Hypoxia tolerance in elasmobranchs. I. Critical oxygen tension as a measure of blood oxygen transport during hypoxia exposure

Environmental hypoxia is a common abiotic stressor affecting the survival and distribution of aquatic species, but its occurrence varies in magnitude and spatio-temporal scale depending on habitat (Diaz and Breitburg, 2009). Consequently, many species of fish have evolved the ability to survive periods of low O₂ exposure but the severity and duration of hypoxia that can be tolerated is highly species specific. Identification of the physiological responses contributing to hypoxia tolerance among fishes is an area of major research interest. The increasing occurrence worldwide of environmental hypoxia due to anthropogenic activities, in particular, has led to the desire for simple physiologic metrics of hypoxia tolerance in fishes and other aquatic organisms. Even so, relatively few comparative studies exist on this topic in fishes.

One simple metric that holds promise in this regard is the critical O₂ tension of whole-animal O₂ consumption rate (), or P_crit, which is the water P_O2 (Pw_O2) at which the of an organism transitions from oxyregulation to oxyconformation. P_crit is thought to reflect the ability of an organism to extract O₂ from the environment to maintain routine as Pw_O2 decreases, with a low P_crit being associated with greater hypoxia tolerance presumably because of improved O₂ uptake and transport to tissues at low Pw_O2. Consequently, P_crit has been employed routinely as an important measure of hypoxia tolerance in aquatic organisms including fishes (Chapman et al., 2002; Mandic et al., 2009; Nilsson and Östlund-Nilsson, 2008; Pörtner and Grieshaber, 1993; Routley et al., 2002). Physiological modifications at any step in the respiratory cascade may affect P_crit, including changes in ventilation, gill surface area, blood O₂ capacity (including blood haemoglobin concentration [Hb] and Hb–O₂ binding affinity), circulation of O₂ (e.g. cardiac output), diffusion into tissues and mitochondrial O₂ turnover (Dejours, 1981; Farrell and Richards, 2009). Whole-blood Hb–O₂P₅₀ (i.e. Hb–O₂ binding affinity), in particular, has received attention as a determinant of P_crit as well as hypoxia tolerance in fishes because of its important role in controlling blood O₂ content and O₂ uptake (Brauner and Wang, 1997). A phylogenetically independent comparison of O₂ transport in sculpins showed that P_crit is strongly correlated with Hb–O₂P₅₀; species possessing a low P_crit typically also have a low Hb–O₂P₅₀ (Mandic et al., 2009). Other fishes known to have a high Hb–O₂ binding affinity typically also have a low P_crit and are often hypoxia tolerant (Burggren, 1982; Jensen and Weber, 1982; Sollid et al., 2005). Overall, however, the relationship between P_crit and Hb–O₂P₅₀ has rarely been directly investigated and it remains incompletely defined. Also, it has not been unequivocally demonstrated that a low P_crit is associated with greater arterial blood O₂ transport during hypoxia exposure in fishes. In fact, little is known in fishes about the responses of blood gas transport at and around P_crit.

At Pw_O2 below P_crit, O₂-independent energy production (e.g. anaerobic glycolysis) is increasingly relied upon to meet energy demands at a time when aerobic metabolism is constrained. Measurements of P_crit and anaerobic end-product accumulation in certain animals provide evidence for the hypothesis that the increased activation of anaerobic metabolism coincides with P_crit (Pörtner and Grieshaber, 1993), although support from other studies is equivocal (McKenzie et al., 2000; Nonnotte et al., 1993). The link between P_crit and increased activation of anaerobic metabolism is thus uncertain, especially in fishes. Also, there are no direct comparisons of the onset of lactate accumulation in hypoxia-tolerant and -sensitive fishes that differ in P_crit.

In the present study, we investigated the relationship between P_crit and arterial O₂ transport properties, including in vivo Hb–O₂P₅₀ and arterial total O₂ content (Ca_O2), during progressive hypoxia exposure in two tropical elasmobranch species with similar lifestyles and activity levels, the epaulette shark, Hemiscyllium ocellatum (Bonnaterre), and the eastern shovelnose ray, Aptychotrema rostrata (Shaw). Epaulette sharks inhabit shallow coral reef environments where nocturnal hypoxia occurs commonly and they can tolerate hours of severe hypoxia (<1.0 kPa) exposure and up to 45 min of anoxia exposure (Renshaw et al., 2002; Routley et al., 2002). In contrast, eastern shovelnose rays are found in generally well-oxygenated coastal sandy and muddy benthic habitats in eastern Australia, including our sampling site of Moreton Bay, where hypoxic events are rare and dissolved O₂ is usually close to air saturation at all depths (Dennison et al., 2004; Gabric et al., 1998; Kyne and Bennett, 2002). Indeed, preliminary observations from the present study showed that shovelnose rays are relatively hypoxia sensitive, succumbing rapidly if held at or below a Pw_O2 of 2.0 kPa for more than approximately 30 min. We predicted that the more hypoxia-tolerant epaulette shark would have a lower P_crit than the shovelnose ray and that the lower P_crit in the epaulette shark would be associated with a lower Hb–O₂P₅₀ and a correspondingly greater Ca_O2 at similar hypoxic Pw_O2. Furthermore, we predicted that at each species' P_crit or at the same percentage of P_crit, Hb–O₂ saturation and Ca_O2 would be similar between the species despite being exposed to different Pw_O2 values. Finally, we measured arterial blood metabolic status including pH, [lactate] and CO₂ status in order to further characterize the physiological correlates of P_crit in fishes and test the hypothesis that P_crit is associated with increased activation of anaerobic metabolism. Overall, the present study provides a comprehensive picture of the physiological responses associated with P_crit in two fishes, elucidating for the first time the relationship between P_crit and arterial blood O₂ transport characteristics during hypoxia exposure, and providing comparative insight into the respiratory and metabolic attributes associated with hypoxia tolerance in fishes.

