Algal enrichment process for newly formed sea ice in the Dalton polynya off East Antarctica during the late summer–early autumn Open Access

The Southern Ocean is characterized by seasonal changes in sea-ice cover, with approximately 80% of the sea-ice zone experiencing annual formation and melting of sea ice (Eayrs et al., 2019). Sea-ice formation occurs via several steps. The initial phase involves atmospheric cooling of seawater that produces tiny crystals termed frazil ice. Within a few days, the frazil ice bind to each other and subsequently become grease ice or nilas depending on the turbulence of the ocean surface. The subsequent congelation growth and rafting make the sea ice thicker, forming larger ice floes (Petrich and Eicken, 2017).

Sea-ice production is most prominent in coastal polynyas, where katabatic winds transport sea-ice floes offshore and promote cooling of seawater (Tamura et al., 2016). During this process, sea ice expels cold and high-saline brine into seawater below, triggering the formation of Antarctic Bottom Water (AABW). The brine released from new ice also can promote the downward export of inorganic carbon associated with AABW (Murakami et al., 2020); moreover, convection brings micronutrients such as iron from the deep ocean to the surface, which may promote primary production (Tagliabue et al., 2014). Hence, sea-ice formation is important for transporting salt, heat, carbon, and nutrients in the Southern Ocean.

Sea ice itself is a place for primary production and supports a wide range of microorganisms including bacteria, protists, and metazoans (Horner, 1985; Arrigo, 2014). High concentrations of microalgae (referred to as ice algae) live in sea ice and contribute to approximately 10% of annual primary production in the sea-ice zone (Pinkerton and Hayward, 2021). Following the melting of sea ice, ice algae are released into seawater, where they become prey for zooplankton (Kohlbach et al., 2019), sink rapidly, exporting carbon to the depths (Tanimura et al., 1990), or seed phytoplankton blooms (Garrison and Buck, 1985; Wilson et al., 1986). These microorganisms, as well as other particulate and dissolved materials incorporated into sea ice, originate from seawater. Phytoplankton, suspended sediment, zooplankton, and trace metals are known to be enriched in the newly formed sea ice, compared to seawater (e.g., Garrison et al., 1983; Reimnitz et al., 1993; Janssens et al., 2016; Janssens et al., 2018; Ito et al., 2019). Previous studies infer that physical incorporation of algae during sea-ice formation is one of the key processes to determine the concentration of ice algae, with reported variations from below detection limit to 138 µg L⁻¹ in the chlorophyll a (Chl a) concentration of newly formed sea ice (e.g., Ackley et al., 1987; Garrison et al., 1989; Ditullio et al., 1998). Therefore, understanding the mechanisms by which algal cells are incorporated into newly formed sea ice is beneficial for assessing the spatial and temporal variations of organic carbon as well as algae in sea ice, especially for the less-studied autumn and winter seasons (Meiners et al., 2012; Janssens et al., 2016; Wongpan et al., 2020).

The high variability in algal concentration in newly formed sea ice stems from several factors such as the formation and growth process of sea ice (Ackley and Sullivan, 1994), phytoplankton species and size (Gradinger and Ikävalko, 1998; Różańska et al., 2008; Janssens et al., 2016), and in situ growth of ice algae (Janssens et al., 2016; Louw et al., 2022). For instance, the presence of large phytoplankton cells results in high enrichment of algae in sea ice, possibly owing to the greater surface area available to attach to the ice surface or greater entrapment in small brine pockets and channels (Krembs et al., 2000; Krembs et al., 2002). Of these factors, particularly frazil ice formation and incorporation of particles (referred to as suspension freezing) has been deemed an efficient mechanism to accumulate suspended particles (e.g., sediments) in sea ice. Weeks and Ackley (1982) suggested that frazil ice crystals, when rising to the surface, can scavenge phytoplankton and suspended sediments efficiently. This scavenging is evidenced from the high particle concentration observed in the granular ice layer, rather than the columnar ice layer (sea ice derived from congelation growth) based on the texture studies of first-year ice (Ackley and Sullivan, 1994; Ito et al., 2019; Takahashi et al., 2023). Field and satellite (surface reflectance of blue and green bands) observations have indicated extremely high algal concentrations in frazil ice (Gersonde, 1986; Ditullio et al., 1998; Lieser et al., 2015; DeJong et al., 2018) supporting the notion that the efficient incorporation of particles by frazil ice (suspension freezing, which has been used mainly for sediment particles) can be extended to include algal cells in seawater.

