Evaluation of Holistic Approaches to Predicting the Concentrations of

Measurements of metals in soils by diffusive gradients in thin films (DGT) have previously been shown to be linearly related to metals measured in sho...
0 downloads 0 Views 197KB Size
Environ. Sci. Technol. 2008, 42, 7649–7654

Evaluation of Holistic Approaches to Predicting the Concentrations of Metals in Field-Cultivated Rice Y U A N T I A N , † X I A O R O N G W A N G , * ,† JUN LUO,‡ HONGXIA YU,† AND HAO ZHANG‡ State Key Laboratory of Pollution Control and Resource Reuse, School of the Environment, Nanjing University, Nanjing 210093, China, and Department of Environmental Science, Lancaster Environment Center, Lancaster University, Lancaster LA1 4YQ, United Kingdom

Received November 6, 2007. Revised manuscript received June 11, 2008. Accepted June 30, 2008.

Measurements of metals in soils by diffusive gradients in thin films (DGT) have previously been shown to be linearly related to metals measured in shoots of plants grown in pots. We examine the relationships between metals measured by DGT and other techniques with metals in the roots and unpolished grains of rice cultivated under field conditions at 18 sites in Jiangsu province, China. Rhizosphere soils of rice were collected and the concentrations of Cd, Cu, Pb, and Zn were determined on soil solution, acetic acid, and calcium chloride (CaCl2) extractions and by DGT. Simple linear regression analyses between concentrations of metals in plants and those measured using DGT and chemical extractions showed a very good fit for DGT measurements of the concentrations of all four metals in both rice roots and unpolished grains. Good fits were also found using soil solution and acetic acid extractions, but the correlation coefficients were lower than those obtained by DGT. CaCl2 extractions provided the poorest fits for all four metals. Multivariate analyses were used to assess the impact of pH, dissolved organic carbon (DOC), soil organic carbon (SOC), cation exchange capacity (CEC), and texture. Two principal components were extracted. The first was well correlated with SOC, DOC, and clay proportion and is therefore representative of “organic matter”. The second primarily correlated positively with pH and negatively with CEC and is representative of “inorganic ions”. When these principle components were included in multiple linear regression, correlation coefficients for plots involving metals in soil solution and in extractions using acetic acid and CaCl2 were improved, but there was little change in the correlation coefficients for comparable plots using metals measured by DGT. These results show for the first time that the DGT measurement quantitatively incorporates the main factors affecting bioavailability.

Introduction It is important to be able to assess the potential bioavailability of metals in soils. Total metal concentrations are recognized to be inappropriate for such purposes (1). Chemical extrac* Corresponding author phone: +86-25-8359-5682; fax: +86-258359-5222; e-mail: [email protected]. † Nanjing University. ‡ Lancaster University. 10.1021/es7027789 CCC: $40.75

Published on Web 09/11/2008

 2008 American Chemical Society

tions of the solid phase and measurements of metal in soil solution can provide alternative approaches. However, these methods have their disadvantages. For example, with chemical extractions it might be difficult to avoid redistribution and readsorption of metals, which can bias the results (2). It is still not clear whether the free metal ion activity in the soil solution can be used to represent the bioavailability of metal to plants (3-7). Uptake of metals by plants involves dynamic interactions between the solid phase, solution and root. The newly introduced diffusive gradients in thin films (DGT) technique accounts for both the soil solution concentration, and the dynamic supply of metal from the solid phase (8). The DGT device is composed of a membrane filter, diffusive gel, and resin gel. When it is deployed on the surface of the soil, it simulates the perturbation of the root in the soil by lowering the local soil solution concentration and, hence, causing the transport of metals from labile pools in the solid phase to solution (9). The metal diffuses through the membrane filter and diffusive gel and is immobilized by the resin gel. The concentration gradient established in the diffusive gel forms the basis for measurement of the accumulation rate of metal. After the required deployment time, the resin gel is eluted and the concentration of metal captured in the resin gel is measured. Calculation of the concentration of DGTmeasured metal, CDGT, is detailed elsewhere (10). The DGT technique has demonstrated the notion that it is not only metal speciation at equilibrium, but also dynamic chemical interactions and resupply of metal from the soil solid phase that determines metal bioavailability (11, 12). It has been successfully used to predict metal concentration in plants cultivated in pot experiments (12-14). However, some authors consider that the technique is no better than conventional analytical extractions (15) or that it fails to predict bioavailability of metals to plants growing under the stress of metal toxicity (16). Many factors, such as pH and organic matter content, affect the uptake of metals by plants by, for example, regulating the concentration of metals in soil solution. Measurements with DGT and chemical extractions do not attempt to measure a single quantity, but rather can be regarded as holistic measurements that are themselves dependent on a range of soil properties. To date, DGT applications have mainly concentrated on predicting accumulation of metal in roots (12) and in whole shoots of plants cultivated in the laboratory (17). It is important to examine its usefulness in predicting metal concentrations in edible plants and edible tissues of plants grown in the field in order to assess risks of metal pollution in terms of food safety. The present work examined the relationships between the concentrations of Cd, Cu, Pb, and Zn in the roots and unpolished grain of rice grown in the field, and the concentrations of the metals in soil solution, acetic acid extracts, and CaCl2 extracts and concentrations measured by DGT. Our objectives were to investigate whether the accumulations of Cd, Cu, Pb, and Zn in the roots and grain of rice can be predicted by these measurements, and to assess the independence of these holistic variables from soil properties.

