Impacts of Basal Melting of the Totten Ice Shelf and Biological Productivity on Marine Biogeochemical Components in Sabrina Coast, East Antarctica

To clarify the impacts of basal melting of the Antarctic ice sheet and biological productivity on biogeochemical processes in Antarctic coastal waters, concentrations of dissolved inorganic carbon (DIC), total alkalinity (TA), inorganic nutrients, chlorophyll a, and stable oxygen isotopic ratios (δ18O) were measured from the offshore slope to the ice front of the Totten Ice Shelf (TIS) during the spring/summer of 2018, 2019, and 2020. Modified Circumpolar Deep Water (mCDW) intruded onto the continental shelf off the TIS and flowed along bathymetric troughs into the TIS cavity, where it formed a buoyant mixture with glacial meltwater from the ice shelf base. Physical oceanographic processes mostly determined the distributions of DIC, TA, and nutrient concentrations. However, photosynthesis and dilution by meltwater from sea ice and the ice shelf base decreased DIC, TA, and nutrient concentrations in surface water near the ice front. These causes also reduced the partial pressure of CO2 (pCO2) in surface water by more than 100 μatm with respect to mCDW in austral summer of 2018 and 2020, and the surface water became a strong CO2 sink for the atmosphere. Phytoplankton photosynthesis changed DIC and TA in a molar ratio of 106:16. Thus, pCO2 decreased mostly as a result of photosynthesis while dilution by glacial and sea ice meltwater had a small effect. The nutrient consumption ratio suggested that photosynthesis was stimulated by iron in the water column, supplied to the surface layer via buoyancy‐driven upwelling and basal ice shelf meltwater in addition to sea ice meltwater.

There are many coastal polynyas around Antarctica (Tamura et al., 2008).In polynyas, primary production by phytoplankton has been shown to be related to the melting rates of adjacent ice shelves and sea ice (Arrigo et al., 2015;Moreau et al., 2019).Primary production in most of the Southern Ocean is low because of iron limitation, and high production is limited to areas where iron is supplied to the surface water (de Baar et al., 1990).Sources of iron to the surface water are basal melting of ice shelves and subglacial discharged water (Arrigo et al., 2015;Herraiz-Borreguero et al., 2016), atmospheric dust (Jickells et al., 2005), vertical mixing in winter (Tagliabue et al., 2014), upwelling associated with fronts (Schallenberg et al., 2018;St-Laurent et al., 2019), and melting of sea ice (Duprat et al., 2020;Lannuzel et al., 2007).Basal melting of ice shelves is considered to be the main source of iron near Antarctic coastal areas with polynyas (Arrigo et al., 2015).In the Pine Island Polynya and Amundsen Polynya in West Antarctica, the areas of highest primary productivity on the Antarctic coast, sediment-derived iron is supplied by mixing with basal meltwater from underneath the adjacent ice shelf.Ice-core drilling at Amery Ice Shelf in East Antarctica has revealed iron-rich marine ice at the ice shelf base, created by subglacial discharged water.Its melting supplies iron to the ocean surface, resulting in the high primary production observed around the Amery Ice Shelf (Herraiz-Borreguero et al., 2016).
Our research group investigated the reduction of the partial pressure of CO 2 (pCO 2 ) due to dilution by glacier meltwater near the Shirase Glacier Tongue in Lützow-Holm Bay, East Antarctica (Kiuchi et al., 2021).In this area, the inflow of modified Circumpolar Deep Water (mCDW) from the outer edge of the continental shelf to the ice shelf grounding line is a major cause of melting at the bottom of the ice shelf (Hirano et al., 2020).The mCDW mixes with the meltwater from the ice shelf and becomes fresher and lighter, and then upwells along the ice shelf base in a vertical circulation called the ice meltwater pump (Lewis & Perkin, 1986).In this process, the dissolved inorganic carbon (DIC) and total alkalinity (TA) of the seawater are diluted by the glacial meltwater supply.As a result, the pCO 2 , which was originally 431 ± 12 μatm in mCDW, decreases by 42 ± 2 μatm because of the influence of the meltwater (Kiuchi et al., 2021).It is therefore clear that variations of the supply of glacial meltwater cause wide fluctuations in the carbonate chemistry of coastal Antarctic seas.
The Totten Ice Shelf (TIS), the focus of this study, is the terminus of the Totten Glacier in East Antarctica.The Moscow University Ice Shelf (MUIS) lies to its east (Figure 1).If the entire ice sheet behind the TIS were to flow into the ocean, the global rise of sea level would be ∼3.5 m (Greenbaum et al., 2015;Li et al., 2015).Basal melting of the TIS and MUIS occurs when warm mCDW flows onto the continental shelf and intrudes the ice cavity (Hirano et al., 2023;Rintoul et al., 2016;Silvano et al., 2017).Basal melting also occurs in ice-shelf cavities where mCDW is absent because the water temperature is warmer than the freezing point.In addition, on the continental shelf, basal melting probably happens through diapycnal mixing with the overlying layers rich in glacial meltwater (Silvano et al., 2018).Recent studies of the mCDW pathways have shown that mCDW is transported across the continental slope and intrudes the coastal area (Hirano et al., 2021(Hirano et al., , 2023;;Nitsche et al., 2017;Silvano et al., 2019).Measurements of biogeochemical components near the TIS were made by the icebreaker Aurora Australis in January 2015.According to Arroyo et al. (2019), who reported the carbonate chemistry near the TIS and MUIS, the surface of the ice front was covered by sea ice even in the summer, and primary production by phytoplankton was limited by the low-light conditions.Therefore, the surface water pCO 2 was supersaturated with respect to the atmosphere.However, better knowledge of the cross-shelf characteristics of the carbonate chemistry and biogeochemical components, which are now poorly known, would allow us to evaluate the effect of dilution by the ice shelf basal meltwater on the carbonate chemistry and primary production by phytoplankton.In addition, the components of the carbonate chemistry have been measured only once on the TIS and MUIS fronts (Arroyo et al., 2019), and seasonal changes due to the presence or absence of sea ice have not been documented.
In this study, we sampled the offshore water intruding the TIS cavity.From these samples, we evaluated the impact of altering the processes of inflow to the ice front and melting of the ice shelf and sea ice on the biogeochemical components of the surrounding sea, including near the MUIS.Our observations extended from the shelf break of the Sabrina Coast to the TIS front.In addition, observations were conducted in December (early summer) and March (late summer) to investigate seasonal changes at the ice front.The effects of differences in the environment, such as the extent of sea ice on the carbonate chemistry, were investigated in detail by conducting the studies for multiple years.We also considered the role of iron supplied by glacial and sea ice meltwater by examining the ratios of nutrients taken up by phytoplankton.

