Multimodal integration: Mechanisms of temperature dynamics and quality formation critical period in Daqu

Abstract

The quality of medium-high temperature Daqu, the core starter for strong-aroma Baijiu, is regulated by the synergistic mechanisms of temperature, physicochemical properties, and microbial activity. In this study, we aimed to integrate dynamic monitoring of indicators, metagenomic analysis, and machine learning modeling to establish a multimodal approach. The systematic analysis of the differential contributions of spatiotemporal factors to Daqu fermentation temperature highlighted the dynamic changes in physicochemical and microbial processes during Daqu fermentation, as well as the critical period for quality control. The influence of temporal factors on Daqu temperature was significantly higher than that of spatial heterogeneity. Additionally, the temperature difference generated by the interaction of dual pathways between environmental changes and microbial metabolic heat production could regulate the Daqu fermentation through a heat-flow positive feedback mechanism. By combining temperatural and physicochemical data, machine learning models identified and validated the early fermentation stage (S2–S3) as the critical period for Daqu quality formation. Consequently, the quality control of Daqu can be effectively predicted and guided through monitoring the temperature in the early stage of fermentation. Metagenomic analysis revealed the two-phase characteristics of medium-high temperature Daqu fermentation: the core microbiota construction was completed in the S1–S3 stages, and the microbiota function then entered a stable period in the S4–S6 stages. This explains the dynamic change regularity of Daqu quality critical period formative from a microscopic perspective.