Determining environmental risk and source of heavy metal(loid)s in the surrounding farmland soil of a zinc smelter in water source area, Northwest China

To explore the impact of metal smelting activities on surrounding environment in the water source area of the Mid–route of South–to–North Water Transfer Project of China, soil samples of farmland around a zinc smelter in this area were collected and the pollution, risk, and source of heavy metal(loid)s (As, Cu, Cr, Pb, Mn, Ni, V, and Zn) in soil samples were determined in this study. The heavy metal(loid)s contents were measured by X-ray fluorescence spectrometry and their pollution levels and ecological risks were assessed by geoaccumulation index, Nemerow synthetic pollution index,and potential ecological risk index. The possible sources of the heavy metals(loid)s were identified by multivariate statistical analysis methods. The results show that the mean contents of all analyzed heavy metal(loid)s in the farmland soil were above the local soil background values except Mn; the contents of As, Cu, Pb, Zn, and Ni in the downwind direction soil decreased with the distance increasing between the sampling site and the zinc smelter; the investigated soils were moderately to seriously polluted by heavy metal(loid)s and the heavy metal(loid)s presented moderately ecological risk as a whole; As, Cu, Pb, and Zn mainly originated from zinc–smelting activities; Cr, Mn, and V primarily derived from natural source; Ni mainly came from zinc–smelting activities, partly from natural source. The zinc–smelting activity influenced the heavy metal(loid)s content, particularly Zn and Pb, in the surrounding farmland soil. The local government should strengthen the cooperative monitoring of heavy metal(loid)s in farmland and agricultural products, as well as pollutant emission monitoring and control.


Introduction
Under the context of global climate change, with the development of industry, the acceleration of urbanization, the increase of agricultural chemicals using, and intensification of human activities, the problems of soil deterioration and pollution have become increasingly prominent and attract more and more attention worldwide in recent decades (Zhao et al. 2020;Kumar et al. 2019;Zhang et al. 2019;Akopyan et al. 2018;Yang et al. 2018;Li et al. 2017). Among all kinds of soil pollutants, heavy metal(loid)s cause more attention due to their high toxicity and non-biodegradability . Heavy metal(loid)s in soil can be absorbed and enriched in plants and transferred to animals and human bodies via food chains (Liu et al. 2017;Zhang et al. 2016). When people eat grains or vegetables polluted by heavy metal(loid)s, their survival and health will be threatened (Baltas et al. 2020;Zhou et al. 2020;Bhatti et al. 2018;Li et al. 2018Li et al. , 2019. High contents of heavy metal(loid) s in the body can interfere with the physiological process of the body, and injure human organs such as heart, bone, kidney, small intestine, reproductive system and nervous system (Csavina et al. 2012). Hence studying and evaluating the ecological risk of heavy metal(loid)s in farmland are important for understanding soil environmental quality, strengthening comprehensive prevention of soil pollution and safeguarding human health.
The major sources of soil heavy metal(loid)s are natural sources and anthropogenic sources (Jiang and Guo 2019).
