9 Aquatic Life

9.1 Toxicity to Aquatic Biota

Selenium toxicity to aquatic organisms is governed by several factors, including:

· form and concentration of selenium,
· type and characteristics (e.g., species and life stage) of organism,
· period and type of exposure to selenium (e.g., acute or chronic), and
· environmental factors (e.g., water hardness, temperature, presence of other substances, etc.).

Many studies have examined the acute and chronic effects of selenium in freshwater and marine biota. However, a great deal of variability exists in the quality of published data. Although many data were found in the literature, only the critical data were examined for the derivation of water quality guideline for the protection of aquatic life. The acute and chronic toxicity data for selenium obtained from the literature are summarized in Tables 9.01 to 9.10 and Figures 9.01 to 9.27. The discussion on acute effects include lethal and sublethal response data obtained on organisms exposed to selenium for more than 96 h; the section on chronic effects discusses lethal and sublethal chronic response data for organisms exposed for a period >96 h.

9.2 Freshwater Biota

9.2.1 Fish

(a) Acute Data

Table 9.01 and Figures 9.01 to 9.04 show that the acute toxicity to fish ranged from of 0.005 to 126.6 mg Se/L. Lemly (1983) reported a sublethal effect as a change in ventilatory rate or breathing activity at a concentration of 0.005 mg/L for both bluegill (Lepomis macrochirus) and largemouth bass (Micropterus salmoides). A sublethal effect, conditional avoidance, was also reported for goldfish (Carassius auratus) exposed to 0.25 mg/L selenite-Se (Weir and Hine 1970). However, the ecological significance of these data is difficult to determine because their impact on the population as whole is not well defined.

The 96-h LC₅₀ ranged from 0.62 to 96.8 mg Se/L for several species of freshwater fish (Figure 9.03). The lowest selenite-Se concentration in water, causing 50% mortality in adult fathead minnows (Pimephales promelas), was reported to be 0.62 mg Se/L by the USEPA (1980) and Kimball (Cited in USEPA 1987). Halter et al. (1980) also reported a 96-h LC₅₀ of 1 mg/L selenite-Se for adult fathead minnows at pH 7.3, 329 mg/L CaCO₃ water hardness, and 25 C.

Figure 9.04 suggests that selenite-Se is more toxic than selenate-Se. This trend may be speculative, as selenate toxicity data are scant. Several studies have shown a clear difference in selenite-Se and selenate-Se toxicity. For instance, the results from studies with zebrafish (Brachydanio rerio) suggest that selenite is significantly more toxic than selenate (Niimi and LaHam 1975, 1976). Fifty percent mortality was observed in the fish exposed to either 20 mg/L selenite-Se or 82 mg/L selenate-Se. Hamilton and Buhl (1990) made similar observations with both coho salmon (Oncorhynchus kisutch) and Chinook salmon (O. tshawytscha). The swim-up fry (8-12 weeks) of coho salmon showed a 50% mortality at 7.8 mg/L of selenite-Se or 32.5 mg/L of selenate-Se; both tests conducted at pH 7.82, 12 degrees C, water hardness of 333 mg CaCO₃/L. These studies also indicated that the younger life stages of both coho and chinook salmon were more sensitive to the toxic effects. In their study, Hamilton and Buhl also demonstrated that, in general, mixtures of selenate and selenite in the ratios of 3.5:1, 6:1, and 14:1 have similar acute toxicity values regardless of fish age. They concluded that the joint acute toxicity of these binary mixtures to the fish could be characterized as strictly additive.

(b) Chronic Data

The available data indicate that chronic toxicity of selenium ranged from 0.005 to 40 mg Se/L for several species of freshwater fish (Table 9.02 and Figures 9.05 to 9.07). Sorensen (1988) reported abnormalities of both liver and ovaries in redear sunfish (Lepomis microlophus) from Martin Lake that was contaminated with selenium at 0.005 mg Se/L. In a study conducted between 1975 and 1976, Cumbie and Van Horn (1978) reported complete reproduction failure for bluegill (Lepomis macrochirus), green sunfish (L. cyanellus), largemouth bass (Micropterus salmoides) and flat bullhead (Ictalurus platycephalus) from Belews Lake that had 0.01 mg Se/L at pH 7.5 and water hardness of 38.2 mg CaCO₃/L. A total of 29 species of fish in this lake decreased in standing crop from 86 680 to 27 099 g/ha for this same time period. Because the levels of selenium in the water were relatively low for the severity of the effects on the fish, it was hypothesized that these deleterious effects were a result of bioaccumulation through the food chain; the phytoplankton from two sites at Belews Lake contained an average of 15 to 70 µg Se/g dry-weight.

