Effects of temperature and food quality on anuran larval growth and metamorphosis

Authors


†Author to whom correspondence should be addressed. E-mail: agnic@correo.uniovi.es

Summary

  • 1Anurans exhibit high levels of growth-mediated phenotypic plasticity in age and size at metamorphosis. Although temperature and food quality exert a strong influence on larval growth, little is known about the interacting effects of these factors on age and size at metamorphosis.
  • 2Plasticity in growth rates, maximum larval mass, mass loss, larval period and size at metamorphosis was examined in Iberian Painted Frogs (Discoglossus galganoi Capula, Nascetti, Lanza, Bullini & Crespo 1985) under different combinations of temperature and diet quality.
  • 3Temperature and diet had strong effects on the maximum size reached by tadpoles throughout the premetamorphic stages. Larval body mass varied inversely with temperature. The effect of diet depended on temperature; larvae fed on a ‘carnivorous’ diet (rich in protein and lipids) achieved a larger size than larvae offered an ‘herbivorous’ diet (rich in carbohydrates) at 17 °C but not at 12 or 22 °C.
  • 4Larval period was insensitive to diet composition, and varied only with temperature. Primarily the interacting effects of food quality and temperature affected size at metamorphosis. Size at metamorphosis varied inversely with temperature under the plant- and the animal-based diets. However, the carnivorous diet resulted in bigger metamorphs at 17 and 22 °C, but did not influence final mass at 12 °C. Maximum size over the larval period explained most of the variation in mass loss after the premetamorphic growing phase.

Introduction

Animals with complex life cycles often experience a critical period for survival at the transition between the larval and adult environments (Arnold & Wassersug 1978; Wilbur 1980). Individuals entering a novel habitat are particularly exposed to uncertainty in finding food or encountering predators, and therefore they must achieve a threshold state defined by a set of ecological and physiological conditions (e.g. minimum body size, energy reserve level and locomotion ability) to reduce potential hazards. For instance, in anuran species, large metamorphs may have either greater ability to withstand starvation and desiccation or to escape predators (Tracy et al. 1993; Semlitsch et al. 1999). However, massive mortality is not restricted to the time of transformation and mortality rates can be very high in the premetamorphic stages (Wilbur 1980; Werner 1986). Predation and pond desiccation have been identified in a number of studies as major causes of larval mortality in permanent (or seasonal) and transient aquatic environments, respectively (Brockelman 1969; Calef 1973; Smith 1983; Newman 1987). Both the risk of mortality by predation and desiccation should increase with the time spent in the pond, thus generating a trade-off between pre- and postmetamorphic survival.

As other ectotherms, anurans do not possess an efficient mechanism for physiological thermoregulation (Brattstrom 1963) and this lack in turn determines a strong dependence of growth and differentiation processes upon temperature. Therefore, temperature can be considered, besides energy uptake, the most important proximal cause of variation in size and age at metamorphosis. Low temperatures retard differentiation more than growth, thereby increasing stage-specific size (Smith-Gill & Berven 1979). As a result, larval anurans grown at cold temperatures have prolonged developmental periods but they are also larger as metamorphs than conspecifics grown at warmer temperatures. This phenomenon makes up one of the most general rules for ectotherms (Atkinson 1994, 1996).

Also the quality and quantity of the food accessible over the larval stage have important effects on the time and size at metamorphosis (Leips & Travis 1994). Larval anuran diets vary widely across taxa and environments; tadpoles can feed on vascular plants, a wide variety of algal forms, precipitates of dissolved organic matter, detritus, and live or dead animals including conspecifics or heterospecific tadpoles (see Kupferberg 1997). The differences in nutritional quality among these food categories are very important; for example, plant tissue is generally higher in carbohydrates and lower in lipids and proteins than animal tissue. Diet composition, and especially the relative amounts of protein, carbohydrate and lipid in the diet, can influence thyroid hormone function, which in turn affects growth and differentiation and, ultimately, metamorphosis (Kupferberg 1997). More specifically, several experimental studies have demonstrated that diets containing a large proportion of protein can produce a twofold effect of acceleration of growth and developmental rates (Nathan & James 1972; Steinwascher & Travis 1983; Pandian & Marian 1985a; but see Crump 1990). In addition, tadpoles fed on protein-rich algae appear to have higher intake rates than conspecifics fed on algae with low protein contents (Kupferberg 1997). In this context, selective foraging would allow tadpoles either to maximize size at metamorphosis or minimize the length of the larval phase. This source of plasticity can be particularly important in species where escape from a drying or overcrowded environment can be mediated through cannibalism (Crump 1983, 1990; Pfennig 1992).

