Clustering in community structure across replicate ecosystems following a long-term bacterial evolution experiment

Experiments to date probing adaptive evolution have predominantly focused on studying a single species or a pair of species in isolation. In nature, on the other hand, species evolve within complex communities, interacting and competing with many other species. It is unclear how reproducible or predictable adaptive evolution is within the context of a multispecies ecosystem. To explore this problem, we let 96 replicates of a multispecies laboratory bacterial ecosystem evolve in parallel for hundreds of generations. Here we find that relative abundances of individual species vary greatly across the evolved ecosystems and that the final profile of species frequencies within replicates clusters into several distinct types, as opposed to being randomly dispersed across the frequency space or converging fully. Our results suggest that community structure evolution has a tendency to follow one of only a few distinct paths.