The coastal polynya in the Amundsen Sea generally forms in November and closes again by March (Arrigo et al., 2012). The peak phytoplankton bloom in the ASP generally occurs in January and then declines from February to March (Arrigo and van Dijken, 2003; Yager et al., 2016). This study was conducted during the phytoplankton bloom period on board the IBRV Araon from January 5 to 10, 2014 (ANA04) and from January 17 to 24, 2016 (ANA06). During the study period, the ASP was closed to the open sea in ANA04 but not in ANA06, resulting in an expanded ice-free area in ANA06 ( Figure 1 ). Water samples for physico-chemical and microbiological analysis were collected at five and eight stations during ANA04 and ANA06, respectively ( Table 1 and Figure 1 ).

Water temperature, salinity, density, and photosynthetically active radiation (PAR) were measured using a conductivity-temperature-depth (CTD) probe (SBE 911 Plus, Seabird Electronics, USA). Euphotic depth was defined as the depth of 1% penetration of the surface on PAR. Mixed layer depth (MLD) was defined as the depth at which the density was 0.05 kg m^-3 higher than the 10-m value (Venables and Moore, 2010). Water samples were collected from 7 – 8 depths in the upper 100 m using Niskin bottles attached to a CTD rosette sampler. The sample bottles were first washed with 10% HCl and rinsed three times with Milli-Q water.

Concentrations of inorganic nutrients ( NO 2 − + NO 3 −   [ NO X ] , NH 4 + , PO 4 3 − and Si ( OH ) 4 ) were measured onboard using a 4-channel continuous Auto-Analyzer (QuAAtro, Seal Analytical). The precisions for the NO_X, NH 4 + , PO 4 3 − , and Si(OH)₄ measurements were ± 0.14, ± 0.02, ± 0.18, and ± 0.29 μmol L^-1, respectively. Samples of chromophoric dissolved organic matter (CDOM) were filtered through a syringe filter (Adventec, 0.2-μm pore size) that had been pre-washed ultra-pure Milli-Q water, and then stored in 150-mL amber bottles in the dark at 2 – 4°C in a refrigerator (Stedmon and Markager, 2001). The absorbance of the samples was measured on board using a double-beam Shimadzu UV-1800 spectrophotometer with a 10-cm quartz cell in the spectral range of 350 – 900 nm. A quartz cell filled with pre-filtered Milli-Q water was used as the reference for all samples. The absorbance at 350 nm relative to distilled water was measured, and then the absorption coefficient at 350 nm (CDOM m^-1) was determined using the following equation (Kowalczuk et al., 2005):

where a(λ) is the absorption coefficient (m^-1), A(λ) is the absorption coefficient at the reference wavelength λ (m^-1), 2.303 is the conversion factor from log10 to log units, and l is the length of the optical path (m).

Samples for DOC and dissolved organic nitrogen (DON) analysis were filtered through pre-combusted (at 550°C for 6 hours) GF/F filters (0.7-μm pore size), followed by hermetic sealing in pre-combusted (at 550°C for 6 hours) 20-mL glass ampoules and were preserved at -24°C until analysis in the laboratory. The concentrations of DOC and total dissolved nitrogen were determined via high temperature combustion on a Shimadzu TOC-L analyzer (Shimadzu Co.) fitted with a chemiluminescence (CLS) detector that was incorporated into the total nitrogen microanalyzer. For the accuracy of the DOC concentration, Milli-Q water blanks and consensus reference material (CRM, 42–45 M, University of Miami) were measured every sixth analysis. The concentration of DON was obtained as the difference between total dissolved nitrogen and dissolved inorganic nitrogen (i.e., NO 2 − , NO 3 − , and NH 4 + ) concentrations.

Chl-a was extracted with 90% acetone for 24 h in the dark after filtration through a GF/F filter (0.7-μm pore size) (Parsons et al., 1984). The Chl-a concentration was determined using a fluorometer (Turner Designs Trilogy).

