Enantioselective toxicity and bioaccumulation of fipronil in fathead minnows (Pimephales promelas) following water and sediment exposures

Authors


Abstract

Fipronil is a widely used, broad-spectrum pesticide that is applied as an equal mixture of two enantiomers. As regulations on older pesticides become more stringent, production and application of fipronil is expected to grow, leading to increased inputs into aquatic environments and complex exposures to biota. To better understand the potential exposures introduced by fipronil contamination, we conducted subchronic toxicity tests with larval fathead minnows (Pimephales promelas) and waterborne fipronil and its enantiomers and exposed juvenile fathead minnows to fipronil-spiked sediment. Enantioselective toxicity was observed in fish after the 7-d subchronic exposure, with increased toxicity of the racemate and (+) enantiomer observed compared with the (−) enantiomer. Curiously, toxicities of the racemate and (+) enantiomer were not significantly different, even though the racemate contains 50% of the (+) enantiomer and 50% of the less toxic (−) enantiomer. During the sediment exposure, racemic fipronil in sediment was transformed primarily to fipronil sulfide, while exposed fish rapidly accumulated fipronil and/or fipronil sulfide and transformed the majority to fipronil sulfone. Using the results of the sediment-exposure experiment, the authors explored a mechanism that may contribute to the interesting trends in enantioselective toxicity observed during the waterborne exposures. In tandem, the aquatic toxicity experiment and the spiked sediment exposure demonstrate the potentially complex behavior of fipronil in sediment and fish. Environ. Toxicol. Chem. 2013;32:222–227. © 2012 SETAC

INTRODUCTION

Pesticide contamination in surface waters and sediment is widespread in the United States. In a 10-year nationwide study, pesticides were detected in 97% of streams sampled in agricultural and urban areas, and pesticide concentrations regularly exceeded water-quality benchmarks for aquatic life 1. Carbofuran and diazinon were two of the most frequently detected insecticides in streams 1, despite their restricted use (http://www.epa.gov/oppsrrd1/REDs/factsheets/carbofuran_ired_fs.htm, http://www.epa.gov/oppsrrd1/REDs/factsheets/diazinon_ired_fs.htm).

Fipronil is a chiral insecticide that is applied increasingly as a replacement for carbofuran and diazinon. In invertebrates, fipronil causes toxicity by interfering with the γ-aminobutyric acid–gated chloride channels of nerve cells and has a strong affinity for invertebrate γ-aminobutyric acid receptors compared with mammals 2, 3. This selectivity contributes to fipronil's popularity because it minimizes toxic effects in humans and domesticated animals at doses that are effective on insects. However, fipronil presents its own complex environmental problems.

Fipronil is produced as a racemate with two enantiomers that are designated (R-), or (−), and (S+), or (+), based on the molecule's three-dimensional structure and rotation of plane-polarized light 4, 5. Although fipronil is applied as a racemate, biological processes within organisms or in the environment can alter its enantiomeric fraction (EF), resulting in enrichment in one enantiomer as the other is selectively biotransformed 6, 7. Therefore, environmental exposures may be from mixtures enriched in either enantiomer, even though fipronil is applied as a racemate. Not long after it was introduced in the United States, fipronil received attention for its toxicity to nontarget organisms after it was implicated in the low crayfish harvest in Louisiana in 1999 8. Subsequent laboratory studies confirmed the acute toxicity of waterborne fipronil to crayfish 9, 10, and further investigation revealed that crayfish are significantly more sensitive to the (+) enantiomer in acute exposures 10. Although acute enantioselective toxicity of waterborne fipronil is well documented 10–12, fewer studies have investigated the chronic toxicity of fipronil, and studies of chronic enantioselective toxicity are even more rare.

