# Rising CO2 enhances hypoxia tolerance in a marine fish

## Abstract

Global environmental change is increasing hypoxia in aquatic ecosystems. During hypoxic events, bacterial respiration causes an increase in carbon dioxide (CO2) while oxygen (O2) declines. This is rarely accounted for when assessing hypoxia tolerances of aquatic organisms. We investigated the impact of environmentally realistic increases in CO2 on responses to hypoxia in European sea bass (Dicentrarchus labrax). We conducted a critical oxygen (O2crit) test, a common measure of hypoxia tolerance, using two treatments in which O2 levels were reduced with constant ambient CO2 levels (~530 µatm), or with reciprocal increases in CO2 (rising to ~2,500 µatm). We also assessed blood acid-base chemistry and haemoglobin-O2 binding affinity of sea bass in hypoxic conditions with ambient (~650 μatm) or raised CO2 (~1770 μatm) levels. Sea bass exhibited greater hypoxia tolerance (~20% reduced O2crit), associated with increased haemoglobin-O2 affinity (~32% fall in P50) of red blood cells, when exposed to reciprocal changes in O2 and CO2. This indicates that rising CO2 which accompanies environmental hypoxia facilitates increased O2 uptake by the blood in low O2 conditions, enhancing hypoxia tolerance. We recommend that when impacts of hypoxia on aquatic organisms are assessed, due consideration is given to associated environmental increases in CO2.

## Introduction

A lack of O2 is one of the greatest challenges that most life can face. In terrestrial ecosystems conditions of low O2 are rare. In contrast low O2, referred to as hypoxia, is much more common in freshwater and marine ecosystems1,2,3,4. Hypoxia occurs because high biological demand for O2 can exceed the rate of O2 supply to the ecosystem, leading to a reduction in environmental O2 levels5,6. However, the challenges of hypoxia are not solely a result of reduced O2. Organisms must also contend with simultaneous but reciprocal changes in the other respiratory gas, CO2.

When O2 decreases in aquatic systems there is a corresponding increase in CO27,8. This is a by-product of respiration, the same process that causes depletion of O2. As such high CO2 during hypoxia is ubiquitous and unavoidable. This coupling of O2 and CO2 has been highlighted numerous times in oceanographic sciences, most recently by Robinson5. Yet unaccountably, despite the known link between decreasing O2 and increasing CO2 during hypoxia8,9, the issue of increased environmental CO2 during periods of low O2 has been relatively overlooked by biologists.

Implications of rising CO2 during hypoxia on aquatic organisms are particularly important to address in the face of human driven climate change. Hypoxic areas are predicted to become more common and more severe, particularly in marine systems, with the de-oxygenation of the world’s oceans recently highlighted as a major component of climate change2,3,10,11,12. In addition, there will be an increase in ambient CO2 as rising atmospheric CO2 is absorbed by the world’s oceans13. Non-linear interactive effects between higher atmospheric CO2 and CO2 accumulation during hypoxia will lead to increased CO2 levels during hypoxia in future oceans14. This means that effects of rising CO2 during hypoxia in marine systems will be amplified by climate change.

Typically, experiments which test responses to hypoxia or impacts from hypoxia on aquatic organisms create hypoxic conditions by off-gassing oxygen from water by gassing with pure nitrogen or a mix of nitrogen (N2) and O2 (for examples see15,16,17,18). This creates low O2 conditions without the concurrent CO2 increase that would be expected in the environment. The lack of studies in which an environmentally realistic simultaneous decrease in O2 and increase of CO2 have been conducted may lead to mismeasurement of responses to hypoxia. Recently, several studies on marine fish and invertebrates have demonstrated interactive effects of low oxygen and increased CO29,19,20, with some species exhibiting loss of equilibrium (LoE) and death at higher O2 concentrations when CO2 is simultaneously elevated19,21. However, these experiments do not give insight into the physiological mechanisms underlying the influence of CO2 on hypoxic responses of fish.

Previously observed impacts of hypoxia-associated rises in CO2 on hypoxia tolerance of fish could be a result of changes in O2 uptake, as CO2 has been shown to impact upon several aspects of organismal biology that are involved in O2 uptake and transport22,23. We aimed to assess whether concurrent increases in CO2 during decreases in O2 affect O2 uptake in a marine fish, the European sea bass (Dicentrarchus labrax) by conducting a standard critical O2 level test (O2crit). Under normal O2 (normoxic) conditions fish maintain a minimum level of O2 consumption rate ($${\dot{{\rm{M}}}{\rm{O}}}_{2}$$), referred to as the standard metabolic rate (SMR), in order to meet maintenance energetic demands of essential processes through aerobic respiration24. As the level of O2 in water drops fish deploy a number of responses (i.e. increased ventilatory water flow and cardiac output, increased haematocrit, functional changes in gill morphology, changes in Hb-O2 affinity) in order to maintain and regulate this minimum level of $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$25,26,27. If environmental O2 continues to drop there comes a point at which fish are unable to regulate $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ to meet minimal energy demands, referred to as the critical O2 level (O2crit). At O2 levels below O2crit fish become oxy-conformers (where $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ is directly proportional to environmental O2 availability) and fish become increasingly reliant on anaerobic metabolism which is unsustainable in the medium to long term. In the past the measure of O2crit has been used as a proxy for overall hypoxia tolerance but recently this approach has been questioned28,29. Nevertheless O2crit does provide information related to the ability of fish to maintain O2 uptake and supply during hypoxia, and its prevalence in the literature allows comparison of responses between species30. Furthermore, we investigated whether any changes in O2crit could be linked to changes in blood acid-base chemistry and blood gas transport via alteration of Hb-O2 binding caused by rising environmental CO2. Our hypothesis was that the simultaneous increase in CO2 during a progressive decrease in O2 would decrease hypoxia tolerance (increase O2crit) and that this response may be a result of blood acid-base disturbance decreasing Hb-O2 affinity and O2 transport.

