Cardiovascular Responses of the Red-Blooded Antarctic Fishes Pagothenia Bernacchii and P. Borchgrevinki

There are over 200 species of fish that are known to live south of the Antarctic convergence (Andriashev, 1987; Eastman, 1991). These fish inhabit a stenothermal environment in which temperatures typically range from -1.9 to 1.0°C. In this environment, the fish benefit from an abundance of food, limited predation from piscivorous fishes and, in terms of oxygen delivery, a high oxygen solubility in water and plasma. While the metabolic demand for oxygen is reduced at low temperatures in ectothermic animals, there is evidence for partial temperature compensation in notothenioid fish (Holeton, 1970, 1972; Hemmingsen and Douglas, 1970, 1972; Macdonald et al. 1987; Forster et al. 1987).

The cardiovascular adaptations that permit antarctic fish to live in these frigid waters are of particular interest. Without any compensating mechanism, the hearts of antarctic fish might be expected to perform proportionately more work, since blood, in common with other fluids, has an increased viscosity at low temperatures (Graham and Fletcher, 1985). Antarctic fish reduce the effective blood viscosity in two ways: they have a high lipid content in the plasma (Rickenbach, 1990) and a reduction in the number of circulating red blood cells. A small number of antarctic fish species (17 species of channichthyids) have haemoglobin-free blood which reduces blood viscosity to 0.3–0.4Pas at 0°C (Hemmingsen and Douglas, 1972; Wells et al. 1990). In other antarctic species haemato-crit values (8–18%) are lower than in temperate species of teleosts (Macdonald et al. 1987). One of these species, Pagothenia borchgrevinki, shows an unusually large increase in haemato-crit when stressed (Davison et al. 1988; Franklin et al. 1991), suggesting that there is an energetic advantage to sequestering large numbers of cells in the spleen to be released only when needed. In contrast, haematocrit values between 25 and 39% were reported for two species of antarctic nototheniids (Notothenia neglecta Nybelin and N. gibberifrons Lonnberg) and for a bathydraconid (Parachae-nichthys charco ti Vaillant) (Holeton, 1970).

Studies of two species of channichthyids, Chaenocephalus aceratus and Pseudo-chaenichthys georgianas, have described some of the cardiovascular adaptations associated with the absence of haemoglobin from the blood. C. aceratus has a high blood volume, estimated to be 2–3 times greater than that of other teleosts (Hemmingsen and Douglas, 1970). Because the oxygen content of the blood is low while oxygen uptake is not, it was predicted that cardiac output (Q̇) would be extremely high (Holeton, 1970). Most recorded values of Q̇ in haemoglobin-free fish are high, but unfortunately they show a large variation and may not be completely free of measurement error. Reported values of Q̇ determined by the Fick principle are 61mlmin⁻¹kg⁻¹ (Holeton, 1970, 1972), 119 ml min⁻¹ kg⁻¹ (Hemmingsen et al. 1972) and 94–104 ml min⁻¹ kg⁻¹ (Hemmingsen and Douglas, 1977). However, determinations of Q̇ using the Fick principle may be erroneously high in fish because of cutaneous gas exchange, gill oxygen consumption and blood shunts within the gills (Randall, 1985). In the case of C. aceratus, skin oxygen consumption appears to be high (Hemmingsen and Douglas, 1970). The reported Q̇ values for C. aceratus based on measurements using electromagnetic flow probes are 20–30 ml min⁻¹ kg⁻¹ [from Hemmingsen et al. (1972), and assuming equal flow distribution to all gill arches] and 67 ml min⁻¹ kg⁻¹ (Hemmingsen and Douglas, 1977); for Pseudochaenichthys georgianas Q̇ is 68 ml min⁻¹ kg⁻¹ (ingsen and Douglas, 1977) at heart rates of 15–20 beats min⁻¹. The relatively large ventricles of channichthyids, 0.3% of the body mass in C. aceratus (Holeton, 1970) and 0.39% in Chionodraco hamatus (Tota et al. 1991b), compared with those of temperate teleost species (Santer, 1985; Farrell, 1991) are also consistent with a high Q̇, especially since ventral aortic blood pressure (P_VA) is relatively low (mean=2.0kPa).

In view of the high Q̇ and low P_VA in C. aceratus, Holeton (1970) also predicted a cardiovascular adaptation in the resistance vessels. The dimensions of the resistance vessels (arterioles) are not known, but the diameter of the capillaries in the axial muscle of C. aceratus was found to be three times greater than that in other teleosts (Fitch et al. 1984).

