Effects of elevated temperature and pCO2 on the respiration, biomineralization and photophysiology of the giant clam Tridacna maxima

In this article, we mimicked current and future warm season conditions and showed that temperature and pCO2 predicted for the end of this century are forcing variables significantly affecting Tridacna maxima’s growth and photophysiology in an order of magnitude that will jeopardize its survival in French Polynesian lagoon.


Introduction
Anthropocene era is characterized by human activities releasing gigatons of CO 2 in the atmosphere and contributing to induce global climate change (Broeker et al., 1979;Caldeira and Wickett, 2003;Sabine et al., 2004;Zeebe et al., 2008;IPCC, 2014). The transfer of CO 2 from the atmosphere to the ocean results consequently to the increase of dissolved CO 2 in the seawater inducing significant pH decreases. On the other hand, due to the 'greenhouse gas' property of CO 2 , terrestrial and sea surface temperature rise constantly (i.e. global warming). According to the last Intergovernmental Panel on Climate Change report (IPCC, 2019), warming had already reached +1 • C by comparison to the pre-industrial period and may reach +1.5 • C around 2040 if the current rate of emission is maintained. Enrichment of dissolved CO 2 in the ocean modifies the carbonate chemistry by decreasing carbonate ion concentration ([CO 3 2− ]) and releasing protons that respectively decrease calcium carbonate saturation state (Ω) and seawater pH, i.e. ocean acidification (Kleypas et al., 1999;Caldeira and Wickett, 2003). Indeed, pH had already decreased by 0.1 pH unit and this drop may reach −0.3 by 2100 (according to the RCP (Representative Concentration Pathway) 8.5 scenario; IPCC, 2019). CO 3 2− is a carbonate ion form involved in biologically controlled calcification process, i.e. biomineralization (Orr et al., 2005) of many marine organisms such as scleractinian corals and molluscs constructing calcium carbonate structures (e.g. exoskeleton, shell, test, spicule). This characteristic of their physiology make them highly sensitive to ocean acidification (Hoegh-Guldberg et al., 2007;Ries et al., 2009;Kroeker et al., 2013;Gazeau et al., 2013;Parker et al., 2013), especially when this skeleton is made of aragonite, which is more sensitive to dissolution than calcite (Morse et al., 2007). Among marine calcifying molluscs (e.g. oysters, mussels, giant clams, abalones, limpets) negative impacts of temperature and pCO 2 have been demonstrated on the survival, growth, biomineralization processes and other key physiological functions on different stages of their life cycle (Bougrier et al., 1995;Kurihara, 2008;McClintock et al., 2009;Gazeau et al., 2010;Welladsen et al., 2010;Melzner et al., 2011;Rodolfo-Metalpa et al., 2011;Schwartzmann et al., 2011;Talmage and Gobler, 2011;Liu et al., 2012;Watson et al., 2012;Kurihara et al., 2013;Fitzer et al., 2014;Le Moullac et al., 2016a, b;Meng et al., 2018;Wessel et al., 2018;Avignon et al., 2020). However, effects of each stressor and their potential synergetic effect on giant clam's physiology are still poorly understood.
Despite their ecological and socio-economic importance, effects of thermal stress and acidification on giant clams still need to be investigated. Thermal stress has shown to decrease fertilization success in T. maxima (Armstrong et al., 2020), decrease oxygen production in Tridacna derasa and Tridacna gigas as well as respiration rates in T. derasa species (Blidberg et al., 2000). In contrast, an increase of respiration with a high photosynthetic rate in the holobiont was reported for Tridacna squamosa (Elfwing et al., 2001). Thermal stress also reduces the abundance of symbionts in T. gigas (Leggat et al., 2003) and Tridacna crocea (Zhou et al., 2019) and a decrease of photosynthate export from symbionts to the host (Leggat et al., 2003). In T. maxima, heat stress was shown to induce changes in the fatty acid composition, lipid pathways and the overexpression of genes encoding reactive oxygen species (ROS) scavengers (Dubousquet et al., 2016). In symbionts hosted in T. gigas, high-light levels and heat stress caused a decrease of symbiont density, cell size and chlorophyll content (Buck et al., 2002). Toonen et al. (2012) demonstrated that low pH combined with high nutrient concentration have different impacts on shell growth rate depending on the Tridacna species. Moreover, Watson (2015) showed that highlight irradiance condition (i.e. high photosynthetically active radiation level) may limit the impact of pCO 2 on shell growth and total animal mass gain. Regarding the potential synergetic effect of temperature and pCO 2 parameters, the only study carried out showed that the survival of T. squamosa juveniles may decrease by ocean warming and acidification (Watson et al., 2012). To our knowledge, no study had investigated the impacts of both parameters, and their potential synergetic effect, on several key physiological parameters of both T. maxima host and its symbionts. To fill this gap, our experimental approach was designed to better understand the physiological mechanisms underlying the response of giant clams to global climate changes.

