Responses to Acute Aquatic Hypoxia In Larvae of the Frog Rana Berlandieri

The respiratory patterns of amphibious vertebrates have attracted considerable attention in recent years (Johansen, 1970; Hughes, 1976; Randall, Burggren, Farrell & Haswell, 1981). A consensus among these and other studies is that the evolution of bimodal respiration was not a unique transitional event in the origin of terrestrial vertebrates from aquatic ancestors, but that it has evolved many times in response to adverse conditions commonly encountered by vertebrates in aquatic environments. Many adult fishes, amphibians and reptiles increase their dependence upon aerial oxygen uptake as aquatic oxygen declines, and show a variety of cardiac and respiratory responses that facilitate this transition. Obviously, the larvae of amphibians are of special interest in testing the generality of conclusions drawn mainly from adult vertebrates (Burggren & Wood, 1981 ; Burggren & West, 1982).

Although larval anurans (tadpoles) may encounter aquatic hypoxia in the field (Crump, 1981; Noland & Ultsch, 1981), aerial oxygen uptake of tadpoles has not been examined until recently (Feder, 1981a, b; Burggren & West, 1982; West & Burggred 1982), perhaps because these organisms seem to be exclusively aquatic and their blood PCO₂, [HCO₃⁻], and pH are characteristic of water breathers (Just, Gatz & Crawford, 1973). Indeed, tadpoles have several respiratory surfaces that are well developed for aquatic gas exchange, including the gills, the vascularized opercula, and the skin (Strawinski, 1956). However, many tadpoles have lungs that they ventilate regularly, even at developmental stages well before the appearance of limbs (Just et al. 1973; Wassersug & Seibert, 1975). Moreover, tadpoles generally increase the frequency of pulmonary ventilation (f_lung) in response to aquatic hypoxia (Wassersug & Seibert, 1975). These morphological and behavioural observations prompted an examination of physiological responses to acute hypoxia in the larvae of a frog, Rana berlandieri, with particular emphasis on air-breathing.

In addition to examining the generality of vertebrate responses to hypoxia, a study of larval anurans may also provide insights into the scaling of the partitioning of oxygen uptake between air and water as well as the limitations imposed on gas exchange by body size. Because scaling of respiratory surface area with body size is allometric in amphibians, some large amphibians may face problems with diminished Oz exchange capacity (Ultsch, 1976). Moreover, factors such as large size, an aquatic existence, high temperatures, activity, lunglessness and hypoxia may exacerbate respiratory supply problems, especially when amphibians must deal with two or more of these factors simultaneously (Ultsch, 1976; Feder, 1977). One possible result of these problems is that large larvae should depend more on aerial respiration than small ones. A second possible result (reviewed by Feder, 1977) is that amphibians lacking lungs would be unable to attain large body sizes. Indeed, a prima facie case for respiratory restriction to small size is evident in anuran larvae (Wassersug & Seibert, 1975). Bufonids (toads) lack lungs as larvae and metamorphose at relatively small sizes, even though adults of some bufonids (e.g. Bufo alvarius, Bufo marinus) are among the largest adult anurans. Generally, non-bufonid larvae, most of which have lungs, become larger than bufonid larvae before metamorphosis.

The present study examines (1) the significance of aerial gas exchange as a response to aquatic hypoxia in tadpoles of Rana berlandieri; (2) alternative responses to hypoxia, including increased gill ventilation and anaerobiosis; and (3) the effects of body size on the relative importance of aerial oxygen uptake. With a similar rationale, Burggren & West (1982) and West & Burggren (1982) have undertaken parallel studies of a congeneric species, Rana catesbeiana.

