Terrestrial Biotic Ligand Model. 2. Application to Ni and Cu Toxicities

(3) have developed the TBLM and applied it to Cu and Ni toxicity to barley root ...... Mike McLaughlin , Steve Lofts , Michael Warne , Monica Amorim ,...
5 downloads 0 Views 230KB Size
Environ. Sci. Technol. 2006, 40, 7094-7100

Terrestrial Biotic Ligand Model. 2. Application to Ni and Cu Toxicities to Plants, Invertebrates, and Microbes in Soil S A G A R T H A K A L I , † H E R B E R T E . A L L E N , * ,† DOMINIC M. DI TORO,† ALEXANDER A. PONIZOVSKY,† CORINNE P. ROONEY,‡ FANG-JIE ZHAO,‡ STEPHEN P. MCGRATH,‡ PEGGY CRIEL,§ HILDE VAN EECKHOUT,§ COLIN R. JANSSEN,§ KOEN OORTS,| AND ERIK SMOLDERS| Center for the Study of Metals in the Environment, Department of Civil and Environmental Engineering, University of Delaware, Newark, Delaware 19716, Agriculture and Environment Division, Rothamsted Research, Harpenden, Hertfordshire, AL5 2JQ, UK, Laboratory of Environmental Toxicology and Aquatic Ecology, Ghent University, 9000 Ghent, Belgium, and Division of Soil and Water Management, K. U. Leuven, 3001 Heverlee, Belgium

The Terrestrial Biotic Ligand Model (TBLM) is applied to a number of noncalcareous soils of the European Union for Cu and Ni toxicities using organisms and endpoints representing three levels of terrestrial organisms: higher plants, invertebrates, and microbes. A comparison of the TBLM predictions to soil metal concentration or free metal ion activity in the soil solution shows that the TBLM is able to achieve a better normalization of the wide variation in toxicological endpoints among soils of disparate properties considered in this study. The TBLM predictions of the EC50s were generally within a factor of 2 of the observed values. To our knowledge, this is the first study that incorporates Cu and Ni toxicities to multiple endpoints associated with higher plants, invertebrates, and microbes for up to eleven noncalcareous soils of disparate properties, into a single theoretical framework. The results of this study clearly demonstrate that the TBLM can provide a general framework for modeling metals ecotoxicity in soils.

Introduction Contamination of soils with metals is a worldwide problem that could threaten the sustainability of essential soil functions (1, 2). Therefore, a risk assessment utilizing appropriate mechanistic soil quality criteria is needed to properly assess the level of risk. The terrestrial biotic ligand model (TBLM) is a theoretical model that provides a general framework for modeling metals ecotoxicity in soils (3). The TBLM assumes that the metal in the soil is in equilibrium * Corresponding author phone: +1-302-831-8449; fax: +1-302831-3640; e-mail: [email protected]. † University of Delaware. ‡ Rothamsted Research. § Ghent University. | K. U. Leuven. 7094

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 22, 2006

with the metal in the soil solution and that the free metal ions (M2+) in the soil solution bind to the receptor site, the biotic ligand (BL), on the organisms and cause a toxic effect. Additionally, other cations compete with the M2+ to bind to the BL sites. In doing so, these cations (principally protons, Ca2+ and Mg2+) provide a protective effect. In the development and application of the Cu and Ni TBLM for barley root elongation, Thakali et al. (3) have reviewed the competitive effects of cations in metal toxicities to plants. Similar interaction of cations in metal toxicity to soil invertebrates has also been reported (4, 5). Steenbergen et al. (4) reported the protective effects of protons and Na+ in acute Cu toxicity to the earthworm Aporrectodea caliginosa in a biotic ligand model (BLM) model developed using a quartz sand-nutrient solution system to represent soil. The protective effect of protons was observed also in Cd toxicity to the growth and reproduction of the springtail Folsomia candida in soils, indicating a BLM-type interaction (5). Similar competitive interactions of cations have also been indicated in metal toxicities to microbial activities or processes in soils (6-8). The critical M2+ activities of Cu, Ni, and Zn for potential nitrification and glucose induced respiration in soils decreased with increasing pH (7, 8). In an analysis of published data, Lofts et al. (6) found a significant negative correlation between critical M2+ activities and the soil pH indicating possible protective effect of protons in metal toxicities to microbial activities in soils. Despite the direct and circumstantial evidence suggesting competitive interaction of cations and subsequent modification of metal toxicities to the biota, a comprehensive treatment of a wide range of toxicological endpoints, metals, and soils is lacking in the literature. Thakali et al. (3) have developed the TBLM and applied it to Cu and Ni toxicity to barley root elongation in a number of soils. In this paper we extend the same TBLM framework to Cu and Ni toxicities to five additional biological endpoints: tomato (Lycopersicon esculentum cv. Moneymaker) shoot yield, reproduction of the redworm Eisenia fetida and the springtail Folsomia candida, potential nitrification rate, and glucose induced respiration by soil microbes.