The few cardiovascular studies of antarctic fish possessing red blood cells indicate that routine oxygen consumption is similar to that of the channichthyids, but that the former tolerate aquatic hypoxia better, because they have a higher venous oxygen reserve. Their ventricle is apparently not unusually large (0.06–0.10% of body mass) (Holeton, 1970; Macdonald et al. 1987).

Very little is known about cardiovascular control in red-blooded antarctic fishes, and there are conflicting reports on the haemoglobin-free fishes. Our objective was to make measurements of Q̇ and P_VA in two species of nototheniid fishes that possess red blood cells, Pagothenia bernacchii and P. borchgrevinki. Their responses to environmental hypoxia, swimming, acute temperature change and pharmacological intervention were studied to obtain information on cardiovascular control.

Two species of nototheniid fish, Pagothenia bernacchii (Boulenger, 1902; also Notothenia bernacchii, Trematodos bernacchii and Pseudotrematomus bernacchii) and P. borchgrevinki (Boulenger, 1902; also Notothenia hodgsoni), were collected from McMurdo Sound in the Ross Sea, Antarctica, during the 1990 summer season. The cryopelagic P. borchgrevinki were caught over deep water by lowering handlines through a hole in the annual ice. P. bernacchii were caught in traps placed on the sea floor in 30 m of water. Fish were kept in sea water (−0.5°C) at Scott Base until they were transported by air to the University of Canterbury, Christchurch, New Zealand. Prior to experimentation, fish were maintained for 2 months in an aquarium system at 0°C and in continuous light, as this corresponded to the summer light regime at Scott Base. All animals were fed once per week.

All surgery was performed in a cold room with an air temperature of 5°C. The anaesthetics were maintained at 0°C. Following anaesthetization in a 1:10000 solution of MS-222 (Sandoz) in sea water, the animals were placed on the operating table and the gills were continuously irrigated with a lighter dose of anaesthetic (1:20000 MS-222). The ventral aorta was chronically cannulated with PE30 tubing via the afferent branchial artery of the third right gill arch (Axelsson and Nilsson, 1986). The tubing was filled with heparinized (50i.u.ml⁻¹) 1.4% NaCl.

The ventral aorta was exposed through a medial incision, and a cuff-type Doppler flow probe was fitted around the vessel. Care had to be taken to avoid a large vein running immediately ventral to the aorta. Miniature Doppler flow probes were preferred both to measurements by the Fick principle and to electromagnetic flow probes as a means of measuring Q̇ in these animals. There are uncertainties using the Fick principle in fish (see Introduction), and electromagnetic flow probes are much too large for fish as small as these. While Doppler flow probes provide reliable information on zero flow and the relative changes in flow, they are difficult to calibrate. To improve the accuracy of our determinations, we carefully sized the probes to the dimensions of the ventral aorta (diameter approximately 1 mm) and calibrated each probe in situ using the range setting employed in the experiment.

To estimate absolute values for blood flow in the ventral aorta, a careful in situ calibration of each flow probe was performed. Each animal was killed by over-anaesthesia and placed ventral side up on a tray of ice. The heart was exposed and the ventricle was cannulated with PE50 tubing for perfusion of the ventral aorta. Perfusion was effected with a Watson Marlow peristaltic pump and heparinized 1.4% NaCl to which blood had been added to give a haematocrit of around 4–5%. Flow rates were determined gravimetrically for at least four different flow levels. The relatively small numbers of absolute values for Q̇, stroke volume and total vascular resistance given in Table 1 represent animals in which linear regression analysis of the data showed a good correlation within the range that was recorded in vivo (r>0.9 in all cases). In cases where the calibration was unsuccessful, only relative changes in Q̇ are presented. However, because relative changes in Q̇ are independent of the calibration of the flow probe, we are confident of the relative changes in Q̇, stroke volume and total vascular resistance that are presented and the information that these values provide about cardio-vascular control.

After surgery, each animal was transferred to an experimental chamber where it was allowed to recover for at least 24 h. The experimental chamber consisted of a small closed-box respirometer (1.91) containing an air-stone to aerate the water continuously (or supply gas mixtures during the hypoxia experiments). The cannula and flow probe leads were passed out of the chamber through a small hole in the lid. The chamber was contained in a constant-temperature waterbath which maintained its temperature at 0°C.

