Source identification and health risks of nitrate contamination in shallow groundwater: a case study in Subei Lake basin

Nitrate pollution of groundwater has become a global concern as it can affect drinking water quality and human health. In this paper, an extensive hydrochemical investigation was performed to assess the spatial distribution, source identification, and health risk of groundwater nitrate pollution in the Subei Lake basin. The prevalent pollutant, nitrate (NO3−), was identified based on descriptive statistical method and box plots, and most of the other parameters of groundwater samples met water standards and can be used for drinking purpose. The results showed that nearly 23.53% of groundwater samples displays the NO3− concentrations higher than the limit of 50 mg/L recommended by the World Health Organization, and the highest nitrate content (199 mg/L) is mainly distributed around the Mukai Lake. Piper triangle diagram demonstrated that the dominated anions of hydrochemical types exhibit a gradual evolving trend from HCO3− to SO42− and Cl− with increasing nitrate concentration. The correspondence analysis suggested that agricultural activities are identified as the most possible source of nitrate contamination, while the higher content of other parameters in individual groundwater samples may be controlled by natural factors. The impacts of pollutant NO3− on human health were quantified using human health risk assessment method, and results showed that the order of non-carcinogenic health risk values through drinking water intake is Infants>Children>Adult males>Adult females, and 65%, 53%, 41%, and 35% of samples exceed the acceptable risk level (hazard quotient=1), respectively. The main findings obtained from this study can provide valuable insight on drinking water safety and groundwater pollution prevention.


Introduction
Freshwater is a vital resource for human survival and economic development. Compared with surface water, the quality of groundwater is more stable and higher, thus, groundwater is an appropriate alternative in arid and semiarid regions where available surface water resource is scarce (Alley et al. 2002;Yang et al. 2016a;Golaki et al. 2022). Unfortunately, fresh groundwater shortage has become a serious problem due to the population growth, climate change, and environmental pollution caused by anthropogenic activities (Yang et al. 2016b;Nath et al. 2018;Adebowale et al. 2019;Abascal et al. 2022). In particular, groundwater quality deterioration has led to severe water security problems, which are threatening human health and ecosystem and influencing socioeconomic development (Li et al. 2015;Chen et al. 2017;Liu et al. 2021;Xiao et al. 2022;Harris et al. 2022). Therefore, groundwater quality Responsible editor: Xianliang Yi * Qingchun Yang qyang@jlu.edu.cn 1 and human health risk assessment have been a primary focus of sustainable water resource development and management. During the process of continuous runoff, groundwater undergoes a series of hydrogeochemical reactions with aquifer media, which determines its major chemical compositions (Liu et al. 2015a;Li et al. 2019;Liu et al. 2021;Wang et al. 2022a). In recent years, high-intensity anthropogenic activities have intensified the impact of anthropogenic inputs on groundwater ions, such as fluoride, nitrate, arsenic, and heavy metal (Gu et al. 2013;Li et al. 2020;Long et al. 2021;Golaki et al. 2022). To ensure safe water supply and mitigate groundwater contamination, a large number of studies on sources identification of groundwater pollution were carried out by different methods, such as principal factor analysis (PCA), hierarchical cluster analysis (HCA), ion proportional coefficient (IPC), correspondence analysis (CA), and correlation matrix (Yang et al. 2016b;Sunkari et al. 2022;Xiao et al. 2022). The successful application of these methods is of importance to identify the source of groundwater contamination.
Nitrate pollution often occurs in groundwater and can affect the overall quality of groundwater (Gu et al. 2013;Li et al. 2019;Adimalla and Qian 2021;Gibert et al. 2022;Wang et al. 2022b), which can enter human body through direct ingestion and dermal contact. It is reported that oral ingestion of nitrate-contaminated water can have serious adverse effects on the human health, such as blue baby syndrome that is most common disease caused by nitrate-contaminated drinking water (Fan and Steinberg 1996;Johnson et al. 2010;Ji et al. 2021;Picetti et al. 2022), especially in shallow groundwater wells with potential nitrate contamination (Burow et al. 2010). Thus, many countries recommended acceptable nitrate concentrations in drinking water based on health related criteria, such as 10 mg/L (US Environmental Protection Agency), 11.3 mg/L (World Health Organization and European Union), and 20 mg/L (Standard for groundwater quality of China). Therefore, it is of great significance to quantify the nitrate associated health risks to local populations for ensuring the safety of drinking water and reducing the risk of groundwater contamination.
Subei Lake basin, located in arid and semi-arid region of north Ordos Basin, is economically and ecologically the most important area for the fast development of Ordos energy base (Liu et al. 2018). Shallow groundwater resource is widely used for drinking, irrigation and industrial purposes because of its convenience. Many recent studies have devoted to analyze the evolution of groundwater hydrochemistry in Subei Lake basin due to the potential risk of groundwater pollution. Based on hydrochemical data, Liu et al. (2015b) found that rock weathering and cation exchange predominated the evolution of groundwater chemistry. A recent study also showed that the nitrate contamination is severe in Subei Lake basin, with a concentration variation of 0.6-147.4 mg/L on average of 29.7 mg/L (Liu et al. 2015a). However, these studies mainly focused on hydrochemistry characteristics of groundwater, and systematic studies of the spatial distribution, source identification, and health risk assessment of groundwater nitrate pollution have been neglected.
To ensure the sustainability of groundwater resource development and management, an extensive hydrochemical investigation is conducted to assess the spatial distribution, source identification and health risk of groundwater pollution by focusing on the three aspects: (1) analyze the content and spatial distribution characteristics of groundwater quality variables; (2) identify the possible sources of groundwater contamination by using the correspondence analysis method; and 3) assess the potential health risk of nitrate in shallow groundwater based on the human health risk assessment method. The results of present study will provide some insights into the source identification and health risk assessment of nitrate in shallow groundwater.

