Temperature and the Physiology of Intracellular and Extracellular Acid-Base Regulation in the Blue Crab Callinectes Sapid Us

It has been widely reported that temperature-related changes in extracellular pH (pHe) roughly parallel those of the neutral pH of water (ΔpN/ΔT = approx. −0·0175 u°C⁻¹ between 10 and 30°C). The phenomenon has been variously explained as the maintenance of ‘constant relative alkalinity’ (e.g. Howell, Baumgardner, Bondi & Rahn, 1970) or of constant fractional dissociation (α) of imidazole groups (‘alphastat’; e.g. Reeves, 1977). Neither theory offers a complete mechanistic explanation. While imidazole is undoubtedly the most concentrated buffer in vivo, regulation of the CO₂/HCO₃⁻ system represents the principal physiological control of acid-base status. The temperature dependence of its effective pK₁′ is very low (∼−0·005 u°C⁻¹) relative to imidazole (∼−0·018 u °C⁻¹), necessitating changes in [HCO₃⁻], or both. In general, air-breathing ectotherms operate a constant CO₂ content system such that is varied by active regulation of ventilation at more or less stable [HCO₃⁻] (cf. Reeves, 1977). The very low O₂vs CO₂ capacitance of water obviously restricts ventilatory control of in aquatic ectotherms; a range of different patterns of and [HCO₃⁻] variation with temperature has been reported (cf. Heisler, 1980; Cameron, 1984a,b). Studies on intracellular acid-base status are relatively few, and no clear temperature pattern has yet emerged. While intracellular pH (pHi) generally falls as temperature rises, ΔpHi/ΔT varies considerably between tissues and species with numerous deviations from the ΔpN/ΔT slope (Malan, Wilson & Reeves, 1976; Heisler, Weitz & Weitz, 1976; Heisler, Neumann & Holeton, 1980; Heisler, 1980; Cameron & Kormanik, 1982; Walsh & Moon, 1982).

The present study examined the influence of temperature on both pHe and pHi in the blue crab, Callinectes sapidus. Temperature effects on pHe have been previously studied in aquatic crabs, but conflicting results have emerged (Truchot, 1973, 1978; Howell, Rahn, Goodfellow & Herreid, 1973; Cameron & Batterton, 1978a; McMahon et al. 1978). There are no previous reports on temperature effects on pHi in any aquatic invertebrate. The present investigation focused on the following points.

(i) The relative importance of extracellular and [HCO₃⁻] changes, with particular attention to possible artifacts induced by inspired variation, a parameter which has not been carefully controlled in previous studies.

(ii) The ‘passive’ physico-chemical characteristics of haemolymph in vitro in an open system. The previous studies have described closed system, constant CO₂ content characteristics, but these may not be particularly relevant to the in vivo situation which more closely resembles an open system in a water breather.

(iii) Critical assessment of the ¹⁴C-DMO/³H-inulin technique for pHi determination (Waddell & Butler, 1959) by examination of marker equilibration, distribution, excretion and possible metabolism. Its only previous use in intact crabs produced peculiarly high values for ‘mean whole body pHi’ (Cameron, 1981).

(iv) The ΔpHi/ΔT relationships in different tissues, and their influence on the response of the whole body.

(v) The pattern of pHe and pHi adjustment following acute temperature change, and the importance of acidic equivalent exchange with the environment in the overall response.

Adult intermoult blue crabs, Callinectes sapidus Rathbun (140–450 g), freshly caught in sea water near Port Aransas, Texas, were held, with daily feeding, in running sea water at seasonal temperature (16–24°C). Five to ten days prior to experimentation, the crabs (groups of 6–10 in 600–1 tanks) were acclimated to the experimental temperatures of 10, 20 or 30 °C, without food. The acclimation sea water was filtered through a charcoal/sand bed and replaced at ∼5%h⁻¹. Salinity was 24–26%o, titration alkalinity was ∼2·4mequivl⁻¹, ⁠, and ⁠. On the day before an experiment, the crabs were fitted with neoprene septa over the pericardial space for arterial haemolymph sampling (Cameron & Batterton, 1978a). They were then allowed to recover overnight in the experimental chambers which were flushed with sea water from the acclimation tanks at −600 ml min⁻¹. These chambers were shielded Plexiglas boxes only slightly larger than the crab itself. A removable port allowed access to the pericardial septum for haemolymph sampling without disturbance. One hour before an experiment, the chamber was connected into a closed recirculating system comprising a pump (output ∼600 ml min⁻¹), aeration reservoir, and heat exchanger which maintained experimental temperature within ± 1 °C and > 120Torr. This closed system was necessary to monitor ¹⁴C-DMO and ³H-inulin excretion for calculation of ‘mean whole body pHi’. In the 20 °C studies only, the air leaving the aeration reservoir was bubbled through a 50cm, 500ml column of 1 moll⁻¹ KOH to trap any ¹⁴C-labelled CO₂ which may have resulted from metabolism of ¹⁴C-DMO.

