Inﬂuencing Factors of Elevated Levels of Potentially Toxic Elements in Agricultural Soils from Typical Karst Regions of China

: Agricultural soils originating from carbonate rocks within karst regions exhibit inherent high concentrations of potentially toxic elements (PTEs) due to geogenic processes. However, the inﬂuencing factors of the elevated levels of PTEs in the naturally contaminated karst regions remain inadequately comprehended. This research investigates the chemical compositions of 278 soils derived from carbonate rocks. Descriptive statistics, stepwise multiple regression, and the random forest (RF) method were applied to screen the signiﬁcant factors that affect the distribution, migration, and enrichment of the PTEs in soils. Cadmium (Cd) and arsenic (As) are the most highly contaminated PTEs in the soils of the study area, and the average contents of Cd and As in soils are 11.5 and 2.92 times the national soil background value, respectively. The pollution risk ranking of PTEs is Cd > As > Cr > Zn > Pb > Cu > Ni > Hg, using the proportion exceeding the risk screening thresholds of agricultural land as the standard. Soil Fe 2 O 3 and Mn contents, soil pH and total organic carbon (TOC) values, and weathering intensity (characterized using the chemical index of alternation, CIA) are the most important factors inﬂuencing the PTE levels in agricultural soils.


Introduction
The contamination of potentially toxic elements (PTEs) in agricultural lands has emerged as a global concern due to the growing recognition of the importance of safe food production. The karst areas in southwestern China have been known to have high geochemical background values of various PTEs (Cd, Hg, As, Cr, Pb, Zn, Ni and Cu) in soils [1][2][3][4][5]. The high geochemical background of PTEs in these areas is related to the local carbonate rock parent materials [6]. Guangxi is one of the most widely distributed karst areas in China. According to the results of the first national soil pollution survey, Cd, Hg, As, Cr, Pb, Zn, Ni and Cu in surface soil in typical karst areas of Guangxi are 4.5, 2.6, 2.0, 1.6, 1.4, 1.4, 1.3 and 1.1 times the national average value [7]. The natural content of PTEs in carbonate-rock-derived soils is mainly controlled by the geochemical background of the parent rocks [8,9].
The weathering process of carbonate rocks displays distinct characteristics. On the one hand, the levels of PTEs in these rocks are generally lower compared to silicate parent rocks [10], with only a few exceptions. For instance, in Switzerland, unusually high Cd contents in carbonate rocks have been identified [11]. On the other hand, a significant majority (over 97%) of the original carbonate content is leached during weathering and subsequent pedogenesis, leading to noteworthy accumulations of PTEs in the remaining materials [2,12]. The elevated concentrations of PTEs in soils originating from karst regions have sparked concerns regarding crop safety, as dietary consumption constitutes the primary avenue of human exposure to these elements. Notably, rice, a pivotal regional cereal, contributes to more than 70% of human Cd exposure [13]. The potential for chronic toxicity arises due to the capability of Cd to accumulate within the human body, potentially resulting in conditions such as kidney disease, lung and prostate cancers, and endocrine disruption [14,15].
The Guangxi karst region, encompassing one of the world's largest karst terrains, has documented elevated levels of PTEs, notably Cd, within its surface and subsurface soils [7]. While past research has explored the geochemical intricacies of the distinct local soil formation process and the PTE enrichments in typical Guangxi karst regions [6,16,17], the underlying factors driving the elevated PTE levels within this high background karst region remain inadequately elucidated.
The weathering processes and pedogenesis in these areas eventually lead to significant changes in soil properties in the carbonate-rock-derived soils, and evidently high levels of PTEs. In order to obtain a deep insight into the influencing factors of elevated levels of PTEs in agricultural soils of karst regions, we apply systematic methodologies, including descriptive statistics, stepwise multiple regression, and the random forest (RF) method. The aim of this work was to identify the influence of soil properties (such as soil pH and organic carbon), mineral composition (such as ferro-manganese oxides), and weathering intensity on the contamination of PTEs in the karst regions.

Study Area
The research is conducted within the karst regions of Guangxi, situated in southwestern China, with elevations ranging from 50 to 200 m. The terrain predominantly comprises karst plains and low hills, while the regional climate features high humidity and a tropical monsoon pattern. The study area is characterized by a prevalence of carbonate rock lithology. The primary soil classifications, adhering to the world reference base for soil resources [18], include Acrisols, Anthrosols, Cambisols, Ferralsols, and Luvisols.