Epaulette sharks and shovelnose rays of mixed sexes were supplied by Cairns Marine (Cairns, QLD, Australia) or Seafish Aquarium Life (Dunwich, QLD, Australia), respectively. The animals were collected under A1 level commercial harvest licences granted by the Department of Primary Industries, Australia. Epaulette sharks were caught on the Great Barrier Reef and transported in flow-through seawater tanks to holding tanks at Cairns Marine, where they were kept for 2 days without feeding before being transported by air and automobile to Moreton Bay Research Station, North Stradbroke Island, QLD, Australia. Shovelnose rays were caught in Moreton Bay, transferred to the mainland in flow-through seawater tanks, and transported to Moreton Bay Research Station by automobile. All fish were held in a recirculating seawater system (28°C) for at least 5 days before experimentation. Animals were fed every other day and fasted for at least 24 h before experimentation. All experiments were conducted according to guidelines set out by the Canadian Council for Animal Care and protocols approved by the University of British Columbia Animal Care Committee and the Griffith University Animal Ethics Committee.

Epaulette sharks (1.29±0.04 kg, N=7) and shovelnose rays (1.54±0.06 kg, N=8) were netted from the holding tanks and anaesthetized in water containing a final concentration of 0.1 g l^–1 benzocaine (initially dissolved in 95% ethanol; 0.001% ethanol in anaesthetic bath). Fish were then moved to a surgery table where the gills were continuously irrigated with aerated seawater (28°C) containing 0.075 g l^–1 benzocaine.

To permit periodic sampling of blood, a PE50 (Clay-Adams, Parsippany, NJ, USA) cannula was fitted in the caudal artery via a lateral incision in the caudal peduncle, as described previously (De Boeck et al., 2001). The cannula was filled with heparinized (50 U ml^–1) elasmobranch saline (in mmol l^–1: 257 NaCl, 7 Na₂SO₄, 6 NaHCO₃, 0.1 Na₂HPO₄, 4 KCl, 3 MgSO₄·H₂O, 2 CaCl₂·2H₂O, 300 urea and 100 trimethylamine oxide). The cannula was exteriorized through a PE160 grommet and sutured to the skin. The incision was closed with silk sutures. The fish were also fitted with ventral aorta flow probes as described in the accompanying paper (Speers-Roesch et al., 2012).

Following surgery, the instrumented fish was immediately moved to a cylindrical acrylic respirometer (18.2 l for epaulette sharks and 28.4 l for shovelnose rays) that was submersed inside an opaque aquarium that received seawater from the same recirculating system used for the fish holding tanks (28°C). A submersible pump inside the aquarium provided a continuous flow of water to the respirometer that ensured complete water mixing inside the respirometer and maintained Pw_O2. The respirometer was covered with black plastic to prevent visual disturbance of the fish. The cannula and the flow probe lead from the fish were exteriorized through a hole in the respirometer fitted with a soft rubber stopper modified with a slit. The fish was allowed to recover from surgery and habituate to the respirometer for at least 12 h before any experimental procedures were performed.

