Food Microbiology 28 (2011) 457e464

Contents lists available at ScienceDirect

Food Microbiology
journal homepage: www.elsevier.com/locate/fm

Dynamics and species diversity of communities of lactic acid bacteria and acetic
acid bacteria during spontaneous cocoa bean fermentation in vessels
Timothy Lefeber a, b, William Gobert a, b, Gino Vrancken a, Nicholas Camu b, Luc De Vuyst a, *
a
Research Group of Industrial Microbiology and Food Biotechnology (IMDO), Faculty of Sciences and Bio-Engineering Sciences, Vrije Universiteit Brussel (VUB),
Pleinlaan 2, B-1050 Brussels, Belgium
b
Société Africaine de Cacao, 6 rue de Pierre et Marie Curie, 01 BP 1045 Abidjan 01, Côte d’Ivoire

a r t i c l e i n f o a b s t r a c t

Article history: To speed up research on the usefulness and selection of bacterial starter cultures for cocoa bean
Received 23 July 2010 fermentation, a benchmark cocoa bean fermentation process under natural fermentation conditions was
Received in revised form developed successfully. Therefore, spontaneous fermentations of cocoa pulp-bean mass in vessels on a 20
13 October 2010
kg scale were tried out in triplicate. The community dynamics and kinetics of these fermentations were
Accepted 19 October 2010
Available online 27 October 2010
studied through a multiphasic approach. Microbiological analysis revealed a limited bacterial species
diversity and targeted community dynamics of both lactic acid bacteria (LAB) and acetic acid bacteria
(AAB) during fermentation, as was the case during cocoa bean fermentations processes carried out in the
Keywords:
Cocoa bean fermentation
field. LAB isolates belonged to two main (GTG)5-PCR clusters, namely Lactobacillus plantarum and
Vessel Lactobacillus fermentum, with Fructobacillus pseudofilculneus occurring occasionally; one main (GTG)5-
Species diversity PCR cluster, composed of Acetobacter pasteurianus, was found among the AAB isolates, besides minor
Community dynamics clusters of Acetobacter ghanensis and Acetobacter senegalensis. 16S rRNA-PCR-DGGE revealed that L.
Lactic acid bacteria plantarum and L. fermentum dominated the fermentations from day two until the end and Acetobacter
Acetic acid bacteria was the only AAB species present at the end of the fermentations. Also, species of Tatumella and Pantoea
were detected culture-independently at the beginning of the fermentations. Further, it was shown
through metabolite target analyses that similar substrate consumption and metabolite production
kinetics occurred in the vessels compared to spontaneous cocoa bean fermentation processes. Current
drawbacks of the vessel fermentations encompassed an insufficient mixing of the cocoa pulp-bean mass
and retarded yeast growth.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction
Raw, freshly harvested cocoa beans have an unpleasant and
astringent flavour. Hence, it is absolutely necessary to ferment, dry,
and roast raw cocoa beans to obtain the desired organoleptic
characteristics (Beckett, 2009; De Vuyst et al., 2009; Thompson
et al., 2007). The cocoa bean fermentation process aims at killing
the seed embryo, which prevents cocoa beans to germinate, and
facilitating removal of the mucilaginous pulp, which surrounds the
cocoa beans. In addition, certain aroma precursors (free amino
acids, reducing sugars, and peptides) within the cotyledons are
formed which contribute, after roasting, to the characteristic
chocolate flavour (Afoakwa et al., 2008; Beckett, 2009).
Nowadays, the cocoa tree, Theobroma cacao L., is cultivated
within a narrow equatorial belt that crosses South and Central
* Corresponding author. Tel.: þ32 2 6293245; fax: þ32 2 6292720.
E-mail address: ldvuyst@vub.ac.be (L. De Vuyst).
America, Africa, and Asia (Wood and Lass, 1985). In the last decade,
the microbial diversity of cocoa bean fermentation processes has
been studied in detail (Ardhana and Fleet, 2003; Camu et al., 2007,
2008b; Daniel et al., 2009; Jespersen et al., 2005; Kostinek et al.,
2008; Lagunes-Gálvez et al., 2007; Nielsen et al., 2005, 2007).
These studies have resulted in a better understanding of the
microbial succession and activities taking place during fermenta-
tion of the cocoa pulp-bean mass (Camu et al., 2007, 2008a,b;
Nielsen et al., 2007). The key microorganisms that are crucial for
successful cocoa pulp fermentation are yeasts, lactic acid bacteria
(LAB), and acetic acid bacteria (AAB) (Camu et al., 2007; Schwan
and Wheals, 2004). Under the initial anaerobic conditions of the
tight pulp-bean mass, yeasts produce ethanol from glucose
(sucrose). Also, the pectinolytic yeast community is held respon-
sible for liquefying the pulp, allowing the pulp sweatings to drain
away out of and air to penetrate into the fermenting pulp-bean
mass (Schwan et al., 1995; Schwan and Wheals, 2004). In parallel,
LAB convert citric acid and residual carbohydrates in the pulp into
mainly lactic acid, acetic acid, and/or mannitol, enabling a slight pH

