The microbiology of Bandji, palm wine of Borassus akeassii from Burkina Faso: identification and genotypic diversity of yeasts, lactic acid and acetic acid bacteria



Labia Irène Ivette Ouoba, Microbiology Research Unit, Faculty of Life Science, School of Human Sciences, London Metropolitan University, 166-220 Holloway Road, London N7 8DB, UK. E-mail:,



To investigate physicochemical characteristics and especially genotypic diversity of the main culturable micro-organisms involved in fermentation of sap from Borassus akeassii, a newly identified palm tree from West Africa.

Methods and Results

Physicochemical characterization was performed using conventional methods. Identification of micro-organisms included phenotyping and sequencing of: 26S rRNA gene for yeasts, 16S rRNA and gyrB genes for lactic acid bacteria (LAB) and acetic acid bacteria (AAB). Interspecies and intraspecies genotypic diversities of the micro-organisms were screened respectively by amplification of the ITS1-5.8S rDNA-ITS2/16S-23S rDNA ITS regions and repetitive sequence-based PCR (rep-PCR). The physicochemical characteristics of samples were: pH: 3·48–4·12, titratable acidity: 1·67–3·50 mg KOH g−1, acetic acid: 0·16–0·37%, alcohol content: 0·30–2·73%, sugars (degrees Brix): 2·70–8·50. Yeast included mainly Saccharomyces cerevisiae and species of the genera Arthroascus, Issatchenkia, Candida, Trichosporon, Hanseniaspora, Kodamaea, Schizosaccharomyces, Trigonopsis and Galactomyces. Lactobacillus plantarum was the predominant LAB species. Three other species of Lactobacillus were also identified as well as isolates of Leuconostoc mesenteroides, Fructobacillus durionis and Streptococcus mitis. Acetic acid bacteria included nine species of the genus Acetobacter with Acetobacter indonesiensis as predominant species. In addition, isolates of Gluconobacter oxydans and Gluconacetobacter saccharivorans were also identified. Intraspecies diversity was observed for some species of micro-organisms including four genotypes for Acet. indonesiensis, three for Candida tropicalis and Lactobacillus fermentum and two each for S. cerevisiae, Trichosporon asahii, Candida pararugosa and Acetobacter tropicalis.


fermentation of palm sap from B. akeassii involved multi-yeast-LAB-AAB cultures at genus, species and intraspecies level.

Significance and Impact of the Study

First study describing microbiological and physicochemical characteristics of palm wine from B. akeassii. Genotypic diversity of palm wine LAB and AAB not reported before is demonstrated and this constitutes valuable information for better understanding of the fermentation which can be used to improve the product quality and develop added value by-products.


Borassus akeassii Bayton, Ouédrago and Guinko, which has been previously misidentified in West Africa as Borassus flabellifer or Borassus aethiopum, is a newly identified species of palm tree found in the south-west of Burkina Faso and occuring as well in other west African countries such as Ivory Coast, Senegal, Mali and Niger (Bayton et al. 2006; Bayton and Ouédraogo 2009). Different parts of the tree can be used for various purposes but one important commercial activity associated with the tree is production of wine from the sap by the local population (Yaméogo et al. 2008). In Africa, palm wine, which is an important beverage in the traditions and economy, is obtained by tapping various palm trees such as Elaeis guineensis, Raphia hookeri, Phoenix dactylifera, B. aethiopum and Cocos nucifera (Mollet et al. 2000; Sambou et al. 2002; Amoa-Awua et al. 2007; Stringini et al. 2009; Ziadi et al. 2011). Elsewhere, palm trees such B. flabellifer in Asia and Acrocomia aculeata in South America are also tapped (Alcántara-Hernández et al. 2010; Naknean et al. 2010). According to the country of origin, palm wine is called by various names such as Toddy, Emu, Ogogoro, Nsafufuo, Nsamba, Mnazi, Yongo, Taberna, Tua or Tubak. Palm trees are usually tapped by extracting the sap from a live upright tree, but in some countries the tree is cut down before tapping (Amoa-Awua et al. 2007; Stringini et al. 2009). There are various ways of tapping palm trees (Dalibard 1999) but in general, tapping involves perforation of the trunk, insertion of a tube in the hole and collection of the sap in a container (gourd, clay pot, plastic container, glass bottle or calabash). The sap, which is originally sweet, ferments within 2–3 h and develops a stronger alcoholic taste and smell with time. If not consumed within 2–3 days, the fermenting sap develops a vinegar taste and the smell is not acceptable to consumers. Palm wine can undergo additional fermentation followed by distillation to produce a local gin (Kadere et al. 2004; Amoa-Awua et al. 2007). It can also be used for production of vinegar, e.g. in some Asian countries such as the Philippines.

