Impacts of ocean acidification and warming on post-larval growth and metabolism in two populations of the great scallop (Pecten maximus L.)

Ocean acidification and warming are key stressors for many marine organisms. Some organisms display physiological acclimatisation or plasticity, but this may vary across species ranges, especially if populations are adapted to local climatic conditions. Understanding how acclimatisation potential varies among populations is therefore important in predicting species responses to climate change. We carried out a common garden experiment to investigate how different populations of the economically important great scallop (Pecten maximus) from France and Norway responded to variation in temperature and pCO2 concentration. After acclimation, post-larval scallops (spat) were reared for 31 days at one of two temperatures (13°C and 19°C) under either ambient or elevated pCO2 (pH 8.0 and pH 7.7). We combined measures of proteomic, metabolic, and phenotypic traits to produce an integrative picture of how physiological plasticity varies between the populations. The proteome of French spat showed significant sensitivity to environmental variation, with 12 metabolic, structural and stress-response proteins responding to temperature and/or pCO2. Principal component analysis revealed seven energy metabolism proteins in French spat that were consistent with countering ROS stress under elevated temperature. Oxygen uptake in French spat did not change under elevated temperature, but increased under elevated pCO2. In contrast, Norwegian spat reduced oxygen uptake under both elevated temperature and pCO2. Metabolic plasticity seemingly allowed French scallops to maintain greater energy availability for growth than Norwegian spat. However, increased physiological plasticity and growth in French spat may come at a cost, as French (but not Norwegian) spat showed reduced survival under elevated temperature. Summary Statement Juvenile scallops from France and Norway differ in their response to warming and acidification. French scallops show more physiological plasticity, adjusting their proteome and metabolism in order to maintain growth.

6 refrigerated van). From here, they were transported to the experimental facility and transferred to 158 tanks maintained at 13°C. 159 Approximately 9000 French spat were collected from sea cages near Sainte-Anne du Portzic, in the 160 Bay of Brest, 6 days after the transport of Norwegian spat. We replicated the most stressful part of 161 the transportation of Norwegian spat (emersion for approximately 6 hours) for French spat before 162 introducing them to tanks. 163 164

Characterisation of experimental system and animal maintenance 165
Following transport or simulated transport, spat were transferred to six 'raceway' flow through tanks 166 (100 L). For each population, spat were split among 18 mesh-bottomed trays (mesh 500 μm), held 167 approximately 2.5 cm from the tank bottom by PVC supports, with populations kept separate initially 168 (6 trays per raceway, 36 total trays). Each raceway drained into an independent header tank (30 L) 169 containing an overflow. The rate of water renewal was regulated by gravity pressure between the 170 input and overflow: filtered UV-sterilized sea water was supplied at a flow of approximately 90 mL 171 min -1 , leading to approximately one renewal per day. Header tanks also received a constant flow of 172 microalgae (equal concentrations of Tisochrysis lutea and Chaetoceros gracilis at a final concentration 173 of approximately 80 000 cells mL -1 in experimental tanks), which was supplied to two different header 174 tanks from 10 L bottles via peristaltic pumps (two delivery tubes per pump, one for each header tank). 