Transgenerational plasticity responses of oysters to ocean acidification differ with habitat

Transgenerational plasticity (TGP) and its role in the rapid acclimation response of marine organisms to climate change stressors has been the focus of many recent studies (Munday, 2014; Ross et al., 2016; Donelson et al., 2018; Byrne et al., 2020). A form of non-genetic inheritance, TGP occurs when the environment experienced by the parent alters the phenotype of the offspring (Ross et al., 2016; Byrne et al., 2020). This phenotypic change in the offspring can prime their physiology to better match the environmental conditions they will face. Several studies report the beneficial effects of parental exposure to climate change stressors for offspring (Donelson et al., 2012; Miller et al., 2012; Parker et al., 2012, 2015, 2017a; Suckling et al., 2014; Putnam and Gates, 2015; Thor and Dupont, 2015; Diaz et al., 2018; Zhao et al., 2018; Putnam et al., 2020; Spencer et al., 2019). Donelson et al. (2012) found that the negative effects of elevated temperature (+1.5 and +3.0°C) on the aerobic scope of the tropical damselfish, Acanthochromis polyacanthus, were completely ameliorated following parental exposure to elevated temperature. Miller et al. (2012) found that the negative impacts of elevated carbon dioxide (CO₂) (1032 µatm) on the survival, growth, weight and resting metabolic rate of juveniles of the clownfish, Amphiprion melanopus were ameliorated following parental exposure to elevated CO₂. Furthermore, Parker et al. (2012) found that parental exposure of the Sydney rock oyster, Saccostrea glomerata, to elevated CO₂ (856 µatm) had beneficial effects on offspring, with faster growth and development rates of larvae and a reduction in the number of abnormal larvae.

Not all studies report beneficial effects of TGP; some studies report detrimental or neutral (no change) TGP outcomes (Dupont et al., 2013; Uthicke et al., 2013; Welch et al., 2014; Chakravarti et al., 2016; Griffith and Gobler, 2017; Bellworthy et al., 2019; Karelitz et al., 2019; Kong et al., 2019; Venkataraman et al., 2019). Chakravarti et al. (2016) found parental exposure of the polychaete, Ophryotrocha labronica, to elevated temperature (+3°C) led to negative effects for the fecundity and volume of eggs, and hatching success of their offspring. Griffith and Gobler (2017) found that parental exposure of the clam Mercenaria mercenaria and the scallop Argopecten irradians to elevated CO₂ (pH 7.6) resulted in larvae that were smaller and had lower survival at elevated CO₂ compared with larvae from non-exposed parents. Uthicke et al. (2013) found, moreover, that parental exposure of the sea urchin Echinometra mathaei to elevated CO₂ (pH 7.7) for 7 weeks had neutral TGP effects on their larvae. Offspring showed no improvement in larval skeletogenesis or abnormality from parents exposed to elevated CO₂ when compared with larvae released from the control parents.

Collectively, these studies show that the nature of TGP responses on offspring performance traits (i.e. beneficial, detrimental or neutral) differs among species. Largely unknown, however, is whether TGP responses to climate change stressors may differ within a species dependent on the environmental conditions experienced in their habitat (Boyd et al., 2016; Byrne et al., 2020). Many marine species have a wide distribution and occupy diverse habitats (e.g. coastal versus estuarine zones; upwelling versus non-upwelling zones; high versus low latitudes; subtidal versus intertidal zones; Boyd et al., 2016; Byrne et al., 2020). The environmental conditions experienced in these habitats can vary dramatically (Smith et al., 2009; Padilla and Savedo, 2013; Boyd et al., 2016). For example, marine species inhabiting both estuarine and coastal habitats may experience vastly different variations in salinity stress. Furthermore, species inhabiting both upwelling and non-upwelling zones will experience variations in the frequency and severity of periods of reduced pH and hypoxia (Kelly et al., 2013).

Previous studies have found that offspring from parents that experience episodes of reduced pH and/or elevated temperature in their environment have a greater tolerance to climate change stressors (Ghalambor et al., 2007; Boyd et al., 2016; Chevin and Hoffmann, 2017; Gaitán-Espitia et al., 2017; Hoshijima and Hofmann, 2019). For example, Gaitán-Espitia et al. (2017) showed that larvae of the sea urchin, Loxechinusalbus, from adults that experienced local variation in seawater pH and P_CO2 had greater tolerance to ocean acidification. In the urchin Strongylocentrotus purpuratus, adults living within kelp forests that experienced stronger, more predictable episodes of diurnal variation in dissolved oxygen and pH produced larvae that were more tolerant to ocean acidification than parents living outside the kelp forest (Hoshijima and Hofmann, 2019). As our oceans continue to acidify and warm, however, parents living in these habitats will be exposed to the combined effects of natural episodes of reduced pH and/or elevated temperature, and climate change stressors. Whether parental exposure to climate change stressors in these habitats will further enhance or diminish the TGP response of offspring, remains unknown.