Study area
The Subei Lake Basin, located in Ordos City, is situated in the northern part of the Ordos Basin, northwestern China. Geographically, the basin covers an area of almost 109 km 2 with a latitude of 39°10′~39°20′ N and a longitude of 108°45′~109°15′ E (Fig. 1). The climate in study area is characterized by continental semi-arid to arid climate, with cold winters and short, hot summers (Li et al. 2010). The mean monthly temperature varies from 11.5 °C (January) to 21.9 °C (July). The study area has an average of 324.3-mm precipitation annually, which is smaller than annual evaporation (2349.1 mm).
The terrain of study area west and east sides is relatively higher with altitudes between 1370 and 1405 m; the terrain of its south side is slightly lower with elevations between 1270 and 1300 m. In terms of actual hydrogeological conditions and groundwater flow field, the main surface water bodies of Chahan, Subei, Mukai, and Baganaoer lakes are inland lakes, belonging to the same watershed.
Geologically, the study area is represented by two distinct geologic units (i.e., the Quaternary unconsolidated sediments and Cretaceous strata). The depth of sampling wells are classified shallow groundwater samples (<120m, including Quaternary groundwater, shallow Cretaceous groundwater) and deep groundwater samples (>120m, deep Cretaceous groundwater) (Hou et al. 2006). The shallow aquifer is the object of this paper due to potable water for convenience of water intake. The recharge source of groundwater in the shallow aquifer is mainly the infiltration recharge of rainfall; it can be also recharged by lateral inflow from groundwater outside the study area. In addition to the above recharge terms, there are other sources of recharge can also provide a small water, such as leakage recharge from the underlying confined aquifer and infiltration recharge of irrigation water. Evaporation of shallow groundwater evaporation is the major discharge way. In addition, lateral outflow, artificial exploitation, and leakage discharge are also included in the main discharge patterns.