The mechanisms underlying the observation of large size or shorter time to metamorphosis in larvae fed on increased protein diets are still puzzling. Crump (1986) suggested that cannibalistic tadpoles received a better balance of essential nutrients, which would result in higher conversion efficiencies. By extension, we could explain the faster growth of tadpoles fed in tissues more similar to their own composition and predict that the addition of animal protein to tadpole diet would enhance growth and development. An alternative hypothesis is that the consumption of animal hormones contained in animal tissues is the proximate mechanism (Kupferberg 1997). The results of experimental studies that aimed to disentangle the direct effects of protein content and thyroid hormones remain controversial. From her experiments with tadpoles fed algae with different food contents, Kupferberg (1997) reported a negative but non-significant correlation between protein content and length of the larval phase. Even so, she hypothesized that changes in size and time at metamorphosis may be mediated through diet-induced changes in thyroid function. However, it should be pointed out that Kupferberg (1997) acknowledges that a diet-based endocrine explanation of changes in metamorphic traits is complex, since nutrition is expected to influence secretion of a number of hormones that have antagonistic developmental effects. For example, in addition to promoting growth in several species of amphibians, prolactin (PRL) antagonizes the action of thyroid hormone (White & Nicoll 1981; Denver 1997) and thus may inhibit differentiation and retard metamorphosis.

Changes in food quantity have been shown to affect amphibian development and, although whether the response is acceleration or deceleration of development depends on the developmental stage at which food levels are changed, most often the net outcome of low resource levels is metamorphosis delay (e.g. Crump 1981; Alford & Harris 1988; Newman 1994; Tejedo & Reques 1994). These results conform well to the Wilbur & Collins (1973) predictions that tadpoles optimally delay metamorphosis either to attain a critical size for metamorphosis or to capitalize on growth opportunities in the aquatic environment. In this sense, a foraging shift towards higher-quality food would provide expectations of enhanced growth conditions.

In this study we examine the effects of food quality and temperature on size and age at metamorphosis of the Iberian Painted Frog (Discoglossus galganoi Capula, Nascetti, Lanza, Bullini & Crespo 1985). This species breeds in small bodies of water, from temporary or semipermanent ponds to very short-lived rain pools. Larval viability relies consequently on rapid development; eggs hatch after 2–6 days at the temperatures most often encountered in the breeding ponds and tadpoles, omnivorous and facultatively cannibalistic, complete their larval development in 2–8 weeks depending on temperature. The occurrence of larvae over a range of aquatic habitats and the tendency of tadpoles to eat conspecifics provide the potential for the evolution of selective foraging and adaptive responses to varying diet quality. Here we addressed two specific questions. First, does developmental rate respond to changes in growth rate mediated by diet quality? If so, there are two possible outcomes. If the shift to an improved diet occurs early in development, tadpoles could delay metamorphosis to capitalize on the growth opportunity. In contrast, if there is an optimal size at metamorphosis (as opposite to a wide range of suitable sizes) that maximizes fitness (Werner 1986) and protein limitation is assumed (Kupferberg 1997), diet enhancement is expected to accelerate development. Second, does diet quality itself influence developmental rate? Metamorphosis is a thyroxine-dependent process; thyroxine promotes differentiation while it inhibits larval growth (White & Nicoll 1981). In turn, low temperatures inhibit the outcome and tissue sensitivity to thyroxine (Kollros 1961). We used a two-factor analysis (temperature and protein content in the diet) to examine the influence of diet quality on size and age at metamorphosis.

