed,
to modify it or to share it with other users.

5 Conclusion
Mykey is a running prototype. It is an efficient additional system to Xper2,
a midfield solution between single access key and free access key. Today a
depository for Xper2 knowledge bases is accessible at http://lis-upmc.snv.
jussieu.fr/xper2/infosXper2Bases/en/index.php to any user. Then the data

110
can be accessible with Mykey. An option modifies the display for output on a
personal pocket palm. Few similar options were encountered (URL: http://www.
phylodiversity.net/palmkey/), and the one proposed by MyKey is perfectible.
Mykey is not a website to access to keys but an online service to produce
keys [17]. In the European project EDIT the functions to create keys were
implemented in the CDM library (see [20] in this book).
Mykey has to be modified to become a web service able to be connected easily
to other softwares. In the future ViBRANT project (Virtual Biodiversity Research
and Access Network for Taxonomy http://vbrant.eu) such identification system
(free access and single access key construction) will be available as a web
service and will allow a more open and flexible use.

Acknowledgement

The authors wish to thank Amandine Sahl for her contribution to this work during her
master PhD, and all the users of this prototype.

References
[1] J. D. Agosti, “Biodiversity data are out of local taxonomists’ reach”. Nature, p. 392, 2006.
[2] J. R. Quinlan, “Induction of decision trees”. Machine learning, vol. 1, pp. 81-106, 1986.
[3] D. W. Goodall, “Identification by computer”. Bioscience, vol. 18(6), pp. 485-488, 1968.
[4] R. J. Pankhurst, “Identification methods and the quality of taxonomic descriptions”. In:
Biological identification with computers. Academic Press, London, 1975.
[5] P. M. Forget, J. Lebbe, H. Puig, R. Vignes and M. Hideux, “Microcomputer-aided identification
/ an application to trees from french Guiana”. Bot. J. Linn. Soc., vol. 93, pp. 205-223, 1986.
[6] M.J. Dallwitz, Overview of the DELTA System, 2009. http://delta-intkey.com/www/overview.
htm, June 2010.
[7] M. J. Dallwitz, T. A. Paine and E.J. Zurcher, User’s Guide to Intkey: a Program for Interactive
Identification and Information Retrieval, vol. 1, 1995.
[8] J. Lebbe R. Vignes and J.P. Dedet, “Computer-aided identification of insect vectors”.
Parasitology Today, vol. 5 (9), pp. 301-304, 1989.
[9] A. R. Brach and H. Song, “eFloras: New directions for on-line floras exemplified by the Flora
of China Project”. Taxon, vol. 55 (1), pp. 188-192, 2006.
[10] R. J. Pankhurst, Practical Taxonomic Computing. Cambridge Univ. Press, Cambridge, 1991.
[11] J. Lebbe and R. Vignes, “State of the art in computer-aided identification in biology”. Oceanis,
vol. 24(4), pp. 305-317, 1998.
[12] K. Causse and J. Lebbe, “Modélisation des stratégies d’identification par la méthode MCC”.
JAVA-95, (Conference proceedings), 1995.
[13] J. Lebbe and R. Vignes, “Génération de graphes d’identification à partir de description de
concepts”. In: Y. Kodratoff and E. Diday (eds.), Induction Symbolique et numérique à partir de
données, Cepadues, pp. 193-239, 1991.
[14] V. Ung, G. Dubus, R. Zaragüeta-Bagils and R. Vignes Lebbe, Xper²: introducing e-Taxonomy.
Bioinformatics, vol. 26(5), pp.703-704, 2010.
[15] R. Vignes, Caractérisation automatique de groupes biologiques. Université Pierre et Marie
Curie, 260 pp. (Thesis), 1991.
[16] J.C. Gower and R.W. Payne, “A comparison of different criteria for selecting binary tests in
diagnostic keys”. Biometrika, vol. 62, pp. 665-672, 1975.
[17] N. Conruyt, D. Sébastien, S. Cosadia, R. Vignes Lebbe and Touraïvane, “Moving from
biodiversity information systems to biodiversity information services”. In: L. Maurer,
K. Tochtermann (eds.), Information and Communication Technologies for Biodiversity
Conservation and Agriculture, Shaker, Aachen, (ISBN: 978-3-8322-8459-6), 2009.
[18] J. Nascimbene, S. Martellos and P. L. Nimis, “An integrated system for automatically producing

111
 user-specific keys - A case study on Italian lichens”. In: P. L. Nimis and R. Vignes Lebbe (eds.),
Tools for Identifying Biodiversity: Progress and Problems, pp. 151-156, 2010.
[19] E. van Spronsen, S. Martellos, D. Seijts, P. Schalk, and P. L. Nimis, “Modifiable digital
identification keys”. In: P. L. Nimis and R. Vignes Lebbe (eds.), Tools for Identifying Biodiversity:
Progress and Problems, pp. 127-131, 2010.
[20] W. G. Berendsohn, “Devising the EDIT Platform for Cybertaxonomy”. In: P. L. Nimis and R.
Vignes Lebbe (eds.), Tools for Identifying Biodiversity: Progress and Problems, pp. 1-6, 2010.
[21] G. Hagedorn, G. Rambold and S. Martellos, “Types of identification keys”. In: P. L. Nimis and
R. Vignes Lebbe (eds.), Tools for Identifying Biodiversity: Progress and Problems, pp. 59-64,
2010.

112
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – p. 113.
ISBN 978-88-8303-295-0. EUT, 2010.

Xper²: managing descriptive

data from their collection to
e-monographs
Visotheary Ung, Florian Causse, Régine Vignes Lebbe

Abstract — Computer Aided Identification (CAI) systems provide

users with the resources to relate morpho-anatomic observations with
names of taxa, and to subsequently access other knowledge about
the organisms. Xper² version 2.1 is one of the most user-friendly
software in its category. It provides a complete environment dedicated
to taxonomic descriptions management. While assisting taxonomists
with knowledge acquisition for identification keys, it also helps with
the publication of descriptive data by providing a large panel of tools,
including analysis facilities (comparison of taxa or descriptors used),
full traceability of information and knowledge (by adding references
or external links). Xper² provides excellent support for automatic on-
line publication of descriptive data and free access keys, as well as
for exporting of datasets for phylogenetic and systematic research.
It focuses on interoperability between systems and can import and
export into structured descriptive data format (TDWG-SDD), and
export to HTML and Nexus formats. Written in Java, it is available
on Windows™ Mac™ or Linux in French, English, and Spanish, and
a new Chinese version can also be downloaded. With its intuitive
interface, Xper² is aimed at professional taxonomists as well as
naturalists who merely want to identify specimens using a ready-
made application. Xper² is free of charge. It can be downloaded at:
http://lis-upmc.snv.jussieu.fr/lis/?q=en/resources/softwares/xper2.

Index Terms — descriptive structured data, knowledge base, interactive

identification key, integrated platform for identification.

————————————————
V. Ung is CNRS engineer in UMR 7207 CNRS/MNHN/UPMC, MNHN Département Histoire de la
Terre, CP48, 57 rue Cuvier, 75005 Paris, France E-mail: [email protected]
F. Causse is UPMC engineer in UMR 7207 CNRS/MNHN/UPMC, MNHN Département Histoire de
la Terre, CP48, 57 rue Cuvier, 75005 Paris, France E-mail: [email protected]
R. Vignes Lebbe is Professor in UMR 7207 CNRS/MNHN/UPMC, MNHN Département Histoire de
la Terre, CP48, 57 rue Cuvier, 75005 Paris, France.

113
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 115-120.
ISBN 978-88-8303-295-0. EUT, 2010.

FRIDA 3.0
Multi-authored digital
identification keys in the Web
Stefano Martellos

Abstract — FRIDA (FRiendly IDentificAtion) is a software for generating

different types of digital identification keys starting from a database of
characters. Its first release dates back to 2003. Since the first version, FRIDA
was constantly developed to improve its core functions, and to add new
features. At the beginning of 2010 it was decided to develop a new version,
3.0, written in PHP language for MySQL engines. The new version will have
new and enhanced features, as well as the possibility to interact with other
software developed in the framework of the European project KeyToNature.
The development is currently in the beta-testing phase, and the final release
is planned for the beginning of 2011.

Index Terms — biodiversity informatics, Dryades, KeyToNature.

—————————— u ——————————

1 Introduction

T
he first approaches to digital identification are recent, dating back to the
‘70s of the last century, and especially to the beginning of the “explosion”
of the World Wide Web, less than twenty years ago. Today there is a
great and continuously increasing number of different digital identification
keys, produced by several research centers: fixed- or free-pathway keys, with
different querying systems, matrix keys, simple textual keys, etc. [1], [2]. They
can be accessible on the Web, stored on CD- or DVD-ROMs, and some of them
can run on mobile devices as PDAs and Smartphones [3], [4], [5], [6], [7], [8],
[9], [10].
The production of identification keys, both “classic” and digital, to large groups
of organisms (e.g. a national flora), normally requires the combined effort of
several authors. In most of the classic, paper-printed keys, each author (or small
groups of authors) develops one or few keys to families, species or genera, which
are then connected in a hierarchical way by a “general key”. This approach has
been succesfully applied to digital keys as well, e.g. in the Flora of China project
————————————————
S. Martellos is with the Department of Life Sciences, University of Trieste, I-34127, Italy. E-mail:
[email protected].

115
[10], in which a different dataset for each family or genus was developed by one
or a few authors. A different approach is that of many authors working together
on a common dataset, e.g. in the project LIAS for lichenized and lichenicolous
fungi [11]. Another interesting experiment to develop a community approach in
building multi-authored digital keys is the BioWikiFarm [12]. In this case, the
keys are digital texts stored on a MediaWiki platform, and a potentially very
large community of authors can edit them, while the system registers and keeps
track of the changes.
FRIDA (FRiendly IdentificAtion) [13] was developed to allow several authors
to work together, but rather independently from each other, while building a
common database of morpho-anatomical data from which it is possible to
produce a virtually unlimited number of different multi-authored digital keys. This
software has been already used to generate hundreds of keys to plants, animals
and fungi in the framework of the European project KeyToNature. FRIDA was
available as a package running on Oracle databases only up to its last version
(2.0). The new version, FRIDA 3.0, currently in beta stage of development,
is written in PHP for MySQL databases, and has several new and improved
features. Furthermore, it will be possible to use it both in stand-alone and on-
line mode.

2 FRIDA
FRIDA (FRiendly IDentificAtion) has been developed since 2003 at the
Department of Life Sciences of the University of Trieste (Italy), in the framework
of project Dryades [7]. Up to its version 2.0 it was written in PL/SQL language,
and developed on an Oracle Database engine.
The most interesting features of FRIDA [13] are:
1. It does not require the learning of any code or programming language.
Input and management of data are in natural language, through simple
Web interfaces written in HTML 4.0.
2. Keys are immediately available on-line since their generation. They are
accessible from the Web by using any common web browser, through a
single-access and a multi-entry query interface [8].
3. Keys are independent from the original data. When a key is produced, it
does exist as a discrete entity, separated from the original database. In
this way it is possible to modify the keys whithout affecting the original
database, or vice-versa.
4. The database of characters has a double-level architecture. Characters
are stored in two levels of information: a) a first level which is common
to all taxa in the database, b) a second level, which is restricted to taxa
belonging to a given “group” (see later). Organisms can be divided into
more homogenous groups (e.g. genera and families, but also fully artifical
groupings) by using several characters of the first level. These groups do
exist as independent entities in the second level, and can be managed
as independent databases by different authors (e.g. specialists of a given
genus or family), which can thus work with a large degree of independence.
5. The weight of characters in the generation of dichotomous keys is decided

116
 by the authors case by case. While it is possible to use an algorithm [14]
to produce the “better” key (e.g. the key with the shortest branches), only
an experienced taxonomist knows which is the weight of a given character
in a particular group of organisms.
6. Keys are portable in the field, both online and in stand-alone versions.
The latter, while less performing, are the better solution when internet
connections are not available or poorly effective. Stand-alone keys can be
also stored on CD- and DVD-ROMs.

3 New features in version 3.0

The development of FRIDA 3.0 started at the beginning of 2010. The new
version is written in PHP language and runs on MySQL engines. It has the same
core features of the previous version, plus some new and enhanced features:
1. It can be also installed and used locally. While the previous version
required an Oracle Database and an Application Server to be installed and
could be used on-line only, this version can run locally on any computer
provided with MySql, PHP and Apache. It can be obviously installed on a
server computer as well.
2. Enhanced multilingual support. While in the previous version it was
possible to input the data into two languages only, in the new version
the number of languages can be decided by the authors, and is virtually
unlimited. The languages are selected in the settings interface, and new
languages can be added at any time.
3. New templates. The development of the previous version focused almost
exclusively on software performance, the interfaces for data management
were rather poor and sometimes tricky for the users. In the new version,
thanks to a new template, the functions are accessible in a far more
rational way. For example, the management of characters can be done by
using a few interfaces, which group all the character and states functions
(Fig. 1), while the old version had a different interface for each function.
4. Improved management of character images. In the previous version the
images of characters could be stored and reused, but the process was
tricky, and sometimes forced the authors to create “dummy” characters
with several fake states. In the new version character images can be
stored in a image archive, and searched by file name or keyword.
5. Easier management of records. In FRIDA each taxon can be described by
several records. Each record, while having the same name, and referring
to the same taxon, differs for at least a character state. In the new version
all the records referring to a taxon can be viewed together, and managed
in a more efficient way (Fig. 2).
6. Commented functions. All the interfaces are provided with short textual
information, to give the authors reminders on the use of the different
functions. A full reference manual to FRIDA 3.0 is in preparation.

117
 Fig. 1 – The  new interface for the management of characters and values. Several
functions are grouped together.

 
Fig. 2 – Each taxon can be described by several records, differing for at least a
character state. The records are managed in a simple interface, which permits to edit,
duplicate and delete them, as well as to add new records.

4 Conclusion
FRIDA 3.0 will be accessible to several research centers, including those
without an Oracle system. It will permit to export the keys both in the Open
Key Editor [15], [16], [17], in the Open Key Player [18] and in the BioWikiFarm
[12] formats, which were developed in the framework of KeyToNature [19],
to contribute to the development of integrated, open networks for digital
identification.
The estimated roadmap for the future development of FRIDA 3.0 is:
• November, 2010 – FRIDA 3.0 Beta 2,
• December, 2010 – FRIDA 3.0 Beta 3,
• January, 2011 – FRIDA 3.0 Release Candidate (RC) 1,
• February, 2011 – FRIDA 3.0 RC 2,
• March, 2011 – FRIDA 3.0 - official release.
While the beta testing phase is closed, the RC versions will be available upon
request to the author.

118
Acknowledgement

This paper was produced in the framework of the project KeyToNature, funded under
the eContentplus programme, a multi-annual Community programme to make digital
content in Europe more accessible, usable and exploitable. — Contract no. ECP-2006-
EDU-410019.

References
[1] M. J. Dallwitz, T. A. Paine, and E. J. Zurcher, Principles of interactive keys. (http://delta-intkey.
com), 2000 (onwards).
[2] M. J. Dallwitz, T. A. Paine and E. J. Zurcher, Interactive identification using the Internet. (http://
delta-intkey.com), 2002 (onwards).
[3] G. Agarwal, H. Ling, D. Jacobs, S. Shirdhonkar, W. J. Kress, R. Russell, P. Belhumeur, A. Dixit,
S. Feiner, D. Mahajan, K. Sunkavalli, R. Ramamoorthi and S. White, “First steps towards an
electronic field guide for plants.” Taxon, vol. 53 (3), pp. 597-610, 2006.
[4] A. R. Brach and H. Song, “ActKey: a Web-based interactive identification key program”. Taxon,
vol. 54 (4), pp. 1041-1046, 2005.
[5] D. F. Farr, “On line keys: more than just paper in the web”. Taxon, vol. 53 (3), pp. 589-596,
2006.
[6] K. Chang-Sheng and H. Song, “Interactive key to Taiwan grasses using characters of leaf
anatomy – the ActKey approach”. Taiwania, vol. 50, pp. 261-71, 2005.
[7] S. Martellos and P. L. Nimis, “KeyToNature: Teaching and Learning Biodiversity. Dryades, the
Italian Experience”. In: M. Munõz, I. Jelínek, and F. Ferreira (eds.), Proceedings of the IASK
International Conference Teaching and Learning 2008, pp. 863-868, 2008.
[8] R. D. Stevenson, W. A. Haber and R.A. Morris, “Electronic field guides and user community in
the ecoinformatics revolution”. Conservation Ecology, vol. 7(1), pp. 3, 2003
[9] M. J. Dallwitz, “A comparison of interactive identification programs”. (http.//delta-intkey.com),
2000 (onwards).
[10] A. R. Brach and H. Song, “eFlora: New directions for online floras exemplified by the Flora of
China Project”. Taxon, vol. 55(1), pp. 188-92, 2006.
[11] G. Rambold, “LIAS – The concept of an identification system for lichenized and lichenicolous
fungi”. In: Anonymous (ed.), The Third Symposium IAL 3. Progress and problems in
Lichenology in the Nineties. Abstracts. - 9. Salzburg, 1996.
[12] G. Hagedorn, G. Weber, A. Plank, M. Giurgiu, A. Homodi, C. Veja, G. Schmidt, P. Mihnev,
M. Roujinov, D. Triebel, R. A. Morris, B. Zelazny, E. van Spronsen, P. Schalk, C. Kittl, R.
Brandner, S. Martellos, and P. L. Nimis, “An online authoring and publishing platform for field
guides and identification tools”. In: P. L. Nimis and R. Vignes Lebbe (eds.), Tools for Identifying
Biodiversity. Progress and Problems, pp. 13-18, 2010.
[13] S. Martellos, “Multi-authored interactive identification keys: the FRIDA (FRiendly IDentificAtion)
package”. Taxon, vol. 59 (3), pp. 922-929, 2010.
[14] M.J. Dallwitz, T.A. Paine and E. J. Zurcher, “User’s guide to the DELTA System: a general
system for processing taxonomic descriptions”. 4th edition. (http://delta-intkey.com), 1993
(onwards).
[15] S. Martellos, E. van Spronsen, D. Seijts, N. Torrescasana Aloy, P. Schalk and P. L. Nimis, “Digital
identification keys to organisms and user-generated content. The KeyToNature approach”. In:
M. Muñoz and F. Ferreira (eds.), Proceedings of the IASK International Conference Teaching
and Learning 2009, pp. 96-102, 2009.
[16] S. Martellos, E. van Spronsen, D. Seijts, N. Torrescasana Aloy, P. Schalk and P. L. Nimis,
“User-generated content in the digital identification of organisms: the KeyToNature approach”.
Int. J. Information and Operations Management Education, vol. 3, 3, pp. 272-83, 2010.
[17] E. van Spronsen, S. Martellos, D. Seijts, P. Schalk and P. L. Nimis, “Modifiable digital
identification keys”. In: P. L. Nimis and R. Vignes Lebbe (eds.), Tools for Identifying Biodiversity:
Progress and Problems, pp. 127-131, 2010.
[18] M. Giurgiu, A. Homodi, E. van Spronsen, S. Martellos and P. L. Nimis, “The Open Key Player:

119
 A new approach for online interaction and user-tracking in identification keys”. In: P. L. Nimis
and R. Vignes Lebbe (eds.), Tools for Identifying Biodiversity: Progress and Problems, pp.
133-136, 2010.
[19] G. Hagedorn, P. L. Nimis and P. Schalk, “KeyToNature: Software, data formats, and
communities”. Biodiversity Informatics Symposium 2008. The Book of Abstracts, Swedish
Museum of Natural History, Stockholm, Sweden, 27, 2008.

120
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 121-125.
ISBN 978-88-8303-295-0. EUT, 2010.

Flora Bellissima, an expert

software to discover botany
and identify plants
Thierry Pernot, Daniel Mathieu

Abstract — To promote and facilitate access to botany for protecting biodiversity

indirectly is the objective of Flora Bellissima. This sofware package is based
on an integrated and complete database with numerous fonctionalities, as
well as on an expert system of recognition of plants called “Ophélie”. Aimed
at beginners, amateurs and experts, Flora Bellissima attempts to show that
it is possible to reconcile scientific rigor and popularization, in order to bring
together everyone interested in botany.

Index Terms — botanical data management, expert system, plant recognition

help, education through play.

—————————— u ——————————

1 Introduction

H
ow to raise awareness in the public of what biodiversity is and of its
protection? A solution is to help them to discover the nature around
them, so they can know and protect it better. To acquire this knowledge,
people need tools which are fun, easy, and based on scientific knowledge. The
tools must be also adaptable to different audiences, from novices to experts,
and they must take into account the progression of each individual’s learning.
It is in this perspective that Flora Bellissima [1], an integrated management
sofware dedicated to botany, has been developed. With a organisation similar
to a “ mini ERP” (Enterprise Resource Planning), this software is based on a
central database composed of a scientific index of plants (BDNFF [2]). Four
modules are organised around this database. They allow  : (i) an educational
discovery of the flora, (ii) assistance in plant determination thanks to the expert
system «  Ophélie  », (iii) a management system for botanical data collected
during field trips, and, (iv) a game module.
————————————————
T. Pernot, the autor of the software, is with Yourproject Informatique 27 rue Saint-Georges 39360
Larrivoire France [email protected] - www.yourprojectinfo.fr. He is a biologist by training
(Master of biology of organisms and populations at the University of Besançon) and computer
scientist (project manager in corporate and SSI).
Daniel Mathieu is the president of Tela Botanica, 163 rue A. Broussonet, 34090 Montpellier,
France, the NGO which publishes and distributes the sofware. [email protected] www.
tela-botanica.org.

121
 This software has been developed for anybody interested in the flora, whatever
their knowledge level. Novices who wish to make their first steps in botany can
learn in an entertaining way thanks to the game module, the image glossary
and the photographs. Amateurs who wish to improve their knowledge can store
the result of their observations (texts and photographs) and build up their own
database. Experts can consult easily and quickly the botanical nomenclature
with all its synonyms and its taxonomic levels : species, sub-species, varieties
and forms. The naturalist organisations can quickly save and exchange their
field botany notes. Everybody can use the expert system Ophélie to identify a
plant and give it a name before recording it.

2 Tool conception

2.1 General specifications

The detailed analysis of the different objectives led us to make the choices
found in the main concepts of Flora Bellissima:
• the software is made like an ERP, i.e. composed of several applications
sharing the same and unique database ;
• It is based on a scientific nomenclatural referential index ;
• It is open to allow the addition of text and photographs ;
• It is multi-level to be accessible to anyone : novices, amateurs and experts ;
• It contains data on the French flora for 1 400 species ;
• It proposes a help tool to determine plants, adapted to all audiences with the
expert system Ophélie ;
• It has several photographs for each plant (9  800 photograhs in total with
general appearance, inflorescence, flower, leaves etc.) to validate the
determination ;
• It is available on PC and is distributed on DVD.

2.2 The flora discovery module

This module helps the discovery of the flora of France through numerous
photographs and allows users to build up their own database based on their
own observations. It also includes more than 30 functionalities which are not
detailed here.
In summary, this module allows the consultation of a flora depending on
different points of view : plant type (tree, shrub, fern, etc.), plant use (medicinal,
cultivated, toxic), botanical family, genus, Latin or vernacular name. It also allows
the consultation and capture of information on different themes  : description,
medicinal properties, protection, and geographical distribution [3]. Note that it is
possible to add and link your own photograhs to each taxon. The three access
levels, the picture glossary and the ergonomy give a strong educational value
to this module.
The important idea emerging from this module is the gathering and
centralisation of information in order to facilitate the access to it.

122
2.3 The « plant station » note management module

This module allows “plant stations” to be defined with their geographical and
ecological characters, then to connect them to “botanical field trip notes”. This
tool has been designed for capturing efficiently the complete Latin names of
plants. Besides its functionalities of classification of stations and botanical field
trip notes, this module proposes the following options :
• plant search among all field trip notes ;
• analysis of the evolution of the plant population of a station through time ;
• copying field trip notes ;
• import/ export of stations and field trip notes ;
• printing field trip notes and exporting them as a Word, Excel, PDF etc file.

2.4 The determination aid module

This module is made up of two main functions  : a system of comparison

between plants and an expert system of determination.
The comparison system allows the comparison of between 2 to 5 plants
displaying all the characters which can differentiate them. This quickly gives
the important characters to differentiate plants from each other. The system,
called Ophélie, is an original tool which helps in the determination of plants
using the Flora Bellissima database. One thousand four hundred species are
now described, with about 30 characters for each plant, i.e. 45 000 characters
described for the whole database.
To make sure that the system is efficient, it is important to take into account
two main constraints: i) not all users are experts, and they can make mistakes,
ii) plant morphology differs from one plant sample to another. Also, to work
properly, the system must be able to accept mistakes and variations in the
descriptions.
The system can work only on a binary way with the characters: true or false.
It must work on all the characters described simultaneously, but not one after
the other, as it is the case with a dichotomous key.
Each entered value allows a “weighting factor” to be allocated to all possible
values of a character. To do this, we need to know “the distances” separating
the different values of the same character. This is done by the similarity matrices
which represent over 11 000 datapoints in Ophélie. In order to be able to enter
such a quantity of datapoints, specific computing tools have been developed.
These tools enable data capturing to be speeded up while checking it at the
same time. Ophélie can manage different types of characters such as matrix,
interval, quantitative values, etc.
The “weighting factors” are allocated to each of the chosen values thanks
to algorithms which depend on the character type: for instance, if we input the
value “pink” to the colour of a flower, this value will have the highest weighting
factor, the value “red” would have a lower weighting factor, and the value “yellow”
an even lower value...
In order to limit the number of questions to which the user must answer, Ophélie
has an algorithm which optimizes the path to the final answer (plant name) by

123
taking into account the previous answers. The system can determine which
criterion - until then unused in the process - can provide the best information.
Anything complex is handled by the computer, not by the user!
Dealing with the large number of descriptive characters to fill in - approximately
700 for 1 400 species, is another issue which had to be worked out. It was
obviously not conceivable to enter all this information manually for each species!
In Ophélie, the solution was to introduce hierarchical levels of description
which allow the factorization of descriptions. Three hierarchical levels are used:
(i) “general” to differentiate families, (ii) “family” to differentiate genera and (iii)
“genus” to differentiate species. This arrangement has led to a great reduction
in the number of descriptive characters to type in, from 700 to approximately 30.
Note that the system works in a global way and that consequently, the
“general” level can be enough for the determination of some species! However,
the management of level changes is a delicate step. To do so, two solutions are
used:
• delay the moment we activate the level change, but not excessively in
order to maintain a reasonable number of questions;
• checking descriptive characters for level change.
Once the determination is finished, the system Ophélie only proposes a plant
name if the difference between this plant and other plants is big enough, and if the
degree of similarity with the description of the given plant has reached a certain
level. This allows the reduction of the risk of getting an incorrect determination.
The quick display of all photographs concerning all plants corresponding closely
to the description is also a good control to validate the result.
The Ophélie system is therefore based on the principle of separation of
species, i.e. individuals of the same species are generally more similar than
individuals from related species.
With this structure, the expert system Ophélie could work with a number of
species superior to 1  400, and must be able to take into account the 6  000
species of the French flora. The performance of this system is solely dependent
on the quality of the information of the database and the sharpness of the
parameter setting.
In conclusion, Ophélie solves the problem of absence of an answer which
occurs when using a determination key and which stops the determination. Also,
with Ophélie, a bad choice for a few answers is not a major problem for the final
determination, because it is the mean of all the answers which is important.

2.5 The game module

This module proposes two games with three levels (Novice/ amateur/ expert).
Their aim is to make users learn how to recognize plants in an entertaining way
through the detailed observation of photographs..

3 Conclusions
Conceived in order to take action on the problems of biodiversity in a
sustainable and indirect way, Flora Bellissima is an attempt to bring together

124
different categories of botanists : novices, amateurs and experts in proposing
an educational and entertaining software based on strong scientific knowlegde.
Flora Bellissima represents 4 000 hours of work including both the software
design and the setting up of the knowledge base. Its extension to the whole
French flora will require the setting up of a collaborative working group in order
to capture characters concerning all taxa. The organization of such a group will
be managed by the Tela Botanica association, which groups together all French-
speaking botanists, and which has the qualifications and abilities required to
achieve such a complex task. Tela Botanica also plans, in the long term, the
online consultation of Flora Bellissima and could propose its application to other
floras in the world within the framework of the research project Pl@ntNet [4] as
a counterpart of existing systems.

Acknowledgement

We wish to thank the Tela Botanica team for their help in the realisation of this project as
well as for its diffusion. We also want to thank Paul Fabre for translating this presentation
in English.

References
[1] Flora Bellissima is a registered trademark of the Yourproject Informatique company.
[2] BDNFF: “Base de Données Nomenclaturale de la Flore de France” conducted by Michel
Kerguélen † and Benoit Bock, in the Tela Botanica network.
[3] Maps of geographic distribution of plants are provided by Tela Botanica.
[4] Pl@ntNet is an Interactive plant identification and collaborative information system supported
bay Agropolis Internatioonal Montpellier, France, built around three core teams that possess
complementary skills : AMAP, IMEDIA and Tela Botanica.

125
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 127-131.
ISBN 978-88-8303-295-0. EUT, 2010.

Modifiable digital
identification keys
Edwin van Spronsen, Stefano Martellos, Dennis Seijts,
Peter Schalk, Pier Luigi Nimis

Abstract — The Open Key Editor (OKE) is a tool for editing and enriching
existing identification keys and to produce localized ‘minikeys’ that apply to
local flora and fauna, such as in parks, nature reserves and school gardens,
or keys that apply to a particular season. The minikeys are easier to use than
their originals, simply because of the fact that they deal with less species,
their language can be adapted to a particular audience (e.g. pupils), and
because they always point to species that are known to be present. OKE
also allows the inclusion of user-generated content in any minikey (new text,
images, hyperlinks etc.). The output of minikeys can be automatically tailored
for display on computers, smartphones or PDA’s.

Index Terms — biodiversity Informatics, field guides, identification tools,

identification software, flora, fauna.

—————————— u ——————————

1 Introduction

I
dentification keys are often written by experts and aim at an ‘academic’
audience. Once they are published, they are more or less carved in stone
and leave little room for adaptation to a specific audience, a particular region
or season. In the case of plants, such a key can - for instance - encompass
1900 species for the Netherlands or more than 6000 species for e.g. Spain or
Italy. Long keys are complicated and have redundant information when used
in a region with fewer species, such as a park or nature reserve, or even a
school garden. An increasing number of identifications tools is being published
on the internet [1], [2], [3] offering an opportunity for tailoring them to particular
audiences and situations. The Open Key Editor [4] is a software package
developed within the KeyToNature European project, that allows users to ‘crop’
a master key and customize it for a given set of species. The ‘cropped’ key can
then be edited for language and illustrations (e.g. to suit a particular user level,
or platform such as the mobile phone).

————————————————
E. van Spronsen, D. Seijts and P. Schalk are with ETI BioInformatics, Amsterdam. E-mail: edwin@
eti.uva.nl, [email protected], [email protected].
S. Martellos and P. L. Nimis are with the Dept. of Life Sciences, Univ. of Trieste, via Giorgieri 10, I
34127 Trieste, Italy. E-mail: [email protected], [email protected].

127
2 The Open Key Editor
The KeyToNature Open Key Editor is an easy-to-use Open-Source tool for
editing and enriching a key with user-generated content. It was developed
since June 2009, is written in the PHP 5.2 language, and runs on a MySQL
5.0 database. The code is Open Source and available under the Creative
Commons Attribution Non-Commercial (CC-BY-NC) license. The program can
import dichotomous and polytomous keys with a compatible structure. It is
downloadable since December 2009 from the Web Portal of project KeyToNature
(http://www.keytonature.eu), together with sample keys. The current version is
1.1.
With the Open Key Editor the user can browse existing master keys and edit
them. The first step in making a customized ‘mini-key’ is to create a filter: this is
a list containing a subset of the species of the main key. Such a list can be made
by selecting species from the original key, or by importing a text file with species
names from an external source. The filters can be stored for later editing, so
many mini-keys can be tailored from the same basic dataset. In the Open Key
Editor new couplets can be added to the key for identifying species that are
absent in the original key.
The unprocessed mini-key will contain three kinds of questions from the
original key:
1. valid ones that still separate (groups of) species.
2. questions that used to be like type 1, but now have only a single remaining
branch.
3. questions that no longer lead to any species at all.
Questions of type 2 and 3 will have to be removed. Once a filter is defined,
the programme starts with the species that were removed and traces them back
until in encounters a question that is still relevant. All questions downstream
are ‘dead wood’ (type 3) and will be removed. The application repeats this
process with the remaining species in order to find questions of type 2. When
it encounters a question that no longer separates at least two species, the
question is removed from the decision tree, but its parent and child questions
are connected in a new branching pattern. Because there is a chance that this
new branching pattern will also contain questions of type 2, the whole process is
reiterated until no more changes have to be made. The result is a key in which
only questions of type 1 remain.
Special problems are reticulated keys. These are keys in which a question
branches to another part of the key that is not ‘downstream’ of the present node.
This problem is solved by controlling the creation of loops and unravelling them
during the processing of the key.

2.1 Key viewer

The Editor is provided with a simple single-access query interface (Fig. 1),
which displays, for each step of the identification process, one question and all
its possible answers, enriched by images when available. Users can retrieve
a list of the remaining species at each step of the identification process.

128
Furthermore, they can produce a printable, illustrated, identification key to the
remaining species.

Fig. 1 – A key as displayed by the Open Key Editor. At any time a minikey, based on the
remaining species, can be generated.

2.2 Editing interface

The main functions of the Editor are devoted to the integration of user-
generated content in the keys. Users can modify, or add user-generated content,
to:
• the text of the key
• the pictures
• the names of the species, their images and descriptions, and links to
external pages of interest
The editing process involves one line of a key at each time. The available
options are displayed in a simple graphic interface (Fig. 2).
Apart from editing keys at the level of lists of species, a second level of editing
is available: all texts of a key and its species descriptions can be changed. Jargon
can be removed for pupils or added for specialists. Photographs or drawings
can be uploaded by the user. They can be added to the original illustrations or
even replace them. Changes in texts or illustrations of a minikey will not interfere
with the original masterkey.

129
 Fig. 2 – The administrative menu of the Open Key Editor.

2.3 Generation of ‘filtered’ keys

The Editor can generate ex-novo a virtually unlimited number of “filtered” keys
from a single “master” key. After filtering, the “cropped” key becomes a separate
entity, which can be edited independently from its “master” counterpart. A filtered
key can be made available on the web, or given to a user for further editing. Any
changes in texts or illustrations of a filtered key will not interfere with the original
master key.

2.4 Export of stand-alone keys

Identification keys can be used at home or in a laboratory, but they can also
be used in the field, while exploring the biodiversity of an area. The Editor can
export the master key, or any of the filtered keys, in the form of stand-alone
packages (Fig. 3). The stand-alone versions can be published on CD- or DVD-
ROMs, or used on mobile devices, such as PDA’s and smartphones. These
devices, when equipped with a camera, can also be used to enrich the key with
original pictures.

130
 Fig. 3 – A single masterkey can produce different stand-alone minikeys, based on
scientific or arbitrary criteria.

3 Conclusion
With the Open Key Editor, existing identification tools can be modified and
their use can be made much easier by removing species that are absent from a
particular region or season. Text of keys and species descriptions can be edited
or translated so as to adapt them to user groups like pupils. Modified keys can
be turned into stand-alone applications for computers, websites, smartphones
or PDA’s.

Acknowledgement

The authors wish to thank all the persons involved in KeyToNature throughout Europe.
Their efforts and input gave us new ideas and energy to develop them. This paper was
produced in the framework of the the project KeyToNature (www.keytonature.eu, ECP-
2006-EDU-410019), funded in the eContentplus Programme.

References
[1] H. H. Visser and H. Veldhuijzen van Zanten, “European Limnofauna”
http://ip30.eti.uva.nl/bis/limno.php?menuentry=sleutel, 2010.
[2] ETI Bioinformatics, “World Biodiversity Database” http://ip30.eti.uva.nl/bis/projects.php, 2010.
[3] S. Martellos and P. L. Nimis, “KeyToNature: Teaching and Learning Biodiversity: Dryades,
the Italian Experience” In: M. Muñoz, I. Jelinek and F. Ferreira (eds.), Proceedings of the
International Association for the Scientific Knowledge (IASK) International Conference
“Teaching and Learning”, Aveiro, Portugal pp. 863-868. (http://www.dryades.eu), 2008.
[4] S. Martellos, E. van Spronsen, D. Seijts, N. Torrescasana Aloy, P. Schalk and P. L. Nimis:
“User-generated content in the digital identification of organisms: the KeyToNature approach”.
International Journal of Information and Operations Management Education (IJIOME), vol. 3,
3, pp. 272 -283, 2010.

131
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 133-136.
ISBN 978-88-8303-295-0. EUT, 2010.

The Open Key Player:

A new approach for online
interaction and user-tracking in
identification keys
Mircea Giurgiu, Andrei Homodi, Edwin van Spronsen,
Stefano Martellos, Pier Luigi Nimis

Abstract — This paper describes a new approach for creating and using
identification keys based on two main components: the Open Key Editor
and the Open Key Player, both of which were created within the European
project KeyToNature. The Open Key Editor can be used to produce custom
identification keys starting from a master key and to add original user-
generated content, while the most important feature of the Open Key Player is
the possibility to track relevant user activities into an eLearning environment,
in order to collect data to improve the design and usability of identification
keys.

Index Terms — identification key, eLearning, user-tracking, Rich Internet

Application.

—————————— u ——————————

1 Introduction

I
dentification keys are used to identify biological entities such as plants and
animals. Since the beginning of the digital era, the keys have undergone a
great improvement from the early, paper-printed versions [1]. Modern digital
keys are easier to access and to use, and can be used into schools as new,
efficient and interactive instruments for teaching biodiversity. However, building
an original identification key is still a task which can be carried on by an expert
only.

————————————————
M. Giurgiu and A. Homodi are with the Telecommunications Department, Technical University of
Cluj-Napoca, Cluj 400027, Romania. E-mail: [email protected].
E. van Spronsen is with ETI Bioinformatics, Amsterdam, E-mail: [email protected].
S. Martellos and P. L. Nimis are with the Department of Life Sciences, University of Trieste,
I-34127, Italy, E-mail: [email protected], [email protected].

133
2 The Open Key Standard
The Open Key Standard from the KeyToNature project has been developed
to make the creation and the use of an identification key easier, as well as to
improve its accessibility (Fig. 1). The standard contains two major components:
The Open Key Editor (OKE) [2], [3] and the Open Key Player (OKP).
The Open Key Editor provides the necessary interfaces to manage an
identification key created from scratch, or imported using the Structured
Descriptive Data (SDD) standard, as well as other file formats. From an original
key, called “master key”, it is possible to create a virtually unlimited number of
derived keys to different lists of taxa, containing new and original user-generated
content, and devoted to different target users.
The identification keys can then be used in the Open Key Player. This is a Flash
application developed by using Adobe Flex, which operates on the database of
the OKE, and displays in a modern, interactive way the keys to the user [4], [5].

3 Open Key Player Design

The Open Key Player has been designed following several requirements:
1) cross platform and browser compatibility, with the possibility to interact with
MySQL databases, and to display data in XML (eXtendable Markup Language)
format [6], [7]; 2) user interaction and information specific for dichotomous keys;
3) tracking the features selected along the identification path, and revising
the selection history; 4) ability to communicate, and to be integrated within an
eLearning platform, such as ILIAS or MOODLE.
A demo version of the Open Key Player running a key on woody plants of
Romania can be found at the URL specified at [9]. The application requires
the input of two parameters: the name of the database which contains the
identification key, and the code of the specific key in the database (since the
same database can store multiple identification keys).
The user interface is divided into two panels (Fig. 2). The left panel contains
a “Reset” button, which starts the identification process from the beginning, and
reports the number of remaining species and the history of the identification
steps (“Selected Features”). The right panel contains the current choices of
the identification process, which can be illustrated by images. The counter
of remaining species is updated, and the last step is added to the “Selected
Features” list at each step of the identification process.
When one of the choices leads to a result (Fig. 3), the scientific name of the
taxon appears in brackets. The information related to the identified taxon is
displayed in a pop-up window at the end of the identification process. It can also
contain an image gallery, when this gallery is available in the database.

134
Fig. 1 – The architecture of the KeyToNature Open Key Standard
(Editor, Player, Conversion tools).

Fig. 2 – The interface for the Open Key Player

(Left: selection history, Right: selection options).

Fig. 3 – The panel with the final result of the identification process.

135
4 User Tracking
One of the most interesting features of the OKP is the possibility of integrating
it in an eLearning environment, thus providing user-tracking. The application
can communicate to the eLearning environment each interaction made by the
user. This feature allows the creators of identification keys to access interesting
statistics on users behaviour, and to use them to improve the keys. The Open
Key Player has been successfully integrated in ILIAS, and was tested in several
Romanian high schools, giving back valuable information about the key of
woody plants of Romania.

5 Conclusions
The OKP has proven to be a valuable asset for the Open Key Standard of
KeyToNature. It provides a modern, user-friendly interface, which it can be
integrated into eLearning environments, providing user-tracking statistics. The
tests carried out in the Romanian high schools showed that the application is an
efficient interactive tool, and that it could be an important component in teaching
and learning biodiversity.

Acknowledgement

This paper has been supported by the project KeyToNature (www.keytonature.eu,

ECP-2006-EDU-410019), in the framework of the eContentplus Programme.

References
[1] S. Martellos, “Multi-authored interactive identification keys: The FRIDA (FRiendly IDentificAtion)
package” Taxon, vol. 59 (3), pp. 922-929, 2010.
[2] S. Martellos, E. van Spronsen, D. Seijts, N. Torrescasana Aloy, P. Schalk and P. L. Nimis,
“Digital identification keys to organisms and user-generated content. The KeyToNature
approach”, Proceedings of the IASK International Conference Teaching and Learning, pp.
96-102, 2009.
[3] S. Martellos, E. van Spronsen, D. Seijts, N. Torrescasana Aloy, P. Schalk and P. L. Nimis,
“User-generated content in the digital identification of organisms: the KeyToNature approach”,
Int. J. Information and Operations Management Education, vol. 3, 3, pp. 272-83, 2010.
[4] J. D. Herrington and E. Kim, Getting started with Flex 3, O’Reilly Media Inc., 2008.
[5] C. E. Brown, The Essential Guide to Flex 3, Apress, 2008.
[6] E. R. Harold, and W. S. Means, XML in a nutshell, O’Reilly Media Inc., 2004.
[7] E. T. Ray, Learning XML, O’Reilly Media Inc., 2003.
[8] D. J. Barrett, MediaWiki, O’Reilly Media Inc., 2008.
[9] Demo of OKE: http://octopus.utcluj.ro:56340/okp/openKeyPlayer.swf?db=oke&key=1, July 2010.

136
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 137-143.
ISBN 978-88-8303-295-0. EUT, 2010.

Improvement of identification
keys by user-tracking
Gerd Schmidt, Mircea Giurgiu, Sónia Hetzner, Fred Neumann

Abstract — Identifications keys are indispensable tools for identifying

organisms and understanding biodiversity. It is therefore advantageous to
give a broad public access to these tools. Due to the fact that these tools are
used by very different target groups – pupils, students, researchers, etc. –
customisation and quality enhancement are crucial for effective and appealing
identification keys. The evaluation of key software by inquiring users is
always influenced by various, often subjective impressions and is usually
restricted to a small number of them. The identification keys developed in the
framework the KeyToNature project are available on electronic devices and
allow registering the actions of single users during the identification process.
By transferring user actions to an Learning Management System (LMS) like
ILIAS OpenSource, where they can be correlated to user-data like experience
in the field of identification, courses visited, age etc., it is not only possible
to identify the problems a individual user did encounter and the mistakes he
made, but also to conduct statistical evaluations to review and substantially
improve the keys, to solve problems with single questions, and to optimise
customisation to the different user groups.

Index Terms — identification keys, user-tracking, eLearning, Rich Internet

Application, customisation, evaluation.

—————————— u ——————————

1 Introduction

I
dentification keys (id-keys) are indispensable tools to identify organisms,
which is a basic process for understanding nature and preserving
biodiversity. The European project KeyToNature aims at developing and
improving identification keys for all media, including electronic devices such
as smartphones and all types of computers. [1]. This form of electronic keys
permits – as a fundamental new possibility – to record every action performed
by users of a key, which is named here “tracking-feature”. Every wrong
alternative chosen by the user, every achieved species name, accepted as
right or rejected as wrong, can be registered. Gathering these information in a
Learning Management Software (LMS) [2], where it can be related to all user-
————————————————
G. Schmidt, S. Hetzner and F. Neumann are with the FIM Innovation in Learning Institute, Nägels-
bachstr. 25b in 91052 Erlangen. E-mail: [email protected], sonia.hetzner@fim.
uni-erlangen.de; [email protected].
M. Giurgiu is with the Universitatea Tehnica din Cluj-Napoca, Dep. Telecomuncatii, Str. Baritiu Nr.
26, 400020 Cluj-Npoca. E-mail: [email protected].

137
specific data (pre-knowledge, courses taken, age) enables a new benchmarking
of identification software and keys.
Therefore an add-on for the LMS ILIAS [3] was developed, that consists
of a logging service to record every action a user performs in an id-key, and
an evaluation service that performs basic statistical analysis for continuous
monitoring and quality enhancement.
The “logging service” in connection with the “tracking feature” inside the Flex
Player [4] or the Open Key Player [5, 6] enables us to track user behaviour for
continuous improvement of id-keys, to correlate the results with user groups:
age, education, previous experiences etc., and to gather an enormous amount
of information about how users use the key.
The “evaluation service” with filtering and exporting features offers automated-
evaluation processing at low efforts, testing the eligibility of keys for user groups
(e.g. age) by the filtering feature and exporting all information for further and
more detailed analysis.

2 What we want to know exactly

The above described services aim at continuously providing updated user
requirements for the further development and customisation of the id-keys. More
concretely, information such as: How long users need to orient themselves in
the key (especially true for multi-access id-keys)? How does the user navigate
in the key (choose an alternative, undo a choice, and restart because of feeling
lost)? How long does a user need to answer every single question? How much
time does a user need for one identification run? Does she or he reach the
correct result at the first attempt? Does she or he feel satisfied or certain to have
correctly identified the organism? How long does she or he use the key at all?
Additionally it is also extremely interesting to analyse the influence of pre-
knowledge and use of the id-key, as well as the specific pedagogical settings.
Furthermore, it is most interesting to investigate further factors that influence the
results, such as age, gender, school type, etc.
To answer these questions, it is necessary to get information about every
single event within the usage of a key, and its duration. All data are gathered by
the logging service, and a first automatic analysis is performed by the evaluation
service that was first developed for ILIAS and in the Open Key Player (single-
access keys) and the Flex Player (multi-access keys).

3 How does the process work

3.1 Integration Overview

The information flow in the collaboration between the LMS (ILIAS) and the key
software is shown in Fig. 1.

138
 Fig. 1 – Information flow LMS - User - Key-Software - LMS - Outside.

3.2 Data transmitted to the Key (Step 1 and 2)

The information transmitted to the Open Key Player software consists of

the name of database to be used by the Player, a reference-id to specify the
ILIAS object representing the key, and a user-specific session-id to correlate
the tracking of data to the individual user when they are sent back to the LMS.

3.3 Logging of User Interactions (Step 3)

The key sends back every user action to the LMS by appending them to the
URL of the logging service, together with the reference-id and the session-id.
The log entry is written by ILIAS only if the session_id is valid. The logging
service is responsible to add a time stamp to every action recorded.

3.4 Evaluation Service (Step 4)

The evaluation service inside the LMS performs the tasks of filtering and
exporting the log events by user profile data (age range, education, precognition,
e.g., lessons or courses attended before the key was used). It is also able
to perform a first analysis of data such as a count of the total number of key
sessions (how often the key was used at all), the average time of a key session
(is the key exciting or boring to the users?) or the average time to answer a
question (is it simple or difficult?). It can also count how often an answer was
revised, etc.

4 Logging events provided by the key respectively by the key-

player software

The information sent back by the key and logged by the LMS was designed to
serve a broad list of types of keys, and should be open for future developments.
Therefore it was decided to log four parameters that can be freely filled with text-

139
strings or numeric values.

4.1 Example for tracking a multi-access key

Multi-access keys were the first type for which user-tracking was established.
In this type of keys, users can select the question they want to answer first,
so that they need some time to orient themselves in the key. At the end of the
identification they can decide if they are confident with the result or if they want
to restart the determination process. With the sorting and filtering feature of the
logging service the log-events can be filtered and exported for further evaluation.

Fig. 2 – Tracking example for a multi-access key

(the lines have been coloured manually).

Data are sorted by user and time, so that the actions of single users within the
key can be easily observed. For example, pupil 1 performs an identification step
but is “NOT satisfied with the result” reached in the first step of the identification.
Thus, he does an “Application Reset” and finishes with the correct result.
Statistical analysis of large amounts of tracking data can identify certain questions
that often lead to wrong results, or are re-visited many times. Accordingly, these
specific questions are improved and closely evaluated in further testing events.

4.2 Example for the tracking of a single-access key

The following table was extracted from the log file of a determination of trees
and shrubs in a Romanian school. The “History revision” (marked in yellow)
represents selections that have been revised. The option number 198 was
independently re-visited by more pupils, which indicates that this alternative is
not clear enough.

140
Fig. 3 – Tracking example for a multi-access key (Grey lines represent tracking events
that have been removed to restrict the length of the table).

5 Feedback from the user tracking to the key builders by analysis

of the logfile

Without any explicit analysis, simply by sorting and filtering functions, the
logfile provides insight into how single users move in the key, how many times
they reached a result, rated it as right or wrong. It also provides a list of the
species they found with the key or how many organisms they tried to identify.
The logfile of the user tracking can be statistically analysed, basically in the LMS
with the evaluation service itself, and in a more extended and detailed way when
exported as an Excel-sheet for further analysis.
The following calculations may be used as feedback to improve the key and
are directly provided by the logging service in the LMS:
1. Key starts / Results assumed as correct or wrong
2. Time to select a question (multi-access keys only)
3. Time to answer a question
4. Time to successfully identify a species
5. Most selected questions (multi-access keys only)
To get additional, statistical information, the log-file can be exported for more
detailed analysing.

141
 Fig. 4 – Log-File of class A after second level analysis. “Species found” refers to the
result the identification.

6 Reliability and Performance

The testing events that took place in a Romanian school have been conducted
with two classes, concurrently using the Open Key Player software that was fed
by a Dutch database, sending the log data to a LMS hosted in Germany. Nearly
4000 datasets were recorded without problems and without significant load to
the servers. The logging service is estimated to be reliable and scalable, and
able to gather large amounts of information about the usability of keys in Europe.

7 Conclusions
The combination of identification tools with tracking features and a logging
and evaluation service in an LMS can give objective information about user
behaviour and the quality of identification keys. The filtering features allow
differentiating the appropriateness of keys for different user groups and enable
the authors of keys to improve them over time, adapting them to the needs of
different target groups. The application is robust and reliable, and can handle a
high number of concurrent users. As the actual evaluation of a single key can
be conducted by hundreds of LMS-Systems, the next step will be to hand back
user-specific quality indicators to the key or the key-database by means of the
key-player software.

142
Acknowledgement

This paper has been supported by the project KeyToNature (www.keytonature.eu,

ECP-2006-EDU-410019), in the framework of the eContentplus Program.

References
[1] L. M. Systems, http://www.e-teaching.org/technik/distribution/lernmanagementsysteme, 2010.
[2] ILIAS: http://www.ilias.de/docu/, 2010.
[3] S. Martellos, E. van Spronsen, D. Seijts, N. Torrescasana Aloy, P. Schalk and P. L. Nimis,
“Digital identification keys to organisms and user-generated content. The KeyToNature
approach”, Proceedings of the IASK International Conference Teaching and Learning, pp.
96-102, 2009.
[4] M. Giurgiu, G. Hagedorn and A. Homodi, “IBIS-ID, an Adobe FLEX based identification tool for
SDD-encoded multi-access keys”. Proc of TDWG 2009, 9-13 November 2009, Montpellier, p.
90, 2009.
[5] M. Giurgiu, A. Homodi, E. van Spronsen, S. Martellos and P. L. Nimis, “Open Key Player: A
new approach for online interaction and user tracking in identification keys”. In: P. L. Nimis
and R. Vignes Lebbe (eds.), Tools for Identifying Biodiversity: Progress and Problems, pp.
133-136, 2010.
[6] Demo version of Open Key Player, http://octopus.utcluj.ro:56340/okp/openKeyPlayer.
swf?db=oke&key=1, July 2010.

143
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 145-150.
ISBN 978-88-8303-295-0. EUT, 2010.

ARIES: an expert system

supporting legislative tasks.
Identifying animal materials
using the Linnaeus II software
Leo W.D. van Raamsdonk

Abstract — Bovine Spongiform Encephalopathy (BSE, Mad cow disease)

is generally considered to be caused by recycling animal by-products as
ingredient in animal, especially ruminant, feed. Feed bans were enforced to
minimize the risk on infections, and monitoring programs are effectuated for
controlling the ban. The only official detection method is visual (microscopic)
examination of the presence of primarily bone fragments, but muscle fibres,
hairs, feather filaments, and fish bones are targeted as well. An expert system
is developed called ARIES (Animal Remains Identification and Evaluation
System) with identification tools for those fragments. ARIES provides
information on procedures, legislation and background documentation,
and descriptions of prohibited materials as well as of a range of confusing
particles. The system allows documentation of the results of evaluations for
future reference and is considered a good platform for training and education.

Index Terms — ARIES, BSE, identification, Linnaeus II software, mad cow

disease.

—————————— u ——————————

1 Introduction

I t is generally supposed that the most likely route of infection of cattle

with Bovine Spongiform Encephalopathy (BSE, mad cow disease) is by
means of feeds containing low levels of processed animal proteins (PAPs).
Because of this likely route of infection, feed bans were enforced, primarily for
ruminant feeds, but later on extended to all feeds for farmed animals. The goal
of legislation is the enforcement of the “species-to-species” ban which prohibits
only the feeding of animal specific proteins to its own species [1]. However,
such a species-to-species ban needs to be supported by species-specific
identification methods.
In the European Union microscopic evaluation is currently the only accepted

————————————————
L. W. D. van Raamsdonk is with RIKILT – Institute of food safety, P.O. Box 230, 6700 AE Wagenin-
gen, the Netherlands. E-mail: [email protected].

145
method for the detection and characterization of PAPs in feeds [2], [3]. This
method is predominantly focused on the presence and characteristics of
bone fragments, although other structures, e.g. muscle fibres, may provide
circumstantial evidence of the respective animal types. Recent developments
are the identification of bone fragments at the level of classes (mammal vs. bird
vs. fish), supported by image analysis of bone characteristics [4].
Identification of bone fragments is based on a series of characteristics,
ranging from the shape and transparency of the bone fragments, to histological
differences between the different species groups. Information on as much
characteristics as possible should be collected when a few bone particles are
detected in a feed sample in order to get an acceptable reliability of a presumed
identification. Nevertheless, in many cases it is impossible to get a 100%
match between all collected information and the profile of one of the targeted
species. Consequently, uncertainty analysis is one of the necessary aspects of
a scientifically sound evaluation of microscopic analyses.

2 Expert system ARIES

2.1 Development

An expert system has been developed to document every aspect of the

detection and identification of prohibited animal materials in feed. The software
package Linnaeus II (ETI, Amsterdam, [5]) has been chosen as platform. A first
stand-alone version of the expert system was released in 2004 with the name
ARIES (Animal Remains Identification and Evaluation System), as a product of
an EU funded project (“STRATFEED”, [6]). ARIES provides a series of modules,
including an Introduction, full documentation on protocol and methods, an
overview of prohibited and confusing materials (e.g. plant hairs, minerals of
animal origin), several identification trees and matching procedures, a glossary,
an overview of legislation and extensive documentation from literature and
internet links [7].
A version 2.0 has been developed as a web-based application in the framework
of the EU funded project “SAFEED-PAP” [8]. The release of ARIES v2.0 was
achieved in 2010 [9].

2.2 Validation of the use of ARIES

ARIES has been used in a validation study of the microscopic method for the
detection of animal proteins in feed according to the official method [10]. In this
study, 25 laboratories investigated a set of 24 blind samples, partly adulterated
with several types of animal proteins in seven different treatments including a
control (blank). Thirteen of these laboratories used ARIES for support of their
detection and identification of the materials, twelve did not. All participants were
asked to report the presence of fish meal, material of terrestrial animals, and if
the latter was present, to indicate whether it was mammalian or avian material.

146
In all cases a “presence”, “absence” or “no result” could be reported. The results
were summarised in accuracy values: the number of correct results divided by
the total number of reports (excluding the number of no results).
This presentation will focus on the results for the proper detection of mammalian
material in the appropriate treatments and the presence of confusing ingredients
(fish meal). Therefore, the results for four different treatments (sample types)
were shown:
1. blank feed
2. feed with 5% of fish meal
3. feed with 0.1% of MBM and 5% of fish
4. feed with 0,1 % of MBM
The overall scores for these treatments and parameters, not stratified for the
use of ARIES, are presented in Tab. 1. The detection at the highest classification
level (fish material, and terrestrial animal material) poses in general no problem.
The detection of terrestrial animal material in the presence of fish material
(0.768) should be improved.
A considerable number of laboratories felt insecure for the identification of
specifically mammalian material, as is shown by the high number of “no results”
in Tab. 1.

n AC
Material terrestrial mammalian fish
blank 100 0.908 (2) 0.933 (7) 0.880 (0)
fish 5% 100 0.857 (2) 0.919 (10) 0.990 (1)
MBM 0.1% + fish 5% 100 0.768 (1) 0.639 (27) 0.970 (0)
MBM 0,1 % 75 0.987 (0) 0.896 (27) 0.920 (0)
Tab. 1 – Basic results expressed in accuracy for the detection of different types of
animal proteins in four differently contaminated feeds. Number of “No results” in
brackets. n: total number of observations.

AC with ARIES AC without ARIES

Material n mammalian n mammalian
blank 52 0.935 (2) 48 0.930 (5)
fish 5% 52 0.891 (2) 48 0.950 (8)
MBM 0.1% + fish 5% 52 0.702 (4) 48 0.520 (23)
MBM 0.1 % 39 0.857 (4) 36 1.0 (23)
Tab. 2 – Basic results expressed in accuracy for the detection of mammalian material
in four differently contaminated feeds split for the use of the ARIES system. Number of
“No results” in brackets. n: total number of observations.

147
 AC’ with ARIES AC’ without ARIES
Material n mammalian n mammalian
blank 52 0.896 48 0.833
fish 5% 52 0.854 48 0.792
MBM 0.1% + fish 5% 52 0.647 48 0.271
MBM 0.1 % 39 0.769 36 0.361
Tab. 3 – Recalculated results expressed in adjusted accuracy for the detection of
mammalian material in four differently contaminated feeds divided in two groups of
users and non-users of the ARIES system. n: total number of observations.

There is no significant difference in the performance of the laboratories that

used or did not use ARIES for the detection of terrestrial animal material in
general, and of fish material. However, for the identification of mammalian
material low numbers of “no result” were reported for the group of ARIES users
compared to the non-users (Tab. 2). The ARIES users apparently made use
of the system as a sufficient source of expertise that allowed them to draw a
(correct) conclusion in most cases.
The accuracy values for non-users are biased by the high number of “no
results”. Therefore corrected accuracy values were calculated (Tab. 3). From
these values it can be concluded that ARIES offers a considerable support to
the detection and identification of mammalian material.

2.3 Evaluation

The participants of the validation study showed, by means of investigating a

training sample set prior to the proper study, to have a sufficient level of expertise
for the type of material that was included in the validation study. This starting
situation explains why there is no significant difference between the group of
users and of non-users for the detection of terrestrial animal proteins in general,
and of fish meal. It should be emphasized, however, that ARIES provides a
much larger range of data and information than the subset that was necessary
for performing well in the validation study. As an example, unintentional
contamination of feeds by rodents can be identified by using the information on
different hair types as included in ARIES.

2.4 Application

A European network of more than 100 official control laboratories monitors

the ban on animal materials in animal feed. The backbone to this monitoring is
provided by an EU network of National Reference Laboratories, one for each
Member State, coordinated by the European Union Reference Laboratory.
Approximately 50 of these laboratories applied for a subscription to the stand-
alone version 1.0. ARIES is used frequently because it offers assistance in the
application of the methodology and the identification of the findings. The system
allows documentation of the results of the evaluation for future reference and is

148
considered a good platform for training and education.

3 Discussion
The application of an expert system for support of the detection of prohibited
animal proteins in feed is an advantage in several ways:
• Support of daily routine analyses by providing procedural information as
well as the evaluation of observations..
• A formalised evaluation of observations as documentation for later reference.
• Training system and a platform for knowledge transition.
In the case of ARIES, the system is designed to be helpful for the experienced
scientists, as well as for training and e-learning of less experienced microscopists.
The development of new information was necessary for a new version 2.0.
The project SAFEED-PAP [8] provided a sufficient amount of data for a principle
improvement and fine tuning. The choice to develop a web application allows to
update the system whenever new information needs to included. A web based
application also implies that a sustainable support should be maintained. It is
the intention to give users access to ARIES 2.0 with a username and password
based on a reasonable annual fee. In this way maintenance can be assured
without having a commercial exploitation.
The performance of the microscopic method as illustrated by the accuracy
indices of Tab. 1 reflect the situation in 2004. After that, improvements have
been achieved, and in a period of five years an accuracy of 0.98 was established
in a blind test for European laboratories for the detection of 0.1% of MBM in
the presence of 5% of fish material [3], [11]. ARIES includes descriptions of
prohibited materials and a range of confusing plant materials in order to minimise
false positive detections. By providing this information, ARIES is one of the tools
for maintaining this high level of performance.

Acknowledgements

This work was supported by the European Commission in the framework of the
European Project SAFEED-PAP (FOOD-CT-2006-036221), “Detection of presence of
species-specific processed animal proteins in animal feed”, funded under the 6th EC FP,
DG RTD.

References
[1] European Union, “Regulation (EC) No 1774/2002 laying down health rules concerning
animal by-products not intended for human consumption”. Official Journal of the European
Communities, 10.10.2002, L 273, pp.1-95, 2002.
[2] European Commission, “Commission Regulation (EC) No 152/2009 of 27 January 2009 laying
down the methods of sampling and analysis for the official control of feed”. Official Journal of
the European Communities L 54, 26.2.2009, pp. 1–130, 2009.
[3] L. W. D. van Raamsdonk, C. von Holst, V. Baeten, G. Berben, A. Boix and J. de Jong, “New
developments in the detection of animal proteins in feeds”. Feed Science and Technology, vol.
133, pp. 63-83, 2007.
[4] A. Campagnoli, C. Paltanin, G. Savoini, A. Baldi and L. Pinotti, “Combining microscopic
methods and computer image analysis for lacunae morphometric measurements in poultry

149
 and mammal by-products characterization”. Biotechnol. Agron. Soc. Environ., vol. 13(S), pp.
25-28, 2009.
[5] ETI bioinformatics. Linnaeus II software package. http://www.eti.uva.nl/, 2010.
[6] Project website: http://stratfeed.cra.wallonie.be/, 2010.
[7] Vermeulen Ph., V. Baeten, P. Dardenne, L. W. D. van Raamsdonk, R. Oger, A.S. Monjoie and
M. Martinez, “Development of a website and an information system for an EU R&D project: the
example of the STRATFEED project”. Biotechnol. Agron. Soc. Environ., vol. 7, pp. 161-169,
2003.
[8] Project website: http://safeedpap.feedsafety.org/, 2010.
[9] RIKILT Institute of food safety. ARIES, Animal Remains Identification and Evaluation System,
http://aries.eti.uva.nl/, 2010.
[10] C. von Holst, L. W. D. van Raamsdonk, V. Baeten, S. Strathmann and A. Boix, “The validation
of the microscopic method selected in the Stratfeed project for detecting processed animal
proteins”. Stratfeed, Strategies and methods to detect and quantify mammalian tissues
in feedingstuffs, chapter 7. Office for Official Publication of the European Communities,
Luxembourg, 2005.
[11] L. W. D. van Raamsdonk, W. Hekman, J. M. Vliege, V. Pinckaers and S.M. van Ruth, “Animal
proteins in feed. IAG ring test 2009”. Report 2009.017, RIKILT, Wageningen, 34 pp., 2009.

150
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 151-156.
ISBN 978-88-8303-295-0. EUT, 2010.

An integrated system for

producing user-specific keys on
demand: an application
to Italian lichens
Juri Nascimbene, Stefano Martellos, Pier Luigi Nimis

Abstract — The identification of lichens is important in several applied fields,

such as the biological monitoring of air pollution and the restoration of open-
air stone monuments. This often creates relevant problems for non-specialists
and technicians which are in charge of routinely applying lichen monitoring
techniques. The coupling of a complex information system (ITALIC), together
with a new software which can automatically produce identification keys
for any subset of species included in a database (FRIDA), is an innovative
approach in the field of identifying biodiversity. ITALIC is able to produce a list
of species which potentially occur under a set of ecological and distributional
conditions specified by the user. The list is automatically transferred to FRIDA,
which generates a user-oriented interactive identification key limited to the
species present in the “virtual habitat” created by the user. The new system
has relevant applications, since it effectively supports the technical personnel
of Environmental Agencies, Nature Parks, Cultural Heritage Conservation
Agencies involved in lichen monitoring throughout the Country.

Index Terms — interactive keys, biomonitoring, restoration of monuments,

lichens, identification.

—————————— u ——————————

1 Introduction

L
ichens are a diverse and species-rich group of fungi living in close
symbiotic relationship with algae or cyanobacteria. Many of them form
crustose thalli whose identification usually requires the analysis of micro-
morphological or chemical characters. However, the identification of lichens
is important in several applied fields, such as the biological monitoring of air
pollution and the restoration of open-air stone monuments.
Epiphytic lichens are highly sensitive to environmental changes and air
pollution, and they are among the most widely used biomonitors in terrestrial
environments [1]. Epi- and endolithic lichens are important in the biodeterioration
————————————————
The authors are with the University of Trieste, Department of Life Sciences, via Giorgieri, 10 – I
34127 Trieste. E-mail: [email protected], [email protected], [email protected].

151
of stone monuments [2], [3]. In Italy their identification is required for restoration
programs, both in the planning phase and for monitoring the effectiveness of
restoration practices [4]. The identification at species level often creates relevant
problems to non-specialists and technicians which are in charge of routinely
applying lichen monitoring techniques.
In Italy, the popularization of lichen biomonitoring was supported and
encouraged by rendering available online information on lichens provided
by researchers. The first relevant contribution was the Italian checklist by
Nimis [5], which was the basis for the creation of ITALIC, the Information
System on Italian Lichens [6], that is freely available online since 2003.
ITALIC provides support to the identification process by offering the possibility
to compare a given specimen with high-resolution pictures and ecological-
distributional information. However, while helpful, this simple comparison is
far from being enough for this difficult groups of organisms, especially for a
layperson.
In this paper we present a system which is able to produce ”keys on demand”,
by coupling lists of species for “virtual habitats” created by users in ITALIC with a
software that automatically generates identification keys to the species in those
lists.

2 Basic information contained in ITALIC

For each lichen species which occurs in Italy (ca. 2350 in total), ITALIC
provides: 1) updated literature on the specimens recorded from each of the 20
administrative regions of the country, 2) a predictive map depicting its potential
distribution in 9 bioclimatic regions of Italy, 3) high-resolution pictures, and 4)
information on biological traits, ecological requirements, and commonness/rarity.
Biological traits include: growth forms (crustose, narrowly-lobed foliose, large
lobed foliose, fruticose, fruticose-filamentose, squamulose), photobiont type
(Cyanobacteria, Trentepohlia, Chlorococcoid green algae), and reproductive
strategy (sexual vs. asexual by soredia, isidia or thallus fragmentation).
Ecological requirements are summarised by 4 ecological indicator values: pH,
air humidity, solar irradiation and eutrophication. Each ecological indicator value
is expressed on a 1-5 ordinal scale. An additional indicator, expressed on a
0-3 ordinal scale, describes the tolerance of a species to human disturbance
(poleophoby or hemeroby). Information on altitudinal range and substrates
(epiphytic, lignicolous, foliicolous, saxicolous, and terricolous) is given as well.
The commonness/rarity of a species is expressed by a scale with 8 classes,
ranging from extremely rare to extremely common in each of the main bioclimatic
areas of the Country. This parameter was assessed on the basis of: a) number
of samples in the TSB lichen herbarium, b) number of literature records, and,
c) expert judgement. The ‘extremely rare’ status is given only to taxa known
from less than 5 localities in Italy, or to those which were not mentioned in the
literature in the last fifty years. Recently-described or dubious taxa are excluded
from this category. The species in the ‘extremely rare’ class are likely candidates
for the ‘threatened species’ category of the IUCN criteria [7], and therefore
important for conservation purposes.

152
3 Automatic generation of species lists for “virtual habitats”
By combining several morphological, ecological and distributional parameters,
ITALIC permits to elaborate complex queries for reconstructing “virtual habitats”
in different parts of the country, the output being in the form of lists of species
which are most likely to occur under the specified conditions. The predictivity
of these virtual lists was tested by Nimis & Martellos [8] using a multivariate
analysis of a matrix of “virtual” and real relevés of epiphytic lichen vegetation.
Virtual relevés were obtained by selecting administrative region, altitudinal belt,
substrate, and appropriate values for each ecological indicator value according
to the main features of the targeted habitats. The results showed that the
“virtual relevés” are highly predictive models, indicating that ITALIC could be
consistently used for generating a large number of lists of lichens potentially
occurring under conditions specified by the user. In the framework of the
“Carta della Natura” project, promoted by the Italian Institute for Environmental
Protection and Research (ISPRA), we recently applied this approach for
providing potential lists of lichens for each of the ca. 160 CORINE-Biotopes
habitats inventoried for Italy at the 1:50.000 scale [9]. The lists were obtained
by combining the most important parameters describing the distribution and
the ecology of Italian lichens: regional distribution, bioclimatic region, substrate
type, ecological indicator values. In a few cases, additional parameters, such as
commonness/rarity and tolerance to human disturbance, were used as well. A
qualitative evaluation of the predictivity of these lists was assessed by checking
real species lists available for well-studied habitats such as coniferous alpine
forests [10], [11]: a good correspondence between the two datasets was found.
Despite the fact that more quantitative evaluations on a large dataset would
be welcome to statistically confirm previous results, this experience supports
the practical utility of using the automatic generation of species lists for applied
purposes.

4 FRIDA: from a morpho-anatomical database to interactive

identification keys

In the last decade, our research team has developed original software for
automatically generating interactive identification tools. A first phase of the
research was conducted in the framework of the national project Dryades,
and then continued in the European project KeyToNature. The most important
software is FRIDA [12], a package which permits to generate interactive
identification keys from a database of morpho-anatomical characters, starting
from any list of species. The huge floristic information contained in several local
checklists or vegetation studies can thus be easily used for generating new and
original identification keys. For example, we used the detailed floristic information
provided by Poldini [13] for the area of the natural reserve of the Val Rosandra
(Trieste, NE Italy), and by Festi & Prosser [14] for the Paneveggio-Pale di San
Martino Natural Park (Trento, NE Italy), to generate digital keys to the 988 and
1451 species of vascular plants known to occur in these two areas, respectively.
In the last years, the continuous development of FRIDA, and of other original

153
software produced in the framework of Dryades and KeyToNature, permitted to
greatly improve both the morpho-anatomical databases and the identification
keys, which are now available in different and user-friendly layouts, such as
those running on iPhones and other portable devices.

5 A step further: the combined use of ITALIC and FRIDA for

generating interactive identification keys on demand

Here we provide an example of the high informative and applicative potential

which derives from the combined use of an ecological-distributional database
(ITALIC) with a morpho-anatomical database (FRIDA). The great novelty is that
it is now possible to obtain from ITALIC lists of species potentially occurring
under specified conditions, which are used for “feeding” FRIDA to generate
personalized, user-oriented interactive identification key on demand.

Fig. 1 – A) The user inputs in ITALIC a combination of parameters defining a “virtual

habitat”, B) ITALIC produces a list of species which are likely to occur under the
specified conditions, C) the list is automatically sent to FRIDA, D) FRIDA produces
an interactive key to the species in the list, E) the key, in its various versions (online,
stand-alone for CD-ROM and for mobile devices) is forwarded to the user.

This novelty can have relevant applicative implications. For example, forest
managers, which are in charge of monitoring epiphytic lichen diversity in spruce
forests of the Italian administrative region Trentino-Alto Adige, could obtain
their potential species list by selecting the administrative region, the ‘subalpine’
bioclimatic belt, ‘epiphytic’ lichens growing on ‘acid bark’ (pH indicator value

154
1-2), in ‘mesic’ (air humidity indicator value 2-4), ‘shaded’ (indicator value for
solar irradiation 2-3), and ‘non-eutrophicated’ (indicator value for eutrophication
1-2) conditions. An interactive identification key to these species, including high-
resolution pictures and ecological notes, is immediately created by FRIDA, and
can be used for routine activities.
Similarly, managers in charge of monitoring the effectiveness of a restoration
program on a Greek temple near the town of Agrigento (Sicily, S Italy), could
obtain the list and the identification key for the species potentially occurring in
that environment. They should just select ‘Sicily’, ‘dry-mediterranean’ bioclimatic
region, ‘saxicolous’ lichens growing on limestone (‘basic’ substrata, indicator
value for pH 5), in ‘dry’ (air humidity indicator 4-5), and ‘sun-exposed’ (indicator
value for solar irradiation 4-5) conditions.
The digital identification keys are immediately ready to be used online.
However, when identification has to be carried out in the field, the Web could
be not the best medium for a key. For this reason, several accessory tools were
developed, which permit the storage of the keys as stand-alone packages on
different media, such as mobile devices (iPhones and other smartphones), as
well as on paper, in the form of printable, illustrated field guides.

6 Conclusion
The identification of lichens is often difficult, both in the field and in the
laboratory, and requires a long period of study and training. The new digital keys
“on demand”, being restricted to a relatively small number of species potentially
occurring in a given habitat and/or in a given area, are generally much more
user-friendly. They proved to be a valuable support for the technical personnel
of Environmental Agencies, Nature Parks, Cultural Heritage Conservation
Agencies, etc. Within KeyToNature we have produced free of charge hundreds
of mini-keys for schools using the same approach. However, the potential market
for such a service in the field of practical applications is quite high: we have
already received dozen of requests by institutions and companies which are
ready to pay for having the possibily of generating “their own” keys on demand.

Acknowledgement

This paper was produced in the framework of the project KeyToNature, funded under
the eContentplus programme, a multi-annual Community programme to make digital
content in Europe more accessible, usable and exploitable. — Contract no. ECP-2006-
EDU-410019.

