Cross-scale interactions between local and regional processes drive the structure of metacommunities

Background/Question/Methods

Scale-dependent approaches largely assume that processes give rise to patterns at corresponding scales. However, when species interactions have significant impacts on the demographic properties of local populations, they can generate dynamical signals that can propagate across scales via dispersal and control the spatial structure of metacommunities, even in the presence of strong regional processes such as large-scale environmental gradients. Current statistical variation partitioning approaches decompose metacommunity structure into spatial and environmental components, and thus tacitly assume that local processes such as species interactions do not affect spatial structure at regional scales. We developed a spatially-explicit metacommunity model and applied an extended version of variation partitioning in order to determine the independent and joint effects of (i) the environment, (ii) space and (iii) non-trophic species interactions. Comparing the results of this extended framework to those generated by the classical approach across different levels of dispersal and species interaction strength allowed us to quantify the degree to which community variation due to the cross-scale effects of species interactions can be incorrectly attributed to space or the environment.

Results/Conclusions

Our analyses revealed that the inclusion of species interaction strength did not increase the total amount of community variation explained because the variation due to species interactions and space were strongly collinear. This result, which was robust across simulations with different levels of dispersal and interaction strength, demonstrates that the spatial fraction in variation partitioning methods cannot always be attributed to regional processes such as dispersal. Instead, the spatial fraction likely reflects a complex combination of local (e.g., species interactions) and regional (e.g., dispersal) processes. Hence, incorporating the effects of species interactions into variation partitioning approaches may promote our ability to detect the cross-scale effects of local species interactions and thus better understand the drivers of metacommunity structure.