Methods

Animals and Rearing Conditions

Over 200 Discoglossus galganoi tadpoles at Gosner (1960) stage 25 were collected from a natural, semipermanent pool near Collada Fumarea, Asturias, northern Spain (43°26′ N, 5°34′ W; 565 m above sea level) on 10 July 1997. The pool is part of a system of rain pools, small semipermanent pools and a few small permanent ponds. Most of the semipermanent pools may dry during short periods in late summer and autumn. Water temperature is highly variable over the year (Álvarez & Nicieza 2002). During the study period, the mean temperature in the breeding site was 17 °C, with average daily temperatures ranging from 15 to 19 °C. In the laboratory, tadpoles were held in plastic trays and fed with algae and rabbit food until the commencement of the experiment, on 15 July 1997 (day 0). Then, they were raised individually in 0·5-l plastic containers. Larvae were fed with food pellets and the excess food removed every second day. Initial ration per tadpole was set at 6 mg day−1. Ration levels were increased as the larval period progressed to keep up with the normal demands of growing and developing animals. Tadpoles were exposed to a 12L : 12D photoperiod throughout the study period and the water in the containers was changed weekly.

Experimental Design and Statistical Analyses

A 2 × 3 factorial design was used to examine the effects of food quality and temperature on larval growth and developmental rates, and postmetamorphic performance. To evaluate the effects of food quality, larvae were fed with either commercial fish food (Trouvit, Trow España S.A., Cojóbar, Burgos, Spain, high protein content, HPC; 46% protein, 22% lipids, 2% carbohydrates, 9% ash) or commercial rabbit food (CUNIASA MATER, ASA S.L., Asturias, Spain, low protein content, LPC; 17% protein, 3% lipids, 15% carbohydrates, 10% ash). A total of 126 tadpoles were haphazardly allocated to each of the six experimental treatments (n = 21). For each temperature, 42 individual vessels (21 for each diet treatment) were randomly placed in two 18-l trays filled with recirculating water. Room (air) temperature was kept below 10 °C, and aquarium heaters were used to raise the temperature to 12, 17 or 22 °C. Water pumps were used to produce water circulation and thus reduce thermal heterogeneity. In addition, to minimize the possible effects of such heterogeneity, the position of the 42 tadpole containers within a given temperature was reassigned at random every 3 days.

Larvae were weighed (±0·1 mg) on 15 July (day 1) and at 6, 13, 17, 22, 32 and 40 days from the commencement of the experiment. Because of the shortening of the larval period at increased temperatures, tadpoles reared at 22 and 17 °C were weighed twice (days 1 and 6) and four times (days 1, 6, 13 and 17), respectively. To analyse the effects of temperature and diet on developmental rate, the number of days elapsed from the start of the experiment (Gosner stage 25) to emergence of at least one forelimb (Gosner stage 42) was recorded. Body mass at Gosner stage 42 was used as a measure of body size at metamorphosis. All containers were checked twice a day for metamorphs. Tadpoles were removed from their container when at least one forelimb had emerged (Gosner stage 42), and then weighed and placed in individual plastic bowls containing a small amount of water (to avoid desiccation) until the tail was completely absorbed (Gosner stage 46). Body size of larvae peaks at some point during late premetamorphic stages (before forelimb emergence) and then declines as they approach the metamorphic climax. Mass decrease associated with metamorphosis was estimated as the difference between the maximum body mass attained throughout the premetamorphic stages (thereafter maximum mass) and body mass at stage 46.

