Invariant Gametogenic Response of Dominant Infaunal Bivalves From the Arctic Under Ambient and Near-Future Climate Change Conditions

Arctic marine ecosystems are undergoing a series of major rapid adjustments to the regional amplification of climate change, but there is a paucity of knowledge about how changing environmental conditions might affect reproductive cycles of seafloor organisms. Shifts in species reproductive ecology may influence their entire life-cycle, and, ultimately, determine the persistence and distribution of taxa. Here, we investigate whether the combined effects of warming and ocean acidification based on near-future climate change projections affects the reproductive processes in benthic bivalves (Astarte crenata and Bathyarca glacialis) from the Barents Sea. Both species present large oocytes indicative of lecithotrophic or direct larval development after ∼4 months exposure to ambient [<2°C, ∼400 ppm (CO2)] and near-future [3–5°C, ∼550 ppm (CO2)] conditions, but we find no evidence that the combined effects of acidification and warming affect the size frequency distribution of oocytes. Whilst our observations are indicative of resilience of this reproductive stage to global changes, we also highlight that the successful progression of gametogenesis under standard laboratory conditions does not necessarily mean that successful development and recruitment will occur in the natural environment. This is because the metabolic costs of changing environmental conditions are likely to be offset by, as is common practice in laboratory experiments, feeding ad libitum. We discuss our findings in the context of changing food availability in the Arctic and conclude that, if we are to establish the vulnerability of species and ecosystems, there is a need for holistic approaches that incorporate multiple system responses to change.


INTRODUCTION
Ocean acidification and warming are synergistic environmental stressors (Byrne et al., 2013a) that can affect whole animal physiology (Pörtner and Farrell, 2008). However, the extent to which reproduction and life history strategy are vulnerable to environmental change has received comparatively little attention ). Reproduction underpins the success of populations over time and, regardless of parental survivorship and tolerance capability, negative species responses to novel circumstances at early life cycle stages have the potential to serve as a bottleneck to long-term population survival (Dupont et al., 2010a). Species responses to changing environmental conditions have been shown to carry a high energetic cost in marine calcifiers (Spalding et al., 2017), especially at higher latitudes (Watson et al., 2017), and at earlier stages in the life cycle Foo and Byrne, 2017). This is particularly concerning in polar environments where species responses to global climate change and ocean acidification are widely considered to be regionally amplified (Miller et al., 2010). Discerning the direction and generality of effect, however, is frustrated by the effects of transgenerational plasticity (Karelitz et al., 2019;Kong et al., 2019;Byrne and Hernández, 2020;, as well as intra-specific variations in sensitivity (Przeslawski et al., 2015) and response (Carr et al., 2006;Campbell et al., 2016;Boulais et al., 2017). In addition, maternal environmental history has been shown to affect egg size and volume (Braun et al., 2013) which, in turn, can induce phenotypic responses in larvae .
Overall, the combined effects of ocean acidification and temperature on early gamete development are poorly constrained (Boulais et al., 2017) with most available information focused on gamete viability post spawning or larval development in a limited number of taxonomic groups (see review by Ross et al., 2011 and meta-analysis by Kroeker et al., 2010). By focusing on the later stages of a species reproductive cycle, sensitivities of gametogenesis or fertilization mechanisms are missed, stimulating debate about the potential for reproductive cycles to be disrupted (Dupont et al., 2010a;Hendriks and Duarte, 2010). Indeed, empirical evidence for the echinoderms indicates that ocean acidification can result in delayed but normal gametogenesis (Kurihara et al., 2013), reduced sperm volumes (Uthicke et al., 2013), lower gonad indices (Stumpp et al., 2012), or smaller eggs (Suckling et al., 2015). However, experimental manipulation of acidification in other taxa, including corals (Jokiel et al., 2008;Gizzi et al., 2017), annelids (Gibbin et al., 2017), molluscs (Parker et al., 2017), crustaceans (Thor and Dupont, 2015), and several miscellaneous species (for comprehensive list see Foo and Byrne, 2017), reveal no effects on egg size, gametogenesis, or development. Hence, considerable uncertainty exists in understanding the impact of climatic forcing on individual species within the context of the wider ecosystem, but it is clear that the earliest stages of gamete development need to be considered whilst adequately addressing variations within populations and regional environmental change (Dupont and Pörtner, 2013).
While many regionally abundant benthic invertebrate species have shown physiological tolerance to environmental forcing, often explained by, or attributed to, their boreal evolutionary histories (Richard et al., 2012), the viability of Arctic populations through their ability to reproduce is not currently known. Here, we used two abundant and functionally important benthic bivalves, Bathyarca glacialis and Astarte crenata from the Barents Sea (Cochrane et al., 2009;Solan et al., 2020), a region undergoing rapid change including ice retreat (Polyakov et al., 2012a), increasing sea surface temperatures (Polyakov et al., 2012b), and ocean acidification (Qi et al., 2017), to examine the combined effects of warming and ocean acidification on gamete development. Both species are reported to have large oocytes all year round indicative of lecithotrophic or direct development, and without seasonal or cyclic patterns of oocyte development (Saleuddin, 1965;Von Oertzen, 1972;Oliver et al., 1980). Our a priori expectation was that the physiological cost of near future conditions would indirectly affect reproduction, expressed via a trade-off with egg size or increased oocyte reabsorption, with consequences for the long-term viability of the population.  Table 2) were maintained in aerated seawater (salinity 35, 1.5 ± 0.5 • C), and returned to the Biodiversity and Ecosystem Futures Facility, University of Southampton. Surficial sediment (less than 10 cm depth: year 2017, mean particle size = 28.06 µm, organic material, 6.74%; 2018, mean particle size = 26.51 µm, organic material, 6.21%; Solan et al., 2020, Supplementary  E, 334 m depth), sieved to remove macrofauna (500 µm mesh), homogenized by stirring, and transported back to the University of Southampton at ambient temperature (1.5 ± 1 • C).
The bivalves were fed consistently throughout the acclimation and experiment period ad libitum three times per week with 100 ml cultured live phytoplankton (mixed Isochrysis sp., Tetraselmis sp., and Phaeodactylum sp.) at peak culture densities of 15.6 × 10 6 cells ml −1 , 8.6 × 10 5 cells ml −1 , and 14.2 × 10 6 cells ml −1 , respectively. This equates to ∼0.197 g algae per day and represents 9.93 and 6.42% of algal weight/bivalve wet mass in Astarte and Bathyarca, respectively. Uneaten food was observed settled on the sediment surface which was used as an indicator of feeding ad libitum, and overlying sea water was replaced weekly (partial exchange, ∼80%) to prevent the accumulation of excess nutrients. Experiments were run for 120 days (B. glacialis, 21/11/2017-20/03/2018) or 135 days (A. crenata, 8/10/2018 -19/02/2019) after which animals were removed and fixed in 4% neutral buffered formaldehyde for approximately 1 month before being prepared for histological examination. No premature mortality was recorded.

