An isotope mixing-based clustering approach for an improved apportionment of nitrate sources in groundwater systems

Several groundwater nitrate (NO3−) studies have focused on apportioning NO3− sources for various sets of groundwater samples statistically clustered (i.e., grouped) based on hydrochemical data. This can be useful when NO3− concentrations in groundwater systems are influenced by the aquifer aqueous geochemistry. Nonetheless, several studies observed independent behavior of NO3− reflected by its insignificant correlation with most major ions in groundwater. Therefore, this study introduces a novel clustering approach using NO3− concentrations and δ15NNO3 values for classifying samples controlled by common mixing of NO3−-containing groundwaters. This clustering, constituting a first step of an approach allowing an improved apportionment of NO3− sources by MixSIAR modeling, is followed by (ii) NO3− source identification based not only on δ15NNO3 and δ18ONO3 tracers, but also on δ11B, which is particularly able to distinguish manure from sewage; (iii) constraining the model by selected NO3− sources and site-specific isotope fractionation effects; and (vi) a separate NO3− sources apportionment processing step for different clustered groundwater samples. The approach is applied to a dataset from a coastal Mediterranean agricultural area affected by several sources of NO3−. Two mixing scenarios with independent clusters were established, and the most realistic one that integrates four different clusters was retained for the subsequent steps. Mixing in three clusters is controlled by two NO3− sources (i.e., manure/sewage for two clusters, and manure/synthetic fertilizers for one cluster), whereas the fourth cluster is governed by mixing of three NO3− sources (i.e., manure, sewage, and synthetic fertilizers). MixSIAR modeling reveals manure as the major source contributing NO3− to groundwater according to the four clusters, but with different contributions varying from 61 to 71 %; consistent with differences in land use of the study area. Ultimately, the approach presented in this study provides a process to implement selective NO3− mitigation strategies, adapted to discriminated sites, rather than a general NO3− mitigation strategy applied to the whole study area.