0740-0020/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.fm.2010.10.010
458 T. Lefeber et al. / Food Microbiology 28 (2011) 457e464
increase (Camu et al., 2007). The LAB species diversity is rather
limited; only strains of Lactobacillus fermentum and Lactobacillus
plantarum dominate Ghanaian spontaneous cocoa bean heap
fermentations (Camu et al., 2007, 2008b). During the aerobic phase,
Acetobacter pasteurianus is the main AAB species participating in
spontaneous cocoa bean fermentation. The AAB oxidize the ethanol
produced by the yeasts and the lactic acid produced by the LAB into
acetic acid (Camu et al., 2007; Schwan and Wheals, 2004). This
volatile short-chain fatty acid is the key metabolite of a cocoa bean
fermentation process. Together with ethanol, it diffuses into the
beans and, together with the temperature increase due to microbial
oxidation of ethanol into acetic acid and overoxidation of the latter
into carbon dioxide and water, it causes the death of the seed
embryo (Camu et al., 2007; Thompson et al., 2007). In parallel,
diffused acetic acid disintegrates cellular membranes inside the
cocoa beans. This induces enzymatic transformations of substrates
within the cotyledons of the beans that lead to formation of
precursors of the characteristic flavour compounds and colour of
fully fermented cocoa beans (Thompson et al., 2007).
Today, the cocoa bean fermentation process is still a sponta-
neous, uncontrolled, on-farm process, without the addition of
starter cultures. Although preliminary experiments of the applica-
tion of defined starter cultures show satisfying results (Buamah
et al., 1997; Dzogbefia et al., 1999; Leal et al., 2008; Samah et al.,
1992, 1993; Sanchez et al., 1985; Schwan, 1998), they have not
been introduced in the field. However, starter cultures may influ-
ence the fermentation rate of raw cocoa beans and thus the taste of
chocolate made of the corresponding fermented dry beans. Testing
starter cultures requires laborious and time-consuming heap and/
or box fermentation experiments to be carried out under difficult
circumstances in the field or a fermentory in the origin countries.
The present study aimed at developing a benchmark cocoa bean
fermentation process to speed up research on the usefulness and
selection of bacterial starter cultures. Therefore, spontaneous cocoa
bean fermentations were set-up in vessels and analysed multi-
phasically, encompassing microbiological analysis (culture-depen-
dent and culture-independent), metabolite target analysis, and
pilot-scale cocoa liquor production.
2. Materials and methods
2.1. Cocoa bean vessel fermentations
Three individual spontaneous cocoa bean fermentations were
carried out in 20 l plastic vessels in a temperature-controlled room
(28  C). The fermentations were carried out simultaneously to
exclude as much environmental factors as possible (harvest season,
external contamination factors). Mature, freshly harvested, healthy
cocoa pods were obtained from the same plantation at the end of
the main-crop of 2007 (January to February 2008). Selected pods
were broken with washed machetes by workers who first washed
their hands, the placenta of the fruits was removed, and the pulp-
bean mass was immediately transferred into a clean, decontami-
nated vessel to obtain a homogeneous mixture of 60 kg of wet
beans. After mixing, 60 kg of the pulp-bean mass was transferred
and divided into three equal vessels (20 kg/vessel) and lasted to
ferment for six days without mixing. The pulp sweatings produced
during fermentation were allowed to drain away through a hole in
the bottom of the vessels. To simulate the anaerobic phase during
the first two days of fermentation, the vessels were closed with a lid
to avoid open air contact with the pulp-bean mass. After 48 h of
fermentation, the lid was removed to allow air ingress into the
pulp-bean mass. After fermentation, the cocoa beans were sun-
dried on coverable platforms for approximately seven to ten days,
depending on the weather conditions. The three vessel
fermentations were monitored online for temperature and pH by
inserting a digital pH 340i sensor (WTW GmbH, Weilheim,
Germany) in the middle of the fermenting cocoa bean mass. As all
three vessel fermentations performed similarly, the courses of one
fermentation are shown below.
2.2. Sampling
Samples (500 g) of the vessel fermentations were taken
according to a fixed time schedule, namely at the start of the
fermentation (time 0) and after 24, 48, 72, 96, 120, and 144 h of
fermentation. Sampling was always done in the middle of the pulp-
bean mass. Each sample was aseptically removed and transferred
into a sterile plastic bag. After 72, 96, 120, and 144 h of fermentation,
2 kg of wet beans were withdrawn from the fermenting pulp-bean
mass, sun-dried, and roasted to check the influence of fermentation
time on fermented dry bean quality and sensory properties of cocoa
liquors made from the corresponding roasted beans.
2.3. Culture-dependent community dynamics
The culture-dependent analysis was performed immediately
after sampling following the protocol of Camu et al. (2007). Briefly,
180 ml of 0.1% (wt/vol) peptone water (Oxoid, Basingstoke, United
Kingdom) was added to 20 g of pulp and beans in a sterile stom-
acher bag that was vigorously shaken for 5 min in a Stomacher 400
(Seward, Worthington, United Kingdom). Samples (1.0 ml) of the
homogenate were serially diluted tenfold in 0.1% (wt/vol) peptone
water, from which aliquots (0.1 ml) were plated on different
selective agar media, which were incubated aerobically at 37  C in
a standard incubator (Jouan, Saint Herblain, France) for enumera-
tion (by recording the number of colony forming units, CFU) of
specific groups of microorganisms responsible for the cocoa bean
fermentation and isolation of individual colonies: malt extract agar
(MEA, Oxoid) plus 100 mg/l of oxytetracycline for yeasts, de Man-
eRogosaeSharpe (MRS, Oxoid) agar plus 200 mg/l of pimaricin for
LAB, acetic acid medium (AAM) [1% D-glucose, 0.5% ethanol, 0.3%
acetic acid, 1.5% bacteriological peptone, and 0.8% yeast extract; wt/
vol] (Lisdiyanti et al., 2003) agar plus 200 mg/l of pimaricin for AAB.
Colonies were picked up from a suitable dilution of each sample on
MRS and AAM agar media and overnight cultures were stored at
20  C in the respective broths. After transport to Belgium, colonies
were further purified through successive transfers in and plating on
the appropriate media, and a catalase test was performed. Potential
LAB (catalase-negative) and AAB (catalase-positive) isolates were
grown in MRS and mannitoleyeast extractepeptone (MYP)
medium [2.5% D-mannitol, 0.5% yeast extract, and 0.3% bacterio-
logical peptone (Oxoid); wt/vol], respectively, and stored at 80  C.
2.4. Culture-dependent identification of LAB and AAB isolates
LAB and AAB isolates were grouped and presumptively identi-
fied through (GTG)5-PCR fingerprinting of genomic DNA, according
to the procedures of Gevers et al. (2001) and De Vuyst et al. (2008),
respectively, taking into account the following procedures. DNA
