Introduction

Most risk assessment procedures for contaminated soils assume that all of the contaminants present in soils, in terms of total (or pseudototal) metal concentrations, are available for absorption by the human gastrointestinal tract (Romic and Romic 2003; Tiller 1989). This assumption can overestimate potential effects and health risks and has implications for the remediation of contaminated sites (Juhasz et al. 2003). The hazard that potentially toxic trace elements (TE) represent to living organisms is determined by their chemical forms (Guo et al. 2006; Gupta et al. 1996; Tiller 1989). In soils, TE are associated with the various components in different ways, and these associations indicate both their mobility in the soils and their availability (Ahumada et al. 1999). To obtain reliable information on the TE toxic effects and the chemical reactivity of their various geochemical forms, sequential extraction procedures are frequently used, which consist of various sequences of different chemical extractions, usually starting with the weakest, least aggressive chemical, and ending with the strongest and most aggressive ones (Davidson et al. 2006). Concerning human health risks, humans assimilate bioavailable TE from contaminated soils through several exposure pathways. Soil ingestion via hand-to-mouth contact by children is a significant direct exposure pathway (Day et al. 1979; Duggan et al. 1985; Wixson and Davies 1994) and can be assessed by measuring the oral bioaccessibility of TE. The bioaccessible fraction of contaminants in soil is defined as the proportion of the contaminant that is desorbed from soil in the human gastrointestinal tract and is potentially available for absorption (Oomen et al. 2002; Paustenbach 2000). Several in vitro approaches have been developed to mimic the effects of the human digestion process (Basta and Gradwohl 2000; Hamel et al. 1999; Medlin 1997; Rodriguez and Basta 1999; Ruby et al. 1993; Wragg and Cave 2003; Yang et al. 2003). However, there are many differences in digestion characteristics between the various in vitro methods that can lead to a wide range of bioaccessibility values (Oomen et al. 2002; van de Wiele et al. 2007). The BioAccessibility Research Group of Europe (BARGE) undertook an international collaborative initiative to develop an unified method (the Unified Barge Method, UBM) capable of providing reproducible, robust, and defensible bioaccessibility data (Cave et al. 2006), to assess human health risk (Button et al. 2009; Denys et al. 2007, 2009). The UBM test has been validated against an in vivo model (young swine) for Cd, Pb, and As. Significant in vivo versus in vitro correlations were obtained against a target organ, i.e., kidney for Cd and Pb and urine for As (Caboche 2009).

The main objective of this study was to contribute to an assessment of the health risks attributed to the soil contamination in an area that was highly affected by the past atmospheric emissions of two smelters in northern France (the lead smelter Metaleurop Nord at Noyelles-Godault and the zinc smelter Umicore at Auby) (Douay et al. 2007; Sterckeman et al. 2000, 2002). The specific objectives were to study TE behavior in 36 agricultural topsoil samples collected surrounding the two smelters in terms of the following: (1) TE accessibility to the human body by using the in vitro UBM test for Cd, Pb, and Zn and (2) TE availability by using sequential extractions. Specific attention was given to the physicochemical soil parameters, which were measured in order to understand how bioavailability is affected by the soil characteristics. The present work completes a first study whose objectives were to evaluate the oral bioaccessibility in urban topsoils from this contaminated area (Roussel et al. 2010).

Materials and methods

Sampling sites

Thirty-six sites were chosen including 25 sites close to the former smelter Metaleurop Nord at Noyelles-Godault (ME) and 11 sites close to Umicore at Auby (UM) (Fig. 1). The sampling sites were chosen in order to represent a large scale of soil metal contamination. The sampled sites were chosen randomly due to agricultural land management. The number of sites sampled close to Umicore is lower, as the contamination ring is more restricted, and the soil use is less agricultural than that of the area surrounding Metaleurop Nord.

