Lophelia pertusa is a scleractinian cold-water coral that is found throughout the world's oceans, typically between 200 and 1,000 m (Rogers, 1999), although it occurs as shallow as 40 m in Norwegian fjords due to the upwelling of deep, cold water (Strømgren, 1971). In the Gulf of Mexico, L. pertusa is most commonly observed between depths of 300 and 600 m, where it creates habitat that greatly increases local diversity (Cordes et al., 2008). Globally, its distribution is best explained by water temperature (Davies and Guinotte, 2011), while in the Gulf of Mexico, its distribution can best be predicted by depth, local topography, and the availability of hard substrata (Georgian et al., 2014).

L. pertusa were collected in the Gulf of Mexico in April and May 2014 on the R/V Atlantis using the DSV Alvin (Woods Hole Oceanographic Institution). All corals were collected at similar depths, temperature and carbonate chemistry conditions (Table 1) from the Viosca Knoll 826 (VK826) site (Cordes et al., 2008), named for the oil lease block governed by the U.S. Bureau of Ocean Energy Management (BOEM). Each collection was carefully placed in a separate temperature-insulated compartment on the submersible, and kept separate throughout the relocation and experiments. All live corals were kept in natural seawater and transported in insulated containers from Gulfport, Mississippi to Temple University, Philadelphia, Pennsylvania.

At the time of arrival, all corals were kept in a 550 L recirculating aquarium system with artificial seawater (ASW) prepared using Instant Ocean® (aquarium sea-salt mixture) and maintained at a temperature of ~8°C, salinity 35 ppt, and total alkalinity (TA) ~2,300 μmol kg⁻¹ (for full description of the system see Lunden et al., 2014b). The corals were fed 20 mL of artificial MarineSnow® (Two Little Fishies) 3 days a week, as per comparable experimental procedures (Lunden et al., 2014a,b; Hennige et al., 2015; Georgian et al., 2016b). Corals were allowed to acclimate to laboratory aquaria conditions for a period of approximately 3 months prior to the start of experiments.

Paired water samples were collected directly above (<2 m) each of the six coral collection sites using vehicle-mounted Niskin bottles. Samples were measured for pH on the total scale (pH_T) using an Orion 5 Star pH meter and glass electrode (ROSS Ultra pH/ATC Triode 8107BNUMD). Total alkalinity was measured by acid-titration on an autotitrator (Mettler-Toledo DL15, 0.1 mol L⁻¹ HCl), with accuracy checked by measuring certified reference material (Dickson Lab, batches 138 and 141; Dickson et al., 2007). CO2Calc software (Robbins et al., 2010) was used to calculate pCO₂, HCO3−, CO32−, and Ω aragonite (Ω_ar) using pH and total alkalinity as input variables with the dissociation constants for boric acid and K1 and K2 from Mehrbach et al. (1973) refit by Dickson and Millero (1987), KHSO₄ from Dickson (1990), total boron from Lee et al. (2000), and pH on the total scale (pH_T). In situ salinity, pressure, and temperature were measured using a vehicle-mounted CTD (Seabird SBE 49). For the full methodology of water sampling and analysis, see Georgian et al. (2016a).

Corals were collected from six discontinuous reef patches within VK826 to increase the probability that different genotypes were sampled. Previous work indicates relatively low levels of clonality among distinct coral colonies at this site as compared to L. pertusa populations in the North Atlantic (Waller and Tyler, 2005; Lunden et al., 2014a). To ensure that distinct genotypes were used in our experiments, each coral fragment was genotyped according to methods used previously in a similar study of L. pertusa (Lunden et al., 2014a). Briefly, coral genotypes were determined using 10 L. pertusa microsatellite loci (Morrison et al., 2008; Molecular Ecology Resources database entry 51059) from total DNA extracted from coral polyps. Of the six collections used in the live coral experiments, all of the microsatellite loci amplified, except for one collection in which only 6 of the 10 amplified. In all pairwise comparisons, the highest number of shared microsatellite genotypes was 2, the average was 0.8, and in 7 of the 15 pairwise comparisons, there were no single-locus genotypes shared between the two individuals. These results indicate that each of the samples represents a distinct genotype.