from bacterial isolates was extracted from cell pellets obtained
through centrifugation (21,036  g, 15 min, 4  C) of overnight
cultures (2 ml) of LAB in MRS medium and AAB in MYP medium. For
cell lysis of AAB, proteinase K (VWR International, Darmstadt,
Germany) was used in an amount of 2.5 mg/ml of TE buffer (10 mM
TriseHCl, 1 mM EDTA, pH 8.0), as described previously (Camu et al.,
2007). All PCR amplifications with the single oligonucleotide
primer (GTG)5 were performed using a T300 Thermocycler (Bio-
metra GmbH, Goettingen, Germany). Numerical cluster analysis of
the (GTG)5-PCR profiles was performed with Bionumerics Version
 T. Lefeber et al. / Food Microbiology 28 (2011) 457e464
5.10 software (Applied Maths, Sint-Martens-Latem, Belgium).
Dendrograms were obtained by means of the unweighted pair-
group method with arithmetic averages (UPGMA) clustering algo-
rithm, with correlation levels expressed as percentage values of the
Pearson correlation coefficient. For identification and verification of
the assigned cluster identities, sequencing of the 16S rRNA gene of
representative isolates of each cluster was performed. The
sequencing results were evaluated through basic local alignment
search tool (BLAST) analysis (http://www.ncbi.nlm.nih.gov/BLAST/).
2.5. Culture-independent community dynamics
Bacterial DNA was extracted directly from 20 g fermented pulp-
bean mass samples as described previously (Camu et al., 2007),
except for purification of the aqueous phase that was performed
using a NucleoSpin column according to the manufacturer’s
instructions (Macherey Nagel GmbH, Düren, Germany). The V3
region of the 16S rRNA gene was chosen as PCR target, using the
forward f357 and reverse r518 universal primers (Muyzer et al.,
1993). PCR-DGGE was performed as described previously
(Vasilopoulos et al., 2008), using a denaturant gradient from 35% to
70%, and an appropriate bacterial reference ladder. DNA of visualised
bands of interest were excised from the gels with a sterile blade,
mixed with 40 ml of sterile water, and placed at 4  C for 24 h to let the
DNA diffuse out of the bands. These bands were vigorously vortexed
for 15 min and centrifuged (21,036  g, 15 s, 4  C) to collect the
aqueous DNA solution. Three ml of this solution was used for a PCR
reaction with the same primers, excluding the GC-clamp used to
carry out DGGE, under the same PCR conditions. The PCR products
were purified using the Wizard Plus SV Minipreps DNA Purification
System (Promega, Madison, WI, USA) and sequenced in a commercial
facility (VIB, Antwerp, Belgium). Searches in the GenBank database
were performed with the BLAST program to determine the closest
known relatives based on the partial 16S rRNA gene sequences.
2.6. Metabolite target analysis
Preparation of samples of fresh pulp, raw cocoa beans and fer-
mented dry cocoa beans to obtain aqueous extracts was carried out
as described previously (Camu et al., 2007).
Ethanol concentrations were measured by gas chromatography
(GC), using a Focus gas chromatograph (Interscience, Breda, The
Netherlands), equipped with a Stabilwax-DA column (Restek,
Bellfonte, PA, USA), a flame ionization detector (FID), and an AS
3000 autosampler. Hydrogen gas was used as carrier gas with
a constant flow rate of 1 ml/min; nitrogen gas was used as make-up
gas. The injector and detector temperatures were set at 150  C and
250  C, respectively. The column temperature program was as
follows: 0 min, 40  C; 10 min, 140  C; 12 min, 230  C; 22 min, 230
 C. The aqueous extracts were diluted (1:4) with acetonitrile to
remove proteins, filtered (0.2 mm filters, Minisart RC 4; Sartorius,
Darmstadt, Germany) prior to injection, and run with the appro-
priate external standards. Injection was performed in split mode
with a split ratio of 20:1; the injected volume was 0.5 ml. Data
processing was performed using the ChromCard for Windows
(version 1.18) software (Thermo Electro Corp., Milan, Italy). All
analyses were performed in triplicate; the average values and
standard deviations are represented as milligrams per gram of fresh
pulp, raw cocoa beans, or fermented dry cocoa beans.
Concentrations of common carbohydrates (sucrose, glucose, and
fructose) and mannitol were measured by high performance anion
exchange chromatography with pulsed amperometric detection
(HPAEC-PAD), mainly as described previously (Camu et al., 2007).
An ICS 3000 chromatograph, equipped with a CarboPacÔ PA10
column, was used (Dionex, Sunnyvale, CA, USA). The mobile phase,
459
run at a flow rate of 1 ml/min, consisted of ultra-pure water (0.015
mS/cm; eluent A), 167 mM NaOH (eluent B), and 500 mM NaOH
(eluent C) with the following gradient: 0.0 min, 87% A and 13% B;
20.0 min, 87% A and 13% B; 21.0 min, 100% C; 27.0 min, 100% C; 27.2
min, 87% A and 13% B; and 30.0 min, 87% A and 13% B. To exclude
matrix interference, quantification was carried out with standard
addition as described by Vrancken et al. (2008) with small varia-
tions. Briefly, four standard solutions with the following composi-
tions were made: ultra-pure water (solution A); 0.100 g/l of
mannitol, glucose, fructose, and sucrose (solution B); 0.200 g/l
of mannitol, glucose, fructose, and sucrose (solution C); 0.300 g/l of
mannitol, glucose, fructose, and sucrose (solution D). For each
sample, 300 ml of diluted supernatant was mixed with 300 ml of
solution A, B, C, or D. To remove proteins, 600 ml of acetonitrile was
added, diluted ten times, centrifuged (21,036  g, 15 min, 4  C), and
filtered (0.2 mm filters, Minisart RC 4, Sartorius).
Concentrations of organic acids (gluconic acid, lactic acid, acetic
acid, and citric acid) were measured by high performance anion
exchange chromatography with conductivity under ion suppres-
sion (HPAEC-CIS), as described previously (Camu et al., 2007). An
ICS 3000 chromatograph with an AS-19 column was used (Dionex).
The mobile phase, at a flow rate of 1 ml/min, consisted of ultra-pure
water (0.015 mS/cm; eluent A) and 100 mM KOH (eluent B) with the
following gradient: 0.0 min, 96% A and 4% B; 14.0 min, 96% A and 4%
B; 20.0 min, 60% A and 40% B; 25.0 min, 60% A and 40% B, 30.0 min,
100% B; 35.0 min, 100% B; 36.0 min, 96% A and 4% B; and 40.0 min,
96% A and 4% B. To exclude matrix interference, quantification was
carried out with standard addition as described by Vrancken et al.
(2008) with small changes. The compositions of the four standard
solutions were: ultra-pure water (solution A); 0.100 g/l of gluconic
acid, lactic acid, acetic acid, and citric acid (solution B); 0.200 g/l of
gluconic acid, lactic acid, acetic acid, and citric acid (solution C);
0.300 g/l of gluconic acid, lactic acid, acetic acid, and citric acid
(solution D). For each sample, 300 ml of diluted supernatant was
mixed with 300 ml of solution A, B, C or D. To remove proteins, 600
ml of acetonitrile was added, centrifuged (21,036  g, 15 min, 4  C),
and filtered (0.2 mm filters, Minisart RC 4, Sartorius).
For all HPAEC analyses, a volume of 10 ml was injected into the
column. The concentrations and the standard deviations of the
