High light alongside elevated P_CO2 alleviates thermal depression of photosynthesis in a hard coral (Pocillopora acuta)

Atmospheric carbon dioxide concentrations are increasing as a result of human activities and the burning of fossil fuels. The oceans have absorbed two-thirds of anthropogenically emitted CO₂, and the resulting perturbation of seawater pH and carbonate chemistry (i.e. ocean acidification) is an unprecedented threat to global marine life (Kleypas et al., 1999). The elevation of the partial pressure of CO₂ in seawater (P_CO2) results in an increase in proton and bicarbonate ion concentrations (Miller and Wheeler, 2012), while decreasing carbonate concentration and the aragonite saturation state (Ω_arag) (Miller and Wheeler, 2012). In coral–algal symbioses, the physiological impacts of elevated P_CO2 have included declines in adult and juvenile calcification (Albright and Langdon, 2011; Cohen and Holcomb, 2009; Zunino et al., 2017), symbiont density (Kaniewska et al., 2012) and lipid biomass (Wall et al., 2017), altered protein metabolism (Edmunds and Wall, 2014) and photophysiology (Iguchi et al., 2012; Noonan and Fabricius, 2016), and compromised sexual reproduction and recruitment (Albright et al., 2010).

Studies of the impacts of elevated P_CO2 on corals have most commonly focused on calcification; however, impacts on other aspects of coral–algal holobiont physiology are often more salient. For instance, a broad survey of physiological responses in the coral Pocillopora acuta demonstrated robustness of calcification under 961 μatm P_CO2 at low- and high-light conditions and found no changes in the density of the intracellular dinoflagellate symbionts (Symbiodiniaceae), total chlorophyll (a+c₂) concentrations or total biomass (Wall et al., 2017). However, declines (of up to 20%) in several forms of energy storage (lipids, protein and total energy content) were observed (Wall et al., 2017). Owing to the central role that the coral–algal symbiosis plays in energy acquisition and carbonate deposition in coral reefs, even moderate impacts on its physiology may result in significant changes to these ecosystems.

Physiological robustness or vulnerability to elevated P_CO2 by itself is noteworthy; however, it is important to recognise that elevated P_CO2 will co-occur alongside other environmental impacts. The impacts of climate change in increasing seawater temperature is well recognised (Hoegh-Guldberg et al., 2014), and the weather conditions that cause heat stress at coral reefs usually also cause very high levels of insolation (Skirving and Guinotte, 2000). However, climate change will have other direct impacts on light levels at reefs. Climate change is predicted to affect cloud cover (Schneider et al., 2019), atmospheric light attenuation (Haywood et al., 2011) and seawater turbidity [via sea level rise (Ogston and Field, 2010) and intensification of both dry spells and precipitation (Fischer and Knutti, 2015)]. Expected decreases in anthropogenic aerosol emissions will also affect irradiance as regulations for reducing air pollution are implemented globally (Rosenfeld et al., 2019; Sato and Suzuki, 2019). These multiple pathways of climate impact will cause irradiance to increase at some reefs and decrease at others, influencing the impact of warming on coral photoinhibition.

Indeed, interactive effects of light with elevated P_CO2, temperature with elevated P_CO2, or light with temperature on coral–algal holobiont physiology have been empirically demonstrated with outcomes that range from antagonistic to synergistic. For instance, Bahr et al. (2016) found that elevated P_CO2 reduced P. damicornis calcification; however, elevated P_CO2 also remedied the negative effect of high temperature on calcification seen at present-day P_CO2 levels (Bahr et al., 2016). Light and temperature also interact to shape symbiont photobiology in hospite (Ban et al., 2014), and light availability influences the susceptibility of coral calcification to P_CO2 effects in juvenile (Dufault et al., 2013) and adult corals (Suggett et al., 2013). Together, these lines of evidence suggest that combinations of light and temperature will mediate how the coral–algal symbiosis will respond physiologically to rising seawater P_CO2.

Owing to the forecast upsurge in mass coral bleaching events under climate change, the combined effects of P_CO2, light and temperature on symbiont densities in corals is of particular interest. Anlauf et al. (2011) documented a synergistic impact of temperature and P_CO2 on symbiont densities in new coral recruits, wherein elevated P_CO2 ameliorated the decrease in symbionts caused by high temperature stress. Elevated P_CO2 synergistically lowered coral luminance (a proxy of bleaching) in two coral species under naturally high irradiances (ca. 1000 µmol photons m⁻² s⁻¹) (Anthony et al., 2008; but see Wall et al., 2014). Meta-analysis has shown that, in general, symbiont densities in corals decline under elevated P_CO2, but there is not yet broad evidence for an interaction between high temperature and elevated P_CO2 in causing symbiont loss (Mason, 2018b).

