Selenium (Se), an essential micronutrient with a narrow therapeutic concentration range (Goyer and Clarkson 2001), is chemically analogous to sulfur and likely follows a similar biogeochemical cycle when released into the aquatic environment (Fan et al. 2002; Young et al. 2010). Dissolved Se exists in either the (IV) or (VI) oxidation states as selenite (SeO3 2−) or selenate (SeO4 2−), respectively (Fan et al. 2002). Studies investigating the effects of these dissolved inorganic Se species on algae, invertebrates, fish (Lemly 1993; Maier et al. 1993; Rosetta and Knight 1995), and waterfowl (Heinz et al. 1988) revealed low toxicity risks. These same studies indicated that toxicity risks increased substantially when aquatic producers biotransformed these inorganic Se-containing compounds into organic Se-containing compounds.

Selenomethionine (SeMet) is the primary organic Se-containing compound formed at the base of aquatic food chains (Young et al. 2010) and is also the most toxic to aquatic life (Heinz et al. 1988; Lemly 1993; Maier et al. 1993; Rosetta and Knight 1995). Se speciation analyses of organisms occupying diverse trophic levels in Kesterson Reservoir (contaminated by agricultural runoff from Se-rich soils; see Ohlendorf 2002) indicated greater SeMet concentrations accumulated in protein stores of top predators than in organisms of lower trophic levels (Fan et al. 2002). This suggested that dietary SeMet exposure was likely responsible for much of the toxicity observed in this system (see Ohlendorf et al. 1988; Hothem and Ohlendorf 1989; Schuler et al. 1990; Fan et al. 2002; Ohlendorf 2002; Janz et al. 2010). The molecular mechanisms responsible for the severity of SeMet toxicity, relative to other Se-containing species, are still unknown (Fan et al. 2002). SeMet toxicity likely results from a combination of improper protein folding and increased cellular oxidative stress, although additional cellular toxicity studies are needed to confirm this hypothesis (Janz et al. 2010).

Amphibians are susceptible to Se exposure in the aquatic environment because they often lay their eggs in Se-contaminated systems, such as coal ash disposal basins or marshes formed from agricultural runoff (Ohlendorf et al. 1988; Rowe et al. 1996, 2002; Hopkins et al. 1998). Amphibian larvae are particularly at risk because they are often confined to small breeding pools with limited dietary resources. Assessing Se toxicity during the larval stage is important because mortality during this stage could negatively affect recruitment to the breeding population (Rowe et al. 2001). Furthermore, studies show that larvae accumulate Se in their tissues and retain it throughout metamorphosis, potentially introducing it into the terrestrial food chain once they emerge as terrestrial or semiaquatic juvenile amphibians (Hopkins et al. 1998; Roe et al. 2005; Unrine et al. 2007).

Much research examining Se accumulation and effects on larval amphibians has focused on Se as a component of trace-element mixtures, such as those present in coal ash, which also contain numerous other elements (see Rowe et al. 1996, 1998, 2001, 2002; Hopkins et al. 1998, 2000; Raimondo et al. 1998; Roe et al. 2005; Unrine et al. 2007). There is a lack of laboratory-based studies specifically addressing the effects of Se contamination on amphibians when other contaminants are not present (Ohlendorf et al. 1988; Fan et al. 2002). The current study was designed to determine the effects of dietary SeMet on a larval amphibian, the Cope’s gray tree frog (Hyla. chrysoscelis). Previous studies revealed that resource limitation, a natural costressor not typically addressed in laboratory toxicity experiments, can augment the toxicities of trace elements to fish and larval anurans (Hopkins et al. 2002; Rowe et al. 2009). To evaluate effects of resource limitation, dietary SeMet was provided in two dietary regimens—food-limited (ration) and ad libitum (ad lib)—using a factorial design. We considered numerous lethal and sublethal end points by which to characterize the toxicities of each SeMet dose and food provision level combination to H. chrysoscelis larvae. Sublethal end points, such as timing of metamorphosis, mass at metamorphosis, and lipid concentrations at metamorphosis, were chosen due to their influence on adult amphibian survival (Scott 1994; Scott et al. 2007). Concentrations of Se accumulated from the diet and retained in tissue through metamorphosis were also quantified.

