Aquatic acidification: a mechanism underpinning maintained oxygen transport and performance in fish experiencing elevated carbon dioxide conditions

The concentration of atmospheric carbon dioxide (CO₂) has been increasing at unprecedented rates since pre-industrial times to approximately ∼400 ppm in 2015 (Dlugokencky and Tans, 2015), the highest level in the atmosphere in 800,000 years (Lüthi et al., 2008). Since the Industrial Revolution, approximately 30% of the excess CO₂ in the atmosphere has been absorbed by the oceans (Sabine et al., 2004), which, owing to their expansive surface area, experience a parallel increase in the partial pressure of CO₂ (P_CO2) with the atmosphere (Doney, 2010). If the current trajectory is maintained, increases in atmospheric CO₂ levels are predicted to exceed 900 ppm by 2100 (Meinshausen et al., 2011). Increasing P_CO2 of marine systems is often accompanied by a decrease in pH of the water, termed ocean acidification (OA) (Doney, 2010). Although it is known that ocean P_CO2 is increasing and pH is decreasing, reliable in situ records of these parameters are lacking (Degrandpre et al., 1995; Hofmann and Todgham, 2010). Moreover, if we continue along the current trajectory predicted by the Intergovernmental Panel on Climate Change IS92a ‘business as usual’ emissions scenario for the year 2100, the ocean pH is predicted to decrease by 0.4 pH units by 2100 (Meinshausen et al., 2011; Pörtner et al., 2014). Predictions for future conditions in marine systems are affected by the composition of seawater, which, when compared with all other aquatic systems, is extremely constant, ranging by ±7% in salinity (Berner and Berner, 2012) and by less than 1% with respect to ion ratios (Millero et al., 2006). Unlike the many predictions and models depicting future ocean scenarios, those for freshwater systems are less clear.

Freshwater chemistry is affected by a multitude of factors, and the P_CO2 varies greatly within and between different systems and locations. Cole et al. (2007) found that the current P_CO2 range in lakes varies from 3.1-fold below to 16-fold above current atmospheric levels, with a mean of ∼1000 μatm. Larger lakes that might be deemed similar to marine systems in terms of the surface area in contact with the atmosphere tend to remain close to atmospheric levels (Cole et al., 2007). However, smaller systems (e.g. rivers and streams) can vary depending on surrounding terrestrial ecosystems, precipitation, surface area, residence time (see Glossary), etc. (Butman and Raymond, 2011; Kokic et al., 2015). Raymond et al. (2013) found that, in over 6700 stream and river systems assessed at a global scale, 95% had a median P_CO2 level higher than current atmospheric levels (∼400 ppm). Given that freshwater systems are so variable and affected by many different environmental and anthropogenic factors, it is difficult to predict how the rising atmospheric CO₂ will influence future freshwater systems. A case study on the Laurentian Great Lakes in North America concluded that larger systems would experience a future increase in P_CO2 similar to that expected for the oceans (Phillips et al., 2015). However, future trends are difficult to predict in smaller systems. What is clear, however, is that, along with increases in atmospheric CO₂, variations and/or increases in P_CO2 will continue to occur in all aquatic systems (Hasler et al., 2016).

Although both long-term data on P_CO2 and pH for marine systems and predictions for freshwater systems are lacking, elevated P_CO2 (hypercapnia; see Glossary) is expected to have detrimental impacts on an array of aquatic organisms, highlighting the need for more research and reliable measurements and/or predictions. Some observed detrimental effects of aquatic acidification (AA; see Glossary) upon both freshwater and marine calcifying molluscs include: reductions in calcification rates, growth, reproduction and immune function, as well as alterations in ion-exchange mechanisms and behaviour (Bibby et al., 2008; Hannan et al., 2016; Michaelidis et al., 2007; Waller et al., 2017; Watson et al., 2013). The tissues and fluids of many aquatic organisms (i.e. molluscs and fishes) become acidified when these organisms are exposed to elevated P_CO2 in their water because of the rapid diffusion of uncharged CO₂ molecules across cell membranes (Ishimatsu et al., 2005). Marine molluscs represent one of the most studied taxa with respect to AA, and molluscs are assumed to be intolerant to elevated P_CO2 when compared with fishes. Freshwater and marine fishes can physiologically compensate for an extreme acidosis (>10,000 µatm) within hours to days (Fig. 1), which has led many to suggest that fish are resilient to elevated environmental P_CO (Heuer and Grosell, 2014). However, negative impacts have been consistently observed in marine fishes at the level of behaviour, even at relatively low levels of P_CO2 (Briffa et al., 2012; Heuer and Grosell, 2014; Munday et al., 2012). Additionally, although far less studied, there is also the potential for freshwater fishes to experience behavioural alterations due to elevated P_CO2 (Tix et al., 2017). These behavioural impairments demonstrate that, despite fish being efficient acid–base regulators, they may not be as tolerant to an acidosis as previously predicted. Indeed, the capacity for acid–base regulation has recently been linked to the observed behavioural impairments related to an ion imbalance at the GABA_A receptor (Fig. 1; also see Heuer et al., 2016; Nilsson et al., 2012).