Here, we explore the impact of simultaneous alteration of P_CO2, light and temperature on a pocilloporid coral (P. acuta) in a factorial design to assess the influence of ecologically relevant scenarios for multiple stressors on performance of the coral–algal holobiont. We measured a range of physiological parameters on which P_CO2, light and temperature are thought to have a major influence in the current understanding of coral–algal holobiont physiology (mechanisms summarised in Table 1): symbiont cell densities, chlorophyll a (chl a) concentrations, respiration rates, light-enhanced dark respiration, photosynthesis, photosynthesis to respiration (P/R) ratio and symbiont to host (S/H) cell ratio. Given the known or theoretical impacts of each stressor in isolation (Table 1), we hypothesised that the combined stressors of elevated P_CO2, irradiance differences and elevated temperature will have an interactive effect on each aspect of physiology of the P. acuta holobiont. For symbiont densities, where single-stressor effects are the most clear-cut (Table 1), we hypothesised that this variable will see its most severe decline under the combination of high light, elevated P_CO2 and high temperature.

Whole colonies of Pocillopora acuta (Lamarck 1816) were collected on 13 and 29 October 2014 (n=5 and 3 colonies, respectively) at 0.2–1.5 m depth, from the windward reef of Moku o Lo‘e in Kāne‘ohe Bay, O‘ahu, HI, USA. Colonies were identified as P. acuta through the prominent pigmented rings around polyps and acute branch tips that distinguish this species from the morphologically similar P. damicornis (Schmidt-Roach et al., 2014). Collections were performed in accordance with permitting guidelines of the Hawai‘i Department of Land and Natural Resources Division of Aquatic Resources under Special Activity Permit 2015–8. At the Hawai‘i Institute of Marine Biology, colonies were fragmented into nubbins (<5 cm), mounted onto bases (as per Wall et al., 2017) and maintained for 21–37 days in outdoor aquaria under ∼60% shade cloth (daily peak of 200–300 µmol photons m⁻² s⁻¹: Putnam et al., 2016) in flow-through seawater pumped from Kāne‘ohe Bay and cooled to 26.05±0.01°C (mean±s.e.m., n=4869). Nubbins were then distributed among 24 indoor tanks (n=8 nubbins per tank with one from each colony). This sample size exceeds that used in most coral tank experiments (3 to 6 per tank), achieving a simulated power of 80% in the detection of a three-factor interaction on symbiont densities at an effect size of ca. −0.36×10⁶ cells cm⁻² (see online code). Nubbins were maintained at 25.78±0.04°C (n=168) under two light regimes (described below) for a tank acclimation period of 25 days prior to the commencement of the experiment. During this period (21 November 2014 to 15 December 2014), the differential between the ambient P_CO2 and the high P_CO2 treatment tanks was partly introduced (Fig. S1). Each tank also contained 14 additional P. acuta nubbins for other studies that were removed on day 32 of the experiment. Corals were not given supplemental food, but dietary items (dissolved organic matter, microbes and plankton <100 µm in size) were present in the flow-through seawater.

CO₂ treatments (431 and 962 µmol P_CO2) reflect present-day P_CO2 and end-of-century ocean acidification under Representative Concentration Pathway 8.5 (Meinshausen et al., 2011). The high light treatment (15.7 mol photons m⁻² day⁻¹) sits within the range of insolation (7-day moving average) seen during the 2014–2015 December–January–February (DJF) period at Moku o Lo‘e at 2 m depth. The low light treatment (7.5 mol photons m⁻² day⁻¹) reflects lower light conditions seen in the more turbid northern Kāne‘ohe Bay during the 2014–2015 DJF period (Cunning et al., 2016). Temperature treatments (25°C and 29°C) reflect a non-stressful sea temperature (near the DJF climatological mean sea surface temperature) and a stressful temperature (2°C in excess of the maximum monthly mean) for the Hawaiian Islands (NOAA Coral Reef Watch, 2019).