Methods

Food Preparation and Feeding Regimens

Larvae were fed dehydrated filamentous green algae suspended in a gelatin/agar matrix. Components were combined in a 0.7:1:5:100 g mass ratio of agar, gelatin, dehydrated algae, and deionized water, respectively. Aqueous solutions of seleno-dl-methionine (SeMet; Sigma-Aldrich) were added as each gel mixture cooled to achieve nominal Se concentrations of 50 and 500 μg Se g−1 dw for low or high treatments, respectively (note that based on an avian toxicity study, seleno-dl-methionine may be somewhat less toxic than seleno-l-methionine (Heinz et al. 1996)). Dosing concentrations exceeded total Se concentrations in potential amphibian food sources collected from coal ash basins (6–12 μg g−1 dw (Rowe et al. 2002; Unrine et al. 2007)) but reflect those measured in potential food sources in the Kesterson Reservoir and surrounding habitats in the San Joaquin Valley, CA (20–332 μg g−1 dw (Saiki and Lowe 1987; Schuler et al. 1990; Ohlendorf 2002)). Solidified food mixtures were partitioned into 4-g spherical pellets and stored at −4 °C. The mean caloric contents of food types were similar (Table 1). Each food-type was provided to larvae by way of a ration and an ad lib dietary regimen. Feeding regimens were developed based on estimates of energetic requirements calculated from metabolic rate measurements of 45 larvae not included in this experiment (see methods in Rowe et al. 1998) and measurements of food caloric content by micro-bomb calorimetry (Parr Instrument). This ensured that diets were ample to support survival and growth. Ad lib tanks received one food pellet every 24 h, whereas ration tanks received one food pellet every 48 h. Typically, no food remained in the ration treatments after 48 h, thus confirming that food was indeed limited in these treatments, whereas there was always excess food present in ad lib treatments.

Table 1 Total Se concentrations and caloric content of food
Full size table

Experimental Protocol

Clutches of freshly laid H. chrysoscelis eggs were collected from a pond in Saint Mary’s County, MD, that has no known history of contamination. Eggs were transported to the Chesapeake Biological Laboratory (Solomons, MD) and held in aerated well water until hatch. Larvae were fed control algae/gel food for 1 month before study initiation. Fifteen Gosner stage 25 (GS 25 (Gosner 1960)) larvae were randomly allocated to each of 24 polypropylene tanks containing 4 l aged, ultraviolet-sterilized well water. Four replicate tanks were assigned per treatment for a total of six treatments: (1) control ad lib, (2) control ration, (3) low ad lib, (4) low ration, (5) high ad lib, and (6) high ration (see Table 2). Tanks were constantly aerated, temperature controlled (23.0 ± 1.3 °C), and maintained under a 12-hour light cycle throughout the study. Tanks were cleaned, tank water replaced, and tank water pH (7.4–8.6), dissolved oxygen (5.1–10.4 mg l−1), and conductivity (278.3–342.4 μS) monitored every fourth day. Water samples were collected from each replicate tank, composited, and acidified with concentrated nitric acid (0.5 % of sample volume) before each tank-cleaning to quantify Se dissolution from food to water throughout the experiment.

Table 2 Proportions of individuals to survive to GS 42, GS 46, and to the end of the study (d 78)
Full size table

The experiment was terminated when all larvae in a single treatment had either died or initiated metamorphic climax (maximum 78 days). Replicate tanks were inspected daily and carcasses removed when observed. Survival was monitored until the experiment was complete or until all larvae in that tank died or initiated metamorphic climax. One control ration tank was spilled during tank-cleaning on exposure day 30, leaving three replicate tanks remaining in this treatment.

Changes to craniofacial and gastrointestinal structures prevent feeding during metamorphic climax (developmental stages GS 42 through GS 46). Therefore, individuals were removed and held unfed in individually labeled containers on attaining GS 42 (initiation of metamorphic climax as indicated by forelimb emergence). As individuals completed metamorphosis at GS 46 (after complete resorption of the tail), they were killed by ventral application of an oral anesthetic (Orajel; Church and Dwight Co.) containing 20 % benzocaine (Brown et al. 2004; Torreilles et al. 2009). Carcasses were stored at −80 °C before lipid and Se content analyses. Wet masses of each metamorphosing individual were recorded at GS 42 and GS 46. Snout–to–vent length (SVL) of each GS 46 individual was also measured before it was killed.

