Source Apportionment and Ecological-Health Risk Assessments of Potentially Toxic Elements in Topsoil of an Agricultural Region in Southwest China

Abstract

Soil potentially toxic element (PTE) contamination remains a global concern, particularly in rural agricultural regions. This study collected 157 agricultural topsoil samples within a rural area in SW China. Combined with multivariate statistical analysis in the compositional data analysis (CoDa) perspective, the PMF model was applied to identify key contamination sources and quantify their contributions. Potential ecological risk assessment and Monte Carlo simulation were employed to estimate ecological-health risks associated with PTE exposure. The results revealed that the main exceeding PTEs (Mercury—Hg and Cadmium—Cd) are rich in urbanized areas and the GFGP (Grain for Green Program) regions. Source apportionment indicated that soil parent materials constituted the dominant contributor (32.48%), followed by traffic emissions (28.31%), atmospheric deposition (21.48%), and legacy agricultural effects (17.86%). Ecological risk assessment showed that 60.51% of soil samples exhibited higher potential ecological risk (PERI > 150), with moderate-risk areas concentrated in the GFGP regions. The elements Cd and Hg from legacy agricultural effects and atmospheric deposition contributed the most to ecological risk. Health risk assessment demonstrated that most risk indices fell within acceptable ranges for all populations, while only children showed elevated non-carcinogenic risk (THI_max > 1.0). Among PTEs, the element As, mainly from traffic emissions, was identified as a priority control element due to its significant health implications. Geospatial distributions showed significant risk enrichment in the GFGP regions (legacy agricultural areas). These findings present associated risk levels in sustainable agricultural regions, providing valuable data to support soil environmental management in regions requiring urgent intervention worldwide.

1. Introduction

Soil, as the material basis for ecosystem and agricultural productivity, plays a critical role in sustaining global socio-economic development [1,2,3,4]. However, rapid industrialization and urbanization have led to the widespread accumulation of potentially toxic elements (PTEs) in soils through anthropogenic activities, including mining operations, fossil fuel combustion, and intensive application of agrochemicals [5,6,7]. These contaminants, characterized by high toxicity, persistence, and bioaccumulation potential [8,9], not only deteriorate the soil environment and disrupt ecological balance but also pose substantial threats to human health via the food chain [10,11,12,13]. Recent studies have shown severe ecological and health impacts associated with soil PTEs across various regions. Refs. [14,15,16,17] found that Hg and Cd were particularly concerning elements in agricultural soils of western Asia, demonstrating significant ecological-health risks. Refs [18] reported that Cd and As contamination in Indian soils caused irreversible DNA damage in plants and elevated carcinogenic risks to human health. Notably, sustainable agricultural development (SAD), such as China’s Grain for Green Program (GFGP) and pollution-free stereoscopic agriculture, has introduced novel paradigms for soil environmental management [19,20]. While these policies have demonstrated efficacy in reducing PTE concentrations [21,22], legacy agricultural effects still cannot be underestimated. Persistent exceedance of PTEs in soils has been observed in some afforested areas [23,24,25,26], meaning that seemingly safe areas may still pose risks to human health or ecosystems. Therefore, the source apportionment and comprehensive risk assessments associated with PTEs need to be quantified for these regions.

Recently, although the research about soil PTEs has made some advanced progress, there are some challenges in accurately quantifying their sources and assessing associated ecological-health risks [27,28]. The main obstacle is attributed to the limitations of traditional source apportionment methods. These approaches often fail to account for the compositional nature of geochemical data, leading to spurious correlations and influenced results [29,30]. For instance, the results of both correlation analysis [31] and principal component analysis (PCA) are usually affected by closure effects [32,33,34]. This limitation obscures the true relationships between PTEs and their sources, especially in complex environments with mixed natural and anthropogenic inputs. On the other hand, ecological environmental improvement is generally evaluated based on visible outcomes such as increased vegetation cover [35]. Furthermore, health risk assessments used to depend on deterministic models, which overlook the variability and uncertainty in exposure parameters such as ingestion rate (Ring), body weight (BW), and exposure frequency (EF) [36,37]. This may make the results of risk estimations inaccurate, particularly for vulnerable populations such as children [38,39].

To address these challenges, recent advancements in compositional data analysis (CoDa) and receptor models have provided valuable solutions for source apportionment and risk assessments [40,41]. CoDa, through transformations such as the centered log-ratio (clr) and isometric log-ratio (ilr), avoids the closure effects in geochemical data, enabling reliable multivariate statistical analysis [42,43]. Meanwhile, receptor models such as Positive Matrix Factorization (PMF) have been powerful tools for quantifying source contributions by decomposing complex raw datasets into source profiles and contributions [44,45]. Additionally, potential ecological risk assessment (PERI) objectively reflects the potential impact on the ecosystem by combining with the toxicity coefficients of the PTEs [46,47,48]. The probabilistic risk assessment method, Monte Carlo simulation, has been increasingly adopted due to considering parameter uncertainty and variability, providing more realistic evaluations of health risks [49,50,51]. For instance, recent studies have integrated CoDa and PMF to distinguish between natural and anthropogenic sources in urban and agricultural soils [41]. PERI is carried out to qualify risk levels of PTEs in soil ecosystems, while Monte Carlo simulations have been used to identify vulnerable populations and prioritize pollutants [52,53,54]. These methods can overcome the limitations of traditional approaches.

The accumulation of PTEs in agricultural soils has emerged as a pressing environmental concern, posing significant threats to ecosystem stability, food safety, and human health [55,56]. In response, the implementation of SAD policies has been a global trend as a strategy to rehabilitate degraded lands and safeguard the soil environment. While these measures have effectively mitigated excessive PTE concentrations in soils, the long-term legacy effects of historical agricultural practices remain understudied, particularly in SAD regions where quantitative risk assessments and source apportionment approaches are limited. Previous studies have primarily focused on PTE contamination in heavily polluted areas, often neglecting regions undergoing sustainable remediation. Moreover, conventional source apportionment methods frequently ignore the compositional nature of soil data, leading to biased interpretations. Crucially, integrated frameworks combining CoDa, advanced receptor modeling, and probabilistic health risk assessment are scarce, leaving gaps in realistic risk evaluations for policy-making. To fill these gaps, the detailed objectives are determined as follows:

• To understand pollution levels and geospatial characteristics of PTE concentrations;
• To accurately identify the main sources and quantify contributions by the PMF model combining multivariate statistical analysis in the CoDa perspective;
• To assess potential ecological risk associated with PTEs through PERI;
• To obtain more realistic assessments of health risks using Monte Carlo simulations.

