Quantitative Identification of Nitrate Sources in a Coastal Peri-Urban Watershed Using Hydrogeochemical Indicators and Dual Isot

Chemosphere 243 (2020) 125364

Contents lists available at ScienceDirect

Chemosphere

journal homepage: www.elsevier.com/locate/chemosphere

Quantitative identiﬁcation of nitrate sources in a coastal peri-urban watershed using hydrogeochemical indicators and dual isotopes together with the statistical approaches

* Zhaofeng Guo a, b, Changzhou Yan a, , Zaosheng Wang a, Feifei Xu a, Fan Yang a a Key Laboratory of Urban Environment and Health, Fujian Key Laboratory of Watershed Ecology, Institute of Urban Environment, Chinese Academy of Sciences, Xiamen, 361021, China b University of Chinese Academy of Sciences, Beijing, 100049, China highlights graphical abstract

MixSIAR model revealed no signiﬁ- cant seasonal difference in NO3 pollution. Reveal the degree and difference of water pollution by using PCA in peri- urban area. Relationships between lower d15N values and nutrients implied nitriﬁ- cation occurred. Manure and sewage were the main NO3 sources of Jimei Lake.

article info abstract

Article history: Surplus nutrient load and complex migration and transformation processes are the challenges for water Received 2 June 2019 quality management in the peri-urban coastal watershed, leading to increasing concerns worldwide. We Received in revised form investigated the spatio-temporal variation of hydrogeochemical parameters in surface water of Jimei 11 October 2019 Lake watershed, and distinguished the sources and transformation of nitrate-N (NO3 -N) using dual Accepted 12 November 2019 isotopes of nitrate (d15N and d18OinNO) with hydrogeochemical indicators. Principal component Available online 16 November 2019 3 analysis (PCA) on hydrogeochemical parameters demonstrated that surface water was seriously polluted Handling Editor: Hyunook Kim by nutrients, especially in the southeast of the downstream. There were signs of seawater intrusion and increased wastewater discharge in the mid-lower reaches with high ammonium concentrations. Nitri- 15 18 Keywords: ﬁcation occurred throughout the monitoring period with lower d N and d O values and NO3 derived Nitrate isotope from mixed pollution sources. Results of Bayesian model showed that dominant NO3 input originated Spatio-temporal variations from manure and sewage (M&S, 71% and 76% in the wet and dry season, respectively) and atmospheric Multivariate statistical analysis deposition (22% and 16%, respectively). This result implied that the controls and treatment of M&S Bayesian model discharges are essential to alleviate of NO3 pollution. The proposed method is helpful to understand the Source apportionment origins of NO3 and may be suitable to develop measures for the reducing of nitrogen loadings in the peri- urban watershed. © 2019 Elsevier Ltd. All rights reserved.

1. Introduction

* Corresponding author. As main input channels of nutrients and chemicals, coastal E-mail address: [email protected] (C. Yan). https://doi.org/10.1016/j.chemosphere.2019.125364 0045-6535/© 2019 Elsevier Ltd. All rights reserved. 2 Z. Guo et al. / Chemosphere 243 (2020) 125364

watersheds have an important influence on the water quality of fertilizers and possible sources of atmospheric NO3 and human/ coastal sea area, and are vulnerable to land use and population animal waste based on NO3 isotopes and chemical indicators (e.g., change. According to statistics, 27% of the coastal watersheds in chloride, sodium, fluoride, sulfate, atrazine). To better evaluate the China did not meet the Class IV water quality standards (Environ- contamination contribution of each nitrogen source, source mental Quality Standards for Surface Water of China, GB3838- apportionment models such as IsoSource, mass-balance mixing 2002), and the main pollutant exceeding the standard was inor- model, and Bayesian model have been applied (Deutsch et al., ganic nitrogen, resulting in over 51% of the sea areas in various 2006; Hale et al., 2014; Nie et al., 2012; Yang and Toor, 2017). degrees of eutrophication (Bulletin on China’s Marine Ecological Compared with other models, MixSIAR model, which applied Environment in 2018). NO3 pollution in coastal environment is Marchain Monte Carlo method to estimate the probability distri- mainly caused by intensive anthropogenic activities (e.g., sewage bution, can be better considered the uncertainties (seasonal vari- discharge from production and living, synthetic fertilizers and ation of isotopes, multiple sources > 3, and isotope fractionation) manure use, and excess nitrogen deposition due to the air pollu- (Parnell et al., 2010). tion) and continuous in-depth change in land use, which has Jimei Lake watershed in southeast China is an important coastal become a widespread environmental issue (Kaushal et al., 2011; area with municipal water supply, landscape water, agricultural Zhang et al., 2015). Elevated NO3 concentrations in water bodies, irrigation and flood control functions. As a typical peri-urban area, especially in coastal and marine environments, cause eutrophica- Jimei Lake watershed is disturbed by intensive human activities. As tion, harmful algal blooms, and hypoxia, ultimately resulting in a a result, Jimei Lake is beset by serious eutrophication. The security series of harmful knock-on effects (Dai et al., 2008; Gutierrez et al., guarantee of water environment in studied areas is causing great 2018). Consequently, to ensure the safety of water quality, it is public concern. Source apportionment of NO3 pollution and nu- essential to take effective measures to reduce inputs of excessive trients discharge is the primary attention for integrative gover- nitrogen. The accurate identification of the main NO3 contributors nance actions. Due to the complex land-use types in peri-urban and its transformations are critical prerequisite for efficient water areas, the source apportionment remains challenging with the quality management. diversification of pollution sources. Locally, the most important To date, existing investigations have frequently been done about contribution of NO3 may be from the production and domestic identification of NO3 sources in groundwater or surface water ac- wastewater, because of the increasing urbanization rate. In addi- 15 18 cording to the typical d N and d O values of NO3 sources (Hale tion, the monthly average of NO3 concentrations in rainwater et al., 2014; Johannsen et al., 2008; Kreitler and Browning, 1983; collected from Xiamen reached 1.84 mg L 1 (Wu et al., 2018), which Pasten-Zapata et al., 2014; Wassenaar, 1995). The successful appli- suggested that nitrogen deposition from the atmospheric system is cation of the isotopic technique for identification of contamination also a major source of nitrogen pollution in Xiamen’s water envi- sources is due to the fact that some NO3 sources exhibit distinctive ronment. In recent years, although the use of agricultural chemical isotopic compositions. Yet, the initial isotope abundance can be fertilizers under the government control has reduced by nearly half greatly altered by isotopic fractionation during biogeochemical since 2010 (Website: http://www.xm.gov.cn/zwgk/tqjj/xmjjtqnj/), transformation (Kendall et al., 2007). Thus, in addition to deter- the amount of nitrogen fertilizer (e.g., ammonium bicarbonate, mining the origins, recently, the transformation pathways of NO3 urea and nitrate fertilizer), phosphate fertilizer (e.g., calcium su- such as denitrification, nitrification, assimilation and mineraliza- perphosphate and calcium magnesium phosphate), potassium tion have been differentiated through the dual isotope model fertilizer, and compound fertilizer still reached 2905, 1396, 2407, (Adebowale et al., 2019; Einsiedl and Mayer, 2006; Mayer and and 5669 tons respectively in 2017. Besides, NO3 pollution caused Wassenaar, 2012; Yi et al., 2017). by organic fertilizer, such as poultry manure, used in agriculture A more bothersome challenge is the overlapping source isotope and aquaculture cannot be neglected. The contribution of other compositions in complex systems, making it difficult to provide nitrogen sources to this watershed remains to be further examined. specific sources using NO3 isotopes. Recent researches have already Here, we conducted an integrated research to investigate the argued that the integration of stable isotopes with hydro- spatio-temporal patterns and source apportionment of NO3 geochemical parameters is regarded as an effective and screening pollution in Jimei Lake watershed by coupling hydrogeochemical tool to reduce the uncertainties in tracing sources of NO3 in water. indicators and dual isotopes together with the statistical ap- Source identification with hydrogeochemistry is of interest proaches. Firstly, the spatio-temporal variations of hydro- considering that besides NO3 , relevant nitrogen sources contain geochemical parameters were evaluated in the study watershed. other characteristic components, which will inevitably change the Secondly, combining principal component analysis (PCA) and content of these components in the environment and have an multiple hydrogeochemical indicators provided preliminary infor- impact on the water quality. In particular, nitrogen sources and mation on potential NO3 sources. Then, the spatio-temporal vari- transformation of NO3 as well as equilibrium fractionations of ations of NO3 sources and related transformations were further isotopic fractionation (Fenech et al., 2012) were closely associated distinguished by combining land-use type and dual stable isotopes. with land use and physicochemical properties of the hydrographic Finally, the quantitative contributions of contamination sources environment. The complexity of anthropogenic sources and envi- were assessed with Bayesian mixing model. ronmental impact factors is reflected in the spatial and temporal diversity of river pollutants. Thus, the high-frequency investigation 2. Materials and methods and screening of relevant of hydrogeochemical factors in study areas are required for analysis of contamination sources and 2.1. Site description and sampling campaigns transport of NO3 in water. For instance, the variation of NO3 /Cl ratio versus Cl concentration was used to interpret different This investigation was conducted in Jimei Lake watershed 0 0 0 0 sources of NO3 (rainfall, synthetic fertilizer, manure and sewage) (24 34 -24 40 N, 117 56 -118 05 E), which is situated in Xiamen (Meghdadi and Javar, 2018). The sewage and manure were enriched City, Fujian Province, southeastern coast of China (Fig. 1). Driven by 2 in Cl and SO4 , and chemical fertilizers were used and increased urban expansion, development of township enterprises, Jimei 2 þ 2 þ SO4 export, resulting in the ratio of Cl /Na and SO4 /Na higher District is characterized by typical peri-urban areas. Jimei Lake than 1 (Zhang et al., 2018). Panno et al. (2001) found that the NO3 watershed originates from the Laoliao Mountains in northwest contamination sources in karst spring was dominated by nitrogen Jimei District and joins the right tributary (Xuxi River) before it Z. Guo et al. / Chemosphere 243 (2020) 125364 3

