International Journal of Food Microbiology 205 (2015) 54–67

Contents lists available at ScienceDirect

International Journal of Food Microbiology journal homepage: www.elsevier.com/locate/ijfoodmicro

The effect of lactic acid bacteria on cocoa bean fermentation Van Thi Thuy Ho, Jian Zhao, Graham Fleet ⁎ Food Science and Technology, School of Chemical Engineering, The University of New South Wales, Sydney, New South Wales 2052, Australia

a r t i c l e

i n f o

Article history: Received 19 January 2015 Received in revised form 25 March 2015 Accepted 30 March 2015 Available online 3 April 2015 Keywords: Cocoa bean fermentation Yeasts Lactic acid bacteria Acetic acid bacteria Nisin Lysozyme

a b s t r a c t Cocoa beans (Theobroma cacao L.) are the raw material for chocolate production. Fermentation of cocoa pulp by microorganisms is crucial for developing chocolate flavor precursors. Yeasts conduct an alcoholic fermentation within the bean pulp that is essential for the production of good quality beans, giving typical chocolate characters. However, the roles of bacteria such as lactic acid bacteria and acetic acid bacteria in contributing to the quality of cocoa bean and chocolate are not fully understood. Using controlled laboratory fermentations, this study investigated the contribution of lactic acid bacteria to cocoa bean fermentation. Cocoa beans were fermented under conditions where the growth of lactic acid bacteria was restricted by the use of nisin and lysozyme. The resultant microbial ecology, chemistry and chocolate quality of beans from these fermentations were compared with those of indigenous (control) fermentations. The yeasts Hanseniaspora guilliermondii, Pichia kudriavzevii, Kluyveromyces marxianus and Saccharomyces cerevisiae, the lactic acid bacteria Lactobacillus plantarum, Lactobacillus pentosus and Lactobacillus fermentum and the acetic acid bacteria Acetobacter pasteurianus and Gluconobacter frateurii were the major species found in control fermentations. In fermentations with the presence of nisin and lysozyme, the same species of yeasts and acetic acid bacteria grew but the growth of lactic acid bacteria was prevented or restricted. These beans underwent characteristic alcoholic fermentation where the utilization of sugars and the production of ethanol, organic acids and volatile compounds in the bean pulp and nibs were similar for beans fermented in the presence of lactic acid bacteria. Lactic acid was produced during both fermentations but more so when lactic acid bacteria grew. Beans fermented in the presence or absence of lactic acid bacteria were fully fermented, had similar shell weights and gave acceptable chocolates with no differences in sensory rankings. It was concluded that lactic acid bacteria may not be necessary for successful cocoa fermentation. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Cocoa beans (Theobroma cacao L.) are the major raw material for chocolate production. Fermentation of the beans is essential for removing the pulp that envelops the beans and for developing the chemical precursors of chocolate flavor (Fowler, 2009; Thompson et al., 2013). The microbial ecology of cocoa bean fermentation is complex and involves the successional growth of various species of yeasts, lactic acid bacteria (LAB), acetic acid bacteria and, possibly, species of Bacillus, other bacteria and filamentous fungi (De Vuyst et al., 2010; Lima et al., 2011; Schwan et al., 2015; Thompson et al., 2013). Despite more than 100 years of research into the microbial ecology of this fermentation, the roles of the different microbial groups and species in contributing to the process, cocoa bean quality and chocolate quality are not fully understood. The results of our previous study (Ho et al., 2014) confirmed the role of yeasts in conducting an alcoholic fermentation of the bean pulp and further demonstrated that this activity was essential for the ⁎ Corresponding author. Tel.: +61 2 9385 5664; fax: +61 2 9385 5966. E-mail addresses: [email protected] (V.T.T. Ho), [email protected] (J. Zhao), g.fl[email protected] (G. Fleet).

http://dx.doi.org/10.1016/j.ijfoodmicro.2015.03.031 0168-1605/© 2015 Elsevier B.V. All rights reserved.

production of good quality beans that gave a typical chocolate character. These conclusions were derived by comparing the quality of cocoa beans obtained from control fermentations with beans from fermentations where yeast contribution to the process was inhibited by natamycin. Fermentations conducted under such conditions gave normal growth of LAB and acetic acid bacteria, but the final beans tasted acidic and lacked the characteristic chocolate flavor. LAB are consistently associated with cocoa bean fermentations and generally grow to populations of 107–108 CFU/g during the first 36–48 h of the process (De Vuyst et al., 2010; Lima et al., 2011; Roelofsen, 1958; Schwan et al., 2015; Thompson et al., 2013). While a diversity of LAB species has been isolated from these ecosystems, Lactobacillus plantarum and Lactobacillus fermentum are most frequently dominant (Ardhana and Fleet, 2003; Camu et al., 2007; Kostinek et al., 2008; Lefeber et al., 2011a; Nielsen et al., 2007; Papalexandratou et al., 2011a,2011b,2011c). De Vuyst et al. (2010) have reviewed the ecology, metabolic activities and potential roles of LAB during cocoa bean fermentations. LAB are thought to conduct three major activities during bean fermentation: (i) they ferment pulp sugars, principally glucose and fructose, to mainly lactic acid and lesser amounts of ethanol and acetic acid, (ii) they utilize citric acid within the pulp to produce mainly

V.T.T. Ho et al. / International Journal of Food Microbiology 205 (2015) 54–67

lactic acid, acetic acid, acetaldehyde, diacetyl, acetoin and 2,3butanediol and (iii) some species may reduce pulp fructose to mannitol (Camu et al., 2007, 2008a; De Vuyst et al., 2010; Lefeber et al., 2011a). Laboratory fermentations using simulated cocoa pulp media and pure culture isolates of LAB from cocoa fermentations have confirmed such metabolic behavior (Adler et al., 2013; Lefeber et al., 2010, 2011b; Moens et al., 2014; Pereira et al., 2012). These reactions change the composition of the pulp, but the impact of these changes on cocoa bean and chocolate quality is not well defined. The excessive production of lactic acid by these bacteria, its migration into the beans, and its non-volatility may contribute to the high acidity of cocoa beans from some geographical regions and poorer chocolate quality (reviewed in Jinap, 1994). However, studies that linked the sensory quality of cocoa beans with their content of organic acids were not conclusive in establishing this impact and suggested that acetic acid may have a greater contribution to bean acidity than lactic acid (Holm et al., 1993; Jinap and Dimick, 1990; Jinap et al., 1995). Acetic acid production in cocoa bean fermentation is largely due to acetic acid bacteria which oxidize the ethanol produced by yeasts (De Vuyst et al., 2010; Lima et al., 2011; Schwan and Wheals, 2004). However, this route of acetic acid production may need reconsideration because Ho et al. (2014) showed that typical levels of acetic acid were produced in fermentations where yeast growth and ethanol production were inhibited. Other studies suggested that LAB could be an important source of acetic acid in cocoa bean fermentations (Roelofsen, 1958) and this role needs further consideration in light of the potential of these bacteria to produce acetic acid by a diversity of metabolic mechanisms (Adler et al., 2013; De Vuyst et al., 2010; Moens et al., 2014). Through their utilization of citric acid and production of lactic acid and acetic acids in the pulp, LAB are considered to be significant in modulating the pH of the bean (De Vuyst et al., 2010). The internal pH of the bean affects the activity of various endogenous enzymes (e.g., proteases, invertase, glycosidases, polyphenol oxidases and others) that impact on chocolate quality (Biehl et al., 1985; Hansen et al., 1998; Voigt and Lieberei, 2015). Of these three acids, only acetic acid is largely taken into the bean, with smaller amounts of lactic acid and almost no citric acid moving into the bean (Camu et al., 2008b; De Vuyst et al., 2010). Lactic acid and mannitol produced by LAB may encourage the growth of acetic acid bacteria and, in this way, they could indirectly affect acetic acid production and bean pH (Camu et al., 2007; De Vuyst et al., 2010). Through various biochemical mechanisms, LAB can produce a diversity of volatile and non-volatile metabolites (De Vuyst et al., 2010) that could directly impact cocoa flavor, but such a role has not been studied in detail and needs investigation. In recognizing the importance of LAB in cocoa fermentations, several authors have used selected strains of L. plantarum and L. fermentum as starter cultures in combination with or without various species of acetic acid bacteria and yeasts to better control and improve the fermentation (Kresnowati et al., 2013; Lefeber et al., 2012; Pereira et al., 2012; Schwan, 1998). However, none of these studies has provided a clearer picture of the specific role of LAB in fermentation. The objectives of this paper are to investigate the role of LAB in cocoa bean fermentations and their contribution to chocolate quality by conducting laboratory scale fermentations under conditions where the growth of LAB did not occur or was significantly restricted. This was achieved by adding nisin and nisin plus lysozyme combinations to the fermentation. Nisin and lysozyme are approved food additives that are particularly effective against Gram positive bacteria, including LAB, but have little effect on Gram negative bacteria (Johnson and Larson, 2005; Thomas and Delves-Broughton, 2005). 2. Materials and methods 2.1. Cocoa bean fermentation Cocoa pods (Trinitario variety) were harvested from plantations in North Queensland, Australia and transported to UNSW Australia,