Palm sap fermentation has been reported to be a lactic-alcoholic-acetic fermentation (Kadere et al. 2004; Aidoo et al. 2006). Presence of micro-organisms such as yeasts, lactic acid bacteria (LAB), acetic acid bacteria (AAB), enterobacteria, Bacillus spp., Micrococcus spp. and Staphylococcus spp. has been reported (Okafor 1972, 1975; Atputharajah et al. 1986; Malonga et al. 1995). However, yeasts, LAB and AAB are believed to play the most important role. The predominant yeast reported is Saccharomyces cerevisiae, but other yeasts such as Schizosaccharomyces pombe, Kodamaea ohmeri, Hanseniaspora occidentalis, Candida tropicalis, Kloeckera apiculata and Pichia ohmeri are also commonly detected (Atputharajah et al. 1986; Aidoo et al. 2006; Amoa-Awua et al. 2007). The predominant LAB reported are Lactobacillus plantarum and Leuconostoc mesenteroides (Aidoo et al. 2006; Amoa-Awua 2007; Ziadi et al. 2011), whereas AAB belong mostly to the genera Acetobacter and Gluconobacter (Okafor 1975; Lisdiyanti et al. 2001; Alcántara-Hernández et al. 2010).

Studies have been performed on various aspects of palm trees in Burkina Faso (e.g. characterization, identification of new species, ecology, uses and commercialization), but to date, to the best of our knowledge, no study has addressed the physicochemical characteristics and microbiology of the wine prepared from the sap of the various types of palm tree present in the country, including the new species B. akeassii. In palm wine fermentation in general, most published information has focused on the role of yeasts, but full identification to species level of AAB is very rare (nonexistent for palm wines from African origin) and published studies on the genotypic diversity of LAB and AAB at species and intraspecies level seem to be nonexistent.

The present study aimed to address the lack of published information on palm wine described in the previous section, thus enhancing understanding of palm sap fermentation to improve the production of palm wine and develop added value by-products.

Materials and methods

Sampling and physicochemical characterization

Different production sites of palm wine were visited in the region of Bobo Dioulasso (town and villages) in Burkina Faso. After discussion with the producers, the technology was recorded and samples were collected in sterile containers, stored immediately on ice and transported to the laboratory for physicochemical and microbiological analyses. Samples from local sellers were also collected for analyses. Samples obtained directly from producers were mainly overnight (14–16 h) fermented samples, and those collected from sellers included overnight fermented samples as well as samples which have been stored and fermenting for additional 1–2 days at ambient temperature (21–30°C during the raining season in the precited region where the samples were collected).

For physicochemical characterization of the wines, methods described by AOAC (1990) and Amoa-Awua et al. (2007) were used.

Microbiological characterization

Enumeration and isolation of micro-organisms

Each palm wine sample (10 g) was mixed for 1 min with 90 g of sterile Maximum Recovery Diluent (MRD; Oxoid CM0733B, Basingstoke, UK) to a final weight of 100 g. Furthermore, ten-fold dilutions (10−2–10−10) were prepared and 100 μl of each dilution was spread-plated on Plate Count Agar (PCA; Oxoid CM0325B) for total mesophilic microflora enumeration. Plates of PCA were incubated aerobically at 37°C and the count recorded after 2 days. Yeasts, LAB and AAB were enumerated and isolated using modified methods described by Amoa-Awua (2007). Yeasts were enumerated after inoculation of serial dilutions on Oxytetracycline Glucose Yeast Extract Agar (OGYEA; Oxoid CM0545B). The plates were inoculated aerobically at 30°C for up to 5 days, followed by enumeration, selection of colonies and purification. For LAB, DeMan, Rogosa and Sharpe Agar (MRS; Oxoid CM0361B) containing 10 μg ml−1 of cycloheximide (Oxoid SR0222C) to suppress yeast growth, was surface- inoculated and incubated anaerobically (Oxoid Gas Kits BR0038B) for 3 days at 30°C. Colonies were counted and separately subcultured several times on MRS agar (MRS; Oxoid CM0361B) without cycloheximide (Oxoid SR0222C) until pure cultures were obtained. AAB were enumerated and isolated by spread plating on YPM (yeast extract, peptone and mannitol) agar at 30°C for 3 days. The YPM agar contained distilled water, 5 g l−1 of yeast extract (Oxoid LP0021), 3 g l−1 of peptone (Oxoid LP0037B), 25 g l−1 of mannitol (Sigma M4125, Gillingham, UK) and 12 g l−1 of agar (Oxoid, LP0013). To the media, 20 ml l−1 of absolute ethanol was added, as well as 10 mg ml−1 (1 ml for 1 l of media) of cycloheximide (Oxoid SR0222C) made up with 50% of ethanol and 20 ml l−1 of penicillin (Sigma P3032-25MU), prepared from a 0·25% stock solution, to inhibit growth of yeasts and LAB respectively. Colonies of presumptive AAB were isolated and purified on YPM agar.

After purification, presumptive yeasts, LAB and AAB were transferred respectively into nutrient broth (Oxoid CM0001B), MRS broth (Oxoid CM0359B) and YPM broth (formulation as above without added agar), all containing 30% glycerol and stored at −20°C until required.


Purified micro-organisms were grown for 2–3 days on appropriate media at 30°C for bacteria and 25°C for yeasts and phenotyped. All micro-organisms were observed for colony and cell morphology. Bacteria were additionally screened for cell motility, Gram stain and catalase reaction. Presumptive AAB were grown on GYC (glucose, yeast extract and calcium carbonate) agar to detect their ability to produce acid, which causes dissolution of calcium carbonate and clearing of the agar. The GYC medium contained distilled water, 10 g l−1 of yeast extract (Oxoid LP0021), 30 g l−1 of CaCO3 (Sigma C6763), 50 g l−1 of glucose (Sigma G8270), 20 ml l−1 of absolute ethanol and 20 g l−1 of agar (Oxoid, LP0013). Yeasts were additionally screened for carbohydrate assimilation using API ID 32 C (BioMerieux, Basingstoke, UK).