Fipronil has been detected in surface water 8, regardless of its hydrophobic characteristics (average log KOC of 2.9 8); and long-term exposure in the environment is likely from sediment-borne fipronil. Microbial degradation in sediment can alter the EF, resulting in sediment concentrations enriched in one enantiomer over the other 7. Further complicating the assessment of fipronil in aquatic environments are its multiple degradation pathways; in sediments and surface waters fipronil may undergo photolysis to desulfinyl fipronil, oxidation to fipronil sulfone, or reduction to fipronil sulfide 7, 13. Although laboratory exposures (e.g., waterborne exposure of racemate, food spiked with the racemate or one of the enantiomers) do not adequately capture the complex exposures that occur in the environment, bioaccumulation of fipronil has been documented via diet in fish 14 and sediment exposure in benthic aquatic macroinvertebrates 15, 16. However, a paucity of data exists on bioaccumulation of fipronil from sediment, and no data are available for fish exposed in a sediment–water system.

The main goal of the present study was to fill the data gaps to better understand exposure to and potential effects of fipronil in aquatic environments. Racemic fipronil is introduced to the aquatic environment via runoff into surface waters and persists in the sediment; therefore, two important components of the aquatic toxicity of fipronil are (1) the enantioselective toxicity of waterborne fipronil and (2) bioaccumulation of fipronil from sediment. We conducted two sets of experiments with fathead minnows (Pimephales promelas) to examine these two components. First, we conducted standardized toxicity tests to evaluate the acute and subchronic enantioselective toxicity of waterborne fipronil to larval fathead minnows. Next, we exposed juvenile fathead minnows to fipronil-spiked sediment for 42 d to determine if they bioaccumulate fipronil from sediment exposure. In addition, we quantified major fipronil metabolites in fish, sediment, and water over the course of the exposure.

MATERIALS AND METHODS

For all experiments, fish were obtained from the U.S. Environmental Protection Agency (U.S. EPA), Region 5 Laboratory, Cincinnati, Ohio. Fathead minnows were chosen for these experiments because of their use as a standard U.S. EPA toxicity test organism and their close association with sediments, which includes sediment ingestion 17. All research in the present study was conducted by protocols approved by the University of Georgia's Institutional Animal Care and Use Committee. Fipronil (± 5-amino-1-[2,6–dichloro–4–(trifluoromethyl)–phenyl]-4[(trifluoromethyl)-sulfinyl]-1H-pyrazole-3-carbonitrile, 97.8% pure), fipronil sulfone (98.2% pure), and fipronil sulfide (95% pure) were purchased from ChemService. The two fipronil enantiomers were separated by Chiral Technologies from the fipronil racemate as described in Konwick et al. 11.

Aquatic toxicity tests

Toxicity tests were conducted in triplicate according to U.S. EPA protocol (method 1000.0 18). Larval fathead minnows were exposed to the fipronil racemate, (+) enantiomer, or (−) enantiomer for 7 d with daily 80% renewal of test solutions. Fish were fed newly hatched Artemia nauplii ad libitum in the evening prior to test initiation and twice daily during the exposure. Fish were not fed during the final 12 h of exposure 18. Based on a range-finding exposure test (data not shown), stock solutions of fipronil racemate and enantiomers were prepared in acetone and diluted in moderately hard water to 50, 100, 200, 400, and 800 µg/L. Acetone concentrations were held constant at 0.1% of final solution volume, and a vehicle control (0.1% v/v acetone in moderately hard water) was tested to evaluate the potential effects of acetone toxicity. Fresh test solutions were prepared daily with moderately hard water. Larval fish were exposed to fipronil in 600-ml glass beakers, with 300 ml of test solution and 10 fish per beaker. We used four replicates (beakers) per concentration and 12 vehicle control replicates. Test vessels were randomly arranged in an incubator and maintained at 25 ± 1°C with a photoperiod of 16 h light and 8 h dark. Water quality (DO [YSI, model 55], pH, and temperature [Orion, model 290a]) was monitored twice daily in one replicate per concentration to minimize disturbance to test organisms. Survival was monitored daily, and dead organisms were removed from test vessels. After 7 d of exposure, surviving larval fathead minnows were killed with buffered MS-222 (Argent Chemical Laboratories), rinsed with deionized water, and placed by replicate on preweighed aluminum pans (FisherBrand; Thermo Fisher Scientific). Fish were dried for 24 h at 60°C (Thelco Laboratory Oven; Thermo Fisher Scientific) and then weighed to the nearest 0.01 mg.