Behaviour is not the only potential negative impact fishes may incur as a result of acidification – an additional stress could come from the energy requirements of acid–base regulation (Fig. 1), which is known to be metabolically costly, even under steady-state conditions (Baker and Brauner, 2012). However, even though there is a prediction for increased energy demand with acid–base regulation under acidification conditions, various fish species have exhibited an array of metabolic responses to elevated P_CO2, ranging from a decrease (Munday et al., 2009) to increases (Couturier et al., 2013; Rummer et al., 2013a) in metabolic performance. Additionally, the responses at the level of aerobic scope (AS; see Glossary) – the total aerobic energy available above basic maintenance costs – to elevated P_CO2 have been varied as well. One explanation for this variability may arise upon closer examination of model teleost fishes.

Studies on model teleost fishes, such as the salmonids, have identified a mechanism responsible for enhanced tissue oxygen (O₂) delivery when fish are exposed to elevated P_CO2 (Alderman et al., 2016; Rummer and Brauner, 2011; Rummer et al., 2013b). Most teleost fishes possess extremely pH-sensitive haemoglobins (Hb; see Glossary), where reductions in pH not only reduce Hb–O₂ affinity (Bohr effect) but also greatly reduce Hb–O₂ carrying capacity (Root effect). In the presence of an acidosis, the extracellular pH (pH_e) decreases initially (Baker et al., 2009), and many fish species release catecholamines (e.g. adrenaline, noradrenaline) into the circulation to activate sodium/proton (Na⁺/H⁺) exchange (βNHE) on the surface of the red blood cells (RBCs) to remove H⁺ and elevate intracellular pH (Rummer et al., 2013b) (Fig. 1). Elevated RBC pH increases Hb–O₂ affinity, and therefore O₂ uptake at the gill and subsequent tissue O₂ delivery is safeguarded. To enhance O₂ delivery at targeted locations (e.g. red muscle, heart, etc.), carbonic anhydrase (CA; see Glossary) that is plasma-accessible in those select locations short-circuits this protective mechanism. The result is a re-acidification of the RBC, decreased Hb–O₂ affinity, and O₂ release to the tissue (Rummer et al., 2013b) (Fig. 1). In salmonids, this mechanism might also be operating under a very mild acidosis as well, one in which catecholamines are not released, perhaps through a more general ‘housekeeping’ (see Glossary) NHE (Rummer et al., 2013b). Regardless of the mechanism, the link between enhanced O₂ delivery and enhanced performance has not yet been made. Additionally, it is unclear whether this mechanism for enhanced O₂ delivery via Root effect Hbs (see Glossary) and plasma-accessible CA exists in more derived teleosts (i.e. those that appeared more recently on the evolutionary time scale; e.g. coral reef fishes) and whether it can be initiated under a low-level acidosis, such as what has been used in OA studies (∼850 to 1000 µatm), where maintained or even increased AS has been documented (reviewed in Heuer and Grosell, 2014). Thus, if enhancing O₂ delivery to tissues during elevated P_CO2 translates to maintained or increased AS across a wide range of fishes, there could be a mechanistic explanation to what has been documented in the literature, which could also aid in predicting how fishes will respond to AA into the future.

