A Biotic Ligand Model Predicting Acute Copper ... - ACS Publications

Nov 30, 2001 - Following values were obtained: log KCuBL = 8.02, log KCuOHBL= 7.45, log KCaBL = 3.47, log KMgBL = 3.58, log KNaBL = 3.19, and log KHBL...
0 downloads 6 Views 91KB Size
Download PDF
Environ. Sci. Technol. 2002, 36, 48-54

A Biotic Ligand Model Predicting Acute Copper Toxicity for Daphnia magna: The Effects of Calcium, Magnesium, Sodium, Potassium, and pH KAREL A. C. DE SCHAMPHELAERE* AND COLIN R. JANSSEN Laboratory for Environmental Toxicology and Aquatic Ecology, Ghent University, J. Plateaustraat 22, B-9000 Ghent, Belgium

The extent to which Ca2+, Mg2+, Na+, K+ ions and pH independently mitigate acute copper toxicity for the cladoceran Daphnia magna was examined. Higher activities of Ca2+, Mg2+, and Na+ (but not K+) linearly increased the 48-h EC50 (as Cu2+ activity), supporting the concept of competitive binding of these ions and copper ions to toxic action or transport sites at the organism-water interface (e.g. fish gill, the biotic ligand). The increase of the EC50 (as Cu2+ activity) with increasing H+, however, seemed to suggest cotoxicity of CuOH+ rather than proton competition. Based on the biotic ligand model (BLM) concept, we developed a methodology to estimate stability constants for the binding of Cu2+, CuOH+, Ca2+, Mg2+, Na+, and H+ to the biotic ligand, solely based on toxicity data. Following values were obtained: log KCuBL ) 8.02, log KCuOHBL) 7.45, log KCaBL ) 3.47, log KMgBL ) 3.58, log KNaBL ) 3.19, and log KHBL ∼ 5.4. Further, we calculated that on average 39% of the biotic ligand sites need to be occupied by copper to induce a 50% acute effect for D. magna after 48 h of exposure. Using the estimated constants, a BLM was developed that can predict acute copper toxicity for D. magna as a function of water characteristics. The presented methodology can easily be applied for BLM development for other organisms and metals. After validation with laboratory and natural waters (including DOC), the developed model will support efforts to improve the ecological relevance of presently applied risk assessment procedures.

Introduction Current water quality standards and risk assessment procedures of metals are predominantly based on total or dissolved metal concentrations. However, there is extensive evidence that neither total nor dissolved aqueous metal concentrations are good predictors of metal bioavailability and toxicity. Copper toxicity to freshwater organisms has been shown to be dependent on a variety of ambient water quality characteristics. For example, increases in hardness (1-4), pH (2, 5, 6), and organic matter content (2, 7, 8) have shown to result in decreased copper toxicity for fish and * Corresponding author phone: 00 32 9 264 39 93; fax: 00 32 9 264 41 99; e-mail: [email protected]. 48

