Cardiovascular adaptations enhance tolerance of environmental hypoxia in the crab Cancer magister

Many aquatic animals face periodically or erratically variable O₂ levels in their natural habitat. Tolerance for reduced environmental oxygen availability is dependent on both phylogenetic and ecological considerations but, within a specific group, the latter factors play a greater role (Herreid, 1980). Animals that must regularly cope with hypoxic exposure possess either physiological or behavioural mechanisms to allow survival during periods of low O₂ availability.

The respiratory responses of several species of decapod crustaceans to environmental hypoxia have been relatively well investigated (Johansen et al. 1970; McMahon and Wilkens, 1975; Taylor, 1976; Dejours and Beekenkamp, 1977; Butler et al. 1978; Wilkes and McMahon, 1982; Bradford and Taylor, 1982), as have some responses of the circulatory system (McMahon and Wilkens, 1975; Butler et al. 1978; Bradford and Taylor, 1982). In general, gill ventilation rate increases and heart-beat frequency is reduced as declines below normal environmental levels. O₂ consumption is maintained at values above the species-dependent critical O₂ tension, P_C, which is affected by other environmental factors such as temperature and salinity as well as endogenous factors including the ability to supply O₂ to the tissues (for a review, see Herreid, 1980). Below the P_C, ventilation and O₂ consumption rates decline and heart rate is further reduced since there is insufficient O₂ available to meet the aerobic demands of the respiratory and circulatory pumps. Depending on the capacity of the organism for anaerobic metabolism, prolonged exposure to such extreme hypoxia is potentially lethal.

In decapod crustaceans, a single ventricle drives haemolymph circulation through several systems of arteries, tissue lacunae, venous sinuses and discrete return channels (Maynard, 1960). Cardioarterial valves at the origin of each arterial system from the heart are independently innervated and contain striated muscle fibres (Alexandrowicz, 1932; Kuramoto and Ebara, 1989). Differential responses of the anterior and posterior cardioarterial valves of isolated lobster (Panulirus japonicus) hearts to amine and peptide neurohormones (Kuramoto and Ebara, 1984) suggest a mechanism for control of haemolymph redistribution in intact crustaceans. Independent changes in haemolymph flow through separate arterial systems in response to both enforced activity and neurohormone treatment have been demonstrated in vivo for the crab Cancer magister (Bourne and McMahon, 1989; Airriess and McMahon, 1992; McGaw et al. 1992) and in response to environmental hypoxia and peptide neurohormones for the lobster Homarus americanus (Reiber et al. 1992; McMahon, 1992).

In the present study, cardiovascular changes associated with hypoxic exposure were investigated using the brachyuran crab Cancer magister Dana. This species is found on the Pacific coast of North America from the littoral zone to a depth of 196 m (Butler, 1961). Its favoured habitats are sandy beaches and estuaries with dense eel-grass (Zostera) cover (Butler, 1984). On sandy beaches, crabs that are stranded by the receding water may burrow for protection (G. Jensen, personal communication; C. N. Airriess, personal observation), leaving only interstitial water available for ventilation of the gills. The O₂ available in this limited supply of water is rapidly depleted (deFur et al. 1983). Estuarine sediments and the waters above them are often extremely hypoxic (Brotas et al. 1990); therefore, crabs in either habitat may face depleted environmental with each tidal cycle.

P_C for O₂ consumption has been estimated at 7–10 kPa for a closely related crab, Cancer pagurus (Bradford and Taylor, 1982). Since C. magister may be more regularly exposed to hypoxic conditions, inhalant values of 3–5 kPa were used in the present investigation to assess the ability of this crab to maintain cardiovascular and respiratory function when O₂ availability is low. Haemolymph flows through all arterial systems, as well as heart-beat and ventilation frequencies, were measured in order to elucidate the mechanisms available to this species to enable it to thrive in euryoxic habitats. These mechanisms may include the ability to control regional haemolymph distribution, using the cardioarterial valves, in order to maximise O₂ delivery to aerobic tissues during hypoxic periods. Simultaneous measurement of haemolymph flow through all arterial systems, which allows direct calculation of cardiac output and stroke volume, has not been reported previously for any decapod crustacean.

