Using functional traits and null models to determine macroinvertebrate community assembly processes at local and regional scales

Background/Question/Methods

Trait-based analyses provide a mechanistic approach to understanding the processes driving the patterns in macroinvertebrate community structure. Whittaker (1975) described differences in taxonomic composition among assemblages as β-diversity. β-diversity may result from taxonomic turnover of ecologically different species over an environmental gradient, or turnover of ecologically similar species along an environmentally homogenous transect (Ackerly and Cornwell, 2007). The former indicates environmental filtering, a local control over community assembly where functionally similar species tend to co-occur. The latter can result from competitive exclusion (local), a heterogeneous source pool (regional), or an environmental gradient interacting with a functional trait that is unaccounted for. Our objectives are (1) to determine which functional traits are non-randomly distributed among assemblages of macroinvertebrates, (2) determine whether traits are convergent (functionally similar species co-occur) or divergent (functionally similar species exhibit negative co-occurrence), and (3) determine how community sorting processes differ between local and regional scales. We assessed local patterns (β-diversity among riffles in headwater streams) in macroinvertebrate community structure from four forested headwater streams in the Nantahala National Forest, and regional patterns (β-diversity among forested watersheds) from 26 watershed scale observations within the Blue Ridge physiographic province. A neutral-lottery model was used as a null model to determine which traits were non-randomly sorted at local and regional scales. The relationships between co-occurrence patterns and functional similarity among taxa were analyzed using partial mantel tests to determine the spatial lag at which taxonomic turnover occurred within an ecotype.

Results/Conclusions

We found no divergent sorting patterns indicative of competitive exclusion in this study. Niche preference traits exhibited convergent sorting patterns at the local scale with a source pool extent <10km. This positive correlation between trait state and co-occurrence was not observed when the null-distribution of trait scores was estimated from a source pool with an extent larger than 10km. Different combinations of niche preference traits were non-randomly sorted when the observational grain size was increased to the watershed scale and the source pool extent was scaled up to the physiographic province. These results suggest different environmental filters influence community assembly at local and regional scales of observation. Furthermore, we can determine the nature and range of the different community assembly processes that make up the nested hierarchy of environmental filters that structure freshwater macroinvertebrate communities.