Haemolymph Acid-Base, Electrolyte and Gas Status During Sustained Voluntary Activity in the Land Hermit Crab (Coenobita Compressus)

The effect of a short burst of exhausting activity on haemolymph acid-base balance and gas transport has been studied in decapod crustaceans (McMahon, McDonald & Wood, 1979; Smatresk, Preslar & Cameron, 1979; Wood & Randall, 1981a,b). The accompanying acidosis is frequently of mixed respiratory and metabolic origin requiring several hours for recovery. Sustained submaximal exercise (swimming for 1 h) has been studied in the blue crab Callinectes sapidus (Booth, McMahon, de Fur & Wilkes, 1984). These authors reported a similar acid-base response [i.e. increases in CO₂ tension and lactate, decrease in total CO₂]-This was most pronounced after 15 min but recovered partially within lh, despite the continued accumulation of lactate because of the transfer of acidic equivalents to the ambient sea water via branchial electroneutral ion exchanges. This continued during recovery to restore pH fully.

Mechanisms employed to maintain O₂ transport and delivery also differ between short-burst and sustained submaximal exercise. During short periods of enforced pedestrian activity, O₂ uptake increases through compensations in ventilation ⁠, gill and tissue perfusion ⁠, mean pressure gradient for diffusion (ΔP_g) and O₂ conductance (⁠ - see review articles by Herreid, 1981; McMahon, 1981; McMahon & Wilkens, 1983). The respiratory pigment haemocyanin (HCy) becomes increasingly involved in O₂ delivery. Postbranchial decreases from saturated values onto the descending portion of the dissociation curve, releasing O₂ from the venous reserve. Nonetheless, a shortfall in O₂ delivery results in a shift towards anaerobic metabolism. It was originally thought that the Bohr shift was the major effector of HCy O₂-binding affinity during exercise. By contrast, sustained swimming in blue crabs is fuelled predominantly aerobically and internal remains high (Booth, McMahon & Pinder, 1982). Furthermore, O₂ delivery is increased through cardiovascular adaptations rather than a change in the role of the pigment since the Bohr effect is minimized by a lactate-induced increase in O₂-binding affinity.

We recently discovered the capability for sustained submaximal walking (up to 3h) in a tropical land hermit crab Coenobita compressus (H. Milne Edwards) and developed a rotating respirometer to characterize respiratory gas exchange (Wheatly, McMahon, Burggren & Pinder, 1985). In the present study we examined haemolymph acid—base and electrolyte status and gas transport during voluntary activity. Hermit crabs transport a substantial water reservoir inside the molluscan shell which they will replenish from an external source (Wheatly, Burggren & McMahon, 1984). By sampling this ‘replacement water’ we were able to determine acidic equivalent exchange, known to be an effective mechanism of acid-base regulation in aquatic species (see reviews by Heisler, 1984; Cameron, 1986).

Land hermit crabs (Coenobita compressus 10·5 ± 1·1g, N=29) were collected and maintained in Panama as described previously (Wheatly et al. 1985). Postdesignated a) and prebranchial (v) haemolymph were sampled for pH, O₂ and CO₂ tension (⁠ and ⁠) and concentrations of lactate and inorganic ions at rest, after 50m (150min at 0·6cms^-1) of voluntary walking in a rotating respirometer or following 1 h of recovery. After removing crabs from their shells, postbranchial haemolymph was sampled from the pericardium via a hole made, in the carapace which had previously been sealed with a latex septum, and prebranchial haemolymph was sampled from the infrabranchial sinus at the base of a walking leg (see McMahon & Burggren, 1979). Only one sample was removed from each animal, constituting an unpaired experimental design. Acid-base exchange with external sea water was assessed in inactive crabs and following voluntary exercise after the introduction of 20 ml of 10% sea water into the bottom of the respirometer. In a parallel series of experiments, crabs were removed from their shells to assess the effects of loading. To assist in interpreting blood gas data, haemolymph CO₂- and O₂-binding characteristics were assessed in vitro on haemolymph removed from 60 crabs which had been transported back to Calgary.

Haemolymph (300μl) was analysed immediately for pH, and using Instrumentation Laboratory electrodes (20982—pH, 20984—02 and 20983—CO₂) thermostatted to 30 ± 2°C and connected to an IL 213 blood gas analyser. The pH electrode was calibrated with Radiometer precision buffers (S1500 and S1510). The CO₂ electrode was calibrated with 3 % CO₂ in N₂ to an arbitrary value of 70 using scale expansion. The calibration gas was then serially diluted twice with air (1 % and 0·3 % CO₂ approximately) and the meter reading regressed against The precise composition of these calibration gases was calculated from measured ⁠.

