Streams in a changing landscape : identifying candidate reference reaches to assess the physical and biotic integrity of Missouri's wadeable streams

Abstract

North American freshwater resources have grown increasingly imperiled as a result of human-induced landscape alterations, with many present-day stream fish and aquatic invertebrate communities representing only a fraction of their historic constituents. Consequently, resource managers need the ability to predict areas of high and low biological integrity to inform management decisions and meet conservation needs. Previous efforts to quantify anthropogenic disturbances to aquatic systems have resulted in indices lacking the ability to identify specific stressor impacts, describe the ways stressors alter the physical and chemical condition of receiving waters, and predict biological integrity. Thus, the need exists for a flexible, quantitative approach to characterizing stream impairment and identifying candidate least-disturbed stream reaches to serve as benchmarks for high quality physical habitat and biological integrity. After accounting for natural sources of biological variation (e.g., watershed drainage area, reach gradient, spring density), we used boosted regression trees to model the influence of channel morphology, substrate, cover, and water quality, and watershed-level flow modification and fragmentation, urbanization, agriculture, and point-source pollution on stream fish and aquatic macroinvertebrate community characteristics of wadeable streams of Missouri. Reach-level environmental predictors explained between 8% and 46% of the variation in our ten biotic metrics, with channel morphology and water quality metrics consistently accounting for more variation than substrate or cover. Biotic metrics related to stream health (e.g., Ephemeroptera, Plecoptera, Trichoptera richness, native lithophilic fish species richness) increased with bankfull width/depth ratios and dissolved oxygen, and decreased with increasing total chlorophyll. Watershed-level environmental predictors accounted for between 4% and 51% of the variation in biotic metrics, with stream health metrics increasing with forest cover, and decreasing with increased densities of headwater impoundments, road crossings, and pasture lands. In general, invertebrate metrics showed higher sensitivity to row-cropping and water quality impairment than did fish, particularly in the Ozark Highlands aquatic subregion. We used the results of our watershed-level models to predict biotic metric values to over 28,000 wadeable stream reaches across the state and rescaled and summed individual metric scores to generate an overall estimate of biological integrity at each site. We identified streams scoring in the top 95th percentile of each stream size class and aquatic subregion to serve as regional candidate least-disturbed reference reaches predicted to exhibit relatively high quality habitat and biotic conditions. Our method represents a novel approach to characterizing and forecasting stream impairment, and represents a critical step in refining existing biological indices, developing a companion physical habitat index, and ultimately conserving the diversity and integrity of Missouri's flowing waters.