## Results

### O2crit tests

There was evidence of enhanced tolerance to hypoxia in sea bass exposed to rising compared to constant CO2 conditions. This was indicated by measurements of O2crit in European sea bass being significantly different between fish exposed to either constant, or rising CO2 levels during O2crit tests when accounting for variation in SMR (Fig. 1; ANCOVA, F1,12 = 7.525, p = 0.0178). A CO2 increase during O2crit tests resulted in a 20% reduction of O2crit (3.88 ± 0.19 kPa O2, 18.7 ± 0.9% air saturation, mean ± S.E.) when compared to tests in which CO2 levels were maintained at ambient levels (4.87 ± 0.22 kPa O2, 23.4 ± 1.1% air saturation, mean ± S.E.).

### Blood chemistry analysis

A comparison of blood chemistry parameters between the two treatment groups indicated that sea bass fully compensated for the rise in CO2 during hypoxia within 5 hours (the period of exposure prior to blood sampling). Blood pH (pHe) was not different between fish exposed to constant ambient CO2 (7.87 ± 0.03) and fish exposed to progressively rising CO2 (7.88 ± 0.02) (Fig. 2A, GLM, F1,14 = 0.23, p = 0.64). The acidifying effect of the ~79% rise in blood pCO2 levels in the rising CO2 regime (0.272 ± 0.019 kPa CO2) compared to the constant ambient CO2 regime (0.152 ± 0.015 kPa CO2) (Fig. 2C, GLM, F1,13 = 23.9, p < 0.001) was fully compensated by elevating blood HCO3 (Fig. 2D, GLM, F1,13 = 40, p < 0.001). Plasma HCO3 was 88% higher under the rising CO2 regime (6.76 ± 0.29 mM) when compared to the constant ambient CO2 (3.60 ± 0.43 mM). There were no differences in haematocrit (Fig. 2B, general linear model, F1,13 = 0.69, p = 0.42) or plasma lactate (Fig. 2E, general linear model, F1,14 = 1.48, p = 0.24) between fish sampled under a constant ambient CO2 regime (haematocrit = 39.4 ± 0.8%, lactate = 0.88 ± 0.26 mM) or rising CO2 regime (haematocrit = 38.2 ± 1.1%, lactate = 0.53 ± 0.12 mM). Blood glucose levels were ~26% lower in fish exposed to a progressively rising CO2 regime (4.50 ± 0.53 mM) when compared to a constant ambient CO2 regime (6.09 ± 0.31 mM) (Fig. 2F, GLM, F1,14 = 6.74, p = 0.021).

### Haemoglobin affinity for oxygen

Oxygen affinity of haemoglobin was increased in fish sampled under the progressively rising CO2 regime (Fig. 3A, GLM, F1,12 = 10.42, p = 0.0073). Haemoglobin P50 was decreased by ~32.5% in fish under the progressively rising CO2 regime (1.64 ± 0.15 kPa O2, 7.9 ± 0.7% air saturation) compared to fish sampled with a constant ambient CO2 (2.43 ± 0.20 kPa O2, 11.7 ± 1% air saturation). There was no significant change in Hills number between treatments (Fig. 3B, GLM, F1,12 = 0.50, p = 0.494).

## Discussion

Our results highlight the biological importance of simultaneously rising CO2 under conditions where O2 levels in water are depleted – a scenario that reflects the natural conditions during hypoxia which will be exacerbated by climate change – by demonstrating that ecologically relevant changes of CO2 impact physiological performance of a marine fish at both the molecular and whole organism level. We hypothesised that rising CO2 during progressive O2 decreases would lead to an increase in O2crit as a result of increased blood CO2, decreased blood pH and the associated Bohr/Root effect of fish haemoglobin (Hb), in which Hb-O2 affinity (Bohr effect) and the total capacity of Hb for O2 (Root effect) are reduced when pH falls. In contrast, we show that increasing CO2 as O2 declined led to enhanced hypoxia tolerance of sea bass with a 20% lower critical oxygen level (Fig. 1). This change in whole organism hypoxic response was accompanied by an increase in Hb-O2 affinity of blood cells in fish exposed to concurrent CO2 rises (Fig. 3). The change in Hb-O2 affinity was not accompanied by a change in blood pH (although the one hour acclimation to treatment pCO2 prior to blood sampling may have contributed to this result). This provides a potential mechanistic basis to explain improved O2crit, enabling sea bass to enhance O2 uptake during hypoxia and thus maintain normal aerobic metabolism to lower environmental O2.

The driver of increased Hb-O2 affinity in sea bass exposed to concurrent O2 decline and CO2 rise is not clear from our results. Several allosteric factors that can modulate the affinity of haemoglobin for O2 could be involved, including pH, organic phosphates and inorganic ions. Fish haemoglobin is highly sensitive to pH, which modulates Hb-O2 affinity and carrying capacity via the Bohr and Root effects31, but we found no differences in blood pH of sea bass between treatment groups. In addition, the in-vivo increase in pCO2 in fish exposed to concurrent CO2 rises during hypoxia led to an opposite response of Hb-O2 affinity than would be expected by an in-vitro rise in pCO2 which would result in pH induced Bohr/Root effects. Increased Hb-O2 affinity could result from increased intracellular pH of erythrocytes32, as acute hypoxic exposure has been shown to stimulate a β-adrenergic stimulated increase in intracellular erythrocyte pH in rainbow trout33. Alternatively, increased Hb-O2 affinity could be due to decreased red cell nucleoside triphosphates (NTPs)31,34, a known hypoxia adaptation, but this can take more than 6 days to complete35. Sea bass may also have a particularly strong β-adrenergic response and/or fast NTP response, although there is little evidence to suggest this because P50 measurements from fish exposed to hypoxia at ambient CO2 levels do not differ from P50 measurements in normoxic fish from the same population (Montgomery et al., Unpublished data). It is possible that rising CO2 during hypoxia may modulate the β-adrenergic response and/or the red cell NTP response within the time frame (~4–6 hours) of our treatments. However, direct measurements of red cell pHi and NTP content of sea bass in each treatment group would be needed to confirm this.