Sample collection and analysis
The study area is a branch of "Hydrogeological Survey of Lake Concentrated Distribution Area in the Northern Ordos Basin." Groundwater Samples were collected by Xi'an Center of Geological Survey from seventeen active shallow wells (W1-W17) during summer 2016 ( Fig. 1). At least half an hour of pumping was taken prior to sampling, and the final groundwater samples were collected after at least three times rinsing the bottles by sampling water. These bottles had been prewashed with 10% HNO 3 − and rinsed with ultrapure water. Parameter pH and total dissolved solids (TDS) were measured in situ by using HANNA HI129 meter. All samples were sealed with adhesive tape and were stored in a portable cooler and transported to the laboratory and refrigerated at 4°C until analysis. The groundwater sample pretreatment and measurements were carried out in a clean room of 1000 class. Potassium (K + ), sodium (Na + ), calcium (Ca 2+ ), magnesium (Mg 2+ ), ammonium (NH 4 + ), fluoride (F − ), chlorine (Cl − ), sulfate (SO 4 2− ), and nitrate (NO 3 − ) were measured using ion chromatography (ICS-2100, Dionex, USA). The concentrations of manganese (Mn), chromium (Cr 6+ ), and iron (Fe) were determined by PerkinElmer Nexion 300D ICP-MS (inductively coupled plasma-mass spectrometry, USA). Bicarbonate concentration in groundwater was determined with the aid of acid-base titration method. All parameter analysis were performed at the Center for Physical and Chemical Analysis of Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences. The reliability of the groundwater data was evaluated by checking ion balances.

Piper triangle diagram
The piper triangle diagram (Piper 1944) is a great effective graphic notation to understand the groundwater basic geochemical characteristics. In the diagram, general chemical characteristics of water samples through the relative content ration of various ions can be directly reflected (in milligram equivalent) (Hong 2012;Zhang et al. 2021). In this study, the AquaChem V4.0 software is selected to make the piper diagram of the study groundwater samples.

Correspondence analysis
Correspondence analysis (CA) method as a multivariate statistical analysis method is often employed to analyze the hydrochemical data for identifying the source of contaminations (Yang et al. 2016b;Wei et al. 2017). CA was developed by French professor Benzecri on the basis of factor analysis in 1970. In CA plane plots, principal factors axes are represented by the top two factors with maximum variance and high resolution respectively. Based on the relative position and distance, CA plane plots can reveal both types of factor analysis features and the relationships between samples and variables (Nowak and Bar-Hen 2005;Angers et al. 1999). The procedures of the CA calculation are subdivided into six steps as below: (I) Acquire the water quality monitoring data matrix X based on Eq.
(1). m and n represent the number of groundwater samples and the number of assessing water quality variables, respectively.
(II) Calculate the sum of each row and column as x i , x j and total T using Eq. (2).
(III) Obtain the data transformation matrix Z based on Eq. (3).
(XXII) Perform the R-type and Q-type factor analysis. In this part, the eigenvalues (λ 1 ≥ λ 2 ≥ … ≥ λ m ) of A are calculated, and obtain k eigenvalues according to the eigenvalue cumulative percentage. Then, the eigenvectors (α 1 , α 2 ,…, α k ) of the k eigenvalue are obtain, and loading matrix of the R-type factor (F) is achieved by using Eq. (6); the Q-type factor analysis is based on the eigenvectors (β 1 , β 2 …β k ) of B by eigenvalues of A, and is expressed as Eq. (7).
(VI) Finally, produce the factor plane diagram.

Human health risk assessment
Human health risk assessment (HHRA) method recommended by the US Environmental Protection Agency (USEPA) has been widely used to assess the health risk of pollutants in groundwater to humans (USEPA 2008;Zhang et al. 2021;Naik et al. 2022). Previous studies showed human's exposure to pollutants via oral ingestion pathway is more harmful than inhalation and dermal contact (Shakoor et al. 2017;Nawale et al. 2021;Xiao et al. 2022). Thus, this study focused on the human health risk associated with the exposure to nitrate-contaminated drinking water via oral ingestion pathway. The non-carcinogenic hazard quotient (HQ) can be calculated by Eq. (8).
where CDI is chronic daily intake of contaminant in drinking groundwater, which is calculated by Eqs. (9) and (10). RfD is reference dose. C is the pollutant concentration in drinking water. IR is the mean drinking rate. EF is the mean exposure frequency. ED is mean exposure duration. BW is the mean weight of human body. AT is the mean exposure time.
The reference values used in HHRA method are detailed in Table S1:

General water quality parameters characteristics
Box plot is one of the most effective visualization graphs to illustrate the dispersal in dataset. The concentrations of sixteen water quality variables are presented in box plots (Fig. 2), statistically summarized in Table S2, and spatially distributed in Fig. 3 (except pH) that were obtained by using ordinary kriging (OK) method in ArcGIS software (Eqs. S1 in supplementary materials). pH values of shallow groundwater samples varied from 7.42 to 8.62 with a mean value of 7.97, indicating an overall weakly alkaline water environment. According to the World health organization (WHO) guideline, 11.76% of groundwater samples exceed the permissible limit of 8.5. Total soluble solids (TDS) is a vital index for classifying water palatability, whose acceptable limit for drinking water is 1000 mg/L. Except groundwater sample W7, the TDS content of other samples varied from 199 to 968.4 mg/L, reaching the permissible drinking standard. The areas with high TDS concentration were distributed in the eastern parts of the study area, which can be attributed to relatively closed hydrogeological condition and strong evaporation effect (Fig. 3). Total hardness (TH) is also an important variable to evaluate the characteristics of the strata. The TH values varied from 35 to 661 mg/L, and higher TH content regions were distributed near the Mukai Lake (northwestern part of the study area), which was consistent with Ca 2+ , Mg 2+ , Cl −, and HCO 3 − (Fig. 3). Only W11 sample had TH concentration falling in the unacceptable level (higher than 450 mg/L) in study area.
The concentrations of F − varied from 0.46 to 1.6 with an average value of 0.85 mg/L. Among groundwater samples, only one sample (W3) had F − concentration higher than 1.5 mg/L, and was located near the Subei Lake (Fig. 3). The semi metal and metal like Mn, Cr 6+ , and Fe concentrations are below the permissible limit, and the corresponding change intervals were 0.001-0.035 mg/L, 0.001-0.005 mg/L, and 0.1-0.2 mg/L, respectively. It was generally noted that the concentrations of nitrate in shallow groundwater samples were high, ranging from 0.7~199 mg/L with an average value of 48.27 mg/L, and about 23.53% of groundwater samples had the NO 3 − concentrations higher than 50 mg/L recommended limit by WHO. The concentrations of NH 4 + varied from 0.02 to 0.05 mg/L, and NO 2 − concentrations in studied wells were lower than the detection value.
In this study, in order to proper interpretation of geochemical processes affecting NO 3 − behavior in groundwater, the groundwater samples are divided into two groups: the first group with low nitrate concentrations less than the value of 20 mg/L; and the second group with high nitrate concentrations greater than the value of 20 mg/L (Bahrami et al. 2020).

Hydrochemical facies
The plots of high nitrate and low nitrate (Fig. 4) revealed that (1) with respect to cations, all the samples are distributed in the lower part of lower-left triangle, indicating that some samples are calcium-type, some samples are sodium-type water and a mixed type predominated. However, in the case of anions, bicarbonate had much contribution to the hydrochemical compositions; (2) all groundwater samples with high nitrate, with one sample exception, fell in zone I, but the plots with low nitrate were relatively scattered. For low nitrate of groundwater samples, only 2 samples (33% of the total samples) fell in zone I, and one sample and 3 samples fell in zone III and IV, respectively. Generally, the dominated cations of groundwater samples with low and high nitrate content were Na + and Ca 2+ (left side of the triangle of piper triangle diagram), respectively, and the dominated anions were observed with a gradual evolving trend from HCO 3 − to SO 4 2− and Cl − with the increase of nitrate content (right side of the triangle of piper triangle diagram). This indicates that the factor increasing NO 3 − concentration may be responsible for the evolution of shallow groundwater hydrochemical facies (Menció et al. 2016).