Stable baseline conditions were confirmed by monitoring routine cardiovascular variables (see Speers-Roesch et al., 2012) for 1–2 h at a normoxic Pw_O2 of between 15 and 16 kPa (74–78% air saturation; 100% air saturation=20.4 kPa=153 Torr). An aortic blood sample (1 ml) was taken at the end of this period for immediate measurement of normoxic resting levels of arterial whole-blood pH, P_O2, [Hb], haematocrit (Hct) and total O₂ content as described below. Plasma was separated by centrifugation (5000 g, 5 min), frozen in liquid nitrogen within 5 min of sampling, transported to Canada in a dry shipper and kept frozen at –80°C for several weeks until analyses of metabolites and [total CO₂] (see below). The red blood cell pellet was re-suspended in elasmobranch saline to a final volume of 1 ml and this was injected via the cannula to replace the blood removed. The respirometer was then closed by connecting the inflow and outflow tubes of the respirometer via a submersible pump that re-circulated the water inside the respirometer. The fish was allowed to consume O₂ in the respirometer and the rate of depletion of Pw_O2 was used to calculate as described below. The changes in water parameters (e.g. pH, P_CO2) potentially associated with the use of closed respirometry have been shown previously to have no effect on P_crit in fish (Henriksson et al., 2008). Indeed, only modest increases in arterial P_CO2 (Pa_CO2) were observed in the present study (see Results and Discussion). Arterial blood samples were taken and treated as described above at regular intervals including Pw_O2 at approximately 11.8, 7.7, 5.8, 3.8 and 2.0 kPa in both species, as well as approximately 1.0 and 0.1 kPa in epaulette sharks (see Table 1). Fish were held in the closed respirometer until Pw_O2 reached ∼0.1 kPa for epaulette sharks and ∼1.6 kPa for the shovelnose rays, which took 135±8 and 71±6 min, respectively, from the point at which the normoxic blood sample was taken and the respirometer was closed. These nadirs of Pw_O2 were chosen based on previous studies on epaulette sharks (e.g. Renshaw et al., 2002) and preliminary observations of hypoxia-exposed animals, which clearly revealed that epaulette sharks tolerated prolonged bouts of severe hypoxia whereas shovelnose rays showed distress or loss of equilibrium if held at or below 2.0 kPa for more than 30 min. The rate of O₂ depletion was similar between species (see Table 1). Once the nadir in Pw_O2 (∼0.1 or ∼1.6 kPa) was reached, normoxic water was reintroduced to the respirometer. A final blood sample was taken at 60 min of recovery in normoxic water and analysed as previously described. In some cases, cannulae were damaged by the movements of the fish during the overnight acclimation period or during the experimental exposure and therefore sample sizes of measured parameters vary slightly (see figure captions for final N values). At the end of the trials the fishes were terminally anaesthetized in seawater containing benzocaine.

Pw_O2 in the respirometer was measured using an Oxyguard probe (Mark IV, Point Four Systems, Richmond, BC, Canada), modified to give a ±1 V output signal, that was placed in a custom-made Plexiglas® chamber connected in line with the circulation pump. The probe output was fed to a Power Lab unit (ADInstruments, Castle Hill, NSW, Australia) and subsequently analysed using LabChart Pro software (v. 6.0; ADInstruments).

Whole-animal during the respirometry trials was calculated from the rate of decline in Pw_O2 over 10 min periods bracketing Pw_O2 at regular intervals from approximately 14.8 to 0.67 kPa in epaulette sharks and approximately 13.7 to 1.9 kPa in shovelnose rays, following the methods of Henriksson et al. (Henriksson et al., 2008). Blanks without fish were run for both chambers and background was subtracted from fish . The corrected was plotted against Pw_O2 and the inflection point at which transitions from being independent to being dependent on Pw_O2 (i.e. P_crit) was calculated using the BASIC program designed by Yeager and Ultsch (Yeager and Ultsch, 1989). At Pw_O2 above P_crit, values were constant and not significantly different from one another within a species, therefore confirming that is independent of Pw_O2 above P_crit. The was not measured at the initial normoxic Pw_O2 (∼15.3–16.0 kPa) or the normoxic recovery period because of the flow-through conditions.

Blood [Hb] was measured spectrophotometrically (Blaxhall and Daisley, 1973). Hct was determined following centrifugation at 5000 g in a sealed capillary tube. Mean cellular Hb content (MCHC) was calculated as ([Hb]/Hct)100. Arterial blood P_O2 (Pa_O2) was measured with a Radiometer P_O2 (E-5046) electrode, thermostatically controlled in a D616 cell at 28°C, in conjunction with a PHM 71 acid–base analyser (Radiometer, Copenhagen, Denmark). Whole-blood Ca_O2 was measured following the method of Tucker (Tucker, 1967). In order to calculate the percentage O₂ saturation of Hb in arterial blood, the quantity of O₂ bound to Hb was calculated by subtracting physically dissolved O₂ in the blood [calculated from measured Pa_O2 and published blood O₂ solubility coefficients (Christiforides and Hedley-Whyte, 1969)] from the Ca_O2 measured in whole blood. The quantity of O₂ bound to Hb was then expressed as a percentage of the theoretical maximum quantity of O₂ bound to Hb, which was calculated using measured [Hb] values and assuming four O₂ bound per Hb tetramer at 100% saturation. In vivo blood P₅₀ and Hill coefficient values were then calculated from Hill plots based on data for arterial P_O2 and Hb–O₂ saturation.

