Metabolic scaling associated with unusual size changes during larval development of the frog, Pseudis paradoxus

Early studies on the diffusion and perfusion limitations of cutaneous breathing indicate that routine rates of aerobic metabolism in amphibians might be limited by the surface area available for O₂ transfer(Piiper and Scheid, 1977; Ultsch, 1976). Amphibians spend a large part of their daily cycle at rest, and many of their routine behaviours seem to use only a fraction of their aerobic capacity(Gatten, Jr et al., 1992). The scaling of resting metabolic rates(Ṁ_O2) with body mass M_b in intra-specific comparisons with adult individuals, described by the allometric equation Ṁ_O2=aM_b^b,has not disclosed any particular tendency towards a 2/3 exponent, and scaling coefficient values (b) on average express an intermediate proportion in the surface area:body mass relationship(Withers, 1992). However,surface limitation may be significant and may restrain metabolic capacity to grow and sustain activity, particularly in those species that rely heavily on cutaneous gas exchange and during larval life stages.

The issue is of interest owing to the recent elucidation of processes contributing to metabolic rate in organisms, and of how body size exerts its influence on energy expenditure at the whole-body level(Hochachka et al., 2003; Darveau et al., 2002; Hulbert and Else, 2000; Rolfe and Brown, 1997). It has become clear that, as organisms increase in size, they adjust their structure and function at all levels of organization, from the external surface for O₂ transfer to the composition of cell membranes and protein function, each with their own characteristic scaling coefficient and control contribution, which in turn, depend on metabolic state. These ideas have been largely developed from inter-specific comparisons of adult organisms, under low and high activity levels; in the ontogenetic context, energy utilization for growth and organogenesis may have a great influence on overall metabolic scaling according to the developmental stage. In the few mammals and fishes examined, metabolic rates change in direct proportion to mass during the early stages (the period of greatest growth), bringing the mass exponent to close to 1.0; thereafter, rates tend to follow the surface rule with the two regressions intersecting after a given proportion of adult body mass is reached (reviewed by Hulbert and Else,2000; Wieser,1995; Wieser,1984).

Interestingly, unlike the situation in mammal and fish species, there appears to be no clear shift in the scaling pattern during the transition from larval to adult life stages in anuran amphibians, the exponent varying from 0.65 to 0.88 (Gatten, Jr et al.,1992; Feder,1982), which suggests possible surface limitations to the increase in cutaneous O₂ transfer in direct proportion to body mass during early development. In addition, experimental data with anuran larvae suggest that the larger the larvae, the greater the contribution of air breathing to the responses to severe aquatic hypoxia. The requirement of air breathing to meet metabolic demands in developing tadpoles has long been controversial;while cutaneous respiration provides 60% or more of overall oxygen uptake in anuran larvae, air breathing does not seem to be essential for growth and activity in their natural settings(Burggren and Just, 1992; Feder, 1984). However, since post-metamorphic growth is usually responsible for 80% or more of body size in anuran amphibians (Werner,1986), it is plausible that growth processes may not influence energy cost at the whole-body level during the larval phase in the species examined, which might explain the apparent discrepancy in the scaling patterns. In the present study, this hypothesis was examined in the frog Pseudis paradoxus. The development of P. paradoxus is unusual in several respects, such as the large larval size and small gain in body mass after metamorphosis, leading to a large contrast in the maximum size of the tadpole compared to the adult frog(Emerson, 1988). Giantism in P. paradoxus larvae may thus represent the anticipation of post-metamorphic growth during the early stages of the life cycle, the adaptive value of which is unknown.

P. paradoxus frogs are exclusively neotropical; in southern Brazil, newly hatched larvae are most abundant in late spring and become scarce by mid-summer, most tadpoles having metamorphosed by the early fall(S.C.R.d.S., unpublished observations). The large size at metamorphosis is attained during a relatively short period, implying use of a large fraction of the energy derived from aerobic sources to support growth processes in larval P. paradoxus. Further, O₂ uptake might be severely affected in the largest tadpoles under hypoxic conditions, as is usual in the natural habitat. The present study is concerned with the question of how O₂ uptake may be reconciled to growth and developmental changes in P. paradoxus tadpoles. We examined the relationship between body mass and aquatic and aerial O₂ uptake in the growth and degrowth phases of the larval cycle, and the effects of aquatic hypoxia on the scaling patterns. In addition, we analysed correlative shifts in the levels of body glycogen and lactate as an indication of supplementary energy production from anaerobic sources. Our findings generally indicate that the scaling of metabolism in P. paradoxus may express the consequences of size on energy demand at the tissue level, suggesting that air breathing must subsidise energy expenditure in the largest individuals, and sustain the increase in metabolism during late metamorphosis.