A third possible driver of increased Hb-O2 affinity in sea bass exposed to concurrent CO2 rise during hypoxia could be decreased erythrocyte chloride (Cl)36. Although erythrocyte Cl was not directly measured in our study, plasma HCO3 was approximately 3 mM higher in fish exposed to concurrent CO2 rises during hypoxia than fish which experienced constant ambient CO2 during hypoxia (Fig. 2D). The higher HCO3 in fish exposed to rising ambient CO2 during hypoxia is likely a result of rapid compensation for a respiratory acidosis due to rising blood pCO2. This change in plasma HCO3 is typically mirrored by a reciprocal change in plasma Cl37,38 which is likely to be followed by a similar decline in erythrocyte Cl.

Analysis of O2crit is a common measure of hypoxia tolerance in fish but concurrent CO2 increases during hypoxia have been generally unaccounted for. A recent meta-analysis by Rogers et al.39 constructed a database of O2crit research of fish (both freshwater and marine). This analysis identified two broad methods employed in O2crit measurements:

1. 1.

Closed respirometry where O2 is usually reduced by the O2 consumption of the fish (52 identified studies) or;

2. 2.

Intermittent or flow-through respirometry in which O2 is usually reduced via gassing with pure N2 or combined N2 and O2 mixes (32 identified studies).

The use of closed respirometry in the majority of studies would result in concurrent CO2 rises as O2 is depleted by fish O2 consumption. The increase in ambient CO2 during closed respirometry is well known and often used as a criticism of this respirometry technique40. In contrast, use of intermittent-flow respirometry in O2crit trials normally necessitates the reduction of O2 in the water by aeration with N2 or a mix of N2 & O2. As a result CO2 would likely decrease during the time course of hypoxia induction (as the gas mixture would contain zero CO2, rather than ~400 µatm present in atmospheric air). Such a change in CO2 during the O2crit trial would be the opposite of that seen in nature. Therefore it may be considered that closed respirometry provides conditions which give a more environmentally relevant measure of O2crit28.

As our results indicate that rising CO2 during hypoxia directly affects the ability of sea bass to maintain O2 uptake, it could be expected that the use of closed respirometry methods would result in lower measurements of O2crit than intermittent-flow methods for the same species. This effect has not been documented for species in which a direct comparison has been made – with either there being no effect of respirometry method on O2crit39,41 or higher O2crit measurements when closed respirometry is used41. However, such comparisons are complicated by differences in the rate of hypoxia induction (RHI) by different studies, which in turn will influence how much time fish have to regulate blood pH when ambient CO2 is rising. For example, Regan and Richards41 have demonstrated that the faster rates of hypoxia induction (RHI) typical of closed respirometry O2crit trials lead to higher values of O2crit (i.e. lower hypoxia tolerance) when compared with longer trials using slower RHI’s typical of the intermittent-flow method. The effect of RHI on O2crit was proposed by Regan and Richards40 as a potential explanation of the results of Snyder et al. (i.e. higher O2crit in closed respirometry compared to intermittent-flow respirometry)42. The speed of RHI during closed respirometry will also effect the speed of CO2 rise. Almost all studies using closed respirometry to measure O2crit do not report changes in CO2 over the course of measurement period. When accumulation of CO2 during a closed respirometry O2crit trial was measured by Regan and Richards, CO2 levels were ~8,000 μatm after ~90 minutes41. However, we should note that this pCO2 level was measured after the O2crit point, when anaerobic metabolism continues to produce CO2 in the absence of O2 consumption, but also metabolic acid production and excretion further drives up water pCO2 in the respirometer due to excess H+ ions titrating ambient HCO3 to CO2. Regardless, at O2 levels above O2crit the rate of CO2 onset will be faster than those used in our current study as a result of the faster RHI. Increased speed of CO2 onset in closed respirometry trials may ‘outstrip’ the ability of fish to acid-base regulate, causing an uncompensated respiratory acidosis during the time of the trial, which in turn would decrease Hb-O2 affinity via the Bohr & Root effects and potentially increase O2crit. Similarly, fish species which have reduced ability to acid-base regulate may have an increased O2crit when rising CO2 is included in trials.

Our results indicate improved hypoxia tolerance during rising CO2 in European sea bass. This contrasts with previous research investigating interactive effects of CO2 and hypoxia on O2crit of fish. Woolly sculpin, Clinocottus analis (an intertidal species that can breathe air), exposed to ~1100 μatm CO2 showed no impact on O2crit after 7 days acclimation but after 28 days had O2crit measurements ~34% higher than fish held in ambient (~400 μatm) conditions43. Higher O2crit after 28 days corresponded with higher RMR and Na+, K+, ATPase activity. This contrast in results could indicate that the beneficial effect of acute rises in CO2 associated with natural hypoxia documented in our study are potentially reversed when fish are exposed to long term constantly high CO2 associated with anthropogenic climate change. Moreover, acute changes in CO2 had no effect on O2crit of the estuarine fish species mummichog, Fundulus heteroclitus, and Norfolk spot, Leiostomus xanthurus when they were exposed to ~8,000–10,000 μatm CO2 immediately prior to an O2crit trial44. As such the effect of CO2 on O2crit will likely depend on differences in physiological responses to CO2 and O2 between species.

Simultaneously rising CO2 also shows variable impacts on non-metabolic responses to hypoxia of several species. Cycling CO2 had no effect on aquatic surface respiration (ASR), the use of the thin surface layer of water for aquatic respiration45, or survival in juvenile Menidia menidia, Fundulus majalis, Fundulus heteroclitus or Morone saxatalis exposed to short term cycles of O246. In contrast, combined hypoxia and acidification resulted in an increase in the O2 level at which Menidia menidia and Menidia beryllina first performed ASR, consistently performed ASR, exhibited LoE, and finally died21. Additionally, combined high CO2 (~2,000 μatm) and hypoxia had no effect on survival of larval Cyprinodon variegatus, an additive negative effect on larval Menidia beryllina, and a synergistic negative effect on larval Menidia menidia47. This variation in effect of CO2 on hypoxia responses could be a result of methodological differences (e.g. constant high CO2 in Dixon et al.46, cycling DO/pH in DePasquale et al.47, and concurrent CO2 rise/O2 decrease in Miller et al.21), the level of CO2 used in studies (e.g. CO2 levels used by Miller et al.21 were ~23,000 μatm which is much higher than levels likely to be commonly found in the environment during hypoxia and may have contributed to the negative effects of rising CO2 noted in the study), differences in species and life stages used (changes in physiological tolerance across life stages have been noted for thermal tolerance by Komoroske et al.48), or possibly variability in response as a result of differences in previous environmental experience49. The role of environmental variability in species sensitivities to CO2 has recently been outlined in the proposed Ocean Variability Hypothesis (OVH)50 and warrants testing on various species in the future.

