Ecotoxicology, 12, 427±437, 2003 # 2003 Kluwer Academic Publishers. Manufactured in The Netherlands.

Toxicity of Lithium to Three Freshwater Organisms and the Antagonistic Effect of Sodium LYNN ADAMS KSZOS,1,y JOHN J. BEAUCHAMP2 AND ARTHUR J. STEWART1 1 Environmental Sciences Division, Oak Ridge National Laboratory, P.O. Box 2008, Bethel Valley Road, Oak Ridge, TN 37831-6422, USA 2 Department of Mathematics, Lipscomb University, Nashville, TN 37204, USA Accepted January 8, 2003

Abstract. Lithium (Li) is the lightest metal and occurs primarily in stable minerals and salts. Concentrations of Li in surface water are typically <0.04 mg l 1 but can be elevated in contaminated streams. Because of the general lack of information concerning the toxicity of Li to common toxicity test organisms, we evaluated the toxicity of Li to Pimephales promelas (fathead minnow), Ceriodaphnia dubia, and a freshwater snail (Elimia clavaeformis). In the laboratory, the concentration of Li that inhibited P. promelas growth or C. dubia reproduction by 25% (IC25) was dependant upon the dilution water. In laboratory control water containing little sodium (2.8 mg l 1), the IC25s were 0.38 and 0.32 mg Li l 1 and in ambient stream water containing 17 mg Na l 1, the IC25s were 1.99 and 3.33, respectively. A Li concentration of 0.15 mg l 1 inhibited the feeding of E. clavaeformis in laboratory tests. Toxicity tests conducted to evaluate the effect of sodium on the toxicity of Li were conducted with fathead minnows and C. dubia. The presence of sodium greatly affected the toxicity of Li. Fathead minnows and Ceriodaphnia, for example, tolerated concentrations of Li as great as 6 mg l 1 when sufficient Na was present. The interaction of Li and Na on the reproduction of Ceriodaphnia was investigated in depth and can be described using an exponential model. The model predicts that C. dubia reproduction would not be affected when animals are exposed to combinations of lithium and sodium with a log ratio of mmol Na to mmol Li equal to at least 1.63. The results of this study indicate that for most natural waters, the presence of sodium is sufficient to prevent Li toxicity. However, in areas of historical disposal or heavy processing or use, an evaluation of Li from a water quality perspective would be warranted. Keywords: lithium; sodium; toxicity; Pimephales promelas; Ceriodaphnia dubia; Elimia clavaeformis Introduction Lithium (Li) is the lightest metal, in its elemental form, and is highly reactive as a pure element. Because of its reactivity, Li does not occur naturally as a pure elementÐit occurs instead in stable minerals and salts (Bleiwas and Coffman, 1986). A review y

To whom correspondence should be addressed: Tel.: (865) 5744784; Fax: (865) 576-8143; E-mail: [email protected]

of the Li resources, distribution and toxicity to aquatic biota is provided in Kszos and Stewart (2003). Lithium is widely used in ceramics, glass and aluminum production (Ober, 2001), and has some applications for synthetic rubber, pharmaceuticals, chemical manufacturing, lubricants, batteries, nuclear reactor coolant and air treatment. Concentrations of Li in surface waters are typically low (usually < 0.04 mg l 1; Durfor and Becker, 1964; Mathis and Cummings, 1973; Emery et al.,

428 Kszos et al. 1981; Hill and Gilliom, 1993; Tanner, 1995), although we have documented one case where ambient levels are elevated due to historical waste-disposal activities at the Department of Energy's (DOE) Y-12 Plant. The toxicity of groundwater from these burial areas to Ceriodaphnia dubia (waterflea; hereinafter referred to as Ceriodaphnia) and Pimephales promelas (fathead minnow) was due primarily to Li (Kszos and Stewart, 2003). Because there was a general paucity of information on Li toxicity to aquatic biota, we conducted this study to evaluate the toxicity of Li to Ceriodaphnia, fathead minnows and Elimia clavaeformis (a freshwater snail). These tests demonstrated that differences in toxicity could be due to the sodium content of the dilution water. Laboratory tests with Ceriodaphnia showed that sodium had an antagonistic effect on Li toxicity that could be described with a model based on the molar ratio of Li to Na. Materials and methods Test organisms Fathead minnow larvae and Ceriodaphnia for tests reported here were obtained and cultured as described in Kszos and Stewart (2003). Freshwater snails, E. clavaeformis, are locally abundant and have been used in other ecotoxicological studies (Burris et al., 1990; Stewart et al., 1993). Snails were collected from a non-contaminated reach of Northwest Tributary located on the DOE Oak Ridge Reservation; they were kept in the laboratory for 48 h in dilute mineral water (DMW; see Kszos and Stewart, 2003), at 25  C, before use. Test methods The toxicity of Li (as LiCl, EM Science, 99% pure) to E. clavaeformis was determined using 72 h laboratory tests that used feeding rate (expressed as mg of lettuce consumed) as the effect endpoint. The tests were conducted by exposing groups of snails (12 snails per test chamber, three replicate test chambers per treatment) to LiCl in DMW. Three fresh lettuce-leaf (Lactuca sativa) disks (2 cm diameter) were stapled together, pre-weighed as a group, and added to each test chamber at the start of the test. The test media was replaced daily, but the lettuce-leaf disks (plus the staples) were left in place. At the end of the test, the