These findings highlight the effects of sustainable agricultural practices in reducing the risks of PTEs to human health and ecosystems, offering critical data-driven insights to support soil environment management in regions worldwide facing similar environmental challenges.

2. Materials and Methods

2.1. Study Area

The study area is located in southwest China and belongs to the rural agricultural area (Figure 1a), which belongs to a subtropical humid monsoon climate. The mean annual temperature and precipitation reach 15.5 °C and 1000 mm, respectively. The terrain exhibits significant elevational variation (611–1346 m), sloping from higher elevations in the north to lower areas in the south. The study area is predominantly underlain by Jurassic strata, including the Suining Formation, Ziliujing Formation, Lianhuakou Formation, and Shaxi Formation, with sparse Quaternary sediments overlying these units. The lithology is predominantly composed of mudstones and sandstones. The strata exhibit a northeast–southwest structural strike (Figure 1c). Land use patterns show distinct spatial differentiation, with forest dominating the northern sector and built-up areas mixed with cropland in the southern sector. Except for the northernmost areas, parts of the northern forest are the GFGP regions (Figure 1b). There are multiple types of soils, particularly paddy soils, accounting for 35% of the total land area in the region. The paddy soils are determined as cropland, which has a proportion of 75.7% of the total cropland areas.

Before implementing the policy, a field investigation found that the local survival focused on traditional agriculture, causing the overuse of agrochemicals, especially phosphate fertilizers. In the past decades, large numbers of studies showed that the PTE concentrations of the study area exhibited significantly exceeding features [57,58]. Nowadays, the economic development in the study area is dominated by pollution-free stereoscopic agriculture, followed by stock farming. The GFGP regions are covered by various commercial crops, including fruits and excellent forages. Additionally, the southern cropland is primarily planted with pollution-free vegetables. These product industries not only reduce excessive rates of soil PTEs but also make tourist industries gradually become the pillar of the economy in this region. There are rich coal reserves in the city governing the study area, and some thermal power stations are built in the urbanized areas. Because of the unique climate, the study area has no source of heavy pollution. Abundant groundwater and surface reservoirs supply the demand of the social economy and daily life with good water quality.

2.2. Soil Sampling and Testing

To ensure the representativeness of soil samples, a systematic network point method was adopted, collecting 157 samples across the study area in January 2022. Each sample, obtained from a depth of 0–20 cm, adhered to a density of 20 samples per 1 km² and a minimum mass of 1 kg. Sampling sites were strategically selected in areas with minimal anthropogenic disturbance to reflect natural conditions. Surface debris was removed before collection, and samples were sealed in pre-cleaned, contamination-free bags. The collected samples were air-dried, sieved, and transported to the laboratory for chemical analysis. The analyzed parameters included the concentrations of PTEs (Arsenic—As, Cadmium—Cd, Chromium—Cr, Copper—Cu, Mercury—Hg, Nickel—Ni, Lead—Pb, and Zinc—Zn), as well as soil pH.

Elemental analysis was conducted using advanced spectroscopic techniques. All analytical-grade reagents used in this study were sourced from China National Pharmaceutical Group Corporation (Sinopharm Group) (Beijing, China) to ensure trace metal-grade purity. For the determination of Cr, Cu, Ni, and Zn, approximately 0.1000 g of soil sample was moistened with distilled water (18.2 MΩ·cm), digested with a mixture of HNO₃ (1.42 g/mL), HF (1.15 g/mL), and HClO₄ (1.67 mg/L), and extracted using 5% HNO₃ before dilution to 25.00 mL and homogenization. The concentrations of these elements were quantified via Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES, iCAP 7600, ThermoScientific, Waltham, MA, USA). For Cd and Pb, 5.00 mL of the supernatant was diluted to 20.00 mL with low-concentration HNO₃ and measured using Inductively Coupled Plasma Mass Spectrometry (ICP-MS, Agilent 7500, ThermoScientific, USA). As and Hg were determined by first oven-drying the sample at 105 °C, followed by digesting 0.2000 g of the dried sample with 10 mL of aqua regia [HNO₃ (1.42 g/mL): HCl (1.19 g/mL) 1:3, v:v] in a 50 mL container; their concentrations were determined by Atomic Fluorescence Spectrometry (AFS, PSAnalytical, Orpington, UK). Soil pH was measured (soil/water ratio of 1:2.5) using the Ion Selective Electrode (ISE) method after suspending 10.0 g of sample in deionized water, stirring vigorously for 1 min, and allowing the mixture to settle for 30 min. The detection limits of these PTEs are 1 mg/kg for As, 0.03 mg/kg for Cd, 5 mg/kg for Cr, 1 mg/kg for Cu, 0.0005 mg/kg for Hg, 2 mg/kg for Ni, 2 mg/kg for Pb, and 4 mg/kg for Zn.

To ensure data reliability, a rigorous quality assurance and quality control (QA/QC) protocol was implemented. Soil samples were analyzed in three batches following the ‘Specification for Multi-Purpose Regional Geochemical Survey (1:250,000)’ (DZ/T 0258-2014) [59], with certified Chinese national standard reference materials (GBW) used for instrument calibration and method validation. The accuracy of the analytical methodology was assessed through the relative deviation between measured and certified values, while precision was evaluated using the coefficient of variation (CV). Both accuracy and precision achieved 100% compliance rates. Additionally, replicate analyses were performed for every batch of 30 samples, with accuracy and precision exceeding 98% for all PTEs. The overall reporting rate for soil samples surpassed 99%, and the repeatability test compliance rate remained above 90%, ensuring the robustness and reproducibility of the results.