Weathering of rocks, volcanic eruptions, splashing waves, vegetation discharge are the main natural sources (Lee et al. 2020), while mining (acid mine drainage, abandoned mine tailing), beneficiation, metal-smelting, electroplating, dyes, textiles, burning of fossil fuels, agricultural pesticides, fertilizers, sludge application and sewage irrigation are the main anthropogenic sources (Wang et al. 2019a, b;Peng et al. 2019;Ettler 2016;Lu et al. 2014a). Among various soil heavy metal(loid) sources, nonferrous metal smelting is one of the significant sources (Kang et al. 2019;Kříbek et al. 2019;Ghayoraneh and Qishlaqi 2017;Shiel et al. 2010;Wang et al. 2010). In the metal-smelting process, large quantities of heavy metal(loid)s discharge into surrounding soil through sewage irrigation, waste residue infiltration, and atmospheric sedimentation, resulting in heavy metal(loid)s pollution in farmland soil around smelter (Shen et al. 2017;Wen et al. 2015). The main reason is that these metal(loid)s cannot be degraded naturally in the soil, so they are continiously enriched to a high level. Many studies have shown that contaminated soils around smelters have become high-risk areas for safe production of agricultural products (e.g. Lee et al. 2020). There are lots of literature on soil heavy metal(loid)s pollution around smelters in China, most of which are concentrated in the eastern and southern areas of the country (e.g. Zhao et al. 2020;Li et al. 2011Li et al. , 2018, whereas, up to now, few researches on farmland soil pollution around metal smelters have been launched in the water source area of the Mid-route of South-North Water Transfer Project and the environmental impact of metal-smelting activities in the water source area remain unclear. South-North Water Transfer Project is a national water resource allocation project for solving the water shortage in North China. The Mid-route of South-to-North Water Transfer Project in China is to divert water from Danjiangkou reservoir of Hubei province to North China to solve the problem of water shortage in Hebei, Tianjin, Beijing and other provinces and cities, and to provide water guarantee for industrial and agricultural production and residents' life in more than a dozen large-and medium-sized cities along the Project. The water security and environmental quality of the water source area are very important for the Water Transfer Project. Therefore, it is extremely necessary to study heavy metal(loid) pollution in farmland soil around metal smelters and the environmental impact of metal-smelting activities in the water source area.
S h a n g l u o c i t y ( 3 3°2 ʹ 3 0 ʺ -3 4°2 4 ʹ 4 0 ʺ N , 108°34ʹ20ʺ-111°1ʹ25ʺ E) is located in the upstream of Danjiang valley, the water source area of the Mid-route of South-to-North Water Transfer Project of China Zhuang and Lu 2020). With abundant non-ferrous metal resources, such as zinc, lead, copper, molybdenum and gold, Shangluo is the important mineral resources base in Northwest China. Many metal mining and smelting industries are distributed in Shangluo, which may cause heavy metal(loid)s pollution in agricultural soil and water. Heavy metal(loid)s in agricultural soil would enter into surface water with runoff or soil erosion excepting pollute agricultural products, leading to the deterioration of water quality. Shangluo has abundant rainfall and complex topography and geomorphology, which result in serious soil erosion in that area. To understand the impact of metal-smelting activities to farmland soil in the water source area, the farmland around a zinc-smelting plant with no other large-scale anthropogenic source in its vicinity in Shangluo, Northwest China was investigated in this study. The primary purposes of this study were (1) to determine the contents of heavy metal(loid)s in the farmland soil around the zinc smelter; (2) to evaluate the contamination degrees and ecological risks of the investigated heavy metal(loid)s in the farmland soil; and (3) to identify the sources of heavy metal(loid)s in the farmland soil. The general idea and methods of this study are as follows: first, collect farmland soil samples on base of the field investigation, second determine the content of heavy metal(loid)s in the soil samples by X-ray fluorescence technology, then evaluate the pollution and risk of heavy metal(loid)s in the farmland using different evaluation indexes, and finally identify the sources of the heavy metal(loid)s by multivariate statistical analysis methods. The findings of this study can reveal the environmental impact of metal-smelting activities in the water source area and would provide scientific basis for regulators in environmental protection and management of the water source area.

Study area
Shangluo is a mountainous valley city with a permanent population of 2.50 million (up the end of 2019) and about 110 km southeast of the provincial capital Xi'an. The climate of Shangluo is semi-humid monsoon mountain climate in warm temperate zone. January is the coldest month with a mean temperature of 7.8 °C while July is the hottest with a mean temperature of 13.9 °C. The annual mean precipitation is 696.8-830.1 mm. Affected by the terrain, the annual wind direction in Danjiang valley is mainly east (E), east-south-east (ESE), and west (W), and the mean wind speed is circa 2.4 m s -1 . The main crops in Shangluo are wheat and corn, and the main cash crops are vegetables and walnuts.
