Experiments with coral fragments (i.e. nubbins) have shown that net calcification is depressed by elevated PCO2. Evaluating the implications of this finding requires scaling of results from nubbins to colonies, yet the experiments to codify this process have not been carried out. Building from our previous research demonstrating that net calcification of Pocillopora verrucosa (2–13 cm diameter) was unaffected by PCO2 (400 and 1000 µatm) and temperature (26.5 and 29.7°C), we sought generality to this outcome by testing how colony size modulates PCO2 and temperature sensitivity in a branching acroporid. Together, these taxa represent two of the dominant lineages of branching corals on Indo-Pacific coral reefs. Two trials conducted over 2 years tested the hypothesis that the seasonal range in seawater temperature (26.5 and 29.2°C) and a future PCO2 (1062 µatm versus an ambient level of 461 µatm) affect net calcification of an ecologically relevant size range (5–20 cm diameter) of colonies of Acropora hyacinthus. As for P. verrucosa, the effects of temperature and PCO2 on net calcification (mg day−1) of A. verrucosa were not statistically detectable. These results support the generality of a null outcome on net calcification of exposing intact colonies of branching corals to environmental conditions contrasting seasonal variation in temperature and predicted future variation in PCO2. While there is a need to expand beyond an experimental culture relying on coral nubbins as tractable replicates, rigorously responding to this need poses substantial ethical and logistical challenges.
Together with climate change (i.e. rising temperature), ocean acidification (OA) threatens tropical reef corals (Hoegh-Guldberg et al., 2007), with effects that augment other anthropogenic disturbances (e.g. Miller, 2015). Understanding the threats posed to net calcification of corals by OA is based on research conducted with coral fragments (i.e. nubbins; Birkeland, 1976) that lend themselves to experiments in aquaria. Such studies have found added relevance to reef restoration efforts that rely on coral ‘fragging’ to increase population sizes (Barton et al., 2017). As fragging produces numerous coral nubbins, this nascent field can benefit from what has been learned through scientific studies conducted with nubbins. Yet, a coral reef is more than countless nubbins growing on hard surfaces.
Most tropical scleractinians are colonial organisms growing through iterations of a modular design (i.e. polyps) to create colonies with emergent properties attributed to corallum morphology (Patterson, 1992a,b; Enríquez et al., 2017; Stocking et al., 2018; Burgess et al., 2017; Dornelas et al., 2017). This body plan was long thought to circumvent allometric constraints on size (Sebens, 1987; Hughes, 2005), thereby providing a convenient rationale for using analyses of coral nubbins to support inferences about larger coral colonies (Spencer Davies, 1984; Shafir et al., 2003). However, there is now evidence that traits such as calcification, respiration and photosynthesis will vary in proportionately dissimilar ways with colony size in corals (Chamberlain and Graus, 1975; Jokiel and Morrissey, 1986; Patterson, 1992a,b; Edmunds and Burgess, 2016; Dornelas et al., 2017). Allometric scaling arises from intrinsic constraints caused by the covariation of organism area and volume (Schmidt-Nielsen, 1984), but it can also arise from extrinsic effects of organism design (i.e. morphology), such as within-colony variation in tissue thickness (Brown et al., 1999; Fitt et al., 2000; Thornhill et al., 2011), and its interactions with the environment (Patterson, 1992a,b; Ong et al., 2017; Burgess et al., 2017).
For more than a decade, nubbins have been used as experimental replicates to study the effects of OA on corals. While many such studies have revealed negative effects on calcification (Chan and Connolly, 2013; Kroeker et al., 2013; Comeau et al., 2014), greater ecological relevance in OA experiments has been sought (Gaylord et al., 2015; Barner et al., 2018), in part, by increasing the size of coral colonies studied. One test of these properties explored the impacts of elevated PCO2 (400 and 1000 µatm) and temperature (26.4 and 29.7°C) on net calcification of Pocillopora verrucosa (identified based on morphological features after Veron and Pichon, 1976) varying in diameter from 2.6 to 12.0 cm (Edmunds and Burgess, 2016). The study revealed size-dependent effects, but net calcification was unaffected by PCO2 or its interaction with temperature. These results contrasted with the inhibition of net calcification by high PCO2 in nubbins of P. verrucosa (Comeau et al., 2014), and Edmunds and Burgess (2016) (see also Burgess et al., 2017) speculated that the different outcome was a product of the emergent properties of intact colonies.