In a three-trophic level and a continuous flow-through test system, Dobbs et al. (1996) exposed alga (Chlorella vulgaris), rotifers (Brachionus calyciflorus), and fathead minnows (Pimephales promelas) for 25 days to various levels of selenium (0, 0.110, 0.207, and 0.396 mg Se/L). They found that the fathead minnow growth was impaired after seven days at 0.207 and 0.396 mg Se/L levels, with 100% mortality by day sixteen. A reduction in the fish biomass, compared to the control, was also apparent at 0.110 mg Se/L after day twenty.

In bluegills exposed to 0.005 mg Se/L in water at 20 degrees C and 5.1 µg Se/g dry-weight in their diet for 180 days, Lemly (1993) reported hematological changes and gill damage in the fish that reduced respiratory capacity while, at the same time, increasing oxygen consumption and respiratory demand. When these tests were conducted at a lower temperature of 4 degrees C, "Winter Stress Syndrome" was seen which was symptomised by an increase in stress-related energy demand, a reduction in feeding, and eventually death as a response to the shorter photoperiod and colder temperature.

Gillespie and Baumann (1986) studied bluegill larvae collected from crosses of adults from Hyco Reservoir with relatively high selenium in the water (0.009-0.012 mg Se/L) and in the diet (25-45 µg Se/g dry-weight), and adults exposed to much lower levels of selenium. They noted that the exposure to Se in the Hyco Reservoir resulted in larvae with gross abnormalities and general edema. They concluded that the symptoms seen in the larvae were due to selenium transferred to the egg from the female parent. Hermanutz et al. (1992) saw similar results with bluegills exposed to 0.01 mg Se/L in experimental streams, which affected both growth and survival of larvae and growth of adults. Schultz and Hermanutz (1990) conducted studies on fathead minnows (Pimephales promelas) which supported these conclusions, resulting in a high incidence of lordosis and edema in larvae at water concentrations of 0.01 mg Se/L in experimental streams. These studies of the uptake of selenite in embryos after hatching, concluded that the high levels seen in embryos were not taken up from the water initially upon hatching, but were in fact passed on from the parent. Hermanutz (1992) observed abnormalities in his studies on the late juvenile and early adult development of fathead minnows from the effect of selenium in water (0.010 mg Se/L) and diet in an experimental stream.

DeFrost et al. (1999) reviewed residue-based selenium toxicity thresholds for freshwater fish. Based on the available literature, they felt that the scientific literature is not supportive of sediment or water thresholds. The reason for this was the site-specific factors that influence Se bioavailability, bioaccumulation, and toxicity in aquatic systems. They were, however, supportive of the tissue-based thresholds and proposed whole body thresholds of 9 mg/kg dw for warmwater fish and 6 mg/kg dw for larval coldwater anadromous fish, an ovary threshold of 17 mg/kg dw for warmwater fish, and fish dietary thresholds of 10 and 11 mg/kg dw for warmwater and larval coldwater anadromous fish, respectively.

Kennedy et al. (2000) studied the effect of elevated concentration of selenium (0.0133 to 0.0145 mg/L - caused by coal mining in the area) on wild cutthroat trout in the Elk River in southeastern British Columbia. The results were compared with the fish caught from the reference site with <0.001 mg Se/L in water. As expected, the Se concentration in females from the Se-elevated sites (e.g., 21.2 plus or minus 20.2 micrograms/g dw in eggs) was much higher than those from the reference (e.g., 4.6 plus or minus 1.8 micrograms/g dw in eggs). Despite the elevated egg Se concentrations, there was no significant effect on fertilization, time to hatch, percent hatch, or egg, larvae, and fry deformities or mortalities. The investigators concluded that the lack of any toxic response might be due to an evolved tolerance of the fish to high levels of selenium in the seleniferous river system.

Figure 9.07 shows a possible trend in selenite being more toxic than selenate. However, available data were scant and no comparative studies were found in the literature.

9.2.2 Invertebrates

(a) Acute Data

The acute toxicity of selenium ranged from 0.006 to <200 mg Se/L for several species of freshwater invertebrates (Table 9.03 and Figures 9.08 to 9.11). A comparison between Figures 9.01 and 9.08 results indicated that selenium was more toxic to invertebrates than fish.

Malchow et al. (1995) reported significant reduction in larval growth of the midge (Chironomus decorus) exposed to 0.01 mg Se/L, as selenite or selenate, in water for a period of 96 hours. In their tests, they found that the larvae were feeding on algae (Selenastrum capricornutum) which had bioaccumulated selenium from water to concentrations of 2.84 µg selenite-Se/g dry-weight and 2.11 µg selenate-Se/g dry-weight, respectively. Algae had made selenium more available to the midge.

Ingersoll et al. (1990) reported a lower range of values for selenite-Se (7.95 to 14.6 mg/L) toxicity than for selenate-Se (10.5-16.2 mg/L) in 48-h LC₅₀ studies, indicating that selenite may be slightly more toxic to the midge (C. plumosus).

