The reproductive ecology of the Antarctic bivalve Aequiyoldia eightsii (Protobranchia: Sareptidae) follows neither Antarctic nor taxonomic patterns

The accepted paradigm for reproduction in Antarctic marine species is one where oogenesis takes 18 months to 2 years, and a bimodal egg-size distribution where two cohorts of eggs are present in female gonads throughout the year. These slow gametogenic traits are driven by low temperature and/or the restriction of resource availability because of extreme seasonality in the marine environment. Here we present data on the reproductive ecology of the common Antarctic bivalve Aequiyoldia eightsii (Jay, 1839) (Protobranchia: Sarepidae) from monthly samples collected between January 2013 and May 2014 at Hangar Cove, Rothera Point on the West Antarctic Peninsula. These data show that A. eightsii is unusual because it does not follow the typical pattern expected for reproduction in Antarctic marine invertebrates, and differs also from closely related nuculanid protobranch bivalves with respect to gametogenic duration and reproductive periodicity. Continuous oogenesis, evidenced by the year-round occurrence of previtellogenic, vitellogenic, and ripe oocytes in female gonads, is supplemented by a seasonal increase in reproductive intensity and spawning in Austral winter (April–May), evidenced by the loss of mature spermatozoa and ripe oocytes from males and females, respectively. The simultaneous occurrence of these contrasting traits in individuals is attributed to a flexible feeding strategy (suspension and deposit feeding) in response to seasonal changes in food supply characteristic of the Antarctic marine environment. Asynchrony between individual females is also notable. We hypothesise that the variability may represent a trade-off between somatic and reproductive growth, and previously reported internal interannual cycles in shell growth.

