Maximum cardiac performance of Antarctic fishes that lack haemoglobin and myoglobin: exploring the effect of warming on nature’s natural knockouts

Abstract Antarctic notothenioids, some of which lack myoglobin (Mb) and/or haemoglobin (Hb), are considered extremely stenothermal, which raises conservation concerns since Polar regions are warming at unprecedented rates. Without reliable estimates of maximum cardiac output (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$\dot{Q}$\end{document}), it is impossible to assess their physiological scope in response to warming seas. Therefore, we compared cardiac performance of two icefish species, Chionodraco rastrospinosus (Hb−Mb+) and Chaenocephalus aceratus (Hb−Mb−), with a related notothenioid, Notothenia coriiceps (Hb+Mb+) using an in situ perfused heart preparation. The maximum \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$\dot{Q}$\end{document}, heart rate (fH), maximum cardiac work (WC) and relative ventricular mass of N. coriiceps at 1°C were comparable to temperate-water teleosts, and acute warming to 4°C increased fH and WC, as expected. In contrast, icefish hearts accommodated a higher maximum stroke volume (VS) and maximum \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$\dot{Q}$\end{document} at 1°C, but their unusually large hearts had a lower fH and maximum afterload tolerance than N. coriiceps at 1°C. Furthermore, maximum VS, maximum \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$\dot{Q}$\end{document} and fH were all significantly higher for the Hb−Mb+ condition compared with the Hb−Mb− condition, a potential selective advantage when coping with environmental warming. Like N. coriiceps, both icefish species increased fH at 4°C. Acutely warming C. aceratus increased maximum \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$\dot{Q}$\end{document}, while C. rastrospinosus (like N. coriiceps) held at 4°C for 1 week maintained maximum \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$\dot{Q}$\end{document} when tested at 4°C. These experiments involving short-term warming should be followed up with long-term acclimation studies, since the maximum cardiac performance of these three Antarctic species studied seem to be tolerant of temperatures in excess of predictions associated with global warming.


