Predictive mapping of the biotic condition of conterminous U.S. rivers and streams

摘要: Understanding and mapping the spatial variation in stream biological condition could provide an important tool for conservation, assessment, restoration of ecosystems. The USEPA's 2008–2009 National Rivers Streams Assessment (NRSA) summarizes percentage lengths within conterminous United States that are good, fair, or poor based on a multimetric index benthic invertebrate assemblages. However, is usually summarized at regional national scales, these assessments do not substantial insight into distribution conditions unsampled locations. We used random forests to model predict probable several million kilometers streams across nearby upstream landscape features, including human-related alterations watersheds. To so, we linked NRSA sample sites StreamCat Dataset; database hundred metrics all 1:100,000-scale their associated watersheds States. data provided geospatial indicators land use, cover, climate, other features modeling. Nationally, correctly predicted class 75% sites. Although evaluations suggested good discrimination among classes, present maps as probabilities condition, given settings. Inversely, can be interpreted probability being watershed alterations. These predictions available download from website. Finally, illustrate how prioritize conservation restoration.