Cyclic pairwise interaction representing a rock–paper–scissors game maintains the population of the vulnerable yeast Saccharomyces cerevisiae within a multispecies sourdough microbiome

ABSTRACT Sourdough starters are spontaneously generated multispecies communities consisting of lactic acid bacteria (LAB) and yeasts. Traditionally used to make baked goods, these communities of microorganisms can be propagated through successive transfers to other containers. Successive transfers result in microbial community dynamics; however, the mechanisms guiding the dynamics are not fully elucidated. This study tracked five species of sourdough LAB-yeast communities (three LAB species and two yeast species) during successive transfers in vitro to explore the interspecies interactions driving the multispecies community dynamics. The microbial dynamics were modeled with a generalized Lotka–Volterra (gLV) equation, which can mathematically identify ecological types of interspecies interactions. The gLV model simulated the population dynamics of each species with Pearson’s correlation coefficient values of ≥0.817. Pairwise experiments revealed that competition (negative/negative) predominated among all pairwise interactions (70%), followed by amensalism (negative/neutral) (30%). In the pairwise community, the LAB species Limosilactobacillus fermentum suppressed Saccharomyces cerevisiae growth through amensalism; however, S. cerevisiae population was maintained when the community comprised 3–5 species, including the LAB species L. fermentum and Pediococcus pentosaceus. The key interspecies interactions for maintaining the S. cerevisiae population were theoretically determined using the gLV model; the three identified species interacted non-transitively with S. cerevisiae in a cyclic pairwise interaction, metaphorically representing a rock–paper–scissors relationship, which systemically supported the S. cerevisiae population in the multispecies community. These theoretical insights indicate that cyclic pairwise interaction is the main driver of LAB-yeast population dynamics, which helps sustain a vulnerable yeast species in a sourdough multispecies community. IMPORTANCE Traditionally, multispecies consisting of lactic acid bacteria and yeasts collaboratively engage sourdough fermentation, which determines the quality of the resulting baked goods. Nonetheless, the successive transfer of these microbial communities can result in undesirable community dynamics that prevent the formation of high-quality sourdough bread. Thus, a mechanistic understanding of the community dynamics is fundamental to engineer sourdough complex fermentation. This study describes the population dynamics of five species of lactic acid bacteria-yeast communities in vitro using a generalized Lotka–Volterra model that examines interspecies interactions. A vulnerable yeast species was maintained within up to five species community dynamics by obtaining support with a cyclic interspecies interaction. Metaphorically, it involves a rock–paper–scissors game between two lactic acid bacteria species. Application of the generalized Lotka–Volterra model to real food microbiomes including sourdoughs will increase the reliability of the model prediction and help identify key microbial interactions that drive microbiome dynamics.