Blood pH was measured using a Radiometer pH micro-electrode thermostatically held at 28°C and displayed on a Radiometer PHM 71 acid–base analyser. True plasma [total CO₂] was measured according to Cameron (Cameron, 1971), immediately following thawing of the plasma on ice. To confirm that freezing of plasma and storage at –80°C has no effect on [total CO₂], we compared [total CO₂] measured in freshly sampled plasma from normocarbic, normoxic tilapia (Oreochromis hybrid sp.) with an aliquot of the same plasma that was frozen in liquid nitrogen and stored at –80°C for 4 weeks. [Total CO₂] was 7.1±0.7 mmol l^–1 in the fresh samples vs 7.0±0.8 mmol l^–1 in the frozen samples (N=6; paired t-test, P>0.05; B.S.-R., unpublished); thus, freeze–thaw of plasma does not appear to affect [total CO₂]. We have seen similar results in hagfish (Eptatretus stoutii) (D. W. Baker and C.J.B., unpublished). Pa_CO2 and [HCO₃^–] were calculated from measured values of pH and [total CO₂] through manipulation of the Henderson–Hasselbalch equation (Brauner et al., 2000) with appropriate constants for elasmobranchs (Boutilier et al., 1984). Plasma [lactate] and [glucose] were measured on deproteinized and untreated plasma, respectively, according to the protocols outlined by Bergmeyer (Bergmeyer, 1983). Plasma [β-hydroxybutyrate] ([β-HB]) was measured on deproteinized plasma following the protocol of McMurray et al. (McMurray et al., 1984).

The effects of species and Pw_O2 on blood gas, acid–base and metabolite parameters were tested for samples 1–6 (at overlapping Pw_O2 between species; see Table 1) using a two-way ANOVA followed by Holm–Sidak post hoc (H–S) tests against species or the normoxic resting values measured at ≥15.3 kPa (Pw_O2 values were not statistically different between species at each sample point, allowing this two-way design; Student's t-test, P>0.05). Samples 7 and 8 in epaulette sharks (see Table 1) were tested separately for significance against the normoxic resting sample 1 (16.0 kPa) using a one-way ANOVA with H–S tests. Species differences in the in vivo blood P₅₀ and Hill coefficient (n_H) values were examined using a Student's t-test and the effect of O₂ on Hb–O₂ saturation was compared within species using a one-way ANOVA with H–S tests (comparisons between species were not made because arterial P_O2 at sample points were not always comparable). The relationship between Hb–O₂ saturation and P_crit was examined via linear regression analysis of Hb–O₂ saturation against Pw_O2 expressed as the percentage of P_crit (using values from individual animals). This analysis was carried out up to a Hb–O₂ saturation of ∼75% because at higher values Hb saturation curves are asymptotic. Recovery values were compared between species and with the normoxic resting values at ≥15.3 kPa within species using a two-way ANOVA with H–S test.

The critical Pw_O2 of were compared between species using a Student's t-test. The effects of species and Pw_O2 on were tested using a two-way ANOVA with H–S tests using data from eight sampling points of overlapping Pw_O2 at approximately 13.6, 12.3, 10.3, 6.1, 4.2, 3.4, 2.5 and 1.9 kPa and the points of statistical comparison are denoted by horizontal braces on the figures. Overlapping Pw_O2 values were not statistically different between species (Student's t-test, P>0.05). Data from other sampling points were omitted from these analyses. However, in order to fully assess the effect of Pw_O2 on in each species, one-way ANOVA were run across all sampling points within each species, with H–S comparisons against the first normoxic resting value. The effect of Pw_O2 on measured parameters was found to be similar for both two-way and one-way ANOVA designs.

Statistical significance was accepted when P<0.05 and analyses were carried out using SigmaStat 3.0 or GraphPad Prism 5.0. Data were log or square-root transformed prior to statistical analyses if assumptions of equal variance or normality were not met. Repeated measures ANOVA could not be carried out because experimental constraints negated the use of data from the same animal at every single sample period. In any case, the standard ANOVA procedures utilized here result in a conservative statistical assessment of our data.

Under normoxic conditions, individuals of both species were generally quiescent. As Pw_O2 decreased, there was a modest and temporary increase in activity level associated with exploratory behaviour in some individuals of each species. As Pw_O2 decreased further, the fishes again became quiescent. In a couple instances in each species, severe agitation occurred temporarily (<10 s) at low Pw_O2, causing the cannula to be dislodged and negating further blood sampling.

A typical relationship between and Pw_O2 was observed in both species, each with a zone of O₂-independent occurring at higher Pw_O2 followed by a zone of O₂ dependence below P_crit, where decreased with decreasing Pw_O2 (Fig. 1). The P_crit was significantly lower in epaulette sharks than in shovelnose rays (Fig. 1). When Pw_O2 was above ∼5 kPa, was similar between species. However, when Pw_O2 was below ∼5 kPa, was consistently greater for epaulette sharks than for shovelnose rays at any similar Pw_O2 (Fig. 1).