Experimental Section Selection of Soils. The selection of soils for this study was based on the applicability of whole soil speciation using WHAM VI (9) for Cu and Ni (3). The soils of the same origin for the Cu and Ni studies differ in their properties because they were sampled at different times and at different locations. A total of 11 and 8 noncalcareous soils for Cu and Ni toxicities, respectively, represented in this study are subsets of the soils used in the bioassay studies (7, 10, 11). Within the subsets, the soils which lacked sufficient data depending on a particular endpoint were also excluded from this study (Table A-1 in Supporting Information). Bioassays. Effect of Cu and Ni on Tomato Shoot Yield. Tomato (Lycopersicon esculentum cv. Moneymaker) was used in the shoot yield study which was based on the ISO-11269-2 method (12), see Rooney et al. (10, 11). The tomato shoot yield (TSY) at each dose is expressed as a percent of the control for each soil. Effect of Cu and Ni on Reproduction of Redworms. Chronic toxicity assays with redworms (Eisenia fetida) were performed following the ISO 11268-2 method (13). After one week of equilibration, 10 adult individuals of E. fetida with a fully developed clitellum were added per test vessel containing 600 g wet weight of soil. At the start of the test, 10 g of finely ground cow dung was supplied. During exposure, all test 10.1021/es061173c CCC: $33.50

