Aerobic energy metabolism
During acute warming, the RMR of N. rossii increased as a result of rising metabolic rate with a Q10 of 3.1 (Figure 1), which is consistent with previous studies on acute thermal responses of different Antarctic fish species [11, 26, 66, 67]. A part of the increase in metabolic rate can likely be attributed to the high costs of the low capacity cardio-vascular system to meet the increasing metabolic oxygen demand, which has been demonstrated for Antarctic eelpouts [13, 68]. Due to limited cardiac scope and increased friction of the vascular system at high blood flow rates, sufficient oxygen delivery at warm temperatures result in a relatively higher workload for the heart. A compensation of cardiac scope during long-term acclimation may alleviate this to some extent, contributing to a lower RMR and Q10 of 2.3 after warm acclimation (Figure 1), which indicate a partial, incomplete compensation of RMR in N. rossii (type 3 after Precht ). The same effect occurred in the long-term warm hypercapnic acclimated N. rossii, providing evidence that the partially compensated RMR in the warm hypercapnic fish was exclusively induced by temperature and not by elevated PCO2.
Following long-term acclimation of N. rossii to higher temperatures and P CO2, animal condition displayed pronounced changes. While control values of HSI and condition factor were within the same range reported recently for N. rossii caught in Potter Cove, King George Island , they were reduced in warm and hypercapnia acclimated N. rossii, although the fish were fed to satiation (control HSI 1.83, condition factor 1.69, warm hypercapnic HSI 0.81, condition factor 1.5; see Table 2). This reduction may be attributed to a reduced aerobic scope caused by elevated RMR (see above) at the high acclimation temperatures chosen.
The fact that the warm-acclimated fish could not completely compensate their RMR to a level comparable to the control animals could indicate beginning limitations in the circulatory system of N. rossii and in oxygen supply to tissues. As a result, the aerobic scope for the SDA response (specific dynamic action) [70, 71] might be limited at warmer temperatures. Consequently, fish may not be capable to ingest sufficient food over time to meet the required energy demand and to sustain basal metabolic rate, even if fed ad libitum. To maintain RMR elevated in warmer water, energy stores such as liver fat may be mobilized , resulting in the observed lower HSI and condition factor (see above).
The paradigm that Antarctic fish have limited acclimation capacity because of their thermal specialization has been challenged by several studies reporting compensatory adjustments of whole animal respiration, cardiovascular response and blood viscosity at elevated temperatures [11, 16, 24, 66, 73]. Most of these studies focused on the cryo-pelagic fish P. borchgrevinki or on several Trematomus species. In N. coriiceps, the congener of N. rossii, an acclimation-induced shift in critical thermal maxima (CTmax) was observed, but the increase was small compared to the shifts observed in other Antarctic species (e.g. P. brachycephalum, Gobionotothen gibberifrons, T. pennellii and T. hansoni, see Bilyk and DeVries  for further details), further corroborating our conclusions for N. rossii.
In general, the measurement of RMR provides a suitable indicator of a species’ thermal tolerance, as limits in oxygen consumption can reflect the onset of whole animal oxygen limitation and associated limitations in circulatory capacity . Nevertheless, the precise determination of aerobic limits under increased temperature and P CO2 benefits from the combined study of several indicators including enzyme and mitochondrial capacities, the limiting factors in ATP supply.
The mitochondrial state III respiration of the control group rose continuously with rising experimental assay temperatures of 0, 6 and 12°C. The RCR values were stable between 0 and 12°C (see Table 3), indicating efficient mitochondrial coupling up to 12°C. A decrease in Q10 from 2.4 (range 0-6°C) to 1.6 (range 6-12°C), indicates that state III respiration became less responsive to temperature at higher assay temperatures, and led to a similar decrease in mitochondrial scope as reported for N. rossii and Lepidonotothen nudifrons beyond 9°C.