DNA extraction

Each isolate was streaked on the appropriate agar and incubated at 30°C for 48 h. The DNA of a pure colony was extracted using InstaGene Matrix (Biorad 732-6030, Hemel Hempstead, UK), according to the manufacturer's instructions. The extracted DNA was stored at −20°C until required.

Amplification of the ITS1-5.8S rDNA-ITS2 and 16S-23S rDNA ITS regions

This was done as described by Ouoba et al. (2010). The PCR mixture composition was similar for both yeasts and bacteria, but with the difference that, for yeasts, the following primers (Jespersen et al. 2005): ITS1 (5′-TCCGTAGGTGAACCTGCGG-3′) ITS4 (5′-TCCTCCGCTTATTGATATGC-3′) were used to amplify the ITS1-5.8S rDNA-ITS2 regions. The PCR was performed in a thermocycler (Gene Amp PCR system 2700; Applied Biosystems, Paisley, UK) under the following conditions: initial denaturation at 94°C for 3 min; 30 cycles of denaturation at 94°C for 2 min, annealing at 60°C for 1 min, extension at 72°C for 2·5 min; final extension at 72°C for 7 min; holding at 4°C. For bacteria, (LAB and AAB) amplification of the 16S-23S ITS region was carried out with primers S-D-Bact-1494-a-S-20 (5′-GTCGTAACAAGGTAGCCGTA-3′) and L-D-Bact-0035-a-A-15 (5′-CAAGGCATCCACCGT-3′) (Ouoba et al. 2010). The cycling program was: initial denaturation at 94°C for 1 min followed by 35 cycles of denaturation at 94°C for 1 min, annealing at 55°C for 1 min and elongation at 72°C for 1 min. The PCR was ended with a final extension at 72°C for 7 min and the amplified product cooled at 4°C. The DNA fragments were separated by applying 10 μl of each PCR product with 2 μl of loading buffer to a 1·5% agarose gel (Biorad 161-310). DNA molecular marker (Direct Load TM Wide Range DNA Marker; Sigma D7058-1VL) was included as a standard. The gel was run in Tris-Borate-EDTA buffer (1× TBE) (Sigma T4415-1L) as described in the previous section (Ouoba et al. 2010). The DNA profiles were observed and all bacteria showing the same profile were clustered in the same group by combined visual observation and cluster analysis using the Bionumerics system: Bio-Numerics 2.50: Dice's Coefficient of similarity with the unweighted pair group method with arithmetic averages clustering algorithm (UPGMA; Applied Maths, Saint-Martens-Latem, Belgium).

Repetitive sequence-based PCR

Yeasts and bacteria were further differentiated by rep-PCR as described by Ouoba et al. (2010). The same primer GTG5 (5′-GTGGTGGTGGTGGTG- 3′) was used for all micro-organisms and amplification was carried out as follows: initial denaturation at 94°C for 4 min followed by 30 cycles of denaturation at 94°C for 30 s, annealing at 45°C for 1 min and elongation at 65°C for 8 min. The PCR ended with a final extension at 65°C for 16 min and the amplified product cooled at 4°C. The DNA fragments were separated as described in the previous section with the difference that the gel was run for 2 h and 30 min at 130 V. The DNA profiles obtained were examined and all bacteria showing the same profile were clustered in the same group as described in the previous section.

26S rRNA gene sequencing

Yeast isolates were identified by sequencing of the D1/D2-region of the 26S rRNA. Primers as described by Jespersen et al. (2005): NL-1 (5′-GCATATCAATAAGCG GAGGAAAAG-3′) and NL-4 (5′-GGTCCGTGTTTCAAGACG G-3′) were used for the amplification. The PCR was carried out by mixing 1·5 μl of each DNA extract with a mixture containing 5 μl of 10× PCR buffer (Applied Biosystems N8080161), 8 μl of dNTP (1·25 mmol l−1), 3 μl of MgCl2 (25 mmol l−1) 0·5 μl of formamide (Sigma F9037), 1 μl of primer NL1 (5 pmol μl−1), 1 μl of primer NL4 (5 pmol μl−1), 0·25 μl of AmpliTaq polymerase (5 U μl−1; Applied Biosystems N808-0161) and 29·75 μl of sterile high purity water. The amplification was as follows: first denaturation at 94°C for 5 min, then 30 cycles at 94°C for 90 s, 53°C for 30 s and 72°C for 90 s. The final extension was carried out at 72°C for 7 min and the product was cooled at 4°C.The PCR product was purified using QIAquick PCR Purification kit (Qiagen 28104, Crawley, UK) and sequenced using primers NL-1 and NL-4. The sequencing was achieved by capillary electrophoresis on a 3730xl DNA Analyser-Titania (Applied Biosystems).

