Resource availability structures microbial competition through genomic niche partitioning

Microbial competition for scarce resources shapes biodiversity patterns and ecosystem function across global biomes, yet quantifying this process from genomic data has remained elusive. Here, we introduce CaCo, a scalable metric that transforms metagenomic carbohydrate-active enzyme profiles into precise measures of niche overlap and competition potential (Resource Partitioning Score, RPS). Analyzing 14,691 high-quality metagenome-assembled genomes spanning Ocean, freshwater, soil, and human gut microbiomes, we reveal a striking macroecological pattern: Niche overlap increases from partitioned specialists in oligotrophic oceans to overlapping generalists in carbon-rich environments, including the human gut. This gradient aligns with classic niche theory, as phylogenetic signals indicate that closely related taxa may compete most intensely. Multitiered validation, spanning BIOLOG phenotypes, synthetic cocultures, and interaction gradients, confirms CaCo's predictive power and captures competitive exclusion. CaCo bridges genomic potential and ecological reality, providing niche-breadth metrics and enabling testable predictions of how resource availability shapes microbial competition and community structure.