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 1, 2002

crustacean species. In most regulatory frameworks, however, most of these toxicity modifying factors are not taken into acount. For example, when deriving site-specific water quality criteria for metals, only a hardness correction is recommended to be carried out by some instances (e.g. U.S. Environmental Protection Agency) (9). A large number of studies explain part of the observed toxicity variability by attributing the toxic effect of copper to differences in the free metal ion activity in the test medium and thus support the Free Ion Activity Model (FIAM), originally formulated by Morel (10) and critically reviewed by Campbell (11). However, copper toxicity does not seem to be determined by the free copper ion activity alone. First, also other copper forms such as hydroxy complexes (12) and complexes with low molecular weight metabolites (11) have been shown to contribute to copper toxicity. Second, and probably more importantly, Ca2+, Mg2+, Na+, and H+ ions are hypothesized to compete with copper ions for binding sites at the organism-water interface (e.g. fish gill) (1, 13-18). Their presence consequently results in a decreased toxicity of the free copper ion (17-20). From the above it is clear that next to copper speciation, which determines the free copper ion activity (12), interactions at the site of toxic action also need to be considered when evaluating bioavailability (18). The recently developed Biotic Ligand Model (BLM) includes both and is therefore gaining increased interest in the scientific and regulatory community (18, 19). The main assumption of the BLM is that metal toxicity occurs as the result of free metal ions reacting with binding sites at the organism-water interface (either physiologically active sites, leading to a direct biological response, or transport sites, leading to metal transport into the cell followed by a subsequent, indirect biological response), which is represented as the formation of a metalbiotic ligand complex. The concentration of this metal-biotic ligand complex directly determines the magnitude of the toxic effect, independent of the physical-chemical water characteristics of the test medium. For fish, for example, the biotic ligand (BL) appears to be sites on the surface membrane of the gill (1, 18). In the BLM, cation-biotic ligand interactions are represented and calculated in the same way as any other reaction of a cation with an organic or inorganic ligand. To that end, the complexation capacity of the BL (analoguous to the concentration of other ligands in solution) and stability constants for the metal- and the cation-BL complexes are required and need to be incorporated in a speciation model like WHAM (21) (see Mathematical Description). Initially BLMs were developed to predict copper toxicity for fish (18). These models combined existing data sets on both gill-copper concentration measurements and stability constants (15) and toxicity (2, 22) and incorporated them in the metal speciation model CHESS (23), modified to include metal-DOC complexation according to WHAM (21). Unfortunately, extensive biotic ligand data sets are not available for other frequently used test organisms such as Daphnia. Therefore, until now, the BLMs developed for fish have been calibrated to fit Daphnia pulex toxicity data (17). This calibration was based on the “fish” stability constants and a modified critical biotic ligand copper concentration resulting in 50% effect. Although this “modified fish BLM” showed promise, two questions remain unanswered: (1) are stability constants for fish and Daphnia really similar and (2) can these constants be derived based on toxicity data alone, i.e., without measuring BL-concentrations. 10.1021/es000253s CCC: $22.00

 2002 American Chemical Society Published on Web 11/30/2001

TABLE 1. Overview of the Chemical Characteristics of the Test Media Used in the Different Bioassay Sets (See Experimental Section) and the Observed 48-h EC50s for Daphnia magna Expressed as Dissolved Copper (CuT) and Free Copper Ion Activity (Cu2+) bioassay set

pH

CO32(µM)

Ca2+ (mM)

Mg2+ (mM)

Na+ (mM)

K+ (mM)

Cl(mM)

SO42(mM)

EC50 nM (CuT)

EC50 nM (Cu2+)

Ca Ca Ca Ca Ca Ca Ca MgSO4 MgSO4 MgSO4 MgSO4 MgSO4 MgSO4 MgSO4 MgCl2 MgCl2 MgCl2 MgCl2 Na Na Na Na Na Na K K K K pH1 pH1 pH1 pH1 pH2 pH2 pH2 pH2 pH2 pH2

6.95 6.95 6.92 6.90 6.94 6.86 6.91 6.77 6.75 6.85 6.83 6.83 6.78 6.83 6.79 6.85 6.87 6.91 6.78 6.78 6.79 6.81 6.79 6.81 6.98 6.96 6.92 6.89 6.44 6.75 7.28 7.91 5.98 6.30 6.70 6.97 7.34 7.92

91.3 89.0 83.4 91.4 89.5 93.4 93.2 104 108 96.3 92.9 97.5 93.1 97.5 87.7 85.3 95.1 92.6 95.6 93.1 95.0 91.3 97.5 93.8 97.7 94.2 85.1 84.3 64.0 97.5 237 1230 56.9 75.3 99.8 176 399 1740

0.25 0.50 0.75 1.0 2.0 3.0 4.0 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25

0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.50 1.0 1.5 2.0 3.0 5.0 0.25 1.0 2.0 4.0 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25

