Neonicotinoids are one of the most frequently used groups of synthetic pesticides of the twenty-first century and have been detected in surface waters in numerous countries (UNEP 2009). Neonicotinoids are a class of systemic insecticides that are taken up and translocated throughout a plant and provide in-plant protection from sucking insect pests, such as aphids, waterflies, leaf- and planthoppers (Gorgens et al. 2008; Kuhar et al. 2002). These insecticides were adopted in the early 1990s to replace carbamate and organophosphate compounds, which are more hazardous to nontarget organisms, particularly mammals (Elbert et al. 2008). Neonicotinoids share the same mode of action as nicotine in invertebrates and vertebrates, binding agonistically to the nicotinic acetylcholine receptors (nAChRs) in the nervous system interfering with acetylcholine neurotransmitter signalling. Neonicotinoids are more neurotoxic to invertebrates, because these insecticides have a greater affinity for invertebrate nAChRs compared with vertebrate nAChRs (Matsuda et al. 2001, 2005). Neonicotinoids are registered in 120 countries and in 2014 neonicotinoids represented more than 25% of the global pesticide market with imidacloprid, clothianidin, and thiamethoxam accounting for 85% of neonicotinoid sales (Bass et al. 2015; Craddock et al. 2019; Jeschke et al. 2011). Neonicotinoids are applied to crops via a variety of methods, including seed treatment, direct spray, soil drenches, and through irrigation systems. Surface level concentrations from 30 studies across nine countries, including Canada, more commonly range from 0.001 to 320 µg/L (Bishop et al. 2018; Crayton et al. 2020; Elbert et al. 2008; Main et al. 2014; Morrissey et al. 2015; Schaafsma et al. 2015; Yamamoto et al. 2012), thus pose a risk to nontarget aquatic wildlife including both invertebrate and vertebrate taxa (i.e., fish, birds, and amphibians).

Currently in Canada, neonicotinoids are undergoing re-evaluation for health and environmental effects and although the decision to protect pollinators was finalized in April 2019, re-evaluations on measures to protect aquatic life are still under special review (Health Canada 2020). Indeed, several studies have demonstrated adverse effects on honeybee populations (Christen et al. 2016; De Prisco et al. 2013; Straub et al. 2019), aquatic invertebrates (Anderson et al. 2015; Cavallaro et al. 2019; Miles et al. 2017; Morrissey et al. 2015), and whole organism effects in many vertebrate taxa (i.e., birds, rodents, fish, frogs; Gibbons et al. 2015; Miles et al. 2017; Morrissey et al. 2015; Roessink et al. 2013; Burke et al. 2018; Marlatt et al. 2019; Eng et al. 2017, 2019). For example, sublethal effects in fish have been reported for neonicotinoid low level impacts, such as early life-stage sockeye salmon exposed to 0.15 µg/L of clothianidin exhibiting elevated whole body 17β-estradiol concentrations (Marlatt et al. 2019) and early life-stage fathead minnows showing decreased body length and weight after exposure to 0.02 µg/L of clothianidin (US EPA 2010). Recent studies in Lithobates sylvaticus (wood frogs) showed 1, 10 and 100 µg/L imidacloprid in a commercial formulation increased survival and delayed metamorphosis at 10 and 100 µg/L (Robinson et al. 2017), but clothianidin and thiamethoxam at 2.5 and 250 µg/L in wood frogs and northern leopard frogs (Lithobates pipiens) had no effects (Robinson et al. 2019). In a field study with a caudate amphibian, a positive correlation between corticosterone concentrations in the skin of two salamander species (34 individual Desmognathus fuscus and 73 individual Desmognathus monticola) and concentrations of imidacloprid in waterways downstream of imidacloprid treated hemlock (Tsuga spp.) stands in West Virginia, United States was recently reported (Crayton et al. 2020). In addition, whole body samples from these two salamander species in waterways without detectable imidacloprid showed evidence of low levels of imidacloprid (mean 15.7 ng/g) providing evidence of bioaccumulation in amphibians (Crayton et al. 2020). However, no controlled laboratory studies have investigated the impact of neonicotinoids on a caudate amphibian. Indeed, the sensitivity of salamanders and caecilians, in general, to known environmental pollutants is poorly studied, and their range in North America and close association with the aquatic environment during sensitive early life-stage development makes them vulnerable nontarget species to aquatic pollutants, such as pesticides.

The native North American salamander, Ambystoma gracile (Northwestern salamander; NWS), is a widespread salamander species with a habitat range from southern Alaska to Northern California (Government of British Columbia 2017) and is a potential candidate model salamander to investigate the sensitivity of North American salamanders to environmental contaminants. The NWS is currently listed as least concern by International Union for Conservation of Nature (2015). It is a carnivorous facultative metamorph that reaches maturity in 1–2 years with adults being one of two morphotypes, a terrestrial sexually mature adult, or a paedomorphic aquatic adult form that is sexually mature with the retention of larval characteristics (i.e., external gills; Gould 1977; Government of British Columbia 2017; Licht and Sever 1991). This caudate species is commonly found breeding in small bodies of water and ephemeral ponds susceptible to run-off events, because they often are adjacent to agriculture (de Solla et al. 2002). However, only a handful of studies have investigated the life cycle of this species let alone its response to environmental pollutants (BC MOE 2014; Moreton and Marlatt 2019). As is the case with many Caudata and Gymnophiona species of amphibians (i.e., salamanders and caecilians, respectively), the modes of action and impacts of pollutants on survival, growth, and development of these taxa is not well documented. One study recently reported by Moreton and Marlatt (2019) showed that after 96 h, larval NWS were insensitive to an acute commercial formulation of diquat dibromide up to 33.72 mg/L. However, during a 21-d subchronic exposure larvae were considerably more sensitive with lethal and sublethal effects on growth occurring in the 1–2 mg/L range similar to larval fish sensitivity (Moreton and Marlatt 2019). Whether NWS demonstrates similar patterns of sensitivity to an insecticide, such as a neonicotinoid, has yet to be studied. This is of particular relevance to the development of protective environmental quality guidelines designed to identify contaminant concentration limits that should not be exceeded in order to be protective of all aquatic wildlife.