In this study, we describe an experiment designed to assess the generality of our previous result, and expand the taxonomic breadth of our analyses (Edmunds and Burgess, 2016, 2018) by working with the common Indo-Pacific coral Acropora hyacinthus (Dana 1846). The colonies of A. hyacinthus also varied in diameter, but unlike P. verrucosa, they had a table-like morphology composed of many small and thin branches, among which the flow of seawater, solute concentrations, metabolite fluxes and zooplankton capture are likely to vary. We focused on an acroporid to test for generality of results for two of the most important functional groups of corals on Indo-Pacific fore reefs (Done, 1983; Veron, 2000; Vercelloni et al., 2019), and used A. hyacinthus because of its high abundance on the fore reefs of Mo'orea when the experiment was conducted (P.J.E., unpublished data).
MATERIALS AND METHODS
The two trials were completed during two field seasons (April to May in 2017 and 2018) at the Richard B. Gump South Pacific Research Station on Mo'orea (17°32'S 149°50'W). Both trials were completed in a mesocosm consisting of 150 l tanks that were independently heated, chilled, illuminated and supplied with CO2 gas. The objective of working with large coral colonies prevented more than one colony of each size class being placed in each tank, because replicate colonies caused the total alkalinity (AT) of the seawater to be drawn down through calcification. It was not possible to compensate for depressed AT through higher rates of seawater turnover in each tank, as this destabilized treatment conditions for PCO2 and temperature. Trial 1 was conducted in 2017 with 8 small (∼6 cm diameter) and 8 large (∼17 cm diameter) colonies of A. hyacinthus that were incubated for 23 days in 8 tanks crossing two temperatures (26.5 and 29.3°C) and two PCO2 levels (∼471 and 1062 µatm). Trial 2 was completed in 2018, and was implemented to replicate the first trial as closely as possible (i.e. 26.4 versus 29.1°C, and 450 versus 1062 µatm PCO2). Trial 2 included 1 small and 1 large coral in each of 7 tanks; an eighth tank was included in an orthogonal design, but was dropped from the analysis following an equipment malfunction.
Net calcification was used as a response variable and was measured by buoyant weighing (Spencer Davies, 1989), and expressed by colony as a function of colony size (tissue surface area). In both experiments, sample sizes were selected based on previous experiments in which statistical power was adequate to test the hypotheses of interest, but the final design was a compromise among the needs for experimental replicates, the importance of statistical independence, and the capacity for high precision in treatment conditions. All coral colonies were allocated randomly to treatment conditions.
Colonies of A. hyacinthus were haphazardly collected from 10 m depth on the fore reef of Mo'orea. Collection targeted small (∼6 cm diameter) and large (∼17 cm diameter) colonies that were bagged individually, and returned to the lab immersed in seawater and shaded from sunlight. Corals for trial 1 were collected on 11 April 2017, and those for trial 2 on 9 April 2018, and in both cases they were kept in a 1000 l tank for 4 days to adjust to laboratory conditions. This tank was maintained at ambient seawater temperature when the experiments were conducted (28.3±0.1°C in 2017 and 29.3±0.1°C in 2018, mean±s.e., N=5 and 26, respectively), supplied with fresh seawater pumped from 14 m depth in Cook's Bay (at 10 l min−1), and illuminated at 452±12 and 618±11 µmol photons m−2 s−1 (mean±s.e., N=5 and 7, respectively) with LED lamps (Aquaillumination SOL white). Light was measured as photon flux density (PFD in the range of photosynthetically active radiation, PAR) using a quantum sensor (LiCor LI 193), and PFD was adjusted to approximate the values recorded at the collection depth on the fore reef of Mo'orea in April (ca. 645 µmol photons m−2 s−1 at noon). The corals were located on a table within the tank that rotated at 2 revolutions day−1 to remove position effects. Following lab adjustment, the corals were buoyant weighted (±1 mg for small corals and ±10 mg for large corals) and placed into the treatment tanks for the 23 day experiment.
One small and one large coral were randomly assigned to each 150 l tank in the mesocosm, with 8 tanks used to create orthogonal contrasts of two temperatures and two PCO2 regimes. Two tanks were nested in each combination of conditions in both trials, but equipment malfunction reduced trial 2 to 7 tanks. Coral colonies were supported upright in the tanks by placing them in PVC cups, and each was identified with a numbered tag. The mesocosm (including lights and PCO2 manipulations) is described elsewhere (Edmunds and Burgess, 2016), and here was used to target treatments of ∼26.5 and ∼29.0°C, which created a contrast of two temperatures within the current annual range of seawater temperatures in Mo'orea (Edmunds, 2017). The PCO2 treatment contrasted present-day (i.e. ∼400 µatm PCO2, ambient) and future PCO2 conditions (∼1000 µatm PCO2). The high PCO2 is a possible outcome by the end of the current century under the pessimistic RCP 8.5 scenario (IPCC, 2014). The temperature treatments tested for the effects on net calcification of ecologically relevant temperatures occurring within a year, but the high value was also selected to reduce the possibility of thermal bleaching. In Mo'orea, coral bleaching occurs when seawater temperature exceeds about 29.8°C (Wyatt et al., 2019), and the use of an upper temperature below this value allowed us to focus on the effects on calcification rather than bleaching.
