Microbial co-occurrence network and spatial metabolism underlie wheat cultivar effects on medium-high temperature Daqu fermentation.

Although wheat cultivar is widely accepted to shape Daqu quality, it remains unclear which cultivar-linked microbial property - taxonomic composition, spatial niche differentiation, or the topology of cross-domain interactions - is the proximate driver of flavor-precursor accumulation. To address this gap, four wheat cultivars (GS19, QM725, QM838 and ZM) were used as natural perturbations, and depth-resolved (core/surface) 16S/ITS profiling, untargeted metabolomics and volatilomics were combined across the 90-day fermentation-maturation trajectory of medium-high-temperature Daqu. Of the four cultivars, only QM725 produced a microbiome with high node density and a positive-edge-dominated topology (88.73% positive correlations versus 76.98% in ZM), here termed a "cooperative-network" phenotype. This phenotype was spatially partitioned into two complementary metabolic compartments: the surface (Group P) was enriched in glycerophospholipid and arachidonic-acid metabolism, driving ester and floral-note formation; the core (Group X) was enriched in α-linolenic-acid and amino-acid metabolism, promoting the accumulation of aldehydes/ketones and higher alcohols. QM725-specific flavor markers (2-heptanone, 6-methyl-5-hepten-2-ol, methyl 13,16-octadecadiynoate) mapped onto these compartments. Co-occurrence analysis further identified an unclassified Thermoactinomycetaceae lineage as a network-level negative regulator of key alcohols and aldehydes, whereas Lactobacillus and Bacillus form complementary co-occurrence modules linking amino-acid metabolism to flavor-precursor accumulation through decarboxylation and Strecker degradation pathways, consistent with our previous bioaugmentation evidence that wheat-origin Bacillus strains causally modulate Daqu protease/amylase activities and amino-acid pools. These results identify cultivar-driven differences in microbial co-occurrence network and spatial metabolism as a hitherto underused dimension of Daqu quality, and introduce a transferable network-phenotype criterion for cultivar selection in Baijiu production.