lettuce remaining was removed with forceps, blotted dry on paper toweling and weighed to the nearest 0.01 mg. These measurements were used to determine lettuce-mass lost due to snail consumption. The nominal concentrations of Li tested were 0, 0.05, 0.1, 0.15, 0.3, 1.0, 2.0, 5.0, and 10.0 mg Li l 1. The toxicity of Li (as LiCl) to fathead minnows and C. dubia was determined using the methods described in EPA (1994) and Kszos and Stewart (2003) with the following additions. Tests were conducted concurrently using two types of water as diluent (East Fork Poplar Creek (EFPC) or DMW). Lithium test solutions were prepared 24 h before test initiation, using stock solutions of LiCl and DMW or EFPC water. The test solutions were stored at 4  C; a subsample of each test concentration was removed daily and warmed to 25  1  C for test solution renewal. The conductivity of each dilution was measured on the first day of the test; the pH, conductivity, alkalinity and hardness were assessed daily for the control water. Fifty-ml samples of the three or four lowest Li concentrations were preserved (pH < 2) with ultrex nitric acid for analysis of Li by inductively coupled plasma (ICP) emission spectroscopy. Concentrations of Li in DMW were 0.32 (measured), 0.93 (measured), 1.87 (measured), 3 (nominal), and 4 (nominal) mg l 1. Concentrations tested in EFCP water were 0.36 (measured), 0.98 (measured), 1.90 (measured), 3.0 measured, and 4 (nominal) mg l 1. In 1995, one experiment was conducted with fathead minnows larvae. The minnows were exposed for 7 d to nominal Li concentrations of 0.5, 1, 2, and 4 mg Li l 1 in combination with five (nominal) Na concentrations (as Na2SO4, JT Baker, 99.9% pure) of 2.5, 10, 40, 80, and 160 mg l 1 (see Kszos and Stewart, 2003 for methodology). Between 1994 and 1997, two sets of tests with Ceriodaphnia were conducted to evaluate the influence of Na on Li's toxicity. In the set of tests, the target nominal Li concentrations were: 1 mg Li l 1 (measured range ˆ 0.92±1.3 mg l 1), 2 mg Li l 1 (measured range ˆ 1.7±2.1 mg l 1), and 4 mg Li l 1 (measured range ˆ 3.4±4.0 mg l 1). Sodium was added to DMW as sodium sulfate (Na2SO4) at increasing concentrations, ranging from 1.7 to 40 mg Na l 1. The combinations of Li (measured) and Na (added) tested were 4 mg Li l 1 with 1.7, 3.2, 7, 9.6, 21, and 40 mg l 1 Na; 2 mg Li l 1 with 1.7, 3.5, 7, and 10 mg Na l 1; and 1 mg Li l 1 with 1.7, 3.5, and 7 mg Na l 1. The Ceriodaphnia reproduction data (not shown here)

Toxicity of Lithium 429

revealed a strong relationship between reproduction and the ratio of Li to Na in the water. The experimental design of these tests did not allow for strong statistical analysis of the effects of Na on the toxicity of Li based on Ceriodaphnia reproduction. Thus, a second set of experiments was conducted specifically to test for the effects of the Li-to-Na ratio on Ceriodaphnia reproduction. To further evaluate the effect of sodium and Li on Ceriodaphnia reproduction, four 7 d Ceriodaphnia experiments were used to characterize the toxicity of the 85 combinations of Li and Na shown in Table 1. The four experiments were too large to be conducted simultaneously, so controls to evaluate test-chamber position (block) and differences in reproduction through time were included in each experiment (160 control animals in total). Each combination of Li and Na listed in Table 1 was randomly assigned to a block by drawing numbers and each of the ten replicates per treatment was randomly assigned to a position in the block using a random number generator. Tests 1±3 contained five treatments that contained 700 mg l 1 Na (not shown in Table 1). The results for these ``high-sodium'' tests are not discussed further because (1) later statistical analyses revealed that the response of Ceriodaphnia to the combinations of Li and 700 mg l 1 Na were controlled by the sodium concentration, and (2) our focus was on responses of the animals in that range of Na : Li where there is an interaction. The toxicity test procedures used in these four experiments were identical to those described earlier, with one exception: the Ceriodaphnia neonates (24 h old) used to initiate the tests were first pooled into a large beaker, and individuals were then randomly selected and placed

into each replicate. Water chemistry and analytical data were obtained as described previously. Statistical analyses The inhibition concentration (IC) was calculated using the Linear Interpolation Method (EPA, 1994) and EPA software (EPA, 1993). Data were expressed as the concentration that inhibits reproduction of Ceriodaphnia or growth of fathead minnows by 25% and 50%. In the tests used to study the effect of Na on the toxicity of Li to Ceriodaphnia, we conducted two types of statistical analyses in order to justify combining data from the four experiments for further analysis. The first analysis used only the animals in the control water. A nested analysis of variance (ANOVA; SAS, 1985) was used to test for test-chamber position effects and time effects in experiments 1±4. The total number of young produced in the first 6 d of the test was the response variable. The ANOVA for test-chamber position effects and time effects for control groups showed no significant difference between tests ( p > 0.70), but did reveal a highly significant test-chamber position effect ( p < 0.01). Examination showed that the mean number of young produced by the controls for one block in test 2 was much lower compared to the controls for all other blocks in that test (10.6 offspring per female, versus a range ˆ 22±30 offspring per female). When data from the entire block in test 2 was excluded, the test for effects due to time and test-chamber position within test was not significant (p ˆ 0.13). Because this analysis showed that the block was suspect, all data from the block were excluded from further consideration.