To determine phytoplankton abundance, water samples were preserved with glutaraldehyde (final concentration, 1%) and stored at 4°C before filtration and staining. The water samples (50 - 150 mL) were filtered through Nuclepore filters (0.8-μm pore size, black, 25-mm diameter) until 5 mL remained in the filtration tower. Then, concentrated DAPI (4’,6-diamidino-2-phenyl-indole) solution (50 μg mL^-1 final concentration) was added, and briefly incubated (5 s) before filtration (Taylor et al., 2011). The filters were mounted on slide glass with immersion oil. Phytoplankton cells were counted using an epifluorescence microscope (Olympus BX 51). Carbon (C) biomass was estimated from the cell biovolume using modified Eppley equations (Smayda, 1978; Hillebrand et al., 1999) as follows: for diatoms, log10 C (pg) = 0.76 log10 [cell volume (μm³)] − 0.352; for other phytoplankton, log10 C (pg) = 0.94 log10 [cell volume (μm³)] − 0.60. The conversion factor used to transform cell numbers of solitary P. antarctica into carbon biomass was 3.33 pg C cell⁻¹ (Mathot et al., 2000).

To elucidate the major phytoplankton groups (Dunbar et al., 2003), disappearance ratios of NO_X (ΔN), PO 4 3 − (ΔP) and Si(OH)₄ (ΔSi) were calculated as the difference between the concentration within the upper 100 m and the concentration at 500-m depth as a proxy for the pre-bloom concentration within the upper 100 m (Lee et al., 2016a). A ΔN:ΔP ratio of 19.2 was used as a proxy for the P. antarctica bloom, whereas ΔN:ΔP of 9.69 was used for the diatom bloom (Arrigo et al., 1999). Similarly, ΔSi:ΔN of < 0.9 was used for the P. antarctica bloom, whereas ΔSi:ΔN of > 2.15 was used for the diatom bloom (Dunbar et al., 2003).

Samples for bacterial abundance (BA) were preserved with glutaraldehyde at a final concentration of 1% and stored at -20 °C (Hyun and Yang, 2003). Samples were stained with DAPI solution (Porter and Feig, 1980), filtered through a Nuclepore filter (0.2-µm pore size, black), and mounted on slide glass with immersion oil (Cargille type A). Bacterial cells were counted using an epifluorescence microscope (Eclipse 80i, Nikon, Tokyo, Japan) equipped with a mercury lamp (HB-10101 AF), an ultraviolet (UV) excitation filter, and a BA 420 barrier filter.

Heterotrophic bacterial production (BP) was determined from bacterial incorporation of ³H-thymidine (³H-TdR) (Fuhrman and Azam, 1980; Fuhrman and Azam, 1982). Duplicate 20-mL water samples were incubated in disposable plastic tubes for 30 min at in situ water temperatures under dark conditions with ³H-TdR (MT-6034; final concentration, 10 nM, Moravek Biochemicals, Inc., Brea, CA, USA). Incubation was stopped by adding cold trichloroacetic acid (TCA, final concentration 5%), and cold TCA-insoluble materials were precipitated for 15 min. Samples were collected by vacuum filtration on 0.2-μm cellulose nitrate membrane filters and rinsed three times with 80% cold ethanol. The filters were placed in scintillation vials. In the lab, after scintillation cocktail (Lumagel Safe; Lumac-LSC, Groningen, The Netherlands) was added to the vials, the activity of the cold TCA-insoluble macromolecules was determined using a liquid scintillation counter (Tri-Carb 2910TR; PerkinElmer, Waltham, MA, USA). Samples treated with 5% TCA solution before the addition of ³H-TdR were used as killed controls. A conversion factor of 8.6×10¹⁷ cells mol^-1 (Ducklow et al., 1999) was used to convert ³H-TdR measurements into bacterial cell production estimates. Bacterial carbon biomass was calculated using a conversion factor of 10 fg C cell^-1 (Fukuda et al., 1998).