175 Bottles of algae were replenished every 2 days. 176 From each header tank, a submerged pump supplied algae enriched seawater to a network of PVC 177 pipes (with small holes drilled in them) overhanging each mesh-bottomed tray in the raceway at a rate 178 language (R Development Core Team 2019), with tray as a random factor to account for variation 223 between trays within each experimental system. The significance of population and environmental 224 variable effects on survival were tested with Wald chi-square tests using the Anova function from the 225 car package (Fox & Weisberg 2011). Because coxme does not allow more than two dependent 226 variables to be tested simultaneously, we first tested for differences between French and Norwegian 227 scallops, before comparing the effects of temperature and pCO2 (plus their interaction) on survival for 228 each population separately. To reduce the effect of sample size differences on statistical power, a 229 random number of French spat, equivalent to the smaller number of Norwegian spat, were included 230 in the survival analysis. Spat (2-6 individuals per tray) were sampled near-daily during the adjustment 231 period and weekly during the experimental treatment for a separate experiment (unpublished data). 232 These individuals were 'left censored' for the purposes of the survival analysis. 233 234

Analysis of whole organism phenotypes 235
At the end of the experiment (day 31), 20-40 individuals were removed from each tray and preserved 236 in 95% ethanol at 4 °C. Three primary traits of shell size, shell weight and soft tissue weight were 237 measured for between 8 and 15 individuals from each tray (mean = 12). Shell height is one of the 238 most-commonly measured morphological phenotypes in bivalves. Bivalve shells grow by marginal 239 accretion and changes in shell height provide an accurate measure of individual growth. Consistent 240 patterns of accretion in the flat valves of scallops allows the estimation of growth over fine temporal 241 scales (Chauvaud et al. 2012). Soft tissue dry weight (dry body weight) and total shell dry weight (total 242 shell weight) are measures of investment in these two compartments. We also used these measures 243 to estimate the condition index (CI: soft tissue dry weight / shell dry weight ratio), which encapsulates 244 the difference in resource allocation to these compartments (Lucas & Beninger 1985). Spat were 245 dissected and soft tissue was dried at 75 °C for 24 hours while both the left valve (flat) and right valve 246 (curved) were air dried for at least 24 hours. The dry bodies, flat and curved valves were then weighed 247 to the nearest 0.0001 g using a digital balance (Mettler Toledo). 248 Multiple high-resolution images of the flat valve at a range of focal depths were obtained using an 249 AxioCam MRC 5 linked to a SteREO Lumar.V12 stereomicroscope (Carl Zeiss) equipped with a 250 motorized stage: the resultant photomosaics were then assembled using AxioVision 4.9.1 software 251 (Carl Zeiss). From these images shell height was measured using ImageJ software. These images were 9 controlled conditions of the experimental facility could be clearly associated with an alteration in the 254 colour of newly calcified shell. Due to the considerable variation in size among scallops from both 255 populations, initial shell height was included as a covariate in whole organism phenotypic analyses. 256 Although Lucas and Beninger (1985) recommend the use of cubed height as a covariate for mass 257 measures (such that variation scales in the same number of dimensions), we found initial height alone 258 (not raised to a power) better accounted for covariation in the data. Phenotypic measures and ratios 259 were plotted against these initial measures as part of an inspection for outliers. Four out of 431 260 samples were removed due to at least one trait showing extreme outlier values. Quantile-quantile 261 plots were assessed to determine probability distributions. Dry body weight and total shell weight 262 were subsequently log transformed to ensure normality. For all analyses, populations were analysed 263 separately because of the strong differences in initial sizes. 264 The dependency of whole organism phenotypes on temperature, pCO2, and initial shell dimensions, 265 as well as their 2-way and 3-way interactions, was assessed using linear mixed-effects models in the 266 lme4 package in R (Bates et al. 2015); tray was included as a random effect. Backwards stepwise term 267 deletion was used to test the importance of interactions and main effects. Statistics were obtained 268 from minimal models fitted with restricted maximum likelihood and P-values were obtained using the 269 lsmeans package (Lenth 2016). When either temperature or the interaction of temperature and pCO2 270 were significant, contrasts between