Investigating the reproduction of a particular species therefore provides insight into its adaptive capacity and potential resilience to environmental change (Byrne 2011;Sanford and Kelly 2011). As the global ocean is under increasing threat from climate change, especially warming and acidification (Byrne 2011), evaluation of a species' reproductive success is key to understanding vulnerability to future climate scenarios (Grange et al. 2004). In the Southern Ocean, the waters around the West Antarctic Peninsula have experienced pronounced warming relative to the global average (see review by Mayewski et al. 2009), with sea surface summer temperatures in the Bellingshausen Sea on the west of the Peninsula increasing by ~ 1 °C since the 1950s (Meredith and King 2005). Antarctic shallow-water invertebrates are particularly susceptible to the effects of climate change, as warming causes perturbations in sea ice dynamics, which could subsequently lead to ecological regime shifts in benthic systems (Barnes and Conlan 2007; Barnes and Souster 2011;Barnes 2016). Presently, there is a lack of knowledge specific to the recovery potential of the benthos after these disturbance events (Clark et al. 2015). Additionally, many studies have demonstrated the physiological constraints and vulnerability of Antarctic benthic invertebrates to subtle temperature increases (e.g. Peck et al. 2004a;Pörtner et al. 2007;Peck et al. 2010Peck et al. , 2014. Benthic assemblages in Antarctica are taxonomically diverse and abundant, with more than 7137 species identified, which accounts for ~ 88% of the total number of identified species in the Southern Ocean (De Broyer and Danis 2011). However, the reproductive ecology and related paradigms regarding invertebrate life history and ecology are far from fully understood (Pearse and Lockhart 2004), with details of reproductive biology, especially in terms of variations between the years, or longer term trends, rarely published.
The reproductive patterns of Antarctic shallow-water benthos are often characterised by a prolonged reproductive cycle (18-24 months), lecithotrophy, and direct development, low fecundity, and seasonal reproductive patterns (Clarke 1979;Brey and Hain 1992;Chiantore et al. 2002;Grange et al. 2004Grange et al. , 2007Higgs et al. 2009;Pearse et al. 2009;Reed et al. 2013a). However, a few species also demonstrate shorter gametogenic periods (12 months or less) (Clarke 1988;Chiantore et al. 2002;Tyler et al. 2003;Grange et al. 2011), planktotrophic larvae (Stanwell-Smith et al. 1999;Bowden et al. 2009), and relatively high fecundity (McClintock and Pearse 1986;Bosch and Pearse 1990;Grange et al. 2004;Orejas et al. 2007). Moreover, the reproductive ecology of Antarctic shallow-water and deep-sea species are similar in several benthic taxa, including asteroids (Bosch and Pearse 1990) and octocorals (Orejas et al. 2012). Several authors have suggested that the abiotic and biotic characteristics shared between the polar and deep-sea environments attribute similarities in reproductive ecology between Antarctic shallow-water and the deep-sea benthos (Bosch and Pearse 1990;Ramirez-Llodra 2002;Orejas et al. 2012). Both realms are characterised by low stable temperatures, low light, and ice scour, and functionally similar benthic communities with slow population turnover rates (Bosch and Pearse 1990;Aronson et al. 2007;Clark et al. 2015). Furthermore, the continental shelf in Antarctica is depressed by the weight of the ice-sheet, such that its average depth is around 500 m compared to around 200 m elsewhere in the world (Heywood et al. 2014). This feature facilitates the connection between shallow and deep areas of the Southern Ocean across the continental shelf, with submerging Antarctic bottom water and emerging circumpolar deep water that may have enhanced colonisation of shallow fauna to deep waters (and vice versa) during glacial cycles of shelf ice advance and retreat (Berkman et al. 2004;Strugnell et al. 2011). Many Antarctic invertebrate taxa including Bivalvia, Gastropoda, Amphipoda, and Decapoda are reported to have a significantly higher degree of eurybathy compared to their temperate counterparts, often being distributed from surface water to hundreds of metres (Brey et al. 1996).
Protobranchia is a subclass of small deposit feeding bivalves commonly found in soft sediment environments with approximately 750 species distributed globally (see review by Zardus 2002). With a Cambrian origin (Carter et al. 2000;Sharma et al. 2013), this subclass has diversified and colonised the deep sea, representing ~ 50 and 90% of bivalve species on the continental slope and abyssal plain, respectively (Allen 1978). Although protobranch bivalves are important bioturbators in the deep sea (Zardus 2002), they remain one of the least studied molluscan groups (Sharma et al. 2013;Reed et al. 2014). Protobranch bivalves are well represented in the Antarctic, but information regarding the ecology of Protobranchia in the Southern Ocean is limited as most species are found only in deeper water (Reed et al. 2013b). One exception is the common shallow-water species Aequiyoldia eightsii (Jay, 1939) (previously Yoldia eightsii), which has been used as a model bivalve species in several investigations. While studies have examined its abundance and ecophysiology (described below), the reproductive ecology and gametogenic cycles in this species are poorly understood, constrained by the problems associated with seasonal sampling of organisms in the Southern Ocean.
Here, we investigated the reproductive ecology of an Antarctic shallow-water protobranch bivalve A. eightsii, which is one of the most abundant benthic species in the maritime Antarctic. Previous records show densities of A. eightsii of up to 1540 m −2 at Signy Island (Peck and Bullough 1993) and biomass contributions equivalent to 56.6% of the total macrofaunal biomass at Faro station in Potter Cove (Pasotti et al. 2015a). The geographical distribution of A. eightsii is patchy around the Antarctic continent (Dell 1990), but is extended to sub-Antarctic regions including King George Island (62 o 10′S) and the Magellan Strait (53 o 28′S) (González-Wevar et al. 2012). With a bathymetric distribution ranging from intertidal sand flats to deep waters (824 m), A. eightsii is most commonly found at depths shallower than 100 m (Dell 1990;Peck and Bullough 1993). Aequiyoldia eightsii has an opportunistic feeding strategy, which typically exhibits deposit feeding but can alternate to a suspension feeding mode when phytoplankton is present in the water column (Davenport 1988). Deposit feeding and burrowing in A. eightsii involve intensive locomotion, where this species is one of the fastest burrowing of all Antarctic marine benthic groups (Peck et al. 2004b). Previous observations have also shown a single specimen can expel sediment up to 13 times its dry tissue mass within 24 h (Davenport 1988). Given its high abundance and wide distribution within Antarctica, A. eightsii is likely to be one of the major bioturbators among the Antarctic shallow-water benthic communities. Moreover, A. eightsii has frequently featured in physiological studies as a comparative species with other Antarctic benthos (Peck and Conway 2000;Peck et al. 2004b;Fattorini et al. 2010;Peck et al. 2014;Román-González et al. 2017), and as a model species in studies designed to analyse the impact of climate change on Antarctic seafloor communities (McClintock et al. 2009;Peck et al. 2009Peck et al. , 2010Clark et al. 2016). A recent study identified an innate endogenous growth rhythm in this species and suggested the pattern may be attributed to the reallocation of energy resources between cycles of somatic and reproductive investment (Román-González et al. 2017). Owing to the use of A. eightsii in a range of Antarctic ecological studies, the reproductive data reported here should inform future studies focused on interpreting the biological and ecological role of the species, and the ecological processes of the Protobranchia and Antarctic benthos as a whole.
The objectives of this study were to (1) describe the reproductive ecology and general anatomy of A. eightsii from a shallow-water site at Rothera Point off the West Antarctic Peninsula using wax histology, (2) to compare the reproductive condition of A. eightsii with other Antarctic shallow-water benthos and protobranch bivalves, and (3) to evaluate, where possible, what may be driving these reproductive patterns.