Experimental Section Soil and Rice Samples. The 18 rice cultivation sites, which had all been established for many years, were chosen along the Yangtze River at Hanjiang (nos. 1-5), Suzhou (nos. 6-10), Taicang (nos.. 11-13), and Yangzhong (nos. 14-18) in VOL. 42, NO. 20, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

7649

Jiangsu, China. There was no obvious point sources of pollution at these sites, except for sites 4 and 5, which may be affected by a chemical plant and a galvanizing plant. Rice was cultivated under field conditions. Growth lasted for three months after emergence. Normal field management practices were used, e.g., submerged in the growth stage and drained in the grainfilling period. After ripening, the whole rice plants were collected, and rhizosphere soil for each plant was sampled as well. The bulk soil was removed by strongly shaking by hand. After that, the rhizosphere soil was obtained by persistent gentle shaking and then stored at -20 °C. Soils were air-dried and sieved to 2 mm before analysis in the laboratory (18). As soon as the rhizosphere soil was sampled, shoots and roots of the rice were thoroughly rinsed with distilled water and then separated. Roots were washed in an ultrasonic system with deionized water (Millipore Milli-Q water, MA) to remove fine soil particles (19). Shoots and roots were dried in an oven at 60 °C for 72 h to constant weight. Each sample was analyzed in triplicate. Roots and grain, unpolished, but after prior removal of chaff, were digested by a mixture of HNO3-HClO4 (1:1) (20). DGT Measurements. Subsamples of the soils were used to determine the maximum water holding capacity (MWHC). DGT devices (DGT Research Ltd., Lancaster, UK) were deployed following the standard procedure (21). Plastic pots were filled with 30 g of soil, and the 18 soil samples were maintained at 40% MWHC for 48 h followed by 80% MWHC for 24 h. DGT devices were firmly pressed on to the soil surface, being careful to ensure complete contact without too much pressure on the gels of the devices. The pots with devices were kept at 25 °C for 24 h. On retrieval, the devices were washed with deionized water to remove adhered soil particles and then dismantled. Resin gels were eluted with 1 mL of nitric acid (1 mol L-1) for 24 h, and then metal concentrations in the eluents were measured by inductively coupled plasma atomic emission spectrometry (ICP-AES) (Jarrell-Ash J-A1100, CO) or inductively coupled plasma mass spectrometry (ICP-MS) (Perkin-Elmer Elan 9000, MA) after appropriate dilution, with rhodium as an internal standard. Calculation of CDGT is specified elsewhere (10, 11). Soil Solution, Acetic Acid, and CaCl2 Extractions. After retrieval of DGT devices, soils were transferred into 25 mL PTFE tubes and then centrifuged for extraction of soil solution. Subsamples of the original air-dried rhizosphere soils were used for acetic acid and CaCl2 extractions and dry soil weights were used. Acetic acid extraction is the first step of the three-step sequential extraction procedure, provided by the Standards, Measurements and Testing Programme (SM&T) of the European Commission (continuation of the Community Bureau of Reference (BCR) and Measurements and Testing Programme) (22). Briefly, 0.5 g of soil was extracted with 20 mL of acetic acid (0.11 mol L-1), stirring at 20 °C for 16 h. CaCl2 extractions are considered to provide a measure of ion exchangeable metal (23). Briefly, 2 g of soil was extracted with 20 mL CaCl2 (0.01 mol L-1) shaking at 20 °C for 2 h (24). For all three extraction methods, the tubes containing the soil mixture were centrifuged at 3000g for 20 min, followed by filtering through a 0.45 µm membrane (Pall Corporation, NY). Concentrations of metals in the filtrate were measured by ICP-AES or ICP-MS. Soil Physical and Chemical Characteristics. Characteristics of soil, including pH, dissolved organic carbon (DOC), soil organic carbon (SOC), cation exchange capacity (CEC), texture,and total concentration of metals in soil and the roots and grain of rice were determined. Soil pH was measured using a combination electrode and pH meter (LeiCi pHS-3C, Shanghai, China) at a ratio of soil to deionized water of 1:5 (v:v), following ISO 10390:2005. DOC in extracted soil solution was measured using a Shimadzu TOC 5000 (Kyoto, Japan). SOC was determined by wet digestion with K2Cr2O7/H2SO4 7650