References
[1] P. L. Nimis, C. Scheidegger and P. A. Wolseley, “An introduction”. In: P. L. Nimis, C.
Scheidegger, and P. A. Wolseley (eds.), Monitoring with Lichens - Monitoring Lichens. NATO
Science Series. IV. Earth and Environmental Sciences, 7, Kluwer Academic Publishers,
Dordrecht, The Netherlands, 2002.
[2] G. Caneva, M. P. Nugari and O. Salvadori, Plant biology for Cultural Heritage. Getty
Conservation Institute, Los Angeles, 400 pp., 2008.
[3] M. R. D. Seaward, C. Giacobini, M. R. Giuliani and A. Roccardi, “The role of lichens in the

155
 biodeterioration of ancient monuments with particular reference to Central Italy”. International
Biodeterioration and Biodegradation, vol. 48, pp. 202-208, 2001.
[4] J. Nascimbene, O. Salvadori and P. L. Nimis, “Monitoring lichen recolonization on a restored
calcareous statue”. Science of Total Environment, vol. 407, pp. 2420-2426, 2009.
[5] P. L. Nimis, The lichens of Italy. An annotated catalogue. Museo Regionale Scienze Naturali,
Torino, Monografie, XII, 897 pp., 1993.
[6] P. L. Nimis and S. Martellos, ITALIC - The Information System on Italian Lichens. Version
4.0. University of Trieste, Department of Biology, IN4.0/1 (http:// dbiodbs.univ.trieste.it/), 2010.
[7] IUCN, IUCNRed List Categories and Criteria: Version 3.1. IUCN Species Survival Commission.
Gland, Cambridge: IUCN. 30 pp., 2010.
[8] P. L. Nimis and S. Martellos, “Testing the predictivity of ecological indicator values. A comparison
of real and “virtual” relevés of lichen vegetation”. Plant Ecology, vol. 157, pp. 165-172, 2001.
[9] P. L. Nimis, Department of Life Science, University of Trieste, 2010. (personal communication)
[10] J. Nascimbene, S. Martellos and P. L. Nimis, “Epiphytic lichens of tree-line forests in the
Central-Eastern Italian Alps and their importance for conservation”. The Lichenologist, vol. 38,
pp. 373-382, 2006.
[11] J. Nascimbene, L. Marini, R. Motta and P. L. Nimis, P. L., “Influence of tree age, tree size
nd crown structure on lichen communities in mature Alpine spruce forests”. Biodiversity
Conservation, vol. 18, pp. 1519–1522, 2009.
[12] S. Martellos, “Multi-authored interactive identification keys: The FRIDA (FRiendly IDentificAtion)
package”. Taxon, vol. 59(3), pp. 922-929, 2010.
[13] L. Poldini, Nuovo Atlante corologico delle piante vascolari nel Friuli-Venezia Giulia. Reg.
Auton. Friuli Ven. Giulia, Udine, 529 pp., 2010.
[14] F. Festi and F. Prosser, La Flora del Parco Naturale Paneveggio Pale di San Martino. Atlante
corologico e repertorio delle segnalazioni. Supplemento agli Annali del Museo Civico di
Rovereto, vol. 13, 438 pp., 2000.

156
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 157-162.
ISBN 978-88-8303-295-0. EUT, 2010.

“Flora Italiana Digitale”:

an interactive identification tool
for the Flora of Italy
Riccardo Guarino, Sabina Addamiano, Marco La Rosa,
Sandro Pignatti

Abstract — The digital facilities of the second edition of Pignatti’s “Flora d’Italia”
are presented. A software, called FID (i.e. “Flora Italiana Digitale”) will link
together a random-access interactive identification tool, a thesaurus, synoptic
tables and one template for each single species, including a distribution map
(referred to the Italian regions), “ecograms”, a text-box and up to 24 high-
resolution colour images. The FID follows a “shareware phylosophy”. All
contents and images can be integrated and/or replaced over time, in order
to continuously improve the diagnostic and qualitative performance of the
provided utilities. Ideally, the community of users should interact on the web,
so that every user could easily become content provider.

Index Terms — Flora of Italy, interactive identification tool.

—————————— u ——————————

1 Introduction

T
he most recent flora of the vascular plants growing in Italy was published
in 1982 [1]. This work consists of 2324 pages in three volumes, where each
of 5599 native and invasive species is briefly described and illustrated
by a regional distribution map and a drawing, the latter mostly taken from the
“Iconographia Florae Italicae” [2].
Since 1982, several changes occurred both in the systematics of vascular
plants and in the floristic exploration of the Country: the records for the
Italian flora raised up to 6700 species [3] and the bulk of knowledge on a
single species includes on one hand more detailed information on ecological
and phytosociological preferences, and on the other hand on molecular and
phylogenetic data and results. Moreover, the public interest and concern for
nature and the biosphere, of which vascular plants are the most visible and
perceivable component (at least for most of the terrestrial ecosystems), has
————————————————
R. Guarino is with the Dept. of Botanical Sciences, University of Palermo, I-90123. E-mail: ric-
[email protected].
S. Addamiano, viale P. Pellini 31, Perugia, I-06124. E-mail: [email protected].
M. LaRosa, via P. Maioli, 36, San Miniato (PI), I-56028. E-mail: [email protected].
S. Pignatti is with the Dept. of Plant Biology, University of Rome “La Sapienza”, I-00165. E-mail:
[email protected].

157
increased consistently in the last three decades also among non-specialists [4].
For these reasons, a second edition of the Pignatti’s “Flora d’Italia” was
planned, making use of the new facilities offered by information technology, in
order to provide an updated inventory where specialists and non-specialists can
easily find the information they search for.
The new work will consist of four volumes with integrated digital utilities and
data-sources, that link together interactive polytomous keys, a thesaurus,
synoptic tables and one template for each single species, including a distribution
map (presence-absence in the Italian regions), “ecograms”, a text-box and up to
24 high-resolution colour images.

2 Digital Utilities for the Italian Flora

Our project started in December 2002. A software, called FID (i.e. “Flora Italiana
Digitale”), has been written in Visual Basic and developed with the inclusion of
client packages for multiplatforms.
The software aims at:
• providing a rich iconography for each of the species of the Italian Flora.
Current settings are limited to max. 24 images/species, but there is virtually
no limit to this number. Images may include digital photographs, optical
and microscopic scannings of diagnostic characters, drawings and any
other depiction that turns out to be useful to identify a given species and to
represent its morphologic variability;
• helping beginners to identify the observed specimens by means of a
random-access interactive identification tool, filtering the species on the
basis of the morpho-anatomical characters selected by the user’s queries;
• letting the user to personalize the contents of the FID, by adding/replacing
texts and/or images, as well as personal data-bases (e.g. own herbarium,
the flora observed during a hike...). These additional contents can be kept
“private only” or, instead, submitted to the scientific committee to be placed
at any other user’s disposal.
• providing exhaustive open-ended descriptions on the species belonging
to the Italian vascular flora, including phylogenetic, morpho-anatomical,
ecological data, conservation issues, as well as information on traditional
uses, common and vernacular names, adversities, importance for human
economy, etc.
Being the digital flora an open-ended work, all contents and images can
be integrated and/or replaced over time, in order to continuously improve the
diagnostic and qualitative performance of the provided utilities. Ideally, the
community of users should interact via web, so that every user could easily
become content-provider, by giving the right to use the provided content in the
frame of the FID. A scientific committee will evaluate the contributions prior to
publication and ensure the visibility of each single contributor. Official updates
will be periodically published online.
Up to now, the digital utilities of the second edition of the “Flora d’Italia” are the
result of the liberal cooperation of more than 150 contributors. One relevant point
is the direct involvement of secondary schools for testing polytomous keys and

158
usability of the contents. A second relevant point is the lack of sponsors, so that
all contents will not only make more accessible the information on plant species,
but also celebrate the praiseworthy synergy of people sharing the same passion
for the beauty of floristic research. In this “shareware phylosophy”, a mutual
aid that has to be particularly mentioned is with the Dryades Project, the Italian
branch of the European project KeyToNature [5]: several contributors collaborate
to both projects and the exchange of know-how and visual contents greatly helped in
the development of the FID.

3 ...How do they work?

Fig. 1 – Opening window of the FID: the options under “trova la tua pianta” (= find
your plant) can be used to go directly to the information on a plant known by the user.
Instead, the option “Cerca la tua pianta” (= search your plant) opens the window of the
interactive identification tool.

Fig. 2 – To identify a specimen, users can select a set of non-hierarchized fields and
options. User’s choices are listed in the left. The filtered species appear by clicking on
“Visualizza figurine” (images) or “Visualizza elenco” (names, printable with “Stampa
elenco”). The centre hosts a table with texts and images (1496 in total).

159
 Fig. 3 – The visual identification of a specimen is possible through the comparison
of thumbnails. A single click will magnify the image. A double-click will open the
informative window of each single species.

Fig. 4 – The informative window includes: a standard text (that can be modified/
personalized by the user); a distribution map; an “ecogram” [6] displaying the ecological
preferences and pollination/dissemination strategies; the thumbs of up to 24 high
resolution images.

160
 Fig. 5 – It is possible to collect up to six selected images in a synoptic table, in order
to compare different features of one or more species. Each image can be zoomed;
colours, contrast and light can be temporarily modified by the user, in order to better
observe diagnostic characters.

Fig. 6 – Some facilities, such as a conceptual map, a thesaurus, the list of common
names can be recalled by the user at any time, in order to make browsing more friendly.

161
Acknowledgement

The authors gratefully acknowledge all the researchers and colleagues that liberally
provided information and images to the Digital Flora of Italy. Without their passionate
support it would not have been possible to collect the (up to now) 80000 digital images
included in our work.

References
[1] S. Pignatti, Flora d’Italia. Edagricole, Bologna, vol. 1-3, 1982.
[2] A. Fiori, Iconographia Florae Italicae, ossia Flora Italiana Illustrata. 3rd ed. Tipografia Editrice
Mariano Ricci, Firenze, 1933.
[3] F. Conti, G. Abbate, A. Alessandrini and C. Blasi (eds.), An annotated Checklist of the Italian
Vascular Flora. Palombi Editori, Roma, 2005.
[4] R. Guarino, S. Addamiano, M. La Rosa and S. Pignatti. “The impact of Information Technology
on the identification of species and archiviation of taxonomic and floristic data”. Bocconea, vol.
23, pp.19-23, 2009.
[5] P. L. Nimis and S. Martellos, “KeyToNature – Dryades” http://www.dryades.eu/home1.html,
2008.
[6] S. Pignatti, H. Ellenberg and S. Pietrosanti, “Ecograms for phytosociological tables based on
Ellenberg’s Zeigerwerte”. Annali di Botanica, vol. 54, pp. 5-14, 1996.

162
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 163-169.
ISBN 978-88-8303-295-0. EUT, 2010.

eFlora and DialGraph, tools

for enhancing identification
processes in plants
Fernando Sánchez Laulhé, Cecilio Cano Calonge,
Antonio Jiménez Montaño

Abstract —This document describes in some detail the eFlora web application,
a powerful tool for the identification of plant species. It incorporates the corpus
of Flora Iberica, a scientific description of the vascular plants living in the
Iberian Peninsula, which is treated as unstructured information and therefore
indexed by a full text search engine tool, in our case Lucene. eFlora also
includes dichotomous keys, which are displayed using Hyperbolic Geometry.
By making intelligent use of the keys, we have created two original and useful
features, the comparison of arbitrarily chosen species, which is resolved by
a dynamic generation of subkeys applied to these selected species, and
the presentation of dichotomous keys in the form of a Virtual Assistant, or
conversational robot, using our solution DialGraph, which allows to non-
academic users an approach in Natural Language, such as chat, or voice
recognition, Text to Speech Synthesis (TTS) or even Automatic Translation
when dealing with a multilanguage context. Concerning the configuration of
the Virtual Assistant, we provide a very intuitive BPM-like graphical design.
This approach to dichotomous keys helps teaching biodiversity science,
enhances the awareness of its importance, and makes citizens emotionally
closer to science.

Index Terms — virtual assistant, hyperbolic geometry, natural language,

dichotomous key, Bot, automatic translation, Flora Iberica, BPM, Text to
Speech Synthesis, TTS.

—————————— u ——————————

1 Introduction

N
owadays there is a major challenge to properly process the large
body of available botanical information, most of which lacks structure
consensus, probably due to both its ancient origins and the inherent
difficulty to structure it in a consistent manner. This document describes in some
detail the eFlora web application, a powerful tool for the identification of plant
species, which incorporates the corpus of Flora Iberica, a scientific description
of the vascular plants living in the Iberian Peninsula. eFlora has been addressed
————————————————
The authors are with Terasoft Consultores S.L., Parque Empresarial Punta Galea
c/.Perú, 4. Loft 3 28230 Las Rozas (Madrid), Spain. E-mail of first author: [email protected].

163
with a full text type approach, and is accessible via an appropriate open source
search engine, Lucene, which allows to launch free text queries, the system
responding with a list of species, ranked in order of relevance. Being Flora
Iberica a work aimed at the identification of species, we have also included
dichotomous keys in graphic form and under hyperbolic geometry. Furthermore,
the original functionality of these keys has been enhanced in the form of
dynamic generation of sub-keys from any arbitrary selection of species. Another
intesting feature, implemented with a dichotomous key to Iberian Conifers, is the
presentation of the key in the form of a conversational robot or virtual assistant,
that interacts with the user in natural language.

2 Flora Iberica

2.1 Descriptions

To give some structure to the information contained in Flora Iberica, while

maintaining the characteristic of free natural language descriptions, the
description part of each species has been separated into a priori-defined fields
(Tab. 1).
Level 1 fields Level 0 fields
Root
Basal stem
Plant
Stems and branches
Spines
Leaves Leaves
Inflorescence
Receptacle
Flower
Calyx
Corolla
Gynaeceum Gynaeceum
Androecium
Androecium
Nectary
Fruit
Fruit
Seed
Tab. 1 – Fields used for organising the descriptions of species in Flora Iberica.

These fields can be grouped into even greater granularity, in anticipation

that some of them may belong to categories of higher order. For example, a
query can be designed as generically directed to the field Flower, or it may be
more detailed, specifying which field in the category Flower the query is to be
launched: Inflorescence, Receptacle, Calyx and Corolla. In this way, queries
can be made either under both full free text scheme, or directed, in free text as
well, towards specific fields.

164
2.2 Dichotomous keys by hyperbolic geometry

The dichotomous keys of Flora Iberica have been digitized. These keys are
specific for all taxonomic levels used in the original work i.e. there are special
keys that identify species within the same genus, keys to identify genera within
a family, etc.
The dichotomous keys have been ported to XML format for processing in the
form of trees in hyperbolic geometry, whose root concept is that it allows to pass
through a point more than one line parallel to another line that does not pass
through that point. With respect to information representation - decision trees in
our case – this means that the hyperbolic space can be represented as a circle
in which the periphery represents infinity. In this geometry, the closer to the edge
(infinity) we are, the smaller is the size of what we represent. This allows to
represent graphs in hyperbolic geometry maintaining the properties of focus and
context. Everything in the center is large, and if we move aside to the periphery,
it becomes smaller, but still visible. For the implementation of hyperbolic trees,
Treebolic, an open source solution, has been used (Fig. 1).

Fig. 1 – General view of dichotomous keys at initial execution from the root node,
displayed using hyperbolic geometry.

3 Indexing

Lucene, an open source solution from Apache Group, was used as the main
search engine. We have created a data model that includes a set of indexed
fields specifically for using in queries. Other fields, such as Flowering periods,
Observations or Bibliography are used only as supplementary information to be
shown when displaying a full description.
As for the ability to query in English, we have added a filter that acts between

165
the user input string and the search engine. In this way, a user not speaking
Spanish, but able to understand botanical information, can query directly in
English.

4 Identification

4.1 By species comparison

While eFlora indexes the data with a full text search engine, queries have
a qualitative difference in relation to those made in a classic form, such as in
Google. In the latter case, some information is sought, and the result can be
found in one or more points from the list of results provided by Google. It is the
surfer who decides what information is to be considered as valid. This means
that sometimes useful information may be scattered among a high number of
returned results. In eFlora the situation is completely different. The question is:
what is the species that we have in our hands? For anwering, it is not sufficient
for the search system to just offer a series of results ranked by relevance. This
is useful, but what is necessary is that the system can tell us what to observe
for differentiating between the items included in the list of results. Otherwise the
user should make a very tedious work examining the full description of every
species in the list, trying to find subtle differences among similar taxa.

Fig. 2 – Example with a list of results ranked by relevance; on the right side, the
repository filled up with three species to be compared.

With eFlora the identification task, which is originated through a query using
terms describing certain observable characteristics of a certain species, begins
with a set of species that have a good chance of containing the targeted
specimen. To differentiate among taxa, users can select those which look more
likely, and bring them to a special repository in which they can perform the
function of comparison (Fig. 2). This function processes the entire tree of the
dichotomous key in real time and extracts a subkey that shows the differences
among the selected species. In this way, the user has a useful tool to identify the
species in question (Fig. 3).

166
Fig. 3 – Once the compare function has been executed, a dynamic subkey is displayed
in hyperbolic geometry.

4.2 The Virtual Assistant

We have developed an original tool for teaching the identification of biodiversity,

based on the use of Virtual Assistants in the form of a conversational bot. A
Virtual Assistant is a software application capable of maintaining a conversation
with users on a specific topic in natural language. We applied DialGraph to
a dichotomous key of Iberian Conifers. DialGraph allows an almost complete
portability between dichotomous keys - considered as a large tree structure - and
its graphical system design. It allows communication with users through Chat,
materialized in a virtual character who pronounces sentences and might even
recognize the voice of the Internet surfer. This system is topologically equivalent
to the traditional dichotomous keys, but for its ability to chain multiple words
appearing in the same sentence, and the freedom to use popular expressions
and to apply automatic translation, it has important advantages for teaching
Biodiversity, mainly when focused towards young users (Fig. 4).
The strongest point of DialGraph is the tool created for designing the
conversation logic, which we call The Designer. This can be accessed through
the web using standard navigators, such us IE, FF, Chrome, etc., and models
the dialog by drag and drop objects within a virtual dashboard. These objects
represent different assistant states and the links among states. By defining in
links as parameters very simple linguistic lexical roots and appropriate messages
in states, the dialog with users remains fully determined as long as the user is
informed about the aim of the assistant (Fig. 5).

167
 Fig. 4 – Moving avatar interacting with users in the web.

Fig. 5 – Designing the flow of dialog for a specific part of the dichotomous key using
DialGraph as a BPM tool.

168
5 Conclusion
TeraSoft Consultores SL has created, together with the Botanic Garden of
Madrid, a set of tools which enhances the identification of plant species. Based
on the digitalization of botanic data “as they are” in the classic form, including
dichotomous keys, it permits to approach science to citisens and to effectively
teach the importance of Biodiversity.

References
[1] E-Flora Iberica: http://www.efloraiberica.es/eflora/, 2010.
[2] DialGraph: http://www.dialgraph.com, 2010.
[3] TeraSoft Consultores SL: http://www.terasoft.es, 2010.

169
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 171-175.
ISBN 978-88-8303-295-0. EUT, 2010.

A catalogue of bird bones:

an exercise in semantic web
practice
Gudmundur Gudmundsson, Seth D. Brewington,
Thomas H. McGovern, Aevar Petersen

Abstract — The vast databases of natural history collections are increasingly

being made accessible through the internet. The challenge is to place this
data in a wider context that may reach beyond the interests of scholars only.
The North Atlantic Biocultural Organization and Icelandic Institute of Natural
History are jointly developing a web based catalogue of bird bones, comprising
digital images, and related information from the museum database. Linking
the bird bone catalogue with the semantic web developed by STERNA will
integrate the bird bone catalogue with diverse information on birds that is
directed towards the general public.

Index Terms — bird bones, collections, natural history, semantic web.

—————————— u ——————————

1 Introduction

N
atural history museums have a long tradition of collecting and
preserving specimens primarily for scientific study and also for public
education and exhibition. Museum research has over time generated
vast collections of specimens and related information, which was mostly
accessible only to a closed community of researchers. The main reason was that
collection information was registered on paper files and index cards. Information
retrieval was thus confined to trained personnel only. Present museums almost
universally rely on computer databases to register and update information of
their vast collections, and almost all institutions are in the process of digitizing
information on paper and card files. In addition to this, diverse information that

————————————————
G.Gudmundsson is with the Icelandic Institute of Natural History, Hlemmur 3, 125 Reykjavik,
Iceland. E-mail: [email protected].
S.Brewington is with the Graduate School and University Canter, CUNY, New York, N.Y. E-mail:
[email protected].
T. McGovern is with the Graduate School and University Canter, CUNY, New York, N.Y. E-mail:
[email protected].
A. Petersen Is with the Icelandic Institute of Natural History, Hlemmur 3, 125 Reykjavik, Iceland
E-mail: [email protected].

171
has resulted from scientific studies on specimens is often integrated to museum
databases, like morphological measurements, chemical analyses and digital
photos. Many collection databases have thus grown to comprise numerous data
tables that share one or more common data fields. For example, a species name
is commonly present in data tables that store information on collection locality,
morphological measurements, chemical analysis, photos, etc. This conceptual
link between data tables creates novel possibilities of filtering out information
from diverse datasets, like species diversity at some locality, correlations in
species distributions, etc. However, collection databases with clearly defined
relations between data sets and standardized data entries can not only be
utilized for research, but may also be utilized for educational purposes through
the web. [1]
Effective dissemination of information entails making information readily
available in such a way that users with different needs have the ability to
comprehend it. Information on museum collections should therefore be provided
in layers of successively greater depth and detail and in a variety of different
contexts. Such a “virtual museum” may be defined as a means to establish
access, context, and outreach by using information technology and to establish
interactive dialog with users. [2]

2 The STERNA Project: Dissemination of Collection Information

Over the last ten years, ever more cultural and scientific resources have been
digitised and made accessible on the internet. However, integrated semantic
search and access to these resources that are hosted in many heterogeneous
databases is still difficult to achieve. The vision of the STERNA project is to
provide cultural and scientific heritage institutions the opportunity to make their
digital collections accessible in a light weight fashion. This will be achieved by
setting up a distributed digital library that is based on semantic web technologies
and standards, such as RDF (Resource Description Framework) and SKOS
(Simple Knowledge Organisation System). STERNA is especially designed
for small and medium sized institutions with limited budget and technical staff.
Thirteen European cultural heritage institutions, multimedia archives, technology
providers and research organizations, from ten countries, are participating in the
STERNA project.
A network of (semantically) related digital resources is accomplished by
connecting data provided by several institutions through a reference structure,
which comprises several kinds of “controlled vocabulary”, from simple word
lists and glossaries to taxonomies, thesauri and ontology. The architecture of
STERNA is thus based on distributed data repositories, which are semantically
connected into a network that can be searched from different perspectives
(faceted search). The network can also be extended by adding new members
as well as tools, instruments and guidelines. Individual organisations can thus
connect to a wider network of content holding organisations and place their data
in a wider and more general context [3].
The STERNA project is partly funded the European Commission’s

172
eContentplus programme. It started in June 2008 and will end in May 2011, and
the participating institutions are:
1. Salzburg Research, Austria (Coordination)
2. Archipelagos, Greece
3. DOPPS BirdLife Slovenia
4. Heritage Malta
5. 5.Hungarian Natural History Museum
6. Icelandic Institute of Natural History
7. Nat. History Museum of the Municipality of Amaroussion Greece
8. Natural History Museum of Luxembourg
9. Naturalis, Natural History Museum of the Netherlands
10. Netherlands Institute of Sound and Vision
11. Royal Museum for Central Africa, Belgium
12. Teylers Museum, Netherlands
13. Wildscreen/ARKive, UK
14. Trezorix, NL

The data sources are of different types and sizes, from natural history
museums, audiovisual archives, research institutions and nature conservation
agencies. The vision is to create a dispersed network of information nodes,
where each is supported and sustained by a member institution. For practical
reasons (limited funds and staff) the STERNA project is focused on data access
on birds, although the general objective is to extend the network to serve
worldwide audience with more general interest in nature and wildlife. The aims
are to:
1. Offer substantial amount of data by the combined effort of several institutions.
2. Linking the data in semantic context.
3. Providing advanced site functionalities, such as faceted search.
4. Offering possibilities for users to contribute additional data.

3 The Bird Bone Catalogue

Bird bones are a constant source of interest both to nature observers and
professionals. Excavation of archaeological sites often yields rich assemblages
of zoological remains. Zooarchaeological studies in Iceland have indicated
that early settlement relied to a greater extent on hunting, than in later periods.
Prior to the eleventh century the faunal remains include a large number of fish
and wild birds in addition to domesticated animals, while in later periods the
presence of wild fauna drops dramatically [4], [5], [6], [7].
Accurate identification of bird bones often requires consultation of reference
specimens in natural history collections. However, this usually requires visit to
the museums, which can be time consuming and expensive. Internet access
to photos of zoological bone specimens and associated information facilitates
proper identification. If these bone data are linked to a wider source of information
on birds, it may be of use to a much larger audience than a closed research
community. It is on these grounds that The Icelandic Institute of Natural History
and NABO, in association with STERNA, decided to develop internet access to

173
a catalogue of bird bones with conceptual relations to diverse bird information
of cultural significance.
NABO (North Atlantic Biocultural Organization) was founded over 20 years ago
in an attempt to cross-cut national and disciplinary boundaries of researchers in
several fields of studies, like archaeology, biology, geology, and anthropology.
NABO has worked to aid in improving basic data comparability, in assisting
practical fieldwork and interdisciplinary ventures, in promoting student training,
and in dissemination of knowledge to other scholars, funding agencies, and the
general public. [8]
The objective of the bone catalogue is to provide basic information on the
internet to aid identification of bird bones. Photos and associated information
are registered and maintained in a relational database, which comprises three
primary data sets:
1. Taxonomical classification of birds, including scientific and vernacular
names in many languages
2. Photographs and descriptions of the major bones in 54 species that have
a long tradition of being utilized by humans and are of cultural relevance.
3. Specific descriptions of individual bones of a particular species – along with
a simple general directory of the major bones in birds: e.g. the skull, bones
of the wings and legs, keel (sternum), pelvic girdle, furcula and coracoid.
Associated with these primary data sets are two secondary (supportive) data
tables:
1. Inventory of available reference specimens of a particular species at the
IINH
2. Exhaustive registry of literature references on Icelandic bird fauna.
The bone catalogue will open on the web by the end of 2010. It is not intended
as a conventional identification key, as it does not provide stepwise guidance
to reach a final identification. Instead, it provides two search options when
looking for images of bird bones: 1) taxon name (species, genus or family) and
2) type of bone. These can be used separately or in combination. A search that
is limited to, say, one type of bone and a single genus will filter and display all
images of skulls of that genus. The associated (secondary) data tables provide
an optional inventory of available specimens at the IINH and a fairly complete
list of literature with relevance to Icelandic populations of birds.
Information on bird bones in museum collections, literature and inventory
of museum specimens are not likely to interest others than archeologists and
ornithologists, with focused reasearch interest on that subject. The intention
with the cooperation between STERNA, NABO and IIHN is to enrich the bone
catalogue by making it a part of a diverse semantic network. Images of bird
bones would then be accessible in a conceptual context to bird enthusiast,
ordinary nature observers, as well as outreach and educational programmes.

Acknowledgement

The authors wish to thank Kjartan Birgisson for technical computing and setting up the
prototype of the bone catalogue. We are also indebted to Þorvaldur Björnsson for his
effort in selecting suitable bones for photographing.

174
References

[1] G. F. MacDonald, G. F. and S. Alsford, “The Museum as Information Utility”. Museum

Management and Curatorship, vol. 10, pp. 305-311, 1991.
[2] W. Scweibenz, “The Virtual Museum 1: New Perspectives For Museums to Present Objects
and Information Using the Internet as a Knowledge Base and Communication System”, In: H. H.
Zimmermann and V. Schramm (eds.), Knowledge Management and Kommunikationssysteme,
Workflow Management, Multimedia, Knowledge Transfer. Proceedings des 6. Internationalen
Symposiums für Informationswissenschaft (ISI 1998), Prag, 3. – 7. November, pp. 185-200,
1998.
[3] A. Mulrenin, STERNA: Semantic Web-based Thematic European Reference Network
Application, Annual Report 1, 1st June 2008 – 31st May, Salzburg Research
Forschungsgesellschaft m.b.h, 2009.
[4] C. M. Tinsley, “The Viking settlement of Northern Iceland”. In: R. A. Housley and G. Coles
(eds.), Atlantic Connections and Adaptations: Economies, environments and subsistence in
lands bordering the North Atlantic. Oxford: Oxbow Books, pp. 191-202, 2004.
[5] T. H. McGovern, S. Perdikaris, A. Einarsson and J. Sidell, “Coastal connections, local fishing
and sustainable egg harvesting: Patterns of Viking Age inland wild resource use in Myvatn
district, Northern Iceland”. Environmental Archaeology, vol. 11, pp. 187-205, 2006.
[6] T. Amorosi, “Icelandic archaeofauna: A preliminary review”. Acta Archaeologica, vol. 61, pp.
272-84, 1991.
[7] T. Amorosi, P. Buckland, A. J. Dugmore, J. H. Ingimundarson and T. H. McGovern. “Raiding
the landscape: Human impact in the Scandinavian North Atlantic.” Human Ecology, vol. 25, 3,
pp. 491-518, 1997.
[8] http://www.nabohome.org, 2010.

175
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 177-181.
ISBN 978-88-8303-295-0. EUT, 2010.

Anthos.es: 10 years showing

Spanish plant diversity
information in the Internet
Leopoldo Medina, Carlos Aedo

Abstract — Anthos is a scientific and technological program that shows

Spanish plant biodiversity information in Internet through the open web page
www.anthos.es. The program has been developed since 1999 with the support
of the Fundación Biodiversidad and the Agencia Estatal Consejo Superior de
Investigaciones Científicas – Real Jardín Botánico and offers 1.3 million plant
data with c. 100.000 annual visitors.

Index Terms — Anthos, Spanish plants, biodiversity, internet.

—————————— u ——————————

1 Introduction

A
nthos is a scientific and technological program that was developed
in accordance with a specific agreement between the Fundación
Biodiversidad (Biodiversity Foundation, from the Ministry of the
Environment) and the Agencia Estatal Consejo Superior de Investigaciones
Científicas (Spanish National Research Council) – Real Jardín Botánico (Royal
Botanic Garden, from the Ministry of Science and Innovation) in order to show
assorted information about the plant diversity of Spain in the Internet.
The program was initiated in 1999 with a public computer application that has
been continually updated up to 1,3 million data concerning plants, using as its
main source the Spanish botanical bibliography. In 2005, in accordance with the
second agreement for the development of the project, a new computer
application was added, developed in a Geographic Information System (GIS),
which became accessible to the public in April 2006. A detailed description of the
previous Anthos system can be consulted at [1].
The new application integrates and improves the procedures and queries of
the previous application, consistently increasing the amount of available data.
Furthermore, the new application combines chorological information with other
information of a cartographic nature concerning environmental variables, and
reference maps. This allows a more accurate location of cited plants, as well as
showing in graphic form the different distribution patterns.
————————————————
The authors are with the Real Jardín Botánico, CSIC, Plaza de Murillo 2, 28014 Madrid (Spain).
E-mail: [email protected], [email protected].

177
 The overall geographical environment chosen for the project is the Iberian
Peninsula and the Macaronesian islands (Canaries, Madeira and the Azores)
as a representation of each of the biogeographical units present in Spain. In this
way, the distribution of a taxon may be studied throughout the entire national
territory and surrounding area, fully integrating the taxon in its geographical
component.

2 Taxonomic Information
The taxonomic framework for the use and management of the names of plants
is still the project Flora iberica (www.floraiberica.es), which provides essential
knowledge in the areas of taxonomy, chorology, cytology, etc. Anthos follows its
criteria faithfully.
Thus, the offered taxonomic treatment is the following:
1. Plants of the Iberian Peninsula and Balearic Islands, within genera
which were already published or are in the process of being published
in Flora iberica. For all other genera, the treatment follows firstly that
of Med-Checklist [2] and the rest Flora Europaea [3], except in cases
where a specific and original treatment is followed.
2. Plants of the Canary Islands, and in general for the entire Macaronesian
area, following the taxonomic scheme offered in the Lista de Especies
Silvestres de Canarias [4].

3 Chorological Information
The distribution maps of plants have been sketched from the chorological
information published in scientific articles and books, along with data from
herbarium collections, reviewed by specialists who submit their data.
The initial bibliographic information came from the database of chorological
citations which the Royal Botanic Garden (CSIC) began to prepare in 1986. This
original information was cleaned up and later greatly extended, thanks to the
Anthos project, until reaching its current number of 1.3 million entries.
Data from herbarium collections are received from critical reviews normally
carried out by authors of genus syntheses for Flora iberica, who submit their
data to us. In some particular cases, herbarium data for some plants have been
added, in order to complete their distribution. The sum of the plant data from
herbaria is close to 36,000 sheets.
Recently, we have also added a great amount of duly contrasted information
from other databases. This information is shown as it was provided (with the
obligatory adaptation for formats), and the source is cited in each record,
so that it can be duly identified. We currently have access to databases as
the database of “Plantas vasculares de los Parques Nacionales” (Vascular
plants of the Spanish National Parks), from the the Organismo Autónomo de
Parques Nacionales, (Autonomous Organism of National Parks) of the Ministry
of the Environment, or the database of the “Plantas vasculares de la cornisa
cantábrica” (Vascular plants of the Cantabrian Cornice), submitted by C. Aedo,
G. Moreno Moral and ó. Sánchez Pedraja, members of the group of experts in

178
botany for the northern part of the Iberian Peninsula, which brings together the
great effort this work group has undertaken in the geographical area comprising
the Cantabrian Mountains. This database has provided about 300.000 plant
references, noticeably improving the quality of the data offered in the area.
Some of the chorological data of plants collected in the botanical bibliography
have shown, over time, to be somewhat unreliable. In these cases, although we
are obliged to display them to the public, we have marked them with the label
“questionable” so that the user may be aware of the fact that the citation needs
verification of some type. This label appears in a distinguishable form both on
the distribution maps and on the lists.

4 Associated information
Besides the distribution maps for the plants, we have incorporated other
information which may be of great interest to users, such as: common
names, chromosome numbers, synonymy, conservation state, drawings
and photographs. The common names were initially taken from the volumes
published in Flora iberica, to which the information contained in the database
“Nombres vernáculos” (Common names) has been added, gathered and
updated by Dr. Ramón Morales (Real Jardín Botánico, CSIC) and his team.
This information has been updated in collaboration with Anthos. The information
regarding chromosome numbers comes from a previously published database
at the Flora Iberica webpage, which was subsequently updated with the most
recent bibliography. The information on synonyms of the accepted names comes
from the database NOMEN of Flora Iberica. For the genera not studied in this
project yet, the system of nomenclature employed by Med-Checklist and Flora
Europaea has been adapted to the structure of Anthos, with the aim of obtaining a
homogeneous nomenclature database. The information about the conservation
state comes from a newly created database in which we have included the
Legal Standards on the protection of plants effective in the Spanish territory,
together with information on books and red lists. Further information about plant
conservation status can be consulted at www.phyteia.es. The illustrations we
offer were submitted from several sources: the black and white plates were
provided by Flora Iberica and were created by different botanical artists. The
coloured plates were submitted from other classical works on the Iberian and
Macaronesian flora which, due to their antiquity, are no longer subject to authors’
or editors’ rights. The photographs of the plants were acquired or submitted
from diverse artists, whose names appear on the caption of each photo, and
which are also responsible for identification of the photographed specimens.
In some cases, due to our interest in completing certain collections of images
of plants in a geographical or taxonomic area, these photographs were taken
within the Anthos project itself, in which case we assume all responsibility for the
identification of the plants displayed.
Download Information. Under the epigraph “listings”, Anthos has developed a
format for the output of data for each consultation on the distribution of a plant.
Thus, the user has access to the information that backs up each of the citations.
This relation may be downloaded in different formats (txt, csv and xml), which

179
allows subsequent editing, using the usual geographical and statistical tools.
Cartographic Information. The Cartographic information comes from free
public services or was submitted by colleagues.
Google Maps is loaded with the corresponding licenses, as is Blue Marble,
and also the climatic variable layers provided by Atlas Climático de la Península
Ibérica (Climatic Atlas of the Iberian Peninsula, from UAB).
The Banco de Datos de la Naturaleza (Nature Data Bank), of the Ministry of
the Environment, provided us with the UTM grid, which we later extended to the
whole area, with information corresponding to Spanish National Parks.
The information in the Geological Map was taken from the SEIS.NET program,
Sistema Español de Información de Suelos de España sobre Internet (Spanish
Information System of Spanish Soils on the Internet, IRNA-CSIC).
We obtain WMS remote visualisation of the orthophotos of the Rural Register
management tool, known as SIGPAC (System of Identification of Agricultural
Plots), for which we have availed ourselves of the generous help of those
responsible for the above-mentioned application in TRAGSATEC.
The Instituto Geográfico Nacional (Spanish National Geographic Institute)
suggested and allowed us the use of the WMS service to load layer information
provided within the framework of the IDEE - Infraestructura de Datos Espaciales
de España, Ministerio de Fomento (Spatial Data Infrastructure of Spain, Ministry
of Public Works)
The DEM (Digital Elevation Model) was made up by Geodata S.L. from
GTOPO30.

5 Use of Information
The information provided in the Anthos Project is distributed on the Internet
freely to the broad public for the benefit of whoever may wish to use it; Anthos
accepts no responsibility for its reliability, which is the sole responsibility of the
authors of the chorological, taxonomic and photographic works.
However, the compilation and management of the above-mentioned
information is the work of Anthos, and we should be grateful to be cited as
an electronic resource in scientific, technical and professional public outreach
works which have availed themselves of the data offered by the program.

Acknowledgements

Since 1999, the year in which Anthos, was initiated, a great number of users have
generously collaborated with us, informing us of errors or faults detected or offering
information of interest. To all of them, as well as to the institutions and the group of
consultants and collaborators both within the Real Jardín Botánico or external to it, we
owe our deepest thanks for the help which, from the outset and up to this day, we have
so generously received. We hope to keep counting on the collaboration of any user, as
we are very much aware that this is one of the best ways of correcting and updating
the extremely complex information in our pages, and that without the assistance of our
collaborators the task would be indeed much more difficult.

180
References
[1] S. Castroviejo, C. Aedo and L. Medina, “Management of floristic information on the Internet:
the Anthos solution”. Willdenowia, vol. 36, pp. 127-136, 2006.
[2] W. Greuter, H. M. Burdet and G. Long (eds.), Med-Checklist. A critical inventory of vascular
plants of the circum-mediterranean countries 1,3,4. Conserv. Jard. Bot. Genève, 1984-2008.
[3] S. Castroviejo (ed.), Flora iberica. Real Jardín Botánico – Consejo Superior de Investigaciones
Científicas. Madrid, 1980-2009.
[4] I. Izquierdo, J. L. Martín, N. Zutita and M. Arechavaleta (eds.), Lista de especies silvestres
de Canarias. Hongos, Plantas y Animales terrestres. Consejería de Medio Ambiente y
Ordenación Territorial, Gobierno de Canarias, 2004.

181
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 183-187.
ISBN 978-88-8303-295-0. EUT, 2010.

An interactive tool for the

identification of airborne
and food fungi
Giovanna Cristina Varese, Antonella Anastasi,
Samuele Voyron, Valeria Filipello Marchisio

Abstract — The growth of fungi may result in several kinds of food-spoilage:

off-flavours, discolouration, rotting and formation of pathogenic or allergenic
propagules. Moreover many foodborne fungi produce mycotoxins and thus
fungal growth in foods and feeds should be avoided. Much interest has also
grown for the fungi present in indoor environments, since exposure to airborne
biological agents in both the occupational and residential environments could
be associated with a wide range of adverse health effects with major public
health impact, including infectious diseases, acute toxic effects, allergies and
cancer. An interactive identification tool was created for food- and airborne
microfungi at the genus and/or species level, based on morphological and
physiological data, using the software FRIDA. The interactive key can also
be stored on CD- or DVD-roms, or used on media such as PocketPCs of
Smartphones. Our key allows the identification of 59 genera/groups and 217
species belonging mainly to Zygomycota and anamorphic and teleomorphic
Ascomycota. The database comes with a set of detailed descriptions of each
genus and species, a rich archive of images, a glossary of the most frequent
mycological terms, and references to descriptions; in addition, culture
condition requirements for identification are provided.

Index Terms — Airborne fungi, food spoilage, fungal identification, indoor

fungi, interactive keys.

—————————— u ——————————

1 Introduction

T
oday a growing interest of microbiologists is turned to the study of fungal
contaminants of food and air. The growth of fungi may result in several
kinds of food-spoilage: off-flavours, discolouration, rotting and formation of
pathogenic or allergenic propagules. Moreover, many foodborne fungi produce
mycotoxins, and thus fungal growth in foods and feeds should be avoided [1],
[2], [3]. In the last decades, much interest has also grown for the fungi present

————————————————
The authors are with the Dipartimento di Biologia Vegetale, Università degli Studi di Torino, viale
Mattioli, 25, 10125 Torino, Italia. E-mail of the first author: [email protected].

183
in indoor environments, since exposure to airborne biological agents in both
the occupational and residential environments could be associated with a
wide range of adverse health effects with major public health impact, including
infectious diseases, acute toxic effects, allergies and cancer [4], [5].
Although food- and airborne fungi, producing toxins or which cause health
hazards, are ubiquitous and belong to the common contamination flora, their
recognition is hampered by an incomplete and often confusing literature [1].
Besides, the still poor understanding of many taxonomical groups and the high
degree of pleiomorphism in response to environmental changes call for endless
floristic, taxonomic and nomenclatural updating.
Moreover, in most of the available books for the identification of fungi the
layout follows a hierarchic approach mainly based on classification; hence this
approach requires a deep theoretical and practical knowledge of mycology. Even
more than for other organisms, therefore, for fungi computer-aided systems are
important for the handling of data useful in their identification and as flexible as
possible, not necessarily founded on traditional systematic criteria.
We have created an interactive tool for the identification of food- and
airborne microfungi at a genus and/or species level. This computer-aided
tool can provide access to and simplify the study of fungi by various kinds of
users: mycologists, but also those concerned with environmental hygiene (i.e.
microbiologists employed in food or pharmaceutical industries), those seeking
to create interactive floras, those concerned with the management, planning
and conservation of natural resources, and teachers at each educational level.

2 Software
Our interactive identification tool stems from databases created on the basis
of morphological, physiological and ecological data of each taxon, using the
program FRIDA [6]. Procedures and functions are written in PL/SQL language,
running on a Oracle Database engine. FRIDA is flexible, its use does not
require the learning of any programming language nor the use of codes to input
information and can automatically generate both interactive identification tools,
accessible online, and traditional paper-printed identification keys. The keys can
be immediately published in the web, and an accessory software was developed
to store stand-alone versions on CD- or DVD-ROMs, PDAs (Personal Digital
Assistants), and smartphones.
As with most programs for interactive identification, the keys produced by
FRIDA are based on a hierarchy of characters, taxa being separated on the
basis of those come first in the hierarchy. In our keys, characters are ranked
according to the simplicity of observation: macroscopic features of colonies, type
of mycelium, presence of ascomata or zygospores, aspect of conidiophores,
conidiogenous cells and conidia, etc.

3 The food- and airborne fungi database

The system has produced an interactive key to food- and airborne microfungi
which allows the identification of 59 genera and 217 species belonging mainly to

184
Zygomycota and anamorphic and teleomorphic Ascomycota. The database has
detailed descriptions of each genus and species, coupled with a rich pictorial
archive of macroscopic and microscopic characters, a brief introduction to the
features of the main fungal phyla, explanations of how to cultivate and examine
fungi preparing microscopical slides, a glossarium of the more cited mycological
terms, and references to descriptions and culture condition requirements.
The interactive keys are usable in two different ways [7]: 1) a simple
identification tool based on a traditional dichotomous system, in which the user
selects between two options which are explained by means of descriptions,
pictures and drawings of the different characters (Fig. 1), 2) a multi-entry query
interface in which the user can operate simultaneously a non-hierarchical
choice of one or more different characters; FRIDA will select all the taxa with the
selected (tagged) characters, and for them it will produce a dichotomous key
coupled with pictures of each taxon.

Fig. 1 – Example of a dichotomous key in which the selection between different options
is supported by pictures, drawings and descriptions of the most difficult terms.

At the end of an identification process, the system displays a taxon (genus/

species) page, reporting the scientific name, a description, and any other
information the author has stored in the system (habitat, micotoxins, etc.), as well
as an image, and a link to all the images of the same species stored in the image
archive. Another important tool of FRIDA is the possibility to insert in the database
geographical and ecological data, physiological features, data on the association
of a fungus to peculiar substrates, etc. These data can be used to create “filters”
which are specific identification pathways encompassing only the species which
share the selected character. The “filters” are very useful, since they simplify the
identification pathway by reducing the set of species included in a key.

185
4 Discussion and Conclusions
Since contamination of fodder and foodstuffs and inhalation of propagules
suspended in the air exposes people and animals to health risks because of
the presence of species producing toxins and MVOC or causing allergies or
infections, the use of our key could be useful for biologists working in local health
units and similar organizations, as well as in the checking of quality control
and environmental hygiene. The prevention of fungi that contaminate indoor
environments and cause food spoilage can only be carried out successfully, if
the fungal species are known [1]. Knowing the properties of the contaminant
species makes it possible to optimize the preservative profile of the food and the
hygienic measures in the indoor environments.
However, the identification of these microfungi by means of traditional methods
still remains problematic, and exclusively accessible only to a small number
of experts. Computer-aided tooks can create a revolution, since they use, in
a multi-dimensional way, a wealth of morphological and physiological data,
plus the ecological information usually hidden in the large ocean of scientific
literature. Traditional keys have several drawbacks that can be avoided by
computer-aided tools [7]:
1. Being printed on paper, their content is frozen and hence nomenclatural-
taxonomic changes and the discovery of new species render them rapidly
outdated. Computerised systems, on the contrary, can be updated and
corrected in real time.
2. Traditional keys are rigid. They contain a huge amount of information which
is fixed into the format and the logical structure selected by the author.
Computerised tools permit to reduce the set of organisms using different
combinations of morphological, physiological, ecological, distributional
characters i.e. special habitats, mycotoxin production and physiological
features (temperature, water activity, pH…).
3. Databases are accumulative. A small database can be the starting point for
future expansions.
4. Outputs can be edited in several different formats, from simple texts to
illustrated books.
In conclusion, our key, especially if integrated with existing systems based
on physiological and molecular criteria, could promote the identification of this
important group of organisms even by unskilled persons who lack specific
mycological expertise.

Acknowledgement

The authors wish to thank Prof. P. L. Nimis and Dr. S. Martellos for their support in the
creation of the database and of the key. This work was supported by a grant from MIUR
(Ministero Istruzione, Università e Ricerca).

References
[1] R. A. Samson, E. S. Hoekstra, J. C. Frisvad, Introduction to food- and airborne fungi,

186
 Centraalbureau voor Schimmelcultures (CBS), 389 pp., 2004.
[2] C. V. Blackburn, Food spoilage microorganisms Woodhead Publishing, 712 pp., 2006.
[3] A. D. Hocking, J. I. Pitt, R. A. Samson and U. Thrane, Advances in food mycology, Springer,
371 pp., 2006.
[4] B. Simon-Nobbe, U. Denk, V. Pöll, R. Rid and M. Breitenbach, “The Spectrum of Fungal
Allergy” Int. Arch. Allergy Immunol., vol. 145, pp. 58–86, 2008.
[5] J. Brett, J. Green, E. R. Tovey, J. K. Sercombe, F. M. Blachere, D. H. Beezhold and D.
Schmechel, “Airborne fungal fragments and allergenicity” Medical Mycology, vol. 44, pp. 245-
255, 2006.
[6] S. Martellos, “Multi-authored interactive identification keys: The FRIDA (Friendly IdentificAtion)
package”, Taxon, vol. 59 (3), pp. 922-929, 2010.
[7] S. Martellos and P. L. Nimis, “KeyToNature: Teaching and learning biodiversity; Dryades, the
Italian experience”, Proceedings of the IASK International Conference Teaching and Learning,
pp. 863-868, 2008.

187
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 189-193.
ISBN 978-88-8303-295-0. EUT, 2010.

The Estonian eFlora

Tiina Randlane, Malle Leht, Andres Saag

Abstract — The Estonian eFlora is an example of the new e-learning

tools prepared by the KeyToNature consortium. It is an interactive digital
identification key for more than 1000 plant species recorded from Estonia.
The tool is freely available on the internet - in Estonian and English - and has
two interfaces (dichotomous and multi-entry), which allow the identification of
species using different approaches. Another tool developed by KeyToNature,
the OpenKeyEditor, allows users not only to edit the text of the existing master
key, but also to produce mini-keys restricted to smaller subsets of taxa (e.g.
the plants of a park, or a scool garden), and to add user-generated content to
them. The reaction to these tools from public media and educational circles
has been very positive.

Index Terms — education, Estonia, e-tools, identification, key, plants.

—————————— u ——————————

I
1 Introduction
n the framework of the three-year EU project KeyToNature (http://
www.keytonature.eu/), where partners from eleven European countries
collaborate to produce practical, user-friendly identification tools targeted
at the wide audience of teachers and learners, national and/or local keys have
been developed for each participating country. Many digital interactive keys
for vascular plants and lichens were created using software FRIDA (FRiendly
IDentificAtion) and the databases for Italian plants (Dryades) and lichens (Italic)
developed at the University of Trieste, Italy [1], [2].
The new e-learning tools produced by KeyToNature have several advantages
when compared to the traditional key-books, allowing to widen the users’ circle
considerably. The new tools can be produced automatically and rapidly by a
computer, since the characters are stored in a database. They are created giving
more weight to those easy-to-observe characters which make the identification
easier to laypersons, such as the colour of flowers, the position and shape of
leaves, etc. Once published online, the resulting keys can be updated or edited
easily and in real time. Furthermore, there are almost unlimited possibilities to
use pictures and drawings to illustrate both the diagnostic characters and the
species.

————————————————
T. Randlane and A. Saag are with the Institute of Ecology and Earth Sciences, University of Tartu,
Lai 38/40, Tartu 51005, Estonia. E-mail: [email protected], [email protected].
M. Leht is with the Institute of Agricultural and Environmental Sciences, Estonian University of Life
Sciences, Kreutzwaldi 5, Tartu 51014, Estonia. E-mail: [email protected].

189
 Estonia is a small North-European country (ca 45 000 km2, population ca 1,4
million, with a vascular flora of about 1500 species) which was selected as a
case study. We wanted to test: 1) whether the database of Italian plants of
Dryades, including ca 7000 species, is suitable for creating identification keys in
other parts of Europe (especially in its northern areas), and 2) whether a digital
national-level identification tool would be accepted and practically used by a
wide circle of professional and non-professional users, including schoolchildren
and students.

2 The Estonian eFlora

2.1 General data

The Estonian eFlora is an interactive digital identification key for ca 1100 plant
species (out of ca 1500 native taxa recorded from Estonia [3]). The key also
includes, besides indigenous and naturalised taxa, ca. 70 species of introduced
trees and shrubs for allowing users to get acquainted with the urban forests.
Some taxa which are difficult to separate even for specialists (e.g. species from
the genera Alchemilla, Crataegus, Hieracium, Rosa, Salix, Taraxacum etc.) are
excluded from this key, as well as several very rare species.
The key is currently freely available online in Estonian and in English (http://
dbiodbs.univ.trieste.it/carso/chiavi_pub21?sc=368). It has both a dichotomous
and a multi-entry interface, which allow identification using different approaches.

2.2 Dichotomous interface

The dichotomous interface is the main instrument of the key, that permits
determination of taxa by selecting step by step between two states of a character.
For making a selection, one has to click on the corresponding statement-button,
after which the next pair of statements is displayed, and so on. A great advantage
for beginners is that most character states are richly illustrated by drawings and
pictures. At every step of the identification process one can see the number of
remaining taxa; clicking on this number, the list of remaining taxa is displayed.
At the end of the identification process, the name and a picture of the identified
species are shown, and by clicking on the name a taxon page will open.

2.3 Multi-entry interface

In the multi-entry interface the user can choose several characters of a plant
in a single step. One just has to specify the characters, click ‘submit’ and wait
for a few seconds. The system “filters” the key and gives back a usually strongly
reduced list of taxa and – upon request - a smaller dichotomous key for them.
This interface can be particularly useful for more expert users, e.g. those who
already know the family or the genus of a plant, since it can also produce keys
for all species within a family or a genus. It is also the quickest way to go directly
to a taxon page by just typing a species’ name.

190
2.4 Taxon pages

Taxon pages have been compiled for each species to provide important
information. The pages in Estonian are more informative than those in English,
since they contain short descriptions with the main diagnostic characters,
distributional and ecological data, as well as the conservation status of the
plant. For most species, distribution maps from the Atlas of the Estonian flora
[4] are also displayed. Another attractive feature of the key is that numerous
illustrations are available for each species. By clicking on an image, this is
strongly magnified, showing even the smallest details. The Dryades picture
archive presently includes ca. 63.000 pictures of ca. 7300 infrageneric taxa,
and is being continuously enriched with new photos and drawings, which in real
time become visible also in the online version of the Estonian key.

2.5 OpenKeyEditor

One of the main aims of KeyToNature is that of introducing computer-aided

identification tools in the educational systems of Europe, including elementary
schools. Therefore, one of the main goals is that of rendering our tools as easy
as possible. The “difficulty” of a key depends on several factors, such as the
selection of characters, the type of interface, the terminology which is used.
However, it is obvious that smaller keys (those with fewer species) are generally
easier than larger ones. We can now automatically produce identification tools
restricted to small subsets of taxa, thanks to the OpenKeyEditor, which we are
using also for the Estonian eFlora. This tool allows users to:
1. view and edit the text of the key;
2. add or remove a dichotomy;
3. create smaller keys which are filtered from the master key;
4. automatically generate stand-alone keys for computers and/or smartphones;
5. add user-generated content to the new mini-keys (e.g. adding photos,
inserting new text, modifying the terminology, etc.).

Modifying the text of an existing key is easy: it does not require any knowledge
of informatic codes or languages, but just the use of a common web browser:
one has just to type the changes into the appropriate window. A further function
of the OpenKeyEditor permits to create a filter for generating a mini-key (e.g. a
key to the plants found in a region of Estonia or in a pond near a school). The
filter is just a list of species. To create it, one has just to flag them in a page which
lists all taxa included in the Estonian eFlora. The generation of a new mini-key
from a filter is easy as well: once the filter is ready, with a single command
(“make a key from a filter”) the mini-key is generated in a few seconds.
The filtered mini-keys are visible online in real time, since they are produced
and hosted by a KeyToNature server. However, users may want a stand-alone
version of their mini-key. The OpenKeyEditor of KeyToNature can produce three
different types of stand-alone versions: 1) a CD-Rom version, usable on any
computer, 2) a version for PDAs, 3) A version for the i-Phone, which can be
disseminated via iTunes.

191
 The use of the OpenKeyEditor for the Estonian eFlora has - for the time being
- a restricted access. If you want to use it, please send an e-mail to the Estonian
KeyToNature contact person ([email protected]).

3 Users’ feedback
The Estonian eFlora online was first presented to a wide audience of students,
teachers and citizens in Tartu (September 2009), on the occasion of a yearly
meeting of KeyToNature. The presentation was broadly and positively reflected
in the national media [5], [6], [7], contributing considerably to the progress of
public interest. In the first week only, more than 1700 users (or just watchers)
visited our site.
The first part of the eFlora, limited to woody plants, was available online much
earlier, with applications for iPhone, iPodTouch and iPad from iTunes [8], which
fostered a great interest of media.
The Estonian eFlora was mainly created for teachers and their students.
Teachers - from school teachers to university professors - feedbacked us through
questionnaires on their experience with the eFlora. Altogether, 19 using events
with about 350 participants have been officially recorded in Estonia on three
different educational levels (primary and secondary schools, and universities).
However, the actual number of users in school lessons is probably much higher,
as not all teachers have filled the questionnaire. According to the answers in
the questionnaire, the computer-based activities for identifying organisms are
very much appreciated by both teachers and students (based on teachers’
judgements). The huge amount of images connected to the keys was seen as
a primary positive aspect. Problems with scientific terms used in the keys but
not understood by pupils occurred especially in the younger classes. Several
teachers had solved the problem by preparing an introductory part to the lesson,
during which specific terms were explained. As the identification practices were
much acknowledged, it has been often proposed to include these activities in
the context of the schools’ official curriculum.

Acknowledgement

The Estonian eFlora and its OpenKeyEditor were prepared within the project
KeyToNature in cooperation between the University of Trieste (Italy), the University of
Tartu and the Estonian University of Life Sciences (Estonia). KeyToNature is financed
by the European Union through the program eContentplus. Special thanks are due to
Aino Kalda, Thea Kull and Ülle Reier who provided important input. Rein Kalamees,
Jaan Liira, Jaanus Paal, Kersti Püssa, Elle Roosaluste, Kai Rünk and Tiina Talve are
acknowledged for putting several plant pictures at our disposal.

References
[1] S. Martellos and P. L. Nimis, “KeyToNature: Teaching and Learning Biodiversity: Dryades,
the Italian Experience”. In: M. Muñoz, I. Jelinek and F. Ferreira (eds.), Proceedings of the
International Association for the Scientific Knowledge (IASK) International Conference
“Teaching and Learning”, pp. 863–868, 2008.
[2] T. Randlane, A. Saag, S. Martellos and P. L. Nimis, “Computer-aided, interactive keys to

192
 lichens in the EU project KeyToNature, and related resources”. In: T. H. Nash III (ed.), The
lives of the lichen symbionts, Bibliotheca Lichenologica, vol. 105, Stuttgart, J. Cramer in der
Gebrüder Borntraeger Verlagsbuchhandlung, (in press), 2010.
[3] T. Kukk, Eesti taimestik. Tartu–Tallinn: Teaduste Akadeemia Kirjastus, 464 pp., 1999.
[4] T. Kukk and T. Kull (eds.), Atlas of the Estonian flora. Institute of Agricultural and Environmental
Sciences of the Estonian University of Life Sciences, Tartu, 527 pp., 2005.
[5] T. Randlane, “An interview in a national radio broadcast”, Environmental Tent, available at
http://www.eseis.ut.ee/k2n_promo/keskkonnatelk20090222_1.mp3, 22.02.2009.
[6] U. Käärt, “Eesti e-taimetark internetis meelitab magnetina tuhandeid loodusesõpru”, Eesti
Päevaleht, Available at http://www.epl.ee/artikkel/480088,13.10.2009.
[7] T. Randlane, “Digitaalsed taime- ja samblikumäärajad ootavad koolides laiemat kasutamist”,
Koolielu, Available at: http://arhiiv.koolielu.ee/pages.php/0710,24405, 25.11.2009.
[8] A. Saag, T. Randlane and M. Leht, “Keys to plants and lichens on smartphones – Estonian
examples”. In: P. L. Nimis and R. Vignes Lebbe (eds.), Tools for Identifying Biodiversity:
Progress and Problems, pp. 195-199, 2010.

193
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 195-199.
ISBN 978-88-8303-295-0. EUT, 2010.

Keys to plants and lichens

on smartphones:
Estonian examples
Andres Saag, Tiina Randlane, Malle Leht

Abstract — The EU project KeyToNature aims to contribute to a better

knowledge of biodiversity by a practical activity, the identification of species.
The project is introducing new tools, including digital keys for mobile media.
In Estonia two applications have been prepared for smartphones: a ‘Key to
trees and shrubs of Estonia’ (for iPhone, iPodTouch and iPad) and a ‘Key to
Estonian epiphytic macrolichens’ (using the Android operating system). A third
tool, the ‘Key for plants of the island Naissaar’ is an example of application of
the Open Key Editor for creating new keys for mobile devices, starting from a
larger master key.

Index Terms — Estonia, e-tools, identification, keys, smartphones.

—————————— u ——————————

1 Introduction

T
he UN has declared 2010 to be the International Year of Biodiversity. The
world is invited to take action to safeguard the variety of life on Earth.
Protecting or managing in a sustainable way the biota in any part of the
world is possible only when their species are not only recorded and recognized
by specialists, but also noticed and appreciated by the widest of audiences – by
everyone.
The three-year EU project KeyToNature (http://www.keytonature.eu/), which
started in 2007 and is finishing in autumn 2010, aims to contribute to a better
knowledge of biodiversity by a practical activity, the identification of species.
The consortium introduces new tools for this purpose, keys that allow the
easy identification of plants, lichens, birds and other organisms using digital,
user-friendly facilities [1]. Hundreds of digital keys have been produced
within KeyToNature, many of them in close cooperation with the University
of Trieste, Italy, using FRIDA (FRiendly IDentificAtion), a software developed
in this University. The keys are published online and are freely accessible to

————————————————
A. Saag and T. Randlane are with the Institute of Ecology and Earth Sciences, University of Tartu,
Lai 38/40, Tartu 51005, Estonia. E-mail: [email protected], [email protected].
M. Leht is with the Institute of Agricultural and Environmental Sciences, Estonian University of Life
Sciences, Kreutzwaldi 5, Tartu 51014, Estonia. E-mail: [email protected].

195
everyone. They can also be stored on CD-ROMs, to be used without an Internet
connection. Those who still prefer paper-printed keys can print out their own
“keybook”. The latest trend is to develop applications which permit the use of
digital keys on mobile media, such as palmtop computers and smartphones,
either online or in stand-alone form.

2 Digital keys on mobile media

2.1 General

A smartphone is a mobile phone that offers more advanced computing ability

and connectivity than a contemporary basic mobile phone. Smartphones allow
the user to install and run various applications based on a specific platform, as
they run complete operating system software providing a platform for application
developers. These advanced mobile devices possess powerful processors,
abundant memory, larger multi-touch screen and a virtual keyboard with e-mail,
web browsing and WiFi connectivity. Today smartphones form the fastest
growing segment of the mobile phone market [2].
Identification keys in the mobile phones show several advantages compared
to the usual digital keys accessible through Internet or on CD-ROMs. Firstly,
they enable the use of identification tools not merely at home, in a classroom
or in a lab, but also in the field. The possibility to identify the exact geographical
location of a plant by GPS (inserted in the mobile phone) and to send additional
information (photos, notes, descriptions) via email – for example, to a botany
expert for the evaluation of an identification – is a valuable supplementary
function of such applications.

2.2 A botanist in your pocket: the key to Estonian trees and shrubs

The Estonian eFlora is an interactive digital identification key for c. 1100

plant species (out of ca 1500 taxa recorded from Estonia), including several
introduced tree species. Some taxa from critical genera are excluded from this
key, as well as several very rare species. The key was presented to a wide
audience of students, teachers etc. in Tartu (Estonia), in September 2009, and
is permanently freely available in Estonian and in English at http://dbiodbs.univ.
trieste.it/carso/chiavi_pub21?sc=368.
Presently, two adaptions for smartphones are available: a ‘Key for trees and
shrubs of Estonia’ (for iPhone, iPodTouch and iPad), and a ‘Key for plants of the
island Naissaar’ (using the Android operating system).
The application ‘Key for trees and shrubs of Estonia’ (Fig. 1) allows users to:
1. Identify more than 140 tree and shrub species in Estonia.
2. See the explanatory files about species and access the photo gallery.
3. Search by taxon name.
4. Post the identified species to Facebook.
5. Take pictures and add field notes to one’s guide.
6. Send multimedia content (pictures and notes) stored in the device via email.

196
 The application was developed by Divulgando Srl (Italy) and released publicly
at the end of 2009. It is available for download from the iTunes App Store for
a symbolic price of 2,39 EUR. During a short period (18.–24. January 2010), it
was even at the very top of the list of Paid Apps in the iTunes App Store and still
today it is top ranking in those for the educational sector.
The ‘Key for plants of the island Naissaar’ is an example of a filtered key derived
from the Estonian eFlora, generated by a new sofware, the OpenKeyEditor
of KeyToNature [3] specifically for a mobile device. This key, which includes
415 plant species, was ordered by a company carrying out nature-educational
training courses on the island Naissaar.
Both tools display a dichotomous interface where each step of the identification
process is richly illustrated with pictures and drawings [3]. As the applications can
be downloaded to the memory card of a smartphone, they can be used in stand-
alone form without additional web-browsing charges (with some limitations in
the access to the image archives, compared to the online keys).

2.3 A lichenologist in your pocket: the key to Estonian epiphytic

macrolichens

Another application for smartphones, using the Android operating system, has
been developed by the Company Mine Avasta (Estonia), based on an internet
key which was produced in cooperation between the University of Trieste (Italy)
and the University of Tartu. This application enables the identification of 115
species of epiphytic macrolichens known to occur in Estonia (Fig. 1). The main
principles are similar to those of the previously described application: it allows
identification of taxa using a simple dichotomous key in which the user has
to decide between two options; it is also possible to search the taxon by its
Estonian or Latin name, and then get additional information about the species
– read a summary of diagnostic characters, and see the photos and distribution
map of the species in Estonia. As the characters of lichens which are used in the
key are less familiar to the wide audience, an explanation of the main characters
of lichens in the form of an illustrated glossary is also provided.
The tool was uploaded into the Android Market in July 2010, and is free of
charge.

3 Conclusion
The two Estonian examples of digital identification keys for smartphones
were meant to attract the attention of a wide circle of non-specialists: pupils,
students, teachers, forestry workers, nature conservation staff, tourists etc. –
to increase public awareness of biodiversity and to allow new approaches in
nature education. Both tools are available not only as smartphone applications
but also on the Internet as interactive keys, freely accessible to everyone
(http://dbiodbs.univ.trieste.it/carso/chiavi_pub21?sc=175, trees and
shrubs; http://dbiodbs.univ.trieste.it/carso/chiavi_pub21?sc=159, epiphytic
macrolichens).
We have already received positive feedback from a large audience through

197
 Fig. 1 – Front pages of the mobile applications ‘Key for trees and shrubs of Estonia’
(left) and ‘Key for the identification of Estonian epiphytic macrolichens’ (right).

public media (newspaper articles, electronic publications, broadcasts) indicating

that the local society is willing to accept and use the new interactive e-learning
devices in everyday life [4], [5], [6].

Acknowledgement

The described facilities have been prepared within the project KeyToNature, financed
by the European Union through the program eContentplus as well as through the
European Regional Development Fund (Center of Excellence FIBIR). Special thanks
are due to Rodolfo Riccamboni (Divulgando Srl, Italy) and Marko Peterson (Mine Avasta,
Estonia) for developing the mobile applications.

References
[1] S. Martellos and P. L. Nimis, “KeyToNature: Teaching and Learning Biodiversity: Dryades,
the Italian Experience”. In: M. Muñoz, I. Jelinek and F. Ferreira (eds.), Proceedings of the
International Association for the Scientific Knowledge (IASK) International Conference
“Teaching and Learning”, pp. 863–868, 2008.
[2] “Smartphone definition from PC Magazine Encyclopedia”. PC Magazine, available at http://
www.pcmag.com/encyclopedia_term/0,2542,t=Smartphone&i=51537,00.asp, July 2010.
[3] T. Randlane, A. Saag, S. Martellos and P. L. Nimis, “Computer-aided, interactive keys to
lichens in the EU project KeyToNature, and related resources”. In: T.H. Nash III (ed.), Together
and separate: The lives of the lichen symbionts, Bibliotheca Lichenologica, vol. 105, Stuttgart,

198
 J. Cramer in der Gebrüder Borntraeger Verlagsbuchhandlung, (in press), 2010.
[4] M. Aeltermann, “Eesti e-Floora määraja võimaldab tuvastada taimeliike”, ERR Uudised,
avalable at http://uudised.err.ee/index.php?06191611, 18 January 2010.
[5] U. Käärt “Telefon määrab teadlaste abiga puu- ja põõsaliike”, Eesti Päevaleht, available at
http://www.epl.ee/artikkel/486512, 19 January 2010.
[6] M. Himma, “Taimed näitavad end nutitelefonis”, Tartu Postimees, available at http://www.
tartupostimees.ee/?id=214092, 20 January 2010.

199
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 201-205.
ISBN 978-88-8303-295-0. EUT, 2010.

IIKC: An Interactive Identification

Key for female Culicoides
(Diptera: Ceratopogonidae) from
the West Palearctic region
Bruno Mathieu, Catherine Cêtre-Sossah, Claire Garros,
David Chavernac, Thomas Balenghien, Régine Vignes Lebbe,
Visotheary Ung, Ermanno Candolfi, Jean-Claude Delécolle

Abstract — In 2006, bluetongue virus (BTV) outbreaks appeared surprisingly

in northern Europe and widely affected most of the European countries.
Correct identification of Culicoides species (Diptera: Ceratopogonidae), known
as BTV vectors, is a key component of all studies intending to understand
vector dynamics and to develop vector control strategies. A computer-based
system, Xper2, was used to develop an Interactive Identification Key (IIKC)
for female Culicoides from the West Palearctic region. The current version of
IIKC includes 108 taxa, 61 descriptors and 837 pictures and schemes. IIKC
is a powerful tool for routinely identifying Culicoides species and for training
young specialized taxonomists.

Index Terms — Culicoides species, identification key, interactive key,

bluetongue.

—————————— u ——————————

1 Introduction

B
luetongue is an arboviral disease affecting ruminants, mainly ovines.
Vectors of bluetongue virus (BTV) are small biting midges belonging to the
genus Culicoides (Diptera: Ceratopogonidae). Excluding the 46 fossils, a
total of 1308 Culicoides species are distributed on every large land mass with
the exception of Antarctica and New Zealand, ranging from the tropics to the
tundra and from sea level to 4000 m [1]. Worldwide, around 60 biting midges
species are suspected or proved to transmit viruses, protozoa or filaria worms

————————————————
B. Mathieu, E. Candolfi and J.-C. Delécolle are with the Institut de Parasitologie et de Pathologie Tro-
picale, Université de Strasbourg, EA 4438, 67000 Strasbourg, France. E-mail: [email protected].
C. Cêtre-Sossah, C. Garros, D. Chavernac and T. Balenghien are with Cirad, UMR Contrôle des
maladies animales exotiques et émergentes, F-34398 Montpellier, France.
R. Vignes Lebbe and V. Ung are with UMR 7207 CNRS-MNHN-Université Pierre et Marie Curie,
75005 Paris, France.

201
[2]. Several of these viruses are of major international significance for animal
health (African horse sickness virus, bluetongue virus, epizootic hemorrhagic
disease virus).
Up to 2010, five autochthonous biting midges species are suspected to transmit
BTV in western and central Europe: Culicoides chiopterus (Meigen, 1830) [3],
C. dewulfi Goetghebuer, 1935 [4], C. obsoletus (Meigen, 1919) [5], C. scoticus
Downes and Kettle, 1952 [6] and C. pulicaris (Linnaeus 1758) [7]. In addition,
along the Mediterranean basin, the main afro-tropical vector Culicoides imicola
Kieffer, 1913 is present and is in charge of BTV transmission. The emergence
and spread of bluetongue disease in Western Europe has highlighted the
taxonomic impediment concerning biting midges. Identification of Culicoides
species is highly important for a clear understanding of virus transmission
and biting midges population dynamics and for surveillance activities as well.
Because of their small size and of their highly specific diversity, morphological
identification of biting midges requires time and expertise.
Despite of the fact that monitoring activities are compulsory, most of the
European countries affected by bluetongue virus have very few taxonomists for
biting midges. They rely on two main identification keys: Campbell and Pelham-
Clinton’s key, published in 1960 [8], and Delécolle’s key, published in 1985 [9].
These two dichotomous keys are remarkable, but they have not been updated
for years and generally, they are not well adapted for non-expert researchers.
Therefore, when using the keys, newly described species or synonymies can be
missed, and scientists face difficulties with diagnostic and described characters.
Since a decade, the development of computer-aided systems marks a turning
point in taxonomy [10; 11]. Interactive identification keys based on multi-entries
are easy to use for experts and non-experts. They allow quick updates and are
easily released to the scientific community through the web. Today, interactive
identification keys have been developed for several arthropods: phlebotomine
sandflies [12], Glossina flies [13], or mosquitoes [14].
The aim of this work is to present the newly developed interactive identification
key for female Culicoides for the West Palearctic region (IIKC). Information on
availability and some recommendations are given.

2 Material and methods

2.1 Morphological characters

60 morphological descriptors coded in 164 morphological states are observed

from wings, abdomen, head and leg.
The wings descriptors are related to the presence or absence of spots,
position, size and shape.
Abdominal descriptors concern the number, size and shapes of spermathecae,
the presence and shape of a sclerotized ring, and some special features like
abdominal sclerites.
The head descriptors gather those of 4 body parts: antennae, eyes, palps and
mouthparts. Antennae are observed for the sensilli distribution (coeloconica,

202
short and long trichodea) and the antennal ratio (length of the first elongated
segment divided by the last short one). On eyes, inter-ocular space and the
interfacetal hairs are observed. Data related to the shape of the palpal segments
and the sensory pits are collected. Teeth are observed on mandibles and
maxilles.
The morphological characters were discussed and validated by 27
entomologists from 14 countries at the taxonomy meeting of the MedReoNet
network at Strasbourg in 2009 (http://medreonet.cirad.fr/).

2.2 Software and collections

The morphological database of female Culicoides was edited with Xper2

version 2.0 [15] which allows the creation of the interactive key.
Slide-mounted specimens observed are from the Callot, Kremer and Delécolle
collections (IPPTS, Strasbourg, France). Pictures were taken with a Zeiss®
microscope equipped with a Motic® camera. All pictures were individually
cleaned up using the software Gimp version 2.6.2.

2.3 Validation

A trial has been proposed to 5 non expert scientists on Culicoides taxonomy. 37

slide-mounted female specimens belonging to the genus Culicoides are anonymously
coded. These selected species span a wide range of morphological diversity of
Culicoides. The identification order of the slides has been randomly determined for
each participant. Correct or false status of identification, estimates of user’s confidence
and the use of definitions and illustrations were also gathered.

3 Results
In total, 60 morphological descriptors have been observed: 27 on the wing, 14
on the abdomen, 16 on the head and 3 on the legs. An additional geographical
descriptor has been added, which allows users to limit the taxa list to one
country. The 60 morphological descriptors are divided into 164 morphological
states illustrated with 403 pictures and schemes. Morphological data of 22
species were collected from stocked species at IPPTS, Strasbourg, France
and, 86 others species were from Callot, Kremer and Delécolle collections. In
total, the current version of IIKC includes 108 taxa. Among them, 8 species with
important morphological variations have been coded as taxa with polymorphic
characters. A total of 76 taxa were illustrated with drawings sheets. 73 species
were illustrated with 434 pictures (mean of 5.9 pictures per taxon): 24 with only
pictures and 49 with both pictures and drawings. Only 8 taxa have not yet been
illustrated. IIKC includes a total of 837 pictures and schemes.
IIKC is still in a validation step at the submission date of this communication
thus, results could not be shown.

203
4 Discussion

4.1 Availability and updates

IIKC will be freely available on a CD-Rom upon request to the authors.

Moreover, a dedicated website under construction will allow the interactive
key as well as updates to be downloaded. To help users with limited computer
capacity, IIKC will also be available on-line without local installation.
A scientific committee has been proposed, and annual meetings will be organized
to validate updates, discuss new species or synonymies, evaluate new systematic
or taxonomic changes. IIKC users are encouraged to contact the authors and the
scientific committee for feedbacks and to inform for new taxonomic information.