Total microbial community respiration rates were calculated from the consumption of dissolved oxygen (DO) over time in ca. 300-mL water samples maintained in BOD bottles (Pomeroy et al., 1994). DO concentration was measured using the Winkler titration method for the ANA04 cruise, while both Winkler titration and the O₂-optode method were applied for samples collected on the ANA06 cruise. In the Winkler titration method, after water was subsampled into six BOD bottles at each depth, duplicate water samples were immediately fixed with Winkler reagent to estimate initial DO concentration. The remaining bottles were incubated at in situ water temperatures under dark conditions. After 12 and 24 h, duplicate water samples were fixed with Winkler reagent, and DO concentrations were determined at a wavelength of 466 nm on a spectrophotometer (Shimadzu UV-1800) (Labasque et al., 2004). In the O₂-optode method (Wikner et al., 2013), DO concentrations were measured using a fiber-optical oxygen meter (FireStingO2, PyroScience GmbH, Germany). The BOD bottles for O₂-optode measurements were incubated for 8 – 14 h in the dark at in situ temperature using a water bath. Prior to measurement, two-point calibration (0 and 100%) was performed. The 0% oxygen saturation was calibrated by adding sodium dithionite (Na₂S₂O₄) to water and the 100% oxygen saturation was calibrated using air-saturated water. The respiration rates measured using the O₂-optode method showed no significant difference from those measured by Winkler titration (p = 0.209, Supplementary Table 1 ). BR was estimated using a bacteria respiration proportion of 0.45 from total microbial community respiration (Robinson, 2008). Bacterial community growth rates were calculated by dividing bacterial production by bacterial carbon biomass (Kirchman et al., 2009a). Bacterial carbon demand (BCD) and bacterial growth efficiency (BGE) were calculated from the following equation: BCD = BP + BR; BGE = BP/BCD (del Giorgio and Cole, 1998; del Giorgio and Cole, 2000).

Statistical analyses were conducted in R (R version 4.0.2, R core Team, using R Studio v.1.3.1073). Data normality was tested using Shapiro-Wilk test. Mann-Whitney U-test was used to compare mean values for chemical and biological parameters between the two periods (ANA04 and ANA06). Spearman’s rho correlation was conducted to examine the relationship between bacterial production and environmental parameters. p < 0.05 was considered significant.

Surface water temperature, salinity, and density exhibited different ranges between the ANA04 and ANA06 periods ( Table 1 ). In ANA04, water temperature, salinity, and density ranged from -0.39 to -0.08°C (average, -0.24 ± 0.13°C), from 33.63 to 33.94 psu (average, 33.81 ± 0.12 psu), and from 27.01 to 27.27 kg m^-3 (average, 27.16 ± 0.10 kg m^-3), respectively. Average water temperature (0.03 ± 0.29 °C) in ANA06 was slightly higher than that of ANA04, but average salinity (33.85 ± 0.25 psu) and density (27.18 ± 0.19 kg m^-3) did not show significant difference between ANA04 and ANA06 (p > 0.05). Average mixed layer depth was not significantly different between ANA04 and ANA06 (p = 0.69, Table 1 ). Euphotic depth in ANA06 (18 m on average) was slightly deeper than that observed in ANA04 (12 m on average) (p = 0.005, Table 1 ), but the median value of surface PAR during the sampling period was six times lower in ANA06 (52 μmol photons m^-2 s^-1) than in ANA04 (290 μmol photons m^-2 s^-1, Supplementary Figure 1 ).

Chemical parameters within the upper 100 m varied in ANA04 and ANA06 ( Figures 2 , 3 ). DIN and DIP concentrations increased with depth ( Figures 2A, B ) and were lower in range in ANA04 (average, 18.15 ± 8.15 μM and 1.48 ± 0.40 μM, respectively) than in ANA06 (average, 24.12 ± 4.95 μM and 1.72 ± 0.40 μM, respectively) ( Figures 3A, B ). Concentrations of CDOM, DOC, and DON were greater in ANA04 than in ANA06 ( Figures 2C, D, E ). CDOM concentrations measured within the upper 50 m were higher in ANA04 (average, 0.53 ± 0.19 m^-1) than in ANA06 (average, 0.23 ± 0.12 m^-1) (p < 0.001, Figure 3C ). DOC concentrations were generally homogenously distributed and DOC measured during ANA04 (average, 50.24 ± 8.09 μM) was higher than that of ANA06 (average, 45.26 ± 3.00 μM) (p = 0.021, Figure 3D ). DON concentrations varied, but were slightly higher in ANA04 (average, 5.65 ± 2.56 μM) than in ANA06 (average, 2.80 ± 1.90 μM) (p = 0.003, Figure 3E ). The range of DOC and DON concentrations for both periods were slightly lower than those reported during the early-mid phytoplankton bloom in the Amundsen Sea (Yager et al., 2016).