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 20, 2008

and CEC using ammonium acetate (20). The texture of soil was measured by particle size analyzer (Malvern Mastersizer 2000, Worcestershire, UK). Soil was digested using a mixture of HNO3-HClO4-HF (1:1:1). Total concentrations of metals were measured by ICP-AES or ICP-MS. Analysis of certified reference materials GBW07429 for soil and GBW07604 for rice (Institute of Geophysical and Geochemical Exploration, Chinese Academy of Geological Sciences, China) gave concentrations of Cd, Cu, Pb, and Zn ranging from 87 to 100%, from 81 to 101%, from 81 to 100% and from 85 to 113%, respectively of the certified values. All samples were above the method detection limits (0.001 µg L-1 for Cd; 0.01 µg L-1 for Cu; 0.01 µg L-1 for Pb; 0.02 µg L-1 for Zn). Certified standard solutions and duplicates of all samples were used to ensure accuracy and precision. Statistical Analyses. Data was analyzed and evaluated with SPSS 13.0 (SPSS Inc., IL. Descriptive statistics, significance analysis, regression, and principal components analysis were performed.

Results and Discussion Characteristics of Soil and Accumulation of Metals in Rice. The total metal concentration, chemical properties, and texture of the soils from the 18 cultivation sites are presented in Table 1. The soils covered a wide range of pH (5.38-7.93) and DOC (6.4-127 mg L-1). Total concentrations of Cd and Zn in soil from sites 4 and 5 were much higher than from other sites, presumably due to industrial sources. Bioconcentration factors (BCFs), namely the ratio of the concentration of metal in rice tissue to the total concentration of metal in soil, were used to quantify accumulation of metals from the surrounding environment. BCFs of roots and unpolished grains cultivated in different soils varied greatly (2.33 ( 2.03 for root Cd; 1.46 ( 1.11 for root Cu; 0.13 ( 0.11 for root Pb; 0.65 ( 0.32 for root Zn; 0.043 ( 0.030 for grain Cd; 0.060 ( 0.020 for grain Cu; 0.011 ( 0.014 for grain Pb; 0.045 ( 0.018 for grain Zn). For both roots and grains the proportion of Pb taken up is less than the other three metals, reflecting the strong partitioning of this metal to solid phases. BCF as defined here using total concentrations does not reflect the real metal bioavailability or toxicity. Total concentrations in soils were not significantly correlated with concentrations of metals in roots or grains (r ) 0.074, p ) 0.785 for root Cd; r ) -0.135, p ) 0.618 for grain Cd; r ) 0.096, p ) 0.705 for root Cu; r ) 0.407, p ) 0.094 for grain Cu; r ) 0.389, p ) 0.110 for root Pb; r ) 0.264, p ) 0.290 for grain Pb; r ) 0.283, p ) 0.288 for root Zn; r ) 0.309, p ) 0.244 for grain Zn). Comparison of Different Techniques for Predicting Metal Bioavailability. Relationships between metal concentrations in roots of rice and the different measurements of metals in soils are shown in Figure 1, Figures S1-S3 and Table 2. For Cd, insets are used to show the cluster of points at low concentration (Figure 1). Correlation coefficients, R2, were obtained using simple linear regressions for all the data of each graph and for the low concentration data of the inserts (inset data) (Table 2). The single data point at high concentrations can have the effect of making plots for all data appear less scattered than plots for inset data. For all data, concentrations of Cd extracted with CaCl2 were linearly related to root or grain Cd (R2 ) 0.48**). However, correlation coefficients were poor for the inset data at the lowconcentration scale (R2 ) 0.25NS and 0.35*). For soil solution measurements, fits were very good for all data and poorer for inset data. The poorer fits at low concentrations are in keeping with the idea that Cd release from the solid phase reflects more the properties of individual soils at low concentrations. The apparent better performance of acetic acid extractions for inset data rather than all data is due to the sole high concentration point for the plant measurements