As with many amphibian species, metamorphosis in NWS development is primarily regulated by the hypothalamic-pituitary thyroid (HPT) endocrine axis and is mainly driven by temperature with individuals in ponds at higher temperatures reaching metamorphic climax more quickly than those at lower temperatures (Goldstein et al. 2017; Marlatt et al. 2013; Miyata and Ose 2012; Page et al. 2008; Thambirajah et al. 2019; Smith-Gill and Berven 1979; Stewart 1956). Indeed, thyroid hormone (TH) signalling is highly important in mediating molecular events contributing to morphological changes early in vertebrate development, including amphibian metamorphosis (Tata 2006). In particular, TH initiates a complex reorganization of physical characteristics (e.g., emergence and differentiation of limbs, tail resorption, and gastrointestinal reorganization) via the binding of TH to thyroid hormone receptors (TRs), specifically TRα and TRβ, that regulate gene expression patterns in numerous tissues leading to changes in developmental fate (Kikuyama et al. 1993; Marlatt et al. 2013; Shi 2000; Thambirajah et al. 2019). In anurans, TRβ is critical for intestinal remodelling and notochord resorption during metamorphosis, whereas TRα found more commonly expressed in the brain and hind limbs, plays a key role in regulating the timing of metamorphosis when unliganded via repressing target gene expression (Nakajima et al. 2019; Shibata et al. 2020; Wang et al. 2008; Wen et al. 2017). Alteration of thyroid function through potential action on disruption of the HPT axis, including binding to and altered expression of TRs, has been observed in amphibian laboratory exposures to various contaminants (Boone et al. 2013; Helbing et al. 2006; Howe et al. 2004; McKinney and Waller 1994; Moriyama et al. 2002). For example, larval Rana catesbeiana exposed to 1 nM or 10 nM acetochlor for 6 days increased expression of TRα and TRβ in the tail and altered the ratio of TRα to TRβ in the brain (Helbing et al. 2006). Additionally, Xenopus laevis tadpoles exposed to ethylenethiourea (ETU; commonly used in rubber and fungicide production) from 1 to 50 mg/L for up to 90 days exhibited a dose response relationship for thyroid gland (e.g., increased follicle size and colloid depletion), developmentally arrested tadpoles, and increased gene expression of thyroid stimulating hormone subunits α and β (Opitz et al. 2006). These studies show some of the potential contaminant impacts on the HPT axis via histological, developmental, and molecular measures commonly applied in Anuran toxicity studies. Yet, few studies have examined the impacts of pollutants on gene expression and developmental processes associated with the HPT axis in salamanders.

The objective of this study was to assess the acute and subchronic toxicity of imidacloprid and the subchronic toxicity of a mixture of neonicotinoids (imidacloprid, clothianidin, thiamethoxam) in the NWS, a North American salamander. This study examined concentrations surrounding and above typical environmental concentrations of neonicotinoids to better understand the lethal and sublethal impacts of these commonly used pesticides on NWS larvae survival, growth, and development. This is the first study, to our knowledge, that investigates neonicotinoid impacts on gene expression in the NWS and the first laboratory study to assess neonicotinoid impacts on apical and developmental endpoints in Ambystomatidae.

Methods

Animal Collection and Culture

Northwestern Salamander (Ambystoma gracile; NWS) egg masses were collected from a pond in the unceded territories of the Stó:lō First Nation currently known as Abbotsford, BC, Canada (49° 01′ 41.4′′ N 122° 17′ 05.9′′ W) under the British Columbia Ministry Environment permit (SU18-290881) and the Simon Fraser University Animal Care Committee Animal Care Protocol #1204B-16. The location of the pond where egg masses were collected is not directly adjacent to agricultural land but an urban landscape within the University of the Fraser Valley Abbotsford campus surrounded by a tree/shrub line of ~ 2 to 5 m. Upon collection of egg masses, there was no recorded use of pesticides, including neonicotinoids. We predict that it is unlikely that there is chemical contamination, except for a low risk of some urban runoff from the surrounding university buildings and parking (i.e., motor vehicle-related contaminants, such as copper, polycyclic aromatic hydrocarbons, etc.). Furthermore, although no analytical chemistry was conducted on water collected from this site, based on 6 years of monitoring this pond and observations of generally consistent numbers of egg masses per year (Marlatt and Danis, unpublished data), we concluded that the water quality and overall habitat was excellent for the NWS. Previous studies of this population and rearing NWS in captivity were optimized by Dr. V.L. Marlatt to demonstrate high survival (> 75%) of hatched larvae for up to 4 months under the following collection and rearing conditions (V.L. Marlatt, personal communication, March 10, 2018) and during acute and chronic (21 day) exposure experiments (Moreton and Marlatt 2019). Egg masses were collected when embryos reached developmental stage 28 (Harrison 1969; Moreton and Marlatt 2019) and were transported in pond water at 16 °C ± 1 °C—the same temperature as the pond water where egg masses were collected—to Simon Fraser University within 2 h of collection. Egg masses were placed in separate 15-L glass aquaria in native pond water with gentle aeration under a photoperiod of 12-h light: 12-h dark and acclimated over a 48-h period from 16 to 18 °C ± 1 °C. After the 48-h acclimation period, a 100% water renewal was performed with municipal dechlorinated water. Water quality (conductivity, dissolved oxygen, pH, and temperature) was monitored daily until hatching using a HQd portable meter (Hach Company, Loveland, CO) and hatched larvae were transferred into 8-L glass aquaria at a loading density of 0.8 g/L suggested for amphibians at water temperatures of 17 °C according to the American Society for Testing and Materials (1996). Ammonia levels in tanks were monitored on a weekly basis. All embryos hatched within 6–12 days after collection and larvae less than 5 days old (developmental stages 40–45; Harrison 1969) were used in the 96-h acute exposure and the modified OECD 21-day amphibian metamorphosis assay (AMA, Test No. 231) subchronic exposures.

