Construction of microecological discrimination models for high-temperature Daqu from different regions

Chinese Baijiu's regional distinctiveness stems from its brewing microecology. In this study, high-temperature Daqu from six regions were analyzed using multi-omics and machine learning. Core microbial taxa (e.g., Bacillus, Thermomyces) were shared, but region-specific signatures emerged (Kroppenstedtia in Sichuan, Oceanobacillus in Guizhou). Bacterial communities responded to short-term environmental shifts, while fungi reflected long-term regional differences. Volatile profiling identified 226 metabolites, dominated by fatty acids (37.8%) and alcohols (26.57%), with Guizhou having the highest content. Network analysis revealed bacterial genera as versatile flavor contributors, while fungi showed specialized roles. Guangxi samples had lower enzymatic activity, correlating with reduced microbial diversity. Machine learning distinguished regions most accurately using volatile markers (Logistic Regression: AUC = 0.99, accuracy = 90%) versus microbial data (Neural Networks, AUC = 0.93, accuracy = 80%). Key discriminators included 18 microbial (e.g., Streptomyces) and 20 volatile biomarkers (e.g., 2-ethyl-6-methylpyrazine). This work deciphers the microecological drivers of Baijiu's regional traits, aiding quality control and origin identification.