Effect of ocean acidification on the early life stages of the blue mussel Mytilus edulis

4CNRS-INSU, Laboratoire d’Océanographie de Villefranche (UMR7093), BP 28, 06234 Villefranche-sur-Mer Cedex, France 2Univ. Pierre et Marie Curie-Paris 6 , Observatoire Océanologique de Villefranche, 06230 Villefranche-sur-Mer Cedex, France 3Univ. of Cambridge, Department of Earth Sciences, Cambridge CB2 3EQ, UK 4Roem van Yerseke BV, Gr. van Zoelenstraat 35, Postbus 25, 4400AA Yerseke, The Netherlands Netherlands Institute of Ecology (NIOO-KNAW), Centre for Estuarine and Marine Ecology, Postbus 140, 4400 AC Yerseke, The Netherlands 6NIOZ Royal Netherlands Institute for Sea Research, Department of Marine Organic Biogeochemistry, P.O. Box 59, 1790 AB Den Burg (Texel), The Netherlands 7Faculty of Geosciences, Utrecht Univ., P.O. Box 80021, 3508 TA Utrecht, The Netherlands


Introduction
The atmospheric partial pressure of CO2 (pCO2) will con tinue to increase with projected values for the end of this century ranging from 500 to 1000 patm, depending on the considered CO2 emission scenario (IPCC, 2007). Because about one third of anthropogenic CO2 emissions (from fos sil fuel, cement production and land-use changes) has been stored in the oceans since the industrial revolution , seawater pH has already declined by 0.1 unit compared with pre-industrial values (Orr et al., 2005) and it is projected to decrease by another 0.35 unit by the end of the century (Caldeira and Wickett, 2003). Ocean

Abstract.
Several experiments have shown a decrease of growth and calcification of organisms at decreased pH lev els. There is a growing interest to focus on early life stages that are believed to be more sensitive to environmental dis turbances such as hypercapnia. Here, we present experimen tal data, acquired in a commercial hatchery, demonstrating that the growth of planktonic mussel (.Mytilus edulis) larvae is significantly affected by a decrease of pH to a level ex pected for the end of the century. Even though there was no significant effect of a 0.25-0.34 pH unit decrease on hatch ing and mortality rates during the first 2 days of develop ment nor during the following 13-day period prior to settle ment, final shells were respectively 4.5 ± 1.3 and 6.0±2.3% smaller at pHnbs ~7.8 (pCC>2 ~ 1100-1200 patm) than at a control pHnbs of ~8.1 (pCC>2 ~ 460-640 patm). Moreover, a decrease of 12.0 ± 5.4% of shell thickness was observed after 15d of development. More severe impacts were found with a decrease of ~0.5 pHnbs unit during the first 2 days of development which could be attributed to a decrease of calcification due to a slight undersaturation of seawater with respect to aragonite. Indeed, important effects on both hatch ing and D-veliger shell growth were found. Hatching rates were 24±4% lower while D-veliger shells were 12.7±0.9% smaller at pHnbs ~ 7-6 (pCO2 ~ 1900 patm) than at a con-Correspondence to: F. Gazeau (f.gazeau@obs-vlfr.fr)^ acidification may have profound impacts on marine biota. Beside the direct effect of decreasing pH on the physiology and metabolism of marine organisms through a disruption of inter cellular transport mechanisms (see Pörtner et al., 2004 for a comprehensive review), calcareous organisms are par ticularly sensitive due to the decreasing availability of car bonate ions (CO3-) driven by increasing p CO2. Indeed, this generates a decrease of the calcium carbonate saturation state (S2): K, (D sp where K 'p is the stoichiometric solubility product, which is a function of temperature, salinity, pressure and the mineral phase considered (calcite, aragonite or high-magnesian calcite). Cold waters will become undersaturated with respect to aragonite (£2arag0nite < 1) in a few decades (Orr et al., 2005). Since the seminal paper of Broecker and Takahashi (1966) reporting a dependency of calcification rates on CaCCH sat uration state, several experimental studies have investigated the effect of a pCC>2 increase on the growth of calcifying organisms. Most studies have investigated primary produc ers (corals, coralline algae and coccolithophores) and have shown a very large range of responses Kleypas et al., 2006;Doney et al., 2009).