55

Sydney where they were stored at 20–25 °C and used within 7–10 days from harvest. The pods were then split for manual removal of the beans. The mass of beans was mixed on a tray and deliberately brought into contact with the outer surfaces of extracted pods to provide a source of natural microbial inoculum. The bean mass was then divided into two ×5 kg batches which were transferred into two plastic boxes (17 × 17 × 20 cm) for fermentation. To restrict bacterial growth, a solution of Nisaplin (Danisco, Denmark) containing 2.5% of pure nisin or a mixture of nisin and lysozyme (Delvozyme, DSM, The Netherlands) was sprayed onto the beans in one box and mixed to give uniform distribution of nisin and lysozyme. The cocoa beans in the other box were not treated with nisin or lysozyme but similarly mixed. The boxes of beans were then incubated for fermentation which developed spontaneously due to growth of the indigenous microbiota. For fermentation, the boxes were covered with lids and incubated at 25 °C (0–12 h), 30 °C (12–24 h), 35 °C (24–36 h), 40 °C (36–48 h), 45 °C (48–72 h) and 48 °C (72–144 h) to simulate the temperature evolution that occurs during most commercial cocoa fermentations (Lima et al., 2011; Schwan and Wheals, 2004). The fermenting beans were thoroughly mixed every 48 h. The fermentations were stopped at day 6, when the beans were removed from their boxes and dried at 30 °C and relative humidity 70% for 5 days. Samples of beans (100 g total, from locations throughout the fermenting mass) were taken daily for microbiological and chemical analyses. Samples for microbiological analysis were used immediately while those for chemical analysis were stored at − 20 °C until examined. More details on fermentation procedures are described by Ho et al. (2014). The fermentations were conducted using (i) cocoa beans harvested in March 2011 and (ii) beans harvested from another plantation in May 2012. For the beans harvested in May 2012, two independent fermentations were conducted, one containing 2500 ppm Nisaplin plus lysozyme and one containing 3500 ppm Nisaplin plus lysozyme. The two 2012 fermentations gave similar microbiological, pH, chemical and sensory results and, in some places, only data for the 2500 ppm Nisaplin fermentation are presented. In addition, several preliminary fermentations were conducted to optimize the conditions for fermentation and drying, chocolate production and sensory assessment, and to select the appropriate concentration of Nisaplin.

2.2. Microbiological analyses Specific details of methods for the enumeration, isolation and identification of microorganisms from the cocoa fermentations are given in Ho et al. (2014). Cocoa beans (25 g) were aseptically mixed with 225 ml of 0.1% peptone water in a Stomacher bag and manually shaken for 5 min to give a uniform suspension of the pulp material. One ml of the suspension was serially diluted in 0.1% peptone water and 0.1 ml samples from each of three consecutive dilutions were spread inoculated onto duplicate plates of different agar media. The enumeration of yeasts was done on plates of Malt Extract Agar (MEA) (Oxoid) containing 100 mg/l of oxytetracycline and Dichloran Rose Bengal Chloramphenicol Agar (DRBC) with incubation at 25 °C for 3–4 days. LAB were enumerated on de Man Rogosa Sharpe (MRS) Agar (Oxoid) containing 100 mg/l of cycloheximide at 30 °C for 3–4 days. Acetic acid bacteria were enumerated on Wallerstein Laboratories Nutrient Agar (WLNA) (Oxoid) and Glucose-Yeast Extract Agar (GYEA) (5% glucose, 0.5% yeast extract and 1.5% agar) containing 100 mg/l of cycloheximide at 30 °C for 4–5 days. After incubation, counts of the yeast and bacterial groups were determined as well as those of individual yeast and bacterial species, based on observations of their colonial and cellular morphologies and subsequent identification. At least three representatives of each colony type were isolated from each sampling time, purified by re-streaking on their appropriate media, and used for identification. Population data reported are the means of duplicate analyses.

56

V.T.T. Ho et al. / International Journal of Food Microbiology 205 (2015) 54–67

Isolates of yeasts and bacteria were identified by a combination of phenotypic and molecular methods. Briefly, the Polymerase Chain Reaction (PCR) was performed to amplify the 5.8S-Internally Transcribed Spacer rRNA gene region of yeast isolates and the 16S rRNA gene of bacterial isolates. DNA amplifications were done with the primers ITS1 and ITS4 for yeasts, 27F and 1495R for LAB and 16Sd and 16Sr for acetic acid bacteria. Following amplifications, PCR products were used for identification of yeasts and bacteria to species by a combination of restriction fragment length polymorphism (RFLP) and sequence analyses. Growth at 37 °C and ascospore production were used as additional tests to differentiate the yeast Hanseniaspora guilliermondii from other species (Cadez and Smith, 2011). The procedures and conditions of DNA extraction, PCR, RFLP and sequence analyses, including the use of reference cultures, were done as described by Ho et al. (2014). 2.3. Chemical analyses The measurement of pH and analyses of sugar, ethanol, glycerol, mannitol and organic acid concentrations were determined on cocoa pulp and nib fractions according to the methods described by Ho et al. (2014). pH was determined with a pH meter on water extracts of pulp and nib fractions obtained from samples of 10 randomly selected whole cocoa beans. The concentrations of sugars, ethanol, glycerol, mannitol and organic acids were analyzed using a high performance liquid chromatography (HPLC) system of Shimadzu with refractive index and photodiode array detectors connected to a Rezex ion-exclusion ROA column (Phenomenex, Torrance, CA, USA). The concentrations of individual analytes were determined by comparison with curves constructed from standard solutions of each compound. Analyses of samples were done in duplicate and average values are reported. We have expressed our data for sugar and ethanol concentrations on a wet weight basis and data for organic acids on a dry weight basis to facilitate comparisons with the literature. Volatile compounds in separated fractions of cocoa pulp and nibs were extracted by head space–solid phase microextraction (HS–SPME) using a 50/30 μm divinylbenzene/carboxene/polydimethylsiloxane fiber (Supelco, Bellefonte, PA, USA) and analyzed by an Agilent 5975 gas chromatography–mass spectrometer (GC–MS) equipped with a Hewlett Packard capillary column (HP-5MS, 30 m length × 0.25 mm ID × 0.25 μm thickness). The conditions of extraction and GC–MS as well as the method for identification of volatile compounds were the same as described by Ho et al. (2014). The volatile compounds in each sample were extracted and analyzed in duplicate and the results are expressed as an average. 2.4. Quality evaluation of cocoa beans The cut test and shell content of dried cocoa beans and sensory evaluation of chocolates were performed as described by Ho et al. (2014) to compare the quality of cocoa beans obtained from fermentations in the presence and absence of bacterial inhibitors. Briefly, the beans (30 for each sample) were cut lengthwise from the middle, and each half of the beans was visually inspected and classified based on their color and defects. The shell content (20 beans for each sample) was calculated as the percentage of shell weight relative to the whole bean weight. Chocolates were prepared by roasting and grinding dried, fermented beans (50%) into cocoa liquor before adding cocoa butter (20%) and icing sugar (30%), followed by conching for at least 6 h and tempering. Chocolates were evaluated for sensory quality using the liking test which asked panelists (30 for each sample) to evaluate color, flavor and overall acceptability of two coded samples of chocolate by giving scores in a seven-point hedonic scale. One of the samples was prepared from beans fermented in the absence of bacterial inhibitors and the other was from beans fermented with bacterial inhibitors (nisin/nisin + lysozyme).