The details of the origin and care of larvae appear elsewhere (Feder, 1981a). The experimental larvae were maintained at a constant temperature, 25 °C, on a constant photoperiod (LD 14:10 centred at 13.00 local time) for at least 1 week before experimentation. All measurements were made under these conditions. The larvae used in the measurements of the rate of oxygen consumption were not fed for 2 days before experimentation to minimize production of faeces and associated microbial (Feder, 1981a). Even so, these larvae were not strictly post-absorptive as they continuously ingested faeces and detritus. The larvae in other experiments were not separated from food until 4h before experimentation. All larvae were at developmental stages before emergence of the forelimbs [stages I-XV of Taylor & Kollros (1946)].

Closed-system respirometers were employed that could be flushed with water at a desired PO₂ and then sealed without disturbing larvae. These respirometers were thermostatted at 25 ± 0·1 °C. Respirometers were Erlenmeyer or round-bottomed flasks fitted with rubber stoppers. The respirometer volumes were between 100–300 ml. The respirometer volume, animal size and the number of animals per respirometer (1–5) were varied to achieve a similar ratio of animal volume to respirometer volume.

Experimental larvae were placed in respirometers and left overnight to become accustomed to these chambers ; aerated water was flushed through the respirometers during this period. Before each measurement of aquatic ⁠, respirometers were flushed with 750–1000 ml of water at a pre-determined PO₂. This water was sterilized and pre-treated with antibiotics to minimize microbial respiration (see Feder, 1981a), and its PO₂ was regulated by bubbling N₂ through it. After the flushing had been completed water samples were withdrawn from the open respirometers into glass syringes, and the respirometers were sealed at the beginning of measurements. The PO₂ of half of each sample was determined immediately. The remaining half was held in the syringe as a control for microbial ⁠. A final sample was taken after 10–45 min had elapsed. The aquatic was calculated from the difference between the PO₂ of the final sample and the control sample, the elapsed time, the respirometer volume and the O₂ capacitance under experimental conditions.

The PO₂ was measured with a thermostatted No. 730 Clark microelectrode and Chemical Microsensor system (Transidyne General Corp., Ann Arbor, MI). The electrode was calibrated between each measurement with air-equilibrated water at 25 ± 0·1°C.

This protocol was varied in three ways, (a) Animals were allowed no access to air, and the PO2 was reduced continually throughout experimentation, (b) Animals were treated as in (a), but a magnetic stirring bar was included in the respirometer. The stirring bar was agitated during measurements (10–20min), and animals responded by swimming vigorously and continuously, (c) Animals had access to air (approximately 10 ml) in the upper neck of a 281 ml round-bottomed flask. Both aerial and aquatic compartments were flushed before measurements. The aerial phase was sampled with gas-tight syringes and its PO₂ determined with a 0-5 ml Scholander analyser (Scholander, 1947). The dead space was filled with mercury to minimize error. The aerial was calculated from the change in aerial PO₂ and airspace volume. All calculations of aerial and aquatic were adjusted for diffusion from air to water; adjustments were based upon the apparent aerial O₂ consumption of a control respirometer containing no animals. The PO₂ of the aerial phase never declined below 130 Torr; the initial PO₂ of the water was reduced continually in consecutive measurements.

Pulmonary ventilation frequency (f_lung), branchial ventilation frequency (f_gill), and heart rate (/heart) were determined visually. The larvae were placed in experimental containers 4 h before observation. The PO₂ was measured with a YSI 54A O₂ meter (Yellow Springs Instruments, Antioch, OH), and was adjusted by bubbling N₂ through the water.

To determine the f_lung, the larvae were placed in individual 1000 ml Erlenmeyer flasks containing water at approximately the desired PO₂, and observed for 1 h. The PO₂ was measured after observations were complete. To measure f_heart and f_gill in larvae with access to air, larvae were placed in a screen cylinder (20 cm diameter × 20 cm height) within a larger aquarium. This arrangement allowed gassing of water but minimized physical disturbance of the larvae. Each tadpole was observed for 5–6 intervals of 1 min, and the frequencies recorded during these intervals were averaged. The body wall beneath the heart was surgically removed sufficiently to visualize the heart beat (Wassersug, Paul & Feder, 1981) in those animals undergoing determination of f_heart. This procedure apparently caused no more trauma to the larvae than the implantation of subcutaneous ECG electrodes (Wassersug et al. 1981; Burggren, Feder & Pinder, 1983). The f_gill was also measured in larvae held without access to air in a submerged, screen-covered box (see Wassersug et al. 1981). The PO2 was reduced between each determination of f_gill and f_heart.