Prebranchial haemolymph (80 μl) was deproteinized in 400 μl of ice-cold perchloric (10 % w/v) and analysed within 2 weeks for [lactate] using a Sigma diagnostic kit (826 UV). In some species, Cu²⁺ liberated from HCy can artificially elevate values unless a modified procedure is adopted (see Graham, Mangum, Terwilliger & Terwilliger, 1983). Remaining haemolymph was frozen and subsequently analysed in Calgary for major inorganic ions using methods outlined in Wheatly et al. (1984).

The net flux of acidic (i.e. H⁺ or NH₄⁺) or basic (OH^- or HCO₃^-) equivalents between the crab and replaced sea water was determined over a 1-h period of rest or recovery from exercise using procedures outlined by McDonald & Wood (1981). Titratable alkalinity [TA—HCO3^-] was determined by titrating 5 ml of water with 0·02 mol I^-1 HC1 to an end pH of 4·0 as determined on a Radiometer GK2402C combination electrode coupled to a PHM 84 pH meter. Theoretically this technique titrates any buffers with pK values between 3 and 9, which includes organic bases (see Heisler, 1984). These, however, are produced primarily at the antennal gland. It is unlikely that the water replaced in the present study comes into contact with the third antennal segment and so there is less possibility of contamination than in an uncatheterized aquatic species. Ammonia [NH3-I-NH4⁺] was determined by the phenolhypochlorite method (Solorzano, 1969). Changes in [TA—HCO3^-] and [NH₃+NH4⁺] were converted to fluxes and combined (signs considered) to give net acidic equivalent flux. This cannot distinguish between net acid excretion and base uptake..

Haemolymph O₂ and CO₂ equilibrium curves were constructed in vitro in Calgary and subsequently used to interpolate O₂ and CO₂ contents (⁠ and respectively) from in vivo measured tensions. CO₂ equilibrium curves and non-bicarbonate buffer curves were determined at 30°C on haemolymph pooled from eight animals using techniques outlined in Truchot (1976) and Randall & Wood (1981). Declotted haemolymph (2·0ml) was equilibrated with humidified gases ⁠; obtained via Wösthoff mixing pumps from analysed mixtures. After 45 min, pH (50-μl samples) and (40-μl samples) were measured with a Radiometer (G299A) liquid-junction capillary electrode connected to a BMS 3 MK 2 Blood Micro System and a Corning 965 CO₂ analyser, respectively.

A conventional gasometric technique was used to determine O₂-combining characteristics of pooled haemolymph (see Wheatly & McMahon, 1982). After equilibration with up to 70 Torr, measurements were made of ⁠, using a Lex-O₂ Con automatic analyser (Lexington Inst. - 80-μl samples), and ⁠, using a Radiometer O₂ electrode and acid-base analyser (E5047; PHM 71/72 - 200-μl samples). CO₂ was used as an exogenous buffer (⁠⁠, 13·3, 26·5 Torr) to study binding over the pH range 7·2-7·7. Haemolymph O₂-carrying capacity was determined on an air-equilibrated sample. Lactate sensitivity was investigated using methodology outlined by Booth et al. (1982). A pooled haemolymph sample was divided into three aliquots and [lactate] adjusted to 3·3, 6·5 or 12·9mmol1^-1 using 0·2 mol 1^-1 neutral (Li⁺) - lithium lactate (the effect of Li⁺ was assumed to be negligible). Oxygen dissociation curves were constructed at equilibration of 2·7 or 26·5 Torr.

All data are expressed as mean±S.E. (number of observations). Samples were tested for homogeneity of variance (F-test) and means compared by Student’s twotailed i-test (unpaired variates) using P=0·05 as the confidence limit. Linear regression was performed by the least-squares method. Slopes from different treatments were compared using an analysis of covariance. In cases where regression lines were parallel, coincidence was tested using the Newman-Keuls multiple comparison procedure.

CO₂-combining characteristics of oxygenated and deoxygenated haemolymph revealed a Haldane effect. Deoxygenated haemolymph contained more CO₂ at a given ⁠, except at low values where the two lines converged (Fig. 1A). Correspondingly, the pH at constant was consistently higher in deoxygenated haemolymph (Fig. IB). The slopes were not significantly different but the elevations were (0·01 >P> 0·005) suggesting that the lines were not coincident even though they did converge around pH 7·92.

