Sensitivity towards elevated p CO 2 in great scallop ( Pecten maximus Lamarck ) embryos and fed larvae

The increasing amount of dissolved anthropogenic CO2 has caused a drop in pH-values in the open ocean known as ocean acidification. This change in seawater carbonate chemistry has been shown to have a negative effect on a number of marine organisms. Early life stages are the most vulnerable, and especially the organisms that produce calcified structures in 10 the phylum Mollusca. Few studies have looked at effects on scallops, and this is the first study presented including fed larvae of the great scallop (Pecten maximus) followed until day 14 post-fertilization. Fertilized eggs from unexposed parents were exposed to three levels of pCO2 using four replicate units: 465 (ambient), 768 and 1294 μatm, corresponding to pHNIST of 7.94, 7.75 (-0.19 units) and 7.54 (-0.40 units), respectively. All of the observed parameters were negatively affected by elevated pCO2: survival, larval development, shell growth and normal shell development. The latter was observed to be 15 affected only two days after fertilization. Negative effects on the fed larvae at day 7 were similar to what was shown earlier for unfed P. maximus larvae. Growth rate in the group at 768 μatm seemed to decline after day 7, indicating that the ability to overcome the environmental change at moderately elevated pCO2 was lost over time. The present study shows that food availability does not decrease the sensitivity to elevated pCO2 in P. maximus larvae. Unless genetic adaptation and acclimatization counteract the negative effects of long term elevated pCO2, recruitment in populations of P. maximus will 20 most likely be negatively affected by the projected drop of 0.06 – 0.32 units in pH within year 2100.


Introduction
The Intergovernmental Panel on Climate Change (IPCC) affirms that the uptake of anthropogenic CO 2 in the ocean has very likely caused elevated seawater CO 2 levels and thereby lowered the average oceanic pH values, termed ocean acidification (IPCC, 2013).A great effort is initiated worldwide to increase our knowledge of how ocean acidification affect coastal marine organisms; producing growing evidence that a high number of species respond negatively to exposure to elevated CO 2 levels (Kroeker et al., 2013).It is crucial to gain more knowledge about the effects on a range of marine organisms in order to get realistic projections of future changes to the marine ecosystems.
Calcifying organisms seem to be more sensitive to elevated CO 2 than non-calcifying organisms, and early life stages are more sensitive than older individuals (Byrne, 2012).Studies on bivalves, especially mussels and oysters, have reported negative effects on the pelagic early life stages (Gazeau et al., 2013;Kroeker et al., 2013;Parker et al., 2013).However, sensitivity to elevated CO 2 is species-specific and can vary greatly in closely related species, between populations and even between individuals (Arnold et al., 2009;Ries et al., 2009;Parker et al., 2011;Agnalt et al., 2013).The great scallop, Pecten maximus, is a commercially exploited species in several European countries (Brand, 2006;Norman et al., 2006;Strand and Parsons, 2006), found in coastal shell sand areas mainly at depths of 10-50 m, making this species highly exposed to changes in the coastal environment throughout its life cycle from the planktonic larvae to the benthic adults.To our knowledge only four studies have been published on the effects of elevated CO 2 on the great scallop, Table 1.Measured and calculated water parameters for three different pCO 2 groups (µatm) given as mean and standard deviation.Carbon chemistry values were computed based on daily measurements (0-14 days) of pH NIST in all replicates (n = 4), means of hourly temperature measurements in three tanks (n = 336), and mean salinity and total alkalinity based on two analyses per treatment at the start and end date (n = 6) in seawater running into the lab.one on unfed larvae (Andesen et al., 2013a), one on juveniles (Sanders et al., 2013) and two on adults (Schalkhausser et al., 2012(Schalkhausser et al., , 2014)).The studies of larvae and adults showed a negative effect of elevated CO 2 on the scallop, while the experiment using juveniles concluded that P. maximus is potentially tolerant to elevated CO 2 when food is unlimited.Blue mussel (Mytilus edulis) juveniles also seem to manage elevated CO 2 better at higher food concentrations (Thomsen et al., 2010(Thomsen et al., , 2013;;Melzner et al., 2011).Thomsen et al. (2010Thomsen et al. ( , 2013) ) showed that food availability outweighs acidification effects in juvenile M. edulis, and Melzner et al. (2011) showed that low food levels gave a negative effect on internal shell dissolution independent of pCO 2 levels in the same species.The importance of the energy budget under conditions of CO 2 stress was also shown in unfed 6-and 8-day-old sea urchin larvae (Strongylocentrotus purpuratus) when they lowered their ATP allocation to protein synthesis and in vivo Na + , K + -ATPase activity compared with fed larvae (Pan et al., 2015).Andersen et al. (2013a) showed that survival and growth in larvae decreased when CO 2 exposure started with fertilized eggs.In addition, the percentage of deformed larvae increased.This study covered the first 7 days of embryonic and larval development in unfed larvae.Since energy levels in larvae are critical for normal growth and development (Delaunay et al., 1992;Nevejan et al., 2003), the question was raised of whether a lack of food may have added stress to the larvae, making them more vulnerable to the elevated pCO 2 levels.In the present study we reared scallop larvae throughout a 14-day period with a feeding regime normally used in aquaculture production (Andersen et al., 2011(Andersen et al., , 2013b) ) aim-ing to elucidate whether larvae offered food would be more resilient to elevated CO 2 .