Introduction
A remarkable trait among vertebrates is the evolutionary loss of haemoglobin (Hb) and myoglobin (Mb) expression within the Channichthyid family of Antarctic notothenioid fishes (Holeton 1970). Mutations leading to the Hb − state occurred once during evolution of icefishes, ∼8.5 MYA (Near, 2004), subsequent to cooling of the Southern Ocean. Loss of Mb expression occurred four times during radiation of the Channichthyids , resulting in 6 of the 16 species of icefishes lacking Mb in their heart . Thus, the Hb − Mb + and Hb − Mb − states are unique to icefishes (Sidell and O'Brien, 2006), although loss of Mb expression is a trait shared by a few temperate teleosts (Grove & Sidell, 2002;Macqueen et al., 2014) whose hearts can function adequately provided blood PO 2 is sufficient (Bailey et al., 1990).
The Hb − and Mb − condition is thought to have restricted Channichthyid distribution primarily to the thermally stable offshore waters of Antarctica, which hover near −1.8 • C below the surface year-round. While cold water may be key to survival of stenothermal species by ensuring high levels of dissolved oxygen, and presumably low oxygen requirements, the Western Antarctic Peninsula region is warming at an unprecedented rate. Thus, endemic species possessing modest thermal plasticity (e.g. Bilyk et al., 2018) may be especially vulnerable to short-term extreme temperature fluctuations, as well as long-term climate warming (Clarke et al., 2007;Beers et al., 2010). Given the limited options for altered geographic distribution, documenting scope for performance provides critical insight into capacity to accommodate shifts in environmental temperature, and hence the potential for longer-term adaptations necessary to allow stock conservation within the current ecological niche.
In terms of physiological performance, cardiac performance is viewed by some as playing a critical role in setting the upper thermal limits of temperate-zone fishes (Wang and Overgaard, 2007;Portner and Farrell, 2008;Farrell, 2009Farrell, , 2016Eliason et al., 2011), though others have challenged this contention (Ern et al., 2016). Indeed, extensive adaptational remodelling of the icefish cardiovascular system is apparent for life in the frigid, but stable Southern Ocean. For example, the 90% decrease in oxygen-carrying capacity associated with the Hb-free blood of icefishes (Ruud, 1954) has been compensated for by a thin-walled ventricle with relative mass 10times larger than other benthic teleosts (Tota et al., 1991;Farrell and Jones, 1992;Axelsson, 2005;Farrell and Smith, 2017) that can eject a very large stroke volume (V S ) and produce an exceptionally high cardiac output (Q) for an animal living at ∼ 0 • C (Hemmingsen and Douglas, 1977). The concern, however, is that a very large routineQ needed to offset a low blood oxygen-carrying capacity might leave little cardiac scope for activity or facing environmental challenges such as warming.
Earlier studies with icefishes have reported on the maximumQ (an integrative measure limiting fish activity), but its response to warming is poorly documented. For example, maximumQ in Chionodraco hamatus (Hb − Mb + ) is impressive at 3 • C: ∼ 300 ml min −1 kg −1 (Tota et al., 1991;Acierno et al., 1997). This maximum performance index is best measured with a working perfused heart given the difficulty of inducing a maximum performance of icefish in vivo (see Joyce et al., 2018b). To determine cardiac scope also requires reliable routineQ values, but these vary considerably, as pointed out by Joyce et al., (2018b), who also reported rou-tineQ for Chaenocephalus aceratus (Hb − Mb − ) at 0.8 • C that was lower than previously reported and increased with both moderate activity and with acute warming. By comparison, routineQ is just 6 ml min −1 kg −1 at ∼ 0 • C in N. coriiceps (Hb + Mb + ), which has an haematocrit of ∼ 16% (Egginton, 1997;Joyce et al., 2018b), although other red-blooded notothenioids of different ecotype and size have a higher routineQ (Axelsson, 2005). Moreover, N. coriiceps can increase routine oxygen uptake,Q, V S and f H appreciably after warm-acclimation (Egginton and Campbell, 2016;Joyce et al., 2018a). Other cardiovascular adaptations of icefishes include wide-bore arteries and capillaries that reduce overall systemic vascular resistance to just 10-20% that of Hb + fishes, and lower arterial blood pressure (Egginton, 1997) and cardiac afterload (Egginton et al., 2002). Thus, massspecific cardiac work (W C ) in icefishes is estimated to range between 0.35 and 2.3 mW g −1 (Axelsson, 2005), at the lower end of the range for teleost fishes (Farrell and Smith, 2017).
Therefore, indications exist that icefishes may have a reasonable cardiac scope and a moderate capacity to respond to warming, but without robust estimates of maximumQ for an Hb − Mb − icefish we cannot test these possibilities. Indeed, maximum cardiac performance has never been assessed in C. aceratus. Therefore, to better understand how loss of Mb and Hb expression might affect vulnerability to ocean warming, we hypothesized that warming impairs maximum cardiac performance of icefishes, especially when compared with the heart of a species possessing Hb and Mb. To test this hypothesis, we compared maximum cardiac performance at 1 • C and 4 • C for three Antarctic notothenioids that naturally express different levels of Hb and Mb. We reasoned that the loss of facilitated oxygen transport, either by systemic convection (Hb) or cellular diffusion (Mb), would limit maximum cardiac capacity. In comparing two icefish species that are Hb − the influence of Mb would be revealed, and the influence of Hb evident by comparison with a red-blooded nototheniid.

Collection of animals
Adult Chaenocephalus aceratus (Hb − Mb − ), Chionodraco rastrospinosus (Hb − Mb + ) and Notothenia coriiceps (Hb + Mb + ) were collected around the Western Antarctic Peninsula during the austral fall and winter of 2015 (see O'Brien et al., 2018). They were held at Palmer Station in ambient seawater temperature (0 ± 1 • C) for a maximum of 3 weeks before experimentation. N. coriiceps were fed cubes of fish muscle every other day, but not 24-48 h prior to surgery. The icefishes did not feed in captivity; they are known to maintain a lower metabolic rate than other notothenioids and likely grow sporadically, feeding at relatively broad but seasonally dependent time intervals (Johnston and Battram, 1993;Campbell et al., 2008).