Age and size at metamorphosis are often highly correlated. Therefore, a multivariate analysis of variance was first conduced, and then univariate two-way anovas for the effects shown significant in manova. To maintain balanced data, cell sizes were equalized by random case removal. Two-way anovas were also employed to analyse temperature and diet effects on maximum body size and mass decrease. In addition to anova, ancova as used whenever there was a major interest in evaluating the effects of diet and temperature on size-independent responses. All data were log-transformed to meet the assumptions of parametric analysis of variance. Because in some cases (anova and ancova for mass decrease) the design had unequal replication, the analyses were carried out using Type III sum of squares. The Sen & Puri non-parametric test was used to check the assumption of homogeneity of variances in multivariate analyses (all P > 0·80), and the Cochran C-test for univariate anovas or ancovas (all P > 0·05). Prior to performing an ancova, the assumption of homogeneity of the six regression slopes was confirmed; the significance values obtained in the tests of parallelism were in all cases greater than 0·40. The statistical program package SPSS (v. 10) was used for all analyses.

Results

The initial assignment of individuals to treatment groups successfully randomized body mass among experimental groups, size differences among the experimental groups at day 1 being negligible (one-way analysis of variance; F5,120 = 0·89, P = 0·49). There were no differences among group variances (Cochran's C = 0·22, P = 0·89). As expected, growth (mass gain adjusted to initial body mass) during days 1–7 increased with temperature (ancova; F2,118 = 44·95, P < 0·000001; Fig. 1). Data did not show a consistent effect of diet quality (F1,118 = 1·78, P = 0·184), but revealed a significant temperature × diet interaction. Post hoc comparisons between food treatments within temperatures indicated only marginally significant differences; HPC larvae showed rapid growth compared with LPC larvae at 17 (P = 0·056) and 22 °C (P = 0·10) but not at 12 °C (P = 0·20; Fig. 1). Survival to metamorphosis was high and uniform across temperature treatments (likelihood ratio χ2 = 0·24, df = 2, P = 0·89). Tadpoles attained a maximum body mass after 6 (22 °C), 17 (17 °C) or 40 days (12 °C) and thereafter there was a continuous decline associated with metamorphosis. Although the correlations between initial (stage 25) and maximum mass were not significant in all treatments (Table 1), a combined probability test (Sokal & Rohlf 1981) indicated that overall there was a reasonably strong correlation between these variables (χ2= 38·65, df = 12, P = 0·00012). Increasing temperature had a strong negative effect on the maximum mass achieved by tadpoles during the premetamorphic phase (anova; F2,66 = 146·56, P < 0·000001; Fig. 2). Also there was a clear diet effect (F1,66 = 11·87, P = 0·00099): overall HPC food produced heavier tadpoles than LPC food. However, the influence of the diet was not uniform across temperature treatments (interaction term; F2,66 = 4·09, P = 0·021; Fig. 2). Tadpoles fed a diet with high protein content were significantly bigger than ‘herbivorous’ tadpoles at 17 °C, but the differences were weak at 12 and 22 °C (Tukey HSD tests, comparisons between diets within temperature levels; 12 °C, P = 0·274; 17 °C, P = 0·0028; 22 °C, P = 0·99). Mass at metamorphosis (stages 42 and 46) tended to be positively correlated with the maximum mass (Table 1); the combined probability tests confirmed overall strong correlations between the maximum size attained over the premetamorphic period and size at metamorphosis (stage 42: χ2= 86·60, df = 12, P < 0·00001; stage 46: χ2= 76·61, df = 12, P < 0·00001).

Figure 1.

Influence of temperature and food quality on maximum mass gain of painted frog tadpoles following least-squares (LS) adjustment for initial body mass (means ± SE). Diet treatment: HPC, high protein content; LPC, low protein content.

Table 1.  Pearson product–moment correlations between maximum mass achieved at any point of the larval phase and mass at Gosner (1960) stages 25, 42 and 46 in different temperature and diet conditions. LPC and HPC are low and high protein content diets, respectively. The significance of estimates within each analysis was assessed using α-levels calculated according to the sequential Bonferroni technique (Rice 1989). The estimates that remained significant after Bonferroni correction are in bold type
TemperatureDietMass at stage 25Mass at stage 42Mass at stage 46
rPnrPnrPn
12 °CLPC0·77< 0·0001210·500·0562150·760·0271 8
 HPC−0·010·9646210·84< 0·0001180·640·025112
17 °CLPC0·340·1464190·85< 0·0001190·85< 0·000119
 HPC0·520·0182200·89< 0·0001200·85< 0·000116
22 °CLPC0·410·0672210·550·0591120·780·001613
 HPC−0·120·5863210·040·9055120·310·328912
Figure 2.