In one experiment, larvae were held in 300 ml screen cylinders within aerated aquaria for 4h before experimentation. The PO₂ was then adjusted to experimental levels, the larvae were held at that PO₂ for 1 h, and the larvae were then killed for determination of lactate content. Feder (1981a) described the lactate determination procedure. The larvae were divided into two groups. In one group, the containers were submerged so that larvae had no access to air during the entire 5 h of experimentation. The other group had access to air during experimentation. An additional group of larvae were placed in screen containers, the containers were submerged overnight in aerated water, and the larvae were killed for determination of lactate the following morning.

To avoid lactate production due to handling of the larvae, morphometric measurements were made on the day before experimentation. These measurements were substituted into a multiple regression equation to calculate the dry mass of the larvae on the day in which lactate was determined (Feder, 1981a).

Larvae were confined without access to air in separate 37 ml compartments of a screen-covered box. After 4h for equilibration, an observer reduced the PO₂ in the container and noted the ‘critical activity point’, at which larvae no longer responded to tactile stimuli.

Larvae were placed individually in a glass aquarium (50×20×20 cm) filled will repeated water. After 4 h for equilibration, an observer noted all locomotor movements the larvae and their duration. Observations were repeated for each larva after reduction of PO₂. A plastic screen shielded larvae from physical disturbance due to bubbling of N₂ and PO₂ measurements.

The larvae were weighed and measured as detailed by Feder (1981a). In some cases the data were transformed by expressing the as a percentage of the routine expected for Rana berlandieri larvae of similar size (Feder, 1982); the range of these data exceeded 100 % of expected routine ⁠. The transformed data were analysed according to several statistical models, including linear, polynomial and logarithmic regressions (Mangum & Van Winkle, 1973), and two line segments with a common join point (Welch, 1979). The latter model accounted for the greatest proportion of the variance in the ⁠, and was used subsequently.

Multiple stepwise regression was performed to determine how body size affected bimodal gas exchange. The (aerial, aquatic and total) of all larvae was first expressed as a proportion of the expected for an animal of mean size. This proportion was the dependent variable in the multiple regression. It was impossible to enter the PO₂ directly into the multiple regression because its effect on was not linear (Figs 1, 2); hence the expected for an animal of mean size was entered in its place as the first independent variable. Next, terms representing dry mass, and the interaction of and dry mass were entered in that order.

Fig. 1 shows the response of Rana larvae to a decline in aquatic ⁠. (Hereafter, refers exclusively to the of the water.) Table 1 presents the equations for the line segments presented in Fig. 1. In normoxic water, aerial gas exchange averaged 18 % of the total ⁠. As the PO₂ declined the aquatic decreased steadily, and animals lost O₂ to the water at low ⁠. By contrast, the aerial increased and actually came to account for more than 100% of the total ⁠. This increase, however, was insufficient to offset the decrement in aquatic ⁠; hence the overall declined under hypoxia.

Also evident was substantial individual variation about this general pattern. For example, in one animal (‘+’, 144 mg dry mass) the aerial accounted for a much larger proportion of the total than in other animals (Fig. 1). Part of this variation may have been due to individual differences in routine activity (see below).

