Nitrogen Legacies in Agricultural Landscapes: A 150-year Longitudinal Study of the Susquehanna and Mississippi River Basins

Abstract:

Global flows of reactive nitrogen (N) have increased significantly over the last century in response to land-use change, agricultural intensification and elevated levels of atmospheric N. Although the use of commercial N fertilizers began to plateau in developed countries in the mid-1980s and despite widespread implementation of a range of conservation measures to mitigate the impacts of N-intensive agriculture, N concentrations in surface waters are in many cases remaining steady or continuing to increase. This lack of correlation between N inputs and outputs is increasingly being attributed to the presence of legacy N stores in subsurface reservoirs, with present-day concentrations being a function of inputs that are many decades old. It has remained unclear, however, what the magnitudes of such stores might be, and how they are partitioned between soil and groundwater reservoirs.

In the present work, we have synthesized agricultural, population, and land-use data to develop a comprehensive, 150-year dataset of N inputs to the land surface of the continental United States. We have concurrently developed a parsimonious, process-based model that utilizes this N input trajectory to simulate biogeochemical transformations of N along subsurface pathways. Model results allow us predict the magnitudes of legacy N in soil and groundwater pools and to predict long-term stream N concentration trajectories over the last century and into the future. We have applied this modeling approach to two U.S. watersheds, the Mississippi River and Susquehanna River Basins, which are major sources of nutrient contamination to the Gulf of Mexico and Chesapeake Bay, respectively. Our results show significant stream N loading above baseline levels in both watersheds before the widespread use of commercial N fertilizers, largely due to 19^th-century conversion of natural forest and grassland areas to row-crop agriculture. However, the temporal patterns of this loading differ between the two watersheds due to differences in trajectories of land-use change. Using the model, we estimate spatiotemporal patterns of N accumulation in both groundwater and soil organic matter in response to increases in N inputs to agricultural soil and explore future scenarios to predict changes in N-loading as a function of these N legacies.