The investigated zinc smelter, located at the hillside of the north bank of Danjing valley about 7 km southeast of Shangluo city, was built in 1987 with 50 thousand tons of zinc productivity per year. Meanwhile, precious metals such as silver (Ag) and indium (In) are recovered and produced in the zinc-smelting process. The main raw ore of the zinc smelter is sphalerite (ZnS), associated galena (PbS) and chalcopyrite (CuFeS 2 ). In addition, it also contains cadmium, indium, silver and other metals.
Considering the influence of topography, wind direction and villages distribution, three plots of farmland (marked A, B and C, respectively) in different directions near the smelter were selected for soil sampling (Fig. 1). Plot A is located on the hillside in the northwest direction of the zinc smelter. There are no villages near this plot and the traffic is inconvenient as only farmland paths lead to there. The altitude of the terraced farmland on the hillside is about 700-720 m and about tens of meters higher than the smelter. Plot B is located in the southwest of the smelter and next to a village. The altitude of plot B is about 647-649 m which is lower than that of the smelter. There are hardened rural cement roads around Plot B. Luxuriant trees and shrubs grow on the hillsides between plot B and the smelter. Plot C is located in a valley to the east of the smelter and a small seasonal stream flows through the valley. The altitude of plot C is about 680-690 m which is slightly lower than that of the smelter. There is barely vegetation between plot C and the smelter. All three plots are wheat and corn rotation, and grains grown on the farmland are consumed by the local residents. The soil type of the investigated farmland is yellow-brown soil with a pH from 7.2 to 8.3. The mean value of soil organic matter, total nitrogen, total phosphorus, available nitrogen, available phosphorus and available potassium was 20.1 g kg -1 , 1.2 g kg -1 , 0.6 g kg -1 , 108.9 mg kg -1 , 17.9 mg kg -1 and 113.7 mg kg -1 , respectively. Texture analysis indicates that soil is silty clay loam, composed of sand (26.3%), silt (56.0%), and clay (17.7%). The mean contents of major elements, i.e. Fe 2 O 3 , CaO, K 2 O, MgO, Na 2 O, SiO 2 , and Al 2 O 3 were 6.7%, 0.9%, 1.5%, 1.8%, 2.0%, 33.1%, and 7.8%, respectively.

Sampling and analytical methods
Soil samples were collected in July 2017. A hand-held GPS was used to record the longitude, latitude, altitude and other information of sampling location. A total of 25 topsoil (0-20 cm) samples were taken from plot A (9 samples, numbered A1-A9), plot B (8 samples, numbered B1-B8) and plot C (8 samples, numbered C1-C8) ( Fig. 1). At each point, a composite soil sample was fully mixed and retained at 1 kg in accordance with the four-point method in an area of about 2 m × 2 m. All soil samples were kept in polythene bags.
In the laboratory, all soils were air-dried naturally, and then the impurities were removed. Agate mortar and pestle were used to ground the samples to pass through 75 μm nylon mesh. The processed samples were stored in polyethylene plastic bags. Preventing from the cross-contamination between different samples, all processing was performed avoiding touching with metallic object. 4 g of treated soil and about 2 g of boric acid which used as the substrate were pressed by the molding press to form a circular sheet sample to be tested (Chen et al. 2013). The contents of heavy metal(loid)s (As, Cu, Cr, Pb, Mn, Ni, V, and Zn) in soil were tested using X-ray fluorescence spectrometer (XRF, Panalytical, PW2403) according to the method in literature (Pan et al. 2017;Lu et al. 2014a, b), with method detection limits (MDL) of 0.5-1.0 mg kg -1 for the selected elements (Pan et al. 2017). The national standard material samples (Chinese standard soil samples GSS6 and GSS8, purchased from the Center of National Standard Reference Material of China) and parallel samples were adopted for quality guarantee and control in the experiment Zhuang and Lu 2020). Relative error between the detected value and reference value of elements in standard sample is less than 10%, and that of the duplicate sample is less than 5%.