All experiments described herein were carried out at the Bamfield Marine Station, Bamfield, British Columbia. Male C. magister (622.1±29.4 g; range 310–880 g) were either obtained by trapping or purchased from local fisherman. Females were not examined in this study because of heavy commercial exploitation and diminishing stocks of this species. Crabs were maintained in outdoor flow-through aquaria at 12±1 °C and 34 ‰ salinity. They were fed fish twice per week and subjected to ambient lighting. Prior to experimental use, crabs were starved for 2 days to avoid changes in metabolism associated with digestion (Ansell, 1973). All animals used in this study remained healthy for at least 1 week following experimental use. After this interval they were dissected for measurement of cardiac and arterial dimensions.

Instantaneous haemolymph flow through the anterior aorta, anterolateral arteries, hepatic arteries, posterior aorta and sternal artery was measured using a pulsed-Doppler flowmeter (University of Iowa Bioengineering). In this technique, bursts of ultrasound generated by 1 mm piezoelectric crystals (Titronics Medical Instruments or Crystal Biotech, Inc.) are directed into the lumen of each artery under investigation. The echo of this transmission from haemolymph-borne particles is received by the same crystal, with its frequency shift being proportional to the velocity of the particles (Hartley and Cole, 1974).

With the exception of that focused on the sternal artery, all Doppler probes were mounted in grooves abraded in, but not penetrating, the carapace overlying each artery (Fig. 1A). The sternal artery probe was fitted inside a sleeve of polyethylene tubing (PE 200) inserted through a hole drilled in the first abdominal segment. Latex dental dam covered the hole to prevent haemolymph loss around the catheter sleeve. The probe tip was guided along the median groove of the last thoracic endophragmal plate until it reached the first descending segment of the sternal artery (Fig. 1B). After lateral adjustment to achieve maximum signal amplitude, the probes were fixed in place using dental periphery wax and cyanoacrylate cement. Vertical focusing of the ultrasonic beam to ensure measurement of centre-stream velocity was accomplished using the flowmeter range control. Optimal focus was verified prior to each measurement.

Heart-beat frequency (f_H) was measured using impedance pneumography. The bared ends of Teflon-coated silver wires (0.05 mm diameter) were inserted approximately 2 mm into the lateral pericardial sinus on each side of the heart. Dental dam was used to cover the holes, both to prevent bleeding and to hold the wires in place. An impedance converter (UFI, model 2991) detected changes in inter-electrode impedance associated with ventricular contraction. Phasic output from the pulsed-Doppler flowmeter could be used as an alternative measure of heart-beat frequency in cases where the impedance electrodes became dislodged. Frequencies determined using either method were identical, but the impedance conversion method was preferable since haemolymph flow was occasionally diminished in all arterial systems in the absence of acardia.

Measurements of scaphognathite (ventilation) rate were obtained using a seawater-filled polyethylene catheter (PE 200) implanted in the right epibranchial chamber and connected to a Statham/Gould PB23Db pressure transducer (modified from Hughes et al. 1969). Dental dam was used to cover a hole drilled in the carapace overlying gills 7 and 8, both to prevent blood loss and to hold the catheter securely in place. This method, evaluated by McDonald et al. (1977), allows measurement of pressure fluctuations within the branchial chamber. These fluctuations correspond to movement of the scaphognathite, allowing determination of scaphognathite beat frequency (f_SC). Branchial-chamber pressure, mean output from the pulsed-Doppler flowmeter and heart-beat frequency were all recorded on a Gould 2600 rectilinear oscillograph.

