Acid–Base Regulation and Blood Gases in the Anuran Amphibian, Bufo Marinus, During Environmental Hypercapnia

Amphibians often encounter environmental hypercapnia as a result of CO2 uptake from the water surrounding their highly permeable skin and/or rudimentary gills (Ultsch, 1976; Heisler, Forcht, Ultsch & Anderson, 1982), but may also become hypercapnic as a result of various innate activities such as diving, exercise or burrowing (Boutilier, Randall, Shelton & Toews, 1979a,b,c; Boutilier, McDonald & Toews, 1980; Boutilier & Toews, 1981; Shelton & Boutilier, 1982; Boutilier & Shelton, 1986). Chronic experimental exposure to elevated environmental CO2 levels generally causes a pronounced fall of arterial plasma pH, which in some species is countered by plasma bicarbonate elevation. Plasma pH compensation in amphibians is far from complete (Toews & Boutilier, 1986) and is certainly lower than in most fish species (Heisler, 1980, 1986b). Indeed, in some amphibians, this type of acid-base regulation is completely lacking (Boutilier, 1981; Heisler et al. 1982; Stiffler, Tufts & Toews, 1983). The relative compensation of plasma pH typically ranges from 0 to 35 % during chronic hypercapnia in amphibians (Toews & Boutilier, 1986), with the one known exception being the Amazonian apodan, Typhlonectes compressicauda, capable of almost 80% compensation (Toews & Macintyre, 1978).

Much of the available information on pH compensation during hypercapnia in amphibians has been collected in the semi-terrestrial toad, Bufo marinus. The degree of extracellular pH compensation during chronic exposure to an environment of 5 % CO₂ is comparatively low (17–0%; Boutilier et al. 1979a), whereas pH in several intracellular body fluid compartments is restored to a much higher degree (Toews & Heisler, 1982). Although these data suggest that pH in intracellular body compartments is better protected against disturbances than the extracellular pH, little information is available as to the source of the intracellular and extracellular bicarbonate increases or on the potential of bicarbonate-equivalent ion transfer processes between the environment and various fluid compartments of the animal.

The urinary bladder plays a key role in generating some of the bicarbonate required for pH compensation, and there is also evidence that mobilization of internal calcareous deposits may contribute to an elevation of the total bicarbonate pool during chronic CO₂ exposure (Tufts & Toews, 1985). Quantification of the contribution of individual mechanisms to the overall compensation is made difficult because of the lack of relevant data. In all of the earlier studies of chronic CO₂ exposure in Bufo mannus, the animals were bathed throughout the experiment in a volume of water approximating to their body mass (about 250–500 ml). As stated by Toews & Heisler (1982) this water must not only have received the excreted end products of nitrogen metabolism, but elimination of H⁺ to or uptake of bicarbonate ions from the environment may have been limited by the availability of adequate amounts of counterions in the water, a factor well-known as being limiting in fish (Heisler, 1986b).

The aim of the present study was to quantify the contribution of transepithelial ion transfer processes to pH compensation, and to evaluate the extent to which the limitation imposed by a small environmental water pool accounts for the much lower degree of extracellular pH compensation in Bufo than generally observed in fishes. This study was also intended to test the suggestion that comparatively incomplete extracellular pH compensation may be related to a bicarbonate ‘threshold’ (Heisler, 1986c) beyond which any further elevation of the bicarbonate concentration is impossible. In addition, we tested the compatibility of arterial blood and O₂ saturation data during hypercapnia with model predictions of O₂ transport in animals with cardiovascular shunts (Wood, 1982).

Specimens of adult Bufo marinus (325–568 g, both sexes) were obtained from Charles D. Sullivan Co. Inc. (Nashville, TN, USA). The animals were held in large terraria at 20°C with free access to water and dry areas for at least 1 month before the experiments. They were treated against internal parasite infestation and infections, and fed regularly on beef liver and mice.

Approximately 48 h prior to an experiment, the animals were anaesthetized by immersion in a 0·75% solution of MS-222 (buffered to pH 7 with NaHCO₃) and intubated intratracheally for ventilation during surgical placement of an indwelling femoral artery cannula (Boutilier et al. 1979a). After surgery the animals were washed with fresh water and artificially ventilated until they regained consciousness.