2.5. Statistical analysis One-way single factor analysis of variance and t-test were used to determine significant differences between means using Microsoft Excel. Significant differences in the concentrations of sugars, organic acids, volatile compounds, shell content and sensory evaluation of chocolates were considered when p b 0.05. 3. Results 3.1. Preliminary experiments Preliminary experiments using cocoa beans with Nisaplin added to give final concentrations of 100, 200, 500 and 1000 ppm indicated that a minimum of 500 ppm was needed to inhibit the growth of LAB. Fermentation of one batch of cocoa beans (March 2011) with the addition of 500 ppm of Nisaplin gave no detectable growth of LAB. However, fermentation of another batch of beans (harvested on another occasion) with 500 ppm Nisaplin did not give satisfactory inhibition of LAB. Therefore, isolates of LAB from the cocoa beans were tested for their sensitivity to nisin and also lysozyme in pure culture agar diffusion assays as described by Pongtharangkul and Demirci (2004). These assays revealed diversity in the responses of these isolates to nisin. It was concluded that a minimum of 2500 ppm Nisaplin was needed to restrict the growth of cocoa isolates of LAB and, consequently, further two fermentations (May 2012) were conducted, one containing 2500 ppm Nisaplin + 1000 ppm Delvozyme and another containing 3500 ppm Nisaplin + 1000 ppm Delvozyme. 3.2. Microbial ecology of cocoa bean fermentations with added nisin and lysozyme Fig. 1 shows the growth of yeasts and bacteria during the fermentation of cocoa beans in the absence and presence of 500 ppm Nisaplin (beans harvested in March 2011). For fermentations not treated with nisin, yeasts (approximately 103–104 CFU/g) and bacteria (102 CFU/g) were present at the commencement of fermentation and grew to populations of 107–108 CFU/g within 48–72 h before declining. LAB were first detected at 24 h, after which they grew to populations of 106 CFU/g and then decreased to about 103 CFU/g at the end of fermentation (Fig. 1a). At 72 h, LAB represented approximately 17% of the total bacterial population and at the completion of fermentation (144 h), they were 100%. The growth of LAB during fermentation of these beans was inhibited by the addition of nisin, but such nisin did not affect the total populations of yeasts and other bacteria (Fig. 1b). Also, nisin did not affect the growth profiles of individual species of yeasts (Fig. 1c, d) and acetic acid bacteria (Fig. 1e, f). The main yeast species were H. guilliermondii, Pichia kudriavzevii and Kluyveromyces marxianus which grew in a similar succession for fermentations in the absence and presence of nisin (Fig. 1c, d). The main species of acetic acid bacteria were Gluconobacter frateurii and Acetobacter pasteurianus which grew similarly in both fermentations (Fig. 1e, f). L. plantarum was the only species of LAB found in the control (no nisin) fermentation and grew to 106 CFU/ml before decreasing to 104–105 CFU/g (Fig. 1e). This species was not detected during fermentation in the presence of nisin (Fig. 1f). Fig. 2 shows the growth of yeasts and bacteria during fermentation of cocoa beans in the absence and presence of 2500 or 3500 ppm Nisaplin + lysozyme (beans harvested in May 2012). The total population of yeasts (maximum about 108 CFU/g) as well as growth of individual species throughout the fermentation were not affected by the presence of nisin + lysozyme (Fig. 2a, b, c, d, e, f). The main species were H. guilliermondii, P. guilliermondii, Saccharomyces cerevisiae and Pichia kudriavzevii which grew in similar succession for both fermentations. For acetic acid bacteria, only G. frateurii was found and it gave weak growth to 104 CFU/g during the first 48 h, after which it was not

V.T.T. Ho et al. / International Journal of Food Microbiology 205 (2015) 54–67

57

Fig. 1. Changes in the populations of total yeasts (●), total bacteria (□) and total lactic acid bacteria (▲) (a, b) and the growth of yeast species (c, d) and bacterial species (e, f) during untreated (a, c, e) and Nisaplin 500 ppm treated (b, d, f) fermentations of cocoa beans harvested in March 2011.

detected. This same pattern of growth occurred for fermentations in the absence and presence of nisin + lysozyme (Fig. 2g, h, j). In the absence of nisin + lysozyme, LAB grew to maximum populations of 107– 108 CFU/g. Three species were found: Lactobacillus pentosus and L. plantarum which grew to 106–107 CFU/g during the first 48 h before dying off to non-detectable levels (b100 CFU/g); and L. fermentum which grew to 107 CFU/g after 96 h and then declined to about 103 CFU/g by the end of fermentation (Fig. 2g). For fermentations in the presence of nisin + lysozyme, no growth of L. pentosus and L. fermentum was detected (b100 CFU/g) and only weak growth (105 CFU/g) of L. plantarum was detected at only one sampling time, 48 h (Fig. 2h, j). 3.3. Chemical changes during cocoa bean fermentation 3.3.1. pH changes in cocoa pulp and nibs The pH of unfermented pulp from beans harvested in 2011 was 3.8 (Fig. 3). During fermentation in the absence or presence of 500 ppm Nisaplin, the pH of the pulp decreased to about 3.0 in the first 48 h, after which it increased to 3.6–3.7 (Fig. 3a). The nibs had an initial pH of 6.8 which decreased to 5.4–5.5 for fermentation in the absence or presence of nisin (Fig. 3c). Similar trends were found for the beans fermented in 2012 without or with 2500 ppm Nisaplin + 1000 ppm Delvozyme, but the initial pH of the pulp was 3.9–4.0 and the final pH was 4.6 for both fermentations (Fig. 3b). The nibs had an initial pH of 6.8 that decreased to 5.8–6.0 for fermentation in the absence or presence of nisin + lysozyme (Fig. 3d). Similar data (not shown) were

obtained for the 2012 beans fermented in the presence of 3500 ppm Nisaplin + 1000 ppm Delvozyme. Although final nib pH values were slightly higher for fermentations in the presence of nisin, the differences were not significant (p N 0.05). It was concluded that pulp or nib pH changes during bean fermentation were not significantly impacted when growth of LAB was prevented or restricted. 3.3.2. Changes in the concentration of sugars, ethanol, glycerol, mannitol and organic acids Fig. 4 shows the changes in concentrations of sugars and ethanol in the pulp and nib fractions during fermentation of cocoa beans harvested in 2011 and 2012. For the 2012 beans, only data for the 2500 ppm Nisaplin + 1000 ppm Delvozyme are presented. Data for fermentations with 3500 ppm Nisaplin + 1000 ppm Delvozyme were similar and are not presented. In unfermented cocoa beans, the main sugars in the pulp were glucose (approximately 50 mg/g, 76 mg/g) and fructose (70 mg/g, 88 mg/g) (Fig. 4a, b, c, d), while those in the nibs were low levels of sucrose (17 mg/g, 16 mg/g), glucose (0.5 mg/g, 3 mg/g) and fructose (1 mg/g, 3.5 mg/g) (Fig. 4e, f, g, h) with the values in parenthesis corresponding to the amounts in beans obtained in 2011 and 2012, respectively. Pulp fructose and glucose were metabolized during fermentation (Fig. 4a, b, c, d). For the 2011 beans, fructose was completely utilized, but this was not the case for glucose. For the 2012 beans, both fructose and glucose were completely utilized. For both 2011 and 2012 beans, the profile of sugar utilization in the pulp was similar for fermentations in the presence or absence of nisin or nisin + lysozyme.

58

V.T.T. Ho et al. / International Journal of Food Microbiology 205 (2015) 54–67

Fig. 2. Changes in the populations of total yeasts (●), total bacteria (□) and total lactic acid bacteria (▲) (a, b, c) and the growth of yeast species (d, e, f) and bacterial species (g, h, j) during untreated (a, d, g), Nisaplin 2500 ppm + Delvozyme 1000 ppm treated (b, e, h) and Nisaplin 3500 ppm + Delvozyme 1000 ppm treated (c, f, j) fermentations of cocoa beans harvested in May 2012.

For the nibs, the initial sucrose concentration dropped below 1 mg/g in all fermentations and the low initial levels of glucose and fructose increased to 3–5 mg/g. These changes were not affected when nisin or nisin + lysozyme was added to the fermentation to inhibit the growth of LAB (Fig. 4e, f, g, h). Ethanol was not detected in the pulp or nibs of unfermented cocoa beans (Fig. 4). For fermentation of the 2011 beans in the absence or presence of nisin, low concentrations of ethanol were detected in the pulp (0.4–0.5 mg/g) and in the nibs (0.2 mg/g) at 24 h. By 48 h, ethanol had increased to 6–8 mg/g in the pulp (Fig. 4a, c) and 5–7 mg/g in the nibs (Fig. 4e, g). By the end of the process, its concentration had decreased to 2–3 mg/g in both the pulp and nibs. There were no statistically significant differences (p N 0.05) between ethanol concentrations for fermentations in the absence or presence of nisin. Similar conclusions were obtained for ethanol production during fermentations of the 2012 beans, but much higher concentrations of ethanol (e.g., 23– 25 mg/g in the pulp; 13–15 mg/g in the nibs) were found (Fig. 4b, d, f, h). Glycerol, at levels up to 7.0 mg/g, was produced in the pulp during fermentation and small amounts (1.5 mg/g) diffused into the nibs (data not shown). Glycerol data were similar for fermentations in the presence and absence of nisin or nisin + lysozyme. Mannitol was not found in the pulp or nibs of beans fermented in 2011. For the 2012 beans, a low level (1.3 mg/g) was produced in the pulp fraction during fermentation without nisin + lysozyme but was not found in the fermentation treated with these bacterial inhibitors

(data not shown). Its presence correlated with the growth of L. fermentum which occurred in the 2012 control fermentations, but was inhibited by nisin + lysozyme. Mannitol was not detected in any nib fractions. Fig. 5 shows the changes in concentrations of organic acids during cocoa fermentations. No lactic acid was detected in the pulp or nibs of unfermented cocoa beans. In the 2011 fermentations, lactic acid was first detected in the pulp at 48 h and increased to about 26 mg/g for beans fermented in the absence of nisin (Fig. 5a). Notably, lactic acid (10 mg/g) was also produced during fermentation of these beans in the presence of nisin (Fig. 5c), where no growth of LAB was detected (see Fig. 1f). Lactic acid produced in the pulp was taken up by the nibs (Fig. 5e, g). For the 2012 fermentations in the absence of nisin + lysozyme, lactic acid was produced in the pulp but at a lower level (approx. 9 mg/g) than that for the 2011 beans (Fig. 5b), despite a greater association of LAB with this fermentation (see Fig. 2g). This lactic acid also transferred to the nibs (Fig. 5f). For fermentation of the 2012 beans in the presence of nisin + lysozyme, no lactic acid was detected in the pulp or nibs (Fig. 5d, h) and this correlated with the restricted growth of LAB in these fermentations (see Fig. 2h, j). Acetic acid was not detected in unfermented pulp or nibs. For the 2011 beans, acetic acid was first detected in the pulp at 48 h and increased to maximum concentrations of about 15 mg/g in the untreated and nisin treated fermentations before declining (Fig. 5a, c). In the nibs, acetic acid was detected at maximum levels of about 20 mg/g by

V.T.T. Ho et al. / International Journal of Food Microbiology 205 (2015) 54–67

59

Fig. 3. Changes in the pH of the pulp (a, b) and nibs (c, d) during untreated (●) and nisin or nisin + lysozyme treated (■) fermentations of cocoa beans harvested in March 2011 (a, c) and in May 2012 (b, d). Data are the means of duplicate analyses ± 0.05.