Tadpoles and froglets of Pseudis paradoxus (Linné 1758)were caught by sieve from the ponds at a fish farm located in Ribeirão Preto, south-eastern Brazil, during the summer. In the laboratory, the animals were maintained in 150 l tanks containing aerated water at 24±1°C and vegetation that served as refuge and support for the late metamorphic stage larvae and froglets. The larvae were fed twice a day with commercial food (Reptomin®) and spinach ad libitum. The intensification of metamorphosis caused a cessation of food intake, confirming observations made on the digestive tube of newly captured animals at late metamorphosis. The minimum and maximum masses at capture were 0.1 g and 72 g, respectively, some mass loss following maintenance in the laboratory. The measurements were made on animals kept in the laboratory for less than 20 days, and deprived of food for 24 h prior to the experiments.

Developmental stage was classified according to Cais after Gosner(Cais, 1982; Gosner, 1960). First stage larvae were very vulnerable to handling and transportation and were not included in the hypoxia experiments; partial data for larval stages 29–32 (body mass=3.7±1.2 g, mean ± s.d.) in normoxic water are given in the Results. The individual data obtained for larvae from the subsequent stages were combined into five developmental groups: the first consisted of stages 33–37 (8.1±4.5 g), all larvae showing inflated lungs; the second group consisted of stages 38–42(17.0±7.3 g), the larvae showing well vascularized lungs, and stage 42 being marked by the emergence of the forelimbs; the fourth included stages 43–44 (14.8±8.2 g), and the fifth stage 45 alone (10.0±2.6 g); both these groups exhibit the most notable anatomical changes that constitute the shape of the adult frog while maintaining a tail of considerable proportions; the sixth group included stage 46 (7.1±2.4 g), corresponding to froglets up to 20 days of complete metamorphosis. According to our observations, surfacing behaviour begins at low frequencies during the early larval stages, and more regularly in tadpoles at stages 38–42; in the latest stages, the tadpoles remain for long periods at the air–water interface, while froglets alternate periods of partial emersion with prolonged submersion.

Aquatic and aerial oxygen uptake, Ṁw_O2 and Ṁa_O2,respectively, were measured by closed system respirometry at 25±1°C(room temperature) on single larvae, each animal being used only once. The metabolic chambers consisted of an acrylic tube with a 10 mm thick wall, and of variable length and diameter. Previous measurements showed that P. paradoxus tadpoles become restless when held with the body lying flat inside a horizontal chamber. The tube was then vertically oriented with the bottom end closed, and a funnel-like lid that fit over the top end over a layer of compressible rubber. For the experiments, the chamber was screwed shut and filled with water leaving a volume of air close to the funnel stem,and the volumes were adjusted according to the size and stage of the tadpole. During measurements, the water was exchanged via outlet and inlet valves, and the samples were withdrawn through a small aperture located at a medium height in the water column, sealed by a silicone septum (Alltech Associates, Inc.,Deerfield, Illinois, USA). The air phase communicated with ambient air via a valve attached to the funnel stem, and air samples were withdrawn via a silicone septum fitted into the valve aperture.

The experiments were conducted on consecutive runs from 08:00 h to 19:00 h. Preliminary measurements were made in nearly constant, normoxic conditions, to check for any sizeable variability related to experimental procedures. The rates were then measured under acute exposure to incremental levels of aquatic hypoxia while the animals had free access to the air volume. The system was opened at 60–90 min intervals and the water carefully exchanged by the inflow of clean, sterilized water at near-neutral pH from a reservoir. The water in the reservoir was constantly mixed by a flow of air or N₂such that the resulting water P_O2 was 146, 114,88, 60, 40 mmHg, the levels representing the global mean of the initial P_O2 values in each run. Further, the air at the top of the chamber was refreshed and O₂ content set to normoxic levels for each subsequent experimental run. The animals were acclimated to the chamber for at least 12 h prior to measurements with a constant flow of aerated water, and the valve to the air phase open to avoid hypoxia. After closing the system, the water P_O2 was allowed to decline with the animal breathing to a limit that nearly matched the initial P_O2 value in the next run, and O₂ content in the air phase was allowed to decline to a limit of 60% below normoxic in each run.

Ṁw_O2 and Ṁa_O2 were measured in triplicate samples of 1 ml and 50 μl, respectively, taken with gas-tight syringes. The air in the gas phase was gently mixed with the syringe before sampling, and the water was mixed for a few seconds with a small,magnetic stirring bar at slow speed, and sampled. O₂ concentration in the air samples were measured using the zirconium sensor of an Oxycon device (Cameron Instrument Co., Port Aransas, Texas, USA), and P_O2 in the water samples was measured using a Clark-type electrode held inside a thermo-stable chamber at 25°C, the signal being converted in a gas analyser (Cameron Instrument Co.). The equipment was calibrated in the early morning and checked for stability during the experiments. During measurements, the animals were shielded by an opaque screen to minimize visual disturbance; the stage 33–42 tadpoles remained mostly quiescent, with the snout upwards and the tail hanging downwards into the water column, while the last stage tadpoles and froglets alternated periods near the water surface with variable periods of submersion to the bottom. In all cases, energy expenditure during the measurements was assumed to be an approximation of resting levels, with ΔO₂ values varying according to the tadpole size and degree of aquatic hypoxia. Micromolar O₂ concentrations were calculated from the O₂tension and the coefficients for distilled water at 25°C, and the total O₂ consumption rates(Ṁ_O2) were calculated from the Ṁw_O2 and Ṁa_O2.

