Characterization and association of bacterial communities and nonvolatile components in spontaneously fermented cow...

In the ecosystem of spontaneously fermented cow milk, the characteristics and relationship of bacterial communities and nonvolatile components at different scales of geographical distances (provincial, county, and village levels) are unclear. Here, 25 sampling sites from Xin Jiang and Tibet, 2 provinces of China, were selected based on the distribution of spontaneously fermented cow milk and used for metagenomic and metabolomic analysis. At the provincial geographical distance, the same predominant species, Lactobacillus delbrueckii ssp. bulgaricus and Streptococcus thermophilus, were detected in Xin Jiang and Tibet. Further, the richness of the bacterial composition of samples from Tibet was higher than those from Xin Jiang; specifically, at the species level, 28 species were identified in Tibet samples but only 7 species in Xin Jiang samples. At the provincial geographical level, we detected significant differences in bacterial structure, shown in principal coordinate analysis plots, and significant differences (Simpson index) in bacterial diversity were also detected. However, at the county and village levels, no significant differences were detected in bacterial communities and diversity, but a difference in bacterial compositions was detectable. This indicates that bacterial communities and diversity of spontaneously fermented milk dissimilarity significantly increased with geographic distance. For the nonvolatile component profiles, the partial least squares discriminant analysis plot (R²Y > 0.5 and Q² > 0.5 for the goodness-of-fit and predictive ability parameter, respectively) showed that samples from different geographical distances (provincial, county, and village) were all separated, which indicated that all the discriminations in nonvolatile components profiles were from different geographical distances. Investigating relationships between lactic acid bacteria and discriminatory nonvolatile components at the county level showed that 9 species were positively correlated with 16 discriminatory nonvolatile components, all species with low abundance rather than the predominant species L. delbrueckii ssp. bulgaricus and Strep. thermophilus, which indicates the importance of the selection of autochthonous nonpredominant bacteria.