Fig. 1
figure 1

Location of the 36 sites in northern France: 26 sites close to Metaleurop Nord at Noyelles-Godault and 11 sites close to Umicore at Auby

Full size image

Physicochemical parameters of topsoils

For each site, a composite topsoil sample was constituted in ploughed layer (0–25 cm). Thirty six soil samples were prepared according to the NF ISO 11464 standard. The samples were air-dried at a temperature below 40°C and crushed to pass through a 2-mm stainless steel sieve. Particle-size distribution was obtained through sedimentation and sieving (NF X 31-107). Soil pH was measured in a water suspension (NF ISO 10390), and organic matter content was obtained by the NF ISO 10694 standard. Total carbonate contents were obtained by measuring the volume of CO2 released after a reaction with HCl (NF ISO 10693). Assimilated P (expressed in g P2O5 kg−1) was measured using an extraction by ammonium oxalate solution and spectrocolorimetric determination (NF X 31-161). These analyses were performed by the INRA Soil Analysis Laboratory (Arras, France). All precautions were taken with respect to the protocol application and the calibration. Quality control was based on the use of certified soil samples (GBW 07401, 07402, 07404, 07405, and 07406) obtained from Standard Materials of Soils Components (Harbin, China), samples from interlaboratory comparisons, internal control samples, and duplicates of the analysis.

Soil extraction procedures

For each of the 36 soil samples, a representative subsample was obtained using an automatic sieve conducted with an ultracentrifugal mill < 250 μm (Retsch type ZM 200, Germany). Subsamples were then used for the determination of Cd, Pb, Zn, Fe, Mn, and Al concentrations in extracting solutions. These concentrations were measured by atomic absorption spectrometry (AAS, AA-6800, Shimadzu, Japan) using a flame (FAAS) (Waterlot et al. 2008).

Pseudototal concentrations of elements

The pseudototal element concentrations (Cd, Pb, Zn, Fe, Mn, and Al) in the soil samples were obtained by microwave-assisted digestion (Berghof SpeedwaveTM MWS-2, Germany). For this, 300 mg of soil subsamples was digested in a mixture of 1.5 mL nitric acid (70%) and 4.5 mL hydrochloric acid (37%) (aqua regia). Quality control was based on a certified sample (BCR CRM 141R).

Single and sequential extractions of elements

Single extraction with a mix of solutions (0.111 mol L−1 sodium bicarbonate, 0.267 mol L−1 sodium tricitrate, and 200 g L−1 sodium dithionite) was performed to extract free Fe, Mn, and Al oxides (Mehra and Jackson 1960).

The Cd, Pb, and Zn fractionation was estimated using an extraction procedure recommended by the Standard Measurement and Testing Program (SM&T) of the European Community, formerly BCR (Rauret et al. 2000). Each fraction was noted as fraction A, B, D, or R and defined respectively as: (a) exchangeable, water- and acid-soluble (0.11 mol L−1 acetic acid), (b) reducible (0.5 mol L−1 hydroxylammonium chloride), (c) oxidizable (8.8 mol L−1 H2O2, followed by 1.0 mol L−1 ammonium acetate at pH 2), and (d) residual (aqua regia). Each suspension was mixed in a mechanical horizontal shaker for 16 h. The BCR CRM 701 was used as standard reference material. The sum of recoveries of individual extraction steps was between 90.2 and 110.8% of the pseudototal concentrations.

In vitro oral bioaccessibility measurement

The bioaccessibility of Cd, Pb, and Zn was measured on subsamples using the in vitro test, based on the Unified Barge Method (UBM protocol, Cave et al. 2006). This protocol consists in two parallel sequential extraction procedures and simulates the chemical processes occurring in the mouth, stomach, and intestine compartments using synthetic digestive solutions according to physiological transit times (Oomen et al. 2003, 2006). The composition of the solutions used is based on the composition found in human physiology (Table 1). During the extraction procedure, temperature was maintained at 37°C.

Table 1 Composition of the digestive solutions used during the phases of the unified barge method (UBM) test (Cave et al. 2006)
Full size table

During the first extraction procedure that simulated the gastric-only phase, soil samples (0.6 g) were mixed with 9 mL of saliva (pH 6.0–7.0). This mix was manually shaken for 5 min. Then, 13.5 mL of gastric solution (pH 0.9–1.0) was added to the soil suspension. The pH of the solution was adjusted at 1.5 with the dropwise addition of 37% HCl or 1 M NaOH. The samples were incubated using an end-over-end rotation at 37°C for 1 h. Then the stomach phase was extracted by centrifuging the suspension at 3,000 g for 5 min. The supernatant was removed and stored at < 4°C prior to the determination of the bioaccessible TE concentrations. To simulate the gastrointestinal phase, a duplicate gastric-phase solution was first produced in parallel and to this 27 mL of intestinal solution plus 9 mL of bile were added. The pH of each suspension was adjusted at 6.3 ± 0.5. Solutions were incubated in the end-over-end rotator for a further 4 h and centrifuged at 3,000 g. The supernatant was then sampled.

