Abstract
Plant microbiomes shape crop performance, but the mechanisms by which host-derived metabolites influence the recruitment and organisation of beneficial microbes-and how these affect crop yield-remain poorly understood. Few studies have linked multi-kingdom microbiome structure, metabolite profiles, and agronomic traits under field conditions. We integrated 16S rRNA and ITS amplicon sequencing with untargeted LC-MS/MS metabolomics across 36 samples from two Zingiber officinale (ginger) cultivars with contrasting yields. Bacterial communities were primarily shaped by stochastic processes (neutral model R² = 0.67-0.68), while fungal assembly was deterministic (βNTI < -2 in roots and rhizomes). The high-yield cultivar exhibited more complex co-occurrence networks (596 vs. 272 edges) and enrichment of keystone taxa, including Talaromyces and Devosia. Metabolomic profiling identified 586 unique compounds, with 24 enriched pathways in the high-yield cultivar, notably isoflavonoid biosynthesis and folate metabolism. Key host metabolites-Niazimin A and 1-oleoyl-lysophosphatidic acid-showed strong positive correlations (r > 0.75, P < 0.01) with nitrogen-fixing and growth-promoting microbes, whereas Oxindole correlated negatively. These findings suggest that host metabolic shifts and keystone microbes co-regulate microbiome structure and nutrient cycling. Our results provide mechanistic insight into microbiome-mediated yield differences and a basis for microbiome-informed crop design.