Effects of aquatic exposure to the insecticide carbaryl are species-specific across life stages and mediated by heterospecific competitors in anurans

Authors


Correspondence author. E-mail: distelca@muohio.edu

Summary

1. Understanding the direct and indirect mechanisms by which anthropogenic stressors interact with species and communities is central to understanding ecology in a modern world. Pesticides have been implicated in amphibian declines worldwide, but no mechanism has yet been identified.

2. We tested the hypotheses that exposure to the insecticide carbaryl (2 mg/L) would mediate competitive outcomes between larvae of two anuran species (Bufo americanus, American toads and Rana pipiens, northern leopard frogs), and that post-exposure effects would carry over to affect juveniles in the terrestrial environment.

3. These hypotheses were tested in controlled mesocosm experiments. Toad survival decreased with insecticide exposure and presence of heterospecific competitors, relative to low-density controls. However, when exposed to the insecticide, toads experienced the greatest survival when leopard frogs were present. Larval period of toads was lengthened by carbaryl exposure and high density of conspecifics. Leopard frog survival was reduced only by high conspecific density, and larval period was shortened by carbaryl exposure. There were no significant effects of the insecticide on periphyton abundance, although phytoplankton abundance increased and zooplankton abundance decreased after exposure.

4. A subset of metamorphs of each species from each aquatic treatment was added to 2 × 2 m terrestrial enclosures to evaluate larval treatment effects on overwinter survival and growth. Despite differences in initial mass of this subset of metamorphs, mass at spring emergence was not significantly affected by any aquatic treatment for either species. However, leopard frog survival to and mass at spring emergence benefited from larger mass at metamorphosis, while toads appeared unaffected.

5. These results suggest that while tadpoles sharing a habitat may experience reduced survival from competition, the presence of multiple amphibian species may ameliorate the effects of pesticide exposure.

6. More generally, maintaining diversity may benefit ecological communities by reducing the impacts of anthropogenic threats to certain taxa.

Introduction

The composition of ecological communities is governed by numerous interacting factors, such as competition, predation, temperature and precipitation. Closely related, co-occurring species are often affected by environmental factors in similar ways due to behavioural and physiological constraints, increasing the likelihood for competitive interactions (Adams & Thibault 2006). However, differential susceptibility to biotic and abiotic factors may lead to competitive asymmetry (McHugh & Budy 2005; Alcaraz, Bisazza & García-Berthou 2008), which can facilitate anything from coexistence (Thomas & Holoway 2005; Krassoi et al. 2008) to competitive exclusion of one species (Taniguchi & Nakano 2000). Therefore, it is critical to understand the interactive effects of biotic and abiotic stressors, which could reduce the abundance of one or more species, potentially increasing their vulnerability to extinction from stochastic events (Lande 1993).

Understanding the impacts of interactions among environmental variables on communities is becoming increasingly important due to anthropogenic production of and manipulation of abiotic environmental stressors. Contaminant release into the environment has the potential to interact with biotic factors to alter community dynamics (Rowe, Hopkins & Coffman 2001). Yet, due to the complex interactions among co-occurring species, the effects of abiotic factors may not be predictable based on single-factor or single-life-stage tests alone. For instance, Warner, Travis & Dunson (1993) found that interspecific competition between tadpoles was mediated by pH and Lefcort et al. (1999) found that predation and heavy metal exposure interacted to change competitive outcomes between tadpoles and snails. Because abiotic factors can mediate or exacerbate the strength of biotic interactions, community structure can be altered in ways that are not predictable from studying the effects on just one species. The last decade has seen increased research on the interactions of contaminants and biotic factors (i.e. competition, predation, parasitism) in aquatic environments (Hanazato 2001; Rohr, Kerby & Sih 2006; Rohr et al. 2008). Contaminants that are directly lethal to some organisms (i.e. insecticides on aquatic arthropods) may cause indirect effects on other species through food web-mediated interactions (Boone & Semlitsch 2003; Relyea 2005).

Pesticides in particular have widespread use and can contaminate aquatic systems outside target agricultural areas via run-off or aerial deposition through both space (Sparling, Fellers & McConnell 2001; Bridges & Little 2005; Freake & Lindquist 2008) and time (Harris et al. 1998; Freake & Lindquist 2008). While many modern pesticides have short half-lives and low bioconcentration factors, their impacts on some taxa within the food web can still be lethal, potentially causing trophic cascades. Even relatively low concentrations of insecticides can eliminate aquatic arthropods (Hanazato & Yasuno 1989; Boone & Semlitsch 2003; Relyea 2005), which can fundamentally alter aquatic communities well after degradation of the chemical (Hanazato & Yasuno 1987; Downing et al. 2008).