For each TE, the bioaccessibility is expressed as a ratio between the extracted concentrations in gastric or gastrointestinal phases and the pseudototal concentrations in the soil. For each sample, tests were carried out in three replicates. For every ten soils, a blank and a reference soil (NIST SMR 2710) were used (Roussel et al. 2010).

Statistical analyses

Statistical analyses were performed using STATISTICA (Statsoft, Tulsa, OK, USA) or XLSTAT 2009.4.06 (Addinsoft). The normality of distribution was checked by means of the Shapiro–Wilk test (Reimann and Filzmoser 1998). Nonparametric data were analyzed using the Mann–Whitney U test. The correlation coefficients were calculated by means of the Pearson’s method. To investigate the main factors determining soil bioaccessibility, a linear multiple regression with stepwise procedure was performed correlating the measured oral bioaccessibility values and soil physicochemical parameters. Before including these variables in the model, the multicollinearity was tested and the significant soil parameters were selected. Statistical analyses were carried out by means of three replicates.

Results and discussion

Physicochemical parameters of the agricultural soils studied

The soils around the former Metaleurop Nord smelter are developed on loessic materials, while those around Umicore originated from alluvial deposits characterized by a large textural variability (Sterckeman et al. 1996, 2000). The main physicochemical parameters of the sampled soils are presented in Table 2. These soils mainly showed a loamy texture with a difference in sandy contents dependent on the industrial site. The studied soils around Umicore had significantly higher sand contents than around Metaleurop Nord. On average, the soil pH was approximately 7.8, and the organic matter (OM) content was approximately 50.2 g kg−1. Significant variations were observed in the values for total carbonates (32.1 ± 67.2 g kg−1), organic matter (50.2 ± 30.0 g kg−1), and assimilated phosphorus (0.388 ± 0.151 g P2O5 kg−1) contents. The Fe, Mn, and Al contents showed a low variation. Regarding the major pollutants (Cd, Pb and Zn), the lowest and highest concentrations were as follows: 2.4 and 13.0 mg kg−1 for Cd, 105 and 824 mg kg−1 for Pb, and 182 and 1,250 mg kg−1 for Zn, whereas the means for the loessic reference soils are 0.4, 38, and 74 mg kg−1, respectively for Cd, Pb, and Zn, and the means for the alluvial reference soils are respectively 0.6, 45, and 78 mg kg−1 (Sterckeman et al. 2002). Moreover, the highest Cd and Pb concentrations were measured in the topsoils around Metaleurop Nord, while the highest Zn concentrations were recorded around Umicore.

Table 2 Mean, standard deviation (SD), minimum, maximum, first quartile (Q1), median, and third quartile (Q3) values of topsoil physicochemical parameters, and pseudototal and human bioaccessible (gastric and gastrointestinal phases) concentrations
Full size table

Cd, Pb, and Zn fractionation in soils

Sequential extractions are frequently used to evaluate the TE distribution within soils and to provide knowledge on metal affinity to the soil components and the strength with which they are bound to the matrix (Davidson et al. 2006). Soil samples were digested in successive extracting solutions to mobilize element fractions with decreasing mobility and availability. Water-soluble and exchangeable fractions are considered readily mobile and bioavailable, while elements incorporated into the crystal lattice are relatively inactive (Gupta et al. 1996). The other TE forms—precipitated as carbonates, occluded in Fe–Mn and Al oxides, or complexed with organic matter, sulfides—are considered to be relatively active fractions, depending on the specific combination of a soil’s physical and chemical parameters (Ahumada et al. 2004; Arain et al. 2008; Jamali et al. 2006, 2007a, b; Kabala and Singh 2001). The distribution of TE in fractions A, B, D, and R, evaluated using sequential extractions, are shown in Fig. 2. Values are expressed for each element as the percentage of pseudototal concentrations. Globally, Cd was mainly present in the exchangeable, water- and acid-soluble fraction (fraction A, 44.1 ± 9.8%) and the reducible fraction (fraction B, 42.7 ± 10.5%), in agreement with findings on other anthropogenically loaded soils (Chlopecka et al. 1996; Kuo et al. 1983; Li and Thornton 2001; Sánchez et al. 1999). Lead was primarily found in fraction B (95.0 ± 10.4%), which is thought to be adsorbed or occluded by iron and manganese oxides. The accumulation of anthropogenic Pb in oxide-bound form is in accordance with earlier findings (Ahumada et al. 1999; Chlopecka et al. 1996; Kaasalainen and Yli-Halla 2003; Karczewska 1996; Ramos et al. 1994; Sahuquillo et al. 1999; Sánchez et al. 1999). The residual fraction (fraction R, related to crystalline structures of minerals) was highest for Zn (29.2 ± 7.4%), and the nonresidual species of Zn were mostly found in fraction B (41.8 ± 9.0%). Ma and Rao (1997) and Narwal et al. (1999) also found Zn to be strongly bound in the residual fraction. A low contribution to the pseudototal metal concentration was defined as the oxidizable form (fraction D, related to organic matter and sulfides), namely, 13.0 ± 5.4%, 3.3 ± 3.6%, and 12.0 ± 2.5%, respectively, for Cd, Pb, and Zn.

