Introduction

In recent years, as the process of industrialization continues to accelerate, economic social development has achieved remarkable results. Meanwhile, the problem of soil contamination has become more and more serious, especially the contamination of potentially toxic elements (PTEs) (Baltas et al., 2020; Cai et al., 2019a). The accumulation of PTEs in the soil will have a significant adverse effect on the physical and chemical properties and the biological characteristics of the soil, resulting in the loss of soil nutrients and the destruction of soil structure and function (Guan et al., 2019). In addition, soil PTEs can enter the human body through three ways—oral intake, oral and nose inhalation, and skin contact—which directly poses a huge threat to human health (Gong et al., 2019). Therefore, soil PTEs have attracted widespread attention due to their adverse effects on the ecological environment and human health. Consequently, determining the pollution status of soil PTEs and quantifying the pollution sources are vital to protecting ecology and human health.

The quantitative source allocation of soil PTEs is of great significance for the determination of priority pollution sources. The positive matrix factorization (PMF) model, as a typical receptor model, quantifies the source contribution of each PTEs and limits the non-negativity of the contribution rate on each data point. It is widely utilized in the identification of soil PTE sources (Jiang et al., 2020; Lv, 2019; Wang et al., 2019). The operation results of PMF help us to identify the contamination source of PTEs and provide a scientific basis for the prevention and control of soil PTE contamination. The potential ecological risk (PER) model combines the concentration of PTEs with their ecological and environmental effects, and toxicological effects, and quantitatively evaluates the potential risks of various PTE pollutants in sediments and soils to organisms. The human health risk (HHR) model quantifies the average daily exposure doses (ADDs) of PTEs for adults and children under the route of oral ingestion, oral and nose inhalation, and skin contact, and calculates the quotient of ADDs for each PTE and its corresponding reference dose (RfD), and the product of ADDs for each PTE and its slope factor (SF) to evaluate the non-carcinogenic and carcinogenic risk of PTE. The physical and chemical properties of soil may be different under different land use types, and the differences in soil physical and chemical properties will affect the accumulation of PTEs in the soil. Therefore, the spatial distribution and accumulation of soil PTEs in different land use types are different, and the concentrations of PTEs were also significantly different (Jin et al., 2019; Song et al., 2019; Wang et al., 2020). However, previous risk assessment studies, including ecological and health risk assessment, were mainly concentrated on a single land use type, such as industrial areas (Jin et al., 2019), farmland (Zhou et al., 2021), and mining areas (Ran et al., 2021). Few studies take into account the influencing factors of different land use types when assessing ecological and health risks. Take Daye smelting area as an example, previous investigations found that dust (Wang et al., 2021), water bodies (Zhang et al., 2014), crops (Yan et al., 2007), etc., around the smelter all accumulated PTEs to varying degrees. Therefore, it is inferred that there will also be accumulation of PTE in the soil around the smelter. But systematic studies on the accumulation characteristics of soil PTEs in different land use types around the smelters and the PER and HHR risks caused by excessive accumulation of PTEs are very rare. Accordingly, this research takes farmland, construction land, and natural land as the research objects to explore the accumulation characteristics of PTE in different land use patterns and its impact on ecological and health risks. Furthermore, because the composition of PTEs in different pollution sources is different, and the toxicity of different PTEs is significantly different, this study explored an integrated approach to quantifying risks from different sources, which combined the PMF model with the risk assessment (ecological and health risk assessment) model, and determined the pollution sources of priority control based on the contribution of PTEs from different sources to ecological risks and health risks, rather than just based on the source contribution of each source.

The purposes of this study were to (1) evaluate the concentration characteristics and contamination levels of soil PTEs under different land use patterns in the study area; (2) quantify the allocation of PTE pollution sources using the PMF model, Pearson correlation analysis, and geostatistics; and (3) quantify the ecological and health risks under different land use types by applying a comprehensive risk assessment model based on PTE pollution sources. The research results will help to determine the priority sources of ecological and health risks in the smelting area, and through scientific management and control of priority pollution sources, the risks of soil PTEs to the ecology and human health will be effectively reduced.

Materials and methods

Study region

Daye City (114°31′ ~ 115°20′ E, 29°40′ ~ 30°15′ N) is located in the southeast of Hubei Province (Fig. S1), with a total area of 1566.3 km2 and a total population of 960,000 (Cai et al., 2019b). It has a typical continental monsoon climate, with an average annual temperature of 17.3 °C and an average annual precipitation of 1007.3 mm. Daye has convenient transportation, obvious geographical advantages, and rich mineral resources. It is an important non-ferrous smelting base in China with a long history of mining. It is known as the cornucopia in the south of the Yangtze River.