Marine molluscs are one of the most vulnerable groups to climate change, particularly during their early life-history stages (Ross et al., 2011; Gazeau et al., 2013; Parker et al., 2013) but have demonstrated capacity for beneficial TGP (Byrne et al., 2020). Along coasts all over the world, molluscs are found in intertidal zones. In the intertidal zone, sessile bivalve molluscs experience episodes of emersion (air exposure) daily as the tide rises and recedes. During low tide, bivalves close their valves, restricting gas exchange. As a result, metabolically induced CO₂ accumulates in the animal, causing hypercapnia and acidosis of the extracellular fluid (Burnett, 1988; Truchot, 1990; Scanes et al., 2017). This metabolically induced hypercapnia and acidosis are very similar to the acid–base changes experienced by bivalves during exposure to ocean acidification (Gazeau et al., 2013; Parker et al., 2013). Previous research has shown that ocean acidification can exacerbate the hypercapnia experienced by organisms in intertidal habitats (Scanes et al., 2017). Despite this, studies to date that have investigated the TGP response of molluscs and other marine organisms to climate change stressors have been done only under subtidal conditions, where adults are continuously immersed (Parker et al., 2012, 2015, 2017a; Thomsen et al., 2017; Diaz et al., 2018; Kong et al., 2019; Zhao et al., 2019).

Here, we investigate the impact of habitat on the TGP response of marine bivalves to ocean acidification, using a model organism, the Sydney rock oyster, Saccostrea glomerata. The Sydney rock oyster is distributed along the south east coast of Australia where it is found from shallow subtidal to high intertidal habitats in bays and estuaries. In the state of New South Wales, S. glomerata forms the largest and oldest aquaculture industry worth approximately 34 million US dollars per annum. Previous studies have found that parental exposure of S. glomerata to elevated CO₂ under subtidal conditions has beneficial TPG outcomes for larval offspring with improved tolerance to elevated CO₂ (Parker et al., 2012, 2015, 2017a). For adult oysters living in the intertidal zone, metabolic hypercapnia is experienced during episodes of emersion, with extracellular pH (pH_e) being approximately 0.75 pH units lower than during episodes of immersion (Scanes et al., 2017). We used a two-part experiment to test the hypothesis that larvae from parents living in the intertidal zone will have a greater existing tolerance to elevated CO₂, and capacity for beneficial TGP to elevated CO₂, than larvae from parents living in the subtidal zone.

Adult Saccostrea glomerata (Gould 1850) were collected from two habitats (subtidal or intertidal habitats) in two locations at Port Stephens, New South Wales (NSW), Australia (location 1: 32°32′42.25″S, 152°03′42.69″E; location 2: 32°41′54.04″S, 152°03′27.71″E). Two locations were used to better ensure a high level of genetic diversity, with oysters pooled from each location according to their habitat (subtidal or intertidal habitats). The height on the shore for collection of oysters from both habitats, subtidal and intertidal, was identified by measuring the height on the rocky shore above the Indian spring low water (ISLW). Subtidal oysters were collected from 0.1–0.4 m above ISLW. Intertidal oysters were collected from 1.1–1.5 m above ISLW. Individuals from each habitat were gently chipped from the rock surface and immediately transferred to the Port Stephens Fisheries Institute (PSFI, NSW, Australia). Once at PSFI, oysters were scrubbed to remove mud and fouling organisms.

To test the hypothesis that larvae from parents living in the intertidal zone will be more tolerant of ocean acidification than larvae from parents living in the subtidal zone, eggs and spermatozoa were obtained from gravid adults from each habitat. Using a scalpel blade, gametes were stripped from the gonads of 10 females and 10 males from each habitat into separate 500 ml containers of 1 µm filtered seawater (FSW). Eggs and spermatozoa were filtered through 60- and 45-mesh screens, respectively, to remove debris. The size of eggs was determined by measuring the diameter of 30 eggs from four females from each habitat on a Sedgewick Rafter slide under a light microscope (Leica 10×). For each habitat, eggs and spermatozoa were then pooled in approximately equal numbers in separate 1 litre containers.

To determine the impact of elevated CO₂ on larvae from parents living in subtidal and intertidal habitats, there were two P_CO2 treatments used in the study: ambient P_CO2 of 400 µatm (which reflects current conditions) and an elevated P_CO2 concentration of 1000 µatm, expected for the end of this century (Collins et al., 2013). Gametes from the subtidal and intertidal habitats were fertilised in two buckets each, at either ambient CO₂ or elevated CO₂ (FSW, 22°C, salinity 34.6). Eggs were divided equally across the buckets and were left to incubate for 10 min. Following this time, spermatozoa were added to the eggs so that three to four spermatozoa could be seen around the perimeter of each egg when viewed under the light microscope (Leica 10×). Eggs and spermatozoa were left for 30 min to allow fertilisation to occur. This resulted in four larval lines: subtidal oysters, ambient CO₂; subtidal oysters, elevated CO₂; intertidal oysters, ambient CO₂; intertidal oysters, elevated CO₂.