Subsamples of test solutions were analyzed to confirm fipronil exposure concentrations. Water samples were collected daily from test beakers and analyzed separately for confirmation of fipronil concentrations during the first toxicity test. In the second toxicity test, two sets of composite water samples were collected per concentration after fipronil renewal by compositing samples from day 0 to 3 and 3 to 6. A single sample was composited for all 7 d for each chemical form and concentration during the third toxicity test. For determination of fipronil concentrations, we used solid phase extraction followed by GC–MS (HP 6890/5973) with a chiral column (BGB-172 Analytik). Analytical procedures are described in detail in Konwick et al. 11. Average recovery of fipronil spiked into water at 100 µg/L was 99 ± 0.1% (n = 3).

Median lethal values (LC50) and associated 95% confidence intervals for fathead minnows exposed to racemic fipronil, (+) enantiomer, and (−) enantiomer were calculated using the trimmed Spearman-Karber method (ToxStat 3.3; University of Wyoming, Laramie, WY, USA). We first calculated LC50s for each test repetition for each form of fipronil; if LC50s were not significantly different (e.g., overlapping confidence intervals), data were combined and analyzed as a single data set to compare toxicities of the fipronil racemate, (+) enantiomer, and (−) enantiomer. If confidence intervals did not overlap, LC50s were considered significantly different 19. Average dry weights of surviving fish were tested for normality (Shapiro-Wilks test) and homogeneity of variance (Bartlett's test). Growth (dry wt) of surviving fish was regressed against exposure concentrations (SAS 9.1; SAS Institute, Cary, NC, USA) to evaluate toxicity trends. To combine tests within each chemical form, weights were expressed as percentages of control weight within each test. Despite a minor residual pattern for the racemate, a linear model of the weight as a percentage of control versus concentration was deemed both appropriate (according to residual plots) and useful (r2 > 0.5). Transformation of the concentration data did not correct the minor residual pattern observed for the racemate, nor did the transformation produce different results versus untransformed data.

Chronic sediment exposure and bioaccumulation test

Sediment was collected from a freshwater pond at the U.S. Department of Agriculture field research station near Watkinsville, Georgia, USA. Sediment from this site was previously characterized by Jones et al. 7 as actively methanogenic with a pH of 6.7, 4% (w/w) total organic carbon, and 0.30% (w/w) total nitrogen. Fipronil was not detected in sediment from this site. Sediment was stored in 5-gallon sealed, plastic buckets with an N2 headspace and held in the dark at 4°C prior to sediment spiking. Prior to spiking and aging, pond sediments were combined and thoroughly mixed with a paint mixer drill for 30 min to ensure homogeneity. Sediment was then divided equally and placed into containers for preparation of control and fipronil-spiked sediments. Fipronil was dissolved in acetone and slowly added to one sediment batch while mixing under a continuous stream of nitrogen gas to maintain anoxic conditions. The target fipronil concentration was 5 mg/kg, and the spiked sediment was mixed for 24 h at 4°C. Although this concentration is 1,000 times the highest reported sediment concentration for fipronil 9, higher concentrations are needed for measuring bioaccumulation by fathead minnows. Control sediment (vehicle control) was prepared by addition of acetone. After the 24-h mixing period, sediment containers were sealed in an N2 atmosphere and sediments were aged for six months at 4°C to allow for maximum sorption of fipronil to the sediment. After sediment aging and prior to test initiation, samples of control and fipronil-spiked sediment were analyzed for concentration of fipronil, transformation products, and determination of fipronil EF.

Juvenile fathead minnows ranging in age from 101 to 120 d were used in the fipronil-exposure/bioaccumulation experiment. Fish were acclimated to laboratory conditions in flow-through laboratory tanks containing no sediment for two months prior to testing. During the holding period, fish were fed twice daily and water quality was monitored daily. Experimental and holding period conditions were as follows: pH 6.9 to 7.5, temperature 18.9 to 24.0°C, DO >4.0 mg/L, photoperiod = 16 h light/8 h dark.