In contrast to the elevation of RMR in the warm normocapnia acclimated N. rossii, maximum mitochondrial respiration rates were not significantly higher than in the control group (Figure 2). This was reflected in similar values of RCR (control: 4.6±0.8, warm normocapnia: 4.0±1.0) and Q10 between these two groups over the whole range of assay temperatures (Q10 from 0 to 12°C, control: 1.6; warm normocapnia: 1.8) (see Table 3). Only the trend towards a reduced thermal slope of mitochondrial state III respiration of the warm normocapnic fish (linear regression analysis of state III respiration from 0 to 12°C assay temperature; warm normocapnic: slope 0.22 [(nmol O2*min-1*mg-1)*1°C], control group: 0.39) could point towards a beginning compensation of mitochondrial respiration. The partial (type III) compensation at the whole animal level (RMR, Figure 1) could therefore originate at the mitochondrial level, possibly underpinning relevant adjustments of the cardiovascular system.
Interestingly, cold hypercapnia acclimation led to significantly reduced state III respiration at acute assay temperatures of 6°C and 12°C compared to the control group, accompanied by a significantly reduced mean RCR over the whole range of assay temperatures (Table 2). Similarly, state III respiration was depressed in the warm hypercapnia acclimated N. rossii, below that of the control group (Figure 2), and showed significantly reduced RCRs, indicating a clear effect of elevated ambient P CO2 on mitochondrial metabolism. This effect did not translate into a change in whole animal RMR but may reflect a decrease in tissue and whole animal aerobic and functional scope. In contrast, a down-regulation in resting aerobic metabolic rate occurred under acute hypercapnic acidosis in muscle tissue of the invertebrate Sipunculus nudus, reflecting a reduction in ATP consuming processes of maintenance metabolism (e.g. anabolic/ catabolic protein metabolism) [75, 76]. Such energy savings might also occur in fishes and affect proteins involved in mitochondrial respiration (e.g. reduced citrate synthase activities in hypercapnia acclimated Sparus aurata), thereby causing a lower state III respiration. Since state IV respiration remained unchanged after cold or warm hypercapnia acclimation (data not shown), the reduced coupling capacities were likely caused by the reduced state III respiration (per mg mitochondrial protein) and not by increased proton leak rates. The reduced COX activities (per mg cellular protein) (Figure 3) in the liver of both cold and warm hypercapnia acclimated N. rossii support this hypothesis and are in line with the projected changes in protein activity, including possible modifications in the mitochondrial membrane.
The differences observed at the mitochondrial level only partially reflected the whole animal level (see Figure 1), specifically in that the mitochondrial studies exclusively concentrated on liver tissue, which only constitutes a fraction of whole animal metabolism. A possible whole organism consequence of such capacity limits in mitochondrial metabolism under conditions of elevated energy demand (e.g. activity, reproduction) may be shifts in metabolic pathways  and a decrease in aerobic scope under long-term elevated P CO2. Further alterations may include a reduction in growth or behavioural capacities under long-term increased P CO2, as observed in coral reef fish (Amphiprion percula & Neopomacentrus azysron) [77–79].
The changes in mitochondrial capacities may be related to shifts in extra- and particularly intracellular acid–base status. The liver pHi in control N. rossii of this study (pHi 7.08, Table 4) were similar to values recorded for the eelpout Z. viviparus (pHi 7.06, ). The pHi values of the white muscle samples (e.g. control group; pHi 7.325) were close to values reported for Antarctic and non-Antarctic fish in other studies (e.g. G. morhua 7.34, ; P. brachycephalum 7.42-7.43 , Harpagifer antarcticus pHi 7.36 at 1°C , 7.33 N. coriiceps). In the warm normocapnia acclimated group, white muscle values followed the α-stat pattern , with a lowering of pH with increasing temperature by −0.014 pH units/°C. Such a rise in body temperature also caused a linear drop of pHi in white muscle of the North Sea eelpout Z. viviparus (−0.016 pH units/°C) . The illustration of intracellular acid–base parameters in the pH-bicarbonate diagram (Figure 4) emphasizes a defence of liver pHi by the non-bicarbonate buffer system (such as proteins or amino acid residues) in the cold hypercapnia acclimated fish, in similar ways as recorded e.g. for G. morhua or freshwater catfish Liposarcus pardalis.
