Acute toxicity tests and meta-analysis identify gaps in tropical ecotoxicology for amphibians
Abstract
Amphibian populations are declining worldwide, particularly in tropical regions where amphibian diversity is highest. Pollutants, including agricultural pesticides, have been identified as a potential contributor to decline, yet toxicological studies of tropical amphibians are very rare. The present study assesses toxic effects on amphibians of 10 commonly used commercial pesticides in tropical agriculture using 2 approaches. First, the authors conducted 8-d toxicity assays with formulations of each pesticide using individually reared red-eyed tree frog (Agalychnis callidryas) tadpoles. Second, they conducted a review of available data for the lethal concentration to kill 50% of test animals from the US Environmental Protection Agency's ECOTOX database to allow comparison with their findings. Lethal concentration estimates from the assays ranged over several orders of magnitude. The nematicides terbufos and ethoprophos and the fungicide chlorothalonil were very highly toxic, with evident effects within an order of magnitude of environmental concentrations. Acute toxicity assays and meta-analysis show that nematicides and fungicides are generally more toxic than herbicides yet receive far less research attention than less toxic herbicides. Given that the tropics have a high diversity of amphibians, the findings emphasize the need for research into the effects of commonly used pesticides in tropical countries and should help guide future ecotoxicological research in tropical regions. Environ Toxicol Chem 2014; 33:2114–2119. © 2014 SETAC
INTRODUCTION
Amphibians are facing global-scale declines of a severity unprecedented among vertebrates 1-3. Approximately 32.5% of amphibians are considered globally threatened 1, and at least 42% of amphibian species are showing population declines 3. While many factors are recognized as widespread threats to amphibian biodiversity, 1 contributor to population declines is contamination from environmental pollutants such as agrochemical pesticides 2, 4-7. Pesticides can have direct lethal effects or chronic sublethal effects on amphibians, either of which may impact population dynamics 2, 8.
Amphibian declines have been particularly severe in the Neotropics 1, 9, 10, where the liberal use of pesticides remains poorly regulated and existing pesticide regulations are ineffectively enforced 11-13. Many studies have examined the effects of toxic pesticides on amphibians 5, yet published studies are strongly biased toward the study of temperate amphibian species and pesticides used in temperate countries. There are broad differences in the types of pesticides used in temperate and tropical countries—differences that derive from different crop requirements, pest diversity, and regulatory structure 11. Yet, because most toxicological research is conducted in temperate regions, there is little information regarding toxicity to amphibian species in tropical regions for even the pesticides most commonly used in tropical agriculture 14. This gap is of concern, given that amphibians are a predominantly tropical group of vertebrates 15 and given that amphibian declines are most severe in tropical regions 1, 3.
Costa Rica, for example, is a tropical country with both very high amphibian biodiversity and very high rates of pesticide usage. Within Costa Rica, amphibian populations have declined both in human-dominated landscapes with intense agricultural activity 9 and in pristine sites such as national parks and protected areas 16, 17. Costa Rica also has the highest rates of pesticide usage per hectare of arable land for any country with available data worldwide 18.
In the present study, we evaluated the toxicity of the 10 most commonly used nonfumigant pesticides in Costa Rica using 2 approaches. First, we conducted toxicity assays using tadpoles native to Costa Rica (red-eyed tree frogs, Agalychnis callidryas) to generate standard toxicity estimates. Second, we conducted a meta-analysis of published toxicity estimates to illustrate trends in research, to identify knowledge gaps in tropical toxicology, and to evaluate the relative toxicity of studied pesticides in a broader perspective. Our goals were to establish a baseline understanding of the toxic effects of commonly used pesticides in tropical regions and to evaluate whether previous research attention is directed toward pesticides posing the greatest risk. These data are critical for a comprehensive evaluation of pesticide impacts to amphibians and for developing appropriate regulations for liming impacts of pesticides to nontarget organisms.
METHODS
Study site
We conducted the present study at La Selva Biological Station, a tropical wet forest reserve in the lowlands of Sarapiqui, Costa Rica. La Selva is predominantly primary forest yet it is surrounded by an agricultural matrix dominated by cattle pastures and banana and pineapple plantations 19, 20. La Selva hosts at least 52 species of amphibians and has experienced a multidecade directional decrease in density of amphibians 17.
