Ecotoxicity of Zinc in Spiked Artificial Soils versus ... - ACS Publications

Not only does terrestrial ecotoxicity vary between species but also the soil characteristics greatly influence the effect concentration of metals by a...
0 downloads 0 Views 80KB Size
Environ. Sci. Technol. 2001, 35, 4295-4300

Ecotoxicity of Zinc in Spiked Artificial Soils versus Contaminated Field Soils KOEN LOCK* AND COLIN R. JANSSEN Laboratory of Environmental Toxicology and Aquatic Ecology, Ghent University, J. Plateaustraat 22, 9000 Gent, Belgium

Metal bioavailability is influenced by soil characteristics and aging period. In the present study, both factors were modeled by comparing metal bioavailability in spiked artificial soil and historically contaminated field soils. The chronic toxicity of zinc to Folsomia candida in spiked artificial soils could be predicted with a model based on pH, cation exchange capacity, and total zinc concentration. However, this model could not adequately predict chronic zinc toxicity in contaminated field soils. Porewater concentration and water- and calcium chloride-extracted zinc fractions of the contaminated field soils were lower than those predicted using models developed for spiked artificial soils, indicating that the effect of aging on metal bioavailability should be taken into account. The reproduction of F. candida in contaminated field soils was lower than predicted with models developed using zinc concentration in the porewater and the water- and calcium chloride-extracted fractions in spiked artificial soils. This suggests that these fractions are not the only bioavailable zinc fractions and that dietary metal exposure might also be an important route of uptake under environmentally relevant conditions. Aging and dietary uptake should be studied urgently in order to be able to perform effect-based risk assessments of metal contaminated soils.

Introduction Environmental risk assessments of metals are currently mainly based on laboratory toxicity data obtained with various test species. Using these data, a species sensitivity distribution is fitted from which a hazardous concentration (e.g., HC5, hazardous concentration for 5% of the species) and a predicted no effect concentration (PNEC) for the ecosystem are derived (1, 2). Not only does terrestrial ecotoxicity vary between species but also the soil characteristics greatly influence the effect concentration of metals by altering their bioavailability. The importance of considering metal bioavailability in risk assessments of contaminated soils is widely recognized (3). Indeed, not accounting for changes in bioavailability may lead to under- or overestimation of the PNEC, which can have serious consequences for the ecological system under consideration. This is especially true when assessing the environmental risks of essential metals such as zinc for which not only toxicity but also possible deficiency effects should be considered. Furthermore, soil quality criteria derivations are usually based on results of ecotoxicity tests in which freshly metal-spiked soils are used. In this type of experiments, the soils have not been * Corresponding author telephone: +32 (0)9 264 37 07; fax: +32 (0)9 264 37 66; e-mail: [email protected]. 10.1021/es0100219 CCC: $20.00 Published on Web 10/02/2001

 2001 American Chemical Society

aged long enough to reach equilibrium. Soil quality criteria based on this type of toxicity data considerably overlap with background concentrations of zinc (4). This is due to the fact that both toxicity data and background concentrations are expressed in total concentrations and not in bioavailable concentrations. Only a few studies have compared the toxicity of zinc in spiked artificial soils with that of historically Zncontaminated field soils (5-7). Zinc toxicity in soils collected near a zinc smelter was much lower in comparison with the spiked soils, but it remained unclear whether this is due to the soil parameters influencing bioavailability or to the effect of aging. In two other studies, Smit et al. (8) and Smit and Van Gestel (9) compared zinc toxicity immediately after spiking and after aging for up to 26 months under field or laboratory conditions. Zinc toxicity decreased considerably after aging; however, as a parallel increase in pH of at least one unit was observed, it remained unclear what the relative contribution to the zinc toxicity was of pH change and aging. Theoretically, the most important soil characteristics influencing metal partitioning are the adsorption phases (clay, organic matter, and metal oxyhydroxides); the amount of available sorption sites, which is related to the cation exchange capacity (CEC); the pH; and the competitive sorbed ions (10). The porewater hypothesis or equilibrium partitioning concept states that the effect of a chemical on soil organisms is related to the concentration in the porewater, which in turn is dependent on the sorption behavior of the chemical (11). However, this porewater hypothesis is only valid when uptake via the porewater is the dominant route of uptake. Spurgeon and Hopkin (5) observed that zinc toxicity for Eisenia fetida decreased with increasing pH and increasing organic matter (OM) content of the soil. Lock et al. (12), using a multivariate test design, reported that the acute zinc toxicity to Enchytraeus albidus is mainly governed by pH and CEC. On the basis of these parameters, they developed a model that could predict zinc toxicity in spiked artificial soils, and their results were validated with spiked field soils. No other multivariate approaches to quantify ecotoxicological effects of metals as a function of soil parameters are available. The main objective of the present study was to derive statistical models capable of predicting chronic zinc toxicity to F. candida as a function of pH, CEC, and zinc concentration. To assess their relevance, these models were validated with historically contaminated field soils in order to take the effect of aging into account. Not only total zinc concentrations in the soils but also porewater concentrations and waterand calcium chloride-extracted zinc fractions were measured to assess their predictive value for chronic zinc toxicity. The application of these models may lead to a more realistic risk assessment of metal-contaminated soils.