Extracellular (i.e. arterial haemolymph) and intracellular acid-base status were examined in crabs acclimated to 10°C (N = 8), 20°C (N = 29) and 30°C (N = 14). At time 0, the volume of the recirculating system was adjusted to a known level (∼1·1 litre) and the crab injected with 20 μCi ³H-inulin (Amersham, ECF marker) and 4 μCi ¹⁴C-DMO (5,5-dimethyl-2,4-oxazolidinedione; ethyl acetate-free; New England Nuclear; pHi marker) in 150 μl of 400mmoll⁻¹ NaCl via the pericardial septum. At 20°C, the temperature at which the behaviour of the markers was examined in detail, simultaneous water, KOH and haemolymph samples (340 μ1) were withdrawn at 1, 2, 3, 4, 6, 8 and 12h. The water and KOH samples (after neutralization with boric acid) were assayed for ³H and ¹⁴C radioactivity and the haemolymph samples for radioactivity, pHa and total carbon dioxide content (C_Ta).

At 12 h, an additional 200μl haemolymph sample and a water sample from the inflow of the experimental chamber were analysed for and ⁠, respectively. To ensure that water ammonia remained below 200 μmol l⁻¹, the system was flushed (2min) immediately after the 4h and 8h samples; additional water samples were then taken to keep track of total ³H and ¹⁴C losses by the animal. To assess possible differences in acid-base status caused by the closed recirculating system, the experimental chambers were returned to an open flow-through system pumped directly from the acclimation tanks for the period from 12–24 h. At 24 h, all haemolymph measurements (and water ⁠) were repeated, a 2 ml terminal haemolymph sample drawn for ionic and protein analyses, and the crab killed for individual tissue pHi measurements. On the basis of the 20 °C results, the KOH trap and the 1, 2, 3 and 4h samples were not employed in the 10 and 30°C studies; the protocols were otherwise identical.

In order to minimize pHi disturbance due to struggling, crabs were killed as rapidly as possible by removal of the legs and dorsal carapace, and excision of the heart. Tissue samples (100–500mg) were dissected out, thoroughly blotted, and then dried to a constant weight at 80 °C in tared paper cups used for subsequent sample oxidation. The following tissues (number of samples per animal) were analysed: cheliped muscle (6–8); walking leg muscle (3–4); heart (1); carapace (from the pericardial and upper branchial chamber regions, scraped clean of soft tissue; 4–6); and dark (2–4) and light (2–4) ‘backfin’ muscle. The ‘backfin’ constitutes the swimming muscle of the 5th periopod, the colour difference between light and dark regions resulting from greater mitochondrial density in the latter (Tse, Govind & Atwood, 1983). The dissection was carried out in a tared dish lined with absorbent paper; subsequent drying to a constant weight yielded the total body water content.

Extracellular and whole body intracellular acid-base, status and acidic equivalent flux to the environment were followed in 20°C-acclimated crabs (N= 10) subjected to a rapid shift to 10 °C. The closed recirculating systems were operated at higher volume (∼3·01) and flushes carried out at 12h intervals to minimize possible disturbing effects on the measured flux rates. For the same reason, 12 h were allowed to elapse between the injection of the ³H-inulin/¹⁴C-DMO stock (as above) and the start of the 12 h control flux period at 20°C. Thereafter, water temperature was lowered to 10 °C over 45 min, the start of which represented time 0. Experimental fluxes were then measured over the following 0–2, 2–4, 4–6, 6–11 and 11–24h intervals. At each time, water samples were assayed for radioactivity, ⁠, total ammonia and titration alkalinity, the latter two allowing calculation of net acidic equivalent flux. Haemolymph samples (340 μl) for radioactivity, pHa, C-Ta and protein analyses were drawn at the start and end of the control period, and at 1, 5, 11 and 24h of the experimental period. Additional 200μl aliquots were analysed for ⁠, at the first control and 24 h experimental sample times. Finally, 2 ml terminal haemolymph samples were drawn for ionic analyses, and the crabs then killed and dried to a constant weight for total body water content. .

The physico-chemical characteristics of haemolymph in an open system in vitro were examined using the blood of six crabs acclimated to 20 °C. These crabs were selected to span the full range of haemolymph protein concentration observed in the in vivo experiments. Approximately 15 ml of venous blood was drawn from the arthrodial membranes of each animal and allowed to clot in a glass tube. After clot disruption and centrifugation at 12 000g for 10 min, the haemolymph was decanted into a spinning tonometer supplied with humidified gas mixtures from Wösthoff pumps. The haemolymph was serially equilibrated to ⁠, 4·5 and 9·0Torr (balance air) at 10, 20 and 30°C; C_T and pH were measured in duplicate after 40–60 min. The order of different temperatures was varied without any obvious effect on the results. As total tonometry time was 10–12 h, two initial values at 20 °C were repeated at the end to check for haemolymph deterioration in two runs. Deviations from the initial values of C_T and pH were within the error of the measurements.

Dried tissue, haemolymph and injection stock samples were combusted with a sample oxidizer (Packard 306), allowing separate assay of ³H and ¹⁴C-radioactivity. Water samples were counted directly in fluor. Dual label quench correction was performed using the external -standard ratio and internal standardization (when required). All radioactivities were measured by a liquid scintillation counter (Packard 3225) and ultimately converted to d.p.m., taking combustion and counting efficiencies into account.

In view of the large number of separate measurements involved in the calculations, the precision of individual tissue pHi determinations is estimated as about ±0·02 units, and of ‘whole body pHi’ values, about ±0·04 units.