Sample collection and analysis

A total of 73 surface soil samples were collected from three land use types (Fig. S1), including 25 for farmland, 27 for construction land, and 21 for natural land. The sampling locations were accurately positioned using GPS. Each sample was composed of 3–5 sub-sampling points within 100 m of the main sampling point. The original weight of the soil sample was greater than 1000 g (Jiang et al., 2021). Three soil profiles were selected in the study area to systematically study the vertical distribution of soil PTEs, with the smelter as the origin, selected P1, P2, and P3 in turn from near to far. The sampling depth of the profile was 100 cm, collected in 5 layers from bottom to top: 0–20, 20–40, 40–60, 60–80, and 80–100 cm. Among them, the same layer was gathered three times and mixed into one soil sample (Chrastny et al., 2012; Magiera et al., 2021), and 15 soil samples were collected in total. The collected samples were sealed in polyethylene bags, numbered uniformly on the sample bags, and sent to the laboratory.

The soil sample was naturally air-dried indoors and then passed through a 2-mm nylon sieve (Cai et al., 2019a). The contents of soil PTEs were determined according to soil and sediment determination of aqua regia extracts of 12 metal elements-inductively coupled plasma mass spectrometry (HJ803-2016) (MEEPRC, 2016). The pretreated sample was digested with a mixed solution of HCl-HNO3 (aqua regia), and then placed in a microwave digestion apparatus for digestion. Finally, an inductively coupled plasma mass spectrometer (ICP-MS) produced by Perkin Elmer, model NexION, was utilized to determine the content of Co, Zn, Mn, Ni, Cu, Pb, As, Cr, Cd, and Sb (Jiang et al., 2020; Wang et al., 2021). Standard samples, parallel samples, and blank samples were applied for quality control. The relative deviation of the parallel samples was less than 20%, and the recovery rate of the standard material was between 92 and 116%. And the correlation coefficients of the standard curves for PTEs were all greater than 0.999, which met the quality requirements of the multi-target regional geochemical survey specifications (DZ/T0258-2014) (CGS, 2014). In addition, the detection limits of Co, Zn, Mn, Ni, Cu, Pb, As, Cr, Cd, and Sb were 0.04, 1, 0.4, 1, 0.6, 2, 0.4, 2, 0.09, and 0.08 mg·kg−1, respectively.

The analysis of variance

The factor is denoted as A in the analysis of variance (ANOVE), and suppose that it has r levels, denoted as A1, A2, ···, Ar. The index Xij examined at each level can be regarded as a whole, and n trial results were obtained at the whole of Ai from each level. The variance is σ2, the mean value is μi, and the required normal population N (μi, σ2) (i = 1, 2, 3……r) is obtained. And ANOVE is calculated as (Exner & Berg-Beckhoff, 2013; Kozak & Piepho, 2018; Rouder et al., 2017):

$${X}_{ij}=\mu + {\partial }_{i}+{\varepsilon }_{ij}(i=1, 2,\cdots,\,r,j=\mathrm{1,2},\cdots,\,{n}_{i})$$
(1)
$${\sum}_{i=1}^{r}{n}_{i}{\partial }_{i}=0$$
(2)
$${\varepsilon }_{ij}\sim N\left(0,{\sigma }^{2}\right), \; \mathrm{each}\;{\varepsilon }_{ij} \;\mathrm{is\;independent\;of\;each\;other}$$
(3)
$$\mathrm{SST}=\mathrm{SSA}+\mathrm{SSE}{\sum}_{i=1}^{r}{\sum}_{j=1}^{{n}_{i}}{(\overline{{x }_{i}}-{\overset=x})}^{2}+{\sum}_{i=1}^{r}{\sum}_{j=1}^{{n}_{i}}{({x}_{ij}-\overline{{x }_{i}})}^{2}={\sum}_{i=1}^{r}{\sum}_{j=1}^{{n}_{i}}{({x}_{ij}-{\overset=x})}^{2}$$
(4)
$$F=\frac{MSA}{MSE}\sim F(r-1,n-r)$$
(5)

Here, the sum of squares of total error is SST, the square of factor error is SSA, and the sum of squares of random error is SSE; MSA and MSE represent the mean square of SSA and SSE, respectively. F is the test statistics. If F > Fα (r − 1, n − r), factor A is considered significant; if F ≤ Fα (r − 1, n − r), factor A is significant.