Overall our results indicate that the environmentally realistic, simultaneous rises in CO2 during a hypoxic event increased the hypoxia tolerance (i.e. reduces O2crit) of European sea bass which is at least partly explained by an enhanced ability of fish to uptake O2 via increased Hb-O2 affinity. Miller et al.21 also demonstrated impacts of concurrent CO2 rise on measurements of hypoxia tolerance, although in an opposite direction to that noted in our study. As concurrent CO2 rises during hypoxia are the norm in nature, evidence that this affects physiology of organisms exposed to hypoxia highlights an important shortcoming of research to predict tolerances to hypoxia of fish. More research on this issue is needed to clarify how common this modifying effect of CO2 on the response to hypoxia is and whether such measurements in the lab are ecologically relevant. A greater understanding of this issue may allow more accurate assessments of the impacts of hypoxic events on marine fish in nature, aiding management and conservation of fish species. With specific regard to measurements of O2crit we believe future studies should include concurrent rising CO2 in the following ways:

1. 1.

Intermittent-flow respirometer studies should include increases of CO2 relevant to hypoxic events that organisms may experience,

2. 2.

Closed respirometry studies should report the start and end CO2 levels in the respirometer.

In addition, both methods should aim to create an environmentally relevant rate of hypoxia induction/CO2 increase for the species studied, and consistently report CO2 levels measured. By incorporating these recommendations we believe that future studies of O2crit will give more representative estimations of species hypoxia tolerance.

## Materials and Methods

### Fish collection and husbandry

We collected juvenile sea bass from estuaries and coastal lagoons on the south Dorset coast and Isle of Wight in June 2017 (Marine Management Organisation permit #030/17 & Natural England permit #OLD1002654). Prior to experimentation, these fish were held in the University of Exeter’s Aquatic Resource Centre in an aerated recirculating aquaculture system and fed a commercial pellet at a ration of ~1–2% body weight per day three times a week (for system water chemistry see Table 1). All fish were starved for a minimum of 72 hours prior to the start of all measurements to ensure their metabolism was not affected by digestion (i.e. specific dynamic action24). All experimental procedures were carried out under home office licence P88687E07 and approved by the University of Exeter’s Animal Welfare and Ethical Review Board.

### Measuring hypoxia tolerance

We determined oxygen consumption rates ($${\dot{{\rm{M}}}{\rm{O}}}_{2}$$) of sea bass using an intermittent-flow respirometer system. The respirometer system set up followed recommendations set out by Svendsen et al.51. Briefly, the system comprised of a sealed 4.515 L respirometer chamber connected to a recirculating loop, including an in-line recirculating pump (Eheim universal 600, Deizisau, Germany), and a measurement chamber into which a temperature-compensated fibre optic oxygen optode (Firesting O2 oxygen meter, Pyroscience GmBH, Germany) was placed. Oxygen optodes were calibrated in water at the start of experiments at 100% air saturation and 0% air saturation according to manual instructions. Respirometry was conducted in a semi-closed system consisting of three 100 L experimental tanks fed by a 100 L sump, with overflowing water from the experimental tanks recirculating back to the sump. A second pump was used to periodically flush the respirometer system with water from the surrounding tank. This pump was controlled by an automated computer program (AquaResp 3, AquaResp®) to intermittently flush the respirometer. Five respirometer chambers were distributed between the three experimental tanks (maximum of 2 chambers per experimental 100 L tank). The sump was temperature controlled (18.27 ± 0.02 °C, mean ± S.E.) using a heater/chiller unit (Grant TX150 R2, Grant Instruments, Cambridge, UK) attached to a temperature exchange coil. Together these tanks formed a 400 L system with the same temperature, oxygen and water chemistry parameters for all respirometers.

Individual sea bass (average mass = 131.2 ± 7.5 g), chosen at random, were placed inside the respirometers and allowed an overnight recovery period, for a minimum of 13 hours, before O2crit tests began. While sea bass were in the respirometers measurements of $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ were conducted every 10 minutes, including a flush period of 300 s, a wait period of 60 s and a measurement period of 240 s. During the wait and measurement period the chamber was sealed by switching off the flush pump and the decline in dissolved O2 within the chamber was continuously measured by the fibre optic O2 electrode.

Following the overnight recovery period a standard O2crit test was conducted. Oxygen levels in the respirometer system were reduced from ~100% air saturation to ~15% air saturation over the course of 6 hours (decline in O2 was ~20% air saturation per hour between ~100% air saturation and ~40% air saturation and ~10% air saturation per hour from ~40% air saturation until the end of trials). Oxygen was regulated by gassing the sump and experimental tanks with a mix of N2 and O2 (G400 Gas mixing system, Qubit Biology Inc.) at a rate of 10 L min−1 following a pre-set automated protocol (Flowvision, Alicat software). Levels of CO2 within the system were controlled under one of two treatments (with 8 fish exposed to Treatment 1 and 8 separate fish exposed Treatment 2):

• Treatment 1 (Constant CO2) - ambient levels of CO2 were maintained by including 0.04% CO2 as part of the gas mix delivered to the respirometer system.

• Treatment 2 (Rising CO2) - the proportion of CO2 in the gas mix was gradually increased as O2 was decreased. This increase in CO2 was designed to reflect environmentally realistic increases in CO2 predicted as a result of depletion of O2 by bacterial respiration (assumed respiratory quotient of 1)52, using the seawater carbonate chemistry calculator CO2sys (see supplementary material for predictions of increased CO2 during hypoxia).