Sample collection
Monthly samples of A. eightsii were scuba diver-collected between January 2013 and May 2014 at 13 to 16 m depths in Hangar Cove, adjacent to the British Antarctic Survey's Rothera Research Station on Rothera Point, Adelaide Island, off the West Antarctic Peninsula (67 o 33′52S, 68 o 07′43 W) (Fig. 1). Fifteen to 20 individuals were collected per month and fixed in 4% buffered formaldehyde in the field. Formaldehyde-preserved samples were then transported to the University of Southampton, UK for storage. Poor ice conditions and inclement weather prevented field sampling in February and June 2013.

Histological preparation
For each specimen, maximum shell length, height, and width were measured using a digital calliper (± 0.01 mm) and the soft tissue separated from the shell using a scalpel. Photographs of specimens before and after shell removal were taken for anatomical study. Aequiyoldia eightsii with shell length > 20 mm have been reported to be reproductively mature . In an effort to maintain consistency, only individuals with a shell length between 20 and 25 mm were selected for processing. None of the bivalves collected in March or August 2013 measured between 20 and 25 mm and were therefore discounted from further study. A total of 95 individuals were selected for histological analysis. Sex of individuals was determined from histological sections.
The gonad of A. eightsii is not a discrete organ and the germinal tissues are infiltrated in the visceral mass. Therefore, whole specimens were examined using standard wax histology techniques following the protocol outlined in Higgs et al. (2009). Preliminary studies demonstrated the denser foot muscle to be more sensitive to the process of clearing when compared to other soft tissues. This tissue was therefore dissected and processed separately. Bivalves were dissected in a longitudinal direction through the body just below the digestive gland, removing the foot muscle from the main body.
Both dissected parts of soft tissue were dehydrated in a graded series of isopropanol, cleared in three washes of histoclear (CellPath, UK) and embedded in wax blocks. Serial sections of 7 µm thick were mounted and stained with Haematoxylin Z (CellPath, UK) and Eosin Y (CellPath, UK), and immediately cover-slipped using a DPX mounting medium (Sigma-Aldrich, UK). Note that the gonads of six female individuals were damaged during histological preparation and were therefore excluded from reproductive analysis.

Gonad morphology
General anatomical observations of A. eightsii were made with respect to gonad morphology from the serial sectioning of both dissected parts of animals under a light microscope [Olympus BHS (BH-2)], and the macrophotography images of soft tissues after shell removal.

Female reproductive analysis
Preliminary analysis indicated that a sexually mature female could simultaneously carry in excess of 3000 oocytes. Therefore, only eight histological sections, evenly distributed across the gonad-digestive complex per specimen (around 1.5 mm apart), were taken for image analysis. Thin tissue sections were photographed using a Nikon D5000 digital camera mounted on a light microscope [Olympus BHS (BH-2)]. Oocyte size, measured as Equivalent Circular Diameter (ECD), of individual females was quantified using the 'Area' tool in ImageJ v1.48 (Rasband 1997(Rasband -2016. ECD assumes the best-fit spherical size of each oocyte and is equivalent to Oocyte Feret Diameter (OFD) used in previous studies (e.g. Reed et al. 2013aReed et al. , 2014. Only oocytes with a visible nucleus were measured to ensure the size of each oocyte was approximated at, or very close to its maximum cross-sectional area. The equation used for calculating ECD from the traced area of each oocyte was where ECD is Equivalent Circular Diameter (µm), and A is the area of an individual oocyte (µm 2 ).
Female maturity was evaluated based on the overall oocyte cellular appearance and description of marine bivalve ovary development outlined in Morse and Zardus (1997). Oocytes < 50 µm ECD, with basophilic cytoplasm and a large nucleus, were defined as previtellogenic. Oocytes between 50 and 80 µm ECD, with eosinophilic cytoplasm and yolk droplets, were identified as vitellogenic, whereas oocytes > 80 µm ECD with an outer vitelline envelope and jelly coat were defined as ripe oocytes ( Fig. 1 in Online Resource 1).