Soil PTE contamination assessment methods

The enrichment factor (EF) is used to assess the enrichment level of soil PTEs and their possible sources (Wang et al., 2021). And it is calculated as

$$EF=\frac{{{(C}_{i}/{B}_{r})}_{\mathrm{sample}}}{{{(C}_{r}/{B}_{r})}_{\mathrm{background}}}$$
(6)

where \({C}_{i}\) and \({B}_{r}\) are the concentration of the target and reference PTE, respectively. Mn was selected as the reference element for the calculation of the enrichment factor due to its characteristics of stable chemical properties and high content. We divided EF into 6 levels: not enriched (EF < 1), slightly enriched (1–2), remarkably enriched (2–5), seriously enriched (5–20), and extremely enriched (EF > 40) (Jiang et al., 2021).

Contamination factor (CF) and pollution load index (PLI) respectively represent the pollution level of a single PTE and the comprehensive pollution characteristics of PTEs (Thomas et al., 2015). CF and PLI are calculated as follows (Baltas et al., 2020; Varol, 2011):

$$CF=\frac{{C}_{i}}{{C}_{b}}$$
(7)
$$PLI=\sqrt[n]{{CF}_{1}\times {CF}_{2}\times {CF}_{3}\cdots \cdots \times {CF}_{n}}$$
(8)

Here, \({C}_{i}\) and \({C}_{b}\) represent the concentrations of PTE and its corresponding background value, respectively. PLI is classified as not contaminated (PLI < 1), moderately contaminated (1–2), seriously contaminated (2–5), and extremely contaminated (PLI > 5).

Potential ecological risk (PER) model based on PMF

The PER model based on PMF can calculate the single-factor potential ecological hazard index \({E}_{{\varvec{r}}}^{i}\) and comprehensive PER of different sources under different land use types. Firstly, the mass contribution of each PTE was computed combining the PTE source contribution rate output by PMF, which is calculated as.

$$*{C}_{jk}^{l}=n\mathrm{\%}\times {C}_{jk}$$
(9)

where *\({C}_{jk}^{l}\) is the mass contribution (mg·kg−1) of the kth PTE from the lth source in the jth sample; n% represents the ratio of the kth PTE of the jth sample from different sources, and \({C}_{jk}\) is the content (mg·kg−1) of the kth PTE of the jth sample.

Next, the PER of soil PTE from different sources under different land use types is calculated as follows (Fu et al., 2009):

$${*PER=\sum \left(*{E}_{r}^{i}\right)}_{jk}^{l}$$
(10)
$$*{E}_{r}^{i}=\sum \frac{{*C}_{jk}^{l}}{{C}_{b}}\times {T}_{r}^{i}$$
(11)

Here, \({({*E}_{r}^{i})}_{jk}^{l}\) is the potential ecological hazard index of the kth PTE of the lth source in the jth sample; \({C}_{b}\) and \({T}_{r}^{i}\) represent the background value and toxicity coefficient of PTE, respectively. In this study, the ecological risk classification criteria of soil PTE were re-adjusted to ensure the accuracy of the evaluation results, according to the toxicity and PTE types. The calculation method is as follows:

$$UTCV= \frac{{l}_{1}}{STC}$$
(12)
$${L}_{1}= {\sum}_{i=1}^{9}TC\times UTCV$$
(13)

Among them, UTCV is the unit toxicity coefficient categorization values; l1 represents the first-level borderline values of Hakanson (1980); STC is the sum of toxicity coefficients of all elements of Hakanson; and TC represents the total value of the toxicity coefficient of PTEs in this research, and the toxicity coefficients of Co, Zn, Mn, Ni, Cu, Pb, As, Cr, Cd, and Sb are 5, 1, 1, 5, 5, 5, 10, 2, 30, and 5. L1 is the adjusted first-level limit value, and then the next-level limit value was calculated based on the first-level limit value, and each level limit value is twice the upper-level limit value. PER is the reclassification standard, and the classification standard is shown in Table S1.

Human health risk model based on PMF

The PMF-based HHR model was used to calculated the human health risks (carcinogenic and non-carcinogenic risks) exposing to soil PTEs from different sources and different ways (ingestion, inhalation, and skin contact) under different land use patterns. The same as the calculation steps of PER, formula (9) needs to be run first. When calculating the HHR of natural land, farmland, and construction land, *\({C}_{jk}^{l}\) represents the average mass contribution of PTEs under the corresponding land use type. And then the ADDs of different channels should be calculated. The reference values of IngR, EF, ED, BW, AT, InhR, PEF, SA, AF, and ABS are displayed in Table S2. The ADDs are calculated as follows:

$${*ADD}_{{jk}_{ing}}^{l}=\frac{{*C}_{jk}^{l}\times IngR\times EF\times ED}{BW\times AT}\times {10}^{-6}$$
(14)
$${*ADD}_{{jk}_{inh}}^{l}=\frac{{*C}_{jk}^{l}\times InhR\times EF\times ED}{PEF\times BW\times AT}$$
(15)
$${*ADD}_{{jk}_{der}}^{l}=\frac{{*C}_{jk}^{l}\times SA\times AF\times ABS\times EF\times ED}{BW\times AT}\times {10}^{-6}$$
(16)