Bringmann and Kuhn (1977b) exposed the water flea (Daphnia magna) to 0.01 mg Se/L as selenite for 24 h at a water hardness of 214 mg CaCO₃/L and reported a change in swimming ability in 50% of the organisms. The quality of these data could not be ascertained because the original source was not found.

Maier et al. (1993) studied mortality in the neonates of the water flea (D. magna) exposed to different forms of selenium in water at pH 8.2, dissolved oxygen (or DO) level of 8.6 mg/L, and 20 degrees C. The 48-h LC₅₀s were as follows: 2.84, 0.55, 0.31 and 2.01 for selenate-Se, selenite-Se, seleno-DL-methionine, and seleno-DL-cystine, respectively. These data show that organic forms of selenium are as toxic, if not more, as the inorganic forms. Immobilization as a sublethal acute response (or 48-h IC₅₀) was observed at concentrations of 0.045 mg/L for seleno-DL-methionine and 0.52 mg/L for seleno-DL-cystine. These investigators also reported that the presence of sulphate in water modified the toxicity of the different forms of selenium. An increase in sulphate concentration resulted in a decrease in toxicity for selenate, varied the toxic effects of selenite, and had no effect on the toxicity of seleno-DL-methionine. Johnston (1987) also showed higher toxicity for selenite than selenate, but the difference was less significant than that reported by Maier et al. (1993). The tests with adult water fleas at 22 degrees C and pH 8.4, yielded 96-h LC₅₀ concentrations of 0.44 mg/L selenite-Se and 0.58 mg/L selenate-Se. More recently, Brasher and Ogle (1993) reported that the 96-h LC₅₀ for the two-month old amphipod (Hyallela azteca) at pH 8.6, DO 7.7 mg/L, 25 degrees C, 133 mg CaCO₃/L hardness, and 302 mg CaCO₃/L alkalinity was three times lower with selenite-Se (0.676 mg/L) than with selenate-Se (1.868 mg/L).

Davis et al. (1988) reported that 0.006 mg Se/L as selenite resulted in 50% mortality of the cladoceran (Daphnia pulicaria) in 96 hours. The same results were obtained with water fleas (D. magna) exposed to 0.04 mg/L selenium as seleno-DL-methionine by these investigators. The 96-h LC₅₀ of 0.006 mg/L selenite-Se was the lowest found in the literature.

Owsley (1984) and Owsley and McCauley (1986) showed that the Se toxicity differed between cladocerans (Ceriodaphnia affinis) whose parents were previously exposed to selenium (48-h LC₅₀=0.35 mg Se/L) and those which weren't (48-h LC₅₀=0.6 mg Se/L). Their tests were carried out at pH 7.9, DO 8.4 mg/L, 21.5 degrees C, and water hardness of 100.8 mg CaCO₃/L and suggested that the toxicity is passed from parent to offspring and this adds to accumulation.

(b) Chronic Data

The chronic toxicity of selenium ranged from 0.002 to 15 mg Se/L for several species of freshwater invertebrates (Table 9.04 and Figures 9.12 to 9.14). Davis et al (1988) reported a decrease in reproduction of the protozoan Entosiphion sulcatum, when exposed to 0.003 mg/L selenite-Se.

Crane et al. (1992) reported changes in abundance of some freshwater invertebrates tested with a 60:40 ratio of selenate-Se to selenite-Se in a 514-day study. Midges (Chironomidae) and water fleas (Eurycercus lamellatus and Graptoleberis testudinaria) were shown to have a decrease in abundance at concentrations 0.002 to 0.025 mg Se/L. The decrease in abundance was also observed between 0.010 and 0.025 mg Se/L for the water flea (Chydorus ovalis), and 0.025 mg Se/L for the water flea (Acroperus harpae), the parasitic copepod (Ergasilus sp.), and the tubificid worm (Tubifex tubifex). It was not apparent whether the decrease in abundance was due to a decrease in reproduction or in growth to adult stage.

Brasher and Ogle (1993) reported that the 10-d LC₅₀ for the two-month old amphipod (Hyallela azteca) exposed to selenium was twice as high with selenate-Se (1.135 mg/L) than with selenite-Se (0.502 mg/L), at pH 8.6, DO 7.7 mg/L, 25 degrees C, 302 mg/L alkalinity, and 133 mg CaCO₃ hardness. Also, a significant reduction in the number of young per female was observed at a concentration of 0.20 mg/L selenite-Se, but no effect on reproduction was observed when exposed to 0.70 mg/L selenate-Se.

In a three-trophic level and a continuous flow-through test system, Dobbs et al. (1996) exposed an alga (Chlorella vulgaris), a rotifer (Brachionus calyciflorus), and a fish (Pimephales promelas) for 25 days to various levels of selenium (0, 0.110, 0.207, and 0.396 mg Se/L). They found that the rotifer population standing crop (measured as dry weight) was significantly reduced at 0.207 and 0.396 mg Se/L by day four and declined to below detection by day seven. A reduction in rotifer biomass, compared to the control, was also apparent at 0.110 mg Se/L after day twenty.