Materials and methods
Local broodstock of the great scallop Pecten maximus were collected in January 2013 from Hardangerfjorden on the west coast of Norway.They were transported to the experimental hatchery at the Institute of Marine Research (IMR) -Austevoll Research Station, cleaned of fouling and deployed in running seawater at an initial temperature of 8.8 • C. The broodstock was conditioned and spawned according to standard procedures at IMR (Andersen et al., 2013b(Andersen et al., , 2011) ) using the same seawater quality (ambient) as was later offered to the larval control group.Spawning was induced by a temperature increase on 6 March.P. maximus is a simultaneous hermaphrodite, and only egg batches with less than 10 % self-fertilization were used in the experiment.The final fertilization (cross and self) was around 87 %.
Fertilized eggs were incubated at a density of 13 eggs mL −1 in 38 L exposure tanks at ambient pH NIST 7.94 (control) and mean pH NIST of 7.74 and 7.54, corresponding to a pCO 2 of 465, 768 and 1294 µatm, respectively (Table 1).Four replicate tanks were used per pH treatment.The pH levels were chosen based on the predicted drop of 0.5 units in the open ocean from today to year 2250 (IPCC, 2013), and kept within the range used by Andersen et al. (2013a).IPCC (2013) has projected the pH levels by 2100 to be 0.06 to 0.32 lower than it is today.
The seawater supply and experimental design are described in detail in Andersen et al. (2013a), and an overview is shown in Fig. 1.The mesh sizes of the outlet sieves in the centre of the larval tanks were 35, 41 and 63 µm at days 0-2, 2-7 and 7-14, respectively.The measured water flow at day 1 was 6.6 ± 0.6 L h −1 (mean ± SD, n = 12), and at days 2, 7 and 8: 11.5 ± 0.7 L h −1 (mean ± SD, n = 36), corresponding to an exchange rate of 7.3 times day −1 .

Seawater parameters
Seawater was pumped from 160 m depth and filtered through a sand filter before temperature was adjusted in a heat pump.The water was aerated and finally filtered through a 50 µm filter.Temperature was recorded every 10 min using a four-detector (one in air and three in exposure tanks) EBI-1 Ebro 4 temperature logger.The overall mean temperature (±SD) calculated from recordings every 10 min in three tanks (at the three treatments) was 15.48 ± 0.16 • C (n = 3903).Daily means based on recordings every 10 min for each treatment (Table 1) were used to calculate pCO 2 values.Salinity was checked daily using a WTW LF330 conductivity meter.
The pH level in each exposure tank was measured daily in a 100 mL sample using a Mettler Toledo equipped with a Mettler Toledo InLab ® ExpertPro pH probe, calibrated with 4.00 and 7.00 buffers (CertiPUR ® buffer solutions, Merck KGaA, 64271 Darmstadt, Germany) traceable to standard reference material from NIST (NBS).The daily means for each treatment (Table 1) were used to calculate pCO 2 values.
Total alkalinity (A T ) was analysed in the three treatments at the start and end of the experiment (n = 6) by a TitraLab radiometer, and the mean value 2321.5 µmol kgSW −1 was used when calculating pCO 2 values.
The pCO 2 values (µatm) corresponding to the pH NIST values (Table 1) were calculated based on the means of temperature ( • C), pH NIST , salinity and A T , and using the macro taken directly from Ernie Lewis' "CO2SYS.BAS" basic program (Pierrot et al., 2006) with the set of constants K1 and K2 from Mehrbach et al. (1973) refitted by Dickson and Millero (1987), the constant for KHSO 4 from Dickson and Millero (1987) and that for total boron (B T ) from Uppstrom (1974).Seawater at different pH levels was produced by mixing seawater with an acid stock solution of pH NIST 5.80, made from mixing CO 2 gas and seawater with an ambient pH NIST of 7.95.The pH in each mixing tank was continuously adjusted to pre-set levels by addition of stock solution with dosage pumps (IWAKI) controlled by feedback from pH electrodes to pH transmitters (Endress & Hauser).

