Effects of Ocean Acidification on Calcification of the Sub-Antarctic Pteropod Limacina retroversa

Ocean acidification is expected to impact the high latitude oceans first, as CO2 dissolves more easily in colder waters. At the current rate of anthropogenic CO2 emissions, the sub-Antarctic Zone will start to experience undersaturated conditions with respect to aragonite within the next few decades, which will affect marine calcifying organisms. Shelled pteropods, a group of calcifying zooplankton, are considered to be especially sensitive to changes in carbonate chemistry because of their thin aragonite shells. Limacina retroversa is the most abundant pteropod in sub-Antarctic waters, and plays an important role in the carbonate pump. However, not much is known about its response to ocean acidification. In this study, we investigated differences in calcification between L. retroversa individuals exposed to ocean carbonate chemistry conditions of the past (pH 8.19; mid-1880s), present (pH 8.06), and near-future (pH 7.93; predicted for 2050) in the sub-Antarctic. After 3 days of exposure, calcification responses were quantified by calcein staining, shell weighing, and Micro-CT scanning. In pteropods exposed to past conditions, calcification occurred over the entire shell and the leading edge of the last whorl, whilst individuals incubated under present and near-future conditions mostly invested in extending their shells, rather than calcifying over their entire shell. Moreover, individuals exposed to past conditions formed larger shell volumes compared to present and future conditions, suggesting that calcification is already decreased in today’s sub-Antarctic waters. Shells of individuals incubated under near-future conditions did not increase in shell weight during the incubation, and had a lower density compared to past and present conditions, suggesting that calcification will be further compromised in the future. This demonstrates the high sensitivity of L. retroversa to relatively small and short-term changes in carbonate chemistry. A reduction in calcification of L. retroversa in the rapidly acidifying waters of the sub-Antarctic will have a major impact on aragonite-CaCO3 export from oceanic surface waters to the deep sea.

Calcium carbonate produced by marine calcifiers, and its subsequent export from the surface waters into the deep, drives the ocean's "carbonate pump" (Holligan and Robertson, 1996;Riebesell et al., 2009). The carbonate pump is one of the key processes in the ocean carbon cycle, and is often described as a CO 2 counterpump because precipitation of CaCO 3 tends to increase the dissolved CO 2 concentration (Zeebe, 2012). The strength of the carbonate pump depends mainly on the relative amount of precipitated CaCO 3 by larger calcifying plankton, as they sink down to the deep relatively quickly because of their size and weight (Bernard and Froneman, 2009;Buitenhuis et al., 2019). As the main planktonic producers of aragonite, thecosome pteropods are important contributors to the carbonate pump (Fabry, 1990;Bednaršek et al., 2012b;Manno et al., 2018;Buitenhuis et al., 2019). The rise in atmospheric CO 2 , and, in turn, decreasing sea surface carbonate concentration are expected to have a profound impact on the calcification efficiency of marine calcifiers, and subsequently affect the strength of the carbonate pump.
Limacina species are the most abundant shelled pteropods in polar and subpolar waters (Bé and Gilmer, 1977;Hunt et al., 2008;Bednaršek et al., 2012b). Recent pteropod research has focused on the Southern Ocean species Limacina helicina antarctica, the most abundant pteropod species south of the Antarctic Polar Front (Hunt et al., 2008). Their shells were found to already start dissolving in the natural environment of the Southern Ocean where waters are seasonally becoming undersaturated with respect to aragonite (Bednaršek et al., 2012a). Other negative responses were found when L. helicina antarctica pteropods were exposed to acidified conditions in experimental incubations, including suppressed metabolism (Seibel et al., 2012), increased mortality and reduced calcification in larval stages (Gardner et al., 2018), and reduced egg development (Manno et al., 2016). Hence, the ecological and biogeochemical importance of L. helicina antarctica in Antarctic waters, and their vulnerability to ocean acidification has been well documented. However, little is known about calcification and vulnerability to ocean acidification of Limacina retroversa (Fleming, 1823), the most abundant shelled pteropod in the sub-Antarctic waters, north of the polar front, and a key species of the region's zooplankton community (Bernard and Froneman, 2009;Roberts et al., 2014). L. retroversa has a bipolar distribution pattern, and, in the Southern Ocean, it is named L. retroversa australis (Eydoux and Souleyet, 1840) from now on referred to as L. retroversa. This species can reach up to 800 individuals/m 3 (Hunt et al., 2008), and display diel vertical migration down to a depth of 150 m (Hunt et al., 2008). A shallowing of the aragonite saturation horizon to a depth of ∼100 m is projected to happen as early as 2050 in the sub-Antarctic waters (Negrete-García et al., 2019). Given L. retroversa's major role in the sub-Antarctic ecosystem, it is important to understand how acidified conditions will affect this species' calcification.
For this study, we incubated sub-Antarctic L. retroversa for 3 days in ocean carbonate chemistry conditions of the past (pH = 8.19; representative of the mid 1880s), present (pH = 8.06), and near-future (pH = 7.93; predicted for 2050) to examine their calcification response. Active calcification was determined using calcein indicator, a fluorescent marker that is only incorporated in the shell where active calcification occurs. The amount of total CaCO 3 was estimated by measuring shell weight. Shell thickness, volume, and density were measured using micronscale computed tomography (Micro-CT) scanning of the same shells. This X-ray technique can quantify variation in calcification of pteropod shells at high spatial resolution (Howes et al., 2017;Oakes and Sessa, 2020;Mekkes et al., 2021), but has not yet been applied to short-term experimental incubations. The combination of these different measurements on the same shells allows for better understanding of how changing ocean carbonate chemistry impacts pteropod calcification in the rapidly acidifying waters of the sub-Antarctic.

