Excess nitrogen (N) from anthropogenic sources deteriorates freshwater resources. Actions taken to reduce N inputs to the biosphere often show no or only delayed effects in receiving surface waters hinting at large legacy N stores built up in the catchments soils and groundwater. Here, we quantify transport and retention of N in 238 Western European catchments by analyzing a unique data set of long-term N input and output time series. We find that half of the catchments exhibited peak transport times larger than five years with longer times being evident in catchments with high potential evapotranspiration and low precipitation seasonality. On average the catchments retained 72% of the N from diffuse sources with retention efficiency being specifically high in catchments with low discharge and thick, unconsolidated aquifers. The estimated transport time scales do not explain the observed N retention, suggesting a dominant role of biogeochemical legacy in the catchments’ soils rather than a legacy store in the groundwater. Future water quality management should account for the accumulated biogeochemical N legacy to avoid long-term leaching and water quality deteriorations for decades to come.

Elevated nitrate concentrations in German water bodies are a widespread problem, potentially resulting from a long history of excess nitrogen (N) inputs. Here, we investigated long-term (1950-2014) N dynamics across 89 German catchments using a process-based model. Results showed that the mean fractions of N surplus (excess) exported to the river, removed by denitrification, accumulated in the soil zone, and accumulated in groundwater across all catchments are 27%, 58%, 14%, and 1%, respectively. Dissolved inorganic N in groundwater could affect the stream N levels over decades as indicated by long groundwater transit times. A cluster identified four catchment groups with distinct archetypal long-term N transport and retention dynamics, which can be partly linked to the catchments’ topographic and geological conditions. This hints at underlying mechanisms that explain spatial differences in the fate of diffuse N inputs to catchments and opens the possibility for better-targeted management

Elevated nutrient inputs and reduced riverine concentration variability challenge the health and functioning of aquatic ecosystems. To improve riverine water quality management, it is necessary to understand the underlying biogeochemical and physical processes and their interactions at catchment scale. We hypothesize that spatial heterogeneity of nutrient sources dominantly controls the variability of instream concentrations among different catchments. Therefore, we investigated controls of mean nitrate (NO), phosphate (PO), and total organic carbon (TOC) concentrations and concentration-discharge (C-Q) relationships from observations in 787 German catchments covering a wide range of physiographic and anthropogenic settings. Using partial least square regressions and random forests we linked water quality metrics to catchment characteristics. We found archetypal C-Q patterns with enrichment dominating NO and TOC, and dilution dominating PO export. Across the catchments, we found a positive but heteroscedastic relation between mean NO concentrations and agricultural land use. We argue that denitrification, particularly pronounced in sedimentary aquifers, buffers high inputs and causes a decline in concentration with depth resulting in chemodynamic, strongly positive C-Q patterns. Consequently, chemodynamic NO enrichment patterns could indicate effective subsurface denitrification. Mean PO concentrations were related to point sources though the low predictive power suggests effects of unaccounted processes. In contrast, diffuse inputs better explained the spatial differences in PO C-Q patterns. TOC levels were positively linked to the abundance of riparian wetlands as well as negatively to NO concentrations suggesting interacting processes. This study shows that vertical concentration heterogeneity mainly drives nutrient export dynamics, partially modified by interactions with other controls.