The same protocol was used in control experiments with no animals in the chambers, allowing adjustment for diffusive changes according to chamber size and degree of aquatic hypoxia. Oxygen changes in the air and in the water phase were measured at 60–90 min intervals under incremental levels of aquatic hypoxia as above; the values obtained were then subtracted from, or added to, the experimental O₂ consumption rates accordingly,averaging 0.9–6.1% of the Ṁa_O2 and 0.1–1.1% of the Ṁw_O2 values.

Total glycogen and lactate contents were measured in larvae at distinct developmental stages in normoxic water (mean P_O2=149 mmHg) or after acute, 3 h exposure to hypoxic water (60 mmHg), during which the animals had free access to normoxic air in the upper phase. Groups of animals were maintained in 60 l tanks containing clean, sterilized water at near-neutral pH and a support made of inert material that allowed access by late metamorphic stage individuals to the surface without increased effort. The animals were killed by pithing and quickly immersed in liquid N₂, then stored at –85°C until analysis.

The frozen larvae were powdered in liquid N₂ and homogenised with ice-cold, 0.6 mol l^–1 perchloric acid (PCA) using a blender. After complete deproteinization, glycogen and l-lactate concentrations were estimated according to standard enzymatic procedures(Bergmeyer, 1984) by following the oxidation rate of NAD⁺ in a spectrophotometer Spectra Max 250(Molecular Devices, Sunnyvale, CA, USA) at 340 nm and 25°C. The final result was expressed as μmol glycosyl units or l-lactate per unit body mass.

Visual inspection suggested the effects of P_O2 on Ṁw_O2 and Ṁa_O2 per unit mass to be non-linear in some of the experimental groups. Without a theoretical model, however, the outcome of the analysis using a different function would be equally arbitrary, and hence, for the purpose of this study,the correlation coefficient `r' was used as a reliable measure of the degree of association between the variables. The data were log-transformed for calculation of the coefficient given the indication of deviation from bivariate normality. The percentage of variance of the whole-body Ṁ_O2, Ṁw_O2, Ṁa_O2, and metabolites attributable to body mass and to developmental stage were estimated by analysis of covariance (ANCOVA), using the data obtained in normoxic and hypoxic water. The relationships between whole-body Ṁ_O2, Ṁw_O2 or Ṁa_O2 and body mass were analysed at two selected water P_O2levels (146 and 60 mmHg) on log-transformed data, using the least-squares,linear regression method. The growth and degrowth phases of the larval period were considered as criteria of biological significance in the calculation of two regression lines and equations, one consisting of the developmental stages 33–42, and the other of stages 43–46, and the regression coefficient r² was used as a measure of functional dependence of O₂ uptake rates with respect to body mass. The effect of hypoxic water on glycogen and lactate per unit mass was tested against normoxic water and as a function of developmental stage by two-way analysis of variance (ANOVA). The scaling relationships of the metabolite concentrations were analysed using the least-squares, linear regression method on log-transformed, whole-body data as above. The analyses were performed according to Zar (Zar, 1999),using SigmaStat (Jandel Scientific Co., San Rafael, CA, USA) and MiniTab(MiniTab Inc., State College, PA, USA) statistical software. The test results and probability of error are given in the text or in the tabulated results,differences being considered significant at P≤0.05.

There were no significant changes in mass-specific O₂consumption rates(Ṁ_O2/M_b)of tadpoles and froglets measured in consecutive experimental runs under normoxic conditions. It was thus assumed that the experimental procedure and its total duration did not contribute significant variation to the data. In normoxic water, Ṁ_O2/M_bmeasured 2.1–2.5 μmol g^–1 h^–1 in stage 29–42 larvae, increasing to 3.9 μmol g^–1h^–1 in stages 43–44 larvae, and peaking at 5.7 μmol g^–1 h^–1 in stage 45 larvae(Fig. 1). Ṁ_O2/M_bvalues decreased sharply on metamorphosis, and those of the froglets were similar to the early larval stages. Acute exposure to water of reduced P_O2 caused no significant change in Ṁ_O2/M_bwithin the P_O2 range analysed, most variability being attributable to developmental stage (F=48.2, P<0.0001).

