Figure 1.
Preparation of the fermented products and sampling locations.
A). Process flow for the preparation of Monkoyo. Chibwantu is produced in the same way. If the product is allowed to be fermented for an additional 5 days an alcoholic drink develops that can be kept for prolonged periods of time. B). Process flow of Mabisi. C). Map of Zambia indicating sampling locations (see Table 1). D). Picture of the fermentation vessel (Calabash).
Table 1.
A total of 37 samples were collected of three product-types; Chibwantu, Mabisi and Munkoyo in various towns across Zambia (Figure 1C).
Figure 2.
Viable counts of bacteria present in product samples. Points show the average of all samples of the same product type (Chibwantu, Mabisi and Munkoyo; Table 1). Error bars show standard errors of the mean. PCA28: plate count agar incubated at 28°C for total viable count; PCA42, plate count agar incubated at 42°C, thermophiles; PCAan, plate count agar incubated at 28°C under anaerobic conditions, total (facultative anaerobes), M17: M17 agar, total (aerobic) lactic acid bacteria, MRSan: MRS agar incubated anaerobically, (facultative) anaerobic lactic acid bacteria; YM: total yeast; VB: total Enterobacteria.
Table 2.
List of strains isolated from product samples using culture based techniques.
Figure 3.
Maximum likelihood tree of bacterial 16S rRNA gene sequences. Triangle heights are proportional to the number of sequences. Numbers adjacent to each node represent aLRT statistics (SH-like supports). Scale bar shows branch lengths (expected number of substitutions per site). See Tables 2 and 3 and the Table S1 for full strain details. Several known strains (from the NCBI database are shown in bold). The inset shows a cluster analysis based on the product type using all strains in the phylogeny.
Table 3.
Number of RFLP patterns observed per clone library.
Table 4.
Diversity indices calculated based on RFLP patterns.
Table 5.
UniFrac significance of comparisons of microbial communities based on clone library sequences in the phylogeny presented in Figure 3.
Figure 4.
Effect of sampling location on product similarity.
Jackknife cluster analysis based on sampling location using pooled data over all clone library sequences. Numbers adjacent to each node represent the estimated support value (%) after 100 permutations.