Ecological Applications, 22(8), 2012, pp.

2221–2236
Ó 2012 by the Ecological Society of America

Developing the multitrait concept for functional diversity:

lessons from a system rich in functions but poor in species
ANNA TÖRNROOS1 AND ERIK BONSDORFF
Åbo Akademi University, Department of Biosciences, Environmental and Marine Biology, Artillerigatan 6, Åbo FIN 20520 Finland

Abstract. Studies focusing on the linkage between numerical and functional trait diversity
frequently consider functional diversity indices but rarely evaluate them empirically or
evaluate the use of other than continuous traits such as body size. Here, we present an
extensive compilation on functional knowledge of benthic macrofauna using the categorical
trait approach and scores of both common and rare species for 25 biological traits, including
102 modalities. We empirically quantify functional trait richness, within-trait species richness
(redundancy), and trait variability on a large regional scale (.1000 km), in three
environmentally different areas (basins of a sea), over a long time-span (10 years). To
develop further the usage of multiple categorical traits as an analysis tool, we examine the
effect of sampling effort for the understanding of the functional properties of the benthic
meta-assemblages. We also evaluate the relationship between species richness and trait
richness in order to understand co-variation between trait modalities and how traits are
packaged within species. Results show that the biological diversity in terms of traits could be
distinguished between areas of higher and lower salinity, higher and lower anthropogenic
stress, and higher and lower species richness. A considerably lower number of samples are
needed to portray the functional structure of an area in relation to the taxonomic structure,
thereby demonstrating the advantage of using traits when considering management and
conservation issues. Using categorical traits empirically requires, as shown within this study,
an understanding of the relationship between species richness and expression of traits, co-
variation of traits at different species richness and composition levels, acknowledgment of
differences in trait expressions between common and rare species, and variability in abundance
of species. Empirical trait-based analysis can reveal large-scale differences and insights into
complexities between assemblage structure and function, and simultaneously be a valid tool
for finding generalities.
Key words: biological traits; common and rare species; multiple environmental stressors; spatial scale;
trait co-variation; trait richness and redundancy; zoobenthos.

INTRODUCTION Functional diversity today is about measuring the

Understanding functioning of ecosystems in relation richness and diversity (evenness and divergence) of
to biodiversity has become a central research theme in functional traits (Schleuter et al. 2010). Functional traits
both aquatic and terrestrial ecology during the recent are components of an organism’s phenotype that
decade (Hooper et al. 2005, Solan et al. 2006). The core influence ecosystem level processes (Violle et al. 2007,
issue in understanding the role of biodiversity for Reiss et al. 2009). Traits can be morphological,
ecosystem functioning ultimately comes down to know- physiological, reproductive, or behavioral aspects of an
ing what organisms do within the constraints of their organism and may directly mediate energy and material
environment (Hooper et al. 2005). Through their diverse fluxes or alter abiotic conditions that govern rates of
abilities to coexist and interact on multiple levels, functions (Symstad et al. 2003). Two different approach-
organisms affect ecosystem processes and influence es to measuring this diversity, using multiple traits, have
ecosystem functioning (Tilman 2000). Human activities emerged during recent years. On the one hand,
affect this diversity and functioning by changing the functional diversity indices have evolved rapidly, com-
environment in both aquatic and terrestrial systems, on prising of indices of functional richness, evenness, and
scales from local to global and immediate to long term divergence (Petchey and Gaston 2006, Laliberté and
(Millenium Ecosystem Assessment 2005, Worm et al. Legendre 2010, Mouchet et al. 2010, Schleuter et al.
2006). 2010). Recommended at the moment is that indices
should only include traits that can be measured on a
continuous scale, (i.e., real-value traits such as body
Manuscript received 15 November 2011; revised 13 April
size), rather than splitting a trait into a priori categories
2012; accepted 18 May 2012. Corresponding Editor (ad hoc): C.
Frid. (i.e., categorical or discontinuous traits). Real value
1 E-mail: anna.m.tornroos@abo.fi traits allow for direct use of species information
2221
 Ecological Applications
2222 ANNA TÖRNROOS AND ERIK BONSDORFF
Vol. 22, No. 8

(measured values) rather than requiring grouping or al. 2000) and marine systems (Bremner et al. 2003), there
coding of species to the categories. On the other hand, is a lack of knowledge in how sampling effort influences
the use of categorical traits have received interest the functional interpretation. Also, to develop and
especially in empirical studies, such as macro-ecological properly utilize the trait concept for understanding
studies (Webb et al. 2009, Tyler et al. 2012). Also studies functional community dynamics on regional scales,
assessing anthropogenic impacts on functioning and quantifying not only the number of traits but also
studies evaluating functional aspects for conservation within-trait richness of species and variability as well as
and management purposes have benefitted from using the functional role of common and rare species is urgent
an approach based on categorical traits (de Juan et al. (Ellingsen et al. 2007, Hewitt et al. 2008).
2007, Frid et al. 2008). In marine and freshwater studies, The general objective of this paper is to develop the
this approach has often gone under the name biological use of categorical traits by exploring the relationship
trait analysis (Townsend and Hildrew 1994, Usseglio- between traits and traditional community variables,
Polatera et al. 2000, Bremner et al. 2003). In this such as species number and composition, under different
approach, all traits are divided into sub-categories or environmental conditions. More specifically, we (1) ask
modalities. Thus, the trait mobility can be divided into how trait richness in relation to species richness is
three modalities; sessile, semi-mobile, and mobile. This affected by sampling effort in a set of environmentally
approach, in comparison to the other, does not measure different areas; (2) ask how trait composition, trait
functional diversity per se, give absolute insight into the richness, as well as species richness and variability within
distribution or variability of traits, or provide a direct traits differ on a large regional scale (.1000 km), in
link to functions. However, categorical traits are the different areas (basins of a sea), over a long time-span
present day substitutes for the data that would enable (10 years); (3) evaluate the relationship between species
the sole use of continuous traits in measurements. For richness and trait richness in order to understand
many functional aspects in space and time, this type of covariation between traits and how traits are packaged
information is often the best available knowledge we and distributed within species to extend the scope of the
have about species (Tyler et al. 2012) if no artificial data functional approach; (4) evaluate trait differences
is used, as is often the case in simulation and theoretical between areas and depths considering abundances of
studies. As it is a group-based approach, it still differs species in order to allow for some generality.
from the trophic or functional group approach (Fau- As a base, we present an extensive compilation of soft-
chald and Jumars 1979, Bonsdorff and Pearson 1999, bottom macrofaunal traits for the brackish water
Cardinale et al. 2006) in that more traits are used, not northern Baltic Sea. Utilizing one of the most studied
only related to feeding interactions and trophic structure areas in the world, the coastal system of the Baltic Sea,
but also environmental gradients. This increases the relatively species-poor by nature due to its short post-
potential for identifying functional differences between glacial history and strong structuring environmental
assemblages on different scales (Bremner et al. 2003). gradients (Bonsdorff 2006, Leppäranta and Myrberg
Although theoretical and experimental studies have 2009), allows for a precise and comprehensive functional
outlined the relationship between species and functions assessment with high taxonomic resolution. Thus, this
on small scales (from petri dishes to several square data provides us the possibility to include both common
meters; Naeem and Li 1997), empirical large scales and rare species as species diversity is naturally low and
(hectares to square kilometers), and long-term studies the natural history of most species is well known.
(years), scales relevant for marine ecosystems are scarce. MATERIALS AND METHODS
Evidence of the connection between habitat or small-
scale environmental heterogeneity and functional trait Distribution and abundance data
diversity in marine systems is robust (Bremner et al. In order to develop a base for the analyses and
2003, Hewitt et al. 2008). However, the importance of compile a species and trait data set, we used data on
this linkage at large scales and long term is less clear soft-bottom macrobenthic fauna from the Finnish
(Hooper et al. 2005). Scaling up from local processes to National Coastal Monitoring Program, available in the
regional and larger scales is potentially more difficult, Environmental Information System (HERTTA) data-
nevertheless critical, in taxonomically diverse, environ- base, maintained by the Finnish Environmental Insti-
mentally variable, and highly impacted systems such as tute. The data spanned a time period of 16 years (1988–
estuaries and coastal seas (Lotze et al. 2006, Worm et al. 2004) and spatially covered the entire Finnish coastline,
2006). To do so, using traits, and particularly categorical in total about 1250 km. Altogether 782 stations and
traits as we will focus on from here onward, still requires 1941 sampling visits were included and distributed
empirical testing and understanding of the procedures across the four main basins of the northern Baltic Sea;
and pitfalls for quantifying trait diversity in comparison from the northernmost basin of the Bothnian Bay and
to species diversity and composition. Whereas, for the Bothnian Sea, to the southern basins of the
example, trait composition has been said to be more Archipelago Sea and the Gulf of Finland (65–598 N,
stable than taxonomic composition over extensive 19–278 E; Fig. 1). The purpose of including data from all
biogeographic gradients in both freshwater (Charvet et sub-basins in this first stage was to allow for describing
December 2012 DEVELOPING THE MULTITRAIT CONCEPT 2223