Histology
Bivalves were selected for histology within a defined size range (A. crenata 25-30 mm shell length; B. glacialis 20-25 mm shell length). For each individual [A. crenata, n = 37 (19 ambient, 18 future); B. glacialis, n = 24 (12 ambient, 12 future)], maximum shell length, height, and tumidity were measured using a digital caliper (±0.01 mm) (Supplementary Table 2), before soft tissue was removed from the shell, wet weighed (±0.001 g), and prepared for histology. Dissection revealed that both species have gonads which infiltrate and partially envelop the digestive diverticula ( Figure 1A) so, as it was not tractable to perform a dissection of the germinal tissue, we adopted whole animal histology for reproductive analysis.
Soft tissue of each specimen was processed for histology according to the protocols described by Lau et al. (2018). In brief, tissue was dehydrated in isopropanol (70-100%), cleared in XTF (CellPath Ltd., United Kingdom) and embedded in 25 mm × 50 mm paraffin wax blocks. Embedded tissue was cut at 6-7 µm, mounted onto slides and stained using hematoxylin Z (CellPath Ltd., United Kingdom), counter stained with eosin Y (CellPath, United Kingdom), and immediately cover-slipped using a DPX mounting medium (Sigma-Aldrich, United Kingdom). Oocytes were captured using a Nikon D5000 digital SLR camera mounted onto an Olympus (BH-2) stereomicroscope and analyzed using ImageJ v 1.48 (Schneider et al., 2012).
Unique oocytes were measured only when a nucleus was visible to ensure the near maximum cross sectional diameter. The size of each oocyte was standardized to the diameter of a circle with an equal aggregate sectional area to the two dimensional section of the imaged oocyte [Equivalent Circular Diameter (ECD), Lau et al., 2018], comparable to the Oocyte Feret Diameter used in previous studies (Higgs et al., 2009;Reed et al., 2014). For each female with more than 100 sectioned oocytes, we calculated the ECD of 100 oocytes in each female (five females and 500 oocytes per treatment in each species). A Chisquared test of independence was conducted between individual females in each experiment treatment to determine a statistically significant association in oocyte size frequencies (Supplementary  Figures 4, 5). Oocyte length frequency distributions for each treatment were pooled to represent the natural variation within individuals, and were analyzed with a Kolmogorov-Smirnov (K-S) test between treatments. All analyses were conducted in R (R Core Team, 2018 v.1.2.5019) and the fishmethods library was used for analysis of the length frequency distribution and K-S test (Nelson, 2019).