Fig. 1. Sketch map of Jimei Lake watershed in Xiamen City with 24 sample locations. The Xuxi (X1-X5, n ¼ 5) is the largest tributary of Jimei Lake watershed; the Zhuxi (Z1-Z3, n ¼ 3) is in the upstream of the study area; the Houxi (H1eH2, n ¼ 2) is in the midstream; the Xinglin Bay (XL1-XL14, n ¼ 14) is in the lower reach of the Jimei Lake watershed. eventually enters Jimei Lake (the Xinglin Bay estuary). It has a experimental analysis. Simultaneously, the pH, dissolved oxygen drainage area of 209.3 km2 and contains a population of about (DO), electrical conductivity (EC), water temperature, salinity and 57,801. The average annual precipitation in the study area is total dissolved solids (TDS) were directly measured in situ using a 1478.8 mm, and specifical information concerning rainfall is shown portable multi-parameter water quality meter (HACH DS5X, USA). in (Fig. S1). Jimei Lake watershed is commonly grouped into four sections: the Xuxi tributary (X1-X5), the upper section (Zhuxi River, Z1-Z3), the middle section (Houxi River, H1eH2) and the lower 2.2. Chemical and stable isotope analysis section, Jimei Lake (also named Xinglin Bay, XL1-XL14), which is surrounded by the seawall as a reservoir. In Jimei Lake watershed, Concentrations of TN, total dissolved nitrogen (TDN), total the land use classes were: forest (40%), waters (14%), residential phosphorus (TP), and solution reactive phosphorus (SRP) in the land (13%), industrial area (10%), transportation (7%), cultivated surface water were determined by standard spectrophotometric land (5%) and grassland (2%). The investigated area is dominated by methods described by (Zhang et al., 2017) with an ultraviolet and granite and volcanics. Soil types are mainly lateritic red soil, red visible spectrophotometer (UV 2450, Shimadzu, Japan). Total soil, saline paddy soil, paddy soil, and coastal saline soil. The Xuxi organic carbon (TOC) was measured using a TOC analyzer (TOC- tributary and the upper reaches are located in the vicinity of mixed VCPH, Shimadzu, Japan), while the chlorophyll-a (chl-a) was areas of forested land, agricultural land, and scattered villages. The analyzed using a PHYTO-PAM Phytoplankton Analyzer (Heinz Walz 2 middle and lower reaches are located in urbanized areas. In addi- GmbH, Eichenring, Germany). Ion concentrations (SO4 ,Cl , fluo- þ þ tion, Jimei Lake also plays a significant role in aquaculture and ride (F ), NO3 , nitrite (NO2 ), ammonium (NH4 ), sodium (Na ), þ þ þ landscape zoning in Xiamen promontory. potassium (K ), calcium (Ca ), and magnesium (Mg2 )) were Considering the diverse types of existing potential pollution determined by ion chromatography (Dionex ICS-3000, USA). sources, land use patterns, and watershed characteristics, we Filtered water with a 0.45 mm membrane filter was acidified (HNO3, selected twenty-four sampling locations covering the whole 1:1 v: v) for the measurement of Boron (B) by the Inductively watershed and recorded their location with a hand-held posi- Coupled Plasma Mass Spectrometer (ICP-MS, Agilent 7500cx, USA). tioning system (GPS). Seven sampling campaigns were undertaken Water samples and rainwater were filtered with 0.45 mm in the wet season (May, July, August, and September) and dry membrane filters prior to isotope analysis. The chemical conversion 15 - 18 - season (January, March, and November) of 2018, respectively. Wa- method for determination of d NeNO3 and d OeNO3 in water ter samples at 0.5 m depth under the surface were collected with was modified from Mcllvin and Altabet (2005). Concretely, filtered 1 L pre-washed polyethylene bottles and stored at 4 C before water samples (pH ¼ 6e8) were transferred into a headspace vial, in which, NO3 was firstly reduced to NO2 under a sufficient 4 Z. Guo et al. / Chemosphere 243 (2020) 125364