The net flux of acidic equivalents to the environment was calculated as the sum of the change in titratable acidity of the water and the ammonia flux, signs considered (cf. Maetz, 1973). This method does not distinguish between ammonia movement in the NHj and NH₄⁺ forms, nor between the net excretion of acidic equivalents and the net uptake of basic equivalents, or vice versa (cf. McDonald & Wood, 1981). Fortunately this does not matter in terms of the net acid-base budget of the animal. Water titration alkalinity (for titratable acidity flux) was determined by titration to pH = 4·00 with 0·02 mol l⁻¹ HC1 as described by McDonald & Wood (1981); ammonia was measured by the phenol hypochlorite method of Solorzano (1969).

Data have been expressed as means ±1 S.E.M. (N) where N equals the number of animals sampled, and Student’s two-tailed paired and unpaired t-tests were used for comparisons within and between groups respectively. Lines were fitted by standard least squares regression, simple Pearson’s correlation coefficients and standard errors for slopes and intercepts were calculated, and differences in slope were assessed by analysis of covariance. A 5% significance level was employed throughout.

At 20 °C, the initial mixing phases for both compounds lasted 3–4 h; thereafter, plots of log haemolymph radioactivity against time were linear. For inulin this equilibration period was illustrated by the initial rise in the ‘instantaneous’ estimate of whole body ECW (Fig, 1A), which from 3h onwards was not significantly different from the ‘mean’ estimate based on the log extrapolation technique using 6–24 h data. The calculated ‘mean whole body pHi’ rose more gradually to a plateau at 6–12h (Fig. 1B), reflecting a slightly longer equilibration phase for DMO. Incomplete inulin distribution was of minor importance relative to incomplete DMO distribution in underestimating pHi, as shown by comparison of the pHi values calculated using the two different ECW estimates (Fig. 1B). Notably, the pHa-pHi gradient was only ∼0·25u (Fig. 1B), about half that seen in most other animals, but very similar to the previous report of Cameron (1981) for two air-breathing crabs. Haemolymph pHa remained stable throughout (Fig. 1B), but both and calculated (not shown) were slightly elevated at 1–4 h relative to later sample times, presumably in response to the injection and/or sampling. For these reasons, haemolymph samples were taken only from 6h onwards in subsequent studies at 10 and 30 °C. At 10 °C, only the 8h and 12 h data could actually be used since DMO equilibration at this colder temperature was not complete at 6h.

There was significant excretion of both DMO and inulin in the 12 h study period (Fig. 1C). Efflux rates varied widely amongst animals but were generally correlated for DMO and inulin and more or less linear over time. The mean losses were comparable at 10 and 20°C but significantly higher at 30°C (Table 1). For inulin, measured losses to the water were very similar to those predictable from K, the rate constant of disappearance from the haemolymph (Table 1), confirming that inulin was lost from the ECW by excretion and not by penetration of the ICW. Inulin clearance rates were 2·5-fold higher at 30 than at 10 or 20 °C, indicating a much greater rate of primary urine formation (Table 1). Radiolabel removal via haemolymph sampling was an additional, though much smaller, route of DMO and inulin loss, amounting to 10 – 30% of the excretion loss rates. Failure to account for these two sources of loss would have introduced significant error. For example, at 30 °C, ‘mean whole body pHi’ would have been overestimated by 0·23 u and ECW by 18%. Loss pf ¹⁴C-DMO by catabolism was evidently insignificant, since no ¹⁴C-radioactivity appeared in the KOH traps.

The haemolymph pHa fell with increasing temperature in Callinectes in the usual manner (Fig. 2A). The overall acid-base vs temperature pattern, however, differed significantly between crabs sampled in the closed recirculating systems and the same crabs sampled 12h later in the open flow-through systems. These effects were attributable to a significant rise in with temperature in the recirculating systems, from 0·74 Torr at 10°C to 1·45 Torr at 30°C, whereas was much more stable in the flow-through systems (0·60 Torr at 10 and 20 °C, 0·81 Torr at 30 °C) (Fig, 2C). Haemolymph to water gradients were identical at comparable temperatures in the two systems, so and were consistently higher under recirculating conditions (Fig. 2B,C). Relative to the flow-through situation, these crabs appeared to be in a state of slight respiratory acidosis, compensated at 30 °C but not at 20 °C (Fig. 2A). Thus, at the low levels characteristic of water breathers, relatively small temperature-related changes in can have marked effects on acid-base status which could easily be confused with direct temperature effects.

The haemolymph acid-base data for the flow-through situation are shown as the standard components of the Henderson-Hasselbalch equation in Fig. 3. The 30 °C data at Torr have been slightly adjusted to the same as the 10 and 20°C data (0·60 Torr), assuming an unchanged gradient and unchanged pHa. (Thus at 30 °C, was adjusted from 2·75 to 2·54 Torr, and [HCO₃⁻]a from 4·10 to 3·79mequivl⁻¹.) The overall ΔpHa/ΔT was -0-0151 u°C⁻¹. Between 10 and 20°C, haemolymph [HCO₃⁻]a fell significantly (Δ [HCO₃⁻]a/ΔT = −0·13 mequiv l⁻¹ °C⁻¹) but did not change at 30°C (Fig. 3B), whereas was identical at 10 and 20°C, but doubled at 30°C (⁠⁠; Fig. 3C). The factorial analysis of Fig. 3A illustrates that between 10 and 20 °C, ΔpHa/ΔT could be almost entirely attributed to a [HCO₃⁻]a change, and between 20 and 30°C to a change, and thus to an increase in the gradient across the gills. Callinectes haemolymph in vivo appeared to operate as an open, constant system between 10 and 20 °C, and as a closed, constant C_T system between 20 and 30°C in temperature-acclimated animals. As expected, the apparent CO₂/HCO₃⁻ pK₁^′, determined experimentally, changed with a relatively low slope ( −0·004 u°C⁻¹; Fig. 3D) and therefore made only a small contribution to the pHa vs temperature relationship (Fig. 3A).