Because Instant Ocean® at salinity 35 yields seawater with a total alkalinity of ~3,600 μmol kg⁻¹, 12.1 N HCl was added to previously prepared artificial seawater (ASW) in order to reduce the total alkalinity to ~2,300 μmol kg⁻¹ before adjusting the pH (Lunden et al., 2014b). The manipulation of the pH was accomplished by bubbling pure CO₂ gas into the different treatment tanks using an automated CO₂ injection system (American Marine Inc, PINPOINT pH Monitor). To reduce pH, pure CO₂ was mixed into a reaction chamber that dissolved the CO₂ into seawater before being delivered into the respective tanks. If pH levels fell below the desired threshold, ambient air was automatically bubbled into the tank. The pH meters underwent a two-point Tris-HCl and AMP-HCl calibration weekly (Dickson et al., 2007). In both experiments the pH_T of the experimental tanks was gradually brought down to the desired treatment conditions at a rate of ~0.1 pH units day⁻¹. Total alkalinity (TA) was measured twice a week as described above. Total alkalinity and pH_T were used to calculate pCO₂, HCO3−, CO32−, and Ω_ar in CO2Calc software (Robbins et al., 2010) as described above. Salinity was measured daily using a handheld refractometer, and temperature was recorded every 5 min using a temperature logger (Onset HOBO Pendant®). A 10% water change was performed weekly for the long-term experiment, and twice weekly for the short-term experiment.

The short-term experiment began in August 2014. Three pH_T treatments were selected: 7.90, which approximates the average in situ pH_T at the VK826 collection site (Lunden et al., 2013); 7.75, which results in seawater near the aragonite equilibrium point (Ω_ar = 1); and 7.6, which generates undersaturated seawater with respect to aragonite (Ω_ar < 1). Each treatment contained 12 fragments from four different genotypes (36 fragments in total). These treatments were established in three separate 50 L tanks housed in a cold room, which was maintained at 8°C. Each of these fragments was weighed at the beginning and end of the 2-week pH exposure to determine net calcification rates.

In July 2014, the six collections were randomly split between two identical 550 L recirculating units (fully described in Lunden et al., 2014b). Two pH_T treatments were selected: 7.90 as the control, and 7.60 as the treatment. The pH was gradually reduced in August 2014 to treatment conditions (control = pH 7.90, acidified = pH 7.60). Fragments (n = 4) from each genotype (n = 6) of five to eight live polyps were set-aside in each tank for buoyant weighing every 2 weeks (48 fragments in total).

The buoyant weight (BW) technique (Jokiel et al., 1978; Davies, 1989) was used to measure net calcification using a Denver Instruments SI-64 closed analytical balance (Precision of 0.1 mg). All fragments from each treatment were weighed before the gradual pH decline, when the pH reached the desired treatment level, after 14 days, and in the long-term experiment, subsequently every 2 weeks for a 6-month time period. The balance was closed within a plexiglass chamber to prevent any outside disturbance from affecting the weight. Each fragment was removed from its treatment tank and transported to the balance in a 4 L beaker without exposure to air. Fragments were gently attached to a monofilament fishing line, and suspended within the water from the bottom of the balance. After 10 s of stabilization, the weight was recorded. Each of the fragments in both the short- and long-term experiments was weighed in triplicate during each buoyant weight trial to minimize error.

Salinity and temperature of the water at time of weighing was recorded, and used to calculate the density of the water and to ensure that the treatment conditions were maintained throughout the weighing process. The dry weight of each fragment (W_a) was calculated using the measured buoyant weight (W_w), the density of the seawater during measurement (D_W), and the coral skeletal density (SD, 2.82 g cm⁻³) (Lunden et al., 2013). Wa=Ww1-DwSD Net calcification of L. pertusa was calculated as the change in weight over each 2-week interval, expressed as % net calcification d⁻¹. The average weight of the fragments at the start of the experiment was 15.81 ± 7.39 g. For the long-term experiment, net calcification rates were averaged monthly to yield a total of six time points. Percent net calcification per day was calculated by the equation described in Lunden et al. (2014a): Nt=100 × Ww2- Ww1 Ww1(T2-T1) where W_w2 and W_w1 is the final and initial standardized buoyant weight respectively, and T₁ and T₂ equal time 1 and time 2, respectively.