Body size had little effect on the aquatic of Rana larvae. Neither large nor small larvae showed a tendency to depart from the general relationships in Fig. 1. Hence, hypoxia affected the aquatic equally in all larvae, regardless of size. In statistical terms, body size and body size- interaction (actually body size-m interaction; see Materials and Methods) accounted for < 1 % of the variance in aquatic (Table 1 ). No effects were attributable to the modest variation in developmental stage the experimental larvae

By contrast, body size markedly affected the aerial and the total of the larvae (see Fig. 2). At normoxic PO₂, the small larvae depended more upon aerial than the large tadpoles. As the PO₂ declined, however, this relationship reversed and large larvae depended disproportionately upon aerial ⁠. The total also reflected these trends (Fig. 2). Large larvae maintained a relatively equable total throughout a wide range of PO₂. Smaller tadpoles had a high relative total in normoxic water but underwent a precipitous decline in total as the PO₂ decreased. In statistical terms, body size accounted for little of the variance in aerial and total ⁠, but the interaction of PO₂ and body size (i.e. -body size interaction) explained 6 % and 11 % of the variation in the aerial and the total ⁠, respectively (Table 1). Developmental stage had little effect on the aerial and total oxygen uptake.

In normoxic water, larvae without access to air were able to augment their aquatic to compensate for the absence of aerial O₂ uptake (Fig. 3). This ability was size dependent. The total (in μl h⁻¹) for these animals and for similar animals that had access to air (Feder, 1982) was compared with analysis of covariance. Access to air per se did not affect the total (F_(1,66) = 1·12; P = 0·29); however, the interaction of body mass and access to air significantly altered the total (F_(1,66) = 4·68; P = 0·03). In other words, both the large and the small larvae without access to air were able to achieve approximately 100 % of the expected routine ; however, the small larvae (but not the large larvae) frequently exceeded 100% of the routine ⁠.

By contrast, the total of larvae without access to air decreased markedly under hypoxia (Fig. 3). This effect was equally evident in larvae of all sizes.

Fig. 3 also shows the aquatic expected when larvae had access to air. As if evident from this Figure, the larvae without access to air did not exceed this level in hypoxic water. To determine whether this situation represented a voluntary reduction in the or a limitation of the aquatic gas exchanger, some animals without access to air were stimulated to vigorous activity. In 14 trials (PO₂ = 44–70Torr; size = 13–596 mg dry mass) the averaged 80 % of the expected routine total VO2, and ranged from −22% to 200%. No correlation was evident between the (in % of expected levels) and either the PO₂ or body size. Obviously, some larvae extracted more O₂ from water than was evident in ‘resting’ measurements (Figs 1, 2). However, it is unclear whether this increase represented a physiological response to activity or inadvertent facilitation of ventilation and diffusion due to stirring of the water.

As indicated by the direct measurements of the total and the aerial ⁠, the larvae breathed air at all PO₂ observed and increased their f_lung in response to low PO2 (Fig. 4). The f_lung was unaffected by the PO₂ above approximately 50–65 Torr; below this range the f_lung increased in inverse proportion to the PO₂. This threshold PO₂ was similar for both large and small larvae; also, larvae of all sizes showed similar changes in the f_lung in response to declining PO_2.

The f_heart was observed in only three animals (Fig. 4). In these animals the /heart increased and then decreased as the PO₂ declined.

Similarly, the f_gill increased from normoxic levels under moderate hypoxia, but declined from maximal levels under severe hypoxia (Fig. 4). The maximum f_gill was approximately twice the f_gill for animals in normoxic water.

The pattern of change in the f_gill was different for large and small larvae. In the smaller larvae, the curves relating the f_gill to the PO₂ were displaced towards lower PO₂ ; i.e., the maximal/pii occurred at relatively low PO₂ and thef_gill was relatively high, even at low PO₂. This apparent trend was confirmed by calculating the PO₂ at the centroid of the polygon bounded by each f_gill-PO₂ curve in Fig. 4 and the X axis. Also analysed were similar measurements (not shown in Fig. 4) for seven additional small larvae. The central PO₂ was positively correlated with the size of larvae (Spearman’s r = 0·587; P < 0·05).