The non-carcinogenic risk is determined by the quotient of the ADD of each PTE from different exposure routes and the RfD. And the total hazard index (THI) is computed as follows:

$$\mathrm{THI}=\sum HI=\sum {HQ}_{{jk}_{i}}^{l}=\sum \frac{{*ADD}_{{jk}_{i}}^{l}}{{RfD}_{i}}$$
(17)

Here, HI > 1 indicates that humanity confronts the non-carcinogenic risk of a single PTE; THI > 1 demonstrates that humanity confronts the non-carcinogenic risk caused by PTEs as a whole. Elsewise, no non-carcinogenic risk.

The carcinogenic risk depends on the sum of the ADD of each PTE and the corresponding SF. And the total carcinogenic risk index (TCRI) is depicted as formula (18):

$$TCRI=\sum CR=\sum {CR}_{jki}^{l}=\sum {*ADD}_{{jk}_{i}}^{l}\times {SF}_{i}$$
(18)

When CR > 10−4 or TCRI > 10−4, indicating that humanity encounters the carcinogenic risk of a single PTE or the overall carcinogenic risk of PTEs. When CR or TCRI is between 10−6 and 10−4, it expresses that humanity is within the acceptable carcinogenic risk range. Otherwise, no carcinogenic risk. Besides, the parameters of formulas (17) and (18) are expressed in Table S3.

Results and discussion

Content characteristics of soil PTEs

The descriptive statistics of topsoil PTEs around the study area are shown in Table 1. The average concentrations of Mn, Ni, and Cr were less than corresponding background values of Hubei Province, while the average concentrations of Co, Zn, Cu, Pb, As, Cd, and Sb were all greater than their background values of Hubei Province (CNEMC, 1990), with 1.16, 2.92, 8.91, 2.81, 2.50, 19.04, and 2.29 times the background value, respectively. This indicated that Co, Zn, Cu, Pb, As, Cd, and Sb have accumulated to various degrees, especially the accumulations of Cu and Cd. The Grade II of the soil environmental quality standard is a limit value for ensuring agricultural production and protecting human health. The content of PTEs in the soil which exceeds the corresponding Grade II will cause harm to the human body. The mean contents of Zn, Cu, and Cd in the study area were high than those of homologous Grade II of the soil environmental quality standard (CEPA, 1995), with 1.22, 5.47, and 10.79 times of standard values, respectively. Among all the samples, the concentrations of Zn (34.24%), Ni (13.70%), Cu (60.27%), Pb (5.48%), As (17.81%), Cr (1.37%), and Cd (91.78%) exceeded the corresponding Grade II standard values, which pointed that the soil in the study area has been highly enriched with PTEs, especially Cu and Cd, and may endanger agricultural production and human health. Furthermore, the coefficients of variation of Zn, Cu, Pb, As, Cd, and Sb > 0.9 were highly variable. This showed that these PTEs were strongly affected by anthropogenic activities.

Table 1 Descriptive statistics of topsoil PTEs in the study area
Full size table

The ANOVE was used to explore the differences of soil PTEs concentration among three land use types (Table S4). The results showed that the average contents of Zn, Cu, Pb, Cd, As, and Sb in construction land were significantly higher than those in farmland and natural land, and the contents of Cu and As in farmland were also relatively high. There were no significant differences for the contents of Co, Mn, Ni, and Cr under different land use methods.

At 20-cm intervals, the profile soil samples around the smelting area were gathered and the sampling depth reached 100 cm. The profile distribution of PTEs in the study area is exhibited in Fig. 1. Zn, Cu, Pb, As, Cd, and Sb were significantly accumulated in the surface layer, and the contents decreased obviously from the surface layer (0–20 cm) to the subsurface layer (20–40 cm), then gradually decreased as the depth increased. This illustrated that these PTEs mainly came from anthropogenic activities such as smelting and accumulated in the top soil. The contents of Co and Mn increased with the increase of depth, and the contents of Ni and Cr varied little between layers, indicating that Co, Mn, Ni, and Cr mainly derived from natural sources.