2.3. Improved Nemerow Index

The Improved Nemerow Index (INI) is employed to evaluate the pollution level of PTEs in soils [60,61]. This enhanced approach addresses the limitations of the conventional Nemerow index by incorporating both the mean and maximum contamination factors [62]. The calculation procedure is shown by following the formulas.

$$\mathrm{I}\mathrm{g}\mathrm{e}\mathrm{o}={\mathrm{l}\mathrm{o}\mathrm{g}}_2{\displaystyle \frac{C_i}{1.5B_i}}$$

(1)

$$\mathrm{I}\mathrm{N}\mathrm{I}=\sqrt{{\displaystyle \frac{{\mathrm{I}}^2\mathrm{g}\mathrm{e}\mathrm{o}\left(\mathrm{m}\mathrm{a}\mathrm{x}\right)+{\mathrm{I}}^2\mathrm{g}\mathrm{e}\mathrm{o}\left(\mathrm{a}\mathrm{v}\mathrm{e}\right)}{2}}}$$

(2)

where C_i represents the measured concentration of PTE i in soil (mg/kg), and B_i denotes the corresponding background value or soil quality standard (mg/kg). In this study, to ensure the accuracy and comparability of the research results, local background levels are selected to calculate the Igeo (geochemical accumulation index). The Igeo is applied to assess the contamination degree of PTEs in soils, considering both their measured concentrations and natural background levels [63]. Igeo(max) and Igeo(ave) are the maximum and average values of the geochemical accumulation indexes, respectively. The detailed classifications of the INI and Igeo are shown in

Table A1. Classification of the PTE contamination risk assessment indexes.

Classification	Description	Classification	Description
Igeo < 0	No contamination	INI ≤ 0.5	No pollution
0 ≤ Igeo < 1	Slight to moderate contamination	0.5 < INI ≤ 1	Warning limit
1 ≤ Igeo < 2	Moderate contamination	1 < INI ≤ 2	Slight pollution
2 ≤ Igeo < 3	Moderate to high contamination	2 < INI ≤ 3	Moderate pollution
3 ≤ Igeo < 4	High contamination	INI > 3	High pollution
4 ≤ Igeo < 5	High to extreme contamination
Igeo ≥ 5	Extremely serious contamination

.

2.4. Compositional Data Analysis

In this study, multivariate statistical analysis includes correlation analysis and principal component analysis (PCA), which are employed to elucidate the relationships among PTE concentrations and the pollutant sources in the soils. Before this analysis, the compositional nature of the data is disposed of using compositional data analysis (CoDa) to avoid unreasonable results arising from closure effects. CoDa is a robust statistical framework specifically designed to handle data that represents parts of a whole. This method emphasizes the transformation of soil PTE concentrations and is realized by multiple pathways, mainly including Centered Log-Ratio (clr), Additive Log-Ratio (alr), and Isometric Log-Ratio (ilr) transformations. In this study, raw PTE concentrations were transformed using the clr transformation to address the issue of closure effects. The mentioned transformation methods are calculated as follows.

$$clr\left(x\right)=\left(\begin{array}{c}ln{\displaystyle \frac{x_1}{g\left(x\right)}},ln{\displaystyle \frac{x_2}{g\left(x\right)}},.,ln{\displaystyle \frac{x_D}{g\left(x\right)}}\end{array}\right)$$

(3)

$$alr\left(x\right)=\left(\begin{array}{c}ln{\displaystyle \frac{x_1}{x_D}},ln{\displaystyle \frac{x_2}{x_D}},.,ln{\displaystyle \frac{x_{D-1}}{x_D}}\end{array}\right)$$

(4)

$$ilr\left(x\right)=Vclr\left(x\right)$$

(5)

where $x_1$, $x_2$, …, and $x_D$ express the concentrations of the first to Dth components in the composition. D is the total number of components in the composition. $g\left(x\right)$ means the geometric mean of all components in the composition $x=(x_1, x_2, \dots , x_D)$. V represents a matrix which defines an orthonormal coordinate system within the clr-plane.

2.5. Ecological Risk Index

The potential ecological risk index (PERI), initially developed by [64], was employed to evaluate the ecological impact of PTEs in soil ecosystems. The method quantifies the risk through two key indices: the single factor pollution index (P_i) and PERI. The formulas are defined as follows:

$$P_i={\displaystyle \frac{C_i}{S_i}}$$

(6)

$$E_i=T_i\times P_i$$

(7)

$$\mathrm{P}\mathrm{E}\mathrm{R}\mathrm{I}={\sum }_{i=1}^n{E}_i$$

(8)

where C_i represents the measured concentration of PTE i in the soils, S_i denotes the reference concentration of PTE i, and T_i is the toxicity coefficient of the PTE i, reflecting its inherent toxicity and the soil’s sensitivity to contamination. Local background levels are determined as the reference concentrations of the PTEs. The toxicity coefficients are 10 for As, 30 for Cd, 2 for Cr, 5 for Cu, 40 for Hg, 5 for Ni, 5 for Pb, and 1 for Zn. E_i represents the ecological risk for a single PTE. The PERI is the integrated ecological risk index, summing the contributions of all PTEs. The detailed classifications [65] are shown in

Table 1. The classification of E_i and PERI.

Ei	PERI	Classification
<40	<150	Low risk
40–80	150–300	Moderate risk
80–160	300–600	Considerable risk
160–320	600–1200	High risk
≥320	≥1200	Extremely high risk

.