Approximately 2.5 L of control or fipronil-spiked sediment were separately added to three 10-gallon aquarium tanks (replicates) per treatment. Test water (18 L of dechlorinated, filtered [5 µm], ultraviolet-treated tap water) was added to each tank slowly to minimize sediment disturbance. Particulates were allowed to settle overnight. The following day, prior to test initiation, 15 fish were killed for analysis (time 0). The remaining 58 minnows were randomly added to each tank. Ammonia, pH, DO, and temperature were monitored three times daily for the first week, after which water quality was monitored twice daily. To maintain water quality, 2 L of test water were exchanged with clean test water twice a week for the first two weeks and once a week thereafter. Fish were fed flake food (Tetrafin; Aquatic Eco-Systems) ad libitum twice daily. On sampling days, fish were not fed within 8 h of sampling.

Fish, sediment, and water were sampled from each tank within each treatment (control and fipronil-spiked sediment) on days 0, 7, 14, 21, 28, 35, and 42. Water (200 ml) and sediment (15 g) samples were collected in glass vials and stored overnight at 4°C prior to analysis. Typically, 10 fish were sampled from each replicate. On day 35, nine fish were removed from one control tank and eight or nine fish were removed from the three fipronil-exposure tanks. On the final day of the experiment, all remaining fish were removed from control tanks (one from the first tank, zero from the second tank, and six from the third tank) and fipronil-exposure tanks (seven from the first tank and five each from the second and third tanks). Collected fish were sorted by treatment and replicate and maintained overnight in clean test water to allow them to empty their gut contents to exclude any fipronil sorbed onto sediment within their gastrointestinal tracts from measurement of fipronil body burdens. Fish were killed with an overdose of buffered MS-222, rinsed with deionized water, placed in plastic storage bags, and transported on ice to the U.S. EPA Office of Research and Development, Ecosystems Research Division in Athens, Georgia, USA for chemical analysis.

Samples of overlying water, sediment, and fish were analyzed for concentrations of both fipronil enantiomers, fipronil sulfide, and fipronil sulfone. Water samples were extracted by solid-phase extraction, and analyte concentrations were determined by gas chromatography–mass spectrometry (GC–MS; HP 6890/5973) using a chiral column (BGB-172 Analytik) as described in detail in Konwick et al. 11. Average recovery of fipronil spiked into water at 100 µg/L was 99 ± 0.1% (n = 3). The minimum detection limit of fipronil in water was 1 µg/L. Fish tissue and sediment samples were analyzed according to Konwick et al. 14. A recovery standard, 2,3,5,6-tetrachlorobiphenyl (PCB 65), was added to the fish and sediment samples prior to extraction to yield a concentration of 0.5 mg/L in the final extract used for GC-MS analysis. Samples (sediment or freeze-dried fish carcasses) were homogenized by sonication twice in methanol (10 ml for sediment and 5 ml for fish). After centrifugation, the methanol was separated, combined, and evaporated to 5 ml (fish) or 2 ml (sediment). This methanol extract was then combined with distilled water (25 ml for fish and 10 ml for sediment) and extracted with methyl-tert-butyl ether (5 ml for fish and 3 ml for sediment). This final extract was then centrifuged to separate as much methyl-tert-butyl ether as possible from the methanol–water layer, and the methyl-tert-butyl ether was analyzed by GC–MS as described above for the water samples. Fish and sediment extract concentrations were corrected to PCB 65 recovery, which averaged 46 ± 2% (mean ± standard error [SE]) over all fish samples and 54 ± 2% (mean ± SE) over all sediment samples. The method detection limits of fipronil, fipronil sulfide, and fipronil sulfone in both fish and sediment were 20, 50, and 50 ng/g, respectively. The EF of fipronil was calculated as EF = [E+] /([E+] + [E−]), where [E+] and [E−] were determined to be the concentrations of the first and second eluting enantiomers, respectively 7.