Concentrations of Chl-a were distributed differently in ANA04 and ANA06 ( Figure 2F ). Chl-a concentrations measured during ANA04 (average, 5.55 ± 3.78 μg L^-1) with lower DIN and DIP concentrations ( Figures 2A, B ) were three times higher than that of ANA06 (average, 1.90 ± 1.24 μg L^-1) ( Figure 3F ).

Major phytoplankton groups in the ASP consist of P. antarctica, diatoms, and D. speculum ( Figure 4 ). P. antarctica dominated carbon biomass in ANA04, but diatoms and D. speculum co-dominated in ANA06 ( Figure 4A ). The relative contribution of P. antarctica was higher in ANA04 (78 ± 14%), but it decreased to 3 ± 2% in ANA06 ( Figure 4B ). In contrast, the relative contribution of diatoms was lower in ANA04 (11 ± 12%), but increased to 45 ± 25% in ANA06 ( Figure 4B ). In ANA06, the average carbon biomass of diatoms (143 ± 94 μg C L^-1) was 12-fold higher than that of P. antarctica (12 ± 7 μg C L^-1) ( Figure 4A ) because of their larger size and higher carbon content (Lee et al., 2016b). In ANA06, D. speculum also exhibited high relative carbon biomass, comprising 48 ± 25% of the total carbon biomass ( Figure 4B ).

BA ranged from 0.25 to 7.50 × 10⁸ cells L^-1 during ANA04 and ANA06 ( Figure 2G ). In ANA04, maximum BA was observed at 15-m depth and decreased rapidly with increasing depth, while BA in ANA06 was distributed relatively consistently with depth. In the upper 100 m, BA in ANA04 (1.95 ± 1.66 x 10⁸ cells L^-1) was slightly higher than that in ANA06 (1.49 ± 0.96 x 10⁸ cells L^-1) (p = 0.038; Figure 3G ). BA was positively correlated with BP (rho = 0.64, p < 0.001), CDOM (rho = 0.51, p = 0.001), and Chl-a (rho = 0.40, p < 0.001), but negatively correlated with NOx (rho = - 0.30, p = 0.014) ( Supplementary Figure 2 ). Depth-integrated bacterial carbon biomass (BCB) from the surface down to mixed layer depth in ANA04 (99 ± 46 mg C m^-2) was two times higher than estimated in ANA06 (45 ± 21 mg C m^-2) ( Table 2 ).

BP ranged from 0.08 to 6.13 µg C L^-1 d^-1 during ANA04 and ANA06 ( Figure 2H ). Like BA, maximum BP in ANA04 was observed at 15-m depth and decreased rapidly with depth, while BP declined only slightly with depth in ANA06. In the upper 100 m of the ASP, BP was four times higher in ANA04 (1.41 ± 1.50 µg C L^-1 d^-1) than in ANA06 (0.39 ± 0.25 µg C L^-1 d^-1) (p <0.001; Figure 3H ). BP was positively correlated with CDOM (rho = 0.90, p < 0.001), Chl-a (rho = 0.74, p < 0.001), DOC (rho = 0.50, p < 0.001), DON (rho = 0.32, p = 0.013), and temperature (rho = 0.38, p = 0.001), but negatively correlated with NOx (rho = -0.62, p < 0.001), PO 4 3 − (rho = -0.63, p < 0.001), salinity (rho = -0.67, p < 0.001), and Si(OH)₄ (rho = -0.39, p = 0.001) ( Table 3 and Supplementary Figure 2 ). When integrated from surface to mixed layer depth, BP was five times higher in ANA04 (73 ± 35 mg C m^-2 d^-1) than in ANA06 (15 ± 8 mg C m^-2 d^-1) ( Table 2 ). The bacterial growth rate in ANA04 (0.76 ± 0.19 d^-1) was two-fold higher than that in ANA06 (0.35 ± 0.11 d^-1) ( Table 2 ).