The texture of sea ice (e.g., granular, columnar or platelet) provides information on sea-ice formation and growth history. Yet most studies have focused on thick first-year ice (FYI) or old ice (OI: second-year or multiyear ice) during the melting season (Meiners et al., 2011; Corkill et al., 2023; Takahashi et al., 2023). As sea ice undergoes a series of environmental changes affecting ice-algal concentration, such as division of algal cells, decomposition of organic matter, and remineralization of nutrients, isolating the mechanisms of material incorporation in thick ice is difficult. The uptake and assimilation of nutrients by ice algae in thin ice has been deemed negligible because of the short period of time since formation (Garrison et al., 1983; Gradinger and Ikävalko, 1998). However, Louw et al. (2022) reported that nutrient assimilation could occur in the thinner pancake ice (approximately 0.2 m thickness) in winter (July) off East Antarctica, where the salinity-normalized silicate concentration was lower than that in seawater. Tison et al. (2017) observed higher ice-algal concentration in a year with thicker snow and lower freeboard, possibly owing to snow depressing sea ice below sea level and enhancing brine circulation and algal accumulation in young and FYI during winter. Therefore, information on the incorporation of materials and sea-ice texture during the earliest development stage of sea ice (thin new ice) is required to better understand the process leading to ice-algal enrichment.

This study sought to determine what controls the variations in algal incorporation into newly formed thin (<0.20 m thickness) sea ice. We collected thin ice and surface seawater from several sites in the Dalton polynya, under different conditions such as ice coverage by old ice floes, air temperature, and wind speed. Algal incorporation was compared with other properties such as sea-ice texture, macronutrient concentrations, and snow fraction. The brash sea ice (broken from FYI/OI that had survived the melting of the season) was also sampled as a point of comparison for algal communities.

Measurements and observations were made in the Dalton polynya in March 2018 and February–March 2020, during the 59th and 61st Japanese Antarctic Research Expedition, respectively (Figure 1). The study sites included five stations where newly formed thin ice was sampled (Stations J59, 39, 43, 108, and S138) and five stations where FYI/OI was sampled (Stations 40, 83, 86, 99, and 101; Figure 1). At each station, 3–9 sea-ice samples were collected. At the thin ice stations, seawater for the analysis of macronutrients (nitrate: NO₃⁻, phosphate: PO₄³⁻, and silicate: Si(OH)₄), chlorophyll a (Chl a), biogenic silica (BSi) (except Station J59), and phytoplankton composition was sampled from a ship intake (approximately 8 m depth). Seawater samples for macronutrients were immediately frozen (−20ºC) and stored until further analysis. For Chl a and BSi measurements, 300–600 mL of seawater was filtered through Whatman GF/F filters and 0.6 µm polycarbonate filters, respectively. The GF/F filters were soaked in 6.0 mL of N, N-dimethylformamide and stored at −20°C (Suzuki and Ishimaru, 1990). The polycarbonate filters for BSi samples were stored in a deep freezer (−80°C). Seawater samples for phytoplankton composition were fixed using neutral-Lugol solution (final concentration of 2%) and stored at 4ºC in the dark.

Sea ice (newly formed sea ice and brash sea ice) was sampled from the ship using a steel cage with ropes (dimensions of 60 cm × 60 cm), which was lowered from the side deck and maneuvered manually (Nomura et al., 2023). Although this method is accepted for biogeochemical studies for thin ice and brash ice (Nomura et al., 2023), small pieces of thin ice may be subject to a release of brine from the ice when they are scooped from seawater due to their high brine volume fraction (Gradinger and Ikävalko, 1998). However, our thin ice samples showed a wide range of salinity (5.2–23.0) that is comparable to previous studies (Gradinger and Ikävalko, 1998; Janssens et al., 2016; Tison et al., 2017), and we surmise the effect of brine release to have been minimal. In this study, newly formed sea ice is defined as grease ice, nilas, pancake ice, and young ice based on their shapes (Worby and Allison, 1999) and thicknesses, where the plate-like thin ice of nilas and pancake ice is <0.1 m and the thickness of young ice is 0.1–0.3 m. Brash ice is assumed to have originally formed through melting, breaking by wave or icebreaker passage, or by collision between ice floes (Nomura et al., 2023). It is defined as a broken piece of FYI/OI that lacks any shape characteristics of newly formed sea ice (Worby and Allison, 1999). Immediately upon sampling, the temperature of thin ice was measured from a hole drilled to 3 cm depth using a thermometer (K320, Tateyama Kagaku Group, Inc., Japan). Thin ice and brash ice were immediately placed in polyethylene bags and stored at −18°C in the dark.