0.077 0.077 0.077 0.077 0.077 0.077 0.077 0.077 0.077 0.077 0.077 0.077 0.077 0.077 0.077 0.077 0.077 0.077 1.08 3.09 5.09 7.60 10.1 15.1 0.077 0.077 0.077 0.077 2.16 2.16 2.16 2.14 1.58 1.58 1.58 1.58 1.58 1.54

0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.500 1.000 2.000 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078 0.078

0.58 1.08 1.58 2.08 4.08 6.08 8.08 0.58 0.58 0.58 0.58 0.58 0.58 0.58 1.08 2.58 4.58 8.58 1.58 3.59 5.59 8.10 10.6 15.6 0.58 1.08 1.58 2.58 2.70 2.66 2.49 1.56 2.12 2.12 2.08 2.00 1.78 0.58

0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.50 1.0 1.5 2.0 3.0 5.0 0.0 0.0 0.0 0.0 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25

81.6 116 113 129 138 172 191 57.6 106 76.5 173 149 221 275 107 174 200 230 91.9 139 142 201 241 333 97.1 108 126 109 106 139 175 671 87.8 104 102 159 221 824

8.60 18.5 19.0 25.3 28.3 43.7 48.7 6.08 19.7 10.2 41.8 32.4 58.3 70.0 18.7 45.4 55.7 64.8 14.3 30.3 30.9 52.4 68.2 90.0 10.9 11.5 14.4 12.4 27.4 31.2 22.1 18.4 32.8 30.5 24.5 29.8 25.6 16.9

Therefore the aims of the present study were 2-fold: (1) to investigate the extent to which calcium, magnesium, sodium, potassium, and hydrogen ions can individually mitigate copper ion toxicity and (2) to use the obtained toxicity data to derive estimates of the parameters necessary to develop a BLM that can predict copper toxicity toward D. magna for a broad range of water quality characteristics. The development (this study) and validation (through future experiments with DOC additions and with natural waters) of such a semimechanistic model could support efforts to improve the ecological relevance of presently applied risk assessment procedures.

Experimental Section Experimental Design. To distinguish between the independent effect of different cations on copper toxicity, one cation concentration at a time was varied, while keeping all other cation concentrations as low and as constant as possible. Seven sets of copper bioassays were performed: 1 Ca-set, 2 Mg-sets, 1 Na-set, 1 K-set, and 2 pH-sets (Table 1). All sets comprised at least four bioassays (four different cation concentrations) that were conducted at the same time to minimize variability. The selected cation concentrations reflected the variability occurring in natural surface waters. Preparation of the Test Solutions. All chemicals were purchased from Vel (Leuven, Belgium) and were reagent grade. All test media were prepared by adding different

volumes of stock solutions of CaCl2, MgCl2 or MgSO4, NaCl, KCl, and NaHCO3 to carbon-filtered and deionized water. Except for the pH-sets, these media were adjusted to pH 6.8 by adding 0.077 mM NaHCO3. pH in the pH-sets was controlled by NaHCO3-buffering. Sodium concentrations were kept constant through additions of NaCl. For each bioassay, the prepared test medium was then used as the dilution water to make a concentration series of copper, added as CuCl2. To obtain near-equilibrium situations, all media were stored in the test cups at 20 °C for 1 day before being used in the toxicity tests. The chemical characteristics of the different testmedia are summarized in Table 1. Acute Toxicity Tests. The acute 48-h immobilization assay with juvenile D. magna (2 mM are high compared to those found in natural waters (2). Effect of pH on Copper Toxicity. In the tested pH range of 5.98-7.92, observed 48-h EC50s ranged from about 88 to 820 nM for dissolved copper and from 17 to 33 nM for free copper ion activity. This means that observed differences in (dissolved) EC50s when pH is varied can at least partially be explained by speciation differences. If the relation between (H+) and 48-h EC50(Cu2+) is considered as being linear, this would indicate the possibility of proton competition at the biotic ligand (Figure 1e). However, it may be suggested that a curvilinear relation with saturating EC50 at higher (H+) would also fit the data. Two, possibly interacting, mechanisms could explain this observation. First, the pH in the microenvironment (of the biotic ligand) of fish (14) and invertebrates (36) can be different from the pH in the bulk solution, leading to a misinterpretation of the interactions of copper and hydrogen ions at the organism-water interface. Playle et al. (14), for example, showed that the pH at the gill of fathead minnows was rendered more basic at acidic pH and more acidic at basic pH (through the release of CO2), thus regulating the pH at the gill within a narrow range. Second, possible toxicity of copper hydroxide complexes (5, 12) would imply that at a higher pH (lower H+), less (Cu2+) would be needed to exert the same toxic effect, leading to a steepening of the slope in this pH range. Within the BLM construct, the possible contribution of copper hydroxide to toxicity can best be accommodated in terms of the formation of a copper hydroxide-biotic ligand complex, [CuOHBL]. In this case, toxicity will be determined by the total amount of copper bound to the BL, being the sum of [CuBL] and [CuOHBL]. Equations 4 and 5 can be transformed to

