Stress hormones and the fitness consequences associated with the transition to a novel diet in larval amphibians

Evolutionary diversification in animals is often accompanied by a change in resource use (Simpson, 1953; Schluter, 2000; Grant and Grant, 2008). For example, closely related, sympatric species typically differ in the prey on which they specialize, presumably as an evolutionary response that minimizes interspecific competition (reviewed by Schluter, 2000). Yet, such species must have had similar dietary preferences before they diverged,suggesting that shifts in resource use often accompany diversification. Despite the importance of dietary shifts, little is known about the physiological changes and fitness consequences that accompany them.

One way to assess the consequences of dietary transitions is to compare diet-induced responses in species or populations that exhibit an ancestral feeding strategy relative to those that have undergone a transition to a novel diet. Some studies that have taken this approach have found that populations may initially suffer a fitness cost as they make this transition. For example,Phillips and Shine (Phillips and Shine,2006) found that when black snakes from Australia first encountered a novel prey item (introduced cane toads) the snakes probably suffered reduced fitness (cane toads are toxic). Black snakes were only able to prey consistently on cane toads once they had evolved behavioral and physiological traits that enabled them to cope with such prey. Furthermore,when organisms are presented with novel dietary resources, they may lack the behavioral, physiological or morphological characteristics for consuming that diet; thus a food-restricted condition either as a result of the inability to capture prey items or extract nutrients from prey items once ingested may be costly (e.g. Carroll et al.,1998). Whether such fitness costs are generally associated with dietary transitions is unclear.

We investigated the physiological changes associated with the transition to a novel diet in spadefoot toad larvae (Pelobatoidea). Most species within the Pelobatoidea clade feed on detritus, which is likely the ancestral diet of anuran larvae (Altig et al.,2007). According to character state reconstruction analysis, Spea is the only genus within the Pelobatoidea that has evolved the ability to exploit macroscopic prey(Ledón-Rettig et al.,2008). The evolution of carnivory in these tadpoles is associated with reduced intra- and inter-specific competition for food(Pfennig, 1992; Pfennig et al., 2007; Martin and Pfennig, 2009). Although this ecological scenario suggests potential fitness benefits to be gained from a carnivorous diet, especially when resources are limited,multiple characteristics are needed to sense, obtain, digest and assimilate nutrients from such prey. In order to better understand how ancestral populations of Spea may have shifted to a carnivorous diet, we examined the physiological changes that occur when larvae of Scaphiopus couchii Cope, a detritivore species in Spea's sister genus(García-Paris et al.,2003), consume anostracan shrimp, a common prey item of modern Spea larvae (Pfennig,1992; Pomeroy,1981).

We hypothesize that the transition to a carnivorous diet in spadefoot tadpoles may have been associated with the inability of tadpoles to ingest or acquire nutrients from such a diet, thus causing nutrient stress that can have negative fitness consequences. Food restriction generates substantial morphological and developmental plasticity among anuran(Wilbur and Collins, 1973; Werner and Anholt, 1996; Kupferberg, 1997) and urodele(Wildy et al., 2001; Rohr et al., 2004) larvae. Generally, studies have shown that food restriction causes reduced body sizes at metamorphosis, but the effects on development time vary depending on when the restriction is experienced. During early developmental stages food restriction extends larval periods, as a minimum body size needs to be reached before metamorphosis can occur (Wilbur and Collins, 1973). By contrast, food restriction experienced during later developmental stages accelerates metamorphic timing [in spadefoots(Newman, 1987; Morey and Reznick, 2004)]. Spadefoot toads typically develop in short-lived ephemeral ponds that not only vary in initial resource abundance, but also become depleted in resources as they dry (Pfennig, 1990; Pfennig et al., 1991). Given that growth and developmental rate are both under strong, positive,directional selection in spadefoots, reduced growth and development caused by nutrient restriction probably represents a fitness cost.

