Oecologia (2002) 131:186–195DOI 10.1007/s00442-002-0876-x

Abstract Anuran larvae exhibit high levels of phenotyp-ic plasticity in growth and developmental rates in re-sponse to variation in temperature and food availability.We tested the hypothesis that alteration of developmentalpathways during the aquatic larval stage should affectthe postmetamorphic performance of the Iberian paintedfrog (Discoglossus galganoi). We exposed tadpoles todifferent temperatures and food types (animal- vs. plant-based diets) to induce variation in the length of the larvalperiod and body size at metamorphosis. In this species,larval period varied with temperature but was unaffectedby diet composition. In contrast, size at metamorphosiswas shaped by the interaction between food quality andtemperature; tadpoles fed on an animal-based diet be-came bulkier metamorphs than those fed on plant-basedfood at high (22°C) but not at low (12°C) temperature.Body condition of newly metamorphosed frogs was un-related to the temperature or food type experienced dur-ing the premetamorphic stage. Frogs maintained at hightemperature during the larval period showed reducedjumping ability, especially when fed on the plant-baseddiet. However, when considering size-independent jump-ing ability, cold-reared individuals exhibited the lowestperformance, and herbivores reared at 17°C the highest.Cold-reared (12°C) frogs accumulated larger amounts ofenergy reserves than individuals raised at 17°C or 22°C.This was still the case after correction for differences inbody mass, thus indicating some size-independent effectof developmental temperature. Despite the higher lipidcontent of the carnivorous diet, the differences in energyreserves between herbivores and carnivores were rela-tively weak and associated with differences in body size.These results suggest that the consequences of environ-mental variation in the larval habitat can extend to the

terrestrial phase and influence juvenile growth and sur-vival.

Keywords Deferred effects · Discoglossus galganoi ·Jumping performance · Lipids · Phenotypic plasticity

Introduction

The evolution of complex life cycles has been deemed asolution to disrupt genetic correlations that arise whenthe same traits must be expressed in contrasting envi-ronments as a result of an ontogenetic niche shift (Ebenman 1992). In these cases, metamorphosis may al-low traits designed for a given function to evolve inde-pendently in different life stages. Nevertheless, eventhough metamorphosis may reduce the ontogenetictrade-offs derived from the genetic coupling of morpho-logical or physiological traits, trade-offs can arise be-tween functionally unrelated traits, as a result of devel-opmental or physiological constraints. To understandthis point, it may be helpful to split complex life cyclesinto the larval, metamorphic and postmetamorphic phas-es. Often larval stages are exposed to intense mortalityintrinsically associated with habitat requirements, andthus rapid development or shortening of the larval phasecan be beneficial (Newman 1989). However, the trunca-tion or acceleration of development can entail a numberof costs that would become effective at the transitionbetween the larval and adult environments (Arnold andWassersug 1978; Wilbur 1980). In many species, this isa critical period for survival because individuals enter-ing a novel habitat can be highly vulnerable to predatorsor the uncertainty of finding food. Therefore, the needto achieve certain threshold states for a set of ecologicaland physiological traits (e.g. body size, energy reserveslevel, and locomotion ability) can constrain the evolu-tion of larval physiology and behaviour. Finally, juve-nile growth and survival to adulthood can be condi-tioned by size-dependent and size-independent perfor-mance in a variety of functions.

D. Álvarez · A.G. Nicieza (✉ )Área de Ecología, Departamento de Biología de Organismos y Sistemas, Universidad de Oviedo, 33071 Oviedo, Spaine-mail: agnic@sci.cpd.uniovi.esTel.: +34-985-104788, Fax: +34-985-104866

E C O P H Y S I O L O G Y

David Álvarez · Alfredo G. Nicieza

Effects of induced variation in anuran larval development on postmetamorphic energy reserves and locomotion

Received: 4 September 2001 / Accepted: 4 January 2002 / Published online: 20 February 2002© Springer-Verlag 2002

187

Anurans are excellent subjects with which to investi-gate ontogenetic trade-offs. During their aquatic phase,heavy mortality occurs due to predation and pond des-iccation (Brockelman 1969; Calef 1973; Wilbur 1980;Smith 1983; Werner 1986; Newman 1987). The risks ofmortality by predation and desiccation increase withthe time spent in the larval stage. However, delayedmetamorphosis often results in larger body size, andlarge metamorphs can have a greater ability to with-stand starvation and desiccation and to escape predators(Tracy et al. 1993; Semlitsch et al. 1999). Thus, this sit-uation should impose a trade-off between pre- and post-metamorphic survival. The environmental conditionsexperienced by larvae can have consequences on post-metamorphic success not only through indirect effectsderived from changes in metamorphic traits (Smith1987), but through their direct effects on traits influenc-ing survival and growth in juvenile stages (Scott 1994).In ectotherms, temperature experienced during embry-onic development has been shown to affect juvenileshape (Swain and Lindsey 1986; Shine and Harlow1993; Blouin and Brown 2000), trunk segmentation(Tåning 1952; Lindsey and Harrington 1972) and mus-cle structure (Stickland et al. 1988; Johnston et al.1997). Moreover, phenotypic variation in these traitshas a significant impact on locomotor performance(Swain 1992a; Shine and Harlow 1993; Tejedo et al.2000a) and is subjected to strong selection associatedwith predation (Swain and Lindsey 1984; Swain 1992a,1992b). Low temperatures retard differentiation morethan growth, thereby increasing stage-specific size(Smith-Gill and Berven 1979). As a result, larval anu-rans grown at cold temperatures have prolonged devel-opmental periods but they are also larger as met-amorphs than conspecifics grown at warmer tempera-tures. In addition, diet composition can influence thy-roid hormone function, thus affecting growth and dif-ferentiation (Kupferberg 1997). Earlier experimentalstudies have suggested that diets containing a large pro-portion of protein can produce a twofold effect of ac-celeration of growth and developmental rates (Nathanand James 1972; Steinwascher and Travis 1983; Pandianand Marian 1985; Pfennig 1992). Therefore it is theo-retically possible that larval anurans can have somecontrol over their age and size at metamorphosis if en-vironmental heterogeneity permits selective foraging ordevelopment of thermal preferences. Despite their po-tential importance, the extent to which environmentalconditions during the larval phase influence postmeta-morphic success remains largely unexplored. As a con-sequence, we have little information on the interdepen-dence, in terms of fitness, of the different stages thatmake up complex life cycles.