Sampling
Oceanographic observations were conducted on the continental shelf slope near the TIS in mid-February 2019 during the tenth Antarctic survey by R/V Kaiyo Maru of the Fisheries Agency (KY1804).Additional studies were carried out from the Japan Maritime Self-Defense Force icebreaker Shirase from the offshore slope to the front of the TIS in early March 2018 during the 59th Japanese Antarctic Research Expedition (JARE59) as well as in December 2019 and early March 2020 during the 61st Japanese Antarctic Research Expedition (JARE61) (Figure 1).The December 2019 observations took place along the east side of the TIS and near the MUIS.
Vertical profiles of temperature and salinity were measured with a conductivity-temperature-depth (CTD) probe (SBE 9plus, Sea-Bird Electronics, Bellevue, WA, USA, from the Kaiyo-maru and SBE19, Sea-Bird Electronics, Bellevue, WA, USA, from the Shirase).In addition, seawater samples were taken to calibrate the salinity sensor.Seawater samples were collected vertically in rosette-mounted 10-L Niskin bottles (Ocean Test Equipment, Inc., Lauderdale, FL, USA) from the Kaiyo-maru and 4-L Niskin bottles (SBE55 ECO, Sea-Bird Electronics, Bellevue, WA, USA) from the Shirase.
Seawater was subsampled into (a) a 200-mL glass vial (Maruemu Co., Ltd., Osaka, Japan) for measurement of DIC and total alkalinity (TA), (b) a 15-mL glass screw-cap vial (Nichiden-Rika Glass Co. Ltd, Kobe, Japan) for measurement of the oxygen isotopic ratio (δ 18 O) of the water, (c) a 10-mL polyethylene screw-cap vial (Eiken Chemical Co.Ltd, Tokyo, Japan) for measurement of inorganic nutrients (NO 3 − , PO 4 3− , and Si(OH) 4 ), and (d) a 300-mL Nalgene polycarbonate bottle (Thermo Fisher Scientific Inc., Waltham, MA, USA) for measurement of chlorophyll a (Chl-a) concentrations.Immediately after subsampling for measurement of DIC and TA, a 6.0% (wt.) mercury chloride (HgCl 2 ) solution (200 μL) was added to stop biological activity.Samples for DIC, TA, and δ 18 O were stored at room temperature (20°C).Samples for nutrient concentrations were stored in a freezer (−30°C).Samples for Chl-a measurements were immediately filtered onto 25-mm diameter Whatman GF/F filters.The chlorophyll on the filters was then extracted with N,N-dimethylformamide (Suzuki & Ishimaru, 1990) for 24 hr in a −80°C freezer.