Sample Collection and Pretreatment
This study encompassed a collection of 278 surface soil samples, sourced from the top 20 cm layer of agricultural lands. The collection process involved gathering soil from 5 distinct sub-sample points, which were subsequently combined to create a composite sample. The sampling sites were primarily situated in the cities of Nanning, Hechi, Liuzhou, Guigang, Laibin, and Guilin ( Figure 1). Following collection, all samples were securely sealed within airtight plastic bags and transported to the laboratory within a 48 h timeframe. The collected surface soil samples were subjected to air drying. Subsequently, these surface soils were sieved through a 10-mesh sieve to eliminate debris and pebbles. To prepare for subsequent analysis, the soil samples were meticulously crushed into a fine powder (<0.075 mm, >200 mesh) utilizing an agate mortar.

Chemical Analysis
Soil pH measurements were conducted using a pH meter on an aqueous suspension, maintaining a soil-to-water ratio of 1:2.5 w/v [19]. Major elemental concentrations within the powdered samples were determined through powder X-ray fluorescence (XRF) analysis, specifically employing the ZSX Primus II instrument (Rigaku, Japan, Osaka). For the assessment of total soil element concentrations, 1.00 g of dried and powdered samples underwent separate analysis. This involved the digestion of the samples in a mixture of HF, HNO3, and HClO4 [20]. Inductively coupled plasma mass spectrometry (ICP-MS) was employed to measure the elemental concentrations of As, Cd, Cr, Cu, Hg, Ni, Pb, and Zn in the digestion solutions. To account for instrumental drift, an internal standard, In (at a final concentration of 10 μg·L −1 ), was employed. Quality assurance and quality control (QA/QC) measures were implemented to assess accuracy and precision. This involved utilizing blanks, duplicate samples, and standard reference materials. Blank assays using reagents produced undetectable results. Accuracy was validated using certified reference materials (GSS-17, GSS-22, GSS-25, and GSS-26). Replicate analysis of samples and standards (performed three times) demonstrated a precision of less than 5% (relative deviation).

Factor Importance Identification
In this research, the random forest (RF) method was employed to ascertain and assess the relative significance of potential variables influencing PTEs in agricultural soils. The RF model represents an advanced approach encompassing classification and regression, originating from the classification and regression tree procedure. The RF model presents distinct advantages compared to alternative models, particularly in the identification of pivotal factors and its resilience against overfitting or unbiased error measurement [21]. Specifically, in this study, RF was exclusively utilized to gauge the importance of potential variables with a focus on Cd and As concentrations in the soils.

Chemical Analysis
Soil pH measurements were conducted using a pH meter on an aqueous suspension, maintaining a soil-to-water ratio of 1:2.5 w/v [19]. Major elemental concentrations within the powdered samples were determined through powder X-ray fluorescence (XRF) analysis, specifically employing the ZSX Primus II instrument (Rigaku, Japan, Osaka). For the assessment of total soil element concentrations, 1.00 g of dried and powdered samples underwent separate analysis. This involved the digestion of the samples in a mixture of HF, HNO 3 , and HClO 4 [20]. Inductively coupled plasma mass spectrometry (ICP-MS) was employed to measure the elemental concentrations of As, Cd, Cr, Cu, Hg, Ni, Pb, and Zn in the digestion solutions. To account for instrumental drift, an internal standard, In (at a final concentration of 10 µg·L −1 ), was employed. Quality assurance and quality control (QA/QC) measures were implemented to assess accuracy and precision. This involved utilizing blanks, duplicate samples, and standard reference materials. Blank assays using reagents produced undetectable results. Accuracy was validated using certified reference materials (GSS-17, GSS-22, GSS-25, and GSS-26). Replicate analysis of samples and standards (performed three times) demonstrated a precision of less than 5% (relative deviation).

Factor Importance Identification
In this research, the random forest (RF) method was employed to ascertain and assess the relative significance of potential variables influencing PTEs in agricultural soils. The RF model represents an advanced approach encompassing classification and regression, originating from the classification and regression tree procedure. The RF model presents distinct advantages compared to alternative models, particularly in the identification of pivotal factors and its resilience against overfitting or unbiased error measurement [21]. Specifically, in this study, RF was exclusively utilized to gauge the importance of potential variables with a focus on Cd and As concentrations in the soils.