 2006 American Chemical Society Published on Web 10/20/2006

vessels were kept at 20 ( 1 °C and a light/dark cycle of 16:8 at 400-800 lux. In all tests, four replicates were tested per concentration. After 4 weeks, the number of cocoons was determined after washing the soil through a 1 mm sieve. The E. fetida cocoon production (ECP) was expressed as a percent of the control. Effect of Cu and Ni on Reproduction of Springtails. Chronic toxicity tests with the springtails (Folsomia candida) were done following the ISO 11267 method (14). One week after spiking, 10 springtails (10-12 days old) were exposed per glass vessel containing 30 g of moist soil. Granulated dry yeast was added weekly on the soil surface as food. At the end of 4 weeks, juveniles were counted after extracting them by flotation. Conditions during exposure were same as the reproduction assay with E. fetida. The F. candida juvenile production (FJP) was expressed as a percent of the control. Effect of Cu and Ni on Glucose-Induced Respiration (GIR) by Soil Microbes. Seven days after spiking, subsamples of each soil at each metal dose were put into separate glass jars, amended with 14C labeled glucose solution, mixed thoroughly, and immediately placed and sealed inside a preserving pot containing 1.0 M NaOH. Each sample was then incubated in the dark at 20 °C for 24 h, after which 1 mL of the NaOH was removed and added to a scintillation cocktail (XT Gold) for activity determination by beta counting. The extent of the added glucose-C respired is calculated from the ratio of the radioactivities. The GIR at each dose is then expressed as the percent of the control, see Oorts et al. (7). Effect of Cu and Ni on the Potential Nitrification Rate (PNR) in Soils. Seven days after spiking, subsamples of each soil were amended with (NH4)2SO4 solution. The soil nitrate was measured colorimetrically in a centrifuged soil extract (1 M KCl). The PNR (mg NO3-N kg-1 fresh soil d-1) was calculated as the slope of the linear regression of soil nitrate concentration against time after substrate addition. The PNR at each dose is then expressed as a percent of the control, see Oorts et al. (7). Extraction of Soil Solutions and Determination of their Properties. After equilibrating for 9 days for TSY (10, 11) and 4 weeks for the rest of the bioassays, soil solutions were separated from the spiked subsamples by centrifugation and immediately filtered through a 0.45 µm filter. The pH, dissolved metal concentrations, and the dissolved organic carbon (DOC) concentrations were then determined. See Rooney et al. (10, 11) and Oorts et al. (7) for details. Estimating the EC50 Values for Individual Soils. The EC50 values were estimated by fitting the log-logistic model in the Toxicity Relationship Analysis Program (TRAP), version 1.0 (15) by using either soil metal (Cu or Ni) concentrations (background + added in mg kg-1 soil) as the dose. The soil solution properties at EC50 for each soil were interpolated from the measured values (Table A-2-A-11 in Supporting Information). The speciation of the cations at EC50 was then computed by WHAM VI using the approach described below. Whole Soil Metal Speciation using WHAM VI. WHAM VI with its default parameters was used to compute the cation activities in the soil solutions. The details of this procedure can be found in Thakali et al. (3). The bulk soil properties required are Cu or Ni (background + added in mg kg-1 of soil) at each dose and the soil organic matter (SOM) content. The SOM was assumed to consist of particulate humic acid (HA) and fulvic acid (HA) in the ratio 84:16. A temperature of 293 K and a CO2 partial pressure of 10-3.5 atm were used, consistent with the experimental conditions. The measured solution pH, dissolved Na, Mg, K, and Ca concentration at each dose were also used as inputs. It was assumed that the soil systems were in equilibrium with amorphous Fe(OH)3 and Al(OH)3 with solubility constants, log *KFe(OH)3 ) 3.0 and log *KAl(OH)3 ) 6.0 (only for Ni speciation) for the respective reactions: Fe(OH)3 + 3 H+ ) Fe3+ + 3 H2O and Al(OH)3 +

3 H+ ) Al3+ + 3 H2O. Additionally, presence of dissolved Cland SO42- concentrations in the ratio 3:1 (molc:molc) was assumed to maintain electroneutrality. Estimation of Model Parameters. The toxicity model underlying the TBLM is presented in Thakali et al. (3). Equation 1 is the mathematical basis for the TBLM R) 1+

(

100 KMBL{M2+}

)

β

+

f50(1 + KMBL{M } + KHBL{H } + KCaBL{Ca } + KMgBL{Mg }) 2+

2+

2+

(1)

where R ) biological response (as % of control), f50 is the fraction of the total biotic ligand (BL) sites bound by the metal (M2+ representing either Cu2+ or Ni2+ in this study) that results in a 50% effect level, β is the dose-response shape factor, Ks are the conditional binding constants of the respective cation binding to the BL sites, and the symbol {} represents the activities. The TBLM constants (β, f50, and Ks) in this study are estimated by relating the response (R) to the cation activities computed using WHAM VI, which incorporates the soil and soil solution interactions. Two additional models were also fitted to the same dataset for comparison to the TBLM. The total metal model (TMM) uses the soil metal concentration (background + added) as the dose (eq 2). The free ion activity model (FIAM) uses the computed M2+ activities as the dose (eq 3). The fitting parameters are EC50 (mg Ni kg-1) and β for the TMM and EC50 (moles L-1) and β for the FIAM.

R)

100 Total Metal 1+ EC50

R)

(

(2)

β

)

100 {M2+} 1+ EC50

(

)

(3)

β

The model parameters were estimated by minimizing the root-mean-squared error (RMSE) of the predicted response (as % of control) using the SOLVER program in Microsoft Excel 2003 and the standard errors were estimated using SOLVER AID (16), see Thakali et al. (3).