We used analysis of variance (ANOVA) followed by Tukey's mean separation test (α = 0.05) to determine significant differences among chemical types (fipronil, fipronil sulfide, and fipronil sulfone) in fish (n = 3) and sediment (n = 3) within each time point and significant differences within chemical types (in fish and sediment) across time points. Proportional EF data were arcsine-transformed, and significant differences of EF between fish and sediment were determined with a t test at each time point. Significant differences of EF within fish and sediment across time points were analyzed with ANOVA followed by Tukey's mean separation test. All ANOVA and t test procedures were conducted with SAS (Ver 9.1). The z statistic, z = (EFmean – 0.50)/σEF, was calculated to determine if measured EFs were significantly different from 0.50. A z statistic >1.96 indicated EFs significantly >0.5.

RESULTS

Aquatic toxicity tests

Daily water quality in toxicity test vessels remained within acceptable values 18 during the three tests. In all test vessels monitored, DO was >4 mg/L, pH ranged 7.30 to 8.00, and temperature was maintained at 25 ± 1°C. Analysis of fipronil solutions from tests 1 and 2 showed that average measured concentrations were within 5% of target concentrations. Although test 3 concentrations deviated from the target concentrations, the LC50s for test 3 calculated with measured concentrations were not significantly different from those produced with nominal concentrations. Furthermore, when comparing fish weights from the three tests by concentration, the average weight of fish from test 3 was similar (i.e., within 1 standard deviation) to fish weights from both tests 1 and 2 (data not shown). Because of the similarity between measured and nominal concentrations in tests 1 and 2 and the consistency of results across all tests within each chemical type, nominal concentrations were used for all statistical analyses.

Fathead minnow mortality increased with increasing concentrations of racemic fipronil and each enantiomer. After 96 h, there were no significant differences in acute toxicity for fipronil racemate or the two enantiomers; LC50 values ranged from 446 to 448 µg/L. After 7 d, the LC50 and 95% confidence intervals were 208 (191–224) µg/L for the racemate, 227 (201–243) µg/L for the (+) enantiomer, and 365 (333–397) µg/L for the (−) enantiomer. The LC50 of the (−) enantiomer was significantly higher than those of the racemate and the (+) enantiomer. Weights of fish exposed to racemic fipronil and both enantiomers decreased with increasing chemical concentration. Linear regression of weights (expressed as a percentage of control weight) and concentration revealed a significantly lower slope (nonoverlapping 95% confidence interval of slope estimate) for the (−) enantiomer compared with similar slopes (overlapping 95% confidence interval of slope estimate) for the racemate and (+) enantiomer (Fig. 1).

Figure 1.

Concentration–response relationships for growth of Pimephales promelas exposed to fipronil racemate, (+) enantiomer, and (−) enantiomer. Each point represents the mean mass relative to controls (percentage of control weight) ± one standard error from three replicate tests. The slopes and 95% confidence intervals for the racemate, (+) enantiomer, and (−) enantiomer were −0.00276 (−0.00322 to −0.00223), −0.00287 (−0.00328 to −0.00247), and −0.00196 (−0.00216 to −0.00176), respectively. The slope of the (−) enantiomer was significantly lower than the slopes of the racemate or (+) enantiomer, implying reduced toxicity of the (−) enantiomer for fish growth.

Sediment exposure and bioaccumulation test

Fish exposed to fipronil-spiked and control sediments experienced sporadic mortality, which never exceeded 12%. Mortality in the three fipronil-spiked tanks was 3, 9, and 9%, and mortality in the control tanks was 3, 7, and 12%. Mortality in all tanks never exceeded the chronic limit of 20% 18. Water quality was similar in all experimental tanks. Temperatures ranged from 19.6 to 23.1°C, with 24-h temperature changes not exceeding ± 1°C. At all times DO was >4.0 mg/L and pH ranged from 6.6 to 7.2. The highest un-ionized ammonia concentration observed in the control tanks was 0.06 mg/L (one tank on day 7) and 0.09 mg/L for the fipronil-spiked tanks (one tank on day 12). By day 23, un-ionized ammonia in the fipronil-spiked and control tanks dropped below 0.002 and 0.007 mg/L, respectively, and remained low for the duration of the experiment. In all tanks, un-ionized ammonia levels were always below documented toxic levels for chronic exposure 20.