After recovery from anaesthesia, the animal was placed in a 2-7-litre water recirculation system (Fig. 1). Water was circulated at l-51min^-1 at20±0·5°C. The experimental chamber was supplied with humidified gases at 20 ± 0·05 °C produced by a gas mixing pump (Wösthoff, Bochum, FRG) at a rate of 300 ml min^-1. A second gas mixing pump delivered gases to the fritted glass bottom of the aeration column. Oxygen concentration was maintained at 20 % during normocapnic and hypercapnic conditions.

Water from the recirculation system was delivered continuously by a roller pump (P₁, Fig. 1) to a ‘delta-bicarbonate’ monitoring system (Heisler, 1978, 1984). The water was fed through a series of three glass columns through the sintered glass bottoms of which a gas mixture of 1 % CO₂ in nitrogen was bubbled. The gas was provided by a Wösthoff gas mixing pump specially selected for long-term stability. The entire system was maintained at 30 ± 0·05°C. A temperature different from that of the animal was deliberately chosen to improve electrode stability and response (see Heisler, 1988a). The water was fed over an electrode chain consisting of a pH glass electrode and a double electrolyte bridge Ag/AgCl reference with sleeve diaphragm (Ingold, Frankfurt, FRG) before being returned to the aeration column. The drift of the electrode chain was experimentally determined to be <0-001 pH unit per 24 h. Electrodes were connected to the recording system by way of high-impedance isolation amplifiers (Model 87, Knick, Berlin, FRG) to avoid ground loops and electromagnetic disturbances. The system was grounded by a platinum electrode placed in the water close to the sensing electrode. Changes in the pH of the water thus standardized for temperature and will be the result of changes in bicarbonate concentration. The system was calibrated in absolute terms by the addition of precisely known amounts of HC1 and NaHCO₃ standards.

The ammonia concentration ([NH₄⁺]) of the water was automatically determined at certain intervals (see below) by means of an ammonia gas electrode (HNU Systems Inc., Newton, MA, USA), and a set of roller pumps, solenoid valves, and a microcomputer assembly. By activating the appropriate pumps and solenoid valves either NH4CI standards or water samples from the recirculation system were introduced into the thermostatted (30 ±0·05 °C) electrode chamber (2 ml) and mixed with a magnetic stirrer with about 2% of 10 mol l^-1 NaOH. Stirring was maintained when the pumps (P₂ and P3, Fig. 1) were stopped after appropriate chamber flushing. Alkalinization of the solution to a pH higher than 12 converted ammonium to physically dissolved ammonia, which readily diffused past the hydrophobic membrane, affecting the pH of the inner electrode solution. The output from this single unit pH electrode was connected to an Orion lonalyser (Orion Research, model 901), which was automatically calibrated and set by the microprocessor with two standards bracketing the ammonium concentrations in the recirculation system. The data output was recorded by means of a chart recorder, and the actual concentration values printed out after stabilization of the electrode reading. Data were thus automatically collected with an accuracy of >0·01 mmol 1^-1 NH₄⁺ for periods of several days. The alkalinized water samples were discarded.

The amounts of bicarbonate-equivalent ions transferred between animal and environmental water were calculated as detailed before (Heisler, 1978, 1984, 1986a).