Sample Analysis
The concentrations of DIC were determined by coulometry (Johnson et al., 1985(Johnson et al., , 1992) using a hand-made CO 2 extraction system (Ono et al., 1998) and a coulometer (CM5012, UIC, Inc., Binghamton, NY, USA).The TA of the seawater was determined by titration (Dickson et al., 2007) with a TA analyzer (ATT-05, Kimoto Electric Co., Ltd., Japan).Both DIC and TA measurements were calibrated against reference seawater materials (Batch AO and AP; KANSO Technos Co., Ltd., Osaka, Japan) traceable to the certified reference material distributed by Prof. A. G. Dickson (Scripps Institution of Oceanography, La Jolla, CA, USA).The standard deviations of the DIC and TA measurements, calculated from the results for 10 subsamples of the reference water with DIC = 1,987.1 μmol kg −1 and TA = 2257.6μmol kg −1 , were less than 2.0 μmol kg −1 for both DIC and TA.The seawater pCO 2 was computed from DIC and TA using the program CO2SYS, version 02.05 (Orr et al., 2018).For this calculation, we used the carbonic acid dissociation constants (K 1 and K 2 ) of Mehrbach et al. (1973) as revised by Dickson and Millero (1987) and the K HSO4 value determined by Dickson (1990).
The concentrations of Chl-a were determined with a fluorometer (Model 10AU, Turner Designs, Inc., Sunnyvale, CA, USA) by the method of Parsons et al. (1984).Standards (0.05-159 μg L −1 Chl-a) prepared from a liquid Chl-a standard (Wako Pure Chemical Industries, Ltd., Osaka, Japan) by stepwise dilution with N,N-dimethylformamide were used to calibrate the fluorometer before Chl-a measurements.

Hydrographic and Biogeochemical Properties
Results of all measurements are presented in Tables S1 and S2 in Supporting Information S1.Figures S1 through S5 in Supporting Information S1 present vertical sections of temperature, salinity, and biogeochemical properties of the ocean in the study area.We discuss these below in separate geographic domains.

The TIS Front and Coastal Areas
In the TIS front and coastal areas (near MUIS), low-temperature and low-salinity water was present in the WW layer, similar to on the continental slope.The presence of high-temperature and high-salinity water in the mCDW layer of the TIS front is consistent with the observations of Rintoul et al. (2016).This mCDW layer had a lower temperature (<+0.16°C) and a lower salinity (<34.62)than the mCDW on the outer edge of the continental shelf.
The DIC concentration at the TIS front and in coastal surface water (20 dbar) was 2086 ± 12 μmol kg −1 in March.In March 2018, the DIC concentration tended to decrease at the stations furthest to the west: Stations TT2 (2,079 μmol kg −1 ) and TT3 (2,068 μmol kg −1 ) (Figure S2c in Supporting Information S1).The DIC was 2,202 ± 4 μmol kg −1 in the WW layer at 200 dbar and 2,233 ± 6 μmol kg −1 in the mCDW layer in the bottom trough.The TA followed a pattern similar to that of the DIC, low on the surface (2,253 ± 11 μmol kg −1 ) and high in the mCDW layer (2,337 ± 7 μmol kg −1 ).The distributions of nutrient concentrations were similar to the distributions of DIC and TA: low at the surface (NO 3 − : 20.0 ± 1.8 μmol kg −1 , PO 4 3− : 1.34 ± 0.10 μmol kg −1 , Si(OH) 4 : 43.8 ± 0.7 μmol kg −1 ) and higher in the mCDW layer (NO 3 − : 32.7 ± 0.7 μmol kg −1 , PO 4 3− : 2.15 ± 0.08 μmol kg −1 , Si(OH) 4 : 87.1 ± 7.8 μmol kg −1 ) (Figure S5h-S5j in Supporting Information S1).The δ 18 O was low in the WW layer and was vertically uniform.The δ 18 O was higher in the deep mCDW layer than in both the subsurface mCDW and WW layers (Figure S5g in Supporting Information S1).The vertical profiles of Chl-a concentrations at 0-100 dbar differed from year to year in their distribution and concentration (Figure S5k in Supporting Information S1).On average, the Chl-a concentrations at each depth in 2018 were about twice those of the other two years.The averaged profiles showed that Chl-a decreased at greater depth and reached a minimum at 100 dbar, although some profiles in 2018 had Chl-a maxima at 50 dbar.
Figure S5 in Supporting Information S1 shows vertical profiles of various water characteristics for December and March at the ice front and in coastal areas.At the ice front, the water temperature increased from −1.5°C to −1.1°C in the surface water (0-100 dbar) from December to March, and the salinity decreased from 34.0 to 33.0.The salinity reduction was likely due to the increased meltwater input that accompanied the rise in water temperature from December to March.In contrast, at depths below 100 dbar, the water temperature and salinity had almost no change between December and March (−1.78 ± 0.06°C, 34.23 ± 0.04), whereas DIC, TA, and nutrient concentrations decreased at the surface but remained constant below 100 dbar.The δ 18 O increased in the surface water between December and March.
From December to March, the salinity of the surface water decreased as the meltwater input increased.Therefore, to remove this dilution effect on DIC and TA, we normalized the DIC and TA to a salinity of 34.3 (the value used as a reference by Arroyo et al. (2019); Figures S5e and S5f in Supporting Information S1).At depths below 100 dbar, there was no change in normalized DIC and TA, but at the surface, normalized DIC decreased and normalized TA increased from December to March., and Si(OH) 4 for each water mass and month.