Bearing the above in mind, our aim here is to highlight potential mechanisms for changes in O₂ transport that could underpin changes in performance in fish in response to elevations in P_CO2. Past reviews on O₂ transport have focused on OA without taking freshwater systems into account, and so both marine oceanic and freshwater fishes, and their respective findings to date, will be discussed here. Furthermore, although elasmobranchs (see Glossary) are not known to possess the Root effect Hbs that are unique to teleosts, some elasmobranch species have been found to be unaffected at AA-relevant P_CO2 levels in terms of AS and metabolic performance; therefore, potential mechanisms for maintaining O₂ transport in elasmobranchs under elevated P_CO2 will also be discussed. For this Review, we have focused on experiments that manipulated water P_CO2 and measured the metabolic responses of fishes. We have reviewed existing studies for exposure periods, types of ecosystems, families of fishes, fish age, CO₂ levels and consequences for metabolic performance.

Our aim was to identify peer-reviewed articles that observed metabolic responses, specifically responses at the level of oxygen uptake and AS, in fishes exposed to elevated P_CO2. It was known a priori that there exists a range of grades of AS in response to elevated P_CO2; therefore, a literature review using techniques typical of a systematic review was completed (Haddaway et al., 2015a). The search strategy used included a Boolean string in Thomson Reuters Web of Science: (‘fish’ OR ‘teleost’) AND (‘oxygen consumption’ OR ‘aerobic scope’ OR ‘MO₂’) AND (‘hypercapnia’ OR ‘low pH’ OR ‘elevated CO₂’ OR ‘ocean acidification’ OR ‘pH’ OR ‘CO₂’ OR ‘carbon dioxide’); a search of the terms (‘aerobic scope’ AND ‘fish’ AND ‘carbon dioxide’) and (‘oxygen consumption’ AND ‘fish’ AND ‘carbon dioxide’) in Google Scholar was also made and the first 80 pages were reviewed for appropriate papers (Haddaway et al., 2015b).

All studies that investigated the effect of elevated P_CO2 on oxygen uptake in fishes were accessed and reviewed. Studies that were immediately excluded based on title alone were those that were conducted in terrestrial environments or on non-fish species, unavailable, syntheses/reviews, or not related to CO₂. Studies that did not specifically investigate the effects of elevated water P_CO2 or were not peer reviewed (e.g. academic theses) were excluded. Studies were then reviewed to determine the ‘type’ of acidification the study used to alter water pH.

The P_CO2 level of 2000 µatm was chosen as the AA target. This level is often chosen as the upper limit for OA-targeted marine studies, but we are lacking predictions in freshwater systems, aside from large systems predicted to mirror OA elevations (Phillips et al., 2015). Thus, until we have better predictions for freshwater systems, some of which we know already experience 1000 µatm, 2000 µatm (AA) will be used here for future conditions in both freshwater and marine systems. All data are reported as means±s.e.m.

The effect of acidification on AS was calculated as the log-transformed response ratio (lnRR), and each SMR and MMR was used to calculate each reported instance of AS. The lnRR is the ratio of the mean effect of the acidification treatment compared with the mean effect in a control treatment (Hedges et al., 1999). The overall mean effect was calculated for each response variable (AS and each respective oxygen uptake rate, SMR and MMR) by weighting each individual lnRR by the inverse of the sum of its sampling variance and the between-experiment variance, and then calculating the weighted mean (i.e. random effects meta-analysis; Hedges and Olkin, 1985). All calculations were performed in Rstudio (version 1.0.136, https://www.rstudio.com/) using the metafor package (according to Viechtbauer, 2010).

Over 238 results were assessed, and 36 studies were identified as having investigated the effects of elevated P_CO2 on oxygen uptake rates of fishes (Table S1). When species, P_CO2 levels and exposure times within each study were separated, 93 reports were examined across the 36 studies. Of the included studies, the majority (89.2%) reported SMR, 44.1% reported MMR and fewer studies (28.0%) reported AS.

Of the 26 instances where responses to elevated P_CO2 at the level of AS were examined, the majority (73.1%) reported no effect of elevated P_CO2 on AS compared with control conditions, whereas 15.3% reported a decrease and 11.5% reported an increase in AS. When the reports were distinguished between those above or below the P_CO2 levels expected by 2300 owing to increased AA, 18 instances reported AS results, 14 reported no change, three observed increases and one observed a decrease in AS as a result of elevated P_CO2. Of the 29 reports not focused on climate-change-relevant P_CO2 levels, only eight species in total were examined, with the majority (58.6%) coming from marine systems.