Statistical Analysis
Basic descriptive statistics were derived to summarize the concentrations of major elements and trace elements in the study area. Descriptive statistics and stepwise multiple regression analysis were performed using IBM SPSS Statistics 24 (IBM Corp, Armonk, NY, USA). Pearson correlation analysis and three-dimensional surface fitting were performed using Origin 2021 (Origin Lab Corporation, Northampton, MA, USA). The RF model was implemented in R 4.2.3 (R Development Core Team, Vienna, Austria) using the randomforest package [22].

Concentrations of PTEs in Agricultural Soils
The statistics of major and trace element content of 278 soil samples in the tillage layer (0-20 cm) in the study area are shown in Table 1. Cadmium (Cd) can accumulate within the human body and cause conditions such as kidney disease or lung and prostate cancers. The soil Cd content in karst areas of Guangxi ranged from 0.12 to 7.69 mg·kg −1 , with an average value of 1.12 mg·kg −1 , which was similar to that of Song et al. [23] in the karst areas of Guangxi (average 0.915 mg·kg −1 ). The Cd concentrations in agricultural soils significantly exceeded the national background value of 0.097 mg·kg −1 in soil layer A and the Guangxi background value 0.267 mg·kg −1 in soil layer A [24]. According to the national standard "Soil environmental quality risk control standard for soil contamination of agricultural land" (GB-15618-2018) issued by the Ministry of Ecology and Environment of China [25], the pollution risk of each PTE was evaluated. The proportion of samples exceeding the background value of soil Cd in Guangxi and China reached 84.2% and 100%, respectively, indicating that the soil Cd content in the study area was very high. The percentage of samples surpassing the risk screening threshold and risk management limit for Cd stood at 76.3% and 7.6%, correspondingly, indicating that the risk of agricultural product safety did not meet the food quality (Table 2). However, according to the results of the previous survey, the safety and quality of crop products in Guangxi karst regions are not as bad as the soils; the exceedance rate of food quality standard is about 10%, which is the national average level [26], which may be because the national standard (GB-15618-2018) is not fully applicable in specific karst high geochemical background areas. Arsenic (As) can cause disorders of cell metabolism, resulting in diseases of important organs such as the nervous system and kidneys, so it has attracted much attention from scholars. Wu et al. [27] conducted a survey in Hechi, Guangxi, and found that soil As pollution was serious, with an exceedance rate of 42.9%. In this investigation, it was observed that the mean soil As concentration in the karst region measured 39.91 mg·kg −1 . A substantial 52.5% of samples exceeded the background value of the A-layer soil in Guangxi (20.5 mg·kg −1 ), with an even greater 84.5% surpassing the background value for the A-layer soil across China (11.2 mg·kg −1 ). Furthermore, 43.5% of samples exceeded the established risk screening threshold ( Table 2). The dispersion of As in the tillage layer soil within the study area was pronounced, evident by its high coefficient of variation (CV) and significant variability across regions (Table 1).
Mercury (Hg), recognized for its high toxicity, also merits consideration. The mean soil Hg content in the study locale was 0.19 mg·kg −1 , notably surpassing both the background values for the A layer soil in Guangxi and China (0.152 mg·kg −1 and 0.065 mg·kg −1 , respectively). Nonetheless, when compared to the established soil pollution risk screening thresholds and risk control limits for agricultural land, these data should be interpreted with caution [25]; the proportion exceeding the soil screening threshold was very low (0.7%), indicating that the safety risk of Hg in crop products was low.
Overall, Cd and As are the most highly contaminated PTEs in the soil of the study area, and their ecological risks are worthy of attention. The average contents of Cd and As in soils are 11.5 and 2.92 times the national soil background value, respectively. Meanwhile, in the study area, Cd contents in soils are also much higher than the average contents of the world's soils (0.35 mg·kg −1 ) and Chinese soil (0.23 mg·kg −1 ) [7,28] However, the safety risk assessed according to the national standard is relatively small. In the karst areas of Guangxi, the average soil Cr content is 178.9 mg·kg −1 , much higher than the background value of 82.1 mg·kg −1 in Guangxi and 61 mg·kg −1 in China. The average Cr content in the world's soil is only 40 mg·kg −1 [28]. The proportion of samples with Cr exceeding the risk screening threshold and risk control limit of agricultural land reached 32.4% and 0.4%, respectively, and there may be a risk that a small number of crop products do not meet food quality and safety standards. The average content of Pb in the soil was 58.29 mg·kg −1 , 5.8% of the samples exceeded the risk screening threshold, and 0.4% of the samples exceeded the risk control limit, which has a certain risk of crop product safety.
Zinc (Zn), Copper (Cu), and Nickel (Ni) are essential elements for life, but excessive accumulation in the soil will have a negative impact on the quality of crop products and the ecological environment. They can also migrate to the human body through the food chain and endanger human health. For example, excessive intake of Zn may cause human cholesterol imbalance [29]. The average contents of Zn, Cu, and Ni in the soil of the karst area in Guangxi are 1.95, 1.58, and 1.46 times the background value of soil in China, respectively (Table 2). Compared to the average Zn (90 mg·kg −1 ), Cu (30 mg·kg −1 ), and Ni (20 mg·kg −1 ) content in the world's soil [28], the Zn and Ni contents in the study area are much higher, and the Cu contents are only slightly higher. According to the excess degree of the average content of PTEs relative to the soil background value, the concentration level of PTEs in agricultural soils is ranked as Cd > As > Cr > Hg > Pb > Zn > Cu > Ni. If the proportion exceeding the risk screening thresholds of agricultural land is used as the judging standard, the pollution risk ranking of PTEs is Cd > As > Cr > Zn > Pb > Cu > Ni > Hg. It can be seen that in the study area, the elevated level and risk of Cd, As, Cr, Pb, and Zn in the surface soil are worthy of attention. Among them, Cd has the highest concentration level and the greatest potential risk, which is the most urgent for evaluation and research.