Chemical and Lipid Analyses

Total Se concentrations in food and GS 46 individuals were determined on a wet-mass basis using inductively coupled plasma mass spectrometry (ICP-MS; Agilent 4500) after microwave digestion of wet samples in high-purity concentrated nitric acid as described by Rowe et al. (2009). Acidified replicate tank water samples collected throughout the experiment were also analyzed using ICP-MS. Digests were diluted with >18 MΩ deionized water to 10 times the original volume before analysis. Each batch of 10 digests included 8 carcass samples, a blank, and a matrix spike. Every fifth batch included a standard reference material (SRM) (NIST 1566b Se = 2.06 ± 0.15 μg g−1 dry mass) in place of one of the carcass samples. Matrix spike Se recoveries averaged 92.6 ± 13.8 %, and replicate SRM Se recoveries averaged 100.9 ± 13.1 %. Total Se concentrations measured in wet food samples were converted to units of dry mass based on the mean moisture content of each food type (Segawa 2008). Total wet mass–based Se concentrations measured in GS 46 individuals were converted to units of dry mass using the mean moisture content of GS 46 individuals regardless of treatment (Rowe et al. 2011).

Lipids were extracted from GS 46 individuals according to the procedure described by Harvey et al. (2012). Lyophilized carcasses were microwave digested in 35 ml dichloromethane and methanol (CH2Cl2:MeOH [2:1]) at 200 psi and 80 °C for 30 min. After extraction, carcasses were removed from microwave digestion vials, rinsed with solvent, and returned to storage at −80 °C for future Se analysis. Postextraction solvent was evaporated from digestion vials. Evaporated lipid extracts were dissolved in approximately 5 ml CH2Cl2:MeOH (2:1), gravity filtered through glass wool, and transferred to 8-ml amber vials. Transfer solvent was evaporated from lipid extracts under N2 gas. Dried extracts remaining in amber vials were diluted to 1.0 ml with CH2Cl2:MeOH (2:1) and stored at −80 °C until analysis.

Separation and identification of major lipid classes in three to five GS 46 individuals per treatment were performed by thin layer chromatography using an Iatroscan MK-V analyzer equipped with a flame ionization detector as described by Ju and Harvey (2004). Sample extracts were separated into five major lipid classes along the length of a silica-coated glass rod (S-III Chromarods; Mitsubishi Chemical Medicine): phospholipids (PL), sterols (ST), free fatty acids (FFA), triacylglycerides (TAG), and wax esters (WE) from bottom to top of the rod, respectively. Extracts were analyzed in triplicate. A 3-μl aliquot of a lipid standard mixture of known class concentrations was analyzed with each batch of three extracts. Lipid sample chromatograms were integrated using HP ChemStation software. Lipid classes in each sample were identified and quantified based on a calibration curve generated from serial dilutions of the concentrated lipid standard mixture. The concentrated standard mixture (7.08 μg μl−1 total lipid [TL]) contained a representative compound from each lipid class: 4.17 μg μl−1 1,2-dipalmitoyl–sn–glycera–3-phosphocholine (PL), 0.40 μg μl−1 cholesterol (ST), 0.17 μg μl−1 nonadecanoic acid (FFA), 2.15 μg μl−1 glycerol trioleate (TAG), and 0.19 μg μl−1 palmityl state (WE) (Sigma-Aldrich) in CH2Cl2:MeOH (2:1). TL concentrations for each sample were derived through summation of the lipid class concentrations quantified. Mean intrasample precision was ±16.4 % or better for each class, and the average percent recovery of lipid classes in the standards mixture was 109.1 ± 12.7 %.

