Quantifying the dominance of local control and the sources of regional control in the assembly of a metacommunity.

To understand the relative roles of local and regional processes in structuring local communities, and to compare sources of dispersal, we studied plant species composition in the context of a field experiment in vernal pool community assembly. In 1999, we constructed 256 vernal pools in a grid surrounding a group of over 60 naturally occurring reference pools. Each constructed pool received a seeding or control treatment. Seeding treatments involved several "focal species" native to vernal pools in this region. Earlier analyses identified local habitat quality (pool depth) and pool history (seeding treatment) as strong predictors of local species composition. For the current analysis, we asked how connectivity among pools might enhance models of focal species presence and cover within pools, using long-term data from control pools and from unseeded transects within a stratified random sample of all constructed pools. We fitted connectivity models for each of four focal species, and compared the relative support for connectivity, seeding treatment, and pool depth as predictors of local species presence and cover. We modeled connectivity in several ways to quantify the relative importance of immigration (1) from constructed pools, (2) from reference pools within the study site, (3) from a cluster of natural pools off-site, and (4) along ephemeral waterways. We found strongest support for effects of connectivity with reference pools. Species presence in a target pool was usually well predicted by an exponential decline in connectivity with distance to source pools, and our fitted estimates of mean dispersal distance indicate strong dispersal limitation in this system. Effects of target and source pool size were also supported in some models, and long-term effects of seeding were supported for most species. However, pool depth was by far the strongest predictor of focal species presence, and depth rivaled connectivity with reference pools as a top predictor of cover after accounting for species presence. We conclude that local species composition was determined primarily by local processes in this system, and we encourage more widespread use of a straightforward method for weighing local vs. regional influences.