The f_gill differed in larvae with and without access to air. In the larvae without access to air (Fig. 5), the f_gill was low in normoxic water and increased steadily as the PO₂ decreased (Friedman’s analysis of variance ; P < 10⁻⁴). Thus the f_gill was not maximal at some intermediate PO₂, as in the larvae with access to air. Moreover, the/pji for exclusively aquatic larvae was significantly lower (P < 0·05) than the/pii for air-breathing larvae at moderate and high PO₂ (57−150 Torr), similar at 38 Torr (P > 0·75), and significantly greater at 19 Torr (P < 0·05 ; one way analysis of variance in each case Considerably smaller larvae (n = 20; mean = 4·5 mg; range =1–8·6 mg dry mass) were also observed while without access to air in individual compartments of a plastic box (Wassersug et al. 1981). These larvae displayed significant increases in the f_gill with declining PO₂ (P < 0·001) that were qualitatively similar to the pattern for larger tadpoles. However, the f_gill was relatively high in the small larvae, ranging from 111–167 % of the mean f_gill of larger larvae at each PO₂.

Larvae varied enormously in the amount of spontaneous activity. At any given PO₂, individual larvae differed in activity; some moved almost continually and others were quiescent. Frequently, individual larvae underwent different amounts of activity at each PO_2. No characteristic pattern of activity was associated with either large or small larvae.

Among the larvae exposed to aquatic hypoxia for 1 h (Table 2), only exclusively aquatic larvae at the lowest PO₂ (19 Torr) showed a significant elevation of whole-body lactate concentration (P < 0·01 ; analysis of covariance with predicted dry mass as the covariate). The lactate concentration of bimodally breathing larvae at the same PO₂ was not elevated similarly. If the larvae at 19 Torr are excluded from the analysis, none of the remaining experimental groups, whether exposed to hypoxic PO₂, denied access to air, or both, differed significantly from one another in lactate concentration (PO₂ = 0-57; analysis of covariance).

In exclusively aquatic larvae, exposure to moderate hypoxia (PO₂ = 38 Torr) for 3 h did not elevate the lactate concentration significantly as compared to the 1 h levels (Table 2). However, exposure to water at 38 Torr for 6h resulted in a five-fold increase in the lactate concentration (P < 0·01; analysis of covariance).

In the larvae held overnight in hypoxic water without access to air, the lactate concentration was positively correlated with body size (Spearman’s r = 0·65). Although the correlation was significant (P < 0·05), the lactate levels were comparable to concentrations in normoxic animals with continual access to air.

With the exception of a single larva (182 mg dry mass) that succumbed at 38 Torr, the critical activity point ranged between 2–17 Torr PO₂ for larvae without access to air. The tolerance of hypoxia was unrelated to the body size. In three separate determinations, the correlation between the PO2 at the critical activity point and the body size was not significant (Spearman’s r = 0·07, 0·13 and 0·31 ; P> 0·05 in each case).

Most textbook accounts and many recent studies have characterized the tadpole as an exclusively aquatic stage that precedes the transition to air-breathing in anurans. As the present study clearly demonstrates, air-breathing contributes significantly to total oxygen uptake in the larvae of Rana berlandieri even in normoxia, and is the sole mode of oxygen acquisition under severe aquatic hypoxia. Similar results have been reported for larvae of Rana catesbeiana (Burggren et al. 1983) and Xenopus laevis (Feder, 1981b). Some tadpoles, in fact, live entirely outside the water for considerable lengths of time (Wassersug & Heyer, 1983). All of the above reports plus several accounts of air-breathing behaviour (Just et al. 1973; Wassersug & Seibert, 1975) refer in part to larvae before metamorphosis [Taylor & Kollros (1946) stages I-XV]. Thus, except in certain lungless tadpoles (Wassersug & Seibert, 1975), bimodal respiration is a common and significant feature of anuran larvae well before the transition to an adult morphology.