Conservation Physiology • Volume 9 2021
Research article Total alkalinity (A T ) is given in mean (±SD) based on weekly measurements for each experimental tank and for each condition. pCO 2 , Ω aragonite and dissolved inorganic carbon (DIC) were calculated using the CO 2 SYS software.
Based on a 65-day crossed-factorial experiment, we investigated physiological responses of 4-year-old T. maxima to temperature and pCO 2 conditions in the French Polynesian lagoons during the present warmer season and those predicted for 2100 by the IPCC 2014: +1.5 • C (RCP 4.5 scenario) and +800 μatm of CO 2 (RCP 8.5 scenario). The effect of each parameter and their potential synergetic effect were evaluated based on respiration, biomineralization and photophysiology by analyzing the holobiont O 2 production and respiration, growth rate, ultrastructure of the shell, symbiont density and photosynthetic yield. In addition, the kinetics of thermal and acidification stress were assessed by performing analyses at different time of exposure (i.e. 29, 41, 53, 65 days).

Biological material
Two hundred juvenile T. maxima (4-year-old, ∼5-6 cm height), with brownish/dark-green colour, were collected from the cultivated stock in Reao lagoon (Tuamotu islands). Then, they were exported to the Centre Ifremer du Pacifique in Tahiti where they were acclimatized in outdoor tank continuously renewed with natural, unfiltered, lagoonal seawater providing a constant input of microflora and fauna at natural concentration. Each individual was placed onto a petri dish on which byssal gland further developed for attachment. All individuals directly opened up right after their transfer into the outdoor tank. No visual sign of stress was observed during the acclimation period except for 4 individuals that died within the first 3 days (corresponding to a 2% mortality rate). Specimens used in this study were collected and held under a special permit (MEI #284) delivered by the French Polynesian government.

Experimental design and rearing system
To study the impact of temperature, pCO 2 , and their synergetic effect on the physiology of the giant clams and their symbionts, 4 experimental conditions were set up by applying 2 temperatures (29.2 • C and 30.7 • C) and 2 levels of pCO 2 (430 ± 22 μatm and 1212 ± 35 μatm). The tested conditions were as follows: (i) control: 29.2 • C, 430 μatm; (ii) acidification stress: 29.2 • C, 1212 μatm; (iii) thermal stress: 30.7 • C, 430 μatm; and (iv) acidification and thermal stress: 30.7 • C, 1212 μatm (Table 1). Temperature was set and maintained with an electronic controller (Hobby Biotherm Professional) connected to an aquarium heater (Shego). The pH was manipulated by bubbling CO 2 in water tanks. This was controlled by a pH-stat system (Dennerle) that continuously monitored pH (calibrated to NIST scale) and temperature to control the quantity of CO 2 to maintain the desired pH. The light was set to obtain a photosynthetically active radiation of 200 ± 20 μmol of photons m −2 .s −1 on a 12:12 h light/dark photoperiod.
After a 3-week acclimation period in an outdoor tank, 96 clams were randomly distributed in the experimental tanks 1 week before starting the experiment. For each condition, we used a 500-l tank containing 4 tanks of 30 l (ecological replicates) renewed with natural unfiltered lagoonal seawater at a flow rate of 50 l/h providing a constant input of microflora and fauna at natural concentration. Each 30l tank contained 6 clams (biological replicates). To avoid physiological shock, targeted temperature and pCO 2 were linearly achieved over 7 days. To evaluate the kinetics of the thermal and/or acidification stress, analyses were performed at 4 different times of exposure, i.e. 29, 41, 53 and 65 days. In total, 64 clams were used for data acquisition corresponding to 4 individuals per condition (1 individual per 30-l tank) and per time of exposure.

Monitoring of temperature, pH and water quality
To insure the stability of experimental conditions, temperature and pH parameters were measured twice a day for each tank at 8:00 am and 4:00 pm using a mercury thermometer certified ISO 9001 (±0.1 • C accuracy) and a pH-meter Consort P603 (±0.01 accuracy). Total alkalinity (TA) was weekly titrated using a 0.01-N HCl solution and a titrator (Schott Titroline Easy). Levels of pCO 2 and aragonite saturation state were calculated from temperature, pH (NBS scale), salinity and mean TA using the CO 2 SYS software (van Heuven et al., 2009). All parameters including seawater carbonate chemistry are reported in Table 1.