Fig. 2
figure 2

Fractionation of Cd, Pb, and Zn in the agricultural topsoils studied (fraction A: exchangeable, water- and acid-soluble fraction, fraction B: reducible fraction, fraction D: oxidizable fraction, fraction R: residual fraction)

Full size image

TE mobility may be obtained on the basis of absolute and relative concentrations of labile forms in which the metal occurs in soil. It may be considered that concentration of metals in the first extracted fraction (fraction A) could be interpreted as an indication of relative metal mobility in soils (Ma and Rao 1997; Narwal et al. 1999). Therefore, an index of relative TE availability (i.e., a mobility factor) was calculated, which corresponds to the ratio of labile fractions (exchangeable, water- and acid-soluble) to the sum of all fractions. High mobility factor values highlighted the relatively high lability and bioavailability of TE in the studied soils. The mobility factor values of Cd, Pb, and Zn were 42.3 ± 9.7%, 1.6 ± 1.2%, and 21.6 ± 8.3%, respectively, for the 36 soils. Mobility factor values are higher for Cd and Zn than for Pb (Cd > Zn ≫ Pb), as was found by several authors (Beesley et al. 2009; Knight et al. 1998; Moreno-Jimēnez et al. 2009; Waterlot et al. 2006). Acetic acid is able to release the remaining specifically adsorbed TE ions. The fraction recovered after acetic acid extraction may contain TE coprecipitated with carbonate minerals but also specifically sorbed to some sites of clay surface, organic matter, and Fe/Mn oxyhydroxides (Pickering 1986). For Pb, acetic acid released little TE from soils. This result could be explained by an incomplete dissolution of Pb bound to carbonates. Indeed, the operating conditions defined in the sequential extraction scheme are acceptable for a low carbonate content of soils (Tack and Verloo 1995), and an incomplete dissolution can be observed in the case of soils with high carbonate contents. In the study, the total carbonate contents varied from 1 to 275 g kg−1 (Table 2). Carbonated species of the various metals have different solubilities, so their dissolution during the acetic acid extraction is sometimes incomplete (Kheboian and Bauer 1987; Gleyzes et al. 2002). Therefore, especially for Pb, carbonate dissolution could continue during the following step of sequential extraction, leaching to an overestimation of the fraction B.

Linear correlations between mobility factor values of Cd and Zn and soil parameters were performed on the 36 soils (Table 3) in order to identify the possible influence of the physicochemical parameters of the soils on relative metal concentrations in soils in fraction A. These results suggest that the potential availability of Cd and Zn (1) increases as the sand content rises and (2) is reduced with increase in clay, carbonates, organic matter, free Fe- and Al-oxides, and pseudototal Fe and Al contents. The increase in the potential availability is due to the fact that sand is a worse sorbent for TE than clay, whereas the decrease in availability may be due to the capacity of clay, organic matter, and oxides to bind metal ions. Moreover, the presence of carbonates indicates a high pH involving conditions in which TE are less mobile than at low pH. However, the mobility factor values for Pb were not significantly correlated with any soil parameter, which can probably be explained by the very low concentrations of Pb in the exchangeable fraction.