96–120 h in fermentations with and without nisin before decreasing to about 10 mg/g (Fig. 5e, g). Although lower concentrations of acetic acid were found in the pulp (2–3 mg/g) and nibs (2–3 mg/g) of the 2012 beans, the profiles of production were similar (p N 0.05) for fermentations in the absence and presence of nisin + lysozyme (Fig. 5b, d, f, h). Citric acid was detected in both the pulp (30–42 mg/g) and nibs (8–17 mg/g) of unfermented beans, with greater amounts in the beans harvested in 2011. Concentrations in the pulp decreased slightly during the first 24 h of fermentation, after which increases occurred (Fig. 5a, b, c, d). For the nibs, citric acid concentrations decreased during fermentation, this being more evident for the 2011 trials, where an increase was observed towards the end of the process (Fig. 5e, f, g, h). The kinetics of the changes in pulp and nib citric acid levels were similar for the untreated and nisin or nisin + lysozyme treated beans (p N 0.05), suggesting that growth of LAB did not significantly affect citric acid behavior. In unfermented beans, succinic and malic acids were present in the pulp and nibs while oxalic acid was only detected in the nibs. Changes in the concentrations of succinic, malic and oxalic acids in the pulp and nibs throughout fermentation were similar for fermentations in the absence and presence of nisin or nisin + lysozyme. For 2011 and 2012 beans, succinic acid occurred in both the pulp (5–8 mg/g) and nibs (9–13 mg/g) of unfermented beans and its concentration in both fractions increased during fermentation (Fig. 5). The levels of malic acid decreased in the pulp (from 8–11 mg/g to 1.5–3 mg/g) (Fig. 5a, b, c, d) but remained relatively unchanged in the nibs (1–1.8 mg/g) (Fig. 5e, f, g, h) during all fermentations. The small amounts of oxalic acid in the nibs (3–7 mg/g) decreased slightly during fermentation (Fig. 5e, f, g, h). 3.3.3. Production of volatile compounds during fermentation The SPME–GC/MS method for volatile detection and measurement enabled the identification of at least 40 compounds in most samples. These compounds were grouped into aldehydes, ketones, esters, alcohols and pyrazines. Fig. 6 shows the concentrations of four groups of volatile compounds (aldehydes, ketones, esters and alcohols) in the cocoa pulp and nibs

during fermentation of the 2012 beans in the absence and presence of nisin + lysozyme. Similar data and conclusions were obtained for the beans fermented in 2011 but were analyzed only for the fermentation times 0, 72 and 144 h (Fig. S1). 3.3.3.1. Higher alcohols. The pulp of unfermented beans contained little higher alcohols but their concentration increased about 12 fold during fermentations in the absence or presence of nisin + lysozyme (Fig. 6a). The main higher alcohols found in fermenting pulps were 2-methyl-1-butanol (approx. 76%, and 83% of total higher alcohols), phenylethyl alcohol (approx. 13.6% and 9% of total) and isoamyl alcohol (6.5% and 5%). The data in parenthesis are the relative amounts at the end of fermentation in the absence and presence of nisin + lysozyme, respectively. Low amounts of higher alcohols were detected in the nibs before fermentation. These increased about 5 fold throughout fermentation before decreasing during drying and roasting (Fig. 6b). Phenylethyl alcohol (69–70% of total higher alcohols), isoamyl alcohol (16–17%) and 2-methyl-1-butanol (8–9%) were the three main alcohols found in the nibs with the relative values given being the amounts after roasting. There were no statistically significant differences between the concentrations of these alcohols in the pulps or nibs for fermentations in the presence or absence of nisin + lysozyme. 3.3.3.2. Esters. Esters were not detectable in either pulp or nibs of the unfermented beans. This concentration increased about 45–50 fold in the pulp during fermentations in the presence or absence of nisin + lysozyme. Five main esters detected in fermenting pulps were ethyl phenylacetate (23%, 28%), ethyl acetate (22%, 18%), ethyl succinate (16%, 15%), isoamyl acetate (10%, 8.5%) and ethyl hexanoate (8%, 12%) with the relative values given being those at the end of fermentations in the absence and presence of nisin + lysozyme, respectively (Fig. 6c). Total ester concentrations in the nibs increased about 12 fold during fermentation and drying but decreased after roasting (Fig. 6d). The main esters detected in roasted nibs of both fermentations were isoamyl acetate (21–22%), ethyl hexanoate (15–17%), ethyl acetate (13–16%) and ethyl octanoate (10–13%) (Fig. 6d). The ester profiles of

60

V.T.T. Ho et al. / International Journal of Food Microbiology 205 (2015) 54–67

Fig. 4. Changes in the concentrations of sucrose (▲), glucose (●), fructose (■), and ethanol (♦), in the pulp (a, b, c, d) and nibs (e, f, g, h) during fermentation in the absence (a, b, e, f) and presence (c, d, g, h) of nisin or nisin + lysozyme of beans harvested in 2011 (a, c, e, g) and 2012 (b, d, f, h). Data are the means of duplicate analyses ± 3 mg/g.

both the pulp and nibs were statistically similar (p N 0.05) for fermentations in the absence or presence of nisin + lysozyme. 3.3.3.3. Aldehydes. Aldehydes were detected in the pulp and nib fractions of unfermented beans. Their levels increased about 9 fold in the pulp but less than 2 fold in the nibs during fermentations in the absence or presence of nisin + lysozyme (Fig. 6e, f). No differences (p N 0.05) in the concentrations of these aldehydes in the pulps or nibs were found between fermentations in the absence and presence of nisin + lysozyme. The two main aldehydes in the pulp were benzaldehyde (69–73%) and phenyl

acetaldehyde (19–20.5%), with the values in brackets corresponding to the concentrations found at the end of fermentations. Three main aldehydes in the nibs were benzaldehyde, phenyl acetaldehyde and hexanal and their relative amounts in roasted nibs of both fermentations were in the range of 67–69%, 16–17% and 16–18%, respectively. 3.3.3.4. Ketones. Considerable amounts of ketones were detected in unfermented pulp and nibs. There were no differences (p N 0.05) in the concentrations of ketones in the pulps or nibs between fermentations in the absence and presence of nisin + lysozyme. In the pulp, their

V.T.T. Ho et al. / International Journal of Food Microbiology 205 (2015) 54–67

61

Fig. 5. Changes in the concentrations of citric acid (■), succinic acid (▲), lactic acid (○), acetic acid (●), malic acid (□) and oxalic acid (△) in the pulp (a, b, c, d) and nibs (e, f, g, h) during untreated (a, b, e, f) and nisin or nisin + lysozyme treated (c, d, g, h) fermentations of beans harvested in 2011 (a, c, e, g) and 2012 (b, d, f, h). Data are the means of duplicate analyses ± 1 mg/g.

total concentration decreased about 2 fold during fermentation (Fig. 6g). Four main ketones detected in the pulp were acetophenone (62%, 64%), 2-heptanone (13%, 12%), 2-nonanone (7%, 9%) and 2-pentanone (9%, 3%). The values in parenthesis are their relative concentrations at the end of fermentations in the absence and presence of nisin + lysozyme, respectively. These compounds were also the main ketones in unfermented nibs and their total concentrations were relatively unchanged during fermentation but decreased throughout drying and roasting (Fig. 6h). The relative amounts of 2-heptanone, acetophenone, 2-pentanone and 2-nonanone in roasted nibs of

fermentations in the absence and presence of nisin + lysozyme were in the range of 30–33%, 29–30%, 21–23% and 15–18%, respectively. 3.3.3.5. Pyrazines. Pyrazines were not detected in unfermented and fermenting pulp or nibs but were formed during roasting of cocoa nibs. Seven pyrazine compounds were detected in the 2011 beans, with 2,3,5,6-tetramethylpyrazine, 2,3,5-trimethylpyrazine, 2,5-dimethyl-3-ethylpyrazine and 2,5-dimethylpyrazine being the predominant ones (Fig. 7a). For the 2012 beans, 2,3-dimethylpyrazine and 2,3,5,6tetramethylpyrazine were dominant among five pyrazine compounds

62

V.T.T. Ho et al. / International Journal of Food Microbiology 205 (2015) 54–67

Fig. 6. Total concentrations of higher alcohols (a, b), esters (c, d), aldehydes (e, f) and ketones (g, h) produced in the cocoa pulp (a, c, e, g) and nibs (b, d, f, h) during fermentation, drying (D) and roasting (R) in untreated ( ) and nisin treated ( ) fermentations of beans harvested in May 2012.