Study species, field collection, and animal care
Agalychnis callidryas is a tropical phylomedusine tree frog distributed from Mexico to Colombia 21. It is categorized by the International Union for Conservation of Nature (IUCN) as a species of least concern, and though its population trend is decreasing, the species remains common in both primary rainforests and sites with significant anthropogenic disturbances 21. We collected newly laid (up to 3 d) egg masses from vegetation surrounding seasonal ponds and other breeding sites around the La Selva reserve between September and December 2010.
We held egg masses until hatching in a laboratory with ambient environmental conditions similar to those of a rain forest understory and allowed egg masses to hatch naturally and without disturbance into bins containing ultraviolet-irradiated and carbon-filtered well water. We initiated all trials with tadpoles at Gosner stage 25, 3 d to 7 d after hatching. We conducted each trial with tadpoles from between 3 and 10 masses to ensure genetic variability in each trial.
Acute toxicity assays
We evaluated the acute toxicity of commercial formulations of each of the 10 most commonly used nonfumigant pesticides from Costa Rica (Table 1). Data on pesticide application rates for all types of pesticides are not available from Costa Rica, but because there is no in-country capacity for pesticide production, pesticide importation rates provide valuable information on pesticide usage. We used data derived from importation records of pesticides into Costa Rica from 1977 to 2005 as the basis for our pesticide selection 11.
| Active ingredient (% active ingredient in formulation) | Formulationa | Type | Rank in Costa Ricab | Rank in United Statesc | Number of LC50 values in literature review | Number of species studied in literature review |
|---|---|---|---|---|---|---|
| Mancozeb (43.5) | Bioman 43.5 SC | F | 1 | 20 | 6 | 3 |
| 2,4-Dimethylaniline (60) | Rimaxil 60 SL | H | 2 | 5 | 30 | 8 |
| Chlorothalonil (50) | Daconil 50 SC | F | 3 | 13 | 2 | 2 |
| Glyphosate (36) | Root Out 36 SL | H | 4 | 1 | 661 | 10 |
| Terbufos (15) | Terbufos 15 G | N | 5 | NAd | 0 | 0 |
| Tridemorph (86) | Calixin 86 OL | F | 6 | NAe | 4 | 1 |
| Paraquat (20) | Ati-La 20 SL | H | 7 | NAd | 40 | 8 |
| Propanil (48) | Proparroz 48 EC | H | 8 | 17 | 6 | 3 |
| Ethoprophos (10) | Mocap 10 G | N | 9 | NAd | 0 | 0 |
| Diuron (80) | Karmex 80 WG | H | 10 | NAd | 2 | 1 |
- a Numbers represent percent active ingredient.
- b Data for Costa Rica span 1977 to 2005 and are derived from Ramirez et al. 11.
- c Data for the United States are from 2001 estimates and are based on US Environmental Protection Agency (2010) 41.
- d Indicates pesticide is not among top 20 pesticides in use.
- e Indicates pesticide is not approved for use in the United States.
- LC50 = 50% lethal concentration; F = fungicide; H = herbicide; N = nematicide; NA = not applicable.
We raised A. callidryas tadpoles individually in a 350-mL solution of water and pesticide for 8 d for our acute toxicity assays. We conducted all trials in a climate-controlled laboratory, with mean water temperature of 19.1 °C (average daily minimum, 18.8 °C; average daily maximum, 19.6 °C), average pH 6.34 (95% confidence interval, 6.13–6.54), and average dissolved oxygen 3.85 mg/L (95% confidence interval, 1.67–6.03). For each pesticide, we first identified the order of magnitude of the toxicity in a range-finding test. Given the results of the order of magnitude trials, we tested 6 concentrations of pesticides that ranged between 100% mortality and 0% mortality in the order of magnitude trial and a control with no pesticide. If the results from the second trial indicated a 50% lethal concentration (LC50) value below 100 mg/L, we then conducted a third trial in which we tested 6 concentrations of the pesticide that ranged between 100% mortality and 0% mortality from the results of the second trial and a control with no pesticide. To make test solutions for each concentration of a given pesticide formulation, we created a stock solution of pesticide formulation and deionized water at the highest concentration used in the trial. We then created dilutions from this stock solution to produce each desired pesticide concentration in a given trial. We tested 2 tadpoles per concentration for each of the order of magnitude trials and 6 tadpoles per concentration for each of the second and third trials. Specific concentrations of pesticides used are listed in Supplemental Data, Table S1.