Results and Discussion Estimation of Model Parameters. The fitted model parameters for Cu and Ni toxicity models are given in Tables 1 and 2, respectively. The estimation procedure, which follows Thakali et al. (3), is elaborated in the Supporting Information. The large standard errors for log KMgBL and log KHBL in Cu toxicity to PNR (Table 1) and in Ni toxicity to PNR (Table 2), respectively, indicate that the additional improvement of the model fit by incorporating the competitive effect the respective cations is too small for a reasonably accurate estimation of these parameters. Dose-Response Relationships. The goodness of fit of the three models is compared based on the RMSE of the predicted response and the R2 values for the linear regression between the predicted and the observed responses (Tables 1 and 2). The dose-response relationships and the predicted vs measured responses are also presented in Figures A1A10 (Supporting Information). Based on the RMSE and R2 values (Tables 1 and 2), the FIAM generally fits the data better than the TMM. Results showing M2+ activities having better correlation than soil metal concentrations with metal toxicities have been reported for Cu and Pb toxicities to crop yield and microbial activities (17) and for Zn, and Cu toxicities to a bacterial biosensor (18, 19). The uptake of metals by soil invertebrates is also through VOL. 40, NO. 22, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

7095

TABLE 1. Dose-Response Parameters (EC50 and β) for the Three Models Considered: Total Cu Model (Total Cu), the Free Ion Activity Model (FIAM) and the Terrestrial Biotic Ligand Model (TBLM)a dose response parameters biological endpoint barley root elongation

model

(BRE)b

total Cu FIAM TBLM tomato shoot yield (TSY) total Cu FIAM TBLM F. candida juvenile production (FJP) total Cu FIAM TBLM E. fetida cocoon production(ECP) total Cu FIAM TBLM glucose induced respiration (GIR) total Cu FIAM TBLM potential nitrification rate (PNR) total Cu FIAM TBLM

EC50

model fit

156.8 ( 16.7 mg Cu (1.50 ( 0.44) ×10-7 M 0.05c 219.3 ( 34.2 mg Cu kg-1 (2.50 ( 0.58) ×10-7 M 0.05c 209.9 ( 40.0 mg Cu kg-1 (1.61 ( 0.33) ×10-6 M 0.05c 184.4 ( 16.5 mg Cu kg-1 (8.30 ( 4.62) ×10-7 M 0.05c 977.3 ( 182.5 mg Cu kg-1 (5.94 ( 2.32) ×10-5 M 0.05c 314.6 ( 54.7 mg Cu kg-1 (1.98 ( 0.72) ×10-6 M 0.05c kg-1

conditional binding constants (log)

RMSEd R 2 e

β 1.61( 0.26 0.54 ( 0.08 0.96 ( 0.11 1.38 ( 0.29 0.76 ( 0.12 1.11( 0.16 0.83 ( 0.16 0.57 ( 0.06 0.70 ( 0.08 1.80 ( 0.31 0.39 ( 0.09 1.14 ( 0.15 0.79 ( 0.13 0.36 ( 0.05 0.58 ( 0.07 1.10 ( 0.23 0.52 ( 0.09 0.78 ( 0.13

23.4 20.2 15.4 30.1 18.0 16.5 24.2 14.5 14.9 18.0 27.5 14.7 21.9 18.9 16.7 29.3 25.2 22.0

0.66 0.75 0.85 0.53 0.83 0.86 0.48 0.82 0.82 0.77 0.47 0.85 0.52 0.65 0.72 0.47 0.61 0.70

KCuBL

KHBL

KMgBL

7.41( 0.23 6.48 ( 0.26 5.65 ( 0.10 4.38 ( 0.21 4.62 ( 0.12 2.97( 0.62 6.50 ( 0.25 5.90 ( 0.29 6.69 ( 0.10

7.50c

4.93 ( 0.48 4.45 ( 0.58 1.64 ( 5.80

a

The conditional binding constants are associated with the TBLM. The associated standard errors ((values) were estimated using Solver Aid (16). b From Thakali et al. (3). c Fixed values (see Supporting Information). d The root-mean-squared error (RMSE) is associated with the predicted responses. e R2 is the coefficient of determination for the linear regression between the predicted and observed responses.