Communities of pond-breeding amphibians provide an excellent system for studying the complex interactions among biotic and abiotic factors because species with similar resource needs, behaviour and morphology (which limits which resources are used) are commonly found together and compete for food. Experiments with competing species with different length in larval period can provide additional insight into the relative importance of interspecific and intraspecific competition. Species with short larval periods may experience greater effects of interspecific density than competing species with longer larval periods (Kupferberg 1997; Lawler et al. 1999; Boone et al. 2004). When co-occurring, the species with the shorter larval period is constantly subjected to interspecific effects, while the species with the longer larval period is only subject to interspecific effects until the other species reaches metamorphosis. Because insecticides can alter aquatic food webs and lead to changes in tadpole food abundance, differences in larval period among species could exacerbate competitive asymmetry. Short-term changes in food abundance could have a relatively greater effect on the species with the shorter larval period because food is altered for a greater proportion of their larval period. Conversely, long-term changes in food could have a greater effect on the species with the long larval period because the other species may achieve metamorphosis before changes in food availability manifest. The carbamate insecticide carbaryl can cause an increase in periphyton abundance shortly after exposure (Boone et al. 2007; Distel & Boone 2009) followed by a long-term decrease (Mills & Semlitsch 2004; Distel & Boone 2009). Because most species of tadpoles feed primarily on periphyton resources (Altig, Whiles & Taylor 2007), short-term increases could benefit species having short larval periods (which could result in larger size and/or earlier time to metamorphosis) while long-term decreases in this resource could negatively impact species having longer larval periods (which could result in smaller size and/or later time to metamorphosis). Because effects on time and size at metamorphosis can have consequences for overwinter survival (Semlitsch 1987; Berven 1990; Scott 1994; Altwegg 2003), changes at metamorphosis can be significant, although effects may be at least partially negated if differences in mass at metamorphosis are overcome in the terrestrial environment (as in Boone 2005; Distel & Boone 2009).

Understanding the mechanisms affecting amphibian fitness and population dynamics is critical because of extinctions and declines worldwide (Collins & Storfer 2003; Stuart et al. 2004) that appear to be occurring orders of magnitude above background extinction rates (McCallum 2007). Correlative evidence strongly suggests an association between pesticide exposure and amphibian declines (Davidson 2004), although no documented mechanism has linked pesticides to declines. Virtually all prior studies of effects of pesticides on amphibians have focused on a single life stage (usually larvae). Few of those studies have followed individuals through multiple life stages, and only a few have examined the effects of larval pesticide exposure on post-metamorphic anurans (e.g. Boone 2005; Distel & Boone 2009), despite evidence that post-metamorphic life stages may be more important in population regulation than embryonic or larval life stages (Biek et al. 2002; Vonesh & De la Cruz 2002). Rohr & Palmer (2005) showed a negative effect of pesticide exposure on juveniles, where larval exposure to atrazine increased the risk of desiccation in post-metamorphic streamside salamanders (Ambystoma barbouri). No study to date has documented the post-exposure effects of pesticide exposure with both intra- and interspecific larval competition.

We tested the hypotheses that interactive effects of carbaryl and competitors on metamorphosis would differ from either factor alone using tadpoles of Bufo americanus (American toads, Fig. 1) and Rana pipiens (northern leopard frogs, Fig. 1) in cattle tank mesocosms. These two species use similar breeding sites, breed at similar times in the spring, and tadpoles can be found together (Werner & Glennemeier 1999; C.A. Distel & M.D. Boone, personal observation). However, these two species differ markedly in the length of their larval periods; toads may spend 50–60 days as tadpoles (Green 2005) while leopard frog larval periods may be 3–6 months or more (Rorabaugh 2005). Additionally, Purrenhage & Boone (2009) showed that slight increases in competitor density (adding six competitors at a time), even at low overall density, could have strong density-dependent effects on these species. We predicted that carbaryl exposure alone would positively affect toad tadpole survival, mass at metamorphosis and larval period because of early increases in periphyton (indirect effect) coincident with the short larval period of toads. Because of a predicted long-term decrease in periphyton abundance based on Distel & Boone (2009), we predicted neutral or negative impacts on leopard frog survival, mass and larval period. We predicted that increased competitor density would have negative impacts on all metamorphic endpoints for both species. When carbaryl exposure was combined with interspecific competitors, we predicted that the net impact for survival, mass and larval period would benefit toads more than leopard frogs (compared with intraspecific competition) because of differences in length of the larval period.