fCuBL ) {KCuBL‚(Cu2+) + KCuOHBL‚(CuOH+)}/ {1 + KCuBL‚(Cu2+) + KCuOHBL‚(CuOH+) + KCaBL‚(Ca2+) + KMgBL‚(Mg2+) + KNaBL‚(Na+) + KHBL‚(H+)} (6) and, given the known stability constant for the CuOH complex (KCuOH, see Table 1), to 50% EC50Cu2+ ) [f CuBL ‚{1 + KCaBL‚(Ca2+) + KMgBL‚(Mg2+) +

KNaBL‚(Na+) + KHBL‚(H+)}]/ 50% [(1 - f CuBL )‚{KCuBL + KCuOHBL‚KCuOH‚(OH-)}] (7)

From eq 7 it can be concluded that, if CuOH+ contributes to copper toxicity, a linear relation should exist between (OH-) and 1/EC50(Cu2+) in the pH range where there is no significant proton competition and provided that other cation activities are constant. Furthermore, the ratio KCuOHBL/KCuBL should be equal to the ratio slope/intercept of that linear regression, divided by KCuOH. The fact that the mentioned linear relation 52

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 1, 2002

was indeed observed (Figure 1f) and that the R 2 (0.86) was higher than the R 2 for the linear regression between (H+) and (Cu2+) (Figure 1e, R 2 ) 0.45) suggests that it is probably better to model the observed toxicity changes in the tested pH range in terms of copper hydroxide cotoxicity than in terms of proton competition (cf. below). KCuOHBL/KCuBL equaled 0.27, which, from a mechanistic point of view, means that CuOH+ is less toxic than Cu2+. Estimation of BLM Parameters. Linear regression analysis of EC50(Cu2+) versus cation activity (e.g. Ca2+) gives an intercept and a slope, which can, according to eq 5, be expressed as (ignoring the small differences between CuOH+ contribution to toxicity caused by small pH changes at pH ∼ 6.8)

InterceptCa )

50% f CuBL 50% (1 - f CuBL )‚KCuBL

‚

{1 + KMgBL‚(Mg2+)Ca + KNaBL‚(Na+)Ca + KHBL‚(H+)Ca} (8) and

SlopeCa )

50% f CuBL 50% (1 - f CuBL )‚KCuBL

‚KCaBL

(9)

where (Mg2+)Ca, (Na+)Ca, and (H+)Ca ) the mean of the calculated cation activities in all tests of the bioassay set under consideration (in this case the Ca-set). Dividing slope by intercept gives the following ratio (RCa):

SlopeCa ) InterceptCa KCaBL {1 + KMgBL‚(Mg2+)Ca + KNaBL‚(Na+)Ca + KHBL(H+)Ca}

) RCa (10)

Similar equations can be derived for the Mg, Na, and H-sets. Rearranging these equations into a matrix form, results in