TABLE 1. Chemical Properties Including pH, Soil Organic Carbon (SOC), Dissolved Organic Carbon (DOC) and Cation Exchangeable Capacity (CEC), Texture and Total Concentration of Cd, Cu, Pb, and Zn in Soils from the 18 Rice Cultivation Sites total concentration of metals (mg kg-1)

Texturea sample

pH

SOC (mg g-1)

DOC (mg L-1)

CEC (cmol kg-1)

clay %

silt %

sand %

Cd

Cu

Pb

Zn

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

7.55 7.57 7.93 7.55 6.00 5.89 5.38 6.36 5.65 6.41 7.65 7.71 7.40 5.90 5.99 6.44 6.70 7.70

9.8 10.8 8.2 10.3 9.9 8.3 7.9 13.0 7.9 9.1 10.1 12.2 12.4 10.4 8.9 9.6 14.6 10.8

26.3 16.8 7.0 14.1 14.0 16.1 6.4 43.7 7.0 25.4 38.4 127 24.4 14.3 16.2 10.9 113 40.9

17.6 15.5 11.6 16.0 17.8 21.6 22.0 15.8 21.8 19.2 17.5 13.8 15.6 17.9 21.0 18.8 17.2 15.5

6.5 6.2 4.4 5.7 4.8 5.4 4.9 6.7 5.4 6.4 6.5 6.8 6.3 6.2 5.9 4.5 7.1 6.7

79.0 77.6 66.1 64.8 58.1 73.5 69.5 75.0 69.3 80.4 72.3 72.4 65.4 74.9 73.8 73.7 75.1 76.2

14.5 16.2 29.5 29.5 37.1 21.2 25.6 18.3 25.3 13.2 21.2 20.8 28.4 18.9 20.3 21.8 17.9 17.1

0.66 0.52 0.44 6.9 33 0.31 0.21 0.30 0.46 0.28 0.63 0.24 0.31 0.42 0.25 0.36 0.30 0.32

37 31 33 44 41 78 33 55 70 44 77 61 33 58 35 36 31 39

40 41 35 67 61 90 33 77 82 61 70 84 34 38 29 35 31 31

129 111 110 693 1093 184 81 145 165 98 180 157 104 123 83 135 89 96

a

Clay, silt, and sand are defined as soils with particles size of 0-2, 2-50, and 50-2000 µm, respectively.

being out of line with the rest of the data. Only for the DGT technique, was there a high probability of fit (p < 0.001) for both data sets and both concentration ranges, suggesting that it is a good predictor of Cd concentrations in roots and grains of rice, even at low-concentrations. Capabilities of predicting accumulation of different metals in rice varied with different analytical techniques. The correlation coefficients followed the descending order: DGT > soil solution > CaCl2 extraction > acetic acid extraction for Cu (SI Figure S1); DGT > acetic acid extraction > soil solution > CaCl2 extraction Pb (SI Figures S2) and DGT ∼ soil solution > acetic acid extraction > CaCl2 extraction for Zn in grain; DGT > soil solution ∼ acetic acid > CaCl2 for Zn in root (SI Figure S3). The values of correlation coefficients for each metal are summarized in Table 2. Generally, the fits of the linear regressions obtained using DGT technique for all the four metals in both roots and unpolished grain were better than the comparable fits obtained for soil solution measurements or chemical extractions. This suggests that interactions between the solid phase and solution may be significant in the rhizosphere soils of rice. The DGT technique reflects the dynamics of these processes. It locally lowers concentrations of metals in the soil solution near the surface of the device and induces both a diffusion gradient in the soil solution and a flux from solid phase to solution to counteract the depletion. The same basic processes occur as roots take up metals in the rhizosphere soil. Free metal ions, inorganic complexes, and a proportion of organic complexes, which roots could take up from the soil system (9), are measured by DGT. Chemical extractions cannot emulate dynamic processes of uptake of metals by roots. However, they may be appropriate for the total accumulation of metals by roots for long-term studies. Whether supply of metals to plants is controlled by the reservoir of available metal (capacity control) or the rate of release (kinetic control) (25), DGT provides a reliable surrogate measurement and is a good predictor of bioavailability of metals to rice. Good predictions have also been found previously, using pot experiments, for wheat (11) and Lepidium heterophyllum (13). The regressions of metal in roots with metals in soil solution gave reasonably good fits. The lower correlation coefficients compared to DGT might be due to methodological problems (26). For example, the use of short equilibration times of one hour (27) or two hours (26) have

