Global patterns in gene content soil microbiomes emerge from ecological interactions

Correlations between the genetic or taxonomic composition of microbial communities and their environment are ubiquitous, with examples ranging from microbiome composition changing with host diet to changes with local nutrient conditions in the ocean. We do not understand how these patterns emerge from the complex interactions present in microbiomes, however. Here, leveraging a global dataset of soil metagenomes, we identify a pattern in gene content of soil microbiomes. The two enzymes that complete the first step in bacterial denitrification (Nar, Nap) exhibit opposite trends with pH: Nar increases at low pH, while Nap decreases. Lab enrichment experiments recapitulate this pattern, confirming that pH is causal. Experiments with isolates possessing these two enzymes reveal the surprising result that Nar-genotypes die during denitrification in acidic conditions – despite their ecological dominance at low pH. However, in communities of both genotypes, Nar dominates at low pH. We show that this outcome emerges from Nar specialization in nitrate utilization and Nap specialization in the utilization of the byproduct nitrite, which is toxic at low pH. Using experiments and simulations, we argue that this resource-sharing interaction arises due to trade-offs that constrain bacterial traits, likely giving rise to the global trend in gene abundances. Our results show how physiology drives ecological interactions that give rise to emergent patterns on a global scale.