Developmental assays using invasive cane toads, Rhinella marina, reveal safety concerns of a common formulation of the rice herbicide, butachlor.

Abstract

Identifying the adverse impacts of pesticide exposure in rice ecosystems is essential to guide regulations that are protective of wildlife and human health. Amphibians are valuable indicators within rice ecosystems because pesticide applications coincide with sensitive reproductive and developmental life stages. We conducted two experiments using wild cane toads (Rhinella marina) to test 1) whether environmentally relevant exposure to a commercial formulation of butachlor, an acetanilide herbicide used extensively in rice, affects amphibian development and 2) whether cane toad tadpoles are capable of acclimatizing to sub-lethal exposure. First, we exposed wild cane toads to 0.002, 0.02, or 0.2 mg/L of butachlor (Machete EC), during distinct development stages (as eggs and hatchlings, as tadpoles, or continuously) for 12 days. Next, we exposed a subset of animals from the first experiment to a second, lethal concentration and examined survivorship. We found that cane toads exposed to butachlor developed slower and weighed less than controls, and that development of the thyroid gland was affected: exposed individuals had smaller thyroid glands and thyrocyte cells, and more individual follicles. Analyses of the transcriptome revealed that butachlor exposure resulted in downregulation of transcripts related to metabolic processes, anatomic structure development, immune system function, and response to stress. Last, we observed evidence of acclimatization, where animals exposed to butachlor early in life performed better than naive animals during a second exposure. Our findings indicate that the commercial formulation of butachlor, Machete EC, causes thyroid endocrine disruption in vertebrates, and suggest that exposure in lowland irrigated rice fields presents a concern for wildlife and human health. Furthermore, we establish that developmental assays with cane toads can be used to screen for adverse effects of pesticides in rice fields.

Graphical Abstract

Capsule Summary:

Environmentally relevant exposures to a commercial formulation of butachlor, an herbicide used extensively in rice, affect development, morphology, and gene expression of cane toad tadpoles. Developmental assays with cane toad tadpoles can be used to screen for adverse effects of pesticides in rice fields.

Introduction

Wild amphibians are valuable indicators of the environmental and human health risks posed by pesticide exposure in rice ecosystems. Native and non-native amphibians occupy rice fields throughout Asia (Bambaradeniya et al., 2004; Holzer et al., 2018; Kanagavel et al., 2017; Khatiwada et al., 2016; Shuman-Goodier et al., 2017) and South America (Dure et al., 2008), and are directly exposed to pesticides as they develop in flooded fields and adjacent microhabitats. Amphibian growth and development are also sensitive endpoints used to identify endocrine disruption in vertebrates (Miyata and Ose, 2012; Opitz et al., 2005; Searcy et al., 2012). Organisms and methodologies that can be used to monitor for adverse effects of pesticide exposure in rice ecosystems are sorely needed in Southeast Asia, where few national monitoring programs exist (Schreinemachers et al., 2015; Schreinemachers and Tipraqsa, 2012).

Endocrine disruption to the hypothalamic pituitary thyroid (HPT) axis is known to perturb development and growth in vertebrates (Boas et al., 2012; Carr & Patiño, 2011; Colborn et al., 1993; Couderq et al., 2020). The herbicide butachlor is among the most common herbicides used in rice, and mounting evidence indicates it is mutagenic (Dearfield et al., 2002) and is an endocrine disruptor that affects development of wildlife and humans at environmentally relevant concentrations (Abigail et al., 2015; Chang et al., 2011; Li et al., 2016; Shuman-Goodier et al., 2017; Zhu et al., 2014). To address these concerns, we used invasive cane toad (Rhinella marina) larvae to test whether exposure to a common formulation of butachlor (Machete EC) affects development, thyroid gland morphology, and gene expression.

While many native amphibians are in critical decline across the world (Blaustein et al., 2011; Wake and Vredenburg, 2008), the cane toad is thriving. It has been introduced in over 40 countries outside of its native range in Central and South America (Lever, 2001; Turvey, 2013), and has successfully established itself in a wide range of novel environments. In the Philippines, the location of this study, cane toads are abundant in irrigated rice fields and in some regions adults reach densities of 1,040 toads per hectare (Diesmos, 2008). Despite their success in agricultural areas, we have found that cane toad larvae are sensitive to pesticides, even more so than the most abundant native counterparts (Shuman-Goodier et al., 2017). Because the cane toad is a wide-spread pest species for which extensive reference resources exist (e.g., annotated genome), it is an ideal candidate for assessing consequences of pesticide exposure in the Philippines and in other rice growing countries where it is abundant.