Boyum and Brooks (1988) reported that organically-bound selenium, when present in water with selenate-Se, significantly reduced the toxicity of the inorganic selenium by altering its rate of incorporation. The tests indicated that an exposure to 0.05 to 1 mg/L selenate-Se decreased the survival of Daphnia magna and reduced the number of offspring produced. When a dietary source of algae, grown in 0.05 to 1 mg/L selenate-S, was added to the test solution, the survival rate for the organisms increased. However, no offspring were produced at 1 mg Se/L, suggesting that reproduction is a more sensitive end point with a dietary source of selenium.

9.2.3 Algae and Macrophytes

The toxicity of selenium to freshwater macrophytes and algae varied from 0.01 to 99 mg Se/L (Table 9.05 and Figures 9.15 to 9.17).

Davis et al. (1988) reported a decrease in cell division for an alga (Ankistrodesmus falcatus) exposed to 0.010 mg Se/L as selenate. For the green alga Selenastrum capricornutum, Foe and Knight (Cited from USEPA 1987) reported a decrease in dry-weight and chlorophyll a at a concentration of 0.075 mg Se/L as selenite.

Kiffney and Knight (1990) exposed cyanobacteria (Anabaena flos-aquae) to seleno-L-methionine, selenate-Se, and selenite-Se for 10 days. The first sub-lethal effect, a decrease in chlorophyll a, was observed at 0.1, 3.0, and 3.0 mg Se/L, respectively. Davis et al. (1988) reported that concentrations of 0.3, 5 and 4, respectively, for these same forms of selenium caused a decrease in cell division of the cynobacterium.

In a three-trophic level and a continuous flow-through test system, Dobbs et al. (1996) exposed an alga (Chlorella vulgaris), rotifers (Brachionus calyciflorus), and fathead minnows (Pimephales promelas) for 25 days to 0, 0.110, 0.207, and 0.396 mg/L selenium from selenate in natural creek water supplemented with nutrients to sustain algal growth. In an initial algal-screening test, they found that the algal growth rates were reduced at 0.082 and 1.053 mg Se/L, with no effect at 0.0085 mg Se/L. In the trophic test, the algal population showed evidence of reduced growth rates at 0.207 and 0.396 mg Se/L, but not at 0.110 mg Se/L.

Richter (1982) exposed the green algae Selenastrum capricornutum to selenate-Se and selenite-Se and reported 4-d EC₅₀s of 0.199 mg Se/L and 2.9 mg Se/L, respectively. Davis et al. (1988) reported a decrease in cell division of 20% for the same species of algae exposed to 0.1 mg/L selenite-Se. In a test solution containing 3.3 mg/L sulphate, Williams et al. (1994) reported a decrease in the growth (weight) of S. capricornutum at a concentration of 0.1 mg/L selenium as 4% selenite and 96% selenate. When the sulphate content of the test solution was raised from 3.3 to 33 mg/L, the bioaccumulation of selenium in the algae decreased significantly and the growth of the algae increased comparatively.

9.3 Marine Biota

Data are limited to assess the effects of toxicity on marine organisms.

9.3.1 Fish

(a) Acute Data

Acute toxicity of selenium to marine fish ranged from 0.6 to about 86 mg Se/L (Table 9.06 acute data only and Figures 9.18-9.21 that include both acute and chronic toxicity data). Cardin (1986) reported the lowest EC₅₀ or LC₅₀ value for the larvae of haddock (Melanogrammus aeglifinus) as 0.599 mg/L selenite-Se. The USEPA (1980) also reported 50% mortality in haddock (M. aeglifinus) exposed to a concentration of 0.6 mg Se/L for 96 hours.

In a study with striped bass (Morone saxatilis), Chapman (1992) exposed the fish to selenite-Se and selenate-Se during different life stages and conditions of inflated or uninflated bladder. The test conditions were: near saturation DO, 17 degrees C, 38.5 to 40.5 g/kg salinity, and 750 to 830 mg CaCO₃/L hardness. For 24-day old fish with an inflated bladder, the 96-h LC₅₀s were reported to be 26.3 mg/L selenite-Se and 3.4 mg/L selenate-Se, as compared to 36.7 mg/L selenite-Se and 3.0 mg/L selenate-Se for the fish with an uninflated bladder. Other life stages displayed similar trends. These results indicate that selenate is significantly more toxic to marine fish than selenite. This is opposite to what was found in fresh water. Figure 9.21 compares the toxicity of various forms of selenium; unfortunately, only limited data were found in the literature.