Fipronil was detected in water samples, although at considerably lower concentrations than in both fish and sediment samples. The highest concentration of fipronil detected in water was 20 µg/L on day 7. From days 7 to 42, the concentration of fipronil steadily decreased for most time points, and fipronil was undetected at day 42. In water samples, the EF of fipronil was greater than the EF of racemic fipronil (0.5) at all sampling time points with detectable fipronil concentrations (Fig. 2), indicating that the amount of (+) enantiomer was higher than that of the (−) enantiomer. The fipronil transformation products fipronil sulfone and fipronil sulfide were not detected in water samples during the course of the experiment. Fipronil sulfide was the most abundant chemical form detected in sediment (Fig. 3) and was detected at significantly higher concentrations (p < 0.0001) than fipronil and fipronil sulfone, beginning on day 21 for the duration of the experiment. Fipronil sulfide concentrations in sediment ranged from 1.2 to 2 mg/kg and did not vary appreciably over the course of the study. Fipronil concentrations in sediment were highest on day 0 (2.0 mg/kg), decreased significantly (p < 0.0001) through day 14, and remained below 0.5 mg/kg and did not vary significantly through the remainder of the exposure. Fipronil was not detected in the sediment on the final day of exposure (day 42). Fipronil sulfone was initially detected in sediment on day 28 (0.2 mg/kg), and concentrations remained below 0.09 mg/kg thereafter (Fig. 3). The EF of fipronil in sediment samples was 0.6 on day 0 of the exposure period; this deviation from the racemic EF of 0.5 was probably caused by enantioselective anaerobic biodegradation during the six-month aging period after sediment spiking. The EF of sediment fipronil was significantly higher than 0.5 (z statistic ranged 3.9–13.6) throughout the exposure period (Fig. 2).

Figure 2.

Average measured enantiomeric fraction (mean ± standard error) of fipronil in fish, sediment, and water by sampling day. Fipronil was below detection in sediment and water on day 42, and there are no data on enantiomeric fractions for this time point.

Figure 3.

Average concentrations (mean ± standard error) of fipronil (equation image), fipronil sulfone (equation image), and fipronil sulfide (equation image) in sediment by sampling day. Fipronil sulfone was detected only in samples collected on day 28, 35, and 42.

Fish removed just prior to sediment exposures (time 0) had no detectable body burdens of fipronil, fipronil sulfide, or fipronil sulfone. The mean concentrations of fipronil and fipronil sulfide in fish sampled at day 7 were 2.0 (± 0.2) and 2.1 (± 0.3) mg/kg, respectively (Fig. 4). However, fipronil sulfone was the most abundant transformation product detected at day 7 (8.4 ± 0.5 mg/kg) and throughout the 42-d period (range, 3.0–8.4 mg/kg). Fish tissue concentrations of fipronil sulfone were significantly (p < 0.0001) higher than either fish or sediment concentrations of fipronil and fipronil sulfide on day 7 through 42 (Figs. 3 and 4). Fish tissue concentrations of fipronil and fipronil sulfide remained below 2 mg/kg over the course of the study (Fig. 4) and were below 1 mg/kg from day 21 to 35. Concentrations of fipronil in fish tissue and sediment were similar for the majority of the sampling time points. Compared to sediment, fish tissue fipronil sulfide concentrations were lower throughout the exposure and much lower on day 35 (p < 0.0001).

Figure 4.

Average concentrations (mean ± standard error) of fipronil (equation image), fipronil sulfone (equation image), and fipronil sulfide (equation image) in fish by sampling day.