The in situ heart preparation
We adopted a well-established in situ working, perfused heart preparation that has been used across a wide variety of fish species, including Arctic fish at low temperatures (e.g. Farrell et al., 1988Farrell et al., , 2013. This preparation retains full anatomical integrity of the heart and generates a maximumQ equivalent to that observed in vivo. Output pressure can be raised to determine the maximum pressure-generating ability and maximum W C (Farrell et al., 1991). All experiments were approved by the University of Alaska, Fairbanks Institutional Animal Use and Care Committee (570217-18).
Briefly, anaesthetized fish (Egginton and Campbell, 2016) were maintained on a surgical operating sling by continuously irrigating the gills with ice-chilled seawater containing MS-222 (50 mg l −1 ). Heparin (∼100 i.u. kg −1 in saline) was injected into the caudal vessels to prevent blood clots. A midline ventral incision, occlusion of the gut arterial blood supply with umbilical tape and removal of the intestines provided access to the hepatic veins, all but one of which were ligated. A saline-filled input cannula [connected to a Marriot bottle containing physiological perfusate (see below) bubbled with 100% oxygen] was advanced into the sinus venosus via the remaining vein and secured in place. Air bubbles sucked into the heart when this vein was cut were prevented from entering gill vessels (which would restrictQ) by severing the ventral aorta at the isthmus and anterior to the pericardium prior to the cut. The Marriot bottle height was adjusted so thatQ was not excessive and the exceptionally thin-walled cardiac chambers were not over filled. Next, the output cannula was inserted into the severed ventral aorta and secured. The ducts of Cuvier were then occluded so that the heart only received perfusate; this procedure also crushed all nerves supplying the heart. Within ∼ 20 min the heart preparation was transferred to and fully immersed in a saline bath (250 mM NaCl) housed inside a 4 • C environmental room. The input cannula was then connected to a pressure head that delivered oxygenated perfusate at a precise cardiac filling pressure (P i ). The bath and the reservoirs containing oxygenated perfusate were maintained at a constant temperature (either 1 • C or 4 • C) with a Neslab RTE recirculating water bath (Thermofisher Scientific, Waltham, MA, USA).
InitialQ was nominally set at ∼ 13 ml min −1 kg −1 body mass by adjusting P i (typically < 1 cm H 2 O and sometimes sub-ambient; 1 cm H 2 O = 0.1 kPa) to set V S while the heart beat with a myogenic rhythm. A constant pressure head device connected to the output cannula set the initial cardiac afterload ( (Egginton, 1997), but information on in vivo ventral aortic blood pressure to set their initial P o is lacking. Regardless, maximum P o was directly measured during the protocol by incremental increases until maximumQ began to fail, suggesting this was a reasonable approach, especially since we repeated this response several times for each preparation. These stable perfusion conditions were maintained for at least 10 min prior to assessing maximum cardiac performance. The pericardium was opened at the end of the experiment to ensure proper placement of input and output cannulae and weigh the heart.

Assessing maximum cardiac performance in situ
Protocols to generate maximum V S , maximumQ, maximum P o and a maximum cardiac work (W C ) are detailed elsewhere (e.g. Farrell et al., 1988). Briefly, incremental increases in P i with a myogenic heartbeat yielded a maximum V S and a maximum Q. All preparations gave a robust Starling response, similar to other teleost working heart preparations, with V S being most sensitive to a P i < 3 cm H 2 O; smaller incremental changes inQ occurred up to 5 cm H 2 O and occasionally up to 8 cm H 2 O. (Note: maximum V S could be higher in vivo if cardiac vagal tonus lowered f H , but maximumQ would not necessarily be greater.) Maximum P o was determined at maximumQ by raising P o in 5-10 cm H 2 O increments until maximum W C was attained, at which point any increase in P o was matched by an equivalent decrease inQ. While hearts can generate a higher P o , but not without disproportionately decreasingQ and W C , icefishes displayed an interesting phenomenon when P o was increased beyond the maximum W C, wherein f H abruptly halved, V S doubled and maximuṁ Q hardly changed; this situation was immediately reversed by reducing P o . After maximum cardiac performance had been assessed, routineQ and P o were rapidly restored, which quickly re-established a stable routineQ.
Adrenaline can enhance maximum cardiac performance of perfused teleost hearts (Farrell et al., 1986), including Arctic fish at 1 • C . However, low circulating levels of catecholamines in Antarctic fishes, even after moderate stress that rose substantially only in extremis (Whiteley and Egginton, 1999), raised the possibility that no adrenergic stimulation would be needed in the present study. Nevertheless, preliminary experiments showed that C. aceratus preparations quickly lost pumping capacity unless they had been stabilized with 50 nM adrenaline in the perfusate, which is a low concentration when compared to the levels induced by stress (∼5 μM). Accordingly, our initial perfusion conditions, including during surgery, were standardized to include 50 nM adrenaline. Thus, the initial f H reported below reflects a tonic adrenergic effect on the myogenic heartbeat. Since the preliminary experiments with all three species showed that adrenaline concentrations > 5 μM elicited no further stimulation of cardiac performance, maximum cardiac performance was assessed with three levels of catecholamine (50nM, 0.5 μM and 5 μM) to provide insight into the modulatory effects of adrenergic stimulation. The need to test a range of concentrations to establish the effect of different levels of adrenergic stimulation reflects a lack of prior data in the literature on these species, so was an essential prerequisite for a robust experimental design. Up to five concentrations have been tested previously without performance deterioration.