Sampling of larvae
Larvae were sampled by collecting 700 mL from each replicate at days 2, 3 and 7.At day 14 all tanks were drained and a total sample from each replicate was collected.Sampling and preservation in 4 % formalin is described in detail in Andersen et al. (2013a).Larvae in the samples were preserved to measure shell size and survival and to determine whether the larval shell was normally developed or deformed.
To calculate the survival based on the initial number of fertilized eggs, larvae concentrated in smaller volumes were counted in 4 × 200 µL droplets at days 3 and 7, and in 10 × 50 µL droplets at day 14.To make our results comparable with the literature on scallop spat production (Andersen et al., 2011;Magnesen et al., 2006) we estimated survival at days 7 and 14 also based on the day 3 yield.
Shell length (parallel to the hinge) and shape was investigated using photos, as described in Andersen et al. (2013a).The number of individuals classified as "live" that was measured from each replicate was 74-104 at day 3, 67-97 at day 7 and 92-156 at day 14.
At day 2 larvae that had not yet developed visible shell valves were classified as "unshelled" larvae.At day 3 larvae that had not developed the muscle to retract the velum were identified, after being preserved, by the presence of a protruded velum.These were classified as larvae with a "protruded velum".The larvae were not considered to be abnormal and the protruded velum not an artifact of preservation, but they were considered to be larvae that had developed slower than shelled larvae or larvae without a protruded velum, respectively.
Larvae at days 3, 7 and 14 were classified into four categories according to shell shape (Andersen et al., 2013a): (1) normal, (2) hinge deformity, (3) edge deformity, and (4) both (edge and hinge deformity).Trochophore larvae at day 2 were only classified as category 1 and 2 (normal and hinge deformity), since the shell edge was often hidden by soft tissue.The number of live larvae classified per replicate (independent of treatment) at day 2 was 96-150, and at days 3, 7 and 14 it was similar to the number of live individuals used for shell length measurements.Larvae were classified as "live" when the shell was filled with soft tissue, and as "dead" when the shell was empty or contained little soft tissue.Deformities in dead larvae were classified only at day 14 based on 29-181 dead individuals (out of 122-333 individuals) from the different replicates, since there were too few dead individuals in the larval samples at days 3 and 7. To describe the relative variation in shell shape categories between replicates for the three treatments, the coefficient of variation (CoV) was calculated as percentage SD of means:

Statistics
To find effects of pCO 2 the parametrical tests one-way ANOVA (ANOVA) and a generalized linear model (GLM) followed by Tukey's HSD post-hoc test to find differences between groups were used if the data or transformed data conformed to normality using Shapiro-Wilk's W test and the variances were homogeneous according to Levene's test.The effect of days on the normal shell category was tested using GLM.A t test was used to determine whether there were differences in shell shape (normal or deformed) between live and dead larvae.Results given as percentages (survival and shell shape categories) were arcsine transformed prior to testing.When parametrical tests were inappropriate, Kruskal-Wallis ANOVA by ranks (K-W ANOVA) was used to test effects of pCO 2 , and differences between groups were then tested using p values for multiple comparisons (twotailed).To find whether normally developed larvae at day 7 post-spawn were different from day 14, the Kolmogorov-Smirnov test was used for the two elevated pCO 2 groups.To find differences between groups, a non-parametric t test, the Mann-Whitney U test, was used when multiple comparisons did not show differences between groups even if the Kolmogorov-Smirnov test showed significant effects.The significance level used in all tests was set to 0.05.Statistica version 11 (Statsoft Inc.) was used to run all statistical tests.