Table 1. Experimental design used to test for the effects of Li-to-Na ratio on Ceriodaphnia reproduction. Values in the body of the table (i.e., 1, 2, 3, 4) indicate experiment number. The combination of 0.20 mg l 1 Li and 13.5 mg l 1 Na, for example, was tested in experiments 1, 2 and 3. Values in parentheses are range of measured values Sodium midpoint (mg l 1) Lithium midpoint (mg l 1) 0.20 (0.18±0.22) 0.39 (0.35±0.43) 0.80 (0.78±0.82) 2.55 (2.5±2.6) 3.45 (3.4±3.5) 4.25 (4.1±4.4) 5.05 (5.0±5.1) 6.05 (5.9±6.2)

13.5 (11±16) 1, 2, 3 1, 2, 3 1, 2, 3 1, 2, 3,4 1, 2, 3

20.0 (19±21)

4 4 4 4

38.5 (37±40)

4 4 4 4 4

50.5 (48±53) 1, 2, 3 1, 2, 3 1, 2, 3 1, 2, 3 1, 2, 3

57.0 (55±89)

4 4 4 4 4

74.5 (73±76)

96.5 (83±110)

4 4 4 4 4

1, 2, 3 1, 2, 3 1, 2, 3 1, 2, 3,4 4 1, 2, 3,4 4 4

430 Kszos et al. We also excluded reproduction data for a few other individual animals from further analysis, based on best professional judgement. Of the 1010 neonates followed in this study, we excluded data on 15 because they were killed accidentally during testing. We also excluded data on nine neonates because (1) they were the only death in the 6 d period for that particular treatment, and (2) these animals produced significantly fewer young, compared to the other nine replicates in that treatment. For example, one animal that we eliminated from statistical analysis produced only one brood of three offspring in 6 d; the remaining nine replicates for this treatment produced, on an average, 20.5 offspring per female. The second analysis involved comparing the response of animals to a treatment that was repeated across all four tests. Treatments were assigned a concentration equal to the midpoint of the measured concentrations, as indicated in Table 1. Due to the experimental design, the only data that fit this criterion were the treatments of 96.5 mg Na l 1 in combination with 2.55 and 4.25 mg Li l 1 (see Table 1). We used ANOVA to evaluate the variation in the total number of young produced in the first 6 d of the test in response to Li concentration, test number and the interaction of Li and test. None of these sources of variation were significant ( p > 0.08), indicating that the animal's response to Li and Na were similar across time. This analysis, in combination with the analysis of the control reproduction noted above, led us to conclude that data from all four tests could be combined. This analysis also indicated that the small variability in measured concentrations for a particular target concentration (see Table 1 for ranges) did not affect the test outcome. Thus, we elected to use the midpoint of the measured concentrations of Li and Na across tests to represent a given concentration (Table 1). Results Lithium toxicity test with snails At higher concentrations, the effect of Li exposure on E. clavaeformis' activity was evident within the first several hours. Lettuce consumption by E. clavaeformis was reduced almost 50% at a Li concentration of 0.15 mg l 1, and was severely inhibited at a Li concentration of 0.3 mg l 1 (Fig. 1). Subsequent tests of Li's inhibitory effects on Pleurocera unicale unicale (a close relative of E. clavaeformis),

Figure 1. Effect of Li on lettuce consumption by the snail, E. clavaeformis (72 h test). Error bars show  SE, for three replicates at each Li concentration tested. C-1 and C-2 represent results for two separate controls.

using a 24 h test method otherwise identical to that used for testing E. clavaeformis, revealed similar results: lettuce consumption was reduced by more than 50% at a Li concentration of about 0.8 mg l 1 (data not shown). Lithium toxicity tests with fathead minnow larvae and Ceriodaphnia Lithium in DMW was 5 to 10 times more toxic than in EFPC water. For Li in DMW, the 25% ICs for fathead minnows and Ceriodaphnia were 0.38 and 0.32 mg Li l 1 respectively. For Li in EFPC water, the 25% ICs for fathead minnows and Ceriodaphnia were 1.99 and 3.33 mg Li l 1 respectively. Likewise, the 50% ICs for fathead minnows and Ceriodaphnia in DMW were 0.57 and 0.72 mg Li l 1 respectively and in EFPC water were 2.47 and >4 mg Li l 1, respectively. The Li and Na concentrations in DMW were < 0.04 mg l 1 and 2.8 mg l 1, respectively. The average Li and Na concentrations in EFPC water < 0.04 mg l 1 and 17.4 mg l 1, respectively. Tests to determine the effects of sodium on lithium's toxicity The results of the fathead minnow test are shown in Fig. 2. The survival of minnows in 80 and 160 mg Na l 1 are not shown because survival was