Figure 1.

Bufo americanus (left; American toad) metamorph with Rana pipiens (right; northern leopard frog) tadpole. Photograph credit: CAD.

We also followed a subset of these metamorphs in the terrestrial environment through their first winter to spring emergence, and predicted that there would be no post-exposure effects of aquatic insecticide exposure, competitor density, or their interaction on overwinter growth and survival of either species, despite potential differences in mass at and time to metamorphosis (sensuBoone 2005; Distel & Boone 2009). Ours is the first experiment to examine the hypothesis that alteration in competitive interactions from insecticide exposure could lead to competitive exclusion as a result of impacts on larval development and juvenile survival.

Materials and methods

Animal collection and husbandry

American toads were collected on 9 April 2008 as 20 partial egg strings (stages 1–7; Gosner 1960) from a flooded no-till agricultural field on Miami University property north of Oxford, Butler Co., Ohio. Northern leopard frog eggs were collected on 12 April 2008 as four partially developed (stages 18–20; Gosner 1960) egg masses from Rush Run Wildlife Area, Somerville, Preble Co., Ohio. All eggs were transported to Miami University and maintained in 5 L plastic tubs with daily water changes and fresh fish flakes until the time of the experiment (6 days for Bufo, 3 days for Rana).

Model pesticide

Carbaryl (1-napthyl N-methylcarbamate) is a widely used carbamate insecticide with the same general mode of action as other cholinesterase inhibitors (carbamates and organophosphates, Cox 2005). These insecticide families have been linked to amphibian declines (Davidson 2004), so exposure to carbaryl and its effects may represent a wide range of influential insecticides. Carbaryl is also an optimum insecticide to use as a model for indirect effects because of its short half-life (hours to days in water), its low bioconcentration factor and the ability of most non-target organisms to eliminate it from the active site in the neuromuscular junction (Cox 2005). Therefore, the direct effects of carbaryl are highly transient, and any post-exposure effects are due to changes in community or ecosystem characteristics.

Experimental communities

Aquatic communities were established in 1300 L cattle tank mesocosms (hereafter ‘ponds’). We filled experimental ponds on 1 to 2 April 2008 with 1000 L of city water and allowed them to age before addition of tadpoles. Each pond was covered with a screen mesh lid to prevent invasion by insects or colonizing anurans. On 3 April 2008 we added 1 kg of mixed deciduous leaf litter raked from a nearby forest and 1 L of plankton inoculum from nearby fishless ponds to each experimental pond. Ponds were maintained in an open field at the Miami University Ecology Research Center, Butler Co., Ohio. Overwinter monitoring took place in 2 m × 2 m terrestrial pens in the same field as ponds at Miami University’s Ecology Research Center. Pens were constructed of stainless steel sheet metal with walls extending 0·9 m above-ground and 0·6 m below-ground to prevent burrowing out by the anurans or burrowing in by potential predators. Each pen had a c. 0·5 m deep × 0·4 m diameter pit in the centre filled with mixed deciduous leaf litter as an anuran hibernaculum. Each pit was covered with an c. 0·4 × 0·6 m piece of plywood. Pens were arranged in blocks of four, in two rows of six blocks.

Experimental design

We established five anuran community treatments: 50 R. pipiens alone, 50 B. americanus alone, 50 R. pipiens and 50 B. americanus together, 100 R. pipiens alone (density control) or 100 B. americanus alone (density control). These tadpole densities are well within the range observed in wild anuran populations (Morin 1983; Werner & Glennemeier 1999), and similar densities have been shown to induce competition in American toads (Distel & Boone 2009). Because changes in tadpole food abundance may be due to top-down food web effects driven by direct effects of pesticides on zooplankton and subsequent competition between periphyton and phytoplankton (Mills & Semlitsch 2004; Boone et al. 2007; Relyea & Diecks 2008), we included a treatment with no tadpoles to examine the effects of carbaryl on the aquatic food webs for a total of six community types. All tadpoles were added on 15 April 2008. Each community was assigned to either a 0 or 2 mg/L carbaryl treatment. Each treatment was replicated four times for a total of 48 ponds. Mean (SE) water pH, dissolved oxygen concentration (D.O.), and temperature immediately prior to carbaryl addition were 7·53 (0·06), 5·5 (0·7) mg/L and 21·4 (0·1) °C respectively.