The physiological mechanisms that enable some organisms, such as the cane toad, to persist in agricultural environments are poorly understood. Phenotypic plasticity, whereby organisms of the same genetic blueprint can generate multiple phenotypes in response to varying environmental conditions (Pigliucci, 2006), may contribute to an organism’s ability to acclimatize to pesticide exposure, and subsequently promote tolerance and survival through future exposures (Vandegehuchte and Janssen, 2014). Phenotypes that confer tolerance specifically to pesticides have been observed in wood frogs (Lithobates sylvaticus), where tadpoles exposed to the insecticide carbaryl developed a phenotype that conferred some tolerance to future exposures of both carbaryl (Hua et al., 2013), and other insecticides (Hua et al., 2014). In Australia, the cane toad exhibits low genetic diversity, but displays large variation in morphology, behavior, and gene expression across different environmental gradients, which likely contributes to its success as an invader (Rollins et al., 2015). In the Philippines, where cane toads are successful in rice agricultural landscapes, plasticity may play an important role in response to pesticide stressors, but this has yet to be tested.

We conducted two experiments to test: 1) whether environmentally relevant exposure to a commercial formulation of butachlor affects cane toad development, thyroid morphology, and differential gene expression, and 2) whether cane toad tadpoles are able to acclimatize to a sub-lethal pesticide exposure. We used the commercial formulation Machete EC (68% Sinochem Crop Protection Inc., Philippines) popular in the Philippines, because it is one of the final butachlor products to which wildlife inhabiting rice fields could realistically be exposed. We hypothesized that exposure would disturb development and growth by influencing the thyroid gland and transcriptome, and that effects would be largest when animals were exposed as eggs and hatchlings because these periods correspond with development of critical tissue and organ systems. We further hypothesized that physiological responses to early pesticide exposure would allow tadpoles to acclimatize to a future exposure, thereby increasing their chances of survival later in life.

Methods

Animal collection and husbandry

Three egg masses were collected from fallow rice fields at the International Rice Research Institute’s experimental station on June 18^th, 2016 (Los Baños, Philippines, 14.11° N, 121.15° E, elevation 21 m ASL) and brought to greenhouse facilities. Annually, two seasons of irrigated rice are grown in the fields in a tropical, humid, lowland environment. Batches of fifteen eggs (five from each egg mass) at Gosner stage 1-5 were counted and transferred to fifty separate 16 L tanks (36.8 x 33.8 x 23.6 cm plastic rice storage bins) containing 8 L of tap water. Treatment tanks were randomly distributed throughout the greenhouse space, and all tap water was allowed to sit for 24 hours before use to allow residual chlorine to evaporate. Eggs were allowed to acclimatize for 12-hours in treatment tanks before butachlor was added from stock solutions to begin Experiment I. Tanks were held at ambient temperature to reflect the wide natural range of water temperatures in nearby rice fields and the temperature was recorded every hour using HOBO Pendant® data loggers and compared to an outside recording. The average 24-hour temperature in tanks throughout the experiment was 29.1 °C (min/max: 20.6/40.9°C) and did not differ cross treatment groups. The average temperature in a nearby rice field was 28.6°C (min/max: 23.5/41.7°C). We recorded dissolved oxygen, pH, conductivity, and total dissolved solids every three days with a HI9829 multiparameter meter (Hannah Instruments, Woonsocket, RI, USA). At the start of the experiment, the dissolved oxygen content in tank water was 9.4 ppm, pH was 8.1, conductivity was 736 uS/cm, and total dissolved solids (TDS) was 369 ppm. After hatching, we fed tadpoles daily ad libitum with a 1:1 mixture of crushed rabbit pellets and fish flakes.

Herbicide application and quantification

We evaluated the effects of the acetanilide herbicide butachlor using the commercial formulation Machete EC (68% active ingredient, 600 mg/mL 2-chloro-2′,6′-diethyl-N-butoxymethyl-acetanilide; Sinochem Crop Protection Inc., Philippines) which is popular in the Philippines (Yap & Demayo, 2015). Commercial formulations of butachlor are important to consider because they are the products that wildlife inhabiting rice fields are realistically exposed to. We refer to quantified concentrations of the active ingredient butachlor in our methods, but emphasize that our results and discussion are relevant to the commercial formulation, which contains butachlor (main active ingredient) and undisclosed additives. We exposed animals to a range of concentrations to which they could be realistically exposed in a rice field. The recommended application concentration of butachlor (molecular weight = 311.89 g/M) in rice fields with 3-4 cm of water ranges from 3.1 to 46.8 mg/L (Alla et al. 2008; Chen et al., 2007), and concentrations in paddy water following application range from 0.121 ± 0.051 mg/L to 0.357 ± 0.338 mg/L (Kaur et al., 2017; Ok et al., 2012). Actual concentrations of butachlor were determined using liquid chromatography triple quadrupole mass spectrometry (see SM for detailed methods) from 25 mL water samples that were frozen at −80°C. Samples were collected from a sub-sample of n = 4 tanks at day 0 in Experiment I, and daily during Experiment II; stock solutions were tested every 3 days (Experiment I) in order to estimate degradation half-life of butachlor (Table S1). For Experiment I, the nominal concentrations were 0.002, 0.02, and 0.2 mg/L, while the actual concentrations were 0.0023, 0.021, and 0.1 mg/L (Table S1). Thus, we refer to the actual concentration of 0.1 mg/L throughout the manuscript. For Experiment II, the nominal concentration was 0.4 mg/L and actual concentration of butachlor was 0.38 mg/L (Table S1).