Both trials lasted 23 days, during which the physical and chemical conditions in the tanks were monitored, and the instantaneous values were used to adjust the control systems to maintain treatment conditions. Seawater temperature was recorded at least daily [using a certified thermometer (±0.05°C), model 15-077, Fisher Scientific, Pittsburgh, PA, USA]. Daily measurements were also taken for salinity (using a conductivity meter, YSI 3100, YSI Inc., Yellow Springs, OH, USA), pH (using a hand-held meter, Orion 3 star meter, Mettler-Toledo, LLC, Columbus, OH, USA, fitted with a Mettler DG115-SC probe) and light (Li Cor, LI-193 sensor attached to a Li Cor LI-1400 meter). Seawater samples (50 ml) were collected every 2–3 days for the analysis of AT using potentiometric titrations (after SOP3b; Dickson et al., 2007) conducted using an open cell, automatic titrator (Model T50, Mettler-Toledo) fitted with a pH probe (Mettler DG115-SC) that was calibrated on the total scale using Tris/HCl buffers (Dickson et al., 2007). The accuracy and precision of the titrations was determined using certified reference materials (CRMs, batches 151, 158 and 172) from A. Dickson (Scripps Institution of Oceanography). The accuracy of the pH measurements obtained with the hand-held meter was determined by periodic comparison with values determined using m-Cresol Purple dye (no. 211761, Sigma-Aldrich, St Louis, MO, USA) according to SOP7 of Dickson et al. (2007). The measurements of seawater pH and AT were used to calculate dissolved inorganic carbon (DIC) parameters using Seacarb (Gattuso et al., 2015) running in the R software environment (v3.6.1; R Foundation for Statistical Computing; http://www.R-project.org/).
Coral colonies were haphazardly moved within the tanks daily to avoid position effects, and periodically were inspected for night-time polyp expansion as an indicator of normal biological function. At the conclusion of the trials (8 May in 2017 and 7 May in 2018), the corals were buoyant weighed, and based on the change in buoyant weight, an empirical determination of seawater density and an aragonite density (Ωarag) of 2.93 g cm−3, the mass increment representing net calcification (in grams) was calculated. The surface area of live coral tissue was calculated by wax dipping (Stimson and Kinzie, 1991) and used as a measure of colony size.
The experiment was analyzed using linear mixed effects models in the nlme package in R. Colony size (surface area in cm2), temperature and PCO2 were modeled as fixed effects. Years and tanks were modeled as random effects. The response variable was net calcification, measured as the average daily change in skeletal mass (mg) over 23 days (i.e. mg day−1). Heteroscedasticity of residuals was modeled as a power variance function with respect to colony size. Analysis of variance was performed on the linear mixed effects model using conditional F-tests, although it should be noted that the P-values are large-sample approximations and are anti-conservative.
The physical and chemical conditions during trials 1 and 2 were stable and differed between treatments for temperature and PCO2 (Table 1). Trial 1 contrasted 26.5±<0.1 versus 29.3±<0.1°C, and 471±26 versus 1062±9 µatm PCO2. Trial 2 contrasted 26.4±0.1 versus 29.1±0.1°C, and 450±9 versus 1062±13 µatm PCO2 (mean±s.e., N=3–4). Mean (±s.e.) AT of the incubation seawater remained close to ambient seawater in Mo'orea (∼2300 µmol kg−1; Doo et al., 2019) during trial 1 (2304±3 µmol kg−1, N=8) and trial 2 (2285±4 µmol kg−1, N=7), indicating that the corals were not depressing AT through calcification. All corals in both trials remained normally colored relative to A. hyacinthus at 10 m depth on the outer reef when the study was completed, although one small colony in trial 1 became pale (Fig. 1C). Overall, there were no signs of tissue loss in either trial, and polyps routinely were seen expanded and feeding at night.