(

-RCa‚(Mg2+)Ca -RCa‚(Na+)Ca

1 2+

-RMg‚(Ca )Mg 1

-RMg‚(Na )Mg -RMg‚(H+)Mg

-RNa‚(Ca2+)Na -RNa‚(Mg2+)Na 1 -RH‚(Ca2+)H

-RH‚(Mg2+)H

-RCa‚(H+)Ca

+

-RNa‚(H+)Na

( )( )

-RH‚(Na+)H

1

KCaBL RCa KMgBL RMg ) KNaBL RNa KHBL RH

)

‚

(11)

The solution of this matrix after speciation calculations and linear regression analyses (Figure 1) resulted in the estimation of the stability constants KCaBL, KMgBL, KNaBL, and KHBL (Table 2). The stability constants were determined by assuming the relation between pH and EC50Cu2+ (Figures 1e,f) to be (1) 100% due to proton competition and (2) 100% due to copper hydroxide toxicity (by excluding the last row out of matrix eq 11). Assuming proton competition, calculations resulted in log KCaBL ) 3.62, log KMgBL) 3.75, log KNaBL ) 3.37, and log KHBL ) 6.52. Although stability constants for Ca2+ and Na+ were more or less similar to those obtained for fish gills (15) and used in the current BLM (17, 18), the constant for proton competition was more than one log-unit higher than the frequently reported cell/membrane/gill surface pKa range from 4 to 5.4, as determined by surface binding analysis or estimated from low pH toxicity test results (15, 37). This large difference again demonstrates that

the relation between (H+) and EC50Cu2+ should rather be explained in terms of the earlier discussed toxicity of CuOH+ than in terms of proton competition. Indeed, cotoxicity of CuOH+ results in the steeper slope at higher pH (lower H+) and, consequently, to the high (calculated) KHBL. Therefore, it is probably better to assume log KHBL ∼ 5.4 (same as for fish gills). This asumption is in accordance with not allowing proton competition in the calculations, since at the lowest tested pH (5.97) proton competition would not be significant anyhow. In this case, calculations resulted in log KCaBL ) 3.47, log KMgBL ) 3.58, and log KNaBL ) 3.19 (Table 2). Again log K’s were in the same range as the constants reported by Playle et al. (15), except for the Mg-constant, which has not been included in current BLMs so far (17, 18). Compared to the values of log KCaBL and log KMgBL, the apparent value for log KNaBL seems unrealistically high, given the relative affinities of Na+, Ca2+, and Mg2+ for typical biogenic chelating ligands (e.g. aspartate, citrate, etc. (27)). This discrepancy suggests that the toxicity protection provided by Na+ is attributable not only to its competition with Cu2+ for the biotic ligand but also possibly to a direct physiological effect. It has been reported that increasing (Na+) could reduce the loss of plasma electrolytes (including Na+), which may be one of the copper toxicity mechanisms in fish (38). Although the effect of Na on copper toxicity might possibly not be a “chemical” effect (competition), it did not significantly influence the calculated values of the (“chemical” competition) stability constants. This is demonstrated by the fact that, when calculated without taking the Na-data set into account (by eliminating the third row from matrix eq 11), values of log KCaBL and log KMgBL (3.43 and 3.53, respectively) were almost identical to the values calculated with the Na-data set included (3.47 and 3.57, respectively). Now, for the final development of a BLM for D. magna, only two parameters remain to be determined: KCuBL and 50% f CuBL . Since these parameters are coupled through eqs 7, 8, and 9, they cannot be derived unambiguously from EC50 data alone. In practice, however, only the ratio

[EC50(Cu2+)]0 )