Fig. 1
figure 1

Profile distribution of PTEs in the study area

Full size image

Contamination assessment of soil PTEs

To further determine the pollution level of PTEs in the soil, EF, CF, and PLI were calculated (Fig. 2). The EF values of soil PTEs from different land use patterns are exhibited in Fig. 2a. Under different land use patterns, the enrichment levels of PTEs were different. The average EF values of Zn, Cu, Pb, As, Cd, and Sb were all significantly greater than 1 in farmland, contractions land, and natural land, and the average EF values of soil Co were slightly higher than 1. The average EF values of Ni and Cr approximated 1 in farmland and were less than 1 in construction land and natural land. The above results illustrated that Zn, Cu, Pb, As, Cd, and Sb were obviously enriched in three land use patterns, indicating that they were significantly affected by humanity activities, while Co, Ni, and Cr showed slight and no enrichment under three land use patterns, meaning that they were likely to derive from natural source. Extraordinarily, the greatest average EF values of Zn (5.14), Cu (16.35), Pb (4.89), As (4.10), Cd (46.13), and Sb (3.92) appeared in construction land, in which Zn, Pb, As, and Sb were moderately enriched, Cu was significantly enriched, and Cd was extremely enriched. In addition, the average EF values of Cd (42.34), As (3.65), and Cu (13.25) in farmland were also relatively high, reaching extreme enrichment, moderate enrichment, and significant enrichment, respectively. The analysis indicated that soil PTEs were enriched in varying degrees around the smelting area, especially the enrichment levels of Cd and Cu which were considerable, and the pollution was serious.

Fig. 2
figure 2

Enrichment and contamination characteristics of soil PTEs in different land use patterns (the panel indicates the boxplot of enrichment and contamination factors of soil PTEs)

Full size image

The CF calculation results of soil PTEs are shown in Fig. 2b. The calculation results indicated that except Mn, Ni, and Cr, the mean CF values of other PTEs under three land use types were all > 1. This illustrated that there was no Mn, Ni, and Cr pollution in the three land use types around the smelting area. The average CF values of Zn (4.07), Cu (12.85), Pb (3.66), As (3.03), Cd (31.40), and Sb (2.88) were the largest in construction land. In addition, the average CFs of Cu, As, and Cd in farmland were 12.85, 3.03, and 31.40, respectively. It indicated that the contamination of Cd and Cu was considerable among three land use methods. The PLI were calculated based on CF values to comprehensively evaluate the pollution degree of PTEs in the study area (Fig. 2b). Farmland (1.84) and natural land (1.61) were at moderate pollution level, and construction land (2.25) was at a severe pollution level. Percentages of 76.92% of farmland, 75.00% of construction land, and 86.36% of natural land were all polluted by soil PTEs at moderate-severe levels. In summary, the study area was severely polluted by PTEs, especially Cd and Cu.

Kriging interpolation was used to make a spatial distribution map of PLI to visually show the pollution status of PTEs in the study area (Fig. S2). Figure S2 shows that the high-value area of PLI was mainly concentrated around the smelter, and the overall distribution characteristics were gradually decreasing around the smelter. This indicates that smelting activities in smelters were the main source of PTE contamination in the study area. Moreover, PLI also had a certain hot spot area in the northwest of the study area. The investigation found that there are several industrial parks of different scales, carrying out industrial production activities such as alloy processing, electroplating, mechanical manufacturing, and electronic and electrical processing. These industrial activities had caused serious PTE pollution in the region.

Source apportionment of PTEs

Correlation analysis

Pearson correlation analysis can reveal the connections between soil PTEs and provide a basis for identifying the source of PTEs. Pearson correlation coefficients of PTEs are exhibited in Table S5. Significantly positive correlations between some element pairs were discovered: Co-Mn (0.46), Co–Ni (0.69), Co-Cr (0.55), Mn-Ni (0.46), Mn-Cr (0.47), and Ni–Cr (0.51) (p < 0.01), which indicated that they may have the same source. This was consistent with the conclusion that (Martin et al., 2006) discovered Co, Mn, Ni, and Cr, as iron family elements, were generally ferrophilic, and there were high degrees of correlation among these PTEs. The correlation coefficients of Zn-Pb, Zn-Cu, Zn-As, Zn-Cd, Zn-Sb, Cu-Pb, Cu-As, Cu-Cd, Cu-Sb, Pb-As, Pb–Cd, Pb-Sb, As-Cd, As-Sb, and Cd-Sb (p < 0.01) were all > 0.45, showing a significant positive correlation. Previous analysis pointed that these PTEs were largely affected by smelting activities. The correlation coefficient between Zn and Pb was as high as 0.85, which indicated that Zn and Pb were likely to come from the same source. Studies have shown that Zn and Pb in the soil usually come from traffic emissions, such as vehicle wear and exhaust emissions (Askari et al., 2020). Similarly, the correlation coefficients of Cd-Sb (r = 0.72) and Cu-As (r = 0.71) were also larger, indicating that Cd-Sb and Cu-As have the same source, respectively.

