Effect of Temperature Upon Carbon Dioxide Stores in the Snake Coluber Constrictor and the Turtle Chrysemys Scripta

Beginning largely with the work by Robin (1962) on the turtle Pseudemys scripta (now Chrysemys scripta), investigators have found that in vivo plasma pH varies inversely with body temperature in ectotherms (see reviews by Reeves, 1977; Heisler, 1986). The actual change in pH varies considerably among species. For example, over a wide temperature range pH changes by −0·005 units°C⁻¹ in the lizard Varanus exanthematicus (Wood, Johansen, Glass & Hoyt, 1981) but in the turtle Chelydra serpentina pH changes by about −0·016 units °C⁻¹ (Howell, Baumgardner, Bondi & Rahn, 1970). In addition, the thermal dependence of plasma pH is not linear in many species; typically pH changes less at lower temperatures (Robin, 1962; Kraus & Jackson, 1980; Ackerman & White, 1980; Nolan & Frankel, 1982; Heisler, 1984; Glass, Boutilier & Heisler, 1985).

In air-breathing ectotherms, declining plasma pH is achieved primarily by reducing ventilation relative to CO₂ production so that varies directly with temperature. Most investigators report little or no change in plasma bicarbonate concentration (see Jackson, 1978; Heisler, 1986). Since bicarbonate represents about 95% of the total plasma CO₂ content it appears that is also little affected by temperature. This apparently stable is a central tenet of the widely cited alphastat hypothesis (Reeves, 1972). Reeves proposed that protein imidazole groups are by far the most important buffer of CO₂ in the extracellular fluid. Consequently, assuming there are no changes in fixed acids, a constant implies the maintenance of a constant net protein charge (termed alpha). He later extended the alphastat hypothesis to intracellular fluid and argued that a constant alpha is biologically important for maintaining enzyme activity, buffer properties and Donnan equilibria across cell membranes (Reeves, 1976b; Malan, Wilson & Reeves, 1976).

More recent work, particularly by Heisler and his coworkers (see Heisler, 1986; Boutilier, Glass & Heisler, 1987), clearly shows that the temperature-dependence of pH in most ectotherm tissues cannot be explained by the alphastat hypothesis. In addition, the use of improper values to calculate plasma bicarbonate concentration has contributed to the view that is little affected by temperature. As recognized by Nolan & Frankel (1982) and by Nicol, Glass & Heisler (1983) the frequently used equation developed by Reeves (1976a) for estimating does not correct for pH. Consequently, the rise in plasma pH in vivo as body temperature falls in ectotherms is not accounted for and the predicted is too high. This error has a large effect upon calculated bicarbonate concentration because of the exponential relationship between and bicarbonate concentration (see Table IV of Howell & Rahn, 1976).

Although errors in can explain why many investigators reported only minor changes in plasma bicarbonate concentration, they do not explain why directly measured plasma did not change with temperature in a number of studies (Reeves, 1972; Kinney, Matsuura & White, 1977; Ackerman & White, 1980; Bickler, 1981; Nicol et al. 1983; Hicks, Ishimatsu & Heisler, 1987). A few studies, however, have reported large changes in directly measured in reptiles and amphibians. Kayser (1940) reported increases in blood when European tortoises, Testudo graeca, and European common frogs, Rana temporaria, were cooled. In the frogs, blood changed by about −0·3 mmol °C⁻¹, and in the tortoise, blood changed by roughly −0·7 mmol °C⁻¹. These changes in Cco₂ were consistent with marked transient depressions in the respiratory exchange ratio (R) that occur when ectotherms are cooled (Dontcheff & Kayser, 1937; Kayser, 1940). Much later, Stinner (1982) reported that blood in gopher snakes, Pituophis melanoleucus, increases 0·13 mmol per °C decrease in body temperature. He also measured large changes in R that agreed with the blood data, and hypothesized that variations in lactate production might be responsible for his results. These findings, like those of Kayser, strongly suggest that blood and whole-body CO₂ stores are dependent upon temperature in at least some air-breathing ectotherms.

In the light of these two opposing views concerning ⁠, the following study was undertaken primarily to investigate the relationships between body temperature, whole-body CO₂ stores and blood ⁠. We measured arterial lactate concentration and plasma pH as well as over a wide temperature range in resting snakes (Coluber constrictor) ranging in size from 104 to 336 g. These results are compared with the effects of changing temperature upon R and upon blood in vitro. Additional measurements of R were made in turtles, Chrysemys scripta, weighing about 1·85 kg.