Carbaryl was added to half of the ponds on 22 April 2008 (experimental day 1) as 8·88 g of liquid Sevin® (22·5% carbaryl; GardenTech, Lexington, KY, USA) dissolved in 5 L of pond water and added evenly over the surface of each treatment pond with a watering can, yielding a nominal concentration of 2·0 mg/L. This concentration is within the ecologically relevant spectrum in surface water (Norris, Lorz & Gregory 1983; Peterson et al. 1994) and does not appear to cause direct effects on survival in northern leopard frogs (Relyea 2004) or southern leopard frogs (R. sphenocephala, Mills & Semlitsch 2004). Insecticide-free pond water was applied to the surfaces of control ponds to control for disturbance. Carbaryl concentration was analysed by high-performance liquid chromatography (Mississippi State Chemical Laboratory) at 1, 24, 48, 72 and 96 h after dosing. Concentrations were 1·2, 0·57, 0·48, 0·31 and 0·22 mg/L, respectively, yielding a half-life of c. 3·4 h.

Periphyton was sampled by applying a 1·3 cm-wide razor blade to the north-west side of each pond 3 cm below the surface and pulling the razor to the surface so that periphyton was not lost into the water column. Each scraping was collected onto a glass fibre filter and stored in buffered acetone. Samples were taken on seven sampling dates and chlorophyll a content was analysed with a 10-AU fluorometer (Turner Designs, Sunnyvale, CA, USA) to estimate periphyton abundance. Phytoplankton samples were taken by collecting three 1 L samples of pond water with a column sampler. Then a 100 mL composite subsample was collected and filtered onto a glass fibre filter by vacuum filtration, and chlorophyll a concentration was analysed with fluorometry as above. Zooplankton abundance was estimated by taking three water column samples as above, passing a 1 L composite over 80 μm mesh, and storing the collected zooplankton in 80% ethanol on three sampling dates. Copepods and cladocerans were counted from each sample under a dissecting microscope. Water in each pond was measured for temperature and D.O. on four sampling dates and for pH on three sampling dates throughout the experiment.

Metamorphosis was defined as the emergence of at least one forelimb (stage 42; Gosner 1960), which was determined by visually observing animals in the ponds. Metamorphs were collected with a small dip net and taken to the laboratory where they were kept until tail resorption was complete (i.e. time to metamorphosis), as in Distel & Boone (2009), at which time they were blotted dry with a paper towel and weighed to the nearest 0·001 g. Mean (SE) time between capture and tail resorption was 3·46 (0·12) days for toads and 4·60 (0·05) days for leopard frogs. The experiment was terminated on 26 September 2008, at which time no toads remained in the ponds and remaining leopard frog tadpoles were collected to determine survival, mass and developmental stage. We assumed that any tadpoles remaining in the experiment would not contribute to the pool of recruits in the natural population because their natal pond was already dry.

A subset of 10 metamorphs from each aquatic treatment was added to the terrestrial pens. The first 40 metamorphs of each species from each treatment received an individual toe clip identification code after weighing. Completion of tail resorption, weighing, toe clipping and addition to the appropriate pen took <1 day for each animal. To control for pond-specific effects on overwinter performance, metamorphs were added to each replicate terrestrial pen in the order that they reached metamorphosis. Therefore, all animals in a pen were from the same aquatic treatment, but not all animals in a pen were reared in the same pond. No pens contained both species, so animals raised in mixed-species ponds were separated into species-specific terrestrial pens. Each aquatic treatment was represented as four replicates in the terrestrial treatments yielding a total of 48 terrestrial pens. Toads were added to terrestrial pens from 4 to 8 June 2008. Leopard frogs were added from 19 June to 2 July 2008.

Pens were left undisturbed for the duration of the winter until metamorphs were observed moving within them (11 March 2009). After visually searching for and collecting juvenile anurans on the ground surface, we removed all leaf litter from the pits and collected any animals that were still burrowed in each pen. Not all animals were captured during the first collection; some had used alternate burrows such as natural cracks in the soil, and remained undetected. Therefore, collections were made twice weekly for the first 2 weeks, weekly for 2 weeks and then every 2 weeks until no more anurans were captured from 11 March to 17 June 2009. All animals collected were weighed to the nearest 0·001 g, then released at their natal site. Any anurans kept in the laboratory for more than 24 h were fed crickets ad libitum and received fresh water and substrate daily.