Pteropod Collection and Calcein Staining
Limacina retroversa was collected during the Atlantic Meridional Transect (AMT) 27 (DY084/085) cruise with the RRS Discovery, on October 30, 2017, at 51 • 56 14.4 S, 35 • 03 37.8 W, in the sub-Antarctic Zone (Figure 1). Using a ring net (200 µm mesh, 1 m diameter, 1 L closed cod-end) three slow, 20 min oblique tows were taken from a maximum depth of 130 m to the surface, between 00:39 and 01:48 AM local time. Undamaged and lively swimming juvenile pteropods with a height of ∼0.65 mm were immediately sorted on crushed ice using a light microscope. A representative subsample of specimens (N = 25) was taken prior to the incubations (T 0 ) to serve as baseline for potential shell growth at the end of the incubations.
Specimens were incubated in 16.8 mL 6-well plates with 0.2 µm filtered sea water with calcein indicator and maintained in the dark for 2 h (MERCK Calcein, CAS 1461-15-0, concentration 50 mg/L in seawater filtered through a 0.2 µm filter). Calcein indicator is commonly used to investigate calcification of pteropods (e.g., Comeau et al., 2009). After staining, specimens were gently rinsed with 0.2 µm filtered ambient seawater, and introduced into the experiment. Calcein is thought to be incorporated into pteropod shells in different ways. One way is by integration of calcein into all shell layers. In previous studies, this was recorded as either a distinct green line formed at the leading edge of the aperture at the time of calcein staining (Comeau et al., 2009), or along the region of apertural extension during FIGURE 1 | Station where Limacina retroversa was collected on October 30, 2017, in the sub-Antarctic Zone (black triangle). Lines represent mean positions of the sub-Antarctic front (SAF) and the polar front (PF) (Orsi et al., 1995). experiments (Lischka et al., 2011). Calcein can also be integrated along the inner prismatic shell layer only, which is assumed to represent thickening of the shell (Lischka et al., 2011).