Our exposure of P. acuta used a staged introduction of multiple stressors. Four treatments, consisting of two P_CO2 levels crossed with two light levels, were administered for 39 days (16 December 2014 to 23 January 2015). During the final 7 days of the experiment, each treatment was divided into two temperature levels (design illustrated in Fig. S2). Twelve 44.8 litre tanks received 400 µmol photons m⁻² s⁻¹ (LL) at the midday light maximum and 12 received 800 µmol photons m⁻² s⁻¹ (HL) at the midday light maximum (Aqua Illuminations Sol Light Emitting Diode lamps; C2 Development Inc., Ames, IA, USA). The photoperiod (12 h:12 h light:dark) consisted of sunrise at 05:00 h, 5 h of linear increase in light intensity, stable peak light from 10:00 to 12:00 h, 5 h of linear decrease in light intensity, and sunset at 17:00 h. During the first 4 days of the acclimation period, a longer duration of peak light during the daytime was used (07:00–15:00 h), with possible paling of tissue observed in the high light treatment. Each tank received flow-through seawater at a mean±s.e.m. of 431±8 µatm P_CO2 (ambient/low, LCO₂) or 962±27 µatm P_CO2 (high, HCO₂) (Table 2), at a flow rate of ca. 1.5 l min⁻¹. The procedures used for modifying P_CO2 levels and for biweekly measurements of tank carbonate chemistry are reported in Wall et al. (2017). During the first stage (32 days), corals in each CO₂×light treatment (n=6 tanks treatment⁻¹) were maintained at seasonally ambient conditions of 24.59±0.06°C (mean±s.e.m., n=153). For the final 7 days, each CO₂×light treatment was split, with three tanks kept at 24.59±0.06°C (n=11; LT) and three tanks raised to 29.05±0.22°C (n=11; HT), using immersion and in-line aquarium heaters.

Alkalinity measurements performed on experimental treatment water at intervals over one 24 h period indicated diel variations in P_CO2 of ca. 150 and 400 µatm in the ambient P_CO2 and elevated P_CO2 treatments, respectively, with minima during midday (Fig. S3). As experimental tanks were downstream from the header sumps (where P_CO2 was controlled) and the residence time (duration to turnover 99% of water; Escobal, 1996) in each tank was moderate (137 min), diel P_CO2 fluctuations in each tank were likely a result of net respiration at night and net photosynthesis during the day. This compares with diel fluctuations of 200–300 µatm P_CO2 on the Kāne‘ohe Bay barrier reef (Shamberger et al., 2011).

Respirometry measurements were performed on four nubbins per tank on days 36 and 37 (non-heated treatments) and days 38 and 39 (heated treatments), in two magnetically stirred 260 ml cylindrical Perspex chambers containing filtered seawater (0.25 µm) from the P_CO2 treatment relevant to each nubbin. To maintain the relevant temperature, chambers were controlled to 24.77±0.03°C (n=45) or 29.68±0.02°C (n=44) via immersion in a water bath. Following a period of low light acclimation (30 min at <10 µmol photons m⁻² s⁻¹), each nubbin was incubated for ∼10 min in darkness (measuring dark respiration, DR), illuminated for ∼10 min at the midday light level of the relevant treatment (measuring daytime net photosynthesis, P_net), exposed to darkness again for ∼10 min (measuring light-enhanced dark respiration, LEDR) and then repatriated to its original treatment tank. Intra-chamber oxygen concentrations were determined at 1 s logging intervals with a Fospor fibre optic probe on a Neofox Oxygen Sensing System (Ocean Optics, Dunedin, FL, USA). Gross photosynthesis under midday light (P_gross) was calculated as P_net–LEDR (assuming no respiration of energy acquired via heterotrophy, and that LEDR approximates light respiration) (but see Stokes et al., 1990). Photosynthesis and respiration measurements were normalised to coral surface area. The ratio of photosynthesis to respiration (P/R ratio) was calculated by dividing P_net by DR.

On the final day of the experiment, nubbins were snap-frozen in liquid nitrogen and stored at −80°C. Tissue was airbrushed off thawed nubbins with a pressurised stream of buffer at 4°C (0.4 mol l⁻¹ NaCl, 0.05 mol l⁻¹ EDTA) and the slurry was processed for ∼10 s with a T25 500 W variable speed homogeniser (IKA Works Inc., Wilmington, NC, USA). The presence of planula larvae (planulae), visible in the coral tissue as it was being removed, was noted when observed. Symbiont cells were pelleted (1 ml of slurry centrifuged at 4500 g for 5 min), extracted in 100% acetone (48 h at −20°C), and chl a was determined as per Jeffrey and Humphrey (1975). Chl a was normalised to coral surface area (chl a density) and to symbiont cell count (chl a per symbiont). Symbiont densities were determined by counts in a Neubauer haemocytometer chamber (n=2–4) (Reichert Inc., Buffalo, NY, USA) and normalised to coral surface area. Coral surface areas were estimated by double wax dipping (Vytopil and Willis, 2001).