Table 3 Pearson’s correlation coefficients between the mobility factor values of Cd and Zn and some topsoil physicochemical parameters
Full size table

Globally, Cd showed higher environmental availability than Pb and Zn. Therefore, a high proportion of mobile Cd may represent a high human bioaccessibility.

Estimation of TE oral bioaccessibility

Quality control of in vitro test for the Cd, Pb, and Zn oral bioaccessibility measurement

The quality control of the unified barge method (UBM) was undertaken at the laboratory. To obtain an insight into the within-laboratory repeatability, three agricultural (coded 09-2954, 09-2955, and 09-2956) and three urban (coded 06-1034, 06-1035, and 06-1092) soils from the study area were chosen. Triplicate extractions (gastric and gastrointestinal) of each soil sample were carried out by one person measuring the same sample repeatedly (Table 4). Compared to the average values obtained in both experiments, the recoveries were between 89.2 and 119.4%.

Table 4 Determination of Cd, Pb, and Zn bioaccessibility (gastric and gastrointestinal phases) by the same operator in agricultural and urban topsoils
Full size table

As for reproducibility, the variability of the oral bioaccessibility (gastric and gastrointestinal) measurement was evaluated taking into account the results recorded by three operators. For this study, four urban soil samples were chosen (Table 5). Statistical analysis (ANOVA) was carried out to assess whether there was a difference in bioaccessibility values among the operators using the Tukey test (α = 5%). No significant difference was observed between bioaccessibility (gastric and gastrointestinal) values according to the three operators.

Table 5 Determination of Cd, Pb, and Zn bioaccessibility (gastric and gastrointestinal phases) by three operators in urban topsoils
Full size table

These results (repeatability and reproducibility) provided evidence for the quality of the UBM test for Cd, Pb, and Zn oral bioaccessibility measurements.

In vitro oral bioaccessibility of Cd, Pb, and Zn

In vitro bioaccessibility of Cd, Pb, and Zn measured in 36 soil samples ranged, respectively, (1) from 1.7 to 9.5 mg kg−1, 58 to 529 mg kg−1, and 40 to 594 mg kg−1 of soil for the gastric phase and (2) from 1.4 to 4.2 mg kg−1, 6 to 172 mg kg−1, and 15 to 251 mg kg−1 of soil for the gastrointestinal phase (Table 2). The mean values of the bioaccessible fractions of Cd, Pb, and Zn during the gastric phase were 83, 55, and 33%, respectively, of the pseudototal concentrations in the soils. During the gastrointestinal phase, the bioaccessible fractions of Cd, Pb, and Zn decreased to 45, 20, and 10%, respectively (Table 6).

Table 6 Mean, standard deviation (SD), minimum and maximum values of Cd, Pb, and Zn oral bioaccessibility in the gastric and gastrointestinal phases (results expressed as the percentage of the pseudototal soil TE concentrations)
Full size table

These results highlighted two behaviors: (1) a difference in oral bioaccessibility of Cd, Pb, and Zn and (2) a significant difference between the TE concentrations obtained in the gastric phase and those recorded in the gastrointestinal phase. The difference in bioaccessibility between the elements (low bioaccessibility of Zn and the higher bioaccessibility of Cd) is most likely due to the different chemical forms in which the TE are bound to the soil constituents. On the other hand, the results recorded by the gastric phase were higher than the data obtained in the gastrointestinal phase. There was a sharp decrease in extracted Cd, Pb, and Zn as they moved from the gastric fluid phase to the intestinal fluid phase of the sequential in vitro extraction. These results may be due to readsorption of TE onto the soil components, complexation by pepsin, or chemical precipitation of elements due to the higher pH environment of the intestinal compartment (Ellickson et al. 2001; Grøn and Andersen 2003; Mushak 1991; Oomen 2000).