Growth trajectories of painted frog larvae reared at 12, 17 or 22 °C and fed on high- or low-protein diets from the start of the experiment (Gosner stage 25) to metamorphosis. Arrows indicate the commencement of metamorphosis (Gosner stage 42).

Body masses at stages 42 and 46 were highly correlated regardless of rearing temperature and diet (Table 2). Some accidental mortality occurred between stages 42 and 46 and therefore most of the analyses presented here are for metamorphic traits measured at stage 42. Anyhow, the results obtained for age and mass at stages 42 and 46 were basically coincident. manova indicated that temperature (Wilks’ lambda = 0·016, P < 0·000001) and diet (Wilks’ lambda = 0·791, P = 0·00049) both significantly affected the multivariate response in age and size at metamorphosis. There was a marginally significant interaction between food and temperature (Wilks’ lambda = 0·873, P = 0·063). Overall, there was no significant correlation between initial size (mass at stage 25) and mass at metamorphosis (χ2= 10·03, df = 12, P = 0·62), but larval period was negatively correlated with initial size (χ2= 49·91, df = 12, P < 0·00001; Table 2). However, inclusion of initial mass as a covariate did not substantially alter the results for the main effects of temperature and diet; the temperature × diet interaction was not significant (mancova; P = 0·11). Developmental rate over stages 25–42 ranged from a mean value of 0·0719 day−1 at 22 °C with the low-protein diet, to 0·0153 day−1 at 12 °C with the high-protein food. Temperature had a strong effect on developmental rate (F2,66 = 1367·65, P < 0·000001). On average, the length of the exogenous feeding phase ranged from 14·5 days at 22 °C to 65·5 days at 12 °C. There were no significant effects of diet (F1,66 = 0·89, P = 0·35) and neither temperature × diet interaction (F2,66 = 0·83, P = 0·44). ancova (initial mass used as covariate) produced identical results. Mass at metamorphosis was affected by temperature (anova; F2,66 = 58·91, P < 0·000001) and diet (F1,66 = 14·14, P = 0·00036). As for maximum mass, mass at metamorphosis varied inversely with temperature, and tadpoles feeding on a rich diet (high protein content) were heavier than tadpoles feeding on a low-protein food (Fig. 3). Again, the effect of diet was not the same at all temperatures (interaction term; F2,66 = 3·26, P = 0·044). In this case, there were not differences at 12 °C (Tukey HSD test, P = 0·99), but ‘carnivores’ had a higher body mass than ‘herbivores’ at 17 °C (Tukey HSD test, P = 0·0044) and tended to be heavier at 22 °C, although the differences were not significant (Tukey HSD test, P = 0·14). The diet effect heightened during metamorphic climax (stage 46: F1,66 = 28·82, P < 0·000001; Fig. 3), which resulted in bulkier tadpoles in the high-protein treatment at 17 and 22 °C (Tukey HSD tests, both P < 0·0003); there was no significant difference at 12 °C (Tukey HSD test, P = 0·96).

Table 2.  Pearson product–moment correlations between mass at stage 42 and mass at stage 46, and between initial mass (stage 25) and age and mass at stage 42 in different temperature and diet conditions. LPC and HPC are low and high protein content diets, respectively. The significance of estimates within each analysis was assessed using α-levels calculated according to the sequential Bonferroni technique (Rice 1989). The estimates that remained significant after Bonferroni correction are in bold type
TemperatureDietMass st 42–mass st 46Mass st 25–mass st 42Mass st 25–days st 42
rPnrPnrPn
12 °CLPC0·8010·0172 80·2010·473215−0·630·012615
 HPC0·8100·0011120·1680·5051180·6390·004018
17 °CLPC0·7450·0003190·2270·350819−0·4470·055119
 HPC0·7640·0007160·3740·1053200·7370·000220
22 °CLPC0·8030·0306 70·0550·864512−0·5320·074312
 HPC0·9010·0054 70·0540·867212−0·2940·351112
Figure 3.