Unlike BP, BR did not show notable vertical variation ( Figure 2I ). The average BR in ANA04 (131 ± 96 µg C L^-1 d^-1) was three-fold higher than that observed in ANA06 (43 ± 20 µg C L^-1 d^-1) (p = 0.025, Figure 3I ). In this study, BR ranged from 12.85 to 320 µg C L^-1 d^-1 in the upper 100 m of the ASP, which was much higher than previously reported (10 – 53 µg C L^-1 d^-1; William et al., 2016).

The dominant phytoplankton group changed from P. antarctica in ANA04 to a mixed community of diatoms and D. speculum in ANA06 ( Figure 4 ). In ANA06, large pennate diatoms (Proboscia alata, Plagiotropus gaussi, and Corethron pennatum) and centric diatoms (Chaetoceros spp. and Thalassiosira spp.) were the major diatom species, rather than small diatoms (Fragilariopsis spp. and Pseudonitzschia spp.) (Lee et al., 2022). The change in dominant phytoplankton biomass from P. antarctica in ANA04 to diatoms and D. speculum in ANA06 was also confirmed using the disappearance ratio of inorganic nutrients (ΔN:ΔP and ΔSi:ΔN) between the two periods. The ΔN:ΔP ratio was 19.5 in ANA04 ( Figure 5A ) similar with the ratio in previous research on P. antarctica-dominant communities (19.2; Arrigo et al., 1999). However, the ΔN:ΔP ratio in ANA06 decreased to 13.4 ( Figure 5A ), which was between the value for a P. antarctica-dominated system and diatom bloom conditions (9.69; Arrigo et al., 1999). In addition, the ΔSi:ΔN ratio changed from 0.25 in ANA04 to 0.96 in ANA06 ( Figure 5B ). The ΔSi:ΔN ratio of < 0.9 represents a P. antarctica-dominated system, whereas the ratio of 0.9 – 2.15 in the water column suggests a mixed phytoplankton community (Dunbar et al., 2003; Fragoso and Smith, 2012). Both microscopic quantification (Lee et al., 2016b; Lee et al., 2022) and chemical proxies using ΔN:ΔP and ΔSi:ΔN ratios clearly revealed a change in the dominant phytoplankton from P. antarctica in ANA04 to diatoms and D. speculum in ANA06.

The change in phytoplankton communities between the two periods was likely driven by changes in light intensity and iron availability, as previously reported in the Southern Ocean (Timmermans et al., 2004; Tagliabue and Arrigo, 2005; Kropuenske et al., 2009; Lee et al., 2022). Previous studies found that P. antarctica can thrive under dynamic light conditions (Kropuenske et al., 2009), while the growth rate of large diatoms and D. speculum were lower than that of P. antarctica at high light intensity (Biggs et al., 2019; Lee et al., 2022). In the present study, surface light intensity was approximately six times lower in ANA06 than in ANA04 ( Supplementary Figure 1 ), and thus phytoplankton biomass was dominated by large pennate diatoms (31%), centric diatoms (14%), and D. speculum (48%) rather than P. antarctica (3%) in ANA06 (Lee et al., 2022).

In addition, several studies reported that P. antarctica has a lower iron requirement than diatoms (Arrigo et al., 2003; Tagliabue and Arrigo, 2005), while the iron requirement of diatoms declines with decreasing cell size (Timmermans et al., 2004). In the present study, the maximum quantum efficiency (Fv/Fm), which is used as an indicator of phytoplankton status to iron stress (Falkowski et al., 2004), was higher in diatom-dominated ANA06 (> 0.5) than in P. antarctica-dominated ANA04 (0.27–0.36) (Lee, 2017). In previous studies, Fv/Fm was approximately 0.55 in the iron-replete Southern Ocean (Coale et al., 2004), but was under 0.35 in the iron-depleted ASP (Alderkamp et al., 2015; Park et al., 2017). Thus, the growth of large diatoms with higher iron requirements is likely to be favored in ANA06 with higher Fv/Fm.