We collected data on mortality every 24 h for 8 d. On day 4 of each trial, we transferred tadpoles into a fresh solution of their respective pesticide–water concentration (i.e., static renewal tests) to ensure consistency of pesticide concentrations after chemical degradation. We fed each tadpole approximately 50 mg of organic alfalfa powder every other day.
On day 8 of the 2 trials, we conducted a scan-sample behavioral assay to determine sublethal pesticide effects on behavior. For each surviving tadpole, we collected behavioral data every 2 min for 20 min. At each observation, we recorded whether tadpoles showed any active movement or no movement at all. We recorded mass of all surviving tadpoles on day 8 of each trial.
Statistical analysis
We calculated 8-d LC50 estimates by analyzing mortality data using binomial linear models with a probit link (i.e., “probit regressions”) in R (R Development Core Team, 2011). We used pesticide concentration as a continuous predictor variable and survival on day 8 as the binomial response. To determine sublethal effects on activity, we used binomial linear models with pesticide concentration as a predictor and activity in behavioral scan samples as a response; lowest-observed-effect concentration (LOEC) values were determined as the lowest concentration showing a significant parameter estimate differing from the control. To determine sublethal effects on growth, we used gaussian linear models with pesticide concentration as the predictor and mass on day 8 as the response; we estimated LOEC values for growth and for activity.
Review of pesticide toxicity data for larval amphibians
We obtained a list of studies from the US Environmental Protection Agency's ECOTOX database (2010) from which we extracted toxicity estimates to compare the LC50 results from our acute toxicity assays with previous work. We only used data from studies that included point estimates of 24-h to 96-h LC50 values, to provide some comparative basis. We included toxicity estimates from studies conducted for any species of amphibian and studies using either the active ingredient or a formulation of the pesticide. We included data from studies that combined pesticide exposure with other natural stressors (i.e., competitors, predators) but not from studies that combined pesticide exposure with other chemical stressors (other pesticides). Particular species (e.g., Xenopus laevis) were heavily represented in the data. To minimize the effects of imbalances of species representation, data for all LC50 values for a single species were condensed into a single average value by taking the geometric mean of all values for that species. The geometric mean was then taken for all the species for each pesticide to obtain an average estimate of the known LC50 values for amphibians. Using geometric means provides a more reliable estimate over arithmetic means, where values span several orders of magnitude. For direct comparison with this review, we used 96-h LC50 estimates from our acute toxicity assays because most reported LC50 estimates are from short (∼96 h) toxicity assays.
RESULTS
Tadpoles in control treatments experienced 100% survival to day 8. Each of the pesticides used in the present study produced 100% mortality at the highest concentrations, although the 8-d lethal concentrations varied considerably among pesticides (Figure 1A and Table 2). Chlorothalonil had the lowest LC50 of the pesticides studied, and 2,4-dimethylaniline had the highest LC50. Nine of our 10 pesticides demonstrated altered effects on tadpole activity, but we found no activity effects for glyphosate (Table 2 and Figure 1A). Nine of our 10 pesticides also showed adverse effects on growth (Table 2 and Figure 1A), but we found no growth effects for mancozeb. While we were able to detect an effect of mancozeb on behavior, we were not able to estimate an LOEC for behavioral effects because of the low number of surviving tadpoles. As expected, 4-d LC50 estimates used for direct comparison with our literature review were higher than or equal to our 8-d LC50 estimates for all pesticides.
| Active ingredient | 8-d LC50 (µg/L) | Significance (mortality) | LOEC activity (µg/L) | Significance (activity) | LOEC growth (µg/L) | Significance (growth) | Environmental concentrations in Costa Rica |
|---|---|---|---|---|---|---|---|
| Mancozeb | 382 113 | <0.0001 | NS | <0.0001 | NS | 0.9704 | No data available |
| 2,4-Dimethylaniline | 536 160 | <0.0001 | 31 250 | <0.0001 | 500 000 | <0.0001 | No data available |
| Chlorothalonil | 59 | <0.0001 | 12.5 | 0.0007 | 12.5 | 0.0102 | 11 µg/L detected in aquatic environments 41 |
| Glyphosate | 6403 | <0.0001 | NS | 0.7500 | 5920 | 0.0042 | No data available |
| Terbufos | 2658 | <0.0001 | 28 | <0.0001 | 28 | <0.0001 | 1.2 µg/L detected after application to banana plantation [42] |
| Tridemorph | 49 360 | <0.0001 | 312 | <0.0001 | 6250 | <0.0001 | No data available |
| Paraquat | 1706 | <0.0001 | 1625 | <0.0001 | 312 | <0.0001 | No data available |
| Propanil | 7821 | <0.0001 | 3120 | <0.0001 | 3120 | <0.0001 | No data available |
| Ethoprophos | 1373 | <0.0001 | 31 | <0.0001 | 93 | <0.0001 | 2.9 µg/L detected in surface waters near banana plantations [42] |
| Diuron | 63 336 | <0.0001 | 31 250 | <0.0001 | 3125 | <0.0001 | 5.6 µg/L detected in surface waters near pineapple plantations [42] |
- a Results are provided for mortality, behavior, and growth. Data on concentrations in aquatic environments are taken from the primary literature.