Holobiont O 2 consumption and production measurements
Giant clams were placed in an ecophysiological measurement system (EMS) to monitor O 2 consumption and production. The EMS consisted of five open-flow chambers. Four giant clams were individually placed into four chambers, while an empty shell was placed into a fifth chamber used as a control. EMS chambers contained water at the same temperature and pCO 2 conditions as in the experimental tanks. The light energy and photoperiod conditions were the same as for the acclimation tanks. Flow rate in all chambers was constantly maintained at 12 l.h −1 . Each chamber was equipped with a two-way electromagnetic valve activated by an automaton (FieldPoint National Instruments). When the electro-valve was opened, the water released from the chamber was analyzed for 3 min using an oxygen sensor (OXI 538, Cellox 325, WTW, Weilheim, Germany) to quantify dissolved oxygen. Oxygen measurements were performed over 48 h. The first 8 h of measurement were discarded due to the animal acclimatization to the chamber. In each chamber, the cycle was completed within 3 min: the first 2 min served to stabilize the measurement and an average of oxygen data was performed on the last minute of acquisition. This cycle was followed by another time frame of 3 min in the control chamber following the sequence specimen #1, control, specimen #2, control, specimen #3, control, specimen #4, control.
Respiration rate (RR) and production rate (PR) were calculated from the data obtained during night time and day time, respectively, using differences in oxygen concentrations between the control and experimental chambers. RR and PR = V(O1 − O2), where O1 is the oxygen concentration in the control chamber, O2 is the oxygen concentration in the experimental chamber and V is the water flow rate. RR and PR data were normalized to tissue dry weight. Once normalized, the terminology becomes O 2 consumption for RR and O 2 production for PR; both expressed in mg O 2 .h −1 .g −1 dry weight.
After O 2 production and O 2 consumption analyses were completed, a piece of the mantle was dissected for further symbiont fluorescence and density analyses (see Section Fluorescence and density measurements of symbionts). The remaining soft tissues were frozen and lyophilized for RR and PR data normalization.

Fluorescence and density measurements of symbionts
Potential effect of temperature and pCO 2 conditions on the photophysiology of the symbionts was studied by comparing fluorescence yield of photosystem II (PSII) between all experimental conditions. After clams were sacrificed, a 1 × 2cm mantle fragment was dissected. The tissue fragment was gently swiped using tissue paper to remove excess mucus and symbionts were collected by doing 5 smears using a sterilized razor blade. Collected symbionts were diluted into 5-ml of 0.2-μm filtered seawater and placed at the obscurity for 10 min to inactive the PSII before light excitation. Samples were homogenized and 3 ml of the homogenate were collected, placed into a quartz-glass cuvette and analyzed with AquaPen fluorometer (APC-100, Photon System Instruments ® , Czech Republic) at a 450-nm wavelength. The minimal fluorescence (F 0 ) and the maximal fluorescence (F m ) were measured. The quantum yield of photosynthesis was calculated as In addition, symbiont densities were evaluated from mantle fragments. For each individual, a circular (5 mm in diameter) piece of mantle was collected using a punch. The piece was weighed, grounded in 0.2-μm filtered seawater and homogenized. Then, 20 μl of the tissue extract were immediately collected and placed into Malassez cells for symbiont counting under optical microscope. For each sample, counting was performed on 4 replicates (3 columns per replicate). Data are expressed in number of symbionts/mg of mantle tissue.

Calcein labelling for evaluating daily shell extension rate
To study the impact of temperature and pCO 2 on shell growth rate, the mineralization front of giant clams was marked using calcein fluorochrome, which is irreversibly precipitated at the CaCO 3 mineralization site. Before the experiment starts, giant clams were immersed in a 100-mg.l −1 calcein solution (Sigma Aldrich) (calcein diluted in 1-μm filtered-seawater) for 8 h in the dark. During the labelling procedure, the bath of calcein solution was aerated using bubblers and water current was created via pumps. Calcein-labelled specimens were then placed into the experimental tanks. At the end of the experiment, for each individual, a 5-mm thick section was cut along the maximal shell growth axis through the right valve using a Swap Top Inland ® diamond saw. All sections obtained were polished and observed under epifluorescence with a Leitz Dialux ® 22 microscope. The distance between the fluorescent calcein mark and the edge of the shell formed during the experiment was measured following the maximal

Shell scanning electron microscopy study
To characterize temperature and pCO 2 effect on the shell ultrastructure, a scanning electron microscopy (SEM) study was carried out. For each individual, a 10-mm thick shell section was cut facing the section used for calcein observations. The section was then fractured along the width, 2 cm below the growing part of the shell (marginal part), using a small chisel and a hammer. Then, the apical fragment was longitudinally fractured and one piece was sonicated in tap water for 10 s, air dried and an additional drying was done overnight at 35 • C. Sample was placed on a stub covered with carbon tape, gold-coated and observed at 15 kV using a Hitachi TM3030 SEM at the Université de la Polynésie française. For each condition, and for each time of exposure, two samples were selected based on their daily shell extension rate, i.e. samples showing the lowest and the highest rate. To evaluate the impact of temperature and/or pCO 2 on the shell ultrastructure, SEM observations were performed for each specimen in two different zones of the crossed-lamellar outer layer. Zone 1 corresponds to the shell formed in situ (i.e. in the shell region located before the calcein mark), while zone 2 corresponds to the shell formed during the experiment. Observations were made at the ultrastructural level and focused on the aspect and integrity of the lamellae of the crossed-lamellar outer layer. In addition, to check the potential effect of temperature and pCO 2 on the external shell surface, the periostracum (i.e. the organic layer covering the shell external surface) of two individuals per experimental condition (65 days of exposure) was observed.