Statistical analyses

We used pond means as the experimental unit to determine effects of treatments and their interactions on survival to metamorphosis, mass at metamorphosis and larval period, as well as water quality measurements. Survival to metamorphosis was arcsine-square-root-transformed and analysed for each species with separate analyses of variance (anova). Mass at metamorphosis and larval period were log-transformed and analysed for each species with separate multivariate analyses of covariance (mancova) with intraspecific survival covariates. As not all Rana tadpoles reached metamorphosis, their survival covariate for the mancova was the per cent total survival to the end of the experiment. Periphyton and phytoplankton abundance were log-transformed. We used a repeated-measures anova to evaluate the effects of carbaryl, community type and their interaction among all treatments with tadpoles on periphyton and phytoplankton. Additionally, we evaluated the effects of carbaryl, tadpole presence/absence, and their interaction on periphyton and phytoplankton to distinguish if changes in algal resources were predominantly a result of changes in zooplankton (without tadpoles) or changes in anuran behaviour with exposure. Zooplankton abundance, pH, D.O. and temperature were similarly analysed with separate repeated-measures anovas (also among tadpole community types and with tadpole presence/absence as above).

We conducted a separate manova on mass at metamorphosis and larval period of the metamorphs of each species used in the terrestrial experiment (using pen means as experimental units) to determine whether any treatment-specific differences were present at the initiation of the terrestrial experiment. Overwinter survival for each pen was arcsine-square-root transformed and analysed with an anova. Size at spring emergence and overwinter growth (defined as difference in mass between metamorphosis and spring emergence) were log-transformed and analysed with ancovas using overwinter survival and date of spring capture as covariates. Each pen was considered an experimental unit, so pen means for mass and growth were used in all analyses. We also used mixed models to determine whether mass at metamorphosis was significantly related to overwinter survival or mass at spring emergence using individual metamorphs as experimental units for both species. Pen was used as a random factor to account for any effects that were not independent on individuals from the same pen. We used SAS 9.1.3 software (SAS Institute, Cary, NC, USA) and α = 0·05 for all analyses.

Results

American toads

Toad survival to metamorphosis was affected by an interaction between carbaryl exposure and community type (Fig. 2a, Table S1, Supporting Information). Toads reared at low density without carbaryl exposure showed the greatest survival (Fig. 2a). In the presence of both carbaryl and interspecific competition, toad survival was 22% higher than with either factor alone (Fig. 2a, Table S1). The multivariate response of mass and time to metamorphosis showed that toad tadpoles were significantly affected by community type (Fig. 3a), but that neither carbaryl exposure nor the interaction was significant (Table S1). This multivariate response was mainly driven by larval period. Toads reared at low density with only conspecifics reached metamorphosis c. 1 day earlier than tadpoles in the other two community types (Fig. 3a, Table S1). Although the multivariate effect of carbaryl was nonsignificant, the univariate analysis indicated that carbaryl exposure lengthened toad larval period by approximately half a day. There were no univariate effects on toad mass at metamorphosis.

Figure 2.

 Analysis of variance results for proportions of (a) American toads (Bufo americanus) and (b) northern leopard frogs (NLFs, Rana pipiens) surviving to metamorphosis after exposure to 0 or 2 mg/L carbaryl and varying tadpole communities. Points sharing the same letter are not significantly different. Error bars are ±1 SE.

Figure 3.

 Analysis of covariance results for date at metamorphosis of (a) American toads and (b) northern leopard frogs (NLFs) after exposure to 0 or 2 mg/L carbaryl and varying tadpole communities. Error bars are ±1 SE.

The subset of metamorphs used in the terrestrial experiment experienced significant effects of aquatic treatments on metamorphosis that by chance differed from the entire cohort of metamorphs described above. There were significant carbaryl, community and interaction effects driven primarily by effects on date at metamorphosis (Fig. 4a), but not on mass at metamorphosis (Fig 5a, Table S2). Neither aquatic treatment affected toad overwinter survival, growth or mass at spring emergence (Table S2). There were also no effects of mass at metamorphosis on overwinter survival (F1,215 = 1·16, = 0·2836) or mass at spring emergence (F1,46 = 0·81, = 0·3731).

Figure 4.

 Analysis of covariance results for effects of aquatic carbaryl and community treatments on date at metamorphosis of the subset of (a) American toad and (b) northern leopard frog (NLFs) metamorphs added to terrestrial pens. Error bars are ±1 SE.

Figure 5.

 Analysis of covariance results for mass of (a) American toads and (b) northern leopard frogs (NLFs) at metamorphosis and at spring emergence. Error bars are ±1 SE.