Statistical Analyses

Mean responses from each replicate tank were statistically analyzed using Minitab 13 statistical software (Minitab, State College, PA). Only replicate tanks with at least one individual to reach GS 42 or GS 46 were included in statistical analyses at these developmental stages (see Table 3 (columns 2 and 7)). All data were tested to ensure satisfaction of model assumptions and transformed if necessary. All statistical differences were evaluated based on α = 0.05. Only one individual from the SeMet low ad lib treatment survived to complete metamorphosis (GS 46); therefore, this treatment was excluded from statistical comparison at this stage of metamorphosis. Both SeMet high treatments, which produced 100 % mortality during the larval period, were excluded from statistical comparisons at all metamorphic stages. Time to forelimb emergence (GS 42), duration of metamorphic climax (GS 42 to GS 46), wet mass at GS 46, and SVL at GS 46 were analyzed for treatment specific differences using general linear model (GLM) analysis of variance (ANOVA) and did not require data transformation. Wet mass at GS 42 required log10 transformation before analysis by GLM ANOVA. Proportions to survive to the end of the study, and proportions to initiate (GS 42) and complete metamorphic climax (GS 46) before the end of the study, were arcsine square root–transformed before analysis by GLM ANOVA. Mean proportions to reach GS 42 and mean overall survival to day 78 in each treatment were calculated based on the total number of replicate tanks per treatment (three for control ration and four replicate tanks for all other treatments). Mean proportions to reach GS 46 in each treatment were calculated based on the number of replicate tanks from which at least one individual reached GS 42 because individuals could not reach GS 46 without first achieving GS 42 (see Table 2 (columns 2, 4 and 6)). The proportions of individuals that died during metamorphic climax and exhibited rear limb deformities were also log10-transformed before analysis by GLM ANOVA. The proportions to display edema at GS 42 were arcsine square root–transformed and analyzed by one-way ANOVA. Total Se body burdens of GS 46 individuals were log10-transformed, and the mean duration of larval exposure was used as a covariate in GLM analysis of covariance to assess differences in accumulation among treatments. Mean total Se concentrations measured in food were also log10-transformed before GLM ANOVA. The limit of detection (1.11 μg Se g−1 dw) was used in place of all “nondetect” values before data transformation. Total Se concentrations in replicate samples of each food type, before and after storage, were analyzed using GLM ANOVA to determine the significance of changes during storage. The relationship between size at GS 42 and survival to GS 46, independent of treatment, was analyzed by binary logistic regression. The within-treatment relationship between wet mass at GS 42 and mortality before GS 46 was also analyzed by binary logistic regression for GS 42 individuals from the low treatments because 100 % of control individuals that initiated metamorphic climax survived to complete it and 100 % of high individuals died before GS 42. General multivariate analysis of variance was used to determine treatment effects on TL content and relative contributions of different lipid classes to TL content of GS 46 individuals.

Table 3 Sublethal endpoints evaluated at Gosner Stage (GS) 42 and GS 46
Full size table

Results

Selenium in Food and Water

Data are listed as means ±1 SE (Table 1). Mean total Se concentrations in food were 50.1 ± 2.8 and 489.9 ± 54.9 μg Se g−1 dry weight (dw) in the low and high treatments, respectively. Mean total Se concentration of control food before storage was lower than the instrumental detection limit (1.11 μg Se g−1 dw). Food was stored at −4 °C for a maximum of 2 months. Se concentrations in SeMet-dosed food were not substantially altered by storage (Table 1; p = 0.4672 and p = 0.6909 for low and high treatments, respectively). A slight increase in the total Se concentration of control food was observed during storage (1.40 μg Se g−1 dw; p = 0.0366). This increase is likely due to sublimation of ice during storage because Se analyses were conducted using thawed, wet samples. After analysis, wet mass–based Se concentrations were converted to Se concentrations on the basis of dry mass as described in methods. The mean total Se concentration of high food was significantly greater than that of low food (p < 0.0001), which was significantly greater than control food (p < 0.0001), both before and after storage (Table 1).

Compared with Se concentrations applied in food, Se dissolution into tank water was minimal. Mean total dissolved Se concentrations were 47 ± 3.0 and 25 ± 2.0 μg l−1 in the high ad lib and ration treatments, respectively, and 10 ± 0.4 and 6 ± 0.3 μg l−1 in the low ad lib and ration treatments, respectively. The maximum total dissolved Se concentrations measured in high ad lib and ration treatments were 53.7 and 28.7 μg l−1, respectively. The maximum total dissolved Se concentrations measured in low ad lib and ration treatments were 15.2 and 8.6 μg l−1, respectively. Both control treatments (ad lib and ration) had the same mean total dissolved Se concentration as that measured in well water before food application (2 ± 0.2 μg l−1). The maximum concentrations of Se measured in control tank water (3.2 and 3.7 μg l−1 in ad lib and ration treatments, respectively) were similar to the maximum background concentration of Se measured in well water (5.0 μg l−1).

