To remain modern, the coexistence program requires modern statistical rigor

Abstract

A recent study by Van Dyke et al.1 paired experimental drought manipulations with demographic models and trait measurements to project major shifts in coexistence among a number of annual plant taxa. However, re-analysis of the data under alternative, more predictive competition models reveals that the authors original conclusions are very sensitive to slight variations in model form. Furthermore, propagating model parameter error into coexistence predictions results in relatively weak support for the majority of coexistence shifts predicted by the authors original model. These results highlight the need for increased statistical rigor when treating binary predictions of species coexistence as observed experimental outcomes, as is commonly practiced in empirical coexistence studies.