Pa_O2 decreased linearly with decreasing Pw_O2 and was not different between species, except at the two highest Pw_O2 where epaulette sharks had higher Pa_O2 compared with shovelnose rays (Fig. 2A). Total Ca_O2 decreased with decreasing Pw_O2 with the exception of between 16.01±0.43 and 11.80±0.06 kPa in epaulette sharks. Total Ca_O2 was the same in both species at the highest Pw_O2, but at all lower Pw_O2 it was greater in epaulette sharks than in shovelnose rays (Fig. 2B). Recovery in normoxic water returned Pa_O2 and Ca_O2 to values similar to those measured at the beginning of the progressive hypoxia trial (Fig. 2). Hb–O₂ saturation decreased as Pa_O2 fell below approximately 9.0 and 6.0 kPa in shovelnose rays and epaulette sharks, respectively (Fig. 3). The in vivo Hb–O₂ binding affinity was significantly higher (lower P₅₀) in epaulette sharks than in shovelnose rays (Fig. 3). The Hill coefficient was similar between species (Fig. 3). At each species' P₅₀, arterial pH was ∼7.82 in epaulette sharks and ∼7.81 in shovelnose rays, and Pa_CO2 was ∼0.23 kPa in both species (extrapolated from pH and Pa_CO2 data at the extrapolated Pw_O2 at P₅₀).

A significant linear relationship was found for each species when Hb–O₂ saturation was regressed against Pw_O2 expressed as a percentage of P_crit, and because the slopes and y-intercepts of these regression lines were not significantly different between species, one line of best fit was applied to the pooled data for the two species (Fig. 4). Above ∼175% of P_crit, the Hb–O₂ saturation became asymptotic so these data were excluded from the linear regression. Thus, Hb–O₂ saturation was similar between species at Pw_O2 values representing the same percentage of each species' P_crit (up to ∼175% of P_crit). Because there were no major differences in [Hb] between species (see below), the same relationship existed for Ca_O2 (data not shown). Overall, these data show that P_crit is predictive of Hb–O₂ saturation and Ca_O2 at Pw_O2 lower than about 175% of P_crit.

In both species progressive hypoxia had no effect on Hct, [Hb] or MCHC (Table 1). Hct and [Hb] were significantly higher in epaulette sharks at sampling points 4 (Pw_O2=5.73±0.02 kPa in epaulette sharks and 5.83±0.09 kPa in shovelnose rays) and 6 (Pw_O2=1.85±0.03 kPa in epaulette sharks and 2.07±0.03 kPa in shovelnose rays) and [Hb] was also higher at sampling point 5 (Pw_O2=3.81±0.12 kPa in epaulette sharks and 3.82±0.07 kPa in shovelnose rays) (Table 1). Recovery [Hb] and MCHC were similar to normoxic resting values but Hct was higher in epaulette sharks during recovery (Fig. 2; Table 1).

Plasma [lactate] was low at Pw_O2 above 8 kPa, then significantly increased above normoxic resting levels by ∼3.7 kPa in both species and increased further with decreasing Pw_O2. Plasma [lactate] remained elevated after 60 min recovery in normoxic water (Fig. 5A). Plasma [lactate] was generally similar between species at all Pw_O2 during progressive hypoxia but was greater in epaulette sharks during recovery (Fig. 5A). A significant decrease in blood pH first occurred at ∼3.8 kPa in both species, decreasing further with declining Pw_O2 and remaining low after 60 min recovery in normoxic water (Fig. 5B). Blood pH was similar between species but was significantly lower in the shovelnose rays at the final sampling point for this species at ∼2.0 kPa (Fig. 5B). Arterial [HCO ^–₃] and [total CO₂] were unchanged during progressive hypoxia exposure in both species (Table 1). An increase in Pa_CO2 occurred in both species and it remained after 60 min recovery in normoxic water (Table 1). In epaulette sharks only, normoxic recovery was associated with significantly lower [HCO ^–₃] and [total CO₂] compared with normoxic resting levels as well as compared with the same parameters in shovelnose rays in recovery (Table 1). Plasma [glucose] or [β-HB] were similar between species and were unaffected by either progressive hypoxia or recovery in normoxic water (Table 1).

The present study demonstrates a link between P_crit and arterial blood O₂ transport characteristics during hypoxia exposure in two elasmobranch species, adding significantly to a growing body of evidence showing that P_crit is an important indicator of hypoxia tolerance in fish (Chapman et al., 2002; Mandic et al., 2009; Nilsson and Östlund-Nilsson, 2008). We provide the first comparative evidence that a lower P_crit is associated with maintenance of greater Ca_O2 during hypoxia exposure, which benefits hypoxia tolerance. Similarly, we show that P_crit is predictive of Hb–O₂ saturation and Ca_O2 (in the observed absence of changes in [Hb]) during hypoxia exposure (Fig. 4), supporting the notion that differences in Hb–O₂ binding affinity determine differences in P_crit. Indeed, the in vivo Hb–O₂P₅₀ of the epaulette shark was lower than that of the shovelnose ray and this is the likely explanation for the greater Ca_O2 observed in the former species at low Pw_O2. The impressive hypoxia tolerance of the epaulette shark is probably attributable, in part, to its enhanced O₂ transport characteristics compared with those of the less tolerant shovelnose ray.