Metamorphosis and Overall Survival

Initiation of Metamorphic Climax (GS 42)

Treatment had a significant effect on the average number of individuals to initiate metamorphic climax (GS 42) by the end of the study (p < 0.0001; Table 2). Initiation of metamorphic climax was significantly decreased by exposure compared with control treatments. No individuals survived to reach GS 42 in the high ad lib or ration treatments. Mean percentages to reach GS 42 were similar between low ad lib and ration treatments (p = 0.0971). Significantly fewer individuals reached GS 42 in the low ad lib treatment than in the control ad lib (p = 0.0002) and control ration (p = 0.0102) treatments. Similarly, fewer individuals reached GS 42 in the low ration treatment than in the control ad lib (p = 0.0072) treatment. However, the proportion to reach GS 42 in the low ration treatment was not significantly different from that of the control ration treatment (p = 0.4427).

Survival Through Metamorphic Climax (GS 42–46)

Mean percent survival from initiation of metamorphic climax through completion of metamorphosis (GS 42–46) is also listed in Table 2. These values are based on the number of replicate tanks in which individuals had reached GS 42 by the end of the experiment. There was a strong treatment effect on mean percentages surviving to complete metamorphosis (GS 46; p < 0.0001). Survival to GS 46 was significantly decreased in the low ad lib treatment compared with the control treatments (p < 0.0001 and p = 0.0005 for control ad lib and ration treatments, respectively). Only 25 % of GS 42 individuals survived to GS 46 in the low ad lib treatment. Similarly, survival to GS 46 was decreased in the low ration treatment compared with control ad lib and ration treatments (p = 0.0002 and p = 0.019, respectively). Of low ration individuals to reach GS 42, 60 % survived to GS 46. Binary logistic regression analysis indicated that regardless of treatment, individuals of smaller mass at GS 42 were less likely to survive to GS 46 (p < 0.0001). To statistically determine the likely cause of mortality during metamorphic climax, the same binary logistic regression analysis was applied separately to treatments in which mortality occurred between GS 42 and GS 46. These analyses indicated that individuals of small masses at GS 42 were more likely to experience mortality before reaching GS 46 than those of larger masses in low ad lib and ration treatments (p = 0.034 and p = 0.026, respectively).

Survival to the End of the Study

The high dose was toxic to H. chrysoscelis larvae independent of food provision levels; the high ad lib and ration treatments reached 100 % mortality after 10 and 22 days of exposure, respectively. Overall survival to day 78, which included all individuals that reached GS 46 and individuals remaining as larvae at the end of the experiment, was significantly influenced by treatment (p < 0.0001; Table 2). Survival to day 78 in the control ad lib treatment was significantly greater than in the control ration treatment (p = 0.0247; Table 2). A greater proportion of individuals survived to day 78 in the low ration treatment than in the low ad lib (p < 0.0001), the high ad lib (p < 0.0001), and the high ration (p < 0.0001) treatments (Table 2). Survival to day 78 was not significantly different in the control ration and low ration treatments (p = 0.0718).

Sublethal End Points

Physical Abnormalities and Malformations

Larvae receiving the high dose, regardless of feeding regimen, displayed abnormal red coloration (possibly due to capillary leakage beneath the integument) within 24 h of study initiation. Abnormal coloration began to fade in surviving individuals after 7 days of dosing. Approximately 25 % of larvae from the high treatments also displayed severe edema before death. These abnormalities were not observed in larvae from any other treatment before GS 42. Edema was observed in some individuals from the low treatments at GS 42. One of the 5 low ad lib individuals to reach GS 42 displayed edema. This individual was 1 of the 4 individuals from this treatment to die during metamorphic climax. Four of 15 low ration individuals displayed edema at GS 42. These individuals died before completing metamorphosis. The proportions of individuals to display edema at GS 42 did not differ between the low ad lib and ration treatments (p = 0.690).

