Dircks 014967987 PDF

Investigation into the fermentation of
Australian cocoa beans and its effect on

microbiology, chemistry and flavour.
Hugh Douglas Dircks
A thesis submitted in fulfillment of the requirements

for the degree of Doctor of Philosophy
University of New South Wales

School of Chemical Sciences and Engineering
Sydney, Australia
2009
Declaration
ORIGINALITY STATEMENT:
'I hereby declare that this submission is my own work and to the best of my
knowledge it contains no materials previously published or written by another
person, or substantial proportions of material which have been accepted for the
award of any other degree or diploma at UNSW or any other educational
institution, except where due acknowledgment is made in the thesis. Any
contribution made to the research by others, with whom I have worked at
UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that
the intellectual content of this thesis is the product of my own work, except to
the extent that assistance from others in the project's design and conception or
in style, presentation and linguistic expression is acknowledged/
Signed
Hugh Dircks
iii
Acknowledgements
There are several important acknowledgements I wish to make at the outset of
this Thesis.
My deep and heartfelt thanks to my supervisor, Professor Graham Fleet. His

guidance and oversight of the project were invaluable. I always benefited from
the time he gave, the encouragement, advice and anecdotes. I thank him for all
his patience and the inspiration he provided, often at unexpected times.
Graham taught me food microbiology, and it was a joy and a privilege to be
able to have been part of his research group.
This study was possible by the generous funding of Cadbury-Schweppes. I am

deeply grateful for their vision and provision for the research. I also want to
thank all those at Cadbury Schweppes, past and present, who personally aided
and guided me: Barry Kitchen, John Aston, Ian Mitchell, Angela Findlay.
A special thanks and acknowledgement to Olivia Roman for her help in setting
up and running the sensory evaluation trials in Cadbury UK.
Acknowledgements to Cadbury-Schweppes in Tasmania and UK for testing of
the Australian chocolate samples.
Thanks to Dennis Nielsen and all at KVL, Copenhagen for the DGGE gels of
our Australian isolates, and advice about their interpretation. Your willingness
to collaborate with us was greatly appreciated, and I was personally
encouraged to have contact with another PhD student overseas.
My field work was performed at the Department of Primary Industries, South

Johnstone. I want to especially thank Donna Campagnolo, as well as Craig
Lemin, and Yan Diczbalis for providing me with a great place to work and lots
of information about cocoa trees. Thanks to Cameron and Luigi for help with
setting up the fermentation and drying equipment, and taking such an interest
in the, at times smelly, fermentations.
iv
Thanks are extended to the lab staff at UNSW: Eileen Emmerson, Camellio
Taraborelli, and Peter Valtchev, for their dependability and cheerfulness in
keeping our labs functioning.
To my fellow PhD Students: AiLin, Peter, Lidia, Sung Sook, Kim, Jeremy, and
Vic: Our friendships were one of the valuable things I experienced during this
project. When things were a hard slog, having you around buoyed me up. And
when we had our little victories, how much better it was to share them.
Thanks to my dear family and friends. You have cared for me and Georgina so
generously through this whole experience. There have been all sorts of
unexpected challenges and trials, and I have appreciated all the phone-calls,
hugs, lunches, exhortations, commiserations, and especially prayers. Thank you
for all of you who were patient.
Finally, and most of all, I want to thank the Fair Georgina, my beautiful wife.
Thank you for sharing with my in this, and in life.
V
Table of contents
DECLARATION i
ACKNOWLEDGEMENTS ii
TABLE OF CONTENTS v
ABSTRACT xiv
PUBLICATIONS AND PRESENTATIONS FROM THIS THESIS xvi
CHAPTER 1 INTRODUCTION 1
CHAPTER 2 LITERATURE REVIEW - A CRITICAL REVIEW

FOCUSED ON THE DEVELOPMENT OF IMPROVED PROCESS FOR 4
COCOA FERMENTATION
2.1 COCOA AND CHOCOLATE 4

2.1.1 Origins 4
2.1.2 Production of cocoa and chocolate 6
2.1.2.1 Cultivation 7
2.1.2.2 Harvest and pod splitting 11
2.1.2.3 Fermentation 12
2.1.2.4 Drying 16
2.1.2.5 Storage and transport 17
2.1.2.6 Roasting, winnowing, grinding 17
2.1.2.7 Pressing and alkalisation 18
2.1.2.8 Chocolate Processing 18
2.2 MARKET AND ECONOMICS OF COCOA AND CHOCOLATE 21
2.2.1 Basic economics of the cocoa market 21
2.2.1.1 Demand for cocoa beans 21
2.2.1.2 Supply and production of cocoa beans 22
2.2.2 Current market trends and opportunities 24
2.2.2.1 Ethically oriented cocoa 24
2.2.2.2 Changing perceptions of the health and nutrition of
chocolate and cocoa - Chocolate marketed for potential health 25
promoting value
2.2.2.4 Chocolate as a sophisticated gourmet food - flavour, dark
and single origin chocolate
2.2.3 Australian domestic chocolate market status 28
2.2.4 Forecast Summary for the Cocoa market 28
vi
2.3 DEVELOPMENT OF COCOA INDUSTRY IN AUSTRALIA 29

2.4 THE BIOCHEMISTRY AND MICROBIOLOGY OF COCOA
FERMENTATION - EFFECTS ON FLAVOUR AND QUALITY
2.4.1 Physical and chemical properties of unfermented cocoa beans 31
2.4.1.1 Physical structure of unfermented cocoa beans 31
2.4.1.2 Chemical composition of unfermented cocoa beans 33
2.4.1.3 Factors affecting the composition of unfermented cocoa ^
beans
2.4.2 Chemical and biochemical changes occurring during cocoa
bean fermentation
2.4.2.1 Metabolism of cocoa pulp by microorganisms and
diffusion of metabolites.
2.4.2.2 Seed death and cellular disruption 43
2.4.2.3 Enzymatic reactions 44
2.4.2.4 Non-enzymatic reactions and the role of drying 47
2.4.3 Transformation of precursors to final flavour compounds 48
2.5 MICROBIOLOGY OF COCOA FERMENTATION 50
2.5.1 Sources of microorganisms 50
2.5.2 Yeasts 51
2.5.2.1 Growth and diversity of yeasts during cocoa bean
fermentation
2.5.2.2 Metabolic activities of yeasts during cocoa bean ^
fermentation
2.5.3 Lactic acid bacteria 57
2.5.3.1 Growth and diversity of lactic acid bacteria during cocoa
58
bean fermentation
2.5.3.2 Metabolic activities of lactic acid bacteria during cocoa ^
bean fermentation
2.5.4 Acetic acid bacteria 63
2.5.4.1 Growth and diversity of acetic acid bacteria during cocoa ^
bean fermentation
2.5.4.2 Metabolic functions of acetic acid bacteria during cocoa ^
bean fermentation
2.5.5 Bacillus species 66
2.5.6 Other bacteria 67
2.5.7 Filamentous fungi 68
vii
2.6 THE IMPROVEMENT OF COCOA BEAN PROCESSING AND

QUALITY - PROGRESS AND OPPORTUNITIES
2.6.1 The industrialisation and improvement of cocoa bean ^
processing
2.6.1.1 Improved approaches to the substrate - selection, ^
cultivation harvesting and post-harvest storage of cocoa beans
2.6.1.2 Improved process design and control 72
2.6.1.3 Controlled inoculation for cocoa bean fermentation 74
2.6.2 Total supply chain management and integrated approaches to ^
the improvement of cocoa quality
2.6.3 Barriers to indutrialisation and improvement of cocoa ^
fermentation
2.7 CONCLUSIONS 78
CHAPTER 3 THE MICROBIOLOGICAL, CHEMICAL AND SENSORY 8Q

PROPERTIES OF AUSTRALIAN COCOA BEAN FERMENTATIONS
3.1 INTRODUCTION 80
3.2 MATERIALS AND METHODS 82

3.2.1 Fermentation of Cocoa Beans 82
3.2.1.1 Source of cocoa beans for fermentation 82
3.2.1.2 Fermentation of cocoa beans in heap, box and barrel 86
3.2.1.3 Addition of microbial nutrients to enhance fermentation 87
3.2.1.4 Effect of mixing on the fermentation of Australian cocoa 88
3.2.1.5 Sampling of fermenting cocoa beans for microbiological ^
and chemical testing.
3.2.1.6 Drying of fermented cocoa beans 88
3.2.1.7 Summary of fermentation experiments 90
3.2.2 Physical Measurements of Fermenting Cocoa beans 92
3.2.2.1 pH of cocoa beans 92
3.2.2.2 Pulp and seed weights 92
3.2.2.3 Moisture content of beans, pulp and seed 92
3.2.3 Microbiological analyses 92
3.2.3.1 Isolation and enumeration of microorganisms in cocoa ^
fermentations
3.2.3.2 Identification of yeast species 98
viii
3.2.3.3 Identification of bacterial species 98

3.2.3.4 DNA extraction for identification by PCR-sequencing 98
3.2.3.5 PCR amplification of DNA for sequence identification 99
3.2.3.6 DNA sequencing 100
3.2.3.7 Denaturing gradient gel electrophoresis (DGGE) analysis
100
of the microbial ecology of cocoa fermentations
3.2.4 Chemical analyses 101
3.2.4.1 Sample preparation 101
3.2.4.2 Carbohydrates 101
3.2.4.3 Organic acids 102
3.2.4.4 Ethanol 102
3.2.5 Quality and sensory assessment of cocoa beans 103
3.2.5.1 Quality assessment 103
3.2.5.2 Chocolate making 104
3.2.5.3 Sensory evaluation of chocolate samples by untrained 104
panelists
3.3 RESULTS 107
3.3.1 Fermentation of cocoa beans in heap, box and barrel 107
3.3.1.1 General fermentation profiles 107
3.3.1.2 Changes in the populations of yeast species 110
3.3.1.3 Changes in the populations of lactic acid bacteria species 112
3.3.1.3 Changes in the populations of acetic acid bacteria and 114
other bacterial species
3.3.1.4 Changes in the concentrations of sugars during heap, box
117
and barrel fermentation
3.3.1.5 Changes in the concentrations of ethanol during heap, box
118
and barrel fermentation
3.3.1.6 Changes in the concentrations of organic acids during
120
heap, box and barrel fermentation
3.3.1.7 Quality evaluation of cocoa beans produced by heap, box
124
and barrel fermentations
3.3.1.8 Sensory evaluation of chocolate made from cocoa beans
127
fermented by heap, box and barrel
3.3.2 Addition of microbial nutrients to enhance fermentation 130
3.3.2.2 Changes in the population of yeast species during
134
fermentation with added microbial nutrients
ix
3.3.2.3 Changes in the population of bacterial species during ^

3.3.2.4 Changes in the concentrations of sugars during
3.3.2.5 Changes in the concentrations of ethanol during ^
fermentation with added microbial nutrients.
3.3.2.7 Quality evaluation of beans fermented with different ^
amounts of added microbial nutrients (Fermaid K®).
3.3.2.8 Sensory evaluation of chocolate made from cocoa beans
fermented using added microbial nutrients
3.3.3 Effects of mixing frequency. 153
3.3.3.2 Growth of yeast species. 155
3.3.3.3 Growth of lactic acid bacterial species. 156
3.3.3.4 Growth of acetic acid bacterial species. 157
3.3.3.5 Changes in the concentrations of sugars 158
3.3.3.6 Changes in the concentration of ethanol 159
3.3.3.7 Changes in the concentration of organic acids 159
3.3.3.8 Quality evaluation of South Johnstone cocoa beans ^
fermented in boxes and mixed at different frequencies.
3.3.3.9 Sensory evaluation of chocolate made from cocoa beans ^
mixed at different frequencies.
3.4 DISCUSSION 166
3.4.1 Physical changes during fermentations 166
3.4.1.1 Temperature 166
3.4.2.2 pH 167
3.4.2.3 Appearance, water content and pulpiseed ratio 168
3.4.3 Microbiological changes during fermentations 169
3.4.3.1 Yeast species 170
3.4.3.2 Lactic acid bacteria 173
3.4.3.3 Acetic acid bacteria 175
3.4.3.4 Other species of bacteria and filamentous fungi 178
3.4.4 Chemical and physiological changes during fermentation 179
3.4.4.1 Sugars 180
3.4.4.2 Ethanol 181
3.4.4.3 Organic acids 182
X
3.4.4.4 Glycerol and mannitol 185

3.4.4.5 Physiological changes in cocoa beans during fermentation. 186
3.4.5 Effects of fermentation on sensory and quality outcomes 186
3.4.5.1 Overall acceptability of Australian cocoa beans 186
3.4.5.2 Effects of fermentation variables on the quality and
sensory properties of cocoa beans grown in Queensland, 188
Australia.
3.4.6 Implications for future cocoa fermentation in Australia 191
3.4.6.1 Baseline established for future reference 191
3.4.6.2 Methods used to examine the microbiology of cocoa ^
fermentations
3.4.6.3 Fermentation Method 193
3.4.6.4 Drying method 195
3.5 CONCLUSIONS 196
CHAPTER 4 EVALUATION OF THE USE OF MIXED YEAST STARTER

CULTURES IN THE AUSTRALIAN COCOA BEANS: EFFECTS ON 199
MICROBIAL ECOLOGY, CHEMISTRY, QUALITY AND FLAVOUR
4.1 INTRODUCTION 199

4.2 MATERIALS AND METHODS 201
4.2.1 Fermentation of cocoa beans using starter cultures 201
4.2.1.1 Laboratory fermentations 201
4.2.1.2 Pilot-scale fermentations 205
4.2.2 Preparation of starter cultures 206
4.2.2.1 Selection and screening of cultures for use as starter
cultures.
4.2.2.2 Preparation of cultures for laboratory fermentations 206
4.2.2.3 Preparation of cultures for pilot-scale fermentations 207
4.2.3 Physical Analyses 208
4.2.3.1 Measurement of pH of pulp and seed 208
4.2.3.2 Pulp and seed weights 209
4.2.3.3 Moisture content of beans, pulp and seed 209
4.2.4 Microbiological analyses 209
4.2.5 Chemical analyses 209
4.2.6 Quality and sensory assessment of cocoa beans 209
xi
4.2.6.1 Quality assessment of dried fermented cocoa beans 209

4.2.6.2 Chocolate Preparations 210
4.2.6.3 Quality evaluation of chocolate samples from pilot ^
fermentations inoculated with defined mixtures of yeast.
4.2.6.4 Sensory evaluation of chocolate samples by untrained ^j j
panelists
4.2.6.5 Quantitative descriptive analysis of chocolate samples 211
4.2.7 SDS PAGE analysis of protein degradation during cocoa bean ^\3
fermentation
4.2.8 Analysis of antioxidant activity of cocoa polyphenols 213
4.2.8.1 Extraction of polyphenols from cocoa beans. 213
4.2.8.2 Analysis of polyphenolic compounds in cocoa extracts 213
4.2.8.3 Determination of total antioxidant capacity (AOC) of ^
cocoa extracts
4.2.9 Gas chromatographic analysis of cocoa aroma volatiles 215
4.2.9.1 Extraction of volatile compounds from roasted cocoa ^j5
beans
4.2.9.2 Gas chromatography-flame ionisation analysis 216
4.2.9.3 Gas Chromatography-mass spectroscopy analysis 216
4.2.10 Summary of experimental plan 217
4.3 RESULTS 218
4.3.2 Laboratory-scale fermentation of cocoa beans by inoculation
Zlo
with defined mixtures of yeast
4.3.2.2 Growth of yeast species during laboratory fermentations 223
4.3.2.3 Growth of bacterial species during laboratory 225
fermentations
4.3.2.4 Changes in the concentration of sugars and ethanol during ^9
laboratory fermentations
4.3.2.4 Changes in the concentration of organic acids during 23 j
laboratory fermentations
4.3.2.4 Quality evaluation of cocoa beans produced by laboratory 23^
fermentations
4.3.2.5 Sensory evaluation of chocolate prepared from laboratory 23 g
fermented cocoa beans
4.3.3 Pilot-scale fermentation of cocoa beans in box and barrel ^,,
241
vessels using defined mixtures of yeast cultures.
xii
43.3.2 Changes in the population of yeast species during ^9

fermentation using defined mixtures of yeast cultures.
4.3.33 Changes in the population of bacterial species during ^54
4.33.4 Changes in the concentrations of sugars during 259
4.33.5 Changes in the concentration of ethanol during 2^4
fermentations using defined mixtures of yeast cultures.
4.33.6 Changes in the concentration of organic acids during
4.3.3.7 Quality evaluation of cocoa beans produced by 277
fermentation using mixed yeast cultures.
4.33.8 Sensory evaluation of chocolate made from cocoa beans „
283
fermented using defined mixtures of yeast cultures.
4.3.4 Quantitative descriptive analysis of cocoa beans fermented
with defined mixtures of yeast cultures.
43.4.1 Appearance of chocolate samples 290
43.4.2 Mouthfeel of chocolate samples 291
4.3.43 Odour characteristics of chocolate samples 291
43.4.4 Flavour characteristics of the chocolate samples 293
43.4.5 Aftertaste characteristics of the chocolate samples 295
43.4.6 Overall characterization of the chocolate samples tested 297
4.3.5 Analysis of breakdown of cocoa storage proteins using SDS- 293
PAGE
4.3.6 Antioxidant activity of cocoa beans fermented with defined ^qq
mixed yeast cultures.
4.3.7 Analysis of volatile compounds from cocoa beans samples ^93
using gas chromatography.
4.4 DISCUSSION 310
4.4.1 Selection of yeast isolates for use in mixed starter cultures 310
4.4.2 Controlled laboratory fermentations to evaluate the use of ^
mixed cultures of yeast.
4.4.2.1 Changes in the temperature of bean mass 312
4.4.2.2 Changes in the pH of cocoa pulp 313
4.4.23 Changes in the populations of yeast and bacterial species 313
4.4.2.4 Changes in the chemistry of fermenting beans 315
4.4.2.5 The quality and sensory characteristics of cocoa beans 316
4.4.2.6 Implications of the laboratory fermentations 318
4.4.3 Pilot-scale fermentations 320
xiii
4.4.3.1 Changes in temperature 321

4.4.3.2 Changes in the pH of the cocoa pulp and seeds. 321
4.4.3.3 Changes in the appearance and aroma of the fermenting
cocoa beans
4.4.3.4 Changes in the populations of yeast and bacterial species 323
4.4.3.5 Changes in the chemistry of fermenting beans 323
4.4.3.6 Quality of beans produced by pilot-scale fermentation, and
the chocolate produced from these beans
4.4.3.7 Sensory evaluation of cocoa beans produced by pilot-scale ^5
fermentations
4.4.3.8 Degradation of cocoa seed proteins during fermentation 328
4.4.3.9 Changes to polyphenols and antioxidant capacity of beans „
32o
during fermentation
4.4.3.10 GC-MS analysis of volatiles extracted from cocoa beans 340
4.4.3.11 Summary and future direction 342
4.5 CONCLUSIONS 343
CHAPTER 5 CONCLUSIONS 346

5.1 KEY CONCLUSIONS AS THEY RELATE TO PROJECT M6
OBJECTIVES AND GAPS IN KNOWLEDGE
5.2 NOVEL ASPECTS 350
5.3 LIMITATIONS 352
5.4 FUTURE WORK 353
CHAPTER 6 REFERENCES 355
Appendix A - Supplementary microbiological data 395

Appendix B - Supplementary chemistry data 402
xiv
Abstract
Cocoa beans are the fermented and dried seeds of the tree Theobroma cacao,
and are the fundamental ingredient in chocolate manufacture. Fermentation
and drying are critical processes during which the beans develop distinctive
chocolate flavour. Worldwide, fermentation is conducted as an uncontrolled
traditional process, giving much variation in bean quality. There is potential to
develop cocoa production in Northern Queensland, Australia using controlled
industrial fermentations.
This thesis investigated the fermentation of cocoa beans grown in Queensland.

To obtain baseline microbiological and chemical information, initial
fermentations were conducted as traditional heap and box methods, in parallel
with a novel process in a barrel. Using cultural and molecular methods, the
microbial ecology of the fermenting cocoa beans was examined along with
changes in pH, temperature and the composition of sugars, ethanol and organic
acids. Consistent with findings in other countries, Australian cocoa bean
fermentations gave a complex microbial ecology. The most frequently found
yeasts were Hanseniaspora guilliermondii, Issatchenkia orientalis, and Saccharomyces
cerevisiae, while the predominant bacteria were Lactobacillus plantarum,
Lactobacillus fermentum and Acetobacter pasteurianus. Unfermented Australian
cocoa beans had similar chemical composition to cocoa beans cultivated in
other countries. During fermentation, physical (pH, temperature) and chemical
(sugars, ethanol, organic acids) changes occurred, consistent with those
observed by other researchers. Sensory evaluations showed that the Australian
cocoa beans gave chocolate of comparable quality and flavour to that made
from industry-standard Ghanaian beans. With optimization of mixing
frequency, the barrel vessel was found to be a more suitable vessel in which to
conduct fermentation
Isolates of H.guillermondii, I.orientalis, and S.cerevisiae were used as starter
cultures in controlled laboratory scale (5 kg beans) and pilot-scale (50 kg beans;
box and barrel vessels) fermentations. While this is not the first study to
inoculate fermentations with mixtures of yeast, it is the first that inoculated
mixed cultures of yeast into non-sterile, commercial scale fermentations.
Fermentations were monitored for microbiological and chemical changes, and

the final dried beans were measured for a range of commercial quality criteria
(fermentation index, moisture, fat and pH). The dried beans were made into
chocolate and subject to statistically based sensory evaluation.
The inoculated fermentations produced beans of acceptable quality and flavour.

This research clearly demonstrated that by using different mixtures of yeasts
distinctly different flavour characteristics can be produced and the fermentation
can be accelerated. Limited data was also obtained suggesting that the use of
different starter cultures can influence the level of procyanidins in the finished
cocoa beans.
This research is the first of its kind performed in Australia. In addition to

providing fundamental microbiological and chemical data, basic processing
parameters were established. This work forms the foundation for any future
industrialised approach to the fermentation of Australian cocoa, using starter
cultures and new fermenter designs.

xvi
Publications and presentations from this thesis

Dircks, H.D and Fleet, G.H. 2008. 'Enhanced bioprocessing of cocoa by use of
starter cultures' Poster presented: 2008 IUFOST Convention, Shanghai, China.
Dircks, H.D. 2008. 'Industrialisation of Australian cocoa fermentation.'

Presented: 2008 IUFOST Convention, Shanghai, China. October 22, 2008.
Fleet, G.H. and Dircks, H.D. (2007) Yeast, cocoa beans and chocolate. Microbiol.
Austr. 48, 48-50.
Dircks, H.D. (2008) Towards industrialised cocoa fermentation in Queensland,

Australia. The Malcolm Bird Award presentation at the 41st Annual Convention
of the Australian Institute of Food Science and Technology, Sydney, 21-24 July.
Dircks, H.D. (2006) The microbial ecology of Australian cocoa bean

fermentation. Oral presentation at the Joint 12th Australian Food Microbiology
Conference and International Risk Assessment Conference on Microbial Risk
Assessment: Food-borne Hazards Sydney, Australia. February 20.
Dircks, H.D. (2004) The Mycology of Australian Cocoa fermentations. Oral

presentation at the Joint Australian Society for Microbiology / Australasian
Mycological Society National Conference, Sydney, 27-28 September.
1
Chapter 1 - Introduction
Cocoa beans are the raw material for chocolate manufacture. Chocolate is a
major confectionary product, and in 2005 the chocolate industry was estimated
to be worth over US$ 44.1 billion (ICCO, 2007).
Cocoa beans are the seeds of the tree Theobroma cacao L. and have been
cultivated for thousands of years. One of the earliest groups to produce cocoa
were the Aztecs, who not only made a drink called chocolatl from the beans, but
also used them as a form of currency. In the late 1500's, as Spanish
conquistadors occupied the Americas, cocoa was introduced to Europe where
its popularity spread quickly (Coe and Coe, 2003). Once established, a number
of innovations were made in cocoa processing, leading to the development of
new products. Several of the world's major chocolate manufacturers - Cadbury,
Nestle, and Lindt - were founded in the 19th century (Beckett, 2000).
While most of the world's chocolate is produced in Europe and the USA, cocoa
beans grow best in the equatorial tropics. Measured by volume of exports, the
main cocoa producing countries are the Ivory Coast and Ghana, with West
Africa accounting for 71% of global production in 2005 (ICCO, 2007).
Chocolate manufacturers require a constant supply of cocoa beans that must
conform to an array of quality criteria. Over the last decade, the cocoa market
has seen an increase in demand, and renewed threats to supply such as political
instability, climate change, and crop loss due to disease. Consequently, many
chocolate companies are interested in diversifying their sources of cocoa bean
raw materials. In this context, it was found that the northern regions of
Australia present climactic and terrain conditions that are favorable for the
cultivation of cocoa trees (Lemin, 2005a). Experimental plantations were
established at two locations in Northern Queensland, Mossman and South
Johnstone.
Before use in chocolate manufacture, cocoa beans must be fermented and dried,
usually close to where they are grown. First, ripe pods are cut from the trees,
broken open, and the beans and associated pulp are removed for fermentation.
The beans, in amounts ranging from 25 kg to 3,000 kg are placed in heaps on the
ground, in wooden boxes (generally l-2m3) or in wooden trays for
fermentation. Fermentation generally lasts for 4-6 days, during which time the
2
pulp liquifies and drains away, and the beans darken in appearance. After
fermentation the beans are dried, usually in the sun, and then sold to chocolate
manufacturing companies. Because most cocoa tree plantations are located in
developing countries, the fermentation continues to be a traditional, village
scale process that frequently lacks any control (Wood and Lass, 1985).
Fermentation is essentially a microbiological process triggering physical,
chemical and biochemical changes within the beans to generate the precursors
that give characteristic chocolate flavour and colour upon roasting (Afoakwa et
al., 2008). It is well established that a succession of yeast and bacterial species
grow throughout the 4-6 day process. The microorganisms responsible for the
fermentation originate from the surface of the pods, as well as the fermentary
environment. Globally, there is some consistency in the species of yeast and
bacteria found to predominate the fermentation. Nevertheless, there is also
significant diversity and variation in the microbial ecology of fermentation and
this can impact on bean and chocolate flavour (Schwan and Wheals, 2004).
Industrialisation of the cocoa fermentation process may allow greater control
over the quality of cocoa beans, and the chocolate derived from them (Schwan,
1998). The concept of industrialisation of traditional fermentation processes to
enhance their performance and efficiency is not new. For example, wine, beer,
cheese, and yoghurt, were at one time all made using traditional processes
(Wood, 1998). Now, these fermentations have been developed into highly
efficient, well controlled processes, in modern fermenter designs often using
defined starter cultures (Steinkraus, 2004).
The potential to cultivate and produce cocoa beans in Queensland, Australia,
brings with it the possibility and challenge to develop modem, industrial
processes for conducting bean fermentation and drying. In order to achieve this
goal, much basic information is required about the microbiology and chemistry
of cocoa beans grown and fermented in the local region. And it is also necessary
to properly link this knowledge to the quality of the cocoa beans and chocolate
made from them.
The overall objective of this thesis was to obtain foundational knowledge for
developing a well controlled process for fermenting cocoa beans cultivated in
Northern Queensland. The specific goals of this thesis were therefore:
3
To gain baseline data on the microbiology, biochemistry and physical

properties of Australian cocoa bean fermentation.
To determine whether Australian fermented cocoa can be processed
into chocolate of acceptable, commercial quality.
Investigate the effects of key processing variables on the fermentation
process, with a focus on those variables important in development of
an industrialised process for cocoa bean fermentation: fermentation
vessel, addition of microbial nutrients, effect of mixing and aeration
and the length of fermentation.
Investigate the effects of defined starter cultures on cocoa bean fer
mentation: microbiology, biochemistry and physical properties.
Assess the effects of novel fermentation methods on duration of fer
mentation, protein degradation, cocoa bean quality, antioxidant lev
els, production of aroma volatiles and final chocolate flavour.
4
Chapter Two
Literature Review - A critical review focused on

the on the development of improved process for
cocoa fermentation.
2.1 Cocoa and Chocolate
2.1.1 Origins
Cocoa beans, the essential ingredient of chocolate, begin as seeds of the tree
Theobroma cacao L. (Family Sterculiaceae). This plant species is thought to have
originated in central and South America. Today, three broad cultivars of cocoa
are commonly recognised: Forastero, Criollo and Trinitario. The cultivars
exhibit differences in the appearance of pods, yields of beans, and in resistance
to pests and disease (Wood and Lass, 1985; Biehl et al., 1989; Aroyeun et al.
2006, de Almeida et al., 2007). Genetic studies of native cocoa populations
across Central and South America show that the Forastero and Criollo cultivars
arose prior to European settlement. Comparison of DNA micro-satellite
markers also suggest that all 'native' cocoa populations originated from the
Amazon basin (Motomayor et al., 2002, 2003). Cultivar diversity is thought to
have resulted from the sporadic human transfer and cultivation of small
populations of plants in isolation from one another, as well as natural
hybridisation between existing varieties (Motomayor et al., 2002, 2003). Further
hybrids and new populations have arisen more recently, as cocoa was spread by
colonial activities of the Spanish, Portuguese, Dutch, British, French and
Germans (Coe and Coe, 2003). Today, populations of cocoa trees exist across the
Americas, the Caribbean, West Africa, the South Pacific and Asia. Due to the
conditions required for their cultivation their distribution is usually limited to
20° north or south of the Equator (Rohan, 1963b; Wood and Lass, 1985; Lemin,
2005b). Figure 2.1 shows a proposed model for the diversification and spread of
cocoa within South America, and internationally.
5
Summary of cocoa domestication

Forastero: Trinitario
(Tnnidad type)
• 1635: Cocoa ICrioIlol
introduced to the West
across northern JlVjfiS by the Spanish
South America
• Unspecified natural
disaster devastates the
cocoa plantations
• New cocoa material
(Forastero) introduced
to replace damaged
Criollo: trees
• Domesticated by • Natural hltaVJiWMW
I Olmec (approx occurred with
rioooBc) remaining Criollo •
• Important to hybrid called
Mayan culture "T nnitario"
’until 900 AD • Trinitario considered
• Grown bv Aztecs productive. Samples
1200 to 1500 AD reintroduced to
mainland lEcuadorl
Spread of cocoa to other countries
1600a ■ Spanish,
ritish, Dutch and
French all
ultivata cocoa in
the West-Indes
1614 Spanish
introduce cocoa
o the Philippines
Expansion of cocoa
production in South
America by Spaniah
colonists:
• 1600's ■ Spanish
colonists expand existing
Native-grown cocoa
plantations in Venezuela lSSO-1930's:
region British introduce
• 1635 - large cocoa to
plantations established Madagascar and
in Mexico East Africa
• 1640's - cocoa
cultivated in Northern
Brazil
• 1750 - plantations in
Bahia region ol Brazil
Figure 2.1 - Origins, domestication and spread of cocoa cultivation: Adapted from
information of Motomayor et al. (2002, 2003) and Coe and Coe (2003).
In the beginning, the main consumers of cocoa were also its main producers.
Records suggest that several Meso-American cultures all cultivated and
consumed cocoa. This included the Olmec (1200BC-400BC), Maya
(250AD-900AD) and the Aztec (1400AD-1500AD) cultures. In all of these
cultures, cocoa was taken as a drink with various additions of chilli, vanilla,
honey and anatto. In the Mayan culture, the cocoa tree played a part in their
creation myths, had a patron deity "Ek Chuah," and was used as a form of
currency. As South and Central America were colonised by the Spaniards in the
1500's, cocoa beans and chocolate were introduced to Europe. The first
appearance of cocoa in Europe was in 1544 when the Kekchi Maya presented
beans to Prince Philip of Spain (Coe and Coe, 2003; Rosenblum, 2005). The first
official shipment of beans from Mexico to Seville followed 41 years later. From
6
Spain, chocolate was spread through the nobility to France, and to a limited
extent, the rest of Europe. From Aztec times until the 19th Century, the
consumption of cocoa was in liquid form, and mostly by the wealthy and elite
of society. During the 19th century, social changes and technological
innovations enabled the average citizen to experience chocolate, and to eat it as
a solid food. Comprehensive reviews of the history of cocoa bean and chocolate
production are given in Chatt (1953), Minifie (1980), Wood and Lass (1985),
Beckett (2000) and Thompson et al. (2007). The historical, social and
gastronomic significance of chocolate and cocoa are well covered by Coe and
Coe (2003), Guittard (2005), Rosenblum (2005) and Roussel-Doutre (2005). The
last two centuries have seen an industrialisation of chocolate production as
illustrated by the following timeline:
1828 - Holland: Van Houten invents the cocoa press producing cocoa butter and
powder.
1866 - Frys (UK) produce the first solid chocolate bar.
1875 - Daniel Peter and Henri Nestle add milk powder to chocolate, and begin mass
producing so called ‘milk chocolate.’
1880 - Rudolphe Lindt invents conching, enhancing flavour and texture of chocolate.
1900 - Milton Hershey - first American to industrially produce chocolate.
1905 - Cadbury releases their first Dairy Milk Bar.
1941 - To prevent melting, the Mars company develops sugar coated chocolate, which it
supplies to American troops during the Second World Wars (M & M’s).
1986 - Valrhona introduces the single origin chocolate bar.
1990 to present - numerous scientific claims regarding the potential health benefits of
chocolate; greater attention paid by international community to the social and environmental
impact of cocoa farming; explosion in the diversity of chocolate products available worldwide.
From: Beckett, 2000; Bright, 2001a; Coe & Coe, 2003; Nestle, 2006; Cadbury -
Schweppes, 2006).
2.1.2 Production of cocoa and chocolate
Raw or poorly processed cocoa beans are of little value to the chocolate
industry as they lack characteristic chocolate flavour and aroma. In order to
obtain the desired flavour, raw cocoa beans must be fermented, dried, and
7
roasted. Additional processing is then required to obtain a commercial product.

Many of these operations have been widely reviewed in the literature (Chatt,
1953; Minifie, 1980; Baker et al., 1994; Lass; 1999; Beckett, 2000; Schwan and
Wheals, 2004; ICCO, 2007c; Thompson et al., 2007). The key unit operations in
cocoa and chocolate production are outlined in this section. A flow-chart of
cocoa bean and chocolate manufacture is given in Figure 2.2. Sequentially, the
main steps are: cultivation of the cocoa trees, harvesting and splitting of the
pods, fermentation and drying of the cocoa beans, roasting, winnowing and
grinding of the cocoa beans to cocoa liquor, and processing of cocoa liquor to
chocolate products.
FACTORY
? ?? ? ROAST&
WINNOW
GRINDING
PLANTATION DRYING
COCOA
PROCESSING LIQUOR
PRESSING
SUGAR
HARVEST COCOA
CHOCOLATE BUTTER
AND SPLIT FERMENT AT ION
Figure 2.2 - Flowchart of cocoa bean and chocolate processing (Rohan, 1963b; Beckett, 2000;
Thompson et al., 2007).
2.1.2.1 Cultivation
Production of cocoa beans ultimately relies on the successful cultivation of

cocoa trees (Theobroma cacao L.). The conditions required for cultivation of
cocoa trees are summarised in Table 2.1. Many of these requirements reflect the
cocoa tree's origin from the lower storey of the Amazonian rainforest.
Protection from strong sun and wind ensures that cocoa producers inter-plant
their cocoa with other trees. In some cases these will be dedicated shade trees
such as Gliricidia sepium, or cocoa may be planted amongst other commodity
8
trees such as banana, plantain, rubber, or oil palms. In other instances, the cocoa
is simply planted on the floor of existing forest, without prior clear-felling
taking place (Rohan, 1963b; Minifie, 1980; Wood and Lass, 1985; Johns, 1999;
Bright, 2001).
Table 2.1 - Climatic and soil conditions required for cultivation of cocoa trees
Temperature Minimum range: 18-21°C
Maximum average range: 30-35°C
Rainfall 1500 - 2500 mm annually.
Irrigation may be needed if 2 or more consecutive months with less than
50mm
Humidity 100% during night time
70-80% during day time
Altitude Generally below 300m.
May be up to 1300m in equatorial regions.
Shade Young seedlings require 50% shading.
Over-shading may reduce yield in mature plants.
Wind Cocoa is sensitive to constant prevailing winds. Windbreaks and inter-planting
with other trees must be established to prevent defoliation.
Soil Cocoa can grow in a diverse range of soils, however must be: Well drained;
topsoil depth >1.5m; C:N ratio >9; No more than 50% sand, 20% silt or 40%
clay. May grow in poorer soils with aid of fertilisers.
Information from Alvim (1977), Minifie (1980) and Wood & Lass (1985).
Ordinarily, cocoa trees take a minimum of three years to produce their first crop
from the time of planting. Maximum crop yields are usually achieved 3-5 years
later, and most cocoa trees will produce commercially acceptable yields until
25-30 years old (Chatt, 1953; Wood and Lass, 1985). Mature cocoa trees flower
and subsequently bear fruit from their trunk and larger branches in habit called
cauliflory. A certain proportion of the fruit do not grow to maturity, but die off
and blacken in a natural process called cherelle wilt. One tree can produce
anywhere from 0.5 kg to 10.0 kg of dried beans, depending on factors such as
genetic variation, rainfall, shade, prevalence of pests and disease, and the
agricultural practices of the farmer (Rohan, 1963b; Wood and Lass, 1985). Like
many other tropical plants, cocoa is cropped over several months. While there
are some variations between countries and varieties, most cocoa producers will
have one large crop about 5 months after the wettest time of year. In some
countries, a second, smaller crop may follow 5 months later (Bridgland, 1953;
Wood and Lass, 1985; Lass, 1999; UNCTAD, 2006).
9
The fruit of the cocoa tree are referred to as "pods," have a thick husk and
contain 30-40 cocoa beans. Fresh cocoa beans consist of several distinct parts.
The seed itself is made up of two cotyledons and an embryo (radicle), which are
in turn encased by a seed coat or testa. Depending on the cultivar, the
cotyledons have a deep to faint purple colour, while the radicle usually appears
white. Attached to the testa is a thick, white mucilaginous endocarp, commonly
termed 'pulp.' (Wood and Lass, 1985; Beckett, 1999; Schwan and Wheals, 2004;
Thompson et al., 2007). Detailed information about the histology of cocoa beans
may be found in the early work of Roelofson (1958), Biehl et al. (1977) and more
recently in Dangou et al. (2002). Figures 2.3, 2.4 and 2.5 show cocoa trees, pods
and beans.
Figure 2.3 - Photo of cocoa tree (Theobrotm cacao L.) growing in northern Queensland, Australia.
(Author's photo)
10
Figure 2.4 - Close-up of cocoa pods. Also note the wilted cherelles (undeveloped cocoa pods).
(Author's photo).
wwm
Embryo
(radicle)
“Seed
Cotyledon fmatcrjar
(mb)
lest a
’(seed coat)
Adhering fruit
(Pulp)
Figure 2.5 - (a) Cocoa pods and cocoa beans, (b) cocoa beans detail, (c) cocoa beans diagram.
(Author's photos and drawing).
11
Until the mid 1980's, nearly all cocoa farmers were employed agricultural
approaches developed in the 19th century (Wood and Lass, 1985). While there
have been improvements in the use of pesticides, fertilisers and phyto-sanitary
principles, many cocoa farmers still do not have access to the knowledge or
capital to improve their methods, inadequate agricultural management
continues to cause high losses of yield due to pests, disease and aging trees
(Lemin, 2005b). In light of problems at the primary production stage, it is
unsurprising that much cocoa research has been agronomic and horticultural in
nature (Montagnon, 2006). Examples of such research include: extensive genetic
studies and breeding programs to develop disease and pest resistant varieties
(Efombagn et al., 2005; Figueira and Alemanno, 2005; Schnell et al., 2005; Paulin
et al. 2008); assessment of farming practices and crop protection (Adejumo,
2005; Olujide and Adeogun, 2006); evaluating improved methods for pest
control (Aikpokpodion et al, 2003; Bateman, 2006). Because Forastero and
Trinitario cultivars give higher yields, and are more resistant to disease, they
currently contribute up to 95% of the world's crop, with Criollo beans making
up the remainder (Hoskin, 1994; Lass, 1999; Schwan and Wheals, 2004).
2.1.2.2 Harvest and pod splitting
Since cocoa pods are indehiscent, and do not abscise naturally, they must be
manually removed from the trees. A range of harvesting practices is employed
in different countries, but usually pods are removed using a machete, hooked
knives or secateurs. Although the possibility of mechanised pod harvesting has
been explored, it is made extremely difficult by the strong attachment of the
pods to the tree, and the irregular fruiting on trunk and branches (Wood and
Lass, 1985; Lemin, 2005b). It is important to harvest only ripe pods (Rohan,
1963b; Lass, 1999), although Ardhana and Fleet (2003) found that a certain
proportion of under-ripe beans did not adversely affect fermentation. After
harvesting, the pods are split, either by striking with a blunt object like a club or
stone, or by cutting with a machete. The beans and the attached pulp are
manually removed together and subsequently used for fermentation. Pod
splitting usually occurs at the location where the fermentation will be
conducted. Sometimes this in the plantation, or the harvested pods may be
transported to a central location before opening (Rohan, 1963b; Carr et al., 1979,
12
1980; Baker et al., 1994; Ardhana and Fleet, 2003; Nielsen et al., 2007b). In some
places, including Indonesia and Brazil, the pods are split in the field and the
freshly obtained beans are transported to a central fermentery. In this scenario,
delays in transport must be minimised since microbial growth in the bean pulp
begins immediately after pod splitting (Rohan, 1963b; Ardhana, 2003; Schwan
and Wheals, 2004). Surveys by Rohan (1963b), Carr et al. (1979,1980), Tomlins
et al.(1993) and Baker et al.(1994) found that many farmers will store harvested,
unopened, pods for a few days to up to two weeks prior to splitting. In some
cases this was to allow a small producer time enough to gather sufficient pods
for fermentation over several weeks of harvest. A more commonly cited reason
for pod storage was that it improved initiation of fermentation, and gave better
quality beans (Biehl et al., 1989; Tomlins et al., 1993). The same surveys also
revealed significant variation in individual practices between, and often within,
different countries (Rohan, 1963b; Carr et al., 1979, 1980; Tomlins et al., 1993;
Baker et al., 1994). Research has confirmed that fermentation may be affected by
ripeness (Wood and Lass, 1985; Ardhana and Fleet, 2003; Schwan and Wheals,
2004) and pod storage (Carr et al., 1979; Biehl et al., 1989). Therefore, a lack of
control at the harvest, storage and splitting stages may be clearly seen to
contribute to the variable quality of cocoa beans produced worldwide.
2.1.2.3 Fermentation
Fermentation is of vital importance in preparing cocoa beans for market, since

unfermented beans are highly bitter, astringent and lacking in full chocolate
flavour and aroma. In essence, cocoa bean fermentation encompasses all those
chemical and microbial changes to the bean pulp and cotyledons, whereby
precursors of chocolate flavor and colour are generated. These changes are
brought about by the biochemical activities of microorganisms that grow during
the process, as well as endogenous biochemical reactions within the cocoa
seeds. The microbiology and biochemistry of cocoa fermentation have been
extensively studied and will be discussed in detail in subsequent sections.
General reviews of the fermentation as a process have been made by Rohan
(1963), Carr (1982), Wood and Lass (1985), Lopez and Dimick (1995), Schwan
and Wheals (2004) and Thompson (2007). This section outlines key process
steps of fermentation, and the general changes the beans undergo.
13
The fermentation is conducted by indigenous microorganisms that originate

from the surface of the cocoa pods, fermentation vessels, hands of workers,
insects, air and soil (Forsyth, 1957; Maravalhas, 1966; Faparusi, 1970;
Maravalhas, 1966; Ardhana and Fleet, 2003; Jespersen et al., 2005). Various
species of yeasts, acetic acid bacteria, lactic acid bacteria and Bacillus spp. are
involved. The yeasts, lactic acid bacteria and acetic acid bacteria grow very
quickly in the pulp material, reaching maximum populations of about 108 cfu/ g
within 24-72 hours. Subsequently, their populations decrease, eventually giving
way to the growth of Bacillus species that are more tolerant of the higher
temperatures (45-50°C) that prevail in the later stages of fermentation (Dougan
et al., 1981; Carr et al., 1979; Schwan, 1998; Nielsen et al., 2007b; Camu et al.,
2007)
This microbial activity during fermentation causes physical and chemical

changes that impact on bean physiology and biochemistry. Degradation of pulp
facilitates the diffusion of oxygen into the bean mass (Quesnel, 1968; Dougan et
al., 1981;). The temperature of the fermenting beans is not controlled by any
external means, aside from any insulation due to the fermentation vessel or
wrappings. Typically the temperature of the beans at the start of fermentation
will be 25°C, increasing to 45-50°C as the fermentation proceeds (Swain, 1957;
Quesnel, 1968; Dougan et al., 1981; Senanayake et al., 1995;). The production
and utilization of acids, especially utilization of citric acid by lactic acid bacteria
and yeasts, causes the pFl of the bean cotyledons to decrease from about 7.0, to
5.0-5.5 (Biehl et al., 1985; Tomlins et al., 1993; Senanayake et al., 1995; Ardhana
and Fleet, 2003; Galvez, 2007; Camu, 2008a, b). Glucose and fructose in the pulp
are metabolised to ethanol and acetic acid by the yeast and bacteria. These
metabolites, combined with the temperature increase, diffuse into the cocoa
seeds, cause a breakdown in the cellular structure of the cotyledons and,
concurrently, kill the embryo. These severe changes allow endogenous cocoa
seed enzymes to act on previously compartmentalised substrates (Biehl et al.,
1977, 1982a,b; Biehl and Voigt, 1996; de Brito et al., 2000). Proteases break down
globulin storage proteins, forming flavour-active peptides and amino acids.
Polyphenol oxidases decrease bitterness and astringent properties by oxidizing
bean polyphenols, such as the catechins, anthocyanins and proanthocyanidins.
Anthocyanidin pigments are glycosylated, and are decolourized by glycosidase
action, assisting in colour development. Invertase hydrolyses seed stores of
14
sucrose to glucose and fructose (Forsyth and Quesnel, 1957a; Lopez et al., 1978;
Voigt and Biehl, 1995; Hansen et al., 1998; Misnawi et al., 2002, 2003). Volatile
microbial metabolites such as 2-phenylethanol, 4-tetramethylpyrazine, guaiacol,
indole and 2-acetyl-l-pyrroline may also diffuse into the cocoa seeds and
contribute positively or negatively to their flavour (Zak et ah, 1972; Lopez and
Quesnel, 1973; Barel, 1985; Flament, 1991; Ouoba, et ah, 2005).
Practically, fermentation involves freshly obtained cocoa beans being placed in
some heap or container for between 3-7 days. In most countries, cocoa beans are
fermented using traditional, village scale operations. In West Africa and in
several other countries, the predominant method for fermenting cocoa beans is
still what is referred to as the 'heap' method. For heap fermentation, 25 -1500
kg of fresh cocoa beans are heaped on banana leaves on the ground, wrapped in
more banana leaves and left for 2-7 days. These heaps may be opened up and
mixed after 48 hours, but with smaller heaps it is common for mixing to not
take place (Carr et al., 1979; Baker et ah, 1994; Nielsen et al., 2007b; Camu et al.
2008a, b). In South-East Asia and Brazil, wooden boxes are mostly used as
fermentation vessels. The dimensions of such boxes can vary significantly, from
small boxes with a capacity of about 50-200 kg, up to large boxes capable of
holding 800-1200 kg of wet bean. In box fermentation mixing usually occurs
several times during fermentation, and this may be facilitated by arranging the
boxes in a cascade formation (Chatt, 1953; Rohan, 1963b; Minifie, 1980; Schwan,
1998; Ardhana and Fleet, 2003). Another less common method, mainly used in
West Africa, is tray fermentation. This is similar to the box method, except that
shallow wooden trays are used to contain the beans.(Rohan, 1963a; Seike, 1973;
Nielsen et al., 2007b). Cocoa beans may also be fermented in other vessels, such
as baskets or troughs, although this is much less common than the heap, box
and tray methods (Carr et al., 1979; Wood and Lass, 1985; Baker et al., 1994;
Schwan and Wheals, 2004). Table 2.2 outlines which fermentation methods are
most commonly employed in each of the main cocoa producing nations. The
duration of fermentation, and frequency of mixing, will depend on a range of
factors, including bean cultivar and the farmer's preference. Mixing during
fermentation aerates the beans and provides some physical uniformity, but its
15
application is not consistent (Said and Samarakhody, 1984; Senanayake et al.,

1997; Hashim et al., 1998b; Camu et al., 2008a, b).
Table 2.2 - Survey of methods used to ferment cocoa bean in producing countries
Country Predominant Predominant Size Typical Mixing References

cultivar method duration frequency
Carr et al.
(1979)
Ghana Heap, tray 50-1200 kg Not mixed, or Tomlins et al.

Forastero* 3-6 days
once at 48 h (1993)
Nielsen et al.
(2007)
Ivory Coast Forastero* Heap, box. 50-1500 kg 3-7 days Not mixed, or Sanchez et
once at 48 h al., 1985
Nigeria Forastero Heap, box. 50-500 kg 2-7 days — Rohan. 1963 b
Schwa
Forastero, (1998),
Brazil Criollo, Box 250-1000 kg 5-7 days Every 24-48 h Schwan and
Trinitario Wheals
(2004)
Venezuala Criollo and Box 100 - 1000 3-7 days — Rohan. 1963b
Trinitario kg
Ecuador Forastero* Heap, box. 100- 500 kg 4-6 days After 48 h and Rohan, 1963b
120 h
Dominican Trinitario Every 24-48 h Galvez et al.
Heap, box. 100 - 800 kg 5-7 days
Republic (2007)
Rombouts,
(1952),
After 48 h and
Trinidad Trinitario Box 100-2000 kg 6-8 days 120 h
Bekele(2003),
(2008)
Toya vessel 2- 3 days forCriollo not

Criollo, for Criollot; Criollo; mixed; Rohan,
Mexico Forastero 50-1000 kg 1963b,
and Box for 3- 7 days for Every 48 h for Rosenblum,
Trinitario Forastero, Forastero and Forastero/ 2005
Trinitario Trinitario. Trinitario

Costa Rica Trinitario Box 100-1000 kg 3-7 days — Rohan, 1963b
Trinitario Ardhana
(1990)
Indonesia and Box 50-1000 kg 4-6 days Every 24 h Ardhana and
Forastero Fleet(2003)
Carr et al.
Malaysia Forastero Box 250-2000 kg 4-6 days Every 24-48 h (1979)
Wood and
Papua New Box, heap Not mixed, or Lass, 1985;
Forastero 5-1000 kg 2-8 days
Guinea every 24-48 h. Hollywood) 1
998)
* The predominant West African strain of Forastero is termed Amelonado (Wood and Lass, 1985;
Efombagn et al., 2006). The strain of Forastero native to Ecuador is called 'Nacional' and is thought to have
diverged genetically from the Amazonian Forastero (Motomayor et al., 2002, 2003).
t The Toya vessel is a large earthenware or wooden trough shaped vessel descended from forms
reportedly used by the Aztec to ferment cocoa (Rohan, 1963b; Coe and Coe, 2003).
16
In spite of the critical role of fermentation in developing the flavour and quality
of cocoa beans, it is usually performed in an uncontrolled manner, yielding
unpredictable results (Bekele, 2003). Furthermore, there is a need for better
linkages between the growth and activity of individual microbial species, and
the final quality of cocoa beans.
2.1.2.4 Drying
At the end of fermentation, cocoa beans contain about 55% water. In order to
prevent mould growth and ensure a good quality product, the beans must be
dried to a final moisture content of between 4-8% (Wood, 1980; Faborode, 1995;
Augier et al., 1998; Garcia-Alamilla, 2007). Dried, fermented beans have two
major components: the nib (the dried, fermented seed material); and the shell
(dried, fermented testa and pulp remnants) (Del Boc, 1962; Anon., 1984; CAA,
2008).
Drying may be performed by solar or artificial means. For solar drying, the
beans are spread in a single layer, on flat surfaces in the sun for 1-2 weeks.
Attention needs to be paid to the beans during this time to deter pests, and
cover the beans during rain. This may or may not be consistently performed,
and several authors have observed farmers failing to take appropriate care
during drying (Quesnel, 1966; Duncan et al., 1989; De Brito et al., 2000;
Mounjouenpou et al., 2008). In areas of high humidity or frequent rainfall, the
beans may be dried in a hot air dryer, or by convection on a heated surface.
Drying cocoa using assisted means usually takes a much shorter time, between
2-4 days (Wood, 1971; Barel, 1995; Faborode, 1995). In some situations, a
combination of solar and artificial drying may be used (Rohan, 1963b; Wood
and Lass, 1985. In addition to removing moisture, the drying process allows a
continuation of many biochemical reactions initiated during fermentation. A
balance is necessary between the need to preserve the beans and the need to
allow these reactions to continue (Jinap et al., 1991; Hoskin and Dimick, 1994;
Flashim et al., 1999; Schwan and Wheals, 2004; Kyi et al., 2005). As with
fermentation, inadequate control negatively impacts the quality of the beans.
Some common examples of drying faults include: overly-rapid drying leading
to weak and/or acid flavours, drying too slowly allowing mould growth
(musty off flavours and mycotoxin production), and smoke from artificial dryer
17
tainting the beans (smokey off flavours) (Biehl, 1984; Wood and Lass, 1985;
Guittard, 2005; Garcia-Alamill et al. 2007; CAA, 2008).
In some parts of Indonesia, Papua New Guinea and Sri Lanka, the fermented
beans are washed prior to drying (Wood and Lass, 1985; Hollywood, 1998;
Ardhana and Fleet, 2003), while in the West Indes beans used to be rubbed or
'danced' as they dried (Rohan, 1963b). The understood reason for these
practices was to reduce the shell content of the dried beans, although the
effectiveness of these practices has been little studied (Hollywood, 1998).
2.1.2.5 Storage and transport
After drying, cocoa beans are usually stored in jute or synthetic sacks, in
quantities of 60-100 kg. If cocoa is to be stored for any length of time,
precautions must be taken prevent mould growth, and losses due to pests. In
order to prevent the hygroscopic beans absorbing moisture, it is recommended
that warehouse humidity be kept below 80% (Roelofsen, 1958; Maravalhas,
1966; Wood and Lass, 1985). A large variety of animal species, both vertebrate
(rats, mice, birds) and invertebrate (moths, beetles and weevils), can cause
losses to stored cocoa. These pests are controlled by ensuring the structural
integrity of the warehouse and storage packaging, and by periodic fumigation
to limit losses from insects (Lass, 1999; Thompson et al., 2007). For international
transport, cocoa beans are typically containerised. Quality loss during
transport can be minimized by using ventilated vessels that are clean and free
from chemical contaminants. For long distance sea transport, properly-
designed, ventilated containers should be used (Wood and Lass, 1985; ICCO,
2007d).
2.1.2.6 Roasting, winnowing, grinding
Cocoa beans are typically tested for basic quality parameters both before and
after transport, After cleaning to remove foreign matter and damaged beans,
the beans are then roasted. Roasting may occur in a batch or continuous
process, usually between 110-140°C for 25-60 minutes (Barrile et al., 1971;
Beckett, 1988; Redgwell, 2003; Krysiak, 2006). Most importantly, the roasting
process completes the process of flavour development that was begun during
fermentation. Methyl-pyrazines and other aroma volatiles are formed during
roasting. The colour of the beans develop through Maillard reactions between
18
reducing sugars and amino acids liberated during fermentation (Ziegleder,

1991; Jinap et al., 1998; Ramli et al., 2006). Physically, roasting also helps to
loosen the shell from the nibs. Since roasting is a critical step in determining
chocolate quality, roasters are designed and operated to ensure even heat
distribution (Minifie, 1980; De Brito et al., 2000).
After roasting, the beans are cracked and winnowed, separating the shell from
the nib of the cocoa beans. This is usually achieved by a combination of
vibrating sieves and air currents. Mechanical separation process is highly
efficient, with as little as 0.5% shell carried through (Minifie, 1980; Beckett, 1988;
Bixler and Morgan, 1999).
The separated cocoa nibs are then ground into a smooth paste called cocoa
liquor or cocoa mass. In addition to reducing particle size, grinding releases the
cocoa fat, which coats non-fat particles and provides a viscosity to the cocoa
liquor (Beckett, 2000). Some manufacturers process cocoa beans to cocoa liquor
at a central location, solidify the liquor into large (0.5-2 tons metric) blocks
called "melters," and then ship these to other plants for chocolate manufacture
(Minifie, 1980; Knight, 1999; Cadbury Schweppes pers.comm, 2004).
2.1.2.7 Pressing and alkalisation
A certain proportion of cocoa liquor is mechanical pressed to separate the cocoa

butter and form a press cake. This press cake is then milled into finely ground
cocoa powder. Cocoa powder can be treated with mild alkali which reduces its
acidity, improves its solubility and lightens its colour. This process, also called
'dutching,' is usually applied where the cocoa powder will be used in the
preparation of beverages (Gabis et al., 1970; Beckett, 1988; Bixler and Morgan,
1999).
2.1.2.8 Chocolate Processing
To make chocolate, cocoa liquor is combined with varying quantities of sugar,

cocoa butter, and milk powder for milk chocolate. These ingredients are mixed
together in a mechanical mixer for 20-30 minutes at 50-60°C. After mixing, the
ingredients are refined or milled in order to reduce the particle size of all
ingredients to 30 microns or less. The amount of sugar and/or milk powder
added will affect the colour, sweetness and flavour of the chocolate. IN some
19
cases, flavours such as vanilla may also be added (Cook, 1972; Lees and
Jackson, 1973; Martin, 1987; Bixler and Morgan, 1999). Next, the liquid chocolate
undergoes conching - a process combining kneading and heating of the
chocolate mass. Conching helps develop the flavour of the chocolate by driving
off undesirable astringent or acidic compounds. Conching also improves the
free flowing properties of liquid chocolate by coating the solid particles with fat.
To assist this process, an emulsifier such as lecithin may be added (Minifie,
1980; Beckett, 2000; Counet et al., 2002; Perego et al., 2004). Advances in the
design and operation of conches have allowed conching time to be as little as a
few hours, although a few manufacturers of fine chocolate still conch for up to 6
days (Beckett, 2000; Counet et al., 2002). Because conching affects the final
quality of the chocolate, manufacturers carefully optimise and control this step.
The crystal structure of solidified cocoa butter is responsible for the texture of
chocolate, and may also affect the release of flavour volatiles (Urbanski, 1992;
Bixler and Morgan, 1999; Beckett, 2000; Anon., 2003). The triacylglycerols in
cocoa butter can form, and convert between, six polymorphic crystallization
forms. One of these, the high-melting |3 form, has the following properties
desirable for chocolate manufacture: a melting point in the range 30-35°C, good
snap, and a glossy appearance (Hoskin, 1994; Beckett, 2000; Anon, 2003). In
order to promote the formation of fat crystals in (3 form, the chocolate is
tempered (Beckett, 2000; Schwan and Wheals, 2004). Specialty, small volume
chocolatiers usually perform tempering by manually mixing molten chocolate
on a marble slab. Industrially, tempering is performed in special heat
exchangers. These tempering machines usually have three zones, sequentially
cooling the chocolate to initiate crystallization, further cooling and shearing to
promote the formation of |3 form cocoa butter, and, finally, heating the chocolate
to 30°C to destroy unstable crystals. The time and temperatures of tempering
are always tightly controlled (Lees and Jackson, 1973; Minifie, 1980; Beckett,
2000; Anon, 2003).
Once tempered, the liquid chocolate can be moulded and shaped into a variety
of products. By solid-filling mould, solid blocks or bars may be produced.
Hollow chocolates, such as eggs or figures are created using spun moulds. Soft
20
centered chocolate products are made using an enrober, which cover a piece of
caramel or fondant in chocolate. These basic shaping and moulding processes
are used both by artisans producing small volumes of premium products, and
industrialised producers manufacturing many thousand of units per hour.
(Beckett, 1988; Coe and Coe, 2003; Schwan and Wheals, 2004).
Finally, the finished chocolate is packaged. The packaging type must take into
account the likely storage conditions, the market and consumers, and the
expected use. Solid bars are usually wrapped in foil, paper, or plastic film
laminates, while cocoa powder is usually sold in air-tight tins (Minifie, 1980;
Beckett, 1988).
21
2.2 Market and economics of cocoa and chocolate

The Cocoa Market
In 2005, global sales of chocolate exceeded $65 billion, with the figure being
closer to $70 billion in 2008. This booming chocolate industry requires large
quantities of cocoa beans, and these are supplied by over 25 countries (Stiff,
2006; ICCO, 2008). A large variety of factors affect the supply and demand of
cocoa beans. An appreciation of these factors is useful, especially the context of
this project, which is part of a greater initiative to establish a new cocoa bean
industry in Australia (Lemin, 2005). This section describes some of the basic
economics, and key market trends, of cocoa production worldwide.
2.2.1 Basic economics of the cocoa market

2.2.1.1 Demand for cocoa beans
Two thirds of all cocoa beans are imported and processed by European
countries (58%) and the USA (22%). Demand has increased in Asia, Eastern
Europe and Latin America. Asia was estimated to have increased its demand for
cocoa beans by a remarkable 15% over the last 5 years (ICCO, 2007a; WCF,
2008). These general increases reflect the increasing consumption of chocolate in
these countries (ICCO, 2007a). The increased demand is also linked to more
intense competition for market shares among chocolate manufacturers. The
current market leaders in terms of sale revenues are Mars Inc., Cadbury
Schweppes, Nestle, Ferrero Group, and FFershey Foods Group (Rogers, 2006).
Market analysis also highlights several trends in chocolate consumption that are
behind changes in demand. For example, the increased sale of single-origin,
dark and premium chocolate products has increased demand for cocoa beans,
both in terms of quantity and quality (Greenwood, 2005; Guittard, 2005;
Pacyniak, 2005; Scully, 2006; Anon, 2007). Other market trends, such as growing
interest in 'ethically'-produced chocolates (Organic, fair-trade, rainforest) have
marginally increased demand for beans produced according to specific
requirements (ICCO, 2005 b, c; ICCO, 2006). These trends suggest an increased
demand for cocoa beans produced under more controlled conditions, whether
for quality or certification (organic, fair-trade) purposes (Nielsen et al., 2007b;
ICCO, 2007b, 2008).
22
2.2.1.2 Supply and production of cocoa beans
To supply the chocolate industry, approximately 3.5 million tonnes of cocoa

beans are produced globally each year, with a total value of US $5-6 billion
(ICCO 2007a, WCF, 2008). Currently, West African countries produce 70% of the
world's cocoa beans (Figure 2.6). This geographical concentration of cocoa
production has developed over the past century, with Cote D'Ivoire accounting
for 40%, Ghana 20% and Nigeria and others the remaining 10% (Schwarz, 1928;
WCF, 2008). The remaining 30% of world production is accounted for by
Indonesia, Ecuador, Malaysia, Papua New Guinea, and a number of other
countries (Beckett, 2000; ICCO, 2007a; COPAL, 2008).
About 70% of world production is grown by small-holders on a low input, low

output basis. Typically, family or village labour is used at relatively little cost,
trees can be individually managed and the quality of bean fermentation is
usually assured. As a rule of thumb, one labourer is required per 2.5 ha of
established cocoa in traditional production systems (Lass, 1999; Lemin, 2005).
Others
11.8%
Ecuador
3.4% . 410,000
tonnes Ivory coast
Brazil
4.5% 37.4%
1.3 Million tonnes
Nigeria
4.5% i
Camerooi 175.000 tonnes

5.0%
440,000 tonnes
720,000 tonnes
Indonesia
12.7%
Ghana
20.7%
Figure 2.6 Share and volume of world cocoa production in the 2006/2007 period.
(Percentages based on total production of 3,480,000 tonnes) (ICCO, 2007a).
23
Figure 2.7 shows changes in the production levels and average price per ton of
cocoa over the last 35 years. The large increase in prices in the period 1976-79
was a result of a sudden increase in chocolate consumption, and a slowing
production of cocoa beans in many countries. Increased production saw prices
fall back to more reasonable levels (Powell, 1983).
r-- lS r- CO OO C>0 OO ON ON ON CD CD
0\ ON ON ON ON ON ON ON ON ON CO CD
« 1 i i i < » < » t » 4 f < i < i I r-H f\]
----- Pnces (S/t) ----- Gross crop (en '000 tons)

Figure 2.7 - Prices ($US/tonne) and production levels (1000’s of tonnes) of cocoa beans
between 1971-2006 (UNCTAD, 2006).
While cocoa production has remained relatively strong over the past 20 years,
several threats to cocoa supply have been identified. The main threats identified
in recent years include ongoing political instability in Cote D'Ivoire, and
problems associated with child labour in many cocoa producing countries
(Anon, 1999; ICCO, 2005a; Boal, 2007). Cocoa production in Brazil and Malaysia
has declined over the last 5-10 years due to the impact of pests such as Witches
Broom (Crinipellis perniciosa), and the replacement of old cocoa plantations with
more profitable crops such as oil palms (ICCO, 2007a). The sudden drop in
prices and subsequent drop in production between 1998 - 2003 underlined
several of these concerns amongst the chocolate industry. In response, several
companies and organizations launched projects aimed at securing or improving
cocoa production around the world (Anon, 1999; Menendez, 2004; ICCO, 2008).
In the period 2003-2008, cocoa production levels have increased and now
generally meet demand (WCF, 2008; COPAL, 2008).
24
2.2.2 Current market trends and opportunities

In 2006 one survey of European consumers revealed that annually, 11% of
respondents ate some single origin chocolate, 11% ate some fair trade chocolate,
and 6% ate some organic chocolate (ICCO, 2007a). As already mentioned, the
demand for cocoa beans produced under more controlled conditions has
increased due to the increased marketing of chocolate products for their ethical-
value (Organic, fair-trade, rainforest), their quality (single-origin) or their
purported health benefits (Pacyniak, 2006). These trends are outlined in more
detail below.
2.2.2.1 Ethically oriented cocoa products
2.2.2.1 A - Organic Cocoa
In 2006 it was estimated that the organic chocolate market had a value of US
$304 million, nearly double the 2002 value of US$171 million (IFOAM, 2008).
Such rapid growth has been attributed to increasing levels of concern amongst
consumers about the safety of their food supply, and about the state of the
global environment. However, organic cocoa still only represents only a very
small share of the total cocoa market (approximately 15,500 - 20,000 tonnes per
year - less than 0.5% of total production) (ICCO, 2006). The International
Federation of Organic Agriculture Movements (IFOAM), acts as an umbrella
organization for participants in the organic market. Together with its members
this body acts to define principles of organic agriculture and ensure that they
are correctly adopted. In addition to certification requirements, producers of
organic cocoa must ensure they comply with any legislation specified by
importing countries regarding 'organic' products. The benefit of certification for
cocoa farmers is that organic cocoa sells at a higher-than-market prices, with
premiums of between US$100-300 per tonne (ICCO, 2006; IFOAM, 2008).
Organic cocoa is produced in a number of countries, including: Madagascar,
Tanzania, Uganda, Belize, Bolivia, Brazil, Costa Rica, Dominican Republic, El
Salvador, Mexico, Nicaragua, Panama, Peru, Venezuela, Fiji, India, Sri Fanka
and Vanuatu (ICCO, 2006; IFOAM, 2008). Green & Black's (Cadbury-
Schweppes),Dagoba and Rapunzel, are examples of chocolate manufacturers
who use organic cocoa beans in their products (Cadbury-Schweppes, 2008;
Zusman, 2008).
25
2.2.2.1 B - Fair Trade cocoa and chocolate
Some consumers show increasing interest in the people who produce

commodities such as cocoa. The Fair Trade system is promoted as a sustainable
model of development which can help in securing the rights of disadvantaged
producers and workers (ICCO, 2005b; FLO, 2008). The essential characteristic of
Fair Trade cocoa is that participating producers receive a higher price for their
cocoa beans. This 'Fair Trade price' includes a premium calculated to allow the
producer(s) to fulfill basic labour conditions and enable social, economic and
environmental development projects. Fair Trade prices are set by The Fairtrade
Labelling Organization (FLO), which was established in 1997. The current fair
trade price for cocoa is US$ 1,750 per tonne, which includes a US$150 per tonne
premium for development projects. Presently, 83% of all fair-trade chocolate is
sold in UK and Europe, but only makes up about 0.1% of all chocolate sold. In
addition to being organic, the cocoa beans used by Green & Blacks are also Fair
Trade certified. (ICCO, 2005,b; FLO, 2008)
2.2.2.1 C Rainforest cocoa
In Cote D'Ivoire, Brazil, and several other South american countries, cocoa is
sometimes grown in the under-story of existing rainforest. Ecological,
environmental and economic research has suggested that such an approach can
lead to better conservation of vulnerable species (Johns, 1999; Bright, 2001).
Certification of such "rainforest cocoa" is performed by the Rainforest Alliance.
Currently only about 30 farms or cooperatives worldwide are "Rainforest
Certified," and the market share of chocolate made from such beans is
miniscule (RA, 2008).
2.2.2.2 Changing perceptions of the health and nutrition of chocolate and
cocoa - Chocolate marketed for potential health promoting value
The potential health benefits of consuming food rich in antioxidant compounds
have been promoted by nutritionists and government health bodies, as well as
by food manufacturers (Serafini, 200; Chenynier, 2005). Targeted foods have
included fruit and vegetables in general, berries, tea, wine, and recently, cocoa
and chocolate (Calver et al., 2005; ICCO, 2005c; Keen, 2005). Given such intense
interest, it is perhaps unsurprising that the polyphenolic compounds of cocoa
and their antioxidant properties have been thoroughly researched (Adamson et
al., 1999; Wollgast and Anklam, 2000; Counet and Collin, 2003; Kris-Etherton,
26
2003; Counet et al., 2004; Gu et al., 2006; Miller et al., 2006; Nazaruddin et al.,
2006). In their review, Lamuela-Raventos et al. (2005) outline the major health
benefits that are claimed to result from consumption of flavanol-rich cocoa and
chocolate, namely: protection against cardiovascular disease (Hollenberg et al.,
2004; Keen et al., 2005), cancer, reduced brain-damage from strokes (Heo and
Lee, 2005) and improved immuno-modulation(Ramiro et al., 2005). It is
suggested that more work must be done before some of the more serious health
claims are accepted widely. Work by Keen et al. (2005), Grassi et al. (2005), and
Vlachopoulous et al., (2006) suggests that any improved health outcome must
be related to the level of polyphenolic compounds present in the cocoa beans.
Surveys of chocolate and cocoa-containing foods found that the level of
polyphenols in a given chocolate product was roughly proportional to the level
of cocoa solids in the product (Lee et al., 2003; Miller et al., 2006). Furthermore,
it was found that that consumption of dark chocolate led to raised blood levels
of these compounds (Kris-Etherton et al., 2004). Such findings have led the
chocolate and cocoa industries to promote ordinary dark chocolate as
potentially health promoting (Calver et al., 2005; ICCO, 2005c; Partos, 2008).
Both Nestle and Mars have sought ways to supplement milk chocolate products
with extra polyphenols, lodging patents for processes to manufacture chocolate
with elevated levels of polyphenols (Kealey et al., 2001; Myers et al., 2001;
Kochnar et al, 2002).
There are several obstacles to the promotion and marketing of chocolate along
health lines. Firstly, many countries including Australia have strict guidelines
regarding the inclusion of specific health claims on food labelling (FSANZ,
2008). Thus, promoting a chocolate product may not be allowed, or may involve
fulfilling costly legislative requirements. Estimating the proportion of chocolate
products marketed for health benefits is therefore very difficult. Secondly,
health authorities continue to highlight the negative consequences of excess
consumption of confectionary, including chocolate: particularly the increased
risk of dental caries, obesity and diabetes (Stiff, 2006). These problems largely
stem from the sugars in the chocolate, rather than the cocoa beans (Schwan and
Wheals, 2004). Furthermore, market data suggest that in spite of these obstacles,
consumers are increasingly viewing chocolate as less unhealthy than other
types of confectionary (Calver, 2005; Partos, 2008).
27
2.2.2.4 Chocolate as a sophisticated gourmet food - flavour, dark and

single origin chocolate.
Chocolate is primarily eaten for its unique and pleasing flavour, and while
ethical or health concerns have had influences, the greatest growth area in
chocolate markets has been in dark and specialty chocolates. Dark chocolate
now comprises 10% of the global chocolate market, with an increase of nearly
10% consumption of dark chocolate consumption in the USA between 2001 -
2005. Furthermore, it is estimated that one third of all new chocolate products
launched in 2006 were dark chocolate (Scully, 2006; ICCO, 2007).
Typically, dark chocolate must have a minimum cocoa solids level of 40%, and
may have up to 99% cocoa solids (Beckett, 1988). Increased consumption of
dark chocolate has clearly been linked to increased demand for cocoa beans,
because of its higher cocoa content (ICCO, 2007a, b). It has also increased
attention paid to the quality of cocoa beans used - the higher the cocoa
percentage, greater the impact the quality of the beans have on the chocolate
(Beckett, 1988, Hoskin and Dimick, 1994). Cocoa bean quality goes from
important, to critical, in the manufacture of single-origin chocolate. Single
origin chocolate is a subtype of dark chocolate and is so-called because it is
made using cocoa beans from a single source. Usually this means a specific
producer or set of producers from a specific region, in a particular country.
Some of the major suppliers of single origin cocoas are Sao Tome, Tanzania,
Ghana, Dominican Republic, Cuba, Papua New Guinea, Grenada and Ecuador
(Doutre-Roussel, 2005; Guittard, 2005). One consequence of the growing
popularity of single-origin chocolate is that more consumers are aware of the
variations of chocolate flavour that exist. Chocolate made with one cultivar of
beans, grown in a particular region, and fermented according to a particular
method have distinct, communicable character that distinguish them from other
beans (Guittard, 2005; Nielsen et al., 2007b; Camu, 2008a, b). This discernment,
similar to the 'terroir' concept in wine, allows for the possibility that the
consumer can taste taints or technical faults in the chocolate that have their
origins in the fermentation or drying methods used (Baker et al., 1994;
Greenwood, 2005; Guittard, 2005; Sukha et al., 2008). Consumers are noticeably
embracing the idea of dark chocolate as an affordable luxury, much like fine
wine, cheese or coffee (Greenwood, 2005; Rosenblum, 2005). The world cocoa
28
market divides cocoa into two types, namely bulk cocoa (used in milk and
lower quality dark chocolate), and fine or flavour cocoa (used in specialty and
premium dark chocolate), with the former accounting for 93.5% of global
productivity (Lass, 1999). Ecuador, Grenada, Trinidad and Tobago, Jamaica,
Venezuela, and Costa Rica are among the 17 countries producing fine cocoa
with aromatic flavour (Perego et al., 2004; ICCO, 2005). Fine cocoa beans, while
making up a smaller share of the market can command premium prices (ICCO.
Investment in production of fine cocoa, however, requires substantial control
and manipulation of cocoa and chocolate flavour, for which the present study
serves.
2.2.3 Australian domestic chocolate market status
In evaluating the feasibility of cocoa bean production, the existence of a stable

and growing market for chocolate and cocoa products in Australia is an
important consideration. The Australian chocolate industry is enjoying a rapid
growth in profits. In 2002 the retail sale value of chocolate products was A $1550
million, contributing 56% of the total value of the confectionery industry. It is
estimated that Australians consume up to 4.4 kg of chocolate a year, with milk-
chocolate remaining the most popular. The Australian chocolate market is
dominated by Cadbury Schweppes, which holds 54% of the market shares, and
is challenged by Masterfoods (Mars) and Nestle Australia, with 19% and 17% of
the market shares, respectively (CMA, 2004).
2.2.4 Forecast Summary for the Cocoa market
Forecasts for the next 5 years predict that cocoa prices will remain steady, with
both supply and demand increasing by about 3% per year. The dominance of
the West Africa region is expected to continue, assisted by recent improvements
to political and social conditions in Cote D'Ivoire. Furthermore, the market
share held by dark and specialty chocolate is expected to continue to increase,
thus also increasing demand for quality cocoa beans (Pacyniak, 2005; ICCO,
2007a; WCF, 2008). Another predicted growth factor is the continued increase of
chocolate consumption in Asian markets (ICCO, 2007b). At the same time,
concerns have been raised over the impact of climate change, the international
economic downturn and a growing awareness of child-labour, on cocoa
production and prices. These all have the potential to reduce supply, and/or
decrease prices gained at market (COPAL, 2008).
29
2.3 Development of a cocoa industry in Australia

Cocoa plants were informally introduced to Australia as early as 1900, although
no commercial propagation or cultivation resulted at that time. In the late 1980's
and early 1990's, the Queensland state government found that some regions of
northern Australia would be suitable for the cultivation of cocoa (Corthorne,
1987, Watson, 1987, Watson, 1990). This possibility was not investigated further
until 1998-99, when several threats and opportunities in the cocoa market were
identified. These market factors, discussed in the previous section, prompted
the establishment of joint government-industry sponsored feasibility studies.
The aim of these projects was to determine if Australia could economically
produce high quality cocoa beans, to supply the domestic market, and/or the
international premium cocoa market (RIRDC, 1999; Lemin, 2005a, 2005b;
Campagnolo and Lemin pers.comm). By 2000, several small plantings had been
established in northern Queensland and Northern Territory regions of Australia
(RIRDC, 1999) (Figure 2.8, Figure 2.9). The planting material used for these
plantings consisted of hybrid Forastero-Trinitario material obtained from Papua
New Guinea, and from the botanical collection at the University of Reading,
Growing Area
MELBOURNE
Figure 2.8 - Region of Northern Queensland where cocoa trials were cultivated.
30
Figure 2.9 — Photograph of Cocoa trees growing at the Mossman plantation
The first Australian cocoa crops were produced in 2002, with high yields being
achieved a year later. While the conditions in the Northern Territory were
considered too dry to maintain good production, agronomic studies have
indicated that commercial cocoa production could be established in northern
Queensland (Lemin, 2005a, 2005b; Campagnolo and Lemin pers.comm, 2006).
Higher labour costs for cocoa production in Australia necessitate an
industrialised approach in order to be economically competitive in international
markets (Lemin, 2005a). A controlled, industrialised process would also help
guarantee a premium quality product with a high market price. While
mechanised harvesting of the cocoa pods was found to be unfeasible, automatic
pod splitting and sorting machinery were developed for use in Australia (Puri
et al., 2007). Having established these important preliminaries, efforts wre then
begun to develop industrialised methods of cocoa fermentation and drying, for
use with the Australian grown cocoa beans. Cocoa beans had never been
fermented before in Australia and there was an absence of basic microbiological
and chemical data available. Therefore this project was commenced to study the
fermentation of Australian cocoa, its microbiology and chemistry and the effects
of these on the final flavour and quality of the beans.
31
2.4 The Biochemistry and Microbiology of Cocoa

Fermentation - effects on flavour and quality.
In order to understand and improve the fermentation, it is necessary to know,
the chemical and physical composition of the bean substrate, what factors affect
these properties, and how these properties affect the fermentation process and
quality of the final product. Cocoa beans are composed of a wide range of
compounds, and their chemistry has been extensively reviewed (Forsyth, 1949;
Quesnel, 1963; Rohan, 1963b; Hoskin and Dimick, 1994; Biehl and Voigt; 1996;
Schwan and Wheals, 2004; Afoakwa et al., 2008). During fermentation many of
these compounds undergo significant changes, by the activities of
microorganisms, and by various enzymic and non enzymic reactions within the
beans themselves. Many of these biochemical changes are responsible for the
final flavour and quality of the cocoa beans (Biehl and Voigt, 1996). Because
some of these changes are the result of microbial activity, monitoring the
chemistry of fermentation can also help clarify the role of microorganisms
(Schwan and Wheals, 2004; Ardhana and Fleet, 2003; Nielsen et al, 2007a, b;
Camu et al., 2007; Camu et al., 2008a, b). The following section briefly reviews
the complex chemistry involved in cocoa fermentation.
2.4.1 Physical and chemical properties of unfermented cocoa beans
2.4.1.1 Physical structure of unfermented cocoa beans
Fresh cocoa beans are the substrate for cocoa fermentation. Discussing the
chemical composition of cocoa beans, and its changes, is complicated by the
physical structure of the beans. In cocoa fermentation the substrate for
microbial growth (the pulp) is physically separated, by the testa, from the
material that is eventually consumed as food (the seed) (C.f. Figure 2.5) (Smith,
1913; Knapp, 1937; Roelofsen, 1958; Rohan, 1963b; Jones and Jones, 1984; Wood
and Lass, 1985). This contrasts to many other food fermentations, where the
substrate the microorganisms grow in, and metabolize, becomes the final food
product. For example, in wine fermentation, grape juice is the substrate and is
transformed to become wine. Similar examples include bread, cheese and
tempeh (Fleet, 2001; Meroth et al., 2003; Steinkraus, 2004; Nout and Kiers, 2005).
32
Because the pulp and seed are physically separate, functionally distinct
materials, there is a strong case that they should be analysed separately. In a
number of studies, the chemical composition of cocoa beans is reported for a
total mixture of pulp and seed (Weissberger et ah, 1971; Reineccius et ah, 1972;
Zak, 1973; Lopez and Quesnel, 1973a, b; Hashim et ah, 1998; Dzogbefia et ah,
1999). This makes interpreting and comparing data problematic. Most studies in
the last decade have performed separate chemical analyses for the pulp and
seed components (Ardhana and Fleet, 2003; Nielsen et ah, 2007b; Camu et ah,
2007, 2008).
The properties of the testa, separating the pulp and seed components, are also
significant when discussing chemical changes during fermentation. Firstly, the
testa acts as a semi-permeable barrier to the flow of substances between the
seed and pulp. It has been demonstrated that the testa is freely permeable to
water, ethanol, acetic and lactic acids, and some volatile organic compounds
(Wood and Lass, 1985; Biehl and Voigt, 1996). This allows substances produced
in the pulp by microbial growth to diffuse into the seed. Transfer of constituents
across the testa, may be a rate limiting factor for certain compounds (Biehl et ah,
1982b). Early research observed that the testa increases in permeability as the
fermentation proceeds. Early research suggested that substances found in the
seed, including polyphenols, peptides, alkaloids, sugars, and some proteins can
diffuse back into the pulp during fermentation (Knapp, 1937; Roelofsen, 1958;
Quesnel, 1966; Kirchoff et ah, 1989a). There is, however, a lack of recent research
regarding the permeability of the testa, and how this changes during
fermentation.
Secondly, the testa acts to contain the substances that are released from the
cocoa seed cells when they are lysed during fermentation (Roelofsen, 1958;
Biehl and Passern, 1982; Biehl et ah, 1982b; Biehl et ah, 1985). Within the seed,
there is further physical or vacuolar separation of various components, such as
lipids, polyphenols and proteins. In fresh, undamaged seeds these vacuoles are
entire, and function to conserve their contents until germination of the seed.
During fermentation, these vacuolar structures are disrupted, allowing their
contents to diffuse into the space between the cells and interact.(Forsyth, 1964;
Biehl and Voigt, 1996; Bucheli, 2001; Dangou et ah, 2002). Figure 2.10 shows a
simple histology of the cocoa bean.
33
Figure 2.10 Histology of the cocoa bean (transverse section including pulp, testa and seed).
From Roelofsen (1959).
Finally, the testa affects the mass transfer rates during the bean drying process.
Several researchers have found that certain methods of drying may make the
testa more impermeable to water and acetic acid, and limit the diffusion of
these substances (Faborode et al., 1995; Augier et ah, 1999).
2.4.1.2 Chemical composition of unfermented cocoa beans
The mucilaginous pulp provides a mixture of substrates for the growth of

microorganisms. Table 2.x gives the average composition of fresh cocoa pulp.
The pulp is rich in fermentable sugars, with a total concentration of glucose,
fructose and sucrose representing 10-15%. The presence of sucrose in cocoa
pulp has been inconsistently reported. Dittmar (1956), Pettipher(1986), Hashim
et al. (1998) and Ardhana & Fleet (2003) found 1-4% sucrose in cocoa pulp
samples. By contrast, Galvez et al. (2007), Nielsen et al. (2007) and Camu et al.
(2008b) found only traces of sucrose in unfermented pulp. Testing by
34
Packiyasothy et al. (1980) suggested that sucrose is present in the pulp of

immature pods, but is converted to glucose and fructose as the pods develop to
maturity. Citric (1-3% or 60 -180 mg/g) and other organic acids gives the pulp
an initially low pH of 3.4-4.0 (Lopez and Dimick, 1995; Thompson et al., 2001;
Schwan and Wheals, 2004). Other organic acids found in fresh cocoa pulp
include malic, oxalic and tartaric acids (~ 0.1% or 5 -10 mg/g). Pectin and
polysaccharide, such as cellulose and hemicellulose, give the pulp a high initial
viscosity. The protein content of the pulp is very low. In a study by Pettipher
(1986), low levels (0.35 - 0.75%) of fat were reportedly found in cocoa pulp,
however this finding has not been followed up in any recent research.
Table 2.2 - Proximate analysis of fresh cocoa beans (pulp and seed)
Component Average concentration (% w/w)
Pulp Seed
Water 80-85% 35-45%

Lipid < 0.5% 45-55%
Sugars (sucrose, glucose and fructose) 10-16% 0.5-2%
Polysaccharides 1.5-3%% 14-20%
Pectin 4-7% 2.0%
Protein 0.6% 1.5-1.7%
Organic acids (predominantly citric acid) 1-3% 0.3-0.9%
Inorganic salt 0.5 - 1.0% 0.5 - 1.0%
Polyphenols <0.1% 7-10%
Alkaloids (theobromine and caffeine) <0.1% 3-3.5%
Adapted from Thompson et al. (2001); Ardhana and Fleet (2003); Schwan and Wheals (2004),
Camu (2008b).
The seed has a much lower water content than the pulp, partly due to the high
fat content (Table 2.x). The majority of the lipids exist in the form of
triglycerides (95%), with small amounts of di- and mono-glycerides,
phospholipids, glycolipids and sterols (<5%). The fatty acid composition of the
triglycerides is typically 26.5% palmitic, 35.4% stearic, 34.7% oelic and 3.4%
linoleic. This composition may vary depending on the average ambient
temperatures at which the beans are grown . This fatty acid composition, in
turn, affects the melting point of the cocoa butter in the finished beans, a
property critical in chocolate manufacture (Lehrian and Keeney, 1980a, b;
Wright et al., 1982; Chin and Zainuddin, 1984; Mohr et al., 1987). The lipids
found in cocoa butter are not absorbed well by the human body and minimally
affect serum cholesterol levels (Kris-Etherton, 1993). Small quantities of glucose
35
(0.5 -1.0%), fructose (0.5 -1.0% and sucrose (1.0 - 2.5%) are present in the seed
of unfermented cocoa beans (Forsyth and Quesnel, 1963; Berbert, 1979; Ardhana
and Fleet, 2003), along with a range of polysaccharides. The main
polysaccharides found in cocoa seeds include: starch (5.0 -10.%), pectins (2.0 -
2.5%), fibre (2.0%), cellulose (1.5 - 2.0%), pentosans (1.5%) and other gums (0.3 -
0.5%) (Knapp, 1937; Schmieder and Keeney, 1980; Lehrian and Patterson, 1983;
Ardhana and Fleet, 2003). Proteins are a significant component of cocoa seeds,
both in quantity (1.5 - 2.0%) and their role in the development of flavour
precursors (Knapp, 1937; Birch, 1941; Wright et al., 1982;). A large number of
different proteins have been isolated from cocoa beans, and they are usually
classified according to their solubility when fractionated, and their
electrophoretic mobility. The main classes of proteins found in cocoa seeds are
two globulins (47 and 31 kDa) and an albumin (21 kDa) (Biehl et al., 1982c; Zak
and Keeney, 1976; McHenry and Fritz, 1992; Buyukpamukcu et al., 2001).
Although lower than in the pulp, citric acid is the predominant organic acid
present in fresh cocoa seeds (0.3 - 2.0%, or 3 - 20 mg/g wet basis, or 15 -100
mg/g dry basis). Smaller amounts (0.05 - 0.5%; 0.5 - 5 mg/g wet basis; 5 - 25
mg/g dry basis) of oxalic, succinic and malic acids may also be present (Lopez
and Quesnel, 1973a, b; Carr et al. 1979; Bucheli et al., 2001; Ardhana and Fleet,
2003) . Spread throughout the cocoa seed are the "tannin cells." These are
storage vacuoles containing flavenoid polyphenols (6.5 - 9.5%), and in purple
beans (Trinitario and Forastero cultivars), anthocyanins (0.4 - 0.5%) (c.f. Figure
2.x) (Roelofsen, 1959; Dangou et al., 2002). The major polyphenolic compounds
contained in cocoa seeds are catechins (3.0 - 6.0 %), leucocyanidins (2.5%) and
tannins (2.0 - 3.5%). The polyphenols have bitter and astringent flavours, and
their antioxidant properties help protect the seed from damage and disease
(Forsyth, 1955; Forsyth and Quesnel, 1957b; Bracco et al., 1969; Kim and Keeney,
1984; Kyi et al., 2005). Unfermented cocoa seeds also contain two alkaloid
compounds - theobromine (2.5 - 3.2%) and caffeine (0.1 - 0.2%). Both are bitter
in taste, and act as mild stimulants on the human central nervous system
(Knapp, 1937; Humphries, 1939a; Senanayake and Wiesekera, 1971; Smit et al.,
2004) .
36
2.4.1.3 Factors affecting the composition of unfermented cocoa beans
The chemical composition of fresh cocoa beans can very, depending on their
cultivar, their maturity, and the conditions under which the cocoa pods were
cultivated (soil, climate, shade) (Rohan, 1963; Zak and Keeney, 1976; Wood and
Lass, 1985; Duncan et ah, 1989; Offem, 1990; Ardhana, 1992).
Cultivar has been observed to affect the following properties of cocoa beans:
a) Pigmentation: Forastero seeds are a deep purple colour due to the

presence of anthocyanins, which are absent from white Criollo beans.
Pods from hybrids will generally contain purple seeds, along with the
occasional white criollo type seeds (Roelofsen, 1958; Wood and Lass,
1985).
b) Polyphenols: Criollo cocoa beans contain 2/3 the amount of polyphenolic
compounds as Forastero type (Nazaruddin et al., 2006).
c) Protein: Criollo cocoa beans have been reported to contain up to half as

much protein as Forastero beans (Hardy and Rodrigues, 1952; Zak and
Keeney 1976a). Offem (1990) reported that different Forastero hybrids
(amelonado, amazon and trinitario) did not significantly vary in protein
content from one another.
The maturity, or ripeness, of the cocoa pods also affects the composition and
properties of the cocoa beans obtained from them. Specifically, as the pods
mature, the following changes occur to the beans:
In the seed, the concentration of sugars and citric acid decrease, while
starch, fat, protein, theobromine and caffeine levels increase
(Schmeider, 1976; Zak and Keeney, 1976; Lehrian and Keeney, 1980a
Bucheli et al., 2001).
In the pulp, the concentration of glucose and fructose increase, while

sucrose and citric acid decrease (Reineccius et al., 1972; Ardhana, 1992;
Hashim et al., 1998).
If cocoa pods are allowed to continued to ripen past full maturity, the water,
sugar and pectin concentrations decrease further, while the pulp pH
37
increases(Biehl et al., 1989). Experiments by Biehl et al., (1989) indicated that

post-harvest pod storage allows these maturation processes (reduction in
amount of pulp and sugar content, increase in pulp pH) to continue. A practical
implication is that slightly under-mature pods can be matured by storing for a
period of 5 -10 days (Barel, 1987; Biehl et al. 1989; Schwan et al, 1995). However,
pods should not be kept for more than 2 weeks after harvest, or senescence and
decay processes will occur, and the beans will become unsuitable for
fermentation (Aroyeun et al, 2006).
Finally, the composition of cocoa beans (pulp and seed) can vary as a result of
different cultivation practices, regional effects such as soil types, weather and
degree of shading. Hardy and Rodrigues (1952) noted that the protein content
of cocoa beans was higher for trees grown in soil with higher organic content.
Offem (1990) observed that the mineral content of cocoa beans reflected the
micro-nutrient balance of the soil, and that cocoa grown under shade had a
higher protein content than cocoa grown without shade.
Research into factors affecting composition of fresh cocoa beans is relatively

scarce, and tends to be more than 10 years old. This is in contrast to other
fermented foods. For example, in wine research, there are several journals
dedicated solely to viticulture, with thousands of articles published on the
factors affecting composition of the grapes prior to fermentation (Pretorius,
2000; Fleet, 2001; Keller et al., 2008; Morlat and Symoneaux, 2008).
38
2.4.2 Chemical and biochemical changes occurring during cocoa

bean fermentation
The quality and flavour of a batch of cocoa beans depends on complex chemical
and biochemical changes, which occur in the cocoa beans during fermentation
and drying. Figure 2.11 outlines the significant changes and indicates how they
relate to one another.
PULP
Pulp substrates Microorganisms Secondary microbial
metabolites
Proteins Sugars
ayat
Polyphenols Enzymes
■■■ ### Embryo
Disrupted cells Polyphenols
/Polyphenol
Invertase / oxidase
oxidised
Peptides and Reduc| sugars polyphenols
amino acids 3 3 ^
Figure 2.11 - Summary of biochemical changes occurring during cocoa fermentation.

