Barium and Cadmium in Tropical Soils Cropped and Under Native Forest

Toxic elements pose a high environmental risk because of their long persistence in soil, water, and food chain. This study aimed to estimated potentially available and pseudototal contents of barium (Ba) and cadmium (Cd) in tropical soil under native forest vegetation, sugarcane and maize crops. Soil samples were collected at 0.00–0.20 m depth in different municipalities in São Paulo State, Brazil, and analyzed for fertility, texture, total iron, iron oxides, pseudototal and available Ba and Cd contents. Heavy metals were extracted using different extraction solutions (Mehlich-1, Mehlich-3, and DTPA). Data were subjected to descriptive and multivariate analyses. Correlations between soil clay content, mineralogy, and fertility were also investigated. Of the three extraction solutions tested, Mehlich-3 was the most effective to estimate the potential availability of Ba and Cd. Ba extracted by Mehlich-3 was negatively correlated with goethite, and pseudototal barium was positively correlated with pH CaCl 2 . Cd extracted by Mehlich-3 was positively correlated with pH CaCl 2 , and pseudototal cadmium was strongly correlated with iron oxide, clay, and organic matter contents.


Introduction
Heavy metals occur naturally in soil but their concentrations can be increased by anthropic activities. Such metals may remain in soil for extremely long periods like thousands of years. Heavy metals half-life varies according to the type of metal (Alloway and Jackson 1991; Brokes 1995). Some of them, such as Cu, Zn, Mo, Se, and Fe, are essential for plants and animals, while others do not have any known bene ts, such as Cd and Ba (Cipollini and Pickering 1986). Any of these elements, however, can be toxic at high concentrations. Heavy metals are among the most studied elements because they pose environmental and health risks. Contamination of soil, water, and crops with heavy metals may damage human health because they can accumulate in the trophic chain (Hooda and Alloway 1998; Chen et al. 2015).
Knowledge of the availability of heavy metals is essential to monitor the environmental impacts caused by these elements (Pascual et al. 2004;Liu et al. 2018). In soil, heavy metals can occur in bioavailable forms, being readily absorbed by plants. Phytoaccumulation is in uenced by adsorption, leaching, and other soil properties (Qian et al. 1996;Hooda et al. 1997;Fontes and Alleoni 2006). The presence of heavy metals has increased in inadequately managed agricultural systems, especially with the indiscriminate use of agricultural inputs as phosphate fertilizers, and contaminated sewage sludge (Halim et al. 2003), possibly leading to reductions in crop yield and increased risks of biomagni cation and bioaccumulation (Coscione et al. 2009).
It is common the use of chemical extraction methods to assess the availability of soil heavy metals to plants (Revoredo and Melo 2006). These techniques allow evaluating the dynamics of nutrients and heavy metals in soil-plant relationships and their mobility in the soil pro le (Rauret 1998). However, one of the main problems encountered in the development of remediation programs for areas contaminated with heavy metals is the low reliability of methods for estimating total and available contents of these metals (Tavares and Oliveira 2017

Material And Methods
Twelve samples of Latossolos under native forest or cropped were collected in ve municipalities in São Paulo state, Brazil. These soils were previously characterized by Oliveira (1977) and Andrioli and Centurion (1999). The samples were taken from areas cropped with sugarcane (Saccharum spp.), maize (Zea mays L.) or under native forest or old reforestation. Sampling sites and sample codes are described in Table 1. Plots with 100 m 2 were demarcated at each native forest site. Soil samples were collected within the plots and in their vicinity with the same soil type but cultivated with sugarcane or maize. Twenty individual samples were collected from each site at 0.00-0.20 m depth using a Dutch auger and combined to form a composite sample. All sampling points were georeferenced. Samples were air-dried, crushed, and sieved through a 2.00 mm mesh screen for analysis of fertility, texture, mineralogy, Cd and Ba availability, and pseudototal contents. Soil fertility (   Results were subjected to descriptive analysis. After standardization of variables to null mean and unit variance (µ = 0, σ = 1), data were subjected to hierarchical cluster analysis using Euclidean distance as a measure of similarity between observations and Ward's method as a clustering strategy. Principal component analysis (PCA) was also performed. The criteria adopted for choosing the number of retained principal components (PCs) were those proposed by Kaiser (1958): eigenvalues greater than 1.00 and cumulative variance greater than 70%. For a better understanding of data behavior, we performed Pearson correlation analysis at the 5% signi cance level. All statistical analyses were carried out using RStudio version 4.0.     Mehlich-3 was found to be the best extraction solution for Ba, followed by DTPA and Mehlich-1. Mehlich-3 was able to extract potentially available Ba from all samples. For Cd, Mehlich-3 also provided the best results, followed by Mehlich-1. DTPA did not extract relevant amounts of Cd.

