You are here:
Predictive Mapping of the Biotic Condition of Conterminous U.S. Rivers and Streams
Citation:
Hill, R., EricW Fox, S. Leibowitz, Tony Olsen, D. Thornbrugh, AND M. Weber. Predictive Mapping of the Biotic Condition of Conterminous U.S. Rivers and Streams. ECOLOGICAL APPLICATIONS. Ecological Society of America, Ithaca, NY, 27(8):2397-2415, (2017).
Impact/Purpose:
The EPA’s National Rivers and Streams Assessment (NRSA) reports on the condition of streams within the conterminous US (CONUS) and within 9 reporting regions. However, these condition assessments do not provide information on the spatial distribution of conditions within each region beyond those sample sites used to develop assessments. Under the Safe and Sustainable Water Resources National Program, work is being conducted to predict the probable biological condition of all streams within the conterminous US (CONUS). We developed predictive models of biological condition based on the condition classes (‘good’ versus ‘poor’) of the NRSA benthic invertebrate multimetric index. These models predict the probability (i.e., values = 0 - 1) of any stream being in good biological condition [Pr(good)] based on nearby and upstream features, such as urbanization or agriculture within the watershed. We produced a map of predictions for all ~1.1 million perennial stream segments within the CONUS. This product leverages the observed condition classes of the NRSA samples to produce spatially explicit predictions for all perennial streams within the CONUS. Evaluations of models and predictions suggest good to excellent performance. These predictions can be queried to identify candidate streams for conservation and restoration. For example, we provide an illustration within the document in which we select all streams within the CONUS that have Pr(good) values above the 95th percentile of values within each of the 9 regions. These streams represent the very best available streams within each region based on our modeling and could be placed on a candidate list for conservation. An additional illustration is provided that combines our predictions with another recent EPA effort that maps indices of watershed integrity and catchment integrity to identify stream segments that are good candidates for restoration. Thus, the predictions of Pr(good) could provide an important tool for prioritizing resources for management efforts. This work is one of three manuscripts contributing to the FY16 SSWR Annual Performance Reporting (APR) product 3.01B.1, “National maps of watershed integrity and stream condition and report and webinar describing these.” The results of this study could be important for three programs within the Office of Water: the National Aquatic Resource Surveys, the Healthy Watersheds Program, and the Biocriteria Program.
Description:
Understanding and mapping the spatial variations in the biological condition of streams could provide an important tool for assessment and restoration of stream ecosystems. The US EPA’s National Rivers and Streams Assessment (NRSA) summarizes the percent of stream lengths within the conterminous US that are in good, fair, or poor biological condition based on an index of benthic invertebrate assemblages. However, conditions are usually summarized at state, regional, or national scales, and do not provide insight into where within these regions these conditions occur. We used random forests to model and predict the probable condition of several million kilometers of streams across the conterminous US based on upstream and nearby landscape features, including human-related alterations to watersheds. This model used condition classes of streams that were assessed with a benthic invertebrate multimetric index as part of the 2008-2009 NRSA. We linked these sites to the US EPA’s StreamCat Dataset; a database of several hundred landscape metrics for all 1:100,000-scale streams and their associated watersheds within the conterminous US. The StreamCat data provided geospatial indicators of nearby and upstream land use, land cover, climate, and other landscape features for modeling. Nationally, the model correctly predicted the biological condition class of 78% of NRSA sites. Regional evaluations of the model suggested similar model performance across the Nation, i.e., there was no indication of large regional biases in model performance. Although model evaluations suggested good discrimination among condition classes, we present maps as predicted probabilities of good condition, given upstream and nearby landscape settings. Inversely, the maps can be interpreted as the probability of a stream being in poor condition, given human-related alterations to the landscape. Finally, we illustrate how these predictions can be used to prioritize streams for conservation and restoration.