During the experiment the ventral aortic cannula was attached to a Bell & Howell pressure transducer (model 4–327) which was calibrated against a static water column. The Doppler flow probe was connected to an Iowa University flowmeter. Oxygen tension in the water was monitored using an IL1302 oxygen electrode connected to a Strathkelvin meter. Blood flow and pressure and oxygen partial pressure of the sea water were continuously monitored on a Devices four-channel recorder and data were continuously sampled with a computerized data acquisition system.

P. bernacchii were allowed to recover from surgery for at least 24 h. Each experiment was started with a recording of the resting variables, which are presented in Table 1 (untreated). Sequential injections of adrenergic and cholinergic agonists and antagonists were used to assess the adrenergic and cholinergic control of the heart and systemic vasculature. The following sequence of injections was used for all animals: adrenaline (10 nmol kg⁻¹) (this is both an α- and a β-adrenoceptor agonist and usually produces a systemic vasoconstriction in fish, Nilsson, 1983); atropine (1.2 mg kg⁻¹) to block the muscarinic receptors; adrenaline (10 nmol kg⁻¹); propranolol (2.5 mg kg⁻¹) to block β-adrenoceptor responses.

All drugs were injected via the ventral aortic cannula through a four-way valve. Sham injections of saline were made in order to determine the effect of injected volumes of saline: these were found to have no effect on the recorded variables.

The choice of doses of drugs to be administered was based on experiments on other species of teleosts (Axelsson and Nilsson, 1986; Axelsson et al. 1988; Axelsson, 1988). However, the dose of adrenaline given was approximately ten times lower than that used previously. We employed this lower dose (10 nmol kg⁻¹) as it provoked a significant response without causing a long-lasting increase in blood pressure. Following atropine and propranolol administration, the fish were given at least 1h to recover from any direct actions of the drugs before the blocking activity of the drug was assessed.

Animals were exposed to hypoxic water before administration of any drugs and then again after injection of atropine. The initial of the sea water in the experimental chamber was about 20–21 kPa. Nitrogen and air mixtures were introduced into the experimental chamber using a Wösthoff gas-mixing pump (Bochum, Germany) to achieve a final of approximately 6.7 kPa (50 mmHg). The total duration of each hypoxia exposure was approximately 40 min, with the time taken to reach the nadir being about 25 min, and a further 15 min elapsing before normoxia was restored. Recordings were made continuously during hypoxia and recovery. The hypoxic level of 6.7 kPa was chosen to allow comparison with previous studies of icefish (C. aceratus). Hemmingsen and Douglas (1970) showed that these animals became agitated at this while Holeton’s (1970) fish died below 6.7 kPa. Davison et al. (1990) have shown that oxygen becomes limiting below about 3.9 kPa in P. borchgrevinki.

P. bernacchii were exposed to temperatures of up to 5°C by heating the water surrounding the experimental chamber. The rate of increase within the chamber was approximately 0.1° min⁻¹. Aeration of the water within the chamber ensured complete mixing. Animals were tested at the end of all other experimental work (i.e. following atropine and propranolol injections). The fish were then allowed to recover from the drugs for several days after which time the heart rate had returned to control values, and then they were subjected to the temperature experiment again. Several fish that had not been previously exposed to drugs were also tested.

Following surgery, these fish were allowed to recover for 24 h in a Blazka-type swimming tunnel. Recordings of resting values were made as for P. bernacchii. Ventilation rate was also noted. Fish were then exercised at approximately 20cms⁻¹ (1 body lengths⁻¹). Continuous recordings were made during the exercise bout and during recovery. Following recovery, each fish was injected with adrenaline. After a suitable recovery time (1–2 h) atropine was administered, and after 1 h the fish were exercised again. Propranolol was then injected, and after a further hour an attempt was made to exercise the fish for a third time.

Fish were subsequently allowed to recover for several days in the aquarium. Heart rate was monitored to assess recovery from atropine. Following this, the animals were transferred to the chamber used for P. bernacchii and their responses to hypoxia and increased temperature were tested as described above.

A computerized data acquisition system (Toshiba model 3200SX running Labtech Notebook) was used for automatic sampling of the data. Further processing and statistical treatment were performed with an analysis program written by P. Thorén (Hässle AB, Sweden).

Evaluation of the statistical significance of differences between observations was made using a Wilcoxon signed-rank test (two-tailed). Results were considered significant if P was less than 0.05. To correct for repeated tests, a sequential rejective Bonferroni test (Holm, 1979) was used to eliminate, as far as possible, the chance of discarding any true null hypothesis (Type I error).