Under normoxic resting conditions (i.e. Pw_O2>P_crit) the of epaulette sharks matches closely with that measured for this species by Routley et al. (Routley et al., 2002). Epaulette sharks and shovelnose rays had similar resting , which fell within the range of temperature-corrected for elasmobranchs of a similar activity level (0.9–3.5 μmol g^–1 h^–1) (Butler and Metcalfe, 1988). In both species, exposure to progressive hypoxia caused a pronounced reduction of below P_crit (Fig. 1), similar to many teleosts (Mandic et al., 2009; Speers-Roesch et al., 2010) as well as at least one other elasmobranch, the spotted catshark (Scyliorhinus canicula) (Butler and Taylor, 1975). The P_crit for epaulette sharks (Fig. 1) matches the lower range of P_crit measured for this species by Routley et al. (Routley et al., 2002). The shovelnose ray had a significantly higher P_crit compared with the epaulette shark (Fig. 1), but it was lower than that of other previously studied elasmobranchs (see Routley et al., 2002). Although the level of O₂ demand can affect P_crit (Thuy et al., 2010), this probably does not explain the observed difference in P_crit because under normoxic conditions was the same in each species (Fig. 1). The difference in P_crit between the hypoxia-tolerant epaulette shark and the comparatively hypoxia-sensitive shovelnose ray is consistent with the notion that a lower P_crit is associated with greater hypoxia tolerance in fish (Mandic et al., 2009; Nilsson and Östlund-Nilsson, 2008). The P_crit of the epaulette shark is generally similar to those of hypoxia-tolerant teleosts and it is the lowest known P_crit among elasmobranchs (Routley et al., 2002; Speers-Roesch et al., 2010). The lower P_crit of epaulette sharks was associated not only with maintenance of routine to a lower Pw_O2 compared with shovelnose rays but also with maintenance of greater at all comparable hypoxic Pw_O2 (i.e. below P_crit), where depression of occurred in both species (Fig. 1). Thus, a low P_crit allows a fish to maintain as high as possible during hypoxia exposure, minimizing reliance on inefficient anaerobic metabolism.

Haematological (i.e. Hct, [Hb], MCHC) and O₂ transport properties (i.e. Pa_O2, Ca_O2 and Hb–O₂ saturation) in arterial blood of epaulette sharks and shovelnose rays under normoxic resting conditions (Figs 2 and 3, values at highest Pw_O2 or Pa_O2; Table 1, sample 1) were generally similar between species and the values were typical for elasmobranchs (e.g. Hct, 10–20%; [Hb], 0.46–0.62 mmol l^–1; MCHC, ∼3.3; Pa_O2, 8–15 kPa; Ca_O2, 3–5 vol. %; Hb–O₂ saturation, 75–100%) (Butler and Metcalfe, 1988; Butler and Taylor, 1975; De Boeck et al., 2001; Lai et al., 1990; Perry and Gilmour, 1996; Routley et al., 2002). The lack of 100% Hb–O₂ saturation in normoxia is consistent with other studies on elasmobranchs (e.g. Lai et al., 1990) and the presence of up to 27% methaemoglobin in fishes has been put forward as an explanation as to why some fish haemoglobins are not saturated to their theoretical maximum in vivo (Graham and Fletcher, 1986). The in vivo P₅₀ values for epaulette sharks and shovelnose rays are higher than previously measured in other elasmobranchs (1.9–2.7 kPa) (Butler and Taylor, 1975; Cooper and Morris, 2004; Lai et al., 1990), but in the previous studies these were measured at lower environmental temperatures compared with the present study and environmental temperature appears to be positively correlated with P₅₀ in marine fishes (Wells, 2005). The Hill coefficients were similar between species and relatively low, which is typical of elasmobranchs (Butler and Metcalfe, 1988).

The changes in Pa_O2 and Ca_O2 during progressive hypoxia (Fig. 2) were similar to those seen in other elasmobranchs (Butler and Taylor, 1975; Perry and Gilmour, 1996; Routley et al., 2002). The roughly linear decrease in Pa_O2 was of similar magnitude in epaulette sharks and shovelnose rays, suggesting ventilatory responses to hypoxia are comparable between species (Fig. 2A). However, at all matched Pw_O2 below the normoxic resting level, Ca_O2 was greater in epaulette sharks than in shovelnose rays, which appears to be due to greater Hb–O₂ saturation resulting from the epaulette shark's higher in vivo Hb–O₂ binding affinity (i.e. lower P₅₀) (Fig. 2B; Fig. 3). This result is consistent with findings suggesting that Hb–O₂P₅₀ is an important component of hypoxia tolerance in fishes, with the most tolerant species possessing the lowest P₅₀, which result in greater blood O₂ loading at low Pw_O2 (Jensen and Weber, 1982; Mandic et al., 2009; Perry and Reid, 1992).