- LC50 = 50% lethal concentration; LOEC = lowest-observed-effect concentration; NS = not significant.
Our literature review yielded valid LC50 estimates for 8 of our 10 pesticides (Figure 1B and Table 2). In general, more LC50 estimates were available for herbicides than other types of pesticides, and the herbicides glyphosate and paraquat were the most heavily studied pesticides. Few studies were available for any of the fungicides (with 2 individual studies on chlorothalonil, 3 for mancozeb, and 1 for tridemorph). There were no available studies on the nematicides investigated in our 8-d toxicity assays (ethoprophos and terbufos).
DISCUSSION
The data generated and synthesized in the present study both provide a base of information for each of the pesticides studied relevant for regulatory agencies and reveal major gaps in tropical ecotoxicology that should guide future research efforts. The toxicity estimates produced in our 8-d toxicity assays span a very large range of values among pesticides, and our meta-analysis data illustrate that available data on toxicity are absent or very poor for the most toxic pesticides investigated in the present study.
The nematicides studied are the most toxic class of pesticides and the group of pesticides least studied by toxicologists. Both terbufos and ethoprophos showed sublethal effects below 100 µg/L. Terbufos affected both growth and activity at 28 µg/L. Ethoprophos affected growth and activity at 93.2 µg/L and 31.2 µg/L, respectively. These toxicity values are within an order of magnitude of environmental concentrations found in agricultural zones in Costa Rica (Table 2), indicating that amphibians are very likely exposed to potentially harmful concentrations of nematicides in these habitats. Unfortunately, no previously published toxicity estimates were available for either of these nematicides, suggesting that nematicides may be an unrecognized threat to amphibian populations in regions experiencing declines.
The fungicides we examined (mancozeb, chlorothalonil, and tridemorph) varied widely in toxicity, though each of our 3 fungicides showed very high toxicity based either on our acute toxicity assays or data in our meta-analysis. Chlorothalonil has the lowest LC50 of any pesticide used in the present study (59.36 µg/L) and had sublethal effects at 12.5 µg/L for growth and activity. Our data for mancozeb showed lower toxicity than that reported in other studies with mancozeb previously, where major effects on growth, development, and survival have occurred at very low concentrations for longer exposure durations, likely the result of pronounced lag effects in toxicity 22. Our LC50 estimate for tridemorph was several orders of magnitude higher than the single existing estimate of tridemorph toxicity to amphibians. The past few years have seen a major increase in research attention on fungicide impacts to amphibians 7, 23-25, and further continuation of this recent trend is likely to help to rapidly close gaps in our knowledge of the impacts of fungicides to amphibians.
The herbicides studied (2,4-dimethylaniline, diuron, glyphosate, propanil, and paraquat) are less toxic as a group than the fungicides or nematicides, although herbicides were by far the most intensively studied class of pesticides. Diuron, 2,4-dimethylaniline, glyphosate, and propanil had no lethal or sublethal effects below 3 mg/L, indicating relatively low toxicity. Paraquat was the most toxic of the herbicides studied and showed sublethal inhibition of growth at 312 µg/L. Previous studies of paraquat have found malformations in tadpoles and inhibition of growth at concentrations as low as 100 µg/L 26 and 62.5 µg/L 27, indicating very highly toxic effects over longer exposure intervals.