FIG. 1. Location of the study area in the northern Baltic Sea with sampling stations (white dots) and sub-areas indicated, as
well as salinity and the pressure gradient. The pressure index is according to the Baltic Sea pressure index (HELCOM 2010).

functioning in terms of traits along the entire coast, a Based on this large data set, we randomly compiled a
natural gradient of environmental and human-induced smaller data set that was used for all analyses. This
stress. The stations were separated into two depth subset combined data from shallow and deep areas in
groups: 0–10 m (hereafter shallow) and 10–55 m three selected basins along the coast: the Bothnian Bay,
(hereafter deep). The divisions of sub-basins and water the Archipelago Sea, and the Gulf of Finland (Fig. 1).
bodies follow the common national and international To include a temporal aspect, data were included from
typology and division of the Baltic Sea and Finnish six years within a 10-year period (1994, 1996, 1998,
coastal areas (Perus et al. 2004). Benthic samples 2000, 2002, 2004). Five visits were randomly selected per
included were all obtained in a standardized way, using year, per depth stratum and area, altogether 30 visits per
depth, i.e., 60 visits (180 samples) per basin in total. An
an Ekman-Birge grab sampler (225–289 cm2) and a 0.5-
exception had to be made for the deep stratum in the
mm mesh sieve. Only visits with three replicate grab
Gulf of Finland, where not enough visits from 2002
samples were included. Taxonomic resolution was set at
could be found in the monitoring data. Thus, for this
species level when possible, and in accordance with depth strata and area, a total of 25 visits were included.
resolution of trait information available. A thorough The three basin areas in the subset were chosen to
taxonomic quality assurance included verification of maximize the contrasts in climate and environmental
taxonomy and scientific names following the European forcing and represent the ‘‘end points’’ in salinity and
Register of Marine Species, World Register of Marine anthropogenically induced stress. As for the whole of
Species, and the Integrated Taxonomic Information the Baltic Sea, the strongest environmental factor in the
System (ITIS) (databases available online).2,3,4 northern Baltic Sea is salinity, decreasing gradually from
5–7 psu in the southwest, the Archipelago Sea, to 1–6
2 www.marbef.org/data/erms.php psu in the Gulf of Finland in the east, and 1–4 psu in the
3 http://www.marinespecies.org/index.php almost limnic Bothnian Bay in the north (Fig. 1;
4 http://www.itis.gov Rönnberg and Bonsdorff 2004, Leppäranta and Myr-
 Ecological Applications
2224 ANNA TÖRNROOS AND ERIK BONSDORFF
Vol. 22, No. 8

berg 2009). The northern basin differs from the two dwellers to certain degrees, hence coded for several
southern basins in annual light, temperature and riverine modalities (Table 1, Supplement). Consequently, the
discharge dynamics, affecting the input of organic maximum number of scores, i.e., number of species
material and nutrient flow to zoobenthos. Not only the expressions, within any modality, can only be the total
deep bottoms, but also the coastal sedimentary areas number of species, though different modalities may
and zoobenthic assemblages are heavily affected by display differing total numbers of score inputs. The
eutrophication and oxygen deficiency, specifically in the maximum number of scores within the whole trait may
southern basins (Karlson et al. 2002, Conley et al. 2011). thus differ depending on if species can express one or
A higher cumulative pressure and impact is found in the several modalities within a trait. Hence traits are not
Archipelago Sea and specially in the Gulf of Finland, equal in the implication of a modality per se. This
compared to the Bothnian Bay, which is the least difference is an important issue specifically when trait
impacted area in the whole of the Baltic Sea (HELCOM richness and trait redundancy are discussed.
2010) (Fig. 1). Hence, the three areas differ largely in Information on species traits was obtained, supple-
background environmental conditions, providing a mented, and reported here using (1) peer-reviewed,
setting for functional analysis, comparable to other
published sources (articles, books, etc.), (2) information
estuarine environments.
based on the species phenotype, e.g., body design, or
Trait data based on other traits, and (3) expert knowledge or
unpublished sources (Supplement). The information was
The principle for gathering information on species
primarily collected from studies conducted in the region
characteristics was to include as wide a range of traits as
in order to appropriately include region-specific charac-
possible in order to be able to describe the trait diversity
teristics. Species in the Baltic Sea might express other
and general functioning of the assemblages in the most
comprehensive way. Therefore, trait information was ranges of modalities compared to their conspecifics in
gathered on the lowest possible taxonomic level, and fully marine areas. Specific to this region is the presence
only later adjusted to appropriate higher taxonomic of insect larvae as a significant component of the
levels. Adjustments were done using all information for zoobenthos community and thus we included these
the species on the accepted taxonomic level, and when based on their larval characteristics. Still, most traits and
differing trait expressions between species/genera was modality divisions were based on other studies (Bremner
found, the taxon was assigned equal probabilities. This et al. 2003, 2006, Frid et al. 2008), and thus kept
was generally an issue affecting number of modalities comparable with other regions outside the Baltic Sea.
rather than the scoring within a modality. Species were To reach our fourth aim of evaluating trait differences
assigned to 25 traits and 102 sub-categories or modal- between areas considering abundances of species, we
ities (Table 1). The traits reflect key components of weighed the functional trait expressions with the
functioning and proxies for ecological processes (Table abundances of the species in the randomly assembled
1; see also Frid et al. 2008). Scoring of species to traits subset. This weighting procedure was done through
was done using the fuzzy coding procedure (Chevenet et multiplying the trait scores of the species with the
al. 1994), with a scoring range of 0 (no affinity) to 3 abundances of the species, producing a sample-by trait
(total affinity). Some species are known to display matrix. Prior to this step, abundances were square root
plasticity in certain traits, and to overcome this on a transformed, in order to down-weigh abundant species
large scale when information on, e.g., sediment charac- (Anderson et al. 2008), and trait scorings standardized
teristics for each sample was scarce, species were to one within a trait for a species (Boström et al. 2010)
assigned equal probabilities to express the different
potential trait modalities. We also acknowledged the Data analysis
fact that traits differ in how many modalities individual To illustrate the effect of sampling effort (visits) on
species can be assigned to within a trait, in three ways. the functional richness in relation to species richness,
First, traits for which a species can only express one sample-based species rarefaction curves for the three
modality of a number of possibilities, e.g., maximum selected areas and two depth strata were calculated using
size, we categorize as a discrete trait (Table 1, EstimateS (Version 8.2; available online).5 In order to
Supplement). Second, for a combinatory trait, species examine the trait richness in relation to area, we
can be assigned to one, several, or all of the modalities, performed the same analysis using trait modalities
for example the trait reproductive technique, as some instead of species, utilizing presence/absence of trait
species can reproduce both sexually and/or asexually.
modalities in the randomly assembled data set. This
The third possibility is a binary trait, where species can
analysis method randomly permutes (50 permutation)
be assigned to only one modality or, if not to that
the order of species or trait modalities considered in an
specific modality, to any combination of the others. For
analysis.
example the trait living habit for which some species can
only be attached and coded only for this modality, but
other species may be both free-living and burrow 5 http://viceroy.eeb.uconn.edu/estimates
December 2012 DEVELOPING THE MULTITRAIT CONCEPT 2225