The occurrence of air-breathing in tadpoles need not imply that aquatic respiration is ineffective. In normoxic water at a variety of temperatures, anuran larvae use aquatic respiration (principally cutaneous gas exchange) to meet the majority of their oxygen demands (Burggren & West, 1982; West & Burggren, 1982; Burggen et al. 1983). In the present study, larvae in normoxic water without access to air were able to sustain the same or greater total than unrestrained larvae with access to air. The larvae of Rana catesbeiana in fact cease ventilating their lungs when in hyperoxic water (West & Burggren, 1982). Nonetheless, even though tadpoles can augment branchial O₂ uptake somewhat in response to moderate hypoxia (West & Burggren, 1982), hypoxia ultimately limits the effectiveness of aquatic O₂ uptake as the PO₂ gradient across the skin and gills declines.

In addition to air-breathing, the acute responses to aquatic hypoxia in tadpoles include direct avoidance of hypoxia (Costa, 1967); undirected movement that may lead larvae from hypoxic water or minimize the boundary layer about the skin (Hutchison, Haines & Engbretson, 1976); and changes in the branchial flow rate, the branchial O₂ extraction efficiency (West & Burggren, 1982), the f_gill and the f_heart. Many of these responses attain their greatest magnitude at moderate hypoxia and diminish under severe hypoxia, since they actually promote loss of O₂ to the water under the latter condition. Anaerobiosis does not occur except in unusual circumstances. Then the total evidently decreases in hypoxia, as it does in some fish (Burggren & Randall, 1978).

The pattern of responses to hypoxia in tadpoles is of special interest because the phylogenetic affinity of tadpoles is with adult anurans, which are predominantly air-breathers (Hutchison et al. 1976; Burggren & West, 1982), but the habitat and habits of tadpoles resemble those of fish in which air-breathing is supplementary. The anuran larvae examined in the present study and elsewhere (Feder, 1981b ; Burggren & West, 1982; West & Burggren, 1982; Burggren et al. 1983) resemble air-breathing fishes in the general pattern of responses to hypoxia. The general pattern is one intermediate between strict metabolic conformity to aquatic PO2 and strict metabolic regulation (McMahon, 1970; Hughes & Singh, 1971; Singh & Hughes, 1971; Graham, Kramer & Pineda, 1977; Gee & Graham, 1978; Stevens & Holeton, 1979). The pattern of increasing dependence upon aerial respiration as the aquatic PO2 declines is also common in air-breathing fishes (Johansen, 1970; Randall, Burggren et al. 1981).

However, fishes and tadpoles typically differ in aspects of branchial ventilation and anaerobiosis. Although some species do not conform to this pattern, fishes respond primarily by increasing the buccal pump stroke volume, with increasing f_gill playing a secondary role (Johansen, Lenfant & Grigg, 1967; Hughes & Saunders, 1970; Cech &Wohlschlag, 1973; Hughes, 1973; Kerstens, Lomholt & Johansen, 1979; Holeton, 1980). By contrast, larvae of every anuran species studied to date vary their f_gill in response to changing respiratory needs (Wassersug & Seibert, 1975; Feder, 198la,b; Wassersugei al. 1981; Burggrenet al. 1983). Changes in branchial stroke volume may also occur, but are inconsistent in their prevalence and direction (West & Burggren, 1982; Burggren et al. 1983). A second difference concerns the end product of anaerobiosis. Lactate production is a common response to severe hypoxia in amphibians (Armentrout & Rose, 1971; Bennett & Licht, 1974; D’Eon, Boutilier & Toews, 1978; Gatz&Piiper, 1979; Feder, 1981b). By contrast, hypoxic fish produce relatively little lactate, and generally accumulate other end products of anaerobiosis (Hochachka, 1980). Given the circumstances under which lactate formation occurs in amphibians and the deleterious consequences of lactate formation, it appears that anaerobiosis in tadpoles is primarily a response of last resort or reflects the increasing stress and muscular activity in severe hypoxia, rather than a routine response to moderate hypoxia.