The UBM test has been validated against an in vivo model (young swine) for Cd, Pb, and As for the gastric and gastrointestinal extraction phases (Caboche 2009). Very good correlations were observed between the in vitro and the in vivo models; nevertheless, higher correlations were recorded with UBM gastric-phase extraction than with gastrointestinal-phase extraction. Thus, for health risk assessment of Cd and Pb, Caboche (2009) recommends to use the more conservative estimate, i.e., to take into account the values of the gastric phase. The UBM test has not been validated for Zn, although Zn may also be a problem for human health (Merian et al. 2004; Pierzynsky et al. 2005). There are no data in the literature for this element. It may be impossible to establish a relationship between UBM bioaccessible Zn and human bioavailable Zn because human will regulate uptake of excess Zn. However, this study concerns agricultural topsoils formerly contaminated by the atmospheric emissions of two lead and zinc smelters, and Zn constitutes a major pollutant. Thus, the UBM test was also applied to characterize the oral bioaccessibility of Zn. The Zn bioaccessible values constitute a first approach when estimating the human exposure, and not bioavailability.

Relationships between TE concentrations extracted with UBM and fractionation in soils

Relationships between Cd, Pb, and Zn oral bioaccessibility and sequentially extracted TE fractions were determined in an attempt to reveal the possible sources of TE available for intestinal absorption (Fig. 3). A multiple regression analysis was performed between these oral bioaccessible TE fractions and the metals’ chemical fractions in the soils. Consistently significant multiple linear regression equations are shown in Table 7. For Cd, Pb, and Zn, the applied equations explained a variation in bioaccessible metal concentrations from 58 to 93%. The fraction A variable positively contributes to the bioaccessibility of the three TEs. In the gastric phase, fraction B contributed positively to the bioaccessible concentrations of Cd, Pb, and Zn. For Zn, fraction R was a variable that contributed to its bioaccessibility in the gastric phase, and fraction D contributed positively to the concentrations of this metal in the gastrointestinal phase.

Fig. 3
figure 3

Comparison between sequentially extracted TE fractions and Cd, Pb, and Zn bioaccessibility (fraction A, exchangeable, water- and acid-soluble fraction; fraction B, reducible fraction; fraction D, oxidizable fraction; fraction R, residual fraction; G phase, gastric phase; GI phase, gastrointestinal phase). (Results expressed as the percentage of pseudototal soil TE concentrations.)

Full size image
Table 7 Relationships between in vitro oral bioaccessible concentrations of Cd, Pb, and Zn (mg kg−1) and sequentially extracted TE fractions (mg kg−1) as given by stepwise regression models (significance level: p < 0.0001)
Full size table

Many studies have assessed bioavailable Pb fractions in soils, but few have looked at the Cd and Zn oral bioaccessibility. It has been reported that in soils, Pb bound to carbonates, to weak organic/inorganic complexes to Fe/Mn oxides and to organic matter could be significantly depleted by in vitro gastrointestinal extraction treatment (Bosso and Enzweiler 2007; Hettiarachchi and Pierzynski 2004; Lamb et al. 2009; Marschner et al. 2006; Ruby et al. 1999). However, the low average values of bioaccessible Pb were attributed to the presence of more stable Pb minerals, e.g., sulfate, sulfide, and phosphate phases (Davis et al. 1993; Porter et al. 2004; Ruby et al. 1999). We can assume that these behaviors could occur for Cd and Zn. Indeed, it has been established that the major factors governing the distribution and dynamics of Cd, Pb, and Zn in soils are pH, organic matter content, inorganic ligands, hydrous metal oxides, specific clay mineralogy, and competition with other metal ions (Abollino et al. 2006; Alloway 1995; Davranche and Bollinger 2000; Lothenbach et al. 1999; Violante et al. 2003). In our study, positive and significant relationships were obtained between bioaccessible TE in the gastric phase and fractions A and B in the sequential extraction procedure. In the gastric solution, where the pH was low (1.5), Cd, Pb, and Zn were released from soil via their desorption and/or dissolution from the carbonates and part of the oxides. For Cd, the sum of fractions A and B was similar to the bioaccessible Cd concentration. For Pb and Zn, the sum was markedly higher than the bioaccessible concentration, which was attributed to the presence of more stable Pb and Zn minerals under gastric conditions. Moreover, a significant positive correlation was observed between bioaccessible Zn and fraction R, which indicated that Zn could be released from soil via its desorption and/or dissolution, from even the clay-mineral phase.