Following fertilisation, eggs from each of the four lines were divided across three 20-litre buckets (12 buckets altogether), to give a final concentration of five fertilised eggs per millilitre in each bucket (FSW, 22°C, salinity 34.6). The buckets were set at the same P_CO2 concentration that the eggs were fertilised at (i.e. ambient fertilisation – ambient larval development; elevated fertilisation – elevated larval development). The elevated P_CO2 level was obtained by directly bubbling CO₂ into the buckets until the desired pH level was reached. The desired pH_NBS level that corresponded to the elevated P_CO2 treatment was determined using the CO₂SYS calculation program developed by Lewis and Wallace (1998). Briefly, the desired elevated P_CO2 level, along with the measured total alkalinity (TA), temperature and salinity of the experimental seawater were entered into the program and the seawater physico-chemical values were calculated using the dissociation constants of Mehrbach et al. (1973) (Table 1). The pH_NBS level in the tanks was measured before and after each water change using a Wissenschaftlich-Technische Werkstätten (WTW) combined Multi meter (3420) and combined electrode (SenTix940). Following the addition of eggs, buckets were immediately sealed with lids to limit gas exchange. After 24 h, a complete water change was done on each bucket. Larvae were collected on a 30 µm screen, gently rinsed with FSW and transferred into new buckets of pre-equilibrated FSW. At this time, larvae were fed an algal diet consisting of 50% Chaetoceros calcitrans, 25% Diacronema lutheri and 25% Tisochrysis lutea at a concentration of 1×10⁴ cells ml⁻¹ (O'Connor et al., 2008). After 48 h, the experiment was stopped and the shell length (antero-posterior measurement) of larvae, percentage development to the D-veliger stage and percentage of abnormal larvae were determined using an ocular micrometer to measure 30 larvae from each replicate on a Sedgewick Rafter slide under a light microscope (Leica 10×). Seawater used in all experiments was collected from Little Beach, Port Stephens, NSW, Australia (32°42′42.75″S, 152°9′26.48″E) and was filtered through 1 µm nominal filters prior to delivery into the hatchery.

To test the second part of our hypothesis, that larvae from parents living in the intertidal zone will have a greater capacity for beneficial TGP to elevated CO₂ than larvae from parents living in the subtidal zone, 200 S. glomerata adults were collected from each of the subtidal or intertidal habitats at the same two locations used for Experiment 1, during the winter months, when they were in a period of gonadal resting.

The oysters from each habitat were returned to the PSFI and divided across 24, 30-litre tanks to acclimate. The tanks were supplied with continuously flowing, recirculating FSW from 750 litre header tanks (one header tank supplied two 30-litre tanks; flow rate 3 l min⁻¹; 22°C; salinity 34.6). Oysters were acclimated at the tidal regime which reflected the diurnal tidal fluctuations they experienced in the field. For the intertidal oysters this involved 9 h of emersion (out of water) followed by 3 h of immersion (submerged in water) twice per day (hereafter called ‘intertidal treatment’). For the subtidal oysters this involved constant immersion (hereafter called ‘subtidal treatment’). During the acclimation period, oysters were fed a maintenance algal diet of 50% Chaetoceros muelleri, 25% D. lutheri and 25% T. lutea twice daily at a concentration of 1×10⁹ cells oyster⁻¹ day⁻¹. Both the water and ambient air temperature in the laboratory were controlled at 22°C. This was done to ensure that the oysters in the subtidal and intertidal treatment experienced the same temperature throughout the experiment.

A fully orthogonal design was used with parents collected from the subtidal and intertidal habitats conditioned under all possible combinations of tidal treatment (intertidal or subtidal) and P_CO2 (400 or 1000 µatm). This was done to determine whether any observed differences in TGP responses of larvae were due to the tidal cycle experienced by the parents during reproductive conditioning (conditioning habitat) or due to differences (genetic and/or plasticity) between the intertidal and subtidal populations (habitat). Following a 2 week acclimation period, six 750 l header tanks were set at the intertidal treatment, and six were set at the subtidal treatment. Within each tidal treatment, three tanks were set at ambient P_CO2 (400 µatm) and three tanks were set at elevated P_CO2 (1000 µatm). Each of the header tanks supplied recirculating FSW to two 30 l tanks, one which housed 16 oysters from the intertidal habitat and one which housed 16 oysters from the subtidal habitat.