Under normoxic conditions, the stage 33–37 tadpoles acquire 96% of O₂ consumed from the water, with a minor contribution from aerial exchange (Table 1; Fig. 2). This value is unchanged in the subsequent stages until the emergence of the forelimbs in stage 42, when gill respiration becomes impaired. Thereafter, the contribution by aerial uptake increases sharply, reaching 73% in stage 45 larvae and in froglets. Partitioning between aquatic and aerial O₂ uptake was altered on exposure to aquatic hypoxia in all groups of stages examined(Table 1; Fig. 2). In stage 33–37 larvae Ṁw_O2/M_bdecreased with P_O2 decrease (r=0.73; P<0.0001), this effect being accompanied by an increase in Ṁa_O2/M_b(r=0.80; P<0.0001) that accounts for 74% of O₂consumed under the most severe hypoxia. Similarly, in larval stage 37–42, there was a positive correlation between Ṁw_O2/M_band the degree of aquatic hypoxia (r=0.68; P<0.0001), Ṁa_O2/M_bshowing an inverse correlation (r=0.61; P<0.0001) and accounting for 77% of O₂ consumed at the lowest P_O2 level. The Ṁw_O2/M_band Ṁa_O2/M_bcorrelation lines cross over at approximately 60 mmHg, when the contributions of aquatic and aerial uptakes are nearly equivalent. Significant decreases in Ṁw_O2/M_bwith decrease in water P_O2 were also observed in stage 43–44 larvae (r=0.64; P<0.0001) and in stage 45 larvae (r=0.40; P<0.0172), the rates reaching less than 10% of O₂ consumed under the most severe hypoxia, with an inverse correlation for Ṁa_O2/M_bin both groups (r=0.43 and 0.44; P<0.0024 and 0.0082). In the froglets, there was no significant change in Ṁw_O2/M_bas a function of decreasing O₂ level in the water even though a small increase in lung ventilation was detected (r=0.37; P=0.0234).

In normoxic water (P_O2=146 mmHg), body mass accounts for 87% of the overall variability in Ṁ_O2 of P. paradoxus larvae and froglets (ANCOVA; F=35.8; P=0.0001); significantly, developmental changes in the gas exchange organs account for 14% (F=4.4; P=0.001). This reflects the strong dependence of Ṁw_O2 on body mass M_b (F=38.0; P=0.0001), with a lesser effect of developmental changes (F=3.4; P=0.004). The Ṁa_O2 is predominantly influenced by developmental stage (F=90.8; P=0.0001), and varies significantly as a function of larval size(F=14.6; P=0.0001) (Table 2; Fig. 3).

Separate analysis of the growth and degrowth phases revealed that the slope for the relationship between Ṁ_O2 and body mass is b=1.02 and 0.97, respectively(Table 2; Fig. 3), implying that O₂ uptake first increases proportionally to body mass increase(r²=0.77; P<0.0001), then decreases in proportion to mass reduction during larval development(r²=0.61; P<0.0001). Ṁw_O2 similarly increases in strict proportion to the increase in body mass during growth stages, with b=1.02 (r²=0.76; P<0.0001), and although representing only a small fraction of the total O₂ uptake, Ṁa_O2 increases out of proportion to body mass increase, with b=1.13(r²=0.67; P<0.0001). During degrowth stages,while the decrease in the Ṁw_O2 with body mass decrease is disproportionate, with b=1.18(r²=0.75; P=0.0001), Ṁa_O2 increases with b=0.77 (r²=0.45; P=0.0002),implying an increasingly higher capacity for O₂ uptake by the lungs during the late metamorphic stages.

In hypoxic water (P_O2=60 mmHg), body mass similarly explains most of the variability (87%) in the Ṁ_O2 of tadpoles and froglets (ANCOVA; F=37.3; P<0.0001). However, the influence of body mass on the Ṁw_O2 is less pronounced (F=4.38; P=0.045), while the mass dependence of Ṁa_O2 is greatly increased (F=17.7; P=0.0001), with 65% of the variability in Ṁa_O2 attributable to body mass compared to 14% in normoxic water. According to the regression analysis, the slope for Ṁ_O2 in hypoxic water declined to b=0.83 over the growth stages(r²=0.69; P<0.0001), implying a variable effect of hypoxia on O₂ uptake as a function of larval size, larger larvae being more affected than smaller ones(Table 2; Fig. 3). Accordingly, the depression in aquatic O₂ uptake is notably large in the larger tadpoles, causing the slope for Ṁw_O2 with body mass to approach zero (b=0.11), and associated with an increase in Ṁa_O2 out of proportion to body mass, with b=1.44 (r²=0.57; P=0.0023). During the degrowth stages, there is a change in the slope for the relationship between Ṁw_O2 and Ṁa_O2 with body mass to b=0.75 (r²=0.127; P=0.087) and 0.99 (r²=0.61; P<0.0001), respectively,suggesting that the effects of aquatic hypoxia are more severe in the large tadpoles than in froglets.

