Detailed analyses of the bacterial populations in processed cocoa beans of different geographic origin, subject to varied fermentation conditions

    Detailed analyses of the bacterial populations in processed cocoa beans of different geographic origin, subject to varied fermentation conditions Cristian Bortolini, Vania Patrone, Edoardo Puglisi, Lorenzo Morelli PII: DOI: Reference:

S0168-1605(16)30350-6 doi: 10.1016/j.ijfoodmicro.2016.07.004 FOOD 7294

To appear in:

International Journal of Food Microbiology

Received date: Revised date: Accepted date:

24 February 2016 6 July 2016 7 July 2016

Please cite this article as: Bortolini, Cristian, Patrone, Vania, Puglisi, Edoardo, Morelli, Lorenzo, Detailed analyses of the bacterial populations in processed cocoa beans of different geographic origin, subject to varied fermentation conditions, International Journal of Food Microbiology (2016), doi: 10.1016/j.ijfoodmicro.2016.07.004

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Detailed analyses of the bacterial populations in processed cocoa beans of different

T

geographic origin, subject to varied fermentation conditions

Istituto di Microbiologia, Facoltà di Scienze Agrarie, Alimentari ed Ambientali, Università

SC R

1

IP

Cristian Bortolini1,2, Vania Patrone1, Edoardo Puglisi1, Lorenzo Morelli1*

Cattolica del Sacro Cuore, Via Emilia Parmense 84, 29122 Piacenza, Italy 2

Ferrero S.p.A, Piazzale P.Ferrero 1, Alba, Italy

NU

*corresponding author

MA

Professor Lorenzo Morelli Istituto di Microbiologia Università Cattolica del Sacro Cuore

TE

D

Via Emilia Parmense 84 29122 Piacenza

CE P

Italy Tel: 0039 0523599248

AC

Fax: 0039 0523599246

e-mail: [email protected]

ACCEPTED MANUSCRIPT Abstract The quality of chocolate is influenced by several parameters, one of which is bacterial diversity

T

during fermentation and drying; a crucial factor for the generation of the optimal cocoa flavor

IP

precursors. Our understanding of the bacterial populations involved in chocolate fermentation can

SC R

be improved by the use of high-throughput sequencing technologies (HTS), combined with PCR amplification of the 16S rRNA subunit. Here, we have conducted a high-throughput assessment of bacterial diversity in four processed samples of cocoa beans from different geographic origins. As

NU

part of this study, we also assessed whether different DNA extraction methods could affect the

MA

quality of our data. The dynamics of microbial populations were analyzed postharvest (fermentation and sun drying) and shipment, before entry to the industrial process. A total of 691,867 high quality sequences were obtained by Illumina MiSeq sequencing of the two bacterial 16S rRNA

TE

D

hypervariable regions, V3 and V4, following paired-read assembly of the raw reads. Manual curation of the 16S database allowed us to assign the correct taxonomic classifications, at species

CE P

level, for 83.8% of those reads. This approach revealed a limited biodiversity and population dynamics for both the lactic acid bacteria (LAB) and acetic acid bacteria (AAB), both of which are

abundant

AC

key players during the acetification and lactic acid fermentation phases. Among the LAB, the most species

were

Lactobacillus

fermentum,

Enterococcus

casseliflavus,

Weissella

paramesenteroides, and Lactobacillus plantarum/paraplantarum. Among the AAB, Acetobacter syzygii, was most abundant, then Acetobacter senegalensis and Acetobacter pasteriuanus. Our results indicate that HTS approach has the ability to provide a comprehensive view of the cocoa bean microbiota at the species level.

Keywords: Cocoa bean, Fermentation, Lactic acid bacteria, Acetic acid bacteria, High-throughput sequencing, 16S rRNA.

ACCEPTED MANUSCRIPT 1. Introduction Cocoa beans are seeds enclosed within the mucilaginous pulp of cocoa pods, which are the

T

fruits of cocoa trees (Theobroma cacao L.). Once harvested, these beans are widely used as the

IP

main raw material in the chocolate industry. The quality of chocolate is influenced by several

SC R

parameters including good agricultural and post-harvest processing practices. The former includes the appropriate maintenance of plant population and phytosanitary state, soil and climate conditions, fruit maturation and harvest. Good post-harvest procedure includes optimized pod opening,

NU

fermentation, drying, and storage conditions (Lima et al., 2011). In particular, fermentation and

MA

drying are crucial stages for the correct formation of cocoa flavor precursors amino acids and shortchain peptides (Lima et al., 2011; Schwan and Wheals, 2004). After the pod has been opened, the cocoa beans undergo a natural fermentation of about 6-7 days, which is important not only for the

TE

D

removal of the mucilaginous pulp surrounding the seeds, but also to facilitate cocoa drying and to initiate the appropriate biochemical changes within the beans (Guehi et al., 2010). Cocoa

CE P

fermentation consists of a succession of different microbial populations (including yeasts, lactic acid bacteria (LAB), and acetic acid bacteria (AAB)), that colonize the mucilaginous pulp as a

AC

substrate for growth, and in doing so, produce several compounds such as ethanol, lactic acid, acetic acid, and aldehydes (Meersman et al., 2013). Following fermentation, the beans may be naturally or artificially dried (in rainy regions) in order to reduce their moisture content from about 60% to 67%. This step is a crucial one in determining the quality of cocoa beans; if drying is too rapid, some of the chemical reactions started during fermentation will not run to completion, with the consequent development of acidity and bitterness. On the other hand, if drying is too slow, contamination with molds can occur with the possible production of mycotoxins and off-flavors, which also incurs substantial production losses (Hamdouche et al., 2014; Lima et al., 2011). Several studies have addressed the diversity of the microbial flora during the spontaneous fermentation of cocoa beans, but most have relied on classical culture-dependent methods that are intrinsically biased because of their inability to cultivate all species. In recent years, culture-

ACCEPTED MANUSCRIPT independent molecular techniques, coupled with classical methods, have enabled a more detailed assessment of the microbial population because of their high throughput advantages and their

T

ability to detect species that may be difficult or even impossible to cultivate (Camu et al., 2007;

IP

Cocolin and Ercolini, 2008; Meersman et al., 2013; Nielsen et al., 2007; Papalexandratou et

SC R

al.,2011a; 2011b; 2013). Until now, polymerase chain reaction denaturing gradient gel electrophoresis (PCR-DGGE) was the most popular molecular technique, despite its low-throughput limitations (De Melo Pereira et al., 2013). More recently, high-throughput sequencing (HTS) has