The relative cholinergic and adrenergic tone on the heart was calculated using a simplification of the method used by Lin and Horvath (1972) (see Axelsson et al. 1988). The heart rate after adrenergic and cholinergic tone had been blocked is referred to as intrinsic heart rate. Total vascular resistance (R_tot) was estimated from P_VA/Q̇ assuming a venous blood pressure of zero. Stroke volume (Vs) was calculated from Q̇/f_H

Twelve fish with average body mass of 50.5±5.1 g (mean±s.E.M.) and ventricle mass 54±8mg were used. The relative mass of the P. bernacchii ventricle was 0.106±0.004%. The atrium (25±4mg) was approximately half the size of the ventricle (relative atrial mass=0.05±0.003%). The haematocrit was 15.4±0.9%.

Control Q̇ was 17.6±7.9mlmin⁻¹kg⁻¹, f_H was 10.5±0.9 beats min⁻¹ and P_VA was 3.09±0.23kPa (Table 1). We were careful to allow sufficient time for fish to acclimate to the experimental chamber (as indicated by a low and steady heart rate) before measurements were made. Heart rates in disturbed fish were much higher than the control value.

Atropine increased heart rate to 27.4±1.0beatsmin⁻¹, but Q̇ and P_VA were unchanged (Table 1). Consequently, stroke volume was 60% lower than the control stroke volume. Subsequent treatment with propranolol reduced heart rate to 21.7±0.9beatsmin⁻¹.

The resting heart rate of P. bernacchii at 0°C was 10.5±0.9 beats min⁻¹(N= 12; Table 1). This was the result of an inhibitory cholinergic tone of 80.4±6.4% and an excitatory adrenergic tone of 27.5±5.1%. The intrinsic heart rate was 21.7±0.9 beats min⁻¹.

During hypoxia, Q̇ was maintained down to a of 6.7 kPa (Fig. 1). There was no sustained hypoxic bradycardia, a response that is typically observed in temperate species of teleosts (Randall, 1970; Satchel!, 1991). In fact, mean heart rate was 2 beats min⁻¹ higher at 6.7 kPa than in normoxia. In some fish, heart rate had a phasic pattern where the beat-to-beat interval was prolonged for 1–3 beats and then followed by a period with shorter than average beat-to-beat intervals (Fig. 2). P_VA and total vascular resistance (P_tot) were unchanged down to 8.0 kPa, but increased at a of 6.7 kPa (Fig. 1).

During the recovery from hypoxia, f_H increased to a maximum of 18 beats min⁻¹, stroke volume was gradually restored to control values and Q̇ was significantly elevated (Fig. 1). P_VA decreased during the later phase of recovery as total vascular resistance (P_tot) decreased significantly.

After atropine treatment, Q̇, Vs and P_VA increased progressively as decreased to 6.7 kPa. The increase in Q̇ occurred even though the experiment began with an already elevated Q̇ as a result of the atropine pretreatment (Fig. 1). This response contrasts with that of the non-atropinized fish, in which there was no change in Q̇. Despite this difference in the response to hypoxia, it is interesting that in both the atropinized and non-atropinized fish Q̇ peaked during recovery (Fig. 1). f_H was constant as decreased except at 6.7 kPa Heart rate was much more regular in the atropinized fish, with no prolonged beat-to-beat intervals. As with the non-atropinized fish, P_tot decreased significantly during recovery.

The water temperature was increased from 0 to 5°C over a period of 40 min (Fig. 3). There was no marked increase in the heart rate of untreated fish until the temperature reached approximately 2.5°C. Above this temperature, heart rate rose rapidly to reach values similar to those recorded in the atropine/propranolol-treated animals. As a result, the Q₁₀ for resting fit was greater than 4 over this temperature range. After the pretreatment with propranolol and atropine, the heart was beating at an intrinsic rate of 21.7 beats min⁻¹ at 0°C. Heart rate increased progressively to a value of 31.5 beats min⁻¹ at 5°C (Fig. 3). The Q₁₀ for the intrinsic heart rate was 1.96. The increase in f_H was accompanied by a progressive decrease in Vs, and Q̇ increased by only 10%. P_VA was relatively unchanged in atropinized fish up to 3°C but declined significantly thereafter along with P_tot.