2.6. Positive Matrix Factorization Model

The Positive Matrix Factorization (PMF) model, conducted with EPA PMF 5.0 software, is a receptor modeling tool that utilizes chemical composition data to identify pollution sources and qualify their contributions to soil environment degradation [66]. This method decomposes the original concentration matrix $X_{ij}$ into three key components: the source contribution matrix $g_{ik}$, the source profile matrix $f_{ki}$, and the residual error matrix $e_{ij}$. The decomposition process aims to minimize the objective function Q, which quantifies the difference between the observed and predicted concentrations. A lower Q value indicates a better fit between the model and the observed data. The mathematical formulation of the PMF model is expressed as:

$$X_{ij}={\sum }_{k=1}^p{g}_{ik}{f}_{kj}+e_{ij}$$

(9)

$$Q={\sum }_{i=1}^n{\sum }_{j=1}^m{\left({\displaystyle \frac{X_{ij}-{\sum }_{k=1}^p{g}_{ik}{f}_{kj}}{{\sigma }_{ij}}}\right)}^2$$

(10)

where n and m represent the number of the samples and the element species, respectively; ${\sigma }_{ij}$ is considered as the uncertainty of the jth element in the ith samples, which can be gained by the following formula: the MDL, concentration, SD, $d_j$, and $c_j$ express the detection limit, the element concentration, the standard deviation, and the ratio coefficient of uncertainty. If the concentration is lower than the MDL, ${\sigma }_{ij}$ of that can be referred to as five-sixths of the MDL.

$${\sigma }_{ij}=MDL+d_j\times X_{ij}$$

(11)

$${\sigma }_{ij}=\sqrt{{(c_j\times SD)}^2+{(0.05\times X_{ij})}^2}$$

(12)

2.7. Health Risk Assessment Model

Health risk assessment, a methodology established by the United States Environmental Protection Agency (US EPA) [67], is employed to quantify the potential adverse effects of environmental pollutants on human health. In this study, elevated concentrations of PTEs in soils are evaluated for their health risks through multiple exposure pathways: direct ingestion (ing), dermal exposure (der), and oral nasal inhalation (inh). These pathways contribute to the average daily dose (ADD) of PTEs absorbed by humans, which serves as a critical metric for risk quantification. The ADD for each exposure route is calculated using the following equations:

$$AD{D}_{ing}=C_s\times {\displaystyle \frac{R_{ing}\times EF\times ED}{BW\times AT}}\times {10}^{-6}$$

(13)

$$AD{D}_{inh}=C_s\times {\displaystyle \frac{R_{inh}\times EF\times ED}{PEF\times BW\times AT}}$$

(14)

$$AD{D}_{der}=C_s\times {\displaystyle \frac{AF\times SA\times ABS\times EF\times ED}{BW\times AT}}\times {10}^{-6}$$

(15)

where C_s expresses the concentration of the PTEs; these parameters, including Ring, Rinh, EF, ED, SA, AF, ABS, PEF, AT, and BW, are shown in

Table A2. Input parameters to characterize the daily exposure dose of PTEs via various exposure pathways.

Parameter		Value	Unit	References
Ingestion rate (Ring)	children	200	mg/day	[123]
	adults	100
Inhalation rate (Rinh)	children	7.5	m3/day	[123]
	adults	14.5
Exposure frequency (EF)	-	350	day/a	[124]
Exposure duration (ED)	children	6	a	[124]
	adults	24
Exposed skin area (SA)	children	2448	cm2	[124]
	adults	5075
Skin adherence factor (AF)	children	0.2	mg/m2·day	[123]
	adults	0.07
Dermal absorption factor (ABS)	-	0.001	unitless	[123]
Particle emission factor (PEF)	-	1.36 × 109	m3/kg	[123]
Average exposure time (AT)	carcinogens	70 × 365	days	[125]
	non-carcinogens	ED × 365
Average body weight (BW)	children	15.9	kg	[124]
	adults	56.8

.

HI and CR represent the non-carcinogenic risk index and carcinogenic risk index, respectively. The calculated equations are as below.

$$H{I}_i={\sum }_i^n{\displaystyle \frac{AD{D}_i}{Rf{D}_i}}$$

(16)

$${CR}_i=\sum _i^n(AD{D}_i\times {SF}_i)$$

(17)

$$\mathrm{T}\mathrm{H}\mathrm{I}=\sum _i^{n}H{I}_i=\sum (H{I}_{ing}+H{I}_{inh}+H{I}_{der})$$

(18)

$$\mathrm{T}\mathrm{C}\mathrm{R}={\sum }_i^{n}C{R}_i$$

(19)

where n is the number of PTEs, and HI_i and CR_i denote the total non-carcinogenic risk index and carcinogenic risk index of all exposure pathways for a single PTE, respectively. RfD_i and SF_i are considered the non-carcinogenic and carcinogenic average daily reference dose for PTE i, respectively (

Table A3. Values of the reference dose (RfD; mg/kg/day) and the slope factor (SF; per mg/kg/day) for PTEs [126].

PTE	RfD (mg/kg/day)			SF (per mg/kg/day)
	Ingestion	Inhalation	Dermal	Ingestion	Inhalation	Dermal
Cr	3.00 × 103	2.86 × 105	6.00 × 105	−	−	−
Hg	3.00 × 104	7.04 × 105	2.10 × 105	−	−	−
Ni	2.00 × 102	9.00 × 105	5.40 × 103	−	−	−
Cu	4.00 × 102	4.00 × 102	1.20 × 102	−	−	−
Zn	3.00 × 101	3.00 × 101	6.00 × 102	−	−	−
As	3.00 × 104	1.23 × 104	1.20 × 104	1.5	15.1	3.66
Cd	1.00 × 103	2.86 × 106	2.50 × 105	6.1	6.3	20
Pb	3.00 × 103	3.52 × 103	5.25 × 104	8.50 × 103	4.20 × 102	4.25 × 101

). Based on the calculated values of the total non-carcinogenic risk (THI) and total carcinogenic risk (TCR), health risks are categorized into distinct grades. For THI, risks are classified as negligible (THI < 1.00) or unacceptable (THI ≥ 1.00). Similarly, TCR values are categorized into negligible risk (TCR < 1.0 × 10⁻⁶), acceptable risk (1.00 × 10⁻⁶ < TCR < 1.00 × 10⁻⁴), and unacceptable risk (TCR > 1.00 × 10⁻⁴). To further refine the risk assessment, a Monte Carlo simulation is employed to evaluate the probabilistic health risks associated with soil PTEs. This method is achieved through using the following Python 3.11.5 packages: Pandas 2.2.3 [68], Numpy 1.26.4 [69], Scipy 1.10.1 [70], and Matplotlib 3.7.1 [71]. The probability distributions of key input parameters, including exposure frequency, body weight, and ingestion rate, are detailed in

Table A4. Distribution settings for each parameter in the Monte Carlo simulation.