Physiologically, this plasticity in developmental timing and size at metamorphosis in response to food restriction (or other environmental stressors) is predominantly controlled by the neuroendocrine stress axis(reviewed by Denver, 2009). Recent studies in anuran tadpoles have shown that the stress axis is activated in response to food restriction (Crespi and Denver, 2005) or increased competition for food(Glennemeier and Denver,2002a), resulting in increased corticosterone (CORT) content (the primary glucocorticoid in tadpoles), increased thyroid hormone secretion and accelerated development rates (Denver,2009). Elevated CORT during food restriction has been observed in other vertebrates, and acts to inhibit growth while increasing gluconeogenesis and mobilizing energy stores from protein or lipid stores, to yield the necessary energetic resources needed to forage or seek more favorable conditions in the absence of nutrient inputs(Denver et al., 2002; McEwen and Wingfield,2003).

In terms of diet transitions, the stress response to a novel diet could invoke morphological and physiological plasticity [i.e. cryptic phenotypic variation (Queitsch et al.,2002)] that facilitates the digestion of the novel food item. In spadefoot toads, for example, intestines of Sc. couchii tadpoles exposed to a novel shrimp diet shorten to look more like the guts of carnivorous tadpoles (Ledón-Rettig et al., 2008), thus sparing developmental resources that would otherwise be squandered on an `unemployed' organ(Diamond, 1991). So while there may a short-term cost incurred by individuals ingesting a novel diet(e.g. reduced growth) the neuroendocrine stress axis could facilitate the evolution of a novel feeding strategy by exposing population variation in traits that would ultimately be adaptive for consuming that resource. Once exposed, this variation may become modified and refined by natural selection[`genetic accommodation', sensu West-Eberhard(West-Eberhard, 2003)].

To test the hypothesis that the neuroendocrine axis is activated when a carnivorous diet is introduced to larvae of spadefoot species with the ancestral feeding strategy, we exposed Sc. couchii tadpoles to shrimp and detritus diets from the time of hatching, and compared their response to that of Spea bombifrons Cope, a spadefoot toad species with a derived, carnivorous feeding strategy. We measured growth, developmental rate and whole-body CORT levels through ontogeny in these tadpoles to assess the physiological responses of each kind of tadpole to each diet. We predicted that Sc. couchii tadpoles would show reduced growth and development rates, and experience increased stress axis activity when exposed to shrimp. At the same time, we restricted the food of tadpoles of both species to assess whether growth, developmental and CORT responses to the novel diet were similar to complete nutrient restriction, or whether tadpoles were able to extract some nutrients and buffer the stress response. Finally, we experimentally manipulated glucocorticoid signaling in both species to determine the direct relationship between CORT and growth in these tadpoles. Ultimately, this set of experiments will resolve how the stress axis mediates the fitness costs incurred by these tadpoles when exposed to a novel diet.

All animals were collected near Portal, Arizona, USA. In this area, Spea bombifrons has undergone trophic character displacement with a closely related species, Sp. multiplicata Cope, making this Sp. bombifrons population particularly carnivorous(Pfennig and Murphy, 2000; Pfennig and Murphy, 2002; Pfennig et al., 2007). Despite the recurrence of shrimp from year to year in this area, Scaphiopus couchii feeds solely on detritus and microorganisms, although they will consume shrimp in the lab if it is the only available resource(Ledón-Rettig et al.,2008). Captured animals had been housed in a colony at the University of North Carolina, Chapel Hill for 1–2 years. To induce breeding, male and female adults were injected with 0.07ml luteinizing hormone-releasing hormone (Sigma L-7134, St Louis, MO, USA) and left for 8 h in nursery tanks. All procedures were carried out in compliance with the Institutional Animal Care and Use Committee at the University of North Carolina, Chapel Hill, under application # 03-0110.