(b) Chronic Data

The chronic toxicity data for marine fish were limited (Table 9.07 chronic data only and Figures 9.18-9.21 that include both acute and toxicity data). The lowest chronic toxicity value was obtained by Klauda (1985), who reported a significant incidence of developmental anomalies of the lower jaw in striped bass (Morone saxatilis) exposed to a concentration of 0.039 mg/L selenate-Se for 9 to 65 days. Also, a significant incidence of severe blood cytopathology was reported in the fish exposed to 1.29 mg/L selenate-Se. Klauda (1986) reported a high number of deformities in the striped bass exposed to a concentration of 0.09 mg Se/L during a period from post-hatch through 60 days.

Ward et al. (1981) reported that a concentration of 0.675 mg Se/L caused sublethal effects in sheepshead minnow (Cyrinodon variegatus) during early life stages.

9.3.2 Invertebrates

(a) Acute Data

The acute toxicity of selenium to marine invertebrates varied from 0.127 to less than 10 mg Se/L (Table 9.08, Figures 9.22-9.24 that includes both acute and chronic toxicity data).

In a life cycle test with opossum shrimp (Mysidopsis bahia), the USEPA (1980) reported an LC₅₀ of 0.127 mg Se/L for the egg stage. Nelson et al., (1988) reported that a concentration of 0.255 mg Se/L as selenite resulted in a 50% mortality of the bay scallop (Argopectin irradians).

Glickstein (1978) reported 30% mortality in the dungeness crab (Cancer magister) when exposed to 0.1 mg Se/L as selenite for a period of 96 hours; the mortality increased to 100% at 5.0 mg Se/L. There were no data available on the effects of selenate-Se in marine invertebrates.

(b) Chronic Data

Few chronic toxicity data were found in the literature for marine invertebrates (Table 9.09, Figures 9.22-9.24). Micallef and Tyler (1990) reported a reduction in filtration rates by 86-100% when the blue mussels (Mytilus edulis) were exposed to 0.05 mg Se/L in the laboratory. Ward et al. (1981) reported a life cycle chronic value of 0.212 mg/L selenite-Se for the mysid Mysidopsis bahia.

Wolfenberger (1986) studied selenium toxicity in the hermit crab (Clibanarius vittatus) at a concentration of 100 mg Se/L, as a function of temperature and salinity. It was reported that survival time decreased as temperature reached 20 degrees C and fluctuated, with no definite trend, with salinity between 100/00 to 400/00.

9.3.3 Algae and Macrophytes

Table 9.10 and Figures 9.25 to 9.27 show the toxicity of selenium to marine algae and macrophytes.

Studies by Wheeler et al. (1982) reported that the effects of selenium on algae depend on the type of algae, the concentration and form of selenium, and the concentration of sulphate. In experiments where the organisms were pre-treated with the sulphate concentration (850 mg/L) normally present in seawater, the most sensitive of the algae tested was the red alga (Porphyridium cruentum) which showed a decrease in growth at 0.01 mg/L selenate-Se. The growth of most of the other algae including the green algae Dunaliella primolecta, Platymonas subcordiformis, and Platymonas sp., and the diatom Tetraselmis chuii, was reduced between 0.1 and 1.0 mg/L selenate-Se with the exception of the green algae (Chlorella sp.). A concentration of 10 mg/L selenate-Se was lethal in 4-5 days to all algae tested. However, when the organisms were pre-treated with a sulphate concentration of one-tenth the normal seawater concentration (i.e., 85 mg/L), Wheeler et al. (1982) found that the dose of 0.1 mg/L selenate-Se reduced the growth of some algae and was lethal to others. Also, a concentration of 0.01 mg/L selenate-Se improved the growth of most algae over the growth usually observed with an 85 mg/L sulphate concentration and no selenium added to the medium.

Figure 9.27 suggests that selenate is more toxic than selenite to marine algae and macrophytes. This corresponds with Figure 9.21 that shows the same trend with marine fish.

9.4 Interactions

Selenium interacts with several metals and non-metals in the environment. In reviewing interactions with various metals, Marier and Jaworski (1983) noted that selenium-interactive elements include metals that (i) are widely used for their conductance or electron-transfer properties in the electric field (e.g., copper, germanium, tungsten), (ii) have a strong affinity for sulphide (e.g., silver, mercury), or can, like Se, easily be reduced by metabolic processes (e.g., arsenic). Also, all elements do not interact with Se in a similar manner. For instance, antimony, copper, germanium, and tungsten alleviate selenium toxicity. Conversely, selenium counters the toxicity of cadmium, mercury (or methyl mercury), silver and thallium. Arsenic, on the other hand, is involved in both types of interactions.

Maier et al. (1993) studied the effects of varying sulphate concentration on the toxicity of selenate-Se, selenite-Se, and seleno-DL-methionine to Daphnia magna. They found that both the chemical form of selenium and the sulphate concentration influenced the toxicity of selenium. Increasing the concentration of sulphate decreased, varied, and left unaffected the toxicity of selenate-Se, selenite-Se, and seleno-DL-methionine, respectively. These investigators also concluded that (a) the toxicological interactions of the various forms of Se were additive, (b) there was a common uptake pathway for sulphate and selenate-Se with simple chemical competition occurring, causing a decrease in selenate toxicity with an increase in sulphate concentration, and (c) there were separate pathways for selenite-Se and seleno-DL-methionine accumulation that were not affected by sulphate concentrations. Williams et al. (1994) also reported antagonism between sulphate and selenate-Se uptake and toxicity in the green alga S. capricornutum.