TABLE 1. List of biological traits and modalities (subcategories).

Key mechanisms
Trait Modality Explanation of functions Function
Morphology/physiology
Maximum size  15 mm small rate of production production
5 mm1 cm small–medium amount of exchange, elemental cycling
facilitation
13 cm medium respiration
35 cm medium–large
.5 cm large
Body design  verminiform, wormlike, lacking true segments rate of production production
unsegmented
verminiform, wormlike, divided into semi- proxy for litter quality decomposition
segmented independent units
bivalved a shell of two calcacerous valves
joined by a flexible ligament
turbinate whorled
articulate jointed, arthrous (composed
segments)
Protection  no protection no protection at all proxy for palatability elemental cycling
tube uses a built tube as protection decomposition
burrow uses loose burrows as protection production
case uses built cases as protection or
shields
soft shell an exoskeleton or hard cuticula
hard shell a hard protective shell, such as,
e.g., a mussel shell
Body flexibility  non-flexible ,10 degrees proxy for palatability production
very flexible .45 degrees
flexible 10–45 degrees
jointed
Fragility  fragile likely to crack as a result of proxy for palatability production
physical impact
intermediate liable to suffer minor damage
robust unlikely to be damage by
physical impact (e.g., hard,
leathery)
Degree of dependency  low totally free above sediment transport elemental cycling
(pelagos-
benthos)
temporary temporary in/on sediment transport elemental cycling
(within
benthos)
permanent permanent in/on sediment
Reproduction/development
Longevity  ,1 yr lives less than one year energy fixations, production
turnover, prod. rate
1–2 yr lives between one to two years elemental cycling
2–5 yr lives between two to five years
5–10 yr lives between five to ten years
Time to maturity  ,20% takes ,20 % of lifetime to reach proxy for turnover production
maturity (propagule supply)
20–50% takes one-half (50%) of lifetime
to reach maturity
.50% takes more than one-half (50%)
of lifetime to reach maturity
Reproductive techniqueà asexual budding/fragmentation, non- proxy for turnover production
fertilized eggs laid by mother/
individual
sexual rate of production
Sexual differentiationà gonochoristic having separate sexes proxy for turnover production
hermaphrodite capable of producing both ova
and spermatozoa either at the
same time
parthenogenetic the ovum develops without
fertilization (a form of asexual
multiplication)
Fertilization typeà nonfertile e.g., parthenogenetic proxy for turnover production
external fertilization outside body
internal fertilization inside body
Developmental fragmentation asexual both fragmentation and proxy for turnover production
techniqueà budding
 Ecological Applications
2226 ANNA TÖRNROOS AND ERIK BONSDORFF
Vol. 22, No. 8

TABLE 1. Continued.

Key mechanisms
Trait Modality Explanation of functions Function
oviparous non-fertilized eggs
(hermaphrodite,
parthenogenetic) or fertilized
eggs laid or spawned
ovoviviparous egg layer/brooder
viviparous develop within and nourishment
from the maternal body,
parental or no care
Larval typeà planktotrophic feeding on material captured proxy for recruitment elemental
from the plankton success transport
lecitotrophic nourished on internal resources, production
yolk
direct direct development of mini
development adults
Reproductive frequency  semelparous breeding only once then dying proxy for turnover production
annual episodic breeds every year in desecrate
periods
annual breeds every year over an
protracted extended or drawn out period
Living
Living habit (how attached adherent to a substratum (95% ability to perform elemental cycling
they live)§ of adult time) transport and (pelagos-
extent benthos)
tube-dweller builds and dwells in a tube(s) elemental cycling
(within
benthos)
burrow dweller builds and dwells in a burrow(s)
case builder builds and dwells in a case
(more pronounced ‘‘housing’’
than a tube)
free living in or on sediment, in water
column
Environmental position infauna (.5 cm) living within the substrate, susceptibility to production
(where they live)à deeper than 5 cm cropping/predation
infauna (2–5 cm) living within the substrate elemental cycling
between 25 cm (within
benthos)
infauna (top 2 living within the top 2 cm of the
cm) substrate
epibenthic living on the surface of the
substrate
bentho-pelagic living in the water column but elemental cycling
(primarily/occasionally) feeds (pelagos-
on the bottom benthos)
Feeding
Feeding position (where suspension obtaining food by actively production
they feed)à feeder sweeping or by holding out a
filter
surface feeder obtaining food from the surface transport from elemental cycling
of the substratum pelagos (pelagos-
benthos)
sub-surface obtaining food from within the transport within elemental cycling
feeder substratum sediment (within
benthos)
selective particle obtaining food by targeting prey facilitation/provision decomposition
feeder or items (in sediment or water of material
column)
miner obtaining food through mining
in, e.g., woody material
parasite
Feeding habit (trophic detritivore feeds on detritus energy fixation/rate of production
position, what they transfer
eat)à omnivore feeds on mixed diet of plant and
animal material
herbivore feeds on plants
carnivore feeds on animals (predator)
scavenger feeds on dead organic material
parasite obtaining food from another
animal (host)
December 2012 DEVELOPING THE MULTITRAIT CONCEPT 2227

TABLE 1. Continued.