RESULTS
Examination of the reproductive organs of A. crenata confirmed 16 females and 21 males with no evidence of hermaphroditism, and one specimen with no discernible gonad tissue. Oocytes were developing in reproductive organs infiltrating the digestive diverticula ( Figure 1A) and consisted of interconnected gonadal alveoli (Figures 1B,C). In four specimens in the future climate treatment, oocytes were loosely held within the supra-branchial chamber and reproductive organs simultaneously ( Figure 1D). Evidence of primary oogenesis was not observed, however, oocytes measured between 46.96 and 185.08 µm (mean ± SD 122.61 ± 22.84 µm, n = 500) in the ambient conditions (Figure 2A), and 44.61-181.93 µm (mean ± SD 122.48 ± 24.08, n = 500) in the future conditions ( Figure 2B), with no notable evidence of atresia. The oocyte size distributions were not treatment specific (2-tailed K-S test, D (116) = 0.062, p = 0.99), and showed a distributional peak between 100 and 150 µm.
Histological examination of reproductive organs of B. glacialis revealed 10 females and 14 males with no evidence of hermaphroditism. Gonads were positioned partially infiltrating the digestive diverticula ( Figure 3A) and were observed in densely packed anterior-posterior tubular pouches up to six oocytes across (Figures 3A,B). Oocytes measured between 39.60 and 144.77 µm (mean ± SD 96.77 ± 14.36 µm, n = 500) in the ambient conditions, and 35.07-144.90 µm (mean ± SD 95.03 ± 18.57 µm, n = 500) in the future conditions, with no notable evidence of atresia. The oocyte size distributions were not treatment specific (2-tailed K-S test, D (94) = 0.122, p = 0.81), but showed a peak at 85 µm following exposure to ambient conditions ( Figure 4A) and 95 µm following exposure to future conditions ( Figure 4B).