Toxicity of Lithium 431 120

100

Survival (%)

80 2.5 mg Na/L 10 mg Na/L 40 mg Na/L

60

40

20

0 0

1

2

3

4

Lithium Concentration (mg/L)

Figure 2. Survival of fathead minnows (P. promelas) in response to lithium and sodium in short-term chronic test.

reduced in the Na ‡ DMW treatments (mean survival was 75% at 80 mg l 1 and 32.5% at 160 mg l 1), thus indicating that sodium was dominating the response. Increasing the concentration of Na clearly decreased the toxicity of Li. For example, at 2 mg Li l 1, survival was 2.5% in DMW containing 2.5 mg Na l 1 but was only slightly reduced when the water contained 10 mg Na l 1 (Fig. 2). Similarly, at 4 mg Li l 1 survival was 0% in water containing only 10 mg Na l 1, but >90% in water containing 40 mg Na l 1 (Fig. 2). The growth data for the minnows in this test are not shown because no clear relationship was identified with varying Li and Na. Growth of minnows was high in the three lowest Na treatments (final mean weights of the larvae were 0.42±0.46 mg per fish), indicating that the test conditions were satisfactory (EPA, 1994). Survival of Ceriodaphnia exposed to Li was strongly affected by the concentration of Na. In lowsodium water (1.7 mg Na l 1), 1 mg Li l 1 killed all of the Ceriodaphnia within 6 d (Fig. 3C). In water having a Na concentration of 40 mg Na l 1, Ceriodaphnia survival was 100% even at a Li concentration of 4 mg l 1 (Fig. 3A). The results of the Ceriodaphnia tests conducted with various combinations of Li and Na are restricted to the consideration of effects on reproduction, because by design, the Li and Na levels were selected to be lower than the acute toxicity thresholds for Li and Na. Dunnett's procedure was used to test for significant differences in Ceriodaphnia reproduction between the control and the treatment groups. A summary of Ceriodaphnia reproduction for four Li

concentrations is given in Fig. 4. The responses of Ceriodaphnia to low concentrations of Li (e.g., 0.20± 2.55 mg l 1) are not shown because at these concentrations there was no significant difference compared to the control, except at Na ˆ 700 mg l 1, where Ceriodaphnia reproduction was low regardless of the concentration of Li. The amelioration of Li toxicity with increasing Na concentration is clearly revealed in Fig. 4. For example, in DMW containing 20.0 mg Na l 1, Li concentrations of 3.45, 4.24, and 5.05 mg l 1 significantly reduced reproduction of Ceriodaphnia. In DMW containing 57 mg Na l 1, however, reproduction was not affected in any of these concentrations. The response of Ceriodaphnia to Li and Na was investigated further by examining the reproductive response to the molar ratio of the two metals (Na : Li). In this analysis, the midpoint concentrations of Li and Na were used, as described in Table 1. The response variable was the mean 6 d reproduction from the replicates for a particular treatment. The number of replicates used to estimate this mean reproduction was retained as the weight to associate with each observed mean. The log of the Na-to-Li ratio was calculated by taking the natural log of the mmol Na to mmol Li. The plot of Ceriodaphnia reproduction versus molar ratio is shown in Fig. 5. We used an exponential model to estimate the dependence of 6 d reproduction on the Na-to-Li ratio: y ˆ a…x ‡ d†b ecx

…1†

where y ˆ 6 day reproduction, x ˆ the Na±Li ratio and a, b, c, and d are parameters to be estimated from the observations on y and x. A weighted non-linear estimation procedure was used to estimate the parameters where the weight of each observation was equal to the number of replicates in the given mean y. Figure 5 shows the observed data and the fitted exponential curve. Table 2 contains the parameter estimates and other associated estimates for this analysis. We also estimated the range of x values where the predicted reproduction from equation (1) did not differ significantly from the mean control reproduction (26.2 offspring per female). This was done by considering the standardized difference between the predicted reproduction and the control reproduction as a function of x, for all values of x where the absolute value of the standardized difference was less than 1.96. Standardization was achieved by dividing the difference by its standard

432 Kszos et al. (A)

100

Survival (%)

80

60 3.2 mg Na/L 7 mg Na/L 9.6 mg Na/L 21 mg Na/L 40 mg Na/L

40

20 A (4 mg Li/L) 0 0

Survival (%)

(B)

1

2

3

4

5

6

7

100 80 1.7 mg Na/L 3.5 mg Na/L 7.0 mg Na/L 10 mg Na/L

60 40 20

B (2 mg Li/L)

0 0

Survival (%)

(C)

1

2

3

4

5

6

7

100 80 1.7 mg Na/L 3.5 mg Na/L 7.0 mg Na/L

60 40 C (1 mg Li/L)

20 0 0

1

2

3

4

5

6

7

Test Day Figure 3. Survival of C. dubia in response to mixtures of sodium and 4 mg Li l was 100%.

error. The process described here was used to determine the estimated range of x values providing estimates not differing significantly from the control reproduction. Using this approach, for x in the interval (1.63, 2.06) and the interval (4.79, 6.68), the predicted 6 d reproduction from (1) and control mean did not differ significantly. Therefore, if the log ratio mmol Na/mmol Li was >1.63, the reproduction of Ceriodaphnia was not different from the control mean. At x ˆ 1.816, the predicted value of y is 26.2 (average reproduction of controls), with 95% confidence limits on the estimated mean of (25.5, 26.9).

