This study selected the Xin’anjiang River Basin section in Huangshan City, Anhui Province, China (118°0′0″E–118°15′0″E, 29°45′0″N–29°52′30″N) as the study area. The region has a subtropical monsoon climate with abundant rainfall and concurrent heat and rain. The annual average rainfall is 1,395–1,702 mm, and the annual average temperature is 15.50 °C–16.40 °C. The terrain is predominantly hilly with an elevation of 105–867 m. The main soil types are yellow soil, mountain yellow-brown soil, and red soil.

The Xin’anjiang River originates in Huangshan City, Anhui Province, with a main stream length of 242 km and a watershed area of 5,856 km² within Huangshan City. The average outflow water volume accounts for more than 68% of the inflow water volume of Qiandao Lake in Zhejiang Province. The Xin’anjiang River is an important water source for the production and living needs of Huangshan City, Anhui Province, and Hangzhou City, Zhejiang Province, and is also a critical ecological barrier and strategic water source for the downstream regions of the Yangtze River Delta. Therefore, this study selected the Huangshan City section of the Xin’anjiang River Basin as the research area.

The soil pH values in the Xin’anjiang River Basin ranged from 4.04 to 8.16, with an average value of 5.87, indicating overall acidic soil. Overall, the average content of Hg, Cr, and Ni in the soil of the Xin’anjiang River Basin was lower than the background values, while the average content of As, Cu, Zn, Cd, and Pb was 1.16, 1.30, 1.43, 2.66, and 1.38 times the background values, respectively. Comparing the variation extent of soil heavy metal content, Cr and Zn showed slight variation, As, Ni, Cu, and Pb showed moderate variation, and Hg and Cd showed high variation. Different farming environments can lead to significant differences in soil environments. Therefore, farmland was divided into paddy and dryland types for comparison (Table 4). The soils in both paddy and dryland areas of the Xin’anjiang River Basin were acidic, but the pH value of paddy soils was higher than that of dryland soils, overall weakly acidic. The average content of Hg, Cd, and Pb in paddy soils was 1.24, 1.24, and 1.16 times that of dryland soils, respectively. As and Zn in paddy soils reached 88.30% and 91.98% of those in dryland soils, respectively. The content of Cr and Ni in the two types of farmland areas showed little difference. By comparing the variation extent of soil heavy metal content in different farmland types, it was found that, except for Zn in paddy fields reaching moderate variation and Pb reaching high variation, the variation extent of soil heavy metal content in the two types of farmland was consistent with the characteristics of the overall situation. The C:N:P ratio of farmland soil in the study area was 14.3:1.4:1, and it was subject to a certain degree of human interference. The coefficient of variation for organic carbon, total nitrogen, and total phosphorus content in the soil reached moderate levels in both paddy fields and dry fields. There were no significant differences in the content of organic carbon, total nitrogen, and total phosphorus between the two types of cropland soils. However, on average, the TP content in upland fields was 1.25 times that of paddy fields, while SOC and TN were 88.42% and 84.80% of those in paddy fields, respectively.

The single factor pollution index can quickly identify the pollution level of various heavy metals in the soil and warn of potential heavy metal pollution risks (Figure 2). Overall, except for the difference in Cd pollution levels between paddy (heavy) and dryland (moderate), the pollution levels of other heavy metals were the same in both types of farmland: As, Cu, Zn, and Pb were mildly polluted, while Hg, Cr, and Ni did not reach pollution levels. Overall, the heavy metal pollution levels in paddy soils were Cd > Cu > Zn > Pb > As > Ni > Cr > Hg, and in dryland soils, they were Cd > Zn > Cu > Pb > As > Ni > Cr > Hg.