Among calcifying species, molluscans are very impor tant both in ecological and economical terms. Shellfish are ecosystem engineers governing energy and nutrient flows in coastal ecosystems, providing habitats for many benthic or ganisms and constituting an important food source for, for instance, birds, crabs, starfishes and fishes (Gutiérrez et al., 2003;Norling and Kautsky, 2007). Moreover, with an av erage annual increase of 7.7% over the last 30 years, global shellfish aquaculture production reached 13.1 million tons in 2008, corresponding to a commercial value of US $ 13.1 bil lion (FISHSTAT Plus vers. 2.31). The Pacific oyster (Cras sostrea gigas) was the most cultivated species in 2008 with a volume of 6.5 million tons or 9.5% of the total world aquaculture production while mussel production represented 1.9 million tons (US $ 390 million). A negative impact of ocean acidification on the growth of these species would, therefore, not only have major consequences for coastal bio diversity and ecosystem functioning and services, but will also cause a significant economic loss (Gazeau et al., 2007;Cooley and Doney, 2009 water column and, thanks to their internal energetic resources (lecithotrophic phase), develop to the ciliated trochophore stage and to the D-shaped veliger (shelled) stage within few days depending on the temperature conditions (Pechenik et al., 1990). These veliger larvae start to feed in the water col umn and gain weight until they reach the pediveliger phase (after few weeks) during which they try to find a place to settle. Larvae become competent to settle at a shell length of ~ 260 pm but can delay metamorphosis and remain in the planktonic compartment until they reach ~ 350 pm (Sprung, 1984). Once the settling conditions are favourable, meta morphosis occurs, plantigrade larvae attach to the substrate thanks to the secretion of the byssus and start to secrete the adult (dissoconch) shell.
Several studies have focused on the effect of projected pH levels on the growth of benthic (e.g. Gazeau et al., 2007: Ries et al., 2009) and planktonic (Comeau et al., 2009(Comeau et al., , 2010a molluscs. Most of these studies have demonstrated a negative effect of ocean acidification on the growth of these organisms although recent experiments (Ries et al., 2009) have sug gested a more complicated story with species-specific sen sitivities to decreasing pH levels and positive effects on cal cification rates in some cases. Early life stages of calcifying organisms are generally considered to be more sensitive to environmental disturbances (Raven et al., 2005). Moreover, amorphous calcium carbonate and aragonite have been iden tified as the main CaCCH mineralization form in molluscs larval stages (Medakovié, 2000). Therefore, as aragonite is 50% more soluble than calcite, these aragonitic larval stages are expected to be more sensitive to ocean acidification than calcitic organisms. Indeed, several recent studies have fo cused on the effect of ocean acidification on the early de velopment of mollusc species (Kurihara et al., 2007: Ellis et al., 2009: Parker et al., 2009: Talmage and Gobler, 2009: Watson et al., 2009 and most of them have reported negative impacts of decreas ing pH levels on the growth and development of these organ isms. So far, there have been no studies on the effect of ocean acidification on the larval development of the blue mussel {Mytilus edulis), the second most cultivated bivalve species in the world after Crassostrea gigas. Blue mussel aquacul ture is very important in The Netherlands and consists al most entirely of bottom-culture, carried out on leased sites in the Wadden Sea and in the Oosterschelde estuary (Smaal, 2002). In the Oosterschelde estuary, mussel beds (both wild and from aquaculture) play a major role in the cycling of nu trients and are able to filter the entire volume of the basin in 4-5 days (Prins and Smaal, 1994). In the last two decades, there has been an overall decline in available mussel seed due to intense fishing strategies that has forced local farmers to initiate the production of spats through hatchery techniques (Pronker et al., 2008). The present study aims to investigate the effects of future ocean pH levels on the development of Mytilus edulis early larval stages in a commercial hatchery.