Ocean Acidification Experiment
To simulate past and future carbonate chemistry, two 60 L sealed barrels filled with 0.2 µm filtered surface seawater were bubbled with a gas mixture of either 180 ppm CO 2 (past) or 795 ppm CO 2 (future) for 12 h inside a temperature-controlled lab onboard. Two additional 60 L barrels were filled with filtered ambient surface seawater, for incubations representing present carbonate chemistry and the control without added pteropods, respectively. Each treatment (including the control) consisted of three replicates of 6 L carboys (3 replicates × 4 treatments = 12 carboys), filled with seawater from the four 60 L barrels right before pteropods were collected. Dried microalgae (a mixture of 33.3% Phaeodactylum, 33.3% Nannochloropsis, and 33.3% Tetraselmis) were added to each of the carboys at a concentration of 0.6 mg/L (3.6 mg per carboy) as a food source. A total of 13-16 calcein-stained L. retroversa were introduced into each carboy randomly, which were subsequently sealed air-tight.
The carboys were incubated for 3 days inside an open tank positioned on the deck of the ship and covered with blackout fabric. During the incubation the ship traveled from 51 • 56 14.4 S, 35 • 03 37.8 W to 53 • 11 05.3 "S, 42 • 24 56.9 W. Using continuous flow-through of ambient sea surface water, incubation temperature remained stable throughout the duration of the experiments (mean temperature ± SD of 1.83 ± 0.49 • C). We chose to conduct a short-term experiment because shelled pteropods are notoriously difficult to keep under laboratory conditions (Howes et al., 2014;Maas et al., 2018). To further reduce captivity effects, pteropods were fed plentiful microalgae to prevent starvation, which could affect calcification (Ramajo et al., 2016).
After 3 days of incubation inside the carboys, swimming pteropods were carefully sucked into a 20 mL glass pipette to be transferred into a Petri dish, and briefly examined under a light microscope to verify that they were undamaged, alive, and actively swimming. Only one of the 37 specimens was found dead after 3 days of incubation under the present conditions, and thus excluded from further analyses. Given the bright green color observed in the stomachs of all pteropods, they had been feeding successfully. At the start and end of the incubations, temperature, salinity, and pH were measured and dissolved inorganic carbon (DIC) samples were collected from all carboys. Pteropods were rinsed with ultrapure (Milli-Q) water prior to being flash frozen in liquid nitrogen. Until analyses, all specimens were stored at −20 • C.

Water Chemistry
Filtered samples for DIC measurements were stored in airtight 5 mL glass vials after preservation with 15 µL mercury (II) chloride (HgCl 2 ). DIC concentrations were analyzed with a Technicon Traacs 800 autoanalyzer (SEAL Analytical, Germany) (Stoll et al., 2001). A benchtop pH meter (HI5522-02, Hanna Instruments, Nieuwegein, Netherlands) with a glass electrode was used for pH measurements on the NBS scale. Total Alkalinity (TA), partial pressure of CO 2 (pCO 2 ), and aragonite saturation state ( Ar ) were calculated from measurements of DIC, pH, salinity, and temperature using the program PyCO2SYS version 1.3 (Humphreys et al., 2020). The calculations used the dissociation constants for carbonic acid (K 1 and K 2 ) of Millero et al. (2002), and the KHSO 4 dissociation constant of Dickson (1990). Furthermore, the calculations included the silicate and phosphate concentrations (31.92 and 1.36 µmol/kg, respectively) measured in the seawater collected for the experiments using a Bran + Luebbe AAIII autoanalyzer.
To compare the incubation experiments with in situ conditions, the same oceanographic variables were sampled at 5, 40, and 60 m depth with a CTD/rosette sampler (Seabird).

Analyses of Calcein Stained Shells
Shell cleaning, weighing, and fluorescent imaging was carried out at the Royal Netherlands Institute for Sea Research (NIOZ), Texel, Netherlands. After freeze-drying, all organic material was removed from the shells by dry oxidation in a Tracerlab low temperature (∼100 • C) asher for 90 min (Fallet et al., 2009). Pteropod shells were individually rinsed with 96% ethanol and Milli-Q water to remove any ash residue, and air dried for at least 24 h in a desiccator. Removal of organic matter is crucial to accurately measure shell weight and to record calcein integration in the shells because organic residues can emit a false fluorescence signal. To demonstrate that all organic material was removed successfully, SEM images were taken of individuals from each of the experimental treatments (Supplementary Figure 1).
A total of 111 shells were imaged using a Zeiss Axioplan 2 microscope with a Colibri light source and fluorescence filter (excitation 485/20 nm, FT 510) (Supplementary Table 1). When exposed to an excitation wavelength of 485 nm (cyanblue), calcein has its fluorescence maximum at 515 nm (green) and marks the active calcification area of the shells during the experiments. Based on the green-emitting light of calcein incorporated in the shells, we distinguished three patterns of active calcification: no glow, apertural glow (leading edge of the last whorl) and full shell glow (Figure 2). Glow categories for each shell were determined by inspecting and rotating the shell under a fluorescence microscope. Representative images of each specimen are included in Supplementary Figure 2.
To demonstrate which shell layers had incorporated calcein, a shell with apertural glow and a shell with full shell glow were embedded in resin and cross-sections were made. This gives more insight into patterns of active calcification. In the case of apertural glow, all shell layers at the aperture are expected to glow, as a reflection of accretionary shell extension. In the case of fully glowing shells, a combination of shell extension (glow of all shell layers at the aperture) and shell thickening (glow of solely the inner prismatic layer across the entire shell) is expected.