Bacterial metabolism in the ocean is primarily controlled by temperature and the availability of organic matter (Pomeroy and Wiebe, 2001). In cold polar seas, despite low temperatures, BP during phytoplankton blooms is as high as that reported in temperate waters (Kirchman et al., 2009b). Likewise, in the present study, although the temperature was lower in ANA04 than in ANA06, Chl-a was higher in ANA04 ( Table 1 ), and BP was also four times higher in ANA04 compared to ANA06 ( Figure 3H ). Therefore, bacterial metabolic activities in the ASP are primarily controlled by the availability of organic matter produced by phytoplankton rather than temperature (Kirchman et al., 2009a; Kirchman et al., 2009b; Ducklow et al., 2012; Hyun et al., 2016).

Interestingly, correlation analysis revealed that the dependence of BP on organic matter source varied between ANA04 and ANA06. BP was significantly correlated with only Chl-a (rho = 0.66, p < 0.001) during ANA04, whereas BP in ANA06 showed a significant correlation with CDOM (rho = 0.84, p < 0.001), Chl-a (rho = 0.59, p < 0.001), and DOC (rho = 0.51, p = 0.001) ( Table 3 ). The difference between the two periods was attributed to variation in DOM quality resulting from the change in the dominant phytoplankton community from P. antarctica in ANA04 to diatoms in ANA06 ( Figure 4 ). A large fraction of the DOM exudated by phytoplankton is in the form of polysaccharides, but the main components of these polysaccharides vary by phytoplankton community (Mühlenbruch et al., 2018). Several previous studies found that Phaeocystis spp. excretes more bioavailable forms of polysaccharides than diatoms (Biersmith and Benner, 1998; Aluwihare and Repeta, 1999; Solomon et al., 2003; Alderkamp et al., 2007). Kinsey et al. (2018) also reported that labile DOM concentration and bacterial abundance were highest in Phaeocystis among phytoplankton culture experiments, including various diatoms, which implies that Phaeocystis supplies more bioavailable DOM for bacterial metabolism than diatoms. Our results also revealed that bacterial growth rates in Phaeocystis-dominated ANA04 (0.76 d^-1 on average) were two-fold that measured in diatom and D. speculum co-dominated ANA06 (0.35 d^-1 on average) ( Table 2 ). In the same period Fang et al. (2020) analyzed ¹⁴C-DOC analysis in surface water and also found a greater amount of freshly produced DOC in ANA04 than in ANA06. Given that BP was significantly correlated with only Chl-a in ANA04 ( Table 3 ), the DOC released directly from P. antarctica is likely a major source of organic matter supporting BP (Ducklow et al., 2012).

In contrast to ANA04, BP in ANA06 was positively correlated not only with Chl-a but also with CDOM and DOC ( Table 3 ). CDOM can be produced via various pathways including direct release from phytoplankton, photodegradation of DOM, microbial degradation, sloppy feeding or excretion by zooplankton, and viral lysis (Moran et al., 2000; Carlson and Hansell, 2015; Lee et al., 2016a; Guallar and Flos, 2019). Although diatoms generate less bioavailable DOM compared with P. antarctica (Biersmith and Benner, 1998; Aluwihare and Repeta, 1999; Solomon et al., 2003; Alderkamp et al., 2007), more zooplankton grazing occurs on diatoms than on Phaeocystis spp. (Stelfox-Widdicombe et al., 2004; Nejstgaard et al., 2007), and the presence of grazing stimulates DOM release from phytoplankton (Strom et al., 1997). Therefore, CDOM might be generated from more diverse sources within food web processes under diatom-dominated conditions in ANA06, providing more support for bacterial metabolic activities.