To date, only two studies have examined the effects of fluctuating (versus constant or elevated) P_CO2 levels on fishes (Methling et al., 2013; Ou et al., 2015). The range of constant P_CO2 levels that have been used to examine the effects on oxygen uptake in fishes so far has spanned 260 to 25,000 µatm (3975.0±627.5 µatm). Most of these instances were examined in marine systems (78%) spanning P_CO2 levels of 600 to 25,000 µatm (3757.3±750.4 µatm). In freshwater systems (22% of all the O₂ uptake rate data), the P_CO2 levels under which fish have been examined have ranged from 260 to 19,000 µatm (4747.9±1040.7 µatm).

Most species studied were from the class of ray-finned fishes (i.e. class Actinopterygii; see Glossary); however, 11 of the total collated data points were for chondrichthyan species (i.e. class Chondrichthyes; see Glossary). Of the 11 instances, five reported AS, and of these, three reported no effect of elevated P_CO2 and two reported an increase in AS.

Finally, the majority of fishes examined were either juvenile or adult stages (43.0% for each). Few studies reported metabolic responses to elevated P_CO2 in larvae or embryos (5.4 and 8.6%, respectively). Furthermore, there were seven reports for adult, 11 for juvenile and two for embryonic fishes in terms of the effects of elevated P_CO2 on AS, and the overall response was that of no effect. There was only one reported increase and one reported decrease in AS in adult fishes, and two increases and one decrease in juveniles.

When all respective metabolic performance responses used to calculate each AS were pooled (i.e. SMR and MMR), elevated P_CO2 had a significant effect on SMR (i.e. SMR responses were elevated compared with those of controls) but no effect on MMR or AS (Fig. 2).

Most studies examined oxygen uptake rates either while fish were resting (SMR) or during and/or following exercise (MMR). While important, these metrics may only be capturing half of the story (Fig. 2). As revealed from this review process, the least-reported O₂ uptake metric was AS, which can aid in predicting whole-animal performance and fitness. When either SMR or MMR are reported alone, any conclusions drawn may not have whole-animal implications. For instance, the lnRR of SMR collected from reports used to calculate AS indicates that fish have higher SMR when exposed to elevated P_CO2, suggesting they require more energy for basic maintenance and are ultimately negatively impacted by elevated P_CO2; however, when MMR and AS are examined from the same datasets, there is no effect of acidification (Fig. 2). If SMR had been reported alone, the conclusions drawn from the data would have suggested that the energy budget of fishes exposed to elevated P_CO2 is negatively affected. However, when AS is examined alone, there seems to be no effect, suggesting that fish are unaffected in terms of whole-animal performance and fitness. The lack of data examining the whole picture makes it difficult to draw whole-animal performance conclusions and/or predictions that are key when considering how fish populations will be affected by future acidification scenarios in aquatic ecosystems. We conclude that observing one oxygen uptake metric alone may not be enough.

Of all 26 instances reporting AS responses to elevated P_CO2, only four reported decreased AS (15.4%), all of which were in marine species, but only one within the range of future predicted conditions arising owing to OA – that for the cardinalfish Ostorhinchus cyanosoma (Munday et al., 2009). The other three that reported decreases in AS were all different species and involved studies using very high (e.g. 2800, 8000 and 9200 µatm) P_CO2 levels (Hamilton et al., 2017; Methling et al., 2013; Tirsgaard et al., 2015, respectively). Thus, in the remaining studies examining AS responses to elevated P_CO2 (84.6%), performance has either been maintained (73.1%) or increased (11.5%) compared with controls for 12 different species of fish. Additionally, exposure time did not seem to factor in to AS responses. The exposure periods ranged from less than a day to a year for all of the AS measurements. The increased AS responses compared with those of controls varied from 17 to 224 days of exposure, and the decreased AS responses compared with controls arose from exposures ranging from 4 to 112 days. This trend could indicate that compensating for the acidosis ensuing from these elevated P_CO2 levels is not a significant additional metabolic cost for fish, and no increase in energy is required to maintain aerobic processes. This may also help support an earlier prediction that fish are relatively tolerant to hypercapnia – in terms of metabolic performance (Heuer and Grosell, 2014) – and are therefore able to maintain whole-animal performance, even under future higher AA levels. The majority of fish species examined thus far were able to maintain performance during acidosis, suggesting that fishes are using one or more mechanism to achieve this. One explanation for this tolerance or insensitivity to elevated P_CO2 could be at the mechanistic level, where Root effect Hbs coupled with plasma-accessible CA can enhance oxygen delivery to select tissues during a mild acidosis (Rummer and Brauner, 2011; Rummer et al., 2013a,b).