At the beginning and end of the long-term experiment, coral fragments from each genotype were selected to assess the potential effects of ocean acidification on enzyme activity. Collected corals were frozen in liquid nitrogen and stored in −80°C until analyses were performed. Frozen coral polyps (2-3 polyps) from each genotype were extracted on liquid nitrogen, ground with a mortar to disrupt membrane cells, and homogenized with 3.5 mL of sterilized artificial seawater (ASW). The homogenate was centrifuged at 4°C for 3 min at 1,500 g. 200 μL was reserved for total protein quantification using bicinchoninic acid assay (Pierce™ BCA assay kit), and 1.5 mL of the remaining homogenate was transferred into a 7 mL scintillation vial and diluted 1:1 with 25 mM veronal buffer. The in-vitro assay for carbonic anhydrase was performed according the protocol described by Weis et al. (1989), and was measured by the decrease in pH resulting from the hydration of CO₂ to HCO3− and H⁺ after the addition of substrate (CO₂−enriched distilled H₂O, pH = 3.5, 4°C). For each assay, 1 mL of buffered homogenate was transferred to a scintillation vial equipped with a magnetic stir bar and further diluted 1:1 with the same veronal buffer. The homogenate was maintained at ~3.5°C before running the assay. The mixture was constantly stirred, and 1 mL of substrate was added to the sample when the pH stabilized and temperature reached 13°C. From this time, the drop in pH was recorded every 10 s for a total of 120 s. As a negative control for CA activity, a subsample of the homogenate (~1.7 mL) was boiled for 5 min to denature the enzyme and treated as above. Enzyme activity was calculated as the difference of the decrease in pH between the sample with the active enzyme and the control with the denatured enzyme, and is given as enzyme units (EU) standardized to total protein content.

For the short-term experiment, a two-way ANOVA was used with net calcification (change in mass as % d⁻¹) as the dependent variable, pH as a fixed factor (three levels: pH = 7.9, 7.75, 7.6) and genotype as a random effect (6 levels). Multiple pair-wise comparisons were then applied to investigate the differential response to pH at the genotype level, using fragments within genotype (n = 12) as replicates. When considered together, the net calcification rate data were not normally distributed, and there were no simple transformations that resulted in a normal distribution. However, when considered separately, the data within each pH level were normally distributed. Furthermore, ANOVA is known to be very robust with respect to violations of this assumption (Schmider et al., 2010), and therefore we proceeded with the analysis as planned.

In the long-term experiments, a three-way factorial ANOVA was performed with growth (% d⁻¹) as the dependent variable, pH and time as fixed factors and genotype as a random effect, using each individual fragment as a replicate (Quinn and Keough, 2002). Tukey's HSD test was used to test for pairwise differences in the response of each genotype to the different pH treatments, where fragments within genotypes (n = 24) were used as replicates. Additionally, to investigate potential differences in the amount of time that a given genotype could maintain positive calcification in undersaturated conditions, survival curves were generated using Kaplan-Meier “time to event” analysis using a Mantel-Cox log rank test followed by a post-hoc Wilcoxon exact test to determine if significant differences were found (Kaplan and Maier, 1958). In this case, the “event” in question was the change from positive to negative calcification rate, and the time when negative calcification was first observed in each of the fragments of a given genotype within a treatment was used as the data in the analysis.

Differences in enzyme activity of L. pertusa at the end of the long-term experiment were assessed with a two-way ANOVA with pH (two levels) and time (initial and final time) as fixed factors. If significant differences were found, genotype was further analyzed by multiple pair-wise comparison. All analyses were applied to the mean standardized values of enzyme activity among replicate fragments of each genotype. For all the analyses, results were considered significantly different with p < 0.05. All data are presented as average ± standard deviation unless otherwise indicated.

Mean values of temperature, salinity and all carbonate chemistry parameters for the different treatments in both short and long-term experimental aquaria were relatively stable during the course of the two experiments. All parameters remained within a very narrow range around the target values of 8°C, 35 ppt, and pH values of 7.9, 7.75, and 7.6 (Table 2). These parameters, along with a total alkalinity near 2,300 resulted in Ω_ar values of approximately 1.5, 1, and 0.8 in the three treatment levels.

Our results did not show a statistically significant response in the short-term (2-week) exposure to low pH in L. pertusa; however, this was primarily due to the extremely high variability in the response of different genotypes within the pH 7.6 treatment. Short-term experiments of ocean acidification with Gulf of Mexico L. pertusa populations have found reductions in net calcification rate with the same pH treatments (Lunden et al., 2014a; Georgian et al., 2016b). Lunden et al. (2014a) found that short-term low pH exposure had an overall negative effect on calcification, although some coral fragments were able to maintain positive calcification rates even at undersaturated levels of aragonite. Moreover, Georgian et al. (2016b) found a 48% decline in net calcification of L. pertusa grown at pH 7.75, and the net dissolution of L. pertusa grown at pH of 7.6. In the low pH treatment (pH = 7.6) in this study, average net dissolution was also observed. However, we observed extremely high variability among genotypes, but consistent responses within genotype. Genotype G4 maintained significantly positive calcification rates in all individuals during the short-term exposure to undersaturated conditions.

