Excessive nitrate (NO3-) concentration in groundwater raises health and environmental issues that must be addressed by all European Union (EU) member states under the Nitrates Directive and the Water Framework Directive. The identification of NO3- sources is critical to efficiently control or reverse NO3- contamination that affects many aquifers. In that respect, the use of stable isotope ratios 15N/14N and 18O/16O in NO3- (expressed as δ15N-NO3- and δ18O-NO3-, respectively) has long shown its value. However, limitations exist in complex environments where multiple nitrogen (N) sources coexist. This two-year study explores a method for improved NO3- source investigation in a shallow unconfined aquifer with mixed N inputs and a long established NO3- problem. In this tillage-dominated area of free-draining soil and subsoil, suspected NO3- sources were diffuse applications of artificial fertiliser and organic point sources (septic tanks and farmyards). Bearing in mind that artificial diffuse sources were ubiquitous, groundwater samples were first classified according to a combination of two indicators relevant of point source contamination: presence/absence of organic point sources (i.e. septic tank and/or farmyard) near sampling wells and exceedance/non-exceedance of a contamination threshold value for sodium (Na+) in groundwater. This classification identified three contamination groups: agricultural diffuse source but no point source (D+P-), agricultural diffuse and point source (D+P+) and agricultural diffuse but point source occurrence ambiguous (D+P±). Thereafter δ15N-NO3- and δ18O-NO3- data were superimposed on the classification. As δ15N-NO3- was plotted against δ18O-NO3-, comparisons were made between the different contamination groups. Overall, both δ variables were significantly and positively correlated (p < 0.0001, rs = 0.599, slope of 0.5), which was indicative of denitrification. An inspection of the contamination groups revealed that denitrification did not occur in the absence of point source contamination (group D+P-). In fact, strong significant denitrification lines occurred only in the D+P+ and D+P± groups (p < 0.0001, rs > 0.6, 0.53 ≤ slope ≤ 0.76), i.e. where point source contamination was characterised or suspected. These lines originated from the 2-6‰ range for δ15N-NO3-, which suggests that i) NO3- contamination was dominated by an agricultural diffuse N source (most likely the large organic matter pool that has incorporated 15N-depleted nitrogen from artificial fertiliser in agricultural soils and whose nitrification is stimulated by ploughing and fertilisation) rather than point sources and ii) denitrification was possibly favoured by high dissolved organic content (DOC) from point sources. Combining contamination indicators and a large stable isotope dataset collected over a large study area could therefore improve our understanding of the NO3- contamination processes in groundwater for better land use management. We hypothesise that in future research, additional contamination indicators (e.g. pharmaceutical molecules) could also be combined to disentangle NO3- contamination from animal and human wastes.

At the catchment scale, a complex mosaic of environmental, hydrogeological and physicochemical characteristics combine to regulate the distribution of groundwater and stream nitrate (NO3-). The efficiency of NO3- removal (via denitrification) versus the ratio of accumulated reaction products, dinitrogen (excess N2) & nitrous oxide (N2O), remains poorly understood. Groundwater was investigated in two well drained agricultural catchments (10km2) in Ireland with contrasting subsurface lithologies (sandstone vs. slate) and landuse. Denitrification capacity was assessed by measuring concentration and distribution patterns of nitrogen (N) species, aquifer hydrogeochemistry, stable isotope signatures and aquifer hydraulic properties. A hierarchy of scale whereby physical factors including agronomy, water table elevation and permeability determined the hydrogeochemical signature of the aquifers was observed. This hydrogeochemical signature acted as the dominant control on denitrification reaction progress. High permeability, aerobic conditions and a lack of bacterial energy sources in the slate catchment resulted in low denitrification reaction progress (0-32%), high NO3- and comparatively low N2O emission factors (EF5g1). In the sandstone catchment denitrification progress ranged from 4 to 94% and was highly dependent on permeability, water table elevation, dissolved oxygen concentration solid phase bacterial energy sources. Denitrification of NO3- to N2 occurred in anaerobic conditions, while at intermediate dissolved oxygen; N2O was the dominant reaction product. EF5g1 (mean: 0.0018) in the denitrifying sandstone catchment was 32% less than the IPCC default. The denitrification observations across catchments were supported by stable isotope signatures. Stream NO3- occurrence was 32% lower in the sandstone catchment even though N loading was substantially higher than the slate catchment.