Experiment I – Developmental effects

To determine whether butachlor exposure affects amphibian development, we exposed cane toads to three environmentally relevant concentrations (0, 0.002, 0.02, and 0.1 mg/L) during different developmental periods: as eggs and hatchlings (days 1-3), as tadpoles (days 4-12), or continuously (from days 1-12). Control animals were kept in aged tap water continuously for the 12-day period (Fig. 1). Experiment I consisted of 50 individual tanks each containing n = 15 eggs, with n = 5 replicate tanks per treatment. Survival was recorded daily. After animals hatched, a water change was conducted on day 4, and butachlor was either refreshed from stock solution (continuous exposure) or added for the first time (exposure as tadpoles); animals that were exposed as eggs and hatchling were removed from butachlor and placed into control water. On day seven, five animals per tank were removed for Experiment II (Fig. S1). On day 12, all animals were euthanized in 0.02% pH buffered MS222 at 250 mg/L, photographed with a 10 mm reference scale for morphological analyses, and weighed to the nearest 0.001g. Three individuals from each tank were frozen at −80°C for transcriptome analysis, and two from each tank were placed in 4% neutrally buffered formalin for 44 hours prior to storage in 70% ethanol for histological analysis of the thyroid gland.

Fig. 1.

The effects of butachlor exposure on A developmental (Gosner) stage and B weight of cane toad (Rhinella marina) larvae. Effects are model predictions that account for the unit of replication (tank), included as random effect. Colors signify exposure periods; animals exposed as eggs and hatchlings, tadpoles, or continuously. Error bars are 95% confidence intervals and stars depict pairwise differences obtained compared to control animals, and are colored to correspond to differences within exposure periods (Table S3, P < 0.05).

Experiment II – Acclimatization

To test whether cane toad tadpoles acclimatize to butachlor, we removed n = 5 individuals/tank from each treatment (n = 25/per treatment) in Experiment I on day 7, and subsequently exposed them to a lethal concentration of butachlor (Fig. S1), 0.4 mg/L (Li et al., 2016). Animals were placed in individual petri dishes containing 10 mL of water and 0.4 mg/L butachlor. Tadpoles were monitored every four hours over a 32-hour period, and mortality and/or severely reduced movement, which we defined as a lack of startle response when touched on the tail three times with a zip tie, were recorded. When mortality or reduced movement were observed animals were removed and euthanized in 0.02% pH buffered MS222 at 250 mg/L. This method of euthanasia was negotiated with IACUC with the intent to remove animals where death was imminent and prevent unnecessary, prolonged suffering. Water was refreshed every 12 hours from stock solution, and after 32 hours all remaining tadpoles were euthanized. All animals were preserved in RNAlater. Acclimatization was inferred when animals previously exposed to butachlor survived, or survived longer compared to controls (see statistical analyses below).

Morphometric and behavioral analyses

We used ImageJ (NIH, Bethesda, MD, USA) to measure the total body length, tail length, snout-vent length, tail width, and hind limb lengths of tadpoles photographed at the end of Experiment I. Tadpoles were photographed on their sides in a petri dish with a built-in grid for scale (Fig S2). Activity level was measured as the number of times any animal crossed a line placed in the center of each tank (Experiment I) or the number of times individual animals crossed a line drawn in the center of each petri dish (Experiment II) during standardized three-minute videos using GoPro cameras (Hero 8). Tanks and petri dishes were filmed one at a time, or on separate tables. In Experiment I, we recorded videos on day 6 and in Experiment II we recorded videos in petri dishes 24-hours after the lethal exposure began. Videos were also examined qualitatively to identify and describe any abnormal swimming behaviors. Morphometric and behavioral analyses were performed blind to experimental treatments.

Thyroid histology

We examined the morphology of the thyroid gland for 62 samples (n = 9 per treatment, with the exception of those exposed as tadpoles to 0.002 mg/L where n = 5, and as eggs and tadpoles to 0.02 mg/L where n = 7). We used ImageJ to quantify total number of follicles, total gland area, follicle area, colloid area and thyrocyte height. Up to five follicles were measured in each individual, and four thyrocyte cells from each follicle were measured in the order of north, south, east, and west to determine thyrocyte height (see SM for detailed methods, Fig. S3).

