The mechanistic rules for species coexistence

Global biodiversity loss requires effective ecological community management, reliant on accurate predictions of species coexistence and community assembly ^1-4 . Traditional methods, predicting coexistence from the effect of one species on another ^5-8 (i.e., phenomenologically), are sensitive to environmental context ^9-13 . This is because they ignore the fundamental processes that can be applied across environments. While mechanistic approaches offer promise ^14-17 , empirical tests remain rare ^18,19 . Here, we integrated a mechanistic consumer-resource model with the growth of 12 phytoplankton species in monoculture over a range of phosphorous, nitrate or ammonium concentrations. We found that the mechanistic approach accurately predicts the composition of 960 communities across species richness and resource conditions. As confirmed by simulations, species competing for substitutable resources (ammonium vs. nitrate) exhibit greater diversity than those competing for essential resources (phosphorus vs. nitrate), especially when initial species richness is high. This is because when competing for essential resources, each species is likely to consume less of the resource that more limit its growth, violating the mechanistic rule of coexistence (each species must consume more of the resource that more limit it ¹⁶ ). Our study highlights the power of the mechanistic approach in understanding and predicting species loss across environments and, ultimately, mitigating its pace.