Postbranchial haemolymph was sampled for analysis of L-lactate concentration in six animals which had been subjected to sham surgery but not instrumented. A hole drilled in the carapace dorsal to the lateral pericardial sinus and covered with dental dam permitted repetitive haemolymph sampling without excessive bleeding. Samples were deproteinated, using ice-cold 0.6 % perchloric acid, and centrifuged. L-Lactate concentration of the supernatant was measured using the Boehringer Mannheim UV test kit (catalogue no. 139084) with absorbance read at 339 nm (method as modified by S. Morris for small haemolymph samples, personal communication).

In order to minimise the disturbance caused to the animals during preparation, crabs were held in a shallow pan of cold aerated sea water for the duration of the set-up period. Following this procedure, the animals were transferred to the experimental chamber and allowed to recover for 24 h.

During experimentation, the crabs were held without additional restraint in a 3 l plastic chamber through which sea water from the main system was allowed to flow at approximately 200 ml min⁻¹. A water jacket maintained temperature at 12±1 °C. The chamber was shielded with black plastic to conceal the experimenter from the animal, but incident light could enter through the rear of the chamber so that the usual light:dark cycle was maintained. A tightly fitting lid with ports for passage of recording leads sealed the experimental chamber from atmospheric air. A large airstone in the path of inflowing water was supplied with either room air, to maintain normoxic conditions, or a precision gas mixture used to lower chamber to the desired level.

Subsequent to the post-operative recovery period, respiratory and circulatory variables were recorded continuously during a 1 h control period. Three measurements obtained at 30 min intervals during this period were averaged to obtain control rates of all variables. Crabs were then subjected to 6 h of hypoxic exposure with stepwise decreases in ⁠. At the start of the experimental period, the supply to the airstone was switched from room air to the outflow from a series of precision gas-mixing pumps (Wösthoff). The percentage of CO₂ in the gas mixture was held constant at 0.03 % while the percentage of O₂ was reduced from normoxic levels to 5, 4 and 3 % (corresponding to values of 4.63, 3.77 and 2.96 kPa, respectively) to subject the crabs to three distinct levels of hypoxic exposure. was held at each level for 2 h and all variables were measured 30, 60 and 120 min after the start of each exposure level. Following 2 h at 2.96 kPa, supply to the airstone was reverted to room air and measurements of circulatory and respiratory variables were obtained after 0.5, 1, 4, 12 and 24 h of recovery.

During control, experimental and recovery intervals, was monitored using a polarographic O₂ electrode (E5046; Radiometer, Copenhagen) protruding through an opening in the lid of the experimental chamber. The electrode was connected to an acid–base analyzer (Radiometer PHM71) and calibrated using N₂-and air-saturated sea water. Output from the analyzer was recorded on a Kipp and Zonen linear chart recorder.

Calibration of the pulsed-Doppler flowmeter system was verified using 12 crabs which were not included in the present study (Airriess et al., 1994). Transducer crystals were implanted as described above and adjusted to produce optimal output. Post-calibration dissection verified the position and orientation of the internal probes. The anterolateral arteries were chosen as representatives of the surface-mounted transducer configuration because of the relative ease of access and large size of these vessels. Both timed collection of haemolymph and retrograde perfusion of the artery via a cannula inserted immediately anterior to the flow probe were used to determine the correlation between actual volume flow rate and flowmeter output (r²=0.996; β=0.963) at flow rates ranging from 0 to 10 ml min⁻¹. For the internal transducer configuration, a peristaltic pump was used to draw haemolymph from the base of a leg and to perfuse the sternal artery of crabs from which a section of the dorsal carapace and the heart had been removed. The ventral pericardial septum remained intact and held the artery in its natural position. Haemolymph flow rates ranging from 0 to 22 ml min⁻¹ were simulated using a pump head, which ensured that pulses of haemolymph were delivered at physiological frequency, again showing a strong correlation between perfusion rate and flowmeter output (r²=0.965; β=0.972).