TABLE 2. Dose-Response Parameters (EC50 and β) for the Three Models Considered: Total Ni Model (Total Ni), the Free Ion Activity Model (FIAM) and the Terrestrial Biotic Ligand Model (TBLM)a dose response parameters biological endpoints barley root elongation (BRE)b

tomato shoot yield (TSY)

F. Candida juvenile production (FJP) E. Fetida cocoon production (ECP) glucose induced respiration (GIR)

potential nitrification rate (PNR)

models total Ni

306.3 ( 61.1 mg Ni (1.44 (

kg-1

model fit β

RMSEd

R2e

1.18 ( 0.28

28.3

0.53

EC50 or f50

conditional binding constants (log)

KNiBL

KHBL

KMgBL

KCaBL

FIAM TBLM total Ni

M 0.05c 103.9 ( 25.2 mg Ni kg-1

1.50 ( 0.23 2.37 ( 0.33 1.14 ( 0.33

13.6 10.8 33.5

0.89 0.93 0.50

3.60 ( 0.53

4.53 ( 0.62

3.81 ( 0.60

1.50c

FIAM TBLM total Ni

(3.05 ( 0.53)×10-5 M 0.05c 671.4 ( 213.9 mg Ni kg-1

1.37 ( 0.29 2.68 ( 0.54 0.71 ( 0.17

20.2 15.9 23.7

0.82 0.89 0.45

6.05 ( 0.15

6.52 ( 0.18

5.23 ( 0.47

5.00c

FIAM TBLM total Ni

(3.10 ( 0.54)×10-4 M 0.05c 491.8 ( 128.2 mg Ni kg-1

1.02 ( 0.18 1.52 ( 0.26 0.92 ( 0.23

17.3 16.8 25.3

0.71 0.73 0.41

5.12 ( 0.06

6.02 ( 0.22

5.00c

FIAM TBLM total Ni

(2.32 ( 0.27)×10-4 M 0.05c 736.9 ( 456.9 mg Ni kg-1

1.48 ( 0.24 2.18 ( 0.35 0.41 ( 0.15

15.3 14.9 31.2

0.81 0.83 0.19

5.33 ( 0.05

6.70 ( 0.12

5.00c

FIAM TBLM total Ni

(5.28 ( 1.27)×10-4 M 0.05c 135.6 ( 21.4 mg Ni kg-1

0.67 ( 0.11 0.99 ( 0.15 1.34 ( 0.29

19.6 17.1 23.3

0.68 0.76 0.68

4.53 ( 0.10

6.09( 0.21

5.00c

(6.44 ( 1.06)×10-5 M 0.05c

1.05 ( 0.17 1.63 ( 0.22

16.5 13.0

0.84 0.90

5.72 ( 0.06

5.10 ( 4.05

5.00c

FIAM TBLM

0.15)×10-4

a

The conditional binding constants are associated with the TBLM. The associated standard errors ((values) were estimated using Solver Aid (16). b From Thakali et al. (3). c Fixed values (see Supporting Information). d The root-mean-squared error (RMSE) is associated with the predicted responses. d R2 is the coefficient of determination for the linear regression between the predicted and observed responses.

the soil pore water (20-22). These findings indicate that the M2+ in the pore waters represents its bioavailability much better than the soil metal concentration. However, some studies have shown a level of toxicity (e.g., EC50) to have more intersoil variations for Cu toxicity to barley root elongation (23), and for Cu and Ni toxicities to PNR and GIR (7) when expressed in terms of M2+ activities in soil solutions than when expressed as soil metal concentrations. This suggests that the FIAM model may not provide the advantage over the TMM that is indicated by the RMSE and R2 measures since the EC50s based on TMM have less variability. However, the best test of a model’s predictive ability is how well it predicts EC50s. This is discussed subsequently in the section on “Prediction of EC50.” 7096

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 22, 2006

By accounting for the competitive binding of cations to the BL sites, the TBLM improves the data fit. The RMSE values are consistently the smallest and the R2 values are the largest for the TBLM for all six endpoints for both Cu and Ni toxicities. In addition, the intercepts (ideally ) 0) and the slopes (ideally ) 1) for the linear regressions between the predicted vs the observed responses (Figures A-1 to A-10 in Supporting Information) indicate improvements by the TBLM over the FIAM, which are similar to those found in Thakali et al. (3) for barley root elongation. These results clearly indicate that the TBLM provides a better approach to modeling Cu and Ni toxicities in soils than the TMM and the FIAM for the six biological endpoints considered in this study.