1

(A), 2 mg Li l

1

(B), and 1 mg Li l

1

(C). Control survival

Discussion The results of this study show that Li is biologically active (i.e., toxic) to various aquatic species in laboratory tests at concentrations in the range of 0.15±0.5 mg Li l 1 (this paper, and Long et al., 1998), yet there is little attention paid to Li in limnological or toxicological texts. We believe that this is due, at least in part, to the fact that Li's toxicity is ameliorated by Na. Fathead minnows and Ceriodaphnia, for example, tolerate concentrations of Li as great as 6 mg l 1 when sufficient Na is present (Figs 2±4). Sodium levels in most ambient waters typically are high enough to afford protection

Toxicity of Lithium 433 13.5 mg/L 20.0 mg/L 38.5 mg/L 50.5 mg/L 57 mg/L 74.5 mg/L 96.5 mg/L

35

Mean Offspring per Female

30 * *

25

*

*

*

5.05

6.05

20 *

15

10

* *

5

*

0 3.45

4.25

Lithium Concentration (mg/L)

Mean 6-d offspring (weighted by number surviving)

Figure 4. Mean (SD) unweighted reproduction (offspring per female) of C. dubia exposed to various combinations of Li and Na. Line represents average reproduction of controls (n ˆ 151). Asterisks () indicate treatments where reproduction was significantly lower than control. Table 2. Parameter estimates for exponential model, y ˆ a(x ‡ d)becx

35

30

Parameter a b c d

25

20

Estimate 23.93 0.8140 0.2375 0.0863

Standard error 1.263 0.1245 0.0414 0.0722

15

10

5

0

0

1

2

3

4

5

6

Log ratio mmole Na/mmole Li

Figure 5. Mean weighted reproduction (offspring per female) of C. dubia in relationship to log ratio of Na-to-Li. Average reproduction of controls in dilute mineral water (n ˆ 151) was 26.2 offspring per female. An exponential model was used to fit the curve to the experimental data.

against the adverse effects of Li in the low part-permillion or sub-part-per-million range. Thus, Li's toxicity might be thought of as being ``hidden'' by levels of Na that are present in most natural systems.

The finding that Li's toxicity to aquatic organisms is strongly affected by sodium is perhaps not surprising given that Li ‡ is known to interfere with the actions of Na ‡ in humans (Klemfuss and Greene, 1991; Bennett, 1997) and other vertebrates (for example, Kinsella and Aronson, 1981; Yu et al., 1993). Because Li is an environmental trace element and effective for treatment of bipolar disorder, much information is available on the toxicity of Li to humans, and on the modifying effect of cations on Li toxicity (Schrauzer and Klippel, 1991). In mammals, the potential action of Li includes almost every process involving the cellular uptake or release of sodium, potassium, calcium, or magnesium, the second messenger system (e.g., compound(s) generated within the cell as a result of an initial binding by a hormone),

434 Kszos et al. hormones, and most of the known neurotransmitters (Klemfuss and Greene, 1991). The monovalent metal cations (Li, Na, K, Rb and Cs) can replace each other in many cellular processes. Li ‡ also interferes with Mg2 ‡ and Ca2 ‡ uptake by replacement at binding sites (Williams, 1973). Functional interactions of Li with other cations include competition for enzyme or membrane binding sites (Klemfuss and Greene, 1991). In mammals, increasing Na intake increases the excretion of Li (Thomsen and Leyssac, 1986, 1987). Thus, Na intake can alleviate many of the toxicological effects of Li (Venugopal and Lukey, 1978). Many of the studies on the toxicity of Li to aquatic biota that we reviewed did not provide information about the concentrations of Na or other ions in the test medium. The study by Long et al. (1998), for example, used water from upper Saginaw Bay of Lake Huron off Whitestone Point as diluent, but the Na concentration in this water was not reported. However, the water was reported to have a conductivity of 160±270 mmho cm 1. Presuming a standard bicarbonate composition (see Hutchinson, 1975, Table 69), the Lake Huron water could reasonably have contained Na at a concentration of 6±11 mg l 1. Tests to determine Li's toxicity to fathead minnows (Fig. 2) showed that Na at 6±11 mg l 1 was sufficient to influence minnow survival, at the Li concentrations of 1.4 and 2.4 mg Li l 1 reported by Long et al. (1998). In fact, if the Na concentration had been as great as 40 mg l 1, perhaps no effect on minnow survival would have been detected at all. Although it is difficult to compare results of the fathead minnow tests by Long et al. (1998) with our results (Long et al. used 26 d tests, while we used 7 d tests), the range of toxic values for Li in the two studies is similar. At a Na level of 10 mg l 1, for example, we found no effect on survival at 1 mg Li l 1 (Fig. 2), and Long et al. (op cit.) reported a 26 d EC50 of 1.0 mg Li l 1 and a NOEC of 0.20 mg Li l 1 (at a presumed Na concentration of 6±11 mg l 1). The results of our fathead minnow tests also are similar to those reported by Emery et al. (1981). In the study by Emery et al. (1981), at a presumed Na level of 3±8 mg l 1 (based on alkalinity of the Columbia River water used as a diluent), juvenile trout survival in a 10 d test was nearly 100% in 0.5 mg Li l 1, but nearly zero when the concentration of Li was greater than 1.0 mg l 1. In our study, at 2.5 mg Na l 1, minnow survival was reduced to