At the start of trial 1, small corals ranged in diameter from 5 to 8 cm, and large corals from 13 to 20 cm; at the start of trial 2, small corals ranged from 5 to 8 cm, and large corals from 16 to 20 cm. At the end of the experiment, the tissue area of trial 1 corals ranged from 93 to 1222 cm2, and for trial 2 from 149 to 1076 cm2. All corals regardless of treatment, tank or trial increased in mass over 23 days, with increments varying from 123 to 1472 mg day−1 in trial 1, and from 59 to 1476 mg day−1 in trial 2 (Fig. 1). Net calcification increased by 1.063 mg day−1 (0.694–1.433 mg day−1, 95% confidence interval, CI) for every cm2 increase in colony surface area at 26.5°C and ambient PCO2, but was unaffected by increases in temperature or PCO2 (Tables 2 and 3, Fig. 1).
After nearly two decades of experimentation, the effects of high PCO2 and temperature on the calcification of reef corals has been tested in numerous studies (Chan and Connolly, 2013; Kornder et al., 2018). The overall outcome of most of these studies is that high PCO2 (relative to ∼400 µatm) depresses net calcification (Chan and Connolly, 2013; Kroeker et al., 2013; Comeau et al., 2014). There are exceptions to this negative trend (Reynaud et al., 2003; Castillo et al., 2014; Strahl et al., 2015; Wall et al., 2017) and indications that its intensity can be attenuated by incubation conditions and nutritional status of the coral (Edmunds, 2011; Dufault et al., 2013; Vogel et al., 2015). In at least two cases, temperature and high PCO2 have been found to act in positive synergy to affect coral calcification, but one of these contrasted temperatures within a normal range for tropical reefs (25.0 versus 28°C; Reynaud et al., 2003) and the other contrasted a naturally occurring temperature with a likely future temperature (26 versus 32°C; Langdon et al., 2018). More commonly, however, temperature and PCO2 have been shown to affect coral calcification independently (Horvath et al., 2016; Okazaki et al., 2017; Kroeker et al., 2013; Anderson et al., 2019). On its own, temperature affects net calcification of corals with a non-linear relationship characterized by a thermal threshold at ∼28°C (Edmunds, 2005; Castillo et al., 2014; Pratchett et al., 2015; Silbiger et al., 2019). Meta-analyses of the effects of high PCO2 (and reduced seawater pH) show negative effects on coral calcification: for 25 studies predating 2011, Chan and Connolly (2013) reported a ∼15% decline for each unit reduction in Ωarag of seawater over the range of 2<Ωarag<4, and for 41 experiments published before 1 January 2012, Kroeker et al. (2013) reported a 32% decline for a ≤0.5 reduction of seawater pH, as is predicted to occur by ∼2100.
Relative to the aforementioned expectations, our previous discovery that net calcification of P. verrucosa ranging in size from 2.0 to 13.1 cm diameter was unaffected by temperature (26.5 versus 29.7°C) and PCO2 (∼400 and 1000 µatm) (Edmunds and Burgess, 2016) was unexpected (cf. Chan and Connolly, 2013; Kroeker et al., 2013; Comeau et al., 2014; Pratchett et al., 2015). Here, we show that a similar outcome resulted from incubating intact colonies of A. hyacinthus ranging in size from 4.6 to 19.9 cm diameter under treatment conditions equivalent to those applied to P. verrucosa (Edmunds and Burgess, 2016). Together, our results make a case for the generalized outcome of our earlier results. For branching corals, net calcification of intact colonies of ecologically relevant sizes is resistant to the effects of high PCO2 interacting with high and a low seawater temperatures within the current seasonal range for the study location.
This conclusion has three important caveats. First, like all experiments testing the effect of temperature and PCO2 on corals, the results are only relevant to the range of treatment conditions and incubation durations employed. While the physical and chemical conditions in our experiment are commensurate with those employed in many other studies of the same topic (e.g. Comeau et al., 2014; Kroeker et al., 2013), from which strong inferences have been made (e.g. Comeau et al., 2014), we do not know whether our results are consistent across a wider range of treatment conditions or over lengthier incubations. Meta-analyses of the biological effects of ocean acidification on calcification, however, downplay the role of incubation duration in affecting the results of the experiment (Kroeker et al., 2013).
Second, while we explicitly sampled the size range of A. hyacinthus colonies found at 10 m depth on the fore reef of Mo'orea when the experiments were conducted, our experiment did not include small fragments or single branches (i.e. nubbins). Instead, we focused on intact colonies. To support the interpretation that it is notable to have null effects of high PCO2 acting across a range of normally occurring temperatures on colonies of A. hyacinthus, we infer that nubbins of this species would have been negatively affected by high PCO2 as in Anderson et al. (2019), who worked with <3 cm diameter fragments of the same species, Comeau et al. (2014) who worked with 5 cm tall branch tips of the congener Acropora pulchra, and the numerous studies compiled in the meta-analyses of Chan and Connolly (2013) and Kroeker et al. (2013).