Paulsson and Lundbergh (1989) found that treatment of lakes with selenium over a period of three years reduced significantly the uptake of mercury by perch (P. fluviatilis), pike (E. lucius) and roach (L. rutilus). The Se levels achieved (at one metre depth) through treatment with sodium selenite ranged from 0.001 to 0.006 mg/L. These investigators also found that the fish accumulated high levels of Se in the tissue during the experiment. As a result, they recommended that for such amelioration procedures the selenium level should be strictly controlled at less than 0.005 mg/L. The investigator did not comment on the any adverse effects of the elevated Se levels in fish.

Gaikwad (1989) found that the toxicity of a mercury, copper, and selenium mixture was much higher than the toxicity of the individual metals.

9.5 Summary of Existing Guidelines

9.5.1 Freshwater Aquatic Life

The Canadian Council of the Ministers of the Environment (CCME, formerly CCREM 1987) has recommended a maximum water quality guideline of 1.0 µg/L total selenium to protect freshwater aquatic life in Canada.

The USEPA (1986) expressed their freshwater criteria for the protection of aquatic life in terms of total recoverable inorganic selenite. They recommended that the 24-hour average concentration should not exceed 35 µg/L, and the maximum concentration to eliminate acute effects should not exceed 260 µg/L at any time.

9.5.2 Marine Aquatic Life

To protect saltwater aquatic life, the USEPA (1986) recommended that the average concentration of inorganic selenite (total recoverable) should not exceed 54 µg/L in 24 hours, and the maximum concentration at any time should not exceed 410 µg/L. The Canadian Council of Ministers of the Environment (CCME) did not establish a selenium guideline to protect marine life.

9.6 Recommended Guideline and Rationale

9.6.1 Freshwater Aquatic Life

It is recommended that the maximum concentration of total selenium in water for the protection of freshwater aquatic life should not exceed 0.002 mg Se/L.

For the protection of aquatic life, it is also recommended that total selenium in the sediment and fish tissue should not exceed 2 micrograms/g dry weight and 1 microgram/g body weight (wet weight), respectively. The sediment guideline will apply to sediments with total organic content of 5%. The sediment and tissue guidelines are interim thresholds that provided the basis for the recommended water quality guideline. Full guidelines will be proposed for these media at a later date using the appropriate protocols.

9.6.2 Marine Aquatic Life

To protect aquatic life in marine environments, it is recommended that the maximum concentration of total selenium in water should not exceed 0.002 mg Se/L.

9.7 Rationale

9.7.1 Freshwater Aquatic Life

Three approaches were used to derive a guideline that will protect freshwater aquatic life from the adverse effects of selenium in water. The proposed guideline of 2 micrograms Se/L was within the range justified in the arguments presented below. This concentration will protect aquatic life in the most sensitive environment of lakes and reservoirs from the adverse effects of Se in water.

A. Water Concentration Approach

In a semi-natural ecosystem of experimental streams, Hermanutz (1992) indicated that a concentration of 0.01 mg Se/L resulted in malformation of late juvenile and early adult fathead minnows. An exposure to 0.01 mg Se/L also caused edema and lordosis in fathead minnow embryos, and reduced adult growth and reproduction in (Schultz and Hermanutz 1990, Hermanutz et al. 1992). In Hyco Reservoir, which was contaminated with effluent from coal-fired power plants, Gillespie and Baumann (1986) found reproduction failure in bluegill and deformity in larvae that were exposed to a Se concentration ranging between 0.009 and 0.012 mg Se/L. Cumbie and VanHorn (1978) observed mortality, deformity, and a reduction in standing crop of threadfin shad, flat bullhead, green sunfish, bluegills and largemouth bass in Belews Lake. Complete reproduction failure was also seen in bluegills and largemouth bass. The concentration of selenium was found to be 0.01 mg Se/L in the main lake. Bringmann and Kuhn (1977b) reported a 24-h EC₅₀ (swimming) of 0.010 mg Se/L for D. magna.

The aquatic systems of Hermanutz (1992), Hermanutz et al. (1992), Schulltz and Hermanutz (1990), Gillespie and Bauman (1986), and Cumbie and VanHorn (1978) can be characterized as standing or slow moving with low flushing rates. Such systems can build up large deposits of organic-rich sediment that can serve as a site for Se storage and remobilization through the benthic-detrital food web and rooted plants (Canton and Van Derveer 1997). Therefore, for derivation of the proposed guideline, it was assumed that the organisms in the above studies were exposed to both Se in water as well as from food sources. (The authors, however, did not state the selenium concentration in foods consumed by the organisms). Furthermore, it was also assumed that in these aquatic systems, the selenium concentrations in the water column and the internal food sources was in equilibrium.

