Beyond Taxonomy: A Matrix–Trait–Function Framework for Predictive Selection of Non-Saccharomyces Yeasts in Food Fermentation

The growing diversity of food fermentation systems has intensified interest in non-Saccharomyces yeasts due to their broad metabolic capabilities and technological potential. However, current understanding of yeast functionality remains fragmented and frequently relies on taxonomy-centered classification, which often provides limited predictive value across fermentation systems. This review critically examines how strain-specific microbial traits, food matrix composition, and process conditions collectively shape fermentation performance across brewing, wine, cereal, plant-based, and functional fermentation systems. Particular emphasis is placed on key determinants of microbial functionality, including carbon metabolism, aroma biogenesis, acidification, enzymatic activity, microbial interactions, and transformation of food-associated bioactive compounds such as glycosides, phenolics, terpenes, and matrix-bound metabolites. The available evidence demonstrates that fermentation-relevant functionality cannot be reliably inferred from species identity alone because microbial performance is strongly modulated by strain variability and matrix-dependent environmental constraints. To address these limitations, this review proposes a matrix–trait–function framework that integrates microbial metabolic capabilities with food matrix characteristics and technological objectives to support a more predictive and application-oriented approach to yeast selection in food fermentation systems.