Northern leopard frogs

Leopard frog survival to metamorphosis was affected by community type, with a >50% reduction when tadpoles were reared at high intraspecific density (Fig. 2b, Table S1). The mancova showed a significant effect of carbaryl on time and mass at metamorphosis, but community type or the interaction did not affect metamorphosis (Table S1). The leopard frog metamorphic response was largely driven by effects on the larval period. Univariate analyses revealed an c. 6-day reduction in larval period with carbaryl exposure (Fig. 3b).

The subset of metamorphs used in the terrestrial experiment experienced significant effects of carbaryl, community and their interaction on metamorphosis, driven largely by mass at metamorphosis (Fig. 5b, where metamorphs from low-density, carbaryl-exposed treatments were c. 24% larger than all other metamorphs), but not date at metamorphosis (Fig. 4b, Table S2). There was a significant effect of carbaryl and a carbaryl × community interaction on the multivariate response of this subset of metamorphs, but there were no significant univariate effects of any aquatic treatment or interaction on overwinter survival, growth or mass at spring emergence (Table S2). There were significant effects of mass at metamorphosis on overwinter survival (F1,214 = 22·31, < 0·0001) and mass at spring emergence (F1,66 = 8·17, = 0·0057; Fig. 6). Metamorphs that survived to spring emergence were c. 15% larger at metamorphosis than those that died [mean (SE) mass at metamorphosis of survivors and non-survivors was 1·111 (0·028) g and 0·961 (0·022) g respectively]. However, any differences in mass disappeared by spring emergence (Fig. 5b).

Figure 6.

 Mixed model results for effect of mass at metamorphosis on mass at spring emergence in northern leopard frogs.

Aquatic community responses

Periphyton abundance was not affected by any treatment or combination (Table S3). Phytoplankton abundance responded to carbaryl exposure in ponds with tadpoles (Table S3). Ponds exposed to carbaryl showed a short-term decrease in phytoplankton abundance immediately after exposure, followed by a long-term increase in abundance over controls (Fig. 7a). Phytoplankton abundance was also significantly affected by tadpole presence and an interaction between carbaryl and tadpole presence (Table S3), where ponds without tadpoles tended to have more phytoplankton than ponds with tadpoles, but carbaryl exposure exacerbated the difference (Fig. 7a).

Figure 7.

 Repeated-measures analysis of variance results for response of (a) phytoplankton abundance and (b) cladoceran zooplankton abundance to 0 or 2 mg/L carbaryl and tadpole presence (phytoplankton only). Error bars are ±1 SE.

Zooplankton abundance differed by major taxonomic group. There were no effects of any treatment, interaction or tadpole presence on abundance of copepod adults or nauplii. Cladoceran zooplankton were most abundant and responded strongly and negatively to carbaryl exposure, but not to community type or their interaction (Fig. 7b, Table S3). Carbaryl significantly reduced cladoceran abundance shortly after exposure, and this reduction was maintained until the last sampling date (49 days after initial exposure).

Water chemistry

Effects on water chemistry were minor. While there were significant time × community and time × carbaryl × community effects on temperature, the greatest differences in temperature between treatments were <0·9 °C on one sampling day (Fig. 8a,b), which we do not believe were biologically significant. There were no effects of tadpole presence on temperature (Table S4).

Figure 8.

 Repeated-measures analysis of variance results for responses of (a, b) temperature, (c, d) dissolved oxygen and (e, f) pH to 0 (a, c, e) or 2 (b, d, f) mg/L carbaryl and five tadpole community arrangements (see legend) with American toads and/or northern leopard frogs (NLFs), or no tadpoles.

In ponds with tadpoles there were no significant effects of treatments or interactions on D.O. (Fig. 8c,d,Table S4). Control ponds without tadpoles tended to have greater D.O. than control ponds with tadpoles (Fig. 8c), but this pattern was reversed in carbaryl-exposed ponds (Fig. 8d, Table S4). However, in ponds with tadpoles the greatest difference in mean D.O. was 0·8 mg/L, which we do not believe would result in a biologically significant effect on anurans.