Key mechanisms
Trait Modality Explanation of functions Function
Resource capture jawed jaws, mandibles consumption production
method (How they siphon elemental cycling
feed)à tentaculate using tentacles decomposition
pharynx both jawed and not jawed
radula rasping
net build nets for capturing food
particles
Movement
Mobilityà sessile temporary (e.g. Mytilus edulis) turnover production
semi-mobile transport elemental cycling
mobile
Movement methodà swimmer movement with fins, legs, facilitation of elemental cycling
appendages via undulatory materials (pelagos-
movements benthos)
rafter/drifter rafting on, e.g., drifting algal elemental cycling
mats, drifting (within
benthos)
crawler movement on the substratum
via muscles, legs, appendages
byssus acknowledged occasional transport
movement by byssus threads
tube-builder moves when in a tube
burrower moves in a burrow
Migration (seasonal, non-migratory resident turnover production
reproductive)§ vertical migratory between sediment and transport elemental cycling
water column (pelagos-
benthos)
horizontal migratory from shallow to deep elemental cycling
(vice versa) (within
benthos)
Dispersal§ non-dispersal resident movement of decomposition
resources
local frequent movement within production
habitat (10–1000 m)
long distance frequent movement out of
habitat (km)
Behavior
Sediment transport§ no transport transport, extent of elemental cycling
bioturbation (within
benthos)
diffusive mixing vertical diffusive transport (e.g., decomposition
reworking, excavating)
surface surface deposition of particles
deposition (e.g., defecation or egestion)
conveyer belt translocation of sediment within
transport the sediment, deep-top (e.g.,
feeding, excavation)
reverse conveyer subduction of particles from
belt transport surface to some depth (e.g.,
feeding, defecation)
Sociabilityà solitary living alone consumption rate elemental cycling
gregarious living in groups (aggregated decomposition
together) occasionally (e.g.,
mussels)
aggregated living in groups, growing in
clusters, schooling
Notes: Labels and explanations are included as well as examples of mechanisms and linkages to functions. Traits (25 in all),
modalities (102 in all), and explanations are adopted from trait literature (see biotic glossary under the MarLin network: http://
www.marlin.ac.uk/biotic/upload.php). Examples of traits linked to key mechanisms and functions are based on Bremner et al.
(2003, 2006) and Frid et al. (2008).
  Discrete trait (species can only express one modality of the trait).
à Combinatory trait (species can be assigned to one, several, or all of the modalities in the trait).
§ Binary trait (species can be assigned to only one modality, or if not to that one modality to any combination of the others in the
trait).
 Ecological Applications
2228 ANNA TÖRNROOS AND ERIK BONSDORFF
Vol. 22, No. 8

To determine the functional composition in the three presented based on the national quantitative monitoring
different end-point areas, we calculated trait richness as data and notes of species occurrence in other literature
the number of modalities expressed (presence/absence) references (Supplement). For trait assignments, we only
and functional redundancy as number of taxa per used the quantitative monitoring data and a division
modality (taxa richness within modalities) over time into 56 taxa was done. This division was inevitable due
for each area and depth. Redundancy is here simply to inconsistency in taxonomic resolution of the moni-
used as shorthand for species richness within trait toring data and the level of trait knowledge found for
modalities. The definition is not a strict one, since any the macrofauna. The 56 taxa expressed on average 29
functional effect, in space and time, of removing a trait modalities per species, indicating that species, in
species is not measured per se, and thus it is not known if general, express more than one modality per trait, which
a function remains exactly the same after the removal highlights the different trait types (Supplement). Out of
(Walker 1992). As a measure of the stability in the 25 traits, 11 belonged to the discrete trait type, 10 to
functional redundancy, variability in number of species the combinatory type, and 4 to the binary type of traits.
per modality was calculated using the coefficient of In general, the total number of taxa scores for single
variation (CV) for each area and depth over time. modalities varied from 1 (four modalities: less than one-
For further evaluating how species richness affect the half lifetime to maturity, attached living habit, sessile
functional understanding, using categorical traits, we mobility though movement with byssus threads), to the
calculated the cumulative number of trait modalities maximum of 53 scores (one modality: sexual reproduc-
expressed by (1) all species, (2) all common species, and tive technique) (Supplement).
(3) all rare species found in the randomly assembled
Effect of sampling effort on species and trait richness
subset, irrespective of area or depth. We defined
common and rare species as species occurring in 50% In total, 53 of the 56 species/taxa signed to traits
and 10% of the visits, respectively. This definition based on the quantitative data set were found in the
implies that a species is, e.g., not necessarily rare in all randomly assembled data set, representing all functional
areas. Since species number differed greatly between important species. The species-accumulation curves
common (10) and rare species (38), we also randomly demonstrated differences in species richness between
drew 10 species out of the rare species pool to be able to the three geographical areas and the two depth strata,
compare these properly. To relate this to the empirical for which Archipelago Sea shallow had the highest
data we randomly drew ten species per area and depth richness (37 species/taxa; Fig. 2a). The different line
from the data subset. These different compilations were curvatures showed the higher number of visits required
analyzed using rarefaction curves in the EstimateS to sample the species-rich Archipelago Sea (higher
program, based on the trait expressions of species in sampling effort) compared to the other two areas, or
terms of presence/absence of a trait modality. In the shallow areas compared to the deep. The trait
comparison to only looking at presence/absence of trait accumulation curves showed an analogous pattern in
modalities we also evaluated the effect of using trait relation to differences between areas for species accu-
scores for species, comparing common and rare species mulation curves (Fig. 2b). The curves leveled off in each
in and NMDS analysis (nonmetric multidimensional area and depth, but all possible modalities, i.e., the
scaling) based on Euclidean distance. predetermined 102, were found only in the shallow area
In order to graphically visualize and compare species of the Archipelago Sea. A marked difference between
abundance patterns in the different sub-areas with the the number of species and the number of trait modalities
abundance-weighted trait patterns, two separate NMDS gained by sampling a similar area (number of visits) is
analyses based on the Bray-Curtis similarity index noteworthy (Fig. 2). Considering all areas and depths, a
matrix were performed on the abundance matrix and minimum between 17 and 25 visits were needed to
the sample-by trait matrix respectively. To test for the obtain 90% of the species, compared to only two to six
variability in the community and functional structure visits to acquire 90% of the trait modalities (Fig. 2).
among areas and depths, a two-way PERMANOVA Trait richness, composition, and redundancy
was performed on both matrices (with the method
unrestricted permutation of raw data, number of The two extremes in the salinity gradient, the Archi-
permutations ¼ 9999). The factors used were area (three pelago Sea and the Bothnian Bay, differed functionally by
levels, random) and depth (two levels, random). All 13 trait modalities (Table 2). When moving from the south
multivariate analyses were done in PRIMER 6.0 and to the north, the identity of the modalities revealed a loss
PERMANOVA (Anderson et al. 2008). in small, flexible organisms with a very long longevity but
short (less than one-half of lifetime) maturity, a parthe-
RESULTS nogenetic sexual differentiation and both plankto- and
lecitotrophic larval types. Lost were also modalities such
General community structure and trait assignment as attached, tentaculate, sessile, movement by byssus
A comprehensive benthic species list for the Finnish threads, and sediment transport characteristics such as
coastal region with a grand total of 143 species/taxa is conveyer belt- and reversed conveyer belt transport. These
December 2012 DEVELOPING THE MULTITRAIT CONCEPT 2229

FIG. 2. Sample-based rarefaction curves for

(a) taxa and (b) traits (modalities) based on a
subset with 30 randomly gathered visits covering
a 10-year period (1994–2004) in each of the three
selected areas and depth zones (key to abbrevi-
ations: AS, Archipelago Sea; GF, Gulf of Fin-
land; BB, Bothnian Bay; S, shallow, 0–10 m; D,
deep, 10–55 m). Note that three samples were
taken in one visit. Only 25 visits were included for
the deep zone in the Gulf of Finland, due to lack
of data from year 2002. Note that the maximum
number possible, 102 modalities, is found in the
Archipelago Sea, shallow. The number of visits
required to sample 90% of the species or the trait
modalities is annotated.