Water chemistry of treatment 2 was monitored once per hour by measuring pHNBS, temperature and salinity as well as taking a 12 mL water sample to measure total CO2 (TCO2)/Dissolved Inorganic Carbon (DIC). Water chemistry of treatment 1 was monitored at the start and end of the treatment to ensure no change in water pCO2 occurred over the time course of the O2crit trial. Seawater DIC analysis was conducted using a custom built system described in detail by Lewis et al.53. These four parameters were then input into the seawater carbon calculator programme, CO2SYS to calculate pCO2 based on the NBS pH scale, equilibration constants from Mehrbach et al. refitted by Dickson and Millero, and KSO4 dissociation constants from Dickson. The reciprocal changes in O2 and CO2 during O2crit tests for each treatment are illustrated in Fig. 4. O2crit tests were stopped once a minimum of 3 $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ measurements showed a transition from an oxy-regulating to oxy-conforming state for each fish or fish showed a large drop in $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ and signs of distress in the respirometer. No fish exhibited LoE during trials. Following completion of O2crit trials experimental tanks were aerated with ambient air to swiftly restore O2 and CO2 levels.

Sea bass were left to recover in respirometers, for a minimum of 1 hour post-trial, until O2 levels reached ~21 kPa O2 (~100% air saturation). Fish were then removed from respirometers and background respiration was measured for a minimum of 1 hour (6 measurement cycles) for all respirometers immediately post trial.

#### Oxygen consumption rate ($${\dot{{\rm{M}}}{\rm{O}}}_{2}$$) analysis

Following each 240 s measurement period $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ was automatically calculated by the AquaResp3 software. A linear regression was fitted to the O2 versus time data for each measurement period. The slope of this regression (s, kPa O2 h−1) was then used to calculate $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ (mg O2 kg−1 h−1) using the equation outlined by Svendsen et al.51:

$$\dot{{\rm{M}}}{{\rm{O}}}_{2}=s{V}_{resp}\,\alpha {m}^{-1}$$

where Vresp is the respirometer volume minus the volume of the fish (L), α is the solubility of O2 in water (mgO2 L−1 kPa−1) for the relevant salinity and temperature, and m is the mass of the fish (kg). Calculations of $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ where s had a R2 of <0.98 were removed from subsequent analysis. For the purpose of establishing O2crit values from a plot of $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ versus ambient O2 level, the oxygen saturation of each measurement period was defined as the average of the dissolved O2 measurement over the measurement period. The average background respiration, over the 1 hour post-trial measuring period, for each respirometer (average background respiration was <2% of fish $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$) was subtracted from $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ measurements to correct fish $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ for background respiration.

We calculated the standard metabolic rate (SMR) in R v3.5.354 using function calcSMR in package fish $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$24. All $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ values from the overnight recovery period and beginning of the O2crit trial in which average dissolved O2 saturation was >80% air saturation were included for SMR calculations. This resulted in approximately 14–16 hours of $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ data to calculate SMR from for each fish. We estimated SMR for all fish using the mean of the lowest 10 $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ measurements during the overnight period. Although Chabot et al.24 recommend use of the mean of the lowest normal distribution (MLND) or quantile (q = 0.2) methods to calculate SMR we chose the mean of the 10 lowest values as it produced a value of SMR that most accurately matched the consistent low $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ measurements of oxy-regulating fish at values of pO2 above the O2crit_point. Additionally, for each fish the coefficient of variation in the mean lowest normal distribution (MLND) was assessed using a ROUT test (Q = 1%) to check whether variation in low $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ measurements was consistent between fish (to account for potential differences in activity in the respirometer prior to O2crit trials). The ROUT test removed one fish from the rising CO2 treatment which displayed a CoV of 34.6. Mean CoV of the remaining 15 fish was 11.85 ± 1.26 (±S.E.M).

We then used function calcO2crit from the package to calculate the O2crit for each individual fish, using the estimated SMR of each individual, as detailed in the supplementary material of Claireaux & Chabot55. This function identifies the portion of the O2crit test where metabolic rate data follows an oxygen conforming relationship and fits a linear regression line through this data, O2crit is then calculated as the oxygen level at which this regression line crosses the calculated SMR of the individual fish. Plots of $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ against O2 during O2crit trials showing calculated O2crit for each individual can be seen in the supplementary material (Supplementary Material Figs 116). Calculations were conducted using a gap limit of 0.83 kPa O2 (4% air saturation) and a maximum number of 7 $${\dot{{\rm{M}}}{\rm{O}}}_{2}$$ points to fit the regression line through the oxygen conforming component of the data used to estimate O2crit.

### Measuring blood chemistry

Following O2crit trials sea bass were moved to individual 7 L chambers, which were aerated and supplied with seawater from the aquarium re-circulating system at a rate of about 4 L min−1. After an overnight acclimation period we then exposed the fish to a decrease in O2 levels to ~6.4 kPa O2 (30% air saturation) over a period of 4 hours (equivalent to the rate of decrease in O2 used in the previous O2crit tests). We chose to blood sample fish at an O2 level above O2crit to ensure that anaerobic metabolism did not influence blood chemistry. This was combined with the same CO2 regime each fish experienced during the O2crit test (i.e. either constant or reciprocally rising CO2). Once an O2 level of ~6.4 kPa O2 (30% air saturation) was achieved the fish were allowed to acclimate for 1 hour before being individually anaesthetised in situ using a dose of 100 mgL−1 of benzocaine. Once fish were judged to be sufficiently anaesthetised (cessation of gill ventilation and lack of response to a pinch of the anal fin) they were immediately transferred within 5 seconds to a gill irrigation table (with the same pO2 and pCO2 levels of the respective treatment), where anaesthesia was maintained with a dose of 37.5 mgL−1 of benzocaine. Gill ventilation was artificially maintained by a micro-pump, so that the operculum were just open and exhalant water flow could just be visualised. Once a stable gill water flow was established blood was sampled by caudal vessel puncture using a 1 ml heparinised syringe. This method has been demonstrated to obtain accurate measurements of blood chemistry parameters comparable to those achieved using cannulation (Davison & Wilson, University of Exeter, personal communication). At the time of blood sampling water pCO2 was 656 ± 44 μatm (mean ± S.E.) for fish in the constant ambient CO2 regime and 1763 ± 43 μatm (mean ± S.E.) for fish in the progressively rising CO2 regime. Water pO2 was 8.1 ± 0.2 kPa (mean ± S.E., ~38% air saturation) for fish in the constant ambient CO2 regime and 8.7 ± 0.2 kPa (mean ± S.E., ~41% air saturation) for fish in the progressively rising CO2 regime. Following blood sampling fish were transferred to seawater isolation tanks containing ~20.8 kPa O2 (100% air-saturated) to recover from the anaesthetic. They were then monitored over a period of 24 hours before we returned them to their original holding tanks.