groups were evaluated with pairwise post-hoc tests using the 271 emmeans package (Lenth 2020). For the interaction between temperature and pCO2, temperature 272 differences at a given pCO2 and pCO2 differences at a given temperature were considered. Phenotypic 273 responses to temperature and pCO2 were plotted using effect size plots in the jtools package (Long 274 2020) based on fully parameterised linear models (temperature, pCO2, initial shell dimensions, and all 275 interactions). These account for covariate variation (including interactions), include confidence 276 intervals, and can be mean centred. Although they do not account for random effect variance, these 277 plots provide an intuitive means of visualising these data when combined with mixed-effects model 278 statistics. 279 280

Analysis of metabolic rates 281
For both populations, the effect of temperature and pCO2 on oxygen consumption (ṀO2), used as a 282 proxy for metabolic rate, was assessed (Rastrick et al. 2018) using randomly selected individuals from 283 highest and lowest temperature treatments . On 284 day 27 of the experiment, three spat from each tray (nine per population/treatment combination, n = 285 72) were selected for metabolic analyses. 286 Spat were placed in individual stop-flow respirometers (volume 100ml) supplied with the same sea-287 water as the respective treatments. Animals were allowed 1 h to recover from handling and regain 288 natural ventilatory behaviour before the flow to each chamber was stopped and the decreases in % 289 oxygen saturation continuously measured using an optical oxygen system (Oxy-10mini, PreSense; 290 labquest 2, Vernier; Rastrick and Whiteley, 2011;Rastrick 2018). The incubation period was 5 h, during 291 which time, oxygen levels of the seawater did not fall below 70% (% air saturation) to avoid hypoxic 292 conditions. A blank chamber with no animal was used to control for the background respiration in the 293 seawater. The decrease in oxygen (% air saturation) within each chamber was converted to oxygen 294 partial pressure (pO2) adjusted for atmospheric pressure and vapour pressure adjusted for relative 295 humidity (measured using a multimeter; Labquest 2, Vernier). This decrease in pO2 was converted to 296 concentration by multiplying by the volume of the chamber, minus the animal volume, and the oxygen 297 solubility coefficient adjusted for the effect of temperature and salinity (Benson and Krause, 1984). 298 Values were standardised to individual dry weight and expressed as µmol O2 h -1 mg -1 ± SEM. At the 299 end of these metabolic experiments, the 72 individuals were sacrificed and dissected. Soft tissue was 300 dried and weighed as described for the phenotypic analyses. 301 ṀO2 values were tested for normality. Although residuals approximated a normal distribution among 302 French spat, they deviated from normality for Norwegian spat. Consequently, the raov function from 303 the package Rfit (Kloke & McKean 2012) was used to provide rank-based estimations of linear models. 304 We initially included population, temperature, pCO2 and all possible interactions in this model. 305 However, to facilitate interpretation of the effects of temperature, pCO2 and their interaction, we also 306 fitted models for each population separately. We used Benjamini-Hochberg-adjusted pairwise 307 Wilcoxon tests to identify differences when the interaction was significant. 308 309 2.6. Analysis of the proteome 310 Spat for proteomic analyses were collected on day 31 from each tray in the 13-normCO2, 13-highCO2, 311 19-normCO2, and 19-highCO2 treatments. For each tray, two samples (each containing a pooled 312 sample of two whole individuals) were flash frozen in liquid nitrogen (48 samples total) and stored at 313 -80°C until analysis. Samples were homogenised by bead beating at 4°C in 500 μl Tris-HCl lysis buffer 314 (100 mM, pH 6.8) containing 1% Protease Inhibitor Mix (GE Healthcare). A full and detailed description samples can be found in Harney et al. (2016), but is described here briefly. Homogenised samples were 317 centrifuged and solubilised proteins from the interphase were quantified using a DC (detergent 318 compatible) protein assay in a micro-plate reader. Then, 800 μg of protein were precipitated and 319 desalted using a 1:1 ratio of sample to TCA/acetone (20% TCA). The supernatant was discarded, and 320 pellets were neutralised by adding Tris-HCl/acetone (80% acetone) containing bromophenol blue as a 321 pH indicator. Pellets were centrifuged once again and air-dried, before being rehydrated in Destreak 322 rehydration solution (GE healthcare) containing 1% IPG (immobilised pH gradient) buffer (pH 4-7). 