Chemicals

All neonicotinoid stock solutions were prepared fresh daily. Clothianidin stock was prepared using ≥ 98.0% pure clothianidin (CAS#: 210,880–92-5, Sigma-Aldrich, Oakville, Ontario, Canada) dissolved in dechlorinated municipal water. Clothianidin is soluble in water (0.327 g/L at 20 °C; US EPA 2003) and has an aqueous photolysis half-life ranging from 0.35 to 3.31 days in simulated mesocosm systems in four seasons in Winnipeg, Canada (Lu et al. 2015) and 25.1–27.7 h in river water in a photoperiod of 9-h light: 15-h dark (Rexrode et al. 2003). Imidacloprid stock was prepared using ≥ 98.0% pure imidacloprid (CAS#: 138,261–41-3, Sigma-Aldrich) dissolved in dechlorinated municipal water. Imidacloprid is soluble in water (0.61 g/L at 20 °C; Krohn and Helpointner 2003; Tomlin 2005), and the parent compound has an aqueous photolysis half-life of 28 h in pond mesocosms (Colombo et al. 2013) and reported as 22 days in water (SERA 2005). Thiamethoxam stock was prepared using ≥ 98.0% pure thiamethoxam (CAS#: 153719-23-4, Sigma-Aldrich) dissolved in dechlorinated municipal water. Thiamethoxam is soluble in water (4.1 g/L at 25 °C; MacBean 2010) and has a half-life of 7.9–39.5 days in surface waters (MacBean 2010). To ensure no photodegradation of neonicotinoids by UV A/B, exposures were conducted in a facility where rooms had no windows or source of UV light that would cause photodegradation (Kurwadkar et al. 2016). In addition to the global high usage rate of neonicotinoids, these pesticides were selected for this study, because they are commonly found locally in the Fraser Valley, BC, which overlaps with the distribution of NWSs (Bishop et al. 2018; British Columbia Ministry of Agriculture 2009, 2012, 2017, 2018). To prepare test vessel water for waterborne NWS larval exposures, 20-L food grade plastic containers were used for mixing neonicotinoid stock solutions with dechlorinated municipal water before addition to glass exposure aquaria. For example, to achieve a 1000-µg/L imidacloprid test vessel exposure concentration, 0.4 L of a 50-mg/L imidacloprid stock solution was added to 19.6 L of water and mixed thoroughly, and 5 L was dispensed into the glass exposure aquarium.

General Experimental Set-Up

NWS larvae exposures took place in 5-L glass tanks measuring 30-cm × 15-cm × 20.5-cm. Exposed sides of the tanks were covered with black plastic to shield larvae from visual disturbances. Tanks were filled with 5 L of water or neonicotinoid(s) dissolved in water. Water quality (conductivity, dissolved oxygen, pH, and temperature) was monitored daily using a HQd portable meter in one of four replicate tanks for each treatment and the water control throughout the acute (96 h) and subchronic (35 d) exposure period. Ammonia levels were measured on the final day before the exposure was terminated in the 96-h exposure and weekly in the 35-d exposure using Seachem MultiTest Ammonia Test Kit (seachem Laboratories, Madison, USA; detection limit 0.05 mg/L). Larvae were fed ad libitum with a mixture of dead Chironomidae larvae (bloodworms; Hikari USA) and dead Mysis diluviana (opossum shrimp; Piscine Energetics) daily. A 12-h light: 12-h dark photoperiod was maintained throughout the experiment and survival checks were performed twice daily (once in the morning, and once in the afternoon). At the end of each exposure period, larvae were euthanized by immersion in 0.4 g/L of tricaine methanesulfonate buffered with sodium bicarbonate to pH 7.0–7.5.

Acute Toxicity of Imidacloprid to Northwestern Salamander Larvae

The experimental design and exposures were adapted based on Test No. 231: The Amphibian Metamorphosis Assay (OECD 2009). Changes included the use of NWS instead of Xenopus spp. and an early termination point of 96 h rather than 21 d to evaluate short-term changes in gene expression as well as apical endpoints. In the 96-h acute exposure stage, 40 (4–7 d old) NWS larvae were randomly assigned to one of the following waterborne treatments: dechlorinated municipal water control; imidacloprid at a concentration of 250, 750, 2250 or 6750 µg/L dissolved in dechlorinated municipal water. Four replicate glass tanks per treatment with three larvae per tank (n = 12 larvae/treatment). After larvae were euthanized, developmental stage was determined using a stereo microscope and stage system by Harrison (1969) using forelimbs as markers. In addition, total length (mm), snout-vent-length (SVL, mm), wet weight (g), and abnormalities were recorded including evidence of cannibalism (missing tail tips), and two whole-body NWS larvae per replicate were immediately frozen on dry ice and stored at − 80 °C for subsequent gene expression analyses. Although we do not expect major development occurring after 4 days, we have little data at this time and can state the change in total length and SVL in larval NWSs between 5 and 11 days is 2.21 ± 0.13 mm and 0.602 ± 0.11 mm, respectively (unpublished data).

Subchronic Toxicity of Neonicotinoids to Northwestern Salamander Larvae

The experimental design and exposures were adapted based on Test No. 231: The Amphibian Metamorphosis Assay (OECD 2009), and changes included the use of NWS instead of Xenopus spp. and a 35-d termination time point rather than 21 d to account for the relatively longer progression of developmental stages in the NWS. In the 35-d subchronic exposures, four or five stage 40 (4–7 d old) NWS larvae were randomly assigned to one of two exposures: (i) imidacloprid at 10, 100 or 1,000 µg/L dissolved in dechlorinated municipal water and a dechlorinated municipal water control; (ii) an equal mixture of imidacloprid, clothianidin, and thiamethoxam (ICT) at a concentration of 10, 100 or 1000 µg/L dissolved in dechlorinated municipal water and a dechlorinated municipal water control. The neonicotinoids selected for the mixture study were based on the prevalence of co-occurrence of these three neonicotinoids in a Canadian agricultural region, the prairie pothole region (Main et al. 2014). We expect similar co-occurrence and similar µg/L concentrations of these three neonicotinoids in many parts of Canada based on sales records of these pesticides (BC MOE 2010, 2015), and the concentrations selected captured these environmentally relevant concentrations as well as higher concentrations to better understand the concentrations effecting growth and development. Four replicate glass tanks per treatment with 4–5 larvae/tank (n = 16–20 larvae/treatment) were included. Water renewals of 80% were performed every 48 or 72 h on a Monday, Wednesday, and Friday weekly cycle. After larvae were euthanized total length (mm), snout-vent-length (SVL, mm), wet weight (g), and abnormalities were recorded including evidence of cannibalism (missing tail tips). A subset of 9–11 larvae from each treatment were preserved in Davidson’s fixative according to OECD (2007; a formalin-based fixative) for 48 h and then transferred to neutral buffered formalin (NBF) for long term storage (Grim et al., 2009) for subsequent developmental staging.