Parameters	Probabilistic Distribution	Reference
Ingestion rate (Ring)	Lognormal	[127]
Inhalation rate (Rinh)	Lognormal	[127]
Exposure frequency (EF)	Triangular	[128]
Exposure duration (ED)	Uniform	[129]
Exposed skin area (SA)	lognormal	[129]
Skin adherence factor (AF)	Beta	[125]
Dermal absorption factor (ABS)	Point	[125]
Particle emission factor (PEF)	Point	[129]
Average exposure time (AT)	Point	[127]
Average body weight (BW)	Normal	[127]

, realizing a comprehensive quantification of uncertainty and variability in the risk assessments.

3. Results

3.1. Contamination Characteristics of Potentially Toxic Elements

The descriptive statistics for PTEs are summarized in Table 2, including minimum (Min), maximum (Max), mean, coefficients of variation (CVs), risk screening values (RSVs), and threshold values (TVs).

The pH values range from 4.26 to 8.29, with a mean of 6.31, indicating predominantly acidic soil conditions. The mean concentrations of PTEs in the soils can be ordered as Zn > Cr > Ni > Pb > Cu > As > Cd > Hg. The high CVs observed for certain PTEs suggest significant variability in their distribution, likely influenced by anthropogenic activities [61,72]. Notably, the element Hg exhibits the highest CV (59.41%), followed by Cu (35.23%) and Cd (33.08%). The concentrations of all PTEs fall within the acceptable ranges of the RSVs, indicating negligible risk to crops. However, the mean concentrations of Cd and Hg elements significantly exceed their TVs, highlighting potential environmental concerns.

Table 2. The descriptive statistics results for PTEs in soils.

Index	pH	As	Cd	Cr	Cu	Hg	Ni	Pb	Zn
Min	4.26	4.43	0.10	35.16	11.19	0.02	13.02	14.00	45.67
Max	8.29	13.95	0.59	81.28	50.82	0.40	46.25	38.48	99.90
Mean	6.31	8.36	0.26	59.16	23.80	0.07	28.47	25.24	70.46
CVs	19.92%	21.46%	33.08%	17.01%	35.23%	59.41%	24.17%	18.44%	16.69%
RSVs	---	20.00	0.30	150.00	50.00	0.50	60.00	70.00	200.00
TVs	---	10.4	0.079	79.00	31.10	0.06	32.60	30.90	86.50

Notes: The pH and CVs are dimensionless. The unit for Min, Max, Mean, RSVs, and TVs is mg/kg. The TVs are from the reference [73].

Table 2. The descriptive statistics results for PTEs in soils.

Index	pH	As	Cd	Cr	Cu	Hg	Ni	Pb	Zn
Min	4.26	4.43	0.10	35.16	11.19	0.02	13.02	14.00	45.67
Max	8.29	13.95	0.59	81.28	50.82	0.40	46.25	38.48	99.90
Mean	6.31	8.36	0.26	59.16	23.80	0.07	28.47	25.24	70.46
CVs	19.92%	21.46%	33.08%	17.01%	35.23%	59.41%	24.17%	18.44%	16.69%
RSVs	---	20.00	0.30	150.00	50.00	0.50	60.00	70.00	200.00
TVs	---	10.4	0.079	79.00	31.10	0.06	32.60	30.90	86.50

Notes: The pH and CVs are dimensionless. The unit for Min, Max, Mean, RSVs, and TVs is mg/kg. The TVs are from the reference [73].

Further analysis using the Igeo reveals that only Cd and Hg exhibit Igeo values above 1, confirming anthropogenic influence on their accumulation (Figure 2a). In contrast, the remaining PTEs show Igeo values below 1, suggesting minimal contamination, and 35.03% of the areas show moderate pollution levels according to the INI (Figure 2b), while geospatial analysis identifies that moderate pollution sites focus on the GFGP regions (Figure 2c). The combined evaluation of CVs, TVs, the Igeo, and the INI underscores the need for prioritized attention toward Cd and Hg in terms of their sources and associated risks.

3.2. Geospatial Distributions of Potentially Toxic Elements

To investigate the spatial patterns of PTEs, the inverse distance weighted (IDW) method is carried out in this study. The GFGP and urbanized regions show alkaline soil (Figure 3a). Previous studies have demonstrated that soil acidification (i.e., decreased pH) enhances the bioavailability of PTEs [74]. Consequently, elevated PTE concentrations in the southern and central regions need particular attention. The element Cd exhibits elevated concentrations predominantly in the GFGP regions (Figure 3c). Cr, Cu, Ni, and Zn demonstrate similar spatial distributions, with higher concentrations clustered in the northernmost and southern zones (Figure 3d,e,g,i). These regions are characterized by natural forests and pollution-free agricultural lands. As and Pb concentrations are primarily enriched in the southwestern areas, particularly near expressway toll gates (Figure 3b,h). The concentration of Hg shows significant accumulation in the southern and eastern urbanized zones (Figure 3f).

4. Discussion

4.1. Soil Acidity and Its Implications for Agriculture

The soil properties in the study area are influenced by multiple factors. For instance, high rainfall accelerates the dissolution of relevant ions in the strata, while long-term agricultural practices create anaerobic conditions that promote sulfide oxidation. These factors may collectively contribute to soil acidification. Strongly acidic soils (pH < 5.5) pose risks to crop production, leading to nutrient deficiencies and increased bioavailability of heavy metals [7], as evidenced by elevated Cd concentrations in the soils of some GFGP regions. To mitigate these issues, agricultural measures such as lime application and organic fertilization are recommended to raise the pH to 6.0–7.0, thereby reducing metal toxicity and enhancing soil fertility [75]. Additionally, regular monitoring of soil pH is essential to assess ecosystem health under acidic conditions.