NU

permitted to achieve a greater resolution and detection sensitivity than the one obtained with the

MA

analyses of DGGE gel banding patterns (Illeghems et al., 2012; Parente et al., 2015; Polka et al., 2015). This type of analysis has proven to be a very powerful instrument for the assessment of microbial populations (Illeghems et al., 2012). However, HTS methodologies have some

TE

D

disadvantages, which include a dependence on DNA extraction methodologies and primer sensitivity, both of which can affect the richness and diversity of the detected microbial populations

CE P

(Delmont et al., 2011; Meersman et al., 2013). The issue of quantity and quality of DNA available for PCR-based analyses is particularly

AC

relevant for complex matrixes such as fermented cocoa beans that are rich in polysaccharides and flavonols (polyphenols) that naturally inhibit PCR (Porebski et al., 1997; Spaniolas et al., 2008; Tortajada et al., 2009). A number of studies comparing different methods of DNA extraction from cocoa or chocolate have been carried out in the past, but they were primarily focused on the detection of genetically modified soybean DNA in chocolate (Gryson et al., 2004; Haymes et al., 2004); relatively few were performed for analyzing the microbial composition of bacteria and filamentous fungi (Fredricks et al., 2005; Tortajada et al., 2009). Here, we have carried out a highthroughput assessment of bacterial diversity for samples of processed cocoa beans derived from Cameroun, Ghana, and Ivory Coast. Our aim was to evaluate whether geographic origin of the sample, and fermentation conditions, could impact bacterial composition, and whether the DNA extraction method used could further bias the accuracy of these analyses.

ACCEPTED MANUSCRIPT 2. Materials and methods

T

2.1 Sample collection

IP

Four samples of processed-dried cocoa beans were collected from different production sites

SC R

in Africa: Ghana (S1), Ivory Coast (S2, S4), and Cameroon (S3) (Table 1). One sample from the Ivory Coast had been fermented with a starter culture, while the other three had undergone spontaneous fermentation processes. Samples originating from Ghana (S1) and Ivory Coast (S2-S4)

NU

were fermented in wooden boxes during the main cropping period (October-November 2014), while the sample from Cameroon (S3) was fermented during the mid-crop (June to July 2014), with the

MA

traditional heap method using banana leaves

D

2.2 DNA extraction

TE

Processed cocoa beans (100 g) were homogenized using a grinder at maximum speed for 60

CE P

sec and the resultant mixture collected in sterile tubes. DNA was harvested in triplicate using the 3 different DNA extraction methods, thus resulting in a total of 9 replicates per sample. DNA

AC

extraction was carried out according to the manufacturer’s protocol for three different kits; the Maxwell® 16 FFS Nucleic Acid Extraction Kit (REF custom Cat. X9431 Promega, USA), the GMO Extraction Kit (REF 4466336, Life Technologies, USA), and the Fast DNA Spin Kit for Soil (REF 116560000 MPbio, USA). For each assay, DNA was eluted in 100 µL of sterile TE buffer (10 mM Tris/HCl, 1 mM EDTA pH 8.0and stored at -20 °C until use. DNA was quantified with the Quant-iT™ HS ds-DNA assay kit (Invitrogen, Paisley, UK) using a QuBitTM fluorometer. Two µL samples of each extract were also resolved by electrophoresis on a 0.8% agarose gel to verify DNA quality.

2.3 DNA amplification

ACCEPTED MANUSCRIPT To analyze bacterial diversity, the V3-V4 region of the 16S rRNA gene was amplified by PCR using the universal primers 343f (5'-TACGGRAGGCAGCAG-3'), and 802r (5'-

T

TACNVGGGTWTCTAATCC-3'). PCR amplification was carried out with the Phusion Flash High-

IP

Fidelity Master Mix (Thermo Fisher Scientific, Inc., Waltham, MA, USA). The reaction mix

SC R

comprised 12.5 µl of Phusion Flash High-Fidelity Master Mix, 1.25 µl of each primer (10µM), 0.2 ng of DNA template, and PCR ultrapure water. In order to perform simultaneous analyses of all samples in a single sequencing run, each sample was amplified using a forward primer with a 9-

NU

base extension at the 5' end, which acts as a tag (Vasileiadis et al., 2015). The possibility of

MA

generating anomalous data, due to non-specific primer annealing, was reduced by incorporation of a two-step PCR, as described in Berry et al. (2011). PCR products generated from the second step were multiplexed as a single pool using equivalent molecular weights (20 ng). This pool was

TE

D

purified using the solid phase reversible immobilization (SPRI) method (Agencourt AMPure XP kit (REF A63880, Beckman Coulter, Italy, Milano)), then sequenced by Fasteris S.A. (Geneva

CE P

Switzerland), using the TruSeq DNA sample preparation kit (REF 15026486, Illumina Inc, San Diego, CA) for amplicon library preparation. Sequencing was performed with the MiSeq Illumina

AC

instrument (Illumina Inc, San Diego, CA) with V3 chemistry generating 300 bp paired-end reads.

2.4 Sequence data preparation and analyses The MiSeq Control software version 2.3.0.3, RTA v1.18.42.0, and CASAVA v1.8.2 were used for base calling, and the Illumina barcode demultiplexing process. As the V3-V4 16S rRNA gene amplicons are shorter than 500 bp, so 300 bp paired-end reads per amplicon were sufficient for re-constructing full-length V3-V4 regions. This task was performed using the “pandaseq” script (Bartram et al., 2011), which imposes a minimum 30 bp overlap between read pairs, with a maximum of 2 allowed mismatches. Sequences were further demultiplexed according to sample indexes and primers using the Fastx-toolkit (http://hannonlab.cshl.edu/fastx_toolkit/). Large homopolymers (≥10), sequences that aligned outside of the targeted V3-V4 region, chimeric

ACCEPTED MANUSCRIPT amplicons, and sequences that were classified by the Bayesian classifier (80% bootstrap cutoff) as non-targeted taxa (versus the Mothur version of the ribosomal database project (RDP)) were purged

T

following their identification using Mothur version 1.32.0 (Claesson et al., 2009; Edgar et al., 2011;

IP

Schloss et al., 2009; Wang et al., 2007). Two approaches were then followed for sequence data

SC R

analyses, the operational taxonomic unit (OTU), and taxonomy-based approaches. Mothur v1.32.0 (Schloss et al., 2009) was used to obtain OTU and taxonomic matrices, while R version 3.0.0 (R Development Core Team, 2011), supplemented with the Vegan package (Dixon, 2003), was used to

NU

perform statistical analyses.

