Variable response in riverine nitrogen exports to long-term temporal changes in watershed inputs

Aquatic nitrogen (N) pollution is a global problem, and understanding where best to implement mitigation strategies to reduce transfers from land to water is becoming increasingly urgent. Watershed nutrient budgets, using commonly available census data, can successfully predict riverine export spatially, albeit with variation related to fractional export. Furthermore, when using long-term data, watershed input-export trajectories vary and can be decoupled. Quantifying which drivers influence fractional export among trajectory types has yet to be explored. Here, we quantified the input-export trajectories of 33 subwatersheds (>25 years) in the St. Lawrence River, Québec, Canada, and assessed how fractional export was associated with land use, climate, and landscape features per trajectory type. Fractional N export ranged from 0.12 to 0.77 (median 0.29) and increased in many subwatersheds with higher input intensities. Annual runoff was associated with fractional export in watersheds with tracking input-export trajectories, and in those where N inputs decreased then increased overtime (nonlinear) while export was stable or increasing. However, shifts in N input components were also associated with fractional export, including in watersheds with decreasing input-increasing export trajectories often associated with legacy N. Higher fractional export was also associated with the runoff ratio in some trajectories likely through increased hydroconnectivity whereas lentification by lakes and dams reduced it in others. Our results suggest that beyond reducing total N inputs, understanding shifts toward leakier N input components as well as considering individual watershed features that govern N transport and retention are also required to minimise aquatic pollution.