Statistical Analyses

We used R version 3.5.1 (R Core Team, 2018) for statistical analyses, and ggplot2 for visualizations (Wickham, 2017). We assessed normality by plotting and visually inspecting residuals using QQplots, and tested for homogeneity of variance (homoscedasticity) by comparing plotted residuals to the fitted terms and using Levene’s tests. Homogeneity of variance was observed, except for the stage endpoint. We accounted for our experimental unit throughout all analyses by including tank as a random effect. To test for differences in developmental stage, weight, behavioral and morphological endpoints across butachlor treatments, we constructed separate linear mixed effects models for each endpoint using the lme4 package and performed pairwise comparisons across concentrations and exposure period (with simultaneous Bonferroni corrections) using the emmeans package (Lenth. et al., 2018). Each model included the categorical fixed effects of exposure concentration × exposure period (egg and hatchling, tadpole, or continuous) and tank included as a random effect. Concentration and lifestage were considered as categorical variables because we did not wish to assume a linear response to butachlor exposure.

The range of development tadpoles reached at the end of our experiment (Gosner stage 27-36) fell within the period of pro-metamorphosis; defined as the period between development of the hind limb and the appearance of the forelimbs associated with increased thyroid gland activity (Fort et al., 2007). To account for potential differences in thyroid gland morphology caused by development alone (i.e. animals further along in development could have larger thyroid glands with more follicles), we included developmental stage as a co-variate in the mixed effects models for morphological endpoints of the thyroid gland (SM).

To visualize survival rates over time (Experiment II), we constructed non-parametric Kaplan Meyer survivorship curves using the survfit function in the R package survival (Therneau, 2016). To test for statistical differences in survival rate across experimental concentrations, we performed three non-parametric Cox proportional-hazards models (Cox 1972), one for each exposure window, using the coxph function. The categorical variable of concentration was entered as a fixed effect, tank was entered as a random effect using the frailty argument, and the control group was entered into all models as the intercept. The exponent of the model coefficients for each categorical fixed effect represented hazard ratios, and provided a useful measure of effect size relative to the control (animals naive to butachlor). In our analyses, a hazard ratio > 1 indicated that previous exposure to butachlor increased the probability of mortality when animals are exposed to butachlor a second time, a hazard ratio < 1 indicated that previous exposure to butachlor decreased the probability of mortality, and a hazard ratio = 1 indicates that previous exposure had no effect on survival rate.

Transcriptome profiling using RNAseq and Illumina Sequence Analyses

We conducted RNAseq to examine the effects of butachlor exposure on gene expression by examining a subset of the original 10 treatment combinations: animals exposed continuously for the 12-day period compared to controls. The low (0.002 mg/l) and medium (0.02 mg/L) butachlor exposures were examined, but because no animals survived the continuous exposure at 0.1 mg/L, this concentration was not represented in the transcriptome analysis. Intestinal tissue was isolated from whole tadpoles that were stored at −80°C in RNAlater, and total RNA was extracted from n = 7 samples per treatment using TRIzol (Invitrogen cat. no. 15596026). Intestinal tissue was used because it undergoes morphological re-modeling during amphibian metamorphosis (Schreiber et al., 2005; Shi and Brown, 1993) and is sensitive to thyroid hormone. Furthermore, butachlor exposure affected expression of a thyroid sensitive gene in cane toad intestines in a previous experiment (Shuman-Goodier and Propper, 2016). Detailed RNA extraction and Illumina library preparation and sequence analysis methods are available in the SM. The top 1,000 differentially expressed transcripts were extracted for each of the three treatment group comparisons for downstream analyses including annotation. Retained transcripts were required to exhibit ≥ 4-fold-change in expression between treatment groups and false discovery rate (FDR)-adjusted P-value ≤ 0.05. Transcriptome sequence data are freely available in the NCBI database (BioProject accession number PRJNA660744).

Annotation and GO-term Enrichment Analysis

The final transcriptome was annotated using Annocript (Musacchia et al. 2015) for organism-level sequence similarity analysis, and the sma3s program (Casimiro-Soriguer et al. 2017) was used to assign functionally relevant gene ontology (GO) terms to each transcript (Camancho et al. 2009). Both programs utilized versions of the Uniref90 database as a reference for BLAST+ (Camancho et al. 2009). Enrichment analysis was then conducted with the resulting GO terms and differential expression results for each transcript using the GOSeq program in R (Young et al. 2010). GOSeq utilizes variations of Fischer’s exact test to look for overrepresentation of GO terms among differentially expressed transcripts based on the total presence of those GO terms, and identifies them with p-values corrected to FDRs. Expression heat maps were created with the ComplexHeatmap package in R (Gu et al. 2016), using the most highly significant GO terms associated with the top 1000 differentially expressed transcripts for each treatment/control comparison.