Black racers (Coluber constrictor) and pond sliders (Chrysemys scripta) were purchased from commercial suppliers and flown to Ohio. Subsequently they were kept in cages provided with light sources that produced basking sites of about 35 °C. Racers were fed mice and small rats weekly and water was available at all times. The turtles were fed canned dog food and lettuce each week and had access to a 250-1 water tank. All animals used in this study appeared in excellent health. Food was withheld for 1 week prior to experimentation.

Metabolic rate was measured in each snake or turtle over a 3-week period. During this time an animal was kept at 30°C for the first few days, then body temperature was lowered to near 5°C where it remained for 1–2 weeks. Subsequently body temperature was returned to 30 °C and measurements were continued for a few more days. Body temperature was monitored by means of a radiotransmitter (Mini-mitter Corp.) that had been force-fed to each animal prior to the metabolic studies. The metabolism chamber was periodically flushed with room air so that O₂ did not fall below 19% and CO₂ did not rise above 1%.

During the 3-week study period the animals were left undisturbed and were not given food or water. They lost approximately 10% of their initial body mass, presumably due mostly to evaporative water loss, but appeared in excellent condition at the end of the study.

For surgery snakes were chilled in crushed ice, and lidocaine (2%) was injected at the site of incision. A PE 10 cannula filled with heparinized saline (1000 units ml⁻¹) was tied occlusively into a branch of the dorsal aorta near the cloaca. The cannulae were inserted through a small (approx. 1 cm) ventral incision and passed to the exterior on the dorsal surface of the tail. Snakes were then allowed to recover for about 1 week at 30 °C. During this time water was available but the snakes were not fed. Following recovery, 13 snakes were taped to sticks and placed inside temperature cabinets (for details see Stinner, 1987a). The trailing ends of the cannulae were left outside the cabinets so that blood could be sampled without disturbing the snakes. An additional seven cannulated snakes were left unrestrained inside temperature cabinets and were picked up at the time of blood sampling. Each snake was tested at 2–3 temperatures and usually 24 h (minimum of 10 h) was allowed for thermal equilibration.

To obtain blood samples, a cannula was thoroughly flushed and approximately 200 ul of blood was drawn directly into heparinized capillary tubes. Blood ⁠, and pH were immediately measured using Radiometer BMS-3 electrodes thermostatted to the snake’s body temperature. The electrode was calibrated with Radiometer -zero solution and aerated water. The and pH electrodes were calibrated with Radiometer gas mixtures and precision buffers, respectively. Calibrations were verified immediately before blood gas measurements. Plasma bicarbonate concentration was calculated from simultaneously determined and pH and the Henderson-Hasselbalch equation ⁠. The solubility coefficient (α) was corrected for temperature using the formula of Boutilier, Heming & Iwama (1984) that was derived from values tabulated by Severinghaus (1965). was corrected for temperature and pH using equation 16 on page 35 of Siggaard-Andersen (1974).

Snakes were cannulated as described above and allowed to recover for 1–3 weeks inside cages provided with heat lamps. Water was available at all times and the snakes were occasionally fed mice. Subsequently the snakes were placed inside individual plastic containers. To prevent movement of the cannulae, a relatively small amount of restraint was employed by taping an 8 cm section of the snake near its tail to the container floor. The container and snake were placed inside a darkened temperature cabinet and blood (approx. 300 μl) was sampled from undisturbed snakes as described above.

was immediately determined manometrically on a 30-μl sample of blood using a Natelson microgasometer (Rolant, 1969). For lactate measurements, a 200μl sample of blood was immediately deproteinized by adding 400 μl of an 8% perchloric acid solution. The supernatant was then stored for not more than 1 week at 2–3°C before analysis with an enzymatic test kit (Sigma no. 826) and Coleman Jr II Model 6/20 spectrophotometer.

Lactate concentration and C_CO2 were measured in 12 snakes and each snake was tested at 2–3 temperatures. Before sampling blood at each temperature, sufficient time was allowed for completion of transients in R caused by changing body temperature (see below). This meant that snakes were at 5°C for 4–5 days and at higher temperatures (13–36°C) for 2–3 days before measurements were made.