Test animals and experimental conditions
To investigate the effect of rising atmospheric CO2 on mussel (Mytilus edulis) larvae, experiments were carried out in mesocosms at the commercial hatchery Roem van Yerseke (Yerseke, The Netherlands) between 18 October and 27 November 2007. A group of approximately 150 ripe, bottom-cultured mussels from the Oosterschelde, a tidal in let, were kept at a constant temperature (10 °C) for about 4 months. These animals originated from a same age-class and were fished in the tidal inlet and cultivated for about 2 years on commercial production plots. Before spawning, mussels (male and female) were cleaned with 1 pm filtered seawater and placed in a spawning tank. Mass spawning was initiated by rapidly raising water temperature from 10 °C to 19 °C. Fertilized eggs were retained on a submerged 30 pm sieve. During each experiment (see below), six enclosures were used, each of them containing 1301 of filtered (1 pm) seawater from the Oosterschelde. Three enclosures were continuously bubbled with ambient air (pCO2 ~ 380 patm) while the three others were bubbled with a mixture of ambi ent air and pure CO2. The flow rates of CO2 were regulated by means of digital thermal mass-flow controllers in order to reach the desired seawater pH.

Bioassay
In a first experiment (experiment #1), the effects of a pH decrease from ~8 .1 (control; pCC>2 ~ 460-640 patm) to, successively, ~ 7.8 G0CO2 ~ 1100-1200 patm) and ~ 7.6 (pCO2 ~ 1900 patm) were investigated during the first two days of development (from eggs to D-shape larvae). After fertilization (see above), embryos (5 7 .7 ± 4 .9 pm of diame ter) were counted, divided into 6 groups and transferred to the enclosures (3 controls, 3 low pH) at a density of approxi mately 10 embryos ml-1 . Embryos were maintained in batch conditions (no feeding, no water flowing) until the popula tion reached the D-veliger stage (initial development of the shell, reached in about 2 days).
In a second experiment (experiment #2), larvae were ex posed to pH values of ~ 8.1 and ~ 7.8 during the two weeks development period following the D-veliger stage. Em bryos were grown at environmental pH (~ 8.0-8.1) during 2 days, then counted and evenly transferred to the 6 enclo sures (3 controls, 3 low pH) at a density of approximately 10 embryos ml-1 . Cultivation period lasted for 13 days (day 2 to day 15 of development) until the population reaches the pediveliger stage. Larvae were fed in a continuous flow through system with a mixture of Isochrysis sp. (T-Iso, CCAP 927/14) and Chaetoceros muellerii (CCMP 1332) (2:1, based on cell counts) at a concentration of approxi mately 80 000 cells ml-1 . From day 4 to the end of the ex periment, larvae were fed with a mixture of Isochrysis sp., Pavlova lutherii (CCAP 931/1) and Chaetoceros muellerii (2:1:2, based on cell counts) at a concentration of approxi mately 150 000 cells ml-1 .

Sampling and analytical measurements
At the end of experiment #1 (day 2) and three times a week during experiment #2, the enclosures were emptied, cleaned with a mixture of diluted acetic acid and HC1 and rinsed with seawater. Water from the tanks was passed through a 90 pm sieve and larvae were concentrated in 21 jars. A sub-sample of 50 ml was fixed in a 5% neutralized-formalin seawater so lution to determine the larval abundance, hatching rates (% of D-veliger larvae) and size. After sampling during experiment #2, we made sure that pH was constant and at the desired pH level before reintroducing the larvae in the enclosures.