detected in this trial (Fig. 7b). Pyrazine production was not significantly different for beans fermented in the absence or presence of nisin or nisin + lysozyme. 3.4. Quality evaluation of cocoa beans Cocoa beans fermented for 6 days in the absence or presence of nisin or nisin + lysozyme were fully fermented as judged by the cut test, with all 30 beans showing a full brown color. The shell weights of dried beans from these fermentations were similar (p N 0.05). The percentage shell

weights were, respectively, in the absence and presence of nisin, 14.92% and 15.32% for the 2011 beans and 11.62% and 11.38% for the 2012 beans (2500 ppm Nisaplin + lysozyme). The beans from all fermentations gave chocolates with typical sensory characters. Statistical analysis (t-test) revealed that the mean liking scores of chocolates made from beans fermented with the growth of LAB (the absence of nisin or nisin + lysozyme) were not significantly different (p N 0.05) from the scores of beans where the growth of LAB was inhibited or restricted during fermentation (presence of nisin or nisin + lysozyme) (Fig. 8). Chocolates prepared from batches of beans

a

3.5E+7

Area unit/g nib (dry basis)

Area unit/g nib (dry basis)

V.T.T. Ho et al. / International Journal of Food Microbiology 205 (2015) 54–67

3.0E+7 2.5E+7 2.0E+7 1.5E+7 1.0E+7 5.0E+6 0.0E+0

63

b

3.5E+7 3.0E+7 2.5E+7 2.0E+7 1.5E+7 1.0E+7 5.0E+6 0.0E+0

Fig. 7. The concentration of pyrazines produced during roasting of cocoa nibs fermented in the absence ( ) and presence of nisin or nisin + lysozyme ( ) of beans harvested in March 2011 (a) and May 2012 (b).

fermented in the presence of 3500 ppm Nisaplin were similar in chocolate flavor and liking to those prepared from beans fermented in the presence of 2500 ppm Nisaplin. 4. Discussion Using controlled, laboratory fermentations and the bacterial inhibitors nisin and lysozyme, we evaluated the contribution of LAB to cocoa bean fermentations. Based on microbiological and chemical criteria, discussed in the following sections, these small scale (5 kg) fermentations followed the behavior of larger, commercial scale cocoa bean fermentations. Pereira et al. (2012) also noted similar conclusions when comparing small and large scale cocoa fermentations. These inhibitors did not change the normal contribution of yeasts and acetic acid bacteria to the fermentations. In one trial (2011 beans), they completely inhibited the growth of LAB while in another trial (2012 beans), they eliminated the contributions of L. pentosus and L. fermentum and restricted the growth of L. plantarum by 100 fold (b106 CFU/g) and to a shorter time frame. Thus, by comparing between fermentations in the absence and presence of these inhibitors, we are able to gain some insight into the contribution of LAB to the process and cocoa bean quality. 4.1. Microbial ecology of fermentations Many studies have described the microbial ecology of cocoa bean fermentations (reviewed in De Vuyst et al., 2010; Lima et al., 2011; Schwan et al., 2015; Thompson et al., 2013). A diversity of yeasts, LAB

a

8

b

8

6

Mean of liking score

Mean of liking score

and AAB has been isolated from cocoa fermentations but, within these groups, a few key species frequently dominate and contribute to the process in a somewhat successional order. While this microbial ecology is generally consistent, worldwide, a critical review of the literature also reveals variations in the presence or absence of some key species in the process, the population levels (103–109 CFU/g) that develop during fermentation, and the timing when different species are found (Lima et al., 2011; Schwan et al., 2015). The reasons for these ecological variations have not been conclusively investigated but probably reside in the many pre-harvest variables (e.g., bean cultivar, plantation location, pod maturity, damage, pesticide residues) and post-harvest variables (e.g., fermentation mode, time, temperature, mixing) that can affect the process (Saltini et al., 2013). Such variations are reflected in our data for the fermentations with beans harvested in 2011 and 2012. The microbial ecologies of our fermentations are broadly consistent with what has been reported previously. For the yeasts, the population levels and the successional growth of H. guilliermondii, P. kudriavzevii and K. marxianus or S. cerevisiae follow the patterns reported in our previous study (Ho et al., 2014) and by others (Dircks, 2009; Jespersen et al., 2005; Moreira, et al., 2013; Nielsen et al., 2007; Papalexandratou et al., 2011b, 2013; Pereira et al., 2013). For the AAB, A. pasteurianus and G. frateurii were the main species found in our fermentations and their growth was consistent with data reported elsewhere (Ardhana and Fleet, 2003; Camu et al., 2007, 2008b; Lefeber et al., 2011a; Nielsen et al., 2007; Papalexandratou et al., 2011b, 2013). However, the weak contribution of these bacteria (b 106 CFU/g) to the 2012 fermentations was notable, although others have also reported such low contributions from these bacteria (Ardhana and Fleet, 2003; Moreira,

4

2

6

4

2

0

0

Chocolate

Overall

Chocolate

Overall

flavour

liking

flavour

liking

Fig. 8. Sensory evaluation of chocolates made from cocoa beans fermented in the absence ( ) and presence of nisin ( ); beans harvested in March 2011 (a) and May 2012 (b).

64

V.T.T. Ho et al. / International Journal of Food Microbiology 205 (2015) 54–67

et al., 2013; Pereira et al., 2012, 2013). The growth profiles of these species of yeast and AAB were the same for fermentations conducted in the absence and presence of nisin and, consequently, we can conclude that, in our studies, contributions of these two groups were not affected by the presence or absence of LAB. As expected, LAB grew during the control fermentations. L. plantarum was the only species associated with fermentation of the 2011 beans, while L. pentosus, L. plantarum and L. fermentum contributed to fermentations of the 2012 beans. The growth of L. plantarum and L. fermentum in cocoa fermentations has been extensively reported (Camu et al., 2007; Dircks, 2009; Garcia-Armisen et al., 2010; Kostinek et al., 2008; Lefeber et al., 2011a; Nielsen et al., 2007; Papalexandratou et al., 2011a,2011b,2011c, 2013; Pereira et al., 2013). L. pentosus is not commonly associated with cocoa fermentation but it was previously reported in studies conducted in the Dominican Republic (Galvez et al., 2007) and Malaysia (Papalexandratou et al., 2013). As mentioned already, the growth of these species was completely inhibited or substantially decreased in the presence of nisin or nisin + lysozyme, thereby enabling judgements to be made about their impacts on cocoa bean fermentation.

4.2. Changes in chemical components during fermentation 4.2.1. Sugars and ethanol Cocoa bean fermentations are characterized by strong utilization of the pulp sugars, glucose and fructose, and their transformation into mainly ethanol which diffuses into the nibs of the beans (Camu et al., 2007, 2008b; Galvez et al., 2007; Ho et al., 2014; Lefeber et al., 2011a; Nielsen et al., 2007; Papalexandratou et al., 2011a,2011b,2011c; Pereira et al., 2013). Our fermentations followed this pattern. The fermentations conducted in 2012 gave greater sugar utilization and higher ethanol production than those in 2011 and this can be explained by the growth of S. cerevisiae in the 2012 trials. Since LAB exhibit strong hexose fermentation (Adler et al., 2013; De Vuyst et al., 2010), it is not unreasonable to expect slower and less utilization of pulp sugars when the growth of these bacteria was absent or decreased. This behavior was not observed in our trials and was also supported by the fact that similar ethanol levels were found for fermentations in the presence or absence of LAB. If LAB were major utilizers of pulp sugars during bean fermentation, then less would be available for yeast metabolism into ethanol. It seems, therefore, that the yeasts can metabolize these sugars before the LAB. Sucrose is the main sugar in unfermented cocoa nibs and it is hydrolyzed to glucose and fructose by bean invertase (Hansen et al., 1998; Voigt and Leiberei, 2015) during fermentation (Camu et al., 2007; Crafack et al., 2013; Ho et al., 2014; Papalexandratou et al., 2011a, 2011b; Pereira et al., 2012, 2013). Our fermentations gave these same changes, even when the growth of LAB was prevented. So it may be concluded that LAB did not impact on sugar metabolism within the nibs. Such metabolism is important because it determines the final concentrations of glucose and fructose in the nibs which undergo Maillard reactions during bean roasting to generate chocolate flavors (Reineccius et al., 1972; Rohan and Stewart, 1967). Mannitol is produced in the pulp during cocoa bean fermentation and it may transfer into the nibs (Camu et al., 2007, 2008a, 2008b; Crafack et al., 2013; Lefeber et al., 2012). Its production is attributed to the reduction of pulp fructose to mannitol by heterofermentative LAB, especially strains of L. fermentum (Adler et al., 2013; Moens et al., 2014). Nisin inhibited the growth of this species during the 2012 fermentations and hence mannitol production. Mannitol can be a substrate for the growth of acetic acid bacteria and, through this mechanism, bean acidity and pH could be impacted (Moens et al., 2014). It is not known how mannitol might impact on the sensory aspects of chocolate quality. Not all species or strains of LAB produce mannitol, so its production during indigenous fermentations can be variable, as we have observed.