From the Continental Slope to the Ice Front and Coastal Areas
Figure 2 shows various cross-sectional views from the continental slope to the ice front and coastal areas.The low-temperature, low-salinity WW layer was present at 100-700 dbar (Figures 2a and 2b).The high-temperature, high-salinity mCDW layer near the seafloor below 700 dbar at the ice front extended continuously from the offshore mCDW layer at the continental slope.The mCDW layer became colder (<+0.16°C) and fresher (<34.62)from the continental slope to the ice front and coastal areas (Figures 2a  and 2b).The distributions of the biogeochemical properties and hydrographic properties were very similar (Figures 2c-2g): concentrations of biogeochemical properties in the mCDW layer were high on the continental shelf and high, but lower, at the bottom of the ice front.In contrast, there were places where the concentrations of NO 3 − and PO 4 3− in the mCDW layer were higher at the bottom of the ice front than on the continental shelf (Figures 2c and 2f).

Variation of pCO 2 by the Dilution Effect
The dilution effect of ice shelf meltwater and sea ice meltwater is an important process affecting the carbonate chemistry in the ocean surface along the Antarctic coast in summer (e.g., Arroyo et al., 2019;Kiuchi et al., 2021;Legge et al., 2017;Shadwick et al., 2017).In this study, observations in December and March showed that DIC and TA decreased with decreasing salinity.Because these changes could be caused by either sea ice meltwater or TIS basal meltwater, we quantitatively evaluated the fraction of TIS basal glacier meltwater, sea ice meltwater, and mCDW in the collected water samples by using salinity and δ 18 O (e.g., Meredith et al., 2008).In addition to TIS basal glacier meltwater, there is direct precipitation as meteoric water.Silvano et al. (2018) examined the contribution of the direct precipitation to the meteoric water in the Sabrina Coast (same area as our study).They concluded that the contribution of the direct precipitation is low, and glacial meltwater is the dominant source of meteoric water (Silvano et al., 2018).The fraction of TIS basal glacier meltwater, sea ice meltwater, and mCDW in the collected water samples was expressed by the following equations:  2021) have pointed out that the δ gmw is the most uncertain end-member and could lie anywhere between −40‰ and −20‰ (Silvano et al., 2018).In our study, we chose an intermediate value of −30‰ for δ gmw , but we also calculated fractions based on δ gmw ranging between −40‰ and −20‰.This analysis suggested that our choice of δ gmw = −30‰ yielded fractions accurate to 1.1%.The 1.1% error was obtained from the full range (maximum) of the uncertainty of the glacial end-member component in all of Antarctica (δ gmw ranging between −40‰ and −20‰).For Totten Glacier (the local area), we expected that the range of the glacial end-member component would be narrower than that for all of Antarctica.Therefore, range of the error will be lower than 1.1%.A similar test for the δ 18 O variation within its standard deviation and the choice of mCDW end-member (i.e., δ mCDW = −0.06± 0.005‰) yielded estimated potential errors of less than 0.1% for the δ 18 O measurement and less than 0.3% for the mCDW end-member.
Figures 3a-3c show F gmw , F simw , and F mCDW , respectively, in the TIS front and coastal areas.The F gmw in the surface water (20 dbar) was 1.53 ± 0.09% (mean ± standard deviation for all stations) in March 2018, 1.64 ± 0.03% in March 2020, and 1.67 ± 0.09% in December 2019.Therefore, there was no seasonal change between December and March.The F gmw was slightly lower and uniformly distributed in 200-500 dbar, and lowest (0.32%) in the deep layer.These results (high F gmw at surface water and low F gmw at deep layer) suggest that the mCDW that intruded from the outer edge of the continental shelf was diluted with glacial meltwater on the continental shelf along the way to the ice shelf cavity.The F simw in the surface water (20 dbar) was 2.91 ± 0.13% in March 2018 and 3.87 ± 0.05% in March 2020.Therefore, sea ice melting was more active in 2020 than in 2018.However, the F simw in December 2019 was as low as 0.18 ± 0.20%.
The longitudinal distributions of F gmw and F simw show that F gmw in the surface layer (20 dbar) in March 2018 was low (1.44%-1.47%)on the east side of the study area (Sts.TT1, 4, 5, and 6) and high (1.63%-1.67%)on the west side (Sts.TT2 and 3) (Figure 3d).This pattern suggests that a large amount of glacial meltwater flowed from the east side to the west side of the TIS.In addition, F gmw in the surface layer in December 2019 was low (1.52%-1.54%)on the east side of the study area (Sts.TV20, 21, and 22) and high (1.72%) on the west side (Sts.TV13, 15, 16, 17 and 19) (Figure 3f), although F gmw in March 2020 was low (1.59%-1.67%)at the same longitudes as well as on the far eastern side (<117.4°E)(Figure 3e).These results suggest seasonal variations in the flow of meltwater and an effect of meltwater from the MUIS (Silvano et al., 2017).