Larval abundance was estimated based on triplicate count ing of 500 pi sub-samples, under a binocular microscope. Larvae shell length (measured on 100 individuals) was mea sured (anterior to posterior dimension of the shell parallel to the hinge) under a microscope (20x; 0.0 1 pm precision in length measurement). Shell thickness was estimated from scanning electron micrograph (SEM) images acquired us ing the JEOL JSE 820 microscope at Cambridge Univer sity. Dried larval shells were mounted onto double sided carbon tape and sectioned using a flat edge needle. Loose organic matter and residual shell were removed with a dry paintbrush. Larval shells were removed using a wet paint brush, reoriented and remounted onto fresh tape attached to aluminium stubs and gold coated. Shell thickness was determined on 20 individuals of each replicate treatment. Hatching rates were defined as the percentage of "normal" D-shape larvae following the criteria proposed by His et al. (1997), after observation of a minimum of 500 larvae.
During the two experiments, pH, temperature and oxygen concentrations were continuously monitored in each enclo sure using Metrohm and Consort electrodes, which were cal ibrated daily on the N. B. S. scale for pH (pH 4 and 7). Salin ity was measured at the beginning of each incubation pe riod (3) in each enclosure. Daily measurements of total al kalinity (TA) in the 6 enclosures were performed by Gran electro-titration on 50 ml samples filtered on GF/F mem branes. Titrations of TA standard provided by A. G. Dick son (batch 82) were within 0.2 2 peqkg_1 of the nominal value (2334.8 ± 3.3 peq kg-1 ; n = 5). pCC>2 was com puted from pHnbs and TA using the software package CO2 1.1 (M. Frankignoulle) and the thermodynamic constants of Mehrbach et al. (1973). The solubility products for calcite and aragonite were from Morse et al. (1980).

Statistical analysis
For final shell length (experiment #1 and #2) and shell thick ness (only experiment #2), differences between replicates of each treatment as well as between control and low pHnbs treatments were tested using one-way ANOVA after testing for normality (Kolmogorov-Smirnov test). No significant differences were found in any of these parameters between the replicate tanks within each experimental condition. For hatching rates and abundances, as only one value was es timated per replicate, this latter was used to obtain grand means and standard deviations (SD) values for each treat ment. Since normality tests could not be applied due to the small sample size, differences of hatching rates between con trol and low pFInbs conditions were tested by means of un paired Student's t-tests using a Welch correction that does not assume equal variance between the two groups (Graphpad Instat software). For all tests, differences were consid ered significant at p < 0.05. In the following section, data are presented as means ± SD.

Results
The environmental parameters as well as parameters of the carbonate chemistry are shown in treatments were slightly outside the range pro jected for the end of the century (500-1000 patm) and must be considered as extreme conditions. The objective of the experiment was to test the effect of a ~0.3 pHnbs unit de crease on these organisms and, in that sense, experimental conditions were successfully set up and controlled.
During experiment #1, a seawater pHnbs decrease of 0.34 unit (pHnbs = 7.81; £2arag0nite = 1.38) had a significant ef fect on mussel larvae development (Fig. 1). Although no significant effect was found on hatching rates (unpaired Stu dent's t-test, p > 0.05), the average shell lengths at the end of the 2-day incubation period at pHnbs 7.81 were signifi cantly lower (ANOVA, n = 100, p < 0.001) than at higher (5 50 Fig. 1. Proportion of embryos that developed to D-shape larvae (A) and average length of D-shape shell (B) at the end of the two incubation periods during experiment # 1, in control (black bars) and low pH (white bars) seawater. During the first incubation, seawater pH^BS was maintained at 8.15 ±0.01 (Control-1) and at 7.81 ±0.01 (Low pH-1). During the second incubation, pH^BS lev els of 8.09 ±0.01 and 7.58 ±0.01 were used (Control-2 and Low pH-2 respectively). Errors bars represent standard deviations of the triplicate enclosures. **Significant difference between control and low-pH groups.