Finally, as the 95% CIs on net calcification (mg day−1) as a function of colony size (cm2) reveal (Fig. 1), the ability to detect a size-specific treatment effect declines with colony size (i.e. the risks of Type II error increase as the effect size in terms of mg day−1 becomes smaller). A larger sample size (i.e. more colonies) would provide a means to resolve two interpretations of the present results: (1) treatment effects were absent because the corals did not respond to the temperature and PCO2 regimes employed (which we posit is a reasonable assertion given the consensus of contemporary literature), or (2) they were obscured by the risks of Type II error. Increasing sample sizes was not possible, however, because there are ethical objections to collecting numerous large coral colonies, and logistical challenges to accommodating them in tanks. The challenge of tank experiments is ensuring independence among corals while manipulating PCO2 without allowing coral calcification to deplete seawater AT. The ethical objections to much-needed experimentation with large coral colonies are unlikely to change, and greater use of in situ chambers (e.g. Camp et al., 2015) and FOCE technology (e.g. Doo et al., 2019) may be required to address the issues highlighted in the present study.
The common features of the present and our previous work (Edmunds and Burgess, 2016) are the use of intact colonies, collected to reflect the range of colony sizes naturally occurring, and the reliance on net calcification as a response variable. The two taxa we have studied, however, are not morphologically identical, and other aspects of their response to environmental conditions are unlikely to be the same. For example, the contrast of corallum structure between our species might explain why area-normalized calcification declined with colony size in P. verrucosa (Edmunds and Burgess, 2016), but remained ca. 0.96 mg cm−2 day−1 across colonies between 93 and 1222 cm2 for A. hyacinthus. In general, it is quite likely that multiple physiological traits will respond in dissimilar ways to OA and temperature (Kroeker et al., 2013)
Although the cause(s) of the discordant results between the present study and the expected outcomes (cited above) remains unclear, there are several hypotheses that could account for these effects. For example, turbulent flow speed could mediate colony size dependency of physiological performance in branching corals (Edmunds and Burgess, 2018). Additionally, interference among closely spaced branches could affect the flux of metabolites (Patterson, 1982a,b; Burgess et al., 2017), zooplanktivory (Sebens et al., 1997) or the concentration of solutes in interstitial seawater between branches that could determine tissue–surface seawater pH (Chan et al., 2016). As larger colonies of branching corals have lower area-normalized calcification than smaller colonies (Edmunds and Burgess, 2016), and the sensitivity of net calcification to high PCO2 is dependent on the absolute rate of calcification (Comeau et al., 2014), it is also possible that increasing colony sizes has a negative feedback on the sensitivity of calcification to high PCO2. Regardless of the causal basis of the results described herein, they have profound implications for the field of experimental coral biology that is built on a foundation of research heavily dominated by analyses of coral nubbins. Large colonies of branching corals probably are not functionally equal to their detached and isolated branches with respect to their response to high PCO2 and elevated temperature.
We thank S. Ginther for technical support, A. Potter, A. Widrick and S. Zimmermann for field assistance, and the staff of the UC Berkeley Richard B. Gump South Pacific Research Station for making our visits to Mo'orea productive and enjoyable. Research was completed under permits issued by the French Polynesian Government (Délégation à la Recherche) and the Haut-Commissariat de la République en Polynésie Française (DTRT) (Protocole d'Accueil 2017-2018).
Conceptualization: P.J.E., S.C.B.; Methodology: P.J.E., S.C.B.; Formal analysis: S.C.B.; Resources: P.J.E.; Data curation: P.J.E.; Writing - original draft: P.J.E., S.C.B.; Writing - review & editing: P.J.E., S.C.B.; Project administration: P.J.E.; Funding acquisition: P.J.E.
This research was funded by the US National Science Foundation through the Mo'orea Coral Reef (MCR) LTER (OCE 12-36905 and OCE 16-37396) and grants for research on coral biology (to P.J.E.: OCE 14-15268 and OCE 18-29898; and to S.C.B.: OCE 18-29867). This is a product of the MCR-LTER and is contribution number 302 of the Marine Biology Program of California State University, Northridge.
Data are available from the EDI Data Portal: https://doi.org/10.6073/pasta/c4490637abcbbd54899e28c7d4f74ae7
The authors declare no competing or financial interests.