Temperature and diet influence on body mass (means ± SE) of painted frogs at forelimb emergence (Gosner stage 42) and complete tail resorption (Gosner stage 46).

Rearing temperature had a marked effect on mass loss during metamorphosis (F2,73 = 114·25, P < 0·000001; Fig. 4). On average, absolute lessening increased with decreasing temperatures. Post hoc tests revealed significant differences between tadpoles reared at 17 and 22 °C (Tukey HSD, P = 0·00011), and between 12 and 17 °C (P = 0·0087). There was no overall effect of diet on mass change (F1,73 = 0·0024, P = 0·96), although the anova revealed a significant interaction between temperature and diet (F2,73 = 7·79, P = 0·00089) that was mainly caused by a deviation in difference between the LPC and the HPC treatments at 22 °C (Tukey HSD, P = 0·023) compared with the other two temperatures (Tukey HSD, P > 0·45). However, the variation in absolute mass reflected mainly variation in maximum size. In fact, the relative loss was strikingly regular (44–48% of the maximum mass) with the exception of tadpoles in the HPC-22 treatment, which lost 22·5% of maximum mass. After adjustment for variation in maximum mass, rearing temperature had no effect on mass loss (ancova; F2,72 = 1·95, P = 0·15; Fig. 4). Moreover, herbivore larvae lost a larger proportion of the tissues gained through premetamorphosis than ‘carnivores’ (ancova; F1,72 = 10·80, P = 0·0016), although this diet effect varied among temperatures (ancova; F2,72 = 6·36, P = 0·003): it was negligible at 12 °C (Tukey HSD; P = 0·86), weak at 17 °C (P = 0·096), and large at 22 °C (P = 0·0001; Fig. 4).

Figure 4.

Temperature and diet influence on absolute (bars: tinted bars, high protein content; open bars, low protein content) and relative (lines: percentage of, and least-squares adjustment to maximum mass attained over the larval period) mass losses of painted frog larvae (means ± SE).

Discussion

Temperature had persistent effects on the development and metamorphic traits of Discoglossus galganoi. As expected, Painted Frogs metamorphosed at an older age and larger body size when reared at low temperature. These results obey a general rule for ectotherms (Atkinson 1994, 1996). That is the expected outcome if differentiation rates are more responsive to temperature than growth rates (Smith-Gill & Berven 1979). The effects of diet quality and the interaction between temperature and diet are vaguely known and less predictable. If feeding rates are unchanged, a shift towards a richer, more energetic food is expected to cause faster growth, therefore allowing for either shorter larval periods and the increase in metamorphic size (Pandian & Marian 1985a) or extended larval periods to capitalize on the rapid-growth opportunity (Wilbur & Collins 1973). Our study revealed that the effects of food quality on larval growth and metamorphosis, far from being clear-cut, were closely dependent on developmental temperature.