Injection of 10 nmol kg⁻¹ adrenaline into the ventral aorta produced immediate increases in R_tot, P_VAand f_H (Fig. 4). At first, Q̇ was unchanged as Vs decreased. After the first 2min, stroke volume began to recover while heart rate continued increasing so that Q̇ had increased by 30% after 3.5 min. Total vascular resistance peaked between 1.0 and 1.5 min after injection, and decreased thereafter (Fig. 4). Therefore, the pressor response to adrenaline initially resulted from a vasocon-striction and was prolonged by an increase in Q̇.

When adrenaline was injected after atropine pretreatment, the changes in P_VA and P_tot were qualitatively similar but quantitatively slightly larger than the corresponding responses measured before the injection of the blocker (Fig. 4). However, the increase in Q̇ was due to a change in stroke volume 2 min after adrenaline injection as f_H was unchanged (Fig. 4).

Three fish with a mean body mass of 64 g and a ventricle mass of 100 mg were used. The mean relative ventricle mass of the P. borchgrevinki heart was 0.156%. The atrium (30 mg) was approximately one-third of the size of the ventricle (relative atrial mass=0.042%). In resting fish prior to swimming, the ventilation rate was 26 ventilations min⁻¹ and Q̇ was 29.6 ml min⁻¹ kg⁻¹ with a heart rate of 11.3 beats min⁻¹ (Table 1). The haematocrit was 17.7±2.3%.

Three fish were swum at 20 cm s⁻¹ (approximately 1 body lengths⁻¹) for 6-10min, after which they appeared exhausted. P. borchgrevinki are labriform swimmers. The sculling rate of the pectoral fins during swimming ranged from 70 to 105 strokes min⁻¹. Ventilation frequency increased approximately twofold to 45–52 ventilations min⁻¹. Q̇ increased immediately at the onset of swimming and plateaued after 3.25 min at a value 75% higher than the control value (Fig. 5). f_H also increased immediately and had almost doubled after 2.25 min of swimming. Vs did not show any changes during swimming. Total vascular resistance (P_tot) decreased, which ensured that there was little change in P_VA during swimming (Fig. 5). In two of the fish, the swimming challenge was repeated after a 24h recovery period. The cardiovascular responses were similar to those during the first swim (not shown).

At the end of the exercise period, haematocrit was slightly higher in each of the three fish (mean value=20.4±2.3%). Venous oxygen content was measured in two fish before and immediately after swimming. It averaged 1.58 vol% before and 0.48 vol% afterwards, a more than threefold decrease.

Recovery from the bout of swimming was characterized by a return of all measured cardiovascular variables towards their resting values. By 15 min post-exercise, f_H was still above the pre-exercise level and Q̇ was still above the control level.

Pretreatment of P. borchgrevinki with atropine increased heart rate and decreased P_VA (Table 1). The fish swam for shorter periods after atropine treatment (3–6min). Resting heart rate was 11.3±2.9beatsmin⁻¹ at 0°C. The intrinsic heart rate was twice this value (23.3±0.9 beats min⁻¹) (Table 1). Resting heart rate was maintained by an inhibitory cholinergic tone of 54.5±12% and an excitatory adrenergic tone of 3.2±2.6%. The fish would not swim after propranolol pretreatment.

Three P. borchgrevinki were exposed to progressive hypoxia after a 2- to 4-day recovery from the swimming experiment. Recovery from the previous iments was indicated by the low heart rates (15 beats min⁻¹). The major difference between P. borchgrevinki and P. bernacchii was the effect of hypoxia on f_H. Two fish showed a bradycardia and only one showed a tachycardia. The individual with the highest resting f_H showed the biggest fall in/H (41%) (Fig. 6). Recovery from the hypoxic exposure was accompanied by a 50% increase in Q̇, with both heart rate and stroke volume increasing.

Previous studies of cardiovascular physiology in antarctic fish have been performed in the Antarctic. Our approach was to bring the fish to an environment where we had greater access to instrumentation. This meant that the fish were transported considerable distances and held in recirculating sea water collected at Lyttleton Harbour near Christchurch, New Zealand, and cooled to 0°C. mens of P. bernacchii have lived for more than 1 year in this system and will take food. It has proved more difficult to keep P. borchgrevinki alive and only three fish were able to swim following surgery.

Because the fish were small, efforts were made to minimize disturbances. The various measuring devices undoubtedly encumbered the animal, but the leads from the Doppler probes were thin and the tubing used for ventral aortic cannulation was the smallest possible that would permit accurate records of pulsatile pressure. Also, both species, including the cryopelagic P. borchgrevinki, spend most of their time lying quietly on the bottom of the aquaria. Thus, the leads and cannulae were not usually a hindrance to the animals. However, during swimming, the drag effects of the leads needed to be taken into account as they would have necessitated a greater swimming effort than that predicted by the water velocity. The frequency and volume of blood samples were kept to a minimum to avoid a major haemodilution. This precluded sampling for catechol-amine analysis, which would have been useful to assess whether the animals were stressed.