Statistical analyses and data processing
Normality of data distribution and homogeneity of variance were tested with the Shapiro-Wilk test and the Bartlett test, respectively. Production and consumption of O 2 data followed the conditions of application of parametric tests, but photosynthetic yields and symbiont densities were transformed using Box Cox transformation while shell extension rates were square root transformed to meet these conditions. Comparisons were done using a three-way ANOVA with interactions (fixed factors: time of exposure, temperature and pCO 2 ). Tukey post hoc comparisons were done at α = 0.05 for all analyses. Correlations between physiological parameters were tested using Pearson method with a threshold of r = 0.25 (α = 0.05).
For all physiological parameters, i.e. O 2 respiration, O 2 consumption, symbiont photosynthetic yield and density, means (±SD) were calculated based on the four biological replicates for each condition and each time of exposure. For the daily shell extension rate, means (±SD) were calculated for each condition and each time of exposure based on the number of biological replicates displaying a detectable calcein mark.

Effect of temperature and pCO 2 on the holobiont oxygen balance
For all sampling time (i.e. 29, 41, 53, 65 days), a cyclic pattern of O 2 production and consumption was observed following the circadian cycle (Fig. 1). This pattern corresponds to oxygen photosynthetically produced and heterotrophically consumed by the holobiont during the day and the night, respectively. No mortality was observed in all tested temperature/pCO 2 conditions during the whole experiment.  Figure 2: Graphs reporting data of (A) oxygen production, (B) oxygen consumption, (C) symbiont photosynthethic yield, (D) symbiont density and (E) daily shell extension rate obtained for each temperature/pCO 2 experimental condition and time of exposure (black and grey columns correspond to 430 μatm and 1212 μatm, respectively). Data are given in mean (+SD) calculated from four replicates for all physiological parameters except for daily shell extension rate for which numbers of replicates are specified.
reported in Table 3. The ANOVA has indicated that oxygen production of the holobiont during day time was significantly altered at 30.7 • C (P = 0.009) but not at 1212 μatm of CO 2 (P = 0.362) ( Table 2). In addition, O 2 production was higher at 29.2 • C than at 30.7 • C for all times of exposure ( Fig. 2A) and decreased over time (P = 0.030, Table 2  The three fixed factors are pCO 2 , temperature and time of exposure. Asterisks denote significant differences ( * P ≤ 0.05, * * P ≤ 0.01, * * * P ≤ 0.001).
by time of exposure (P = 0.533) and remained stable throughout the whole experiment (Fig. 2B). No interaction effect between the three tested factors was found to affect the holobiont oxygen balance (Table 2).

Symbiont density and photosynthesis under different temperature/pCO 2 conditions
After 2 months of exposure, the symbiont photosynthetic yield and density were significantly impacted at 30.7 • C (P < 0.0001), high pCO 2 (1212 μatm) (P = 0.017 and P = 0.005, respectively) and by the time of exposure (P = 0.021 and P = 0.001, respectively). During the whole experiment and for both pCO 2 conditions, the means of photosynthetic yield (Fig. 2C) and symbiont density (Fig. 2D) tended to be higher at 29.2 • C than at 30.7 • C. However, an inter-individual variability was observed.
Concerning the interaction parameters, only interaction between temperature and time of exposure parameters had a significant effect on the symbiont photosynthetic yield (P = 0.025) ( Table 2). No parameter interaction was found to affect the symbiont density ( Table 2).

Effect of pCO 2 on daily shell extension rate
Calcein mark was detectable in 55 over 64 shell sections. Statistical analyses on shell extension rate showed that pCO 2 and the time of exposure had a significant effect on the shell extension rate (P = 0.010 and P = 0.007, respectively, Table 2).
As shown in Fig. 2E, the mean values of shell extension rates tended to be lower at high pCO 2 .