Locomotor performance and the level of energy re-serves are probably two crucial traits in determining thesurvival and early growth of newly metamorphosedanurans (Pfennig 1992; Watkins 2001). Moreover, theconsequences of variation in these two traits are mostlikely interdependent. Jumping ability can have a posi-

tive influence on food acquisition (Walton 1988) andpredator avoidance (Wassersug and Sperry 1977). Inturn, energy reserves will become increasingly impor-tant if food acquisition is constrained, whether this isdue to lack of foraging experience, food shortage, or anelevated risk of predation associated with foraging. Se-lection on both locomotor performance and lipid stor-age can be decoupled, at least partially, with selectionon traits maximizing survival during the larval period(Shaffer et al. 1991). For example, it has been suggest-ed that the maintenance of within-population variationin diet can be mediated by a trade-off between rapidgrowth and development and lipid storage (Pfennig1992). Moreover, if the morphology or the muscularphenotypes of metamorphs are affected by temperatureand the rate of energy acquisition experienced by larvae(see Emerson 1986), we can expect that the influenceof these factors continues beyond metamorphosisthrough a direct effect on locomotor performance (Lutzand Rome 1994; Tejedo et al. 2000a).

In this study we set out to investigate the effects offood quality and temperature experienced by larval Ibe-rian painted frogs (Discoglossus galganoi) on locomotorperformance and energy reserves at the onset of terres-trial life. We measured the total amount of nonpolar lip-ids and jumping performance of newly metamorphosedfrogs to explore the potential coupling of larval andpostmetamorphic performance. This species breeds insmall bodies of water, from temporary or semi-perma-nent ponds to very short-lived rain pools. As a conse-quence, larval viability relies on rapid development;eggs hatch after 2–6 days at the temperatures most oftenencountered in the breeding ponds and tadpoles are om-nivorous but facultatively cannibalistic. Larval develop-ment is usually completed in 2–10 weeks depending ontemperature. The occurrence of larvae over a range ofaquatic habitats and the tendency of tadpoles to eat con-specifics provide the potential for the evolution of selec-tive foraging and adaptive responses to varying dietquality. Here we test the hypothesis that environmentalfactors affecting larval growth and development canalso influence size-independent attributes of juveniles.These effects could either boost or counteract the effectsof size. In a previous study based on the same designand source of experimental animals we showed a strongeffect of temperature on the age at metamorphosis of D. galganoi, whereas body size was primarily affectedby the inter-action between temperature and food com-position (D. Álvarez and A. G. Nicieza, unpublishedmanuscript). Therefore we expect at least a change inmetamorphic energy reserves or locomotor ability as aconsequence of the alteration of the timing of develop-ment. In addition to a trade-off between leaving theaquatic habitat early and doing it at a larger size, wesuggest that trade-offs may exist also between develop-mental traits and measures of size-independent perfor-mance in the terrestrial phase. Costs of large body sizesor short larval periods can be manifested as a subopti-mal morphology in the postmetamorphic stage resulting

in decreased locomotor performance and reduced levelsof energy reserves.

Materials and methods

Animals and rearing conditions

We collected over 200 Discoglossus galganoi tadpoles at Gosner(1960) stage 25 from a breeding site at Collada Fumarea, north-ern Spain (43°26′N, 5°34′W; 565 m a.s.l.) on 10 July 1997. Thepool is part of a system of rain pools, small semi-permanentpools and a few small permanent ponds. The water temperatureis highly variable over the year, while daily fluctuations are lessthan 9°C (Fig. 1). In the laboratory, tadpoles were held in 12-lplastic trays and fed with algae and rabbit food until the com-mencement of the experiment, on 15 July 1997. Then, they wereraised individually in 0.5-l plastic containers. We fed larvae withfood pellets and removed the excess food every second day. Ra-tions were set initially at 6 mg·day–1 (≈7% of tadpole bodymass), and then were increased as the larval period progressed tokeep up with the normal demands of growing animals. Tadpoleswere exposed to a 12-h light:12-h dark photoperiod throughoutthe study period (the natural photoperiod at that time of year var-ied from 15-h light:9-h dark to 11-h light:13-h dark) and the wa-ter in the containers was changed weekly.