4.2 Recommendations

IIKC helps in identifying adult female Culicoides species. Identification of

Culicoides genus among other genus of the Ceratopogonidae family is not
included.
IIKC helps in identifying slide-mounted specimens and users are recommended
to use a microscope for good morphological observations. Stereomicroscope
observations of biting midges preserved in alcohol limit the observation on wing
patterns only.
IIKC is a multi-entry key. Compared to a dichotomous key, this key allows to
choose the descriptors the user wants to observe. If the specimen has damaged
parts, identification can go on with others ones. Users can also select a group
of descriptors (wing, abdomen, leg, head, or geography). Three optimized list
of characters classifying descriptors according to their discriminating power,
are available as an option and leads to quick identification. Users are strongly
recommended to use the option “Xper original sort”. When activated, a number
into brackets appears for each descriptor (from 0 to 1) representing the
discriminating power. The highest numbers are the most powerful descriptors
(i.e. the ones that will best discriminate the taxa).
Users are strongly recommended to use reference collections and national
experts to confirm their identification when dealing with new recorded or
observed species. Authors strongly encouraged future users to build regional
reference collections and to help in exchanging material between collections
to improve our systematic and taxonomic knowledge of the genus Culicoides.

5 Conclusion
IIKC is a newly developed morphological identification key allowing the
identification of 108 taxa of Culicoides (Diptera: Ceratopogonidae). Largely
illustrated with 837 pictures, drawings and schemes, this interactive identification
key is based on a multi-entry system, with optimized list of characters (including
geographical distribution). The richness of illustrations is a great advantage to
train taxonomists. The development of identification tools for Culicoides and more

204
generally for arthropods involved in pathogen transmission will help scientists in
identifying species and therefore will give better insights into the bioecology and
dynamics of these groups, helping in designing more appropriate vector control
strategies.

Acknowledgements

IIKC was developed in the framework of a surveillance network of reoviruses,

bluetongue and African horse sickness in Mediterranean basin and Europe (acronym
MedReoNet, contract no. SSPE-CT-2006-044285. The authors would like to acknowledge
all the partners of the WP 2, partners of MedReoNet, and especially R. Meiswinkel,
F. Schaffner, M. Miranda. Special thanks to X. Allène, S. Carpenter, D. Delécolle, L.
Gardès, K. Huber, I. Rakotoarivony, M.L. Setier-Rio, R. Vénail for testing the beta version
and the validation step.

References
[1] A. Borkent, World species of biting midges (Diptera: Ceratopogonidae), Belmont University,
The Ceratopogonid web page, pp. 236, 2009.
[2] A. Borkent, “Chapter 10. The biting midges - The Ceratopogonidae (Diptera)”, in: W.C.
Marquardt, (ed.), Biology and disease vectors, Elsevier Academic Press, pp. 113-126, 2005.
[3] E. Dijkstra, I. J. van der Ven, R. Meiswinkel, D. R. Holzel and P. A. Van Rijn, “Culicoides
chiopterus as a potential vector of bluetongue virus in Europe”, Vet. Rec., vol. 162, p. 422,
2008.
[4] A. Stephan, P. H. Clausen, B. Bauer and S. Steuber, “PCR identification of Culicoides dewulfi
midges (Diptera: Ceratopogonidae), potential vectors of bluetongue in Germany”, Parasitol.
Res., vol. 105, pp. 367-371, 2009.
[5] S. Carpenter, H. L. Lunt, D. Arav, G. J. Venter and P. S. Mellor, “Oral susceptibility to bluetongue
virus of Culicoides (Diptera: Ceratopogonidae) from the United Kingdom”, J. Med. Entomol.,
vol. 43, pp. 73-78, 2006.
[6] S. Carpenter, C. McArthur, R. Selby, R. Ward, D. V. Nolan, A. J. Luntz, J. F. Dallas, F. Tripet and
P. S. Mellor, “Experimental infection studies of UK Culicoides species midges with bluetongue
virus serotypes 8 and 9”, Vet. Rec., vol. 163, pp. 589-592, 2008.
[7] S. Caracappa, A. Torina, A. Guercio, F. Vitale, A. Calabro, G. Purpari, V. Ferrantelli, M. Vitale
and P. S. Mellor, “Identification of a novel bluetongue virus vector species of Culicoides in
Sicily”, Vet. Rec., vol. 153, pp. 71-74, 2003.
[8] J. A. Campbell and E. C. Pelham-Clinton, “A taxonomic review of the british species of
Culicoides Latreille (Diptera: Ceratopogonidae)”, Proc. R. Soc. Edinburgh, vol. 68, pp. 181-
302, 1960.
[9] J. C. Delécolle, Nouvelle contribution à l’étude systématique et iconographique des espèces
du genre Culicoides (Diptera: Ceratopogonidae) du Nord-Est de la France, PhD dissertation,
U.F.R. sciences de la vie et de la terre, Université Louis Pasteur de Strasbourg I, pp. 229,
1985.
[10] D. Agosti, “Biodiversity data are out of local taxonomists’ reach”, Nature, vol. 439, p. 392,
2006.
[11] D. E. Walter and S. Winterton, “Keys and the crisis in taxonomy: extinction or reinvention?”,
Annu. Rev. Entomol., vol. 52, pp. 193-208, 2007.
[12] R. Vignes Lebbe and C. Gallut, Computer Aided Identification of Phlebotomine sandflies of
Americas (CIPA), Université Pierre et Marie Curie, Paris, France. http://lis-upmc.snv.jussieu.fr/
xper2/infosXper2Bases/en/, 1997.
[13] J. Brunhes, D. Cuisance, B. Geoffroy and J. Hervy, Les glossines ou mouches tsé-tsé
(réédition), IRD Editions, Montpellier, France (CD-Rom), 2009.
[14] F. Schaffner, G. Angel, B. Geoffroy, J. Hervy, A. Rhaiem and J. Brunhes, The mosquitoes
of Europe. An identification and training software, IRD Editions and EID Méditerranée,
Montpellier, France (CD-Rom), 2001.
[15] V. Ung, G. Dubus, R. Zaragueta-Bagils and R. Vignes Lebbe, Xper2: introducing e-taxonomy,
Bioinformatics, vol. 26, pp. 703-704, 2010.

205
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 207-211.
ISBN 978-88-8303-295-0. EUT, 2010.

Indochinese bamboos:
biodiversity informatics to assist
the identification
of “vernacular taxa”
My Hanh Diep Thi, Régine Vignes Lebbe, Ha Phuong Nguyen,
Bich Loan Nguyen Thi

Abstract — Bamboo (Bambusoides – Gramineae) is one of the most

important natural resource in Southeast Asia. However, bamboo identification
has many difficulties. In the area of SEP programme «Indochina Bamboos»
(2008-2010), the final objective is to update the bamboo flora of Indochine
(Vietnam, Laos, Cambodege) and to publish an e-flora including free access
keys, digital images and information about the traditional and economic use
of bamboos. During field trips in Vietnam, Laos and Cambodgia, samples,
pictures and morphological description sheets were collected in various
locations and previously assigned to the local vernacular names. We use
an informatic program, Xper2, to assist the comparison and identification of
“vernacular bamboo taxa” based on morphological characteristics.

Index Terms — bamboo, Indochinese, taxonomy, common name, computing,

XPER2..

—————————— u ——————————

1 Introduction

B
amboo is extensively used in traditional handicrafts in Southeast Asia.
Nowadays, bamboo is also used in others fields: construction, medicine,
etc. With a better understanding of this group of plants, we could propose
recommendations on conservation measures and find species to be developed
for industrial exploitation and economic benefits [4].
Bamboo is in the family Gramineae (Poaceae), subfamily Bambusoideae, tribe
Bambuseae. Since Linné time, its taxonomy is based on flower characteristics.
However, bamboo is characterized by infrequent flowering. The taxonomy
————————————————
M. H. Diep, is with the University of of Sciences of HCMC, Centre de Recherche pour la Conser-
vation des ressources naturelles (CRC). Vietnam and Director of the Phu An Plant Conservation
Centre, Vietnam.
R. Vignes Lebbe is with UMR 7207 CNRS/MNHN/UPMC, MNHN Département Histoire de la
Terre, CP48, 57 rue Cuvier, 75005 Paris, France.
H. P. Nguyen is a master student in the Université Pierre et Marie Curie, Paris –VI, France
B. L. Nguyen Thi is master student in the University of of Sciences of HCMC, Vietnam.

207
of Indochinese Bamboo has not been completed; it is basically based on E.
G. Camus and A. Camus, «Flore Générale d’Indochine, Vol 7 Gramineae»
[9] describing 14 genera and 73 species [6]. In the 1970s, Professor Pham
Hoang Ho mentioned more than 120 bamboo species in «Vietnamese Plants»
[8]; almost 200 species with illustrative pictures are recorded in Nguyen Hoang
Nghia, «Vietnamese Bamboos» [7].
Facing the ongoing disappearance of many traditional uses of bamboo, and
its shrinking natural environment, Dr. Diep Thi My Hanh has decided to establish
the Bamboo Ecology Museum and the Plant Conservation Centre in Phu An, to
collect a variety of bamboo species and other endangered precious plants in the
Southeast. The project is jointly undertaken by Rhône-Alpes (France), the Binh
Duong Province (Vietnam), the Pilat Natural Garden, and the Natural Science
University of HCMC. In 2003-2007, the project has gathered a large amount
of information on Vietnamese bamboos in the North, Central Part, Highlands,
Mekong Delta, and the Southeast, with 301 dry specimens in a botany collection
and 157 samples of bamboos growing in the Conservation Centre [2].
Since 2007, a project to achieve the revision of Indochine Bamboos is in
progress, in collaboration with Laotian, Cambodgian and Vietnamian biologists.
During many field trips, morphological description sheets with pictures and
information on the bamboo applications in various locations have been made.
With a few exceptions, most samples have their common names in each
location. A crucial task is then to assign a scientific name to all gathered data.
This paper describes the methodology and the results of the project.

2. Methodology

2.1 Data collection

Data were gathered from literature, collections and field trips. Field trips
were conducted for the most part in Vietnam (all the regions) and in Laos and
Cambodia as well. The exploration needs to be completed in some locations.
The literature was consulted and analysed to collect all characters proposed by
botanists to define and identify bamboo species. This task was completed by
the observation of specimens (including type material) in the main reference
collections, such as the Royal Botanic Gardens Kew (UK), the Laboratoire de
Phanérogamie de Paris (France), and other botany collections in Asia.
2.2 Proposal of a standardised description form to describe specimens and
Bamboo species
A list of 90 morphological characters, divided into 11 groups, has been
established and documented by texts and images (Fig. 1). The botanic
terminology was controlled by the botanist Soejatmi Dransfield. The database is
now translated into five languages (Vietnamese, English, French, Laotian, and
Cambodian).

208
  
Fig. 1 – The 11 groups of characters describing Bamboos.

2.2 Digitalization of the Bamboo descriptions

Xper2 appears well adapted to manage our structured descriptions, texts and
images. Following the standardized list of characters, we edit the descriptions of
the species and also the descriptions of specimens with their common names.

2.3 Assigning scientific names to vernacular names

The scientific identification of each specimen is time-consuming and requires

highly skilled and adequately trained scientific personnel. To facilitate this task,
we use the facilities offered by Xper2 to compare descriptions (see Fig. 2).
We also use the free access key of Xper2 to associate the specimens of the
references collections to “vernacular taxa”. Similarity measurements between
descriptions are used to group vernacular and scientific descriptions. All these
results are compared, to propose one or few scientific names for each vernacular
name. This work is already in progress.

2.4 DNA analyses

To complete and to verify some identifications, DNA analyses are conducted

iin collaboration with the MNHN and the laboratory of Créteil University. This
early approach enables us to conduct more intensive studies on the Indochinese
Bamboo evolution based on analyzing the molecular evolution.

209
   Fig. 2 – The automatic comparison of descriptions displays in a visual table the
characters which are common or different between two or more descriptions. Here the
comparison of a “vernacular” entity and the species Dendrocalamus giganteus.

2.5 Computer-aided identification for Indochine bamboos

All the information collected in the project is already digitalized in a structured

format. The automatic HTML export of Xper2 and Xper2 online free access keys
will be combined to offer a e-bamboo-flora.

3. Conclusion
The project “Indochine Bamboos” is still progressing. Three new species have
been detected and will be published.
Presently, the Centre’s collection has about 350 specimens from Vietnam,
Laos and Cambodia. Few additional field trips are planned to complete the
live collection in the Plant Conservation Centre in Phu An with typical bamboo
species in Indochina.
The validation of the approach to identify vernacular names to scientific names
could be proposed for other taxa, and made more automated.

210
 Two master students and a PHD student are working on the subject. The project
also offers the opportunity of organising training courses on the identification
tools for students and young researchers coming from the participant institutions.
Two workshop trainings for using the software Xper2 were organized in 2008
and 2010. The participants were from Vietnam, Laos and Cambodia. This type
of tools is attracting students interested in botany, enhancing their capabilities to
analyse characters and taxonomic data.

Acknowledgement

The authors wish to thank Mrs Dransfield, Florian Causse and all participants of the
“Bambous d’Indochine” project. This work is supported by the French initiatives Sud-
Expert Plantes granted by the French. Ministry of Foreign Affairs.

References
[1] M. H. Diep, et al., Collection des variétés de bambou du Viet Nam. Rapport scientifique après
3 ans de prospection des Bambousa du Viet Nam, 155 pp., 2005.
[2] M. H. Diep and M. L. Nguyen thi, Ethnobotanique du bambou du Viet Nam. Rapport scientifique
de la Conférence scientifique de l’Université des Sciences Naturelles, novembre 2006.
[3] M. H. Diep, Biodiversité du Bambou du Viet Nam. Rapport scientifique de la Conférence
scientifique de l’Université des Sciences Naturelles, novembre 2006.
[4] S. Dransfield, and E. A. Widjaja, Plant Resources of South-East Asia. No 7 – Bamboos.
Backhuys Publishers, Leiden. 189 pp., 1995.
[5] J. Lebbe and R. Vignes, “Modelling taxonomic description for identification”. In: P. Bridges,
P. Jeffries, D. R. Morse and P. R. Scott (eds.), Information Technology, Plant Pathology and
Biodiversity, pp. 37-46, 1998.
[6] H. Le Comte, Flore Générale de l’Indochine. Editeur Masson et Cie, 630 pp., 1912-1923.
[7] H. N. Nguyen, Bamboos of Viet Nam. Agriculture Editions, 199 pp., 2005.
[8] P. Hoang Ho, Flore du Viet Nam. Montréal Edition. 735 pp., 1992.
[9] A. Camus, E. G. Camus and H. Lecomte, Flore générale de l’Indochine, Masson et Cie, Paris,
pp. 581-650, 1912-1923.

211
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 213-216.
ISBN 978-88-8303-295-0. EUT, 2010.

Identification tools as part of

Feedsafety research:
the case of ragwort
Leo W.D. van Raamsdonk, Patrick Mulder, Michel Uiterwijk

Abstract — Ragwort (Senecio jacobaea) and related species of the genus

Senecio are the main source of pyrrolizidine alkaloids. These plants grow in
road verges, meadows and production fields and they show up in parties of
roughage: grass and alfalfa. Monitoring can be carried out during the field
production and harvesting stages. The final objective is to reject parties with
a too high contamination level. Identification tools can support the decision to
accept or refuse materials for the food production chain. A ragwort model has
been developed for the mobile application Determinator. This identification
model includes the relevant objects (species of the genus Senecio), and a
range of so called confusing objects in order to minimise the chance of false
positive identifications.

Index Terms — Determinator, diagnosis, identification, ragwort.

—————————— u ——————————

1 Introduction

S
afe feed is one of the cornerstones of a healthy food production chain,
and as important side-effect it supports the desired welfare of husbandry
animals. In a lot of cases in the history of feed and food production emerging
risks were initially detected by visual surveillance. Also in a majority of those
cases visual inspection was replaced by more dedicated chemical detection
methods.
Nevertheless, new risks still emerge and visual inspection is still at hand at
the same moment that surveillance is needed. Recent examples are Ambrosia
seeds in bird feeds, packaging materials in overdue materials, precatory bean
(Abrus precatorius, in 2009 included in legislation) and ragwort in roughage
and in salads for human consumption. The well known problem of animal
byproducts is still primarily based on visual control. Identification tools are an
essential support for these diagnosing problems.

————————————————
L. W. D. van Raamsdonk and P. Mulder are with RIKILT – Institute of food safety, P.O. Box 230, 6700
AE Wageningen, the Netherlands. E-mail: [email protected], [email protected].
M. Uiterwijk is with Alterra, 6700 AE Wageningen, the Netherlands. E-mail: [email protected].

213
2 The Ragwort problem

2.1 Background

Ragwort (Senecio jacobaea) is one of the sources of pyrrolizidine alkaloids

and it can occur in parties of roughage: grass and alfalfa. Pyrrolizidine alkaloids
are toxic for animals and a long term lethal effect is observed, especially in
horses. Many other species of Senecio (e.g. common groundsel, S. vulgaris)
and species of the family Boraginaceae (Symphytum, Echium) will produce
pyrrolizidine alkaloids as well [1], [2].
Monitoring of the ragwort / pyrrolizidine alkaloids problem can be effected at
several stages in the feed production chain. Visual inspection can be carried out
during the field production and harvesting stages, when fresh materials are still
present. Chemical LC-MS/MS analysis can be applied to trade parties of dried
and processed roughage. In any situation visual screening can be followed by
chemical confirmation.

 
Fig. 1 – Inflorescence of ragwort, Senecio jacobaea.

2.2 Strategy

Monitoring of production fields or road sides where Senecio species might

occur is effective during pre-harvesting and harvesting times. On the spot
identification and qualitative risk assessment can be achieved with a mobile
knowledge system. Based on the prevalence of ragwort or other species it can
be decided to use a harvested party for feed production or to ignore and destroy
contaminated parties. In this way, identification tools can be useful for early

214
warning systems, so that costs in subsequent parts of the production chain can
be avoided.

2.3 Knowledge system Determinator

Support of monitoring in those pre-harvesting stages is provided by the

knowledge system Determinator. A datamodel is developed for this knowledge
system including five Senecio species, and a series of 22 different yellow-
flowering species, which can be confused with ragwort. Determinator can be
used in laboratory (Windows XP, Vista) as well as in field situations (Windows
mobile).
Determinator is a program package that assists the user in “determining”
or identifying an object. A final conclusion is reached by entering answers to
questions associated to the objects included in the dataset used. A match is
calculated between the object as described by the user and each of the targets
included in the chosen dataset. The process of identifying an object is supported
by the possibilities to browse the included targets, and to compare two targets
in every combination.
Every target in a fully developed datamodel is available with a description, with
one or more images, and with one or more states for every feature (character).
The descriptions and images are used to document the targets after the option
Browse. The lists of feature states are being used to Compare two targets, and
to Identify an object chosen by the user.
Determinator and the ragwort datamodel are freely available: www.
determinator.wur.nl/UK/. Some screenshots are shown in Fig. 2.

 
Fig. 2 – Screenshots of Determinator on a Windows mobile-based smartphone.

3 Discussion
The ragwort datamodel developed for Determinator is a highly dedicated
identification tool. It is an example of an open classification model: only diversity
is included that can directly support the final decision. Open classification models
can function only in a situation where closed classification systems exist that

215
included all the existing diversity [3], [4], [5]. The flora of the Netherlands and
the flora of the British Isles in Linneaus II [6] are examples of such classification
systems that support the selected diversity in the ragwort datamodel. Another
example of an open classification model is the decision support system ARIES
[7] designed to support the ban on animal by-products as feeding stuff.
The philosophy of the ragwort datamodel and of ARIES is to include two types
of objects. The first type of objects includes the species of Senecio or all types
of animal by-products, respectively. The second type of objects added to the
datamodel consists of a range of confusing objects. These confusing objects
are meant to minimise false positive identifications.
Open classification models provide a good support for certain types of risk
assessments, where information on identification is necessary. They can be
developed in a relatively short time, exclusively targeted information should be
included, and a connection exists with closed classification systems providing a
full view on the relevant diversity.

References
[1] D. Frohne and H. J. Pfänder, Poisonous Plants, second edition, London, Manson Publishing,
2005.
[2] P. B. Pelser, H. de Vos, C. Theuring, K. Vrieling, T. Hartmann and T. l. Beuerle, “Frequent gain
and loss of pyrrolizidine alkaloids in the evolution of Senecio section Jacobaea (Asteraceae)”
Phytochemistry, vol. 66, pp. 1285–1295, 2005.
[3] L. W. D. van Raamsdonk, “The effect of domestication on plant evolution”. Acta. Bot. Neerl.,
vol. 44, pp. 421-438, December 1995.
[4] L. W. D. van Raamsdonk and T. de Vries, “Cultivar classification in Tulipa”. Acta Bot. Neerl.,
vol. 45(2), pp. 183-198, June 1996.
[5] W. L. A. Hetterscheid, R. G. van den Berg and W. A. Brandenburg, “An annotated history of
the principles of cultivated plant classification”. Acta Bot. Neerl., vol. 45(2), pp. 123-134, June
1996.
[6] ETI bioinformatics. Linnaeus II software package. http://www.eti.uva.nl/, 2010.
[7] RIKILT Institute of food safety. ARIES, Animal Remains Identification and Evaluation System,
http://aries.eti.uva.nl/, 2010.

216
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – p. 217.
ISBN 978-88-8303-295-0. EUT, 2010.

Two identification tools applied

on Mascarene’s corals genera
(Xper2) and species (IKBS)
Yannick Geynet, Noël Conruyt, David Grosser,
Gérard Faure, David Caron

Abstract — For future biodiversity studies relying on species

identification, environmental officers and researchers will only be left
with monographic descriptions and collections in museums. This is
why a knowledge base on the zooxanthellate scleractinian corals of
the Mascarene Archipelago is being developed. This project offers
results for both biologists/taxonomists and students or MPA-teams.
Two online computer-based applications permit to identify genera
and species. The first identification tool, called Xper², was developed
by LIS (Informatic and Systematics Laboratory) in Paris, and is used
for identifications to genera. The second tool, named IKBS (Iterative
Knowledge Base System), was developed by IREMIA (Institute
for Research in Applied Mathematics and Computer Science) in
La Réunion, and is used for identifications from families to species. The
tools presently work for Astrocoeniidae, Pocilloporidae, Acroporidae
(only Acropora + Isopora), Psammocoridae, Siderastreidae (owns
Psammocoridae as genera), Fungiidae, Poritidae, Faviidae Faviinae,
Faviidae Montastreinae, Mussidae. We plan to start a new phase to
add the last families, fully translate the web site in English and extend
the Xper² identification to all the western Indian Ocean genera.

Index Terms — identification tool, IKBS, Mascarene archipelago, scleractinian

corals, Xper2.
—————————— u ——————————

Acknowledgement

This work was supported in most part by the EU (FEDER), the French Ministry of
National Education, Advanced Instruction and Research, and the Réunion regional
council. Web site: http://coraux.univ-reunion.fr/

————————————————
Y. Geynet., N. Conruyt, D. Grosser and D. Caron are with the IREMIA lab from the Réunion Uni-
versity - PTU, 97490 Ste Clotilde – La Réunion. E-mail: [email protected].
G. Faure is retired from the Univ. of Sciences Montpellier 2, 34000 Montpellier, E-mail: faure@
cegetel.net.

217
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – p. 219.
ISBN 978-88-8303-295-0. EUT, 2010.

Interactive, illustrated, plant

identification keys: an example
for the Portuguese flora
Maria Helena Abreu Silva, Rosa Maria Ferreira Pinho, Lísia Graciete,
Martins Pereira Lopes, Paulo Cardoso da Silveira

Abstract — At the University of Aveiro, a multimedia tool was

developed to help the teaching of botany [1], [2]. It includes a
Dichotomous Interactive Key (DIK), for 390 taxa of vascular plants
occurring at “Ria de Aveiro” lagoon system. This key is linked to two
different glossaries, one descriptive, the other illustrated. At the end
of the identification process, the student is conducted to a webpage
including photographs, descriptions and other relevant information
about the taxa. The students considered it “very useful”, and the
successful identifications increased in the assessment tests in the
two years after the introduction of the DIK. This seem to be an
effective tool in the teaching of botany, namely in plant identification
at secondary and post-secondary level.

Index Terms — interactive keys, illustrated keys, morphology, plant

identification.

—————————— u ——————————

References
[1] P. Silveira, H. Silva, R. Pinho and L. Lopes, Chaves Ilustradas. Identificação das plantas
vasculares do Baixo Vouga Lagunar, CD-ROM. Colecção Biorede, Universidade de Aveiro,
ISBN: 972-789-211-6, 2006.
[2] H. Silva, R. Pinho, L. Lopes and P. Silveira, “ Illustrated plant identification keys: an interactive
tool to learn botany” Computers & Education, submitted for publication.

————————————————
All authors are with the Department of Biology, University of Aveiro, 3810-193, Aveiro, Portugal
E-mail [email protected]., P. Silveira and H. Silva are also with the CESAM (Centre for Environment
and Marine Studies), University of Aveiro, 3810-193, Aveiro, Portugal.

219
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – p. 221.
ISBN 978-88-8303-295-0. EUT, 2010.

The ORCHIS software used to

identify 100 orchids species of
Lao PDR
Pierre Bonnet, André Schuiteman, Boukhaykhone Svengsuksa,
Daniel Barthélémy, Vichith Lamxay, Soulivanh Lanorsavanh,
Khamfa Chanthavongsa, Pierre Grard

Abstract — The identification of plants, especially of orchids, is a

major obstacle to the regulation of their trade. Lao PDR recently
signed the CITES (Convention on International Trade in Endangered
Species of Wild Fauna and Flora), and is facing many difficulties in
its application at its borders, precisely because of lack of resources
and information about its flora. This is particularly marked in
Laos, where the pressures on the country’s natural resources are
particularly important because of the growth of this region. The
National Herbarium Netherlands, the National University of Laos, and
CIRAD have joined within the European project ORCHIS (http://www.
orchisasia.org/) to develop, test and disseminate original tools to
identify 100 orchids species in Laos. This identification tool, primarily
targeted for managers of protected areas and customs authorities of
the Country, is adapted to people with only a limited knowledge of
botany. The tool, based on the IDAO software developed by CIRAD,
enables the creation of a graphic sketch of the plant that is sought,
thus allowing to overcome the constraints of language and knowledge
of a specialized terminology.

Index Terms — orchids, Laos, biodiversity informatics, taxonomic database,

identification tool.

————————————————
P. Bonnet and D. Barthélémy are with INRA, UMR AMAP, Montpellier, F-34000, France. E-mail:
[email protected], [email protected].
A. Schuiteman ([email protected]) is with Royal Botanic Gardens, Kew, Richmond, Surrey,
TW9 3AB, UK.
B. Svengsuksa ([email protected]), V. Lamxay ([email protected]), S. Lanorsavanh,
and K. Chanthavongsa are with National University of Lao PDR, Faculty of Science of Lao PDR,
Department of Biology, P.O. BOX 7322, Vientiane, Lao PDR.
P. Grard is with CIRAD, UMR AMAP, Montpellier, F-34000, France.

221
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – p. 223.
ISBN 978-88-8303-295-0. EUT, 2010.

A collaborative and distributed

identification tool for plants
Philippe Laroche

Abstract — We are developing a website/smartphone application

where users can deposit pictures of plants. Volunteers can then
answer questions relating to these pictures: text questions, or
drawing comparisons, or comparisons with other pictures. When
sufficient data are available (50 or 100 of such answers) we can
then propose an identification. The volunteers give answers based
on “naive” observations of plant details. Example on a flower: do
you see 3 petals, 4 petals, 5-8 petals, many petals ? For leaves,
we display a small set of drawings, the set changes when the firsts
questions provide sufficient information. To finish the identification,
we allow for direct comparison with other pictures of similar detail.
The questions are inside Iframe, and so are easily included on a
network of participating web sites. Depending on the total number of
visits to these sites, we shall have identifications within a few minutes
or a few hours. We have now a set of 300 plants which are well-know
in France. An important point about the method is that the volunteers
help to classify the pictures we have as well as new pictures, so that
the system is bootstrapping and scales easily. Moreover, we can
easily experiment with new questions or drawings, to find those that
reduce entropy most rapidly.

Index Terms — collaborative tool, PDA application, biodiversity collection

building.

————————————————
The author is with Agoralogie, 6 rue de Candie, F 75011 Paris, France. E-mail: philippe.laroche@
agoralogie.fr.

223
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 225-229.
ISBN 978-88-8303-295-0. EUT, 2010.

Alternative 2D and 3D Form

Characterization Approaches to
the Automated Identification of
Biological Species
Norman MacLeod

Abstract — Few have sought to compare the performance of alternative types

of morphological data for biological species identification. This investigation
contrasts results of form characterization via form factors, superposed
landmark coordinates, landmark-registered semilandmark outlines, 3D
semilandmark networks, and raw digital images for a test set of seven Recent
planktonic foraminifer species. While all data types performed better than the
qualitative assessment of morphological variation by human taxonomists,
landmark-registered semilandmark outlines and raw digital images delivered
the best performance in the context of approaches that could reasonably
serve as the basis for fully automated species identification systems.

Index Terms — automated identification, landmark coordinates, Foraminifers,

taxonomy.

—————————— u ——————————

T
he automated identification of biological objects (individuals) and/or groups
(e.g., species, guilds, characters) has been a dream of systematists’ for
centuries. The goal of some of the first multivariate biometric methods
was to address the perennial problem of group identification and inter-group
characterization [1], [2]. Despite much preliminary work in the 1950s and 60s,
progress in designing and implementing practical systems for fully automated
specimen identification has proven frustratingly slow. However, as recently as
2004 Dan Janzen updated the dream for a new audience [3].

“The spaceship lands. He steps out. He points it around. It says

‘friendly–unfriendly-edible–poisonous–safe–dangerous–living–
inanimate’. On the next sweep it says ‘Quercus oleoides–Homo
sapiens–Spondias mombin–Solanum nigrum–Crotalus durissus–
Morpho peleides– serpentine’. This has been in my head since
reading science fiction in ninth grade half a century ago.” (p. 731)

————————————————
The author is with the Palaeontology Department, The Natural History Museum, Cromwell Road,
London SW7 5BD, [email protected].

225
 Janzen’s solution to this classic problem involved building machines to identify
species from their DNA. His predicted budget and proposed research team are
“US$1 million and five bright people.” (p. 731). However, recent developments
in computer architectures, as well as innovations in software design, have
placed the tools needed to realize Janzen’s vision in the hands of the scientific
community not in several years hence, but now; and not just for DNA barcodes,
but for digital images of organisms.
A recent survey of small-scale automated species identification system trials
(<50 taxa), shows an average reproducible accuracy of over 85 percent with
no significant correlation between accuracy and the number of included taxa or
the type of group being assessed (e.g., butterflies, moths, bees, pollen, spores,
foraminifera, dinoflagellates, vertebrates) [4]. These figures should be compared
with the disturbingly few blind test studies of accuracy and consistency of
human taxonomist identifications that have been published to date [5], [10].
Human cognition studies [5] suggest that human experts who are routinely
engaged in particular discriminations can return accuracies in the range of 84
to 95 percent. But in the (far more common) cases in which trained personnel
must deliver identifications for species they are not dealing with on a day-to-day
basis self-consistencies drop to 67-83 percent and consensus consistencies
between identifiers to 43 percent. Moreover, semi-automated and automated
identifications–often involving thousands of individual specimens–can be made
in a fraction of the time required by human experts and can be done on site, on
demand, anywhere in the world.
Is there a need for such systems? After all, biology has been getting by
without them for millennia. What makes anyone think computers can – much
less should – replace human taxonomists or that the taxonomic communities
efforts would not be better spent lobbying for increased government funding for
tried and true traditional α-taxonomy?
If evidence existed to reassure the scientific community that most taxonomic
identifications are accurate and consistent current identification practices
situation might be tolerable. There is little such evidence. For example, 1997
a group of geologists organized a blind test to try to resolve a controversy over
whether marine animals went extinct before or after the meteorite impact that
marked the end of the Cretaceous Period.9 Four taxonomic experts were asked
to identify species of microscopic foraminifera in a set of rock samples without
being told the age of the samples. No consensus on when the animals died out
was established – not because of any flaw in the test’s design, but because the
species lists produced were so different as to be incomparable, in some cases
with just 25 percent of species names in common.
Contrary to some voices within the systematics community, these developments
could not have come at a better time. As all scientists already know, the world
is running out of specialists who can identify the very biodiversity whose
preservation has become a global concern. In commenting on this problem in
palaeontology as long ago as 1993, Roger Kaesler recognized12 …

“… we are running out of systematic paleontologists who have

anything approaching synoptic knowledge of a major group of

226
 organisms” (p. 329). “Paleontologists of the next century are
unlikely to have the luxury of dealing at length with taxonomic
problems … [Paleontology] will have to sustain its level of
excitement without the aid of systematists, who have contributed
so much to its success.” (p. 330).

This expertise deficiency cuts as deeply into those commercial industries that
rely on accurate identifications (e.g., agriculture, biostratigraphy) as it does into
a wide range of pure and applied research programmes (e.g., conservation,
biological oceanography, climatology, ecology).
If truth be told, it is commonly, though informally, acknowledged that
the technical, taxonomic literature of all organismal groups is littered with
examples of inconsistent and incorrect identifications. This is due to a variety
of factors, including taxonomists being insufficiently trained and skilled in
making identifications (e.g., using different rules-of-thumb in recognizing
the boundaries between similar groups), insufficiently detailed original group
descriptions and/or illustrations, inadequate access to current monographs and
well-curated collections and, of course, taxonomists having different opinions
regarding group concepts. Peer review only weeds out the most obvious errors
of commission or omission in this area, and then only when an author provides
adequate representations (e.g., illustrations, recordings, gene sequences) of
the specimens in question.
Few have sought to compare the performance of alternative types of
morphological data for biological species identification. This investigation
contrasts results of form characterization via form factors, superposed landmark
coordinates, landmark-registered semilandmark outlines, 3D semilandmark
networks, and raw digital images for a test set of seven Recent planktonic
foraminifer species. While all data types performed better than the qualitative
assessment of morphological variation by human taxonomists, landmark-
registered semilandmark outlines and raw digital images delivered the best
performance in the context of approaches that could reasonably serve as the
basis for fully automated species identification systems.
Systematics too has much to gain, both practically and theoretically, from
the further development and use of automated identification systems. It is now
widely recognized that the days of systematics as a field populated by mildly
eccentric individuals pursuing knowledge in splendid isolation from funding
priorities and economic imperatives are rapidly drawing to a close. In order to
attract both personnel and resources, systematics must transform itself into a
“large, coordinated, international scientific enterprise” [13] (p. 4). Many have
identified use of the internet–especially via the world-wide web–as the medium
through which this transformation can be made. While establishment of a virtual,
GenBank-like system for accessing morphological data, audio clips, video
files and so forth would be a significant step in the right direction, improved
access to observational information and/or text-based descriptions alone will
not address either the taxonomic impediment or low identification consistency
issues successfully. Instead, the inevitable subjectivity associated with making
critical decisions on the basis of qualitative criteria must be reduced, or at the

227
very least, embedded within a more formally analytic context.

 
Fig. 1 – Example of the DAISY system interface displaying a planktonic foraminifer
specimens from the test dataset. For this group, chamber arrangement, primary
aperture position, and wall texture are among the primary taxonomic characteristics
used to identify species.

Properly designed, flexible, and robust, automated identification systems,

organized around distributed computing architectures and referenced to
authoritatively identified collections of training set data (e.g., images, gene
sequences) can, in principal, provide all systematists with access to the electronic
data archives and the analytic tools necessary to handle routine identifications
of common taxa. Properly designed systems can also recognize when their
algorithms cannot make a reliable identification and refer that image to a
specialist. Such systems will, inevitably, include elements of artificial intelligence
that will allow them to improve their performance the more they are used. Most
tantalizingly, once morphological (or molecular) models of a species have been
developed and demonstrated to be accurate, these models can be queried to
determine which aspects of the observed patterns of variation and variation
limits are being used to achieve the identification, thus opening the way for the
discovery of new and (potentially) more reliable taxonomic characters.
As has been demonstrated repeatedly through human history, scientific
progress lies, in part, in constructing machines that do what machines do best
and allowing humans to do what humans to best. Far from making taxonomists
obsolete, the creation of automated identification systems will free them from the
drudgery of delivering routine identifications to focus on the more conceptually

228
difficult issues of discovering, revising and describing species concepts,
understanding how species fit into higher taxonomic and ecological groups
and establishing how species function within natural systems. Getting high-
throughput machine-learning systems on the agenda of research communities
and scientific research funding councils, as well as into the study programmes
of all sorts of disciplines, is required if taxonomy is to regain the sense of mission
that will allow it to fulfil its potential as a twenty-first century science.

References
[1] R. R. Sokal and P. A. Sneath, Principles of numerical taxonomy. W. H. Freeman, San
Francisco, 1963.
[2] P. H. A. Sneath and R. R. Sokal, Numerical taxonomy: the principles and practice of numerical
classification. W. H. Freeman, San Francisco, 1973.
[3] Janzen, D. H., “Now is the time”. Philosophical Transactions of the Royal Society of London,
Series B, vol. 359, pp. 731–732, 2004.
[4] K. J. Gaston and M. A. O’Neill, “Automated species identification–why not?” Philosophical
Transactions of the Royal Society of London, Series B, vol. 359, pp. 655–667, 2004.
[5] P. F. Culverhouse, R. Williams, B. Reguera, V. Herry and S. González-Gils, “Do experts make
mistakes?” Marine Ecology Progress Series, vol. 247, pp. 17–25, 2003.
[6] W. P. Colquhoun, “The effect of a short rest pause on inspection efficiency”. Ergonomics, vol.
2, 367–372, 1959.
[7] W. J. Zachariasse, W. R. Riedel, A. Sanfilippo, R. R. Schmidt, M. J. Brolsma, H. J. Schrader,
R. Gersonde, M. M. Drooger and J. A. Brokeman, Micropaleontological counting methods and
techniques – an exercise on an eight meters section of the Lower Pliocene of Capo Rossello,
Sicily. Utrecht Micropaleontological Bulletins, vol. 17, pp. 1–265, 1978.
[8] R. Simpson, P. F. Culverhouse, R. Ellis and R. Williams. Classification of Euceratium gran.
pp. 223-230. Neural Networks. IEEE International Conference on Neural Networks in Ocean
Engineering. IEEE, Washington, D. C., 1991.
[9] R. N. Ginsburg, “Perspectives on the blind test”. Marine Micropaleontology, 29, pp. 101–103.
[10] Kelly, M. G. 2001. “Use of similarity measures for quality control of benthic diatom samples”.
Water Research, vol. 35, pp. 2784–2788, 1997.
[11] K. W.Gobalet, “A critique of faunal analysis; inconsistency among experts in blind tests”.
Journal of Archaeological Science, vol. 28(4), pp. 377-386, 2001.
[12] R. L. Kaesler, “A window of opportunity: peering into a new century of paleontology”. Journal
of Paleontology, vol. 67, pp. 329–333, 1993.
[13] Q. D. Wheeler, “Transforming taxonomy”. The Systematist, vol. 22, pp. 3–5, 2003.

229
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 231-236.
ISBN 978-88-8303-295-0. EUT, 2010.

VeSTIS: A Versatile Semi-

Automatic Taxon Identification
System from Digital Images
Nikos Nikolaou, Pantelis Sampaziotis, Marilena Aplikioti,
Andreas Drakos, Ioannis Kirmitzoglou, Marina Argyrou,
Nikos Papamarkos, Vasilis J. Promponas

Abstract — In this work we present a flexible Open Source software platform

for training classifiers capable of identifying the taxonomy of a specimen from
digital images. We demonstrate the performance of our system in a pilot
study, building a feed-forward artificial neural network to effectively classify
five different species of marine annelid worms of the class Polychaeta. We
also discuss on the extensibility of the system, and its potential uses either as
a research tool or in assisting routine taxon identification procedures.

Index Terms — digital image analysis, open source, semi-automatic taxon

identification.

—————————— u ——————————

1 Introduction

A
utomated taxon identification (ATI) can be defined as the process of
automating the routine identification of specimens [1] through the
exploitation of modern computer science technologies and domain
knowledge. ATI methods are based on mathematical descriptors of morphological
[1], [2], [3], behavioural [4] or genetic [5] characters. These data are used as
input into pre-processing and analysis pipelines, which are most often based on
statistical or machine learning methods. ATI procedures are quickly becoming a
necessity in the effort to understand and monitor global biodiversity.
So far, several research efforts to deal with ATI from digital images have been

————————————————
N. Nikolaou, I. Kirmitzoglou, V.J. Promponas are with the Bioinformatics Research Laboratory,
Department of Biological Sciences, University of Cyprus, P.O. Box 20537, 1678 Nicosia, Cyprus.
E-mail: [email protected], [email protected], [email protected].
M. Aplikioti, A. Drakos, M. Argyrou are with the Department of Fisheries and Marine Research, 101
Vithleem Street, 1416 Nicosia, Cyprus. E-mail: [email protected], andreas_drakos@
hotmail.com, [email protected].
P. Sampaziotis, N. Papamarkos are with the Department of Electrical and Computer Engineer-
ing, Democritus University of Thrace, 67100 Xanthi, Greece. E-mail: [email protected],
[email protected].

231
reported [2], with three major ones focusing on the implementation of semi-
automatic species identification systems; (i) DAISY [3], (ii) SPIDA web [6], and (iii)
ABIS [7]. Important drawbacks of such systems are that they are either suitable
for a relatively narrow taxonomic range (e.g., SPIDA, ABIS) or unavailable for
public use (e.g. DAISY, ABIS). Nevertheless, both these shortcomings could
be eliminated in a community-based approach with the availability of suitable
extensible platforms open for further development. Extensibility can be achieved
in a dual manner: (i) at the software component level (e.g. by an Open Source
modular software), and (ii) at the data level, with a flexible scheme to permit
incorporation of novel data types regarding the taxonomic range accepted by
the system, or data and feature types utilized in the ATI task.
In this work, we present our progress in designing and implementing such
an Open Source computer system, VeSTIS. We demonstrate VeSTIS in the
systematic identification of 5 species of the Class Polychaeta (Phylum Annelida),
a marine macroinvertebrate group well known for the identification difficulties it
presents.

2 Materials and Methods

2.1 Description and Key Features of the System

VeSTIS is intended to be a generic user-friendly platform capable of virtually

identifying any taxonomic unit. It currently embeds a large number of state-of-
the-art digital image analysis, enhancement and pattern recognition algorithms
making it independent from the use of commercial software. Moreover, VeSTIS
incorporates an SQL-based database schema and client-server technology
to allow multiple users working simultaneously. The database schema was
specifically designed to aid easy storage and retrieval of meta-data and to allow
publishing of its contents on the Internet. This adds to the extensibility of the
platform by facilitating the development of web-modules, such as a national
biodiversity portal or a web-application for remotely identifying specimens
through the users’ browser. Finally, a very important characteristic is the ability
to train VeSTIS with user-selected features in order to optimize the ATI process.

2.2 Species Selection, Sample Collection and Image Acquisition

In order to test the functionality of the system, five Polychaete species were
used: Nematonereis unicornis (Smarda, 1861), Marphysa bellii (Audouin &
Milne-Edwards, 1833), Polyophthalmus pictus (Dujardin, 1839), Armandia
polyophthalma (Kükenthal, 1887) and Terebellides stroemi (Sars, 1835). These
species were selected due to: (i) their high abundance in the coastal waters of
Cyprus, and (ii) the relatively few problems in their identification compared to
other Polychaete species.
Samples were collected with a Van Veen grab from a number of coastal
sampling stations at depths of 25-35m in soft substrates. They were then
sieved with a 0.5mm sieve, fixed and properly preserved. Finally, all Polychaete

232
specimens were identified to species level with the use of stereoscopes and
microscopes.
Prior to finalising the exact photo-shooting conditions, we evaluated a series
of factors directly related to the quality of the shots; i.e. various magnifications,
background colour, lighting source and homogeneity, specimen body parts and
their orientation, as well as specimen fixation. The best results were obtained by
fixing the specimens between slides against a uniformly black background. For
illuminating the system we used two Leica CLS150X cold light sources with the
optic fibres oriented in a way that minimized shadows. For this demonstration,
we focused on the frontal body part of the animals and specifically on the head
and the first 10 segments.
All images used for training and validating VeSTIS were acquired using a
Leica DFC290 camera mounted on a Leica MZ7.5 stereo-microscope. Photos
were taken under specimen-size dependent magnifications (in the 12.6x-32x
range) with the maximum resolution supported by the camera (3.2 MP) through
the Leica Application Suite (LAS) software. Image pre-processing was carried
out within VeSTIS.

2.3 Image Pre-processing and Feature Extraction

Object (specimen) orientation correction: Image orientation is corrected, for

the specimen to lay in a horizontal direction (Fig. 1A and 1B). This is important for
object contour representation (see below).
Image segmentation and object isolation: In order to isolate the object in the
image, we used the Otsu binarization method [8]. This is a segmentation process
which automatically creates a black (object) and white (background) image
(Fig. 1C) based on the image histogram. Using connected component analysis,
VeSTIS locates and isolates the object.
Object contour representation & feature vector generation: Upper/lower
object profile features are computed by recording the distance of the lower
boundary of the bounding box to the furthest/closest object pixel for each image
column.

Fig. 1 – (A) Original, and (B) Corrected specimen orientation. (C) Segmentation and object
isolation. (D) Contour representation. (E) An image classified as bad due to curvature.

233
 All values vary between 0 and 1, since they are normalized by the height of
the object (Fig. 1D). These two profiles can be considered to form a closed
curve, allowing the use of Fourier descriptors [9] to mathematically describe the
object’s contour. Fourier descriptors allow bringing the power of Fourier theory
to shape parameterisation by characterising a contour with a set of numbers
that represent the frequency content of a whole shape. They are invariant to
rotation, scale, and translation and are used as the input vector for the feed-
forward artificial neural network (FFANN).

2.4 Generation of training and validation data sets

For generating training/validation sets, we manually classified all images

based on species, specimen, orientation and condition. For four of the species
in question, orientation was either dorsal or ventral. For T. stroemi only lateral-
view photos were taken, mainly because of the species’ morphology. Images
were classified as good (G) whenever the specimen was in a good condition
or bad (B) if the specimen was curved or moderately destroyed (Fig. 1E).
We then created 3 training sets based on specimens’ orientation, using only
images flagged as good. Following a similar procedure we generated 9 different
sets for evaluation purposes using both good & bad images (Tab. 1). We only
included good images in training sets to reduce noise and test the ability of our
approach to correctly classify problematic images/specimens. Multiple images
were acquired for each specimen. However, a single image of each individual
was included in either the training or the validation sets, in order to (i) avoid
over-fitting during training, and (ii) minimize any bias on the estimation of the
performance of the classifier. Thus, any pair of training-validation sets was
strictly disjoint.

Training sets Validation sets

TS1 TS2 TS3 VS1 VS2 VS3 VS4 VS5 VS6 VS7 VS8 VS9

DLG VLG DVLG DLG VLG DVLG DLB VLB DVLB DLGB VLGB DVLGB

Tab. 1 – Image types included in different training and validation data sets. D, V, L =
Dorsal, Ventral, Lateral view; G, B = Good, Bad image classes.

3 Results and Discussion

Two FFANNs were trained for each training set in batch mode with the
resilient back-propagation learning algorithm [10], each initialized with different
random weights. A fully connected architecture, with a single hidden layer of
30 neurons and a sigmoid activation function, proved to be good choices after
experimentation. Five output units served for classifying each specimen to the
respective species using a ‘winner-take-all’ output encoding scheme. FFANNs
were trained for 2000 epochs, and in all cases the mean squared error of desired
versus predicted outcomes of the networks converged to very small values. The

234
performance of each FFANN was evaluated with the independent validation
sets (Tab. 2). We also observed the performance of a simple ensemble average
of independent classifiers trained with different types of data. In several cases
the performance was drastically improved (Tab. 2).

FFANN-TS1 FFANN-TS2 FFANN-TS3 ENSEMBLE

VS1 0.702 (0.017) 0.667 (0.034) 0.702 (0.051) 0.738

VS2 0.693 (0.016) 0.727 (0.000) 0.727 (0.000) 0.705

VS3 0.698 (0.000) 0.698 (0.016) 0.715 (0.025) 0.709

Tab. 2 – Evaluation of identical FFANNs trained with different training sets on

independent validation data sets. For each FFANN the performance reported
corresponds to the average overall performance (and standard deviation) of two
independently trained networks initialized with different random weights. A simple
ensemble averaging approach often seems to outperform individual classifiers. Data
sets are described in Tab. 1. Specimen orientation seems to be an important factor
affecting classification accuracy. As expected, results obtained with bad images were
clearly inferior (data not shown).

Species identification is a painstaking and time-consuming task, which

requires highly skilled and adequately trained scientific personnel. Although
the design and implementation of reliable and accurate ATI methods is a
challenging problem, it will definitely give rise to more experimentation and thus
to the growth and evolution of systematics. It is anticipated that Open Source
solutions will boost development, applicability and usage of ATI methods similar
to what has been experienced in the field of computational molecular biology.
We are currently working on adding more software components to VeSTIS
(feature extractors, classifiers, etc.). We specifically plan to address the feature
selection task, since classification quality is expected to depend mainly on the
features rather than the classifier. This is also an attempt to cover a gap in the
literature that mainly deals with the effectiveness of classifiers.
VeSTIS, although currently in alpha phase, is being actively developed. We
expect to release the first beta binaries and source code at the url http://troodos.
biol.ucy.ac.cy/BRL/ within late 2010.

Acknowledgement

This work was co-funded by the Republic of Cyprus and the EU European Regional
Development Fund (ERDF) through a grant from the Cyprus Research Promotion
Foundation (AEIFORIA/FISI/0308(BIE)/10).

References
[1] K. J. Gaston and M. A. O’Neill, “Automated Species Identification: Why Not?”, Philos. Trans.
R. Soc. Lond. B. Biol. Sci., vol. 359, pp. 655-67, 2004.
[2] N. MacLeod, Automated Taxon Identification in Systematics: Theory, Approaches and
Applications. Boca Raton, FL: CRC Press, 2008.
[3] A. T. Watson, M. A. O’Neill and I. J. Kitching, “Automated Identification of Live Moths

235
 (Macrolepidoptera) Using Digital Automated Identification System (Daisy)”, Systematics and
Biodiversity, vol. 1, pp. 287-300, 2004.
[4] J. Tanttu, J. Turunen, A. Selin and M. Ojanen, “Automatic Feature Extraction and Classification
of Crossbill (Loxia spp.) Flight Calls”, Bioacoustics, vol. 15, p. 251, 2006.
[5] A. Valentini, F. Pompanon and P. Taberlet, “DNA Barcoding for Ecologists”, Trends in Ecology
& Evolution, vol. 24, pp. 110-117, 2009.
[6] K. N. Russell, M. T. Do, J. C. Huff and N. I. Platnick, “Introducing Spida-Web: Wavelets, Neural
Networks and Internet Accessibility in an Image-Based Automated Identification System”. In:
N. MacLeod (ed.), Automated Taxon Identification in Systematics: Theory, Approaches and
Applications, Boca Raton, FL: CRC Press, pp. 131-152, 2008.
[7] T. Arbuckle, S. Schroder, V. Steinhage and D. Wittmann, “Biodiversity Informatics in Action:
Identification and Monitoring of Bee Species Using Abis”. In: 15th International Symposium
Informatics for Environmental Protection, Zurich, pp. 425-430, 2001.
[8] N. Otsu, “A Threshold Selection Method from Gray-Level Histograms”, IEEE Transactions on
Systems, Man and Cybernetics, vol. 9, pp. 62-66, 1979.
[9] O. Petkovic and J. Krapac, Shape Description with Fourier Descriptors, Technical Report,
2002.
[10] M. Riedmiller and H. Braun, “A Direct Adaptive Method for Faster Backpropagation Learning:
The Rprop Algorithm”. In: IEEE International Conference on Neural Networks, San Francisco,
1993.

236
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 237-242.
ISBN 978-88-8303-295-0. EUT, 2010.

Iterative Search with Local

Visual Features for Computer
Assisted Plant Identification
Wajih Ouertani, Pierre Bonnet, Michel Crucianu,
Nozha Boujemaa, Daniel Barthélémy

Abstract — To support computer assisted plant species identification in a

realistic, uncontrolled picture-taking condition, we put forward an approach
relying on local image features. It combines query by example and relevance
feedback to support both the localization of potentially interesting image
regions and the classification of these regions as representing or not the
target species. We show that this approach is successful, and makes prior
segmentation unnecessary.

Index Terms — assisted identification, biodiversity informatics, local features,

local query, object localization, relevance feedback.

—————————— u ——————————

1 Introduction

G
iven the large volume and increasing accessibility of biodiversity data -
e.g. Encyclopaedia of Life [1], Atlas of living Australia [2], or ZipcodeZoo
- gathered from all over the world, it is even more important to explore,
master and capitalize this type of knowledge [3]. Joint efforts of biologists,
information science and data-mining communities are required for solving
significant common problems. As biological image databases are increasing
rapidly [4], automated species identification based on digital data is of great
interest for accelerating biodiversity assessment, research and monitoring [5].
We put forward here an interactive identification approach in which a botanist
having a partially annotated a large image database is assisted by a Relevance
Feedback search mechanism to identify a plant species. The botanist can then
easily select the relevant unlabeled images (without having to go through the
entire database) and label them at once with the name of the species.

————————————————
W.Ouertani, M. Crucianu and N. Boujemaa are with INRIA, IMEDIA Project, BP 105, 78153 Le
Chesnay cedex, France. E-mail: (Wajih.Ouertani, Michell.Crucianu,
Nozha.Boujemaa)@ inria.fr.
P. Bonnet, D. Barthélémy and W.Ouertani are with INRA, Amap Joint research unit, CIRAD,TA
A-51/PS2, 34398 Montpellier cedex 5, France. E-mail: (Pierre.Bonnet, Daniel.Barthelemy)@cirad.fr.

237
2 Context and resulting challenges

2.1 Content-based image retrieval and interactive identification

In a query by visual example (QBVE), an example image is first provided to

the search engine as a visual query. The engine returns images that are visually
similar to the query image, using a metric on the space of the low-level features
that represent the images. Motivated by the “semantic gap” issue, i.e. the fact
that such features seldom reflect the user’s intention, a Relevance Feedback
(RF) [6] mechanism includes the user in the retrieval process. In a RF session,
the search result is iteratively refined. For a given query, the system first retrieves
a set of images ranked according to the predefined similarity measure between
the query vector and feature vectors of images in the database. Then, the user
provides feedback regarding this result, by qualifying the returned images as
either “relevant” or “irrelevant”. From this feedback, the engine iteratively learns
the visual features of the images, and returns improved results to the user. A
good RF mechanism should find the user intention with minimal interaction [7].
This retrieval refinement technique was applied to botanical databases
with pictures taken in controlled conditions [8], but it has important limitations
resulting from the global image description. To remove such restrictions on
picture-taking conditions, we extend here RF to the use of local features (LF).
This is a more adequate representation of image regions, which allows users to
provide a precise feedback by freely selecting relevant and irrelevant regions of
interest in images.

2.2 Challenges

We address here learning and recognition challenges that come from strong
variations in viewpoint, picture-taking conditions, interactivity and generalization
requirements. Recent work on plant species identification requires reliable prior
segmentation of informative organs such as leaves [9], [10] (with controlled
picture-taking conditions) or flowers [11] (less restrictive conditions). With such
well-controlled pictures, the shape of a leaf, its margins, or several local and
region-based features of flowers are employed for recognition. In general,
due to variations in the natural environment, plant accessibility, picture-taking
system and intention, an object of interest (a plant or a plant part) may appear
on different backgrounds and cover a potentially small part of the image (see
first row in Fig. 1). This supports the use of LF to focus on the target object. Also,
in a botanical identification context, some images illustrate global aspects of a
plant or of an inflorescence, while others show details having different visual
attributes. A same object of interest could thus be represented in various poses
and at different scales (see second row in Fig. 1).
Relevance feedback brings in two additional challenges. First, the search
engine should respect the interactivity requirement, i.e. quickly respond during
each round. Even if joint object segmentation and recognition (e.g. [12]) could
improve identification, its additional cost makes it inappropriate for interactive

238
retrieval. Second, at each RF round the user only labels a few images. For the
retrieval session to be successful, the system should generalize well from these
few examples. In the next section we propose an approach that addresses
these problems using LF.

Fig. 1 – Background variations of an inflorescence of Habenaria species (1st row), scale

and pose variation of an inflorescence of Cleisomeria lanatum (Lindl. ex G.Don, 2nd row).

3 Identification Approach
We propose to jointly use search by example with local queries and supervised
classification (with Support Vector Machines, SVM). Every RF round thus
consists of two stages: (1) QBVE using as query the LF that were previously
found relevant; (2) result re-ranking by the SVM decision function, applied to
the potentially relevant set of features in every returned image. This joint use of
QBVE and SVM classification serves two purposes. First, it allows to locate, in
the returned images, the potential regions of interest (see Fig. 2, green and red
points) that have to be evaluated by the SVM. A region of interest is here the
set of LF that were found to be individually similar to some LF in the query. An
image can indeed contain objects from multiple classes; our approach will focus
on the potentially relevant parts and ignore other, irrelevant parts (blue points in
Fig. 2). In this context, the task of the SVM is to solve ambiguity and distinguish
sets of LF that belong to the target specie (Fig. 2, middle) from sets composed
of LF that are individually similar to relevant LF but, when considered together,
do not correspond to the target species (Fig. 2, right).
Second, QBVE can be very fast with an appropriate index structure - we rely
here on a posteriori multi-probe locality sensitive hashing [13] - and only images
containing hit points (i.e. points that are individually similar to relevant LF)
have to be evaluated by RF rather than all the images in the database, which
significantly improves scalability.

239
Fig. 2 – Region of interest localization: user target (left) and two candidate images with
LF belonging to the target (middle) or not (right). The other LF are ignored.

We assume that the distribution of LF in the selected sets brings relevant

discriminating information with respect to the joint presence of LF, so we employ
the pyramidal matching kernel (PMK, [14]) or the kernel based on random
histograms (RH, [15]). The SVM has thus to downgrade image regions (sets
of LF) whose LF are individually similar to LF of the target species, but whose
distribution does not correspond to this target.

4 Experimental Evaluation
We employed two different image databases for the evaluations. The first
one was produced by AMAP Joint Unit on Laos orchid’s reproductive organs
(mainly inflorescences and flowers). It contains 1913 images for 181 orchid
species. There are significant variations in scale, pose and lighting (see Fig. 1,
2). Botanists manually labelled 2347 regions of interest. The second database
is Oxford flowers 17 (www.robots.ox.ac.uk/~vgg/data/flowers/17/), consisting of
17 flower categories with 80 images each. The database includes common UK
flowers; there is a significant variation within a same class and close similarity
between several classes. There is also a ground truth showing fine flower
segmentation for a subset of the images [11].
We compare RF with global image description (GF_RF) to RF with local
descriptions (LF_RF_QVE_Harris, LF_RF_QVE_SIFT). The global image
description employed (named “joint description” below) concatenates a
Laplacian weighted RGB histogram, a Fourier-based histogram and a Hough
histogram [2]. Two types of LF were employed: (i) joint description (with coarser
histograms) obtained in the neighbourhood of Harris colour points, and (ii) SIFT
[16]. The experiments were performed by using the ground truth to emulate user
feedback under realistic conditions. Each RF session consists of 8 iterations.
At every iteration, the emulated user labels the first 3 relevant and the first 3
irrelevant unlabelled regions. Fig. 3 shows the mean average precision (MAP)
of system’s responses where recall equals precision (MAP at R=P), for the three
RF mechanisms. Only the 10 orchid classes having enough image examples
were used for generating RF sessions. Fig. 3 (left) shows that, even with few
iterations (1st to 4th, less than 50% of the available training data), RF with LF
outperforms global RF. We also note that the results obtained with SIFT (features
ignoring colour!) are better than those with Harris points whose description
includes colour. This is due to the fact that scale and shape variations within a
same class are more important than colour differences between classes in this
dataset.

240
Fig. 3 – MAP evolution over RF iterations. Left: on Orchids database. Right: on Oxford
flowers 17 database, with and without segmentation masks in prediction stage.

Using LF_RF_QVE_Harris and the fine segmentation ground truth provided

in Oxford flowers 17 database, we performed two experiments in which we use
segmented objects as training examples and, for the prediction stage, we either
(i) use only hit points (retrieved by QBVE) that fall in pre-segmented objects
of interest in a candidate image (TP_GTMasks), or (ii) use all the hit points
retrieved in a candidate image (T_GTMasks). As can be seen in Fig. 3 (right),
the object localisation given by the QBVE stage allows to reach a performance
that is close to the one obtained with fine prior segmentation. We also find that
the inclusion of a small part of object’s neighborhood provides a relevant context
that increases recognition accuracy.

Fig. 4 – Object localization examples on Oxford Flower 17 database, points showing

the object of interest. From left to right: Colt’s Foot, Daisy Flower, Buttercup, Tiger Lily.

5 Conclusion
Content-based image search can provide a significant contribution to plant
species identification. However, to make it successfully applicable to realistic
contexts, we argue that it is necessary to let the user interact with the system on
the basis of local image descriptions that allow to focus on the relevant part of
an image. We proposed a relevance feedback method relying on local images
features. It also makes use of an LF retrieval stage in order to locate potentially
interesting image regions and improve scalability to larger image databases.
We have shown that this approach can be successful and that it makes prior
segmentation unnecessary. The results also show how important it is to devise
local features that are robust to most of the variations that can be expected
when pictures are taken in more general, uncontrolled conditions.

241
Acknowledgements

This work is part of the flagship project of Agropolis fondation: Pl@ntNet, http://www.
plantnet-project.org.

References
[1] E. O. Wilson, “The encyclopedia of life”, Trends in Ecology and Evolution, 18 (2), 2003.
[2] Anon., 2008, “Atlas of Living Australia – sharing biodiversity knowledge to shape our future”,
Proc. R. Soc. Western Australia, Nov. 2008.
[3] N. F. Johnson, “Biodiversity informatics”. Annu. Rev. Entomol., vol. 52, pp. 421-438, 2007.
[4] S. J. Baskauf and B. K. Kirchoff, “Digital plant images as specimens: toward standards for
photographing living plants”. Vulpia, vol. 7, pp. 16–30, 2008.
[5] K. J. Gaston and M. A. O’Neil, “Automated species identification: why not?” Phil. Trans. R.
Soc. B., vol. 359, pp. 655-667, 2004.
[6] X. S. Zhou and T. S. Huang, “Relevance feedback for image retrieval: a comprehensive
review”, Multimedia Systems, vol. 8, no. 6, pp. 536-544, 2003.
[7] M. Ferecatu, Image retrieval with active relevance feedback using both visual and keyword-
based descriptors. PhD thesis, Université de Versailles, France, 2005.
[8] M. Coutaud, P. Bonnet, A. Joly, R. Enficiaud, N. Boujemaa and D. Barthélémy, “Advances in
taxonomic identification by image recognition with the generic content-based image retrieval
IKONA”. In: e-Biosphere 09: Intl. Conf. on Biodiversity Informatics, London, 2009.
[9] I. Yahiaoui, N. Hervé and N. Boujemaa, “Shape-based image retrieval in botanical collections”.
In: 7th Pacific Rim Conf. on Multimedia, LNCS vol. 4261, pp. 357–364, 2006.
[10] P. N. Belhumeur, D. Chen, S. Feiner, D. W. Jacobs, W. J. Kress, H. Ling, I. Lopez, R.
Ramamoorthi, S. Sheorey, S. White and L. Zhang, “Searching the world’s herbaria: A system
for visual identification of plant species”. In: European Conf. on Computer Vision, LNCS vol.
5305: pp. 116–129. Springer, 2008.
[11] M.-E. Nilsback and A. Zisserman, “Automated flower classification over a large number of
classes”. In: 6th Indian Conf. on Computer Vision, Graphics & Image Proc., pp. 722–729,
Washington, DC, USA.IEEE Computer Society, 2008.
[12] J. Shotton, J. Winn, C. Rother and A. Criminisi, “Textonboost for image understanding: Multi-
class object recognition and segmentation by jointly modeling texture, layout, and context”. Int.
J. Comput. Vision, vol. 81(1), pp. 2–23, 2009.
[13] A. Joly and O. Buisson, “A posteriori multi-probe locality sensitive hashing”. In: 16th ACM intl.
conf. on Multimedia, pp. 209–218, New York, NY, USA, 2008.
[14] K. Grauman and T. Darrell, “The pyramid match kernel: Efficient learning with sets of features”.
J. Mach. Learn. Res., vol. 8, pp.725–760, 2007.
[15] W. Dong, Z. Wang, M. Charikar and K. Li, “Efficiently matching sets of features with random
histograms”. In: 16th ACM intl. conf. on Multimedia, pp. 179–188, New York, NY, USA, 2008.
[16] D. G. Lowe “Distinctive image features from scale-invariant keypoints”. Intl. J. Comp. Vis., vol.
60(2), pp. 91-110, 2004.

242
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 243-248.
ISBN 978-88-8303-295-0. EUT, 2010.

Image data banks

and geometric morphometrics
Anna Loy, Dennis E. Slice

Abstract — This paper examines the opportunities offered by recent

advances in digital image processing to allow access to natural history
museum collections without direct handling of specimens. It specifically refers
to two- and three-dimensional data recording and analysis in the frame of
geometric morphometrics.

Index Terms — geometric morphometrics, 2D and 3D image recording,

landmarks, osteological collections.

—————————— u ——————————

1 Introduction

O
steological collections, especially skull collections, represent an ideal
material for the study of morphological variation in both time and space
in a variety of vertebrates. Its multiple functional properties (protection
of the brain and sense organs, feeding and respiratory structures) make the
skull a highly informative structure where both highly conservative and plastic
characters coexist [1]. These qualities led to a rich production on the intra- and
interspecific variation of vertebrates based on skull features [1] and to a precise
coding of traditional quantitative characters [2]. The geometric morphometrics
revolution in the ’90s [3], [4] offered a new powerful tool to investigate the
variation of biological forms, allowing the distinction between size and shape
variation. Geometric Morphometric (GM hereafter) studies have found their
elective applications in the analysis of osteological collections devoted to
clarification of clarify phylogenetic and evolutionary patterns among vertebrates,
[5], [6], [7], [8]. Basic data for GM are usually recorded either from 2D or 3D
images. Taking advantage of the tremendous advances in digital technologies,
museums can play a fundamental role for future GM studies by offering an easy
and rapid remote access to their collections [9], [10].

————————————————
A.Loy is with the Università del Molise, Pesche, Italy, I-86090. E-mail: [email protected].
D.E.Slice is with the Florida State University, Tallahassee, FL 32306-4120. E-mail: [email protected].

243
2 Geometric Morphometrics
GM uses sets of Cartesian coordinates, such as (semi-)landmark locations,
outlines, curves, and surfaces, to capture the geometric information about
biological structures and preserves that information throughout the analyses,
including the multivariate treatments of data [4], [5], [11]. Most multivariate methods
of GM are linearizations of statistical analyses of distances and directions in the
Kendall’s shape space. Each point in this shape space represents the shape of
a configuration of points (landmarks) in some Euclidean space, irrespective of
size, position, and orientation [5]. In shape space, scatters of points correspond
to scatters of entire landmark configurations (specimens), not merely scatters of
single landmarks, and differences among shape configurations are most often
expressed as cord distances relative to a curved, generalized Procrustes space
[4], [12], [13], [14], [15].
Configurations are described by either two (x,y) or three (x,y,z) Cartesian
coordinates of homologous points (landmarks). The advantage of working
with 2D landmarks is that these data are easily recorded from digital pictures
through easily accessible and friendly software, e.g., TpsDig [16]. Meanwhile,
studies using coordinates of 3D points are becoming standard in some fields,
such as physical anthropology [17], [18]. A distinct advantage of the use of
3D coordinates is that the definitions of landmark points are often much less
arbitrary in three dimensions than they are in 2D projections [5]. An historical
disadvantage of three dimensional landmarks is that they can only be recorded
directly from the objects by means of devices like the 3D Microscribe or Polhemus
digitizers or gathered from 3D pictures obtained from very expensive scanners.
Unfortunately, statistical methods for dealing with such additional data types
(surfaces and volumes) are still in their infancy. Moreover, we also still lack
effective methods for the visualization of genuinely 3D shape variation whether
for points or more complicated data structures [5].

3 2D and 3D digital images for GM analyses

Datasets used for GM analyses can be derived from 2D and 3D digital images
(Fig. 1). Digital imaging has undergone an explosive development in the last
decade, allowing access to high resolution and low-cost devices, especially in
the case of 2D pictures.
Digital images used to collect data for GM analyses have to comply with some
basic requirements, as imaging systems include artifacts related to acquisition,
storage, and display processes [7], [21], [22], [23], [24]. The use of high-quality
optical equipment and caution in specimen position and distance from the
camera will help to reduce the effect of these artifacts.

244
  
Fig. 1 – Devices used for 2D and 3D data recording for GM. A. Binocular microscope
connected to a digital camera and a pc. B. Digital camera mounted on a tripod at a
fixed distance from the skull. C. Microscribe used to get 3D coordinates directly from
the skull.

The choice of digital camera should take into account resolution, accuracy,
tonal range, color purity and accuracy, white balance, and image noise (see
[23] and http://www.imaging-resource.com). Shadows and the reflecting surfaces
of the object (bone surfaces often reflect a lot) may impede the view of important
features like skull sutures. Lighting that maximizes the visibility of the whole
object is recommended.
Once the equipment is calibrated, a standard protocol should be adopted for
image recording that will retain all information needed for GM analyses (Fig. 2).

 
Fig. 2 – Flow chart of 2D data recording and processing for GM.

The protocol should closely adhere to the following recommendations:

1. Place the object on a soft and possibly dark substrate to allow the object to
be placed exactly horizontal to the plane.
2. Images should always include a scale factor and the specimen label.
3. Keep the object as much as possible in the centre of the image, and place
the camera at such a fixed distance that distortion effects do not occur at
image margins [24].

245
 Three-dimensional images are more complicated, both in their nature and
acquisition. They can be volumetric data encoding some spatial property,
e.g., x-ray attenuation (CT scans) or water content (MRI-Magnetic field
and Radiowave pulses), or they may consist of coordinates of point clouds
representing the surface of a specimen (preferably with texture information).
The production of such images has been prohibitively expensive, but is
becoming increasingly affordable and accessible. Micro-CT scanners with
micron-scale resolution are appearing within universities (e.g.http://www.ucalgary.
ca/mousegenomics/3DMorphometrics, http://www.ctlab.geo.utexas.edu/index.php,
http://micro-ct.at/). Usable surface scanning devices are ever more affordable
(e.g., http://www.nextengine.com/). Students at FSU last year built a working scanner
for only 15.00 USD (http://www.david-laserscanner.com/). Just as with 2D images,
though, care must be taken to calibrate and test the product of any 3D scanning
modality, and this information should be made available along with the image.

4 Images databanks, museums and GM analyses

Museum’s collections play a fundamental role in geometric morphometric
studies. By creating specific digital image databanks, museums would greatly
speed up the data collection of large samples, reduce the costs of the research,
and minimize damage to collections due to specimen handling. They could also
benefit from extra income, as access to images could be regulated and charged,
and advice from specialists could rapidly solve diagnostic and systematic
problems often posed by specimen labels. To be suitable to GM analyses, image
data banks should meet some basic requirements, including standards for image
recording, acquisition, storage, and analysis of morphometric information. A
number of excellent image repositories exist today. These include:
• Morphbank (http://www.morphbank.net/),
• MorphoBank (http://www.morphobank.org),
• and eSkeletons (http://www.eskeletons.org/),
designed primarily for two dimensional images, and:
• Aves3d (http://aves3d.org/),
• DigiMorph (http://digimorph.org/), and:
• the Open Research Scan Archive (http://plum.museum.upenn.edu/~orsa/
ORSA/),
for (mainly) CT-based 3D images.
Such web-accessible repositories are valuable resources for any study
concerned with the anatomical variation of the archived material, but there is
much that could be improved. None of these (nor any other) repositories have
incorporated into their structure the capacity for the direct acquisition, storage,
and analysis of morphometric information. Neither do any of these provide the
ability to easily interface with other archives. What is very much needed is an
interface standard that would allow the query and retrieval of material from
multiple online archives. Perhaps more practical, and maybe even better, would
be the development of a meta-interface that could map the idiosyncrasies of the
individual archives to a common access tool. Scientific research would be greatly
enhanced, then, with even the basic capacity to search for and download images

246
(2- or 3D) and associated data for local analysis with standalone morphometric
tools. Better still would be the direct support of such morphometric tools within
the common interface and a secure, quality-controlled extensibility to the root
archives to support morphometric annotation, e.g., labeled point (=landmark)
coordinates, curve descriptors, etc.
As the growing use of medical imaging has allowed the production of 3D
images of extant and fossil specimens, the analysis and visualization of 3D data
and the combination of (semi-)landmarks, outlines, and surfaces are expected
to yield a better description of changes in biological complexes. 3D analyses
are affording several new perspectives in the field of human paleontology and
physical anthropology (see for example [19], [20]). Progress is also expected in
the study of covariation between subsets of landmarks [25] and the extension
of landmark-based morphometrics to the analysis of articulated structures [26].
This progress would be greatly facilitated by standards-based, online archives
that either directly incorporate or otherwise support the acquisition, storage, and
analysis of morphometric data.

References
[1] J. J. Hanken and K. Hall (eds.), The skull. Functional and evolutionary mechanisms. vol. 3.
Univ of Chicago Press, 1993.
[2] O. Thomas, “Suggestions for the nomenclature of cranial length measurements and the cheek
teeth of mammals” Proc.Biol.Soc.Wash., vol. 18, pp. 191-196, 1905.
[3] F. J. Rohlf and L. F. Marcus, “A revolution in morphometrics”. T.R.E.E, vol. 8, pp. 129-132,
1993.
[4] F. L. A Bookstein, “Hundred years of morphometrics”. Acta Zool. Acad. Scient. Hungar., vol.,
44, pp. 7-59, 1998.
[5] D. C. Adams, F. J. Rohlf and D. E. Slice, “Geometric Morphometrics: Ten Years of Progress
Following the ‘Revolution’” Ital. J.Zool., vol. 71, pp. 5-16, 2004.
[6] D. E. Slice (ed.), Modern Morphometrics in Physical Anthropology. Kluwer Academic/Plenum,
2005.
[7] P. L. Forey and N. MacLeod (eds.), Morphology, shape and phylogeny. Syst. Ass. Sp. Vol. Ser.
64. London: Taylor and Francis, 2002.
[8] F. J. Rohlf “Geometric morphometrics and phylogeny”. In: P. L. Forey and N. MacLeod (eds.),
Morphology, shape and phylogeny. Syst. Ass. Sp., Vol. Ser. 64. London: Taylor and Francis
pp. 175-193, 2002.
[9] A. Loy, “Morphometrics and theriological collections. Homage to Marco Corti”. Hystrix- It. J.
Mammal., vol. 18, pp. 115-136, 2007.
[10] S. Elton and A. Cardini, “Anthropology from the desk? The challenges of the emerging era of
data sharing”. J. Anthr.Sci., vol. 86, pp. 209-212, 2008.
[11] F. L. Bookstein “Size and shape spaces for landmark data in two dimensions”. Stat. Sci., vol.
1, pp. 181-222, 1986.
[12] D. E. Slice, ”Landmark coordinates aligned by procrustes analysis do not lie in Kendall’s shape
space”. Syst. Biol., vol. 50, pp. 141-149, 2001.
[13] D. G. Kendall, ”Shape-manifolds, Procrustean metrics and complex projective spaces”. Bull.
Lond. Math. Soc., vol 16 pp. 81-121, 1984.
[14] D. G. Kendall, “Exact distributions for shapes of random triangles in convex sets”. Adv. Appl.
Prob., vol. 17, pp. 308-329, 1985.
[15] F. L. Bookstein, Morphometric tools for landmark data: geometry and biology. Cambridge
University Press, Cambridge, 1991.
[16] F. J. Rohlf, tpsDig ver. 2.10. Ecology and Evolution at SUNY Stony Brook, 2006.
[17] D. E. Slice, ”Geometric morphometrics”. Ann.Rev. Anthr., vol. 36, pp. 261-281, 2007.