PMF analysis

The PMF model was run 20 times, and 4 factors were finally determined according to the minimum value and stable Q value produced during the running process. Meanwhile, the residual values of PTEs from all samples were between − 3 and 3. The signal-to-noise ratio (S/N) of 10 PTEs were all greater than 2, and they were classified as “Strong,” and the fitting coefficients (R2) of PTEs were all above 0.75. This ensured the reasonableness of the effect for model fitting. The above showed that it was appropriate to use the model to explain the information contained in the initial data, and the results of the model were reliable. The factor profiles and origin contributions of PTEs based on PMF are exhibited in Fig. 3.

Fig. 3
figure 3

Factor profiles and origin contributions of soil PTEs based on PMF. a Factor 1. b Factor 2. c Factor 3. d Factor 4

Full size image

Factor 1 accounted for 19.15% of the total source contribution and was composed of Cu (23.66), As (36.50%), and Cd (28.93%) (Fig. 3a). The average contents of soil Cu, As, and Cd in the study area were 8.91, 2.5, and 19.04 times of the soil background value, respectively. The EF values of As, Cu, and Cd showed enrichment, especially in farmland, and the EF showed that As, Cu, and Cd in farmland were moderately, significantly, and extremely enriched, respectively. This indicated that the As, Cu, and Cd in the study area were principally affected by anthropogenic activities. In addition, the ANOVA denoted that the average contents of Cu, As, and Cd in farmland soil were relatively high. Therefore, Cu, As, and Cd in the soil of the study area probably related to agricultural activities. As an additive to livestock feed (Guan et al., 2018), Cu can enter the soil along with livestock manure. Luo et al. (2009) reported that 69% of Cu in the soil came from livestock manure. In addition, the application of Cu-containing pesticides such as bordeaux mixture can also cause Cu accumulation in the soil (Cai et al., 2019a). Cd is widely present in phosphate fertilizers as an impurity (Cai et al., 2019b). According to statistics, the total amount of Cd bringing into the farmland through phosphate fertilizers was estimated to be as high as several hundred tons in the past 30 years (Zhuang, 2015). Furthermore, nitrogen and potash fertilizers also contain a certain amount of Cd (Liu et al., 2021). Consequently, the application of chemical fertilizers was an important reason for the accumulation of soil Cd. As is generally regarded as the identifying element of agricultural activities. The applications of phosphate fertilizer and zinc sulfate fertilizer were the primary way for As to migrate into the soil (Qi & Donahoe, 2008). And inorganic arsenic compounds, such as calcium arsenate, lead arsenate, and sodium arsenate, were widely utilized in pesticides and herbicides (Liu et al., 2021). Furthermore, local farmers abused chemical fertilizers and pesticides, resulting in the loss of large amounts of chemical fertilizers and pesticides to the soil. Accordingly, factor 1 was attributed to agricultural activities.

Factor 2 was dominated by Zn (66.98%) and Pb (58.88%), explaining 23.07% of the total source contribution (Fig. 3b). The average EF values of Zn and Pb in construction land were significantly higher than those of farmland and natural land. And the spatial variation trends of Zn and Pb were relatively consistent (Fig. S3), with hot spots close to main roads and smelting areas with frequent transportation. Studies indicated that the accumulations of Zn and Pb in the soil were usually related to traffic emissions (Chen et al., 2010; Kaur et al., 2021). Zn is utilized as a reinforcing agent and activator in the tire manufacturing industry. The wear of the tire releases Zn-containing particles. Furthermore, the corrosion of automobile galvanized parts and the wear of zinc alloy parts also release a large amount of Zn into the environment. Pb, as the identification element of transportation, mainly came from the emission of automobile exhaust. In addition, the wear of vehicle engines and brakes and the application of lead-acid batteries were also the main reasons for the accumulation of soil Pb. In summary, factor 2 was considered to be traffic emissions.

Factor 3 was characterized by Co (46.91%), Mn (60.89%), Ni (58.06%), and Cr (59.37), which accounted for 33.46% of the total source contribution (Fig. 3c). Among these PTEs, except for the average content of Co which was slightly higher than the soil background value, the average contents of the other PTEs were all lower than their soil background value. Meanwhile, EF exhibited that these PTEs were at no and slightly enriched levels under three land use types. In addition, the coefficients of variation of Co, Mn, Ni, and Cr also showed that they were relatively less affected by human factors but were largely affected by natural factors such as geological background and topography. This was coincident with the research results of Zhuang and Lu (2020), Fang et al. (2021a), and Fang et al. (2021b). Based on the above analysis, the PTEs in factor 3 were mainly derived from natural sources.