We used a 2×3 factorial design to examine the effects of foodtype (plant- vs. animal-based diets) and temperature (12, 17, and22°C) on larval development and postmetamorphic performance.To evaluate the effects of food quality, we fed larvae with eithercommercial fish food [Trouvit 2 (Trouw-E), high protein content(HPC); 46% protein, 22% lipids, 2% carbohydrates, 9% ash] orcommercial rabbit food [Cuniasa, low protein content (LPC);17% protein, 3% lipids, 15% carbohydrates, 10% ash]. A total of126 tadpoles were haphazardly allocated to each of the six ex-perimental treatments (n=21 per treatment). For each tempera-ture, we randomly placed 42 individual vessels (21 for each diettreatment) in two 18-l trays filled with re-circulating water. Wekept the room (air) temperature at <10°C, and used aquariumheaters to raise the temperature to 12, 17, or 22°C. Water pumpswere used to produce water circulation and thus reduce thermalheterogeneity. In addition, to minimize the possible effects ofsuch heterogeneity, the position of the 42 tadpole containerswithin a given temperature was re-assigned at random every3 days. We checked all containers twice a day for metamorphsand recorded larval period (measured as the number of dayselapsed from the start of the experiment) and body mass(±0.1 mg) at Gosner stage 42. At that point tadpoles were trans-ferred to individual plastic bowls containing a small amount ofwater to avoid desiccation, and placed in a 16°C (±2°C) temper-ature-controlled room until they completed tail resorption (Gosner stage 46). Metamorphs were fed ad libitum with adultDrosophila melanogaster for 7–10 days prior to being subjectedto the jumping trials.

Data collection and analyses

We examined locomotor performance and lipid content of newlymetamorphosed frogs to ascertain if the environmental condi-tions experienced by the larvae can affect various aspects of ju-venile performance. As a measure of locomotor performance, weused jumping ability because this trait can influence escape frompredators and foraging success in this species, and hence canprovide a functional link between selection of diet and thermalenvironment and fitness. To assess jumping performance, weplaced the frogs on a clean, flat surface and chased them to in-duce an escape response; we recorded a total of five jumps perindividual. The locomotor performance of each frog was deter-mined using two response variables: (1) maximum jump dis-tance, defined as the length of the longest leap, and (2) average

188

jump distance, defined as the mean length of the five leaps. Weruled out all the small jumps that were unequivocally far belowthe actual capacity of individuals (about less than five bodylengths). The temperature in the room was 22–26°C, a rangecommonly experienced in the field. Frogs were moved into theroom to acclimatize to the new temperature for 60 min before theperformance measurements. Following the jumping trials, themetamorphs were measured for snout-vent length (SVL;±0.01 mm; three repeated measurements per individual) andfresh body mass (±0.1 mg).

Total nonpolar lipid levels were determined by extraction ofdried samples using petroleum ether, which is a highly efficientsolvent for the extraction of nonpolar (storage) lipids, with littleremoval of polar (structural) lipids (Dobush et al. 1985). Ani-mals were oven-dried at 35°C for 4 days to a constant mass,weighed (±0.01 mg) before and after extraction, and placed inEppendorf tubes. Lipids were extracted by adding 2 ml solventto each tube, removing the ether after 60 min, and repeating for atotal of six ether washes (Heulett et al. 1995). The lipid contentwas calculated as the difference between fat and lean masses. Tovalidate this “cold-extraction” method, we conducted a secondextraction in a Soxhlet apparatus for 3 h, and recalculated theamount of remaining lipids as above.

Because we were interested in both absolute and size-inde-pendent performance, leap length and lipid storage were analy-sed by two-way ANOVAs and analyses of covariance (ANCO-VAs). We used fresh body mass as a covariate for jump distance,and dry mass after total extraction of nonpolar lipids as a covari-ate for lipid mass. All data were ln-transformed to meet the as-sumptions of parametric ANOVA. Because in some cases the de-sign had unequal replication, we carried out the analyses usingtype III sum of squares and, in the ANCOVA, we computed theleast-squares estimates of marginal means. Individual repeatabil-ity of locomotor performance was evaluated by using the intra-class correlation coefficient (Sokal and Rohlf 1981; Lessells andBoag 1987). We used Cochran’s C-test to check the assumptionof homogeneity of variances in ANOVA or ANCOVA (allPs>0.05). Prior to performing ANCOVA to compare regressionelevations, we confirmed the assumption of homogeneity of the

Fig. 1 Water temperatures in the source pond. The solid line rep-resents the daily average values and the shaded area indicates dai-ly maximum and minimum temperature. Temperature data wererecorded every 30 min by Onset Optic StowAway data-loggers(range: –5 to +37°C). Vertical and horizontal lines show the dura-tion of the experiment and the experimental temperatures, respec-tively. Pond temperatures during the study period oscillatedaround the intermediate experimental temperature (17°C). NovNovember, Dec December, Jan January, Feb February, MarMarch, Apr April, Jun June, Jul July, Aug August

189

six regression slopes; the significance values obtained in thetests of parallelism were in all cases >0.17.