39
As shown, microorganisms grow in the pulp and produce a diversity of

metabolites, along with substantial heat. The metabolites and heat diffuse into
the cocoa seeds, killing them and disrupting their cellular integrity. Cellular
disruption and seed death initiate various enzymatic and non-enzymatic
reactions between the seed components. These reactions develop a range of
flavour pre-cursors (peptides, amino acids, reducing sugars and polyphenols),
and also affect the colour of the cocoa beans. During roasting, the flavour
precursors undergo further transformations to form the final chocolate flavour
compounds. Because the endogenous reactions are triggered and controlled by
microbial fermentation of the pulp, unfermented cocoa beans do not develop a
characteristic chocolate flavour and aroma, even after roasting (Thompson,
2007; Afoakwa et al., 2008).
The first proper studies of chemical changes during cocoa fermentation were
performed nearly a century ago and focussed on the development of ethanol
and organic acids in the pulp (Preyer, 1913; Knapp, 1926). The following
sections summarise current knowledge of the chemical and biochemical
changes that occur during cocoa fermentation, and how these changes
contribute to the flavour of the cocoa beans.
2.4.2.1 Metabolism of cocoa pulp by microorganisms and diffusion of

metabolites.
During fermentation, microorganisms grow in the cocoa pulp, produce a range

of metabolites and cause changes to the pH, temperature and oxygen levels.
The physiology of specific microorganisms will be described later. The main
chemical changes in the cocoa pulp, caused by microbial growth and
metabolism, are:
- The degradation of pectin by microbial pectinases, causing liquefaction and

drainage of some (10-50%) of the pulp (Knapp, 1937; Sanchez et al., 1984;
Wood and Lass, 1985; Schwan et al., 1997; Da Silva et al., 2005; Outtara et
al., 2008).
- A decrease in the concentration of glucose, fructose and sucrose as they are
consumed by the microorganisms. Most researchers report > 70% reduction
in the level of pulp sugars by 36-48 h of fermentation (Forsyth, 1949;
40
Roelofsen, 1958; Berbert 1979; Carr et al., 1979; Schwan et al., 1995; Camu et
al., 2006; Nielsen et al., 2007b; Galvez et al, 2008).
- A decrease in the concentration of citric acid (30-50% reduction by the end
of fermentation) (Roelofsen, 1958; Quesnel, 1966; Pettipher, 1986; Holm et
al., 1993; Jinap, 1994; Thompson, 2007).
- An increase in the concentration of ethanol, lactic, acetic and, to a lesser
extent, other organic acids as they are produced by microorganisms
(Roelofsen, 1958; Forsyth and Quesnel, 1963; Schwan and Wheals, 2004).
Ethanol and acetic acid tend to reach peak levels during the middle of
fermentation (24-60h), and then decrease again. In contrast, maximum
levels of lactic acid are usually reached at the end of fermentation (Carr et
al., 1979; Schwan et al., 1995; Ardhana and Fleet, 2001; Nielsen, 2007b).
- The production of small quantities of various other substances by
microorganisms. This includes polysaccharides, enzymes, proteins, fatty
acids, and volatile organic compounds (Yoo et al., 1998; Blanco et al., 1999;
Bonvehi, 2005; Da Silva et al., 2005; Thompson et al., 2007; Afoakwa, 2008;
Camu et al., 2008b).
- An increase in the pH of the pulp, caused by the consumption of citric acid
(final pH 4.5 - 6.5) (Rombouts, 1953; Rohan, 1963; Biehl, 1984; Ardhana and
Fleet, 2001; Nielsen et al., 2007b).
- An increase in temperature, due to the exothermic nature of many of the
metabolic processes (maximum temperature 45-52°C) (Rohan, 1963; Minifie,
1980; Wood and Lass, 1985; Schwan and Wheals, 2004; Thompson et al.,
2007).
Table 2.2. summarises the chemical changes to pulp composition observed
during cocoa fermentations in various countries. The fundamental changes
listed above occur in all cocoa fermentations. There are, however, variations in
the amounts of metabolites produced. Comparison of the data in the table
highlights some of these differences. During fermentations in Indonesia and
Brazil, ethanol and acetic acid reached much greater concentrations in the pulp
compared to fermentations conducted in Ghana. The final concentrations of
acetic and lactic acid were similar in all studies.
41
Table 2.2 - Table comparing the key biochemical changes occurring in cocoa pulp
during fermentation in various countries.
Dominican
Country Ghana Brazil Indonesia
republic
Camu et al. Schwan (1998) Galvez et al. Ardhana and

Reference
(2007) (2007) Fleet (2003)
Temperature, initial °C 26 25 21 25
Temperature, final °C 44 45 45 48
Temperature, maximum °C 46 50 51 50
pH, initial 3.7 3.6 4.2 3.8
pH, final 4.3 5.8 4.9 4.8
Ethanol, maximum % w/w 2.30% 4% 1% 6.50%
Ethanol, final %w/w 0.50% <0.01% <0.01% 1.5
acetic acid, maximum mg/g 6 30 22 12
wet basis
(mg/g dry basis in brackets)* (30) (150) (110) (60)
acetic acid, final mg/g 6 10 6 12
wet basis
lactic acid, final mg/g 8 10 10 6
wet basis
citric acid, final mg/g 2 8 5 11
wet basis
Reduction in citric acid (%) 78% 60% 70% 54%
* dry basis quantities estimated assuming water content of 80%, based on data from REFS
The differences between the Indonesian and Brazilian, and Ghanaian

fermentations, were probably caused by different composition of fresh pulp
(more sugars in Brazilian and Indonesian beans), fermentation in boxes instead
of a heap, the more frequent mixing (mixing every 24h, compared to not
mixed), and differences in the microbial ecology. The data above agree with
older literature, which notes that Asian and Brazilian beans tend to be more
acid than West African beans (Wood and Lass, 1985; Ardhana and Fleet, 2003;
Schwan and Wheals, 2004).
The observed differences probably also reflect, to varying extents, differences in
the methods of analysis. For example, the maximum concentration of 1%
ethanol, detected by Galvez et al.(2008), is lower than would be expected for a
maximum acetic acid concentration of 22 mg/g (2.2%). One explanation is that
evaporation of ethanol occurred during sample preparation and analysis. Carr
et al. (1979) and Ardhana (1990) both noted that care should be taken when
42
testing for ethanol, acetic acid and other volatile analytes, to prevent losses
prior to testing.
Many of the microbial metabolites produced in the pulp rapidly diffuse into the
cocoa seeds. In particular, the concentrations of ethanol, lactic and acetic acid
increase in the cocoa seeds during fermentation (Liau, 1976; Carr et al., 1979;
Biehl, 1984; Ardhana and Fleet, 2003; Thompson, 2007; Camu et al., 2008b). The
influx of organic acids causes the pFI of the seeds to decrease from 6.5 to
between 4.5- 5.8. Some authors suggest that other products of microbial
metabolism, such as enzymes (Yusep et al., 2002; de Brito et al., 2004) and
volatile compounds (Yoo et al., 1998; Afoakwa et al., 2008) also diffuse into the
seeds during fermentation. Table 2.3 compares the chemistry of cocoa seeds
obtained from fermentations conducted in Ghana and Indonesia.
Table 2.3 - Comparison of key biochemical changes occurring in cocoa seeds during
fermentation in Ghana and Indonesia.
Country Ghana Indonesia
Reference Camu et al. (2007) Ardhana and Fleet (2003)
pH, initial 6.5 6.4
pH, final 5.2 5.1
Ethanol, maximum % w/w 1% 2.80%
Ethanol, final %w/w 0.50% 1.60%
acetic acid, maximum mg/g wet basis 6 25

(mg/g dry basis in brackets)* (10) (42)
acetic acid, final mg/g wet basis 6 25

(mg/g dry basis in brackets)* (10) (42)
lactic acid, final mg/g wet basis 3 4

(mg/g dry basis in brackets)* (5) (6.7)
citric acid, final mg/g wet basis 2.5 4

(mg/g dry basis in brackets)* (4.2) (6.7)
Reduction in citric acid (%) 58% 55%
* dry basis quantities estimated assuming water content of 40%, based on figures from REFS
Higher concentrations of ethanol and organic acids were found in the

Indonesian seeds, compared to the Ghanaian seeds. As would be expected, this
corresponded to the differences noted in the pulp (C.f. Table 2.2). Due to the
scarcity of data, there is a need for more thorough research linking changes in
the chemistry of the pulp to changes in chemistry of the cocoa seeds.
43
As will be seen, the production and diffusion of metabolites into the seed,
particularly ethanol, and acetic and lactic acids are critical for optimum flavour
formation. However, excess acids can negatively impact flavour and quality
(Wood and Lass, 1985; Schwan and Wheals, 2004; Thompson et ah, 2007). Aside
from interfering with enzymatic reactions, excess acetic and lactic acids directly
contribute undesirable sour flavours. Much research has been performed to
attempt to alleviate the problem of excess acid production during fermentation
(Liau, 1978; Duncan et al., 1989, Meyer et al., 1989; Jinap and Dimick, 1990;
Holm et al., 1993). Some of the strategies developed to prevent over
acidification will be discussed in the final section of this review.
2.4.2.2 Seed death and cellular disruption

The changes in the pulp, and the corresponding changes in the seeds, initiate
and regulate a number of physical and biochemical processes leading to the
formation of chocolate flavour-precursors. Before these reactions can occur, the
substrates must be made available. In unfermented cocoa seeds, cell walls and
membranes are intact and contain and separate their contents from one another.
The diffusion of microbial metabolites from the pulp, combined with elevated
temperatures, kill the seeds and disrupts their cellular integrity (Forsyth and
Quesnel, 1957; Lopez and Dimick, 1995; Thompson et al., 2001). Although some
reviews confuse it, this process actually comprises two distinct phases.
First, the embryo is killed, primarily due to the toxicity of the ethanol (Lehrian,
1989). The death causes a loss of germinative power. This prevents certain
flavour and quality defects by preventing radicle protrusion, and the
metabolism of lipid and protein stores (Thompson et al., 2007). The role of
germination processes in cocoa are poorly studied, and the interaction between
the embryo and the rest of the seed's metabolism may shed more light on some
of the biochemistry involved in flavour development.
Roughly parallel to the death of the embryo, the cellular integrity of the cocoa
seed disintegrates. Cellular de-compartmentalisation occurs as a result of loss of
cellular regulation by embryo, and the effects of heat, ethanol and acetic acid on
the cellular membranes. One indication this has occurred is that under
microscopic magnification, the purple anthocyanin cells lose definition against
the other cells, as they leak their contents (Biehl et al., 1982a, 1982b, 1982c; Biehl
and Voigt, 1996). At a macro level, cellular breakdown is indicated by the
44
presence of a fluid within the testa, swelling the cocoa bean, and by the seed
becoming softer and more spongy in texture (Biehl et al, 1982a; Camu et ah,
2008a). The breakdown of cellular integrity allows the mixing of various
substrates and enzymes, initiating various reactions, both enzymic and non-
enzymic. If insufficient acetic acid, ethanol or heat are produced during
fermentation, cellular disruption will not occur, these reactions will not be
initiated, and flavour precursors will not form (Biehl et al., 1985; Voigt et al.,
1994a; Thompson et al., 2007; Afoakwa et al., 2008).
2.4.2.3 Enzymatic reactions
Enzyme-catalysed reactions are critical in the formation of chocolate flavour
precursors. The main enzymes of cocoa seeds, namely endoprotease,
carboxypeptidase, invertase, polyphenol oxidase, and glycosidases, are well
documented (Forsyth and Quesnel, 1957; Lopez and Dimick, 1995; Hansen et al,
1998; Thompson et al, 2007). Table 2.x summarizes the essential characteristics
of these enzymes.
Table 2.4 - Summary of the major enzymatic changes that occur during cocoa bean
fermentation.
Enzyme Function Optimal conditions Stability during
fermentation3
Endoprotease Hydrolysis of vicilin-class pH 3.5; 45°Cb Active throughout

globulin protein to fermentation
oligopeptides.
Carboxypeptidase Hydrolysis of oligopeptides to pH 5.8; 45°Cb 85% reduction in

smaller peptides and amino activity after 4 days
acids - formation of flavour
precursors
Invertase Hydrolysis of seed sucrose to pH 4.5a; 37°C for Inactivated after 2

glucose and fructose glucose0; 52°C for days
fructose0
Glycosidase Hydrolytic cleavage of sugar pH 3.8-4.5; 45°Cd Active throughout
group from anthocyanin - loss femrentation
of purple colour
Polyphenol oxidase Oxidation of polyphenols - pH 6; 35.5°C° 95% reduction in

enzymic browning activity after 3 days
Information from aHansen et al. (1998); bVoigt et al. (1994a); cLopez and Dimick (1995); dForsyth
and Quesnel (1957).
It has long been recognised that the formation of characteristic cocoa flavours
are dependent on the activity of endogenous cocoa-seed proteases (Biehl, 1982b;
Biehl and Passern, 1982; Biehl et al., 1985; Biehl, 1990). Several studies were
45
subsequently undertaken to identify and characterize these proteases and their

activity in cocoa seeds (Laloi et alv 2002; Guilloteau et ah, 2005). In vitro and in-
vivo experiments have demonstrated that the combined activities of an aspartic
endoprotease and a carboxypeptidase are critical in the formation of cocoa-
specific aroma precursors during fermentation (Voigt et ah, 1994a, 1994b; Voigt
and Biehl, 1995). These enzymes hydrolyse the vicilin (7S) class-globulins
storage proteins in cocoa seeds into a mixture of hyrophilic peptides and
hydrophobic amino acids (Voigt et ah, 1994a, 1994b; Hansen et ah, 1998). The
aspartic endo-proteases are present at high concentrations in storage vacuoles
in the cotyledon material (Guilloteau et ah, 2005), and are activated during
cellular disruption following seed death. Laloi et al (2002) identified the
monomeric TcAP2 polypeptide as the main aspartic protease of cocoa seeds.
Aspartic endo-proteases actively degrade the globulin proteins into
predominantly hydrophobic oligo-peptides. The carboxypeptidase then
preferentially degrades these oligopeptides into hydrophilic peptides and
hydrophobic amino acids (Bytof et ah, 1995; Yusep and Jiinap, 2002). Research
by Kirchoff et ah, (1989), Voigt et al.(1994a, b), Hansen et ah (2001) and Yusep et
ah, (2002) strongly suggested that strong cocoa flavour develops during
roasting, only where there are sufficient levels of hydrophilic peptides and
hydrophobic amino acids (leucine, alanine, Phenylalanine and tyrosine) present
in the seeds. In this regard, the degradation of cocoa seed vicilin-globulins is a
critical step in the development of chocolate flavour and aroma. The activities
and stability of endoproteases and carboxypeptidases are, however, strongly
pH and temperature dependent (Bytof et ah, 1995; Hansen et ah, 1998). The
aspartic endoprotease exhibits optimal activity at pH 3.5 - 3.8 and 42-47°C,
while the carboxypeptidase at pH 5.5 - 5.8 (Hansen et ah, 1998; Guilloteau et ah,
2005). Under fermentation conditions these enzymes will be moderately active,
even though the endoprotease is outside its pH optima. However, if the beans
become too acidic (pH 4.0 - 5.0), the development of strong cocoa aroma
precursors is strongly inhibited (Voigt et ah, 1994b; Hansen et ah 1998). Because
changes in seed pH are governed by diffusion of organic acids from the pulp,
the overproduction of lactic and acetic acid during fermentation ultimately lead
to inhibition of proteolysis, and thus weak flavour (Biehl et ah, 1985; Yusep and
Jinap, 2002; Schwan and Wheals, 2004; Afoakwa, 2008).
46
Prior to the fermentation, the cotyledons may contain up to 2% sucrose. During

fermentation, nearly all sucrose is hydrolysed by native cocoa seed invertase
into the reducing sugars glucose and fructose (Hansen et ah, 1998). Together
with the peptides and amino acids, these reducing sugars are required in order
for Maillard reactions to occur during roasting (Rohan and Stewart, 1967b;
Afoakwa et al., 2008).
Cocoa glycosidases are also active following the death of the cocoa seed. These
native enzymes hydrolytically cleave sugar groups from the anthocyanins
contained in the cocoa seeds. The main anthocyanins are (name them). This
reaction results in a 'bleaching' or loss of purple colour for the cocoa seeds. This
is one reason why the presence of purple-coloured beans at the end of the
fermentation process may indicate under-fermentation. The loss of
anthocyanins does not contribute significantly to the development of cocoa
flavour compounds (Forsyth and Quesnel, 1957; Lopez and Dimick, 1995;
Afoakwa).
Polyphenols are present in large concentrations of 120-180 mg/g in
unfermented cocoa beans (Kim and Keeney, 1984) and can be classified into
three groups: catechins, anthocyanins, and proanthocyanidins (Wollgast and
Anklam, 2000). During fermentation, cocoa polyphenols are oxidized by
polyphenol oxidase (PPO). PPO is activated as oxygen diffuses into the beans,
following disruption of the polyphenol storage cells. The polyphenols are first
oxidized into o-quinones which then complex with amino acids, proteins, and
flavonoids, resulting in brown, water insoluble tannins (Lehrian and Patterson,
1983; Lopez and Dimick, 1995; Hansen et al., 1998; Schwan and Wheals, 2004;
Bonvehi, 2005; Kyi et al., 2005; Afoakwa et al., 2008). Polyphenol oxidase
achieves its optimal activity at pH 6 and 35.5°C (Lopez and Dimick, 1995). The
activity of the enzyme quickly diminishes during fermentation, probably due to
the increase in temperature and the inhibitory effect of tannins. In one study,
only 6% of the original activity remained after 48h (Hansen et al., 1998).
Residual polyphenol oxidase is completely inactivated when the beans are
dried (Lopez and Dimick, 1995; Hansen et al., 1998; Kyi et al., 2005). Because the
polyphenols are both astringent and bitter, a certain reduction in the level of
polyphenols is required to achieve cocoa beans with a good flavour (quesnel
and Jugmohunsingh, 1970). Additionally, the production of the brown quinone
compounds, contributes to the colour of the cocoa beans.
47
2.4.2.4 Non-enzymatic reactions and the role of drying
A range of non-enzymatic changes occur in the cocoa seeds, contributing to the

development of final cocoa bean flavour and quality.
A proportion of polyphenols are consumed in tanning reactions with the
albumin proteins. This renders them insoluble and reducing the astringency of
the cocoa beans to a palatable level(Forsyth et al., 1958; Quesnel, 1966; Zak and
Keeney, 1976; Misnawi et ah, 2005
In describing the chemistry of cocoa fermentation, most reviews emphasise the
diffusion of substances and heat from the pulp into the seed. It is worth noting,
however, that numerous substances also diffuse out of the seed into the pulp.
Some of these outgoing diffusion processes play important roles in the
development of cocoa bean quality and flavour. For example, diffusion of
(-)epicatechin from the seed into the shell accounts for a significant proportion
of polyphenols losses. Again, this contributes to a reduction in astringency (Kyi
et al., 2005;). Amino acids and peptides may also diffuse out of the seed,
accounting for the losses observed in some studies (Wollgast and Anklam, 2000;
Schwan and Wheals, 2004).
Drying is a critical step in cocoa fermentation, and involves the diffusion of
water out of the seed, into the pulp where it evaporates. The progressive loss of
water is the ultimate cause of termination of all enzymic changes inside the
beans, and microbiological growth outside the beans (Misnawi et al., 2003).
Experiements by Misnawi et al., (2002) in wetting and incubation dired cocoa
beans, suggested that endogenous reactions begun during fermentation will
continue until the water content drops below ~10 % in the seed. In this way,
much of the drying process may be thought of as an extension of the
fermentation, allowing the formation of the maximum level of precursor
molecules as possible. Supporting this theory are studies indicating that slower
drying processes result in beans with a stronger chocolate-character (Duncan et
al., 1989; Jinap et al., 1991; Tomlins et al., 1993; Baker et al., 1994; Fabo rode,
1995; Hashim et al., 1999). Drying also allows the loss of acetic acid, which can
help reduce sour-acid flavour characteristics (de Brito et al., 2000; Nganhou et
al., 2003; Afoakwa et al., 2008).
48
2.4.3 Transformation of precursors to final flavour compounds
The precursor compounds formed during the fermentation and drying stages
undergo further transformations when the beans are processed at the factory. In
particular, the roasting is responsible for the development of compounds from
the precursors that give chocolate its distinct, final flavour.
Has been good reviewed extensively by (Rohan, 1969; Hoskin and Dimick,
1994; Biehl and Voigt, 1996; Dimick and Hoskin, 1999), and more recently by
Afoakwa et al. 2008.
Prior to roasting beans have bitter, acidic, astringent and nutty/beany flavours
(Rohan, 1969). During roasting, the precursors formed during fermentation and
drying undergo changes to form chocoalte flavour and aroma compounds.
One function of roasting is to cause a loss of volatile organic acids, particularly
acetic acid, but not non-volatile acids (citric, lactic, tartaric, succinic, oxalic).
Additionally, roasting can effectively inhibit the enzyme polyphenoloxidase,
which adversely affects the formation of pyrazines(Misnawi et al., 2004b;
Bonvehi, 2005).
The most important set of changes that occur during roasting, however, are the
Maillard reactions, which are critical to the formation of typical chocolate
flavour, and involve complex and inter-related changes (2008). The basic
requirements for Maillard reactions are heating (typically above 100°C),
pH>3.0, reducing sugars (glucose or fructose), and amino groups from
peptides, amino acids or proteins. These precursors are produced during
fermentation and the endogenous reactions within the seed, as already
discussed (Voigt et 1., 2993, 1994). The initial processes involve an amine-
assisted degradation of reducing sugars and the formation of Schiff-
bases(Beckett, 2000; Ramli et al., 2006; Granvogl et al., 2006). Reducing sugars
and amino acids also undergo reactions to form various intermediat
compounds, in particular, the reductones and dehydroreductons (1-DH, 3-DH,
5-DH). Even though the intermediate reactions are complex and poorly
understood, they have a strong effect on the final outcome, and are affected by
pH, temperature, and substrate availability (Ramli et al., 2006; Stark et al., 2006).
These intermediate molecules then undergo Strecker degradations, which are
key to the formation of characteristic chocolate flavours. Strecker degradations
49
involve amino-derivatives (1-DH, 3-DH) being split in to three fragments. The

amine component strongly influences the final flavour formation. For example,
leucine and glucose yield flavour compounds with flavour notes described as
"sweet chocolate;" threonine and glucose yield compounds contributing
"chocolate" flavour notes, while valine and glucose react to give compounds
with "penetrating chocolate" flavour notes (Dimick and Hoskin, 1999).
The final stages of the Maillard reactions are poorly understood, but the
generally accepted features are condensation of the aldol intermediates, and
either cyclization leading to the formation of heterocyclic amines, or simpler
polymerisation leading to dark brown melanoidin pigments. Dimick and
Hoskin (1999) proposed that the relative composition of the precursors in the
beans (quantitative amounts, and relative ratios of the different sugars and
amino acids) affects the composition of flavour compounds produced during
roasting. The particular mix of precursors produced in the cocoa beans during
fermentation favours the formation of pyrazines during roasting. At least 80
different pyrazines have been identified in roasted cocoa beans, although the
concentrations in beans from different sources varies (Counet et ah, 2002; Stark
et ah, 2006).
In addition to the pyrazines, more than 500 flavour-active compounds have
been identified from the volatile and non-volatile fractions - hydrocarbons,
alcohols, aldehydes, ketones, esters, amines, oxazoles, sulfur compounds
(Hoskin and Dimick, 1994; Dimick and Hoskin, 1999; Afoakwa, 2008; Stark et
ah, 2006).
Conching also plays an important role in the chemistry of flavour development.
There is some substantial evidence showing that conching allows a decrease in
volatiles, especially acetic acid, but also volatile pheonls, esters and alcohols
(Hoskin and Dimick, 19841). There is also some evidence to suggest that levels
of 2-phenyl-5-methyl-2-hexenal, furaneol and branched pyrazines increase
during conching, while Strecker aldehydes decrease. In this way, some of the
reactions begun during roasting may continue during conching (Beckett, 2000;
Counet et ah, 2002)
50
2.5 Microbiology of cocoa fermentation
The growth and metabolic activities of a diversity of microorganisms are critical

to the successful fermentation of cocoa beans. The microbial ecology of the
fermentation is quite complex, and involves contributions from yeasts, lactic
acid bacteria, acetic acid bacteria, Bacillus species and filamentous fungi. Clearly
defining the contributions of the various microorganisms involved is important
if any attempt is to be made to improve or industrialise the fermentation
process. This section reviews the microbiology of cocoa bean fermentation.
2.5.1 Sources of microorganisms
Freshly obtained cocoa beans begin to undergo fermentation as they are

removed from the pods and become inoculated with a diverse range of yeast,
bacteria and filamentous fungi. The most commonly listed sources of these
microorganisms include the surface of the cocoa pods, machetes and tools,
workers' hands, the leaves used to wrap heap fermentations, and containers
such as baskets, fermentation boxes,and trays (Maravalhas, 1966; Faparusi,
1974; Jespersen et al., 2005; Thompson et ah, 2007). To a lesser extent,
microorganisms may also enter the cocoa fermentations from environmental air
and soil. Jespersen et ah (2005), using DGGE, conclusively demonstrated that
certain yeast species isolated during cocoa fermentations in Ghana came from
the surface of pods, and from the trays used for fermentation. Many researchers
assert that cocoa pulp is sterile until the pods are opened (Martelli and Dittmar,
1961; Ostovar and Keeney, 1963; Carr et ah, 1980; Schwan, 1998; Thompson et
ah, 2007). Jespersen et ah (2005) however, demonstrated the presence of Candida
krusei and Pichia membranifaciens from intact cocoa pods. There is a need for
studies examining the source of bacterial microflora, as well as further
investigations into the presence of microorganisms in apparently intact pods.
Wood and Lass (1985) noted that the presence of various beetles and flying
insects in fermenting cocoa beans was commonplace in many locations. It is
therefore surprising that insects are consistently overlooked as potential vectors
for the introduction of microorganisms into cocoa fermentations. Nicholson
(1913) recognized fruit flies, Drosophila melanogaster, as capable of carrying
yeasts and inoculating the beans. Aside from this, there are no other references
in the literature to the role of insects in cocoa fermentation.
51
2.5.2 Yeasts
The association of yeasts with cocoa bean fermentation has studied for over 100
years. In the early 20th century, cocoa bean fermentation, which was then called
"sweating," was attributed primarily to the activity of yeasts. Nicholson (1913)
and Loew (1913) identified Saccharomyces theobromae (now classified as
S.cerevisiae), S. ellipsoidens, and S. apiculatus (now classified as Kloeckera apiculata)
as the major yeasts found in cocoa fermentations. These same early workers
also suggested that it was these yeasts that converted the pulp sugars into
ethanol and CO2 during the early stages fermentation. More recently, there has
been a focus on the application of non-cultural, molecular methods, such as
DGGE, to examine the diversity of species present throughout fermentation
(Jespersen et al., 2005; Nielsen et al., 2007b).
2.5.2.1 Growth and diversity of yeasts during cocoa bean fermentation
Over the last 50 years, there have been only about 5-10 significant studies
reporting the isolation and identification of yeast species from various cocoa
producing regions of the world (Rombouts, 1952; Carr et al., 1979; Sanchez et
al., 1985; Lopez and Dimick, 1995; Schwan et al., 1995; Schwan et al., 1998).
Several reviews tabulated the yeast species reported in the older studies
(Fowler, 1998; Schwan and Wheals, 2004; Thompson et al., 2007). Table 2.5 is an
extension of this older material, showing species identified from more recent
investigations of cocoa bean fermentations in Indonesia (Ardhana and Fleet,
2003), Ghana (Jespersen et al., 2005; Nielsen et al., 2007b) and the Dominican
Republic (Galvez et al., 2007). These studies are significant in demonstrating the
diversity of species associated with the fermentation, and showing that some
species (Candida krusei, Hanseniaspora spp., Kloeckera spp., and Saccharomyces
cerevisiae) are consistently found in all fermentations. The table clearly shows
that the same species were isolated from fermentations conducted in widely
different regions, and according to different methods. While assessing
taxonomic diversity is important, so too is obtaining the growth profiles of
individual species. Such data are reported in the studies of Sanchez et al. (1985),
Schwan et al. (1995,1998), Ardhana and Fleet (2003), Jespersen et al.(2005), and
Nielsen et al. (2007b). Comparison of these data reveal a common pattern of
sequential growth and death of yeast species.
52
Table 2.5 - Yeast species isolated from cocoa bean fermentations in different countries
Country Fermentation Details
Yeast species isolated
Fermentation
Main (Predominant species indicated in bold type)
Study method
cultivar
(approximate size)
Heap Candida sp., Hansenula sp., Kloeckera sp.,

Carr et al., Pichia sp., Saccharomyces sp.,
Ghana (500 kg), and Forastero
1979 Saccharomycopsis sp.,
Box (2500kg) Schiiosaccharomyces sp., Torulopsis sp.
C.intermedia, C.krusei, C.parapsilosis,

C.quercitrusa, C.silvicola, C.stellimalicola,
Jespersen et Cryptococcus humicolus, Cr.laurentii,
Heap (100 kg,
al. 2005, Hanseniaspora guilliermondii, Pichia
Ghana 500 kg) and Tray Forastero
Nielsen et al. guilliermondii, Pkluyveri,
(100kg)
2007b P.membranifaciens, Rhodotorula glutinis,
Saccharomyces cerevisiae, Trichosporon
asahii
Candida lactativorus, C. krusei, C. sake,

Ivory Sanchez et C.valida, Kloeckera apiculata, K.cortids,
Box (70 kg) Forastero
Coast al., 1985 K.wickerhamii, Pichia membranaefaciens,
Saccharomyces cerevisiae var. chevalieri.
Candida sp., Debaryomyces sp.,

Carr et al., Box Hanseniaspora sp., Hansenula sp.,
Malaysia Trinitario
1979 (2500 kg) Kloeckera sp., Rhodotorula sp.,
Saccharomyces sp., Torulopsis sp.
Candida tropicalis, C.pelliculosa,

Trinitario
Ardhana and Box C.humicola, Kloeckera apis, K.javanica,
Indonesia and
Fleet, 2003 (1000 kg) K.africana, Rhodotorula rubra, R. glutinis,
Forastero
Saccaromyces cerevisiae
Candida spp. (C.bombi, C.rugopelliculosa,

C.pelliculosa, C.rugosa), Kloeckera
Schwan et
apiculata, Kluyveromyces marxianus,
al., 1995; Box
Brazil Forastero K.thermotolerans, Lodderomyces
Schwan, (200-800 kg)
elongisporus, Pichia fermentans,
1998.
Saccharomyces cerevisiae, S. cerevisiae
var. chevalieri, Torulaspora pretoriensis
Candida krusei, Hansenula anomala,

Pichia fermentans, Pichia
Rombouts, Box membranifaciens, Saccharomyces
Trinidad Trinitario
1952 (100-500 kg) cerevisiae var. ellipsoides,
Schiiosaccharomyces pombe, Trichosporon
pullulans
Hanseniaspora guilliermondii, H.
Dominican Galvez et al., valbyensis, Candida krusei, C.glabrata,
Box (100 kg) Trinitario
Republic 2008 C.zeylanoides, C.inconspicua, Pichia
fermentans, Yarrowia lipolytica
Yeast dominate the microbial ecology in the early stages (0-60 h) of

fermentation, due to the high concentrations of sugars, low pH, and low oxygen
53
availability in the cocoa pulp. It is estimated that 70-90% of the total

microorganisms in the first 24-48 hours of fermentation are yeast (Balasimha,
1983; Schwan et ah, 1995; Ardhana and Fleet, 2003; Nielsen et al, 2007b)During
the first 24-36 h of fermentation, apiculate yeasts (Hanseniaspora / Kloeckera spp.)
rapidly grow to maximum populations of 107- 108 cfu / g. There are still
questions as to why these apiculate yeasts grow and dominate the early stages
of fermentation so consistently. As the concentration of ethanol and temperature
increase, the apiculate yeasts die off, and are replaced by more ethanol and
heat-tolerant species. In many studies, the more resistant yeast include
S.cerevisiae, I.orientalis (C.krusei) and P.membranifaciens. These species reach
maxiumum populations of 106 - 107 cfu/g between 36-48h, and may persist at
these levels until the end of fermentation. In some fermentations, the
temperature of the bean mass exceeds 48°C after approx. 72h, causing even the
thermo-tolerant yeasts to decline rapidly. The decline in sugars, and increase in
acetic and lactic acids acts to prevent any increase in yeast populations after
about 72-84 h.
In addition to these predominant yeast species, a large diversity of other yeasts
grow during fermentation. These are listed in Table 2.5, in normal type, and
include a range of Candida, Pichia, Rhodotorula, Torulaspora, Debaryomyces and
Schizosaccharomyces species. Many of these yeasts were isolated during the first
48h of fermentation, at maximum populations of 103-105 cfu/g. They are usually
transient and die off within 24 h of appearing.
Variations on this common growth pattern have been observed. These
variations are caused by a number of factors, including the method of
fermentation, batch size, bean homogeneity (mixing), and the properties of the
cocoa beans used (ripeness, pod storage, cultivar).
The fermentation methods used, whether cocoa beans are fermented in heaps,
boxes, trays or some other means, has been shown to affect the diversity and
populations of yeast. During a recent study conducted in Ghana, the growth
curves produced for parallel heap and tray fermentations were significantly
different. The heap fermentations were dominated by C. krusei. In contrast, the
dominant species during tray fermentations were S. cerevisiae and P.
membranifaciens. There were also general difference in the population dynamics
54
between the heaps and the tray fermentations, but there were difficult to
characterise clearly (Jespersen et ah, 2005; Nielsen et al, 2007b).
Batch size, and the degree of homogeneity may also affect the development of
yeast during fermentations. In large heaps or boxes (> 500kg), the fermenting
mass of cocoa beans is not homogenous, with the center of the fermenting mass
more limited in oxygen and higher in temperature than the edges. Jespersen et
al. (2005) and Nielsen et al. (2007) observed that maximum populations of
yeasts were reached at 24h, compared with 72h for the outer part. At a species
level, S.cerevisiae had the highest populations in the centre of the heap, while in
the outer parts, P.membranifaciens dominated (Nielsen et al., 2007b). Reasons for
these differences were not proposed, but they seem to reflect the different
species' oxygen tolerance, as well as localised differences in temperature and
concentrations of ethanol and organic acids. The inhomogeneity of yeast
growth during heap fermentations may be one reason why the quality of cocoa
produced by such traditional methods is highly variable.
Mixing frequency, and the effect on cocoa quality, have been reasonably well
researched (Sadi and Samarakhody, 1984; Wood and Lass, 1985; Duncan et al.,
1989; Baker et al., 1994; Senanayake et al., 1997; Hashim, 1998b; Lass, 1999).
None of these studies however, reported how mixing affected the yeast ecology
during the fermentations. In a recent study by Camu et al. (2008a, b) mixing of
heap fermentations was observed to cause accelerated decline of yeast
populations in the fermenting beans. Mixing is often described as a critical
variable in determining cocoa bean quality, and more investigation into its
effects on yeast ecology are warranted.
The effects of the properties of the fresh cocoa beans on the growth of yeast are
also poorly researched. Ardhana and Fleet (2003) compared fermentations
conducted using beans of Trinitario and Forastero cultivars, grown on different
plantations. They also tested the effect of fermenting beans of different maturity
levels. Minor differences in the yeast ecology were noted, but there was
insufficient data to draw firm conclusions as to the effects of pulp composition
on yeast ecology. Again, this is an area requiring greater investigation.
Although much progress has been made in understanding the yeast ecology of
cocoa bean fermentations, research on the yeast ecology of other fermented
products suggests that this ecology is more complex than currently appreciated.
55
First, it is likely that there is significant strain variation within each species as
they evolve throughout the fermentation. For example, it is now known that
multiple strains of S. cerevisiae and other yeast species contribute to each
individual wine fermentation (Fleet, 2003; Fleet, 2007). Differing sensitivities to
ethanol and heat strongly guide the sequential growth and decline of the
different species of yeasts during cocoa fermentation. However the yeast
ecology is certainly affected by additional types of microbial interactions. Firstly
there are yeast-yeast interactions. For example, there may be differences in the
production of, and sensitivity to killer-factors (Magliani, 1997; Fleet, 2003); cell
cell interactions may occur via quorum sensing molecules (Hogan, 2005); or
there may be at work some spatial phenomena that are still the subject of
conjecture (Arneborg et al., 2005). Work by Howell et al. (2006) demonstrated
the significance of yeast-yeast and interactions in affecting wine flavour. A
range of yeast-bacterial interactions can also affect the growth of yeast. In wine
fermentations, some lactic acid bacteria produce substances in addition to lactic
acid that inhibit the growth of yeasts (Fleet, 2003; Alexandre et al, 2004; Fleet,
2007). Similar interactions probably occur during cocoa fermentations, and the
impact of these on the flavour and quality of the cocoa beans deserves greater
attention.
In recent years, molecular methods have contributed significantly to
understanding the yeast ecology of foods and beverages. For example, the use
of culture independent methods such as denaturing gradient gel electrophoresis
(DGGE) has aided in detecting viable-but-non-culturable species. PCR-
sequencing of the D1 /D2 domain of the 26S rDNA gene, and chromosome
length polymorphism, have assisted in obtaining more reliable taxonomic
identifications, and in accurately identified yeast isolates to species, subspecies
and strain levels (Karlin and Altschul, 1990; Altschul et al., 1990; Kurtzmann
and Robnett, 1998; Muyzer and Smalla, 1998; Lopez et al., 2003; Meroth et al.,
2003; Jespersen et al., 2005). With the exception of the recent work by Jespersen
et al.(2005) and Nielsen et al. (2007b), all previous yeast ecological
investigations of cocoa bean fermentations have used classical cultural
methods. These recent studies confirmed that the diversity of yeasts detected by
culture-based examinations closely correlated with the results of non-cultural
methods such as denaturing gradient gel electrophoresis (DGGE). One
exception to this, was that Nielson et al. (2007) were only able to detect
56
populations of the yeast C. zemplinina using DGGE. These studies also

highlighted the heterogeneity of the microbial ecology of cocoa fermentations.
Jespersen et al.(2005) noted that the isolates of C.krusei obtained from Ghanaian
cocoa fermentations were genotypically close to C.krusei strains isolated from
other indigenous african fermented foods. Molecular methods could be useful
tools in clarifying the effects of geographical separation on the diversity of
species found in cocoa fermentations, performed in different countries.
2.5.2.2 Metabolic activities of yeasts during cocoa bean fermentation
The main activity of yeasts in cocoa bean fermentation is the metabolism of

sugars in the pulp into ethanol, and CO2. Yeast species are also responsible for
the breakdown of citric acid, and the production of pectinases, and a vast array
of volatile secondary end products (Nichols, 1913; Loew, 1913; Schwan and
Wheals, 2004; Thompson et al., 2007; Afoakwa et al., 2008).
The production of ethanol by yeasts affects the contributes to the cocoa

fermentation in several ways, including the inhibition of certain microbial
species, the death of the cocoa seeds and their de-compartmentalisation, and as
a substrate for the formation of acetic acid. The production of CO2 also acts as a
selective pressure, providing conditions that favour the growth of lactic acid
bacteria and yeasts, and delaying onset of acetic acid production (Lopez and
Quesnel, 1973a; Carr et al., 1979; Schwan et al., 1995; Schwan et al., 1998;
Ardhana and Fleet, 2003; Nielsen et al., 2007b; Camu et al., 2008b).
Yeasts also produce also produced a wide variety of secondary end products
that include higher alcohols, organic acids, esters, aldehydes, ketones, aromatic
volatiles, sulphur volatiles and volatile amines These probably contribute
significantly to the final flavour profiles of cocoa beans, and chocolate made
from them, but their contributions are not well researched. Yeast ecology has
been correlated to the profiles of volatile compounds produced, for fermented
foods such as wine (Rojas et al., 2001; Swiegers et al., 2005; Howell et al., 2006),
and indigenous African foods (Ouoba et al., 2005). These studies all
demonstrated that the mixture of volatiles compounds produced, and resulting
flavour profiles, were directly influenced by the composition of yeasts present
during the fermentations. Similar research correlating yeast growth with the
presence of specific volatiles, during fermentation and in processed beans,
could lead to increased control over the flavour of cocoa beans.
57
Yeasts are considered to have a role in the degradation of pectinous material in

the pulp. The degradation of pectic substances in the cocoa pulp, decreases the
amount of pulp and increases aeration of the fermenting mass, allowing the
growth of acetic acid bacteria (Sanchez et ah, 1984; Ravelomanana et ah, 1986;
Schwan et ah, 1997). Pectin degradation requires the action of several enzymes,
polygalacturonase, pectin methylesterase, pectin lyase, and pectate lyase, as
described previously (Schwan et al., 1997; Blanco et ah, 1999; Da Silva et al.,
2005). The ability of certain yeasts to degrade pectin in cocoa pulp has been
recognized. A study on fermentations in the Ivory Coast found the yeasts S.
cerevisiae, C. saitoana, K. marxianus, and P. norvegensis to be pectinolytic (Sanchez
et al., 1984). In Brazil, pectinolytic yeasts included K. marxianus, S. cerevisiae var
chevalieri, C. rugopeliculosa, and K. thermotolerans (Schwan et al., 1997). K.
marxianus has been found to be the most active producer of heat stable
endopolygalacturonase and the production is constitutive, although the yeast
cannot grow solely on pectin or galacturonic acid as the carbon source (Schwan
et al., 1995; Schwan and Wheals, 2004).
Finally, some yeasts can utilize and/or produce various organic acids. For
example several Candida and Pichia spp. can both produce and utilize citric acid,
and probably contribute to its metabolism, leading to an increase in pFI from 3.5
to 4.2 (Schwan et al., 1995; Anastassiadis et al., 2002; Schwan and Wheals, 2004).
Organic acids such as oxalic, phosphoric, succinic, malic, and acetic acids may
also be produced in small quantities by yeast, contributing to buffering of the
pulp, and the death of the seeds (Jinap and Dimick, 2990; Schwan and Wheals,
2004).
2.5.3 Lactic acid bacteria
While the possibility of lactic acid production was raised by Preyer (1913)
almost a century ago, the early review of cocoa fermentation by Knapp
(1937)downplayed the significance of lactic acid bacteria. Roelofsen (1958)
isolated rod-shaped bacteria that produced acid, and were later described by
Forsyth and Quesnel (1963) as Lactobacillus fermentii. Studying cocoa
fermentations in Trinidad, Forsyth and Rombouts (1951) incubated plates
anaerobically, resulting in improved recovery of lactic acid bacteria isolates.
Between 24-36 h of fermentation, they observed these isolates reached
populations of 107-108cfu/g, which represented only 20% of the total microflora
58
at that time. This, and their apparent transience, meant that these isolates were
considered unimportant and remained identified until a year later (Rombouts,
1952).
Work by Ostovar (1971) and Ostovar and Keeney (1973) constituted the first
thorough taxonomic study of lactic acid bacteria associated with cocoa bean
fermentations, and included detailed growth curves for each of the main
species of lactic acid bacteria isolated. A realisation that residual lactic acid may
be responsible for flavour defects in Asian and Brazilian cocoa led to increased
attention being given to the role of lactic acid bacteria. Carr et al. (1979), and
Passos et al. (1984) investigated the role of lactic acid bacteria in determining
the quality of cocoa beans produced in in Malaysia and Ghana, and Brazil
respectively.
More recently, a number of studies have applied a range of molecular-methods
to cocoa fermentations, with the aim of further improving understanding of the
genetic diversity, and growth dynamics of lactic acid bacteria in cocoa bean
fermentations (Camu et al., 2007; Nielsen et al., 2007b; Camu et al., 2008a). This
led to a number of discoveries, including the description of several new species.
2.5.3.1 Growth and diversity of lactic acid bacteria during cocoa bean
fermentation
Table 2.6 provides an updated summary of the major taxonomic studies

undertaken in the last 50 years, including those published in the last 2 years.
Based on these data, the species of lactic acid bacteria most frequently isolated
from cocoa fermentations include Lactobacillus plantarum, Lb. fermentum, Lb.
brevis, Lb. casei, Leuconostoc pseudomesenteroides, and Pediococcus acidilactici.
Leuconostoc, Pediococcus, and Lactococcus spp. had previously only been detected
in Brazil and Belize. Nielsen et al. (2007b) recently became the first to report the
involvement of these species in West African fermentations. A variety of other
lactic acid bacteria (Lactobacillus, Weisella, Streptococcus, Leuconostoc,
Brevibacterium, Enterobacter spp.) have been detected during cocoa bean
fermentations, but these species were found only at the start of fermentation (Oh
samples), or were transient and present at low populations (<W cfu/ g).
59
Table 2.6 - Lactic acid bacteria isolated from cocoa bean fermentations in different
countries
Country Fermentation Details
Yeast species isolated
Fermentation
method (Predominant species indicated in bold
Study Main cultivar type)
(approximate
size)
Heap
Carr et al., Lactobacillus collinoides, Lb. mali,
1979 Lb. fermentum, Lb. plantarum.
Box (2500kg)
Lactobacillus fermentum, Lb. plantarum,
Lactococcus lactis, Leuconostoc
Heap (100 kg,
Nielsen et al. pseudoficulneum, Leuc.
Ghana 500 kg) and Tray Forastero
2007b Pseudomesenteroides, Pediococcus
(100kg)
acidilactici
Enterobacter casseliflavus, Lactobacillus

brevis, Lb. fermentum, Lb. plantarum,
Camu et al. Heap (100-1000
Ghana Forastero Leuconostoc pseudomesenteroides,
2007, 2008a kg)
Weissella confusa, W. paramesenteroides
Carr et al., Box

Malaysia Forastero Lactobacillus collinoides, Lb. plantarum.
1979 (2500 kg)
Lactobacillus cellobiosus
Ardhana and Box Trinitario and (Lb.fermentum), Lb. plantarum,
Indonesia Lb. Iiilgardii,
Fleet, 2003 (1000 kg) Forastero
Lactobacillus acidophilus, Lb. brevis,

Lb. .casei, Lb. delbrueckii, Lb. lactis,
Passos et al.,
Brazil Box (100 kg) Forastero Lb. plantarum, Leuconostoc
1984a
mesenteroides, Pediococcus acidilactici,
P. cerevisiae, Streptococcus lactis.
Brevibacteriumm ammoniagenes,
Lactobacillus acidophilus, Lb. bulgaricus,
Rombouts, Box Lb. casei, Lb. fermentum, Lb. lactis,
Trinidad Trinitario
1952 (100-500 kg) Lactococcus lactis, Leuconostoc
mesenteroides, Pediococcus acidilactici,
Streptococcus thermophilus
Dominican Lactobacillus brevis, Lb. paracasei subsp.
Galvez et al. Box (100 kg) Trinitario
Republic casei, Lb. pentosus, Lb. plantarum.
Of the studies listed in Table 2.6, only some monitored the quantitative growth
of individual species throughout the fermentation process (Carr et al., 1979;
Passos et al., 1984; Ardhana and Fleet, 2003). Others monitored the total
populations of all lactic acid bacteria during fermentation and then used DGGE
to estimate the proportion of total lactics that each species represented (Nielsen
et al., 2007; Camu et al., 2008). In comparing the data from these studies, a
common growth pattern becomes apparent. Like yeasts, species of lactic acid
bacteria also occur in sequence, conforming to their adaptation to the changing
60
fermentation conditions. During the first 24-48h, abundant carbohydrates, low

pH, and micro-aerophilic conditions caused by yeast CO2 production provide a
habitat that encourages the growth of the Lactobacillus species. The most
commonly isolated species during this stage were Lb. plantarum and Lb.
fermentum. In most of the studies surveyed, L.fermentum was the dominant
species. The lactobacilli grew most rapidly after 24h of fermentation, and
reached maximum populations of 108-109 cfu/ g by 48-64 h (Ardhana and Fleet,
2003; Schwan and Wheals, 2004; Nielsen et al., 2007b). After 72 h, aeration and
temperature increase, as do the concentrations of ethanol and acetic acid. These
changes caused the populations of lactic acid bacteria to decline. In some cases,
this decline was complete, with no lactic acid bacteria detected after 96 h (Carr
et al., 1979; Thompson et ah, 2007). In Ghana (Nielsen et al.,2007b) and
Indonesia (Ardhana and Fleet, 2003), populations of Lb.plantarum and
Lb.fermentum persisted, at levels of 104-106cfu/ g, until the end of fermentation.
In addition to being affected by changing aeration, temperature, ethanol and
acetic acid, the growth of lactic acid bacteria in cocoa beans may also be affected
by other forms of microbial interaction. For instance, during wine and
sourdough fermentations, the death and autolysis of yeast cells releases
vitamins and other nutrients that stimulate the growth of lactic acid bacteria
(Alexandre et al., 2004; Paramithiotis et al., 2006; Fleet, 2007) Inversely, yeast
may also can inhibit the growth of lactic acid bacteria, although in wine the
relationship is very much strain-dependent, at both the yeast and bacteria
levels. The molecular basis of this inhibition require further study, but seem to
involve yeast production of inhibitory short chain fatty acids, sulphur dioxide,
peptides and proteins (Fornachon, 1968; Dick et al., 1992; Markides, 1993;
Capucho and San Romao, 1994; Lonvaud-Funel et al., 1998; Guilloux-Benatier et
al., 1998). In spite of the complex microbial ecology of cocoa bean fermentation,
these kind of interactions have not been considered.
Certain process variables, such as homogeneity of the bean mass, mixing and
source of beans have the potential to affect the growth dynamics of lactic acid
bacteria during cocoa bean fermentation. Nielsen et al.(2007) noted that
Lactobacillus plantarum was isolated less frequently, and in lower populations,
from the centre of large heap fermentations, than from the surface. Because
heap fermentations are mixed infrequently, it is reasonable that such
inhomogeneity should occur. This study also found that lactic acid bacteria
61
persisted at higher populations in tray fermentations than in heap

fermentations. Camu et al. (2008a, b) observed that mixing appeared to
stimulate the growth of L.fermentum, but did not significantly affect the growth
sequence of the lactic acid bacteria. Ardhana and Fleet (2003) tested the effects
of maturity on the growth of lactic acid bacteria, but their results were
inconclusive. They did observe, however, that Lb.plantarum and Lb.fermentum
survived until the end of fermentations using Forastero beans, whereas during
fermentations with Trinitario beans, no lactic acid bacteria were detected after
72 h.
Recently, molecular methods have been applied to the study of lactic acid
bacteria in cocoa bean fermentations. Two studies, using DGGE to examine the
microbial ecology of Ghanaian cocoa bean fermentations, discovered new
species of lactic acid bacteria, Lactobacillus ghaneiisis (Nielsen et al., 2007a) and
Weissella ghanaensis (Camu et al., 2007). Both of these studies also obtained new
phylo-genetic information on the lactic acid bacteria isolated during
fermentation. This data, combined with phenotypic testing seemed to suggest
that there is considerable strain variation present, particularly amongst
Lb.fermentum. These studies also raise some questions: Are the new species
geographically specific to Ghanaian cocoa fermentations? Are these, or any of
the other strains or species of lactic acid bacteria, specific to to the ecological
niche of fermenting cocoa beans? Are there species of lactic acid bacteria present
in cocoa fermentations in other countries that are specific to those countries?
The application of DGGE and other molecular methods to cocoa bean
fermentations in other countries may help answer these questions
The impact of bacteriophages on many food fermentations involving lactic acid
bacteria, such as cheese, fermented sausages and sourdough bread, has received
much attention (Fiammes and Hertel, 2007). While bacteriophage
contamination has not been an issue during vegetable fermentations
(Buckenhuskes, 1993), no attempt has been made to assess the possibility of
bacteriophage presence during cocoa bean fermentations. Similarly the
production and role of bacteriocins by lactic acid bacteria (Vandenbergh, 1993)
during cocoa bean fermentation also remains uninvestigated. Research into
these areas would not only improve understanding of the microbial ecology, but
would also be assist in attempts to industrialise the fermentation.
62
5.5.3.2 Metabolic activities of lactic acid bacteria during cocoa bean

fermentation
The biochemical roles of lactic acid bacteria during cocoa bean fermentation has
been described by numerous researchers (Passos et alv 1984; Arunga, 1992;
Schwan et alv 1995; Thompson et al., 2007; Camu et al., 2008b). The homo-
fermentative species (Lactobacillus plantarum, Lactobacillus fermentum, Pediococcus
acidilactici, Lactococcus lactis) metabolize Glucose via the Emden-Meyerhof
pathway to yield lactic acid. These species also metabolise lactic acid to a small
extent to produce acetaldehyde and CO2. The heterofermentative species
(.Leuconostoc pseudomesenteroides, Weissella spp.) metabolise glucose by the
pentose phosphate pathway to give lactic acid, acetic acid, glycerol, mannitol
and CO2. Heterofermentative lactic acid bacteria metabolise citric acid to acetic
acid (Vandenbergh, 1993; Wisselink, 2002; Hammes and Hertel, 2007;
Thompson et al., 2007).
The primary end products modify the conditions in the pulp, influencing the
growth of other microbial species as already mentioned. The lactic acid also
diffuses into the cocoa seeds.lowering their pH, and contributing to the
breakdown of cell structure (Biehl et al., 1985; Thompson et al., 2007). If the
fermentation is longer than 5 days, and the populations of lactic acid bacteria
do not decline after 72h, excess lactic acid is often produced. This in turn causes
excess lactic acid in the cocoa seeds, seed pH values of less than 5.0, and an
inhibition of enzymic reactions (Biehl, 1984; Passos et al., 1984; Jinap et al., 1991,
Schwan and Wheals, 2004). Furthermore, because lactic acid is non-volatile,
such an excess is not reduced during drying or conching, and high
concentrations of residual lactic acid will impart a sour flavour to chocolate
(Augier et al. 1998; Afoakwa et al., 2008). This is one reason why dried
fermented beans are tested for pH as an indicator of quality. Mannitol was
detected in large quantities during one recent study, although the role of this
metabolite is not clear (Camu et al., 2008a, b). The breakdown of citric acid by
lactic acid bacteria lowers the pH of the pulp (Hugenholtz, 1993; Thompson et
al., 2007).
Studies of sourdough fermentation using pure cultures of Lactobacillus

plantarum, showed that lactic acid bacteria can also produce a vast array of
volatile aroma compounds. Some key metabolites identified were higher
63
alcohols such as hexanol, esters such as ethyl acetate, ethyl lactate, as well as n-
hexanal and 2-pentyl-furan (Hansen and Hansen, 1994). When the
fermentations were repeated with a combination of S.cerevisiae and L.plantarum,
an even larger array of compounds were formed, emphasising the importance
of microbial interactions in flavour development during food fermentation.
While such interaction studies have been performed in wine (Fleet, 2003), and
cheese (Addis et al., 2001) no studies of a similar nature have been performed
on cocoa bean fermentation.
2.5.4 Acetic acid bacteria
After 48-72 h, fermenting cocoa beans develop a strong vinegar aroma,

indicating the presence of acetic acid (Rohan, 1963; Lehrian and Patterson, 1983;
Wood and Lass, 1985; Lopez and Dimick, 1995; Lass, 1999; Afoakwa et al., 2008).
In the early part of this century, researchers attributed this acetic acid
production to the presence of small rod shaped bacteria, following what they
described as the 'alcoholic' phase of fermentation (Bainbridge and Davies, 1912;
Loew, 1913; Nicholls, 1913). Roelofsen and Geisberger (1947) made the first
taxonomic and quantitative study of the acetic acid bacteria involved in cocoa
bean fermentation. Carr et al., (1979) made a detailed comparison on the
populations and diversity of species in fermentations performed in Ghana and
Malaysia. Recent studies applied various molecular methods to the study of
acetic acid bacteria in cocoa fermentation, and discovered two new
species(Camu et al., 2007; Cleenwerck et al., 2007; de Vuyst et al., 2008).
2.5.4.1 Growth and diversity of acetic acid bacteria during cocoa bean
fermentation
The diversity of acetic acid bacteria found in cocoa bean fermentations, globally,
and within any particular fermentation, is much less than the diversity of yeast
or lactic acid bacteria. Table 2.7 presents an updated summary of the taxonomy
of acetic acid bacteria in cocoa fermentation, as described by major studies over
the last 50 years. In most studies, only one or two species of acetic acid bacteria
were isolated from each fermentation. Across all studies, Acetobacter pasteurianus
was isolated significantly more frequently than any other species, showing a
clear association between this species and fermenting cocoa beans. This species
is also associated with cider fermentation (Passmore and Carr, 1975).
64
Table 2.7 - Acetic acid bacteria isolated from cocoa bean fermentations in different
countries
Country Fermentation Details Yeast species isolated
Fermentation (Predominant species indicated in bold
Main
Study method type)
cultivar
(approximate size)
Heap
Carr et al., Acetobacter rancens, A.ascendens,
1979 A.xylinum, Gluconobacter oxydans.
Box (2500kg)
Heap (100 kg, 500

Nielsen et al. Acetobacter malorum, A. pasteurianus,
Ghana kg) and Tray Forastero
2007b A. syzigii, A. tropicalis, G. oxydans
(100kg)
Camu et al. Heap Acetobacter ghanensis, A. pasteurianus,

Ghana Forastero
2007, 2008b (100-1000 kg) A. senegalensis
Acetobacter rancens, A. lovaniensis,

Carr et al., Box A. xylinum, Gluconobacter oxydans.
Malaysia Forastero
1979 (2500 kg)
Trinitario Acetobacter aceti, A. pasteurianus.

Ardhana and
Indonesia Box (1000 kg) and
Fleet, 2003
Forastero
Acetobacter aceti subsp. Liquefaciens,
Schwan and A. pasteurianus, A. peroxydans,
Brazil (200-800 kg) Forastero
Wheals, 2004 Gluconobacter oxydans subsp.
su boxydans.
Acetobacter aceti, A. roseus, A.

Rombouts, sub oxydans.
Trinidad Box (100-500 kg) Trinitario
1952
Dominican Galvez et al.

Box (100 kg) Trinitario Acetobacter lovaniensis
Republic 2008
Various other Acetobacter spp., along with Gluconobacter oxydans were isolated
less frequently. As with the lactic acid bacteria, a few of the studies presented in
the table quantitated individual species during fermentation (Carr et al., 1979;
Ardhana and Fleet, 2003), while others estimated what proportion of the total
count each species comprised (Nielsen et al., 2007b; Camu et al., 2008a, b). There
were similarities between the growth data that all of these studies presented,
allowing a general growth curve to be described.
At the beginning of fermentation, the initial populations of acetic acid bacteria

(< lCPcfu / g beans) are much lower than the initial populations of yeast or lactic
acid bacteria. In most fermentations, one species dominated, often
A.pasteurianus (Ardhana and Fleet, 2003; Nielsen et al., 2007b; Camu et al.,
2007), but also A. rancens (Carr et al., 1979), A.aceti (Schwan et al., 2004) and
A.louaniensis (Galvez et al., 2008). This dominant species rapidly increased in
population after about 24-48 hours, reaching maximum populations of 106-107
65
cfu/ g by 48-72h. This growth corresponded to increased aeration due to

drainage of the pulp, increased ethanol from yeast growth and metabolism, and
increasing temperatures. From around 72h onwards, the temperature of many
fermentations exceeded 45°C. Acetic acid bacteria exhibit optimal growth at
30-35°C and are not tolerant to high temperatures (Drysdale and Fleet, 1988;
Kersters et ah, 2007; Thompson et ah, 2007). Consistent with this, the dominant
acetic acid bacteria died off rapidly once the temperature of the fermenting
beans exceeded 45-50°C (Carr et al., 1979; Ardhana and Fleet, 2003; Nielsen et
al., 2007b). In a few studies, acetic acid bacteria were isolated until the end of
fermentation, at populations of between 103 -106 cfu/g (Nielsen et al., 2007;
Galvez et al., 2008). In most studies, apart from the single dominant species,
additional species of acetic acid bacteria were only isolated at the beginning of
fermentation, or transiently. In some cases, however, a second species grew
parallel to the dominant species, at populations 10 to 100 fold less (Carr et al.,
1979;Ardhana and Fleet, 2003; Nielsen at al., 2007b).
Most of the studies did not determine how processing factors, such as batch
size, mixing, or fermentation method affect the growth and diversity of acetic
acid bacteria in cocoa bean fermentations. One exception was Camu et al.
(2008a, b), who reported that mixing of heap fermentations accelerated the
growth of acetic acid bacteria, with higher maximum populations reached 12-24
h earlier than in un-mixed heaps. This in turn led to increased acetic acid
production in the mixed heaps.
Across a series of studies on Ghanaian cocoa bean fermentations, two novel
species, Acetobacter ghanensis and A. senegalensis, were discovered. The
development and application of the (GTG)5-rep-PCR fingerprinting technique
allowed for rapid classification and identification of acetic acid bacteria isolates
obtained in these studies (Camu et al., 2007; Cleenwerck et al., 2007; de Vuyst et
al., 2008). Nielsen et al.(2007b) used DGGE as a rapid means of assessing the
total microbial ecology of fermenting cocoa beans, without having to perform
extensive culture-based examinations. These molecular methods had clear
advantages in the case of studying acetic acid bacteria, which are notoriously
difficult to maintain in the laboratory (Cleenwerck et al., 2002; Kersters et al.,
2007).
66
2.5.4.2 Metabolic functions of acetic acid bacteria during cocoa bean

fermentation
Acetic acid bacteria oxidize ethanol to acetic acid, which diffuses through the
testa, causing the death of cocoa beans, the disruption of cellular integrity.
Acetic acid also lowers the pH of the seed (Schwan et ah, 1995; Schwan and
Wheals, 2004; Ardhana and Fleet, 2003). These reactions are exothermic, and
raise the temperature of the cocoa beans high as 52°C. This temperature
increase, along with the production of acetic acid, exerts a selective pressure
against all microorganisms. Only some spore forming bacteria may survive
these conditions. The increased temperature further contributes to cell
disruption and the triggering of endogenous biochemical reactions inside the
cocoa seeds (Schwan et al., 2004; Thompson et al., 2007; Afoakwa et al., 2008).
Acetic acid bacteria can also oxidise compounds in the pulp, and those
produced during fermentation to form a range of aldehydes, ketones and other
volatiles (Kersters et al., 2007). The affect of microbial interactions on the
formation of these secondary products, and their impact on cocoa flavour, are
poorly understood.
2.5.5 Bacillus species

Aerobic, spore-forming bacteria, dominate the microflora at the end of
fermentation process, when selective pressures have eliminated many other
species. The exothermic biochemical reactions of acetic acid bacteria heat the
fermenting mass up to 50°C (Schwan et al., 1995). Oxidation of acetic acid
increases the pH to 5.0. As most of the pulp has drained out, oxygen tension
increases. Species of the spore forming genus Bacillus are well suited to these
conditions and dominate the microflora at the end of fermentation process.
Preyer (1913) appears to be the first to suggest that Bacillus spp. may play a role
during the latter stages of cocoa bean fermentations. The growth of Bacillus
species may be inhibited by organic acids, and ethanol produced during
fermentation. They may also be inhibited by bacteriocins produced by the lactic
acid bacteria (Vandenbergh, 1993).
Examples of isolated species include Bacillus stearothermophilus, B. cereus, and B.
megaterium in Trinidad (Ostovar and Keeney, 1973); B. subtilis, B. lichenformis,
67
and B. coagulans in Brazil (Schwan et al., 1986); B. pumilus, B. licheniformis, B.

subtilis, B. cereus, and B. coagulans in Indonesia (Ardhana and Fleet, 2003).
The production of tetramethylpyrazines by B. subtilis has been confirmed
recognized by in vitro studies (Kosuge and Kamiya 1962; Adams and de Kimpe,
2007), as well as field studies correlating increasing concentrations of
tetramethylpyrazines with increasing populations of B. subtilis during
fermentation (Zak et al., 1972;REFS). Bacillus spp. have also been shown to
produce a range of other volatile compounds during cocoa fermentations,
including C3-C5 free fatty acids and 2,3-butanediol, which contribute to
undesirable hammy notes in over-fermented beans (Schwan et al., 1986; Perego
et al., 2004; Schwan and Wheals, 2004). Outtara et al. (2008) recently
demonstrated that a large number of Bacillus spp. found in cocoa fermentations
produce pectinases and probably contribute to the liquefaction of the pulp
during fermentation. This study also emphasised the need to better investigate
the functional properties of the microorganisms previously thought to play a
minor role in fermentation.
2.5.6 Other bacteria
A range of other bacteria grow through the course of the fermentation process,
although they do not dominate the microflora. The elevated temperature of the
fermenting mass at the end of fermentation process particularly favours the
growth of thermotolerant bacteria, such as Streptococcus thermophilus. However,
little attempt has been made to isolate and characterize these bacteria from
cocoa bean fermentations. Examples of identified species are Zymomonas mobilis
in Brazil (De Camargo et al., 1963) and Trinidad (Ostovar and Keeney, 1973);
Aerobacter aerogenes, Arthrobacter oxydans, Micrococcus conglomerates, M. roseus,
Staphylococcus aureus, S. capitis, M. kristinae, and Pseudomonas cepacea in
Indonesia (Ardhana and Fleet, 2003). Usually, these bacteria have only been
isolated from the beginning of fermentations (0-12h), or transiently and at low
populations. Ardhana and Fleet (2003) found that some strains of Staphylococus
and Pseudomonas spp. isolated from Indonesian fermentations were lipolytic
and proteolytic. The contribution of many of these bacteria to the cocoa
fermentation remain both unclear and un-investigated.
The presence of Zymomonas mobilis in cocoa fermentations has been given some
68
attention, due to this bacterium's ability to efficiently ferment sugars into

ethanol (De Camargo et ah, 1963; Ostavar and Keeney, 1973, Ardhana, 1990).
2.5.7 Filamentous fungi
Many recent reviews and research papers give little or no attention to the role of
filamentous fungi during cocoa fermentation. Yet the association of filamentous
fungi with cocoa bean production has been recognized since the 1930's
(Bunting, 1928; Dade, 1928; Knapp, 1937). Several studies have reported their
isolation and identification from freshly extracted beans (Rombouts, 1952;
Roelofsen, 1958; de Camargo et al., 1963; Maravalhas, 1966; Ogundero, 1983;
Schwan et al., 1995). Aspergillus, Mucor; Penicillium and Rhizopus spp. are most
commonly isolated at this stage. One reason these species are often considered
to have little positive contribution to bean fermentation is because they usually
die off within 12 h of the start of fermentation. The inability of many of these
species to survive during the fermentation is due to their aerobic metabolism,
low ethanol tolerance and sensitivity to low pH and heat (Schwan and Wheals,
2004; Schwan et al., 2005; Ardhana and Fleet, 2003).
Nevertheless, the presence of fungal mycelia on the surface of fermenting beans
cocoa beans is regularly observed during fermentation in many regions (Rohan,
1963; Wood and Lass, 1985). Furthermore, a small number of papers report the
growth of some fungal species in the bean mass, during fermentation, grow in
the well-aerated parts of the fermenting mass (Schwan et al., 1995; Ardhana and
Fleet, 2003; Mounjouenpou et al., 2008).
In Indonesia, Penicillium citrinum, Aspergillus versicolor, and A. zventii, grew to
106-107 cfu/ g between 0-36h (Ardhana and Fleet, 2003). After 36 h, the
populations of these species declined, which was attributed to an increase in
temperature to above 40°C. In Ghana, the most commonly detected species of
filamentous fungi was Ceratocystis paradoxa, which was found during the first
48h of fermentation, at levels of 106 cfu/g (Nielsen et al, 2005). Filamentous
fungi may also grow in the fermenting cocoa beans towards the end of
fermentation, if the temperature falls below about 40°C. For example, Schwan et
al. (1995) isolated low populations (103-104cfu/g) of various species between
72-160h of fermentation. The species detected included A. fumigatus, A. niger,
Lasiodiplodia theobromae, Fusarium moniliforme, F. oxysporum, Mucor racemosus, P.
citrinum, and Mycelia sterilia.
69
The impact of filamentous fungi on cocoa fermentation has been only

sporadically studied. Ardhana and Fleet (2003) examined the enzyme activities
of several fungal isolates and found that many strains of P. citrinum had
pectinolytic and proteolytic activities, while strains of A. wentii were both
amylolytic and proteolytic (Ardhana and Fleet, 2003). Schwan et al. (1995) also
reported on the pectinolytic acitivities of several species of fungi isolated from
cocoa. Most of the attention given to filamentous fungi in cocoa fermentation
relates to their negative impacts. For example, many species of mould produce
carbonyls which contribute hammy flavour notes, and anisoles which impart
musty and mouldy aromas (Schwan et al., 1995; Misnawi et al., 2003; Afoakwa
et al., 2008). The most well researched aspect of filamentous fungi in cocoa
beans is the ability of many species to produce mycotoxins. Amezqueta et al.
(2004, 2005) and Bonvehi (2004) reported high concentrations of Ochratoxin A in
cocoa samples. More recently, Mounjouenpou et al. (2008) positively confirmed
that growth of Aspergillus niger and A. carbonarius during cocoa fermentation
was accompanied by the production of significant levels of Ochratoxin A. There
is some evidence to suggest that simply shelling the cocoa beans significantly
reduces levels of Ochratoxins (Amezqueta et al., 2005). It would be preferable to
control the growth of the causative microorganisms. The only effective control
measure currently available is to ensure the fermented cocoa beans are properly
and promptly dried, once fermentations ceases and the temperature of the
beans drops. There is a need to develop simple and effective measures for
preventing mould growth during fermentation.
The interactions of filamentous fungi with other microorganisms during cocoa
bean fermentation has not been studied. This is in contrast to other foods,
including cheese (Fleet et al., 2007), coffee beans (Masoud and Latoft, 2006) and
tempeh (Nout and Kiers, 2005). By understanding these interactions, these
studies were able to improve product flavour and quality, or develop starter
cultures that can inhibit mycotoxigenic moulds.
70
2.6 The improvement of cocoa bean quality - progress and

opportunities
Market trends have fueled overall demand for cocoa beans. At the same time,
much greater attention is being paid to the quality of the cocoa beans being
produced worldwide. Over the last 50 years, much of the research into cocoa
bean fermentation, drying and processing has been aimed at solving certain
quality or flavour problems. This section outlines the progress that has been
made in improving cocoa quality, focussing on the role of fermentation, and to a
lesser extent, drying. Opportunities to further improve cocoa bean fermentation
are noted, along with suggestions about how these opportunities might be
addressed.
2.6.1 The industrialisation and improvement of cocoa bean

fermentation
Most of the world's cocoa beans production remains uncontrolled and largely
unchanged from how it was performed centuries before. This is in contrast to
other fermented foods such as bread, wine, cheese, tempeh and natto, which
have all been successfully industrialised (Hesseltine and Wang, 1986; Beuchat,
1995; Wood, 1998; Fleet, 2001; Steainkraus, 2004; Nout and Kiers, 2005).
Industrialisation, by improving control of these fermentations, has allowed
higher quality products to be obtained more consistently. The industrialisation
of any fermented food requires that three main factors be understood and
controlled: the substrate and its properties; the physical processes or unit
operations carried out; and the microbiology of the fermentation (Hesseltine &
Wang, 1986; Steinkraus, 2004). With regards to cocoa fermentation, some of
these areas have been well addressed, while there remains significant gaps in
others.
2.6.1.1 Improved approaches to the substrate - selection, cultivation,

harvesting, storage of cocoa beans
The composition and properties of a food substrate may strongly affect the
fermentation process, ultimately determining many of the properties of the
fermented food (Hesseltine and Wang, 1986; Wood, 1998; Steinkraus, 2004).
There are numerous opportunities for further research into the effects of fresh
71
(unfermented) cocoa bean composition on fermentation and final cocoa bean

quality.
Firstly, the composition of fresh cocoa beans are strongly affected by genetic
factors. There are a large diversity of genetic subtypes recognised to exist within
Theobroma cacao, and via the application of molecular methods to track
variations, new hybrids are constantly being developed (Figueira and
Alemanno, 2005; Schnell et al., 2005; Efombagn, 2006). The majority of genetic
research on Theobroma cacao, however, focusses only cataloguing genetic
variation within cocoa growing regions, and screening new hybrids for
desirable agronomic properties, particularly yield and disease resistance
(Gotsch, 1997). In contrast, much research has gone into understanding the
effects of genetic variation of grapes on wine flavour and quality (Fleet et al.,
2001). A recent study by Assemat et al.(2005) aimed to investigate the effects of
genetic variation in Guyana and Africa on cocoa flavour and quality. They
found that cocoa beans of different varieties had predictably different flavour
chacracteristics. Improved understanding of the flavour characteristics of
individual cocoa hybrids would allow the producer to choose a fermentation
suited to that variety.
The composition of unfermented cocoa beans will also be affected by their
growing conditions (soil, climate, shade, pests). There have been no adequate
investigations into the effects of these things on the fermentation, or the final
flavour or quality of cocoa beans.
A limited number of studies have assessed the affect of bean maturity on
fermentation and cocoa quality (Packiyasothy et al., 1981; Ardhana and Fleet,
2003), but the data from these studies was relatively inconclusive.
The impact of post harvest treatments on fresh cocoa beans, and the effects of
these treatments on fermentation and final bean quality have been well
investigated. Three basic processes have been evaluated for the treatment of
fresh cocoa beans prior to fermentation - pod storage, mechanical depulping
and enzymatic depulping (Rohan, 1963; Wood and Lass, 1985; Schwan and
Wheals, 2004). All three of these treatments were developed or investigated in
attempts to reduce the problem of acidity in dried fermented cocoa beans.
Overacidity in processed cocoa beans has been linked to the production of high
levels of lactic and acetic acid during fermentation. By removing a portion of
72
the pulp, or reducing the fermentable sugar content of the beans, it has been
shown that less acid is produced during fermentation, leading to less acid beans
(Duncan et al., 1989; Sanagi et al., 1997). Removal of up to 20% of the cocoa pulp
from fresh Brazilian cocoa beans significantly improved the flavour quality of
the beans produced (Schwan and Wheals, 2004). Methods for mechanically
depulping fresh cocoa beans include presses (Rohan, 1963; Wood and Lass,
1985), centrifuges(Schwan et al., 1995), or simply spreading beans onto a flat
surface for several hours prior to fermentation (Biehl et al., 1990). Alternatively
pectinases may be added to the fresh beans prior to fermentation, causing a
significant increase in the 'sweatings produced in the first 24 h of fermentation.
In addition to reducing acidity, benefits of de-pulping include shorter
fermentations and increased efficiency, and the ability to use the excess pulp in
the manufacture of jams, marmalade, pulp juice, wine or cocoa soft drink
(Buamah et al., 1997; Schwan and Wheals, 2004; Dias et al., 2007).
Storage of cocoa pods before the beans are removed for fermentation can also
be beneficial to fermentation outcomes (Sanagi, 1997). It has been shown that
upon storage, the pulp volume per seed decreases, due to water evaporation
and inversion of sucrose (Biehl et al., 1989) and the total sugar content is
diminished, reducing acid production during fermentation. The flavour quality
of Malaysian beans were improved by pod storage for up to 12 days prior to
fermentations (Barel. et al., 1987; Duncan et al., 1989; Lass, 1999; Aroyeun et al.,
2006)
2.6.1.2 Improved process design and control

The physical processes, configurations and treatments applied to a fermenting
food material will affect the microbial ecology of the fermentation, it's
chemistry and the final qualities of the food. Most studies aiming to improve
the quality of cocoa beans have focussed their attention on such processing
variables. Each of the main processing steps or variables examined by major
studies are briefly described and evaluated below.
a) Mechanisation of harvest and pod-splitting
Because industrialisation of cocoa fermentation would proved the ability to

process large amounts of beans with lower labour costs, a few studies have
been conducted to examine the possibility of mechanisation of pod harveting
(Lemin, 2005b; Wan Ishak, 2008) and pod splitting and bean-removal
73
(Maduako, 1994; Puri, 2008; Wan Ishak, 2008). Due to the unusual fruiting and
delicate nature of cocoa trees, mechanical harvesting has proved to be
impractical so far.
b) Fermenter design
The most commonly used means for containing cocoa bean fermentations are
boxes, or banana-leaf wrapped heaps. Because of the inhomogeneity of these
systems, neither are particularly suitable for controlled fermentations. Similar
fermenters are described by Wood and Lass, (1985) although in that case, no
comment is made as to their efficacy. The main criteria for the fermentation
vessel is that it should contain the beans, insulate them to some degree against
heat and water loss, and protect them from the elements and vertebrate pests. It
should also be easy to clean, be robust, easy to move or operate with minimum
labour and easy to repair with locally available materials (Steinkraus, 2004).
c) Mixing and aeration
Of all the variables in the cocoa fermentation process, mixing has received the
most attention. The earlier papers attempting to improve cocoa fermentations
all included a consideration of the frequency of mixing (Rohan, 1963; Biehl et
al., 1977; Carr et al., 1979; Said and Samarkhody, 1984; Passos et ah, 1984b;).
This is partly because of the link between aeration, mixing and the development
of acetic acid. All of these early studies emphasised the need to optimise mixing
freqency in order to ensure the fermentation produced beans witha good
chocolate character. More recently, the findings were confirmed by Hashim et al.
(1998) and Camu et al., (2008a, b) who found that mixing less frequently
introduces less oxygen into the fermentation and results in lower concentrations
of acetic acid. Hashim et al. (1998) equated this lower acid to higher
concentrations of pyrazines in dried and roasted beans. Camu et al. (2008a, b)
equated lower acid with stronger chocolate character, and less objectionable
flavours in chocolate made from the cocoa beans that were not mixed during
fermentation. Baker et al.(1994) also noted a postive correlation between less
frequent mixing and stronger cocoa flavour. However, Senanayake et al. (1997)
found that when cocoa bean fermentations in Sri Lanka were not mixed, off
flavours developed. These results suggest that mixing frequency may be a
critical process variable, but may need to be optimised in each situation. It also
suggest that more research should be done to better understand the inter-
74
relationships between mixing, aeration, microbial ecology, acid production and

flavour development.
d) Time/temperature control
The fermentations must proceed for a minimum amount of time, and must
reach > 45°C for at least 2 days in order to develop a good flavour (Shepherd,
1976; Dougan et ah, 1981. These conditions are required in order for the
endogenous enzymatic changes to be triggered (Biehl et al., 1977). One
application of this is that temperature changes may be used as rough indicators
of fermentation index. For example, when comparing fermentation methods for
use with Criollo beans, Portillo et al. (2005) used an temperature sustained
above 46°C for 2 days to indicate complete fermentation.
One difficulty in studying these process variables is the large variation in
fermentation methods used. This makes comparison between the different
studies difficult. Nevertheless, Rohan et al.,(1963), Wood and Lass (1985) and
Baker et al. (1994) suggested that all forms of cocoa fermentation practices
shared some common features, and that with variables like mixing the same
general rules apply in all cases, building on these principles, Barel et al. (1995),
designed an integrated fermenter-dryer, in a horizontal cylinder configuration.
Aside from this small number of works, there is a severe lack of any published
research into new fermentation vessels or configurations.
2.6.1.3 Controlled inoculation for cocoa bean fermentation
Pod storage, de-pulping, fermenter design, and mixing frequency all indirectly
affect the microbial ecology of fermentation and thus the quality of cocoa beans
produced (Said and Samarakhody, 1984; Biehl, 1989; Barel, 1995; Thompson et
al., 2007; Afoakwa et al., 2008). However, it is also important to consider how
the microbiology of fermentation may be controlled more directly. Improved
understanding of the microbiology of cocoa bean fermentation have led to the
first investigations into the use of starter cultures for cocoa bean fermentation.
Some of these involve replacement of the natural fermentation by a somewhat
artificial process, which is carried out with defined microbial inocula.
Most attempts to improve cocoa bean fermentation with defined microbial

cocktails have placed emphasis on the use and selection of yeasts. Sanchez et al.
(1985) fermented cocoa beans by pure culture yeasts, including Saccharomyces
75
chevalieri (now classified as S. cerevisiae var chevalieri), Candida zeylanoides, and

Kluyveromyces fragilis (now classified as K. marxianus). Each yeast culture was
inoculated at a population of 5xl06 cfu/g of cocoa beans which were obtained
from previously washed cocoa pods. The fermentations were conducted for 7
days with frequent mixing, after which the beans were sun-dried for 15 days.
The study indicated that S. cerevisiae var. chevalieri rapidly degraded the pulp
and effectively fermented the beans, and the chocolate made from cocoa beans
fermented by this strain was comparable in quality with chocolate from
traditionally fermented beans (Sanchez et al., 1985).
Buamah et al. (1997) and Dzogbefia et al. (1999) used pectinolytic yeasts K.
fragilis, Candida norvegensis (now classified as Pichia norvegensis), Torulopsis
Candida (now classified as Candida saitoana), and S. chevalieri in controlled
fermentations of cocoa beans in sterile funnels under laboratory conditions.
Cocoa pods were washed with detergent and sterilized together with
equipment by UV irradiation. S. chevalieri and K. fragilis both significantly
increased the yields of cocoa sweatings, which could be commercialized as jam,
marmalade, or syrup (Buamah et al., 1997). Furthermore, the inoculation did
not alter the physico-chemical properties of the degraded pulp. The exclusion of
acetic acid bacteria from the fermentation also resulted in low levels of acetic
acid (Dzogbefia et al., 1999). The quality of fermented beans was evaluated by
anthocyanin content and acidity of dried cotyledons. Beans with lowest
anthocyanin content were those fermented with S. chevalieri, either by itself or
in combinations with other yeasts. Interestingly, K. marxianus inoculated
fermentations resulted in beans with poorer in quality, as they had higher
anthocyanin content than beans from other treatments (Buamah et al., 1997).
Schwan (1998) conducted cocoa bean fermentations with a defined microbial

inoculum of yeasts and bacteria, which contained the yeast S. cerevisiae var
chevalieri, and the bacteria L lactis and L. plantarum as lactic acid bacteria, and A.
aceti and G. oxydans as acetic acid bacteria. The inoculum was added either
together as a cocktail at time zero or sequentially at different times during the
fermentation process. The phased addition was found to be less successful than
zero-time inoculum, as the former resulted in poorer chocolate. As for zero-time
inoculum, the microbial succession and speed of change in their populations
were found to closely follow those of the natural process. The yeast S. cerevisiae
76
var chevalieri was chosen for its fermentative ability and pectinolytic activity.
Another active and constitutive producer of pectinases is K. marxianus, and it
has been suggested that this species might be used instead of S.cerevisiae var.
chevalarii (Schwan, 1998) An important factor in chocolate quality was the
duration of fermentation process, which was suggested to be a full 7 days
(Schwan, 1998).
The major difficulty associated with inoculated fermentations is the removal of
natural flora. In small-scale laboratory fermentations, this could be achieved by
washing the beans with hypochlorite treated water and using sterilized
equipment, as applied by Schwan (1998). However, as on-farm fermentations
usually involve large amounts of beans, bean washing could be a laborious and
ineffective process and require intensive training of farmers.
Nevertheless, inoculated and controlled fermentations are expected to improve
the reliability and quality consistency of cocoa fermentations. Furthermore, they
could greatly reduce the possibility of contamination and growth of undesirable
microorganisms (Schwan, 1998). This would improve not only the quality but
also the shelf-life of fermented cocoa beans. An objective for further studies
could be production of chocolate with different flavour characteristics by
varying microbial inocula or fermentation conditions.
Schwan (1998) found that beans of acceptable quality could be produced, but
did not compare the effects of different organisms. Sanchez et al.(1985) found
that using K.marxianus gave beans with lower cut test scores, but did not
examine the effects of mixtures of yeast and did not perform any other quality
testing. Finally, Buamah et al. (1997) found that fermentation with different
mixtures affected the anthocyanin content of the beans, but failed to suggest
any possible reasons for the differences.
Sanchez et al. (1985) provided some data regarding yeast - yeast interactions in
their research on the use of single yeast cultures to perform cocoa fermentation.
However, there are no data available in the literature regarding the interactions
between the inoculated yeast and the indigenous bacteria.
77
2.6.2 Total supply chain management and integrated approaches to

the improvement of cocoa quality.
While the concepts of Quality Assurance, and total supply chain management
are not new, they have only recently begun to be applied to cocoa bean
fermentation and the cocoa supply chain. Part of the reason for this is that many
large chocolate manufacturers wish to have more control and predicability over
their suppliers (Beckett, 1999; ICCO, 2007c, 2007d). It is also related to the trend
of increased processing of cocoa beans by the growing countries. It is now
estimated that as much as 30% of cocoa beans are processed further in the
countries of origin (ICCO, 2007d)
Over the last 5-10 years there have been increased co-ordinated efforts to aid
cocoa growers obtain a better, more consistent product to sell. This includes
European aid and development projects by EU-INCO "COCOQual"(Nielsen et
al. 2007) and CIRAD (CIRAD, 2008).
In addition, there have been numerous projects investigating cocoa quality
management, funded by international Cocoa bodies such as the ICCO (ICCO,
2007d) and WCF, and industry (Camu et al. 2007, 2008a, b).
In addition to these general initiatives, a small number of studies on cocoa
fermentation have attempted to link together the microbiology, and chemistry
of fermentation, with the flavour and / or quality outcomes of those
fermentations (Carr et al., 1979; Schwan et al., 1998; Ardhana and Fleet, 2003).
These studies are important in developing effective quality assurance
approaches, by clarifying the relationships between changes to the substrate,
the process, and the microorganisms involved. These integrated studies were
used as models upon which this current study was designed
2.6.3 Barriers to industrialisation and improvement of cocoa

fermentation
Economic factors are the single greatest cause of low industrialisation and
improvement of cocoa bean fermentation (Schwan et al., 1998; Schwan and
Wheals, 2004; ICCO, 2008). However, some sources predict that as demand for
cocoa beans continues to grow, the higher prices will result in better farming
practices and subsequent increases in yields (ICCO, 2007a). As trade
78
liberalization continues, a key factor in better prices for farmers will be the
ability to produce higher quality cocoa beans (ICCO, 2007d).
With an estimated 70% of the world's cocoa produced by small-holders, there is
a basic lack of capital amongst producers to implement new technologies
(Anon., 1999; Guyton, 2007; ICCO, 2007a). FOr example, special fermentation
vessels may require significant cost to build, while the use of starters can be
expensive. It seems likely that improved approaches to cocoa fermentation will
be first applied to large plantations or co-operatives where sufficient beans are
processed to allow some economies of scale, and to justify some significant
capital expenditure (Schwan et al., 1998).
Another very significant barrier to the implementation of better practices in
cocoa fermentation is inadequate access and transfer of information. There is a
huge diversity of research has been done on cocoa publications tend to be
scattered in a huge range of literature, of varying quality and relevance. In
addition to the various abstracted journals, this frequently includes provincial
publications, institute reports, trade journals, proceedings of regional
conferences devoted to broad topics, and non-scientific literature such as
newspapers, correspondence and travel reports. A key consequence of these
communication issues is that technology transfer is limited, haphazard or even
purely a matter of chance. In addition, the majority of farmer education in cocoa
processing is geared towards phyto-sanitary and horticultural practices. There
are only a handful of programs that recognize the significance of fermentation
and drying in impacting the effective value of a cocoa crop, such as the one
established in Vietnam with the assistance of the World Cocoa
Foundation(Guyton, 2007).
2.7 Conclusion
The final quality of cocoa and chocolate is the results of a number of processing
steps, each governed by a combination of factors. In the first place, cultivation
of cocoa requires good agricultural practices and suitable climate conditions.
Furthermore, appropriate cocoa bean curing and processing methods are of
crucial importance in the sensory quality of cocoa and chocolate.
The fermentation process has a key role in flavour quality of cocoa beans and
chocolate. The linkage between microbial ecology of fermented beans and the
79
flavour compounds has been well established in the literature. The microbial
succession during fermentation creates changes in pH, temperatures, oxygen
availability of fermenting mass, which facilitates the formation of cocoa-specific
aroma precursors. The variety and concentrations of these precursors
compound is determined by microbial activity. If the control of microflora and
conditions of fermentations comes into commercial practice, it is possible that
chocolate with desired flavour characteristics be economically produced.
The literature however has some gaps that were apparent.
Firstly, there is absolutely no information regarding the chemistry or
microbiology of cocoa beans cultivated in Australia. Such information is critical
if fermentations are to be successfully industrialised.
Second, there is insufficient, or contradictory, information available on ideal
frequency of mixing, length of fermentation, preferred fermentation
configuration, and the effects of these on final cocoa bean quality. Such
information as exists does not provide a clear guide for the establishment of a
cocoa fermentation operation from first principles. Furthermore, there is scant
research into the design of alternative fermentation vessles, and
implementation of industrialised approaches to cocoa bean fermentation.
Third, few researchers have provided good comparisons between model, or
laboratory, fermentation of cocoa beans, and larger pilot, or commercial
fermentations. In particular, many researchers assess model or trial
fermentations only according to chemical or microbiological criteria, without
regard for the effects on flavour.
Finally, certain forms of controlled inoculation have been under-researched. In
particular, the addition of mixtures of yeast only, without addition of bacterial
species, and without attempting to remove indigenous microflora.
These areas were therefore highlighted as priorities to be investigated in this
research.
80
Chapter Three
The microbiological, chemical and sensory
properties of Australian cocoa bean fermentations.
3.1 Introduction
Cocoa beans are the raw material of chocolate manufacture and are obtained
from the fruit (called pods) of the tree, Theobroma cacao L. After being harvested,
the beans are removed from the pods and subjected to the process of
fermentation, followed by drying (Wood and Lass, 1988). Cocoa fermentation is
a microbiological process that generates conditions which kill the beans and
induces an array of endogenous and exogenous biochemical reactions. These
reactions produce the chemical precursors of characteristic chocolate flavour
and colour (Beckett, 2000). Cocoa research performed during the last century
has described aspects of the microbiology, biochemistry and chemistry of cocoa
bean fermentations and attempted to link this knowledge to attributes of cocoa
and chocolate quality (Schwan and Wheals, 2004). Despite significant progress
being made in understanding the scientific foundations of cocoa bean
fermentation and chocolate production, many gaps in knowledge exist and
cocoa bean fermentation remains a traditional, village scale process (Thompson
et al. 2007). This lack of technological improvement and innovation, is in sharp
contrast with progress made in production of other fermented foods and
beverages, such as cheese, yoghurt, wine and beer; in these foods, scientific
understanding has led to the development of modern, well controlled and
industrialised processes (Wood, 1998; Steinkraus 2004).
Recently, it has been established that cocoa beans can be successfully cultivated
in Northern Queensland, Australia and this finding has provided an incentive
to develop a linked cocoa-bean and chocolate industry in that region (Lemin,
2004). Accompanying this opportunity is the need to develop controlled,
efficient, industrialised processes of fermentation and drying that give high
quality cocoa beans which are suitably priced for sale in international markets.
Currently, there is no baseline scientific information on cocoa beans cultivated

in Queensland, and it is not known if these beans can be fermented to produce
beans that give chocolate of an acceptable quality.
81
This chapter investigates the fermentation of Queensland cultivated cocoa

beans, with the goal of producing beans that give acceptable chocolate quality.
Traditional heap and box methods are compared with that of a novel method
conducted in a barrel fermenter. The frequency of bean mixing, and the
addition of microbial nutrients to the process, are examined.
Baseline information on the microbial ecology of the fermentation, and the
chemical and physical changes that occur in the cocoa pulp and seeds
throughout fermentation are reported, and correlated with attributes of bean
and chocolate quality.
82
3.2 Materials and Methods

3.2.1 Fermentation of Cocoa Beans
During the period from November 2003 to June 2005, a series of experimental
cocoa fermentations were performed at the Department of Primary Industries
(DPI) Research Station, South Johnstone, Queensland, Australia. This research
station was well equipped, with facilities for cocoa fermentation, drying of
cocoa beans and basic microbiological testing. The fermentation studies at
South Johnstone were conducted during several visits to the station, each being
approximately 2 weeks in duration. A range of samples were collected during
these field trips and they were transported back to the University of New South
Wales for further analysis.
3.2.1.1 Source of cocoa beans for fermentation.
The cocoa beans used in the fermentation experiments were harvested from two
small plantations located in northern Queensland, Australia (Figure 3.1). The
first plantation was located near Mossman (Figure 3.2), while the second
plantation was located onsite at the South Johnstone research station (Figure
3.3). Although the two plantations were 175 km apart, the overall seasonal and
climatic influences were similar.
The cocoa trees grown at these two plantations were Trinitario hybrids, with the
majority of trees bearing Trinitario type pods. The Trinitario type pods were
warty, elongate and red-violet when unripe, becoming yellow-orange when ripe
(Figure 3.4, b). Because of the genetic variation within hybrid planting material,
about 30% of the trees produced Forastero type pods. These pods were smooth,
melon shaped and light green when unripe, becoming yellow when ripe (Figure
3.4, a). Both varieties of pods contained approximately 30 cocoa beans, the
average weight of which was 3g (±lg).
Cocoa pods of both cultivars were manually harvested using secateurs from
trees at both plantations, and were not separated into individual cultivars for
fermentation. Within 48 hours of harvesting, the pods were transported to the
research station, where they were stored for 5 days at ambient temperature (25 -
32°C). A period of pod storage of between 5-10 days is recommended prior to
fermentation, as a means of improving the quality of the cocoa produced [Wood
and Lass, 1985; Jinap and Dimick, 1990].
83
Pacific Ocean
Mossman
(76 km)a .Cairns
'South Johnstone
2 (101 km)a
Kilometres
QueensTand
0 100
200
a Distance from Cairns _Tronic Qi CQ&riCi
Fig. 3.1. Map of Northern Queensland, Australia; Location of the cocoa plantations
involved in this study.
Fig. 3.2. Cocoa plantation located at Mossman, Queensland, Australia.