Developmental Staging Analysis

At the onset of the experiment the Harrison (1969) developmental guide for A. maculatum was used; however, upon the completion of the 35-d exposure period, the majority of the NWS larvae were beyond the described stages outlined by Harrison (1969). Therefore, conservative developmental stage categories were developed based on weekly live and preserved NWS larvae observations using a stereo microscope at 20× magnification (Danis et al. in prep) based on developmental stage delineations reported for similar salamanders by Nye et al. (2003; A. mexicanum) and Russell and Watson (2000; A. maculatum). Briefly, upon termination of this experiment after 35 days, the 9–11 larvae per treatment preserved in neutral buffered formalin (NBF) were examined using a stereo microscope to determine developmental stage based on the morphology of each forelimb and hindlimb. Specifically, to assess the stage of each limb, forelimb digit and hindlimb digit number was recorded. Forelimb development was categorized into five developmental stages (Supplementary data, Table S1): club/stub (forelimb present but no individual digits evident and rounded distal end); two digits (forelimb with 2 distinct digits); third digit formation (digit 3 on forelimb begins to develop and appears as a swelling or three digits evident); fourth digit formation (digit 4 on forelimb begins to develop and appears as swelling); and digit growth (4 distinct digits on forelimb). Hindlimb development was also categorized into five developmental stages (Supplementary data, Table S1): emergence and growth of hindlimb (hindlimb buds appear as thickenings on the flanks posterior to the vent); two digits (hindlimb has single indentation or 2 distinct digits); third digit formation (3rd digit appears as fibular bulge/rounded swelling); third digit indentation (indentation formed between 3rd digit and 2nd digit); three digits (the hindlimb bears 3 distinct digits). Developmental staging was conducted blindly by one observer and 10% of the individuals were blindly checked a second time by the same observer and by a second observer to account for inter- and intra-observer variability. Developmental categories were consistent across observations.

Isolation of RNA, cDNA Synthesis, and Real-Time Quantitative Polymerase Chain Reaction (qPCR)

Whole body gene expression analyses were performed on cDNA prepared from the larvae in the 96-h acute imidacloprid exposure experiment (n = 3 larvae per replicate tank for each treatment; 4 tanks/treatment). Total RNA isolation was performed by homogenizing whole NWS larvae in 1.0 mL of TRIzol® Reagent (Invitrogen, Burlington, ON, Canada) and two 1-mm tungsten-carbide beads in a 2.0-mL RNase free safe-lock Eppendorf microcentrifuge tube using a Retsch Mixer Mill MM 400 (Fisher Scientific, Ottawa, ON, Canada) set to 30 Hz for 12 min (chambers were stopped after 6 min and rotated 180° and then continued for another 6 min). Total RNA was then isolated according to TRIzol® Reagent manufactures protocol. The final RNA obtained from the TRIzol® RNA isolation procedure was reconstituted in 30 µL of DNAase/RNase-free water and stored at − 80 °C. Quantity and purity of RNA were determined by measuring the optical density unit and ratios (OD260, OD260/280, and OD260/230) with an Epoch 2 Microplate Spectrophotometer and Gen5 2.06 read software (BioTek, Winooski, VT).

For each RNA sample, a total of 0.75 µg of RNA was DNase treated using TURbO DNA-free kits™ (Ambion, Austin, TX) to remove any coextracted DNA. The quality of DNAse treated RNA was then evaluated using a Bio-Rad Experion™ Automated Electrophoresis System and Experion software (version 3.20; Bio-Rad, Mississauga, ON, Canada). The RNA integrity values (RIN's) for all samples used in this study ranged from 8–9.7. In accordance with Minimum information for publication of quantitative real-time PCR experiments (MIQE) guidelines outlined in Bustin (2010) total DNase-treated RNA samples devoid of contamination (i.e., OD260/280 and OD260/230 ratios ranging from 1.8–2.1) and RNA degradation (i.e., RIN scores > 8.0) were used in subsequent qPCR experiments. All DNase treated RNA samples were stored at − 80 °C until cDNA synthesis. Reverse transcription of 0.75 µg of DNase-treated total RNA into cDNA was performed using the qScript™ cDNA SuperMix (Quanta Biosciences, Beverly, MA) according to the manufacturer’s instructions. All cDNA samples were stored at − 20 °C until subsequent qPCR experiments.

Relative gene expression analysis was performed using qPCR according to methods previously described by Marlatt et al. (2019) and MIQE guidelines (Bustin et al. 2009; Bustin 2010) using the Bio-Rad CFX384TM Real-Time PCR Detection System and CFX ManagerTM Software in 384-well plates. The relative quantification of each target gene of interest in whole body A. gracile larvae cDNA sample was measured using the ΔΔCq method via the Bio-Rad CFX384™ Real-Time PCR Detection System and CFX Manager™ Software (Bio-Rad Laboratories Ltd, Mississauga, ON, Canada), including normalizing the data to 3 reference genes. Expression analysis demonstrated that the combined M-value for three reference genes (cytoplasmic β-actin, (CBA), glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and ribosomal protein lateral stalk subunit P0 (RPLP0)) was 0.41 with a coefficient of variation (CV) = 0.15. Thus, all three of these reference genes were used for normalization purposes for all target genes tested in this study. The gene expression targets examined in this study, thyroid receptor alpha (TRα) and thyroid receptor beta (TRβ), were selected because of their relevance to various biological processes, including growth and development. A complete list of these genes is presented in Table 1. Primers with optimal annealing temperature between 55 and 58 °C were designed using Integrated DNA Technologies (IDT) OligoAnalyzer 3.1 (www.idtdna.com/calc/analyzer) and sequences obtained from GenBank National Center for Biotechnology Information (https://www-ncbi-nlm-nih-gov.proxy.lib.sfu.ca/) database for the NWS when available or from closely related salamander species. Primer sets were tested for efficiency using a 4- to 8-point standard curve generated by a fourfold dilution of a 50-ng cDNA/µL of water (comprised of whole body samples from larval NWS). The criteria for acceptance of the standard curve included in single peak melt curve, efficiencies between 90 and 110%, amplification in at least four concentrations of the standard curve, and an R2 of the standard curve > 0.900 (Bustin 2010). Primers that satisfied these criteria are shown in Table 1, which also specifies product size, PCR efficiency and correlation coefficient (R2). Primer sets were evaluated for specificity by conducting PCR using a T100 Thermal Cycler (Bio-Rad; activation at 95 °C for 10 min, denaturing at 94–96 °C for 30 s, annealing at 52 °C for 30 s, extension at 72 °C for 1 min, for 42 cycles) and Immolase™ DNA Polymerase (Meridian Bioscience, Cincinnati, OH) for each primer set on NWS larval cDNA and each amplicon was sequenced. All sequencing was conducted using Sanger Sequencing at the University of British Columbia Sequencing and Bioinformatics Consortium (Vancouver, BC, Canada). Sequencing results were aligned using the ClustalW2 nucleotide alignment tool (https://www.ebi.ac.uk) with a closely related Ambystoma species (A. mexicanum) nucleotide sequence obtained from GenBank National Center for Biotechnology Information for each gene. All qPCR reactions for each gene were performed according to Marlatt et al. (2019) on the Bio-Rad CFX384TM Real-Time PCR Detection System using three technical replicates per individual larval sample.