been questioned and the use of much longer periods have been suggested (28). Moreover, extraction methods that neglect the influence of water holding capacity ignore the actual growth conditions and may result in poor fits. In our study there should have been adequate time for equilibration, as soils were equilibrated with water for a total of 72 h, the first 48 h at 40% MWHC, and then 24 h at 80% MWHC. Furthermore, soil solutions were extracted following retrieval of DGT devices and therefore must have reflected the same conditions as DGT. Reasonably good fits were obtained for the regression of Cd in roots and unpolished grain with Cd in CaCl2 extractions due to one data point (contaminated soil from site 5). However, at lower concentration range, the plots (Figure 1d and h, inset) showed no significant correlations. The different shape of these plots to those obtained for soil solution measurements is partly due to the different dependence of the different soil measurements on soil properties, such as pH. Reasonably good correlation coefficients for Zn and Cd (excluding 1 high data point) extracted by acetic acid indicate that the acetic acid liberates these metals from the solid phase relevant to the pool available during plant uptake. Extremely low extractability of acetic acid (about 1% of the total metal) was observed for soil 5 and even lower extractability of CaCl2 (about 0.2%) for soil 4. It could be due to (i) the very high total concentration of Cd and Zn and (ii) insufficient shaking time for extractions in those heavily contaminated soils. In soil 5, CaCl2 extracted Zn is higher than concentration obtained by acetic acid extraction. This is unexpected and difficult to explain. Although the precision was good (see error bars in SI Figure S3), other measurements on different samples of this soil have demonstrated high variability with respect to the total and extractable concentrations for Zn, showing its heterogeneity. The precise form of the contaminating Cd and Zn is unknown and they may lead to the unusual ratios of metals extracted by acetic acid and CaCl2. Multivariate Analysis. Principal components analysis (PCA) was used to reduce multidimensional soil properties to lower-dimension parameters. Properties of both solid phase and solution, including pH, SOC, DOC, CEC, and texture might affect the process of uptake. For texture, only the clay proportion was selected, because metals are bound to clay more effectively than to silt or sand. Using PCA, the variance of the data set with interrelated variables (pH, SOC, DOC, CEC, and clay proportion) could be explained by fewer VOL. 42, NO. 20, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

7651

rotation using varimax with Kaiser Normalization, the factors of PC1 for pH, SOC, DOC, CEC, and clay proportion were 0.184, 0.868, 0.876, -0.203, and 0.878; the factors of PC2 were 0.931, 0.283, 0.176, -0.941, and 0.112. PC1 is clearly well correlated with SOC, DOC, and the clay proportion, and therefore is representative for “organic matter”; PC2, which primarily correlates positively with pH and negatively with CEC, is interpreted as “inorganic ions”. Conceptually, this is reasonable, as metals are complexed by “organic matter”, with the complexation being affected by competing “inorganic ions”. These processes may be the main factors determining the bioavailability of metal to roots. Stepwise multiple linear regressions were employed to study how PCs influence the relationships between measurements by chemical analytical techniques and the accumulations of metals by roots. Two PCs and one analytical technique were used as inputs and root metals as outputs. Regressions were respectively performed for DGT measurements, soil solution, acetic acid, and CaCl2 extractions. Results showed that the measurements of Cu by chemical extractions and the DGT technique were significantly influenced by PC2 (eqs 1-4), acetic acid extractable Pb by PC2 (eq 5) and soil solution Pb by both PC1 and PC2 (eq 6). Root Cu ) 6.12 DGT-measured Cu - 20.42 PC2 + 32.84 (R2 ) 0.85***) (1) Root Cu ) 0.64 soil solution Cu - 24.69 PC2 + 34.35 (R2 ) 0.86***) (2) Root Cu ) 3.26 acetic acid extractable Cu - 30.60 PC2 + 55.91 (R2 ) 0.80***) (3) Root Cu ) 15.90 CaCl2 extractable Cu - 29.53 PC2 + 45.81 (R2 ) 0.82***) (4) Root Pb ) 2.21 acetic acid extractable Pb - 2.18 PC2 + 4.40 (R2 ) 0.44**) (5) Root Pb ) 1.22 soil solution Pb - 1.69 PC1 - 2.23 PC2 + 0.90 (R2 ) 0.73***) (6)