323 After one hour, samples were ready for isoelectric focusing (IEF) on the IPGphor3 system (GE 324 healthcare). After IEF, IPG strips were bathed in a rehydration solution (50 mM Tris-HCl pH 8.8, 6 M 325 urea, 30% glycerol, 2% SDS and 0.002% Bromophenol Blue) for two 15 min periods, first with 10 mg/ml 326 dithiothreitol, and then in the same solution containing 48 mg/ml iodoacetamide. Strips were then 327 deposited on a 15cm × 15cm lab-cast SDS-PAGE gel containing 12% acrylamide and migrated. Protein 328 spots were stained with Coomassie Blue (PhastGel, GE Healthcare). Gels were bleached with baths of 329 H2O/methanol/acetic acid (70/30/7) and photographed using G:BOX (SynGene). The 32 clearest gels 330 were taken forward for analysis (4 per population per treatment) and protein spots in the resulting 331 images were aligned using Progenesis SameSpots v3.3 software (Nonlinear Dynamics, Newcastle upon 332 Tyne, UK) and then manually verified. The effects of population, temperature, and pH were evaluated 333 by running ANOVAs for each spot (combined population analysis). Due to the large number of tests 334 involved, P values were adjusted using the false discovery rate (FDR), and fold change values were 335 determined. Proteins which differed significantly in abundance between the populations, or between 336 temperature or pCO2 treatments (FDR ≤ 0.05) were excised from the gels and analysed using mass 337 spectrometry. 338 Gel pieces were first washed in 50 mM ammonium bicarbonate (BICAM), and then dehydrated in 100% 339 acetonitrile (ACN). Gel pieces were vacuum-dried, rehydrated with BICAM containing 0.5 μg MS-grade 340 porcine trypsin (Pierce Thermo Scientific), and incubated overnight at 37°C. Peptides were extracted 341 from the gels by alternatively washing with 50 mM BICAM and ACN, and with 5% formic acid and ACN. 342 Between each washing step, the supernatants from a given gel piece were pooled and finally 343 concentrated by evaporation using a centrifugal evaporator (Concentrator 5301, Eppendorf). 344 Mass spectrometry (MS) experiments were carried out on an AB Sciex 5800 proteomics analyzer 345 equipped with TOF/TOF ion optics and an OptiBeamTM on-axis laser irradiation with 1000 Hz 346 repetition rate. The system was calibrated immediately before analysis with a mixture of Angiotensin 347 μL volume of this peptide solution was mixed with 10 μL of 5 mg/mL α-cyano-4-hydroxycinnamic acid 350 matrix prepared in a diluent solution of 50% ACN with 0.1% TFA. The mixture was spotted on a 351 stainless steel Opti-TOF 384 target; the droplet was allowed to evaporate before introducing the 352 target into the mass spectrometer. All acquisitions were taken in automatic mode. A laser intensity of 353 3400 was typically employed for ionizing. MS spectra were acquired in the positive reflector mode by 354 summarizing 1000 single spectra (5 × 200) in the mass range from 700 to 4000 Da. Tandem mass 355 spectrometry (MS/MS) spectra were acquired in the positive MS/MS reflector mode by summarizing 356 a maximum of 2500 single spectra (10 × 250) with a laser intensity of 4200. For the MS/MS 357 experiments, the acceleration voltage applied was 1 kV and air was used as the collision gas. Gas 358 pressure was set to medium. The fragmentation pattern was used to determine the sequence of the 359 peptide. 360 Database searching was performed using the MASCOT 2.4.0 program (Matrix Science). A custom 361 database consisting of an EST database from a previous study was used (Artigaud et al. 2014c) and a 362 compilation of the Uniprot database with Pecten maximus as the selected species. The variable 363 modifications allowed were as follows: carbamidomethylation of cystein, K-acetylation, methionine 364 oxidation, and dioxidation. "Trypsin" was selected as enzyme, and three miscleavages were also 365 allowed. Mass accuracy was set to 300 ppm and 0.6 Da for MS and MS/MS mode, respectively. Protein 366 identification was considered as unambiguous when a minimum of two peptides matched with a 367 minimum score of 20. False discovery rates were also estimated using a reverse database as decoy. 