50% f CuBL 50% (1 - f CuBL )‚KCuBL

is needed for the prediction of EC50(Cu2+) with eq 7 (since KCuOHBL/KCuBL ) 0.27, cf. above). Mechanistically, [EC50(Cu2+)]0 can be interpreted as the EC50(Cu2+) in the hypothetical case where no competing ions are present in the test medium and where contribution of CuOH+ to toxicity is unsignificant. Based on observed EC50(Cu2+) and ion activities for each bioassay and using the stability constants for Ca, Mg, and Na and using KCuOHBL/KCUBL ) 0.27, (cf. above), [EC50(Cu2+)]0 was calculated with (7) to have a mean of 6.1 nM with a standard deviation of 1.8 nM (n ) 38). This value is lower than the observed intercepts (eq 8, Figure 1a-c) because there was always a small amount of ions present in both Ca, Mg, and Na bioassay sets. This value is also lower than the value calculated for P. promelas with eq 12 (19.9 nM, Table 2), which may reflect the higher copper sensitivity of D. magna to copper. 50% Nevertheless, good approximations of KCuBL and f CuBL are important, since these provide more information on the mechanistic understanding of copper bioavailability and toxicity. Therefore, for every treatment (38 bioassays × 5 Cu-concentrations) the fraction of the BL occupied by copper was calculated with eq 6 for varying log KCuBL. A priori it was assumed that the best approximation of the “real” KCuBL would result in the best correlation between the calculated fCuBL and the logit of the percent immobilized D. magna after 48 h of exposure (Figure 2). Values for log KCuBL ) 8.02 (and thus

FIGURE 2. Relationship between the calculated fraction of the biotic ligand sites occupied by copper (fCuBL) and the logit of the observed percent Daphnia magna immobilized after 48 h of exposure. fCuBL was calculated for every treatment with eq 6 using the outputs from the speciation calculations with BLM version a008 (30) and using the stability constants obtained in this study (Table 2). 50% ) 0.39 resulted log KCuOHBL ) 7.45) and the associated f CuBL 50% 2 in the best fit (R ) 0.72) and were retained. While f CuBL for D. magna is more or less similar to that for fish, log KCuBL was found to be 0.6 log-units higher, which possibly reflects the higher copper sensivity of D. magna (cf. above). From the good agreement between BLM constants of Santore et al. (17) and those presented in this study (see Table 2), it can be suggestedsas done by Santore et al. (17)s that as a first approximation the daphnid biotic ligand has the same affinity for metals and cations as that of fish. This assumption would only lead to incorrect toxicity predictions at pH values where CuOH+ toxicity becomes important, since Santore et al. (17) did not take CuOH+ toxicity into acount. This may be illustrated as follows: when calculated (with eq 50% 6) with the constants derived in this study f CuBL for the bioassays of the pH set varied between 0.34 and 0.44 and between 0.09 and 0.17, when calculated with the model of Santore et al. (17). This means that in the model of Santore et al. (17) fCuBL is not a better mortality predictor than free copper ion activity (see Table 1) when pH is varied, since both show a difference of about a factor 2 between the lowest and the highest value at 50% effect. Validation and Applicability of the Developed BLM. 50% Finally, all stability constants and f CuBL (Table 2) were included in version a008 of the BLM computer program (30) to predict 48-h EC50s (as dissolved copper) for the water quality characteristics of the bioassays conducted in this study (Table 1). The predicted EC50s differed from the observed EC50s with a factor of less than 1.5 (Figure 3). It should however be recognized that the BLM-parameters were especially estimated to fit the observed EC50 data. Although the developed Daphnia BLM shows promise, further validation experiments with a broader range of exposure conditions and field collected water samples (including variations of DOC concentrations) are required. Indeed, in natural waters (Cu2+) (and thus copper bioavailability and toxicity) is to a large extent determined by complexation to natural organic matter. In the tests performed in this study, DOC concentrations were assumed to be constant (138 µg DOC L-1) and to have the same metal complexation properties as natural organic matter (fulvic acid, cf. above). To assess the effect of this assumption on the calculations of the BLM constants, we repeated all calculations for the upper and lower 95% confidence limit of the DOC concentration, i.e., 101 and 175 µg DOC L-1 and the following were obtained: log KCuBL ) 7.75 and 8.50, log KCuOHBL) 6.98 and 7.97, log KCaBL ) 3.34 and 3.82, log KMgBL ) 3.47 and 3.93, log KNaBL ) 3.04 and 3.53, 50% f CuBL ) 0.37 and 0.43, and [EC50(Cu2+)]0 ) 2.4 and 10.5 nM. It is clear that calculated constants are more (KCuBL, KCuOHBL, 50% and [EC50(Cu2+)]0) or less (KCaBL, KMgBL, KNaBL, and f CuBL )