Bicarbonate concentration (effectively [HCO₃^-- +CO₃^2-]) was calculated from these data as (using a CO₂ solubility coefficient, αCO₂, of 0·0352 mmol I^-1 Torr^-1) and plotted versus pH (Fig. 1C). The slopes of these non- bicarbonate buffer lines were not significantly different but the elevations were (0·01 >P> 0·005). The combined non-bicarbonate buffer value (β) was 23 mequiv 1^-1 pH unit^-1 (slykes) over the physiological pH range.

Using in vitro CO₂-binding data, [HCO₃^- + CO₃^2-] was calculated from measured values by rearrangement of the Henderson-Hasselbalch equation using and αCO₂ values of 6·04 and 0·0352 mmol 1^-1 Torr^-1, respectively.

Inactive haemolymph acid-base status in crabs carrying shells (Table 1) was typical of other decapods (McMahon & Wilkens, 1983). Interpolating mean and (2·5 and 2·7 Torr) onto the oxygenated and deoxygenated CO₂ dissociation curves (Fig. 1A) provided for the theoretical release of 1·9mmol CO₂1^-1via the Haldane effect, assuming that postbranchial blood was air-equilibrated and became completely deoxygenated at the tissues. The existing gradient, however, was only 4 Torr (see Table 3 below), reducing the magnitude of the Haldane effect to 0·04 mmol CO₂1^-1. Haemolymph from crabs without shells was significantly acido-tic (by 0·25 pH units) with half the control levels of circulating bicarbonate (Table 1) ; this effect may be attributable to trauma (see Wheatly et al. 1985).

Sustained exercise caused significant changes in acid-base status (Table 1). pH decreased on average by 0·4units; [HCO₃⁻+CO₃²⁻] and [lactate] increased by 9-, 3- and 3- to 4-fold, respectively. Post/prebranchial differences in pH, and [HCO₃^- + CO₃^2-], which were originally —0·05pH units, +0·2Torr and +0·3 mequiv I^-1, increased to —0·07 pH units, +4·8 Torr and +1·1 mequiv 1^-1 after exercise. When and values were interpolated onto the in vitro CO₂ dissociation curves, the functional Haldane effect increased to 0·3 mmol CO₂1^-1. Unloaded crabs accumulated less lactate but otherwise experienced a comparable acid-base disturbance.

Acid-base balance was re-established within 1 h after exercise irrespective of access to external sea water at this time (Table 1). Although remained significantly elevated it had substantially recovered from exercised levels. levels were significantly lower in crabs recovering without water. Changes in postbranchial acid-base’status are summarized graphically in Fig. 2. The non-bicarbonate buffering line from Fig. 1C is superimposed. The exercise acidosis is due largely to an increase in respiratory CO₂ which is partially offset by accumulation of metabolic HCO₃^-. Changes in the concentration of metabolic H⁺ buffered in the haemolymph (ΔH⁺m) were calculated according to McDonald, McMahon & Wood (1979) and correlated with changes in [lactate]. An H⁺ deficit of 10mequiv1^-1 was apparent at the end of exercise (i.e. Δlactate> ΔH⁺m). During recovery, [lactate] decreased more rapidly, leaving an excess of metabolic H⁺.

Although inactive hermit crabs exchange their branchial/shell water with an external source, there is no net exchange of acidic equivalents since an ammonia efflux of 440 ± 100μequivkg^-1 h^-1 is countered by an approximately equivalent uptake of titratable acidic equivalents (Fig. 3). Following exercise, however, this was a major route for H⁺ excretion at rates of 4800 ± 500 μequivkg^-1 h^-1 in crabs carrying shells (2800 ± 350μequivkg^-1 h^-1 without shells). In both cases, 80% of the acidic equivalents were excreted in a titratable form. Ammonia excretion doubled in crabs carrying shells.

Electrolyte levels were significantly higher in inactive crabs removed from their shells (Table 2). In animals transporting shells, [Na⁺] was just elevated after exercise (P = 0·05) but all other levels were unchanged. This had the effect of raising the [Na⁺] : [Cl^-] ratio from 0·75 to 0·85. Control values were re-established rapidly during recovery if water was available; otherwise levels rose further. Since [Na⁺] and [Cl^-] were both now significantly elevated, the ratio between the two was restored.