After placing the toad in the experimental chamber, the recirculation system was flushed with fresh water for approximately 12 h. Thereafter, the system was filled with an artificial medium prepared by adding sodium bicarbonate and choline chloride to Gottingen tapwater, bringing the concentrations of Na⁺ and Cl^- to 3·5 mmol 1^-1 and the [HCO₃¹] to 2·5 mmol 1^-1. This ambient water was recirculated and completely exchanged with fresh water of the same ion concentration every 12 h. During an initial 24 h period control measurements of bicarbonate and ammonia release were monitored and the experiment was continued only when the data indicated that the animals had reached steady-state conditions. Two blood samples were withdrawn from each animal towards the end of this control period and analysed for blood gases and pH. The gas supply to the aerial and aquatic portions of the chamber was then switched from air to a gas mixture of 5 % CO₂, 20% O₂ and 75 % N₂. This time was defined as time zero for all subsequent measurements. Blood samples (220μl) and water samples (for ammonia determination) were taken at 15 and 30 min, 1, 2, 4, 8 and 24h. An additional water sample was taken at 16 h. The bicarbonate concentration of the ambient water was monitored continuously. At the end of this second period the toads (hypercapnia maintained) were infused intraarterially (at 0’35 ml min^-1 by a roller pump) with 4·5 mmol kg^-1 body mass of NaHCOj solution in about 8–10 min. Blood and water samples were analysed at the end of the infusion period (new time zero). Blood samples were analysed 15 and 30 min 1, 2, 4 and 24h post-infusion and water samples at 1, 2, 3, 4 and 24h post-infusion.

Blood samples were analysed at the same temperature as the toad for pH, and using appropriate microelectrodes (Radiometer, BMS3). The oxygen and carbon dioxide electrodes were calibrated with humidified gas mixtures (20 ±0·05 °C) produced by a Wosthoff pump, with each measurement being referenced to a gas mixture close to the expected value. The pH electrode was calibrated with Radiometer precision buffer solutions S1500 and S1510, with each sample being referenced to the high pH buffer. A 10μ1 portion of each well-mixed sample was used for haematocrit determination using a micro-haematocrit centrifuge (Compur). Blood used for measurement of gas tensions (175 μl) was suspended in heparinized saline and re-infused into the animal without any apparent haemolytic effects. Nevertheless, haematocrit declined from 26 to 18% over the course of the experiments. Plasma bicarbonate concentrations were calculated by application of the Henderson–Hasselbalch equation using values for αCCT and from the equations derived by Heisler (1984; note: the sign of the last line term of the tr-formula was misprinted, and should read +).

In vitro oxygen dissociation curves were constructed using the mixing method (Haab, Piiper & Rahn, 1960) in which various blood O₂ saturations are prepared for measurement by anaerobically mixing known proportions of oxygenated and deoxygenated whole blood (Scheid & Meyer, 1978). Samples of blood, pooled from four animals, were equilibrated in intermittently rotating glass cuvettes (20 ± 0·5 °C) with gas mixtures containing 1 % or 5 % CO₂ in either air or nitrogen. Two complete curves were constructed from measurements of at prepared saturations of 10, 30, 50, 70 and 90%. Measurements of and pH were made on P₅₀ samples.

Upon exposure to 5 % CO₂ in both the water and the air space of the recirculation system, arterial levels rose from the normocapnic control of 9·3 mmHg to values of 35 and 37·5 mmHg after 1 and 2h, respectively, and gradually approached levels of about 40 mmHg after 24 h (Fig. 2). During the first 4 h of hypercapnia, the average arterial pH (pHa) decreased from 7·907 to 7·397, rising slightly to 7·433 by 24h. Arterial plasma bicarbonate concentrations increased from 25 ·8 mmol 1^-1 at time 0 to approximately 34 mmol 1^-1 within the first 2h of hypercapnic exposure, reaching levels of 36·5 mmol 1^-1 after 24 h.

Exposure to hypercapnia was also accompanied by a marked elevation in arterial from control levels of 42–44 mmHg to values in excess of 80 mmHg after 15 and 30min (Fig. 3). Thereafter levels declined in an exponential fashion, reaching 72 mmHg after 8h and 59 mmHg after 24 h. During the initial respiratory acidosis, an important response of the animal is a considerable CO₂-stimulated hyperventilation (Macintyre & Toews, 1976; Boutilier et al. 1979a), which leads to an increase in and corresponding decrease in the difference between arterial blood and inspired gas (Fig. 4). The elevation of following acute exposure to hypercapnia appears to be a slow process, requiring at least 1-2 h before complete equilibrium is attained. However, following the initial period of CO₂ disequilibrium, it is clear that the animal must continue to hyperventilate, since even after 24 h of exposure, the gradient between inspired and arterial blood is only 60 % of that seen when animals are breathing air (Fig. 4).