Effect of long-term exposure to temperature and pCO 2 on the periostracum and the shell microstructure
As a first approach, SEM observations of the periostracum (external shell surface) were made in order to evaluate potential effect of the treatments. These observations were performed at two different places of the shell: (i) the ventral extremity part and (ii) a lower part of the shell. Two individuals per experimental condition (65 days of exposure) were studied. For all the samples, the periostracum was of the same colour (based on visual observations). Under SEM, at the ventral extremity in control condition (29.2 • C, 430 μatm), the periostracum did not appear altered ( Supplementary Fig.  S2A, C; Table 4), in the lower part it displayed a 'mushy' aspect with numerous micro-borer galleries ( Supplementary  Fig. S2B, D). These features were observed in all the other experimental conditions (i.e. 29.2 • C, 1212 μatm; 30.7 • C, 430 μatm; 30.7 • C, 1212 μatm; Fig. S2) and for all the samples studied, which may signify that the treatments did not significantly affect the periostracum.
To strengthen the results above ultrastructural observations of shell fractures were done in two zones: zone 1 corresponding to the shell formed in situ (i.e. before the experiment) and zone 2 corresponding to the shell formed during the experimental conditions. In the control condition (i.e. 29.2 • C, 430 μatm), after 29 days of exposure, the lamellae formed before the experiment (zone 1) and during the experiment (zone 2) were well cohesive and displayed an elongated shape, a smooth surface and sharp or slightly rounded outlines ( Fig. 3A-C). At 29 days, no difference was observed between both zones for all temperature/pCO 2 conditions (example given for 29.2 • C, 1212 μatm in Fig. 3D-F; Table 4).
From 41 days and after, and for all temperature/pCO 2 experimental conditions, the shell formed before the experiment (zone 1) displayed the same microstructural features like the ones described above (example given for 29. The effects of significant parameters were tested as time of exposure alone and combined to the temperature. The letter annotations correspond to the significances between conditions (P < 0.05).
in all temperature/pCO 2 conditions, appeared less cohesive (Fig. 3N, O showing shell formed at 30.7 • C, 1212 μatm; Table 4) and some of them additionally displayed a pronounced granular aspect of the surface even in the control condition (Fig. 3H, I and Fig. 3K, L showing shell formed at 29.2 • C, 430 μatm and 30.7 • C, 430 μatm, respectively; Table 4). These modifications were observed for the majority of shells formed in all temperature/pCO 2 conditions, at 53 and 65 days (Table 4). Two states classifying ultrastructural differences between the shell formed before and during the experiment were defined as follows: (i) ND, no difference observed between zones 1 and 2; and (ii) D, difference observed between zones 1 and 2 meaning that the lamellae were less cohesive in zone 2 compared to zone 1 (see Table 4). In total, 19 samples over 32 displayed differences in lamellae aspect between both zones.

Effect of time of exposure on oxygen balance, biomineralization and symbiont photophysiology
Time of exposure to thermal and acidification stress was shown to have a significant impact on O 2 production (P = 0.030), photosynthetic yield (P = 0.021), symbiont density (P = 0.001) and shell growth rate (P = 0.007) ( Table 2).
Post hoc tests showed that the kinetics of thermal or acidification stress varies depending on the physiological parameter measured: O 2 production and photosynthetic yield were significantly different between 29 and 65 days while symbiont density and shell growth rate were both impacted from 53 days of exposure (Table 3).

Relationship between physiological parameters
To establish if relationships exist between the various physiological parameters measured, a correlation matrix was generated (Table 5). The photosynthetic yield of symbionts was strongly correlated to their density (r = 0.667). O 2 production was also strongly correlated to symbiont density (r = 0.492) and photosynthetic yield (r = 0.331). Concerning the circadian functioning of the holobiont, its nocturnal oxygen need was strongly correlated to the diurnal O 2 production (r = 0.629). No significant correlation was found between shell extension rate and other physiological parameters.

Discussion
In the present study, we investigated the physiological responses of T. maxima (i.e. 4-year-old specimens) to temperature and pCO 2 conditions in the French Polynesian lagoons under current warmer season conditions and those predicted for 2100 by the IPCC 2014 (+1.5 • C and +800 μatm of CO 2 ). Present thermal conditions were stressful enough to induce an acclimatization process characterized 'A' and 'B' represent the two samples observed for each treatment and time of exposure; 'ND' means that no significant difference is observed between the shell formed before and during the experiment for the lamellae of the outer shell layer; 'D'means that a notable difference is observed between both parts of the shell (i.e. the lamellae are less cohesive). For the periostracum, 'no alteration' means that the periostracum covering the ventral extremity part of the shell is well preserved.
by a slight regulation of the symbiont density without the collapse of the photosynthesis. However, the predicted thermal and pCO 2 conditions were much more stressful and induced strong disturbances of the processes linked to the phototrophic symbiosis and the biomineralization.
Indeed, the 30.7 • C treatments have significantly impacted the holobiont by inducing a strong bleaching response illustrated by the reduction of its O 2 production, symbiont density and photosynthetic yield. High pCO 2 (+800 μatm of CO 2 ) was shown to alter the symbiont photosynthetic yield and  Correlation was tested using the Pearson method (threshold of r = 0.250, α = 0.05) and significant relationships between parameters are indicated by an asterisk ( * ).
density and affect its biomineralization process by decreasing the shell growth rate.