VOL. 36, NO. 1, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

53

FIGURE 3. Relationship between observed and predicted 48-h EC50s (expressed as dissolved copper, both axes are on a logarythmic scale) for Daphnia magna indicating the predictive capacity of the BLM developed in this study. Predicted EC50s were calculated with BLM version a008 (30) using the stability constants obtained in this study (Table 2). The full line indicates a perfect match between observed and predicted EC50s, the dashed lines indicate ratios of 0.5 and 2 between observed and predicted EC50s. sensitive to the choice of the DOC-concentration, which controls the range of the calculated (Cu2+) values. This sensitivity demonstrates that further experimental work is needed to obtain more accurate values of BLM constants and a mechanistic understanding of the BLM for D. magna developed in this study. This work could include both the identification of the biotic ligand for D. magna, the measurement of BL-concentrations of copper at different exposure conditions, and the calibration of the model to toxicity data of bioassays performed with DOC additions in which EC50s are determined based on measured Cu2+ activities. The main conclusion from this study is that, with the presented methodology, BLM constants can indeed be estimated from toxicity data alone, provided that valid metal ion activities are used as a starting point for the calculations. Further BLM development for other organisms and metals, using the presented methodology, could support efforts to improve the ecological relevance of presently applied risk assessment procedures.

Acknowledgments Karel De Schamphelaere is the recipient of a Ph.D. grant obtained from the Flemisch Institute for the Promotion of Scientific and Technological Research in Industry (IWT). Additional support was provided by the International Copper Association and the European Copper Institute. The authors wish to thank Van Wichelen Tom for the metal analyses.

Literature Cited (1) Pagenkopf, G. K. Environ. Sci. Technol. 1983, 17, 342-347. (2) Erickson, R. J.; Benoit, D. A.; Mattson, V. R.; Nelson, H. P., Jr.; Leonard, E. N. Environ. Toxicol. Chem. 1996, 15, 181-193. (3) Barata, C.; Baird, D. J.; Markich, S. J. Aquat. Toxicol. 1998, 42, 115-137. (4) Welsh, P. G.; Lipton, J.; Chapman, G. A.; Podrabsky, T. L. Environ. Toxicol. Chem. 2000, 19, 1624-1631