Immediately after sampling, whole blood pO2 was measured at 18 °C in a temperature-controlled system (Strathkelvin 1302 electrode and 781 meter; Strathkelvin Instruments Ltd, Glasgow, UK). We measured extracellular pH on 30 µL of whole blood using an Accumet Micro pH electrode and Hanna HI8314 pH meter at 18 °C calibrated to pHNBS 7.04 and 9.21 specific buffers. Three 75 µL micro capillary tubes were then filled with whole blood and sealed with Critoseal capillary tube sealant (Fisher) and paraffin oil and centrifuged for 2 minutes at 10,000 rpm. Haematocrit was measured using a Hawksley micro-haematocrit reader. Plasma was extracted from these tubes for analysis of TCO2 using a Mettler Toledo 965D carbon dioxide analyser. Plasma pCO2 and HCO3 were then calculated from TCO2, temperature and blood pH using the Henderson-Hasselbalch equation with values for solubility and pK1app based on Boutilier et al.56,57. Haemoglobin content of the blood was also assessed by the cyanmethemoglobin method (using Drabkin’s reagent, Sigma). Half the remaining whole blood was then centrifuged at 10,000 rpm for 2 minutes at 4 °C. The resulting plasma was separated and snap frozen in liquid nitrogen and stored at −80 °C before later being used to measure plasma glucose and lactate using a YSI 2900D Biochemistry Analyzer (Xylem Analytics, UK). All measurements or storage of blood occurred within 10 minutes of blood sampling.

### Measuring Hb-oxygen binding

We measured Hb-O2 affinity using a Blood Oxygen Binding System (BOBS, Loligo systems), detailed in Oellermann et al.58. A sample of the same whole blood used for blood chemistry measurements was diluted at a ratio of 1:4 in its own plasma. 1 μL of this blood was then used for measurements. The BOBS exposed this blood sample to gas mixes with a progressive increase in O2 whilst measuring absorbance of light across a spectrum ranging from 200 to 800 nm. For each individual fish the gas mix that blood was exposed to matched the calculated pCO2 of the blood sample. The change in absorption of light at a wavelength of 435 nm was used to assess changes in oxygenation of Hb, as previously used by Verhille & Farrell59. Background changes in absorption of the blood sample were corrected using the isosbestic wavelength of 390 nm59. Following measurements the BOBS calculated the oxygen equilibrium curve of the sample using Hill’s formula before estimating P5058.

### Statistical analysis

We conducted all statistical analysis in R v3.5.354. There was no significant difference in mass of fish between the treatment groups (One-Way ANOVA, F1,13 = 2.821, p = 0.117) or of estimated SMR of fish between treatment groups (Unpaired t-test, t = 0.8455, d.f. = 13, p = 0.413). There was, however, a significant correlation between estimated SMR and calculated O2crit for all fish (Pearson’s correlation, df = 13, t = 3.32, R = 0.68, p = 0.0056). As such the effect of CO2 treatment on O2crit was assessed using an ANCOVA with SMR as a covariate. All other analyses were conducted using general linear modelling (GLM). All values in the text are reported as mean ± standard error (S.E.).

## Data availability

Data is available via the University of Exeter’s online repository at: https://doi.org/10.24378/exe.1523.

## References

1. 1.

Diaz, R. J. Overview of Hypoxia around the World. J. Environ. Qual. 30, 275–281 (2001).

2. 2.

Breitburg, D. et al. Declining oxygen in the global ocean and coastal waters. Science (80-). 359 (2018).

3. 3.

Diaz, R. J. & Rosenberg, R. Spreading Dead Zones and Consequences for Marine Ecosystems. Science (80-). 321, 926 (2008).

4. 4.

Townhill, B. L. et al. Consequences of climate-induced low oxygen conditions for commercially important fish. Mar. Ecol. Prog. Ser. 580, 191–204 (2017).

5. 5.

Robinson, C. Microbial Respiration, the Engine of Ocean Deoxygenation. Frontiers in Marine Science. 5, 533 (2019).

6. 6.

Rabalais, N. N., Turner, R. E., Díaz, R. J. & Justić, D. Global change and eutrophication of coastal waters. ICES J. Mar. Sci. 66, 1528–1537 (2009).

7. 7.

Sunda, W. G. & Cai, W.-J. Eutrophication Induced CO2-Acidification of Subsurface Coastal Waters: Interactive Effects of Temperature, Salinity, and Atmospheric PCO2. Environ. Sci. Technol. 46, 10651–10659 (2012).

8. 8.

Wallace, R. B., Baumann, H., Grear, J. S., Aller, R. C. & Gobler, C. J. Coastal ocean acidification: The other eutrophication problem. Estuar. Coast. Shelf Sci. 148, 1–13 (2014).

9. 9.

Gobler, C. J. & Baumann, H. Hypoxia and acidification in ocean ecosystems: coupled dynamics and effects on marine life. Biol. Lett. 12 (2016).

10. 10.

Altieri, A. H. & Gedan, K. B. Climate change and dead zones. Glob. Chang. Biol. 21, 1395–1406 (2015).

11. 11.

Oschlies, A., Brandt, P., Stramma, L. & Schmidtko, S. Drivers and mechanisms of ocean deoxygenation. Nat. Geosci. 11, 467–473 (2018).

12. 12.

Oschlies, A., Schulz, K. G., Riebesell, U. & Schmittner, A. Simulated 21st century’s increase in oceanic suboxia by CO2-enhanced biotic carbon export. Global Biogeochem. Cycles 22 (2008).

13. 13.