Shell Biometry
A subset of intact shells was used for weighing and Micro-CT scanning as visual inspection by light microscopy showed that some shells were damaged during handling for fluorescence imaging. To estimate shell growth, between 13 and 15 undamaged shells were selected from each treatment (a total of 42 individuals; Supplementary Table 1) and weighed using a Sartorius microbalance Model PRO 11 (Sartorius AG, Göttingen, Germany) with a resolution of 0.001 mg. The 25 shells collected at T 0 were weighed to estimate the mean shell weight prior to the incubations. Each specimen was weighed three times to account for random error of the weight measurements (±0.493 µg). Subsequently, the shells were scanned using a micro-CT scanner (SkyScan, model 1172, Aartselaar, Belgium) at Naturalis Biodiversity Center, Leiden, Netherlands. Scans were carried out using 60 kV and a scan resolution of 1.0 µm voxel FIGURE 2 | Active calcification as detected by calcein incorporation using fluorescence microscopy. Calcein incorporation in Limacina retroversa showed three different patterns after a 3-day incubation: full glow, apertural glow, and no glow. size was achieved. X-ray projections were reconstructed using NRecon ver. 1.6.6.0 (SkyScan) and introduced into Avizo 9.5 3D software (ThermoFisher Scientific, Waltham, MA, United States) to generate 3D models of the shells for calculating shell thickness, volume, and density. The embedded thickness-measuring tool "Thickness map" in Avizo was used to calculate shell thickness (µm) for each voxel in a binary image, defined as the diameter of the largest sphere containing the voxel. Subsequently, the average shell thickness per shell was calculated. Height of the shells was used as an indication of size and was measured using the measuring tool of Avizo. Shell density was calculated based on the formula: ρ = m/v where ρ is density, m is the shell weight, and v is the volume of calcified material (derived from the micro-CT scanned shells).

Statistics
A full pairwise matrix of Pearson's correlation coefficients for all carbonate variables was determined to assess the relationships among these variables. For potential differences in calcification patterns between treatments, Pearson's Chi-square test was used. To establish whether pteropod shell weight, height, thickness, and density varied among stations, a one-way ANOVA was performed, and a subsequent Tukey's HSD post hoc test to indicate differences between treatments. Normality of the residuals in the ANOVA was tested with a Shapiro-Wilk test and homogeneity of variance with Levene's test. Statistical analyses were conducted in R (R Core Team, 2018), using the packages lme4 (Bates et al., 2015) and vegan (Oksanen et al., 2018).

Carbonate Chemistry
Carbonate chemistry measurements show that the applied treatments achieved realistic past, present and near-future conditions ( Table 1). Under the applied past conditions, pH was 8.19, which is 0.13 pH units higher than under the applied present conditions and representative of the mid-1880s (assuming a decrease in pH of 0.001 units per year in sub-Antarctic waters; Kitidis et al., 2017). Under the applied near-future conditions, pH was 7.93, which is ∼0.13 pH units lower than present conditions and similar to predicted conditions for 2050 under IPCC Representative Concentration Pathway RCP8.5 (Hartin et al., 2016). The water was supersaturated with respect to aragonite ( Ar > 1) in the past and present conditions, but close to the aragonite saturation point ( Ar ≈ 1) in the near-future condition ( Table 1). The carbonate chemistry parameters ( Ar , pH, pCO 2 , DIC) in the experiment were all strongly correlated with each other, except for TA (Supplementary Table 2). After 3 days of incubation, most experimental parameters in the treatment under present conditions were still very similar to the in situ values measured in seawater at the sampling location (Table 1), although pCO 2 and the DIC concentration in the incubations were slightly higher and Ar was slightly lower than the in situ values (possibly due to respiration by the incubated community).