The measurement of total vascular resistance does not allow any distinction to be made between changes in gill and systemic resistances. A change in R_tot could reflect a change in either the gill circulation or the systemic circulation or some combination of the two.

Studies of C. aceratus have been used to exemplify cardiovascular adaptations to the antarctic environment. We can now compare red-blooded and haemoglobin-free antarctic fishes to ascertain whether there are any common adaptations. As in C. aceratus, P_VA. in P. bernacchii (3.09 kPa) and P. borchgrevinki (3.60 kPa) is low compared to that of temperate species of teleosts, even though the nototheniids have haematocrits that are not exceptionally low by comparison. Furthermore, since Q̇ is not unusually low in either species of nototheniid, a low vascular resistance is a common feature in all antarctic species examined thus far, whether or not they have red blood cells. C. aceratus has a lower P_VA (2.13–2.46 kPa) than the red-blooded nototheniids and, since Q̇ is higher in C. aceratus, its vascular resistance (35 Pa min kg ml⁻¹, using a P_VA of 2.13 kPa and Q̇ of 61 ml min⁻¹ kg⁻¹) is 5–6 times lower. Interestingly, the total vascular resistance of the antarctic fishes is more similar to that of hagfishes than to that of temperate species of teleosts (Axelsson et al. 1990).

It is possible to estimate whether the difference in R_tot between the red-blooded and haemoglobin-free notothenioids can be attributed entirely to differences in blood viscosity and blood flow rates. To do this, the component of vascular resistance due to viscosity must be separated from that due to vascular geometry. Wells et al. (1990) measured the viscosity of blood from P. bernacchii as 0.68Pas at −1.8°C at a shear rate of 225 s⁻¹ and a haematocrit of 15.4%. This was 80% higher than the blood viscosity of two haemoglobin-free channichthyids measured under similar conditions. Such a difference in blood viscosity clearly does not account for the five-to sixfold difference in R_tot. Thus, differences in vascular geometry must account for the majority of the higher P_tot of red-blooded compared to haemoglobin-free antarctic fish.

The low vascular resistance due to vascular geometry in red-blooded antarctic fish more than compensates for the high blood viscosity at 0°C and keeps cardiac work at a level comparable to that found in temperate species of teleosts. The resting cardiac power output is similar for P. bernacchii (0.82 mW g⁻¹ ventricle mass), P. borchgrevinki (1.05 mW g⁻¹ ventricle mass), C. aceratus (0.72 mW g⁻¹ ventricle mass), Oncorhynchus mykiss (1.7 mW g⁻¹ ventricle mass) and Plemitrip-terus americanas (1.0 mW g⁻¹ ventricle mass) (Farrell et al. 1988; Farrell, 1991). In Chionodraco hamatus, the peak power output of the heart in vitro exceeds 3 mW g⁻¹ ventricle mass, but stroke work dimishes at afterloads exceeding 3.0 kPa, for although there is cardiomegaly, the myocardium is of the spongy type alone (Tota et al. 1991a,b). To what degree red-blooded antarctic fish can maximally increase myocardial power output is not clear. The maximum increase in myocardial power output observed was less than 100% in P. bernacchii and approximately 100% in P. borchgrevinki. In contrast, the rainbow trout and sea raven can increase myocardial power output by three-to fourfold (Farrell, 1991).

Perhaps red-blooded antarctic fishes have a much lower scope for cardiac output than temperate species. In this regard it is interesting to note that recovery from exercise and hypoxia are accompanied by large reductions in R_tot, which tend to keep blood pressure and cardiac work low, even though cardiac output and haematocrit (in P. borchgrevinki) are elevated.

