Assembly processes inferred from eDNA surveys of a pond metacommunity are consistent with known species ecologies

Technological progress is enabling ecologists to create repeated, large-scale, structured, and standardised community surveys. However, it is unclear how best to extract information from these novel community data. We metabarcoded 48 vertebrate species from their eDNA in 320 ponds in England and applied the ‘internal-structure’ approach, which uses joint species distribution models to explain community compositions as the outcome of four metacommunity assembly processes: environmental filtering, dispersal, species interactions, and stochasticity. We find that the environment plays an important role in community assembly and that the inferred environmental preferences of species are consistent with their ecologies. We also infer negative biotic covariances between fish and amphibians, which is consistent with predator-prey interactions, and high spatial autocorrelation for the palmate newt, which is consistent with its hypothesised relictual distribution. Comparing sites in the metacommunity, environmentally and spatially distinctive sites are better explained by their environmental covariates and geographic locations, respectively, revealing sites where environmental filtering and dispersal limitation act more strongly. Furthermore, species belonging to different trait groups differ in how well environmental covariates, biotic covariances, and geographical locations explain their distributions. Overall, our results highlight the value of a modern interpretation of metacommunity ecology that embraces the fact that assembly processes differ between individual species and sites. We discuss how novel community data make feasible several study-design improvements that will strengthen the inference of metacommunity assembly processes from observational data. ### Competing Interest Statement D.W.Y. is a cofounder of NatureMetrics, which provides commercial eDNA services.