Consequently, the significance of mannitol to the fermentation process, and chocolate quality needs further consideration. 4.2.2. Organic acids The utilization and production of organic acids during fermentation have important implications for the process and chocolate quality. The main acids of interest are citric, acetic and lactic. Citric acid occurs naturally in the pulp and nibs of cocoa beans, but acetic and lactic acids are products of microbial fermentation. The key issue is the acidity of the nibs at the end of fermentation. A decrease in nib pH from about 7.0 to about 5.5 is considered important to activate the various endogenous enzymes necessary for the development of chocolate flavor precursors, but excessive acidification depreciates chocolate quality (De Vuyst et al., 2010; Lopez and Dimick, 1995; Jinap, 1994; Schwan and Wheals, 2004; Thompson et al., 2013; Voigt and Lieberei, 2015). The concentrations reported for these acids in pulp and nib fractions vary considerably (Holm et al., 1993; Jinap, 1994; Thompson et al., 2013). While there are sound biological factors (e.g., bean cultivar, climate, pod maturity) that account for this variation, the methods for extraction, analyses and reporting of these organic acids are generally non-standardized, thereby compromising valid comparisons. Recent data for sugars and ethanol concentrations are mostly reported on a wet weight basis while those for organic acids have been reported as either wet or dry weight amounts. Citric acid (5–40 mg/g) is the main acid in the pulp of unfermented cocoa beans and it is thought to be partially or completely utilized during fermentation by the growth of yeasts and LAB (De Vuyst et al., 2010; Schwan and Wheals, 2004). Some species of LAB, especially L. fermentum, are strong utilizers of citric acid, and their growth often correlates with utilization of pulp citric acid and an increase in pulp pH (Crafack et al., 2013; De Vuyst et al., 2010; Moens et al., 2014; Moreira, et al., 2013; Pereira et al., 2012, 2013). Although we observed the increase in pulp pH (Fig. 3), it did not correlate with a decrease in citric acid concentration, despite significant growth of L. plantarum in the 2011 fermentations and L. fermentum and L. pentosus in the 2012 fermentations (Fig. 5). These differences might be explained by strain variation within species of LAB, as not all strains metabolize citric acid (Lefeber et al., 2010, 2011b; Pereira et al., 2012). Moreover, our data show production of citric acid in the pulp fraction during the later stages of fermentation. Such observations are not peculiar to our work, and increases in pulp citric acid can be seen in the studies of Camu et al. (2007), Dircks (2009), Papalexandratou et al. (2011b, 2011 c), and Pereira et al. (2012, 2013). Further research is needed to provide an explanation for these findings. Some yeasts produce citric acid (Anastassiadis et al., 2002; Jinap, 1994; McKay et al., 1990; Whiting, 1976) and it would be of interest to determine if species associated with cocoa (e.g., P. kudriavzevii, K. marxianus) fall into this category. Pulp citric acid does not diffuse into the cocoa nibs (De Vuyst et al., 2010; Lehrian and Patterson, 1983), so external changes to its concentration are not likely to impact on bean acidity or chocolate quality. The nibs of unfermented cocoa beans also contain significant levels of citric acid (2–9 mg/g) which decrease to a partial but significant extent during fermentation as shown in Fig. 5 and similarly reported elsewhere (Camu et al., 2007, 2008a; Dircks, 2009; Ho et al., 2014; Lefeber et al., 2011a; Pereira et al., 2012). The mechanism for this decrease has not been discussed in the literature and requires study since it could affect bean acidity and chocolate quality. Despite this decrease, acidity of the nibs increases, suggesting a greater influence of other acids on nib pH. Considering changes in the concentrations of citric acid in either pulp or nibs, we observed no differences between fermentations in the presence, absence or restricted growth of LAB. Lactic acid is not found in the pulp or nibs of unfermented beans. During fermentation, its concentration progressively increases in the pulp from which it transfers into the nibs, as found in our study (Fig. 5) and by others (Ardhana and Fleet, 2003; Camu et al., 2007, 2008a, 2008b; Crafack et al., 2013; Ho et al., 2014; Lefeber et al.,

V.T.T. Ho et al. / International Journal of Food Microbiology 205 (2015) 54–67

2011a; Papalexandratou et al., 2011a, 2011b, 2013; Pereira et al., 2013). Metabolism of pulp glucose and fructose by LAB is considered to be the main mechanism of lactic acid production, especially by facultative heterofermentative species such as L. plantarum and to a lesser extent by obligate heterofermentative species such as L. fermentum where a major part of the fructose is converted to mannitol instead of lactic acid (De Vuyst et al., 2010; Lefeber et al., 2011b; Moens et al., 2014). These outcomes are reflected in our data (Figs. 1, 2 and 5) where more lactic acid was produced in the 2011 control fermentations due to growth of L. plantarum, in contrast to a prevalence of L. fermentum in the 2012 fermentations. As might be expected, therefore, restriction of the growth of LAB by inhibitors restricted the production of lactic acid in the pulp and transfer to the nibs (Fig. 5). However, these differences were not reflected as significant changes to nib pH (Fig. 3) or sensory acceptability of the chocolate prepared from beans coming from the two types of fermentation (Fig. 8). In this context, our findings are inconsistent with the view that, through lactic acid production, LAB might negatively impact on bean acidity and chocolate quality (De Vuyst et al., 2010; Holm et al., 1993; Jinap and Zeslinda, 1995). Possibly, there was not enough production of lactic acid in our fermentations for these impacts to be observed. Surveys of dried fermented cocoa beans, globally, show lactic acid concentrations ranging from about 1–10 mg/g dry weight with most in the range of 2–5 mg/g. It has been difficult to find a conclusive relationship between the content of lactic acid in these beans, their pH and sensory quality, but beans with lactic acid exceeding 5 mg/g dry weight tend to have a pH value less than 5.0 and have lesser quality (Holm et al., 1993; Jinap, 1994; Jinap et al., 1995). The fully fermented, dried beans from our 2011 trials had lactic acid concentrations of 5–12 mg/g dry weight (Fig. 5) and a pH of 5.4–5.5 (Fig. 3). A novel observation was the production of small amounts of lactic acid during fermentation of the 2011 beans where the growth of LAB was inhibited. Such production could be attributed to the growth of yeasts, especially K. marxianus in this case (Fig. 1c, d), which are known to produce lactic acid from hexose metabolism (Plessas et al., 2008; Radler, 1993; Whiting, 1976). Consequently, yeasts could be a potential source of lactic acid during cocoa fermentation and more research is needed to determine their influence on such production. Some yeasts, such as K. marxianus and P. kudriavzevii, also have the potential to utilize lactic acid (Kurtzman, 2011; Lachance, 2011). This factor could also impact on the changes to its concentration during cocoa fermentation. Production of acetic acid during cocoa bean fermentation is considered essential because this acid diffuses into the nibs, causing their death and decrease in pH, which are factors needed for the endogenous biochemical reactions that give chocolate flavor precursors (Lopez and Dimick, 1995; Thompson et al., 2013). It is generally considered that most of this acid is produced by acetic acid bacteria that oxidize the ethanol produced by the yeasts (De Vuyst et al., 2010; Lima et al., 2011; Schwan and Wheals, 2004). LAB also have the capacity to generate this acid by heterofermentative metabolism of hexose sugars and from citric acid metabolism (Adler et al., 2013; De Vuyst et al., 2010; Lefeber et al., 2010; Schleifer and Ludwig, 1995). However, no evidence was obtained in our study to suggest that LAB were a major contributor to production of this acid. As expected, acetic acid was produced in the pulp and transferred to the nibs during control cocoa bean fermentations, but the same profiles were also obtained when the growth of LAB was inhibited (Fig. 5). The concentrations and profiles found for acetic acid production in the pulp and nibs were similar to those reported elsewhere (Camu et al., 2007, 2008a, 2008b; Crafack et al., 2013; Lefeber et al., 2011a; Papalexandratou et al., 2013; Pereira et al., 2013). The lower levels obtained for this acid during our 2012 trials are consistent with the weak association of acetic acid bacteria with these fermentations (Fig. 2). The lower concentration of acetic acid in the nibs of these beans may account for their higher pH (5.8–6.0, Fig. 3) and less sensory appeal of chocolate prepared from them (Fig. 8) compared with the beans

