The net-phytoplankton samples were collected during the Antarctic krill survey conducted by the Chinese krill fishing vessel Fu Rong Hai around the SSI from 8 to 12 in March 2020. Samples were collected at 40 stations with 21 stations in the west of SSI and 19 stations in the BS ( Figure 1 ). Net-phytoplankton samples were collected with a standard net III (net area 0.1 m², mesh size 76 μm) by vertical hauling from 200 m depth or from the bottom to the surface when the depth was less than 200 m. The collected samples were preserved in 1 L bottles with 5% formaldehyde solution.

The krill density was estimated using acoustic data collected from a hull-mounted Simrad EK60 echosounder onboard F/V Fu Rong Hai along the transects survey as shown in Figure 1 . Krill backscatters were identified using the swarm-based method (Krafft et al., 2021). The acoustic backscatter at 120 kHz attributed to krill were then integrated as nautical area scattering coefficient (NASC, m²/n. mile²) from the surface exclusion layer (15 m) to the lower limit (250 m), and exported at an elementary distance sampling unit (EDSU) of 1 n.mile. In the subsequent correlation analysis, the NASC values of each station were averaged by 6 n. miles (3 n. miles before and after of each station).

Environmental data, including the SST and SIC, were obtained from the Copernicus Marine Data Store (https://resources.marine.copernicus.eu/). The spatial resolution of the original data was 0.05° × 0.05°, and temporal resolution was daily mean.

In the land-based laboratory, the phytoplankton samples were settled for more than 48 h, and the supernatant was aspirated off. The volume of the concentrated samples was about 100 ml, 0.5 mL of which was analyzed and counted under the Nikon Eclipse Ti2-U inverted microscope with 200× to 400× magnification. The taxa identification was based on species morphology referred to the books and literatures on phytoplankton classification and the website of www.algaebase.org.

The taxa abundance (A) was calculated as:

where n_i is the number of cells in species i, V1 is the analyzed volume 0.5 mL and V2 is the concentrated volume. V is the filtration volume when hauling the net which was calculated as 200 m or the depth of water minus two meters and then multiply by the net area.

Shannon-Weaver index (H’) was used to evaluate the species diversity of phytoplankton community. H’ is calculated as (Shannon and Weaver, 1949):

Dominance (Y) was calculated as:

where N is the total number of phytoplankton cells; S is the number of species and P_i is the ratio of the number of cells i to the total numbers, n_i is the number of cells in species i and f_i is the frequency of species i.

Phytoplankton community structure was examined by carrying out a multivariate analysis on abundance. Clustering was performed for each dataset based on the Bray- Curtis similarity matrix of log (x+1) transformed phytoplankton abundance and the average linkage group classification (Field et al., 1982; Shi et al., 2020) to distinguish the phytoplankton communities. In the process of analysis, stations in the west of SSI were defined as cluster 1, and the others were defined as cluster 2. ANOSIM (analysis of similarities) procedure was used to test the difference of phytoplankton community. To understand the circumstances of the dissimilar species that caused the difference between stratums, a similarity percentage analysis (SIMPER) was conducted. The species causing the difference between clusters were listed. The BIO-ENV analysis with Spearman rank correlation was carried out between the taxonomic and environmental data, to evaluate the best sets of environmental factors including SST, SIC and acoustic density of krill on phytoplankton communities.

Eighty-one taxa were identified in this study and the composition of phytoplankton was listed in Table 1 . Diatoms were the dominant group with 67 species. Others groups including dinoflagellates, silicoflagellates and cryptophytes were less abundant. The most abundant species were all diatoms in chains including Chaetoceros atlanticus, C. criophilus, C. dichaeta, Fragilariopsis kerguelensis, and Pseudo−nitzschia lineola.

The distributions of different phytoplankton groups abundance were exhibited in Figure 2 . Total phytoplankton abundance ranged from a minimum of 100 cells/m³ at T13-2 station to a maximum of 2.64×10⁷ cells/m³ at T5-1 station. Phytoplankton abundance in the west of the SSI was apparently higher than that in the BS. The distribution patterns of different groups (diatoms, dinoflagellates and most abundant species) were similar to the total abundance. Compared with pennate diatoms, centric diatoms were more abundant in the BS.

