Effects of Metacommunity Networks on Local Community Structures: From Theoretical Predictions to Empirical Evaluations

Abstract Although empirical support for it is at an early stage of development, metacommunity theory has become a mainstream component in ecological thinking. Four paradigms encapsulate proposed metacommunity mechanisms: (1) colonization–competition trade-off in “patch dynamics,” (2) community-dependent species–environment fitting in “species sorting,” (3) increase in local persistence through immigration from more successful populations in the “mass effect,” and (4) trait-independent assemblages of “neutral processes.” Using metapopulation models, we highlight the wide range of patterns predicted by different mechanisms. Neutral and mass effects should be enhanced by dispersal, thereby constraining the conditions for species sorting. The relative effects of dispersal on dominant and subordinate species determine the weakening or strengthening of patch dynamics. Potential methodological approaches to field-data evaluation of metacommunity theoretical predictions are considered here. The key difference among paradigms lies in the putative role of traits on species performance in different communities. We argue for the use of functional biodiversity analyses by maximum entropy (MaxEnt) procedures as powerful tools for distinguishing metacommunity mechanisms. The effect of dispersal on metacommunity mechanisms also requires realistic field-testing. The centrality–isolation gradient may directly determine dispersal, while graph theory provides measures of isolation and network structure; nevertheless, estimation of metacommunity networks is not straightforward. Percolation distances, minimum spanning trees, and physical connections between patches are excellent approaches toward this problem of estimation, but the configurations they describe may differ from the functional metacommunity network. We propose an estimation method based on maximization of network–local pattern coherence, and we consider further alternatives. The combined use of MaxEnt and graph theories will likely contribute significantly to advances in metacommunity concepts through rigorous validations based on field-collected data.