Interaction of Cations with the Biotic Ligand. The inclusion of proton competition in Cu toxicity improved the data fit remarkably, especially in the case of Cu toxicity to ECP, where the FIAM actually achieved a worse data fit than the TMM (Table 1). Further inclusion of either Ca2+ or Mg2+, however, did not improve the data fit appreciably (Table 1). This indicates that the competition due to Ca2+ and Mg2+, if any, is insignificant compared to that of the protons for the conditions of these experiments. The relative affinity of the BL sites for the cations, Cu2+ > H+ . Ca2+ and/or Mg2+ in decreasing order, is similar to that in Cu toxicity to plants (24, 25). Reported ameliorative effects of protons in Cu toxicity to soil invertebrates (4, 5) are consistent with the findings of this study. For acute Cu toxicity to the earthworm A. caliginosa, Steenbergen et al. (4) observed significant competition of protons and Na+ but not for Ca2+ and Mg2+. For the chronic Cu toxicity in this study Na+ competition was not apparent. The presence of proton competition observed for the microbial activities in this study is also consistent with the significant correlations observed between critical metal concentrations and the pH (7, 8). The interaction of the cations in Ni toxicity is more complex than in Cu toxicity as indicated by the significant values of Ca and Mg conditional binding constants in Table 2, but not in Table 1. Both Ca2+ and Mg2+ competitions, in addition to that of the proton, found in TSY are similar to the case of Ni toxicity to BRE (3). Only proton and Ca2+ competitions are observed for FJP and PNR, and only proton and Mg2+ competitions are observed for ECP and GIR (Table 2). The relative affinity for cation binding by the BL sites associated with TSY follows the order H+ > Ni2+ > Mg2+ > Ca2+. For ECP and FJP, the order is H+ > Ni2+ > Ca2+ > Mg2+ and for PNR and GIR there is no apparent order. The competitive binding of cations to plasma membranes of plant roots (26) and bacterial cell walls (27, 28) have been studied extensively and modeled. The TBLM approach may be extended to these studies by treating metal accumulation as the endpoint. The advantage of the TBLM over other cation complexation models is its ability to link the toxicity, pore water chemistry, and soil chemistry without having to identify the nature of the BL sites and quantify their density. The identification of the BL associated with the biological endpoints of our study and the mechanisms of the toxicities will no doubt validate the TBLM approach. However, we refrain from speculation since any direct measurements are not available. We simply rely on the evidence for BL-type interactions reported in the literature for similar endpoints (4-8). Extent of the Protective Effect of the Competing Cations. An analysis based on the binding constants (Tables 1 and 2) can illustrate the importance of competitive binding of each of these cations in Cu and Ni toxicity to the six endpoints. The EC50 based on M2+ activities can be determined by solving eq 1 with R )

{M2+}50 )

f50 (1 - f50)KMBL

(1 + KHBL{H+}50 +

KCaBL{Ca2+}50 + KMgBL{Mg2+}50) (4) Equation 4 can also be expressed as

{M2+}50 ) R50(1 + 10(log KHBL-pH50) + 10(log KCaBL-pCA50) + 10(log KMgBL-pMg50)) (5) where R50 is the theoretical EC50 M2+ activity when there is no cation competition (eq 4) and “p” denotes the negative of the base 10 logarithm (e.g., pCa50 ) -log {Ca2+}50). Equation 5 indicates that the competition due to a cation becomes

FIGURE 1. Comparison of the binding constants for (a) Cu toxicity and (b) Ni toxicity.

