Evaluation of spatial complexity to improve predictions of nutrient export across watersheds in New York State

Water quality impairment from excess nutrient loading disrupts aquatic ecosystems, recreational activities, and potable water supplies. To better manage land-based loading, there is a strong interest in identifying whether a disproportionate fraction of nutrient loads can be attributed to only a small portion of the watershed. Comparing across 50 watersheds, we evaluated the relationship between nutrient export and anthropogenic land cover as a function of increased spatial complexity. Three key nutrients, total phosphorus, total nitrogen, and soluble reactive phosphorus, in river discharge were examined. Concentration-discharge (CQ) data from over 50 watersheds with long-term discharge records in New York State were used to calculate three different nutrient export metrics: median concentration, CQ curve slope, and annual area weighted load. Median concentration was the export metric most readily predicted with watershed characteristics, while CQ slope was difficult to predict. Aggregate, watershed-scale characteristics were better predictors of export rather than predictors that assessed only features within stream riparian areas or at the intersection of select land cover, soil, and slope features. Our results build upon a growing body of literature that suggests increased spatial complexity does not greatly enhance predictions of watershed-scale nutrient export.