In the human gastrointestinal tract, absorption of TE occurs predominantly at near-neutral pH. At this pH, TE speciation changes and the most bioaccessible free Cd2+, Pb2+, and Zn2+ are transformed to less soluble species. A near-neutral pH also favors metal readsorption on remaining soil particles or other indigestible materials present in chyme (Davis et al. 1997), decreasing solubility and absorption. Marschner et al. (2006) compared data of sequential extraction with those of bioaccessible Pb concentrations and observed that bioaccessibility was highly correlated with the amount of Pb in the third fraction, i.e., easily reducible oxides. We observed a similar relationship (r = 0.744, p < 0.0001). Moreover, for Zn, bioaccessibility was highly correlated with the amount of Zn in the fraction D (r = 0.691, p < 0.0001), showing that oxidative processes in the intestine could be more relevant for Zn absorption than the initial dissolution in the stomach. As suggested by Marschner et al. (2006), the concentration of soluble TE in the intestinal phase depends on the availability of ligands rather than on metal solubility during the gastric phase. Indeed, at the low pH of the gastric phase, the main TE binding sites in soils (oxides, clays, carbonates, organic matter), are partly solubilized. The solubilized TE are readsorbed or precipitated at the neutral pH in the intestinal phase. The demobilization of TE during the transit from the gastric to the intestinal phase should result in a redistribution of TE among the fractions of the sequential extraction. Therefore, the concentration of soluble TE in the intestinal phase should be more a function of the availability of ligands than that of TE solubilized during the gastric phase.

Relationships between TE concentrations extracted with UBM, pseudototal metal concentrations, and selected soil parameters

A multiple regression analysis was performed between bioaccessible TE, pseudototal metal concentrations, and soil physicochemical parameters. Because some soil parameters can be intercorrelated, a first statistical analysis was used to evaluate the effects of multicollinearity and to select the significant parameters (Table 8). Specifically, a significant linear correlation was observed between organic matter and clay contents (p < 0.0001) that Stevenson (1994) explained by the formation of stable complexes between these constituents. Moreover, significant correlations were also observed between free Al oxides and clay contents (p < 0.0001), which points out that most of the clay fraction is aluminosilicate minerals. Additionally, significant correlations were observed between free Mn oxides and pseudototal Mn (p < 0.0001), free Al oxides and pseudototal Al (p < 0.0001), and pseudototal Al and pseudototal Fe (p < 0.0001). The variance inflation factor (VIF) and the corresponding tolerance values were determined to quantify the severity of the observed multicollinearity. Thus, only the following soil parameters, having a VIF < 5, were included in the multiple regression analysis: pH, contents of organic matter, total carbonates, sand, assimilated P, free Al- and Fe oxides, and pseudototal Fe and Mn (Table 9). Significant correlations (r 2 of 0.64–0.93) were found between concentrations extracted with UBM, pseudototal TE concentrations, and selected physicochemical parameters of the studied soils (Table 9). These interrelations showed that (1) Cd bioaccessibility was affected by sand and free Al oxide contents for the gastric phase and by organic matter and sand contents for the gastrointestinal phase, (2) Pb bioaccessibility was impacted by free Al oxide and carbonate contents for the gastric phase and by pseudototal Fe contents for the gastrointestinal phase, and (3) Zn bioaccessibility was affected by OM contents for the gastric phase and by carbonates, assimilated P, and organic matter contents for the gastrointestinal phase. The negative correlations observed with the soil parameters such as carbonates, free Al oxide, pseudototal Fe, organic matter, and assimilated P contents provide evidence for a decrease in TE bioaccessibility in soils with high values of these physicochemical parameters resulting in their capacity to bind metal ions. For example, the negative relationship between soil organic matter content and bioaccessible Zn suggests that increasing organic matter content has the potential to reduce Zn bioaccessibility in these soils. The negative correlation between total CaCO3 content and bioaccessible Pb in the gastric phase suggests that carbonates may have increased the gastric pH. Indeed, before to be adjusted to 1.5, the pH of the soil–saliva–gastric solution mix varies according to soil parameters. In particular, a significant positive correlation was observed between pH (soil–saliva–gastric solution mix) and CaCO3 content (r 2 = 0.96, p < 0.0001), which indicates that soil samples with high CaCO3 contents result in an increase in the gastric pH. For Pb, there was no correlation between soil organic matter content and bioaccessible Pb. This may be attributed to the large Pb concentrations of soils and the subsequent supersaturation of organic matter binding sites.