oscillation for 20 min by adding 0.8 mL of 20 g L 1 cadmium 3. Results and discussion chloride, 0.8 mL of 250 g L 1 ammonium chloride and zinc plate in turn; subsequently 2 mL of sodium azide buffer solution (2 M NaN3: 3.1. Spatio-temporal variation of hydrogeochemical parameters 20% CH3COOH; 1:1 v: v) was injected into the produced NO2 , so that 15 18 the NO2 was converted to N2O. d N and d O values of N2Owere The concentration distribution of hydrogeochemical parameters analyzed by an Isotope Ratio Mass Spectrometer (IRMS, Thermo in Jimei Lake watershed during the wet season (May to September) Fisher MAT 253) equipped with a GasBench ІІ device (Thermo and the dry season (November to March) were investigated in this Fisher) at the Third Institute of Oceanography, Ministry of Natural study (Table S2). Cl varied between 11.54 and 10066.32 mg L 1 2 1 Resources. The stable isotopic rates (R) were expressed as d nota- (average of 2403.62), and SO4 were 6.75e3353.19 mg L (average þ tion and a per mil (‰): of 331.16). As for Na , it changed between 8.77 and 4808.91 mg L 1 þ . (average of 982.77). Other cations varied comparatively little: K 1 2þ dð = Þ¼ ranged from 3.29 to 130.30 mg L (average of 36.32), Ca ranged 0 00 Rsample Rs tan dard 1 1000 (1) þ from 2.05 to 605.54 mg L 1 (average of 109.78), and Mg2 ranged from 11.67 to 298.42 mg L 1 (average of 73.06). Boron concentra- where R and R are the heavy to the light isotope ratios sample standard tions varied between 1.86 and 1879.20 mgL 1, with an average of sample and standard, respectively. For N and O, the reference value of 453.49 mgL 1. Boron concentrations at downstream sites standard is atmospheric nitrogen and Vienna standard mean ocean (XL5-XL14) were all higher than 500 mgL 1 (WHO, 2011) exceeding water (VeSNOW), respectively. The international references, the given guideline value and threatening human health. In gen- including USGS32 (d15N ¼þ180‰; d18O ¼þ25.7‰) and air V-SMOW eral, boron concentration is relatively low in natural surface water, USGS34 (d15N ¼1.8‰; d18O ¼27.9‰), were applied to air V-SMOW and mainly depended on the geochemical nature, anthropogenic correct the detected values of d15N and d18O. Additionally, the input (agricultural runoff, industrial and domestic sewage precision of d15N and d18O analysis was 0.3‰ and 0.5‰, discharge) and seawater intrusion (Voutsa et al., 2009). The boron respectively. concentration in the lower reaches was abnormally high compared with the upper and middle reaches, but it was not higher as the concentration in the seawall, indicating that the sources of boron in the lower reaches simultaneously influenced by seawater intrusion 2.3. Bayesian mixing models: MixSIAR and wastewater discharge. DO varied from 3.50 to 8.10 mg L 1 (average of 6.16) in the wet season, and from 2.34 to 7.64 mg L 1 The proportional contribution of different contamination sour- (average of 5.23) in the dry season. From the spatial perspective, ces to NO3 in different seasons was evaluated by applying a higher concentrations of anions and cations in this area were Bayesian mixing model, which was implemented by the Stable mostly located in the lower reaches, resulting in EC values above Isotope Analysis in R graphical user interface package (MixSIAR 1000 mscm 1 at these points. In this study, the high standard de- version 3.1.9) (Parnell et al., 2013; Stock and Semmens, 2016). viation values revealed that most of the hydrogeochemical pa- 15 18 The dual-isotope values (d N and d OofNO3 ) and four po- rameters exhibited a greatly spatial distribution, which was caused tential NO3 sources (atmospheric deposition, synthetic fertilizer, not only by the difference in pollution discharge, but also by the soil nitrogen, and manure and sewage) in this study were used to seawater intrusion at the downstream sampling sites with higher quantify the contributions of NO3 in the Jimei Lake watershed Parnell et al., 2013 salinity (average of 6.00‰). No statistically sig- during the two seasons. In the Bayesian mixing model, measured nificant (p > 0.05) differences were observed in these hydro- isotope values were designated as the “consumers”, and the mean geochemical parameters for the two seasons. and standard deviation obtained from other literature were With the aim to explore whether the coastal watershed at risk of “ ” designated as sources (Table S1). The transformation processes seawater intrusion is suitable for irrigationpffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi purpose, we used so- such as denitrification and nitrification influencing isotopic char- dium absorption ratio (SAR ¼ Naþ= ðCa2þ þ Mg2þÞ=2; Richards, þ acteristic values were discussed in this study (see Section 3.3). 1954) and sodium percentage (Na% ¼ Na 100/ þ þ þ þ (Ca2 þ Mg2 þ Na þ K ); Richards, 1954) and the classifications of the United States Department of Agriculture for further comprehensive analysis. The agricultural fields around the up- 2.4. Multivariable statistical analysis stream had good irrigation water, whereas except for the upstream part, other samples were suffered from high salinity-hazard prob- Multivariate statistical approaches were applied to sufficiently lems (Fig. S2). Worse still, high sodium-hazard also observed in mine the potential information from the experimental data sets to lower reaches indicated that these rivers are unsuitable for irriga- provide more detailed interpretation pertaining to the relationship tion because high salt content increases soil solution osmotic among the variables and sampling sites (Matiatos et al., 2014). pressure and slows plant growth (Srivastava, 2019). Overall, the Normality and homogeneity were examined using Shapiro-Wilk surface water in the tributaries and mid-lower reaches applied for and Levene test, respectively. The spatio-temporal differences of irrigation can be mixed with fresh water to reduce its salinity below variables were statistically compared by using ANOVA (p < 0.05). the threshold. When normality or homogeneity assumptions were not satisfied, Concentrations of NO3eN showed not seasonal significant the non-parametric Kruskal-Wallis test was required. Spearman’s (p > 0.05) difference, ranging from 0.47 to 2.65 mg L 1 in the wet rank correlation coefficients were calculated to analyze the rela- season and from 0.55 to 3.77 mg L 1 in the dry season, with a high tionship among hydrogeochemical parameters and isotope values coefficient of variation during the wet season (47%) and the dry with 0.01 and 0.05 significance level. Based on the correlation and season (49%). Spatially, the NO3eN concentrations presented a relative significance of variables, principal component analysis firstly increased and then decreased trend from the upper reaches (PCA) was performed to summary orthogonal factors and to extract to the lower reaches in both seasons, with a higher value in the principal components. IBM SPSS (version 24.0) and OriginPro tributary (2.04e3.77 mg L 1) than that in the mainstream 1 (version 9.0, OriginLab) were performed for statistical analyses and (0.47e2.47 mg L )(Fig. 2). In addition, NO3eN concentrations scientific mapping respectively. were significantly (p ¼ 0.0001) higher in the tributary and Z. Guo et al. / Chemosphere 243 (2020) 125364 5