16S rRNA gene sequencing

The analysis was performed as described by Ouoba et al. (2010). A 940 bp portion of conserved regions of the 16S rRNA gene was amplified using primers pA (100 μmol l−1; 5′-AGAGTTTGATCCTGGCTCAG-3′) and pE (100 μmol l−1; 5′-CCGTCAATT CCTTTGAGTTT-3′), at the following conditions: initial denaturation at 95°C for 5 min, then 35 cycles at 94°C for 1 min, 55°C for 1 min and 72°C for 1 min. The final extension was carried out at 72°C for 5 min and the product was cooled at 4°C. The PCR product was purified using QIAquick PCR Purification kit (Qiagen 28104). Sequencing to generate 550 bp of nucleotides with primer pD (5′-GTATTACCGCGG CTGCTG-3′) was carried out, followed by use of the ABI Big Dye Terminator ver. 3.1 Cycle Sequencing Kit (Applied Biosystems 4337455) to stop the reaction. The reaction was achieved by 35 PCR cycles using the following program: 95°C for 2 min, then 35 cycles at 96°C for 15 s, 40°C for 1 s and 60°C for 4 min. The sequencing was achieved by capillary electrophoresis on a 3730xl DNA Analyser-Titania (Applied Biosystems).

Sequencing of gyrB gene

Sequencing of gyrB gene that encodes the subunit B protein of DNA gyrase, an enzyme important in DNA replication, was performed on selected bacteria, mainly to differentiate closely related species of bacteria and to confirm the identity of bacteria obtained by 16S rRNA gene sequencing. Primers described by Yamamoto and Harayama (1995) and Thorsen et al. (2011) were used. The PCR was carried out by mixing 3 μl DNA with 5 μl 10× Dream Taq green buffer (EP0712; Fermentas, Cambridge, UK), 4 μl dNTP (2·5 mmol l−1), 4 μl MgCl2 (25 mmol l−1), 1 μl formamide (Sigma F9037), 2 μl of each primer (10 pmol μl−1) UP1-F (5′-GAAGTCATCATGACCGTTCTGCAYGCNGGNGGN AARTTYG A-3′) and UP2-R (5′-AGCAGGGTACGGATGTGCGAGCCRTCNACRTCN GCRTCNGTCAT-3′), 0·6 μl of Dream Taq green DNA polymerase (5 U μl−1; Fermentas EP0712) and 28·4 μl sterile high purity water. The reaction was achieved by 30 PCR cycles with the following program: 94°C for 2 min, then 30 cycles at 94°C for 1 min, 66°C for 1 min and 72°C for 2 min. The final extension was carried out at 72°C for 7 min and the product was cooled at 4°C. Positive amplicons were checked by electrophoresis, purified and sequenced using the forward primer UP1-F and/or the reverse primer UP2 R (3·2 pmol μl−1). The sequencing was achieved by capillary electrophoresis and the data analysed as described in the previous section.

Database search for identification of bacteria

For all sequences (16S rRNA, 26S rRNA and gyrB genes) a database search was performed in GenBank (National Center for Biotechnology Information, Bethesda, MD, USA) using the Blast program (Zhang et al. 2000). A second database search for 16S rRNA gene sequences was performed using EzTaxon server, which contains a manually curated database of type strains of prokaryotes and provides identification tools using a similarity-based search (Chun et al. 2007).


In Burkina Faso, the sap from the palm tree B. akeassii is commonly allowed to ferment naturally into palm wine. For tapping the sap, a bamboo ladder is used by producers to reach the top of the tree to perforate the trunk and insert a tube into the hole to deliver the sap into a receiver (traditionally, a gourd), followed by fermentation, which usually occurs within 2–3 h. The gourd remains attached to the bamboo tube and the fermenting sap is collected twice daily (early in the morning, and in the afternoon). Sap collected from different trees is mixed in plastic containers, redistributed in bottles (always kept open as a result of gas production) and sold. Freshly collected sap is sweet but with time, fermentation continues in the container and a more pronounced alcoholic smell and taste are noticed. A vinegar taste and smell are detectable 1 day after collection, and within 3–4 days these become very pronounced, at which stage the wine is on the verge of spoilage. Producers and consumers report that usually after 3 days Bandji is not consumable.

The physicochemical characterization revealed an acidic pH in all studied samples within the range 3·48–4·12, with a titratable acidity between 1·67 and 3·50 mg KOH g−1 and a concentration of acetic acid between 0·16 and 0·37%. The percentage of total alcohol varied between 0·30 and 2·73%. The concentration of carbohydrates expressed as degrees Brix was between 2·70 and 8·50. It was observed that samples with lower percentages of alcohol contained higher concentrations of sugar and vice versa.

The microbial analysis revealed a total count of mesophilic bacteria ranging between 2·0 × 106 and 6 × 109 CFU g−1, LAB ranging between 2·0 × 107 and 1·3 × 108 CFU g−1, AAB in the range of 1·2 × 105 and 1·0 × 106 CFU g−1, and finally yeasts ranging between 6 × 104 and 2 × 107 CFU g−1.

Identification of bacteria and yeasts by phenotyping and genotyping revealed diversity among the isolates at genus, species and intraspecies levels.