Rear limb deformities were observed in some individuals from both the low ration and ad lib treatments (Table 3). The femurs of these individuals appeared to be partially dislocated from the acetabulum of the pelvic girdle, rendering their limbs weak and often nonfunctional. Incidences of rear limb deformities were not significantly different between low ad lib and ration feeding regimens (p = 0.220). On average, 33.3 ± 33.3 % of the individuals to reach GS 42 in the low ad lib treatment developed deformed rear limbs and an average 33.3 ± 33.3 % of these died before completing metamorphosis. In the low ration treatment, an average of 70.4 ± 12.4 % of individuals to reach GS 42 developed rear limb deformities and these individuals died before completing metamorphosis. One of the low ad lib individuals and four of the low ration individuals developed both rear limb deformities and edema on reaching GS 42. These individuals did not survive to complete metamorphosis.

Sublethal Endpoints at GS 42

There was a significant difference (p = 0.036) in mean days of exposure before GS 42 between treatments (Table 3). Individuals in the low ad lib treatment took an average of 27.5 ± 7.8 days after exposure initiation to reach GS 42, which is significantly less than the 41.8 ± 3.5 days averaged in the control ad lib treatment (p = 0.0303). Mean wet mass of individuals at GS 42 differed significantly among treatments (p = 0.003). Individuals to reach GS 42 in the low ration treatment were of a lower mean wet mass than individuals in the control ad lib (p = 0.0019) and control ration (p = 0.0391) treatments. Mean wet mass of individuals at GS 42 in the low ad lib treatment was not significantly different from that of any other treatment.

Sublethal Endpoints at GS 46

The sublethal toxicity end points evaluated for individuals surviving to GS 46 are listed in Tables 3 and 4. Treatment did not significantly affect mean duration of metamorphic climax (GS 42 to GS 46; p = 0.552). Treatment significantly affected the mean wet masses and SVLs of individuals at GS 46 (p = 0.002 and p < 0.0001, respectively; Table 3). control ad lib individuals were of significantly greater mean wet mass at GS 46 than control ration and low ration individuals (p = 0.0448 and p = 0.0015, respectively). Mean wet masses at GS 46 in the control ration and low ration treatments were not significantly different (p = 0.0711). The mean SVL of individuals at GS 46 in the low ration treatment was significantly smaller than the mean SVL of control ad lib and ration treatments (p < 0.0001 and p = 0.0001, respectively). The TL content of GS 46 individuals did not vary by treatment (p = 0.633) nor did concentrations of individual lipid classes (see Table 4).

Table 4 Mean TL and lipid class concentrations in GS 46 individuals
Full size table

Selenium Body Burdens After Metamorphosis

Mean Se body burdens of GS 46 individuals from each treatment are listed in Table 5. There was a significant difference in mean Se body burdens of individuals at GS 46 between treatments (p < 0.0001); however, larval exposure duration did not have a significant influence on these values (covariate p = 0.102). Mean tissue Se concentration in GS 46 individuals from the low ration treatment (80.8 ± 14.4 μg Se g−1 dw) was significantly greater than mean tissue Se concentrations in GS 46 individuals from control ad lib and ration treatments (2.0 ± 0.1 and 2.1 ± 0.2 μg Se g−1 dw; p = 0.0003 and p = 0.0004, respectively). Mean Se body burdens of the control ad lib and ration treatments were not significantly different (p = 0.9868). Only one individual from the low ad lib treatment survived to complete metamorphosis, but it was used during development of the lipid extraction method and therefore was not available for Se analysis. Although tissue concentrations were derived from whole-body analyses, the nonfeeding period of metamorphic climax (3.7 ± 0.3 days in this study) provided a sufficient period to eliminate contributions of gut content.

Table 5 Mean total Se tissue concentrations and larval exposure durations of GS 46 individuals
Full size table

Discussion

Total Se concentrations in food from this study (Table 1) exceeded concentrations measured in periphyton, algae, and macrophytes (6–12 μg g−1 dw) found in coal ash–contaminated basins (Rowe et al. 2002; Unrine et al. 2007) but were similar to Se concentrations (20–332 μg g−1 dw) measured in algae, plankton, aquatic plants, and macroinvertebrates at Kesterson Reservoir, San Joaquin Valley, CA (Saiki and Lowe 1987; Schuler et al. 1990; Ohlendorf 2002). Furthermore, Se doses applied in this study represent a “worst-case scenario” because the Se applied to food was in the most toxic chemical form, SeMet, rather than a less toxic, inorganic form (Heinz et al. 1988; Lemly 1993; Maier et al. 1993; Rosetta and Knight 1995).