84
Fig. 3.3. Cocoa plantation located at South Johnstone, Queensland, Australia. The tall
trees in the background are Gliricidia sepium shade trees.
Fig. 3.4. Variation in cocoa pods borne by the Australia cocoa trees: a) Forastero type
pods; b) Trinitario type pods
85
A mechanised pod splitter-sorter was used to rapidly and conveniently remove

the cocoa beans from the pods. This machine was developed by the Queensland
Department of Primary Industries, QLD Australia, and its proprietary design is
protected by patent (Puri, 2007). Mouldy, damaged, under-mature or small
pods were manually removed prior to splitting. Unsorted fragments of placenta
or shell were manually removed from the cocoa beans after mechanical splitting
and sorting (Figure 3.5). The design of the automatic pod splitter/sorter meant
that the exterior of the pods intermingled with the fresh pulp for up to 5
minutes before the pulp exited the machine. The maximum time between the
removal of beans from the pod and the commencement of a fermentation was
90 minutes.
Fig. 3.5. Freshly obtained Australian cocoa beans (from Mossman pods). Note the
pieces of placenta and broken pod present as a result of the automatic pod splitting
process. These were manually removed prior to fermentation
The fresh beans obtained by these methods were subject to different

experimental fermentations, in order to determine how certain processing
variables affected the microbiology, chemistry and final quality of the cocoa
beans.
86
3.2.1.2 Fermentation of cocoa beans in heap, box and barrel
Batches of cocoa beans from the Mossman plantation, were fermented using
heap box and barrel methods.
For heap fermentation, beans (75 kg) were placed in a cone shaped heap on top
of a layer of banana leaves and then covered with more banana leaves. The
heap had a radius of 50 cm and a height of 40 cm. (Figure 3.6, a). Beans were
mixed once during fermentation, at which time the heap was unwrapped,
mixed on the spread banana leaves and then rewrapped using fresh leaves.
For box fermentation, beans (75kg) were placed in a wooden box of dimensions
50 cm long x 50 cm wide x 50 cm deep, and the top covered with clean plastic
sheeting to prevent heat and moisture loss. The box was made of untreated
hardwood slats, with 5mm gaps to allow aeration of beans and drainage of the
pulp (Figure 3.6, b). In the box, beans were thoroughly mixed using sterile
gloves and a sterile wooden paddle rinsed with ethanol. Particular care was
taken to redistribute the beans at the edges of the fermentation.
For barrel fermentation, beans (75 kg) were placed in a cylindrical barrel of 65
cm diameter x 85 cm height, and covered with a fitted lid to prevent heat and
moisture loss. The barrel was made of food grade polypropylene and was
mounted on a steel frame to allow it to rotate end-over-end. For aeration and
drainage of pulp, the barrel was perforated with 1 cm diameter holes, 15 cm
apart on the sides and 5 cm apart on the base. (Figure 3.6, c). In the barrel, beans
were mixed by closing the lid and tumbling the barrel, end over end, 10 times to
ensure thorough redistribution of the beans inside the barrel.
The beans were fermented by these methods for 5 days (120 hours), and were
mixed every 24 hours, except for the heap which was mixed every 48 hours.
To prevent sudden changes in ambient conditions, the fermentations were
conducted in an insulated room where the ambient temperature remained
between 30 - 35°C and the relative humidity 70 - 90%. These controls were not
applied to the heap fermentation which was conducted in the field.
The temperature of each fermentation was monitored using electronic
temperature recorders (Gemini Data Loggers, UK) that were inserted into the
centre of the bean mass.
87
Fig. 3.6. Heap (a), box (b) and barrel arrangements used in cocoa fermentation experiments.
3.2.1.3 Addition of microbial nutrients to enhance fermentation.
To investigate the effects of nutrient addition on cocoa bean fermentation,

Fermaid K® (Lallemand, S.A.S France) was added to selected fermentations.
Fermaid K® is a food grade additive that is widely used in the fermentation of
wine and other alcoholic beverages to fortify the levels of assimilable nitrogen
and vitamins present in the fermentation substrate.
One hundred kilogram (100 kg) quantities of beans from the Mossman
plantation were fermented in barrels as described previously. Three
fermentations were conducted simultaneously, thus:
(i) Control (no Fermaid K® added)

(ii) Fermaid K® added at rate of 25g/ 100 kg fresh cocoa beans
(iii) Fermaid K® added at rate of 50g/ 100 kg fresh cocoa beans
88
A similar experiment was performed using beans harvested from the South
Johnstone plantation, but only the control (no Fermaid K® added) and Fermaid
K® added at 25g/100 kg were conducted in this case.
3.2.1.4 Effect of mixing on the fermentation of Australian cocoa
Two fermentations (75 kg of beans) were performed in boxes as described
previously, using beans harvested from the South Johnstone plantation, and
mixed at different intervals:
(i) every 24 hours, or
(ii) every 12 hours.
The boxes were mixed using sterile gloves and a sterile wooden paddle, at the
prescribed frequency. Care was taken to ensure that beans at the edges, corners
and bottom of the boxes were thoroughly redistributed.
3.2.1.5 Sampling of fermenting cocoa beans for microbiological and
chemical testing.
From the beginning of fermentation, at time zero (Oh), samples of the cocoa
beans were taken for microbiological and chemical analyses.
Samples (1 kg) of beans were taken from the bean mass at daily (24h) intervals
throughout the fermentations. Samples were aseptically taken by hand using
sterile plastic gloves, from several different locations in the beans mass and
combined to give a representative sample. A portion (250g) of the sample was
immediately used for microbiological analysis, and the remainder (750g) was
frozen (-20°C) for later chemical analyses.
3.2.1.6 Drying of fermented cocoa beans
After each fermentation, the beans were dried by the following methods to
ensure a final moisture content of between 5-7%:
(i) Solar drying. In this method the beans were spread at a depth no greater
than 2 cm on large mesh trays (100 cm x 200 cm) in the sun, and solar
dried for 4-5 days. At night or during intermittent showers, large canopies
were used to protect the drying beans from precipitation (Figure 3.7, a).
(ii) Oven drying. Fermented beans were spread approximately 5 cm deep on
trays ( 35 cm x 75 cm), and placed in a temperature controlled, forced air
oven for 48h at 38°C, and then 60°C for another 48h. (Figure 3.7, b)
89
The end point of drying was judged by bean breaking properties as described in
Wood and Lass (1985). The dried beans were bagged in airtight plastic bags and
stored in a cool, dark room free from strong odours. Samples of dried fermented
beans were transported to UNSW to be made into chocolate for sensory
evaluation, while other samples were sent to the Cadbury-Schweppes quality
assurance laboratories in Singapore for quality evaluation.
Solar drying was the preferred method for drying of the fermented beans.
However, in rainy and humid weather it was sometimes necessary to directly
dry the beans in the oven, or partly dry them under solar conditions and then
complete the process in the oven. The drying conditions used in each
experiment are given in the results section and appropriate controls were
conducted to determine any effect of drying on the quality or sensory
properties of the beans.
Fig. 3.7 Drying arrangements used: a) Solar drying trays arranged outside. A large
sliding canopy was able to be placed over the trays overnight, or during rain;
b) Oven drier, showing one of four drying compartments, with removable shelves
containing the cocoa beans.
90
3.2.1.7 Summary of fermentation experiments
Table 3.1 summarises the experiments performed, the source of cocoa beans and
the drying method(s) used, and the dates the pods were harvested.
Table 3.1 - Matrix of experiments investigating the effects of processing variables on the
microbiology, chemistry and quality of Australian cocoa fermentation.
Date Experiment description Source of beans Drying method(s)

(date harvested)
May 2004 Effect offermentation vessel
Trial 1: Fermentation in heap, box and Mossman Oven drying only

18/5/04 to 23/5/04 barrel vessels. (10/5/04)
Trial 2: Fermentation in box and ban-el South Johnstone i)Solar drying;

24/5/04 to 29/5/04 vessels. (17/5/04) ii) Oven drying
November 2004 Effect of adding microbial nutrients
Trial 1: Og, 25g or 50g of Fermaid K® Mossman i) Combined drying ;

15/11/04 to 20/11/04 added per 100 kg of cocoa beans (6/11/04) ii) Oven drying
Trial 2: Og or 25g of Fermaid K.® added South Johnstone i)Solar drying;
22/11/04 to 27/11/04 per 100 kg of cocoa beans (14/11/04) ii) Oven drying
June 2005 Effect of mixing frequency
14/6/05 to 20/6/05 Fermentations mixed every South Johnstone i) Combined drying ;

24 h, or 12h. (6/6/05) ii) Oven drying
Figure 3.8 outlines the range and sequence of microbiological, chemical,

physical, quality and sensory analyses performed on samples of cocoa beans
obtained during the fermentation experiments.
91
Fig. 3.8. Flow chart of procedures for testing the microbiological, chemical and
sensory properties of cocoa beans fermented at South Johnstone, Australia.
DPI research station, South Johnstone, Australia
Harvest and storage

of cocoa pods
Splitting of Pods
Fermentation Temperature
Sampling
External lab
(Singapore)
Frozen and
Microbiological
transported to Quality
analyses
UNSW testing Testing
Food science laboratories, UNSW Sydney, Australia
Additional
microbiology Dried, fermented cocoa
bean samples
Fresh (frozen) cocoa
bean samples
Quality testing
Physical HPLC analysis:

Chocolate making
analyses: • Sugars
• % H20 • Ethanol
• Bean:pulp • Organic acids
ratio Sensory evaluation
92
3.2.2 Physical Measurements of Fermenting Cocoa beans

3.2.2.1 pH of cocoa beans
The pH of pulp and seed samples extracted from the beans were measured
according to the method of Ardhana and Fleet (2003). The pulp and seeds were
manually separated from 30 beans. The pulp or seed material was mixed in a
Waring Blender (Waring products, Torrington CT) with distilled water at a ratio
of 1:1 by weight. The pH of the homogenate was measured using an Activon
209 digital pH meter (Activon, Sydney). Standard buffers of pH 4.0 and 7.0
were used to calibrate the pH probe.
3.2.2.2 Pulp and seed weights
The pulp-bean ratio was calculated by the following procedure: 50 g of whole

cocoa beans were weighed, then the pulp was manually separated from the
seed and weighed. The specific weight of the beans and pulp were used to
calculate the pulp-bean ratio. In the calculation of the ratio, the testa is included
as part of the pulp.
3.2.2.3 Moisture content of beans, pulp and seed
The moisture contents of beans, pulp and seeds were determined according to
standard methods described by AOAC (1997). Empty aluminium dishes and
lids were oven dried for 4h at 110°C, cooled in a desiccator over silica gel, and
then weighed. Five grams of sample (seed or pulp) were spread uniformly in
the prepared dish, the lid replaced and then weighed. The dishes, with lids ajar,
were the dried in a vacuum oven (Thermoline) for 24 h at 70±1°. After drying,
the dishes were cooled in a desiccator over silica gel, before being reweighed.
3.2.3 Microbiological analyses

3.2.3.1 Isolation and enumeration of microorganisms in cocoa
fermentations
Isolation and enumeration of microorganisms from the fermenting cocoa beans

were conducted on location at the research station in South Johnstone,
Queensland, Australia. Beans (250g) with attached pulp were suspended in
250mL of sterile 0.1% peptone water (Oxoid, Melbourne, Australia) in a plastic
Stomacher bag (Seward Lab Systems, London UK). The suspension was
vigorously shaken and massaged for 5 minutes to form a uniform homogenate.
93
Five mL of the resulting supernatant was then diluted into 45 mL of sterile 0.1%
peptone water (10-1). Subsequently, l.OmL samples of the homogenate were
subject to serial, decimal dilutions as required in 0.1% peptone water. Aliquots
(O.lmL) of the dilutions were then spread inoculated, in duplicate, over the
surface of appropriate plates of agar media as follows:
Yeasts and other fungi were specifically isolated and enumerated on plates of
Malt Extract Agar (Oxoid) containing 100 ppm of added oxytetracycline
(Sigma) to control bacterial growth, and on Dichloran Rose Bengal
Chloramphenicol Agar (Oxoid). The plates were incubated at 25°C for 4-5 days.
Total bacteria were isolated and enumerated on Plate Count Agar (Oxoid,)
supplemented with cycloheximide (Sigma) at 100 ppm to suppress the growth
of yeasts and fungi. These plates were incubated at 35°C for 3-4 days.
Acetic acid bacteria were isolated and enumerated on Wallerstein Laboratories
Nutrient Agar (Oxoid) containing 100 ppm of cycloheximide (Pallmann et al.,
2001). These plates were incubated at 30°C for 4-5 days.
Lactic acid bacteria were enumerated and isolated on MRS agar (Oxoid)
containing cycloheximide at 100 ppm . This medium was also used to
successfully isolate acetic acid bacteria. These plates were incubated at 35°C for
3-4 days under micro-aerophilic conditions achieved by use of a candle-jar.
The plates were incubated under the specified conditions until prominent
colonies were observed. Colonies were presumptively grouped according to the
media on which they were isolated, cellular and colony morphologies, and the
results of the Gram stain, catalase and oxidase tests. The predominant colony
types were thus recorded and are described in Tables 3.2 - 3.4. Populations were
determined by counting each colony type. Representative samples (3-5 of each
colony type obtained from each sample) were taken and purified by re
streaking. More than 1000 isolates were obtained in this manner. The collected
isolates were transported to the University of New South Wales, Sydney,
Australia, where they were re-streaked to verify purity. Purified cultures were
maintained by regular subculture on appropriate growth media and storage
with 30% glycerol at -80°C.
Figure 3.9 outlines the main steps in the isolation, enumeration and
identification of microbial species in the cocoa fermentations.
94
Fig. 3.9. Flow chart for microbiological testing of fermented cocoa beans
1 kg sample from
fermentation >t
Isolates transported to
University of New South
Wales, Sydney
250g of beans
in 250mL
0.1% peptone -80°C
Purification
water storage
Broth API
culture test kit
i 5m L of
homogenate DNA
into 45mL extraction
0.1% peptone
water y
PCR and
sequencing
Serial dilution to 10 6
Colony morphology
Counting Cell morphology
> & Gram stain
0.1 ml_ spread onto selection Catalase test
media in duplicate of isolates Oxidase test
95
Table 3.2 - Presumptive grouping of yeast isolates. Isolates were grouped according to their colony
morphology on MEA and DRBC at 30°C, and cell morphology under phase contrast light microscopy.
Isolate Colony morphology Cellular morphology Identification*

Number (Malt extract agar - brown; (Phase contrast light
DRBC medium - pink) microscope)
1
Hanseniaspora
guiUiermondii
(Confirmed by growth at
37°C)
Shiny, off-white umbonate Lemon shaped yeast,

colonies; 5-10 mm diameter, distinctive bottle or bowling
pin shape.
Matte, off-white, flat, Rod or block-like yeast,

irregular filamentous edged occasionally forming
colonies; 10-20 mm pseudo-hyphae
Shiny, white, rounded Rounded ovoid shaped yeast,

colonies; 5-10 mm. multi-polar budding
Matte, off white, round, Slightly elongate, ovoid

raised filamentous colonies; yeast
5-10 mm.
* Representative samples of the isolates were identified at UNSW to species level by PCR-sequencing of
26S rDNA and confirmed by API ID32 carbohydrate profile kit.
96
Table 3.3 - Presumptive grouping of lactic acid bacteria isolates. Isolates were grouped according to their
colony morphology on MRS agar under micro-aerophilic conditions at 37°C, their cell morphology under
phase contrast light microscopy, and their reaction to Gram’s stain, catalase and oxidase tests. Tentative
lactic acid bacteria were Gram positive, catalase negative and oxidase negative.
Isolate Colony morphology Cellular morphology Identification*

dumber (MRS agar - brown) (Phase contrast light
microscope, 100 x)
Translucent, shiny brown, Rounded ovoid cells, in pairs

domed, round colonies; and short chains.
2-5 mm diameter
Shiny off-white, round, Straight rods with mainly in

colonies; 2-10 mm pairs
diameter.
Small translucent, shiny, Short straight rods in pairs

off-white, pointed colonies; and short chains.
2-4 mm diameter
Small shiny, rounded bright Cocci, in pairs and tetrads

white colonies; 1-3 mm
16S rDNA and confirmed by API CH50 carbohydrate profile kit.
97
Table 3.4 - Presumptive grouping of acetic acid bacteria and Bacillus spp. isolates. Isolates were grouped
according to their growth on WLNA and MRSA under aerobic conditions at 37°C, their cell morphology
under phase contrast light microscopy, and their reaction to Gram’s stain, catalase and oxidase tests.
Tentative acetic acid bacteria were Gram negative, catalase positive and oxidase negative. Tentative
Bacillus spp. were Gram positive, catalase positive and oxidase positive.
Isolate Colony morphology Cellular morphology Final identification*

Number (On WLNA medium - green; (Phase contrast)
on MRSA medium - brown)
Round shiny, translucent Short rounded rods in pairs and

domed colonies; 2-5 mm individually.
Small round shiny colonies Small rounded rods in pairs,

with dark centres; 1-3 mm
No photo available No photo available Asaia siamensis
Very small, round domed, Very small cocci-bacilli in pairs,

shiny, pale pink colonies; 1-2
mm diameter
4
Pantoea agglomerans
No photo available
Shiny yellow, irregular, flat Small thin straight rods,

colonies; 2-10 mm diameter individually.
5
Bacillus spp.
(Mixture of
B./icheniformis,
B.subtilis,
B. megaterium &
B. cereus)
Large flat spreading, Rods. Older colonies (>4 days)
irregular, matte off-white exhibited large numbers of
colonies; 5-25 mm; smelly. spores (bright), and rods (dark)
containing spores.
16S rDNA; confirmed by API CH50 kit and formation of clear zones on GYC medium (1.5% ethanol).
98
3.23.2 Identification of yeast species
Yeast isolates were differentiated into sub-groups according to colony and

cellular morphology. As described in Tables 3.2, four colony types were
repeatedly observed during the fermentation experiments. Representative
isolates of each colony type were identified to species level by sequence analysis
of the D1/D2 region of the 26S rDNA. Species identifications were confirmed
on the basis of assimilation profiles obtained with the API ID 32 C kit
(BioMerieux, Marcy-L'Etolie, France), and in some cases by additional
physiological tests. Homology data and accession numbers for the different
species and samples analysed are shown in Appendix A. The results of
carbohydrate assimilation profiling using the API test kits (Biomerieux) are
shown in Appendix A.
3.2.3.3 Identification of bacterial species
Bacterial isolates were examined for cell morphology and Gram stain, catalase
and oxidase reactions according to standard procedures (Tables 3.3 and 3.4)
(Smitbert and Krieg 1994) and correlated with sequence data. Isolates were
identified to species level using the results of sequencing of the partial 16S
rDNA sequences. Species identification was confirmed using API CH 50 test
strips (BioMerieux) on selected isolates. Homology data and accession numbers
for the different species and samples analysed are shown in Appendix A. The
results of carbohydrate assimilation profiling using the API CH 50 kits
(Biomerieux) are shown in Appendix A.
3.2.3.4 DNA extraction for identification by PCR-sequencing
For yeast, genomic DNA was extracted according to the method described by
Cocolin et al. (2000). Yeast cell pellets were prepared from 24 h cultures of
isolates in Malt Extract Broth (1 mL, centrifuged at 12, 000 g, 10 min, 4°C) and
resuspended in 200pL of breaking buffer (1% SDS, 2% Triton X-100, lOOmM
NaCE, lOmM Tris, ImM EDTA, pH 8.0) containing 0.3 g glass beads (0.5 mm
diameter, BioSpec Products, Inc. Bartlesville OK). Cells were homogenised at
6,000 rpm for 1 min in a bead beater (Mini-BeadBeater-1™, Biospec) in the
presence of 200pL of phenol/chloroform/isoamylalcohol (25:24:1). 200pL of TE
buffer (lOmM Tris, ImM EDTA, pH 8.0) were added, and the mixture was
centrifuged at 12,000 for 10 min at 4°C. The aqueous phase was collected and
DNA was precipitated by the addition of 0.6 volumes of isopropanol and
99
recovered by centrifugation at 12,000 g for 10 min at 4°C. The DNA pellet was
washed with 70% ethanol, dried and resuspended in 50 pL of TE buffer. DNA
extracted from pure cultures was used in PCR reactions for the amplification of
partial 26S ribosomal DNA.
For bacterial isolates, the method used for DNA extraction was that described
by Lopez et al (2003) and was identical to that for yeast, with the following
variations: (i) instead of Malt Extract Broth, MRS broth (Oxoid) or Tryptic Soy
Broth (Oxoid) were used to grow the pure cultures; (ii) the glass beads used for
the bacterial DNA extraction were smaller (diameter 0.1 mm; Daintree
Scientific, Tasmania, Australia); and (iii) the homogenising step involved 2
rounds of bead-beating for 40s at 4,600 rpm. The extracted bacterial DNA was
precipitated, washed, dried and resuspended in the same manner as the yeast
DNA
3.2.3.5 PCR amplification of DNA for sequence identification.

Genomic DNA from yeast cultures was amplified by PCR with universal
primers NL1 (5'-GCATATCAATAAGCGGAGGAAAAG-3') and NL4 (5'-
GGTCCGTGTTTCAAGACGG-3') for identification by partial 26S rDNA
sequence analysis (Kurtzman and Robnett 1998). Bacterial genomic DNA was
amplified by PCR with universal primers flanking the region between 968
(5'-AACGCGAAGAACCTTAC-3') and 1401 (5'-CGGTGTGTACAAGACCC-3')
(E.coli numbering) for identification by 16S rDNA sequence analysis as
described by Bae et al. (2004).
For PCR of extracted yeast DNA, each reaction contained PCR buffer (lOmM
Tris-Cl, 50mM KC1), 0.2 pM of each primer, 200pM of each DNTP (Roche
Diagnostics, Indianapolis, IN), 1.5 mM MgC12,1.25 U Gold Taq DNA
polymerase (AmpliTaq™, Roche Molecular Systems, Brachburg, NJ), and lOng
of DNA template in 50pL final volumes. Amplifications were carried out in a
9600 Thermal Cycler (Applied Biosystems), with the following program: initial
denaturation at 95°C for 7 min, 36 cycles at 95°C for 1 min (denaturation), 52°C
for 2 min (Annealing) and 72°C for 2 min (extension), followed by a final
extension for 10 min at 72°C. The presence of PCR amplicons was confirmed by
1.5% agarose gel electrophoresis.
The above method was also used for the PCR amplification of bacterial DNA,
except that the thermal program was as follows: initial denaturation at 94°C for
100
7 min, 10 cycles at 94°C for 30 s, 50°C for 30 s, and 72°C for 45 s, followed by 20
additional cycles where each successive cycle had an additional 5s of elongation
time. The final elongation was conducted at 72°C for 10 min. The presence of
PCR amplicons was confirmed by 1.5% agarose gel electrophoresis, before being
used in sequencing analysis. Yeast primers were obtained from Proligo Primers
and Probes (Sydney, Australia), and the bacterial primers were obtained from
Sigma Genosys (Sydney, Australia).
3.2.3.6 DNA sequencing
Sequence analysis was performed on PCR amplicons from yeast and bacterial
isolates, using the ABI PRISM® BigDye™ Terminator v3.1 Cycle Sequencing kit
(Applied Biosystems, Foster City CA), according to the manufacturer's
directions. Labeled DNA fragments were sequenced at the Automated DNA
Analysis Facility, School of Biotechnology and Biomolecular Sciences, UNSW.
The sequence of nucleotide bases, 26S rDNA sequences for yeast and 16S rDNA
sequences for bacteria, were compared with available sequences in Genbank
using the NCBI BLAST utility (Altschul et al., 1990; Karlin and Altschul 1990).
3.2.3.7 Denaturing gradient gel electrophoresis (DGGE) analysis of the

microbial ecology of cocoa fermentations
DGGE was used to qualitatively determine the microbial ecology of the

fermentations described in this chapter. This was done to compliment the
culture based enumeration methods, and provide further confirmation of
identifications made by sequencing of rDNA and use of API kits.
For the DGGE-investigations, pulp was aseptically washed from the beans in
sterile double distilled water, centrifuged (20,000g for 15 min, 4°C), the cell
pellet removed and freeze dried, and the nucleic acid extracted following the
protocol of Nielsen et al. (2005). The dried samples of extracted nucelic acid
were then sent to The Royal Veterinary and Agricultural University,
Department of Dairy and Food Science, Food Microbiology, Copenhagen,
Denmark for analysis by PCR -DGGE, according to the method described by
Nielsen et al. (2005). To assist in comparison, standards were also run on the
gels. These standards consisting of amplicons from four yeast isolates
(Ha?iseniaspore guilliermondii, Saccharomyces cerevisiae, Issatchenkia orientalis and
Pichia membranifaciens), and 8 bacterial isolates (Lactobacillus plantarum,
Pediococcus acidilactici, Leuconostoc pseudomesenteroides, Acetobacter pasteurianus,
101
Acetobacter tropicalis, Gluconobacter oxydans, Pantoea agglomerans and Bacillus

subtilis). The DGGE profiles generated from these analyses are included in
Appendix A, and discussed in Section 3.4, Discussion.
3.2.4 Chemical analyses

3.2.4.1 Sample preparation
Samples to be analysed for concentrations of carbohydrates and organic acids

were prepared according to the method of Ardhana and Fleet (2003). After
thawing of frozen samples, approximately 50 cocoa seeds were manually
separated from their pulp. Samples (50g) of each were separately ground in a
mortar (diam. 18.5 cm) for 10 min. Twenty grams of the ground sample (seed or
pulp) and 60 mL of double distilled water (Millipore, Billerica MA) was
homogenised in a Waring blender (Waring Products, Torrington, CT) for 4 min.
The homogenate was centrifuged (Beckman-Coulter, Fullerton CA) at 4°C,
20,000 x g for 20 minutes and the supernatant retained. The sediment was
washed twice with 20 mL of double distilled water by repeated suspension and
centrifugation, and the supernatants, retained and combined with the first
supernatant. The combined supernatants were clarified by vacuum filtration
through a 0.45g non-sterilized filter membrane (Millipore), followed by vacuum
filtration through an Amicon YM5 membrane (Amicon Corporation, Lexington,
MA, USA). The filtrate was passed through a syringe driven Sep-pak C18
cartridge (Waters, Milford MA) before analysis.
3.2.4.2 Carbohydrates
The concentrations of monosaccharides and disaccharides in seed and pulp

samples were determined using high performance liquid chromatography
(HPLC). The instrumentation consisted of a pump and controller (Waters 600),
auto-injector (Waters 717 Autosampler PLUS), a differential refractive index
detector (Waters 2414 RID), and a Silica Radial-Pak Column (Waters 8 SI, 10(a)
compressed in a radial compression module (Waters RCM 100). A computer
running Waters Millennium™ software controlled the equipment and recorded
data. The column was eluted at 2.5 mL/min with a mobile phase of acetonitrile-
water (75:25 v/v) containing 1% (v/v) Silica modifier (Waters, SAM 1). Each
sample was analysed in duplicate.
102
The concentrations of individual carbohydrates eluted from the column were

determined by reference to the injection and elution of standard solutions of
glucose, fructose, sucrose and maltose (Sigma) at concentrations of 1.5%, 1.0%
and 0.5% (w/ v) for each sugar. Standard curves were constructed, and
calculations performed using Waters Millennium™ software. The procedure
from extraction to HPLC determination recovered greater than 98% of an
internal standard (lactose). For comparison to the literature, the concentration of
sugars was expressed as % w / w. The maximum variability between duplicate
samples was ± 0.2% w/ w.
3.2.4.3 Organic acids
The concentration of individual organic acids was determined using HPLC. The
chromatography instrumentation consisted of a pump (Waters 600), auto
injector (Waters 717 Autosampler PLUS), a UV detector (Waters 996 PDA)
operating at 210 nm, a column heater and controller (Waters, TCM temperature
controller), and an Aminex Ion Exclusion HPX-87 H stainless steel column (Bio-
Rad, Richmond CA). A computer running Waters Millennium™ software
controlled the equipment and recorded data. The column was eluted at 65°C
using 0.08% H3P04 (Ajax Chemicals, Sydney Australia) at a flow rate of
0.5mL/ min. All samples were analysed in duplicate.
The concentrations of individual organic acids eluted from the column were
determined by reference to the injection and elution of standard solutions of
oxalic, citric, tartaric, malic, succinic, lactic, and acetic acids at concentrations of
0.2, 0.1, 0.05, 0.025% (w/v) for each organic acid. Standard curves were
constructed and calculations performed using Waters Millennium™ software.
The procedure from extraction to HPLC determination recovered greater than
96% of the internal standard, fumaric acid. For comparison to the literature, the
concentration of organic acids was expressed as mg/g(dry basis). The
maximum variability between duplicate samples was ± 3 mg/g.
3.2.4.4 Ethanol
Ethanol concentrations were determined in samples of cocoa pulp and seed

taken from the fermentations. The samples were prepared as described in
Section 3.2.4.1, except that vacuum filtration was not used. Instead, 2mL
aliquots of supernatant were centrifuged an additional time at 4°C, 15,000 x g
for 10 min. Ethanol concentrations were determined using an enzymatic assay
103
kit, following the manufacturer's instructions (R-Biopharm AG, Germany). For

comparison to the literature, the concentration of ethanol was expressed as
%w/ w. The maximum variability between duplicate samples was ± 0.1% w/ w.
3.2.5 Quality and sensory assessment of cocoa beans

3.2.5.1 Quality assessment
Samples of fermented and dried cocoa beans were sent to Cadbury-Scwheppes
laboratories in Singapore for quality analysis, where the fermentation index and
appearance of the bean samples were assessed using the cut test. Other tests
routinely used to evaluate commercial cocoa beans were also performed as
follows: Moisture content of whole beans and nibs, bean size, % shell, fat
content and nib pH. The methods used for these tests were developed by
Cadbury-Schweppes but were based on AO AC methods (AOAC, 1984) and
other literature as follows: Cut test (Wood and Lass, 1985; Senenayake et al.,
1995); bean size and % shell by weight (Wood and Lass, 1985); moisture content
by oven drying, and nib pH (AOAC, 1984); fat content by nuclear magnetic
resonance (Kirk and Sawyer, 1991). Ideal and industry standard values for these
parameters were obtained during the literature review and are included later
(Table 3.9).
A basic assessment was also made of the external appearance of the beans, and
of their aroma. To assess the aroma, lOOg of dried fermented beans were ground
in a knife-mill for 30s and smelt. The strength of cocoa aroma was noted, along
with any unpleasant aromas or taints.
The cut test was repeated at UNSW for all beans samples, in order to verify the
results provided by the external laboratory. To perform the cut test 100 beans
were randomly selected from the sample being evaluated. Each bean was cut in
half laterally using secateurs and visually inspected. The beans were graded
according to the visual guide in Wood and Lass (1985). A sample was judged to
be acceptable if it contained 0% mouldy or insect infested beans, less than 2%
slaty beans and less than 20% fully purple beans. Based on these results, the
highest quality cocoa bean samples from each experiment were selected and
made into chocolate for sensory evaluation.
104
3.2.5.2 Chocolate making
The chocolate making methods used in this study were provided by Cadbury-
Schweppes, but were very similar to protocols outlined in the literature (Holm
et al., 1993; Jinap et ah, 1995). Fermented and dried cocoa beans were roasted in
a digitally controlled oven at 120°C for 55 minutes.
The shell of roasted beans was manually removed from nibs. Chocolate was
made by the procedures as follows:
One hundred and eleven grams of roasted and peeled cocoa beans (also called
nibs) were transferred into a mechanical mortar and pestle (Supplied by
Cadbury-Schweppes, Tasmania, Australia). After an initial grinding period of
30 minutes, 45g of deodorized cocoa butter (Cadbury Schweppes, Melbourne,
Australia) and 72g of pure icing sugar (Bundaberg, Australia) were added. The
mixture was then conched in the mortar for at least 5 hours until a fine liquid
chocolate was obtained. The mortar was heated to 30-40°C during the whole
process. The liquid chocolate was transferred to a stainless steel bowl for
tempering. Tempering is necessary to create the desired texture and suitable
melting point for chocolate, both of which affect flavour perception (Hoskin,
1994). The steel bowl of liquid chocolate was placed in a water bath with water
supplied from hot and cold taps, and the temperature controlled by the aid of a
digital thermometer . Continuous mixing of the chocolate was essential to
distribute the heat evenly throughout the chocolate mass. Firstly, the chocolate
container was placed in hot water and mixed until the temperature of chocolate
was 45-46°C (about 5 min). Then, hot water was replaced with cold water to
cool the liquid to 27°C (~10 min). Finally, the chocolate was warmed to 33°C (2
min) before being poured into small paper moulds and allowed to set in a
refrigerator at 10°C. Moulded chocolate was stored in an odour free refrigerator
for at least 2 weeks prior to tasting to allow stabilisation of flavours.
3.2.5.3 Sensory evaluation of chocolate samples by untrained panelists
Two types of sensory evaluation were performed on the chocolate samples:

discrimination testing and affective testing, according to the methods described
by Carpenter et al. (2000) and Stone and Sidel (2004)
In the discrimination tests, panelists were required to taste 3 chocolate samples
and choose the one that was different from the other two. This is termed a
105
'Triangle' test (Carpenter, 2000). This assessed the ability of panelists to

successfully distinguish between chocolate samples made from different
batches of cocoa beans, solely on the basis of their flavour. The samples were
numerically coded and presented on a serving tray, along with warm rinse
water, plain water biscuits, a score card and a pen (Figure 3.10, Figure 3.11).
The sensory evaluation session was carried out in sensory booths with suitable
lighting to maximize the objectiveness and reliability of the results (Carpenter et
al., 2000). Between samples, panelists cleansed their palate with the water and
crackers. Panelists had no prior training in sensory evaluation. Before each
session, all panelists were screened to ensure that they regularly consumed dark
chocolate (at least monthly) and had not consumed any strongly flavoured
foods in the 90 minutes before the session. A pool of 50 panelists was used, and
each sample set was assessed 10 -15 times. Two discrimination tests were
performed in a session.
Fig. 3.10. Arrangement of chocolate samples for sensory evaluation. A similar

arrangement was used for both difference (triangle) tests and liking tests.
106
warm water Plain biscuits
Instructions
Sample 308 Sample 512 Sample 781
and score-card
Fig. 3.11. Presentation of chocolate samples on serving trays for sensory evaluation.
Sample codes are for illustration only.
In addition to discrimination testing, panelists also performed affective, or

diking,' testing. The test consisted of two samples of chocolate made from the
Australian beans, and one sample made from Ghanaian cocoa beans. Panelists
rated their degree of liking for the chocolate flavour of each sample using a
linear scale. The graphic scale corresponded to numerical values where a score
of 0 indicated "intense dislike," a score of 10 indicated "intense liking" while a
score of 5 indicated basic acceptance without strong like or dislike.
The samples were coded, and their identities were not known by the panelists.
Serving orders were randomized and balanced and all samples were presented
an equal number of times. A maximum of one affective (liking) test was
performed at a time, and each set of samples was assessed for liking at least 15
times. Panelists were instructed to record on their forms any distinctive flavours
that were noted in individual samples.
The Ghanaian samples were included to provide a standard point of reference
against which acceptability of the Australian chocolates could be assessed. Such
a comparison is standard practice in industry, because:
"Ghanaian cocoa is considered to be the standard for bulk cocoa and is
characterised by a high chocolate flavour, and low acidity and
astringency" (Tomlins et al. 1993).
Dried, fermented cocoa beans from Ghana were supplied by Cadbury-
Schweppes and were obtained by randomly sampling their stocks of
commercial quality beans.
107
3.3 Results
3.3.1 Fermentation of cocoa beans in heap, box and barrel.
Batches of cocoa beans were simultaneously fermented by well established
heap and box fermentation methods (Rohan, 1963), and by a novel barrel
method.
3.3.1.1 General fermentation profiles
Figure 3.12 compares the growth of total populations of yeast and

bacteria,changes in the temperature of the bean mass and pH of pulp, for the
three methods of fermentation. For the first trial, beans from the Mossman
plantation were used.
All fermentations started at 22°C, and by 96 hours had reached 45-47°C.

The kinetics of temperature increase were similar for the heap and the barrel
fermentations, while the temperature of the box fermentation increased at a
faster rate. The box fermentation reached its maximum temperature (45°C) 24h
earlier than the heap and barrel fermentations. The pH of the pulp was initially
3.45 and increased to 4.1 - 4.2 for the barrel and box fermentations. The final pH
of the pulp for the heap fermentation was higher, with a value of 4.85.
With regards to microbial ecology, the fermentations showed some similarities.

The initial population of both yeast and bacteria in these vessels was 105 cfu / g
beans for all three fermentations. The populations of yeast and bacteria
increased to maxima of 107-108 cfu / g beans after 48 hours. In the case of the box
and heap fermentations, these populations held relatively steady until the end
of fermentation at 120 hours. However, in the case of the barrel fermentation,
the yeast population declined from 48 hours onwards, and no yeast (< 102 cfu/g
beans) were detected after 96 hours.
108
s' 3 -
E 25 -
o 2 -
Fermentation time (hours) Fermentation time (hours)
2 35 -
b 30 -
't 3 -
U 40 ■
H 35 -
b 30 -
's' 3 -
© 2 -
Fig. 3.12. Fermentation of Australian cultivated cocoa beans in: (a) heap, (b) box and (c)
barrel; Total yeast population ♦; Total bacterial population EfTemperature ■; pulp pH O.
Beans sourced from the Mossman plantation.
The fermentation trials were repeated using beans sourced from the South
Johnstone plantation (Figure 3.13). In these trials, the heap fermentation was not
conducted due to a shortage of beans, and since it was not considered a
practically controllable method for the future development of an industrial
process.
109
a 40 -
s 35 ■
2 30 -
Fig. 3.13. Fermentation of Australian cultivated cocoa beans in: (a) box (b) barrel;
Total yeast population ♦; Total bacterial population □; Temperature ■; pulp pH O; Beans
sourced from the South Johnstone plantation.
A faster temperature increase, and higher maximum temperature (47°C) was

observed for the box fermentation, compared to the barrel fermentation which
achieved a much lower final temperature of 42°C. The initial pH of the pulp
was 3.7, similar to that for the Mossman beans (Figure 3.12). By the end of
fermentation (120h) the pH of the pulp in the box fermentation (pH 5.0) was
higher than that of the barrel (pH 4.3), and of the box using beans from the
Mossman plantation (pH 4.5). The final pH of the barrel fermentations was 4.3,
whether the beans were from the Mossman or the South Johnstone plantations.
The initial populations of yeast were higher, and the bacteria lower than in the
first trial (Figure 3.12). The yeast reached maximum populations (107 cfu / g
beans) after about 24 hours, and then died off in both box and barrel. In contrast
to the first trial, the yeast in the box fermentation of this trial died off after 48
hours, while bacteria reached maximum populations (108 cfu/g beans) after 72
hours. During fermentations using South Johnstone beans (Figure 3.13) the
bacterial population in the box declined sharply after 96 hours.
110
3.3.1.2 Changes in the populations of yeast species
Figure 3.14 shows the growth of individual yeast species during the heap, box
and barrel fermentation of beans from the Mossman plantation, and box and
barrel fermentation of beans from the South Johnstone plantation. Three yeast
species were found to be predominant in all of the fermentations. These were
Hanseniaspora guiltiermondii, Issatchenkia orientalis and Saccharomyces cerevisiae.
Fermentation time (hours)
2 -

Fig. 3.14. Growth of yeast species during fermentation of cocoa beans in (a) heap [Mossman
beans], (b) box [Mossman beans], (c) barrel [Mossman beans], (d) box [S.Johnstone beans], (e)
barrel [S. Johnstone beans]; Hanseniaspora guiltiermondii ♦; Saccharomyces cerevisiae □;
Issatchenkia orientalis O.
Ill
Hanseniaspora guilliermondii grew to populations of 107-108 cfu/g in the first 24

hours, after which it declined to non-detectable levels (<102 cfu/g). This decline
occurred more slowly in the heap fermentation, compared to the box and barrel
fermentations. Generally, H.guilliermondii was not detected in beans from the
box or barrel after 48-72h.
In all fermentations, S.cerevisiae reached a maximum population after 48 hours.

There was some variation observed in the initial populations of S.cerevisiae
isolated. For the fermentations using Mossman beans, no S.cerevisiae were
detected in the heap at the beginning of fermentation, while 103 cfu/g were
detected in the box, and 105 cfu/g were detected in the barrel. The levels of
S.cerevisiae in these fermentations remained between 106- 107cfu/g beans until
120h. In the barrel, S.cerevisiae populations declined after 48 hours.
Issatchenkia orientalis was initially detected in the beans of fermentations at

levels of 104-105 cfu/g beans. It steadily increased in population from Oh until
48-72 hours. In the fermentations using beans from the Mossman plantation, the
populations of l.orientalis remained steady in the heap and box
fermentations(107-108 cfu/g beans) between 72-120 hours .
In the fermentations using beans from South Johnstone, the populations of all
species of yeast declined after 48 hours. The box and barrel fermentations also
had higher initial populations of yeast (105 - 106 cfu/g) than when Mossman
beans were used (104 -105 cfu/g).
112
3.3.1.3 Changes in the populations of lactic acid bacteria species
Four species of lactic acid bacteria dominated the fermentations and were
readily differentiated by their colony and cell morphologies, described in Table
3.3 in the Methods section 3.2.3.I. These were Lactobacillus plantarum,
Lactobacillus fermentum, Pediococcus acidilactici and Leuconostoc
pseudomesenteroides. Figure 3.15 shows changes in populations of these bacteria
during fermentation.

Fig. 3.15. Growth of lactic acid bacteria species during fermentation of cocoa beans in; (a) heap
[Mossman beans), (b) box [Mossman beans], (c) barrel [Mossman beans], (d) box [S.Johnstone
beans], (e) barrel [S.Johnstone beans]; Pediococcus acidilactici ♦; Lactobacillus plantarum □;
Lactobacillus fermentum O; Leuconostoc pseudomesenteroides A.
113
For the Mossman beans, Leuconostoc pseudomesenteroides grew from initial

populations of 104-105 cfu/ g to maximum populations of 107- 108cfu/ g beans
in the first 48h. It then declined at rates that differed for the different
fermentation methods, with the slowest decrease occurring in the heap.
In the heap, box and barrel fermentations, Lactobacillus plantarum was initially
present at 103 cfu/g, and after 72 hours it reached a maximum population of
104-105 cfu/g. In the heap and box fermentation, the population of L.plantarum
remained steady, but in the barrel it decreased to 104 cfu/g beans by 120h
(Figure 3.15 a,b,c).
Pediococcus acidilactici was detected at initial populations of 103- 104cfu/ g and

slowly increased to maximum populations of 107-108 cfu/g in the heap and box
fermentations, where it persisted at these populations until the end of
fermentation (120h). In the barrel fermentation the population of P.acidilactici
reached a maximum population of 106-107 cfu/g at 72h, and then decreased to
104-105cfu/ g beans.
Lactobacillus fermentum was not consistently isolated during fermentations of the

Mossman cocoa beans, but was detected at populations of 102 cfu/g in the
samples taken from the heap and box fermentations at Oh.
When the box and barrel fermentations were repeated using beans harvested
from the South Johnstone plantation, the lactic acid bacteria produced similar
but slightly different growth profiles (Figure 3.15 d,e).
Leuconostoc pseudomesenteroides grew in both box and barrel fermentations,
reaching maximum populations at 24h and then dying off. In these
fermentations, the maximum population was less (106-107 cfu/g) and the rate of
decline was faster. Lc. pseudomesenteroides were not detected after 48h, compared
to 72h in the fermentations with Mossman beans (compare b & d, c & e in
Figure 3.15).
In the box, Pediococcus acidilactici was the predominant species of lactic acid
bacteria, growing throughout the fermentation, with a maximum population
(108 cfu/g beans) at 72 hours, and declining slowly from this point.
114
Lactobacillus plantarum and L. fermentum also grew in the box, reaching

maximum populations of 108 cfu/ g, at 48h and 72h respectively, and then died
off.
In the barrel, a similar growth pattern was observed, except that L. plantarum
was the predominant species. In this fermentation, P. acidilactici reached a
maximum population of 107cfu / g at 48h and then slowly declined to a final
population of 103 cfu/g at 120h. Meanwhile L. fermentum was detected 48h later
compared to the box, reaching maximum populations of 107 at 96 hours and
remaining at this level until the end of fermentation (Figure 3.15 d, e)
Overall, slightly lower initial populations of lactic acid bacteria were found in
the South Johnstone beans, but they established much faster growth.
3.3.1.3 Changes in the populations of acetic acid bacteria and other

bacterial species
Figure 3.16 shows the growth of acetic acid bacteria, and other bacterial species,
during the heap, box and barrel fermentation of cocoa beans. The acetic acid
bacteria Acetobacter pasteurianus and Gluconobacter oxydans, along with Pantoea
agglomerans and some members of Bacillus spp. were predominant species
isolated from the fermenting beans. Not all of these species were isolated from
every fermentation.
A. pasteurianus was the most prevalent and predominant species of acetic acid
bacteria in all fermentations. However, in the box and barrel fermentations
using Mossman beans, it was only detected after 24 hours of fermentation had
already occurred. Maximum populations (107-108 cfu/g) were usually found
after 48-72 hours into the fermentation, and remained at this level until the end
(120h). However, in the box using beans from the South Johnstone plantation, it
declined after 96h.
115
a 4 -

Fig. 3.16 Growth of acetic acid bacteria and other bacterial species during fermentation of cocoa
beans in; (a) heap [Mossman beans], (b) box [Mossman beans], (c) barrel [Mossman beans], (d)
box [South Johnstone beans], (e) barrel [South Johnstone beans]; Pantoea agglomerans ♦;
Acetobacter pasteurianus □; Gluconobacter oxydans A; Bacillus spp. O.
Gluconobacter oxydans was the other main species found to grow at significant
levels throughout the fermentations. This species was significant in the box and
barrel fermentations of cocoa from the South Johnstone plantation, but was
found only in the box fermentation of beans from the Mossman plantation
(compare b, d and e in Figure 3.16). In fermentations using South Johnstone
beans, G.oxydans reached maximum populations of 107 cfu/ g around 48-72h
116
before dying off., while in the box fermentation of Mossman beans, it grew to
maximum populations of 107cfu/g beans by the end of fermentation.
Pantoea agglomerans was isolated from the Oh samples of cocoa beans from the
Mossman plantation. In the first 24-48 hours of fermentation it grew to 105-106
cfu/g and then died off (Figure 3.16, a, b & c). When beans from the South
Johnstone plantation were used, P..agglomera?is was only isolated from the Oh
samples, and did not grow in either the box or barrel.
Populations of Bacillus spp. were also detected in the cocoa beans, using the
same media as for the isolation of acetic acid bacteria. The isolates obtained
were a mixture of B.licheniformis, B.pumilis, B.subtilis, B. megaterium and
occasionally, B.cereus. The maximum populations of Bacillus spp. were
approximately
104 cfu/g, and these maxima were usually detected at the end of fermentation
(120h). However, in one fermentation, Bacillus spp. were detected earlier in the
fermentation (24h) at a population of 6 x 104 cfu/g, but then died off
(Figure 3.16, b).
117
3.3.1.4 Changes in the concentrations of sugars during heap, box and

barrel fermentation
Figure 3.17 shows the change in concentration of sugars during fermentation of

cocoa beans obtained from the Mossman plantation, using box, barrel and heap
methods.
Pulp Seed
-O 5.0 -
* 2.0 - £ 2.0 -
^ 1.0 ^
§ 5.0 - § 5.0 -
4.0 - « 4.0 -
'T' 3.0 - 'r' 3.0 -
£ 2.0 - £ 2.0 -
^ 1.0-
0.0 Et
£ 5.0 - £ 5.0 -
tJ 4.0 - -S 4.0 -
'r' 3.0 -
«i 2.0 - £ 2.0 -
^ 1.0 - ^ 1.0 -
Figure 3.17 Changes in the concentration of sugars in the pulp (a, c, e) and seed (b, d, f) of
Mossman cocoa beans during fennentation by heap (a, b), box (c. d) and barrel (e, f) methods;
glucose ♦; fructose □; sucrose O.
Glucose (6.0%) and fructose (7.0%) were the main sugars found in freshly
obtained cocoa pulp. They were utilised throughout fermentation and could not
118
be detected in the pulp after 72h. Faster utilisation of the pulp sugars was
observed in the barrel fermentation.
Sucrose (1.3%) was the main sugar found in the seeds. Its concentration
decreased throughout fermentation until it could no longer be detected (Figure
3.17, b, d, f). This decrease in concentration was fastest in the box fermentation
(Figure 3.17, d)., and slowest in the heap fermentation (Figure 3.17 b). The
utilisation of sucrose was accompanied by slight increases in the concentrations
of glucose and fructose in the seed. Traces of sucrose(0.5%), glucose(0.25%) and
fructose (0.3%) were detected in the seed of the cocoa beans at the end of
fermentation (Figure 3.17 b, d, f).
Similar concentrations of sugars were found in the pulp and seeds of cocoa
beans obtained from the South Johnstone plantation. The kinetics of sugar
utilization when this cocoa was fermented were similar to those for the
Mossman beans (data not shown). As before, the rate of sugar utilization in the
pulp was slightly faster in the barrel fermentation compared to that of the box.
3.3.1.5 Changes in the concentrations of ethanol during heap, box and

barrel fermentation
Ethanol was produced in the pulp throughout fermentation. It reached

maximum concentrations at 48h for all of the heap, box and barrel methods
(Figure 3.18).
The highest pulp concentration of ethanol was recorded in the barrel (7.0%),
followed by the box (6.5%) and heap (5.5%) fermentations. Its concentration
declined sharply after 48h and by 120h only about 0.5% was detected in the
pulp of any fermentation.
The concentration of ethanol in the seed increased throughout fermentation and

reached maximum levels at 72h for all fermentation methods. The highest seed
concentration of ethanol was found in the barrel fermentation (3.5%), followed
by the box (2.0%) and heap (1.5%) fermentations.
119
tJ 4.0 -
£ 2.0 -
^ 1.0 -
£ 2.0 -
I 5.0 -
« 4.0 -
£ 2.0 -
^ 1.0 -
Figure 3.18 Changes in the concentration of ethanol in the pulp and seed of Mossman cocoa
beans during fennentation by heap (a), box (b) and barrel (c) methods; pulp ♦ and seed □.
Similar dynamics of ethanol production were observed when cocoa beans from
the South Johnstone plantation were fermented using box and barrel methods
(data not shown). As before, higher concentrations of ethanol were observed in
the seed and pulp from the barrel fermentation compared to those from the box.
120
3.3.1.6 Changes in the concentrations of organic acids during heap, box and
barrel fermentation
Figure 3.19 shows the changes in concentrations of organic acids during heap,
box and barrel fermentation of cocoa beans from the Mossman plantation.
Table 3.5 summarises the changes in concentration of these acids from
beginning (Oh) to end of the fermentations (120h). In the unfermented cocoa
pulp, the following acids were detected: citric acid (55 mg/g), malic acid (16 -
17 mg/g), tartaric acid (5 mg/g) and oxalic acid (4-5 mg/g).
£ 70 :
30 -
£ 20 £ 10 -
100 n
.52 80 - .52 40 -
£ 70 -
£ 20 £ 10 -
100
£ 70 -
30 -
£ 20 £ 10 -

Figure 3.19 Changes in the concentrations of organic acids in the pulp (a, c, e) and seed
(b, d, f) of Mossman cocoa beans during fermentation by heap (a, b), box (c, d) and barrel (e, f)
methods; citric acid ♦; acetic acid □; lactic acid O; malic acid ■
121
The concentration of citric acid in the pulp increased slightly during the first
24h of fermentation, with the heap method giving the greatest increase.
Between 24-72h, there were notable decreases in the concentration of citric acid
for both heap and box fermentations, after which slight increases were observed
(Figure 3.19 a, c). For the barrel fermentation, the concentration of citric acid in
the pulp declined slowly from 24h until the end of fermentation. The final
concentrations of citric acid in the pulp were 15 mg/g for the heap, 31 mg/g for
the box and 41 mg/g for the barrel fermentations.
The concentration of malic acid in the pulp increased during the first 24h for
both heap and box fermentations and during the first 48h for the barrel
fermentation. Thereafter, the concentration of malic acid decreased until 72h
when its concentration increased again, especially in the barrel and box
fermentations.
Lactic acid was produced in the pulp of the cocoa beans throughout
fermentation. The fastest increase in lactic acid concentration was noted in the
heap fermentation (Figure 3.19 a). In this fermentation, the concentration of
lactic acid subsequently decreased after reaching a maximum at 48h.
Acetic acid was also produced during the fermentations, with its concentration
in the pulp progressively increasing throughout the process. It was the main
acid in the pulp at the end of fermentation, especially in the barrel fermentation
The greatest increase was noted for the barrel fermentation, followed by the box
and heap. The concentrations of tartaric and oxalic acids remained constant in
the pulp of the cocoa beans, throughout fermentation.
In the unfermented cocoa seed, the following organic acids were found: citric
acid (10 -11 mg/g), malic acid (1-2 mg/g), tartaric acid (1 mg/g) and oxalic
acid (4 -5 mg/g). The seed concentrations of lactic and acetic acids increased
during fermentation, while those of citric, malic, tartaric and oxalic acids
remained relatively constant. At the beginning of fermentation (Oh), only traces
of acetic and lactic acid were detected in the seed samples tested (< 3 mg/g).
Levels of lactic and acetic acids increased in the seeds during fermentation. By
the end of fermentation, the largest increase in concentration of acetic acid was
observed for the barrel fermentation, followed by the box and the heap. Similar
levels of lactic acid (13 -15 mg/g) were found in the seeds from all three
fermentations.
122
Similar concentrations for these organic acids, were found in cocoa beans from
the South Johnstone plantation except for tartaric acid which was found at a
concentration of 10 mg/g in the pulp of unfermented beans. Changes in the
concentrations of these acids during fermentation were also similar to the data
from fermentations using Mossman cocoa beans (compare Tables 3.5 and 3.6).
In the South Johnstone beans, the concentration of citric and malic acids in the
pulp decreased during fermentation, as also observed in the Mossman beans.
Acetic and lactic acids increased to a greater degree in the pulp and seed of the
barrel fermentation, compared to samples from the box fermentation (Table 3.6)
Table 3.5 - Summary of changes in concentration of organic acids from the start (Oh) to end
(120h) of fermentation of Mossman cocoa beans using heap, box and barrel methods.
Fermentation Organic Pulp Seed

method acid (mg/g) (mg/g)
Start End Change Start End Change
(Oh) (120h) (Oh) (120h)
Heap Citric 55 15 -40 11 9.6 -1.4
Malic 16 8.8 -7.2 1.0 1.0 0
Tartaric 5.0 4.7 -0.3 1.0 1.0 0
Oxalic 4.4 2.4 -2 4.8 4.7 -0.1
Lactic 2.0 21 19 2.4 14 11.6
Acetic 1.1 28 26.9 1.2 12 10.8
Box Citric 52 31 -21 10 7.5 -2.5
Malic 17 23 6 1.1 1.2 0.1
Tartaric 5.1 4.3 -0.8 1.0 1.0 0
Oxalic 4.8 3.8 -1 4.1 3.8 -0.3
Lactic 1.2 36 34.8 2.3 13 10.7
Acetic 0.0 45 45 0.0 19 19
Barrel Citric 54 40 -14 11 12 1
Malic 16 17 1 1.5 1.7 0.2
Tartaric 4.8 5.0 0.2 1.0 1.0 0
Oxalic 3.8 2.8 -1 5.1 5.6 0.5
Lactic 1.3 46 45 2.2 15 12.8
Acetic 3.4 92 89 2.3 34 31.7

123
(120h) of fermentation of South Johnstone cocoa beans using heap, box and barrel methods.


(Oh) (120h) (Oh) (120h)
Box Citric 58 18 -40 1.0 3.7 2.7
Malic 12 6.7 -5.3 0.0 0.0 0
Tartaric 9.3 9.7 0.4 0.9 0.9 0
Oxalic 1.5 0.0 -1.5 0.8 1.1 0.3
Lactic 1.6 36 34.4 2 13 11
Acetic 0.0 38 38 0.0 15 15
Barrel Citric 62 21 -41 1.8 5.8 4
Malic 14 10 -4 0.0 0.0 0
Tartaric 11 11 0 0.9 0.9 0
Oxalic 2.0 0.3 -1.7 0.7 1.1 0.4
Lactic 1.8 51 49.2 1.9 17 15.1
Acetic 0.0 66 66 0.0 22 22

124
3.3.1.7 Quality evaluation of cocoa beans produced by heap, box and barrel
fermentations
After heap, box or barrel fermentation and subsequent drying, cocoa beans
were tested for several quality parameters routinely used to evaluate
commercial cocoa beans. These tests were done at the quality assurance
laboratories of a chocolate processing company in Singapore (Cadbury -
Scwheppes). For verification purposes, the cut test was repeated in the
laboratories of UNSW, as were general observations of the beans (appearance,
aroma).
Fermented beans from the Mossman plantation were oven dried, as previously
described, because excessive rain at the time of the experiments prevented solar
drying of the beans. Fermentations trials with the South Johnstone beans were
conducted a week later when the weather was more conducive to solar drying.
In order to permit comparison to the Mossman beans, a portion of the South
Johnstone beans were also dried using the oven method.
Figure 3.20 shows data for the fermentation index of the dried fermented cocoa
beans from both the Mossman and South Johnstone plantations. Measures of
significant differences in the fermentation indices of different bean samples
were determined using the Student's T-test (95% confidence interval) (data not
shown).
Heap
(oven)
II1
Box (oven) Barrel
(oven)
Box (oven) Barrel
(oven)
Box (sun) Barrel (sun)
Fig. 3.20. Cut test results for cocoa beans fermented in a heap, box or barrel;
a) Mossman beans; b) South Johnstone beans; ■ % beans fully brown,
11% beans part brown/part purple. Brackets indicate drying method used.
125
Heap fermentation gave beans with a small but significantly lower proportion
of fully brown beans, compared to beans from the box or barrel fermentations.
Furthermore, solar drying resulted in cocoa beans with a significantly higher
proportion of fully brown beans compared to oven dried beans.
Based on the cut test, all samples of beans were judged to be acceptable, with
0% mouldy or insect infested, less than 1% slaty and less than 20% fully purple
(Wood and Lass, 1985).
The effects of fermentation and drying methods on the other commercial

quality parameters are shown in Tables 3.7 and 3.8.
As might be expected, oven dried beans had significantly lower moisture

contents and significantly more shell than solar dried beans. The nibs of heap
fermented cocoa beans had a significantly higher pH (5.5) than the box or barrel
fermented beans (pH 5.0 ± 0.2).
Apart from these variations, neither the plantation from which the cocoa beans
were obtained, nor the fermentation and drying methods appeared to affect the
basic physical quality of the cocoa beans produced. All Australian samples were
found to meet the prescribed ideal values for bean size, moisture content (bean
and nib), fat content, appearance and aroma parameters (Table 3.9).
Table 3.7 - Quality evaluation of Mossman cocoa fermented by heap, box and barrel methods,
and dried by oven method.
Standard
Quality parameter Heap (oven) Box (oven) Barrel (oven) Mean
Deviation
94 96 94
Bean size - no. per 1 OOg 95 1.41
% Bean moisture 5.1 4.6 5.8 5.17 0.60
4.3 4.0 5.1 4.47 0.57

% Nib moisture
13.7 13.2 13.3 13.4 0.26

% Shell content
% Nib fat (dry' basis) 54.9 55.7 55.2 55.3 0.40
Nib pH 5.5 5.1 4.9 5.17 0.30
acceptable acceptable acceptable ... ...

External appearance
acceptable acceptable acceptable ... ...

Aroma assessment
126
Table 3.8 - Quality evaluation of South Johnstone cocoa fermented by box and barrel methods,
and dried by solar and oven methods
Barrel Box Barrel Standard

Quality parameter Box (oven) Mean
(oven) (solar) (solar) Deviation
Bean size - no. per lOOg 96 97 98 99 98.0 1.00
% Bean moisture 3.9 4 5.4 6 5.13 1.02
% Nib moisture 3.7 3.3 5.1 5.4 4.60 1.14
% Shell content 14.9 14.8 12.8 13.4 13.7 1.02
% Nib fat (dry basis) 56.9 57.4 55.8 54 55.7 2.40
Nib pH 4.9 4.8 5.2 5.0 5.00 0.2
External appearance acceptable acceptable acceptable acceptable ... ...
Aroma assessment acceptable acceptable acceptable acceptable _ _
Table 3.9 - Industry prescribed values for the commercial assessment of cocoa bean quality.
Quality parameter Ideal value or range* Ghana standardb Indonesia standardb
Bean size - no. per 1 OOg <110 beans per 1 OOg 87 90
% Bean moisture < 7.5% 6.4 6
% Nib moisture < 7.5% 5.1 5.3
% Shell content 11 - 13% 11.5 12
% Nib fat (dry basis) 54 - 58% 55.3 55.5
Nib pH pH 5.3 -6.0 5.8 5.3
Free from surface mould, foreign ...

External appearance —
material and live insects
Absence of taints, especially:

Aroma assessment smoky, mouldy, hammy and — ...
ammoniacal aromas
a Adapted from the chocolate manufacturers' quality guidelines in Anon (1984) and CAA (2008).
b Industry standard values for regional beans provided by Cadbury-Schweppes.
However, all samples had shell contents slightly higher than the recommended
values, while only the heap fermented beans had a pH value close to that
recommended (5.5). Whilst these data suggest a need for further optimisation of
the fermentation and drying methods for Australian cocoa beans, most
properties indicated that the Australian beans were of commercially acceptable
quality.
The dried fermented Mossman cocoa beans were selected for use in chocolate
making and sensory evaluation, because they were of acceptable quality and
represented a more complete comparison of heap, box and barrel methods.
127
3.3.1.8 Sensory evaluation of chocolate made from cocoa beans fermented by
heap, box and barrel.
Samples of the Mossman cocoa beans fermented by heap, box and barrel
methods were made into chocolate and subjected to two types of sensory
evaluation (difference and affective tests). The triangle test was used to
determine whether untrained panelists could detect a significant difference
between chocolate samples made from cocoa beans fermented by different
methods. Table 3.10 summarises the results of this difference testing.
Table 3.10 - Sensory evaluation (triangle test) performed on the chocolate samples made with
Mossman cocoa fermented by heap, box and barrel methods.
Number of Correct judgements

Number
Chocolate sample comparison correct needed for significance*
of tests
judgements (99% confidence level)
May 2004 - Fermentation method
Box Vs. Barrel fermentation 10 9 8
Box Vs. Heap fermentation 10 9 8
Barrel Vs. Heap fermentation 10 10 8
* Adapted from a table in Stone and Sidel (2004). A set of samples was determined to taste significantly
different from another if: the “number of correct judgments” was equal to or greater than the “number of
judgements needed for significance.”
Difference testing of the chocolate samples demonstrated that beans fermented

using box, barrel and heap methods could be distinguished from one another
on the basis of flavour. Therefore, fermentation in either heap, box or barrel
caused significant differences in the flavour of the cocoa beans.
Panelists also evaluated the chocolate samples according to liking (affective
assessment). The mean and standard deviation were calculated for the liking
scores of all samples, and Student's T-test was applied to the data. Tables 3.11
and 3.12 shows the results of these statistical analyses. The T-test results
describe whether the mean liking score of one chocolate sample was
significantly greater than that of a second sample. The data thus indicated a
significant difference in the mean liking scores of the following sample pairs:
Ghana and Heap; Ghana and Box; Heap and Box; Heap and Barrel; Box and
Barrel. There was no significant difference between the mean liking scores of the
Ghana and Barrel cocoa beans.
128
Table 3.11 - Mean liking scores for chocolate made from Mossman cocoa fermented by heap,
box and barrel methods, compared to chocolate made from Ghanian cocoa.
Sample Mean liking score Standard Deviation

(out of 10)*
Ghana 4.96 2.21
Heap 7.42 1.28
Box 6.41 1.89
Barrel 4.83 2.73
* A score of 0 indicated that the panelist strongly disliked the chocolate sample, while 10 indicated that
the panelist strongly liked the sample.
Table 3.12 - Student’s T-test analysis of liking scores for chocolate made from Mossman cocoa
fermented by heap, box and barrel methods, compared to chocolate made from Ghanian cocoa.
Chocolate samples T t* T-test result at 95% confidence level: T > t

compared (Direction of difference for one tailed test)**
Ghana, vs. Significantly different

4.71 1.68
(Heap > Ghana)
Heap

2.17 1.68
(Box > Ghana)
Box
Ghana, vs.
0.160 1.71 Not significantly different
Barrel
Heap, vs. Significantly different

1.71 1.70
(Heap > Box)
Box
Heap, vs. Significantly different

3.32 1.72
(Heap > Barrel)
Barrel
Box, vs. Significantly different

1.84 1.71
(Box > Barrel)
Barrel
* One tailed critical value.
** Significant difference is established if T > t. The direction of any significant difference between mean
liking scores is described in parentheses.
129
By combining these data, the chocolate samples may be ranked according to

their mean liking scores, as follows:
Ranked order of chocolate samples: Heap > Box > Ghana ~ Barrel
Mean liking scores: 7.42 > 6.41 > 4.90 ~ 4.83
These results demonstrated a basic acceptability of the chocolate produced from

the cocoa beans grown in Queensland, Australia, compared to the chocolate
made from Ghanian cocoa beans.
The significant differences observed between the mean liking scores of the
different chocolate samples verifies the significant differences observed during
the triangle (difference) tests. That is, the chocolate samples received
significantly different liking scores, which suggests that the samples tasted
significantly different from one another (as demonstrated by the difference
testing).
Since fermentation of cocoa from the South Johnstone plantation gave similar
chemistry and microbiology data, only the Mossman beans were subject to
sensory evaluation. Furthermore, the Mossman cocoa beans provided a more
complete comparison between the heap, box and barrel methods, since a heap
fermentation was not performed using the South Johnstone cocoa beans.
130
3.3.2 Addition of microbial nutrients to enhance fermentation
The availability of nitrogen and vitamin nutrients can be limiting in food and
beverage fermentations. As a consequence, the rate of fermentation can be
decreased, or the fermentation may not go to completion, leading to spoilage by
species of bacteria and filamentous fungi (Bisson, 1999). To address this
possibility, commercial preparations of microbial nutrients have become
available for use in assisting food and beverage fermentations. One widely used
product is the Fermaid K preparation sold by Lallemand. The recommended
rate of addition of this product is 25g per 100 kg (or 100 L) of food fermentation.
This section describes the addition of Fermaid K to cocoa beans at rates of 25

and 50 g/100 kg of cocoa beans, to determine the effects of this
supplementation on the kinetics and microbial ecology of the fermentations.
The experiments were conducted using the barrel fermentation method only,
since this method has the greatest potential for industrialisation. Cocoa beans
from both Mossman and South Johnstone plantations were used in these
experiments.
131
Figures 3.21 and 3.22 show changes in the total populations of yeast and
bacteria, along with changes in temperature and pulp pH of the fermentations.
^ 7 -
U 40 -
2 35 -
g 30 -
U 40 -
2 35 -
g 30 -
U 40 -
2 35 -
2 30 -
Fig. 3.21. Effect of adding commercial microbial nutrients (Fermaid K®) on the fermentation
of cocoa beans obtained from the Mossman plantation. Fermentations were conducted in barrels
with (a) control (no added nutrients), (b) 25g Fermaid K® and (c) 50g Fermaid K®;
Total yeast population ♦; Total bacterial population □; Temperature O; pH ■.
132
a 6 -
s 6 -
Fig. 3.22. Effect of adding commercial microbial nutrients (Fermaid K®) on the fermentation
of cocoa beans obtained from the South Johnstone plantation. Fermentations were conducted in
barrels with (a) control (no added nutrients) and (b) 25g Fermaid K®; Total yeast population ♦;
Total bacterial population □; Temperature O; pH ■.
At the commencement of fermentation, the temperature of the Mossman cocoa

beans was 25-27°C and after 72-96 hours it had reached a maximum of 47°C in
the control, 45.5°C in the barrel with 25g Fermaid K, and 47°C in the barrel with
50g Fermaid K (Figure 3.21 a, b, c). After reaching these maximum values, the
temperature remained fairly steady until the end of fermentation (120h).
At the beginning of fermentation, the pH of the cocoa pulp was 3.9 and
gradually increased to a final value of 4.1. The rate of change of pH differed
little between the treatments (Figure 3.21). This increase in the pH of the pulp
(by pH 0.2) was much smaller than typically reported, and suggested that the
level of acids in the pulp decreased only slightly during fermentations
133
The initial populations of both yeast and bacteria were around 105 cfu/ g beans.
Maximum populations of yeasts (107cfu/g beans) were reached at 24 h, after
which the yeasts died off.
Maximum populations of bacteria(108 cfu/g beans) were reached after 48 hours.

The total population of bacteria remained between 107-108 cfu/g until the end
of fermentation (120h).
Similar trends were observed when the experiments were repeated using beans
from the South Johnstone plantation (Figure 3.22). Due to limited supply of
cocoa beans, only the control (no added Fermaid K) and 25g Fermaid K®
treatments were repeated. As before, both fermentations reached a maximum
temperature of 46°C and then remained steady until the end of fermentation.
Maximum temperatures were reached 24 hours earlier in these fermentations,
compared to those using cocoa from the Mossman plantation. Furthermore the
pH of the South Johnstone beans was slightly higher at Oh (pH 4.0) and was
observed to increase to a greater extent by the end of fermentation (pH 4.6)
(Figure 3.22 a, b).
In the fermentations using cocoa beans from South Johnstone, bacteria reached
maximum populations (108cfu/g beans) after 48 hours and remained at this
level until the conclusion of fermentation. The populations of yeasts again
peaked between 24-48 hours, and then declined. In contrast to the fermentations
using cocoa beans from the Mossman plantation, this decline was not complete,
and at 120 hours the yeast population increased again (compare Figure 3.21 to
Figure 3.22).
In summary, the general profiles of cocoa bean fermentations appeared very

similar, regardless of the amount of Fermaid K® added to the fermentation.
134
33.2.2 Changes in the population of yeast species during fermentation with

added microbial nutrients
The growth of individual yeast species during fermentation with the addition of
microbial nutrients are shown in Figure 3.23.

Fig. 3.23. Effect of adding commercial microbial nutrients (Fermaid K®) on growth of yeast
species during the fermentation of cocoa beans. Fermentations were conducted in barrels with
(a) control (no added nutrients) [Mossman beans], (b) 25g Fermaid K® [Mossman beans], (c)
50g Fermaid K® [Mossman beans], (d) control (no added nutrients) [S.Johnstone beans] and (e)
25g Fermaid K® [S.Johnstone beans]: Hanseniaspora guiUiermondii ♦; Saccharomyces
cerevisiae □; Issatchenkia orientalis O; Pichia membranifaciens A
135
There were four species of yeasts predominant in all fermentations as shown in

Figure 3.23. They were Hanseniaspora guiltiermondii, Saccharomyces cerevisiae,
Issatchenkia orientalis and Pichia membranifaciens.
In the Mossman cocoa beans, all species of yeasts reached maximum
populations after 24-48 hours of fermentation, and then declined to varying
extents. H.guiltier mondii recorded maximum populations of 107 cfu/g beans,
while S.cerevisiae, 1.orientalis and P.membranifaciens all recorded maximum
populations of 105 - 106 cfu/ g beans (Figure 3.23 a, b, c). At the end of
fermentation (120h), S.cerevisiae was only detected in those fermentations to
which Fermaid K® had been added, but not in the control. Furthermore,
H.guiltier mondii and I.orientalis survived 24 hours longer in the fermentations
using Fermaid K® compared to the control.
When the experiments were repeated using cocoa beans from South Johnstone,
similar results were obtained, with some variations. Firstly, the initial
populations of all species were similar to those observed in the Mossman beans
at Oh (Figure 3.23; compare a, b & c with d & e).
In the fermentations using South Johnstone cocoa beans, H. guiltier mondii and
P.membranifaciens reached maximum populations after 24h of fermentation, with
maximum populations of 107cfu/g and 105~106 cfu/g beans, respectively. After
reaching these maxima both species decreased to undetectable levels
(<102 cfu/g). This decline was faster in the South Johnstone beans than in the
Mossman beans (Figure 3.23; compare a, b & c with d & e).
S.cerevisiae reached maximum populations of 106cfu/g beans in the
fermentation to which 25g of Fermaid K® had been added. The control only
gave a maximum S.cerevisiae population of 104cfu/g beans. S.cerevisiae also
survived for 24 hours longer, in the fermentation to which 25g Fermaid K® had
been added, than in the control (Figure 3.23 d, e).
Finally, I.orientalis was also detected in the unfermented South Johnstone beans,
and throughout fermentation. It reached a higher maximum population in the
control (106 cfu/g beans), compared to the fermentation with 25g Fermaid K®
added (105 cfu/g beans). At the end of fermentation, I.orie?italis was the
dominant yeast species, accounting for the increase in total yeast population
during the last 24 h (Figure 3.23 d, e).
136
3.3.2.3 Changes in the population of bacterial species during fermentation

with added microbial nutrients
Figure 3.24 shows the changes in population of the four species of bacteria that
were predominant in fermentations conducted using different amounts of

added Fermaid K®.

Fig. 3.24. Growth of bacterial species - Effect of adding commercial microbial nutrients
(Fermaid K®) on the fermentation of cocoa beans. Fermentations were conducted in barrels
with (a) control (no added nutrients) [Mossman beans], (b) 25g Fermaid K® [Mossman beans],
(c) 50g Fermaid K® [Mossman beans], (d) control (no added nutrients) [S.Johnstone beans] and
(e) 25g Fermaid K® [S.Johnstone beans]. Acetobacterpasteurianus^\ Gluconobacter oxydans
□ ; Lactobacillusplantarum A; Lactobacillusfermentum O.
137
The four predominant species of bacteria were: Acetobacter pasteurianus,

Gluconobacter oxydans, Lactobacillus plantarum and Lactobacillus fermentum.
In the fermentations using beans obtained from the Mossman plantation,

bacteria were initially detected at levels of 103-105 cfu/ g beans. L. plantarum and
G. oxydans reached maximum populations of 108 cfu /g beans after 48 hours, in
all fermentations. The population of these species then declined.
(Figure 3.24 a, b, c)
L. fermentum reached maximum populations (107 cfu/g beans) in all

fermentations after 24h. By 72h, L. fermentum had fallen to undetectable levels
(<102 cfu/g) in the fermentations to which Fermaid K had been added. By
contrast, in the control fermentations L. fermentum reappeared at a population
of 107cfu/g beans at the end of fermentation (120h).
Regardless of the amount of Fermaid K added, A. pasteurianus grew in a similar

fashion in all fermentations using cocoa from the Mossman plantation. At the
beginning of fermentation (Oh), A.pasteurianus was isolated at populations of
103 cfu/g. After 48 hours, the populations of A.pasteurianus had increased in all
fermentations to 106-107 cfu/g. It remained at these populations until 96h before
increasing to a final maximum of 108 cfu/g beans.
Similar data were obtained using beans from the South Johnstone plantation.
In these fermentations, higher populations of L fermentum were detected in the
control than in the fermentation to which 25g of Fermaid K had been added.
It was also observed that after reaching a maximum (107-108), the population of
L. plantarum remained steady in both fermentations (Figure 3.24 d, e). This was
in contrast to the fermentations using Mossman cocoa, in which the populations
of L.plantarum declined after 72h (Figure 3.24 a, b, c).
138
3.3.2.4 Changes in the concentrations of sugars during fermentation with

Figures 3.25 and 3.26 show the changes in concentration of these sugars during
fermentation of cocoa beans with different amounts of added Fermaid K®.
Fructose (5.3%) and glucose (4.8%) were the predominant sugars detected in
unfermented Mossman cocoa pulp (Oh). Small amounts of sucrose (1 -1.5%),
glucose (0 - 0.5%) and fructose (0.5 -1%) were detected in the seed fraction of
unfermented Mossman cocoa beans.
^ 6.0 - ^ 3.5 -
1 3.0 -
« 4.0 -
h 2.0
r 3.o -
£ 1.5
-
£ 2.0 -
^ 1.0-
0.0 t
« 4.0 -
£ 2.0 -
X 3.0 - £ 1.5 -
£ 2.0 -
^ 1.0 -
^ 6.0 -
g 5.0
4.0 -
r 3.o - 2-°
£ 1.5
'
£ 2.0 -
1.0-
0.0

Fig. 3.25. Changes in the concentration of sugars in the pulp (a, c, e) and seed (b, d, f) of
Mossman cocoa during barrel fermentation with added Fermaid K®; Negative control (a, b);
25g Fermaid K® (c, d); 50g Fermaid K® (e, f); glucose ♦; fructose □; sucrose O.
139
During fermentation, the concentrations of glucose and fructose in the pulp

decreased throughout (Figures 3.25 a, c, e). By the end of fermentation (120h) no
sugars were detected in the pulp of any of the treatments. A slight difference
was observed in the rate at which the sugars were utilised in each fermentation.
The concentrations of glucose and fructose decreased slightly faster in the
fermentations containing added Fermaid K®, than in the control.
When the experiments were repeated using cocoa from the South Johnstone
plantation, the data obtained were similar to those from the experiments using
Mossman beans. In unfermented South Johnstone cocoa beans, similar levels of
fructose (5.2%) and glucose (4.9%)were found in the pulp, while in the seed
fractions sucrose (1%), glucose (0.5%) and fructose (0.5%) were again detected.
During these fermentations, the sugars decreased at similar rates to those
observed for the Mossman cocoa (compare Figure 3.25 to Figure 3.26). Sugars
were utilised at similar rates in botfi the control fermentation and in the
fermentation with 25g Fermaid K (Figure 3.26; compare a & b with c & d).
^ 6.0 -
a 3.o -
« 4.0 -
Jt 2.0 -
'r' 3.0 -
£ 1.5 -
£ 2.0 -
* 1.0 -
1.0-
Q 6.0 -
§ 4.0 -
'r' 3.0 -
£ 2.0 -
£ 1.0 -
-m
120
Fig. 3.26. Changes in the concentration of sugars in the pulp (a, c) and seed (b, d) of South
Johnstone cocoa during barrel fermentation with added Fermaid K®; Negative control (a, b);
25g Fermaid K® (c, d); glucose ♦; fructose □; sucrose O.
140
3.3.2.5 Changes in the concentrations of ethanol during fermentation with

Figures 3.27 shows the changes in concentration of ethanol in fermentations

conducted using different amounts of Fermaid K®.
£ 5.0 - £ 5.0 -
« 4.0 - a> 4.0 -
£ 2.0 - £ 2.0 -
cS 5.0 -
§ 5.0 -
£ 4.0 -
4.0 -
£ 2.0 -
£ 2.0 -
^ 1.0 -
g 5.0 -
4.0 -
£ 2.0 -
^ 1.0 -
Fig. 3.27. Change in ethanol concentration in pulp ♦ and seed □ fractions during barrel
fermentation of cocoa beans obtained from the Mossman (a, b, c) and South Johnstone (d, e)
plantations, with added Fermaid K.®: (a, d) Control (no added nutrients), (b, e) 25g Fermaid K®
and (c) 50g Fermaid K®.
141
In the Mossman cocoa beans, no ethanol was detected at the start of

fermentation (Oh) (Figure 3.27 a, b, c). During fermentation, ethanol was
produced, with maximum concentrations of 6.5% ethanol detected in the pulp
at 48h. From 48h onwards, the concentration of ethanol in the pulp declined
sharply. At the end of fermentation, low levels (1 - 2%) of ethanol were still
present in the pulp.
As ethanol was produced in the pulp, its concentration also increased in the
seed fraction of the cocoa beans, reaching maximum levels between 48-72h. In
all of the fermentations using Mossman beans, the maximum concentration of
ethanol found in the seeds was 3% (± 0.2).
Similar trends were found during fermentation of beans from the South
Johnstone plantation (Figure 3.27). At the start of these fermentations, traces
(< 1%) of ethanol were detected in the pulp, probably caused by rapid
fermentation after pod splitting, but before sampling and freezing of the beans.
Ethanol was produced during fermentation and reached maximum levels at 48h
for both treatments. The maximum concentration of ethanol recorded in the
fermentation with 25g Fermaid K was 6.5%, while in the control it was 5.8%
(Figure 3.27 d,e).
In the seeds of the South Johnstone cocoa beans, maximum concentrations of

4% ethanol were obtained between 48-72h in both fermentations. This was
higher than that observed in the seeds of the fermenting Mossman cocoa
(compare to Figure 3.27; a, b, c). After reaching maximum levels, the
concentration of ethanol decreased and, by the end of fermentation, only traces
(<1%) were detected in pulp and seed of both treatments (Figure 3.27 d, e).
Generally, the addition of Fermaid K had little effect on the kinetics of ethanol
production and loss, in the pulp and seeds of the cocoa beans.
142

fermentation with added microbial nutrients.
HPLC analysis of cocoa beans samples taken from the fermentations conducted
using different amount of Fermaid K® detected levels of citric, malic, tartaric,
oxalic, lactic and acetic acids. Figures 3.28 and 3.29 show the changes in
concentration of various organic acids, while Tables 3.13 and 3.14 summarise
the changes from the beginning to end of fermentation.
60 -
X) 50 - x 50 -
x 50 - X) 50 -
20 - m 20 -
x 50 - x 50 -
^ 20 - ^ 20 -
Fig. 3.28. Changes in the concentration of organic acids in the pulp (a, c, e) and seed (b, d, f) of
Mossman cocoa during barrel fermentation with added Fermaid K®; Control (0g)(a, b);
25g Fermaid K® (c, d); 50g Fermaid K® (e, f); citric acid ♦; acetic acid □; lactic acid O; malic
acid ■.
143
-o 50 - -o 50 -
an J0 -
20 -
-o 50 -
^ 20 - 2P 20 -

Fig. 3.29. Changes in the concentration of organic acids in the pulp (a, c) and seed (b, d) of
South Johnstone cocoa during barrel fermentation with added Fermaid K®; Control (a, b); 25g
Fermaid K® (c, d); citric acid ♦; acetic acid □;
lactic acid O; malic acid ■.
The concentration of some organic acids remained constant throughout the

fermentations, regardless of the amount of Fermaid K® added or the source of
cocoa. These are not shown in Figures 3.28 & 3.29, and are listed as follows:
- in the pulp, tartaric (5 mg/g) and oxalic (10 mg/g) acids;
- in the seeds, malic (5 mg/g), tartaric (5 mg/g) and oxalic (5 mg/g) acids.
During the fermentations using Mossman cocoa beans, the concentration of

citric acid in the pulp decreased between 48-72h from 45 mg/g to 35 mg/g,
before increasing again to a final concentration of about 50 mg/g.
The concentration of malic acid increased slightly from initial levels of 10 mg/g
(Oh) to maximum concentrations at the end of fermentation (120h). The highest
concentration of malic acid in the pulp was recorded for the control (17 mg/g),
144
followed by the fermentations with 25g Fermaid K® (14 mg/g) and 50g
Fermaid K® (11 mg/g).
Lactic and acetic acids were produced in the pulp during these fermentations,
their concentrations increasing most rapidly between 48-120h. Maximum
concentrations of both lactic and acetic acids were detected in the pulp at the
end of fermentation (120h) (Figure 3.28 a, c, e). The final concentration of acetic
acid in the pulp was similar for all treatments (approximately 50 mg/g). In
comparison, the final concentration of lactic acid was lower in the control
fermentation (61 mg/g) compared to the fermentations to which Fermaid K®
had been added (71-72 mg/g) (Table 3.13).
In the seeds of the Mossman cocoa beans, the concentration of citric acid
decreased as fermentation proceeded from initial levels of 15 mg/g to
10 mg/g at the end of fermentation.
As acetic and lactic acids were produced in the pulp, their concentrations also
increased in the cocoa seeds (Figure 3.28 b, d, f), reaching maximum levels at
the end of fermentation (120h). The final concentration of acetic acid in the
seeds was approximately 30 mg/g for all treatments. As was observed in the
pulp, the final concentration of lactic acid in the seeds of the control
fermentation (11 mg/g) was slightly lower than in the fermentations to which
Fermaid K® was added (both 15 mg/g) (Table 3.13).
Similar trends were observed when the fermentations were conducted using
cocoa from the South Johnstone plantation (Figure 3.29).
The initial levels of citric acid in the pulp of the South Johnstone cocoa beans
were lower (30 mg/g) than in the Mossman cocoa beans (40 mg/g) (Tables 3.13,
3.14). During fermentation, the concentrations of citric acid in the pulp
increased slightly before decreasing again to final levels of 25 mg/g in all
fermentations (Figure 3.29).
145
(120h) of fermentation of Mossman cocoa beans using the barrel method and with different
amounts of Fermaid K® added.

(Oh) (120h) (Oh) (120h)
Control Citric 41 53 12 16.0 9.7 -6.3
Malic 8.1 17 8.9 5.2 5.0 -0.2
Tartaric 5.0 5.0 0 4.8 4.8 0
Oxalic 10.0 10.0 0 4.8 4.7 -0.1
Lactic 2.3 61.0 58.7 4.6 11 6.4
Acetic 0.0 51.0 51 0.0 30 30
25g Fermaid K.® Citric 53.0 58.0 5 15.0 9.0 -6
Malic 8.0 14 6 5.0 5.6 0.6
Tartaric 5.1 4.3 -0.8 4.9 5.0 0.1
Oxalic 10.0 10.0 0 4.1 4.8 0.7
Lactic 2.5 73.0 70.5 1.9 15.0 13.1
Acetic 0.0 54.0 54 0.0 31.0 31
50g Fermaid K.® Citric 47.0 51.0 4 15.0 11.0 -4
Malic 8.4 10 1.6 5.4 5.5 0.1
Tartaric 4.8 5.0 0.2 5.1 5.0 -0.1
Oxalic 9.9 10.0 0.1 5.1 5.6 0.5
Lactic 2.9 72.0 69.1 3.4 15.0 11.6
Acetic 0.0 48.0 48 0.0 28.0 28
The concentration of malic acid increased in the fermentations from 10 mg/g at

Oh to 19 mg/g in the control and 13 mg/g in the fermentation with 25g
Fermaid K® added.
Lactic and acetic acids were produced in the pulp during fermentation,
reaching maximum concentrations at 120h. The final concentrations of lactic
and acetic acids were close to 45 mg/g in both fermentations, with only minor
variations (Table 3.14). However, the acids were produced at different rates,
depending on whether Fermaid K® was used.
146
In the seeds of the South Johnstone cocoa beans, the concentration of citric acid
decreased, while acetic and lactic acid increased, as the fermentations
proceeded. Except for acetic acid, the final concentration of organic acids in the
seeds was similar for both treatments (Table 3.14). The final concentration of
acetic acid in the control (21 mg/g) was higher than the fermentation with 25g
of Fermaid K® added (16 mg/g).
The total concentration of acids in the seeds at the end of fermentation was
slightly higher for cocoa beans from the Mossman plantation (~70 mg/g) than
for the cocoa beans from the South Johnstone plantation (~60 mg/g).
Overall, the addition of Fermaid K® appeared to slightly increase the

production of lactic and acetic acids in the pulp, and resulted in slightly higher
levels of lactic acid in the seeds. While small, these differences were greater than
the variability between the duplicate samples, and were probably significant.
(120h) of fermentation of South Johnstone cocoa beans using the barrel method and with
different amounts of Fermaid K® added.


(Oh) (120h) (Oh) (120h)
Control Citric 28 23 -5 14 11 -3
Malic 9.0 19 10 6.5 6.0 -0.5
Tartaric 5.3 4.7 -0.6 4.9 4.9 0
Oxalic 10 9.9 -0.1 5.0 5.1 0.1
Lactic 7.8 42 34.2 4.6 17 12.4
Acetic 0.0 44 44 0.0 21 21
25g Fermaid K.® Citric 30 26 -4 15 11 -4
Malic 9.5 13 3.5 5.7 5.2 -0.5
Tartaric 4.9 4.6 -0.3 4.9 4.9 0
Oxalic 10 10 0 4.9 5.1 0.2
Lactic 6.8 43 36.2 4.1 20 15.9
Acetic 0.0 48 48 0.0 16 16

147
33.2.7 Quality evaluation of beans fermented with different amounts of

added microbial nutrients (Fermaid K®).
After fermentation, beans from the Mossman plantation were dried by two
methods. To provide comparison with previous experiments, half of these beans
were dried by the oven method. The remainder of the beans were initially
placed outside to undergo solar drying. However, after 2 days the onset of
heavy rain necessitated that drying be completed in the oven. The weather
permitted the fermented South Johnstone beans to be completely dried using
the solar method. In order to permit comparison to the Mossman beans, a
portion of the South Johnstone beans was also dried using the oven method.
The fermented, dried cocoa beans were tested for several quality parameters
that are routinely used in commercial quality testing of cocoa beans. As
previously, this testing was performed at Cadbury-Schweppes laboratories in
Singapore. For verification purposes, the cut test was repeated in the
laboratories of UNSW.
Figure 3.30 shows data for the fermentation index of the dried fermented cocoa
beans from both the Mossman and South Johnstone plantations. The results of
the commercial quality testing are presented in Tables 3.15 and 3.16. When
making comparisons of the various quality parameters, measures of significant
difference were determined using the Student's T-test (95% confidence interval)
(data not shown).
The results of the cut tests indicated that the addition of microbial nutrients had
no discernible effects on the fermentation indices of the cocoa beans. Mossman
beans fermented with 25g or 50g of added Fermaid K® had a similar proportion
of fully brown beans as the control (~30%). Likewise S. Johnstone beans
fermented with 25g of Fermaid K, and solar dried, had a similar proportion of
fully brown beans to the solar dried control beans (Figure 3.30).
148
100
Control (oven) '25' (oven) '50' (oven) Control(solar&oven) ’25'(solar&oven) '50' (solar&oven)
Control (oven) '25' (oven) Control (solar) '25' (solar)
Fig. 3.30. Cut test results for cocoa beans fermented with added microbial nutrients (Fermaid
K®); a) Mossman beans; b) South Johnstone beans. Control - beans fermented without
Fermaid K®, ‘25’ - beans fermented with 25g Fermaid K®, and ‘50' - beans fermented with
50g Fermaid K®; H % beans fully brown, beans part brown/purple.
The beans fermented with 25g of Fermaid K®, and oven dried, had a
significantly higher proportion of fully brown beans than the oven dried,
control fermentation beans. This was likely caused by this sample of beans
being located in an, overheated section of the oven, as temperatures of >60°C
have been shown to accelerate breakdown of cocoa pigmentation (Kyi et ah,
2005). Excluding this point as an anomaly, the cut test also indicated that the
South Johnstone beans had a higher mean fermentation index (53% fully brown
beans) compared to the Mossman beans (34% fully brown beans).
In this experiment, it appeared that drying method had little effect on the
fermentation indices of the beans. For the Mossman beans, it was determined
that, at the 95 % confidence level (calculations not shown), there was no
significant difference between the fermentation indices of the oven dried beans
and the beans dried by a combination of solar and oven methods. In the case of
the South Johnstone beans, the anomaly in the cut test data prevented analysis
of the effect of drying method on the fermentation indices of these beans.
149
Finally, the data indicated that on the basis of the cut test, all of the samples
were commercially acceptable, with no sample containing mouldy or insect
infested beans, and all samples having <1% slaty beans and <20% fully purple
beans (Figure 3.30) (Wood and Lass, 1985).
Table 3.15 - Quality evaluation of Mossman cocoa beans fermented with different amounts of
added Fermaid K®.
Combination of solar and

Quality parameter Oven drying
oven drying
Control '25'* '50'* Control '25' '50' Mean St. Dev.
Bean size - no. per lOOg 92 93 90 94 98 104 95.17 5.08
% Bean moisture 4.9 5.2 5.5 6.3 6.1 5.9 5.65 0.55
% Nib moisture 4.7 4.6 4.6 5.6 5.3 5.5 5.05 0.47
% Shell content 12.4 12.9 13.5 13 12.2 12.2 12.70 0.52
% Nib fat (dry basis) 53.7 55.2 55.8 54.7 54 54.4 54.63 0.77
Nib pH 4.8 4.9 4.8 4.9 4.9 5 4.88 0.1
External appearance acceptable acceptable acceptable acceptable acceptable acceptable ... ...
Aroma assessment acceptable acceptable acceptable acceptable acceptable acceptable ... ...
* ‘25’ Beans fermented with 25g Fermaid K®; ‘50’ Beans fermented with 50g Fermaid K®
Table 3.16 - Quality evaluation of South Johnstone cocoa beans fermented with different
amounts of added Fermaid K®.
Control '25'* Control '25'*

Quality parameter Mean St. Dev.
(oven) (oven) (solar) (solar)
Bean size - no. per 1 OOg 134 128 119 110 122.2 9.71
% Bean moisture 6.0 6.2 6.8 6.8 6.48 0.57
% Nib moisture 4.8 4.8 5.9 5.8 5.46 0.48
% Shell content 13.5 13.2 12.8 12.7 12.9 0.26
% Nib fat (dry basis) 55 55.2 54.8 54.8 54.8 0.83
Nib pH 5.4 5.2 5.4 5.4 5.32 0.1
acceptable acceptable acceptable acceptable ... ...

External appearance
Aroma assessment acceptable acceptable acceptable acceptable ... ...
* ‘25’ Beans fermented with 25g Fermaid K®; ‘50’ Beans fermented with 50g Fermaid K®
Comparison of the data in Tables 3.15 and 3.16 indicated that addition of
Fermaid K® did not cause any significant changes to bean size, moisture
150
content, shell content, fat content or nib pH. When the cocoa beans used were
obtained from the Mossman plantation, drying method made no significant
difference to any of the quality parameters. For the South Johnstone beans,
solar drying led to significantly higher moisture content, and larger beans with
a lower % shell, than the oven dried beans.
Taken together, these data also suggested that the combination drying method
(solar and oven) is probably no different to oven only drying, in its effects on
bean quality parameters.
It was noted that some of the quality parameters for the dried fermented
Mossman cocoa beans were significantly different to those obtained for the
dried fermented South Johnstone beans, as follows:
- The mean size of the Mossman beans was significantly smaller than the
South Johnstone beans, and
- The mean pH of the Mossman cocoa nibs (4.88) was significantly lower
than the mean pH of the South Johnstone nibs (5.32).
These data suggested that the source of beans (Mossman or South Johnstone)
could have affected the quality of the beans. Given the small sample size and
the presence of anomalies, further study is recommended to confirm this
observation. Finally, the Australian samples were found to meet the prescribed
ideal values for moisture content (bean and nib), fat content, appearance and
aroma parameters (c.f Table 3.9).
However, in some parameters the beans were less than ideal. The South
Johnstone beans had nib pH values close to the recommended value of 5.5, but
were very small (>110 beans/100 g). The Mossman beans were of good size, but
had a pH lower than the recommended level. While most of the data indicated
that the Australian beans were commercially acceptable, the size and pH data
suggest that further improvements to harvest, fermentation and drying could
be made.
151
3.3.2.8 Sensory evaluation of chocolate made from cocoa beans fermented

using added microbial nutrients
The dried fermented Mossman cocoa beans were made into chocolate samples
and subjected to sensory evaluation. Only the Mossman beans were used
because they represented a more complete comparison of the different
treatments. The triangle test was performed to determine if the addition of
Fermaid K caused any significant difference to the flavour of the cocoa beans.
Table 3.17 summarises the results of this difference testing.
Table 3.17 - Sensory evaluation (triangle test) performed on the chocolate samples made with
Mossman cocoa fermented using the barrel method, with different amounts of Fermaid K®.

Number of
Samples correct needed for significance*
tests
November 2004 - Addition of microbial

nutrients
Control Vs. 25g Fermaid K 10 4 8
Control Vs. 50g Fermaid K 10 7 8
25g Vs 50g Fermaid K 10 5 8
different from another if the “number of correct judgments” was equal to or greater than the “number of
Chocolate samples made from cocoa fermented using 25g or 50g of

Fermaid K® did not taste significantly distinguishable from one another, or
from the chocolate made from the control beans. Thus, the addition of microbial
nutrients did not cause a significant difference in the flavour of cocoa beans.
As described previously, simple assessments were also made as to the liking of

the different chocolate samples, with reference to chocolate made from
Ghanaian cocoa beans. The Student's T-test was used to determine significant
difference between the means liking scores, at a confidence level of 95%. Tables
3.18 and 3.19 show results of statistical analyses of the liking scores.
152
Table 3.18 - Mean liking scores for chocolate made from Mossman cocoa beans fermented
using the barrel method and with different amounts of Fermaid K® added.
The Australian chocolate samples were compared to chocolate made from Ghanian cocoa.
Sample Mean liking score* Standard Deviation

(out of 10)
Ghana 5.01 2.28
Control (no Fermaid K.®) 4.78 2.02
25g Fermaid K / 100 kg beans 5.35 2.47
50g Fermaid K / 100 kg beans 4.81 2.12

Table 3.19 - Student’s T-test analysis of liking scores for chocolate made from Mossman cocoa
beans fermented using the barrel method and added microbial nutrients (Fermaid K®).
Sample Pair T t* T-test result at 95% confidence level:

T>t
(Direction of difference for one tailed test)
Ghana, vs. Control 0.335 1.68 Not significantly different
Ghana, vs. 25g Fermaid K® 0.442 1.70 Not significantly different
Ghana, vs. 50g Fermaid K® 0.294 1.68 Not significantly different
Control, vs. 25g Fermaid K® 0.691 1.70 Not significantly different
Control, vs. 50g Fermaid K® 0.034 1.70 Not significantly different
25g Fermaid K® vs. 0.647 1.70 Not significantly different

50 g Fermaid K®
Preliminary comparison of the mean liking scores suggested that the all of the
samples tested were liked to a similar extent (Table 3.18). The T-test confirmed
that there were no significant difference between the liking scores of any of the
Australian chocolate samples, compared to each other or the Ghana standard
(Table 3.19). The Australian chocolates received liking scores statistically similar
to the scores for the Ghana chocolate( ~5). These scores confirmed the basic
acceptability of the Australian cocoa beans, compared to the Ghana standard. It
may be concluded that, with respect to the improvement of chocolate flavour,
there appears to be no advantage in the addition of microbial nutrients to
Australian cocoa fermentations.
153
3.3.3 Effects of mixing frequency.

Mixing of cocoa beans has been previously recognised as an important variable
in the fermentation, able to affect bean quality and flavour. Experiments were
conducted to determine how mixing frequency affects fermentation and
resultant bean quality.
Fermentations were conducted in boxes with beans harvested from the South
Johnstone plantation. Limited supply of cocoa beans prevented the
simultaneous repetition of these experiments in barrels.
Figure 3.31 shows the changes in total populations of yeast and bacteria, as well
as changes in pH and temperature during cocoa fermentations mixed at
different frequencies. In this fermentation, the pH of both pulp and seed were
monitored.
^ 7 -
U 40 -
a 35 -
2 30 -
120 144 120 144

U 40 -
g 30 -
120 144 120 144

Fig. 3.3 1 - Effect of mixing frequency on the fermentation of cocoa beans from the South
Johnstone plantation. Fermentations were conducted in boxes and mixed every (a) 24 hours,
(b) 12 hours ; Total yeast population ♦; Total bacterial population □; Temperature ■;
pulp pH O; seed pH A.
154
The temperature and pH changes were similar for both fermentations. Fresh
(unfermented) pulp had a pH of 3.9, while the seed had a pH of 6.3. During
fermentation, the pH of the pulp increased, and the pH of the seed decreased.
Both fermentations gave final pulp and seed pH values of 4.3 and 4.6,
respectively. The seed pH started to decrease earlier, but at a slower rate, in the
fermentation mixed every 24 hours compared with the fermentation mixed
every 12 hours.
Mixing frequency appeared to have some effect on the changes in fermentation

temperature. Initially at 22°C, the temperature of the beans increased during
both fermentations. In the fermentation mixed 24 hourly, this increase occurred
at a steady rate, and by the end of fermentation a maximum temperature of
45°C was reached. In the 12 hourly mixed fermentation, the increase occurred in
stepwise fashion, and the final maximum temperature was slightly higher
(47°C) (Figure 3.31 a, b).
Both fermentations had similar initial populations of yeast and bacteria: 104
cfu/ g beans and 106 cfu/ g beans, respectively. In the fermentation mixed every
24h, bacteria and yeast steadily increased in population, peaking at 72h for
yeast, and 96h for bacteria. In the fermentation mixed every 12h, maximum
populations of yeast and bacteria were detected at 96h. The rate of growth
during the first 72h, for both bacteria and yeast, was more variable in the 12
hourly mixed box compared to the box mixed every 24h. After reaching their
maxima, the populations of yeast and bacteria steadily decreased until the end
of fermentation, with the yeasts declining more rapidly than the bacteria.
In the 12 hourly mixed box, the maximum level of yeast (5xl07cfu/g beans)
was greater than the maximum level detected in the 24 hourly mixed box (7xl06
cfu/g beans). The maximum detected level of bacteria was slightly higher in the
beans mixed every 24h (9xl08 cfu/g) compared to the 12h mixed beans (2xl08).
155
3.3.3.2 Growth of yeast species.
Figure 3.32 shows the growth of individual yeast species during fermentations
which were mixed with differing frequency. The same three yeast species were
found as during previous fermentation experiments: H.guilliermondii, S.cerevisiae
and I.orientalis.
120 144 120 144

Fig. 3.32 - Effect of mixing on the growth of yeasts during fermentation of Australian cocoa
Fermentations were conducted in boxes and mixed every (a) 24 hours, (b) 12 hours ;
Hanseniaspora guiltiermondii ♦; Saccharomyces cerevisiae □; Issatchenkia orientalisO.
At the beginning of fermentation, I.orientalis and H.guilliermondii were found in

populations of 103 cfu/ g beans, while S.cerevisiae occurred at 104 cfu/g beans.
H. guilliermondii grew, reaching a maximum population of 106cfu/ g beans by
48h. After 48 hours, H.guilliermondii decreased to non-detectable levels. Similar
growth profiles were found for this yeast species in both fermentations
I. orientalis was the next species in the succession of yeast. In both fermentations,
I. orientalis outgrew H.guilliermondii after 48h, attaining a maximum population
of about 107cfu/ g beans at 72 hours. Slightly higher populations were formed
with the 12 hourly mixing. This yeast died off faster when mixing was
performed at 24 hourly intervals compared with 12 hourly intervals.
S.cerevisiae was the last species in the succession, and grew according to a
similar pattern regardless of mixing frequency. During the first 48 hours of
fermentation, the population of S.cerevisiae actually decreased approximately 10
fold. It then increased in population to a maximum level of 10 5~ 106 cfu/g beans.
The population of S.cerevisiae then declined.
Less than 102 cfu yeasts/g beans were isolated at the conclusion of the
fermentations (144h).
156
3.3.3.3 Growth of lactic acid bacterial species.
Three species of lactic acid bacteria were isolated from these fermentations:.
Pediococcus acidilactici, Lactobacillus plantarum and Lactobacillus fermentum (Figure
3.33).
cc 5 ■ oo 5 ■
120 144 120 144

Fig. 3.33 - Effect of mixing on the growth of lactic acid bacteria during fermentation of South
Johnstone cocoa beans. Fermentations were conducted in boxes and mixed every (a) 24 hours,
(b) 12 hours; Lactobacillus fermentum ♦, Pediococcus acidilactici O; Lactobacillus plantarum □.
L.fermentum was only isolated from the 24 hourly mixed fermentation.