A total of 44 yeast isolates presenting variable macroscopic and microscopic characteristics were identified. Using ITS-PCR, 11 groups of yeasts representing different species were recorded (Table 1). Sequencing of the 26S rRNA gene led to identification of various genera and species and revealed a diversity of yeasts according to the sample (Table 1). The predominant yeast was S. cerevisiae (27·27%), followed by Arthroascus (Ar.) fermentans (15·9%), Issatchenkia orientalis (11·36%), C. tropicalis (11·36%), Trichosporon (T.) asahii (11·36%), Candida pararugosa (4·54%), Hanseniaspora uvarum (4·54%), K. ohmeri (2·27%), S. pombe (2·27%), Trichosporon asteroides (2·27%), Trigonopsis (Tr.) variabilis (2·27%), Galactomyces geotrichum (2·27%) and Candida quercitrusa (2·27%). Saccharomyces cerevisiae was isolated from all samples except sample M2 (Table 1). Using rep-PCR, variable DNA profiles were observed for some yeast species including three different profiles for C. tropicalis, two for S. cerevisiae, two for T. asahii and two for C. pararugosa (Table 1, Fig. 1). Using ITS-PCR, it was not possible to differentiate C. pararugosa and K. ohmeri as well as T. asahii and T. asteroides (Table 1). However, they were clearly differentiated by rep-PCR as seen in Fig. 1. As seen from Table 2, fermentation of sugars varied amongst the isolates. All were able to ferment glucose whereas none fermented melibiose. The fermentation of the remaining sugars varied according to genus and species.

Figure 1.

Dendrogram of cluster analysis of rep-PCR fingerprints of yeast isolates isolated from Bandji. The dendrogram is based on Dice's Coefficient of similarity with the unweighted pair group method with arithmetic averages clustering algorithm (UPGMA).

Table 1. Identification of yeasts isolated from Bandji
SamplesYeastsITS-PCR groupsRep-PCR groupsIdentification by 26S rDNA sequencing/Blast searching
M1YM1111·1 Saccharomyces cerevisiae
YM1211·1 S. cerevisiae
YM1366·2 Kodamaea ohmeri
YM1422·1 Ar. fermentans
YM1555·1 Issatchenkia orientalis
YAM1133·1 Trichosporon asahii
YAM1433·1 T. asahii
M2YM2122·1 Ar. fermentans
YM2322·1 Ar. fermentans
YM2422·1 Ar. fermentans
YM2522·1 Ar. fermentans
YM2622·1 Ar. fermentans
M1PYM1P111·1 S. cerevisiae
YM1P288·1 Schizosaccharomyces pombe
V1YV1133·2 T. asahii
YV1233·3 T. asteroides
YV1366·1 Candida pararugosa
YV1411·2 S. cerevisiae
YV1511·2 S. cerevisiae
YAV1544·2 C. tropicalis
YAV1677·1 Hanseniaspora uvarum
YAV1777·1 H. uvarum
V2YV2144·1 C. tropicalis
YV2255·1 I. orientalis
YV2344·1 C. tropicalis
YV2411·1 S. cerevisiae
YV2511·1 S. cerevisiae
YAV2122·1 Ar. fermentans
YAV2244·3 C. tropicalis
YAV2344·2 C. tropicalis
YAV241111·1 Trigonopsis variabilis
YAV261010·1 Galactomyces geotrichum
P1YP1166·1 C. pararugosa
YP1211·1 S. cerevisiae
YP1355·1 I. orientalis
YP1455·1 I. orientalis
YP1511·1 S. cerevisiae
YAP1533·1 T. asahii
YAP1633·1 T. asahii
P2YP2111·1 S. cerevisiae
YP2211·1 S. cerevisiae
YP2355·1 I. orientalis
YP2499·1 C. quercitrusa
YP2511·1 S. cerevisiae
Table 2. Carbohydrate assimilation profile of yeast species from Bandji
SugarsYeast species
Candida tropicalis Trichosporon asahii Saccharomyces cerevisiae Arthroascus fermentans Kodamaea ohmeri Candida pararugosa Candida quercitrusa Trigonopsis variabilis Schizosaccharomyces pombe Hanseniaspora uvarum Issatchenkia orientalis
  1. V: variable.


A total of 30 LAB were identified. They were mostly Gram-positive, catalase negative rods except six isolates, which were Gram positive, catalase negative cocci. Four genera and seven species of LAB corresponding to seven groups of ITS-PCR were identified (Table 3). The predominant genus was Lactobacillus representing 86·67% of the total isolates followed by the genera Leuconostoc (10%), Fructobacillus (6·67%) and Streptococcus (3·33%). Lactobacillus plantarum, found in all samples, was the dominant species and represented 46·67% of the total isolates, followed by Lactobacillus fermentum (20%), Lactobacillus paracasei (10%) and Leuc. mesenteroides (10%), Fructobacillus durionis (6·67%), Lactobacillus nagelii (3·33%) and Streptococcus mitis (3·33%). Using primer GTG5, intraspecies genotypic diversity was observed for Lact. fermentum isolates, which were divided into three groups by rep-PCR (Fig. 2, Table 3). The sequencing of gyrB gene allowed clear distinction between Lact. plantarum and Lactobacillus pentosus. These were not clearly differentiated by 16S RNA gene sequencing. Sequencing of gyrB gene also confirmed the identity of all other LAB as revealed by 16S rRNA gene sequencing /EzTaxon searching except for the isolates of Lact. paracasei which were not differentiated from Lactobacillus casei. However, these latter two isolates were very clearly differentiated by 16S rRNA gene sequencing/EzTaxon searching. For all isolates, the similarity of the sequences with those in the databases was between 98·90 and 100%.