Active Calcification
The patterns of active calcification differed consistently between the treatments (Pearson's Chi-square test: χ 2 = 14.841, df = 4, p < 0.01; Figure 3 and Supplementary Table 3). Pteropods exposed to past conditions either calcified over their complete shell (full glow: ∼48% of all shells) or extended their shell at the aperture (apertural glow: ∼48% of all shells) and only one of the 31 shells did not show any calcification. Pteropods from present and near-future conditions predominantly extended their shells at the aperture (apertural glow: ∼67 and ∼70% of all shells, respectively) and few individuals showed active calcification over their entire shell (full glow: ∼17 and ∼14%, respectively). Moreover, a much larger proportion of shells from present and near-future conditions did not show any calcification at all (∼17 and ∼16%, respectively) compared to past conditions (∼3%).
Cross-sections of a shell with full glow and a shell with apertural glow show that they incorporated calcein differently (Figure 4). Fluorescence images show that the shell with full glow incorporated calcein solely at the inner prismatic layer across most of the shell (Figure 4B; see region 1 in Figure 4C for a detailed view), but calcein was integrated into all layers (outer prismatic, middle crossed-lamellar, inner prismatic) near the aperture (region 2 in Figure 4C). This indicates that this fully glowing shell calcified across the entire inner shell and extended its shell by accretionary growth at the aperture. For the shell with apertural glow, calcein was incorporated in all shell layers (prismatic, middle crossed-lamellar, inner prismatic) along the complete region of increment near the aperture, indicating accretionary shell growth only (Figures 4D,E).

DISCUSSION
We found pronounced differences in calcification of L. retroversa during short-term exposure to ocean carbonate chemistry conditions representative of the past (mid-1880), present, and near-future. Conditions remained supersaturated with respect to aragonite, and pH was above 8, for the past and present experimental conditions, but closely approached aragonite undersaturation in the near-future treatment with pH values declining to ∼7.9 as expected for the sub-Antarctic by the year 2050 (Hartin et al., 2016;Hauri et al., 2016;Negrete-García et al., 2019). Specimens incubated under these near-future conditions had a lower shell weight and lower shell density than under present and past conditions. Moreover, L. retroversa calcified over their entire shell under past conditions, while they mainly extended their shells by apertural growth only under present and near-future conditions. This suggests that their calcification may already be affected in the present-day sub-Antarctic compared to past conditions. Since our findings are independent of shell size, and the shell weight of L. retroversa incubated in near-future conditions remained similar to initial values at T 0 , it appears that L. retroversa invested in extending their shells at the expense of a lower shell weight, density and volume when exposed to near-future ocean conditions.

Different Patterns of Active Calcification
Our results show that active calcification shifted from whole shell calcification in many pteropods exposed to past conditions toward a predominance of apertural calcification in pteropods exposed to present and near-future conditions. Calcein studies have revealed a variety of active calcification patterns in pteropods (e.g., Comeau et al., 2009;Lischka et al., 2011;Bednaršek et al., 2017). For example, in L. helicina, calcein integrated into all shell layers, leaving a distinct line on the growing edge (Comeau et al., 2009) or along the complete apertural extension (Lischka et al., 2011). Instead of apertural shell extension, a patchy calcification across the entire shell was reported for Pacific L. helicina  and Atlantic L. retroversa (Maas et al., 2018). In our study, the cross section of a fully glowing shell showed incorporation of calcein along the inner layer of the entire shell, except at the aperture where calcein was incorporated in all shell layers (Figure 4). In the cross section of a shell with apertural glow, calcein was incorporated in all shell layers, but at the apertural extension region only. These results indicate that the shift from fully glowing shells under past oceanic conditions to apertural glow under present and near-future conditions can be interpreted as a loss of calcification along the inner layers of the older (non-accretionary) part of the shell. Such a distinctive loss of an active calcification area in response to increasingly acidified conditions is a pattern that has not been recorded before.
Calcification efforts are likely enhanced when L. retroversa pteropods are exposed to high pH and Ar conditions, resulting in full shell calcification in past conditions. Shifting calcification to only the aperture in lower pH and Ar in present and near-future conditions could result from energy reallocation under less optimal conditions. By growing larger-sized shells, growth is sustained, while the energetically expensive process of calcification over the complete shell is reduced (Wood et al., 2008). Moreover, an increased proportion of shells in present and near-future conditions did not show any active calcification, suggesting that calcification may cease entirely  during energetically more expensive times when exposed to suboptimal calcification conditions, as reported by Lischka and Riebesell (2012).
To our knowledge, this is the first incubation experiment in which past carbonate chemistry conditions resulted in more active calcification and a larger shell volume of pteropods than present conditions. Two earlier studies also exposed pteropods to past ocean chemistry conditions (Lischka et al., 2011;Manno et al., 2012). In the study of Lischka et al. (2011), Arctic L. helicina did not produce a larger shell diameter in past conditions compared to present conditions, although their study incubated pteropods in seawater with a similar shift in pH from past conditions (pH 8.27 at 3 • C) to present conditions (pH 8.12) as in our study. Northern hemisphere L. retroversa from a Norwegian Fjord also showed no difference in calcification (based on shell mass) between pre-industrial conditions (pH 8.2) and present conditions (pH 8.0) (Manno et al., 2012), despite the larger difference in pH (0.2 units) compared to our study (0.13 units). Differences in calcification responses to acidified conditions between southern and northern hemisphere L. retroversa could indicate evolutionary and ecological differences between the two geographically disjunct populations, and might suggest that southern hemisphere L. retroversa are more susceptible to changes in carbonate chemistry than its northern counterpart.