P.acidilactici was initially present in both fermentations at a population of 102
cfu/ g beans. In the 24 hourly mixed fermentation, P.acidilactici increased to a
maximum population of 107 cfu/g beans at 72 hours,before declining. The same
growth behaviour was observed in the fermentation mixed 12 hourly, except
that the maximum population peaked 24 hours later (96h)..
L.plantarum exhibited very similar growth profiles in both fermentations.

Initially isolated at 106cfu/ g beans, the population of L.plantarum increased to a
maximum of 108 cfu/g beans, before dying off. In the fermentation mixed at 24
hourly intervals, the maximum population was reached 24 hours earlier than in
the fermentation mixed at 12 hourly intervals.
In the fermentation mixed at 24 hourly intervals, the population of L.fermentum

rapidly increased between 24-48 hours to a maximum of about 107cfu/ g beans.
The population of L.fermentum remained steady at this level for 48 hours, before
declining.
157
3.3.3.4 Growth of acetic acid bacterial species.
Figure 3.34 shows the growth of individual species of acetic acid bacteria
during fermentations that were mixed with different frequency. Three species of
acetic acid bacteria were found in all fermentations: Acetobacter pasteurianus,
Gluconobacter oxydans, and Asaia siamensis. These species exhibited similar
growth profiles in 24 h and 12 h mixed fermentations (Figure 3.34 a,b).
*2 4 -
120 144 120 144

Fig. 3.34. Effect of mixing on the growth of acetic acid bacteria during fermentation of South
Johnstone cocoa beans. Fermentations were conducted in boxes and mixed every (a) 24 hours,
(b) 12 hours; Acetobacter pasteurianus ♦; Gluconobacter oxydans □; Asaia siamensis. O.
In both fermentations, A. siamensis. quickly grew to maximum populations of

106 cfu/ g beans after 24 hours, and then died off. A.pasteurianus and G.oxydans
grew concurrently, both reaching maximum populations after 96 hours. In the
24 hourly mixed fermentation, G.oxydans and A.pasteurianus reached maximum
populations of about 108cfu/ g beans at 96h, before declining slightly to a final
populations of about 106cfu/g beans. In the 12 hourly mixed fermentation,
G.oxydans and A.pasteurianus approached a maximum population of 107cfu/ g,
and remained at this level until the end of fermentation.
158
3.3.3.5 Changes in the concentrations of sugars

Figure 3.35 shows the changes in concentrations of sugars in the pulp and seeds
of cocoa beans fermented in boxes and mixed at different frequencies.
24 48 72 96 120 144 24 48 72 96 120 144

Time (hours) Time (hours)
24 48 72 96 120 144 24 48 72 96 120 144

Time (hours)
Fig. 3.35. Effect of mixing on the concentration of sugars in the pulp (a, c) and seeds (b, d) of
fermenting South Johnstone cocoa beans. Fermentations were conducted in boxes and mixed
every 24 hours (a, b), or 12 hours (c, d); glucose ♦; fructose □; sucrose O.
In the unfermented cocoa pulp, glucose and fructose were detected at
concentrations of 4.5% and 6%, respectively. In the seed fraction, sucrose,
glucose and fructose were detected at low concentration (all about 1%).
During both fermentations, the concentration of these sugars decreased in the
pulp and the seed. This decrease in sugar concentration was most rapid in the
fermentation which was mixed every 12 h. After 72 h, the fermentation mixed
every 12 h, contained only about 0.5% of each glucose and fructose in the pulp.
This contrasts to the the fermentation mixed every 24h that, at 72 h, contained
around 1.5% glucose and 4% fructose. The rate at which the concentration of
sugars decreased in the seeds was faster in the fermentation mixed 12 hourly
(Figure 3.35). By the end of fermentation (144 h), no sugars were detected in the
pulp or seeds of either fermentation.
159
3.3.3.6 Changes in the concentration of ethanol
Figure 3.36 shows that the ethanol was produced during fermentation,
increasing in concentration in the pulp and seeds of both fermentations during
the first 72h.
0 24 48 72 96 120 144 0 24 48 72 96 120 144

Fig. 3.36. Effect of mixing on the concentration of ethanol in pulp ♦ and seed □ fractions of
fermenting cocoa beans. Fermentations were conducted in boxes and mixed every (a) 24 hours;
(b) 12 hours.
The maximum concentration of ethanol was similar for both fermentations

(approx. 5%), and in both the pulp and seeds. After reaching maximum levels,
the concentration of ethanol decreased in the pulp and seeds, giving final
concentration of 1% (144h).
3.3.3.7 Changes in the concentration of organic acids
Citric, malic, tartaric, oxalic, lactic and acetic acids were all detected in the pulp
and seeds of cocoa beans fermented in boxes and mixed at different frequencies.
Figure 3.37 and Table 3.20 show the changes in concentration of these acids.
The pulp concentration of citric acid decreased from initial levels of 35 mg/g to
final levels of 10 mg/g in the fermentation mixed every 24h, and 21 mg/g in the
fermentation mixed every 12h. The concentration of malic acid in the pulp
increased from 5 mg / g to 20 mg/g in both fermentations. Lactic and acetic
acids were produced in the pulp during fermentation.
Maximum concentrations of acetic acid were reached at 144h, in both
fermentations. The final concentration of acetic acid in the pulp of the
fermentation mixed every 12h (53 mg/g) was nearly twice that found in the
fermentation mixed every 24h (32 mg/g) (Figure 3.37 a, c).
160
_$£> 20 - ,$p 20 -
24 48 72 96 120 144 24 48 72 96 120 144

b 30 ■ b 30-
_oo 20 - .op 20 ■
24 48 72 96 120 144 24 48 72 96 120 144

Fig. 3.37. Effect of mixing on the concentration of organic acids in the pulp (a, c) and seeds (b,
d) of fermenting South Johnstone cocoa beans. Fermentations were conducted in boxes and
mixed every 24 hours (a, b), or 12 hours (c, d); citric acid ♦;acetic acid □; lactic acid O; malic
acid ■; tartaric acid 0; oxalic acid ▲.
In the fermentation mixed every 24h, a maximum concentration of lactic acid(31

mg/g) was recorded at 72h, after which it decreased to 22 mg/g). In the
fermentation mixed every 12h, lactic acid was produced at a slower rate, and
achieved a maximum concentration at the end of fermentation (20 mg/g). In
contrast to previous fermentations, oxalic acid and tartaric acids were not
detected in the pulp.
In the cocoa seeds, the concentrations of citric acid (9 -10 mg/g), malic acid (5
mg/g), tartaric acid (1 mg/g) and oxalic acid (4 mg/g) remained constant
during fermentation, and therefore are not shown in Figure 3.27. As lactic and
acetic acids were produced in the pulp, their concentrations also increased in
the cocoa seeds, and reached similar maximum concentrations of 15 - 25 mg/g
at the end of fermentation (Figure 3.37 b, d; Table 3.20).
161
Table 3.20 - Summary of changes in concentration of organic acids from the beginning (Oh) to
end (120h) of fermentation of South Johnstone cocoa beans. Fermentations were conducted in
boxes and mixed every 24 h or 12 h.______________________________________________
method acid (mg/g dry basis) (mg/g dry basis)

(Oh) (120h) (Oh) (120h)
Mixing every 24h Citric 35 10 -25 9.7 9.1 -0.6
Malic 5 18 13 4.7 4.5 -0.2
Tartaric 0.0 0.0 0 1.0 1.0 0
Oxalic 0.0 0.0 0 3.8 3.9 0.1
Lactic 4 22 18 0.0 18 18
Acetic 0.0 32 32 0.0 14 14
Mixing every 12h Citric 35 21 -14 10.0 10 0
Malic 7.0 19 12 5.0 5.0 0
Tartaric 0.0 0.0 0 1.0 1.0 0
Oxalic 0.0 0.0 0 4.1 4.7 0.6
Lactic 3.6 20 16.4 0.0 18 18
Acetic 0.0 53 53 0.0 26 26
3.3.3.8 Quality evaluation of South Johnstone cocoa beans fermented in

boxes and mixed at different frequencies.
After fermentation, the cocoa beans were dried by two methods. To provide
comparison with previous experiments half of these beans were dried by the
oven method. The remainder of the beans were initially placed outside to
undergo solar drying. However, the onset of heavy rain after 2 days
necessitated that drying of these beans be completed in the oven.
The fermented dried cocoa beans were tested for commercial quality at
Cadbury-Schweppes laboratories in Singapore. For verification purposes, the
cut test was repeated at the laboratories of UNSW.
Figure 3.38 shows the fermentation indices of beans fermented with mixing at
different frequencies, as determined by the cut test. The results of the other
commercial quality testing are presented in Table 3.21. When comparing the
quality data, measures of significant difference were determined using the
Student's T-test (95% confidence interval) (calculations not shown).
162
100 n
03
<U
aj
60
■2
p
<u
Q.
24 hr mixing 12 hr mixing 24 hr mixing 12 hour mixing

(solar & oven) (solar & oven) (oven) (oven)
Fig. 3.38. Cut test results for cocoa beans fermented in boxes and mixed at differing
frequencies; % beans fully brown H; % beans part brown/purple H; % beans fully purple □ .
The beans from the fermentation mixed every 24 hours had a significantly
higher mean cut test score (50% fully brown beans) than those from the
fermentation mixed every 12 hours (20% fully brown beans).
Also, a significantly higher proportion of fully brown beans occurred in the

samples dried with the combination method (50%), compared to those which
were only oven dried (20% fully brown). From these data, it may be inferred
that both frequency of mixing and drying method affected the fermentation
indices of the South Johnstone cocoa beans.
Finally, on the basis of the cut test, only the beans mixed every 24h were
determined to be of acceptable commercial quality. Although no samples
contained any mouldy, insect infested or slaty beans, the fermentations that
were mixed every 12h had >20% fully purple beans present.
Data from the additional commercial quality tests revealed that mixing
frequency also affected bean size and nib pH. When the beans samples were
compared on the basis of fermentation method, it was observed that the beans
mixed every 24 h were significantly larger than beans which were mixed every
12h. The pH of the beans mixed every 24h was significantly higher than the pH
of the beans mixed every 12h. The frequency of mixing did not cause a
significant change to any other parameters.
163
Table 3.21 - Quality evaluation of South Johnstone cocoa beans fermented in boxes and mixed
at different frequencies.
24h mixing 12h mixing 12h

24h mixing
Quality parameter (solar and (solar and mixing Mean St. Dev.
(oven)
oven) oven) (oven)
% Bean moisture 7.5 7.5 7.3 7.51 7.5 0.10
% Nib moisture 6.71 6.75 6.59 6.7 6.7 0.07
% Shell content 15.5 14.5 16 16 15.5 0.71
% Nib fat (dry basis) 55.6 55.7 55.5 55.5 55.6 0.10
Nib pH 5.5 5.2 5.4 4.8 5.2 0.31
External appearance acceptable acceptable acceptable acceptable ... ...
Aroma assessment acceptable acceptable acceptable acceptable ... ...
While drying method affected the cut test results, no significant differences
were observed between the commercial quality parameters of beans dried by a
combination of solar and oven methods, compared to beans dried only by oven.
Compared to the prescribed 'ideal' industry values, the quality of these beans
was found to be mostly acceptable (c.f. Table 3.9). An exception was that the
mean shell content of the beans samples was 15.5%, significantly higher than
recommended by the literature. Furthermore, the pH of the beans mixed every
12h was lower than the ideal value of pH 5.5. This particular observation,
combined with the cut test results, suggest that mixing every 12h is too frequent
and produces beans of inferior quality. The data also suggest that, while the
beans mixed every 24h are generally acceptable, the methods for fermentation
and drying could be further optimised to decrease the final shell content of the
beans.
The beans dried using combined solar and oven methods were selected to be
made into chocolate and sensory evaluation, since their cut test scores were
higher than the beans dried by oven only.
164
3.3.3.9 Sensory evaluation of chocolate made from cocoa beans mixed at

different frequencies.
Chocolate samples made from the dried fermented cocoa beans were subjected
to the triangle test. This test was used to determine whether the frequency of
mixing during fermentation had significantly effects on the flavour of the cocoa
beans. The results of this testing indicate that chocolates made from cocoa
mixed every 24h tasted significantly different from chocolate made from beans
mixed every 12h, on the basis of flavour (Table 3.22).
Table 3.22 - Sensory evaluation (triangle test) performed on chocolate made from cocoa beans
fermented in boxes and mixed at different frequencies.
Number of Number of Correct judgements

Samples tests correct needed for significance
June 2005 - Frequency of mixing
Mixing every 24h Vs. every 12h 32 22 18
different from another if the “number of correct judgments” was equal to or greater than the “number of
As before, panelists also scored the different chocolate samples on the basis of
liking. The liking scores were collated, and then analysed using the Student's T-
test (95% confidence interval) to determine any significant difference between
the mean scores (Tables 3.23 and 3.24).
Table 3.23 - Mean liking scores for chocolate made from cocoa beans fermented with mixing at
different frequencies, compared to chocolate made from Ghanian cocoa.
Sample Mean liking score Standard Deviation
(out of 10)*
Ghana 5.48 1.84
Mixing every 24h 3.65 2.11

Mixing every 12h 4.4 2.24
165
Table 3.24 - Student’s T-test analysis of liking scores for chocolate made from South Johnstone
cocoa beans fermented in boxes and mixed at different frequencies.
T-test result at 95% confidence level: T > tcritical

Sample Pair T tcritical
(Direction of difference for one tailed test) **

2.21 1.75
(Ghana > Mixing 24hrly)
Mixing 24hrly

2.06 1.75
(Ghana > Mixing \2hrly)
Mixing 12hrly
Mixing 24hrly, vs.

Mixing 12hrly
The mean liking scores of Ghana chocolate were, significantly greater than for
either of the Australian bean chocolates (12 or 24 hour mixing). The data also
indicated that chocolate made from 24h mixed beans was not liked significantly
more than chocolate made from the 12h mixed beans. Furthermore, the mean
liking scores for both Australian samples were significantly less than 5.
Together, the liking data indicated that chocolate made from cocoa mixed at
different frequencies had a less than acceptable flavour.
While the lower cut test and nib pH data suggested that beans mixed every 12h
would be of inferior quality, it was somewhat unexpected that the beans mixed
every 24h would also have an inferior flavour.
166
3.4 Discussion
This chapter reports, for the first time, fermentation of cocoa beans cultivated in
North Queensland, Australia, and shows that commercially acceptable
chocolate can be obtained from the fermented beans. Baseline information on
the microbiology of the fermentation, and on changes in the chemical
composition of the beans during fermentation has been determined. The effects
of several variables on the conduct of the fermentation and final cocoa quality
were also examined: Fermentations by heap and box were performed and
compared to a barrel fermentation. In order to determine if microbial nutrients
could improve cocoa fermentation, experiments with the addition of Fermaid K
were performed. Finally, given the importance of aeration for successful
fermentation, experiments were performed to determine the effects of mixing
frequency.
3.4.1 Physical changes during fermentations
Physical changes observed during fermentation included increases in

temperature, changes to the pH of the pulp and seeds, changes to the mass and
water content of the beans, and changes to the appearance and aroma of the
beans.
3.4.1.1 Temperature
During fermentation, the temperature of the bean masses increased due to the
metabolic activities of microorganisms in the pulp, and of the cocoa seeds
themselves. Maximum temperatures of 43-48°C were reached by 60-72h. After
reaching maxima, the temperatures then remained steady, or decreased to
35-43°C. The maximum temperatures of the fermentations compared well with
those reported in Ghana (Nielsen et al., 2007b; Camu et al., 2008a) but were
lower than the 50°C reported for commercial cocoa fermentations conducted in
Indonesia (Ardhana and Fleet, 2003) and Brazil (Schwan, 1998). The lower
temperatures were probably caused by the smaller masses of beans fermented
(75-100 kg), compared to the 100-1000 kg of commercial fermentations (Rohan,
1963; Nielsen, 2007b).
Slight variations in the temperature profiles were observed when cocoa beans
were fermented by heap, box or barrel. In particular, higher temperatures were
167
achieved in the heap compared to the box. These results were similar to those of
Said and Samarakhody (1984), Portillo et al. (2005) and Nielsen et al. (2007b),
where temperature was also affected by fermentation method (boxes of
different surface areas and depths; tray, heaps).
Mixing cocoa fermentations at different frequencies also affected the
temperature. Compared to the fermentation mixed every 24 hours, mixing
every 12 hours resulted in a lower maximum temperature, and an erratic rate of
temperature increase. Similar effects of mixing on fermentation temperature
were recorded by Passos et al. (1984b), Said and Samarakhody (1984), Portillo et
al. (2005) and Camu et al. (2007, 2008a). The literature proposes that while
mixing is needed to initiate the temperature increase, too frequent mixing
increases cooling, and interrupts the exothermic reactions (Wood and Lass,
1985)
The addition of microbial nutrients (Fermaid K®) did not affect the temperature
changes during fermentation.
Small variations in temperature were observed across all fermentations,
probably reflecting seasonal or regional differences in the beans, and slight
differences in the microbiology and biochemistry of the process. During these
experiments, the effect of changing ambient conditions was minimised by
conducting all the fermentations in a temperature controlled room, as described
in Methods, Section 3.2.1.2.
The development of chocolate flavour precursors requires temperatures greater
than 40°C for more than 24h, and several studies have suggested that
temperatures above 45°C are optimal (Biehl et al., 1977; Biehl and Voight, 1996).
All fermentations conducted in this study met these conditions.
3.4.2.2 pH
The pH values of fresh cocoa pulp (3.5 - 3.9) and seeds (6.1 - 6.2) were consistent
with the general ranges reported in the literature (Rohan, 1963; Wood and Lass,
1985; Schwan, 1998; Ardhana and Fleet, 2003; Camu, 2007, 2008a, b).
During all of the fermentations, the pH of the pulp increased, while the pH of
the seed decreased. The final pH was typically between 4.0 - 5.0 for the pulp,
and 6.0 - 5.0 for the seed. Similar final values have been described in Brazil
(pulp pH 5.0; Schwan, 1998), Ghana (pulp pH 4.2 - 4.5; Nielsen et al. 2007b), Sri
168
Lanka (pulp pH 5.0, seed pH 5.5; Senanayake et al. 1995) and Indonesia (pulp
pH 3.9 - 4.9, seed pH 5.0 - 5.1; Ardhana and Fleet, 2003). The changes in pH are
caused by the microbial utilization of citric acid in the pulp, and the microbial
production of lactic and acetic acid, which then diffuse into the seed (Roelofsen,
1958; Biehl, 1984; de Brito et al., 2000, Schwan and Wheals, 2004; Camu et al.
2007).
Fermentation by heap, box or barrel affected changes in the pH of the cocoa
pulp. The heap fermented beans had the highest pulp pH of all three methods.
Beans fermented by box had the next highest pulp pH, while beans fermented
in barrels had the lowest pulp pH. The observed differences in pH were linked
to differences in the concentrations of citric, lactic and acetic acids in the beans.
These results agreed with previous work by Carr et al.(1979) and Tomlins et al.
(1993), who found that fermentation in heaps gave beans with higher pH than
beans fermented in boxes. These studies attributed this to an increased aeration
in the boxes, leading to increased production of acetic acid.
The addition of Fermaid K did not significantly affect the pH changes observed
during fermentation.
Mixing frequency affected the rate of pH changes during fermentation. In the
fermentation mixed every 24 hours, the pH of the seed decreased more slowly
than in the fermentation mixed every 12 hours. Such a rapid decrease in the pH
of the seed is undesirable, since it may inhibit development of flavour pre
cursors (Biehl, 1984; Biehl and Voigt, 1996). The finding that both fermentation
configuration and mixing frequency affected cocoa bean pH agreed with work
by Said and Samarakhody (1984), Duncan et al. (1989), and Senanayake et al.
(1995).
Slight differences were observed in the pH of fresh pulp and beans from the
Mossman and South Johnstone plantations (Table 1, Appendix B). Specifically,
the pH of pulp (X= 3.7) of cocoa beans harvested from the Mossman plantation
was consistently lower than the pH of pulp (X= 3.9) from South Johnstone
beans, suggesting regional effects.
3.4.2.3 Appearance, water content and pulprseed ratio
The physical appearance of the cocoa beans changed during fermentation. The
pulp decreased in quantity, and became darker and drier. The seeds, initially
169
firm in texture and having a strong purple colour, became soft and pale purple
before turning brown. In the later stages of fermentation, a brown liquid was
exuded when the cocoa beans were cut in half. These observations were
consistent with those made for cocoa fermentations by previous workers (Wood
and Lass, 1985; Ardhana and Fleet, 2003; Schwan and Wheals, 2004; Nielsen,
2006). Interestingly, during box or heap fermentations, it was noticed that cocoa
beans near the edges of the bean mass appeared dry and/or mouldy. By
contrast, beans fermented in the barrel had a very uniform appearance, with no
noticeable drying or mould growth. By preventing mould growth during
fermentation, the use of a barrel vessel may improve the quality of the beans
produced.
The water content of the pulp decreased and that of the seed increased as the
fermentations proceeded. This was consistent with observations made by
Roelofsen (1958), Biehl et al. (1982b) and Ardhana (1990). Most of the water was
lost from the pulp as it liquified and drained away during the first few days of
fermentation. Water uptake by the seed is necessary to initiate the endogenous
biochemical reactions responsible for the formation of chocolate flavour
precursors (Biehl et al., 1982b; Biehl and Voigt, 1996).
Changes to the pulp-bean ratio confirmed the visual observation that the
quantity of pulp decreased during fermentation (Table 1, Appendix B). It also
revealed differences between cocoa from the two different plantations: for beans
from the Mossman plantation, the pulp represented 45-75% of the total bean
mass; for beans from the South Johnstone plantation, the pulp represented
40-45% of the total bean mass. For fully fermented beans, the proportion of
residual pulp was 20-30% of the total bean mass, which agreed with data given
by Wood and Lass (1985), Ardhana (1990) and Schwan et al. (1995).
In general fermentation configuration (heap, box or barrel), addition of Fermaid
K, and mixing frequency did not significantly affect the changes in water
content or pulp: seed ratio.
3.4.3 Microbiological changes during fermentations
The present study is novel in providing the first descriptions of the microbial
ecology of cocoa bean fermentation in Australia. Species of yeast, and lactic and
acetic acid bacteria, are commonly found in cocoa fermentations around the
170
world (Schwan and Wheals, 2004), and similar species were also isolated from
the Australian fermentations. Common trends were observed in the changes of
populations of individual species of yeast and bacteria, during the
fermentations. The most rapid growth of yeasts and bacteria occurred during
the first 24-48h of the fermentations, after which the microbial populations
remained steady or decreased. The microbiology of the fermentation was
affected by changes to parameters such as mixing, and the type of fermentation
vessel used.
3.4.3.1 Yeast species
Yeast species grew in all of the cocoa fermentations examined, with the most
rapid growth occurring in the first 24-48h. Quantitatively significant
populations (107-108 cfu/ g) were reached after this time, indicating that the
yeasts and their metabolic activities are important in influencing the final
quality and flavour of the beans.
Four species of yeast were consistently isolated from the Australian
fermentations: Hanseniaspora guillermondii, Issatchenkin orientalis, Saccharomyces
cerevisiae and Pichia membranifaciens. A further six species were isolated
sporadically, usually in the first 24h of fermentation, but did not persist in any
of the fermentations. These species were: Hanseniaspora uvarum, Pichia burtonii,
Pichia anomala, Candida sorbisivorans, Candida quercitrusa and Zygoascus hellenicus.
This diversity was similar to that described in recent studies of cocoa
fermentations in Indonesia (Ardhana and Fleet, 2003), Ghana (Jespersen, 2005;
Nielsen et al., 2007b), Brazil (Schwan, 1998) and the Dominican Republic
(Lagunes-Galvez et al., 2007).
The yeast species grew according to a common pattern of succession in all
fermentations. Hanseniaspora guillermondii dominated the first 24-48h of
fermentation and then died off rapidly. This was followed by the growth of
I.orientalis and S.cerevisiae, usually concurrently, that typically reached
maximum populations between 48 - 72 hours. These species are quite
temperature and ethanol tolerant (Okuma, 1986; Alexandre et al., 1994;
Kurtzman and Fell, 1998; Pina et al., 2004) and they persisted in the
fermentations much longer than H.guillermondii.
171
The growth pattern observed in this study was similar to that found in Ghana
(Jespersen et al. 2005), although the populations of S.cerevisiae were higher in
the Ghanaian fermentations. There were also similarities to the Indonesian
fermentations described by Ardhana and Fleet (2003), except that in that study
I.orientalis (Candida krusei) was not detected and other Candida spp. dominated.
In other countries, such as Malaysia (Carr et al. 1979), Brazil (Schwan, 1998) and
Dominican Republic (Lagunes-Galvez et al., 2007), Saccharomyces cerevisiae was
much more dominant than in the Australian fermentations, and the diversity of
non-Saccharomyces spp. was quite different.
One distinctive feature of the Australian fermentations was that Pichia
membranifaciens was only found in one set of fermentations. This was in contrast
to the ecology of fermentations in Belize (Thompson, 2007), Brazil (Sanchez et
al., 1985) and Ghana (Jespersen et al., 2005, Nielsen et al., 2007b) where
P.membranifaciens was consistently isolated at high populations, from the
majority of fermentations examined.
The baseline yeast data confirmed the observations of previous workers that
several species of yeast grow during successful fermentation of cocoa beans
(Roelofsen, 1958; Ostovar and Keeney, 1973; Schwan et al., 1995; Ardhana and
Fleet, 2003; Jespersen et al., 2005). This finding supports the hypothesis that
starter cultures developed for use in cocoa fermentation should be a mixture of
yeast species, rather than a single species (Schwan, 1998).
It is widely accepted that, during cocoa fermentation, yeasts ferment the pulp
sugars to ethanol, and also utilise citric acid in the pulp (Schwan and Wheals,
2004). The chemical analyses of the fermentations in this study confirmed that
these processes occurred.
Much of the literature describing the roles of yeast during cocoa fermentation
also emphasizes the importance of pectinase production for the breakdown of
the pulp (Sanchez et al, 1984; Schwan et al., 1995; Schwan et al, 1997). Simple
screening of the enzyme activities of Australian yeast isolates found several
strains of S.cerevisiae and I.orie?italis to be pectinolytic (Huynh, N.T.A., Fleet,
G.H. and Dircks, H.D., unpublished results). Hanseniaspora guillermondii,
I.orientalis and S.cerevisiae can also produce a large variety of volatile secondary
metabolites, which are known to have strong flavor impacts in other fermented
beverages and foods (Steinkraus, 2004; Swiegers et al., 2005; Fleet, 2007). Given
172
that high populations of these species were found in the cocoa beans, it is likely
that they contributed to the flavour characteristics of the beans.
Some processing variables affected the growth of yeast species during
fermentation of Queensland cocoa beans:
Firstly, growth of yeast species was affected by the fermentation method -heap,
box or barrel. In the heap fermentation, H.guillermondii persisted for a much
longer time, than in either the box or barrel. In the heap and box fermentations
using cocoa beans from the Mossman plantation, it was observed that I.orientalis
and S.cerevisiae grew until the very end of fermentation, while in the barrel
these species died off around 96 h. It was interesting to note that when the
experiment was repeated using South Johnstone cocoa beans, the yeast ecology
of the box and barrel were very similar to one another. The cause of this
difference was unclear, particularly when the temperature and pH profiles of
the fermentations were only slightly different from one another.
Secondly, the addition of microbial nutrients (Fermaid K®) caused slight, but
significant differences in growth and survival of yeast species, especially
S.cerevisiae. This result is sensible, since this nutrient mixture was originally
formulated to boost the growth of S.cerevisiae during wine fermentation (Bisson
and Butzke, 2000; Lallemand, 2007). Application of nutrients to better control
the process is a novel concept in cocoa fermentation, but a well-established
practice in other industrialized fermentations (Hesseltine and Wang, 1984;
Bisson, 1999; Steinkraus, 2004). They may prove useful in the future, for fixing
stuck or sluggish cocoa fermentations during large-scale, industrialised
operations, or for improving viability of added yeast starter cultures.
Thirdly, it was observed that mixing frequency affected the growth profile of
yeast species during fermentation. In the fermentation mixed every 24 hours,
H.guillermondii, I.orientalis and S.cerevisiae grew according to the baseline
pattern. In the fermentation mixed every 12 hours, the growth of I.orientalis was
delayed. Much has been published on the effects of bean mixing in general and
on the growth of bacterial species. In contrast, the specific effects of mixing on
yeast ecology has been poorly considered. The current data suggests it be given
more careful consideration in the future.
173
The data were inconclusive as to the effects of cocoa bean source (South
Johnstone or Mossman plantations) on the growth profiles of the yeast species.
There was some indication that the yeast species died off more rapidly in the
South Johnstone beans, compared to the Mossman beans, but further
experiments are necessary to confirm this trend. Such regional variations have
previously been noted in the literature and attributed to differences in the
chemical composition of the pulp, or differences in the micro-flora present on
the surface of pods (Wood and Lass, 1985; Ardhana, 1990; Schwan and Wheals,
2004; Camu et al. 2008).
3.4.3.2 Lactic acid bacteria
Four species of lactic acid bacteria were consistently isolated from the
Australian fermentations: Lactobacillus plantarum, Lactobacillus fermentum,
Pediococcus acidilactici and Leuconostoc pseudomesenteroides
These lactic acid bacteria grew, to different extents, throughout the
fermentations. Generally, Lc.pseudomesenteroides grew during the first 24-48h,
reached maximum levels of 105-106 cfu/g, and then rapidly died off due to its
sensitivity to ethanol, acid and temperature. The remaining species grew to
higher populations (108-109 cfu/g) between 48-72h of fermentation. Frequently,
105-107 cfu/g lactic acid bacteria remained at the end of fermentation. The non-
Leuconostoc species of lactic acid bacteria often grew in succession, but the order
of this succession varied considerably between fermentations.
The dominance of Lactobacillus spp., particularly L.plantarum, has been widely
reported for cocoa fermentations in Ghana (Camu et al., 2007; Nielsen et al.,
2007a, 2007b), Brazil (Passos et al. 1984a; Schwan, 1998), Indonesia (Ardhana
and Fleet, 1993) and the Dominican Republic (Lagunes-Galvez et al., 2007).
Several of these studies also reported high levels of L.fermentum and Lc.
pseudomesenteroides. The isolation of P.acidilactici was a distinctive feature of the
Australian fermentations. While Pediococcus spp. are well represented in other
fermented foods (Flolzapfel, 2002; Holzapfel et al., 2007), their occurrence in
cocoa fermentation is rare. The only other recorded occurrences of Pediococcus
spp. was in Trinidad (Ostovar and Keeney, 1973) and Brazil (Passos et al.,
1984a).
174
A few presumptive Lactobacillus spp. isolates gave low (<90%) matches with the
GenBank database. The scope of the project precluded any further identification
of these isolates. However, the recent discovery of Lactobacillus ghanensis sp.nov.
by Nielsen et al.(2007a) from Ghanaian cocoa beans demonstrated that regional
cocoa fermentations may contain an underestimated diversity of species. It
seems reasonable, therefore, that the Australian cocoa fermentations may
harbour some novel species of lactic acid bacteria, which are unique to the local
ecology. Further investigation of this possibility is recommended.
In addition to the four predominant species, the following lactic acid bacteria
were also sporadically isolated: Lactobacillus brevis, Leuconostoc pseudoficulneum
and Lactococcus lactis. The presence of these species was consistent with the
literature (Schwan, 1998; Ardhana and Fleet, 2003; Schwan and Wheals, 2004;
Nielsen et al., 2007b; Camu et al., 2007). The low populations at which these
species were found means that their contribution to the quality and flavour of
the beans was probably not significant.
During cocoa fermentation, lactic acid bacteria metabolise the pulp sugars,
producing mainly lactic acid which then diffuses into the cocoa seeds Schwan
and Wheals, 2004. They can also degrade citric acid, and produce significant
amounts of diacetyl and acetaldyde, both of which could affect the sensory
properties of the beans (Hugenholtz, 1993). Lactic acid bacteria and their
metabolites can also inhibit the growth of other bacterial species, including
Bacillus spp., as demonstrated in tempeh (Nout and Kiers, 2005) and traditional
African cereal products (Kostinek et al. 2005).
The growth profile of species of lactic acid bacteria varied to some extent
according to the configuration used for fermentation. Firstly,
Lc.pseudomesenteroides persisted for much longer in the heap fermentation.
Secondly, both L.plantarum and P.acidilactici died off more quickly in the barrel,
compared to the box and heap. The potential for fermentation configuration to
affect growth of lactic acid bacteria has also been noted by Passos et al. (1984b),
Ardhana (1990) and Nielsen et al. (2007b).
The addition of Fermaid K® slightly improved the growth and survival of the
lactic acid bacteria. L.plantarum was most strongly affected, rapidly reaching
higher maximum populations in the presence of Fermaid K. This is a novel
observation. Since lactic acid bacteria prefer enriched substrates (Hammes and
175
Hertel, 2007), it is not unexpected that elevated levels of vitamins and nitrogen
would stimulate their growth.
Mixing of fermentations also affected the growth of lactic acid bacteria. When
the fermentations were mixed every 12 h, instead of every 24 h, L.fermentum
was not detected, and L.plantarum and P.acidilactici grew more slowly, and
reached lower populations. This agreed with the work of Passos et al. (1984b),
who also observed that mixing twice daily suppressed the growth of lactic acid
bacteria. This suppression was most likely caused by increased oxygenation of
the bean mass, since both Lactobacillus and Pediococcus spp. are sensitive to
oxygen (Passos et al., 1984b; Hammes and Hertel, 2007; Holzapfel et al., 2007).
Only small variations were observed in the growth of lactic acid bacteria during
fermentation of beans from different locations (South Johnstone or Mossman
plantation), and the data were inconclusive as to any specific relationships.
3.4.3.3 Acetic acid bacteria
Only three different species of acetic acid bacteria were isolated from the
Australian fermentations: Acetobacter pasteurianus, Gluconobacter oxydans and
Asaia siamensis. Several recent studies also found a limited diversity of acetic
acid bacteria present during cocoa fermentation (Ardhana and Fleet, 2003;
Lagunes-Galvez et al., 2007; Camu et al. 2007; Nielsen et al, 2007b).
Asaia siamensis appeared only during the first 24-48h of a few fermentations,
reaching maximum populations of 106 cfu/g, before dying off. This species had
not previously been isolated from cocoa, although it has been isolated from
other foods, including fermented rice and bottled fruit water (Kersters et al.,
2007).
A recent study by Camu et al. (2007) reported the identification of two new
Acetobacter species: A.senegalensis and A.ghanensis. These species were only
rarely isolated and their detection and characterisation was assisted by the use
of non-cultural molecular methods. Although these species were not detected in
this study, their recent discovery emphasises the need for more detailed
investigations of the microbial diversity of Australian cocoa fermentations.
176
Acetobacter pasteurianus was found in all fermentations, while G.oxydans was

detected in approximately two thirds of the fermentations. Both A.pasteurianus
and G.oxydans grew in the latter part of fermentation (48h onwards).
A.pasteurianus typically reached maximum populations of 108 cfu/g, and in
many cases remained present until the end of fermentation. Gluconobacter
oxydans was usually present at levels 10 fold lower than those of A.pasteurianus,
and died off much more rapidly. These findings are consistent with those of
Schwan et al. (1995), Nielsen et al. (2007) and Camu et al. (2007) who found
high levels of acetic acid bacteria in the latter stages of fermentation.
Contrasting data were obtained by Ardhana and Fleet (2003) who found lower
levels of acetic acid bacteria (105-106 cfu/ g) much earlier in the fermentations
(within 12 h), and Galvez et al.(2007) who reported that acetic acid bacteria
reached maximum populations of 108 cfu/g around 48 h and then died off.
Capable of growth at moderate temperatures (20-30°C), many species of acetic
acid bacteria only display rapid growth at temperatures above 35°C (Carr and
Passmore, 1979; Ardhana and Fleet, 2003; Camu et al., 2007; Kersters et al.,
2007). This observation was confirmed in this study, with maximum
populations of acetic acid bacteria usually recorded after the temperature of the
beans had exceeded 35°C.
The acetic acid bacteria in general, and A.pasteurianus in particular, grew in a
similar fashion in all of the fermentations examined, and any variations were
relatively minor. Neither fermentation configuration (heap, box, barrel),
addition of Fermaid K, mixing frequency, nor source of cocoa beans (South
Johnstone or Mossman plantation) significantly affected the growth of these
species. Carr et al. (1979,1980) found that more frequent mixing every 24 h
resulted in higher levels of acetic acid bacteria, compared to mixing every 48 h.
Flowever, apart from this current study, no data is available on whether 12
hourly mixing would further stimulate the growth of acetic acid bacteria. It may
be that mixing every 24 hours provides sufficient aeration to meet the growth
requirements of the acetic acid bacteria, and that additional aeration stimulates
additional acetic acid production, but not additional biomass.
These results differed from those obtained by Camu et al. (2008a) which found
that fermentation method, and in particular, mixing, strongly influenced the
growth of acetic acid bacteria. This was most likely because in those
177
experiments, the fermentations involved significantly larger quantities of beans

(150 kg), compared to the experiments described in this chapter (75 kg). In
smaller fermentations, there is a higher surface area to volume ratio, which
naturally gives greater aeration, whereas larger fermentations require mixing in
order for all parts of the bean mass to be aerated (Rohan, 1963; Carr et al. 1979;
Said and Samarakhody, 1984 ; Wood and Lass, 1985).
In this study, growth of A.pasteurianus, and G.oxydans was accompanied by the
production of acetic acid and heat, and a reduction in the concentration of
ethanol. This agrees with the classic function of these bacteria as described in
the literature (Quesnel, 1966; Lopez and Quesnel, 1973a; Ardhana and Fleet,
2003; Schwan and Wheals, 2004; Thompson, 2007). These processes are
important, since they contribute to the internal breakdown of the cocoa seeds
and the formation of flavour precursors (Biehl and Voigt, 1996). The presence of
residual acetic acid can also give cocoa beans acid flavours, which, if excessive,
can be undesirable (Liau, 1978; Jinap, 1994). Furthermore, both A.pasteurianus
and G.oxydans are capable of producing a wide range of secondary metabolites
(Passmore and Carr, 1975; Drysdale and Fleet, 1988; Amoa-Awua et al., 2006
Kersters et al. 2007), and these probably contribute to the flavour of the beans.
In most studies, the isolation and enumeration of acetic acid bacteria has mainly
relied on the use of GYC-CaCCL medium. In this study, a combination of
WLNA and MRS agar media were successfully used to detect, isolate and
enumerate the acetic acid bacteria. WLNA was originally designed for the
isolation and enumeration of bacteria from brewery samples (Holzapfel, 1992;
Pallman et al., 2001). The nutritive status of this medium supported the growth
of acetic acid bacteria, the Bacillus spp. and a limited number of
Enterobacteriaceae, such as Erwinia and Pantoea spp.
While MRS was designed as a medium for isolation of lactic acid bacteria
(Holzapfel, 1992), this study found it to be highly effective in isolating and
maintaining cultures of Acetobacter and Gluconobacter spp. In many cases, the
recoveries of these bacteria using MRS agar were 101 -102 higher than the other
media (GYC-CaCOs or WLNA).
178
3.4.3.4 Other species of bacteria and filamentous fungi
In several fermentations, the growth of Bacillus species was observed during the
last 24-48 hours. This only occurred where the population of lactic acid bacteria
had begun to decrease by around 72 h. Maximum populations of 104 cfu/ g
Bacillus spp. were observed. The main species isolated were B.licheniformis,
B.megaterium, B.pumilis and B.subtilis. A single isolate was identified as B.cereus.
These observations were consistent with the behaviour of Bacillus spp. observed
in cocoa fermentations in Brazil, Indonesia and Ghana (Schwan et al., 1986;
Ardhana and Fleet, 2003; Nielsen et al, 2007). The growth of Bacillus spp. at the
end stages of fermentation reflects their ability to produce heat-resistant and
ethanol resistant endospores (Slepecky and Hemphill, 2007). During this study,
the growth of Bacillus spp. appeared to be inhibited by high populations of
lactic acid bacteria and/ or yeast. This was consistent with research that found
lactic acid bacteria inhibited the growth of Bacillus spp. during other food
fermentations (Vandenbergh, 1993, Nout and Kiers, 2005; Kostinek et al. 2005).
Growth of low populations of Bacillus spp. may have a positive impact since
they have been shown to produce pectinases (Ouattara et al., 2008) and flavour-
active volatiles (Kosuge and Kamiya, 1962; Romancyzk et al., 1995; Adams and
De Kimpe, 2007). However high population of Bacillus spp. leads to the
production of off flavours (Zak and Keeney, 1972; Wood and Lass, 1985; Schwan
et al., 1986; Romancyzk et al., 1995; Schwan and Wheals, 2004). Therefore, the
ability of lactic acid bacteria to control their growth may prove useful during
industrialised fermentations.
While the lactic and acetic acid bacteria were the dominant bacterial species
found during fermentation, a number of other bacterial species were isolated.
These included: Pantoea agglomerans, Erwinia carotovorum, Serratia marcescens,
Pseudomonas putida, and Staphylococcus aureus. The species Pt.agglomerans,
E.carotovorum and Ps.putida are all member of the family Enterobacteriaceae and
are commonly found as endophytes of plants (Grimont and Grimont, 1992;
Kado, 2007). This supports the notion that these species originate from the
surface of the cocoa pods and entered the pulp during pod splitting.
Staphylococcus aureus commonly inhabits mammalian hosts, including humans.
Its presence is an indicator of human contact with the beans during the pod
removal process, and was observed in Indonesian cocoa fermentations
179
(Ardhana and Fleet, 2003). Low levels (102-103) of these bacteria were detected
in the Oh samples, but they died off rapidly and were not detected after 24h.
In a few fermentations, Pt.agglomerans continued to grow during the first 24-48
h of fermentation and reached maximum populations of 106 cfu/ g. At such
levels, it had the potential to compete with the other microbial species or to
contribute significant levels of metabolites. For example, Pt.agglomerans can
produce copious quantities of exo-polysaccharides (Kado, 2007) and these have
the potential to alter the aeration or mass-transfer properties of the fermenting
bean mass. Growth of Pt.agglomerans during the early stages of cocoa
fermentations may be analogous to the growth of Enterobacteriaceae during the
first 24-48 h of sauerkraut fermentation (Farnworth, 2003). Only one other study
has noted the growth of this species during cocoa fermentation (Camu et al.,
2007).
A range of filamentous fungi were infrequently isolated during fermentation
and drying of the cocoa beans. This contrasted to the work of Ogundero (1983)
and Ardhana and Fleet (2003) who reported significant populations of mould
throughout fermentation. This difference may have been due to a lower load of
moulds on the Australian cocoa pods, or the conditions under which they were
processed. Due to project constraints, these fungi were identified to genus level
only, and included representatives of: Aspergillus, Fusarium, Penicillium and
Rhizopus spp. The populations of these species were always low (<103 cfu/g)
and transient, dying off within 24h of detection. The isolation of mould from the
fermentations corresponded to times at which mycelia were visible on the
surface of the beans mass: usually between 0-24h, or 96-120h. Given the ability
of these species to produce my cotoxins (Ogundero, 1983; Pitt and Hosking,
1985; Mounjouenpou et al., 2008), future work into Australian cocoa processing
should more closely investigate the exact species present, the levels of any
mycotoxins and the effectiveness of control measures to reduce this risk.
3.4.4 Chemical and physiological changes during fermentation
Analyses of cocoa pulp and seed samples revealed a variety of chemical

changes that occurred during the fermentations, which were similar to those
described in previous studies. The observed chemical changes could be linked
to, and explained by, the microbiology of the fermentations.
180
3.4.4.1 Sugars
The concentrations of glucose (4.6 - 6.0%) and fructose (5.0 - 6.7%) in the
unfermented pulp of Australian cocoa beans were consistent with those
reported elsewhere (Dittmar, 1956; Rohan, 1963; Pettipher, 1986; Goto, 2002;
Ardhana and Fleet, 2003; Camu, 2007). As noted in previous studies, fructose
was the most abundant sugar in the pulp. Sucrose was never detected any
sample of Australian cocoa pulp, which agreed with observations in Brazil
(Dittmar, 1956; Berbert, 1979), in the Dominican Republic (Lagunes-Galvez et al,
2007) and Ghana (Camu, 2007), but contrasted with the work of Ardhana and
Fleet (2003) who did detect sucrose in the cocoa pulp of Indonesian beans.
Slight variations in the initial levels of both pulp and seed sugars were noticed
throughout the duration of the project ( 2003-2006) (Appendix B - Table 1).
These variations probably reflect seasonal and regional effects on the
composition of the cocoa beans. During fermentation, the glucose and fructose
were utilised by the yeast and bacterial species present. By the end of
fermentation, only traces of glucose were detected in the pulp.
In the unfermented cocoa seeds, glucose (0.2 - 0.8%), fructose (0.2 -1.0%) and
sucrose (1.0 -1.7%) were all detected at low levels. By the end of fermentation,
these sugars had decreased to trace or undetectable levels. In many
fermentations, a decrease in the concentration of sucrose coincided with
increased concentration of glucose and fructose. This has been noted by
previous workers (Hashim et al., 1998; Ardahana and Fleet, 2003) and,
according to Hansen et al. (1998), is caused by the action of native cocoa seed
invertase.
Fermentation configuration, addition of Fermaid K® and mixing frequency

affected the rate of sugar utilisation:
The pulp sugars (glucose and fructose) were utilised slightly faster in the barrel,
compared to the box and the heap fermentations. The reason for this was not
clear, but it may be linked to the dominance of S.cerevisiae in the barrel, given
that S.cerevisiae is capable of rapid fermentation of sugars (Fleet, 2001; Pina,
2004).
The pulp sugars were utilised slightly faster in the fermentations to which
Fermaid K® had been added. This can also be linked to the slightly accelerated
181
growth of yeast in the supplemented fermentations. The effect of microbial

nutrients to increase sugar utilisation has been noted in wine research literature
(Bisson, 1999; Lallemand, 2007).
Finally, sugar utilisation was accelerated in the fermentation mixed every 12
hours. This was probably caused by the increased growth of S.cerevisiae in the
fermentations mixed every 12 h compared with those mixed every 24h.
3.4.4.2 Ethanol
The kinetics of ethanol production in the pulp and seeds of Australian cocoa
beans corresponded with the observed growth of yeast species and utilisation
of pulp sugars. Ethanol was not detected in fresh cocoa beans, but was
produced during fermentation. Maximum levels of ethanol were recorded, in
both the pulp (5 - 7%) and the seeds (2 - 5%), between 48 - 72h for all
fermentations. The concentrations of ethanol were lower when seasonal or
regional differences caused lower initial levels of sugars in the pulp.
At the end of fermentation, the concentration of ethanol in the pulp and seeds
had decreased to low (0.25 - 0.75%) or undetectable(<0.1%) levels. This loss was
caused by a combination of evaporation and oxidation to acetic acid. Diffusion
of ethanol from the pulp to the seeds helps to trigger the endogenous
biochemical reactions responsible for flavour development in the beans (Biehl et
ah, 1982c; Lopez and Dimick, 1995). The kinetics of ethanol production and
utilization in the Australian fermentations were consistent with observations
made elsewhere (Roelofsen, 1958; Lopez and Quesnel, 1973a; Carr et al., 1979;
Lopez and Dimick, 1995; Ardhana and Fleet, 2003; Afoakwa, 2008; Camu et al.,
2008b).
There were differences in the concentrations of ethanol detected in the beans
fermented by heap, box or barrel. The highest concentrations of ethanol were
recorded in beans fermented by barrel, with the lowest being recorded in beans
from the heap. This may have been caused by the differences in microbiology
observed. Another explanation is that ethanol evaporated more easily in the
heap and box fermentations, compared to the barrel. This is supported by the
literature (Wood and Lass, 1985; Lagunes-Galvez et al., 2007; Camu et al.,
2008b). It is also supported by the observation that beans in the barrel remained
moist, while beans at the surface and edges of the box and heap appeared dry.
182
The addition of Fermaid K® did not appear to affect the levels of ethanol
produced. Slight differences were observed when beans from the South
Johnstone plantation were used, and 25 g of Fermaid K added. Ffowever, the
results were not reproduced when beans from the Mossman plantation were
used.
Mixing frequency did not appear to affect the ethanol levels. This contrasted to
experiments by Camu et al. (2008a), where mixing led to increased loss of
ethanol. In that case, however, the comparison was between no mixing, or
mixing every 48 hours. In the present study, the mixing intervals were more
frequent (every 12 or 24 hours), suggesting a threshold effect: for small
quantities of beans (<100 kg), mixing more frequently than 24 hours did not
significantly affect ethanol production and loss.
Finally, the source and time of harvest of the cocoa beans affected the amount of
ethanol produced during fermentation. In general, the higher the concentration
of sugars present in the pulp, the higher the amounts of ethanol produced
during fermentation. This relationship has been extensively described in the
literature (Roelofson, 1958; Lopez and Quesnel, 1973a; Dougan, 1979; Carr et al.,
1980; Wood and Lass, 1985; Ardhana and Fleet, 2003; Schwan and Wheals, 2004;
Afoakwa et al., 2008).
3.4.4.3 Organic acids
Citric acid was the most abundant acid in the pulp (30 - 55 mg/g) and seeds(10
-15 mg/g) of fresh Australian cocoa beans. As with the pH and sugars, slight
variations in the concentration of citric acid were noted. Such variations
suggested subtle regional and seasonal effects on the composition of the cocoa
beans. The concentration of citric acid decreased during fermentation, probably
through utilisation by S.cerevisiae, L.plantarum and L.fermentum (Jinap, 1994;
Ardhana and Fleet, 2003; Hammes and Hertel, 2007). In a few fermentations,
the concentration of citric acid in the pulp increased during the final 48 hours.
This might be explained by production by L.plantarum and/or I.orientalis
(C.krusei) (Kaneuchi et al., 1988; McKay et al., 1990; Jinap, 1994; Anastassiadis et
al.; 2002). As previously described, the concentration of citric acid in the cocoa
seeds decreased only slightly during fermentation (Jinap, 1994; Ardhana and
Fleet, 2003; Camu et al., 2007). The results of this research indicated that there
was little diffusion of citric acid between the pulp and seed.
183
Malic acid was detected at levels of 10 -15 mg/g in the pulp of unfermented
Australian cocoa beans. The levels of malic acid generally decreased during
fermentation, although sometimes significant fluctuations were observed. The
decrease can be explained as the utilisation of malic acid by I.orientalis (Seo et al.
2007) and Lactobacillus spp.(Hammes and Hertel, 2007). The increases,
occasionally observed, may have been due to malic acid production by yeast
species: both S.cerevisiae (Fatichenti et al., 1984; Zelle et al., 2008) and
H.guillermondii (Albergaria et al., 2003) can produce malic acid. The level of
malic acid in unfermented seeds was typically low (around 5 mg/g) and did
not change significantly during the fermentations. This suggested that malic
acid did not readily diffuse into the seed.
Acetic acid was not found in the pulp of fresh Australian cocoa beans, but was
produced in the pulp as the acetic acid bacteria utilised ethanol. The kinetics of
acetic acid production tended to be linked to the production of ethanol by the
yeast, with the most rapid increases occurring after the concentration of ethanol
peaked (48-72h). The final concentration of acetic acid varied between
fermentations (30 - 90 mg/g in the pulp,TO - 40 mg/g in the seed). These levels
were comparable with those found in Brazil and Malaysia (Jinap and Dimick,
1990), and Indonesia (Ardhana and Fleet, 2003). The concentration of acetic acid
usually peaked at the end of fermentation, but occasionally did so earlier (72-96
h) and then decreased. These decreases were probably caused by over
oxidization of the acetic acid by acetic acid bacteria, and evaporation (Lehrian
and Patterson, 1983).
No acetic acid was detected in unfermented cocoa seeds. Its concentration
increased during fermentation, as it was produced in the pulp and diffused into
the seeds. In some cases, the concentration of acetic acid in the seeds was higher
than in the pulp, suggesting that the cocoa seeds had an affinity for uptake and
storage of this acid. Such an affinity was not observed for other acids, and was
previously noted by Ardhana (1990). By the end of fermentation, acetic acid was
usually the most abundant acid in the cocoa seed. Less acetic acid was
produced when the initial concentrations of sugars in the pulp were lower. This
confirmed the observations of previous workers (Liau, 1980; Wood and Lass,
1985; Biehl et al., 1990;Ardhana and Fleet, 2003).
184
Lactic acid was also absent from the seeds and pulp of fresh Australian cocoa
beans. During fermentation, the production of lactic acid corresponded to the
growth of lactic acid bacteria, and the utilisation of glucose and fructose in the
pulp. The maximum levels of lactic acid produced in the pulp varied widely
between different fermentations (20 - 70 mg/g). Higher concentrations of lactic
acid were recorded in fermentations that contained higher populations of
L.plantarum, and these variations were usually caused by differences in the
processing methods used. The observations were consistent with those of
Passos et al.(1984), Jinap (1994) and Ardhana and Fleet (2003).
Low levels (2-5mg/ g) of tartaric and oxalic acids were also found in the pulp
and seeds of Australian cocoa beans. The concentrations of these acids did not
change significantly during fermentation.
The changes in concentrations of organic acids in the seeds and pulp caused
changes in pH of the pulp and seeds, consistent with the explanations of Rohan
(1963), Biehl (1984) and Jinap and Dimick (1990). Overall, the levels of organic
acids observed in Australian beans were lower than in Indonesia (Ardhana and
Fleet, 2003) and but slightly higher than in Ghana (Camu, 2007; 2008b). Small
regional and seasonal variations in the levels of organic acids were observed
during this study (Table 1 - Appendix A). These differences help explain the
variations in the initial pH of beans from different plantations.
Fermentation configuration (heap, box or barrel) affected the metabolism of
citric acid and the production of acetic and lactic acids during fermentation:
Firstly, the concentration of citric acid decreased more in the heap and box,
compared to the barrel. The incomplete utilisation of citric acid corresponded to
the premature death of L.plantarum observed in the barrel fermentation. Second,
less acetic acid was produced in the heap, compared to the box or barrel. This
indicated a difference in the degree of aeration of heap, which was mixed only
once, compared to the box and barrel, which were mixed every 24h. Thirdly,
more lactic acid was produced in the heap fermentation, compared to the box or
barrel. This supported the assertion that the heap was less aerated than the box
and barrel, since lower oxygen levels promote lactic acid production by lactic
acid bacteria (Passos et al., 1984b; Arunga, 1992; Camu, 2008b).
185
The addition of Fermaid K caused slightly increased production of lactic acid in

the pulp - by stimulating the growth of lactic acid bacteria. The levels of lactic
acid in the seed were also affected, but to a lesser degree.
Increasing the frequency of mixing from 24 hourly, to 12 hourly resulted in
reduced utilisation of citric acid in the pulp by the lactic acid bacteria. The
increased aeration also caused increased production of acetic acid in the beans,
consistent with the growth of acetic acid bacteria.
3.4.4.4 Glycerol and mannitol
In some of the Australian cocoa fermentations, glycerol and mannitol were

produced.
Between 48 - 72h of fermentation, glycerol was detected at levels of 1-5 % in the
pulp and 0.5-1% in the seeds. Higher levels of glycerol were detected in the
barrel fermentations. Glycerol was probably a co-product with ethanol of yeast
metabolism. The production of glycerol by yeast is common in wine
production, where it can affect mouth-feel, or reduce the ethanol yield
(Michnick, 1997). By contrast, the significance and effects of glycerol in cocoa
fermentation is not well understood. The production of glycerol during cocoa
fermentation has not been reported in the literature consulted in this study .
Mannitol was detected in the pulp of some Australian fermentations,
particularly those with high populations of L.fermentum. Lactobacillus fermentum
has previously been described as a potentially strong producer of mannitol,
particularly under stress conditions such as lower moisture or higher ethanol
concentrations (Wisselink et al., 2002). Maximum concentrations of 3-5%
mannitol were typically found at the end of fermentation (120h), although the
concentration was higher (6-8%) when the fermentations were conducted in
barrels or mixed every 12 h. The presence of mannitol in cocoa beans has not
been widely reported, although, in a recent study, Camu et al. (2008b) observed
that increased mixing of heap fermentations led to increased production of
mannitol. Mannitol has a sweet flavour, and like many other sugar-alcohols it
can also have a cooling effect in the mouth (Wisselink et al., 2002). For these
reasons, the presence of high levels of mannitol in the beans at the end of
fermentation may affect their sensory properties.
186
3.4.4.5 Physiological changes in cocoa beans during fermentation.
According to the currently accepted model of chocolate flavour formation,

several conditions are required to trigger the endogenous reactions which in
turn produce the chocolate flavour precursors. The requisite conditions include:
increased temperature, a decrease in the seed pH, and increased concentrations
of acetic acid and ethanol. These conditions are achieved by the timely growth
in the cocoa pulp of a diverse range of yeasts, lactic and acetic acid bacteria
(Forsyth and Quesnel, 1956; Wood and Lass, 1985; Lopez and Dimick, 1995). As
already discussed, the Australian fermentations met the basic conditions of
microbiology, temperature, pH, and chemistry described in the literature.
Qualitative observations indicated that these conditions successfully triggered

endogenous biochemical reactions inside the cocoa seeds: beans obtained at 72
h of fermentation were found to be filled with a brown liquid. When seed
tissue from these liquid-filled beans were examined microscopically, it was
observed that over 90% of the pigment cells had been lysed. These observations
were consistent with those made by Biehl (Biehl et al., 1977; Biehl et al, 1982a)
who positively linked these specific physical changes to the formation of
flavour precursors.
3.4.5 Effects of fermentation on sensory and quality outcomes
Cocoa beans represent a novel crop for Australia, and a key goal of this project
was to determine whether cocoa beans of acceptable quality and flavour could
be produced. Since cocoa beans are the raw ingredient of chocolate, the ultimate
success of any cocoa fermentation must be judged by the quality of the beans
and their suitability for use in chocolate manufacture (Wood and Lass, 1985;
Beckett, 2000). Therefore, the beans produced in this study were carefully
evaluated using methods and criteria adopted from industry, combined with a
scientifically based sensory evaluation program. Testing was also performed to
determine how certain fermentation practices affect the quality of the beans
produced.
3.4.5.1 Overall acceptability of Australian cocoa beans
Using industry standard tests, the quality of the Australian cocoa beans was
found to be largely acceptable. Based on the cut test, all samples of beans except
187
one1 were judged to be acceptable, with zero % mouldy or insect infested, less
than 1% slaty and less than 20% fully purple (Wood and Lass, 1985). The size,
moisture content, fat content, bean pH, general appearance and aroma of the
Australian beans were also found to be acceptable when compared to the ideal
prescribed values commonly used in industry. The quality data for the
Australian beans also compared favourably to the data obtained for cocoa beans
from both Ghana and Indonesia (Table 3.9, Section 3.3.1.8).
A minor undesirable feature of the Australian beans was a higher-than-ideal
shell content. The higher shell content was probably a result of the drying
procedure used. In Ghana, fermented cocoa beans are frequently mixed during
drying (Nielsen, 2006), while in Indonesia, beans are washed prior to drying
(Ardhana and Fleet, 2003). Both of these procedures have been demonstrated to
reduce shell content, and similar methods are employed in other cocoa
producing countries (Rohan, 1963). In this study,the beans were mixed only
minimally during drying and were not washed, and these factors may explain
the higher shell content.
Ultimately, the success of a cocoa fermentation will be judged by the aroma and
flavour of the beans produced and chocolate derived from them - the flavour of
beans strongly determines the price paid for them (Wood and Lass, 1985;
Buamah, et ah, 1997; Beckett, 2000; Guittard, 2005). While physical measures of
bean quality, such as cut test, shell content and pH are useful first indicators of
quality, they should not be used at the exclusion of sensory analyses (Baker et
ah, 1994; Schwan, 1998; Am ores et al., 2007; Sukha et al., 2008). Therefore, the
overall acceptability of the flavour of the Australian cocoa beans was evaluated,
by comparison to the flavour of cocoa beans from Ghana. As discussed
previously, Ghana beans are considered to represent a standard for bulk cocoa
with respect to chocolate flavour (Carr et al., 1979; Wood and Lass, 1985; Pino et
al, 1992; Baker et al., 1994; Jinap et al., 1995).
In nearly all cases, the chocolate tasters prepared from samples of the
Australian beans were liked equally or greater than the chocolate made from
Ghanaian beans. The goal of producing Australian cocoa beans of commercially
acceptable quality was, therefore, seen to be achievable.
1 Unacceptable sample was from mixing experiment: box fermentation, mixing every 12h, oven
drying.
188
The sensory evaluation also suggested that the Australian cocoa beans might
have their own unique flavour characteristics. Qualitative feedback from
sensory panelists suggested that the strengths of chocolate flavour, bitterness
and astringency of the Australian chocolate were similar to the Ghanaian
samples. Approximately a third of panelists also noted distinctive floral, yeasty
or fruity characteristics in the chocolate prepared from Australian samples.
Given these observations, it is recommended that future sensory testing involve
the development of quantitative flavour profiles, of both the Australian and
Ghanaian cocoa beans. Such profiles could be used to clarify how local
fermentation practices and microbial ecology might affect the final flavour of
beans.
3.4.5.2 Effects of fermentation variables on the quality and sensory properties

of cocoa beans grown in Queensland, Australia.
Beans fermented by heap, box and barrel met the industry guidelines for cut
test, moisture content, bean size, fat content and pH. The flavours of all samples
were rated as equal to or greater than the flavour of the Ghana beans. As this
was the first experiment conducted, these results provided the initial indication
that Queensland beans could be successfully fermented. Furthermore,
production of acceptable beans using the barrel confirmed its suitability as a
fermentation vessel. The triangle tests, on chocolate made from the different
beans, indicated that the different samples tasted significantly different from
one another. Sensory analysis revealed that beans produced by heap
fermentation gave chocolate with a flavour that was rated much higher than the
Ghana standard. By comparison, chocolate made beans fermented by box or
barrel were similar to the Ghana beans. Qualitative feedback from some
panelists suggested that chocolate made from the box and barrel fermented
beans had a slightly acid aftertaste. The differences in flavour directly
corresponded to the differences in acetic acid levels observed earlier, and
suggested that the box and barrel were significantly more aerated than the
heap. The link between aeration, acetic acid levels and flavour has long been
recognised in the literature (Quesnel, 1968; Shepherd, 1976; Biehl et al., 1977;
Wood and Lass, 1985; Duncan et al., 1989; Tomlins et al., 1993; Senanayake et al.,
1997; Afoakwa et al., 2008; Camu et al., 2008b). In light of the quality and
sensory data from this experiment, it is suggested that further optimisation of
189
the process to reduce acidity could improve the quality of beans produced by
box and barrel fermentations.
The addition of Fermaid K® caused no differences in quality or flavour, but all

beans from this experiment were found to be commercially acceptable. This
reflected the similarity of the temperature, pH, microbiology and chemistry
observed during the fermentations. Furthermore, the similarity in flavour of the
control and treatments suggested that the minor variations in microbiology and
chemistry were not significant in the context of final flavour or quality.
Together, the data showed that the addition of microbial nutrients conferred no
benefits in this case. The similarity of the control and treated beans also
demonstrated that addition of Fermaid K® had no negative effects on quality or
flavour. Fermaid K®, and similar microbial nutrients, are commonly added to
wine and other alcoholic fermentations to assist the growth of yeast species
under unfavorable conditions (Lallemand, 2007). The lack of difference between
the control fermentations, and those supplemented with Fermaid K®, suggests
that the initial conditions of the beans were already favourable for microbial
growth. However, stuck and sluggish fermentations have been reported as a
problem during cocoa processing (Rohan, 1963; Quesnel, 1972), and microbial
nutrients might be useful in their prevention or cure (Bisson, 1999).
Mixing frequency during fermentation was found to significantly affect the

flavour of the cocoa beans. Triangle testing showed that beans mixed 24 hourly
had a significantly different flavour from those mixed 12 hourly. Unfortunately,
it was also found that all beans produced in this experiment were of inferior
quality and flavour. For the first time in this study, a batch of beans failed to
pass the cut test due to an unacceptably high proportion of purple beans.
Furthermore, sensory evaluation revealed that both treatments (24 and 12
hourly mixing) resulted in beans which tasted worse than the Ghana standard
beans. Feedback from panelists indicated that the chocolate made with these
beans had strong acid flavours and weak chocolate character. This anecdotal
evidence also indicated that the flavour faults were more pronounced in the
beans that had been mixed every 12 h. Together with the chemistry data, the
sensory results suggested that the poor flavour was caused by high levels of
acetic acid in the seeds, due to over-aeration during fermentation. As well as
190
contributing acid flavours, the excess acidity inhibited the development of

chocolate flavour-precursors. The observation that the beans mixed every 24
hours were also considered unacceptable caused a re-examination of earlier
results.
It was noted that previous fermentations had been mixed every 24 hours, yet
still produced beans of an acceptable flavour and quality. The heap
fermentation, which was mixed every 48 hours, produced beans with the
highest mean liking score (7.42). Interpreting these observations together, it is
suggested that mixing every 48 hours might be better than every 24 hours.
This conclusion is supported by Shepherd (1976), Wood and Lass (1985),
Duncan et al. (1989) and Baker et al. (1994) who all recommended an optimum
mixing frequency of once every 48 hours. However, it refuted previous
recommendations to mix every 12 hours (Senanayake et al., 1997) or every 24
hours (Rohan, 1963; Said and Samarakhody, 1984; Portillo et al., 2005).
Interestingly, the studies that recommended more frequent mixing only
evaluated quality using the cut test, and other physical measurements. In
contrast, the studies recommending mixing every 48 hours or less had all
conducted sensory analysis to confirm the effects on the flavour of the beans.
This stresses the importance of including adequate sensory analysis in any
attempt to improve cocoa fermentation (Rohan, 1963; Wood and Lass, 1985;
Am ores et al., 2007; Sukha et al., 2008).
This point was further stressed by a comparison of the sensory evaluation and
quality data obtained during this study. First, it was found that the heap
fermentation produced beans with a flavour that was significantly better than
any other beans in this study. However, the cut test score for the heap
fermented beans was quite low (34%). Secondly, during the mixing experiment,
beans mixed every 24 hours had a high cut test score (70%), but had an
unacceptable flavour. In both cases, the percentage of brown beans correlated
poorly with the flavour obtained. The cut test was designed as a way of
excluding beans of unacceptable quality (Wood and Lass, 1985). Many
researchers, however, have used cut test data to infer the presence of positive
attributes, without conducting any form of sensory analysis (Said and
Samarakhody, 1984; Barel, 1987; Buamah et al., 1997; Senanayake et al., 1997;
Portillo et al., 2005). The contradictions between the cut test and sensory
191
evaluation data in this study reinforce the limitations of the cut test as a tool for
evaluating cocoa quality.
3.4.6 Implications for future cocoa fermentation in Australia
3.4.6.1 Baseline established for future reference
This research has established baseline information about the physical,

microbiological and chemical changes that occurred during fermentation. This
collected baseline information is significant for a number of important reasons:
- The information was obtained during fermentation of locally grown
Queensland cocoa beans, under local conditions. This regionally specific
knowledge will assist future cocoa research in Australia, by reducing reliance
on data obtained under foreign conditions. The effects of different
fermentation variables, or the use of starter cultures can be compared to this
Australian baseline.
- Australia is geographically and ecologically isolated from the traditional
cocoa producing-regions. The microbiological data, and large number of
isolates obtained during this study, may contribute to the discovery of new
species of yeast and bacteria unique to Australia.
- Understanding the basic features of Australian cocoa bean fermentations will
help identify any characteristics that are unique to the Australian beans.
These unique properties would be useful in marketing Australian cocoa beans
internationally.
3.4.6.2 Methods used to examine the microbiology of cocoa fermentations
This study used a culture based method for examining the microbiology of
fermentations. Various precautions were taken to ensure an accurate survey of
the microbial ecology: a range of agar media were used, samples were plated
within 1 hour of being obtained, and serial dilutions were used to obtain
quantitative measures of the microbial populations. Large numbers of
representative isolates were obtained and these were identified using both
phenotypic and molecular (genotypic) methods. A similar approach was
successfully used by Jespersen et al. (2005), except in that case only yeast
species were examined.
192
Many recent investigations of the microbial ecology of cocoa fermentation used

non-cultural methods, particularly PCR-DGGE (Nielsen et ah, 2005; Nielsen et
ah, 2007b; Camu et ah, 2007; Camu et ah, 2008a, b). These studies are part of a
broader trend to apply non-cultural methods to fermented foods (Giraffa and
Neviani, 2001). The potential advantages of these methods, over cultural
examination include: a better estimate of microbial diversity, the ability to
detect injured or non-culturable cells, speed of analysis, ability to analyse frozen
samples shipped from the field, and an ability to distinguish between closely
related species or strains (Beh et ah, 2006).
However, PCR-DGGE has some potential limitations which made it less
suitable for use in this study. In particular, research by Prakitchaiwattana et al.
(2004) revealed that PCR-DGGE would not accurately detect species present at
populations of less than 104 cfu/g, and would not accurately indicate cell
viability. Furthermore, accurate assessment of microbial ecologies by PCR-
DGGE requires appropriately designed primers, and the use of poorly targeted
primers will incorrectly estimate the microbial diversity present (Beh et ah,
2006).
In this present study, it was most important to obtain accurate quantitative data
concerning the growth, survival and death of the dominant species found
during fermentation. This information was particularly important in deciding
which species would be most suitable for use in starter cultures. The need to
develop a working fermentation meant that detection of the full diversity of
species was less important. Finally, the availability of well-equipped laboratory
facilities allowed extensive testing to occur on-location, and meant that results
could be obtained more rapidly by cultural methods. For these reasons, culture-
based methods were used for this study.
Nevertheless, in order to provide some comparison of a culture based method
with popular non-cultural methods,a collaboration was made with researchers
at the Royal Agricultural and Veterinary University of Denmark (KVL). Nucleic
acid was extracted from cocoa bean samples as described in Methods Section
3.2.3.7. These extracts were sent to KVL for PCR-DGGE analysis using the
method of Nielsen et al. (2007b). The results are presented in the Appendix A
(Section A.3, Figures 1-3). The molecular and cultural methods yielded
comparable results. The banding patterns produced by the DGGE gels showed
193
that presence of H.gnilliermondii, I.orientalis, S.cerevisiae, L.plantarum,

Lc.pseudomesenteroides, P.aciddactici, A.pastreurianus and Pagglomerans. The
appearance and disappearance of several of the bands roughly corresponded to
the population dynamics described by the growth curves. Some differences
between the cultural and non-cultural data were also observed. In particular,
the presence of additional bands on the gels suggested the presence of species
of yeast and bacteria that were not detectable by the cultural methods. Nielsen
et al. (2007b) also reported good comparability between cultural and non-
cultural methods when investigating the microbial ecology of cocoa
fermentation. The similarities between the results obtained by the two methods
further vallidate the culture-based methods used in this study.
3.4.6.3 Fermentation Method
The experiments described in this chapter examined the effects of fermentation

configuration (heap, box or barrel), the addition of microbial nutrients (Fermaid
K) and frequency of mixing. The results obtained had several practical
repercussions.
Firstly, although heap fermentation was able to produce good quality beans it
was quickly determined to be unsuitable for large scale commercial
fermentations. Large amounts of manual labour were required to prepare the
heap at the beginning of fermentation (lay out the banana leaves, ensure the
heap is the correct shape, and wrap the heap). Furthermore, mixing of the heap
required that it be fully dis-assembled and re-assembled each time.
Furthermore, the heap fermentation was observed to be very heterogenous.
This would make implementing the use of starter cultures, or process control
principles very difficult.
Secondly, mixing frequency and aeration was identified as critical variables in

determining the outcome of cocoa fermentation:
- Based on the results obtained in this chapter, it is recommended that
fermentations of 50-100 kg are mixed no more than once every 48 hours.
- The aeration of a mass of fermenting cocoa beans appears to be related not
only to mixing frequency, but also the the fermentation vessel used. Therefore
if larger masses of beans are to be fermented in the future it is probable that
mixing will need to be re-optimised.
194
- In addition to aeration, mixing helps reduce heterogeneity in the beans and

facilitates various mass-transfer processes. Therefore more research should be
performed into methods of mixing that do not introduce large volumes of air
into the bean mass during fermentation.
Thirdly, a number of practical advantages in using the barrel vessel were

observed:
- The barrel allowed for easier mixing and manual handling of the beans
during fermentation. The ability to tumble the barrel mean that mixing could
take place without opening the barrel. This eliminated human contact with
the beans and reduced the risk of cross-contamination during fermentation.
The beans could very easily be removed at the end of fermentation, simply by
inverting the barrel. Given that I.orientalis is an opportunistic human
pathogen (Kurtzmann and Fell, 1998), reducing human contact with the
fermenting beans mass also has occupational health and safety advantages.
- The barrel provided greater homogeneity in the fermenting bean mass. The
lack of corners, or cool spots in the barrel meant that the temperature was
much more consistent throughout the barrel. There was much less drying of
the beans, and mould growth was much less frequent during barrel
fermentations compared to box fermentations. This will help improve
consistency in the quality of beans produced.
- The barrel vessel is a much closer physical configuration to large-scale

industrial fermenters than either the box or the heap. The mechanisation of a
barrel type vessel can easily be obtained by attaching a motor to the turning
axis. Automation can then be developed by linking sensors (temperature, pH,
oxygen or other variables) to a control circuit or computer. Several other
workers have developed industrial cocoa fermenters in this manner, and all of
these were based around some form of cylindrical vessel (Wood and Lass,
1985; Barel, 1995; Schwan and Wheals, 2004).
Finally, some potential improvements to the experimental design were

identified:
- It is recommended that further comparisons be made between the box and

barrel vessels. The box represents a common method used internationally,
195
and would allow a point of reference for any developments made using the
barrel.
- More detailed sensory analysis: future experiments in fermenting Queensland
cocoa beans should involve a quantitative, descriptive analysis of the beans
produced.
3.4.6.4 Drying method
For this project, two methods of drying were used: solar drying and oven
drying. These methods were chosen for their similarity to methods used in
commercial cocoa production worldwide. While the focus of the project was on
the fermentation, drying is widely considered to be critical to the final quality of
the cocoa beans (Rohan, 1963; Wood and Lass, 1985; Thompson, 2007).
Specifically, slower drying allows the continued formation of flavour precursors
(Misnawi et al., 2002, 2003) and a greater loss of acetic acid (Duncan et al., 1989;
Augier et al., 1998; Nganhou et al. 2003; Garcia-Alamill et al., 2007). However,
the timing must be carefully managed, since moist beans will eventually
support the growth of moulds, which can present both a health and quality
hazard (Rohan, 1963; Hollywood, 1998; Schwan and Wheals, 2004)
In this study, solar drying consistently produced better quality beans, although
oven drying also produced cocoa beans of an acceptable quality. Comparison of
cut test data revealed that solar dried beans had a consistently higher
fermentation index compared to oven dried beans. It was also observed that
solar dried beans tended to be larger and have less shell than beans dried by
oven. One disadvantage of solar drying was that it could be interrupted or
delayed by long periods of rain. At such times, the drying of these beans had to
be completed in the oven. Beans dried by this combination of solar and oven
methods did not have the same high quality as solar dried beans. Observations
in this study support the general conclusions of previous investigations of the
drying process (Rohan, 1963; Duncan et al, 1989; Tomlins et al., 1993; Hashim,
1999). Much of the literature recommends slow drying processes, to allow the
enzymatic reactions to achieve completion.
Since drying method can significantly affects cocoa quality, the development of
a mechanical drier that can mimic the solar drying process would be beneficial
for Australian cocoa production. If such a process were to be developed, it is
196
also recommended that the beans be mixed or agitated during drying, to break
up clumps of beans and help reduce the shell content of the beans (Wood and
Lass, 1985; Barel, 1995).
The scope of the project prevented a detailed investigation of the microbiology
of drying beans. Nevertheless, a small number of beans samples were taken at
24hr and 72 h after the commencement of drying, and examined using the same
methods as the fermentations (Section 3.2.3). This limited microbiological
survey revealed that significant populations of I.orientalis (104-106 cfu/g) and
Bacillus spp. (105-106 cfu/ g) remained during the early stages of drying. Little
attention has been given to the microbiology of drying cocoa beans, and
research into this area may reveal novel ways to improve the safety and quality
of cocoa beans.
3.5 Conclusions
A) This study is the first to report on the fermentation of cocoa grown in
Queensland, Australia. A variety of baseline information was collected,
describing cocoa fermentation as it occurred under local conditions:
i. pH - The pH of the cocoa pulp increased during fermentation, while
the pH of the cocoa seeds decreased.
ii. The temperature of the fermenting beans increased from 25°C at the
beginning of fermentation to maximum temperatures of between 40 -
48°C.
iii. As the fermentation proceeded, the quantity of pulp decreased, and it
became darker in colour. The fermentations initially smelt alcoholic in
the first 24-48 hours, and then developed a vinegar aroma between
48-96 hours.
iv. The main yeast isolated from Australian cocoa fermentations were
H.guillermondii, I.orientalis and S.cerevisiae. P.membranifaciens was also
isolated from a small number of fermentations.
v. Representatives of the lactic acid bacteria (L.plantarum, L.fermentum,
P.acidilactici, and Lc.pseudomesenteroides) and acetic acid bacteria
(A.pasteurianus, G.oxydans and As.siamensis) were isolated from the
Australian cocoa fermentations. Several isolates of lactic acid bacteria
197
were not able to be identified by sequencing or phenotypic

characterisation. Further characterization of these isolates may yield
the discovery of a new species, and is strongly recommended.
vi. Comparison of the microbiological data obtained by PCR-DGGE, with
data obtained by the cultural methods used in this study suggested
that the cultural methods were appropriate. However, results from the
non-cultural methods suggested the presence of unidentified species
and a diversity of strain types. More detailed investigations of the
microbial diversity of Australian cocoa fermentations, using
molecular methods to identify new strains or species, is
recommended.
vii. As the fermentations proceeded, sugars were utilised by the yeast and
bacterial species, producing ethanol and lactic acid. Acetic acid
bacteria oxidised ethanol to acetic acid. The lactic acid bacteria and
yeast utilised the citric acid present in the pulp. The ethanol, lactic
and acetic acids diffused into the seed from the pulp. This triggered
the breakdown of the seed cellular structure and killed the embryo,
triggering endogenous changes.
viii. These general trends observed during fermentation of Australian
cocoa beans agreed with those reported in the literature for
fermentations in other countries. However, the Australian
fermentations also had several features that distinguished them from
fermentations performed elsewhere
B) Commercially acceptable beans were able to be produced by fermentation of
cocoa grown in Queensland, Australia. The beans met standard commercial
criteria for quality (cut test, bean size, moisture content, nib pH, fat content).
The flavour of the beans was judged acceptable compared to the flavour of
beans from Ghana. Initial results indicated that the Australian cocoa had a
flavour that was distinct from the flavour of beans produced in other
countries. More detailed sensory evaluation should be performed to build a
quantitative flavour profile of the Australian cocoa beans.
C) Fermentation of Queensland cocoa beans by different configurations (heap,

box or barrel) significantly affected the microbiology, chemistry, quality and
flavour of the beans produced. The results suggested that the different
198
fermentation configurations affected the aeration of the beans. Although the

heap fermentation produced beans with the highest rated flavour, the barrel
provided a number of practical advantages. Further optimisation of the box
and barrel methods is recommended.
D) For the first time, the effect of added microbial nutrients (Fermaid K®) on
cocoa fermentation was evaluated. The use of microbial nutrients conferred
no clear advantages under the conditions of this study. Only minor
differences were observed in the microbiology and chemistry of the different
treatments. The addition of nutrients did not significantly affect the quality
or flavour of the beans produced. Flowever, given that no negative effects
were observed, microbial nutrients might be used in the future to prevent or
fix, stuck or sluggish cocoa fermentations.
E) Experiments to evaluate the effects of mixing frequency found that mixing
every 12 or 24 hours produced beans of poor quality and flavour. This
finding contradicts several studies recommending that fermentations be
mixed every 12 or 24 hours. It is recommended that future fermentations be
mixed every 48 hours. Mixing is suggested to be a critical fermentation
variable, since it strongly affected aeration and acid production.
199
Chapter Four
Evaluation of the use of mixed yeast starter
cultures in the fermentation of Australian cocoa
beans: effects on microbial ecology, chemistry,
quality and flavour.
4.1 Introduction
Quality chocolate is fast becoming an internationally recognised gourmet
commodity, taking its place alongside fine wine, cheese, coffee and artisan
bread (Guittard, 2005; Rosenblum, 2005; Doutre-Roussel, 2005; ICCO, 2007b). To
meet changing consumer demands, the larger scale chocolate manufacturers are
increasing production of dark and specialty chocolates (Anonymous, 2007;
ICCO 2008). Yet, even as demand for higher quality chocolate increases, the
quality of cocoa beans sold on the global market remains highly variable
(Beckett, 2000; ICCO, 2006). The major causes of inconsistent quality are
inadequate controls over the bean fermentation and drying processes. The use
of starter cultures in cocoa bean fermentation has been suggested as a way of
improving control over the process, and enhancing the quality of cocoa
produced (Schwan and Wheals, 2004). The complex microbial ecology of cocoa
fermentation presents a challenge in selecting suitable species for application as
starter cultures.
The importance of yeasts in cocoa bean fermentations is well recognised but, to

date, little is understood about the influence of individual species or strains on
bean quality and chocolate flavour. In several studies, pectinolytic strains of
Saccharomyces cerevisiae were evaluated as starter organisms, reportedly giving
cocoa beans that made acceptable chocolate (Sanchez et al. 1985, Buamah et al.,
1997). However, the influence of indigenous yeasts and bacteria that grew
during these fermentations were not adequately considered, and the starter
cultures consisted of a single species or strain only. Furthermore, many of the
strains used were not isolated from cocoa, but were taken from culture
collections. Experiments by Schwan (1998) were better designed and more
successful, but relied on obtaining sterile cocoa beans as a starting material. In a
200
commercial cocoa fermentary, sterility would be extremely difficult, and

probably uneconomical, to achieve. Freshly extracted beans would invariably
be contaminated with some level of indigenous microflora.
This situation is typical in many food fermentations, and a good analogy exists
in modem wine production methods. Yeast starter cultures are used extensively
in wine fermentations and have been shown to successfully compete with the
indigenous grape microorganisms (Fleet, 2007). There is a need to evaluate the
ability of added cultures to compete with indigenous microflora during cocoa
fermentations. The benefits of using added cultures also need to be better
demonstrated.
Chapter 3 has described some basic knowledge on the chemistry and microbial
ecology of cocoa bean fermentations using beans cultivated in Queensland.
After optimization of some fermentation and drying conditions, beans were
produced that met basic quality criteria, and which could be processed into
chocolate of acceptable quality. Chapter 3 also reported the possibility of
conducting the fermentation in barrels. Barrel fermentation allowed more
convenient handling and mixing of the beans, and provided a platform for
improved process control, based on the use of selected starter micro-organisms.
This chapter describes a series of experiments to develop and evaluate the use
of defined mixtures of yeast cultures for the fermentation of Queensland grown
cocoa beans. Several mixed cultures, based on species isolated from previous
cocoa fermentations, are considered and compared. The traditional box
fermentation process and the novel barrel fermentation process are also
compared.
The microbial ecology of the fermentation, and the chemical and physical
changes that occur in the cocoa pulp and seeds during fermentation are
reported and correlated with attributes of bean and chocolate quality.
Quantitative and descriptive sensory analyses were performed on chocolates
prepared from the beans to determine whether specific cultures contributed to
specific flavours. Additional chemical and biochemical analyses were
performed to determine the effects of starter cultures on the breakdown of
cocoa seed proteins, antioxidant activity of the beans, and on the flavour
volatiles present in roasted bean samples.
201
4.2 Materials and Methods

4.2.1 Fermentation of cocoa beans using starter cultures
Two series of experiments were conducted to investigate the use of starter
cultures in fermenting Australian cocoa beans: 1) laboratory fermentations
performed at UNSW, Sydney; and 2) larger, pilot scale fermentations
conducted at South Johnstone, Queensland. The cocoa pods for the
fermentations were obtained from Trinitario X Amazonia hybrid trees grown at
South Johnstone and Mossman plantations, as described in Chapter 3.
4.2.1.1 Laboratory fermentations
Cocoa pods were harvested from the plantations and packed into cardboard
boxes, each box containing 20 kg of pods. These boxes were transported by
truck to UNSW within 5 days of harvest, and stored for a further 2 days at 20°C.
Ripe and intact pods were selected and pre-washed with lukewarm water to
eliminate dust. The pods were then immersed in 0.1% hypochlorite solution for
10 minutes, followed by a rinse in warm potable water. After rinsing, cocoa
pods were placed on a bench surface that had been sterilised by spraying with
70% ethanol (Figure 4.1 a).
One minute prior to opening, the cocoa pods were sprayed with 70% ethanol.
The pods were opened by cutting with a sterilised stainless steel knife, and the
beans and pulp aseptically removed by hand using sterile surgical gloves. Three
kilograms of freshly extracted cocoa beans, were deposited in plastic containers
with dimensions 25 cm x 16 cm x 11 cm (4.5L) (Figure 4.1 b).
Fig. 4.1. a) Cocoa pods prior to washing and surface sterilisation; b) Sterile plastic container
containing freshly obtained cocoa beans (Photos: Huynh, T.N. - used with permission)
202
The base of these containers was drilled with 10 holes, each 1 cm in diameter
and 5 cm apart to allow drainage of fermentation sweatings. Before use, the
containers and their lids were sequentially washed in aqueous solutions of 10%
acetic acid, and 0.1% hypochlorite, swabbed with 70% ethanol, and dried in an
oven at 60°C.
The fermentation process was initiated by inoculating the beans with pre
prepared microbial cultures and the beans were then thoroughly mixed with a
sterilized spatula. Section 4.2.3. describes the selection and preparation of these
cultures. Immediately after inoculation, the boxes were covered with their fitted
lids to avoid contamination from air-borne microorganisms. The boxes were
placed on sterilised trays to allow the sweatings to escape through the holes in
the bottom of the boxes. These trays, and the bottom of the fermentation vessel
were wrapped with sterile aluminium foil to minimize environmental
contamination. Table 4.1 outlines the mixtures of cultures that were used to
inoculate each fermentation, in each box.
Table 4.1 - Combinations of starter cultures used for laboratory scale cocoa bean fermentations
Starter culture
Species comprising the starter culture mixture
designation*
“H.guilliermondii Hanseniaspora guilliermondii, Issatchenkia orientalis, Lactobacillus,

& I. orientalis ” plantarum, Pediococcus acidilactici, Acetobacter pasteurianus
“S.cerevisiae Saccharomyces cerevisiae, Kluyveromyces marxianus, L. plantarum,

& K.marxianus ” P. acidilactici, A. pasteurianus
H. guilliermondii, /. orientals, S. cerevisiae, K. marxianus, L. plantarum,

“All yeast”
P. acidilactici, A. pasteurianus
Control** No starter culture added.
* The culture mixtures were named according to the mixture of yeast species present, since all culture
mixtures contained the same combination of bacterial species.
** For the control, an additional box of cocoa beans was prepared, but not inoculated with prepared
cultures. Rather, the beans used were non-aseptically extracted from unwashed pods and fermentation
occurred by the growth of indigenous microflora.
The focus of this experiment was the evaluation of defined mixtures of yeast
species for use as starter cultures. However, to replace the bacterial microflora
that was presumed absent due to aseptic preparation methods, mixed cultures
of Acetobacter pasteurianus, Lactobacillus plantarum and Pediococcus acidilactici
were added along with each of the mixed yeast cultures.
203
Control fermentations were also conducted, without the addition of any starter
culture mixture (Figure 4.3). Cocoa pods used in the control fermentations were
not washed or sterilized, and the beans were removed without aseptic
techniques. To ensure the transfer of pod microflora to these cocoa beans, some
pod fragments were massaged into the bean mass, and then removed. The
beans were placed in plastic containers identical to those used for the
inoculated fermentations. As before, the fermentation box was closed with a lid
and the bottom half wrapped with aluminum foil.
Once prepared, all boxes (control and inoculated), underwent fermentation for
6 days (144h). To control the temperature of fermentation, the plastic boxes were
placed in incubators and the temperature adjusted according to the following
program: 0 - 24h, 25°C constant; 24 - 72h, 25°C to 37°C; 72 - 120h, 37°C to 45°C;
120 - 144h, 45°C to 50°C. The temperature of the fermentations were checked
every 24h using a sterile glass thermometer with a range of 15 - 50°C, and
gradients of 0.2°C. Samples of beans were taken every 24h from the beginning
(Oh) until the completion (144h) of fermentation using the following method:
The cocoa beans in each fermenter box were mixed thoroughly with a sterilized
spatula. Using sterile gloves, 50g of cocoa beans were aseptically removed from
the fermentation and placed into a stomacher bag for immediate
microbiological analysis. Due to the limited size of the laboratory scale
fermentations, samples (50g) for chemical analysis were only taken at the
beginning(Oh) and end (144h) of fermentation. The pH of the cocoa pulp was
measured every 24h, using the method of Ardhana and Fleet (2003). Thirty (30)
beans were mixed in a beaker with distilled water at a ratio of 1:1 by weight.
The pH of the homogenate was measured using an Activon 209 digital pH
meter (Activon, Sydney). Standard buffers of pH 4.0 and 7.0 were used to
calibrate the pH probe. Two fermentation trials were performed in this manner.
The first trial used cocoa beans harvested from the South Johnstone plantation,
and the second trial used cocoa beans harvested from the Mossman plantation.
Figure 4.3 gives an overview of the experimental plan used for the laboratory
fermentations.
The cocoa beans produced during laboratory fermentations at the University of
New South Wales were solar dried according to the following method: After
fermentation, the cocoa beans were spread onto 0.5 m x 1 m plastic trays and
placed in a well ventilated glasshouse for 7 days to dry (Figure 4.2 a, b).
204
The beans were examined daily for changes in appearances and mouldy beans
were discarded. The dried beans were placed in airtight plastic bags and stored
in a cool, dark room free from strong odours. Samples of dried fermented beans
were made into chocolate for sensory evaluation, while other samples were sent
to the Cocoa Quality Laboratory at MacRobertson's Foods Ltd., Singapore for
quality evaluation. The laboratory fermentations were conducted with the
assistance of Honours students TuAnh Ngoc Huynh, and Nessa Suriadi
Fig. 4.2 a) Cocoa beans in the plastic containers at the termination of fermentation (144h);
b) Cocoa beans spread onto trays for solar drying. (Photos: Huynh, T.N. - used with permission)
COCOA PODS'
Inoculated Uninoculated
fermentation fermentation
Wash with 0.1% hypochlorite
I
Cut open with sterilized equipment Cut open with non-sterile equipment
(sprayed with 70% ethanol)
COCOA BEANS
I COCOA BEANS
Deposit in containers
I
Deposit in containers
Inoculate starter cultures
ncubate at suitable temperatures-
I
Sun dry
I
Store at cool temperature, low humidity
Fig. 4.3. Experimental plan for the controlled laboratory fermentation of cocoa beans, with or
without added mixtures of yeast species.
205
4.2.2.2 Pilot-scale fermentations.
Cocoa beans harvested from the Mossman plantation were predominantly used
in these experiments because the South Johnstone plantation was destroyed by
a cyclone in April 2006. The beans were stored for 7 days at 25°C, followed by
mechanised pod splitting and bean removal using the same methods described
in Section 3.2.1.1, Chapter 3. One series of experiments were conducted with
beans harvested in August 2006 and another series was conducted with beans
harvested in November 2006.
The fermentations were performed in the same box and barrel vessels already
described in detail in Section 3.2.1.2 of Chapter 3. At the beginning of
fermentation, one box and one barrel, each containing 50 kg of beans, were
inoculated with mixtures of yeast cultures or left as uninoculated controls. Table
4.2 outlines the fermentations as conducted with different yeast inocula.
Table 4.2 - Experiments investigating the effects of added starter cultures on the microbiology,
chemistry and quality of pilot-scale cocoa fermentations.
Fermentation
Date Starter culture mixture
vessels used*
August
Hanseniaspora guilliermondii and Issatchenkia orientalis Boxes; barrels
2006
November Saccharomyces cerevisiae and Kluyveromyces marxianus;

Boxes; barrels
2006
November H.guilliermondii, I. orientalis, S.cerevisiae and

Barrels
2006 K. marxianus
Due to experimental design and favourable conditions, only solar drying was used in these experiments.
Cyclone damage severely reduced the crop size from the South Johnstone plantation.Therefore these
experiments used beans sourced from the Mossman plantation.
* For each experiment one box and/or one barrel was inoculated; the other was uninoculated (control).
The yeast cultures were selected and prepared as described in Section 4.2.3.
All fermentations were mixed at Oh, 48h and 96h. The beans in box
fermentations were mixed using large sterile wooden paddles, while the barrel
fermentations were mixed by turning the barrel on its axis 10 times. The
temperature of each fermentation was monitored using electronic temperature
recorders (Gemini Data Loggers, UK) that were inserted into the centre of the
bean mass. Care was taken to avoid any cross contamination of the
fermentations. Representative samples (1 kg) for microbiological and chemical
206
analyses were aseptically taken from the fermenting bean mass at regular
intervals throughout the process, as previously described in Section 3.2.1.2 of
Chapter 3. As before, microbiological testing of the bean samples was
conducted onsite in Queensland. Samples obtained for chemical testing were
frozen and transported back to UNSW.
The beans produced from the pilot scale fermentations were solar dried,
according to the methods described in Chapter 3. Oven drying was not used for
these experiments, only solar drying. The beans were dried to final moisture
content of less than 7%. After drying, beans were stored in self-sealed 15 cm x
30 cm plastic bags in a seed storage room at a temperature of 20°C, and a low
relative humidity of 15-25%. The bags of dried fermented beans were
transported to UNSW to be made into chocolate for sensory evaluation, while
other samples were sent to the Cocoa Quality Laboratory at MacRobertson's
Foods Ltd., Singapore for quality evaluation.
4.2.2 Preparation of starter cultures
4.2.2.1 Selection and screening of cultures for use as starter cultures.
Three different yeast species were chosen for evaluation in mixed yeast starter
cultures. Previous studies on the use of yeasts as starter cultures for cocoa
fermentation have focused on pectinolytic yeasts, especially strains of S.
cerevisiae and K.marxianus (Sanchez et al., 1985; Buamah et al., 1997; Schwan,
1998; Dzogbefia et al., 1999). It was, therefore, decided that one of the mixtures
to be evaluated should be comprised of pectinolytic isolates of S.cerevisiae and
K.marxianus (one isolate of each). In spite of their predominance in cocoa
fermentations in other countries, H.guilliermondii and I.orientalis have not
previously been considered as candidates for starter cultures. Therefore,
another mixture comprised an isolate of H.guilliermondii and one of I.orientalis.
The third mixture comprised a cocktail of all four yeast species, namely,
H.guilliermondii, I.orientalis, S.cerevisiae and K.marxia?ius.
Sub-groups of isolates, comprising each of the four species was chosen for
potential use as starter cultures (10 isolates each of H.guilliermondii, I.orientalis
and S.cerevisiae; 1 isolate of K.marxianus). These selected isolates were then
subjected to a basic screening protocol: Their identities were re-confirmed using
the same methods used to study the microbial ecology of the fermentations in
207
Chapter 3 (Section 3.2.3); and each isolate was screened for enzyme activity
(lipase, protease and pectinase) according to the methods of Charoenchai et al.,
1997. For the lipase activity test, the following modification was made to the
butter-fat agar: butter fat was replaced with cocoa butter. Due to project
constraints, the isolates were not characterised any further.
The identities of all isolates were successfully re-confirmed. The results of the
enzyme screening are reported in Table 5, Appendix A.
Finally, one isolate of each species was selected. In the case of the
H.guillermoiidii and I.orientalis, the isolates were positive for protease activity,
but negative for lipase and pectinase (pH 5.0 and 7.0) activities. The selected
S.cerevisiae isolate was positive for pectinase activity (pH 5.0 and 7.0), but
negative for lipase and protease activity. The K.marxianus isolate was positive
for pectinase (pH 5.0 and 7.0), weakly positive for protease, and negative for
lipase activity.
4.2.2.2 Preparation of cultures for laboratory scale fermentations
Cultures of the yeasts Hanseniaspora guiltiermondii, Issatchenkia orientalis,

