Selection for Coexistence May Help Explain Smooth Functional Landscapes of Microbial Ecosystems

Understanding how the function of microbial communities arises from the individual microbes has been an active and important area of research in ecology. Recent work from Skwara et al. (2023) found that the structure-function mapping of multiple empirical examples was surprisingly "smooth" (well fit by models as simple as linear or quadratic regression over presence/absence of species). Understanding what contributes to this smoothness is important because it might enable easier design of microbial communities and shed light on the significance of interactions to community function. One notable feature laboratory experiments of synthetic ecosystems share with natural ecosystems is that the set of species being assayed is non-random. Species in experiments are typically chosen to favor their ability to coexist, while in nature evolution selects for communities that coexist. Here, we adopt a simulation-based approach to test the likely impact of such species selection on the smoothness of community functional landscapes, and find that, at least in our model, filtering species for coexistence rendered the landscapes smoother. We discuss the implications for applications of community design and for understanding the species-function mapping.