Whole body ECW was similar at 10°C [232·1 ± 12·1 (8)gH₂0kg⁻¹] and 20°C [245·3 ± 12·1 (11)] amounting to 35% of total body water, but increased significantly by one-third at 30°C [320·1 ± 16·9 (14)]. This increase occurred entirely at the expense of the ICW, because total body water remained the same at the three temperatures (670–688 ml H₂O kg⁻¹). There was also a significant increase in trapped ECW at 30°C, in cheliped and dark backfin muscles, heart and carapace (Table 2). Total haemolymph protein concentration (mainly haemocyanin) did not vary significantly with temperature (Table 2), suggesting that synthesis of new haemocyanin may have accompanied the rise in ECW at 30°C. Haemolymph [Na⁺] and [Cl⁻] both fell by ∼30 mequiv l⁻¹ between 10 and 20 °C with no further change at 30°C (Table 3). Concentrations of Ca²⁺ and Mg²⁺ fell by a greater relative proportion between 10 and 20 °C, and then rose again at 30 °C (significant only for Ca²⁺). K⁺, present only at very low levels, was significantly greater at 20 and 30 than at 10°C. Interestingly, similar electrolyte levels to those in the 10 °C-acclimated crabs were seen in animals only 24 h after acute transfer from 20 to 10 °C in the temperature change experiment (Table 3). The strong ion difference (Stewart 1978), here defined as [Na⁺+Ca²⁺ + Mg²⁺ + K⁺ – Cl⁻] in mequivl⁻¹, was ∼30 mequiv l⁻¹ (or about seven-fold [HCOj⁻]a; Fig. 3A) and did not vary significantly with temperature. These electrolyte patterns seen during short-term laboratory acclimation were generally very similar to those occurring in Callinectes during long-term seasonal temperature acclimation in the wild (Lynch, Webb & Van Engel, 1973; Colvocoresses, Lynch & Webb, 1974).

The remarkably high ‘mean whole body pHi’, and resultant small pHa-pHi gradient, was explained (see Discussion) by the discovery of a large, highly alkaline fluid compartment in the carapace (Fig. 4). Whether or not this is a true intracellular compartment is uncertain, but it was not penetrated by inulin, and with a pHi 0·3–0·4u above pHa, served as an important sink for DMO. In one immediately pre-moult crab at 30°C, pHi in the soft underlying new carapace was 7·55 (relative to 8·33 in the old hard carapace) pointing to a role for mineralization in the origin of this high pHi (or vice versa;Cameron & Wood, 1985).

Heart and skeletal muscle tissues had a more usual pHa-pHi gradient ≃0·5–0·6 u (Fig. 4). Nevertheless, there was significant heterogeneity within skeletal muscle. At all temperatures, pHi was consistently lower in light than in dark backfin or cheliped, while walking leg was intermediate (Fig. 5). As the colour difference between light and dark backfin results from greater mitochondrial density in the latter (Tse et al. 1983), and mitochondrial pH is generally greater than cytosol pH (Roos & Boron, 1981), these inter-muscle pHi differences may reflect relative mitochondrial volume.

Intracellular pH fell significantly with increasing temperature in all compartments (Figs 4, 5; Table 4). There was a general tendency for a greater slope between 20 and 30 °C than between 10 and 20 °C, but this was significant only for heart (−0·0222 ± 0·0051 vs −0·0093 ± 0·0050 u°C⁻¹) and walking leg (−0·0167 ± 0·0033 vs −0·0076 ± 0·0036u°C⁻¹). Over the whole range of 10–30°C, the slopes for all tissues except carapace were not significantly different from those for haemolymph (ΔpHa/ΔT = −0·0151 u°C⁻¹). However ΔpHi/ΔT values for all skeletal muscle tissues were significantly less than for constant relative alkalinity (i.e. ΔpN/ΔT ≃ −0·0175 u°C⁻¹). The slope for carapace was extremely flat, and this was clearly reflected in the slope for ‘mean whole body pHi’ (Fig. 4; Table 4); both were significantly less than either ΔpHa/ΔT or ΔpN/ΔT.

When haemolymph was equilibrated at constant in vitro (i.e. open system conditions), [HCO₃⁻] fell as temperature rose (Fig. 6B). Although the absolute [HCO₃⁻] increased with greater ⁠, Δ[HCO₃⁻]/ΔT was independent of but a direct function of haemolymph protein concentration (Fig. 7A). This is interpreted as a direct consequence of rising temperature increasing the dissociation of H⁺ ions from the haemocyanin, resulting in the removal of HCO₃⁻ from the system as gaseous CO₂. As such it is considered a ‘passive’, physico-chemical effect - i.e. not involving ‘active’ regulation of and/or acidic equivalent exchange by the animal. The wide range of haemolymph protein levels, selected to span the normal range of occurrence, explained most of the variability in the averaged data of Fig. 6. At the mean protein concentration observed in vivo (Table 2), this ‘passive’ Δ[HCO3⁻]/ ΔT would have been about −0·17mequivl⁻¹ °C⁻¹ (Fig. 7A), slightly greater than the in vivo slope of −0·13 mequiv 1⁻¹ °C⁻¹ at constant between 10 and 20°C (Fig. 3B,C). The change in haemolymph pH with temperature in vitro associated with this constant Δ[HCO₃⁻]/ ΔT was proportionately greater at lower values (Fig. 6A), as could also be predicted from the Henderson-Hasselbalch equation. However, even at the lowest tested (1·5Torr), ΔpH/ΔT (−0·0071 u °C⁻¹; Fig. 6A) was less than half that observed in vivo (−0·0151 u °C⁻¹; Fig. 3A) between 10 and 20 °C where was similarly constant. This difference was explained partly by the lower in vivo (1·2Torr), since small changes in have large effects on pH in this range, and partly by the greater variability in vitro.