One sibship each of Sp. bombifrons and Sc. couchii were bred for this experiment. After hatching (3 days after breeding), larvae from each sibship were transferred to individual 3 oz (∼84 g) plastic cups,assigned a particular treatment, and then randomized and interspersed on racks in the same room maintained at 26°C and on a 14 h:10 h L:D photoperiod. In most anuran larvae feeding is precluded in early development by a plug of endoderm in the esophageal region (Wright,2005). This plug is cleared in spadefoot larvae only after hatching, thus no food was administered during the first 3 days of development. On day 4 after breeding, larvae were fed either brine shrimp nauplii, ground fish food (hereafter, detritus), or received no food; brine shrimp resemble the fairy shrimp that Spea feed on in nature, whereas ground fish food resembles detritus in form and nutrition [fish food produces growth rates in Spea that are similar to growth rates found in natural ponds (Pfennig et al.,2006; Pfennig et al.,1991)]. Unfed tadpoles of both species were only raised until the third day of treatment. All other larvae were fed their respective diets, ad libitum, for the next 7 days. In the natural ponds where spadefoots breed and develop, anostracan shrimp undergo metamorphosis from nauplii to more complex, adult shrimp. Thus, 4 days after the initial feeding,both Sc. couchii and Sp. bombifrons tadpoles, which were initially fed nauplii, were switched to adult shrimp to emulate the natural development of shrimp in field conditions. An additional group of Sc. couchii tadpoles, initially fed nauplii, was switched to no food at all,to determine whether a shrimp diet presented any additional stress, beyond nutritional restriction, after 4 days. Yet another group of Sc. couchii tadpoles initially fed nauplii was kept on nauplii to control for the possibility that elevated CORT in Sc. couchii was due to the presence of the larger, more active adult shrimp, and not a shrimp diet per se (Fig. 1).

Tadpoles of both species were sampled 1, 2, 3 and 7 days after the onset of feeding. Each tadpole was transferred to a 12×75 mm Falcon tube, rapidly frozen by immersion in an ethanol and dry ice slurry, and transferred to a–80°C freezer. For each treatment (species:diet:time) 8–12 larvae were used. All samples were taken at the same time of day(approximately 21:00 h). Weight and developmental stage(Gosner, 1960) were recorded for tadpoles before they were individually transferred to glass tubes for radioimmunoassay (RIA).

Corticosterone was extracted from whole tadpoles following the method of Denver (Denver, 1998) with modifications. Briefly, total lipids were extracted by homogenizing each tadpole in 2 ml ethyl acetate; samples were centrifuged and the supernatant was reduced by rapid evaporation. Samples were spiked with 35,000 cpm tritiated CORT ([³H]CORT, Perkin-Elmer, Waltham, MA, USA) to determine CORT migration distance on silica thin-layer chromatography (TLC)plates (JT Baker Si250F, Phillipsburg, NJ, USA) exposed to a toluene:cyclohexane (1:1) solvent system, followed by a chloroform:methanol(9:1) system (exposed twice). We located the CORT migration distance by scraping silica in 1 cm increments from two extra samples on the ends of the plate, and determining the radioactive peak using a Beckman LS6500 scintillation counter. We then scraped off the region of silica containing CORT from each sample, extracted CORT from the silica with 5 ml anhydrous ether, and dried samples under nitrogen (repeated to improve efficiency). Samples were resuspended in 0.5 ml 0.2 mol l^–1phosphate-buffered saline with 1% gelatin for RIA. We included samples from each diet and time group on each TLC plate and RIA run; samples from each species were run on separate plates and assays. We conducted eight assays in total; the intra-assay coefficient of variation was 9.7%, and inter-assay variation was 11.5%.

Because Sp. bombifrons and Sc. couchii vary drastically in size and developmental speed, species were considered separately for all described analyses. For each treatment (unique time point and diet), the sample size was eight for Sc. couchii and ranged from 8–12 for Sp. bombifrons. To determine the effects of diet on mass in each species, we performed ANOVAs with natural log-transformed mass as the dependent variable and assay, time, diet and the time by diet interaction as factors. We sequentially removed non-significant terms from each model,beginning with higher order terms. To identify specific pairwise differences,we used Tukey's test [multcomp package(Hothorn et al., 2008)] that included all combinations of time points and diets within each species (thus preserving the experiment-wise error rate at 0.05).

To determine how diet affected developmental rate in each species, we used non-parametric Pearson's χ² tests since Gosner stage does not conform to parametric assumptions. Visual inspection of the developmental trends indicated that the most substantial differences in stage occurred 3 and 7 days after the onset of feeding (Fig. 2), so comparisons were made between diet groups collected at those times.

To determine the effect of diet on whole-body CORT, we performed ANCOVAs on each species with natural log-transformed CORT as the dependent variable. Assay, time, diet and the time by diet interaction as factors, and natural log-transformed mass and stage (as a categorical factor) were used as covariates. As with mass, we sequentially removed non-significant terms and performed Tukey's tests between all time points and diets to determine specific pairwise effects.