Fig. 2. The variations of NO3eN, NH4eN, and TN concentrations during the wet season (May, July, August, and September) and dry season (January, March, and November) of 2018. Boxplots illustrate the 25th, 50th, 75th percentiles; the black dot represents the mean concentration. upstream, and significantly (p < 0.001) lower in the mid-lower 3.2. Using hydrogeochemical parameters identify origin reaches as compared to NH4eN concentrations, which could be information on anthropogenic influence explained by the elevated NO3eN concentrations due to the agricultural activities in the tributary and upstream. By contrast, mean Based on the Spearman’s correlation analysis of the hydro- NH4eN concentrations in the mid-lower reaches ranged between geochemical parameters, we found that significant positive re- 1 1 2 þ 2.11 and 19.81 mg L with an average of 6.92 mg L , far exceeding lationships were observed among dissolved ions (Cl ,SO4 ,Na , þ þ þ the water quality standard (class V, 2 mg L 1). Similarly, the severity K ,Ca2 , and Mg2 ) and boron in water (Table S3). As is known, of the TN pollution was also pronounced as TN content these chemical variables have been used as suitable indicators for (2.02e25.58 mg L 1) of all samples exceeded 2 mg L 1 (class V, anthropogenic influences. In particular, the Cl is considered as a GB3838-2002). Our data also showed that NO3eN and NH4eN were biogeochemical inactive yet valuable marker in tracking sources the predominant nitrogen forms at all sampling sites, accounting because previous studies have found that the animal/human for 57e84% of TN with a mean value of 70%. The results indicated sewage and fertilizers (potash or KCl) are rich in high levels of Cl þ that the impact factors of water eutrophication in Jimei Lake could (Minet et al., 2017). The mean Na /Cl ratios (Fig. S3) for most be attributed to the high concentration of dissolved inorganic ni- water samples in the tributary and middle-upper reaches were trogen, especially the NO3eN and NH4eN. Individually, the most located in the lower right of 1:1 line (halite dissolution). Combined severe nitrogen contamination occurred in downstream with dense with land-use and soil types, Cl may be introduced from fertil- residential communities and industries (site XL1, XL5, and XL9), izers, manure or sewage contamination in the study area. However, þ which had a higher proportion of NH4eN to TN. Overall, the the Na /Cl ratios from the upstream (Z1 and Z2) samples in both NO3eN, NH4eN, and TN concentrations for most sampling sites seasons and the tributary (X2) samples in the wet season fell þ were lower during the wet season than those in the dry season. slightly above the halite dissolution line, suggesting additional Na There are several reasons for such phenomena, such as dilution inputs and possibly due to the application of Na-based fertilizers effect and denitrification accompanying with atmospheric precip- (Ogrinc et al., 2019) and a small amount of halite dissolution. Be- þ itation in the wet season (Ding et al., 2014), and the release of sides, elevated Na concentrations may also be related to the nutrients into water bodies due to the decomposition of aquatic external inputs of organic effluents (Robertson et al., 1991), which þ plant residue during the dry season (Correll, 1998). could be proved by the positive correlation between Na and TOC þ (r ¼ 0.718, p < 0.0001). Although the Na and Cl concentrations in tributaries and middle-upper reaches were below the guideline þ levels recommended by the WHO (Na : 200 mg L 1,Cl: 250 mg L 1), there was insufficient evidence that the pollution in 6 Z. Guo et al. / Chemosphere 243 (2020) 125364 relation to non-nitrogen waste effluents and fertilizers had no impact on this river, as the concentrations at these sampling sites were more than twice that of site Z1 as the background value. The Cl contents at all water samples in the lower reaches were much þ þ higher than Na contents, and Na /Cl ratios fell at the lower right of 1:1 line. In addition, EC, TDS, and salinity were positively correlated with aforesaid parameters, which justified our result that the salinization phenomenon existed in downstream waters due to the seawater intrusion. Because boron was usually considered as an important tracer for wastewater and sewage, strong positive relationship between boron and Cl also indicated that industrial and municipal effluents were the origin of the Cl . Interestingly though, there were several stronger negative corre- lations between NO3 and aforementioned ions. These relationships were unexpected because NO3 concentrations are supposed to increase with these increasing anthropogenic ions. During this study, NO3 /Cl ratios were used to distinguish whether NO3 derived from agricultural inputs, soil nitrogen or sewage inputs (Liu et al., 2006). There was a negative correlation (r ¼0.734, Fig. 3. Principal component analysis (PCA) of 18 variables in water samples from Jimei p < 0.0001) between NO3 and Cl , which could be due to the Lake watershed. The lines and arrows represent the PC1 and PC2 loading of each mixing process affected the transformation of NO3 .NO3 /Cl molar variable (arrows labeled with variables symbol). The dots represent the PC1 and PC2 ratio versus Cl molar concentration showed the NO3 in part of scores for each sampling site. sampling sites, such as sites of Z1, Z2, X3, and X4, were influenced by soil nitrogen, with low NO3 /Cl ratios and Cl concentrations (Fig. S3). Whereas some of the sampling sites in the midstream and highest scores of PC1 were located in the inner bay, possibly because the concentration of salts in the coastal sampling sites was tributary exhibited a higher Cl concentrations but lower NO3 /Cl diluted by the surrounding sewage discharge. Then, PC2 accounted ratios indicating the potential sources of NO3 could be associated with manure and sewage. In the downstream, an extremely low for 25.91% of the total variance, exhibiting strong positive loading (0.79) on NH4eN, TN, SRP, TP, and TOC, as well as a strongly NO3 /Cl ratios and very high Cl concentration were observed. All negative loading on DO (0.845), which could be regarded as the ratios of NO3 /Cl were less than 1, which could not explain the contribution of agricultural inputs sources. Additionally, the lower contribution from nutrient inputs and organic pollutants. Spatial distribution of PC2 scores (Fig. 4b) showed that significantly higher NO3 /Cl ratios in the mid-lower reaches than that in the upstream could be attributed to the urbanization and industrialization scores values were located at sites XL1, XL9, XL4, and XL5, and these development and the overload discharge of municipal wastewater sampling sites were close to the residential land. While the PC2 in this areas (Li et al., 2019). Note that the salinity was strongly yet scores from the inner bay presented the lowest values in sites of XL8, XL11, and XL14. The PC2 score values of sampling sites located negatively correlated with NO3 (r ¼0.699, p ¼ 0.0001) and an increase in the salinity from freshwater to estuarine sites was in agricultural areas were relatively small compared with that in found, especially in the downstream, showing a conservative the mid-lower reaches. Hence, the sources of nutrient inputs and mixing behavior of freshwater and seawater in the downstream organic pollutants were dominated by manure and sewage. (Archana et al., 2018). Therefore, the lower reaches were affected by both seawater intrusion and untreated water effluents. Unlike NO3 , 3.3. Investigating the origin and transformation of nitrate using NH4eN positively correlated with the above-mentioned ions dual stable isotope (r > 0.4, p < 0.05) and salinity (r ¼ 0.475, p ¼ 0.019). This is because NH4eN was the dominant species of TN in contaminated waters. In Along with hydrogeochemical tracers, the isotopic compositions þ addition, as reported by Jani and Toor (2018),NH4 in the high saline can further identify the source and transformation of nitrogen 15 waters might be replaced by the salts from the anionic sites, so the contamination. During the wet season, measured d NeNO3 values contents of NH4eN in the estuarine waters will increases. for the studied surface water ranged from 5.57‰ to 6.08‰ 18 In order to further explore the relationships among hydro- (average of 1.96‰) and d OeNO3 ranged from 0.96‰ to 24.30‰ 15 geochemical parameters and the hot spot areas severely contami- (average of 7.36‰)(Fig. 5). The d NeNO3 was 0.37‰~4.28‰ 18 nated within Jimei Lake watershed, PCA was created, which (average of 2.19‰) and d OeNO3 was 1.81‰~13.11‰ (average of presented two considered components (Eigenvalues > 1) that 2.57‰) in the dry season. Although there was no distinct seasonal 15 described approximately 86% of the total variance. Based on Fig. 3, difference (p > 0.05) in the values of d NeNO3, it existed spatial we observed that PC1 explained 59.77% of the total variance, variability (p < 0.05) across the different sampling sites, whereas þ þ 2þ 2þ 18 dominated by Cl ,Na ,K ,Ca ,Mg , boron, TDS, salinity, EC, and d OeNO3 was just the opposite. Spatial patterns (Fig. 6a and c) 15 pH, all with positive loading values > 0.86. In light of the above showed that the d NeNO3 values in the tributary and upstream hydrogeochemical parameters analysis, it can be concluded that characterized by low density residential and agricultural river did PC1 seemed to mainly represent anthropogenic influences, and not drop as expected compared to those in the middle and down- water salinization attributed to the mixing effect of freshwater and stream, reflecting that the fertilizer and soil contributed less to the 15 seawater. Furthermore, according to the spatial differences of PC1 nitrate sources. The highest value of d NeNO3 was observed in the scores shown in Figs. 3 and 4a, we noticed that the sample scores at downstream sample (XL5) during the wet season and in the trib- the downstream were dramatically higher than that those at the utary sample (X1) during the dry season. Moreover, we found that, 15 mid-upper reaches and tributaries, but the scores did not increase regardless of the season, relatively low d NeNO3 values in the gradually as the distance to the seawall decreased. Especially in the midstream and the inner bay where there was high flow transect. 15 upper bay, this further confirmed additional anthropogenic input On the whole, the d NeNO3 values at all sampling sites were low from the surrounding area. In addition, the sampling sites with the and varied in a narrow range, indicating that there may be nitrate Z. Guo et al. / Chemosphere 243 (2020) 125364 7