The elevated P_CO2 treatment was controlled by pH negative-feedback systems (Aqua Medic, Aqacenta Pty Ltd, Kingsgrove, NSW, Australia; accuracy ±0.01) according to Parker et al. (2012; 2018). These systems control P_CO2 in each tank by bubbling CO₂ until a set pH is reached. The pH_NBS level was measured twice daily in each header tank using a WTW combined Multi meter (3420) and combined electrode (SenTix940) and total alkalinity was measured at each water change (Table 1). Adults were fed a conditioning algal diet of 50% C. muelleri, 25% D. lutheri and 25% T. lutea twice daily at a concentration of 2×10⁹ cells oyster⁻¹ day⁻¹. Water was changed in each tank every 2 days using pre-equilibrated FSW and adults remained in the combined tidal and P_CO2 treatments for 5 weeks until they reached gravid stage. At this time, the adult oysters were removed from the tanks in preparation for spawning.

To determine the impact of transgenerational exposure of the subtidal and intertidal parents to elevated CO₂ and a subtidal and intertidal treatment we measured the size of eggs, shell length (antero-posterior measurement), percentage of D-veligers, percentage abnormality and metabolic rate of 48-h-old larvae following exposure to ambient and elevated CO₂.

Eggs and spermatozoa were obtained from adults from each habitat (subtidal or intertidal), tidal treatment (subtidal or intertidal) and parental CO₂ treatment (ambient or elevated) combination by strip spawning and were fertilised at both ambient and elevated CO₂ using the same methods as in Experiment 1. This resulted in the production of 16 larval lines (Fig. 1).

Following fertilisation, fertilised eggs from each of the 16 lines were divided across three 20-litre buckets (48 buckets altogether), to give a final concentration of five fertilised eggs per millilitre in each bucket (FSW, 22°C, salinity 34.6). The buckets were set at the same P_CO2 level that the eggs were fertilised at. After 24 h, a complete water change was done on each bucket. At this time, larvae were fed an algal diet consisting of 50% C. calcitrans, 25% D. lutheri and 25% T. lutea (O'Connor et al., 2008). After 48 h, the experiment was stopped and the morphological (shell length, percentage of D-veligers, percentage abnormality) and physiological (metabolic rate) larval parameters were measured using the methods described in Experiment 1.

where MR is the oxygen consumption normalized to per larva values (pmol O₂ larva⁻¹ h⁻¹), V is the volume of the vial (l), ΔCw_O2 is the change in water oxygen concentration measured (pmol O₂ l⁻¹), Δt is measuring time (h) and l is the number of larvae in the vial. For the purposes of the calculation, the measured change in percentage P_O2 in air saturation was converted to pmol O₂ l⁻¹ using the Loligo­ online oxygen converter.

The diameter of eggs from Experiment 1 was analysed using ANOVA where ‘habitat’ (subtidal or intertidal) was the single fixed factor, and ‘replicate individual’ (N=4) was used as a random factor. The percentage development to the D-veliger stage and the percentage of abnormal larvae were each analysed using a separate ANOVA with ‘habitat’ (subtidal or intertidal) as the first fixed factor and ‘larval CO₂ treatment’ (ambient or elevated) as the second fixed factor. Data for the shell length of D-veliger larvae did not meet the assumptions of ANOVA due to a non-normal distribution as determined by the Shapiro–Wilk normality test and non-homogenous variances as determined by Cochran's test. Data on the shell length of D-veliger larvae were therefore analysed using a non-parametric aligned-rank transform (ART) ANOVA (Wobbrock et al., 2011) using the ARTool package in R software (https://github.com/mjskay/ARTool). ART ANOVA used ‘habitat’ (subtidal or intertidal) as the first fixed factor and ‘larval CO₂ treatment’ (ambient or elevated) as the second fixed factor and included ‘replicate tank’ (N=3) as a random factor.

The diameter of eggs in Experiment 2 also did not meet the assumptions of ANOVA and were analysed using ART ANOVA with ‘habitat’ (subtidal or intertidal), ‘tidal treatment’ (subtidal or intertidal) and ‘parental CO₂ treatment’ (ambient or elevated) as fixed factors and ‘replicate tank’ (N=3) as a random factor. The percentage development to the D-veliger stage, the percentage of abnormal larvae and the metabolic rate of larvae in Experiment 2 were each analysed using a separate orthogonal ANOVA with ‘habitat’ (subtidal or intertidal), ‘tidal treatment’ (subtidal or intertidal), ‘parental CO₂ treatment’ (ambient or elevated) and ‘larval CO₂ treatment’ (ambient or elevated) as fixed factors; the analysis of D-veliger shell length included ‘replicate tank’ (N=3) as a random factor. Post hoc pairwise comparisons of estimated marginal means were made using the ‘emmeans’ package (Lenth and Lenth, 2018) with Tukey-adjusted P-values to determine significance among levels for factors or interactions of interest (α<0.05). The Shapiro–Wilk normality test was used to check normality and Cochran's test was used to confirm homogeneity of variances, and residual plots were used to assess error distribution for all analyses. For analyses that included random factors, the random effect explained <0.02% of model variance. All analyses were done using the ‘stats’, ‘ARTool’ and ‘emmeans’ (Lenth and Lenth, 2018) packages in R 3.5.3 statistical software; critical P-values were set at α=0.05.