Next, we quantitatively evaluated the effects of glacial meltwater and sea ice meltwater on the carbonate chemistry from the following relationships between F gmw , F siw , DIC, TA, and salinity (Kiuchi et al., 2021).
In the above equations, ΔS gmw and ΔS simw are changes of salinity by glacial meltwater and sea ice meltwater, respectively, ΔDIC mix and ΔTA mix are changes of DIC and TA by mixing of glacial meltwater and sea ice meltwater with mCDW, and M DIC-S and M TA-S are the slopes of the DIC-salinity and TA-salinity relationship between mCDW and the origin.
Using the changes of ΔS gmw , ΔS simw , ΔDIC mix , and ΔTA mix calculated with Equations 4-7, we used CO2SYS, version 02.05 (Orr et al., 2018) to calculate the pCO 2 gmw when the mCDW flowing under the ice shelf was mixed with the basal meltwater of the TIS as well as the pCO 2 simw when the mCDW was mixed with sea ice meltwater.
Figure 4 the pCO 2 and its change (ΔpCO 2 ) when mCDW on the ice front was mixed with each water mass.The pCO 2 gmw was 440 ± 0.7 μatm (ΔpCO 2 gmw : −11 ± 0.6 μatm) in December 2019, 440 ± 0.2 μatm  4b).At the ice front, the pCO 2 in seawater therefore decreased consistently by 11 μatm via the dilution effect of the glacial meltwater from the ice shelf.There were no seasonal changes between March and December or interannual changes between 2018 and 2020.
The pCO 2 gmw+simw when sea ice and glacial meltwaters were mixed was 438 ± 1.4 μatm (ΔpCO 2 gmw+simw : −13 ± 1.4 μatm) in December 2019, 418 ± 0.5 μatm (ΔpCO 2 gmw+simw : −33 ± 0.5 μatm) in March 2020, and 424 ± 1.2 μatm (ΔpCO 2 gmw+simw : −27 ± 1.2 μatm) in March 2018 (Figures 4a and 4b).These results indicate that the ice shelf front reduced the pCO 2 in seawater via the dilution effect of mixing with meltwater, but the resulting seawater was still supersaturated with respect to atmospheric pCO 2 (saturation is at about 390 μatm).The dilution effect of glacial meltwater on the ice front in March corresponded to a 2.4 ± 0.1% decrease of pCO 2 , and the dilution effect of sea ice meltwater corresponded to a 4.1 ± 0.6% decrease.In contrast, the dilution effect of the basal meltwater from the Shirase Glacier Tongue calculated in Lützow-Holm Bay corresponded to a decrease of 9.6 ± 0.4% of pCO 2 (Kiuchi et al., 2021).In our study, the pCO 2 observed at each surface observation point (pCO 2 obs ) was 377 ± 46 μatm in December 2019, 246 ± 2.1 μatm in March 2020, and 232 ± 21 μatm in March 2018 (Figure 4a).Therefore, the pCO 2 was decreased in the ice front by other factors besides meltwater dilution, especially photosynthesis of phytoplankton.In the following section, we discuss the effect of photosynthesis on the surface seawater pCO 2 .

Factors Controlling the Changes of DIC and TA at the Ice Front and Coastal Areas
In the ocean surface layer along the Antarctic coast in summer, primary production by phytoplankton and the precipitation and dissolution of calcium carbonate (CaCO 3 •6H 2 O) are important biological processes that affect the carbonate chemistry of the ocean (e.g., Arroyo et al., 2019;Legge et al., 2017;Shadwick et al., 2017).When the DIC/TA ratio changes as a result of photosynthesis by phytoplankton, the phytoplankton take up DIC (C) and nitrate (N) in the Redfield ratio (global average C:N = 106:16 by atoms) (Redfield et al., 1963).The ratio of the change of DIC to the change of TA is therefore 106:16 on a molar basis.In addition, when DIC/TA changes because of the precipitation and dissolution of calcium carbonate, it changes in the ratio DIC:TA = 1:2 (Zeebe & Wolf-Gladrow, 2001).
Figure 5 shows the relationship between normalized DIC (nDIC) and normalized TA (nTA), which were standardized to a salinity of 34.3, after the dilution effect of the meltwater was removed.Because the plot of the ice front surface layer (0-100 dbar) did not follow the slope of nDIC:nTA = 1:2, the precipitation and dissolution of calcium carbonate was not the main factor influencing the carbonate chemistry of the surface layer along the Sabrina Coast.This result  is consistent with observations made in the area from 31 December 2014 to 1 January 2015 (Arroyo et al., 2019).However, the fact that the plot followed the slope of nDIC:nTA = 106:16 indicates that the nDIC and nTA changed as a result of photosynthesis of phytoplankton from the previous winter to summer.
In addition, the concentrations of nutrients (NO 3 − , PO 4 3− , and Si(OH) 4 ) in the surface layer in March were lower than those in the WW layer (Figures S5h-S5j in Supporting Information S1).These results indicate that photosynthesis by phytoplankton was vigorous in the ice and the surrounding sea area in austral summer, consuming DIC and nutrients in the seawater.