Fig. 4. Spatial distribution of PC1 (a) and PC2 (b) scores for each monitoring stations within Jimei Lake watershed.

IN was almost double that in 2012 (Wu et al., 2018). As an important source of nitrogen pollution in our studied areas, atmospheric nitrogen mainly derived from fuel combustion and automobile 18 exhaust (Butler et al., 2005; Kirchner et al., 2005). d OeNO3 was 2 þ þ 2þ positively correlated with dissolve ions (Cl ,SO4 ,Na ,K ,Ca , þ Mg2 , and boron), giving proof that seawater mixing affected the isotopic values. Based on these nitrate isotopes signatures, we noted that the 15 18 d NeNO3 and d OeNO3 values of many surface water sample in Fig. 5 overlapped across three different nitrate sources: soil nitrogen, inorganic fertilizer, and manure/sewage. Furthermore, the above data interpreted that the contribution of atmospheric deposition to the nitrate sources should also be taken into account in our studied waters. In view of the existence of multi-source uncertainties, the Bayesian isotope mixing model (MixSIAR) was used to quantitatively assess the proportional contributions of the above four potential nitrate sources. Modeling outputs illustrated that the contributions from manure and sewage (M&S) were 15 18 Fig. 5. Plot of d NeNO3 versus d OeNO3 in 24 sampling locations during the maximal with an average proportion of 71% in the wet season and monitoring period. The end member values of potential NO3 sources are captured from 76% in the dry season, and then followed by atmospheric precipi- Kendall et al. (2007) and Liu et al. (2018). tation (AP) (22% and 16% for the wet and dry seasons, respectively), whereas the contributions from synthetic fertilizers and soil nitrogen (SN) to NO3 in aquatic systems was extremely low, with an transformations (nitrification, nitrogen fixation) or other sources average values of 3% and 4%, respectively (Fig. 7). Hence, the NO causing this scenario. By contrast, the d18OeNO values varied 3 3 pollution of Jimei Lake watershed was predominately derived from widely with a complex temporal variation (Fig. 6b and d), because 18 anthropogenic emissions and nitrogen deposition, which accoun- the microbial transformation process occurred. The d OeNO 3 ted for 89e96% of NO contribution. Except for AP, the contribution values in the wet season were higher than those in the dry season. 3 of M&S, SF, and SN during the wet season were slightly lower than Compared with the dry period, we believe that the mixed dilution that in the dry season, elucidating that precipitation to a certain effect of atmospheric deposition is the reason for the lower NO eN 3 extent has an effect on the mixed dilution of NO pollution sources. concentrations and higher d18OeNO values in the wet season. In 3 3 On the other hand, due to the large proportion from M&S, the Xiamen, annual inorganic nitrogen (IN) deposition in 2016 was seasonal variation of the contributions was not obvious. During the 36.45 kg N ha 1 in town and 35.92 kg N ha 1 in suburb, accounting wet season, the contributions of M&S were higher in the tributary for 7.1e13.3% of IN loading of the Xiamen Bay, and the total influx of 8 Z. Guo et al. / Chemosphere 243 (2020) 125364

15 18 Fig. 6. Geographical plot showing the spatial distribution of d NeNO3 and d OeNO3 in the wet season (a, and b) and dry season (c, and d) for Jimei Lake watershed. and upstream than that in the mid-down reaches, indicating that as green belt sprinkling (Website: http://www.xm.gov.cn/zfxxgk/ nonpoint source pollution from untreated wastewater has a greater xxgkznml/gmzgan/tjnj/). Domestic sewage discharged from tribu- impact on the rural areas than urban areas. According to statistics, tary was the highest, about 8.58 million tons, including COD 679.94 the total amount of domestic sewage discharged from Jimei Lake is tons yr 1, TN 282.29 tons yr 1 and TP 12.67 tons yr 1 from raw 10.06 million tons, only about half of which was treated by pipes, sewage. According to the compilation of environmental statistics in and the partially or completely treated efﬂuents were mainly used Xiamen City (2016), TN ﬂowed into the river from domestic sewage Z. Guo et al. / Chemosphere 243 (2020) 125364 9

Fig. 7. Fractional contributions of four potential NO3 sources estimated by MixSIAR in the Jimei Watershed during the wet season (a) and dry season (b).

