Using generalized additive models to investigate drivers and controls on stream water nitrate concentrations in four agricultural catchments

Nitrogen (N) is essential for agricultural production and additional inputs of N-containing mineral or organic fertilizers as part of modern agricultural practice followed by the subsequent loss of surplus N has led to a deterioration in water quality globally. In order to design effective mitigation measures and accurately assess progress towards meeting water quality improvement goals, an understanding of the processes that govern N loss is prerequisite. However, the complexity which governs N cycling in agricultural catchments and the timescales over which they occur make understanding the relative importance of the drivers and controls of N loss challenging. Here, we used an eight-year dataset to investigate stream water nitrate concentrations across four catchments with contrasting characteristics where agriculture accounted for greater than 95 % of the land use. We subdivided each catchment into subcatchments and investigated trends in stream water nitrate concentrations using explanatory variables representing farming intensity, land management, climatic conditions, and soil drainage. We adopted a systematic approach using generalized additive mixed models (GAMM) to capture complex relationships between explanatory variables and nitrate concentrations within each catchment and across all catchments. We found no clear relationship between source N loadings and stream water nitrate concentrations, with the most likely explanation for this being that N cycling in the study sites were transport- and not source-limited. Our results highlighted the key role played by climate and the hydrological characteristics of catchments driving N loss from agricultural catchments. We also found that site specific characteristics mediated the relative importance of the drivers and controls of N losses which suggests that effective mitigation measures should be determined by hydraulic properties of (sub)catchments. • Generalized additive mixed models were used to investigate N cycling • Agricultural intensity, land management, climate and soil drainage were considered • Climatic forcing was found to exhibit a key control on N cycling • The impact of climatic forcing was mediated by (sub)catchment characteristics • Mitigation strategies should consider climate and (sub)catchments characteristics