In addition to steady-state measurements, we examined the time course of changes. In four snakes, serial blood samples were taken when body temperature was lowered from 30 to 5 °C and then returned to 30 °C. The experimental set-up was like that used for the lactate and CO₂ studies except that body temperature was monitored using a temperature probe (Yellow Springs Instrument Co.) inserted into the cloaca. Twenty-five 50-μl blood samples were taken from each snake over 5 days. The blood samples were immediately analysed for using the Natelson microgasometer.

All blood sampling was done between 07.00 and 20.00h local time. At the completion of these experiments the snakes appeared in good health and generally began feeding within a few days to 1 month. The tape and cannulae were removed when the snakes shed their skins and many of the snakes were later released.

Minimal rates of for the four snakes at 5 and 30·5 °C were 4·1 ± 0·8 (x̄ ± S.D.) and 57·5 ± 13·5 μlsTPD g⁻¹h⁻¹ (x̄ mass = 291 g, range 238 –336 g). Mean ± S.D. for R during steady-state conditions was 0·75 ± 0·06 at 5°C and 0·75 ± 0·02 at 30-5°C. Lowering temperature in C. constrictor produced a marked decrease in R (Fig. 1). R fell to between 0·15 and 0·36 in the first few hours during which temperature was falling, then gradually increased back to steady state by about 60 h. Elevating body temperature from 5 to 30·5 °C produced a rapid two-to three-fold increase in R (Fig. 2). Within about 6h R had fallen back to below 0·8 and by about 24 h it reached steady-state values near 0·75. These large changes in R suggest that total body CO₂ stores are higher at lower temperatures. In the four snakes, the average change in CO₂ stores, ±S.D., predicted from the transients in R is 4·50 ± 1·72 mmol kg⁻¹. This value represents the difference between the animal’s actual CO₂ elimination and the CO₂ elimination that would have occurred if R had remained at steady state (i.e. 0·75) when temperature was changed. The calculated change in CO₂ stores was dependent upon the absolute change in temperature (J. N. Stinner & R. L. Wardle, unpublished results) but was unaffected by the direction of the temperature change.

R followed the same general pattern in the turtles (Figs 3,4). The change in CO₂ stores predicted by the transients in expired CO₂ is 2·81 ± 0·94 mmol kg⁻¹.

Total of whole blood in 16 snakes was inversely related to body temperature (Fig..5). From 5 to 35°C, fell by more than 30%. Least-squares regression analysis yields (P < 0·01, N = 16, x̄ mass = 167 g, range 128–226 g), where is in mmol l⁻¹ and T_B is in °C. The large decrease in cannot be explained by the amount of hydrogen ions equivalent to the changes in lactate concentration. From 15 to 35°C there was a relatively small (0·75 mmol l⁻¹) rise in blood lactate concentration (Fig. 6). Linear regression yields [lactate] = 0·18 ± 0·0377T_B (P<0·05, N=12, x̄ mass = 162 g, range 128–202 g), where [lactate] is in mmol l⁻¹. Near 5 °C, lactate concentration appears to be higher and exhibits considerably more scatter. Mean ± S.D. for lactate concentration in C. constrictor at 1–7°C is 2·60 ± 1·68 mmol l⁻¹.

Lowering the body temperature of the snakes from 30 to 5 °C produced a gradual rise in blood (Fig. 7). Initially this increase was relatively rapid and corresponded to the time of falling temperature (approx. 3 h). Subsequently, when body temperature was low, there appeared to be a monoexponential rise in until steady state was reached at about 45–50 h. Elevating body temperature from 5 to 30°C produced a very rapid decrease in (Fig. 8). In three snakes, fell to about 63% of its initial value and then increased to steady-state values. This ‘overshoot’ was not evident in the fourth snake in which appears to have decreased in a simple exponential fashion. Approximately 7h was required to reach steady state in the four snakes.

Blood at 15 °C and a of 12·5 mmHg averaged 17·5 ± 2·5 mmol l⁻¹ and blood equilibrated at 35°C and a of 23 mmHg averaged 15·5 ± 2·6 mmol l⁻¹. The average decrease in of 2·0 ± 0·5 mmol 1⁻¹ caused by the 20°C increase in temperature was significant (P<0·01, paired t-test) but is considerably less than the 4·6 mmol l⁻¹ decrease that occurs in vivo. Lactate concentration was significantly lower (P<0·05, paired t-test) at 15°C than at 35°C, averaging 1·44 ± 0·19 and 3·20 ± 0·74 mmol l⁻¹, respectively. In 10 blood samples (five at 15°C and five at 35°C) the ratio averaged 0·57 ± 0·14 and was not significantly different at the two temperatures (P>0·05, paired t-test).