Saccharomyces cerevisiae and Kluyveromyces marxianus, and the bacteria
Lactobacillus plantarum, Acetobacter pasteurianus, and Pediococcus acidilactici were
chosen from the isolates obtained during previous fermentations, as described
in Chapter 3.
Prior to use as starter cultures, all isolates were checked for purity by streaking
onto plates of Malt Extract Agar (MEA) for yeasts, and MRS Agar for bacteria.
The agar plates were incubated for 24h at 25°C for yeasts and 37°C for bacteria.
Purified biomass from freshly cultured plates was used to inoculate liquid
cultures as follows: Yeasts were grown in 1L of sterile YEPD medium (lOg
peptone, 20g yeast extract, and 20g glucose per 1L of distilled water, with pH
adjusted to 4.0 with 0.1N HC1) in Erlenmeyer flasks. Bacteria were grown in 1L
of sterile MRS medium (Oxoid, Melbourne) in Erlenmeyer flasks. After
inoculation, the flasks were covered with 10 cm x 10 cm pieces of sterile
Parafilm 'M' (Lab Depot Inc., USA) to exclude environmental contamination.
The cultures were incubated for 24h in a Gallenkamp orbital incubator
(Gallenkamp, Australia) at 25°C for yeasts, and 37°C for bacteria. The shaker
speed was set to 150 rpm. After growth, microbial cells were harvested
208
aseptically by centrifugation in sterilized tubes at 10,000 rpm for 20 minutes at

20°C (Beckman-Coulter, Fullerton CA). The supernatant was discarded and the
sedimented cells retained. For each fermentation, aliquots of the harvested cells
were re-suspended in 20mL of sterilized distilled water before being inoculated
into the cocoa beans. The concentration of cells in each suspension was
estimated microscopically using a counting chamber (Improved Neubauer,
Heinz Optics), and adjusted so that the initial population after inoculation and
mixing into beans was approximately 1x106/g (for each microbial species).
4.2.2.3 Preparation of cultures for pilot-scale fermentations
No facilities were available at the South Johnstone laboratories for large-scale

cultivation, centrifugation and harvesting of microbial cells from liquid
cultures. Consequently, cell biomass was cultured and harvested from the
surface of agar cultures. Selected isolates of Hanseniaspora guilliermondii,
Issatchenkia orientalis, Saccharomyces cerevisiae and Kluyveromyces marxianus were
transported as slants to the South Johnstone research station. Isolates were re
checked for purity by streaking onto plates of ME A. Biomass from these
cultures was then spread inoculated over the surface of fresh plates of MEA and
incubated at 30°C for 36 hrs to give a lawn culture For each yeast species, 15
plates of MEA lawn cultures were prepared. Yeast biomass from the 15 plates
was aseptically scraped off using a sterile cover slip and suspended in lOOmL of
sterile distilled water. Samples (50ml) of the suspension were then immediately
used to inoculate the beans (50 kg) in boxes or barrels according to the protocols
shown in Table 4.2. The concentration of yeast cells in suspension was estimated
microscopically using a counting chamber (Improved Neubauer, Heinz Optics)
and adjusted so that the initial population after inoculation and mixing into the
beans was approximately lxlO6 cfu/ g (for each yeast species).
4.2.3 Physical Analyses

4.2.3.1 Measurement of pH of pulp and seed
The pFI of samples of cocoa pulp and seed from each fermentation was
measured every 24h, as described already (Chapter 4, Section 4.2.2.1).
209
4.2.3.2 Pulp and seed weights
The pulp-bean ratio was calculated according to the methods described in

Chapter 3, Section 3.2.2.2.
4.2.3.3 Moisture content of beans, pulp and seed
The moisture contents of whole beans, pulp and seeds were determined
according to standard methods described by AO AC (1984). For more details of
this method, refer to Section 3.2.2.4 in Chapter 3
4.2.4 Microbiological analyses
The samples taken from the cocoa fermentations described in Section 4.2.1 were
analysed for their populations of yeasts, lactic acid bacteria, acetic acid bacteria
and total bacteria according to the methods given in Section 3.2.3 of Chapter 3.
Representative isolates of the different microorganisms were identified to genus
and species level by their morphological, biochemical and physiological
properties and rDNA sequencing as described in Section 3.2.3.
4.2.5 Chemical analyses
Samples of beans taken during fermentation were prepared and analysed for
sugars, ethanol and organic acids, in both the pulp and seed fractions,
according to the methods described in Section 3.2.4 in Chapter 3. To assist in
comparison with the literature, the concentration of sugars and ethanol were
expressed as % w/w, while the concentrations of organic acids were expressed
as mg/g dry basis. The maximum variability between duplicate samples was
±0.2% for sugars, ±0.1% for ethanol and ±3 mg/g for organic acids.
4.2.6 Quality and sensory assessment of cocoa beans
4.2.6.1 Quality assessment of dried fermented cocoa beans
Samples of dried cocoa beans from pilot scale fermentations were analysed for
commercial quality criteria by the Cocoa Quality Laboratory, MacRobertson's
Foods Ltd., Singapore. The following quality parameters were tested:
fermentation index by cut test; moisture content of the bean and nib, % shell by
weight, nib fat content and nib pH. The methods used for these tests are
210
described in Section 3.2.5.1, Chapter 3. Basic subjective assessments were also

made of the external appearance of the beans, and of their aroma. To assess the
aroma, 100g of dried fermented beans were ground in a knife-mill for 30s and
then the strength of cocoa aroma was noted, along with any unpleasant aromas
or taints by smelling by trained staff.
To verify the results from the external laboratory, the cut test was repeated on
all samples at the Food Science Laboratories, School of Chemical Sciences and
Engineering, UNSW Sydney.
The beans produced by laboratory scale fermentations were tested for quality at
UNSW, as there was insufficient sample for dispatch to the Singapore
laboratory. All tests apart from fat content were performed. The methods
described in Section 3.2.5.1 (Chapter 3) were used, except that 50 beans were
tested rather than 100.
4.2.6.2 Chocolate Preparations
The dried cocoa bean samples produced by the experimental fermentations

were made into chocolate for sensory evaluation. The chocolate was made
using the same methods described in Section 3.2.5.2, without variation or
modification.
When making chocolate from the cocoa beans produced by the laboratory-scale
fermentations, Mossman and South Johnstone cocoa beans from each culture
mixture were combined. For example, South Johnstone beans fermented using a
culture of H.guiltiermondii and l.orientalis were blended with Mossman beans
fermented using the same culture. This blend of beans fermented by the same
yeast species was used to make chocolate.
4.2.6.3 Quality evaluation of chocolate samples from pilot fermentations
inoculated with defined mixtures of yeast.
In order to confirm their commercial suitability and safety, additional tests were
performed on these chocolate samples. Samples from each batch of chocolate
were sent to Analytical Laboratories at Cadbury-Schweppes, Claremont,
Tasmania Australia. This laboratory is a commercial chocolate testing laboratory
that routinely performs quality and safety testing.
211
The following physical and chemical tests were performed in this laboratory on
the chocolate samples: % nitrogen (protein); % fat; % sucrose; % lactose; %non-
fat solids; particle size; solid fat content and fat melting curves. The test
methods were all based on AOAC methods (AO AC, 1997), except for the fat
melting curves which used the Padley-Timms method (Padley-Timms, 1980).
Basic microbiological testing was also performed on the samples to further

assess quality and safety: Standard plate count, osmophilic yeasts and moulds,
coliforms, and Salmonella spp. (ICMSF 2000; Standards Australia 1991, 1994).
4.2.6.4 Sensory evaluation of chocolate samples by untrained panelists
The chocolate samples were subjected to sensory evaluation using methods

described in Section 3.2.5.3 of Chapter 3. Both triangle (difference) and liking
tests were performed. The same liking scale was used as before; a liking score of
0 indicated that a chocolate sample was "strongly disliked," while a score of 10
indicated it was "strongly liked." A score of 5 indicated basic acceptability.
4.2.6.5 Quantitative descriptive analysis of chocolate samples
Descriptive sensory analysis was also performed on the chocolate samples as

follows: Samples (1.5 kg) of each batch of chocolate were sent for testing to the
Cadbury-Schweppes Global Science Centre, The Lord Zuckerman Research
Centre, University of Reading, Whiteknights, Reading UK. The descriptive
analysis was performed by trained panelists who generated profiles of melted
chocolate samples based on attributes of aroma, texture, taste and aftertaste.
Twenty five grams of chocolate sample were placed in 75 x 26 mm soda glass
specimen tubes, stoppered and labeled with a random sample code.
The sample tubes were placed in Techne-Driblock heaters set at 56°C and left to
melt for 15 minutes. The melted chocolate samples were then evaluated
immediately. The technique of Quantitative Descriptive Analysis was used to
define and rate the sensory characteristics of the samples (Carpenter et al., 2000;
Stone and Sidel, 2004). A list of terms for describing the appearance, odour,
mouthfeel, flavour and aftertaste of chocolate samples was derived during a
series of round-table sessions (Table 4.3). Trained assessors, in individual
sensory evaluation booths, then scored the intensity of the key characteristics
associated with each sample. Samples were assessed in duplicate, with the
order of sample presentation balanced across assessors and replicates. Data
212
were collected via the Compusense data capture system (Compusense 5,

Canada) and analysed with Senpak using analysis of variance (ANOVA). The
Fisher least significant difference (LSD) was calculated in order to determine
those attributes which were significantly different (p<0.05). Results for the
main attributes and comparisons of specific products were plotted as spider
plots given in Figures 4.32 to 4.34 in Section 4.3.5.
It is worth noting that many of the primary flavour attributes identified during
the round-table sessions (astringent, cocoa, acidity, bitterness, brown fruit)
(Table 4.3) were the same as those used in previous studies (Lopez and
McDonald, 1981; Duncan et al., 1989; Hoskin, 1994; Schwan, 1998; Assemat et
al., 2005; Fraudorfer and Schieberle, 2006; Ramli et al., 2006; Stark, 2006).
Table 4.3 - List of sensory attributes used in Quantitative Descriptive Analysis of chocolate
Attribute Category* Definition
Colour App Overall intensity of colour assessed from

light brown (low score) to dark brown (high score)
Body MF Viscosity of melted chocolate
Drying MF Degree to which the chocolate dries inner surfaces of mouth after
swallowing
Astringent MF Degree to which throat and mouth are irritated during swallowing
Fruity O, F, A Associated with apple, pineapple, mango exotic-fruity
Dusty O, F, A Typically associated with dusty aroma
Doughy O, F, A Typically associated with uncooked bread dough
Nutty O, F, A Typically associated with roasted nuts and nut skins
Spicy O, F, A Typically associated with dried spices
Estery O, F, A Typically associated with estery/nail varnish
Brown Fruit O, F, A Typically associated with sultanas and raisins
Cocoa Degree of roast cocoa liquor aroma/flavour present in the

O, F, A
chocolate
Nutskins/grain F,A Typically associated with peanut skins and/or cereal grain
Acidity F, A Basic taste of acidity
Sweetness F, A Basic taste of sweetness
Bitterness F, A Basic taste of bitterness

Sugar O The smell of caramelised sugar
Acetic O The smell of acetic acid, typically associated with white vinegar
* A pp-Appearance O-Odour; MF - Mouthfeel; F-Flavour; A-Aftertaste
213
4.2.7 SDS PAGE analysis of protein degradation during cocoa

bean fermentation
SDS-PAGE was used to monitor the breakdown of cocoa seed proteins during
the pilot scale fermentations. The methods for protein extraction and
electrophoresis were adapted from Buyukpamukcu et al. (2001). Cocoa bean
samples were obtained from the beginning (Oh), middle (48 or 72h) and end (96
or 120 h) of the fermentation period. The pulp was removed from the seeds by
hand peeling, and the seeds (20g) were ground in a cooled mortar and pestle to
a rough paste (largest particle size = 4mm). Ground cocoa was stored in the
dark in airtight containers at -20 °C until use. Aliquots (100 mg) of ground
cocoa seed samples were mixed with lmL of extraction buffer [100 mM Tris-
acetate, pH 8.1, 1% sodium dodecyl sulfate (SDS) and 1% dithiothreitol].
Mixtures were heated at 95 °C for 30 min to denature the proteins. Samples
were allowed to cool and subsequently centrifuged (18000g for 5 min).
Supernatants were filtered (0.2 |um) and prepared for SDS-PAGE by addition of
4 volumes of sample running buffer (glycine system buffer, BioRad
Laboratories, Hercules, CA), containing 0.005% (w/ v) bromophenol blue as
tracking dye. SDS-PAGE gels were prepared and run according to the method
of Laemmli (1970), using a PROTEAN II xi cell and PowerPac 3000 power
supply (BioRad). The gels were run for 4 hrs at 40mA, and were cooled at 4°C.
Molecular weight standards (Cat. No. 161-0304, BioRad) were run on each gel
for the calculation of cocoa protein-fraction size. Gels were stained with
Coomassie Blue and visualised using an Alphalmager 2200 (Alpha Innotech,
San Leandro CA).
4.2.8 Analysis of antioxidant activity of cocoa polyphenols
4.2.8.1 Extraction of polyphenols from cocoa beans.
A method described by Counet et al. (2004) was adapted for extraction of the
polyphenolic compounds from samples of fermented cocoa beans. Samples
(lOOg) of frozen cocoa beans obtained from the fermentations were thawed and
hand peeled, to separate the seeds from the pulp material. For Oh and 72h
samples, lOOg of beans from each fermentation were mixed together, and then
lOOg of the composite sample was taken and peeled.
214
Seed samples (lOOg) were freeze-dried using a Lyovac GT 2 (Leybold-Heraeus,

Koln, Germany) for 24 h. Freeze dried beans (50g) were reduced to a powder by
grinding in a mortar and pestle. Twenty grams of each sample were defatted in
a Soxhlet apparatus with diethyl ether (375 mL) for 5 hours. Polyphenolic
compounds were extracted from the defatted cocoa samples (10 g) with 50 mL
of acetone:water:acetic acid (70:28:2, v/ v). Each sample was extracted three
times for 1 hour at room temperature. After each extraction, the suspension was
centrifuged for 10 min at 10,000g, and the supernatant was collected. After
filtration to remove residual particles, the combined supernatants were
concentrated by rotary evaporation under partial vacuum (40 °C) to obtain
about 50 mL of extract. A Cl8 Sep-Pack cartridge (Waters, Millipore) was
preconditioned by washing with methanol followed by double distilled water
(MilliQ, Millipore). Aliquots (5mL) of extracts were loaded onto the cartridge,
and sugars were removed by elution with 20 mL of deionized water.
Polyphenols were then eluted with 3 mL of acetone-water-acetic acid (70:28:2,
v/ v). The eluates were concentrated by rotary evaporation under partial
vacuum (40 °C) and freeze-dried.
4.2.8.2 Analysis of polyphenolic compounds in cocoa extracts
Detection of polyphenolic compounds in cocoa extracts was performed by

HPLC according to a method adapted from Counet et al. (2004).
The HPLC was performed using a LC-10AT system (Shimadzu, Kyoto Japan)
equipped with a DGU-14A de-gasser, a FCV-10AL pump system and a RF-10A
XL combined fluorescence/diode array detector. The system was controlled
with Class VP software (Shimadzu). The polyphenols were separated on a 5 pm
normal-phase Luna silica column, 250 x 4.6 mm i.d. (Phenomenex, Torrance
CA) at 25 °C. Separations were carried out at a flow rate of 1 mL/min with a
linear gradient from A (dichloromethane) to B (methanol) and a constant 4%
level of C (acetic acid and water, 1:1, v/ v). The elution gradient was: 0-30 min,
14-28% B; 30-60 min, 28-50% B; 60-65 min, 50-86% B; 65-70 min, isocratic. Cocoa
extracts were diluted in methanol before injection (10 mg/mL) into the HPLC.
The excitation and emission wavelengths were 276 and 316 nm, respectively, for
fluorescence detection, while the photodiode array detector was set to 280 nm.
Procyanidins were quantitated against (-)-catechin (Sigma) in aqueous standard
solutions of 0.5, 1.0, 2.0 and 4.0 mg/g.
215
4.2.8.3 Determination of total antioxidant capacity (AOC) of cocoa extracts
The total antioxidant capacities (AOC) of the polyphenol extracts

(Section 4.2.8.2) were analysed using a modification of the oxygen radical
absorbance capacity (ORAC) assay described in Brand-Williams et al. (1995)
and Lohachoompol (2007). The antioxidant activities were determined using 2,
2-diphenyl-l-picrylhydrazyl (DPPH) as a free radical. The stock solution was
prepared by dissolving 2.4 mg in 10 ml methanol (spectrophotometric grade).
The stock solution was kept at -20 °C until used. The working solution was
obtained by mixing 10 ml of the stock solution with 45 ml of methanol to obtain
an absorbance of 1.10+ 0.020 at 515 nm. Diluted solution (0.15ml) of the extract
was allowed to react with 2.85 ml of the working solution for 24 hours in the
dark at ambient temperature. The absorbance was then taken at 515nm. The
decrease in the quantity of the radical (DPPH) due to the antioxidants was then
calculated. All samples were tested in duplicate and the antioxidant activity
was expressed as pmol Trolox1 equivalent/ g dried weight.
4.2.9 Gas chromatographic analysis of cocoa aroma volatiles
4.2.9.1 Extraction of volatile compounds from roasted cocoa beans
Volatile aroma compounds were extracted from dried, fermented cocoa bean
samples produced during the August and November 2006 pilot scale
fermentations using a modification of the method described by Yoo et al. (1998)
and Jinap et al. (1998).
Approximately 250g of manually peeled cocoa bean nibs were placed in a steel
tray (35 cm x 20 cm) and spread in a single layer. The samples were then roasted
one at a time using the same method and conditions as for the production of
chocolate - a temperature of 120°C for 55 minutes. The roasted samples were
cooled in closed containers, and the oven was vented for lh between samples to
prevent aroma transfer. Eighty (80) grams of each roasted cocoa sample, along
with 800g of distilled water were blended in a Waring blender for 2 minutes to
create a suspension of cocoa in the water. This suspension was poured into a
round bottom flask and then attached to a Likens-Nickerson apparatus for
continuous steam distillation extraction (SDE) to extract the cocoa aroma
1 Trolox is an antioxidant used as a standard reference in AOC tests(Brand-williams et al., 1995).

216
volatiles. The solvent flask contained 50 mL of dichloromethane as the co

extraction solvent. The water baths for the sample and solvent flasks were
100°C and 50°C respectively. Dodecane (1.2 mg) was added as an internal
standard for quantitation. The suspension of cocoa sample was heated and co
extracted for 4 hr into the dichloromethane. After cooling to room temperature,
the volatile extracts were concentrated in a rotary evaporator at 30°C for 10
minutes and then further concentrated by nitrogen stream for a further 2
minutes.
4.2.9.2 Gas chromatography-flame ionisation analysis
The volatile extracts were analysed on a Shimadzu GC-17A gas chromatograph

(Shimadzu, Kyoto, Japan), equipped with a flame ionisation detector (FID). The
method used was based on that of Yoo et al. (1998) and Jinap et al. (1998). A
fused silica capillary column was used, having dimensions of 30 m x 0.2 mm
I.D., and a 0.5 pm Supelcowax coating (Supelco, Bellefonte, PA).
The operation conditions were as follows: injector temperature 250°C; detector
temperature 280°C; The GC oven was initially held at 25 for 5 minutes, and
ramped at 1°C/ min to 80°C and then at 5°C/ min to 260°C, which was held for
10 min. Nitrogen was used as a carrier gas, with a constant flow rate of 3.9cm /
minute. Two micro-litres (2pL) of sample were injected in a lined glass injector
port, with a split ratio of 1:15. Volatile component profiles were identified by
using gas chromatograph-mass spectrometry. Quantities of individual
compounds were estimated by comparing relative peak areas against that of the
internal standard, dodecane.
4.2.9.3 Gas Chromatography-mass spectroscopy analysis
The extracted cocoa aroma volatiles were also analysed using an Agilent 5975
gas chromatograph-mass spectrometer according to the methods of Yoo et al
(1998). Gas chromatograph conditions were as before. Mass spectra were
obtained by mass-ionisation at 70 eV; ion-source temperature was 180°C, the
filament emission current was 41.mA. Data from the mass spectrometer was
recorded and analysed using HP Chemstation™. All mass spectra were
identified by comparison with the NIST library. The results were analysed by
principal component analysis using the StatisticXL™ software package
(Broadway-Nedlands, WA, Australia).
217
4.2.10 Summary of experimental plan
Figure 4.4 illustrates the various components of the experiments and analyses
described in this chapter.
Harvest and storage of

cocoa pods
Measuring
Temperature Splitting of Pods
Fermentation
Microbiological
testing (See
Sampling
separate
flowchart)
Singapore for
quality testing
Frozen and transported to UNSW for testing
Quality testing
Antioxidant Physical
activity analyses: Chocolate making
• % H20
• Bean:pulp ratio
Sensory testing
SDS-PAGE of
cocoa proteins
HPLC analysis:
• Sugars Additional quality
• Ethanol assessment
• Organic acids
GC/MS of cocoa Quantitative,

aroma volatiles descriptive
sensory analysis
Fig. 4.4. Outline of methodology used for the sampling and testing of cocoa fermentations.
Sections in RED indicate methods previously described in Chapter 3. Sections in BLACK
indicate methods new to Chapter 4.
218
4.3 Results
4.3.2 Laboratory-scale fermentation of cocoa beans by inoculation
with defined mixtures of yeast
Controlled laboratory fermentations were conducted at UNSW to evaluate the

survival and growth of defined mixtures of yeasts, and their effects on cocoa
bean and chocolate quality. Data were obtained regarding the microbial ecology,
chemistry and quality of the cocoa beans. Sensory evaluations were also
performed to determine how different mixtures of yeast might affect the flavour
of chocolate prepared from these cocoa beans.
The results from laboratory fermentations were subsequently used in designing
the pilot-scale fermentations that also used defined cultures of yeasts.
Figures 4.5 and 4.6 show the changes in temperature, pH and total populations
of yeast and bacteria during these fermentations.
During fermentation, the temperature of the South Johnstone cocoa beans
increased from an initial temperature of 25°C to a final value of 50°C.
The control (uninoculated) fermentation increased in temperature more rapidly
than the inoculated fermentations, reaching 45°C at 72 h, and then increasing
more slowly to 50°C at 144 h (Figure 4.5 d). In the inoculated fermentations, the
temperature increased in an incremental fashion, mirroring the increasing
temperature program of the incubators (Figure 4.5 a, b, c).
The pulp of the South Johnstone beans had an initial pH of 3.9. In all
fermentations, except the one inoculated with S.cerevisiae and K.marxianus, the
pH of the pulp increased to a final value of 4.4. In the fermentation using
S.cerevisiae and K.marxianus, the final pH was 4.0 (Figure 4.5 b).
In the uninoculated (control) fermentation, the initial population of yeast was

103 cfu/ g beans, while for bacteria it was 105 cfu/g beans. These populations
increased to maximum levels of 107 and 108 cfu/g, respectively, during the first
48 hr, and then declined (Figure 4.5 d). No yeasts (less than 50 cfu/g) were
detected after 96 h and only 101 cfu/g bacteria were found at the end of
fermentation.
219
As expected, the inoculated fermentations had higher initial populations of

yeast (106cfu/g) and bacteria (107 cfu/g), compared to the control. In all of the
inoculated fermentations, maximum populations of yeast (107-108 cfu/g) were
reached after 24h.
In the fermentations inoculated with H.guilliermondii and I.orientalis, and
H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus, the yeast population
remained between 106-108 until 96h, after which it declined rapidly
(Figure 4.5 a, c). A similar pattern of yeast growth was observed for the
fermentation inoculated with S.cerevisiae and K.marxmius, although the decline
occurred 24 h earlier (72h) (Figure 4.5 b).
The bacterial population increased in the fermentation inoculated with
H.guilliermondii and I.orientalis, to a maximum of 109 cfu/g beans at 24 h. After
this, the population of bacteria decreased to 107 cfu/g at 72 h, and then to
undetectable levels by the end of fermentation (144 h) (Figure 4.5 a).
In the fermentation inoculated with S.cerevisiae and K.marxianus, the population
of bacteria increased to a maximum of 108 cfu/g, remained steady until 72 h,
and then died off (Figure 4.5 b).
Finally, in the fermentation inoculated with H.guilliermondii, I.orientalis,
S.cerevisiae and K.marxianus, the total population of bacteria did not increase
much past 107 cfu/ g beans, but remained steady until 96 h, and then died off
(Figure 4.5 c).
220
O' 40 -
% 35 -
24 48 72 96 120 144 24 48 72 96 120 144

U 40 -
H 35 -
25 * t-r1
120 144
U 40 -
K 35 -
120 144 24 48 72 96 120 144

^ 7 -
u 40 -
£ 35 -
120 144 24 48 72 96 120 144

Fig. 4.5. Laboratory-scale fermentation of cocoa beans inoculated with different yeast species:
(a) H.guilliermondii and I.orientalis, (b) S.cerevisiae and K.marxianus, (c) H.guilliermondii,
I.orientalis, S.cerevisiae and K.marxianus, (d) uninoculated (control).
Cocoa beans obtained from the South Johnstone plantation; — Incubation temperature;
Total yeast population ♦; Total bacterial population □; Temperature ■; pulp pH O.
221
When the laboratory-scale fermentations were repeated using cocoa beans from
the Mossman plantation, similar results were obtained. As before, the
temperature increased from 25°C to 50°C over the course of the fermentation. In
the control, and in the inoculated fermentations, this increase was observed to
be slightly incremental. Slower rates of temperature increase in the
fermentations coincided with periods of constant incubator temperature
(48-72h, 96-120h)(Figure 4.6. a, b, c & d).
In all fermentations (control and inoculated), the pH of the pulp increased from
an initial value of 3.9, to final values of 4.3-4.4. This increase occurred at similar
rates in all four fermentations (Figure 4.6. a, b, c & d).
The initial populations of yeast and bacteria in the control fermentation using
Mossman beans were 8xl02 cfu yeast/g beans and 104 cfu/g beans respectively.
These levels were about ten-fold lower than those recorded for the South
Johnstone control (compare Figure 4.6, d with Figure 4.5, d). From Oh, the
populations of yeast and bacteria increased, reaching maximum levels at 48h
(107 and 108 cfu/g beans respectively). After this, the levels of yeast and bacteria
decreased to final populations of 103 cfu/g beans. This decrease was fastest for
the bacteria between 72-96 h, and fastest for the yeast between 120-144h (Figure
4.6 d).
Higher initial populations of yeast and bacteria were found in the inoculated
fermentations (about 106 cfu/g beans for both yeast and bacteria). In the
fermentation inoculated with H.guilliermondii and I.orientalis, maximum
populations of both yeast (107cfu/g) and bacteria (108 cfu/g) were detected
after 24h (Figure 4.6 a). They then declined until 144h, when the final
populations of yeast and bacteria were each found to be 102 cfu/g beans. In the
fermentations inoculated with S.cerevisiae and K.marxianus, and H.guilliermondii,
I.orientalis, S.cerevisiae and K.marxianus, the total populations of yeast and
bacteria followed similar patterns (compare Figure 4.6 b & c).
The yeast increased to maximum populations of 107-108 between 24-48 h and
then died off. No yeast (<50 cfu/g) were detected at the end of these
fermentations. Similarly, the bacteria grew to maximum populations of 6xl08
cfu/g beans at 24 h in both of these fermentations. The bacterial populations
then decreased slightly to 107 cfu/ g and remained at this level for 24 h.
From 72 h onwards the bacterial populations in these fermentations died off.
222
'S' 7 -
U40 -
S 35 -
120 144 24 48 72 96 120 144

'S 7 -
U 40 -
% 35 -
120 144 24 48 72 96 120 144

U 40 -
g 35 -
§ 30 -
120 144 24 48 72 96 120 144

1? 7 -
U 40 -
^ 35 -
120 144 24 48 72 96 120 144

Fig. 4.6. Laboratory-scale fermentation of cocoa beans inoculated with different yeast species:
(a) H.guilliermondii and I.orientalis, (b) S.cerevisiae and K.marxianus, (c) H.gui/liermondii,
I.orientalis, S.cerevisiae and K.marxianus, (d) uninoculated (control).
Cocoa beans obtained from the Mossman plantation; — Incubation temperature;
Total yeast population ♦; Total bacterial population □; Temperature ■; pulp pH O.
223
4.3.2.2 Growth of yeast species during laboratory fermentations
Four species of yeast were consistently isolated from the laboratory-scale

fermentations: Issatchenkia orientalis, Hanseniaspora guilliermondii, Saccharomyces
cerevisiae and Kluyveromyces marxianus. Figure 4.7 shows the changes in
population of each of these species.
In the control fermentations, the population of H.guilliermondii increased from

initial levels of around 102 cfu/ g beans to maximum populations, and then died
off. When cocoa beans from South Johnstone were used, the maximum
population in the control was 1 x 103 cfu/g beans at 24h. When cocoa beans
from Mossman were used, the maximum population in the control was 3 x 106
cfu/g beans at 48h (Figure 4.7 d, h). In the fermentations inoculated with
mixtures of H.guilliermondii and Lorientalis, the population of H.guilliermondii
increased to 1 x 107 cfu/g beans at 24h, and then died off over the next 48h
(Figure 4.7 a, e).
In the fermentations inoculated with mixtures of H.guilliermondii, Lorientalis,
S.cerevisiae and K.marxianus, the population of H.guilliermondii did not increase,
but instead decreased from Oh onwards (Figure 4.7 c, g). Low levels (lOMO3
cfu/g beans) of H.guilliermondii were also detected in the fermentations
inoculated with S.cerevisiae and K.marxianus, and these died off by 48 h (Figure
4.7 b, e).
In the control fermentations, S.cerevisiae increased in population from initial
levels of 103 cfu/g beans to maximum populations of approx. 107 cfu/g beans at
48h, after which it died off. The initial population of S.cerevisiae in the
fermentations inoculated with S.cerevisiae and K.marxianus, or H.guilliermondii,
Lorientalis, S.cerevisiae and K.marxianus was 106 cfu/g beans. As the inoculated
fermentations proceeded, the population of S.cerevisiae increased to maxima of
107 - 108 cfu/g beans, then decreased from 48h onwards (Figure 4.7 b, c, f & g).
S.cerevisiae was also detected in the fermentations inoculated with
H.guilliermondii and Lorientalis, where it grew from low levels (< 103 cfu/g
beans ) to maximum populations of between 105 - 105 cfu/g beans, and then
died off (Figure 4.7 a, e).
224
24 48 72 96 120 144 24 48 72 96 120 144

24 48 72 96 120 144 24 48 72 96 120 144

ox) 5 -
% 3 -
24 48 72 96 120 144 120 144

ox) 5 -
0 24 48 72 96 120 144 0 24 48 72 96 120 144

Fig. 4.7. Growth of yeast species during laboratory-scale fermentation of cocoa beans
inoculated with different yeast species: (a)&(e) H.guilliertnondii and I.orientalis, (b)&(f)
S.cerevisiae and K.marxianus, (c)&(g) H.guilliermondii, I. orientalis, S.cerevisiae and
K.marxianus, (d)&(h) Uninoculated (control); Hanseniaspora guiUiermondii ♦;
Issatchenkia orientalis O; Saccharomyces cerevisiae □; Kluyveromyces marxianus A .
(a) to (d) cocoa beans obtained from the South Johnstone plantation;
(e) to (h) cocoa beans obtained from the Mossman plantation.
225
K.marxianus was only detected in those fermentations to which it had been

added as part of a mixed culture (Figure 4.7 b, c, f & g). The initial population of
K.marxianus varied somewhat between the fermentations, in a range of 104 -106
cfu/g beans. In the fermentations inoculated with Sxerevisiae and K.marxianus,
the population of K.marxianus rose slightly between 0 - 24h, before decreasing
again. In the fermentations inoculated with the four species of yeast, the
population of K.marxianus decreased from Oh onwards. K.marxianus was not
detected in any fermentation after 48h.
It was apparent from these results that yeast species were isolated from
fermentations to which they had not been added. For example, S.cerevisiae was
isolated from the fermentations inoculated with I.orientalis and H.guilliermondii
(Figure 4.7 a, e). Similarly, I.orientalis and H.guilliermondii were isolated from the
fermentations inoculated with S.cerevisiae and K.marxianus (Figure 4.7 b, f). This
observation will be considered in the discussion.
Some kind of competitive inhibition was also observed in the fermentations

from which both I.orientalis and S.cerevisiae were isolated. For example, in the
fermentations inoculated with I.orientalis and H.guilliermondii, the populations
of S.cerevisiae were lower than in the controls (Figure 4.7 a, e). Likewise, in the
fermentations inoculated with S.cerevisiae and K.marxianus the contaminating
populations of I.orie7italis were severely inhibited (Figure 4.7 b, f).
4.3.2.3 Growth of bacterial species during laboratory fermentations
Figure 4.8 shows the changes in population of individual species of bacteria that
were isolated from laboratory-scale fermentations conducted using mixed
defined microbial cultures.
Four bacterial species consistently isolated from the fermentations were:

Acetobacter pasteurianus, Lactobacillus plantarum, Pediococcus acidilactici and
Lactobacillus fermentum. Several Bacillus species were also isolated from some of
the fermentations.
Acetobacter pasteurianus was found in the control fermentations and the

inoculated fermentations. In both of the control fermentations, A.pasteurianus
grew according to a similar pattern. It was initially found in both controls at
104 cfu/g beans, reached maxima of 108 cfu/g beans at 48h, before again
declining in population from 72h onwards. This decline was faster in the
226
fermentation using cocoa beans from the Mossman plantation (Figure 4.8 d, h).
In the inoculated fermentations using South Johnstone beans, A.pasteurianus
was detected at 106 - 107cfu/g beans (Figure 4.8 a, b, c). In the inoculated
fermentations using Mossman beans it was initially detected at 105 cfu/ g beans
(Figure 4.8 e, f, g). In the inoculated fermentations, A.pasteurianus reached
maximum populations between 24 - 72h, depending on the fermentations, and
then died off.
In the control fermentation using South Johnstone cocoa beans, Lactobacillus
plantarum was initially detected at 2 x 104 cfu/g beans, increased to a maximum
population of 105 cfu/g beans at 24h, and then died off (Figure 4.8 d).
L.plantarum was not detected in the control fermentation using Mossman cocoa
beans (Figure 4.8 h). L.plantarum grew in a similar fashion in all inoculated
fermentations. In the inoculated fermentations using South Johnstone beans,
L.plantarum was initially found at 106 - 107cfu/g beans, and 105 - 106 cfu/g
beans in the inoculated fermentations using Mossman beans. In all
fermentations, maximum populations of L.plantarum were reached at 24h, after
which it died off. This decline was faster in the fermentations using cocoa beans
from the South Johnstone plantation (compare Figure 4.8 a, b, c with e, f, g).
Pediococcus acidilactici was only detected in the fermentations to which it had
been added , and was not found in the (uninoculated) controls. P.acidilactici was
also observed to grow differently depending on the source of the cocoa beans.
In the fermentations using South Johnstone cocoa beans, P.acidilactici was
initially detected at 106 - 107 cfu/g beans. It increased to maximum populations
of 108 at 24-48h and then died off (Figure 4.8 a, b, c). In the fermentations using
Mossman cocoa beans, P.acidilactici steadily decreased from initial populations
of 103 - 104 cfu/g beans and was not detected after 48h (Figure 4.8 e, f, g).
Lactobacillus fermentum grew according to common pattern in most of the
fermentations. Initially this species was detected at levels ranging from 102 -104
cfu/g beans. It then increased to maxima of 107 cfu/g beans between 24 - 72h,
before dying off. In the fermentation using Mossman cocoa beans, and
inoculated with H.guilliermondii and I.orientalis, the maximum population
reached was lower, at 4 x 105 cfu/g beans (compare Figure 4.8 e with a, b, c, d, f,
g & h). L.fermentum was not detected in the inoculated fermentations using
cocoa from the South Johnstone plantation.
227
120 144 120 144

24 48 72 96 120 144 24 48 72 96 120 144

OX) 2 -
120 144 120 144

120 144 120 144

Fig. 4.8. Growth of bacterial species during laboratory-scale fermentation of cocoa beans
inoculated with different yeast species: (a)&(e) H.guilliermondii and I.orientalis,
(b)&(f) S.cerevisiae and K.marxianus, (c)&(g) H.guilliermondii, I.orienta/is, S.cerevisiae and
K.marxianus, (d)&(h) uninoculated (control);
Acetobacterpasteurianus ♦/ Lactobacillusplantarum □; Pediococcus acidi/actici O;
Lactobacillus fermentum A; Bacillus spp. ▲.
Graphs (a) to (d) - cocoa beans obtained from the South Johnstone plantation;
Graphs (e) to (h) - cocoa beans obtained from the Mossman plantation.
228
In the fermentations using cocoa beans from the Mossman plantation, various
members of the genus Bacillus were isolated. The most commonly isolated
Bacillus species were B.subtilis, B.licheniformis and B.megaterium.
In the control, Bacillus spp. were found at levels of between 102 - 103 cfu/ g, for
the duration of fermentation (0-144h). In the fermentations inoculated with the
yeast, H.guilliermondii and I.orientalis, and S.cerevisiae and K.marxianus, various
Bacillus spp. were detected at 96h (101 -102 cfu/g beans). Maximum populations
(103 cfu/g) were detected at 120h, after which there was a decline. Bacillus spp.
were not isolated from the fermentation inoculated with the culture containing
H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus.
The following general trends were apparent in the bacterial growth curves
shown in Figure 4.8:
(a) The bacterial species grew according to quite different patterns in the
control fermentations, compared to the inoculated fermentations.
(b) In the South Johnstone control, L. plantarum was detected, but Bacillus
spp. were not. In the Mossman control, L. plantarum was absent while
Bacillus spp. were isolated throughout.
(c) L. fermentum grew for a shorter period in the control fermentation using
Mossman cocoa beans, compared to the control using South Johnstone
cocoa beans.
(d) Finally, during the fermentations using cocoa from the Mossman
plantation, there were greater variations in the bacterial growth curves,
compared to the growth curves for the same bacteria growing in cocoa
beans from the South Johnstone plantation.
229
4.3.2A Changes in the concentration of sugars and ethanol during laboratory

fermentations.
Samples taken at the beginning (Oh) and end (144h) of the laboratory-scale
fermentations were tested to determine levels of sugars and ethanol in the pulp
and seeds. The results from these tests are presented in Tables 4.4 and 4.5.
Two sugars were detected in the unfermented cocoa pulp of cocoa beans
obtained from the South Johnstone plantation: glucose (5.1%) and fructose
(5.8%). In the control (uninoculated) fermentation, traces of glucose and
fructose remained at the end of fermentation (144h). In the three fermentations
that were inoculated with defined yeast cultures, glucose and fructose had been
completely utilised by 144h (Table 4.4)
Low levels of glucose (0.37%), fructose (0.92 - 0.95%) and sucrose (1.4%) were
found in the seeds of the unfermented South Johnstone cocoa beans. In all
fermentations, whether inoculated or uninoculated, the concentration of these
sugars had decreased to undetectable levels by the end of fermentation (144h)
(Table 4.4).
Traces of ethanol were detected in both pulp (0.5 - 1.2%) and seed (0.2% - 0.8%)
of the South Johnstone cocoa beans at 144h. These data are consistent with
previous observations of fermentations, which suggest that maximum levels of
ethanol, in both seed and pulp, are reached around the middle of fermentation.
It is to be expected that most of the ethanol produced during fermentation
dissipates or is converted to acetic acid.
230
Table 4.4 - Concentrations of sugars and ethanol in pulp and seeds at beginning (Oh) and end
(144h) of laboratory fermentation of South Johnstone cocoa beans inoculated with different
mixtures of yeast species.
Yeast species Component Pulp (% w/w wet basis) Seed (% w/w wet basis)
inoculated
(Oh) (144) (Oh) (144)
H.guilliermondii Glucose 5.1 0 -5.1 0.37 0 -0.37
andl. orientalis
Fructose 5.8 0 -5.8 0.92 0 -0.92
Sucrose 0 0 0 1.4 0 -1.4
Ethanol 0 0.75 0.8 0.12 0.9 0.8
S.cerevisiae and Glucose 5 0 -5 0.37 0 -0.37

K. marxianus
Fructose 5.8 0 -5.8 0.92 0 -0.92
Sucrose 0 0 0 1.4 0 -1.4
Ethanol 0 1.2 1.2 0.12 0.84 0.72
H. guilliennondii, Glucose 5 0 -5 0.37 0 -0.37

l. oriental is,
Fructose 5.8 0 -5.8 0.95 0 -0.95
S. cerevisiae and
K. marxianus Sucrose 0 0 0 1.4 0 -1.4
Ethanol 0 0.58 0.58 0.12 0.54 0.4
Uninoculated Glucose 5.1 0.59 -4.5 0.37 0 -0.37

(control) Fructose 5.8 0.86 -4.9 0.93 0 -0.77
Sucrose 0 0 0 1.4 0 -1.4
Ethanol 0 0.51 0.51 0.1 0.3 0.2
Similar data were obtained for the concentrations of sugars and ethanol in the
cocoa beans obtained from the Mossman plantation, (c.f. Table 4.5) The initial
pulp concentrations of glucose (5.4%) and fructose (5.7%) were similar to those
observed for the South Johnstone beans. Again, by 144h these pulp sugars were
completely utilised in all of the inoculated fermentations, and again traces were
found in the control.
The levels of sugars in the seeds of the Mossman beans (glucose 1.2%,
fructose 1.7% and sucrose 1.6%) were slightly higher than those found in the
seeds of the South Johnstone beans. As before, these sugars were absent from all
seed samples taken at 144h. Finally, low levels (0.5 -1.3%) of ethanol were again
detected in the pulp and seed of the Mossman cocoa beans at the end of
fermentation (Table 4.5).
231
Table 4.5 - Concentrations of sugars and ethanol in pulp and seeds at beginning (Oh) and end
(144h) of laboratory fermentation of Mossman cocoa beans inoculated with different mixtures
of yeast species.
Yeast species Component Pulp (% w/w wet basis) Seed (% w/w wet basis)
inoculated

(Oh) (144h) (Oh) (144h)
H.guilliermondii Glucose 5.4 0 -5.4 1.2 0 -1.2

and I. orientalis
Fructose 5.7 0 -5.7 1.7 0 -1.7
Sucrose 0 0 0 1.5 0 -1.5
Ethanol 0 0.6 0.6 0 0.48 0.48
S.cerevisiae and Glucose 5.4 0 -5.4 1.2 0 -1.2

K. marxianus
Fructose 5.7 0 -5.7 1.6 0 -1.6
Sucrose 0 0 0 1.4 0 -1.4
Ethanol 0 0.6 0.6 0 0.6 0.6
H. guilliermondii, Glucose 5.5 0 -5.5 1.2 0 -1.2

1. orientalis,
S. cerevisiae and Fructose 5.7 0 -5.7 1.8 0 -1.8
K. marxianus
Sucrose 0 0 0 1.6 0 -1.6
Ethanol 0 1.3 1.3 0 0.87 0.87
Uninoculated Glucose 5.4 0.6 -4.8 1.2 0 -1.2

(control)
Fructose 5.7 0.8 -4.9 1.7 0 -1.7
Sucrose 0 0 0 1.6 0 -1.6
Ethanol 0 0.5 0.5 0 0.72 0.72
4.3.2.4 Changes in the concentration of organic acids during laboratory

fermentations.
Tables 4.6 and 4.7 summarise the changes in concentrations of organic acids in
the pulp and seed of cocoa beans during laboratory-scale fermentations.
The unfermented pulp of cocoa beans obtained from the South Johnstone
plantation contained citric acid (45 mg/g) and malic acid (15 mg/g). Unlike
previous fermentations (Sections 3.3.1.6, 3.3.2.6 and 3.3.3.6, Chapter 3) tartaric
and oxalic acids were not detected in the pulp in this case.
At the conclusion of all fermentations (144h) the concentration of both citric and
malic acids had decreased. This decrease was greatest in the control and in the
232
fermentation inoculated with the mixture of H.guilliermondii and l.orientalis.

Smaller decreases in the pulp concentrations of citric and malic acids were
observed for the fermentations inoculated with S.cerevisiae and K.marxianus, and
H.guilliermondii, l.orientalis, S.cerevisiae and K.marxianus.
Acetic acid and lactic acid were produced during fermentation, and were
detected in all pulp samples taken from the boxes at 144h. Similar levels of
lactic acid were produced in the inoculated fermentations (22 - 24 mg/g). The
level of lactic acid in the control fermentation was significantly less (16 mg/g)
(Table 4.6).
High levels of acetic acid (~50 mg/g) of acetic acid were found in the pulp of
the South Johnstone cocoa beans that were inoculated with S.cerevisiae and
K.marxianus, and H.guilliermondii, I.orie?italis, S.cerevisiae and K.marxianus. Less
acetic acid was found in the pulp of the fermentation inoculated with
H.guilliermondii and l.orientalis (24 mg/g), and even less in the control
fermentation (14 mg/g) (Table 4.6).
The seeds of unfermented cocoa beans from the South Johnstone plantation
were found to contain the following organic acids: citric (15 mg/g), malic acid
(5 mg/g) oxalic acid (10 mg/g) and traces of lactic acid (2 mg/g).
By the conclusion of all fermentations (144h), the seed concentration of citric
acid decreased slightly, while the concentrations of lactic and acetic acids had
increased. The concentration of malic and oxalic acids in the seeds did not
change significantly.
In the seeds of the fermentation inoculated with S.cerevisiae and K.marxianus, the
concentration of citric acid decreased only slightly to 14 mg/g. In all other
fermentations, it decreased to a final concentration of 10 mg/g.
The concentrations of acetic and lactic acids increased in proportion to their

increased concentrations in the pulp. The concentration of lactic acid in the
seeds of the various inoculated fermentations was similar (21 - 23 mg/g), while
the control had a much lower concentration (15 mg/g).
233
The highest concentration acetic acid was found in the seeds of the
fermentations inoculated with S.cerevisiae and K.marxianus, and H.guilliermondii,
I.orientalis, S.cerevisiae and K.marxianus (~ 45 mg/g). The final concentration of
acetic acid in the seeds of the fermentation inoculated with H.guilliermondii and
I.orientalis was about half as great (20 mg/g), with half as much again found in
the seeds of the control fermentation (12 mg/g) (Table 4.6).
Table 4.6 - Concentrations of organic acids at the beginning (Oh) and end (144h) of laboratory'
fermentation of South Johnstone cocoa beans inoculated with different mixtures of yeast
species.
Yeast species Organic Pulp (mg/g) dry basis Seed (mg/g) dry basis
inoculated acid

(Oh) (144h) (Oh) (144h)
H.guilliermondii Citric 45 22 -23 15 10 -5
and I.orientalis
Malic 15 2 -13 5 5 0
Oxalic 0 0 0 10 11 1
Lactic 0 24 24 0 21 21
Acetic 0 24 24 0 20 20
S. cerevisiae and Citric 45 28 -17 15 14 -1

K. marxianus
Malic 15 10 -5 5 5 0
Oxalic 0 0 0 10 8 -2
Lactic 0 22 22 0 23 21
Acetic 0 48 48 0 46 46
H. guillierm ondii, Citric 45 25 -20 15 10 -5
I.orientalis,
S.cerevisiae and
Malic 15 8 -7 5 5 0
K. marxianus
Oxalic 0 0 0 10 10 0
Lactic 0 24 24 0 23 21
Acetic 0 50 50 0 45 45
Uninoculated Citric 45 24 -21 15 10 -5
(control)
Malic 15 5 -10 5 5 0
Oxalic 0 0 0 10 10 0
Lactic 0 16 16 0 15 15
Acetic 0 14 14 0 12 12
234
When the laboratory-scale fermentations were repeated using cocoa from the
Mossman plantation, similar data were obtained. The initial concentration of
citric acid (48 mg/g) and malic acid (20 mg/g) in the pulp of the Mossman
beans were very similar to the concentrations found in the South Johnstone
cocoa (c.f. Table 4.6 and 4.7).
The concentration of malic and citric acids in the pulp decreased, from the
beginning (Oh) to the end (144h) of the fermentations. As before, larger
decreases in the concentrations of these acids were observed in pulp of the
control fermentation and the fermentation inoculated with H.guilliermondii and
I.orientalis (Table 4.7).
As before, lactic and acetic acids were produced in the pulp during
fermentation, with the amounts produced varying depending on the inoculum
used. As in the fermentations using South Johnstone cocoa, the final
concentration of lactic acid was lower in the control (12 mg/g), compared to the
inoculated fermentations (18 - 20 mg/g).
As before, the highest concentrations of acetic acid were detected in the pulp of
the fermentations inoculated with S.cerevisiae and K.marxianus (50 mg/g), and
H. guilliermondii, I.orientalis, S.cerevisiae and K.marxianus (57 mg/g). The
concentration of acetic acid in the pulp of the control fermentation, and the
fermentation inoculated with H.guilliermondii and I.orientalis was about half as
great (24 mg/g) again (Table 4.7).
In the seeds of Mossman cocoa, the changes in concentration of organic acids
were similar to those observed for the South Johnstone cocoa. The concentration
of citric (10 mg/g), malic (6 mg/g) and oxalic (10 mg/g) acids remained
relatively constant from beginning to end of the fermentations.
Lower concentrations of lactic acid were detected in seeds taken from the
control (13 mg/g), compared to seeds from the inoculated fermentations
(19-20 mg/g).
Finally, more acetic acid was detected in the seeds of the fermentations
inoculated with S.cerevisiae and K.marxianus (45 mg/g), and H.guilliermondii,
I. orientalis, S.cerevisiae and K.marxianus (42 mg/g). Less acetic acid was detected
in the seeds of the control fermentation (20 mg/g) and the fermentation
inoculated with H.guilliermondii and I.orientalis (18 mg/g) (Table 4.7).
235
Table 4.7 - Concentrations of organic acids at the beginning (Oh) and end (144h) of laboratory
fermentation of Mossman cocoa beans inoculated with different mixtures of yeast species.
Yeast species Organic Pulp (mg/g) dry basis Seed (mg/g) dry basis
inoculated acid
(Oh) (144h) (Oh) (144h)
H.guilliennondii Citric 48 19 -29 10 9.8 -0.2
and I. orientalis
Malic 20 8 -12 6 5 -1
Oxalic 0 0 0 10 10 0
Lactic 0 18 18 0 20 20
Acetic 0 24 24 0 18 18
S.cerevisiae and Citric 48 34 -14 10 10 0

K. marxianus
Malic 20 15 -5 6 6 0
Oxalic 0 0 0 10 11 1
Lactic 0 18 18 0 19 19
Acetic 0 50 50 0 45 45
H.guil/iermondii, Citric 48 36 -12 10 9.9 -0.1

I.orientalis,
S.cerevisiae and Malic 20 14 -6 6 5.8 -0.2
K. marxianus
Oxalic 0 0 0 10 12 2
Lactic 0 20 20 0 19 19
Acetic 0 57 57 0 42 42
Uninoculated Citric 48 30 -18 10 10 0

(control)
Malic 20 10 -10 6 6 0
Oxalic 0 0 0 10 10 0
Lactic 0 12 12 0 13 13
Acetic 0 24 24 0 20 20
236
4.3.2A Quality evaluation of cocoa beans produced by laboratory

fermentations
The cocoa beans produced during laboratory fermentations were subject to a

range of quality tests, including the cut test to determine the fermentation
index, and measurement of bean size, moisture content, shell content and the
pH of the dried nibs. A basic assessment was also made of the appearance and
aroma of the beans. The cut test results for these cocoa bean samples are
presented in Figure 4.9.
Hg and Io Sc and Kmx All yeast Control Hg and lo Sc and Kmx All yeast Control
Fig. 4.9. Cut test results for cocoa beans produced by laboratory-scale fermentation using
mixed yeast cultures: (a) beans from South Johnstone plantation (b) beans from Mossman
plantation; ■ % beans fully brown, 11 % beans part brown/purple. All beans were solar dried.
“Hg and Io” = H.guilliermondii and l.orientalis; “Sc and Kmx” = S.cerevisiae and K.marxianus\
“All yeast” = {H.guilliermondii, l.orientalis, S.cerevisiae and K.marxianus).
The cut tests revealed little differences between the fermentation indices of the
cocoa bean samples, regardless of the yeast cultures used in the fermentation.
All samples had very high fermentation indices (proportion of fully brown
beans > 90%). Beans fermented with all four yeast species gave a slightly lower
cut test score (94% fully brown) compared to the other samples (99% fully
brown). T-test analysis (95% confidence) of the cut test data suggested that this
difference was not significant.
Additional quality parameters for the dried, fermented beans are collated in
Tables 4.8 and 4.9. All of the fermentations produced beans that were of
acceptable appearance, but had weak cocoa aroma in spite of the high cut test
scores. The shell content of all beans produced during laboratory fermentation
were much higher than the recommended maximum of 13%. The reasons for
this were not clear.
237
Table 4.8 - Quality evaluation of South Johnstone cocoa beans produced during laboratory-
scale fermentation using mixed yeast cultures.
H.guilliermondii S.cerevisiae and

Mixture of yeast species used All yeast* Control Mean St. Dev.
and /. orientalis K.marxianus
Quality parameter
% Bean moisture 6.27 6.3 6.32 6.26 6.29 0.00
% Nib moisture 6.2 6.2 6.2 6.2 6.20 0.00
% Shell content 16 13 14 15.5 14.63 1.38
Nib pH 5.2 4.78 4.79 5.3 5.02 0.27
External Appearance Acceptable Acceptable Acceptable Acceptable ... ...
weak cocoa weak cocoa weak cocoa weak cocoa

Aroma evaluation ... ...
aroma aroma aroma aroma
* This mixture contained the yeast H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus
All cocoa beans from this experiment were dried using solar-drying.
Table 4.9 - Quality evaluation of Mossman cocoa beans produced during laboratory-scale
fermentation using mixed yeast cultures.
H.guilliermondii S.cerevisiae and

Mixture of yeast species used All yeast * Control Mean St. Dev.
and I.orientalis K.marxianus
Quality parameter
Bean size - no. per lOOg 88 86 86 87 86 JS 0.96
% Bean moisture 7.38 7.4 6.4 6 6.80 0.71
% Nib moisture 6.7 6.68 5.7 5.3 6.10 0.71
% Shell content 16 16 16 16.5 16.1 0.24
Nib pH 5.26 4.89 4.85 5.35 5.09 0.25
External Appearance Acceptable Acceptable Acceptable Acceptable ... ...
weak cocoa weak cocoa weak cocoa weak cocoa

Aroma evaluation ... ...
aroma aroma aroma aroma
* This mixture contained H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus

All cocoa beans from this experiment were dried using solar-drying.
The addition of cultures did not cause any significant differences to the bean
size, bean and nib moisture contents, shell contents, appearance or aroma scores
of the beans produced. Significant differences were observed in the nib pH,
depending on the inoculum used. The beans fermented with mixtures of
S.cerevisiae and K.marxianus, and H.guilliermondii, I.orientalis, S.cerevisiae and
238
K.marxianus had pH values significantly lower than the beans fermented with
H.guillermondii and I.orientalis or without any inoculum (control).
When the quality data were analysed on the basis of cocoa bean source, it was
found that cocoa beans from the Mossman plantation were significantly larger,
and had significantly higher nib pH compared to those from South Johnstone
plantation.
All significant differences were verified using the Student's T-test at 95%
confidence interval (a = 0.05).
4.3.2.5 Sensory evaluation of chocolate prepared from laboratory
fermented cocoa beans
Chocolate samples made with the cocoa beans from the laboratory
fermentations were subjected to difference and liking sensory evaluation, as
described in Section 4.2.5.3 of methods.
The triangle test was performed to determine whether chocolate made using
different samples of beans tasted significantly different from one another. The
results of the difference (triangle) tests are presented in Table 4.10, while the
results of the liking tests are presented in Tables 4.11 and 4.12.
Table 4.10 - Sensory evaluation (triangle test) of chocolate samples made from cocoa beans
produced during laboratory-scale fermentations using mixed yeast cultures.

Number of
Sample correct needed for significance*
tests
Description of defined mixed cultures
Control Vs. H.guilliermondii and I.orientalis 10 6 7
Control Vs. S.cerevisiae and K.marxianus 10 7 7

Control Vs. H.guilliermondii, I.orientalis, 10 10 7
S.cerevisiae and K.marxianus
H.guilliermondii and I.orientalis Vs.
10 10 7
* Adapted from a table in Stone and Sidel (2004). For a pair of samples to taste significantly different
from one another,“number of correct judgments” > “number of judgements needed for significance.”
Chocolate made from the control beans was found to taste significantly
different from chocolate made from beans fermented with a mixture of
S.cerevisiae and K. marxianus, or mixtures of H.guilliermondii, I.orientalis,
S.cerevisiae and K.marxianus (Table 4.10).
239
No significant difference was found in the flavour of chocolate prepared from

control beans compared to chocolate made using beans fermented with cultures
of H.girilliermondii and I.orientalis. Fermentation with H.guilliermondii and
l.orientalis resulted in cocoa beans with a flavour that was not significantly
different from the flavour of cocoa beans fermented by indigenous microflora
(control). Finally, the chocolate prepared from beans fermented with
H.guilliermondii and I.orientalis tasted significantly different from chocolate
prepared from beans fermented with S.cerevisiae and K.marxianus (Table 4.10).
Table 4.11 - Mean liking scores of chocolate samples made from cocoa beans produced during
laboratory-scale fermentations using mixed yeast cultures.
Mean liking score*

Sample Standard deviation
(out of 10)
Ghana 6.92 2.10
Beans fermented with a mixture of

H.guilliermondii and I.orientalis
6.74 1.50

4.94 2.42

H.guilliermondii, /. orientalis, 4.87 2.98
Beans fermented without added
starter cultures (control) 6.37 1.76
* A liking score of 0 indicated that a chocolate sample was “strongly disliked.” while a score of 10 indicated it was
“strongly liked.”
The mean liking scores presented in Table 4.11 show that chocolate made from
Ghanaian cocoa beans had the highest mean liking score. Chocolates from
"Control" and "Hg & Io" fermented beans had similar mean liking scores
compared to the "Ghana" sample. By contrast, chocolate made from beans
fermented with mixtures of S.cerevisiae and K.marxianus , or H.guilliermondii,
I.orientalis, S.cerevisiae and K.marxianus, had lower mean liking scores compared
to chocolate prepared from the Ghanaian beans.
All samples received mean liking scores of around 5-7. This indicated that all of
the samples tested were neither strongly liked, nor strongly disliked by any of
the panelists surveyed.
T-test analysis of the liking data was performed to determine whether these
observations were significant. The results of these statistics are presented in
Table 4.12
240
Table 4.12 - T-test analysis of liking scores for chocolate made from cocoa beans produced
during laboratory-scale fermentations using mixed yeast cultures. The chocolate made from
Australian cocoa was compared to chocolate made from Ghanaian cocoa.
Sample Pair T-test result at 95% confidence level: T > t

T t*
Cocoa beans description (Direction of difference for one tailed test)
Ghana, Vs. 0.560 1.80 Not significantly different

H.guilliermondii and I.orientalis
Ghana, Vs. Significantly different

2.52 1.80
S. cerevisiae and K. marxianus (S.cerevisiae and K.marxianus < Ghana )
Ghana, Vs. H.guilliermondii, Significantly different

I.orientalis, S.cerevisiae and 2.56 1.80 (H.guilliermondii, I.orientalis , S.cerevisiae and
K. marxianus K.marxianus < Ghana)
Ghana, Vs. 1.11 1.80 Not significantly different

Control (uninoculated)
H.guilliermondii and I.orientalis Significantly different
Vs. 2.52 1.80 (S.cerevisiae and K.marxianus <
S.cerevisiae and K.marxianus H.guilliermondii and I.orientalis )
The T-test analyses confirmed that chocolates prepared from beans fermented
with cultures containing S.cerevisiae and K.marxianus received mean liking
scores significantly lower than those of chocolate produced from the Ghana
beans. It can be inferred, therefore, that these samples had a less preferred
flavour compared to Ghanaian cocoa (Table 4.12).
The T-test results also showed that the mean liking scores of chocolates
prepared from uninoculated beans and beans fermented with a mixture of
H.guiltiermondii and I.orientalis were not significantly different from the
chocolate made using Ghanaian beans. Therefore, these samples were judged to
have a flavour that was basically acceptable, and compared to the flavour of
Ghanaian cocoa beans (Table 4.12).
T-tests also showed that chocolate prepared from beans inoculated and
fermented with S.cerevisiae and K.marxianus had a significantly lower mean
liking score compared to the chocolate made from beans inoculated and
fermented with H.guilliermondii and I.orientalis. This indicates that, on average,
the panelists liked the samples fermented with H.guilliermondii and I.orientalis
more than the samples fermented with S.cerevisiae and K.marxianus (Table 4.12).
241
4.3.3 Pilot-scale fermentation of cocoa beans in box and barrel

vessels using defined mixtures of yeast cultures.
Experiments with different fermentation vessels demonstrated that is possible
to successfully ferment Australian cocoa beans (Section 3.3.1, Chapter 3).
Having selected three mixtures of yeast, and demonstrated their ability to
ferment cocoa beans in laboratory scale experiments, large pilot-scale
fermentations were now investigated. This section describes the pilot-scale box
and barrel fermentations with the following mixtures of yeast: H.guilliermondii
and I.orientalis; S.cerevisiae and K.marxianus; and H.guilliermondii, I.orientalis,
S.cerevisiae and K.marxianus.
Temperature, pH and microbial ecology were monitored during the
fermentations. Chemical analyses were performed on samples taken at the
beginning, middle and end of fermentation. The fermented beans were dried
and assessed for quality, according to industry criteria. Finally, the dried
fermented beans were made into chocolate and subject to sensory analysis to
determine whether the added cultures affected the flavour of chocolate
prepared from the cocoa beans. Beans from these pilot-scale fermentations were
further tested for their antioxidant activity, the breakdown of proteins during
fermentation, and their profile of volatile aroma compounds.
4.3.3.1 (A) Box and barrel fermentations using mixtures of Hanseniaspora
guilliermondii and Issatchenkia orientalis
Figure 4.10 shows the changes in pH (pulp and seed), temperature, and total
populations of yeasts and bacteria that occurred during the pilot scale box and
barrel fermentations with the mixture of H.guilliermondii and I.orientalis. The
inoculated box fermentation was terminated after 96 h to prevent spoilage of
the beans, the onset of which was indicated by blackening of beans at the
corners of the fermentation, and the presence of a faint ammonia aroma.
At Oh, the pH of the fresh pulp was 3.75, while the pH of the fresh seed was
6.25. As the fermentations proceeded, the pH of the seed decreased and the pH
of the pulp increased. In the box inoculated with H.guilliermondii and I.orientalis,
the final pH of the pulp was much higher (4.64) than that of the control (3.63).
The final pH of the seed from the inoculated box(6.08) was close to the control
242
(5.98) (Figure 4.10 a, b). Similarly, in the barrel inoculated with H.guilliermondii
and I.orientalis, the final pH of the pulp (4.52) was higher than the control (3.71).
The pH of the seed from the inoculated fermentations (6.00) was slightly higher
than the control (5.76) (Figure 4.10 c, d). The different fermentation vessels (box
and barrel) did not greatly affect the pH changes in pulp and seed (compare
Figures 4.10 a & c; b & d).
The temperature of the beans changed in a similar manner in all fermentations
except for the inoculated box. In both barrel fermentations and the uninoculated
box fermentation the temperature was initially at 23°C. The temperature at the
centre of these fermentations progressively increased to a maximum of 45°C,
between 72-84 h. Between 96 -120 h, it decreased to 40°C (Figure 4.10 a, c &d).
In the inoculated box fermentation, the temperature gradually increased from
23°C at Oh to a maximum of 42.5°C at 72h, and then slowly decreased to 40°C at
96h (Figure 4.10 b).
The overall microbiology of the fermentations was affected by the addition of
cultures of H.guilliermondii and I.orientalis.
In the control (uninoculated) box and barrel fermentations, the initial
populations of both yeast and bacterial species were 104 cfu/g beans. In each of
the control fermentations, the populations of yeast increased to maximum
populations of 107 - 108 cfu/ g (72-96h), and remained at this level until
completion of fermentation. In the controls, the bacterial species reached
maximum populations of 107 cfu/g beans after 48h. They remained between
106 - 107 cfu/g beans until the end of fermentation (Figure 4.10 a, c).
In the inoculated fermentations, the initial populations of yeasts were 106 cfu/g
beans, while the initial population of bacteria was around 104 cfu/g beans.
In the inoculated box and barrel, the yeasts grew to maximum populations of
108-109 by 24h and then decreased slightly to final populations of 1x10s cfu/g in
the box and 5xl07 cfu/g in the barrel (Figure 4.10 b, d).
The total bacterial population in the inoculated box fermentation behaved quite
differently from that in the inoculated barrel. In the box, the bacteria remained
around 104 cfu/g beans until 72 h, and then increased to 106 cfu/g beans at 96h.
By contrast, the inoculated barrel fermentation reached maximum populations
of bacteria at 72h (107 cfu/g beans) before declining slight to about 106 cfu/g
beans at the end of fermentation (compare Figure 4.10 b, d).
243
f- 25 -
35 -
H 25 -
<u 6 -
H 25 -
H 25 -
* This fermentation was terminated at 96h, as described in the methods section 4.2.1.2
Fig. 4.10. Fermentation of cocoa beans using defined mixtures of yeast. Cocoa was fermented
in: (a) Box, no added starter culture, (b) box, H. guilliermondii & I. orientalis added, (c) barrel,
no added starter culture, and (d) barrel, H. guilliermondii & I. orientalis added. Total yeast
population ♦; total bacterial population □; temperature ■; pulp pH O; seed pH A.
244
4.3.3.1 (B) Box and barrel fermentations using mixtures of Saccharomyces

cerevisiae and Kluyveromyces marxianus
populations of yeast and bacteria that occurred during box and barrel
fermentations inoculated with the mixture of S.cerevisiae and K.marxianus.
The temperature of the beans increased from 25°C to maximum temperatures at
around 72h. Lower maximum temperatures were recorded in the inoculated
fermentations.
In the control (uninoculated) box fermentation, the temperature steadily
increased to a maximum of 45°C at 72h, fell to 40°C at 84h and decreased again
during the last 12h to 33°C (Figure 4.11 a). In the inoculated box fermentation
the maximum temperature reached was 40°C, after which it decreased to a final
temperature of 36°C(Figure 4.11 b). In both the control and inoculated barrel
fermentations, the temperatures increased to maxima of 45°C and 43°C
respectively and finally decreased to 40°C at 120 h (Figure 4.11 c, d).
In all fermentations, the pH of the seeds decreased, while the pH of the pulp
increased. There was, however, considerable variation in the pH changes that
occurred in the different fermentations. In all fermentations, the pH of the pulp
and seed were initially 3.75 and 6.15, respectively.
In the uninoculated box fermentation, the pH of the pulp increased to a final
value of 4.83, while the pH of the seed decrease to a final value of 5.10 (Figure
4.11 a). The final pH of the pulp and seed from the inoculated box fermentation
were 4.6 and 5.8, respectively (Figure 4.11 b).
The final pH of the pulp from the uninoculated barrel was 4.56, and was very
similar to the final pH of the pulp from the inoculated barrel (4.58). The final pH
of the seeds from the uninoculated barrel (5.66) was much higher than the final
pH of the seeds from the inoculated barrel (5.13) (Figure 4.11 c & d)
The growth of yeast and bacteria followed similar trends in all fermentations,
with only minor variations. The uninoculated box and barrel fermentations had
lower initial populations of yeast (104 and 105 cfu/g beans, respectively),
compared to the inoculated fermentations (both 106 cfu/g beans). In all of the
fermentations, the populations of yeast species increased to maxima of
107 cfu/g by 24 h, and remained at this level until the end of fermentation at
120h (Figure 4.11)
245
The initial population of bacteria was 1 x 105 cfu/ g beans in all fermentations,
except for the barrel inoculated with S.cerevisiae and K.tnarxicmus, where the
initial population was 2 x 104 cfu/g beans. Between 0 - 24 h, the total population
of bacteria actually decreased by a factor of 10, in all fermentations, after which
it slowly increased to 107 cfu/g beans at 120h (Figure 4.11 a, b, c & d). A second,
temporary decrease in the total population of bacteria was observed during the
uninoculated box fermentation at 72 h(Figure 4.11 a).
246
24 48 72 96 120
f- 25
A 35 -
H 25
cH 4 k
in: (a) box, no added starter culture, (b) box, S.cerevisiae & K.marxianus added, (c) barrel, no
added starter culture, and (d) barrel, S.cerevisiae & K.marxianus added. Total yeast population
♦ ; total bacterial population □; temperature ■; pulp pH O; seed pH A.
247
4.3.3.1 (C) Barrel fermentations using a mixture of Hanseniaspora guilliermondii,

Issatchenkia orientalis, Saccharomyces cerevisiae and Kluyveromyces
marxianus
populations of yeast and bacteria that occurred during barrel fermentations
inoculated with the mixture of H.guilliermondii, l.orientalis, S.cerevisiae and
K.marxianus. Due to limited supply of beans, parallel box fermentations were
not performed.
The temperature of the beans at Oh was 25°C. In the control barrel, the
temperature increased to a maximum of 45°C at 72h after which it decreased to
40°C, where it remained until the end of fermentation (Figure 4.12 a). For the
fermentation inoculated with H.guilliermondii, l.orientalis, S.cerevisiae and
K.marxianus, the maximum temperature reached was 42°C, and in the final 24 h
the temperature dropped to 35°C (Figure 4.12 b).
The pH of the fresh cocoa pulp was 3.76, while that of fresh seeds was 6.12.
The pH of the seed decreased as the pH of pulp increased during fermentation.
In the uninoculated barrel the final pH of the pulp was 4.56, significantly lower
than the final pH of the pulp in the inoculated barrel (5.00). The final pH of the
seeds from the control barrel (5.66) was also lower than that of the seeds from
the inoculated barrel (5.81) (compare Figure 4.12 a & b).
Changes in the total populations of yeast and bacteria were very similar in both
barrel fermentations. As expected, lower initial populations of yeast (105 cfu/ g
beans) were present in the control barrel, compared to the inoculated barrel (106
cfu/g beans). In both fermentations, the population of yeasts increased to
maxima of 107 cfu/g, where they remained until the end of fermentation at 120h
(Figure 4.12)
Initially, bacterial populations were 105 cfu/g beans. The population of bacteria
in both fermentations decreased slightly during the first 24h, before slowly
increasing to 107 cfu/g beans at 120h (Figure 4.12 a & b).
Differences in the microbiology of these fermentations become more apparent
when examined at the species level.
248
cH 4 -
35 -
H 25
in: (a) barrel, no added starter culture, and (b) barrel, H.guilliennondii, I.orientalis, S.cerevisiae
& K.marxianus added. Total yeast population ♦; total bacterial population □; temperature ■;
pulp pH O; seed pH A.
249
4.3.3.2 Changes in the population of yeast species during fermentation using

defined mixtures of yeast cultures.

guilliermondii and Issatchenkia orientalis.
Figure 4.13 shows the growth of individual yeast species during the
fermentations inoculated with a mixture of Hcuiseniaspora guilliermondii and
Issatchenkia orientalis. The three species of yeast consistently isolated from these
fermentations were Hanseniaspora guilliermondii, Issatchenkia orientalis and
Saccharomyces cerevisiae.
The initial population of H.guilliermondii was around 104 cfu/ g beans in the
control fermentations, and 106 in the inoculated fermentations. During all of the
fermentations, H.guilliermondii grew, reaching maximum populations after 24h.
In the inoculated fermentations, H.guilliermondii reached higher maximum
populations (108 cfu/g beans), and persisted 24h longer (until 96h), compared
to the control fermentations (compare Figures 4.13 a & c with b & d).
The growth of I.orientalis also followed a repeated pattern in the control and
inoculated fermentations. In the uninoculated (control) box and barrel
fermentations, I.orientalis slowly increased from initial populations of 102 cfu/g
beans to maximum levels of 107 cfu/g beans. The populations of
H. guilliermondii remained steady at around 107 cfu/g until the end of
fermentation (Figure 4.13 a, c). In the inoculated box and barrel fermentations,
I. orientalis grew from initial populations of 106 cfu/g beans to maximum levels
of 108 cfu/g at 24h, and then remained between 107-108 cfu/g until the end of
fermentation (Figure 4.13 b, d).
In the control fermentations, S.cerevisiae steadily increased in population, from
an initial level of 102 cfu/g beans. It reached a maximum population of
107 cfu/g beans (96h) in the uninoculated box and 5 x 106 cfu/g (120h) beans in
the uninoculated barrel (Figure 4.13 a, c). In contrast to the control
fermentations, S.cerevisiae did not grow well in the inoculated fermentations. In
the inoculated box fermentation, it was not detected until 72h (1 x 103 cfu/g
beans), after which it increased to a maximum population of 2 x 107 cfu/g beans
at 96h (Figure 4.13 b). In the inoculated barrel fermentation, S.cerevisiae was not
detected until 24h (5 x 102 cfu/g beans), after which it grew to a maximum
250
population of 5 x 104 cfu/ g beans at 48h, and then died off over the next 24h
(Figure 4.13 d).
In summary, these yeast species grew differently, depending on whether the
fermentations were uninoculated or inoculated. The figures also show that
I.orientalis and S.cerevisiae grew more rapidly in the box fermentations
compared to the barrel fermentations.
cH 4 -
* This fermentation was terminated at 96h. as described in the methods section 4.2.1.2
Fig. 4.13. Growth of yeast species during fermentation of cocoa beans using defined mixtures
of yeast. Cocoa beans were fermented in: (a) box, no added starter culture,
(b) box, H. guilliermondii & I. orientalis added, (c) barrel, no added starter culture, and (d)
barrel, H. guilliermondii & /. orientalis added. Hanseniaspora guilliermondii ♦; Issatchenkia
orientalis O; Saccharomyces cerevisiae □.
251

Figure 4.14 shows the changes in population of the four species of yeast that
were most commonly isolated from these fermentations. These species were
Hanseniaspora guilliermondii, Issatchenkia orientalis, Saccharomyces cerevisiae and
Kluyveromyces marxianus.
In all fermentations, H.guilliermondii was initially isolated at populations of 104
cfu/ g beans. It grew to maximum populations by 24h and then died off.
The maximum populations of H.guilliermondii were slightly higher in the
control box (1.5xl07 cfu/g) and barrel (2xl07 cfu/g) compared to the inoculated
box (2.5xl06 cfu/g) and barrel (5xl06 cfu/g) (compare Figure 4.14 a & c, b & d).
I.orientalis was initially detected in the control and inoculated box fermentations
at populations of 103 cfu/g. In the control box it grew to a maximum population
of 107 cfu/g at 48h, and remained between 106-107 cfu/g until the end of
fermentation(120h)(Figure 4.14 a) In the box inoculated with S.cerevisiae and
K.marxianus, the population of I.orientalis increased throughout fermentation
and reached a maximum population of 3 x 107cfu/ g at 120 h (Figure 4.14 b).
In the control and inoculated barrels, I.orientalis was not detected until 24h
(103cfu/g), after which it increased to 107 cfu/g, and remained at this level until
completion of fermentation(120h) (Figure 4.14 c, d).
The initial population of S.cerevisiae was much lower in the control
fermentations (103 cfu/g), compared to those inoculated with a mixture of
S.cerevisiae and K.marxianus (106 cfu/g). S.cerevisiae was not detected in the
uninoculated barrel until 24 h of fermentation. In the control (uninoculated) box
and barrel fermentations, S.cerevisiae gradually increased to maximum
populations at the end of fermentation (108 and 105 cfu/g respectively) (Figure
4.14 a, c). In the fermentations inoculated with mixtures of S.cerevisiae and
K.marxianus, the populations of S.cerevisiae reached maxima of 107 cfu/g at 24h,
and then remained at this level until the end of fermentation (Figure 4.14 b, d).
K.marxianus was only detected in the inoculated fermentations, where it was
found at initial populations of 105 cfu/g beans. In these fermentations it then
increased to maxima of 106 - 107 cfu/g beans by 24h. In the inoculated box,
K.marxianus died off after 96h, while in the inoculated barrel it remained steady
at 106 cfu/g beans until the end of fermentation (Figure 4.14 b, d).
252
a 4 -
of yeast. Cocoa beans were fermented in: (a) box, no added starter culture, (b) box, S.cerevisiae
& K.marxianus added, (c) barrel, no added starter culture, and (d) barrel, S.cerevisiae &
K.marxianus added. Hanseniaspora guilliermondii ♦; Issatchenkia orientalis O;
Saccharomyces cerevisiae □; Kluyveromyces marxianus A.

marxianus
Hanseniaspora guilliermondii, Issatchenkia orientalis, Saccharomyces cerevisiae were

found in both fermentations, while Kluyveromyces marxianus was only isolated
from the inoculated barrel fermentation. Figure 4.15 shows the changes in
populations of these yeast during the course of the fermentations.
In the control (uninoculated) barrel, H,guilliermondii grew from an initial

population of 3xl04 cfu/ g to a maximum population of 6xl06 at 24h, and then
declined. It was not found in the control after 72h (Figure 4.15 a).
H.guilliermondii was initially detected at 8xl05 cfu/g in the inoculated barrel. By
24h it had increased to a population of lxlO7 cfu/g. Between 24-72 h the
population of H.guilliermondii in the inoculated barrel had decreased to ~105
253
cfu / g, and remained at this level until the end of fermentation. This was the
only Australian fermentation recorded in which H.guiltiermondii was detected at
the end of fermentation (Figure 4.15 b)
Issatchenkia orientalis was first detected in the control at 24h (103 cfu/g). It grew
rapidly in the first 48h to a population of 107 cfu/g, after which it continued to
increase more slowly (Figure 4.15 a). In the inoculated barrel, I.orientalis was
initially detected at 0 h, at much higher populations (7xl05 cfu/g) than the
control. After 24 h it had grown to a population of 107 cfu/g where it remained
until 72h, after which it increased again to a final population of 108 cfu/g at
120h (Figure 4.15 b).
Saccharomyces cerevisiae was not detected in the control fermentations until 24h
(5xl03 cfu/g). It grew slowly, reaching a final population of 105 cfu/g at 120h
(Figure 4.15 a). In the inoculated barrel, S.cerevisiae grew from an initial
population of 8x 105 cfu/g up to 107 cfu/g at 24h, and then remained between
106-107 cfu/g until the end of fermentation (Figure 4.15 b).
Kluyveromyces marxianus was not detected in the uninoculated barrel
fermentation. In the inoculated barrel fermentation, it was initially detected at
2xl05 cfu / g and grew during the first 24 h of fermentation to a maximum
population of 7xl06 cfu/g. Finally, it died off slowly and was not detected at the
end of fermentation (Figure 4.15 b).
of yeast. Cocoa beans were fermented in: (a) barrel, no added starter culture, and
(b) box, H.guiltiermondii, I.orientalis, S.cerevisiae & K. marxianus added; Hanseniaspora
guilliermondii ♦; Issatchenkia orientalis O; Saccharomyces cerevisiae □; Kluyveromyces
marxianus A.
254
4.3.3.3 Changes in the population of bacterial species during fermentation

using defined mixtures of yeast cultures.

Acetobacter pasteurianus, Gluconobacter oxydans, Pediococcus acidilactici,

Lactobacillus plantarum and Lactobacillus fermentum were the predominant
bacterial species isolated during this experiment. Bacillus spp. were also isolated
from the inoculated box fermentation. Figure 4.16 describes the changes in
population of these species over the course of the fermentations.
Fig. 4.16. Growth of bacterial species during fermentation of cocoa beans using defined
mixtures of yeast. Cocoa beans were fennented in: (a) box, no added starter culture,
(b) box, H. guilliermondii & 1. orientalis added, (c) barrel, no added starter culture, and
(d) barrel, H. guilliermondii & I. orientalis added. Acetobacter pasteurianus ♦;
Gluconobacter oxydans □; Pediococcus acidilactici A; Lactobacillus plantarum O;
Lactobacillus fermentum ■; Bacillus spp. #.
A.pasteurianus demonstrated different growth patterns in each fermentation:

In the uninoculated box, A.pasteurianus was initially present at 1 x 103 cfu/ g
255
beans and increased to 4 x 106 cfu / g beans at 48h. At 96h it decreased in

population to 105 cfu/g beans, before again increasing to a final population of
107 cfu/g beans (120h) (Figure 4.16 a). In the inoculated box, A.pasteurianus was
initially present at 104 cfu/g beans. It remained at a population of between 103 -
104 until 72 h, and then increased to final a population of 106 cfu/g in the last
24h of fermentation (Figure 4.16 b).
A.pasteurianus was initially detected at at 8 x 103 cfu/g beans in the
uninoculated barrel. By 24 h it had grown to a maximum population of 7 x 106
cfu/g beans, before slowly decreasing to a final population of 105 cfu/g beans
(120h) (Figure 4.16 c). Finally, in the inoculated barrel A.pasteurianus was
initially detected at 103 cfu/g beans, after which it increased to maximum
population of 3 x 107 cfu/g beans and then declined to a final population of 5 x
103 cfu/g beans (Figure 4.16 d).
G.oxydans was isolated from all fermentations except for the inoculated box. The
growth of this species was similar in all three fermentations: It was initially
detected at 103 cfu/g beans, increased to maximum populations of between 106
-107 cfu/g beans at 120h. G.oxydans was detected 24h earlier in the control box
(48h) than in the barrels (72h) (compare Figure 4.16 a with c & d).
P.acidilactici was only detected in the uninoculated (control) box and barrel
fermentations, growing quite differently in each. In the control box P.acidilactici
was initially at found at 72h (5 x 103 cfu/g beans) and grew to 105 cfu/g at 96h.
After this the population fell to 104 cfu/g beans, then remained at this level
until the end of fermentation (Figure 4.16 a). In the control barrel, P.acidilactici
was initially detected at 96 h (103 cfu/g), and then increased to a maximum of 2
x 106 cfu/g at the end of fermentation (120h)(compare Figure 4.16 a, b).
L.plantarum was only detected in the inoculated barrel. It was initially found at
72 h (103 cfu/g), grew to a maximum population at 96h (2 x 105 cfu/g), and then
died off (Figure 4.16 d).
L.fermentum was only isolated from the uninoculated (control) box and barrel
fermentations. It increased from initial populations of 103 - 104 cfu/g, to
maximum populations of approximately 106 at 48h, and then died off (Figure
4.16 a, c).
Finally, Bacillus spp. were isolated only from the inoculated box. The species
isolated were found to be a mixture of B.subtilis, B.licheniformis and
256
B.megaterium. These species were initially detected at 8 x 102 cfu/g beans (72h).
Between 72 - 96 h the population of Bacillus spp. increased to a final levels of
5 x 103 cfu/ g (96h)(Figure 4.16 b).
Figure 4.17 shows the changes in population of the six most commonly isolated
species of bacteria over the course of the fermentations. The five species isolated
during the fermentations were: Acetobacter pasteurianus, Acetobacter tropicalis,
Asaia siamensis, Lactobacillus plantarum and Lactobacillus fermentum.

mixtures of yeast. Cocoa beans were fennented in: (a) box, no added starter culture,
(b) box, S.cerevisiae & K.marxianus added, (c) barrel, no added starter culture, and
(d) barrel, S.cerevisiae & K.marxianus added. Acetobacter pasteurianus ♦; Acetobacter
tropicalis □; Asaia siamensis A; Lactobacillus plantarum O; Lactobacillus fermentum ■.
A.pasteurianus was first detected at 24 h in the inoculated fermentations, and at

48 h in the uninoculated (control) fermentations. In each case the initial
257
populations were ~103 cfu/ g beans. In all fermentations, A.pasteurianus grew to

maximum populations of 107cfu/g beans. In the control box (Figure 4.16 a), and
in the inoculated barrel (Figure 4.17 d), maximum populations were at 120 h.
In the inoculated box (Figure 4.17 b) and the control barrel (Figure 4.17 c),
maximum populations were reached at 96 h, after which the population of
A.pasteurianus decreased slightly.
A. tropicalis was isolated for the first time during this set of fermentations. It was
found at initial populations of 105 cfu/g beans in all fermentations except the
inoculated barrel, where it was initially found at 2 x 104 cfu/g beans. The
populations of A.tropicalis in the first 24 h and then decreased again. Finally,
after 48 h, the populations of A. tropicalis increased again and reached maxima
at 120h in all fermentations. In the control box (Figure 4.16 a) this maximum
was 108 cfu/g beans, while in all other fermentations it was 107 cfu/g beans
(Figure 4.16 b, c & d).
The acetic acid bacteria Asaia siamensis was only found in the inoculated barrel
fermentation. It was first detected at 48 h, at a population of 103 cfu/g beans.,
and increased to a maximum of 106cfu/g beans at 96 h. Between 96-120 h, the
population of As.siamenesis decreased to 105 cfu/g beans (Figure 4.17 d).
L.plantarum was isolated from all fermentations except for the control box.
In the inoculated box (Figure 4.17 b) L.plantarum was initially detected at 96h
(5xl03 cfu/g) and increased to a final population of 105 cfu/g at 120h). In the
control barrel, L.plantarum was detected at the beginning of fermentation
(105 cfu/g), but died off in the first 24h. At 96h it was detected again (103 cfu/g),
and increased to a final population of 106 cfu/g (Figure 4.17 c). L.fermentum, was
first detected in the inoculated barrel at 24h (3 x 103 cfu/g). It increased in
population to 106 cfu/g at 48h, and then continued to increase slowly until the
end of fermentation (Figure 41.6 d)
Populations of L.fermentum were found in all fermentations. In both of the box

fermentations (control and inoculated), L.fermentum was first isolated at 72h
with populations of 103cfu/ g beans. It then grew to final populations of 106
cfu/g beans in the control box, and 105 cfu/g beans in the inoculated box
258
fermentations (Figure 4.16 a, b). In the control barrel fermentation, L.fermentum

grew from an initial population of 1 x 103 cfu/ g beans at 24h, to a maximum
population of 9 x 105 cfu/g beans at 48h. It then decreased to around 104 cfu/g
beans and remained at this level until the end of fermentation (Figure 4.17 c).
In the inoculated barrel fermentation, L.fermentum was only detected during the
last 2411, at about 104 cfu/g beans (Figure 4.17 d).
4.3.3.3(C) Barrel fermentations using a mixture of Hanseniaspora guilliermondii,
Issatcienkia orientalis, Saccharomyces cerevisiae and Kluyveromyces
marxienus
Figure 4.18 shows the growth of Acetobacter pasteurianus, Acetobacter tropicalis,

Lnctobccillus plantarum, Lactobacillus fermentum and Bacillus subtilis during these
barrel fermentations.
mixture; of yeast. Cocoa beans were fermented in: (a) barrel, no added starter culture, and (b)
barrel, 3.guilliermondii, I.orientalis, S.cerevisiae & K.marxianus added. Acetobacter
pasteuranus ♦; Acetobacter tropicalis LA; Lactobacillus plantarum O; Lactobacillus
fermentim ■/ Bacillus subtilis #.
In botl the uninoculated (control) and inoculated barrel, A.pasteurianus was

first isolated at 24h. In the control barrel, it grew from an initial population of
103 cfu/g, reaching 107 cfu/g at the end of fermentation (120 h) (Figure 4.18 a).
In the noculated barrel, A.pasteurianus was initially present at 9xl03 cfu/g and
slowly grew to a maximum population of 8xl06 cfu/ g at 96h, before declining
slight!7 to 7x 106cfu/g at 120 h (Figure 4.18 b).
259
Acetobacter tropicalis was initially detected at populations of 105 cfu/g beans in

both fermentations. In both the control and inoculated fermentations, its
population first decreased (0-24 h), and then increased again(24-120 h), reaching
maxima of 107 cfu/g at 120h in both fermentations (Figure 4.18 a, b).
In the uninoculated barrel fermentation, L.plantarum only appeared in the last
24h fermentation, reaching a final population of 4xl05 cfu/g. In the inoculated
barrel, L.plantarum was first detected at 24 h, at a population of 103 cfu/g. As the
fermentation proceeded, L.plantarum grew and reached a maximum population
of 106 cfu/g at the end of fermentation (Figure 4.18 b).
In the control (uninoculated) barrel, populations (103 cfu/g) of L.fermentum
were detected at 72h of fermentation. By the end of fermentation, the
population of L.fermentum had increased to 8xl05 cfu/g (Figure 4.18 a).
L.fermentum was not detected in the barrel fermentation inoculated with
H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus.
Bacillus spp. were not detected in the control barrel fermentation. In the
inoculated barrel, B.subtilis was initially detected at 48 h (103 cfu/g), increased
to 104 cfu/g (72h), and then died off (Figure 4.18 b).
4.3.3.4 Changes in the concentrations of sugars during fermentation using
Figure 4.19 shows the changes in concentrations of sugars in the pulp and seed
of samples taken from the pilot-scale fermentations with a mixture of
H.guilliermondii and I.orientalis cultures.
In the pulp of the unfermented cocoa beans (0 h), glucose (5.3 %) and fructose
(6.2 %) were the only sugars detected. By 72h, glucose and fructose had
decreased to undetectable levels in the pulp of all fermentations (or by 48 h, in
the inoculated box) (Figure 4.19 a, c, e, g).
In the seed of unfermented cocoa beans, small amounts of glucose (0.05%) and
fructose(0.1%) and sucrose (1.7%) were found. For all fermentations except the
control barrel, the seed concentrations of sucrose, fructose and glucose had
decreased to levels of less than 0.1% by 72 h (Figure 4.19 b, d, h).
In the control barrel, sucrose only decreased to 1.3% by 72 h, and then
decreased further to 0.2% by 120h (Figure 4.19 f).
260
24 48 72 96 120
Time (h) Time (h)
24 48 72 96 120 24 48 72 96 120
Time (h) Time (h)
24 48 72 96 120
Time (h) Time (h)
24 48 72 96 120
Time (h) Time (h)
Fig. 4.19 Changes in the concentration of sugars during fermentation of cocoa beans using
defined mixtures of yeast, (a) Pulp and (b) seed from box, no added starter culture;
(c) pulp and (d) seed from box, H. guiUiermondii & /. orientals added; (e) pulp and (f) seed
from barrel, no added starter culture; (g) pulp and (h) seed from barrel, H. guilliermondii & /.
orientalis added. Glucose ♦; fructose □; sucrose O.
261

Glucose, fructose and sucrose were detected in samples of cocoa beans

fermented with defined mixed cultures containing S.cerevisiae and K.marxianus.
Figure 4.20 shows the changes in concentration of these sugars.
The concentration of these sugars decreased, in both pulp and seed fractions of
the cocoa beans, as the fermentations proceeded. This decrease in sugar
concentration occurred at similar rates in all four fermentations.
In the unfermented cocoa pulp, glucose and fructose were initially detected at
concentrations of 5.0 % and 6.6 % respectively. In the seed fraction, glucose,
fructose and sucrose were detected at low concentrations (0.5%, 0.6% and 1.7%
respectively).
At 120h only traces (< l%w/w) of any sugars were detected in the pulp or seed
of all fermentations, except for the barrel inoculated with S.cerevisiae and
K.marxianus. Samples of pulp taken from the inoculated fermentation at 120 h,
contained glucose and fructose (both 1.0% ±0.2).
The decrease in concentration of the sugars during fermentation suggested that

the sugars were utilised by microorganisms growing in the pulp, and during
endogenous biochemical reactions inside the cocoa seeds.
In the seeds fermented in barrels, sucrose were utilised slightly faster than in
the seeds fermented in boxes (compare Figure 4.20 f & h with b & d).
262
48 72 96 120 48 72 96 120
Time (h) Time (h)
48 72 96 120 48 72 96 120
Time (h) Time (h)
48 72 96 120 24 48 72 96 120
Time (h) Time (h)
48 72 96 120 24 48 72 96 120
Time (h) Time (h)
Fig. 4.20. Changes in the concentration of sugars during fermentation of cocoa beans using
defined mixtures of yeast, (a) Pulp and (b) seed from box, no added starter culture;
(c) pulp and (d) seed from box, S.cerevisiae & K.marxianus added; (e) pulp and (f) seed from
barrel, no added starter culture; (g) pulp and (h) seed from barrel, S.cerevisiae & K.marxianus
added. Glucose ♦; fructose □; sucrose O.
263

marxianus
Glucose (5%) and fructose (6.8%) were the only sugars detected in the fresh
pulp at the beginning of these fermentations. These sugars were consumed
during fermentation, and by the end of fermentation only traces remained
(<0.1%) (Figure 4.21 a, c).
Sucrose (1.6%), glucose (0.6%) and fructose (0.6%) were found in the
unfermented seeds at Oh.
In the control barrel, the concentration of sucrose in the seeds decreased to
undetectable levels by 72h. The concentration of glucose also decreased slightly
to 0.4%, while the concentration of fructose increased slightly to 0.7% By the
end of fermentation, only traces (<0.1%) of these sugars remained in the seed
(Figure 4.21 b).
In the seeds of the beans fermented in the inoculated barrel, the concentration
sucrose decreased more slowly than in the control. As in the control, the seed
concentration of fructose increased slightly to 0.8% at 72h. At completion of
fermentation, less than 0.1 % of fructose and sucrose and 0.4% glucose were
detected (Figure 4.21 d).
48 72 96 120 48 72 96 120
Time (h) Time (h)
Fig. 4.21. Changes in the concentration of sugars during fermentation of cocoa beans using
defined mixtures of yeast, (a) Pulp and (b) seed from barrel, no added starter culture; (c) pulp
and (d) seed from barrel, H.guilliermondii, 1.orientalis, S.cerevisiae & K.marxianus added.
Glucose ♦; fructose □; sucrose O.
264
4.3.3.5 Changes in the concentration of ethanol during fermentations using


Ethanol was not detected in the pulp or seed of fresh cocoa bean (Oh) samples,
but was produced during the course of all four fermentations. The change in
ethanol concentration in each of these fermentations is shown in Figure 4.22.
Maximum concentrations of ethanol were recorded at the mid-point of

fermentation (48 or 72h), in both pulp and seed fractions. Higher concentrations
of ethanol were recorded in the box and barrel fermentations inoculated with
mixtures of H.guilliermondii and I.orientalis, compared to the controls (compare
Figure 4.22 b & d with a & c).
At 120 h, the concentrations of ethanol in the seed and pulp of the barrel
fermentations (inoculated and control) were higher than in the box
fermentations (inoculated and control) (compare Figure 4.22 c & d with a & b).
24 48 72 96 120 24 48 72 96 120
Time (h) Time (h)
24 48 72 96 120 24 48 72 96 120
Time (h) Time (h)
Fig. 4.22. Changes in the concentration of ethanol during fermentation of cocoa beans using
defined mixtures of yeast, (a) box, no added starter culture; (b) box, H. guilliermondii & I.
orientalis added; (c) barrel, no added starter culture; (d) barrel, H. guilliermondii & I. orientalis
added. Seed ♦; Pulp □.
265

Figure 4.23 shows the changes in concentration of ethanol in the pulp and seed
of each of the fermentations. At Oh of fermentation, traces (< l%w/ w) of
ethanol were found in the pulp. This may indicate that some fermentation of
the beans occurred in the 60-90 minute period it took to extract the beans from
the pods.
After 72 h, the concentration of ethanol had increased to between 3.5 - 7% in the

pulp and 2 - 5% in the seed. Fermentation vessel did not greatly affect the
concentration of ethanol produced (compare Figure 4.23 a with c).
The addition of cultures of S.cerevisiae and K.marxianus led to increased levels

of ethanol being produced in the pulp, and diffused into the seed
(compare Figure 4.23 b & d with a & c).
The highest concentrations of ethanol occurred in the 72h pulp and seed
samples from the inoculated barrel (7% and 4.9% respectively)(Figure 4.23 d).
In all fermentations, the concentration of ethanol in the pulp and seed had
fallen to <1.0% by 120 h.
48 72 96 120 48 72 96 120
Time (h) Time (h)
defined mixtures of yeast, (a) box, no added starter culture; (b) box, S.cerevisiae &
K.marxianus added; (c) barrel, no added starter culture; (d) barrel, S.cerevisiae & K.marxianus
added. Seed ♦; pulp □.
266

marxianus
Figure 4.24 shows the changes in concentration of ethanol in the pulp and seed
of each of the fermentations. Traces (< l%w/w) of ethanol in the pulp at Oh may
indicate that some fermentation of the beans occurred between pod splitting
and placing the beans in barrels. Ethanol was produced during both
fermentations, and maximum levels of ethanol were recorded in pulp and seed
samples taken at 72h of fermentation.
In the control barrel, the concentration of ethanol in the pulp and seed at 72 h
were 4.2% and 3.6% respectively (Figure 4.24 a).
In the inoculated barrel, the concentration of ethanol detected in the pulp at 72h
was 4.68% - significantly higher than the control. The concentration of ethanol
detected in the seed at 72h (3.58%), however, was similar to the control
(Figure 4.24 b). At the end of both fermentations, only traces (<1.0%) of ethanol
were detected in the seed or pulp samples from either fermentation.
24 48 72 96 120 48 72 96 120
Time (h) Time (h)
defined mixtures of yeast, (a) barrel, no added starter culture; (b) barrel, H.guilliermondii,
1.orientalis, S.cerevisiae & K.marxianus added. Seed ♦; pulp □.
267
4.3.3.6 Changes in the concentration of organic acids during fermentation


Figure 4.25 and Table 4.13 show the change in concentration of organic acids in
the pulp and seed of cocoa beans fermented with or without added cultures of
H.guilliermondii and I.orientalis. Citric, malic, lactic, acetic and oxalic acids were
detected in the beans.
Fresh cocoa pulp (0 h) contained citric acid at a concentration of 55 mg/g.