DNA was extracted from 300 µl of tissue slurry using a CTAB–chloroform protocol (dx.doi.org/10.17504/protocols.io.dyq7vv). Symbiont to host (S/H) cell ratios were determined by quantitative PCR (qPCR) on a StepOnePlus Real-Time PCR machine (Applied Biosystems, Foster City, CA, USA). We used taxon-specific primers that separately target the actin gene of dinoflagellate symbionts within Symbiodiniaceae clade C (genus Cladocopium) and clade D (genus Durusdinium) and P. acuta (primers and qPCR protocol as per Cunning and Baker, 2013). Clades C and D are the only clades typically detected in the genus Pocillopora (Cunning et al., 2013; Pochon et al., 2010; Stat et al., 2008, 2015; M. Stat, personal communication). qPCR reactions were excluded if they had a cycle threshold (C_T) within three cycles of the C_T (if any) observed in the respective negative control. We corrected the S/H cell ratios with the symbiont to host DNA extraction efficiency ratio for CTAB–chloroform extraction, 0.828 (Cunning and Baker, 2013). We assumed the same actin copy number (three) as measured in P. damicornis by these authors owing to its close taxonomic affinity to P. acuta. Actin gene copy numbers for clades C and D were estimated using a singular value decomposition method (Cunning et al., 2016).

Differences between the two CO₂ treatments for carbonate chemistry variables (P_CO2, pH_T and total alkalinity) were assessed via t-tests. To explore tank effects on these variables, one-way ANOVAs were performed to compare all tanks within each of the two CO₂ treatments. Analyses of physiological responses (symbiont density, chl a per symbiont, chl a density, P_net, P_gross, DR, LEDR, P/R and S/H ratios) were performed via mixed-effects models in the package lme4 (Bates et al., 2015) in R (https://www.r-project.org/). Temperature, light and P_CO2 were designated as fixed effects; tank and colony were designated as random effects, and random effects were retained or dropped according to the function {step} (lmerTest). Normality of distribution was checked using quantile–quantile plots and transforms were applied if needed, and homoscedasticity was checked using plots of residuals versus fitted values. Type III ANOVAs were performed in the package lmerTest (Kuznetsova et al., 2017). S/H cell ratio was log-transformed prior to analysis (Cunning and Baker, 2013). Post hoc tests of significant interactive effects were performed using least-square means with the Tukey method for P-value adjustment.

Seawater carbonate chemistry was effectively controlled to deliver consistent conditions at each P_CO2 level (Table 2). Mechanical issues in one HCO₂–HL–HT tank and one HCO₂–HL–LT tank towards the end of the experiment led to the a priori removal of these tanks and the nubbins within from further analyses, leading to final replication of two tanks per treatment for these treatments and three for all other treatments. The two carbonate chemistry treatments differed in P_CO2 (P<0.001) and pH_T conditions (P<0.001). Total alkalinity was not affected by CO₂ treatment (P=0.147). P_CO2 and pH_T did not differ among replicate tanks in the HCO₂ treatment (P≥0.668). At ambient CO₂, P_CO2 and pH_T differed between two tanks (tank 14: 376 μatm P_CO2 and pH 8.05; tank 21: 498 μatm P_CO2 and pH 7.95) (one-way ANOVA P≤ 0.026, post hoc P=0.036). However, these differences were within an acceptable range for the LCO₂ treatment and no pairwise difference in P_CO2 or pH_T was found between either tank and any other LCO₂ tanks (post hoc P≥0.064), nor between any other pairs of LCO₂ tanks (post hoc P≥0.264).

Symbiont density was not affected by P_CO2 (F_1,21=0.40, P=0.536), but declined by 23 and 25% at high light (F_1,21=18.20, P<0.001) and high temperature (F_1,22=30.69, P<0.001), respectively (Tables 5 and 6). Symbiont density experienced no interactive effect among any factors (F_1,21≤3.78, P≥0.065) (Table 3). Chl a per symbiont did not change under high P_CO2 (F_1,14=0.09, P=0.763), decreased by 17% at high light (F_1,14=21.82, P<0.001) (Table 5) and increased by 24% at high temperature (F_1,14=28.24, P<0.001) (Table 6). Chl a per symbiont was not affected by an interaction among any factors (F_1,14≤0.51, P≥0.485) (Table 3). Chl a density declined by 36% at high light (F_1,20=47.71, P<0.001) (Table 5) but was not affected by P_CO2 (F_1,20=0.68, P=0.418), temperature (F_1,20=3.02, P=0.097) nor any interactive effects (F_1,20≤3.59, P≥0.073) (Table 3).