Factor 4 accounted for 31.33% of the total factor contribution, of which PTEs with high contribution rates were Co (36.06%), Cu (74.25%), As (52.21%), Cd (56.87%), and Sb (60.47%) (Fig. 3d). The average contents of soil Cu, As, Cd, and Sb in the study area were 8.91, 2.50, 19.04, and 2.29 times their soil background value, respectively, and the accumulation was serious. In the three land use types, EF exhibited that Cu, As, and Cd reached medium or above enrichment level. And Sb and Co were slightly or above enriched. Besides, the spatial distribution of these PTEs was relatively similar, and high-value areas were all mainly located near smelters (Fig. S3). This indicated that Co, Cu, As, Cd, and Sb were mainly affected by human factors such as smelting activities. Substantial particles containing Co, Cu, As, Cd, and Sb were produced during smelting activities, which entered the soil through atmospheric deposition. In fact, besides the impact of smelting activities, there were alloy processing, electroplating, machinery manufacturing, plastic production and other industries near the smelter. The three wastes generated by these industrial activities were also important reasons for the accumulation of soil PTEs in the study area. Co is often used as a chemical analysis reagent in electroplating, battery, alloy industry, and pigment industry (Chen et al., 2019). The combustion of coal was generally considered to be an important source of As in the environment. According to the survey, the As content from coal in China ranges from 0.24 to 70.83 mg·kg−1 (Wang et al., 2005). Moreover, As was also utilized as alloy material in copper and lead alloys (Men et al., 2018; Sun et al., 2019). A large amount of Cd was discharged into the surrounding environment in the form of three wastes through industrial activities such as electroplating, alloy processing, and paint and machinery manufacturing, and eventually pollutes the soil (Jiang et al., 2020). Sb was widely applied in varieties of industries, such as hardeners in the production of metals and alloys, and flame retardants in the production of textiles and plastics (Jiang et al., 2017). Consequently, factor 4 was identified as industrial activities.

Potential ecological risk estimation from different sources

The potential ecological risks of soil PTEs from 4 sources under 3 land use patterns in the study area were quantitatively analyzed, and the results are shown in Table 2. The PER of farmland (505.91), construction land (862.59), and natural land (370.46) all reached extremely high ecological risk levels. Under different land use patterns, the contribution of different PTE sources to PER showed a similar pattern: factor 4 >  > factor 1 > factor 2 > factor 3, and factor 4 accounted for 55.66%, 56.30%, and 55.36% of farmland, construction land, and natural land, respectively. This was because the toxicity coefficients of Cd and As emitted by factor 4 (industrial activities) were much higher than those of other PTEs. Furthermore, the contributions of different PTEs to PER in different land use patterns were all expressed as Cd >  > Cu > As > Pb > Sb > Co > Ni > Zn > Cr > Mn. In conclusion, industrial activities were the largest contributor to PER in the study area, and Cd was the most dangerous element for PER. Consequently, priority should be given to monitoring and managing the smelting activities of the smelter and other surrounding industrial activities in order to reduce the PER of soil PTEs around the smelting area.

Table 2 Potential ecological risks of soil PTEs from different sources under different land use patterns
Full size table

Health risk assessment from different sources

Based on the source contribution of PTEs, the humanity carcinogenic and non-carcinogenic risks exposed to PTEs from different sources under different land use patterns and different exposure pathways were quantified (Tables 3 and S6; Fig. 4). In this study, seven soil PTEs, Zn, Ni, Cu, Pb, As, Cr, and Cd, were selected for non-carcinogenic risk assessment (US EPA, 2011). From Table S6, the hazard quotient (HQ) of different exposure routes of these soil PTEs is presented as oral intake risk > skin contact > oronasal inhalation. Therefore, oral intake was the main exposure route for non-carcinogenic risks. The hazard index (HI) value of each PTE for adults and children was shown as As >  > Cr > Pb > Cu > Cd > Ni > Zn. The HI values of all PTEs for adults were less than 1, illustrating that none of the non-carcinogenic risks caused by Zn, Ni, Cu, Pb, As, Cr, and Cd had a significant impact on adults. However, the HI value of As for children was 1.32 greater than 1, indicating that As had caused serious non-carcinogenic damage to local children. Furthermore, Ni, Pb, As, Cr, and Cd were selected to estimate the carcinogenic risk (US EPA, 2011). The carcinogenic risk (CR) values of As, Cr, and Cd exhibited oral intake risk > skin contact > oral and nasal inhalation. Therefore, oral intake was also the primary way of carcinogenic risk. The CR values of the PTEs for adults and children were As > Cr >  > Cd > Pb >  > Ni. Among them, the CR values of As and Cr for adults and As, Cr, and Cd for children were within the carcinogenic risk threshold range (1 × 10−6–1 × 10−4), and Cd, Pb, and Ni for adults and Pb and Ni for children had no carcinogenic risk.