The composition of analyzed samples in OTUs and taxonomic groups were used to estimate α

MA

and β diversity values. Observed Richness (S) and the Simpson's Diversity Index (D) were calculated as measures of diversity. Simpson’s Index quantifies the biodiversity of a habitat and is

TE

D

based on richness and evenness of species (Heip et al., 2001). The Good's coverage estimate was calculated to assess the "percentage diversity" captured by sequencing (Good, 1953). Metastats

AC

treatments.

CE P

(Paulson et al., 2011) was then applied to identify features that were significantly different between

3. Results and Discussion 3.1 Data processing and taxonomic classification A total of 691,867 sequences variously distributed among replicates were obtained. Preliminary analyses for the entire data set revealed a consistent coverage of ≥ 97%. In order to avoid biases related to the analysis of samples having different sizes, the dataset was downscaled to 2318 sequences per sample, i.e., the number of sequences in the sample S2_life-techW. Two samples had a number of sequences lower than 2318: namely S2_life-techG (134 sequences) and “S3_mp-bioW” (870 sequences), but we found that a downscale to either one of them would have resulted in an unacceptable loss of coverage. On the contrary, a downscale to 2318 sequences per sample resulted in the loss of these two samples but maintained a consistent coverage of ≥ 97%.

ACCEPTED MANUSCRIPT Almost all of the Illumina derived sequences were correctly classified to family level (99.9%), with a slight decrease at genus level (96.7%), then a dramatic reduction at species level

T

(36.9%). Given that Acetobacter and Lactobacillus were the most abundant genera in our samples,

IP

in agreement with the literature (Camu et al., 2007; Hamdouche et al., 2014; Illeghems et al., 2012),

SC R

the reference database was improved by adding the sequences available in RDP for all of the species belonging to these two genera. This database “upgrade” allowed us to increase taxonomical identification at the species level to 83.8%. A similar approach was taken by Polka et al. (2015),

NU

who executed a manual curation of their 16 rRNA database in order to improve the characterization

MA

of bacteria identified in fermented meats. Our data allow us to draw two conclusions. First, microbial communities associated with end-process cocoa beans are primarily composed of species with known 16S sequences that are represented in public databases. Second, sequencing of the two

TE

D

hypervariable regions (V3-V4) is sufficient to appreciate the biodiversity of these samples, as already shown in other ecological studies using HTS with fermented foodstuffs (Bassi et al., 2015;

CE P

Greppi et al., 2015; Polka et al., 2015).

AC

3.2 DNA extraction effects

Until now, few papers have compared DNA extraction methods in terms of their efficacy for cocoa bean analyses (Gryson et al., 2004, 2007; Tortajada et al., 2009), and none have investigated how DNA extraction might affect micro-biome capture by HTS. In the present study, we assessed three different commercial DNA extraction methods and compared them in terms of DNA yield and quality. Regarding DNA recovery per gram of sample, no significant differences were found as a function of the DNA extraction method used (data not shown). In assessing whether the structure of bacterial communities was altered depending on the extraction method used, a db-RDA (distancebased redundancy analysis) model was applied to relative OTU distributions using, as a fixed effect, the DNA extraction method used (supplementary data Fig. S1). The db-RDA model was significant (P<0.005), although a limited percentage of variance could be explained (>12%). Analogously, α

ACCEPTED MANUSCRIPT diversity analysis based on Simpson’s index (D) and Observed Richness (S) (Fig. 1) showed that in all cases, the DNA extraction method didn’t result in significant differences for the calculated

T

indexes. Together our data suggest that all three DNA extraction methods are equally suitable for

SC R

IP

investigating cocoa bean bacterial ecology.

3.3 Bacterial ecology in processed cocoa beans of different geographic origin In accordance with evidence from Illeghems et al. (2012), analyses of phyla revealed an

NU

abundance of Firmicutes and Proteobacteria, and low numbers of Actinobacteria (supplementary

MA

data Fig. S2), while at the order level, Alphaproteobacteria were most abundant followed by Bacilli and Gammaproteobacteria (data not shown). A comprehensive picture of the most abundant families (≥5% in at least one sample), derived from the three geographic areas, and processed with

TE

D

the three DNA extraction kits, is shown in Fig. 2. Although the samples showed a high heterogeneity, some commonalities were identified. All samples revealed a strong predominance of

CE P

Acetobacteraceae, followed by Lactobacillaceae; key players during the acetification and lactic acid fermentation phases, respectively (Camu et al., 2007). The general picture of the most

AC

abundant bacteria at genus level revealed a high percentage of Acetobacter, ranging from approximately 58% in Ghanaian samples to ~36% in Ivorian samples that employed a starter culture (Fig. 3a and 3b). In order of declining abundance we then identified Lactobacillus, Bacillus, Rummeliibacillus,

Acinetobacter,

Enterococcus,

Klebsiella,

Weissella,

Lysinibacillus,

Corynebacterium, Staphilococcus, and Stenotrophomonas. Interestingly, the samples from Ivory Coast that incorporated a starter culture showed a high percentage (~12%) of Acinetobacter. In order to better estimate the potential of HTS to define the composition of bacterial communities at species level, specific analyses were carried out to investigate the species composition of acetobacteria and lactobacilli. This fine-scale approach was permitted by the taxonomic assignments highlighted in Fig. 1b, which showed that up to 83.8% of all sequences could be correctly classified at species level. Regarding the composition of acetobacteria

ACCEPTED MANUSCRIPT (supplementary data Fig. S3), all samples were dominated by Acetobacter syzygii, followed by Acetobacter pasteurianus, Acetobacter senegalensis, Acetobacter sicerae, and Acetobacter

T

tropicalis. Clustering of samples based on their composition of these species underlined their well-

IP

defined and homogeneous composition regardless of DNA extraction method used. Regarding

SC R

geographical origin, as noted for classifications at species level, we observed a clustering of all the samples originating from Ghana (S1). However, lactobacilli were more heterogeneously distributed than acetobacteria, with no clusters detected. Lactobacillus fermentum was predominant in all but

NU

two samples, followed by Lactobacillus plantarum / paraplantarum, Lactobacillus nagelii,

MA

Lactobacillus vaginalis and Lactobacillus ghanensis (supplementary data Fig. S4). Total bacterial communities in the four cocoa beans samples were also studied by an OTUbased approach. Firstly, a db-RDA model was applied in order to test if the geographic origin of the

TE

D

samples had any effect on their pattern of bacterial ecology. As outlined in Fig. 4, the model was significant (P=0.01), but again with a limited explanation of variance (11% of total variance).