Based on the lowest observed effect level of 0.01 mg Se/L and a safety factor of 5², a guideline of 0.002 mg/L of selenium was proposed to protect aquatic life in fresh water.

Several studies reported adverse effects of selenium on aquatic life at concentrations below 0.01 mg Se/L (see below). These studies were not considered in the derivation of the proposed water quality guideline. The reasons for their rejection were several-fold:

· they lacked details about the data and their original source (e.g., Davis et al. 1988);
· they failed to provide details on experimental methodology (re: Se effects), which made the data suspect (Hyne et al. 1993);
· chronic endpoints were indefinite or could have been influenced by factors other than those considered in the study (e.g., Crane et al. 1992), and
· test organisms, while exposed to waterborne selenium, were also fed a diet containing selenium (an external source of selenium; e.g., Lemly 1993, Alaimo et al. 1994).

Crane et al. (1992) and Hyne et al. (1993) reported the lowest observed effect level for selenium at 0.002 mg /L. In the former study, the authors reported a decrease in abundance in several invertebrates exposed to Se for a long period of time (about 1.5 years); these results could have been influenced by other factors such as predation. The latter study with daphnia (Moinodaphnia macleyi) had very little information in the source paper about materials and methods used in the experiment. Davis et al. (1988) reported that 0.003 mg Se/L decreased reproduction in the protozoan E. sulcatum and 0.006 mg Se/L as a 96-h LC₅₀ for the cladoceran D. pulicaria. However, this paper was a review of the literature and the original sources of data were not reported. Sorensen (1988) reported that a waterborne concentration of 0.005 mg Se/L caused substantial alterations in the livers (necrosis, cytoplasmic, vacuolation, etc.), as well as in the ovaries (atretic follicles, connective tissue hypertrophy, etc.), of redear sunfish. These endpoints were not accepted for the derivation of water quality guidelines because their ecological significance could not be ascertained. Lemly (1993) reported hematological changes and gill damage in bluegills exposed to a waterborne concentration of 0.005 mg Se/L. This study was rejected because the fish were also fed an external diet containing 5.1 µg Se/g (dry-weight).

B. Tissue Concentration Approach

In an aquarium study, Hamilton and Wiedmeyer (1990) reported no-effect concentrations of 35 microgram/L and 70 microgram/L total selenium for growth and survival of chinook salmon in both freshwater and well water. This no-effect waterborne concentration range was considered to be equivalent to a no-effect, whole-body selenium concentration of 3-5 microgram/g body weight (bw) on a dry weight (dw) basis in well water and 2-4 microgram/g bw dw in freshwater. The threshold of 3-5 microgram Se/g bw on a dry weight basis was corroborated by Hamilton et al. (1990) in a feeding study with chinook salmon. Brix et al. (2000) combined Hamilton et al. (1990) data with similar data from the literature and reworked them using sophisticated statistical procedures. They, then, proposed a whole-body toxicity threshold (EC10) of 6 microgram/g dw for the chinook salmon based on the 60-day exposure data. Assuming 80% moisture content, the safe level on a wet weight (ww) basis was determined to be 1.2 microgram Se/g ww (i.e., 0.2*6 microgram Se/g dw)³. A bioaccumulation factor (BAF) for chinook salmon exposed to Se from both water and food simultaneously was not found in the literature. Hermanutz et al. (1992), however, reported a BAF of about 500⁴ (based on wet weight of Se in the whole body) for bluegill exposed to selenium from water and food in an experimental stream environment. Given a safe tissue (whole body) level of 1.2 microgram Se/g ww and BAF of 500, it was estimated that a concentration of 2.4 microgram/L selenium in water would protect freshwater life from selenium toxicity due to its accumulation in tissue.

C. Sediment Concentration Approach

Selenium exhibits a strong association with particulate matter. Hence, the most prominent fate of dissolved Se in an aquatic environment is to bind and complex with particulate matter and settles out. The accumulation of Se in sediment followed by its movement into the food chain is the primary cause of Se toxicity in the aquatic environment. Hence, Canton and Van Derveer (1997) argued in favour of a sediment-based guideline for the protection of aquatic life and wildlife from selenium. Using data collected from streams of the middle Arkansas River basin in Colorado, Van Derveer and Canton (1997) showed that waterborne Se (WSe-microgram/L), sediment Se (SSe-microgram/g dw), and total organic carbon in sediment (TOC-% of dry weight) were highly correlated. The final form of the model, validated with data from other regions, is shown below:

Ln SSe = Ln(WSe*TOC) * 0.657 - 0.877 (1)

From the literature data, Van Derveer and Canton also noted that the 10^th percentile of sediment Se associated with predicted adverse effects is 2.5 microgram/g dw. This is only slightly above the 2 microgram/g dw upper limit reported for uncontaminated sediment and soil. The following table was created from the above relationship, assuming that 2 microgram Se/g in sediment would protect aquatic life from adverse effect of Se bioaccumulation in the food chain.