Eggs were significantly larger from parents in the intertidal habitat compared with the subtidal habitat (Fig. 2; F_1,158=33.7, P<0.0001).

Size and percentage development of D-veliger larvae of S. glomerata decreased and the percentage of abnormal larvae increased when exposed to elevated CO₂ for 48 h (Fig. 3A–C). These impacts were less severe in larvae from parents living in the intertidal habitat as shown by the habitat×CO₂ (larvae) interaction (Table 2). There was a 9% reduction in size of larvae exposed to elevated CO₂ from subtidal parents compared with only a 4% reduction from intertidal parents. Larvae from intertidal parents reared at elevated CO₂ were similar in size to larvae from subtidal parents reared at ambient CO₂ (Fig. 3A).

Exposure to elevated CO₂ caused a 22% reduction in the percentage of D-veligers that developed and a 31% increase in the percentage of abnormal larvae from subtidal parents. In contrast, elevated CO₂ had no effect on these developmental traits in larvae from parents living in the intertidal zone (Fig. 3B,C).

Parents that were conditioned in the intertidal treatment produced larger eggs than those in the subtidal treatment at both ambient and elevated CO₂, irrespective of their collection habitat (Fig. 4, Table 3). There was no significant difference in the egg size of oysters collected from the intertidal and subtidal habitat or conditioned at ambient or elevated CO₂.

Parents exposed to elevated CO₂ in the subtidal treatment produced larvae that were larger in size at elevated CO₂ than larvae from parents exposed to ambient CO₂ (Fig. 5A). In contrast, parents exposed to elevated CO₂ in the intertidal treatment produced larvae that were similar in size at elevated CO₂ than larvae from parents exposed to ambient CO₂. Importantly, however, larvae from parents conditioned in the intertidal treatment under ambient CO₂ were larger in size at elevated CO₂ than larvae from parents conditioned in the subtidal treatment under ambient CO₂. Larvae from the intertidal parents were larger in size than larvae from the subtidal parents when parents were conditioned in the subtidal treatments. When parents were conditioned in the intertidal treatments, however, this superior growth of the larvae from the intertidal parents was no longer present, with larvae from the subtidal and intertidal parents being similar in size. Finally, larvae were generally larger in size at ambient compared with elevated CO₂ (Fig. 5A).

The percentage of D-veligers that developed after 48 h was greater than 97% in all ambient CO₂ larval treatments (Fig. 5B). The percentage of larvae that developed to the D-veliger stage was lower at elevated compared with ambient CO₂ when parents were conditioned in the subtidal treatment under ambient CO₂ (82±2% D-veligers). When the parents in the subtidal treatment were conditioned under elevated CO₂, however, the impact of elevated CO₂ on the percentage development to the D-veliger stage was significantly less (91±2% D-veligers). There was no significant effect of elevated CO₂ on the percentage development of larvae from parents conditioned in the intertidal treatment at ambient or elevated CO₂ (Fig. 5B, Table 3). There was also no difference in the percentage development to the D-veliger stage between larvae from intertidal and subtidal parents.

The percentage of abnormal larvae that developed was not affected by elevated CO₂ in larvae from parents conditioned in the intertidal treatment, with low levels of abnormal larvae (0–3%) at both ambient and elevated CO₂ (Fig. 6A, Table 3). For larvae from parents conditioned in the subtidal treatment under ambient CO₂, however, exposure to elevated CO₂ caused up to a 29% increase in abnormal larvae. Parental exposure to elevated CO₂ almost completely ameliorated this impact with larvae from parents conditioned in the subtidal treatment displaying only a 3% increase in abnormality at elevated CO₂ when their parents were also exposed to elevated CO₂ (Fig. 6A, Table 3). Parent habitat had no effect on the percentage of abnormal larvae that developed.

Overall, larvae reared at elevated CO₂ had a greater metabolic rate than larvae reared at ambient CO₂ across all parent habitat and tidal treatments (Fig. 6B, Table 3). When intertidal parents were conditioned under ambient CO₂, larvae had a greater metabolic rate than those from subtidal parents (Table 3). When intertidal parents were conditioned under elevated CO₂, however, this pattern was reversed, and larvae had a lower metabolic rate than those from the subtidal parents (Fig. 6B, Table 3). Exposure of subtidal parents to elevated CO₂ had no effect on the metabolic rate of the larvae. In contrast, exposure of intertidal parents to elevated CO₂ caused a significant reduction in the metabolic rate (Table 3). Finally, there was no effect of tidal treatment on the metabolic rate of larvae from intertidal or subtidal parents.