Although the Root effect has been relatively well studied in teleost fishes, elasmobranchs are not known to possess these unique Hbs, which has been evidenced by the mere <10% reduction in Hb–O₂ saturation in elasmobranchs under an acidosis when compared with the >40% reduction noted for teleosts (Berenbrink et al., 2005). Thus, Root effect Hbs would not be capable of playing a role in enhancing or maintaining O₂ delivery in elasmobranchs during times of elevated P_CO2. However, in all studies examining the effects of elevated P_CO2 on elasmobranchs to date, AS was maintained (83.3%) or increased (16.7%) (Di Santo, 2015, 2016; Green and Jutfelt, 2014). Although there are minimal data – only five reports considered the effects of elevated P_CO2 on AS – the evidence that is available suggests that, similar to teleost fishes, elasmobranchs can maintain or even enhance performance when exposed to likely future OA conditions. We know that elasmobranchs do not possess Root effect Hbs (Berenbrink et al., 2005), yet they retain the ability to maintain or increase performance. This may suggest that elasmobranchs possess a different mechanism that allows them to maintain performance when P_CO2 is elevated, and it is possible that some basal lineages of living ray-finned fishes may use this as well, as evidenced by their minimal decrease in Hb–O₂ saturation with a decrease in pH, suggesting a modest to non-existent Root effect (Berenbrink et al., 2005). Identifying these mechanisms becomes important as these populations are predicted to experience elevations in P_CO2 owing to AA, and this lack of knowledge on how these important meso- and apex predators (see Glossary) are able to maintain performance under acidified conditions points to a need for further research to identify the underlying mechanism.

It is important to point out that the mechanism associated with Root effect Hbs and plasma-accessible CA represents just one explanation for increased O₂ delivery to tissues in teleost fishes during exposure to elevated P_CO2. Although often implied, no studies to date have actually linked this mechanism directly to maintained O₂ uptake and/or performance metrics. Other proposed mechanisms to explain maintained O₂ uptake metrics may have a genetic basis. A recent study by Schunter et al. (2018) suggests that transcriptional changes in the brain during acidification may be linked to the oxidative stress response [e.g. stanniocalcin (stc2)]. Additionally, an increase in brain fibroblast growth factor 1 (fgfI) can promote blood glucose reduction. This aids in the ability of fishes to rebalance gluconeogenesis and glucose homeostasis, which the authors suggest is key to adapting to new environmental conditions (Schunter et al., 2018). Furthermore, the physiological changes that occur at the branchial level are also crucial in determining the efficiency of gas exchange, ion balance and acid–base regulation, all of which underpin O₂ transport. Specifically, both the diffusion distance of the respiratory epithelium and the permeability of the gas through the epithelium can impact gas exchange (Perry et al., 2009). Esbaugh et al. (2016) suggest that, when fish are exposed to hypercapnic conditions, branchial plasticity can reduce diffusion distance, enhancing CO₂ extrusion. For example, the branchial diffusion distance was reduced by 32% in fish exposed to elevated P_CO2 when compared with control fish. Another mechanism that can aid in maintaining O₂ transport is increased cardiac flow-generating capacity during exposure to acidic conditions (Gräns et al., 2014). The mechanisms responsible for enhanced cardiac pumping capacity in fishes acclimated to elevated P_CO2 conditions remain unknown (Gräns et al., 2014); however, this finding emphasizes that more research is needed to tease apart the mechanisms fish use to maintain oxygen transport during an acidosis.

Although elasmobranchs have been insufficiently studied in terms of the effects of elevated P_CO2 (Rosa et al., 2017), a more obvious data gap is for the freshwater fishes. Only three data points from one study on one species showed the response of AS to elevated P_CO2 in freshwater systems. Thus, in addition to lacking predictions of how the P_CO2 of freshwater systems will change in the future, studies investigating how possible elevations will affect the whole-organism performance of these fishes, which would have important fitness implications, are lacking. Furthermore, in all freshwater reported results (20), only five reports were attained for non-adult fish and in only six instances were freshwater fishes exposed to elevated P_CO2 conditions for more than 2 days. Collectively, this information further highlights the need for firm predictions as to how P_CO2 will change in freshwater systems in the future; we urge that studies should focus on an array of age classes and use an array of exposure times.