Cardiac output was calculated by summation of arterial flows from trials in which all arterial systems had been successfully instrumented. Heart stroke volume was calculated by dividing by f_H for the same trials. All rates except f_H and f_SC were normalised to crab wet mass.

Data were analyzed for the effects of and time in a nested two-way analysis of variance (ANOVA). ANOVAs showing significant effects of (P<0.05) were further analyzed using Tukey’s HSD test with α=0.05. Unless otherwise stated, all data are shown as mean ± 1 S.E.M.

Fig. 2 illustrates the changes in f_H, and recorded from one crab during the 24 h period following instrumentation. The measurements at 0 h were obtained as soon as possible after the animals had been placed in the experimental chamber and all recording leads had been connected. The initial values of (52 ml kg⁻¹ min⁻¹) and (0.6 ml kg⁻¹ beat⁻¹) were high relative to those obtained after the animals had been allowed to recover fully, but f_H was no higher than the typical resting value for this variable. and fluctuated greatly over the recovery period but showed an overall declining trend, reaching minima of 19.7 ml kg⁻¹ min⁻¹ and 0.3 ml kg⁻¹ beat⁻¹, respectively, at 21 h. f_H increased steadily over the first 12 h of recovery, then also reached a minimum (70 min⁻¹) at 21 h. The extremely low values of and recorded at 1 h were a result of spontaneous cessation of flow through all vessels except for the anterolateral arteries just prior to measurement and were atypical of the values recorded before and after the 1 h sample. The values for 21 h were recorded just prior to ‘sunrise’ in the laboratory when the lights were switched on. High levels of and f_H were maintained for at least 3 h following this visual disturbance.

After 24 h of recovery from instrumentation, all crabs showed low, variable rates of arterial haemolymph flow, low f_H and intermittent ventilation with frequent reversals (Table 1, Fig. 3). Resting haemolymph flow through the sternal and anterolateral arteries was generally the highest. Control rates of haemolymph flow through the anterior and posterior aortas and hepatic arteries were comparatively low, although spontaneous periods of high perfusion of the latter arteries occurred in conjunction with vastly reduced flow through all other systems (not shown).

Following measurement of control levels of all variables for settled crabs, the of the water in the experimental chamber was reduced to 4.63 kPa over a 10 min period. f_H decreased by 20 % within 30 min of onset of hypoxic exposure (F=37.82, P<0.01) and bradycardia continued until was returned to normoxic levels (Fig. 4). As soon as aeration resumed in the experimental chamber, f_H returned to its pre-hypoxic rate. The increase in f_H accompanying restoration of normoxic conditions usually occurred within 60 s, in contrast to the more gradual decline in f_H associated with declining ⁠. f_H was linearly related to at the three hypoxic levels tested (Fig. 5) and reached a minimum of 38.2±4.0 min⁻¹ at 2.96 kPa. Despite the reduction in f_H, appeared to increase slightly during the 6 h treatment period (F=0.86, P>0.05) owing to a marked increase in cardiac (F=12.79, P<0.01) over the same interval (Fig. 4), although the change in was statistically insignificant. showed an inverse-linear relationship to over the tested range from 4.63 to 2.96 kPa and, at the lowest experimental ⁠, was 2.6 times higher than its control value (Fig. 5). Following restoration of normoxic conditions, remained elevated over the control value for at least 1 h; this, combined with a slight overshoot in f_H, resulted in a further increase in during the first hour of recovery (Table 1, Fig. 4).

f_SC increased from a control rate of 82.1±4.3 min⁻¹ to a maximum of 125.9±8.9 min⁻¹ 4 h after the onset of hypoxic exposure (F=13.56, P<0.01) at a of 3.77 kPa (Fig. 5). This elevated rate was maintained, even at the 2.96 kPa treatment level, suggesting that sufficient O₂ was delivered to the ventilatory pump to allow it to metabolise aerobically despite the reduced ⁠. In most cases, f_SC remained elevated for several hours following restoration of normoxic conditions. Along with a small but significant elevation in haemolymph L-lactate concentration (F=3.75, P<0.05; Table 2), this post-treatment tachypnea indicates that the animal did incur an O₂ debt in some tissues during hypoxic exposure.