FIGURE 2. Protective effects of the competitive cations in (a) Cu toxicity and (b) Ni toxicity. The legend “None” represents the theoretical EC50 when none of the competitive cations are present. The reduction in Cu toxicity is computed at pH 6.5, 5.5, 4.5, and at 4.5 with Ca ()2.87 mmol L-1) and Mg ()0.73 mmol L-1) indicated by the legend “pH 4.5*” in (a). The reduction in Ni toxicity (b) is computed at pH 5.5 (“only H”), with only Ca ()2.87 mmol L-1) (“Only Ca”), with only Mg ()0.73 mmol L-1) (“Only Mg”) and with Ca ()2.87 mmol L-1), Mg ()0.73 mmol L-1) at pH 5.5 (“All”). effective only when pX50 < log KXBL (where X is H, Ca, or Mg). The EC50 ({M2+}50) is doubled, i.e., the toxicity is halved, when the log KXBL ) pX50 and then increases in multiples of approximately 10, i.e., the toxicity is reduced by 10 fold, with each unit decrease in pX50. Consequently, the higher the log KXBL, the higher the pX50 (i.e., the lower its activity) that results in significant competitive effect. Additionally, it is also clear from eq 5 the stronger the metal-BL interaction, i.e., larger the KMBL, the lower the value of R50, i.e., it is more toxic. The log KHBL values are generally greater than 4 (Figure 1a) in Cu toxicity, except for FJP. Therefore, the proton competition is likely to be significant in Cu toxicity for the rest of the endpoints at ambient pH conditions. In the case of Ni toxicity, the log KHBL values are greater than 5 in all cases except for BRE (Figure 1b). Therefore, the proton competition is expected to be significant in Ni toxicity to these endpoints as well. The values of log KCaBL or log KMgBL, where present, are also significant (Figure 1b). Consequently, Ca2+ and Mg2+ will also reduce Ni toxicity significantly. The extent of the competitive effects of the cations to Cu and Ni toxicities in an ambient condition is illustrated in Figure 2. The median values of the pH (5.5) and the dissolved Ca and Mg concentrations (2.87 and 0.73 mmol L-1, respectively) from 75 soils with pH < 7 were used. These 75 soil pore water data include the control data (i.e., no metal added) from this study and those from Wolt (29) and Peijnenburg et al. (30). Figure 2a illustrates the effect of proton competition in alleviating Cu toxicity. The first bar represent the theoretical EC50 pM (-log of R50 in eq 5) with no cation competition. For each of the endpoints, except for FJP, decreasing the pH (i.e., increasing the proton activity) reduces the EC50 pCu (i.e., lowers the toxicity). The reduction in pCu is pronounced in the case of BRE, ECP, and GIR and is more pronounced when the pH is decreased from 5.5 to 4.5 than when the pH is reduced from 6.5 to 5.5. This can be explained by the magnitude of the log KHBL values for each of these endpoints. VOL. 40, NO. 22, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

7097

FIGURE 3. Comparison of the predicted vs observed EC50 soil metal concentrations for Cu toxicities to the six endpoints. The predictions are based on (a) the TMM(b) the FIAM, and (c) the TBLM. The solid lines represent the 1:1 ratio and the dotted lines represent a factor of 2 variation above and below the 1:1 lines.

FIGURE 4. Comparison of the predicted vs observed EC50 soil metal concentrations for Ni toxicities to the six endpoints. The predictions are based on (a) the TMM, (b) the FIAM, and (c) the TBLM. The solid lines represent the 1:1 ratio and the dotted lines represent a factor of 2 variation above and below the 1:1 lines. The unit reduction in pH corresponds to an increase in the H+ activity by multiples of approximately 10. Since it is the product of the KHBL and H+ activity that determines the extent of its ameliorative effect, this effect becomes increasingly pronounced at pH [dlt] log KHBL. Therefore, even at pH ) 4.5, the reduction in Cu toxicity is not pronounced for TSY, FJP, and PNR as log KHBL is