Results

Jumping ability

The mass of newly metamorphosed individuals wasstrongly influenced by temperature and diet quality (Ta-ble 1). However, food quality and developmental tem-perature did not affect body condition; ln-body mass ad-justed to ln-SVL did not differ among temperatures norbetween diets, and nor was the interaction significant(Table 2). Preliminary analyses revealed similar within-treatment variances for the average jump and the maxi-mum jump, and similar relationships between these vari-ables and mass or SVL (Table 3; Fig. 2). There was nosignificant correlation between initial mass (stage 26)

Table 1 Two-way ANOVA(model I) for temperature anddiet effects on juvenile perfor-mance. All the variables are ln-transformed. MASS Fresh bodymass of juvenile frog, JMAXmaximum jump distance of ju-venile frog, NPLS size of thenonpolar lipid store, LDM leandry mass of juvenile frog

Trait Source df Type III MS F P

MASS Temperature 2 0.9952 24.78 <0.0001Food 1 1.0469 26.07 <0.0001Temperature×food 2 0.1013 2.52 0.0912Error 46 0.0406

JMAX Temperature 2 0.2938 9.82 <0.0003Food 1 0.1403 4.69 0.0355Temperature×food 2 0.0410 1.67 0.1991Error 46 0.0299

NPLS Temperature 2 53.8213 19.65 <0.0001Food 1 10.5860 3.86 0.0553Temperature×food 2 1.4617 0.53 0.5888Error 46 2.7390

LDM Temperature 2 486.1922 15.39 <0.0001Food 1 503.7674 15.94 0.0002Temperature×food 2 65.5493 2.07 0.1371Error 46 31.5912

Table 2 Two-way analysis ofcovariance (ANCOVA) (mod-el I) for temperature and dieteffects on juvenile perfor-mance. All the variables (in-cluding covariates) are ln-trans-formed. SVL Snout-vent lengthof juvenile frog; other abbrevi-ations are given in Table 1

Trait Source df Type III MS F P

MASSa Temperature 2 0.711 1.95 0.1539Food 1 0.027 0.07 0.7875Temperature×food 2 0.202 0.55 0.5782Covariate: SVL 1 168.344 462.30 <0.0001Error 45 0.364

ANCOVA for homogeneity of regression slopes F5,40=1.08 P=0.383

JMAXa Temperature 2 6.116 3.63 0.0345Food 1 3.010 1.79 0.1879Temperature×food 2 0.670 0.40 0.6739Covariate: MASS 1 61.782 33.69 <0.0001Error 45 1.684

ANCOVA for homogeneity of regression slopes F5,40=1.63 P=0.175

NPLS Temperature 2 36.253 12.96 <0.0001Food 1 8.877 3.17 0.0816Temperature×food 2 1.521 0.54 0.5842Covariate: LDM 1 0.119 0.04 0.8374Error 45 2.797

ANCOVA for homogeneity of regression slopes F5,40=1.53 P=0.201a MASS and JMAX meansquares multiplied by 100

Fig. 2 Maximum jump distances of recently metamorphosed Ibe-rian painted frogs as a function of wet mass. Animals reared at17°C with a high protein-content diet (17-HPC) or low protein-content diet (17-LPC) jumped significantly farther than thosereared at 22°C (22-HPC and 22-LPC)

190

Table 3 Pearson product-moment correlations between maximumjump distance and initial mass [Gosner (1960) stage 26], SVL andwet mass for groups of frogs reared in different temperature anddiet conditions. The significance of estimates within each analysis

was assessed using α-levels calculated according to the sequentialBonferroni technique (Rice 1989). The estimates that remainedsignificant after Bonferroni correction are in italics. LPC Lowprotein content, HPC high protein content

Table 4 Repeatability of jump-ing performance for five jumpsperformed by juvenile frogsreared in different temperatureand diet conditions. For abbre-viations, see Table 3

Temperature Diet n Wet mass stage 26 SVL Wet mass

r P r P r P

12°C LPC 7 0.427 0.340 0.711 0.073 0.824 0.022HPC 10 –0.224 0.533 0.780 0.008 0.795 0.006

17°C LPC 11 0.697 0.017 0.083 0.804 0.353 0.282HPC 11 0.309 0.354 0.944 <0.001 0.920 <0.001

22°C LPC 7 –0.477 0.279 0.491 0.263 0.197 0.692HPC 6 0.319 0.537 0.414 0.414 0.516 0.295

and maximum or average jump length (Table 3). There-fore, we present here the results for maximum jump dis-tance; it can be more tightly related to the ability to es-cape predators and hence is a better predictor of the actu-al locomotor performance and survival than averagejump distance (Semlitsch et al. 1999; Watkins 2001).Nevertheless, the conclusions are the same if average,rather than maximum, leap length is used. There weresignificant differences in jumping ability among met-amorphs that had been reared at different temperatures orfed on different diets (Table 1; Fig. 3). On average, frogsraised at 22°C jumped shorter distances than thosereared at 17°C [Tukey honestly significant difference(HSD) test; P<0.0009] or 12°C (P=0.072), but in the lat-ter case these differences were just marginally signifi-cant. HPC frogs tended to jump farther than the LPCones at all temperatures, although the difference was rel-atively clear only for the 22°C groups (Tukey HSD,P=0.121; Fig. 3). There was no significant interactionbetween temperature and diet. ANCOVA with mass ascovariate indicated no effects of diet or diet×temperatureinteraction, but still a significant temperature effect (Ta-ble 2). As expected from the ANOVA results, post hoccomparisons showed that for a given body size, frogsraised at 12°C jumped shorter distances on average thanfrogs raised at 17°C (P=0.013; Fig. 3). Jumping perfor-mance of juvenile frogs was highly repeatable within in-dividuals of each temperature and diet group. Valuesranged between 0.61 and 0.85 in five out of the sixgroups (P<0.0001 for all values; Table 4). Leap lengthswere less repeatable in frogs raised at 22°C and fed onplant-based food (P=0.0016).