Haemolymph flow through the hepatic and anterolateral arteries showed decreasing trends with declining (Fig. 6), although there was large inter-crab variability and the changes were statistically insignificant. Flow through the right hepatic artery appeared to be slightly reduced at 2.96 kPa with respect to the control rate (F=0.88, P>0.05) and haemolymph flow through the left anterolateral artery fell from a control rate of 3.5±0.4 ml kg⁻¹ min⁻¹ to 2.1±0.5 ml kg⁻¹ min⁻¹ (F=1.77, P>0.05) at 2.96 kPa. The rate of haemolymph flow through the anterior aorta remained relatively stable (F=0.11, P>0.05) and was not obviously affected by hypoxic exposure (Fig. 6).

Haemolymph flow rate through the posterior aorta decreased by 33 % with exposure to of 4.63 kPa, but then increased, from 0.19±0.04 to 0.36±0.09 ml kg⁻¹ min⁻¹, over the range from 4.63 to 2.96 kPa, although the changes were again statistically insignificant (F=1.42, P>0.05; Fig. 7). The rate of haemolymph flow through the sternal artery almost doubled in response to a of 2.96 kPa (F=3.06, P<0.05; Fig. 7) and reached a maximum of 7.3±1.2 ml kg⁻¹ min⁻¹ after 4 h of hypoxic exposure. Unlike the case with the posterior aorta, flow through the sternal artery (which supplies the scaphognathites and ventral nerve cord) was significantly higher than the control rate at all experimental O₂ levels.

The increase in haemolymph flow through the sternal artery corresponded to an increase in instantaneous distribution to this system from 36±6 to 63±5 % at the most severe level of hypoxia tested (Fig. 8). This increase was coincident with a decrease in haemolymph allocation to the anterolateral arteries, which received the greatest proportion of during both control and recovery periods, from 55±6 to 29±4 %. The percentage of delivered via the other arterial systems was comparatively low (Fig. 8) and showed little change in response to hypoxic exposure. distribution to the posterior aorta increased from 2±0.5 to 4±1 % at the most severe treatment level.

Following return to normoxic conditions, haemolymph flow through all arterial systems showed a rapid increase in rate (Fig. 3), overshooting control values in most cases. Haemolymph flow rate through the sternal artery continued to increase and remained elevated for several hours (Table 1), but all other flow rates returned to their pre-treatment values within 30 min. The distribution of approximately regained the pre-treatment pattern after 12 h of recovery (Fig. 8).

Voltage output from the pulsed-Doppler flowmeter system is linearly related to kHz Doppler shift of the reflected ultrasonic beam, which is in turn related linearly to blood velocity (Hartley and Cole, 1974). Zero flow can be determined electronically, eliminating the problem of calibration drift associated with electromagnetic flow meters (Milnor, 1989). In situ calibration of the system at physiological flow rates shows that there is a strong linear relationship between actual volume haemolymph flow and flowmeter output for both the carapace-mounted and internal transducer configurations. It is unlikely that a fully laminar haemolymph flow profile develops upstream from the measurement sites, given their close proximity to the ventricle and cardioarterial valve flaps; therefore, accurate determination of volume haemolymph flow is possible without correction for velocity profile. Unlike determinations based on the Fick principle or thermal dilution (see below), which estimate average over a relatively long period, calculation of by summing the minute haemolymph flow through all arteries leaving the heart allows immediate detection of adjustments in both and its distribution.