Table 8 Pearson’s correlation coefficient (r) among selected soil properties (values are significant at * α = 0.05, ** α = 0.01
Full size table
Table 9 Relationships between bioaccessible concentrations of Cd, Pb, and Zn and soil parameters as given by stepwise regression models (significance level: p < 0.0001)
Full size table

The TE bound to soil components seemed to be stable, but they might dissolve during the gastric phase, involving in the intestinal fluid possible readsorption of TE onto these soil components, thereby reducing their bioaccessibility. The observed relationships can be explained by the interactions of the TE with the different soil compartments. Indeed, Cd, Pb, and Zn occur in soil as a complex mixture of solid-phase chemical compounds of varying particle size and morphology including discrete mineral phases, coprecipitated and sorbed species associated with soil minerals or organic matter, and dissolved species that may be complexed by a variety of organic and inorganic ligands (Alloway 1995; Hickey and Kittrick 1984; McBride et al. 1997; Rieuwerts et al. 2000; Ruby et al. 1999). The distribution of TE among these various phases, as well as the physical relationship between the phases and the soil compounds, controls its dissolution and, hence, its oral bioaccessibility.

In soil, generally, TEs compete for available binding sites (on a soil), resulting in an increase or decrease in oral bioaccessibility of pollutants. In this study, an increase in pseudototal Cd concentrations of soil also resulted in an increased oral bioaccessible Pb and a decreased bioaccessible Zn in the gastrointestinal phase. However, an increase in pseudototal Zn concentrations resulted in a decreased bioaccessible Pb in this same phase. These results provide evidence for competition processes between TE and various ligands present in the digestive solutions. These correlations were improved with input from additional significant soil components as regression parameters.

Poggio et al. (2009) estimated the human bioaccessible concentrations of Pb, Cu, Zn, Ni, and Cr in agricultural and residential topsoils. They performed a regression analysis (n = 56, p < 0.05) taking into account the pseudototal concentrations and selected soil parameters (pH, contents of OM, clay, and sand). They showed that (1) Pb bioaccessibility was affected by organic matter, clay, and pseudototal Pb concentrations for the gastric phase (r 2 = 0.53) and pH, organic matter, sand, and pseudototal Pb concentrations for the gastrointestinal phase (r 2 = 0.49) and (2) Zn bioaccessibility was impacted by organic matter, sand, and pseudototal Zn concentrations for both phases (r 2 = 0.42 and 0.34, respectively). These results are different from our own. Indeed, we introduced additional soil parameters and showed that they may increase the explained variance for studied TE (Table 9).

Conclusion

The contaminated agricultural topsoils studied were investigated in order to establish the Cd, Pb, and Zn environmental availability and oral bioavailability together with the soil parameters commonly known to influence TE solubility and mixed pollutant interactions. The main conclusions obtained were that:

  1. (a)

    The three TE presented a varied chemical distribution: Cd occurred in the more available fractions, Pb was mostly present as adsorbed or occluded by Fe/Mn oxides, and a significant contribution to the pseudototal Zn concentrations was defined as the unavailable residual form related to crystalline structures of minerals.

  2. (b)

    Mobility factor values (assessed by the first step of sequential extractions) followed the sequence Cd > Zn ≫Pb and highlighted a relatively high lability and bioavailability of Cd and Zn.

  3. (c)

    The metal bioaccessibility (as % of pseudototal concentration) followed the sequence Cd > Pb > Zn. Significant differences were observed between bioaccessible values in the gastric phase and those in the gastrointestinal phase. These results were linked to several parameters, particularly to the different chemical forms in which the elements are bound to the soil constituents. Moreover, the concentration of soluble TE in the intestinal phase depended on the availability of ligands rather than on metal solubility during the gastric phase. For risk assessment, it may be more relevant to use the more conservative estimate, i.e., to take into account the values of the gastric phase.

  4. (d)

    The in vitro oral bioaccessible concentrations were affected by the pseudototal TE concentrations and by the physicochemical soil parameters. The multiple regression analysis provided evidence for good interrelations between bioaccessible fractions and significant soil parameters such as sand, carbonates, organic matter, assimilated P, free Al, and pseudototal Fe contents.

The concepts of bioavailability and bioaccessibility are important in quantifying the risks associated with exposure to environmental pollutants. According to these results, Cd showed higher environmental availability and oral bioaccessibility than Pb and Zn in the highly contaminated area studied. Further studies are needed to validate and generalize the relationships obtained on TE human bioaccessibility.