Arterial in C. constrictor increased from about 12·5 mmHg at 15 °C to about 28 mmHg at 40°C (Fig. 9). There was no significant difference between values measured in taped animals and in those handled during blood sampling (ANCO-VAR). The cubic polynomial expression for the combined data is: (P<0·01, N = 20, x̄ mass = 163 g, range 104–260 g). Arterial plasma pH of snakes decreased with rising body temperature, especially above 25°C (Fig. 10). From 15 to 25°C pH decreased by only about 0·0037 units °C⁻¹, but from 30 to 40°C pH declined by approximately 0·0085 units °C⁻¹. As with ⁠, there was no significant difference between pH in taped and handled snakes (ANCOVAR). The quadratic polynomial expression for the combined data is pH = 7·508 + 0·0115T_B − 0·00038T_B² (P<0·01). Plasma bicarbonate concentration was linearly affected by temperature from about 22 mmol l⁻¹ at 15°C to about 16 mmol l⁻¹ at 40°C (Fig. 11). Linear regression yields bicarbonate concentration (mmol l⁻¹) = 25·0 − 0·224T_B (P<0·01). Thus both declining bicarbonate concentration and rising produce the negative temperature dependence of plasma pH.

Systemic arterial was markedly affected by temperature in snakes (Fig. 12). In undisturbed (taped) snakes nearly tripled when temperature was increased by 20°C, rising from roughly 32 mmHg at 15°C to 89 mmHg at 35°C. Least-squares regression analysis yields (P<0·01, N= 11, x̄ mass = 156 g, range 115-200g). Unlike and pH, was significantly affected by handling the snakes (P< 0·001, ANCOVAR). Linear regression of data from handled snakes between 15 and 35°C yields (P<0·01, N = 7, x̄ mass = 168g, range = 104–260 g). Thus, although disturbing the snakes at 15°C had little effect upon ⁠, predicted at 35 °C was 34 mmHg higher in the handled snakes.

The rise in (Fig. 9) and fall in plasma pH (Fig. 10) that occurs in C. constrictor is a now familiar response of reptiles and amphibians to rising body temperature (see Introduction). However, unlike most studies, blood ⁠. was found to be markedly affected by temperature so that both changing bicarbonate concentration and changing produce the thermal dependence of pH. Raising body temperature produced a −0·23 mmol l⁻¹ change in ⁠. per °C (Fig. 5) and at the same time exceeded that predicted from metabolism (i.e. R = 0·75, Fig. 2). Body CO₂ stores are also temperature-dependent in Chrysemys scripta (Figs 3,4). These changes in CO₂ stores are not the result of titrating bicarbonate by hydrogen ions originating from dissociation of lactic acid, as previously suggested (Stinner, 1982), because blood lactate concentration was little affected by temperature in the snakes (Fig. 6). In animals near 5°C blood lactate concentration averaged 2·6 mmol l^−l and in animals close to 35 °C lactate concentration was about 1·5 mmol l⁻¹.

One mechanism partly responsible for changing at least in C. constrictor, is thermally induced shifts of the blood CO₂ dissociation curve. When in vitro blood samples at physiological were warmed by 20°C, fell by roughly 2·0 mmol 1⁻¹. Since blood in vivo falls by about 4·6 mmol 1⁻¹ when the snakes are warmed by 20°C, it is apparent that shifting the CO₂ dissociation curve is responsible for approximately 43% of the total in vivo blood change. The remaining 57% is presumably due to exchange between extracellular fluid and other tissues (e.g. skeletal muscle). It has also been reported that in blood from tortoises (T. graeco) and frogs (R. temporaria) falls by about 0·5 mmol l⁻¹ per °C rise in temperature (Kayser, 1940), which is about five times higher than that in C. constrictor blood. The higher value probably results from the use of the same equilibration gas at the different temperatures.