247
[18] P. Mitteroecker and P Gunz, “Advances in Geometric Morphometrics”. Evol.Biol. vol. 36, pp.
235-247, 2009.
[19] P. Gunz, P. Mitteroecker and F. L. Bookstein, “Semilandmarks in three dimensions”. In: D. E.
Slice (ed.). Modern Morphometrics in Physical Anthropology, Kluwer, 2005.
[20] P. Gunz and K. Harvati, “The Neanderthal “chignon”: Variation, integration and homology. J.f
Hum. Evol. vol. 52, pp. 262-274, 2007.
[21] F. J. Rohlf and F. L. Bookstein (eds.), Proceedings of the Michigan morphometrics workshop.
Sp. Publ. N. 2. Univ. of Michigan Museum of Zoology, Ann Arbor, 1990.
[22] J. M. Becerra, E. Bello and A. Valdecasas “Building your own machine image system for
morphometric analysis: a user point of view”. In: L. F. Marcus, E. Bello, A. Valdecasas (eds.),
Contribution to Morphometrics, Monographias 8, Museo Nacional de Ciencias Naturales, pp.
66-92, 1993.
[23] A. Garcìa-Valdecasas, “Two-dimensional imaging: an update”. In: L. F. Marcus, M. Corti, A.
Loy, G. J. P. Naylor and D. E. Slice (eds.), Advances in Morphometrics, NATO ASI Series A:
Life Sciences vol. 284. Plenum: New York: 71-81, 1996.
[24] M. L. Zelditch, D. L. Swiderski, H. D. Sheets and W. L. Fink, Geometric Morphometrics for
biologists: a primer. London: Elsevier Academic Press, 2004.
[25] F. L. Bookstein, P. Gunz, P. Mitterocker, H. Prossinger, K. S Chafer and H. Seidler, “Cranial
integration”. In Homo: singular warps analysis of the midsagittal plane in ontogeny and
evolution. J.Hum. Evol., vol. 44, pp. 167-187, 2003.
[26] D. C. Adams, “Methods for shape analysis of landmark data from articulated structures”. Evol.
Ecol. Res., vol. 1, pp. 959-970, 1999.

248
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 249-250.
ISBN 978-88-8303-295-0. EUT, 2010.

Outline analysis for identifying

Limodorum species
from seeds
Sara Magrini, Sergio Buono, Emanuele Gransinigh,
Massimiliano Rempicci, Silvano Onofri, Anna Scoppola

Abstract — Limodorum trabutianum Batt. is an orchid species of the

Italian flora, with a central-western stenomediterranean distribution,
that is sporadic in the western part of the distribution area of the more
common L. abortivum (L.) Sw., an eurimediterranean species. It occurs
in Italy only with a few populations in Tuscany, Latium, Umbria, Sicily
and Sardinia [1], often with L. abortivum [2], [3], [4] from which it is
easily recognizable only during anthesis for the denser inflorescence
spike, the ribbon-like lip without differentiation in epychile and
hypochile, and for the spur that is very short or absent [5]. On the
contrary, the identification of these two taxa during the fruiting phase
is rather difficult or even impossible. The aim of this study is to verify
the taxonomic value of Limodorum seeds, particularly of their shape,
as highlighted from recent studies for other orchids [6], [7], in order to
establish its usefulness for recognizing the two species.
We have identified 5 Italian populations of the two taxa: 2 populations
of L. trabutianum, one within the Marturanum Regional Park
(Barbarano Romano, Viterbo), the other near Cortona (Arezzo), and
3 populations of L. abortivum, near S. Martino al Cimino (Viterbo),
in the M. Casoli Reserve (Bomarzo, Viterbo), and in the same site
of L. trabutianum within the Marturanum Park. The phenology of
these populations was monitored to collect mature seeds from
naturally dehiscing capsules. The intra- and interspecific variability of
seed shapes was analyzed with the methodology of Elliptic Fourier
descriptors [8], which allows to describe in terms of harmonics each
two-dimensional shape with a closed outline. For this outline analysis
we used the software package SHAPE 1.3 [9]. An average of 100
seeds from each species and from each site was photographed with
a NIKON Coolpix 5000 camera mounted on a LEITZ-ARISTOPLAN
microscope, obtaining 500 digital images with a resolution of 300
dpi and a size of 800 x 1000 pixels. All images were prepared using
Adobe Photoshop 7.0: as a first step, every foreign element was
eliminated from the picture, thereby isolating the single seed, then its
contrast with the background was maximized, and finally all images
were saved in .bps format (24bit). The color images were converted

————————————————
S. Magrini, S. Onofri and A. Scoppola are with the Tuscia Germplasm Bank, Botanical Gardens of
Viterbo - University of Tuscia, largo dell’Università s.n.c. - blocco C, Viterbo I 01100, Italy. E-mail of
S. Magrini: [email protected].
S. Buono, E. Gransinigh and M, Rempicci are with GIROS, Section “Etruria meridionale”.

249
 to binary with Chain Coder before tracing the outlines in Chain-code,
a coding system that describes the geometrical information of the
shapes. Then the Chain-code file was transformed into a Normalized
Elliptic Fourier file using Chc2Nef using 20 harmonics. The matrix of
the harmonic coefficients underwent a process of data normalization
based on the first harmonic, to transform the data into shape
variables. Subsequently, a PCA was performed on the variance-
covariance matrix of normalized coefficients using PrinComp, which
gives a graphical output of the principal components (average shape
± standard deviations).
The first results of the outline analysis confirm a low intraspecific
variability of seed shape, but show a very high interspecific variability:
L. abortivum seeds are very elongated, from fusiform to filiform, while
L. trabutianum seeds are much wider and have a very lower length/
width ratio. These results allow to distinguish between these two
species even during the fruiting phase, simply using seed shape as
a diagnostic character, avoiding the use of traditional morphometric
analysis which need microscopic measurements.

Index Terms — image analysis, orchids, identification, plants.

—————————— u ——————————

References
[1] A. Scoppola and G. Spampinato (eds.), Atlante delle specie a rischio di estinzione, CD-ROM,
2005.
[2] R. Romolini and C. Merlini, GIROS Notizie, vol. 16, pp. 26-27, 2001.
[3] S. Buono and E. Gransinigh, GIROS Notizie, vol. 23, pp. 3-5, 2003.
[4] G. Ratini, GIROS Notizie, vol. 31, pp. 26-27, 2006.
[5] GIROS, Orchidee d’Italia, Il Castello, Cornaredo (MI), 303 pp., 2009.
[6] M. Aybeke, J. Pl. Biol., vol. 50, pp. 387-395, 2007.
[7] T. A. Akçin, Y. Ozdener and A. Akçin, J. Belgian. Bot., vol. 142, pp. 124-139, 2010.
[8] F. P. Kuhn and C. R. Giardina, Comp. Graph. Ima. Proc., vol. 18, pp. 236-258, 1982.
[9] H. Iwata, Y. Ukai and J. Heredity, vol. 93, pp. 384-385, 2002.

250
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 251-256.
ISBN 978-88-8303-295-0. EUT, 2010.

Geometric morphometrics as
a tool to resolve taxonomic
problems: the case of
Ophioglossum species (ferns)
Sara Magrini, Anna Scoppola

Abstract — A modern method, geometric morphometrics, was used to clarify

the taxonomic position of the European Ophioglossum species: O. azoricum,
O. lusitanicum, and O. vulgatum. The identification of these taxa by traditional
methods is rather difficult, due to different taxonomic interpretations. Sterile
leaf shapes were investigated using a landmark-based method and the
Fourier analysis of outlines. Both methods highlight the shape and the base of
the leaf as an important diagnostic character.

Index Terms — geometric morphometrics, landmark, Ophioglossum, outline

analysis.

—————————— u ——————————

1 Introduction

G
eometric morphometrics [1], [2] is a technique for multivariate shape
analysis that preserves the integrity of biological shape, avoiding its
reduction to linear and angular measures that do not include information
concerning the geometric relationships of the entire subject. allowing to identify
the anatomical areas of morphological remodelling. Two different kinds of
geometric morphometrics are most widely used: landmark-based methods
such a Thin Plate Splines analysis [3], and Fourier analysis of outlines [4]. The
theoretical formulation has been developed in recent decades by the synthesis
of multivariate analysis methods of covariance matrices and methods for direct
visualization of changes in the biological shape [1], [2], [5]. Multiple applications
exist in the biomedical field, in paleontology, anthropology and zoology. In
contrast, geometric morphometrics was almost absent from botanical research
until a few years ago [3], with a few exceptions for Dactylorhiza [6], Quercus [7],
and other groups.
————————————————
S. Magrini is with the Tuscia Germplasm Bank, University of Tuscia, Viterbo, I 01100. E-mail:
[email protected].
A. Scoppola is with the Viterbo Botanical Gardens, University of Tuscia, Viterbo, I 01100. E-mail:
[email protected].

251
 Many groups of plants still require more intensive taxonomic studies. One of
them is the genus Ophioglossum C. Presl, an ancient fern genus of about 25-
30 species of Ophioglossaceae, with a cosmopolitan but primarily tropical and
subtropical distribution. The simple morphology of the sporophyte - a single leaf or
a few - has limited the number of characters available for building classifications
on morphological ground and clarifying relationships, forcing investigators to rely
on details of frond size or sporangia number as diagnostic markers. Problems
associated with the evaluation of these often subtle differences gave rise to
different taxonomic interpretations of the genus [8]. Furthermore, the number
of taxonomic studies of Ophioglossum species is relatively low in Europe [9],
[10], so that many questions concerning Ophioglossum azoricum C. Presl and
its relationships with the other species (O. vulgatum L. and O. lusitanicum L.),
remain unresolved. Our study aims at testing a modern method - geometric
morphometrics - for resolving some of these problems and clarifying the
taxonomic position of these Ophioglossum species.

2 Materials and Methods

2.1 Collecting materials

We have analysed three species: O. azoricum, O. lusitanicum, and O.

vulgatum, as well as some critical populations from Tuscany (Moriglion di
Penna and Monte Pianello, Lucca) and Latium (Selva del Lamone, Viterbo).
We used 100 exsiccata from 9 European herbaria. We photographed with a
digital camera all the sterile leaves in the specimens of Cagliari (CAG), Florence
(FI), Genoa (GE), Pisa (PI), Siena (SIENA), Viterbo (UTV), and Valencia (VAL),
and we collected images from the website of Paris (P), and Vienna (W). We
obtained over 203 images, standardized with a resolution of 300 dpi and a size
of 800x1000 pixels, prepared using Adobe Photoshop 7.0. As a first step, every
foreign element was eliminated from the picture, thereby isolating the sterile
leaf, and then the contrast with background was maximized.

2.2 Landmark-based method

The TPS - Thin-Plate Splines software package was used for landmark-based
morphometrics analysis: TpsDig2.12 to digitize landmarks [11] and TpsRelw 1.42
for statistical analysis [12]. In our case, we could find only 3 homologous points
(landmarks), due to the shape of the leaf, with an entire margin, and the lack of
evident veining. A grid of 11 equally spaced horizontal lines was superimposed
on all images, taking care to position it perpendicular to the longitudinal axis
of the frond and to align the apex and the base with the top and the bottom
line of the grid, respectively, to digitize other 18 semi-landmarks (points defined
relatively to other landmarks). The next step was the development of matrix of
landmark coordinates through TpsRelw 1.42. This program operates through an
algorithm used to describe the mechanical deformation of thin metal foils (thin
plate spline or TPS), building linear combinations of the landmark coordinates.

252
The new variables - called “shape variables” - allow the creation of deformation
grids that visually illustrate the change in shape compared to the average
shape of the sample (consensus). We have analysed the entire dataset and the
partial datasets corresponding to each species and population to defy inter- and
intraspecific variability, respectively, and we have performed a Relative Warps
Analysis on the covariance matrix.

2.3 Outline Analysis

For Outline Analysis we used the software package SHAPE 1.3 [13]. based
on the methodology of Elliptic Fourier descriptors, which allows to describe - in
terms of harmonics - each type of two-dimensional shape with a closed outline.
All images were saved in .bps format (24bit) and were converted to binary with
Chain Coder before tracing the outlines in Chain-code, a coding system that
describes the geometrical information on the shapes. Then, the Chain-code file
was transformed into a Normalized Elliptic Fourier file with Chc2Nef, using 20
harmonics on 77 components. The matrix of the harmonic coefficients underwent
normalization based on the first harmonic, to transform the data into shape
variables. Subsequently, a PCA was performed on the variance-covariance
matrix of normalized coefficients (Elliptic Fourier Descriptors) using PrinComp,
that gives a graphical output of the average shape ± the standard deviation.

3 Results
We report the results of the Relative Warps Analysis (RWA) on the entire
dataset. The analysis of the first two RWs (accounting for 85.10% and 6.07% of
the total variance, respectively), shows a clear separation between O. vulgatum
and O. lusitanicum along RW1, while in the centre of the plot several samples of
O. azoricum are very close to O. lusitanicum. The samples from Lamone (LAM)
and M. Pianello (PIA) overlap with O. azoricum, while those from Moriglion di
Penna (MOR) are distributed along RW1 in the areas of O. vulgatum and of
O. azoricum. Moving along RW1 from negative values, corresponding to O.
lusitanicum, to the positive values of O. vulgatum, the associated deformation
grids show a vertical contraction and an expansion in width, particularly in the
proximal part of the leaf (Fig. 1).
The results of the Outline Analysis of the entire dataset (Fig. 2a) are
consistent with those from RWA: PC1 (accounting for 88.33%) shows the 2
extreme shapes diversified in the proximal part of the leaf, corresponding to O.
lusitanicum for the negative end, to O. vulgatum for the positive end. The results
for the single species show a lower variability in exsiccata of O. azoricum “typus”
(PC1=56.15%) (Fig. 2b), and of O. lusitanicum (PC1=61.34%) (Fig. 2e), while
the greater variability is in O. vulgatum (PC1=70.67%) (Fig. 2d), confirming the
difficulty of a correct identification by traditional morphological methods due to
the different taxonomic interpretations.
The results were used to identify the critical samples from LAM, PIA and MOR.
The LAM outlines (Fig. 2f) have a low variability (PC1=64.15%), they are akin
to O. azoricum outlines (Fig. 2b-c), as the PIA samples (Fig. 2g), which have a

253
greater variability (PC1=80.67%). On the contrary, the MOR samples (Fig. 2h)
are diversified (PC1=91.54%), showing 2 extreme shapes, corresponding to O.
vulgatum (Fig. 2d) and O. azoricum outlines (Fig. 2 b-c).

Fig. 1 – Deformation grids along the first Relative Warp with vectors of deformation
relative to average shape of fronds, showing the anatomical areas of morphological
remodelling.

Fig. 2 – Results from Outline Analysis. First principal component of: (a) the entire
dataset, PC1=88.33%; (b) O. azoricum typus, PC1=56.15%; (c) O.azoricum,
PC1=64.98%; (d) O. vulgatum, PC1=70.67%; (e) O. lusitanicum, PC1=61.34%; and
Ophioglossum sp. from: (f) LAM, PC1=64.15%; (g) PIA, PC1=80.67%; (h) MOR,
PC1=91.54%.

4 Discussion
The geometric morphometrics analysis gave results that lead to a better
characterization of the 3 Ophioglossum species.
The landmark-based analysis does not completely discriminate among the three
species. It clearly separates O. lusitanicum from O. vulgatum, but O. azoricum
and O. lusitanicum are widely overlapping.
More information derives from the deformation grids that highlight a diagnostic
character like the relative width of the leaf, especially in its proximal part, as

254
confirmed by the Outline Analysis. The graphical outputs of the PCA of outlines
show the differences between the three species, confirming shape and base
of the leaf as the main diagnostic characters. O. vulgatum is well distinct from
the other two species by the shape of the lamina, which is from lanceolate to
broadly ovate with a large round and attenuated base. O. azoricum and O.
lusitanicum have a more or less cuneate base, but they are differentiated in the
leaf shape: from lanceolate to narrow ovate in O. azoricum, lanceolate-linear
in O. lusitanicum. Geometric morphometrics also allows to identify the critical
samples: LAM and PIA samples can be referred to O. azoricum, as confirmed
by both Relative Warp and Outline Analysis. Instead, the Outline Analysis of
MOR samples shows a higher variance, although this could be attributed to the
variability of O. vulgatum; probably in this site O. azoricum and O. vulgatum
coexist.

5 Conclusion
The study led to a more accurate characterization of the three species, with
the identification of valid diagnostic characters, and the presumable discovery
of 2 new sites of O. azoricum in Italy, M. Pianello (Lucca) and Selva del Lamone
(Viterbo), even if other analyses (e.g. molecular and caryological) are suggested.
The research will continue with the revision of all samples that showed abnormal
or out-of-range values, and by increasing the dataset with exsiccata from other
herbaria. These results also will allow understanding and defining the actual
distribution of Ophioglossum species in Italy and Europe.

Acknowledgement

The authors wish to thank the curators of the cited herbaria and dr. L. Peruzzi (University
of Pisa) for providing images or exsiccata for this study.

References
[1] F. Bookstein, Morphometric Tools for Landmark Data. Cambridge Univ. Press, 1991.
[2] F. J. Rohlf and L. F. Marcus, “A Revolution in Morphometrics”. Trends Ecol. Evol., vol. 8, pp.
129-132, 1993.
[3] D. C. Adams, F. J. Rohlf and D. E. Slice, “Geometric Morphometrics: Ten Years of Progress
Following the ‘Revolution’”. Ital. J. Zool., vol. 71, pp. 5-16, 2004.
[4] R. J. Jensen, K. M. Ciofani and L. C. Miramontes, “Lines, Outlines and Landmarks:
Morphometric Analyses of Leaves of Acer rubrum, Acer saccharinum (Aceraceae) and Their
Hybrid”. Taxon, vol. 51, pp. 475-492, 2002.
[5] L. R. Rabello, B. Bordin and S. Furtado, “Shape Distances, Shape Spaces and the Comparison
of Morphometric Methods”. Trends Ecol. Evol., vol. 15, pp. 217-220, 2000.
[6] A. B. Shipunov and R. M. Bateman, “Geometric Morphometrics as a Tool for Understanding
Dactylorhiza (Orchidaceae) Diversity in European Russia”. Biol. J. Linnean Soc., vol. 85, no.
1, pp. 1-12, 2005.
[7] V. Viscosi, O. Lepais, S. Gerber and P. Fortini, “Leaf Morphological Analyses in Four European
Oak Species (Quercus) and Their Hybrids: a Comparison of Traditional and Geometric
Morphometric Methods”, Plant Biosyst., vol. 143, no. 3, pp. 564-574, 2009.
[8] W. D. Hauk, C. R. Parks and M. W. Chase, “Phylogenetic Studies of Ophioglossaceae:
Evidence from rbcL and trnL-F Plastid DNA Sequences and Morphology”. Mol. Phyl. Evol.,

255
 vol. 28, pp. 131-151, 2003.
[9] A. M. Paul, “The Status of Ophioglossum azoricum (Ophioglossaceae: Pteridophyta) in the
British Isles”. Fern Gaz., vol. 13, pp. 173-187, 1987.
[10] L. Peruzzi, G. Cesca and D. Puntillo, “Isoëtes (Isoetaceae), Ophioglossum and Botrychium
(Ophioglossaceae) in Calabria (S Italy): More Karyological and Taxonomical Data”. Caryologia,
vol. 56, no. 3, pp. 355-359, 2003.
[11] F. J. Rohlf, TpsDig2, Digitize Landmarks and Outlines, Version 2.12. Dep. of Ecology and
Evolution, State University of New York at Stony Brook, 2007.
[12] F. J. Rohlf, TpsRelw, Version 1.42. Dep. of Ecology and Evolution. State University of New
York, Stony Brook, 2005.
[13] H. Iwata and Y. Ukai, “SHAPE: a Computer Program Package for Quantitative Evaluation of
Biological Shapes Based on Elliptic Fourier Descriptors.”, J. Heredity, vol. 93, pp. 384-385,
2002.

256
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 257-261.
ISBN 978-88-8303-295-0. EUT, 2010.

Geometric morphometric
analysis as a tool to explore
covariation between shape and
other quantitative leaf traits in
European white oaks
Vincenzo Viscosi, Anna Loy, Paola Fortini

Abstract — In this study, 2-Block Partial Least-Squares analysis was used to

explore the pattern of covariation between two sets of leaf morphological data
(landmarks and traditional measures), sampled on 273 trees in a mixed forest
of Quercus frainetto, Q. petraea and Q. pubescens, in southern-central Italy.
Two groups of related variables were detected and the three oak species were
highly significant discriminated by CVA computed on dimensions extracted by
2B-PLS analysis. Q. frainetto was characterized by obovate leaf blade with a
short petiole and leaves were greater than in the other species. Q. petraea
was differentiated by acute basal and apical regions, while Q. pubescens has
higher values of leaf compactness, pubescence and length of trichomes. The
high degree of classification accuracy of this combined approach advocates
its extension to other problematic species and highlights its importance as an
exploratory tool in plant ecology, physiology and taxonomy.

Index Terms — geometric morphometrics, leaf shape, Quercus, Two-Block

Partial Least-Squares.

—————————— u ——————————

1 Introduction

T
he morphological variability in subgenus Quercus has been studied for a
long time. Several analyses, usually based on leaf morphological data,
supported the existence of high variability within and among these oak
species ([1], [2], [3], [4], [5]). Here a landmark based geometric morphometric
approach is used to analyse the leaf morphological pattern of covariation

————————————————
V. Viscosi is with the Museo Erbario del Molise, Department STAT, University of Molise, Pesche
(IS) IT-86090. E-mail: [email protected].
P. Fortini is with the Museo Erbario del Molise, Department STAT, University of Molise, Pesche (IS)
IT-86090. E-mail: [email protected].
A. Loy is with Department STAT, University of Molise, Pesche (IS) IT-86090. E-mail: vincenzo.
[email protected]. E-mail: [email protected].

257
between shape and other traditional traits in three white oak species. We used
Two-Block Partial Least-Squares analysis (2B-PLS) [6] (i) to summarize the
greater part of leaf phenotypic variability in few linear dimensions, (ii) to detect
the degree of differentiation among species, (iii) to obtain a useful classification
method for unidentified specimens.

2 Materials and Methods

2.1 Plant material and data sampling

A total 273 oaks were sampled in a woody community (Italy, Molise) hosting
three hybridizing white oak species. Species assignment followed Schwarz
(1996): Q. petraea (n=122), Q. pubescens (n=57), Q. frainetto (n=56), hybrids
(38). Ten leaves for each tree were photographed for morphological analyses.
Eleven landmarks and 25 traditional morphological characters were collected
on each leaf [5]. Eight morphological characters (area, perimeter, roundness,
elongation, feret diameter, compactness, major axis length, minor axis length)
were automatically recorded on digital images by means of ImageTool ver. 5.1
[7]; five variables were recorded manually by a digital calliper: lamina length
(LL), petiole length (PL), lobe width (LW), sinus width (SW), length of lamina
at largest width (WP), number of lobes (NL) and number of intercalary veins
(NV). Five variables were obtained by ratios between previously measured
variables [1]: lamina shape (OB), petiole ratio (PR), lobe depth ratio (LDR),
percentage venation (PV), lobe width ratio (LWR); four variables were observed
and recorded as ordinal variables ([1], [8],): basal shape (BS), pubescence of
abaxial surface, adaxial surface and petiole. Finally, the length of hairs was
measured [9].

2.2 Statistical analysis

For each tree, the mean of 25 leaf variables and the landmark consensus
configuration was computed. 2B-PLS analysis was used to investigate the
covariation pattern among the 11 shape descriptors and the 25 traditional
variables. Linear combinations of partial warps and other quantitative traits were
visualized as deformation grids through tpsPLS 1.18 [10].
Finally, the linear dimensions, derived from 2B-PLS analysis, were analysed by
Canonical Variate Analysis (CVA) to detect the degree of species differentiation
among species and to obtain a classification rule for unidentified specimens.

3 Results
The first two significant dimensions (2B-PLS) accounted for 95.34% and 2.96%
of total covariance, respectively. The correlations between shape descriptors
and morphological variables for the two first dimensions were 0.884 (p < 0.001)
and 0.780 (p < 0.001), respectively.
The ordination plot showed a clear separation of Q. frainetto, Q. pubescens

258
and Q. petraea (Fig. 1): Q. frainetto is clearly distinguished from both Q. petraea
and Q. pubescens, while the two latter partially overlapped. Moreover, two main
groups of variables that covariated in relation to leaf shape were detected (Fig.
2). Weight along these dimensions in relation to each deformation grid are
shown in Fig. 3.
Linear dimensions derived from 2B-PLS analysis were subjected to CVA (Fig.
4): the two CVs explained 76.8% (Wilks’λ = 0,006; df = 22; p<0.0001) and 23,2%
(Wilks’λ = 0,136; df = 10; p<0.0001) of total variability, respectively. All three
species were highly significant discriminated. The cross validation confirmed
the result and 100% of specimens were correctly classified.

Fig. 1 – Ordination plot for the projections of the pure species onto first two significant
dimensions. Filled circles = Q. frainetto; crosses = Q. petraea; empty squares = Q.
pubescens. Hybrids are not shown.

Fig. 2 – Loadings plot of 25 morphological variables onto the first two significant
dimensions.

259
  
Fig. 3 – Shape variation of leaves along the first two significant dimensions of 2B-PLS.
Deformation grids are shown for the negative (left) and positive (right) extremes of D1
(top) and D2 (bottom). The bar-plot show the degree of covariation between shape and
the 25 traditional characters (see text for codes).

Fig. 4 – Ordination plot of the pure species onto the first two canonical variates. Filled
circles = Q. frainetto; crosses = Q. petraea; empty squares = Q. pubescens. Hybrids
are not shown.

4 Conclusion
This approach results as a useful tool to analyze traditional morphological
variables in relation to leaf shape, one of the most important visual attributes
to characterize white oak species, playing an important role in pattern of
recognition, and in the systematics of this plant subgenus.
2B-PLS results allowed to define a typical leaf shape for each oak species,
and to associate this shape to other morphological traditional features.
Results confirmed the high diagnostic power of the geometric morphometric
approach. Moreover, the visualization of the leaf shape differences, associated

260
to other groups of correlated morphological traits, allowed to obtain a clear
diagnosis of leaf morphology for each species.
Q. frainetto was characterized by grater leaves and short petiole, obovate leaf
blade and high deepness of lobes. Q. petraea was characterized by a longer
petiole, more acute basal and apical regions, and a more deeply lobed lamina
than Q. pubescens, which has higher values of leaf compactness, pubescence
and length of trichomes.
The high degree of classification accuracy of this combined approach advocates
its extension to other problematic species and highlights its importance as an
exploratory tool in plant ecology, physiology and taxonomy.

References
[1] A. Kremer, L. J. Dupouey, J. D. Deans, J. Cottrell, U. Csaikl, R. Finkeldey, S. Espinel, J.
Jensen, J. Kleinschmit, B. Van Dam, A. Ducousso, I. Forrest, U. L. de Heredi, A. J. Lowe, M.
Tutkova, R. C. Munro, S. Steinhoff and V. Badeau, “Leaf morphological differentiation between
Quercus robur and Quercus petraea is stable across western European mixed oak stands”.
Ann. Sci. For., vol. 59, pp. 777-787, 2002.
[2] P. Bruschi, G. Vendramin, F. Bussotti and P. Grossoni, “Morphological and Molecular
Differentation between Quercus petrea (Matt.) Liebl. and Quercus pubescens Willd.
(Fagaceae) in Northern and Central Italy”. Ann. Bot., vol. 85, pp. 325-333, 2000.
[3] S. Ponton, J.-L. Dupouey, E. Dreyer, “Leaf morphology as species indicator in seedlings of
Quercus robur L. and Q. petraea (Matt.) Liebl.: modulation by irradiance and growth flush”.
Ann. For. Sci., vol. 61. pp. 73-80, 2004.
[4] D. Uribe-Salas, C. Sáenz-Romero, A. González-Rodríguez, O. Téllez-Valdéz and K. Oyama,
“Foliar morphological variation in the white oak Quercus rugosa Née (Fagaceae) along a
latitudinal gradient in Mexico: Potential implications for management and conservation”. For.
Ecol. Manag., vol. 256, pp. 2121-2126, 2008.
[5] V. Viscosi, P. Fortini, D. E. Slice, A. Loy and C. Blasi, “Geometric morphometric analyses of
leaf variation in four oak species of subgenus Quercus (Fagaceae)”. Plant Biosyst., vol. 143,
pp. 575-587, 2009.
[6] F. J. Rohlf and M. Corti, “Use of two-block partial least-squares to study covariation in shape”.
Syst Bio., vol. 49, pp. 740–753, 2000.
[7] S. B. Dove, “UTHSCSA ImageTool program 3.0”, 2000.
[8] P. Kissling, “Les poils des quatre espèces de chênes du Jura (Quercus pubescens, Q. petraea,
Q. robur et Q. cerris)”. Ber. Schweiz. Bot. Ges., vol. 87, pp. 1-18, 1977.
[9] P. Bruschi, G. G. Vendramin, F. Bussotti and P. Grossoni, “Morphological and molecular
diversity among Italian populations of Quercus petraea (Fagaceae)”. Ann. Bot., vol. 91, pp.
707-716, 2003.
[10] F. J. Rohlf, “TpsPLS 1.18”. Department of Ecology and Evolution, State University New York.
Stony Brook, 2006.

261
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 263-268.
ISBN 978-88-8303-295-0. EUT, 2010.

Landmark based morphometric

variation in Common dolphin
(Delphinus delphis L.,1758)
Paola Nicolosi, Anna Loy

Abstract — In this study we compare Mediterranean stocks of Delphinus

delphis (L., 1758) with other populations of the same species coming from
different seas using a geometric morphometrics method. The aim is to
define the patterns of geographical variation of Delphinus delphis through a
geometric morphometrics analysis of the skulls of 124 individuals from seven
marine areas (West and East Pacific Ocean; North-east and South-East
Atlantic Ocean, West and East Indian Ocean, Mediterranenan Sea).

Index Terms — common dolphin, geometric morphometrics, skulls.

—————————— u ——————————

1 Introduction

T
he Mediterranean sea has experienced significant changes in the last
decades in terms of biodiversity, due to a combination of environmental
and anthropogenic influences. In this project we focus the attention on
the common dolphin, Delphinus delphis, whose Mediterranean population was
drastically reduced starting from the Sixties and is considered as “Endangered”
from 2003 “[3], [4]”. Analyses where devoted to clarify the pattern of geographic
variation of the species through a geometric morphometric approach, and to
evaluate any specific differentiation/adaptation of the Mediterranean stock
with respect to other populations across the range of the species. Due to
the difficulties related to data collection and records in the field, the museum
collections represented the primary source of information, as in many other
Cetacea“[6]”.

————————————————
P. Nicolosi is with the Museo di Zoologia, Università degli Studi di Padova, 35121 Padova, Italy.
E-mail: [email protected].
A. Loy is with the Dipartimento di Scienze e Tecnologie per l’Ambiente e il Territorio, Università
degli Studi del Molise, Pesche (IS), Italy. E-mail: [email protected].

263
2 Materials and Methods
A total 124 skulls of adult specimens from seven marine areas across the
distribution range of the species (Tab. 1) were photographed on dorsal, ventral
and lateral projections with a digital camera using a standard procedure to avoid
the effects of distortion. Previous analyses on the absence of sexual dimorphism
in the shape of the skull “[9]” allowed to pool males and females.
The analysis of 24 two-dimensional cartesian coordinates (landmarks) have
been recorded on the various projections using the software tpsDig “[10], [11]”
(Fig. 1). Data have been translated, rotated and superimposed through a General
Procrustes Analysis, GPA “[11]” using the tpsRelw software “[10]”. Centroid sizes
were stored for allometric and size variation evaluations. Multivariate ordination
of specimens was performed through Relative Warp Analysis on the weight
matrix of aligned specimens.

Tab. 1 – Number of specimens and their geographical location: 14 from the

Mediterranean sea (Italians Naturalistic and Zoological Museums), 70 from the Atlantic
Ocean (Lisbon Natural History Museum, 26; Zoological Museum of Amsterdam,
25; Zoological Museum of Copenhagen, 19), 33 from the Pacific Ocean (Zoological
Museum of Amsterdam, 31; Zoological Museum of Copenhagen, 2), 7 from the Indian
Ocean (Zoological Museum of Amsterdam, 3; Zoological Museum of Copenhagen, 4.

3 Results
Fig.2 shows the results of ordination analysis of the residual from GPA for
the dorsal view of the skulls, while Fig.3 shows the results of classification
analysis run on the Mediterranean, Atlantic, Indian, and Pacific stocks. The first
two PC (retaining 37,7% and 10,5% of cumulative variation respectively) do
not allow a clear identification of different stocks except for the Indian ocean

264
sample. Nevertheless Mahalanobis distances among groups derived from CVA
scores are highly significant (Tab. 2). Procrustes distances among populations
confirm the Indian stock as the most divergent from all other samples, while
the Mediterranean is the most different with respect the Atlantic and the Pacific
dolphins.

Fig.1 – Landmark recorded on the dorsal projection of the skull.

Atlantic Indian Mediterranean Pacific

Atlantic - 0,0552 0,0211 0,0220

Indian 5,0605*** - 0,0521 0,0440

Mediterranean 2,7085*** 5,2529*** - 0,0247

Pacific 3,0821*** 5,6897*** 3,2302*** -

Tab. 2 – Above diagonal: Procrustes distances among populations; below diagonal:

Mahalanobis distances among groups derived from CVA scores. *** P < 0.0001.

The deformation grid on the left in the graph (Fig.2) is referred to the shape
changes characterizing the Indian Ocean dolphins. The skulls of these
specimens shown an elongation of intermaxilla bones and infraorbita foramina
aligned to the antorbital notch respect the mean.

265
Fig. 2 – Results from PCA run on the residuals from GPA for the dorsal view of the
skull: the deformation grid on the left is referred to the Indian population.

Fig. 3 – Results from the first two canonical axes extracted from residual from GPA.
Symbols refer to the groups are the same for PCA and CVA.

266
4 Conclusion
Geometric morphometrics has shown significant differences in the shape of
the skulls of Delphinus delphis populations from different geographical areas.
Differences are particularly evident in the dolphins from the Indian ocean which
appear the most divergent among all.
Other authors used the morphometric approach to identify shape differences
between populations in the same dolphin species living in different geographical
area “[7], [12], [13]” and also for phylogenetic and evolutionary studies “[2]”.
Many papers also underline the importance of morphometric analysis to support
the genetic, ecological and ethological results as a powerful tool to describe and
understand the mechanism of morphological differentiation “[1], [5], [8]”.
These preliminary results show the need to include the other projections of
the skulls to better elucidate the degree and the pattern of geographic variation,
as well as adaptive traight involved in this pattern and to analyse in depth the
degree and pattern of asimmetry in the region involved in the acousticmotor
complex.

Acknowledgement

The authors wish to thank Dr. Graca Ramalhinho of the Natural History Museum
of Lisbon, Dr. Hans J. Baagøe, Mr. Mogens Andersen and Mrs. Katrine Mohr of the
Zoological Museum of Copenhagen, Dr. Ronald Vonk, Dr. Wendy van Bohemen, Dr.
Roland Sluys of the Zoological Museum of Amsterdam, all the curators of the Italian
Natural History Museums, Dr. Luigi Cagnolaro for information on the Italian cetological
collections and Dr. Andrea Cardini for the Synthesys supporting statement and technical
suggestions. This work was supported in part by a grant from the European Commission’s
Integrated Infrastucture Initiative programme SYNTHESYS.

References
[1] D. C. Adams, F. J. Rohlf and D.E. Slice, “Geometric morphometrics: 10 years of progress
following the ‘revolution’”, Ital. J. Zool., vol. 71, pp. 5-16, 2004.
[2] A. R. Amaral, M. M. Coelho, J. Marugàn-Lobon and F. J. Rohlf, “Cranial Shape Differentiation
in Three Closely Related Delphined Cetacean Species: insights into Evolutionary History”,
Zoology, vol. 112, pp. 38-47, 2009.
[3] G. Bearzi, “Delphinus delphis (Mediterranean subpopulation)”, IUCN Red List of Threatened
Species, http://www.iucnredlist.org, 2009.
[4] G. Bearzi, R. R. Reeves, G. Notarbartolo di Sciara, E. Politi, A. Cañadas, A. Fratzis and B.
Mussi, “Ecology, Status and Conservation of Short-beaked Common Dolphins (Delphinus
delphis) in the Mediterranean Sea”, Mammal Review, vol. 33(34), pp. 224-252, 2003.
[5] A. Cardini, D. Nagorsen, P. O’Higgins, P. D. Polly, R. W. Thorington and P. Tongiorgi, “Detecting
biological uniqueness using geometric morphometrics: an example case from the Vancouver
Island marmot”, Ecology, Ethology and Evolution, DOI 10.1111/j.1439-0469.2008.00503, 2009.
[6] A. Loy, “Morphometrics and Theriology Homage to Marco Corti”, Hystrix It. J. Mamm., vol. 18
(2), pp. 115-136, 2007.
[7] A. Loy, A. Tamburelli, R. Carlini and D. E. Slice, “Craniometric Variation of some Mediterranean
and Atlantic Populations of Stenella caeruleoalba (Mammalia, Delphinidae): a 3D Geometric
Morphometrics Analysis”, Marine Mammal Science, submitted for publication. (pending
publication)
[8] A. Natoli, A. Cañadas, C. Vaquero, E. Politi, P. Fernandez-Navarro and A. R. Hoelzel,
“Conservation Genetics of Short-beaked Common Dolphin (Delphinus delphis) in the

267
 Mediterranean Sea and in the Eastern North Atlantic Ocean”, Conserv. Genet. DOI 10.1007/
s10592-007-9481-1, 2008.
[9] P. Nicolosi, A. Loy, “Morphometric Variation of Mediterranean vs Atlantic Stocks of Common
Dolphin (Delphinus delphis Linnaeus, 1758)”, Paleontologia I Evolució, memòria especial, vol.
3, pp. 97-99, 2009.
[10] F. J. Rohlf, TpsDig, TpsRelw, Department of Ecology and Evolution. State University of New
York at Stony Brook. http://life.bio.sunysb.edu/morph/, 2008.
[11] F. J. Rohlf, D. E. Slice, “Extensions of the Procrustes Method for the Optimal Superimposition
of Landmarks”, Systematic Zool., vol. 39, pp. 40-59, 1990.
[12] A. J. Westagate, “Geographic Variation in Cranial Morphology of Short-beaked Common
Dolphins (Delphinus delphis) from the North Atlantic”, Journal of Mammology, vol. 88(3), pp.
678-688, 2007.
[13] K. A. Viaud-Martinez, L. R.Jr. Brownell, A. Komnenou and A. J. Bohonak, “Genetic Isolation
and Morphological Divergence of Black Sea Bottlenose Dolphins”, Biological Conservation,
vol. 141, pp. 1600–1611, 2008.

268
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 269-273.
ISBN 978-88-8303-295-0. EUT, 2010.

DNA barcoding: theoretical

aspects and practical
applications
Maurizio Casiraghi, Massimo Labra, Emanuele Ferri,
Andrea Galimberti, Fabrizio De Mattia

Abstract — DNA barcoding is a molecular-based identification system,

recently introduced in the scientific community. The method is not completely
new to science, but the real innovation is not in the discrimination system itself:
DNA barcoding can be considered as the core of an integrated taxonomic
system, where bioinformatics plays a key role. Time is now ripe for a real
collaboration of all the different forces working in taxonomy, towards a “next
generation systematics”.

Index Terms — DNA barcoding, DNA taxonomy, molecular identification,

species identification.

—————————— u ——————————

1 Introduction

T
he classification and monitoring of biodiversity are playing a key role in
different contexts (e.g.: biological, social, economical), even if several
aspects linked to these topics are far to be completely understood. A
common assumption is that the central unit of taxonomy is the species, and the
unequivocal association of a scientific name to a biological entity is an essential
step to build a reliable reference system of biological information [1].
In the last 250 years, since Carl Linnaeus’ classification system, about 1.7
million species have been formally described by taxonomists, but it is largely
accepted that this number probably represents only a small fraction of the
real biodiversity present on the planet (presently estimated in tens of millions
of species) [2]. To help discovering this hidden biodiversity and in order to
provide a useful and standardized tool for species identification, a molecular
and bioinformatical tool called DNA barcoding has been proposed in 2003 [3].
The basic idea of this approach is quite simple (and not completely new

————————————————
The authors are with the Department of Biotechnologies and Biosciences, University of Milan-
Bicocca, Milan, Italy. E-mail:[email protected], [email protected].

269
to science): through the analysis of the variability in a single or in a few
standard molecular marker(s), it is possible to discriminate biological entities
(hopefully belonging to the taxonomic rank of species). This method relies
on the assumption that the genetic variation between species exceeds that
within species. Consequently, the ideal DNA barcoding analysis mirrors the
distributions of intra- and interspecific variabilities separated by a distance called
‘DNA barcoding gap’ [4], [5]. The original idea was to apply DNA barcoding
systematically to all metazoans, by the use of one or a few (mitochondrial)
markers (e.g. coxI, [1]). Rapidly, but with less coherent results, the idea was
extended to flowering plants [6], [7] and fungi [8], and now the DNA barcoding
initiative can be considered as a tool suitable for all of the tree kingdoms.
Efforts in DNA barcoding development and management are coordinated by the
Consortium for the Barcode of Life (CBoL; http://barcoding.si.edu/).
One of the major properties of a DNA barcode is the possibility to easily
associate all life history stages and genders, to identify organisms from part/
pieces, or to discriminate a matrix containing a mixture of biological species.
Quite soon it became clear that DNA barcoding was suitable for two different
purposes: (1) the molecular identification of already described species [1], and
(2) the discovery of undescribed species [9].
A lot of rumours raised around this approach, but what is the revolution
introduced by DNA barcoding? In our opinion, the big leap forward is not only
the discrimination power itself, but the joint use of three innovations of modern
taxonomy: (1) molecularization (i.e. the use of variability in a molecular marker
as a discriminator); (2) computerization (i.e. the non redundant transposition of
data using informatic supports) and (3) standardization (i.e. the extension of the
approach to vast groups of not deeply related organisms). For the first time, by
DNA barcoding, it is possible to introduce in taxonomy a generalization, allowing
researchers specialized in different fields to work on a shared framework.
In the space of few years, DNA barcoding has moved from fantasy to reality. In
some of the first enthusiastic reports, DNA barcoding was even claimed as the
way to make true the dreams of Gene Roddenberry, the creator of the science
fiction drama Star Trek: the creation of a tool for organism identification, the DNA
barcoder, as a homologous to the fictional Tricorder [10]. A few years later we
are not yet in the spaceship Enterprise, but DNA barcoding has deeply impacted
the scientific community, becoming a widely used approach.
Presently, the most relevant DNA barcoding tool, The Barcode of Life Data
Systems, BOLD (http://www.barcodinglife.org/, [11]) is still in constant evolution
and update.

2 The ‘DNA barcoding molecular entity’ versus ‘species’ debate

Most of the questions raised by the use of DNA barcoding are directly linked
to the essence of an identification method. In a strict sense, to identify simply
means to differentiate, but the choice of the discriminator is essential, because
the difficulty is in giving a biological meaning to what it has been discriminated
[12].
Even if not always fully acknowledged, DNA barcoding implies two different

270
approaches to discrimination. DNA barcoding sensu stricto is a simple sorting
method that could differentiate biological entities. It is not significantly different
from a dichotomic key in the traditional taxonomical framework. On the other
hand, DNA barcoding sensu lato represents a system that reflects the true
sense of taxonomy. The discrimination method itself can be considered as an
epiphenomenon - and the subject of major criticisms (DNA barcoding sensu
stricto) - but it also becomes a system implementing all the aspects of taxonomy
towards the representation of the living world as a whole (DNA barcoding sensu
lato). It should be clear to users which kind of DNA barcoding philosophy they
are going to adopt.

3 The ‘DNA barcoding molecular entity’ versus ‘species’ debate

It is well known that no identification method (morphological, biochemical,
genetic, etc.) can truly identify species, because species are entities in
continuous evolution and it is theoretically impossible to define statically such
a dynamic matter. DNA barcoding, in its original generalization, follows the
typological species approach, a concept that theoretically fails because it freezes
the evolutionary continuum of species. To cope with these limitations, some
development of DNA barcoding has shifted towards other species concepts [13].
The entities identified by molecular approaches have been named in several
ways: ‘Genospecies’; ‘Phylospecies’, ‘Recognizable Taxonomic Units’, RTUs,
‘Phylotypes’ sensu, ‘Molecular Operational Taxonomic Units’, ‘MOTUs’ [12].
A general naïve assumption considers ‘molecular entities’ and ‘species’ as
synonyms. This is the (almost) insurmountable problem for DNA barcoding
sensu stricto: the biological meaning of the identified ranks cannot be directly
derived, unless we have clearly and unequivocally linked a species to the
variability pattern of a single DNA barcoding marker. In all other cases, we need
DNA barcoding sensu lato [12].
The identification and then the interpretation of molecular entities is the
main goal of DNA barcoding. This can be reached only by users with a sound
theoretical background on what this technique is able to identify.

4 The choice of the Barcode marker

DNA barcoding is not coxI only. A precise portion in the 5’ end region of this
mitochondrial gene has been proposed as a standard for metazoans [1]. Even
if coxI has proven to be useful to discriminate species in most of the tested
groups, its limits in some animal taxa are already evident (e.g. [14]). The choice
of regions usable for DNA barcoding has been little investigated in many other
eukaryotes. For instance, a marker was already available in fungi: the nuclear
ITS region, which has been now confirmed as the main DNA barcode for
this group [15]. In terrestrial plants, compared to animals, mitochondrial DNA
has slower substitution rates and shows intramolecular recombination [16].
The search for an analogous to coxI or ITS in plants, that matches with the
DNA barcoding criteria, has focused attention on the plastid genome. Several
plastid genes have been proposed, such as the most conserved rpoB, rpoC1

271
and rbcL or a section of matK showing a rapid rate of evolution, but in some
plant families these genes showed amplification problems. At the same time,
the intergenic spacers such as trnH-psbA, atpF-atpH and psbK-psbI were also
tested for their rapid evolution [17. Recently, the CBoL Plant Working Group [18]
provided a recommendation on a standard plant barcode suggesting the 2-locus
combination of rbcL and matK.

5 Biological and bioinformatical Repositories for DNA barcoding

data

DNA barcoding data are meant to be easily and widely accessed. To reach
this aim, a proper sequence submission procedure is available for GenBank
(http://www.ncbi.nlm.nih.gov/WebSub/?tool=barcode). This procedure
slightly modifies the standard sequence submission procedure, introducing a
DNA barcoding label to the sequence in order to simplify database querying
and searching. Moreover, additional data are requested to link barcode
sequence data to its voucher specimen. This standardization is mirrored by
the establishment of the Registry of Biological Repositories initiative (http://
www.biorepositories.org/), an on-line registry of organisms linked to DNA
sequences. DNA barcoding sequences can also be deposited as projects in
BOLD databases, characterized by an automatic submission tool to publish
sequences to GenBank. By December 2009, BOLD database encompassed
more than 760,000 sequences, corresponding to more than 65,300 formally
described ‘species’. The amount of data managed by the BOLD database is
impressive: it collects, for a large amount of deposited barcode sequences,
specimen details such as morphology, photographs, geographical distribution,
collection points and others [11].

6 Conclusion
DNA barcoding is not a “perfect” method, but it has deeply impacted the
scientific community, becoming a widely used approach, characterized by many
relevant aspects of uniformity and generalization. A critical knowledge of the
method is essential for a proper use of it.

Acknowledgement

The authors are grateful to the ZooPlantLab staff, students and supporters.

References
[1] Q. D. Wheeler, “Taxonomic triage and the poverty of phylogeny”. Phil. Trans. R. Soc. Lond. B,
vol. 359, pp. 571-583, 2004.
[2] R. Vernooy, E. Haribabu, M. R. Muller, J. H. Vogel, P. D. N. Hebert, et al., “Barcoding Life
to Conserve Biological Diversity: Beyond the Taxonomic Imperative”. PLoS Biol., vol. 8(7):
e1000417. doi:10.1371/journal.pbio.1000417, 2010.
[3] P. D. N. Hebert, A. Cywinska, S. L. Ball, et al., “Biological identifications through DNA barcodes”.
Proc. R. Soc. London, Biol. Sci. Series B, vol. 270, pp. 313–321, 2003.

272
[4] C. P. Meyer and G. Paulay, “DNA Barcoding: error rates based on comprehensive sampling”.
PLoS Biol., vol. 3, e422, 2005.
[5] M. Wiemers and K. Fiedler, “Does the DNA barcoding gap exist? – a case study in blue
butterflies (Lepidoptera: Lycaenidae)”. Frontiers in Zoology, vol. 4, p. 8, 2005.
[6] W. J. Kress, K. J. Wurdack K., E. A. Zimmer, et al., “Use of DNA barcodes to identify flowering
plants”. PNAS, vol. 102, pp. 8369-8374, 2005.
[7] M. L. Hollingsworth, A. Clark, L. L. Forrest, et al., “Selecting barcoding loci for plants: evaluation
of seven candidate loci with species level sampling in three divergent groups of land plants”.
Mol. Ecol. Res., vol. 9, pp. 439-457, 2009.
[8] X. J. Min and D. A. Hickey, “Assessing the effect of varying sequence length on DNA barcoding
of fungi”. Mol. Ecol. Notes., vol. 7, pp. 365–373, 2007.
[9] P. D. N. Hebert, E. H. Penton, J. M. Burns, et al., “Ten species in one: DNA barcoding reveals
cryptic species in the neotropical skipper butterfly Astraptes fulgerator.” PNAS, vol. 101, pp.
14812-14817, 2004.
[10] K. J. Gaston and M. A. O’Neill, “Automated species identification: why not?” Phil. Trans. R.
Soc. Lond. B, vol. 359, pp. 655-667, 2004.
[11] S. Ratnasingham and P. D. N. Hebert, “BOLD: The Barcode of Life Datasystem (www.
barcodinglife.org)”. Mol. Ecol. Notes, vol. 7, pp. 355-364, 2007.
[12] M. Casiraghi, M. Labra, E. Ferri, A. Galimberti and F. De Mattia, “DNA barcoding: a six-question
tour to improve users’ awareness about the method.” Brief Bioinform., vol. 11(4), pp. 440-453.
Epub 2010 Feb 15, 2010.
[13] J. M. Padial, A. Miralles, I. De la Riva and M. Vences, “The integrative future of taxonomy.”
Front Zool., vol. 7, p. 16, 2010.
[14] T. L. Shearer and M. A. Coffroth, “Barcoding corals: limited by interspecific divergence, not
intraspecific variation.” Mol. Ecol. Resour., vol. 8, pp. 247-255, 2008.
[15] R. H. Nilsson, M. Ryberg, E. Kristiansson, et al., “Taxonomic Reliability of DNA Sequences in
Public Sequence Databases: A Fungal Perspective.” PLoS One, vol. 1, e59, 2006.
[16] G. D. D. Hurst and F. M. Jiggins, “Problems with mitochondrial DNA as a marker in population,
phylogeographic and phylogenetic studies: the effects of inherited symbionts.” P. Roy. Soc.
Lond. B Bio, vol. 272, pp. 1525-1534, 2005.
[17] A. J. Fazekas, K.S. Burgess, P. R. Kesanakurti, et al. “Multiple multilocus DNA barcodes from
the plastid genome discriminate plant species equally well.” PLoS One, vol. 3, e2802, 2008.
[18] CBoL Plant Working Group. “A DNA barcode for land plants.” PNAS, vol.106, pp. 12794-
12797, 2009.

273
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 275-280.
ISBN 978-88-8303-295-0. EUT, 2010.

Strength and Limitations

of DNA Barcode under the
Multidimensional Species
Perspective
Valerio Sbordoni

Abstract — DNA barcoding aims at providing an efficient method for species-

level separation using a partial sequence of the mitochondrial COI gene. The
efficiency of the barcode in separating species is based on the amount of
genetic distance among samples. While in many taxa the species can be
efficiently identified through the barcode, other situations cannot be treated by
this approach. The causes for such discrepancy appear to be mostly related to
the nature of speciation events and to the different roles of the genetic system,
natural selection and evolutionary time. Thus, DNA barcode represent just
one important descriptor in the framework of the multidimensional species
approach.

Index Terms — DNA barcode, molecular systematics, species concepts,

taxonomic procedures.

—————————— u ——————————

1 Introduction

S
ince 2003, many research groups started to accumulate molecular data with
the aim of setting up a sort of inventory of life that might itemize biodiversity
as a sequence of species-specific DNA. In particular, Paul Hebert from
the Canadian University of Guelph [1], [2],proposed to use a sequence of the
COI mitochondrial gene, codifying for cytochrome oxidase 1,as a “molecular
signature” to identify a species. The selection of this gene for exploring limits
between species allows for the practical advantages of using mitochondrial
DNA together with the previous wide use of this gene in a large variety of
organisms. COI sequences are also currently used at different taxonomic levels,
in phylogeny, phylogeography and population genetics studies, due to the great
advantage offered by the availability of amplification protocols, as well as a large
number of sequences ready for barcoding.
At present there are many ongoing DNA barcoding projects reported on the
————————————————
V. Sbordoni is with the Department of Biology, University of Roma Tor Vergata, Italy.
E-mail: [email protected].

275
website of the Barcode of Life Data Systems (www.boldsystems.org), an online
workbench that supports collection, management, analysis, and use of DNA
barcodes. An enormous bulk of barcode data for a wide array of organisms has
already been made available to the scientific community. Thus, a sequence of
the mitochondrial COI gene has become the most used mitochondrial marker,
especially for animals. The same marker, preferably associated with nuclear
DNA sequences, is commonly used also in more wide phylogenetic studies.
The choice of a marker specific for plants and fungi is more problematic, but it
currently seems oriented, at least for angiosperms, to the trnH-psbA sequence,
an intergenic spacer of plastidic DNA [3], [4].

2 Molecular Systematics and the DNA Barcode

By identifying genetic differentiation threshold values that would include
individuals of the same species, the barcode approach allows investigating
and analysing in some detail inter-species delimitation and its many related
problems. A great advantage offered by DNA barcoding is the possibility of
identifying cryptic species, that is to distinguish as belonging to different species
individuals that, due to their similar morphology, were considered as belonging
to a unique species. This would be feasible by identifying genetic distance limit
values within which two individuals can be considered as belonging to the same
species, while outside these limits they should be considered as belonging to
different species. Although the limit seems to be somehow taxon dependent, it
has been observed that a value of genetic distance between two DNA-barcode
sequences equal to or higher than three per cent (D≥0.03) identifies distinct
species.
However, it has become clear that the currently used genetic distance
approaches by means of DNA barcodes have strong limitations, particularly
when it comes to defining species boundaries [5]. One reason is that mtDNA
rates of evolution vary substantially between and within species and between
different groups of species, thus resulting in broad overlaps of intra- and
interspecific distances [6], [7], [8]. But there are other reasons worth to be briefly
discussed.
To what extent DNA barcode may solve the problem of identification of
interspecific limits? And, will this approach be applicable to the different
typologies of species? Based on the data produced up to day, it seems that
DNA barcoding may solve problems of specific classification in a wide range of
organisms and situations, but many taxa cannot be treated by this approach [9].
Let’s try to understand why. Biologists and evolutionists know that species are
the result of historical processes, mainly speciation and range dynamics. The
knowledge of these processes is of help for inferring the essence of species
whose properties reflect the “signature” of speciation where the following factors
have more or less predominant roles:
1. The genetic system.
2. Natural selection.
3. Evolutionary time.
The genetic system, i.e. the organization of genes and of other chromosomal

276
genetic material may favour, by acceleration, the speciation process especially
when it is based on Robertsonian translocations, or inversion polymorphisms
or, as with the extreme case of instantaneous speciation, on allopolyploid
mechanisms, so frequent in plants.
Natural selection may have a determinant role in those sympatric speciation
phenomena connected to the shift of the trophic niche, that are frequent
especially in phytophagous insects. In these cases selective pressure may
significantly accelerate the adaptive diversification process of genotypes in their
new evolutionary path.
Finally, in the absence of diverging selective pressure, even the simple
allopatric condition of two originally conspecific populations may lead to
speciation in the long run if the genic flow barrier persists. This is where the
evolutionary time finds its role. The longer the isolation time, the higher the
number of different nucleotidic substitutions in the DNA sequences of the two
genic pool involved. Accumulation of diverging mutations in the whole genome
will inevitably lead to speciation by genetic drift, and the speed of this process
will be inversely proportional to the population effective size.
This brief analysis of times and modes of speciation is of help for interpreting
the biological meaning of genetic distance estimates based on the DNA barcode
concept. A limit of 3% genetic divergence is relatively well working to separate
species that are the result of geographic speciation events driven by gradual
accumulation of diverging mutations. Emblematic examples are diverging
populations and species of animals adapted to cave life [10], with special regard
to the thoroughly studied Dolichopodacave crickets [11], [12], [13].
On the other hand, much smaller values, up to a ten factor, do not allow the
discrimination by barcoding of species differing by one or more chromosomal
inversions, as it is the case with some Anopheles [14], or of recent species
originated sympatrically by adaptive shift, as it has been reported in fruit flies
of the genus Rhagoletis [15]. Most probably, in the case of Dolichopoda, the
divergence process may have involved most of the genome and, consequently,
mitochondrial genes as well. Conversely, in the other two examples, the
divergence should have involved, in relatively short times, only a few genes
target of selection, or particular combinations of these genes, showing no effects
in mitochondrial DNA sequences. Indeed, mitochondrial DNA is often used as a
molecular clock [16], [17].

3 The Multidimensional Approach

The debate on species’ properties shows many different aspects: a more
classical one opposing the “biological concept of species, BSC” to the “typological,
TSC” one [18], and a more modern one that seeks a conflict between BSC and
the “phylogenetic concept” [19].
But when analysing the nature of these contrasts it must be agreed that
the three concepts focus on properties of the species that do not contrast
and which are coherent with a vision of the species as the in itinere product
of the evolutionary process. Therefore it should be agreed that a species,
beside representing “the smallest monophyletic group of common ancestry”

277
[20], expresses its distinct gene pool (BSC) by having eventually acquired the
distinctive characters emphasized in the typological species concept.
Actually, in routine work, many taxonomists tend to use operational approaches
toward species, based on morphological characters that are unique and shared
or their hierarchies. Although this is not explicitly declared, these species are
based on philosophies ranging from the classic typological concept up to the
phylogenetic and the phenetic ones. Traditional phenetic definitions of species
are based essentially on the numeric recognition of intervals separating clusters
of phenetically similar individuals [21], [22]. For this evaluation many different,
“not weighed” types of taxonomic characters are examined, but their biological
meaning is not evaluated.
The phenetic “concept” has been strongly criticized because it was not
considered as sufficient to describe the complex interrelations existing
among clusters of similar populations. Nonetheless, this kind of approach has
operational advantages of some practical value and consequently it is widely
adopted and used by systematists.
Being an intrinsically complex entity, the species requires a multidimensional
approach taking into account the whole set of taxonomic characters. Nowadays,
the huge progress in multivariate analysis, together with the wide choice of
technologies to measure parameters related to the ecology of niche, or to sexual
behaviour, and/or to many other crucial features of species-specificity, have
richly endowed the kit of characters available to the modern taxonomist. The
technical and conceptual progress in the field of molecular biology has made
relatively easy and rapid both the acquisition and the phylogenetic interpretation
of sequence data containing an enormous quantity of information. The routine
use of these characters has substantially empowered the understanding of the
species’ genetic structure and has brought to the discovery of the existence of
cryptic species.
However, as previously discussed, DNA sequences are not the only
depositories of evolutionary history. Any other kind of character is potentially
suitable to give its own contribution to the species definition. For instance,
much of the ecological role (niche) of an organism, is written in its morphology,
although there is no assurance that the ecological divergence between two
similar species corresponds to a genetic gap, even though this coincidence
shows up in the majority of cases. The diagnostic value of each character varies
by taxon or specific evolutionary, geographical and ecological situations. For
most taxa specialists know very well which characters are more kin to represent
the species’ biological properties.
This discussion makes us re-consider, although with due caution and
adjustments, the usefulness of the phenetic approach. Since different
descriptors have different values for a given organism, one cannot rely on
automatic discriminating procedures, while one can rely on the availability of the
whole set of algorithms and multivariate procedures set up in the systematic and
ecological fields. Yet, the responsibility of the final decision will inevitably have
to rely on the competence and experience of the specialist. It must be stressed
that different taxonomic characters do not necessarily vary in a coordinated
way, yet they are often conflicting. Both evaluation and weighing of characters

278
have always had and continue to have a fundamental role in the process of
species’ delimitation [11].
Based on this logic and premises, species are considered as “clusters of
individuals that are effectively separated from other clusters in the space defined
by their descriptors” [23]. Alike the preceding phenetic definitions, species are
seen as clouds of probability in an hyperspace. Here though characters are
weighed and a value is assigned to genetic, and inter-reproductive descriptors,
i.e. exactly those characterising the species as a monophyletic cluster, as a
cluster of genotypes and as a cluster of individuals sharing a special relation
with their environment. An “ad hoc” reduction of this hyperspace makes this
multidimensional concept operative. For instance, the typical biological species
becomes a particular case where intra-population genetic and reproductive
relationships are quantified and analysed as a sub-set of a wider set of descriptors.
Yet, the use of the multidimensional approach should be particularly useful for
organisms with asexual or uniparental reproduction, including bacteria, protists,
fungi, rotifera and many parthenogenetic taxa, to which it is traditionally difficult
or impossible to apply the biological concept of species, thus overcoming the
tie of amphigonic reproduction and allowing, not only in theory, the evaluation
of clusters defined by appropriate descriptors. The literature on taxonomy, and
not only the recent one, offers many examples of this approach, adopted with
success in cave crickets [11], butterflies [24], fishes [25], fossil Ostracoda [26],
Rotifera[27], etc.
Many species’ definitions, privileging either properties, can be accommodated
within this approach, but I want here to recall in particular a somewhat unknown
definition by Alfred Russell Wallace, incidentally quoted in one of his writings
where he disputes with Galton: “A species … is a group of living organisms,
separated from all other such groups by a set of distinctive characteristics,
having relations to the environment not identical with those of any other group
of organisms, and having the power of continuously reproducing its like” [28].
This definition, dated many years before the Synthetic Theory, refers to all
the emerging properties of species: a set of distinctive characters (highlighted
by the typological concept), the relationship with the environment (ecological
concept), and finally the power of reproducing its own characteristics, which
implies the properties of the hereditary material. Compared with many others,
Wallace’s definition has certainly the merit of stressing multidimensionality,
a concept that expresses the best operational solution to the problem of the
delimitation of species.

References
[1] P. D. N. Hebert, A. Cywinska, S. L. Ball and J. R. DeWaard, “Biological Identifications Through
DNA Barcodes”, Proc. R. Soc. Lond. B., vol. 270, pp. 313-321, 2003.
[2] S. Ratnasingham and P. D. N. Hebert, “Barcoding Bold: The Barcode of Life Data System”,
Molecular Ecology Notes, doi: 10.1111/j.1471-8286.2006.01678.x, 2007.
[3] W. J. Kress, K. J. Wurdack, E. A. Zimmer, L. A. Weigt and D. H. Janzen, “Use of DNA Barcodes
to Identify Flowering Plants”, PNAS, vol. 102, pp. 8369-8374, 2005.
[4] M. W. Chase and M. F. Fay, “Barcoding of Plants and Fungi”, Science, vol. 325, pp. 682–683,
2009.

279
[5] J. D. Witt, D. L. Threloff and P. D. Hebert, “DNA Barcoding Reveals Extraordinary Cryptic
Diversity in an Amphipod Genus: Implications for Desert Spring Conservation”, Mol. Ecol., vol.
15, pp. 3073–3082, 2006.
[6] W. W. Kipling and D. Rubinoff, “Myth of the Molecule: DNA Barcodes for Species Cannot
Replace Morphology for Identification and Classification”, Cladistics, vol. 20, pp. 47–55, 2004.
[7] D. Rubinoff, “Utility of Mitochondrial DNA Barcodes in Species Conservation”, Conserv. Biol.,
vol. 20, pp. 1026–1033, 2006.
[8] D. Rubinoff, S. Cameron and K. Will, “A Genomic Perspective on the Shortcomings of
Mitochondrial DNA for “Barcoding” Identification”, J. Hered., vol. 97, pp. 581–594, 2006.
[9] J. Waugh, “DNA Barcoding in Animal Species: Progress, Potential and Pitfalls”, BioEssays, vol.
29, pp. 188–197, 2007.
[10] V. Sbordoni, “Advances in Speciation of Cave Animals”. In: C.Barigozzi (ed.), Mechanisms of
Speciation, New York, A.R. Liss Inc., pp. 219-240, 1982.
[11] V. Sbordoni, G. Allegrucci and D. Cesaroni, “A Multidimensional Approach to the Evolution and
Systematics of Dolichopoda Cave Crickets”. In: G. M. Hewitt et al. (eds.), Molecular Techniques
in Taxonomy, NATO ASI Series, vol. H57, Berlin, Springer, pp. 171-199, 1991.
[12] G. Allegrucci, M. Rampini, P. Gratton, V. Todisco and V. Sbordoni, “Testing Phylogenetic
Hypotheses for Reconstructing the Evolutionary History of Dolichopoda Cave Crickets in the
Eastern Mediterranean”, J. Biogeography, vol. 36, pp. 1785-1797, 2009.
[13] L. Martinsen, F. Venanzetti and A. Johnsen, V. Sbordoni and L. Bachmann “Molecular Evolution
of the pDo500 Satellite DNA Family in Dolichopoda Cave Crickets (Rhaphidophoridae)”, BMC
Evolutionary Biology, vol. 9, 301 (14 pp.), 2009.
[14] M. Coluzzi, A. Sabatini, A. della Torre, M.A. Di Deco and V. Petrarca, “A Polytene Chromosome
Analysis of the Anopheles gambiae Species Complex”, Science, vol. 298, pp. 1415–1418,
2002.
[15] J. J Smith and G.L. Bush, “Phylogeny of the Genus Rhagoletis (Diptera: Tephritidae) Ifrom
DNA Sequences of Mitochondrial Cytochrome Oxidase II”, Mol. Phyl. Evol., pp. 33-43, 1997.
[16] L. Bromha and D. Penny, “The Modern Molecular Clock”, Nature Rev. Genet., vol. 4, pp. 216–
224, 2003.
[17] A. Caccon and V. Sbordoni, “Molecular Biogeography of Cave Life: a Study Using Mitochondrial
DNA from Bathysciine Beetles”, Evolution, vol. 55, pp. 122–130, 2001.
[18] E. Mayr, Systematics and the Origin of Species, from the Viewpoint of a Zoologist, New York,
Columbia University Press, 1942.
[19] M. J. Donoghue, “A Critique of the Biological Species Concept and Recommendations for a
Phylogenetic Alternative”, Bryologist, vol. 88, pp. 172-181, 1985.
[20] K. de Queiro and M. J. Donoghue, “Phylogenetic Systematics and Species Revisited”, Cladistic,
vol. 6, pp. 83-90, 1990.
[21] C. D. Michener, “Diverse Aapproaches to Systematics”, Evol. Biol., vol. 4, pp. 1-38, 1970.
[22] E. H. A. Sneat and R. R. Sokal, Numerical Taxonomy, W. H. Freeman, San Francisco, 1973.
[23] V. Sbordoni, “Molecular Systematics and the Multidimensional Concept of Species”,
Biochemical Systematics and Ecology, vol. 21, pp. 39-42, 1993.
[24] D. Cesaroni, M. Lucarelli, P. Allori, F. Russ and V. Sbordoni, “Patterns of Evolution and
Multidimensional Systematics in Graylings (Lepidoptera: Hipparchia)”, Biol.J.Linn.Soc., vol.
52, pp. 101-119, 1994.
[25] M. Barluenga, K. N. Stolting, W. Salzburger, M. Muschic and A. Meyer, “Sympatric Speciation
in Nicaraguan Crater Lake Cichlid Fish”, Nature, vol. 439, pp. 719-23, 2006.
[26] M. Gross, K. Minati, D.L. Danielopo and W. E. Piller, “Environmental Changes and
Diversification of Cyprideis in the Late Miocene of the Styrian Basin (Lake Pannon, Austria)”,
Palaeobiodiversity and Palaeoenvironments, vol. 88, pp. 161-181, 2008.
[27] D. Fontaneto, E. A. Herniou, C. Boschetti, M. Caprioli, G. Melone, C. Ricc and T. G. Barraclough,
“Independently Evolving Species in Asexual Bdelloid Rotifers”, Plos Biology, doi: 10.1371/
journal.pbio.0050087, 2007.
[28] A. R. Wallace, “The Method for Organic Evolution”, Fortnightly Review (N.S.), vol. 57, pp. 435-
445, London, 1895.

280
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 281-287.
ISBN 978-88-8303-295-0. EUT, 2010.

DNA Barcoding and Phylogeny

of Patellids from Asturias
(Northern Spain)
Yaisel Juan Borrell, Fernando Romano, Emilia Vázquez,
Gloria Blanco, Jose Antonio Sánchez Prado

Abstract — The main role for genetics in marine invertebrates is the

identification of species and groups of interbreeding individuals. In Asturias
exists an ancient culinary tradition of consumption for limpets (Patella s.s.)
but there is a lack of studies about these species. We have sampled Asturian
marine Patella s.s. specimens and conducted sequencing of the mtDNA COI
gene. We have confirmed the presence of four Patella s.s. in the Asturian
coasts (P. vulgata, P. depressa, P.aspera, and P. rustica). Our work raises
concerns about the current state of the P. vulgata populations in Asturias,
where it is exploited, due to its low levels of genetic variation. Our phylogenetic
analyses using Bayesian approaches confirmed that patellids belong to four
different clades but gives a new version about how these clades are related
inside the genus aiming for the necessity of more work to address this issue.

Index Terms — Patellids, COI gene, Bayes phylogenies, Asturias.

—————————— u ——————————

1 Introduction

T
he family Patellidae contains most of the common limpets on the temperate
rocky shores of Europe. It contains around 37 species distributed in four
morphological clades: Helcion, Cymbula, Scutellastra and Patella [1]. The
phylogeny and taxonomy of this family has been revised and modified many
times and it is not completely resolved yet [1], [2], [3]. Several morphological
characters have been used to differentiate these species (e.g. radular
morphology, headfoot, sperm, etc); however, the external form of the shell had
been the principal character used in the species-level taxonomy of patellids [1],
[4]. Despite this, it is known that shells are highly variable and usually lead to
taxonomy confusions (see Mauro et al. [4]).
The main role for genetics in marine invertebrates is the identification of species
and groups of interbreeding individuals [5]. A public library of sequences linked
to named species, and the promotion of portable devices for DNA barcoding, will
————————————————
All authors are with the Laboratorio de Genética Acuícola, Departamento de Biología Funcional,
IUBA, Universidad de Oviedo, 33071 Oviedo, Spain. E-mail: [email protected].

281
also considerably help in the management and conservation of marine species
[6]. The cytocrome c oxidase I gene (COI) seems to be a useful genetic tool to
be used in patellids molecular genetics. The universal primers for this gene are
very robust and COI appears to possess a greater range of phylogenetic signal
than any other mitochondrial gene [7], [8]. Kouffopanou et al. [2] used 12S and
16S genes for constructing the first molecular phylogeny of patellids. More
recently Sá-Pinto et al. [3] used also COI and pointed to Patella s.s. showing
five strongly supported clades (I: P. candei, P. lugubris, P. caerulea, P. depressa;
II: P. ulyssiponensis (P. aspera); III: P. vulgata. IV: P. rustica, P. ferruginea; V: P.
pellucida).
The Patella s.s. show problems for their conservation and management
today. Conservation of declining stocks of P. candei and P. aspera has become
a concern on the Atlantic islands [9], while in the Mediterranean, P. ferruginea is
seriously endangered [10] and in the eastern Pacific P. mexicana may be locally
extinct on parts of the mainland coast of Mexico [1]. In Asturias, northern Spain,
exists an ancient culinary tradition of consumption for limpets (Patella s.s.).
Even when they have been under commercial exploitation for decades, there
is not any previous genetic data about them. There is only a few data about the
composition of the genus and the more recent morphological studies dates from
1980 [11], [12], [13]. It had been reported four species (P. vulgata, P. depressa,
P. aspera and P. rustica) of which the principally harvested is P. vulgata. We have
sampled Asturian marine Patella s.s. specimens and conducted sequencing of
the mtDNA COI gene. Our aims were to identify/confirm the patellids species
present in our coasts using molecular methods and to analyze the phylogenetic
status of the genus using a Bayesian approach.

2 Material and methods

Fourty five Patella s.s. specimens were collected in four areas (western
Asturias: Punta La Cruz (10 inds), central Asturias: Moniello (8 inds), Antromero
(18 inds) and eastern Asturias: Tereñes (9 inds)) (Fig. 1). Patellid individuals
were morfologically identified by an experienced zoologist. DNA extractions
were carried out using the ZR Genomic DNA KitTM. We used the universal
primers described by Folmer et al. [7] for the amplification of a 658 bp fragment
of the Citochrome Oxidase subunit I gene (COI). Sequences were aligned in
a multiple alignment with the BioEdit software package. After alignment and
trimming, the final sequence length used was 642 bp. We used Collapse 1.2
for collapsing sequences to haplotypes. A Global search in the Barcoding of
Life system (CBOL) resource for the identification of species using the COI
sequences was carried out. Nucleotide (π) and haplotype (h) diversities were
calculated using Arlequin 3.11. For Bayesian phylogenetic inferences MrBayes
3.1.2 was used evaluating tree topologies and models of nucleotide substitution
for the aligned dataset. The model of nucleotide evolution was the general time-
reversible model (GTR+G) with gamma-distributed rate variation across sites
but not proportion of invariable sites.

282
 Fig. 1 – Map showing the geographic localization for patellids samples in Asturian
coastal area (northern Spain) 1- Punta la Cruz. 2- Moniello. 3- Antromero 4- Tereñes.
Limpet individuals assigned to each species of the Patella genus after COI identification
are indicated by areas in the same order.

