Landscape controls on dissolved nutrients, organic matter and major ions in a suburbanizing watershed

Understanding the relative importance of anthropogenic and natural landscape features that drive spatial variability in water quality is a central challenge in studying the biogeochemistry of heterogeneous landscapes. We quantified the average annual flux and concentration of dissolved inorganic nitrogen (DIN), dissolved organic nitrogen (DON), dissolved organic carbon (DOC), phosphate-P (PO4-P), sodium (Na+) and chloride (Cl-) at ~40 stream sites in three major (51 to 903 km2) NH basins. We used GIS to quantify anthropogenic (e.g. human population density, % impervious surface cover and % agriculture) and natural (e.g. % forest, % wetlands and soil C:N) landscape features for each sub-basin and then employed multiple-regression analysis to relate water quality parameters to landscape characteristics. Anthropogenic features were strong predictors of DIN flux and Na+ and Cl- concentrations, whereas wetland cover (a natural feature) was a significant, but weak predictor of DOC (r2=0.26, p<0.01) and DON (r2 = 0.14, p<0.05) flux. Anthropogenic features could not explain a significant amount of variance in DON or DOC flux. Mean PO4-P concentrations were surprisingly low (<0.015 mg P/L) when compared to the larger range in mean DIN concentrations (0.03 to 0.96 mg/L) and consequently no landscape characteristics could explain a significant amount of spatial variability in PO4-P flux or concentration. Human population density was the single best predictor of DIN flux (r2=0.76, p<0.01), and together with % impervious surface and % agriculture explained 86% (p<0.01) of the total variance. Among all sites, % road pavement was a strong predictor of stream Na+ and Cl- concentrations (r2 = 0.75 to 0.78, p<0.01) and % impervious surface was a stronger predictor (r2 = 0.86 to 0.92, p<0.01) among a subset of sites. Our results suggest that DIN and DON result from different sources in the landscape and although sources of DON and DOC are similar, DON and DOC concentrations respond differently to hydrologic events. Concentrations of DON were negatively correlated with discharge (r2= 0.06, p < 0.01), but DOC concentrations did not respond to discharge, suggesting there is much to learn about the decoupling of C and N rich dissolved organic matter (DOM). Our landscape models do not account for spatial variability in dissolved inorganic carbon (DIC), dissolved organic P (DOP) or particulate C, N and P; however, at our site with the longest record, the Lamprey River, DIC and particulate C account for only 25% and 5% respectively of the total C flux. Among three sites ranging from 3-16% development, particulate N only accounted for 12-18% of total N, but particulate P accounted for 30-50% and DOP accounted for an additional 40-50% of total P. Landscape characteristics may in fact be better predictors of spatial variability in particulate-P and DOP than PO4-P. Nonetheless, the landscape models that we have developed for DIN and DOM flux can be used by local watershed managers to predict changes in water quality in response to changes in land use.