latter modalities are expressed by, e.g., the blue mussel, group of 12 modalities, with a median over two species/
Mytilus edulis, or polychaete worms, both lost when taxa was found to be general to all areas over time,
moving in the south-north gradient (Supplement). In the hence representing the typical functional characteristics
most anthropogenically stressed area, the Gulf of Finland, and reveals a ‘‘type organism’’ of the region (Table 3).
six modalities were absent compared to the Archipelago Based on these modalities (shown in boldface type in
Sea (Table 2). Aside from the differences in absence of Table 3), the typical northern Baltic Sea benthic
traits, the three areas had 72 trait modalities in common, organism is of medium size, reproduces sexually and is
i.e., these constitute the common functional pool in the gonochoristic (separate sexes), has larvae that are direct
region (Table 2). developers, and lives as an adult permanently in or on
Richness of taxa within trait modalities, or redun- the sediment, both in the top 1 cm of the sediment and
dancy, revealed a different pattern compared to the one epibenthically. It is further a mobile detritivore with
found for the number of modalities expressed. In total, local dispersal and diffusively mixes the sediment as it
48 modalities (47% of all) displayed a median of over lives solitarily.
two taxa per modality over the 10-year period (Table 3).
The area with most modalities and also overall highest Covariation of traits and ‘‘trait packaging’’ within species
redundancy was the Archipelago Sea. In this area a total The accumulation curves based on all 53 species
of 34 modalities had a median over two species/taxa, found in the subset shows that as more species are added
and the maximum number of taxa found within one more trait modalities are expressed, hence, trait richness
such modality was eight (sexual reproductive technique; is determined by species richness (Fig. 3a). The fact that
Table 3). The area displaying the lowest redundancy was the curve for all 53 species levels off before reaching the
the most stressed area, the Gulf of Finland, with 12 predetermined limit of 102 modalities shows that traits
modalities corresponding to only about 12% of the total are also packaged into species, since a new species does
number of 102 modalities (Table 3). The variability not necessarily add a new trait modality. This follows
across years, measured as CV, was similar and low for theory (Schleuter et al. 2010) and is also illustrated in the
all areas varying between 0.21 and 0.63 (Table 3). A calculations above of trait richness and redundancy
 Ecological Applications
2230 ANNA TÖRNROOS AND ERIK BONSDORFF
Vol. 22, No. 8

TABLE 2. Traits and modalities present (X) and absent (blank) in the different areas (AS, Archipelago Sea; GF, Gulf of Finland;
BB, Bothnian Bay; S, shallow, 0–10 m; D, deep, 10–55 m).

Traits Modalities ASS ASD GFS GFD BBS BBD

Maximum size  small X X
medium, large X X X X X
Body design  bivalved X X X X X
turbinate X X X X X
Protection  burrow X X X X X
case X X X X
hard shell X X X X X
Body flexibility  non-flexible X X X X X
flexible X X X
Longevity  very long X X X X
Time to maturity  less than half lifetime X X
Sexual differentiationà parthenogenetic X X X X
Larval typeà planktotrophic X X X X
lecitotrophic X X X X
Reproductive frequency  annual episodic X X X X X
Living habit§ attached X X
case builder X X X
Environmental positionà benthic pelagic X X X X
Feeding habità parasite X X
scavenger X X X X X
Feeding positionà parasite X X
Resource capture methodà siphon X X X X X
tentaculate X X X X
radula X X X X X
Mobility  sessile X X
semi mobile X X X X X
Movement methodà byssus X X
Sediment transport§ conveyor belt transport X X X X
reverse conveyor belt transport X X X X
Sociability  aggregated X X X X X
Other traits (19–25) other modalities (31–102) X X X X X X
Total traits 18 (max. 25) for sub-areas total modalities 30 (max. 102) for sub-areas 0 4 8 11 15 26
Total traits 18 (max. 25) for areas total modalities 30 (max. 102) for areas 0 6 13
Note: Modalities in boldface type are absent in all areas but the Archipelago Sea. The last two rows indicate the total number of
traits and modalities displayed in the table. Out of the total number of traits (18 out of 25 possible) and modalities (30 out of 102
possible) displayed in the table, the first row indicates the number of traits displayed per sub-area (e.g. ASS, ASD, etc.). The second
row indicates the number of traits displayed per area (i.e., AS [Archipelago Sea], GF [Gulf of Finland], and BB [Bothnian Bay]).
  Discrete trait (species can only express one modality of the trait).
à Combinatory trait (species can be assigned to one, several, or all of the modalities in the trait).
§ Binary trait (species can be assigned to only one modality, of if not to that one modality to any combination of the others in the
trait).

(Tables 2 and 3). This raises the question of covariation number (higher loss) of trait modalities than expected
of trait modalities, i.e., that there are sets of traits that (89 compared to 99), while Bothnian Bay showed a
covary or to certain degrees accompanies others so that higher number (lower loss) of modalities (73 compared
if, e.g., the trait modality bivalved (body design) is to 79).
present then the modality hard shell (protection) is also. Complicating this further, the result of the accumu-
Loosing one or a few species might, depending on lation curves for the common and 10 randomly drawn
species composition, result in a sharp loss in trait rare species differed. Common species showed a higher
modalities, as the lost species are the only ones number of species per modalities than did the rare
expressing particularly those modalities. The curve species and, hence, expressing more modalities indepen-
implies that this might be true when species are few dently of species richness (Fig. 3a). Common and rare
(the beginning of the curve), this will be less likely when species had 78 trait modalities in common, 13 trait
there are more species. This loss is not easily deduced at modalities were only expressed by common species, and
low species level, as shown for the accumulation curves eight modalities only by the rare species. These results
for 10 randomly drawn species from the species pool of suggest that number of modalities can also depend on
the different areas and depths (Fig. 3b). Archipelago Sea the number of common and rare species included in the
shallow still showed the highest trait richness but the analysis, though this still requires further testing. A
maximum number of trait modalities that the 10 distinction between common and rare species could also
randomly drawn species expressed differed from what be seen using the trait scores, which portray relevant
was expected based on the trait accumulation curves for ecological variation and not only expression or not, of a
the areas (Fig. 2b). Archipelago Sea showed a lower trait modality (Fig. 4).
December 2012 DEVELOPING THE MULTITRAIT CONCEPT 2231

TABLE 3. Number of species and modalities displaying redundancy in the different areas.