In normoxia (P_O2=149 mmHg), glycogen and lactate concentrations per unit mass change remarkably with developmental stage (Fig. 4). Glycogen represents 2.33± 0.51 glycosyl units g^–1 body mass(mean ± s.e.m.) in stage 33–37 larvae, increasing sixfold in stages 38–42 (15.32±4.41), and almost tenfold in stages 43–45 (22.32±7.53), a sharp decrease following in the newly metamorphosed froglets (4.48±0.75). Lactate measures 2.28± 0.26μmol lactate g^–1 body mass (mean ± s.e.m.) in stage 33–37 larvae and is similar during stages 38–42, then increases sharply twofold (4.48±1.7) and threefold (7.63±0.64) in the late metamorphic stages, remaining similarly elevated in the young froglets(Fig. 4).

Developmental change accounts for 35% of the variability in glycogen at the whole-body level (F=3.56; P<0.019), while 55% of the variability derives from the effect of size (ANCOVA; F=5.58; P<0.026). In Table 3, the mass exponents for the relationship between glycogen and body mass suggest changes out of proportion both in the larval growth stages,during which glycogen content increases with body mass according to b=2.05 (r²=0.53; P=0.002), and in the stages of size reduction, during which glycogen content decreases sharply with body mass decrease, according to b=2.04 (r²=0.84; P=0.0001). Lactate content also increases out of proportion to the increase in body mass in the larval growth stages according to b=1.47(r²=0.77; P=0.000), then decreases in strict proportion to body mass decrease during late metamorphosis, with b=1.03 (r²=0.73; P=0.0001).

The 3 h exposure to aquatic hypoxia (P_O2=60 mm Hg) with access to normoxic air showed an interaction effect with developmental stage on glycogen concentration per unit mass (ANOVA; F=2.99; P=0.028); however, a large, interindividual variability is associated with this response(Fig. 4). Multiple comparisons among mean values revealed a significant decrease of 66% during larval stages 38–42 compared to values in normoxic water, a similar tendency being seen in stages 43–44. Hypoxia exposure may differentially affect the whole-body glycogen content of growing larvae according to body size, the slope for the regression being less steep than in normoxic conditions,suggesting that large tadpoles are more severely affected than smaller ones. However, the coefficient r²=0.147 does not predict functional dependence between the variables, and the small sample size and mass range in this group prevent a solid conclusion. The regression for the degrowth phase is likewise less steep after hypoxia exposure(r²=0.52; P=0.002), and in this case, the calculated equation predicts a 50% reduction in body glycogen in 20 g larvae,and no change in 5 g animals, typically froglets.

Hypoxia exposure caused no significant change in lactate content per unit mass compared to normoxic values (Fig. 4). However, the large change in the mass exponent for the whole-body content, to b=2.04 (r²=0.46; P=0.021), suggests that larger tadpoles accumulate surplus lactate in hypoxic water (Table 3). According to the calculated equations, body lactate content would increase by 84% in 20 g larvae relative to normoxic levels. In the degrowth stages,hypoxia exposure caused a decline in slope to b=0.79(r²=0.64; P=0.0001), and the equations indicate that under these conditions 20 g larvae contain 36% less lactate compared to normoxic levels, while the contents would be similar in 5 g froglets.

The present study shows that multiple adjustments of metabolism during the early development of P. paradoxus correlate with body mass changes resulting from the growth and degrowth processes. The scaling patterns reveal a general tendency for O₂ uptake to increase in direct or more than direct proportion to mass, and may express the consequences of size on energy demand at the tissue level. One consequence of these patterns may be that air breathing must provide potential for any further expansion of energy expenditure in the largest individuals, and for the large increase in metabolism during late metamorphosis. This may be particularly meaningful in the context of fluctuating O₂ availability in the aquatic environment, as suggested by remarkable changes in the scaling patterns, and by evidence of a size-dependent increase in lactate content, related to hypoxia exposure. These findings are examined below, and the means by which energy derived from aerobic and anaerobic sources may be reconciled with growth and developmental changes in P. paradoxus tadpoles, ensuring the large size at transformation defined by natural selection, is discussed.

The mass-specific O₂ consumption rates of larval P. paradoxus are fairly constant and low in the early larvae until the emergence of the forelimbs, values lying within the ranges reported for other frogs, despite large inter-specific differences in the degree of size variation during the larval period and in the adult mode of life(Gatten Jr et al., 1992). This phase constitutes more than three quarters of the larval period in P. paradoxus, lasting for 120 days under laboratory conditions(Cais, 1982) and possibly less in the natural environment, during which the larvae undergo minor developmental changes and body size increases; in the present study, the minimum and maximum weights at capture were 0.1 g (stage 26) and 72 g (stage 40), respectively. By contrast, Rana catesbeiana tadpoles can grow at very slow rates to a large size by increasing the developmental time during one or two overwintering cycles, as seen under temperature conditions that lead to interruption of metamorphosis(Emerson, 1988). Overall, then,the great size quickly attained by P. paradoxus tadpoles may result from higher rates of whole body protein accretion.