FIGURE 1. The relationships between Cd concentrations in rice (root and unpolished grain, respectively) and Cd in soils: (a and e) measured by DGT, (b and f) in soil solution, (c and g) extractable by acetic acid, (d and h) CaCl2 extractable. The bars are the standard deviations for three replicates. The inserts show the relationships for low concentrations. Correlation coefficients are presented in Table 2. independent variables called principal components (PC). Taking eigenvalues >1 as extraction criterion, two PCs were extracted. They cumulatively explained 85% of the total variance of the data set, of which the first (PC1) and second PC (PC2) accounted for 60 and 25%, respectively. After

Multiple regressions explain more variance than simple regressions because the former takes into account additional impacts of properties of soils on uptake of metals by roots. For example, correlation coefficients for Cu measurement of soil solution, acetic acid and CaCl2 extractions rose to 0.86, 0.80, and 0.82, respectively, from 0.52, 0.31, and 0.35. This effect of pH and CEC on the accumulation agrees with other workers and is consistent with the approaches used to investigate the biotic ligand model (29). There is also the possibility that analytical techniques alone could explain the bioavailability of metals to root, without the need for any additional soil properties. Incorporating PC1 and PC2 into the regressions gave virtually no improvement in the fit for DGT for Cd, Pb, and Zn and a modest improvement for Cu (from R2 ) 0.75*** to 0.85***). This indicates that the single measurement of

TABLE 2. Correlation Coefficients (R2) Obtained from Linear Regressions of Plots between Metal Accumulations in Roots/ Unpolished Grains and Measurements of Cd, Cu, Pb, and Zn in Soilsa root measurements

Cdb

Cdc

Cu

DGT soil solution acetic acid extraction CaCl2 extraction

1.00f

0.81f [16]

0.93f 0.27d 0.48e

0.32d [15] 0.60f [16] 0.25NS[15]

unpolished grain Pb

Zn

Cdb

Cdc

Cu

Pb

0.75f

0.85f

0.74f

0.99f

0.82f [16]

0.71f

0.56f

0.52f 0.31d 0.35e

0.40e 0.26d 0.17NS

0.59f 0.60f 0.18NS

0.95f 0.22d 0.48e

0.61f [15] 0.58f [16] 0.35d [15]

0.42e 0.31d 0.14NS

0.15NS 0.08NS 0.15NS

a

Zn 0.64f 0.65f 0.50e 0.07NS

Number in the square bracket is number of data points for regressions. b All data. c Low-concentration data (inset). Denote not significant, p < 0.05, p < 0.01, p < 0.001, respectively (see Supporting Information (SI) Figures S1-S3 for detailed results).

NS,d,e,f

7652

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 20, 2008

DGT automatically takes all the important soil factors into account. With other techniques additional factors must be considered to achieve the same predictive capability as DGT. Taking Pb as an example, no additional soil properties are needed using the DGT technique (R2 ) 0.85***), while soil solution extraction has to rely on both PC1 and PC2 (from R2 ) 0.40** to 0.73***). For acetic acid extraction, even with PC2 it is not particularly good (R2 ) 0.44**). For CaCl2 extraction, even with PC1 and PC2, the relationship is not well established (p > 0.05). During the deployment of the DGT device, Pb transfers from solid phase to solution, through the membrane filter and diffusive gel, and then binds to the resin gel. Soil properties relevant to this dynamic transfer, which also occurs during plant uptake, are embraced by the DGT measurement. Chemical extractions reflect the equilibrium state for a new chemical matrix and therefore may not account for the influence of relevant soil properties. The better prediction of bioavailability of Cd, Zn (11) to wheat, Zn (30) to Lepidium sativum and Cu (13) to Lepidium heterophyllum by DGT compared to other measurements was attributed to the DGT technique fully considering the resupply dynamics of metals from solid phase to solution. Zhao et al. (12) also found that bioavailability of Cu to barley was predicted better by the DGT measurement than either free Cu2+ activity or soil solution Cu. Consideration of soil pH enhanced the prediction of free Cu2+ activity significantly, showing the importance of other factors. However, DGT may sometimes fail to accommodate all the factors that affect uptake by plant. Implications. The present study showed that the DGT technique gave a better prediction of the concentrations of Cd, Cu, Pb, and Zn to both roots and grains of rice cultivated in the field than soil solution, acetic acid and CaCl2 extractions. Inclusion of principle components derived from pH, SOC, DOC, CEC, and clay proportion into stepwise multiple linear regressions improved predictions for measurements based in soil solution and extractions using acetic acid and CaCl2. This analytical approach could not appreciably improve the predictions obtained from direct measurements using DGT, suggesting that the DGT measurement incorporates the main factors affecting bioavailability. We recognize that we have only sampled the soil at maturation and that practical predictions might be based more usefully on DGT measurements performed on the soil prior to planting and during the inundation phase when redox-related processes will be important. Nevertheless, the results indicate that the DGT technique is a promising tool for predicting bioavailability of metals to rice cultivated in the field. In-situ application of the DGT technique merits further study, especially in respect to the challenges posed by the inherent heterogeneity of the soils.