Results

Experiment I: Development, growth, and morphology

Cane toad development was affected by butachlor exposure, the timing of exposure, and there was an interaction between concentration and timing of exposure (P = < 0.001; Fig. 1A; Table S2). For ease of interpretation, we refer to the exposure concentrations as the low (0.002 mg/L), medium (0.02 mg/L) or high (0.1 mg/L) concentrations throughout the results and discussion. Animals exposed as eggs and hatchlings developed slower than controls, but only when exposed to the low concentration. Animals exposed as tadpoles developed slower, but only when exposed to the high concentration. Animals exposed continuously throughout the experiment developed slower than controls across all concentrations (Fig. 1, Table S4). Notably, the high concentration caused 100% mortality when animals were exposed continuously; thus, this treatment was removed from all subsequent analyses. The high concentration also caused 58% mortality when animals were exposed as eggs and hatchlings, and, as such, morphological and developmental comparisons were interpreted with attention to the fact that sample sizes were lower in this treatment. Mortality was not observed in other treatments.

Cane toads weighed less compared to controls when they were exposed to the medium and high concentrations of butachlor (Fig. 1; P < 0.001) across all exposure periods (P < 0.001). Exposure to all concentrations of butachlor affected morphometric endpoints (Fig. 2, P < 0.05), whereas exposure lifestage did not. Total body length was smaller across all exposed animals compared with the controls, except for animals exposed as eggs and hatchlings to the high concentration. Pairwise comparisons revealed that animals exposed as eggs and hatchlings to the low concentration of butachlor were significantly smaller than animals exposed to the medium concentration (Fig. 2, P < 0.05). When cane toads were exposed as eggs and hatchlings, only the low concentration resulted in shorter snout-vent length, tail width, hind limb and tail lengths. In contrast, when animals were exposed as tadpoles, the high concentration resulted in shorter snout-vent length and hind limb lengths. Cane toads exposed to butachlor continuously were shorter than control across all measured morphological endpoints and exposure concentrations (Fig. 2; P < 0.05). No other endpoints were affected. Results from statistical tests for all endpoints are provided in the SM (Tables S2 & S3).

Fig. 2.

The effects of butachlor exposure on the morphometric endpoints (in mm) of A total body length, B tail length C snout-vent length, D tail width, and E hind limb length of cane toad (Rhinella marina) larvae. Effects are model predictions that account for the unit of replication (tank), included as random effect. Colors signify different exposure periods; animals exposed as eggs and hatchlings, tadpoles, or continuously (n = 35 per treatment except for the 0.1 mg/L egg and hatching exposure, where n=5). Error bars are 95% confidence intervals and stars depict pairwise differences compared to control animals, and are colored to correspond to differences within exposure periods (Table S3, P < 0.05).

Thyroid gland histology

We observed a consistent non-monotonic dose response where the thyroid gland was affected by exposure to the low concentration of butachlor across all morphological endpoints, for two exposure periods: animals exposed as eggs and hatchlings and animals exposed continuously (Fig. 3). Cane toads in these treatments had reduced follicle, colloid, and total gland areas and shorter thyrocyte cells (P < 0.01). Interestingly, animals from all exposure periods exposed to the low concentration also had more follicles overall (P < 0.05). Animals exposed as tadpoles to the high concentration had less total gland area and shorter thyrocyte cells, but thyroid morphology was generally not affected by exposure to low and medium concentrations, with the exception of follicle number (P < 0.05). Thyroid figures and pairwise comparisons are based on stage-corrected data that accounted for observed effects of development on each morphological endpoint (Fig. S4, Table S3).

Fig. 3.

Representative images of the effects of butachlor treatment and exposure period on thyroid development in cane toad (Rhinella marina) larvae matched by developmental stage (GS = Gosner Stage). Thyroid glands analyzed from controls not treated with butachlor (panels, A, E & I, n = 8 individuals) contained an average of 4 follicles/gland and larger thyroid glands and longer thyrocyte cells. Thyroid glands analyzed from tadpoles treated continuously (I-K) or as eggs and hatchlings (A-D) were most affected by butachlor exposure, and had smaller thyroid glands, follicle areas, and thyrocyte cells as well as follicular hyperplasia (multiple cell layers observed surrounding the follicle). A proliferation of small follicles was observed in the thyroid glands from tadpoles treated with the lowest concentration, 0.002 mg/L (n = 6-8/treatment).

Behavior

Activity did not differ across treatments or lifestage (Fig S5, Table S2). However, a strong interaction was observed for animals that were exposed as eggs and tadpoles to the high concentration of butachlor (P < 0.01). These animals were significantly less active than controls. Although activity level did not differ significantly across concentration and lifestage, abnormal circular swimming behavior was observed qualitatively the low and medium exposures. Circular swimming behavior was not observed in control animals. Results from statistical tests for all endpoints are provided in the SM (Tables S2 & S3).