Male reproductive analysis
Eight histological sections evenly distributed across the gonad-digestive complex of each male A. eightsii were photographed for image analysis. Five stages of gonad maturity were identified according to the rationale for the overall development and distribution of spermatozoa in marine bivalve testes development described in Morse and Zardus (1997) (Fig. 2

Statistical analysis
A χ 2 test of independence was conducted between both month using pooled females and individual females within each month, and oocyte size frequency distribution to determine whether there was a statistically significant association between these variables. For instances, where expected cell frequencies were less than five, neighbouring oocyte size bin ranges were collapsed and the procedure re-run until this assumption was met. Statistical significance is reported at p < 0.05. For instances, where a statistically significant association was reported, the strength of association between variables was evaluated using Cramer's V (Cohen 1988), with a value of 0 and 1 indicating "no association" and "complete association", respectively (i.e. small association = 0.1, moderate association = 0.3, and large association = 0.5). In addition, the deviation of individual cells from independence was reported using adjusted standardised residuals according to Agresti (2007), where values > 3 were deemed indicative of cells that significantly deviate from independence. All statistical analyses were performed using IBM SPSS Statistics version 24 for Mac OS X.

Gonad morphology
The overall anatomy of A. eightsii is presented in Fig. 2. The gonad envelops the centre of the alimentary canal. A thin layer of gonadal tissue (up to 3-4 cells thick) covers the outer layer of the digestive gland and envelops the stomach where the gonad tissue thins to approximately 1 cell thick. A strip of gonadal cells is located within the anterior adductor in the upper body (Figs. 2b,d; 3a).
While the diameter of the stomach reduces as it penetrates the centre of the foot muscle, the gonadal tissue layer around the stomach thickens (up to 10-12 cells thick; Fig. 3b). The adjacent section of hindgut in the foot muscle is also surrounded by gonadal tissue, but fewer cells are present (around 5-6 cells thick). No gonadal cells are found near the base of the half gut loop in the foot muscle.

Sex ratio
All specimens (n = 95) selected for histological analysis had distinguishable gonads and were separately sexed. Of the 95 individuals, 48 (50.5%) were female and 47 (49.5%) were male.

Female reproductive analysis
Qualitative assessment of the female gonad revealed the presence of oocytes of all developing stages throughout the study period. Previtellogenic oocytes (and other smaller   Although oocytes of all maturity stages were observed in the gonad throughout the study period, the distribution of oocyte sizes varied between months (Fig. 5). A χ 2 test of independence was conducted between month, using pooled females, and oocyte size frequency distribution. All expected cell frequencies were greater than five. There was a statistically significant association between month and the frequencies of oocytes distributed across the range of oocyte size (χ 2 = 8156.288, df = 36, p < 0.0001). The association was moderately strong (Cohen, 1988), Cramer's V =0.349, and 46 of 52 cells had adjusted standardised residuals > 3 (Table 1) indicating a significant deviation from independence.
Statistically significant associations between individual females, sampled in the same month, and their oocyte size frequency distributions were also recorded, where the strength of associations was generally moderate (p < 0.05; see Table 1 in Online Resource 2).

Male reproductive analysis
Seasonal changes in male maturity stage were interpreted to show a spawning event in April-May, as indicated by spent testicular acini (stage V) in 2013 and newly developed spermatogonia (stage I) in 2014 (Fig. 6). Testes matured from an early developmental (stage I) to mature stage (stage IV) between July 2013 and April 2014, and from a late developmental (stage III) to mature stage (stage IV) between January and April 2013.