The tail represents 78% of body length in the largest tadpoles of P. paradoxus (stages 38–42), and its wet muscle mass increases exponentially (1.19) with body mass in the absence of significant changes in dry:wet tissue mass ratio (C.M.K. and S.C.R.d.S., unpublished data). Using available data on developmental time (Cais,1982) and a tail muscle protein content of 10 mg% obtained from our samples, a rough calculation gives 1.4 mg day^–1 for the rate of tail muscle protein synthesis in tadpoles growing at 1% per day. The fraction of Ṁ_O2expended on this process would be between 19 and 42%, employing a minimum cost of peptide bond formation of between 7 and 16 mmol O₂g^–1 of protein (Houlihan,1991). While representing a smaller fraction of the body mass,other sites of protein synthesis to be considered in the growing larvae are the liver and the gastrointestinal tract, as protein turnover by these tissues reaches very high rates compared to white muscle(Houlihan, 1991). Further,tadpoles are typically herbivores, exhibiting elevated rates of intestinal nutrient transport to extract the amino acids required for growth from large quantities of plant matter (Toloza and Diamond, 1990). The costs of growth would then include the energy expended on eating, digesting and absorbing food in excess of that needed for maintenance, which amounts to 60–80% of metabolic rate in young toads, Bufo bufo (Jørgensen,1988) and may represent a large fraction of the routine rates of metabolism in P. paradoxus tadpoles.

Oxygen consumption increases sharply in P. paradoxus tadpoles on emergence of the forelimbs concomitant with the abolishment of gill respiration and food intake, and may remain elevated until completion of metamorphosis over a period of approximately 10 days(Cais, 1982). Compositional changes during the advanced stages of metamorphosis, such as those in water content and in the relative size of internal organs, may partially explain this increase in metabolic rate. The dry:wet muscle tissue ratio increases slightly, both in the degenerating tail and in the growing leg of P. paradoxus tadpoles. In addition, there is a 19-fold increase in liver tissue mass from the early to the latest larval stages (C.M.K. and S.C.R.d.S.,unpublished data), associated to an increase in the liver somatic tissue index from 0.5 to 2% of body mass, implying that the biosynthetic capacity of this organ relative to body mass would increase in late metamorphosis. The heart size decreases with decreasing body size according to b=0.72, the net effect being a 1.7-fold increase in relative organ mass in later stage larvae(C.M.K. and S.C.R.d.S., unpublished data), and together these changes may increase O₂ consumption by an increase in the relative mass of metabolically active tissue. Accordingly, studies focusing on phenotypic plasticity and physiological adaptation have shown that the relative size of the heart, liver and kidney may explain much of the inter-individual and inter-specific variability seen in O₂ demand at the whole-body level (Selman et al., 2001; Williams and Tieleman,2000).

Despite the extensive tissue reorganization, however, O₂consumption decreases on completion of metamorphosis in P. paradoxus,and the mechanisms promoting the sharp increase in energy expenditure in the tadpoles, prior to metamorphosis, remain unknown. The late metamorphic events in anurans are marked by tail reabsorption, when more than half of the body weight is lost; the energy requirements of caudal proteolysis may be substantial in P. paradoxus tadpoles, given the large relative mass of muscle tissue that is degraded in a matter of a few days. Increases in the rates of anabolic processes underlying the extensive tissue reorganization,such as in the gastrointestinal tract, may also contribute significantly to the increase in metabolic rates at the whole body level(Hourdry et al., 1996). Besides, the amino acids resulting from tail resorption provide the animal with a continuous source of carbohydrate via the gluconeogenic pathway; recently we have found increased enzyme activity related to gluconeogenesis in the liver of tadpoles during late metamorphosis (F. M. Oshiro, S. C. R. de Souza, J. E. P. W. Bicudo and M. S. C. Bianconcini,unpublished observations). In mammals, nearly one-quarter of body O₂ use takes place in the liver and gastrointestinal tract under standard conditions (Rolfe and Brown,1997); assuming a similar proportion in the organs of P. paradoxus tadpoles, a fourfold increase in the rates of liver and gastrointestinal tissue respiration would be necessary to fully account for the twofold increase in O₂ consumption rates at the whole-body level during late metamorphosis.