Transcriptomics and Differential Expression

The transcriptome assembly, based on contaminant-filtered reads, resulted in 1,251,217 trinity genes (isoforms) and 1,688,777 transcripts; of these genes, 42,806 were expressed by at least 5 transcripts per million (TPM) in any single sample. Total GC content was 48.55%. The mean contig length was 597.48 and the contig N50 was 796.

Across all comparisons (i.e., control, and continuous low and medium butachlor exposure concentrations) there were 32616 transcripts identified as differentially expressed. The majority of differentially expressed transcripts were upregulated in control samples relative to samples treated with butachlor, whereas only 1.68 and 5.95 transcripts were also upregulated relative to the control in the 0.002 and 0.02 mg/L exposure groups. Of the top 1,000 differentially expressed isoforms, 56 were upregulated and 944 were downregulated in the treatment group exposed continuously to the medium concentration of butachlor as compared to the control group. For the treatment group exposed to the low concentration, 23 genes were upregulated and 977 were downregulated as compared to the control group. In comparing the two butachlor treatment groups, 93 genes were upregulated and 907 downregulated in the group exposed to the medium as compared to the low concentration (Fig. 4, SM).

Fig. 4.

Heatmap showing the top 1,000 differentially expressed (DE) transcripts generated for each pairwise comparison: 0.002 and 0.02 mg/L butachlor exposure vs. control, 0.02 vs 0.002 mg/L exposure, for cane toads (Rhinella marina) exposed continuously. Rows represent unique DE transcripts, and columns represent replicate animals. The top gray horizontal bars identify clear clustering in the transcriptome profile of the control samples (n = 7) versus samples exposed to 0.02 mg/L (n = 7) and 0.002 mg/L (n = 6). Heatmap colors correspond to scaled fragments per kilobase of transcript per million mapped reads (RPKM) values for each transcript, where blue transcripts are upregulated in at least one pairwise comparison between the three treatments and red transcripts are downregulated. Top enriched gene ontogeny (GO) terms that correspond to the DE transcripts are displayed by colored hashes to the right of the heatmap.

Gene Ontology (GO) Clustering

GO term annotation yielded 16,471 differentially expressed transcripts assigned to at least one GO term, representing 50.5% of the total number of differentially expressed transcripts. Transcripts in cane toad samples exposed continuously to both low and medium concentrations of butachlor were represented by similar enriched GO terms when compared to controls. These included the biological responses of translation, biosynthetic processes, metabolic processes, anatomic structure development, cellular protein modification, cell differentiation, response to stress and immune system processes and the molecular functions of ion and RNA binding, ribosomal activity, enzyme binding, and oxidoreducatase and kinase activity (Fig. 5). To identify enriched GO terms associated with the phenotypic differences we observed between butachlor exposure concentrations, we also compared transcripts of cane toad samples exposed continuously to the low versus medium concentration of butachlor. The top enriched GO terms for this comparison included the biological processes of translation, biosynthetic processes, metabolic processes, response to stress and the molecular function of ion and pigment binding and an increase in oxidoreductase activity. While enriched GO terms analyses were completed across all differentially expressed transcripts (both up and downregulated), most enriched terms in butachlor exposed animals were associated with downregulated transcripts (Fig. 4).

Fig. 5.

The top enriched gene ontology (GO) terms categorized as either biological processes or molecular functions from each pairwise comparison for cane toads (Rhinella marina) exposed continuously to butachlor. Terms are ranked from left to right by the lowest FDR adjusted P-value and the largest percentage of differentially expressed, annotated transcripts represented by each GO term. Darker gray indicates more significantly enriched terms, where the largest percentages of DE transcripts were represented by the given GO term.

Experiment II: Acclimatization

We observed that cane toad tadpoles previously exposed to sub-lethal concentrations of butachlor (Experiment I) exhibited tolerance when they were exposed a second time to a lethal dose (Experiment II), but the magnitude of the effect, as represented by hazard ratios, varied depending on previous exposure period and concentration (Fig. 6). Cox proportional hazard models revealed that cane toads previously exposed to the high concentration of butachlor as eggs and hatchlings, to the medium concentration as tadpoles, and to the low concentration continuously had a significantly lower probability of mortality compared with controls when exposed to butachlor a second time (Table S4). Activity level, measured in Experiment II at 24 hours, did not differ significantly across treatments.

Fig. 6.

Nonparametric Kaplan–Meier survival curves depicting the percentage of cane toad (Rhinella marina) tadpole survival over a 32-hour exposure to a lethal dose of butachlor, 0.4 mg/L. The grey scale depicts the sub-lethal concentrations of butachlor animals were exposed to in Experiment I. Errors bars depict 95% confidence intervals and stars represent treatments where survival was improved compared to control animals (naive to butachlor) based on Cox proportional hazard regressions (P < 0.05, Table S4).