Table 1 Primer sets used to examine the acute effects of imidacloprid on various target genes in Ambystoma gracile early stage larvae
Full size table

Statistical Analysis

Visual inspection of raw data was conducted to determine evidence of outliers. Potential outliers were converted to z-scores and omitted if they fell outside a ± 3 range (Cousineau and Chartier, 2010). All statistics were carried out in program R v3.0.2 (Bates et al., 2015; R Core Team, 2013). Normality for each data set was assessed by plotting data for visual inspection along with a Shapiro–Wilk test, and a Levene’s test was used to evaluate homogeneity of variance. Data was log10 transformed if it failed test parameters and retested for normality. To evaluate effects on survival, we used a GLMM with binomial distribution and logit link function (R package “lmer4”). Linear mixed models (using the “nlme” package in R) with Gaussian distributions were used to determine the effects of treatment on total length (L, mm), snout-vent-length (SVL, mm), wet weight (W, g), and condition factor (K). Condition factors for NWS larvae were calculated using Fulton’s condition factor (K) where K = 100(W/L3) (W = total wet body weight and L = snout to tail body length; Fulton, 1904). In all LMM analyses, nonindependence of larvae in treatment tanks was accounted for by including tank replicate as a random effect to determine the independence from fixed effects (treatments). We checked the diagnostics for all our models following Zuur et al. (2013) using residual plots, quantile–quantile plots, and density plots to assess homogeneity of variance and goodness of fit.

The relative quantification of each gene of interest in NWS larvae samples from the 96-h imidacloprid exposure measured in qPCR experiments was obtained using the ΔΔCq method, and by normalizing this data to the expression levels of three reference genes in each sample (CFX Manager™ Software Gene Expression Analysis). Differences in normalized gene expression between imidacloprid treatments were determined by one-way analysis of variance (ANOVA; p < 0.05) followed by a Tukey's post-hoc.

Developmental stage in preserved larvae from the 35-d imidacloprid and mixture exposure were analysed using Fisher’s exact tests to determine differences between neonicotinoid treatments and the water control. Individual Fisher’s exact tests were performed on the following for the 35-d imidacloprid experiment: right forelimb; left forelimb; right hindlimb; and left hindlimb. Within each control treatment of imidacloprid and ICT mixture exposures, both left and right forelimbs, and left and right hind limbs were compared to determine whether there was evidence of natural background asynchronicity in development of untreated animals.

Results

Water Quality

In the 96-h imidacloprid exposure, the range of water quality values were: dissolved oxygen from 8.94 to 9.39 mg/L; pH from 7.03 to 7.51; conductivity from 36.9 to 52.4 µS/cm; water temperature ranged from 15.9 to 17.4 °C, and ammonia levels were below detection limit (< 1 µg/L; Table 2). In the 35-d imidacloprid only exposure the range of water-quality values were: dissolved oxygen from 8.77 to 10.26 mg/L; pH from 6.64 to 7.68; conductivity from 30.4 to 34.7 µS/cm; water temperature ranged from 13.7 to 19.8 °C; and ammonia levels ranged from 0.0184 to 0.618 µg/L (Table 2; below British Columbia’s Water Quality Guideline value of 1920 µg/L at a pH of 7; CCME 2001). Lastly, water quality values in the 35-d ICT equal part mixture exposure the range of water quality values were: dissolved oxygen from 8.56 to 10.08 mg/L; pH from 6.48 to 8.64; conductivity from 30.7 to 61.4 µS/cm; water temperature ranged from 14.4 to 21.5 °C; and ammonia levels ranged from 0.238 to 0.640 mg/L (Table 2). Dissolved oxygen and pH values were within the acceptable range of values highlighted in the OECD Test No. 231 across all experiments (OECD 2009), and on any given day the temperature measured in each pesticide test concentration replicate did not vary more than 1 °C from the control replicate for all experiments (Supplemental Fig. S1).

Table 2 Water quality from 96-h and 35-d waterborne, static renewal neonicotinoid exposure experiments of A. gracile larvae
Full size table

Acute Toxicity of Imidacloprid to Northwestern Salamander Larvae

No mortality of NWS larvae was observed in the four replicate control tanks (3 larvae/tank) during the 96-h imidacloprid exposure experiment. There were no significant differences in survival between any treatments for this acute exposure (data not shown; p > 0.05). There was little evidence of cannibalism across all treatments in the 96-h imidacloprid exposure (average evidence of cannibalism per tank ± SE was 4.26 ± 2.42%, range = 0% to 12%). In addition, there was no evidence of acute effects of imidacloprid on mean wet weight (ranging from 0.028 to 0.047 g; data not shown; p > 0.05), total body length (ranging from 12 to 18 mm; data not shown; p > 0.05), or mean snout-to-vent length (ranging from 8 to 15 mm across all treatments; Supplementary Table S2; p > 0.05). However, mean condition factor (K) ± SE of larvae in the 250 (4.70 ± 0.20; p = 0.04) and 750 (4.56 ± 0.11; p = 0.05) µg/L imidacloprid treatments was increased compared with the control larvae (3.83 ± 0.28).