Figure 2.

Dendrogram of cluster analysis of rep-PCR fingerprints of lactic acid bacteria or isolated from Bandji. The dendrogram is based on Dice's Coefficient of similarity with the unweighted pair group method with arithmetic averages clustering algorithm (UPGMA).

Table 3. Identification of lactic acid bacteria isolated from Bandji
SamplesBacteriaITS-PCR groupsRep-PCR groupsIdentification by 16S rRNA gene sequencing/Eztaxon searchingIdentification by gyrB gene sequencing/Blast searchinga
  1. a

    gyrB gene sequencing was performed on selected isolates from each repetitive sequence-based PCR group.

M1LM11IIII.1 Lactobacillus fermentum
LM12II.1Lact. plantarum/Lact. pentosus
LM13IIII.1 Lact. fermentum
LM14II.1Lact. plantarum/Lact. pentosus Lact. plantarum
LM15IIII.1 Lact. fermentum Lact. fermentum
LM16II.1Lact. plantarum/Lact. pentosus
M2LM23II.1Lact. plantarum/Lact. pentosus
LM24II.1Lact. plantarum/Lact. pentosus
LM25IIII.2 Lact. fermentum Lact. fermentum
LM26IIII.2 Lact. fermentum
LM28IIII.3 Lact. fermentum Lact. fermentum
M1PLM1P1II.1Lact. plantarum/Lact. pentosus
LM1P3IIIIII.1 Lactobacillus paracasei Lact. casei/Lact. paracasei
LM1P5IVIV.1 Leuconostoc mesenteroides Leuc. mesenteroides
V1LV11VIVI.1 Streptococcus mitis Strep. mitis
LV12II.1Lact. plantarum/Lact. pentosus
LV13VV.1 Fructobacillus durionis F. durionis
LV14IVIV.1 Leuc. mesenteroides
V2LV21II.1Lact. plantarum/Lact. pentosus
LV25II.1Lact. plantarum/Lact. pentosus Lact. plantarum
LV27IIIIII.1 Lact. paracasei
LV28IIIIII.1 Lact. paracasei
P1LP11II.1Lact. plantarum/Lact. pentosus
LP13VV.1 F. durionis
LP14IVIV.1 Leuc. mesenteroides
LP15VIIVII.1 Lact. nagelii Lact. nagelii
P2LP22II.1Lact. plantarum/Lact. pentosus Lact. plantarum
LP23II.1Lact. plantarum/Lact. pentosus Lact. plantarum
LP24II.1Lact. plantarum/Lact. pentosus
LP25II.1Lact. plantarum/Lact. pentosus

All 22 AAB screened were Gram-negative, catalase positive rods and showed clearing on GYC agar, revealing their ability to dissolve calcium carbonate by producing acid. They were all clearly identified by 16S rRNA gene sequencing/EzTaxon searching, with 99·13–100% similarity. Various genera and species were identified (Table 4) with the predominant genus being Acetobacter (Acet.) (90·90%). Other genera were Gluconobacter (G) (4·54%) and Gluconoacetobacter (Gl) (4·54%). Eleven species corresponding to 11 ITS-PCR groups were identified, with Acetobacter indonesiensis as the predominant species (31·81%). Other species included Acetobacter tropicalis (18·18%), Acetobacter estunensis (9·09%) Acetobacter ghanensis (9·09%), Acetobacter aceti (4·54%), Acetobacter lovaniensis (4·54%), Acetobacter orientalis (4·54%), Acetobacter pasteurianus (4·54%), Acetobacter cerevisiae (4·54%), Gluconobacter oxydans (4·54%) and Gluconobacter saccharivorans (4·54%). Intraspecies genotypic diversity was observed for Acet. indonesiensis and Acet. tropicalis isolates, which were respectively divided into four and two groups by rep-PCR (Fig. 3). It was noticed that gyrB gene sequencing using the described primers was not suitable for identification of AAB. Although the PCR revealed very clear positive amplicons leading to good quality sequences, the Blast search did not allow a conclusive identification (results not shown).

Figure 3.

Dendrogram of cluster analysis of rep-PCR fingerprints of mainly acetic acid bacteria isolated from Bandji. The dendrogram is based on Dice's Coefficient of similarity with the unweighted pair group method with arithmetic averages clustering algorithm (UPGMA).

Table 4. Identification of acetic acid bacteria isolated from Bandji
SamplesBacteriaITS-PCR groupsRep-PCR groupsIdentification 16S rRNA gene sequencing/EzTaxon searching
M1AM13HH.1 Acetobacter cerevisiae
AM15BB.1 Acetobacter tropicalis
M2AM21DD.1Acetobacter estunensis
AM22DD.1 Acet. estunensis
AM23JJ.1 Gluconacetobacter saccharivorans
AM24FF.1 Acetobacter pasteurianus
AM25AA.1 Acetobacter indonesiensis
AM26AA.4 Acet. indonesiensis
M1PAM1P2II.1 Gluconobacter oxydans
V1AV11AAA.3 Acet. indonesiensis
AV11BAA.3 Acet. indonesiensis
AV12BB.1 Acet. tropicalis
AV14CC.1 Acetobacter ghanensis
V2AV27KK.1 Acetobacter orientalis
AV28BB.2 Acet. tropicalis
ALV24EE.1 Acetobacter lovaniensis
P1AP13AA.2 Acet. indonesiensis
AP14GG.1 Acetobacter aceti
ALP12AA.1 Acet. indonesiensis
ALP16CC.1 Acet. ghanensis
P2AP23AA.2 Acet. indonesiensis
ALP26BB.1 Acet. tropicalis

