The study area covers the continental shelf to the abyssal plain in East Antarctica south of the PF (water depth ranging from 219 to 4642 m) (Fig. 1; Table 1). Regarding the surface sediment sampling in this study, the study area is divided into three subareas, i.e., the Cosmonauts Sea, the Cooperation Sea, and the Mawson Sea (Fig. 1b).

The Cosmonauts Sea, situated between 30 and 50∘ E in the Southern Ocean (Fig. 1b), encompasses Lützow–Holm Bay (LHB). This bay is characterized by a deep glacial trough in the center that is ∼ 15 km wide and more than 600 m deep and deepens southward to ∼ 1200–1400 m (Hirano et al., 2020). To the east of the Cosmonauts Sea lies the Cooperation Sea, which includes Prydz Bay (PB). Prydz Bay is a vast embayment covering an area of approximately 80 000 km2. Within Prydz Bay, the continental shelf slopes gradually toward the land, with water depths ranging from approximately 200 to 400 m at the continental shelf break, increasing to about 600 to 1000 m at the inner shelf (Fretwell et al., 2013). The Mawson Sea, situated to the east of the Cooperation Sea, encompasses the Sabrina Coast (SC). The continental shelf in the Sabrina Coast region is typically over 120 km wide (Close et al., 2007). The topography in the SC is similar to that of PB and LHB, with the seabed deepening landwards from the outer shelf at depths ranging from 200 to 500 m to inner shelf depths exceeding 1400 m in this study area of SC (116–122∘ E).

The Southern Ocean south of the PF is mainly occupied by three water masses, i.e., Antarctic Surface Water (AASW), Circumpolar Deep Water (CDW), and AABW (Fig. 2). The AASW is a cold and low-salinity mass that forms at the surface south of the PF (Fig. 2). To the north of the southern boundary (SB), the surface water temperature rises to approximately 0 ∘C due to summer solar radiation, and the subsurface cold layer (often referred to as winter water) is relatively thin and not extremely cold. South of the SB, the subsurface cold layer, which formed in winter at nearly -2 ∘C is much thicker (Fig. 2). CDW is a warm (1.5–3 ∘C) and salty (salinity ∼ 34.7) water mass located below a water depth of 2000 m north of 40∘ S and is mainly derived from North Atlantic Deep Water. South of 40∘ S, CDW shoals due to wind-driven upwelling (Greene et al., 2017; Marshall and Speer, 2012). As CDW moves poleward, it undergoes mixing with AASW and continental shelf waters, leading to a modification of its core properties (Orsi and Wiederwohl, 2009), resulting in what is known as modified Circumpolar Deep Water (mCDW). AABW is a water mass generally occurring below 4000 m, with temperatures ranging from -0.8 to 0 ∘C and salinities ranging from 34.6 to 34.7. AABW, which originates from mCDW and dense shelf water characterized by temperatures as low as -1.8 ∘C and high salinity (>34.6), is formed in various Antarctic coastal polynyas, including those in the Adélie Coast (Mertz Polynya) and Prydz Bay (Cape Darnley polynya).

The water structure on the continental shelf region is affected by different processes, such as mCDW intrusion, sea ice formation in polynyas, and freshwater discharge from ice shelves. In this study area, polynyas (Dalton and Cape Darnley polynyas; Fig. 1) are present in SC and PB (Tamura et al., 2008). In LHB, mCDW (>0 ∘C) intrudes the continental shelf along the trough (Hirano et al., 2020). In SC, mCDW also intrudes the continental shelf along the depression from east to west, with the near-surface cold layer largely affected by Dalton Polynya (Silvano et al., 2018). Conversely, the intrusion of mCDW is less pronounced in the Prydz Bay (PB) area, primarily due to significant winter heat loss in this region (Herraiz-Borreguero et al., 2015; Guo et al., 2019).