Age and Size at Metamorphosis

Our data suggest that growth rates of D. galganoi are influenced by both temperature and diet quality, but developmental rates respond only to temperature. The diet-based enhancement of growth further translates into a larger size at metamorphosis without change in age at metamorphosis, which is mostly influenced by temperature effects on differentiation rates. Temperature can affect metamorphic size in two ways. First, large differences in size are the result from its differential effects on growth and differentiation (Smith-Gill & Berven 1979). Second, temperature influences the extent that food quality can affect growth and thereby size at metamorphosis; protein-rich food had a positive effect on metamorphic size at intermediate and high temperatures, but its influence became negligible at moderately low temperatures. It should be pointed out that the growth trajectories of larvae fed on animal and plant diets diverged during the phase of rapid growth at 22 °C and especially at 17 °C, but carnivores and herbivores showed parallel trajectories when reared at 12 °C (Fig. 2). Therefore, the interacting effects of diet and temperature on size at metamorphosis basically reflect the influence of these factors on growth rates. Mass decrease through metamorphosis was influenced also by the diet × temperature interaction and, again, this is evidenced as a greater differentiation of herbivorous and carnivorous individuals at the highest temperature. In fact, whereas at 22 °C carnivores experienced mass decreases of about a quarter of their maximum mass, mass decrease in the rest of the groups consistently approached one-half of the premetamorphic maximum mass. Because all the frogs were maintained in a common environment between stages 42–46, an intuitive explanation is that the lower mass loss of carnivores at 22 °C could be a result of a relatively high retention of water during metamorphosis. However, the dry mass and lipid content of newly metamorphosed Painted Frogs increased with decreasing temperatures and increasing protein and lipid content in the larval diet (Álvarez & Nicieza 2002), a pattern that resembles the changes reported here for metamorphic fresh mass. In addition, temperature and diet had no significant effects on body condition of juvenile frogs (Álvarez & Nicieza 2002). All together, these observations suggest that the relative water content at metamorphosis could be similar in all experimental groups. Moreover, most probably, frogs from a given treatment would lose a lower proportion of water during metamorphosis if they had lower water content during the premetamorphic stages as a consequence of, for example, the replacement of water by lipids. However, the lipid content of frogs reared at high temperature (17 or 22 °C) was lower than for cold-reared individuals (12 °C) even after correction for differences in body size (Álvarez & Nicieza 2002). Therefore it seems that mass decrease through metamorphosis was qualitatively similar, in terms of water and whole dry mass, for all the experimental groups. If so, a carnivorous diet could confer a superior efficiency during metamorphosis in warm environments where high developmental temperatures can have a negative influence on body size, locomotor performance and the amount of lipid reserves (Álvarez & Nicieza 2002). In fact, whereas tadpoles reared at 22 °C as carnivores produced frogs that were similar in lean mass and lipid content to those emerging from the 17 °C tanks, the 22 °C herbivorous individuals metamorphosed with about half of the lean mass and lipid content of those.

Food quality did not affect the length of the larval period of D. galganoi, although an animal-based diet enhanced their growth rates. This suggests that the differentiation rate of D. galganoi is relatively unresponsive to the protein and lipid content of the diet. Interestingly, within some limits, differentiation was insensitive to variation in growth rates. From the observation of non-significant tendencies some previous work suggested, but did not demonstrate, that an increase in either the protein content of the food or the fraction of animal tissues in the diet could shorten the period of premetamorphic growth (Nathan & James 1972; Kupferberg 1997). Experiments conducted with Hyla chrysoscelis tadpoles fed on different amounts of low-protein and high-protein food over a 7-day period revealed that particular ratios of protein to carbohydrate can in fact alter both developmental and growth rates (Steinwascher & Travis 1983). However, the experimental design used by Steinwascher & Travis (1983) was not the most appropriate to evaluate the effects of food quality, since these could be at least partially confounded with food quantity effects. Pandian & Marian (1985b) reported larval periods varying between 24 and 96 days in groups of Rana tigrina tadpoles reared at 27 °C and fed on either plant or animal tissues, but remarkably they did not make any additional reference to the influence of food type on larval period. Obviously, this range is meaningless without information about mean values (not provided) and information on the use by tadpoles of the different food types. In fact, the large difference reported (24–96 days) suggests that either the range included individuals with abnormally fast or slow development, or some of the food types had very low digestibility for tadpoles, thus confounding diet quality with food quantity. More refined experiments revealed that increase in protein content in the diet had no strong effect on developmental time (Crump 1990; Babbitt & Meshaka 2000). In conclusion, there is no evidence that an acceleration of growth mediated by food quality could result in faster larval development. That can be viewed as contrary to the expectations for species that, like D. galgonoi, are associated with ephemeral breeding habitats in which any reduction of the larval period would probably convey increased fitness.