Relationships between Major Elements and PTEs in Agricultural Soils
The composition of major elements has always been considered to be an important factor affecting the concentration levels of PTEs in soil. Some studies have found that the influence of Fe 2 O 3 and FeO on the contents of PTEs such as Cr, Zn, Cd, and Pb in carbonate-rock-derived soils is particularly obvious [30]. Quezada-Hinojosa et al. [31] found that Al has a good positive correlation with Cr, Ni and other trace elements, while Ca has a significant negative correlation with PTEs in the carbonate-rock-derived soil in the Swiss high-geochemical-background area.
The Pearson correlation coefficient is widely used to determine the relationship between major and trace elements of samples. The Pearson correlation coefficients of major and trace elements, TOC, and pH in agricultural soil samples in the study area are shown in Figure 2. K 2 O, Na 2 O, and CaO showed no significant correlation with PTEs, which indicated that the soil minerals incorporated with K 2 O, Na 2 O and CaO as the main components, such as feldspar, had poor adsorption capacity for the PTEs. Al 2 O 3 is an important component of aluminum-containing minerals in soil. The relationship between Al 2 O 3 and various PTEs can explain the influence of soil aluminumcontaining minerals as carriers on the elevated levels of PTEs to a certain extent. The common aluminum-containing minerals in soil mainly include clay minerals and feldspar minerals. Many studies posit that, unlike feldspar minerals, clay minerals in soil are important carriers of PTEs and can significantly enhance their adsorption capacity [32,33]. It can be seen from Figure 2 that except for Cd, Al 2 O 3 has a significant positive correlation with other trace elements, while SiO 2 has a significant negative correlation with each PTE, indicating that clay minerals, such as gibbsite formed after mineral desiliconization, may be important carriers of trace elements. It is worth noting that there is a significant positive correlation between Cd and CaO in soil, indicating that secondary carbonate in soil may be an important carrier of Cd. MgO also shows a significant positive correlation with various PTEs. Due to the high degree of weathering of the soil in the karst area of Guangxi, it usually does not contain magnesium-containing minerals such as mica. The magnesiumcontaining minerals in soils of the study area are generally vermiculite or transition phase mineral hydroxy-aluminum vermiculite [6]. These minerals may be important carriers of PTEs, which is consistent with the study of Zn-rich limestone-derived soils by Jacquat et al. [34].
Moreover, the Chemical Index of Alteration (CIA), computed as the molar ratio Al 2 O 3 /(Al 2 O 3 + CaO* + Na 2 O + K 2 O) × 100%, serves as an effective gauge for assessing the levels of chemical weathering [35]. The correlation between CIA values and PTEs, except for Cu, was positive. In this study, we used soil properties, major and trace elements data, for stepwise multiple regression analysis to screen the influencing factors of the elevated levels of PTEs. According to the results, soil Fe 2 O 3 and Mn contents, soil pH and TOC values, and weathering intensity (characterized using CIA) are the most important factors influencing the PTE levels in agricultural soils in the study area.
The Pearson correlation coefficient is widely used to determine the relationship between major and trace elements of samples. The Pearson correlation coefficients of major and trace elements, TOC, and pH in agricultural soil samples in the study area are shown in Figure 2. K2O, Na2O, and CaO showed no significant correlation with PTEs, which indicated that the soil minerals incorporated with K2O, Na2O and CaO as the main components, such as feldspar, had poor adsorption capacity for the PTEs.

