The scale-dependent importance of habitat factors and dispersal limitation in structuring Great Lakes shoreline plant communities

Niche-based and neutral models of community structure posit distinct mechanisms underlying patterns in community structure; correlation between species' distributions and habitat factors points to niche assembly while spatial pattern independent of habitat suggests neutral assembly via dispersal limitation. The challenge is to disentangle the relative contributions when both processes are operating, and to determine the scales at which each is important. We sampled shoreline plant communities on an island in Lake Michigan, varying the extent and the grain of sampling, and used both distance-based correlation methods and variance partitioning to quantify the proportion of the variation in plant species composition that was attributable to habitat factors and to spatial configuration independent of habitat. Our results were highly scale dependent. We found no distance decay of plant community similarity at the island scale (1-33 km). All of the explained variation (32%) in species composition among samples at this scale was attributed to habitat factors. However, at a site intensively sampled at a smaller scale (5-1,200 m), similarity of species composition did decay with distance. Using a coarse sampling grain (transects), habitat factors explained 40% of the variation, but the purely spatial component explained a comparable 22%. Analyzing plots within transects revealed variation in species composition that was still jointly determined by habitat and spatial factors (18 and 11% of the variance, respectively). For both grain sizes, most of the habitat component was spatially structured, reflecting an abrupt alongshore transition from sandy dunes to cobble beach. Space per se explained more variation in species composition at a second site where the habitat transition was more gradual; here, habitat acted as a less selective filter, allowing the signal of dispersal limitation to be detected more readily. We conclude that both adaptation to specific habitat factors and habitat-independent spatial position indicative of dispersal limitation determine plant species composition in this system. Our results support the prediction that dispersal limitation-a potentially, but not necessarily, neutral driver-is relatively more important at smaller scales.