Acknowledgments This work is financially supported by the National Natural Science Foundation of China (grant 20577021), Natural Science Foundation of Jiangsu Province (grant BK2004091), People’s Government of Jiangsu Province, and China Geological Survey of the Ministry of Land and Resource of P.R. China (grant 20031230008 and grant 20031230000903), Scientific Research Foundation of the Graduate School of Nanjing University (grant 2005CL09), EPSRC DHPA grant (NE/C506999/1) and NoMiracle (a project funded by the European Commission under contract 003956). We thank Renzhang Lin and Xuhai Zhou for their assistance with sampling in the field, and Xueyuan Gu for paper revision.

Supporting Information Available Figures S1-S3 show the relationship between accumulations in roots/grains and measurements in soils for Cu, Pb, and Zn, respectively. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Schuwirth, N.; Hofmann, T. Comparability of and alternatives to leaching tests for the assessment of the emission of inorganic soil contamination. J. Soils Sediments 2006, 6, 102–112. (2) Chomchoei, R.; Shiowatana, J.; Pongsakul, P. Continuous-flow system for reduction of metal readsorption during sequential extraction of soil. Anal. Chim. Acta 2002, 472, 147–159. (3) Rice, M. S.; Higgins, L. S.; Snowbarger, J. T. Chalcone synthase gene expression compared in dgt and VFN8. Abstr. Pap. Am. Chem. Soc. 2000, 219, U359–U359. (4) MacDonald, D.; Taillon, K.; Johnson, D.; Hale, B.; Hendershot, W. Modeling the effect of trace metal emissions on boreal forest soils. Hum. Ecol. Risk. Assess 2003, 9, 723–747. (5) McBride, M. B. Cupric ion activity in peat soil as a toxicity indicator for maize. J. Environ. Qual. 2001, 30, 78–84. (6) Batley, G. E.; Apte, S. C.; Stauber, J. L. Speciation and bioavailability of trace metals in water: Progress since 1982. Aust. J. Chem. 2004, 57, 903–919. (7) Collins, R. N. Separation of low-molecular mass organic acidmetal complexes by high-performance liquid chromatography. J. Chromatogr., A 2004, 1059, 1–12. (8) Zhang, H.; Davison, W.; Knight, B.; McGrath, S. In situ measurements of solution concentrations and fluxes of trace metals in soils using DGT. Environ. Sci. Technol. 1998, 32, 704– 710. (9) Ernstberger, H.;.; Zhang, H.;.; Tye, A.;.; Young, S.;.; Davison, W. Desorption kinetics of Cd, Zn, and Ni measured in soils by DGT. Environ. Sci. Technol. 2005, 39, 1591–1597. (10) Zhang, H.; Davison, W. Performance characteristics of diffusion gradients in thin films for the in situ measurement of trace metals in aqueous solution. Anal. Chem. 1995, 67, 3391–3400. (11) Nolan, A. L.; Zhang, H.; McLaughlin, M. J. Prediction of zinc, cadmium, lead, and copper availability to wheat in contaminated soils using chemical speciation, diffusive gradients in thin films, extraction, and isotopic dilution techniques. J. Environ. Qual 2005, 34, 496–507. (12) Zhao, F. J.; Rooney, C. P.; Zhang, H.; McGratht, S. P. Comparison of soil solution speciation and diffusive gradients in thin-films measurement as an indicator of copper bioavailability to plants. Environ. Toxicol. Chem. 2006, 25, 733–742. (13) Zhang, H.; Zhao, F. J.; Sun, B.; Davison, W.; McGrath, S. P. A new method to measure effective soil solution concentration predicts copper availability to plants. Environ. Sci. Technol. 2001, 35, 2602–2607. (14) Cornu, J. Y.; Denaix, L. Prediction of zinc and cadmium phytoavailability within a contaminated agricultural site using DGT. Environ. Chem. 2006, 3, 61–64. (15) Koster, M.; Reijnders, L.; van Oost, N. R.; Peijnenburg, W. Comparison of the method of diffusive gels in thin films with conventional extraction techniques for evaluating zinc accumulation in plants and isopods. Environ. Pollut. 2005, 133, 103–116. (16) Almas, A. R.; Lombnaes, P.; Sogn, T. A.; Mulder, J. Speciation of Cd and Zn in contaminated soils assessed by DGT-DIFS, and WHAM/Model VI in relation to uptake by spinach and ryegrass. Chemosphere 2006, 62, 1647–1655. (17) Sonmez, O.; Pierzynski, G. M. Assessment of zinc phytoavailability by diffusive gradients in thin films. Environ. Toxicol. Chem. 2005, 24, 934–941. (18) Wang, Z. W.; Shan, X. Q.; Zhang, S. Z. Comparison between fractionation and bioavailability of trace elements in rhizosphere and bulk soils. Chemosphere 2002, 46, 1163–1171. (19) Wang, W. S.; Shan, X. Q.; Wen, B.; Zhang, S. Z. A method for predicting bioavailability of rare earth elements in soils to maize. Environ. Toxicol. Chem. 2004, 23, 767–773. (20) Bremner, J. M. In Methods of Soil Analysis. Part 3. Chemical Methods; Sparks, D. L., Ed.; SSSA and ASA: Madison, WI, 1996; pp 1085-1121. (21) OwenK. WarnkenK. ZhangH. Developing a protocol for assessing metal availability in soils by DGT J. Environ. Monit. 2007, submitted (22) Quevauviller, P. Operationally defined extraction procedures for soil and sediment analysissI. Standardization. TrAC, Trends Anal. Chem 1998, 17, 289–298. VOL. 42, NO. 20, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

