Why ecologists struggle to predict coexistence from functional traits.

The rationale behind trait-based ecology is that shifting focus from species' taxonomic names to their measurable characteristics ('functional traits') leads to greater generality and predictive power. This idea has been applied to one of ecology's most intractable problems: the coexistence of competing species. But after 20 years, we lack clear evidence that functional traits effectively predict coexistence. Here, we present a theory-based argument for why this might be the case. Specifically, we argue that coexistence often depends on special quantities called 'process-informed metrics' (PIMs), which combine multiple traits and demographic characteristics in non-intuitive ways, obscuring any direct ties between individual traits and coexistence. We then lay a path forward for trait-based coexistence research that builds on mechanistic models of competition.