Context is Everything: Interacting Inputs and Landscape Characteristics Control Stream Nitrogen.
Assessing the effectiveness of actions designed to improve water quality requires adequate monitoring of aquatic ecosystems paired with information about landscape and climate drivers. To understand spatial and temporal trends in stream total nitrogen (TN) concentrations across the conterminous US, we combined summer low flow TN samples from 5093 streams across three survey periods from the US EPA’s National Rivers and Streams Assessment (NRSA) with driver information. Watershed N inputs alone explained 51% of the variation in stream TN concentrations (log transformed), and the directional change in inputs matched the directional changes in stream TN in 47% of sites revisited between surveys. By including climate and additional landscape factors, random forest (RF) modeling explained 71% of variations in stream TN concentrations. RF models suggested that the dominance and direction of relationships between watershed drivers and stream TN concentration varied by N input rate. For example, wetlands appeared to be a TN source in low input watersheds and a sink in high N input watersheds, while base flow appeared to be a TN source in high input watersheds. Understanding of stream TN concentrations was improved by incorporation information about N inputs along with other watershed factors, land use and climate.