Temperature treatments
The three species were assessed at 1 • C, which was near their habitat temperature (0 ± 1 • C). Attempts to more closely match test and exposure temperature in a 4 • C environmental chamber produced unwanted ice formation. Acute warming to 4 • C of heart preparations from fish acclimated to 0 ± 1 • C were assessed with the experimental bath and perfusate reservoirs maintained at 4 • C), but only with C. aceratus (Hb − Mb − ) and N. coriiceps (Hb + Mb + ) because too few C. rastrospinosus (Hb − Mb + ) were caught. In addition, hearts from both N. coriiceps and C. rastrospinosus were similarly tested at 4 • C after an exposure period of a minimum of 5 days at 4 ± 0.5 • C in 700 l insulated recirculating seawater tanks (4-6 fish per tank); fish were initially held at 0 ± 1 • C for 24 h before increasing the temperature by 1 • C daily using 3-kW Elecro Titanium inline heaters (Aqualogic, San Diego, CA, USA). C. aceratus, which was intolerant of more than 1-2 • C rise in temperature, proved difficult to acclimate to 4 • C.

Measurements and analysis
Perfusate outflow was measured with an in-line Transonic probe (4.0 mm diameter) and flowmeter (T206, Transonic Systems, Ithaca, NY, USA), gravimetrically calibrated at 1 • C and 4 • C to adjust for temperature effects on factorycalibrated flow readings. P i and P o were measured near the input and output cannulae, respectively, via fluid-filled tubes connected to pressure transducers (DP6100, Peter von Berg, Medizintechnik, Germany) and referenced to saline level in the bath. Recorded pressures were corrected to provide true values using individual calibrations of cannula resistance. Flow and pressure signals were amplified (4CHAMP amplifier, Somedic, AB, Hörby, Sweden) and processed on-line (custom LabView program; National Instruments Sweden AB, Kista, Sweden) to display real time raw values of P i , P o , V S ,Q and f H (determined from the pulsatile flow trace) throughout the experiment. W C , the product of [Q (ml min −1 ) × (P o − P i ) × 0.00167]) was displayed on-line and used to determine maximum performance when P o was elevated. Data were analysed off-line for 5-10 stable, consecutive heartbeats for each incremental change in P i and P o . W C was corrected for ventricular mass (mW g −1 ventricle mass), andQ was corrected for body mass (ml min −1 kg −1 body mass).

Statistical analyses
All data are presented as mean ± standard error (s.e.m.), analysed using SPSS v.23 (SPSS Inc., Chicago, IL, USA). Intraspecific comparisons (temperature × adrenaline), or interspecific comparisons for similar exposure and test temperatures (species × temperature), used a two-way ANOVA with Tukey's Kramer post-hoc tests. A one-way Welch's ANOVA was used to compare differences between 1 • C-acclimated animals tested at 1 • C and 4 • C, and between 4 • C-exposed animals tested at 4 • C; the dose-response to adrenaline was tested by one-way repeated measures ANOVA with Sidák posthoc tests at each test temperature. Relative ventricular mass (RVM) was transformed by Lngamma function to achieve normality for statistical comparisons. Statistical significance was assigned to P ≤ 0.05.
Interspecific comparisons of maximum performance with 5 μM adrenaline at 1 • C f H was significantly (P < 0.05) lower in C. aceratus (Hb − Mb − ) than in either C. rastrospinosus (Hb − Mb + ) or N. coriiceps (Hb + Mb + ), whose f H was not significantly different from each other (Fig. 4). Maximum V S and maximumQ were highest for C. rastrospinosus, intermediate for C. aceratus and lowest for N. coriiceps. N. coriiceps generated a significantly (P < 0.05) and substantially higher maximum P o and maximum W C than both icefish species, which were similar to each other in this regard.
Response of maximum cardiac performance on acute warming to 4 • C

C. rastrospinosus (Hb − Mb + )
When 4 • C-exposed fish were tested at 4 • C, the initial f H (37.3 ± 3.3 min −1 ) was significantly (P < 0.05) higher compared with fish acclimated to 1 • C and tested at 1 • C (26.0 ± 0.5 min −1 ; Fig. 2). While 5 μM adrenaline produced a small, significant (P < 0.05) decrease in f H (33.8 ± 3.4 min −1 ) at 4 • C when compared with 50 nM adrenaline, maximum V S and maximumQ increased significantly (P < 0.05) by a small amount without significantly changing maximum W C . Even so, neither maximum V S (6.0 ± 0.7 ml kg −1 ), maximumQ ( : Maximum cardiac performance of the in situ perfused heart of C. aceratus (Hb − Mb − ) acclimated to 1 • C as a function of adrenaline concentration in the perfusate, tested at two temperatures. A significant difference between adrenaline concentrations is indicated by dissimilar numbers; an asterisk indicates a significant difference between the performance at 1 • C (n = 8) and 4 • C (n = 7) for a given adrenaline concentration.