Traits Modalities ASS ASD GFS GFD BBS BBD

Maximum size  medium 4 (0.37) 5 (0.33) 3 (0.31) 3 (0.38) 3 (0.34) 3 (0.30)
Body design  verminiform segmented 3 (0.45) 3 (0.39) 3 (0.37) X 3 (0.32) 3 (0.21)
Protection  no protection X X 3 (0.38) X 3 (0.28) 3 (0.21)
hard shell 3.5 (0.49) X X X X
Body flexibility  non-flexible 3.5 (0.49) X X X X
Fragility  intermediate 3 (0.54) 4 (0.31) X X 3 (0.41) 3 (0.47)
robust 4 (0.48) X X X X X
Degree of dependency  temporary 3 (0.57) X X X X X
permanent 6 (0.41) 6 (0.26) 3 (0.34) 4 (0.35) 4 (0.27) X
Longevity  short 4 (0.45) 4 (0.41) X X 3 (0.42) X
Time to maturity  half of lifetime 5 (0.45) 3 (0.46) X X X X
longer than half of lifetime 3 (0.53) 4 (0.39) X X 3 (0.45) X
Reproductive techniqueà sexual 8 (0.40) 6 (0.34) 4 (0.36) 4 (0.37) 4.5 (0.36) 4 (0.34)
Sexual differentiationà gonochoristic 7 (0.41) 6 (0.35) 3 (0.43) 3 (0.50) 3 (0.39) 3 (0.46)
Fertilization typeà external 5 (0.45) 3 (0.40) X X
internal 3.5 (0.58) 4 (0.45) 3 (0.47) X 5 (0.38) 4 (0.34)
Developmental techniqueà oviparous 2.5 (0.63) 3 (0.40) 3 (0.45) X 3 (0.42) 3 (0.21)
ovoviviparous 5 (0.50) 4 (0.47) X X X X
Larval typeà planktotrophic 2.5 (0.59) X X X X X
direct development 5 (0.46) 5.5 (0.37) 3 (0.45) 3 (0.42) 5 (0.38) 4 (0.34)
Reproductive frequency  semelparous X 3 (0.40) X X 3 (0.39) X
annual protracted 4 (0.50) 3 (0.57) X X X X
Living habit§ burrow dweller 3 (0.46) 3 (0.34) X X X X
free 4 (0.53) 3 (0.59) X X 3 (0.57) X
Environmental positionà infauna middle 3 (0.38) 3 (0.32) X X X X
infauna top 6 (0.33) 6 (0.25) 3.5 (0.33) 4 (0.35) 4.5 (0.31) 4 (0.26)
epibenthic 4.5 (0.48) 4 (0.51) 3 (0.43) 3 (0.44) 4 (0.39) 3 (0.30)
Feeding positionà suspension 3 (0.45) X X X X X
surface 4 (0.40) 4 (0.39) X X X X
Feeding habità detritivore 7 (0.39) 6 (0.31) 3 (0.44) 3 (0.32) 3 (0.41) 3 (0.34)
omnivore 3 (0.47) 3.5 (0.41) X X 4 (0.39) 3 (0.34)
carnivore 3 (0.50) 3 (0.45) X X X X
Mobility  semi mobile 2.5 (0.53) X X X X
mobile 6 (0.39) 6 (0.38) 3 (0.44) 3 (0.34) 4 (0.35) 4 (0.34)
Movement methodà swimmer X 3 (0.45) X X 3 (0.43) 3 (0.44)
rafter/drifter 2.5 (0.54) X X X X X
crawler 5.5 (0.40) 4 (0.52) X X 4 (0.42) 3 (0.34)
burrower 3 (0.47) 3 (0.40) X X X X
Migration§ nonmigratory 6.5 (0.46) 4 (0.46) X X X X
vertical X 3 (0.36) X X X X
Dispersal§ non dispersal 3 (0.54) X X X X X
local 7 (0.37) 6 (0.38) 4 (0.39) 3 (0.32) 5 (0.36) 4 (0.34)
long distance 3.5 (0.49) 3 (0.58) X X X X
Sediment transport§ no transport 4 (0.51) 3 (0.55) X X X X
diffusive mixing 6 (0.35) 5.5 (0.29) 4 (0.35) 3 (0.38) 4.5 (0.31) 4 (0.29)
surface deposition 3 (0.48) X X X X X
Sociability  solitary 6 (0.41) 5 (0.33) 3.5 (0.35) 3 (0.35) 4 (0.44) 3 (0.26)
gregarious 5 (0.40) 4 (0.37) X X X X
Total traits 24 (max. 25) for total modalities 48 (max. 102) 44 37 16 12 24 20
sub-areas for sub-areas
Total traits 24 (max. 25) for total modalities 48 (max. 102) 34 12 20
areas for areas
Notes: Values are given only for modalities with a median of .2 species (with CV in parentheses). ‘‘X’’ marks modalities in that
are present but with a median ,2 species; other modalities are absent (see Table 2 for comparison). Boldface type indicates
modalities where a median of .2 species is found in all areas and thus represent the ‘‘type organism’’ of the northern Baltic Sea. The
only trait not represented here, i.e., with a modality displaying a median of species .2, is resource capture method. The last two
rows are as in Table 2.
  Discrete trait (species can only express one modality of the trait).
à Combinatory trait (species can be assigned to one, several or all of the modalities in the trait).
§ Binary trait (species can be assigned to only one modality, of if not to that one modality to any combination of the others in the
trait).

Trait differences in relation to species abundance analysis indicated a significant difference between the
species rich Archipelago Sea and the two other areas for
The NMDS analysis for the abundance subset and for
the interaction between area and depth both in the
the sample-by-trait matrix was in line with the findings analysis based on species abundance (F2, 169 ¼ 3.6944, P
in the numerical analysis of differences between areas ¼ 0.0002) and the abundance-weighted sample-by-trait
and depths (Fig. 5a and b). The PERMANOVA matrix (F2, 169 ¼ 3.2288, P ¼ 0.0026).
 Ecological Applications
2232 ANNA TÖRNROOS AND ERIK BONSDORFF
Vol. 22, No. 8

FIG. 3. Rarefaction curves based on species

trait profiles (presence/absence of trait modali-
ties) for (a) all species, common species, rare
species, and a randomized draw of 10 rare species
in the subset, irrespective of area and depth, and
(b) a randomized draw of 10 species from the
different areas and depth zones (AS, Archipelago
Sea; GF, Gulf of Finland; BB, Bothnian Bay; S,
shallow, 0–10 m; D, deep, 10–55 m).

DISCUSSION This study concerned the diverse shallow photic areas

as an alternative to the more often investigated open sea
Our study identified biological traits for all function-
areas in the Baltic Sea, which are known to inhabit only
ally important species (56 taxa both rare and common) a handful of species (Bonsdorff 2006, Villnäs and
along the entire Finnish coastline. The analyses showed Norkko 2011). The bulk of the species along the Finnish
that the biological diversity in terms of traits could be coast is also found in fully marine areas, making
distinguished between areas of higher and lower salinity, comparisons relevant over and between systems. We
higher and lower anthropogenic stress, and higher and also included an often forgotten (see Nohrén et al. 2009),
lower species richness. Using categorical traits, as is but highly interesting faunal component, namely the
done within the biological trait analysis, require an insects, both aquatic and terrestrial species with aquatic
understanding of the relationship between species benthic larval stages. This group, also present in marine
estuaries (Nohrén et al. 2009), was represented in this
richness and expression of traits, covariation of traits
study primarily by midges (e.g., Chironomidae and
at different species richness and composition, and
Certaopogonidae) and identified as a general and
acknowledgment of both common and rare species and important functional contributor along the entire
variability in abundance of species. However, all of these coastal and salinity gradient. Our results reflect the
issues need further evaluation before this approach can common hypothesis of decreasing diversity with salinity,
be implemented, e.g., into management. as noted in the southern Baltic Sea and other estuarine
December 2012 DEVELOPING THE MULTITRAIT CONCEPT 2233

FIG. 4. Nonmetric multidimensional scaling (NMDS) plot of rare (solid circles) and common (open triangles) species trait
composition based on trait scores.