4.2. Source Apportionment of Potentially Toxic Elements

4.2.1. Compositional Associations of the PTEs

To elucidate the relationships among PTEs in soils, this study employs correlation analysis (p < 0.05) and PCA within the CoDa framework. After accounting for closure effects inherent in soil compositional data, significant positive correlations persist between specific PTEs, including Ni-Zn, Ni-Cu, Cr-Ni, and Cd-Pb (Figure 4). These correlations suggest potential co-occurrence patterns [76], particularly implicating Ni as a key element in shared contamination sources. The PCA results further clarify some associations (Figure 4), with Bartlett’s test of sphericity (p < 0.001) and the Kaiser–Meyer–Olkin measure (KMO = 0.707) confirming the validity and appropriateness of the factor analysis [44]. Three PCs are retained in the PCA based on the eigenvalues greater than 1. PC1, accounting for 32.1% of the total variance, exhibits strong loadings for Ni, Cu, Zn, and Cr. Given that Cr and Ni are predominantly associated with soil parent materials in numerous studies [77,78,79], PC1 likely represents natural sources. PC2 explains 21.5% of the variance and is primarily influenced by Pb and Cd. These elements are well documented to originate from anthropogenic activities such as fossil fuel combustion, agricultural activities, traffic emissions, and industrial processes [80,81,82,83]. PC3 (17.44%) is mainly impacted by the element As, highlighting a specific source. This element is generally associated with non-ferrous metal mining, fossil fuel combustion, and industrial smelting processes [84]. Thus, PC2 and PC3 are possibly interpreted as reflecting anthropogenic contributions.

4.2.2. Source Identification Based on the PMF Model

In this study, a PMF model is used to quantitatively assess the contributions of potential sources to soil PTEs (Figure 5). The PMF model only involves special parameters expressed as concentration. Thus, the source apportionment considers these parameters, including As, Cd, Cr, Cu, Hg, Ni, Pb, and Zn. Before model execution, the uncertainty of each element was evaluated by applying an error fraction of 0.3, and all elements were classified as having a ‘strong’ signal-to-noise ratio (S/N). The model was iterated 20 times with four predetermined factors to ensure stability. The robustness of the model was confirmed through several validation metrics. Specifically, the residuals for all PTEs fell within the acceptable range of −3 to 3. In addition, except Cr (0.46), the remaining PTEs had R² values spanning from 0.80 to 0.99. The ratio of Q_true to Q_expected (Q_true/Q_exp = 0.16) was below the threshold of 2, further supporting the reliability of the source apportionment results.

Factor 1 is predominantly associated with Cu (63.29%), Ni (48.94%), Cr (36.74%), and Zn (37.78%), with a high contribution of 32.48%. The concentrations of these PTEs in nearly all samples are below both TVs and RSVs. Additionally, samples with elevated concentrations are primarily located in natural forests and pollution-free agricultural areas. The observed enrichment pattern in natural forest soils indicates that natural sources predominantly contribute to these PTEs. The northeast/southwest structural strike of the strata further aligns with the spatial distribution of associated PTE hotspots, supporting a lithogenic origin. The existing literature generally attributes Ni and Cr to geogenic origins, such as soil parent materials [85]. Therefore, factor 1 can be interpreted as soil parent materials.

Factor 2, accounting for 28.31% of the total variance in PTEs, is dominated by the element As (51.95%), followed by Pb (33.42%), Cr (38.43%), Ni (38.67%), and Zn (31.46%). In general, As and Pb are derived from combustion processes, such as coal burning and fossil fuel combustion [86]. The concentrations of Cr, Ni, and Zn in soils originate from multiple sources, including natural weathering, vehicular component wear, and lubricant additives [87,88]. In this study, these PTEs exhibited higher concentrations near the southern toll gates of expressways, likely due to intensive traffic activity and prolonged exposure to vehicle emissions. Thus, factor 2 can be attributed to traffic emissions.

Factor 3 contributes the most to Hg (73.42%), with a percentage of 21.48%. The strong spatial variability and enrichment of Hg suggest significant anthropogenic influence. Previous studies have proved that Hg content in soils is controlled by industrial and mining activities [89,90]. Notably, the concentrations of Hg are particularly high in southern and eastern urbanized regions, where there are some thermal power stations. Empirical field investigations demonstrate that soil Hg concentrations display a significant negative relationship with increasing distance from thermal power station emission sources. Given that coal combustion serves as a primary energy source for residents, long-term emissions such as fly ash deposition have likely contributed to Hg accumulation in surface soils. Therefore, factor 3 can be recognized as atmospheric deposition.

The contribution of factor 4 (17.86%) largely focuses on the element Cd (57.36%) in soils, followed by the element Pb (26.63%). Notably, Cd exhibits stronger spatial variability and higher exceedance of threshold values compared to other PTEs. Additionally, the geospatial distribution of Cd hardly exhibits similarity with the structural strike of the strata, potentially meaning the impact of human activities. Field investigations reveal excessive application of phosphate fertilizers in former croplands converted to orchards through afforestation programs. Historical application of fertilizers in these farmlands contributed to persistent PTE accumulation in soils, particularly the element Cd. Currently, thousands of fruit trees are planted to improve the soil environment of the northern part. However, this special way is without the need for frequent tilling and irrigation and induces soil compaction problems due to fruit picking in orchards [91], potentially slowing Cd degradation rates. Therefore, combined with the mentioned findings, factor 4 represents legacy agricultural effects.

Integrated with correlation analysis and PCA, the PMF model reveals distinct source apportionment patterns for PTEs. Zn, Cr, and Ni exhibit mixed sources, combining contributions from soil parent materials and traffic emissions. Cd predominantly originates from legacy agricultural effects, while Cu derives mainly from soil parent materials, and Hg shows a strong association with atmospheric deposition. As and Pb are primarily attributed to traffic emissions, with Pb additionally influenced by secondary contributions from soil parent materials and legacy agricultural effects.

4.3. Potential Ecological Risk Assessment

Source apportionment analysis of PTEs reveals persistent legacy effects of agricultural activities in the GFGP regions. To evaluate the associated environmental impacts, a potential ecological risk assessment was conducted. According to the mean E_i values, PTEs are ranked by ecological risk contribution as follows: Cd > Hg > As > Ni > Pb > Cu > Cr > Zn (Figure 6a). Notably, Cd and Hg from legacy agricultural effects and atmospheric deposition exhibit moderate to high ecological risks (E_i ≥ 40), while other PTEs remain at low-risk levels (E_i < 40). This phenomenon can be attributed to the fact that Cd and Hg possess significantly higher toxicity coefficients and more stringent environmental quality standards compared to other PTEs. The comprehensive assessment using the PERI demonstrates that 60.51% of PERI values (mean > 150) exceed the threshold for low ecological risk (Figure 6b).