carbohydrates, mannitol, and organic acids in the original aqueous
extracts were calculated as described by Vrancken et al. (2008) and
are reported as milligrams per gram of fresh pulp, raw cocoa beans,
or fermented dry cocoa beans.
2.7. Quality assessment of fermented dry cocoa beans and
production and sensory analysis of cocoa liquors
Fermented dry cocoa beans were checked for appearance (bean
count per 300 g of cocoa beans, and bean quality; Barry Callebaut,
Abidjan, Côte d’Ivoire), making use of the cut test as described
before (Camu et al., 2008a).
Fermented dry cocoa beans (500 g) were whole-bean roasted
(140  C, 30 min) in a hot-air oven. Cocoa beans were deshelled by
breaking and winnowing of the nibs. Roasted nibs were ground into
cocoa liquor using a ball miller (Retsch GmbH, Haan, Germany).
Sensory analysis of cocoa liquors was performed by a trained panel
of five members of Barry Callebaut France (Louviers, France), as
described before (Camu et al., 2008a).
3. Results
3.1. Culture-dependent community dynamics
Enumeration and isolation. Fig. 1A shows a representative
course of the different microbial groups selected for thorough
460
A 10
9 reached at the end of the fermentations (Fig. 1B).
(lo g C F U g -1 p u lp -b ea n m a ss)
C o m m u n ity d y n a m ics
8 3.2. Culture-dependent identification of LAB and AAB isolates
7 Out of a total of 211 isolates from MRS agar, 156 isolates could be
6
5
4
3
0 24
B 50
45
Te m p e r a t u r e ( ° C )
40
35
30
25
20
0 24
Fig. 1. (A) Microbial cell counts on MEA, ( ), MRS, (-), and AAM, (:) during spon-
taneous cocoa bean vessel fermentations, represented by the laboratory vessel
fermentation 2. (B) Temperature (black line) and pH (grey line) in the cocoa pulp-bean
mass.
plating during vessel fermentation 2, which is representative of the
three spontaneous cocoa bean vessel fermentations performed
(including pH and temperature). For comparison, the lowest and
highest values of the cell counts for the three fermentations per-
formed are included below. Yeast counts on MEA agar varied from
below the detection limit to 3.60 log CFU/g pulp-bean mass at the
start of the fermentations. After 48 h, the yeast community grew to
6.99e7.37 log CFU/g pulp-bean mass in the three fermentations.
Upon further fermentation, yeast counts remained stable and
decreased slightly at the end of the fermentations, except for vessel
3 wherein the yeast community dropped to 4.65 log CFU/g pulp-
bean mass. The initial level of the LAB counts on MRS agar was
between 3.30 and 3.48 log CFU/g pulp-bean mass. LAB grew to
a maximum size of 8.82e9.13 log CFU/g pulp-bean mass after
72e96 h. After 144 h of fermentation, the community was still
7.48e7.70 log CFU/g pulp-bean mass. Although AAM agar was used
to monitor the growth of AAB, apparently a lot of catalase-negative
bacteria (later identified as LAB) grew on this medium too. No
colonies of AAB were found on AAM agar at the beginning of the
fermentation. A maximum bacterial community of 8.71e8.90 log
CFU/g pulp-bean mass on AAM agar was reached after 72e96 h of
fermentation. At the end of the fermentations, counts on AAM agar
were between 7.46 and 8.12 log CFU/g pulp-bean mass.
The initial pH of the pulp was  3.6 at the start of the fermen-
tations. The pH increased during the first 48 h of fermentation,
followed by a short decrease, in turn always followed by a slow
increase to  3.9 at the end of the fermentations (Fig. 1B).
T. Lefeber et al. / Food Microbiology 28 (2011) 457e464
At the start of the fermentations, the temperature of the pulp-
bean mass was  25  C. The maximum temperature of  41  C was
recovered and purified and represented catalase-negative LAB. The
(GTG)5-PCR fingerprints of these LAB isolates clustered into two
main groups, whose identity was confirmed by 16S rRNA gene
sequencing as L. plantarum and L. fermentum (Fig. 2). The L. fer-
mentum cluster represented 59% of the total number of LAB isolates,
while the cluster of L. plantarum represented 34%. The third cluster
was identified as Fructobacillus pseudoficulneus (6%). Besides
48 72 96 120 144 L. fermentum and L. plantarum, F. pseudoficulneus was present
Time (h) during the first 72 h of vessel fermentation 3 only. In this vessel,
L. fermentum dominated the last 48 h of the fermentation, while
5.0 L. plantarum dominated the first 96 h. A fourth cluster only repre-
sented 1% of the LAB isolates and was identified as L. plantarum, the
(GTG)5-PCR fingerprints of which were different from those of the
4.5 main cluster of L. plantarum present in these fermentations.
In total, 180 colonies were picked up from AAM agar. Although
41% of the isolates did not even survive transport from Côte d’Ivoire
to Belgium, 43% of the survivors were catalase-negative bacteria
pH
4.0
and only 16% (29 isolates) were AAB (catalase-positive). Hence, as
mentioned above, AAM counts were not representative for the
growth of AAB. Almost all AAB isolates were picked up during the
3.5 last 48 h of fermentation. No more than three clusters were
obtained and identified through (GTG)5-PCR fingerprinting, namely
A. pasteurianus (58%), Acetobacter ghanensis (35%), and Acetobacter
senegalensis (7%) (data not shown).
3.0
48 72 96 120 144
Time (h) 3.3. Culture-independent community dynamics
 The 16S rRNA-PCR-DGGE results showed that L. plantarum and L.
fermentum were the dominating LAB throughout the fermentation
processes (Fig. 3). Bands for Acetobacter species were found at the
end of the fermentation courses. In the beginning of the three
fermentations, DGGE bands for species of Tatumella and Pantoea
were found too (Fig. 3).
3.4. Metabolite target analysis of the spontaneous cocoa bean
vessel fermentations
Tables 1 and 2 are representative for the course of the metab-
olite concentrations of the three spontaneous cocoa bean vessel
fermentations performed. For comparison, the lowest and highest
values of substrate and metabolite concentrations for the three
fermentations are included below. Variations within one fermen-
tation and between different fermentations were mainly due to
sample collection, as the fermenting mass was not mixed during
fermentation and hence could not be kept homogeneous. Pectin
degradation by yeast (not measured) caused pulp sweating that
could drain away through a hole in the bottom of the vessels.
At the start of the fermentations, no sucrose was found. The
initial concentrations of glucose and fructose in the pulp were 92.4
 7.09 to 131  18.5 mg/g pulp and 116  6.39 to 183  24.6 mg/g
pulp, respectively, and both carbohydrates were consumed simul-
taneously and completely exhausted after 96 h of fermentation.
Most of the pulp was liquefied during the first 72 h of fermentation.
Initially, the concentration of citric acid was 8.46  1.06 to 32.8 
7.58 mg/g pulp. Citric acid disappeared as well during the first 72 h
of fermentation. Ethanol reached a maximum concentration of 14.3
 0.32 to 18.7  1.01 mg/g pulp after 72 h of fermentation. Lactic acid
reached a maximum concentration of 1.59  2.59 to 40.0  3.84 mg/
 T. Lefeber et al. / Food Microbiology 28 (2011) 457e464 461

