Abstract
Species sorting underlies changes in microbial community composition under varying environments, yet predicting the species diversity and their functional outcomes when exposed to novel conditions remains challenging. We explored this using mabisi, a Zambian traditional fermented milk, by propagating a shared starting mabisi microbial community across five novel substrates - raw bovine milk (control), low-fat milk, full-cream milk, and the infant formulas F100 and S26 – under static conditions and fixed dilution for ~ 33 generations, at three rural farm sites. The microbial community composition was profiled through 16 S rDNA sequencing, and community-level functioning through volatile organic compounds, pH, and consistency. We observed a substrate-driven divergence of microbial communities, with low-fat milk enriched with Acetobacter, full-cream milk enriched with Lactococcus, and S26 enriched with Lactiplantibacillus and Leuconostoc. This pattern exhibited a temporal community succession but remained consistent between sites. Similarly, the community-level functionality diverged but remained largely repeatable to properties known for mabisi. Our findings suggest that novel environments drive a species sorting process in natural microbial communities, and this process persists over time. Additionally, the repeatability of its community-level functionality points to underlying functional redundancy within diverse and successive microbial communities. Our study not only advances the understanding of how microbial communities adapt to environmental changes but also provides a basis for harnessing the mabisi-derived microbial community for broad biotechnological applications, particularly supporting the feasibility and effectiveness of its use in an in-house formulated F100 infant formula for treating malnourished children in developing regions. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s11274-026-04830-3.