Whole body mRNA expression levels of TRα were not significantly different between treatments in the 96 h imidacloprid exposure (Fig. 1a; p > 0.05). However, a significant 1.70- and 2.33-fold decrease in TRβ transcript levels was detected in the 750 and 2250 µg/L imidacloprid treated larvae, respectively, compared with the water control larvae (Fig. 1b; p = 0.009 and 0.0003, respectively). The average normalized expression ± SE for the 750 µg/L imidacloprid was 0.00364 ± 2.64 × 10−4 and 0.00265 ± 5.22 × 10−4 for the 2,250 µg/L imidacloprid exposure compared with the water control (0.00619 ± 7.2 × 10−4).

Fig. 1
figure 1

Effects of waterborne 96-h imidacloprid exposures on a thyroid receptor alpha (TRα) and b thyroid receptor beta (TRβ) whole-body gene expression levels in Ambystoma gracile larvae. Normalized gene expression was calculated based on the ΔΔCq method for relative quantitation of a target gene using three reference genes (β-actin, GAPDH, and RPLP0). Box plots depict the medians (horizontal line), the lower and upper 25th and 75th percentiles (boundaries of the box), the minimum and maximum values (whiskers), and outliers (black circles). Biological replicates ranged from 7 to 8 per imidacloprid treatment (µg/L: 0, n = 8; 250, n = 7; 750, n = 8; 2250, n = 7; 6750, n = 8). Different letters indicate significant differences between treatments (one-way analysis of variance followed by Tukey’s post-hoc test, p < 0.05)

Full size image

Subchronic Toxicity of Neonicotinoids to Northwestern Salamander Larvae

Exposure to imidacloprid did not affect survival of NWS larvae across all subchronic treatments (data not shown; p > 0.05). Furthermore, the control mean (± standard deviation) survival was 95% ± 8.7%. This was similar to the survival in imidacloprid treatments (Imid-10 = 88.75 ± 11.4%, Imid-100 = 100 ± 0%, Imid-1000 = 95 ± 8.66%). There was little evidence of cannibalism across all treatments in the 35-d imidacloprid exposure (average evidence of cannibalism per tank ± SE was 6.80 ± 3.04%, range = 0–17%), except for one organism in the control treatment that was removed from analysis based on visual plots and a z-score that was well above acceptable variation in data (z-score = 6.93 > 3). In addition, there was no observed effect of subchronic exposures on mean total length (ranging from 13 to 30 mm; Fig. 2a; p > 0.05), mean snout-to-vent length (ranging from 7 to 18 mm; Fig. 2c; p > 0.05), mean wet weight (ranging from 0.023 to 0.194 g; Fig. 2e; p > 0.05), and mean condition factor (K; ranging from 1.304 to 11.8; data not shown; p > 0.05). Similarly, NWS larvae exposed to an equal part mixture of imidacloprid, clothianidin, and thiamethoxam (ICT) for 35 d at 10, 100 or 1000 µg/L mixture, or a water control had no significant impact on survival (Supplemental data, Table S2; p > 0.05). Specifically, the control mean survival was 90 ± 17.3% and higher survival was observed in the ICT mixture treatments (ICT-10 = 100 ± 0%, ICT-100 = 93 ± 10.83%, ICT-1000 = 95 ± 8.66%). There was little evidence of cannibalism across all treatments in the 35-d ICT mixture exposure (average evidence of cannibalism per tank ± SE was 1.67 ± 1.49%, range = 0–8.3%). In addition, there was no observed effect of subchronic exposures for any ICT concentrations on mean total length (ranging from 14 to 29 mm; Fig. 2b; p > 0.05), mean snout-to-vent length (ranging from 8 to 17 mm; Fig. 2d; p > 0.05), mean wet weight (0.022 to 0.128 g; Fig. 2f; p > 0.05), and mean condition factor (ranging from 0.948 to 6.25; data not shown; p > 0.05) compared with water controls (Supplemental Data, Table S2).

Fig. 2
figure 2

Mean a, b total length, c, d snout-to-vent length and e, f wet weight of Ambystoma gracile after 35-day exposures to waterborne imidacloprid (0, 10, 100, 1000 µg/L), and an equal part mixture of imidacloprid (Imid.), clothianidin (Cloth.), and thiamethixam (Thia.) (ICT; 0, 10, 100, 1000 µg/L). Box plots depict the medians (horizontal line), the lower and upper 25th and 75th percentiles (boundaries of the box), the minimum and maximum values (whiskers), and outliers (black circles). No significant differences were observed (generalized linear mixed model; p < 0.05)

Full size image

Developmental stage was assessed based on all four limbs independently. To investigate natural variation in development in this species, statistical analyses were performed on control group larvae from the 35-d imidacloprid and ICT mixture experiments. These analyses revealed that the right forelimb developmental stage in control larvae from the 35-d imidacloprid only exposure had more individuals at the club/stub developmental stage (72%) compared with the left forelimb (18%; p = 0.043; data not shown). Based on these findings from the control animals in this experiment, it appears that early in development (i.e., 40-d post-hatch at ~ 16 °C) asynchronous left and right forelimb development may naturally occur. However, upon comparing left and right forelimb development within the control larvae from the 35-d ICT mixture experiment, the control animals showed no differences in left and right forelimb developmental stage (data not shown; p > 0.05). In addition, when pooling the control larvae from both 35-d exposure experiments, there were no differences in the developmental stage between left and right hindlimb and forelimb (n = 21, data not shown; p > 0.05). Comparisons of developmental stage between all treatments in the 35-d imidacloprid experiment revealed that imidacloprid alone did not alter left forelimb development (Fig. 3a), but 100 µg/L imidacloprid significantly advanced development in the right forelimb (p = 0.0026; Fig. 3b). However, the ICT mixture did not cause any changes in limb development during the 35 d ICT experiment (p > 0.05; Fig. 4).