PHnbs-The relative decrease in shell length after 2 days has been estimated to 4.5 ± 1.3%. In contrast, a decrease of 0.51 unit (pH^BS = 7.58; ^aragonite = 0.81) had large effects on both the hatching and growth rates. The hatching rates de creased by 24 ± 4 % (unpaired Student's t-test, p < 0.001), while D-veliger shells were 12.7 ±0.9% smaller (ANOVA, n = 100, p < 0.001). From day 2 to day 15 (experiment #2, Figs. 2 and 3), a decrease of seawater pH^BS by 0.25 (pHnbs = 7.78; £2aragonite = E37) also did not have signifi cant effects on larvae survivorship (unpaired Student's t-test, ft = 3 ,/?> 0.05) while a significant effect was found for final shell length (ANOVA, n = 100, p < 0.001), corresponding to a relative decrease of 6.0 ±2.3% . This relative decrease of shell length was statistically significant after day-13 of de velopment. Growth rates, calculated as the difference in shell length between 2 sampling times divided by the time elapsed (d), decreased with increasing shell length (Fig. 2c) under both control and low-pH conditions. Statistically significant linear relationships between growth rates and initial shell length showed a shift to lower growth rates under low-pH conditions which was maintained throughout the experimen-

Discussion
In the past few years, several papers have reported on the impacts of seawater acidification on the growth and devel opment of shellfish early life stages. Kurihara et al. (2007Kurihara et al. ( , 2008 have demonstrated that a pH^ßs decrease to ~ 7.4 (-0.7 as compared to control values) caused a significant alteration of Crassostrea gigas and Mytilus galloprovin cialis early (up to 6 d) larval development, with significant decreases in hatching rates and shell growth. It has to be noted that at this pH level, which is much lower than the levels projected for the end of the century, the seawater was clearly undersaturated with respect to aragonite (^aragonite ~ 0.68). Parker et al. (2009)  . They actually showed that, although the growth of the 3 studied species (Mercenaria mercenaria, Argopecten irradians, and Crassostrea virginica) was neg atively affected, they did not exhibit the same sensitivity to a decrease of up to 0.6 pH unit. This species-specific sensi tivity to ocean acidification has also been observed by Miller et al. (2009) who showed that the development (from 96 h to ^ 30 d) and growth of the Eastern oyster (Crassostrea vir ginica) was significantly reduced at lowered pH levels (up to a 0.4 pHnbs unit decrease), while the Suminoe oyster (Cras sostrea ariakensis) did not appear to be sensitive to the same acidified conditions. In this study, we show that ocean acidification has a signif icant effect on the blue mussel larval development although the observed decrease in growth rates both in terms of length and thickness was not accompanied by a decrease of hatch ing rates and an increase in mortality rates as long as seawater remained oversaturated with respect to aragonite. Although no effect on hatching and mortality rates have been observed after 2d and after 15d of development, the consequences, in the field with the presence of predators, of a potential de crease of shell resistance and/or an augmentation of the time spent in the water column (delay in settlement) due to a re duction in growth as observed for a 0.25-0.34 pH unit de crease, are still unknown. Since the experimental period did not extend to the settlement and metamorphosis of the or ganisms, it is impossible to know if the observed decrease in growth rates would translate in a miniaturization of the spats and/or an increase of the time spent in the planktonic com partment. Nevertheless, both effects could have major con sequences for the survival of the populations. Suspensionfeeding benthic invertebrates can be important predators of pelagic larvae. In the Oosterschelde estuary, it has been shown that larviphagy from adult bivalves is a major source of mortality for bivalve larvae (Troost et al., 2009). However, several studies showed that, thanks to their shell, larvae could be rejected unharmed with the feces (Mackenzie, 1981). A reduction of the shell both in terms of length and thickness has therefore the potential to increase mortality rates during the planktonic larval stage. Finally, decreases in size dur ing the early developmental stages of marine organisms have been shown to effect juvenile fitness by reducing competi tive ability and increasing postsettlement mortality (Anil et al., 2001).