The H’ ranged from 0.0747 to 4.0176 ( Figure 3 ). In general, diversity indices were clearly higher in transects T1-T6 than those calculated in transects T7-T13. Interestingly, sampling stations with extremely low species diversity were almost dominated by one species namely Corethron pennatum. For example, T11-3 was the station with the lowest diversity index, where Corethron pennatum accounted for 99.33% of the total abundance. And the second lowest index was found at the station T13-1, where C. pennatum accounted for 98.90% of the total abundance.

Two clusters were classified at 25% of the similarity level shown in Figure 4 , which were significant different between cluster 1 and 2 (ANOSIM: R=0.513, p=0.001). In summary, cluster 1 mainly assembled the stations in the west of the SSI whereas cluster 2 in the BS.

To understand the circumstances of the dissimilar species that caused the difference between two clusters, we then conducted a similarity percentage analysis (SIMPER). The average dissimilarity was 76.65% between two clusters. In view of the significant difference, the species that contribute more than 3% were listed in Table 2 and their accumulated contribution added to 37.13%. Some diatom species contributed more for the dissimilarity between the clusters at sampling stations ( Table 2 ), mainly referred to centric diatoms (Rhizosolenia styliformis, Chaetoceros criophilus, C. atlanticus, R. antennata f. semispina, Proboscia alata, and Corethron pennatum).

The SST ranged from -0.35 to 3.88 °C. The SST in the BS was apparently lower than that in the west of the SSI ( Figure 5 ). The distribution of SIC was totally different from the SST spatial distribution. The SIC values in most stations were mostly 0, namely there was no ice cover. Only several stations including T4-3, T8-1, T9-1, T10-1 near the SSI still had a few bits of ice floes.

Acoustic density of krill, ranged from 0 to 1194.4 m²/n. mile², showed obvious spatial difference ( Figure 6 ). In general, acoustic density was higher in the BS than on the west side of SSI and varied greatly between stations. More than half of stations had NASC values less than 10 m²/n. mile², even the values in 13 stations were 0. And the high values occurred at transects T8, T9 and T10 in the BS (see Figure 1 ).

SST was the best environmental variable to explain the variance in the study area (P=0.01). Acoustic density was analyzed with the abundance of phytoplankton communities and abundant species respectively. It was found that the acoustic density was the best factor to explain the pennate diatoms (P=0.03).

The phytoplankton communities in the study area were mainly composed of the micro-diatoms in chains like Chaetoceros spp., Fragilariopsis spp., Pseudo−nitzschia spp. and Rhizosolenia spp. Diatom assemblages in Antarctic waters exist transitional characteristics. Their ecological types included eurythermic species, cold-water species and endemic species in Antarctic. The abundant species were mainly endemic species in Antarctic and cold-water species, which reflected the survival strategy of phytoplankton in the Southern Ocean. Firstly, due to larger cells are more resistant to sinking, they can stay in the euphotic layer (Sun et al., 2003). Secondly, large diatoms in chains with high ratio of superficial and volume of cells are conductive to the absorption of nutrients, especially to the absorption of limiting nutrients such as iron (Sun et al., 2003; Luan et al., 2012). Finally, some larger cells would be less susceptible to ingestion, while some cells have thick cell wall that make digestion difficult for zooplankton. In addition, we missed the pico- and some nanophytoplankton cells to some extent as we analyzed net-phytoplankton by the net sampling. These factors caused the net- phytoplankton communities to be dominated by larger cells or diatoms in chains (Sun et al., 2003).