1 with 1535 tons yr , followed by agricultural and industrial efflu- predominant transformation process of NO3 in the surface water, ents sources (543.8 and 62.16 tons yr 1, respectively). Therefore, TN and whether exactly the assimilation process took place remains to pollution from the effluents input was major from domestic sewage be studied. sources. The dual-isotope signals of NO3 are not only applied to trace 4. Conclusions various sources, but can help identify the transformation process (Li fi et al., 2010). In the hypoxic condition, microbial denitri cation The present study reported the results on the status of water preferentially consumes the light isotopes and thus makes the 15N 18 quality and source appointment of nitrogen in Jimei Lake water- and O enriched (Xue et al., 2009). If present, there is a linear shed. Data showed that the mean concentrations of TN and TP in d15 e d18 e trajectory between N NO3 and O NO3, with slopes from surface water were above the class V standards of surface water 1.3:1 to 2.1:1 (Aravena and Robertson, 1998; Liu et al., 2006). (GB3838-2002, China). Through the application of multivariate Further, the decrease of NO3 is accompanied by the increase of statistical methods (PCA and correlation analysis), it was found that d15 e d18 e þ þ þ þ þ N NO3 and O NO3, and it also demonstrated the occurrence pollutants (Cl ,SO2 ,Na ,K ,Ca ,Mg2 ,NH , TP, and boron) and fi 4 4 of denitri cation (Lehmann et al., 2003). Our data showed that physical parameters (EC, salinity, and TDS) in the downstream are e d15 e ¼ NO3 N was positively correlated with N NO3 (r 0.453, significantly higher than those in other reaches because of seawater ¼ fi p 0.026), but signi cantly and negatively correlated with intrusion and anthropogenic input such as manure and sewage d18 e ¼ ¼ O NO3 (r 0.633, p 0.001). Whereas, no correlation was (M&S) discharge. In addition, the problems of high salinity-hazard d15 e d18 e ¼ > found between N NO3 and O NO3 (r 0.081, p 0.05) and sodium-hazard in the downstream make it difficult to achieve fi with a slope of 0.127, illustrating that no denitri cation took irrigation purposes. The observed hydrogeochemical data and dual place. This was further confirmed by the fact that DO contents were 1 isotope values indicated that NO3 derived from mixed pollution above 2 mg L at all water samples and was not conducive to sources, and the dominant nitrogen transformation processes were fi e denitri cation. As the NO3 N concentrations decreased, the Chl-a nitrification without microbial denitrification. The results of d18 e concentrations and O NO3 values increased from upstream to Bayesian mixing model manifested that there was a relatively downstream (Fig. S4), and the high Chl-a concentrations and constant yet high contribution of M&S (71% and 76% in the wet d18 e O NO3 values mainly appeared in the downstream (especially in season and dry season, respectively) to Jimei Lake watershed, fol- the southeast of Xinglin Bay), which may be a sign of phytoplankton lowed by nitrogen deposition (22% and 16% in the wet season and assimilation. However, we could not observe a similar positive dry season, respectively), soil nitrogen, and synthetic fertilizers. d15 e correlation between Chl-a and N NO3, which implied that other The high proportion contribution of M&S during two monitoring fl transformation processes corporately in uencing the isotopic seasons has resulted in no significant seasonal and spatial varia- e compositions occurred. The DO was related with negative NH4 N tions in the river, so reducing the output of M&S to Jimei Lake ¼ < d15 e ¼ (r 0.881, p 0.0001) and positive N NO3 (r 0.266, watershed is the key to pollution control. > fi p 0.05), revealing nitri cation by microbial activity was occurring Combining hydrogeochemical indicators and dual isotopes e e 14 since it could convert NH4 NtoNO3 N and accumulate N. The together with the statistical approaches might provide us a useful d15 e low N NO3 values of the water samples during the two periods screening tool to reduce the uncertainties in tracing sources of NO fi 3 further indicated that the effect of nitri cation process on nitrogen contamination in a complex peri-urban coastal watershed. The cycle could not be neglected. Besides that, Ji et al. (2017) found that results from our study can serve as a reference by decision makers d18 fi the calculated theoretical O values originated from nitri cation to remediate and mitigate of NO contamination in surface water ‰ ‰ 3 should vary between 0.05 and 7.77 , and most of the with a reliable approach. To further reduce the uncertainty of d18 measured O values in our study fell within this typical range. source apportionment, in the future study, the ranges of isotopic Overall, the microbial denitrification obviously had a minor influ- signatures from the main NO3 sources and isotope fractionation ence on the dual isotopic values, while nitrification was the factor should be detected to calculate the contribution of NO3 10 Z. Guo et al. / Chemosphere 243 (2020) 125364