A total of 36 surface sediment samples (0–2 cm depth) from the area of 59–68∘ S and 38–145∘ E (water depths ranging from 219 to 4642 m) were used in this study (Fig. 1; Table 1). There were 17 samples collected from the continental shelf, using a grab sampler during the 61st Japanese Antarctic Research Expedition (JARE-61) cruise of icebreaker (I/B) Shirase in 2019–2020. Then, 19 samples were collected from the continental shelf, continental slope, and abyssal plain during cruises conducted by the Japan National Oil Corporation as part of the Antarctic Geological and Geophysical Research Project (1989–2000) aboard R/V Hakurei Maru, using piston or gravity corers. Their core top samples were used as surface sediments.

Freeze-dried subsamples were weighed (approximately 1–2 g), treated with 15 % H2O2 to remove the organic matter, and then treated with an HCl solution to remove the calcium carbonate. The samples were then wet-sieved (45 µm mesh size), after which two types of permanent slides were made from the residue to quantify the abundance (Q-slide) and for faunal analysis (F-slide) using the method described by Itaki et al. (2018). Briefly, the Q-slides were prepared by transferring all residue to a 200 mL beaker containing 100 mL distilled water. The solution was well mixed, and a 0.5 mL sample was taken from the suspension using a micropipette and dropped onto a glass slide. The sample was then dried and mounted with Norland Optical Adhesive. The F-slides were made from the remaining residues in the beaker and then mounted.

Radiolarian skeletons were observed under an optical microscope at 100×, 200×, and 400× magnifications. The radiolarian concentration, which represents the total number of radiolarians present in 1–2 g of dry sediment, was estimated using the following equation: RC (specimens g-1) is equal to the total number of specimens in the Q-slide × 200 / sample weight (g). The relative abundance (percent of total assemblage) of radiolarian taxa was estimated by counting and identifying more than 300 specimens on the F-slide; however, as many individuals as could have been identified were counted in the F-slide when radiolarian abundances were scarce.

A plankton tow with a mesh size of 60 µm and a collecting and closing system (with a frame diameter of 1 m) was vertically towed at eight stations in the Mawson Sea of the Southern Ocean (Fig. 1) during the 16th Kaiyodai Antarctic Research Expedition aboard training vessel (TV) Umitaka Maru in January 2013. At all stations, three–four vertical tows were carried out at a maximum speed of 0.5 m s-1 through intervals of 0–50, 50–100, 100–200, and 200–500 m. A calibrated flow meter (Rigosya, Japan) was employed to determine the volume of filtered water.

The plankton samples were preserved in borate-buffered formalin–seawater solution (final concentration 5 %) and were stained with rose bengal to distinguish living specimens from dead ones. Observation slides were prepared using the following process. First, larger zooplankton were excluded from the original samples using a 1 mm mesh. Second, specimens smaller than 1 mm were split into aliquots between 1/16th and 1/128th of the original quantity. Third, specimens were wet-sieved using a 45 µm mesh. Finally, the sample was dried and mounted with Norland Optical Adhesive. Polycystine radiolarian species were identified and counted using an optical microscope at magnifications of 100× or 200×. All plankton specimens on a slide were identified and counted, and their individual counts were converted to abundance (number of specimens per m3).

During the JARE-61 cruise, vertical hydrographic data, including temperature, salinity, and dissolved oxygen (DO), were collected using a conductivity–temperature–depth profiler (CTD) (Sea & Sun Technology CTD90M), which was equipped with the grab sampler. Salinity and DO values were calibrated using regression lines derived from salinity and DO data obtained from both Niskin bottle bottom-water samples and a sensor installed in the grab sampler.

The hydrographic dataset (including temperature, salinity, and DO) from the World Ocean Atlas 2018 was also utilized to depict the distribution of water masses across a broad region. Although these hydrographic datasets provided annual, winter, spring, summer, and autumn data, the austral summer datasets were used to compare with the austral summer data observed by the JARE-61 cruise. In addition, the austral summer datasets were used because sediment trap studies in the Southern Ocean have demonstrated that radiolarian flux in austral summer contributes most of the annual flux (Abelmann and Gersonde, 1991; Abelmann, 1992).