CE P

Samples were generally not clustered, with the exception of S1 (of Ghanaian origin) and S4 (Ivory coast with starter), which clustered separately from each other.

AC

Detailed analyses were then carried out using a Metastats model to identify any significant differences between samples in terms of their distribution of the most abundant OTUs (i.e., those covering 95% of total bacterial diversity, Fig. 5). The same OTUs are also reported in Table 2, where a bibliographic assessment of their possible ecological role in cocoa bean fermentation is presented. Among the most abundant OTUs, Pediococcus spp., previously detected during cocoa fermentation in Brazil (Papalexandrotou et al.,2011a), have been shown to possess antifungal activity related to their production of bacteriocin and other anti-microbial molecules (Essia Ngang et al., 2014). Interestingly, both samples from Ivory Coast (S2-S4) showed a higher percentage of Pediococcus spp compared with samples from Ghana (S1) and Cameroun (S3). Additionally, we identified an abundance of two particular OTUs classified respectively as Enterobacteriaceae and Staphylococcus saprophyticus in the Cameroonian sample (S3). S3 was ground fermented, using

ACCEPTED MANUSCRIPT banana leaves as a casing; a method that is difficult to standardize (compared with wooden box fermentation) and susceptible to increased contamination from soil and manual handling.

T

Contaminants can include Enterobacteriaceae that comprise many potential animal and plant

IP

pathogens and S. saprophiticus, previously isolated from fermented sausages (Rebecchi et al.,

SC R

2015), but now identified in processed cocoa beans. Arthrobacter solii, isolated for the first time from wastewater reservoir sediment (Roh et al., 2008) was also abundant in S3, as well as S4. Among the Enterobacteriaceae, DNA of Klebsiella pneumoniae was found in all four samples, but

NU

with a trend that reversed that seen for OTU25 (unclassified Enterobacteriaceae). K. pneumoniae is

MA

involved in vitamin B12 synthesis during soybean food fermentation in Indonesia (Wei et al., 2012), and in Mexican fermented drinks (Alcantara-Hernandez et al., 2010). We also reported the presence of Corynebacterium variabile that contributes to the development of flavor and textural properties

TE

D

in smear-ripened cheeses (Schröder et al., 2011). C. variabile was most abundant in sample S2. The statistical differences between S2 and S4, both originating from Ivory Coast, could be explained by

CE P

the use of a starter culture for S4, which inhibits extant microbes in order to drive fermentation in a defined way that supports flavor acquisition. Among the Acinetobacter species, the most

AC

representative were Acinetobacter iwoffi, Acinetobacrter radioresistens, and Acinetobacter guillouiae. A. radioresistens, known to produce an alkaline lipase (Li et al., 2005) was previously identified during the fermentation of pulque, a Mexican alcoholic beverage (Escalante et al., 2008). The percentage of A. radioresistens differed significantly between S1 and S3, compared with samples from Ivory Coast (S2/S4). Among the novel species identified for the first time on cocoa beans were Lysinobacillus boronitolerans, which also differed significantly between S2/S3/S4 and S1. L. boronitolerans was previously isolated from traditional Korean fermented soybean products (Nam et al., 2012) and was proposed as a potential biological control agent for diseases that affect cocoa (Melnick et al., 2011). Among the Bacillus species, the most representative were B.circulans, B.safensis and B.clausii. B.circulans is well represented in sample S4, and is known for its amylase activity during

ACCEPTED MANUSCRIPT traditional Vietnamese alcohol fermentation (Thanh et al., 2008), and its action during alkaline fermentation of soybean foods in India (Chettri and Tamang, 2015). The role of B. safensis is

T

currently unknown, however it was recently isolated from fermented seeds of Hibiscus sabdariffa,

IP

which are used as additives to produce Kantong, an acid fermented seed condiment in Ghana.

SC R

Consistent with these data, we found a relatively high percentage of B. safensis in the sample originating from Ghana (S1) (Kpikpi et al., 2014). B. clausii, as our second mostly abundant OTU, was most frequently found in S2, and was previously detected (by its 16S RNA phylogenetic

NU

affiliation) during the industrial processing of cocoa beans to cocoa powder (Lima et al., 2012).

MA

This finding suggests that B. clausii establishes itself after fermentation, probably during drying and subsequent stages.

Among LAB species we reported the presence of L. vaginalis, L. nagelii, L. plantarum/

TE

D

paraplantarum, Enterococcus camelliae, Weissella paramesenteroides, Enterococcus casseliflavus and L. fermentum. L. vaginalis showed a significant difference between S1 and the remaining

CE P

samples, and has been reported in spontaneously fermented buckwheat and teff sourdoughs (Moroni et al., 2011). In agreement with the literature (Papalexandratou et al., 2011b), L. nagelii was found

AC

among the LAB species with a significant difference between samples originating from Ivory Coast (S2/S4), and S1 and S3, for which the abundance of L. nagelii was reduced. According to the literature (Nielsen et al., 2007) L. plantarum/paraplantarum, previously detected in processed cocoa beans, was most abundant in the Cameroonian sample (S3), which was heap fermented, compared with S1 and S2 (both box fermented). OTU0016 (L. plantarum/paraplantarum) was one of the most abundant LAB in our samples, in agreement with the literature (De Vuyst and Weckx, 2016; Visintin et al., 2016;). Enterococcus camelliae has been previously isolated from fermented tea leaves in Thailand (Sukontasing et al., 2007), and, interestingly, we found a significant percentage in S4 (starter inoculation), suggesting a possible role during lactic acid fermentation with starter cultures. Weissella paramesenteroides, previously isolated during cocoa beans fermentation in Malaysia (Papalexandratou et al., 2013), was significantly higher in S1 and S3,

ACCEPTED MANUSCRIPT compared to S2 and S4. Enterococcus casseliflavus, previously isolated by Camu et al. (2007), is often associated with plant material, and was most abundant in S4. As suggested, the role of this

T

wide range of LAB species might be to promote the growth of yeasts through acidification, while

IP

suppressing the growth of food spoilage bacteria.

SC R

In accordance with published evidence (De Vuyst and Weckx, 2016; Papalexandratou et al., 2011c) Lactobacillus fermentum was the most abundant LAB in the samples analyzed. The only significant difference was found between sample S1 and S3, which may be explained by the well-

NU

defined fermentation undergone by S1, which promotes a defined bacterial community favoring L.

MA

fermentum as the most abundant LAB, versus the heap fermentation used for S3, which results in a more heterogeneous population.