During fermentation the concentration of citric acid decreased to 40 mg/g in all
fermentations, except the uninoculated (control) box, where the final
concentration of citric acid was ~30 mg/g (compare Figure 4.25 a with c, e, g).
Malic acid was initially present in the pulp at concentrations of ~10 mg/g.
Except in the uninoculated box, the concentration of malic acid increased
during fermentation to a final level of 20 mg/g (Figure 4.25 c, e, g). In the
uninoculated box the concentration of malic acids remained around 10 mg/g
for the whole fermentation (Figure 4.25 a).
Small amounts of oxalic acid (not shown in Figure 4.27) were found at a
constant concentration of 2 mg/g in the pulp of the fermenting cocoa beans.
Acetic acid production differed between the box and barrel fermentations. In
the box fermentations, maximum acetic acid levels were recorded at the end of
fermentation. In the barrel fermentations, maximum levels were recorded at
72 h, after which the concentration of acetic acid decreased. Fermentation with
cultures of H.guilliermondii and I.orientalis led to a more rapid production of
acetic acid in the box only (Figure 4.25 a,c). The final concentration of acetic acid
in the pulp from the control box fermentation (21 mg/g) similar to the the
inoculated box fermentation (24 mg/g) (Figure 4.25 a,c). The control barrel had
a slightly higher final concentration of acetic acid (15 mg/g) compared to the
inoculated barrel (5 mg/g).
Lactic acid was not found in unfermented cocoa pulp (Oh), but was produced
during fermentation. When the fermentations were conducted in a box, lactic
268
acid increased gradually from the beginning of fermentation. When barrel

vessels were used, lactic acid was only produced in the last 48h of
fermentations. Fermentation with cultures of H.guiltiermondii and I.orientalis did
not significantly affect the production of lactic acid in the pulp.
Organic acids were also detected in the seed fraction of fresh and fermenting
cocoa beans. In all fermentations, the concentration of citric acid in the seed
decreased from 16 mg/g at the beginning of fermentation(0 h), to ~10 mg/g at
the end. This decrease did not appear to be affected by fermentation vessel or
the addition of cultures.
The concentration of malic acid in the seed remained constant throughout the
fermentations, and was similar for all treatments (15 mg/g). Traces of oxalic
acid were also found in the seed of the cocoa bean samples, and remained at
constant concentration of 1 mg/g during all four fermentations (Table 4.13).
The concentration of both lactic acid and acetic acid in the seed increased
during fermentation. In all of the fermentations, the final concentration of lactic
acid was in the range 10 - 15 mg/g. The final concentration of acetic acid in the
seed was higher in the beans fermented in boxes (8 mg/g) compared to those
fermented in barrels (1-5 mg/g ) (compare Figure 4.25 b & d with e & h).
The addition of cultures did not cause a significant difference in the final
concentration of acetic or lactic acids, compared to the controls (Table 4.13).
During the fermentations with H.guilliermondii and I.orientalis, it was observed

that the final concentrations of lactic acid were equal to or greater than those of
acetic acid, particularly in the seed. This was in contrast to previous
fermentation experiments, in which the final concentration of acetic acid in the
seed was always greater than the concentration of lactic acid in the seed.
269
3 30 ■
E 10 ■
24 48 72 96 120 24 48 72 96 120
Time (h) Time (h)
24 48 72 96 120 24 48 72 96 120
Time (h) Time (h)
,gf> 20 ■
0 24 48 72 96 120 0 24 48 72 96 120
Time (h) Time (h)
0 24 48 72 96 120 0 24 48 72 96 120
Time (h) Time (h)
* This fermentation was terminated at 96h. as described in the methods section 4.2.1.2
Fig. 4.25. Changes in the concentration of organic acids during fermentation of cocoa beans
using defined mixtures of yeast, (a) Pulp and (b) seed from box, no added starter culture;
(c) pulp and (d) seed from box, H. guilliermondii & I. orientalis added; (e) pulp and (f) seed
from barrel, no added starter culture; (g) pulp and (h) seed from barrel, H. guilliermondii &
I. orientalis added. Citric acid ♦; acetic acid □; lactic acid O; malic acid ■.
Note: The following acids remained at constant concentrations during fermentation and are not
shown in the graphs: oxalic acid (pulp 2 mg/g & seed 1 mg/g); and malic acid (seed 3-4 mg/g).
270
Table 4.13 - Summary of changes in organic acids in inoculated cocoa bean fermentations
Fermentation Method Acid Pulp (mg/g) dry basis Seed (mg/g) dry basis
(Description of defined Start End Change Start End Change

mixture of yeast) (Oh) (120h) (Oh) (120h)
Box, uninoculated Citric 54 8 -46 16 10.8 -5.2

(control)
Malic 11 11 0 3.5 3.5 0
Oxalic 2 2.3 0.3 1 1 0
Lactic 0 24 24 1 11 10.1
Acetic 0 21 21 0 8.8 8.8
Box inoculated with Citric 54 36 -18 16 11.5 -4.5

H.guilliermondii and
Lorientalis Malic 11 18 7 3.5 3 -0.5
Oxalic 2 2 0 1 1 0
Lactic 0 29 29 0.8 8.3 7.6
Acetic 0 24 24 0 9.5 9.5
Barrel, uninoculated Citric 54 38 -16 16 9.6 -6.4

(control)
Malic 11 20 9 3.5 4 0.5
Oxalic 2 2.1 0.1 1 1 0
Lactic 0 31 31 0.9 15 14.1
Acetic 0 15 15 0 4.9 4.9
Barrel inoculated with Citric 54 37 -17 16 11.1 -4.9

H.guilliermondii and
/. orientalis Malic 11 21 10 3.5 4.2 0.7
Oxalic 2 2.1 0.1 1 1 0
Lactic 0 23 23 0.8 11 10.2
Acetic 0 5 5 0 1.1 1.1

271

cerevisiae and Kluyveromyces marxianus.
Citric, malic, and oxalic acids were detected in the pulp and seed of the
unfermented cocoa beans used in these fermentations. Acetic and lactic acid
were produced in the pulp during fermentation, and diffused into the seeds.
Figure 4.26 and Table 4.14 show the change in concentrations of these organic
acids in cocoa beans fermented with, or without, added cultures of S.cerevisiae
and K.marxianus.
The concentration of citric acid in the fresh cocoa pulp was 41 mg/g. After 72h,
the concentration of citric acid decreased in all fermentations. In all
fermentations except for the uninoculated (control) barrel, the final
concentration of citric acid in the pulp was ~30 mg/g (Figure 4.26 a, c, g). In the
uninoculated barrel fermentation (Figure 4.26 e), the decrease in citric acid was
greater, and the final concentration was significantly lower (18 mg/g). Malic
and oxalic acids were initially present in the fresh pulp, at concentrations of 10
mg/g and 1.5 mg/g respectively. The concentration of these acids did not
change significantly during fermentation.
Acetic acid was not detected in the fresh pulp samples, but was produced,
particularly from 72 h onwards. Except for the uninoculated (control) barrel,
similar amounts of acetic acid were detected in the pulp of all of the
fermentations (21 - 22 mg/g) upon completion (Figure 4.26 a, c, g). In the
control barrel, the final concentration of acetic acid in the pulp was lower (14
mg/g) (Figure 4.26 e). Fermentation with S.cerevisiae and K.marxianus did not
appear to affect the production of acetic acid in the pulp, under the conditions
tested.
Fermentation with S.cerevisiae and K.marxianus, and use of box or barrel,

affected the production of lactic acid during fermentation.
In the control box, lactic acid was produced between 0-72h, after which it
decreased in concentration, to a final level of 15 mg/g (Figure 4.26 a). In the box
inoculated with S.cerevisiae and K.marxianus, lactic acid was produced mainly
272
after 72h, to a final concentration of 22 mg/g (Figure 4.26 c). This was
significantly higher than the control box. In the control barrel fermentation,
lactic acid was again produced mainly after 72h, to a final concentration of 24.6
mg/g. This was significantly higher than in the control box, suggesting that
more lactic acid was produced in the barrels, compared to the boxes.
In the inoculated barrel, lactic acid was mainly produced in the first half of
fermentation, reaching 30 mg/g by 72 h. Between 72 -120 h it increased slightly,
to a final concentration of 32 mg/g (Figure 4.26 g). This was significantly higher
than the final lactic acid concentration in the inoculated box.
In the cocoa seeds, the concentration of citric (10 mg/g), malic (5 mg/g) and
oxalic (1.5 mg/g) acids did not change significantly during fermentation.
As acetic acid was produced in the pulp, it diffused into the seeds. Acetic acid
was only detected in the seeds from 72 h onwards. In all fermentations the final
concentration of acetic acid in the seeds was similar: between 5-10 mg/g.
Flowever, seeds from the uninoculated box had a final concentration of acetic
acid (9.9 mg/g) that was significantly greater than the rest.
Fermentation with S.cerevisiae and K.marxianus, and use of box or barrel,
affected the final concentration of lactic acid in the seeds. In the control box, the
concentration of lactic acid in the seeds increased steadily throughout
fermentation to a final value of 4.9 mg/g (Figure 4.26 b). In the seeds from the
inoculated box, a higher final concentration of lactic acid was detected (7.4 mg/
g) (Table 4.14). A significantly higher concentration of lactic acid was found in
the seeds from the inoculated barrel (12.5 mg/g) compared to the control barrel
(8.8 mg/g) (Table 4.14). Finally, the final concentration of lactic acid in seeds
from the barrels (control and inoculated) were significantly higher than those
from the boxes.
All significant differences noted in the text above were confirmed using the
Student's T-test at a 90% confidence interval(a=0.1)(calculations not shown).
273
3 30 ■
3 30 ■
E 10-
24 48 72 96 120 24 48 72 96 120
Time (h) Time (h)
3 30 ■
> 20 ^ E 10-
24 48 72 96 120 24 48 72 96 120
Time (h) Time (h)
3 30 -
3 30 ■
> 20 ^ E 10 -
24 48 72 96 120 24 48 72 96 120
Time (h) Time (h)
3 30 -
E 10 ■
24 48 72 96 120 24 48 72 96 120
Time (h) Time (h)
using defined mixtures of yeast, (a) Pulp and (b) seed from box, no added starter culture; (c)
pulp and (d) seed from box, S.cerevisiae & K.marxianus added; (e) pulp and
(f) seed from barrel, no added starter culture; (g) pulp and (h) seed from barrel, S.cerevisiae &
K.marxianus added. Citric acid ♦; acetic acid □; lactic acid O.
shown in the graphs: citric acid (seed 10 mg/g); malic acid (pulp 10 mg/g & seed 5 mg/g);
oxalic acid (pulp 1.5 mg/g & seed 1-2 mg/g);.
274
Table 4.14 - Summary of the changes in the concentration of organic acids during fermentation
of Australian cocoa beans using defined mixtures of yeast.
Fermentation Method Acid Pulp (mg/g) dry basis Seed (mg/g) dry basis
(Description of defined Start (Oh) End Change Start (Oh) End Change
mixture of yeast) (120h) (120h)
Box, uninoculated Citric 41 28 -13 10 9.5 -0.5
(control)
Malic 10 9.9 -0.1 5 4.9 -0.1
Oxalic 1.5 1.5 0 1.5 1.5 0
Lactic 0 15 15 0.3 5.2 4.9
Acetic 0 22 22 0 9.9 9.9
Box inoculated with Citric 41 30 -11 10 10 0

S.cerevisiae and
Malic 10 9.9 -0.1 5 4.9 -0.1
Kmarxianus
Oxalic 1.5 1.5 0 1.5 1.4 -0.1
Lactic 0 22 22 0.45 7.8 7.4
Acetic 0 21 21 0 5.2 5.2
Barrel, uninoculated Citric 41 18 -23 10 10.1 0.1

(control)
Malic 10 9.9 -0.1 5 4.9 -0.1
Oxalic 1.5 1.4 -0.1 1.5 2.1 0.6
Lactic 0 25 25 0.1 8.9 8.8
Acetic 0 14 14 0 4.9 4.9
Barrel inoculated with Citric 41 30 -11 10 10.1 0.1

S.cerevisiae and
Malic 10 9.9 -0.1 5 5 0
K marxianus
Oxalic 1.5 1.5 0 1.5 1.1 -0.4
Lactic 0 32 32 0.5 13 12.5
Acetic 0 21 21 0 7.8 7.8

275

marxianus.
Citric, malic and oxalic acids were detected in the pulp and seeds of the fresh
cocoa beans used in this experiment. As before, acetic and lactic acids were
produced in the pulp during fermentation, and also diffused into the seeds. The
changes in the concentrations of these organic acids are shown in Figure 4.27
and Table 4.15. It is apparent from the figure that the use of a culture of
H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus had little effect on these
organic acids.
Citric acid was detected in the fresh pulp at a concentration of 41 mg/g and
decreased in concentration after 72 h. This decrease was greater in the control
barrel fermentation, where the final concentration of citric acid was 18 mg/g
(Figure 4.27 a), and lesser in the inoculated barrel fermentation, where the final
concentration of citric acid was 26 mg/g (Figure 4.27 c).
The concentration of malic and oxalic acids in the fresh pulp were initially 10
mg/g and 1.5 mg/g respectively and did not change significantly during
fermentation.
Acetic acid was not detected in unfermented cocoa pulp at Oh, but was
produced acetic acid was produced after 72 h. It was produced at similar rates
in both fermentations and the final concentration of acetic acid in the pulp from
the control barrel fermentation (14 mg/g) was only slightly lower than in the
inoculated barrel fermentation (18 mg/g). This difference was not statistically
significant.
Fermentation with H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus did
not affect the production of lactic acid during fermentation. In pulp from
control barrel fermentation, the final concentration of lactic acid was 25 mg/g,
while in the inoculated barrel fermentation it was 22 mg/g (Figure 4.27 a, c).
In the seeds from both the control and inoculated barrels, the concentrations of
citric (10 mg/g), malic (5 mg/g) and oxalic (1.5 mg/g) acids remained constant
throughout fermentation.
276
The concentration of acetic acid in the seed increased between 72-120 h of each
of the fermentations. At completion of fermentation, the concentration of acetic
acid in the seeds from the control barrel (4.9) was only slightly lower than in the
seeds from the inoculated barrel (6.5) (compare Figure 4.27 b & d).
In both the control and inoculated barrels, the concentration of lactic acid
steadily increased in the seed during fermentation. This increase occurred at a
similar rate in both fermentations, and the final concentration of lactic acid in
both was around 8 mg/g (Table 4.15).
3 30 -
3 30 J
> 20 -
E 10 ■
24 48 72 96 120 24 48 72 96 120
Time (h) Time (h)
3 30 -
3 30 J
E 10-
24 48 72 96 120 24 48 72 96 120
Time (h) Time (h)
using defined mixtures of yeast, (a) Pulp and (b) seed from barrel, no added starter culture; (c)
pulp and (d) seed from barrel, H.guilliermondii, I. orientalis, S.cerevisiae & K.marxianus added.
Citric acid ♦; acetic acid □; lactic acid O; malic acid ■.
shown in the graphs: oxalic acid (pulp 1.5 mg/g & seed 1-2 mg/g); malic acid (seed 5 mg/g).
277
Table 4.15 - Summary of the changes in the concentration of organic acids during fermentation
of Australian cocoa beans using defined mixtures of yeast.
Fermentation method Acid Pulp (mg/g) dry basis Seed (mg/g) dry basis
(Description of defined Start (Oh) End Change Start (Oh) End Change
mixture of yeast) (120h) (120h)
Barrel, uninoculated Citric 41 18 -23 10 10.1 0.1
(control)
Malic 10 9.9 -0.1 5 4.9 -0.1
Oxalic 1.5 1.4 -0.1 1.5 2.1 0.6
Lactic 0 25 25 0.1 8.9 8.8
Acetic 0 14 14 0 4.9 4.9
Barrel, inoculated with Citric 41 26 -15 10 10 0

H.guilliermondii,
Malic 10 9.9 -0.1 5 5 0
I.orientalis, S.cerevisiae
and Kmarxianus Oxalic 1.5 1.1 -0.4 1.5 2 0.5
Lactic 0 22 22 0 8.2 8.2
Acetic 0 18 18 0 6.5 6.5
4.3.3.7 Quality evaluation of cocoa beans produced by fermentation using

The dried fermented beans produced during the pilot-scale fermentations were
subjected to a range of quality assessments.
This included the cut test, and measurements of bean size, moisture content,
shell content, fat content and nib pH. Further quality tests (protein, sugar
content, microbiological quality, fat melting profile) were also performed on
samples of chocolate made with these beans
Results from the cut test performed on the beans from this experiment are
shown in Figure 4.28.
The cut test showed no fully purple, slaty, germinated or insect infested beans
in any of the samples evaluated. These data, indicated that the beans were of
acceptable quality. The cut test results also suggested that the addition of mixed
yeast cultures could affect the fermentation index of cocoa beans. The beans
fermented with cultures of H.guilliermondii and I.orientalis had a higher
278
proportion of completely brown beans (~65%) compared to the uninoculated

controls (~45%). This occurred for both box and barrel fermented beans.
100
g 90
2 80
£ 70
o 60
& 50
& 40
8 30
S3 20
^ 10
0
Box Box Barrel Barrel
(uninoculated) (inoculated) (uninoculated) (inoculated)
Fig. 4.28. Cut test results for cocoa beans fermented using a mixture of Hanseniaspora
guilliertnondii & Issatchenkia orientalise ■ % beans fully brown, 11% beans part brown/part
purple.
Other quality data are presented in Table 4.16. The beans were well dried, with
a moisture content less than 7.5% as recommended by Wood and Lass (1985)
and the ISO standards for cocoa beans (ISO, 1977).
Table 4.16 - Quality evaluation of cocoa beans fermented using a mixture of
Hanseniaspora guilliermondii & Issatchenkia orienta/is.

Quality parameter Mean St. dev
Bean size 89
81 80 88 83.0 4.36
(Number per lOOg)
% Bean moisture 5.1 4.3 4.8 4.4 4.73 0.40
% Nib moisture 4.3 3.6 4.2 3.6 4.03 0.37
% Shell content 12.9 11.3 14.1 12.4 12.8 1.40
% Nib fat (dry basis) 54.5 55 54.3 55 54.6 0.36
Nib pH 5.8 6.3 5.9 6.2 6.00 0.26
External appearance Acceptable Acceptable Acceptable Acceptable ... ...
Aroma evaluation Acceptable Acceptable Acceptable Acceptable ... ...
At 80-89 beans / lOOg, the beans were considered to be quite large (Wood &
Lass, 1985), with those fermented in the barrel being significantly larger than
those fermented in boxes.
279
Beans fermented with H.guiltiermondii and I.orientalis had significantly lower

bean and nib moisture levels, less shell, and higher nib pH levels compared to
the controls. The fat contents of the cocoa beans were not affected by either
fermentation vessel or inoculation with yeast cultures. All significant
differences mentioned were confirmed using the Students T-test, at a confidence
level of 95% (alpha = 0.05) (data not shown here).
The aroma and appearance of all samples were determined to be acceptable,
based on the guidelines for the commercial assessment of cocoa bean quality.
The results from the cut test, and other quality tests, are presented in Figure 4.29
and Table 4.17.
100 n
90 -
caj
CD
x>
4-1
<
o
CD
3
c
CD
O
<D
0_

Fig. 4.29. Cut test results for cocoa beans fermented using a mixture of Saccharomyces
cerevisiae & Kluyveromyces marxianus; ■ % beans fully brown, !H% beans part brown/part
purple.
The cocoa beans produced during these fermentations were of acceptable

quality: no sample contained any slaty, insect affected or germinated beans, and
less than 20% of the beans were fully purple(Wood and Lass, 1985). The cut test
data also showed that the proportion of fully brown beans was similar in all
bean samples. This suggests that the use of cultures containing S.cerevisiae and
K.marxianus had little effect on the fermentation index of the beans produced.
The control box gave the highest proportion of fully brown beans (78%), while
the barrel inoculated with the mixture of S.cerevisiae and K.marxianus gave the
lowest proportion of fully brown beans (68%).
280
Comparison of the quality data in Table 4.17, on the basis of fermentation

vessel, suggested that the cocoa beans fermented in barrels were smaller than
those fermented in boxes. The data also indicated that the beans were of
acceptable commercial quality. The moisture content of both whole beans and
nibs was below 7.5%, and the appearance and aroma criteria were acceptable
(Wood & Lass, 1985). With respect to the properties of bean size, moisture
content, shell content, fat content and nib pH, there were only slight differences
between the beans from control and inoculated fermentations, but these were
not statistically significant (Student's T-Test, 95% confidence level).
Table 4.17 - Quality evaluation of cocoa beans fermented using a mixture of Saccharomyces
cerevisiae & Kluyveromyces marxianus.

Quality parameter Mean St.dev.
Bean size - no./lOOg 89 88 92 96 913 0.71
% Bean moisture 5.5 4.9 5.3 5.3 S3 0.42
% Nib moisture 4.6 4.3 4.8 4.2 43 0.21
% Shell content 11.8 11.5 11.7 12.3 113 0.21
% Nib fat (dry basis) 57.6 58.2 56.7 57.1 57.4 0.42
Nib pH 5.5 5.9 6.1 5.6 5.8 0.28
External appearance Acceptable Acceptable Acceptable Acceptable -- --
Aroma evaluation Acceptable Acceptable Acceptable Acceptable -- --

marxianus.
The results from the cut test, and other quality tests, performed on the beans
fermented with H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus are
presented in Figure 4.30 and Table 4.18. Cut test data indicated that the cocoa
beans produced during these fermentations were of acceptable quality, as no
sample contained any slaty, insect affected or germinated beans, and less than
20% of the beans were fully purple(Wood and Lass, 1985). Beans fermented in a
barrel inoculated with H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus
had a higher proportion of fully brown beans (82%), than the uninoculated
barrel (72%). However, this difference was not statistically significant.
281
Barrel Barrel
(uninoculated) (inoculated)
Fig. 4.30. Cut test results for cocoa beans fermented using a mixture of Hanseniaspora
guilliermondii, Issatchenkia orientalis, Saccharomyces cerevisiae & Kluyveromyces marxianus;
■ % beans fully brown, beans part brown/part purple.
Small differences were noted in the bean size, moisture content, shell content,
fat content and nib pH of beans from the inoculated and control fermentations.
However, analysis of the data using Students T-test (confidence level of
95%)found that these differences were not statistically significant.
The data in Table 4.18 also indicated that the beans were of acceptable
commercial quality. The moisture content of both whole beans and nibs was
below 7.5%, and the appearance and aroma of the beans were acceptable. The
data obtained for these beans also compared favourably against quality data
obtained for cocoa bean from Ghana and Indonesia (Section 3.3.1.8, Chapter 3).
Table 4.18 - Quality evaluation of cocoa beans fermented using a mixture of Hanseniaspora
guilliermondii, Issatchenkia orientalis, Saccharomyces cerevisiae & Kluyveromyces marxianus.
Barrel Barrel
Quality parameter Mean St.dev
(uninoculated) (inoculated)
Bean size - no./100g 92 99 955 4.95
% Bean moisture 5.3 5.7 55 0.28
% Nib moisture 4.8 5 4.9 0.14
% Shell content 11.7 12.1 11.9 0.28
% Nib fat (dry basis) 56.7 56.5 56.6 0.14
Nib pH 6.1 6 6.1 0.07
External appearance Acceptable Acceptable - --
Aroma evaluation Acceptable Acceptable -■ •*

282
4.3.3.7 (D) Chocolate quality assurance testing results.
Chemical analysis of all chocolate samples, prepared from beans fermented in

the pilot-scale fermentations, revealed that they were similar in their
composition. It was found, using the Student's T-Test (99% confidence), that
there was no significant difference between any of the samples, including the
Ghana standard. The average composition data were: sucrose, 30%; fat 47%;
cocoa solids non-fat, 23%; protein in non-fat solids, 7%. Microbiological testing
of the chocolate samples found that all samples, including the Ghana reference,
met the required standards, with less than 100 cfu/g yeasts or moulds, less than
10 cfu / g coliforms and no Salmojialla spp. detected (in 25g of sample) (Data not
shown here).
The samples were also subjected to the Padley-Timms method for determining
the solid fat content (Padley and Timms, 1980) (Figure 4.31). The chocolate
samples contained only cocoa butter, based on their fat melting profiles. The
tests also revealed that the fat from the Queensland beans had a significantly
lower melting temperature compared to the Ghanaian beans.
100 T
Temperature
Figure 4.31 - Fat melting curves for chocolate made with beans from Queensland ■, Ghana 0.
283
4.3.3.8 Sensory evaluation of chocolate made from cocoa beans fermented

The dried fermented cocoa beans were made into chocolate samples and
subjected to sensory evaluation. Difference testing was performed to determine
if untrained panelists could distinguish between different samples. Liking
testing was performed to assess the basic acceptability of the samples compared
to chocolate made from Ghanaian cocoa beans (Section 4.2.6).

The results of the difference (triangle) tests are presented in Table 4.19. The
majority of untrained panelists correctly distinguished between the control
samples and samples fermented with cultures of H.guilliermondii and I.orientalis
(Table 4.19). Therefore, fermentation with mixtures of H.guilliermondii and
I.orientalis produced cocoa beans with a significantly different flavour compared
to the cocoa beans fermented by indigenous microflora (uninoculated control).
Table 4.19 - Sensory evaluation (triangle test) of chocolate samples made with cocoa beans
fermented in boxes or barrels inoculated with a mixture of Hanseniaspora guilliermondii and
Issatchenkia orientalis.

Number of
Sample correct needed for significance*
tests
Box (control) Vs. box (inoculated) 30 24 17

Barrel (control) Vs. barrel (inoculated) 30 24 17
* Adapted from a table in Stone and Sidel (2004). For one pair of samples to taste significantly different
from another, the “number of correct judgments” must be equal to or greater than the “number of
Liking tests were also performed on the chocolate samples by untrained

panelists (Tables 4.20 and 4.21). The mean liking scores indicated that all
chocolates made with cocoa beans from these pilot-scale fermentations achieved
higher scores than the chocolate made from Ghanaian beans. Furthermore,
chocolate made from the beans fermented in the barrel with H.guilliermondii and
I.orientalis received a mean liking score of 8.1. This indicates that, on average,
most panelists scored the flavour of this chocolate sample as "strongly liked,"
and suggests the beans were of particularly high quality.
284
Table 4.20 - Mean liking scores for chocolate made from cocoa beans fermented in boxes or
barrels inoculated with a mixture of Hcmseniaspora guilliermondii and Issatchenkia orientalis,
compared to chocolate made from Ghanian cocoa.
Mean liking score

Sample Standard deviation
(out of 10)
Ghana 4.98 2.37

Box (control) 6.33 1.80
Box (inoculated) 6.88 1.72
Barrel (control) 6.29 2.02
Barrel (inoculated) 8.10 2.85
To determine whether these observations were significant, T-test analysis of the

liking data was performed (Table 4.21). Chocolate made from the beans
fermented in a box with no added yeast culture received a mean liking score
that was not significantly different from the mean score of chocolate made using
Ghanaian cocoa beans. The chocolate made from all other cocoa bean samples
had mean liking scores that were significantly greater than for the chocolate
made from Ghanaian cocoa beans. It was concluded that the cocoa beans
produced in these experiments were of equal or greater quality than the
industry standard (Ghanian cocoa beans).
Table 4.21 - Results of Student’s T-test on the liking data for chocolate made from cocoa beans
fermented in boxes or barrels inoculated with a mixture of H. guilliermondii and /. orientalis.
T-test result at 95% confidence level: T > t

Sample pair T t*
Ghana, vs.
box (control)
2.25 1.76
box (inoculated) (Box (inoculated) > Ghana)
1.76 1.76
barrel (control) (Barrel (uninoculated) > Ghana)
4.19 1.76
barrel (inoculated) (Barrel (inoculated) > Ghana)
Box (control), vs.
box (inoculated)
Barrel (control),vs. Significantly different
2.41 1.76
barrel (inoculated) (Barrel (inoculated) > Barrel (uninoculated))
*One tailed critical value.
285
The T-test analyses also indicated that the chocolate made from beans
fermented in the barrel with H.guilliermondii and I.orientalis received a mean
liking score that was significantly higher than the sample fermented in a barrel
without added culture. More detailed information regarding differences in the
flavor profiles of chocolate prepared from beans fermented by the various
methods are presented in Section 4.3.5.
The results of these tests are presented in Tables 4.22, 4.23 and 4.24.
fermented in boxes or barrels inoculated with a mixture of Saccharomyces cerevisiae and
Kluyveromyces marxianus.

Number of
tests
Box (control) Vs. box (inoculated) 30 24 17

Difference testing of the chocolate samples found that untrained panelists could
correctly distinguish between the control samples and the samples that had
been fermented with S.cerevisiae and K.marxianus. Therefore, conducting
fermentations with S.cerevisiae and K.marxianus resulted in cocoa that tasted
significantly different to cocoa beans fermented by indigenous microflora alone
(control). Chocolate made from cocoa beans fermented in a barrel with
S.cerevisiae and K.marxianus also tasted significantly different from the control.
Therefore this mixture of yeast was also able to affect the flavour of the cocoa
beans produced.
Chocolate samples made with cocoa beans from these pilot-scale fermentations
achieved higher mean liking scores than the chocolate made from Ghanaian
cocoa (Table 4.23). Most of the samples received mean liking scores of between
7-8 out of 10. The highest mean liking score (9.18) was recorded for chocolate
made from beans fermented in a box without added yeast cultures (control).
286
Table 4.23 - Mean liking scores for chocolate made from cocoa beans fermented in boxes or
barrels inoculated with a mixture of Saccharomyces cerevisiae and K/uyveromyces marxianus,
compared to chocolate made from Ghanian cocoa.
Mean liking score Standard

Sample
(out of 10) deviation
Ghana 4.73 2.65
Box (control) 9.18 1.33
Box (inoculated) 7.64 2.00
Table 4.20. shows the results of T-test analysis of the liking data. These tests
confirmed that the cocoa beans produced from these pilot-scale fermentations
had a flavour quality equal to or greater than that of the Ghanian beans.
T-test analysis of the mean liking scores also indicated that panelists preferred
chocolate made from cocoa beans fermented in the box, without added cultures,
compared to chocolate made from cocoa beans fermented in the box and
inoculated with S.cerevisiae and K.marxianus.
fermented in boxes or barrels inoculated with a mixture of Saccharomyces cerevisiae and
K/uyveromyces marxianus.
T-test result at 95% confidence level **

Sample pair T t*
6.11 1.76
box (control) (Box (control) > Ghana)
2.72 1.76
box (inoculated) (Box (inoculated) > Ghana)
4.01 1.76
barrel (control) (Barrel (control) > Ghana)
2.95 1.76
barrel (inoculated) (Barrel (inoculated) > Ghana)
Box (control), vs. Significantly different
2.60 1.76
box (inoculated) (Box (uninoculated) > Box (inoculated))
Barrel (control),vs.
barrel (inoculated)
** Significant difference is established if T > t.The direction of any significant difference between mean
287

marxianus.
The cocoa beans produced in these experiments were made into chocolate and
subjected to difference and liking sensory tests. The results of these tests are
presented in Tables 4.25, 4.26 and 4.27.
fermented in boxes or barrels inoculated with a mixture of Hanseniaspora guilliermondii,
Issatchenkia orientalis, Saccharomyces cerevisiae and Kluyveromyces marxianus.

Number of
tests
Untrained panelists correctly distinguished between the control samples and

the samples that had been fermented with cultures of H.guilliermondii,
I.orientalis, S.cerevisiae and K.marxianus. This showed that cocoa beans fermented
in the barrel with this mixture of yeast produced cocoa beans with a flavour
that was significantly different to the flavour of the beans produced in the
uninoculated barrel.
The mean liking scores for the chocolate samples are shown in Table 4.26 and
the T-test analyses of these liking data are shown in Table 4.27. Chocolate made
from cocoa beans fermented in a barrel and inoculated with H.guilliermondii,
I.orientalis, S.cerevisiae and K.marxianus, had a mean liking score that was
significantly higher than that of the chocolate made from the Ghanaian cocoa
beans. This suggested that the cocoa beans produced during these experiments
had a better flavour than the Ghanian cocoa beans. The T-test data also showed
that chocolate made from the inoculated beans was not liked significantly more
than the chocolate made from the uninoculated beans. That is, the control and
inoculated beans tasted different from one another, but were liked by panelists
to a similar extent.
288
Table 4.26 - Mean liking scores for chocolate made from Australian cocoa fermented in barrels
inoculated with a mixture of Hanseniaspora gui/liermondii, Issatchenkia orientalis,
Saccharomyces cerevisiae and Kluyveromyces marxianus, compared to chocolate made from
Ghanian cocoa.
Mean liking score Standard
Sample
(out of 10) Deviation
Ghana 4.73 2.65
fermented in barrels inoculated with a mixture of Hanseniaspora guilliermondii, Issatchenkia
orientalis, Saccharomyces cerevisiae and Kluyveromyces marxianus.
T-test result at 95% confidence level **

Sample pair T t*
4.01 1.76
barrel (control) (Barrel (uninoculated) > Ghana)
Ghana, vs.
barrel (inoculated)
Barrel (control),vs.
barrel (inoculated)
** Significant difference is established if T > t.The direction of any significant difference between mean
289
4.3.3.8 (D) Comparison of chocolate samples made from cocoa beans fermented
in barrels and inoculated with different mixtures of yeast.
Chocolate made from beans fermented in barrels with added cultures were also
compared to one another across all pilot-scale experiments. This comparison
was made using the triangle test, to determine whether the different samples
tasted significantly different from one another. Table 4.28 presents the results of
these comparisons. Fermentation with different yeast cultures produced cocoa
beans that were distinctively different from one another, on the basis of flavour.
The liking scores for chocolate samples prepared from these beans were also
compared using the Student's T-test (a=0.05). These analyses (data not shown
here) indicated that there was no significant difference between the liking scores
for these samples. Therefore, while the different mixtures of yeast produced
cocoa beans with different flavours, all of the beans were liked to a similar
extent. Finally, the mean liking score for all samples fermented in barrels and
inoculated with mixtures of yeast was 7.7. This was significantly greater than
the liking score for the Ghana bean chocolate (4.8).
Table 4.28 - Comparison (triangle test) in the flavour of chocolate samples made with cocoa
beans fermented in barrels inoculated with different mixtures of yeast.
Number of Number of Correct judgements

Samples tests correct needed for significance*
Barrel (H.guilliermondii & /. orientalis), Vs. 29 17

30
barrel (S.cerevisiae & K.marxianus)
Barrel {H.guilliermondii & I.orientals), Vs.

barrel {H.guilliermondii, I. orientalis, 30 25 17
S.cerevisiae & K.marxianus)
Barrel {S.cerevisiae & K.marxianus) Vs.

barrel {H.guilliermondii, I.orientals, 30 26 17
S. cerevisiae & K. marxianus)
from another, the “number of correct judgments’' must be equal to or greater than the “number of
290
4.3.4 Quantitative descriptive analysis of cocoa beans

fermented with defined mixtures of yeast cultures.
Chocolate samples prepared from cocoa beans produced during the pilot-scale
fermentations, were evaluated to construct quantitative sensory profiles. These
analyses were done at the Cadbury-Schweppes Global Science Centre, UK.
The chocolate samples were analysed in two groups:
1. Group 1 samples were made from cocoa beans fermented with mixtures of
H. guilliermondii and I.orieiitalis cultures (box and barrel vessels).
2. Group 2 samples were made from cocoa beans fermented with mixtures of
S.cerevisiae and K.marxianus (box and barrel vessels), and H.guilliermondii,
I. orientalis, S.cerevisiae and K.marxianus cultures (barrel only).
The chocolate samples made from these cocoa beans were also compared to
chocolate made from Ghanaian cocoa beans (industry standard).
Figures 4.33, 4.33 and 4.34 present the sensory profiles as spider plots. All
sensory characteristics which were capable of discriminating between the
samples at a significant level are marked.
Measures of significant difference were calculated using Fisher's least
significant difference (p<0.05)(calculations not shown).
4.3.4.1 Appearance of chocolate samples
All chocolate samples prepared from Australian beans had a significantly

darker brown colour than chocolate made from the Ghanaian beans.
Within the Australian samples, chocolate made from beans fermented with
mixtures of H.guillermondi and I.orientalis, S.cerevisiae and K.marxianus, and
H.guilliermondii, I.orie?italis, S.cerevisiae and K.marxianus, were all significantly
darker than the chocolate samples made from the control (uninoculated) beans.
There were no significant differences between the appearance of the chocolates:
a) made from beans harvested in different months and fermented with any of
the mixtures of yeast.
b) made from different batches of control beans (uninoculated).
291
c) made from beans fermented in barrels or those fermented in boxes

(Figure 4.32 A, B).
4.3.4.2 Mouthfeel of chocolate samples
No significant differences were noted in the mouthfeel attributes (body, drying

or astringency) of any of the samples.
In general, the "body" of chocolate is determined by its physical processing,

and fat content. Since all samples were subject to the same preparation method,
it would be expected that the samples would have the same basic consistency
and texture.
4.3.4.3 Odour characteristics of chocolate samples
All of the Australian samples, except for the control box (Group 1) sample, had
significantly stronger fruity and doughy odours than the Ghanaian sample
When comparing the odour of the samples on the basis of the yeast inoculum,
across both groups, the following differences were noted:
• Cocoa beans fermented with H.guilliermondii and I.orientalis gave chocolate

with a significantly stronger doughy odour than the control samples
(Figure 4.32, A).
• Cocoa beans fermented with H.guilliermondii and I.orientalis produced

chocolate with significantly stronger doughy odours than the beans
inoculated with S.cerevisiae and K.marxianus (Figure 4.32, A).
• Cocoa beans fermented with mixtures of S.cerevisiae and K.marxianus

produced chocolate with significantly stronger acetic odours than the
Ghanaian sample.
• Within the group 2 samples only, cocoa beans fermented in a box, without
added yeasts, gave chocolate with a significantly stronger brown-fruit
odour than the other Australian samples in this group (Figure 4.32, B).
There were no clear, significant differences between the odour characteristics of

samples when compared on the basis of fermentation in barrels versus boxes.
292
Appearance - Colou?’ ^ -•—Box (control)

B 60 Box (Sc & Km)
Odour - Acetic Mouthfeel - Body -•—Barrel (Control)
—-— Barrel (Sc & Km)
— Barrel (All yeast)
■ Ghana
Odour - doughy Mouthfeel - Drying
Odour - fruity Mouthfeel - Astringent
Odour - brownfruit Odour - Toffee
Odour - spicy Odour - cocoa
Fig.4.32 Appearance, mouthfeel & odour of chocolate samples: A) Group 1 samples; B) Group 2 samples
a. Attributes with significant differences between the control and inoculated cocoa beans;
b. Attributes with significant differences between the Australian and Ghanaian cocoa beans.
“Control” - beans fermented without added yeast culture.
“Hg & Io” - beans fermented with added Hanseniaspora guilliermondii and Issatchenkia orientalis;
“Sc & Km” - beans fermented with added Saccharomyces cerevisiae and Kluyveromyces marxianus',
“All yeast” - beans fermented with added H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus.
293
4.3.4.4 Flavour characteristics of the chocolate samples
Significant differences in the flavour of the samples were identified. Here,

'flavour' refers to the taste while the chocolate remained in the mouth. Flavour
differences correlated to use of different mixtures of yeast in the fermentations.
The Australian samples were significantly less sweet, more acid, more bitter
and had stronger doughy and/or grainy flavours than the Ghanaian sample.
When comparing the flavour of the samples on the basis of the type of
inoculum used, across both groups, the following differences were noted:
• Cocoa beans fermented with H.guilliermondii and I.orientalis produced
chocolate with significantly stronger doughy flavour, and significantly less
bitterness, than the control (uninoculated) samples (Figure 4.33, A).
• Cocoa beans fermented with H.guilliermondii and I.orientalis, or
H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus resulted in
chocolate with significantly stronger doughy flavours than the samples
inoculated with S.cerevisiae and K.marxianus (Figure 4.33 A, B).
• Cocoa beans fermented with S.cerevisiae and K.marxianus, or
H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus resulted in
chocolate with significantly stronger grainy/nutskin flavours than the
samples inoculated with H.guilliermondii and I.orientalis (Figure 4.33 A, B).
The following comparisons were only valid within the Group 2 samples:
• Cocoa beans fermented with S.cerevisiae and K.marxianus produced
chocolate with significantly stronger nutskin flavours, and significantly
weaker doughy flavours, than the control samples (Figure 4.33 B).
• Cocoa beans fermented in a barrel without added culture, or in a barrel
with added H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus
produced chocolate with stronger acid, bitter and doughy flavours than
the samples fermented with S.cerevisiae and K.marxianus
(Figure 4.33 B).
When the chocolate samples were compared on the basis of fermentation in
barrels versus boxes, it was found that:
• Cocoa beans fermented in barrels gave chocolate with significantly
stronger cocoa flavour than beans fermented in boxes (Figure 4.33 A, B).
294
SweetnesJ3 —•—Box (control)

—■—Box (Hg & lo)
—■—Barrel (Control)
Nutskins/Grain Acidity —•—Barrel (Hg & Io)
■ Ghana
Bitterness’ ^
Doughy Cocoa
Brownfruit
Sweetness
—«—Box (control)
Acidity
■ Box (Sc & Km)
■ Barrel (Sc & Km)
—■—Barrel (All yeast)
—■—Ghana
Bitterness 3’ ^
Doughy' Cocoa3, b
Brownfruit
Fig. 4.33. Flavour characteristics of chocolate samples: A) Group 1 samples; B) Group 2 samples
“Hg & lo” - beans fermented with added Hanseniaspora guilliermondii and Issatchenkia orientalis',
“Sc & Km” - beans fermented with added Saccharomyces cerevisiae and Kluyveromyces marxianus;
“All yeast” - beans fermented with added H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus.
295
4.3.4.5 Aftertaste characteristics of the chocolate samples
Significant differences were found in the aftertaste profiles of the different

chocolate samples. 'Aftertaste' refers to the taste after the chocolate had been
swallowed. Differences in aftertaste correlated mainly with the use of different
mixtures of yeast as starter cultures. Overall, the Australian cocoa beans
oroduced chocolate with an aftertaste significantly less sweet, and more bitter,
doughy, cocoa, nutty and nutskin/ grainy than chocolate made from Ghanaian
aeans (Figure 4.34 A, B).
When comparing the aftertaste of the samples on the basis of the type of
moculum used, across both groups, the following differences were noted:
• Cocoa beans fermented with any defined yeast mixture produced

chocolate with a significantly less bitter and more cocoa aftertaste, than
any of the uninoculated (control) samples (Figure 4.34, A, B).
• Cocoa beans fermented with H.guilliermondii and Lorientalis produced

chocolate with a significantly less acid aftertaste than the control samples.
• Cocoa beans fermented with H.guilliermondii and Lorientalis produced

chocolate with an aftertaste that was significantly more doughy compared
to the beans fermented with S.cerevisiae and K.marxianus (Figure 4.34 A, B).
• Cocoa beans fermented with S.cerevisiae and K.marxianus, or

H.guilliermondii, Lorientalis, S.cerevisiae and K.marxmius produced
chocolate with a significantly stronger nutskin/grain aftertaste than the
samples fermented with H.guilliermondii and Lorientalis.
• In Group 1 only, cocoa beans from the uninoculated (control) barrel
fermentation produced chocolate with a significantly stronger spicy
aftertaste than the control beans from this group (Figure 4.34 A).
• In Group 2 only, cocoa beans inoculated with S.cerevisiae and K.marxianus

produced chocolates with a significantly less sweet and more bitter, cocoa
and grainy aftertaste than the control beans from this group (Figure 4.34 B).
'There were no significant differences between the aftertaste of the chocolate

samples, when compared on the basis of fermentation in barrels versus boxes.
296
—1»—Box (control)
—1■— Box (Hg & Io)
■ *■ Barrel (Control)
—♦—Barrel (Hg & Io)
—*— Ghana
Bitterness3
Cocoa5
Brownfruit
a, b
Sweetness
B 30 —•—Box (control)
—■*— Box (Sc & Km)
Nutskins/Grain' Acidity
—"—Barrel (Sc & Km)
—»— Barrel (All yeast)
—■—Ghana
Nutty Bitterness3, ^
Doughy Cocoa’b
Brownfruit5
Fig. 4.34. Aftertaste characteristics of chocolate samples: A) Group 1 samples; B) Group 2 samples
“Hg & Io” - beans fermented with added Hanseniaspora gui/liermondii and Issatchenkia orientalis;
“Sc & Km” - beans fermented with added Saccharomyces cerevisiae and Kluyveromyces marxianus\
“All yeast” - beans fermented with added H.gui/liermondii, I.orientalis, S.cerevisiae and K.marxianus.
297
4.3.4.6 Overall characterization of the chocolate samples tested
The quantitative, descriptive analysis of the chocolate samples indicated that

the flavour of the Australian cocoa beans compared favourably to the Ghana
beans in basic criteria such as cocoa flavour and astringency.
All cocoa beans fermented with mixtures of yeasts produced chocolate that was
acceptable when compared to chocolate made from Ghanaian beans.
However, the results also indicated that the Australian samples possessed their
own unique flavour characteristics. Furthermore, different yeast cultures
resulted in different tasting chocolates, each with distinct and describable
sensory characteristics.
A summary comparison of the chocolate samples is given in Table 4.29.

Table 4.29- Summary of the main sensory differences between the chocolate samples produced
from cocoa beans fermented with defined mixtures of yeasts
Distinctive Distinctive
Samples Colour Distinctive odours
Flavours Aftertaste
Strong cocoa and

bitter; faint doughy
very faint doughy Moderate cocoa
Control Brown flavour; boxes had
odour and bitter aftertaste
distinct brownfruit
flavour
Inoculated with Same as control; Moderate cocoa,

H.guilliermondii and Dark brown Mild doughy odour stronger doughy bitter and doughy
/. orientalis flavour aftertaste
same as control; no Strong cocoa and

Inoculated with
very faint doughy doughy flavour; bitter aftertaste;
S.cerevisiae and Dark brown
odour moderate grain/ mild grain/nutskin
K. marxianus
nutskin flavour aftertaste
Inoculated with Strong cocoa and Strong cocoa and

H. gu i 11ierm ondii, bitter; moderate bitter aftertaste;
Dark brown Mild doughy odour
I. orientalis, S.cerevisiae doughy flavour; faint mild doughy
and K. marxianus grain flavour aftertaste
Moderate cocoa
Strong cocoa and
Strong cocoa odour; and brown fruit
Ghana Red-brown brownfruit flavour;
no doughy odour aftertaste; weak
weak bitter flavour
bitter aftertaste
298
4.3.5 Analysis of breakdown of cocoa storage proteins using
SDS-PAGE
Proteins were extracted from cocoa bean samples taken during the pilot-scale
fermentations and analysed using SDS-PAGE ( Figures 4.35 and 4.36).
MW markers (kDa)
Fig. 4.35. SDS-PAGE of proteins extracted from cocoa beans sampled at different stages of
fermentation. Fermentations were performed in boxes and barrels, with or without added
cultures of H.guilliermondii & I.orientalis (“Hg & Io”). Lanes 1-9: 1, Composite sample Oh;
2, Box (uninoculated) 48h; 3, Box (Hg & Io) 48h; 4, Barrel (uninoculated) 48h;
5, Barrel (Hg & Io) 48h; 6, Box (uninoculated) 120h; 7, Box (Hg & lo) 96h;
8, Barrel (uninoculated) 120h; 9, Barrel (Hg & Io) 120h.
* 1 23456789 10 11 *
* MW markers (kDa)
Fig. 4.36. SDS-PAGE of proteins extracted from cocoa beans sampled at different stages of
fermentation. Fermentations were performed in boxes and barrels, with or without added
cultures of S.cerevisiae & K.marxianus (“Sc &Km”), and H.guilliermondii, I.orienta/is,
S.cerevisiae & K.marxianus (“All yeast”). Lanes 1-11: 1, Composite sample, Oh;
2, Box (uninoculated) 48h; 3, Box (Sc & Km) 48h; 4, Barrel (uninoculated) 48h;
5, Barrel (Sc & Km) 48h; 6, Barrel (All yeast) 48h; 7, Box (uninoculated) 120h;
8, Box (Sc & Km) 120h; 9, Barrel (uninoculated) 120h; 10, Barrel (Sc & Km) 120h;
11, Barrel (All yeast) 120h;
299
SDS-PAGE of extracts from Oh (unfermented) beans resulted in four major

bands at 46, 32, 22, and 14 kD. These bands corresponded to banding patterns
found in previous studies of cocoa proteins, with the 46 and 32 kD proteins
previously identified as vicilin-class proteins (Biehl et ah, 1982). The fading of
these bands, in the 48h and 120h samples, indicates that the vicilins were
hydrolysed as the fermentations proceeded. The process of hydrolysis of cocoa
vicilin proteins has been causally linked to the formation of chocolate flavour
precursors (Voigt and Biehl, 1995). The 22 kDa band, previously identified as an
albumin (Zak and Keeney, 1976), was not degraded to the same extent and was
still detected at the end (120h) of all fermentations. Since no additional bands
formed between 46 -14 kDa, in the 48h or 120h samples, it may be inferred that
the proteins were mainly hydrolysed to peptides with masses less than 14 kDa.
These results are consistent with cocoa protein gel banding patterns in the
literature (Buyukpamukcu et al. 2001). Both gels exhibited a common banding
pattern, suggesting that the breakdown of cocoa seed proteins occurred in a
similar fashion in all of the fermentations tested.
The gels also highlight some differences between the fermentations.
In Figure 4.35, the bands representing the vicilin proteins are weaker in lanes 5,
7 and 9. These lanes correspond to samples taken from the fermentations
inoculated with mixtures of H.guilliermondii and l.orientalis. The rapid
disappearance of these bands, compared to the lanes containing the control
samples, suggests that the addition of starter cultures accelerated the hydrolysis
of the vicilins. Of all samples tested, the fastest and most complete degradation
of cocoa storage proteins occurred in the barrel fermentation inoculated with
H.guilliermondii and I.orie?italis (Figure 4.35 lanes 5, 9).
The next fastest degradation occurred in the barrel fermentation inoculated

with H.guilliermondii, l.orientalis, S.cerevisiae and K.marxianus, as indicated by the
weakened bands in lanes 6 and 11 of Figure 4.36.
The addition of mixtures of S.cerevisiae and K.marxianus only slightly
accelerated the breakdown of cocoa seed proteins (Figure 4.36, lanes 3, 5, 8 &10)
compared to the control (Figure 4.36 lanes 2, 4, 7 &9).
While the breakdown of proteins may be indicative of flavour potential, sensory
analysis must be performed in order to confirm its effects. Too much proteolysis
may actually lead to loss of flavour potential (Hansen et al., 2000).
300
4.3.6 Antioxidant activity of cocoa beans fermented with defined

HPLC analyses were performed to estimate the level of procyanidins in

unfermented (Oh) and fermented (120h) cocoa beans. Figure 4.37 shows the
chromatogram produced by UV-Fluorescence analysis of polyphenolic
compounds extracted from unfermented Queensland (Mossman plantation)
cocoa beans.
Minutes
Fig. 4.37. HPLC-Fluorescence chromatogram of polyphenols extracted from unfermented (Oh)
Mossman cocoa beans, showing PI (catechin) monomers to P5 procyanidin polymers.
Standards of mixed polymeric procyanidins are not commercially available.

Therefore, the total procyanidins present in each sample were estimated using
standard solutions of a procyanidin monomer (in this case (-)-catechin). The
results are presented in Figures 4.38 and 4.39.
More procyanidins were detected in unfermented cocoa beans harvested in
August 2006, compared to the beans harvested in November 2006. This
suggested that there may be a seasonal effect on the polyphenol content of the
cocoa beans. As the fermentations progressed, the amount of procyanidins
decreased. Cocoa beans inoculated with H.guilliermondii and Lorientalis retained
a higher level of procyanidins compared to the control. Inoculation with
S.cerevisiae and K.marxianus, or H.guilliermondii, Lorientalis, S.cerevisiae and
K.marxianus did not significantly affect the level of procyanidins.
301
Oh 120h Box 96h Box 120h Barrel 120h Barrel

(control) (Hg & Io) (control) (Hg & lo)
Fig. 4.38 Total procyanidin content of polyphenol extracts from cocoa beans fermented with
H.guilliermondii and I.orientalis (“Hg & lo”) [beans harvested in August 2006].
** Total procyanidins are calculated as (-)-catechin equivalents.
P 40 -
120h Box 120h Box (Sc 120h Barrel 120h Barrel 120h Barrel
(control) & Km) (control) (Sc & Km) (All yeast)
Fig. 4.39. Total procyanidin content of polyphenol extracts from cocoa beans fermented with
S.cerevisiae & K.marxianus (“Sc & Km”), and H.guilliermondii, I.orientalis, S.cerevisiae &
K.marxianus (“All yeast”) [beans harvested in November 2006].
** Total procyanidins are calculated as (-)-catechin equivalents.
The antioxidant potential of the polyphenol extracts was also estimated using
the DPPF4 assay. The data from this assay are presented in Figures 4.40 and 4.41.
These results describe the relative capacity of the cocoa samples to absorb
oxygen radicals, and are expressed in units of Trolox equivalents (TE). The
antioxidant capacity of the cocoa beans decreased as the fermentations
proceeded. Cocoa beans that had been fermented with cultures of
H.guilliermondii and I.orientalis gave extracts with greater antioxidant activities
302
at compared to the control samples. Fermentation with S.cerevisiae and

K.marxianus also produced cocoa beans with slightly higher antioxidant
activities, compared to the controls (Figure 4.40). Finally, it was found that
cocoa beans fermented in barrels gave extracts with slightly higher antioxidant
capacities compared to those fermented in boxes (Figure 4.41).
3500
3000 ■
"§> 2500 ■
2000 ■
500 ■
Oh 120h Box 96h Box 120h Barrel 120h Barrel

(control) (Hg & Io) (control) (Hg & Io)
Fig. 4.40. Oxygen radical absorbance capacity (ORAC) of polyphenol extracts from cocoa
beans fermented with H.guildermondii and I.orientalis (“Hg & Io”).
3500
3000 -
5 2500 -
2000 -
500 -
120h Box 120h Box 120h Barrel 120h Barrel 120h Barrel
(control) (Sc & Km) (control) (Sc & Km) (All yeast)
Fig. 4.41. Oxygen radical absorbance capacity (ORAC) of polyphenol extracts from cocoa
beans fermented with S.cerevisiae & K.marxianus (“Sc & Km”), and H.guilliermondii,
I.orientalis, S.cerevisiae & K.marxianus (“All yeast”).
303
4.3.7 Analysis of volatile compounds from cocoa beans samples
using gas chromatography.

Extracts of roasted Australian and Ghanaian cocoa beans were prepared using
simultaneous steam-distillation-extraction and analysed by GC MS (Section
4.2.8). Approximately 90 volatile compounds were detected in the Ghanaian
sample, while an average of 120 were found in the Australian sample. Examples
of the chromatograms produced during these analyses are shown in Figures
4.42 and 4.43. Fifty peaks were large enough to be conclusively identified by
comparison of their mass spectral patterns with the NIST library, and their
Kovats retention indices with values found in the literature. Table 4.30 lists the
42 volatile compounds that were most frequently isolated from the cocoa bean
extracts. Of the compounds identified, 18 were selected as suitable for further
comparative analysis. These compounds were selected on the basis that they
occurred in all samples in reasonable quantities, and that they represented
compounds previously identified as significant contributors to cocoa flavour
(Bonvehi, 2005). Table 4.31 lists the identity of the chosen compounds, their
retention indices, descriptions of their odours, and the odour thresholds of each
as given in the literature. The numbers (1-19) assigned to each of the
compounds in Table 4.31 correspond to peaks labelled in Figures 4.41 and 4.42.
Table 4.32 gives the relative abundance of each compound, as calculated against
the peak area of the internal standard, dodecane.
Comparison of the chromatograms (Figures 4.42 and 4.43) reveals that different
peaks patterns were produced for the different samples depending on:
fermentation vessel, starter culture mixture, or whether the beans were from
Ghana or Australia. These differences are also apparent in the numerical data
shown in Table 4.32. Comparison of individual quantities or peak areas can be
time-consuming, and can generate misleading correlations (Howell et ah, 2006).
Therefore, to compare these differences the relative abundance data of 18
selected volatile compounds (Table 4.32) were subjected to principal component
analysis (PC A). PC A is a commonly used method for the simplification and
comparison of large data sets. The term "principal component" refers to an
artificial variable, generated by PCA, and representing multiple real variables
(in this case, the relative abundance of weighted groups of volatiles) (Stevens,
1986; Rencher, 1995; Yoo et al., 1998).
304
90
80
70
60
3
(IS)
30
9 13
20 .iUuJhuL. JLJL
15 20 25 30 35 40 45 50 55 60 65 70 75 80
Figure 4.42 - Chromatograms of volatile aroma compounds extracted from roasted cocoa beans
and analysed using GC-F1D; a. Cocoa beans fermented in a box (control); b. Cocoa beans
fermented in a box inoculated with H.guildermondii and I.orientalise c. Ghanaian cocoa beans
(industry std.). Peak numbers correspond to compounds listed in Tables 4.31 and 4.32
305
11 14,
5 6 7
5 6 7
Figure 4.43 - Chromatograms of volatile aroma compounds from roasted cocoa beans analysed
using GC-FID; a. Cocoa beans fermented in a barrel (control); b. Cocoa beans fermented in a
barrel inoculated with S.cerevisiae and K.marxianus\ c. Ghanaian cocoa beans (industry std.).
Peak numbers correspond to compounds listed in Tables 4.31 and 4.32
306
Table 4.30 - Predominant volatile organic compounds isolated from extracts of roasted cocoa
beans
Compound name
1-Phenyl-ethanone Acetic acid Hexanoic acid

2,3,3-Trimethy 1-2-butanol Benzaldehyde Isoamyl acetate
2,3,5-Trimethylpyrazine Benzeneacetaldehyde Isoamyl alcohol
2,5-Dimethylpyrazine Benzoic acid Isobutyl acetate
2,6-Dimethylpyrazine Benzonitrile Linalool
2-Acetylfuran Benzyl alcohol Phenylethyl Linalool oxide (trans furanoid)
2-Heptanone alcohol Methyl sulfone
2-lsobutyl-3-methoxy-pyrazine Bicyclohexan-2-ol acetate Methylpyrazine
2-Methyl-3-buten-2-ol Ethanol Pentanoic acid
2-Pentanol acetate Ethyl benzoate p-hydroxyacetophenone
2-Pentanone Ethyl palmitate Ethyl pentanoate Tetradecanoic acid
2-Phenylethanol Ethyl phenylacetate Tetram ethy 1 py raz i n e
3-Methoxy-2-butanol Ethyl Propianoate Tricyclooctane
4,6-dimethyl-2-hydroxypyrimidine Formic acid Trimethylthiazole
Table 4.31 - Volatile aroma compounds most commonly isolated from extracts of Australian
and Ghanaian cocoa beans using GC-MS.
Peak Compound name KI Sensory attributes Odour threshold

No. (gg/L) *
1 2 - pentanol acetate a 933 Fruity, banana, pearc -
2 Isoamyl acetate a 1117 Fruity, banana, pear, apple c -
3 Dodecane (internal standard) 1200 Internal Standard -
4 Isoamyl alcoholab 1211 Faint whisky, alcoholic c 5000
5 2 - Methylpyrazine ab 1251 Nutty, cocoa, chocolate, roasted nuts d 60
6 2,5 - Dimethylpyrazine ab 1297 Cocoa, roasted nuts d 1700
7 2,6 - Dimethylpyrazine ab 1304 Nutty, coffee d 9000
8 2,3,5 - Trimethylpyrazine ab 1377 Cocoa, roasted nuts, peanutd 1800
9 Tetramethylpyrazine ab 1438 Chocolate, cocoa, coffee d 1000
10 Linalool oxide (trans furanoid)a 1450 Refreshing, floral, citrus, tea d 5
11 Benzaldehyde ab 1493 Bitter d 350
12 Benzonitrile b 1583 Almond d -
13 Ethyl benzoate ab 1752 Fatty, floral, fruity d -
14 Pentanoic acid b 1785 Putrid, faecal, sweaty, rancid d 3000
15 Hexanoic acid b 1823 Pungent, sickening, rancid, sourd 3000
16 Ethyl phenylacetate ab 1885 Sweet, honeyd, resinous 650
17 Benzoic acid b 2100 Faint balsamic, urine-liked, resinous 3000
18 2 - Phenylethanol b 2200 Rose, honey, floral, fragrantd 1100
19 Benzeneacetaldehyde b 2300 Floral, green, sweet, grainy d 10
* Odour thresholds in aqueous solution (micrograms per litre) (Bonvehi, 2005).
a. Previously isolated from cocoa beans (Counet et al., 2002).
b. Previously isolated from cocoa beans (Bonvehi, 2005).
c. Sensory attribute as described in Stevenson (2008).
d. Sensory attribute as described in Bonvehi (2005).
KI - Kovats retention indices.
307
to
-a
3 VO OO O it, it, m ■— in -— cn m —■—1 _ cn m in O'
-C — CN <N cddddddd<Ndd”—^(N—”d
a
-52
O I « C
r-; on o 3- no cn no — cn
3- — NO CN CN C— ON “ I/O
& ^ o — 'N d d d d d d © CN CN d (N oc d I -2
Table 4.32 - Relative abundance of principal aroma compounds isolated from roasted cocoa bean samples
Control - uninoculated fermentation; Io & Hg - fermentation inoculated with a mixture of Hanseniaspora guilliertnondii and Lorn
CQ —
Sc & Km - fermentation inoculated with a mixture of Saccharomyces cerevisiae and Kluyveromyces marxianus; All yeast - fermenta
77 <
§
■o8
-oo 3
(N-rt-c^
—' —
— NDONfn —
cN^ddd—:dddddddd<NNod
tNONfNooTtr^r'NDmTfr-
3 no
m o
T3 o w
§ o CN X—N
to
15 _D CD E 2 t- *n ooofi^-m-ooos<x)0 in «n cn no no in
d 7: «NtToOOOOO(NO d — d — >n d
oE E CQ §
3 S
c > o
ts 2
c E
3
ofi x ^ m o ir> <n O' cn oo 3; NO oo — cn oo no cn ^ r-
3 d ~ *n d d d d d d ^ d d 2 o
•O
D
C/i
o oX ~E OO ^
• CN
o 00 cn _ >n no — I— cn no on no in e' NO d
inoculated with a mixture of H.guilliertnondii, l.orientalis, S.cerevisiae and K.marxianus.

c rr> <N ddd'—1 cn en _
1/J
03
O
3
Q. _
d 3C
D
C/5
oov£)Qjg3''3"'3'in3f'3‘r^r~' ^r^oocncNd'3;
C ofN^NdddddddddddddcnNod
o CQ o
8-
d
13 &g — r— o n ^ O' I— t— cn O' __ in no oo on r~- n
I no
o cvi<N<NTfddddd d d d CN d 3-' >n o
o CQ S
CN
to
3
OX) 'or.
3 x I 3- cn cn r*~i ■ — n) — rn cn sq rn
< d
m
— <N
O'
>n ddddddcN'ddddddood
no
X S m, in oo ^ no m, Tj- oq ni O^ oo — rn ON CN
° c
CQ § doo<N'“d°dddd(Ndd cn cn <d d
■§
s: in
"O Q D cn
c C CN ~oD>>
<D 3 N ‘n D
E o
a. a> 2 3 3
15 E 3 2
N
D •g
CD
o d o
JO N 3 Cl
CL N
>N D 23
In 2
_3
£ U 3 d _Q.
D
15
O -a
o. £ £
X. X >N >> D
_Q. X. .-3 N
o
§ s; 15 £ X. E
X>
.3
D - l .= g D D JC iD _D>N
o o
<3
O x. -C .1 .3
§ “ CL
<D E D Q 5 g E
2
o 15
E3 >, § s g
© 3
2 <A Ni | is*
.3 D D 8 £
CN Q (N CN (N h- I d QQ CQ W x d CN
308
Rather than comparing the quantities of specific, individual compounds, PCA

was used to compare the overall similarity or difference of the total profiles
generated by GC-MS (Yoo et ah, 1998; Howell et ah, 2006). PCA of the gas
chromatographic data showed that 57.2% of the variation between samples
could be explained by two principal components (PCI vs PC2). The first and
second principal components were plotted against one another (bi-plot) and are
shown in Figure 4.44. The bi-plot shows that PCA of the gas chromatography
data was able to separate the different samples into two distinct groups:
• Group 1 was comprised of the Australian cocoa bean samples.
PCA revealed that this group was characterized by relatively high
concentrations of the compounds: isoamyl alcohol, benzonitrile,
benzeneacetaldehyde and 2-phenylethanol (Table 4.32).
• Group 2 was comprised of the Ghanaian cocoa bean sample and was
characterized by high concentrations of the compounds: 2-pentanol
acetate, linalool oxide, ethyl benzoate and ethyl phenylacetate (Table 4.32).
In comparing the Australian samples with the Ghanaian sample, PCA
suggested that the following differences were significant in distinguishing
between the various samples:
• In the Australian samples the relative abundance of 2,5-methylpyrazine
(mean = 0.5), 2,6-methylpyrazine (mean = 0.4) and tetramethylpyrazine
(mean = 0.4) were equal to, or significantly greater than, the relative
abundances of these compounds in the Ghanaian sample (0.5, 0.1 and 0.1
respectively).
• The Ghanaian cocoa beans had significantly higher levels of fruity/floral
volatiles (2-pentanol acetate, linalool oxide, ethyl benzoate) compared to
the Australian samples.
• The relative abundance of 2-phenylethanol in extracts from the Australian
cocoa beans (mean value = 7.1), was significantly higher compared to the
extracts from the Ghanaian beans (value = 0.9).
Using PCA to compare the Australian cocoa bean samples, a distinct sub group
was observed. This was comprised of the beans fermented with added
H.guilliermondii and Lorientalis, and the beans fermented in an uninoculated
barrel (November 2006) (Figure 4.44 points 3, 5, 8).
309
This led to the observation that the samples inoculated with H.guilliermondii
and I.orientalis, had significantly higher levels of isoamyl alcohol, isoamyl
acetate and 2-phenylethanol compared to the corresponding control samples
(Table 4.32). While other differences may be noted in Table 4.32, these were not
determined to be significant in the overall comparison of total GC profiles, as
determined by PC A.
PCA 1 (32.3%)
Fig. 4.44. Principal component bi-plot showing grouping of Australian (Group 1: points 2-10)
and Ghanaian (Group 2: point 1) cocoa bean samples. Each data point represents a different
batch of cocoa beans:
1. Ghana reference beans;
2. Beans fermented in a box, uninoculated [Aug. 2006];
3. Beans fermented in a box, inoculated with H.guilliermondii and I.orientalis [Aug. 2006];
4. Beans fermented in a barrel, uninoculated [Aug. 2006];
5. Beans fermented in a barrel, inoculated with H.guilliermondii and I.orientalis [Aug. 2006];
6. Beans fermented in a box, uninoculated [Nov. 2006];
7. Beans fermented in a box, inoculated with S.cerevisiae and K.marxianus [Nov. 2006];
8. Beans fermented in a barrel, uninoculated [Nov. 2006];
9. Beans fermented in a barrel, inoculated with S.cerevisiae and K.marxianus [Nov. 2006];
10. Beans fermented in a barrel, inoculated with H.guilliermondii, I.orientalis, S.cerevisiae and
K.marxianus [Nov. 2006];
310
4.4 Discussion
Chapter 4 reports the use of selected yeast species as starter cultures for the
laboratory and semi-commercial (pilot) scale fermentation of cocoa beans
grown in Queensland, Australia. The development of appropriate starter
cultures has been a key criterion in the successful industrialization of other
traditional fermented foods, such as wine, cheese, bread, soy sauce and tempeh
(Wood, 1998; Fleet, 1999; Holzapfel, 2002; Steinkraus, 2004; Nout and Kiers,
2005). The main reasons why starters cultures are used include: greater control
of the microbial ecology of the fermentation, increased process efficiency,
greater control and predictability of product quality, and reduced variability in
quality between batches (Lucke et al., 1990; Schwan, 1998; Holzapfel, 2002;
Schwan and Wheals, 2004). The development and use of starter cultures was
identified as one way to achieve the practical goal of developing a controlled
process for the fermentation of Australian cocoa beans. However, in order to
develop starter cultures, a basic knowledge of the food to be fermented is
essential (Hesseltine and Wang, 1986; Lucke et al., 1990; Steinkraus, 2004). In
Chapter 3, the Australian cocoa bean fermentation was investigated, yielding
detailed information on the following: the composition of the pulp and seed of
fresh and fermented beans; the physical changes that occur during fermentation
(pH, temperature); qualitative and quantitative data on the microbial ecology of
the fermentation; key metabolic products of fermentation (eg. ethanol, lactic
and acetic acid); and the effects of these changes on the quality of the beans and
flavor of the chocolate prepared from them. This information led to
improvements to the process (fermentation in a barrel, optimisation of mixing
frequency) and, also, the identification of several species of yeasts as potential
candidates for development into starter cultures.
4.4.1 Selection of yeast isolates for use in mixed starter cultures
An important part of the work in this Chapter was the design of the starter
cultures for use in conducting cocoa fermentation. It was decided that these
starter cultures would be comprised of mixtures of yeast species. The decision
to use yeast as the initial culture was justified given their role as predominant
microorganisms in the early to mid stages of fermentation (Fowler, 1998;
Ardhana and Fleet, 2003; Jespersen et al., 2005; Nielsen, 2007b). Furthermore,
311
the decision to focus on yeast species was consistent with the approach of
previous research into the use of starter cultures for cocoa fermentation
(Sanchez et ah, 1985; Buamah et ah, 1997; Schwan, 1998; Dzogbefia et ah, 1999).
The particular yeast species selected were chosen because, either: a) they had
been used in previous studies (S.cerevisiae and K.marxianus) (Schwan, 1998), or
b) they were found to be predominant species during fermentations of
Australian cocoa beans - (H.guiltiermondii and I.orientalis) (Chapter 3).
Furthermore, Saccharomyces cerevisiae and Kluyveromyces marxianus can have
pectinolytic activities (Schwan et ah, 1997).
The isolates chosen for inclusion in these starter cultures were subjected to a
basic screening protocol as described in Section 4.2.3.1 of this Chapter. Due to
project constraints, however, the isolates were not characterised any further. In
particular, molecular characterisation of the isolates at the strain level was not
conducted. It was reasonably assumed that the strains inoculated into the bean
mass dominated and conducted the fermentation. Experiences in the wine
industry where starter cultures of yeasts are inoculated into grape juice
containing an indigenous microflora support this assumption (Fleet, 2003;
2007). For the present investigation, the focus was at the species level, and the
growth data clearly demonstrate quantitative dominance of the inoculated
species. As will be seen, this approach still revealed that the starter cultures
caused significant changes to the microbiology, chemistry and sensory
properties of the cocoa beans. However, circumstances can arise when the
inoculated strain may not dominate the fermentation, and contributions from
indigenous strains could be significant. Systematic molecular characterization
of strains isolated throughout the entire course of fermentation is needed to
determine such contributions, and such studies represent the next stage of
research to characterize the yeast ecology of cocoa bean fermentations
Work in Chapter 3 resulted in the assembly of a large collection (thousands) of

isolates. Only four isolates, however, were selected for use in this study. The
strains of H.guilliermondii, I.orientalis and S.cerevisiae were selected from the
yeast isolates obtained from earlier Australian cocoa fermentations. The strain
of K.marxianus used was from a culture collection (Australian Wine Research
Institute - Isolate CY69). In other fermented foods and beverages, process
efficiency and product quality can be significantly influenced by strain
312
properties (Pretorius, 2000; Holzapfel, 2002; Dias et aL, 2007). Therefore it is

recommended that more of the isolates obtained be characterized, to determine
differences between strains in their effects on fermentation and cocoa bean
flavour, and to properly estimate their potential as starter cultures.
4.4.2 Controlled laboratory fermentations to evaluate the use of mixed

cultures of yeast.
To evaluate the suitability of the three mixtures of yeasts, small-scale

fermentations of Australian cocoa beans were conducted in the laboratory,
under closely controlled conditions. During these laboratory fermentations,
changes were observed in the temperature, pH, microbiology and chemistry of
the cocoa beans. The effects of the different cultures on the quality and flavour
of the dried and fermented beans were also assessed. Although the focus of
these experiments was to evaluate the yeast cultures, a defined mixture of
bacterial cultures was also added, on the assumption that the preparation of the
pods (washing, surface sterilisation) would have eliminated the indigenous
bacterial microflora. These cultures consisted of two species of lactic acid
bacteria (Lactobacillus plantarum and Pediococcus ncidilactici) and one acetic acid
bacterium (Acetobacter pasteurianus). Based on previous work by Schwan (1998),
the bacterial cultures were inoculated at a level of 106-107 cfu/ g beans.
4.4.2.1 Changes in the temperature of bean mass
Previous attempts have been made to ferment small masses (<20 kg) of cocoa
beans (Rohan, 1963; Quesnel and Lopez, 1975). In these studies, it was noted
that the temperature of bean mass did not increase normally, often only
reaching 35 - 37°C. Correct bean temperature was maintained during small
scale fermentations by better insulation of vessels or conducting the
fermentations in an incubator (Rohan, 1963; Quesnel and Lopez, 1975). In the
present study, the laboratory fermentations were therefore placed in a
temperature-controlled incubator. The incubator was programmed to increase
in temperature during fermentation. The program chosen was designed to
mimic the increases observed in traditional fermentation, allowing the
successive growth of yeast (25 - 37°C; 0-72 h), followed by lactic and acetic acid
bacteria (37 - 45°C; 72 - 120h). The change in the temperature of the beans, in all
fermentations, closely followed the changing temperature of the incubator. In a
313
few cases (Figure 4.5 c, d), the heat produced by microbial metabolism was
sufficient to raise the temperature of the beans by a further 5 - 10°C.
4.4.2.2 Changes in the pH of cocoa pulp
In contrast to previous fermentations, where the pH of the Mossman cocoa pulp

was slightly lower than that of the South Johnstone cocoa pulp, here, both had
an initial pH of 3.9. During the laboratory fermentations, the pH of the cocoa
pulp increased to a final value of 4.3 - 4.4. This was similar to the pH values
observed during previous fermentations in Australia (4.2 - 4.9)(Chapter 3) and
Ghana (4.1 - 4.5) (Camu et al. 2008a), but was lower than those in Brazil (4.5 -
4.6) (Schwan et al., 1995) and Indonesia (4.8 - 4.9) (Ardhana and Fleet, 2003).
4.4.2.3 Changes in the populations of yeast and bacterial species
The microbiology of the control (non-inoculated) laboratory fermentations was

similar to that observed in previous, larger scale indigenous fermentations of
Queensland cocoa beans. The predominant yeasts were Hanseniaspora
guiltiermondii, Issatchenkia orientalis, and Saccharomyces cerevisiae, while the
predominant bacteria were Lactobacillus fermentum, Lactobacillus plautarum,
Acetobacter pasteurianus and a mixture of Bacillus spp. The initial levels of yeast
(103 cfu/ g) and bacteria (102 - 104 cfu/ g) were lower than observed during
previous Australian fermentations, probably as a result of the hygiene of the
laboratory, compared to the fermentary environment. These similarities
suggested that the small-scale laboratory fermentations provided a suitable
analogy to larger scale fermentations, with regards to microbial ecology. Having
a suitable model system allowed evaluation of the starter cultures under
controlled conditions prior to their use in the field.
In the inoculated fermentations, the added yeast cultures were all observed to
survive, grow and dominate the process. Microbiological testing at the
beginning (Oh) of the fermentations indicated that the inoculated species had
initial populations close to the target of 106 cfu/g. The higher initial populations
caused the inoculated yeast species to reach higher maximum populations, 24
hours earlier than in the control. The bacterial cultures added to the inoculated
fermentations, along with the yeast mixtures, were observed to survive and
grow. This fulfilled the primary goal of adding these cultures, which was to
314
ensure the presence of lactic and acetic acid bacteria that were presumed absent
as a result of pod surface sterilisation.
Inoculated H.guilliermondii and K.marxianus died off after about 24 h, probably
due to their sensitivity to ethanol. By contrast, I.orientalis and S.cerevisiae, with
their greater tolerance to temperature and ethanol, persisted much longer in the
fermentations to which they were added, only dying off after 96 -120 h.
The isolation of H.guilliermondii, I.orientalis and S.cerevisiae, from fermentations
into which they had not been inoculated was unexpected. Insufficient hygiene,
leading to cross contamination during mixing, is one possible explanation. This
possibility seems unlikely, given that the contamination was evident from Oh,
the initial load of these yeasts was lower than in the control, and that extreme
care was taken during preparation and mixing of these fermentations. Another,
more likely explanation, is that the cleaning and surface sterilisation method
used on the pods was insufficient in removing all indigenous flora. Finally, it is
possible that these yeasts may have been present in low numbers inside
apparently intact and healthy pods, as previously observed by Jespersen et al.
(2005). The presence and growth of these contaminating species must be
considered. In the fermentations inoculated with S.cerevisiae and K.marxianus,
the levels of indigenous H.guilliermondii and I.orientalis were low, and the
growth of these species was strongly suppressed by the added cultures. In the
fermentations inoculated with H.guilliermondii and I.orientalis, the populations
of indigenous S.cerevisiae were also suppressed, but to a lesser extent.
The inoculated lactic acid bacteria typically increased rapidly to maximum
populations of 107-108 cfu/ g by 24 h, and then gradually died off. In the
inoculated fermentations using South Johnstone beans, P.acidilactici was the
predominant lactic acid bacteria, while in fermentations using Mossman beans,
Lplantarum dominated. In the fermentations using Mossman beans, the initial
population of P.acidilactici was below the expected dose of 106 cfu/g, and
declined rapidly. This suggests that the culture was in decline before addition,
possibly as a result of phage contamination at some stage during its
preparation. A.pasteurianus grew in a similar fashion in all fermentations to
which it was added, and tended to reach maximum populations 24 - 48 h after
the lactic acid bacteria.
315
This sequence of microbial growth corresponded to the physico-chemical

changes observed: at the beginning of fermentation, the lower pH and oxygen
availability was most suited to the growth of lactic acid bacteria. As the yeasts
degraded the pulp and produced ethanol, conditions became more conducive
to the growth of A.pasteurianus.
Bacillus spp. were also detected in the inoculated fermentations, typically

during the last 48 hours of fermentation, at populations of 102-103 cfu/g. As in
previous Australian fermentations, the appearance of Bacillus spp. in the
laboratory fermentations coincided with the death of the lactic acid bacteria.
This supports the hypothesis that lactic acid bacteria could inhibit the growth of
Bacillus spp. during cocoa fermentation, in a similar manner observed in other
fermented foods (Vandenbergh, 1993, Nout and Kiers, 2005; Kostinek et al.
2005).
Although it was not added as part of the inoculum, populations of L.fermentum

were detected in the fermentation using Mossman beans. As with the yeast
contaminants, the presence of non-inoculated flora in the inoculated
fermentations suggests that the procedure to wash and surface sterilize the
pods was insufficient. The growth of L.fermentum to significant levels (105 - 107
cfu/ g) did not appear to interfere with the growth of the inoculated cultures. At
populations of 106 cfu/g, L.fermentum would be expected to contribute
significant metabolic activity to the fermentation, and this needs to be taken
into account when evaluating the fermentations.
4.4.2.4 Changes in the chemistry of fermenting beans

Chemical testing showed that the beans (pulp and seeds) used in the laboratory
fermentations had levels of sugars and organic acids, comparable to previous
batches of Australian cocoa beans (Chapter 3). During all of the fermentations,
some common changes in chemistry were observed.
Firstly, glucose and fructose in the pulp were utilised. In the pulp of the control
fermentations, glucose and fructose were incompletely utilised with traces
remaining at 144h (0.6% and 0.8% respectively). By contrast, the sugars were
completely utilised in the inoculated fermentations.
The results suggested that ethanol was produced during fermentation. Since
only samples from the beginning (Oh) and end (144h) of fermentation were
316
tested, the maximum level of ethanol produced was not directly determined.
The presence of higher concentrations of ethanol earlier in the fermentation was
inferred from the presence of 0.5-1% ethanol at the end of fermentation, and
from the observation that the fermentations smelt distinctly alcoholic between
24- 72h. Observations made during previous fermentations suggested that a
residual level of 0.5 -1.0% ethanol at the end of fermentation corresponded to a
maximum level of 5.0 - 7.0% ethanol between 24 - 72h. Due to these limitations,
however, it is not possible to determine whether the addition of different
cultures affected the level of ethanol production during fermentation.
During fermentation, citric acid in the pulp was metabolised, while lactic and
acetic acid were produced in the pulp, and diffused into the seeds.
More lactic and acetic acid was produced in the inoculated fermentations,
compared to the controls. This observation corresponds to the higher levels of
lactic and acetic acid bacteria observed in the inoculated fermentations
compared to the controls. Fermentation with mixtures of S.cerevisiae and
K.marxianus, and H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus
resulted in significantly higher levels of acetic acid, compared with the controls
or fermentations using H.guilliermondii and I.orientalis. This was probably due to
increased ethanol production, leading to in increased acetic acid production.
This agrees with chemistry data from Chapter 3, and the literature (Forsyth and
Quesnel, 1963; Carr et al., 1980; Ardhana and Fleet, 2003) which note a positive
correlation between ethanol and acetic acid production. Acetic acid production
may have been further stimulated by increased aeration due to mixing every 24
hours and the small size of these fermentations (increased surface area: volume
ratio).
4.4.2.5 The quality and sensory characteristics of cocoa beans
The uninoculated (control) laboratory fermentations ultimately produced beans

that were largely acceptable, as judged by the cut test and testing of other
parameters commonly used in industry. All samples had a very high proportion
of fully brown beans (>90%). This was probably due to the slower than usual
sun-drying, which took about 12 days. By comparison, sun drying at the North
Queensland site usually took 6-7 days.
317
The inoculated laboratory fermentations gave no significant differences in any

of the quality parameters tested, apart from pH. While the pH of the seeds was
not monitored during fermentation, the pH of the dried nibs was tested, and
provided a guide as to the final acidity levels inside the seeds at the end of
fermentation and drying. These indicated that the beans fermented with
S.cerevisiae and K.marxianus, and H.guilliermondii, I.orientalis, S.cerevisiae and
K.marxianus had significantly lower pH values (~ pH 4.8) compared to the
controls, or the fermentations inoculated with H.guilliermondii and I.orientalis
(~pH 5.3). This corresponded to the greater amount of acetic acid produced in
the fermentations inoculated with S.cerevisiae and K.marxianus.
During quality testing, it was noted that all of the beans produced by laboratory
fermentation had a weak cocoa aroma. Nevertheless, chocolate made from
beans from the control fermentations, and the fermentations inoculated with
H. guilliermondii and I.orientalis, were considered acceptable when compared to
chocolate made from Ghanaian beans. By contrast, chocolates made from beans
fermented with S.cerevisiae and K.marxianus, or H.guilliermondii, I.orientalis,
S.cerevisiae and K.marxianus were unacceptable. During sensory evaluation of
these latter samples, many panelists commented on the weak chocolate flavour
and sourness of these samples. The observations agreed with the organic acid
and nib pH data, and support the assertion that inoculation with S.cerevisiae and
K.marxianus resulted in more acetic acid. This acidity was probably also
responsible for the weak chocolate flavour, since high levels of acetic acid can
inhibit the production of chocolate flavour precursors (Chick, 1981; Lopez, 1983;
Biehl, 1984; Voigt and Biehl, 1995).
Triangle tests demonstrated that chocolate prepared from the beans fermented
using cultures of H.guilliermondii and I.orientalis tasted similar to that prepared
from beans in control fermentations. In contrast, beans fermented using cultures
of S.cerevisiae and K.marxianus, and H.guilliermondii, I.orientalis, S.cerevisiae and
K.marxianus gave chocolate that tasted significantly different from chocolate
made from control beans or those fermented, with H.guilliermondii and
I. orientalis. It seems likely that these differences were mostly due to the acidity
or sourness of the beans. Subtle differences in flavour may also have been
caused by volatile metabolites produced by the yeast and bacterial species. This
was supported by the qualitative observations of some panelists, who noted
318
that chocolate made from beans fermented with S.cerevisiae and K.marxianus
had fruity and nutty flavour notes that were weak or absent in the other
samples. The same panelists also noted that chocolate made from beans
inoculated with H.guilliermondii and I.orientalis had subtle, yeasty flavour notes.
Overall, it was found that the quality of the beans produced by the laboratory
fermentations was lower than expected. Further consideration of the methods
used indicated that this was probably linked to the mixing frequency, since
these laboratory fermentations were mixed every 24 hours, rather than every 48
hours as recommended at the end of Chapter 3. In this case, the small volumes
and mixing were justified by a need to obtain uniform samples for analyses
4.4.2.6 Implications of the laboratory fermentations
The microbiology and chemistry of the uninoculated laboratory fermentations,

were similar to the microbiology and chemistry of previous larger scale
indigenous fermentations (Chapter 3). These similarities suggested that the
small-scale laboratory fermentations provided a suitable model system in
which to study the growth and survival of starter cultures for cocoa
fermentation. Similar systems have previously been used to research other
aspects of cocoa fermentation such as the effects of cultivar, pod storage, pulp
removal and acid production (Rohan, 1963; Quesnel and Lopez, 1975; Carr et
al., 1979). The main advantage of the laboratory fermentations was that they
allowed trials to be performed at a time when only small quantities of beans
were available.
The primary aim of the laboratory fermentations was to evaluate three different
mixtures of yeasts for subsequent use as starter cultures in larger pilot-scale
fermentations (Hanseniaspora guilliermondii and Issatchenkia orientalis,
Saccharomyces cerevisiae and Kluyveromyces marxianus, and H.guilliermondii,
I.orientalis, S.cerevisiae and K.marxianus). The yeast mixtures consistently
demonstrated both survival and growth when inoculated into freshly obtained
cocoa beans. This suggested that the selected cultures were suitably adapted to
the conditions of fresh cocoa beans (low pH, moderate temperature, sugars). It
also suggested that the initial inoculum dose of 106 cfu/g yeasts was
appropriate for this application. This agreed with previous research into the use
of starter cultures for cocoa fermentation (Sanchez et al., 1985; Schwan, 1998). A
correct inoculation rate is essential in order for the inoculated species to have
319
good survival capacity and be able to compete with and dominate the
indigenous flora (Sanchez et ah, 1985; Fleet, 2003; Steinkraus, 2004). Defined
mixtures of bacterial cultures were also added to the inoculated fermentations.
Flowever, unlike the study by Schwan (1998), a close study of their effects on
fermentation was not the aim of these experiments. Nevertheless the results of
the current study suggested that the species selected (L.plantarum, P.acidilactici
and A.pasteurianus) were generally appropriate and behaved similarly to those
found in traditional/indigenous fermentations. However, the data also
suggested that the inoculation level (106- 107cfu/g) was too high. In particular,
the inoculated lactic acid bacteria grew more rapidly and reached higher
maximum populations in the inoculated fermentations compared to the
controls. Since the intended purpose of these bacteria was to simulate the
missing indigenous bacterial species, an inoculation rate closer to the
populations usually found at the beginning of fermentation (e.g. 103-104 cfu/ g)
may have been more appropriate.
A secondary aim of these laboratory fermentations was to determine whether

the use of different mixtures of yeasts caused any differences in the chemistry,
quality and flavour of the beans. It was found that the beans fermented with a
mixture of H.guillermondii and I.orientalis were similar to the control in their
chemistry, quality and flavour. In contrast, fermentation with a mixture
Saccharomyces cerevisiae and Kluyveromyces marxianus, or H.guilliermondii,
I.orientalis, S.cerevisiae and K.marxianus gave beans with significantly different
properties. Although the quality (pH, cut test scores and flavour) of the
chocolate prepared from beans produced by the latter two mixtures was lower,
this was tentatively attributed to mixing frequency and aeration of the
fermentation, leading to elevated levels of acetic acid.
Given the stringent measures to ensure the sterility of the cocoa beans (washing
in 0.1% hypochlorite, followed by surface sterilization with 70% ethanol), the
detection of contaminating species of bacteria and yeast in the inoculated
fermentations was unexpected. In the study by Schwan (1998), the same
measures were reported to be effective at eliminating indigenous species of
yeast and bacteria, giving fresh beans with negligible contamination (<10 cfu/
g). Our experiments, in a clean laboratory environment, were not able to
reproduce these conditions. This, at least, suggested that the elimination of
320
natural microbial contamination might not be consistently or predictably

achieved during commercial processing of cocoa grown in Australia. Also,
while centralised, industrialised processing would allow for the maintenance of
clean-rooms and the sterilization of equipment, such measures would require
intensive training of workers, increased capital investment and higher running
costs. Furthermore, the removal of the indigenous microflora, even if feasible,
may not be desirable. In many food fermentations, even those where starter
cultures are added, differences in the background flora are important in
differentiating the flavour of products from different regions or producers
(Lucke et al., 1990; Wood et al., 1998; Holzapfel, 2002; Steinkraus et al., 2004).
For these reasons, it was decided that the pilot-scale fermentations should be
performed under non-sterile conditions, to test the ability of the yeast cultures
to influence the fermentations, even in the presence of indigenous yeast and
bacterial species. Such an approach is routinely used in the fermentation of beer,
wine, cheese, and vinegar (Wood, 1998; Fleet, 2003; Vanderhaegen et al. 2003).
4.4.3 Pilot-scale fermentations

Having demonstrated their ability to grow in cocoa beans under laboratory
conditions, the starter culture mixtures were used to conduct larger scale (50 kg)
fermentations. The goal of these trials was to determine whether the starter
cultures could produce high quality cocoa beans, under pilot-scale conditions,
and to identify any advantages associated with their use. In particular, the
ability for different mixtures of yeast to produce beans with different flavour
characteristics was examined. It was envisaged that the results of these
experiments would provide the foundations for an industrialised process for
the large scale, commercial processing of cocoa beans grown in Queensland,
Australia. The pilot-scale fermentations were conducted in both barrels and
boxes, as recommended at the end of Chapter 3, in order to provide a further
comparison of the traditional vessel (box) to the novel vessel (barrel). The
starter cultures used in the pilot-scale fermentations were prepared by a
different method to that used in the laboratory fermentations. Facilities for the
preparation of large volumes of liquid culture were not available at the South
Johnstone site. Therefore, the cultures were prepared on solid media. This
method gave inoculum biomass similar to that produced by liquid media
321
culture, and has been used by several other researchers in the past (Subden et
ah, 1987; Kushnirov, 2000).
4.4.3.1 Changes in temperature
In all of the pilot-scale fermentations, the temperature of the beans increased

from 25°C at Oh, to maxima of 40-45°C around 72h. These temperatures were
above the 40°C sufficient to trigger and support the formation of flavour
precursors (Biehl et ah, 1977; Lopez and Dimick, 1995). While some minor
temperature variations were observed between the different fermentations,
these could not be attributed to either the addition of different yeast cultures, or
the use of boxes versus barrels. These results support previous research by
Sanchez et al. (1985) and Schwan (1998) who found that addition of starter
cultures did not significantly affect the temperature profile of fermentation.
Rather, fluctuations in fermentation temperatures were probably more strongly
influenced by bean maturity (Barel, 1987; Ardhana and Fleet, 2003; Portillo et
al., 2005), fermentation mass (Bridgland, 1959; Forsyth and Quesnel, 1963;
Quesnel and Lopez, 1975; Hashim, 1998b) and turning frequency (Rohan, 1963;
Quesnel, 1968; Senanayake, 1997; Camu, 2008a, b).
4.4.3.2 Changes in the pH of the cocoa pulp and seeds.
The initial pH values of the pulp (3.75) and beans (6.2) were similar to those
reported in Chapter 3 (pulp pH 3.5 - 3.9; seed pH 6.1-6.2) and the literature
(Ardhana and Fleet, 2003; Schwan and Wheals, 2004; Afoakwa et al., 2008).
During the pilot-scale fermentations, the pH of the pulp increased while the pH
of the seed decreased, following a similar pattern to that observed in Chapter 3.
In general, the final pH of the pulp was significantly higher for these pilot-scale
fermentations (mean value 4.8), compared to the final pH of the pulp in
previous fermentations (mean value 4.3). The chemistry data will show that this
was a result of decreased acetic acid production due to reduced mixing and
aeration. The formation of chocolate flavour precursors best occurs when the
seed pH drops below 6.0 (Biehl et al., 1985; Guilloteau et al., 2005). These
conditions were met in all pilot-scale fermentations.
Fermentation of cocoa beans with H.guilliermondii and I.orientalis, in either box
or barrel vessels, resulted in higher final pulp pH (4.7) compared to the controls
(3.8). A similar effect was observed for beans fermented in a barrel with a
322
mixture of S.cerevisiae and K.marxianus. By contrast, beans fermented in a box

with a mixture of S.cerevisiae and K.marxianus had lower pulp pH (4.6) and
higher seed pH (5.8) than the beans from the uninoculated (control) box (4.8
and 5.0 respectively). The addition of a mixture of H.guilliermondii, l.orientalis,
S.cerevisiae and K.marxianus did not have a significant effect on the pH of the
pulp or seed, compared to the control. Similarly, fermentation in boxes versus
barrels did cause significant differences in pulp or seed pH values. Each of these
differences in pH can be linked to observed differences in the concentrations of
organic acids detected in the different fermentations. This will be discussed
later in Section 4.4.3.6.
Given the ease with which pH may be monitored, surprisingly few studies have
examined the effects of added cultures on the pH of cocoa fermentations.
Sanchez et al. (1985) and Samah et al.(1992) also found that addition of
K.marxianus or S.cerevisiae cultures led to a lower final pH in cocoa pulp, but did
not report the effect on the pH of the seed material. From this study, it is
strongly recommended that the pH of the pulp and seed material always be
measured separately.
4.4.3.3 Changes in the appearance and aroma of the fermenting cocoa beans
Inoculation with mixtures of H.guilliermondii and l.orientalis affected the internal

and external appearance, and aroma, of cocoa beans during fermentation. The
inoculated beans became swollen, browner in colour and had lost their
adhering pulp faster than the control beans. These effects may have caused by
enzymes produced by the inoculated yeast species. When samples of the
inoculated beans were cut open, between 48 - 72 h, a brown exudate was
observed. The swelling of the beans, and the presence of this exudate only
appeared in the control beans between 96 - 120h. Cocoa beans fermented with
H.guilliermondii and l.orientalis had faint yeasty aroma, not present in the
control, between 24-72 h. There was no apparent acceleration in browning
(internal or external) of the beans inoculated with S.cerevisiae and K.marxianus,
or H.guillierifiondii, l.orientalis, S.cerevisiae and K.marxianus, compared to the
controls. It was observed, however, that cocoa beans inoculated with S.cerevisiae
and K.marxianus had a pineapple-estery aroma between 24-72h that was not
present in the control fermentations. During all inoculated fermentations, a
vinegar-like aroma was detected 24 h earlier than in the control fermentations.
323
This corresponded to the earlier detection of acetic acid in these fermentations.