Net Community Production
To estimate the changes of DIC due to biological processes, net community production (NCP) was calculated using NO 3 − concentrations normalized to a salinity of 34.3 (nNO 3 ) (e.g., Shadwick et al., 2014).We defined NCP as the integrated change of the nNO 3 concentration at water depths of 20-100 m from winter to summer: In Equation 11, z is the water depth, [nN] winter is the nNO 3 − concentration in winter, and [nN] obs is the nNO 3 − concentration at the time of observation.The temperature minimum layer is generally used to compare water masses in winter and summer (e.g., Bates et al., 1998;Ishii et al., 2002), but over the Antarctic continental shelf, it is difficult to define the temperature minimum layer (Murakami et al., 2020).Therefore, in this study, because the NO 3 − concentrations at a water depth of 200-300 dbar at the time of observation were vertically uniform at each observation date, we averaged the NO 3 − concentrations (31.9 ± 0.4 μmol kg −1 for March 2018, 30.3 ± 0.3 μmol kg −1 for December 2019, and 30.4 ± 0.7 μmol kg −1 for March 2020) and assumed that these were the NO 3 − concentrations at depths of 20-100 dbar in winter.NCP was converted to a carbon basis by using the Redfield ratio (C:N = 106:16 by atoms).We then calculated NCP per day by assuming that photosynthesis started on 1 November 2019 (e.g., Arroyo et al., 2019).
Figure 6 shows the NCP at the ice front and coastal areas.The NCP for March 2018 was +28.7 ± 3.8 mmol m −2 day −1 (Figure 6a).It was particularly high at Station TT3 in the western part of the study area (+35.9 mmol C m −2 day −1 ) (Figure 6b).The NCP for March 2020 was +39.9 ± 3.7 mmol m −2 day −1 , higher than the NCP for March 2018 at all stations (Figures 6a and 6c).In contrast, the NCP for December 2019 was −1.6 ± 9.0 mmol C m −2 day −1 (Figure 6a).In particular, in the western part of the study area (Sts. TV 13,15,16,17,and 19), the negative values of NCP indicated the predominance of respiration over photosynthesis.In the eastern part of the study area (Sts.TV20, 21, 22), the NCP was positive (Figure 6d).The NCP calculated in the same area from 31 December 2014 to 1 January 2015 (−3.8 to +6.6 mmol m −2 day −1 ) (Arroyo et al., 2019) is consistent with the December NCP calculated in this study.
We next used satellite images to compare the state of the ocean surface at each observation time.There was less sea ice in March than in December (Figures 1b-1d).Moreau et al. (2019) have reported that NCP is positively correlated with the proportion of sea ice meltwater.Consistent with their result, we found that NCP was positively correlated with F simw (Figure 7a).It appears that photosynthesis and NCP increased from December to March because the open water surface area was greater and the light environment was better because of the melting of sea ice.In contrast, NCP was not correlated with F gmw for December and March (Figure 7b).As indicated in Section 4.1, F gmw was the same in December and March; therefore, this result was expected.
The relationship between the fractions of meltwater (F gmw , F simw ) and NCP was apparent only in the surface layer (20 dbar) in March 2018 and March 2020 (Figures 7b and 7c).Both F gmw and F simw were positively correlated with NCP, and the higher the fractions of meltwater, the higher the NCP tended to be.This result was likely due to stratification caused by the influence of the basal meltwater from the ice shelf and the melting of sea ice, the difference in the open water surface area, and a stable environment for the phytoplankton.It may also have been affected by specific substances in the two meltwaters that promote phytoplankton growth, such as iron (Herraiz-Borreguero et al., 2016;Lannuzel et al., 2007).The relationships between NCP and the proportions of the two meltwaters (Figures 7c and 7d) differed in their slopes: F gmw versus NCP had a slope of 53.9 and F simw versus NCP had a slope of 12.2.This result suggests that glacial meltwater had a greater effect on biological productivity than sea ice meltwater.In Section 4.4, we therefore assess the role of iron supplied by glacial meltwater by examining the ratios of nutrients taken up by phytoplankton.

