Abstract
The adult stage of Trichoptera is valuable for assessing the biotic integrity of streams; however, it is not clear which specific metric(s) have the greatest value for doing so. In this study, >500,000 adult caddisfly specimens reflecting 299 species were sampled and identified from 903 stream sites throughout the northcentral United States. Specimen data were compiled into 31 water quality metrics encompassing taxonomic richness, diversity indices, pollution tolerance, percent dominant taxa, and relative abundance of functional feeding groups. Each metric was individually tested for its ability to predict the known percentage of undisturbed habitat upstream of each sampling site using simple linear regression modeling. Most metrics were statistically significant but had R 2 values <0.30. The highest performing models were taxonomic richness at the species (R 2 = 0.40), genus (0.52), and family (0.59) levels and the Hilsenhoff Biotic Index (HBI) (0.37). The family level of taxonomic resolution produced models with a higher R 2 value than genus or species for all four of the metrics where taxonomic resolution was tested. Multiple linear regression models of all 31 metrics (R 2 = 0.65) and of combined family richness, HBI, and the ratio of shredders to filtering collectors (0.62) exhibited modest improvements over using family richness exclusively. These results indicated that simple taxonomic richness metrics constitute the most effective predictors of undisturbed upstream habitat when using adult caddisfly data, and that family richness may be the most effective of all due to low stochastic variation and ease of use.