Among the 25 most representative OTUs, two species belonging to acetic acid bacteria

TE

D

(AAB) have been detected: Acetobacter senegalensis and Acetobacter syzygii. A. senegalensis is able to perform a slow oxidation of ethanol into acetic acid and a fast oxidation of lactic acid into

CE P

acetoin (De Vuyst and Weckx, 2016), and has been previously isolated in Ghanaian cocoa beans during fermentation by Camu et al. (2008). Our data confirm the restricted biodiversity of AAB

AC

species during spontaneous cocoa bean fermentation. In accordance with Camu et al. (2008) we found a higher relative percentage of A. senegalensis in the Ghanaian S1 sample. Interestingly, the lowest abundance was found in sample S4, generated with a starter culture, suggesting that the initial inoculation inhibits the proliferation of extant species including the principal groups that ordinarily play a role during fermentation such as AAB. Acetobacter syzygii is able to perform a slow oxidation of ethanol into acetic acid and a slow oxidation of lactic acid into acetoin /acetic acid (De Vuyst and Weckx, 2016). It was the most abundant OTU, ranging from 38% (S4) to 56% (S1) of the total bacterial community; this difference could explain why, in the db-RDA model, the S1 sample from Ghana and the S4 (starter) culture from Ivory Coast-starter clustered separately (Fig.5). We detected the highest relative percentage of A. syzygii in the Ghanaian S1 sample.

ACCEPTED MANUSCRIPT This study shows some similar outcomes as well as differences with previously published data. First, previous studies noted the presence of L. plantarum/paraplantarum (Camu et al., 2007;

T

Hamdouche et al., 2014; Nielsen et al., 2007; Papalexandratou et al., 2011b, 2011c, 2013; Visintin

IP

at al., 2016), and the dominance of L. fermentum among the LAB (Camu et al., 2007, 2008;

SC R

Hamdouche et al., 2014; Nielsen et al., 2007; Papalexandratou et al., 2011c, 2013; Visintin et al., 2016). Our results confirm the preponderance of L. fermentum and the presence of L. plantarum/paraplantarum as one of the most abundant LAB. L. plantarum/paraplantarum is known

NU

to be present at the beginning of the fermentation process while L. fermentum is able to persist

MA

during further fermentations performed by AAB (De Vuyst and Weckx, 2016); for this reason it is present at a higher percentage in end-process cocoa beans. In accordance with the literature (Camu et al., 2007, 2008; Nielsen et al., 2007) we also found among LAB species, W. paramesenteroides

TE

D

and E. casseliflavus. Previous studies noted the presence of A. pasteurianus (Camu et al. 2007, 2008; De Vuyst and Weckx, 2016; Hamdouche et al., 2014; Illeghems et al., 2013; Meersman et al.

CE P

2013; Nielsen et al. 2007; Papalexandratou et al. 2011a, 2011b, 2011c, 2013; Visintin at al., 2016) and A. syzygii (De Melo Pereira et al., 2012; De Vuyst and Weckx, 2016; Hamdouche et al., 2014;

AC

Visintin at al., 2016) as most prevalent among the AAB. We detected a strong predominance of A. syzygii, while A. senegalensis, A. pasteurianus, and A. tropicalis were present, but less abundant (supplementary data Fig. S3). For the first time we found A. sicerae. It is interesting to note that almost all of our samples revealed abundant Acetobacter and in particular A. syzygii. This is probably due to the fact that our samples were end process cocoa beans, post sun drying, storage, and shipment. The dominance of Acetobacter species at the end of this process could be explained by the strong resistance of this genus to acidity and heat (Hamdouche et al., 2014). It is important to stress that HTS analyses carried out on all the samples did not show any specific trend until we performed analyses of distribution at species level. In fact when we performed the hierarchical clustering of samples based on Acetobacter species (supplementary data Fig. S3) all Ghanaian bean

ACCEPTED MANUSCRIPT samples clustered separately from the others thanks to a relative high abundance of A. senegalensis

IP

T

.

SC R

4. Conclusions

The present work was undertaken with the aim of assessing the total bacterial community in fermented cocoa bean samples originating from different geographic origins by means of HTS of

NU

16S rRNA amplicons. An average coverage of 97.39% was achieved for all the samples, with

MA

sequences correctly identified up to species level in 83.8% of cases. It is important to emphasize that this study has been developed in order to define a practical method with which to obtain the optimal results from HTS data, and that this approach could be used alongside culture-dependent

TE

D

methods in order to give the most complete picture of microbial ecology. Our results indicate that HTS approach has the ability to provide a comprehensive view of the cocoa

CE P

bean microbiota at the species level. The possibility to exploit this ability in order to discriminate cocoa beans samples on the basis of their geographical origin and/or biochemical transformations

AC

should be tested by further studies involving a high number of samples.

ACCEPTED MANUSCRIPT 5. Bibliography Alcántara-Hernández, R.J., Rodríguez-Álvarez, J.A., Valenzuela-Encinas, C., Gutiérrez

T

Miceli, F.A., Castañón-González, H., Marsch, R., Ayora-Talavera, T., Dendooven, L., 2010. The

IP

bacterial community in 'taberna' a traditional beverage of Southern Mexico. Letters in Applied

SC R

Microbiology 51, 558-563.

Bartram, A.K., Lynch, M.D., Stearns, J.C., Moreno-Hagelsieb, G., Neufeld, J.D., 2011. Generation of multi-million 16S rRNA gene libraries from complex microbial communities by

NU

assembling paired-end Illumina reads. Applied and Environmental Microbiology 77, 3846-3852.

MA

Bassi, D., Puglisi, E., Cocconcelli, P.S., 2015. Understanding the bacterial communities of hard cheese with blowing defect. Food Microbiology 52, 106-118. Berry, D., Mahfoudh, K.B., Wagner, M., Loy, A., 2011. Barcoded primers used in multiplex

TE

D

amplicon pyrosequencing bias amplification. Applied and Environmental Microbiology 77, 78467849.

CE P

Camu, N., De Winter, T., Verbrugghe, K., Cleenwerck, I., Vandamme, P., Takrama, J.S., Vancanneyt, M., De Vuyst, L., 2007. Dynamics and biodiversity of populations of lactic acid

AC

bacteria and acetic acid bacteria involved in spontaneous heap fermentation of cocoa beans in Ghana. Applied and Environmental Microbiology 73, 1809–1824. Camu, N., Gonzalez, A., De Winter, T., Van Schoor, A., De Bruyne, K., Vandamme, P., Takrama, J.S., Addo, S.K., De Vuyst, L., 2008. Influence of turning and environmental 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, 86–98. Claesson, M.J., O'Sullivan, O., Wang, Q., Nikkila, J., Marchesi, J.R., Smidt, H., De Vos, W.M., Ross, R.P., O'Toole, P.W., 2009. Comparative analysis of pyrosequencing and a phylogenetic microarray for exploring microbial community structures in the human distal intestine. PLoS ONE 4, e6669.