In this study we tested whether the capacity of the Sydney rock oyster, S. glomerata, to acclimate to ocean acidification depends on the vertical distribution, from the subtidal to intertidal habitat. First, we measured the response of eggs and larvae from parents collected from the subtidal and intertidal habitat to determine whether exposure to metabolic hypercapnia in the intertidal zone led to differences in existing tolerance to elevated CO₂. Second, using a transgenerational experiment, we assessed whether parental conditioning under elevated CO₂ in an intertidal treatment produced an enhanced beneficial TGP response in larval offspring to elevated CO₂ compared with larvae from parents conditioned under elevated CO₂ in a subtidal treatment.

The existing tolerance of S. glomerata larvae to elevated CO₂ depends on the habitat, with the impacts on larvae exposed to elevated CO₂ being less severe when from parents in the intertidal habitat. Larvae from intertidal parents were larger in size when exposed to elevated CO₂ than larvae from subtidal parents. Furthermore, larval development and abnormality from intertidal parents were not affected when exposed to elevated CO₂ (compared with a reduction in the percentage of D-veligers and increase in percentage abnormality in larvae from subtidal parents). These results are consistent with those of previous studies showing that marine organisms which experience episodes of reduced pH and elevated CO₂ in their habitats (e.g. upwelling zones, CO₂ vents, tide pools, intertidal zone) have a greater existing tolerance to ocean acidification (Evans et al., 2013; Kelly et al., 2013; Cole et al., 2016; Gaitán-Espitia et al., 2017; Kapsenberg et al., 2017; Thomsen et al., 2017).

Following settlement, sessile molluscs such as oysters in the intertidal zone experience diurnal episodes of valve closure during periods of emersion (Burnett, 1988; Truchot, 1990; Scanes et al., 2017). This valve closure results in the accumulation of metabolic CO₂, acidifying the extracellular pH (pH_e) of the organism to levels lower than that induced by near-future ocean acidification (Scanes et al., 2017). For example, Scanes et al. (2017) showed that during emersion, the partial pressure of CO₂ level in the extracellular fluid (Pe_CO2) of adult S. glomerata can reach as high as 0.75 kPa (compared with 0.18 kPa during immersion; Scanes et al., 2017). This exceeds the Pe_CO2 level recorded in S. glomerata during exposure to ocean acidification (under a subtidal treatment; Pe_CO2=0.59 kPa; Scanes et al., 2017). The intertidal parents used in this study have been exposed to this CO₂ environment since settlement and as such, may have undergone genetic selection, via post-settlement mortality, to elevated CO₂. Alternatively, the greater tolerance of larvae from intertidal parents could be a phenotypic plasticity response, with parental exposure to metabolic CO₂ in the intertidal zone inducing beneficial TGP effects from parents to their offspring (Kelly et al., 2013). Strong evidence for phenotypic plasticity has been found in numerous studies that have compared the physiology of bivalve species living in the intertidal and subtidal zone (Widdows and Shick, 1985; Lesser, 2016; Scanes et al., 2017). For example, when the mussel Mytilus edulis was transplanted from a subtidal to an intertidal treatment, they were shown to adjust their standard metabolic rate (SMR) to suit their new environment within 14 days (Widdows and Shick, 1985). Furthermore, M. edulis from different tidal heights were observed to have different metabolic enzymes and glycogen stores. When the mussels were transplanted to a common tidal height for 21 days, however, these metabolic indices converged (Lesser, 2016). Specific to our study, assessment of subtidal and intertidal adults of S. glomerata collected from the same populations as those used here, showed that there was no difference in their physiological tolerance to elevated CO₂ when held in the laboratory under a common tidal treatment (Scanes et al., 2017). Collectively, these studies suggest that bivalves are highly plastic, and the naturally greater tolerance of the S. glomerata larvae from the intertidal habitat shown in this study probably reflects beneficial TGP and not genetic selection.