368 As well as carrying out analysis of variance for the two populations combined, we also ran separate 369 analyses of variance for each population. The overall effect of temperature, pCO2 and their interaction 370 on protein abundance in each population were tested through permutational multivariate analysis of 371 variance (Permanova) using the adonis2 function in vegan (Oksanen et al. 2019). Then separate 372 ANOVAs were fitted for each protein considering the effects of temperature, pCO2 and their 373 interaction, with P values adjusted using FDR. For all proteins with significant environmental effects 374 (FDR < 0.05), differences between the four treatments were quantified with post-hoc tests in 375 emmeans (Lenth 2020). 376 To provide a clearer view of population responses to environmental variation, we ran additional 377 exploratory and statistical analyses for each population separately using differentially abundant and 378 successfully annotated proteins from the combined population proteomic analysis. We initially looked 379 for correlations among proteins using principal component analysis (PCA), carried out in R using the 380 packages FactoMineR (Lê et al. 2008) and factoextra (Kassambara & Mundt 2020), with spot size data 381 scaled to unit variance. Correlations between proteins were identified by high loadings values (> 0.65 confidence ellipses (the multidimensional space in which we expect to find the mean 95% of the time, 385 given the underlying distribution of the data). 386 387

Differences in survival 389
Survival was significantly higher among Norwegian spat than French spat ( 2 = 154.22, df = 1, P < 390 0.0001). For Norwegian spat, temperature, pCO2, and their interaction did not affect survival. For 391 French spat, pCO2 did not affect survival (as a main effect or through its interaction with temperature); 392 however, temperature did have a significant effect ( 2 = 18.79, df = 1, P < 0.0001), with mortality 393 increasing with temperature. Survival curves are shown in Figure 1. 394 395

Variaition in whole-organism phenotypes 396
For the three primary traits of shell height, dry body weight and total shell weight, French and 397 Norwegian spat differed markedly in their responses to pCO2 and temperature, although the effect of 398 initial height was always highly significant (P < 0.0001). Among French spat, none of the primary traits 399 responded strongly to temperature (table 2; Fig. 2A, 2C and 2E), and while elevated pCO2 had a 400 positive effect on dry body weight (F = 4.92, df = 1, P = 0.041), it did not influence shell height or total 401 shell weight. On the other hand, pCO2 effects were much stronger among Norwegian spat, where they 402 interacted with initial height and temperature (table 2). For shell height, the pCO2 x temperature 403 interaction was significant during the model selection process (in which ML estimates were used; F = 404 5.40, df = 2, P = 0.014), but the interaction was not significant once optimal models were refitted using 405 REML estimates (F = 3.71, df = 2, P = 0.055). Yet the fact that the effects of temperature and pCO2 406 appear similar for all three primary traits in Norwegian spat (Fig. 2B, 2D and 2F) suggests that the pCO2 407 x temperature interaction for shell height, though weak, may be biologically meaningful. Thus, we 408 report the pCO2-dependent temperature contrasts and temperature-dependent pCO2 contrasts for all 409 three traits in table S1. At 19°C, elevated pCO2 had a significant negative effect on both shell height (t 410 = 2.68, df = 11.9, P = 0.020) and dry body weight (t =2.42, df = 11.6, P = 0.033), and the effect was 411 marginally non-significant for total shell weight (t = 2.12, df = 11.7, P = 0.056). In contrast, at 13°C, 412 elevated pCO2 resulted in a greater total shell weight (t = -2.24, df =12.8, P = 0.043). Furthermore, at elevated pCO2 there was a decrease in dry body weight at 19°C relative to 13°C (t = 3.75, df 12.0, P = 414 0.007), and at normal pCO2 there was an increase in total shell weight at 16°C relative to 13°C (t = -415 2.664, df 13.2, P = 0.0474). 