Discussion
The present study of the reproductive ecology of the Antarctic shallow-water protobranch bivalve, A. eightsii, demonstrates a reproductive mode that differs from previously described protobranch species and other Antarctic shallowwater invertebrates. Our results indicate that the Hangar Cove population of A. eightsii exhibits continuous oogenesis, with a period of increased reproductive intensity and spawning between April and May, with asynchrony in oocyte development among females. Although we acknowledge reduced sample sizes in some months, this was to ensure all samples were comparable with respect to shell lengths and therefore sexual maturity, based on previous studies (Peck et al. 2010). Of those specimens which fit our selection criteria, there was no control over gender at the point of collection. Within the individual females measured, efforts were made to measure in excess of 1000 oocytes where possible to reduce the weighting of inherent variability. All specimens in this study were dioecious with an overall female-to-male ratio approximating unity and with no evidence of hermaphroditism found.
Continuous investment in oogenesis was evidenced by the simultaneous, year-round occurrence of previtellogenic, vitellogenic, and ripe oocytes in the female gonads. This reproductive trait contrasts with observations of seasonal oogenic development reported in most shallow-water Antarctic marine invertebrates studied to date (e.g. Chiantore et al. 2001;Ahn et al. 2003;Tyler et al. 2003;Grange et al. 2004Grange et al. , 2007Kang et al. 2009;Grange et al. 2011;Reed et al. 2013a, b), and other closely related protobranch bivalves including the cold water Yoldia hyperborea (Jaramillo and Thompson 2008), where cohorts of oocytes distinguishable in their maturity stage dominate the gonad at different times of year. Continuous oogenesis is observed, however, in some deep-sea protobranch bivalves (Tyler et al. 1992;Zardus 2002;Scheltema and Williams 2009;Reed et al. 2014). Significant associations between individual females and their oocyte size frequency distribution were also observed indicating asynchrony in oogenesis between individuals sampled in the same month. This asynchrony could be attributed to an aseasonal reproductive rhythm. However, we did not observe any evidence for aseasonality in the form of multiple spawning events in the females (Fig. 6), nor male maturity stage (Fig. 2 in Online Resource 1). Instead, we observed a seasonal periodicity in the reproduction of both male and female A. eightsii (described below). In the case of the females studied, this seasonal periodicity was observed simultaneously to the continuous investment in oocyte development described above. We therefore propose that the asynchrony observed between individual females is consistent with an innate, long-term multi-year reproductive cycle, related to the partitioning of energetic resources between somatic and reproductive investment (as hypothesised in Román-González et al. 2017 and discussed below).
Observed simultaneously to the continuous oocyte development described above was a seasonal increase in reproductive intensity (spawning) during early Austral winter. This was demonstrated by an investment in the size and presence of ripe oocytes and their subsequent loss from the gonad, and a statistically significant association between month and oocyte size frequency distribution. Despite this observation, previous studies have identified very few bivalve veligers or larvae in winter in shallow waters adjacent to Rothera Research Station (Bowden et al. 2009). However, the pericalymma larva of protobranchs differs from the typical veliger larva of most bivalves in identifying features, being barrel-shaped and cloaked with an outer test enveloping the larva (Zardus and Morse 1998). This atypical appearance and the well-established difficulties associated with field identification of larvae render this larval type susceptible to misidentification, underestimation, and/or being missed entirely in plankton samples. Winter spawning events have been previously reported for other Antarctic shallow benthic invertebrates, including the seastar Odontaster validus (Pearse 1965), the ascidian Cnemidocarpa verrucosa (Sahade et al. 2004), and the octocoral Malacobelemnon daytoni (Servetto and Sahade 2016), and confirmed in some cases by the presence of larval stages in the water column (e.g. for O. validus; Bowden et al. 2009). These events have been attributed to flexible feeding, energy storage, and/or extended gametogenic periods. In the case of A. eightsii, based on observations in this study, there is no evidence of energy storage or an extended gametogenic period; however, flexible feeding in the form of both suspension and deposit feeding has been reported (Davenport 1988), and we speculate that the peculiar reproductive ecology of the bivalve is underpinned by a long-term internal cycle related to energy allocation.