3 Results
The COI sequence data have been deposited in the GenBank nucleotide
sequence database with accession numbers EF462952 to EF462975 (23
patellids haplotypes). The 45 limpets (Patella s.s.) collected in the Asturian
coasts were morphologically assigned to the species P. vulgata (20), P. depressa
(5), P. aspera (15) and P. rustica (5). This was concordant with assignments
using the molecular method and the CBOL resource. However, two individuals
showed more than a 2% of sequence divergence with its species sequences
(PA-1205 and PA-3105 assigned to P. aspera with a 94.4% and 97.8% of
similarity, respectively).
The levels of COI genetic variation for each of the mentioned above species
are shown in Tab. 1. Inside the Patella genus we found an 8.8% and 88.1%
of nucleotide (p) and haplotype (h) diversity, respectively. The most variable
species was P. aspera (p =1.60%) while P. vulgata appeared as the less variable
one (p =0.07%) (Tab. 1).
The COI phylogenetic tree using the Bayesian approach showed two main
nodes: one contains the P. aspera individuals and another contains the P.
vulgata ones and also a group of P. depressa and P. rustica individuals (Fig. 2).
One individual (PA-1205), morphologically classified as P. aspera, was located
inside the P. vulgata branch although it showed only a 93.6% of similarity with
the available P. vulgata sequences (CBOL).

4 Discussion
Our sampling and the identification methods used here revealed four Patella
species in Asturias (P. vulgata, P. depressa, P.aspera, and P. rustica) confirming

283
 Fig. 2 – Consense Phylogenetic tree after Bayesian analysis using Cytochrome
Oxidase I (COI) sequences in the genus Patella. C.safiana as output group. Numbers
represent more than a 70% of branch support. A P.aspera individual grouping with the
P.vulgata group is indicated by a discontinuous circle.

284
 bp n π h

Haplotypes numbers

Polymorphic. sites
Patella Species Ts Tv Sb Id

Genbank AN
Gene

COI 642 45 23 EF462952- 156 139 38 177 0 0.0885 0.8808

75

P.vulgata 20 5 EF462957- 5 4 1 5 0 0.0007 0.4474

62

P.depressa 5 4 EF462972- 8 5 3 8 0 0.0049 0.9000

75

P.rustica 5 5 EF462952- 7 5 2 7 0 0.0046 1.0000

56

P.aspera 15 9 EF462963- 68 53 16 69 0 0.0160 0.8857

71

Tab. 1 – Genetic variation in patellids from the Astrurian Coasts.

bp: base pairs. n: number of samples. Genbank AN: Genbank accession numbers for
haplotypes. Ts: transitions. Tv: transversions. Sb: substitutions. Id: indels. π: Nucleotide
diversity. h: Haplotype diversity. COI: Cytocrome Oxidase I.

previous morphological studies from the 80s [12], [13]. Two individuals fall
apart from the criterion of less than a 2% of divergence for correct species
classifications (following the CBOL recommendations) and the PA-1205
haplotype is clearly out of its putative species branch in the phylogenetic COI-
trees. This could point to cases of genetic introgression and to the necessity to
clarified and revise the taxonomic classifications in Patella species [3].
The patellids found showed a dissimilar species distribution in the Asturian
coastal area with P. rustica being only present in the Asturian/Galician frontier.
Possibly, this species is the most sensible to changes in the sea surface
temperature that determine its reproductive success and hence its dispersal
potential [14]. The commercially exploited species (P. vulgata) is the genetically

285
less variable. This raises concerns about the health of this species in Asturias.
The Patella genus is monophyletic inside the Patellidae family. Sá-Pinto et al.
[3] showed five strongly supported clades for patellids although relationships
between them were not well supported. Working with four species included
in the Sá-Pinto et al. [3] study we have recovery the expected four clades
phylogenetic tree. However, we have well supported branches (more than 80%)
indicating different relationships among these clades. Our results revealed
two main nodes: one included P. aspera (P. ulyssiponensis) as an entity and
another included P. vulgata, P. depressa and P. rustica species. Our results
differ from the closeness between P. depressa and P. vulgata species proposed
by Koufopanou et al. [2] (P. depressa is more close to P. rustica in our work) and
also from the relationships between clades showed by Sá-Pinto et al [3]. They
showed P. depressa (clade I) together with P. aspera (Clade II) and these two
grouped to P. vulgata (Clade III), while P. rustica (Clade IV) was an independent
entity. It will be necessary much more work and different approaches to ascertain
which the relationship between clades inside the genus Patella is. This will help
to clarify taxonomy and will give us clues about speciation patterns and origins
inside the genus. All this information will be vital for its adequate conservation
and management.

5 Conclusion
Molecular methods are useful tools for species identification when morphological
analyses lead to taxonomic confusions. Using COI gene sequences we have
confirmed the presence of four patellids in the Asturian coasts (P. vulgata, P.
depressa, P. aspera, and P. rustica). Our work raises concerns about the current
state of the P. vulgata populations in Asturias, where it is exploited, due to its low
levels of genetic variation. Our phylogenetic analyses confirmed that patellids
belong to four different clades; however our work gives a new version about how
these clades are related inside the genus aiming for the necessity of more work
to address this issue.

Acknowledgements

The authors wish to acknowledge the “Consejería de Medio Rural y Pesca de Asturias,
España” for permissions to obtain biological samples. Thanks also to Anadon N. from
the Departamento de Organismos y Sistemas, Universidad de Oviedo, who carried out
morphological classifications.

References
[1] S. A. Ridgway, D. G. Reid, J. D. Taylor, G. M. Branch and A. N. Hodgson, “A cladistic phylogeny
of the family Patellidae (Mollusca: Gastropoda)”. Proc. R. Soc.Lond., B., vol. 353, pp. 1645-
1671, 1998.
[2] V. Koufopanou, D. G. Reid, S. A. Ridway and R. H. Thomas, “A molecular phylogeny of
the Patellidae limpets (Gastropoda: Patellidae) and its implications for the origins of their
antitropical distribution”. Mol. Phyl. Evol., vol. 11, pp. 138-156, 1999.
[3] A. Sá-Pinto, M. Branco, J. D. Harris and P. Alexandrino, “Phylogeny and phylogeography of
the genus Patella based on mitochondrial DNA sequence data”. J. Exp. Mar. Biol. Ecol., vol.

286
 325, pp. 95-100, 2005.
[4] A. Mauro, M. Arculeo and N. Parrinello, “Morphological and molecular tools in identifying the
mediterranean limpets Patella caerulea, Patella aspera and Patella rustica”. J. Exp. Mar. Biol.
Ecol., vol. 295, pp.131-143, 2003.
[5] J. P. Thorpe, A. M. Solé-Cava and P. C. Watts, “Exploited marine invertebrates: genetics and
fisheries”. Hydrobiologia, vol. 420, pp. 165-184, 2000.
[6] P. D. N. Hebert, A. Cywinska, S.L. Ball, J. R. DeWaard, “Biological identifications through DNA
barcodes”. Proc. R. Soc. Lond. B Biol. Sci., vol. 270, pp. 313-321, 2003.
[7] O. Folmer, M. Black, W. Hoeh, R. Lutz and R. Vrijenhoek, “DNA primers for amplification of
mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates”. Molecular
Mar. Biol.Biotech., vol. 3, pp. 294-299, 1994.
[8] J. P. Wares and C. W. Cunningham, “Phylogeography and historical ecology of the North
Atlantic intertidal”. Evolution, vol. 12, pp. 2455–2469, 2001.
[9] S. J. Hawkins, H. B. S. M. Côrte-Real, F. G. Pannacciulli, L. C. Weber and J. D. D. Bishop,
“Thoughts on the ecology and evolution of the intertidal biota of the Azores and other Atlantic
islands”. Hydrobiologia, vol. 440, pp. 3-17, 2000.
[10] F. Espinosa and T. Ozawa, “Population genetics of the endangered limpet Patella ferruginea
(Gastropoda: Patellidae): taxonomic, conservation and evolutionary considerations”. J. Zool.
Syst. Evol. Res., vol. 44, pp. 8-16, 2006.
[11] E. Fischer-Piette and J. Gaillard, “Les Patelles au long des cotes Atlantiques Ibériques et nord
Marocaines”. J. Conchyliologie, vol. 99, pp. 135-200, 1959.
[12] M. P. Miyares, Biologia de Patella intermedia y P vulgata (Mollusca, Gasteropoda) en el litoral
asturiano (N. de España) durante un ciclo anual (Diciembre de 1978 a Noviembre 1979). Bol.
Cienc. Nat. I.D.E.A., vol. 26, pp. 173-192, 1980.
[13] J. A. Ortea, “El género Patella Linné 1758 en Asturias”. Bol. Cienc. Nat. I.D.E.A., vol. 26, pp.
57-72, 1980.
[14] F. P. Lima, N. Queiroz, P. A. Ribeiro, Hawkins and A. M. Santos, “Recent changes in the
distribution of a marine gastropod, Patella rustica Linnaeus, 1758, and their relationship to
unusual climatic events”. J. Biogeogr., vol. 33 (5), pp. 812-822, 2006.

287
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 289-294.
ISBN 978-88-8303-295-0. EUT, 2010.

Molecular Identification of Italian

Mouse-eared Bats
(genus Myotis)
Andrea Galimberti, Adriano Martinoli, Danilo Russo,
Mauro Mucedda, Maurizio Casiraghi

Abstract — Despite the fact that the genus Myotis (Mouse-Eared bats) is
one of the most investigated microchiropteran groups, recent molecular
studies highlighted the presence of several cryptic species with substantial
implications for ecological and conservation issues. Our dataset includes 55
coxI sequences from 11 morphologically-identified Italian Mouse-Eared bats
species. We applied an integrated approach comparing data from a traditional
morphological identification and molecular variability in a fragment of the
mitochondrial coxI gene (DNA barcoding). Our results clearly show a strong
coherence between the two identification approaches for almost all of the
examined species, and revealed interesting patterns of intraspecific variability
within the species M. nattereri. Finally, we successfully tested the efficacy of
our identification method on undetermined individuals sampled in the field.

Index Terms — cryptic species, DNA barcoding, integrated taxonomy,

Vespertilionidae.

—————————— u ——————————

1 Introduction

M
ammals are usually considered as one of the best-known animal groups.
However, several studies provided clear evidences that bats (order
Chiroptera) are characterized by a high incidence of overlooked taxa
due to their cryptic morphology and habits [1]. A clear example of this situation is
given by the recent taxonomic changes within the family Vespertilionidae (one of
the best-studied taxonomic group of bats) in the Western Palearctic. Thanks to
the development of recent molecular techniques, the number of species within

————————————————
A.Galimberti and M. Casiraghi are with the Department of Biotechnologies and Biosciences, Uni-
versity of Milan-Bicocca, Milan, Italy. E-mail: [email protected].
A. Martinoli is with the Department of Environmental Health and Safety, Universities of Insubria-
Varese, Italy. E-mail: [email protected].
D. Russo is with the Department Ar bo Pa Ve, Laboratory of Applied Ecology, Agrarian faculty,
University of Naples “Federico II”, Naples, Italy. E-mail: [email protected].
M. Mucedda is with the Centro per lo Studio e la Protezione dei Pipistrelli in Sardegna,Sassari,
Italy.

289
this family has risen from 37 to 54, with at least 8 new cryptic species identified
in Europe [2], [3], [4].
Within the Mediterranean basin, the Italian peninsula is one of the most
important biodiversity hotspots for bats and other taxa [5]. It has been
hypothesized that this peninsula would have provided stable habitats during ice
ages, where species survived leading to the generation of new cryptic lineages
[5]. 33 microchiropteran species out of the almost 40 currently known to live
in Europe are reported for Italy [5], the family Vespertilionidae being the most
diverse and abundant (8 genus and 27 species). As showed in recent studies,
the family Vespertilionidae is characterized by high levels of cryptic diversity
[3], [4] and in particular in Italy, at least 11 different species are included in the
group of Mouse-eared bats (genus Myotis). This taxon is the most problematic
concerning species identification, due to the presence of cryptic species like in
the ‘Whiskered-bats’ complex (i.e.: M. mystacinus, M. Brandtii, M. alcathoe),
peculiar biogeographical histories (i.e.: M. Myotis and M. Blythii; [6]) and
genetically-uncharacterized lineages (i.e.: M. Nattereri; [2], [4]).
Despite the fact that the genus Myotis includes several threatened species,
the compilation of realistic action plans for their conservation is biased by some
practical difficulties: bats are hard to observe because of their elusive habits.
Moreover, several species are cryptic and sometimes it is impossible to reach
a correct identification in the field, especially for juveniles or females [5], [7].
Despite the fact that morphological identification keys are available for European
bats (e.g. [7]), integration with molecular approaches has proven to be efficient
in detecting morphologically cryptic species [2], [3], [4], [8], [9]. An efficient and
widely used molecular tool in species identification is DNA barcoding [10]. This
technique is based on the analysis of the variability in the nucleotide sequence
of a short, standardized region of the genome (among metazoans the 5’- end
of the mitochondrial subunit 1 of cytochrome c oxidase), to evaluate differences
among species [10]. A few studies have shown the efficacy of coxI in identifying
bats species, but any (e.g.:[1]) work was conducted with a standardized DNA
barcoding approach for European and Italian Myotis and other Vespertlionids.
The main objectives of our study are: i) to compile a reference dataset of
coxI sequences from all the Italian Myotis species; ii) to test the coherence
between a molecular approach and the morphologically-based taxonomy and
iii) to investigate the intraspecific molecular variability of the barcode region to
reveal the presence of undescribed cryptic lineages.

2 Materials and Methods

2.1 Sampling, DNA extraction, sequencing and alignment

The samples analysed in this study derive from 55 bats belonging to 11 Myotis
species collected during 2006-2007 from 16 Italian localities distributed along
the Italian peninsula. Bats were identified by researchers of the GIRC (Italian
Chiroptera Research Group). Tissue samples (i.e.: ‘punches’ of patagium, 3mm
large) were stored in ethanol 96%. According to the protocol specified by the

290
Biorepositories initiative (http://www.biorepositories.org) each sample was
vouchered with the id Institution name ‘MIB:ZPL:’ followed by a progressive
numeric code. Sixteen unidentified individuals belonging to the genus Myotis
were also collected.
Total genomic DNA was extracted using a guanidinium thiocyanate and
diatomaceous earth protocol [11]. coxI amplification and sequencing were
obtained following the laboratory protocols provided by [1]. Sequences were
checked and aligned following the approach described in [12] and, after
checked for the presence of pseudogenes and numts (i.e. nuclear mitochondrial
pseudogenes), alignment was cut to 560 bp in order to have all the sequences
of the same length.

2.2 Molecular dataset definition and DNA barcoding analyses

To evaluate the efficacy of a DNA barcoding approach as a molecular tool

to identify Italian Mouse-Eared bats species, we assembled a ‘Reference
Dataset’ including only coxI sequences from Italian morphologically identified
bats belonging to the genus Myotis. An ‘Optimum Threshold’ value of molecular
divergence (OT) was then calculated (following [12]) directly from the whole
range of molecular variability of the molecular dataset. OT value maximizes
the coherence between the morphological identification and the molecular
divergence minimizing, at the same time, the total amount of possible mismatches
(see Minimum Cumulative Error analysis in [12]). To avoid biases during OT
calculation, we removed Myotis nattereri barcode sequences because of its
high intraspecific variability, probably due to the presence of undescribed cryptic
lineages (as clearly shown in [2] and [4])
OT was then used to perform a DNA barcoding analysis on the ‘comprehensive
molecular dataset’, containing all reference barcodes and also the coxI
sequences of the 16 unidentified Italian Myotis sp. individuals and the 9 Myotis
nattereri samples. Intraspecific molecular variability was then analyzed to test
the congruence with previously described species. OT has also been used to
predict potentially new taxa within the dataset. Finally, a neighbor-joining (NJ)
tree (Fig. 1) was generated from the comprehensive molecular dataset using
MEGA 4.1 according to the parameters provided in [12].

3 Results and Discussion

Any coxI sequence exhibited indels, stop-codons or numts interferences. The
‘Reference Dataset’ included 46 barcode sequences 560 bp long belonging to all
the Mouse-eared bat species distributed in Italy, except for M. nattereri (average
number of coxI sequences for every single species: 5, standard deviation: 3.05,
range: 1-9).
Bioinformatic analyses on the ‘Reference dataset’ allowed us to infer the
following parameters: mean K2P distance within species 0.6% (standard
deviation: 0.7%; range: 0% – 2%), mean K2P distance between species 15.8%
(standard deviation: 2.8%; range: 2% – 20.1%) and coxI overall mean diversity
14% (standard deviation: 1.1%). Concerning the OT calculation, we obtained

291
 Fig. 1 – NJ Tree. Neighbour joining tree based on coxI sequences of Italian Mouse-
eared bats generated with MEGA 4.0 (Tamura et al, 2007). Unidentified samples are
indicated as “Myotis sp.”. Cryptic molecular lineages inferred using OT Threshold are
indicated with square brackets.

292
a minimum cumulative error of 4.36% at a Optimum Threshold value of 4.2%.
Only one identification mismatch occurred, due to the low interspecific variability
(lower than OT) observed between the species M. myotis and M. blythii (mean
K2P distance between species: 2.0%, standard deviation: 1.7%, range: 0% –
4.1%). A similar result has been previously reported in other molecular studies
on these taxa (e.g.: [6], [8]) and relies on the fact that a series of introgression
events having occurred repeatedly during the recent colonization of Europe by
M. blythii from Asia. Hybridization is still ongoing in the areas of sympatry (e.g.:
in Italy), therefore suggesting an unclear taxonomic status of these taxa in the
Western Palearctic.
The application of the OT value on the ‘comprehensive dataset’ allowed
to assign the unidentified specimens (Fig.1) to 12 M. mystacinus and 4 M.
alcathoe. These are two cryptic sympatric species of Mouse-eared bats one of
which (M.alcathoe) has been recently described [13] and which status in Italy is
almost unknown.
Moreover, OT revealed the presence of two cryptic lineages within the taxon
M. nattereri (here tentatively named ‘Lineage I’ and ‘Lineage II’) exclusive of
Northern and Southern localities of the peninsula respectively (Fig.1). The
mean K2P distances within each lineage are 0.7% and 0.4% for Lineage I and
II respectively, while the mean K2P distance between the two lineages is higher
than OT: 5.6%. Garcia-Mudarra and colleagues [4] recently identified at least
four European cryptic molecular lineages within this taxa and they concluded
that ecological as well as morphological studies would be desirable before
any definitive conclusions can be drawn about its taxonomic status. Moreover,
preliminary molecular comparisons among our lineages and other mitochondrial
sequences available in GenBank (i.e: ND1 and cytb) revealed that the Southern
Italian ‘Lineage II’ discovered by our DNA barcoding approach is completely
undescribed and could represent a new cryptic Myotis species for the Western
Palearctic (data not shown).

4 Conclusions
Our study provides clear evidences that DNA barcoding is a reliable and
efficient tool for the discrimination of almost all the Italian Mouse-eared bats,
showing a high strength of coherence between data based on classical morphology
and variability in the mitochondrial coxI barcode region. OT value calculated from
our dataset allows to infer a clear taxonomic assignment for all the morphologically-
unidentified individuals collected in the field. Moreover, the OT value inferred from the
molecular dataset is efficient to reveal the presence of undescribed cryptic lineages
within known species, like the case of M. nattereri. These results suggest that DNA
barcoding could be successfully used as a reliable support to ecological studies in
order to develop efficient conservation strategies for endangered bats populations.

Acknowledgement

This work was in part supported by a grant from Fondazione Cariplo.

293
References
[1] E. L. Clare, B. K. Lim, M D. Engstrom, J. L. Eger and P. D. N. Hebert, “DNA barcoding of
Neotropical bats: species identification and discovery within Guyana”, Molecular Ecology
Notes, vol. 7, pp. 184-190, Mar 2007.
[2] C. Ibanez, J. L. Garcia-Mudarra, M. Ruedi, B. Stadelmann and J. Juste, “The Iberian
contribution to cryptic diversity in European bats”, Acta Chiropterologica, vol.8, pp. 277-297,
2006.
[3] F. Mayer, C. Dietz and A. Kiefer, “Molecular species identification boosts bat diversity”,
Frontiers in Zoology, vol. 4(1), p. 4, 2007.
[4] J. L. Garcia-Mudarra, C. Ibanez,and J. Juste, “The Straits of Gibraltar: barrier or bridge to
Ibero-Moroccan bat diversity?”, Biological Journal of The Linnean Society, vol. 96, pp. 434-
450, Feb 2009.
[5] P. Agnelli, A. Martinoli, E. Patriarca, D. Russo, D. Scaravelli and P. Genovesi (eds.), “Guidelines
for bat monitoring: methods for the study and conservation of bats in Italy”, Min. Ambiente –
Ist. Naz. Fauna Selvatica, Rome and Ozzano dell’Emilia (Bologna), Italy, Quad. Cons. Natura
Series, vol. 19bis, 2006.
[6] P. Berthier, L. Excoffier and M. Ruedi, “Recurrent replacement of mtDNA and cryptic
hybridization between two sibling bat species Myotis myotis and Myotis blythii”, Proc. R. Soc.
B., vol. 273, pp. 3101-3109, 2006.
[7] C. Dietz and O. von Helversen, “Illustrated identification key to the bats of Europe”, available at
http://www.fledermaus-dietz.de/publications/publications.html, version 1.0., Dec. 2004.
[8] F. Mayer and O. von Helversen, “Cryptic diversity in European bats”, Proc. R. Soc. B., vol. 268,
pp. 1825-1832, Sept. 2001.
[9] S. M. Goodman, C. P. Maminirina, N. Weyeneth, H. M. Bradman, L. Christidis, M. Ruedi and
B. Appleton, “The use of molecular and morphological characters to resolve the taxonomic
identity of cryptic species: the case of Miniopterus manavi (Chiroptera, Miniopteridae)”,
Zoologica Scripta, vol. 38(4), pp. 339-363, Jul. 2009.
[10] P. D. N. Hebert, S. Ratnasingham and J. R. de Waard, “Barcoding animal life: cytochrome
c oxidase subunit 1 divergences among closely related species”, Proc. R. Soc. B., vol. 270,
suppl. 1, pp.S96-S99, Aug. 2003.
[11] U. Gerloff, C. Schlötterer, K. Rassmann, I. Rambold, G. Hohmann, B. Fruth and D. Tautz,
“Amplification of hypervariable simple sequence repeats (microsatellites) from excremental
DNA of wild living bonobos (Pan paniscus)”, Molecular Ecology, vol. 4, pp. 515-518, 1995.
[12] E. Ferri, M. Barbuto, O. Bain, A. Galimberti, S. Uni, R. Guerrero, H. Ferté, C. Bandi, C.
Martin and M. Casiraghi, “Integrated taxonomy: traditional approach and DNA barcoding for
the identification of filarioid worms and related parasites (Nematoda)”, Frontiers in Zoology,
vol.6(1), Jan 2009.
[13] O. von Helversen, K. G. Heller, F. Mayer, A. Nemeth, M. Volleth and P. Gombkötö, “Cryptic
mammalian species: a new species of whiskered bat (Myotis alcathoe n. sp.) in Europe”,
Naturwissenschaften, vol.88, pp. 217-223, May 2001.

294
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 295-299.
ISBN 978-88-8303-295-0. EUT, 2010.

Identifying algal symbionts in

lichen symbioses
Martin Grube, Lucia Muggia

Abstract — Lichens are a ubiquitous terrestrial symbiosis of fungi

with photoautotrophic microorganisms. The identification of the hosted
photoautotrophs is notoriously difficult. Molecular data to clarify evolutionary
relationships on the involved algal and cyanobacterial lineages are
accumulating, but the assignment to species is challenging for various
reasons. One of the challenges is the limited knowledge on the alpha diversity
of photoautotrophs. New lineages are being discovered with increasing
amounts of sequencing. Identification tools could incorporate these aspects,
by routinely updating the assignment process. We propose the establishment
of a classification tool using algal sequence data from public databases.

Index Terms — lichens, symbionts, photobionts, ITS, actin.

—————————— u ——————————

1 Introduction

L
ichens are symbioses of fungi and photoautotrophic partners (algae and/
or cyanobacteria). Lichens are widespread in all climatic zones and cover
more than 8% of the land surface [1]. Lichens are generally named after
the morphology-determining fungal partner which represents more than 18.800
known species of Ascomycetes [2]. Contrarily, the knowledge about photobiont
species diversity is still limited. The determination of lichen photobionts is
complicated due to the lack of diagnostic characters for routine analyses. Algae
in lichenized stage do not express useful characters at all, and cultivation of
algae is time-consuming and not yet possible for some lineages [3].
Recent DNA sequence analyses have studied phylogenetic diversity of algal
partners in lichen symbioses. About 50% of the lichen fungal species associate
with single-celled green algae, and most of these belong to the genus Trebouxia
(Trebouxiophyceae, Chlorophyta). Although morphologically similar, different
genetic lineages of these photobionts are detected in wide geographic ranges
of the same lichen fungal species.
Algal symbiont sequence information is usually obtained by using algal
specific primers for amplification from total lichen (=holobiont) extracts, which
avoids the contamination by sequences of the fungal partner, and multiple co-
occurring bacteria. The phylogenetic analyses of the internal transcribed spacer
————————————————
The authors are with the Institut für Pflanzenwissenschaften Karl-Franzens-Universität Graz
Holteigasse 6, 8010 - Graz, Austria. E-mail: [email protected].

295
(ITS) nuclear region uncovered relationships among trebouxioid photobionts
and selectivity of the fungal partners for their algae. Sequence data of this
group of algae have accumulated significantly in the past years and meanwhile
a search in GenBank using “Trebouxia ribosomal ITS” returns 1356 hits (as of
07.07.2010).
Phylogenetic analyses are used to assign sequenced strains to named
species. This is not at all a trivial procedure, because sequence divergence
among recognized species is far from equal: some taxa are separated by few
nucleotide changes while more pronounced sequence divergence can be
detected among other species. Thus there is uncertainty about the assignment
of sequences within the range of divergence among two species.
Previous study found that sequence divergence of Trebouxia is recognized in
several main phylogenetic clades, which have been designated by letters A, I,
G, S [4]. Within these clades, subclades are distinguished by numerals. This has
lead to a fairly resolved phylogenetic classification of lineages in Trebouxia. This
phylogeny does not agree perfectly with phenotypical classification and species
taxonomy. It has also been revealed that diversity of these algae was previously
underestimated and includes many yet to be described species. This includes
entirely new lineages as well as the better characterization of yet cryptic lineages
within broadly understood species. Nevertheless, new names for species are
rarely introduced, e.g. when morphological characters correlate with distinct
phylogenetic positions. Fig. 1 (modified from [5]) displays the challenges. In a
recent publication [3] we could show that the sister clade (Trebouxia sp. 1) of
Trebouxia arboricola cannot be distinguished by ultrastructural data of cultured
algae from that species, whereas the distinct clade Trebouxia sp. 2 could not be
cultured with standard methods. Thus the phylogenetic suggest a distinct clade
but phenotypic support is still missing. Whether the basal lineages of that clade
could represent a further species still needs to be awaited and supported by
further sequence data.
Variation within an algal species will more precisely be estimated with more
sequence data. Because sexual stages are cryptic in lichenizing Trebouxia,
sequence evolution in clonal lineages could blur species delimitation. We expect
that species are increasingly recognizable as ‘clusters’ in the sequence space
with appropriate gene loci. Sequence divergence of ITS is suitable for DNA
barcoding of green algal lichen symbionts. We therefore suggest establishing
an automated assignment tool that tests query sequences against a regularly
updated database of lichen algal ITS sequences. Automated classifiers have
been incorporated in the RDP database of bacterial rDNA [6], but do not yet exist
for eukaryotes. Moreover, assignments have to consider the growing amount
of environmental sequences without assignment to taxonomic names. We are
therefore exploring methods to assess the coherence of related sequence as
clusters in the sequence space, and the confidence of their assignment to
species names. This work is still in progress and more details will be presented
at the Bioidentify meeting in Paris.

296
Fig. 1 – Phylogenetic tree of Trebouxia species as algal symbionts in lichens (modified
from [3]). The tree is constructed using ITS rDNA sequence data. Symbionts in
Mediterranean samples of the lichen Tephromela atra are named informally as
Trebouxia sp. 1 and sp. 2. Further information is required for taxonomic recognition of
these cryptic species.

2 Multiple phototrophs in lichens

Lichen fungi may internalize more than one algal symbiont. This is clearly
observed when a green algal lichen thallus contains nitrogen-fixing cyanobacteria
in specialized organs. Less clear, however, are cases where several algae of the
same green algal lineages are involved. Culture based studies have previously
shown that a single individual of lichen can host several lineages of Trebouxia.
This variation is often poorly detected with conventional PCR approaches
using whole thallus DNA extracts, whereby usually only one distinct sequence

297
is detected. Any additionally occurring algal sequences are obscured by the
exponential amplification of the most common sequence during PCR. Because
diagnostic characters of algae in the lichenized stage are hardly available it is
still unclear how multiple strains of algae are distributed in lichens. Are additional
algae merely epibionts, are they evenly distributed in low abundance throughout
the thallus, or are they localized in certain parts of a lichen thallus?

 
Fig. 2 – Identification of multiple algal symbionts in the lichen Lecanora muralis by
single strand conformation polymorphism (SSCP) detection. External (odd lanes) and
internal (even lanes) areoles from 5 lichen individuals were analysed. Bands of equal
position represent distinct algal genotypes. Lecanora muralis associates with several
algal genotypes, and areoles 1, 7, 9, and 10 display heterogeneity for the algal partner.

Several methods are available to analyse the composition of microbial

communities. One of these is single strand conformation polymorphism (SSCP)
analysis. With this method single-stranded DNA fragments can be separated
on a gel according to their nucleotide sequence variation. The separated bands
can then be excised from the gels for sequencing. Each band with different run-
length represents a different sequence/strain of algae. Best results are obtained
with sequence fragments of 200-300 nucleotides length, i.e. within the size
range of the ITS subregions flanking the internal 5.8S rDNA. We are now using
this method to explore the algal composition in lichens in greater detail. Fig. 2
shows a detail of a SSCP analysis of individual areoles of 5 specimens of the
euryoecious lichen Lecanora muralis. Odd lanes are from external areoles, even
lanes are from internal areoles. Symbiont heterogeneity is observed among
thalli and within thalli (especially in the two samples representing lanes 7-10).
We assume that different strains are not evenly distributed at dissimilar
abundances in the entire thallus. We rather think that thallus outgrowths can
newly associated with algae, and that different algae could be detected in
different areoles or lobes of a lichen individual. We expect that this flexibility
could contribute to the ecological adaptivity of euryoecious lichens.

3 Conclusion
The ITS rDNA sequences of lichen photobionts are useful DNA barcodes to
study partner selectivity in symbioses. Here we focused on algal partners in
lichens. The major challenge of this work is the still unsettled taxonomy of algae,
and that several algal symbionts may be present in lichen individuals. A self-
organising classification tool that uses regularly updated sequence information

298
on algal lichen symbionts is under development.

Acknowledgement

We thank Toby Spribille (Graz) for comments on the text. This work was supported in
part by a grant from FWF (P17601).

References
[1] J. V. Ahmadjian, “Lichens are more important than you think”. Bioscience vol. 45, pp. 123–124,
1995.
[2] T. Feuerer and D. L. Hawksworth, “Biodiversity of lichens, including a world-wide analysis of
checklist data based on Takhtajan’s floristic regions.” Biodiversity and Conservation, vol. 16,
pp. 85–98, 2007.
[3] L. Muggia, G. Zellnig, J. Rabensteiner and M. Grube, “Morphological and phylogenetic
study of algal partners associated with the lichen-forming fungus Tephromela atra from the
Mediterranean region”. Symbiosis, on line first, 2010.
[4] G. Helms, Taxonomy and Symbiosis in Associations of Physciaceae and Trebouxia. Univ.
Göttingen (Doctoral thesis), 2003.
[5] L. Muggia, M. Grube and M. Tretiach, “Genetic diversity and photobiont associations in selected
taxa of the Tephromela atra group (Lecanorales, lichenised Ascomycota)”. Mycological
Progress, vol. 7, pp. 147-160, 2008.
[6] J. R. Cole, Q. Wang, E. Cardenas, J. Fish, B. Chai, R. J. Farris, A. S. Kulam-Syed-Mohideen,
D. M. McGarrell, T. Marsh, G. M. Garrity and J. M. Tiedje, “The Ribosomal Database Project:
improved alignments and new tools for rRNA analysis”, Nucleic Acids Research, vol, 37
(Database issue), D141-D145; doi: 10.1093/nar/gkn879, 2009.

299
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 301-305.
ISBN 978-88-8303-295-0. EUT, 2010.

Identification of polymorphic
species within groups of
morphologically conservative
taxa: combining morphological
and molecular techniques
Kim Larsen, Elsa Froufe

Abstract — Identification of small species with high levels of intra-specific

polymorphism within groups of inter-specific morphologically-conservative
taxa, presents numerous obstacles for biodiversity and ecological studies.
This is particularly true for deep-sea studies that often reveal a great number
of species but only few numbers of specimens. It is here proposed to deal
with such cases by extrapolating information obtained from highly detailed
baseline studies. Such baseline studies should include information about
sexual and ontogenetic variation and should include a combination of both
morphological and molecular techniques.

Index Terms — baseline studies, polymorphism, sibling species, species

identification.

—————————— u ——————————

1 Introduction

T
he identification of species can be problematic enough when dealing with
taxa which include a large number of morphologically similar species. The
obstacles can increase manifolds with smaller taxa that display few stable
characters and show tendencies towards reductions. Adding the complications
of substantial sexual and ontogenetic variations, the results are often misleading
to the point of being meaningless. This is particularly true for deep-sea studies
that often reveal numerous species but only few specimens for each species.
One example of such a problematic group is the Tanaidacea (Crustacea:
Peracarida), but there are many other, similar difficult taxa, among the smaller
invertebrates. In the Tanaidacea species, differentiation is notoriously difficult,

————————————————
The authors are with CIIMAR (Center for Interdisciplinary Investigation of the Marine environment),
LMCEE (Laboratory of Marine Community Ecology and Evolution), Rua dos Bragas 289. 4050-
123, Porto, Portugal. E-mail: [email protected], [email protected].

301
and males/juveniles often share no species-specific characters with females;
even family level identification of males can be hazardous [1]. In many families,
multiple polymorphic males exist- the consequence of a peculiar reproductive
strategy involving protogynous hermaphrodites- and this causes additional
problems [2]. As males locate females by roaming the substrate, they are
exposed to high predation pressure that - combined with their non-feeding
life-style - makes the life span of males short. In situations where depletion
of males from the population occurs, some females may molt into males at
several different instars, each resulting in a morphologically different male (up to
four different male morphs have been recorded in one species) [2]. Ontogenetic
variations among adult females are also known to cause problems [1].
At the same time, the tanaidaceans are infamous for creating species
complexes containing many, often sympatric, species that display a very
conservative inter-specific morphology. This again makes species identification
exceedingly difficult [1].
Tanaidacea are particularly common in deep-sea substrates, where they
constitute a major proportion, up to 22 % by some estimates, of the total fauna
(in terms of biodiversity) [3], [4]. It is clearly undesirable for any scientific study
(particularly in biodiversity and ecology) that such a large proportion of the fauna
cannot be identified, but the solution to this problem is not apparent due to
time constraints and lack of available expertise. Many large scale biodiversity
programs rely heavily on cheap (and poorly trained/supervised) student help
to process the often enormous material of small benthic invertebrates. Clearly,
given the problems inherent with taxa displaying such troublesome attributes as
described above, such personnel have little chance of successful identification
(we have personally observed as much as 50% misidentifications in collections
from biodiversity studies).

2 Methods
The methods we suggest here for species identifications are not new but
make use of an expanded procedure. Firstly the samples should be screened
and identified to order. Thereafter, samples with large number of specimens
should be given priority. Those high-value samples should then be sent to
taxonomical experts for ‘baseline’ processing. Do NOT use untrained student
assistants for this part. Once the experts have reported the baseline study, make
the identifying personnel use these for comparisons with each single species.
Singletons should not be dissected, but not assigned species rank either (like
for example ‘Sp. A’), until comparisons have been made with other singletons of
different instars that may belong to the same species.

2.1 Morphological Methods

The baseline study should be conducted in the utmost details, including

the dissection, description, and illustration of ANY appendages (not just
those normally regarded as taxonomically informative) for ALL developmental
stages encountered. Both the lateral and dorsal view of the body should be

302
drawn. Dissection should include appendages from BOTH sides of the body.
Appendages should be mounted in glycerin dyed with clorasol black, sealed
with nail polish, and stored for further studies. All character transformations seen
from manca-juvenile-adults should be noted and illustrated as a transformation
diagram. The baseline study should result in the manufacturing of a guide to
identification of genders and developmental stage, and supplied to the personal
conducting the identifications of the entire material.

2.2 Molecular methods

2.2.1 DNA Extraction

The main problem with DNA extraction from such specimens is the very low
yield of starting tissue available (for the smaller taxa, the entire animal has to be
used, since a leg or other appendages do not yield enough DNA). Therefore, the
extraction is crucial for further analyses and usually requires some modifications
to frequently used protocols. There are several DNA isolation techniques. Here
we describe our modifications to one frequently used protocol: silica columns.
The most crucial points are as follows: VERY thorough grinding of samples,
prolonged periods in the several steps stated in the DNA extraction Kit (we use
JETQUICK Tissue DNA kit) and also prolonged periods for the final elution step.
Insufficient disruption of starting material leads to low yield and purity, therefore
this step is crucial; we use hand-made hard-plastic cylinders which are efficient
in disruption and homogenization of the hard crustaceans exoskeleton and also-
because of their small size- can be used in micro-centrifuges tubes avoiding the
risk of contamination (they can also be autoclaved) and avoiding loosing tissue
(the same micro-centrifuges tubes can be used for proteinase-k digestion).
Extraction can be performed according the JETQUICK protocol but should
be modified by increasing the time length of each step, from incubation with
proteinase K to each centrifuge step (we used double time). Due to the low
final DNA concentration; the same elution solution should then be used for the
DNA elution and the same for the second elution step (pre-warmed at 70ºC for
five minutes). Densitometric measurements are not useful for detection of small
amounts of DNA [5] so the “Qubit” flurometer is ideal (requires only 1µl DNA
elution).

2.2.2 PCR

The basic “PCR rules” HAVE to be employed when dealing with these kind of
samples, e.g., cleaning the bench top with alcohol before setting up reactions,
using plugged tips for all PCR reagents (to avoid contamination), always including
a sample without template as a negative control to check for contamination
of the reagents. The most crucial points are as follows: short length of PCR
products (optimum of 300-350 bp) and higher number of PCR cycles.
The amount of DNA used will depend on the concentration of the sample. It
is best to use a “hot start” Taq that will provide increased sensitivity, specificity

303
and yield. Due to the high numbers of PCR cycles needed the quality of the Taq
is also important (we used Platinum Taq DNA Polymerase). Finally in order to
avoid adding enzyme inhibitors that may be present, we recommend the use of
a high PCR final volume (20 µl).

2.2.3 DNA Sequencing

At this stage, the products must be checked for both quantity and quality.
Agarose gel electrophoresis can be used to visualize the amount and size of
DNA fragments present in the sample, and since usually the amount of final PCR
concentration is low when using these type of samples, we recommend to dry
up the total PCR product (use a vacuum centrifuge) into a loading agarose gel
volume and excised the PCR gel band. We used several different commercial
Gel extraction kits, with no significant results among them. The only modification
is the final elution step, which should be no higher than 10µl (we used 5 µl). The
DNA sequencing can proceed as usual hereafter.

3 Discussion
Given the large material of ‘difficult taxa’ often encountered during biodiversity/
ecological studies (particularly from deep-sea environments), the limited
expertise available on many such taxa, and the financial restraints, it is not
possible to have specialists processing all the material. Therefore we propose to
deal with these problems by extrapolating information obtained from the highly
detailed baseline-studies described above. We are not so much suggesting new
‘methods’ for species identifications, but rather a different overall procedure of
dealing with large amounts of small troublesome taxa. Instead of dealing with
samples from one end to the other, we suggest discriminating between samples
of ‘low’ and ‘high value’, the latter to be dealt with in great details by specialist,
and with priority over ‘low’ value samples. High value samples are those which
contain lots of specimens. Particularly deep-sea collections often reveal many
species but few specimens and thus offer only few such targets for detailed
studies of inter-specific variation. However, due to the patchy distribution often
encountered in the deep sea, a few samples (maybe 1 in 100) will contain lots
of specimens and most often these will belong to one or two species. These
are the samples worth their weight in gold, and those species of which much
material exist should be examined (and the species described/redescribed)
in great detail, including dissections, illustrations and descriptions of several
individuals, of several developmental stages, and of both sexes. At the same
time specimens (males, females, juveniles, and mancae) should be processed
for molecular studies to verify con-specificity with absolute certainty. Since
most families have not been studied in such detail, these baseline studies are
needed to provide the detailed information required for processing other species
of the same phylogenetic groups but encountered in fewer numbers during the
specific survey. Once such a baseline study has been made, other members
of the same family can be processed ’normally’ by comparing the characters of
whatever instar or gender with the information provided by the baseline study.

304
If an adult female singleton is encounter, then it can be compared with an adult
female from the baseline study; if a manca is encounter, it can be compared with
a mancae from the baseline study and so forth. We will thus have the information
at hand which is needed for firstly correctly identifying the actual instar/gender,
and, secondly, for species identification knowing now which characters are
stable or not. These may well vary between higher taxonomical groups but are
likely to be similar (or more similar at least) within phylogenetically close groups.
We would like to end this paper with a note on descriptions of new taxa. The
senior author recently participated in a workshop regarding the description of
peracaridean crustacean. The participants received the following request by the
one of the organizers:
“We recently collected several thousand deep-sea species of which we
estimate half of them to be new to science. We would like to describe these
new species but it is a monumental task that we just don’t have the time for.
We would therefore like the participants of this workshop to come up with some
guidelines to how to describe ‘bulk’ new species in a short abbreviated and
timely fashion”.
After a short debate the participants unanimously came up with the only
possible answer:
“Please don’t do that!”.
While the person in charge of this overwhelming material only had good
intentions and was indeed faced with an impossible task, abbreviated descriptions
can only lead to chaos. If descriptions of such small and difficult creatures are to
have any value what so ever- now and in the future- it is absolutely paramount
that new species are described thoroughly and in minute details.

References
[1] K. Larsen, “Morphological and Molecular Investigation of Polymorphism and Cryptic Species
in Tanaid Crustaceans: Implications for Tanaid Systematics and Biodiversity Estimates”.
Zoological Journal of the Linnean Society, vol. 131(3), pp. 353–379, 2001.
[2] J. Sieg, “Evolution of Tanaidacea”. In: F. R. Schram (ed.) Crustacean Issues 1, Crustacean
Phylogeny, Rotterdam, 1983.
[3] T. Wolff, “Diversity and composition of deep-sea benthos”. Nature, London, vol. 267, pp. 780–
785, 1977.
[4] K. Larsen, Deep-Sea Tanaidacea (Peracarida) from the Gulf of Mexico. Crustaceana
Monographs. (Brill, Leiden), vol. 5, p. 387, 2005.
[5] U. M. Csaikl, M. Bastian, R. Brettscheider, S. Gauch, A. Meir, M. Schauerte, F. Scholz, C.
Sperisen, B. Vornam and B. Ziegenhagen, “Comparative analysis of different DNA extraction
protocols: A fast universal maxi preparation of high quality plant DNA for genetic evaluation and
phylogenetic studies”. Plant Molecular Biology Reporter, vol. 16(1), pp. 69–86, 1998.

305
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 307-313.
ISBN 978-88-8303-295-0. EUT, 2010.

Coffee species
and varietal identification
Patrizia Tornincasa, Michela Furlan, Alberto Pallavicini,
Giorgio Graziosi

Abstract — There are serious economical reasons to pretend warranties in

coffee species and varieties authenticity. Arabica adulteration with Robusta
coffees, intentional or not, is carried out at different steps of the coffee chain,
from plantation to beverage. We present a method based on a real-time PCR
technique to perform: a) a qualitative analysis to evaluate the presence/
absence of a species in a sample; b) a quantitative analysis to amplify Robusta
samples only, making possible the detection of less than 5% of Robusta in
a mixture. Arabica cultivars sold as specialty coffees can be identified and
certified by fingerprinting using SSR markers.

Index Terms — adulteration, Coffea arabica, Coffea canephora,

microsatellites, multiplex PCR, real time PCR, roasted coffee, traceability.

—————————— u ——————————

1 Introduction

C
offee is one of the most important products in the international market.
The annual consumption exceeds 5 billions kilograms, wich corresponds
to 500 billions cups. The genus Coffea contains more than 100 species,
only two of which, Coffea arabica (known as Arabica coffee) and C. canephora
(known as Robusta coffee) are commercially cultivated. Arabica produces
high quality coffee compared to Robusta, and contributes with about 70% of
the total world coffee production, being consequently sold at 2-3 times higher
prices. Also, among Arabica coffee some cultivars are considered as specialty
coffees with peculiar organoleptic characteristics and a very high commercial
value. Thus, there are serious economical reasons to pretend warranties in the
authenticity of coffee species and varieties. Arabica adulteration with Robusta
coffees can be intentional or not and is carried out at different steps of the
coffee chain, from plantation (one or both species can be cultivated by the same
producer) to coffee beverage.
Green coffee authentication can be very useful for roasters, while that of
roasted coffee (beans or ground) should be very interesting for retailers and
consumers.

————————————————
The authors are with the Department of Life Sciences, University of Trieste, P.le Valmaura, 9, I
34127 Trieste, Italy. E-mail of A.Pallavicini: [email protected].

307
 The methods to distinguish Arabica from Robusta in coffee blends are presently
based essentially on the chemical analysis of compounds such as sterols [1],
chlorogenic acid and caffeine [2], fatty acids [3], tocopherol [4], etc, but these
do not always give reliable results. Nowadays there is no totally reliable method
to guarantee coffee authenticity; the only reference is the production chain of
coffee, trusting on what the sellers declare through labels of already packed
products, or tasting the drink.
Concerning Arabica varieties, it is not possible to distinguish them from the
morphology of the seed or from plant phenotype and agronomy. The only
method to assess product quality wich is available for dealers is to roast the
seeds and taste the coffee beverage. Importers also cannot know whether the
small testing sample effectively corresponds to the many hundreds of coffee
bags received afterwards.
In recent years, food forensics requires DNA-based methods for molecular
analysis. The aim of this approach is to guarantee authenticity of commercially
important foods that can be contaminated accidentally or by fraud. Generally,
molecular tecniques based on DNA analysis are more effective and reliable than
those considering phenotipic characteristics. In particular molecular markers
such microsatellites or SSR (simple sequence repeats) are the most suitable for
their features: abundance in eucariotic genomes, high level of polymorphism,
codominance, locus specificity, PCR detection and high results reproducibility.
SSR are widely used in the characterisation of plant species such as rice [6],
potato [7], and wheat [8]
Research on coffee in this field is still at the beginning, and only one method
based on PCR-RFLP is available [5]. Here a Real-time PCR based method is
described for blend coffee analysis.

2 Materials and Methods

2.1 Discerning Arabica and Robusta

Real time PCR can be used to analyse coffee blends for establishing the
relative presence of Arabica and Robusta coffee. This method relies on DNA-
based probes which are complementary to target sequences in a region internal
to PCR primers. Each probe has a fluorescent reporter at one end and a
quencher of fluorescence at the opposite end of the probe. The close proximity
of the reporter to the quencher prevents detection of its fluorescence; breakdown
of the probe by the 5’ to 3’ exonuclease activity of the Taq polymerase breaks
the reporter-quencher proximity and thus allows unquenched emission of
fluorescence, which can be detected after excitation with a laser. An increase in
the product targeted by the reporter probe at each PCR cycle therefore causes
a proportional increase in fluorescence due to the breakdown of the probe and
release of the reporter.

308
2.2 Qualitative analysis of Arabica and Robusta blends

The method is based on the amplification of both Arabica and Robusta

samples through the use of Real Time PCR technology. The identification of
these two species is given by tree different probes, each one presenting a
specific fluorochrome: one is an universal probe, that recognizes both Arabica
and Robusta, index of amplification efficiency; a specific probe for C. arabica,
that binds a sequence present only in Arabica but absent in Robusta; a specific
probe for Robusta, that binds a piece of DNA present in C. canephora and absent
in Arabica. This method allows to carry out qualitative analysis to evaluate the
presence/absence of a species in a sample. Additionally, it can also amplify and
detect relative quantities of the two species in coffee blends.

2.3 Quantitative analysis of Arabica and Robusta blends

The second method permits only Robusta species amplification also in Arabica/
Robusta blends, through the use of Real Time PCR technology. The detection of
the DNA products is given by an universal probe, that recognizes both species.
In addition, to avoid undesired amplification of Arabica, a LNA (Locked Nucleic
Acid) oligonucleotide clamp was added. This clamp can hybridize in a DNA
region present only in Arabica but absent in Robusta, and does not permit
primer annealing. This method inhibits the amplification of the most abundant
species (usually Arabica) in favour of the less abundant one (Robusta) that can
be present in case of fraudulent contamination. Using this method it is possible
to give an estimation of the percentage of Robusta present in a mixture.

2.4 Arabica coffee fingerprinting

DNA was extracted from leaves and seeds of 320 different plants of Coffea
arabica that constitute the Arabica collection of Laboratory of Genetics
(University of Trieste, Italy).
Two multiplex PCR reactions were performed on each sample. The M1
reaction involves 9 couples of primers and allows to amplify at the same
time 9 microsatellite loci. The second multiplex, PCR M2, contains 7 couples
of primers and allows amplification of 7 loci microsatellites. Moreover, the
amplification products of M1 and M2 primers are studied to have amplicons with
non-overlapping molecular weights. Since this is not always possible, primers
are labelled with different fluorophores to distinguish amplification products with
similar size. This permits to mix both M1 and M2 products and to analyze them
in a single electrophoretic run through the genetic analyzer. The sequences on
which primers contained in M1 and M2 were designed are covered by patent
[9]. The advantages of this technique are: PCR easy to perform; possibility
to analyze 16 microsatellites using only 2 PCR reactions and one run by the
genetic analyzer, saving reagents, time and costs.

309
3 Results and Discussion

3.1 Qualitative and quantitative analysis of Arabica and Robusta blends

Two methods were developed to distinguish Arabica and Robusta species. The
qualitative approach permits immediate verification of the presence of Arabica,
Robusta or of both in a mixture. Furthermore, we can approximately estimate
the quantity of the two species up to 20 % of Robusta. Fig. 1A and Fig. 1B show
the specificity of the two fluorescent probes for Robusta and Arabica coffee,
respectively. The universal probe (indicated in Fig. 1A and 1B with number 1)
gives an amplification efficiency higher than the other two (numbered with 2 and
3).

Fig. 1A – Amplification of a Robusta sample. The amplicon detection is given by the

universal probe (1) and the C. canephora specific one (2), while the Arabica specific
probe (3) shows no signal.

Fig. 1B – Amplification of an Arabica sample; the detection in this case is achieved by

the universal probe (1) and the Arabica specific one (3), while the C. canephora specific
probe (2) gives no signal.

310
 The quantitative method was developed to amplify only Robusta samples,
making possible the detection of less than 5% of Robusta in a mixture. The
addition of the oligo clamp was required to inhibit Arabica amplification performed
by this system. Fig. 2 shows that increasing concentrations of this oligo clamp
progressively inhibit amplification of Arabica.

Fig. 2 – Chart showing amplification of Robusta and Arabica with the quantitative
system. Curves 1 and 2 represent Robusta and Arabica amplifications, respectively.
Curves numbered from 3 to 5 display progressive reduction of Arabica amplification
with addition of increasing oligo clamp amounts, in particular: 0,06 µM (3), 0,6 µM (4), 1
µM (5).

Conversely, the oligo clamp doesn’t interfere with Robusta amplification (data
not shown).
This method can be useful to detect a wide range of Robusta percentages that
can be present in a mixture.

3.2 Arabica coffee fingerprinting

Multiplex PCR reaction was performed successfully, giving a genetic profile

for each of the 320 Arabica samples, as shown in Fig. 3. Alleles were used
to calculate genetic distances and to design a cladogram showing genetic
relationships between the samples. These varieties, whose origin is known over
confidence, constitute a “genetic bank” that can be used to compare samples
with unknown origin.

311
 Fig. 3 – Example of sample amplification by Multiplex PCR.

4 Conclusions
All methods described in this paper can be successfully applied on green
and roasted coffee beans with DNA extraction and analysis protocols set up
in the Laboratory of Genetics of the University of Trieste. The main goals of
these analyses aim towards food traceability. The authenticity of Arabica,
Robusta or blends of the two species, is an important topic for producers and
customers. Several molecular markers are available to establish the origin of
coffee varieties for scientific aims, but none of them was used and validated for
commercial purpose so far. The possible applications are many: analysis of a
coffee stock proposed to a wholesaler, commercial coffee analysis to protect the
dealer from unfair competition (food traceability). The analysis can be also used
to determine the variety in a gourmet coffee lot.

312
References
[1] F. Carrera, M. Leon-Camacho, F. Pablos and A. G. Gonzalez, “Authentication of green coffee
varieties according to their sterolic profile” Anal. Chim. Acta, vol. 370, pp.131-139, 1998.
[2] M. J. Martin, F. Pablos and A. G. Gonzalez, “Discrimination between arabica and robusta
green coffee varieties according to their chemical composition” Talanta, vol. 46, pp. 1259-
1264, 1998.
[3] M. J. Martin, F. Pablos, A. G. Gonzalez, M. S. Valdenebro and M. Leòn-Chamacho, “Fatty
acid profiles as discriminant parameters for coffee varieties differentiation” Talanta, vol. 54,
pp. 291–297, 2001.
[4] G. N. Jham, J. K. Winkler, M. A. Berhow and S. F. Vaughn, “γ-Tocopherol as a marker of
Brazilian coffee (Coffea arabica L.) adulteration by corn” J. Agric. Food Chem., vol. 55, pp.
5995-5999, 2007.
[5] S. Spaniolas, S.T. May, M. J. Bennett and G. A. Tucker, “Authentication of Coffee by Means of
PCR-RFLP Analysis and Lab-on-a-Chip Capillary Electrophoresis” J. Agric. Food Chem, vol.
54(20), pp. 7466-7470, 2006.
[6] B. K. Chakravarthi and R. Naravaneni, “SSR marker based DNA fingerprinting and diversity
study in rice (Oryza sativa. L)” African J. Biotech., vol. 5(9), pp. 684-688, 2006.
[7] V. Ashkenazi, E. Chani, U. Lavi, D. Levy, J. Hillel and R. E. Veilleux, “Development of
microsatellite markers in potato and their use in phylogenetic and fingerprinting analyses”
Genome, vol. 44, pp. 50–62, 2001.
[8] M. M. Manifesto, A. R. Schlatter, H. E. Hopp, E. Y. Suárez and J. Dubcovsky, “Quantitative
Evaluation of Genetic Diversity in Wheat Germplasm Using Molecular Markers” Crop Science,
vol. 41, pp. 682-690, 2001.
[9] “Method for the discrimination between the varieties of Coffea arabica based on polymorphisms
of nuclear DNA” Patent n. *PD2008A000336*.

313
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 315-322.
ISBN 978-88-8303-295-0. EUT, 2010.

Mislabelling in megrims:
implications for conservation
Victor Crego-Prieto, Daniel Campo, Juliana Perez,
Eva Garcia-Vazquez

Abstract — Mislabelling of fish catch and commercial seafood products is

relatively frequent worldwide and can affect fisheries management exploitation
when stock estimates are based on landings. In this study we have analyzed
genetically 239 commercial lots of two morphologically similar species of
megrims (genus Lepidorhombus) that are caught together in mixed fisheries.
A high proportion of mislabelling was detected, suggesting enormous
underreported exploitation of one of the species, which can be endangered if
the problem persists. These results highlight the urgency of applying currently
available species-specific molecular tools in fisheries sciences for preventing
biodiversity losses in exploited species.

Index Terms — exploitation, genetic identification, megrims, mixed fisheries,

species-specific markers.

—————————— u ——————————

1 Introduction

I
ndustrialized fisheries typically reduce community biomass by 80% within 15
years of exploitation, and, as a consequence, large predatory fish biomass
today is only about 10% of pre-industrial levels [1]. Depletion of fish stocks
is due to many different factors, some of them anthropogenic. For example, due
to factors ranging from climate change [2] to pollution or overfishing, exploited
natural populations are in decline in many marine areas. Possible solutions
for environmental challenges fall out of the scope of this study. Solutions for
overfishing, however, exist and are relatively simple, although they may have
a short-term socioeconomic cost for the fishery sector. Some solutions really
work, as demonstrated by the recovery of Atlantic herring after its depletion in
the late 70s and further implementation of protective measures [3]. Restricted
fisheries effort and protection of spawning areas and juveniles are some of the

————————————————
V. Crego-Prieto is with the Department of Functional Biology, University of Oviedo,C/ Julian Clav-
eria s/n, 33003-Oviedo, Spain. E-mail: [email protected].
D. Campo is with the Department of Molecular and Computational Biology, University of Southern
California, 1050 Childs Way, RRI. Los Angeles (CA), 90089, USA. E-mail: [email protected].
J. Perez is with the Department of Natural Resources Conservation, Holdsworth Hall, Amherst MA
01003, USA.
E. Garcia-Vazquez is with the Department of Functional Biology, University of Oviedo,C/ Julian
Claveria s/n, 33003-Oviedo, Spain. E-mail: [email protected].

315
possible approaches for allowing natural stocks to recover. However, population
estimation techniques are not exact, leading to inaccurate estimates of stock
size. Eggs and larvae of different species with overlapping spawning areas are
often morphologically similar, and methods of species identification in addition
to visual identification are needed for accurate stock assessment [4], [5], [6].
The same problem exists when estimates of fishery effort are based on
reported catch data. In some cases, such as for sharks in Hong Kong markets,
high concordance between trade and specific names may allow the use of
market records for monitoring species-specific trends in trade and exploitation
rates [7], [8]. This method, however, cannot be generalized. Sometimes the
adults of two species caught simultaneously, for example in trawl fisheries, are
so similar that it is difficult to identify them so that mislabelling may occur, as
shown for example in hakes [9]. Once mislabelled at landing, the error persists
along the entire seafood chain to the consumer, who buys a marketed product
which does not correspond to the species marked on the label.
DNA variants can be revealed employing many different techniques. Here it is
almost impossible to describe all of them in detail, but some specific examples
of molecular techniques used for revealing species-specific variations in fish
are listed in Tab 1. They could be useful to fisheries if assessment of trade fish
products is used to estimate stock exploitation.
The aim of this study was to analyze in detail a case where application of
species-specific markers to fisheries science seems necessary and likely
urgent. Species misidentification was detected from landings to commercial
products, suggesting underreported exploitation of megrim species (genus
Lepidorhombus), whose exploitation rates are largely based on catch reports.
We also assessed the possible consequences of these likely inadvertent errors
for long-term sustainability of fish stocks.

2 Material and methods

2.1 Mixed-fisheries case studies

We have focused our study on two morphologically similar fish species that
are caught in different areas of the Atlantic Ocean. In Europe, the two species
of megrim, Lepidorhombus whiffiagonis (megrim) and L. boscii (four-spotted
megrim), are flatfishes of the Scophthalmidae family (Pleuronectiformes) having
overlapping distributions (Fig. 1).
As with other species, they are caught together in trawl fisheries. Most
landings correspond to Spain, followed by the UK, the two countries having
together approximately 70% of European catches. In the period 2000-2004,
catches of 58,180 tons of L. whiffiagonis were reported (FAO catch statistics,
available at http://www.fao.org/fishery/statistics/global-capture-production)
compared to only 40,187 tons of L. boscii (40.85 % of megrim catches). Little is
known about population structure and/or population size of megrims, although
the existence of a separate stock of L. whiffiagonis in the Mediterranean sea has
been demonstrated employing genetic markers [18] and differences in growth

316
among distributional areas have been described [19], [20].

2.2 Samples analyzed

Reference samples for each species (Tab. 2) were obtained in the context
of research cruises for the European Project MARINEGGS. The specimens
were obtained from at least four different locations covering roughly the
Atlantic distribution range of each species and identified by local experts in fish
taxonomy. A piece of gill or muscle tissue (about 3 g) was taken from each
specimen and stored in absolute ethanol. The reference samples are deposited
in the laboratory of the research team at the University of Oviedo (Spain).
Spain was chosen for the survey because it is the top country in megrim
fisheries (43% of total landings, FAO catch statistics 2008). Marketed products
of both megrim species, labelled with species names, were directly purchased
from landings in Asturias (North Spain) in 2004. A total of 239 landings were
analyzed. A piece of tissue (approx. 3 g) was taken from each sample and
stored in absolute ethanol until analysis.

2.3 DNA analyses

DNA extraction was carried out employing the resin Chelex [21]. For
identification of megrims we used differences in sequence length of the PCR-
amplified fragment of a conserved locus, the 5S rDNA coding for small ribosomal
RNA (5S rRNA), as the species-specific marker. The locus is composed of the
coding sequence, typically 120 base pairs (bp) long and highly conserved among
species, and the non transcribed spacer (NTS) which can differ in sequence
and length among closely related species. PCR amplification of the 5SrDNA
locus was carried out in a GeneAmp PCR system 9700 (Applied Biosystems),
employing the primers designed by Pendas et al. [22], in a total volume of 20
μl containing 0.5 μl GoTaq Polymerase at 5U/ml (Promega), 2μl of 10x Buffer,
2μl of 25 mM MgCl2, 2μl of dNTPs, 100 pmol of each primer and approximately
5 ng of genomic DNA. PCR amplification conditions were: initial denaturation
at 95oC for 5 min, then 35 cycles of denaturation at 95oC for 20 s, annealing at
65º for 20 s and extension at 72oC for 30 sec, and a final extension at 72oC for
20 min. When agarose methodology was employed, products were run in 2.5%
agarose gels at 100 V and visualized by staining with 2 μl ethidium bromide (10
mg/ml). The size of the amplified fragments was estimated by comparison with a
standard 100 bp DNA marker (Promega). In the Genetic Analyzer (Sequencing
Unit, University of Oviedo), fragment sizes were also directly visualized in a
chromatogram employing the GeneScan 3.7 Analysis Software (Applied
Biosystems).

317
3 Results

3.1 Species-specificity of the markers assayed

Species-specificity of the marker was confirmed for the two species studied.
All the individuals morphologically identified as belonging to a given species
yielded the same genetic pattern. The 5S rRNA locus amplification yielded two
DNA fragments of 435 and 217 base pairs (bp) for Lepidorhombus whiffiagonis
and two main fragments 331 and 233 bp long (plus some secondary heavier
shorter fragments) for L. boscii (Fig. 2, amplification fragments visualized in an
agarose gel).

3.2 Mislabelling in landings

A high level of mislabelling was found in the 239 commercial landings

analyzed. Although declared landings were 60% L. whiffiagonis and 40% L.
boscii, the actual proportion of each species was 49 and 51% for L. whiffiagonis
and L. boscii, respectively (Tab. 3).

4 Discussion
The 5S rDNA can be considered a good species-specific marker. It has
already been used for example in fishes like the Genus Leporinus [23], the
Sciaenidae family [24] or in shark species [25], and also in many other taxa.
The patterns obtained here for both megrims are in concordance with patterns
described for this species by Garcia-Vazquez et al. [26]. Mislabelling in these
megrim species is likely accidental, as they are morphologically similar and
often difficult to separate by visual inspection. The trade price is the same for
the two species, therefore intentional mislabelling for purposes of commercial
fraud cannot explain the detected differences between declared and real
commercialized species. Although inadvertent, this type of mislabelling could
however produce serious errors in fisheries assessments. If we assume that the
individuals analyzed are representative of landings, the divergence between
declared and actual catches would be thousands of tons of megrims (Fig. 3).
Figures corresponding to estimated “actual” catch can be obtained based on
species content in the commercial landings analysed (in percent), multiplied by
the total catch (in tons) of each species.
Another point to consider is the direction of mislabelling, which was deviated
incrementing the catch data of L. whiffiagonis in a high percentage, and so
decreasing the catch data corresponding to L. boscii. Underreported exploitation
of a species leads to overexploitation and, in the long term, to exhaustion of
stocks, fisheries decline and eventual extinction of the overexploited species
[27]. For purposes of fisheries management, these data should clearly be taken
into account.
Stock sizes are not estimated separately for this two species in annual
surveys in their respective area of occurrence. The two Lepidorhombus are not

318
genetically distinguished in routine plankton surveys, although there are recent
studies describing species-specific markers for these species [5] that clearly
demonstrate that visual identification is not accurate. L. whiffiagonis was the only
megrim species identified in Bay of Biscay plankton samples [28], and was also
the megrim species confounded with hake eggs in other plankton surveys [5].
Absence to date of genetically analyzed L. boscii in plankton samples could be
interpreted as a signal of its scarcity, but those studies were based on a limited
number of samples and cannot be taken as an indicator of real abundance of
that species.
Genetic identification of specimens in landings is even more important for
species like those studied in this work, whose production in aquaculture is not
forecasted at short-term. As demersal species, their cultivation is not easy.
For megrims, cultivation assays have not been carried out as far as we know.
Thus, although aquaculture seems to be a solution for obtaining seafood protein
at a global scale, as for other marine species [29], production of megrim at
commercial scale will likely rely on extractive fisheries in the forthcoming years.
Application of species-specific markers to fisheries science seems necessary
and likely urgent, and stock evaluation based on catch records will require
application of genetic markers for improving its utility for sustainable exploitation
of these valuable marine species.

5 Conclusion
DNA analysis revealed high percentage of mislabelling in megrim landings.
These results suggest underreported exploitation of four-spotted megrim L.
boscii, a species whose exploitation rates are largely based on catch reports
and which could become endangered if the problem persists. We highlight
the urgency of applying currently available species-specific molecular tools in
fisheries sciences.

Acknowledgements

We thank Paula Alvarez (AZTI, Spain), Francisco Sanchez (IEO, Santander, Spain)
and Placida Lopes (IPIMAR, Portugal) for providing megrim samples. This study was
supported by the FICYT project IB09-0023 (Asturias, Spain). Ivan Gonzalez Pola
provided help with laboratory analyses. Eva Garcia-Vazquez was a Grantee from the
Spanish Ministry of Research and Innovation (PR2008-0239) in 2008.

References
[1] R. A. Myers and B. Worm, “Rapid worldwide depletion of predatory fish communities”. Nature,
vol. 423, pp. 280-283, 2003.
[2] C. M. O’Brien, C. J. Fox, B. Planque et al., “Fisheries: climate variability and North Sea cod”.
Nature, vol. 404, p. 142, 2000.
[3] J. A. Hutchings, “Collapse and recovery of marine fishes”. Nature 406, pp. 882-885, 2000.
[4] C. J. Fox, M. I. Taylor, R. Pereyra et al., “TaqMan DNA technology confirms likely overestimation
of cod (Gadus morhua L.) egg abundance in the Irish Sea: implications for the assessment of
the cod stock and mapping of spawning areas using egg-based methods”. Molecular Ecology,
vol. 14(3), pp. 879–884, 2005.

319
[5] J. Perez, P. Alvarez, J. L. Martinez et al., “Genetic identification of hake and megrim eggs in
formaldehyde-fixed plankton samples”. ICES Journal of Marine Science 62, 908-914, 2005a.
[6] M. Kochzius, M. Nölte, H. Weber et al., “DNA Microarrays for Identifying Fishes”. Marine
Biotechnology, vol. 10(2), pp. 207-217, 2008.
[7] D. L. Abercrombie, S. C. Clarke and M. S. Shivji, “Global-scale genetic identification of
hammerhead sharks: Application to assessment of the international fin trade and law
enforcement”. Conservation Genetics, vol. 6, pp. 775-788, 2005.
[8] S. C. Clarke, M. K. McAllister, E. J. Milner-Gulland et al., “Global estimates of shark catches
using trade records from commercial markets”. Ecology Letters, vol. 9, pp. 1115-1126, 2006.
[9] G. Machado-Schiaffino, J. L. Martinez and E. Garcia-Vazquez, “Detection of mislabeling
in hake seafood employing mtSNPs-based methodology with identification of eleven hake
species of the genus Merluccius”. Journal of Agriculture and Food Chemistry, vol. 56(13), pp.
5091-5095, 2008.
[10] F. Teletchea, “Molecular identification of fish species; reassessment and possible applications”.
Reviews in Fish Biology and Fisheries (on line first DOI 10.1007 1160-009-9107-4), 2009.
[11] D. Blohm, F. Bonhomme, G. Carvalho et al., “Assessment of tools for identifying the genetic
origin of fish and monitoring their occurrence in the wild”. In: T. Svåsand, D. Crosetti, E.
Garcia-Vazquez and E. Verspoor (eds.), Genetic impact of aquaculture activities on native
populations (Genimpact final scientific report, EU contract n. RICA-CT-2005-022802). http://
genimpact.imr.no/, pp. 128-134, 2007. Accessed 12 March 2009.
[12] M. M. Ferguson and R. G. Danzmann, “Role of genetic markers in fisheries and aquaculture:
useful tools or stamp collecting?” Canadian Journal of Fisheries and Aquatic Sciences, vol.
55(7), 1553-1563, 1998.
[13] Z. J. Liu and J. F. Cordes, “DNA marker technologies and their applications in aquaculture
genetics”. Aquaculture, vol. 238, pp. 1-37, 2004.
[14] R. S. Rasmussen and M. T. Morrissey, “DNA-based methods for the identification of commercial
fish and seafood species”. Comprehensive Reviews in Food Science and Food Safety, vol. 7,
pp. 280-295, 2008.
[15] F. Aranishi, T. Okimoto and S. Izumi, “Identification of gadoid species (Pisces, Gadidae) by
PCR-RFLP analysis”. Journal of Applied Genetics, vol. 46(1), pp. 69-73, 2005.
[16] M. I. Taylor, C. Fox, I. Rico and C. Rico, “Species-specific TaqMan probes for simultaneous
identification of (Gadus morhua L.), haddock (Melanogrammus aeglefinus L.) and whiting
(Merlangius merlangus L.)”. Molecular Ecology Notes, vol. 2(4), pp. 599-601, 2002.
[17] J. E. Magnussen, E. K. Pikitch, S. C. Clarke et al., “Genetic tracking of basking shark products
in international trade”. Animal Conservation, vol. 10(2), pp. 199-207, 2007.
[18] E. Garcia-Vazquez, J. I. Izquierdo and J. Perez, “Genetic variation at ribosomal genes
supports the existence of two different European subspecies in the megrim Lepidorhombus
whiffiagonis”. Journal of Sea Research, vol. 56, pp. 59-64, 2006a.
[19] J. Landa and C. Piñeiro, “Megrim (Lepidorhombus whiffiagonis) growth in the North-eastern
Atlantic based on back-calculation of otolith rings”. ICES Journal of Marine Science, vol. 57(4),
pp. 1077-1090, 2000.
[20] J. Landa, N. Perez and C. Piñeiro, “Growth patterns of the four spot megrim in the northeast
Atlantic”. Fisheries Research, vol. 55, pp. 141-152, 2002.
[21] A. Estoup, C. R. Largiader, E. Perrot. et al., “Rapid one-tube DNA extraction protocol for
reliable PCR detection of fish polymorphic markers and transgenes”. Molecular Marine Biology
and Biotechnology, vol. 5(4), pp. 295-298, 1996.
[22] A. M. Pendas, P. Moran, J. L. Martinez and E. Garcia-Vazquez, “Applications of 5S rDNA
in Atlantic salmon, brown trout, and in Atlantic salmon x brown trout hybrid identification”.
Molecular Ecology, vol. 4, pp. 275-276, 1995.
[23] I. A. Ferreira, C. Oliveira, P. C. Venere, P. M. Galetti Jr and C. Martins, “5S rDNA variation and
its phylogenetic inference in the genus Leporinus (Characiformes: Anostomidae)”. Genetica,

320
 vol. 129(3), pp. 253-257, 2006.
[24] F. A. Alves-Costa, C. Martins, F. Del Campos de Matos, F. Foresti, C. Oliveira and A. P. Wasko,
“5S rDNA characterization in twelve Sciaenidae fish species (Teleostei, Perciformes): depicting
gene diversity and molecular markers”. Genetic Molecular Biology, vol. 31(1) Suppl. 0, 2008.
[25] D. Pinhal, O. B. F. Gadig, A. P. Wasko, C. Oliveira, E. Ron, F. Foresti and C. Martins,
“Discrimination of shark species by simple PCR of 5S rDNA repeats”. Genetic Molecular
Biology, vol. 31(1), pp. 361-365, 2008.
[26] E. Garcia-Vazquez, J. I. Izquierdo and J. Perez, Genetic variation at ribosomal genes
supports the existence of two different European subspecies in the megrim Lepidorhombus
whiffiagonis”. Journal of Sea Research, vol. 56, pp. 59-64, 2006b.
[27] D. J. Agnew, J. Pearce, G. Pramod, T. Peatman, R. Watson, J. R. Beddington and T. J. Pitcher,
“Estimating the Worldwide Extent of Illegal Fishing”. PLoS One www.plosone.org, vol. 4(2),
pp. 1-8, 2009.
[28] E. Garcia-Vazquez, P. Alvarez, P. Lopes. et al., “PCR-SSCP of the 16S rRNA gene, a simple
methodology for species identification of fish eggs and larvae”. Scientia Marina, vol. 70 (Suppl.
2), pp. 13-21, 2006c.
[29] R. Goldburg and R. Naylor, “Future seascapes, fishing, and fish farming”. Frontiers in Ecology
and the Environment, vol. 3, pp. 21-29, 2005.
[30] F. Aranishi, T. Okimoto and S. Izumi, “Identification of gadoid species (Pisces, Gadidae) by
PCR-RFLP analysis”. Journal of Applied Genetics, vol. 46(1), pp. 69-73, 2005.
[31] L. Asensio, I. González, M. A. Rodríguez, B. Mayoral, I. López-Calleja, P. E. Hernández, T.
García and R. Martín, “Identification of grouper (Epinephelus guaza), wreck fish (Polyprion
americanus), and Nile perch (Lates niloticus) fillets by polyclonal antibody-based enzyme-
linked immunosorbent assay”. Journal of Agriculture and Food Chemistry, vol. 51(5), pp. 1169-
1172, 2003.
[32] L. Asensio, I. González, M. A. Pavón, T. García and R. Martín, “An indirect ELISA and a PCR
technique for the detection of Grouper (Epinephelus marginatus) mislabeling”. Food Additives
and Contamination, vol. 25(6), pp. 677-683, 2008.
[33] E. Carrera, T. García, A. Céspedes et al., “Differentiation of smoked Salmo salar, Oncorhynchus
mykiss and Brama raii using the nuclear marker 5S rDNA”. International Journal of Food
Science and Technology, vol. 35, pp. 401-406, 2000.
[34] A. G. F. Castillo, J. L. Martinez and E. Garcia-Vazquez, “Identification of Atlantic hake
species by a simple PCR-based methodology employing microsatellite loci”. Journal of Food
Protection, vol. 66(11), pp. 2130-2134, 2003.
[35] M. Carrera, B. Cañas, C. Piñeiro, J. Vázquez and J. M. Gallardo, “Identification of commercial
hake and grenadier species by proteomic analysis of the parvalbumin fraction”. Proteomics ,
vol. 6(19), pp. 5278-5287, 2006.
[36] M. Carrera, B. Cañas, C. Piñeiro, J. Vázquez and J. M. Gallardo, “De novo mass spectrometry
sequencing and characterization of species-specific peptides from nucleoside diphosphate
kinase B for the classification of commercial fish species belonging to the family Merlucciidae”.
Journal of Proteome Research, vol. 6(8), pp. 3070-3080, 2007.
[37] M. J. Chapela, A. Sánchez, M. I. Suárez, R. I. Pérez-Martín and C. G. Sotelo, “A rapid
methodology for screening hake species (Merluccius spp.) by single-stranded conformation
polymorphism analysis”. Journal of Agriculture and Food Chemistry, vol. 55(17), pp. 6903-
6909, 2007.
[38] Z. Hubalkova, P. Kralik, J. Kasalova and E. Rencova, “Identification of gadoid species in fish
meat by polymerase chain reaction (PCR) on genomic DNA”. Journal of Agriculture and Food
Chemistry, vol. 56(10), pp. 3454-3459, 2008.
[39] D. F. Hwang, H. C. Jen, Y. W. Hsieh and C. Y. Shiau, “Applying DNA techniques to the
identification of the species of dressed toasted eel products”. Journal of Agriculture and Food
Chemistry, vol. 52(19), pp. 5972-5977, 2004.

