Long-term riverine nitrogen dynamics reveal the efficacy of water pollution control strategies

•WRTDS model had satisfactory accuracies for predicting long-term riverine N dynamics.•Riverine NH4+ level decreased since 2010 due to enhanced sewage treatment.•Riverine TN and NO3− levels continuously increased in 1980–2019 due to legacy effects.•Point source pollution control efforts demonstrated improved effectiveness.•Nonpoint source pollution control requires additional efforts for legacy N issues. Identification of long-term water quality trends in response to watershed anthropogenic interventions is crucial for developing and adapting water pollution control strategies. This study represents the first use of the Weighted Regressions on Time, Discharge, and Season (WRTDS) model to evaluate trends and sources of riverine nitrogen (N) levels over the 1980–2019 period in the Yongan River watershed of eastern China. The WRTDS model showed satisfactory accuracies for predicting daily riverine total N (TN), NH4+ and NO3− concentrations/loads (R2 > 0.55, n = 366). Modeled flow-normalized riverine NH4+ concentration increased by 789% from 1980 to 2009 and then decreased by 63% in 2010–2019. This changing trend for riverine NH4+ concentration was mainly attributed to a 43% decrease of wastewater NH4+ discharge load in 2010–2019 due to establishment of three new WWTPs in urban areas and enhanced rural domestic sewage collection/treatment. Although chemical N fertilizer use decreased by 49% and domestic animal numbers decreased by 73% in 2000–2019, flow-normalized riverine TN and NO3− concentrations progressively increased by 161% and 232% in 1980–2019, respectively. The paradox between decreasing N inputs and increasing riverine TN/NO3− concentrations is attributed to inputs of legacy N from soil and groundwater. This is supported by the 3.8-fold increase of riverine NO3− concentration in 1980–2019 (86% increase in 2000–2019) following 10-days with no-precipitation (representing groundwater contributions to baseflow) and a 4.1-fold increase of riverine NO3− concentration in 1980–2019 (91% increase in 2000–2019) following the first rainstorm after 10-days of no-precipitation (representing soil flushing). These results document that point-source pollution control efforts were effective, whereas benefits from nonpoint-source pollution control were masked by inputs from legacy N pollution. The WRTDS model was demonstrated to be a useful tool for assessing long-term riverine N pollution dynamics and sources, thereby providing decision-makers with critical informat