The of Coenobita haemolymph was 1·51 ±0·07mmoll^-1(N = 11); of this, saturated HCy transported 1·37 ± 0·07 mmol I^-1. A Bohr shift (ΔlogPjo/ ΔpH) of -0·44 was determined (Fig. 4). However, contrary to findings in other species (e.g. Booth et at. 1982; Mangum, 1983a), physiological lactate levels (3—13 mmol I^-1) had no detectable effect on binding affinity either at low or high (Fig. 5). The P₅₀ of the combined data was 12 Torr at pH 7·7 and 19Torr at pH 7·2, corresponding closely with the Bohr plot of Fig. 4.

The in vitro equilibrium curves were used to interpolate HCy-bound O₂ concentrations from and pH measurements made in the field (Table 3; Fig. 6). At mean haemolymph pH (=7·8), the HCy was saturated at 28 Torr. Post- and prebranchial HCy were 61·8 ±7·9% (8) and 37·8 ±12·3% (8) saturated, respectively (Fig. 6A). was 0·329 mmol I^-1, accounting for 98% of the total O₂ delivery. After exercise and both increased, though not significantly. A rightward shift in the O₂ dissociation curve increased from 0·333 to 0·503 mmol I^-1 by a depression in prebranchial saturation to 29·3 ± 11·4% (8).

Acid—base recovery following exercise was the same with or without access to water. was significantly elevated above the exercised level. increased significantly to 26 Torr. Since circulating pH was now restored, these values fell around the shoulder of the O₂ dissociation curve [95·0 ± 7·3 % (4) and 84·0 ± 6·3% (6) saturated, respectively]. was half the pre-exercised level (0·124 mmol I^-1) and the venous O₂ reserve doubled [1·183 ±0·087 (4) mmol 1^-1versus 0·530 ± 0·170 (6) mmol 1^-1].

Crabs removed from their shells (Table 3; Fig. 6B) remained acidotic throughout the experimental procedure. Pre-exercised crabs had a higher mean than animals with shells; correspondingly HCy was saturated [84·0± 10·9% (3)]. Nonetheless, the gradient across the gill and O₂ content difference were approximately the same. During exercise, tensions and contents both decreased, releasing a similar quantity of O₂ (0·593mmol 1^-1) though this time originating from incursion into the venous reserve rather than via a Bohr shift in binding affinity.

The Haldane effect has been described in a number of crustacean and molluscan haemocyanins (Brix, Lykkeboe & Johansen, 1981; Truchot, 1983) and has been quantitatively assessed in relation to the (-carrying capacity of the blood. In Coenobita, under physiological conditions, 1·22 mmol CO₂ are released for each mmolO₂ bound - This value is double that reported in other decapods (e.g. Truchot, 1976; Randall & Wood, 1981; Booth et al. 1982) and is closer to values in cephalopod molluscs (Brix et al. 1981), which may relate to the elevated protein concentration. Since the Haldane coefficient represents the amount of protons liberated during oxygenation, it should theoretically correspond to the Bohr effect, which reflects the quantity of oxygen liberated by proton binding. The Bohr factor (Δlog P₅₀/ΔpH), however, was only -0·44. When non-correspondence has been previously reported (e.g. Truchot, 1976; Brix et al. 1981) the Bohr factor is generally larger, which can be explained if CO₂ is not the only source of protons. The only explanation we can offer in the present case is a non-linear relationship between O₂ saturation and proton release; this effect has been substantiated in human haemoglobin. Values for β and are greater than those found in the related species C. clypeatus (McMahon & Burggren, 1979), making them the highest so far reported in decapods (McMahon & Wilkens, 1983). The correlation between these two indices agrees well with equations determined by Truchot (1976) and Wood & Randall (1981b).

Inactive (Table 1) is atypically low for a terrestrial decapod (see table 1 of Cameron, 1986) but resembles levels recorded in air in the intertidal crab, Carcinus (Taylor & Butler, 1978). Elimination of CO₂via the water reservoir may explain this as well as the low apparent respiratory quotient previously measured in the gas phase (Wheatly et al. 1985). For shell water to serve as a CO₂ sink, it requires consistent convection which is achieved in the land crab Cardisoma by the action of the flabellae, gill movements and pulsations of the branchial chamber wall (Wood & Randall, 1981a).