The quantitative significance of the microbial loop is often evaluated using the BP : PP ratio, in which a higher BP : PP ratio implies that relatively more PP is channeled to nano- and microzooplankton via microbial food web processes and/or mineralized to CO₂ within the microbial loop (Kirchman et al., 2009b). In the present study, the BP : PP ratio was three-fold higher in Phaeocystis-dominated ANA04 (0.09 on average) than in diatom- and D. speculum-co-dominated ANA06 (0.03 on average) ( Table 2 ), which indicated that the microbial loop is relatively more significant under Phaeocystis-dominated conditions than diatom-dominated conditions. The results are consistent with a previous study on the ASP (Yager et al., 2016) that reported a relatively lower BP : PP ratio at diatom-dominated stations (0.03) than in P. antarctica-dominated stations (0.06). Considered with respiration, the BCD : PP ratio was 7.45 in ANA04 and 3.80 in ANA06, BCD exceeded PP in both periods, and BGE was very low (0.02 in ANA04, 0.01 in ANA06)( Table 2 ). The BCD : PP ratio and BGE in the present study were much higher and lower, respectively, than the previously reported values in ASP (BCD : PP = 0.43, BGE = 0.11; Williams et al., 2016). These results indicate that bacteria utilize a large amount of semilabile DOM for respiration. According to Chen et al. (2019), although massive labile CDOM production occurs in ASP, semilabile DOC accumulation is low due to the high bioavailability and rapid turnover of DOM. Williams et al. (2016) also suggested that the low DOM accumulation in ASP resulted from rapid DOM turnover by bacteria, and thus a high CDOM production and rapid DOM turnover could explain high BR and low BGE. Consequently, although semilabile DOM consumption by bacterial respiration is significantly large in ASP, the relatively higher BP : PP and BCD : PP ratios in ANA 04 than ANA06 indicate that carbon consumption through the microbial loop is more active during the P. antarctica-dominated condition. As the BP : PP ratio and export flux are inversely related (Cho et al., 2001; Kirchman, 2008), the BP : PP ratio also provides quantitative information about the efficiency of export flux. In the present study, we found a clear inverse relationship between BP : PP ratio and the diatom carbon biomass ( Figure 6A ). The high BP : PP ratio in the Phaeocystis-dominated system and low ratio in the diatom-dominated system strongly implies that the significance of the microbial loop and the efficiency of export flux (i.e., the biological pump) are directly controlled by changes in dominant phytoplankton group. DOM excreted from P. antarctica is rapidly utilized by bacteria compared to diatom-derived DOM (Biersmith and Benner, 1998; Aluwihare and Repeta, 1999; Kinsey et al., 2018). In addition, the zooplankton grazing rate on P. antarctica is lower (Tagliabue and Arrigo, 2003; Tang et al., 2008) and its sinking rate slower (Becquevort and Smith, 2001; Reigstad and Wassmann, 2007) than those of diatoms. Thus, under P. antarctica-dominated conditions, most DOC derived from P. antarctica is rapidly mineralized by bacteria within the water column before it reaches the seafloor (Kirchman et al., 2001). In contrast, more fresh DOC is likely to be transported to deeper ocean during a fast-sinking diatom bloom (Alldredge and Gotschalk, 1989). Fang et al. (2020) during the same periods also found that a greater amount of fresh DOC was transported to the deeper ocean in ANA06 (i.e., the diatom-dominated condition) than in ANA04 (the P. antarctica-dominated condition). These results suggest that as the carbon biomass of diatoms increases, less organic carbon is degraded by bacterial metabolism in the euphotic layer, which results in an increase of POC export flux.

Previous studies in the ASP also revealed that the POC flux efficiency varied depending on the dominant phytoplankton species in the ASP (Ducklow et al., 2015; Kim et al., 2015). In the P. antarctica-dominant polynya, only 1.6% of net primary production (NPP) is delivered to the deep layer, and most of the exported organic matter is mineralized between depths of 60 and 150 m via microbial respiration (Ducklow et al., 2015). However, in the sea ice zone, where diatoms were a dominant phytoplankton group, approximately 18% of NPP is delivered to the deep layer (Kim et al., 2015). Similarly, in the Ross Sea, DeJong et al. (2017) reported that carbon export flux in diatom-dominated regions (7.3 ± 0.9 mol C m^-2 d^-1) was two times higher than that measured in Phaeocystis-dominated regions (3.4 ± 0.8 mol C m^-2 d^-1). In the present study, BR also showed a significant decreasing trend with increasing diatom proportion (y=191.44e^(-0.08x), r² = 0.55, p = 0.043; Figure 6B ). These results imply that the change in major phytoplankton communities from P. antarctica to diatoms weakens the role of the microbial loop and enhances the efficiency of biological pump (i.e., POC export flux) (Passow et al., 2007).