No synergetic effect between temperature and pCO 2 on giant clam and symbiont physiological parameters
Increase in atmospheric CO 2 impacts both the seawater temperature and pH inducing ocean warming and acidification (Sabine et al., 2004;IPCC, 2014). Therefore, the analysis of the synergetic effect of both stressors on marine bivalve's physiology is crucial. Watson et al. (2012) reported in T. squamosa juveniles a decrease of giant clam's survival rate induced by exposure to high-temperature or high-pCO 2 conditions. The lowest survival rate (i.e. <20%) was observed under the highest pCO 2 condition (+600 μatm of CO 2 ) at +1.5 • C and +3 • C, after 60 days of exposure.
In our study, no synergetic effect was detected for all measured physiological parameters. Moreover, no mortality was observed in all tested temperature/pCO 2 conditions during the whole experiment which leads us to suggest that a temperature of 30.7 • C and a pCO 2 of 1212 μatm seem to be non-lethal, at least over 65 days of exposure. However, we cannot exclude that some specimens, especially the bleached individuals, may have died after longer exposure to the experimental conditions.

Effect of temperature on holobiont oxygen balance, symbiont photophysiology and shell ultrastructure
Temperature is one of the most important parameters driving bivalve physiology (Aldridge et al., 1995;Bougrier et al., 1995;Watson et al., 2012;Gazeau et al., 2013;Le Moullac et al., 2016a;Latchère et al., 2017). However, since giant clams live in symbiosis with photosynthetic symbionts (Klumpp et al., 1992;Klumpp and Griffiths, 1994;Soo and Todd, 2014), the comparison at the metabolic level with nonsymbiotic bivalves can be misleading. Discussion on elevated temperature effect on the holobiont oxygen balance should therefore integrate the effect on both giant clam and its symbionts (Jones and Hoegh-Guldberg, 2001). Even though physiological and molecular responses to thermal stress may differ, the comparison with scleractinian corals, which are also symbiotic and calcifying organisms, makes sense to better understand the impact of temperature on the holobiont physiology.
We observed that high temperature (+1.5 • C) significantly reduced the holobiont O 2 production, the density and the photosynthetic yield of symbionts from 29 days of exposure. Additionally, partial or total bleaching was observed for the majority of individuals exposed to 30.7 • C (at both ambient and high pCO 2 , see Supplementary Fig. S3). This suggests that thermal stress has a significant impact on the holobiont photophysiology. The reduction of the symbiont photosynthetic yield could reflect photoinhibition. Photoinhibition was previously linked to the degradation of the D1 protein of the reaction center of the PSII altering the photosynthetic apparatus functioning in symbiotic corals and sea anemone subjected to thermal stress (Warner et al., 1999;Smith et al., 2005;Richier et al., 2006;Ferrier-Pagès et al., 2007). Decrease of photosynthetic activity reduced the symbiont O 2 production and consequently, impacted the holobiont O 2 production. Comparable results were obtained from a 24-h experiment showing that an increase of 3 • C affected the oxygen production of T. gigas and T. derasa (Blidberg et al., 2000). Photoinhibition led to the production of ROS by symbionts, which are known to pass through cellular membranes, cause oxidative damages (Lesser et al., 1990;Lesser, 1996;Downs et al., 2002) and impact the PSII (Richter et al., 1990). In corals, Downs et al. (2002) showed that high levels of oxidative damage associated with coral bleaching. Indeed, ROS such as hydrogen peroxide (H 2 O 2 ) may play a role in signalling molecule activating the symbiosis dissociation (Fitt et al., 2001;Smith et al., 2005;Perez and Weis, 2006). Expulsion of symbionts by the host may be a strategy to limit oxidative stress and damage to ultimately survive environmental stress (Downs et al., 2002;Perez and Weis, 2006). Two recent works support this hypothesis. Firstly, Zhou et al. (2019) related a decrease in symbiont density in response to an excess of oxidative stress in thermally stressed T. crocea. Secondly, Dubousquet et al. (2016) showed that T. maxima overexpressed genes encoding ROS scavengers in response to thermal stress. Interestingly, the cellular and molecular mechanisms enhanced before the bleaching response (e.g. the loose of the symbionts) seem similar between giant clams and corals and conduct to