Following the 24 h post-operative period, ventilatory and circulatory parameters exhibited the low and variable rates characteristic of quiescent C. magister in normoxic sea water (McDonald et al. 1977; McMahon et al. 1979), indicating that the crabs had recovered from the handling necessary for preparation. The high variability recorded in each of these parameters resulted from periods of spontaneous cessation of ventilatory and cardiac activity (‘pausing behaviour’) as well as reversed ventilation and associated alteration of arterial flow rates. These types of behaviour are typical for aquatic decapods resting in well-aerated water (McMahon and Wilkens, 1972, 1977; McMahon et al. 1978; McMahon and Burnett, 1990) and may represent a mechanism for energy conservation when O₂ availability is high and metabolic demands are low (Burnett and Bridges, 1981).

Resting values obtained in the present study (9.8±1.6 ml kg⁻¹ min⁻¹; range 2.9–31.8 ml kg⁻¹ min⁻¹) were at the low end of the range of values reported for this and other decapod species in previous studies utilising either the Fick principle (Johansen et al. 1970; Burnett, 1979; McMahon et al. 1979; Bradford and Taylor, 1982; Wilkes and McMahon, 1982) or indicator dilution (Burnett et al. 1981) to estimate indirectly. Fick principle estimates ranging from 29.5 (Johansen et al. 1970) to 72±25 ml kg⁻¹ min⁻¹ (McMahon et al. 1979) have been reported for quiescent C. magister, whereas among other decapods values of 46 ml kg⁻¹ min⁻¹ for C. pagurus (Bradford and Taylor, 1982), 430 ml kg⁻¹ min⁻¹ for the spider crab Libinia emarginata (Burnett, 1979), 122 ml kg⁻¹ min⁻¹ for the lobster Homarus americanus (McMahon and Wilkens, 1975) and 45.8 ml kg⁻¹ min⁻¹ for the crayfish Orconectes rusticus (Wilkes and McMahon, 1982) have been reported. In a study utilising the thermal dilution technique (Burnett et al. 1981), estimates of ranging from 11 to 60 ml kg⁻¹ min⁻¹ were obtained for C. magister (data obtained by vectorisation of their Fig. 2E). The relatively low reported for this species, both in previous investigations and in the present study, may be due to the high haemolymph haemocyanin concentration as well as to the low blood convection requirements (Burnett, 1979). Cardiac values obtained in the present study (0.15±0.03–0.41±0.06 ml kg⁻¹ beat⁻¹) also fell within the range of previously reported values (0.2–0.9 ml kg⁻¹ beat⁻¹) of for this species determined using the thermal dilution method (Burnett et al. 1981).

In the present study, crabs were completely unfettered within the experimental chamber and no additional manipulation was required to obtain continuous recordings of all variables. In addition, minimal surgery was required because all but one of the pulsed-Doppler transducers were mounted outside the body wall. These factors may have contributed to the comparatively low values reported here, since both Fick principle and thermal dilution methods cause repeated disturbance to the animal during the course of determination. Although attempts to minimise disturbance of the animals were made in the investigations employing these techniques, both methods of estimation require intervention by the experimenter, either for blood sampling or for indicator injection, and in many cases the animals were heavily restrained. Responses to even modest visual stimuli are dramatic in these animals (see Fig. 2), and tactile stimuli such as handling and vibration cause cardiac and ventilatory activity to become greatly elevated above resting levels (Cumberlidge and Uglow, 1977).

Recent Fick principle determinations have usually employed a mask fitted over the anterior of the animal for measurement of branchial water flow rate and exhalant ⁠. Crabs fitted with such a mask took much longer to acclimate to experimental conditions and often failed to attain respiratory patterns typical of unstressed crabs (McDonald, 1977). Prebranchial sampling of haemolymph from decapods also causes major disturbance to the animals (McMahon et al. 1978; Bradford and Taylor, 1982), and repeated sampling of postbranchial haemolymph, although less overtly intrusive, has been linked to haemolymph acidosis in lobsters (McMahon et al. 1978). Chronically implanted catheters, although potentially allowing simultaneous sampling of pre-and postbranchial haemolymph with minimal disturbance, are not feasible because of the rapidity of clotting and the lack of anti-coagulatory agents for crustaceans. Chronic intraventricular catheter implantation, required for thermal dilution determinations of ⁠, may cause massive clotting and extensive damage to the heart (McDonald, 1977; C. N. Airriess, personal observation), although immediate visual and tactile disturbance of the animals may be less problematic than for Fick principle estimations of ⁠. The decrease in cardiac activity observed during the 24 h post-operative period in the present investigation suggests that truly ‘resting’ levels may be attained by these freely moving, undisturbed crabs and also illustrates the huge range of variability in (7.6–108.0 ml kg⁻¹ min⁻¹) possible for a single animal.