N. coriiceps (Hb + Mb + )
When tested at 4 • C, the initial f H (36.0 ± 1.0 min −1 ) was significantly (P < 0.05) higher when compared with fish tested at 1 • C (Fig. 3). Stimulation with 5 μM adrenaline had no significant effects on maximum W C , maximum V S or maximuṁ Q, but produced a small, significant (P < 0.05) decrease in f H when compared with 50 nM adrenaline. While maximum W C (4.6 ± 0.3 mW g −1 ) with 5 μM adrenaline was significantly (P < 0.05) higher after a period of exposure to 4 • C compared with fish acclimated to 1 • C, neither maximum V S nor max-imumQ were different (P > 0.05). Moreover, none of the maximum performance parameters were different (P > 0.05) when 4 • C-exposed fish tested at 4 • C were compared with fish acclimated to 1 • C and acutely warmed to 4 • C (Fig. 4).

Interspecific comparisons with a period of exposure to 4 • C
While 4 • C-exposed N. coriiceps had a significantly (P < 0.05) higher maximum W C and maximum P o than 4 • C-exposed C. rastrospinosus, maximumQ and maximum V S were significantly (P < 0.05) lower (Fig. 4). In general, cardiac performance at 4 • C was similar whether acutely exposed : Maximum cardiac performance of the in situ perfused heart of N. coriiceps (Hb + Mb + ) as a function of the adrenaline concentration in the perfusate. Fish were acclimated either to 1 • C and tested at 1 • C (n = 7) and 4 • C (n = 6), or exposed to 4 • C for an acclimation period and tested at 4 • C (n = 6). Significant differences are identified as Fig. 1.
to this temperature change or measured after a period of exposure to 4 • C.

Maximum cardiac performance
We provide the first estimate of maximum cardiac pumping capacity for an icefish that lacks Mb, and the first comparison of maximum in situ cardiac performance among notothenioids differing in their expression of Hb and Mb. This comprehensive study of maximum cardiac function of notothenioid fishes now allows a better assessment of the likely scope for acclimation during Southern Ocean warming. While we show that N. coriiceps had a robust cardiac solid circle) as a function of test temperature, using 5 μM adrenaline in the perfusate. Dashed lines connect fish tested at their acclimation temperature (either 1 • C or 4 • C), solid lines connect fish acclimated to 1 • C and acutely warmed to 4 • C for testing. A significant difference is indicated by dissimilar numbers between species when tested at 1 • C, by dissimilar lower letters between species when tested at 4 • C and by an asterisk between performance at 1 • C and 4 • C for each species. performance at all of the temperature conditions we tested, the extremely large hearts associated with loss of Hb expression in icefishes incur a cost to maximum W C , which was considerably lower than in N. coriiceps. While the low vascular resistance of icefish (e.g. in C. hamatus it is only 10-20% that of Hb + fishes; Acierno et al., 1997) will offset this to an extent, low W C may still limit scope for activity despite an unusually high maximumQ near 0 • C. In this regard, the additional loss of Mb in C. aceratus may create a more precarious situation than C. rastrospinosus in a warmer world. Specifically, the maximumQ values recorded here are in the lower range of those previously published for icefishes (70-300 ml min −1 kg −1 ; Axelsson, 2005). Moreover, a functional consequence for the evolutionary loss of Mb may be our novel finding of a significantly higher maximumQ at 1 • C for C. rastrospinosus (Hb − Mb + ; 128 ml min −1 kg −1 ) than for C. aceratus (Hb − Mb − ; 80 ml min −1 kg). Routine Q in vivo is 27 ml min −1 kg −1 and maximumQ 80 ml min −1 kg −1 for C. aceratus at 1 • C (Joyce et al., 2018b), suggesting that cardiac factorial scope could be as high as 3 (i.e. 80/27) and similar to other teleosts (Farrell and Smith, 2017). However, routine Q in vivo for N. coriiceps at 1 • C is 7 ml min −1 kg −1 , while the maximumQ is 40 ml min −1 kg −1 , suggesting a higher cardiac factorial scope of 5.7. These comparisons suggest that the red-blooded N. coriiceps has the potential to withstand a warming climate better than icefishes, where conservation efforts within the Southern Ocean should be focused.