e sources more accurately. The NO3 sources in wastewater is further Environ. Sci. Technol. 48, 6211 6219. https://doi.org/10.1021/es501039t. refined by combining the isotope method with organic pollutants Jani, J., Toor, G.S., 2018. Composition, sources, and bioavailability of nitrogen in a longitudinal gradient from freshwater to estuarine waters. Water Res. 137, such as food additives and pharmaceuticals. In addition, long-term 344e354. https://doi.org/10.1016/j.watres.2018.02.042. monitoring of groundwater and surface water is conducive to a Ji, X.L., Xie, R.T., Hao, Y., Lu, J., 2017. Quantitative identification of nitrate pollution more systematic understanding of the nitrogen cycling in this sources and uncertainty analysis based on dual isotope approach in an agricultural watershed. Environ. Pollut. 229, 586e594. https://doi.org/10.1016/ region. j.envpol.2017.06.100. Johannsen, A., Dahnke,€ K., Emeis, K., 2008. Isotopic composition of nitrate in five Declaration of competing interests German rivers discharging into the North Sea. Org (Tokyo. 1967). Geochem 39, 1678e1689. https://doi.org/10.1016/j.orggeochem.2008.03.004. Kaushal, S.S., Groffman, P.M., Band, L.E., Elliott, E.M., Shields, C.A., Kendall, C., 2011. The authors declare that there is no known competing financial Tracking nonpoint source nitrogen pollution in human-impacted watersheds. interests or personal relationships that could have appeared to Environ. Sci. Technol. 45, 8225e8232. https://doi.org/10.1021/es200779e. fl Kendall, C., Elliott, E.M., Wankel, S.D., 2007. Tracing anthropogenic inputs of ni- in uence the work reported in this paper. trogen to ecosystems. Stable Isot. Ecol. Environ. Sci. 375e449. https://doi.org/ 10.1002/9780470691854.ch12. Second Edition. Acknowledgments Kirchner, M., Jakobi, G., Feicht, E., Bernhardt, M., Fischer, A., 2005. Elevated NH3 and NO2 air concentrations and nitrogen deposition rates in the vicinity of a highway in Southern Bavaria. Atmos. Environ. Times 39, 4531e4542. https:// This project was supported by the Xiamen Jimei Environmental doi.org/10.1016/j.atmosenv.2005.03.052. Protection Bureau for providing research materials and funding, Kreitler, C.W., Browning, L.A., 1983. Nitrogen-isotope analysis of groundwater nitrate in carbonate aquifers: natural sources versus human pollution. J. Hydrol and was also partly funded by the Strategic Priority Research Pro- 61, 285e301. https://doi.org/10.1016/0022-1694(83)90254-8. gram of the Chinese Academy of Sciences (Grant No. Lehmann, M.F., Reichert, P., Bernasconi, S.M., Barbieri, A., McKenzie, J.A., 2003. XDA23030203), and the National Natural Science Foundation of Modelling nitrogen and oxygen isotope fractionation during denitrification in a lacustrine redox-transition zone. Geochem. Cosmochim. 67, 2529e2542. China (NSFC) (21377125). We are grateful for providing us with the https://doi.org/10.1016/S0016-7037(03)00085-1. sampler boat by the Kenfair Co., Ltd. of Xiamen. The authors would Li, C., Li, S.L., Yue, F.J., Liu, J., Zhong, J., Yan, Z.F., Zhang, R.C., Wang, Z.J., Xu, S., 2019. like to express their gratitude for these supports to implement this Identification of sources and transformations of nitrate in the Xijiang River e research. using nitrate isotopes and Bayesian model. Sci. Total Environ. 646, 801 810. https://doi.org/10.1016/j.scitotenv.2018.07.345. Li, S.L., Liu, C.Q., Li, J., Liu, X.L., Chetelat, B., Wang, B.L., Wang, F.S., 2010. Assessment Appendix A. Supplementary data of the sources of nitrate in the Changjiang river, China using a nitrogen and oxygen isotopic approach. Environ. Sci. Technol. 44, 1573e1578. https://doi.org/ 10.1021/es902670n. Supplementary data to this article can be found online at Liu, C.Q., Li, S.L., Lang, Y.C., Xiao, H.Y., 2006. Using d15N- and d18O-values to identify https://doi.org/10.1016/j.chemosphere.2019.125364. nitrate sources in karst ground water, Guiyang, southwest China. Environ. Sci. Technol. 40, 6928e6933. https://doi.org/10.1021/es0610129. Liu, J., Shen, Z.Y., Yan, T.Z., Yang, Y.C., 2018. Source identification and impact of References landscape pattern on riverine nitrogen pollution in a typical urbanized watershed, Beijing, China. Sci. Total Environ. 628e629, 1296e1307. https://doi.org/ Adebowale, T., Surapaneni, A., Faulkner, D., McCance, W., Wang, S., Currell, M., 2019. 10.1016/j.scitotenv.2018.02.161. Delineation of contaminant sources and denitrification using isotopes of nitrate Matiatos, I., Alexopoulos, A., Godelitsas, A., 2014. Multivariate statistical analysis of near a wastewater treatment plant in peri-urban settings. Sci. Total Environ. the hydrogeochemical and isotopic composition of the groundwater resources 651, 2701e2711. https://doi.org/10.1016/j.scitotenv.2018.10.146. in northeastern Peloponnesus (Greece). Sci. Total Environ. 476e477, 577e590. Aravena, R., Robertson, W.D., 1998. Use of multiple isotope tracers to evaluate https://doi.org/10.1016/j.scitotenv.2014.01.042. denitrifi cation in ground water: study of nitrate from a large-flux septic system Mayer, B., Wassenaar, L.I., 2012. Isotopic characterization of nitrate sources and plume. Gr. Water 36, 975e982. https://doi.org/10.1016/0014-5793(83)80292-0. transformations in Lake Winnipeg and its contributing rivers, Manitoba, Can- Archana, A., Thibodeau, B., Geeraert, N., Xu, M.N., Kao, S.J., Baker, D.M., 2018. Ni- ada. J. Gt. Lakes Res. 38, 135e146. https://doi.org/10.1016/j.jglr.2012.02.004. trogen sources and cycling revealed by dual isotopes of nitrate in a complex Mcllvin, M.R., Altabet, M.A., 2005. Chemical conversion of nitrate and nitrite to urbanized environment. Water Res. 142, 459e470. https://doi.org/10.1016/ nitrous oxide for nitrogen and oxygen isotopic analysis in freshwater and j.watres.2018.06.004. seawater. Anal. Chem. 77, 5589e5595. https://doi.org/10.1021/ac050528s. Butler, T., Likens, G., Vermeylen, F., Stunder, B., 2005. The impact of changing ni- Meghdadi, A., Javar, N., 2018. Quantification of spatial and seasonal variations in the trogen oxide emissions on wet and dry nitrogen deposition in the northeastern proportional contribution of nitrate sources using a multi-isotope approach and USA. Atmos. Environ. Times 39, 4851e4862. https://doi.org/10.1016/ Bayesian isotope mixing model. Environ. Pollut. 235, 207e222. https://doi.org/ j.atmosenv.2005.04.031. 10.1016/j.envpol.2017.12.078. Correll, D.L., 1998. The role of phosphorus in the eutrophication receiving waters: a Minet, E.P., Goodhue, R., Meier-Augenstein, W., Kalin, R.M., Fenton, O., review. J. Environ. Qual. 27, 261e266. https://doi.org/10.2134/ Richards, K.G., Coxon, C.E., 2017. Combining stable isotopes with contamination jeq1998.00472425002700020004x. indicators: a method for improved investigation of nitrate sources and dy- Dai, M.H., Wang, L.F., Guo, X.H., Zhai, W.D., Li, Q., He, B.Y., Kao, S.J., 2008. Nitrification namics in aquifers with mixed nitrogen inputs. Water Res. 124, 85e96. https:// and inorganic nitrogen distribution in a large perturbed river/estuarine system: doi.org/10.1016/j.watres.2017.07.041. the pearl river estuary, China. Biogeosciences 5, 1227e1244. https://doi.org/ Nie, Y.P., Chen, H.S., Wang, K.L., Yang, J., 2012. Water source utilization by woody 10.5194/bg-5-1227-2008. plants growing on dolomite outcrops and nearby soils during dry seasons in Deutsch, B., Mewes, M., Liskow, I., Voss, M., 2006. Quantification of diffuse nitrate karst region of Southwest China. J. Hydrol 420e421, 264e274. https://doi.org/ inputs into a small river system using stable isotopes of oxygen and nitrogen in 10.1016/j.jhydrol.2011.12.011. nitrate. Org. Geochem. 37, 1333e1342. https://doi.org/10.1016/ Ogrinc, N., Tamse, S., Zavadlav, S., Vrzel, J., Jin, L.X., 2019. Evaluation of geochemical j.orggeochem.2006.04.012. processes and nitrate pollution sources at the Ljubljansko polje aquifer Ding, J.T., Xi, B.D., Gao, R.T., He, L.S., Liu, H.L., Dai, X.L., Yu, Y.J., 2014. Identifying (Slovenia): a stable isotope perspective. Sci. Total Environ. 646, 1588e1600. diffused nitrate sources in a stream in an agricultural field using a dual isotopic https://doi.org/10.1016/j.scitotenv.2018.07.245. approach. Sci. Total Environ. 484, 10e18. https://doi.org/10.1016/ Panno, S.V., Hackley, K.C., Hwang, H.H., Kelly, W.R., 2001. Determination of the j.scitotenv.2014.03.018. sources of nitrate contamination in karst springs using isotopic and chemical Einsiedl, F., Mayer, B., 2006. Hydrodynamic and microbial processes controlling indicators. Chem. Geol. 179, 113e128. https://doi.org/10.1016/S0009-2541(01) nitrate in a fissured-porous karst aquifer of the franconian alb, southern Ger- 00318-7. many. Environ. Sci. Technol. 40, 6697e6702. https://doi.org/10.1021/es061129x. Parnell, A.C., Inger, R., Bearhop, S., Jackson, A.L., 2010. Source partitioning using Fenech, C., Rock, L., Nolan, K., Tobin, J., Morrissey, A., 2012. The potential for a suite stable isotopes: coping with too much variation. PLoS One 5, e9672. https:// of isotope and chemical markers to differentiate sources of nitrate contami- doi.org/10.1371/journal.pone.0009672. nation: a review. Water Res. 46, 2023e2041. https://doi.org/10.1016/ Parnell, A.C., Phillips, D.L., Bearhop, S., Semmens, B.X., Ward, E.J., Moore, J.W., j.watres.2012.01.044. Jackson, A.L., Grey, J., Kelly, D.J., Inger, R., 2013. Bayesian Stable Isotope Mixing Gutierrez, M., Biagioni, R.N., Alarcon-Herrera, M.T., Rivas-Lucero, B.A., 2018. An Models. Environmetrics 24, 387e399. https://doi.org/10.1002/env.2221. overview of nitrate sources and operating processes in arid and semiarid Pasten-Zapata, E., Ledesma-Ruiz, R., Harter, T., Ramirez, A.I., Mahlknecht, J., 2014. aquifer systems. Sci. Total Environ. 624, 1513e1522. https://doi.org/10.1016/ Assessment of sources and fate of nitrate in shallow groundwater of an agri- j.scitotenv.2017.12.252. cultural area by using a multi-tracer approach. Sci. Total Environ. 470e471, Hale, R.L., Turnbull, L., Earl, S., Grimm, N., Riha, K., Michalski, G., Lohse, K.A., 855e864. https://doi.org/10.1016/j.scitotenv.2013.10.043. Childers, D., 2014. Sources and transport of nitrogen in arid urban watersheds. Richards, L.A., 1954. Diagnosis and improvement of saline and alkali soils. US Dep. Z. Guo et al. / Chemosphere 243 (2020) 125364 11