An analysis of arterial blood O₂ saturation by plotting in vivo values on O₂ dissociation curves determined in vitro at pH and levels characteristic of control and hypercapnic conditions is shown in Fig. 5. O₂ saturation levels at 4 and 8 h (after CO₂ equilibrium had occurred, see Figs 2 and 4) were slightly higher than those seen prior to CO₂ treatment, whereas after 24 h of CO₂ exposure, the O₂ saturation of arterial blood was not significantly different from control conditions (Fig. 5). Certainly, the analysis assumes two steady curve positions even though there are small deviations in pHa and from each of the curves. However, given the Bohr factor (ΔlogP₅₀/ΔpH) of –0·23, it is equally certain that such deviations would hardly affect the general conclusions of this analysis.

During the normocapnic control period, bicarbonate was released from the animals to the water at a rate of 84 μmol h^-1 kg^-1 body water, which is only about 12 % of the average rate of NH₄⁺ efflux of 662 μmol h^-1 kg^-1 body water, resulting in an H⁺-equivalent excretion (the difference between NH₄⁺ and HCO₃^- release) of 578 ± 66 μmol h^-1 kg^-1 body water (x̄±S.E., Fig. 6). Upon exposure to hypercapnia, the release of bicarbonate stopped and reversed to an initial, statistically significant, but small bicarbonate gain which thereafter subsided at a rate of 2·7 μmol h^-1 kg^-1 body water between hours 1 and 8 of hypercapnia. Over the last 16 h of hypercapnia, the bicarbonate concentration of the water began to rise again at a rate of 137μ mol h⁻¹ kg^-1 body water. NH₄⁺ excretion, however, was increased above the control rate over the entire 24-h period of hypercapnia (771 μmolh^-1 kg^-1 body water. This pattern of bicarbonate and NH₄⁺ release resulted in a net (hypercapnic minus control) H⁺-equivalent excretion of 3·24 mmol kg^- body water over the 24-h hypercapnic period, equivalent to a net gain of bicarbonate from the environment.

When the acid-base status had reached a quasi steady-state after 24 h of hypercapnia (Fig. 2; Boutilier et alToews & Heisler, 1982), the animals were subjected to an intra-arterial infusion of 4·5 mmol kg^-1 body water of NaHCO₃, resulting in transient rises of arterial pH, and bicarbonate concentration (Fig. 7). Analyses of the changes in bicarbonate and NH₄⁺ content in the environmental water showed that about 90% of the infused bicarbonate (4·0 mmol kg^-1 body water) was excreted by the animal within the first 2h post-infusion (Fig. 7). Although arterial pH and the plasma bicarbonate concentrations decreased following NaHCO₃ infusion, both were slightly but significantly elevated above the pre-infusion values 24 h post-infusion (Fig. 7). Bicarbonate infusion had no effect on arterial levels which ranged from 57·8 ±4·9 to 63·4 ±4·6 mmHg (x ± S.E.) at the sample periods shown in Fig. 7.

Exposure of Bufo marinus to elevated levels of environmental CO₂ lèads to a progressive, albeit relatively slow, elevation in arterial values, which attain higher than environmental levels only after 1–2 h of hypercapnia (Figs 2, 4). This initial 1-2 h of net CO₂ uptake by the animal represents the time required for CO₂ equilibration of the various tissue compartments (Toews & Heisler, 1982). The marked increase in ventilation which occurs upon exposure of Bufo marinus to elevated ambient CO₂ levels (Macintyre & Toews, 1976) is almost certainly responsible for the considerable reduction in the difference between arterial and inspired during continued CO₂ exposure (Fig. 4; Boutilier et al. 1979a), which probably also applies to the large rise in during hypercapnia. An alternative, though not mutually exclusive, explanation of the rise in during hypercapnia can be based on existing models of O₂ transport in animals with cardiovascular shunts (Wood, 1982; Wood & Hicks, 1985; Wood, Glass, Andersen & Heisler, 1987), or a redistribution of R–L shunt fractions to the systemic cardiac outflow vessels (see Ishimatsu, Hicks & Heisler, 1987).