Species comparisons of maximum performance
In vitro measurements previously showed thatQ in icefishes at −0.5 • C (V S ∼ 3.5 ml kg −1 ) declined appreciably when afterload was increased to > 30 cm H 2 O (Acierno et al., 1997; cf . maximum P o in the present study ∼ 35 cm H 2 O). Interestingly, poisoning Mb with nitrite in C. rastrospinosus (Hb − Mb + ) significantly reduced W C to 0.8 mW g −1 (P o =29 cm H 2 O) from 1.1 mW g −1 (P o = 31 cm H 2 O), whereas W C in C. aceratus (Hb − Mb − ) was unaffected and remained at 1.4 mW g −1 with a similar afterload (Acierno et al., 1997). Unfortunately, this study did not assess maximumQ, but the present findings suggest that C. aceratus may have been working near maximum, unlike C. rastrospinosus. Previous in vitro estimates of maximumQ in an icefish are limited to those for C. hamatus (Hb − Mb + ) at 3 • C (200-300 ml min −1 kg −1 and maximum V S of 8-11.5 ml kg −1 ; Tota et al., 1991). Our in situ values for maximumQ at 4 • C were considerably lower for C. rastrospinosus (Hb − Mb + ; 175 ml min −1 kg −1 ) and still lower in C. aceratus (Hb − Mb − ; 98 ml min −1 kg −1 ); whether this reflects an important species variation or methodological differences is unclear. However, an in situ heart preparation is likely a superior preparation to placing a heart in an artificial pericardium, as potential damage during isolation is avoided. As observed, icefish hearts are easily overstretched by excessive filling pressure even when the pericardium is intact, resulting in a spontaneous decrease in f H . In addition, in vitro heart perfusions place an input cannula in the atrium at the sino-atrial junction, potentially impairing activity of pacemaker cells located there (Haverinen and Vornanen, 2007;Farrell and Smith, 2017); intrinsic f H for isolated heart preparations was far more variable for C. aceratus at −0.5 • C (22-30 min −1 ; Acierno et al., 1997) and C. hamatus at 0.1 • C (19.5-26.3 min −1 ; Tota et al., 1991) compared with the stable initial f H of our study. Notably, neither Tota et al. (1991) nor Acierno et al. (1997) used adrenaline in their in vitro studies, although it is plausible that catecholamines released from cardiac tissues could have accumulated in recycled perfusate.
Both C. hamatus and C. rastrospinosus have a much higher maximum V S compared with C. aceratus, a difference that may be related to Mb expression facilitating a somewhat higher cardiac afterload, while a 30% larger relative ventricular mass and a ∼ 60% higher maximumQ at 1 • C (C. rastrospinosus vs. C. aceratus; above) are also of potential benefit. Although both these species have similar values of maximum P o and W C , Mb has a much higher oxygen affinity than Hb and enhances both oxygen storage and diffusion rate within cells. Other species differences among notothenioids also might affect maximal cardiac performance, e.g. icefish hearts lack mitochondrial creatine kinase (O'Brien et al., 2014) and energy charge of N. coriiceps hearts is typically higher than C. aceratus (O'Brien et al., 2018). In addition, the Mb + state (independent of Hb loss) is associated with a 22% higher initial f H at 1 • C (26 min −1 in C. rastrospinosus vs. 19 min −1 in C. aceratus), a difference also seen at 4 • C and between N. coriiceps and C. aceratus. Because the sinoatrial node was intact in the in situ preparation, we are confident that f H was driven by a steady pacemaker activity, justifying our novel suggestion of a species difference in initial f H among icefishes. A similar species difference was observed in vivo after blockade of cardiac vagal tone (24 min −1 for N. coriiceps and 14.2 min −1 for C. aceratus) (Joyce et al., 2018b), while initial in situ f H for C. aceratus (19.0 min −1 ) compares well with in vivo f H measured immediately following activity (17.6 min −1 ; Joyce et al., 2018b). Of potential value in setting resting metabolic rate, and hence scope for activity, comparisons of initial f H values are confounded by the tonic level (50 nM) of adrenaline used to stabilize the in situ heart preparation; f H of C. aceratus may simply be less sensitive to tonic adrenergic stimulation at concentrations < 50 nM than C. rastrospinosus (both species were unresponsive to further additions of adrenaline). In contrast, with a similar preparation, M. scorpius f H increased by 22% at 1 • C and by 27% at 6 • C with maximal adrenergic stimulation, stimulatory effects that were fully reversed by β-adrenergic blockade .
Maximum W C of N. coriiceps at 1 • C was impressive when compared with those of temperate-water teleosts at much warmer temperatures (Farrell and Jones, 1992;Axelsson, 2005;Farrell and Smith, 2017), maintaining maximumQ against a comparable afterload. Indeed, it shows a remarkably similar maximum cardiac performance to M. scorpius (Hb + Mb + ) from Greenland, with comparable adaptive response observed on acute warming . Despite a lower maximum V S andQ, N. coriiceps hearts generated more than twice the relative work output compared with both C. rastrospinosus and C. aceratus, in part due to an enhanced pressure-generating ability compared with icefish hearts, as reported earlier (Acierno et al., 1997). However, with the much larger ventricle of icefishes, absolute maximum cardiac work output for a 1 kg representative fish was similar for all three species. Whether this translates to a similar resilience to environmental change will depend on the origin of physiological challenges to be overcome.
Fish hearts are exquisitely sensitive to cardiac filling pressure (venous return), and perfused hearts can generate an output even with a sub-ambient P i (C. hamatus: Tota et al., 1991;C. aceratus: Acierno et al., 1997 and present study;C. rastrospinosus and N. coriiceps: present study). However, neither icefish species examined here displayed the sensitivity to preload shown by C. hamatus (Tota et al., 1991), which increased V S up to 11.5 ml kg −1 (Q up to 300 ml min −1 kg −1 ) with just a 0.25 cm H 2 O (≈0.025 kPa) increase in P i , although why this difference exists is unclear. Adrenergic stimulation can increase sensitivity to P i (Farrell and Jones, 1992), as seen in the present study and with M. scorpius , by increasing maximum V S . This effect may be particularly important for icefishes where high concentrations of nitric oxide resulting from lack of Hb conversion to nitrate (Beers et al., 2010) may reduce adrenergic-stimulated contractility in vivo (Rastaldo et al., 2007). The importance of adrenergic stimulation for cardiac contractility has been demonstrated in notothenioids: maximum tension of isolated ventricular strips from C. aceratus increased by 47% with adrenaline concentrations similar to those of the present study, while N. coriiceps maximum tension increased by a remarkable 377% with 1 mM adrenaline (Skov et al., 2009). These positive inotropic effects were more marked and speciesspecific than the modest stimulatory effects of adrenaline on V S orQ (20-30% increases) seen in situ, so may reflect a potentially powerful role for tonic adrenergic stimulation of contractility not fully realized in vivo. Differences in adrenergic sensitivity could reflect species difference as well as thermal effects: adrenaline responsiveness increased at 4 • C in N. coriiceps (present study) and at 6 • C in M. scorpius , although a dependency on tonic adrenergic stimulation existed in rainbow trout (Oncorhynchus mykiss) hearts at 5 • C, but not at higher temperatures (Graham and Farrell, 1989). Ventricular strips of both C. aceratus and N. coriiceps developed alternans at high-pacing frequencies, an impairment exacerbated by adrenaline (Skov et al., 2009). A protective effect may be provided by the modest negative chronotropic effect of adrenaline, seen in both N. coriiceps and C. rastrospinosus at 4 • C (present study), perhaps due to a stimulation of cardiac α-adrenoceptors or a prolongation of the ventricular action potential (Farrell and Jones, 1992). Clearly, the subtle differences in adrenergic stimulation of the heart that exist among polar fish species need further study to better understand their autonomic cardiac regulation, as sympathovagal balance helps define a species ecotype (Campbell et al., 2009).