416 Although condition index (CI) was based on the ratio of two of the primary traits (dry body weight over 417 total shell weight), it revealed new effects that were not identified from analyses of primary traits. 418 Specifically, analysis of CI identified a significant temperature effect in both French ( Fig. 2G; F = 5.90, 419 df = 2, P = 0.014) and Norwegian spat ( Fig. 2H; F = 16.63, df = 2, P < 0.001), with CI declining as 420 temperature increased, particularly when comparing 13°C and 19°C treatments (table 2, table S1). CI 421 also responded positively to elevated pCO2 in French spat (F = 16.92, df = 1, P = 0.001), mirroring the 422 result found in dry body weight. Furthermore, initial height was not a significant covariate in explaining 423 CI for Norwegian spat and had a weaker effect than temperature and pCO2 among French spat (F = 424 4.94, df = 1, P = 0.027). 425 426

Differential accumulation of proteins: separated populations analyses 473
The Permanova revealed that both temperature (F = 3.0444, df = 1, P = 0.002) and pCO2 (F = 3.3187, 474 df = 1, P = 0.007) significantly influenced overall patterns of protein abundance in French spat, but not interaction between temperature and pCO2 was not significant for either population. Among French 477 spat, individual ANOVAs for the 79 proteins revealed 12 with potential temperature or pCO2 effects 478 (FDR < 0.05; Fig. 6; table S3), six of which were also significant in the combined population analysis. 479 Fold changes were greater than 1.5 in eight of these proteins. Although none of the 12 proteins 480 showed a significant response to both temperature and pCO2 at our statistical threshold (FDR < 0.05), 481 differing responses to elevated temperature and pCO2 were suggested by post-hoc tests (table S4). 482 These showed that elevated temperature and pCO2 either had opposite effects that offset each other 483 when combined (Fig. 6 A-G), or similar effects that exacerbated one another additively (Fig. 6 H-I). 484 Conversely, among Norwegian spat only 11 proteins showed potential responses to temperature, 485 pCO2 or their interaction (P < 0.05; Fig. S2), and none of these were significant following correction for 486 multiple testing (FDR < 0.05). 487

488
The 12 environmentally dependent proteins in French scallops were ATP synthase, triosephosphate 489 isomerase (TPI), medium-chain specific acyl-CoA dehydrogenase (ACAD), complex I, Retinal 490 dehydrogenase 2 (RALDH2), MnSOD, PPIase, EPDR, an isoform of actin, an isoform of myosin, and 491 putative isoforms of Kyphoscoliosis peptidase (KY) and von Willebrand factor type A (vWA). For these 492 12 proteins only one main effect, temperature (nine out of 12) or pCO2 (three out of 12), was 493 significant at our stringent statistical cut-off (FDR < 0.05). Despite this, post-hoc tests indicated that 494 both variables frequently had an impact on protein abundance (table S3). For seven proteins (Fig. 6A-495 G), the most significant difference in abundance occurred between 13-highCO2 and 19-normCO2 496 treatments, while the comparison of 13-normCO2 and 19-highCO2 did not differ significantly. This 497 suggests that pCO2 and temperature had opposite effects that offset each other when both were 498 elevated. For all these proteins (except ACAD, Fig. 6A), increasing temperature had a positive effect 499 on abundance while elevated pCO2 had a negative effect. On the other hand, in two proteins (complex 500 I and PPIase; Fig. 6H-I) the 13-normCO2 and 19-highCO2 treatments differed the most, suggesting that 501 temperature and pCO2 both had additive negative effects on abundance. 502 503

Principal component analyses of protein abundance 504