The Nemero composite pollution index can comprehensively evaluate the pollution status of various heavy metals, providing a more overall judgment of soil heavy metal pollution based on the weights of different heavy metals (Figure 2). The heavy metal pollution levels in paddy and dryland soils in the Xin’anjiang River Basin were mostly at the warning and mild pollution levels. The Nemero composite pollution index of heavy metals in paddy soils ranged from 0.57 to 6.65, with a median and mean of 1.34 and 1.78, respectively. The Nemero composite pollution index of heavy metals in dryland soils ranged from 0.43 to 10.57, with a median and mean of 1.17 and 1.54, respectively.

Based on Kriging interpolation, the spatial distribution of heavy metals and nutrient factors in the study area was evaluated (Figure 3). Hg is primarily concentrated in the foothills of the western part of Lantian Town, as well as in the basin areas of the southern parts of Haiyang Town and Wanan Town. As is distributed relatively uniformly in the eastern part of Xidi Town, as well as in Lantian Town, Haiyang Town, and Wanan Town. Cr is primarily concentrated in the eastern part of Xidi Town and the western part of Lantian Town, with limited distribution in other areas. Ni is distributed relatively uniformly throughout the study area. Cu is primarily concentrated in the mountainous valleys at the junction of the eastern part of Xidi Town and the western part of Lantian Town, similar to Zn and Cd, which are also concentrated in these areas. Pb is primarily distributed in the western part of Lantian Town and the southern basin areas of Haiyang Town and Wanan Town. Soil organic carbon in the study area is primarily distributed in the mountainous valleys at the western boundary between Lantian Town and Xidi Town, as well as in the northern foothill regions of Wanan Town. Soil total nitrogen concentrations are higher in the mountainous valleys at the western boundary between Lantian Town and Xidi Town. Regions with higher soil total phosphorus concentrations are primarily the foothill areas west of Lantian Town and the basin regions east of Wanan Town. The spatial distribution of the Nemero index showed that heavy metal composite pollution was more severe in the eastern part of Xidi Town and the western part of Lantian Town, while pollution levels were lower in other areas.

This study performed PMF source apportionment of the eight heavy metals in the soil. The model parameters showed that when the number of sources was set to 6, the correlation coefficient r² of the heavy metals, except for Cu (0.822), was above 0.9, indicating good model performance (Table 5). The results of the PMF source apportionment (Figure 4) showed that soil heavy metal pollution was mainly influenced by traffic sources, natural sources, atmospheric sources, industrial sources, agricultural sources, and solid waste sources. Atmospheric sources contributed the most to Hg, accounting for 72.8% of the total contribution. As was mainly influenced by industrial sources, with a contribution rate of 73.9%. Natural sources were the main contributors to Cr, Ni, and Cu, accounting for 52.5%, 51.7%, and 41.8%, respectively. Zn was mainly derived from natural and traffic sources, accounting for 31.0% and 42.8%, respectively. Cd was primarily affected by agricultural activities, with agricultural sources contributing 83.9%. Pb was mainly influenced by natural and solid waste sources, accounting for 40.9% and 43.1%, respectively.

Soil C, N, and P content and their stoichiometric ratios are core indicators for assessing soil function, health, and nutrient cycling efficiency. In this study, the soil C, N, and P characteristics (Figures 5, 6) showed no significant differences in nutrient profiles between dryland and paddy soils. Overall, soil N and P levels showed little variation, with their primary ranges (25th–75th percentiles) being 0.85–1.41 g/kg and 0.58–1.06 g/kg, respectively. In contrast, soil C content displayed a clear gradient, ranging mainly from 8.80 to 15.20 g/kg. The soil nutrient stoichiometric ratios were primarily characterized by a C:P gradient (10.66–19.64), while C:N and N:P ratios were more narrowly distributed.

A linear regression model (Figure 6) linking the heavy metal composite pollution index to soil factors helps clarify the relationship between soil nutrients and heavy metal pollution. In this study, the P_N index showed significant positive correlations with soil pH, SOC, TN, and TP, and a significant negative correlation with C:P (p < 0.001).