Lipid storage

Developmental temperature had a strong effect on non-polar lipid levels (Table 1; Fig. 4). Lipid mass wasgreater for individuals from the 12°C group than formetamorphs reared as larvae at 17°C (Tukey HSD test;P=0.002), but the difference between 17°C and 22°Canimals was nonsignificant (P=0.29). In conjunctionwith the temperature effect, juveniles from the HPCtreatments had higher lipid levels than those from theplant-based diet, but the differences were marginallysignificant (Table 1). There was no interaction betweendiet and temperature. Variation in lipid levels paral-leled, to some extent, the effects of temperature andfood quality on lean dry mass; the lean dry mass of Ibe-rian painted frogs increased with temperature (Table 1;Tukey HSD: 12>17>12°C, P<0.016) and protein con-tent in diet. However, moderate and high developmen-tal temperatures negatively affected lipid storage evenafter correction for differences in lean dry mass (Ta-ble 2), which indicates some size-independent effect ofdevelopmental temperature. Again, post hoc compari-sons revealed significant differences between 12°C and17°C (P=0.0001) but not between 17 and 22°C(P=0.13). Size-independent lipid mass was not affectedby food quality or the diet×temperature interaction (Ta-ble 2).

Group MSwithin df F-ratio P Repeatabilitya

12°C, LPC 0.01180 6, 28 28.54 <0.0001 0.8512°C, HPC 0.01123 9, 40 13.95 <0.0001 0.7217°C, LPC 0.00916 10, 44 8.75 <0.0001 0.6117°C, HPC 0.00967 10, 44 20.67 <0.0001 0.8022°C, LPC 0.02616 6, 28 4.89 0.0016 0.4422°C, HPC 0.02241 5, 24 9.25 <0.0001 0.62

a Repeatabilities were calculated using ANOVA as r=SA2/(S2+SA

2), where S2=MSwithin and SA2=

(MSamong–MSwithin)/n

191

Fig. 3A, B Influence of tem-perature and food quality onjumping performance of newlymetamorphosed Iberian paintedfrogs. A Absolute maximumjump distances; B maximumjump distances following least-squares (LS) adjustment to bodymass (means±SE)

Fig. 4 Energy reserves(amount of nonpolar lipids;means±SE) of newly metamor-phosed Iberian painted frogsreared at 12, 17 or 22°C andfed on HPC or LPC diets. Forabbreviations, see Fig. 2

192

Discussion

Conditions for larval growth, especially temperature, af-fected postmetamorphic performance of D. galganoi,both directly and through plasticity in body size. Para-doxically, size-dependent and size-independent effects oftemperature on jumping ability worked in opposition toone another. In amphibians, body size at metamorphosisaffects adult reproductive success and survival (Bervenand Gill 1983; Smith 1987) and thus variation in bodysize may contribute to variation in fitness. However,there is very little information on the mechanistic link-age between size at metamorphosis and survival to adult-hood or reproductive success. The consequences of sizeat metamorphosis for subsequent growth and survivalcan be mediated at least in part by locomotor perfor-mance and storage of energy reserves. Only a few stud-ies have examined the effects of factors operating duringthe larval period on measures of performance of recentlymetamorphosed anurans (John-Alder and Morin 1990;Goater et al. 1993; Semlitsch et al. 1999; Tejedo et al.2000a; Tejedo et al. 2000b) and to our knowledge noneof these evaluated the effects of developmental tempera-ture or larval diet on postmetamorphic performance. Al-though our results show, as others have before (e.g. Emerson 1978; Goater et al. 1993; Miller et al. 1993),that body size is an important determinant of metamorphjumping capacity, they also reveal size-independent ef-fects of temperature on locomotor performance. Here,LPC and high temperature produced small metamorphsthat jumped shorter distances than large metamorphsfrom the high protein or moderate temperature treat-ments. Therefore, conditions in the larval environmentthat influence size at metamorphosis (e.g. larval density,parasitic infection, starvation, diet quality, temperature)are expected to influence juvenile jumping ability. How-ever, D. galganoi reared at 17°C jumped farther thanthose reared at 12°C despite the latter being bigger. Infact, cold-reared individuals had inferior size-adjustedjumping performance compared to the other groups.Since these differences were not accounted for by varia-tion in body condition and larval periods were signifi-cantly longer at 12°C than at 17°C (Álvarez and Nicieza,unpublished manuscript), the mechanism by which de-velopmental temperature could have influenced locomo-tor performance is most likely physiological. In ecto-therms, some traits important in locomotion, such as thenumber of body segments and muscle structure, arehighly responsive to the temperature experienced earlyin development (e.g. Lindsey 1966; Lindsey and Harrington 1972; Stickland et al. 1988; Shine and Harlow 1993; Partridge et al. 1994; Crill et al. 1996; Johnston and McLay 1997). Temperatures close to theupper and lower thermal tolerance limits generate a highprobability for phenodeviants as a result of developmen-tal stress (Smith-Gill 1983; Møller and Swaddle 1997).In our case, the poorest performance in terms of relativejump length was associated with the low temperature,where the abnormal prolongation of the larval period