65

fermented in 2011. Nib pH should decrease to about pH 5.5 for optimum activity of the endogenous enzymes that generate chocolate flavor precursors (Lopez and Dimick, 1995; Thompson et al., 2013). However, nib pH and acetic acid content may not necessarily have a direct relationship because our data for the 2011 beans show sharp decreases in acetic acid content of nibs in the later stages of fermentation, without corresponding impact on nib pH (Figs. 3c, 5e, g). Such decreases in the concentration of nib acetic acid are also evident in the data of other researchers (Camu et al., 2008b; Crafack et al., 2013; Moreira, et al., 2013; Pereira et al., 2012, 2013) and further investigation is needed to explain this behavior and the relationship between nib pH and acetic acid content. Three other organic acids, malic, succinic and oxalic, were found in the pulp and nibs of our beans. The profiles of their changes were not affected by LAB. We draw attention to the relatively high concentration of succinic acid in pulp and nib fractions and its increase during fermentation. Its significance to the process and chocolate quality has, previously, been overlooked. The utilization of pulp malic acid during fermentation might be due to P. kudriavzevii which is a strong utilizer of this acid (Kim et al., 2008). 4.2.3. Volatile compounds In addition to producing primary metabolites such as ethanol, lactic and acetic acids during cocoa fermentation, yeasts and bacteria also produce a vast array of volatile secondary metabolites such as higher alcohols, fatty acids, esters, aldehydes, ketones, and thiols (Rodriguez-Campos et al., 2011, 2012) that could diffuse into the nibs and influence cocoa flavor. Although such volatiles would be largely lost during bean drying and roasting processes, there are suggestions that they could still have an important impact on the estery, fruity characteristics of chocolate flavor (Crafack et al., 2013; Lima et al., 2011). Previously, we demonstrated that yeasts were the major producers of such flavor volatiles during cocoa fermentation (Ho et al., 2014). In fermentations where yeast growth was inhibited and only LAB and AAB grew, there was little production of these secondary metabolites. With respect to LAB, these findings were confirmed in the present study where no differences in the production of these volatiles were found for fermentations in the presence or absence of LAB. Consequently, we conclude that LAB do not have a major role in their production and may not influence chocolate character by this mechanism. Nevertheless, further research is recommended on this topic. While our method to determine these volatiles detected a mixed array of some 50 metabolites, it may not have detected some of the more specific aldehydes, ketones, thiols etc. that can be produced by LAB (Annan et al., 2003; Smit et al., 2005). 4.2.4. Quality of fermented cocoa beans and chocolate After fermentation and drying, beans fermented in the presence (control) or absence of LAB had similar physical appearance, showed full fermentation by the cut test and gave similar shell weights that were within specification guidelines (less than 16%) (Fowler, 2009). Chocolates prepared from the two sets of beans were considered acceptable and were not significantly different in terms of chocolate flavor and overall liking as evaluated by sensory panelists. Based on these results, it can be concluded that the growth of LAB during cocoa bean fermentation was not necessary to give beans with an acceptable chocolate flavor. 5. Conclusions By integrating ecological, chemical and sensory analyses of cocoa beans fermented in the presence and absence of LAB, this study has demonstrated that the contributions of these bacteria may not be necessary for the production of beans that give typical chocolate character. In this context, it supports suggestions of early researchers (e.g., Barel, 1998 cited in De Vuyst et al., 2010; Nicholls, 1913) that these bacteria

66

V.T.T. Ho et al. / International Journal of Food Microbiology 205 (2015) 54–67

are probably not essential for successful cocoa bean fermentations. Further studies of cocoa fermentations under controlled conditions using inoculated starter cultures of particular species of LAB are needed to confirm this possibility. While the potential impact of LAB on the process and bean quality through modulation of acidity and pH is not in question, our data suggest the need for further research to better understand the roles of other microbial groups, such as yeasts, in this context and to better understand the biochemical reactions within the bean nib that affect their organic acid content. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ijfoodmicro.2015.03.031. Acknowledgments The authors gratefully acknowledge Mr. Yan Diczbalis, Department of Agriculture and Forestry, Queensland, Australia and Mr. Alan Mortimer, the Australian Blending Company for assisting in supplying cocoa beans for this project. Ms. Van Ho thanks the Department of Education, Employment and Workplace Relations, Australian Government for providing an Endeavour PhD Scholarship to conduct this research. References Adler, P., Bolten, C.J., Dohnt, K., Hansen, C.E., Wittmann, C., 2013. Core fluxome and metafluxome of lactic acid bacteria under simulated cocoa pulp fermentation conditions. Appl. Environ. Microbiol. 79, 5670–5681. Anastassiadis, S., Aivasidis, A., Wandrey, C., 2002. Citric acid production by Candida strains under intracellular nitrogen limitation. Appl. Microbiol. Biotechnol. 60, 81–87. Annan, N.T., Poll, L., Sefa-Dedeh, S., Plahar, W.A., Jakobsen, M., 2003. Volatile compounds produced by Lactobacillus fermentum, Saccharomyces cerevisiae and Candida krusei in single starter culture fermentations of Ghanaian maize dough. J. Appl. Microbiol. 94, 462–474. Ardhana, M.M., Fleet, G.H., 2003. The microbial ecology of cocoa bean fermentations in Indonesia. Int. J. Food Microbiol. 86, 87–99. Barel, M., 1998. Première transformation du cacao. Formation de l’arôme du cacao. In: Pontillion, J. (Ed.), Cacao et Chocolat. Lavoisier Tec. & Doc., Collection Sciences et Techniques Agroalimentaires, Paris, France, pp. 96–115. Biehl, B., Brunner, E., Passern, D., Quesnel, V.C., Adomako, D., 1985. Acidification, proteolysis and flavor potential in fermenting cocoa beans. J. Sci. Food Agric. 36, 583–598. Cadez, N., Smith, M., 2011. Hanseniaspora Zikes (1912). In: Kurtzman, C.P., Fell, J.W., Boekhout, T. (Eds.), The Yeasts: A Taxonomic Study, 5th ed. . vol. 2. Elsevier, Amsterdam, pp. 421–434. 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 bacteria and acetic acid bacteria involved in spontaneous heap fermentation of cocoa beans in Ghana. Appl. Environ. Microbiol. 73, 1809–1824. Camu, N., De Winter, T., Addo, S.K., Takrama, J.S., Bernaert, H., De Vuyst, L., 2008a. Fermentation of cocoa beans: influence of microbial activities and polyphenol concentrations on the flavour of chocolate. J. Sci. Food Agric. 88, 2288–2297. 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 environmental contamination on the dynamics of populations of lactic acid and acetic acid bacteria involved in spontaneous cocoa bean heap fermentation in Ghana. Appl. Environ. Microbiol. 74, 86–98. Crafack, M., Mikkelsen, M.B., Saerens, S., Knudsen, M., Blennow, A., Lowor, S., Takrama, J., Swiegers, J.H., Petersen, G.B., Heimdal, H., Nielsen, D.S., 2013. Influencing cocoa flavour using Pichia kluyveri and Kluyveromyces marxianus in a defined mixed starter culture for cocoa fermentation. Int. J. Food Microbiol. 167, 103–116. De Vuyst, L., Lefeber, T., Papalexandratou, Z., Camu, N., 2010. 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, Iowa, USA, pp. 301–325. Dircks, H.D., 2009. Investigation Into the Fermentation of Australian Cocoa Beans and Its Effect on Microbiology, Chemistry and Flavour. The University of New South Wales, Australia. Fowler, M.S., 2009. Cocoa beans: from tree to factory. In: Beckett, S.T. (Ed.), Industrial Chocolate Manufacture and Use, 4th ed. Blackwell Science, Oxford, UK, pp. 10–47. Galvez, 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. Int. J. Food Microbiol. 114, 124–130. Garcia-Armisen, T., Papalexandratou, Z., Hendryckx, H., Camu, N., Vrancken, G., De Vuyst, L., Cornelis, P., 2010. Diversity of the total bacterial community associated with Ghanaian and Brazilian cocoa bean fermentation samples as revealed by a 16 S rRNA gene clone library. Appl. Microbiol. Biotechnol. 87, 2281–2292. Hansen, C.E., del Olmo, M., Burri, C., 1998. Enzyme activities in cocoa beans during fermentation. J. Sci. Food Agric. 77, 273–281. Ho, V.T.T., Zhao, J., Fleet, G., 2014. Yeasts are essential for cocoa bean fermentation. Int. J. Food Microbiol. 174, 72–87.