We also used linear regressions to determine the unique contribution of mass and developmental stage to whole-body CORT in our tadpoles. Species were analyzed separately, but data from all time points and diet treatments were considered. Because mass and CORT change over time, they can possibly generate a spurious correlation between each other(Yule, 1926). Therefore, to remove time-dependent trends, we regressed time with natural log-transformed mass and CORT, separately, and regressed the residuals of these models against each other. In a separate equation, to assess the contribution of development to whole-body CORT, natural log-transformed CORT was regressed on Gosner stages (transformed as Helmert contrasts, which allow for comparisons between qualitative groups that have an a priori ranking), and natural log-transformed mass (to control for the effect of mass).

All statistics were performed in the R 2.8.1 statistical language(R Development Core Team,2008), except for the Pearson's χ² tests, which were performed in JMP 7.0.1 (SAS Institute, Inc., Cary, NC, USA).

To determine whether variation in CORT causes changes in larval growth rate, we treated tadpoles of both species with exogenous CORT (Sigma, C2505)or a glucocorticoid receptor antagonist, RU-486 (mifepristone; Sigma, M8046). Solutions of 250 nmol l^–1 CORT and 100 nmol l^–1 RU-486 were prepared by first dissolving each solute in ethanol, and then diluting each in 0.1× Marc's modified Ringer solution(MMR; final dilution of ethanol: 0.25^–3×). A 0.1×MMR and 0.25^–3× ethanol solution was prepared for the control treatment. The concentrations of CORT and RU-486 were based on those used in previous studies (Crespi and Denver, 2004; Das and Brown,2004), and were adjusted for the relative size and age of tadpoles used in our experiment. Because CORT and RU-486 are lipophilic compounds they are able to permeate tadpole tissues and cell membranes. Adding CORT to water in which tadpoles are housed has been shown to elevate their whole-body CORT with physiological effects (Krain and Denver, 2004).

One sibship each of Sp. bombifrons and Sc. couchii were used for this experiment. Two days after fertilization, 100 larvae were transferred to individual 3 oz plastic cups containing the previously prepared CORT, RU-486, or 0.1× MMR solutions. After 24 h, these individuals were fed 0.25 ml (approximately 10 mg dry mass) brine shrimp nauplii or 10 mg detritus. Cups were randomized and interspersed on racks at 26°C and on a 14 h:10 h L:D photoperiod. Twenty-four hours after feeding, tadpoles that had cleared their cups of food (ensuring that all analyzed individuals had consumed the same amount; N=12–24 per species:diet:hormone treatment) were euthanized with tricaine methosulfonate (MS-222), fixed in 4%paraformaldehyde, and rocked at 4°C overnight. Tadpoles were then dehydrated and stored in 100% ethanol at –20°C. A ventral photograph of each tadpole was captured with a Leica (Wetzlar, Germany) DFC480 R2 camera(magnification ×2.5), and the snout–vent length (SVL), a proxy for growth, was measured using NIH ImageJ(http://rsb.info.nih.gov/ij/).

Separate ANOVAs were conducted for each species using natural log-transformed SVL as the response variable and hormonal treatment, diet and the hormone by diet interaction as the independent variables. The minimum adequate model was found by systematically eliminating non-significant terms,as previously described. Tukey's tests were then performed to determine whether hormonal treatments differentially influenced tadpole growth.

Mass in both species was influenced by assay, time, diet and a time by diet interaction (ANOVA: Sp. bombifrons, assay: F_(3,82)=8.45, time: F_(3,82)=440.24,diet: F_(2,82)=469.35, time by diet: F_(5,82)=15.98; Sc. couchii, assay: F_(3,87)=1.32, time: F_(3,87)=43.58,diet: F_(4,87)=63.25,time by diet: F_(5,87)=6.87, all P-values <0.0001). Unfed tadpoles grew poorly for the first 3 days of the experiment, weighing significantly less than fed tadpoles in both species(Fig. 2). Tadpoles fed either shrimp or detritus exhibited similar growth up to day 3 in both species, but thereafter, only Sc. couchii tadpoles diverged in a diet-dependent manner. Sc. couchii tadpoles fed detritus for 3 and 7 days grew more than similar tadpoles fed shrimp (Fig. 2). There were no differences in body mass between Sc. couchii tadpoles that were fed adult shrimp, nauplii, or no food between 4 and 7 days (Fig. 3).