While larvae from intertidal parents appear to have a greater existing tolerance to elevated CO₂, the results of our transgenerational experiment show that there was no further improvement in this tolerance following transgenerational exposure to elevated CO₂. As seen in previous transgenerational studies on S. glomerata (Parker et al., 2012, 2015, 2017a), adults conditioned in a subtidal treatment and exposed in the laboratory to elevated CO₂, passed beneficial carryover effects to their larval offspring. In contrast, there was no improvement in the larvae from parents conditioned in the intertidal treatment and exposed to elevated CO₂. Like the results of Experiment 1, we found that conditioning parents in the laboratory in the intertidal treatment under ambient CO₂ for 5 weeks improved the size, development rate and abnormality of their larval offspring at elevated CO₂. This occurred in larvae from parents collected from both the subtidal and intertidal habitats, further highlighting the highly plastic nature of oysters. Importantly, however, when these parents were exposed to the intertidal treatment and elevated CO₂, there was no further improvement in the tolerance of their larval offspring. Consequently, our results demonstrate that the nature of TGP responses (beneficial, detrimental or neutral) on larval tolerance traits to elevated CO₂ depends on the habitat experienced by a population during reproductive conditioning.

The nature of transgenerational responses in metazoan species to elevated CO₂ have previously been shown to differ, even between closely related species (Donelson et al., 2018; Byrne et al., 2020). For example, parental exposure to elevated CO₂ was found to increase the size of larval offspring in the mussel Mytiluschilensis (Diaz et al., 2018) but reduce the size of larval offspring in the mussel M. edulis (Thomsen et al., 2017). The results of our study are the first to show, however, that the nature of TGP responses within a species depend on the habitat experienced during reproductive conditioning. This highlights the importance of considering the habitat of a species when assessing transgenerational responses.

The neutral TGP response observed in this study in larvae of S. glomerata from parents conditioned under elevated CO₂ in the intertidal treatment may reflect a limitation on the phenotypic traits which respond transgenerationally (DeWitt, 1998; Munday, 2014). For example, parental exposure of anemonefishes to elevated CO₂ has been shown to lead to a complete restoration of life-history and metabolic traits in juvenile offspring, but the impacts of elevated CO₂ on some behavioural effects were not fully restored (Allan et al., 2014). In this study, parental exposure of S. glomerata to the intertidal treatment (exposed to both ambient and elevated CO₂) completely ameliorated the negative impacts of elevated CO₂ on the development rate and percentage abnormality of their larval offspring. The impact of elevated CO₂ on the shell length of these larvae, however, was only partially ameliorated, suggesting that shell length may have limited plasticity to elevated CO₂ (Munday, 2014). In the absence of genetic adaptation, complete restoration of growth effects of elevated CO₂ on larvae of S. glomerata may not be possible via TGP, at least for populations living in the intertidal zone.

Alternatively, TGP to elevated CO₂ for parents exposed to an intertidal treatment may not be limited. Instead, parents living in the intertidal zone, which already experience intermittent episodes of metabolic CO₂ accumulation, may not perceive CO₂ from ocean acidification as an additional threat. As a result, these parents may not provide additional beneficial carryover effects to their larval offspring. Indeed, despite the beneficial TGP effects observed in the morphological trait responses of larvae from parents conditioned under elevated CO₂ in the subtidal treatment, larvae from parents conditioned under elevated CO₂ in the intertidal treatment remained more tolerant to elevated CO₂ across all traits measured. Importantly, in this study we maintained both the water and ambient air temperature at 22°C, to ensure that the oysters in the subtidal and intertidal treatments experienced the same temperature throughout the experiment, allowing us to adequately assess the impacts of elevated CO₂ on the TGP response of S. glomerata in each habitat in isolation of temperature. In the natural environment, however, temperature may also influence the TGP response of intertidal populations to elevated CO₂ as intertidal oysters experience greater temperature stress than those in the subtidal habitat (Somero, 2002). If periods of emersion lead to elevated body temperatures, for example, then because of Q₁₀ effects, the SMR of intertidal oysters could be higher than that of subtidal oysters (Parker et al., 2017b). This could in turn lead to faster rates of gametogenesis in the intertidal oysters and/or a reduction in the energy available to provision into eggs (Parker et al., 2017b, 2018).

The mechanisms that underlie the different TGP responses of S. glomerata in this study are likely to be complex. For many marine organisms, beneficial TGP has been linked to adaptive maternal effects, whereby mothers prime their offspring for suboptimal conditions by investing more energy into each egg (Podolsky and Moran, 2006; Allen et al., 2008; Moran and McAlister, 2009; Zhao et al., 2019). For example, larger egg size was observed in Musculistasenhousia following transgenerational exposure to elevated CO₂ (Zhao et al., 2019). This increased maternal investment is thought to be advantageous during exposure to elevated CO₂ as it results in larger, faster developing larvae that spend less time in the water column (Podolsky and Moran, 2006; Allen et al., 2008; Moran and McAlister, 2009; Zhao et al., 2019). Furthermore, offspring have more energy available to cope with the suboptimal conditions (Gazeau et al., 2013; Kroeker et al., 2013; Parker et al., 2015, 2017a,b). In our study, the egg size was similar between parents exposed to ambient and elevated CO₂, but parents exposed to the intertidal treatment produced larger eggs than parents exposed to the subtidal treatment. This suggests that increased maternal provisioning may be occurring in the intertidal treatments and may be at least partially responsible for the increased CO₂ tolerance of larvae of S. glomerata from these parents. This increase in maternal provisioning by mothers conditioned in the intertidal treatment is perhaps surprising given in this treatment mothers have much shorter periods to feed each day compared with mothers from the subtidal treatment (6 h compared with 24 h per day). Similar results were observed in larvae of the coral, Stylophora pistillata, with parents that were unfed during reproductive conditioning producing larvae with greater total fatty acid content than larvae from parents that were fed (Bellworthy et al., 2019). This increase in fatty acid content of larvae, however, came at the cost of fecundity, with unfed parents producing fewer larval offspring. It is important to note here that eggs were obtained in this study by a method of strip-spawning. While this method was necessary to allow a comparison of the impacts of conditioning habitat and elevated CO₂ on the TGP response of larvae of S. glomerata, it may lead to a bias in the size of eggs if treatment combinations have developed at a different rate (Parker et al., 2018).

