10.1 Toxic Effects
The toxicity data for wildlife are summarized in Table 10.01. Studies conducted in the Kesterson Reservoir by Ohlendorf et al. (1986a and 1988) revealed that water entering the Kesterson ponds, from the contaminated San Luis Drain, had average levels of 0.3 mg Se/L. The concentration of selenium in plants, invertebrates and fish collected from the ponds ranged from 22-175 µg Se/g. The analysis of wildlife inhabiting the area revealed 16.7% dead embryos in the nests of mallards (Anas platyrhynchos) and 59.3% dead embryos in the nest of American coots (Fulica american). The authors also noted that no nests were found in 1984 to 1985 in an area where 92 nests had been found in 1983.
Fairbrother and Fowles (1990) reported that 3.5 mg/L selenite-Se in the drinking water of mallards did not affect immune function, but it increased the serum alanine amino-transferase (ALT - indicative of renal or to a lesser degree hepatic dysfunction in birds) level. However, 2.2 mg Se/L as selenomethionine appeared to affect certain aspects of mallard immune response system, including a suppression in delayed type hypersensitivity (DTH), increased activity of serum ALT, and an increased plasma glutathione peroxidase (GPX) activity.
Peterson and Nebeker (1992) modeled thresholds for waterborne selenium toxicity to aquatic birds and mammals. The model considered the duration of time the birds spent in the contaminated areas, the proportion of their diet comprised of foods from contaminated aquatic environments, trophic position, food ingestion rates (function of body mass), and energy demands associated with activities such as reproduction. The toxicity thresholds for the marsh wren (Cistothorus palustris), belted kingfisher (Ceryle alcyon), osprey (Pandion haliaetus), bald eagle (Haliaeetus leucocephalus) and the mallard were estimated to be 0.8, 0.9, 1.3, 1.9, and 2.1 µg/L dissolved selenium, respectively. The toxicity thresholds for mammals in the same environment were 0.9 µg Se/L for bats and shrews, 1.1 µg Se/L for mink, and 0.7 µg Se/L for river otters. Lemly and Smith (1987), DuBowy (1989), and Skorupa and Ohlendorf (1991) also suggested that waterborne selenium (dissolved) concentrations between about 1 and 3 µg/L were protective of aquatic birds.
Heinz et al. (1989) observed those mallard ducklings on control diets with no added selenium, but born to parents who were fed 8 to 10 µg/g selenomethionine-Se (dry-weight) in their diet, had a significantly reduced survival. Females fed control diets with no added selenium had an average of eight ducklings that lived to 6 days of age (at which time all ducklings were sacrificed) compared to females fed 8 µg/g of selenomethionine-Se, who had an average of 4.6 young. There were no young born to females fed 16 µg/g selenomethionine-Se. Also, diets containing 8 and 16 µg/g (dry-weight) Se as selenomethionine caused malformations in 6.8% and 67.9%, respectively, of unhatched eggs compared with 0.6% for controls. The authors concluded that the dietary threshold necessary to impair reproduction in mallards is between 4 and 8 mg/g (dry-weight) selenium as selenomethionine. These investigators also suggested that when eggs from a wild population contain 1 µg Se/g (wet-weight), reproductive impairment may be possible and should be evaluated in that population. At 5 µg Se/g (wet-weight) in eggs, reproductive impairment is much more likely to occur.
Heinz and Fitzgerald (1993a) reported 25% mortality in mallards exposed to 20 µg Se/g (dry-weight) as selenomethionine for 16 weeks and 95% mortality at twice the concentration. These investigators also observed that, after one week of exposure, the 20 µg Se/g treatment compared to the control or the 10 µg Se/g treatment significantly depressed the body weight of the birds. However, four weeks after being returned to an untreated diet, the body weight of birds fed 20 µg Se/g had increased to the point of being statistically inseparable from the weight of the control birds. Finally, the concentrations of selenium in their blood were related to dietary selenium, but mortality was not clearly related to a threshold concentration of selenium in blood.
Heinz et al. (1987) compared the toxic effects of selenomethionine and selenite-Se at 10 µg/g (dry-weight) in mallards. Whereas selenite caused mainly embryotoxic effects, selenomethionine was mostly teratogenic, causing hydrocephaly, bill defects, eye defects, and foot and toe defects. They also noted that males accumulated more selenium because females can eliminate some in their eggs. Heinz (1993) exposed adult male mallards to a diet of 15 µg Se/g, as seleno-DL-methionine, for 21 weeks. After this initial exposure, the mallards were fed untreated food for 12 weeks and then were re-exposed to the selenium at 100 µg/g of diet for five weeks. The author observed that the earlier (i.e., 15 µg Se/g) exposure of mallards to selenium had no effect on the birds' response to the later (i.e., 100 µg Se/g) treatment. It was concluded that one should be concerned primarily about the current exposures of birds to selenium and not the previous exposures.