321
[40] R. Murgia, G. Tola, S. N. Archer, S. Vallerga and J. Hirano, “Genetic identification of grey
mullet species (Mugilidae) by analysis of mitochondrial DNA sequence: application to identify
the origin of processed ovary products (bottarga)”. Marine Biotechnology, vol. 4(2), pp. 119-
126, 2002.
[41] J. Perez and E. Garcia-Vazquez, Genetic identification of nine hake species for detection of
commercial fraud. Journal of Food Protection, vol. 67, 2792-2796, 2004.
[42] M. Perez, J. M. Vieites and P. Presa, “ITS1-rDNA-based methodology to identify world-wide
hake species of the Genus Merluccius”. Journal of Agriculture and Food Chemistry, vol.
53(13), pp. 5239-5247, 2005b.
[43] C. Piñeiro, J. Barros-Velázquez, R. I. Pérez-Martín et al., “Development of a sodium dodecyl
sulfate-polyacrylamide gel electrophoresis reference method for the analysis and identification
of fish species in raw and heat-processed samples: a collaborative study”. Electrophoresis,
vol. 20(7), pp. 1425-1432, 1999.
[44] C. Piñeiro, J. Vázquez, A. I. Marina, J. Barros-Velázquez and J. M. Gallardo, “Characterization
and partial sequencing of species-specific sarcoplasmic polypeptides from commercial hake
species by mass spectrometry following two-dimensional electrophoresis”. Electrophoresis,
vol. 22(8), pp. 1545-1552, 2001.
[45] I. Rodushkin, T. Bergman, G. Douglas, E. Engström, D. Sörlin and D. C. Baxter, “Authentication
of Kalix (N.E. Sweden) vendace caviar using inductively coupled plasma-based analytical
techniques: evaluation of different approaches”. Analytica Chimica Acta, vol. 583(2), pp. 310-
318, 2007.
[46] P. Sebastio, P. Zanelli and T. M. Neri, “Identification of anchovy (Engraulis encrasicholus L.) and
gilt sardine (Sardinella aurita) by polymerase chain reaction, sequence of their mitochondrial
cytochrome b gene, and restriction analysis of polymerase chain reaction products in
semipreserves”. Journal of Agriculture and Food Chemistry, vol. 49(3), pp. 1194-1199, 2001.

322
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 323-326.
ISBN 978-88-8303-295-0. EUT, 2010.

Seeds in subtribe Orchidinae

(Orchidaceae): the best
morphological tool to support
molecular analyses
Roberto Gamarra, Emma Ortúñez, Ernesto Sanz,
Iris Esparza, Pablo Galán

Abstract — Seeds of the genera belonging to the subtribe Orchidinae

(Orchidaceae) have been studied using scanning electron microscopy.
Qualitative data concerning the general morphology of the seed and testa
cells, and ornamentation of periclinal and anticlinal walls, are characters that
allow to recognize genera and taxa at infrageneric levels. The results show
that seed micromorphology is the best tool to support the molecular analyses
published in recent years for this taxonomic group.

Index Terms — qualitative data, seed micromorphology, SEM, subtribe

Orchidinae.

—————————— u ——————————

1 Introduction

T
he family Orchidaceae is probably the largest among flowering plants,
with 25.158 species [1]. The systematics has undergone many changes
along the last few decades. The latter taxonomic proposals were published
by Dressler [2] and Szlachetko [3]. In the subfamily Orchidoideae, Dressler [2]
divided the tribe Orchideae into two subtribes: Orchidinae with 34 genera and
370 species, and Habenariinae with 23 genera and 930 species. Within this
tribe, Philip Cribb [4] recognizes 62 genera and near 1800 species. This is the
principal orchid group of the north temperate area, also with considerable
diversity in Africa, South Asia and Australia; only the genus Habenaria occurs in
South America.
Flower morphology (lip, anther, stigma, rostellum) and vegetative characters
(habitus, inflorescence, tuberoids) have been used to elucidate the systematics
of this tribe. According to Dressler [2], the morphology of tuberoids is essential
————————————————
R. Gamarra, E. Ortúñez, E. Sanz and I. Esparza are with the Departamento de Biología, Univer-
sidad Autónoma de Madrid, C/ Darwin, 2, E-28049 Madrid, Spain. E-mail: [email protected].
P. Galán is with the Departamento de Producción Vegetal: Botánica y Protección Vegetal, E.U.I.T.
Forestal, Universidad Politécnica, E-28040 Madrid, Spain, E-mail: [email protected].

323
to recognise the genera in the subtribe Orchidinae, while in Habenariinae the
stigmas are more important. Szlachetko [3] proposes 6 subtribes (Orchidinae,
Herminiinae, Bartholininae, Androcorytinae, Platantherinae, Habenariinae)
based on stigma and rostellum morphology.
Recently, molecular analyses have changed the taxonomy of several genera
and species in this tribe [5], [6], e.g., the monotypic genus Coeloglossum Hartm.
is integrated into Dactylorhiza Neck. ex Nevski, while some species attributed to
the genus Orchis L. are included into Anacamptis Rich. and Neotinea Rchb. fil.
Beer [7] published the first study about the seed morphology in Orchidaceae.
In his book, the figures show the great diversity in genera belonging to different
subfamilies.
Clifford & Smith [8] proposed the first methodology to analyze the testa
morphology studying 49 species of the subfamily Epidendroideae. They showed
a strong correlation between qualitative characters of seeds and taxa above
genus level.
Later, Barthlott [9] studied 58 genera using SEM, demonstrating the great
diagnostic value of seed morphology and its phylogenetic significance, principally
at tribe and subtribe level. He also indicated that this is a useful taxonomic tool
to recognise the genera. In 1979, Joseph Arditti and colleagues established
the methodology for quantitative analyses, related to the sizes and volumes of
seeds and embryos [10]. Several authors published different papers about seed
morphology in the family Orchidaceae [11], [12], [13]. All these papers show the
great diagnostic value of the qualitative and quantitative characters of the seed,
to approach phylogenetic studies in this family. Arditti & Al-Ghani [14] published
an overview of previous works, and they conclude the importance to continue
this research.
More recently. Krishna Swamy et al. [15] described seeds of the genus
Cymbidium using SEM and morphometric data, and Tsutsumi et al. [16] compared
the phylogenetic proposal using molecular markers with seed morphology of the
japanese species of the genus Liparis.
Different authors have obtained similar results in other flowering plants, as
in the genus Phyllocladus (Phyllocladaceae) [17], in the tribe Massonieae
(Hyacinthaceae) [18], in the genus Veronica (Plantaginaceae) [19], and in
the genus Moehringia (Caryophyllaceae) [20]. These papers show that seed
morphology is an important tool to elucidate the taxonomy of these genera.
Since 2003, our research group initiated a study of seed micromorphology of
iberian orchids using SEM and the methodology proposed by former authors.
We have published data about genera of subfamilies Cypripedioideae [21]
and Orchidoideae [22], [23], some of them previously presented in the 17th
International Botanical Congress, held in Wien in 2005. Our studies show
that qualitative and quantitative characters strongly support the results using
molecular markers. These characters are of good diagnostic value to recognize
many taxa, principally above the species level.
Presently, the main aim of our research is the study of the seed micromorphology
in all groups within the genera of the tribe Orchideae.

324
2 Results
In the studied genera, the seed morphology varies from elongate fusiform
(Dactylorhiza, Platanthera, Neotinea, Anacamptis), to shortly fusiform and
almost ovoid (Gymnadenia, Pseudorchis, Amitostigma). Generally, in the
elongate fusiform, the medial cells are longer than apical and basal cells; in the
other morphologycal type, are similar or slightly longer.
The apical pole mainly consist of short and polygonal cells. Only the genus
Platanthera finished in a truncated cell.
The chalazal end is opened, with short and polygonal cells. Exclusively, in the
genus Ophrys, a distinct asymmetry is showed.
In several genera (Orchis, Pseudorchis, Gymnadenia), the periclinal walls
are unsculptured. However, many genera present a type of ornamentation,
with prominent and spaced ridges (Ophrys), to slight ondulations (Serapias,
Amerorchis, Comperia). The distribution of ridges and ondulations varies from
lax (Platanthera, Himantoglossum) to dense (Serapias, Steveniella, Comperia),
and from transversal (Neotinea, Steveniella) to oblique (Anacamptis,
Himantoglossum, Aorchis). Only the genus Dactylorhiza shows a great variation,
from unsculptured periclinal walls (D. incarnata group) to different types of
ornamentation (D. maculata group, D. majalis group).
The morphology of the anticlinal walls varies from straight (Platanthera,
Himantoglossum, Aorchis) to undulate (Gymnadenia, Orchis p.p., Anacamptis
p.p.) The las type is more typical in the cells of apical pole. Also, a distinct type
of lamella can be found in these walls (Ophrys).

3 Conclusion
The morphological study, including qualitative and quantitative characters, of
the seed coat in the genera of the subtribe Orchidinae has showed that each
genus and each subgroup within this, have its own morphological type. Each
type fully agrees with the clades obtained in the recently published molecular
analyses [5], [6]. For example, it is consistent with the inclusion of the genus
Coeloglossum with the species of the Dactylorhiza incarnata group, Nigritella
into Gymnadenia or Barlia into Himantoglossum. Also, within genera with many
species as Anacamptis, Orchis or Dactylorhiza, each clade is consistent with
each morphological seed type. Likewise, it also supports the monophyly of
genera such as Ophrys and Serapias.

Acknowledgement

We are much indebted to Esperanza Salvador, Enrique and Isidoro, the technical staff
in the SEM laboratory at SIDI-UAM. To Jeff Wood and Mauricio Velayos, the curators
of Royal Botanic Gardens at Kew and Madrid, respectively. We are also grateful to all
orchidologists who sent us seeds of different genera.

325
References
[1] P. Cribb and R. Govaerts, “Just how many orchids are there”, Proc. 18th World Orchid
Conference, pp. 161-172, 2005.
[2] R. Dressler, Phylogeny and classification of the orchid family, Cambridge, Cambridge
University Press, 1993.
[3] D. Szlachetko, “Systema Orchidalium”, Fragm. Florist. Geobot., Suppl. 3, pp. 1-152, 1995.
[4] A. M. Pridgeon, P. Cribb, M. W. Chase and F.N. Rasmussen, “Genera Orchidacearum. 2.
Orchidoideae (Part 1)”, Oxford, Oxford University Press, 2001.
[5] A. M. Pridgeon, R. M. Bateman, A. V. Cox, J. R. Hapeman and M. W. Chase, “Phylogenetics
of subtribe Orchidinae (Orchidoideae, Orchidaceae) based on nuclear ITS sequences. 1.
Intergeneric relationships and polyphyly of Orchis sensu lato”. Lindleyana, vol. 12, pp. 89-109,
1997.
[6] R. M. Bateman, P. M. Hollingsworth, J. Preston, L. Yi-Bo, A. M. Pridgeon and M. W.
Chase, “Molecular phylogenetics and evolution of Orchidinae and selected Habenariinae
(Orchidaceae)”, Bot. J. Linn. Soc., vol. 142, pp. 1-40, 2003.
[7] J. G. Beer, Beiträge zur Morphologie und Biologie der familie der orchideen, Vienna, Druck
und Verlag von Carl Gerold’s Sohn, 1863.
[8] H. T. Clifford and W. K. Smith, “Seed morphology and classification of Orchidaceae”,
Phytomorphology, vol. 19, pp. 133-139, 1969.
[9] W. Barthlott, “Morphologie der Samen von Orchideen im Hinblick auf taxonomische und
funktionelle Aspekte”, Proc. 8th World Orchid Conference, pp. 444-455, 1976.
[10] J. Arditti, J.D. Michaud and P. L. Healey, “Morphometry of orchid seeds. I. Paphiopedilum and
native California and related species of Cypripedium”, Amer. J. Bot., vol. 66, pp. 1128-1137,
1979.
[11] H. Tohda, “Seed morphology in Orchidaceae I. Dactylorchis, Orchis, Ponerorchis, Chondradenia
and Galeorchis”, Sci. Rep. Tohoku Univ., 4th ser., Biology, vol. 38, pp. 253-268, 1983.
[12] H. Kurzweil, “Seed morphology in Southern African Orchidoideae (Orchidaceae)”, Pl. Syst.
Evol., vol.185, pp. 229-247, 1993.
[13] M. Molvray and P. J. Kores, “Character analysis of the seed coat in Spiranthoideae and
Orchidoideae, with special reference to the Diurideae (Orchidaceae)”. Amer. J. Bot., vol. 82,
pp. 1443-1454, 1995.
[14] J. Arditti and A. K. Al-Ghani, “Numerical and physical properties of orchid seeds and their
biological implications”, New Phytologist, vol. 145, pp. 367-421, 2000.
[15] K. Krishna Swamy, H. N. Krishna Kumar, T. M. Ramakrishna and S. N. Ramaswamy, “Studies
on seed morphometry of epiphytic orchids from Western Ghats of Karnataka”, Taiwania, vol.
49, pp. 124-140, 2004.
[16] C. Tsutsumi, T.  Yukawa, N.  Lee, C.  Lee and M.  Kato, “Phylogeny and comparative seed
morphology of epiphytic and terrestrial species of Liparis (Orchidaceae) in Japan”, J. Pl. Res.,
vol. 120, pp. 405-412, 2007.
[17] A. V. Bobrov, A. P. Melikian and E. Y. Yembaturova, “Seed morphology, anatomy and
ultrastructure of Phyllocladus L. C. and A. Rich. ex Mirb. (Phyllocladaceae (Pilg.) Bessey) in
connection with the generic system and Phylogeny”, Ann. Bot., vol. 83, pp. 601-618, 1999.
[18] W. Pfosser, W. Wetschnig, S. Ungar and G. Prenner, “Phylogenetic relationships among
genera of Massonieae (Hyacinthaceae) inferred from plastid DNA and seed morphology”. J.
Pl. Res., vol. 116, pp. 115-132, 2003.
[19] L. M. Muñoz-Centeno, D. C. Albach, J. A. Sánchez-Agudo and M. M. Martínez-Ortega,
“Systematic significance of seed morphology in Veronica (Plantaginaceae): a phylogenetic
perspective”. Ann. Bot., vol. 98, pp. 335-350, 2006.
[20] L. Minuto, S. Fior, E. Roccotiello and G. Casazza, “Seed morphology in Moehringia L. and its
taxonomic significance in comparative studies within the Caryophyllaceae”, Pl. Syst. Evol., vol.
262, pp. 189-208, 2006.
[21] E. Ortúñez, E. Dorda, P. Galán and R. Gamarra, “Seed micromorphology in the iberian
Orchidaceae. I. Subfamily Cypripedioideae”, Bocconea, vol. 19, pp. 271-274, 2006.
[22] R. Gamarra, E. Dorda, A. Scrugli, P. Galán and E. Ortúñez, “Seed micromorphology in the
genus Neotinea Rchb. f. (Orchidaceae, Orchidinae)” Bot. J. Linn. Soc., vol. 153, pp. 133-140,
2007.
[23] R. Gamarra, P. Galán, I. Herrera and E. Ortúñez, “Seed micromorphology supports the splitting
of Limnorchis from Platanthera (Orchidaceae)”, Nord. J. Bot., vol. 26, pp. 61-65, 2008.

326
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 327-331.
ISBN 978-88-8303-295-0. EUT, 2010.

Lentils biodiversity:
the characterization
of two local landraces
Vincenzo Viscosi, Manuela Ialicicco, Mariapina Rocco,
Dalila Trupiano, Simona Arena, Donato Chiatante, Andrea Scaloni,
Gabriella Stefania Scippa

Abstract — A multi-disciplinary approach was used to characterize two

autochthonous lentil landraces from Molise region (Central Italy). Different
mature seed populations for each landrace were provided by the Molise
Germoplasm Bank at the University of Molise (Pesche, Italy), and analyzed
at the morphological and molecular (DNA and protein) levels. Nuclear ISSR
markers were used to assess genetic differences, whereas phenotypic
variability was detected by biochemical (proteomics) and morphological
analyses. The genetic and phenotypic diversity of the two lentil landraces
were well assessed in relation to their geographical provenance, supporting
further studies to identify landrace markers.

Index Terms — ISSR markers, Lens culinaris, seed morphology, proteomics.

—————————— u ——————————

1 Introduction

L
ens culinaris Medik. has been cultivated around the Mediterranean basin
since at least the seventh century B.C. and its cultivation area expanded
to Middle East, Ethiopia and the Indian Subcontinent ([1], [2]). Local
landraces are characterized by high genetic variability and high adaptation
to different environmental conditions evolving in adaptive gene complexes
[3]. However, in the industrialized countries the cultivation of many different
local landraces has progressively decreased, becoming at high risk of genetic
erosion ([4], [5]). In this paper, the diversity of two lentil landraces is analyzed to
define and quantify differences between groups of populations coming from two
geographical areas of Molise region (Central Italy). In particular, this study aims
————————————————
V. Viscosi, M. Ialicicco, D. Trupiano and G. S. Scippa are with the Department S.T.A.T. of Univer-
sity of Molise, Pesche, I-86090. E-mail: [email protected], [email protected],
[email protected], [email protected].
M. Rocco is with the Department of Biological and Environmental Sciences, University of Sannio,
Benevento, Italy. E-mail: [email protected].
S. Arena and A. Scaloni are with the Proteomics and Mass Spectrometry, ISPAAM, National Re-
search Council, Naples, Italy.
D. Chiatante is with the Department S.C.A., University of Insubria, Como, Italy.

327
to deepen the knowledge about morphological, genetic and proteomic markers
that differentiate local lentil landraces in relation to their provenance.

2 Material and Methods

Twelve lentil seed populations from Molise region (Central Italy) were
studied in relation to their provenance: five from Conca Casale and seven
from Capracotta. The total sample was analyzed by a multi-disciplinary
approach, useful to characterize the phenotypical and genotypical traits of the
two landraces. For the genetic analysis, nuclear ISSR markers were used as
described in a previous paper [3], whereas the biochemical investigation was
carried out on total seed proteins extracted according to the method of Rabilloud
and resolved by 2DE [3]. For each population, gels were run in triplicate, and
the mean value computed by software-assisted (PD-Quest) analysis was used
to obtain a standardized matrix of the abundance (relative volume) of protein
spots. One hundred seeds of each landrace population were used to measure
eight morphological traits: area, perimeter, major axis length, minor axis length,
and roundness, 100-seed weight, 100-seed volume and density (g/ml).

2.1 Statistical analysis

For molecular data, Principal Component Analysis was computed on Nei’s

genetic distance (1972) and the hierarchical partition of genetic variation among
and within populations was obtained by means of the analysis of molecular
variance (AMOVA). For biochemical and morphological data, a standardized
matrix was subjected to univariate (ANOVA) and multivariate statistical analysis.
Principal Component Analysis was computed on significant variables (detected
by ANOVA) and the extracted Principal Components (eigenvalues > 1) were
used in Canonical Variate Analysis.

3 Results
The results of ISSR analysis pointed out the genetic relationship between the
two landraces. As shown in Fig. 1, the PCA highlighted a clear separation of the
populations sampled in Conca Casale and Capracotta. In particular, along the
first two PCs the total variance accounted for 45.57% and 18.04%, respectively.
Molecular degree of differentiation between the two groups of populations
(AMOVA) showed a significant molecular discrimination (PhiPT = 0.438; p =
0.001). Moreover, it resulted that the genetic variability was greater within (56%)
than among (44%) groups of landrace populations.
The comparison of total seed proteomic maps of the Conca Casale and
Capracotta populations revealed a total of 193 differentially expressed proteins.
The biochemical data set (193 proteins) was subjected to ANOVA, to identify
biochemical markers useful to distinguish lentil populations from different
provenances. It resulted that 25 proteins were significant to discriminate
Capracotta from Conca Casale lentils. PCA was computed on a correlation
matrix, using these 25 significant proteins; the first two PCs explained 53.79%

328
and 11.62% of total variance, respectively, and the scatter plot of these two PCs
indicated a clear distinction between the two groups of lentils (Fig. 2). Differences
between the two landraces were tested by canonical variate analysis (CVA)
computed on the extracted PCs. They were significantly discriminated (Wilks’
λ = 0.028; df=5; p< 0.0001) as shown by the test of cross-validation (100% of
cases were correctly classified).
The eight morphological variables were subjected to ANOVA in relation to
population provenance. The two groups of lentil populations from Capracotta
and Conca Casale were significantly discriminated by six morphological traits:
seed density, roundness, volume, major axis length, perimeter and minor axis
length. These six variables were used to compute a PCA: respectively, PC1
and PC2 explained 80.45% and 18.30% of total variance, highlighting a clear
separation between landraces (Fig. 3). Then, the PCs were used in CVA and
results indicated significant differences between the two population groups
(Wilks’ λ = 0.047; df = 2; sig.< 0.0001). Moreover, the test of cross-validations
showed a high significance of the CVA reporting that 100% of cases was
correctly classified.

Fig. 1 – Scatter plot of specimens (cross = Conca Casale; point = Capracotta) ordered
along the first two principal components; PCA from molecular data.

Fig. 2 – Scatter plot of specimens (cross = Conca Casale; point = Capracotta) ordered
along the first two principal components; PCA computed on 25 proteins.

329
 Fig. 3 – Scatter plot of specimens (cross = Conca Casale; point = Capracotta) ordered
along the first two principal components; PCA computed on six morphological variables.

4 Conclusion
Autochthonous plant germplams, characterized by a wide genetic variability
and high adaptation to different environmental conditions, are often more
subjected to genetic erosion risks. In Italy, several different lentil landraces
evolved thanks to the combination of different geographical characteristics.
The literature reports a wide variety of methods that have been used to
investigate genetic similarities and relations among landraces of L. culinaris
Medik.
Different methods have different powers of genetic resolution and provide
different information: neutral DNA markers are useful tools to describe genetic
relations in terms of time divergence [6], whereas phenotypic markers can provide
information about adaptive responses to macro-environmental conditions [7].
In this study we used a combination of genetic and phenotypic analyses to
characterize two autochthonous lentil landraces of two different provenances
within a small region such as Molise.
The integration of genetic markers analysis with seed morphology and
proteomic traits provided a high resolution approach to dissect lentil biodiversity
[3]. The diversity between groups of populations, coming from two very close
geographical areas, was well assessed and quantified. In addition, differences
between the two local landraces were principally related to their sites of origin,
where climate conditions and human activity may have selected the local
accessions characterised by specific morphological and biochemical traits of
seeds. Work is in progress to deepen the relation between these phenotypic
markers and the environmental characteristics of the landrace provenance
areas, and to identify the seed proteome markers.

References
[1] G. Ladizinsky, “The origin of lentil and its wild genepool”, Euphytica, vol. 28, pp. 179-187, 1979.
[2] Y. Duran, R. Fratini, P. Garcia and M. Perez de la Vega, “An intersubspecific genetic map of

330
 Lens.”, Theor. Appl. Genet., vol. 108, pp. 1265-1273, 2004.
[3] G. S. Scippa, D. Trupiano, M. Rocco, V. Viscosi, M. Di Michele, A. D’Andrea and D. Chiatante,
“ An integrated approach to the characterization of two autochthonous lentil (Lens culinaris)
landraces of Molise (south-central Italy)”, Heredity, vol. 101, pp. 136-144, 2008.
[4] G. Ladizinsky, “Wild lentils”. Crit. Rev. Plant Sci., vol. 12, pp. 169-184, 1993.
[5] A. R. Piergiovanni, “The evolution of lentil (Lens culinaris Medik.) cultivation in Italy and its
effects on the survival of autochthonous populations”, Genet. Resour. Crop Evol., vol. 47, pp.
305-314, 2000.
[6] H. Thiellement, N. Bahrman and C. Damerval, “Proteomics for genetic and physiological
studies in plants”, Electrophoresis, vol. 20, pp. 2013-2026, 1999.
[7] J. L. David, M. Zivy, M.L. Cardin and P. Brabant, “Protein evolution in dynamical managed
population of wheat: adaptative responses to macro-environmental conditions”,Theor. Appl.
Genet., vol. 95, pp. 932-941, 1997.

331
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 333-339.
ISBN 978-88-8303-295-0. EUT, 2010.

A model study for tardigrade

identification
Roberto Bertolani, Lorena Rebecchi, Michele Cesari

Abstract — Using tardigrades from a single moss sample as a case study,

we propose a new method for tardigrade species identification, which is
often problematic, due to the low number of morphological characters.
Identification at generic level was carried out on adults, while morphological
analyses were performed on animals (LM) and eggs (LM and SEM), including
hologenophores, vouchers used also for molecular analysis of COI mtDNA.
This multi-approach method revealed the presence of three species of the
“Macrobiotus hufelandi group” instead of the two species identified in a
previous study. The validity of the method is shown, indicating that it could be
applied to studies of problematic meiofauna taxa.

Index Terms — COI, DNA barcoding, morphology, Tardigrades, taxonomy.

—————————— u ——————————

1 Introduction

T
ardigrades consist of more than 1,000 described species [1], [2] colonizing
marine, limnic and terrestrial environments, including “hostile to life” and
unpredictable habitats. In the seventies, a new evaluation of the
intraspecific variability and new morphological characters for species
identification were proposed [3], [4], [5], which led the number of tardigrade
species to increase from less than 500 species described to that date to the
current number. An example of this improvement in identifying species can be
found in Macrobiotus hufelandi, the first described [6] and most commonly
identified tardigrade species. What was considered a single species is currently
represented by more than 25 species. Nonetheless, tardigrade identification at
the species level is often problematic due to the low number of taxonomic
characters. During our work it was not rare to find in the same moss sample
more than one tardigrade species, not only in the same genus but also in the
same species group, creating problems of species identification. For this reason
we have begun to identify species by coupling a detailed evaluation of animal
and egg shell morphology with DNA barcoding [7]. Using one moss sample as
a case study, we propose a new method for tardigrade species identification
and, in general, for identification of meiofaunal taxa whose morphological
————————————————
The authors are with the Department of Biology, University of Modena and Reggio Emilia, 41125
Modena, Italy. E-mail: [email protected]; [email protected]; michele.cesari@
unimore.it.

333
characters are often very limited.

2 Material and methods

The moss sample was collected at Andalo (Central Alps; province of Trento,
Italy, 46°N 10.133, 011°E 00.017, 1050 m) on a rock already examined by us
(Fig. 1) [7], [8]. Several tardigrade species were present in the sample, belonging
to different genera of eutardigrades (Macrobiotus, Minibiotus, Ramazzottius,
Milnesium) and heterotardigrades (Echiniscus) but only the specimens
belonging to the so-called “Macrobiotus hufelandi group” have been used in this
study. Two species in this group, Macrobiotus macrocalix Bertolani & Rebecchi,
1983 (amphimictic) and M. cf. terminalis (parthenogenetic), were already found
in the previous collections [7], [8].

Fig. 1 – The rock located in Andalo (Italy), with the moss patch that was used in the
study.

Morphological analyses of paragenophore voucher specimens (sensu Pleijel

et al. [9]) were carried out by mounting animals and eggs (that have species-
specific shell processes) in Faure-Berlese fluid for light microscopy (LM)
observations. Other eggs were fixed and dehydrated for scanning electron
microscopy (SEM) analysis. Additional animals were stained with acetic lactic
orcein for gender identification.
For molecular analysis, DNA was extracted from single entire animals.
Some of these specimens were newborns hatched from isolated eggs, whose
shells were mounted in Faure-Berlese fluid, obtaining hologenophore voucher
specimens (sensu Pleijel et al. [9]). Amplification and sequencing of 684 bp
of the COI mtDNA gene were carried out, following the procedures described
in Cesari et al. [7]. Kimura 2-parameters distances between haplotypes were
scored by using MEGA4 [10], while neighbor joining and maximum parsimony
dendrograms were computed using PAUP* 4.01b10 [11], also using sequences
retrieved from GenBank (EU244599; FJ435804-7; AY598773-5; FJ176203-17).
For possible further investigations, a fragment of the moss sample was stored
at -80°C.

334
3 Results
Observations of animals stained with acetic lactic orcein confirmed the
presence of males and females morphologically attributable to M. macrocalix by
the presence of a strong buccal armature, with thick crests and large bands of
evident teeth and with a relatively wide buccal tube, also observed in mounted
specimens (Fig. 2a). In addition, males were also found among the specimens
characterized by a weaker buccal armature and narrower buccal tube (Fig. 2b).

Fig. 2 – Buccal-pharyngeal apparatuses (Faure-Berlese fluid, phase contrast). a:

Strong buccal armature and wide buccal tube. b: Weak buccal armature and narrow
buccal tube. Scale bars = 10 µm.

Molecular analysis of specimens and eggs belonging to the “Macrobiotus

hufelandi group” revealed three clearly distinct haplogroups (a-c, Fig. 3), with
very high genetic distances among them (Tab. 1).

1 2 3 d
1 Haplogroup a 0.005
2 Haplogroup b 0.193 0.000
3 Haplogroup c 0.169 0.181 0.001

Tab. 1 – Kimura 2-parameters distances computed among (under the diagonal) and
inside (column d) haplogroups.

A detailed analysis of egg shell morphology both of the hologenophores and

of other voucher specimens (paragenophores) showed the presence of three
types of eggs, all bearing processes as inverted goblets on the shell (Fig. 4).
One type of egg exhibited high (9.6-10.7 µm) and wide (7.9-8.6 µm) processes
with large (7.0-8.0 µm) smooth distal discs and very large pits located only
around the process bases (typical of M. macrocalix) (Fig. 4a, d). The second
type of egg was characterized by clearly smaller processes (7.4-8.4 µm) than

335
those of M. macrocalix and having an irregularly edged distal disc (6.3-7.0 µm
in diameter) and a non-uniform reticulated egg shell with a thick meshwork (Fig.
4b, e). The third type of egg had small processes (5.0-5.3 µm in height) with a
slightly irregular edge on the distal disc (4.7-5.2 µm in diameter), and a very
uniform reticulated egg shell with a very thin meshwork (Fig. 4c, f).

Fig. 3 – Dendrogram combining neighbor joining (NJ, ME score: 0.731) and maximum
parsimony (MP, consistency index: 0.743; retention index: 0.920; rescaled consistency
index: 0.684) analyses. Numbers above branches indicate mutational steps, while
numbers in parentheses show bootstrap values computed after 2000 replicates (above
branches: MP; below branches: NJ). a-c denote different haplogroups, while H denotes
individuals for which hologenophore voucher specimens are available. Names in bold
indicate specimens pertaining to the studied moss.

336
 Fig. 4 – Voucher specimens consisting of egg shells. a-c: Faure-Berlese fluid (LM,
phase contrast). a: Macrobiotus macrocalix, haplogroup a (hologenophore H1).
b: M. cf. terminalis, haplogroup b (paragenophore). c: M. sandrae, haplogroup c
(paragenophore). d-f: SEM (paragenophores). d: M. macrocalix. e: M. cf. terminalis. f:
M. sandrae. Scale bars = 5 µm.

4 Discussion
The sex ratio analysis of the tardigrades belonging to the “Macrobiotus
hufelandi group” in the moss sample revealed a much more complicated
situation than that known from the literature [7], [8]. Nevertheless, by comparing
the results of a detailed morphological analysis with those obtained by DNA
barcoding, and in particular by sequencing the newborns’ DNA and linking their
sequences to the related egg shell shapes (hologenophores), the problem can
finally be solved.
The distance values among the three different haplogroups are very high,
far exceeding the 3% threshold and the 10x rule proposed by Hebert et al.
[12], [13], [14], thus supporting the specific rank of the three haplogroups. Two
species, M. macrocalix and M. cf. terminalis (currently being described as a
new species), morphologically correspond to what was previously found on the
same rock at Andalo [7], [8]. With regards to the third species, the animals look
similar to the specimens of M. cf. terminalis (even through probably smaller), but
the eggs are quite distinguishable and allow us to attribute them to Macrobiotus
sandrae Bertolani & Rebecchi, 1983. This species is known to be amphimictic

337
[8], a situation consistent with the presence of males among the animals with a
weaker buccal armature and narrower buccal tube.

5 Conclusions
The methods described here allow us to solve intricate tardigrade identification
problems, validating our new approach based on linking morphological
and molecular data. The use of voucher specimens, and in particular of
the hologenophores, is critical for obtaining a correct species diagnosis. A
hologenophore can also be obtained by culturing an isolated female until
oviposition, mounting it as voucher and using its developing eggs either for
molecular analysis and/or as further vouchers. Further information important
for identification can also be obtained from other tardigrades, which can be
photographed in vivo up to maximum magnification (100x objective) before
being used in molecular investigations.
In our opinion, our multi-approach method for tardigrade identification can be
easily applied to other meiofaunal taxa, whose few morphological characters
can generate problems in species identification.

Acknowledgement

The authors wish to thank Diane R. Nelson, East Tennessee State University, U.S.A,
for her help in the English revision and for her suggestions. This work was supported by
a grant from the Fondazione Cassa di Risparmio di Modena (Italy) and the University
of Modena and Reggio Emilia (Italy): “MoDNA project (Morphology and DNA): DNA
barcoding and phylogeny of tardigrades, basic research and applications”.

References
[1] R. Guidetti and R. Bertolani, “Tardigrade taxonomy: an updated check list of the taxa and a list
of characters for their identification”, Zootaxa, vol. 845, pp. 1-46, 2005.
[2] D. R. Nelson, R. Guidetti and L. Rebecchi, “Tardigrada”. In: J. H. Thorp. A. P. Covich (eds.),
Ecology and Classification of North American Freshwater Invertebrates, Elsevier Inc.,
Amsterdam, The Netherlands, pp. 455-484, 2010.
[3] G. Pilato, “Structure, intraspecific variability and systematic value of the buccal armature of
eutardigrades”, Z. f. Zool. Systematik u. Evolutionforschung, vol. 10, pp. 65-78, 1972.
[4] G. Pilato, “Redescription of Haplomacrobiotus hermosillensis May, 1948, and consideration
of the genus Haplomacrobiotus (Eutardigrada)”, Z. f. Zool. Systematik u. Evolutionforschung,
vol. 11, pp. 283-286, 1973.
[5] G. Pilato, “On the taxonomic criteria of the Eutardigrada”, Mem. Ist. Ital. Idrobiol., vol. 32,
Suppl., pp. 277-303, 1975.
[6] C. A. S. Schultze, Macrobiotus Hufelandii animal e crustaceorum classe novum, reviviscendi
post diuturnam asphyxiam et ariditatem potens, C. Curths, Berlin, 6 pp. 1 tab., 1834.
[7] M. Cesari, R. Bertolani, L. Rebecchi and R. Guidetti, “DNA barcoding in Tardigrada: the
first case study on Macrobiotus macrocalix Bertolani & Rebecchi 1993 (Eutardigrada,
Macrobiotidae)”, Mol. Ecol. Resour., vol. 9, pp. 699-706, 2009.
[8] R. Bertolani and L. Rebecchi, “A revision of the Macrobiotus hufelandi group (Tardigrada,
Macrobiotidae), with some observations on the taxonomic characters of eutardigrades”, Zool.
Scripta, vol. 22, pp.127-152, 1993.
[9] F. Pleijel, U. Jondelius, E. Norlinder, A. Nygren, B. Oxelman, C. Schander, P. Sundberg
and M. Thollesson, “Phylogenies without roots? A plea for the use of vouchers in molecular
phylogenetic studies”, Mol. Phylogenet. Evol., vol. 48, pp. 369-371, 2008.

338
[10] K. Tamura, J. Dudley, M. Nei and S. Kumar, “MEGA4: Molecular Evolutionary Genetics
Analysis (MEGA), software version 4.0”, Mol. Bio. Evol., vol. 24, pp. 1596-1599, 2007.
[11] D. L. Swofford, “PAUP* Phylogenetic analysis using parsimony (*and other methods)”, Version
4.0b10 win32”, Sinauer Associates, Sunderland, USA, 2002.
[12] P. D. N. Hebert, A. Cywinska, S. L. Ball and J. R. deWaard, “Biological identifications through
DNA barcodes”, P. Roy. Soc. Lond. B Biol., vol. 270, pp. 313-321, 2003.
[13] P. D. N. Hebert, S. Ratnasingham and J. R. deWaard, “Barcoding animal life: cytochrome c
oxidase subunit 1 divergences among closely related species”, P. Roy. Soc. Lond. B Bio., vol.
270, pp. 596-599, 2003.
[14] P. D. N. Hebert, M. Y. Stoeckle, T. S. Zemlak and C. M. Francis, “Identification of birds through
DNA barcodes”, PLoS Biol., vol. 2, e312, 2004.

339
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – p. 341.
ISBN 978-88-8303-295-0. EUT, 2010.

DNA Barcoding of
Philippine plants
Esperanza Maribel G. Agoo

Abstract — DNA barcoding is a technique that uses DNA sequence

data for species-level identification, for analyzing phylogenies and
interspecific variation and in population genetic studies. The DNA
barcodes determined by the Consortium of the Barcode of Life
(CBOL) to be the most effective in achieving these goals in plants
are the plastid genes namely trnH-psbA, rbcL, matK, accD, rpoB,
rpoc1, and trnL(UAA)-trnF(GAA). The goal of this barcoding project
is to test the genes, trnH-psbA, rbcL, and matK, in identification and
describing variation in some Philippine noteworthy indigenous plant
groups such as orchids, gingers, aroids, cinnamons, and cycads.
DNA are extracted and processed using standard protocol set by the
CBOL. The DNA are then kept in a cold storage facility in the DLSU-
CENSER laboratory. Voucher specimens are also collected and are
now deposited in the DLSU-Manila Herbarium. Results of this study
show that these candidate barcodes can successfully discriminate
species including probable novelties of cycads, aroids, and
cinnamons. Quantitative analysis also suggest that rbcL and trnH-
psbA are very variable genes and can reflect greater interspecific
variation thus are the more useful barcodes in these plant groups.

Index Terms — barcoding, Philippines, DNA, identification, flora.

————————————————
The author is with the Biology Department, De La Salle University-Manila, 2401 Taft Avenue,
Manila. E-mail: [email protected].

341
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – p. 343.
ISBN 978-88-8303-295-0. EUT, 2010.

Molecular and ecophysiological

characterisation of the Tunisian
bee: Apis mellifera intermissa
Mohamed Chouchene, Naima Barbouche,
Lionel Garnery, Michel Baylac

Abstract — This study concerns the morphological identification, the

molecular features and the eco-physiology of the Tunisian bee Apis
mellifera intermissa, focused on 655 colonies from 7 populations:
Kroumirie Moogod, North East Cap Bon, Ridge and Tell, high steppe,
lower steppe, Atlas Chainon, Jeffara and Ouarra. The geometric
morphometry of the interior wing of the bee shows polymorphism in size
and shape. The size polymorphism is essentialy related to beekeeping
practices. The characterization by means of a cytoplasmic molecular
marker - mitochondrial DNA (DNA m t) - showed that the Tunisian
bee originated from lineage A, which contradicts its membership to
lineage M as demonstrated by a study based on biometric data only
(Ruttner, 1988). There is a genetic polymorphism of the Tunisian bee
in the presence of four haplotypes: A1, A8, A9 and A4. The distribution
of the A4 and A9 haplotypes depends on ecological conditions.
Foreign haplotypes are present in the region of Ghardimaou near the
Algerian border (C7 haplotype). The study of some ecophysiological
parameters in colonies of Apis mellifera intermissa from 5 sites showed
that the Tunisian bee is endowed with a very marked disregard for all
haplotypes (A1, A4, A8 and A9). However, we report the existence of a
difference between these haplotypes in thermoregulation, oviposition
and respiration of solitary bees. The temperature of the A1 and A8
haplotypes brood nest is around 36°C while the A9 and A4 haplotypes
brood nest has a temperature of 34°C when weather conditions
are extreme. The A4 and A9 haplotypes fall into hibernation, the
temperature of the brood nest ranging between 22 and 28°C. The A1
and A8 haplotypes have a high tendency to lay A9 and A4 haplotypes,
which however is variable, ranging from zero to average depending
on climatic conditions. A study of respiration of isolated honeybees
showed a difference in oxygen consumption between haplotypes A1/
A8 and A4/A9 at low temperatures.

Index Terms — Apis mellifera intermissa, DNA, ecophysiology, Tunisia bee,

haplotypes.
————————————————
M. Chouchène, N. Barbouche are with the INAT, 43 Avenue Charles Nicolles, 1082 Tunis, Tunisia.
E-mail: [email protected].
L. Garnery is with the Laboratoire Evolution, Génomes et Spéciation, UVSQ - CNRS. E-mail:garnery@
pge.cnrs-gif.fr.
M. Baylac is with the MNHNP,36 Rue Geoffroy St Hilaire, 75005 Paris.

343
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 345.
ISBN 978-88-8303-295-0. EUT, 2010.

Biological identifications through

mitochondrial and nuclear
molecular markers: the case of
commercially important crabs
from Indian EEZ
Sherine Sonia Cubelio, K. K. Bineesh, K. Raj, Suraj Tewari,
Achamveettil Gopalakrishnan,
Valaparambil Saidumohammad Basheer, Wazir Singh Lakra

Abstract — The ability of a 600- 700 base pair section of

mitochondrial and nuclear molecular markers (COI, 16S & ITS-I)
to provide species level identifications has been demonstrated for
large taxonomic assemblages of animals such as insects, birds,
fishes and crustaceans. In the Indian context, there had been no
comprehensive attempt to determine the molecular sytematics,
evolutionary relationships or phylogeny within the crabs of genera
Scylla, Portunus and Charybdis which support the commercial
fisheries. The present study is the first attempt to test the suitability
of using a DNA barcode approach to discriminate accurately the
edible marine crabs from Indian waters. Partial sequence of COI,
16S and ITS-I revealed distinct species specific profiles supporting
the morphological data with low levels of intraspecific genetic
diversity. Differentiation of two species of Scylla, Scylla serrata and S.
tranquebarica using taxonomic tools is problematic especially based
on the adult morphology. DNA barcoding using COI and other regions
such as 16SrRNA and nuclear ITS fragment proved to be efficient in
discriminating the species. The study revealed that mitochondrial and
nuclear molecular genes will be an effective tool for discriminating
species of commercial importance and thus aiding in scientific
management of marine fishery resources.

Index Terms — molecular markers, crabs, fishery, Indian Ocean.

————————————————
S. S. Cubelio is with the National Bureau of Fish Genetic Resources (NBFGR) Cochin Unit,
CMFRI Campus, P.B. No. 1603, Kochi 682 018, Kerala, India. E-mail: [email protected].
K. K. Binesh is with CMFRI, P.B. No.1603, Ernakulam, Kochi 682 018, Kerala, India
The other authors are with NBFGR, Canal Ring Road, Dilkusha P.O., Lucknow 226 002, U. P.,
India.

345
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – p. 347.
ISBN 978-88-8303-295-0. EUT, 2010.

Barcoding Fauna Bavarica –

Capturing Central European
Animal Diversity
Lars Hendrich, Michael Balke, Gerhard Haszprunar, Axel
Hausmann, Paul Hebert, Stefan Schmidt

Abstract — The Barcoding Fauna Bavarica (BFB) is an All Species

Barcoding campaign ran by the Zoologische Staatssammlung in
Munich and the Canadian Centre for DNA Barcoding (www.
faunabavarica.de). Core funding comes from the Bavarian Ministry
for Science, Research and the Arts and from Genome Canada
through the Ontario Genomics Institute. The initial funding period is
from 2009–2013. Bavaria has the highest biodiversity of all German
states, with at least 35000 animal species reported, representing a
significant portion of the central European species diversity.
Ecoregions include high altitude biomes, foothill areas and forested
lowlands. The Zoologische Staatssammlung (ZSM) is one of the
largest German natural history research institutions. It holds the
world’s largest collection of Lepidoptera and Germany’s largest
Hymenoptera collection. Since mid-2009, the BFB project has
contributed DNA barcode records from 7208 specimens representing
3000 species and is therefore, after less than one year, one of the
most comprehensive sources for local DNA barcode data. The focus
groups for the initial phase were Lepidoptera (1820 species
barcoded), bees (316 species), ants (39 species) and aquatic insects
(322 species). Work on these focal groups will continue during 2010,
with the goal to complete 80% of the Bavarian focal group species by
the end of the year. New focal groups are Diptera, Mollusca, all
Vertebrata and terrestrial Coleoptera, targeting 2000 species in 2010.
Most tissue samples come from specimens in the ZSM collection,
and where this was not feasible from freshly collected and identified
specimens. This rapid progress reflects the strong involvement of
taxonomists throughout the process, which is one of our key missions.
We have implemented a system which co-ordinates vouchers stored
in our main collection, with tissues as well as DNA samples in our
DNA bank.

————————————————
L. Hendrich, M. Balke, G. Haszprunar, A. Hausmann and S. Schmidt are with the Zoologische Staats-
sammlung, Münchhausenstraße 21, 81247 München, Germany. E-mail: [email protected].
P. Hebert is with the Biodiversity Institute of Ontario, 579 Gordon Street, University of Guelph,
Guelph, Ontario, Canada N1G 2W1, E-mail: [email protected].

347
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – p. 349
ISBN 978-88-8303-295-0. EUT, 2010.

Molecular techniques for

identifying North Sea fauna
Thomas Knebelsberger, Sandra Ditzler, Silke Laakmann,
Inga Mohrbeck, Michael J. Raupach

Abstract — Accelerated biodiversity assessment is the key to

understanding the relationship between biodiversity and ecosystem
functioning, especially in times of rapid climate change and habitat
destruction. For the marine fauna of the North Sea, morphological
species identification is impaired by the small size of many taxa,
morphological convergence, intraspecific variation and larval stages
which often elude morphological identification. Accordingly, the use
of molecular methods presents highly promising tools for fast and
accurate species identification. The aim of the new established
research group “molecular taxonomy of marine organisms” at
the German Centre of Marine Biodiversity Research is to test and
develop molecular methods for the identification of the marine fauna
of the North Sea, aiding efforts to monitor biodiversity patterns and
changes. The research will focus on the analysis and identification
of specimens using DNA barcodes, and environmental samples,
in particular zooplankton, using next-generation DNA sequencing
techniques. In addition it is planned to develop molecular methods
for a fast and routine identification of larvae of selected invertebrate
and vertebrate taxa of economic value.

Index Terms — molecular methods, DNA barcodes, North Sea, fauna.

————————————————
The authors are with the Senckenberg am Meer, Deutsches Zentrum für Marine Biodiversitätsfor-
schung, Südstrand 44, 26382 Wilhelmshaven, Germany TK. E-mail:[email protected].

349
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – p. 351.
ISBN 978-88-8303-295-0. EUT, 2010.

DNA Bank Network – connecting

biological collections and
sequence databases by long-
term DNA storage with online
accession
Matthias Geiger, Nicolas Straube

Abstract — In times of increasing numbers of methods and tools for

the molecular identification of organisms, it is inevitable that
researchers have to deal with an additional flood of samples: the
extracted DNA of organisms which are in the researcher’s focus.
Today, voucher specimens – the specimens from which DNA was
extracted – have to be placed in adequate biological collections and
organisms’ sequence data can officially be deposited in online
databases such as Genbank, often a prerequisite for publication of
results in peer-reviewed journals. DNA extracts do not underlie such
rules, but adequate housing of DNA extracts, especially of rare or
difficult to obtain species, will be a major task in the near future.
The DNA Bank Network bridges the gap between natural history
collections and molecular sequence databases by providing online
references to analysed specimens and inferred molecular data.
DNA samples are linked to their respective vouchers and inferred
molecular data are stored in public sequence databases, facilitating
taxonomic verification of molecularly analysed organisms.
We provide the opportunity for long-term storage of DNA in the
DNA Bank Network, giving other reseachers the opportunity to
access DNA for further projects dealing with the same organisms.
In this way, multiple sampling can be avoided and there is a direct
link between the three main sources of information, i.e. the sampled
organism, the DNA, and the sequence data. Here we present the
functioning and layout of the DNA Bank Network, which currently
connects DNA banks of four research museums in Germany: the
Bavarian State Collection of Zoology (ZSM), the Botanic Garden and
Botanical Museum Berlin-Dahlem (BGBM), the German Collection
of Microorganisms and Cell Cultures Braunschweig (DSMZ), and
the Zoologisches Forschungsmuseum Alexander Koenig (ZMFK).
Presently, the DNA Bank Network allows to access DNA samples of
more than 35.000 DNA samples and 11.000 taxa.

Index Terms — DNA, databases, molecular identification.

The authors are with the Bavarian State Collection of Zoology, Münchausenstrasse 21, 81247
Munich, Germany. E-mail: [email protected].

351
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 353-354.
ISBN 978-88-8303-295-0. EUT, 2010.

Mitochondrial DNA sequences

for forensic identification of
the endangered whale shark,
Rhincodon typus (Smith, 1828):
A Case study
Kavungal Abdulkhadar Sajeela, Chandran Rakhee,
Janardanan Nair Rekha, Achamveettil Gopalakrishnan,
Valaparambil Saidumohammad Basheer, Joe Kizhakkudan Shoba,
Kizhakkudan Joe, Wazir Singh Lakra

Abstract — The whale shark (Rhincodon typus), the largest fish in

the ocean, has become susceptible to over-exploitation and has a
global conservation status of ‘vulnerable to extinction’ as listed by
World Conservation Union in the Red list of threatened species. The
increase in demand for its meat, skin and fins in international trade
is a severe threat to the animal and its indiscriminate capture will
have to be taken seriously as they may have a major impact in the
marine ecosystem. Rhincodon typus was nominated in Appendix
II of Convention on International Trade in Endangered Species
(CITES) in April 2000, to enable adequate regulation of trade of
whale shark products. Whale shark (Rhincodon typus) is enlisted as
one of the protected species in India and its fishing prohibited under
Schedule Ι of the Indian Wildlife Protection Act, 1972, according to
the Order No.1-2/2001 WL1 Dated 28.05.2001, Govt. of India, so as
to conserve the species in Indian waters. still illegal fishing prevails
in Indian waters and the catch is processed in the vessel itself
and sold in markets as meat chunks. To curb the illegal trade and
marketing of fishery products from whale shark, for devising good
management practices and for the strict law enforcement, accurate
and reliable species identification methods using molecular tools are
of paramount importance. In an effort to establish a comprehensive
identification data set, we have generated a species-specific partial
sequence data of the mitochondrial genome of properly identified
stranded whale shark samples, covering the 16S rRNA (546 bp),

————————————————
K.A. Sajeela, C. Rakhee, A. Gopalakrishnan and V.S. Basheer are with the National Bureau of
Fish Genetic Resources (NBFGR) Cochin Unit, CMFRI Campus, P.B. No.1603, Kochi 682 018,
Kerala, India. E-mail: [email protected].
J.N. Rekha, S.J. and J. Kizhakudan are with CMFRI, P.B. No.1603, Ernakulam, Kochi 682 018,
Kerala, India.
W.S. Lakra is with NBFGR, Canal Ring Road, Dilkusha P.O., Lucknow 226 002, U. P., India.

353
Cyt b (541bp), COI (600bp) genes as the reference genetic profile
helping in accurate identification of any body parts of the species.
In the year 2008, flesh suspected as that of the Wildlife protected
whale shark (Rhincodon typus) was seized from fishermen by the
Forest Range Officer (Govt. of Kerala), Kannur, Kerala, India and
was brought before the Judicial First Class Magistrate, Thalassery,
Kannur, Kerala, India. The detailed sample analysis and confirmation
of species was carried out at NBFGR Cochin Unit (R.P.330/08, dt 29.
09. 2008). Based on DNA sequencing of 16S rRNA(525bp) and COI
(600bp) Cyt b(541bp) genes and comparing with the sequences earlier
generated by NBFGR (FJ375724, FJ375725, FJ375726, FJ456921,
FJ456922, and FJ456923), the suspected sample was identified as
that of endangered Whale Shark (Rhincodon typus) and the result
was communicated to the court. This is the first criminal case in India
in which scientific evidence was sought in forensic identification of
the meat of an aquatic organism enlisted in the Wildlife Protection Act
of India and the DNA markers reiterated their ability to reliably identify
product/meat sample of a species, thus helping in curtailing illegal
trade of the endangered organisms.

Index Terms — DNA markers, whale shark, identification.

354
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 355-360.
ISBN 978-88-8303-295-0. EUT, 2010.

An assignment-based
e‑learning course on the use of
KeyToNature e-keys
Pencho Mihnev, Nadezhda Raycheva

Abstract — This article conceptualises a common approach to train teachers

and university lecturers in integrating the use of e-keys produced by the
EU-funded project KeyToNature in the design, development, and delivery
of e-learning modules within their school practice. The approach is based
on setting and administering learning scenarios developed around real
practice assignments that have at their core the identification of organisms
and end up with the creation of meaningful biodiversity information products,
rooted in concrete environmental contexts. The learning products may even
include the creation of specific sub-keys from the initially used e-keys. The
paper briefly describes the application and the real testing of the approach
in an experimental teacher training course conducted in Bulgaria within the
KeyToNature project.

Index Terms — active learning, assignment, biodiversity, e-keys, e-learning

course, identification tools, learning design, teacher training.

—————————— u ——————————

1 Introduction

T
he KeyToNature project shares a big variety of electronic identification
tools and instruments for their handling in different modes and on different
hardware and software platforms. Using and integrating these tools within
different learning scenarios and contexts is a concrete task of the teachers. Some
or all stages of the applied learning scenarios and the resulting final products are
often not digitised, which has obvious negative effects. Another consideration
related to the digitisation is the use of different e-tools not pertaining to a united
system. This complicates the organisation of learning, leads to efficiency issues,
and may lower the re-usage potential of the achieved results.

————————————————
P. Mihnev is with BIKAM Ltd., 46, Oborishte str., Sofia 1505, Bulgaria. E-mail: pmihnev@gmail.
com – N. Raycheva is with the Department for In-Service Teacher Training, Sofia University “St.
Kl. Ohridski”, 224, Tzar Boris III bld., Sofia, Bulgaria. E-mail: [email protected].

355
2 The Use of a Learning Management System as a Learning
Platform for the Creation of Effective Scenarios
The identification process is “a means to an end”, not “the end” itself. An
implication about learning by using identification tools is that the identification
needs to be integrated in wider contexts of meaningful learning scenarios.
Designing good scenarios requires good pedagogical skills, knowledge of the
subject matter, adequate technological skills, and the availability of technological
facilities for implementation. A scenario should also reflect the active learning
and motivational findings researched and described by learning and motivation
theories, e.g. [1], [2], and [3].
The identification process is per se an active learning. The keys themselves
become more complex, integrating add-ons that facilitate re-usage and complex
learning activities.
In order to maximise the positive effects of the e-keys, one can use the support
of e-learning platforms. A newer and richer concept about the use of e-learning
platforms is that activities and not the content are in the core of an e-learning
platform.
This concept directly corresponds both to the active learning methods and to
the creation of contexts which exploit positive motivation factors.
Another feature of all widespread e-learning platforms is the inter‑linkage of all
resources, activities, communication, organisation, and assessment tools within
the platform. This enables more complex learning events and activities to be
organised and conducted as a “learning whole”.
The authors have designed an experimental curriculum for teacher training to
develop and use e-learning modules for students having as their core activity
identification processes based on the use of e-keys. The Moodle learning
management system (e-learning platform) was chosen for the concrete course
design.
The main teacher training target group consisted of biology teachers that have
average ICT-skills, are able to work with e-keys, but are not trained to work with
any e-learning platform.
We have chosen to use only a few, but very important and powerful e-tools of the
platform (resources and activities) that can add real value in the implementation
of meaningful learning scenarios for the use of e-keys.

3 The Teacher Training Curriculum – the Course Study

Programme
The teacher training course curriculum consisted of two main parts: the first
part puts the teacher in a student role to alter her/his viewpoint in experiencing
learning. The second part trains teachers on how to construct and conduct an
e-Learning course.
The first part is based on Assignment 1:
“As a trainee in the first part of this course you - preferably in a team of 2 to 3
persons - should:

356
 • Produce an electronic “Profile of a tree”, following a predefined structure,
by using the information previously entered for that purpose by your
team in the e-learning platform.
• Publish and present in an attractive and appealing way your electronic
“Profile of a tree” by using the tools and instruments of the e-learning
platform.
• Edit and complement the identification key with collected and/or
personally developed material.”
This assignment consists of activities in two different environments: in the field
and in the classroom.
Activities in the filed include tree identification with the e-key and observation
of the characteristics that are used to identify it. Activities in the classroom
include working on the structure of a tree profile and editing a e-key. The tree
profile describes the following information:
1. Name of the tree: Latin, Bulgarian; 2. Classification (levels of detail: on
learner’s own judgment); 3 Photos (minimum 3 taken by the learner, and 2 from
the e-key) - they should present a natural view of tree, leaf - margin, upper
and lower surface, flower and fruit; 4. Description (following a worksheet); 5.
Importance for mankind; 6. Do you know that… (interesting facts/information
about the species); 7. Additional information – personal comments (personal
opinion); 8. References and resources used.
The second part is based on the Assignment: “Develop your own e-module on
identification for your specific case (subject, grade, and students)”.
The curriculum, structured in chapters and activities, can be reviewed in the
folder “Materials” of [4].

4 The Course
An experimental course was conducted on July 8th and 9th 2010 at the
Department of Information and In-Service Teacher Training of the Sofia
University. The Department campus includes a small park in front of the buildings
where the field work was carried out.
Twelve participants took part in the course – 8 Biology school teachers, 2
Biology trainers from a training centre, 1 university lecturer in Botany, 1 Science
expert from a Regional Educational Inspectorate. The course was held by the
authors of this article.
The entry level of the course participants with respect to their ICT knowledge
and skills was as follows:
1. All participants were able to work with the e-keys of KeyToNature;
2. No participant worked previously with the Open Key Editor1;
3. All participants had an “average” skill level for working with ICT, namely:
Windows, e-mail, Internet, MS Word, MS PowerPoint;
4. Only one participant worked previously with an e-learning platform - Moodle

————————————————
The Open Key Editor is a software developed within KeyToNature. It permits to edit already existing
keys or sub-keys extracted from them, by changing the text, adding images, adding new species,
changing the structure of the e-key, and even creating new e-keys from scratch.

357
- without using it after the training.
The aim of the course was to test the developed curriculum and to receive
feedback on effectiveness and efficiency of the course.
The twelve participants were grouped into 5 permanent groups; the work was
conducted by each group as a whole.
In broad terms, the course time frame was set in the following way:
1. First half of the first day – work in the field: trees’ identification, gathering
additional information (taking photos, observing the environment, taking notes),
and filling-in the “terrain” part of Worksheet 1 (profile of the tree);
2. Second half of the first day – work as an user of the Moodle platform
and of the e-learning course: collecting the requested additional information
from Internet and from the e-key, entering the collected data in the interactive
geographic map of Moodle and in the prepared course multimedia database;
working with the Open Key Editor and developing a sub-key.
Homework: development of a short PowerPoint presentation about the profile
of the identified tree.
3. First half of the second day – Design of e-learning modules in Moodle:
e-Course setting, student enrollment, work with selected resources (labels,
folders, hyperlinked text, access to different study materials); developing an
e-course programme/syllabus.
4. Second half of the second day – work with selected activities in Moodle:
setting up interactive geographic maps, developing a database, developing an
assignment, setting up a glossary. Starting the design of each trainee’s own
e-learning module.
Homework: Full design and preparation of the e-learning module, ready for
use by students.
It was very encouraging to see that all groups managed to perform well the
required activities and to develop the products envisaged in the course.
The results of the work of each group from the course can be seen in the
Bikam’s KeyToNature Moodle platform [4]. Photos from the teacher training
course can be seen at the web-address provided in [5].

5 Evaluation of the Teacher Training Course

5.1 COLLES - Constructivist On-Line Learning Environment Survey

We used COLLES [6] as one of the two survey instruments to evaluate the
course. COLLES comprises 24 statements grouped into six scales. The six
groups are Relevance, Reflection, Interactivity, Tutor Support, Peer Support,
and Interpretation. The concrete survey questions grouped by categories can
be reviewed at [7]. Graph charts with all survey results can be viewed at [4].
Important feedback from the teachers’ answers was that:
1. In terms of the question categories the course scores higher than the middle
value of occurrence in the scale (“Sometimes”).
2. The highest, almost maximal, score in the survey is assigned to the course
relevance category. According to J. Cole and H. Foster [8], p. 192, the Relevance

358
category is the most important with respect to the assessment of the course
design.

5.2 Free-Form Opinion Survey

In the second, free-form opinion survey, participants were asked about the
strong and weak sides of the course and its methodology.
The articulated strong sides were:
1. The power of the e-learning platform to offer electronic means that unite into
a learning whole and serve the overall learning process.
2. The power of attractivity to students of the final products created within (or
with the support of) the e-learning platform.
The most often mentioned weak sides were related to:
1. The eventual lack of sufficient hardware for the implementation of e-learning
scenarios - laptops for field work, the availability of computer labs.
2. About the course delivery – the very short duration of the course – only 2
days. The participants didn’t know that the course was intentionally compressed
to last only two days in order to test the possibility of achieving the main goals
in such a short training time.

6 Summary and Conclusions

The e-course proved to be a success, especially having in mind the intentionally
imposed extreme constraints (unskilled trainees to work with e-platform, and
short duration of the course).
The use of an activity-based e-learning platform as Moodle maximises the
learners’ motivation, effectiveness, and efficiency of both students’ learning and
teacher preparation. The teachers have to be trained accordingly, preferably
initially through a simple, not overwhelming training that utilises only some, yet
powerful, learning activities in the e-learning platform.
Two other important implications are that our curriculum & learning design
approach proved to be:
1. “e-key independent”. It can be used with any interactive identification key,
irrespective of the organisms it identifies.
2. “e-learning platform independent”, providing that the corresponding platform
has a set of functionalities/instruments that can realise appropriate learning
activities.
Furthermore, the opportunity to mix/blend the course could be considered, to
organise its delivery partly face-to-face, partly by distance.
The course was prepared in levels, in a modular way. This article has discussed
only the first, initial level of the curriculum, which was also tested with teachers
in the course. The next two course levels are meant to update and upgrade
the knowledge and skills of the teachers, to fully employ the functionalities and
instruments of the platform.
As a next step in learning design and delivery inquiries, it will be interesting
to test the delivery of an entirely distance-online course, following the outlined
curriculum.

359
Acknowledgement

This work has been supported by the KeyToNature project, ECP-2006-EDU-410019,

in the frame of the eContentplus Programme.

References
[1] R. L. Hanna, “Active Learning = Remembering = Learning”, Las Positas College, Livermore,
California, USA. http://lpc1.clpccd.cc.ca.us/lpc/hanna/learning/activelearning.htm, July 2010.
[2] S. Hidi, “Interest and Its Contribution as a Mental Resource for Learning”, Review of
Educational Research, vol. 60, 4, pp. 549-571, Winter 1990.
[3] S. Reiss, Who am I: The 16 basic desires that motivate our actions and define our personalities,
New York: Tarcher/Putnam, ISBN 1-58542-045-X, 288 p., 2000.
[4] KeyToNature Moodle e-learning platform. http://k2n.bikam.com/moodle-new, July 2010.
[5] Photo gallery “Training KeyToNature”. http://picasaweb.google.bg/k2n.bulgaria/
TrainingKeyToNature#, July 2010.
[6] P. C. Taylor and D. Maor, “The Constructivist On-Line Learning Environment Survey (COLLES)”
Curtin University of Technology, Pert, Australia. http://surveylearning.moodle.com/colles/, July
2010.
[7] P. C. Taylor and D. Maor, “Example COLLES (Preferred and Actual)”. Curtin University of
Technology, Pert, Western Australia. http://surveylearning.moodle.com/mod/survey/view.
php?id=3, July 2010.
[8] J. Cole and H. Foster “Using Moodle: Teaching with the Popular Open Source Course
Management System”, 2nd Edition. O’Reilly Media Inc., USA, 2008.

360
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 361-365.
ISBN 978-88-8303-295-0. EUT, 2010.

User needs for interactive

identification tools to organisms
employed in the EU-Project
KeyToNature
Astrid Tarkus, Emanuel Maxl, Christian Kittl

Abstract — The EU-funded project KeyToNature is developing and

optimising interactive tools for identifying organisms, making them suitable
for being usable in the field of formal education across Europe. To define the
requirements of the target audience, research was conducted in 11 partner
countries during an initial project phase. Teachers and lecturers from primary
schools to university level were asked to express their views about selected
existing identification tools in a qualitative survey. The target audience was
asked about perception, strengths and optimisation options, output channels
and pedagogical application fields. The results showed that the adaption
of the tools to the range of local organisms and the native language of the
audience represents a fundamental step.

Index Terms — education, EU project, identification keys, KeyToNature, user

needs.

—————————— u ——————————

1 Introduction

T
he 3-year EU-funded project KeyToNature deals with interactive tools
designed to identify organisms. These software and web-based tools
are to be incorporated within educational structures with the objective
of improving knowledge of biodiversity. KeyToNature aims to provide easy
access to identification tools, to optimize their educational efficiency and ease
of use. A further objective is to provide for the interconnection of these tools,
so that multilingual access is possible and usage across Europe as a whole is
enhanced.
A total of 14 project partners from 11 countries – Austria, Belgium, Bulgaria,
Estonia, Germany, Italy, the Netherlands, Romania, Slovenia, Spain and the
————————————————
A. Tarkus and E. Maxl are with the Market and Usability Research Department of evolaris next
level Competence Centre, Graz, Austria; E-mail: [email protected], emanuel.maxl@
evolaris.net.
C. Kittl is CEO at the evolaris next level Competence Centre, Graz, Austria; E-mail: christian.kittl@
evolaris.net.

361
United Kingdom – are participating in this project [1], [2].

2 Background Situation
The success achieved through the use of educational tools in the classroom
is determined to a large extent by the competence of the teachers and the
pedagogical concepts they employ [3], [4]. In their international review of
education, for example, Kugeman & Fisher [5] established that the systematic
involvement of teachers is crucial for the success of ICT tools – from setting
up the learning processes and creating pedagogical concepts at the start, to
reviewing and verifying the contents and results provided by the learners.
During the KeyToNature Project start phase, a state of the art presentation of
the tools was undertaken [6]; a selection of the tools that were to be used for
research purposes was put forward for review.
As many aspects of the existing tools and prototypes for pedagogical use
have not previously been considered in the international context, we decided to
conduct a user requirements analysis as a first step towards involving teachers
in the development of the identification tools.
This paper presents the results of our user requirements survey in connection
with the KeyToNature project. Our main objective was to identify specific details
of how the tools are perceived. What do teachers think about the general concept
of the tools and their use within the curriculum? Which didactic framework is
suitable for the identification tools and how can they be implemented in lessons?
Which medium (e.g. mobile phone, website) is perceived as most appropriate?

3 Research Design
In order to meet the target groups’ needs, we decided to concentrate on a
user-oriented design approach [7]. In particular, data on the pedagogical and
educational requirements, as these relate to the identification of biodiversity
were gathered by means of qualitative analysis in all partner countries.
The target audiences were lecturers who were recruited by the project partners
in their respective countries. At each educational level - primary, secondary and
university - focus groups were formed in all 11 participating countries in order to
obtain input for all end-user segments [8].
Where it was not feasible to form focus groups, qualitative interviews – face-
to-face, per telephone or email – were used as an alternative [9]. The tools
were presented and afterwards discussed. Specific guidelines that covered
the survey questions were employed. The collected data were subsequently
analysed by the partner countries and summarised in a detailed report.
A total of 219 teaching staff participated in the survey that was conducted
in the period October 2007 to February 2008. Of these, 152 were interviewed
within focus groups, 33 were surveyed in face-to-face interviews and 11 were
surveyed by means of email.

362
4 Tools Presented
The material under consideration consisted of a collection of existing tools:
these were presented to the target audience, where possible, either in form of
online prototypes or as concepts using PowerPoint slides - depending on their
development status.
The identification tools (i.e. software-based identification keys) use different
techniques to identify organisms. These include dichotomous keys which
provide identification based on only two different possible selection options per
stage, and multi-criteria keys which enable users to select several characters
at once. Another option is to provide a free input field so that users can search
for results on the basis of their own entered search criteria (e.g. taxon name).
The presented tools were:
1. Walking with woodlice (primary school level): a simple web-based key to
woodlice in the UK with a colourful interface.
2. Key to Trees and Shrubs (primary and secondary school levels): a web-
based tool that makes extensive use of pictures and graphics.
3. Earthworm Survey (secondary school level): a dichotomous key that uses
PowerPoint slides.
4. Key to the Flora of Val Rosandra (university level): this employs the same
software and UI as the “Key to Trees and Shrubs”, but includes substantially
more species (c. 1000).
5. E-Flora iberica (university level): a web-based key to the plants of the Iberian
Peninsula with free text search option and a powerful browsing mode.
6. Bumblebee (university level): a key to bumblebees in the form of an
interactive flash application.

Fig. 1 – Screenshot of a dichotomous key to woody plants (with options “leaves

opposite” and “leaves not opposite”).
The current versions of the tools can be viewed at www.keytonature.eu.

363
5 Results
Perception of interactive identification tools: The dichotomous keys were
perceived as an easily understandable, target group-oriented concept.
Those surveyed considered that pupils/students would find this tool easier to
use than the multi-criteria tools as they would be required to deal with less
information at one time. The multi-criteria selection option should be provided
for more advanced pupils/students, also because there was only limited pictorial
information on individual characteristics available. The free text input option also
met with a positive response, but was considered to be a more difficult tool
to use for identification purposes, requiring more knowledge by the user with
regard to the criteria that are crucial for differentiation.
Adaptation to the educational level: Our survey population considered that
the woodlice key and the earthworm key would be suitable for a younger target
group if the present application were improved - translations to native language
and age-specific design. The idea of organism identification was generally seen
as suitable at all levels, but the texts and designs needed to be tailored to the
specific target audience.
Suitable media: In general, the fact that the target audience would need to be
able to use a computer was not seen as problematic – even at primary school
level – although these are not always available at all schools.
With regard to the media format, the CD-ROM was liked best by our survey
population because this ensures there is no distraction by other web sites or
services. Nevertheless, the web application was also perceived as positive
in view of the better availability of updates, the platform it provides for online
activities and its community-related aspects, such as the option for links with
discussion forums, specialists etc. The mobile versions (for mobile phones or
PDAs) were seen as outstanding in comparison with the other media because
they can be used in the field; however, negative aspects were cited, such as
the cost of data transfer and the limited screen size that makes it difficult to
recognise organism details in images.
Pedagogical framework and educational applications: Several potential
applications were identified; one that was frequently mentioned was the use for
project work (at home, in the field or at school) at the primary and secondary
school level, as current school curricula do not provide sufficient time to cover
identification of organisms. Consequently, elective subjects would provide a
perfect environment to work with identification tools. It was suggested that these
tools could be used to present group projects in front of the class, thus helping
improve pupils’ presentation skills. However, one crucial aspect specified
was that pupils at primary and secondary school level would need to be first
instructed in the use of the tools by a teacher.

6 Conclusions
Several recommendations for optimisation based on the proposals made by
our survey group were subsequently implemented. However, the survey itself
had certain limitations. The tools were presented to the survey group only once,

364
and the individual teachers and lecturers were not given the option of trying out
the tools over the long term in their educational institutes. Within the project
progress, hundreds of tools were subsequently adapted to local requirements;
local organisms were included and additional languages were integrated in the
system with the help of some of the teachers and other associated members of
KeyToNature who had been recruited for the project. Higher quality, high definition
images and photographs and more interactive features were also added. At a
subsequent phase of the project, an end-user evaluation was conducted by
means of trials with students as subjects in order to investigate practicability,
cognitive level of difficulty, and look-and-feel of the tools. A generally accessible
platform was put in place in order to provide an arena for communication
and for showcasing of ideas for the educational use of the tools. This has the
potential to be further expanded and developed in future. Teachers and students
were encouraged to participate by inputting descriptions, nomenclature and
ecological data, distribution maps, images, sources and feedback. The purpose
was to establish an international network of those interested in biodiversity and
promote the dissemination of knowledge in this field.

Acknowledgements

The authors wish to thank all project partners that contributed in the user requirements
analysis for valuable inputs and a great cooperation. KeyToNature is funded under the
eContentplus programme, a multiannual Community programme to make digital content
in Europe more accessible, usable and exploitable (ECP-2006-EDU-410019).

References
[1] Project website and platform for identification tools KeyToNature, http://www.keytonature.eu,
2010.
[2] S. Martellos and P. L. Nimis, “KeyToNature: Teaching and Learning Biodiversity: Dryades, the
Italian Experience”, Proc. International Association for the Scientific Knowledge (IASK) Intern.
Conf. Teaching and Learning, pp. 863-868, Aveiro, 2008.
[3] D. H. Jonassen, Modeling with technology: Mindtools for conceptual change, Ohio, Prentice-
Hall, 2006.
[4] J. M. Cox and C. Abbott, ICT and attainment: A review of the research literature, British
Educational Communications and Technology Agency/Department for Education and Skills:
Coventry and London, 2004.
[5] W. F. Kugemann and T. Fischer, e-Learning 2005-2007: Eine Bestandsaufnahme der
europäischen Beobachtungsplattform HELIOS
http://www.bibb.de/dokumente/pdf/a32_fernlernen_helios_fischer.pdf, presented at 5. BIBB-
Fachkongress, Düsseldorf, 2007.
[6] P. L. Nimis, S. Martellos, G. Hagedorn, M. Brugman, M. Ferrer, A. Saag, T. Randlane, P.
Schalk, B. Press, A. Barry and T. Trilar, KeyToNature, Inventory of educational products http://
www.keytonature.eu/w/media/9/97/D.3.1_K2N_IDTools_Survey_Report.pdf, 2006.
[7] J. Nielsen, Usability Engineering. San Francisco, Morgan Kaufmann, 1994.
[8] T. L. Greenbaum, Moderating Focus Groups. A Practical Guide for Group Facilitation, London,
SAGE Publications, 2000.
[9] G. Mey and K. Mruck, “Qualitative Interviews”. In: G. Naderer and E. Balzer (eds.), Qualitative
Marktforschung in Theorie und Praxis Grundlagen, Methoden und Anwendungen, Wiesbaden,
Gabler, pp. 249-276, 2007.