In comparison with fish and mammals, the comparatively low arterial saturation in amphibians (see normocapnic curve, Fig. 5) can be attributed to systemic venous admixture with pulmonary venous blood (right to left cardiac shunt) within the undivided ventricle. Under these physical conditions, the of a mixture of pulmonary and mixed venous blood is dependent upon the resulting O₂-saturation (Wood, 1982), the principle being the same as for the in vitro mixing method for O₂ curve construction (Haab et al. 1960; Scheid & Meyer, 1978). is thus a dependent variable of O₂ saturation (Wood, 1982) and increases in arterial oxygen tensions during hypercapnia could be interpreted as the result of the pH-induced decrease in haemoglobin O₂-affinity. Hyperventilation, however, could play an important role in elevating the of the pulmonary venous return, and therefore the Pao₂, by providing a more saturated precursor for intraventricular mixing. The relationships are complex, however, as the carbon dioxide contents, and therefore pH values, of the various possible bloodstreams (see Ishimatsu et al. 1987; G. M. Malvin & N. Heisler, in preparation) may also change in a dynamic fashion. Regardless of the underlying mechanism, the end result is that the levels of arterial O₂ saturation during a condition of respiratory acidosis were maintained at, or slightly above, those seen in control animals breathing air. Furthermore, it is evident that there is a reciprocal interaction between the two respiratory gases; the gradual increase in and corresponding decrease in as hypercapnia continues (Fig. 4) are probably the result of diminished ventilation. The latter is well-established for chronic hypercapnic states in mammals where adjustments in cerebrospinal fluid pH, through increases in cerebrospinal fluid bicarbonate concentrations, cause some decrease in ventilatory drive with time.

The reduction in the inspired–arterial difference during hypercapnia (Fig. 4) can be viewed as an initial compensatory response to offset further decreases in arterial pH. Though evidence is lacking as to a direct response of the animals to elevated CO₂ levels per se, or to a CO₂-mediated change in arterial oxygenation (Fig. 5), the common effect of both mechanisms is a reduction of the hypercapnia-induced change in plasma pH. The initial increase of plasma bicarbonate within the first hour of CO2 exposure can, at least in part, be attributed to physicochemical action of the extracellular non-bicarbonate buffers. The slope of the apparent extracellular buffer line (β = Δ[HCO₃^-]/ –ΔpH) during this period has previously been found to vary between 6·2 (Boutilier et al. 1979a) and 14mequivpHunit^-11^-1 (Toews & Heisler, 1982). The apparent buffer slope determined in the course of the present study was not significantly different from that of true plasma (20·4mequiv pH unit^-11^-1) when blood samples were equilibrated in vitro (Boutilier et al. 1979a). It is difficult to judge to what extent the elevated levels of counterions and larger volume of the ambient water in the present study were responsible for the comparatively greater degree of initial compensatory accumulation of plasma bicarbonate (Fig. 8).

Some indirect evidence suggests that improved reabsorption of bicarbonate is partially responsible for the improved early pH compensation. Tufts & Toews (1985) described a marked reduction of the initial apparent buffer slope (1·9 mmol pH unit^-11^-1) in bladder-bypassed toads, indicating the key role of net bicarbonate reabsorption through the bladder wall. Indeed it has been shown that the apparent in vivo buffer slope for intact animals without access to water (6·4 mmol pH unit^-11^-1) was not significantly different from that observed for their hydrated counterparts (Boutilier et al. 1979b). Further evidence for the acid-base-relevant potential of the toad bladder comes from in vitro studies on hemibladder preparations, showing increased rates of H⁺ and NH4⁺ excretion during respiratory and metabolic acidoses (Frazier & Vanatta, 1971, 1973). The skin of amphibians is also a well-known exchanger of acid-base-relevant ions (see Evans, 1986, for a review). Taken together, these data suggest that mechanisms additional to passive blood buffering, such as transfer of bicarbonate-equivalent ions between the environment and/or various body fluid compartments, may influence the initial in vivo buffer slope to a considerable extent. Although an activity-related metabolic acidosis could, in principle, explain the lower in vivo buffer slopes described in earlier studies, the increased potential for cutaneous net bicarbonate uptake due to higher sodium and bicarbonate levels in the ambient water of the present study certainly facilitated an earlier rise in plasma bicarbonate (Fig. 8).