Hypoxia-induced hyperventilation was maintained even at the lowest O₂ level tested, suggesting that the P_C for ventilatory performance is lower than 2.96 kPa for C. magister. Although heart rate decreased, cardiac output increased slightly during hypoxic exposure, which also suggests a high degree of oxy-independence. Previous reports for this species (Johansen et al. 1970) predicted that survival for periods of 30 min would be possible at O₂ partial pressures as low as 3.75 kPa. These authors speculated that longer exposure to such low O₂ levels would be lethal. C. pagurus has been shown to have a P_C for ventilatory performance of 5–7.5 kPa (Bradford and Taylor, 1982). The present study indicates that C. magister is much more independent of environmental O₂ levels than previous conspecific evaluations or comparisons with C. pagurus suggest.

Homarid lobsters also respond to reduced environmental with hyperventilation and bradycardia (McMahon and Wilkens, 1975; Butler et al. 1978). The increase in f_SC reported for H. americanus was equivalent to that observed for C. magister in the present investigation at levels as low as 3.75 kPa, but below that level the ventilation rate fell dramatically in lobsters whereas it was maintained in C. magister. The higher tolerance of this crab for hypoxic exposure compared with C. pagurus and H. americanus may be related to the natural habitat of each species. Of these decapods, C. magister probably encounters the most severe levels of environmental hypoxia in its natural habitat. C. pagurus inhabits rocky subtidal habitats (Pearson, 1908) and H. americanus lives under rocks and in burrows on rocky, sandy or muddy substrata (Scarratt, 1984). Preliminary studies (C. N. Airriess and B. R. McMahon, unpublished results) indicate that the of water at the sediment interface of muddy estuaries frequented by C. magister may fall below 1.25 kPa during extremely low tides.

The time course of hypoxic exposure was much shorter in the present experiment than in some previous investigations, but was chosen to emulate naturally encountered conditions based on the tidal cycle of the north-eastern Pacific Ocean. The shorter period of exposure may have contributed to the great hypoxia-tolerance observed in this study.

Maintained hyperventilation during the most severe level of hypoxia suggests that enough O₂ was delivered to the scaphognathite muscles to maintain aerobic metabolism in these active tissues. The low level of L-lactate production during the exposure period also indicates that anaerobic metabolism was minimal. For comparison, the haemolymph lactate concentration of C. magister increased from 0.7 to 11.1 mmol l⁻¹ following 20 min of enforced exercise (McDonald et al. 1979). The muscles responsible for scaphognathite movement are highly vascularised by very fine branches of the ventral thoracic artery (B. J. K. DeWachter, C. N. Airriess and I. J. McGaw, unpublished observations), which is the anterior branch of the sternal artery. Thus, a shift in the majority of distribution to the sternal artery increases perfusion of the scaphognathites, reflecting the aerobic needs of these organs and accounting for their ability to remain active during the deepest hypoxic level tested. Interestingly, maximum haemolymph flow through the sternal artery during the treatment period coincided with the maximum recorded scaphognathite beat frequency, 4 h subsequent to the onset of hypoxia. The ventral nerve cord, a major component of the central nervous system, is supplied with haemolymph originating from the sternal artery (Fig. 1B); therefore, a change in haemolymph distribution to favour perfusion of the sternal artery may help to protect the central nervous system (CNS) from the effects of hypoxic exposure. Haemolymph from the sternal artery is also delivered to the pereiopods, which may have been important sites of gas exchange in primitive aquatic crustaceans and retain a vital role in allowing modern decapods to escape locally hypoxic waters.