Experimental Section Toxicity Assays. The culture of the springtail Folsomia candida Willem 1902 (Insecta, Collembola) was obtained from Aquasense B.V. (Amsterdam, The Netherlands). Animals are cultured on a substrate of plaster of Paris and pulverized chemical activated charcoal in a ratio of 8:1 (w:w). Granulated dry yeast is added weekly as a food source. Cultures have been maintained in our laboratory for at least 5 yr at 20 °C and in complete darkness. Chronic toxicity tests with F. candida were carried out according ISO (13). Ten 10-12-day-old synchronized springtails were exposed per glass vessel containing 30 g wet weight of soil. Tests were carried out at a soil moisture content of 50% of the water holding capacity. Soil moisture content VOL. 35, NO. 21, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4295

was adjusted twice a week by replenishing weight loss with the appropriate amount of deionized water. Granulated dry yeast was added weekly on the soil surface as a food source. During exposure, vessels were kept at 20 ( 1 °C and a light: dark cycle of 16:8 at 400-800 lux. The reproduction test with F. candida takes 4 weeks to complete. At the end of the test, juveniles were photographed and counted after flotation. The number of juveniles in the controls always exceeded the prescribed minimum of 100 instars per vessel. EC50 values were calculated using the moving average method (14). NOECs were calculated by Kruskal-Wallis ANOVA followed by post-hoc multiple comparisons (15). Experimental Design. To develop a second-order mathematical model relating the effect concentrations of zinc to the environmental variables, a central composite design (CCD) was applied (16). In a full factorial design, all factors are tested at all possible combinations of settings with two levels per variable. In a CCD, center points are added to the cube design points of a factorial design to allow explicit testing for curvature. In a second-order CCD, the cube is enhanced by a star; these are the design points that allow estimation of the nature of the curvature. These star points are chosen so that the design is rotatable. Specifically, they allow one to estimate the quadratic components of the relationships between the factors and the dependent variable. The obtained surface-response relationships can be written as follows:

28-d EC50 or NOEC repr. F. candida ) 2

2

intercept + aP1 + bP2 + cP1 + dP2 + eP1P2 where the intercept and the factors a-e are the parameters of the surface and P1 and P2 are the values of the tested parameters. In the CCD developed in this study, the pH was varied from 4 to 7, and CEC was varied by adding 0-10% OM to assess the influence of these environmental parameters on the chronic toxicity of zinc to F. candida. Models relating the porewater concentration and the extracted fractions of zinc to CEC, pH, and total zinc concentrations as well as models relating the reproduction of F. candida to the different zinc fractions were developed using partial least squares projection to latent structures (PLS) (17). PLS is a generalization of regressions based on latent variables to find a linear or polynomial relationship between a set of prediction variables X (NobservationsKvariables) and a set of response variables Y (NobservationsMvariables) or one singleresponse variable y. The R2 of the model is a measure for the variance explained by the model while Q2 is measure for the variance of the variables that can be predicted by the model. The Q2 for a significant model has to be greater than a critical value (Q2 limit ) 0.097, which corresponds with p < 0.05; 17). PLS has some important advantages as compared to multiple linear regression: PLS allows the number of prediction variables to be greater than the number of observations, PLS can work with high numbers of correlated X variables (multiple collinear data), and PLS tolerates a certain amount of missing data. Multiple linear regressions do not work adequately under these conditions. For the analyses in the present study, CEC and zinc concentrations were log-transformed prior to PLS modeling; pH was not transformed because it is already based on a log scale. The CEC rather than the OM content was used in the model development because this parameter is a measure of the amount of available sorption sites and thus incorporates the clay-metal oxyhydroxides as well as OM content. Test Soils. Artificial soils were composed with the same ingredients as those recommended by OECD (18). Finely ground peat was used as OM, pH was adjusted ((0.1 unit) with CaCO3, and coarse sand was used to make up the mixture to 100%. Zinc was added as aqueous solutions of chloride salt (ZnCl2, Merck, Leuven, Belgium, pro analysis). Zinc was 4296

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 21, 2001

TABLE 1. pH, CEC, and Zinc Concentrations in the Different Fractions of Contaminated Field Soils place Basel Dilzen Hoboken Hoboken Kruibeke Kruibeke Kruibeke Lommel Lommel Lommel Lommel Lommel Lommel Lommel Lommel Overpelt Overpelt Overpelt Overpelt a

total H2O extd CaCl2 extd (mg/kg CEC (mg/kg porewater (mg/kg dry wt) dry wt) pH (cmol/kg) dry wt) (mg/L) 6.09 7.03 4.23 5.90 5.05 5.63 5.48 5.64 5.70 3.63 5.76 5.73 4.70 4.65 4.46 5.95 5.93 6.17 4.27

24.6 16.2 1.78 6.11 21.4 18.3 23.9 4.00 3.11 2.01 5.11 4.67 3.33 7.28 5.31 5.78 6.44 0.71 1.85

157 409 93.5 274 508 581 222 761 490 133 5180 1140 81.2 206 63.6 847 787 3720 556

nda nd 12.7 2.90 4.43 8.03 nd 13.0 11.7 44.1 177 223 5.78 44.6 36.8 6.20 1.60 427 449

5.94 4.22 3.91 1.94 4.95 5.25 6.55 1.86 2.57 4.73 40.4 42.4 2.51 8.37 5.99 2.29 1.70 29.4 50.4

5.84 2.01 19.1 19.2 24.0 31.4 6.56 126 104 77.4 286 211 10.7 82.0 13.1 80.1 73.0 320 223

nd, not detected.

spiked in logarithmic series with four concentrations per order of magnitude (... 100, 180, 320, 560, 1000, ...). The models developed on the basis of the spiked artificial soils were validated using historically contaminated field soils (mainly zinc). These field soils were selected in such a way that the main parameters determining bioavailability varied considerably: pH ranged from 3.63 to 7.03, and CEC ranged from 0.71 to 24.6 cmol/kg (Table 1). The soils from Lommel, Overpelt, and Hoboken are sampled in former smelter areas while the Dilzen soil is contaminated by sludge of the river Meuse, and the Basel and Kruibeke soils are contaminated by sludge of the river Scheldt. Characterization of Soil Parameters. At the end of the exposures, pH and CEC were measured. pH (KCl) was measured (Consort pH meter, P407, Turnhout, Belgium) at a 1:2.5 soil:liquid ratio with 1 M KCl (19). CEC was determined with the AgTu method in 0.4 M ammonium acetate (20). The water holding capacity was determined by measuring the water content of the soil after inundating it for 3 h in water and subsequently draining it for 2 h (21). The soil water content was determined by weighing the sample, drying it to constant mass at 105 °C, and reweighing it. Metal Analysis. Soils were digested in hot acid (HCl:HNO3 1:5 v/v, microwave heating), and soil zinc concentrations were measured using flame atomic absorption spectrometry (Varian, SpectrAA-100, Victoria, Australia) with a deuterium background correction. A calcareous loamy soil (CRM 141 R, Community Bureau of Reference, Brussels, Belgium) was used as the certified reference material. Measured concentrations in the spiked soils and the reference material never differed more than 10% from the nominal and the certified values, respectively. Water-soluble and calcium chloride-extracted fractions were obtained by shaking 5 g of soil in 50 mL of deionized water or 50 mL of a 0.01 M CaCl2 solution for 2 h at 200 rpm and subsequent filtration through a 0.45-µm cellulose membrane filter (Gelman Sciences, Ann Arbor, MI). Porewater was collected by centrifuging the moist soil for 45 min at 1800g (10 °C) using polypropylene centrifuge tubes equipped with a 0.45-µm polyether sulfone membrane filter (Eppendorf, Laborimpex, Brussels, Belgium). Zinc concentrations in the extracts and the porewater were measured by flame atomic absorption spectrometry (Varian, SpectrAA-100, Victoria, Australia) with deuterium background correction.