These results were similar to those of Sanchez et al., (1985) and Buamah et al.
(1997) who observed that addition of cultures led to a greater proportion of
cocoa pulp being lost as sweatings. This supports the notion that strains of these
yeasts may produce pectinases. It was also noticed that beans in box
fermentations appeared more swollen than those in the barrels. This swelling
has been suggested to be one indicator of successful fermentation (Rohan, 1963;
Wood and Lass, 1985; Hashim, 1998), and is caused by the uptake of water by
the cocoa seeds (Biehl, 1982b).
4.4.3.4 Changes in the populations of yeast and bacterial species
4.4.3.4 A Control (uninoculated fermentations)
The microbial ecology of the control pilot-scale fermentations followed a similar

pattern to that described for the heap fermentation in Chapter 3, and to the
laboratory fermentations described earlier. H.guilliermondii dominated the first
24 h of fermentation, reaching populations of 107 cfu/ g before dying off and
being replaced by 107 - 108 cfu/g of I.orientalis and S.cerevisiae, both of which
persisted until the end of fermentation. The lactic and acetic acid bacteria grew
from the beginning of fermentation, and reached maximum populations
between 24-72 hours. In all control fermentations, the main species of lactic
acid bacteria isolated was Lactobacillus fermentum. This was in contrast to the
baseline established in Chapter 3, and to the laboratory-scale fermentations,
where the most prominent species of lactic acid bacteria was Lactobacillus
plantarum. Acetobacter pasteurianus was the most frequently isolated species of
acetic acid bacteria, and grew in a pattern similar to the baseline established in
Chapter 3 and the laboratory fermentations. Acetobacter tropicalis and
Gluconobacter oxydans were also isolated. In contrast to Chapter 3, where it grew
in the first 48-72 h of fermentation, G.oxydans typically grew towards the end of
fermentation and had lower populations than A.pasteurianus. In the
fermentations from which it was isolated, A.tropicalis grew more rapidly, and to
higher maximum populations than A.pasteurianus. These pilot-scale
fermentations were the first time A.tropicalis was isolated from Australian cocoa
beans. The unexpected appearance of this species may have been due to low
level infection of pods, since A.tropicalis has been reported to infect and spoil
fruits including coconuts, figs and limes (Cleenwerck, 2002). Other possible
324
sources of this species include insect hosts, particularly Drosophila spp.(Gilbert,

1980), aerial contamination (Schwan and Wheals, 2004) or from the fermentary
environment (Roelofsen, 1958; Ostovar and Keeney, 1973). The decreased
mixing frequency (48 hourly, from 24 hourly in Chapter 3), did not significantly
reduce the populations of acetic acid bacteria detected, but did affect the
production of acetic acid, as discussed in the next section.
The control fermentations demonstrated some differences between box or barrel
vessels with respect to the growth of yeast and bacterial species. Firstly,
H.guilliermondii reached higher populations and survived for longer in the
barrels compared to the boxes. Second, l.orientalis dominated over S.cerevisiae to
a greater extent in the barrels compared to the boxes. Thirdly, populations of
lactic acid bacteria found during the box fermentations differed from those
found during barrel fermentation, although no common pattern emerged.
Finally, higher populations of acetic acid bacteria were isolated from barrel
fermentations compared to box fermentations. This research (Chapters 3 and 4)
represents the first time that cocoa bean fermentations in box and barrel vessels
have been compared, and that the effects on the microbial ecology have been
reported. However, these results were analogous to those of Nielsen et al.
(2007b) who observed that the microbial ecology of heap fermentations was
significantly different to the ecology of tray fermentations. Similarly, earlier
studies by Carr et al. (1980) observed differences between the ecology of heap
and box fermentations. The most common explanation for these data is that the
different fermentation configurations have different levels of oxygenation that
affect the development of the yeast and bacterial species (Carr et al. 1980;
Ardhana, 1990; Camu et al, 2008a,b). However, oxygen concentration is not the
only variable at work, and there is a clear need for more research to answer this
question more precisely. More research could be performed into the effect of
fermentation vessel on the following: the homogeneity of the bean-mass, the
rate of diffusion of acetic acid and ethanol from pulp to seed, and the retention
and transfer of heat within the bean-mass.
4.4.3.4 B Inoculated fermentations
The data indicated that inoculation with mixtures of yeast influenced the
growth and survival of both the inoculated yeast, and indigenous species of
yeast and bacteria.
325
This study is probably the first to report cocoa fermentation conducted with
added cultures of Lorientalis and H.guilliermondii. It is also the first time that
inhibition of S.cerevisiae by inoculated species has been reported in cocoa
fermentations. In the fermentations inoculated with H.guilliermondii and
Lorientalis, these species reached maximum populations of 107 - 108 cfu/g after
only 24 hours, exhibiting much earlier dominance of the microflora than in the
control. It was also found that indigenous S.cerevisiae was strongly inhibited in
these fermentations. This was consistent with a similar observation made
during the laboratory fermentations. The potential for Hanseniaspora spp. to
interrupt the growth of S.cerevisiae has long been recognised in some wine
fermentations (Bisson, 1999). Fermentation with H.guilliermondii and Lorientalis
led to increased growth of A.pasteurianus, but decreased growth of lactic acid
bacteria. Suppression of the lactic acid bacteria was greater in the box
fermentation than in the barrel, with no (<102 cfu/ g) species of lactic acid
bacteria detected in the latter case. The production of lactic acid in these
fermentations, described later, suggests that low populations of lactic acid
bacteria were present, but were not detected. Possibly, their populations were
below the detection limit of the method (<102 cfu/g between 24-120h, due to
dilutions), or the cells may have entered a viable-but-not-culturable (VNBC)
state. This inhibition was unexpected, particularly since Hounhouigan et al.
(1999) found that Lorientalis stimulated the growth of L.fermentum. Possible
mechanisms for the stimulation or inhibition of the bacterial species will be
mentioned later.
Inoculation of beans with S.cerevisiae and K.marxianus allowed these species to

establish high populations more rapidly than the indigenous yeast species
(H.guilliermondii and Lorientalis). The growth of H.guilliermondii and Lorientalis
was only slightly suppressed, however, and they grew in a similar pattern to
that in the controls. This was in contrast to the laboratory fermentations, where
inoculation with S.cerevisiae and K.marxianus did suppress the indigenous
yeasts. Nevertheless, the earlier predominance of these yeasts after inoculation,
allowed them 24 hours in which to influence the fermentation, prior to the
growth of the indigenous yeast. S.cerevisiae and K.marxianus grew and survived
much better in this study compared to the studies by Sanchez et al. (1985) and
Schwan (1998), where the inoculated species died off after 24-48 hours.
326
Fermentations with S.cerevisiae and K.marxianus stimulated the growth of

Acetobacter spp. and L.plantarum, but inhibited the growth of L.fermentum (Figure
4.17). Jespersen (2003) previously noted that some strains of S.cerevisiae
encouraged the growth of lactic acid bacteria in indigenous, African fermented
foods, but this observation has not been reported in cocoa fermentation.
In the fermentation inoculated with H.guilliermondii, I.orientalis, S.cerevisiae and

K. marxianus, all species rapidly grew to a maximum population of 107 cfu/g
after only 24h. As the fermentation proceeded, K.marxianus slowly died off, due
to increased temperature and ethanol concentration. The other, heat and ethanol
resistant, species remained present at 106-107 cfu/g until the end of
fermentation. This fermentation was the first recorded in which viable
populations of H.guilliermondii were isolated from the end of fermentation. It
also represented the first cocoa bean fermentation deliberately inoculated with a
mixture of more than 3 yeast species. Fermentation with H.guilliermondii,
I.orientalis, S.cerevisiae and K.marxianus affected the bacterial microflora in the
same manner as the fermentations with S.cerevisiae and K.marxianus. This
general growth pattern was similar to that observed earlier during the
laboratory fermentations, except that all yeast species persisted for longer in the
pilot-scale fermentation. The lower maximum temperature of the pilot
fermentation (45°C), compared to the laboratory fermentation (50°C), likely
caused this to occur.
There were large differences in the bacterial species isolated (diversity and
populations) from the inoculated pilot-scale fermentations compared to the
laboratory fermentations. These differences were to be expected, given that the
laboratory fermentations were inoculated with 106cfu/ g each of A.pasteurianus,
L. plantarum and P.acidilactici. In contrast, the bacteria isolated from the pilot-
scale fermentations were indigenous species, typically present at 103-104 cfu/g
cfu at the beginning of fermentation.
4.3.4.4 C Microbial interactions during inoculated pilot-scale fermentations
While Sanchez et al. (1985) made some observations regarding yeast - yeast
interactions, there are no data available in the literature regarding yeast-bacteria
interactions during fermentation with starter cultures. This study has provided
some data on this phenomenon. The following possibilities were identified as
causes for the inhibition of L.fermentum by the added yeasts: (a) rapid utilisation
327
of pulp sugars by the inoculated yeast, reducing the amount of substrates

available for L.fermentum, (b) increased ethanol production by the inoculated
yeast leading to increased death or inhibition of L.fermentum, and (c) the
production of sulphur dioxide, fatty acids, proteins or peptides by the
inoculated yeast, having an inhibitory effect on the L.fermentum. Although there
is a need to confirm the molecular basis of this inhibition in cocoa fermentation,
it is known that yeast can inhibit growth of lactic acid bacteria during wine
fermentation by the mechanisms proposed above (Fornachon, 1968; Dick et ah,
1992; Markides, 1993; Capucho and San Romao, 1994; Lonvaud-Funel et ah,
1998; Guilloux-Benatier et ah, 1998; Alexandre et ah, 2004).
The more rapid growth of the acetic acid bacteria (AAB), and L.plantarum
observed in these pilot-scale, inoculated fermentations, was probably caused
by: (a) increased ethanol production by the inoculated yeast, providing more
substrate for the growth and metabolism of the bacteria, (b) increased pectinase
production by the inoculated yeast, leading to faster loss of pulp and greater
aeration of the bean mass, and (c) the release of nutrients by autolysis of the
yeast cells, stimulating the growth of a range of bacteria. Although the literature
does not adequately discuss microbial interactions during inoculated cocoa
fermentations, it does provide general support for points (a) and (b) in noting
that the AAB tend to increase in population in the presence of aeration and
ethanol (Lopez and Quesnel, 1973; Carr et al., 1979,1980; Dougan et al., 1981;
Camu et al., 2008b). With regards to point (c), the effect of yeast autolysis in
stimulating bacterial growth has long been recognised in the wine industry, but
has not been investigated with respect to cocoa fermentation. (Fornachon, 1968;
Guilloux-Benatier et al., 1993; Patynowski et al., 2002 ; Sponholz, 1993;
Fugelsang, 1997; Fleet, 1998; Du Toit and Pretorius, 2000; Crouigneau et al.,
2000). The ability for yeast starter cultures to influence the growth of bacterial
species could lead to improved control of the fermentation, without the need
for adding bacterial cultures. Clearly, more information is needed regarding the
interactions between inoculated yeast cultures and the bacterial flora of cocoa
fermentations. In particular, more research should be undertaken to determine
the molecular basis for these interactions, beyond the simplistic explanations
proposed.
328
4.3.4.4 D Auxiliary microbiological observations
The consistent observation of an unidentified basidiomycete in the early stages

of fermentation of Australian cocoa is also worthy of further investigation. Its
colony and cellular morphology matched that of a similar filamentous fungus
isolated by Ardhana (1992) from Indonesian cocoa fermentations. It was
suggested that this species, along with other filamentous fungi may play a role
in the early stages of fermentation (Ardhana and Fleet, 2003).
It is also recommended that further attention be paid to the microbiology of
drying. During this study, a range of yeast, filamentous fungi and Bacillus spp.
were isolated during drying, often at high (106-108) populations. There is little
information available on the potential for microorganisms to contribute to the
flavour and quality of cocoa beans during drying. The concentration of
mycotoxins in cocoa beans is of public health concern (Amezqueta t al, 2004,
2005; Bonvehi, 2004; Mounjouenpou et al., 2008), and fungal growth on beans
during drying and subsequent storage requires more systematic research.
During the course of this research, it was incidentally observed that bean
inoculated with H.guillermondii and l.orientalis were less prone to mould growth
during drying. The potential for Picliia and Hanseniaspora spp. to inhibit mould
growth in coffee beans has been reported by Masoud and Kaltoft (2006). It may
be that mycotoxin production in cocoa beans could be prevented through the
use of these yeast, as part of starter cultures.
4.4.3.5 Changes in the chemistry of fermenting beans
Chemical changes during the uninoculated (control) pilot-scale fermentations
followed the patterns observed in Chapter 3 and for the control laboratory
fermentations noted in this chapter. Fructose, glucose and citric acid were
utilized by yeast and bacterial species with ethanol, and lactic and acetic acid
being the main metabolic products. Maximum concentrations of ethanol were
reached by 48-72 h. At the end of fermentation, only traces of glucose and
fructose were detected in the pulp. Similar amounts of lactic acid were
produced in these fermentations as noted in previous ones (25-30 mg/g in the
pulp; 10-15 mg/g in the seeds), but, the levels of acetic acid produced (10-20
mg/g in the pulp; 5-10 mg/g in the seeds) were much lower than those
produced during the fermentations described in Chapter 3 (30-60 mg/g in the
pulp; 20-30 mg/g in the seeds). This difference can be explained by the
329
decreased rate of been mixing (48 hr cf. 24 hr) in the pilot scale fermentations
with the reduced aeration limiting the rate at which ethanol would be oxidised
to acetic acid.
Higher concentrations of ethanol were found in the control barrel fermentations
(5-6% in both pulp and seeds) compared to those in the box fermentations (3-4%
in both pulp and seeds). This observation was consistent with those made in
Chapter 3. The barrel fermentations are more contained and enclosed than the
box configuration and, possibly, less ethanol is lost via evaporation. Slightly
higher levels of acetic acid (pulp ~ 20 mg/g, seeds ~10 mg/g) and lower levels
of lactic acid (pulp 15-20 mg/g, seeds 5-10 mg/g) were found in the control box
fermentations, compared to the control barrel fermentations (acetic acid in pulp
~ 15 mg/g, seeds ~5 mg/g; lactic acid in pulp 20-30 mg/g, seeds 10-15 mg/g).
These data suggest that the open structure of the box allowed better passive
aeration compared to the barrel which is more enclosed, and essentially agreed
with Camu et al. (2008b) who found that increased aeration led to more acetic
acid and less lactic acid production.
Different starter cultures affected the chemical changes that occurred during
bean fermentation. Fermentations with H.guilliermondii and I.orientalis gave
faster utilisation of sugars, higher levels of ethanol, acetic acid (seed and pulp)
and malic acid (pulp only) and lower levels of lactic acid compared with both
the control, and the other inoculated fermentations. Compared to S.cerevisiae,
these yeasts are not generally considered strong producers of ethanol. It may be
that the strains selected, or the fermentation conditions used, allowed increased
ethanol production. The higher acetic acid levels corresponded to the higher
ethanol levels, while the increased levels of malic acid supports the hypothesis
that strains of H.guillermondii might produce malic acid during cocoa
fermentation. The lower levels of lactic acid correlated with the suppression of
the lactic acid bacteria observed in these fermentations. However, the presence
of lactic acid in the beans fermented in the box with H.guilliermondii and
I.orientalis suggests that lactic acid bacteria were present, even though they were
not detected using culture based methods.
Fermentation with S.cerevisiae and K.marxmius caused higher levels of ethanol
compared to the uninoculated controls, but did not appear to affect the rate of
sugar utilisation. It is possible that faster utilisation did occur, but was not
330
detected since samples were only tested at 0, 72, and 120 h. The higher levels of
ethanol resulted in slightly higher levels of acetic acid in these fermentations.
Fermentation with S.cerevisiae and K.marxianus resulted in slightly higher
concentrations of lactic acid in the pulp. This may reflect the higher populations
of L.plantarum that occurred when these yeast were used, compared to control
fermentations.
Finally, fermentation with H.guiltiermondii, I.orie?italis, S.cerevisiae and
K.marxianus led to slightly higher concentrations of ethanol, but did not appear
to affect the rate of sugar utilisation, or the production of organic acids during
fermentation compared to the control.
The increased levels of ethanol in the inoculated fermentations compared to the
controls, (>1% higher than the controls), were probably caused by the higher
yeast population. Similar maximum ethanol concentrations were observed for
all three mixtures, supporting the hypothesis that, under the conditions tested,
yeast population rather than yeast species had the greatest effect on ethanol
concentration. Schwan (1998) did not report increased production of ethanol
with the addition of cultures, although Dias et al. (2007) observed that added
cultures led to increased levels of ethanol during fermentation of cocoa pulp
and sweatings.
4.4.3.6 Quality of beans produced by pilot-scale fermentation, and the

chocolate produced from these beans.
All of the cocoa beans produced by these pilot-scale fermentations met, or
exceeded the industry standards for quality testing. The cut test results
indicated that, for all batches of beans, there were no fully purple, slaty,
germinated or insect infested beans. The moisture content of all batches of
beans was below 7%, the fat content was between 54-58%, the minimum bean
size was above 100 beans / lOOg and the pH values of the nibs were between 5.5
and 6.3. The shell content of most batches of beans was within the
recommended limit of 11-13%. The reduction in shell was largely due to
increased mixing during the process of solar drying. Basic subjective
assessment of the appearance and aroma of the beans found that all batches
were acceptable, having no apparent mould growth or undesirable odours.
These results confirmed that the beans produced by the pilot-scale
331
fermentations were suitable for use in commercial chocolate manufacture. One

interesting observation was that beans fermented in boxes were slightly larger
than those fermented in barrels. This corresponded with the earlier observation
that beans fermented in boxes tended to swell more than those fermented in
barrels. The reason for this is not clear, although it suggests that beans
fermented in boxes had better water uptake than the beans in the barrels (Biehl
et al., 1982b). In general, conclusions about the general quality of the fermented
beans agreed with those of Schwan (1998), who also found that fermentation
with defined mixtures of microorganisms could produce beans of an acceptable
quality.
Quality testing revealed that using yeast starter cultures affected some quality
parameters. Beans fermented with H.guilliermondii and I.orientalis had a
significantly higher proportion of fully brown beans, a lower shell content and
higher nib pH compared with the control beans. The differences in cut test and
nib pH correlated with the lower levels of acids found in the seeds of these
beans, compared to the control beans. The lower shell content of the inoculated
beans was consistent with the increased loss of pulp that was observed during
fermentation with H.guillermondii and I.orientalis. This reason for this
accelerated loss of pulp is unclear, since the inoculated yeasts were previously
found to be negative for pectinase production. It should be noted that certain
strains of Candida and Hanseniaspora spp. have been found to be strong
producers of pectinases (Schwan and Wheals, 2004; Maoud and Jespersen,
2006). The addition of H.guillermondii and I.orientalis may have stimulated the
growth of wild strains that had strong pectinase activity. It is recommended that
further screening be performed on the hundreds of H.guillermondii and
I.orientalis isolates collected during this research.
Fermentation with mixtures of S.cerevisiae and K.marxianus, or H.guilliermondii,
I.orientalis, S.cerevisiae and K.marxianus gave inconsistent effects on bean quality
properties. Some differences were, however, noted in the nib pH values (lower
pH in the control box and the barrel inoculated with S.cerevisiae and
K.marxianus), and these were consistent with the chemistry results from the
fermentations (more acetic and lactic acid in the seeds of these fermentations).
Interestingly, while these inoculated species were pectolytic, they did not
significantly lower the shell content of the beans produced. In contrast to
332
research by Sanchez et al. (1984) and, Schwan and Wheals (2004), this study
found that, under the conditions examined, fermentation with pectinolytic
K.marxianus and S.cerevisiae did not give increased pulp loss. Possibly, the
pectinolytic properties described in the literature become more important when
the fermentations are performed in batches larger than 75 kg. Therefore, it
would be beneficial to investigate the performance of pectinolytic yeast for
fermentation of Australian cocoa beans, under a larger range of conditions than
those tested here. In particular, the effect of pulp pH on the pectinolytic activity
of the yeasts should be more closely examined.
Because little research has been performed into the effects of starter cultures on
cocoa bean quality, there is only limited information with which to compare
results. For example, the increased cut test scores of beans fermented with
added H.guilliermondii and I.orientalis, or H.guilliermondii, I.orientalis, S.cerevisiae
and K.marxianus, broadly agreed with those of Buamah et al. (1997), who noted
that different mixtures of yeasts affected the fermentation index of beans. Also,
the observation that mixtures containing K.marxianus could produce beans of
acceptable quality was in contrast to the results of Sanchez et al. (1985) and
Buamah et al. (1997) who found its use resulted in beans of inferior quality.
In addition to testing the quality parameters of dried, fermented beans, quality

evaluation was also performed on the chocolate tasters made from these beans.
This followed the usual industry practice, where samples of processed
chocolate are routinely tested for compliance to quality specifications. This
evaluation found that different starter cultures, or fermentation vessels, did not
have any effect on general composition data such as the nitrogen, sugar or fat
content of the chocolate samples. This agrees with the literature, in that the
proximate composition of chocolate is less influenced by the beans used, and
more influenced by the recipes and processes used in its manufacture (Minifie,
1980; Wood and Lass, 1985; Beckett, 1988, 2000). In contrast, small differences in
other components of the beans, especially the volatile aroma compounds,
resulted in significant flavour variations between the different batches.
Microbiological testing showed that all chocolate samples were free from
Salmonella spp., and were of good microbiological quality (<10 cfu/ g coliforms,
<100 cfu/g yeast, moulds and total plate count). Given the hygiene of the
laboratory environment, the roasting conditions (120°C for 55 min.) and the care
333
take in packaging the chocolate samples for transport, these results were as
expected.
The Padley-Timms test for solid fat content was originally designed to detect
the use of cocoa butter equivalents in chocolate manufacture (Padley and
Timms, 1980). However, in this study it revealed that the cocoa-fat from
Australian cocoa beans had a significantly lower melting temperature than the
cocoa fat from the Ghanaian beans (~3.5 °C difference) (Figure 4.31, Section
4.3.3.7). This difference is probably due to the hotter climate of Ghana compared
to that of Australia, and agrees with the work of Lehrian and Keeney (1980b)
who noted that fat from cocoa beans grown in higher temperature climates had
higher melting points than those grown in cooler environments.
4.4.3.7 Sensory evaluation of cocoa beans produced by pilot-scale
fermentations
The quality of cocoa beans must ultimately be evaluated in terms of their
flavour. Therefore, extensive sensory evaluations were performed on chocolate
made from beans produced by the pilot-scale fermentations. This initially
involved testing whether the samples tasted significantly different from one
another (triangle test), and a simple determination of the flavour quality of the
samples compared to chocolate made from Ghanaian beans (overall liking
testing).
The triangle tests revealed that fermentations with different yeast mixtures
significantly affected the flavour of the resulting cocoa beans, and the chocolate
made with them. Beans fermented with mixtures of H.guiltiermondii and
I.orientalis, S.cerevisiae and K.marxianus, or H.guilliermondii, I.orientalis,
S.cerevisiae, and K.marxianus all gave chocolate that tasted significantly different
from chocolate prepared from beans fermented without added cultures.
Furthermore, the different yeast mixtures gave chocolates that tasted
significantly different from one another. The general finding that different
yeasts resulted in beans with different flavours agreed with work by Sanchez et
al. (1985).
Overall liking tests revealed that all chocolates made with the Queensland
beans were liked significantly more than the chocolate made with Ghanaian
beans. Furthermore, the liking scores from this study were significantly higher
334
than the scores achieved by beans from previous fermentation experiments

(Chapter 3). In Chapter 3, untrained panelists identified a large disparity in the
acidity of some Australian samples compared to the Ghanaian samples. Overt
acid flavours were not reported for the samples produced from the current
pilot-scale fermentations. This suggests that changing the mixing frequency
from 24 hourly to 48 hourly was successful in reducing the acid flavours of the
beans.
The liking results also indicated that beans fermented in a barrel, with
H.guilliermondii and I.orientalis gave chocolate that was particularly liked,
suggesting that both fermentation vessel and the use of cultures may have
contributed to the improvement in acceptability. In the study by Sanchez et al.
(1985), the use of Kluyveromyces marxianus resulted in beans of unfavourable
flavour. In the current study, the use of K.marxianus did not result in poor
flavour, but this may have been because it was used as part of a yeast mixture,
rather than by itself.
While these basic sensory evaluation tests were able to confirm that the starter
cultures caused differences in flavour, they were not able to provide more
specific information about the nature or possible causes of these differences.
Therefore, the chocolate samples were also subjected to quantitative descriptive
analysis (QDA). QDA is a form of trained panel sensory evaluation whereby
detailed sensory profiles of a food may be constructed (Stone et al., 1974;
Carpenter et al., 2000; Stone and Sidel, 2004). Performing QDA on the chocolate
samples helped to clarify and explain the differences in the flavour of beans
fermented with different mixtures of yeasts.
Chocolate made from beans produced by the pilot-scale fermentations had

basic flavour profiles similar to chocolate made from Ghanaian cocoa beans. In
particular, the trained panel found that the Ghanaian and Australian chocolates
had similar intensities of cocoa, astringency, bitterness, sweetness and acidity
(Figures 4.32, 4.33, 4.34). These five characteristics are broadly considered to be
the most important in determining the general acceptability of a chocolate
sample (Lopez and McDonald, 1981; Duncan et al., 1989; Hoskin, 1994; Assemat
et al., 2005; Fraudorfer and Schieberle, 2006; Ramli et al., 2006; Stark, 2006). The
similarity of the Australian samples to the Ghanaian samples in these specific
criteria corresponded to the acceptability of the Australian samples as judged
335
by simple liking testing. The sensory profiles confirmed that the beans
produced during the pilot-scale fermentations did not have strong acid flavours
or aftertastes. This observation corresponded to the chemical data showing that
less acetic acid was produced during the pilot-scale fermentations compared to
the fermentations reported in Chapter 3, and the laboratory fermentations. As
noted before, these improvements can be attributed to changing the mixing
frequency from 24 hourly to 48 hourly, thereby reducing aeration and
production of acetic acid.
There were some differences between the sensory profiles of the Australian and
Ghanaian beans. Firstly, beans produced by the pilot-scale fermentations gave
chocolate with a darker colour than the chocolate made from Ghanaian beans.
This was probably due to their higher pH compared to the Ghana beans, since
chocolate becomes darker brown at higher pH; and lighter red-brown at lower
pH (Minifie, 1980; Beckett, 1988). Secondly, the Australian beans had flavour
characteristics that distinguished them from the flavour of the Ghanaian beans.
Specifically, chocolate made using the Australian beans had doughy, nutty and
grainy flavours that were weak or absent in the chocolate made using Ghanaian
beans.
The profiles generated by QDA also demonstrated that different mixtures of

yeasts gave cocoa beans, and chocolate, with distinctly different flavours.
Cocoa beans fermented with H.guilliermondii and I.orientalis resulted in

chocolate with distinct doughy odour, flavour and aftertaste. These samples
were also less bitter and had a less acid aftertaste than the control. The less acid
aftertaste was consistent with the lower levels of acetic acid in these
fermentations compared to the controls. Fermentations with S.cerevisiae and
K.marxianus produced beans that gave chocolate with distinct nutskin/grainy
flavours and aftertastes. Finally, the mixture of H.guilliermondii, I.orientalis,
S.cerevisiae and K.marxianus produced beans that gave chocolate with a
combination of doughy and nutskin/grainy characteristics. The data indicated
a direct link between the distinctive flavour characteristics and use of particular
mixtures of yeasts. The characteristic 'doughy' flavour was linked to the use of
cultures of H.guilliermondii and I.orientalis, while the 'nutskin/grainy' flavours
were linked to the use of S.cerevisiae and/or K.marxianus. Previously, Sanchez et
al. (1985) proposed that different yeasts might cause flavour differences, while
336
Schwan (1998) found that the use of a mixture of S.cerevisiae and various
bacterial species gave cocoa beans with a similar flavour to that produced by
natural (un-inoculated) fermentation. The current work is novel in that it
confirms that inoculation with different yeast species can produce cocoa beans
of acceptable quality, whilst at the same time causing specific differences in
flavour. This work compliments that being done by Camu et al. (2008a, b) who
have suggested that particular bacterial species might result in particular
flavour characteristics during cocoa fermentation.
The lack of use of starter cultures to control flavour in cocoa bean processing is
in sharp contrast to other fermented products such as wine, beer, bread and
cheese, where inoculation with particular species or strains of microorganisms
is routinely used to obtain desired sensory properties (Urbach, 1995; Rojas et al.,
2001; Fleet, 2003; Vanderhaegen et al., 2003; Hemme & Foucaud-Scheunemann,
2004; Steinkraus, 2004; Swiegers et al., 2005; Howell et al., 2006; Fleet, 2007). Up
until now, control of flavour characteristics in chocolate has relied upon the
blending of cocoa beans with different characteristics, to obtain the desired
profile. The sensory evaluation data of this study show that the use of yeast
starter cultures can allow the primary producer to have greater control over the
flavour of the beans produced. In the Australian context, this can help ensure
that a high quality product can be consistently produced. Such quality
assurance is important for cocoa production in a context where labour costs are
high and the profit margins rely on a premium price being obtained.
4.4.3.8 Degradation of cocoa seed proteins during fermentation
Proteolysis of the cocoa seed proteins is critical to the development of chocolate

flavour (Zak and Keeney, 1976; Biehl et al., 1982; Voigt et al. 1994; Voigt and
Biehl, 1995; Hansen et al., 2000; Afoakwa et al., 2008). Consistent with beans
produced in other countries, Australian cocoa beans exhibited the presence of
the vicilir. and globulin proteins.. The enzymatic proteolysis of vicilin-class
proteins during fermentation produces peptides and amino acids that, when
roasted, give cocoa beans their distinctive flavour (Voigt et al., 1994a, 199b;
Voigt and Biehl, 1995). The protein gels generated by SDS-PAGE qualitatively
indicated :hat these proteins were degraded during fermentation, according to
the expected pattern from the literature (Zak, 1973; Zak and Keeney, 1976a,
1976b; Buyukpamukcu et al., 2001).
337
In the fermentations inoculated with H.guiltiermondii and I.orientalis, the

proteolysis was more rapid, while fermentations with S.cerevisiae and
K.marxianus and H.guillier mondii, I.orientalis, S.cerevisiae and K.marxianus
increased the rate of proteolysis, but to a lesser degree. This is the first time that
the use of defined mixtures of yeast to conduct cocoa fermentation, has been
linked to accelerated proteolysis in the cocoa seeds. According to explanations
given by Voigt et al. (1995), Misnawi et al. (2002, 2003), and Afoakwa et al.
(2008), faster proteolysis is usually triggered by acceleration of external
fermentation processes (changes to pH, temperature, sugars, ethanol, acids),
such as were observed in the inoculated fermentations during this present
study. Therefore, one explanation for the more rapid proteolysis in the
fermentations using mixtures of yeasts, is that they indirectly triggered faster
proteolysis by accelerating the general rate of the fermentation.
An alternative explanation that deserves consideration is that proteases

produced by the inoculated yeast may have penetrated the testa and
contributed to the proteolytic reactions within the seeds. This hypothesis is
supported by the observation that yeasts isolated from Australian cocoa
fermentations (and used in the starter cultures) may produce proteases similar
to those found in the cocoa seeds. Strains of I.orientalis are able to produce a
range of proteases including an aspartic endoprotease (Pichova et al., 2001), and
it is well established that S.cerevisiae can produce carboxypeptidases (Magliani,
1997; Yusep et al., 2002). Both of these enzymes have been shown to be critical
in the formation of peptides and amino acids that act as chocolate flavour
precursors (Voigt et al., 1994a, 1994b, 1995; Hansen et al., 1998). H.guillier mondii
also secretes a range of proteases and peptidases (Dizy and Bisson, 2000) that
could affect cocoa seed proteins. Strains of H.guillier mondii, I.orientalis and
S.cerevisiae, isolated from the Australian cocoa fermentations were positive for
protease activity (Table 5, Appendix A). Furthermore, the specific strains of
H.guillermondii and I.orientalis used in the starter culture mixture were
proteolytic. The above hypothesis is also supported by the work of Yusep at al.
(2002) and de Brito (2004) who found that the application of yeast proteases
(particularly carboxypeptidases) to under-fermented cocoa beans gave
complete proteolysis, and a better flavour potential. Finally, early work by
Rohan (1963), Quesnel (1966) and Kirchoff (1989a) suggested that cocoa seed
338
proteins are leached into the pulp during fermentation. There appear to be no
reports into whether proteases of microbial origin might contribute to the
degradation of cocoa seed proteins. The hypothesis that the testa remained
impermeable to enzymes was made in early literature (Chatt, 1953, Holden,
1959, Lehrian and Patterson, 1983) but has not been investigated further. It is
therefore recommended that this topic be re-examined.
Regarding flavour development, the results from the protein gels could not be
clearly linked to the sensory evaluation results. Specifically, while the gels
suggested differences in the rates of proteolysis, sensory evaluation indicated
that all samples had a similar strength of cocoa flavour (c.f. Section 4.3.4,
Chapter 4). This suggests that while the rates of proteolysis differed between
fermentations, a similar endpoint of proteolysis was reached for all beans by the
completion of drying. That is sun-drying for 5 days allowed ample time for the
completion of flavour-precursor reactions, in all treatments. Nevertheless, the
protein gels suggest that using a combination of proteolytic starter cultures, and
an appropriate drying method, could produce beans with an acceptable flavour
in a shorter time (at least 24h shorter). The ability to shorten the processing time
would be a great advantage in a large-scale commercial operation, such as
envisaged in the Australian context.
4.4.3.9 Changes to polyphenols and antioxidant capacity of beans during

fermentation
Cocoa seeds are rich in a range of polyphenolic compounds. These compounds
are significant because of their role in contributing to the flavour of the beans
(Wollgast and Anklam, 2000; Luna et al., 2002; Counet et al., 2004; Stark et al.,
2005; Nazaruddin et al., 2006). More recently, greater attention has been given to
cocoa polyphenols due to claims about their potential health benefits (Serafini,
2004; ICCO, 2005c; Keen et al., 2005; Lamuela-Raventos et al., 2005;
Vlachopoulos et al., 2006). During fermentation, drying and roasting, the
polyphenolic compounds present in the cocoa seeds usually undergo
considerable changes (Forsyth and Quesnel, 1963; Lehrian and Patterson, 1983;
Misnawi et al., 2003).
In this study, the concentration of procyanidins in unfermented beans was
estimated at about 40 mg/g. While the initial level can vary by +5 mg/g due to
339
regional and seasonal factors, this is still lower than the 60 mg/g average
estimated previously (Nazaruddin et ah, 2006). This suggests losses of
procyanidins occurred during sample preparation. One weakness of the
method used in this study - based on Counet et al. (2002) - was that it fails to
estimate losses during sample preparation. Another problem, noted by Counet
et al.(2002) was the absence of a consistent, commercially available standard
mixture of polymeric procyanidins. This necessitated the estimation of total
procyanidins against catechin (procyanidin monomer), which may have a
different response factor to the polymeric forms. In spite of these limitations, it
was still possible to make the following qualitative observations:
First, during the pilot-scale fermentations the concentration of total
procyanidins in the cocoa seeds decreased by approximately 40-50 %, which is
consistent with the earlier estimates of Kim and Keeney (1984). The loss of
procyanidins in the beans correlated qualitatively with a decrease in the
antioxidant capacity of the beans. This generally agreed with Counet & Collin
(2002) and Gu et al.(2005) who found that the procyanidins are largely
responsible for the antioxidant activity of cocoa.
Second, the decrease of procyanidins, and antioxidant activity, during
fermentation was slightly less for fermentations conducted with H.guiltiermondii
and Lorientalis, or S.cerevisiae and K.marxianus. The reasons for this are unclear.
Possibly, faster degradation of seed proteins in the inoculated fermentation
resulted in less formation of polyphenol-protein complexes. Also, the starter
cultures may have caused more rapid inactivation of the cocoa seed polyphenol
oxidase, which reduces the rate at which the polyphenols are degraded by this
enzyme.
Residual polyphenols contribute bitterness and astringency to cocoa beans
(Quesnel, 1972a, 1972b; Kim and Keeney, 1984; Luna et ah, 2002; Counet et ah,
2004). In this study, no correlation could be made between the differences in
polyphenol levels observed during fermentation, and the flavour of the
chocolate produced from the different beans. In particular, all of the beans gave
chocolate with similar levels of bitterness and astringency. It should be noted
that the levels of polyphenols in the beans after drying were not measured in
this study. It is possible that the beans reached similar levels of polyphenols
during the drying process, in spite of the differences observed during
340
fermentation. Previously, Nazaruddin et al. (2006) found that pod storage and
duration of fermentation affected the loss of polyphenols, while Luna et al.
(2002) suggested that fermentation method can affect the procyanidin content of
the beans. There have been no previous attempts to determine the influence of
fermentation microbiology on polyphenol levels. Given this, and the limitations
of the analyses made in this study, future experiments into the use of starter
cultures for cocoa fermentation should more closely examine the effects of
different cultures on the levels of polyphenolic compounds and their
antioxidant activity.
4.4.3.10 GC-MS analysis of volatiles extracted from cocoa beans
A diversity of volatile compounds present in roasted cocoa beans contributes to

their flavour (Rohan, 1965, 1969; Flament, 1991; Jinap et al. 1998; Bonvehi, 2005).
In this study, GC-MS was used to estimate the volatile compounds found in
roasted Australian cocoa beans. The volatile compounds found in the samples
were qualitatively consistent with those found in beans from Malaysia (Jinap et
al., 1998), Ghana (Yoo et al., 1998; Bonvehi, 2005) Papua New Guinea, Trinidad,
Indonesia, Venezuela, Cameroon, Ivory Coast, Brazil and Ecuador (Counet et
al., 2004; Bonvehi, 2005; Afoakwa et al., 2008). In particular, pyrazines, typically
responsible for characteristic cocoa aroma (Flament, 1991; Hashim and
Chaveron, 1994; Hashim et al. 1998a, b, 1999), were isolated from the Australian
samples at similar levels to those in the Ghanaian sample. This observation
correlated with the sensory evaluations which found that the Australian beans
had a similar strength of chocolate character compared to Ghanaian beans.
Quantitative data from the GC-MS were analysed using principal component
analysis (PCA), in order to compare the overall profiles, rather than comparing
compounds individually. Using PCA to analyse the GC-MS data, the Ghanaian
beans were clearly distinguished from the Australian beans. It was also found
that the Australian samples were grouped into various sub-clusters on the PCA
bi-plot. One of these sub-clusters corresponded to the fermentations conducted
with H.guiltiermondii and I.orientalis. The data were inconclusive concerning the
other clusters. In a limited way, these results corresponded to the sensory
analyses, which were able to distinguish between Australian and Ghanaian
beans, and beans fermented with different mixtures of yeast.
341
As well as allowing overall comparison of large data sets, PCA can also help in
correctly determining the weight, or significance level, of individual variables
within the set (Stevens, 1986; Rencher, 1995; Yoo et ah, 1998). Therefore, while
the focus was to compare the overall GC profiles, rather than comparing
individual points, PCA revealed that some differences between individual
compounds were statistically significant. Firstly, the higher levels of 2-pentanol
acetate, linalool oxide, ethyl benzoate, and lower levels of 2-phenylethanol in
the Ghanaian beans were significant in distinguishing them from the Australian
beans. There were insufficient data to clearly link this observation to differences
in flavour observed during sensory analysis. Secondly, the higher levels of
isoamyl alcohol, isoamyl acetate, and 2-phenylethanol detected in the samples
fermented with H.guilliermondii and I.orientalis were significant in
distinguishing these samples during PCA analysis. In this case, the GC-MS
results could be related to the microbiology of the fermentations. The
microbiology of these fermentations was characterised by a predominance of
H.guilliermondii and I.orientalis. H.guilliermondii is a strong producer of 2-
phenylethanol, isoamyl alcohol and isoamyl acetate (Rojas et al., 2001; Swiegers
et al., 2005), and these results suggest that it played a significant role in
determining the sensory profile of these beans, as one of the most populous
yeast species (Fleet, 2003). The GC-MS data were not sufficient to allow similar
correlations with beans fermented with other mixtures of yeasts
Metabolomic approaches combining GC-MS and PCA have been used for
studying various fermented foods, and to correlate differences in microbial
ecology to differences in volatile aroma compounds (Steinkraus, 2004; Ouoba et
al., 2005). In spite of the problems and limitations, this present study has
demonstrated the usefulness of GC-MS in clarifying the contribution of volatile
compounds to cocoa beans, by the microorganisms responsible for
fermentation. GC-MS and PCA have also been used to study the interactions
between mixed yeast cultures in wine fermentation, their metabolites and the
effects of these interactions on the flavour of the wine produced (Howell et al.,
2006). Such 'metabolic profiling' has not been applied to cocoa fermentation,
but could prove useful in further characterisation of yeast strains for starter
culture development.
342
4.3.4.11 Summary and future direction

In this study, the primary goal of the pilot-scale fermentations was achieved:
fermentation with mixtures of yeasts (H.guilliermondii and I.orientalis, S.cerevisiae
and K.marxianus, and H.guilliermondii, I.orientalis, S.cerevisiae and K.marxianus)
produced cocoa beans of good commercial quality and having a good flavour.
A key implication of this research is that by the use of starter cultures, cocoa
fermentation could be made more consistent, and possibly faster. Furthermore,
the different mixtures of yeasts gave cocoa beans having distinctively different
flavour profiles. This represents a novel finding, one which would allow the
primary producer to have greater control over the beans produced. These
findings have both practical and economic consequences for cocoa production
in Australia, and abroad. It is recommended that the findings of this research be
applied in developing a full-scale industrial process for use in Queensland,
Australia.
To achieve a truly industrialized process, however, further work needs to be

done. In particular, production-scale equipment needs to be designed and
tested. As mentioned already, a mechanized pod splitter has been developed
(DPI, 2004). Building on the potential of the barrel configuration, a cost-
effective, full scale (>200 kg) fermenter should be designed and tested. The
fermenter should be trialled under a range of conditions. In particular, results
from this study suggest that aeration and mixing will have be optimized to
ensure appropriate levels of acetic acid. There is also a need to develop an
effective system for bean drying suitable for use in Northern Queensland. It
seems likely that some form of mechanical dryer would be preferred, because of
high humidity and rainfall in Northern Queensland. Such a dryer should
ideally be energy efficient, compact, and should conduct drying in a fashion
that mimics the processes of sun-drying. Finally, effective control systems
should be developed to allow for an integration of the different processing
steps, and to improve the consistency of quality. Methods for monitoring critical
processing variables such as fermentation pFI, temperature and basic
microbiology should be developed. Such a control system might evolve into a
quality and/or safety management system.
In addition to these practical improvements, there is potential to further

research certain questions not adequately addressed in this study. In particular,
343
this study did not explore the potential contribution of different species or
strains of lactic acid bacteria and acetic acid bacteria on process kinetics, or
cocoa flavor. Recent work by Camu et al. (2007; 2008a, b) has investigated this
with respect to Ghanaian cocoa fermentation. They believe that these bacteria
are key to the production of characteristic flavours during cocoa fermentation.
More research is needed to determine the impacts of specific bacteria on cocoa
bean processing and chocolate quality, and the potential for their development
as starter cultures. The isolation of strains of lactic acid bacteria and acetic acid
bacteria from the Australian fermentations (Chapter 3) that could not be
identified by available criteria underscores the importance for such research.
Discovery of novel species of lactic and acetic acid bacteria from cocoa
fermentations in Ghana also suggests that there may be geographical variation
in the strains present in this ecological niche (Camu et al.; 2007; Nielsen et al.,
2007b).
4.5 Conclusions
A) Three mixtures of yeast species were selected for use as starter cultures. The
mixtures were comprised of yeasts isolated from previous Australian cocoa
fermentations, and one strain of Kluyueromyces marxianus isolated from wine
fermentations. The mixtures were: H.guilliermondii and l.orientalis, S.cerevisiae
and K.marxianus, and H.guilliermondii, l.orientalis, S.cerevisiae and
K.marxianus. It is recommended that the isolates used in these mixtures, be
more fully characterised, including detailed molecular profiling as well as
cataloging important phenotypic characteristics such as production of
hydrolytic enzymes or specific volatile aroma compounds. This would
facilitate the development of further starter cultures, and help explain
specific impacts on chocolate flavor.
B) Laboratory fermentations confirmed that the selected mixtures of yeasts
would survive and grow when inoculated into Australian cocoa beans.
These small-scale fermentations proved to be an appropriate model for the
evaluation of starter cultures, and may be useful in conducting other
controlled experiments not possible in the field. Cocoa beans produced by
the laboratory fermentation varied in quality, and it is recommended that the
344
method used for laboratory fermentation be further optimised to reduce

aeration and improve the quality of beans produced.
C) The pilot-scale fermentations produced cocoa beans of good quality and
flavour. The physical, microbiological and chemical changes observed
during the control fermentations reflected those observed during previous
Australian fermentations (Chapter 3).
D) Conducting pilot-scale fermentations with added mixtures of yeasts affected
the microbiology. Firstly, the inoculated species dominated the early stages
of fermentation. The microbiology data also suggested that there were yeast-
yeast interactions between the inoculated species and the indigenous yeast
species. Specifically, inoculation with H.guilliermo?idii and Lorientalis
suppressed the growth of S.cerevisiae, and inoculation with S.cerevisiae and
K.marxianus slightly suppressed the growth of H.guiltiermondii. Fermentation
using mixtures of yeasts also affected the growth of lactic acid bacteria. In
particular, the growth of L.fermentum was suppressed in the fermentations
inoculated with H.guilliermondii and Lorientalis.
E) Fermentation with different mixtures of yeasts affected the chemistry and
biochemistry of the beans.
F) Fermentation with different mixtures of yeast resulted in cocoa beans with
distinct differences in flavour. This experiment confirmed the potential for
carefully designed starter cultures to be used to control the flavour and
quality of cocoa beans.
G) Analysis of the degradation of cocoa seed proteins suggested that
fermentation with a mixture of H.guilliermondii and Lorientalis might result in
accelerated proteolysis reactions. This in turn could allow for more rapid
processing of cocoa beans compared to traditional methods. These results
also suggested that microbial proteases might diffuse into the cocoa seeds
and accelerate the breakdown of cocoa seed proteins.
H) Changes in the levels of polyphenols during fermentation indicated that
fermentation with mixtures of H.guilliermondii and Lorientalis might allow
increased retention of these compounds. Further research is recommended to
confirm this result.
345
I) Analysis of cocoa bean samples by gas-chromatography mass spectrometry

(GC-MS) revealed significant differences in the volatile aroma profiles. Some
of these differences correlated with the use of different starter cultures. There
are clear opportunities for the application of 'metabolic profiling' to be
applied to cocoa fermentation. This would allow microbial ecology data to
be better linked to the volatile profiles of the cocoa beans produced.
346
Chapter 5
Conclusions
The major objectives of this thesis were to collect fundamental information
about the fermentation of cocoa beans cultivated in Queensland, Australia, and

to develop approaches to fermentation that would be suitable for
industrialisation. Both overall objectives were fulfilled. While the core goals
were achieved, some limitations and opportunities for further research were
also identified.
5.1 Key conclusions as they relate to project objectives and gaps in

knowledge
• Cocoa beans cultivated in Queensland, Australian possessed physical and

chemical properties comparable to cocoa beans grown elsewhere in the
world. For this research, cocoa pods were obtained from a selected group
of Trinitario-Forastero hybrid trees, grown at two plantations, South
Johnstone and Mossman, Queensland. The Australian cocoa beans were
found to have a similar composition to cocoa beans cultivated in other
countries. Unfermented pulp had a pH of 3.5 - 3.9, contained glucose

(5-6%), fructose (6-7%) citric acid (40-50 mg/g) and traces of malic,
tartaric and oxalic acids. The seed material of unfermented beans
contained only traces of sugars and organic acids (<1%). Minor

variations in the pH, and concentration of sugars and organic acids were
observed in different batches of beans. The Mossman plantation
produced cocoa beans with more pulp, a lower pulp pH and higher
levels of sugars and citric acid than beans from the South Johnstone
347
plantation.
o The Australian cocoa beans underwent fermentation due to
indigenous yeasts and bacteria. During fermentation a range of
physical, chemical and microbiological changes were observed,
and were consistent with the general changes typically observed

during cocoa fermentation in other countries:
o The quantity of pulp adhering to the beans decreased during
fermentation, and was lost as sweatings. The temperature of the

cocoa beans started at 23 - 25 °C, increased to 45 - 48°C by 72 hr
and then decreased to 40°C by the end of fermentations (120 h).
The pH of the cocoa pulp increased during fermentation (from pH
3.5-3.9 to pH 4.2 - 5.0), while the pH of the cocoa seeds decreased
(from pH 6.0 to pH 5.0-5.5).
o Fermenting Australian cocoa beans (in heaps, boxes or barrels)
were found to have a complex microbial ecology involving the

growth of yeast and bacteria. A diversity of yeast species grew,
and reached maximum populations of 107-108 cfu/ g after 36-48h.
The predominant species were Hanseniaspora guiltiermondii,

Issatchenkia orientalis and Saccharomyces cerevisiae, with Pichia
membranifaciens isolated occasionally. Lactic acid bacteria also
grew and reached maximum populations of 108-109 cfu/g by 48 h.

The main lactic acid bacteria were Lactobacillus plantarum,
Lactobacillus fermentum, Leuconostoc pseudomesenteroides and
Pediococcus acidilactici. After 36-48h the microbial ecology was

dominated by growth of acetic acid bacteria, particularly
348
Acetobader pasteurianus, Acetobader tropicalis, and, less frequently,
Gluconobader oxydans. These species usually reached maximum
populations of 107-108 cfu/ g. From 48h onwards. In the later
stages of some fermentations, various Bacillus species were
isolated, in particular B.subtilis, B.licheniformis and B.megaterium.
Filamentous fungi were occasionally isolated at the beginning and
end of fermentations, and from drying beans.
o During the fermentations, ethanol was produced, reaching
maximum levels by 48 h of 4-7% in the pulp, and 2-5% in the
seeds. Lactic and acetic acids were also produced in the pulp
during fermentation and diffused into the seeds, reaching
maximum levels around the end of fermentation. The maximum
levels of lactic and acetic acids were 20-40 mg/g and 30-90 mg/g
in the pulp, and in the seed, 10-15 mg/g and 15-30 mg/g
respectively.
• This research confirmed that Australian fermented cocoa could be
processed into chocolate of acceptable, commercial quality. The quality of
dried, fermented cocoa beans, from these experiments, were deemed
acceptable when measured using standard industry criteria such as cut-
test, moisture and fat content, and nib pFL The Australian cocoa beans
also compared favourably to samples of Ghanaian beans, which are
commonly used as a benchmark in industry. Sensory evaluation
demonstrated that chocolate made with the Australian beans had an
acceptable flavour, compared to the flavour of chocolate made with
Ghanaian beans. Sensory analysis also showed that the Australian cocoa
349
beans had a flavour distinctly different from the Ghanaian beans.
• Investigation of key processing variables confirmed that fermentation

vessel and mixing frequency affected the microbial ecology, chemistry,
quality and flavour of the cocoa beans produced. Based on these
experiments, it was found that an uninoculated barrel fermentation, with

mixing every 48 h, for a period of 5 days (120h), produced high quality
cocoa beans having a flavour comparable to, or more acceptable than, the
Ghanaian standard.
• Three mixtures of yeasts were designed and tested for use as starter
cultures in the fermentation of Australian cocoa beans: Hanseniaspora
guilliermondii and Issatchenkia orientalis, Snccharomyces cerevisiae and
Kluyveromyces marxianus, and H.guilliermondii, I.orientalis, S.cerevisiae and
K.marxianus. Except for K.marxianus (strain obtained from wine grapes),
all yeast strains used were isolated during the earlier fermentation
experiments. The isolates used were subjected to basic strain
characterisation screening.
• Controlled, laboratory fermentations confirmed that the selected mixtures

of yeasts survived and grew when inoculated into Australian cocoa
beans. These small-scale fermentations proved to be an appropriate
model for the evaluation of starter cultures, and may be useful for
conducting other controlled experiments not possible in the field.
• Pilot-scale (50kg) fermentations using the three mixtures of yeasts were
performed in box and barrel configurations. Although no specific efforts

were made to exclude indigenous species, the inoculated species grew
350
and dominated the early stages of fermentation. Fermentation with
mixtures of H.guiltiermondii and I.orientalis suppressed the growth of
S.cerevisiae and lactic acid bacteria. Fermentation with S.cerevisiae and
K. marxianus slightly suppressed the growth of H.guildermondii and
L. fermentum, but stimulated the growth of A.pasteurianus and L.plantarum
• The use of the starter cultures affected the chemistry of the fermentations.
The pH of the pulp increased to a greater degree in the inoculated
fermentations, compared to the controls. In the fermentation with added
yeasts, higher levels of ethanol and acetic acid were produced, compared
to the controls. Fermentation with a mixture of H.guilliermondii and
I.orientalis led to reduced lactic acid production compared to the controls,
which corresponded to the inhibition of lactic acid bacteria observed in
these fermentations. Accelerated proteolysis was observed in the seeds
fermented with H.guilliermondii and I.orientalis.
• All of the cocoa beans produced by these pilot-scale fermentations met, or
exceeded the industry standards for basic quality criteria (cut test,
moisture content, fat content and pH). Chocolate made with cocoa beans
fermented with the different mixtures of yeast had an acceptable flavour
compared to chocolate made from Ghanaian beans. Fermentation with
different mixtures of yeast resulted in cocoa beans, and chocolate, with
distinct differences in flavour.
5.2 Novel aspects
• This study represents the first major investigation on the fermentation of
cocoa beans grown in Queensland, Australia. With no prior history of

351
cocoa fermentation in Australia, significant gaps in knowledge existed.
In particular, this thesis provides the first descriptions of microbial
ecology under local conditions. The data from this project provides a
useful baseline for any future cocoa research in Australia. An improved
understanding of Australian cocoa beans, and their fermentation, may
also help identify characteristics unique to the local product.
• The barrel fermentation vessel investigated in this project was shown to be
comparable to traditional heap and box methods, in terms of the basic
microbiology and chemistry observed. Furthermore, beans fermented in
barrels were of acceptable quality and flavour compared to beans
fermented by traditional methods, and compared to Ghanaian beans.
The barrel vessel had several practical advantages over the traditional
heap and box methods.
• Microbial nutrients are commonly used in many food fermentations to
prevent or ameliorate stuck fermentations. This was the first time such
supplements have been evaluated with regards to their effects in the
fermentation of cocoa beans. Their addition slightly improved the
growth of some species, in particular S.cerevisiae and L.plantarum.
• While research has been performed into the use of starter cultures for
cocoa fermentation, certain features of this project set it apart. The starter
cultures used here were mixtures of yeasts, rather than single strains. As
much as possible, the isolates originated from cocoa, rather than other
ecological niches. The starters were deliberately inoculated into non-
sterile substrate, rather than attempting to first eliminate the indigenous
microflora, as would be required in commercial operations. In particular,

352
this study represents the first in-depth study on the use of a mixture of
H.guilliermojidii and I.orientalis to conduct commercial-scale cocoa
fermentation. The distinctively different flavour profiles caused by the
different mixtures of yeast also represented a novel finding - one which
could allow the primary producer greater control over product character.
5.3 Limitations
• The relatively small amount of cocoa pods available each season
constrained the experiments in several ways. Firstly, the size of
fermentations that could be performed was restricted to between 50-75
kg. While this corresponds to fermentations practiced by small holders,
many commercial operations ferment 250 - 500 kg quantities of beans.
Second, damage to trees by a tropical cyclone, prevented separate testing
of Mossman and South Johnstone pods across all experiments. Finally,
there were insufficient pods from individual Australian hybrid types to
test the effects of genetic type on fermentation and quality.
• The scope of the project, and methodology used, limited the evaluation of
microbial ecology. The diversity and significance of filamentous fungi
was not investigated in-depth, in spite of their potential significance at
the beginning and end of fermentation. Microbial diversity throughout
bean fermentation was only examined at the species level and further
investigation of diversity at the strain level is needed. Several isolates of
lactic acid bacteria were obtained that were not able to be fully identified.
Further testing of these isolates, and further surveys of microbial ecology
are necessary to evaluate whether these isolates represent novel species.
In particular, the microbial ecology of drying was not investigated and

353
warrants attention, especially in relation to control of the growth of
filamentous fungi and their potential to produce mycotoxins.
• Due to project constraints, the isolates chosen for use as starter cultures
were subjected only to a basic screening protocol. In particular, full
metabolic profiling and molecular characterisation of the isolates at the
strain level was not conducted. Further characterization, for the purposes
of strain monitoring is recommend, since circumstances can arise when
the inoculated strain might not dominate the fermentation.
5.4 Future work
• Further work is necessary in order to consolidate the results presented
here, and develop fully industrialised methods for cocoa fermentation in
Australia. In particular, there is a need to design and optimize
fermentation vessels capable of processing larger quantities (250-1000 kg)
of cocoa beans, such as would be encountered in an industrial facility. It
is recommended that more cocoa beans be obtained to allow these scale-
up experiments, and the mixing frequency of these larger fermentations
be carefully optimised. Appropriate sensing and control systems must be
designed and incorporated into such industrial fermenters. A well-
controlled mechanical drying system should also be developed so that
beans can be dried efficiently, without interrupting the formation of
flavour precursors, but with control of mould growth.
• The starter cultures should be further developed for industrial
fermentation of cocoa beans. In particular, the yeasts should be screened
for favorable attributes, such as enzyme production, production of

354
volatile compounds, resistance to ethanol and heat, and optimum
growth rate. More work should be undertaken to investigate the role of
bacteria in affecting the flavour of cocoa beans. This could lead to the
development of specific bacterial starters, whether individual species, or
mixtures. It is recommended that this work could be performed in
collaboration with a company specializing in starter production, in order
to ensure supply of the organisms on a commercial scale.
• It is strongly recommended that as more cocoa plantations are established
in Queensland, the effects of genetic variety on fermentation and flavour
should be considered as much as possible. This will allow selection of the
best variety not only in terms of yield, but also quality of beans
produced. The effects of agricultural practices on the composition of
unfermented cocoa beans, and their fermentation could also be
evaluated.
• Finally, there are clear opportunities for further investigation into the
microbiology of cocoa beans cultivated in Queensland, Australia. Areas
of interest include the microbiology of cocoa beans during drying, the
development and application for rapid methods to study cocoa
fermentation, the microbial interactions occurring during fermentation
and their effects on development of aroma volatiles. Metabolomic
studies, linking the growth of certain species to detailed GC-MS profiles
of flavour volatiles have been useful in studying other food
fermentations, such as wine, and traditional African foods. Such an
approach may prove useful in studying the contributions of
microorganisms to the flavour of cocoa beans.

355
Chapter Six
Bibliography
Abeygunasekera, D.D. and Jansz, E.R. (1989) The effect of the maturation
process on fermented cocoa beans II: Acids, sugars, xanthines, pyrazines and
anthocyanins. J. Natn. Sci. Coun. Sri Lanka 17, 35-42.
Adams, A. and De Kimpe, N. (2007) Formation of pyrazines and 2-acetyl-l-

pyrroline by Bacillus cereus. Food Chem. 101,1230-1238.
Adams, M.R. and Moss, M.O. (2000) Food Microbiology, 2nd edition. Royal Society
of Chemistry, Cambridge UK.
Adamson, G.E., Lazarus, S.A., Mitchell, A.E., Prior, R.L., Cao, G., Jacobs, P.H.,
Kremers, B.G., Hammerstone, J.F., Rucker, R.B., Ritter, K.A. and Schmitz, PEEL
(1999) HPLC method for the quantification of procyanidins in cocoa and
chocolate samples and correlation to total antioxidant capacity. J. Agric. Food
Chem. 47, 4184-4188.
Adejumo, T.O. (2005) Crop protection strategies for major diseases of cocoa,
coffee, and cashew in Nigeria. Afr. J. Biotech. 4,143-150.
Addis E., Fleet G.H., Cox J.M.C., KolakD. and Leung, T. (2001) The growth,
properties and interactions of yeasts and bacteria associated with the
maturation of Camembert and blue-veined cheeses. Int. J. Food Microbiol. 69,
25-36.
Afoakwa, E.O., Paterson, A., Fowler, M. and Ryan, A. (2008) Flavour formation
and character in cocoa and chocolate: a critical review. Crit. Rev. Food Sci. Nutr.
48, 840-857.
Aikpokpodion, P.O, Badaru, K., and Eskes, A. (2003) Improving brown budding
efficiency in cacao, Theobroma cacao L.: effects of twig manipulation and some
control treatments of fungal infections on new sprouts. Crop Prot. 22, 1-6.
Albergaria, H., Torrao, A.R., Flogg, T. and Girio, F.M. (2003) Physiological
behaviour of Hanseniaspora guilliermondii in aerobic glucose-limited continous
conditions. FEMS Yeast Res. 3, 211-216.
Alemanno, L., Ramos, T., Gargadenec, A., Andary, C. and Ferreire, N. (2003).
Localization and identification of phenolic compounds in Theobroma cacao L.
somatic emryogenesis. Ann. Botany 92, 613-623.
Alexandre, H., Rousseaux, I. and Charpentier, C. (1994) Relationship between

ethanol tolerance, lipid composition and plasma membrane fluidity in
Saccharomyces cerevisiae and Kloeckera apiculata. FEMS Microbiol. Lett. 124, 17-22.
Alexandre, H., Costello, P.J., Remize, F., Guzzo, J., Guilloux-Benatier, M. (2004).
Saccharomyces cerevisiae-Oenococcus oeni interactions in wine: current knowledge
and perspectives. Int. J. Food Microbiol. 93,141-154.
356
Allison, H.W.S. and Kenten, R.H. (1963) Tray fermentation of cocoa. Trop. Agric.
(Trinidad) 40,15-24.
de Almedia, A-A. F. and Valle, R.R. (2007) Ecophysiology of the cacao tree. Braz.
J. Plant Physiol 19, 425-448.
Altschul, S.F, Gish, W., Miller, W., Myers, E.W. and Lipman, D.J. (1990) Basic
local alignment search tool. J. Mol. Biol. 215, 403-410.
Alvim, P. de T. (1977) Cacao. In Ecophysiology of Tropical Crops eds. Alvim, P de T.
and Kozlowski, T.T. pp. 279-314. New York: Academic Press.
Amoa-Awua, W.K., Sampson, E. and Tano-Debrah, K. (2006) Growth of yeasts,
lactic and acetic acid bacteria in palm wine during tapping and fermentation
from felled oil palm (Elaeis guineensis) in Ghana. /. Appl. Microbiol. 102, 599-606.
Amezqueta, S., Gonzalez-Penas, E., Marillo, M. and Lopez de Cerain, A. (2004)
Validation of a high performance liquid chromatography analytical method for
ochratoxin A quantification in cocoa beans. Food Addit. Contain. 21, 1096-1106.
Amezqueta, S. Gonzalez-Penas, E., Murillo, M. and Lopez de Cerain, A. (2005)
Occurrence of ochratoxin A in cocoa beans: effect of shelling. Food Addit.
Contain. 22, 590-596.
Amin, I., Jinap, S., Jamilah, B., Flarikrisna, K. and Biehl, B. (2002) Analysis of
vicilin (7S)-class globulin in cocoa cotyledons from various genetic origins. /.
Sci. Food Agric. 82, 728-732.
Amores, F., Butler, D., Ramos, G., Sukha, D., Espin, S., Gomez, A., Zambrano,
A., Hollywood, N., Van Loo, R. and Seguine, E. (2007) Project to determine the
physical chemical and organoleptic parameters to differentiate between fine and bulk
cocoa - project completion report. London: International Cocoa Organization.
Anastassiadis, S., Aivasidis, A. and Wandrey, C. (2002) Citric acid production by
Candida strains under intracellular nitrogen limitation. Appl. Microbiol.
Biotechnol. 60, 81-87.
Annan, N.T., Poll, L., Sefa-Dedeh, S., Plahar, W.A. and Jakobsen, M. (2003)
Influence of starter culture combinations of Eactobacillus fermentum,
Saccharomyces cerevisiae and Candida krusei on aroma in Ghanaian maize dough
fermentation. Eur. Food Res. Technol. 216, 377-384.
Anon. (1984) Cocoa Beans: Chocolate manufacturers' quality requirements. London
UK: The Cocoa, Chocolate and Confectionary Alliance.
Anon. (1999) Deficit in World Cocoa Production; Cadbury Looks to Australia
for Production. Food & Drink Weekly [online]. [Accessed 12 December 2007].
Available from World Wide Web: <http:/ / www.allbusiness.com/retail-trade/
food-beverage-stores / 289650-1 ,html>.
Anon. (2003) Optimisation of chocolate manufacturing: in-situ angle X-ray
scattering (SAXS) of cocoa butter crystallisation. Australian Synchrotron, Victoria
[online]. [Accessed 1 November 2008]. Available from World Wide Web:
357
<http: / / www.synchrotron.vic.gov.au/files/documents/Case chocolate 4-

final.pdf>.
Anon. (2007) Chocolate Trends 2007. Confectionery News.Com [online]. [Accessed

1 May 2008] Available from the World Wide Web: chttp:/ /
www.confectionerynews.com / news>.
AO AC. (1997) 'Official methods of analysis.' Washington D.C.: Association of

Official Analytical Chemists.
Ardhana, M.M. (1990) Microbial ecology and biochemistry of cocoa bean

fermentation. PhD Thesis. University of New South Wales, Sydney Australia.
Ardhana, M.M. and Fleet, G.H. (2003) The microbial ecology of cocoa bean
fermentations in Indonesia. Int. J. Food Microbiol. 86, 87-99.
Aremu, C.Y., Agiang, M.A. and Ayatse, J.O.I. (1995) Nutrient and antinutrient
profiles of raw and fermented cocoa beans. Plant Foods Hum. Nutr. 48, 221-223.
Arneborg N., Siegumfeldt H., Andersen G.H., Nissen P., Daria V.R., Rodrigo P.J.,
and Gluckstad J. (2005) Interactive optical trapping shows that confinement is a
determinant of growth in a mixed yeast culture. FEMS Microbiol. Lett
245,155-159.
Arikiah, A. Tan, Y.P, Sharma, M. and Clapperton, J.F. (1994) Experiments to

determine influence of primary processing parameters and planting material on
the flavour of cocoa beans in Malaysia. Cocoa Growers' Bull. 48, 36-46.
Aroyeun, S.O., Ogunbayo, J.O. and Olaiya, A.O. (2006) Effect of modified
packaging and storage time of cocoa pods on the commercial quality of cocoa
beans. Brit. Food J. 108,141-151.
Arunga, R. O. (1992). Lactic acid bacteria in coffee and cocoa fermentation. In

The Lactic Acid Bacteria ed. Wood, B.J.B. pp. 409-430. London & New York:
Elsevier Applied Science.
Assemat, S., Lachenaud, Ph., Ribeyrel, F. Davrieux, F., Pradon, J.L. and Cros, E.
(2005) Bean quality traits and sensory evaluation of wild Guianan cocoa
populations (Theobroma cacao L.) Genet. Resour. Crop Evol. 52, 911-917.
Augier, F., Nganhou, J., Barel, M., Benet, J.C. and Berthomieu, G. (1998)
Reducing cocoa acidity during drying. Plantations, Recherche, Developpement. 5,
127-133.
Bae, S., Fleet, G.H. and Heard, G.M. (2004) Significance and occurrence of
Bacillus thuringiensis on wine grapes. Inti. J. Food Microbiol. 94, 301-312.
Baigerie, B.D. and Rumbelow, S.J. (1987) Investigation of flavour defects in

Asian cocoa liquors. J. Sci. Food Agric. 39, 357-368.
Bainbridge, J.S. and Davies, S.H. (1912) The essential oil of cocoa. J. Chem. Soc.
101, 2209-2220.
358
Baker, D.M., Tomlins, K.I. and Gay, C. (1994) Survey of Ghanaian cocoa farmer
fermentation practices and their influence on cocoa flavour. Food Chem. 51,
425-431.
Balasimha, D., Bhat, N.T. and Bopaiah, B.M. (1983) Comparative chemistry and
microbiology of cocoa fermentation in India. J. Plantat. Crops. 11, 55-59.
Bangerter, V., Beh, B.H., Callis, A.B. and Bilkington, I.J. (1994) Treatment of
cocoa beans for improving fermentation. US Patent 5342632.
Barel, M.A. (1987) Delay in opening pods. Effect on yields and quality of raw
and roasted cocoa. Cafe, Cacao, The. 31, 141-150.
Barel, M.A. (1995) Cocoa processing using an integrated fermenter and dryer.
Plantation Recherche Developpement 2, 35-42.
Barel, M.A., Leon, D. and Vincent, J.C. (1985) The influence of cocoa
fermentation time on the production of pyrazines in chocolate. Cafe, Cacao, The.
29, 277-286.
Barrile, J.C., Ostovar, K. and Keeney, P.G. (1971) Microflora of cocoa beans
before and after roasting at 150°C. J. Milk Food Tech. 34, 369-371.
Bart-Plange, A. and Baryeh, E.A. (2003) The physical properties of category 'B'
cocoa beans. J. Food Eng. 60, 219-227.
Bateman, R.P. (2006) Spray application to cocoa pods and other small targets
using cone nozzles (International Advances in Pesticide Application 2006).
Aspects App. Biol. 77, 79-84.
Baylis, C.L., MacPhee, S., Robinson, A.J., Griffiths, R., Lilley, K. and Betts, R.P.
(2004) Survival of Escherichia coli 0157:H7, 0111:H- and 026:H11 in artificially
contaminated chocolate and confectionary products. Int J. Food Microbiol. 96,
35-48.
Beckett, S.T. (1999) Industrial chocolate manufacture and use. London: Blackie.
Beckett, S.T. (2000) The Science of Chocolate. Cambridge UK: The Royal Society of
Chemistry.
Beckett, S.T. (2003) Is the taste of British milk chocolate different? Int. J. Dairy
Technol. 56,139-142.
Beh, A.L., Fleet, G.H., Prakitchaiwattana, C. and Heard, G.M. (2006) Evaluation
of molecular methods for the analysis of yeasts in food and beverages. Adv. Exp.
Med. Biol. 571, 69-106.
Bekele, F.L. (2003) The history of cocoa production in Trinidad and Tobago - a
report from the University of The West Indes, Trinidad, [online] [Accessed 10
June 2006] Available from the World Wide Web: <http:/ /sta.uwi.edu/ cru/fb-
hocp-ncnp.pdf>.
359
Berbert, P.R.F. (1979) Contribution to the knowledge of the sugars component of

cocoa beans and 'sweating.' Revista Theobroma (Brazil). 9, 55-61.
Beuchat, L. R. (1995) Indigenous fermented foods. In Biotechnology (2nd edn)
eds. Rehm, H.-J., Reed, G., Puhler, A and Studler, P. pp. 505-559. Weinheim,
Germany: VCH.
Biehl, B. (1984) Cocoa fermentation and the problem of acidity,
overfermentation and low cocoa flavour. International Conference on Cocoa and
Coconuts, Kuala Lumpur, Malaysia. Pp. 1-8.
Biehl, B., Brunner, E., Passem, D., Quesnel, V.C. and Adomako, D. (1985)
Acidification, proteolysis and flavour potential in fermenting cocoa beans. J. Sci.
Food Agric. 36, 583-598.
Biehl, B., Meyer, B., Crone, G., Pollman, L. and Said, M.B. (1989) Chemical and
physical changes in the pulp during ripening and post-harvest storage of cocoa
pods. J. Sci. Food Agric. 48,189-208.
Biehl, B., Meyer, B., Said, M., and Samarakoddy, R.J. (1990). Bean spreading: A
method for pulp preconditioning to impair strong nib acidification during
cocoa fermentation in Malaysia. J. Sci. Food Agric. 51, 35-45.
Biehl, B. and Passern, D. (1982) Proteolysis during fermentation-like incubation
of cocoa seeds. J. Sci. Food Agric. 33,1280-1290.
Biehl, B., Passern, D. and Sagemann, W. (1982a) Effect of acetic acid on
subcellular structures of cocoa bean cotyledons. J. Sci. Food Agric. 33,1101-1109.
Biehl, B., Quesnel, V.C., Passem, D. and Sagemann, W. (1982b) Water uptake by
cocoa seeds during fermentation-like incubation. J. Sci. Food Agric. 33,1110-1116.
Biehl, B. Passern, U. and Passern, D. (1977) Subcellular structures in fermenting
cocoa beans. Effect of aeration and temperature during seed and fragment
incubation. J. Sci. Food Agric. 28,41-52.
Biehl, B. and Voigt, J. (1996). Biochemistry of chocolate flavour precursors. In:
International Cocoa Conference, Salvador, Bahia, November 1996.
Biehl, B. Wewetzer, C. and Passern, D. (1982c) Vacuolar (storage) proteins of
cocoa seeds and their degradation during germination and fermentation. J. Sci.
Food Agric. 33, 1291-1304.
Biomerieux, Inc. (2005) ApiWEB™ - an online rapid phenotypic identification
database. Biomerieux Corporation, USA [online]. [Accessed 6 April, 2007]
Available from the World Wide Web: < www.apiweb.biomeriux.com>.
Birch, H.F. (1941) Changes in the nitrogenous components of Forastero cocoa
during fermentation. Tenth Annual Report on Cocoa Research, Trinidad. Pp22-33.
Bisson, L.F. (1999) Stuck and sluggish fermentations. Am. J. Enol. Vitic. 50,
107-199.
360
Bisson, L.F. and Butzke, C.E. (2000) Diagnosis and rectification of stuck and
sluggish fermentations. Am. J. Enol. Vitic. 51, 168-177.
Bixler, R.G. and Morgan, J.N. (1999) cacao bean and chocolate processing In
knight etc....
Bjorkkroth, J. and Holzapfel, W. (2007) Genera Leuconostoc, Oenococcus and

Weissella. In The Prokaryotes (3rd Edn) eds. Dworkin, M., Falkow, S., Rosenberg,
E., Schleifer, K.-H. and Stackebrandt, E. New York: Springer.
Blanco, P., Sieiro, C., and Villa, T.G. (1999). Production of pectic enzymes in
yeasts. FEMS Microbiol. Eett. 175,1-9.
Boal, C. (2007) Ghana addresses cocoa labour issues. Confectionery News.Com

[online][Accessed 1 May 2008]. Available from the World Wide Web: <http: / /
www.confectionerynews.com / news>.
Bonvehi, J.S. (2004). Occurrence of ochratoxin A in cocoa products and

chocolate. J. Agric. Food Chem. 52, 6347-6352.
Bonvehi, J.S. (2005). Investigation of aromatic compounds in roasted cocoa

powder. Euro. Food Res. Technol. 221,19-29.
Bracco, U. Grailhe, N., Rostagno, W. and Egli, R.H. (1969) Analytical evaluation
of cocoa curing in the Ivory Coast. J. Sci. Food Agric. 20, 713-717.
Brand-Williams, W., Cuvelier, M.E. and Berset, C. (1995) Use of a free radical
method to evaluate antioxidant activity. Eebensm. Wiss. Technol. 28, 25-30.
Bridgland, L.A. (1959) Cacao processing - history and principles. Agricultural

Journal (Papua New Guinea) 12, 49-85.
Bright, C. (2001) Chocolate could bring the forest back. World Watch. Nov/Dec
2001 Reprinted by Ohio State University [online]. [Accessed 2 October 2006].
Available from the World Wide Web: chttp:/ / www.oardc.ohio-state.edu/
cocoa/main ftr.htm>.
Brill, FI. (1915) Enzymes of cacao. Phillipines J. Sci. 10,123-133.
Brown, H.B. (1954) Separation of pigment cells of cacao. Nature 173,492.
Brown, H.B. (1957) Changes observed in cocoa due to fermentation and their
relation to chocolate flavour. The Cocoa Conference, Eondon, 1957. The Cocoa,
Chocolate and Confectionary Alliance/Office International du Cacao et du Chocolate,
London UK. Pp. 165-175.
Buamah, R., Dzogbefia, V.P. and Oldham, J.H. (1997) Pure yeast culture
fermentation of cocoa (Theobroma cacao L.): effect on yield of sweatings and
cocoa beans quality. World J. Microbiol. Biotechnol. 13,457-462.
Bucheli, P, Rousseau, G., Alvarez, M., Laloi, M. and McCarthy, J. (2001)

Developmental variations of sugars, carboxylic acids, purine alkaloids, fatty
361
acids and endoprotease activity during maturation of Theobroma cacao L. seeds.

J. Agric. Food Chem. 49, 5046-5051.
Buckenhuskes, H.J. (1993) Selection criteria for lactic acid bacteria to be used as
starter cultures for various food commodities. FEMS Microbiol. Rev. 12, 253-272.
Bunting, R.H. (1928) Fungi occurring in cacao beans. In. Department of

Agriculture of the Gold Coast Year Book. Pp. 44-62.
Buyukpamukcu, E., Goodall, D.M., Hansen, C.E., Keely, B.J., Kochhar, S. and
Wille, H. (2001) Characterisation of peptides formed during fermentation of
cocoa bean. J. Agric. Food Chem. 49, 5822-5827.
Buzzini, P. and Martini, A. (2002) Extracellular enzymatic activity profiles in

yeast and yeast-like strains isolated from tropical environments. J. Appl.
Microbiol. 93, 1020-1025.
Bytof, G.; Biehl, B., Heinrichs, H., and Voigt, J. (1995) Specificity and stability of
the carboxypeptidase activity in ripe, ungerminated seeds of Theobroma cacao
L. Food Chem. 54,15-21.
CAA (2008) The Cocoa Association of Asia - quality standard for cocoa beans
[online]. [Accessed 3 March 2008]. Available from the World Wide Web:
<http:/ /cocoa-association-asia.org/QltyStd.html>.
Cadbury-Schweppes (2006). Company Overview [online]. [Accessed 5

September 2006]. Available from the World Wide Web: < http: / /
www.cadburyschweppes.com / EN / AboutUs / CompanvOverview / >.
Caligiani, A., Cirlini, M., Palla, G., Ravaglia, R. and Arlorio, M. (2007) GC-MS
detection of chiral markers in cocoa beans of different quality and geographic
origin. Chirality 19, 329-334.
Calver, S. (2005) Yes, chocolate is good for us. bit. Food Ingred. 2, 8-9.
Camu, N., DeWinter, T., Verbrugghe, K., Cleenwerck, I., Vandamme, P.,
Takrama, J.S., Vancanneyt, M. and De Vuyst, L. (2007) Dynamics and
biodiversity of populations of lactic acid bacteria and acetic acid bacteria
involved in spontaneous heap fermentation of cocoa beans in Ghana. Appl. Env.
Microbiol. 73,1809-1824.
Camu, N., Gonzalez, A., DeWinter, T., Van Schoor, A., De Bruyne, K.,
Vandamme, P., Takrama, J.S., Addo, S.K., DeVuyst, L. (2008a). Influences of
turning and environmental contamination on the dynamics of populations of
lactic acid and acetic acid bacteria involved in spontaneous cocoa bean
fermentation in Ghana. Appl. Env. Microbiol. 72, 86-98.
Camu, N., DeWinter, T., Addo, S.K., Takrama, J.S., Bernaert, H. and De Vuyst, L.
(2008b) Fermentation of cocoa beans: influence of microbial activities and
polyphenol concentrations on the flavour of cocoa. J. Sci. Food Agric. 88,
2288-2297.
362
Capucho, I., San Romao, M.V. (1994) Effect of ethanol and fatty acids on
malolactic activity of Leuconostoc oenos. Appl. Microbiol Biotechnol. 42, 391 - 395.
Carpenter, R.O., Lyon, D.H., and Hasdell, T.A. (2000). Guidelines for Sensory
Analysis in Food Product Development and Quality Control. Maryland, USA: Aspen
Publishers, Inc.
Carr, J.G. (1982) Cocoa In Economic Microbiology. In: Vol. 7. Fermented Foods ed
Rose, A.H. pp. 275-292. London UK: Academic Press.
Carr, J.G., Davies, PA. and Dougan, J. (1979) Cocoa fermentation in Ghana and
Malaysia - Part 1. A report available from Long Ashton Research Station, Long
Ashton, Bristol UK.
Carr, J.G., Davies, P.A. and Dougan, J. (1980) Cocoa fermentation in Ghana and
Malaysia. Part 2 - Further microbiological methods and results. A report available
from Long Ashton Research Station, Long Ashton, Bristol UK.
Carr, J.G. and Passmore, S.M. (1979) Methods for identifying acetic acid
bacteria. In Identification Methods for Microbiologists, 2nd edition eds. Skinner, F.A.
and Lovelocke, D.W. pp. 33-47. London UK: Academic Press.
Casey, J. and Dobb, R. (1992) Microbial routes to aromatic aldehydes Enzyme.
Microbiol. Technol. 14, 739-747.
Charoenchai, C., Fleet, G.H., Henschke, P.A. and Todd, B.E.N. (1997) Screening
of non-Saccharomyces wine yeast for the presence of extracellular hydrolytic
enzymes. Austral. J. Grape Wine Res. 3, 2-8.
Chatt, E.M. (1953) Cocoa: Cultivation, processing, analysis. Interscience
Publishers Inc., New York USA.
Cheesman, E.R. (1944) Notes on the nomenclature, classification and possible
relationships of cacao populations. Trop. Agric. (Trinidad) 21,144-159.
Chenynier, V. (2005) Polyphenols in foods are more complex than often thought.
Am. J. Clin. Nutr. 81, 223S-229S.
Cheraiti N., Guezenec S. & Salmon J.M. (2005) Redox interactions between
Saccharomyces cerevisiae and Saccharomyces uvarum in mixed culture under
enological conditions. Appl. Environ. Microbiol. 71, 255-260.
Chick, W.H., Mainstone, B.J. and Wai, S.T. (1981) Mitigation of cocoa acidity in
peninsular Malaysia. Proceedings of the Eighth International Cocoa Research
Conference, Cartagena, Colombia. Pp. 759-764.
Chin, A.H.G. and Zainuddin. (1984) Characteristics of Malaysian cocoa butter.
International Co?iference on Cocoa and Coconuts, Kuala Lumpur, Malaysia. Pp. 1-14.
Chong, C.F. Shepherd, R. and Poon, Y.C. (1978) Mitigation of cocoa bean
acidity-fermentary investigations. Proceedings of the biternational Conference on
Cocoa and Coconuts, Kuala Lumpur, Malaysia. Pp. 22-27.
363
Ciferri, R. (1931) Studies on cocoa: actinomycetes on cacao beans. J Dept. Agric.

(Puerto Rico) 15, 223-286.
CIRAD (2008) CIRAD - agricultural research for developing countries [online].
[Accessed 5 March 2008]. Available from the World Wide Web: <http:/ /
www.cirad.fr/ en/ index.php>.
Clark, RJ. (2007) Lessons from chocolate processing. Food Technol. 12, 89-91.
Cleenwerck, I., Camu, N., Engelbeen, K., DeWinter, T., Vandemeulebroecke, K.,
Vos, P. and de Vuyst, L. (2007) Acetobacter Ghanensis sp. nov. a novel acetic acid
bacterium isolated from traditional heap fermentations of Ghanaian cocoa
beans. Int. J. Syst. Evol. Microbiol. 57, 1647-1652.
Cleenwerck, I., Vandemeulebroecke, K., Janssens, D. and Swings, J. (2002) Re
examination of the genus Acetobacter, with descriptions of Acetobacter cerevisiae
sp. nov. and Acetobacter malorum sp. nov. Int. J. Syst. Evol. Microbiol. 52,
1551-1558.
Cocolin, L. Bisson, L.F. and Mills, D.A. (2000) Direct profiling of the yeast
dynamics in wine fermentations FEMS Microbiol.Lett. 189, 81-87
Coe, S.D. and Coe, M.D. (2003). The True History of Chocolate. New York: Thames
and Hudson.
Collins, C.H. and Lyne, P.M. (1984) Microbiological Methods, 5th edition. London
UK: Butterworths.
Confectionery Manufacturers of Australasia Limited (2004). Australian
confectionery industry profile 2004 [online]. [Accessed 2 October 2006].
Available from the World Wide Web: http: / / www.candy.net.au/cma/content/
20041015142659.asp
Contreras, C., Ortiz de Bertorelli, L., Graziani de Farinas, L. and Parra, P. (2004)
Fermentadores para cacao usados por los productores de la localidad de
Cumboto, Venezuela. Agronomia Trop. (INIA), 54, 219-232.
Cook, R.L. (1972) Chocolate Production and Use. New York USA: Books for
Industry Inc.
COPAL (2008) Cocoa Producers' Alliance. COPAL homepage [online]. [Accessed 5
December 2008]. Available from World Wide Web: chttp: / / www.copal-
cpa.org>.
Corthorne, T.G. (1987) Prospects for growing cocoa in northern Queensland.
Internal report. Queensland Dept, of Primary Industries.
Counet, C., and Collin, S. (2003) Effect of the number of flavanol units on the
antioxidant activity of procyanidin fractions isolated from chocolate. J. Agric.
Food Chem. 51, 6816-6822.
Counet, C., and Collin, S. (2005) Flavour retention and haze formation by
chocolate polyphenols. AgroFood bidust. Hi-tech. 16, 12-15.
364
Counet, C., Callemein, D., Ouwerx, C. and Colin, S. (2002) Use of gas
chromatography-olfactometry to identify key odorant compounds in dark
chocolate. Comparison of samples before and after conching. J. Agric. Food
Chem. 50, 2385-2391.
Counet, C. Ouwerx, C. Roseux, D. and Collin, S. (2004) Relationship between

procyanidin and flavour content of cocoa liquors from different origins. J.Agric.
Food Chem. 52: 6243-6249.
Cowan, S.T. and Steel, K.J. (1970) Manual for the Identification of Medical Bacteria.
London UK: Cambridge University Press.
Crouigneau, A. A., Feuillat, M., Guilloux-Benatier, M. (2000) Influence of some

factors on autolysis of Oenococcus oeni. Vitis 39,167 - 171.
Crozier, A., Borges, G. and Stewart, A.J. (2004) Absortion, metabolism and
potential bioavailability of dietary flavonols and anthocyanins. AgroFood Indust.
Hi-tech 15,16-19.
Cuatrecasas, J. (1964) Cacao and its Allies - A Taxonomic Revision of the Genus
Theobroma. Washington D.C.: Smithsonian Institute.
Dade, H.A. (1928) Internal moulding of prepared cocoa. The Department of

Agriculture of the Gold Coast Year Book, Bulletin 16, 74-100.
Dakin, K. and Wichmann, S. (2000) Cacao and chocolate: an Uto-Aztecan

perspective, Ancient Mesoamerica 11: 55-75.
Dangou, J.S., Hocher, V., Ferriere, N., Fulcheri, C., Morard, P. and Alemanno, L.
(2002) Flistological and biochemical characterization of Theobroma cacao L.
endosperm during seed development. Seed Sci. Res. 12, 91-100
Da Silva, E.G., De Fatima Borges, M., Medina, C., Piccoli, R.H., and Schwan, R.F.
(2005). Pectinolytic enzymes secreted by yeasts from tropical fruits. FEMS Yeast
Res. 5, 859-865.
De Brito, E. S., Garcia, N. H.P., Gallao, M. I., Cortelazzo, A. L., Fevereiro, P. S.

and Braga, M. R. (2000) Structural and chemical changes in cocoa (Theobroma
cacao L.) during fermentation, drying and roasting. J. Sci. Food Agric. 81, 281-
288.
De Brito, E.S., Garcia, N.H.P. and Amancio, A.C. (2004) Use of a proteolytic
enzyme in cocoa (Theobroma cacao L.) processing. Brazil. Arch. Biol. Tech. 47,
553-58.
De Camargo, R. Leme, J. and Filho, A. Martinelli. (1963) General observations

on the microflora of fermenting cocoa beans (Theobroma cacao) in Bahia
(Brazil). Food Tech. 17,116-118.
Del Boca, C. (1962) Cocoa beans: quality requirements and methods of

assessment. Int Choc. Rev. 17, 218-221.
365
Deheuvals, O., DEcazy, Bv Perez, R., Roche, G. and Amores, F. (2004) The first
Ecuadorean 'National' cocoa collection based on organoleptic characteristics.
Trop. Sci. 44, 23-27.
De Vuyst, L., Camu, N., DeWinter, T., Vandemeulebroecke, K., Vandeperre, V.,
Vancanneyt, M., De Vos, P. and Cleenwerck, I. (2008). Validation of the (GTG) 5-
rep-PCR fingerprinting technique for rapid classification and identification of
acetic acid bacteria, with a focus on isolates from Ghanaian fermented cocoa
beans. Int. J. Food Microbiol 125, 79-91.
De Witt, K.W. (1952) The enzyme of cacao tissue. In: Report on Cacao Research,
1945-1956 ed. De Witt, K.W. P.114 Trinidad: The Imperial College of Tropical
Agriculture.
De Witt, K.W. (1957) Nitrogen metabolism in fermenting cocoa. In: Report on

Cacao Research, 1955-1956 ed. De Witt, K.W. pp. 54-57 Trinidad: The Imperial
College of Tropical Agriculture.
Dias, D.R., Schwan, R.F., Freire, E.S. and dos Santos Serodio, R. (2007)
Elaboration of a fruit wine from cocoa (Theobroma cacao L.) pulp. Int. J. Food Sci.
Tech. 42, 319-329.
Dick, K.J., Molan, PC., Eschenbruch, R. (1992) The isolation from Saccharomyces
cerevisiae of two antibacterial cationic proteins that inhibit malolactic bacteria.
Vitis 31,105 - 116.
Dickson, K.B. (1960) Cocoa in Ghana. PhD. Thesis. The University of London,
London UK.
Dimick, PS. and Hoskin, J.C. (1999) The Chemistry of Flavour Development in
Chocolate. In. Beckett, S.T. Industrial Chocolate Manufacture and Use. 3rd Edition.
Pp. 137-152. Oxford: Blackwell Science.
di Tomaso, E., Beltramo, M., Piomelli, D. (1996) Brain cannabinoids in chocolate.