ACCEPTED MANUSCRIPT Chettri, R., Tamang, J.P., 2015. Bacillus species isolated from tungrymbai and bekang, naturally fermented soybean foods of India. International Journal of Food Microbiology 197, 72-76.

T

Cocolin, L., Ercolini, D., 2008. Molecular Techniques in the Microbial Ecology of

IP

Fermented Foods. Springer, New York.

SC R

De Melo Pereira, G.V., Magalhães-Guedes, K.T., Schwan, R.F., 2013. rDNA-based DGGE analysis and electron microscopic observation of cocoa beans to monitor microbial diversity and distribution during the fermentation process. Food Research International 53, 482–486.

NU

De Vuyst, L., Weckx, S., 2016. The cocoa bean fermentation process: from ecosystem

MA

analysis to starter culture development. Journal of Applied Microbiology. doi:10.1111/jam.13045 Delmont, T.O., Robe, P., Cecillon, S., Clark, I.M., Constancias F, Simonet, P., Hirsch, P.R., Vogel, T.M., 2011. Accessing the soil metagenome for studies of microbial diversity. Applied and

TE

D

Environmental Microbiology 77, 1315–1324. Dixon, P., 2003. VEGAN, a package of R functions for community ecology. Journal of

CE P

Vegetation Science 14, 927-930.

Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., Knight, R., 2011. UCHIME improves

AC

sensitivity and speed of chimera detection. Bioinformatics 27, 2194-2200. Escalante, A., Giles-Gómez, M., Hernández, G., Córdova-Aguilar, M.S., López-Munguía, A., Gosset, G., Bolívar, F., 2008. Analysis of bacterial community during the fermentation of pulque, a traditional Mexican alcoholic beverage, using a polyphasic approach. International Journal of Food Microbiology 124, 126-134. Essia Ngang, J.J., Yadang, G., Sado Kamdem, S.L., Kouebou, C.P., Youte Fanche, S.A., Tsochi Kougan, D.L., Tsoungui, A., Etoa, F.X., 2015. Antifungal properties of selected lactic acid bacteria and application in the biological control of ochratoxin A producing fungi during cocoa fermentation. Biocontrol Science and Technology 25, 245-259.

ACCEPTED MANUSCRIPT Fredricks, D.N., Smith, C., Meier, A., 2005. Comparison of Six DNA Extraction Methods for Recovery of Fungal DNA as Assessed by Quantitative PCR. Journal of Clinical Microbiology

T

43, 5122–5128.

IP

Good, I.J., 1953. The population frequencies of species and the estimation of population

SC R

parameters. Biometrika 40, 237-264.

Greppi, A., Ferrocino, I., La Storia, A., Rantsiou, K., Ercolini, D., Cocolin, L., 2015. Monitoring of the microbiota of fermented sausages by culture independent rRNA-based

NU

approaches. International Journal of Food Microbiology 212, 67-75.

MA

Gryson, N., Messens, K., Dewettinck, K., 2004. Evaluation and optimization of five different extraction methods for soy DNA in chocolate and biscuits Extraction of DNA as a first step in GMO analysis. Journal of the Science of Food and Agriculture 84,1357–1363.

TE

D

Gryson, N., Dewettinck, K., Messens, K., 2007. Influence of cocoa components on the PCR detection of soy lecithin DNA. European Food Research and Technology 226, 247–254.

CE P

Guehi, T.S., Zahouli, I.B., Ban-Koffi, L., Fae, M.A., and Nemlin, J.G., 2010. Performance of different fermentation methods and the effect of their duration on the quality of raw cocoa beans.

AC

International Journal of Food Science and Technology 45, 2508–2514. Haymes, K.M., Ibrahim, I.A., Mischke, S., Scott, D., Saunders, J.A., 2004. Rapid isolation of DNA from chocolate and date palm tree crops. Journal of Agriculture and Food Chemistry 52, 5456 -5467. Hamdouche, Y., Guehi, T., Durand, N., Kedjebo, K.B.D., Montet, D., Meile, J.C., 2014. Dynamics of microbial ecology during cocoa fermentation and drying: Towards the identification of molecular markers. Food Control 48, 117-122. Heip, C.H.R., Herman, P.M.J., Soetaert, K., 2001. Indices of diversity and Evenness. Oceanis 24, 61-87.

ACCEPTED MANUSCRIPT Illeghems, K., De Vuyst, L., Papalexandratou, Z., Weckx, S., 2012. Phylogenetic analysis of a spontaneous cocoa bean fermentation metagenome reveals new insights into its bacterial and

T

fungal community diversity. PLoS ONE 7, e38040.

IP

Kpikpi, E.N., Thorsen, L., Glover, R., Dzogbefia, V.P., Jespersen, L., 2014. Identification of

International Journal of Food Microbiology 180, 1-6.

SC R

Bacillus species occurring in Kantong, an acid fermented seed condiment produced in Ghana.

Li, C.Y., Chen, S.J., Cheng, C.Y., Chen, T.L., 2005. Production of Acinetobacter

NU

radioresistens lipase with repeated fed-batch culture. Biochemical Engineering Journal 25, 195-199. Lima, L.J.R., Almeida, M.H., Rob Nout, M.J. and Zwietering, M.H., 2011. Theobroma

MA

cacao L., “The food of the gods”: quality determinants of commercial cocoa beans with particular reference to the impact of fermentation. Critical Reviews in Food Science and Nutrition 55, 731-

TE

D

761.

Lima, L.J.R., Van der Velpen, V., Wolkers-Rooijackers, J., Kamphuis, H.J., Zwietering,

CE P

M.H., Rob Nout, M.J., 2012. Microbiota dynamics and diversity at different stages of industrial processing of cocoa beans into cocoa powder. Applied and Environmental Microbiology 78, 2904-

AC

2913.

Melnick, R.L., Suárez, C., Bailey, B.A., Backman, P.A., 2011. Isolation of endophytic endospore-forming bacteria from Theobroma cacao as potential biological control agents of cacao diseases. Biological Control 57, 236-245. Meersman, E., Steensels, J., Mathawan, M., Wittocx, P.J., Saels, V., Struyf, N., Bernaert, H., Vrancken, G., Verstrepen, K.J., 2013. Detailed Analysis of the Microbial Population in Malaysian Spontaneous Cocoa Pulp Fermentations Reveals a Core and Variable Microbiota. PLoS ONE 8, e81559. Moroni, A.V., Arendt, E.K., Bello, F.D., 2011. Biodiversity of lactic acid bacteria and yeasts in spontaneously-fermented buckwheat and teff sourdoughs. Food Microbiology 28, 497-502.

