Variation in NO3 export from flowing waters of vastly different sizes:: Does one model fit all?

Flowing waters receive nitrogen (N) from the surrounding watershed and ultimately export much of this N to coastal waters, which in turn can be substantially affected by these inputs. Although the control of N export is complex, for large rivers among-system variation is predicted relatively well by simple models of human activity. Using data from 249 predominantly North Temperate watersheds that varied in size from 0.1 to over 1,000,000 km(2), we examined whether these simple models lose their predictive power at smaller scales. We found that the relationship between human population density and NO3 export becomes weaker at smaller scales, and that for watersheds less than 100 km(2), it explains only 8% of the 1000-fold variation in NO3 export. However, NO3 export predicted from a simple loading model related well to measured NO3 export across all scales; linear regressions of log modeled versus log measured export for small (less than 100 km(2)), mid-sized (100-10,000 km(2)), and large (more than 10,000 km(2)) watersheds were all highly significant (P < 0.01) and had r(2) values of 0.78, 0.63, and 0.77, respectively. For the smallest systems, however, the model was biased and predicted higher NO3 export than was measured. The bias suggests slightly greater storage or gaseous N loss in smaller watersheds, whereas the tight correlation between predicted and measured export indicates that for small as well as large systems, among-system variation in NO3 export is controlled primarily by anthropogenic N loads rather than site-specific variations in soil or vegetation characteristics. Across all scales, however, predictive models can be improved by the inclusion of these local parameters.