The correlation between heavy metal content and important soil factors in the farmland soils of the Xin’anjiang River Basin is shown in the Figure 7. As, Cr, Ni, Cu, Zn and Cd showed significant correlations with each other (p < 0.05). In the correlation analysis of heavy metals in soil, Pb showed no significant correlation with Cr but exhibited significant positive correlations with other heavy metals. Hg only had significant negative correlations with Cr and Ni in soil and significant positive correlations with Cu and Pb (p < 0.05). Regarding the correlation between soil nutrient factors and heavy metals, soil SOC, TN, and TP showed significant positive correlations with heavy metals except for Hg and As (p < 0.05). From a stoichiometric perspective, only C:N showed significant negative correlations with Cr and Ni, while C:P and N:P showed significant negative correlations with As (p < 0.05).

As shown in Figure 8 redundancy analysis can clarify the overall explanatory ability of soil factors on heavy metals. In this study, TN, N:P, TK, C:N and pH contributed 61.4%, 15.7%, 8.3%, 6.2%, and 4.0% of the variation in soil heavy metal content, respectively (p < 0.05). After dimensionality reduction, the comprehensive explanatory degree of soil factors reached 21.54%. The heavy metals in the soil are mainly distributed across two quadrants. Based on their projection distance and direction relative to the interpretive axes and environmental factors, they can be preliminarily divided into five groups:Hg lies close to the second interpretive axis; in the first quadrant, Pb has a positive correlation with TP and SOC, Zn, Cu, and Cd are close to the first interpretive axis; Cr and Ni are in the fourth quadrant, influenced by the first axis of interpretation and pH; As has a small interpretive degree.

By constructing a structural equation model between soil factors and different heavy metals (Figure 9), the potential relationships between soil factors and heavy metals can be identified, and the influence coefficients of each factor on soil heavy metals can be clarified. The soil Cu content was significantly directly linked to pH, TN, and C:P, with path coefficients of 0.10, 0.64, and −0.28, respectively. C:P was significantly influenced by pH, SOC, and TP, with path coefficients of −0.10, 0.90, and −0.84, respectively (p < 0.05). Cd had a similar model to Cu, with influence coefficients of 0.23, 0.56, and −0.28 for pH, TN, and C:P, respectively. Pb was directly affected by SOC and TK (path coefficients of 0.54 and 0.13, respectively, p < 0.05) and driven by C:P, C:N, and N:P (path coefficients of −0.74, 0.27, and 0.62, respectively, p < 0.05). TN directly affected Ni content (path coefficient of 0.66) and indirectly suppressed it via N:P (path coefficient of −0.22). Zn content was directly influenced by TN, TP, and TK (path coefficients of 0.20, 0.45, and 0.22, respectively) and had no significant direct path relationship with C:N:P. The structural equation model of soil nutrient factors with the P_N index, similar to Cu and Cd, was directly path - affected by pH, TN, and C:P (0.20, −0.41, and 0.52, respectively).

The results of the single-factor pollution index and Nemero pollution index in this study (Figures 2, 3) indicate that Cd is the primary heavy metal contaminant in the farmland soils of the study area, with generally low levels of heavy metal composite pollution. Only a small portion (<25%) of the region experiences moderate to severe contamination. The spatial distribution patterns (Figure 3) show that different heavy metals have distinct spatial distribution characteristics, yet they also exhibit similar distribution patterns. This is because soil heavy metal pollution levels are influenced by multiple sources, leading to regional variations (Yuan et al., 2021). Positive matrix factorization (PMF) analysis for the eight heavy metals in this study (Figure 4) shows that Hg is predominantly sourced from atmospheric origins, aligning with previous studies (Hu et al., 2018). As is often present as a by-product or raw material in chemical industry processes, such as mining and manufacturing of glass and paints (Ma et al., 2019), matching the industrial structure of the study area. Cr and Ni, with similar configurations, are widely present in soil formation processes and parent materials due to natural sources (Yi et al., 2024). This is consistent with the low coefficient of variation and uniform distribution observed in this study, and the significant positive correlation between Cr and Ni (Table 4; Figures 3-7). Cu and Zn also exhibit significant positive correlations and similar source contributions (Figures 3-7), likely due to inputs from vehicle tire wear, galvanized component corrosion, and automobile exhaust emissions, making traffic sources a significant contributor to Cu and Zn contamination (Fan et al., 2022). Previous studies have shown that the application of chemical fertilizers and pesticides leads to Cd accumulation in agricultural soils (Du et al., 2025), and the addition of straw and organic fertilizers, while enhancing soil nutrients, also increases the risk of heavy metal pollution (Li J. Q. et al., 2024). This is consistent with the significant positive correlations observed in this study between Cd and TN, TP, TK, and SOC (p < 0.05) (Figure 7). Regarding Pb source identification, leaching from solid waste is likely the main source. Previous studies have indicated that urban construction materials, municipal solid waste, and waste electrical and electronic products are important sources of soil Pb (Miao et al., 2022; Yeganeh et al., 2025).

