Abstract
The gut microbiota functions as a complex adaptive system where microbes form structural modules known as "guilds." Each guild comprises taxonomically distinct microbes that work together as cohesive functional units, contributing to overall system function. Traditional taxon-based microbiome analyses often yield inconsistent associations with disease, limiting mechanistic insights. To address this, we compared guild-based and taxon-based approaches using datasets from a time-restricted feeding (TRF) study in mice. C57BL/6 J male mice were assigned to ad libitum feeding or TRF groups, with metabolic parameters and gut microbiota composition assessed over 12 weeks. Isocaloric TRF improved glucose tolerance and reduced weight gain in high-fat diet (HFD)-fed mice while maintaining metabolic stability in normal-fat diet-fed mice. To examine microbial contributions, 293 prevalent amplicon sequence variants (ASVs) from the 16S rRNA gene's V3-V4 regions were clustered into 34 co-abundance groups (CAGs), representing potential microbial guilds and accounting for 96% of the total sequence abundance. By contrast, the taxon-based approach classified 660 ASVs into 126 genera, capturing only 78% of the total sequence abundance while omitting 22% of sequences representing novel microbes. The 34 CAGs preserved community-level information more effectively than the 66 prevalent genera, as demonstrated by Procrustes analysis. Five CAGs correlated with improved metabolic phenotype under TRF, including unclassifiable ASVs. Notably, two key CAGs exhibited conserved diurnal rhythmicity under TRF. In contrast, ASVs within putative health-relevant genera displayed opposing TRF responses. This study underscores microbial guilds as key mediators of TRF's metabolic benefits and emphasizes the need to recalibrate taxon-based microbiome analysis biomarker discovery.