TABLE 2. OM Content, CEC, and pH of Artificial Soils Used for CCDa soil parameters OM (% dry wt)

CEC (cmol/kg)

pH

28-d NOEC

1.46 1.46 0 5 5 5 8.54 10 5 8.54

3.03 3.06 0.06 7.77 8.05 8.18 12.2 14.1 9.00 13.9

4.4 6.6 5.5 4.0 5.5 5.5 4.4 5.5 7.0 6.6

32 56 100 100 320 320 180 320 320 560

toxicity data 28-d EC50 50.5 (37.0-63.0) 103 (71.4-135) 133 (100-180) 202 (160-251) 297 (245-339) 336 (298-369) 416 (385-450) 600 (436-767) 656 (456-883) 1000 (817-1220)

a Chronic toxicity of zinc to F. candida expressed as 28-d NOEC and 28-d EC50 values for reproduction (mg/kg dry wt) with indication of the 95% confidence intervals.

FIGURE 1. Simulation runs of the model describing the 28 d of EC50 of zinc to Folsomia candida as a function of the CEC and the pH.

Results Surface-Response Modeling (CCD). Reproduction in the controls of the artificial soils used in the CCD was not significantly influenced by pH (p ) 0.23) nor by OM content (p ) 0.10); therefore, differences found in effect concentrations between soils will essentially be related to the bioavailability of zinc rather than to the soil characteristics. A CCD was used to model the surface-response relationship between pH and OM content and the chronic toxicity (expressed as EC50 and NOEC values) of zinc to F. candida (Table 2). OM content, pH, and the interaction between both parameters are the significant factors contributing to the overall variability of the ecotoxicity of zinc to F. candida (p < 0.05). Toxicity significantly decreases with increasing OM content and pH. Moreover, the significant interaction coefficient (OM × pH) indicates that the effect of OM content and pH reinforce each other. The quadratic factors (pH2 and OM2) were not significant (p > 0.05) and were therefore not retained for the development of the model. The following surface-response relationship equations were derived:

28-d EC50 (mg of Zn/kg dry wt) ) -26.3pH - 127OM + 35.4pH × OM + 174

(R2 ) 0.92, p ) 0.00012)

28-d NOEC (mg of Zn/kg dry wt) ) -34.3pH - 96.4OM + 23.7pH × OM + 264 (R2 ) 0.81, p ) 0.0028) with OM ) % organic matter (w/w).

FIGURE 2. Observed concentrations of zinc in the porewater (A), the CaCl2-extracted fraction (B), and the water-extracted fraction (C) of the spiked artificial soils (b) and the contaminated field soils (O) vs the concentrations predicted using the PLS models developed for the spiked artificial soils. There is a linear relationship between the OM content and the CEC; i.e., CEC (cmol/kg) ) 1.41OM (% dry wt peat) + 0.93 (R2 ) 0.98) (12). The equations could therefore be rewritten using CEC instead of OM: VOL. 35, NO. 21, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4297

FIGURE 3. Observed reproduction (% of control) of Folsomia candida after 28 d of exposure to spiked artificial soils (b) and contaminated field soils (O) vs the reproduction predicted using the PLS models developed for the spiked artificial soils with porewater concentrations (A), CaCl2-extracted concentrations (B), water-extracted concentrations (C), and total concentrations (D) of zinc.