In the liver of the warm hypercapnia acclimated N. rossii (Figure 4), pHi was compensated by intracellular HCO3- accumulation, in parallel to the findings in the blood (pHe) and muscle pHi. This compensation of chronically increased P CO2 of both cold and warm groups may have contributed to the observed shifts in metabolic steady state towards slightly alkaline pH values. The long-term reaction to acute changes in acid–base status may include shifts in the use of metabolic substrates by favouring oxidative decarboxylation of dicarboxylic acids (malate, glutamate/ aspartate) [75, 84]. These reactions could help to reduce the elevated proton load under chronic elevated P CO2, thereby playing an important role in the buffering of changes in the acid–base status. Nevertheless, such modifications appear to be insufficient to maintain full mitochondrial capacities in N. rossii, paralleled by the observed reduced COX activities and RCR of the cold/ warm hypercapnic mitochondria.
During long-term elevated ambient P CO2, CO2 enters the mitochondria by diffusion, yielding an increase in proton and [HCO3- levels. Taking into account the pH and total CO2 gradient maintained between mitochondria and the cytoplasm under control conditions, liver mitochondrial [HCO3- of the warm hypercapnic animals were up to 10 mmol/l higher than in the warm normocapnic group (or 4 mmol/l in the cold hypercapnic group compared to their controls); calculated after [47, 85]. Earlier studies of the acute effects of alkaline pH and increased [HCO3- on liver mitochondria revealed inhibitions in the TCA-cycle , thereby lowering mitochondrial respiratory capacities and their capacity to supply ATP. In trout (O. mykiss) hepatocytes, acutely increased [HCO3- at 1 kPa P CO2 also depresses mitochondrial metabolism via interruptions in the TCA-cycle, possibly caused by alterations in citrate and phosphate transport .
Intracellular acid–base regulation is supported by the respective adjustments in extracellular acid–base status, e.g. the accumulation of extracellular bicarbonate during compensation for the respiratory acidosis . The same shift in ‘set points’ towards alkaline values observed at the intracellular level occurred in the blood. In contrast to other pHe values recorded for temperate marine fish (e.g. cod 7.95 , flounder 7.78 , seabream 7.65 ), the extracellular pH of N. rossii was quite low in the present study (pH 7.44 in the control group). A study by Egginton  revealed a low blood pH of 7.5 for N. coriiceps directly after capture, which increased to 7.7 over 96 hours during recovery from landing stress. In cannulated N. coriiceps pHe increased from 7.5 to 8.0 during recovery, a value consistent with the blood pH of 8.01 measured for N. rossii by Egginton et al. . pH values measured in the present study may be lower than these values due to ‘grab and stab’ effects, as the cannulation of animals was experimentally not possibly. Nevertheless, these handling effects should have affected measurements in all experimental groups in similar ways and thereby still allow for comparison between the different acclimation groups.
As expected for marine teleost fish (e.g. Conger conger, G. morhua or Sparus aurata), the acute acidosis evoked by higher environmental P CO2 was compensated for by a significant increase in plasma [HCO3- in both the cold and warm hypercapnic groups. The depiction in the pH-bicarbonate diagram (Figure 5) shows that the increase in plasma [HCO3- cannot solely be attributed to extracellular non-bicarbonate buffering. Instead, combined acid–base parameters were positioned above the non-bicarbonate buffer line, likely due to the involvement of proton equivalent ion transfer processes . Although the pattern of compensation is similar for many teleost fish, the [HCO3- reached differ between species: e.g. levels reached 22 mM in C. conger, and 32 mM in cod, respectively  when exposed to 1 kPa CO2. The exposure to a moderate P CO2 of 0.2 kPa led to lower but still significantly elevated [HCO3- of 11.3 mM in cold hypercapnic N. rossii.