Interestingly, the species differences in Hb–O₂ saturation and Ca_O2 disappeared when these parameters were plotted against Pw_O2 expressed as a percentage of P_crit (i.e. the relationships overlapped between species) (Fig. 4). Thus, when measured at a Pw_O2 of the same percentage of each species' P_crit, values of Hb–O₂ saturation or Ca_O2 (because changes in [Hb] were not apparent) were the same in epaulette sharks and shovelnose rays. These results suggest that P_crit is predictive of Hb–O₂ saturation and Ca_O2 during hypoxia exposure, with species with lower P_crit having a greater capacity for arterial blood O₂ transport at similar hypoxic Pw_O2. These results also suggest that the differences in arterial Hb–O₂ saturation between epaulette sharks and shovelnose rays may represent the basis of species differences in P_crit. This observation, as well as the fact that at P_crit the Pa_O2 in both species was similar to their respective in vivo Hb–O₂P₅₀ (Fig. 2A; Fig. 3), agrees with Mandic and colleagues' (Mandic et al., 2009) discovery of a close relationship between P_crit and Hb–O₂P₅₀ in sculpins and supports the idea that P₅₀ is an important determinant of P_crit in fishes.

The lack of a change in Hct or [Hb] during progressive hypoxia (Table 1) is consistent with previous studies on epaulette sharks and other elasmobranchs (Perry and Gilmour, 1996; Routley et al., 2002; Short et al., 1979) and thus adjustments of these parameters may have a minimal role in improving O₂ transport in elasmobranchs exposed to hypoxia. Nonetheless, Hct and [Hb] were approximately 25% higher at some hypoxic Pw_O2 values in epaulette sharks compared with shovelnose rays (Table 1), probably contributing to the higher Ca_O2 seen in the former species. However, the difference of Ca_O2 between species was greater (40–100% higher in epaulette sharks) at these hypoxic Pw_O2, affirming an important role for Hb–O₂P₅₀ in enhancing O₂ supply in the epaulette shark.

An increased reliance on anaerobic glycolysis and a consequent metabolic acidosis during hypoxia exposure was indicated by a large increase in plasma [lactate] and a decrease in blood pH at or below P_crit in both species (Fig. 5), similar to results observed in other hypoxia-exposed fishes (Routley et al., 2002; Scott et al., 2008). At the lowest Pw_O2 exposure of shovelnose rays, blood pH was significantly lower compared with epaulette sharks (Fig. 5B), suggesting that the latter species may possess more effective acid–base regulation or less reliance on anaerobic energy production, potentially due to a superior O₂ supply. Nevertheless, at lower Pw_O2 epaulette shark blood pH dropped precipitously and like other hypoxia-tolerant vertebrates this species must be able to tolerate severe metabolic acidosis during hypoxia exposure (Driedzic and Gesser, 1994). The Pw_O2 at which epaulette sharks first showed a significant increase in plasma [lactate] was approximately the same as found previously for this species (Routley et al., 2002) and coincided with its P_crit (Fig. 5A). In shovelnose rays, however, the increase in plasma [lactate] did not coincide with its P_crit but rather occurred at a similar Pw_O2 to that of the epaulette sharks and accumulated at a similar rate (Fig. 5A). Although our data do not provide solid evidence for the hypothesis that P_crit correlates with increased activation of anaerobic metabolism in fishes (Pörtner and Grieshaber, 1993), they are consistent with this activation not occurring above P_crit. It is also possible that a non-equilibrium state during early progressive hypoxia exposure including interspecific differences in blood volume or lactate handling may have obscured our ability to detect the onset of an increase of anaerobic metabolism at P_crit in shovelnose rays. This may also explain why the rate of plasma lactate accumulation was similar between species despite lower blood O₂ content in shovelnose rays. We consider the effect of changes in fish activity (see Results) on plasma [lactate] to be negligible because of the low measurement variability and the consistently low values at Pw_O2 above P_crit (Fig. 1). Also, most fish activity was moderate and in the few instances of excessive activity, cannula dislodgement meant that plasma [lactate] was no longer measured.

The metabolic acidosis seen in epaulette sharks and shovelnose rays during progressive hypoxia resulted in a downward trend in [HCO ^–₃] and [total CO₂] (Table 1). In both species Pa_CO2 increased, possibly due to a build-up of CO₂ in the closed respirometer. This elevation in CO₂ probably had no major effect on other measured parameters, because the magnitude of the Pa_CO2 increase was not large until the final sample point and no major differences were observed between the final sample point and the previous sample point for any measured parameters. In any case, hypercarbia commonly accompanies environmental hypoxia so in this regard the present exposures are more ecologically relevant. The responses of blood CO₂ parameters to progressive hypoxia in the present study differ from the respiratory alkaloses seen in hypoxic spiny dogfish (Squalus acanthias) (Perry and Gilmour, 1996) or spotted catshark (Butler et al., 1979), but these other studies utilized moderate rather than severe hypoxia that probably mitigated metabolic acidosis – Butler et al. (Butler et al., 1979) found no increase in plasma [lactate]. The relatively low normoxic resting levels of Pa_CO2, [HCO ^–₃] and [total CO₂] in epaulette sharks and shovelnose rays are typical of elasmobranchs (Butler and Metcalfe, 1988; De Boeck et al., 2001; Lai et al., 1990).