Diet significantly delayed the development of shrimp-fed Sc. couchii tadpoles at 3 days (Pearson's test: χ²=7.62, P=0.05) and 7 days (χ²=24.00, P=0.0005) after the onset of feeding (Fig. 2). This was not the case for Sp. bombifrons tadpoles, which showed no significant difference in developmental stage at either 3 days(χ²=4.67, P=0.20) or 7 days (χ²=6.48, P=0.09) from the beginning of feeding. Likewise, there were no differences in developmental stage between tadpoles that were feed adult shrimp, nauplii, or no food between 4 and 7 days(Fig. 3).

CORT levels in Sp. bombifrons were significantly influenced by mass, stage and assay (ANCOVA: P=0.02, P=0.03, mass: F_(1,80)=18,095.68, P<0.0001, stage: F_(8,70)=3.87, P=0.02, assay: F_(3,80)=100.75, P<0.0001) and marginally by diet and time (diet: F_(2,80)=2.62, P=0.083, time: F_(3,80)=2.31, P=0.08). Unfed Sp. bombifrons tadpoles generally had higher levels of CORT than fed tadpoles, but there was no difference between the CORT levels of shrimp- and detritus-fed tadpoles (Fig. 2). Although CORT was influenced by the same parameters in Sc. couchii(time: F_(3,78)=5.64, mass: F_(1,78)=264.35, P<0.0001, stage: F_(8,78)=3.87, P=0.0007, assay: F_(3,78)=31.02, P<0.0001, diet: F_(4,78)=7.46, P<0.0001), tadpole CORT varied in diet-dependent manner by 7 days after feeding began (time by diet interaction: F_(5,78)=4.02, P=0.003), with shrimp-fed tadpoles having significantly higher levels of CORT than detritus-fed tadpoles. We found no differences in CORT content between Sc. couchii tadpoles that were fed adult shrimp, nauplii, or no food from day 4 to 7 day (Fig. 3).

Mass had a significant effect on whole-body CORT in Sp. bombifrons(F_(1,95)=1,813.77, R²_adj=0.95, P<0.0001) and Sc. couchii (F_(1,102)=28.62, R²_adj=0.27, P<0.0001) tadpoles after removing the effect of time (Fig. 4). Likewise, developmental stage influenced CORT in both species(Helmert contrasts, Sp. bombifrons, P=0.05; Sc. couchii,P=0.002). When assessed individually, all contrasts between developmental stages in Sp. bombifrons were negative, and two were significant,indicating that CORT concentrations decline as these tadpoles develop. Similarly, in Sc. couchii, six of eight contrasts were negative, and three of these were significant (no positive contrasts were significant),again suggesting that CORT declines during early spadefoot development.

The results of the hormonal manipulation indicated that, regardless of food intake, glucocorticoid exposure affected growth in both species of tadpoles(Fig. 5; ANOVA: Sp. bombifrons, F_(2,110)=67.10; Sc. couchii,F_(2,105)=90.48, both P-values <0.0001). CORT significantly reduced growth in larvae of both species fed either diet after 2 days of exposure. By contrast, blocking glucocorticoid signaling with RU-486 significantly increased growth in Sc. couchii tadpoles. Sp. bombifrons tadpoles also tended to grow more when exposed to RU-486, but this effect was not significant (P=0.09).

In order to investigate the changes that may have accompanied the evolutionary transition to a novel diet, we characterized growth, development and neuroendocrine stress response in two species of spadefoot toads –one that normally consumes only detritus and plankton (Sc. couchii)and a closely related species that can also consume shrimp (Sp. bombifrons) – when both species were given either an herbivorous or carnivorous diet. As predicted, Sc. couchii suffered growth and developmental costs when raised on shrimp but Sp. bombifrons did not. Also as predicted, and also in contrast to Sp. bombifrons, Sc. couchii experienced higher levels of whole-body CORT when raised on shrimp. These increased levels of CORT may be at least partially responsible for the reduced growth and development observed in shrimp-fed Sc. couchii.