Nature 382, 677-678.
Dittmar, H.F.K. (1956) The composition of the pulp of various varieties of Bahia
cocoas. Gordian 56, 18-19.
Dizy, M. and Bisson, L.F. (2000) Proteolytic activity of yeast strains during grape
juice fermentation. Am. J. Enol. Vitic. 51, 155-167.
Dougan, J. (1979) A comparitive study of the fermentation of Amelonado and Amazon

cocoa carried out at the cocoa research institute, Tafo, Ghana. London UK: Cocoa,
Chocolate and Confectionary Alliance. Pp. 1-73.
Dougan, J., Duncan, R.J.E., Jardine, N.J. and Sammons, PM. (1981) Fermentation
Trials in Malaysia 1981. London UK: Cocoa, Chocolate and Confectionary
Alliance.
Dougan, J., Duncan, R.J.E., Jardine, N.J., Robinson, J.M., Sammons, P.M. and
Woodage, C. (1981) The relationship between oxygen, temperature, acetic and lactic
366
acid during cocoa fermentation - A record offermentation trials performed at CRIG,

Tafo, Ghana 1980. London UK: Cocoa, Chocolate and Confectionary Alliance.
Doutre-Roussel, C. (2005) The Chocolate Connoisseur. Piatkus: London.
Doyle, M.P., Beuchat, L.R. and Montville, T.J. (2007) Food Microbiology:
Fundamentals arid Frontiers, 3rd edition. Washington DC: ASM Press.
Drysdale, G.S. and Fleet, G.H. (1988) Acetic acid bacteria in wine making: a
review. Am. J. Enol. Vide. 39,143-154.
Duncan, R.J.E, Godfrey, G., Yap. T.N., Pettiphar, G.L. and Tharumarajah, T.
(1989) Improvement of Malaysian cocoa bean flavour by modification of
harvesting, fermentation and drying methods - The Sime Cadbury Process. The
Planter, Kuala Lumpur 65,157-173.
Du Toit, M., Pretorius, I.S. (2000) Microbial spoilage and preservation of wine:
using weapons from nature's own arsenal—a review. S. Afr. J. Enol. Vitic. 21, 74
-96.
Dworkin, M., Falkow, S., Rosenberg, E., Schleifer, K.-H. and Stackebrandt, E.
(2007) The Prokaryotes, 3rd edition. New York: Springer.
Dzogbefia, V.P., Buamah, R. and Oldham, J.H. (1999) The controlled

fermentation of cocoa (Theobroma cacao L.) using yeasts: Enzymatic process and
associated physico-chemical changes in cocoa sweatings.' Food Biotechnology
13,1-12.
Efombagn, M.I.B., Sounigo, O., Nyasse, S., Manzanares-Daulex, M., Cilas, C.,
Eskes, M.A.B. and Kolesknikova-Allen (2006) Genetic diversity in cocoa
germplasm of southern Cameroon revealed by simple sequences repeat (SSRS)
markers. Afr. J. Biotech. 5,1441-1449.
Faborode, M.O., Favier, J.F., Ajayi, O.A. (1995) On the effects of forced air drying
on cocoa quality. J. Food Eng. 25, 455-472.
Fadel, H.FLM, Abdel Mageed, M.A., Abdel Samad, A.K.M.E. and Lotfy, S.V.
(2006) Cocoa substitute: evaluation of sensory qualities and flavour stability.
Eur. Food Res. Technol. 223, 125-131.
Faparusi, S.I. (1974) The Yeasts associated with cacao (Theobroma cacao) pods.
Egypti. J. Microbiol. 9, 105-115.
Farnworth, E.R. (2003) Handbook of Fermented Functional Foods. London UK: CRC
press.
Fatichenti, F., Farris, G.A., Deiana, P. and Ceccarelli, S. (1984) Malic acid
production and consumption by selected strains of Saccharomyces cerevisiae
under anaerobic and aerobic conditions. Appl. Microbiol. Biotechnol. 19,427-M29.
Figueira, A. and Alemanno, L. (2005) Theobroma cacao cacao In: Biotechnology of

Fruit and Nut Crops pp 639-669. Wallingford, UK: CAB International.
367
Flament, I. (1991). Coffee, cocoa, and tea. In: Volatile Compounds in Foods and
Beverages ed. Maarse, H. pp 617-653. New York: Marcel Dekker, Inc.
Fleet, G.H. (1999) Microorganisms in food ecosystems. Int J. Food Microbiol. 50,
101-117.
Fleet, G.H., (2001) Wine. In: Food Microbiology Fundamentals and Frontiers, eds.
Doyle, M.P., Beuchat, L.R. and Montville, T.J. pp. 747 - 772. Washington DC:
ASM Press.
Fleet, G.H., (2003) Yeast interactions and wine flavour. Int. J. Food Microbiol. 86,
11-22.
Fleet, G.H. (2007) Yeasts in foods and beverages: impact on product quality and
safety. Curr. Opinion. Biotech 18, 170-175.
Fleet, G.H., and Praphailong, W. (2001). Yeasts. In: Spoilage of Processed Foods:
Causes and Diagnosis eds. Moir, C.J., Andrew-Kabilafkas, C., Arnold, G., Cox,
B.M., Hocking, A.D., and Jensen, I. Sydney, Australia: AIFST Food Microbiology
Group.
FLO (2008) The Fairtrade Labelling Organization [online]. [Accessed: 1 May

2008] Available from the World Wide Web: chttp:/ / www.fairtrade.net>.
Fornachon, J.C.M. (1968) Influence of different yeasts on growth of lactic acid

bacteria in wine. J. Sci. Food Agric. 19, 374 - 378.
Forsyth, W.G.C. (1949) A method for studying the chemistry of cacao

fermentation. Nature 164, 25-26.
Forsyth, W.G.C. (1953) Purple beans. Report to the Cocoa Conference, London. The
Cocoa, Chocolate and Confectionary Alliance, London UK. Pp. 32-34.
Forsyth, W.G.C. (1955) Cacao polyphenolic substances: 3 - Separation and

estimation on paper chromatograms. Biochem. J. 60, 108-111.
Forsyth, W.G.C. (1957) An appraisal of fundamental research on cocoa curing at

the colonial microbiological research institute. Report to the Cocoa Conference,
London. The Cocoa, Chocolate and Confectionary Alliance, London UK. Pp. 145-150
Forsyth, W.G.C. and Rombouts, J.E. (1951) Our approach to the study of cocoa
fermentation. Fourth Session of the Cocoa Conference, London. The Cocoa, Chocolate
and Confectionary Alliance, London UK. Pp. 73-81.
Forsyth, W.G.C. and Quesnel, V.C. (1957a) Variations in cocoa preparation. Sixth
Meeting of the Inter-American Cacao Committee, Bahia, Brazil. Pp. 157-168.
Forsyth, W.G.C. and Quesnel, V.C. (1957b) Cacao polyphenol substances 4: The
anthocyanin pigments. Biochem. J. 65,177-179
Forsyth, W.G.C. and Quesnel, V.C. (1957c) Cacao glycosidase and colour
changes during fermentation. J. Sci. Food Agric. 8, 505-509.
368
Forsyth, W.G.C. and Quesnel, V.C. (1963) The mechanism of cocoa curing. Adv.
Enzym. 25,457-492.
Forsyth, W.G.C., Quesnel, V.C. and Roberts, J.B. (1958) The interaction of
polyphenols and proteins during cacao curing. J. Sci. Food Agric. 9,181-184.
Fowler, M.S., Leheup R, and Cordier J.-L. 'Cocoa, coffee and tea.' In: (1998)
Microbiology offermented foods, Volume 1, 2nd edition ed. Wood B.J.B. pp. 128 -145.
London UK: Thompson Science.
Franzen, M. and Mulder, M.B. (2007) Ecological, economic and social
perspectives on cocoa production worldwide. Biodivers. Conserv. 16, 3835-3949.
Fraudorfer, F. and Schieberle, P. (2006) Identification of the key aroma
compounds in cocoa powder based on molecular sensory correlations. J. Agric
Food Chem. 54, 5521-5529.
FSANZ (2008) FSANZ and health claims [online]. [Accessed 1 November 2008].
Available from World Wide Web: chttp: / / www.foodstandards.gov.au /
foodmatters / healthnutritionandrelatedclaims / index. cfm>.
Fugelsang, K.C. (1997) Wine Microbiology. New York: Chapman & Hall.
Gabis, D.A., Langlois, B.E. and Rudnick, A.W. (1970) Microbiological
examination of cocoa powder. Appl. Microbiol. 20, 644-645.
Gaeraert, E. and Sandra, P. (1987) Capillary Gas Chromatography of
triglycerides in fats and oils using a high temperature phenylmethylsilicone
stationary phase. Part II - The analysis of chocolate fats. J. Am. Oil Chem. Soc.
64,100-105.
Galvez, S.L., Loiseau, G., Paredes, J., Barel, M. and Guiraud, J.-P. (2007) Study
on the microflora and biochemistry of cocoa fermentation in the Dominican
Republic. Int. J. Food Microbiol. 114, 124-130.
Garcia-Alamill, P., Salgado-Cervantes, M.A., Barel, M., Berthomia, G.,
Rodriguez-Jimines, G.C. and Garcia-Alvarado, M.A. (2007) Moisture, acidity
and temperature evolution during cacao drying. J. Food Fng. 79, 1159-1165.
Gauthier, B. Guiraud, J., Vincent, J.C., Parvais, J.P and Galzy, P. (1977)
Remarques sur la flore de leuvres de la fermentation traditionanelle du cacao en
Cote d'Ivoire. Rev. Ferment. Indep. Alim. 32,160-163.
Gilbert, D.G. (1980) Dispersal of yeasts and bacteria by Drosophilia in a
temperate forest. Oecologia. 46, 135-137.
Giraffa, G. and Neviani, E. (2001) DNA-based, culture-independent strategies
for evaluating microbial communities in food- associated ecosystems. Int. J.
Food Microbiol. 67,19-34.
Glendinning, D.R. (1967) Technical aspects of the breeding programme at the
Cocoa Research Institute, Tafo, Ghana: breeding methods. Fuphytica 16, 76-82.
369
Goto, Y., Aihara, T., Okuyama, S., Kamiwaki, T., Tsukada, K., and Uzawa, M.
(2002) Compositional changes of Venezuelan cacao beans during the
fermentation. Nippon Shokuhin Kagaku Kogaku Kaishi 49, 731-735.
Gotsch, N. (1997) Cocoa biotechnology: status, constraints and future prospects.
Bioteclmology Advances 15, 333-352.
Granvogl, M., Bugan, S. and Schieberle, p. (2006) Formation of amines and
aldehydes from parent amino acids during thermal processing of cocoa and
model systems: new insights into pathways of the Strecker reaction. J. Agric.
Food Chem. 54,1730-1739.
Grassi, D., Lippi, C., Necozione, S., Desideri, G. and Ferri, C. (2005) Short term
administration of dark chocolate is followed by a significant increase in insulin
sensitivity and a decrease in blood pressure in healthy persons. Am. J. Clin.
Nutr. 81, 611-614.
Greenwood, H. (2005) Hot Chocolates - the single-origin concept takes one of
life's great pleasures to a whole new level. The Sydney Morning Herald. Feb 1,
2005.
Griffiths, L.A. (1957) On the optimal conditions for microfermentation: a critical
review. Rep. Cocoa Res. Trinidad 1957/58, 70-73.
Gu, L., House, S.E., Wu, X., Ou, B. and Prior, R.L. (2006) Procyanidin and
catechin contents and antioxidant capacity of cocoa and chocolate products. J.
Agric. Food Chem. 54, 4057-4061.
Guilloteau, M., Laloi, M., Michaux, S., Bucheli, P, and McCarthy, J. (2005).
Identification and characterization of the major aspartic proteinase activity in
Threobroma cacao seeds. J. Sci. Food Agric. 85, 549-562.
Guilloux-Benatier, M., Le Fur, Y., Feuillat, M., (1998) Influence of fatty acids on
the growth of wine microorganisms Saccharomyces cerevisiae and Oenococcus oeni.
J. Ind. Microbiol. Biotech. 20,144-149.
Guittard, G.W. (2005) Origin cocoa: vive la difference. Manuf. Confect. 85, 81-88.
Guyton, W. (2007) World Cocoa Foundation - encouraging sustainable,
responsible cocoa growing [online]. [Accessed 20 December, 2008]. Available
from World Wide Web: <http:/ / www.worldcocoafoundation.org>
Hammes, W.P and Hertel, C. (2007) The genus Lactobacillus. In: The Prokaryotes,
3rd edition eds. Dworkin, M., Falkow, S., Rosenberg, E., Schleifer, K.-H. and
Stackebrandt, E. New York: Springer.
Hankin, L. and Lacy, H. (1984) Pectinolytic microorganisms. In Compendium of
methods for the microbiological examination offoods. 2nd edition, ed. Speck, M.L. pp.
176-183. Washington DC: American Public Health Association.
Hansen, A.P. (1975a) Microbiological activity and its effect on cocoa beans.
Manufacturing Confectioner 55, 35-39.
370
Hansen, A.P. (1975b) Understanding the microbiological deterioration of cacao.

Candy and Snack Industry 140,44-47.
Hansen, B. and Hansen, A. (1994) Volatile compounds in wheat sourdoughs

produced by lactic acid bacteria and sourdough yeasts. Z. Lebensm. Unters.
Forsch. 198, 202-209.
Hansen, A.P. and Welty, R. (1970) Microflora of raw cocoa beans. My copath.
Mycol. Appl. 44, 309-316.
Hansen, C.E., Del Olmo, M., and Burri, C. (1998) Enzyme activities in cocoa
beans during fermentations. J. Sci. Food Agric. 77, 273-281.
Hansen, C.E., Klueppel, A. and Raetz, E. (1999) Enzymatic treatment of cocoa. US

patent 660901, 5,888,562.
Hansen, C.E., Manez, A., Burri, C. and Bousbaine, A. (2000) Comparison of

enzyme activities involved in flavour precursor formation in unfermented
beans of different cocoa genotypes. J. Sci. Food Agric. 80,1193-1198.
Hardy, F and Rodrigues, G. (1952) Quantitative variations in nitrogenous

components of the cocoa bean: effect of genetic type and soil type. A report on
Cocoa Research 1945-1951. pp. 89-91. Trinidad: The Imperial College of Tropical
Agriculture.
Harmon, A.D. (1997). Solid-phase microextraction for the analysis of flavours.

In: Techniques for Analyzing Food Aroma ed. Marsili, R. pp. 81-112. New York:
Marcel Dekker, Inc.
Harrigan, W.F. and McCance, M.E. (1976) Laboratory Methods in Food and
Dairy Microbiology. London UK: Academic Press.
Hashim, L. and Chaveron, H. (1994) Extraction and determination of

methylpyrazines in cocoa beans using coupled steam distillation-
microdistillator. Food Res. Int. 27, 537-544.
Hashim, P., Jinap, S., Muhammad, S.K.S. and Ali, A. (1998a) Changes in free
amino acid, peptide-N, sugar and pyrazine concentration during cocoa
fermentation. J. Sci. Food Agric. 78, 535-542.
Hashim, R, Jinap, S., Muhammad, S.K.S. and Ali, A. (1998b) Effect of mass and
turning time on free amino acid, peptide-N, sugar and pyrazine concentration
during cocoa fermentation. J. Sci. Food Agric. 78, 543-550.
Hashim, P, Jinap, S., Muhammad, S.K.S. and Ali, A. (1999) Effect of drying time,
bean depth and temperature on free amino acid, peptide-N, sugar and pyrazine
concentrations of Malaysian cocoa beans. J. Sci. Food Agric. 78, 543-550.
Hemme, D. and Foucaud-Scheunemann, C. (2004) Leuconostoc, characteristics,

use in dairy technology and prospects in functional foods. Int. Dairy. J. 14,
467-494.
371
Heo, J.H. and Lee, C.Y. (2005) Epicatechin and catechin in cocoa inhibit amyloid
3 protein induced apoptosis. J.Agric. Food Chem. 53,1445-1448.
Herraiz, T. (2000) Tetrahydro-3-carbolines, potential neuroactive alkaloids in

chocolate and cocoa. J. Agric. Food Chem. 48, 4900-4904.
Hesseltine, C.W. and Wang, H.L. (1986) Indigenous fermented foods of non-western
origin. Berlin; J-Cramer.
Hi, C.L., Rahman, R.A., Jinap, S. and Che-Man, Y.B. (2006) Quality of cocoa
beans dried using a direct solar dryer at different loadings. J. Sci. Food Agric. 86,
1237-1243.
Hogan D.A. (2005) Quorum sensing: alcohols in a social situation. Curr. Biol. 16,
R457-R458.
Holden, M. 1959. Processing of raw cocoa III - Enzymic aspects of cocoa

Hollenberg, N.K., Schmitz, H., Macdonald, I. and Poulter, N. (2004) Cocoa,

flavanols and cardiovascular risk. Brit. J. Cardiol. 11, 379-386.
Hollywood, N. (1998) The effect of fermentation time and washing of cocoa

prior to drying on cocoa quality in Papua New Guinea. Cocoa Grow. Bull. 51,
23-32.
Halm, M., Hornbaeker, T., Arneborg, N., Sefa-Dedeh, S. and Jespersen, L. (2004)
Lactic acid tolerance determined by measurement of intracellular pH of single
cells of Candida krusei and Saccharomyces cerevisiae isolated from fermented
maize dough. Int. J. Food Microbiol. 94, 97-103.
Holm, C.S., Aston, J.W., and Douglas, K. (1993). The effects of the organic acids
in cocoa on the flavour of chocolate. J. Sci. Food Agric. 61, 65-71.
Holmes, K.E. (1950) Determination of theobromine in cocoa products. Analyst

75, 457-461.
Holzapfel, W.H. (1992) Culture media for non-sporulating Gram-positive food

spoilage bacteria. Int. J. Food Microbiol. 17, 113-133.
Holzapfel, W.H. (2002) Appropriate starter culture technologies for small scale
fermentation in developing countries. Int. J. Food Micro. 75,197-212.
Holzapfel, W.H., Franz, C.M.A.P, Ludwig, W., Back, W. and Dicks, L.M.T.
(2007) The genera Pediococcus and Tetragenococcus. In: The Prokaryotes, 3rd edition
eds. Dworlan, M., Falkow, S., Rosenberg, E., Schleifer, K.-H. and Stackebrandt,
E. New York: Springer.
Hoskin, J.C. (1994). Sensory properties of chocolate and their development. Am.
J. Clin. Nutr. 60,1068S-1070S.
372
Hoskin, J.C., and Dimick, P.S. (1994). Chemistry of flavour development in

chocolate. In: Industrial chocolate manufature and use. 2nd Edition, eds. Beckett, S.
T., (Ed), pp 102-115. New York: Van Nostrand Reinhold.
Hounhouigan, D.J., Nout, M.J.R., Nago, C.M., Houben, J. H., Rombouts, F.M.
(1999) Use of starter cultures of lactobacilli and yeast in the fermentation of
mawe, an African maize product. Trop. Sci. 39, 220 - 226.
Howell, K.S., Cozzolino, D., Bartowsky, E.J., Fleet, G.H. and Henschke, RA.
(2006) Metabolic profiling as a tool for revealing Saccharomyces interactions
during wine fermentation. FEMS Yeast Res. 6, 91-101.
Hoynak, S., Polansky, T.S. and Stone, R.W. (1941) Microbiological studies of
cacao fermentation. Food Res. 6,471-479.
Huang, Y.-C., Edwards, C.G., Peterson, J.C. and Haag, K.m. (1996) Relationship
between sluggish fermentations and the antagonism of yeast by lactic acid
bacteria. Am. J. Enol. Vitic. 47,1-10.
Hugenholtz, J. (1993) Citrate metabolism in lactic acid bacteria. FEMS Microbiol.
Rev. 12:165-178.
Humphries, E.C. (1939a) Changes in fat and theobromine content of the kernel
of the cocoa bean during fermentation and drying. 8th Annual Report on Cocoa
Research, Trinidad. Pp. 34-36.
Humphries, E.C. (1939b) A new method for estimating total alkaloids in cacao.
8th Annual Report on Cocoa Research, Trinidad. Pp. 36-37.
ICCO (2005a) ICCO Annual report 2003/2004. International Cocoa Organization,
Fondon U.K.
ICCO (2005b) ICCO document: facts and figures on fair-trade cocoa.
International Cocoa Organization, Fondon U.K.
ICCO (2005c) ICCO document: inventory of the health and nutritional
attributes of cocoa and chocolate. International Cocoa Organization, London, U.K.
ICCO (2006) ICCO document: a study on the market for organic cocoa.
International Organization, Fondon, U.K.
ICCO (2007a) ICCO document: assessment of the movements of global supply
and demand. International Cocoa Organization, Fondon, U.K.
ICCO (2007b) ICCO document: Annual forecasts of production and
consumption and estimates of production levels to achieve equilibrium in the
world cocoa market. International Cocoa Organization, Fondon, U.K.
ICCO (2007c) ICCO document: overview of “best known practices" in cocoa
production. International Cocoa Organization, Fondon, U.K.
ICCO (2007d) ICCO document: supply chain management for total quality
cocoa in Africa. International Cocoa Organization, Fondon, U.K.
373
ICCO (2008) ICCO document: progress report on projects. International Cocoa

Organization, London, U.K.
ICMSF (2000) Microorganisms in foods: microbial ecology of food commodities

(2nd ed.) Kluwer Academic, New York. Pp. 467-479.
IFOAM (2008) The International Federation of Organic Agriculture Movements

[online]. [Accessed 20 December, 2008]. Available from World Wide Web:
chttp:/ / www.ifoam.org>.
Jabugun, J.A. (1965) A mechanical cocoa pod sheller. Niger. Agric. J. 2, 44-55.
Jespersen, L. (2003) Occurrence and taxonomic characteristics of strains of

Saccharomyces predominant in African indigenous fermented food and
beverages. FEMS Yeast Res. 3, 191-200.
Jespersen, L., Nielsen, D.S., Honholt, S., and Jakobsen, M. (2005). Occurrence
and diversity of yeasts involved in fermentation of West African cocoa beans.
FEMS Yeast Res. 5,441-453.
Jinap, S. (1994) Organic acids in cocoa beans - a review. ASEAN Food J. 9, 3-12.
Jinap, S. and Dimick, P.S. (1990a) Analysis of non-volatile organic acids in

fermented and dried cocoa beans by high performance liquid chromatography.
Pertanika 13,107-111.
Jinap, S and Dimick, P.S. (1990b). Acidic characteristics of fermented and dried
cocoa from different countries of origin. J. Food Sci. 55, 547-550.
Jinap, S., Dimick, PS., and Hollender, R. (1995). Flavour evaluation of chocolate
formulated from cocoa beans from different countries. Food Control 6,105-110.
Jinap, S. Rosli, W.I.W., Russly, A.R. and Nordin, L.M. (1998) Effect of roasting
time and temperature on volatile component profiles during nib roasting of
cocoa beans (Theobroma cacao). J.Sci Food Agric. 77: 441-448.
Jinap, S., Thien, J. and Yap, T.N. (1991) Effect of drying on acidity, volatile fatty
acid and pyrazine content of cocoa beans In: 45th P.M.C.A. Production conference.
Pp 71-78.
Johns, N.D. (1999) Conservation in Brazil's chocolate forest: the unlikely

persistence of the traditional cocoa agroecosystem. Env. Manag. 23, 31-47.
Jones, K.l and Jones, S.E. (1984) Fermentations involved in the production of
cocoa, coffee and tea. Prog. Indust. Microbiol. 19, 411-433.
Kado, C.I. (2007) Erwinia and related genera. In: Dworkin, M., Falkow, S.,
Rosenberg, E., Schleifer, K.-H. and Stackebrandt, E. (Eds.) The Prokaryotes (3rd
Ed'n) New York: Springer.
Kaneuchi, C., Seki, M. and Komagata, K. (1988) Production of succinic acid from
citric acid and related acids by Lactobacillus strains. Appl. Environ. Microbiol 54,
3053-3056.
374
Karen, I.I., Liau, T.L.H and Lee, M.T. (1983) A brief account of the reduction of
cocoa bean acidity and recent experience in two fermentaries in Sabah. Seminar
on Primary Cocoa Processing and Quality Control, Tazvau Sabah. Pp. 1-9.
Karlin, S. and Altschul, F. (1990) Methods for assessing the statistical
significance of molecular sequence features by using general scoring schemes.
Proc. Natl. Acad. Sci. U.S.A. 87, 2264-2268.
Katsura, K., Kawasaki, H., Potacharoen, W., Saono, S., Seki, T., Yamada, Y.,
Uchimara, T. and Komagata, K. (2001) Asaia siamensis sp. nov., an acetic acid
bacterium in the alpha proteobacteria. Int. J. System. Evol. Microbiol. 51, 559-563.
Kealey, K.S., Snyder, R.M., Romancyzk, L.J., Geyer, H.M., Myers, M.E.,
Whitacre, E.J., Hammerstone, J.F. and Schmitz, H.H. (2001) Cocoa components,
edible products having enriched polyphenol content, methods of making same
and medical uses. US Patent 6312753, 254353.
Keen, C.L., Holt, R.R., Oteiza, PI., Fraga, C.G. and Schmitz, H.H. (2005) Cocoa
antixidants and cardiovascular health. Am. J. Clin. Nutr. 81(Suppl.): 298S-303S.
Keeney, PG. (1976) Extraction and fractionation of cocoa proteins as applied to
several varieties of cocoa beans. J. Agric. Food Chem. 24, 479-483.
Keeney, P.G. and Zak, D.L. (1974). Studies on the proteins of cocoa beans. Int.
Kongress Kakao-Schokoladefrosch., [Vorabdrukel], Institut fiir. Lebensmitteltechnologie
und Verpackung, Germany. 1, 135-139.
Keller, M., Smityhman, R.P. and Mills, L.J. (2008) Interactive effects of deficit
irrigation and crop load on Cabernet Sauvignon in an arid climate. Am. J. Enol.
Vitic. 59, 221-234.
Kenten, R.H. and Powell, B.D. (1960) Production of heat during fermentation of
cacao beans. J. Sci. Food Agric. 11, 396-400.
Kersters, K., Lisdiyanti, P, Komagata, K. and Swings, J. (2007) The family
Acetobacteriaceae: The genera Acetobacter, acidomonas, Asaia, Gluconobacter, and
Kozakia. In: Dworkin, M., Falkow, S., Rosenberg, E., Schleifer, K.-H. and
Stackebrandt, E. (Eds.) The Prokaryotes (3rd Ed'n) New York: Springer.
Kerti, K. (2001) Investigating isothermal DSC method to distinguish between
cocoa butter and cocoa butter alternatives. J. Therm. Anal. Calor. 63, 205-219.
Kim, H. and Keeney, P.G. (1984). (-)-Epicatechin content in fermented and
unfermented cocoa beans. J. Food Sci. 49,1090-1092.
King, A. and Dickinson, J.R. (2000) Biotransformation of monoterpene alcohols
by Saccharomyces cerevisiae, Torulaspora delbrueckii and Kluyveromyces lactis. Yeast.
16,499-506.
Kiransree, N., Sridhar, M., Venkateswar Rao, L. (2000). Characterisation of
thermotolerant ethanol tolerant fermentative Saccharomyces cerevisiae for ethanol
production. Biopr. Eng. 22, 243-246.
375
Kirchoff, P.M., Biehl, B. and Crone, G. (1989) Peculiarity of the accumulation of

free amino acids during cocoa fermentation. Food Chem. 31, 295-311.
Kirchoff, P.M., Biehl, Ziegeler-Berghausen, H., Hammoor, M. and Lieberi, R.
1989. Kinetics of the formation of free amino acids in cocoa seeds during
fermentation. Food Chem. 34, 161-179.
Kirk, R.S. and Sawyer, R. (1991) Pearson's composition and aiialysis offoods 9th
Ed'n. Harlow, UK: Longman.
Knapp, A.W. (1926) Experiments in the Fermentation of cacao. /. Soc. Chem. Ind.
45,140T-142T.
Knapp, A.W. (1934) Cacao fermentation in West Africa. /. Soc. Chem. Ind. 53,
151T-158T.
Knapp, A.W. (1937) Cacao fermentation. John Bale, Sons and Cumow Ltd.,
London.
Knight, I. (ed.) (1999) Chocolate and Cocoa: Health and Nutrition. Oxford, UK:
Blackwell Science.
Kochnar, S., Hansen, C.E. and Juillerat, M.A. (2002) Cocoa polypeptides and
their uses in the production of cocoa and chocolate flavour European Patent
EP1253200.
Kostinek, M., Specht, I., Edward, V.A., Schillinger, U., Hertel, C., Holzapfel,
W.H. and Franz, C.M.A.P. (2005) Diversity and technological properties of
predominant lactic acid bacteria from fermented cassava used for the
preparation of Gari, a traditional African food. Syst. Appl. Microbiol. 28, 527-540.
Kochhar, S., Gartenmann, K. and Juillerat, M.A. (2000) Primary structure of
abundant seed albumin of Theobroma cacao by mass spectrometry. /. Agric Food
Chem. 48, 5593-5599.
Kosuge, T. and Kamiya, H. (1962). Discovery of pyrazine in a natural product:
Tetramethylpyrazine from cultures of a strain of Bacillus subtilis. Nature 193, 776.
Kreger-van Rij, N. J. W. (ed.). (1984) The yeasts: a taxonomic study, 3rd ed.
Amsterdam, The Netherlands: Elsevier.
Kreiser, W.R. and Martin, R.A. Jr. (1978) Cacao Products. High Pressure Liquid
Chromatographic determination of theobromine and caffeine in cocoa and
chocolate products. J. Assoc. Off. Anal. Chem. 61, 1424-1427.
Krieg, N.R. and Holt, J.G. (ed.) (1984) Bergey's Manual of Systematic Bacteriology.
Vol. 1. London, UK: Williams and Wilkins Co.,
Kris-Etherton PM, Derr JA, Mitchell DC, et al. (1993) The role of fatty acid
saturation on plasma lipids, lipoproteins and apoliproteins. I. Effects of whole
food diets high in cocoa butter, olive oil, soybean oil, dairy butter and milk
chocolate on the plasma lipids of young men. Metabolism. 42,121-129.
376
Kris-Etherton, P.M., Lefevre, M., Beecher, G.R., Gross, M.D., Keen, C.L. and
Etherton, T.D. (2004) Bioactive compounds in nutrition and health research
methodologies for establishing biological function: the antioxidant and
antiinflammatory effects of flavonoids on arthersclerosis. Annu. Rev. Nutr. 24,
511-538.
Krysiak, W. (2006) Influence of roasting conditions on coloration of roasted

cocoa beans. J. Food Eng. 77, 449-453.
Kurtzmann, C.P. and Fell, J.W. (1998) The Yeasts - A Taxonomic Study (4th ed).
New York: Elsevier.
Kurtzmann C.P. and Robnett C.J. (1998) Identification and phylogeny of

ascomycetous yeasts from analysis of nuclear large subunit (26S) ribosomal
DNA partial sequences. Antonie van Leeuwenhoek 73, 331 - 371.
Kushnirov, V.V. (2000) Rapid and reliable protein extraction from yeast Yeast 16,
857-860.
Kyi, T.M., Daud, VV.R.W, Mohammad, A.B., Samsudin, M.W., Kadhum, A.A.H.
and Talib, M.Z.M. (2005) The kinetics of polyphenol degradation during the
drying of Malaysian cocoa beans, hit. J. Food Sci. Tech. 40, 323-331.
Laloi, M., McCarthy, J., Morandi, O., Gysler, C., and Bucheli, P. (2002).
Molecular and biochemical characterisation of two aspartic proteinases TcAPl
and TcAP2 from Theobroma cacao seeds. Planta 215, 754-762.
Lallemand (2007) Fermaid K - product data sheet [online]. [Accessed: 1 fune,

2008]. Available from the World Wide Web: <http:/ / www.lallemandwine.com /
catalog /products /view /1999>.
Lamuela-Raventos, R.M., Romero-Perez, A.I., Andres-Lacueva, C. and Tornero,

A. (2005) Review: Health effects of cocoa flavonoids. Food Sci. Tech. Int. 11,
159-179.
Lass, R.A. (1999). Cacao growing and harvesting practices. In: Knight, I. (Ed).
Chocolate and Cocoa: Health and Nutrition. Oxford, UK: Blackwell Science.
Laycock, T. 1930. Experiments on the fermentation and moulding of cacao.

Ninth Annual Bulletin of the Agricultural Department of Nigeria. Pp. 5-26.
Lei, V., Amoa-Awua, W.K.A. and Brimer, L. (1999) Degradation of cyanogenic

glycosides by Lactobacillus plantarum strains from spontaneous cassava
fermentation and other microorganisms. Int. J. Food Microbiol. 53, 169-184.
Lerceteau, E., Rogers, J., Petiard, V. and Crouzillat, D. (1999) Evolution of cacao
bean proteins during fermentation: a study by two-dimensional electrophoresis.
J. Sci Food Agric. 79, 619-625.
Lee, K.W., Kim, Y.J., Lee, H.J. and Lee, C.Y. (2003) Cocoa has higher
polyphenolic phytochemicals and a higher antioxidant capacity than teas and
red wine. J. Agric. Food Chem. 51, 7292-7295.
377
Lees, R. and Jackson, E.B. (1973). Sugar Confectionery and Chocolate

Manufacture. UK: Leonard Hill Books.
Lehrian, D.W. and Keeney, RG. (1980a) Changes in lipid components of seeds
during growth and ripening of cacao fruit. /. Amer. Oil Chem. Soc. 57, 61-65.
Lehrian, D.W. and Keeney, P.G. (1980b) Triglyceride characteristics of cocoa

butter from cocoa fruit matured in a microclimate of elevated temperature. J.
Amer. Oil Chem. Soc. 57, 66-69.
Lehrian, D. W., and G. R. Patterson. (1983) Cocoa fermentation, In G. Reed (ed.),

Biotechnology, a comprehensive treatise, Vol. 5. Pp. 529-575. Basel, Switzerland:
Verlag Chemie.
Lemin, C. (2005a) Cocoa: a potential new crop for Northern Australia? [online].
[Accessed 12 December, 2007]. Available from the World Wide Web: <http: / /
www2.dpi.qld.gov.au/horticulture/ 6223.html>.
Lemin, C. (2005b) Growing cocoa [online]. [Accessed 12 December, 2007].

Available from the World Wide Web: <http: / / www2.dpi.qld.gov.au/
horticulture/ 6222.html>.
Leston, D. (1970) Entomology of the cocoa farm. Annu. Rev. Entomol. 15, 273-294.
Levanon, Y. and Rossetini, S.M.O. (1965) A laboratory study of farm processing

of cocoa beans for industrial use. J. Food Sci. 30, 719-722.
Li, C. and Sun, W.Q. (1999) Desiccation sensitivity and activities of free radical
scavenging enzymes in recalcitrant Theobroma cacao seeds. Seed Sci. Res. 9,
209-217.
Liau, H.T.L. (1976) Raw cocoa processing: Acidity and flavour. Proceedings of the
East Malaysia Planters Association Cocoa-Coconut Seminar, Tawau, Sabah.
Pp. 184-196.
Liau, H.T.L. (1978) The criteria and mechanism for the removal of cocoa bean
acidity. Malaysian Agricultural Research and Development Institute Paper No. 24.
International Conference on Cocoa and Cocoanuts, Kuala Eumpur. Pp. 425-439.
Liendo, R., Padilla, F.C. and Quintana, A. (1997) Characterization of cocoa

butter extracted from Criollo cultivars of Theobroma cacao L. Food Res. Int.
30, 727-731.
Loew, O. (1913) Chapter II. In: The Fermentation of Cacao. Smith, H.H.(ed.). Pp.
32-65. London UK: John Bale, Sons and Danielson.
Lohachoompol, V. (2007) Effects of Drying on Anthocyanins in Blueberries [PhD

Thesis]. The University of New South Wales, Sydney.
Lonvaud-Funel, A., Joyeux, A., Desens, C. (1998) Inhibition of malolactic

fermentation of wines by products of yeast metabolism. J. Sci. Food Agric.
44,183-191.
378
Lopez, A.S. (1972) The development of chocolate aroma from non-volatile

precursors. 4th International Cocoa Research Conference, Trinidad. Pp.640-646.
Lopez, A.S. (1979) Fermentation and organoleptic quality of cacao as affected by
partial removal of pulp juices from the beans prior to curing. Rev. Theobroma
9, 25-37.
Lopez, A.S. (1983) Factors associated with cocoa bean acidity and the possibility
of its reduction by improved fermentation. Rev. Thebroma 13, 233-248.
Lopez, A.S. and Dimick, P.S. (1995). Cocoa fermentation. In: Biotechnology - A
Multivolume Comprehensive Treatise. Volume 9: Enzymes, Biomass, Food and Feed.
Rehm, H.J. and Reed, G. (Eds). Pp. 562-577. Weinheim, Germany: Chemie-
Verlag.
Lopez, A.S., Lehrian, D.W. and Lehrian, L.W. (1978) Optimum temperature and
pH of invertase of the seeds of Thebroma cacao L. Rev. Theobroma 8,105-112.
Lopez, A.S. and Quesnel, V.C. (1973a) Production of ethanol and acetic acid
during fermentation. Annual Report on Cocoa Research, University of West bidies,
Trinidad. Pp.39-50.
Lopez, A.S. and Quesnel, V.C. (1973b) Volatile fatty acid production in cocoa
fermentation and the effect on chocolate flavour. J. Sci. Food Agric. 24, 319-326.
Lopez, A.S. and McDonald, C.R. (1981) A definition of descriptors to be used for
the qualification of chocolate flavours in organoleptic testing. Rev. Theobroma 11,
209-217.
Lopez, I., Ruiz-Larrea, F., Cocolin, L., Orr, E., Phister, T., Marshall, M,
VenderGheynst, J. and Mills, D.A. (2003) Design and evolution of PCR primers
for analysis of bacterial populations in wine by denaturing gel electrophoresis.
Appl. Environ. Microbiol 69, 6801-6807.
Liicke, F.-K., Brummer, J.-M., Buckenhuskes, H., Garrido Fernandez, A.,
Rodrigo, M., Smith, J.E. (1990) Starter culture development. In: Processing and
Quality of Foods - Volume 2: Food Biotechnology: Avenues to Healthy and Nutritious
Products. Zeuthen, P, Cheftel, J.C. Eriksson, C. Gormley, T.R. Linko, P. and
Paulus, K. (eds.). Pp. 2.11 - 2.36. London: Elsevier Applied Science.
Luna, F., Crouzillat, D., Cirou, L. and Bucheli, P. (2002) Chemical composition
and flavour of Ecuadorian cocoa liquor. J. Agric. Food Chem. 50, 3527-3532.
McHenry, L. and Fritz, P.J. (1992) Comparison of the structure and nucleotide
sequences of vicilin genes of cocoa and cotton raise questions about vicilin
evolution. Plant Molec. Biol. 18, 1173-1176.
McKay, I.A., Maddox, I.S. and Brooks, J.D. (1990) Citrate production by yeast.
In: Fermentation Technologies: Industrial Applications. Yu, P.L.(ed.). Pp. 285-291.
London: Elsevier Applied Science.
Maduako, J.N. and Faborode, M.O. (1994) Characterization of the breaking
behaviour of whole cocoa pods. J.Agric. Eng. Res. 59, 89-96.
379
Magliani, Wv Conti, S., Gerloni, M., Bertolotti, D. and Polonelli, L. (1997) Yeast
Killer systems. Clin. Microbiol Rev. 10, 369-380.
Makivuokko, H., Kettunen, H., Saarinen, M., Kamiwaki, T., Yokoyama, Y.,
Stowell, J. and Rautonen, N. (2007) The effect of cocoa and polydextrose on
bacterial fermentation in gastrointestinal tract simulations. Biosci. Biotechnol.
Biochem. 71,1834-1843.
Maravalhas, N. (1966) Mycological deterioration of cocoa beans during

fermentation and storage in Bahia. Int. Choc. Rev. 21, 375.
Maravalhas, N. (1972) Amino acids in fermented and unfermented cocoa beans.

International Chocolate Review 27, 22-23.
Markides, A. (1993) Factors influencing the growth of malolactic bacteria and

malolactic activity in wine—interactions between wine yeast and lactic acid
bacteria. Aust. Grapegrow. Winemak. 352,108 - 111.
Martelli, H.L., and Dittmar, H.F.K. (1961). Cacao fermentation. V. Yeasts isolated
from cacao beans during the curing process. App. Microbiol. 9, 370-371.
Martin, R.A. (1987) Chocolate. Adv. Food Res 31, 211-342.
Masoud, J. and Jespersen, L. (2006) Pectin degrading enzymes in yeasts

involved in fermentation of Coffea arabica in East Africa. Int. J. Food Microbiol.
110, 291-296.
Masoud, W. and Kaltoft, C.H. (2006) The effects of yeasts involved in the
fermentation of Coffea arabica in East Africa on growth and ochratoxin A (OTA)
production by Aspergillus ochraceus. Int. J. Food Microbiol. 106, 229-234.
Menendez, J. (2004) No sweet life for cocoa farmers. BBC News Online [online].
[Accessed 2 October 2006]. Available from the World Wide Web: <http: / /
www.bbc.co.uk>.
Meroth, C.B., Hammes, W.P. and Hertel, C. (2003) Identification and population
dynamics of yeasts in sourdough fermentation processes by PCR-denaturing
gradient gel electrophoresis. Appl. Env. Microbiol. 69, 7453-7461.
Mian, M.A., Fleet, G.H. and Hocking, A.D. (1997) Effect of diluent type on
viability of yeasts enumerated from foods or pure culture. Int. J. Food Microbiol.
35,103-107.
Michnick, S., Roustan, J-L., Remize, F., Barre, P. and Dequin, S. (1997)
Modulation of glycerol and ethanol yields during alcoholic fermentation in
Saccharomyces cerevisiae strains overexpressed or disrupted for GPD1 Encoding
glycerol 3-phosphate dehydrogenase. Yeast 13, 783-793.
Miller, K.B., Stuart, D.A., Smith, N.L., Lee, C.Y., McHale, N.L.m Flanagan, J.A.,
Ou, B. and Hurst, W.J. (2006) 'Antioxidant activity and polyphenol and
procyanidin contents of selected commercially available cocoa-containing and
chocolate products in the United States.' J. Agric. Food Client. 54, 4062-4068.
380
Minifie, W. (1980) Chocolate, Cocoa and Confectionary: Science and Technology, 2nd
ed'n. Connecticut USA: AVI Publishing Co. Inc.
Misnawi, Jinap, S., Nazamid, S. and Jamilah, B. (2002) Activation of remaining
key enzymes in dried under-fermented cocoa beans and its effect on aroma
precursors formation. Food Chem. 78, 4-7-417.
Misnawi, Jinap, S., Jamilah, B. and Nazamid, S. (2003) Effects of incubation and
polyphenol oxidase enrichment on colour, fermentation index, procyanidins
and astringency of unfermented and partly fermented cocoa beans. Int. J. Food
Sci. Tech. 38, 285-295.
Misnawi, Jinap, S., Jamilah, B. and Nazamid, S. (2004). Effect of polyphenol
concentration on pyrazine formation during cocoa liquor roasting. Food Chem.
85, 73-80.
Misnawi, Jinap, S., Jamilah, B. and Nazamid, S. (2005) Changes in polyphenol

ability to produce astringency during roasting of cocoa liquor. J. Sci. Food Agric.
85, 917-924.
Mohr, E., Wichmann, G. and Roche, G. (1987) Cocoa butter - Influence of

fermentation on the properties of cocoa butter. Gordian 87, 77-82.
Montagnon, C. (2006) What are the priorities for cocoa research? [online].
[Accessed 20 December, 2008]. Available from the World Wide Web: chttp:/ /
www.cirad.fr/en/ actualite/communique.php?id=491>.
Moore, D.S., and McCabe, G. (2003). Introduction to the Practices of Statistics 4th
Edition. Pp. 491-570. New York, USA: WH Freeman and Company.
Moore, J.E., McCalmont, M., Xu, J., Millar, C. and Heaney, N. (2002) Asaia sp., an
unusual spoilage organism of fruit-flavoured bottled water. Appl. Env. Microbiol.
68, 4130-4131.
Moores, R.G. and Campbell, H.A. (1948) Determination of theobromine and
caffeine in cacao materials. Analytical Chemistry 20, 40-47.
Moores, R.G., Greninger, D.M. and Rusoff, I.I. (1952) Cacao oxidase. J. Am.
Chem. Soc. 74, 928-930.
Morlat, R. and Symoneaux, R. (2006) Lon-term additions of organic
amendments in a Loire Valley vineyard on a calcareous sandy soil III: Effects on
fruit composition and chemical and sensory characteristics of cabernet franc
wine. Am J. Enol. Vitic. 59, 375-386.
Motomayor, J.C., Risterucci, A.M., Lopez, P.A., Ortiz, C.F., Moreno, A. and
Lanaud, C. (2003) Cacao domestication II: progenitor germplasm of the
Trinitario cacao cultivar. Heredity 91, 322-330.
Motomayor, J.C., Risterucci, A.M., Heath, M. and Lanaud, C. (2002) Cacao
domestication I: the origin of the cacao cultivated by the Mayas. Heredity 89,
380-386.
381
Mounjouenpou, P., Gueule, D., Fontana-Tachon, A., Guyot, B., Tondje, P.R. and
Guiraud, J-P. (2008) Filamentous fungi producing ochratoxin A during cocoa
processing in Cameroon. Int. J. Food Microbiol. 121, 234-241.
Mulato, S., Atmawinata, O., Yusianto, Handaka, Pass, T., Muehlnauer, W. and
Esper, A. (1999) Development of a solar cocoa processing centre for cooperative
use in Indonesia. Planter 75, 875, 57-74.
Muyzer, G. and Smalla, K. (1998) Application of denaturing gradient gel
electrophoresis (DGGE) and temperature gradient gel electrophoresis (TGGE) in
microbial ecology. Antonie van Leeuwenhoek 73,127-141.
Myers, M.E., Nwosu, C.V., Whitacre, E.J. and Hammerstone, J.F. Jr. (2001) Food
products having enhanced cocoa polyphenol content and processes for
producing the same. US Patent Application 20010007693.
Natsume, M., Osakabe, N., Yamagushi, M., Takizana, T., Nakamura, T.,
Miyatake, H., Hatano, T and Yoshida, T. (2000) Analyses of polyphenols in
cacao liquor, cocoa and chocolate by normal phase and reversed-phase HPLC.
Biosci. Biotechnol. Biochem. 64, 2581-2587.
Nazaruddin, R., Seng, L.K., Hassan, O and Said, M. (2006) Effect of pulp
preconditioning on the content of polyphenols in cocoa beans (Theobroma cacao)
during fermentation. Indust. Crops Products 24, 87-94.
Nestle (2006). History [online]. [Accessed 19 January 2008]. Available from the
World Wide Web: chttp:// www.nestle.com/All About/History/History
+introduction.htm>.
Neukam, K., Stahl, W., Tronnier, H., Sies, H. and Heinrich, U. (2007)
Consumption of flavanol-rich cocoa acutely increases microcirculation in
human skin. Fur. J. Nutr. 46, 53-56.
Nicholls, L. (1913) Chapter VII. In: The Fermentation of Cacao. Smith, H.H. (ed.).
Pp. 221-251. London, UK: John Bale, Sons and Danielson.
Nielsen, D.S., Honholt, S., Tano-Debrah, K. and Jespersen, L. (2005) Yeast
populations associated with Ghanaian cocoa fermentations analysed using
denaturing gradient gel electrophoresis (DGGE). Yeast 22, 271-284.
Nielsen, D.S., Schillinger, U., Franz, C.M.A-R, Bresciani, J., Amoa-Awua, W.,
Holzapfel, WH. and Jakobsen, M. (2007a) Lactobacillus ghanesis sp. nov., a motile
lactic acid bacterium isolated from Ghanaian cocoa fermentations. Int. J. Syst.
Evol. Microbiol.
57,1468-1472.
Nielsen, D.S., Teniola, O.D., Ban-Koffi, L. Owusu, M., Andersson, T.S. and
Holzapfel, W.H. (2007b) The microbiology of Ghanaian cocoa fermentations
analysed by using culture-dependent and culture independent methods. Int.J.
Food Microbiol. 114,168-186.
Niles, E.V. (1981) Microflora of imported cocoa beans. J. Stor. Prod. Res
17,147-150.
382
Niyam, P. (2000) Cocoa and coffee fermentations In: Encyclopedia of Food

Microbiology. Robinson, R.K., Batt, C.C. and Patel, PD. (Eds.). New York:
Academic Press.
Nganhou, J., Njomo, D., Benet, J.C., Augier, F. and Berthomieu, G. (2003)
Perfecting a method of micro-analysis of water and acetic acid in a cocoa bean
in the course of drying: applying to determine coefficients. Heat Mass Transf.
39, 797-803.
Nout, M.J.R. and Kiers, J.L. (2005) Tempe fermentation, innovation and
functionality: update into the third millenium. J. Appl. Microbiol. 98, 789-805.
Ogundero, V.W. (1983) Thermophilic fungi and fermenting cocoa beans in

Nigeria. Mycopathologia 82: 159-165.
Okuma, Y., A. Endo, H. Iwasaki, Y. Ito, and S. Goto. (1986) Isolation and
properties of ethanol-using yeasts with acid and ethanol tolerance. J. Ferment.
Technol. 64, 379-382.
Olujide, M.G. and Adeogun, S.O. (2006) Assessment of cocoa growers' farm
management practices in Ondo State, Nigeria. Span. J. Agric. Res. (INIA) 4,
173-179.
O'Mahony, M. (1986). Sensory Evaluation of Foods: Statistical Methods and

Procedures. New York: Dekker.
Ostovar, K. (1971) Isolation and characterisation of microorganisms involved in

the fermentation of Trinidad's cacao beans. [PhD Thesis] Pennsylvania State
University, Philadelphia University Microfilms International, Ann Arbor,
Michigan USA.
Ostovar, K. and Keeney, P.G. (1973) Isolation and characterisation of

microorganisms involved in the fermentation of Trinidad's cacao beans. Journal
of Food Science 38, 611-617.
Ouattara, H.G. Koffi, B.L., Karou, G.T., Sangare, A., Niamke, S.L. and Diopoh,
J.K. (2008) Implication of Bacillus spp. in the production of pectinolytic enzymes
during cocoa fermentation. World J. Microbiol. Biotech. 24,1753-1760.
Ouoba, L.I.I., Diawara, B., Annan, N.T., Poll, L. and Jakobsen, M. (2005) Volatile
compounds of Soumbala, a fermented African locust bean (Parkia biglobosa) food
condiment. J. Appl. Microbiol. 99,1413-1421.
Packiyasothy, E.V., Janz, E.R., Senenayake, U.M., Wijesundara, R.C. and

Wickremasinghe, P. (1981). Effect of maturity on some chemical components of
cocoa. J. Sci. Food Agric. 32, 873-876.
Pacyniak, B. (2005) Taking turns at tacking. Candy Indust [online]. [Accessed 1

August 2005]. Available from the World Wide Web: chttp: / /
www.candyindustry.com>.
Padley, F.B. and Timms, R.E. (1980) The determination of cocoa butter
equivalents in chocolate. J. Amer. Oil Chem. Soc. 57, 286-293.
383
Pallmann, C.L., Brown, J.A., Olienka, T.L., Cocolin, Lv Mills, D.A. and Bissdon,
L.F. (2001) Use of WL medium to profile native flora fermentations. Am. J. Enol.
Vitic. 52,198-203.
Paramithiotis, S.,Gioulatos, S., Tsakalidou, E. and Kalantzopoulos, G. (2006)

Interactions between Saccharomyces cerevisiae and lactic acid bacteria in
sourdough. Proc. Biochem. 41, 2429-2433.
Partos, L. (2008) Health boom sets sales soaring for dark chocolate. Confectionery
News.Com [online] [Accessed 9 July 2008] Available from the World Wide Web:
chttp:/ / www.confectionervnews.com/news>.
Passmore, S. M., and J. G. Carr. (1975) The ecology of acetic acid bacteria with
particular reference to cider manufacture. J. Appl. Bacteriol. 38,151-158.
Passos, F.M.L., Silva, D.O., Lopez, A., Ferreira, C.L.L.F., and Guimares, W.V.
(1984a). Characterization and distribution of lactic acid bacteria from traditional
cocoa bean fermentation in Bahia. J. Food Sci. 49, 205-208.
Passos, F.M.L., Lopez, A.S. and Silva, D.O. (1984b). Aeration and its influence on
the microbial sequence in cacao fermentations in Bahia, with emphasis on lactic
acid bacteria. Journal of Food Science 49: 1470-1474.
Patzold, R. and Bruckner, H. (2006) Gas chromatographic determination and

mechanism of formation of D-amino acids occurring in fermented and roasted
cocoa beans, cocoa powder, chocolate and cocoa shell. Amino Acids 31: 63-72.
Patynowski, R.J., Jiranek, V., Markides, A. (2002) Yeast viability during

fermentation and sur lie ageing of a defined medium and subsequent growth of
Oenococcus oeni. Aust. J. Grape Wine Res. 8, 62 - 69.
Paulin, D., Ducamp, M., and Lachenaud, P. (2008) New sources of resistance to
Pyhtophtora megakarya identified in wild cocoa trees populations of French
Guiana. Crop Prot. 27,1143-1147.
Perego, P., Fabiano, B., Cavicchioli, M. and Del Borghi, M. (2004) Cocoa quality
and processing: a study by solid-phase microextraction and gas
chromotography analysis of methylpyrazines. Instit. Chem. Eng. 84, 291-297.
Pettipher, G.L. (1986) Analysis of cocoa pulp and the formulation of a

standardised artificial cocoa pulp medium. J. Sci. Food Agric. 37, 297-309.
Pichova, I., Pavlfckova, L., Dostaa, J., Dolejsf, E., Hruskova-Heidingsfeldova, O.,
Weber, J., Ruml, T. and Soucek, M. (2001) Secreted aspartic proteases of Candida
albicans, Candida tropicalis, Candida parasilosis and Candida lusitaniae - inhibition
with peptidomatic inhibitors. Eur. J. Biochem. 268, 2669-2677.
Piggott, J.R. (1984). Sensory Analysis of Foods. New York: Elsevier Applied
Science.
Pina, C., Santos, C., Couto, J.A., and Hogg, T. (2004). Ethanol tolerance of five
non-Saccharomyces wine yeasts in comparison with a strain of Saccharomyces
cerevisiae - influence of different culture conditions. Food Microbiol. 21, 439-447.
384
Pino, J., De Villavincencio, N. and Roncal, E. (1992) Pattern recognition of GC

profiles for classification of cocoa powder of Ghanaian and Cuban varieties.
Nahrung 36, 302-306.
Pino, J. and Roncal, E. (1992) Linalool content in roasted cocoa butter as a

characteristic of several flavour grade cocoas. Nahrung 36, 299-301.
Pitt, J.I. and Hocking, A.D. (1985) Fungi and Food spoilage. Sydney, Australia:
Academic Press.
Portillo, E., Graziani de Farinas, L. and Betancourt, E. (2005) Effect of the post
harvest treatments on the temperature and the index of fermentation in the
quality of the cocoa criollo porcelana (Theobroma cacao L.) in the south of
Maracaibo's Lake. Rev. Fac. Agron. (LUZ). 22, 388-398.
Powell, B.D. (1983) Changes in cocoa bean availability. Manufacturing

Confectioner 9, 63-69.
Prakitchaiwattana, C.J., Fleet, G.H. and Heard, G.M. (2004) Application and
evaluation of denaturing gradient gel electrophoresis to anlayse the yeast
ecology of wine grapes. FEMS Yeast Res. 4: 865-877.
Presecki, A.V., Zelfc, B., Vasfc-Racki, D. (2007) Comparison of the 1-malic acid
production by isolated fumarase and fumarase in permeabilized baker's yeast
cells Enz. Microbial Technol. 41: 605-612. (production of malic acid by
S.cerevisiae)
Pretorius, I.S. (2000) Tailoring wine yeast for the new millenium: novel
approaches to the ancient art of winemaking. Yeast 16, 675-729.
Preyer, A. (1913) Chapter I, In The fermentation of cacao. Pp. 1-31. Smith, H.H.
(ed.), John Bale, Sons and Danielson, London, UK.
Pridmore, R.D., Crouzillat, D., Walker, C., Foley, S., Zink, R., Zwahlen, M.-C.,
Brussow, H., Petiard, V. and Mollet, B. (2000) Genomics, molecular genetics and
the food industry. J. Biotechnol. 78, 251-258.
Pritchard, G. and Collbear, T. (1993) The physiology and biochemistry of the

proteolytic system in lactic acid bacteria. FEMS Microbiol. Rev. 12,179-206.
Pszczola, D.E. (2006) Thinking outside the box (of chocolates). Food Tech.
60, 50-61.
Puri, K. (2007) NACDA Cocoa pod splitter [online]. [Accessed 2 March, 2008].
Available: chttp:/ /www.dpi.qld.gov.au/cps/rde/xchg/dpi/hs.xsl/
30 5997 ENA HTML.htm>
Purr, A., Helfenberger, A and Nadai, J. (1963) USe of vacuum infiltration of

cocoa beans with special reference to enzymatic changes and the formation of
flavours and flavour precursors. Rev. Int. Choc. 18,194-207.
385
Quesnel, V.C. (1958) An index of completion of the fermentation stage in cacao

curing. Septima Conferencia Interamericana de Cacao. Palmira, Columbia.
Pp. 512-516.
Quesnel, V.C. (1960) The anaerobic phase in cocoa fermentation. Chemistry and
Industry Pp.101-102.
Quesnel, V.C. (1963) Biochemistry. Annual Report on Cocoa Research, University of
The West Indies, Trinidad. Pp. 57-61.
Quesnel, V.C. (1965) Agents inducing the death of cacao seeds during
Quesnel, V.C. (1966) Fermentation and drying of cocoa. J. Agric. Soc. Trinidad
Tobago 67, 527-545.
Quesnel, V.C. (1968) Oxygen consumption and heat production during the
fermentation of cacao. Turrialba 18,110-114.
Quesnel, V.C. (1970) Slimy fermentation. Annual Report on Cocoa Research,
University of The West Indies, Trinidad. Pp. 47-48.
Quesnel, V.C. (1972a) Cacao curing in retrospect and prospect. IV International
Cocoa Research Conference, Trinidad. Pp. 602-606.
Quesnel, V.C. (1972b) Slimy fermentation. Cocoa Grower's Bulletin 18, 19-24.
Quesnel, V.C. and Jugmohunsingh, K. (1970) Browning reaction in drying
cocoa. J. Sci. Food Agric. 21, 537-541.
Quesnel, V.C. and Lopez, A. (1975) A sweat-box for fermenting small samples of
cacao. Trop. Agric. 52, 309-316.
RA (2008) The Rainforest Alliance [online]. [Accessed 21 December 2008].
Available from the World Wide Web: chttp: / / www.rainforest-alliance.org>.
Ramiro, E., Franch, A., Castellote, C., Andres-Lacueva, C., Izquierdo-Pulido, M.
and Castell, M. (2005) Effect of Theobroma cacao flavonoids on immune
interaction of a lymphoid call line. Brit J. ofNutr. 93, 859-866.
Ramli, N., Hassan, O., Said, M., Samsudin, W. and Idris, N.A. (2006) Influence
of roasting conditions on volatile flavour of roasted Malaysian cocoa beans. /.
Food Proc. Preserv. 30, 280-298.
Raper, K.B. and Fennell, D.I. (1965) The genus Apergillus. Baltimore USA:
Williams and Wilkins.
Raper, K.B. and Thom, C. (1968) A manual of the Penicillia. London: Flafner
Publishing.
Ravelomanana, R., Guiraud, J.P. and Galzy, P. (1986) Isolation of a pectin-
utilising yeast from cocoa beans. Sys. Appl. Microbiol. 8, 230-233.
386
Ravelomanana, R., Guiraud, J.P., Vincent, J.C. and Galzy, R (1985) The yeast
flora of cocoa bean fermentation in the Ivory Coast. MIRCEN J. Appl. Microbiol.
Biotech. 1, 319-326.
Redgwell, R.J. and Hansen, C.E. (2000) Isolation and characterisation of cell wall
polysaccharides from cocoa (Theobroma cacao L.) beans. Planta 210, 823-830.
Redgwell, R.J., Trovato, V. and Curti, D. (2003) Cocoa beans carbohydrates:
roasting induced changes and polymer interactions. Food Chem. 80, 511-516.
Reineccius, G.A., Keeney, P.G. and Weisberger, W. (1972) Factors affecting the
concentration of pyrazines in cocoa beans. J. Agric. Food Chem. 20, 202-206.
Rencher, A.C. (1995) Characterising and displaying multivariate data.: In
Methods of multivariate a?ialysis. Rencher, A.C. (ed.) New York, USA: John Wiley
and Sons.
RIRDC (1999) Cocoa research across northern Australia [online]. [Accessed 14
July 2006] Available from the World Wide Web: chttp:/ / www.rirdc.gov.au/
pub /media releases/21julv99.html>.
Roelofson, PA. (1958) Fermentation, drying and storage of cacao beans. Adv.
Food Res. 8, 225-296.
Rogers, P. (2006)Top 100 global confectionery companies - Setting new
milestones. Candy Industry 171, 29-32.
Rohan, T.A. (1957) Observations on the fermentation of West African
Amelonado Cocoa. Report of the Cocoa Conference, The cocoa, Chocolate and
Confectionary Alliance, London UK. Pp.203-211.
Rohan, T.A. (1963a) The fermentation of cocoa in trays. Cocoa Growers' Bull.
1, 22-26.
Rohan, T.A. (1963b) Processing of raw cocoa for the market. Rome: FAO.
Rohan, T.A. (1965) Application of gas chromatography to quantitative
assessment of the chocolate aroma potential of unroasted cocoa beans. Food
Teclmol. 19,122-124.
Rohan, T.A. (1969) The flavour of chocolate, its precursors and a study of their
reaction. Gordian 69, 500-501, 542-544.
Rohan, T.A. and Stewart, T. (1965) The precursors of chocolate aroma: The
distribution of free amino acids in different commercial varieties of cocoa beans.
J. Food Sci. 30, 416-419.
Rohan, T.A. and Stewart, T. (1967a) The precursors of chocolate aroma:
Production of free amino acids during fermentation of cocoa beans. J. Food Sci.
32, 385-398.
387
Rohan, T.A. and Stewart, T. (1967b) The precursors of chocolate aroma:

Production of reducing sugars during fermentation of cocoa beans. J. Food Sci.
32, 399-402.
Rojas, V., Gil, J.V., Pinaga, F. and Manzanares, P. (2001) Studies on acetate ester
production by non-Saccharomyces wine yeasts. Int. J. Food Microbiol. 3, 283-289.
Romancyzk, L.J., McClelland, C.A., Post, L.S. and Aiken, W.M. (1995) Formation
of 2-acetyl-l-pyrroline by several Bacillus cereus strains isolated from cocoa
fermentation boxes. J. Agric. Food Chem. 43, 469-475.
Rombouts, J.E. (1952) Observations on the microflora of fermenting cacao beans

in Trinidad. Proceeds of the Society of Applied Bacteriology 15,103-111.
Rosenblum, M. (2005) Chocolate - A Bittersweet Saga of Dark and Light. New York
North Point.
Roura, E., Andres-Lacueva, C., Jauregi, O., Badia, E., Estrach, R., Izquierdo-
Pulido, M. and Lamuela-Raventos, R.M. (2005) Rapid liquid chromotography
tandem mass spectrophotometery assay to quantify plasma (-)-epicatechin
metabolites after ingestion of a standard portion of cocoa beverage in humans.
J. Agric. Food Chem. 53, 6190-6194.
Said, M.B. and Samarakhody, R.J. (1984) Cocoa fermentation - effect of surface
area, frequency of turning and depth of cocoa masses. International Conference on
Cocoa and Coconuts, 1984. 29, 1-15.
Samah, O. A., M. F. Ptih, and J. Selamat. (1992) Biochemical changes during

fermentation of cocoa beans inoculated with Saccharomyces cereuisiae (wild
strain). J. Food Sci. Tech. 29, 341-343.
Samah, O.A., Ibrahim, N., Alimon, H. and Abdul Karim, M.I. (1993a)
Comparative studies on fermentation products of cocoa beans. World J.
Microbiol. Biotech 9, 361-362.
Samah, O.A., Ibrahim, N., Alimon, H. and Abdul Karim, M.I. (1993b)
Fermentation studies of stored cocoa beans. World J. Microbiol. Biotech 9, 603-604.
Samson, R.A., Hoekstra, E.S. and van Oorschot, C.A.N. (1984) Introduction to
Food-borne Fungi. 2nd ed. Baarn: Central Bureau voor Schimmelcultures.
Sanagi, M.M., Hung, W.P, and Md Yasir, S. (1997). Supercritical fluid extraction
of pyrazines in roasted cocoa beans - Effect of pod storage period. J. Chromat.
785, 361-367.
Sanchez, A-H., Rejano, L., Montano, A. and de Castro, A. (2001) Utilization at

high pH of starter cultures of lactobacilli for Spanish-style green olive
fermentation. Int. J. Food Microbiol. 67,115-122.
Sanchez, J., Daguenet, G. Guiraud, J.P., Vincent, J.C. and Galzy, P. (1985) A study
of the yeast flora and the effect of pure culture seeding during the fermentation
process of cocoa beans. Lebens. Wiss. Tech. 18, 69-75.
388
Sanchez, J. Guiraud, J.P. and Galzy, P. (1984)A study of the polygalacturonase

activity of several yeast strains isolated from cocoa. Appl. Microbiol. Biotech.
20, 262-267.
Saposhnikova, K. (1952) Changes in the acidity and carbohydrate during

growth and ripening. Agron. Trop. (Paris) 2,185-195.
Schmeider, R.L. and Keeney, P.G. (1980) Characterisation and quantification of

starch in cocoa beans and chocolate products. J. Food Sci. 45, 555-563.
Schuier, M., Sies, H., Illek, B. and Fischer, H. (2005) Cocoa related flavonoids
inhibit CFTR-mediated chloride trasnport across T84 human colon epithelia.
Am. Soc. Nutr. (2005), 2320-2325.
Schutz, G.P., Prinsen, A.J. and Peter, A. (1970) The spectrophotometric

determination of caffeine and theobromine in cocoa and cocoa product. Int.
Choc. Rev. 25, 7-11.
Schwan, R.F. (1998). Cocoa fermentations conducted with a defined microbial

cocktail inoculum. Appl. Environ. Microbiol. 64,1477-1483.
Schwan, R.F., Vanetti, M.C.D., Silva, D.O., Lopez, A. and De Moraes, A. (1986)
A research note - Characterisation and distribution of aerobic, spore-forming
bacteria from cacao fermentations in Bahia. J. Food Sci. 51,1583-1584.
Schwan, R. F., Rose, A.H. and Board, R.G. (1995) Microbial fermentation of
cocoa beans, with emphasis on enzymatic degradation of the pulp. J. Appl.
Bacteriol. Symp. Suppl. 79, 96S-107S.
Schwan, R. F., Cooper, R.M. and Wheals, A.E. (1997) Endopolygalacturonase

secretion by Kluyveromyces marxianus and other cocoa pulp-degrading yeasts.
Enzyme and Microbiology Technology 21, 234-244.
Schwan, R.F. and Wheals, A.E. (2004). The microbiology of cocoa fermentation
and its role in chocolate quality. Crit. Rev. Food Sci. Nutr. 44, 205-211.
Schwarz, L.J. (1928) Cocoa in West Africa. ll.S. Bureau of Foreign and Domestic
Commerce, Trade Promotion Series No. 69. Pp. 1-44. Washington D.C.: U.S
Government Printing Office.
Schwart, P.B., Wong, D.Y.C. and Bridger, R. (1981) Some notes on maturation of
cocoa beans as a means to reduce acidity. The Planter (Kuala Lumpur) 57, 584-587.
Scully, C. (2006) Nothing but noir. Candy Indust [online]. [Accessed 27 July 2006]
Available from the World Wide Web: chttp: / / www.candyindustry.com>.
Seikie, K. (1973) Chemical changes during cocoa beans fermentation using the
tray method in Nigeria. Inti. Choc. Rev. 28, 38-42.
Senanayake, M., Jansz, E.R. and Buckle, K.A. (1995) Effect of variety and
location on optimum fermentation requirements of cocoa beans: an aid to
fermentation on a cottage scale. J. Sci. Food Agric. 69, 461-465.
389
Senanayake, M., Jansz, E.R. and Buckle, K.A. (1997) Effect of different mixing
intervals on the fermentation of cocoa beans. J. Sci. Food Agric. 74, 42-48.
Senanayake, U.M. and Wijesekera, R.O.B. (1971) Theobromine and caffeine
content of the cocoa bean during its growth. J. Sci. Food Agric. 22, 262-263.
Seo, S-H., Rhee, C-H. and Park, H-D. (2007) Degradation of malic acid by
Issatchenkia orientalis KMBL 5774, an acidophilic yeast strain isolated from
Korean grape wine pomace. /. Microbiol. (Korea) 45, 521-527.
Serafini, M. (2004) Polyphenols from tea and cocoa: antioxidants for human
health? AgroFood Industry Hi-tech. 15,13-15.
Sharpe, M.E. (1979) Identification of the lactic acid bacteria. In: Identification
Methods for Microbiologists, 2nd Ed'n. Skinner, F.A. and Lovelock, D.W. (eds.) Pp.
231-259. London UK: Academic Press.
Shepherd, R. (1976) Large scale processing of cocoa beans - temperature and
acidity trends. The Planter (Kuala Lumpur) 52, 311-322.
Shepherd, R. and Yap, T.N. (1984) Utilisation of byproducts of cocoa bean
processing. International Conference on Cocoa and Coconuts, Kuala Lumpur
Malaysia. Pp. 1-17.
Singh, S., De Vries, J., Hulley, J.C.L., and Yeung, P. (1977). Coffee, Tea, and Cocoa:
Market Prospects and Development Lending. Baltimore: The John Hopkins
University Press.
Skinner, F.A. and Lovelock, D.W. (1979) Identification Methods for Microbiologists.
2nd Ld. London UK: Academic Press.
Skoog, D. A., West, D.M., and Holler, F.J. (1996). Fundamentals of Analytical
Chemistry. Saunders College Publisher, USA.
Slepecky, R.A. and Hemphill, H.E. (2007) The Genus Bacillus - Nonmedical. In:
The Prokaryotes (3rd Edn), Dworkin, M., Falkow, S., Rosenberg, E., Schleifer, K.-
H. and Stackebrandt, E. (Eds.) New York: Springer.
Smit, H.J., Gaffa, E.A. and Rogers, P.J. (2004) Methylxanthines are the psycho-
pharmocologically active constituents of chocolate. Psychopharmacology 176,
412-419.
Smitbert, R. and Krieg, N. (1994) Phenotypic characterisation. In: Methods for
General and Molecular Bacteriology. Gerhardt, P. Murray, R.G.E., Wood, W.A. and
Krieg, N.R. (Eds.) Pp614. American Society of Microbiology, Washington DC.
Smith, H.H. (1913) Fermentation of Cacao. London: John Bale, Sons and
Danielson.
Schnell, R.J, Olano, C.T., Brown, J.S., Meerow, A.W., Cervantes-Martinez, C.,
Nagai, C. and Motomayor, J.C. (2005) Retrospective determination of the
parental population of superior cacao (Theobroma cacao L.) seedlings and
390
association of microsatellite alleles with productivity. J. Am. Soc. Hort Sci. 130, 2,
181-190.
Speck, M.L. (1984) Compendium of methods for the microbiological examination of
foods. 2nd Ed. Washington: American Public Health Association.
Spencer, J.F.T., and Spencer, D.M. (1997). Yeasts in Natural and Artificial Habitats.
New York: Springer.
Sponholz, W. (1993) Wine spoilage by microorganisms. In: Wine Microbiology and
Bioteclmology. Fleet, G.H. (Ed.) Pp395-420. Switzerland: Harwood Academic
Publishers.
Standards Australia (1991a) Australian Standard 1766. Methods for the
microbiological examination offood. AS 1766.2.1 Standard plate count. Standards
Australia, North Sydney, Australia.
Standards Australia (1991b) Australian Standard 1766. Methods for the
microbiological examinatioji offood. AS 1766.2.5 Examination for specific organisms -
Salmonellae. Standards Australia, North Sydney, Australia.
Standards Australia (1994) Australian Standard 1766. Methods for the
microbiological examination offood. AS 1766.2.2 Colony count of yeasts and mould.
Standards Australia, North Sydney, Australia.
Stark, T., Bareuther, S. and Hofmann, T. (2005) Sensory guided decomposition of
roasted cocoa nibs (Theobroma cacao) and structure determination of taste active
polyphenols. J. Agric. Food Chem. 53, 5407-5418.
Stark, T., Bareuther, S. and Hofmann, T. (2006) Molecular definition of the taste
of roasted cocoa nibs (Theobroma cacao) by means of quantitative studies and
sensory experiments. J. Agric. Food Chem. 53, 5407-5418.
Stark, T. and Hofmann, T. (2005a) Isolation, structure determination, synthesis,
and sensory activity of N-phenylpropenoyl-L-amino acids from cocoa
(Theobroma cacao). J. Agric. Food Chem. 54, 5530-5539.
Stark, T. and Hofmann, T. (2005b) Structure, sensory activity and dose/response
functions of 2,5-diketopiperazines in roasted cocoa nibs (Theobroma cacao). J.
Agric. Food Chem. 53, 7222-7231.
Steinkraus, K.H. (2004). Industrialization of Indigenous Fermented Foods,
Second Edition. Boca Raton: CRC Press.
Stevens, J. (1986) Applied multivariate statistics for the social sciences. Hillsdale, NJ:
Lawrence Erlbaum Associates.
Stephenson, T. (2008) Aromas and Flavours [online]. [Accessed 1 February
2008]. Available from the World Wide Web: <http: / / www.wine-pages.com /
guests / tom / taste.htm>.
Stiff, P. (2006). Developing markets primed for growth. Candy Industry 171,
25-28.
391
Stone, H., and Sidel, J.L. (2004). Sensory Evaluation Practices. New York:
Elsevier.
Stone El., Sidel J.L., Oliver S., Woolsey A. and Singleton R.C. (1974) Sensory
evaluation by quantitative descriptive analysis. Food Technol. 28, 24-34.
Subden, R.E., Charlebois, R.L. and Carey, C.K. (1987) Selection in yeast I:
Assessing genetic stability and relative fitness of commercial yeast. J. Ind.
Microbiol Biotech. 2,159-165.
Sukha, D.A., Butler, D.R., Umaharan, P. and Boult, E. (2008) The use of an
optimised organoleptic assessment protocol to describe and quantify different
flavour attributes of cocoa liquors made from Ghana and Trinitario beans. Eur.
Food Res. Technol. 226, 405-413.
Summa, C., Raposo, F.C., McCourt, J., LoScalzo, R., Wagenr, K-H., Elmadfa, I. ad
Anklam, E. (2006) Effect of roasting on the radical scavenging activity of cocoa
beans. Eur. Food Res. Technol. 222, 368-375.
Swain, T. (1957) Cacao-fermentation. Chemistry and Industry May, 543-545.
Swiegers, J.H., Bartowsky, E.J., Henschke, P.A. and Pretorius, I.S. (2005) Yeast
and bacterial modulation of wine aroma and flavour. Aust. J. Grape Wine Res.
11,139-173.
Thompson, S.S., Miller, K.B. and Lopez, A.S. (2007). Cocoa and coffee. In: Food
Microbiology Fundamentals and Frontiers 3rd Ed'n. Doyle, M.P and Beuchat, L.R.
Pp. 837-850. Washington, DC: ASM Press.
Timbie, D.J. and Keeney, P.G. (1980) Comparison of several types of cocoa beans
relative to fractionated protein components. J. Agric. Food Chem. 28, 472-474.
Timbie, D.J., Sechrist, L. and Keeney, P.G. (1978) Application of high pressure
liquid chromatography to the study of variables affecting theobromine and
caffeine concentrations in cocoa beans. Journal of Food Science 43, 560-565.
Tomlins, K.I., Baker, D.M., Daplyn, P. and Adomako, D. (1993) Effect of

fermentation and drying practices on the chemical and physical profiles of
Ghana cocoa. Food Chem. 46, 257-263.
Tsunetsugu, Y., Ashitani, H., Shimada, M., Kamiwaki, T., Morikawa, T., Kojima,
T. and Miyazaki, Y. (2005) Relationship as seen from perspectives of both
gender and individual personality to the subjective comfortable feeling for the
taste and smell of chocolate. Nippon Shokuhin Kagaku Kaishi 52, 347-354.
United Nations Conference on Trade and Development (2006). Cocoa Prices

[online]. [Accessed 18 August 2007]. Available from the World Wide Web:
chttp:/ / www.unctad.org/infocomm/anglais/cocoa/prices.htm>.
Urbach, G. (1995) Contribution of lactic acid bacteria to flavour compound

formation in dairy products. Int. Dairy J. 5, 877-903.
392
Urbanski, J. (1992) Chocolate flavour: origins and descriptions; the effect of

process and bean source. Manuf Confect. 72, 69-82.
Urquhart, D.H. (1962) Cocoa. London UK: Longmans.
Valero, E., Schuller, D., Cambon, B., Casal, M and Dequin, S. (2005)
Dissemination and survival of commercial wine yeast in the vineyard: A large-
scale, three-years study. FEMS Yeast Res. 5, 959-969.
Vandenbergh, P.A. (1993) Lactic acid bacteria, their metabolic products and
interference with microbial growth. FEMS Microbiol. Rev. 12, 221-238.
Vanderhaegen, B., Neven, H., Coghe, S., Verstrepen, K.J., Derdelinckz, G. and
Verachtert, H. (2003) Bioflavouring and beer refermentation. Appl. Microbiol.
Biotechnol. 62,140-150.
Viljoen, B. (2006) Yeast ecological interactions. Yeast-yeast, yeast-bacteria, yeast-

fungi interactions and yeast as biocontrol agents. In Yeasts in Food and Beverages
Querol, A. and Fleet, G.H. (Eds.) Pp83-100. Berlin: Springer-Verlag.
Vlachopoulos, C., Alexopoulos, N. and Stefanidis, C. (2006) Effect of dark

chocolate o arterial function in healthy individuals: cocoa instead of ambrosia?
Curr. Hypertension Rep. 8, 205-211.
Voigt, J., Heinrichs, H., Voigt, G., and Biehl, B. (1994a). Cocoa specific aroma
precursors are generated by proteolytic digestion of the vicilin-like globulin of
cocoa seeds. Food Chem. 50,177-184.
Voigt, J., Biehl, B., Heinrichs, H., Kamaruddin, S., Gaim Marsoner, G., and Hugi,
A. (1994b). In-vitro formation of cocoa-specific aroma precursors: aroma-related
peptides generated from cocoa seed protein by co-operation of an aspartic
endoprotease and a carboxypeptidase. Food Chem. 49, 173-180.
Voigt, J. and Biehl, B. (1995). Precursors of the cocoa-specific aroma components

are derived from the vicilin-class (7S) globulin of the cocoa seeds by proteolytic
processing. Botanica Acta 108, 283-289.
Willi, G.C., Berger, A. Di Marzo, V. and Bisogno, T. (2001). Lipids in neural

function: modulation of behavior by oral administration of endocannabinoids
found in foods. Nestle Nutr. Workshop Ser. Clin. Perform. Programme 5:169-187.
Walker, G.M. and Van Dijck, P. (2006). Physiological and molecular responses of
yeasts to the environment. In: The Yeast Flandbook-Volume 2: Yeasts in Foods and
Beverages. Querol, A. and Fleet, G.H. (Eds.). Pp. 111-152. Berlin: Springer-Verlag,
Wampler, T.P. (1997). Analysis of food volatiles using headspace-gas

chromatographic techniques. In: Techniques for Analyzing Food Aroma.
Marsili, R. (Ed.) Pp. 59-80. New York, USA: Marcel Dekker.
Wan Ishak (2008) Curriculum Vitae [online]. [Accessed 20 December 2008]

Available from the World Wide Web: chttp: / / webed.eng.upm.edu.my / files/
kbp/ wanishakcv.pdf>.
393
Watson, B.J. (1987) Review: Cocoa potential and possible research direction in
north Queensland. Internal report, Queensland Dept, of Primary Industries.
Watson, B.J. (1992) Introduction, evaluation and management of cocoa cultivars

for far north Queensland. Final Report for Rural Industries Research and
Development Corporation, Project DAQ39, Queensland Dept, of Primary Industries.
Weissberger, W, Kavanagh, T.E. and Keeney, RG. 1971. Identification and

quantification of several non-volatile organic acids of cocoa beans. J. Food Sci.
36, 877-879.
Welsh, F.W., Murray, W.D. and Williams, R.E. (1989) Microbiological production
of flavor and fragrance chemicals. Crit Rev. Microbiol. 9,105-169.
Werber, D., Dreesman, J., Feil, F., van Treeck, U., Fell, G., Ethelberg, S., Hauri,
A.M., Roggentin, P., Prager, R., Fisher, I.S.T., Behnke, S.C., Bartelt, E., Weise, E.,
Ellis, A., Siitonen, A., Andersson, Y., Tschape, H., Kramer, M.H, and Ammon, A.
(2005). International outbreak of Salmonella Oranienburg due to German
chocolate. BMC Infect. Dis. 5, 7.
Wigley, R.C., Dass, C.R., Cogan, T.M., Uraz, T. and Ozer, B.H. (2005) Starter
Cultures. In: Encyclopedia of Food Microbiology. Robinson, R.K., Batt, C.C. and
Patel, P.D. (Eds.) Pp. 2084-2114. New York: Academic Press.
Wisselink, H. W., Weusthuis, R.A., Eggink, G., Hugenholtz, J. and Grobben, G.J.
(2002) Mannitol production by lactic acid bacteria: a review, hit. Dairy J. 12, 151—
161.
Wollgast, J. and Anklam, E. (2000). Review of polyphenols in Theobroma cacao:

changes in composition during the manufacture of chocolate and methodology
for identification and quantification. Food Res. Int. 33, 423-447.
Wood B.J.B (Ed) (1998) Microbiology of Fermented Foods (2nd edition). London UK:
Thompson Science.
Wood, G.A.R. (1971) Ikiliwindi: Cadbury-Schweppes plantation in Cameroon. 3.

Fermentation and drying. Cocoa Growers' Bull. 15, 25-29.
Wood, G.A.R. and Lass, R.A. (1985) Cocoa. 4th ed. London UK: Blackwell
Science.
Wright, D.C., Park, W.D., Leopold, N.R., Elasegawa, P.M. and Janick, J. (1982)
Accumulation of lipids, proteins, alkaloids and anthocyanins during embryo
development in vivo of Thebroma cacao L. J. Amer. Oil Chem. Soc. 59,475-479.
Yang, X., and Peppard, T. (1994). Solid-phase microextraction for flavour

analysis. J. Agric. Food Chem. 42, 1925-1930.
Yoo, S.S., Kim, K., Lee, S.Y., Hong, S.K., Lee, M.C., Chang, Y.Y., Kwon, I.B. and
Pyun, Y.R. (1998) Characterization of flavour components in cocoa mass
produced by better taste and color treatment using GC/MS and principal
component analysis. Food Sci. Biotechnol. 7, 248-252.
394
Yusep, I. Jinap, S., Jamilah, B. and Nazamid, S. (2002) Influence of

carboxyopeptidases on free amino acid, peptide and methyl pyrazine contents
of under-fermented cocoa beans. J. Sci. Food Agric. 82,1584-1592.
Zak, D.L., Ostovar, K., and Keeney, P.G. (1972). Implication of Bacillus subtilis in
the synthesis of tetramethylpyrazine during fermentation of cocoa beans. J. Food
Sci. 37, 967-968.
Zak, D.L (1973) Studies on the proteins of cocoa beans. [PhD. Thesis]
Pennsylvania State University, Philadelphia, Pennsylvania USA.
Zak, D.L. and Keeney, P.G. (1976a) Extraction and fractionation of cocoa
proteins as applied to several varieties of cocoa beans. J. Agric. Food Chem. 24,
479-482.
Zak, D.L. and Keeney, P.G. (1976b) Changes in cocoa proteins during ripening
of fruit, fermentation and further processing of cocoa beans. J. Agric. Food Chem.
24,483-486.
Zelle, R.M., Hulster, E., van Winden, W.A., de Waard, P, Dijkema, C., Winkler,
A.A., Geertman, J.M.A., van Dijkens, J.P., Pronk, J.T. and van Maris, A.J.A.
(2008) Malic acid production by Sacclmromyces cerevisiae: engineering of
pyruvate carboxylation, oxaloacetate reduction, and malate export. Appl. Env.
Microbiol. 74, 2766-2777.
Zhu, Q.Y., Holt, R.R., Lazarus, S.A., Ensunsa, J.L., Hammerstone, J.F., Schmitz,
H.H. and Keen, C.L. (2002) Stability of the flavan-3-ols epicatechin and catechin
and related dimeric procyanidins derived from cocoa. J. Agric. Food Chem. 50,
1700-1705.
Ziegleder, G. (1990) Linalool contents as characteristic of some flavour grade
cocoas. Z. Lebensm. Unters. Forsch. 191, 306-309.
Ziegleder, G. (1991) Composition of flavour extracts of raw and roasted cocoas.
Z. Lebensm Unters. Forsch. 192, 521-525.
Appendix A 395
Appendix A
Supplementary data for the confirmation of
identity of yeast and bacterial isolates.
Appendix A.l - API test kit results
The results from the API test kits are presented below as carbon assimilation
profiles. Yeast isolates were tested using the API ID32 C kit. Bacterial isolates
were tested using the API 50 CH kit. The resulting carbohydrate profiles were
compared to the API system database (Biomerieux, 2005) and the literature
(Kurtzmann and Fell, 1998; Dworkin et al., 2007).
Table 1 - Assimilation profiles of selected yeast isolates obtained during cocoa
fermentation
Carbon compounds3 Assimilation (ratio of positive isolates)6
H.guillermondii /. orientalis S.cerevisiae P. membranifaciens K.marxianus
Galactose - 2/15 7/15 - 5/5

Actidione - - - - 5/5
Saccharose - - 15/15 - 5/5
N-acety 1-glucosamine - 15/15 1/15 8/8 -
DL-lactate - 15/15 1/15 - 5/5
L-arabinose - - - - 2/5
Cellobiose 15/15 - - - -
Raffinose - - 15/15 - 5/5
Maltose - - 15/15 - -
Trehalose - - 1/15 - -
2-keto-gluconate 5/15 - - - -
a-methyl-D-glucoside 4/15 - 1/15 - -
Mannitol - - - - 5/5
Lactose - - - - 5/5
Inostiol - - - - -
Sorbitol - - - - 4/5
D-xylose 1/15 - - - 4/5
Ribose - - - - -
Glycerol - 15/15 - 7/8 4/5
Rhamnose - - - - -
Palatinose - - 1/15 - -
Erythritol - - - - -
Melibiose - - - - -
Glucoronate - - - - -
Melezitose - - - - -
Gluconate 15/15 - - - -
Levulinate - 8/15 - - -
Glucose 15/15 15/15 15/15 8/8 5/5
Sorbose - 8/15 - 4/8 -
Glucosamine - 4/15 - 8/8 -
Esculin 15/15 11/15 15/15 8/8 5/5
a Assimilation of carbon compounds was determined by use of the API ID 32 C kit.

b Number of positive isolates/number of isolates tested.
Appendix A 396
Table 2 - Assimilation profiles of selected bacterial isolates

Carbon compounds3 Assimilation (ratio of positive isolates)15
L.plantarum L.fermentum Pacidilactici A.pasteurianus G. oxydans
Glycerol - - - 3/10 -
Erythritol - - - - -
D-arabinose - - - - -
L-arabinose 20/20 10/10 - - -
D-ribose 20/20 10/10 15/15 - -
D-xylose 4/20 8/10 15/15 - -
L-xylose - - 15/15 - -
D-adontiol - - - - -
Methyl-pD-xylopyranoside - - - - -
D-galactaose 10/20 9/10 15/15 - -
D-glucose 20/20 10/10 15/15 9/10 5/5
D-fructose 20/20 10/10 15/15 - -
D-mannose 8/20 - 15/15 - -
L-sorbose - - - - -
L-rhamnose - - 4/15 - -
Dulcitol - - - - -
Inositol - - - - -
D-mannitol 20/20 - - - -
D-sorbitol 20/20 - - - -
Methyl-aD-mannopyranoside 18/20 - - - -
Methyl-aD-glucopyranoside 2/20 - - - -
N-actyl-glucosamine 20/20 - 15/15
Amygdalin 20/20 - - - -
Arbutin 20/20 - 15/15 - -
Esculin 20/20 - 15/15 - -
Salicin 20/20 - 15/15 - -
D-cellobiose 20/20 - 15/15 - -
D-maltose 20/20 10/10 - - -
D-lactose 20/20 - - - -
D-melibiose 20/20 10/10 - - -
D-saccharose 20/20 10/10 - - -
D-trehalose 20/20 2/10 15/15 - -
Inulin - - - - -
D-melezitose 19/20 - - - -
D-raffinose 15/20 8/10 - - -
Starch - - - - -
Glycogen - - - - -
Xylitol - - - - -
Gentobiose 20/20 - 15/15 - -
D-turanose 20/20 - - - -
D-lyxose - - - - -
D-tagatose - - 15/15 - -
D-fucose - - - - -
L-fucose - - - - -
D-arabitol - - - - -
L-arabitol - - - - -
Potassium gluconate 20/20 7/10 - - -
Potassium 5-ketogluconate - - - - -
Potassium 2-ketogluconate - 7/10 - - -
a Assimilation of carbon compounds was determined by use of the API CH 50 kit.

b Number of positive isolates/number of isolates tested.
Appendix A 397
Appendix A.2 - Ribosomal DNA sequence identification of

yeast and bacterial isolates obtained during fermentation of
cocoa beans cultivated in Queensland, Australia.
Table 3 - Partial 26S rDNA sequence identification of yeast isolates from cocoa
fermentations by plate culture.
Species Accession number Sequence similarity (%)
Candida quercitrusa* AY529522 99-100
Candida sorbisivorans* AJ783433 96
Hanseniaspora guilliermondii AY529510 99-100
Hanseniaspora uvarum* U84229 99-100
Issatchenkia orientalis AY529500 100
Kluyveromyces marxianus** AY894822 99
Pichia membranifaciens AY529506 99-100
Pichia anomola* AF330115 99
Pichia burtonii* AB438159 98
Saccharomyces cerevisiae AF533067 100
Zygoascus hellenicus* U40125 98
* Indicates species isolated infrequently at the beginning of fermentation

** K.marxianus was not detected in any spontaneous fermentations, but was re-isolated
from fermentations to which it was inoculated
Appendix A 398
Table 4 - Partial 16S rDNA sequence identification of bacterial isolates obtained

from cocoa fermentations by plate culture.
Species Accession number Sequence similarity (%)
Acetobacter pasteur ianus AB117968 99
Acetobacter syzigii*** DQ523496 98
Acetobacter tropical is DQ523494 98
Asaia siamensis AY360335.1 99
Bacillus cereus DQ523499 98
Bacillus licheniformis DQ523501 94-100
Bacillus megaterium AY553118.1 99
Bacillus pumilis DQ523500 94-100
Bacillus subtilis DQ523502 99
Erwinia carotovorum * ECA233411 100
G/uconobacter oxydans AB166739.1 96-98
Lactobacillus brevis DQ523492 96-97
Lactobacillus fermentum AY929282.1 99
Lactobacillus piant arum AY735409.1 99
Lactobacillus sp. * * DQ523487 84-89
Lactococcus lact is AB008214 98
Leuconostoc pseudomesenteroides DQ523483.1 98
Leuconostoc pseudoficulneum AY 169967 95-97
Pantoe a agglomerans AY974376.1 99
Pediococcus acidilactici AJ784922.1 99
Pseuomonas putida* EU836174 96-98
Serratia marcescens* EF204297 95
Staphylococcus aureus* DQ226212 96-98
* Indicates species isolated infrequently at the beginning of fermentation

** This unidentified Lactobacillus spp. returned poor matches from GenBank. Initial
investigations suggest it may represent a new species, possibly indigenous to
Australian cocoa fermentation. Further investigation is strongly recommended.
Appendix A 399
Appendix A.3 - Denaturant gradient gel electrophoresis

(DGGE) examination of Australian cocoa fermentations.
Samples of frozen cocoa beans obtained from Australian cocoa fermentations
were prepared locally according to the methods described in Section 3.2.3.7,
Chapter 4. The prepared nucleic acid samples were shipped to the Royal
Veterinary & Agricultural University of Denmark (KVL). Workers at these
laboratories processed these samples for PCR and DGGE, according to the
methods described by Nielsen at al. (2005) and Nielsen et al. (2007b). We
acknowledge the contribution of Dr Dennis Nielsen in performing the DGGE,
and in assisting with the interpretation of the gels.
Figures 1, 2 and 3 show the DGGE profiles obtained from the heap, box and
barrel fermentations conducted using beans from the Mossman plantation.
In each case, the predominant species of yeast and bacteria identified by
cultural methods were confirmed by culture independent PCR-DGGE analyses.
The gels also show several sets of bands that did not match the banding
patterns of our isolates. These bands may represent species of microorganisms
that were not able to be detected using cultural methods.
Yeast Species Bacterial Species
h
h
h
h
4: 7 2 h
4: 72 h
5: 9 6 h
6 :1 2 0
6 :1 2 0
h
96
3 :4 8
3 :4 8
2 :2 4
1 :0
5:
M* ««* *»<*
• * * i hi M
Miju^
H guilliermondii. u«a«~ ~~***»~*&2.
S. cerevisiae .# 11 , -iP addilactici
Pt agglomerans
I. onentalis A. pasteurianus
Figure 1 - DGGE profiles (35-65% denaturant) representing 26S rRNA (yeast) and 16S rRNA
(bacteria) fragments of cocoa pulp sampled at 24 hour intervals during 120 hours of heap
fermentation. Identified fragments labelled, see arrows. Beans sourced from Mossman
plantation.
Appendix A 400

•c c. c. c.
C\
CO
rt
CM l£>
C>
s*“K O CM
CD CM
f*.
ip
05
N
«-
N n » tb to »- cm n V in ib
• J
rr:
—— L.plantarum
H. guilliermondii. *«»**
—Lc. pseudomesenteroides
S. cerevisiae MIR
P. addilactici
..... . ■ Pf. agglomerans
I. orientalis
—ali«l ..........A pasteurianus
(bacteria) fragments of cocoa pulp sampled at 24 hour interv als during 120 hours of box
plantation.

_ c C. C. C. p
^ ^ CO CM to K
O CM h- 0> «-
v- cm <n 't uni co
— Lptantarum
H. guilliermondii Lc. pseudomesenteroides
S. cerevisiae -P. addilactici
I. orientalis
is" ’A. pasteurianus
(bacteria) fragments of cocoa pulp sampled at 24 hour intervals during 120 hours of barrel
plantation.
Appendix A 401
Appendix A.5 - Enzyme Screening results
Selected strains of yeast isolates were screened for pectinase, protease and
lipase activity. These tests were performed according the the methods described
in Chapter 4 (Section 4.2.3.1).
This Data was collected with the assistance of TuAnh Ngoc Huynh, as part of
an Honours project conduected at the Univrsity of New South Wales.
Table 5 -Screening selected yeast isolates3 for pectinase, protease and lipase
activities.
Species Enzyme activity (isolates with positive activity/number of isolates tested)
Pectinase (pH 5)b Pectinase (pH 7)c Protease (SMA) Lipase(cocoa

butter agar).
Hanseniaspora 0/10 0/10 7/10 3/20
guillermondii
Issatchenkia orientalis 1/20 1/20 13/20 1/20
Saccharromyces 1/10 1/10 2/5 1/10

cerevisiae
Pichia 0/10 0/10 1/10 0/10
membranifaciens
Kluyveromyces 1/1 1/1 1/1 0/1
marxianus (AWRI
strain CY69)
a. All isolates except K. marxianus were isolated as indigenous strains from

cocoa bean fermentations conducted in Nth Queensland.
b. Pectin degradation at pH 5 nominally indicates polygalacturonase activity.
c. Pectin degradation at pH 7 nominally indicates pectate lyase activity.
Appendix B 402
Appendix B
Comparison of the chemical and physical properties of cocoa
beans harvested from the South Johnstone and Mossman
plantations, Queensland.
The data collected in Table 1 show variations in the chemical and physical properties of
cocoa beans harvested from the two plantations. Typically, beans from the Mossman
plantation had more pulp and lower pulp pH compared to beans from the South
Johnstone plantation. As both plantations became older, the average pH of the pulp
increased. Differences in chemical and physical properties were consistently observed
between batches of beans harvested during the same month, but from different
plantations.
~o
a> o in vO n n VO vlt in
'§60 00
cn O GO
<u Q.
3 cd CO o oo O' o in o o
> CN —
— CL
a
-a T3
a> >n o in o VO o
<D — ~ •“V V v~
C/3
CJ 6X)
a ___
IS «n o o in <n oo vt
3 »n CO Tt <n cn ■3- ■3" >n vt
u CL
-o oo
<u O a> — VO 3 o CN rq rq rq
C/3 > o' o'
CO
2 *
_Q_ o
o o o o o O o o
CL o © © © o cd cd o cd
•a m
O <L) VO © VO oo r~~ >n CN in
<u o © o o © o <d
O * CO
t3 £
CL r- o rq oo oo © rq CN vf
3
CL vd r-t in vd «n n vd vd vd
-a
<L> rq <n (N <n oo CN © — vo
C/3 S <D © o O © © cd cd o
CO
8 *
a. o o Ov vq vq 3 Ov rq o
a£ 3
vd vd Tf vd vd cd in *n
CL
I — cn CN <n rq CN —
a> vd vd vd vd vd vd vd
CO
DC
a. D.
3 vo tq OV o Ov oo vq rq oo
CL rd rd rd vd rd rd cd rd rd
S--S-2 rq O Ov oo >n oo o in
3 <D Cd cd o © cd cd
£L 5/3 J—
a> ■3- tt vO
ft
3 w
o o o
TO
CN CN vo n. i
(N
o
ts Tf
o c5 »n vo
o
vo
o O a>
1 o X x o o o CN X)
(N CN o (N CN
§ E E CN to E
r= r= 3
DC !_ X > > OS) >
o*v
w c o
LL
3
< < Z Z —o < < < Z
c
o § c 3 c C c 3 § §
o o o o
I E -3 to E E E E
•4—* C/3 X to X to X to C/3 C/3 C/3
% n C/3 C tG C ti C C/3 C/3 C/3
cdi o 5 -c O n n •*= O O O
CL 2 &S, 5-S S S 2

Dircks 014967987 PDF

Uploaded by

Copyright:

Available Formats

Dircks 014967987 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dircks 014967987 PDF

Uploaded by

Copyright:

Available Formats

Investigation into the fermentation of

Australian cocoa beans and its effect on

Hugh Douglas Dircks

A thesis submitted in fulfillment of the requirements

University of New South Wales

knowledge it contains no materials previously published or written by another

award of any other degree or diploma at UNSW or any other educational

institution, except where due acknowledgment is made in the thesis. Any

contribution made to the research by others, with whom I have worked at

UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that

in style, presentation and linguistic expression is acknowledged/

My deep and heartfelt thanks to my supervisor, Professor Graham Fleet. His

This study was possible by the generous funding of Cadbury-Schweppes. I am

My field work was performed at the Department of Primary Industries, South

CHAPTER 2 LITERATURE REVIEW - A CRITICAL REVIEW

2.1 COCOA AND CHOCOLATE 4

2.3 DEVELOPMENT OF COCOA INDUSTRY IN AUSTRALIA 29

2.6 THE IMPROVEMENT OF COCOA BEAN PROCESSING AND

CHAPTER 3 THE MICROBIOLOGICAL, CHEMICAL AND SENSORY 8Q

3.2 MATERIALS AND METHODS 82

3.2.3.3 Identification of bacterial species 98

3.3.2.3 Changes in the population of bacterial species during ^

3.4.4.4 Glycerol and mannitol 185

3.5 CONCLUSIONS 196

CHAPTER 4 EVALUATION OF THE USE OF MIXED YEAST STARTER

4.1 INTRODUCTION 199

4.2.6.1 Quality assessment of dried fermented cocoa beans 209

43.3.2 Changes in the population of yeast species during ^9

4.4.3.1 Changes in temperature 321

CHAPTER 5 CONCLUSIONS 346

CHAPTER 6 REFERENCES 355

Appendix A - Supplementary microbiological data 395

This thesis investigated the fermentation of cocoa beans grown in Queensland.

inoculate fermentations with mixtures of yeast, it is the first that inoculated

mixed cultures of yeast into non-sterile, commercial scale fermentations.

Fermentations were monitored for microbiological and chemical changes, and

The inoculated fermentations produced beans of acceptable quality and flavour.

This research is the first of its kind performed in Australia. In addition to

cultures and new fermenter designs.

Publications and presentations from this thesis

Dircks, H.D. 2008. 'Industrialisation of Australian cocoa fermentation.'

Dircks, H.D. (2008) Towards industrialised cocoa fermentation in Queensland,

Dircks, H.D. (2006) The microbial ecology of Australian cocoa bean

Dircks, H.D. (2004) The Mycology of Australian Cocoa fermentations. Oral

To gain baseline data on the microbiology, biochemistry and physical

Literature Review - A critical review focused on

Summary of cocoa domestication

Spread of cocoa to other countries

2.1.2 Production of cocoa and chocolate

roasted. Additional processing is then required to obtain a commercial product.

Production of cocoa beans ultimately relies on the successful cultivation of

2.1.2.2 Harvest and pod splitting

Fermentation is of vital importance in preparing cocoa beans for market, since

The fermentation is conducted by indigenous microorganisms that originate

This microbial activity during fermentation causes physical and chemical

application is not consistent (Said and Samarakhody, 1984; Senanayake et al.,

Country Predominant Predominant Size Typical Mixing References

Ghana Heap, tray 50-1200 kg Not mixed, or Tomlins et al.

once at 48 h al., 1985