Despite amassing detectible body burdens of Se, there was no indication of Se toxicity in control animals. Overall survival and rates of successful metamorphosis were high in both control ad lib and ration treatments (Table 2). This was not unexpected because Se is an essential micronutrient to most organisms (Goyer and Clarkson 2001). Furthermore, control animals from a dietary-dosing study using larvae of a closely related species, H. versicolor, accumulated comparable Se body burdens (2.7 μg Se g−1 dw) after receiving similar concentrations from the diet (1.0 μg Se g−1 dw), but they also did not experience toxic effects (Rowe et al. 2011).

Recently metamorphosed individuals from the low ration treatment did accumulate significantly greater Se body burdens than individuals from both control ad lib and ration treatments. Although loss of the only GS 46 individual from the low ad lib treatment precluded comparisons between mean postmetamorphosis tissue Se concentrations in ad lib and ration feeding regimens, the high Se concentrations retained after metamorphosis by low ration individuals indicate that terrestrial juvenile amphibians could transport potentially toxic concentrations of Se from the aquatic environment into terrestrial food webs (Hopkins et al. 1998; Roe et al. 2005; Unrine et al. 2007). Mean Se concentrations measured in GS 46 individuals from the low ration treatment were approximately 10 times the United States Environmental Protection Agency (USEPA)–proposed concentration for the protection of aquatic life (7.91 μg Se g−1 dw in whole-body fish tissue; USEPA 2004), but they were similar to concentrations measured in Rana catesbeiana livers (45.0 μg Se g−1 dw) collected from habitat neighboring Kesterson Reservoir (Ohlendorf et al. 1988).

Limiting the amount of food provided at each feeding had opposing effects on developing H. chrysoscelis larvae in this study depending on whether food was dosed with SeMet or not. As expected, food-limitation stress in the control ration treatment decreased overall survival to the end of the study and probably contributed to the decreased mean wet mass of individuals at GS 46 compared with control ad lib individuals. However, resource limitation in the control ration feeding regimen was not severe enough to decrease the mean proportion of individuals that survived to complete metamorphosis. Although both high treatments (ad lib and ration) caused equal mortality (100 %), the consequences of rationing the low-dose SeMet–amended food were unexpected. Rationing the low dose of SeMet increased overall survival to day 78 compared with the low ad lib treatment. In addition, apart from a decrease in mean mass at GS 42, the toxicity of the low ration treatment was not significantly different from the control ration treatment based on evaluated sublethal end points. These results are in contrast to those of a similarly designed dietary exposure study investigating vanadium (V) toxicity to larval R. sphenocephala. Larvae that received a rationed diet of V-dosed food exhibited decreased growth rates, survival through metamorphic climax, and TL concentrations, whereas larvae that received the same V-dosed food ad lib did not display these effects (Rowe et al. 2009). Unlike previous studies (Hopkins et al. 2002; Rowe et al. 2009), decreasing the amount of food provided in the low ration treatment likely benefitted individuals by decreasing the total SeMet dose to which they were exposed.

