The metacommunity concept provides a framework to link ecological processes across spatial scales, explicitly recognizing that both local and regional processes simultaneously act to shape patterns of local community composition.  The relative influence of regional spatial structure and local environment is a function of both organism and habitat attributes that affect connectivity, or dispersal between local sites. Bottomland habitats provide an informative setting to examine the relationship between local and regional processes in a well-connected habitat, since bottomland habitats are by definition connected in a landscape context, longitudinally arranged along the course of a river. The hydrologic dynamics of riparian systems increase functional connectivity, as the movement of water provides an effective dispersal vector for many species.  Additionally, as spatial scale increases, bottomland habitats become explicitly disconnected, once river basin boundaries are crossed, providing a useful dichotomy of scales to compare the relative influence of spatial structure and environment. We explored the role of spatial and environmental variables in explaining variation in plant community composition in bottomland habitats at these two spatial scales: across river basin boundaries and within single river basins. We used Mantel tests to investigate correlations between local vegetation composition and environmental and spatial variables and redundancy analysis (RDA) to partition variation in community composition between the two explanatory datasets. We then compared these results to other studies in the literature from both well-connected and patchy systems.  

Results/Conclusions

We found that in alluvial habitats, environmental variables explain more variation in plant species composition and structure than does spatial structure . Each of our analyses supported this same conclusion.  When we used in-stream distance instead of Euclidean distance for Mantel tests, correlation between vegetation composition and geographic distance further declined. RDA results indicate that local edaphic conditions explain the largest portion of total variance explained (21% of total variance).  In addition, the explanatory power of local environment increases when restricted to the scale of a single river basin where habitat connectivity and functional connectivity are highest.  Our results indicate similar explanatory power of environmental variables as found in other contiguous forested systems (mostly tropical forests), but still fail to account for a considerable portion of the variation in plant community composition (RDA residuals account for 70% of variation in abundance data).