(>5 times the larval period at 22°C and >3 times that at17°C) could have acted as a source of stress. There aretwo ways by which temperature is expected to affectjumping ability. First, temperature can affect muscle cel-lularity and ultrastructure. Although this possibility hasnot been adequately explored in anurans, fish studiessuggest that both the post-larval number of muscle fibresand their cross-sectional area is determined at least inpart by the thermal experience of the embryo (Johnstonet al. 1996). Second, the effect of temperature on loco-motor performance can be mediated through changes inthe morphology and relative length of hind limbs; frogswith lower ratios of hind-limb length to SVL jump short-er distances than frogs with higher ratios (Emerson1978). This is not surprising, since the hormones thatregulate metamorphosis also control limb developmentand they are very sensitive to temperature (Emerson1986). Moreover, in frogs from density experiments theearly metamorphosing individuals have higher ratios ofhind-limb to SVL than do the later metamorphosing individuals, so that time to metamorphosis and relative hind-limb size show a negative phenotypic correlation (Emerson 1986; but see Blouin and Loeb 1991). Blouinand Brown (2000) demonstrated that the cold-inducedextension of the larval period in Rana cascadae resultsin shorter relative tibio-fibula lengths at metamorphosis.Therefore, the lower jumping performance of cold-reared individuals can result from an alteration in met-amorph morphology. Regardless of whether size-inde-pendent temperature effects on locomotor performanceare mediated by changes in muscle structure or biome-chanical traits, it seems that larval development andgrowth history affect traits that subsequently influencegrowth and survival rates in the terrestrial juvenile phase(see also Goater et al. 1993; Tejedo et al. 2000b). Like-wise, our results suggest that size-independent maycounterbalance size-dependent effects. Here, a crucialpoint is to know how long the phenotypic alteration un-derlying the size-independent effect can persist; if thephenotypic modification is not compensated soon aftermetamorphosis it might offset the advantages of largesize at metamorphosis or magnify the drawback of sub-optimal sizes.

This study focussed on the postmetamorphic effectsof factors influencing larval (as opposed to embryonic)development. Because tadpoles were exposed to differ-ent temperatures and diets only after the achievement ofstage 25, we cannot discard additional effects mediatedthrough the alteration of embryonic (pre-hatching) devel-opment. In turn, this does allow us to identify a narrowersegment of development during which the alteration ofenvironmental conditions can have important conse-quences for postmetamorphic performance. On the otherhand, it is unlikely that growth, development and alloca-tion strategies of tadpoles could have been differentiallyinfluenced by the temperature and food conditions expe-rienced before the attainment of stage 25. Firstly, all theexperimental animals were collected over a very smallarea (0.2 m2), and stage-25 larvae have reduced mobility,

193

so they probably experienced very similar conditions offood and temperature. Moreover, mouthparts of anuranlarvae begin to develop in stage 23 and are essentiallycomplete by stage 25 (Duellman and Trueb 1986); stag-es 24–25 mark the transition from a larva sustained byyolk reserves to an exogenously feeding tadpole. Inagreement with this, there were no significant differ-ences in body mass at the start of the experiment.

Survival during the postmetamorphic period can bestrongly dependent on the size of non-polar lipid store,which in turn can vary inversely with developmentalrates (Pfennig 1992). Strikingly, the pattern of variationin lipid levels in D. galganoi metamorphs resembles thetrends observed for jumping ability, but in the oppositedirection. Low-temperature individuals stored largeramounts of fat than those reared at 17°C or 22°C. How-

ever, lipid accumulation and jumping performance wereapparently set independently (i.e. there is no causal linkbetween them) because temperature did not affect thebody condition of metamorphs. Here, it should be notedthat larval thermal conditions affected lipid storage buthad no apparent effect on juvenile body condition. Onepossible explanation is that temperature or food condi-tion could have affected water retention or body mor-phology in a way that obscured the differences in lipidstorage; this may occur if, for example, low temperatureshad resulted in more elongated trunks and heads. How-ever, very little information is available in this respect,and the existing data suggest no effect of developmentaltemperature on head morphology (Blouin and Brown2000). A more plausible explanation is that, although thedifferences in lipid storage caused by differences in tem-perature are patent, the total amount of lipids in recentlymetamorphosed anurans is very small. Therefore, evenmoderate individual variation in water content and shapecould easily obscure the effects on lipid content.

At the time of metamorphosis, there were two funda-mental differences between individuals maintained at12°C versus 17°C or 22°C: their larger body size andtheir expanded larval period. Nevertheless, the differ-ences in mass between frogs reared at 12°C and 17°Cseem to be of minor importance compared to the differ-ences between these and 22°C individuals (Fig. 4). Incontrast, the differences in larval period for these groupsfit well with the hypothesis that lipid accumulation isconstrained by development rates. In spadefoot toads(Scaphiopus multiplicatus), carnivore larvae developmore rapidly than omnivores at the expense of accumu-lating significant lipid reserves, and thus suffer higherpostmetamorphic mortality despite their larger size atmetamorphosis (Pfennig 1992). In D. galganoi, the car-nivorous diet did not influence the larval period, and car-nivores and herbivores had similar amounts of lipids af-ter correction for differences in lean mass. In contrast,developmental temperature had a strong, negative effect

Fig. 5 Schematic representation of the chains of effects of envi-ronmental factors on postmetamorphic traits (third level) mediatedthrough age and size at transformation (metamorphic traits; sec-ond level) and growth and development rates (larval traits; firstlevel). The hypothesized bi-directional link between food qualityand temperature refers to the observed interaction effects of thesefactors on metamorphic traits; the net influence of protein or ener-gy content depends on the developmental temperature experiencedby larvae, just as optimum temperature will depend on the qualityof the available food. A single factor can have opposite effects onpostmetamorphic traits. For example, temperature has a “direct”(size-independent) positive effect on jumping ability; matched forsize, fast-developing individuals jumped farther than slow-devel-oping ones. Therefore, we hypothesize that the negative influenceof temperature on jumping performance can be a cost of very slowdifferentiation at unusually low temperatures (dashed link). In ad-dition, temperature has a negative effect on size at metamorphosisbecause differentiation rate changes with temperature faster thandoes growth rate, so producing a negative effect on jumping abili-ty. Because rapid development has a cost in terms of lipid accumu-lation, a trade-off can arise between energy reserves and locomo-tor performance. Tadpoles and breeding adults could have somecontrol on this trade-off by spatial and temporal selection of watertemperatures. Food quality affects postmetamorphic performancevia growth rate and its influence on body size

on size-adjusted lipid stores. This provides experimentalconfirmation that differences in lipid accumulation canbe primarily modulated by variation in the length of thelarval period.