A more significant difference between anuran larvae and fishes is that the skin is a major site of organismal gas exchange in anuran larvae (Burggren & West, 1982; Burggren et al. 1983), while in fishes it is not (Kirsch & Nonnotte, 1977). Unlike the skin of fishes, which is covered by scales and a layer of mucus, tadpole skin is relatively thin and highly vascularized (Strawinski, 1956), as in most amphibians. This arrangement may facilitate cutaneous O₂ uptake in normoxia (Burggren & West, 1982), but evidently renders impossible the curtailment of O₂ loss to hypoxic water. Fishes may prevent most O₂ loss to hypoxic water simply through cessation of branchial gas exchange. This may occur behaviourally by reducing water flow over the gills or through evolution of a reduced branchial surface area (Randall, Burggren et al. 1981). Thus loss of O2 to hypoxic water in fishes, while it may sometimes happen (Burggren 1979; Stevens & Holeton, 1979; Randall, Cameron, Daxboeck & Smatresk, 1981), is relatively uncommon. Many tadpoles also decrease branchial ventilation in response to severe aquatic hypoxia (West & Burggren, 1982; M. E. Feder, unpublished data for Bufo woodhousei andXenopus laevis), but this response leaves cutaneous exchange unencumbered. One means of minimizing cutaneous O₂ loss is the shunting of blood away from the skin. However, although some adult amphibians can control the perfusion of the skin to alter gas exchange (Moalli, Meyers, Jackson & Millard, 1980), such control (if present at all) is ineffective in preventing O₂ loss in larvae of Rana and of other species (Feder, 1981b). Larvae are only able to offset aquatic O₂ loss by greatly increasing aerial ventilation. Thus in some circumstances anuran larvae that have lungs may be committed to aerial respiration to a far greater extent than are most air-breathing fishes.

The responses to hypoxia in the Rana larvae suggest that the lack of lungs per se does not restrict larvae, such as those of Bufo (see Introduction), to small size. For lunglessness to restrict anuran larvae to small sizes, the aquatic gas exchangers (gills and skin) should be inadequate in large larvae. If so, then large larvae should be more susceptible to hypoxia or show greater efforts at compensating for hypoxia than small larvae (Ultsch, 1976; Feder, 1977). Most data for the Rana larvae, however, were contrary to these predictions. The large and the small larvae showed no consistent differences in critical PO₂, underwent equivalent depressions in when denied access to air in hypoxic water, and were similarly tolerant of hypoxia. Although the large larvae had greater lactate concentrations than the small larvae after prolonged submergence in normoxic water, this difference was quantitatively negligible. Among Rana larvae with access to air, the large larvae seemed more effective in their use of bimodal gas exchange than the small larvae. The large larvae derived a relatively high proportion of their total from aerial sources, and decreased their f_gill at a low PO₂ more readily than the small larvae. The relatively precise regulation of the total in the large larvae (Fig. 2) reflected these differences.

No effects in the present study were attributable to the developmental stage (as opposed to body size). The absence of any ontogenetic trends may reflect no more than the similar developmental stages of the larvae chosen as experimental subjects. Other studies that intentionally varied the developmental stages of larvae have shown major ontogenetic changes in gas exchange associated with the development of lungs and the atrophy of gills (Burggren & West, 1982; West & Burggren, 1982).

Pulmonary respiration is obviously an important component of the responses to aquatic hypoxia in anuran larvae. Moreover, in larvae that can use their lungs, large larvae gain more benefit from lungs than do small larvae. Even so, in larvae that are exclusively aquatic, the disadvantages of lunglessness seem to affect larvae of all sizes equally. A small lungless larva, such as Bufo, would be no better off than a large tadpole without lungs, although both might be at a disadvantage compared to larvae with lungs. Hence, the small size of bufonid larvae may be related primarily to ecological and energetic considerations (Feder, 1976) rather than a direct consequence of lunglessness.

I thank Rashel D. Paul for her expert technical assistance, and Drs Warren But gren, Richard Wassersug and an anonymous reviewer for their criticism of the manuscript. Research was supported by NSF Grant DEB 78-23896, The Andrew Mellon Foundation, and The Louis Block Fund, The University of Chicago.