The improved tolerance of larvae from parents conditioned in the intertidal treatment suggests a rapid phenotypic plasticity response, driven by the tidal habitat experienced by the parents during reproductive conditioning and not genetic differences which may exist between the intertidal and subtidal populations due to post-settlement mortality and/or phenotypic differences due to their history of exposure to different habitats. This is because the improved tolerance occurred in these larvae irrespective of whether their parents originated from the intertidal or subtidal zone. Interestingly, and in contrast, however, this was not the case for metabolic rate (MR). Following transgenerational exposure, the MR of larvae reared at elevated CO₂ was greater than larvae reared at ambient CO₂. Larvae from intertidal parents exposed to elevated CO₂ (conditioned under both the intertidal and subtidal treatment), however, had a 2.5-fold reduction in MR compared with larvae from parents exposed to ambient CO₂; a result that was not observed in larvae from subtidal parents. Unlike the results observed for size, development rate and abnormality, this contrasting MR response of larvae from subtidal and intertidal parents may reflect genetic and/or phenotypic differences that exist between the populations due to their history of exposure to different habitats. The reduction in MR in larvae from intertidal parents that were exposed to elevated CO₂ is the first example of metabolic acclimation in this species to elevated CO_2. This indicates that these larvae are more energy efficient at maintaining homeostasis at elevated CO₂ (Pörtner, 2010; Sokolova, 2013). Consequently, these larvae may have a higher aerobic scope and capacity to tolerate a wider range of stressors in their environment (Pörtner et al., 2004; Sokolova, 2013). Metabolic acclimation following transgenerational exposure to elevated CO₂ has previously been shown in the mussel M. senhousia (Zhao et al., 2019), the copepod Pseudocalanus acuspes (Thor and Dupont, 2015), and the anemonefish A. melanopus (Miller et al., 2012). The reduction in energy expenditure in these species was suggested to reflect a non-genetic improvement in energy acquisition and assimilation (Zhao et al., 2019) and/or a more efficient ion and acid–base exchange capacity or mitochondrial function during exposure to elevated CO₂ (Miller et al., 2012). For intertidal bivalve species such as S. glomerata, metabolic acclimation may occur, at least in part, due to enhanced activity of carbonic anhydrase (CA) (Zhao et al., 2019). Bivalve species that are adapted to living in the intertidal zone have been shown to compensate for the metabolic CO₂ that accumulates during emersion via CA-catalysation of CO₂ to HCO₃⁻ (Booth et al., 1984; Burnett, 1988). Furthermore, transgenerational exposure to elevated CO₂ has been shown to enhance the activity of CA for some bivalve species (Goncalves et al., 2016, 2017; Zhao et al., 2020). This more rapid conversion of metabolic CO₂ to HCO₃⁻ may contribute to the total calcifying dissolved inorganic carbon (DIC) pool, and reduce the cost of calcification (i.e. by reducing the need to actively pump DIC from seawater) (Zhao et al., 2018, 2020).

The number of studies that measure the transgenerational response of marine organisms to climate change stressors is rapidly growing. These studies, however, do not consider the habitat in which marine organisms live. Here, we show that the TGP effects of elevated CO₂ on larvae of the Sydney rock oyster, S. glomerata, depend on the habitat, with episodes of metabolic hypercapnia experienced in the intertidal treatment leading to a greater tolerance of larvae exposed to elevated CO₂ but limiting beneficial TGP. This shows that studies on the nature of TGP effects need to consider the natural habitat of a species to accurately predict their TGP response to climate change over this century. Forecasting the capacity for marine organisms to persist in a changing ocean via TGP is a leading imperative, yet the natural complexity of marine habitats makes accurately predicting this capacity difficult.

We wish to acknowledge the tremendous support of staff at the Port Stephens Fisheries Institute.