A significant interactive effect among all three factors was found for P_net (F_1,80=5.26, P=0.024). Post hoc tests indicated that high temperature caused declines in P_net at LCO₂–HL (of 40%) and HCO₂–LL (62%), but not at HCO₂–HL or LCO₂–LL (Fig. 1). Whilst main effects are difficult to interpret under a significant interaction, a significant main effect of temperature (F_1,80=31.53, P<0.001, decline of 41% at HT) (Table 6) but not light (F_1,80=0.39, P=0.536) or CO₂ (F_1,80=1.04, P=0.311) was found, and no two-factor interactions were detected (F_1,80≤1.85, P≥0.178) (Table 3).

A significant three-factor interaction among light, temperature and CO₂ affected P_gross (F_1,80=4.08, P=0.047). Post hoc tests found that high temperature depressed P_gross by 34% at HCO₂–LL, but not at LCO₂–LL, LCO₂–HL or HCO₂–HL (Fig. 2). For the main effects, P_gross increased by 12% at high P_CO2 (F_1,80=4.78, P=0.032) (Table 4) but did not change under high light (F_1,80=1.44, P=0.233) and decreased by 20% under high temperature (F_1,80=12.72, P<0.001) (Table 6). No two-factor interactions were detected (F_1,80≤1.79, P≥0.184) (Table 3).

Dark respiration was not affected by P_CO2 (F_1,76=1.81, P=0.183) or light (F_1,76=1.21, P=0.276), but increased by 24% at high temperature (F_1,76=6.86, P=0.011) (Table 6). No significant two- or three-factor interactions were found (F_1,76≤2.03, P≥0.158) (Table 3). LEDR increased by 15% at high P_CO2 (F_1,76=5.05, P=0.028) (Table 4), was not affected by light (F_1,76=1.34, P=0.250), and increased by 21% at high temperature (F_1,76=6.63, P=0.012) (Table 6). No significant interactive effects among any factors affected LEDR (F_1,76≤1.40, P≥0.241) (Table 3).

The P/R ratio was not affected by P_CO2 (F_1,76=0.01, P=0.905) or light (F_1,76=0.24, P=0.629), and decreased by 56% at high temperature (F_1,76=26.47, P<0.001) (Table 6). No two- or three-factor interactions affected P/R ratio (F_1,76≤3.25, P≥0.075) (Table 3).

The S/H cell ratio was not affected by P_CO2 (F_1,149=1×10⁻⁴, P=0.991) nor light (F_1,149=4×10⁻⁴, P=0.984), but increased by 65% at high temperature (F_1,149=6.31, P=0.013) (Table 6). No two- or three-factor interactions affected S/H cell ratio (F_1,149≤1.71, P≥0.192) (Table 3).

Though not statistically significant, S/H cell ratio values at HCO₂–LL–HT (0.16±0.04, mean±s.e.m.) and LCO₂–HL–HT (0.15±0.03) were higher compared with those at LCO₂–LL–HT (0.10±0.03) and HCO₂–HL–HT (0.11±0.04). All treatments at 25°C had mean S/H cell ratios between 0.06 and 0.09.

In order to project how climate change will influence coral reef ecosystems, it is necessary to determine how multiple environmental parameters interact to attenuate and exacerbate climate change stressors. However, experimental explorations of such interactions are faced with the challenge that elevated P_CO2 owing to human activities is a chronic stressor, whereas high temperature is an acute phenomenon. As impacts of elevated P_CO2 may take some time to manifest (Anthony et al., 2008), the application of heat stress as a chronic stressor could obscure any effect of elevated P_CO2 (e.g. by continually exerting a strong downward pressure on symbiont population size). The present study utilised an experimental design for the analysis of the combined impacts of elevated P_CO2, light and temperature that provided time for physiological changes to elevated P_CO2 and light to occur (over 32 days) before heat stress was applied (additional 7 days), thus providing the proper context of elevated P_CO2 and heat as chronic and acute impacts, respectively.

A significant three-way interaction between elevated P_CO2, light and temperature was detected for net and gross photosynthesis. For P_net, declines were detected at 29°C versus 25°C within the P_CO2–light combinations of LCO₂–HL and HCO₂–LL. Circumstantial evidence suggests that elevated S/H cell ratios may have been linked to depressed P_net. S/H cell ratio was elevated at 29°C versus 25°C (significant main effect); however, at LCO₂–HL and HCO₂–LL, this difference in S/H cell ratio was far more pronounced, 74–141% (trend only), compared with 18–24% in the other P_CO2–light combinations. As the three-factor interaction for S/H cell ratio was not statistically significant, further work will be required to investigate this link.