28-d EC50 (mg of Zn/kg dry wt) ) 13.0pH - 31.3CEC + 15.5pH × CEC - 113

2

(R ) 0.89, p ) 0.00037)

28-d NOEC (mg of Zn/kg dry wt) ) -15.3pH 39.1CEC + 12.1pH × CEC + 126 (R2 ) 0.80, p ) 0.0033) (with CEC in cmol/kg). Simulation runs with these models show that the 28-d EC50 of zinc for F. candida decreases with decreasing pH or CEC (Figure 1). From these simulations, it is also obvious that the pH effect increases with increasing CEC. Similar results were obtained for the 28-d NOEC values (not shown). Zinc Partitioning (PLS). To test if the porewater concentration and the water- and calcium chloride-extracted zinc concentrations could be predicted from pH, CEC, and total zinc concentrations, PLS models were used. The porewater concentration (R2 ) 0.90, Q2 ) 0.83), the waterextracted fraction (R2 ) 0.93, Q2 ) 0.88) as well as the calcium chloride-extracted fraction (R2 ) 0.90, Q2 ) 0.87) of the spiked artificial soils could be predicted very accurately (Figure 2, b). However, if this PLS model, developed for the spiked artificial soils, was applied to the contaminated field soils, the zinc concentration in all these fractions was generally 4298

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 21, 2001

lower than that predicted by the model, especially at the highest concentrations (Figure 2, O). Zinc Toxicity (PLS). To assess the relationship between the different zinc fractions and the reproduction of F. candida, PLS models were applied. The porewater concentration (R2 ) 0.78, Q2 ) 0.76) (Figure 3A), the calcium chloride-extracted fraction (R2 ) 0.70, Q2 ) 0.66) (Figure 3B), and the waterextracted fraction (R2 ) 0.61, Q2 ) 0.57) (Figure 3C) were good predictors of the chronic toxicity of zinc in spiked artificial soil (b) and is in contrast to the total zinc concentration of the soil (R2 ) 0.34, Q2 ) 0.33) (Figure 3D), which is a poor predictor of toxicity. When the reproduction of F. candida in the contaminated field soils was predicted using the models developed for the spiked artificial soils (Figure 3, O), the reproduction was often lower than predicted. As the porewater zinc concentrations and the calcium chloride-extracted zinc concentrations in some of the contaminated field soils were lower than in all the spiked soils, the PLS models predicted a very high reproduction of F. candida in these soils. As the control soils themselves were not included in the PLS models, the predicted reproduction was sometimes even higher than 100%. A PLS model based on the same parameters as the CCD (CEC, pH, and total zinc concentration) resulted in good

predictions for the spiked artificial soils (R2 ) 0.84, Q2 ) 0.78), but predictions were very bad for the contaminated field soils (not shown).