There was a significant time × carbaryl effect on pH in ponds with tadpoles (Table S4), but mean differences were never greater than 0·15 pH units. There were no effects of community type or interaction effects in ponds with tadpoles. When comparing ponds without tadpoles to ponds with tadpoles there were significant effects of both carbaryl and the carbaryl × tadpole presence interaction (Table S4). Overall, pH was higher in control ponds than in carbaryl-exposed ponds on sampling dates after exposure (Fig. 8e,f), but mean differences were never greater than 0·36 pH units. After exposure to carbaryl, control ponds containing tadpoles had significantly lower pH than control ponds without tadpoles (Fig. 8e), but carbaryl-exposed ponds showed the opposite effect (Fig. 8f). While there were differences as great as 0·9 pH units between some of these treatments, the treatments containing tadpoles were still no greater than 0·15 pH units. Therefore, we believe that there was no biologically significant effect of differences in pH on anurans.

Discussion

This is the first study to test the relative impacts of conspecific and heterospecific presence under pesticide stress on amphibian performance across multiple life stages. We show that despite biotic and abiotic differences in the aquatic environment, juvenile anurans reach similar sizes by spring emergence even if differences in mass existed at metamorphosis, similar to previous studies (Beck & Congdon 1999; Boone 2005; Distel & Boone 2009). Juvenile survival may be one of the most important determinants of population success (Biek et al. 2002; Vonesh & De la Cruz 2002). Interestingly, our results suggest that studies designed to determine the effects of the aquatic environment on anuran survival may focus largely on measuring survival to and mass at metamorphosis alone, because the number of juveniles surviving to breed may be most heavily affected by these end points. However, our study did not address every life stage, and impacts of the embryonic environment (Bridges 2000) and post-exposure effects of contaminant exposure at any life stage (Budischak, Belden & Hopkins 2008) on reproductive adults must still be considered.

Contrary to our initial predictions and previous studies (Boone & Semlitsch 2002; Boone et al. 2004), our results indicate that survival to metamorphosis of the species with the shorter larval period suffered more severely from insecticide exposure than that of the species with the long larval period (consistent with Mills & Semlitsch 2004). The carbaryl concentration used in the present study was not expected to have direct effects on tadpoles (Mills & Semlitsch 2004; Relyea 2004), but toads appeared to have suffered lethality in carbaryl-exposed ponds. As there were no reductions in periphyton abundance to explain reductions in survival due to starvation, it appears that 2·0 mg/L carbaryl may be directly lethal to a proportion of a toad tadpole cohort in outdoor environments. Surprisingly (and contrary to our predictions), the reduction in toad survival did not lead to greater mean mass or shorter mean larval period of either species, or to greater periphyton abundance in carbaryl-exposed ponds. These results suggest that periphyton may not be the main resource for which these species compete, and therefore does not limit tadpole metamorphosis (Altig, Whiles & Taylor 2007). Tadpoles may switch to alternate food sources due to changes in behaviour or periphyton composition in exposed ponds, although we did not measure behaviour or periphyton composition. Alternatively, periphyton may remain limiting, but direct effects of exposure may outweigh any indirect benefits.

There appeared to be condition-specific effects on tadpole survival. Carbaryl interacted with community type to alter the effects of competitors on toads. Therefore, toads may be at greater risk than leopard frogs in contaminated systems, but the presence of leopard frogs reduced the impact of carbaryl on the toads. Unexposed toads showed reduced survival with increased heterospecific tadpole density. This outcome may be due to greater corporeal absorption of carbaryl by the leopard frog tadpoles, reducing the concentration to which toad tadpoles were exposed. However, as we did not measure body burdens of carbaryl in the tadpoles, this possibility remains unconfirmed. Interestingly, all density-dependent effects on toad survival disappeared in treatments exposed to carbaryl. One possible explanation for this phenomenon is insecticide-induced niche partitioning through species-specific differences in foraging behaviour. While we did not collect behavioural data, niche partitioning could theoretically happen in more than one way in these systems. If one species capitalized on the abundant phytoplankton more than the other species in exposed systems, competition for periphyton would be reduced. Alternatively, insecticide exposure could have directly affected the behaviour of leopard frogs (i.e. impaired movement through cholinesterase inhibition; Bridges 1997), temporarily reducing competitive pressure on toads. Contrary to our predictions, carbaryl exposure did not exacerbate the effects of competition on either tadpole species. Whether this indicates an amelioration of competitive effects by the insecticide or merely indicates that a certain proportion of a tadpole cohort is likely to be prone to increased stress remains unclear (i.e. both the insecticide and competitors eliminated the same proportion of tadpoles from each toad cohort, and both stressors in combination did not affect hardier tadpoles). The leopard frogs responded differently than toads, and showed effects of only conspecific density, as we predicted. Therefore, the relative larval periods of each species may influence the relative importance of inter- vs. intraspecific competition.