For P_gross, a decline at 29°C versus 25°C was supported for HCO₂–LL but not LCO₂–HL or other P_CO2–light combinations. Several explanations exist for why P_net and P_gross differed with respect to the LCO₂–HL treatment. Firstly, differences in light respiration among treatments, if any, may introduce real differences between P_net and P_gross. Secondly, LEDR, used here as a proxy for light respiration to calculate P_gross, may underestimate light respiration rates (Schrameyer et al., 2014). Finally, P_gross, as calculated here, contains measurement error from both P_net and LEDR and this added variation may have served to reduce the statistical power of the post hoc tests.

A consideration of how the experimental treatments may limit the light and dark reactions of photosynthesis could help to explain the interactive effects observed for P_net and P_gross. Increases in temperature are known to decrease the capacity for photosynthetic electron transport, such as by damaging the D1 protein (Warner et al., 1999). Under high light conditions, this can lead to overreduction of the remaining intact electron transport chains and to further damage through the production of oxygen radicals (Hoegh-Guldberg, 1999), decreasing the areal rate of photosynthesis (as was seen for P_net at elevated temperature and ambient P_CO2). However, under high substrate availability (elevated P_CO2), an enhanced capacity for Calvin–Benson activity may help to keep electron transport chains sufficiently open to cope with high light influx, offsetting the inhibiting effect of high temperature. Enhanced turnover rate of the Calvin–Benson cycle at moderately elevated temperature (Dusenge et al., 2018), and/or faster repair of damaged D1 proteins owing to an improved energetic status of the organism, might also contribute (Hoogenboom et al., 2012). This interpretation offers an explanation for why photosynthesis (both P_net and P_gross) increases with irradiance under elevated temperature combined with elevated P_CO2. It is likely that these effects would only occur within a range of elevated light that is sufficient to deliver adequate reducing power, but still low enough to avoid substantive photodamage. As a case in point, under very high light conditions, ocean acidification and warming were found to synergistically decrease net photosynthetic productivity in Porites lobata by Anthony et al. (2008).

In previous studies involving fully crossed levels of P_CO2 and temperature, the effect of these variables on photosynthesis was independent in some coral species but interactive in others. In Seriatopora hystrix, P_net and P_gross decreased under high temperature and increased under high P_CO2, with both stressors having an additive, but not interactive, effect (Noonan and Fabricius, 2016). In Acropora millepora, no interactive effect of temperature and acidification nor of light (at low levels) and acidification on P_net and P_gross has been found (Noonan and Fabricius, 2016; Vogel et al., 2015), but P_gross may decrease under acidification at high light (Kaniewska et al., 2012). In Stylophora pistillata, P_net increased under high temperature and decreased under high CO₂, but no interaction between the two factors was found (Reynaud et al., 2003). In Acropora intermediata and P. lobata, net productivity decreased over a gradient of low to high P_CO2 at ambient temperature, with warming interactively altering this pattern by effecting an increase in net productivity at intermediate P_CO2 in A. intermediata and exacerbating the size of the net productivity decline in P. lobata (Anthony et al., 2008). P_gross increased across a range of coral species at a natural CO₂ seep by 23–37% (Biscéré et al., 2019), moderately more than the increase in P_gross under acidification (12%) observed as a main effect in the present study. Physiological differences such as in carbon-concentrating mechanisms or symbiont photophysiology may account for inter-specific differences in P_net and P_gross responses (Brading et al., 2011).

Multiple stressors can have a compound effect on corals through interactive effects, where present, or through additive impacts where each stressor exhibits a main effect. The latter was seen with chl a per symbiont (discussed later) and LEDR. LEDR increased by 15% under elevated P_CO2, and increased by 21% at elevated temperature. Increased LEDR can relate to several forms of photochemical quenching. Increased rates of respiration can occur as a result of increased Calvin–Benson cycle activity or production of ATP and NADPH by thylakoid electron transport under illumination (Parys and Jastrzębski, 2006). LEDR may also indicate oxygen consumption through the alternate photochemical quenching mechanisms of chlororespiration and the Mehler ascorbate–peroxidase cycle, or re-oxidation of spent electron carriers (Roberty et al., 2014; Stokes et al., 1990). Significantly, in a prior study of Acropora muricata, LEDR increased under acidification whilst DR remained constant, and evidence from xanthophyll cycling suggested that increased Mehler ascorbate–peroxidase cycle activity could have been the cause (Crawley et al., 2010). This may also occur in the P. acuta holobiont; however, the increase in P_gross under acidification leaves open the possibility of increased photosynthetic substrate as a cause of elevated LEDR. As LEDR increased under elevated temperature, likely owing to the increased kinetic rate of metabolic reactions (Gillooly et al., 2001), an additive impact of increased temperature and elevated P_CO2 on LEDR in some coral–algal holobionts is to be expected under future global change.