areas (Elmgren and Hill 1997, Bonsdorff and Pearson theory (Loreau 2004). For the Baltic Sea, the argument
1999), but highlight the loss, particularly of polychaetes in general has been exactly this, that loss or drastic
and individual mollusc species, rather than the gain of reduction of a single species most probably leads to loss
taxa, such as insects. of a whole functional group and thus greatly effect
Trait diversity and redundancy are, as are other ecosystem functioning (Elmgren and Hill 1997, Bons-
measures of diversity, scale dependent (Fonseca and dorff and Pearson 1999). On the contrary, the ecosystem
Ganade 2001). Since assemblages are defined for a given is simple with a short successional history, and hosts
area at a given time, sampling a larger area means more estuarine species with mixed origins living on the edge of
species and thus possibly more traits. Our assessment of their stress tolerances and hence, not co-evolved or
how sampling effect could affect trait analysis showed highly specialized (Elmgren and Hill 1997, Bonsdorff
that considerably fewer numbers of samples are needed 2006). This reasoning would imply that there might be
to portray the functional structure of an area in relation open niches and that species tolerating the conditions
to the biotic (taxonomic) structure. Sampling a larger could also be more general and similar in their trait
area may not necessarily reveal additional trait modal- expressions, hence give room for redundancy (Norkko
ities. In this respect, this type of trait analysis is a cost- et al. 2011). Though we found general differences in
effective tool for the understanding of coastal soft- traits between areas, recognizing both common func-
bottom communities, because large scale coverage may tional characteristics (type organism) and characteristics
not be required in order to obtain a picture of how the seldom expressed (e.g., traits specific to rare and
systems differ in functional aspects. common species) is important, especially in a rapidly
We found that a species/taxa expressed on average changing environment. For species to withstand chang-
more than one modality per trait. For one modality a es, possessing some of the common characteristics of the
total of 53 taxa scores were found, i.e., almost all 56 taxa type organism might be enough, however to be a winner
in the analysis expressed it, possibly indicating that there one might need e.g., traits expressed by the rare species.
is potential for redundancy across the whole region for Empirical studies such as ours, simultaneously inves-
some traits. It has been argued theoretically that tigating trait richness and redundancy, encompassing a
functional redundancy is not expected on large scales wide range of ecosystem processes, are fragmentary and
since spatial and temporal environmental variation help thus in demand (Fonseca and Ganade 2001). However,
maintain coexistence (Loureau 2004). How this relates treating trait richness and redundancy levels with
to natural ecosystems is not clear, however (Hooper et caution is important as classification schemes and
al. 2005). In the shallow coastal zones and benthic scoring methods will affect the interpretation of patterns
communities of the northern Baltic Sea, both small and observed (Fonseca and Ganade 2001, Bremner et al.
large scales are influenced by the variability in salinity 2006). As we highlighted for the multitrait method, the
and abiotic factors affecting food supply and hydro- types of traits chosen for the analysis will also affect the
graphic disturbance (Bonsdorff et al. 2003), variables results. If a modality reflects the number of species
that should not allow for redundancy, according to directly, as for some types of traits where species can
 Ecological Applications
2234 ANNA TÖRNROOS AND ERIK BONSDORFF
Vol. 22, No. 8

FIG. 5. NMDS plot of (a) taxon abundance composition and (b) biological trait composition based on the subset data of
randomly gathered benthic stations covering a 10-year period (1994–2004).

only express one modality (the discrete or binary type), dependent on species richness and could be relevant
higher redundancy in such a modality may be important information when species richness is low. In relation to
for functioning in that it directly buffers for loss of a this, our results imply that common and rare species
function. In contrast, a trait for which one species can express different traits and that this could also be
express several modalities to varying degrees, the important in comparing different areas and warrants
probability of several species covering the modality further analysis, as indicated earlier (Ellingsen et al.
and potential function is higher. Another issue that 2007). Our results demonstrate that the multitrait
needs to be resolved is how much the correlation of method shows potential to develop theoretically and
traits, i.e., one trait complementing others, influences the also for practical management purposes, but protocols
functional understanding of a system. Based on our for analyses are still needed (Bonsdorff and Pearson
results, we highlight that a pre-analysis on how many 1999, Bremner et al. 2006, Naeem and Bunker 2009).
and which traits that covary could be useful as a means We conclude that the biological trait analysis proved
to eliminate duplicated information. However, this is to be useful in describing the functional patterns on
December 2012 DEVELOPING THE MULTITRAIT CONCEPT 2235

large to small regional scales between areas differing in de Juan, S., S. F. Thrush, and M. Demestre. 2007. Functional
salinity and human impact, in a comparatively species- changes as indicators of trawling disturbance on a benthic
community located in a fishing ground (NW Mediterranean
poor system, the Baltic Sea. The species and trait data Sea). Marine Ecology Progress Series 334:117–129.
set developed (Supplement) can hopefully aid further Ellingsen, K. E., J. E. Hewitt, and S. F. Thrush. 2007. Rare
research on functional diversity not only in the region, species, habitat diversity and functional redundancy in
but also allow for comparisons between marine areas, marine benthos. Journal of Sea Research 58:291–301.
Elmgren, R., and C. Hill. 1997. Ecosystem function at low
since the traits and modalities described are not specific,
biodiversity—the Baltic example. Pages 319–336 in R. F. G.
but rather applied to the Baltic Sea. We see our data set Ormond, J. D. Gage, and M. Angel, editors. Marine
and analysis approach as an additional base and tool for biodiversity patterns and processes. University Press, Cam-
management of the coastal soft-bottom areas in the bridge, UK.
region. Our study serves as a marine empirical example Fauchald, K., and P. A. Jumars. 1979. The diet of worms: a
study of polychaete feeding guilds. Oceanography and
in the trait debate, which has to date mostly been Marine Biology: An Annual Review 17:193–284.
advanced by theoretical simulation studies or experi- Fonseca, C. B., and G. Ganade. 2001. Species functional
mental studies in terrestrial environments. redundancy, random extinctions and the stability of ecosys-
tems. Journal of Ecology 89:118–125.
ACKNOWLEDGMENTS Frid, C. L. J., O. A. L. Paramor, S. Brockington, and J.
We thank Julie Bremner, Marie Nordström, and Katri Bremner. 2008. Incorporating ecological functioning into the
Aarnio for providing fruitful comments, which greatly im- designation and management of marine protected areas.
proved the manuscript. We also thank the reviewers for an in- Hydrobiologia 606:69–79.
depth examination of the analysis and manuscript. Thanks also HELCOM. 2010. Ecosystem health of the Baltic Sea 2003–
to Jens Perus for help with data gathering and compilation. 2007: HELCOM Initial Holistic Assessment. Baltic Sea
Financial support was received from the Maj and Tor Nessling Environmental Proceedings No. 122. Helsinki Commission,
Foundation and Stiftelsen för Åbo Akademi. Helsinki, Finland.
Hewitt, J., S. F. Thrush, and P. K. Dayton. 2008. Habitat
LITERATURE CITED variation, species diversity and ecological functioning in a
marine system. Journal of Experimental Marine Biology and
Anderson, M. J., R. N. Gorley, and K. R. Clarke. 2008. Ecology 366:116–122.
PERMANOVAþ for PRIMER: guide to software and Hooper, D. U., et al. 2005. Effects of biodiversity on ecosystem
statistical methods. PRIMER-E. Plymouth, UK. functioning a consensus of current knowledge. Ecological
Bonsdorff, E. 2006. Zoobenthic diversity-gradients in the Baltic Monographs 75:3–35.
Sea: continuous post-glacial succession in a stressed ecosys- Karlson, K., R. Rosenberg, and E. Bonsdorff. 2002. Temporal
tem. Journal of Experimental Marine Biology and Ecology and spatial large-scale effects of eutrophication and oxygen
330:383–391. deficiency on benthic fauna in Scandinavian and Baltic
Bonsdorff, E., A. O. Laine, I. Vuorinen, and A. Norkko. 2003. Waters—a review. Oceanography and Marine Biology: An
Zoobenthos in the outer archipelago waters (N. Baltic Sea)— Annual Review 40:427–489.
the importance of local conditions for spatial distribution Laliberté, E., and P. Legendre. 2010. A distance-based
patterns. Boreal Environmental Research 8:135–145. framework for measuring functional diversity from multiple
Bonsdorff, E., and T. H. Pearson. 1999. Variation in the traits. Ecology 91:299–305.
sublittoral macrozoobenthos of the Baltic Sea along envi- Leppäranta, M., and K. Myrberg. 2009. Physical oceanography
ronmental gradients: a functional-group approach. Austra- of the Baltic Sea. Praxis Publishing, Chichester, UK.
lian Journal of Ecology 24:312–326. Loreau, M. 2004. Does functional redundancy exist? Oikos
Boström, C., A. Törnroos, and E. Bonsdorff. 2010. Inverte- 104:606–611.
brate dispersal and habitat heterogeneity: Expression of Lotze, H. K., H. S. Lenihan, B. J. Bourque, R. H. Bradbury,
biological traits in a seagrass landscape. Journal of Exper- R. G. Cooke, M. C. Kay, S. M. Kidwell, M. X. Kirby, C. H.
imental Marine Biology and Ecology 390:106–117. Peterson, and J. B. C. Jackson. 2006. Depletion, degradation,
Bremner, J., S. I. Rogers, and C. L. J. Frid. 2003. Assessing and recovery potential of estuaries and coastal seas. Science
functional diversity in marine benthic ecosystems: a compar- 312:1806–1809.
ison of approaches. Marine Ecology Progress Series 254:11– Millennium Ecosystem Assessment. 2005. Ecosystems and
25. human well being: biodiversity synthesis. World Resources
Bremner, J., S. I. Rogers, and C. L. J. Frid. 2006. Methods for Institute, Washington, D.C., USA.
describing ecological functioning of marine benthic assem- Mouchet, M. A., S. Villéger, N. W. H. Mason, and D.
blages using biological trait analysis (BTA). Ecological Mouillot. 2010. Functional diversity measures: an overview
Indicators 6:609–622. of their redundancy and their ability to discriminate
Cardinale, B. J., D. S. Srivastava, J. E. Duffy, J. P. Wright, community assembly rules. Functional Ecology 24:867–876.
A. L. Downing, M. Sankaran, and C. Jouseau. 2006. Effects Naeem, S., and D. E. Bunker. 2009. TraitNet: furthering
of biodiversity on the functioning of trophic groups and biodiversity research through the curation, discovery, and
ecosystems. Nature 443:989–992. sharing of species trait data. Pages 281–289 in S. Naeem,
Charvet, S., B. Statzner, P. Usseglio-Polatera, and B. Dumont. D. E. Bunker, A. Hector, M. Loreau, and C. Perrings,
2000. Traits of benthic macroinvertebrates in semi-natural editors. Biodiversity, ecosystem functioning, and human
French streams: an initial application to biomonitoring in wellbeing. Oxford University Press, New York, New York,
Europe. Freshwater Biology 43:277–296. USA.
Chevenet, F., S. Soledec, and D. Chessel. 1994. A fuzzy coding Naeem, S., and S. Li. 1997. Biodiversity enhances ecosystem
approach for the analysis of long-term ecological data. reliability. Nature 390:507–509.
Freshwater Biology 43:277–296. Nohrén, E., L. Pihl, and H. Wennhage. 2009. Spatial patterns in
Conley, D. J., et al. 2011. Hypoxia is increasing in the coastal community structure of motile epibenthic fauna in coastal
zones of the Baltic Sea. Environmental Science and habitats along the Skagerrak—Baltic salinity gradient.
Technology 45:6777–6783. Estuarine and Coastal Shelf Science 84:1–10.
 Ecological Applications
2236 ANNA TÖRNROOS AND ERIK BONSDORFF
Vol. 22, No. 8