Changing Shell Properties
Shells incubated under past and present conditions increased in weight by 57 and 31%, respectively, compared to the initial shell weight prior to the experiments ( Figure 5B). However, shell weight under near-future conditions was similar to the initial weight prior to the experiments. All shells were transparent and SEM images did not reveal dissolution marks ( Supplementary  Figure 1), suggesting that the low shell weight under nearfuture conditions was likely not due to dissolution, but rather the result of ceased aragonite deposition. Similar to the 34% lower shell weight that we found in near-future compared to present conditions, Heliconoides inflatus pteropods in the Mediterranean Sea decreased their calcification by 37% when incubated in near-future (pH 7.9) compared to present-day conditions (pH 8.1) for 3 days (Moya et al., 2016). Here, geneexpression analysis showed that genes linked to calcification were upregulated, presumably to counter a further reduction of the calcification rate. A similar finding was recorded for natural populations of North Atlantic L. retroversa, where lower shell condition and upregulation of biomineralization genes was found in response to seasonally lower Ar state (Maas et al., 2020). Hence, lower shell weight of sub-Antarctic L. retroversa in near-future conditions likely indicates that the organisms are less able to maintain calcification rates when exposed to more acidified conditions. In our study, shell density was 31% lower when exposed to acidified near-future conditions. This exceeds the decline in shell density of 15% between samples 91 years apart for Cavolinia inflexa from the Mediterranean Sea, where pH decreased by 0.1 unit (Howes et al., 2017). In our experiment, pH was 0.13 and 0.26 units lower in near-future compared to present and past conditions, respectively, which could account for the more pronounced decline in shell density. Shell thickness remained similar across treatments, whereas shell density and weight were significantly lower in near-future conditions. This indicates that the lower weight and density of L. retroversa in near-future conditions was not caused by thinning of the shells. A similar result was obtained for Creseis acicula and Diacavolina longirostris sampled between 1963 and 2009 along the Australian coast, where porosity increased for both pteropod species as a function of decreased Ar , but shell thickness remained the same (Roger et al., 2012). We argue that this could be the result of a change in packing of the aragonite nanofibers, which affects shell density while shell thickness remains unaltered. Corals also produced less dense aragonite under lower pH conditions Holcomb et al., 2009). Here, coral calcification was less dense, as the packing of the aragonite nanofibers was less densely structured. Therefore, analyzing the organization or packing of the aragonite nanocrystals of L. retroversa may give more insight into how shell density changes when pteropods are exposed to more acidified conditions.
In our experiments, shell weight and density of L. retroversa increased significantly under past and present conditions but was similar to T 0 for near-future conditions. In contrast, a field study using sediment-traps in the sub-Antarctic Zone over the period from 1997 to 2007 showed that shell weight of L. retroversa became heavier despite a slight decline in pH from 8.13 to 8.08 (Roberts et al., 2014). This may result from factors other than carbonate chemistry conditions, such as an increase in food availability, which could enhance calcification (Howes et al., 2014;Ramajo et al., 2016;Oakes and Sessa, 2020). As pH was considerably lower in the nearfuture conditions of our study (pH 7.93) and we supplied all three treatments with the same amount of food, changes in carbonate chemistry likely had a more pronounced effect on calcification in our experiments than in Roberts et al. (2014). However, even stronger negative effects of acidified conditions on L. retroversa calcification may have been mitigated by successfully feeding them in the experiments (bright green algae in their stomachs), as previous studies have shown that food shortages can severely suppress the metabolism of pteropods (Maas et al., 2011).
Similar to Moya et al. (2016), we chose a 3-day incubation period to obtain an impression of the effects of different ocean carbonate chemistry scenarios on calcification. Shelled pteropods are notoriously hard to maintain in captivity (Howes et al., 2014), and longer-duration experiments could affect calcification over time through captivity effects (Maas et al., 2018). Our observations of calcein integration are similar to the results of ocean acidification experiments by Lischka et al. (2011), who used a longer incubation period of 29 days. Lischka et al. (2011) sampled Arctic L. helicina in autumn, close to the seasonal cessation of shell growth during winter time. This differs from our study, where juvenile pteropods were sampled in austral spring. According to Dadon and de Cidre (1992), L. retroversa at the Southern Argentinian coast, near the sub-Antarctic, produces two generations per year. The first generation is produced in austral spring, when a rapid increase in size is essential to reach fertility and produce offspring for the second generation in late summer (Dadon and de Cidre, 1992;Seibel et al., 2007). The same applies to L. retroversa in the Gulf of Maine, where the phytoplankton spring bloom is thought to enable rapid growth of the juvenile pteropods (Maas et al., 2020). We mimicked a scenario of high food abundance in our experiments, by successfully feeding the incubated pteropods with microalgae (as shown by bright-green stomach contents and mucous webs) and all pteropods were still swimming lively after 3 days. Hence, under these conditions, incubating pteropods for only a few days appears sufficient to assess their calcification responses.