365
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 367-371.
ISBN 978-88-8303-295-0. EUT, 2010.

Teaching biodiversity with online

identification tools
from KeyToNature:
a comparative study
Felicia Boar, Adelhaida Kerekes

Abstract — This paper reports on new pedagogical approaches in teaching

biodiversity by using online interactive identification tools developed by the
European project KeyToNature. A comparative educational study was carried
out on two classes of students, revealing the educational value of the interactive
identification tools. A systematic evaluation of both the educational processes
and the acquired skills was conducted, and the results are presented here.

Index Terms — biodiversity, online identification keys, e-learning, assessment.

—————————— u ——————————

1 Introduction

B
iodiversity represents the variety and variability of living organisms at the
taxonomic and ecosystem levels [4]. Every species has a well-defined
niche in natural ecosystems. Relations among species are complex and
of different kinds, the disappearance of some species can bring imbalance in
ecosystems [2].
Knowledge of biological diversity through curricular and extra-curricular
activities is one of the main objectives of environmental education. Education on
biodiversity is important for raising environment-aware citisens: the students
have to be aware that every species has its own place and role in the maintenance
of the ecological balance of the Earth, and that biodiversity safeguards the order
of the planet affecting climate change, keeping the air clean, providing food,
resources, medicines and potable water.
The teaching of biodiversity is a complex process: it needs a blend of classic
educational methods (observation, simulation, explanation, field-work, visits to
botanic gardens, zoos, natural history museums, etc.) with modern approaches
such as the use of educational software and related questioning and team-
————————————————
F. Boar is Professor of biology with the Vocational Environmental Protection High School, 400202
Cluj, Romania. E-mail: [email protected].
A. Kerekes is Inspector of biology with the Cluj Inspectorate for Education, 400200 Cluj, Romanian
E-mail: [email protected].

367
research projects. These allow the forming of abilities in measurement,
phenomenon recognition, ecosystem research, and – a basic task - identification
of species.

2 Educational process

2.1 Didactic methods

Didactic methods – when used skillfully - lead to systematic and progressive

enrichment of knowledge. They can be defined as “ways of acting, by means of
which students can - independently or under the guidance of teachers - acquire
knowledge, build up their skills, abilities and attitudes, develop their world and
life outlook” [5]. Modern methods of teaching and learning with educational
software (Fig. 1) are important in this context. They help students by providing a
large amount of well-organized and structured opportunities for practicing skills
in various fields of knowledge, allowing an easy and often playful simulation
of processes and phenomena which are often hard or impossible to access
directly [1], [3].

Fig. 1 – Using the identification tools of KeyToNature.

We have conducted a comparative study in teaching the identification of

Spermatophytes to two classes of students. Teaching methods were identical in
the two classes (see Tab. 1): observation, modelling, a study tour, explanation, a
botanical atlas, worksheets. For identification, however, a class used a pictorial
botanical atlas, the other class used computer-assisted learning tools developed
by the European project KeyToNature [6, 7].

2.2 Learning activities

The developed learning activities (see Tab. 1) aimed at teaching the general
characteristics, recognising some representative species, discussing the
ecology and economical/ecological importance of Spermatophytes.

368
 Systema- Topic / time Training scenario
tic group Class A Class B
Gymno- G e n e r a l ● grouping students in four groups (collaborative tasks)
sperms characteristics ● the students study the available biological resources
(1 hour) ● they observe the morphology of leaves and cones identifying
the main characteristics of gymnosperms; these were written
down afterwards on a dedicated worksheet

Identification & ● place: biology lab ● place: computer lab

description of ● usage guidelines ● usage guidelines
the species ● use of a botanic atlas ● use of a KeyToNature interactive
(1 hour) to identify five species key [7]
● use of worksheets for ● using the worksheets for Class B
Class A (e.g. Tab. 2)

Angio- G e n e r a l ● grouping students in four groups (collaborative tasks)

sperms characteristics ● the students study the available biological resources
(1 hour) ● they observe the morphological characteristics of leaves, flowers
and fruits, identifying the main characteristics of Angiopsperms;
these were written or drawn (leaves, flower) afterwards on the
worksheet.

Identification & ● place: biology lab ● teamwork - computer lab

description of ● usage guidelines ● usage guidelines
the species (1 ● use of a botanic atlas ● use of a KeyToNature interactive
hours) to identify five species key
● worksheets ● worksheets

Sperma- field training (1 ● visit in the park near the school ● identification of the species
tophytes hour) studied in the lab ● comparing the characteristics of different
species (Gymnosperms vs Angiosperms) ● discussing the
adaptations to climate.

Evalution (1 hour) ● evaluation test (see section 2.3 in this paper)

Tab. 1 – Design of the teaching scenario.

The training activities were organised in a learning unit of six hours, combining:
in-class training, visit outdoors, and evaluation. Tab. 2 shows a sample of the
student worksheet for Class B (studying Gymnosperms).

No. Name of Bole Leaves Seeds/ Areal/

species Cones Importance
1. Taxus Straight Acicular, A single seed Hardwood
baccata ascending flattened with a woody used for fine
skin, covered carpentry and
by a red, sculpture....
fleshy Aryl
2. Thuja Straight Small, scaly, Several seeds Cultivated as
orientalis have a included into a ornamental
dimple on the cone plant in
underside... gardens and
parks.
3. ... ... ... ... ...

Tab. 2 – Example from the worksheet used by students to fill in their responses.

369
2.3 Methods and tools used for evaluation

Knowledge evaluation was carried out using different methods: systematic

observation, self-assessment, correctness of identifications, as well as written
paper. Both classes received the same test. The evaluation tool contained:
short answer items, double choice items, pair items, questions on identifying
and classifying, other structured questions. The evaluation was centered
on: a) recognition of morphological characteristics of leaves and flowers of
Gymnosperms and Angiosperms, b) description of the general characteristics of
Spermatophytes, c) identification of species and their assignment to the correct
systematic group.
Several questions concerned the recognition of morphological characteristics
of leaves and flowers. The students in Class B (those who learned through the
KeyToNature tools) identified better the morphological characteristics of leaves
(Fig. 2) rathen than those of flowers (Fig. 3). This probablly depends on the fact
that the interactive keys of KeyToNature were mostly based on leaf morphology.
The knowledge of the general characteristics of Spermatophytes was checked
through short answers and double choice answers, with four statements on
Gymnosperms and four on Angiosperms. For Gymnosperms, students in Class
B had 25% more of correct answers compared with those of Class A, whereas for
Angiosperms the results were similar in both claasses. This probably depends
on the fact that Gymnosperms are a much smaller group with distinctive
characteristics, which are easily appreciated by all students.
Finally, we have evaluated the ability of students to use morphological
characteristics for identification and classification in the correct systematic
group. The students had to identify four species of gymnosperms and four of
angiosperms. Students from Class B had better results: most of them identified
and classified correctly all species (Fig. 4). For example, 18 students correctly
identified all species of Gymnosperms, 2 students correctly identified only 3
species, and 1 student only didn’t identify any species (Fig. 4).

20 20
18 18
16 16
14 14
No. students

No students

12 12
Class A Class A
10 10
Class B Class B
8 8
6 6
4 4
2 2
0 0
Correct answ er Wrong answ er Correct answ er Wrong answ er

Fig. 2 (left) – Evaluation item: leaves morphology.

Fig. 3 (right) – Evaluation item: flower morphology.

370
 20
18
16
14

No. students
12
Class A
10
8 Class B
6
4
2
0

species

species
Wrong

Wrong
answer

answer
species

species

species

species
4

3
2

2
Gymnosperms Angyosperms

Fig. 4 – Evaluation item: systematic recognition and classification of species (number of

correctly identified and classified species from a total of four).

These results indicate that the use of identification keys for the study of
biodiversity allows students to develop observing, research and analytical skills,
to enhance their intellectual work, and to improve their digital skills.

3 Conclusions
The study of biodiversity is one of the major goals of the scientific community
and of worlwide policies. Educational goals can be achieved through trans-
disciplinary approaches by bringing together traditional methods and innovative
teaching. The use of computer assisted-learning such as the identification keys
of KeyToNature proved to increase the quality and efficiency of the teaching–
learning– assessment process.

Acknowledgement

This work was carried out during the testing activities in the framework of the
KeyToNature Project (www.keytonature.eu), ECP-2006-EDU-410019. The authors
wish to thank the project coordinator, Prof. Pier Luigi Nimis, for the development of
the identification keys, and Prof. Mircea Giurgiu for providing access to the eLearning
environment and for continuous local collaboration.

References
[1] L. Cohen and L. Manion, Research Methods in Education, 4th edition, Routledge, London and
New York, 1994.
[2] V. Cristea and S. Denaeyer, De la biodiversitate la OMG-uri?, Eikon, Cluj-Napoca, 2004.
[3] L. Ezechil, Prelegeri de didactică generală, Paralela 45, Piteşti, 2003.
[4] V. Ghidra and M. Botu, Biodiversitate şi Bioconservare, Academic Press, Cluj-Napoca, 2004.
[5] M. Ionescu and I. Radu, Didactica modernă, Dacia, Cluj-Napoca, 2001.
[6] P. L. Nimis, S. Martellos, F. Boar, A. Kerekes, F. Crisan and M. Giurgiu, “A guide to the
Pteridophytes of Romania”, http://dbiodbs.univ.trieste.it/carso/chiavi_pub21?sc=412, July 2010.
[7] F. Boar, A. Kerekes, M Giurgiu and A. Homodi, “Spontaneous and Cultivated
Gymnosperms”, http://www.keytonature.eu/tools/gymnospermae/index.html, July 2010.

371
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 373-378.
ISBN 978-88-8303-295-0. EUT, 2010.

Digital Tools in the Botanical

Garden of Madrid
Marina Ferrer, Esther García

Abstract — This paper describes the different roles of the Real Jardin Botanico
of Madrid in the creation and use of digital tools related to biodiversity. The
strongest point of this historical institution is the large amount of scientific data
and items in its possession, which makes an excellent starting point for the
creation of digital tools. The involvement of the Jardin Botanico in the creation
of digital tools is centered on data providing, but some original tools have
been developed as well, such as E-Flora Iberica, an ambitious system based
on Flora Iberica. Another important role of the Botanical Garden of Madrid
is that of testing the digital identification tools with users, since it receives
about 500.000 visitors yearly, 50.000 of which are involved in formal education
activities.

Index Terms — data provider, digital tools, dichotomous keys, education,

Flora Iberica, wiki, workshops.

—————————— u ——————————

1 Introduction

T
he Botanical Garden of Madrid [1] has been involved in the use and
creation of digital tools related to the identification of organisms during the
last 3 years, as a partner of the European project KeyToNature. On one
hand, it has created its own tools, on the other hand, it has served as a test-bed
for identification tools developed in other contexts.

2 Data Providing
Being an institution with 250 years of history and a leading research center,
the Botanical Garden of Madrid [1] owns an rich heritage of botanical resources:
photographs, drawings, herbarium specimens, etc. Its most valuable contribution
comes from Flora Iberica [2], a large project started in 1980, that gathers
information on the flora of the Iberian Peninsula, with c. 4000 taxa described.
The quality and quantity of the data produced in this project allow to create
different tools for the identification of plant diversity. All of these data are freely

————————————————
The authors are with the Scientific Culture Department, Botanical Garden of Madrid, Claudio
Moyano 1, 28014 Madrid. E-mail: [email protected], [email protected].

373
available online, and thus completely accessible to the broad public. Since the
Iberian Peninsula is a biodiversity hotspot in Europe, this contribution covers
about 60% of the European flora.
Dichotomous keys and taxon pages for 130 families and 732 genera, for a
total of 3.560 species and 940 subspecies, were submitted to the online archive
of KeyToNature, as well as 12.978 images of the flora of the Iberian Peninsula.
In total, 242.330 metadata have been submitted.

3 Original identification tools

 
Fig. 1 – Distribution of metadata submitted to the KeyToNature online archive.

3.1 E-Flora Iberica

This is a digital identification tool which uses Flora Iberica as a raw material.
Using e-Flora Iberica [3] one can obtain a long dichotomous key that includes all
species, but can also create “mini identification keys” after setting up a series of
filters such as province, family, etc. Another important feature is that the system
can create a “minikey” out of some chosen species one wants to compare,
providing a dichotomous key based on the differences among those species.

3.2. Wiki-keys

Using a wiki template developed in the framework of KeyToNature, two keys

were created:
1. Wiki key of Gymnosperms of the Real Jardin Botanico - This is a dichotomous
key of the gymnosperm species present in the Garden. It first leads us to genera
(37 genera in the key), and for the genera that have more than one species
(most of them), another subkey is shown, leading to the species. Information
about the species and pictures (for most of them) are displayed in the taxon
pages.
2. Wiki key for the ferns of the Flora of Equatorial Guinea [5] - Based on
a modern flora of Equatorial Guinea, this wiki-key also follows a dichotomous
structure ans is enriched with images.

374
2.3 The virtual assistant [6]

An experimental “virtual assistant” to identify plants was developed in the late

phase of the KeyToNature project. This is a web application created from a very
simple key: Conifers of the Iberian Peninsula, in which a “digital lady” literally
“talks” to users, which can answer back by writing free text. Images can be
implemented during the “conversation” to illustrate the process. The advantage
with respect to a dichotomous key is that the user can skip several steps,
depending on the amount of information she or he gives through the process
of identification. The virtual assistant also has the quality of being amusing and
attractive for youngsters.

4 Test Bed For Tools

The Botanical Garden of Madrid [1] offers excellent conditions for tools trial
and testing, since it receives about 500.000 visitors yearly, 50.000 of them are
involved in public education activities.
Several workshops centered on the use of digital identification tools were held
at the Botanical Garden of Madrid [1] since the beginning of KeyToNature. Our
main interest on these workshops was to develop learning experiences based
on identification processes. The workshops participants’ came from all grades
of secondary school. 15 years old students were the most abundant age group
(4th year Secondary School in Spain), with 449 pupils.

 
Fig. 2 – Distribution of the different school grades participating in the bioidentification
workshops.

Most participants came from Madrid (65%), the rest from other municipalities
(35%). 92% were from public schools, 7% from private schools and 1% from
cultural centers.

375
4.1 Observation and Identification workshops

In total, 75 workshops took place under the framework of KeyToNature during

3 years. 69 of them were at the Botanical Garden of Madrid [1]. The workshops
were carried with by groups of about 20 students accompanied by their teachers
and assisted in the Garden by two previously trained instructors. The workshop
consisted in 3 different stages: [7]
1. Brief explanation: The students were briefly introduced to the project and to
the main features of the plants they would observe during the exercise: types of
leaves, their disposal along the branch, whether they fall in winter or not, color
and texture of tree bark.
2. Observation: Once the students knew what to pay attention at, they start
drawing the plants or trees that the instructors had previously chosen. The
identification tags of the plants were covered, to keep them “anonymous”.
3. Identification: When sketches were finished, the group is conducted to the
computer room of the Botanical Garden. There, they use an online identification
key to identify the specimens. This key was produced by the Dyades project, the
Italian branch of KeyToNature, for the Botanical Garden of Madrid [1], and in a
simple and an advanced version [8]. The key is enriched with many images the
pupils can compare with their drawings, in a didactic and entertaining process
were the concepts learnt on the first two steps are efficiently fixed. This part
lasts about 45 minutes.

4.2 Winter workshops

Thinking of the worst winter conditions in which the workshops could take
place, a set of laminated leaves was prepared, which allowed students to have
a close look at the leaves also in winter. During this exercise, even when they
stood in front of a naked trunk, students had the opportunity to look at a well-
preserved little branch. These waterproof resources served as a great support
for the activity.

4.3 Other Workshops

Similar workshops took place elsewere: in the Environmental Center

Villaviciosa de Odón, the Botanical Garden of Alcalá and in the Natural Area “El
Mesto” near Madrid by the Secondary Centre “San Agustín de Guadalix”. These
other centers had their own keys created especially for their needs, also in this
case by the Dryades project [9].

Finally, the Dryades project also created another key for Spain: the Key for
Trees and Shrubs of Catalonia [9], translated both in Spanish and Catalan, and
posted in the web page of a public Catalan Centre of Science, the Centre de
Documentació i Experimentació en Ciències.

376
   Fig.3 – Location of the workshops.

5 Keys on mobiles
We had the chance to try the Dryades key developed for the Botanical Garden
of Madrid [1] on a PDA. This experience was carried out on the “Scientific
Weekend”, a science fair that took place in May 2010 at the National Museum
of Science and Technology. An instructor was placed on a KeyToNature stand
with two PDAs and a series of laminated leaves. The public that approached the
stand (kids and adults) was able to try the key by using the mobile device. The
activity attracted many visitors who found it very interesting.

6 Conclusion
The participation of the Real Jardin Botanico of Madrid [1] in the KeyToNature
project as a data provider has put at disposal of the project the taxonomic,
ecologic and biogeographical information for c. 4000 taxa of vascular plants
from the Iberian Peninsula, as well as their images. This is an optimum data
set for the development of digital tools for identification, due to the quality and
rigour of the information generated by the Flora Iberica project along 30 years.
Furthemore, the Garden has created the “e-flora” digital tool, and different
identification keys in a wiki-format. It also performed as a tester of the digital
tools generated under the KeyToNature project, with 73 experiences carried out
with c. 1.400 students. The interest that these tools generate and their efficacy
for the teaching of Natural Sciences was evident.

Acknowledgement

This paper was produced in the framework of the project KeyToNature, funded by the EU
in the eContentplus programme.

377
References
[1] Botanical Garden of Madrid: http://www.rjb.csic.es/jardinbotanico/jardin/, 2010.
[2] Flora Iberica: http://www.floraiberica.org/, 2010.
[3] E-Flora Iberica: http://www.efloraiberica.es/eflora/, 2010.
[4] Wiki key for Gymnosperms of RJB: http://www.keytonature.eu/wiki/Clave_de_Gimnospermas_
RJB, 2010.
[5] Wiki key for the ferns of the Flora of Equatorial Guinea: http://www.keytonature.eu/wiki/Clave_
de_familias_de_Pteridophyta_de_la_Flora_de_Guinea_Ecuatorial_%28RJB%29, 2010.
[6] Virtual assistant: http://www.dialgraph.com/av/?idAsistente=1404151, 2010.
[7] Video “Observation and Identification Workshop” http://www.youtube.com/user/
RJBCSIC#p/u/1/EDSVIJXsjEk, 2010.
[8] Links for the keys used at the Workshops: Simple version: http://dbiodbs.units.it/carso/chiavi_
pub21?sc=119. Advanced version: http://dbiodbs.units.it/carso/chiavi_pub21?sc=165, 2010.
[9] Other Dryades tools created for Spain: Trees and shrubs of Catalonia: http://dbiodbs.units.it/
carso/chiavi_pub21?sc=346. Key for the natural area “El Mesto” (Madrid): http://dbiodbs.units.
it/carso/chiavi_pub21?sc=296. Key for the Nat. Area “El Forestal” (Vilaviciosa de Odón): http://
dbiodbs.units.it/carso/chiavi_pub21?sc=311, 2010.

378
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 379-381.
ISBN 978-88-8303-295-0. EUT, 2010.

Use of KeyToNature
Identification Tools in the
Schools of Slovenia
Irena Kodele Krašna

Abstract — This paper presents some results of the testing of several new
interactive e-tools for learning and teaching biodiversity in the schools of
Slovenia. The tools were developed in the frame of the ongoing eContentplus
European project KeyToNature, in such a way to make them tailored to
the needs of different educational users. We found out that identification of
organisms with these tools is not only easy for students, but also for primary
school children who had just learned to read.

Index Terms — biology, education, lesson, identification, KeyToNature,

natural science.

—————————— u ——————————

1 Introduction

I
n the initial phase of the project KeyToNature we examined the need for
identification tools in biology teaching. After this phase, we produced the first
interactive identification keys in Slovene, and we tested them in elementary and
secondary schools. In educational seminars for teachers we presented the new
tools and the possibilities of their use. We showed teachers how our identification
tools could be customized and adapted to their needs. Until now, over 50 tools
were created in the Slovene language. All of them are freely accessible via the
Internet and teachers are widely using them. Based on the responses of teachers,
we greatly improved our identification tools, also developing several scenarios for
their use in the educational system of the Country.

2 Methods
New interactive identification tools were tested in primary and secondary
schools in Slovenia from June 2008 to June 2010. Teachers were invited to
participate through the media and personal at meetings of study groups.
Analysis of usability and applicability of our tools occurred on the basis of
questionnaires sent back by teachers. Testings on the upper level of primary
school and secondary schools were done by the teachers themselves, while on
————————————————
The author is with the Slovenian Museum of Natural History, Prešernova 20, P.O.Box 290, 1001
Ljubljana, Slovenia.

379
the primary level (pupils aged 7–10) biologist (a KeyToNature project assistant
in the Natural History Museum of Slovenia) helped the teachers. The usability
and applicability of the identification tools were tested in different environments -
in the computer lab and in the field using laptops and mobile phones. In the case
of larger groups (over 40 students), for field-work we used printed versions of
the identification keys. Pupils of the lower grades of primary schools have used
smaller, customized keys (e.g. that to the plants of the lawn in front of the school
in the village of Budanje).

3 Results
Completed questionnaires were returned by 19 teachers, 5 from the lower
level and 12 from the upper level of elementary school, and 2 from high school
teachers. A total of 710 students participated in the testing. The most used
identification tool was an Interactive Guide to woody plants of Slovenia (31%).
In 5 cases (26%) students worked with a key which was designed for a specific
school (customized key). We mainly used the dichotomous interface of the keys
(90%), because we soon realised that keys with a more complex approach -
such as multi-entry keys - require more background knowledge and experience
from students.
The interactive keys proved to be a very useful tool both in regular classes
and in the other activities as science days and biology circles. After a brief
introduction and explanation of basic terms used in the keys, pupils of the lower
level of primary schools were capable to autonomously identify the organisms.
While using a key, students learned about plants and animals specific to a certain
habitat (meadow, forest, stream ...), observed their morphology, classified
plants into the system. They also developed a functional way of reading,
practicing cooperative learning and self-study. All teachers who participated in
the survey stated that they would like to repeat the activity. Pupils enjoyed using
the interactive keys because they had an active role and were independent at
work. The activities proved to be successful independently from the form of the
identification key which was used (interactive keys on the computer or printed
on paper) and from where the activity took place (in the classroom or outdoors)
(Fig. 1). The vast majority (80%) of teachers noticed that students were more
interested in biodiversity after having used our keys, and several students even
used them at home with their families.

Fig. 1 – Activities with the KeyToNature identification keys (a – pupils of the lower level
of primary schools with a printed key, b – students of the higher level of primary schools
with a stand-alone identification key on laptop, c – secondary school student working
with an online identification key).

380
4 Conclusion
The identification tools – once adapted to users’ needs – proved to be well
suited for self-study or cooperative learning. They enable individualization and
differentiation of lessons. The students like them very much, and the keys do
not only add variety to lessons, but also make learning and the achievement of
educational objectives easier: knowledge acquired in this way is more lasting.

Acknowledgement

I wish to thank Tomi Trilar for giving me the opportunity of working in KeyToNature. I am
very grateful to The University of Trieste (Italy), especially Pier Luigi Nimis and Stefano
Martelos, who have created for us several keys in the Slovene language. I would like to
thank also Sonia Hetzner and Gerd Schmidt, who prepared the questionnaires, and the
teachers who contributed in testings and gave us important feedback information. This
paper was produced in the framework of the the project KeyToNature (www.keytonature.
eu, ECP-2006-EDU-410019), funded in the eContentplus Programme.

References
[1] V. Trošt Vidic, “Learning about grassland plants in Podnanos, Slovenia”, KeyToNature
Teacher’s Handbook, http://www.keytonature.eu/handbook/Learning_about_grassland_
plants_in_Podnanos,_Slovenia, 2010.
[2] S. Mozetič, “Learning about grassland plants in Šempeter”, KeyToNature Teacher’s
Handbook, http://www.keytonature.eu/handbook/Learning_about_grassland_plants_
in_%C5%A0empeter, 2010.
[3] N. Šefer, “Trees and shrubs around the school”, KeyToNature Teacher’s Handbook, http://
www.keytonature.eu/handbook/Trees_and_shrubs_around_the_school, 2010.
[4] M. Žohar, “Trees and shrubs in Laško”, KeyToNature Teacher’s Handbook, http://www.
keytonature.eu/handbook/Trees_and_shrubs_in_La%C5%A1ko, 2010.
[5] A. Sodja, “Natural science lessons and identification keys”, KeyToNature Teacher’s
Handbook, http://www.keytonature.eu/handbook/Natural_science_lessons_and_
identification_keys, 2010.
[6] K. Prosen, “Let’s learn about the families of vascular plants”, KeyToNature Teacher’s
Handbook, http://www.keytonature.eu/handbook/Let%E2%80%99s_learn_about_the_
families_of vascular_plants, 2010.

381
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 383-387.
ISBN 978-88-8303-295-0. EUT, 2010.

New key-tools for pollen

identification in research
and education
Jade Dupont, Nathalie Combourieu Nebout, Jean-Pierre Cazet,
Florian Causse, Régine Vignes Lebbe

Abstract — Pollen ID offers a free and easy access to various palynological

information and compiles in the same web-space a pollen database and
different services through a friendly user interface. Pollen ID proposes,
or will propose, pollen and plant descriptions, terminology learning with an
illustrated glossary and interactive images, identification keys, pollen analysis,
pollen diagram construction, links with vegetation and climate data. The
Pollen ID project is presently restricted to the European and Mediterranean
geographical area, but it will be extended to other regions as well. This project
is still in progress; its content and user interface – presently in French - will be
soon available in English. In its final shape, the Pollen ID project will include
palynological applications such as pollen determination tests, several original
pollen analysis exercises with representations in diagrams and an easy
interpretation of vegetation and climate. Pollen ID is accessible on http://lis-
upmc.snv.jussieu.fr/pollen/.

Index Terms — pollen, free access key, identification.

—————————— u ——————————

1 Introduction

P
alynology is commonly used in allergology, ecology, environmental
reconstruction, climatology, and geology. Recently, it has been added
in the current college program in France. Pollen identification by using
books and online database is now largely used in palynology. Nevertheless
numerous websites do not provide resource access for a large audience, from
school education to research. Moreover, most of the websites do not link pollen
data with the plant description and do not associate pollen applications to the
descriptive content (our preliminary review has selected and analysed the
————————————————
Jade Dupont was a master student of the University Paris VI. E-mail: [email protected].
Nathalie Combourieu Nebout, Jean-Pierre Cazet are with the LSCE - Laboratoire des Sciences
du Climat et de l’Environnement, UMR 8212 CNRS/CEA/UVSQ, Domaine du CNRS, avenue de la
Terrasse, F-91198 Gif sur Yvette cedex, France. E-mail: [email protected].
Florian Causse and Régine Vignes Lebbe are with the LIS - Laboratoire Informatique et Systéma-
tique, UMR 7207 CNRS/MNHN/UPMC, MNHN Département Histoire de la Terre, CP48, 57 rue
Cuvier, 75005 Paris, France.

383
content of 16 websites dedicated to the pollen description and identification).
The Pollen ID project tries to take the challenge to fill this gap.

2 What is Pollen ID ?
Pollen ID offers a free and easy access to various palynological information
and compiles in the same web-space a pollen database and different services
through a friendly user interface. Pollen ID proposes, or will propose, pollen
and plant descriptions, terminology learning with an illustrated glossary
and interactive images, identification keys, pollen analysis, pollen diagram
construction, links with vegetation and climate.

2.1 Pollen ID components

Resources are stored in two types of information systems: a relational

database (MySQL) and knowledge bases for descriptive data (Xper2) (Fig. 1).

  Fig. 1 – Pollen ID information technology architecture.

The relational database stores 1) nomenclatural data, plant information,

photos and movies for the taxa, 2) definitions, pictures with image maps for the
glossary terms, and 3) texts for the UI.
The descriptions of pollen are stored in two Xper2 knowledge bases, and
presently concern 94 Mediterranean species documented by texts, images and
67 videos:
• The first knowledge base is dedicated to beginners for training on pollen
identification. 56 descriptors were selected, documented by text and
images.
• The second one is more exhaustive and designed for expert users. It
includes 115 descriptors with definitions and images.
The web pages are then dynamically constructed from the different resources
(see Fig. 4).

384
3 Pollen ID use
The Pollen ID website user interface provides an original and large access to
the complementary resources. Through the interactive buttons of the Home page,
the user will discover numerous information from generalities on palynology,
pollen descriptions and images to applications (in future developments) (Fig. 2)

Fig. 2 – Pollen ID interface

(1) the menu, (2) Access to on-line identification services, (3) direct access to photos
and videos, (4) details of the selected menu, (5) about Pollen ID.

The pollen ID project pays a special attention to beginners with the production
of a rich information on palynology, from pollen extraction techniques to a
glossary and interactive images and films for basic training.

3.1 Learning pollen morphology

In pollen ID, the user can explore easily all data: definitions, drawings, and
images when necessary. The glossary has been inspired from [2]. In each page,
hyperlinks are coloured and can redirect towards definitions and their associated
drawings. Interactive drawings are managed by the rollover technique allowing
the users to explore, to discover and to be familiarized with the terminology
of pollen anatomy (Fig. 4). Then the beginner can learn the basic concepts of
palynology.
The interface also combines real views (pollen and plant photos and pollen
observation movies). The videos are constructed from a sequence of pictures
(microscope X60), about 50 photos for each pollen, to have a good view of the
total volume of pollen. The user can stop the movies as he wants, to compare
with drawings describing anatomic structures. In the future, the ID project will
intend to propose pollen photos with superimposed drawings in order to show
the pollen characters directly on pollen views.

385
 Fig. 3 – Example of an interactive drawing. The rollover technique highlights the
morphological structure pointed by the user and displays the related definitions.

3.2 Pollen ID identification keys

Pollen ID includes at this step: two interactive free access keys, one dedicated
to beginners and the other to advanced researchers. These two keys have been
refined after [1]. They use the Xper2 identification process and are a free access
system available on-line. At the end of identification process, users can access
the complete information on the taxon (Fig. 4).

3.3 Pollen ID Taxonomic forms

Pollen ID produces taxonomic forms that gather nomenclatural data,

geographical information, textual pollen descriptions, pollen images and

Fig. 4 – Taxonomic form example. (1) textual description (the part “DESCRIPTION” is
produced in natural language from the structured Xper2 knowledge base), (2) pollen
photos access, (3) movies access, (4) plant photos access, (5) external links. Some
words are hyperlinked to definitions in the glossary.

386
movies, plants photos and links to external websites (Fig. 4). All textual pollen
descriptions are automatically generated from the Xper2 structured descriptive
data with added hyperlinks to the pollen glossary. Thus, the user can go through
the different items to have a look at all information, and all the parts of the
website are consistent.

3.4 Taxonomic research

The user can also find directly a taxon by choosing it in the lists of forms, photos
or movies. A classification is available with scientific and vernacular names. In
the future the project will include an interface to construct diagrams from pollen
inventories, and links with vegetation and climate data for environmental studies
and archeo-paleo climatic reconstructions.

4 Conclusion
The Pollen ID project is presently restricted on the European and Mediterranean
geographical area, but it will be enlarged to other regions as well. This project
is still in progress; its content and user interface – presently in French - will be
available soon in English. In its final shape, the Pollen ID project will include
palynological applications such as pollen determination tests, several original
pollen analysis exercises with representations in diagrams, and an easy
vegetation and climate interpretation. Pollen ID is accessible at http://lis-upmc.
snv.jussieu.fr/pollen/.

Acknowledgement

This work has been done with CNRS, MNHN and UPMC funds. We thank JP Cazet for
pollen process, and pollen photos and movies production..

References
[1] J. Lebbe, S. Nilsson, J. Praglowski, R. Vignes and M. Hideux, “A microcomputer-aided method
for identification of airborne pollen grains and spores”, Grana, vol. 26, pp. 223-229, 1987.
[2] W. Punt, P. P. Hoen, S. Blackmore, S. Nilsson and A. Le Thomas, “Glossary of pollen and
spore terminology”, Review of Palaeobotany and Palynology, vol. 143 (1-2), pp. 1-8, 2007.

387
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 389-393.
ISBN 978-88-8303-295-0. EUT, 2010.

The UK urban tree survey

Bob Press

Abstract — Public communication of science shares various educational

aspects with more formal learning experiences but differences in both the
aims and the target audiences make for subtly different approaches. The UK
urban tree survey is an example of the requirements and possibilities of a
public-orientated project relying on the use of identification tools.

Index Terms — trees, identification tools, survey, public communication of

science.

—————————— u ——————————

1 Introduction

D
epending on the target audience, projects with an educational aspect
need to adopt different approaches to learning. Public communication
of science differs in small but significant ways from projects exclusively
aimed at formal learning in schools (Tab. 1).

Schools (formal learning) Public (informal learning)

Not voluntary for participants Entirely voluntary for participants

Linked to curriculum Linked to personal interest

Narrow within-group ability range Broad within-group ability range

Direct support available Little/no direct support available

Participants use results Participants do not use results

Tab. 1 – Considerations for formal versus informal learning.

As a project relying on the active participation of the general public, the UK

tree survey (launched in July 2010) is an example of a wide-ranging educational
activity but with a focus on public participation and demonstrates some of the
ways in which these differences can be addressed.

2 Context
Why survey trees at all – especially in the UK which has one of the best known
————————————————
Bob (J.R.) Press is with the Natural History Museum, Cromwell Road, London, SW7 5BD. E-mail:
[email protected].

389
and most recorded floras in the world? Trees are currently a focus of interest
in the UK, following studies such as that on urban health and forestry [1], and
investigations into the potential role of trees in areas such as carbon off-setting.
The Department of the Environment, Fisheries and Agriculture (Defra) recently
announced plans to plant 1 miilion trees across the UK.
In addition to the wild flora, there are copious (though patchy) data relating
to trees managed by such organisations as the Forestry Commision and local
authorities. Despite this, the urban forest remains relatively little known. In
particular, data on trees in private gardens represent something of a black hole.
Against this background, the Natural History Museum in London decided to
launch a web-based survey of urban trees. For a survey of this type public
participation is vital since only members the public have access to many of
the areas of interest – especially to private gardens. Since public participation
in such surveys is entirely voluntary, it is intimately linked to personal interest.
Previous experience (Bluebells survey, OPAL surveys) indicates a strong
willingness by the public to take part in what they perceive as ‘real science’ as
long as the scientific reasons behind the project are clear.
The scientific aims of the urban tree survey are:
1. Gaining a more precise understanding of the make-up of the urban forest
e.g. the constituent species
2. Changes in tree demographics (and perhaps their causes)
3. Potential impact on (and changes in) other wildlife relying on trees
4. To gain phenological data and insights into the effects of climate change
(changes in the seasons as indicated by flowering times, which species are
now flourishing where in the UK etc).
Trees are highly visible within the urban environment and have well-known
benefits for urban populations. The scientific aims of the project are easily seen
as relevant to the public at large.

3 The urban tree survey

A major factor in designing the survey is that the public user has no on-the-spot
expert to consult directly such as a botanist or teacher; essentially the user is
working alone. For the identification and mapping tools in particular, clarity and
ease of use are essential, with an obvious progression through the necessary
steps. Clear, unequivocal instructions are needed at every major decision point
lest the user lose interest and stop.
A pilot survey focusing exclusively on cherries was run from April to June
2010. This garnered extensive media attention at a national level and received
huge support from the public. The experience was used to modify the main
survey before it was launched in July 2010 (see 6. Lessons learned).

3.1 Species included in the survey

A total of 80 ‘trees’ were included in the survey. Some of these are individual
species, others groups of species and yet others are genera. This apparent
complication is necessary for three reasons. 1) Given that we do not know

390
precisely which taxa occur in the areas to be surveyed, producing a definitive
key to individual species is impossible. One of the reasons for the survey is to
gain an initial idea of which taxa are present and then to refine the keys in the
light of these data. 2) A key including every possibility would be too large and
unwieldy. 3) There are some taxa e.g. Sorbus which are simply too difficult for
non-experts. A reduction to groups enables users to cope with taxonomically
difficult trees.

3.2 Identification tools

Prevailing wisdom is that keys for the public must be short, simple and entirely
devoid of technical terminology. We know from KeyToNature’s work that this is
not necessarily true for schools and it is no more true for the public. However,
keys which are obscure or fussy are confusing and counter productive for non-
experts, so presentation is a major consideration. The choices at each step
need to be clear and it must be possible to retrace the steps to correct any
wrong turning.
The key provided is interactive and uses simple illustrations, images and text.
As each step is negotiated it recedes on the screen while the next step takes
centre screen. This guides the user through the key while previous steps remain
visible to aid backtracking. An interactive version for i-phones and a printable
key are also available.

3.3 Factsheets

A key alone is insufficient for identification purposes when there is no expert

present to confirm the result. Thus, each exit point from the key is linked to a fact
sheet for that species, enabling the user to confirm their identification.

3.4 Mapping

Familiarity breeds confidence. We used Google Earth, a mapping tool already

familiar to many people. Our system is a hybrid of aerial photographs and street
maps. Offering easy access is important so locations can be identified by several
methods: by post code; grid reference; latitude/ longitude; or by simply zooming
in. The simple system encourages users to continue and helps to avoid errors
when adding records.

3.5 Recording data

A tree is added to the map with a click of a mouse and the user is asked to
record data relating to it. While sufficient data to make the survey worthwhile
is required there are limitations on what the users can provide due to their lack
of expertise. There may also be a lack of support or encouragement (school
children have a teacher to prompt them to go one step further, the public have
no such presence).
We restricted data to eight fields including date of the record, identification,

391
type of site (private garden, street, park etc) and size of tree. This is in line with
the concept of keeping the effort required to achieve a result to a minimum. The
fields use drop-downs to ensure consistency of data entry.
When dealing with the public, plant names probably cause more
misunderstandings than any other factor. The public prefer to use vernacular
names raising all manner of complications. A particular problem arose with
cultivars, of which there are many hundreds. They are often sold under names
such as Prunus ‘Amanogawa’ (= Prunus serrulata) or Prunus ‘pandora’ (=
P. yedoensis). Some users were convinced these were species names and,
instead of keying the tree out, simply believed their species had been omitted
and did not enter a record.
We provided fields for both vernacular and scientific names, with an autofill
function for the field not selected. Again, drop-down lists prevent ‘false’ names
from being added.
There is an option to upload up to three images of the specimen which helps
the survey team to monitor records. No free text was allowed in any field. This is
a safeguard against individuals posting abuse on the web site.

4 Other support
Given that the target audience is working without direct support, it is important
to provide as much additional information as possible. The web site offers tips
on tree identification, an on-line identification forum, glossary and references to
other identification tools. There is an entire section devoted to learning support
for schools wishing to use the survey.

5 Results
To date (July 2010), over 5000 trees have been recorded and mapped [2].
There have been more than 100,000 web site visits, with a bounce rate of 25%.
This is a baseline survey and one in its early stages. Statistical validity
considerations aside, we are only able to make a generalised analysis of results
for now. However, from the pilot survey we already have a snap-shot of Prunus
species in the UK urban forest, with some ideas of which species are most
frequent in different site types, which are most widespread and so on.
Providing feedback is vital to maintain the connection with the audience. The
public do not themselves use the results of surveys such as these but they do
want to see what difference their efforts have made. The initial results of the
cherry tree survey have already been posted on the web site.

6 Lessons learned
The pilot project was informative both in terms of the data gathered and in the
challenges of conducting a survey of this kind. The positive lessons learned are:
1. The public are keen to participate – as long as they can be attracted by the
project (it is voluntary)
2. The public can cope with quite sophisticated methods

392
3. The project must be linked to a real scientific question/investigation and tie
in with personal interest rather than a curriculum
4. Sufficient data can be collected
5. Unexpected data appeared e.g. on harvesting and cultural practices
associated with cherries
6. Contacts with other groups were made (e.g. The Orchards Initiative, local
authorities, tree warden groups)
There are also negatives:
1. Data accuracy and data usability - since the public do not actualy use the
data, they may be less interested in its accuracy
2. Verification of data – a major factor but one we must live with. Only the
recorders have access to the sites although posting images is of considerable
help in weeding out wrong or false records
3. Technical problems e.g. Explorer initially failed to cope with the number of
records on the map. This and the point above relate to lack of support or
how we provide support
4. Such projects risk continually re-inventing the wheel
5. There is no clear system for migrating data upwards/outwards to other
organisations and potential users e.g. GBIF

7 Next steps

The cherry tree survey has now been subsumed within the similar but larger
project covering all urban trees in the UK [3]. Both surveys will be refined in light
of the results received by the end of 2010 and repeated for two more years. All
data will be made available to other users.

8 Conclusion
Surveys based on identification of organisms can be remarkably successful
on two fronts: in gathering scientific data and in communicating science to the
public. As with any such project, careful consideration of the audience’s needs
are paramount. Clear, consistent presentation of information (including the
reasons behind the tasks) and using familiar technology are also requirements.

Acknowledgement

The author wishes to thank Kate Evans, Claire Gilby, Sam Rae, Sheila Sang, Mike
Sadka, and Philippa Watson for contributions to preparing and launching the web site.
This work was supported in part by a grant from the Gulbenkian Foundation.

References
[1] Liz O’Brien, Kathryn Williams and Amy Stewart, “Urban health and health inequalities and the
role of urban forestry in Britain: A review”. Forest Research, 2010.
[2] http://www.nhm.ac.uk/nature-online/british-natural-history/urban-tree-survey/results-map/
index.php, 2010.
[3] http://www.nhm.ac.uk/nature-online/british-natural-history/urban-tree-survey/index.html,
2010.

393
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 395-400.
ISBN 978-88-8303-295-0. EUT, 2010.

Tree School – A new Innovation

for Science and Education
Della Hopkins, Karen James

Abstract — The current decline in the popularity of science at school has

resulted in many current activities to encourage and enthuse children in these
subjects. One such initiative has been born from a partnership between the
Natural History Museum, London, and the Cothill Educational Trust, which
aims to engage children in current biodiversity science. This project, named
Tree School, combines traditional methods used in botanical taxonomy with
new research activities using DNA barcoding to investigate local biodiversity,
and the applications of these techniques on a larger scale. The challenges
faced and potential opportunities realised have been investigated with a
series of pilot workshops, in preparation for the two-year operational phase
in 2011-12.

Index Terms — DNA barcoding, identification, trees.

—————————— u ——————————

1 Introduction

T
he current decline in popularity of the science subjects for school age
children has been recognised both in the education sector and also by
scientists. With fewer children choosing to continue to study science, there
may be a shortage of scientifically literate and trained professionals in the future.
A recent Research Councils report highlights this predicament, encouraging
the development of innovative initiatives between scientific researchers and
schools [1].
Scientists at The Natural History Museum (NHM) in London were keen to
address this problem, and began to develop ideas into how modern scientific
techniques could be effectively communicated to school children. Public
engagement and learning already play a huge role within the Museum; extending
these principals through the scientific researchers was an obvious development.
Simultaneously, the Cothill Education Trust was considering ways to excite
and encourage the children at their schools in science. A partnership was formed
between the NHM and the Cothill Trust, and a new initiative developed with the
aim of enabling current scientific practices to be understood and undertaken by

————————————————
D. Hopkins is with the Natural History Museum, London, SW7 5BD. E-mail: [email protected].

395
children. The Tree School project forms a continuing collaboration between the
two parties, to encourage a greater interest in science in children.

2 The Tree School

The scientific aim of Tree School was to design and pilot ways of involving
children and other non-experts in biodiversity science, specifically international
DNA barcoding campaigns. Pupils from the Cothill Educational Trust schools, as
well as others from schools across the country, collaborate with NHM researchers
to collect scientifically relevant data in this emerging scientific field. This is
achieved by attending a five-day workshop at the Old Malthouse School, a fully-
equipped educational centre in Dorset, which combines traditional classroom,
practical and outdoor teaching techniques. The Old Malthouse accommodates
the pupils, teachers and scientists comfortably, allowing a relaxed atmosphere
for both study and leisure time.
The workshop incorporates the more traditional methods used in taxonomy
and identification with modern laboratory-based work. This enables a deeper
understanding of the subject, and also demonstrates how different research
overlaps and compliments the other. The trees situated on the Old Malthouse
grounds are being used, with the children investigating different techniques to
identify the trees to species level.

2.1 Taxonomy

Field work is carried out at the beginning of the week, where the tried and
tested methods still used by botanists in the field today are demonstrated. Each
pupil chooses a tree, and captures the relevant data, such as a GPS point,
relevant observations and the correct species name. Identifications are made
using a variety of binomial keys and pictures, namely:
KeyToNature: Key to trees at the Old Malthouse [2],
The Field Studies Council: ‘Tree Name Trail’ key to common trees [3],
The Woodland Trust: Leaf identification swatch book [4],
The Collins Tree Guide [5].
A specimen from the tree is also taken and pressed, which is later made into
a herbarium sheet incorporating all of the gathered information.

2.2 DNA Barcoding

Leaf samples are also taken from the trees for use in DNA barcoding. This is a
compelling new tool, promising dramatic improvements in the rate and accuracy
of biodiversity inventories, and the potential to identify samples to a species
level. Within the fully-equipped laboratory, the children carry out DNA extraction
and PCR amplification and evaluation for their own specimens. These extracts
are then taken to the NHM sequencing laboratory, and the DNA sequences
produced. The results are added to TreeBOLD [6], the international barcode of
life data system, where they will be available to the scientific community.

396
3 Pilot Workshops
In 2011, a series of pilot workshops were undertaken to develop these ideas,
concentrating on the suitability of the techniques and the resources used.
Opinion has been sought from the teachers, children and scientific researchers
attending these workshops, with the findings and lessons learnt used to finalise
preparations of a series of workshops to take place during 2011-2012. It is
anticipated that pupils from both private and state schools will attend these
workshops.
A series of questionnaires were carried out both before and after attending
Tree School for the children and the teachers. Thirty five children aged between
eleven and twelve were questioned, from two different schools and currently in
year 7 of the UK system. A series of background questions were asked upon
their arrival, including their favourite subject, whether they enjoyed science at
school, whether they would like to work in science when they are older, and their
perception of a scientist. Three teachers present for the entirety of the course
participated in the survey, comparing their perceptions of Tree School with their
findings having attended.

3.1 Teachers

The common concern prior to Tree School was that the work would be too
advanced for the age group attending. Conversely, each participant stated that
any opportunity for students to broaden their horizons can only be a positive
experience. The teachers also expressed a frustration that there is little
opportunity for them to explore new aspects of science, or develop children’s
areas of interest, due to syllabus constraints. Most work taught during term-time
is exam-driven, which of course is a necessity.
At the end of the week, the teachers were pleasantly surprised at the amount
the children had understood, and their level of engagement in both classroom
lessons and practical sessions.

3.2 Pupils

Upon arrival at Tree School, the most popular subject proved to be French,
with around one third of the children naming it as their favourite subject. The next
most popular subjects were History, Geography and Art, followed by Science
with four votes. English, Maths and Latin were also mentioned. The majority of
the children stated that they enjoy science at school, although very few have
considered continuing with science in the future. Most of the group were unsure
about what they wanted to do as a career.
An interesting observation is the high number of children who chose French
as their favourite subject. The Cothill Trust has one school in France, the
Château de Sauveterre, where the children spend one term learning French
intensively to a much greater extent than can be covered in normal lesson time.
This has obviously had a positive effect on their enjoyment of the subject. It can
therefore be assumed that with extra insight and learning into a subject, children

397
can easily be enthused and encouraged to pursue a subject in the future. With
72% of the children already enjoying science at school, but the same number
unsure as to what they would like to do as a career, it is possible that more
children could choose to study science at a higher level if given the opportunity
to engage in real science at a younger age.

3.3 Scientists

The workshops also proved an enlightening experience for the research

scientists from the NHM. It was felt that the information was communicated
effectively, supported by informal reports back from the teachers, students and
even the students’ parents. The level of engagement with the children grew
over the week, leading to some thought provoking discussions. It provided
a new perspective on the subject, as children ask very different questions to
those usually posed by peers and other academics, and resulted in a two-way
enthusiasm between the scientists and the students.
It was also interesting to see how the children reacted to real-life experimental
problems faced by scientists. Schools usually teach canned experiments with
known outcomes, so pupils are usually able to achieve the correct result, or
know where they may have made a mistake. Often, a species may not appear
in a guide book, which encourages the child to think about how they may be able
to identify their tree. Also, the processes used to extract and amplify the DNA
can often fail, for no obvious reason, which can often happen.

4 The Workload
It is obviously imperative to ensure that the aims of the Tree School are fully
met for each group of children attending the workshops. It is important that they
end the week with a proven increased knowledge of identification and scientific
methodologies, and also hopefully an increased enthusiasm for botanical
science.
Much of this can be gauged from the question and answer sessions during
classroom and outdoor activities. In addition to queries for clarification, inquisitive
questioning increased and many sensible further and leading questions were
asked as the week progressed. This was particularly exciting to see, as many
of the children became engaged in the project and were inspired to find out
additional information.
Feedback from the children was important in ascertaining the work load,
and level of understanding of the classroom aspects. All attending pupils were
questioned on the level of the work undertaken – was it too easy, about right, or
too difficult. Over 90% of the children claimed the work was about right, stating
that although much of it sounded complicated, the methods were easy to carry
out with well-explained instructions and help. Three participants thought the
work was too easy, and none that it was too hard. Whilst this is a positive result,
it is important to consider preparing some additional information for those pupils
wishing to investigate the subject further.
Direct testing was also carried out at the end of the week in the form of a

398
tree identification quiz. A selection of trees was labelled, and the children were
given the task of identifying them using the resources and techniques given
to them previously. The results were encouraging, with many achieving full
marks. Again, the children were questioned about this task, and the different
keys available to them. The KeyToNature key, developed especially for the Tree
School project, was a clear favourite with 23 of the 35 children choosing this as
the most user-friendly resource.

5 Opportunities and Challenges

There is a huge opportunity to develop further projects using the same
template at the Old Malthouse. The school is a short distance from the
Jurassic coast, affording opportunities to study geological formations and other
geographic principals. Additional biological courses could also be offered, for
example studies into the invertebrate communities of the site, associated with
different habitats or plant species. Many of these ideas could be developed in
conjunction with other departments within the NHM or could be provided by
expertise available directly to the Cothill Trust.
Whilst the opportunity to develop the project at the Old Malthouse is an
exciting and achievable prospect, the possibilities of extending the format on
a wider scale, even internationally, is not straightforward. Whilst the concept is
easily expanded, the varying requirements of different educational structures
must be established fully before undertaking the work. This will involve working
closely with the key educators to determine prior knowledge, as curriculum may
vary greatly in different countries.

6 The Bigger Scientific Picture

Whilst the Tree School project aims to provide scientific skills to children,
and engage them in cutting-edge science, the project also is invaluable to
current scientific research. Currently, an international consortium of scientists
are working to build libraries of short, specimen-linked DNA sequences against
which unknown specimens might be identified to the species level. This will only
succeed as an identification tool for unknown specimens if a comprehensive
library of DNA barcodes is first developed. All DNA sequences generated by the
children will be entered into this database.
It is also thought that heavy sampling can improve the overall efficacy of a
DNA barcode database, and will help to test the expectation of low intraspecific
variation as compared to interspecific variation.  If sampled intensively as part
of this project, the British tree flora could serve as a test-bed for the effects of
intensive specimen sampling on DNA barcode performance.

Acknowledgement

The authors wish to thank the Cothill Educational Trust for their support and
collaboration with this project.

399
References
[1] Anon. Research Councils UK: Engaging Young People with Cutting Edge Research: a guide
for researchers and teachers. www.rcuk.ac.uk/per, 2010.
[2] P L. Nimis, B. Press and S. Martellos, KeyToNature: Key to trees at the Old Malthouse. http://
dbiodbs.units.it/carso/chiavi_pub21?sc=446, 2010.
[3] J. Oldham and C. Roberts, The tree name trail: A key to common trees (2nd edition). Field
Studies Council / Forestry Commission. FSC Publications, Shrewsbury, Shropshire, 2003.
[4] The Woodland Trust: Leaf identification swatch book.
[5] O. Johnson and D. More, Collins Tree Guide. Collins, UK, 2006.
[6] TreeBOLD: http://www.boldsystems.org/views/login.php, 2010.

400
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 401-404.
ISBN 978-88-8303-295-0. EUT, 2010.

Engaging Schools in Cutting

Edge Science: From the
Educator’s Perspective
Adrian Richardson, Della Hopkins

Abstract — The field of scientific research, by definition, is constantly

developing new techniques and adapting current thinking in order to address
pertinent issues. With curriculum constraints and exam-based teaching, it is
becoming increasingly challenging to engage young people in new ideas and
methods, and thus facilitate them in becoming the scientists of the future. A new
project developed though collaboration between the Cothill Educational Trust
and The Natural History Museum aims to develop a deeper understanding of
biodiversity science in pre-GCSE aged children, kindling a real excitement for
the science subjects at school.

Index Terms — education, engaging and enthusing, project development,

scientific research.

—————————— u ——————————

1 Introduction

A
recent government report has highlighted the current predicament faced
by science and education. Whilst it is accepted that science and scientists
are crucial globally at economic, environmental and social levels, less
children are studying science subjects at school. This is especially true at GCSE
and higher levels, where children choose their subjects. The danger is that if this
trend continues, there will be a shortage of science professionals in the future
[1]. It is therefore imperative that children have the opportunity to engage in
science at an early age; to enthuse and inspire them to choose to continue to
study science, and to realise the career opportunities available to them.
This view is shared within the scientific community, with an awareness that
students are often unaware of current methods routinely used by scientists.
“Exciting new areas of science typically do not appear in science classrooms
and textbooks until many years after their inception. The result is that too
many students are never afforded opportunities to learn about the cutting-edge
discoveries that make biology so exciting to professional scientists [2].
————————————————
A. Richardson is Principal of the Cothill Educational Trust, Abingdon, Oxfordshire, OX13 6JL. E-
mail: [email protected].
D. Hopkins is the project manager of Tree School on behalf of the Natural History Museum, Lon-
don, SW7 5BD. E-mail: [email protected].

401
2 The Partnership
In order to address these problems, a partnership was formed between two
like-minded parties: the Cothill Educational Trust [3] and The Natural History
Museum of London (NHM). Through a shared interest in combining research,
learning and public engagement, a project began to develop to bring together
educators and scientists.

2.1 The Educators

The Cothill Educational Trust was established in 1967 by the Trustees of

Cothill House, a boys’ boarding school, with the aim of providing first class
education to children up to the age of 14. The Trust now incorporates seven
schools, including the Château de Sauveterre in France, and the Old Malthouse,
a science education centre used for the new collaboration between the Cothill
Trust and the NHM.

2.2 The Scientists

The Natural History Museum is a well-known and respected organisation.

Not only does it house internationally-respected scientists, but the association
provides the project with an identifiable name within the public realm. A
collection of scientists from the NHM are collaborating on the project, with a
range of specialist expertise, including molecular genetics, botanical taxonomy
and biodiversity conservation.

3 Needs Identified and Aims Agreed

In early 2009 a series of meetings were held to explore the potential of the
project and agree the aims between the two partners. These meetings not only
included representatives from the Cothill Trust and NHM scientists, but also
involved staff from other NHM departments including Learning and Interactive
media.
The needs and requirements of both parties were identified. Whilst the Cothill
Trust required the support of the museum, both in terms of scientific expertise
and project management, the NHM needed the experience of an educational
partner and access to a suitable place in which to carry out the teaching. It
was also imperative for the museum to involve suitable staff for the teaching
aspect of the project. It required enthusiastic and personable presenters who
can interact with children, but who are deemed specialists in order to provide
“cutting edge” science and thus a premium “product”.
Once the needs and provisional capacity of each partner had been determined,
the combined project aims and deliverables could be decided. It was agreed
that the NHM - Cothill Educational Trust Project ‘Tree School’ would establish
proof-of-concept in joining scientific research objectives together with science
education imperatives through botany and DNA barcoding. Specifically:
To design, pilot, optimise and communicate methods for involving schoolchildren

402
and other non-experts in international DNA barcoding campaigns.
To promote the development of a scientifically and environmentally literate
citizenry.
To increase the scale on which biodiversity science can be undertaken.
A start-up phase was established in order to develop the relationship between
Cothill and NHM, with the Cothill Educational Trust providing infrastructure,
equipment, logistics, teachers, pupils and the Natural History Museum
contributing the science and learning.

4 The Role of the Cothill Educational Trust

4.1 Location

The Old Malthouse, situated on the Isle of Purbeck, Dorset, has been developed
into a field centre fully equipped to provide five-day residential courses for up
to 32 children at any one time. The Old Malthouse was a boarding preparatory
school, but closed in 2007. It has been completely refurbished, with dormitories
for the children and individual rooms for the teachers and visiting scientists. The
concept when redesigning the interior was to provide a safe and comfortable
environment for all participants, to create a relaxed atmosphere for enhanced
learning and enjoyment. The classroom blocks were also updated to create
a laboratory for the DNA barcoding, and a herbarium area for the storage of
specimens.

4.2 Schools

The boarding and day schools managed by the Cothill Trust will be amongst
the first schools participating in Tree School. Attendance will also be extended
to state schools with links to Cothill, funded by charitable donations to the Trust.
Following planned publicity, it is expected that schools will pay to attend Tree
School, although state schools will continue to be subsidised by the trust.
The importance of the accompanying teaching staff can not be underestimated.
They will maintain overall control of the classroom, in terms of discipline, but can
also act as an intermediary between the children and the scientists. They can
refer to recent lessons, to demonstrate how topics interact and overlap, and can
also ask leading questions of their own to engage with the scientists.

5 Challenges and Benefits

The initial phase of this project has seen a series of three trial workshops
through the summer of 2010. The scientists have worked in collaboration with
three of the Cothill Trust schools in order to ensure the smooth running of the
project during 2011-2012.

403
5.1 The Challenges

The principal challenge faced at the onset of the project was to find a suitable
location, which was able to provide accommodation, laboratory space and enough
outdoor space and suitable ‘field’ locations. This need also brings with it a huge
financial requirement, not only for the initial set up, but also for the continuing
maintenance and upkeep of the site. The recruitment of enthusiastic scientists
from the NHM proved to be straightforward, and a high demand from the teachers
and pupils was established.
Modifications have been made between workshops, in order to pitch the science
at the right level for the group attending. Timetable alterations have been needed,
due to varying group sizes and unpredictable weather conditions.

5.2 Potential Benefits

The immediate benefits of Tree School are apparent, with the provision of
learning for an increasing number of children.
The concept also has the potential for development, both for children and
adults. A workshop specifically for teachers could provide ideas and training,
and incorporate Continuing Professional Development opportunities. Interest
has also been expressed by other non-scientific adults, which could perhaps be
developed as a summer school event. The project design can also be applied to
further scientific projects, and other academic subjects, all with the principal aim
of engaging young people in the excitement of cutting edge research.

6 Conclusion
This endeavour is a marked change from the ‘canned’ experiments schools
are required to provide during scientific learning. Whilst these lessons are an
important aspect of understanding science, the inclusion of current, innovative
scientific principals and experiments allow an insight into science and research,
as well as an opportunity for the children to challenge their questioning skills and
feed their desire for learning. It is important to fire an interest in science at a young
age in order to secure the scientists of the future.
Following the pilot workshops, a continued enthusiasm for the subject has been
reported once back at school. Many of the methods used at Tree School can be
replicated away from the field centre, such as trying to identify and map the trees
on the school premises. It is planned that once schools become engaged with this
project, and have a long-term relationship with the NHM, ideas will be developed
and lead to further collaborative projects.

References
[1] Anon. Nurturing tomorrow’s scientists, Department for children, schools and families. DCSF
Publications, Nottingham, 2007.
[2] A. Jurkowski, A. H. Reid and J. B. Labov, From the National Academies: Metagenomics: A Call
for Bringing a New Science into the Classroom (While It’s Still New). Center for Education,
National Research Council, Washington, DC. CBE—Life Sciences Education. vol. 6, 260 -265,
Winter, 2007.
[3] The Cothill Educational Trust. http://www.cothill-trust.demon.co.uk/index.html, 2010.

404
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 405-409.
ISBN 978-88-8303-295-0. EUT, 2010.

Educational or emotional
languages?
An interactive experiment with
the Lucanian flora (S-Italy)
Riccardo Guarino, Patrizia Menegoni, Sandro Pignatti

Abstract — In the frame of dissemination activities for a still-in-progress work

on the Sites of Community Importance (see EU Directive 92/43) of Basilicata
(a region of Southern Italy), an interactive tool (IIT) for the identification of
vascular plants growing there has been illustrated to two groups of people,
following two different approaches: one focused on textual parts and on
scientific accuracy, the other on images and on the visual comparison
of different objects. The reactions were measured in terms of number of
accesses to the IIT, elapsed time from the demonstration to the first individual
access, and number of queries in the first week after the IIT was distributed.
The most clicked options were recorded as well. People who followed the
emotional/visual approach proved to be significantly more interested in the
IIT than those who followed the descriptive/scientific approach. It seems that
to raise the interest of non-experts to the identification of plant species and,
more in general, to the study of biodiversity, words should be kept at minimum,
while the quality of the images and their “appeal” are essential.

Index Terms — Basilicata, flora, interactive identification tools, people’s

reaction.

—————————— u ——————————

1 Introduction

T
here is general agreement that for scientists it is important to foster
public knowledge on biodiversity and ecosystem’s functioning. However,
all too often educators think about this focus in a fragmented manner,
either as an important end in itself, or as a contribution for enhancing people’s
awareness on their responsibility towards nature and on the effects of human
impact. In the first case, a classical academic approach is followed and often
————————————————
R. Guarino is with the Dept. of Botanical Sciences, University of Palermo, I-90123. E-mail: ric-
[email protected].
P. Menegoni is with E.N.E.A. - C.R. Casaccia, Via Anguillarese 301, S. Maria di Galeria (Rome),
I-00123. E-mail: [email protected].
S. Pignatti is with the Dept. of Plant Biology, University of Rome “La Sapienza”, I-00165. E-mail:
[email protected].

405
the efforts towards popularization are limited to the simplification of concepts
and to a drastic reduction of the provided information. Nature and biodiversity
tend to be depicted as a special selection of vertebrates and big, colourful
invertebrates that sometimes interact with the most attractive plants growing in
a given place, neglecting a myriad of other living organisms. In the second case,
a paternalistic approach is followed: the few who know provide strong evidence
that the survival of a relevant percentage of living organisms is at risk, planning
informative campaigns on most striking examples (polar bears, coral reefs,
tropical rainforests...), following the theory that what does not raise people’s
interest has no value. The most typical, although not very logical, conclusion
of these campaigns is that humans should respect any form of life not only for
ethical reasons, but also because preserving the integrity of natural ecosystems
is an essential need for the survival of ourselves.
We think that to foster people’s knowledge on biodiversity and ecosystem’s
functioning is important per se, but that the success of these non-academic
outcomes can be better achieved through the development of new social
and emotional learning techniques, which do not necessarily imply excessive
simplification and reduction of concepts and information.
In order to test what languages and type of information best stimulate people’s
response and intellectual behaviour, a simple experiment has been carried out
by means of an interactive identification tool on a regional flora. The results are
presented here.

2 Material and methods

An interactive identification tool (IIT) on the vascular flora of Basilicata (a region
of Southern Italy) has been presented and distributed to 54 people attending a
seminar on the Sites of Community Importance (see EU Directive 92/43) of the
same region.
The IIT was an excerpt of the digital utilities developed for the second edition
of Pignatti’s Flora d’Italia [1]. Core of the IIT is a multi-entry key where users can
filter the species by making their own choices in a set of non-hierarchized fields
and options (see [1], Fig. 2). The template of the multi-entry key consists of four
main components: a number of fields, a list of options for each field, a scroll-
down table describing and illustrating every single option, a directory command
leading to the filtered species.
For the presentation of the IIT, attendants were divided into two groups of 27
persons each. They were working in a computer lab of the University of Palermo.
Both groups consisted of post-graduates with similar age and instruction.
Two slightly different versions of the IIT were presented: in the first one, the
scroll-down table of the template of the multi-entry key was describing but
not illustrating the single options, and the directory command was leading to
a classical dichotomous key; in the second one, the scroll-down table was
describing and illustrating the single options, and the directory command was
leading to a panel where the identification of a species took place through the
visual comparison of different images and, after that, through the (optional)
reading of the diagnostic characters.

406
 The “style” adopted during the presentation of the IIT was also different: in the
first case, more emphasis was given to the scientific accuracy of the information
provided, and the identification of the specimens selected for the experiment
was carried out as an individual activity (each participant had his own computer
and specimen); in the second case, more emphasis was given to the images,
and the identification was carried out as a group activity, with discussions on the
available options and plenty of jokes on the visual skills of the people involved.
People’s reactions were measured in terms of number of accesses to the IIT,
time elapsed from the demonstration to the first individual access, number of
queries in the first week after the IIT was distributed. The most clicked options
were recorded, as well. The significance of the differences observed in the two
groups were checked with Student’s t test.

3 Results
People who followed the visual/participatory approach (group 2) seemed to be
significantly more interested to the IIT than those who followed the descriptive/
individual approach (group 1), at least concerning the number of accesses to
the IIT in the first week and the time elapsed from the demonstration to the first
individual access. The number of queries, i.e. the number of options (out of
1496 possible ones) experimented by each user was not significantly different
(see Tab. 1 for details) within the two groups, nor were the most clicked options.
The most clicked options pertained to the following fields: “regional distribution”,
“life form”, “group” (a field including 12 options based on simple floral characters,
like symmetry and number of floral parts), “colour of the flower”, “veining of the
leaves”.

Mean St. Dev. 0,05 C.I.

Nr. of accesses - group 1 9.04 5.54 2.09
Nr. of accesses - group 2 13.87 5.22 1.97
Elapsed time (hrs) - gr. 1 87.70 38.99 14.71
Elapsed time (hrs) - gr. 2 62.44 42.88 16.17
Nr. of queries - group 1 212.07 127.56 48.12
Nr. of queries - group 2 245.90 115.82 43.69
Nr. acc. El. time Queries
Pearson’s correlation -0.0501 0.0675 -0.1298
Stat t 3.1791 -2.5447 0.9597
P(T≤t) - one tail 0.0020 0.0087 0.1732
P(T≤t) - two tails 0.0039 0.0175 0.3464

Tab. 1 – Responses of the two groups to the considered parameters and their
significance.

407
4 Discussion
Two relevant facts influenced users’ behaviour in our experiment: the use of
images and the presentation of the IIT as a kind of “social game”.
In the academic communication and in the related formative activities,
descriptive models are still largely based on textual forms and the learning
process is all too often seen as the result of individual efforts.
Attractive colours and images, integrated with the innovative tools made
available by information technology, can create a supportive environment where
experiential activities can be carried out with a social and emotional involvement.
This will more easily convey a durable acquisition of knowledge [2].
A community approach is vital in a learning process. Many recreational groups
with online forums are already fostering the botanical culture of people. Some
meaningful examples are, for Italy country: GIROS (www.giros.it), Acta Plantarum
(www.actaplantarum.org), Flora delle Alpi Marittime (www.floramarittime.
it), F.A.B. (www.floralpinabergamasca.net), G.M.Lu (gmlu.wordpress.com),
Natura Mediterranea (www.naturamediterraneo.com), Botanica Italiana (www.
botanicaitaliana.it). Each of these websites counts thousands of visitors and
relies on a permanent virtual community with hundreds of supporters.
The sharing of images and experience enhances the individual learning attitude.
The identification and characterization of species becomes a participatory
process to which everyone can contribute with images, observations, new
findings and, finally, with the correct identification of the diagnostic traits of a
given species.
This process is complex and operates at multiple levels. A good IIT should be
well-calibrated on different level of fruition: the availability of information must
be easy, and contents have to be interesting for the whole community, from
beginners to experienced scientists. For these reasons, the starting questions
when implementing an IIT should be: “Which kind of user do we address? What
information users are looking for? Where/how do they expect to find it?”. The
answers to these questions can help in designing a gradual availability of the
contents, able to raise the interest of multi-level users, with no need to sacrifice
the completeness of the information for ensuring better usability [3].
Italo Calvino said that, in order to be effective, information must be: light, short,
exact, visible, coherent [4]. If applied to an IIT, Calvino’s sentence means that the
functions of multimedia objects must be perceived as simple and immediate by
the user (light); assets and files should occupy a few bytes, in order to be loaded
and recalled very quickly (short); textual parts should be essential, precise and
organized in small blocks, with keywords and concepts well highlighted and
illustrated (exact); the hierarchy of the fields and options to filter the species,
as well as the whole structure of functions, commands and graphic templates
should be evident (visible). Finally, the sense of innovation stays in the ability of
creating harmonic and complete communication paths through the integration of
heterogeneous components, keeping at the same time the unitary consistence
of a project (coherent).

408
5 Conclusion
The commonest way to cultivate a hobby, or a scientific interest, is to share it
with other people. Universities, schools, associations are social places where
learning is, intrinsically, a social process. Members of a community do not learn
alone, but rather in collaboration with their teachers, in the company of their
peers, and with the support of their friends.
Emotions can facilitate or hamper the learning process and the ultimate
success in the amount of knowledge acquired. Because social and emotional
factors play such an important role, interactive tools aimed at popularizing
scientific knowledge will be most successful when they integrate efforts to
promote people’s academic, social, and emotional learning.
Our experiment suggests that, in order to raise the interest of non-experts in
the identification of plant species, the learning object must be visually attractive
and the learning process must be “blended”: words should be well calibrated
with illustrations, concepts must be clear and essential. The quality of the images
and their “appeal” are essential, as well as the possibility for users to interact
online, to share information and contents with other users and with scientists,
to become members of a botanical virtual forum that keeps active thanks to the
inputs of a large number of people.

References
[1] R. Guarino, S. Addamiano, M. La Rosa and S. Pignatti, “Flora Italiana Digitale”: an interactive
identification tool for the Flora of Italy. In: P. L. Nimis and R. Vignes Lebbe (eds.), Tools for
Identifying Biodiversity: Progress and Problems, pp. 157-162, 2010.
[2] N. M. Haynes, M. Ben-Avie and J. Ensign (eds.), How social and emotional development add
up: Getting results in math and science education. Teachers College Press, New York, 2003.
[3] R. E. Mayer, The Cambridge Handbook of Multimedia Learning. Cambridge University Press,
2005.
[4] I. Calvino, Lezioni americane. Sei proposte per il prossimo millennio. Oscar Mondadori,
Milano, 1993.

409
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 411-416.
ISBN 978-88-8303-295-0. EUT, 2010.

Online sharing educational

content on biodiversity topics:
a case study from organic
agriculture and agroecology
Nikos Palavitsinis, Nikos Manouselis, Kostas Kastrantas,
John Stoitsis, Xenofon Tsilibaris

Abstract — E-Learning Technologies and Standards are emerging as the

dominant way to make educational content widely available. Approaches
to these technologies should be domain-independent and easily adaptable
to different contexts. Organic.Edunet aims at making content on Organic
Agriculture and Agroecology widely available through a single point of
reference. To achieve this, the project has adopted and adapted Open
Software solutions and has built upon them to offer the Organic.Edunet Web
Federation Portal and the Repository Suite of Tools. This paper presents the
tools that were developed in the frame of Organic.Edunet project, serving as
a guide for all individuals that aim at establishing similar tools in a field such
as the biodiversity.

Index Terms — repositories, education, open access, metadata.

—————————— u ——————————

1 Introduction

T
he Organic.Edunet project aims to facilitate access, usage and exploitation
of digital educational content related to Organic Agriculture (OA) and
Agroecology (AE). From the technical viewpoint of the project’s objectives,
Organic.Edunet aims to support stakeholders producing content about OA & AE
in order to publish it in an online federation of learning repositories and describe
it according to multilingual, standard-complying metadata. This objective is
accomplished through the deployment of the Repository Suite of Tools.
Also on the same basis, the project has deployed a multilingual online environment
(the Organic.Edunet Web portal) that facilitates end-users’ search, retrieval,
access and use of the content in the learning repositories. Both tools deployed (the
Repository Tool and the Web Portal) are already running smoothly on the web while

————————————————
All authors are with the Greek Research & Technology Network, Athens, Greece. E-mail: pa-
[email protected], [email protected], [email protected], [email protected], [email protected].

411
small changes are being made to ensure a smooth operation. Having completed
the biggest part of the work involved, the Organic.Edunet partners have a clear view
of all the complexity, problems, issues and challenges that were faced during the
deployment of the tools.
This paper aims to briefly describe the tools produced in terms of their main
characteristics and to present a part of the work that was carried out for launching
them successfully. More specifically, the metadata application profile that was used
to deploy the Repository Tools will be described, and the main parts of the Organic.
Edunet Web Portal will also be analyzed. Overall, this proposal aims to demonstrate
a complete process of making educational content available online through the use of
e-learning technologies and standards. This paper also aims to serve as a reference
point for ongoing or future projects that will deal with the issue of making educational
content available, regardless the application domain. The proposed methodology and
tools can be deployed in order to support education on biodiversity.

2 Tools and project results

2.1 Application profile

The IEEE LOM standard has been chosen as the basis for the metadata
application profile to be used in Organic.Edunet. The schema is therefore termed
as Organic.Edunet Application Profile (AP). It adopts many of the elements of
LOM, specializing several of them in order to better describe learning resources
on organic agriculture and agroecology. In each one of the nine (9) categories
of LOM elements, a number of elements have been refined, in order to be used
in Organic.Edunet [1].

2.2 Repository tool

The Organic.Edunet AP is implemented in a repository management tool

that is being used for the organisation and management of learning objects.
The repository tool can be used by content providers to create learning
resources, collections, and complex learning objects such as online exhibits
and educational paths. Moreover, the tool allows content providers to design
and integrate educational templates that implement specific educational paths
and digital exhibits. Using the repository tool, content providers can generate
complex learning objects by reusing existing learning objects in their repository.
The repository tool is used to describe resources with appropriate metadata,
and to publish resources in their own learning repository. When the resources
are published in the individual repositories, they can be harvested from the
Organic.Edunet Portal. When a resource is harvested, it is published on the
portal where users can access it.
The repository tool allows content providers to connect their repository to
the Organic.Edunet federation of repositories. In addition, the tool supports the
connection of the repository with other federations of learning repositories. This
is achieved through the adoption of open standards and specifications for the

412
exchange of search queries and the harvesting of metadata. These standards
and specifications include the Open Access Initiative Protocol for Metadata
Harvesting (OAI-PMH, http://www.openarchives.org) and the Simple Query
Interface (SQI) [2].