80% at 1.0 mg Li l 1, and survival was only 2.5% at 2 mg Li l 1 (Fig. 2). One of the highest reported Li tolerance limits is that for Colorado squawfish, razorback sucker and bonytail (Hamilton, 1995). In this evaluation, the investigators used reconstituted Middle Green River basin water as diluent. This water contained Na at a nominal concentration of 49 mg l 1. Our results suggest that the toxicity of Li to these species would be much greater if the water had contained less Na. Very few published data are available with which to compare our Ceriodaphnia test results. Anderson (1946) gives a value of 6.5 mg l 1 for (Na and K) in Lake Erie water, which was used as a diluent for the 64 h threshold concentration of 1.2 mg Li l 1 for D. magna (Anderson, 1948). In our tests, at 48 and 72 h, we observed 80% and 60% survival of Ceriodaphnia, respectively, in water containing 2 mg Li l 1 and 3.5 mg Na l 1 (Fig. 3B). Given that 1.64 was the lowest calculated log ratio of (mM Na : mM Li) calculated to be not different from controls for Ceriodaphnia, one can predict that a Na concentration of about 20.5 mg l 1 would have negated the effects of the Li at 1.2 mg Li l 1 in the study by Anderson (1948). We found no information that the interaction of Li and Na in the aquatic environment had been considered or measured at the organism level. Most research using invertebrates has focused on understanding ion transport mechanisms at the cellular level by using excised membranes (e.g., Shetlar and Towle, 1989; Ahearn and Clay, 1989; Ahearn and Franco, 1991; Davis et al., 1992). It is reasonable to hypothesize that the amelioration of Li's toxicity by Na is based upon competitive interactions related to Na‡/H‡ exchange. Grinstein and Wieczorek (1994) provide a review of Na‡/H‡ exchange in animals. In vertebrates, this exchange occurs with a one-to-one stoichiometry and is therefore electroneutral (op. cit.). Ahern et al. (1994) identified a similar electroneutral Na‡/H‡ exchange in apical membranes of crayfish gills. Additionally, other investigators have identified a unique type of exchange in some invertebrates, wherein two extracellular Na ions are exchanged for one intracellular protein (for review see Ahearn et al., 1994). Based on evidence that Daphnia (closely related to Ceriodaphnia) respiratory surfaces are similar to those of crayfish (Kikuchi, 1983;

Toxicity of Lithium 435

Dickson et al., 1991), we expect that a transport mechanism such as the Na ‡ /H ‡ exchange system would be present in Ceriodaphnia tissues. Although we found no studies that evaluated the ability of Li to inhibit the ion-exchange process in crayfish, Duerr and Ahearn (1996) reported that external Li ‡ was an effective competitive inhibitor of the exchange process in lobster (Homarus americanus). Thus, we speculate that the basis for inhibitory effect of Li on Ceriodaphnia reproduction probably stems from competition of Li ‡ with Na ‡ for the exchange of H ‡ . At the whole organism level, many studies have shown that the toxicity of metals such as Zn, Cu and Cd can be ameliorated by elevated levels of calcium (e.g., Rand and Petrocelli, 1985). However, relatively few specifically address the amelioration of a metal's toxicity by sodium (but see Bury et al., 1999; Grosell et al., 2000). In a review by Hall and Anderson (1995), data on the influence of salinity on the toxicity of chemicals to aquatic animals were summarized. Similar to our study, the general trend was that the toxicity of metals such as Cd, Cu, Ni, and Zn increased with decreasing salinity. The authors (op. cit.) attribute this finding to an increase in the activity of the free metal ion (e.g., Cd2 ‡ ) under lower-salinity conditions. Others have proposed models based upon competition of cations for biotic ligands present on fish gills and in the environment (Santore et al., 2001). The biotic ligand model (op. cit.) showed good agreement with experimental data when evaluating the effect of Ca and Na with the acute toxicity of copper to fathead minnows. The authors concluded that Na competes with Cu for copper binding sites on the fish gill. A similar evaluation with Na and Cu was not possible with an invertebrate such as Ceriodaphnia but it is certainly likely that such a competition occurs as discussed previously. It would be interesting to include Li in such mixture studies to evaluate whether Li competes for metal binding sites or interferes with a separate Na‡/ H‡ exchange system. The results of laboratory toxicity tests reported here and by others indicate a potential for Li to impact aquatic ecological communities in some cases. In such cases, the potential for Na to ameliorate the toxicity of Li should be taken into account. Using the model we developed, most natural waters contain Na at concentrations greater than 1 mg l 1, which is