The differences between the present study and others utilizing the same degree of hypercapnia (5 % CO₂) are largely due to an earlier contribution by active compensatory mechanisms in the present study (Fig. 8). However, the overall degree of extracellular pH compensation at the end of 24 h of hypercapnia (about 30%) was not appreciably different from that observed by Toews &,Heisler (1982). A similar degree of compensation was achieved in dehydrated toads (about 28%), which was slightly more than in their hydrated counterparts (Fig. 8). This phenomenon is not easily explained, particularly when one considers that the dehydrated animals experienced a more severe acidosis (Boutilier et al. 1979b). There can be little doubt that the bladder plays a dominant role in the overall correction of such acid-base disturbances, since elimination of the bladder as an exchange surface prevents any compensation from occurring in otherwise intact and freely moving animals (Tufts & Toews, 1985). The latter data also imply a relatively low potential for acid-base compensation by the skin and kidneys.

In contrast to fish, a general trend of limited pH compensation during hypercapnia in air-breathing lower vertebrates is apparent (Heisler, 1986c). Since air-breathing in lower vertebrates is usually coupled with higher levels of extracellular and HCO₃^- for a given pH (Robin, Bromberg & Cross, 1969), the limited compensation in these animals may reflect an inability to accumulate bicarbonate to the extremely high levels that would be required for complete pH restoration during hypercapnia. This limitation does not hold for water-breathing fish with their considerably lower levels of plasma bicarbonate. Their potential for further bicarbonate elevation during hypercapnia, without causing intolerable disturbances in electrolyte balance (i.e. too low [Cl^-]), must accordingly be much greater than that of bimodally breathing vertebrates (Heisler, 1986c). The plasma bicarbonate concentration in Bufomarinus does not become elevated beyond about 30 mmol 1^-1, irrespective of the levels of hypercapnic exposure (D. P. Toews, personal communication). This maximal level of bicarbonate or ‘bicarbonate threshold’ appears to be class-specific, ranging from 22–27 mmol 1^-1 in fish to 22-36 mmol 1^-1 in amphibians and 40-50 mmol 1^-1 in higher vertebrates (Heisler, 1986c).

The fact that amphibians are less capable of increasing plasma bicarbonate during hypercapnia may, however, not be related to the constraints of the electrolyte homeostasis or of electrophysiological mechanisms, but may simply be an expression of the inability to gain sufficient amounts of bicarbonate-equivalents from the environment. This study has shown that the overall pH compensation during chronic hypercapnia could not be improved by external intervention, at least in the short term. The higher environmental bicarbonate concentration in the present experiments had little effect on the time taken to compensate as compared to other studies (Fig. 8). Bicarbonate directly infused into the bloodstream increased the plasma concentration only transiently, before being almost completely eliminated to the environment (Fig. 7). Both of these findings are quite consistent with the idea of a ‘bicarbonate threshold’. These data, together with comparable results from similar experiments on the aquatic salamander Siren lacertina (Heisler et al. 1982), the lacertilian reptile Tupinambis nigropunctatus (Glass & Heisler, 1986) and the teleost fish Cyprinus carpio (Claiborne & Heisler, 1986), supply increasing evidence for a ‘bicarbonate threshold’ as a major limiting factor for pH compensation during hypercapnia in lower vertebrates.

The main objectives of this study were to quantify the net bicarbonate transfer between animal and environment as a regulatory measure of the acid-base homeostatic system (Fig. 6), to differentiate between ammonia and non-ammonia transfer mechanisms, and to relate the amount gained from the environment to the total rise in the body bicarbonate pool. As suggested earlier, the increases in extracellular and intracellular bicarbonate concentrations during hypercapnia in Bufo are probably larger than can be expected on the basis of CO₂-buffering by intracorporeal non-bicarbonate buffers (Toews & Heisler, 1982). This idea has been pursued by estimating the net changes in bicarbonate pools, the bicarbonate produced by non-bicarbonate buffering, and the bicarbonate transferred by ionic mechanisms between animal and environment (for details of the approach see Heisler, 1982, 1986a, 1987a, b).