Doney, S. C., Fabry, V. J., Feely, R. A. & Kleypas, J. A. Ocean Acidification: The Other CO2 Problem. Ann. Rev. Mar. Sci. 1, 169–192 (2009).

14. 14.

Melzner, F. et al. Future ocean acidification will be amplified by hypoxia in coastal habitats. Mar. Biol. 160, 1875–1888 (2013).

15. 15.

Brandt, S. B., Gerken, M., Hartman, K. J. & Demers, E. Effects of hypoxia on food consumption and growth of juvenile striped bass (Morone saxatilis). J. Exp. Mar. Bio. Ecol. 381(Suppl), S143–S149 (2009).

16. 16.

Almeida, L. Z., Guffey, S. C., Sepúlveda, M. S. & Höök, T. O. Behavioral and physiological responses of yellow perch (Perca flavescens) to moderate hypoxia. Comp. Biochem. Physiol. Part A Mol. Integr. Physiol. 209, 47–55 (2017).

17. 17.

Chabot, D. & Dutil, J. D. Reduced growth of Atlantic cod in non-lethal hypoxic conditions. J. Fish Biol. 55, 472–491 (1999).

18. 18.

Domenici, P., Steffensen, J. F. & Batty, R. S. The effect of progressive hypoxia on swimming activity and schooling in Atlantic herring. J. Fish Biol. 57, 1526–1538 (2000).

19. 19.

Gobler, C. J., DePasquale, E. L., Griffith, A. W. & Baumann, H. Hypoxia and Acidification Have Additive and Synergistic Negative Effects on the Growth, Survival, and Metamorphosis of Early Life Stage Bivalves. PLoS One 9, e83648 (2014).

20. 20.

Steckbauer, A. et al. Synergistic effects of hypoxia and increasing CO2 on benthic invertebrates of the central Chilean coast. Front. Mar. Sci. 2, 49 (2015).

21. 21.

Miller, S. H., Breitburg, D. L., Burrell, R. B. & Keppel, A. G. Acidification increases sensitivity to hypoxia in important forage fishes. Mar. Ecol. Prog. Ser. 549, 1–8 (2016).

22. 22.

Nikinmaa, M. Oxygen and carbon dioxide transport in vertebrate erythrocytes: an evolutionary change in the role of membrane transport. J. Exp. Biol. 200, 369–380 (1997).

23. 23.

Rummer, J. L. & Brauner, C. J. Root effect haemoglobins in fish may greatly enhance general oxygen delivery relative to other vertebrates. PLoS One 10, e0139477 (2015).

24. 24.

Chabot, D., Steffensen, J. F. & Farrell, A. P. The determination of standard metabolic rate in fishes. J. Fish Biol. 88, 81–121 (2016).

25. 25.

Perry, S. F., Jonz, M. G. & Gilmour, K. M. Chapter 5 Oxygen Sensing And The Hypoxic Ventilatory Response. In Fish Physiology (eds J. G. Richards, A. P. F. and C. J. B.) Volume 27, 193–253 (Academic Press, 2009).

26. 26.

Wells, R. M. G. Chapter 6 Blood‐Gas Transport and Hemoglobin Function: Adaptations for Functional and Environmental Hypoxia. In Fish Physiology (eds Jeffrey G. Richards, A. P. F. and C. J. B.) Volume 27, 255–299 (Academic Press, 2009).

27. 27.

Gamperl, A. K. & Driedzic, W. R. Chapter 7 Cardiovascular Function and Cardiac Metabolism. In Hypoxia (eds Richards, J. G., Farrell, A. P. & Brauner, C. J. B. T.-F. P.) 27, 301–360 (Academic Press, 2009).

28. 28.

Wood, C. M. The fallacy of the Pcrit – are there more useful alternatives? J. Exp. Biol. 221, jeb163717 (2018).

29. 29.

Speers-Roesch, B., Mandic, M., Groom, D. J. E. & Richards, J. G. Critical oxygen tensions as predictors of hypoxia tolerance and tissue metabolic responses during hypoxia exposure in fishes. J. Exp. Mar. Bio. Ecol. 449, 239–249 (2013).

30. 30.

Regan, M. D. et al. Don’t throw the fish out with the respirometry water. J. Exp. Biol. 222, jeb200253 (2019).

31. 31.

Pelster, B. & Weber, R. E. The Physiology of the Root Effect. In Advances in Comparative and Environmental Physiology 51–77, https://doi.org/10.1007/978-3-642-75900-0_2LB-Pelster1991 (Springer Berlin Heidelberg, 1991).

32. 32.

Nikinmaa, M. Adrenergic regulation of haemoglobin oxygen affinity in rainbow trout red cells. J. Comp. Physiol. 152, 67–72 (1983).

33. 33.

Tetens, V. & Christensen, N. J. Beta-adrenergic control of blood oxygen affinity in acutely hypoxia exposed rainbow trout. J. Comp. Physiol. B 157, 667–675 (1987).

34. 34.

Weber, R. E. & Lykkeboe, G. Respiratory adaptations in carp blood influences of hypoxia, red cell organic phosphates, divalent cations and CO2 on hemoglobin-oxygen affinity. J. Comp. Physiol. 128, 127–137 (1978).

35. 35.

Soivio, A., Nikinmaa, M. & Westman, K. The blood oxygen binding properties of hypoxicSalmo gairdneri. J. Comp. Physiol. 136, 83–87 (1980).

36. 36.

Mairbäurl, H. & Weber, R. E. Oxygen Transport by Hemoglobin. Comprehensive Physiology, https://doi.org/10.1002/cphy.c080113 (2012).

37. 37.

Esbaugh, A. J., Heuer, R. & Grosell, M. Impacts of ocean acidification on respiratory gas exchange and acid-base balance in a marine teleost, Opsanus beta. J. Comp. Physiol. B. 182, 921–934 (2012).

38. 38.

Perry, S. F., Goss, G. G. & Laurent, P. The interrelationships between gill chloride cell morphology and ionic uptake in four freshwater teleosts. Can. J. Zool. 70, 1775–1786 (1992).

39. 39.

Rogers, N. J., Urbina, M. A., Reardon, E. E., McKenzie, D. J. & Wilson, R. W. A new analysis of hypoxia tolerance in fishes using a database of critical oxygen level (Pcrit). Conserv. Physiol. 4 (2016).