sufficient to ameliorate the effects of Li up to about 0.06 mg l 1 for Ceriodaphnia. However, if the potential exists for higher concentrations of Li to enter the environment, a concurrent increase in Na would be necessary to prevent toxicity. Additional research with more sensitive or region-specific species may be warranted in such cases. Because removal of Li from the aquatic environment is difficult (Kszos and Stewart, 2003), it will be important to maintain best management practices for preventing dissolved lithium from entering the aquatic environment. This will become increasingly important as the use and processing of Li-rich materials increases. Acknowledgments This research was supported by the Y-12 Plant Toxicity Control and Monitoring Program and the Y-12 Biological Monitoring and Abatement Program (BWXT Y-12 Environment, Safety and Health Division). Gail Morris, Belinda Konetsky, Kara Kenney, Wilena Session, Jim Sumner, Terry Phipps, and Natasha Hunt provided outstanding technical assistance. We thank Joe Kszos for his invaluable support, and Sylvia Talmage (Oak Ridge National Laboratory), Joyce Ober (US Geological Survey), and three anonymous reviewers for thoughtful, constructive reviews that improved the manuscript. Oak Ridge National Laboratory is managed by UT-Battelle, LLC, for the US Department of Energy under contract DE-AC0-00OR22725. References Ahearn, G.A. and Clay, L.P. (1989). Kinetic analysis of electrogenic 2 Na‡±1 H‡ antiport in crustacean hepatopancreas. Am. J. Physiol. 257, R484±93. Ahearn, G.A. and Franco, P. (1991). Electrogenic 2Na‡/H‡ antiport in echinoderm gastrointestinal epithelium. J. Exp. Biol. 158, 495±507. Ahearn, G.A., Zhuang, A., Duerr, J. and Pennington, V. (1994). Role of the invertebrate electrogenic 2 Na ‡ ±1 H ‡ antiporter in monovalent and divalent cation transport. J. Exp. Biol. 196, 319±35. Anderson, B.G. (1946). The toxicity threshold of various sodium salts determined by the use of Daphnia magna. Sewage Works J 18, 82±87. Anderson, B.G. (1948). The apparent thresholds of toxicity to Daphnia magna for chlorides of various metals when

436 Kszos et al. added to Lake Erie water. Trans. Am. Fish. Soc. 78, 96±113. Bennett, W.M. (1997). Drug interactions and consequences of sodium restriction. Am. J. Clin. Nutr. 65(suppl. 2), 678S± 681S. Bleiwas, D.I. and Coffman, J.S. (1986). Lithium availability± market economy countries, Bureau of Mines Information Circular 9102, 23 pp. Burris, J.A., Bamford, M.S. and Stewart, A.J. (1990). Behavioral responses of marked snails as indicators of in situ toxicity. Environ. Toxicol. Chem. 9, 69±76. Bury, N.R., McGeer, J.C. and Wood, C.M. (1999). Effects of altering freshwater chemistry on physiological responses of rainbow trout to silver exposure. Environ. Toxicol. Chem. 18, 49±55. Davis, B.A., Hogan, E.M. and Boron, W.F. (1992). Activation of Na±H exchange by intracellular lithium in barnacle muscle fibers. Am. J. Physiol. 263, C246±56. Dickson, J.S., Dillaman, R.M., Roer, R.D. and Roye, D.B. (1991). Distribution and characterization of ion transporting and respiratory filaments in the gills of Procambarus clarkii. Biol. Bull. 180, 154±66. Duerr, J.M. and Ahearn, G.A. (1996). Characterization of a basolateral electroneutral Na ‡ /H ‡ antiporter in atlantic lobster (Homarus americanus) hepatopancreatic epithelial vesicles. Jr. Exp. Biol. 199, 643±51. Durfor, C.N. and Becker, E. (1964). Public water supplies of the 100 largest cities in the United States, 1962, U.S., Geological Survey Water Supply Paper 1812. 364 p. Emery, R.D., Klopfer, C. and Skalski, J.R. (1981). The incipient toxicity of lithium to freshwater organisms representing a salmonid habitat. PNL-3640, UC-11, Pacific Northwest Laboratory: Richland, WA. EPA (US Environmental Protection Agency) (1993). A linear interpolation method for sublethal toxicity: The inhibition concentration (ICp) approach (version 2.0), National Effluent Toxicity Assessment Center Technical Report 03-93, Environmental Research Laboratory: Duluth, MN. EPA (US Environmental Protection Agency) (1994). Short-term methods for estimating the chronic toxicity of effluents and receiving water to freshwater organisms (third edition). EPA/ 600/4-91/002. U.S. Environmental Protection Agency, Cincinnati, OH, Environmental Monitoring Systems Laboratory. Grinstein, S. and Wieczorek, H. (1994). Cation antiports of animal plasma membranes. J. Exp. Biol. 196, 307±18. Grosell, M., Hogstrand, C., Wood, C.M. and Hansen, H.J.M. (2000). A nose-to-nose comparison of the physiological effects of exposure to ionic silver versus silver chloride in the European eel (Anguilla anguilla) and the rainbow trout (Oncorhynchus mykiss). Aquat. Toxicol. 48, 327±42. Hall, L.W. and Anderson, R.D. (1995). The influence of salinity on the toxicity of various classes of chemicals to aquatic biota. Crit. Rev. Toxicol. 24, 281±49. Hamilton, S.J. (1995). Hazard assessment of inorganics to three endangered fish in the Green River, Utah. Ecotoxicol. Environ. Safety 30, 134±42. Hill, B.R. and Gilliom, R.J. (1993). Streamflow, dissolved solids, suspended sediment, and trace elements, San Joaquin River, California, June 1985±September 1988, United States Geologial