Holm, C.S., Aston, J.W., Douglas, K., 1993. The effects of the organic-acids in cocoa on the flavor of chocolate. J. Sci. Food Agric. 61, 65–71. 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 Res. 5, 441–453. Jinap, S., 1994. Organic acids in cocoa beans — a review. ASEAN Food J. 9, 3–12. Jinap, S., Dimick, P.S., 1990. Acidic characteristics of fermented and dried cocoa beans from different countries of origin. J. Food Sci. 55, 547–550. Jinap, S., Zeslinda, A., 1995. Influence of organic-acids on flavor perception of Malaysian and Ghanaian cocoa-beans. J. Food Sci. Technol. 32, 153–155. Jinap, S., Dimick, P.S., Hollender, R., 1995. Flavor evaluation of chocolate formulated from cocoa beans from different countries. Food Control 6, 105–110. Johnson, E.A., Larson, A.E., 2005. Lysozyme. In: Davidson, P.M., Sofos, J.N., Branen, A.L. (Eds.), Antimicrobials in Foods. CRC Press, New York, USA, pp. 361–388. Kim, D.H., Hong, Y.A., Park, H.D., 2008. Co-fermentation of grape must by Issatchenkia orientalis and Saccharomyces cerevisiae reduces the malic acid content in wine. Biotechnol. Lett. 30, 1633–1638. 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. Curr. Microbiol. 56, 306–314. Kresnowati, M.T.A.P., Suryani, L., Affifah, M., 2013. Improvement of cocoa beans fermentation by lactic acid bacteria starter addition. J. Med. Bioeng. 2, 274–278. Kurtzman, C.P., 2011. Pichia. E.C. Hansen (1904). 5th ed. In: Kurtzman, C.P., Fell, J.W., Boekhout, T. (Eds.), The Yeasts: A Taxonomic Study vol. 2. Elsevier, Amsterdam, pp. 685–707. Lachance, M.A., 2011. Kluyveromyces van der Walt (1971). 5th ed. In: Kurtzman, C.P., Fell, J.W., Boekhout, T. (Eds.), The Yeasts: A Taxonomic Study volume 2. Elsevier, Amsterdam, pp. 470–481. Lefeber, T., Janssens, M., Camu, N., De Vuyst, L., 2010. Kinetic analysis of strains of lactic acid bacteria and acetic acid bacteria in cocoa pulp simulation media toward development of a starter culture for cocoa bean fermentation. Appl. Environ. Microbiol. 76, 7708–7716. Lefeber, T., Gobert, W., Vrancken, G., Camu, N., De Vuyst, L., 2011a. Dynamics and species diversity of communities of lactic acid bacteria and acetic acid bacteria during spontaneous cocoa bean fermentation in vessels. Food Microbiol. 28, 457–464. Lefeber, T., Janssens, M., Moens, F., Gobert, W., De Vuyst, L., 2011b. Interesting starter culture strains for controlled cocoa bean fermentation revealed by simulated cocoa pulp fermentations of cocoa-specific lactic acid bacteria. Appl. Environ. Microbiol. 77, 6694–6698. Lefeber, T., Papalexandratou, Z., Gobert, W., Camu, N., De Vuyst, L., 2012. On-farm implementation of a starter culture for improved cocoa bean fermentation and its influence on the flavour of chocolates produced thereof. Food Microbiol. 30, 379–392. Lehrian, D.W., Paterson, G.R., 1983. Cocoa fermentation. In: Reed, G. (Ed.), Biotechnology, a Comprehensive Treatise vol. 5. Verlag Chemie, Basel, Switzerland, pp. 529–575. Lima, L.J.R., Almeida, M.H., Nout, M.J.R., Zwietering, M.H., 2011. Theobroma cacao L., “the food of the gods”: quality determinants of commercial cocoa beans, with particular reference to the impact of fermentation. Crit. Rev. Food Sci. Nutr. 51, 731–761. Lopez, A.S., Dimick, P.S., 1995. Cocoa fermentation. In: Rehm, H.J., Reed, G. (Eds.), Biotechnology, 2nd ed. VCH Publishers, Weinheim, Basel, Switzerland, pp. 561–577. McKay, I.A., Maddox, I.S., Brooks, J.D., 1990. Citrate production by yeast. In: Yu, P.L. (Ed.), Fermentation Technologies: Industrial Applications. Elsevier Applied Science, London, pp. 285–291. Moens, F., Lefeber, T., De Vuyst, L., 2014. Oxidation of metabolites highlights the microbial interactions and role of Acetobacter pasteurianus during cocoa bean fermentation. Appl. Environ. Microbiol. 80, 4702–4716. Moreira,, I.M.d.V., Miguel, M.G.d.C.P., Duarte, W.F., Dias, D.R., Schwan, R.F., 2013. Microbial succession and the dynamics of metabolites and sugars during the fermentation of three different cocoa (Theobroma cacao L.) hybrids. Food Res. Int. 54, 9–17. Nicholls, L., 1913. Chapter VII. In: Smith, H.H. (Ed.), The Fermentation of Cacao, 1st ed. John Bale, Sons and Danielson, London, UK, pp. 221–351. 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 culturedependent and culture-independent methods. Int. J. Food Microbiol. 114, 168–186. Papalexandratou, Z., Camu, N., Falony, G., De Vuyst, L., 2011a. Comparison of the bacterial species diversity of spontaneous cocoa bean fermentations carried out at selected farms in Ivory Coast and Brazil. Food Microbiol. 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 microbiota associated with traditional Ecuadorian spontaneous cocoa bean fermentations. Appl. Environ. Microbiol. 77, 7698–7714. 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 species diversity of lactic acid bacteria and acetic acid bacteria. Food Microbiol. 28, 1326–1338. Papalexandratou, Z., Lefeber, T., Bakhtiar, B., Seng, L.O., 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 Microbiol. 35, 73–85. Pereira, G.V.D., Miguel, M., Ramos, C.L., Schwan, R.F., 2012. Microbiological and physicochemical characterization of small-scale cocoa fermentations and screening of yeast and bacterial strains to develop a defined starter culture. Appl. Environ. Microbiol. 78, 5395–5405. Pereira, G.V.D., Magalhaes, K.T., de Almeida, E.G., Coelho, I.D., Schwan, R.F., 2013. Spontaneous cocoa bean fermentation carried out in a novel-design stainless steel tank: influence on the dynamics of microbial populations and physical–chemical properties. Int. J. Food Microbiol. 161, 121–133.

V.T.T. Ho et al. / International Journal of Food Microbiology 205 (2015) 54–67 Plessas, S., Fisher, A., Koureta, K., Psarianos, C., Nigam, P., Koutinas, A., 2008. Application of Kluyveromyces marxianus, Lactobacillus delbrueckii ssp. bulgaricus, and L. helveticus for sourdough bread making. Food Chem. 106, 985–990. Pongtharangkul, T., Demirci, A., 2004. Evaluation of agar diffusion bioassay for nisin quantification. Appl. Microbiol. Biotechnol. 65, 268–272. Radler, F., 1993. Yeasts — metabolism of organic acids. In: Fleet, G.H. (Ed.), Wine Microbiology and Biotechnology. Harwood Academic Publishers, London, UK, pp. 165–182. Reineccius, G., Keely, B.J., Weissberger, W., 1972. Factors affecting the concentration of pyrazines in cocoa beans. J. Agrigc. Food Chem. 20, 202–206. Rodriguez-Campos, J., Escalona-Buendia, H.B., Orozco-Avila, I., Lugo-Cervantes, E., Jaramillo-Flores, M.E., 2011. Dynamics of volatile and non-volatile compounds in cocoa (Theobroma cacao L.) during fermentation and drying processes using principal components analysis. Food Res. Int. 44, 250–258. Rodriguez-Campos, J., Escalona-Buendia, H.B., Contreras-Ramos, S.M., Orozco-Avila, I., Jaramillo-Flores, E., Lugo-Cervantes, E., 2012. Effect of fermentation time and drying temperature on volatile compounds in cocoa. Food Chem. 132, 277–288. Roelofsen, P.A., 1958. Fermentation, drying and storage of cacao beans. Adv. Food Res. Int. 8, 225–296. Rohan, T.A., Stewart, T., 1967. Precursors of chocolate aroma: production of reducing sugars during fermentation of cocoa beans. J. Food Sci. 32, 399–402. Saltini, R., Akkerman, R., Frosch, S., 2013. Optimizing chocolate production through traceability: a review of the influence of farming practices on cocoa bean quality. Food Control 29, 167–187.

67

Schleifer, K.H., Ludwig, W., 1995. Phylogenetic relationships of lactic acid bacteria. In: Wood, B.J.B., Holzapfel, W.H. (Eds.), The Genera of Lactic Acid Bacteria. Blackie Academic and Professional, London, UK, pp. 5–18. Schwan, R.F., 1998. Cocoa fermentations conducted with a defined microbial cocktail inoculum. Appl. Environ. Microbiol. 64, 1477–1483. Schwan, R.F., Wheals, A.E., 2004. The microbiology of cocoa fermentation and its role in chocolate quality. Crit. Rev. Food Sci. Nutr. 44, 205–221. Schwan, R.F., Pereira, G.V.M., Fleet, G.H., 2015. Microbial activities during cocoa fermentation. In: Schwan, R.F., Fleet, G.H. (Eds.), Cocoa and Coffee Fermentations. CRC Press, Florida, USA, pp. 129–192. Smit, G., Smit, B.A., Engels, W.J., 2005. Flavour formation by lactic acid bacteria and biochemical flavour profiling of cheese products. FEMS Microbiol. Rev. 29, 591–610. Thomas, L.V., Delves-Broughton, J., 2005. Nisin. In: Davidson, P.M., Sofos, J.N., Branen, A.L. (Eds.), Antimicrobials in Foods, 3rd ed. CRC Press, Florida, USA, pp. 237–274. Thompson, S.S., Miller, K.B., Lopez, A., Camu, N., 2013. Cocoa and coffee. In: Doyle, M.P., Beuchat, L.R., Montville, T.J. (Eds.), Food Microbiology: Fundamentals and Frontiers, 4th ed. ASM Press, Washington DC, USA, pp. 881–899. Voigt, J., Lieberei, R., 2015. Biochemistry of cocoa fermentation. In: Schwan, R.F., Fleet, G.H. (Eds.), Cocoa and Coffee Fermentations. CRC Press, Florida, USA, pp. 193–226. Whiting, G.C., 1976. Organic acid metabolism of yeasts during fermentation of alcoholic beverages — a review. J. Ins. Brew. 82, 84–92.