11
the same phenomenon of symbiosis dissociation. However, since giant clams and corals display an extra-and an intracellular symbiosis, respectively, the mechanisms of symbiosis dissociation may differ. This constitutes an interesting example of evolutionary convergence of a stress response in two very distant organisms. In this context, it would be interesting to test if the bleaching response in giant clam can be adaptive as it was proposed for corals (Fautin and Buddemeier, 2004) and if variations in giant clam thermotolerance are correlated to the composition of the symbiotic population (Baker, 2003).
Concerning the effect of temperature on the ultrastructure of the crossed-lamellar structure, the lamellae formed in all experimental conditions in the first 29 days were well cohesive with elongated shape and a slight granular aspect, which is consistent with the description made by Faylona et al. (2011) in T. maxima, Belda et al. (1993 in T. gigas and Agbaje et al. (2017) in T. derasa. However, after 29 days, the lamellae observed in the majority of the shells formed in both temperature conditions (whatever the pCO 2 condition) appeared to be less cohesive with pronounced granular aspect. We suggest that these features are due to a lack of organic matrix between the lamellae (i.e. inter-lamellar organic matrix) and embedding the nano-grains forming the lamellae in the crossed-lamellar structure. Indeed, the biomineralization of calcium carbonate structures involves the transport of ions to the mineralization site and the synthesis of macromolecules referred to as 'organic matrix' (Allemand et al., 2004). The organic matrix, consisting of 0.9 wt% and 1.83 wt% in T. derasa and T. gigas shell, respectively (Agbaje et al., 2017(Agbaje et al., , 2019, is mainly composed of macromolecules commonly found in molluscan shell organic matrix such as lipids, polysaccharides and proteins displaying various levels of glycosylation (Marin et al. 2012;Agbaje et al., 2017). The formation of these macromolecules may be energetically costly for molluscs such as marine gastropods (Palmer, 1992).
We suggest that in our tested experimental temperature, while oxygen balance is significantly altered, less energy may be available and allocated to the synthesis of macromolecules involved in biomineralization, which may explain the lack of organic matrix embedded into the giant clam shell.

Effect of pCO 2 on giant clam biomineralization and symbiont photophysiology
Results of the present study have indicated that in T. maxima high-pCO 2 condition (+800 μatm) induced a significant decrease of the shell extension rate. This observation is in accordance with those obtained for giant clams in three different experiments using various CO 2 enrichment levels and exposure durations. Watson (2015) has exposed juveniles T. squamosa to +250 and +550 μatm of CO 2 for 8 weeks under various light levels. Under mid-light level, the author has reported a reduced survival and growth while under high-light levels the growth only was affected. Kurihara and Shikota (2018) also reported a negative effect of pCO 2 on shell growth (i.e. shell height) in juveniles T. crocea when exposed to +600 μatm and +1600 μatm for 4 weeks. Toonen et al. (2012) demonstrated that young specimens of T. maxima and T. squamosa had lower shell growth rates, compared to those reported in the literature under natural pCO 2 /pH conditions, when kept 1 year in +350 to +1000 μatm pCO 2 conditions.
As mentioned above, high-pCO 2 condition (+800 μatm) affected giant clam's shell growth rate and also photosynthetic yield and density of symbionts. Negative effect of pCO 2 on the shell growth rate may be linked to two non-exclusive hypotheses: (i) the physiological adjustment needed to thrive with a seawater at 1212 μatm of CO 2 instead of 430 μatm and/or (ii) an alteration of symbiont photophysiology leading to a potential reduction of the 'light-enhanced calcification' (LEC) phenomenon. Concerning the former, the increase of CO 2 dissolution in the water column leads to the modification of the carbonate chemistry equilibrium and to the increase of H + concentration (Cyronak et al., 2016). These strong environmental changes affect the whole organism's homeostasis, but more particularly physiological functions, such as biomineralization, where the control of carbonate concentration and pH are the most essential. To form their mineralized structures, calcifying organisms modify the carbonate composition at the mineralization site to promote CaCO 3 precipitation. These modifications are performed by several enzymatic reactions, including the removal of H + to locally increase the pH at the mineralization site for maintaining chemical conditions to enhance CaCO 3 precipitation (Allemand et al., 2011;Taylor et al., 2012). Corals are known to regulate the pH at their mineralization site (Venn et al., 2013;Holcomb et al., 2014) but a decrease of coral calcification was also linked to a decline in pH in the calcifying fluid (Ries, 2011;McCulloch et al., 2012). The maintenance of biomineralization under these non-optimal environmental conditions are therefore energetically costly, which may result in a general decrease of growth (Vidal-Dupiol et al., 2013). In our case, we suggest that giant clam T. maxima exposed to high pCO 2 may allocate more energy to maintain a proper pH of the extrapallial fluid for nucleation and deposition of aragonite. Regarding the LEC (Vandermeulen et al., 1972), this phenomenon observed in Symbiodiniaceaehost symbiosis has been extensively described in corals and represents the capacity of symbionts to stimulate the host calcification (Allemand et al., 2004). The symbionts stimulate the host metabolism and calcification by providing energy resources and/or O 2 (Chalker and Taylor, 1975). Moreover, symbionts may promote the aragonite precipitation by providing inorganic carbon, nitrogen and phosphorus and by synthetizing molecules used as precursor for the synthesis of skeletal organic matrix (Pearse and Muscatine, 1971;Cuif et al., 1999;Furla et al., 2000;Muscatine et al., 2005). They also facilitate CaCO 3 precipitation by influencing the dissolved inorganic carbon (DIC) equilibrium by removing the CO 2 via photosynthesis (Goreau, 1959 has been also described in giant clams (Ip et al., 2006;Ip et al., 2017). In T. squamosa, LEC increased the pH and reduced the ammonia concentration at the interface between the inner mantle and the shell in the extrapallial fluid, where the biomineralization occurs (Ip et al., 2006). Recently, Chew et al. (2019) reported a light-enhanced expression of carbonic anhydrase, i.e. CA4-like, in the inner mantle of T. squamosa and suggested that this enzyme is involved in giant clam biomineralization by catalyzing the conversion of HCO 3 − to CO 2 . In this context, an altered photophysiology of the symbionts can rationally alter LEC and consequently results in a decrease of the shell growth rate. Finally, one can suggest that under acidification stress, giant clam may reduce some physiological functions such as biomineralization and allocate more energy to essential functions for its survival. In our study, temperature also altered photophysiology and holobiont O 2 production, but did not significantly affect shell growth rate. Therefore, the most plausible hypothesis explaining the negative effect of pCO 2 on shell growth rate may be related to the low aragonite saturation state at high pCO 2 .
Concerning the effect of pCO 2 on the shell microstructural integrity, in the temperate bivalve Mytilus edulis, an exposure to +150, +350 and +600 μatm of CO 2 for 6 months induced disorientation in the shell of the newly formed calcite crystals of the prismatic layer (Fitzer et al., 2014). In the giant clam T. maxima, high pCO 2 had no significant effect on the integrity of the aragonitic lamellae of the crossed-lamellar layer during the first 29 days of exposure. From 41 days of exposure, its potential impact remains unresolved as differences were noticed in the shells formed under future high-temperature, high-pCO 2 and even under today's temperature/pCO 2 conditions. The fact that differences were reported for shells formed in all experimental conditions from 41 days suggests that they may be due to a long-term exposure to 29.2 • C and 30.7 • C.