54

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 1, 2002

(5) Meador, J. P. Aquat. Tox. 1991, 19, 13-32. (6) Schubauer-Berigan, M. K.; Dierkes, J. R.; Monson, P. D.; Ankley, G. T. Environ. Toxicol. Chem. 1993, 12, 1261-1266. (7) Winner, R. E. Water Res. 1985, 19, 449-455. (8) Ma, H.; Kim, S. D.; Cha, D. K.; Allen, H. E. Environ. Toxicol. Chem. 1999, 18, 828-832. (9) Ambient Water Quality Criteria for Copper - 1984. U.S. Environmental Protection Agency: Washington, DC, 1994; EPA/ 823/B-94/001. (10) Morel, F. M. M. Principles of Aquatic Chemistry; John Wiley and Sons: New York, 1983. (11) Campbell, P. G. C. In Metal Speciation and Bioavailability in Aquatic Systems; Tessier, A., Turner, D. R., Eds.; John Wiley & Sons: Chichester, 1995; pp 45-102. (12) Allen, H. E.; Hansen, D. J. Water Environ. Res. 1996, 68, 42-54. (13) Zitko, V.; Carson, W. G. Chemosphere 1976, 5, 299-303. (14) Playle, R. C.; Gensemer, R. W.; Dixon, G. D. Environ. Toxicol. Chem. 1992, 11, 381-391. (15) Playle, R. C.; Dixon, D. G.; Burnison, K. Can. J. Fish. Aquat. Sci. 1993, 50, 2678-2687. (16) Campbell, P. G. C.; Stokes, P. M. Can. J. Fish. Aquat. Sci. 1985, 42, 2034-2049. (17) Santore, R. C.; Di Toro, D. M.; Paquin, P. R. Environ. Toxicol. Chem. Accepted for publication. (18) Di Toro, D.; Allen, H.; Bergmann, H.; Meyer, J.; Paquin, P.; Santore, R. Environ. Toxicol. Chem. Accepted for publication. (19) Playle, R. C. Sci. Total Environ. 1998, 219, 147-163. (20) Meyer, J. S.; Santore, R. C.; Bobbit, J. P.; Debrey, L. D.; Boese, C. J.; Paquin, P. R.; Allen, H. E.; Bergmann, H. L.; Di Toro, D. Environ. Sci. Technol. 1999, 33, 913-916. (21) Tipping, E. Comput. Geosci. 1994, 20, 973-1023. (22) MacRae, R. K.; Smith, D. E.; Swoboda-Colberg, N.; Meyer, J. S.; Bergman, H. L. Environ. Toxicol. Chem. 1999, 18, 1180-1189. (23) Santore, R. C.; Driscoll, C. T. In Chemical Equilibrium and Reaction Models; Loeppert, R., Schwab, A. P., Goldberg, S., Eds.; American Society of Agronomy: Madison, WI; pp 357-375. (24) Guideline for Testing of Chemicals, No. 202; OECD: Paris, 1984. (25) Elendt, B. P.; Bias, W. R. Water Res. 1990, 24, 1157-1167. (26) Standard Methods for the Examination of Water and Wastewater, 20th ed.; Clesceri, L. S., Greenberg, A. E., Eaton, A. D., Eds.; American Public Health Association: Washington, DC, 1998. (27) Stumm, W.; Morgan, J. J. Aquatic Chemistry: Chemical equilibria and rates in natural waters; John Wiley & Sons: New York, 1996. (28) Martell, A. E.; Smith, R. M.; Motekaitis R. J. Critical Stability Constants of Metal Complexes Database, version 4.0. NIST Standard Reference Database 46; National Institute of Standards and Technology: Gaithersburg, MD, 1997. (29) Fish, W.; Morel, F. M. M. Can. J. Fish. Aquat. Sci. 1983, 40, 1270-1277. (30) Biotic Ligand Model (BLM) User’s Guide for Version a008; HydroQual: Mahwah, NJ, 1999. (31) Dwane, G. C.; Tipping, E. Environ. Int. 1998, 24, 609-616. (32) Hamilton, M. A.; Russo, R. C.; Thurston, R. V. Environ. Sci. Technol. 1977, 11, 714-719. (33) Meyer, J. S. Environ. Sci. Technol. 1999, 33, 908-912. (34) Markich, S. J.; Jeffree, R. A. Aquat. Toxicol. 1994, 29, 257-290. (35) Hille, B.; Woodhull, A. M.; Shapiro, B. I. Philos. Trans. R. Soc. London B 1975, 270, 257-290. (36) Gensemer, R. W.; Playle, R. C. Crit. Rev. Environ. Sci. Technol. 1995, 29, 315-450. (37) Cusimano, R. F.; Brakke, D. F.; Chapman, G. A. Can. J. Fish. Aquat. Sci. 1986, 43, 1497-1503. (38) Paquin, P. R. Current structure of the BLM and model assumptions - sodium balance model; Presented at the Biotic Ligand Model Colloquium, Washington, DC, January 2001, 28-30.

Received for review October 25, 2000. Revised manuscript received October 10, 2001. Accepted October 11, 2001. ES000253S