Discussion

Our results provide strong evidence in line with our hypotheses that exposure to environmentally relevant concentrations of butachlor can disturb development and growth in cane toads, particularly at early developmental stages. We observed lethal and sub-lethal effects on cane toad larvae well within the range of concentrations reported in rice fields after spraying (0.121 ± 0.051 mg/L to 0.357 ± 0.338 mg/L (Kaur et al., 2017; Ok et al., 2012), and which were consistent with results from our previous experiment (Shuman-Goodier et al., 2017). Surface water concentrations outside of rice fields are lower, ranging from no detection to 0.0003 mg/L, (Li et al., 2016; Mamun et al., 2009; Tang et al., 2018), perhaps due to strong absorption to soil (Armanpour & Bing, 2015). However, in a survey of surface water near agricultural soils in Hainan, China, butachlor remained one of the most frequently detected contaminants, and samples were found to exceed the regulatory threshold of 0.0001 mg/L (Huadong et al., 2020).

Concentrations of butachlor reported to be lethal to 50% of an amphibian population (LC50s) range from 0.53 mg/L (96 hrs, Microhyla ornate) to 2.67 mg/L (24 hrs, Polypedytes megacephalus) for amphibian embryos and larvae. However, sub-lethal effects of butachlor exposure on development, growth, and endocrine function have been observed at lower concentrations (Abigail et al., 2015; Geng et al., 2005; Li et al., 2016). Identifying sub-lethal effects, like the growth developmental and endpoints we targeted in this study, are important because they are directly linked to survival and fitness. For example, larger body sizes at metamorphosis improve chances of survival for cane toads (Cabrera-Guzmán et al., 2013).

For pesticide regulations to be protective of wildlife in rice ecosystems, it is crucial to consider a range of concentrations and sensitive life history stages. Testing concentrations that spanned three orders of magnitude allowed us to detect a non-monotonic dose response that occurred when animals were exposed as eggs and hatchlings; where the largest sub-lethal effects occurred when animals were exposed to the low concentration of butachlor (0.002 mg/L). This phenomenon is a consistent outcome when evaluating exposure to endocrine disrupting compounds, and may be explained in part by the adaptation of the vertebrate endocrine systems to respond to very small concentrations of hormones (Vandenberg et al., 2012). Similarly, our findings that cane toads were sensitive to low concentrations throughout a range of early development, but particularly as eggs and hatchlings, suggest that butachlor could produce long-term impacts that affect organisms later in life through organizational mechanisms. Chemicals that induce organizational effects typically cause irreversible structural alterations to tissues and organs during sensitive developmental windows early in life (Guillette et al., 1995). For amphibians that inhabit rice fields, breeding and development correspond with the application of pre-emergence herbicides and insecticides after fields are irrigated. Thus, direct exposure to pesticides like butachlor is most likely to occur during early developmental stages when amphibians are also vulnerable to organizational endocrine disruption.

Across the literature, butachlor exposure consistently affects development of exposed fish and amphibians, but in varying directions. In X. laevis tadpoles, exposure to 0.1 mg/L butachlor for 14 and 21 days promoted metamorphosis, affected expression of thyroid sensitive genes, and increased circulating thyroid hormone (T₃ and T₄) levels (Li et al., 2016). However, wild Fejervarya limnocharis tadpoles (a wild species commonly found in rice fields) exposed to butachlor at 0.025 and 0.1 mg/L developed slower (Liu et al., 2011); a result comparable to our own. Differences across experimental studies are likely due to physiological differences in species used for assessments (Leney et al., 2006), exposure timing and duration, and temperature. Notably, our study was the first to examine the morphology of the thyroid gland in response to butachlor exposure, and to utilize temperatures reflective of realistic exposure scenarios in tropical rice growing countries.

Differences in sub-lethal effects could also be due to undisclosed additives in commercial formulations of butachlor. Proprietary additives can interact with the main ingredient or cause additional toxicity, but are often overlooked in standard toxicology tests (Mesnage & Antoniou, 2018). Commercial formulations, such as the one we used (Machete EC), represent the final herbicide products applied in the environment, and to which inhabitants of rice fields are most realistically exposed. However, a comparison of the toxicity of analytical grade butachlor with the commercial formulation Machete EC in the freshwater fish, Channa punctate, did not find any evidence that the additives produce toxic effects beyond those of butachlor alone (Tilak et al., 2007).