The additive impact on LEDR of increased temperature and P_CO2 may lead to energetic consequences for the P. acuta holobiont. In the case of the latter stressor, Wall et al. (2017) identified that lipids and energy content (measured per unit biomass and thus robust to the effects of skeletal extension) are depleted in P. acuta under acidification at two light levels. Depletion of energy reserves could be due to decreased availability of heterotrophically derived energy, or to an increased rate of energy usage, which, in most cases, will manifest as an increased rate of respiration. Our finding of increased LEDR under elevated P_CO2 suggests a greater rate of metabolism of photosynthates immediately after their production, and thus could provide a plausible hypothesis for the decreased energy reserves of P. acuta observed by Wall et al. (2017).

Maintenance of calcification rate may explain the lack of decline in DR and the increase in LEDR under elevated P_CO2. The lack of decline in DR [also observed in P. damicornis (Comeau et al., 2017)], and its increase under elevated P_CO2 at 29°C (Fig. 2) is notable as the increased seawater proton concentration under elevated P_CO2 could make respiration more energetically costly (Mackey et al., 2015). Pocillopora damicornis and P. acuta maintain their rate of calcification, normalised to both biomass and skeletal area, and experience declines in biomass or energy reserves, under elevated P_CO2 (Comeau et al., 2013; Comeau et al., 2014a,b; Wall et al., 2017). This could be a signature of the energetic cost of the maintenance of calcification under elevated P_CO2 (Comeau et al., 2013; Galli and Solidoro, 2018) and/or the increased energetic costs of respiration. In free-swimming larvae (i.e. a non-calcifying life stage) of Pocillopora spp., respiration rate displays a trend to decrease under elevated P_CO2 (Putnam et al., 2013), supporting the idea that the metabolic requirements of maintaining calcification under acidification contributes to stability in DR in adults.

Other vital processes may have also contributed to a changed energetic balance at high P_CO2. Incipient planula larvae were observed within the adult tissue of many coral specimens at the end of this study. We were unable to determine from these observations whether P_CO2 level influenced the quantity of incipient larvae within coral tissues, the number of larvae released over the course of the experiment or their maturation time. However, their presence in this study suggests that gametogensis warrants investigation as a process that may potentially effect or be affected by changes in adult coral condition under elevated P_CO2.

Our hypothesis that symbiont density loss would be greatest at the combination of elevated P_CO2, high temperature and high light was not fully supported by the data. Symbiont density loss was the greatest under high light combined with high temperature, with statistical support, and the two stressors together had an additive but not interactive effect on this variable (also see Ban et al., 2014). Unlike light and temperature, elevated P_CO2 had no statistically significant impact on symbiont cells. However, a decline in symbiont densities under elevated P_CO2 has been shown in a meta-analysis across multiple studies and coral species (Mason, 2018b), indicating that an effect of P_CO2 on symbiont densities is often present. In our study, an effect of P_CO2 on symbiont densities may have required a greater period of incubation at high P_CO2 or may have been moderated by our particular combination of coral and symbiont species.

Of the three stressors tested (elevated P_CO2, light and temperature), changes in temperature exerted the greatest number of main effects, significantly affecting almost every response variable. The ratio of P_net under midday irradiance to dark respiration declined to 1.01 at 29°C, which could suggest that phototrophy may not meet energy expenditure at 29°C if integrated over a full day–night cycle. The decrease in symbiont densities with an increase in temperature (from 25°C to 29°C) was accompanied by an increase in chl a content per symbiont cell. Increased chl a per symbiont under stressful temperature has been observed in some coral species [Seriatopora caliendrum (Baghdasarian et al., 2017), S. hystrix (Hoegh-Guldberg and Smith, 1989) and the temperate coral Balanophyllia europaea (Caroselli et al., 2015)], and likely indicates a photoacclimation response to decreased symbiont densities. In constrast to temperature, high light caused a decrease in both symbiont cell density and chl a per symbiont, possibly indicating some form of light-induced photosynthetic dysfunction, or rapid linear extension and concomitant thinning of symbiont densities and chl a per symbiont.

