Disentangling within- and between-species components of spatial community variation reveals processes driving community assembly

Background/Question/Methods

Although it is generally understood that spatial patterns of community structure reflect a mixture of both deterministic and stochastic driving processes, it is unclear how the relative importance of these processes may be estimated from distributional data. Here we couple a novel statistical framework that simultaneously quantifies the within- and between-species components of spatial community variation and a spatially explicit community simulation model to explain avian community structure.  Specifically we used the simulation model to infer the relative importance of dispersal limitation and environmental filtering in generating the spatial structure of the empirical avian communities. A priori it is unclear what the strengths of dispersal limitation and environmental filtering are in the empirical community, and therefore we examined what community patterns resulted from all combinations of 10 environmental niche breadths and 10 dispersal breadths. The ranges of parameters were chosen such that a full range of processes strengths were considered.  After simulating hundreds of communities we used a pattern-matching approach to identify which sets of simulation parameters resulted in communities which most closely matched the observed spatial avian community structure. We carried out this analysis using the USGS breeding bird survey data across the lower 48 states. 

Results/Conclusions

Our analyses suggest that the avian communities we examined typically display a significant signature of within-species aggregation and positive between-species association. The strong positive species associations are driven primarily by a lack of negative species associations rather than an over abundance of positive species associations. This is particularly true for distances between 350 and 600 km in which positive associations are typically as frequent as expected under the null hypothesis of species independence. Our simulations indicated that this particular spatial pattern of community covariance is typical when environmental filtering is of intermediate strength and dispersal limitation is weak. We will discuss the implications of these results as well as the basis of the systematic geographic variation in community covariance we observed.