Impact of Reduced Calcification on the Carbonate Pump
The relocation of active calcification to only the aperture in present and near-future conditions, and the substantially lower amount of precipitated shell in near-future conditions, suggests that L. retroversa has difficulties maintaining calcification under current and increasingly more acidified conditions. Therefore, this study demonstrates a high sensitivity of juvenile L. retroversa to shallow undersaturated conditions projected to emerge in sub-Antarctic waters from 2050 onward (Negrete-García et al., 2019). A decrease of CaCO 3 precipitation by the most abundant sub-Antarctic pteropod species would cause a reduction in the amount of CaCO 3 formed in surface water and exported to the deep. Since pteropods are major contributors to the carbonate pump in the Southern Ocean (Hunt et al., 2008;Bednaršek et al., 2012b;Manno et al., 2018), reduced calcification of L. retroversa would have a profound impact on the carbonate pump in these waters.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. Micro-CT scans could be obtained on reasonable request. All fluorescence images of shells are available in the Supplementary Information.

AUTHOR CONTRIBUTIONS
LM, DW-P, and KP designed the study and performed the experiment. LM, GS-R, G-JB, WR, and LD carried out sample preparation and analysis. LM, GS-R, JH, and EL carried out data analysis. All authors contributed to interpretation of the data and writing of the manuscript, and approved the submitted version.

FUNDING
The Atlantic Meridional Transect is funded by the United Kingdom Natural Environment Research Council through its National Capability Long-term Single Centre Science Program, Climate Linked Atlantic Sector Science (grant number NE/R015953/1). This study contributes to the international IMBeR project and is contribution number 336 of the AMT program. This research was funded by a Vidi grant (016.161351) from the Dutch Research Council (NWO) awarded to KP. This project has received funding from the European Union's Horizon 2020 Research and Innovation Program under the Marie Skłodowska-Curie grant agreement no 746186 (POSEIDoN, DW-P). LD was supported by the Netherlands Earth System Science Centre (NESSC), grant number: 024.002.001 from the Dutch Ministry of Education, Culture and Science (OCW).

ACKNOWLEDGMENTS
We are very grateful to the captain, crew, and scientists of cruise DY084/085 (AMT27) on board the RRS Discovery, and to Vassilis Kitidis (Plymouth Marine Laboratory) for discussion on pH changes and carbonate chemistry calculations. We express our gratitude to Matthew Humphreys (NIOZ) for help with PyCO2SYS, Lennart de Nooijer (NIOZ) for arranging usage of instruments at the NIOZ (Texel), and Michele Grego (NIOZ) for his support during fluorescence microscopy. We would also like to thank Bertie Joan van Heuven and Rob Langelaan for their valuable insights and expertise in the lab.