In conclusion, this study illustrates the potential im-portance of events in the larval aquatic phase in affectinggrowth and survival in the terrestrial stage. We revealeda complex chain of induced modifications that may re-sult in conflicting effects (Fig. 5). This can lead to multi-ple trade-offs between traits affected simultaneously orsequentially by an environmental factor. For example,the same change in larval temperature may be advanta-geous in terms of locomotor performance but detrimentalbecause of reduced accumulation of energy reserves.Likewise, the advantages of short permanence in theaquatic habitat or commencement of the terrestrial phaseearly in the growing season can be offset by relativelysmall sizes or reduced energy stores. The relative impor-tance of each of these changes in metamorph ability willpresumably be dependent on key characteristics of theaquatic and terrestrial environments, like pond duration,risk of predation, or food availability (Pfennig 1992).Tadpoles of several anuran species select microhabitatsand aggregate in apparent response to temperature. Froman adaptive perspective, if larvae prefer temperaturesthat maximize growth and survival to adulthood, ap-proaches considering only the performance in the larvalphase may undervalue adaptive behaviours. Developingthe relationships between population-specific thermalpreferences and environmental variables in the terrestrialand aquatic habitats will permit a more mechanistic un-derstanding of the impact that developmental modifica-tions exert on fitness.

Acknowledgements We thank N. B. Metcalfe and an anonymousreviewer for useful comments on the manuscript. This researchwas supported by DGES research grant PB96-0861 to M. Tejedoand A. G. Nicieza.

References

Arnold SJ, Wassersug RJ (1978) Differential predation on meta-morphic anurans by garter snakes (Thamnophis): social behav-ior as a possible defense. Ecology 59:1014–1022

Berven KA, Gill DE (1983) Interpreting geographic variation inlife-history traits. Am Zool 23:85–97

Blouin MS, Brown ST (2000) Effects of temperature-inducedvariation in anuran larval growth rate on head width and leglength at metamorphosis. Oecologia 125:358–361

Blouin MS, Loeb MLG (1991) Effects of environmentally-in-duced development-rate variation on head and limb morpholo-gy in the green tree frog Hyla cinerea. Am Nat 138:717–728

Brockelman WY (1969) An analysis of density effects and preda-tion in Bufo americanus tadpoles. Ecology 50:632–644

Calef GW (1973) Natural mortality of tadpoles in a population ofRana aurora. Ecology 54:741–758

Crill WD, Huey RB, Gilchrist GW (1996) Within- and between-generation effects of temperature on the morphology andphysiology of Drosophila melanogaster. Evolution 50:1205–1218

Dobush GR, Ankney CD, Krementz DG (1985) The effect of ap-paratus, extraction time, and solvent type on lipid extractionsof snow geese. Can J Zool 63:1917–1920

Duellman WE, Trueb L (1986) Biology of amphibians. McGraw-Hill, New York

Ebenman B (1992) Evolution in organisms that change their nich-es during the life cycle. Am Nat 139:990–1021

Emerson SB (1978) Allometry and jumping in frogs: helping thetwain to meet. Evolution 32:551–564

Emerson SB (1986) Heterochrony and frogs: the relationship of alife history trait to morphological form. Am Nat 127:167–183

Goater CP, Semlitsch RD, Bernasconi MV (1993) Effects of bodysize and and parasite infection on the locomotory performanceof juvenile toads, Bufo bufo. Oikos 66:129–136

Gosner KL (1960) A simplified table for staging anuran embryosand larvae with notes on identification. Herpetologica 16:183–190

Heulett ST, Weeks SC, Meffe GK (1995) Lipid dynamics andgrowth relative to resource level in juvenile eastern mosquito-fish (Gambusia holbrooki: Poeciliidae). Copeia 1995:97–104

John-Alder HB, Morin JP (1990) Effects of larval density onjumping ability and stamina in newly metamorphosed Bufowoodhousii fowleri. Copeia 1990:856–860

Johnston IA, McLay HA (1997) Temperature and family effectson muscle cellularity and first feeding in Atlantic salmon (Salmo salar L.). Can J Zool 75:64–74

Johnston IA, Vieira VLA, Hill J (1996) Temperature and ontogenyin ectotherms: muscle phenotype in fish. In: Johnston IA, Benett AF (eds) Animals and temperature. Phenotypic andevolutionary adaptation. Cambridge University Press, Cam-bridge, Mass. pp 153–181

Johnston IA, Cole NJ, Vieira VLA, Davidson I (1997) Tempera-ture and developmental plasticity of muscle phenotype in her-ring larvae. J Exp Biol 200:849–868

Kupferberg SJ (1997) The role of larval diet in anuran metamor-phosis. Am Zool 37:146–159

Lessells CM, Boag PT (1987) Unrepeatable repeatabilities: a com-mon mistake. Auk 104:116–121

Lindsey CC (1966) Temperature-controlled meristic variation inthe salamander Ambystoma gracile. Nature 209:1152