The compensation of higher ambient P CO2 via elevated [HCO3-e and [HCO3-i can lead to an increase in ATP demand for ion exchanging processes to maintain [HCO3- at this higher level, as it was reported for long-term hypercapnia acclimated eelpout (Z. viviparus) . The reaction to this constantly higher ATP demand could be a shift in energy budget with reduced ATP consuming processes, e.g. protein turnover or anabolism . This new metabolic equilibrium under increased metabolic demands for acid–base regulation could result in shifted ‘set points’, as we observed in the warm or cold hypercapnia acclimated N. rossii, with pH shifted towards alkaline values and thus a constant, slight metabolic alkalosis in both groups.
Both temperature and hypercapnia influence blood parameters in Antarctic fish, which has been demonstrated for blood osmolarity after thermal acclimation in notothenioids . This explains the observed decreased serum osmolarities in our warm acclimated animals. The unaffected osmolarities in cold hypercapnia acclimated N. rossii are in line with earlier findings by Larsen et al.  for cod (G. morhua) exposed to 1 kPa CO2. Although we observed hypercapnia induced changes in ion regulation, the higher [HCO3-e in the blood are too small to significantly alter total osmolarity. Hence, the changes in osmolarity can exclusively be attributed to long-term warm acclimation.
The haematocrit levels of N. rossii were unaffected by warm and/ or hypercapnia acclimation, and within the range reported for its sympatric sister species N. coriiceps[26, 95] (Table 2). While acute warming causes an elevation of haematocrit in red-blooded notothenioids , long-term warm acclimation leaves haematocrit levels constant, consistent with results from other studies on Antarctic notothenioids . Thus, the oxygen carrying capacities of the blood of warm and/ or hypercapnia acclimated N. rossii do not seem to be limiting under these conditions.
It has been assumed that the extracellular non-bicarbonate buffering is mostly accomplished by proteins in the blood , and thus strongly depends on haematocrit , which varies greatly between fish species, also among Antarctic fish species [26, 28, 98, 99]. The haematocrit levels measured in N. rossii (28–31, see Table 2) thus result in high blood βNB values (30.3 mmol/l pH). Similarly, the red-blooded Antarctic fish Dissostichus mawsoni and P. borchgrevinki showed higher βNB values (~28 and 18 mmol/l pH, respectively) than the haemoglobinless icefish Pagetopsis macropterus (βNB ~3 mmol/l pH) .
Some possible limitations in oxygen availability may have occurred at the intracellular level in the warm normocapnia acclimated fish, where liver pHi was lower than in the control group. This pH difference cannot be exclusively attributed to α-stat regulation , as pHi changed by −0.032 pH units/°C. Possibly, the high acclimation temperature of 7°C led to limiting oxygen supply to the liver tissue as a consequence of elevated metabolic demand,, resulting in a slight contribution of anaerobic metabolism and thereby lactate production, thus shifting the pHi of the warm normocapnic group to acidic values. Nevertheless, other tissues with lower metabolic loads than liver, such as white muscle, may still be able to metabolise anaerobic end products to some degree, allowing the animals to survive at these warmer temperatures (4–6 weeks acclimation time in this study).
We did, however, not observe elevated lactate values in the blood of warm normocapnic/ hypercapnic fish, and the generally low lactate values of N. rossii were similar to those measured in N. rossii and N. coriiceps earlier [28, 101]. Only in the cold hypercapnia acclimated animals, lactate levels were slightly elevated, but are likely the result of minor handling stress and do not originate from a beginning anaerobic metabolism in the liver, as they are still close to the levels of 1 mM reported for N. coriiceps under natural conditions [26, 90].
A higher ATP demand under conditions of elevated temperature in combination with an intracellular acidosis might shift or even impair liver functionality over a longer time-scale in the warm-acclimated animals (normocapnia/ hypercapnia), which could relate to the reduced HSI in the animals of the present study (see Table 2).