This estimate is based on the assumptions that the intracellular compartment of white muscle represents the total intracellular space, that the apparent buffer slope for skeletal muscle of Bufo marinus, determined during the initial stages of CO₂ exposure (Toews & Heisler, 1982), is representative of the intracellular non-bicarbonate buffer slope, and that about 50% of the rise in intracellular bicarbonate can be attributed to intracellular non-bicarbonate buffering of CO₂ (Table 1 ). On the basis of this estimate, the change in the intracellular bicarbonate pool is about 4-2 mmol kg^-1 body water. The low apparent buffer value for skeletal muscle (Toews & Heisler, 1982) as compared with other animals (see Heisler, 1982, 1986a) leads us to view this estimate with caution. The non-bicarbonate buffer values (β_NB) determined for several fish species, for instance, are of the order of 40–60 mmol pH unit^-11^-1 (Heisler, 1980, 1982, 1986a,c), substantially higher than the apparent in vivo buffer slope of toad skeletal muscle after 24 h of hypercapnic exposure (33 mmol pH unit^-11^-1 ; Toews & Heisler, 1982). Thus, if the non-bicarbonate buffer value of toad skeletal muscle was of the same order as those found in fish, a considerable quantity of the bicarbonate produced through non-bicarbonate buffering of CO₂ in toad skeletal muscle would have had to have been transferred to other body fluid compartments. In fact, when the animal is modelled as a two-compartment system (ICS and ECS) surrounded by a closed third compartment (the environmental water), the model predicts that the initial and longer term increases in the extracellular bicarbonate pool should occur at the expense of that of the intracellular compartment.

From the balance sheet in Table 1, the changes in the total bicarbonate pool (about 7·8 mmol kg^-1 body water) are estimated to be greater than that which can be attributed to non-bicarbonate buffering plus uptake from the environment (6’4 mmol kg^-1 body water). Tufts & Toews (1985) suggested that the increases in calcium concentrations in urine and plasma of hypercapnic Bufo marinus represented dissolution of calcium carbonate deposits and estimated that approximately 1 mmol kg^-1 animal of bicarbonate would be generated by this process over the 24 h period of hypercapnia. This factor, together with the assumption of more realistic intracellular buffer values (see above), would make up for all of the difference in the balance sheet in Table 1.

The main limitation of the balance sheet approach displayed in Table 1 is the limited information about the intracellular tissue compartments. A more reliable approach comes from the directly measured ionic transfer, the bicarbonate concentration changes, and the amount of bicarbonate produced in the extracellular space by non-bicarbonate buffering (see Heisler, 1984, 1986a, 1987a,b) and indicates (Fig. 9) that the main part of the rise in extracellular bicarbonate level is due to transepithelial ion transfer. The overall transmembrane transfer is limited to transient shifts of smaller magnitude, and any contribution of bicarbonate originating from the environment is progressively reduced from about 8h of hypercapnia towards the end of the experiment. The maximal amount transferred into the intracellular fluid space is similar to that observed in fishes (Heisler, Weitz & Weitz, 1976; Toews, Holeton & Heisler, 1983), but the compensatory effect has to be considered to be much lower as a result of the several-fold higher intracellular control bicarbonate concentration in Bufo.

The net uptake of bicarbonate-equivalent ions plays an accordingly important role in the limited extracellular pH compensation brought about by exposure to elevated levels of ambient CO₂ (Fig. 9; Table 1). The present experiments have indicated, however, that the incomplete compensation of arterial pH in Bufo is not appreciably limited by the availability of bicarbonate or ion exchange counterions in the surrounding water, but is rather attributable to a constitutional inability to accumulate bicarbonate above a certain threshold value.

The authors gratefully acknowledge the skilful technical assistance of Mr G. Forcht and Mrs S. Glage. Supported by Deutsche Forschungsgemeinschaft.