The EF of fipronil in fish was significantly higher (z statistic ranged 3.7–26.2) than the value for racemic fipronil (EF = 0.5) throughout the experiment (Fig. 2). Overall, the fipronil EF in fish increased significantly (p < 0.0001) from day 7 to 21, after which it decreased. The fish EF values for the 7- and 14-d sampling times were lower than the EF values for both sediment and water. The 7-d EF value was 0.6, the same as that for sediment on day 0 of fish exposure. The final EF value measured at day 35 was similar for tissue and sediment.

DISCUSSION

Acute toxicity of fathead minnows (LC50) to racemic fipronil is similar to values measured for other fish species. In our study of fipronil toxicity, fathead minnows were slightly more sensitive to racemic fipronil (96-h LC50 = 448.49 µg/L) than channel catfish (96-h LC50 = 560 µg/L) and more tolerant than bluegill (96-h LC50 = 83 µg/L), sheepshead minnow (96-h LC50 = 130 µg/L), rainbow trout (96-h LC50 = 246 µg/L), or Japanese medaka (96-h LC50 = 94.2 µg/L) 12 (see http://www.epa.gov/EPA-PEST/1996/June/Day-12/Factsheet.pdf). Based on acute test comparisons, fathead minnows are not extremely sensitive to racemic fipronil. Furthermore, even the highest known concentration of racemic fipronil detected in surface waters (6.4 µg/L 21) is well below the 96-h and 7-d LC50s for fathead minnows determined in the present study. Enantioselective toxicity was observed only after 7 d of exposure, and enantioselective toxicity is not well documented in fish. In an acute exposure with Japanese medaka 12, enantioselectivity was not observed in 96-h LC50s for fipronil and its two enantiomers; however, this exposure was terminated after 96 h. Perhaps with longer exposure times (i.e., 7 d) enantioselective toxicity may have occurred in medaka, as was observed in the present study.

We observed an interesting trend of the (−) enantiomer being less toxic in waterborne exposures than both the racemate and the (+) enantiomer. The slopes of fish growth plotted versus concentration (Fig. 1) may give some insight into the observed trends of enantioselective toxicity 22. The higher (steeper) slopes of the concentration–response curves for the racemate and (+) enantiomer compared to the significantly lower (flatter) slope of the concentration–response curve for the (−) enantiomer imply a similar relationship between toxicity and increasing concentration for the racemate and (+) enantiomer compared to the (−) enantiomer. Higher slopes generated by the (+) enantiomers and racemate indicate that relatively small increases in their concentrations can result in larger increases in toxicity. The lower slope for the (−) enantiomer indicates that larger increases in chemical concentration are required to produce the same increases in toxicity observed for the other forms of fipronil. For both acute and subchronic exposures, the (−) enantiomer was consistently the least toxic.

Data from the fish bioaccumulation experiment may shed some light on the interesting trends in enantioselectivity observed in the toxicity experiment. In the sediment exposure, fish rapidly bioaccumulated fipronil and/or fipronil sulfide from the sediment and most of it was transformed in vivo and stored in their tissue as fipronil sulfone. Also, both fish and sediments were enriched in the (+) enantiomer of fipronil. By making a few assumptions, these results may be used to explain the enantioselective trends observed in the aquatic toxicity experiment. The first assumption is that the (−) enantiomer is selectively metabolized in the fish to fipronil sulfone. Fipronil sulfone is a known oxidation product in mice, invertebrates, and fish 14, 23, 24. This assumption is supported by the increase in EF of fipronil in fish from 0.6 to >0.8 over the experiment time (Fig. 2), indicating faster biotransformation of the (−) enantiomer. (As mentioned earlier, the initial EF of the sediment was not the racemic EF of 0.5 but 0.6.) This increase may have been caused by preferential uptake of the (+) enantiomer by the fish; however, enantioselective transport of this kind has not been reported in the literature.