DISCUSSION
We have demonstrated, for two abundant species of Arcticboreal bivalve, evidence of gametogenic resilience to projected near-future atmospheric carbon dioxide (550 ppm CO 2 ) and sea temperature (+3 • C), after a 20 week incubation. Our  observations show no difference in oocyte size frequency or physical structure, and in this respect are consistent with other studies (Kurihara et al., 2013;Verkaik et al., 2017). However, the interpretation of reproductive resilience based on a gametogenic response risks the generalization of a fundamental physiological output impacting on population dynamics, and does not take into account prolonged developmental cycles in cold water (Peck, 2016;Moran et al., 2019), or the effects on viability, fertilization, and larval development (Dupont et al., 2010a). The maximum oocyte sizes in A. crenata and B. glacialis are slightly lower than those reported previously (∼200 and 170 µm, respectively, Von Oertzen, 1972;Oliver et al., 1980; see Supplementary  Table 4), likely representing different stages of maturity, and are consistent with the current understanding of reproduction in these species. Both species have egg sizes which suggest direct development or short pelagic development (i.e., lecithotrophic) (Ockelmann, 1965), and the small variation in oocyte frequency observed within treatments (Supplementary Figures 4, 5) are akin to continuous spawners with an overlying seasonal intensity in reproduction (Lau et al., 2018), or natural variations in reproductive fitness. However, the presence of eggs in the suprabranchial chamber in four specimens of A. crenata exposed to future conditions is indicative of brooding, a previously unreported reproductive trait that is also supported by the large egg sizes, and formally hypothesized by the presence of adherent eggs and observed internal fertilization within Astartidae (Ockelmann, 1958;Marina et al., 2020).
It is tempting to conclude that our findings indicate resilience of the reproductive stage examined to near-term climatic forcing, but our observations of the successful progression of gametogenesis took place under standard laboratory conditions which, following accepted protocols (e.g., Pansch et al., 2018), include a constant supply of food. This may have inadvertently provided a sufficient supply of energy to overcome the metabolic costs of environmental stress (Cominassi et al., 2020) and mitigated the impact on gametogenesis. Increasing temperature and carbon dioxide concentrations affect species physiology through increased metabolism (Parker et al., 2013;Jager et al., 2016;Leung et al., 2020), and sometimes the suppression of feeding (Stumpp et al., 2012;Kurihara et al., 2013;Appelhans et al., 2014), which directly affects per offspring investment (Moran and McAlister, 2009;Pettersen et al., 2019), and gamete behavior post spawning (Verkaik et al., 2016). Energy stored as gametes can also be reabsorbed and act as a trade-off with fecundity (Stumpp et al., 2012;Verkaik et al., 2017;Rossin et al., 2019). However, considerable physiological resilience to ocean acidification has been demonstrated at various life-cycle stages in bivalves (Dell'Acqua et al., 2019), echinoderms (Verkaik et al., 2017), and corals (Gizzi et al., 2017), and during short incubations, appears to show no significant effects on growth and reproduction in benthic invertebrates (Dell'Acqua et al., 2019), even in food limited scenarios (Goethel et al., 2017). Laboratory experiments have shown that higher food quality and availability has a role in buffering the physiological effects of climate change and ocean acidification (Asnaghi et al., 2013), with positive effects reported in Calanus copepods (Pedersen et al., 2014), bivalves (Thomsen et al., 2013), and barnacles . Further, a recent study has demonstrated that ad libitum feeding mediated fish growth rates in ocean acidification and warming scenarios, and suggest that this standard method may not reliably detect the impacts of environmental change in laboratory experiments (Cominassi et al., 2020). In our study, the supply of sufficient and nutrient rich food, common to laboratory experiments, is likely to have moderated the effects of nearfuture carbon dioxide and temperature controls (Thomsen et al., 2013;Ramajo et al., 2016;Cominassi et al., 2020), and provided the necessary nutrients for successful gamete development. Nevertheless, the physiological fitness of a species and production of gametes does not imply their viability, successful development, or recruitment to the environment (Caroselli et al., 2019).
Environmental change does not only have a direct physical effect on species physiology (Pörtner and Farrell, 2008), but also changes the wider ecosystem, including food-web structures FIGURE 4 | Size frequency of oocytes for Bathyarca glacialis from the Barents Sea following incubation (20 weeks) under (A) ambient and (B) representative future climate conditions. There was no significant difference in distribution between treatments (2-tailed K-S test, D (94) = 0.122, p = 0.81). (Wassmann et al., 2006). In polar regions, the seasonal input of nutrient rich primary production originating from ice algae contributes an important seasonal input of organic matter to the benthos (Wassmann et al., 2011;Degen et al., 2016), impacting on biomass (Kêdra et al., 2013), growth (Blicher et al., 2010;Carroll et al., 2011aCarroll et al., ,b, 2014, benthic community physiology Carroll and Peterson, 2013), and reproduction (Boetius et al., 2013). It has been consistently shown in polar environments that food has a greater impact on invertebrate physiology than temperature (Brockington and Clarke, 2001;Blicher et al., 2010), and can drive multi-decadal scale patterns in growth Carroll et al., 2009) and recruitment (Skazina et al., 2013;Dayton et al., 2016). Associated with environmental forcing in the Arctic, there is expected to be a shift in the timings and quality of organic matter input to the benthos, from nutrient-rich ice algae to pelagic phytoplankton derived primary productivity (Arrigo and van Dijken, 2015), associated with thinner sea ice (Lange et al., 2019), and the transition to ice free conditions (Grebmeier et al., 2006;Leu et al., 2011;Polyakov et al., 2012a). Arctic phytoplankton assemblages may display resilience to ocean acidification through natural tolerances and intraspecific diversity (Hoppe et al., 2018), but the increasing unpredictability in quality of organic matter input impacts on the tight pelagic-benthic coupling which characterizes the Arctic (Tamelander et al., 2006;Wassmann et al., 2011;Kêdra et al., 2015). However, the observation that benthic species such as Astarte spp. and Bathyarca glacialis display feeding plasticity also ensures efficient use of available food input throughout the year (Gaillard et al., 2015;De Cesare et al., 2017), which may result in reproductive viability in otherwise unfavorable conditions.