Survey Open-File Report, Water-Resources Investigations Report 93-4085. Denver, CO. Hutchinson, G.E. (1975). A Treatise on Limnology, Vol 1, Part 2Ð Chemistry of Lakes. New York: John Wiley. Kikuchi, S. (1983). The fine structure of gill epithelium of a fresh-water flea, Daphnia magna (Crustacea: Phyllopoda) and changes associated with acclimation to various salinities. I. Normal fine structure. Cell Tissue Res 229, 53±68. Kinsella, J.L. and Aronson, P.S. (1981). Interaction of NH4‡ and Li‡ with the renal microvillus membrane Na‡±H‡ exchanger. Am. J. Physiol. 241, C220±6. Klemfuss, H. and Greene, K.E. (1991). Cations affecting lithium toxicity and pharmacology. In G.N. Schrauzer and K.F. Klippel (eds), Lithium in Biology and Medicine, pp. 131±45. New York: VCH Publishers. Kszos, L.A. and Stewart, A.J. (2003). Review of lithium in the aquatic environment: Distribution in the United States, toxicity and case example of groundwater contamination. Ecotoxicology 12, 439±447. Long, K E., Brown, R.P. Jr and Woodburn, K.B. (1998). Lithium chloride: a flow-through embryo-larval toxicity test with the fathead minnow, Pimephales promelas Rafinesque. Bull. Environ. Contam. Toxicol 60, 312±17. Mathis, B.J. and Cummings, T.F. (1973). Selected metals in sediments, water and biota of the Illinois River. J. Water Pollution Control Federation 45, 1573±83. Ober, J.A. (2001). LithiumÐ2000. In Minerals Yearbook, Metals & Minerals.U.S. Geological Survey, Minerals Information: Reston, VA, URL: http://minerals.usgs.gov/minerals/pubs/commodity/ lithium/. Rand, G.M. and Petrocelli, S.R. (1985). Fundamentals of Aquatic Toxicology. Hemisphere: Washington, 666 pp. SAS (1985). SAS1 User's Guide: Statistics, Version 5 Edition. SAS Institute Inc: Cary, 956 pp. Santore, R.C., Di Toro, D.M., Paquin, P.R., Allen, H.E. and Meyer, J.S. (2001). Biotic ligand model of the acute toxicity of metals. 2. Application to acute copper toxicity in freshwater fish and Daphnia. Environ. Toxicol. Chem. 20, 2397±402. Schrauzer G.N. and Kippel K.F. (eds) (1991). Lithium in Biology and Medicine, 209 pp, New York: VCH Publishers. Shetlar, R.E. and Towle, D.W. (1989). Electrogenic sodium-proton exchange in membrane vesicles from crab (Carcinus maenas) gill. Am. J. Physiol. 257, R924±31. Stewart, A.J., Hill, W.R. and Boston, H.L. (1993). Grazers, periphyton and toxicant movement in streams. Environ. Toxicol. Chem. 12, 955±7. Tanner, D.Q. (1995). Surface-Water-Quality Assessment of the Lower Kansas River Basin, Kansas and Nebraska: Distribution of Trace-Element Concentrations in Dissolved and Suspended Phases, Streambed Sediment, and Fish Samples, May 1987±April 1990, Lawrence, Kansas: USGS Water-Resources Investigations Report 94-4201, Denver, CO. Thomsen, K. and Leyssac, P.P. (1986). Acute effects of various diuretics on lithium clearance-studies in rats on medium and low sodium diet. Renal Phys. Biochem. 9, 1±8. Thomsen, K. and Leyssac, P.P. (1987). Effect of sodium-poor diet on lithium clearance at different dietary potassium conents in rats. Pharmacol. Toxciol. 61, 260±4.

Toxicity of Lithium 437 Venugopal, B. and Luckey, T.D. (1978). Metal Toxicity in Mammals. Vol. 2, New York: Plenum Press. Williams, R.J.P. (1973). The chemistry and biochemistry of lithium. In S. Gershon and B. Shopsin (eds), Lithium: Its Role in Psychiatric Research and Treatment, pp. 15±31. New York: Plenum Press.

Yu, F.H.Shull, G.E.Orlowski, J. (1993). Functional properties of the rat Na/H exchanger NHE-2 isoform expressed in Na/H exchanger deficient Chinese hamster ovary cells. J. Biol. Chem. 268, 25536±41.