The analysis of the 16S RNA gene sequences using both Blast and EzTaxon searches revealed that although Blast is a good tool for identification of micro-organisms, EzTaxon (a database containing only type strains) is an excellent support to Blast and leads to an improved differentiation of closely related species of bacteria. In fact, closely related species such as Lactparacasei and Lact. casei, Lact. nagelii and Lactobacillus ghanensis, Acet. cerevisiae and Acet. pasteurianus, Acet. ghanensis and Acet. lovaniensis, Gl. saccharivorans and Gl. europaeus, which could not be differentiated by Blast were very well separated by EzTaxon.

DNA sequences of selected isolates have been deposited in EMBL database and can be consulted through the following accession numbers: HE979549, HE979550, HE979551, HE979552, HE979553, HE979554, HE979555, HE979556, HE979557, HE979558, HE979559.


The results from the present study revealed that production of Bandji from B. akeassii does not significantly differ from wine production from other types of palm trees where sap is collected from a live upright tree. Production of palm wine by felling the tree before tapping as done in Ghana (Amoa-Awua et al. 2007) is not practiced in Burkina Faso.

The acidic pH of Bandji is similar to that of other palm wines (Amoa-Awua et al. 2007; Stringini et al. 2009; Adedayo and Ajiboye 2011) and may be attributed mainly to the activity of LAB but also AAB with production of organic acids such as lactic and acetic acids. The total alcohol content of our samples (0·3–2·73%) is similar to those of Amoa-Awua et al. (2007) collected during the first day of tapping (1·4–2·82%) and those of Stringini et al. (2009) 1–2 days after tapping (2·7–3·2%). However, this alcohol content is relatively low compared with the alcohol content of samples accumulated overnight in the study reported by Amoa-Awua et al. (2007), which showed values of up to 4·75–6·00%. Higher alcohol contents have also been reported in other studies (Stringini et al. 2009; Adedayo and Ajiboye 2011). This relatively low alcohol content of the Bandji samples could be related to several factors, such as the nature of fermentation, diversity of organisms comprising the microbiota, composition of sap, type of palm tree, type and season of tapping, time of collection and time between collection and analysis of the samples.

A diverse yeast population was detected in Bandji samples, with S. cerevisiae occurring in all samples except one. The presence of S. cerevisiae in palm wine fermentation has been reported by various authors who designated the species as responsible for the fermentation and aroma of the wine (Aidoo et al. 2006; Amoa-Awua et al. 2007; Stringini et al. 2008, 2009). In fact, the presence of S. cerevisiae in almost all Bandji samples indicates the importance of this species in fermentation of palm sap. However, the other yeasts detected in the samples also probably play a determinant role in the fermentation. Moreover, in sample M2 where S. cerevisiae was not detected, the only species of yeast isolated was Ar. fermentans. This yeast species was originally isolated from soil in Thailand (Lee et al. 1994) and to the best of our knowledge, has not been reported before in palm wine fermentation. Jolly et al. (2006) reported that the non Saccharomyces yeasts naturally occurring in wine fermentation are metabolically active and can contribute positively to fermentation. Schizosaccharomyces pombe, K. ohmeri and H. occidentalis have been reported to be as important producers of ethanol in toddy (another term for palm wine) as is S. cerevisiae (Aidoo et al. 2006) and could significantly contribute to the fermentation and aroma of the wine. In Bandji, I. orientalis, which seems not to have been previously reported in palm wine, as well as Candida spp., may also be important yeasts, as they occur in the majority of Bandji samples.

Using rep-PCR, significant genotypic diversity was observed within the species C. tropicalis, which showed three subgroups for five isolates. Limited genetic diversity was observed for S. cerevisiae and T. asahii and no diversity was observed for Ar. fermentans and I. orientalis. However, these species showed phenotypic diversity in sugar fermentations, with C. tropicalis still demonstrating the greatest diversity. Stringini et al. (2009) reported a low level of diversity at species level within the yeast isolates of palm wine of E. guineensis, but high levels of genetic diversity within S. cerevisiae, by amplifying minisatellite genes encoding cell-wall proteins. By this method, they differentiated 15 different strains of S. cerevisiae. The results from the present and previous studies revealed that multi-yeast starters are involved in palm sap fermentation, whether at the species or subspecies/strain level.