Developmental changes in O₂ consumption of frogs do not share a common pattern, and the available data are far from exhaustive(Burggren and Just, 1992). A study with Xenopus laevis illustrates the contrasts with P. paradoxus tadpoles, showing a peak in Ṁ_O2 per unit mass on hatching, and another in newly metamorphosed frogs, while in the more advanced larval stages, rates decline slightly due to a presumed decrease in energy expenditure linked with cessation of gill respiration(Hastings and Burggren, 1995). The significance of the distinct metabolic patterns seen during the metamorphic transition of anurans is difficult to interpret based on the information available, and species-specific characteristics may also be involved, together with plasticity of morphological and physiological characters in response to the environment.

In normoxic water, the early larvae of P. paradoxus rely almost exclusively on aquatic O₂ exchange to maintain routine metabolism,as do other anuran larvae studied (Feder,1984). On emergence of the forelimbs, the contribution of aerial breathing increases sharply together with an equivalent decrease in aquatic O₂ uptake. Under aquatic hypoxia, gill ventilation is apparently inhibited and lung breathing stimulated at P_O2values around 90 mmHg and lower, Ṁ_O2 per unit mass in the developmental groups remaining nearly constant over the P_O2 range analysed. Previous data from Rana catesbeiana reveal a remarkable capability of the larvae to counteract the effects of aquatic hypoxia by increasing gill ventilation during forced submersion in water of decreasing P_O2, with effective O₂ regulation occurring down to a critical P_O2 of around 30 mmHg(Crowder et al., 1998). This pattern may differ among ranid frogs, as suggested by the study with R. berlandieri, showing a marked decrease in total O₂ uptake during submersion in low P_O2 water and insufficient capacity to offset the effects of aquatic hypoxia by air breathing (Feder, 1983). Despite the influences of phylogeny per se, and the more obvious differences in methodological approach, there is evidence of dissimilarities in the ability of anuran larvae to compensate for environmental O₂fluctuations by gill ventilation. In P. paradoxus tadpoles,adjustments in gill and lung ventilation are apparent in water P_O2 levels at which gas exchange across the integument is presumably effective, and may indicate early reliance on aerial O₂ uptake associated with the large larval size on transformation. Hypoxia exposure affects growth rates in many ways, as shown by reductions in feeding rate, respiration rate, faecal production and protein synthesis(Zhou et al., 2001). In the habitat where P. paradoxus tadpoles are found, O₂ supply is neither homogeneous nor stable, circumstances in which air breathing may be favoured.

Body mass in P. paradoxus tadpoles covaries with the effects of developmental stage on whole-body O₂ uptake, and accounts for some 90% of the variability under normoxic conditions, similar to observations on other anuran larvae (Feder,1982). However, the scaling patterns seen in P. paradoxusare distinct, and reveal a general tendency for O₂ uptake to increase in direct or more than direct proportion to mass, as in the early ontogenetic stages of mammal and fish species(Hulbert and Else, 2000; Wieser, 1995). The gas exchange organs of mammals and fishes presumably increase their efficacy with growth and development, while in metamorphosing amphibians, growth coincides at some point with gill degeneration. Despite this, and the inherent limitation of the skin for gas exchange, aquatic O₂ uptake increases in direct proportion to the increase in mass in P. paradoxus tadpoles, and the relationship nearly overlaps the corresponding curve for total O₂ uptake under resting conditions. Since this pattern does not prevent the developing animal from increasing metabolism with activity changes and metabolic state, some reserve capacity may be available at the O₂ transfer step. One consequence of this pattern, however, may be that air breathing must subsidise any further increase in energy expenditure in the largest individuals, as the scaling for aerial O₂ uptake in normoxia appears to indicate (b=1.13). Under hypoxia, the depression in cutaneous O₂ uptake is notably large in the larger tadpoles, causing the slope of the aquatic O₂uptake in the growth stages to change from 1.02 to almost zero(b=0.11), and they may depend disproportionately on air breathing to maintain O₂ consumption rates, as the slope of the aerial uptake indicates (b=1.44).

During the degrowth stages of late metamorphosis, the scaling relationships for aquatic O₂ uptake imply that the mass-specific rates of aquatic O₂ uptake are twice as great in the latest larval stages than in froglets (b=1.18). P. paradoxus larvae not only have unusually shaped tails, with the keel extending further onto the head compared to tadpoles of R. catesbeiana of similar size, but tail reabsorption takes longer than in other frogs (Emerson,1988). In the present study, tail length corresponds to 63% of total body length in stages 43–45, the oldest larvae resembling adult frogs with a larval tail hanging in the water column. Since its size and surface area actually increase relative to body mass during these stages of body mass reduction, the tail appears to function as an important site of O₂ transfer until completion of metamorphosis. In water of low O₂ content, however, cutaneous O₂ uptake becomes limiting and late-stage larvae may increase lung ventilation more than newly metamorphosed frogs to sustain their much higher metabolic rate, as suggested by the change in the mass exponent for aerial O₂ uptake(b=0.77) in normoxia to close to unity under hypoxia.