Our findings suggest that the reduced growth and development rates of shrimp-fed Sc. couchii tadpoles were likely due to nutritional restriction. Tadpoles fed shrimp showed similar growth and development rates as those that were not fed. Because our tadpoles were fed ad libitum,we cannot say whether dietary restriction was caused by behavior (avoiding the diet) or the inability to assimilate nutrients once the diet was consumed. However, when forced to consume shrimp, most Sc. couchii tadpoles assimilate nutrients less efficiently than they will from the same amount of detritus (Ledón-Rettig et al.,2008). Therefore, at the very least, this species lacks the ability to digest or extract nutrients from shrimp at the level of the gut. Although it is not clear what traits aid Spea tadpoles in digesting shrimp (e.g. nutrient transporters or enzymes), they possess a number of behavioral and morphological features for consuming shrimp, thus rendering shrimp a higher quality diet. For example, compared with Scaphiopus,Spea tadpoles exhibit a stronger behavioral preference to eat shrimp(C.C.L.-R. and D.W.P., unpublished data), and develop enlarged jaw muscles and mouthparts for capturing shrimp (Martin and Pfennig, 2009). Presumably, these features make Speatadpoles better able to capture and ingest shrimp, thereby maintaining growth and development.

In addition to reduced growth and attenuated development, the significant elevation of glucocorticoid production in shrimp-fed Sc. couchiitadpoles after 7 days of feeding suggests that these individuals were inefficiently extracting or assimilating nutrients from their diet. Although increased competition for food (i.e. increased density) and food deprivation elevates CORT levels in larvae of other anuran species [Rana pipiens(Glennemeier and Denver,2002a); Spea hammondii(Crespi and Denver, 2005)],here we show for the first time that CORT level is affected by the quality of a given diet. In this case, the quality of diet is dependent upon a species'evolutionary history with that diet. This increase in corticosterone was likely stimulated by physiological factors indicating a state of negative energy balance and was associated with the mobilization of energy from stored fuels to keep these tadpoles functioning(McEwen and Wingfield, 2003). By 7 days of growth and development, the energetic demands of Sc. couchii tadpoles probably outstripped the nutrients they could extract from a shrimp diet such that development slowed, growth ceased, and CORT synthesis and secretion increased to a level similar to that of tadpoles that were not fed. The novelty of giving tadpoles adult shrimp after 3 days of eating nauplii did not appear to cause the stress response; tadpoles that were raised solely on nauplii also showed increases in CORT.

Although it appears that diet quality clearly affects mass and the activation of the neuroendocrine stress axis in Sc. couchii, it is not entirely clear which causal pathways link these variables. Our multiple regression analysis indicated that CORT concentrations varied inversely with body mass (and to a lesser extent with development stage) in tadpoles, an outcome that is consistent with a recent study of wood frog tadpoles(Belden et al., 2007) and other vertebrates (Kitaysky et al.,1999; Moore et al.,2000). However, the effect of diet on Sc. couchiiwhole-body CORT levels after 7 days of feeding was significant even after statistically controlling for differences in tadpole mass. Thus, it appears that although diet quality has a distinct effect on body mass, additional physiological or central nervous system cues that signal nutrient restriction[e.g. neuropeptide Y (see Crespi et al.,2004)] or unique chemical cues from shrimp are probably stimulating the increases in CORT levels in tadpoles receiving a suboptimal diet. Furthermore, we showed experimentally that exogenous CORT decreases growth in both species, and in Sc. couchii, this decrease in growth was prevented by blocking CORT signaling with a glucocorticoid receptor antagonist. If CORT suppresses growth in Sc. couchii after 7 days of feeding (as it did in our exogenous hormone treatment), then the shrimp diet may have additional negative effects on body mass, either by stimulating an increase in metabolism or inhibition of growth factors associated with glucocorticoid signaling (McEwen and Wingfield, 2003).