Discussion In contrast to the potworm E. albidus (22), reproduction of the springtail F. candida was not influenced by soil characteristics such as pH and OM content. F. candida therefore seems to be better suited to assess the ecotoxicity of contaminated field soils with varying soil properties. Several studies report the chronic toxicity of zinc to F. candida in artificial soils. The observed EC50 values are compared to the EC50 values predicted by the surfaceresponse model developed in the present study, which predicts the EC50 as a function of pH and OM content (CEC was not reported in these studies). Sandifer and Hopkin (23) assessed the influence of pH on zinc toxicity to F. candida and found 28-d EC50 values of 590, 600, and 900 mg of Zn/kg dry wt at a pH of 4.5, 5, and 6, respectively, while the model predictions were 379, 543, and 870 mg of Zn/kg dry wt, respectively. A 28-d EC50 for F. candida of 626 (526-744) mg of Zn/kg dry wt was reported by Van Gestel and Hensbergen (24) while the pH in their experiment varied from 5.37 to 6.23; the model predictions were 664 and 946 for these pH values. Smit and Van Gestel (9) reported a 28-d EC50 for F. candida of 487 (407-582) mg of Zn/kg dry wt while the model predicted 870 mg of Zn/kg dry wt. A comparison of the literature data with the model predictions indicates that, despite inter-laboratory variations, the developed model gave good predictions of the chronic zinc toxicity to F. candida. Lock et al. (12), using a fractional factorial design, found that pH and CEC were the most important soil parameters affecting acute zinc toxicity to the potworm E. albidus. By varying pH and CEC according to a CCD, a surface-response model for zinc toxicity was developed, which was also done in the present study. A good correlation was found between the 14-d LC50 for E. albidus (12) and the 28-d EC50 for F. candida (present study) exposed to the same artificial soils spiked with zinc (R2 ) 0.89). These results may indicate that, if the sensitivity of a species is taken into account, it is possible to extrapolate the toxicity of zinc from one species to another, provided the metal uptake routes are similar. Porewater concentrations and water- and calcium chloride-extracted zinc fractions of the contaminated field soils were lower than those predicted by PLS models developed for spiked artificial soils and using pH, CEC, and total zinc concentration. This is probably due to the effect of aging: with time, zinc fixation results in a lower bioavailability. Fixation of zinc is probably mainly related to the type and content of clay and metal oxyhydroxides (25-28). The rate and extent of metal fixation, however, has to be studied in more detail to allow prediction of metal bioavailability in soils. If the zinc concentration in the porewater or the extracted fractions would be the only bioavailable fraction, toxicity would be easily predictable on the basis of these zinc fractions. Indeed, in contrast to the total metal concentrations in the soil, the reproduction of F. candida in spiked artificial soils could be well-predicted by PLS models based on these fractions, independent of the pH and the CEC. However, reproduction in the contaminated field soils was lower than predicted with these models, indicating that these fractions are not the only bioavailable zinc fractions and that other routes of exposure should be considered. For sediments, for example, it was found that dietary exposure is an important route of metal uptake and that, especially at environmental relevant concentrations, bioaccumulation could not be predicted by porewater metal concentrations alone (29, 30). In most terrestrial toxicity experiments, high concentrations of metals are added, and equilibrium times are short. These

experimental features result in acute toxicity from artificially high porewater concentrations before the implications of dietary exposure can become clear. The porewater concentration-based approach may be appropriate for protecting soil organisms from acute toxicity associated with exposure in extremely contaminated soils. However, the most common regulatory and scientific challenge is the need to determine the ecological implications of moderately contaminated soils. Clearly, the contribution of dietary metal uptake to chronic metal toxicity needs to be investigated for the different feeding guilds of soil invertebrates.

Acknowledgments This research was partially supported by the Public Waste Agency of Flanders. We thank Tom Van Wichelen and Emmy Pequer for the metal analyses and Stijn Verbeke for the practical assistance.