Other short-term experiments with cold-water corals including different populations of Lophelia pertusa from the North Atlantic and Mediterranean Sea have shown contrasting results. Despite these inconsistencies, there is evidence for declines in calcification as pH decreases (Maier et al., 2009; Form and Riebesell, 2012). For example, a study using the reef-building cold-water coral Madrepora occulata, found that calcification rates are 50% slower than in pre-industrial times, and suggesting that ocean acidification has already impaired cold-water coral calcification (Maier et al., 2012). Moreover, Maier et al. (2009) found that lowering pH by 0.15 to 0.3 units (from an initial value of pH 8.1) according to future scenarios of ocean acidification resulted in a decrease in L. pertusa's gross calcification by 30 and 56% respectively.

In our study, coral fragments in the control treatment (pH = 7.9) and the slightly acidified treatment (pH = 7.75) maintained similar calcification rates. The similar calcification rates between the control and the 7.75 treatment can likely be attributed to the fact that both treatments were still supersaturated with respect to aragonite (Ω_ar), therefore positive calcification was still energetically favored and dissolution was not likely to occur. These results are similar to the in situ deep-water (McCulloch et al., 2012; Wall et al., 2015) and shallow-water (Cohen and McConnaughey, 2003; Venn et al., 2013; Holcomb et al., 2014) studies that have found corals growing in low pH conditions, presumably due to specialized cellular mechanisms for maintaining elevated pH and carbonate ion concentration within the calcifying fluid.

In the short-term experiment, two of the three replicate fragments from G1 displayed a very high net dissolution rate, while all of the replicates of G4 had positive net calcification within the low pH treatment. It should be noted that the overall net calcification rates reported here for the high pH treatments, designed to simulate ambient in situ conditions, were lower than those found in the previous studies of Gulf of Mexico L. pertusa populations, although within the range of these and other studies (Table 3). Since these studies were conducted in the same laboratory using the same methodology by many of the same scientists, it is possible that some of the coral fragments examined here were in a poorer condition at the time of collection than those collected in previous years, or that they included some genotypes that simply do not perform as well under ambient conditions.

The fact that genotype G4 was able to calcify under low pH and undersaturated conditions suggests that these corals may have started with larger energy reserve, such as carbohydrates, lipids, or proteins, allowing them to allocate additional energy to the calcification process, and as has been found in shallow-water calcifying corals that maintain calcification rates under environmental stress (Rodrigues and Grottoli, 2007; Schoepf et al., 2013). When comparing Gulf of Mexico and Norwegian L. pertusa populations, Georgian et al. (2016b) attributed the higher calcification rates of the Norwegian corals to higher food supply at the shallower Norwegian sites and an elevated metabolic response of the corals. However, increased food supply did not confer an increased resilience to lower pH in Madrepora oculata from the Mediterranean Sea (Maier et al., 2016). It is also possible that certain genotypes are more resilient to low-pH stress via mechanisms not yet understood, but plausibly related to the up-regulation of the calcification pathway or certain stress genes and pathways (Carreiro-Silva et al., 2014) that warrant further investigation.

The finding of a variable response of L. pertusa genotypes to the undersaturated treatment reflects the disparate results from the literature. In the short-term experiment here, two of the genotypes exhibited net dissolution (G1 and G3), while one of the genotypes (G4) was capable of maintaining significant positive net calcification in undersaturated conditions (Ω_ar = 0.81, range 0.76–0.85). This is similar to previous experiments with L. pertusa from the Gulf of Mexico (Lunden et al., 2014a), where a few of the genotypes tested were capable of positive calcification in undersaturated conditions, although the former experiment was not designed to include genetic replicates. Maier et al. (2009) found that L. pertusa from two different populations in the North Atlantic were able to calcify in slightly undersaturated levels of aragonite (Ω_ar = 0.97) albeit at reduced calcification rates. However, the authors used ⁴⁵Ca to measure gross calcification over a 24-h period, which does not take into account possible dissolution effects, while the buoyant-weight technique used here integrates both biologically mediated calcification and chemical dissolution. Therefore, is plausible that all of our coral fragments in the highly acidified treatment (pH = 7.6) were still able to calcify, but at slower rates that are insufficient to compensate for skeletal dissolution.