The non-bicarbonate buffer value (β = − Δ [HCO₃⁻]/ ΔpH) was calculated from the linear change in [HCO₃⁻] with pH at the three equilibration levels of ⁠. At all three temperatures, β was a linear function of total haemolymph protein concentration (Fig. 7B), with intercepts not significantly different from zero. This is not surprising, as the only other likely non-bicarbonate buffer, inorganic phosphate, was present in negligible quantities [0·18 ±0’03 (5) mmol l⁻¹]. The value of clearly increased with temperature, the regression slope rising significantly from 1·819 ±0·121 at 10°C to 2·637 ± 0·201 slykes 100ml⁻¹ g⁻¹ protein at 30°C. The mean values of β in vivo based on these relationships and the haemolymph protein levels of Table 2 were ∼11·5slykes at 10 and 20°C and ∼15’4slykes at 30°C.

In acclimated animals, the acid-base change between 20 and 10°C in vivo was entirely a [HCO₃⁻]a rather than a adjustment (Fig. 3); the ‘passive’ generation of basic equivalents by haemolymph at constant in vitro was large enough to explain this A[HCO₃⁻]a/ΔT, and the ECW did not change. The response to a step change from 20 down to 10 °C was therefore followed in the whole animal to identify the time course of adjustment and to assess the magnitude of any acidic equivalent exchange with the environment which might occur.

An immediate drop in was the most prominent extracellular effect of the rapid decrease from 20 to 10°C (Fig. 8C), very different from the situation seen in fully acclimated animals (Figs 2, 3). While this was partially due to a fall in in the recirculating system, the more important component was a 50% decrease in the gradient across the gills (Fig. 8C). Initially, Cq_a rose only slightly and non-significantly (Fig. 8B), so the intact animal in some ways initially behaved in a manner similar to a classic in vitro closed system. However the situation was complicated by the fall in ⁠, which contributed to the decrease in and thereby masked an effective [HCO₃⁻]a adjustment. Had the change in gradient alone occurred (i.e. absence of change), together with the observed pHa change, [HCO₃⁻]a would have risen with a slope of −0·15mequivl⁻¹°C⁻¹ between 20 and 10°C over the first hour. The net effect of this combined and [HCO₃⁻]a adjustment was a large initial overshoot in pHa, which rose with a slope of −0·0218 u°C⁻¹ during the first hour (Fig. 8A). Thereafter pHa gradually fell, accompanied by a slow rise in both and the latter reflecting an increase in the gradient. By 24 h, this gradient had returned to the 20°C value, had risen significantly, and pHa had reached a level which gave a 20 to 10°C slope of −0·0104 u°C⁻¹. Adjusting the data to constant as before, the effective Δ[HCO₃⁻]a/ΔT value between 20 and 10°C after 24 h was −0·18mequivl^−loC⁻¹. Thus after 24h, the extracellular in vivo situation was very similar to that predicted by the immediate ‘passive’ response of haemolymph in an open system in vitro to this same temperature change. Throughout the experiment, there was no significant change in haemolymph protein concentration, which averaged 4·79 ± 0·50 (10) g 100 ml⁻¹ in this group.

‘Mean whole body pHi’ followèd a very similar pattern to pHe, with an initial overshoot followed by later decline (Fig. 8A), though the absolute changes were smaller. Over the first hour, ΔpHi/ΔT was −0·0164 u°C⁻¹, but this decreased to −0·0077 u°C⁻¹ by 24 h, close to the value observed in temperature-acclimated crabs (Table 4). Assuming equilibration across the extracellular/intracellular interface, this pattern can be explained by the same factors as in the ECW – i.e. an initial, and later corrected, undershoot of against a background of [HCO₃⁻]i accumulation. At constant the effective A[HCO₃⁻]i/ΔT between 20 and 10°C after 24h was −0·12mequiv 1^−1OC⁻¹.

At 20°C, the titratable flux was just balanced by the ammonia flux, so the net acidic equivalent flux between the crab and the external sea water was not significantly different from zero (Fig. 9B). Two hours after the temperature shift from 20 to 10°C, this situation changed dramatically. A significant net uptake of acidic equivalents (or excretion of basic equivalents) from the environment began (Fig. 9B), accompanied by a 90% decrease in ammonia output (Fig. 9A). By 24h, the ammonia output had recovered only slightly. The acidic equivalent flux, while still significantly elevated, had dropped to about half the peak level of + 200 μequivkg⁻¹ h⁻¹. The net uptake of acidic equivalents over the 24h period following the temperature change amounted to 2240 μequivkg⁻¹ relative to the control rate at 20°C.

