A method for evaluating sediment-induced macroinvertebrate community composition changes in Idaho streams

Human activities can increase sediment delivery to streams, changing the composition, distribution, and abundance of stream aquatic life. Few U.S. states have numeric water quality standards for streambed sediment under the Clean Water Act, so managers often need to develop local application-specific benchmarks. This study developed stream surface fine sediment <2 mm (sand and fines, SF) and macroinvertebrate fine sediment biotic index (FSBI) benchmarks and an application framework to test for sediment-induced macroinvertebrate community composition changes in 1^st-4^th order Idaho streams. FSBI reference benchmarks were calculated as the 25^th percentile FSBI value among reference sites within three ecoregion-based site classes. Two approaches were used to develop SF benchmarks. Quantile regression was used to define reach-specific SF benchmarks representing an upper bound value expected under reference conditions. In addition, logistic regression was used to predict SF values with 50% and 75% probability that FSBI is worse than reference within each stream order and site class. The strength of association between SF benchmarks and macroinvertebrate community condition was evaluated by calculating relative risk using multiple datasets and examining responses of multiple macroinvertebrate indicators to SF benchmark status. SF reference benchmarks generally had stronger associations with poor macroinvertebrate condition than SF stressor-response benchmarks. Across datasets and macroinvertebrate indicators, poor macroinvertebrate condition was 1.8-3 times more likely when SF reference benchmarks were exceeded than when achieved. We propose rating the strength of evidence for a surface fine sediment-induced macroinvertebrate community composition change at the sample event scale as 'unlikely' if both SF and FSBI reference benchmarks are achieved, having 'mixed evidence' if only one reference benchmark is achieved, and 'likely' if both reference benchmarks are not achieved. We recommend combining ratings with other relevant data in a weight-of-evidence approach to assess if sediment impairs aquatic life.