The above table suggests that the safe level for Se in water varies with the total organic carbon content of the sediment. In a lake with an average TOC concentration of 5%, aquatic life would be protected if selenium in the water were at or below 2 microgram/L.

9.7.2 Marine Aquatic Life

The proposed guideline was determined using the chronic LOEL of 0.01 mg Se/L and a safety factor of 5 as in the case of freshwater life.

Wheeler et al. (1982) reported a 23-35% decrease in the growth of a red alga (Porphyridium cruentum) exposed to 0.01 mg/L selenate-Se over a 14-day period in test marine water containing sulphate comparable to that normally found in seawater (850 mg/L sulphate). A lower decrease (12%) in growth was also observed in a green alga (Chlorella sp.) exposed to 0.01 mg/L selenite-Se in test water with similar characteristics.

Glickstein (1978) reported that a selenium concentration of 0.01 mg/L increased abnormalities in the Pacific oyster (C. gigas) from 9.3% in the control to 12.9%. However, Glickstein's study produced inconsistent results with respect to the dose-response relationship and the ecological significance of the effect (i.e., abnormal development) could not be ascertained from the data.

The interim guideline for sediment and tissue for the freshwater environment (Section 9.7.1) were accepted as interim guidelines for the marine environment. There were not sufficient data for the marine environment to conduct an analysis such as in Section 9.7.1. However, there is some evidence in the literature in support of this argument. For instance, in modeling Se bioavailability to a benthic bivalve from particulate and solute pathway, Luoma et al. (1992) suggested that Macoma balthica ingesting sediment with >1.5 microgram/g dw of selenium in the San Francisco Bay could achieve steady state tissue burdens approaching the level toxic to fish.

9.8 Application of the Aquatic Life Guidelines

As can be seen above, selenium toxicity in aquatic environments will depend on many factors, including its interaction with other elements. Although selenium has been shown to bioaccumulate, the potential for its bioaccumulation through food in the aquatic environments may also vary with several factors (lotic versus lentic environments). For instance, McDonald and Strosher (1998) found that in a fast-flowing river system (lotic environment) Se levels in sediments, algae, insects, and fish tissue increased modestly (2-5 times) despite 100 to 200 fold increase in waterborne selenium below a coal mine. Situations have also be found where fish in certain (e.g., lotic) environments may survive Se levels that have been found to be toxic to fish elsewhere (e.g., lentic environments - Lemly 1993), suggesting higher tolerance to naturally occurring Se (McDonald and Strosher 1998, Kennedy et. al. 2000).

Obviously, in such environments, there is a need to develop site-specific water quality objectives to account for both background water quality conditions and effect concentrations in the local. However, in a watershed, both lotic (fast-flowing water) and lentic (still or slow-moving water) can exist simultaneously. A site-specific water quality guideline protective of the most sensitive environment should be more appropriate in such situations.

2 CCME (1991) recommended a safety factor (SF) of 10 to derive a water quality guideline from chronic data. Instead, a smaller safety factor of 5 was used because selenium is recognized as a substance with a small effect concentration to background concentration ratio. Furthermore, the SF of 10 will yield a guideline that is at the background level (~1 microgram/L) or will fall below it in certain mineralized area (Webber 1996a,b). Another argument in favour of a lower safety factor was based on the Canton (1999) and Canton and Van Derveer (1997) observations. These investigators pointed out that while the 10 microgram/L selenium in the main Belews Lake water was toxic to the fish, no discernable selenium toxicity to fish was apparent in the arm of the lake that contained 5 microgram Se/L.

3 The toxicity threshold of 6 microgram/g dw on whole body weight basis was chosen for the argument presented in this section, because this number was based on a sound statistical procedure that includes not one but all available data from the literature collected at various locations.

4 Using global data and on-step model, Adams et al. (1998) found that water selenium (WS - microgram/L) was correlated with mean egg selenium (MES - mg/kg dry weight) as follows: log (MES) = 3.366+0.561*log (WS). By expressing WS in the units of mg/L, the relationship in its curvilinear form can be written as: MES (mg/kg dw) = 48.2 [WS (mg/L)]^0.561. Further manipulation will yield the following relation:
MES(mg/kg dw) / WS (mg/L) = 48.2 WS^-0.439. This relation suggests that Se bioaccumulation (i.e., ratio MES/WS) in eggs is a function of it concentration in water. At a water concentration of 0.001 mg Se/L, this relation will yield a bioaccumulation factor of about 1000 on dry weight basis or 400 on wet weight basis (considering 80% moisure content). These BAFs are close to 500 used in this section.