Conclusion
This study enables the evaluation of T. maxima's physiological responses to realistic temperature and pCO 2 predicted at the end of this century. We demonstrated that high temperature mimicking temperature encountered during the warmer months of the year has a significant negative impact on symbiont densities and photosynthetic capacities. This negative impact on the symbiont physiology induces a decrease in the net holobiont O 2 production.. Therefore, by influencing symbiont physiology, the temperature may affect the energetic needs of the giant clam host. The high pCO 2 has a negative impact on shell growth rate, symbiont densities and photosynthetic capacities. Shell microstructure is not affected by temperature nor by the pCO 2 in the first 29 days of exposure. However, for all temperature/pCO 2 conditions, a longer exposure (≥41 days) modified the shell ultrastructure. These observations support our hypothesis that 29.2 • C is a temperature that already affects giant clam metabolism, at least over a long-term exposure. However, no synergetic effect was found between temperature and pCO 2 parameters. All these observations suggest that temperature and pCO 2 influence different physiological functions and that giant clam populations may dramatically suffer from the temperature and pCO 2 conditions predicted for the next decades. This is especially true for the temperature since the populations of giant clams (wild and farmed) of the Tuamotu Archipelago are already confronted with temperatures of ∼29.2 • C every year. Additionally, the threshold of 30.7 • C applied in our study and that will be encountered annually in the future corresponds to what is measured currently during abnormally warm years (Fig S1). During the heatwave event in 2016, 80% of the farmed clam populations have bleached resulting in a high level of mortality (personal communication). To complement these results on the effects of temperature on the giant clams, it is now essential to conduct integrative analyses that will take into account the acclimatization and adaptive potential of the whole holobiont. The tools of transcriptomic, genomic and epigenomic in association with ecologically relevant experiments both at the individual and population levels will be particularly relevant to address these questions (Torda et al., 2017). These will allow a better understanding of the fundamental physiological processes of the holobiont and its response to future changes. Results from these studies may help in adapting local policies and management to maintain sustainability of giant clam populations and their exploitation, especially in the Eastern Tuamotu islands where bleaching events have been observed at an increasing and alarming rate.

Funding
This work was supported by the Ifremer Institution (Politique de site program, GECO project) and the Université de la Polynésie française (MAPIKO and CLAMS projects).