Feeding plasticity may overcome the food limitation that characterises shallow Antarctic waters and poses resource limits for many species during winter, and may also facilitate the continuous oogenesis observed in A. eightsii. Aequiyoldia eightsii exploits phytoplankton in the summer by suspension feeding, and deposit feeds on organic material in the sediment, mostly composed of microphytobenthos and macroalgal fragments all year round (Davenport 1988;Corbisier et al. 2004;Pasotti et al. 2015b). Fragments of phytoplankton/algae were observed in the stomach and hindgut of all specimens in the current study, supporting continuous feeding throughout the year. These findings are consistent with previous laboratory experiments where specimens aged younger than 24 years fed primarily on phytoplankton < 20 µm in diameter (Cripps and Priddle 1995). As the A. eightsii specimens analysed in the current study had shell lengths between 20 and 25 mm, representative of 10-33 years of age (Peck and Bullough 1993), it is likely that the phytobenthos is the primary food source. Nanoplankton concentrations in the waters around Rothera Research Station peak in the late summer (~ March), and are suggested to have an important influence on the ecology of benthic grazers in the area (Clarke et al. 2008). The settlement and subsequent deposit feeding on nanoplankton in sediment therefore could trigger a reproductive response in A. eightsii, such as spawning in the Hangar Cove population, which could explain the increased reproductive intensity and spawning detected in April and May.
While feeding plasticity enables A. eightsii to sustain continuous oogenesis, and the seasonal availability of food, i.e. nanoplankton production, acts as a selective pressure driving a superimposed seasonal reproduction, recent evidence of a multi-year cycle in shell growth of the protobranch bivalve (Román-González et al. 2017) suggests the reproductive asynchrony observed in females is not underpinned by environmental factors. Nor can this reproductive pattern be explained by the accepted paradigm of an 18to 24-month period needed for oogenesis in several other Antarctic invertebrates (Peck et al. 2005; for examples see Grange et al. 2004Grange et al. , 2007Brockington et al. 2007;Orejas et al. 2007). The reproductive ecology of A. eightsii also differs from other shallow-water and deep-sea nuculanid protobranch bivalves that have a 12-month gametogenic cycle, often with seasonal investment in reproduction (Tyler et al. 1992;Nakaoka 1994;Jaramillo and Thompson 2008;Scheltema and Williams 2009). Instead the asynchrony observed here may be linked to an innate, endogenous growth rhythm (e.g. 9.06 year) that is asynchronous between individuals, not keyed with environmental variability, and attributable to cycles in reproductive output (Román-González et al. 2017).
The maximum egg size of A. eightsii measured in this study was 176.7 μm. In marine invertebrates, larval development mode is often inferred from egg size (Moran and McAlister 2009). However, in Protobranchia, egg sizes can vary extensively within individual modes of development and even within closely related species (Gustafson and Reid 1986; Scheltema and Williams 2009). This subclass is known to only exhibit two modes of larval development to date: pelagic lecithotrophy and brooding (Zardus 2002;Scheltema and Williams 2009). Since gametogenesis of A. eightsii shows synchrony in seasonality between males and females, the loss of a relatively large proportion of gametes between April and May and the absence of brooded embryos suggest this bivalve is a broadcast spawner with lecithotrophic larvae. Winter spawning of lecithotrophic larvae may facilitate the avoidance of predators feeding on larvae associated with the summer bloom (Sahade et al. 2004), ice-mediated disturbance (Barnes and Souster 2011), and a greater availability of free substratum for larval settlement (Bowden 2005). As adult A. eightsii can inhibit larval settlement via density-dependent control (Peck and Bullough 1993), larvae spawned during periods with more available habitat could increase the chance of settlement. The reproductive ecology of A. eightsii exhibits continuous oogenesis, with a period of increased reproductive intensity and spawning between April and May, and asynchrony between females. These characteristics contrast with the seasonal reproductive traits reported for many shallowwater Antarctic invertebrates and closely related protobranch bivalves. We hypothesise that the asynchrony documented between females is underpinned by a multi-year periodicity attributed to an innate, endogenous rhythm related to energetic allocation between somatic and reproductive investment. However, the ability to switch between suspension and deposit feeding modes, according to the quality and quantity of food available, facilitates a continuous investment into oogenesis, whilst being responsive to the seasonal pulse in primary production. This study provides key information on a scientifically important species of bivalve that has been commonly used for baseline ecological investigations in Antarctica. Knowledge of its reproduction provides insight into the biological flexibility of the species and contributes to our wider understanding of adaptations in polar marine environments.