Fig. 2. Dendrogram based on the numerical analysis of generated, digitized (GTG)5-PCR fingerprints from isolates present during cocoa bean vessel fermentations and identified as
LAB. Banding patterns were clustered together with the reference strains, using the unweighted pair-group method with arithmetic averages (UPGMA) algorithm, with correlation
levels expressed as percentage values of the Pearson correlation coefficient. Species validation of representative strains of each cluster was carried out by 16S rRNA gene sequencing
(indicated by *).

g pulp after 72e96 h of fermentation. After 72 h, production of
mannitol out of fructose increased. The maximum concentration of
mannitol in the pulp was 7.93  3.89 to 36.6  4.42 mg/g pulp.
Acetic acid reached a concentration of 25.0  6.81 to 56.9  9.13 mg/
g pulp. At the end of the fermentations, lactic acid was oxidized by
AAB. At the start of the fermentations, production of gluconic acid
was found (except for vessel fermentation 1), with a maximum
concentration of 8.52  2.77 to 21.3  0.88 mg/g pulp.
The main carbohydrate inside the beans was sucrose with an
initial concentration of 13.5  2.90 to 19.9  2.10 mg/g beans. The
concentrations of glucose and fructose were approximately
constant during the fermentations. The maximum concentrations
of penetrated ethanol and lactic acid were 7.19  0.63 to 10.7  1.30
mg/g beans and 11.7  1.57 to 14.4  3.00 mg/g beans, respectively.
After 72 h, small amounts of lactic acid were measured, maximally
2 mg/g beans. The citric acid level inside the beans was constant;
462 T. Lefeber et al. / Food Microbiology 28 (2011) 457e464