Survival
The mean survival at day 3 post-spawn based on the initial egg count varied between 27.6 and 31.1 % for the three pCO 2 groups (Fig. 2).After day 3 survival decreased more at 1294 µatm than at lower pCO 2 and was only 3.1 % by day 14, while it was 18.2 and 14.9 % in the ambient group (465 µatm) and at 768 µatm, respectively.Mean survival was lower at 768 µatm than in the ambient group (465 µatm) at all days, but the differences were not significant.Significant differences in survival between pCO 2 groups were only found between the highest pCO 2 and the other two groups at day 7 (p = 0.001 for 465 and 1294 µatm; p = 0.028 for 768 and 1294 µatm) and day 14 (p < 0.001 for all).
Based on survival at day 3 the present study gives a survival at day 14 of 62, 57 and 10 % for the groups at 465, 768 and 1294 µatm.

Shell development and length
Larvae that had not yet developed a visible shell (unshelled larvae) at day 2 ranged from 4.0 % in the ambient group to 9.8 and 23.4 % in the 768 and 1294 µatm groups, respectively (Fig. 3).A significant difference in unshelled larvae was only found between the lowest and highest pCO 2 (p = 0.032).At day 3 no unshelled larvae were observed, but larvae with a protruded velum were observed in all groups (Fig. 3).The percentage of larvae with a protruded velum was affected by pCO 2 (p = 0.025) and increased from 4.6 and 4.5 % in the ambient and 768 µatm group, respectively, to 33.7 % at 1294 µatm.Larvae with a protruded velum were also found at day 7, but in only 3 out of a total of 1134 individuals, and were not observed at day 14.
At day 3, shell length (SL) of larvae was 109.9, 107.2 and 94.6 µm at 465 (ambient), 768 and 1294 µatm, respectively (Fig. 4).However, shell growth at 768 µatm was slower than in the ambient group and the difference in SL between the two increased until day 14, when the 768 µatm group was similar in shell length to the 1294 µatm group (Fig. 4), with average values of 123.7 and 122.7 µm, respectively.At day 14 SL of the ambient group was 134.4 µm.
There was an effect of pCO 2 at all days post-spawn (day 3: p < 0.001; day 7: p = 0.002; day 14: p < 0.001).All pCO 2 groups were significantly different in SL at day 3 (p < 0.013  for all groups), at day 7 only SL at the highest pCO 2 was significantly different from the ambient group (p = 0.002), but at day 14 the SL of both elevated groups was significantly different from the ambient group (p < 0.001).The mean shell growth rate for the days 3-14 was 2.2, 1.5 and 2.5 µm day −1 in the pCO 2 groups 465, 768 and 1294 µatm, respectively, and it was significantly lower in the 768 µatm group than in the other two groups (p < 0.004 for both groups).Larvae in the two elevated pCO 2 groups showed different shell growth patterns, with a higher daily growth rate at the most elevated level, but starting from a smaller SL at day 3 (Fig. 4).

Shell shape categories
On day 2, 90, 89 and 65 % of the live larvae at ambient, 768 and 1294 µatm, respectively, had a normal hinge, and the most elevated group was significantly different from the other two (p < 0.001 for both).The percentage of live larvae classified into the category normal at day 3-14 decreased when pCO 2 increased independent of days, and the values were lowest at day 7 in all groups (Fig. 5a-c), but there was no significant effect of days on normal in any of the treatments.The range of normal in the ambient group ( 437 for day 3-14 was 71-80 %, and the effect of pCO 2 was significant on all days (p < 0.001 for day 3; p = 0.007 for both day 7 and day 14).Only the most elevated group was significantly different from the ambient at all days (p < 0.001 for day 3; p = 0.005 for both day 7 and day 14), except at day 3, when all groups were significantly different (p = 0.043 for 465 and 768 µatm, p < 0.001 for both 465 and 768 µatm, as well as 1294 µatm).
Edge was the most frequently occurring deformity category at day 3-14, observed in 15-25 % of the ambient larvae and in 32-63 % of the elevated groups (Fig. 5a-c).The hinge category was always highest in the most elevated CO 2 group with a range of 5-30 % at day 3-14 (Fig. 5a-c), and the values were mostly more than double the values in the other two groups (range 2-6 %).Few larvae were classified into the category both, but the ranges increased with an increase in pCO 2 level: 0.3-1.5, 1.1-7.0 and 9.4-31.7 % at 465, 768 and 1294 µatm, respectively (Fig. 5a-c).
The variation between replicates in the different shell categories was relatively high in the two groups at elevated CO 2 .The coefficient of variation (CoV) ranged between 2.4 % at day 2 (at 768 µatm) and 115.5 % at day 7 (at 1294 µatm).For the three deformity categories edge, hinge and both, the CoV range was 18.9-146.4%.In general, the variation was highest at day 7.In two out of four replicates in the most elevated group at day 7 we did not observe any live larvae in the category normal (n = 94 and 83), but in the same replicates there were normally shaped dead larvae (6.3 and 20.0 %).
In dead larvae at day 14 normal decreased from 55 % at 465 µatm (ambient) to 25 % at 1294 µatm (Fig. 5d).There was an effect of pCO 2 on normal (p = 0.018), and 465 µatm (ambient) was significantly different from 1294 µatm (p = 0.018).Edge was also the major category observed in dead larvae, and ranged from 38 % in the ambient group to 54 % at 1294 µatm (Fig. 5d).Shell deformity in dead larvae was significantly higher than in live larvae at day 14 (Fig. 5c and  d) in the ambient (p = 0.011) and the 768 group (p < 0.001), but not in the most elevated group.