A Q-mode factor analysis (Imbrie, 1971; Imbrie et al., 1973) was conducted on surface sediment samples to illustrate the relationships between oceanographic conditions in each factor analysis interval and the ecology of the radiolarians found in those intervals. For factor analysis, the size of the dataset of counted radiolarians was reduced to a total of 24 taxa. In accordance with the procedure outlined by Abelmann and Gowing (1997), species with abundances below 1 % and those occurring in only one or two samples with abundances below 2 % were removed from the dataset. The groups of juvenile nassellarians and spumellarians, which could not be identified at the species level, were not included in the statistical data. Moreover, the dataset was simplified by consolidating certain taxa to minimize counting uncertainties and enhance the reproducibility of counts for taxa with intricate taxonomy (e.g., Acanthosphaera spp.). Among the 36 samples, 17 had fewer than 100 radiolarians on the F-slide and were excluded from subsequent analyses. In the remaining 19 samples, between 100 and 300 specimens were identified at the species level. The PAST statistical software package (Paleontological Statistics v. 4.4) (Hammer et al., 2001) was used for the Q-mode factor analysis. Prior to the analysis, the data were normalized, following the method described by Itaki et al. (2008, 2010).

AASW with low temperature (-2 to 1 ∘C) and low salinity (∼ 34.5) was observed in the surface layer (∼ 400 m) at all continental shelf sites south of SB (Fig. 3) (hereafter nearshore AASW). In certain locations on the continental shelf of PB and SC, relatively warmer water (around 1 ∘C) was present in the surface layer (upper 100 m) as a result of summer solar heating. This warm surface water was particularly prominent in the polynya areas, such as St. X23-KG and St. CD4-KG (refer to the Supplement). In addition, the surface layer (upper 100 m) had low salinity (33.2–34.2) due to meltwater from sea ice and glacial ice at most sites. In the abyssal plain area north of SB, the temperature in AASW, which occupies the surface layer (upper 100 m), increased from south to north (Fig. 2) (hereinafter offshore AASW).

CDW with high salinity (∼ 34.7) and warm temperature (1–2 ∘C) was observed below AASW in the abyssal plain sites north of SB (water depth of 200–3000 m). This CDW shoaled toward the south due to wind-driven upwelling (Fig. 2). Warm and high salinity water below 400 m was observed at continental shelf sites in SC and LHB, reflecting mCDW intrusion into the continental shelf. In particular, high temperature (>0 ∘C) was observed at St. 83-KG and St. X23-KG (Fig. 3; see the Supplement), indicating a significant intrusion of relatively warm mCDW into these sites. At St. LH1a-KG, relatively warm (>0 ∘C) water was recognized in the bottom layer, indicating an intrusion of mCDW (Fig. 3; see the Supplement).

Conversely, in the PB area, the bottom layer is marked by high salinity (>34.5) and low temperature (< -1.5 ∘C) (Fig. 3), signifying the presence of dense shelf water (DSW) and the subsequent production of AABW in this vicinity. AABW produced in this area sinks along the continental slope and spreads northward to fill most of the abyssal ocean deeper than 4000 m (Fig. 2).

As the result of factor analysis for surface sediments, F1 shows its highest and lowest loadings at continental shelf and abyssal plain samples, respectively (Fig. 5a). Lithelius nautiloides, Antarctissa denticulata, and Antarctissa strelkovi have relatively high scores for this factor (Table 3). Lithelius nautiloides is abundant in surface sediments south of the PF (Hays, 1965; Petrushevskaya, 1971). Antarctissa denticulata is present in surface sediments south of the PF and is more frequent south of SB (Petrushevskaya, 1971). Abelmann et al. (1999) showed that Antarctissa strelkovi has high abundances at 53–58∘ S and is limited south of the PF in the Weddell Sea (40∘ W–10∘ E). In contrast, Petrushevskaya (1971) showed that the abundance of Antarctissa strelkovi in surface sediment samples gradually increases from north to south (data from longitude 10–180∘ E) (Petrushevskaya, 1971). Dow (1978) concluded that Antarctissa strelkovi is abundant south of the PF, based on factor analysis of the surface sediment (80–110∘ E). Although the reasons for the regional differences in the distribution of Antarctissa strelkovi are unclear, this species is more abundant in the higher latitudes (continental shelf sites) of our study area (30–150∘ E). In F1, it was indicated that these three species are related to water conditions south of the SB, which is in agreement with the other abovementioned studies.

