Resource competition can explain simplicity in microbial community assembly

Predicting the composition and diversity of communities is a central goal in ecology. While community assembly is considered hard to predict, laboratory microcosms often follow a simple assembly rule based on the outcome of pairwise competitions. This assembly rule predicts that a species that is excluded by another species in pairwise competition cannot survive in a multispecies community with that species. Despite the empirical success of this bottom-up prediction, its mechanistic origin has remained elusive. In this study, we elucidate how this simple pattern in community assembly can emerge from resource competition. Our geometric analysis of a consumer-resource model shows that trio community assembly is always predictable from pairwise outcomes when one species grows faster than another species on every resource. We also identify all possible trio assembly outcomes under three resources and find that only two outcomes violate the assembly rule. Simulations demonstrate that pairwise competitions accurately predict trio assembly with up to 100 resources and the assembly of larger communities containing up to twelve species. We then further demonstrate accurate quantitative prediction of community composition using harmonic mean of pairwise fractions. Finally, we show that cross-feeding between species does not decrease assembly rule prediction accuracy. Our findings highlight that simple community assembly can emerge even in ecosystems with complex underlying dynamics. Significance Multispecies microbial communities play an essential role in the health of ecosystems ranging from the ocean to the human gut. A major challenge in microbial ecology is to understand and predict which species can coexist within a community. While a simple empirical rule utilizing only pairwise outcomes successfully predicts multispecies laboratory communities, its mechanistic origin has remained unexplained. Here, we find that the observed simplicity can emerge from competition for resources. Using a generic consumer-resource model, we demonstrate that community assembly of highly complex ecosystems is nevertheless well predicted by pairwise competitions. Our results argue that community assembly can be surprisingly simple despite the potential complexity associated with competition and crossfeeding of many different resources by many different species.