Geospatial distribution patterns further identify GFGP implementation areas as the moderate zones, while southern urbanized regions belong to the moderate- to high-risk area (Figure 6c). This spatial correlation highlights historical agricultural practices continue to pose significant ecological risks despite the reduction in PTE concentration through land-use conversion. The findings underscore the long-term environmental impacts of legacy agricultural contamination in ecologically vulnerable regions. Notably, the ecological risk degree in urbanized areas should be paid more attention than in the GFGP regions.

4.4. Human Health Risk Assessment

Soil PTEs present significant human health hazards, necessitating a comprehensive evaluation of both non-carcinogenic and carcinogenic risks. Through the health risk assessment methodology provided by USEPA [67], all PTEs in this study are used to conduct non-carcinogenic risk assessment. Given the toxicity, source, and chemical forms of these PTEs, some PTEs involved in carcinogenic risk assessment include As, Cd, and Pb elements.

4.4.1. Deterministic Human Health Risk Assessment

The non-carcinogenic risk assessment reveals distinct contributions from PTEs, with mean HI values ranking as As > Cr > Pb > Ni > Cu > Cd > Zn > Hg (

Table 3. The statistical results of non-carcinogenic and carcinogenic risks.

PTEs	Population	Non-Carcinogenic Risk			Carcinogenic Risk
		Min	Max	Mean	Min	Max	Mean
As	Children	0.18	0.56	0.34	6.92 × 10−6	2.18 × 10−5	1.3 × 10−5
	Adults	2.52 × 10−2	7.92 × 10−2	4.75 × 10−2	3.89 × 10−6	1.22 × 10−5	7.33 × 10−6
Cd	Children	1.31 × 10−3	7.84 × 10−3	3.52 × 10−3	6.23 × 10−7	3.73 × 10−6	1.67 × 10−6
	Adults	1.95 × 10−4	1.17 × 10−3	5.24 × 10−4	3.5 × 10−7	2.1 × 10−6	9.41 × 10−7
Cr	Children	0.16	0.37	0.27	---	---	---
	Adults	2.35 × 10−2	5.44 × 10−2	3.96 × 10−2	---	---	---
Cu	Children	3.40 × 10−3	1.54 × 10−2	7.24 × 10−3	---	---	---
	Adults	4.78 × 10−4	2.17 × 10−3	1.02 × 10−3	---	---	---
Hg	Children	8.74 × 10−4	1.65 × 10−2	2.73 × 10−3	---	---	---
	Adults	1.24 × 10−4	2.34 × 10−3	3.87 × 10−4	---	---	---
Ni	Children	7.97 × 10−3	2.65 × 10−2	1.74 × 10−2	---	---	---
	Adults	1.14 × 10−3	3.79 × 10−3	2.49 × 10−3	---	---	---
Pb	Children	5.71 × 10−2	0.16	0.10	1.38 × 10−7	3.8 × 10−7	2.49 × 10−7
	Adults	8.04 × 10−3	2.21 × 10−2	1.45 × 10−2	8.11 × 10−8	2.23 × 10−7	1.46 × 10−7
Zn	Children	1.86 × 10−3	4.07 × 10−4	2.87 × 10−3	---	---	---
	Adults	2.62 × 10−4	5.72 × 10−4	4.04 × 10−4	---	---	---
Total	Children	0.53	1.09	0.74	8.48 × 10−6	2.35 × 10−5	1.5 × 10−5
	Adults	0.07	0.16	0.11	4.77 × 10−6	1.32 × 10−5	8.42 × 10−6

Note: “---” refers to not being included in the associated calculation.

). The THI for children ranges from 0.53 to 1.09, with an average of 0.74, indicating potential non-carcinogenic risks from soil PTEs. In contrast, the THI values for adults remain below the safety threshold (1.00). Carcinogenic risk assessment identified that As exhibits risk values exceeding those of Cd and Pb by approximately 7- to 52-fold despite all elements remaining below the 1.00 × 10⁻⁴ safety threshold. The TCR values for both populations fall within acceptable ranges, suggesting generally manageable risk levels under current exposure scenarios.

In summary, the integrated risk assessment demonstrates that the element As from traffic emissions represents the predominant risk factor, serving as the primary contributor to both non-carcinogenic and carcinogenic health risks. This is mainly attributed to the fact that As has a high geological background and SF, as well as a low RfD. Cr and Pb emerge as secondary contaminants of concern, exhibiting significantly higher risk potentials than other PTEs. Notably, children demonstrate greater susceptibility to PTE exposure than adults, as evidenced by both non-carcinogenic risk (HI_child/HI_adult = 6.73) and carcinogenic risk ratios (CR_child/CR_adult = 1.78). These findings are consistent with previous studies [92,93,94,95].

The geospatial distributions of THI and TCR can effectively locate high-risk areas, providing critical insights for targeted environmental management. As illustrated in Figure 7, elevated non-carcinogenic and carcinogenic risks predominantly cluster in northern regions, where legacy agricultural practices represent the primary contamination source. These spatial patterns demonstrate that health problems caused by legacy agricultural effects need to be given importance in the study area despite decreased exceeding rates for PTEs with the help of ecological restoration initiatives.

A comparative analysis of land-use practices reveals striking differences between the northern (high-risk) and southern (low-risk) regions. The southern area’s success in maintaining lower risk levels can be attributed to the implementation of pollution-free stereoscopic agriculture, which integrates vertical farming techniques with strict contamination controls. This contrast underscores the effectiveness of sustainable agricultural practices in mitigating soil pollution and associated health risks.

4.4.2. Probabilistic Human Health Risk Assessment

Given the variations such as body weight (BW) and chemical intake from diverse populations, probabilistic human health risk assessment is carried out to accurately qualify health risk degrees to the human body. In the results of non-carcinogenic risk assessment, the HI caused by each PTE is less than 1.00, showing that single PTEs in soils cannot pose a potential health risk (Figure 8). Consistent with deterministic results, As, Cr, and Pb contribute the most to the non-carcinogenic risk levels (Figure 8a,c,g), in which the maximum risk value of the element As is close to 1.00 and should be of concern (Figure 8a). The THI values for children have a maximum value exceeding the limit of 1.00, for which the percentage reaches 12.82% (Figure 8i). Compared to children, the total non-carcinogenic risks of these PTEs to adults are below the limit due to their strong immune systems.