Temperature effects
The present study revealed a clear (30-40%) increase in f H when notothenioid species were tested at 4 • C, independent of whether the fish had been acutely warmed or experienced an exposure period at 4 • C. Temperature has a variety of direct cardiovascular effects, but perhaps the most important and universal is an increase in f H during acute warming (e.g. Sandblom and Axelsson, 2007;Farrell, 2009;Eliason et al., 2011;Keen and Gamperl, 2012). Notothenioids are clearly no exception in this regard, e.g. isolated C. hamatus hearts warmed from 0.6 • C to 5.8 • C, increased f H by 21% to 24 min −1 (Tota et al., 1991) and C. aceratus doubled f H to 36 min −1 when exposed for 48 h to 10 • C (Hemmingsen et al., 1972). Routine f H andQ both increased with acute warming from −1 • C to 8 • C in another Antarctic fish, Pagothenia borchgrevinki (Franklin et al., 2007). However, its scope to increase f H , V S andQ following burst activity all decreased with increasing temperature because the increase in maximum cardiac capacity did not match the greater routine oxygen uptake. In the present study, both C. aceratus and N. coriiceps increased f H by a similar amount when acutely warmed. Notably, only N. coriiceps significantly increased W C (by 64%), largely through an increase in maximum P o , while C. aceratus maintained maximum V S and maximum P o , but increased maximumQ. Interestingly, the cardiac response to warming in N. coriiceps was similar to M. scorpius, which is considered a temperature generalist (Franklin et al., 2007) and maintained scope forQ with forced activity over a 10 • C acute warming, whereas two closely related Arctic sculpins with a more restricted range could not do so . These comparisons suggest that the ability to tolerate ocean warming may differ even among red-blooded polar species, as well as among icefishes.
The period of exposure used here was short relative to the slow responses known for some Antarctic fishes (Franklin et al., 2007;Franklin and Seebacher, 2009;Egginton and Campbell, 2016), even though they involved only small temperature changes. Therefore, conclusions concerning cardiac thermal acclimation for C. rastrospinosus and N. coriiceps must be treated with caution. For example, acclimation of P. borchgrevinki (Hb + Mb + ) to 4 • C for 4-5 weeks reduced peak Q after exercise at −1 • C (factorial scope of 1.4 vs. 2.6 for −1 • C acclimated fish), while peakQ was maintained on acute exposure to 4 • C and 8 • C (Franklin et al., 2007). Thermal compensation maintained cardiac performance at the new acclimation temperature and allowed scope forQ change up to 8 • C, but at the expense of reduced cardiac performance when acutely cooled to −1 • C (Franklin et al., 2007). For N. coriiceps maximum performance with acute warming and exposure to 4 • C were not statistically different; in contrast, thermal tolerance increased by 1.  (Bilyk and DeVries, 2011). Like acute warming of C. aceratus, C. rastrospinosus maintained maximum V S , maximumQ and maximum P o after exposure at 4 • C (present study). Clearly, a more extensive exploration of thermal acclimation effects in polar fishes is warranted, which may allow the capacity for different niche selection to be predicted, although facilities in remote areas remain a challenge for such detailed studies.
It is possible that over evolutionary time icefishes could accommodate warming as the Patagonian icefish Champsocephalus esox apparently has done, where surface water temperature is 10-12 • C over its summer range. In addition, some notothenioids have escaped the stable cold waters of the Southern Ocean and successfully colonized warmer temperate water. Stranded following a northerly extension of the ice front the black cod, Notothenia angustata, has established a viable population around the southern coast of New Zealand.