Literature Cited (1) Van Straalen, N. M.; Denneman, C. J. Ecotoxicol. Environ. Saf. 1989, 18, 241-251. (2) Aldenberg, T.; Slob, W. Ecotoxicol. Environ. Saf. 1993, 25, 4863. (3) Fairbrother, A.; Glazebrook, P. W.; Van Straalen, N.; Tarazona, J. V. Summary of the SETAC Workshop on Test Methods for Hazard Determination of Metals and Sparingly Soluble Metal Compounds in Soils, June 19-23, 1999, San Lorenzo de El Escorial, Spain; Society of Environmental Toxicology and Chemistry (SETAC): Pensacola, FL, 1999. (4) Lock, K.; Janssen, C. R. Environ. Toxicol. Chem. 2001, 20, 19011908. (5) Spurgeon, D. J.; Hopkin, S. P. Pedobiologia 1996, 40, 80-96. (6) Smit, C. E.; Van Gestel, C. A. M. Appl. Soil Ecol. 1996, 3, 127136. (7) Spurgeon, D. J.; Hopkin, S. P. Ecotoxicology 1995, 4, 190-205. (8) Smit, C. E.; Van Beelen, P.; Van Gestel, C. A. M. Environ. Pollut. 1997, 98, 73-80. (9) Smit, C. E.; Van Gestel, C. A. M. Environ. Toxicol. Chem. 1998, 17, 1132-1141. (10) Janssen, R. P. T.; Peijnenburg, W. J. G. M.; Posthuma, L.; Van Den Hoop, M. A. G. T. Environ. Toxicol. Chem. 1997, 16, 24702478. (11) Van Gestel, C. A. M.; Rademaker, M. C. J.; Van Straalen, N. M. In Biogeodynamics of Pollutants in Soils and Sediments; Salomons, W., Stigliani, W. M., Eds.; Springer-Verlag: New York, 1995; pp 171-192. (12) Lock, K.; Janssen, C. R.; De Coen, W. M. Environ. Toxicol. Chem. 2000, 19, 2666-2671. (13) ISO 11267. Soil Quality: Effects of Soil Pollutants on Collembola (Folsomia candida): Method for Determination of Effects on Reproduction; International Organisation for Standardisation: 1999. (14) Stephan, C. Methods for Calculating an LC50; ASTM Special Technical Publication 34; American Society for Testing and Materials: Philadelphia, 1977; pp 65-84. (15) Conover, W. J. Practical Non-parametric Statistics; John Wiley & Sons: New York, 1980. (16) Statsoft. Statistica; Tulsa, OK, 1994; Vol. III, pp 3613-3682. (17) Simca. SIMCA-P for Windows: Graphical Software for Multivariate Process Modelling; Umetri AB: Umea, Sweden, 1996. (18) Earthworm, Acute Toxicity Tests; OECD 207; Organisation for Economic Cooperation and Development: 1984. (19) Soil Quality-Determination of pH; ISO 10390; International Organisation for Standardisation: 1994. (20) Chhabra, R.; Pleysier, J.; Cremers, A. In Proceedings of the International Clay Conference in Mexico; Bailly, S. W., Ed.; Applied Publications: 1975; pp 439-449. (21) Soil Quality-Effects of Pollutants on Earthworms (Eisenia fetida). Part 2: Determination of Effects on Reproduction; ISO 11268-2; International Organisation for Standardisation: 1996. (22) Dirven-Van Breemen, E. M.; Baerselman, R.; Notenboom, J. Onderzoek naar de Geschiktheid van de Potwormsoorten Enchytraeus albidus en E. crypticus (Oligochaeta, Annelida) in Bodemecotoxicologisch Onderzoek; Report 719102025; Rijksinstituut voor Volksgezondheid en Milieuhygie¨ne: Bilthoven, The Netherlands, 1994. (23) Sandifer, R. D.; Hopkin, S. P. Chemosphere 1996, 33, 24752486. VOL. 35, NO. 21, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4299

(24) Van Gestel, C. A. M.; Hensbergen, P. J. Environ. Toxicol. Chem. 1997, 16, 1177-1186. (25) Ford, R. G.; Bertsch, P. M.; Farley, K. J. Environ. Sci. Technol. 1997, 31, 2028-2033. (26) Martı´nez, C. E.; McBride, M. B. Environ. Sci. Technol. 1998, 32, 743-748. (27) Martı´nez, C. E.; McBride, M. B. Environ. Sci. Technol. 1999, 33, 745-750. (28) Trivedi, P.; Axe, L. Environ. Sci. Technol. 2000, 34, 2215-2223.

4300

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 21, 2001

(29) Lee, B. G.; Griscom, S. B.; Lee, J. S.; Choi, H. J.; Koh, C. H.; Luoma, S. N.; Fisher, N. S. Science 2000, 287, 282-284. (30) Lee, B. G.; Lee, J. S.; Luoma, S. N.; Choi, H. J.; Koh, C. H. Environ. Sci. Technol. 2000, 34, 4517-4523.

Received for review January 17, 2001. Revised manuscript received May 22, 2001. Accepted July 16, 2001. ES0100219