Species-specific traits predict whole-assemblage detritus processing by pond invertebrates.

Abstract

Functional trait diversity determines if ecosystem processes are sensitive to shifts in species abundances or composition. For example, trait variation suggests detritivores process detritus at different rates and make different contributions to whole-assemblage processing, which could be sensitive to compositional shifts. Here, we used a series of microcosm experiments to quantify species-specific coarse and fine particulate organic matter (CPOM and FPOM) processing for ten larval caddisfly species and three non-caddisfly species in high-elevation wetlands. We then compared trait-based models including life history, dietary, and extrinsic traits to determine which traits explained interspecific variation in detritus processing. Finally, we compared processing by mixed caddisfly assemblages in microcosms and natural ponds to additive predictions based on species-specific processing to determine if single-species effects are additive in multi-species assemblages. We found considerable interspecific variation in biomass-specific CPOM (13-fold differences) and FPOM (8-fold differences) processing. Furthermore, on a mass-specific basis, amphipods, chironomids, and caddisflies processed similar amounts of detritus, suggesting non-shredder taxa could process more than previously recognized. Trait models including dietary percent detritus, development rate, body size, and wetland hydroperiod explained 81 and 57% of interspecific variation in CPOM and FPOM processing, respectively. Finally, species-specific additive predictions were strikingly similar to mixed-assemblage processing in microcosms and natural ponds, with the largest difference being a 15% overestimate. Thus, additivity of species-specific processing suggests single-species rates may be useful for understanding functional consequences of shifting assemblages, and a trait-based approach to predicting species-specific processing could support generating additive predictions of whole-assemblage processing.