Elevating C. constrictor’s body temperature by 25°C results in about 4·50 mmol CO₂kg⁻¹ being eliminated in excess of that predicted from an R of 0-75, although blood in vivo falls by 5·75 mmol l⁻¹. The following calculations show that the estimated change in of extracellular fluid and red blood cells is not great enough to account for the excess CO₂ elimination. Interstitial is close to that of plasma ⁠. In blood taken from C. constrictor, the of red blood cells averaged 57% of that in plasma ⁠, which agrees closely with Jesuits from oxygenated human blood (see fig. 4 of Van Slyke & Sendroy, 1928). By substituting _p for in equation 2, plasma can be calculated if blood and haematocrit (100 × F_C) are known. In resting C. constrictor haematocrit is about 23% (Stinner, 1987a) and predicted is 21·75 mmol l⁻¹ at 5°C and 16·0 mmol l⁻¹ at 30°C. Thus, plasma (and interstitial fluid) equals 1·11 × SO that at 5 and 30°C, is 24·1 and 17·8 mmol l⁻¹, respectively. Assuming that the interstitial fluid volume represents 15% of the body mass and blood volume is 6% of the body mass, then warming a 1-kg snake from 5 to 30°C should produce an excess CO₂ elimination of 0·945 mmol from the interstitium and 0·345 mmol from the blood. Thus, less than one-third (29%) of the 4·50 mmol comes from the extracellular fluid and red blood cells. Obviously, of other tissues is inversely affected by temperature. Despite the large effect of temperature upon of tortoise and frog blood in vitro obtained by Kayser (1940), he also concluded that of other tissues must be inversely proportional to body temperature.

changes in tissues other than blood could explain why R does not appear to be at steady state although blood is constant. When C. constrictor’s body temperature was reduced from 30 to 5 °C approximately the same amount of time was required to reach steady-state R and blood values (compare Figs 1 and 7). The gradual rise in and R may be limited by the low metabolic rate and hence slow accumulation of CO₂ at 5°C. Alternatively, the rise in CO₂ could be limited by transmembrane bicarbonate transfer processes, i.e. ion pumps. When temperature was increased back to 30°C, blood achieved steady state in about 7h (Fig. 8). However, R did not appear to reach steady state until approximately 24 h (Fig. 2). This time difference may reflect continued CO₂ elimination from intracellular fluid.

As discussed in the Introduction many investigators have concluded that blood is relatively unaffected by temperature in reptiles and amphibians. We believe that at least four factors have contributed to this view. The first is the use of incorrect values to calculate plasma bicarbonate concentration from the Henderson-Hasselbalch equation (see Introduction). Plasma values have not been determined for C. constrictor but, in a study of Chrysemys picta, Nicol et al. (1983) found close agreement between their measured values and those predicted by the equation of Siggaard-Andersen (1974). Use of this equation to estimate bicarbonate concentration in C. constrictor yields values that agree well with directly measured (compare Figs 5 and 11). The regression equations for bicarbonate concentration and demonstrate about the same thermal dependence (0·224 and 0·230, respectively) and predicted plasma bicarbonate concentration is about 2·2 mmol l⁻¹ higher than whole-blood ⁠. The lower blood is expected because of the much lower in red blood cells compared with plasma (Southworth & Redfield, 1925; Van Slyke & Sendroy, 1928). In C. constrictor blood the plasma was found to be (see above). Since predicted is 19·5 mmol l⁻¹ at 15 °C and 14·9 mmol l⁻¹ at 35 °C, calculated values are 21·6 and 16-5 mmol l⁻¹, respectively. Subtraction of dissolved CO₂ (i.e. ⁠) yields plasma bicarbonate concentrations of 20·9 and 15·8 mmol l⁻¹ which are reasonably close to the values of 21·7 and 17·1 mmol 1⁻¹ predicted from regression analysis of data presented in Fig. 11.

A second factor that can affect the outcome of measurements is the wide variation in normally encountered among different individuals kept at the same temperature (Fig. 5). Thermal effects can easily be obscured by individual differences if is not investigated over a wide temperature range (for example 20°C) in the same animal. Third, it is obvious from Figs 7 and 8 that to make an accurate estimate of the influence of temperature upon sufficient time must be allowed for reaching steady-state conditions. A fourth factor that can influence measurements is struggling by the experimental animals. However, this problem might be especially significant at higher body temperatures. The result would be elevated lactate and reduced bicarbonate concentrations. Hence the thermal dependence of blood would be overestimated. This may be responsible for the larger changes in blood obtained by Kayser (1940) (see Introduction) since blood was collected by heart puncture. Anaerobic metabolism does not appear to be elevated in our study because lactate values are well within those reported for resting reptiles (Bennett & Dawson, 1976).

