ABSTRACTMetacommunity ecology has become an important subdiscipline of ecology, but it is increasingly evident that its foundational theoretical and analytical frameworks do not adequately incorporate a realistic continuum of environmental and biotic process at play. We propose an approach that develops stronger links between theoretical and statistical frameworks to shift the focus towards the study of the ‘internal structure’ of metacommunities by dissecting how different species and different sites contribute to overall metacommunity structure. To illustrate this, we simulate data from a model that includes environmental variation, dispersal, biotic interactions, and stochasticity as the basic ecological processes that influence species’ (co)-distributions. We analyze the simulated data with hierarchical community models and propose a new method to visualize and analyze the simultaneous role of species co-distribution, environment, space, and stochasticity in this emerging statistical approach. We focus in particular on quantifying how species affect the overall structure of the metacommunity via differences in their dispersal and niche traits, and how environmental filtering, dispersal and species interactions varies from site to site in relation to environmental conditions and connectivity. Although there are still challenges ahead, this framework provides a roadmap for a more comprehensive approach by jointly developing more mechanistic theory based on community assembly processes and the analytical tools needed to map these concepts onto data in diverse landscapes and systems.