4. Discussion

Today, cocoa bean fermentation is still carried out spontane-

Fig. 3. 16S rRNA-PCR-DGGE profiles (35e70% denaturant gradient) of bacteria present
during vessel fermentation 2, sampled during six days. All 16S rRNA gene fragments
were amplified by the f357-r518 universal primer pair. The ladder (L) consisted of (a)
Lactobacillus plantarum LMG 6907T, (b) Lactobacillus fermentum LMG 6902T, (c) Leu-
conostoc pseudomesenteroides isolate 274 (Camu et al., 2007), (d) Acetobacter pasteur-
ianus LMG 1262T, (e) Gluconoacetobacter europaeus LMG 18890T. The closest relatives of
the fragments sequenced (the percentages of identical nucleotides compared to
sequences retrieved from the GenBank database through BLAST analysis are shown in
parentheses) were as follows: L. plantarum (1, 100%), L. fermentum (2, 100%), Tatumella
sp. (3), plant DNA (4), Pantoea sp. (5), and Acetobacter sp. (6).
only at the end of the fermentations the concentration lowered. No
mannitol was found inside the beans.
Comparing the final composition of beans dried after different
fermentation times, sucrose and acetic acid concentrations inside
the beans differed. The longer the beans resided in the pulp, the
lower the sucrose content was and the higher the acetic acid
concentration was. Lactic acid and mannitol concentrations were
approximately similar. No ethanol was found, indicating that it was
evaporated during drying and storage.
3.5. Quality assessment of fermented dry cocoa beans and
production and sensory analysis of cocoa liquors
Cocoa beans of good quality were produced in all cases, repre-
senting an average bean count of 98/100 g beans and only less than
20% of beans lacking fermentation. The spider diagram of Fig. 4 is
representative for the four cocoa liquors produced from fermented
dry beans of the three fermentations. These cocoa liquors displayed
a fruity note and a low chocolate flavour. In the samples that were
only fermented for 72 h, a raw flavour was found. Based on these
sensory analyses, the vessel fermentations could be considered as
finished after 96 h of fermentation.
ously (De Vuyst et al., 2009; Schwan and Wheals, 2004). Hence, this
uncontrolled cocoa bean processing depends on farming practices,
not only with respect to pod/bean pre-processing (harvest time,
storage time, pod selection and opening, bean selection, etc.) but
also regarding fermentation method (e.g., heap or box). To control
and further steer cocoa bean fermentations by means of starter
cultures, a uniform process is necessary. During the present study,
the spontaneous cocoa bean fermentation process was mimicked in
plastic vessels successfully.
First, it was shown through culture-dependent and -indepen-
dent microbiological analysis that the same species diversity and
community dynamics occurred in the vessels compared to spon-
taneous cocoa bean fermentation processes. Moreover, it was
shown that precautions taken in the laboratory to carry out the
vessel fermentations, for instance concerning the materials used to
open the cocoa pods and to collect and weigh the pulp-bean mass
as well as decontamination of the workers’ hands and vessels to
exclude disadvantageous cross-contamination, had no influence on
the species diversity of the LAB and AAB involved in the fermen-
tation process. The bacterial species encountered were mainly
L. plantarum and L. fermentum as LAB species, and A. pasteurianus,
A. ghanensis, and A. senegalensis as AAB species. These bacterial
species have been shown to be prevalent in Ghanaian spontaneous
cocoa bean heap fermentations too (Camu et al., 2007, 2008b). It
indicates that the microbial inoculum of the laboratory vessel
fermentations could only be derived from the cocoa pod surfaces.
Cocoa pods have been reported to be the main inoculation source of
spontaneous cocoa bean heap fermentations before (Camu et al.,
2008b), besides air transfer through microorganisms (Schlegel
and Jannasch, 2006). At the farms, knives, baskets, boxes, etc.,
contaminated with the remains and desired microorganisms from
previous fermentations, contribute to the accidental contamination
of the pulp-bean mass (Wood and Lass, 1985). In addition, in the
case of fermentations at the farms, banana or plantain leaves are
used to cover the pulp-bean mass, which are another source of
microorganisms (Camu et al., 2007). The community dynamics
consisted of a succession of yeasts, LAB, and AAB, the growth of
each of which declined after having reached a maximum. However,
the relatively stable size of the yeast community in the vessels
compared to spontaneous fermentations in the field, where yeast
growth rapidly declines, as well as the late boost of AAB growth and
no pronounced elevated temperature at the end of the fermenta-
tions, may be the result of insufficient mixing of the cocoa pulp-
bean mass in the vessels. The transient dominance of Fructobacillus
pseudofilculneus in one of the three vessel fermentations can be
ascribed to its association with higher levels of L. plantarum than of
L. fermentum in the beginning of that fermentation. In general, the
species F. pseudofilculneus, which is adapted to fructose-rich envi-
ronments, as it actively grows on fructose (Endo and Okada, 2008),
mainly occurs in the beginning of the cocoa bean fermentation

Table 1
Course of metabolite concentrations in the pulp during laboratory vessel fermentation 2.

Fermentation time (h) Concentration (mean  standard deviation in mg/g)

Glucose Fructose Mannitol Ethanol Lactic acid Acetic acid Citric acid Gluconic acid
0 95.6  4.09 182  24.5 0 0 0 0 11.3  1.54 0
24 44.5  1.53 70.2  1.96 2.13  0.39 2.30  0.09 3.08  0.32 4.34  0.82 14.5  0.81 8.52  2.77
48 30.7  2.84 9.11  2.94 0.14  0.18 8.99  0.66 0.35  0.11 0.73  0.35 2.53  0.53 3.08  0.58
72 7.77  1.02 12.4  1.78 3.53  0.20 16.6  0.49 12.4  1.23 6.89  0.44 0.42  0.30 3.24  0.52
96 0.36  0.20 1.11  1.12 17.6  2.37 14.3  0.37 16.3  1.97 38.5  2.30 0.58  0.49 3.58  1.10
120 0.51  0.19 1.76  1.46 15.9  3.85 5.95  0.09 6.52  1.73 18.8  0.58 0.31  0.11 2.46  0.55
144 0 0 5.62  2.43 3.68  0.19 3.74  1.10 10.6  0.47 0.63  0.39 1.42  0.61
 T. Lefeber et al. / Food Microbiology 28 (2011) 457e464 463

Table 2
Course of metabolite concentrations in the beans during laboratory vessel fermentation 2.

Fermentation time (h) Concentration (mean  standard deviation in mg/g)

Sucrose Glucose Fructose Ethanol Lactic acid Acetic acid Citric acid
0 19.8  2.10 2.32  0.49 3.60  0.53 1.00  0.04 0 0 7.39  0.88
24 14.3  0.69 1.60  0.19 2.44  0.28 2.01  0.14 0.11  0.08 2.61  0.17 7.81  1.39
48 6.02  0.96 1.24  0.28 1.77  0.35 3.60  0.14 0.03  0.06 0.38  0.17 6.69  0.69
72 3.85  0.63 0.65  0.28 1.09  0.26 5.13  0.08 0.23  0.01 1.71  0.21 5.59  1.59
96 2.42  0.68 1.35  0.47 1.97  0.55 7.19  0.63 0.63  0.08 8.02  0.54 5.42  0.43
120 3.73  0.64 1.83  0.21 2.00  0.32 5.93  0.53 0.82  0.13 12.6  2.85 5.77  0.89
144 1.61  0.26 2.16  0.42 2.12  0.34 2.65  0.06 0.76  0.08 14.1  2.55 5.41  0.70