The morphological differences in the thyroid glands of butachlor exposed animals (smaller thyroid glands, follicles, and thyrocyte cells) as well as the downregulation of genes related to anatomical structure development, metabolism, and cell differentiation explain the reduced developmental rates we observed in our study. The size of the thyroid is directly related to gland activity (Klaren et al., 2008) and the production of thyroid hormones (T3 and T4) that underlie vertebrate development and amphibian metamorphosis (Fort et al., 2007). Interestingly, we observed both up and down regulation of genes associated with cell differentiation (Fig. 4) and a proliferation of small follicles in cane toad tadpoles with the phenotypes most altered by butachlor (those exposed to the low concentration, Fig. 3). This may represent the beginnings of a compensatory thyroid response to maintain adequate production of thyroid hormones. For example, three-spine sticklebacks (Gasterosteus aculeatus) and African clawed frogs (X. laevis) chronically exposed to the EDC perchlorate during development compensate by increasing the number of thyroid follicles and the size of thyrocyte cells (Petersen et al., 2015; Tietge et al., 2005). Further evidence in support of a compensatory thyroid responses to butachlor exposure comes from our previous study exposing cane toad tadpoles to 0.2 mg/L, where we observed longer exposure time led to increased relative gene expression of thyroid hormone receptor beta (Shuman-Goodier et al., 2017). Compensatory responses could cause an acceleration in metamorphosis, if exposure were prolonged, which has been observed in the African clawed frog (Li et al., 2016).

In order for amphibians to acclimatize to pesticides, exposure concentrations would need to be high enough to induce adaptive gene transcription, but low enough that exposure did not cause harm to physiology, development and survival that outweigh adaptive stress and immune responses. Our findings that cane toads previously exposed to butachlor were more likely to survive a second, higher exposure than naive animals from the same population was supported by our RNAseq analyses for the continuously exposed animals. Comparing the transcriptome profiles of cane toads exposed continuously to low (0.002 mg/L) versus medium (0.02 mg/L) concentrations of butachlor provides an in-depth picture of the differentially expressed genes that may have facilitated acclimatization. For instance, genes that were only upregulated in animals exposed to the low concentration of butachlor but not in controls or animals exposed to the medium concentration, were enriched for functions pertaining to biosynthetic and metabolic processes. In addition, of the genes that were upregulated over 12-fold in animals exposed to the low versus the medium concentration of butachlor, one was GDI2, which codes for a protein involved in innate immunity, and another was carnitine-acylcarnitine translocase, which codes for a protein critical in lactate metabolic processes. The activation of specific immune and metabolic functions in this treatment (although the majority of these functions remained downregulated compared to controls) may explain why we only observed acclimatization when animals were exposed to a low concentration of butachlor (Tilak et al., 2007).

This research demonstrates that developmental assays using cane toad larvae can be used to screen for adverse effects of pesticide exposure in rice fields. Cane toads are abundant in rice fields at all life stages, and larvae are more sensitive to herbicide exposure than larvae of co-occurring native species (Shuman-Goodier et al., 2017). Their use for toxicology studies is also preferable over the use of native rice field amphibians, because they are an introduced species that may indirectly damage rice crops by consuming beneficial predators (Shuman-Goodier et al., 2019) and because their wide global distribution provides a standard animal for cross-country comparisons. The extensive genomic resources available for the cane toad, including a fully sequenced and annotated genome (Edwards et al., 2018), allow for comprehensive, mechanistic analyses of differential expression and physiological pathways. We suggest the cane toad can be used throughout their native and introduced range to test whether pesticides currently used in rice agriculture affect vertebrate development.

Conclusions

Exposure to environmentally relevant concentrations of a common formulation of the herbicide butachlor, Machete EC, affected development and growth of cane toad tadpoles: exposed animals developed slower and were smaller. A consistent non-monotonic dose response was observed when animals were exposed as eggs and hatchlings or continuously, where the lowest concentration produced the largest effect. In contrast, when animals were exposed only as tadpoles, the highest concentration produced the largest effect. Exposure affected the morphology of the thyroid gland and expression of genes responsible for biosynthetic processes, anatomical structure development, metabolism, signal transduction, cell differentiation, and stress and immune responses. Last, cane toads acclimatized to exposure, whereby animals exposed continuously to the low concentration, 0.002 mg/L, or as tadpoles to the medium concentration, 0.02 mg/L, of butachlor were more likely to survive a second, lethal exposure than naive animals. Our findings demonstrate that exposure to environmentally relevant concentrations of a popular commercial formulation of butachlor induce changes to tadpole development, morphology, and gene expression. Furthermore, we confirm that the cane toad can be used as an indicator of the physiological consequences of pesticide exposure in rice agroecosystems.

Highlights.

• Realistic exposure to butachlor affected development of cane toad tadpoles

• Animals developed slower, weighed less, and the thyroid gland was affected

• Non-monotonic response; low concentration, 0.002 mg/L, produced the largest effects

• Exposure caused extensive downregulation of transcripts

• Cane toad tadpoles can acclimatize to some herbicide exposures