Species and strain-specific phenotypes shape performance and output of a five-species yeast consortium

Spontaneous wine alcoholic fermentation is initiated by complex yeast communities, which shape the fermentation process and fermentation outcome. While this yeast ecosystem has been described extensively, the biotic factors impacting population dynamics, and the relative ecological importance of the taxonomic levels of species and of strains in defining these parameters remain poorly resolved. Here, we established a consortium of five yeast species that are commonly identified as major contributors to spontaneous wine fermentations: Saccharomyces cerevisiae, Lachancea thermotolerans, Hanseniaspora uvarum, Starmerella bacillaris, and Torulaspora delbrueckii. Each species in this consortium was represented by two phenotypically diverging strains. Two systematic consortium perturbations, single-species drop-out, and single-strain substitutions were applied, and fermentation kinetics, sugar utilization, population dynamics, primary metabolites, and major volatile compounds were characterized. Both species drop-out and strain substitution produced significant impacts on population dynamics, growth behaviors, and metabolic outputs. Two-way analysis confirmed that the non-Saccharomyces substitutions affected the metabolic output more broadly than the S. cerevisiae strain. Sugar utilization and population data jointly revealed competitive interactions among species, while chemical outputs tracked the identity of each consortium comprising different species and strains. The data support a hierarchical interpretation in which species identity establishes consortium structure, and strain identity modulates its effect size.IMPORTANCEMicrobial communities drive important processes in food production, yet we still poorly understand how these complex communities behave. Wine fermentation offers a tractable model, as it is carried out by multiple yeast species whose combined activity determines the metabolic outcome. It is commonly assumed that knowing which species are present is sufficient to predict community function, while differences among strains of the same species are often overlooked. By establishing a defined community of five wine yeast species and systematically removing species or substituting individual strains, we show that species identity sets the basic structure of the community, while the particular strain of each species tunes the size of that effect. This distinction matters for anyone seeking to design or predict the behavior of microbial communities, and it shows that strain-level differences must be accounted for when engineering predictable microbial fermentations.