Table 3 Non-cancer and cancer risk of soil PTEs from different contamination sources under different land use patterns
Full size table
Fig. 4
figure 4

The contribution of various pollution sources to human health risks (noncancer risk and cancer risk) of children (a) and adults (b) under different land use patterns

Full size image

As shown in Table 3, the non-carcinogenic risks values (THI) of adults were all < 1 under three land use patterns. However, the THI of children were all > 1, which were 2.02 for farmland, 2.56 for construction land, and 1.60 for natural land. This indicated that under three land use types, children were exposed to serious non-carcinogenic risks, while the non-carcinogenic risks to adults were still at an acceptable level. Except the carcinogenic risk (TCRI) of children in construction (1.06E − 04) land exceeded 10−4, the TCRI of children in farmland and natural land and the TCRI of adults in three land use types were all within the carcinogenic risk threshold range (1 × 10−6–1 × 10−4). This revealed that children had a slight carcinogenic risk in construction land. Besides, the TCRI of children in farmland (9.06 × 10−5) was very close to the threshold (1 × 10−4), which required attention. The carcinogenic and non-carcinogenic risks of children from four different sources were distinctly higher than those of adults, which were because the special behaviors of children are more likely to be exposed to PTEs (Cai et al., 2019b). Thence, it is vital to guide children to develop good hygiene habits. Moreover, the carcinogenic and non-carcinogenic risks of three land use patterns were all expressed as construction land > farmland > natural land.

As shown in Fig. 4, there were similar trends for carcinogenic and non-carcinogenic risks from the four sources under the three land use types. More accurately, both for adults or children, industry activities (factor 4) were the primary source of PTEs under the three land use types, which were represented by children: 41% of farmland, 42% of construction land, and 40% of natural land, as well as by adults: 41% of farmland, 43% of construction land, and 40% of natural land. In addition, the contribution of agricultural practices (factor 1) to the non-carcinogenic risk of adults and children was second only to industrial activities. For the carcinogenic risk of adults and children, the four factors in farmland and natural land were all expressed as factor 4 > factor 3 > factor 1 > factor 2, and in construction land, they were expressed as factor 4 > factor 1 > factor 3 > factor 2. Therefore, industrial activities were the priority source under the three land use types, followed by agricultural practices in construction land and nature source in farmland and natural land. Thus, optimizing the industrial structure based on smelting activities and improving the treatment level of industrial three wastes are essential to protect humans from carcinogenic risks. And the management and control of agricultural activities cannot be ignored.

As shown in Figs. 3 and 4, it can be concluded that the human health risks caused by industrial activities were significantly higher than other sources, but industrial activities only rank second in the contribution of PTEs sources. It indicated that the main source of PTEs was not necessarily the biggest contributor to health risks, and this is perhaps due to that the As and Cr released by industrial activities were more dangerous (Gong et al., 2019). Therefore, if the priority pollution sources are determined based on source contribution rate of PTEs, human health may not be effectively protected. Furthermore, this study lacks in-depth research on the occurrence form of soil PTEs in smelting areas, PTEs in major crops, and population health risks. Therefore, the next stage of investigation and research can be carried out in the soil-crop system in the region.

Conclusions

This study explored the pollution level of soil PTEs, and quantified the ecological risks and human health risks of 4 sources under farmland, construction land, and natural land. The results of PLI and EF exhibited that farmland (1.84) and natural land (1.61) were moderately polluted and construction land (2.25) was severely contaminated, and the pollution of Cd and Cu was worst. Four sources of PTEs were allocated by the PMF model: nature source (33.46%) > industries activities (31.33%) > traffic emission (23.07%) > agricultural practice (12.14%). Ecological and health risks from 4 sources under three land use patterns were quantified through PER and HHR models based on PMF. PER of farmland (505.91), construction land (862.59), and natural land (370.46) all reached extremely high risk, and industrial activities were the main source of PER. Oral intake was the main exposure route for non-carcinogenic and carcinogenic risks. The THIs of children were all greater than risk limit under three land use patterns, children were exposed to serious non-carcinogenic risks, and As had the greatest impact on non-carcinogenic risk of children. The TCRI of children in construction land exceeded 10−4, children were exposed to slight carcinogenic risk in construction land, and the carcinogenic risk (9.06E − 05) of children in farmland also needs to attract attention. Adults have no obvious health risk. Industrial activities were the cardinal contributor to the health risks of adults and children in the study area, and As and Cr were the primary carcinogenic risk elements. Accordingly, priority control and management of industrial activities dominated by nonferrous metal smelting are significantly necessary to protect the ecological environment and human health. And the management and control of agricultural activities cannot be ignored.