Plasma [glucose] was unaffected by progressive hypoxia (Table 1), which is unlike most teleosts but mirrors previous findings in hypoxia-exposed epaulette sharks and spiny dogfish (Routley et al., 2002; Speers-Roesch and Treberg, 2010). In some hypoxia-tolerant teleosts, large decreases in plasma [non-esterified fatty acid], an important aerobic fuel, occur during hypoxia exposure (Speers-Roesch et al., 2010). There was no similar response in epaulette sharks or shovelnose rays for plasma concentration of β-HB, which in elasmobranchs serves an important role as an alternative lipid-derived aerobic fuel (Speers-Roesch and Treberg, 2010).

Epaulette sharks and shovelnose rays showed comparable recovery of respiratory parameters after progressive hypoxia and followed a trajectory similar to that of other fishes (Hughes and Johnston, 1978; Van Raaij et al., 1996). Following 60 min of recovery in normoxic water, normoxic arterial O₂ parameters were restored in both species (Fig. 2B; Fig. 3); however, recovery from metabolic acidosis and high [lactate] was incomplete, particularly in epaulette sharks, which had experienced more severe hypoxia and hypoxia-induced acidosis and lactate accumulation (Fig. 5B). Arterial [HCO ^–₃] in recovery remained lower than normoxic levels in epaulette sharks, probably as a consequence of the persistence of acidosis, whereas shovelnose rays had recovered (Table 1).

The present study on the hypoxia-tolerant epaulette shark and the hypoxia-sensitive shovelnose ray provides the first evidence that P_crit is predictive of arterial Hb–O₂ saturation and Ca_O2 during hypoxia exposure in fishes. At the same level of hypoxia, fishes with a low P_crit can maintain higher Ca_O2 than fishes with a higher P_crit and this appears to be due to the presence of a lower in vivo Hb–O₂P₅₀ that allows greater Hb–O₂ saturation (Fig. 2B; Fig. 3). Additionally, at Pw_O2 of the same percentage of P_crit, Hb–O₂ saturation (and Ca_O2) is the same between species (Fig. 4). These results suggest that the interspecific differences in Hb–O₂ saturation may represent the basis of species differences in P_crit, providing strong support for the notion that Hb–O₂P₅₀ is a major determinant of P_crit as well as hypoxia tolerance in fishes (Mandic et al., 2009).

Our finding of a link between P_crit, Hb–O₂ saturation and Ca_O2 provides a mechanistic explanation for the argument that P_crit is a good indicator of hypoxia tolerance in fishes. A low P_crit is associated with greater Ca_O2 and therefore improved O₂ delivery to tissues during hypoxia exposure, which presumably enhances energy supply and reduces the accumulation of deleterious anaerobic end products, thus improving hypoxia tolerance. Importantly, our findings also show that hypoxia exposures that are standardized to P_crit will yield comparable levels of arterial hypoxaemia, facilitating cross-species comparative analyses. However, the results of the accompanying study (Speers-Roesch et al., 2012) warn that even when hypoxia exposures are scaled to P_crit, tissue-specific differences may occur in the metabolic response to the same amount of circulating O₂.

Although our measurements of plasma [lactate] showed no difference in the onset Pw_O2 or the rate of accumulation of lactate between species despite differing P_crit, our ability to detect such a difference may have been obscured by non-steady-state lactate dynamics during initial hypoxia exposure. Accumulation of lactate in each species did occur at or below P_crit, providing equivocal support for the hypothesis that P_crit is associated with increased activation of anaerobic metabolism (Pörtner and Grieshaber, 1993). Investigations of lactate turnover in hypoxia-exposed fish, including consideration of the potential effect of metabolic depression (which could blunt lactate accumulation), are needed to further test this hypothesis.

The superior O₂ uptake and blood O₂ capacity during hypoxia exposure in epaulette sharks probably explains in part the renowned hypoxia tolerance of this elasmobranch (Nilsson and Renshaw, 2004). Epaulette sharks also show enhanced hypoxic cardiovascular function compared with shovelnose rays (Speers-Roesch et al., 2012), which further improves O₂ delivery to tissues. Maintenance rather than depression of O₂ supply and aerobic metabolism at low levels of Pw_O2 may be an important component of hypoxia tolerance in fishes.

 
• β-HB

  β-hydroxybutyrate

 
• Ca_O2

  arterial blood O₂ content

 
• Hb

  haemoglobin

 
• Hb–O₂P₅₀

  haemoglobin–O₂ binding affinity

 
• Hct

  haematocrit

 
• MCHC

  mean cellular haemoglobin content

 
• 

  whole-animal O₂ consumption rate

 
• n_H

  Hill coefficient

 
• Pa_CO2

  arterial blood P_CO2

 
• Pa_O2

  arterial blood P_O2

 
• P_CO2

  partial pressure of CO₂

 
• P_crit

  critical O₂ tension of whole-animal O₂ consumption rate

 
• P_O2

  partial pressure of O₂

 
• Pw_O2

  water P_O2

Kevin and Kathy Townsend and their staff at the Moreton Bay Research Station provided exemplary logistical and technical assistance. Craig Franklin kindly lent us equipment needed for a portion of this study. Miles Gray assisted with hypoxia exposures and sampling.