Although gametogenesis may remain unaffected or mediated by food supply as a consequence of near future environmental change, the viability of fertilization and larval development under projected environmental conditions could still compromise successful recruitment (Dupont et al., 2010b;Kroeker et al., 2010;Albright, 2011). Fertilization across taxa has often shown negative responses to increasing carbon dioxide and temperature (e.g., Kurihara et al., 2007;Ericson et al., 2012;Guo et al., 2015;Graham et al., 2015), but results are not always consistent between species (Clark et al., 2009), populations (Thor et al., 2018), sexes (Verkaik et al., 2016), or individuals (Campbell et al., 2016;Boulais et al., 2017). Meanwhile "carry-over" effects and transgenerational plasticity may affect subsequent life cycle stages Kong et al., 2019), shifting the development "bottleneck" to later stages, or forcing tradeoffs with alternative reproductive traits such as fecundity or egg volume (Chakravarti et al., 2016). Larval development and early life history have also shown inconsistencies in their response to ocean acidification and increasing temperature. The Antarctic urchin Sterechinus neumayeri showed no differences in growth of reproductive tissue (Morley et al., 2016), or larval skeletal development (Clark et al., 2009) after exposure to ocean acidification and increased temperature, but a significant decrease in fertilization, developmental success, and increased developmental aberrations in alternative experiments have been recorded (Ericson et al., 2012;Byrne et al., 2013b). Larval type is also considered important, and planktotrophic larvae which are reliant on pelagic food are considered to be more susceptible than direct developing or non-feeding lecithotrophic larvae (Gutowska and Melzner, 2009;Dupont et al., 2010c;Gray et al., 2019). Consequently, biogeographical variations in larval responses to environmental change are likely to follow global and regional patterns of dominant larval types (Marshall et al., 2012). Here, both A. crenata and B. glacialis have oocyte sizes akin to non-feeding lecithotrophic or direct development (brooding) (Ockelmann, 1965) which is common in Polar species (Marshall et al., 2012), and suggests that food availability will have limited impact on larval development directly.
The lack of consistency between studies demonstrates the complex relationships between ocean acidification and temperature as synergistic stressors on individual reproductive performance (Byrne et al., 2013a;Harvey et al., 2013) and/or natural variations in tolerances within populations (Smith et al., 2019). Recruitment and recovery from disturbance in polar environments is often very slow (years -decades) (Barnes and Kukliński, 2005;Konar, 2013) although our knowledge of Arctic invertebrate reproductive biology remains limited (Kuklinski et al., 2013). The rapid rates of environmental change, however, may be further exacerbated by extreme longevity of organisms at high latitudes (Moss et al., 2016), To understand the effects of environmental change on reproductive ecology it will be important to consider all life history stages (Dupont et al., 2010a), and the role of population variability and plasticity as mechanisms of population resilience to environment change . The role of changing food resources in determining reproductive viability in regions experiencing rapid change is presently under appreciated, but will be necessary to understand the links between the environment and reproductive/larval physiology (Goethel et al., 2017). The complex interactions between physiology, the environment, and climate change (Byrne et al., 2013a) will determine future population distributions and local extinction risks (Murdoch et al., 2020). We put forward, therefore, that holistic approaches including projected changes to regional food sources, are required to understand how future conditions may affect reproduction and modify interactions with whole animal physiological characteristics (Pörtner and Farrell, 2008;Dupont and Pörtner, 2013).

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. All histology image data are to be made openly available from the Discovery Metadata System (https://www.bas.ac.uk/project/ dms/), a data catalogue hosted by The UK Polar Data Centre (UK PDC, https://www.bas.ac.uk/data/uk-pdc/).

AUTHOR CONTRIBUTIONS
AR and LG conceived the idea. AR conducted the laboratory work and image analysis and wrote the manuscript. JG and AR analyzed the data and produced the figures. MS, LG, and JG reviewed and provided critical comments on the manuscript prior to submission. All of the authors have contributed to the manuscript.

FUNDING
This work was supported by "The Changing Arctic Ocean Seafloor (ChAOS) -how changing sea ice conditions impact biological communities, biogeochemical processes, and ecosystems" project (NE/N015894/1 and NE/P006426/1, 2017-2021) funded by the Natural Environment Research Council (NERC) in the United Kingdom.

ACKNOWLEDGMENTS
We thank the crew of cruise JR16006 and JR17007, RRS James Clark Ross, Robbie Robinson (University of Southampton) for assistance with the design of our experimental systems, Tom Williams and Matthew Carpenter-Liquorice for maintaining the experiments, and National Marine Facilities, Southampton and the British Antarctic Survey, Cambridge for logistical support.