Other functions have been attributed to surfacing behaviour in developing tadpoles, such as prevention of lung collapse during organ development in the early larval stages (Burggren and West,1982; Bruce et al.,1994; Crowder et al.,1998). The emergence of central chemoreceptive reflexes regulating O₂ and CO₂ by lung ventilation precedes the disappearance of central gill respiratory regulation in R. catesbeiana, and as development advances, the central lung pattern generator gradually becomes more functional, dominating in the late metamorphic stages (Torgenson et al., 1997). In P. paradoxus, there is a disproportionate increase in the aerial O₂ uptake to size increase during larval growth stages, and a sharp increase in air breathing rates on forelimb emergence, reaching 20-fold in 30 g larvae, according to the calculated equation. Some plasticity in the ontogeny of respiratory control may have developed in P. paradoxus tadpoles, associated with their large size at metamorphosis. Other physiological and biochemical adjustments may also contribute to providing adequate O₂ transfer and delivery to the body tissues. Burggren et al.(Burggren et al., 1992)described changes in blood pressure variables in P. paradoxus of dramatic speed and magnitude, causing a threefold increase in mean, systemic,arterial blood pressure in late-stage larvae compared to younger larvae. This may be explained in part by the developmental changes in the relative heart mass, described above, and thus scaling effects on adjustments by the cardiovascular system may promote increased blood perfusion and contribute to an increase in the O₂ uptake and delivery steps to the body tissues in the later stages of P. paradoxus metamorphosis.

Body glycogen content rises in P. paradoxus larvae from stages 38 to 42, and remains high over the period of late metamorphosis, dropping in newly metamorphosed froglets. Lactate content increases sharply after emergence of the forelimbs and remains high after metamorphosis. Together,these changes express increased anaerobic potential and power generation for burst activity in the late larval stages. Given that food intake is interrupted during late metamorphosis in P. paradoxus, the high,sustained glycogen content suggests that an endogenous substrate is required for carbohydrate synthesis during the fasting period, possibly amino acids resulting from protein catabolism in the tail, in agreement with our recent findings of increased enzyme activity related to gluconeogenesis in the larvae of late stages (F. M. Oshiro, S. C. R. de Souza., J. E. P. W. Bicudo and M. S. C. Bianconcini, unpublished observations).

The lack of significant changes in lactate concentration per unit mass after 3 h of aquatic hypoxia is not unexpected, given the ability of P. paradoxus tadpoles to maintain aerobic, resting metabolic rates by increasing lung ventilation. Lactate does not build up in other anuran larvae,except after prolonged exposure to severe hypoxia(P_O2<10 mmHg)(Crowder et al., 1998), or on exertion, at rates dependent on the intensity and duration of exercise(Gatten, Jr, 1985; Quinn and Burggren, 1983). In this context, however, the significant decrease in body glycogen in P. paradoxus larvae in stages 37–42 exposed to hypoxic water appears paradoxical. One explanation is that carbohydrate stores may be mobilised and used to fuel oxidative processes, due to the higher energy yield per O₂ molecule from glucose(Hochachka and Somero, 2002);this mechanism is apparently essential to the development of hypoxia tolerance in animals (Hochachka et al.,1996; West and Boutilier,1998). An alternative interpretation comes from the suggestion that whole-body glycogen and lactate in larval P. paradoxus are markedly affected by body size. In the growing larvae, the calculated equations predict a large increase in lactate and glycogen levels of 2.8 and 4.1-fold, respectively, for a twofold increase in body size, similar to the pattern observed in teleost fish (Somero and Childress, 1980). After hypoxia exposure, there is a tendency for body glycogen to decrease more in the larger larvae, and a clear suggestion of a change in the mass exponent for lactate, implying that larger tadpoles produced surplus lactate after exposure to hypoxic water, calculated to be 1.8-fold of lactate in 20 g larvae compared to normoxic conditions. In the stages of larval growth, the tadpoles ascend periodically in the water column to ventilate their lungs, and apparently increase the frequency of ascending during hypoxic exposure. This behaviour tends to disappear during late metamorphosis, and the greater reliance on air breathing of the largest tadpoles might explain the differential production of lactate as a function of size in P. paradoxus.

In conclusion, multiple adjustments at the O₂ transfer step during the early development of P. paradoxus correlate with body mass changes resulting from the growth processes, similar to observations of metabolic scaling on fishes and mammals. Developmental programs operating through changes in organs and body parts may ultimately effect adjustments in the metabolic capacities of tissues and related systems, so that O₂transfer and energy expenditure may be reconciled with the large larval size and metamorphosis.

This study was financed by a research grant from the Foundation for Research Support of São Paulo State (FAPESP) to S.C.R.d.S. C.M.K. received a scholarship from FAPESP. The authors are deeply grateful to Wagner Bianchini for access to his fish farm and for valuable contributions during field work. Constructive criticism of two anonymous referees greatly improved the clarity of the article.