ACCEPTED MANUSCRIPT Nam, Y.D., Seo, M.J., Lim, S.I., Lee, S.Y., 2012. Genome sequence of Lysinibacillus boronitolerans F1182, isolated from a traditional Korean fermented soybean product. Journal of

T

Bacteriology 194, 5988-5988.

IP

Nielsen, D.S., Teniola, O.D., Ban-Koffi, L., Owusu, M., Andersson, T.S., Holzapfel W.H.,

SC R

2007. The microbiology of Ghanaian cocoa fermentations analysed using culture-dependent and culture-independent methods. International Journal of Food Microbiology 114, 168– 186. Papalexandratou, Z., Camu, N., Falony, G., De Vuyst, L., 2011a. Comparison of the

NU

bacterial species diversity of spontaneous cocoa bean fermentations carried out at selected farms in

MA

Ivory Coast and Brazil. Food Microbiology 28, 964–973.

Papalexandratou, Z., Falony, G., Romanens, E., Jimenez, J.C., Amores, F., Daniel, H.M., ., De Vuyst, L., 2011b. Species diversity, community dynamics, and metabolite kinetics of the

TE

D

microbiota associated with traditional Ecuadorian spontaneous cocoa bean fermentations. Applied and Environmental Microbiology 77, 7698–7714.

CE P

Papalexandratou, Z., Vrancken, G., De Bruyne, K., Vandamme, P., De Vuyst, L., 2011c. Spontaneous organic cocoa bean box fermentations in Brazil are characterized by a restricted

AC

species diversity of lactic acid bacteria and acetic acid bacteria. Food Microbiology 28, 1326-1338. Papalexandratou, Z., Lefeber, T., Bahrim, B., Lee, O., Daniel, H.M., De Vuyst, L., 2013. Hanseniaspora opuntiae, Saccharomyces cerevisiae, Lactobacillus fermentum, and Acetobacter pasteurianus predominate during well-performed Malaysian cocoa bean box fermentations, underlining the importance of these microbial species for a successful cocoa bean fermentation process. Food Microbiology 35, 73-85. Paulson, J.N., Pop, M., Bravo, H.C., 2011. Metastats: an improved statistical method for analysis of metagenomic data. Genome Biology 12, 1-27. Parente E., Cocolin L., De Filippis F., Zotta T., Ferrocino I., O’Sullivan O., Neviani E., De Angelis M., Cotter P. D., Ercolini D., 2015. FoodMicrobionet: A database for the visualisation and

ACCEPTED MANUSCRIPT exploration of food bacterial communities based on network analysis. International Journal of Food Microbiology 219, 28-37

T

Połka, J., Rebecchi, A., Pisacane, V., Morelli, L., Puglisi, E., 2015. Bacterial diversity in

IP

typical Italian salami at different ripening stages as revealed by high-throughput sequencing of 16S

SC R

rRNA amplicons. Food Microbiology 46, 342-356.

Porebski, S., Bailey, L.G., Baum, B.R., 1997. Modification of a CTAB DNA extraction protocol for plants containing high polysaccharide and polyphenol components. Plant Molecular

NU

Biology Reporter 15, 8–15.

MA

R Development Core Team, 2011. R: a Language and Environment for Statistical Computing, Reference Index Version 2.14.1. R Foundation for Statistical Computing, Vienna, Austria.

TE

D

Rebecchi, A., Pisacane, V., Miragoli, F., Polka, J., Falasconi, I., Morelli, L., Puglisi, E., 2015. High-throughput assessment of bacterial ecology in hog, cow and ovine casings used in

CE P

sausages production. International Journal of Food Microbiology 212, 49-59. Roh, S.W., Sung, Y., Nam, Y.D., Chang, H.-W., Kim, K.H., Yoon, J.H., Jeon, C.O., Oh, H.-

AC

M., Bae, J.W., 2008. Arthrobacter soli sp. nov., a novel bacterium isolated from wastewater reservoir sediment. Journal of Microbiology 46, 40-44. Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., Lesniewski RA, Oakley, B.B., Parks, D.H., Robinson, C.J., Sahl, J.W., Stres, B., Thallinger, G.G., Van Horn, D.J., Weber, C.F., 2009. Introducing mothur: opensource, platform-independent, community-supported software for describing and comparing microbial communities. Applied and Environmental Microbiology 75, 7537-7541. Schröder, J., Maus, I., Trost, E., Tauch, A., 2011. Complete genome sequence of Corynebacterium variabile DSM 44702 isolated from the surface of smear-ripened cheeses and insights into cheese ripening and flavor generation. BMC genomics 12, 545.

ACCEPTED MANUSCRIPT 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 44,205–221.

T

Spaniolas, S., Tsachaki, M., Bennett, M.J., Tucker, G., 2008. Evaluation of DNA extraction

IP

methods from green and roasted coffee beans. Food Control 19, 257–262.

SC R

Sukontasing, S., Tanasupawat, S., Moonmangmee, S., Lee, J.S., Suzuki, K.I. 2007. Enterococcus camelliae sp. nov., isolated from fermented tea leaves in Thailand. International Journal of Systematic and Evolutionary Microbiology 57, 2151-2154.

NU

Thanh, V.N., Mai, L.T., Tuan, D.A. 2008. Microbial diversity of traditional Vietnamese

MA

alcohol fermentation starters (banh men) as determined by PCR-mediated DGGE. International Journal of Food Microbiology 128, 268-273.

Tortajada, M., Martínez-Culebras, P.V., Navarro, V., Monzó, H., Ramón, D., 2009.

TE

D

Evaluation of DNA extraction methods for PCR detection of fungal and bacterial contamination in cocoa extracts. European Food Research and Technology 230,79–87.