Systemic arterial ⁠. in C. constrictor rises sharply with increasing body temperature (Fig. 12). The rise in ⁠. is due to a rightward shift of the O₂ dissociation curve in the presence of a right-to-left intracardiac shunt (Wood, 1984; Wood & Hicks, 1985; Stinner, 1987b; Boutilier et al. 1987). The displacement of the O₂ dissociation curve is a direct effect of rising temperature as well as increasing H⁺ concentration and (Bohr effect). It is evident from Fig. 12 that also increases when the snakes are disturbed by handling. This increase is probably due to a rightward shift of the O₂ dissociation curve caused by some metabolic acidosis (J. N. Stinner & R. L. Wardle, unpublished results). However, the decrease in pH is relatively small; pH values in the handled snakes are not significantly different from those in the undisturbed snakes.

The rise in arterial with temperature in C. constrictor (Fig. 9) is not caused by the same mechanism as that for ⁠. Because of the greater slope of the CO₂ dissociation curve compared with the O₂ dissociation curve within the physiological range of gas tensions, even a relatively large intracardiac shunt does not appreciably affect systemic arterial ⁠. Fig. 13 illustrates this point. The three CO₂ dissociation curves were constructed using the formulae given by Weinstein, Ackerman & White (1986) in a study of Chrysemys blood at 15, 25 and 35 °C. Haematocrit, and presumably haemoglobin concentration, are similar in Chrysemys scripta and Coluber constrictor. Hence their non-bicarbonate blood buffering is quantitatively similar. Consequently, if allowance is made for differences in blood ⁠, then CO₂ dissociation curves in the two species should be comparable. Based upon this reasoning, adjustments were made for the lower in C. constrictor by substituting the appropriate in vivo and steady-state values into the formulae of Weinstein et al. (1986) and then solving for the intercepts. Also shown in Fig. 13 are pulmonary venous and systemic arterial values. The pulmonary venous points were estimated using the arterial – venous O₂ content differences measured in C. constrictor (Stinner, 1987a), an R of 0·75, and an assumed right-to-left shunt equivalent to 25% of the systemic blood flow. It can be seen that this relatively large shunt has little effect upon arterial ⁠. At 15 °C systemic arterial is only about 0·5 mmHg higher than in pulmonary venous blood and at 35°C there is a roughly 2-0mmHg difference. Thus, the rightward displacement of the CO₂ dissociation curve in the presence of the shunt produced a 1·5 mmHg rise in ⁠. This is less than 15% of the total increase in that occurs when C. constrictor is warmed from 15 to 35 °C. Systemic arterial is therefore determined primarily by lung ventilation, and is dependent upon the and the position of the CO₂ dissociation curve. In contrast, systemic arterial ⁠. is dependent upon the arterial O₂ content (which is a function of the shunt) and the position of the O₂ dissociation curve (Wood, 1984).

The primary goal of this study was to examine the thermal dependence of total body CO₂ stores in reptiles. In agreement with earlier work, notably by Kayser (1940), we found that CO₂ stores vary inversely with temperature in a snake (C. constrictor) and a turtle (C. scripta). In the snakes about 10% of the change in CO₂ stores can be accounted for by shifts of the blood CO₂ dissociation curve, and 90% of the change represents intracellular adjustments in CO₂. It appears, then, that when temperature increases there is some decrease in extracellular caused by a rightward shift of the CO₂ dissociation curve and there is also a net decrease in within the intracellular compartment that results in further lowering of extracellular ⁠. The decrease in blood ⁠, and thus bicarbonate concentration, along with an increase in ⁠, produce declining pH with rising temperature. Although there is undoubtedly considerable interspecific variation, we believe that total body CO₂ stores in most reptiles and amphibians are. negatively dependent upon body temperature. Hence air-breathing ectotherma must be viewed as open systems with respect to CO₂, i.e. they do not regulate to maintain a constant in the face of changing body temperature. Our findings support the view that the alphastat hypothesis does not explain acid-base regulation in these animals.