Another interesting pattern that was revealed by our study was that the relationship between mass and whole-body CORT varied in the two species. Sc. couchii showed far more variation than Sp. bombifrons in this correlation, even within diet groups. It is unlikely that Sc. couchii larvae are more sensitive to environmental conditions, and that this variation is due to environmental differences among the individual cups in which they were raised. Previous studies have shown that species of Spea are similarly sensitive to even subtle environmental variation,such as water levels (Denver,1998; Boorse and Denver,2004; Gomez-Mestre and Buchholz, 2006). Furthermore, tadpoles of both species were collected at the same time and were spatially interspersed in this study,thus, if microenvironmental variation caused variation in CORT among Sc. couchii tadpoles, it should have done so among Sp. bombifronstadpoles, as well.

An alternate possibility is that whole-body CORT is less coupled to mass in Sc. couchii as a result of their unique life history. For Sc. couchii, maintaining a constant developmental rate in spite of nutritional condition is essential for survival: the larvae of both species develop in ephemeral desert ponds, where the risk of mortality by desiccation is high (Bragg, 1965; Newman, 1987) but Sc. couchii tadpoles develop in ponds that are, in general, smaller and of shorter duration (Bragg, 1965). Possibly, as a result of this strong selection to metamorphose before their pond dries, Sc. couchii has evolved a developmental rate that is independent of food supply and, by extension, mass(Morey and Reznick, 2004). A tight link between whole-body CORT and mass might compromise the ability of Sc. couchii to metamorphose in time; thyroid hormone, which is responsible for metamorphic timing, and CORT are controlled by the same molecule [corticotropin releasing factor(Denver et al., 2002)]. Thus,relatively high variation in whole-body CORT content of Sc. couchiitadpoles may be the result of a relatively weak relationship between an individual's nutritional status and the neuroendocrine axis, which can vary between life stages in spadefoot toads(Crespi and Denver, 2005).

In addition to diet and mass, we also found that CORT content varied significantly with ontogeny. In both species, CORT content (per unit body mass) decreased dramatically with development during the early larval stages(stages 25–35). This pattern is similar to that of Xenopus laevis, where CORT is detectable just before hatching (stage 19),increases to its highest concentration (stage 26), and declines rapidly thereafter (Kloas et al.,1997; Glennemeier and Denver,2002b). By contrast, in Rana pipiens or R. catesbiana CORT is low during early stages and becomes elevated as metamorphosis approaches (Krug et al.,1983; Glennemeier and Denver,2002b). Spadefoot toads and Xenopus are more closely related to each other than either is to Rana, which may indicate that ontogenetic patterns of CORT are, to some degree, phylogenetically conserved. However, few studies have addressed the ecological and evolutionary causes of variation among amphibian neuroendocrine stress axes(Wada, 2008), and only additional studies encompassing a broad range of taxa will shed light on the origins of such variation.

Because Scaphiopus is the most closely related taxonomic group to Spea that maintains the ancestral feeding strategy, it is in a unique position to shed light on the diet-induced responses of ancestral Spea populations. If ancestral Spea populations were like Sc. couchii in their growth, developmental and hormonal responses to shrimp, they may have had to surmount these physiological consequences as they transitioned to a novel diet. These physiological effects could, in turn,contribute to fitness costs, such as mortality resulting from insufficient food or small body size if tadpoles reach metamorphosis. A general issue requiring clarification is how taxa eventually offset these fitness costs. One evolutionary scenario is that populations may endure prolonged periods of reduced fitness before variation arises (through mutation, recombination, or introgression with another species) that facilitates the utilization of a novel resource. An alternative scenario is that natural selection may act on standing variation to produce a population that is specialized for the novel diet. Such standing variation may often be expressed (and, therefore, exposed to selection) through phenotypic plasticity, only after a population actually encounters a novel resource or environment(Waddington, 1959; Rutherford and Lindquist,1998). Diet-induced stress hormones may be a mechanism of mediating the expression of this phenotypic variation. For example, it has been suggested that stress-induced thyroid hormone fueled developmental and morphological plasticity in ancestral groups of spadefoot toads –plasticity that was ultimately responsible for diversification among species(Gomez-Mestre and Buchholz,2006). Future studies will determine whether Scaphiopuspopulations possess heritable variation in diet-induced CORT and its downstream effects on physiology, behavior and morphology. Demonstrating such heritable variation would suggest that diet-induced CORT played a substantial role in ancestral Spea's transition to a novel diet.