Compared to the mean C:N:P ratio of 38:3.2:1 in subtropical farmland soils (Song et al., 2020), the study area exhibits higher P content and lower SOC content (C:N:P = 14.3:1.4:1). This may be related to the study area’s location in an intermontane basin (Figure 1). Phosphorus, cycling predominantly as a sedimentary cycle, is primarily present in rocks and minerals as phosphate. The relatively enclosed topography of the intermontane basin favors soil accumulation, and phosphorus in the soil undergoes redistribution through processes such as leaching and precipitation. Some areas may experience phosphorus enrichment due to precipitation effects (Li J. S. et al., 2023).

The coordination and driving relationship between soil nutrient factors and heavy metal content: On the one hand, heavy metal emission sources can contribute compounds containing C, N, and P to the soil. On the other hand, heavy metals in the soil are influenced by nutrient factors, altering their speciation and mobility (Tian et al., 2025). In this study, soil SOC, TN, and TP showed significant positive correlations with Ni, Cu, Zn, Cd, and Pb (p < 0.05) (Figure 7), and similar results were indicated by the structural equation model. The reasons may be as follows: from the perspective of pollution sources, traffic activities (such as automobile exhaust emissions and tire wear) release nitrogen oxides into the soil, increasing the nitrogen content contributed by traffic sources (Li Y. R. et al., 2025). This is consistent with the significant correlations between Cu, Ni, Zn, and Pb and TN in this study (Figures 7–9). Regarding solid waste sources, studies have shown that urban sewage sludge compost can increase soil organic carbon, total nitrogen, and total phosphorus, but also poses a risk of Pb pollution (Xu et al., 2016), which may explain the positive path relationships between TOC and TK and Pb in this study. In agricultural activities, wastewater irrigation and the application of straw or livestock manure as alternatives to chemical nitrogen fertilizers can introduce Cd into the soil while also adding extra C, N, and P (Drabesch et al., 2024). Notably, straw, as a high C:N compound, can increase the C:N and C:P ratios in the soil upon application (Wang et al., 2023). Additionally, the use of phosphate fertilizers containing Cd impurities is considered an important cause of Cd contamination in farmland (Kubier et al., 2019). These results suggest that there are complex interactions between heavy metals and nutrients in soil.