The conditions at which the larvae were exposed in our experiment must be regarded as optimal. In the field, mus sels usually spawn in spring when the water temperature is ~8 -1 8°C and chlorophyll-a concentrations vary between 0.5 and 19pgl-1 (April-June, 5 stations, monthly measure ments; see Table 1 for experimental levels). Therefore, as both experimental parameters were significantly higher than the ones encountered in situ at the time of spawning, the ex trapolation, to the field, of our laboratory-based observations on the effects of decreasing pH on the blue mussel larval de velopment, must be performed with caution. Indeed, Parker et al. (2009) have shown that the effects of ocean acidifica tion on the growth of Sydney rock oyster larvae were greater at sub-optimal temperatures. Moreover, as food availability is a very critical parameter in limiting larval development, the fact that, in the present experiment, food concentrations were optimal could have led to a high resistance of mussel lar vae to decreasing pH levels. The experimental pH level used for the control incubations in this study also does not reflect the conditions experienced by larvae in situ. Indeed, at the time of spawning, the spring bloom occurring in the Ooster schelde estuary, drives seawater /7CO2 to values below atmo spheric equilibrium corresponding to an average pH level of 8.27 ±0.09 (April-June, 5 stations, monthly measurements), a value much higher than the one used as a control during the incubations. In order to evaluate the potential effect of ocean acidification on this species fitness, there is a great need to conduct future experiments under conditions similar to the ones experienced by the organisms in the field.
At p Hnbs ~ 7.6, both hatching success and growth rates (as estimated by shell length decrease after 2d of devel opment) exhibited an important decrease coinciding with a F. Gazeau et al. : Effect of ocean acidification on the early life stages of the blue mussel M ytilus edulis 2057 slight undersaturation of seawater with respect to aragonite. At this point, it can only be speculated that the observed de crease in larval developmental success for a ~ 0.5 p Hnbs unit decrease is due to the seawater undersaturation with respect to aragonite. It must be stressed that our data do not al low discriminating between the physiological impact of pH decrease alone via a disruption of inter cellular transport mechanisms and the impact on calcification resulting from aragonite undersaturation, on the larval development of this species. More studies are needed to disentangle these differ ent aspects of the potential effect of ocean acidification on marine organisms.
As the different pressures exerted by the environment and predators in the field result in a considerable mortality, ap proaching 99% (Bayne, 1976) during the free-swimming lar val period, an additional 24% decrease in hatching rates as observed at a pHnbs of 7.6 can therefore compromise the survival of the population. Indeed, relatively small fluctua tions in the abundance of these larval stages are known to reg ulate the size of the population (Green et al., 2004). Shellfish predominantly inhabit coastal regions, which usually exhibit lower pH values than the open ocean because of permanent or episodic low pH water inputs from rivers (Salisbury et al., 2008), from upwellings (Feely et al., 2008) and due to intense rates of organic matter degradation and/or nitrification (Hof mann et al., 2009). While many estuaries already have high and variable p CO2, atmospheric CO2 enrichment will shift the baseline toward even higher values  that could lead to extended periods of undersaturation with respect to aragonite, although it has recently been suggested that, in some areas, eutrophication can counter the effects of ocean acidification (Borges and Gypens, 2010). There fore, these species will most likely be exposed to suboptimal growth conditions in the coming years. In order to assess socio-economic and ecological effects of ocean acidification on shellfish, it is therefore crucial to predict accurately the evolution of pH as well as the saturation state of the ocean and its coastal waters with respect to aragonite in the near future.
Our observation of no significant effect of a ~ 0.3 pHnbs unit decrease on both hatching rates and survivorship stands in contrast with results obtained by Parker et al. (2009) on Saccostrea glomerata during the first 48 h of develop ment and by Talmage and Gobler (2009) on Mercenaria mercenaria and Argopecten irradians between the veliger and metamorphosed stages (18-20 days). Our results are consistent with those from Talmage and Gobler (2009) on Crassostrea virginica who observed significant effects on growth rates but no significant effects on mortality for a ~ 0.3 pH unit decrease. Altogether, these different studies re veal, similar to what is observed for adult stages, that the ef fects of ocean acidification of molluscans early life develop ment are species-specific (Kurihara, 2008) and that the sen sitivity of the organisms might depend on the pH variability at which they are naturally exposed in the field.