Concluding remarks
We provide robust estimates of maximal cardiac performance in two species of Channichthyid icefishes lacking facilitated systemic oxygen transport. While exhibiting the expected rise in heart rate on warming, our results indicate a potential to accommodate some degree of climatic change, which would have significant bearing on species survival in a warming world. While the loss of Hb is associated with large ventricles that deliver high cardiac output, retention of Mb allows additional improvements in cardiac function. Although both icefish species increased cardiac pumping capacity when warmed to 4 • C, their hearts were unable to tolerate a high afterload, suggesting limited capacity to cope with stressful situations where sympathetic drive may increase peripheral resistance. However, during acute warming with activity C. aceratus increased in vivoQ 3-fold and vascular conductance 5-fold, suggesting that short-term cardiovascular compensation is possible (Joyce et al., 2018b). In contrast, a closely related species expressing both Hb and Mb was capable of higher maximum work capacity, an enhanced performance that may convey greater resilience to near-future ocean warming. Although these data demonstrate a limited capacity to withstand more frequently occurring short-term thermal extremes, and perhaps the potential to accommodate a gradual warming of the Southern Ocean, long-term temperature acclimation experiments are required to understand interspecific limits for stock conservation.
This novel information allows greater insight into plasticity of fish species previously considered to be living perilously close to their upper thermal limits as a consequence of unique physiological constraints. The issue is of some urgency, as they inhabit ocean regions around Antarctica that are experiencing the most dramatic effect of global warming; compelling reasons why we need to understand the ecophysiology of such species before conservation efforts become untenable. In addition, there has been relatively little research on ther-mal resilience of such important predators in the simplified food chain, and we provide the most reliable cross-specific comparison on maximal cardiac performance, a key factor in ecological fitness, to date.