Whilst symbiont density decreased at high temperature, the S/H cell ratio increased. Decreasing symbiont density and increasing S/H cell ratio together suggest that host cell density per cm² decreased at elevated temperature, but at a rate greater than the loss of symbiont cells. This pattern is consistent with past observations of physiological responses of coral–algal holobionts to seasonal warming, which is thought to cause host cell loss, increasing the S/H ratio but also triggering a regulatory downwards adjustment in the symbiont population (Cunning and Baker, 2013). An alternative explanation, that S/H ratio increases owing to rapid linear extension and consequent thinning of symbiont cells, is unlikely as 29°C is above the calcification optima of Kāne‘ohe Bay corals (Jokiel and Coles, 1977). Increasing S/H cell ratios confirm the nuance revealed by chl a per symbiont that the symbiont density loss at high temperature is not yet indicative of a state of coral bleaching, though loss of host cells is a sign that bleaching may be imminent (Ainsworth et al., 2008).

It is clear that under future climates, where temperature and acidification have both already increased, the way that light is affected by climate change will prove to be important to P. acuta, impacting the degree to which warming will decrease photosynthetic rate. Light can decrease under climate change through sea level rise, which may cause increased tidal resuspension of sediments (Ogston and Field, 2010), through increased turbidity owing to stronger precipitation events (Fischer and Knutti, 2015), and through increased atmospheric water vapour content, which decreases light transmission (Haywood et al., 2011). In those cases, P. acuta will clearly not experience the alleviation of temperature-induced photosynthesis depression. In contrast, cloud cover could decrease in many areas owing to warming (McCoy et al., 2017; Schneider et al., 2019), and future reductions in aerosol emissions will tend to increase light penetration and reduce cloud cover (Rosenfeld et al., 2019; Sato and Suzuki, 2019). Changes in high cirrus clouds that trap heat but reduce irradiance could also be particularly important (Kärcher, 2017; Liou, 1986). We note that 800 µmol photons m⁻² s⁻¹ is relatively strong irradiance but still within the moderate range of light that reefs experience, and future increases in irradiance will only be beneficial to the extent that they move P. acuta into this range, but not beyond it (to photoinhibitive light levels).

Coral species/genotypes and their partner symbionts vary in their temperature tolerance and photophysiology, such as non-photochemical quenching and carbon concentrating mechanisms. As such, it is difficult to infer how general these responses would be among other coral species/symbiont combinations. However, increase in photosynthesis under acidification has been observed in some other species (Biscéré et al., 2019; but not all, see Sweet and Brown, 2016), and a decrease in photosynthesis at stressful temperatures is a common phenomenon (e.g. Hoegh-Guldberg, 1999). Therefore, whilst demonstrated in only one species here, a degree of commonality among corals in displaying the alleviation of thermally depressed photosynthesis by strong light combined with high P_CO2 is an intriguing possibility.

Owing to the roles of light and CO₂ as (respectively) power source and substrate for photosynthesis, and the role of temperature in optimising or disrupting this process, the influences of P_CO2, light and temperature on marine photosynthesisers remains a research imperative under climate change. Of the three factors, temperature exerted the dominant influence on the physiology of the P. acuta–Symbiodiniaceae symbiosis. Importantly, these effects were observed at a heat stress level that had not caused bleaching, and hence may reflect physiological responses to moderate but year-round elevations in temperature under climate change. Interactions among all three factors were detected for some key physiological processes. Elevated P_CO2 caused an increase in photosynthesis in P. acuta (as found in some other coral species), whilst stressful temperature caused a decrease. However, our results demonstrate that light can alter this impact: 800 µmol photons m⁻² s⁻¹ at high (but not ambient) P_CO2 ameliorated the decline in photosynthesis caused by high temperature, an effect that was clearly not seen at the lower light level. This is a significant finding, illustrating that the well-known negative impacts of high temperature on photosynthesis can be alleviated by moderately strong irradiance if accompanied by elevated seawater P_CO2.

Changes in energetic status at the organism level could have flow over effects to the broader ecosystem. Coral–algal holobionts supply vast quantities of energy to other levels of the ecosystem, through the release of mucus (Crossland et al., 1980) and dissolved organic carbon (Crossland, 1987), and corallivory (Cole et al., 2011). The changes in the rate of photosynthesis but also of metabolism of energy (LEDR) at elevated P_CO2 observed in this study indicates the possibility of changes in the release of energy through these other pathways. Measurement of mucus release, dissolved organic carbon release and rates of corallivory under ocean acidification will be useful to ascertain the impacts of these holobiont-level changes on the coral reef ecosystem.

We thank D. Schar, J. Dulin, H. Putnam, R. Ritson-Williams, J. Davidson, B. Lenz, and W. Skirving for constructive discussions and logistical support, and G. Davies for performing DNA extractions and qPCR runs. Portions of this research formed a part of the PhD thesis of R.A.B.M. (Mason, 2018a). This is Hawai‘i Institute of Marine Biology (HIMB) contribution number 1822 and School of Ocean and Earth Science and Technology (SOEST) contribution number 11136.