CE P

Vasileiadis, S., Puglisi, E., Trevisan, M., Scheckel, K.G., Langdon, K., McLaughlin, M., Lombi, E., Donner, E., 2015. Changes in soil microbial communities and diversity in response to

AC

long-term silver exposure. FEMS Microbiology Ecology 91, fiv114. Visintin, S., Alessandria, V., Valente, A., Dolci, P., Cocolin, L., 2016. Molecular identification and physiological characterization of yeasts, lactic acid bacteria and acetic acid bacteria isolated from heap and box cocoa bean fermentations in West Africa. International Journal of Food Microbiology 216, 69-78. Wang, Q., Garrity, G.M., Tiedje, J.M., Cole, J.R., 2007. Naive Bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Applied and Environmental Microbiology 73, 5261-5267. Wei, C.L., Chao, S.H., Tsai, W.B., Lee, P.S., Tsau, N.H., Chen, J.S., Lai, W.L., Ching Yueh Tu, J., Tsai, Y.C., 2013. Analysis of bacterial diversity during the fermentation of inyu, a high-

ACCEPTED MANUSCRIPT temperature fermented soy sauce, using nested PCR-denaturing gradient gel electrophoresis and the

T

plate count method. Food Microbiology 33, 252-261.

IP

Acknowledgements

SC R

The authors are grateful to Sotirios Vasileiadis for providing data analyses tools and scripts used in

AC

CE P

TE

D

MA

NU

the processing of Illumina data.

ACCEPTED MANUSCRIPT Figures & Tables

T

Table 1. Description and labels used to identify the four processed beans samples analyzed, together with

IP

details about geographical origin and fermentation conditions. For each sample, different DNA extraction

SC R

methods were compared.

Table 2. Overview of the most abundant OTUs detected in the cocoa samples, with highlighted their

NU

ecological role during fermentation in cocoa or other foodstuffs.

MA

Fig. 1. Observed richness (a, b), and Simpson’s indexes (c, d), for the analyzed samples. Data are

D

classified according to DNA extraction method (a, c), or sample origin (b, d).

TE

Fig. 2. Hierarchical clustering of classified sequences using the average linkage algorithm at family

CE P

classification level for taxa contributing at least 5% to a single sample. Taxa beneath this threshold were added to the sequence group denoted “other”.

AC

Fig. 3a. Hierarchical clustering of classified sequences using the average linkage algorithm at genus classification level for taxa contributing at least 5% to a single sample. Taxa beneath this threshold were added to the sequence group denoted “other”. 3b.Metastats model to assess the effects of the sample on the relative abundances of genera comprising 95% of the bacterial diversity in the processed cocoa beans. Genera showing significant differences according to the false discovery rate correction are highlighted with letters. Fig. 4. db-RDA (distance-based Redundancy analyses) to test the hypotheses that geographic origins, as a significant effect, could alter the composition of the bacterial OTU community.

ACCEPTED MANUSCRIPT Fig. 5. Metastats model assessing the effects of the sample on the relative abundances of the 26 most abundant OTUs observed in the cocoa beans for genera contributing 95% of the total

T

measured bacterial diversity. OTUs showing significant differences according to the false discovery

IP

rate correction are highlighted with letters. For each OTU, the assigned species was indicated where

SC R

possible. Lactic acid bacteria (LAB) are highlighted with blue dots, and acetic acid bacteria (AAB)

AC

CE P

TE

D

MA

NU

with green dots.

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

S2

Ghana

Ivory

no

no

time

Wooden box

Wooden box

S3

Cameroon no

Ground/banana

Ivory

yes

AC

CE P

Coast Table 1

Wooden box

TE

S4

D

leaves

Main

Main

MA

Coast

SC R

S1

method

6 days

NU

label

Crops Fermentation Drying DNA

T

Starter Fermentation

IP

Sample Origin

Mid

Main

6 days

7 days

time

extraction method

4 days

MP-Bio/RBio/LifeTech

4 days

MP-Bio/RBio/LifeTech

5 days

MP-Bio/RBio/LifeTech

6 days

4 days

MP-Bio/RBio/LifeTech

ACCEPTED MANUSCRIPT

Bacillus safensis

Otu0015

Bacillus circulans

Otu0016 Otu0017

Lactobacillus plantarum/paraplantarum Corynebacterium variabile

Otu0019 Otu0021 Otu0023

Klebsiella pneumoniae Lactobacillus nagelii Lactobacillus vaginalis

development of flavor and textural properties vitamin B12 production lactic acid fermentation lactic acid fermentation

Otu0026

Pediococcus spp.

antifungal activity

Table 2

TE D

CE P

Recently isolated from fermented seeds of Hibiscus sabdariffa used as additive to produce the alkaline fermented seed condiments amylase activity; alkaline food fermentations and spoilage of foods acid/ethanol-tolerant LAB

AC

cocoa

Nam et al., 2012; Melnick et al., 2011

fermented drinks cocoa

Escalante et al., 2008; Li et al., 2005 Camu et al., 2007; Nielsen et al., 2008

cocoa, mango fermented tea leaves ('miang') Kantong, an acid fermented seed condiment produced in Ghana alcohol fermentation starters; fermented-soybean-foods cocoa

Papalexandratou et al., 2011 Sukontasing et al., 2007

smear-ripened cheeses

Schröder et al., 2011

fermented drinks cocoa, wine sorghum, rye and rice sourdoughs, buckwheat cocoa

Wei et al., 2012 Papalexandratou et al., 2011b Moroni et al., 2010

T

Otu0014

acetic fermentation lactic acid fermentation

Reference De Vuyst et al., 2016; Visintin et al., 2016 Lima et al., 2012 De Vuyst et al., 2016; Visintin et al., 2016 Camu et al., 2007; Nielsen et al., 2008

IP

Otu0012 Otu0013

Lysinobacillus boronitolerans Acinetobacter radioresistens Weissella paramesenteroides Acetobacter senegalensis Enterococcus camelliae

Food matrix cocoa cocoa cocoa cocoa

CR

Otu009 Otu0010 Otu0011

Ecological role acetic fermentation post processing contamination strictly heterofermentative LAB lactic acid fermentation from soil and leaves potential biological control agents for diseases that affect cacao Possible lipolytic activity hetero-fermentative LAB

US

Closest taxonomical hit Acetobater syzygii Bacillus clausii Lactobacillus fermentum Enetrococcus casseliflavus

MA N

OTU Otu001 Otu002 Otu003 Otu007

Kpikpi et al., 2014

Thanh et al., 2008; Chettri et al., 2015 De Vuyst et al., 2016; Visintin et al., 2016

Essia Ngang et al., 2014 Papalexandratou et al., 2011a

ACCEPTED MANUSCRIPT

Highlights:

IP

T

 High-throughput sequencing (HTS) assessed bacterial diversity in four processed samples of cocoa beans from different geographic origins.

SC R

 Specific pattern of acetic acid bacteria (AAB) and lactic acid bacteria (LAB).  HTS analyses of bacterial 16S DNA may represent a novel approach to define

AC

CE P

TE

D

MA

NU

marker strains for geographic traceability.

33