Fig. 3
figure 3

Proportion of Ambystoma gracile larvae at five different developmental stage categories after 35-d continuous waterborne exposure to imidacloprid based on individual limb development: a left forelimb; b right forelimb; c left hindlimb; and d right hindlimb. Different shading indicates developmental stage categories indicated on the right side of figures. Five developmental stages of forelimb: club/stub (forelimb present but no individual digits evident and rounded distal end); 2 digits (forelimb with 2 distinct digits); 3rd digit formation (digit 3 on forelimb begins to develop and appears as a swelling or 3 digits evident); 4th digit formation (digit 4 on forelimb begins to develop and appears as swelling); and digit growth (4 distinct digits on forelimb). Hindlimb development was also categorized into 5 developmental stages: emergence and growth of hindlimb (hindlimb buds appear as thickenings on the flanks posterior to the vent); 2 digits (hindlimb has single indentation or 2 distinct digits); 3rd digit formation (3rd digit appears as fibular bulge/rounded swelling); 3rd digit indentation (indentation formed between 3rd digit and 2nd digit); 3 digits (the hindlimb bears 3 distinct digits). Significant difference between the control versus treatment at a given imidacloprid concentration is indicated by an asterisk (0.01 < p < 0.05)

Full size image
Fig. 4
figure 4

Proportion of Ambytsoma gracile larvae at five different developmental stage categories after 35-d continuous waterborne exposure to equal part imidacloprid (Imid.), clothianidin (Cloth.), and thiamethoxam (Thia.) (ICT) based on individual limb development: a left forelimb; b right forelimb; c left hindlimb; and d right hindlimb. Different shading indicates developmental stage categories indicated on the right side of figures. Staging categories are identical to those highlighted in Fig. 3 caption and Developmental staging section of methods. Significant difference between the control versus treatment at given ICT concentration is indicated by an asterisk (0.01 < p < 0.05)

Full size image

Discussion

In this study, we have shown expedited development in NWS larvae after exposure to 100 µg/L imidacloprid for 35 days, providing some evidence of disruption of the thyroid endocrine axis at an environmentally relevant concentration. In addition, we found changes in the developmental gene expression pattern based on decreased gene expression of TRβ after acute exposures to imidacloprid (750 and 2250 µg/L), although this occurred at concentrations two to seven times higher than typically reported environmental concentrations. Furthermore, our experiments demonstrate that imidacloprid alone does not cause mortality up to 6750 µg/L, nor does an equal parts mixture of imidacloprid, clothianidin and thiamethoxam up to 1000 µg/L. These lethality data are well below several water quality guidelines including, the CCME water-quality guideline for imidacloprid of 0.23 µg/L, the USEPA aquatic life benchmarks of 35 (acute) and 1.05 µg/L (chronic), and the range of neonicotinoid concentrations reported in various surface waters in North American, Asian, and Australian studies (0.0035–320 µg/L) (CCME 2007; Crayton et al., 2020; de Perre et al., 2015; Hladik et al., 2014; Main et al., 2014; Miles et al., 2017; Morrissey et al., 2015; Sánchez-bayo and Hyne, 2014; Schaafsma et al., 2015; Starner and Goh, 2012; Van Dijk et al., 2013). Collectively, direct impacts of imidacloprid and a mixture of imidacloprid and clothianidin and thiamethoxam on survival of NWS larvae through pre- and pro-metamorphosis at typical environmental levels are unlikely. Ultimately, there is a paucity of studies conducted examining the sensitivity of salamanders to pollutants. Thus, this study reports novel findings that will contribute to understanding the sensitivity of a Caudate amphibian model to a common environmental pollutant and demonstrates the potential of this species in laboratory toxicity testing.

The acute toxicity of neonicotinoids in several frog models generally corroborates the findings of the present study showing that the LC50 of imidacloprid is > 6750 µg/L in the NWS and that genotoxic effects and gene expression changes occur at concentrations less than their respective LC50s. For example, the LC50 in ~ 1-month-old tadpoles exposed to imidacloprid for 96 h was 82,000 µg/L and 129,000 µg/L for Hypsiboas limnocharis and Pelophylax nigromaculata, respectively (Feng et al. 2004). These aforementioned frog species studied by Feng et al. (2004) were wild collected near Nanjing city in China. In additional 7-d waterborne exposures by Feng et al. (2004) with P. nigromaculata tadpoles to 2000, 8000 and 32,000 µg/L of imidacloprid, blood was collected. An erythrocyte micronucleus assay revealed that 8000 and 32,000 µg/L caused genotoxic effects. Further in vitro assays incubating adult frog (P. nigromaculata) erythrocytes for 1 h with imidacloprid ranging from 50 to 500 µg/L also demonstrated DNA damage measured via the Comet assay occurred at all concentrations in a dose-dependent manner (Feng et al. 2004). Taken together, Feng et al. (2004) showed that genotoxic impacts after in vivo and in vitro erythrocyte experiments occur at concentrations at least 10 times lower than the reported LC50 for P. nigromaculata. Similar results have been observed in a wild collected Argentinian frog species, Hypsiboas pulchellus, in experiments initiated at Gosner stage 36 (prometamorphosis) during acute exposure to a commercial formulation of imidacloprid, reporting a 96 h LC50 of 48,470–58,180 µg/L and DNA damage occurring in erythrocytes at concentrations as low as 500 µg/L using the Comet assay (Pérez-Iglesias et al. 2014). Ruiz de Arcaute et al. (2014), also observed genotoxic effects via the micronucleus assay and Comet assay on peripheral circulating blood erythrocytes collected from these H. pulchellus larvae in the range of 15,000–30,000 µg/L of 95.1% pure imidacloprid, again much lower than the LC50 ranging from 77,200 to 93,040 µg/L. To date, it appears as though the concentrations that cause genotoxic effects in these anuran species are still much higher than what would likely be encountered in the environment.