process, as does L. plantarum in general (Camu et al., 2007), when
fructose is plentiful and the pH is low (due to the presence of citric
acid in the pulp). Further, fructose is used as an alternative external
electron acceptor by strictly heterofermentative L. fermentum,
thereby being converted into mannitol, upon further fermentation
(Camu et al., 2007; Wisselink et al., 2002).
Also, it was shown through metabolite target analyses that
similar substrate consumption (pectin, sucrose, glucose, fructose,
and citric acid) and metabolite production kinetics (ethanol, lactic
acid, acetic acid, mannitol, and gluconic acid) occurred in the
vessels compared to spontaneous cocoa bean fermentation
processes. However, the normal succession of microbial activities
seemed to be retarded, compared with the natural fermentations
previously described (Ardhana and Fleet, 2003; Camu et al., 2007,
2008b; Lagunes-Gálvez et al., 2007). In particular, yeast growth
was prolonged, while LAB and AAB activities took place normally.
Consequently, fruity flavour compounds could be formed by the
yeasts (Schwan and Wheals, 2004), which were reflected in the
taste of the cocoa liquors. Gluconic acid was most probably
produced by the Pantoea and Tatumella species present at the
beginning of the fermentations (Kageyama et al., 1992).
The cocoa bean vessel fermentation set-up presented in the
present paper can serve as a benchmark process for further studies
on starter culture development, for instance to compare the influ-
ence of different starter cultures of potential use for controlled
cocoa bean fermentation, which will otherwise be a very laborious
task. The use of starter cultures has been studied preliminarily
before, in particular with respect to yeasts, to enhance the pecti-
nolytic activity for improving the yield of pulp drainage during
Sour
3
Green/Wood/Raw Astringent
2
Smoked Bitter
1 Scientific Research e Flanders, the Institute for the Promotion of
Mouldy/Sack Alcalin
0
fermentation for the production of wine, jam, alcohol, syrup, and
marmalades (Buamah et al., 1997; Dzogbefia et al., 1999; Leal et al.,
2008; Samah et al., 1992; Sanchez et al., 1985). Moreover, in these
studies, yeasts have been obtained from culture collections, which
have not the same properties as wild-type strains. In general, pure
cultures isolated from mixed populations of traditional fermented
foods differ in metabolic activities, growth rate, adaptation to the
substrate, and flavour formation, traits that even vary among
strains of the same species (Leroy and De Vuyst, 2004). Therefore,
natural isolates should be tested thoroughly. Only in one particular
case yeasts have been used in combination with LAB and AAB, all
isolated from previous cocoa bean fermentations (Schwan, 1998).
These ‘controlled’ wooden box fermentations mimicked the spon-
taneous cocoa bean box fermentation perfectly. Unfortunately, it
has never been shown that these fermentations are accelerated or
that the chocolates produced from the concomitant fermented dry
beans have an elevated taste. Therefore, the laborious selection of
wild-type strains and formulation of an appropriate starter culture
can now be investigated directly on the pulp-bean mass through
‘controlled’ fermentations in plastic vessels.
5. Conclusions
To summarize, spontaneous cocoa bean fermentations can be
mimicked in plastic vessels. Additional turning of the pulp-beans
mass to enhance its aeration and mixing will certainly lead to
a better simulation of the natural process regarding fermentation
time and microbial succession. Nonetheless, the simulated labora-
tory fermentation process will be a proper tool for the selection of
microbial strains to formulate starter cultures with applications in
controlled cocoa bean fermentation processes.
Acknowledgements
This research was funded by the Research Council of the Vrije
Universiteit Brussel (OZR, GOA, and IOF projects), the Fund for
Innovation through Science and Technology in Flanders (IWT-
080357), and Barry Callebaut N.V.
References