Norkko, J., D. Reed, K. Timmermann, A. Norkko, B. Townsend, C. R., and A. G. Hildrew. 1994. Species traits in
Gustafsson, E. Bonsdorff, C. Slomp, J. Carstensen, and D. relation to habitat templet for river systems. Freshwater
Conely. 2011. A welcome can of worms? Hypoxia mitigation Biology 31:265–275.
by an invasive species. Global Change Biology. http://dx.doi. Tyler, E. H. M., P. J. Somerfield, E. Vanden Bergeh, J.
org/10.1111/j. 1365-2486.2011.02513.x Bremner, E. Jackson, O. Langmead, D. Lourdes, M.
Perus, J., S. Bäck, H.-G. Lax, V. Westberg, P. Kauppila, and E. Palomares, and T. Webb. 2012. Extensive gaps and biases
Bonsdorff. 2004. Coastal marine zoobenthos as an ecological in our knowledge of a well-know fauna: implications for
quality element: a test of environmental typology and the
integrating biological traits into macroecology. Global
European Framework Directive. Coastline Reports 4:27–38.
Ecology and Biogeography. http://dx.doi.org/10.1111/j.
Petchy, O. L., and K. J. Gaston. 2006. Functional diversity:
back to basics and looking forward. Ecology Letters 9:741– 1466-8238.2011.00726.x.
758. Usseglio-Polatera, P., M. Bournaud, P. Richoux, and H.
Reiss, J., J. R. Birdle, J. M. Montoy, and G. Woodward. 2009. Tachet. 2000. Biological and ecological traits of benthic
Emerging horizons in biodiversity and ecosystem functioning freshwater macroinvertebrates: relationships and definition
research. Trends in Ecology and Evolution 24:505–514. of groups with similar traits. Freshwater Biology 43:175–205.
Rönnberg, C., and E. Bonsdorff. 2004. Baltic Sea eutrophica- Villnäs, A., and A. Norkko. 2011. Benthic diversity gradients
tion: area-specific ecological consequences. Hydrobiologia and shifting baselines: implications for assessing environ-
514:227–241. mental status. Ecological Applications 21:2172–2186.
Schleuter, D., M. Daufresne, F. Massol, and C. Argillier. 2010. Violle, C., M.-L. Navas, D. Vile, E. Kazakou, C. Fortunel, I.
A user’s guide to functional diversity indices. Ecological Hummel, and E. Garnier. 2007. Let the concept of trait be
Monographs 80:469–484. functional. Oikos 116:882–892.
Solan, M., D. G. Raffaelli, D. M. Paterson, P. C. L. White, and Walker, B. H. 1992. Biodiversity and ecological redundancy.
G. J. Pierce. 2006. Marine biodiversity and ecosystem
Conservation Biology 6:18–23.
function: empirical approaches and future challenges. Intro-
Webb, T. J., E. H. M. Tyler, and P. J. Somerfield. 2009. Life
duction. Marine Ecology Progress Series 311:175–178.
Symstad, A. J., F. S. Chapin, III, D. H. Wall, K. L. Gross, L. F. history mediates large scale population ecology in marine
Huenneke, G. G. Mittelback, D. P. C. Peters, and D. Tilman. benthic taxa. Marine Ecology Progress Series 396:293–306.
2003. Long-term and large-scale perspectives on the rela- Worm, B., E. B. Barbier, N. Beaumont, J. E. Duffy, C. Folke,
tionship between biodiversity and ecosystem functioning. B. S. Halpern, J. B. Jackson, H. K. Lotze, F. Micheli, S. R.
BioScience 53:89–98. Palumbi, E. Sala, K. A. Selkoe, J. J. Stachowicz, and R.
Tilman, D. 2000. Causes, consequences and ethics of biodiver- Watson. 2006. Impacts of biodiversity loss on ocean
sity. Nature 405:208–211. ecosystem services. Science 314:787–790.

SUPPLEMENTAL MATERIAL
Supplement
Data on species occurrence and trait information of northern Baltic Sea coastal benthos (Ecological Archives A022-120-S1).