In 1986-1987, dominant microphytoplankton were diatoms and the average cell density was 4.406×10⁶ cells/m³ in the BS and adjacent waters of EI (Zhu, 1993). Cefarelli et al. (2011) found diatoms were dominated in the mixed layer (1.06×10⁹–2.09×10⁹cells/m³) and small centric diatoms were also highly abundant in the northwestern Weddell Sea between 10 March and 1 April 2009. Luan et al. (2012) used the same method we used to collect and analyze the phytoplankton community during austral summer 2010. Phytoplankton abundance varied from 387 to 1.04×10⁷ cells/m³ which was similar to our results. Thalassiothrix antarctica, Gymnodinium sp., Chaetoceros sp., Pseudo-nitzschia lineola, Fragilariopsis kerguelensis, Chaetoceros criophilus, Corethron inerme and Fragilariopsis curta were the most abundant species. These species were also occurred in our results. Compared with previous studies above, there was little difference in composition of phytoplankton community. The dominant species or genus were similar to our study. Due to the difference in survey area, season and sampling methods, the phytoplankton abundance are varied. Water samples were collected for HPLC/CHEMTAX pigment and microscopic analysis around the tip of the AP during February/March 2008 and 2009. Phytoplankton assemblages were generally dominated by diatoms especially at coastal stations, while nanoflagellates replaced diatoms in open-ocean areas (Mendes et al., 2012). According to Mascioni et al. (2019), the highest phytoplankton abundance and biomass values were mainly represented by nanophytoflagellates, and the abundance of large bloom-forming diatoms was low in the relatively unexplored nearshore sites of the western AP during late summer of 2016 and during the spring-summer 2016-2017. As aforementioned, there are discrepancies among different results which might attribute to the conditions of sampling and methods of analyzing. So it seems to be necessary to have a long-term observation by same method at changeless location. Actually, rates of warming and sea ice loss are fastest in the southwest Atlantic sector with the impact of climate changes (Flores et al., 2012). Several studies have described a shift from large phytoplankton (diatoms) to smaller flagellated species (Moline et al., 2004; Monte-Hugo et al., 2009; Rozema et al., 2017b; Biggs et al., 2019). Therefore, long-term observation of the abundance of microphytoplankton is important to know more about size dynamics.

As shown in the results, SST was the major environmental variable to explain the variance. In both laboratory (Eppley, 1972; Berges et al., 2002) and field investigations (Montagnes and Franklin, 2001; Hernando et al., 2018), temperature has been found to play an essential role in the growth of organisms, which can promote enzyme activity and metabolic processes. Higher temperature leads to accelerate phytoplankton growth and increase the matter accumulation (Winder and Sommer, 2012). With the increase of temperature, the biomass increased. In addition, Lionard et al. (2012) also found that high temperature was more favorable for the growth of large centric diatoms in phytoplankton assemblages in temperate environments. Therefore, SST might have promoted the growth of many net- phytoplankton taxa, which was highlighted by great contribution of microdiatoms.

The study area is hydrologically complex, with multiple water masses flowing from the Weddell Sea and the Bellingshausen Sea (Sangrà et al., 2011; Mendes et al., 2012). Within the surface layer of the BS, there are two major transitional waters being detected: Transitional Weddell Water (TWW) dominated by relatively cold and salty water mass flows north and west along the AP, and Transitional Bellingshausen Water (TBW) dominated by a relatively warm and fresh water mass, flows east (Gonçalves-Araujo et al., 2015). Thermohaline difference between TWW and TBW could be reflected in the phytoplankton communities. There is a well-mixed water column in the TWW where nanoplanktonic flagellates was dominant and exhibited lower chl a. On the contrary, microplanktonic diatoms were dominant and contributed higher chl a in the TBW because of the strong pycnocline and shallow upper mixed layers (Gonçalves-Araujo et al., 2015) within. In the west of the SSI, the high nutrients brought by Circumpolar Deep Water (CDW) and the deep water of subtropical Pacific Ocean accelerated the bloom of phytoplankton (Luan et al., 2012).

Krill is a potential resource and their feeding behavior is complex, not only filtering phytoplankton and protozoa but also preying on zooplankton (Cleary et al., 2018). In addition, krill feed on algae and detritus from sea ice and seabed (Price et al., 1988; Stretch et al., 1988; Clarke and Tyler, 2008; Schmidt et al., 2014). During the period of phytoplankton bloom, krill mainly feed on diatoms. While in the scarcity phase of phytoplankton, they also choose flagellates, copepods and detritus to sustain life (Schmidt and Atkinson, 2016). Analysis of stomach contents (Meyer and El-Sayed, 1983) and studies of the comparison of krill and phytoplankton distribution (Schmidt and Atkinson, 2016) suggested that krill feeding selective.