Food availability has been proved to affect larval developmental rates in several anuran species. In most cases, the effect of enhanced growth conditions is faster development (Wilbur 1977; Travis 1984; Newman 1989, 1994; Leips & Travis 1994). However, experimental manipulations of ration size at different points of ontogeny have revealed much more complex outcomes (Alford & Harris 1988; Newman 1994; Tejedo & Reques 1994; Audo et al. 1995), which generally support the Wilbur & Collins (1973) model for amphibian metamorphosis. This model assumes that metamorphosis must proceed within a given range of sizes. If the environmental conditions allow for rapid growth, tadpoles will postpone metamorphosis to capitalize on the opportunity of entering the terrestrial habitat with a larger body size. In contrast, under poor growth conditions, the effect of food shortages on larval period will depend on the particular time at which the food reduction occurs. Before the attainment of the minimum size for metamorphosis suboptimal intake rates will result in longer larval periods. Under poor growth conditions, metamorphosis will proceed as soon as tadpoles reach the critical size, and therefore if the food shortage occurs late in development the result is a shortening of the larval period. Moreover, temporal changes in food availability can affect the length of the larval period; for example, in Rana temporaria metamorphosis occurs at an earlier age when food is available only during daytime than when it is continuously accessible (Nicieza 2000). Therefore, the lack of strong evidence for food quality effects on the rates of development is, at least, surprising. In our study, carnivorous tadpoles grew faster and presumably reached critical sizes earlier than herbivorous larvae. According to the Wilbur–Collins model we would expect that thereafter these fast-growing tadpoles had delayed the onset of metamorphosis compared with the herbivorous, slow-growing larvae. Hence, the Wilbur–Collins model provides a conceptual framework to understand how differences in food quality that induce changes in growth rates can alter final body size without change in the overall rate of development. Obviously, this assumes that the delay in the time of metamorphosis is compensated by the faster growth in early development (i.e. a shorter time to reach the threshold size). Here, the important reading is that the larval period could be fixed (as short as possible) with regard to minor variations in feeding conditions. This may be attained by reducing the minimum size at metamorphosis, which is consistent with a high risk of desiccation or predation in the breeding habitats.

An important question is why differences in food composition that alter growth rates at 17 °C are insufficient to produce an equivalent effect at lower temperatures. Differences in the time necessary for digestion of these foods could explain the absence of effects at constant low temperature. Transit times of the food through the digestive tracts of ectotherms increase exponentially with decreasing temperatures (e.g. Elliott 1972; Nicieza, Reiriz & Braña 1994). At low temperatures an increase in the time required to process food can set a limit for intake rates, which would compensate or reduce the gain associated with the higher energy value of the diet. Above a threshold temperature food processing is not slow enough to curtail intake rates and therefore it is unlikely tadpoles experience a trade-off between food quality and quantity. A simpler explanation is that, because food intake is very low at 12 °C, moderate differences in food composition become irrelevant. Moreover, digestion efficiencies increase along with an increase in transit times associated with decreasing temperatures (Nicieza et al. 1994), so that the relative digestibility of different food types can change differentially with temperature. Finally, the diet–temperature interaction might be related to quality-dependent food intake. There is some evidence that feeding rates of larval anurans could increase with food quality (Kupferberg 1997). In warm water, short transit times allow for further increase in daily intake, so carnivores would take a larger amount of richer food than herbivorous tadpoles. However, if food processing is long enough to constrain intake rates, these are unlikely to be quality-dependent and therefore herbivores and carnivores would take similar amounts of food. We have also provided evidence that initial larval size has a positive influence on maximum tadpole size, which in turn correlates with metamorphic size. Despite that, size at metamorphosis was unrelated to initial larval size; instead of maintaining their size advantage, large larvae metamorphosed earlier than small larvae. In any case, in conjunction with the results presented in a related study (Álvarez & Nicieza 2002), our data suggest that selection of either the thermal habitat or the diet could allow tadpoles to have some control on their future performance in the terrestrial environment and ultimately on the chance of juvenile survival.

Acknowledgements

This research was supported by DGES research grant PB96-0861. We thank Penelope J. Watt, Neil B. Metcalfe and an anonymous reviewer for discussion and comments on an earlier version of the manuscript.

Ancillary