Factors Influencing Elevated Levels of PTEs in Agricultural Soils
To identify and quantify the key factors impacting the elevated levels of PTEs, the RF model was applied. It produced robust results that explained the relationships between the factors and Cd/As and the predicator variables ( Figure 3). CIA was the most important explanatory variable for Cd, and the pH was a less but still important factor in controlling Cd contents. Mn, Fe 2 O 3 , and TOC were also relatively important factors for Cd. However, the relative importance of all nine factors differed considerably for As. Fe 2 O 3 was the highest-ranked factor for As, and the importance of Mn, TOC, and pH was also high. CIA was a moderately important factor for As contents in soils, whereas SiO 2 , Al 2 O 3 , K 2 O, and CaO played a weak role in determining both As and Cd concentrations (Figure 3). These results were consistent with our previous findings using Pearson correlations and stepwise regression.

Soil Ferro-Manganese Oxides
In addition to clay minerals in soil, iron oxides and manganese oxides have also been shown to be the main adsorption carriers of PTEs. The ferro-manganese oxides play a key role in soil adsorption and immobilization of PTEs due to their large surface area and adsorption capacity [36]. Iron oxide is one of the main factors affecting the trace elements in soil, since Fe is a variable-valence element, and the redox effect of iron oxide has an important impact on the adsorption and fixation of trace elements. Studies have also shown that in carbonate-rock-derived soil samples, the effect of FeO on Cr and Zn accumulation is stronger than that of Fe 2 O 3 [32,37]. In Figure 2, there is a significant positive correlation between Fe 2 O 3 T and PTEs in the soils of the study area; the correlation coefficient between As and Fe 2 O 3 T is as high as 0.798. Some studies have shown that goethite has a strong enrichment and adsorption effect on As [36], and goethite is an