We also assume increased toxicity of the sulfone metabolite compared to the parent compound in fathead minnows. This assumption is based on U.S. EPA data, which demonstrated that fipronil sulfone is 6.3 times more lethal to rainbow trout and 3.3 times more lethal to bluegill (Lepomis macrochirus) than the parent compound (http://www.epa.gov/EPA-PEST/1996/June/Day-12/Factsheet.pdf). Finally, we know from the present study that the (+) enantiomer is significantly more toxic than the (−) enantiomer in fathead minnows. Combining assumptions with fact, we hypothesize that the fathead minnows in the present study were selectively metabolizing the less toxic (−) enantiomer to the more toxic sulfone metabolite. Based on this hypothesis, we can rationalize the relative tissue concentrations of each enantiomer and the sulfone metabolite in fathead minnows in the present study, based on their exposure: (+) enantiomer, (−) enantiomer, or racemate. Fathead minnows exposed to the pure (−) enantiomer would retain some of this less toxic (−) enantiomer while converting some of it to the more toxic sulfone metabolite, resulting in a body burden of the less toxic (−) enantiomer and the more toxic sulfone metabolite. Fish exposed to the (+) enantiomer would retain relatively more of the more toxic (+) enantiomer, resulting in a body burden of primarily the toxic (+) enantiomer. Fish exposed to the racemate would retain more of the toxic (+) enantiomer and convert more of the (−) enantiomer to the toxic sulfone metabolite, resulting in a body burden of primarily the more toxic forms (the [+] enantiomer and the sulfone metabolite) and a smaller portion of the less toxic (−) enantiomer. Therefore, fish exposed to the (+) enantiomer and the racemate (which were both significantly more toxic than the [−] enantiomer in the present study) could potentially retain greater relative amounts of the more toxic forms of fipronil (and its metabolite) and thus exhibit increased toxic effects relative to fish exposed to the (−) enantiomer.

CONCLUSIONS

The two experiments presented herein suggest that there are several potential scenarios for fipronil exposure in aquatic environments. Although waterborne fipronil is toxic to larval fish, known environmental concentrations are not currently at levels likely to produce acute toxicity. As with many environmental contaminants, the true threat to aquatic organisms lies in the sediment, and fipronil presents several potential exposure scenarios from sediment. While the present study demonstrated the fate of fipronil and potential exposures in a simplified sediment–fish system, it also introduced several important factors to be considered when evaluating the potential deleterious effects of fipronil in the aquatic environment. First, fish may be a source of fipronil sulfone (fipronil sulfone was not detected in sediment until well after it was detected in fish), presenting an increased risk to sediment-dwelling organisms with greater sensitivity to fipronil sulfone. For example, fipronil sulfone is slightly (though not significantly) more toxic than fipronil to crayfish (Procambarus clarkii) when exposed via water in acute exposures 9 and in both acute and chronic end points of larval midges (Chironomus tentans) when exposed via sediment 25. Based on data for bluegill and rainbow trout (http://nepis.epa.gov/Adobe/pdf/p1001KCY.pdf), conversion of fipronil to fipronil sulfone may also be detrimental to the fish itself. Also, fathead minnows accumulated body burdens of fipronil, fipronil sulfone, and fipronil sulfide, allowing for possible trophic transfer and/or bioaccumulation of all three chemicals if fish were consumed by predator organisms. Although the present study adds to the body of knowledge demonstrating the variable fate of fipronil in the aquatic environment, more research on the toxicity of the sulfide and sulfone metabolites to nontarget organisms will aid in understanding the potential deleterious effects of fipronil in different exposure scenarios.

Acknowledgements

We thank the Office of Research and Development, Ecosystems Research Division, U.S. Environmental Protection Agency for in-kind support and the U.S. Environmental Protection Agency for donation of the fathead minnows used in this research. Funding for S. Baird was provided by the University of Georgia's Interdisciplinary Toxicology Program and the Department of Environmental Health Science. Daily monitoring of toxicity experiments was conducted with the help of L. Choi, B. Cantrell, S. Dulson, and A. Kulkarni. We thank J. Overmyer for valuable guidance in the design and conduct of these experiments and R. Cooper and M. Newman for valuable direction in the statistical analysis of the data. The present study has been reviewed in accordance with the U.S. Environmental Protection Agency's peer and administrative review policies and approved for publication. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.

Ancillary