Assessing methods of pollution and ecological risk
The geoaccumlation index (I geo ) and Nemerow synthetic pollution index (NSPI) were, respectively, used to assess the single pollution level and comprehensive pollution level of heavy metal(loid)s in the farmland soil. The I geo is calculated using the following formula Chakraborty et al. 2017;Duodu et al. 2016;Lu et al. 2009): where C i is the content of heavy metal(loid) i in the soil (mg kg -1 ); Constant k is the change of referencevalue possibly caused by diagenesis, which is taken as 1.5 here; B i is the reference value of heavy metal(loid) i in the parent rock, and here is the background value of soil elements in Shaanxi Proveince (CNEMC 1990). The grades of pollution were distinguished in six levels proposed by Müller (1969) ( Table S1 in Supplementary Materials).
The NSPI is calculated by formula (2)  where P iave and P imax, respectively, represents the mean and the maximum of single pollution index P i of all heavy metal(loid)s in a sample. P i is equal to the ratio of the content of heavy metal(loid) i in the farmland soil and its corresponding background value in Shaanxi soil (CNEMC 1990). The grades of P i and NSPI, as well as the corresponding pollution levels are listed in Table S2 and Table S3 (Supplementary Materials). Potential ecological risk index (PERI), proposed by Hakånson (1980), to evaluate the ecological hazard effect resulted from toxic pollutants in sediments, has been widely applied to assess the environmental risk of heavy metal(loid) s in soil Ke et al. 2017;Suresh et al. (1) where C s i represents the determined value of heavy metal(loid) i in the soil, while C b i represents the corresponding reference value (CNEMC 1990). T r i is the toxic response factor of different substance (As, Cu, Cr, Mn, Ni, Pb, V, and Zn are 10, 5, 2, 1, 6, 5, 2, and 1, respectively) (Yuan et al. 2014;Lu et al. 2014b). E r i is the potential ecological risk index of heavy metal(loid) i, while PERI is the comprehensive potential ecological risk index. The evaluation criteria for E r i and PERI are listed in Table S4 (Supplementary Materials) (Yuan et al. 2014;Lu et al. 2014b).

Statistical analysis methods of data
Descriptive statistical analysis, Pearson correlation analysis, cluster analysis (CA) and principal component analysis (PCA) were conducted using the commercial statistics software package SPSS version 21.0 for Windows (IBM Company, Chicago, USA). Descriptive statistics of heavy metal(loid) contents in the farmland soil samples, including minimum (Min), maximum (Max), mean, standard deviation (SD), coefficient of variation (CV), skewness and kurtosis, were applied to analyze the content characteristics of heavy metal(loid)s in the farmland soil. Pearson correlation analysis, CA and PCA, the extensively used multivariate statistical analysis methods in pollution studies (Wang et al. 2019a, b;Zhang et al. 2019;Ke et al. 2017;Pan et al. 2017;Lu et al. 2014a), were adopted to determine the relationships among heavy metal(loid)s and to identify their possible sources combining their content characteristics and pollution levels in the farmland soils around the zinc smelter. The soil element background values of Shaanxi published in 1990 (CNEMC 1990) were used in this study to distinguish the natural source and anthropogenic source of heavy metal(loid)s in the farmland soils around the zinc smelter and to evaluate the pollution and risk of heavy metal(loid)s in the soil. At present, it is difficult or even impossible to find the farmland soil unaffected by human activities.