Soil Ferro-Manganese Oxides
In addition to clay minerals in soil, iron oxides and manganese oxides have also been shown to be the main adsorption carriers of PTEs. The ferro-manganese oxides play a key role in soil adsorption and immobilization of PTEs due to their large surface area and adsorption capacity [36]. Iron oxide is one of the main factors affecting the trace elements in soil, since Fe is a variable-valence element, and the redox effect of iron oxide has an important impact on the adsorption and fixation of trace elements. Studies have also shown that in carbonate-rock-derived soil samples, the effect of FeO on Cr and Zn accumulation is stronger than that of Fe2O3 [32,37]. In Figure 2, there is a significant positive correlation between Fe2O3 T and PTEs in the soils of the study area; the correlation coefficient between As and Fe2O3 T is as high as 0.798. Some studies have shown that goethite has a strong enrichment and adsorption effect on As [36], and goethite is an important mineral component in the soil in the study area, which proves that iron oxides have a significant impact on the accumulation of PTEs in carbonate-rock-derived soils.
According to the three-dimensional surface between the Cd, Zn and Fe concentration in the tillage layer soil in the study area (Figure 4), when the Fe content is less than 12%, there is a very positive correlation between Cd, Zn and Fe. However, when the Fe content is greater than 12%, the relationship between the Cd, Zn and Fe content began to deteriorate, mainly because the soil type of the samples in this area was mainly Ferralsols. Due to the existence of tiny ferro-manganese nodules in the soil, Cd and Zn concentrations in these ferro-manganese nodules are significantly higher than those in ordinary soils [38][39][40]. The nodules and soil particles were analyzed after grinding and digestion together, thus affecting the relationship between Fe, Cd and Zn in the soil. According to the three-dimensional surface between the Cd, Zn and Fe concentration in the tillage layer soil in the study area (Figure 4), when the Fe content is less than 12%, there is a very positive correlation between Cd, Zn and Fe. However, when the Fe content is greater than 12%, the relationship between the Cd, Zn and Fe content began to deteriorate, mainly because the soil type of the samples in this area was mainly Ferralsols. Due to the existence of tiny ferro-manganese nodules in the soil, Cd and Zn concentrations in these ferro-manganese nodules are significantly higher than those in ordinary soils [38][39][40]. The nodules and soil particles were analyzed after grinding and digestion together, thus affecting the relationship between Fe, Cd and Zn in the soil.
Compared with other adsorption carriers in soil, manganese oxides have a stronger ability to adsorb cations of PTEs. For instance, the ability of manganese oxides to adsorb Pb is more than 40 times that of iron oxide. Manganese oxide is also an important component of ferro-manganese nodules. The concentrations of Cd, As, Zn and other trace metals in nodules are much higher than those in the surrounding soil [41][42][43][44]. In Figure 2, there is a significant positive correlation between Mn and PTEs in the soils of the study area. In particular, the correlation coefficient between Mn and Ni is as high as 0.678, which shows that manganese oxides can significantly influence the concentrations of PTEs in carbonate-rock-derived soils. The adsorption of manganese oxides to PTEs in soil is also affected by the soil pH, competing cations in soil solution, soil mineral types and surface negative charges.  Compared with other adsorption carriers in soil, manganese oxides have a stronger ability to adsorb cations of PTEs. For instance, the ability of manganese oxides to adsorb Pb is more than 40 times that of iron oxide. Manganese oxide is also an important component of ferro-manganese nodules. The concentrations of Cd, As, Zn and other trace metals in nodules are much higher than those in the surrounding soil [41][42][43][44]. In Figure 2, there is a significant positive correlation between Mn and PTEs in the soils of the study area. In particular, the correlation coefficient between Mn and Ni is as high as 0.678, which shows that manganese oxides can significantly influence the concentrations of PTEs in carbonaterock-derived soils. The adsorption of manganese oxides to PTEs in soil is also affected by the soil pH, competing cations in soil solution, soil mineral types and surface negative charges.

Soil pH and Organic Carbon
Soil organic carbon and pH have been considered to be important factors affecting the accumulation of PTEs in soil [33,45]. Liu et al. [46] found that there was a good correlation between concentrations of trace metals, especially Cd, and organic matter in the soils from high geochemical background areas in Chongqing, China. Soil pH is an important factor affecting the distribution, migration, and enrichment of PTEs, since the hydrogen ion concentration in soil solution may significantly affect the adsorption of metal ions such as Cu 2+ , Zn 2+ , and Cd 2+ by soil minerals such as kaolinite, alumina and silica [47]. The general trend in the soil mineral adsorption capacity of metal ions such as Cu 2+ , Zn 2+ , and Cd 2+ is that with the increase in soil pH, the adsorption capacity of metal ions increases and has a characteristic change range (1-2 pH units) with pH change. The adsorption capacity increases suddenly in this range, even up to 100%, mainly because the pH affects the morphology of metal ions in soil solution. Figure 2 shows that soil pH values in the study area are significantly correlated with Cr, Ni, Zn, Cd, As, and Hg, indicating that the increase in soil pH is the main reason for promoting the concentrations of PTEs in carbonate-rock-derived soils. This is