7653

(23) Houba, V. J. G.; Lexmond, T. M.; Novozamsky, I.; van der Lee, J. J. State of the art and future developments in soil analysis for bioavailability assessment. Sci. Total Environ. 1996, 178, 21–28. (24) Houba, V. J. G.; Temminghoff, E. J. M.; Gaikhorst, G. A.; van Vark, W. Soil analysis procedures using 0.01 M calcium chloride as extraction reagent. Commun. Soil Sci. Plan 2000, 31, 1299– 1396. (25) Lehto, N. J.; Davison, W.; Zhang, H.; Tych, W. Analysis of micronutrient behaviour in the rhizosphere using a DGT parameterised dynamic plant uptake model. Plant Soil 2006, 282, 227– 238. (26) Crommentuijn, T.; Doornekamp, A.; VanGestel, C. A. M. Bioavailability and ecological effects of cadmium on Folsomia candida (Willem) in an artificial soil substrate as influ-

7654

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 20, 2008

(27) (28) (29) (30)

enced by pH and organic matter. Appl. Soil Ecol. 1997, 5, 261– 271. Bakker, E.; Buhlmann, P.; Pretsch, E. Carrier-based ion-selective electrodes and bulk optodes. 1. General characteristics. Chem. Rev. 1997, 97, 3083–3132. Meers, E.; Du Laing, G.; Unamuno, V. G.; Lesage, E.; Tack, F. M. G.; Verloo, M. G. Water extractability of trace metals from soils: Some pitfalls. Water Air. Soil Pollut. 2006, 176, 21–35. Peijnenburg, W.; Jager, T. Monitoring approaches to assess bioaccessibility and bioavailability of metals: Matrix issues. Ecotox. Environ. Safe. 2003, 56, 63–77. Zhang, H.; Lombi, E.; Smolders, E.; McGrath, S. Zn availability in field contaminated and spiked soils. Environ. Sci. Technol. 2004, 38, 3608–3613.

ES7027789