Heavy metal(loid) contents
The determination results of the heavy metal(loid) contents from the farmland soil samples together with the background value of Shaanxi soil (CNEMC 1990) are presented in Table 1. The mean contents of As, Cu, Cr, Pb, Mn, Ni, V, and Zn were 17.3, 39.5, 135.7, 143.3, 571.1, 35.8, 82.4, and 771.9 mg kg -1 , respectively. Except for Mn, the mean contents of As, Cu, Cr, Pb, Ni, V, and Zn were above their corresponding background values, which were 1.6, 1.8, 2.2, 6.7, 1.2, 1.2, and 11.1 times the background values, respectively. The coefficient of variation (CV), the ratio of the standard deviation and the mean, is a statistical measure reflecting the variability of the observed value. In general, a CV ≤ 20% indicates low variability, a CV range of 21-50% indicates moderate variability, a CV range of 51-100% shows high variability and a CV > 100% is regarded as very high variability (Pan et al. 2017;Phil-Eze 2010). Table 1 shows that the contents of heavy metal(loid)s determined in the farmland soil around the zinc smelter presented different variance. The contents of Zn, Pb, and Cu presented high variability (51% < CV ≤ 100%), the content of As presented moderate variability, and the other heavy metal(loid) s showed low variability. The large CV values of Zn, Pb, Cu, and As reveal their heterogeneity in the farmland soil around the zinc smelter, demonstrating the impact of anthropogenic sources (Pan et al. 2017(Pan et al. , 2020Karim et al. 2014). Kurtosis and skewness values indicate whether the concentrations of the studied heavy metal(loid)s follow a normal distribution (Pan et al. 2017;Chen et al. 2013). The kurtosis values of As, Cu, Pb, and Zn are 2.7, 3.7, 1.8, and 2.5, respectively, showing that these four elements deviated from the normal distribution. The absolute kurtosis values of Cr, Mn, Ni, and V are less than 1, indicating that their contents in the soil samples were close to the normal distribution. Skewness values of 9.9, 15.9, 5.1, and 8.8 further indicated that As, Cu, Pb, and Zn were skewed positively towards the higher concentration, while, the skewness values of Cr, Mn, Ni, and V are close to 0, showing that the content distribution of these four elements was relatively uniform. Table 1 displays a comparison of heavy metal(loid)s content in farmland soil around the zinc smelter of Shangluo, China with some similar studies reported in the literature (Kang et al. 2021;Kříbek et al. 2019;Wang et al. 2012Wang et al. , 2019aGhayoraneh and Qishilaqi 2017;Li et al. 2015). It can be found that the contents of Cr, Cu, V, Zn and Pb in the farmland soil around the zinc smelter of Shangluo are higher than that around other zinc or lead-zinc smelters, except for Pb in the soil around Kabwe lead-zinc smelter, Zambia. The difference of heavy metal(loid) content in the soil around the smelters may be controlled by a variety of natural and anthropogenic factors, such as soil parent material, soil type, background value of soil elements, terrain, wind direction and speed, precipitation, composition of ore raw materials, smelting process, chimney height, gas and particle emission control technology and efficiency, etc. Figure 2 shows that the investigated heavy metal(loid) s in the farmland soils of three plots have diverse variation features. The contents of As, Cu, Pb, and Zn in the farmland soils of three plots have remarkable difference, while the  Figure 3 shows that the contents of As, Cu, Pb, Zn*, and Ni in the farmland soil samples from plot A present decreasing trend with the increase of the distance between the sampling site and the zinc smelter, while the contents of Cr, Mn, and V in the farmland soil have not distinct trend. The contents of As, Cu, Pb, and Zn decreased by 88%, 76%, 71%, and 83%, respectively, and Ni contents declined by 28% in the farmland soil with the distance of soil sampling site to the zinc smelter increasing from 100 to 300 m (Fig. 3). Plot A is situated in the downwind direction of the zinc smelter. Considering no other pollution sources existing in the vicinity of the zinc smelter, we think the aforementioned variation characteristics of heavy metal(loid) content in the farmland of plot A reflecting the impact of zinc-smelting activity on the contents of As, Cu, Pb, Zn, and Ni. The elevated As, Cu, Pb, and Zn in the downwind farmland soils (Plot A and Plot C) of the zinc smelter and their contents in the downwind soils decreasing with the increase of distance, indicate that zinc-smelting activities have caused the enrichment of these metal(loid)s in the surrounding farmland soils, in particular Zn and Pb, and the emission of flue gas and particles during zinc-smelting is the main reason for the accumulation of heavy metal(loid)s in the farmland soils.