Since there is no evidence for the presence of vasoconstrictive musculature in decapod crustacean blood vessels (Maynard, 1960; Martin et al. 1989; Shadwick et al. 1990), mechanisms for alteration of regional haemolymph distribution almost certainly involve the cardioarterial valves. These muscular semilunar valves receive separate innervation from the central nervous system (Alexandrowicz, 1932) and respond differentially to monoamine and peptide neurohormones in isolated lobster (Panulirus japonicus) heart preparations (Kuramoto and Ebara, 1984, 1989). Several neurohormones have been identified in the pericardial organs of decapods, which are ideally situated to release their product directly into the path of venous haemolymph return to the heart (Alexandrowicz, 1953). These include the monoamines 5-hydroxytryptamine, octopamine and dopamine (for a review, see Beltz and Kravitz, 1986) and the peptides proctolin (Belamarich and Terwilliger, 1966), crustacean cardioactive peptide (Stangier, 1991), F1 and F2 (Mercier and Russenes, 1992).

Neural control of the cardioarterial valves has not been investigated in vivo, but changes in the distribution of in intact C. magister infused with the pericardial neurohormones octopamine and proctolin (Airriess and McMahon, 1992; McGaw et al. 1992) closely parallel the responses observed in isolated lobster hearts. The monoamine octopamine relaxes the muscle of the anterior cardioarterial valves and causes contraction of the posterior valve muscle of lobster hearts (Kuramoto and Ebara, 1984) and shifts the distribution of to the anterior arterial systems in intact crabs (Airriess and McMahon, 1992). Conversely, the peptide proctolin causes contraction of anterior valve muscle in lobster hearts (Kuramoto and Ebara, 1984) and increases the proportion of directed through the sternal artery and posterior aorta of intact C. magister (McGaw et al. 1992). Infusion of proctolin directly into the heart in an in situ preparation of the crab Carcinus maenas with an intact CNS causes bradycardia in conjunction with a marked increase in ventricular pressure (M. A. Saver, C. N. Airriess and J. L. Wilkens, in preparation). The magnitude of these cardiac adjustments is very similar to that of the responses to hypoxic exposure shown in the present investigation. Proctolin, therefore, is a strong candidate for the effector substance responsible for hypoxia-induced changes in and haemolymph redistribution in decapod Crustacea.

It is clear that C. magister is well able to cope with the severe levels of environmental hypoxia it may face on a regular basis. Physiological mechanisms, such as the ability selectively to alter haemolymph distribution to maximise O₂ supply to the limbs, ventral nerve cord and respiratory pumps, combined with hyperventilation and maintenance of via increased allow this crab to maximise O₂ uptake during hypoxic exposure and to thrive in euryoxic habitats that may be intolerable to other decapods. Techniques making possible the determination of distribution provide the basis for revision of the concept that crustaceans do not increase tissue perfusion during hypoxic exposure (Herreid, 1980). At least decapods, and possibly isopod crustaceans, which have complex neural control of the cardioarterial valves (Kihara and Kuwasawa, 1984), are capable of selectively increasing haemolymph perfusion of specific tissues to compensate for environmental stress. Control of blood distribution at the level of the heart is a mechanism which performs a function analogous to the peripheral control of tissue perfusion in vertebrates and suggests additional complexity in the cardiovascular system of decapod crustaceans.

The authors would like to Dr G. B. Bourne for the use of equipment and comments on the manuscript. T. A. Rawlings supplied useful discussion and criticism. Financial support was provided by NSERC grant A5762, NSERC and AHFMR fellowships to C.N.A. and by the Bamfield Marine Station.