In plankton tow data, Antarctissa denticulata and Antarctissa strelkovi were in high abundance in surface-to-subsurface water (50–100, 100–200, and 200–500 m), showing that these two species live in the surface water south of the PF (Fig. 7; see the Supplement). Previous studies from plankton tow also show that three of the species having high scores in F1 in our data from Q-mode factor analysis of surface sediments (Antarctissa denticulata, Antarctissa strelkovi, and Lithelius nautiloides) dominantly occur in surface water (∼ 200 m) south of the PF (Morley and Stepien, 1985). The cold water cooled by meltwater from ice sheets and sea ice is located in the nearshore AASW (Fig. 2), indicating that Antarctissa denticulata, Antarctissa strelkovi, and Lithelius nautiloides inhabit the nearshore AASW, which is even colder than offshore AASW.

The result of Q-mode factor analysis from surface sediments shows high and low loadings in the abyssal plain and continental shelf samples, respectively (Figs. 5b and 6b; Table 2), suggesting that the taxa showing high scores in F2 live in the open ocean north of SB. The AASW, CDW, and AABW are located offshore (Fig. 2), but because waters rich in radiolarian are generally limited to ocean depths shallower than 1000 m (Boltovskoy, 2017), the AABW, which is located in lower layers, is thought to have few radiolarian species. Thus, the 13 taxa displaying relatively high scores in F2 (>1.0) could be living in the offshore AASW or CDW.

F3 shows high loadings at continental shelf samples in SC and LHB and low loadings at the continental shelf in PB and abyssal plain samples (Figs. 5c and 6c). For this factor, Larcopyle buetschlii, Lithomelissa sp. A, Actinomma leptoderma, and Spongodiscus resurgens show high scores (1.04, 2.85, 2.06, and 2.75). Our plankton tow data suggest that these taxa were found in shallow and deep waters and were abundant in 200–500 m (Fig. 7; see the Supplement). As mentioned above, Larcopyle buetschlii (adult) dwells in deep water, and it was identified as a species that dwells in the mesopelagic zone. Actinomma leptoderma has been reported in high-latitude oceans, such as the Arctic Ocean and Greenland Sea (Hülsemann, 1963; Tibbs, 1967; Petrushevskaya, 1969), and observed in the relatively warm mesopelagic zone (Jørgensen, 1905). It has been reported that Lithomelissa sp. A dwells in the mesopelagic zone in the Southern Ocean (∼ 55∘ S) from factor analysis of plankton tow data (Abelmann and Gowing, 1997). These insights suggest that these taxa inhabit relatively deep waters.

High loadings in F3 were found at St. 83-KG, St. X23-KG, and LH1a-KG (0.87, 0.52, and 0.56, respectively). The vertical profiles of temperature and salinity suggest the intrusion of mCDW onto these sites (Fig. 3; see the Supplement). Therefore, if these taxa showing high scores in F3 also live to the south of SB in the deep ocean, then they can be identified as mCDW taxa. However, plankton tow data at the continental shelf are required to further understand the habitat depth of this species south of 65∘ S.

F4 shows the highest loadings in coastal samples (St.14B-KG, St. 15-KG, St. 12B-KG, St. 17-KG, and CD4-KG). Moreover, Plectacantha oikiskos and Lithomelissa sp. A. Plectacantha oikiskos, showing high scores in F4, have been observed in coastal areas at high latitudes, such as fjords of the Norwegian Sea, Southern Ocean (e.g., Jørgensen, 1905; Nishimura et al., 1997), and are reported to be common in the plankton at the ice edge in the Greenland Sea (Swanberg and Eide, 1992). Lithomelissa sp. A was abundant in the coastal area in the Southern Ocean (Petrushevskaya, 1967).

The sites with high loadings in F4 (St. 14B-KG, St. 15-KG, St. 12B-KG, St. 17-KG, and CD4-KG) were located near polynya and the edge of glacier areas, suggesting that these taxa showing high scores in F4 (Plectacantha oikiskos and Lithomelissa sp. A) may be related to the ice edge. Lithomelissa sp. A exhibited high loadings in both F3 and F4 and may be influenced by the ice edge environment, as well as the water masses.