Pollution and ecological risk of heavy metal(loid)s
The results of P i and NSPI are shown in Table 2 and Fig. 4. The determined heavy metal(loid)s presented pollution at different extent in three plots, except for As in plot B and Mn in plot C. The mean P i values of As, Cu, Pb, and Zn in the farmland soil decrease in the order of plot A > plot C > plot B. Especially Pb and Zn, their P i values are 10.31 and 15.92 in plot A, and 6.80 and 14.64 in plot C, respectively, presenting severe pollution. The pollution diversity of heavy metal(loid)s in three plots may be related with their position and terrain. Plot A is located in downwind of the zinc smelter and its terrain is higher than the zinc smelter. The special topography hinders the further diffusion of smoke and dust with heavy metal(loid) particles and promote pollutants sinking, which may be the cause of severe soil pollution in plot A. Plot B is not in the dominant wind direction, and there are lush trees and vegetation between plot B and the smelter. Plot C is situated in the east of the smelter (downwind) and its terrain is lower than the smelter. The diffusion of pollutants discharged from the smelter in plot C is easier than in plot A. Figure 4 shows that the NSPI values of heavy metal(loid)s determined in the farmland soils from plot B are in 2-3 presenting moderate pollution, while which of heavy metal(loid)s in the farmland soils from plot A and plot C are larger than 3 indicating serious pollution. The comprehensive pollution degree of heavy metal(loid)s in the farmland soils of three plots decreases in the order of plot A > plot C > plot B. The computed results of I geo for elements in the farmland soil around zinc smelter of Shangluo are presented in Table 3 and Table S5 (Supplementary Materials). In plot A, the I geo of Mn and V are all below zero, indicating no pollution of these two metals in samples from plot A. Most of the I geo values of Ni are under zero, which indicated no pollution, except two I geo values 0.03 (A4) and 0.09 (A7) ( Table S5 in Supplementary Materials) slightly greater than zero. The I geo values of As range from − 0.44 (A3) to 1.66 (A7) which corresponded to level 0 of unpolluted to level 2 of moderately polluted. The mean I geo value of 0.31 indicated As is slightly polluted. As for Cr, the I geo value of which range from 0.30 to 0.68 with the mean value of 0.55, indicated slight Cr pollution in plot A. The mean I geo value of Cu is 0.51, which corresponded to level 1 of slightly polluted. Sample from A7 is an exception, the I geo value of which is 2.03 corresponded to level 3 of moderately severely polluted. Pb and Zn showed more accumulation in plot A as compared to other seven elements. The mean I geo value of Pb is 2.66 which corresponded to level 3 of moderately severely polluted. The max I geo value of 3.95 (A7) corresponded to level 4 of severely polluted. The mean I geo value of Zn is 3.12 which corresponded to level 4 of severely polluted. Sample from A7 was extremely polluted with the I geo value of 5.14.
In plot B, the I geo of As, Mn, Ni, and V are all below zero, which indicated no samples are polluted by these four elements in plot B. The mean I geo value of Cu is − 0.11, corresponding to level 0 of unpolluted, but the I geo value 0.10 and 0.05 from B5 and B8 are two exceptions. As for Cr, Pb, and Zn, the I geo of them are all between 0 and 1, which  . 4 The NSPI values of heavy metal(loid)s in the farmland soils showed samples from plot B were slightly polluted by these elements.
In plot C, the I geo values of Mn, Ni, and V are all below zero, which indicated no samples are polluted by these elements in plot C. For As, Cr, and Cu, their mean I geo values are between 0 and 1, indicating that these three elements corresponded to level 1 of slightly polluted. Samples from C2 and C3 were unpolluted with As. The I geo value of Cr has little change, and the maximum and minimum values are 0.46 and 0.33, respectively. As for Cu, the I geo value of samples from C1, C2, C3, and C4 are < 0, corresponding to level 0 of unpolluted. The I geo value of samples from C5, C6, C7, and C8 are between 0 and 1 ( Table S5 in Supplementary Materials), corresponding to level 1 of slightly polluted. Similar to plot A, Pb and Zn are also accumulated obviously in plot C. The I geo value of Pb ranged from 1.05 (C3) to 2.68 (C8) which corresponded to level 2 of moderately polluted and level 3 of moderately to severely polluted. The mean I geo value 2.06 of Pb indicated moderately polluted in plot C. The mean I geo value of Zn is 3.15 which corresponded to level 4 of severely polluted. The same as Pb, the minimum I geo value of Zn 1.74 appears at C3 and maximum I geo value 4.07 appears at C8.