Results And Discussion
According to Tavares and Oliveira (2017), environmental agencies that monitor metal-contaminated areas usually opt for extraction methods that afford the highest contents. The recovery rates of Ba and Cd obtained by each extraction solution as percentage of pseudototal extraction are depicted in Fig. 1 established that the prevention value for nonagricultural soils is 120 mg kg −1 Ba and the intervention value is 500 mg kg −1 Ba. In the present study, the pseudototal content of Ba was 171.69 mg kg −1 in native forest soil and 222.08 mg kg −1 in cropped soil. In evaluating the available Ba contents of these sites, we found that the recovery rate was lower than 50% (Fig. 1). Biondi et al. (2011) argued that in-depth analyses are needed to understand metal mobility and availability in undisturbed soils with high heavy metal contents.
We used multivariate exploratory analysis to better characterize soils with regard to data variability. Fig. 2 shows the dendrogram and phylogenetic tree of pseudototal Ba content and Ba contents obtained using Mehlich-1, Mehlich-3, and DTPA solutions. Three distinct groups were formed by using a Euclidean distance close to 4. Group 1 was formed by LV JC and LV JF ; group 2 by LV RC , LA IF , LA IC , LVA IF , and LA MF ; and group 3 by LV SC , LV SF , LA MC , LVA IC , and LV RF . Group 1 soils had higher pseudototal Ba contents, standing out from the other groups.
Principal component analysis revealed only one factor with an eigenvalue greater than one. However, for biplot construction, we used PC1 and PC2 (Fig. 3), which together explained 94.97 % of the total variance contained in the original data. The position of variables on the biplot con rmed the results of cluster analysis. PC1 explained 78.65 % of the variance in data, correlating positively with contents extracted by the USEPA method (pseudototal contents), DTPA, and Mehlich-3, in descending order. PC2 accounted for 16.32 % of the original variance, positively correlating with Mehlich-1 extraction results. PCA afforded two PCs with eigenvalues greater than 1. Variables had a strong correlation (>0.7) with PC1 and PC2, as shown by the biplot in Fig. 5. Both PCs explained 80.61% of the total variance in data, 55.50% of which was explained by PC2 and 25.11% by PC1. PC1 was more strongly correlated with Mehlich-1, DTPA, and pseudototal content, in descending order, whereas PC2 was correlated with Mehlich-1 only.
The results of Pearson correlation analysis (Fig. 6) demonstrated that available Ba determined by Mehlich-3 extraction had a negative correlation with goethite/(goethite + hematite) ratio, suggesting that soils with minor goethite contents have higher Ba availability. Available Ba correlated positively with pseudototal Ba. Pseudototal Ba showed a positive correlation with pH CaCl 2 , indicating that soils with higher pH had higher pseudototal Ba contents.
Cd extracted by Mehlich-3 showed a positive correlation with pH CaCl 2 and pseudototal content. Pseudototal Cd, in turn, was highly correlated with total iron, clay, and organic matter contents. According to Pichtel et al. (2000), Cd availability is in uenced by total Cd concentration, medium pH, and oxyreduction potential.
Harter and Naidu (2001) reported that the soil attributes that most affect metal availability are pH, solution composition and ionic strength, element species and concentration, and presence of ligands and competing ions. In particular, pH is positively correlated with adsorption of metals in soil; the availability and mobility of heavy metals decrease as the pH increases in soils with variable loads (Sposito, 2008). It is noteworthy that the pH of the study soils ranged from 4.1 to 5.6 (acidic soils), except for LV JF , which had a pH of 6.4 (     Pearson correlation matrix for Fe2O3, clay, kaolinite/(kaolinite + gibbsite) [kt/(kt + gb)] ratio, goethite/(goethite + hematite) [gt/(gt + hm)] ratio, P resin, organic matter (OM), pH in CaCl2, cation-exchange capacity (CEC), cadmium, and barium in Latossolos in São Paulo State, Brazil. Nonsigni cant correlations (p > 0.05) are marked with an "X"