1. INTRODUCTION

Nitrate contamination of drinking water sources is a serious global problem (Denk et al., 2017; Torres-Martínez et al., 2021). The World Health Organization stipulates that the nitrate nitrogen content in drinking water should not exceed 10 mg/L (WHO,2017), and if the nitrate content in drinking water exceeds the standard, it can easily lead to diseases such as methemoglobinemia in infants and children and gastric cancer in adults (Nestler et al., 2011; Bu et al., 2017). In addition, high concentrations of nitrate in surface waters can lead to eutrophication (Bailey et al., 2016). There are two predominant sources of nitrogen pollution in aquatic environments: natural and anthropogenic. The natural source originates from atmospheric deposition and soil organic nitrogen, whereas the anthropogenic source is the result of human activities, including the use of fertilizers, domestic sewage discharge, livestock manure, industrial effluent, and leaching from solid waste. The crucial to the study of nitrate pollution is to identify the contribution of human activities. Nitrogen undergoes a variety of physical, biological, and chemical reactions from pollution sources to water, and accurately tracing the source of nitrogen remains a scientific challenge when it is transported and transformed in the environment (Stoliker et al., 2016; Wang et al., 2020; Cao et al., 2021).
After decades of development, great progress has been made in the research on the source of nitrate pollution and transport and transformation patterns in surface water and groundwater. As early as the 1960s, there was research on nitrate pollution in the world. Early studies generally chose hydrochemical and statistical methods. Nitrogen isotopes were first applied to quantify the contributions of nitrogen pollution sources in 1971, but the results were questioned due to the neglect of factors such as isotope fractionation (Kohl et al., 1971). In order to distinguish the sources of overlapping nitrate pollution areas, Amberger first measured the oxygen isotopes in nitrate (Amberger et al., 1987). The introduction of the Bayesian isotope mixing models has solved the problem of determining the contribution of multiple potential nitrogen sources to nitrate pollution in water bodies. Commonly used models include SIAR (Parnell et al., 2010), MixSIAR (Moore et al., 2008), among others. In the first application of the SIAR model to the source apportionment of nitrate in water, Xue quantified the contributions of atmospheric deposition, soil nitrogen, nitrate fertilizers, ammonium fertilizers, and fecal and domestic wastewater to nitrate pollution and its contribution rates (Xue et al., 2012). Since then, a large number of studies have started to use hydrochemical combined with nitrogen and oxygen isotopes to quantify the sources of nitrate in watershed water (Zhang et al., 2018; Yi et al., 2020; Guo et al., 2022). Furthermore, some studies have attempted to improve the accuracy of quantifying nitrate source apportionment, including the introduction of multiple isotopes such as 34S and13C (Torres-Martínez et al., 2020; Ren et al., 2022). In addition, uncertainty analysis of isotope mixing models has gradually begun to be applied (Ji et al., 2017; Zhang et al., 2018; Shang et al., 2020; Ji et al., 2022). The analytical process described above is well established for analyzing the nitrate sources of one water body alone. However, in a watershed scale, surface water and groundwater typically interact, and focusing solely on one water body may lead to substantial inaccuracies. Therefore, a comprehensive approach that takes into account both surface water and groundwater is necessary. Additionally, financial constraints often result in relatively low sample densities when conducting research at the watershed scale. It is particularly important to maximize the accuracy of identifying nitrate sources in a watershed using limited resources.
The study area of this research, the Chuzhou Hushan Hydrological Experimental Watershed, is located in the southwestern part of the Jianghuai Hilly area, and is a typical agricultural small watershed. It was established as one of the primary hydrological experimental watersheds in China based on the International Hydrological Programme (IHP) after the country’s founding. This study aims to: (1) quantify the contributions of different sources of nitrate using a Bayesian isotope mixing model, and (2) quantify the uncertainty of the model results and analyze the impact of input variability on the model outputs. The findings provide scientific guidance for controlling nitrate pollution in the watershed.