The complexity of soil environments can lead to contradictory results in linear relationships, making it necessary to clarify the impact paths of different factors on heavy metals. In this study, SOC showed a significant positive path relationship only with Pb (p < 0.05). Previous studies have indicated that increased SOC can significantly reduce the bioavailability of Pb but may also lead to its accumulation in soil (Cui et al., 2024). Phosphorus is not only an essential soil nutrient but also an environmental factor closely related to soil minerals (Luo et al., 2023). In this study, TP was found to directly and positively drive Zn content in the soil, aligning with previous research showing that TP can promote Zn immobilization by increasing the content of amorphous iron oxide-bound Zn in soil (Lu et al., 2015). The soil TK exhibited significant positive path relationships with Ni, Zn, and Pb (Figure 9), consistent with the correlation results (Figure 7). Beyond the direct contributions of pollution sources (Yin et al., 2024), TK, as a fertility factor, can positively drive the content of these heavy metals when its content increases (Gong et al., 2024), which is consistent with the results of this study showing that TK has the same directional path relationships as TN and SOC. For Cd and Cu, high TN content in the soil can reduce their effectiveness and mobility by promoting plant growth and the secretion of organic acids from plant roots that can bind with Cd (Zheng et al., 2020). Additionally, Cu can promote total nitrogen fixation in the soil (Elrys et al., 2024). However, in the structural equation model (Figure 9), C, N, and P did not show fully significant direct relationships with heavy metals. This may be due to the complexity of nutrient factors driving changes in the availability of soil heavy metals. Further exploration of the relationship between nutrient factors and soil heavy metal content from a stoichiometric perspective is needed. From the perspective of stoichiometry, the results indicate that C:P has a negative path relationship with Cd and Cu. On one hand, high carbon content can promote soil organic matter formation. While organic matter can adsorb and immobilize Cd and Cu, dissolved organic matter can also form Cu complexes, increasing Cu outflow in runoff (Neaman et al., 2024; Xie et al., 2024). On the other hand, low phosphorus content can reduce the chances of Cd and Cu forming soluble compounds with phosphorus, thereby lowering their bioavailability and limiting their accumulation in the soil (Elrys et al., 2024; Xie et al., 2024). For Pb, the path mechanism of C:P may be that increased P can effectively transform the exchangeable forms of heavy metals into residual forms, significantly reducing the concentration of exchangeable heavy metals. This effect is more pronounced for Pb-contaminated soils than for Cd-contaminated soils (Qiu et al., 2024), which is consistent with the results of this study showing that C:P has a higher path coefficient for Pb than for Cd. In this study, soil N:P negatively drove the accumulation of Ni, but TN showed a positive direct path relationship. This may be due to high TP content, as increased phosphorus can immobilize heavy metals in the form of P-mineral complexes, reducing their reactivity and mobility (Boostani et al., 2025).

In terms of soil characteristics, soil pH is a crucial factor influencing the speciation of heavy metals in soil. In this study, an increase in pH led to increased contents of Cu, Cd, etc. (Figures 6-9). This is likely because the reduction in hydrogen ion concentration in the soil decreased the competitive adsorption of Cd and Cu on soil colloids. Additionally, Cd more readily forms insoluble hydroxide precipitates with hydroxide ions in the soil. This process lowers the mobility and bioavailability of Cd and Cu, promoting their precipitation (Liu et al., 2025; Sereni et al., 2021). In terms of farming systems, the results of this study indicate that Cd contamination is more severe in paddy fields than in dry fields. These results are consistent with previous studies indicating that the high clay content and anaerobic conditions of paddy soil promote Cd adsorption (Long et al., 2024).

The structural equation model results for P_N and soil nutrient factors (Figure 9) show a model path similar to that of Cd. The main reason is that soil heavy metal composite pollution is primarily influenced by Cd (Figure 2). In this study, attempts were made to construct structural equation models for Hg and As with soil nutrient factors, but the fitting results showed R² < 0.3. This may be because the primary sources of Hg and As are atmospheric deposition and mineral extraction, respectively, and their direct associations with SOC and TN are weak. Additionally, the methylation of Hg and the redox state transformations of As in the soil are dominated by microbial activities, which may mask their direct relationships with SOC and TN (Li B. W. et al., 2025; Qin A. et al., 2023). In redundancy analysis and PMF model results, Cr and Ni showed similar results, but the structural equation model constructed based on Cr had a low explanatory power (R² < 0.3) (Figures 4-8). This may be due to its low content in the soil. Therefore, further research on the coupling relationship between soil heavy metals and nutrient factors can be conducted from perspectives such as the different forms and activities of heavy metal elements, plant roots, and nutrient element cycle-related functional microorganisms.