While cardiac power output is not unusually high in antarctic fishes, they have a relatively large heart. The relative ventricle mass in C. aceratus is 0.3% (Holeton, 1970; Harrison et al. 1991) and in Chionodraco hamatus it is 0.39% (Tota et al. 1991a): this is three times larger than in a rainbow trout at 10°C (0.1%) and is comparable to the largest values in teleosts (yellowfin tuna, 0.3%; skipjack tuna,

O. 4%) (Brill and Bushnell, 1991; A. P. Farrell, P. S. Davie, C. E. Franklin, J. A. Johansen and R. W. Brill, in preparation). The relative ventricle mass of P.bernacchii (0.1%) is as large as that in rainbow trout and is larger than that in temperate species of benthic teleosts (0.035–0.07%; Santer, 1985; Farrell, 1991). There are at least two explanations for the large heart. One relates to the observation that antarctic fish are characterized by relatively large stroke volumes (3–6 ml kg⁻¹ in C. aceratus ‘, 2.19 ml kg⁻¹ in P. borchgrevinki; 1.56 ml kg⁻¹ in P. bernacchii) compared with temperate species, in which the maximum stroke volume is more often 1 ml kg⁻¹ or less (Farrell, 1991). Assuming that Laplace ‘s law can be applied to the fish heart, a disproportionately thicker ventricle will be needed to generate the same wall tension with the greater radius (internal volume) associated with a higher stroke volume.

Another explanation for the larger ventricle in antarctic fish may be related to muscle hypertrophy which could compensate for the effect of temperature on muscle contractility. Low temperatures reduce cardiac contractility in fish. Hearts from rainbow trout acclimated to 5°C generate only half the maximum massspecific power output of hearts from fish acclimated to 15°C. However, cold-acclimated rainbow trout compensate for this by having a ventricle that is 50% larger and, as a result, total cardiac power output is only 34% lower than in the warm-acclimated fish (Graham and Farrell, 1989).

Antarctic fish are also characterized by low heart rates. Heart rates in P. bernacchii (10.5 beats min⁻¹) and P. borchgrevinki (11.3 beats min⁻¹) are lower than those reported for C. aceratus (14–17 beats min⁻¹). Steady heart rates as low as 3–6 beats min⁻¹ were observed in some P. bernacchii. The low heart rate in the red-blooded nototheniids was clearly a result of a high inhibitory cholinergic tone. In fact, the cholinergic tone found in P. bernacchii (80.4%) is the highest so far observed in a teleost fish. The low heart rate was not related to a low intrinsic heart rate. The isolated perfused heart of the icefish C. hamatus beat at 26 beats min⁻¹ (Tota et al. 1991a), suggesting that the channichthyids too may have a high inhibitory cholinergic tone. Factors other than catecholamines and acetylcholine may affect the heart, but the two major influences on the heart are the adrenergic and cholinergic systems (Farrell, 1991). Once their effects are blocked, it is likely that the intrinsic heart rate is revealed.

The intrinsic heart rate in P. bernacchii has a Q₁₀ value of 1.96. It is readily apparent that the intrinsic heart rate at 5°C (32 beats min⁻¹) is very similar to those reported for rainbow trout at 5°C (36 beats min⁻¹, Graham and Farrell, 1989) and for sea raven at 3°C (24 beats min⁻¹, Graham and Farrell, 1985). Thus, the intrinsic pacemaker frequency in antarctic fish is probably not very different from that of temperate species of teleosts. The excitatory adrenergic component of the control of heart rate in P. bernacchii is similar to that found in other teleosts (Axelsson et al. 1988). The low level of excitatory adrenergic tone in P. borchgrevinki, while unusual, does not preclude either adrenergic inotropic effects or adrenergic chronotropy under other conditions.

Several other aspects of cardiovascular control in addition to those discussed above were revealed by the present study. Perhaps the most unusual observation was the lack of bradycardia associated with hypoxia for P. bernacchii. Temperate and tropical teleosts typically respond to aquatic hypoxia with a reflex decrease in f_H (Randall, 1970; Daxboeck and Holeton, 1978; Smith and Jones, 1978; Bushnell and Brill, 1991; Farrell, 1991; Satchell, 1991), which may improve oxygen extraction by the working myocardium and oxygen transfer across the gill lamellae. The absence of hypoxic bradycardia in P. bernacchii may reflect the fact that f_H is already under a large inhibitory cholinergic tone, rather than the absence of the afferent arm of the reflex. Some fish, particularly those with the highest initial heart rates, showed periods of phasic increases and decreases in heart rate (Fig. 2). These phasic decreases in heart rate could be viewed as a form of hypoxic bradycardia followed by periods of escape from the vagal tone. The observation that atropinized fish showed a decrease in/H at the lowest values of (which continued into the recovery period) suggests a direct effect of hypoxia on the heart.