The results of the toxicity assays and meta-analysis show that the 3 very highly toxic pesticides (the fungicide chlorothalonil and the nematicides terbufos and ethoprophos) represent the principal gaps in the literature and that these gaps appear to be representative of trends for lack of data on fungicides and nematicides. Relyea and Hoverman 28 reviewed ecotoxicology studies and showed that of the research on single-pesticide toxicity, 65% of studies have focused on insecticides, 33% on herbicides, and only 2% on fungicides; nematicides were not evaluated in this review of ecotoxicological studies, presumably because research attention to nematicide toxicity has been negligible. In addition to high toxicity, the gap in research attention for chlorothalonil, terbufos, and ethoprophos is of concern because our LOECs are all within an order of magnitude of concentrations found in aquatic environments in Costa Rica (Table 2).
The most commonly used pesticides in Costa Rica differ markedly from the most commonly used pesticides in the United States, where ecotoxicological studies of amphibians are conducted more commonly than in tropical regions (Table 1). Overall, tropical agriculture appears to rely more heavily on the use of fungicides and nematicides than does temperate agriculture, likely because warmer and wetter tropical climates and tropical biodiversity necessitate a greater need for these classes of pesticides. As a trend, commonly used pesticides in the United States (i.e., glyphosate, 2,4-dimethylaniline) received considerable attention in toxicological research, yet even the most commonly used pesticides in Costa Rica (i.e., mancozeb, chlorothalonil) received very little research attention. This temperate bias in ecotoxicological studies is important. Most amphibian species (including most amphibian species experiencing declines and extinctions) inhabit tropical rather than temperate zones and are therefore more likely to face environmental exposure to pesticides used in tropical agriculture than pesticides used in temperate agriculture. Unfortunately, the gap in research on commonly used pesticides in tropical areas may inhibit efforts to link pesticides to widespread amphibian declines in these regions.
The present study represents a critical step toward investigating toxicological impacts of amphibians in tropical regions, with critical importance for amphibian declines. Our data from toxicity assays, linked with scattered reports of environmental concentrations from Costa Rica, should illustrate that in agricultural regions amphibians may be exposed to harmful concentrations of fungicides and nematicides (Table 2). This should provide compelling evidence that in agricultural landscapes tropical amphibians (and other aquatic organisms 5) may well be exposed to harmful concentrations of understudied pesticides.
Furthermore, and more controversially, pesticides have long been suggested to be costressors in widespread amphibian declines outside of agricultural areas 29-31. In Costa Rica, pesticides are applied primarily in lowland regions where agricultural activity is most intense 11. However, long-distance atmospheric transport of pesticides and tropical mountain cold trapping result in peak pesticide concentrations in higher-elevation forests 32, 33; this spatial pattern in pesticide residues is consistent with the spatial pattern of the most intense amphibian population declines in mid- to high-elevation tropical forests 34-36. While many of these amphibian population declines in montane tropical regions have been linked to the emergence of a potentially lethal fungal pathogen, Batrachochytruim dendrobatidis 34, 35, controversy lingers about whether B. dendrobatidis is acting alone or synergistically with other amphibian stressors 36-38. In particular, pesticide residues have been demonstrated to lower amphibian immune response 23 and to increase susceptibility to emerging diseases 39, 40. Given the large number of pesticides in use and the limited information on the toxicity of common pesticides used in tropical agriculture, links between pesticide residues and emerging disease have been difficult to establish.
In addition to providing basic data on lethal and sublethal effects for the most commonly used pesticides in Costa Rica, the present study identifies a concerning gap in efforts to link amphibian declines to pesticide contamination: the most toxic groups of pesticides (fungicides and nematicides) are either generally or entirely overlooked by toxicologists and amphibian conservation biologists. This gap likely results from major differences in pesticide-use regimes between temperate and tropical regions and because ecotoxicologists are predominantly based in temperate regions and focus on regional threats to the environment. We suggest that attempts to link pesticide contamination to amphibian declines should focus on contaminants that most likely threaten amphibians in regions where amphibian declines have occurred and are ongoing.
SUPPLEMENTAL DATA
Table S1. (92 KB DOC).
Acknowledgment
The present work was supported by the Florida International University Dissertation Year Fellowship to S.M. Whitfield. A. Brenescoto and M. Arguedas provided assistance in data collection. C. Bruhl, M. Boone, and 2 anonymous reviewers provided comments on the manuscript. The Organization for Tropical Studies provided logistical support. The Ministerio de Ambiente, Energía y Mares de Costa Rica and the Florida International University Institutional Animal Care and Use Committee provided research permits.