Although exceptions exist, development and growth of embryos seems to be generally unaffected by P_CO2 (Bignami et al., 2013; Frommel et al., 2013; Heuer and Grosell, 2014; Munday et al., 2011). However, data on O₂ uptake metrics are also lacking for embryonic and larval marine fishes exposed to elevated P_CO2. Adding to our knowledge of how elevated P_CO2 affects O₂ uptake metrics in early-life-stage fish is especially important as these life stages are some of the most sensitive to environmental perturbations (Ishimatsu et al., 2005). Additionally, the role of Root effect Hbs in O₂ delivery has predominantly been examined in adult fishes, largely salmonids (Alderman et al., 2016; Berenbrink et al., 2005, 2011; Rummer and Brauner, 2015; Rummer et al., 2013b; Waser and Heisler, 2005). However, larval pink salmon decrease MMR upon exposure to elevated P_CO2 (Ou et al., 2015), which suggests that there is less energy available for maximum performance, which has implications for predator escape, settlement, foraging, etc. Note that this is just one metric – and we previously mentioned the danger of drawing conclusions based on one metric. Ontogenetic shifts may allow salmonids and other fishes to maintain and increase performance under elevated P_CO2. Other metrics such as swimming performance, growth and survival are also important during early development, with wide implications for the structure of fish communities, which could be even more important if there is an ontogenetic shift in performance in response to elevated P_CO2.

Based on our meta-review, it is clear that there are many gaps in the literature regarding the metabolic responses of fish to elevated P_CO2. However, the general trend, regardless of species, exposure time, CO₂ level and ontogeny, is that performance is largely maintained, with over 60% of all the reports for each trait (SMR, MMR and AS) showing no effect upon exposure to elevated P_CO2.

Despite there being only a few empirical studies focused on the AS responses of fishes to AA-relevant P_CO2 levels, the only increases in AS observed were upon exposure to the range encompassing 900 to 2000 µatm, whereas the majority of the decreases in AS were upon exposure to levels exceeding 2000 µatm. This suggests that fish have a mechanism to maintain and even increase AS under future higher levels of AA; however, once those P_CO2 thresholds are exceeded, metabolic performance could start decreasing. This is not to say, however, that fish will be unable to use phenotypic plasticity and transgenerational acclimation and, ultimately, adapt to increasing P_CO2 over time (Donelson et al., 2011, 2012; Miller et al., 2012).

This Review highlights many gaps in the literature that need to be addressed, but furthermore, serves to reveal a current consensus of studies to date that have found minimal adverse effects of AA on AS. Additionally, we have put forward a hypothesis that could explain how fish maintain aerobic performance during an acidosis. O₂ uptake rates are often used as a proxy for metabolic rates, and, from these, we can estimate the amount of energy available to an animal under different environmental conditions. The amount of energy available to an animal can affect all life processes (e.g. growth, feeding, digestion, swimming, avoiding predators, reproduction, etc.), and thus, if altered during exposure to a certain environmental condition, there can be dramatic implications for future populations. Additionally, there has been recent emphasis on determining ‘winners’ and ‘losers’ during challenging environmental conditions and the traits responsible. These traits have the potential to make an organism more successful and thus increase their survival and reproductive success. If these traits are heritable, the population will start changing.

Variations in behaviours upon exposure to elevated P_CO2 have been linked to differential expression of genes controlling circadian rhythm (Schunter et al., 2016). Offspring of behaviourally tolerant parents displayed a flexibility of gene expression compared to offspring of P_CO2 sensitive parents, allowing them to phase shift their circadian clock and avoid detrimental reactions to elevated P_CO2 conditions. Thus, there is evidence that individual variation in CO₂ sensitivity could facilitate adaptation of fishes and has the potential to influence populations in the future (Schunter et al., 2016). This combination of mechanistic and behavioural research is crucial for predicting the future fate of fish populations. In terms of oxygen uptake, studies on Root effect Hbs have been almost exclusively mechanistic, and the studies on oxygen uptake, swimming performance and behaviour in response to climate change have largely been separate from mechanistic work. The variability of traits that mechanistic research can illuminate can help highlight the traits that selection will be acting on that will drive future adaptations and/or populations. Indeed, collaborative research combining molecular, physiological and ecological approaches will allow us to better predict how fish communities will be affected in the future by OA, and will aid the identification of the systems or species that will be able to adapt and the species that are vulnerable and in need of management.

We are grateful to Erin Walsh for the illustrations used for this contribution.