Hermit crabs experienced similar changes in haemolymph pH and after 3 h of voluntary activity to those seen in other land crabs after a 10-min burst of high-speed activity (Smatresk et al. 1979; Wood & Randall, 1981b). The inability of exercising crabs to excrete CO₂ may be associated with loss of shell water which was noticeably spilled inside the respirometer. The acidosis was primarily attributable to respiratory hypercapnia (Fig. 2). [Lactate] increased during the O₂ deficit early in the exercise period (Table 1) but did not accumulate progressively (M. G. Wheatly, unpublished observations). This suggests that exercise was fuelled aerobically, which correlates with the short lag time in kinetics, small O₂ deficit and a steady-state capable of fulfilling the total energy requirement (see Wheatly et al. 1985), characteristics which are diagnostic of sustained submaximal locomotion in homéothermie vertebrates and insects (e.g. Cerretelli, Pendergast, Paganelli & Rennie, 1979). The extended time course in the present study enabled the initial hypercapnia to be partially compensated by metabolic bicarbonate accumulation (cf. Cameron, 1978).

A similar H⁺ deficit was found after sustained swimming in Callinectes (Booth et al. 1984) and strenuous activity in other crabs (McDonald et al. 1979; Wood & Randall, 1981b). It has been variously attributed to differential efflux rates of lactate and protons from exercising muscle, buffering of H⁺ by CaCO₃ stores, and excretion of protons into the external medium. Intracellular sequestration of H⁺ was not examined in the present study and circulating Ca²⁺ remained constant (Table 2), which eliminates the first two possibilities. Proton excretion, however (Fig. 3), could account for the H⁺ deficit, assuming that it occurs at the same rate during exercise as measured in recovery. Based on a circulating volume of 30% body weight, a net excretion of 33 μequiv during exercise would account for the measured H⁺ deficit. This is an efflux rate of 1257 μequivkg^-1 h^-1, which is a quarter of the rate measured during recovery. The reduced efflux rate may arise from water spillage.

The increased ammonia efflux during recovery (Fig. 3) could indicate some impairment during exercise or that proton/amino acid reserves were utilized as metabolic substrate. Additionally, it may reflect changes in NH₃ or NH₄⁺ gradients, although circulating levels were not measured in this study. However, since ammonia only accounted for 30 % of total acidic equivalent efflux, other counterions were predominantly exchanged. Branchial electroneutral ion exchanges (i.e. Na⁺/H⁺ or NH₄⁺; C1^-/HCO₃^- or OH^-) have now been well documented in ion and acid-base balance in crustaceans (Cameron, 1986).

Considering that most exercise regimes require 6-8 h for recovery of haemolymph acid—base balance, [lactate] and (McMahon, 1981), it was surprising to discover re-establishment of settled levels within 1 h in Coenobita. This may relate to the voluntary nature of the activity, which did not drive animals to any physiological limit, as well as the elevated experimental temperature. Since [lactate] is not typically excreted in decapods and we found no evidence for oxidation (⁠ not elevated in recovery, Wheatly et al. 1985), one must infer that it was removed by gluconeogenesis. This was the conclusion of a recent study on Uca by Full & Herreid (1984).

Comparison of electrolyte levels in protected and exposed hermit crabs (Table 2) confirmed the role of the adopted molluscan shell in preventing dehydration from the soft, moist pagurid abdomen (Reese, 1969). A thickening of the abdominal cuticle apparently occurs when glaucothoea of the closely related monospecific coenobitid genus Birgus dispense with the ancestral behaviour of shell dwelling or when hermit crabs are reared for extended periods without shells. [Na⁺]: [Cl^-] ratios around 0·75 are typical in other decapods acclimated to hyposaline media (Mantel & Farmer, 1983). Although exercise caused crabs to spill shell water, the resulting increase in [Na⁺] (Table 2) may not be a simple consequence of haemoconcentration since [Cl^-] remained at pre-exercised levels, indicating an ability for differential ion regulation, perhaps by transcellular Cl^-/lactate exchange (Jackson & Ultsch, 1982). Since [Ca²⁺] remained constant, there was no evidence for mobilization of exo-skeletal CaCO₃ to buffer extracellular acidosis, although this has been seen in other terrestrial species (Henry, Kormanik, Smatresk & Cameron, 1981).

Oxygen-binding characteristics (Fig. 4) were similar to those reported for C. brevimanus (McMahon & Burggren, 1980). Compared with other terrestrial crabs, however (e.g. table IV, Mangum, 1983b), the Bohr factor is relatively small. Although lactate sensitivity has been demonstrated in other members of the Paguroidea (e.g. Petrolisthes and Emerita), it is by no means universal. In a recent review of the subject, Mangum (1983a) concludes that lactate sensitivity is commonly associated with a large Bohr shift and reliance on anaerobic metabolism - neither of which was observed in the present study.

