Competition and immigration lead to clusters of similar species, not trait separation

Patterns of trait distribution among coexisting species can potentially reveal the processes by which species assemble into communities. The dominant paradigm, that competition causes species to differ more than expected by chance, has limited empirical support. Here we show that when competition acts in concert with stochasticity and dispersal, communities spontaneously organize into clusters of similar species. Further, we show clusters feature generally across different nichemechanisms, and even under the confounding influence of environmental filters. While clusters have been previously reported as transient or else persisting under restricted circumstances, our results show they persist broadly under immigration. Previous stochastic nichestudies missed this effect because they only explored extremely low immigration. We provide parameter-free metrics for detecting clusters in field data, which we validate using simulations. We conclude that clusters are a more general pattern than overdispersion, and trait-based searches for niche differentiationmay be more successful once they account for this fact.