Early studies about krill gut content established the suitable phytoplankton species for feeding (Schmidt and Atkinson, 2016). In the South Georgia, microphytoplankton was the predominant component of gut contents. Solitary and colonial cells of Nitzschia spp., Thalassiosira spp. and Fragilariopsis kerguelensis were the most abundant (Pakhomov et al., 1997). It was also found that Thalassiosira spp. are preferred by krill feeding and some small pennate diatoms such as Navicula spp. and Nitzschia spp. are barely fed (Opaliński et al., 1997). In the stomach contents study of Cleary et al. (2018), krill have a diatom-based diet, while the occasional presence of copepod suggests carnivorous supplemented diet. Compared with cryptophytes or prymnesiophytes, diatoms are recognized as high quality food for zooplankton (Ross et al., 2000). Indeed, diatom bloom and gonad development of krill occurs simultaneously in spring (Cuzin-Roudy and Labat, 1992; Schmidt et al., 2012) and the accumulation of polyunsaturated fatty acids of krill by feeding on diatoms was more effective than that by feeding on copepods (Schmidt et al., 2014). In addition, krill were more likely to feed on chain-forming diatoms than solitary phytoplankton species (Stuart, 1989; Haberman et al., 2003a). This result should be mainly concerned with the cell size. For cells with large size greater than 70 μm, krill were incapable of ingestion, while the size was favored by krill at 20-40 μm (Meyer and El-Sayed, 1983; Drits and Pasternak, 1993; Maciewska and Opalinski, 1993 and Opaliński et al., 1997).

Compared with phytoplankton distribution, there was a negative correlation between krill abundance and primary production during the survey in the South Georgia (Pakhomov et al., 1997). And the abundance of phytoplankton community dominated by diatoms rapidly decreased due to the feeding of the krill swarm only in a few hours according to the observation of the scientific cruise in the Scotia-Weddell Sea (Smetacek and Veth, 1989). However, striking differences were observed between the stomach contents of krill collected in fjords and in adjacent open waters which could not be explained by differences in the surface water phytoplankton (Cleary et al., 2018). These findings are inconclusive. The distribution of krill is concentrated along the AP, and their spatial and temporal distribution is highly variable due to the ability of krill swarms migration (Ross et al., 1996; Moline et al., 2004). Large krill swarms may contain up to 10000–30000 individuals m^-3 (Hamner et al., 1983), and can rapidly clear phytoplankton up to a few litres per hour (Quetin et al., 1994). In fact, this phenomenon occurred mostly in local area. Spreading over larger areas, krill are difficult in grazing down phytoplankton (Atkinson et al., 2014). In the meanwhile, copepods were consistently part of krill diet (Schmidt et al., 2014). The distribution of zooplankton in waters also need to be considered, which may influence krill diet and increase the pressure on phytoplankton. In conclusion, it is hard to find the corresponding relationship between phytoplankton and krill density in the voyage survey even within the same region and season. This could explain why there was no correlation between phytoplankton abundance and krill density in our study.

Our research further analyzed the correlation between the distribution of pennate diatoms and the acoustic density of krill by BIO-ENV analysis. In the BS, krill density was high while the abundance of phytoplankton dominated by Corethron pennatum was low. In contrast, phytoplankton including pennate diatoms were abundant while the krill density was low in the west of SSI. Aforementioned findings were speculated that the phytoplankton assemblages were related to the krill grazing pressure. Biological processes, such as zooplankton grazing, superimposed to physical and chemical changes, can modify the abundance and dominance of different taxonomic assemblages (Cefarelli et al., 2011). Diatoms such as Thalassiosira spp., Fragilariopsis spp. and Chaetoceros spp. are feeding targets of krill, the low abundance of them and the dominance of Corethron pennatum in the BS may be the results of krill selective grazing, which seems to be the signal after krill grazing. Corethron pennatum may be the species krill refuses to eat. The structure of phytoplankton community is the result of consumption of higher trophic level including krill selective grazing. These conjectures may provide a new research direction for krill selective grazing and more evidence need to be explored.