The carcinogenic health risk assessment model reveals that the percentages of CR < 1 for As, Cd, and Pb are 100%, indicating low carcinogenic risks to both populations (Figure 9). Likewise, As is identified as the major contributor of carcinogenic risk levels (Figure 9a). The maximum values of TCR for children and adults are 4.00 × 10⁻⁵ and 2.30 × 10⁻⁵, respectively, demonstrating that the total carcinogenic risks from PTEs are negligible for various populations (Figure 9d).

4.4.3. Sensitivity Analysis

In this study, a sensitivity analysis was used to evaluate influential levels of various factors on health risk assessments. The results revealed that the Ring emerges as the most sensitive parameter for both non-carcinogenic and carcinogenic risks (Figure 10), suggesting that oral ingestion serves as the predominant exposure pathway for PTEs in soils. This finding is particularly significant for pediatric populations, as children exhibit heightened vulnerability due to frequent hand-to-mouth behaviors and contact with contaminated surfaces in hygienically compromised environments [96]. The inverse correlation between body weight (BW) sensitivity and health risk levels further explains the increased susceptibility of children compared to adults. Zn and Cr have much greater sensitivities than other PTEs, suggesting the risk accumulation in the human body (Figure 10a). This pattern suggests that long-term exposure to these elements may pose substantial health concerns despite their concentrations below the safety threshold values.

4.5. Comparison with Some Agricultural Regions

Incorporating regional comparisons into this study can enhance the objectivity of this research result and potentially reveal potential patterns and drivers of soil contamination, such as the types of land use and anthropogenic activities. This can provide a more comprehensive understanding of pollution dynamics.

Numerous studies have established that Cd represents the predominant PTE in agricultural systems, with its accumulation primarily driven by prolonged application of phosphate fertilizers [97,98,99,100,101]. Meanwhile, Cd is considered a priority control pollutant due to its high toxicity and bioavailability [102,103,104]. Additionally, according to various findings of human health risk assessments, children are exposed to a greater risk compared to adults, while the element As is listed as the predominant contributor to potential health risks [105,106]. These conclusions are consistent with these results. Inversely, this research exhibits lower PTE concentrations (especially Cd, Hg, and As) compared to most other agricultural regions, demonstrating minimal ecological-health risk levels (Figure 11). The relatively lower risk levels observed in this study area can be linked to the successful implementation of SAD.

4.6. Limitations and Future Scope

The findings of this study reveal the characteristics of PTE concentrations, the main sources of pollutants, and the ecological-health risk levels. These results provide scientific evidence for the favorable soil environment in the study area. However, this research has certain limitations. Specifically, it does not account for long-term risk assessment based on PTE transport dynamics or detailed changes in water quality. Therefore, future studies should incorporate continuous soil and water monitoring, along with machine learning techniques and numerical simulations, to better characterize PTE mobility and associated environmental risks [119]. On the other hand, while this study focused on total PTE concentrations for source/risk nexus analysis, future investigations could incorporate soil texture, organic matter content, and CEC measurements to assess bioavailable fractions and refine risk predictions [74].

To maximize the broader environmental and socio-economic benefits, subsequent studies should explore synergies between PTE remediation and global sustainability agendas, such as carbon neutrality, peak carbon emissions, and new quality productive forces (NQPF). For example, the carbon sequestration potential of the GFGP regions could be quantified to support climate goals, while sustainable agricultural activities could be optimized to align with low-carbon development strategies [120,121]. The energy sector could develop NQPF and promote carbon neutrality to reduce pollution sources [122]. By aligning soil contamination remediation with these forward-looking policies, future studies can contribute to a holistic approach that simultaneously addresses environmental degradation, climate change, and sustainable development.

5. Conclusions

This research conducted quantitative source apportionment and comprehensive risk assessments in a typical SAD region of SW China. An integrated analytical approach combining multivariate statistical analysis with CoDa techniques and the PMF model was employed to accurately identify the sources of PTEs and quantify source contributions. Health risk assessment models incorporating Monte Carlo simulation were used to evaluate both non-carcinogenic and carcinogenic risk levels across different populations. The key findings of this study are summarized as follows:

• Elevated elemental concentrations of Cd (mean: 0.26 mg/kg; CV: 33.08%) and Hg (mean: 0.07 mg/kg; CV: 59.41%) were observed, exceeding their respective threshold values (Cd: 0.079 mg/kg; Hg: 0.06 mg/kg). Spatial analysis revealed distinct distribution patterns, with Cd enrichment predominantly occurring in GFGP implementation areas, while Hg accumulation was concentrated in urbanized areas.
• Source apportionment identified four primary contributors to PTE concentrations: soil parent materials (32.48%), traffic emissions (28.31%), atmospheric deposition (21.48%), and agricultural activities (17.86%). Cd, Cu, and Hg are mainly sourced from legacy agricultural effects, soil parent materials, and atmospheric deposition, respectively. Zn, Cr, and Ni exhibited mixed sources (soil parent materials and traffic emissions). The elements As and Pb were primarily associated with traffic emissions, with Pb showing secondary influences from soil parent materials and legacy agricultural effects.
• In total, 60.51% of the sample sites exhibited PERI values (mean > 150) exceeding the threshold for low ecological risk, falling in legacy agricultural regions.
• The PTEs in soils were found to pose a potential non-carcinogenic health risk exclusively to children, while carcinogenic risk assessment results indicated negligible risks for both children and adults. Notably, the risk levels for children consistently exceeded those for adults, highlighting their greater vulnerability to PTE exposure. Among the elements analyzed, As was identified as the most hazardous PTE. Geospatial analysis revealed that the relatively high-risk areas were predominantly concentrated in the GFGP region.

Although this study cannot present dynamic risk assessment results, the research findings still provide valuable insights into an agricultural region conducting ecological restoration policies.