Main content area

Linking terrestrial phosphorus inputs to riverine export across the United States

Metson, Genevieve S., Lin, Jiajia, Harrison, John A., Compton, Jana E.
Water research 2017 v.124 pp. 177-191
agricultural land, algal blooms, aquatic ecosystems, crops, data collection, detergents, eutrophication, fertilizers, humans, hypoxia, mechanistic models, monitoring, phosphorus, pollution load, prediction, runoff, sewage, variance, watersheds, waterways, United States
Humans have greatly accelerated phosphorus (P) flows from land to aquatic ecosystems, causing eutrophication, harmful algal blooms, and hypoxia. A variety of statistical and mechanistic models have been used to explore the relationship between P management on land and P losses to waterways, but our ability to predict P losses from watersheds often relies on small scale catchment studies, where detailed measurements can be made, or global scale models that that are often too coarse-scaled to be used directly in the management decision-making process. Here we constructed spatially explicit datasets of terrestrial P inputs and outputs across the conterminous U.S. (CONUS) for 2012. We use this dataset to improve understanding of P sources and balances at the national scale and to investigate whether well-standardized input data at the continental scale can be used to improve predictions of hydrologic P export from watersheds across the U.S. We estimate that in 2012 agricultural lands received 0.19 Tg more P as fertilizer and confined manure than was harvested in major crops. Approximately 0.06 Tg P was lost to waterways as sewage and detergent nationally based on per capita loads in 2012. We compared two approaches for calculating non-agricultural P waste export to waterways, and found that estimates based on per capita P loads from sewage and detergent were 50% greater than Discharge Monitoring Report Pollutant Loading Tool. This suggests that the tool is likely underestimating P export in waste the CONUS scale. TP and DIP concentrations and TP yields were generally correlated more strongly with runoff than with P inputs or P balances, but even the relationships between runoff and P export were weak. Including P inputs as independent variables increased the predictive capacity of the best-fit models by at least 20%, but together inputs and runoff explained 40% of the variance in P concentration and 46–54% of the variance in P yield. By developing and applying a high-resolution P budget for the CONUS this study confirms that both hydrology and P inputs and sinks play important roles in aquatic P loading across a wide range of environments.