Of the three P. borchgrevinki that were exposed to hypoxia, only the fish with the highest heart rate showed a marked reduction in f_H (41%). Observations in C. aceratas are not clear-cut with respect to hypoxic bradycardia. In C. aceratas, Holeton (1972) reported an atropine-sensitive bradycardia associated with both hypoxia and hyperoxia, heart rate decreasing by 45–85%. In Pseudochaenichthys georgianas, Q̇ heart rate and ventral aortic blood pressure are reduced by up to 50% during hypoxia (Hemmingsen and Douglas, 1977). In contrast, Hemmingsen et al. (1972) and Hemmingsen and Douglas (1977) reported a slight decrease in heart rate and a modest increase in Q̇ in C. aceratas, while Hemmingsen and Douglas (1972) reported no change in heart rate during hypoxia.

Adrenaline, through an increase in vascular resistance, usually produces a pressor response that is often associated with a reflex bradycardia. This is presumably a barostatic reflex (Wood and Shelton, 1980a,b; Farrell, 1991). It is noteworthy that such a response was absent in P. bernacchii. Instead, a decrease in stroke volume reduced the pressor response. Since f_H increased immediately after the injection of adrenaline, it seems likely that there was a decrease rather than the expected increase in inhibitory cholinergic tone to the heart associated with the pressor response. This is further suggested by the lack of tachycardia in response to adrenaline after pretreatment with atropine. This was not the case in the two P. borchgrevinki examined. Heart rate was either unchanged or decreased under these conditions. The physiological basis for this interspecific difference is worth further study.

Farrell (1991) concludes that temperate fishes generally regulate cardiac stroke volume to a greater degree than heart rate, with tunas presented as an exception to this generality. Based on the present study, red-blooded nototheniids may be another exception. Because heart rate is under such a large inhibitory cholinergic tone, P. bernacchii could increase f_H about threefold. We observed twofold increases in f_H during swimming in P. borchgrevinki. Such changes are greater than those observed in temperate teleosts. Furthermore, four observations support the idea that red-blooded nototheniids alter /H to a greater degree than they alter stroke volume. In P. bernacchiii, the recovery from hypoxia was associated with tachycardia, with Vs only returning to, but not increasing beyond, resting values. In P. borchgrevinki, swimming was associated with a much larger increase in f_H than in Vs. In both P. bernacchii and P. borchgrevinki, Q̇ was well regulated when f_H. was elevated after atropine treatment. Lastly, in many of the manipulations, increases in/H were accompanied by decreases in Vs. A question still to be resolved is whether these fish can increase stroke volume much beyond the relatively small (25%) increases observed here.

It is unlikely that antarctic fish will be exposed to hypoxia in their well-oxygenated, cold and stenothermal environment. Nevertheless, they respond to experimentally induced hypoxia in a similar manner to temperate species. In P. borchgrevinki exposed to a of 8 kPa for 11–14 days there was a 93% increase in haematocrit and the P50 of the blood was decreased by 33%, a change associated with a 25% reduction in erythrocyte ATP concentrations (Wells et al. 1989). In the present study, cardiac output increased and total vascular resistance declined on recovery from hypoxia as was restored to 20–21 kPa, results that are consistent with an improved oxygen delivery to tissues that previously may have either run down their oxygen stores or switched to anaerobic metabolism. There was a rise in P_VA at a of 6.7 kPa in both control and atropinized fish. Gadus morhua also showed a transient rise in P_VA. in response to hypoxia, which was brought about by an increase in systemic resistance (Fritsche and Nilsson, 1989). A rise in P_VA might be expected to result in lamellar recruitment at the gill.

Since P. bernacchii is a benthic species and P. borchgrevinki is a cryopelagic species, cardiovascular differences were expected. P. bernacchii would not swim in the swimming tunnel, a response typical of benthic fish. Although most of the work was performed on P. bernacchii, sufficient observations were made on P. borchgrevinki to note significant species differences. In keeping with the more active life style of P. borchgrevinki, Q̇P_VA, myocardial power output, haemato-crit and the sizes of the atrium and ventricle were relatively greater than for P. bernacchii. The difference in PVA is not accounted for by the small difference in haematocrit (and thus blood viscosity). Instead, it is largely due to a lower total vascular resistance in P. borchgrevinki. Whether the branchial or systemic resistances are markedly different in the two species remains to be determined.

This work was support by grants from Natural Sciences and Engineering Research Council, Canada, by an Erskine Visiting Fellowship to A.P.F. and by University Grants Committee awards to W.D. and M.E.F. We thank Gavin Robinson for his help in setting up facilities and Professor Rufus Wells for drawing to our attention various papers in press.