Disturbance moderates biodiversity–ecosystem function relationships: experimental evidence from caddisflies in stream mesocosms

We present empirical evidence that the direct and indirect effects of species richness on ecological processes depend on a system's disturbance regime. We manipulated the number of species of freshwater suspension feeders (net‐spinning caddisflies) in laboratory stream mesocosms where communities were either subject to a periodic mortality or were left undisturbed. We hypothesized that disturbance would alter the direct effect of suspension‐feeder species richness on the flux of their resource, particulate organic matter (POM), and alter the indirect effects of suspension‐feeder species richness on algal productivity and biofilm respiration (via nutrient recycling). The direct effect of suspension‐feeder species richness on POM flux did indeed change with disturbance regime. The flux of POM from the water column to stream bed was not influenced by species richness in the absence of disturbance, but increased as species richness was increased in disturbed streams. This apparently occurred because a high taxonomic evenness in disturbed streams minimized current “shading” (i.e., the blocking of flow from upstream to downstream neighbors) that, in turn, allowed diverse assemblages to capture a greater fraction of suspended resources. Disturbance also altered the indirect effect of suspension feeder species richness on algal productivity. In the absence of disturbance, increasing species richness led to dominance by a competitively superior suspension feeding taxon that had particularly low rates of nutrient excretion. Reduced nutrient availability, in turn, caused algal productivity to decline as a function of species richness. Periodic disturbances, however, prevented taxonomic dominance and eliminated this indirect effect of suspension feeder species richness on stream productivity. Our results suggest that disturbance can moderate relationships between biodiversity and ecosystem functioning by (1) increasing the chance that diversity generates unique system properties (i.e., “emergent” properties), or (2) suppressing the probability of ecological processes being controlled by a single taxon (i.e., the “selection‐probability” effect).