Lindsey CC, Harrington RW (1972) Extreme vertebral variationinduced by temperature in a homozygous clone of the self-fer-tilizing cyprinodontid fish Rivulus marmoratus. Can J Zool50:733–744

Lutz GJ, Rome LC (1994) Built for jumping: the design of thefrog muscular system. Science 263:370–372

Miller K, Monteforte PB, Landis LF (1993) Scaling of locomotorperformance and enzyme activity in the leopard frog, Ranapipiens. Herpetologica 49:383–392

Møller AP, Swaddle JP (1997) Asymmetry, developmental stabili-ty and evolution. Oxford University Press, Oxford

Nathan JM, James VG (1972) The role of Protozoa in the nutritionof tadpoles. Copeia 1972:669–679

Newman RA (1987) Effects of density and predation on Scaphi-opus couchii tadpoles in desert ponds. Oecologia 71:301–307

Newman RA (1989) Developmental plasticity of Scaphiopus cou-chii tadpoles in an unpredictable environment. Ecology 70:1775–1787

Pandian TJ, Marian MP (1985) Predicting anuran metamorphosisand energetics. Physiol Zool 58:538–552

Partridge L, Barrie B, Fowler K, French V (1994) Evolution anddevelopment of body size and cell size in Drosophila melano-gaster in response to temperature. Evolution 48:1269–1276

Pfennig DW (1992) Polyphenism in spadefoot toad tadpoles as alocally adjusted evolutionary stable strategy. Evolution 46:1408–1420

Rice WR (1989) Analyzing tables of statistical tests. Evolution43:223–225

Scott DE (1994) The effect of larval density on adult demographictraits in Ambystoma opacum. Ecology 75:1383–1396

Semlitsch RD, Pickle J, Parrish MJ, Sage RD (1999) Jumping per-formance and short-term repeatability of newly metamor-phosed hybrid and parental leopard frogs (Rana sphenoceph-ala and Rana blairi). Can J Zool 77:748–754

194

Swain DP, Lindsey CC (1986) Meristic variation in a clone of thecyprinodontid fish Rivulus marmoratus related to temperaturehistory. Can J Zool 64:1444–1455

Tåning ÅV (1952) Experimental study of meristic characters infishes. Biol Rev Camb Philos Soc 27:169–173

Tejedo M, Semlitsch RD, Hotz H (2000a) Covariation of morphol-ogy and jumping performance in newly metamorphosed waterfrogs: effects of larval growth history. Copeia 2000:448–458

Tejedo M, Semlitsch RD, Hotz H (2000b) Differential morpholo-gy and jumping performance of newly metamorphosed frogsof the hybridogentic Rana esculenta complex. J Herpetol34:201–210

Tracy CR, Christian KA, O’Connor MP, Tracy CR (1993) Behav-ioral thermoregulation by Bufo americanus: the importance ofthe hydric environment. Herpetologica 49:375–382

Walton M (1988) Relationships among metabolic, locomotory, andfield measures of organismal performance in the Fowler’s toad(Bufo woodhousei fowleri). Physiol Zool 61:107–118

Wassersug RJ, Sperry DG (1977) The relationship of locomotionto differential predation on Pseudacris triseriata (Anura: Hyl-idae). Ecology 58:830–839

Watkins TB (2001) A quantitative genetic test of adaptive decou-pling across metamorphosis for locomotor and life-historytraits in the Pacific tree frog, Hyla regilla. Evolution 55:1668–1677

Werner EE (1986) Amphibian metamorphosis: growth rate, preda-tion risk, and the optimal size at transformation. Am Nat128:319–341.

Wilbur HM (1980) Complex life cycles. Annu Rev Ecol Syst 11:67–93

195

Shaffer HB, Austin CC, Huey RB (1991) The consequences ofmetamorphosis on salamander (Ambystoma) locomotor perfor-mance. Physiol Zool 64:212–231

Shine R, Harlow P (1993) Maternal thermoregulation influencesoffspring viability in a viviparous lizard. Oecologia 96:122–127

Smith DC (1983) Factors controlling tadpole populations of thechorus frog (Pseudacris triseriata) on Isle Royale, Michigan.Ecology 64:501–510

Smith DC (1987) Adult recruitment in chorus frogs: effects of sizeand date at metamorphosis. Ecology 68:344–350

Smith-Gill SJ (1983) Developmental plasticity: developmentalconversion versus phenotypic modulation. Am Zool 23:47–55

Smith-Gill SJ, Berven KA (1979) Predicting amphibian metamor-phosis. Am Nat 113:563–585

Sokal, RR, Rohlf FJ (1981) Biometry. Freeman, San Francisco,Calif.

Steinwascher K, Travis J (1983) Influence of food quality andquantity on early growth of two anurans. Copeia 1983:238–242

Stickland NC, White RN, Mescall PE, Crook AR, Thorpe JE(1988) The effect of temperature on myogenesis in embryonicdevelopment of the Atlantic salmon (Salmo salar L.). AnatEmbryol 178:253–257

Swain DP (1992a) The functional basis of natural selection forvertebral traits of larvae in the stickleback Gasterosteus acu-leatus. Evolution 46:987-997

Swain DP (1992b) Selective predation for vertebral phenotype inGasterosteus aculeatus: reversal in the direction of selection atdifferent larval sizes. Evolution 46:998-1013

Swain DP, Lindsey CC (1984) Selective predation for vertebralnumber of young sticklebacks, Gasterosteus aculeatus. Can JFish Aquat Sci 41:1231–1233