The ¹⁴C-DMO/³H-inulin technique worked satisfactorily in Callinectes, providing certain precautions were observed. The first was adequate marker equilibration time, 4–6 h at 20°C (Fig. 1) and up to 8 h at 10°C, comparable or slightly longer than those for teleost fish (Cameron & Kormanik, 1982; Walsh & Moon, 1982; Hōbe, Wood & Wheatly, 1984) and crayfish (Gaillard & Malan, 1983). Most of this delay probably reflected convective mixing in the ECW, since, at least in fish tissues, DMO equilibration time across the extracellular/intracellular boundary is rapid (<15 min; Walsh & Moon, 1983; C. L. Milligan & C. M. Wood, unpublished data). An additional problem in crustaceans may be slow marker penetration of the carapace, the pHi of which so markedly influenced ‘mean whole body pHi’ (see below). Nevertheless, the plateaux reached in the whole body pHi and ECW curves by 12 h (Fig. 1A,B) clearly indicated that marker distribution in the carapace was at equilibrium by the time (24 h) this tissue was sampled. A second important precaution was correcting for the significant excretion of ¹⁴C-DMO and ³H-inulin which occurred before equilibration; the neglect of this could result in considerable overestimation of ‘mean whole body pHi’ and ECW. The measured DMO excretion rates in Callinectes were 1·2 to 2·8-fold greater than in trout (Hôbe et al. 1984). The route of inulin loss is exclusively via the antennal gland (Cameron & Batterton, 1978b; Wheatly, 1984) and the close correlation of DMO and inulin loss rates in individual crabs suggests that this is also true for DMO.

Compared with microelectrodes, DMO yields slightly lower but consistent values of pHi in isolated crustacean muscles (Boron & Roos, 1976; Rodeau, 1982). Mean pHi values in several soft tissues of intact Callinectes (heart, cheliped muscle; interpolated to 13°C; Fig. 4; Table 4) in the present study were very similar to those obtained by Gaillard & Malan (1983) in the crayfish Astacus at 13°C using the same methodology. The pHa-pHi gradient of 0·5–0·6u for skeletal muscle in Callinectes was in the normal range, but that for ‘mean whole body’ was only about half as large as normally found in vertebrates (Roos & Boron, 1981). Cameron (1981) reported an identical situation in two air-breathing crabs and suspected either a large heterogeneity of the overall ICW or a systematic error. The current results show that the former is the correct explanation, because there is a large, highly alkaline (pHi = 8·1–8·3) fluid compartment in the carapace markedly influencing the whole body pHi value (Fig. 4; Table 4).

Waddell & Bates (1969) have noted that a subcompartment with a high pHi may bias the estimate of ‘mean whole body pHi’ with the DMO method because the distribution of DMO is a logarithmic, rather than a linear, function of pH. A rough estimate of the extent of such bias may be made for the 20 °C data using the total weight and water content of Callinectes carapace reported by Cameron & Wood (1985), the trapped ECW corrections of Table 2, the whole body ICW, and assuming that the mean pHi of the whole carapace was equal to that of the carapace areas sampled and that the mean pHi of all soft tissues was equal to that of skeletal muscle (Fig. 4). Given these conditions, the true ‘mean whole body pHi’ should have been 7·51, whereas the estimate based on equation (8) was 7·54, indicating that the upward bias, while present, was not particularly serious.

It is uncertain whether the carapace pHi values were representative of true intracellular fluid or simply of a bounded fluid space not penetrated by inulin. True cellular pHi values in this range (8·1–8·3; Fig. 4; Table 4) are unusual but not unprecedented (Roos & Boron, 1981). The cuticular epidermis certainly contains some cells (Dennel, 1960), but it is unlikely that all the carapace water (∼90%, Table 2; equal to about 91 ml kg⁻¹ body weight, Cameron & Wood, 1985) lying outside the trapped ECW is contained within cells, as assumed in the pHi calculation. Whatever the exact nature of this pool, it is probably in dynamic equilibrium with the massive CaCO₃ stores which occur in the carapace (Cameron & Wood, 1985), and either or both may be an important source of mobilizable base for extracellular buffering. Calculated ‘intracellular’ [HCO₃⁻] in the carapace fluid space was ∼12mequivl⁻¹ at 20°C in comparison to ∼4mequivl⁻¹ in the haemolymph and ∼lmequivl⁻¹ in skeletal muscle. The measured CO2 (mainly CO₃²⁻) store in the whole carapace was 2·09×10⁶μequivkg⁻¹ body weight ( 1·04×10⁶μmol l⁻¹; Cameron & Wood, 1985) or over 1000 times the HCO₃⁻ content of the rest of the body. There is now considerable evidence that mobilization of carapace buffer base can occur during acidosis in crabs (Defur, Wilkes & McMahon, 1980; Henry, Kormanik, Smatresk & Cameron, 1981; Wood & Randall, 1981); this whole area clearly deserves further investigation.

In the present study the value of ΔpHa/ΔT for acclimated Callinectes was −0·0151 u°C⁻¹, close to that for constant relative alkalinity (≃ −0·0175 u°C⁻¹).