Possibility of Iron Supply at the TIS Front
Primary production of Antarctic surface water is restricted mainly by iron (Martin, 1990;Moore et al., 2013).The sources of iron to surface water include basal melting of ice shelves and subglacial discharged water (Arrigo et al., 2015;Herraiz-Borreguero et al., 2016), atmospheric dust (Jickells et al., 2005), vertical mixing in winter (Tagliabue et al., 2014), upwelling associated with fronts (Schallenberg et al., 2018), and melting of sea ice (Duprat et al., 2020;Lannuzel et al., 2007).Iron stimulates primary production in the surface layer of the Southern Ocean.In general, the uptake ratio of nutrients by phytoplankton differs under iron-limited and iron-replete conditions; Si/N and Si/P values are larger under iron-limited conditions than under iron-replete conditions (Takeda, 1998).To evaluate the presence or absence of iron supplied by meltwater in the TIS front surface layer (20 dbar), we calculated the nutrient consumption ratios from winter to summer (ΔSi/ΔN, ΔSi/ΔP) using Equations 12 and 13 and the nutrient concentrations in seawater: ) in the WW layer, and nSi obs , nN obs , and nP obs are the nutrient concentrations normalized to the salinity of the WW (34.21 for March 2018, 34.18 for March 2020) in the surface layer (20 dbar) at the times of our observations.In addition, the average nutrient concentrations in the WW layer (62.3 ± 3.8 μmol kg −1 for Si(OH) 4 , 31.9 ± 0.4 μmol kg −1 for NO 3 − , and 1.96 ± 0.04 μmol kg −1 for PO 4 3− for March 2018, and 61.1 ± 4.2 μmol kg −1 for Si(OH) 4 , 30.4 ± 0.7 μmol kg −1 for NO 3 − , and 2.10 ± 0.01 μmol kg −1 for PO 4 3− for March 2020) were used as the nutrient concentrations in the surface layer in winter.
We evaluated the possibility of iron supply by meltwater by examining the relationships between seasonal nutrient differences (ΔSi/ΔN, ΔSi/ΔP) and meltwater fractions (F gmw , F simw ).Our comparison showed negative correlations: ΔSi/ΔN and ΔSi/ΔP were lower when F gmw and F simw were higher (Figure 9).Moreover, F gmw was more strongly correlated with ΔSi/ΔN and ΔSi/ΔP than was F simw .These results suggest that glacial meltwater, rather than sea ice meltwater, supplied iron to the ocean surface and contributed to biological productivity.Takeda (1998) has proposed that ΔSi/ΔN lower than 1.9-2.3mol mol −1 and ΔSi/ΔP less than 16-42 mol mol −1 represent iron-replete conditions based on both in vitro and laboratory culture experiments.By this definition, ΔSi/ΔN indicated iron-replete conditions (Figure 8a) but ΔSi/ΔP did not (Figure 8c).Takeda's (1998) laboratory culture experiments used filtered and sterile seawater and two specific phytoplankton strains (Chaetoceros dichaeta and Nitzschia sp.) under continuous light conditions to support the ΔSi/ΔP range of 16-642 mol mol −1 .Instead, we decided to use the control value (26 mol mol −1 ) obtained by Takeda's (1998) in vitro (unfiltered seawater) experiment in the Southern Ocean.Using that criterion, the study area could be divided into eastern and western parts with the boundary around Station TT3 (Figure 8c).The results then suggested that the western part of the study area was iron-replete in March 2018 and 2020 and the surface water was iron-limited in the eastern part of the study area in March 2018.In support of these results, Makabe et al. (2020) indicated that the large size phytoplankton was dominated with respect to the total (10 μm pore size Chl-a/total Chl-a) around Station TT2 (Station S106 and 107 in Makabe et al., 2020) than at stations in the eastern side (Table S2 in Supporting Information S1).We also evaluated the ΔDIC/ΔP of 84.0 ± 7.9 mol mol −1 around Stations TT2-6.Arrigo et al. (1999) indicated the ΔDIC/ΔP of 94.3 ± 20.1 mol mol −1 in diatoms dominated waters, whereas the ΔDIC/ ΔP of 147 ± 26.7 mol mol −1 in Phaeocystis antarctica dominated waters in the Ross Sea.These results suggest that the diatoms were main phytoplankton in our study area (around Stations TT2-6).
An east-west gradient of ΔSi/ΔN and ΔSi/ΔP (Figures 8a and 8c) is plausible given the ocean currents in our study area.The mCDW transport path that causes basal melting of the ice shelf transports mCDW from the outer edge of the continental shelf to the depression on the continental shelf because of factors such as eddies and seafloor topography (Hirano et al., 2021;Hirano et al., 2023).The mCDW flows along the trough from the eastern side of the ice shelf (Hirano et al., 2023;Rintoul et al., 2016;Silvano et al., 2017), where it causes melting, and then flows further west onto the shelf after exiting the cavity of the TIS.Such circulation has also been observed in Lützow-Holm Bay (Hirano et al., 2020;Kiuchi et al., 2021).These results suggest that primary production is promoted by the iron supplied by glacial meltwater as it flows to the western flank of the TIS front.
The Amundsen Polynya and Pine Island Polynya, which are near the ice tongue/shelf off West Antarctica, are areas of high productivity (Arrigo et al., 2015).The supply of iron to the surface layer of these polynyas following an influx of basal meltwater from the adjacent Pine Island Ice Shelf accounts for their high production rates (Gerringa et al., 2012).Likewise, the flow of the source water for the mCDW beneath the Pine Island Ice Shelf causes basal melting (Jacobs et al., 1996(Jacobs et al., , 2011)).It has been shown that in Prydz Bay, East Antarctica, dissolved and particulate iron concentrations are high in the marine ice that forms on the bottom of the Amery Ice Shelf.Meltwater from that ice supplies iron to the surface layer that is thought to account for the high productivity of the Mackenzie Polynya (Herraiz-Borreguero et al., 2016).Furthermore, Kanna et al. (2020) have shown that the iron input from a marine-terminating glacier in Greenland, which is supplied by a subglacial discharge plume, has the potential to fuel phytoplankton blooms in a glacial fjord.
Our study presents a detailed description of the potential supply of nutrients and iron to stimulate photosynthesis by phytoplankton in coastal Antarctic waters.The stimulation results from buoyancy-driven upwelling (Lewis & Perkin, 1986) and mixing of nutrient-rich mCDW with iron-rich subglacial discharge, ice shelf basal meltwater, and sea ice meltwater (Figure 10).The supply of iron and nutrients plays an important role in the development  of extensive phytoplankton blooms in wind-driven coastal upwelling systems (Bruland et al., 2001;Fitzwater et al., 2003;Johnson et al., 1999).Likewise, near the ice front of the Antarctic coast, subglacial discharge, upwelling plumes, and sea ice melting supply abundant iron and macronutrients to the euphotic zone that sustain high productivity.Although we did not measure the iron concentrations in the water we sampled, a research project to examine the role of iron in Antarctic coastal waters is planned for future JARE expeditions.That study will facilitate understanding of iron dynamics in not only the TIS system but also other Antarctic coastal waters, because the rapid melting of ice shelves in recent years (e.g., Pritchard et al., 2012;Rignot et al., 2019) has provided large amounts of freshwater to the coastal areas and high productivity was observed (e.g., Arrigo et al., 2015).