2.3 The Portal

The Organic.Edunet Web portal (http://portal.organic-edunet.eu) aims to

facilitate access, usage and exploitation of digital educational content related
to Organic Agriculture and Agroecology. It is a multilingual online federation
of learning repositories, populated with quality content from various content
producers. Its main purpose is to facilitate end-users’ search, retrieval, access
and use of the content in the learning repositories.
The development of the Organic.Edunet Web portal addresses one of the
main objectives of the project. That is to integrate and specialize state-of-art
technologies of the World Wide Web, in order to provide end-users with a single
European reference point that will offer advanced services such as ontology-
based searching and social recommendation and facilitate search, retrieval and
use of the collected content. The global architecture of Organic.Edunet is thus
composed of 2 major subsystems (shown in Fig. 1):
• Organic.Edunet Web portal (top part): It provides services to the
portal users (such as learners, teachers, professionals, etc.), ranging
from searching through the content in various ways (different search
functionalities), to tagging of resources and bookmarking of resources
in a personal online space. The users are able to see the resources and
their metadata because this information is fed into the portal from the
federation of repositories.
• Federation of Repositories (lower part): This subsystem includes the
repositories containing either the learning resources uploaded by the
content providers of the project, or their metadata or both.
The potential users of the Organic.Edunet Web Portal include: teachers,
students, pupils, researchers, OA&AE professionals, general public, etc. The
Web Portal is internally made of four components: the semantic repository Ont-
Space, the semantic navigation module, the portal infrastructure and the social
navigation module.

413
 Fig. 1 – Overview of Organic.Edunet technical architecture.

2. 4 Semantic Repository Ont-Space

Ont-space is a software framework for the deployment of semantic Learning

Object Metadata Repositories. Ont-space is based on LOMR [3] a semantic
learning object metadata repository developed as part of the LUISA project. The
semantic repository is driven by ontologies that contain the specification of the
learning object schema used such as Dublin Core, IEEE LOM, etc. Ont-space
enables specialized components, such as custom query managers and result
composers, to benefit from the availability of different, heterogeneous ont-space
instances.

2.5 Semantic Navigation Module

The semantic navigation module is responsible for the semantic search

capabilities integrated in the architecture. The main objective of the semantic
navigation module is to provide an ontology-based semantic interface for
performing interactive searching sessions. This is carried out through the
navigation of ontology elements (classes, individuals and properties) and their
relations. This process is totally ontology-independent, so any valid ontology in
OWL format can be navigated, and particularly the OA&AE ontology developed
for Organic.Edunet [4]. This type of semantic search provides more accurate and
meaningful results, eliminating non-relevant results and avoiding the existence
of long lists with repeated elements in the search outcomes.

414
2.6 Portal Infrastructure

The Organic.Edunet Web portal has been designed to be used by non-expert

users (usability has been a priority) and developed over a customized Joomla
CMS (Content Management System) installation. The technical partners of
the project developed some Joomla add-ons in order to provide customized
functions in the Organic.Edunet Web portal architecture.
As the Web application based on Joomla needs to communicate with other
architecture elements, there is a need to establish some communication
guidelines. The interaction between the Ont-Space module and the Joomla-
based Web portal is asynchronous and it takes place through the new
components developed and integrated in Joomla. Those components have
been implemented in PHP and AJAX, follow the Joomla MVC guidelines. The
communication between Ont-Space and Joomla is based on the exchanging
of JSON objects that contains the information sent from Ont-Space in an
asynchronous way as result of a request from Joomla to a concrete remote
servlet in Ont-Space.

2.7 Social Navigation Module

The Web Services API of the Organic.Edunet Social Navigation Module

are available only within the scope of the OrganicEdunet’s suite of tools. The
available methods of this Module’s API that are exposed as Web Services are
grouped in 5 categories: General Methods, Resource Annotation Methods,
Prediction Methods, Recommendation Methods, and Population Methods.

3 Discussion
The Organic.Edunet Tools are fully deployed and are being used by the
Organic.Edunet partners. More specifically, the repository tool has been set
up and is used by the project partners to upload resources and links, also
annotating them with metadata. Presently, a translation of the repository tool is
underway to include more languages (i.e. French, Czech, Slovenian, Bulgarian,
Turkish, Dutch and Hindi).
The Organic.Edunet Metadata Application Profile is also completed and
deployed in the repository tool. Some work is also being carried out for the
translation of the AP, so that the repository tool can be set up for different
languages, thus helping additional communities connect their material with
Organic.Edunet. The same languages that are being added for the repository
tool are also translated for the metadata AP.
As far as the Organic.Edunet Portal is concerned, this is currently being
used by a great number of users, as the Organic.Edunet Open Days are
underway. The participants of the Open Days are also asked to provide their
feedback using an online questionnaire. All the results are being evaluated by
the responsible Organic.Edunet partner, making the necessary revisions to the
portal. Additional languages are also being deployed, building on the existing
support offered by the Joomla community. More specifically, all the portal texts

415
are being translated to open the Organic.Edunet Portal to stakeholders that are
interested in connecting their collections to Organic.Edunet Portal.

4 Conclusion
This paper presented the basic tools – technologies that were developed
during the Organic.Edunet project. Its main aim was to provide a brief overview
of the whole system deployed through Organic.Edunet that can serve as
a reference point for institutions trying to open their content to the world, by
setting up learning repositories and by making their content available through
open access portals. A possible direction to take advantage of the finding of this
paper, is to adapt the process to biodiversity, in order to explore how well will this
approach fit to this domain.

Acknowledgements

The work presented in this paper has been funded with support by the European
Commission, and more specifically the project No ECP-2006-EDU-410012 “Organic.
Edunet: A Multilingual Federation of Learning Repositories with Quality Content for the
Awareness and Education of European Youth about Organic Agriculture and Agroecology”
of the eContentplus Programme.

References
[1] H. Ebner, N. Manouselis, M. Palmér, F. Enoksson, N. Palavitsinis, K. Kastrantas and A. Naeve,
“Learning Object Annotation for Agricultural Learning Repositories”, in Proc. of 9th IEEE
International Conference on Advanced Learning Technologies (ICALT2009), Riga, Latvia,
accepted for publication.
[2] S. Ternier, E. Duval, “Interoperability of Repositories: The Simple Query Interface in ARIADNE”,
International Journal on E-Learning, vol. 5(1), pp. 161-166, 2006.
[3] M. A. Sicilia, S. Sanchez, S. Arroyo and S. Martín-Cantero, Deliverable D4.3: LOMR
architectural prototype specification (D4.3). LUISA Learning Content Management System
Using Innovative Semantic Web Services Architecture (IST- FP6 - 027149). Madrid, Spain:
Atos Origin, 2007.
[4] A. Steen, S. Sanchez-Alonso, M. A. Sicilia and G. Lieblein, ECP-2006-EDU-410012 Organic.
Edunet Deliverable D2.2.3, Initial version of Organic Agriculture and Agroecology Domain
Model Representation. Online at: http://www.organic-edunet.eu/organic/files/document/
OrganicEdunet _D2.2.3a_final.pdf, 17 November 2008.

416
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 417-418.
ISBN 978-88-8303-295-0. EUT, 2010.

JSTOR Plant Science

Michael Sean Gallagher

Abstract — JSTOR Plant Science is an online environment that brings

together content, tools, and people interested in plant science. It provides
access to foundational content vital to plant science – plant type specimens,
taxonomic structures, scientific literature, and related materials, making them
widely accessible to the plant science community as well as to researchers
in other fields and to the public. It also provides an easy to use interface with
powerful functionality that supports research and teaching, including the ability
to measure and record plant specimens, share observations and objects with
colleagues and classmates, and investigate global plant biodiversity.

Index Terms — tools, database, resources, type specimens.

—————————— u ——————————

J STOR Plant Science is an online environment that brings together

content, tools, and people interested in plant science. It provides access
to foundational content vital to plant science – plant type specimens,
taxonomic structures, scientific literature, and related materials, making them
widely accessible to the plant science community as well as to researchers
in other fields and to the public. It also provides an easy to use interface with
powerful functionality that supports research and teaching, including the ability
to measure and record plant specimens, share observations and objects with
colleagues and classmates, and investigate global plant biodiversity.
JSTOR Plant Science strives to be a comprehensive online research tool for
aggregating and exploring the world’s botanical resources, thereby dramatically
improving access for students, scholars, and scientists around the globe. It
is useful for those researching, teaching or studying botany, biology, ecology,
environmental and conservation studies.
A significant portion of the content available on JSTOR Plant Science has
been contributed through an effort known as the Global Plants Initiative (GPI).
GPI is an international undertaking by leading herbaria to digitize and make
available plant type specimens and other holdings used by botanists and others
working in plant science every day. Partners include more than 150 institutions
in 52 countries on 5 continents. There are two partner networks in place and
contributing today: the African Plants Initiative which focuses on plants from
Africa and the Latin American Plants Initiative which contributes plants from
Latin America. GPI and has also expanded to Asia. GPI has received funding
and guidance from The Andrew W. Mellon Foundation.
————————————————
M. S. Gallagher, 100 Campus Drive, Suite 100 - 08540 - New Jersey, USA, E-mail: michael.gal-
[email protected].

417
 Currently, JSTOR Plant Science has more than 900,000 specimens.
When complete, there will be an estimated 2.2 million. Further there are
foundational reference works and books such as The Useful Plants of West
Tropical Africa, Flowering Plants of South Africa, and illustrations from Curtis’s
Botanical Magazine. JSTOR Plant Science also includes a significant set of
correspondence, including Kew’s Directors’ Correspondence which included
hand-written letters and memorandum from the senior staff of Kew from 1841 to
1928. The JSTOR Plant Science team is developing tools to extract contextual
data from this aggregation, which will hopefully enhance its use to the botanical
community.

418
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – p. 419.
ISBN 978-88-8303-295-0. EUT, 2010.

ecoBalade: Towards a workflow

for Citizen Science Nature Trails
Julie Chabalier, Khaled Talbi, Patrick Peters, Amandine Sahl,
Olivier Coullet, Olivier Assunçao, Olivier Rovellotti

Abstract — In the context of Citizen Science, where potential

users may not be familiar with traditional keys and criteria, it seems
necessary to provide a simplified interface in order to guide the users
through the complexity of natural biodiversity.
Tacit knowledge [1] necessary for citizen science projects; is well
known to be difficult to transfer to another person by means of
writing it down or verbalizing it. The ecoBalade solves that constraint
by putting the user in a problem solving situation in the context of
determination, observation, and recording of field data.
In order to be successful the workflow must include three stages; a
filter of the potential taxa by an expert naturalist, support thought-
out the first identification process and a finally visualization of the
field observations by the subject. In a typical ecoBalade scenario,
an expert naturalist will survey the potential species beforehand,
and will generate a list of potential taxa criteria and pictures. This is
then formalized in a XML semantic structure used by a PDA software
Pocket eReleve [2] to guide the users thought the identification
process.
This concept has been successfully tested on the field in Saint
Mandrier with approximately twenty novice users and three PDAs,
it has generated in the course of two hours 32 observations [3]. The
experience will be generalized to other locations in the months to
come.

Index Terms — citizen science, education, biodiversity, XML standard, mobile

solutions, identification, survey.

—————————— u ——————————

References
[1] M. Polanyi, “The Tacit Dimension”, New York, Anchor Books. (108 + xi pp.), 1967.
[2] G. H Kipré and al., “Pocket eReleve, Nouvelle approche de collecte de données sur le
terrain” Geomatique expert, vol. 69, pp. 24-27, 2009.
[3] “Prototypage de l’Eco-balade … Et plus si affinités …”,
http://territoiresenresidences wordpress.com/2010/06/24/.

————————————————
All authors are with the company Natural Solutions, Marseille, CP 13002.
E-mail: [email protected].

419
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 421-422.
ISBN 978-88-8303-295-0. EUT, 2010.

Electronic data recording tools

and identifying species
in the field
Alexander Kroupa, Anke Hoffmann, Juan Carlos Monje,
Christoph L. Häuser

Abstract — The “European Distributed Institute of Taxonomy”

(EDIT) is an initiative of 28 European, North American and Russian
institutions to build a network in “Taxonomy for Biodiversity and
Ecosystem Research”, with the objective to reduce the fragmentation
in taxonomy through institutional integration in Europe (www.e-
taxonomy.eu). European Commission funding (FP6) for this “Network
of Excellence” has started in March 2006, and runs for 5 years. For
EDIT Work Package 7 (WP 7) “Applying Taxonomy to Conservation”
the aim is to strengthen the input of taxonomic expertise in Europe for
biodiversity conservation, by organizing the participation of individual
taxonomists and experts in biodiversity inventory and monitoring
efforts in conservation areas (www.atbi.eu).
For biodiversity inventories and monitoring, the advantage of using
digital field recording tools is to simplify data recording as well as to
improve data quality. The use of electronic field tools and software
should be promoted to help minimizing error rates, in particular to avoid
mistakes from the beginning of the recording chain. Many errors may
be avoided by using authority lists, e.g. for countries, habitat-types
or taxa that can already be determined in the field. Automated geo-
referencing and recording of date and time in standardized formats
already in the field will also avoid errors when importing or retyping
such data into a database. Relevant software should be usable for
tools such as mobile phones with GPS (Global Positioning System)
functionality up to water resistant PDAs - Personal Digital Assistant
(e.g. Magellan - Mobile Mapper; Trimble – Juno, Nomad).
Examples presented here for more efficient electronic data recording
in the field include the application of mobile recording devices with
customized forms, which are tested for field work in ATBI+M (All Taxa
Biodiversity Inventories + Monitoring; www.atbi.eu) sites operated
in the EDIT project. This is a general approach for recording geo-

————————————————
A. Kroupa is with the Museum für Naturkunde, Invalidenstr. 43, 10115 Berlin. E-mail: alexander.
[email protected].
A. Hoffmann is with the Museum für Naturkunde, Invalidenstr. 43, 10115 Berlin. E-mail: anke.
hoffmann@ mfn-berlin.de.
C.J. Monje is with the Staatliches Museum für Naturkunde, Rosenstein 1, 70191 Stuttgart. E-mail:
[email protected].
C.L. Häuser is with the Museum für Naturkunde, Invalidenstr. 43, 10115 Berlin. E-mail: christoph.
[email protected].

421
referenced, individual species data using customized forms for
ESRI ArcPad applications. Species names can be selected from a
taxonomic authority list provided in a file in dBASE-format. Such files
can be easily created, modified, and exchanged to allow individual
researchers to use regional or otherwise customized species lists.
Fields and field formats correspond to ABCD standards so that
exports of recorded locality, event, and species data can be directly
integrated into a central database and applications for individual
ATBI+M websites (e.g. www.atbi.eu/mercantour-marittime/ or www.
atbi.eu/gemer/). The authority species lists may be customized for
a geographic area (e.g., a nature reserve) and/or a group of taxa
(e.g., larger birds). This allows each expert to choose the species
list needed for his/her research. Problems remain with observation
records which cannot be reliably determined in the field. Therefore
identification help should be made available on the PDA at least for
difficult taxa.

Index Terms — biodiversity, digital data capture, fieldwork, inventory, species

authority lists.

422
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 423-428.
ISBN 978-88-8303-295-0. EUT, 2010.

Cost Assessment of the Field

Measurement of Biodiversity:
a Farm-scale Case Study
Stefano Targetti, Davide Viaggi, David Cuming

Abstract — Attention to the effects of agriculture on biodiversity is currently

increasing. Yet the measurement of biodiversity is both time-consuming and
costly. Considering the limited budgets available for biodiversity conservation,
it is timely to focus on the cost analysis of biodiversity indicators in order to
ensure the optimization of the scarce funds available. We present the cost
analysis of operational data from the fieldwork efforts undertaken in the
measurement of biodiversity indicators at farm-scale. Methodological issues
are discussed.

Index Terms — biodiversity measurement, earthworm indicator, spider

indicator, research costs.

—————————— u ——————————

T
1 Introduction
he growing societal demand for environmental services provided by
agriculture focuses attention on the implementation of sound agro-
environmental schemes based on reliable information with respect to
the effects of different agricultural practices on biodiversity [1]. The
existing gap between need and availability of funds for biodiversity monitoring
and assessment highlights the importance of the optimization of resources [2].
The cost analysis of biodiversity measurement, in particular if undertaken by
way of a cost-effectiveness analysis, can ensure the optimization of scarce
available funds and the selection of the most efficient indicators of biodiversity
[3], [4]. Nevertheless, the cost-effectiveness of biodiversity measurement is a
practically unstudied issue [5] and only a few examples exist [6] which propose
a methodological approach to its analysis.
The assessment of the costs of measuring biodiversity at farm-scale is one
of the specific tasks of the BioBio project (Indicators for biodiversity in organic
and low-input farming systems -UE-FP7- http://www.biobio-indicator.wur.nl). In
this paper we propose a methodology for the cost assessment of biodiversity

————————————————
The authors are with the Department of Agricultural Economics and Engineering of the Univer-
sity of Bologna, Italy. E-mail: [email protected], [email protected], david_cuming@
hotmail.com.

423
measurement, and discuss its practical application to the spider and earthworm
indicators measured through the BioBio project protocol.

2 Material and methods

The BioBio project involves 12 case studies (CS) throughout Europe
concerning organic, or low input, and conventional agricultural systems. Here
we present the preliminary results from the French CS (Midi- Pyrénées Region).
The cost assessment is related to the field measurement of the spider (SP)
and the earthworm (EW) biodiversity indicators carried out in Spring 2010. Data
covers 4 arable farms (Tab. 1), where wheat and sunflower are the main crops.
Distance from the research centre (driving time in minutes) was similar for
each farm (about 1 hour). Survey stratification was performed through habitat
mapping in order to cover the different habitat conditions of the surveyed farm
sites (see [7] for further details). In the aggregate, 537 samples (345 for SP and
192 for EW) were gathered.

Distance from
Number of
Farm Area (ha) Type research centre
samples
(minutes)

A 23 organic 53 113

B 19 organic 57 88

C 27 conventional 60 111

D 146 conventional 68 225

Tab. 1 – Main features of the 4 farms studied and number of samples (SP + EW)
gathered during the spring fieldwork.

Spider sampling was carried out with the aid of a modified vacuum shredder
(Stihl SH 86-D), and 5 suction samples were taken in each plot (each suction
had a suction area of approximately 0.1 m² and lasted 30 seconds). The samples
were stored separately in a cool-box and transferred to a laboratory. Spiders
were sorted out in the laboratory and placed in vials with 70% alcohol [8], [9].
Three survey sessions were scheduled in the project protocol. Here we present
data from the first session. The sampling team was composed of 3-4 persons.
Earthworm sampling was carried out by way of two methods: 1) stirring up
an allyl-isothiocyanate and ethanol solution into metal frames (30 cm X 30 cm)
which were placed in the ground, and collecting the earthworms that came
upward during the first 10 minutes; 2) extracting the soil core (20 cm depth)
from the sampling site and hand-sorting the earthworms on a plastic sheet.
Samples were placed in cold containers with oxygenated water and transferred
to refrigerators in the laboratory [10], [11]. The sampling team was composed
of 5 persons.
The cost assessment methodology was organised in such a way as to
allow for an analytical assessment of actual costs, as well as the subsequent
simulation of costs with standardised costs. For this reason, both physical units

424
of resources used and related prices were collected on a regular basis. Data
collection was performed through the collection of records related to staff time,
distance and transport time, consumables and equipment. Time spent (and
costs) for fieldwork organisation and preparation and taxonomy identification
is not included. Field staff filled-in a weekly cost-form which was entered into a
relational data-base. Data collection was organised in order to retrace the costs
related to each single farm and each single indicator. Each record contained:
date, farm site, staff qualification, time spent per field-worker and was linked
to different tables indicating the salary band of the staff, the distance of the
farm site from the research centre, transport time, equipment and consumable
costs, and the type of work (fieldwork, laboratory, etc.). The cost of the indicator
measurement was composed of three resource categories: 1) equipment
and consumables, 2) labour time investment (fieldwork, laboratory-work and
transport), 3) worker categories (permanent, temporary).
Equipment and consumables included all the materials used during the
fieldwork as well as the field lunches for the staff. The cost of the vacuum
shredder was calculated as: cost per suction = cost of the vacuum new / number
of suctions over its lifetime. This was approximated to 0.038€ per suction. The
gross salary of the staff was approximated to 36€ per hour for permanent
workers and 13.8€ per hour for temporary workers. Vehicle costs were charged
at 0.32€ per km and included fuel, car insurance and vehicle depreciation. All
the costs are related to 2010.

3 Results
The composition of the costs for the field measurement of the two biodiversity
indicators in the four farms studied are presented in Tab. 2. The cost per sample
of the earthworm indicator was 3.5 times higher than the spider indicator. EW
costs per hectare were only 2 times higher than SP because of the lower number
of samples gathered for the EW indicator. Although the spider indicator required
a higher permanent work effort (1 hour of permanent work for every 2.6 hours
of temporary work for SP vs. 1 hour of permanent work for every 4 hours of
temporary work for EW), the labour load was higher for the EW indicator (the
labour cost was 83% of total cost for EW vs. 57% for SP). The portion of the
other costs were always lower (max. 10% of total costs), except for lab work and
preparation of samples which constituted an important component of costs for
the spider indicator (23% of total costs).
The cost of transportation (vehicle, highway tolls and work time for transfer of
fieldworkers from the research centre) was a consistent portion of costs for the
measurement of biodiversity (Tab. 3). This cost was about 30% of total costs.
Accordingly, the cost of the measurement of the indicators was strongly tied to
the organisation of the fieldwork (number of sessions, distance of farms from
research centre, etc.). The portion of transportation + transfer of fieldworkers
with respect to the total costs was higher for SP than for EW (34% vs. 28%)
because the research unit was equipped with only one vacuum tool. As a result
only one sampling team could be organised for fieldwork each day. The EW
measurement was more flexible as several sampling teams per day could be

425
arranged. Thus, the differences in costs between the two indicators were more
evident when considering the effective costs of fieldwork (resources spent in
field measurement after transport costs): 13.3€ ha-1 for SP vs. 28.1€ ha-1 for EW
(ratio 1:2.1).

Biodiversity indicator Spiders Earthworms

Cost per sample 12,6 44,3

Cost per ha 20,2 39,6

Labour 618 1756

Permanent / Temporary ratio 1: 2,6 1: 4

Consumables and equipment 110 187

Labwork 253 35

Vehicle and tolls 105 148

Sum of costs 1086 2126

Tab. 2 – Composition of costs (mean values per farm) for the field measurement of
biodiversity indicators (values in €) and permanent vs. temporary work effort ratio.

Transport costs
Biodiversity (vehicle + Percentage of Effective cost per Effective cost
indicator displacement of total costs (%) sample (€) per ha (€)
fieldworkers, €)

Spiders 369 34 8,3 13,3

Earthworms 618 28 31,4 28,1

Tab. 3 – Analysis of costs of the field measurement of biodiversity. Share of

transportation and transfer of fieldworkers with respect to total costs (mean values per
farm) and effective costs of fieldwork (effective costs are: total costs – transport and
transfer of fieldworker costs).

The comparison of effective costs of biodiversity measurement between

organic and conventional farms pointed out a consistent higher effort of field
sampling for the organic farms (Tab. 4). Even if the mean number of samples
was higher in the conventional farms (84 vs. 50), sampling effort in organic
farms was 1,5 times higher concerning the cost per hectare and 2,4 times higher
considering the days person-1 ha-1. This is probably related to a higher variability
of habitats for the organic farms which required a more intense sampling than
the conventional farms.

426
 Effective days per
Samples Effective cost ha-1
person ha-1

Organic 50 41,6 0,62

Conventional 84 27,5 0,26

Tab. 4 – Comparison of costs of field sampling in organic and conventional farms.

Number of samples, effective cost ha-1 and effective days person-1 ha-1 of effort required
(mean values of Sp + EW sampling per farm, effective cost is: total cost – transport and
transfer of fieldworker costs).

4 Conclusion
The first important result concerns the relevance of costs that were in the
thousands of Euros per farm.
The ex-post assessment of costs of the field measurement of biodiversity is of
significant importance both for the organisation of the sampling sessions as well
as for the cost-effectiveness analysis. The cost assessment could be a valid tool
for the optimisation of the use of available resources. This evidence is of great
importance considering the gap between the need and the availability of funds
for biodiversity. It is our opinion that the increased availability of cost data could
be of great assistance in the advancement of the effectiveness of biodiversity
assessments.
The share of transportation costs (vehicle and transfer time of staff) suggests
that a careful organisation of fieldwork should be considered essential for the
optimisation of available resources.
Our preliminary analysis clearly identified lower costs, coupled with a higher
number of samples (thanks to the vacuum tool), for the spider indicator. However,
this information is incomplete without an assessment of the effectiveness of the
measurement. Moreover, the cost of SP will be much higher considering the
other two survey sessions which are scheduled in the BioBio project protocol.

Acknowledgement

This work was supported by a grant from EU-FP7, BioBio - Indicators for biodiversity
in organic and low-input farming systems. The authors wish to thank J.P. Sarthou, J.P.
Choisis, C. Pelosi and S. Ledoux for their cost data gathering.

References
[1] OECD, “OECD Expert Meeting on Agri-biodiversity Indicators”, Zurich, 5-8 November 2001,
http://www.oecd.org/dataoecd/16/56/40339943.pdf, 2010.
[2] T. B. Gardner, J. Barlow, I. S. Araujo, T. C. Avila-Pires, A.B. Bonaldo, J. E. Costa, M. C.
Esposito, L. V. Ferreira, J. Hawes, M. I. M. Hernandez, M. S. Hoogmoed, R. N. Leite, N. F. Lo-
Man-Hung, J. R. Malcolm, M. B. Martinus, L. A. M. Mestre, R. Miranda-Santos, W. L. Overal, L.
Parry, S. L. Peters, M. A. Ribeiro-Junior, M. N. F. Da Silva, C. Da Silva Motta and C. A. Peres,
“The Cost-Effectiveness of Biodiversity Surveys in Tropical Forests”, Ecology Letters, vol. 11,
pp. 139-150, 2008.
[3] P. J. Ferraro and S. K. Pattanayak, “Money for Nothing? A Call for Empirical Evaluation of
Biodiversity Conservation Investments“ Plos Biology, vol. 4, pp. 482-488, April 2006, http://

427
 www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.0040105, 2010.
[4] B. S. Halpern, C. R. Pyke, H. E. Fox, J. C. Haney, M. A. Schlaepfer and P. Zaradic, “Gaps and
Mismatches Between Global Conservation Priorities and Spending” Conservation Biology, vol.
134, pp. 96-105, 2007.
[5] A. Juutinen and M. Mönkkönen, “Testing Alternative Indicators for Biodiversity Conservation
in Old-Growth Boreal Forests: Ecology and Economics”, Ecological Economics vol. 50, pp.
35-48, 2004.
[6] A. Qi, J. N. Perry, J. D. Pidgeon, L. A. Haylock and D. R. Brooks, “Cost-efficacy in Measuring
Farmland Biodiversity – Lessons from the Farm Scale Evaluations of Genetically Modified
Herbicide-tolerant Crops“ Annals of Applied Biology, vol. 152, pp. 93-101, 2008.
[7] R. Jongman and R. G. H. Bunce, Farmland Features in the European Union. A Description
and Pilot Inventory of their Distribution, Alterra report 1936, ALTERRA, Wageningen UR, 2009.
[8] M. H. Schmidt-Entling and J. Dobeli, “Sown wildflower areas to enhance spiders in arable
fields”, Agriculture Ecosystems & Environment, vol. 133, pp.19-22, 2009.
[9] M. H. Schmidt and D. T. Tscharntke, “The Role of Perennial Habitats for Central European
Farmland Spiders” Agriculture Ecosystems & Environment, vol. 105, pp. 235-242, 2005.
[10] E. R. Zaborski, “Allyl isothiocyanate: an alternative chemical expellant for sampling earthworms”,
Applied Soil Ecology, vol. 22, pp. 87-95, 2003.
[11] C. Pelosi, M. Bertrand, Y. Capowiez, H. Boizard and J. Roger-Estrade, “Earthworm collection
from agricultural fields: Comparisons of selected expellants in presence/absence of hand-
sorting”, European Journal of Soil Biology, vol. 45, pp. 176-183, 2009.

428
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 429-435.
ISBN 978-88-8303-295-0. EUT, 2010.

Markets for biodiversity

information products:
real or imaginary?
Bill Hominick, Peter Schalk

Abstract — In the past decade, a large range of biodiversity information

products and services has become available. Some were developed with EC
subsidies, some with national funds, and others as initiatives of universities,
research institutes or private persons. Few, if any, were developed in the
‘commercial world’ based on a business plan. Some sponsors and funders
ask for exploitation and sustainability plans after the development phase
or at the end of the project. The success of these depends on whether the
products meet real demands and serve actual markets. Market potential is
often misjudged or overestimated and many products are developed without
a prior needs or market analysis. This negatively affects the sustainability
for biodiversity information services. In this paper we review the marketing
of some biodiversity information products in a commercial environment, to
assess demands and size of the markets.

Index Terms — biodiversity information products, user needs, markets, return

on investment, sustainability.

—————————— u ——————————

1 ETI Information Services: a specialized sales Company

E
TI Information Services Ltd (ETIIS) was a subsidiary of ETI BioInformatics,
a not-for-profit organisation initiated by the Netherlands’ Government
and UNESCO. Its aim is to make authoritative biodiversity information
broadly accessible and usable by using information technology. Initially Springer
Verlag marketed and distributed ETI information products, but recognising the
unique market for e-media and specific requirements to reach the audience,
ETI developed its own marketing and distribution subsidiary. ETI Information
Services Ltd was established in 2001 and relied on the catalogue on the
website of ETI. The company launched its own website to support sales on
1 April, 2004. The company ceased trading on 31 March, 2009. ETI products
are currently marketed by Margraff, Germany. The purpose of this paper is to
share the commercial experience gained in marketing electronic biodiversity
————————————————
B Hominick was Executive Director of ETI Information Services Ltd, UK (2001-2009). E-mail: bill@
etiis.org.
P Schalk is managing Director of ETI BioInformatics, University of Amsterdam, Netherlands. E-
mail: [email protected].

429
information products, to discuss the size of this niche market, and the marketing
requirements to reach it. The analysis is based on commercial sales of selected
multimedia biodiversity titles to 1705 customers from over 50 countries from 2001
to 2008 in a period of 77 months. During that time 9970 items were sold. Sales
were direct to customers online, by mail, e-mail or phone orders, or through re-
sellers reaching particular markets. All resellers sold multimedia products as a
minor part of their main business, which was usually selling books. ETIIS relied
exclusively on income from the niche market for electronic resources related to
biodiversity.
The income generated by sales is compared with product development costs
to get an impression of the sustainability of such products.

2 Users of Biodiversity Information

ETIIS focused on three categories of users of biodiversity e-products: Teachers
and students in schools, colleges and universities (Formal Education); outdoor
enthusiasts such as walkers, wildlife observers and workers in Nature reserves
(Popular); and, professional specialists, researchers, regulatory officials and
dedicated amateurs (Science). Each of these markets has different sizes and
requirements and needs to be considered individually. The Formal Education
market is not considered in this communication.

3 Popular Market
The general population is a large and important market for biodiversity
information. A number of ETI wildlife field guides should be of interest to non-
specialists, i.e. individuals who have an interest in some aspect of nature, and
are committed to learn more. This is a highly competitive market with numerous
printed products available, so the e-products must be priced competitively. The
following titles, showing the price range (including VAT) charged by different
distributors, fit these criteria:

- Birds of Europe (£14.95 - £25.95)

- Interactive Guide to Butterflies of Europe (£12.95 - £25.95)

- Interactive Guide to Mushrooms and other fungi (£12.95 - £25.95)

- Interactive Flora of the British Isles (£29.95 - £39.95)

- Five Kingdoms – a multimedia guide to life on earth (£14.95 - £25.95)

ETIIS used several resellers specialized in particular markets to sell some or

all of these titles. Their sales should indicate the interest of their customers in
biodiversity information in an e-format. Tab. 1 lists the main resellers and sales
data, and is followed by some observation on sales and markets.

430
Tab. 1 – Number of e-guides sold by ETIIS and resellers with information on publishing
date and marketing methods. NS = Not Stocked.

Observations regarding Tab. 1:

1) The best-selling popular title by far is Birds of Europe. DreamDirect markets
carefully-selected products to a mass market by distributing up to 36 million
catalogues per year. Remarkably, Birds of Europe is one of the first titles
produced by ETI and over ten thousand (in various updates) have been sold
since 1992.
2) In contrast to DreamDirect’s mass marketing approach, Alana Ecology is
specialized, and targets outdoor enthusiasts. It features books and CD-ROMs
as a subset of its products, which are mainly equipment. For their customers the
Interactive Flora of the British Isles (IFBI) is a favourite; they had less interest in
birds, butterflies and mushrooms, field guides that could be used by this group.
3) NHBS supplies shops, institutions and private customers. It lists all ETI titles
amongst the book categories; customers have to seek them out. Though they
have numerous books and field guides on birds, they sold only 3 copies of Birds
of Europe. Are paper field guides the preferred medium for these customers?
IFBI and Butterflies are their best lines for ETI products, promoted strongly in
their catalogue and website. Sales indicate that most of their e-products achieve
very small sales.
4) The RHS has a large, customer base of gardeners and horticulturalists and
is known for its large selection of plant titles. The shop has a small number of
e- products, and does not market them actively. IFBI was their best-selling ETI
title. The other sales are likely to be impulse buys.
5) Summerfield Books targets the plant sciences market, and e-products are
uncommon in their catalogue. However, they actively market titles and have
achieved strong sales of IFBI. The BSBI strongly supported the IFBI. Apparently,
an audience can be persuaded to purchase these e-products if they are aimed
specifically at them and their field of interest.

431
 6) Sales by ETIIS itself demonstrate the importance of marketing in boosting
sales. In this selection, IFBI is the best-selling popular product. It was heavily
promoted and received strong reviews in popular and scientific journals; it is
now recognized as one of the standard (e)flora for the UK. Advertisements were
taken in relevant publications (eg Plant Talk) and it was promoted via Google
Adwords. The Interactive Guide to Butterflies of Europe is the other best-seller
of the popular titles. It was promoted mainly by a repeated advertisement in
Butterfly, the magazine of Butterfly Conservation (readership 17,000). However,
such promotion is expensive and the cost of promotion exceeded the income
generated by increased sales!
Sometimes it is implied that e-products are preferred over printed versions.
This is not the case. For example, compare the book Flora of the British Isles
First Edition: Sales 7400 (1991-1997) and Second Edition: Sales 7350 (1997-
2004) with the Interactive Flora of the British Isles DVD-ROM: Sales 1474
(2004-2008). Similarly, the book Flora of the Netherlands print run was 18,000
while Heukel’s Interactieve Flora CD-ROM sold over 5,000 copies in four years.
E-products have a higher access barrier than books, and the interest of retail
shops in stocking e-titles is limited as the products are considered too specialized
and with limited sales potential. This prejudice towards the print medium must
be overcome if sales for biodiversity e-information are to increase.
We conclude that there are many potential users of biodiversity information in
the general population, and some are willing to pay for and use it in an e-format.
Amongst the 612 biodiversity information e-product ETIIS sold, birds were by
far the most popular subject. Price, within a limit of £40, does not appear to
be an issue for the customer. However, the single most important fact is that
while there are many potential users of biodiversity e-information, they need
to be made aware of the existence of the product, and then be persuaded to
purchase it. Most are not actively seeking to purchase multimedia biodiversity
information. Hence marketing is critical. Of course, if the information was free,
the situation could be different. ETI’s website ‘soortenbank.nl’ freely offers
detailed information, identification keys and distribution maps on almost 7,000
species in the Netherlands. It attracts well over 3,000 unique users daily, a
number that is still growing.

4 Science Market
The science market is fragmented and specialized, with small sales to be
expected for the vast majority of similarly specific e-products. The largest
category of products is from ETI’s World Biodiversity Database series: 80+
e-publications. These e-products are taxonomic monographs aimed at specialists
and by their nature will have a limited market. It is difficult and costly to reach a
small audience. Sales do not appear to be strongly price-sensitive. Like books,
sales patterns show that most sales are achieved in the first few years for a
title, and then enter a steady state of a few sales per year. Reducing prices can
have a temporary effect on lagging sales. Special bulk sales, at a discount, can
have a large affect on sales. Specialist training courses are obvious targets, but
usually rely on the author’s support or knowledge of training efforts. Authors

432
can be extremely helpful and work hard to promote their publication. They
supply mailing and e-mailing lists for contacts, contact their colleagues and
promote their titles at relevant meetings. We collated information for the best-
selling scientific titles in 2002-2008 listed in Tab. 2 together with sales, prices,
publication dates and months available.

Tab. 2 – Best-selling science titles of ETIIS.

Observations regarding Tab. 2:

- Arthropod titles were promoted by e-mails to addresses provided by the
author and to e-mail lists created by data mining. Specialist bookshops and
fliers at relevant meetings were also used to promote the titles.
- Chironomids and Oligochaetes. A remarkable number of sales were achieved
for these three products. While these are very specialized groups, they have
great importance for environmental monitoring of water quality. This market was
reached by building up e-mail addresses of water quality specialists. Titles of
practical importance are less price-sensitive as they are frequently “must have”
rather than “would like to have” products.
- Although World Seaweed Resources is a fairly recent publication, it is the
best-selling science title. It was priced low deliberately to encourage sales, and
this was possible because a company provided sponsorship for its production.
While the low price undoubtedly helped sales, the author has been very important
in promoting it. In addition the product was bundled with Cultivation and Farming
of Marine Plants at reduced price, which had a remarkable effect on sales of
Cultivation and Farming of Marine Plants, one of the poorest-selling scientific
titles, with only 14 sold separately over a period of 77 months compared to the
97 sold as part of the special package.
Scientific e-titles, by their nature, cover very specialized groups and have a
small, fragmented market. Even best-selling titles rarely achieve sales beyond
100. In comparison: scientific books usually have print runs of 500-800. For

433
the foreseeable future, books will outsell e-products even for the same title:
books have the clear advantage in portability, comfort of handling, familiarity,
shelf life and identifiable prestige on the shelf. The trend in scientific publishing
is towards smaller print runs, pitched to a known market and then straight to
Publish on Demand. E-products are ideal for this trend, as they can be produced
easily at little cost in small numbers, and they can also be updated easily.
After the first few years of sales, there will be a long period when sales
continue at a trickle rate. Significantly reducing prices can promote the level of
these sales, while linking to e-mail marketing can enhance the effect. However,
the costs of producing a database for e-mail marketing are substantial, and are
not met by the small numbers of sales. Hence, even if funding is available, one
should ask, “Who wants the information?” Always ask, “Is this a need-to-have or
a would-like-to-have, product?” All biodiversity information is not equal. It could
be argued that it is nice to know about the butterflies or birds in a region, but it
is essential to know about pests, crop protection, insect vectors of disease, etc.
So, if investment is required, some priorities will also be required.

5 Product devevelopment costs

The selected biodiversity information products discussed here use the
Linnaeus II taxonomic information management system as a vehicle to compile,
organize, share, and e-publish (on disk, web, mobile) biodiversity data. Linnaeus
II was developed in the 90-ties (initially with subsidies from UNESCO and the
Dutch government) as a freely available data management tool for scientists
to create ‘e-monographs’. The website: http://www.eti.uva.nl/products/linnaeus.
php provides further information. This software is updated in 3-4 year cycles.
Current version is 2.65 (2008), while a wholly new (web-based) version
(Linnaeus NG) is expected in 2011. When calculating the development costs for
e-products as described here, the generic costs of (maintaining) the information
management system must be taken into account. ETI estimates the cost of an
update cycle of Linnaeus to approximately 50,000 Euros. The cost for technical
user support (helpdesk) amounts to 15,000 euro per annum. ETI distributes
the additional costs equally over all e-products amounting to 5,000 Euros per
e-product.
Tab. 3 gives an indication of the product development costs of the popular-
educational and scientific e-products discussed. Popular-educational products,
created by ETI staff in collaboration with external specialists and contributors,
are more expensive to develop than the scientific titles which are often built and
delivered by the authors to ETI in an almost publishing-ready state. However,
when the author’s research costs and institutional overheads are taken into
account especially the science products would be extremely expensive.

434
 Title Costs* Revenues Results

Birds of Europe (1) 120,000 302,000 + 182,000

HIFN+IFBI (2) 191,000 97,000 - 93,000

Butterflies (3) 43,000 51,000 + 8,000

Mushrooms (3) 91,000 51,000 - 40,000

Scientific Titles (5) 20,000 7,000 - 13,000 (av. per title)

Tab. 3 – Overview of developments costs for several products compared to revenues

from sales. Notes: (*) includes 5,000 Euros ‘system maintenance costs’ see text. (1)
Birds of Europe was produced in the English, Dutch, German and Italian language. (2)
The Interactive Flora of the British Isles built upon and extended an earlier developed
Interactive Flora of the Netherlands. (3) The Butterflies and Mushrooms e-guides were
produced in the English and Dutch language. The multi-language approach increased
the market potential for the content. (5) The development costs for purely scientific titles
is based upon the average of the e-products listed in Tab. 2; similar for the revenues.

ETI has been relatively successful in marketing biodiversity information

e-products. Still, the development costs of many less popular e-products exceed
the income generated by sales as demonstrated by Tab. 3. Sustainability (i.e.
updating the products, keeping them available) is therefore an issue. Academic
developers not always properly calculate full development costs (i.e. all hours,
all overheads) when considering product exploitation and sustainability issues.
A way to address this is: cooperation, increased efficiency by standardization
of data (exchange) formats, shared software (developments), joint marketing
approaches, in combination with serving the markets needs. The KeyToNature
project (www.keytonature.eu) demonstrated that a collaborative approach is
possible, as for computer based species identification products.

Acknowledgement

The authors wish to thank Dr Christian Kittl, Prof Pier Luigi Nimis, Mr Nicola Dorigo,
Dr Wim Backhuys, Mr Paul van Bruggen, Dr George Tippet for stimulating discussions.

435
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 437-443.
ISBN 978-88-8303-295-0. EUT, 2010.

A Basic Business Model for

Commercial Application
of Identification Tools
Christian Kittl, Peter Schalk, Nicola Dorigo Salamon,
Stefano Martellos

Abstract — Within the three-year EU project KeyToNature various identification

tools and applications in formal education for teaching biodiversity have been
researched and developed. Building on the competencies of the involved
partner organisations and the expertise gained in this domain, the paper
outlines a business model which aims at commercially exploiting the project
results on a broader scale by describing the value proposition, products &
services, value architecture, revenue model and the intended market.

Index Terms — business model, identification tools, exploitation, sustainability,

EU project.

—————————— u ——————————

1 Introduction

F
rom September 2007 to the end of 2010 the KeyToNature project mobilises
14 partners from 11 EU countries in the eContentplus Programme, with
a total budget of 4.8 Million Euros. The main objectives of KeyToNature
are to: 1) increase access and simplify use of e-learning tools for identifying
biodiversity, 2) improve interoperability among existing databases for the
creation of identification tools, 3) optimise educational efficiency and increase
quality of educational contents, 4) add value to existing identification tools by
providing multilingual access, and 5) suggest best practices against barriers
that prevent the use, production, exposure, discovery and acquisition of the
digital contents required for designing the identification tools [1].
Software packages developed in recent decades, which enable the rapid and
easy creation of interactive identification tools, are the driving forces behind

————————————————
C. Kittl is with evolaris next level GmbH, Hugo-Wolf-Gasse 8, A-8010 Graz. E-mail: christian.kittl@
evolaris.net.
P. Schalk is with ETI BioInformatics, Mauritskade 61, 1092 AD Amsterdam. E-mail: pschalk@eti.
uva.nl.
N. Dorigo is with T&B e Associati srl, c/o AREA SCIENCE Park, Padriciano 99, Building H
I-34012, Trieste. E-Mail: [email protected].
S. Martellos is with Department of Life Sciences, University of Trieste, I-34127, Trieste. E-mail:
[email protected].

437
the switch from traditional paper-based keys to multimedia and online versions
with their many advantages. These interactive identification tools are not only
important for the educational sector, but can also be used to solve identification
problems in many industrial application areas. The basic assumption of the
KeyToNature project, namely that these tools should not only be usable by a
few experts, but made applicable for pupils and students by reengineering them
to fit their needs and wants, also applies when aiming to reach a broad audience
of potential customers in industry: usability, support of multiple languages,
aesthetic appeal, the possibility to easily enhance or change parts of the derived
keys by adding user generated content, etc., are all factors that are equally
important, irrespective of whether the tools are applied in the classroom or in
professional environments.
In order to exploit the knowledge gained in the project in the best possible
way and thus being able to keep the developed services and tools up to date
and usable, a sustainable business model is needed. This will ensure that
solutions for the educational field, which are already now used by a large
number of KeyToNature associated members like schools and universities all
over Europe, can be provided even after Community funding ends. This could
also help interested project partners to generate returns on their investments in
the project, as about half of the budget was financed by their own resources.
In order to analyse the potential of a business, traditionally two different
approaches were used: 1) the resource-based view, which focuses on the core
competencies and the unique access to resources a company has in order to
build competitive advantage (for an overview of the most important works see
[2]), and 2) the market-based view, which emphasises the industry with its
competitors, customer segments and regulations in which the company has
to be successfully positioned [3]. Both approaches are still important when
developing a new business, although the main innovation often lies in the
business model, especially when modern ICT (Information and Communication
Technologies) plays a crucial role in the business [4].
Chapter 2 thus first describes the products developed within the KeyToNature
project in order to better understand the resources which can be used for
exploitation when designing the business. The following chapter then introduces
a hypothetical business model and briefly outlines the market strategy.

2 Identification Tools and Results of KeyToNature

The identification of organisms is fundamental in biology. An accurate
identification provides a correct name through which detailed information such
as taxonomic descriptions, ecological relations, economic values, conservation
status, legislation status and genetic code can be unlocked [5].
Identification is done by observing certain characters of an organism and
subsequently using the respective character states in a structured way by
applying a key to retrieve the correct name. This approach can not only be
applied for identifying organisms in the field of biology, but basically applies to
any identification problem. The tools generated and the expertise gained within
KeyToNature can thus in principle be generalised to other fields of application.

438
The following figure gives an overview of the main elements of the KeyToNature
system architecture:

 
Fig. 1 – The KeyToNature system architecture [6].

The heart of the system architecture is formed by the keys. The whole system
is primarily aimed to be used online, but the keys can also be made available
offline, e.g. for use abroad in the field with mobile phones, where Internet
connection fees are expensive or when no Internet connection is available at
all. They can thus be web-based, paper-based (i.e. printed out), provided on
CD-ROM or on PDAs/mobile devices.
The keys are normally based on data stored in relational databases and
software packages like FRIDA and LINNAEUS which dynamically generate the
actual sequence of questions step per step. Within the KeyToNature consortium
the following two tools are being developed by project partners:
FRIDA (FRiendly IDentificAtion) is an original and flexible program developed
by S. Martellos and patented (2002) by the University of Trieste. FRIDA is
based on a relational database and can automatically generate both interactive
identification tools accessible online, and traditional, dichotomous paper-printed
identification keys.
LINNAEUS II is developed and sold as a product by ETI. There are three
‘modules’ of Linnaeus II: the ‘Builder’ to manage data and to create an information
system, the ‘Runtime’ engine to publish completed information systems on CD-
ROM/DVD-ROM, and the ‘Web Publisher’ to publish a completed project as a
Web site.
Besides the primary data (information that generates the actual key) these
software packages make use of secondary data like pictures, drawings, sound

439
and text files in order to present the user with supporting information to the
current identification step and the finally derived organism. In KeyToNature,
a FEDORA (Flexible Extensible Digital Object Repository Architecture) media
repository is used to this end as a supplement to multimedia data stored directly
in the key database. FEDORA is a conceptual framework that uses a set of
abstractions about digital information to provide the basis for software systems
that can manage digital information.
In order to make the output of software packages like FRIDA and LINNAEUS
compatible and editable in other tools, a certain standard needs to be adopted.
KeyToNature has decided to adopt the SDD (Structured Descriptive Data)
standard proposed by TDWG (Biodiversity Information Standards, formerly
Taxonomic Database Working Group). The goal of the SDD standard is to allow
capture, transport, caching and archiving of descriptive data in all forms, using a
platform- and application-independent, international standard [6].
The keys generated by software packages capable of producing SDD files
can then be adapted to produce localized minikeys, for specific applications
such as school gardens, parks and reserves, or enhanced with user-generated
data by using the Open Key Editor.
Many keys made available by the data providers are ‘master keys’ including
many species. Long keys are complicated and have redundant information
when used in an area with fewer species, such as a park or nature reserve, or
a school garden. The Open Key Editor allows users to ‘crop’ a master key and
customize it for a given set of species. The ‘cropped’ key can then be edited for
language and illustrations (e.g. to suit a particular user level, or platform such
as the mobile phone). With the Open Key Editor the user can browse existing
master keys and edit them.
The Open Key Editor has two further important features: 1) it permits to
largely solve the problem of translation: once a “large” key has been translated
into a given language, it is possible to derive from it a high number of smaller
keys adapted to the users’ needs without the need of further translations, 2) It
permits users to add to the key user-generated content in their own language,
thus enhancing considerably the degree of interaction between users and
KeyToNature identification tools. The Open Key Editor was developed and
optimized jointly by the University of Trieste and ETI and is based on open
source software. In addition to access on the Internet, output on mobile platforms
was included [6].
Keys for mobile devices form further important elements of the system
architecture. Many of the keys can be used in the field on PDAs or iPhones, both
in stand-alone and in online versions. Tools like the MobilePackager developed
by KeyToNature partner GIUNTI Labs increase the benefit of the keys by giving
users the important possibility of adding user-generated content (in this case
geo-referenced pictures) directly in the field, using their mobile devices.
Another useful tool developed in the course of the KeyToNature project is
IBIS-ID (Interactive Biodiversity Identification Software): it is a “key player
software tool” created to help the users in the process of identification of species
or other taxa, by using the multi-access keys described in a SDD (Structure of
Descriptive Data) file. It is based on the Adobe Flex technology, a well suited

440
candidate because of its effectiveness for data driven interactive applications
and native support for dealing with data organized in XML (eXtensible Markup
Language) structured files through the support of the ECMA e4x (ECMAScript
for XML) standard [7].
Last but not least it is also possible to comfortably view and edit keys in
Wikis, the so called “wiki-keys”. KeyToNature developed two tools to this end,
the jKey player and jKey Editor: The jKey key player (http://www.keytonature.
eu/wiki/JKey_Player) is a small javascript, that allows wiki-keys - in addition to
their printable overview display - to be also “played” step-by-step, similar to the
FRIDA and Open Key functionality. The complementary jKey Editor (http://www.
keytonature.eu/wiki/Wiki-based_identification_key_editor) allows form-based
editing of the identification keys.

3 A Basic Business Model for Exploitation

A business model is a model of an existing business or a planned future
business. A model is always a simplification of the complex reality. It helps to
understand the fundamentals of a business or to plan how a future business
should look [8].
In the following the key elements of a business model for the possible
commercial application of some identification tools and expertise developed
within KeyToNature is presented:
The value proposition, which describes the key value generated for the main
customers and business partners: In our case the key value is supporting
customers in solving complex identification problems by utilising expertise and
tools of the value chain partners.
Product/market design: The primary end user groups for the KeyToNature
project are pupils/students and their teachers in formal learning environments.
The main products generated in the project are software keys for identifying
organisms, which are embedded in learning management systems to provide an
adequate context for the respective target group. The formal education market is
a difficult one in terms of generating revenues. However, the authors believe that
it can be continued to be served for free if maintenance and further development
of the software tools can be mainly financed through other commercial activities.
Although the identification tools could in principle be applied to any industry
with identification problems (e.g. through software keys for identifying bacteria
in medicine, or for identifying diseases starting from a series of symptoms) the
first step will be to offer the existing tools to a market where there is already
relevant domain expertise in the partner network through existing data providers
(e.g. plants, certain animals, microfungi). The new business shall thus start off
with software keys for desktop and mobile devices in this field as products.
The customers could be sought in the B2B area, while individual partners of
the network may also address the B2C market directly. Examples of potential
customers are nature parks, who want to provide their visitors with specific
mobile keys to their flora and fauna. Already in the past it could be proved that
this market is worth addressing.
The value creation architecture outlines in which steps the product will be

441
generated and which partners and competencies are necessary. The following
figure outlines the value chain:

  Fig. 2 – Value chain of the proposed business model.

When a customer wants support in solving identification problems (for

example, a key to the plants in a nature park in Catalonia), a specific software
identification tool is provided as a product. It is based upon data provided by one
of the data providers who own primary data (i.e. master keys) and secondary
data (multimedia files), in our example this might be the Royal Botanical Garden
of Madrid with the eflora Iberica database. If the customer cannot provide a list
of all species that are in the relevant set, a domain expert (who might be from
the Royal Botanical Garden or some other institution), it will be necessary to
analyse the situation locally. The master key data together with the list containing
the relevant subset of species is then passed on to a key generating software
(e.g. the University of Trieste with their FRIDA software or ETI with LINNAEUS,
etc.), who produces the specific interactive key. Finally, a developer (e.g. the
KeyToNature project partner EVOLARIS or Divulgando, a University of Trieste
spin-off) creates an application and customized user interface according to the
customer’s needs and wants (e.g. an iPhone application).
The proposed business model thus draws upon expertise of various partners
to solve the problem of the customer. If one of the partners could serve the
customer’s needs alone, then there is no need for cooperation and the business
model doesn’t work there. Sometimes this may be the case for data providers,
who are at the same time key generator software vendors. But in many cases
they may either lack the required master data, resources to build the list
containing the subset of relevant species, or the know-how to build the final
application with outstanding user experience, e.g. for Apples iPad.
The value architecture could be set up in a way that a newly founded company
following the proposed business model could flexibly collaborate with any
organisation able to deliver value in one of the four stages through classic
buyer-seller-relationships. In the first instance however, the main suppliers of
this company could be those KeyToNature project partners who decide to join
the value creation network.
Revenue model: In order to be able to offer the products and services described
above to the market, a company would need to be established, ideally a Ltd. or
the like to limit personal liability of the involved individuals in case of potential
losses. This company could have the following main objectives: marketing &
sales, i.e. finding suitable customers and contracting with them, and sourcing
the necessary expertise and tools within the partner network. The company
should thus be kept very lean regarding production capacities – here all steps
should be outsourced to KeyToNature partners and potentially further network
members, especially KeyToNature associate members. The new company
thus would need to manage the partner network to orchestrate the value chain,

442
negotiate contracts with the customers and ensure proper service through
service level agreements. All partners in the network could freely negotiate what
they want to get in return for providing their expertise and tools to the company
so that the final product could be created. For its services the company should
retain the difference between the revenues it could generate from the market
and the costs incurred for the services of the business partners.

4 Conclusion
The paper presents tangible outcomes of the KeyToNature project and a basic
business model for their commercial exploitation. Based on this model a sound
business plan with detailed analysis of costs and revenue forecast needs to
be developed in order to establish a sustainable business. All KeyToNature
partners and further interested suppliers of data, expertise, and tools are invited
to collaborate with the company, if and when this shall be founded.

Acknowledgement

This work has been supported by the KeyToNature project, ECP-2006-EDU-410019,

in the frame of the eContentplus Programme.

References
[1] P. L. Nimis and S. Martellos, ”KeyToNature a European Project for Teaching Biodiversity”. In:
A. L. Weitzman and L. Belbin (eds.), Proceedings of TDWG, Abstracts of the 2007 Annual
Conference of the Taxonomic Databases Working Group, Bratislava, Slovakia, 16-22, p. 67,
September 2007.
[2] N. J. Foss (ed.), Resources, Firms and Strategies: A Reader in the Resource-Based Perspective.
Oxford University Press, Oxford, 1998.
[3] M. E. Porter, Competitive Strategy: Techniques for Analyzing Industries and Competitors. The
Free Press, New York, 1980.
[4] C. Kittl, Kundenakzeptanz und Geschäftsrelevanz als Grundlage ökonomisch sinnvoller
Geschäftsmodelle für digitale Dienste, Gabler, 2009.
[5] P. Schalk, P. L. Nimis and W. Addink, “KeyToNature Species Identification e-Tools for
Education”. In: Biodiversity Information Standards (TDWG) Annual Conference 2008, http://
www.keytonature.eu/w/media/4/4b/KeyToNature_Species_Identification_e-Tools_for_
Education_TDWG_2008.pdf, 2008.
[6] P. L. Nimis, N. Dorigo Salamon, C. Kittl, G. Hagedorn and P. Schalk, D1.6. Annual Report, 2nd
Annual Report to the eContentplus project ECP-2006-EDU-410019/KeyToNature, 2009.
[7] M. Giurgiu, A. Homodi and G. Hagedorn, “IBIS-ID, an Adobe FLEX based identification
tool for SDD-encoded multi-access keys”. In: Biodiversity Information Standards (TDWG)
Annual Conference 2009, 9-13 November 2009, Montpellier, France. http://www.tdwg.org/
fileadmin/2009conference/documents/PreProceedings2009.pdf, 2009
[8] P. Stähler, Geschäftsmodelle in der digitalen Ökonomie: Merkmale, Strategien und
Auswirkungen. Josef Eul, Lohmar, 2002.

443
Nimis P. L., Vignes Lebbe R. (eds.)
Tools for Identifying Biodiversity: Progress and Problems – pp. 445-450.
ISBN 978-88-8303-295-0. EUT, 2010.

Keys to Nature:
A test on the iPhone market
Rodolfo Riccamboni, Alessio Mereu, Chiara Boscarol

Abstract — Several keys running on mobile devices were developed by

the KeyToNature project. Most of them are freely downoadable online.
The rapid spread of smartphones has opened up new opportunities in the
production and distribution of multimedia applications for the educational
sector, including interactive keys to identify organisms. This market is new,
still partly unexplored and changing fast. The Department of Life Sciences
of the University of Trieste and Divulgando Srl, have tested its potential by
uploading in the iTunes Store different types of keys for the iPhone, some
of them for free, others for sale. This paper introduces the issue of global
market applications, summarizes the experience gained in our case-studies,
and suggests ways to make these applications economically viable.
 
Index Terms — smartphones, mobile store, iPhone applications, education,
global market, Apple, Android.

—————————— u ——————————

1 Introduction

O
ne of the most successful products developed by KeyToNature are
identification keys to plants, animals and fungi running on mobile devices
(PDAs and smartphones). Since they can be used in the field, they
proved to be useful in schools, also attracting the attention of many associate
members of KeyToNature such as Natural Parks and Botanic Gardens, as a
means to advertise their biodiversity heritage. Presently, several hundreds of
applications for PDAs, most of which specifically created for a single school,
are freely downloadable online from the Italian portal of KeyToNature (www.
dryades.eu).
The rapid spread of smartphones has opened up new opportunities in the
production and distribution of multimedia applications for the educational
sector, including interactive keys to identify organisms. This market is new, still

————————————————
R. Riccamboni is with the Department of Life Sciences, University of Trieste, I-34127, Italy. E-mail:
[email protected].
A. Mereu is President of Divulgando S.r.l., Corso Italia, 31, I-34122, Trieste, Italy. E-mail: mereu@
divulgando.eu.
C.Boscarol is Content Manager of Divulgando S.r.l., Corso Italia, 31, I-34122, Trieste, Italy. E-mail:
[email protected].

445
partly unexplored and changing fast. The Department of Life Sciences of the
University of Trieste and Divulgando Srl, have tested its potential by uploading
in the iTunes Store different types of keys for the iPhone, some of them for free,
others for sale.
This paper introduces the issue of global market applications for mobile
devices, summarizes the experience gained through four case-studies, and
suggests ways to make these applications economically viable.

2 The market of mobile applications

In the past 12 months, the global market for applications has greatly increased
its sales, mainly due to Apple Inc. and its iPhone App Store. At the end of April
2009, downloads from the App Store exceeded 1 billion, a year later they were
over 4 billions [1]. This market leadership is due to several factors:
1. developers create applications for a single operating system running on
all devices in the iOS market (something that happens on Nokia Symbian
as well),
2. procedures for purchasing an application are very simple,
3. the hardware is of excellent quality, the software is very user-friendly,
4. the process is reversed: people buy the iPhone because there are many
iPhone applications,
5. the satisfaction of customers buying the iPhone is 73% compared with
39% of HTC (High Tech Computer Corporation) [2].
Juniper Research, an International Company of statistics and market forecasts,
states that the App Store market - worth nearly $ 10 billion in 2009 - will be worth
over $ 32 billion in 2015, with an increase of over $ 22 billion dollars in 6 years
[1]. The development of mobile applications is not an ephemeral fashion, but a
strong reason for business and visibility. A challenge against Apple’s App Store
is now posed by all the major mobile phone companies, such as Samsung Rim,
Microsoft, Nokia, and Intel. Currently, developers prefer the Apple App Store of
Apple, but Google Android is rapidly increasing in terms of both applications and
devices sold.

3. Aim, data and methods

We tried to assess the potential of the market for iPhone applications by
developing and making available on the App Store of iTunes several identification
tools to vascular plants developed by KeyToNature, differing in:
• geographic coverage (nationwide to very local),
• number of species involved (from c. 100 to more than 2000),
• dimension of market (which mainly depends on language: Estonia vs.
Italy-The Netherlands),
• price (for sale vs. for free).
All applications were developed by the University of Trieste and Divulgando
Srl, except one, developed by ETI Bioinformatics. The results are based on
the absolute number of downloads for each type of application sold, and on
their relative rank in the educational sector of the respective national markets

446
(which varies daily depending on the number of downloads). This information is
provided to developers by Apple Inc. every day. The ranking of applications is
related to both the category where they are included (e.g. entertainment, travel,
games, education) and to the total number of applications sold in each category.
Each country has its own history each store has its own ranking.
Developers of applications in the Apple App Store can change the pricing
of an application almost in real time: we took advantage of this opportunity
by changing the price of some applications for short periods of times, to test
whether the price had an influence on the number of downloads.

4 Case-study applications

4.1 Woody plants of Estonia (Eesti eFloora I. Puud ja põõsad)

This key to over 140 trees and shrubs growing in Estonia [3] was our first
testbed. The key is dichotomous, richly ilustrated. It includes taxon pages with
notes (in Estonian), distribution maps and pictures of every species. It can also
interacts with users, who can add user-generated content in “their” own key in
the form of textual field-notes and pictures, and permits users to share their
input in the Web community (Facebook).
With this key we wanted to test the iPhone market in the worst conditions:
• in a very small market (the key is written in Estonian),
• in an unfavourable period (January 2010, when most of Estonia was
covered by snow),
• charging a fixed cost of € 1.59.
The result was surprising: in a couple of days this became the best-seller
in the educational sector of the App Store in Estonia. Still now, at a distance
of 8 months, it still keeps a leading position among the 10 bestellers in this
sector. The total number of downloads (c. 370) is small, but it ranks high in the
educational sector of the Estonian market. A peak with over 120 downloads
(almost 1/3 of the total) was recorded on May 13th, 2010, on the occasion of a
national TV service, which shows how important advertising is for the iPhone
market.
Lesson for us: the market for identification keys on mobiles is interesting.

4.2 The flora of the Netherlands

This application, produced by another partner of KeyToNature - ETI

Bioinformatics - is an interactive flora with a nationwide coverage, including
more than 1.800 species, which manages a large amount of data (375 Mb).
It was uploaded in the Apple App Store in June 2010, at the rather high price
of 9.99 €. Apple Inc. has set a threshold of 20 Mb for direct download on the
iPhone via UMTS/GPRS. Applications that exceed this limit must be installed by
connecting to a computer, and in general lightweight applications (less than 20
Mb) tend to be downloaded more than those exceeding the threshold of 20 Mb.
In spite of this and of the rather high price, the market responded well: from June

447
15th to August 15th the key had ca. 700 downloads.
Lesson for us: the market for identification keys is not only interesting, but also
economically rentable.

4.3 A “very local” key for free in the Italian market (100 plants in the
Botanical Garden of Catania - Sicily)

At this point we wanted to test the potential of another large market, that of
Italy. The first experiment was based on a very local-special key, that to 100
woody plants occurring in the Botanical Garden of Catania (Sicily). Prepared
in collaboration with colleagues from the University of Catania, this key was
originally meant to be used only for educational activities organised by the
Garden. We thought that it would be of little interest on the Italian iPhone market,
but the results were surprising: the application was made available for free on
the App Store in May 2010: in the first 20 days it had over 1000 downloads, and
still now (August) is having 20-40 downloads a day. On August 2, 2010 the total
downloads were more than 4000.
Lesson for us: the market for identification keys in Italy is very interesting.

4.4 Other “Local” keys for sale in the Italian market (Flora of the
Trieste Karst area, NE Italy)

The previous lessons told us that the Italian market is wide and economically
interesting. Thus, we tried to “sell” on the Apple App Store some local keys to
plants, those occurring in the Karst area near Trieste, a small enclave at the
eastern border of Italy which however is poorly known by Italians.
The test was based on 2 keys:
• a “large” key to c. 1000 species occurring in the Val Rosandra Natural
Reserve near Trieste, sold at 2.39 €
• two smaller keys (200-250 species) to the plants of special habitats in the
Karst area (dry grasslands, woody habitats), sold at 1.59 Euros.
These keys were uploaded on iTunes at the end of June 2010.
The general results were disappointing: the smaller keys had a constant
average of c. 3 daily downloads, while the downloads of the larger key were
even smaller (average 1.5 a day). At this point we have lowered the price of all
keys to 0.79 Euros, without any relevant change in the number of downloads.
Finally, we made the applications downoadable for free. The experiment lasted
3 days only (August, 18-20). The result was surprising: in 3 days the number of
daily downloads rose dramatically, with an average of 60 downloads a day per
key and an increasing trend. There was a significant difference in the number
of daily downloads between the two very similar smaller keys (dry grasslands
and woody habitats). The latter had many more downloads a day.This difference
may be due to the titles of the 2 keys: the former was entitled “La landa carsica”,
where “landa” is a term for dry grasslands which is used only locally in the
Karst area, while the second was entitled “Piante di sottobosco” (plants of
the understorey), where the term “sottobosco” is well known nationwide. This
suggests to pay attention to title and keywords when uploading a key on the

448
Apple App Store.
Lesson for us: “local” keys have no market if sold for money, but may have a
large impact if made available for free.

5 Conclusions
The results of our tests are summarised in Tab. 1. The most relevant
conclusions from our tests are:

GEOGRAPHIC NUMBER Nr. of

KEY MARKET PRICING
COVERAGE OF SPECIES DOWNLOADS

Estonia nationwide medium small for sale very high

Flora of the
nationwide very high wide for sale very high
Netherlands

Catania Botanical
very local small wide free very high
Garden

Karst flora local small wide free very high

Flora of the Val

very local high wide free high
Rosandra Reserve

Karst flora local small wide for sale low

Flora of the Val

very local high wide for sale very low
Rosandra Reserve

Tab. 1 – Synthesis of the results.

1. The interest for identification tools in the iPhone market is potentially high.
2. The geographic coverage of a key has a great impact on the market:
nationwide keys can be profitably sold, while local keys seem to be a poor
source of revenues.
3. However, when local keys are made downloadable for free, their impact
increases dramatically, and they are downloaded also by persons outside
of the area of interest of the key.
On the light of these considerations, we have changed our market strategy as
follows:
1. We have started the production of keys with a nationwide coverage, which
will be placed on the market for sale.
2. We are proposing to the many Parks and Natural Reserves for which
we have already created a key a further service, that of developing an
application for the iPhone, which will be made available for free as a
powerful means to advertise their biodiversity heritage. Several Parks
ready showed a keen interest, and are redy to pay for such a service.

449
Acknowledgement

This work has been supported by the KeyToNature project, ECP-2006-EDU-410019, in

the frame of the eContentplus Programme.

References
[1] W. Holden, A World of Apps. Whitepaper from Mobile App Stores, Business Model, Strategies
& Market Segmentation 2010-2015, Juniper Research Ltd, June 2010.
[2] J. Crumrine and P. Carton. Explosive Changes in Consumer Demand Shake Up Smart Phone
Industry. ChangeWave Research Ltd. http://www.changewaveresearch.com/articles/2010/07/
smart_phones_20100714.html, July 14, 2010.
[3] A. Saag, T. Randlane and M. Leht, “Key to plants and lichens on smartphones: Estonian
examples”. In: P. L. Nimis and R. Vignes Lebbe (eds.), Tools for Identifying Biodiversity:
Progress and Problems, pp. 195-199, 2010.

450
 Author Index

A Borrell Y. J., 281

Addamiano S., 157 Boscarol C., 445

Addink W., 55 Boujemaa N., 237

Aedo C., 99, 177 Brandner R., 13

Agoo E. M. G., 341 Brewington S. D., 171

Anastasi A., 183 Buono S., 249

Aplikioti M., 231

C
Arena S., 327
Argyrou M., 231 Cabezas F., 99

Assunçao, O., 419 Campo D., 315

Candolfi E., 201
Atanacio R., 31
Cano Calonge C., 163
Caron D., 217
B
Casiraghi M., 269, 289
Bailly N., 31
Causse F., 113, 383
Balenghien T., 201
Cazet J.-P., 383
Balke M., 347
Cesari M., 333
Barberá P., 99
Cêtre-Sossah C., 201
Barbouche N., 343 Chabalier J., 419
Barthélémy D., 221, 237 Chanthavongsa K., 221
Basheer V. S., 345, 353 Chavernac D., 201
Baylac M., 343 Chiatante D., 327
Berendsohn W. G., 1, 7 Chouchene M., 343
Bertolani R., 333 Combourieu Nebout N., 383
Bineesh K. K., 345 Conruyt N., 65, 71, 217
Bisby F. A., 37 Coullet O., 419
Blanco G., 281 Crego-Prieto V., 315
Boar F., 367 Crucianu M., 237
Bonnet P., 221, 237 Cubelio S. S., 345

Borg J. J., 25 Cuming D., 423

451
D Garcia-Vazquez E., 315

De Giovanni M., 25 Garnery L., 343

De la Estrella M., 99 Garros C., 201

De Mattia F., 269 Geiger M., 351

Delécolle J. -C., 201 Gérard D., 107

Diep M. H, 207 Geynet Y., 217

Ditzler S., 349 Giurgiu M., 13, 19, 83, 133, 137

Dorigo Salamon N., 437 Gopalakrishnan A., 345, 353

Drakos A., 231 Graciete L., 219

Dupont J., 383 Gransinigh E., 249

Grard P., 221
E Graziosi G., 307
Grosser D., 65, 71, 217
Esparza I., 323
Grube M., 295

F Guarino R., 157, 405

Gudmundsson G., 171
Faure G., 217
Güntsch A., 7
Fero M., 99
Ferrer M., 373
H
Ferri E., 269
Filipello Marchisio V., 183 Hagedorn G., 13, 19, 59, 77, 83, 89

Fortini P., 257 Haszprunar G., 347

Fradin H., 7 Häuser C. L., 421

Froese R., 31 Hausmann A., 347

Froufe E., 301 Hebert P., 347

Furlan M., 307 Hendrich L., 347

Hetzner S., 77, 137

G Hoffmann A., 421

Hoffmann J., 43
Galán P., 323
Hoffmann N., 7
Galimberti A., 269, 289
Hominick B., 429
Gallagher M. S., 417
Homodi A., 13, 19, 83, 133
Gamarra R., 323
García E., 373 Hopkins D., 395, 401

452
I Loy A., 243, 257, 263

Ialicicco M., 327

M
J MacLeod N., 225

James K., 395 Magrini S., 249, 251

Jong Y. de, 49 Maiocco Ô., 7

Manouselis N., 411
K Martellos S., 13, 59, 77, 115, 127,
133, 151, 437
Kastrantas K., 411
Martin P. A., 65
Kerekes A., 367
Martinoli A., 289
Kirchhoff A., 7
Mathieu B., 201
Kirmitzoglou I., 231
Mathieu D., 121
Kittl C., 13, 361, 437
Maxl E., 361
Kizhakudan J., 353
McGovern T., 171
Kizhakudan S. J., 353
Medina L., 177
Knebelsberger T., 349
Menegoni P., 405
Kodele Krašna I., 379
Mereu A., 445
Kohlbecker A., 7
Mihnev P., 13, 355
Kroupa A., 421
Mohrbeck I., 349
Kuntzelmann E., 7
Monje J. C., 421
Montaño A. J., 163
L
Morris R. A., 13
La Rosa M., 157
Mucedda M., 289
Laakmann S., 349
Muggia L., 295
Labra M., 269
Mulder P., 209
Lakra W. S., 345, 353
Müller A., 7
Lamxay V., 221
Mulrenin A., 25
Lanorsavanh S., 221
Laroche P., 223 N
Larsen K., 301
Nascimbene J., 151
Leht M., 189, 195
Neumann F., 137
Lopes M., 219

453
Nguyen B. L., 207 Randlane T., 189, 195
Nguyen H. P., 207 Raupach M. J., 349
Nicolosi P., 263 Raycheva N., 355
Nikolaou N., 231 Rebecchi L., 333
Nimis P. L., 13, 19, 77, 127, 133, Rekha J. N., 353
151
Rempicci M., 249
Reyes R. Jr., 31
O
Riccamboni R., 445
Onofri S., 249 Richardson A., 401
Ortúñez E., 323 Roberts D., 53
Ouertani W., 237 Rocco M., 327
Romano F., 281
P
Roskov Y. R., 37
Palavitsinis N., 411 Roujinov M., 13
Pallavicini A., 307 Rovellotti O., 419
Papamarkos N., 231
Russo D., 289
Perez J., 315
Pernot T., 121 S
Peters P., 419
Saag A., 189, 195
Petersen A., 25, 171
Sahl A., 419
Pieterse S., 25
Sajeela K. A., 353
Pignatti S., 157, 405
Sampaziotis P., 231
Pinho R.M., 219
Sánchez Laulhé F., 163
Plank A., 13, 77
Sánchez Prado J. A., 281
Press B., 77, 389
Sanz E., 323
Promponas V. J., 231
Sbordoni V., 275
Scaloni A., 327
R
Schalk P., 13, 55, 127, 429, 437
Raj K., 345
Schmidt G., 13, 137
Rakhee C., 353
Schmidt S., 347
Ralambondrainy H., 71
Scholz H., 43
Rambold G., 59
Schuiteman A., 221

454
Scippa G. S., 327 van Spronsen E., 13, 55, 127, 133
Scoppola A., 249, 251 Varese G. C., 183
Seijts D., 127 Vázquez E., 281
Silva M. H., 219 Veja C., 13, 19
Silveira P., 219 Velayos M., 99
Slice D. E., 243 Venin M., 7
Smith V., 53 Viaggi D., 423
Steinmann R., 25 Vignes Lebbe R., 7, 107, 113, 201,
207, 383
Stoitsis J., 411
Viscosi V., 257, 327
Strasser A., 25
von Mering S., 77
Straube N., 351
Voyron S., 183
Svengsuksa B., 221

W
T
Weber G., 13, 77, 89
Talbi K., 419
Targetti S., 423
Z
Tarkus A., 361
Zammit N., 25
Teage I., 25
Tewari S., 345 Zelazny B., 13

Tornincasa P., 307

Trayler A., 25
Triebel D., 13
Trilar T., 95
Trupiano D., 327
Tsilibaris X., 411

U
Uiterwijk M., 213
Ung V., 113, 201

V
van Raamsdonk L. W. D., 145, 213

455