This may be compared with previous values for Callinectes of −0·0120 (Cameron & Batterton, 1978a) and −0·0260 (Howell et al. 1973), for Carcinus of −0·0142 (Truchot, 1978), −0·0162 (Truchot, 1973) and −0·0190 (Howell et al. 1973), and −0·0180u°C⁻¹ for Cancer (McMahon et al. 1978). Haemolymph [HCO₃⁻] fell with rising temperature in all studies but with a slope varying widely from −0·085 mequiv l⁻¹ °C⁻¹ (Truchot, 1973) to −1·;29 mequiv l⁻¹ °C⁻¹ (Howell et al. 1973). either rose a lot (Truchot, 1973, 1978), a little (Cameron & Batterton, 1978a), remained the same (McMahon et al. 1978) or fell (Howell et al. 1973). In none of these previous studies was closely controlled or measured. In our experience, despite vigorous aeration, tends to increase with temperature in recirculating systems (cf. Fig. 2). Because of the very low levels and gradients across the gills in crabs, minor changes in can have very large effects on pHa, which may or may not be compensated. For example, using the 20°C data of Fig. 3, an increase in from 0·60 to 1·20 Torr would have decreased pHa from 7·85 to 7·68; after metabolic compensation, [HCO₃⁻]a would have risen from 3·99 to 5·92 mequiv l⁻¹. These effects are larger than measured differences at temperatures 10°C apart when Pico₂^was stable (Fig. 3). The comparison of recirculating and flow-through system data in Fig. 2 emphasizes the possible errors involved.

At constant the haemolymph acid-base adjustment in acclimated Callinectes was almost entirely the result of a [HCO₃⁻]a change between 10 and 20°C and a change between 20 and 30°C. This pattern has previously only been observed in two amphibious crabs (Cardisoma and Coenobita; McMahon & Burggren, 1981), where it was attributed to a switch from an aquatic (i.e. HCO₃⁻ exchange) to an aerial mode of regulation (i.e. ventilatory control) at higher temperature, an explanation which cannot apply to Callinectes. Rather, it appears that between 10 and 20°C, the [HCO₃⁻]a change could be explained by the ‘passive’ effect of temperature on H⁺ dissociation from haemocyanin in an open system (see below), and between 20 and 30°C, the change could be explained by classic closed system behaviour of the haemolymph (Reeves, 1977). It is not clear why the latter should occur in a water breather. Relative ventilation (as expressed by the ventilatory convection requirement for O₂) is unaffected by temperature in Carcinus (Truchot, 1978) and Callinectes (Cameron & Batterton, 1978a). However, as Truchot (1978) points out, may depend on (expired) ⁠, which in turn varies inversely with μCO₂, the latter decreasing with temperature. Thus the fall in at high temperature observed by Cameron & Batterton (1978a) in Callinectes was probably associated with a rise in ⁠.

In Callinectes, as in most other poikilotherms which have been examined (see Introduction), pHi in individual tissues fell with increasing acclimation temperature (Figs 4, 5; Table 4). For most tissues, ΔpHi/ΔT was less than ΔpN/ΔT but not significantly different from ΔpHa/ΔT, and there was some evidence of changing slope over the full 10–30°C range. A general fall in pH with temperature is to be expected in virtually any buffer system. In intact crayfish, Gaillard & Malan (1983) have demonstrated considerable capacity within individual tissues to regulate pHi independently of pHe, presumably by transmembrane acidic equivalent flux. Without a detailed knowledge of the extent of such fluxes, metabolic adjustments, changes and the relative proportions of protein (imidazole), phosphate and bicarbonate buffers in the cytosol, no definite conclusions can be drawn. However, in the carapace, where buffering by the bicarbonate system undoubtedly predominated, the very flat temperature slope was probably a direct consequence of the low temperature sensitivity of this buffer system. The marked influence of carapace pHi on ‘mean whole body pHi’ contributed greatly to the low ΔpHi/ ΔT of the latter.

The HCO₃⁻ pools within acclimated crabs at 10, 20 and 30°C (Table 5) were estimated by assuming that pHi was internally uniform and that ^was equilibrated across the extracellular/intracellular interface, and by using intracellular pK₁′ values taken from pH vs pK₁′ regressions fitted to the in vitro haemolymph data. The analysis indicated that over half the HCO₃⁻ pool lay in the extracellular compartment at each temperature, and that the total pool varied by less than 30% at the three temperatures. The rise in the total pool between 20 and 30°C did not indicate that the pattern of [HCO₃⁻] regulation in the intracellular compartment was different from that of the extracellular compartment, but rather that the volume of the ECW relative to the ICW had increased.

The fall in [HCO₃⁻]a with rising temperature from 10 to 20°C in acclimated crabs in vivo need not reflect active mechanisms of HCO₃⁻ regulation. These changes could result simply from the ‘passive’ effect of temperature on H⁺ dissociation from haemocyanin, presumably from the histidine residues. These released H⁺ ions would combine with HCO₃⁻ and the resulting CO₂ would leave in an open, constant system. In accordance with this interpretation, A[HCO₃⁻]/ ΔT in the open system was independent of (Figs 6B, 7A) but directly related to haemolymph protein concentration (Fig. 7A). This ‘passive’ A[HCO₃⁻]/ΔT of −0·17mequivl⁻¹°C⁻¹ at normal in vivo protein levels was sufficient to explain the observed in vivo change of −0·13 mequiv l⁻¹ °C⁻¹ at constant between 10 and 20°C in acclimated animals (Fig. 3B). The Henderson-Hasselbalch equation illustrates that at constant ‘passive’ Δ [HCO₃⁻]a/ ΔT, the accompanying change in pHa will be greater, the lower the absolute value of ⁠, because of relative changes in the [HCO₃⁻]a/S_CO2 ratio. In practice in Callinectes, it can be calculated that ‘passive’ ΔHCO₃⁻/ ΔT = 0·17 mequiv l ⁻¹⁰C⁻¹ between 10 and 20°C would produce ΔpHa/ ΔT ≃ ΔpN/ ΔT at ⁠, which was close to the observed situation between 10 and 20°C (Fig. 3). At higher ⁠, as at 30°C (Fig. 3C), the influence of this ‘passive’ mechanism on pHa would be greatly reduced (e.g. Fig. 6A) and thereby overwhelmed by ‘active’ mechanisms such as the Paco₂ adjustment itself.