It is important to notice that even under aragonite under saturated conditions, mussel larvae were able to produce a shell, highlighting that molluscs exert a control over calci fication (McConnaughey and Gillikin, 2008) and are there fore not completely dependent on environmental conditions. This does not appear as a surprise since most freshwater mol luscs are clearly well adapted to such conditions and bivalve growth has been showed by Tunnicliffe et al. (2009) under the extremely undersaturated conditions of deep hydrother mal sites. Most calcifying species, including molluscs, are able to concentrate Ca2+ and CO3-ions at the site of cal cification. Adult molluscs appear to use conventional cal cification physiology by pumping protons from the calci fication site (extrapallial fluid), largely through Ca2+/2H+ exchange catalyzed by Ca2+ ATPase (McConnaughey and Gillikin, 2008). The elevation of pH in the extrapallial fluid (Misogianes and Chasteen, 1979) allows an elevation of the concentration of CO3-that favours calcification. However, as this mechanism requires energy, this can lead to sub stantial energy shifts from other processes and to important costs for the growth of the organism as observed by Wood et al. (2008) for the brittlestar Amphiura ßliformis. Although the regulation of calcification by this mechanism is well doc umented for adults, few studies have focused on the mech anisms of larval calcification and on the capacity of bivalve larvae to regulate calcification rates by controlling the car bonate chemistry at the site of calcification. There is, how ever, some indication that biomineralization of Mytilus edulis larvae is physiologically controlled, as the activity of the car bonic anhydrase, an enzyme that catalyses the reversible hy dration of CO2 to HCO((" and H+ , reaches a maximum at the end of each developmental stage connected with biominer alization (Medakovic, 2000). This study also reported that these larvae, as showed for other molluscs and echinoderms larvae (Weiss et al., 2002), produce mainly amorphous cal cium carbonate during the first 2-3 days of development and aragonite in the following days. As the solubility of amor phous calcium carbonate is 30 greater than that of aragonite (Brecevic and Nielsen, 1989), early larval stages should be much more vulnerable than older larval stages and adults that precipitate aragonite and/or calcite. Again, the fact that 2days old larvae were able to produce a shell under aragonite undersaturation highlights the strong regulation capacity of these organisms under sub-optimal growth conditions.
As mentioned previously, in the Oosterschelde estuary, adults are exposed to a relatively narrow range of pH with winter pH levels never falling below ~ 7.9 and high pH lev els in springtime (~ 8.3) when spawning and larval devel opment occur. There is, therefore, a great need to evaluate the adaptive capacity of this species to low pH conditions. This could be achieved by comparing the responses, to a de crease in seawater pH, of populations originating from areas with contrasting environmental conditions with respect to the carbonate chemistry and/or by performing such experiments over several generations. F. Gazeau et al.: Effect of ocean acidification on the early life stages of the blue mussel M ytilus edulis Finally, in the present study, we show that shell increase, by linear extension, which is the most commonly measured parameter in ocean acidification related studies for mol luscs larvae, should not be the only measured parameter if one wants to investigate the effects of acidification on shell growth. Indeed, shell thickness appeared to be affected as well with a relative decrease twice the relative decrease ob served in shell length. This is consistent with  findings who estimated a much higher decrease of shell weight (estimated as the amount of calcium in the shells) than shell area (respectively -42 and -16% between pre-industrial and end of 21st century projected pH level for C. virginica) . However, it must be stressed that, in our study, shell thickness measurements could not be performed on Dveliger (2 days old) larvae and appeared to be limited to large pediveliger larvae using our protocol. Calcium content in the shells, as an estimator of shell weight, is also an interest ing parameter to follow and has been successfully applied by Miller et al. (2009). However, again, this technique has been applied to large larvae, and there is still a need to test its validity for smaller veliger larvae. Finally, more precise techniques such as 45 Ca labelling, recently used on pteropods (planktonic molluscs; Comeau et al., 2009) are promising and might be valuable tools to accurately evaluate the effect of ocean acidification within the range of expected levels for 2100 on calcification rates of mollusc early life stages.