40. 40.

Steffensen, J. F. Some errors in respirometry of aquatic breathers: How to avoid and correct for them. Fish Physiol. Biochem. 6, 49–59 (1989).

41. 41.

Regan, M. D. & Richards, J. G. Rates of hypoxia induction alter mechanisms of O2 tension of goldfish. J. Exp. Biol. 220, 2536 LP–2544 (2017).

42. 42.

Snyder, S. et al. Effect of closed v. intermittent-flow respirometry on hypoxia tolerance in the shiner perch Cymatogaster aggregata. J. Fish Biol. 88, 252–264 (2016).

43. 43.

Hancock, J. R. & Place, S. P. Impact of ocean acidification on the hypoxia tolerance of the woolly sculpin, Clinocottus analis. Conserv. Physiol. 4, cow040–cow040 (2016).

44. 44.

Cochran, R. E. & Burnett, L. E. Respiratory responses of the salt marsh animals, Fundulus heteroclitus, Leiostomus xanthurus, and Palaemonetes pugio to environmental hypoxia and hypercapnia and to the organophosphate pesticide, azinphosmethyl. J. Exp. Mar. Bio. Ecol. 195, 125–144 (1996).

45. 45.

Kramer, D. L. & McClure, M. Aquatic surface respiration, a widespread adaptation to hypoxia in tropical freshwater fishes. Environ. Biol. Fishes 7, 47–55 (1982).

46. 46.

Dixon, R. L., Grecay, P. A. & Targett, T. E. Responses of juvenile Atlantic silverside, striped killifish, mummichog, and striped bass to acute hypoxia and acidification: Aquatic surface respiration and survival. J. Exp. Mar. Bio. Ecol. 493, 20–30 (2017).

47. 47.

DePasquale, E., Baumann, H. & Gobler, C. J. Vulnerability of early life stage Northwest Atlantic forage fish to ocean acidification and low oxygen. Mar. Ecol. Prog. Ser. 523, 145–156 (2015).

48. 48.

Komoroske, L. M. et al. Ontogeny influences sensitivity to climate change stressors in an endangered fish. Conserv. Physiol. 2 (2014).

49. 49.

Murray, C. S., Malvezzi, A., Gobler, C. J. & Baumann, H. Offspring sensitivity to ocean acidification changes seasonally in a coastal marine fish. Mar. Ecol. Prog. Ser. 504, 1–11 (2014).

50. 50.

Baumann, H. Experimental assessments of marine species sensitivities to ocean acidification and co-stressors: how far have we come? Can. J. Zool. 399–408, https://doi.org/10.1139/cjz-2018-0198 (2019).

51. 51.

Svendsen, M. B. S., Bushnell, P. G. & Steffensen, J. F. Design and setup of intermittent-flow respirometry system for aquatic organisms. J. Fish Biol. 88, 26–50 (2016).

52. 52.

Del Giorgio, P. A. & Duarte, C. M. Respiration in the open ocean. Nature 420, 379–384 (2002).

53. 53.

Lewis, C., Clemow, K. & Holt, W. V. Metal contamination increases the sensitivity of larvae but not gametes to ocean acidification in the polychaete Pomatoceros lamarckii (Quatrefages). Mar. Biol. 160, 2089–2101 (2013).

54. 54.

R Core Team. R: A language and environment for statistical computing (2019).

55. 55.

Claireaux, G. & Chabot, D. Responses by fishes to environmental hypoxia: integration through Fry’s concept of aerobic metabolic scope. J. Fish Biol. 88, 232–251 (2016).

56. 56.

Boutilier, R. G., Heming, T. A. & Iwama, G. K. Appendix-physicochemical parameters for use in fish respiratory physiology. Fish Physiol. 10, 403–430 (1984).

57. 57.

Boutilier, R. G., Iwama, G. K., Heming, T. A. & Randall, D. J. The apparent pK of carbonic acid in rainbow trout blood plasma between 5 and 15 °C. Respir. Physiol. 61, 237–254 (2016).

58. 58.

Oellermann, M., Pörtner, H.-O. & Mark, F. C. Simultaneous high-resolution pH and spectrophotometric recordings of oxygen binding in blood microvolumes. J. Exp. Biol. 217, 1430 LP–1436 (2014).

59. 59.

Verhille, C. & Farrell, A. P. The in vitro blood-O2 affinity of triploid rainbow trout Oncorhynchus mykiss at different temperatures and CO2 tensions. J. Fish Biol. 81, 1124–1132 (2012).

## Acknowledgements

This work was supported by a NERC GW4+ Doctoral Training Partnership studentship from the Natural Environment Research Council [NE/L002434/1] with additional funding from CASE partner, The Centre of Fisheries and Aquaculture Science (Cefas) to D.W.M., and from the Biotechnology and Biological Sciences Research Council (BB/D005108/1 and BB/J00913X/1) and NERC (NE/H017402/1) to R.W.W. The authors would like to thank the Marine Management Organisation and Natural England for granting permits to collect wild sea bass for use in this study, Simon Pengelly and the Southern Inshore Fisheries and Conservation Authority for assistance with fish collections. We would also like to thank the aquarium staff, particularly Steven Cooper, Alice Walpole and Rebecca Turner, of the Aquatic Resource Centre at Exeter University for assistance with fish husbandry and system water chemistry, and finally Will Davison, Dr. Cosima Porteus and Harriet Goodrich for assistance with blood chemistry measurements.

## Author information

D.W.M. contributed to study concept and design, conducted all data collection, analysed all data and drafted the manuscript. S.D.S. contributed to study concept and design, and revisions of the manuscript. G.H.E. and S.N.R.B. contributed to study concept and design, data analysis and revisions of the manuscript. R.W.W. was responsible for study supervision, contributed to study concept and design, assisted with data collection, helped with interpretation of data, and contributed to revisions of the manuscript.

Correspondence to Daniel W. Montgomery or Rod W. Wilson.

## Ethics declarations

### Competing interests

The authors declare no competing interests.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## Rights and permissions

Reprints and Permissions