Strategies used by C. compressus to increase O2 delivery during sustained voluntary walking (Table 3; Fig. 6) were similar to those identified in swimming blue crabs (Booth et al. 1982). Both contrast with the typical response to intense periods of activity (see review by McMahon, 1981), suggesting that marathon and exhausting exercise are functionally dissimilar. The majority of inactive decapods saturate the pigment so that O₂ delivered to respiring tissues originates from physical solution. During exercise, the pigment increases its role in O₂ delivery as circulating decreases onto the rising phase of the dissociation curve (Mangum, 19836; McMahon & Wilkens, 1983). in terrestrial hermit crabs is 2- to 4-fold higher than in other decapods (see tables III, IV of McMahon & Wilkens, 1983). Binding affinity and haemolymph are comparable, however, which means that and are located around the P₅₀. value on the dissociation curve. The pigment therefore accounts for most of the O₂ released under resting conditions.

Both Coenobita and Callinectes maintain haemolymph during sustained exercise on account of a rapid ‘on-response’ in and steady states of that can completely fuel the activity bout (see Booth et al. 1982; Wheatly et al. 1985). In Callinectes, the Bohr shift is offset by a lactate sensitivity which partially restores the overall HCy transport function. Since is only increased by 10%, the increase in O₂ delivery is attributed primarily to cardiovascular compensations. In Coenobita, the Bohr shift was unopposed and was increased 50 % by depressing prebranchial saturation (Fig. 6A). Since increased by threefold (Wheatly et al. 1985), the cardiovascular system is similarly implicated in enhancing transport O₂ Crabs removed from their shells exhibited the response typical of bursting activity (Fig. 6B). Several lines of evidence suggest that preexercised crabs were severely traumatized by removal from their shells which resulted in higher circulating values of ⁠.

Resting cardiac output was calculated, using the Fick principle, as 274 ml kg^-1 min^-1 (taking from Wheatly et al. 1985). This is almost double that of other species (McMahon & Wilkens, 1983), which is somewhat unexpected since generally decreases with Problems with this method of estimation have recently been discussed by Taylor (1982). doubled during sustained exercise in Coenobita and increased further still during recovery with access to external water. This suggests that gill perfusion is geared towards gill exchange functions occurring at this time (Fig. 3).

Haemolymph convection requirement was low (3·01 mmol^-1 O₂) in control animals due to the high ⁠. increased to 8·11 mmol^-1O₂ during exercise. Since the design of the respirometer made it impractical simultaneously to monitor respiratory and cardiac frequencies or ventilatory airflow and gas tensions, the following analysis of gas exchange has incorporated data from previous studies (e.g. Wood & Randall, 1981a). Since internal did not change during exercise, one can assume that the mean O₂ diffusion gradient was constant. Therefore increased must be primarily attributed to more favourable conditions for gas transfer across the respiratory surface. Assuming that branchial gas had the same as Cardisoma (Wood & Randall, 1981a), the transfer factor increased from 0·7 to 2·5 μmol O₂ kg^-1 min^-1 Torr^-1, which is similar to the increase observed in Callinectes (Booth et al. 1982). Both are higher than reported during sprinting in Cardisoma (Wood & Randall, 1981a).

For CO₂ excretion to occur in resting animals, the of branchial water must be below 2 Torr. A value halfway between circulating and ambient would produce a of 1·2Torr, corresponding to a value of 34·7μmolCO₂kg^-1 min^-1 Torr^-1. Based on diffusivities alone, one would expect to exceed by a factor of 10-fold. The 50-fold discrepancy would therefore confirm our earlier suspicion that conditions across the branchial epithelium favour CO₂ elimination. It is difficult to estimate changes in CO₂ conductance during exercise without more detailed information on the CO₂ diffusion gradient.

This work was undertaken at the Naos Laboratory of the Smithsonian Tropical Research Institute, Republic of Panama, in April and May of 1983 at which time the senior author was an Alberta Heritage Foundation Postdoctoral Fellow. Agencies gratefully acknowledged for financial support are AHFMR (MGW), NSERC (no. A 5762 to BRM), NSF (no. PCM-80-03752 to WWB) and the University of Massachusetts (WWB and AWP). The artwork was prepared by Sharon Harrison and Bill Adams.