Afoakwa, E.O., Paterson, A., Fowler, M., Ryan, A., 2008. Flavor formation and char-
Earthy Chocolate
acter in cocoa and chocolate: a critical review. Critical Reviews in Food Science
and Nutrition 48, 840e857.
Ardhana, M.M., Fleet, G.H., 2003. The microbial ecology of cocoa bean fermentations
in Indonesia. International Journal of Food Microbiology 86, 87e99.
Wine/Liquorice Grilled/Roasted Beckett, S.T., 2009. Industrial Chocolate Manufacture and Use. John Wiley & Sons,
Ltd, Chichester.
Buamah, R., Dzogbefia, V.P., Oldham, J.H., 1997. Pure yeast culture fermentation of
Aromatic Fruity cocoa (Theobroma cacao L.): effect on yield of sweatings and cocoa bean quality.
World Journal of Microbiology & Biotechnology 13, 457e462.
Fig. 4. Flavour profiles of the cocoa liquors produced from fermented dry cocoa beans Camu, N., De Winter, T., Verbrugghe, K., Cleenwerck, I., Vandamme, P., Takrama, J.F.,
derived from cocoa bean vessel fermentations. Cocoa beans were withdrawn from the Vancanneyt, M., De Vuyst, L., 2007. Dynamics and biodiversity of populations of
vessels after three (A), four (,), five (6) and six (B) days of fermentation. lactic acid bacteria and acetic acid bacteria involved in spontaneous heap
464 T. Lefeber et al. / Food Microbiology 28 (2011) 457e464
fermentation of cocoa beans in Ghana. Applied and Environmental Microbi-
ology 73, 1809e1824.
Camu, N., De Winter, T., Addo, S.K., Takrama, J.F., Bernaert, H., De Vuyst, L., 2008a.
Fermentation of cocoa beans: influence of microbial activities and polyphenol
concentrations on the flavour of chocolate. Journal of the Science of Food and
Agriculture 88, 2288e2297.
Camu, N., Gonzalez, A., De Winter, T., Van Schoor, A., De Bruyne, K., Vandamme, P.,
Takrama, J.S., Addo, S.K., De Vuyst, L., 2008b. Influence of turning and envi-
ronmental contamination on the dynamics of populations of lactic acid and
acetic acid bacteria involved in spontaneous cocoa bean heap fermentation in
Ghana. Applied and Environmental Microbiology 74, 86e98.
Daniel, H., Vrancken, G., Takrama, J.F., Camu, N., De Vos, P., De Vuyst, L., 2009. Yeast
diversity of Ghanaian cocoa bean heap fermentations. Fems Yeast Research 9,
774e783.
De Vuyst, L., Camu, N., De Winter, T., Vandemeulebroecke, K., De Perre, V.V.,
Vancanneyt, M., De Vos, P., 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.
International Journal of Food Microbiology 125, 79e90.
De Vuyst, L., Lefeber, T., Papalexandratou, Z., Camu, N., 2009. The functional role of
lactic acid bacteria in cocoa bean fermentation. In: Mozzi, F., Raya, R.R.,
Vignolo, G.M. (Eds.), Biotechnology of Lactic Acid Bacteria: Novel Applications.
Wiley-Blackwell, Ames, pp. 301e326.
Dzogbefia, V.P., Buamah, R., 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, 1e12.
Endo, A., Okada, S., 2008. Reclassification of the genus Leuconostoc and proposals of
Fructobacillus fructosus gen. nov., comb. nov., Fructobacillus durionis comb. nov.,
Fructobacillus ficulneus comb. nov and Fructobacillus pseudoficulneus comb. nov.
International Journal of Systematic and Evolutionary Microbiology 58,
2195e2205.
Gevers, D., Huys, G., Swings, J., 2001. Applicability of rep-PCR fingerprinting for
identification of Lactobacillus species. Fems Microbiology Letters 205, 31e36.
Jespersen, L., Nielsen, D.S., Honholt, S., Jakobsen, M., 2005. Occurrence and diversity
of yeasts involved in fermentation of West African cocoa beans. Fems Yeast
Research 5, 441e453.
Kageyama, B., Nakae, M., Yagi, S., Sonoyama, T., 1992. Pantoea punctata sp. nov.,
Pantoea citrea sp. nov., and Pantoea terrea sp. nov. isolated from fruit and soil
samples. International Journal of Systematic Bacteriology 42, 203e210.
Kostinek, M., Ban-Koffi, L., Ottah-Atikpo, M., Teniola, D., Schillinger, U.,
Holzapfel, W.H., Franz, C., 2008. Diversity of predominant lactic acid bacteria
associated with cocoa fermentation in Nigeria. Current Microbiology 56,
306e314.
Lagunes-Gálvez, S.L., Loiseau, G., Paredes, J.L., Barel, M., Guiraud, J.P., 2007. Study on
the microflora and biochemistry of cocoa fermentation in the Dominican
Republic. International Journal of Food Microbiology 114, 124e130.
Leal, G.A., Gomes, L.H., Efraim, P., Tavares, F.C.D., Figueira, A., 2008. Fermentation of
cacao (Theobroma cacao L.) seeds with a hybrid Kluyveromyces marxianus strain
improved product quality attributes. Fems Yeast Research 8, 788e798.
Leroy, F., De Vuyst, L., 2004. Lactic acid bacteria as functional starter cultures for the
food fermentation industry. Trends in Food Science & Technology 15, 67e78.
Lisdiyanti, P., Katsura, K., Potacharoen, W., Navarro, R.R., Yamada, Y., Uchimura, T.,
Komagata, K., 2003. Diversity of acetic acid bacteria in Indonesia, Thailand and
the Philippines. Microbiology and Culture Collection 19, 91e98.
Muyzer, G., Dewaal, E.C., Uitterlinden, A.G., 1993. Profiling of complex microbial-
populations by denaturing gradient gel-electrophoresis analysis of polymerase
chain reaction-amplified genes-codes for 16S ribosomal-RNA. Applied and
Environmental Microbiology 59, 695e700.
Nielsen, D.S., Honholt, S., Tano-Debrah, K., Jespersen, L., 2005. Yeast populations
associated with Ghanaian cocoa fermentations analysed using denaturing
gradient gel electrophoresis (DGGE). Yeast 22, 271e284.
Nielsen, D.S., Teniola, O.D., Ban-Koffi, L., Owusu, M., Andersson, T.S., Holzapfel, W.H.,
2007. The microbiology of Ghanaian cocoa fermentations analysed using
culture-dependent and culture-independent methods. International Journal of
Food Microbiology 114, 168e186.
Samah, O.A., Putih, M.F., Selamat, J., 1992. Biochemical changes during fermentation
of cocoa beans inoculated with Saccharomyces cerevisiae (wild strain). Inter-
national Journal of Food Science and Technology 29, 341e343.
Samah, O.A., Putih, M.F., Selamat, J., Alimon, H., 1993. Fermentation products in
cocoa beans inoculated with Acetobacter xylinum. ASEAN Food Journal 8,
22e25.
Sanchez, J., Daguenet, G., Guiraud, J.P., Vincent, J.C., Galzy, P., 1985. A study of the
yeast flora and the effect of pure culture seeding during the fermentation
process of cocoa beans. LWT-Food Science and Technology 18, 69e75.
Schlegel, H.G., Jannasch, H.W., 2006. Prokaryotes and their habitats. In:
Dworkin, M., Falkow, S., Rosenberg, E., Schleifer, K.-H., Stackebrandt, E. (Eds.),
The Prokaryotes: Symbiotic Associations, Biotechnology, Applied Microbiology.
Springer, New York, pp. 137e184.
Schwan, R.F., Rose, A.H., Board, R.G., 1995. Microbial fermentation of cocoa beans,
with emphasis on enzymatic degradation of the pulp. Journal of Applied
Bacteriology 79, 96e107.
Schwan, R.F., 1998. Cocoa fermentations conducted with a defined microbial
cocktail inoculum. Applied and Environmental Microbiology 64, 1477e1483.
Schwan, R.F., Wheals, A.E., 2004. The microbiology of cocoa fermentation and its
role in chocolate quality. Critical Reviews in Food Science and Nutrition 44,
205e221.
Thompson, S.S., Miller, K.B., Lopez, A.S., 2007. Cocoa and coffee. In: Doyle, M.P.,
Beuchat, L.R., Montville, T.J. (Eds.), Food Microbiology: Fundamentals and
Frontiers. ASM Press, Washington, pp. 721e733.
Vasilopoulos, C., Ravyts, F., De Maere, H., De Mey, E., Paelinck, H., De Vuyst, L.,
Leroy, F., 2008. Evaluation of the spoilage lactic acid bacteria in modified-
atmosphere-packaged artisan-type cooked ham using culture-dependent and
culture-independent approaches. Journal of Applied Microbiology 104,
1341e1353.
Vrancken, G., Rimaux, T., De Vuyst, L., Leroy, F., 2008. Kinetic analysis of growth and
sugar consumption by Lactobacillus fermentum 130101 reveals adaption to the
acidic sourdough ecosystem. International Journal of Food Microbiology 128,
58e66.
Wisselink, H.W., Weusthuis, R.A., Eggink, G., Hugenholtz, J., Grobben, G.J., 2002.
Mannitol production by lactic acid bacteria: a review. International Diary
Journal 12, 151e161.
Wood, G.A.R., Lass, R.A., 1985. Cocoa. Longman Group Limited, London.