Fingerprints of high-dimensional coexistence in complex ecosystems

Abstract

The coexistence of many competing species is a long-standing puzzle in ecology. Classic niche theory explains coexistence by trade-offs between a few essential species traits. Here we study an unexplored frontier of this theory: we assume that coexistence is intrinsically high-dimensional, arising from many traits and trade-offs at once. Species interactions then appear almost random, but their disorder hides a diffuse statistical structure: competitors that become successful start by subtly favoring each other, and partitioning their impacts on other species. We validate our quantitative predictions using data from grassland biodiversity experiments. We conclude that a high biodiversity can be attained through a pattern of collective organization that cannot be understood at the species level, but exhibits the fingerprint of high-dimensional interactions.