Discussion
The present study shows the effects of elevated pCO 2 on Pecten maximus embryos and fed larvae during a 14-day period, approximately two-thirds of the larval life cycle.Although larvae may experience a lack or scarcity of food in their natural environment, it is important to know whether sufficient food supply decreases larval sensitivity towards elevated pCO 2 as the early life stages of bivalves seem to be very sensitive to elevated levels of CO 2 (Fabry et al., 2008;Kurihara, 2008;Talmage and Gobler, 2009, 2010, 2011;Parker et al., 2010;Beniash et al., 2010;Gazeau et al., 2010;Gaylord et al., 2011;Andersen et al., 2013a;White et al., 2013White et al., , 2014)).A number of studies have shown effects on marine invertebrate larvae (Brennand et al., 2010;Crim et al., 2011;Stumpp et al., 2012), but only Andersen et al. (2013a) have studied P. maximus embryos and larvae.Andersen et al. (2013a) presented a comparison of studies concluding that the responses to elevated pCO 2 seem to vary little between bivalve species, but the magnitude of the responses may differ.In addition, the pCO 2 level, temperature and rearing volume in the studies vary, and one should therefore be careful in drawing conclusions about effects between studies.

Food availability
Scallop larvae were fed about 6-10 cells µL −1 , which was within the standard range described by Magnesen et al. (2006) and Andersen et al. (2013b).Andersen et al. (2013b) showed that feed concentrations of 3-20 cells µL −1 in large rearing volumes (2800 L) affected the lipid content in the larval populations but not larval survival or total yield of juveniles 4 weeks after metamorphosis.Based on these results feed concentrations in the present study should be sufficient to maintain growth and survival at this larval concentration and age.

Survival
In the present study survival in the ambient and 768 µatm groups was not significantly different throughout the experimental period, while survival in the 1294 µatm group was lower than in the ambient group at day 7 and day 14.When pCO 2 increased from 465 (ambient) to 1294 µatm the survival decreased with a factor of 0.3 on day 7 and 0.2 on day 14, similar to what was observed in Andersen at al. (2013a).The negative effect of elevated pCO 2 on survival relative to the ambient group was similar in both studies, thus we could not detect any positive effect of feeding.Mean survival at day 7 in the ambient group was 21 % of the initial egg count, less than half of that reported by Andersen et al. (2013a) (45 %) using the same rearing system without food supply.The differences in survival of the ambient groups between the two studies may have been caused by slight changes in experimental conditions or by parental effects between larval groups (Andersen et al., 2011).
The incubated eggs yielded 28-31 % larvae collected on a 35 µm mesh screen at day 3 post-spawn, independent of pCO 2 treatment.The survival until day 3 in our study was on the lower side of what was earlier reported for larger rearing systems (Andersen and Ringvold, 2000;Andersen et al., 2000;Magnesen et al., 2006), and may be due to the smaller rearing volumes or other unknown factors.Survival of more than 60 % at day 14 in the ambient group, based on the larval count at day 3, was in accordance with survival in batches of viable P. maximus larvae in large-scale hatchery rearing systems (Andersen et al., 2011(Andersen et al., , 2013b)).This indicates that the veligers in our study were viable and healthy.The survival at day 14 based on day 3 fits well for the two lowest pCO 2 groups with the relationship between growth rate and survival that was shown in Marshall et al. (2010), but survival at the highest pCO 2 was much lower than they described.