Agric. https://doi.org/10.1126/science.120.3124.800. Handbook 60. Xue, D., Botte, J., De Baets, B., Accoe, F., Nestler, A., Taylor, P., Van Cleemput, O., Robertson, W.D., Cherry, J.A., Sudicky, E.A., 1991. Ground-water contamination from Berglund, M., Boeckx, P., 2009. Present limitations and future prospects of stable two small septic systems on sand aquifers. Gr. Water 29, 82e92. https://doi.org/ isotope methods for nitrate source identiﬁcation in surface and groundwater. 10.1111/j.1745-6584.1991.tb00500.x. Water Res. 43, 1159e1170. https://doi.org/10.1016/j.watres.2008.12.048. Stock, B.C., Semmens, B.X., 2016. MixSIAR GUI user manual. https://github.com/ Yang, Y.Y., Toor, G.S., 2017. Sources and mechanisms of nitrate and orthophosphate brianstock/MixSIAR. transport in urban stormwater runoff from residential catchments. Water Res. Srivastava, S.K., 2019. Assessment of groundwater quality for the suitability of 112, 176e184. https://doi.org/10.1016/j.watres.2017.01.039. irrigation and its impacts on crop yields in the Guna district, India. Agr. Water Yi, Q.T., Chen, Q.W., Hu, L.M., Shi, W.Q., 2017. Tracking nitrogen sources, trans- Manag. 216, 224e241. https://doi.org/10.1016/j.agwat.2019.02.005. formation, and transport at a basin scale with complex plain river networks. Voutsa, D., Dotsika, E., Kouras, A., Poutoukis, D., Kouimtzis, T., 2009. Study on dis- Environ. Sci. Technol. 51, 5396e5403. https://doi.org/10.1021/acs.est.6b06278. tribution and origin of boron in groundwater in the area of Chalkidiki, Northern Zhang, Q., Sun, J., Liu, J., Huang, G., Lu, C., Zhang, Y., 2015. Driving mechanism and Greece by employing chemical and isotopic tracers. J. Hazard Mater. 172, sources of groundwater nitrate contamination in the rapidly urbanized region 1264e1272. https://doi.org/10.1016/j.jhazmat.2009.07.132. of south China. J. Contam. Hydrol. 182, 221e230. https://doi.org/10.1016/ Wassenaar, L.I., 1995. Evaluation of the origin and fate of nitrate in the Abbotsford j.jconhyd.2015.09.009. 15 18 - Aquifer using the isotopes of N and OinNO3. Appl. Geochem. 10, 391e405. Zhang, W.Q., Jin, X., Liu, D., Lang, C., Shan, B.Q., 2017. Temporal and spatial variation https://doi.org/10.1016/0883-2927(95)00013-A. of nitrogen and phosphorus and eutrophication assessment for a typical arid WHO, 2011. Guidelines for drinking-water quality, fourth edition. World Health river-Fuyang River in northern China. J. Environ. Sci. 55, 41e48. https://doi.org/ Organization (WHO), p. 323. 10.1016/j.jes.2016.07.004. Wu, S.P., Dai, L.H., Wei, Y., Zhu, H., Zhang, Y.J., Schwab, J.J., Yuan, C.S., 2018. Atmo- Zhang, Y., Shi, P., Li, F.D., Wei, A.L., Song, J.X., Ma, J.J., 2018. Quantiﬁcation of nitrate spheric ammonia measurements along the coastal lines of Southeastern China: sources and fates in rivers in an irrigated agricultural area using environmental implications for inorganic nitrogen deposition to coastal waters. Atmos. Envi- isotopes and a Bayesian isotope mixing model. Chemosphere 208, 493e501. ron. 177, 1e11. https://doi.org/10.1016/j.atmosenv.2017.12.040. https://doi.org/10.1016/j.chemosphere.2018.05.164.