Excessive nitrogen (N) inputs to coastal waters can lead to severe eutrophication and different chemical forms of N exhibit varying levels of effectiveness in fueling primary production. Efforts to mitigate N fluxes from coastal watersheds are often guided by models that predict changes in N loads as a function of changes in land use, management practices, and climate. However, relatively little is known on the impacts of such changes on the relative fractions of different N forms. We leveraged a long-term dataset of N loads from over 100 river stations to investigate how the  fraction, that is, the ratio of  to total N (/TN), changes as a function of spatio-temporal changes in TN loads in the Chesapeake Bay watershed. We built a hierarchical model that separates the response of  to changes in TN load occurring at different scales: Across river stations, where differences in TN loads are largely driven by spatial differences in anthropogenic inputs, and within stations, where inter-annual variability in hydrology is a key driver of changes in TN loads. Results suggest that while increases in TN loads resulting from changes in anthropogenic inputs lead to an increase in the  fraction, a decrease in the  fraction may occur when increases in TN loads are driven by increased streamflow. These results are especially relevant in watersheds that may experience changes in N loads due to both management decisions and climate-driven changes in hydrology.