Larvae development and shell size
At the earliest shelled stage, the muscles are not sufficiently developed and thus the larvae are unable to retract the velum (Cragg, 2006, and references therein).The percentages of unshelled larvae at day 2 and larvae with a protruded velum at day 3 were significantly higher in the 1294 µatm group, which is most likely a result of delayed development caused by elevated pCO 2 , is in accordance with the reports of slower development at elevated pCO 2 levels reported in the earlier study of great scallop larvae (Andersen et al., 2013a) and also in other bivalve larvae (Talmage and Gobler, 2011;Kurihara, 2008).
Larvae in the ambient group and at 768 µatm were larger than larvae at 1294 µatm at days 3 and 7.At day 14, larvae at 768 µatm were smaller than in the ambient group, but similar in size to the larvae in the 1294 group.The change in growth rate after day 7 as seen in the 768 µatm group has not been shown for scallop larvae earlier, as Andersen et al. (2013a) ended their study at day 7.This indicates that the growth rate in larvae in the 768 µatm group may have been able to compensate to some extent for the elevated pCO 2 level for the first 7 days.Later the growth rate was affected by the carbonate chemistry, even if survival seemed unaffected.It may then be worth discussing whether the endogen reserves for the current larval groups were insufficient for embryo and first-stage larvae development at the highest pCO 2 level, since survival decreased and the remaining larvae developed slower compared to the two lower pCO 2 groups.Andersen et al. (2013b) showed that larvae supplied with no food after day 3 stopped growing at day 6, while the survival was affected only after day 15.This indicated that endogen reserves from the eggs (Cragg, 2006) were important for larval growth and development until day 6, and that exogenous energy from food would be more important after that day.The change in growth rate after day 7 for the 768 µatm group also shows the importance of longer exposure duration to find indications of future success for larval groups.Larvae in the 1294 µatm group developed slower than the other two groups until day 3, and the smaller size was not compensated for until day 14, when the experiment ended.This was also shown by White et al. (2014) when they exposed one group of bay scallop larvae, Argopecten irradians, to elevated pCO 2 from 11 h post-fertilization and for only 3 days.The negative effect on size was not compensated for when larvae were transferred back to ambient conditions, although they were pulsefed daily.Our larvae in the 1294 µatm group reach the same size at day 14 as larvae in the ambient group reached at day 7.
Shell length for the two lowest pCO 2 groups in our study was only slightly higher than reported by Andersen et al. (2013a) at day 7 (119 µm in our ambient group vs. 115 µm in theirs, and 115 µm in our 768 µatm group vs. 110 µm in their 821 µatm group), and shell length was the same for the most elevated groups (105 µm).The similar shell length at day 7 for the two larval batches suggests that their shell growth rate was similar.Andersen et al. (2013) showed that food availability did not affect larval shell growth the first 6 days after fertilization, supporting the idea that feeding probably did not cause any difference in growth rate between the two larval batches.

Shell deformities
The effect of pCO 2 on shell shape of day 2 larvae seemed to be less in our study compared with Andersen et al. (2013a).The range of normally developed hinges in the present study was 65-90 %, while Andersen et al. (2013a) reported 28-68 %.Since food was distributed after day 2, this was not www.biogeosciences.net/14/529/2017/Biogeosciences, 14, 529-539, 2017 due to feeding, but it may have been caused by other factors such as genetic variation or energy status.It is known that the variation in performance between larval batches of P. maximus is very high (Andersen et al., 2011), and a variation in egg quality have been suggested as an explanation by Robert and Gérard (1999).The proportion of deformed live larvae in the ambient group was 20-29 % at days 3-14.This may seem high in comparison to His et al. (1997), who found that it was 5-10 % in the control groups of 18 and 48 h old embryos or larvae of the oyster C. gigas and mussel M. galloprovincialis, respectively, when kept in static seawater in 25 mL containers.A relatively high deformity level in the ambient group indicates that other factors than the treatment may also cause physiological stress to the embryo and larvae.However, the deformity level in our larvae at ambient pCO 2 was lower than the nearly 40 % reported for unfed scallop larvae in the same rearing system at day 7 (Andersen et al., 2013).This could indicate that food counteracts deformity at ambient levels of pCO 2 , or it could simply be a natural difference between larval groups due to genetics or parental effects.
The different types of deformity seemed to show different patterns with an increase in pCO 2 .Edge deformity was higher in the two elevated groups than in the ambient group, while hinge deformity was higher in the most elevated group than in the two other groups.The percentage of normal followed a more linear pattern with an increase in pCO 2 independent of days.The percentage of larvae in the edge category in our study on day 7 was similar to the percentages reported by Andersen et al. (2013a), 25-63 % in ours vs. 30-57 % in theirs.In hinge our percentages were lower (2-5 % vs. 5-22 %) and in both our percentages were higher (1-32 % vs. 1-10 %).Again, differences between family groups may explain the variation found between the two experiments.The high CoV also indicates that factors differing between replicates may partly contribute to deformities.However, the present study shows that more larvae develop shell deformities with elevated pCO 2 and that feeding does not seem to counteract this effect.
The aragonite saturation at 0.82 can possibly add energetic stress to the larvae, since calcium carbonate dissolves at saturation below 1 (Andersson et al., 2011).However, the carbonate shell in live larvae is covered by a protein layer, the periostracum (Mouëza et al., 2006;Silberfeld and Gros, 2006), and the effect of a reduced aragonite saturation may not be significant.