The presence and role of LAB in palm sap fermentation from different types of palm tree has been documented for different dominant species according to the origin of the palm wine studied. In the study by Amoa-Awua et al. (2007) and in the study herein reported, the dominant LAB occurring in all samples was Lact. plantarum. The presence of Leuc. mesenteroides has also been reported and, with Lact. plantarum, this species seems to play an important role in the fermentation (Okafor 1975; Aidoo et al. 2006; Amoa-Awua et al. 2007; Ziadi et al. 2011). From the Tunisian date palm sap, Leuc. mesenteroides and Lact. delbrueckii have been reported, whereas in the Mexican coyol palm wine, LAB present were identified as F. durionis, F. fructosus, Lact. nagelii, and Lact. sucicola. (Alcántara-Hernández et al. 2010; Ziadi et al. 2011). In Congolese oil palm wine, no Lactobacillus or Leuconostoc were detected but only Lactococcus lactis (Malonga et al. 1995) was detected. Presence of Streptococcus spp. has also been reported in Nigerian and Congolese palm wine (Okafor 1975; Malonga et al. 1995); however, no published study seems to have reported the presence of Lact. fermentum, which was the second most dominant species in Bandji and which is important in production of other African alcoholic beverages such as dolo and pito (Sawadogo-Lingani et al. 2007). There are also no reports of presence of Lactparacasei in others palm wines.

Lactic acid bacteria are recognized as the main micro-organisms responsible for the pH decrease in fermenting palm sap through production of organic acids, which give a sour taste to the product, which is to some extent acceptable and desirable in Bandji. These bacteria also control growth of undesirable micro-organisms such as enterobacteria by acid and H2O2 production (Sanni et al. 1999; Amoa-Awua et al. 2007; Alcántara-Hernández et al. 2010; Naknean et al. 2010; Ziadi et al. 2011). Moreover, they have been associated with the aroma, consistency and colour of palm wine by production of gums at the beginning of the fermentation (Lasekan et al. 2007).

Some research has been done on the genotypic diversity of yeast from palm wine, but no published work seems to have been done on diversity of LAB and AAB in the product at subspecies/strain level. In the present study, intraspecies genotypic diversity was observed for Lact. fermentum isolates using rep-PCR and this seems to be specifically sample-related, as a subgroup observed in a particular sample was not observed in others. However, no diversity was observed amongst the predominant Lact. plantarum and the other species comprising multiple isolates. One reason could be that this strain of Lact. plantarum is dominant and common to Bandji in that region, although the samples were from different producers and locations.

Various studies have reported presence of AAB in palm sap fermentation with identification mainly at the genus level. Full identification at the species level is unusual in general, nonexistent for African palm wines and there is a dearth of published information on their genotypic diversity. Reports of AAB during palm sap fermentation are inconsistent. Amoa-Awua et al. (2007) did not detect AAB in palm wine until the third day after tapping, whereas Stringini et al. (2009) detected AAB immediately after tapping. The genera of AAB reported in palm wine from others palm trees such E. guineensis, C. nucifera, B. flabellifer and A. aculeata are Acetobacter and Gluconobacter (Okafor 1975; Atputharajah et al. 1986; Aidoo et al. 2006; Amoa-Awua et al. 2007; Kadere et al. 2008; Alcántara-Hernández et al. 2010). The limited publications on species identity of AAB in palm wine include those of Lisdiyanti et al. (2001), who reported presence of Acet. lovaniensis, Acet. indonesiensis and Acet. tropicalis in Indonesian palm wine and Alcántara-Hernández et al. (2010), who recorded presence of Acet. pasteurianus in Mexican palm wine. Occurrence of the genus Gluconacetobacter and other species of Acetobacter (Acet. cerevisiae, Acet. estunensis, Acet. ghanensis, Acet. orientalis, Acet. aceti) identified in this study seems not to have been reported before in palm sap fermentation. The reason could be related to the limited number of advanced identification studies on AAB recovered from the product and also to the origin of the sap. The predominance of Acetobacter species in palm wine indicates that these species may play a determinant role in fermentation. Their metabolic activity may be variable as a high level of diversity was observed at the species and intraspecies level in their ITS and rep-PCR profiles. Acetobacter indonesiensis may be particularly important in Bandji as it occurred in the majority of samples.

The precise role of AAB during fermentation of palm sap remains unclear. Like LAB, they probably contribute to the acidification of the wine leading to inhibition of undesirable micro-organisms. Acetic acid comprises part of the aroma volatiles of palm wine (Naknean et al. 2010) and this can be attributed to metabolic activity by AAB. However, in general, AAB have been considered as spoilage micro-organisms in winemaking as they can adversely influence the quality of wine (Drysdale and Fleet 1988; Du Toit and Lambrechts 2002). In palm wine, AAB have also been considered as spoilage bacteria (Kadere et al. 2008), due to their metabolic activity, which leads to a high level of acetic acid after 3 days of fermentation, conferring on the product a strong vinegar taste and smell not acceptable for consumption. However, this metabolic activity, although precluding consumption of the wine, can be valuable for production of palm vinegar.

The present study revealed the presence of a diversity of yeasts, LAB and AAB in Bandji. Many microbial species such as Ar. fermentans, I. orientalis, Lact. fermentum, Lact. paracasei, Acet. cerevisiae, Acet. estunensis, Acet. ghanensis, Acet. orientalis, Acet. aceti, Gl. saccharivorans, have not been reported before in palm wine from other palm trees and may be specific to sap from B. akeassi.


The support for this research of London Metropolitan University, London, UK; Département Technologie Alimentaire/IRSAT/CNRST, Ouagadougou and Bobo Dioulasso, Burkina Faso; Mr Faustin Yé (contact with producers and sellers in Bobo Dioulasso) and all producers/sellers, is gratefully acknowledged.