This is the first long-term experiment performed on L. pertusa from the Gulf of Mexico, although long-term experiments have been conducted using L. pertusa from the North Atlantic (Form and Riebesell, 2012; Hennige et al., 2015), and the Mediterranean sea (Maier et al., 2013; Movilla et al., 2014). Our results showed a significant effect of time and pH, where corals in the control treatment (pH 7.9) maintained positive net calcification, while, on average, corals in the acidified treatment (pH = 7.6) experienced declining net calcification rate over time and net dissolution by the end of the 6-month period. On average, corals were not able to maintain calcification even after only 1 month in the low pH treatments, however there was a variable response among genotypes over time. Genotype 3 maintained positive net calcification for 2 months, and G4 displayed positive net calcification for approximately 3 months before net dissolution started to occur, longer than any of the other genotypes (Figure 4). This provides evidence for an adaptive capacity in this population of L. pertusa that would improve its probability of surviving predicted levels of ocean acidification.

These results contrast with what has been found so far in other long-term experiments where there has not been a significant negative response in calcification for L. pertusa (Form and Riebesell, 2012; Maier et al., 2013; Movilla et al., 2014; Hennige et al., 2015) and for other cold-water coral species (Carreiro-Silva et al., 2014; Rodolfo-Metalpa et al., 2015; Gori et al., 2016). However, the majority of the studies performed in cold-water corals to date have used pH levels that still yielded Ω_ar saturation states at or near 1. Form and Riebesell (2012) found that in the long-term (six months), corals exposed to low pH conditions (pH: 7.75; Ω_ar: 0.93) experienced no significant change in calcification compared to control conditions suggesting an acclimation potential. Maier et al. (2013) and Movilla et al. (2014) in different experiments with L. pertusa and M. oculata, did not find any significant difference in net calcification after 9 and 6 month experiments respectively, however, all of the treatments in those studies were saturated with respect to aragonite (Ω_ar > 1). Likewise, Carreiro-Silva et al. (2014) performed a long-term experiment (8 months) on the cold-water coral Desmophylum diantus subjected to two levels of pH that simulated ambient (pH 8) and acidified conditions (pH 7.7) and found no significant differences in average net calcification, however their experiment were never subjected to undersaturation for aragonite. In this study, none of the L. pertusa genotypes examined were capable of maintaining positive net calcification beyond 90 days in undersaturated conditions. Although we did not measure gross calcification in the present study, it is clear that dissolution is an important factor controlling overall calcification and net calcification rates in the natural environment, and measurements of gross calcification should be incorporated into future studies (Rodolfo-Metalpa et al., 2015).

The most recent direct comparison to the present study, because of the inclusion of a treatment with Ω_ar < 1, measured net calcification in L. pertusa populations from the North Atlantic subjected to undersaturated conditions (Hennige et al., 2015). Although they did not find evidence for a significant response in net calcification rate, their data showed a negative trend with calcification rates approaching zero at undersaturated levels of aragonite. However, it is important to bear in mind that this maintenance of calcification under ocean acidification scenarios comes with a significant energetic cost. Norwegian populations of L. pertusa were found to be able to maintain calcification only by increasing their feeding and respiration rates, while the Gulf of Mexico populations did not exhibit this physiological response (Georgian et al., 2016b).

Environmental differences between the regions may account for some of the variability among experiments conducted in different regions, as relatively isolated L. pertusa populations (Morrison et al., 2011; Lunden et al., 2014a) may have acclimated or adapted to local pH and carbonate conditions. The corals used here were collected from 400 to 500 m at an ambient pH of approximately 7.9 (Table 1), as opposed to the North Atlantic and Norwegian corals that were collected from a depth of 70–150 m and a pH of approximately 8.0 to 8.1 (Maier et al., 2009; Form and Riebesell, 2012). Positive calcification rates were maintained in an experiment on Norwegian corals due to an increase in feeding rate and higher metabolic rate (Georgian et al., 2016b), and these mechanisms will remain effective as long as food supplies are sufficient. Model simulations indicate that L. pertusa reefs induce downwelling of organic-carbon rich waters, which would serve to ensure a consistent food supply (Soetaert et al., 2016), at least while surface productivity rates remain consistent. Even though overall net calcification is positive in lab experiments at low pH and undersaturation of aragonite, dissolution of coral fragments has been noted, particularly in areas that were not covered by tissue (Hennige et al., 2015). There was also a decrease in the breaking strength of new skeletal structure, changes in the crystallographic organization, as well as thinner corallites grown in high CO₂ (Hennige et al., 2015). Furthermore, Gulf of Mexico L. pertusa populations (Georgian et al., 2016b) and Mediterranean M. oculata populations (Maier et al., 2016) do not appear to be able to increase feeding rate to compensate for the increased energetic cost of calcification at low saturation states. These changes have the capacity to alter the stability of reef frameworks in the future.