Conclusions
Hydrographic observations were conducted from the continental shelf slopes of the Sabrina Coast to the ice shelf front in March 2018, December 2019, and March 2020.The mCDW, a water mass with relatively high temperature and high salinity, was transported from the outer edge of the continental shelf into a depression on the shelf and then flowed into the deep layers of the TIS along a trough.At the ice shelf front, biogeochemical components changed significantly as a result of mixing of mCDW with meltwater from sea ice and TIS basal meltwater.The surface layer was strongly influenced by these mixing and dilution effects and also by biological activity, especially photosynthesis by phytoplankton.During the observation period, the pCO 2 in seawater was reduced by mixing with glacial meltwater and sea ice meltwater at the ice shelf front and in the surrounding surface layer.The dilution effect of mixing with basal meltwater corresponded to a 2.4%-2.6%reduction of the pCO 2 in seawater over the observation period.The dilution effect of mixing with sea ice meltwater was small in December, but in March it exceeded the dilution effect of mixing with basal meltwater and corresponded to a 3.6%-4.8%reduction of the pCO 2 .In March, as the open water surface area expanded because of the melting of sea ice, pCO 2 decreased significantly as a result of photosynthesis by phytoplankton, and the pCO 2 in seawater was undersaturated with respect to the atmosphere.The ratios of nutrients taken up by phytoplankton suggest that iron contained in the basal meltwater of the ice shelf in addition to the sea ice meltwater helped to stimulate photosynthesis by phytoplankton, especially on the west side of the TIS front.

Figure 7 .
Figure 7. Relationship between NCP and (a) F simw and (b) F gmw for December 2019 and March 2018 and 2020, (c) detail of (a) showing relationship for March 2018 and 2020, and (d) detail of (b) showing relationship for March 2018 and 2020.Dashed lines representing regression lines are accompanied by annotations.

Figure 10 .
Figure 10.Schematic illustration of nutrient and iron transport in the study area.

Table 1
Mean and Standard Deviation for DIC, TA, δ 18 O, NO 3 , and Si(OH) 4 for Each Water Mass and Month