Correspondence analysis
The results of correspondence analysis are presented in Table S3 and Fig. 5, of which F1 and F2 represent the principal factor loadings of groundwater indexes, and G1 and G2 indicate the ones of collected water samples, respectively. Meanwhile, the first two eigenvalues of cumulative variance contribution rate reached 82.6%.
In CA plane plots, axis F1 plays a vital role in water quality assessment. The maximum absolute value of F1 was NO 3 − in the negative direction, while the positive direction was SO 4 2− (Fig. 5). The absolute value of NO 3 − was greater than SO 4 2− , which further is therefore considered to be the primary contaminations of groundwater. The closer the sample is to the negative direction of F1, the higher the content is more seriously polluted by NO 3 − , while the closer to the positive direction of F1, the higher the degree of pollution by SO 4 2− . The axis F2 can also assess the groundwater pollution, but the effect is less than axis F1. The load of F − was the largest in the positive direction, and the absolute value of SO 4 2− was the largest in the negative direction. The five indexes with lower factor loadings (TH, Mn, Cr 6+ , Cl − , and NH 4 + ) were located in the center of axes F1 and F2, meaning that they are not the dominating indexes of water pollution in study area. From the samples of view, 17 samples were obviously divided into four subzones: Zone I (W1, W5, W14 and W15), Zone II (W2, W4, W6, W10, W12 and W17), Zone II (W3, W8, W9, W13 and W16), and Zone IV (W7 and W11), respectively. The nitrate content of groundwater samples in Zone I was high (88.8 ~ 199 mg/L), indicating that private residential wells are seriously contaminated as a result of anthropogenic activities and should be regarded as contamination hot spots of special concern.
It can be seen that TH, Ca 2+ , and Mg 2+ , the primary water quality parameters in negative direction of axis F1, are related to the samples in Zone II which indicates medium nitrate concentration (22.9 ~ 49.2 mg/L). The research performed by Yang et al. (2016b) has reported that high groundwater total hardness and nitrate are most likely caused by agricultural activities. In Zone III, the water samples were closely related to HCO 3 − and F − , among which the former can be considered as the main anion of most water samples in the region. The high F − and low nitrate content in groundwater is assumed to be influenced by local geological conditions. Samples in Zone IV are closely related to TDS, Na + , and SO 4 2− . Mineral dissolution, evaporation and concentration, and cation exchange result in higher Na + in groundwater that can provide explanation for the contribution of Na + to the hydrochemical compositions in some samples. According to the research of Liu et al. (2015a), rock weathering, as a dominant factor, contributes greatly to the increase of salinity in groundwater. Sulfate in the main constituent of dissolving salt, and lower strata is rich in oil and natural gas, which may lead to sulfate exceeding the standard in individual groundwater samples (Yang et al. 2016b).