Standing genetic variation and local adaptation may play a role in species persistence in a changing ocean. Kelly et al. (2013) tested the responses of larval Strongylocentrotus purpuratus cultured under varying pCO₂ conditions, and observed differences in larval length among individuals collected from two different populations characterized by different CO₂ regimes. The authors found that urchins from naturally high-CO₂ environments were less sensitive to OA, implying a capacity for adaptation to ocean acidification. In a related study using several populations of S. purpuratus, Pespeni et al. (2013) observed rapid selection of genes encoding proteins involved in calcification under acidification, demonstrating a potential for rapid evolution to occur in this species. Generally, in slow-growing organisms with long-generation times such as cold-water corals, persistence in a changing ocean will likely hinge on the maintenance of genetic diversity and the success of resilient genotypes, the potential for individual acclimatization, or local adaptation among populations in the presence of both selection and gene flow (Palumbi and Sotka, 2006). While it is unknown if the observed resilience within certain genotypes to climate change stressors in the present study of L. pertusa is heritable, the general consistency in the response of different nubbins of the same genotype within and among different experiments provides evidence for the adaptive capacity of this species.

Identification of the variability of responses due to genetics may lead to future studies of the mechanisms underlying such variation. One mechanism responsible for this could be differential expression of calcification genes. A recent study using next-generation sequencing techniques observed significant effects of OA on gene expression in larvae and primary polyps of the scleractinian coral Acropora millepora (Moya et al., 2012). Notably, genes involved in CaCO₃ deposition showed patterns of differential expression under OA, including down-regulation of genes encoding proteins that support the skeletal organic matrix. It is unclear if this study employed true biological replicates, and different genotypes will likely have different thresholds for their transcriptomic response. An interesting question for future work would be to investigate the potential for different individuals or populations to balance these processes under acidification, as has been observed in natural coral populations responding to thermal stress (Barshis et al., 2013).

We did not detect any significant change in the overall activity for carbonic anhydrase in the long-term experiment. Lophelia pertusa maintained stable activity after a 6-month exposure, but exhibited higher variability in the low pH treatment. While our measured values (between 0.002 and 0.031 units of enzyme activity) are low compared to studies of zooxanthellate corals, it has been found that CA activity in zooxanthellate corals is 29 times higher than in azooxanthellate species (Weis et al., 1989). Reported values for CA activity in the azooxanthellate coral Tubastrea aurea were 0.08 units of enzyme activity Weis et al. (1989) reported similar values of 0.06 units for different species of azooxanthellates, which are close to the values obtained here.

In the low pH treatment, there was a slight increase in overall CA activity due to two genotypes that showed increased activity despite a decrease in net calcification. Calcification in scleractinian corals occurs in a biologically mediated environment (Allemand et al., 2011) and carbonic anhydrase actively participate in the calcification process (Tambutté et al., 2007; Moya et al., 2008; Bertucci et al., 2013). In symbiotic scleractinians, reductions in calcification are correlated with decreased enzyme activities (Goreau, 1959; Weis et al., 1989), where calcification is tightly linked with photosynthesis and CA acts as a CO₂ supplier for the endosymbiont (Weis et al., 1989). However, metabolic CO₂ is not the main source of DIC for calcification in deep-water corals (Adkins et al., 2003; Mueller et al., 2013). An increase in CO₂ will cause an increase in DIC, specifically by producing HCO3− and H⁺, which ultimately will alter the internal chemistry and lower the pH inside the cells (Allemand et al., 1998). It is plausible that the role of the isoform of αCA found in corals (Bertucci et al., 2013), is to regulate the build-up of H⁺ ions that result from the hydration of CO₂ to HCO³⁻ in order to avoid internal acidosis. This has been suggested for the scleractinians Stylophora pistillata (Moya et al., 2008), and the cold water coral Desmophyllum dianthus (Carreiro-Silva et al., 2014), where an over-expression of the genes that encode for CA was found when high levels of H⁺ were present in the internal compartments of the cells.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.