Exposure to increased concentrations of SeMet caused physical abnormalities (abnormal coloration, edema, and rear limb deformities) to develop throughout the larval period, and exposure to the high dose was lethal to larvae, under both the ad lib and ration feeding regimens, before initiation of metamorphic climax. Abnormal coloration displayed by larvae in the high treatments may be attributed to capillary hemorrhaging beneath the integument because similar symptoms were observed in gill lamellae of Se-exposed fish (Lemly 2002). Abnormal coloration was not observed in larvae from the low treatments, indicating that this dose was not sufficient to elicit this particular response. The edema observed in high larvae and low GS 42 individuals might be a nonspecific response to Se exposure (Janz et al. 2010), similar to that observed in Se-exposed larval and adult fish and embryonic birds (Lemly 1993, 2002; Ohlendorf 2002; Hamilton 2004). Although dissimilar from the spinal and craniofacial malformations observed in developing anuran larvae from coal ash–contaminated sites (Rowe et al. 1996; Raimondo et al. 1998; Hopkins et al. 2000), rear limb malformations exhibited by some individuals in our study share similarities with rear limb malformations observed in two R. clamitans larvae from a laboratory-based coal ash–exposure study conducted by Snodgrass et al. in 2004. These individuals developed shortened femurs that also appeared to be dislocated from the pelvic girdle, but this was not associated with coal ash exposure (Snodgrass et al. 2004). In our study, rear limb deformities were only associated with SeMet exposure in low treatments (no individuals in the high treatments survived to initiate metamorphosis). Mortality before GS 46 was strongly associated with the presence of rear limb deformities. These results indicate a relationship between rear limb deformities in H. chrysoscelis larvae, mortality during metamorphosis, and exposure to increased concentrations of SeMet.

All individuals that initiated metamorphic climax in the control ad lib and ration treatments survived to complete it, whereas mortality was observed during metamorphic climax in both the low ad lib and ration treatments. SeMet exposure may have prevented larvae from amassing sufficient energy stores to sustain them through metamorphosis, or exposure may have increased energetic demands beyond those typically associated with metamorphosis. This hypothesis is further supported by the observation that individuals from the low ration treatment were of significantly smaller wet mass at the beginning of metamorphic climax (GS 42) than individuals from other treatments. Low ad lib individuals reached GS 42 faster than control ad lib individuals. Larval period plasticity is a life history strategy evolved by many anuran species to increase survival by rapidly achieving the terrestrial life stage. By decreasing the time spent in the fully aquatic life stage, exposure to aquatic predators may be limited, desiccation risk in ephemeral pools might be decreased, and competition for resources might decrease (Wilbur and Collins 1973). Therefore, if the toxicants invoke physiological responses similar to those imposed by natural stressors, it is also possible that rapid metamorphosis might result in decreased exposure to toxicants in the larval habitat.

At metamorphic completion (GS 46), individuals from resource-limited treatments were of significantly decreased mean wet mass compared with individuals fed ad lib. Field studies with salamanders demonstrated a strong relationship between size at metamorphosis and adult survival to reproductive age (Scott 1994). Assuming this trend is also true for H. chrysoscelis, our results indicate that resource-limited individuals may be less likely to reach reproductive maturity. However, based on GS 46 lipid measurements, a resource-limitation effect on survival to adulthood seems less likely. Lipids are the primary energy stores used during nonfeeding periods of metamorphic climax and early terrestrial life (Fitzpatrick 1976; Scott et al. 2007) and thus may be a better determinant of adult success than body mass at GS 46. Although a study conducted by Thomas and Janz (2012) revealed that SeMet exposure increased triacylglyceride stores in zebrafish, neither SeMet exposure nor resource limitation altered concentrations of specific lipid classes or TL concentrations in H. chrysoscelis individuals at GS 46. In fact, lipid class concentrations measured in this study agreed with those reported for recently metamorphosed R. tigrina collected from the wild (Sawant and Varute 1973). Lipid concentrations were only determined for individuals that reached GS 46 in our study. Therefore, additional research determining the correlation between lipid stores amassed before GS 42 and metamorphic success are recommended.

Conclusion

This study was designed primarily to determine the potential toxic effects of SeMet in a representative species of an understudied taxonomic group. The results of this work established that larval exposure to relatively high dietary SeMet concentrations (50.1 and 489.9 μg Se g−1 dw) have the potential to induce rear limb malformations and edema in H. chrysoscelis larvae. In addition to inducing physical malformations, SeMet exposure decreased survival through the larval period, decreased the number of individuals to initiate (GS 42) and complete metamorphic climax (GS 46), and decreased wet mass and SVL at GS 46. Contrary to expectation, resource limitation did not augment the toxicity of the SeMet doses applied. Further investigation is warranted to determine if lower dietary SeMet concentrations would elicit the severe physical malformations, decreased metamorphic success, and overall survival observed in this study. We also recommend that studies be conducted to investigate the toxicity of dissolved SeMet on amphibians because dissolved concentrations as low as 5 μg Se l−1 can lead to accumulation and toxicity in fish (Lemly 2002).