The E r i values of heavy metal(loid)s in the farmland soils are shown in Fig. 5. It can been found from Fig. 5 that the E r i values of Cr, Mn, Ni, and V in soil samples and Cu in most soil samples (except soil sample A7 in plot A) are < 15, indicating low ecological risk level. As and Zn in some soil samples of plot A and plot C have moderate and considerable ecological risk. Pb in the farmland soils from three plots possesses different ecological risk levels, i.e. in plot A presenting considerable to high ecological risk, in plot B presenting low ecological risk, while in plot C presenting moderate to considerable ecological risk. The PERI values of heavy metal(loid)s determined in the farmland soil around the zinc smelter range from 43.46 to 261.49 with a mean of 83.49 (Table S6 in Supplementary Materials), showing the investigated soils have wide ecological risk levels, i.e. low to high ecological risk. The comprehensive ecological risk levels of heavy metal(loid)s in the farmland soils of three plots are low in plot B, moderate in plot C, and considerable in plot A. Pb is the main contributor of the comprehensive ecological risk; then there are As, Cu, and Zn, which, respectively, contribute 37.2%, 19.3%, 11.7%, and 11.3% to PERI.

Pearson's correlation analysis results
Studying the correlation between elements can predict whether the sources of heavy metal(loid)s are same owing to that different heavy metal(loid)s in soil have different migration and enrichment trends. If there is a correlation between them, the source may be similar; otherwise, the source may be different (Dong et al. 2019). The Pearson correlation analysis results of heavy metal(loid)s in the farmland soil around the zinc smelter are displayed in Table 4. Table 4 displays that there are very significantly positive correlations (P < 0.01) among As, Cu, Pb, and Zn, and the correlation between each other is highly correlated. Similarly, positive correlations (P < 0.01) were found among Cr, Mn, and V. Different from the other seven elements, Ni appeared in moderately positive correlation with As, Cu, and Pb (P < 0.01). At the significant level of 0.05, Ni has a positive correlation with Cr, Mn, V, and Zn, but the correlation degree is low.
To some extent, the correlation between heavy metal(loid) s reflects the similarity of pollution degree or the similar sources of heavy metal(loid)s. The correlation coefficient analysis also suggested that As, Pb, Cu, and Zn may have the same source or similar geochemical properties. Pb-Zn minerals are often accompanied by heavy metal(loid) elements such as Cu, As, etc., which are the main characteristic pollutants of nonferrous smelting industry (Félix et al. 2015;Tembo et al. 2006). The correlation coefficient analysis also indicated that Cr, Mn, and V may have the same source. In this study, Pearson correlation analysis among studied heavy metal(loid)s (As, Cr, Cu, Mn, Ni, Pb, V, and Zn) and Al 2 O 3 and Fe 2 O 3 , two major and conservative elements in the environment (Pan et al. 2017;Fan et al. 2021), was also conducted and the results (Table 4) further confirm the preceding explanation. Table 5 shows the PCA results of heavy metal(loid)s determined in the farmland soil around the zinc smelter. It can be seen from Table 5, two factors with the eigenvalues > 1 were extracted in PCA, which explains 92.2% of the total variance. The contribution rate of the first factor is 52.9%, and the factor loads of As,Cu,Ni,Pb,and Zn are 0.982,0.961,0.679,0.972,and 0.960,respectively. The second factor explains 39.3% of the total variance and the factor loads of Cr, Mn, V, and Ni are 0.956, 0.950, 0.974, and 0.575, respectively.