Tadpoles of both species from high-density treatments were predicted to be smaller than in low-density treatments (Wilbur & Collins 1973). While there were no significant effects on mass of either species, the significant effects on survival indicate that those tadpoles surviving in the greater density treatments or carbaryl-exposed treatments were able to reach a comparable mass to metamorphs in the low-density control treatments. These results on mass may also be tied in to what resources tadpoles actually consume (which may not be limited to periphyton; Altig, Whiles & Taylor 2007), or periphyton may not have been limiting for the tadpoles in our experiment.

Aquatic communities that did not contain tadpoles did not have significantly different trajectories for algal abundance or water chemistry measurements than communities with tadpoles. Relyea (2009) suggested that carbaryl-mediated differences in periphyton abundance may be due to reduced foraging activity by tadpoles. Therefore, in ponds without tadpoles, no effects of carbaryl should be observed on periphyton. We did observe a short-term decrease in phytoplankton abundance in exposed ponds. This change is not consistent with the hypothetical prediction that insecticide-mediated elimination of zooplankton drives a trophic cascade through which phytoplankton abundance immediately increases. Additionally, the empirically based prediction of decreased periphyton abundance later in the experiment (Boone et al. 2007; Relyea & Diecks 2008; Distel & Boone 2009) was not observed. Because we did not observe differences in the relative abundance of certain community members (i.e. phytoplankton, zooplankton) between ponds with and without tadpole grazers, our results suggest that changes in grazer behaviour were not drivers of observed insecticide-induced changes in algal dynamics.

Leopard frog overwinter survival and mass at spring emergence were greater in juveniles that reached metamorphosis at a greater mass, consistent with Werner (1986). These results may have profound effects on leopard frog populations if factors in the aquatic environment affect timing and size at metamorphosis (such as competition and pesticides; Boone et al. 2004; Relyea & Diecks 2008; Distel & Boone 2009). Berven (1990) found that wood frogs (R. sylvatica) exhibited both greater juvenile survival and greater adult body size with greater size at metamorphosis (≤4 mm body length). Even apparently slight size advantages at metamorphosis can translate into positive effects on fitness by increasing size at first breeding, and reducing the time to sexual maturity (relative to animals that were smaller at metamorphosis; Smith 1987; Berven 1990). Our results supported the prediction that there would be no post-exposure aquatic treatment effects on the endpoints that we measured in terrestrial juveniles. Morey & Reznick (2001) reared tadpoles at low and high density, and found that the ability of metamorphs to overcome differences in mass at metamorphosis within the first year of life was dependent on both gender and on terrestrial food availability. As we did not determine the gender of our metamorphs, it is possible that the variability associated with mass at spring emergence could be a factor of each pen containing individuals of each gender. Given the close proximity of our terrestrial pens to one another and their equal size, we assumed food availability to be approximately equal between pens. Finally, our sample sizes were considerably lower than field studies of natural populations used to evaluate the effects of mass at metamorphosis (where effects were significant), potentially reducing our ability to detect differences.

In conclusion, species-specific differences in sensitivity to contaminants can affect the relative roles of intra- and interspecific competition for one species while another species remains unaffected. These asymmetric effects indicate that insecticide–competitor interactions on tadpoles may not be predictable based on studies of one factor alone. However, studies evaluating the effects on metamorphic endpoints alone may be sufficiently descriptive to provide insight into population-level processes. Species with short larval periods may experience reduced survival from insecticide exposure, increased interspecific competitor density or both, while species with longer larval periods may be unaffected for the same fitness criteria. Additionally, if species with longer larval periods receive survival benefits from metamorphosing at larger masses, the pesticide-mediated changes in the food web could benefit these species. These changes in survival, along with pesticide-mediated changes in abundance of food resources, could potentially lead to exclusion of the species with the short larval period that is negatively impacted. However, our study suggests that diverse communities buffer species from extinctions and suggest that loss of some species may make others more vulnerable to elimination from the system.

Acknowledgements

We thank Nick Webber, Jennifer Purrenhage, Holly Puglis, Tamara McPeek, Ben Bulen, Sarah Distel, Mark Mackey and Rodney Kolb for their valuable help on the execution of this project. We thank the Miami University Ecology Research Center for use of their facilities. This project was funded by the U.S. Geological Survey (Amphibian Specialist Group Seed Grant to C.A.D., administered by Conservation International and Partners in Amphibian and Reptile Conservation), National Science Foundation (DEB #0717088 to M.D.B.) and Miami University. All animal handling and care was in accordance with Miami University IACUC protocol #740.

Ancillary