In the present study, decreased gene expression of TRβ was observed at 750 and 2250 µg/L imidacloprid in NWS larvae after 96-h exposures, but not at the higher (6750 µg/L) or lower concentration tested (250 µg/L). The alteration of mRNA expression of receptors for thyroid hormone that are key components of the HPT axis indicates that imidacloprid has the potential to disrupt this endocrine axis, albeit at concentrations approximately two to three times higher that environmental concentrations reported in in North America. It is possible that temporal changes in TRβ throughout the 96-h imidacloprid exposure were not captured in the present study. However, the nonmonotonic concentration–response relationship observed in TRβ induction in the present study has been observed for many endocrine disruptors and endogenous hormones. Although the underlying explanation for this non-monotonic phenomenon is beyond the scope of the present study, it has been shown that genes containing estrogen or thyroid hormone response elements in their promoters may include one or more of the following underlying modes of action that may account for this: negative feedback regulation mechanisms related to physiological control of hormone actions; hormone receptor desensitization; plurality of molecular targets; and/or, metabolic modulation of hormones (Lagarde et al. 2015). Amphibians undergo a complex reorganization of physical characteristics (e.g., emergence and differentiation of limbs, tail resorption, and gastrointestinal reorganization) that are controlled in large part by thyroid hormones, which exert their actions via thyroid hormone receptors (Kikuyama et al. 1993; Marlatt et al. 2013; Shi 2000). It is well known that the expression of TRα in anuran development and metamorphosis varies depending on tissue type and throughout development (Navarro-Martín et al. 2012; Tata 2006; Thambirajah et al. 2019). For example, an increase in TRα gene expression in the head, middle portion of the body, tail, and skin occurs through to premetamorphosis and is followed by a sharp decrease when tail resorption begins to occur at metamorphic climax in Xenopus spp., a well studied anuran model (Kawahara et al. 1991; Suzuki et al. 2009; Tata 2006; Wen et al. 2017). Similarly, TRβ mRNA expression levels in Xenopus spp. are low early in anuran development and increase drastically in the head, body, tail, and skin when tadpoles reach premetamorphosis and tail resorption at metamorphic climax (Kawahara et al. 1991; Suzuki et al. 2009; Tata 2006; Wen et al. 2017). Although few studies of TR expression in caudates are reported, one study suggests a possible decreasing trend of TRs as development proceeds. Specifically, a study in Ambystoma mexicanum, a caudate more closely related to NWS, exhibited a decrease of TRα and TRβ mRNA expression in skin from the head region in a linear trend shortly after hatching and during ongoing exposure to exogenous waterborne thyroxine (T4) at 50 nM for 30 days during pre- and pro-metamorphosis (Page et al. 2009). Although this down-regulation of TRs maybe be due to exogenous T4 exposure causing inhibition of the HPT axis overall including TR expression, it does more closely mirror the results of the present study whereby whole body TRβ decreased after 96-h exposure to imidacloprid at two intermediate concentrations tested (i.e., 750 and 2250 µg/L). It is hypothesized that this change in TRβ gene expression upon exposure to waterborne imidacrloprid may have contributed to the advanced development observed in the NWS 35-d imidacloprid exposure in the 100-µg/L imidacloprid treatment. Nonetheless, due to the paucity of data on NWS development and thyroid-mediated gene expression profiles during development, additional studies are required to determine the extent of whole organism adverse effects of acute imdacloprid exposure suppressing TRβ gene expression. In addition, whether the underlying mechanism causing suppression of TRβ is caused by genotoxicity as observed in frog species (Feng et al. 2004; Pérez-Iglesias et al. 2014; Ruiz de Arcaute et al. 2014) or imidacloprid interacting more specifically with one or more aspects of the HPT axis merits further study.

This is one of the first studies to examine development in the larval NWS after exposure to a toxicant, and although few overt changes were evident, expedited development was observed after exposure to imidacloprid at 100 µg/L. To further characterize development in this species based on limb development, left and right limb development were analyzed separately with digit formation as the main measure. It appears that based on examination of control larvae from both subchronic exposures that both forelimbs and hindlimbs develop in a synchronous manner. Interestingly, the significant advancement in the 100-µg/L imidacloprid treatment alone was only observed in the right forelimb and not the left, and no changes were observed in either hindlimbs. Although various studies have evaluated the direct effects of neonicotinoids on several frog species growth and development under subchronic and chronic exposure scenarios, few effects below 1000 µg/L have been reported and none have been conducted using salamanders. Previous studies conducted on Lithobates sylvaticus (wood frogs) showed 1, 10 and 100 µg/L imidacloprid in a commercial formulation increased survival and delayed metamorphosis at 10 and 100 µg/L, but no effects on body size were observed (Robinson et al. 2017). The wood frogs appear to support the results observed in the present study, except that salamander development was expedited after exposure to 100 µg/L imidacloprid rather than delayed as observed in the wood frogs by Robinson et al. (2017). Ultimately, these two studies are not directly comparable since the present salamander study was of shorter duration (i.e., 35 days not to metamorphosis vs. 42 days to metamorphosis) and entailed pure active ingredient and not a commercial formulation. More recent experiments examining the effects of commercial formulations of clothianidin and thiamethoxam on wood frogs and northern leopard frogs (Lithobates pipiens) showed that individual treatments of these pesticides at 2.5 and 250 µg/L did not have any effects on larval survival, growth, and development through to metamorphosis (Robinson et al. 2019). Studies in Blanchard’s cricket frogs (Acris blanchardi) chronically exposed to 1000 µg/L of imidacloprid had larger mass at metamorphosis compared to controls (Boone 2018) and at high concentrations, 9000 µg/L, significant mortality occurred. In the present study NWS larvae were unaffected in terms of body size and mortality after 35-d exposures up to 1000 µg/L of imidacloprid and a mixture of neonicotoinoids and also after acute exposure up to 6750 µg/L of imidacloprid. Nonetheless, additional exposure experiments through to metamorphosis with the NWS are needed to compare directly to wood, northern leopard, and Blanchard’s cricket frog studies that were of longer duration to verify no further latent effects on growth and development ensued at metamorphic climax in the NWS. Furthermore, although the lack of concentration response in thyroid hormone-mediated processes is commonly observed in toxicant exposures in vertebrates (Lagarde et al. 2015), the absence of changes in development in the left forelimb, and no changes in the neonicotinoid mixture exposure collectively do not provide clear evidence of altered development due to imidacloprid exposure in the NWS. Additional studies examining whether the observed expedited development persists through metamorphosis after 100 µg/L imidacloprid exposure, as well as studies characterizing the changes in development in response to exogenous thyroid hormone would aid in understanding the implications of asynchronous forelimb development on mobility and ultimately survival in the NWS.

The present study provides evidence of sublethal impacts on the Northwestern salamander based on advanced development and thyroid receptor gene expression changes after imidacloprid exposure and contributes to the growing body of literature suggesting some sub-lethal effects of neonicotinoids on nontarget wildlife. Furthermore, the novel toxicity data demonstrating that the current reports of environmental concentrations of imidacloprid, thiamethoxam, and clothianidin are not likely causing direct mortality in the NWS. Nonetheless, in light of the reports of sublethal effects in many taxa (i.e., fish and anurans), including the NWS in this study, future studies are necessary to accurately assess the risk of these pesticides on poorly studied caudate amphibians.