Concluding remarks and future work
Scallop embryos and larvae seem highly sensitive to elevated pCO 2 at a very early stage in life.Our study confirms that even when food is supplied, increased pCO 2 levels that may be reached within the next 50-100 years in the open ocean (Zondervan et al., 2001;IPCC, 2013) have a negative effect on scallop larvae similar to that found in unfed larvae (Ander-sen et al., 2013a).In our study, negative effects on survival were observed after 7 days, size was affected after 3 days at the highest pCO 2 and development rate and normal shell development were affected after only 2 days at 1294 µatm.This shows that slow development rate and abnormal shell development are very early indicators of sensitivity towards stressors like elevated pCO 2 .If this sensitivity is also true for natural populations, it could have serious implications for recruitment of natural populations and for future aquaculture production.In aquaculture, however, scallop spat is produced in land-based nurseries, where seawater quality can be adjusted.
Our study ended at day 14, around 1 week prior to metamorphosis.Future studies should focus on how well exposed larvae groups succeed through the energy-demanding process of metamorphosis, as this may be one of the main bottlenecks in the recruitment process since low survival in this stage is shown in spat production (Andersen et al., 2011).Based on earlier studies (Parker et al., 2011(Parker et al., , 2012;;Suckling et al., 2014;White et al., 2014;Jager et al., 2016) it seems that a better understanding of energy budgets, genetic drifting and adaptation, and effects of parental exposure on both gametes and larvae will be crucial to predicting recruitment success in scallop populations exposed to an increasing ocean acidification.
In most studies, pCO 2 levels are kept constant during the experimental period, while in the natural environment the levels fluctuate on both shorter and longer temporal scales, especially in coastal areas.Fluctuations may cause extra stress to organisms exposed to a rising pCO 2 in the environment (Almén et al., 2014), and should be included in future experiments.Not only are the fluctuations different between open ocean and coastal areas, but average levels in coastal areas are also different from open ocean levels.The few reports on the situation in near-shore waters show pH values as low as 7.6, already exceeding the expected average values for the open ocean within year 2100 (e.g.Thomsen et al., 2010;Gazeau et al., 2011;Reum et al., 2014).These data are so far based on very few coastal monitoring stations, and effort should be made to increase the monitoring of highly productive coastal areas in the future to reveal the pCO 2 levels the coastal epibenthic species are in fact exposed to.

Data availability
The underlying research data for survival, shell length, larval development, shell shape categories and seawater chemistry can be accessed in the Supplement.
The Supplement related to this article is available online at doi:10.5194/bg-14-529-2017-supplement.

Figure 1 .
Figure 1.An overview of the experimental design showing both the mixing room for production of seawater with different levels of pCO 2 and the exposure room with fibreglass tanks for larvae.

Figure 5 .
Figure5.Larvae classified into four different shell categories: normal, edge (deformities), hinge (deformities) and both (edge and hinge deformities) at three pCO 2 levels -465 (ambient), 768 and 1294 µatm -in live larvae at day 3, day 7, day 14, and in dead larvae at day 14.The figure is based on the sum of four replicates in each pCO 2 group.

Table 2 .
Larval food concentration sampled from inside the tank and at the outlet.Values are means and standard deviation (n = 12).