Sources of nitrate in shallow groundwater
As discussed above, nitrate is the dominant pollutant influencing the shallow groundwater quality. Although there is no existing literature on the background values of nitrate in the study area, a large number of studies reported that nitrate concentration of more than 3 mg/L may be caused by anthropogenic activities (Burkart and Kolpin 1993;Liu et al. 2005). Thus, the origins and mechanisms controlling the hydrochemistry of nitrate enriched groundwater in the Subei Lake basin need to be further studied.
Anthropogenic activities can exercise a dominant influence on groundwater NO 3 − pollution (Gu et al. 2013;Li et al. 2019;Liu et al. 2021;Ren et al. 2022;Zhang et al. 2022). It is well known that intensive anthropogenic activities can lead to elevated Cl − and SO 4 2− , which may also originate from the dissolution of evaporate (gypsum and halite), the oxidation of sulfide and industrial activities, whereas NO 3 − mainly originates from agricultural activities and inputs of domestic sewage (Zhang et al. 2021). Thus, water bodies polluted through anthropogenic activities are generally characterized by higher molar ratios of Cl − /Na + and NO 3 − /Na + . In the Fig. 6a, high nitrate samples (the content of NO 3 − >20 mg/L) were primarily located near the upper right corner of the 1:1 line, indicating that high NO 3 − in groundwater of study area is mainly contributed by agricultural activities. Meanwhile, the relationship between NO 3 − /Na + ratios and SO 4 2− /Na + ratios also showed industrial activities also affect the groundwater chemical characteristics to a certain extent (Fig. 6b). In addition, according to field investigation, the population distribution in the study area is scattered, and there are no other large factories except some alkali factories around the lake in the middle of the study area. Therefore, agricultural activities are supposed to be the most possible source of nitrate pollution due to the long-term use of fertilizers such as organic (slurry and manure) or synthetic fertilizers. In addition to fertilizer, livestock, manure and atmospheric sources, nitrate levels in groundwater also fluctuate widely from area to area depending on many factors, such as precipitation, evaporation, and soil type (Wakida and Lerner 2005;Harris et al. 2022;Wang et al. 2022b).

Health risk assessment of nitrate in groundwater
The results of statistical and correspondence analysis show that the groundwater has been contaminated by nitrate in the study area, with nitrate concentrations of up to 199 mg/L in well 14. Meanwhile, previous studies indicated that groundwater nitrate pollution is mainly distributed in northern part of the Ordos Basin (Liu et al. 2015a;Yang et al. 2016b). Given the potential health risk of NO 3 − -N in shallow groundwater, it is thus important to assess the non-carcinogenic health risks of exposure to nitrate in drinking water.
In this study, hazard quotient (HQ) for individuals from NO 3 − -contaminated drinking water computed (Fig. 7). The HQ values of nitrate showed a variation of 0.04-11.65 (mean = 2.61), 0.03-7.2 (mean = 1.61), 0.02-5.17 (mean = 1.16), and 0.02-6.17 (mean = 1.38) for infants, children, adult females, and adult males, respectively. Overall, under the same circumstances, the order of non-carcinogenic health risk values of different groups is Infants>Children>Adult males>Adult females. The results of HQ demonstrated that the younger groups are at a higher health risk of NO 3 − contamination than the adults due to their less developed enzymatic metabolism and lighter body weight (Zhang et al. 2021).
According to USEPA, when HQ higher than 1, the potential non-carcinogenic risk should not be neglected. According to statistical classification of shallow groundwater samples, 65%, 53%, 41%, and 35% of samples surpassed the permissible limit for nitrate associated health risk (HQ=1) for infants, children, adult males, and Adult females, respectively, indicating a high risk of NO 3 − in groundwater. As shown in Fig. 7, the areas of high risk were concentrated on the northwest (near the Mukai Lake), northeast (near the Bagannaoer Lake) and southern parts of the study area, which is essentially consistent with that of groundwater NO 3 − concentrations.

Conclusions
In this study, the hydrochemistry characteristics, the possible sources of contamination, and the human health risk of shallow groundwater were investigated. The results showed that most of groundwater samples are within the permissible limit and suitable for drinking purpose regardless of nitrate content in shallow groundwater.
Piper triangle diagram demonstrated that the dominated anions are observed with a gradual evolving trend from HCO 3 − to SO 4 2− and Cl − with the increase of nitrate content. Agricultural activities were identified as the most possible source of nitrate pollution in groundwater. The health risk assessment showed that the risk of exposure to nitrate in drinking water displays a basic trend: Infants>Children>Adult males>Adult females, and 65%, 53%, 41%, and 35% of samples exceed the acceptable risk level (HQ=1) for the four groups, respectively. To protect water environment and public health in the study area, local government should adopt measures to prevent the overuse of fertilizers and pesticides in agriculture for reducing the risk of groundwater nitrate pollution. Fig. 7 The HQ spatial distribution maps of non-carcinogenic health risk for different age groups. a Infants, b children, c adult females, and d Adult males