To further explore correlations in protein abundance we carried out PCA using the 25 proteins that 505 were not annotated as actin, myosin or paramyosin. Correlations were stronger among French spat, 506 where the first two principal components (PC1 and PC2) together explained 54.7 % of the total had high loading values (> 0.65 or < -0.65) for PC1 (Fig. 7A), and treatments also showed separation 509 along this axis in the individual coordinate plot (Fig. 7C). The confidence ellipse for 19-normCO2 510 treatment was associated with higher PC1 values than any other treatment, and the confidence ellipse 511 for 13-highCO2 treatment was associated with lower PC1 values than ellipses for either 19-normCO2 512 or 19-highCO2. Transaldolase (TALDO), TPI_1, TPI_2, ATP synthase, RALDH1, RALDH2, glutathione S-513 transferase (GST), KY and vWA were positively correlated with PC1, while glucose-6-phosphate 514 isomerase (GPI), ACAD, and an isoform of NADP-dependent isocitrate dehydrogenase (IDH_1) were 515 negatively correlated with PC1. Among Norwegian spat, nine proteins had high loading values for PC1 516 (Fig. 7B), but there was no separation of treatments in the individual coordinate plot (Fig. 7D). 517 Variation in the influence of temperature and pCO2 on metabolism was also detected in the proteome. oxidative stress enzyme MnSOD and reduced abundance of the oxidative phosphorylation enzyme 572 complex I (also known as quinone oxidoreductase and NADH dehydrogenase), the proteome of French 573 spat showed far greater plasticity than that of Norwegian spat, again highlighting potential 574 evolutionary differences between the populations. Among French spat, temperature effects were 575 generally greater than pCO2 effects, but acidification frequently exerted a subtle effect in the opposite 576 direction to heating, with increased temperature and pCO2 offsetting one another. proteins were always more abundant in Norwegian spat. As they grow, scallop spat increasingly use 626 their adductor muscle for swimming, which can lead to an increase in muscle condition (Kleinman et 627 al. 1996). This could explain why Norwegian scallops (which were slightly older and larger at the start 628 of the experiment) had elevated levels of these proteins. While no isoforms of myosin or paramyosin 629 responded to temperature treatment, there was some evidence of pCO2 sensitivity in one isoform of 630 myosin (more abundant at elevated pCO2 in French spat) and one isoform of paramyosin (more 631 abundant at elevated pCO2 in the combined population analysis). Several recent studies from diverse 632 marine invertebrates have linked increases in myosin and/or paramyosin transcript (Wäge et al. 2016, 633 Bailey et al. 2017) and protein (Timmins-Schiffman et al. 2014, Zhao et al. 2020) abundance with the 634 response to elevated pCO2. While the mechanism by which myosin and paramyosin abundance aids Drawing on phenotypic results from whole organismal, metabolic, and proteomic scales, we show 640 clear differences in how French and Norwegian Pecten maximus spat respond to increases in 641 temperature and pCO2. Although some proteomic and organismal responses were common to both 642 populations, such as the increase in MnSOD and decrease in complex I abundance at high 643 temperature, or the corresponding decline in condition index, French spat seem to acclimate better 644 to temperature and pCO2 variation and more precisely adjust their energy metabolism than 645 Norwegian spat. By putatively altering their carbon metabolism to deal with increased redox stress 646 associated with higher temperatures, and by increasing oxygen consumption at elevated pCO2, 647 potentially to ensure cellular homeostasis, French spat appear better able to maintain growth under 648 OAW conditions. In contrast, Norwegian spat did not appear to fine-tune their proteome, but reduced 649 oxygen consumption if temperature or pCO2 increased. This corresponded with negative effects on 650 growth, with reduced body weight and shell height when high temperature and pCO2 were combined. 651 The experiments were carried out during July, when SST in the Bay of Brest is 2°C higher (16°C) (2021) found that urchins exposed to elevated pCO2 and high temperatures 658 after 4 weeks increased metabolism (similar to the French spat in our study), but that after 12 weeks, 659 the combined stress lead to reduced metabolism. The costs of maintaining metabolic function and 660 growth at elevated temperatures could also have contributed towards the reduced survival we 661 observed in French spat. These two populations are known to be genetically divergent (Morvezen et 662 al. 2015), with some genetic differentiation at loci associated with environmental variation in mean 663 annual SST and dissolved organic carbon (Vendrami et al. 2019). This could therefore suggest some 664 adaptive differentiation of these scallop populations in response to environmental variation.