Raptors may be exposed to excessive selenium from contaminated prey. Wiemeyer and Hoffman (1996) exposed captive eastern screech owls (Otus asio) to diets containing 0 (control), 4.4, and 13.2 mg/kg (wet-weight) added selenium as seleno-DL-methionine. Adult mass at sacrifice and reproductive success of birds receiving 13.2 mg Se/kg were significantly depressed compared to the controls. Parents fed 4.4 mg Se/kg produced no malformed nestling, but the femur lengths of the young were shorter than those of the controls. Also, the 5-day-old nestling from parents fed 4.4 mg Se/kg indicated oxidative stress in the liver (indicator for the onset of Se-related toxicosis), including a 19% increase in glutathione peroxidase activity, a 43% increase in the ratio of oxidized glutathione to reduced glutathione, and a 17% increase in lipid peroxidation.
Selenomethionine is a major form of selenium in some plant material such as wheat seeds and soybeans. In nature, selenomethionine is found almost exclusively in the `L' form, which is one the two stereoisomer forms it can take. The other stereoisomer, the `D' form, is found in equal amounts in manufactured selenomethionine (Heinz and Hoffman 1996). In comparing the toxicity of various forms of selenium on reproduction of mallards, Heinz and Hoffman (1996) reported that seleno-DL-methionine and seleno-L-methionine were equally toxic; however, both were more toxic than selenium from selenized yeast. Also, regardless of the source, a diet of 10 mg Se/kg (dry-weight) posed a serious threat to the reproductive success of mallards.
In their assessment of Se transfer in the food chain to avian wildlife species at 15 sites, Adams et al. (1998), Skorupa and Ohlendorf (1991) and Ohlendorf et al. (1993) reported a strong correlation between water and mean egg Se (or MES) concentrations. Adams et al. (1998) used global data and suggested that a concentration of 6.8 microgram/L selenium in water was associated with a threshold concentration of 20 microgram/g dw in eggs (MES) for reproductive effects in birds. They also concluded that the proposed concentration of 6.8 microgram Se/L would protect 90% of the sites and species from the adverse effects of selenium. The threshold MES of 20 microgram/g dw in eggs from Adams et al. (1998), was higher than the contamination threshold concentrations5 of 3 microgram/g dw in eggs from Skorupa and Ohlendorf (1991) and 8 microgram/g dw in eggs (impaired egg hatchability) from Ohlendorf and Sontolo (1994). This difference in the threshold concentrations can be attributed to different endpoints chosen to be protected and the models used by the investigators. In a recent publication, based on data published in the literature, Fairbrother et al. (1999) reported an EC10 of 16 microgram/g dw in eggs (MES) and an EC50 of 21 microgram/g dw in eggs (MES), using the most sensitive endpoint of chick mortality. Again, this differs from the threshold of 6 microgram/g dw in eggs proposed by Skroupa (1998).
10.2 Summary of Existing Guidelines
Water quality guidelines for the protection of wildlife (excluding aquatic life) were not found in the literature.
10.3 Recommended Guidelines
To protect wildlife it is recommended that the maximum concentration of total selenium should not exceed 0.004 mg Se/L.
Selenium in eggs is the prime indicator of its toxicity to avian species. It is recommended that a concentration of 1.4 mg Se/kg wet weight in eggs (i.e., 7 mg Se/kg dry weight6, assuming 80% moisture content) be used as an alert level. When average concentration of selenium in avian eggs approaches this level, a detailed site-specific investigations should be undertaken to assess the effects of selenium in the environment.
10.4 Rationale
The proposed guideline was based on the toxicity thresholds of birds exposed to selenium in water and food in aquatic systems. Using global data, Adams et al. (1998) determined that 0.0068 mg Se/L will protect 90% of avian species exposed to selenium. This corresponds to an average concentration of 0.020 mg/g selenium (dry weight) in the eggs of the avian species. This concentration is slightly greater that the threshold concentration (EC10) of 0.016 mg/g dw in eggs reported by the same group of investigators (Fairbrother et al. 1999). Bird eggs from selenium-normal environments rarely average greater than 0.003 mg/g dw (mean egg selenium or MES) (Skorupa and Ohlendorf 1991). Using the Skorupa and Ohlendorf (1991) model relating mean egg Se and water Se concentrations, Adams et al. (1998) determined that an upper 95% confidence limit of 0.0023 mg Se/L in water will correspond to 0.003 mg/g dw MES. The proposed guideline of 0.004 mg Se/L to protect avian wildlife from the adverse effects of selenium is an average of the two values (0.0023 and 0.0068 mg Se/L).
Peterson and Nebeker (1992) modeled a threshold of about 0.001 mg/L in water for birds and mammals that bioaccumulate selenium by feeding on contaminated aquatic life. Peterson and Nebeker's data were not accepted for the guideline derivation purposes, since the assumptions used in the model could not be verified. For instance, it was assumed that selenium accumulates in wildlife foods in a general and predictable manner, and that wildlife consumes only contaminated foods. Because of the uniform treatment (i.e., assumptions used) of all wildlife in their model, we accepted their proposition that the protective concentration for all wildlife (birds and mammals) was similar in magnitude.
5 Contamination threshold refers to concentration that may be found in naturally occurring selenium environments. It is also called as `natural background index'.
6 Geometric mean of EC10 concentration of 16 microgram Se/g dw in eggs (Fairbrother et al. 1999) and egg Se concentration of 3 microgram/g in contaminated-free environment (Skorupa and Ohlendorf 1991).