CA results
The CA results are shown in Fig. 6. As, Cu, Pb, Zn, and Ni are of the first category, which can be further divided into two sub-clusters, i.e., As-Zn-Pb-Cu and Ni. Cr, V, and Mn are of the second category. The results of cluster analysis are consistent with that of PCA.

Heavy metal(loid) source identification
Eight elements were divided into two groups based on the multivariate statistical analysis results together with their content characteristics in the soil. The first group of elements are As, Cu, Pb, Zn, and Ni. In Pearson's correlation analysis and PCA, these five elements have strong positive correlation and are clustered together in cluster analysis. There was basically no correlation between these five heavy metal(loid)s and Al 2 O 3 and Fe 2 O 3 except Ni, implying anthropogenic sources of these elements. Combining the content characteristics and pollution levels of As, Cu, Pb, Zn, and Ni in the soil, as well as their spatial variation trend in the downwind farmland of the zinc smelter, we think these elements in the farmland soil around the zinc smelter mainly came from the emission of zinc smelter. It should be noted that there is a certain difference between Ni and other four metals. The results of correlation analysis and PCA indicate that Ni in the farmland soil has other source yet.
The second group is composed of Cr, Mn, and V. Compared with the first group of elements, the coefficients of variation of these three metals are comparatively small, indicating that their content variations in the investigated soil samples are minor. Combining the results of multivariate statistical analysis and pollution assessment, as well as the field investigation, we consider that these metals in the farmland soil mainly originated from natural source. Al 2 O 3 and Fe 2 O 3 , two main conservative elements in the soil, are the indicators of natural source, i.e. related to the composition of the local soil parent material (Pan et al. 2017). In this study, Cr, Mn, and V are in significant positive correlation with Al 2 O 3 and Fe 2 O 3 (Table 4), which further confirms the natural source of Cr, Mn, and V in the farmland soil. Ni is positively correlated with Cr, Mn, V, Al 2 O 3 , and Fe 2 O 3 at the significant level of 0.05 (Table 4) and is moderately correlated with the second factor of PCA (with the factor load of 0.575) ( Table 5), showing that Ni in the farmland soil partly originated from natural source.

Conclusions
This study investigated the pollution, risk and source of heavy metal(loid)s in farmland soil around a zinc smelter in Shangluo City, the water source area of the Mid-route of South-to-North Water Transfer Project of China. The main findings are as follows: The farmland soils had elevated heavy metal(loid)s except Mn and the content of Zn and Pb in the downwind soils of the smelter exceeded the risk screening values for soil contamination of agricultural land of China, indicating that zinc-smelting activities have affected the environmental quality of the surrounding soils and there are certain environmental risks for heavy metal(loid)s in the soils, which are mainly caused by the atmospheric deposition of flue gas and particles discharged by the smelter; the investigated farmland was moderately to seriously polluted by Zn, moderately polluted by Pb, slightly polluted by Cr and Cu, and unpolluted by other heavy metal(loid)s; The comprehensive pollution of heavy metal(loid)s in the farmland were extreme as a whole; The comprehensive ecological risk of heavy metal(loid)s in the farmland soils was moderate as a whole, which was mainly contributed by Pb, then As, Cu, and Zn; As, Cu, Pb, and Zn in the soil mainly originated from the emission of the zinc smelter; Cr, Mn, and V primarily came from natural source, while Ni in the farmland soil mainly derived the emission of the zinc smelter and partly came from natural source. According to this research, the local government should pay attention to the elevated As, Cu, Pb, Zn, and Ni content, particularly Zn and Pb, in the farmland soil caused by zinc-smelting activities and strengthen the cooperative monitoring of heavy metal(loid)s in farmland and agricultural products, as well as pollutant emission monitoring and control, especially the control of air pollutant emission from the smelter. In addition to, the environmental impact and risk of other mining and smelting activities in the water source area should be investigated in the future. Of course, the speciation, bioavailability, leaching and precipitation capacity of heavy metal(loid) s in farmland soils should be further studied in the future to accurately evaluate the migration capacity of heavy metal(loid)s in soils and their impact on the quality of agricultural products and water environment.