Soil pH and Organic Carbon
Soil organic carbon and pH have been considered to be important factors affecting the accumulation of PTEs in soil [33,45]. Liu et al. [46] found that there was a good correlation between concentrations of trace metals, especially Cd, and organic matter in the soils from high geochemical background areas in Chongqing, China. Soil pH is an important factor affecting the distribution, migration, and enrichment of PTEs, since the hydrogen ion concentration in soil solution may significantly affect the adsorption of metal ions such as Cu 2+ , Zn 2+ , and Cd 2+ by soil minerals such as kaolinite, alumina and silica [47]. The general trend in the soil mineral adsorption capacity of metal ions such as Cu 2+ , Zn 2+ , and Cd 2+ is that with the increase in soil pH, the adsorption capacity of metal ions increases and has a characteristic change range (1-2 pH units) with pH change. The adsorption capacity increases suddenly in this range, even up to 100%, mainly because the pH affects the morphology of metal ions in soil solution. Figure 2 shows that soil pH values in the study area are significantly correlated with Cr, Ni, Zn, Cd, As, and Hg, indicating that the increase in soil pH is the main reason for promoting the concentrations of PTEs in carbonate-rock-derived soils. This is owed to the decrease in H+ increasing the negative charge on the soil surface and promoting the adsorption of metal ions by various minerals in the soil [48].
Soil organic carbon can adsorb and stabilize trace elements such as Cd and Hg; moreover, organic carbon in soils can be used as a reservoir of PTEs to provide a steady stream of minerals. The TOC content is the result of a balance between the amount of organic matter entering the soil in the form of plant residues and the amount of organic matter lost due to microbial decomposition. In Figure 2, TOC was significantly correlated with Ni, Cu, Zn, Cd, and Hg, indicating that the increase in soil organic carbon content was the main reason for promoting the accumulation of PTEs in carbonate-rock-derived soils, because organic matter can adsorb PTEs or form stable complexes with humus in the soil, which reduces the amount of leaching migration of PTEs in the soils [33,49].

Weathering Intensity
The duration of weathering directly affects the development of weathering crust. Therefore, the content of major elements such as Al, K, Ca, and Na in the soil and the relative ratio reflect the weathering intensity. The degree of soil weathering restricts the geochemical behavior of PTEs such as As, Cd, Ni, Pb, and Zn. The CIA values of laterite from southwestern China changed from approximately 75% to 95%, while those of loess from northeastern China ranged from approximately 50% to 75%, and the average CIA value of soil in the study area reached 90.8%, indicating that Ca, Na and K components had strong leaching with respect to Al during the weathering process of the parent rocks. The weathering degree of the tillage layer soil in the study area is quite high. Particularly, with the increase in soil weathering intensity, the content of PTEs in soil also increases.
During the process of soil weathering, the distribution, migration, and enrichment of PTEs can vary greatly. The widely distributed carbonate-rock-derived soils in the karst area of Guangxi have been through complex weathering and soil formation stages. During the early stage, the soluble minerals of carbonate rocks are dissolved and lost in the karst water system, and the insoluble components remain. Then, the soil solution changes from alkaline to acidic, and the soil mineral changes from illite to kaolinite. During the final weathering stage, the aluminosilicate minerals are completely decomposed to form gibbsite and manganese oxides, which, together with clay minerals, form a ferro-manganese oxide layer covering the soil surface. The contents of ferro-manganese minerals increase continuously and the ferro-manganese nodules in the soil also increase; thus, the contents of Fe 2 O 3 and Mn in soils are very remarkably high. Therefore, as the degree of weathering increases, the importance of Fe 2 O 3 and Mn in controlling the concentrations of PTEs, especially As, Cr, Ni, Pb, and Zn, also increases.

Conclusions
This research delved into the chemical compositions of 278 soils originating from carbonate rocks in Guangxi. The objective was to discern the factors contributing to the elevated concentrations of PTEs in agricultural soils within karstic zones. Notably, the PTE contents within the Guangxi karstic soils markedly exceeded the established national soil background values. Of particular significance were Cd and As, which emerged as the predominant PTEs in the study region. Due to their high toxicity, the ecological risks of the PTEs, especially for Cd and As, are worthy of attention. The average content of Cd and As reached 11.56 and 3.54 times the national soil background value, respectively. The most significant factors affecting the distribution, migration, and enrichment of the PTEs in soils include soil properties (soil pH and organic carbon), mineral composition (ferro-manganese oxides), and weathering intensity. The weathering intensity and soil iron oxide played the most important roles in determining Cd and As concentrations in agricultural soils, respectively. This implied that the Cambisols, which are young soils at an incipient stage of soil formation with lower Fe contents, may be the ideal soil type to grow safe and healthy crops. These findings in typical karst regions shed light on the geochemical behaviors of PTEs and provided the scientific basis for regional agricultural land management strategies.

Data Availability Statement:
The data is unavailable due to confidentiality agreement.