A similar passive mechanism has recently been identified in tuna blood (S. F. Perry, unpublished data) but.appears to have been overlooked in previous studies on water breathers, since open systems have been little studied. Two exceptions are the data of Cameron & Batterton (1978a) showing an in vitro open system slope of −0·12mequivl⁻¹⁰C⁻¹ in Callinectes haemolymph, close to their in vivo slope of −0·16 mequiv l⁻¹ °C⁻¹, and the report of Randall & Cameron (1973) that ΔpH/ ΔT in trout blood in vitro was −0·019u°C⁻¹ at 0–1 Torr, but −0·005 and −0·004 u °C⁻¹ at 11 or 24 Torr.

Like Δ [HCO₃⁻]/AT in the open system, haemolymph β was also a linear function of protein concentration (Fig. 7B), which seems reasonable as the same buffer sites could be involved in both phenomena. Truchot (1976b) also demonstrated a linear relationship between total protein and β in Carcinus haemolymph. However, in the present study, β, at constant protein concentration, clearly increased with temperature, an effect which could result from slight conformational changes in the haemocyanin molecule opening up more buffer sites. In rats, Saborowski, Lang & Albers (1973) found that the extracellular β per unit haemoglobin rose by 48% between 21·5 and 37°C, a proportionally similar change to that of haemolymph in vitro between 10 and 30°C (Fig. 7B). A similar effect has recently been documented in purified human haemoglobin solutions and attributed to the increased participation of carbamino groups on valine residues in buffering at higher temperatures (Castaing, Bursaux & Poyart, 1982). Whether this is a feature of proteins in general remains to be seen. If so, analyses of temperature vs acid-base status which assume constant values of β should be re-evaluated (e.g. Heisler & Neumann, 1980; Cameron, 1984b).

In the 24 h period following an acute shift from 20°C down to 10°C, crabs took up 2240 μequivkg⁻¹ net acidic equivalents from the environment (Fig. 9B). Thus the animal effectively excreted base (i.e. acidic equivalent uptake) at a time when it was accumulating HCO₃⁻internally. However, as outlined above, the ‘passive’ change in haemolymph [HCO₃⁻] with temperature at constant ⁠, would account for all the observed HCO₃⁻accumulation in the ECW. One possible explanation is that the ‘passive’ readjustment of the intracellular buffer pools produced more HCO₃⁻ than was needed for pHi adjustment, and this excess was effectively moved into the external environment by ‘active’ mechanisms. With an effective A[HCO₃⁻]i/ΔT slope of −0·12 mequiv l⁻¹ °C⁻¹ 24 h after the acute change, the actual elevation in the intracellular HCO₃⁻ pool was only + 530 μequivkg⁻¹. In fully acclimated crabs, the intracellular elevation was even smaller (+276 μequiv kg⁻¹; Table 5). If this is the correct explanation, then ‘passive’ intracellular A[HCO₃⁻]/ΔT would have to be approximately three- to four-fold greater than the observed extracellular value of −0·17 mequiv 1⁻’°C⁻¹.’ This may not be unreasonable, as passive extracellular Δ[HCO₃⁻]/ ΔT seemed to vary in parallel with β (Fig. 7), and intracellular β is typically three- to four-fold greater than extracellular β in poikilotherms (e.g. Heisler & Neumann, 1980; Cameron & Kormanik, 1982). However, other explanations are equally possible (e.g. metabolic adjustments, shifts in buffer composition, variation); again more detailed intracellular studies are obviously needed to settle the question.

The only comparable study on crabs is that of Truchot (1978) who found no change in net acidic equivalent flux after Carcinus was subjected to a 20 to 10°C shift. However, Truchot did not take ammonia flux into account. If this were reduced by 90% in Carcinus as in Callinectes (Fig. 9A), his results would have been very similar to ours. On the other hand Heisler (1978) and Cameron & Kormanik (1982) measured smaller net acidic equivalent fluxes of opposite sign (ammonia considered) in dogfish and catfish respectively subjected to similar temperature decreases. The reasons for these differences are unknown.

The pattern of acid-base regulation initially seen after the acute temperature shift was very different from that seen in acclimated animals. This difference was characterized particularly by an initial undershoot of the gradient across the gills and associated overshoot of pHa and pHi, both of which were corrected over the following 24h (Fig. 8). Whether this resulted from acute temperature effects on metabolism, ventilation or both cannot be determined at present. Nevertheless the observation emphasizes both the complexity of the adjustments made by the animal and the importance of acclimation time in defining the pattern of these adjustments.

We thank Anna M. Garcia for excellent technical assistance, Drs S. Gaillard and A. Malan for helpful communication and access to their manuscript prior to publication, and Drs S. F. Perry and C. E. Booth for constructive comments. The work was supported by NSF Grant PCM80-20982 to JNC and an NSERC operating grant to CMW. CMW’s visit to Port Aransas was supported by an NSERC International Collaborative Research Fellowship.