Newly formed sea ice (excluding grease ice) was cut into 10-cm thick sections for thin-section analyses in a cold room at −15°C. Photographs of the vertical sea-ice thin sections (0.5 mm thickness) were obtained using cross-polarizing light (Lange, 1988). The rest of the sea ice was crushed and mixed well. Then, 200–800 g of sea ice were melted in a 4°C room with 8 times the volume of filtered seawater prepared using a 0.2 µm capsule filter. Algal compositions and Chl a concentration of ice algae for all stations and BSi concentration for Stations 39 and 108 were analyzed for this filtered seawater-buffered meltwater and corrected for the dilution factor. The remaining crushed sea ice was melted at room temperature without the addition of filtered seawater (Roukaerts et al., 2019) and then subsampled for macronutrients, stable oxygen isotope (δ¹⁸O), and salinity for all stations, and BSi concentration for Stations 43 and S138. The pre-treatment (filtering and storing) of samples for Chl a and BSi analyses were carried out in the same manner as for seawater.

After the extraction of pigments for >24 h using N, N-dimethylformamide (Suzuki and Ishimaru, 1990), Chl a concentration of seawater and sea ice was measured using a pre-calibrated fluorometer (10-AU, Turner Designs, USA) following the method of Welschmeyer (1994). The PC filters for BSi analysis were soaked in 0.2 mol L⁻¹ of sodium hydroxide at 40°C for 100 min. The dissolved silica concentration was measured following the molybdenum blue method (Hansen and Koroleff, 1999) using the autoanalyzer (QuAAtro-Marine 5ch, SEAL Analytical, USA). The species composition and cell abundance (cells L⁻¹) of microalgae from seawater and sea ice were investigated using an inverted light microscope at 400× magnification. A sample volume of 10–100 mL was concentrated using an Utermöhl chamber (Edler and Elbrächter, 2010). At least two transects were traversed to observe more than 200 fields and 400 algal cells; frustules of diatoms were excluded from the count. The concentrations of macronutrients (NO₃⁻, PO₄³⁻, and Si(OH)₄) were determined following the method described by Shimada et al. (2022) for sea ice at Station J59 and for all seawater samples. For the other sea-ice samples, macronutrient concentrations were measured following the method of Nomura et al. (2023), and some of the salinity and nutrient data were cited from their study. The δ¹⁸O of sea-ice meltwater was determined using cavity ring-down spectrometry (Picarro L2130-i and vaporizer; Santa Clara, CA, USA). The δ¹⁸O in per mil (‰) was defined using the ¹⁸O/¹⁶O ratio of Vienna standard mean ocean water (VSMOW2) as the standard. The standard deviation of the δ¹⁸O from 10 subsamples taken from reference freshwater (δ¹⁸O value of −10.5‰) was 0.02‰.

The meteoric fraction of snow in sea ice (%) was calculated according to Nomura et al. (2023) based on a mass balance equation model (Jeffries et al., 1994; Jeffries et al., 2001):

where F is the mass fraction for meteoric water or seawater that ranges from 0.0 to 1.0, except F_obs which was 1.0; δ is the δ¹⁸O value, and the respective subscripts “snow,” “sea,” and “obs” are used for snow, seawater, and the measured bulk ice. For δ_sea, we used the δ¹⁸O values of the temperature minimum layer during summer (remnant of winter water) of the Antarctic seasonal ice zone (−0.4 ± 0.1‰ from Nomura et al., 2023) and added a fractionation factor to correct the preferential entrapment of the heavier atoms (¹⁸O) during the freezing of seawater (Toyota et al., 2013). The fractionation factor, however, depends on the growth rate of sea ice and is lower for rapidly growing newly formed sea ice (Toyota et al., 2013). Hence, we first estimated the growth rate of thin ice following the equation by Anderson (1961) using the sea surface temperature, air temperature (°C) averaged over 24 h preceding the sampling, and thickness of the thin ice. Then we determined fractionation factor for each thin ice sample (Toyota et al., 2013). Although Toyota et al. (2013) calculated this factor only from congelation growth, the average fractionation factor was close to the minimum and showed little variation (1.6 ± 0.1‰); hence we applied it for the calculation of F_snow. For determination of snow δ¹⁸O, we collected snow accumulated on the ship deck around 66°S, 120°E during the period of February 22 to March 7, 2020. We observed snowfall 7 times during this period and used the mean of 7 snow samples (−14.4‰) as the end-member for snow δ¹⁸O value.

The volumetric brine fraction (V_b/V) of thin ice was calculated using the temperature and salinity of the sea ice. The calculation followed the method used by Vancoppenolle et al. (2019).

where V and V_b indicate volume of bulk ice and brine, respectively and $∅br$ is the mass fraction of brine. The terms $ρbr$ and $ρi$ are the brine and pure ice densities at the measured temperature of thin ice referring to Zubov (1945) and Pounder (1965).

We quantified the degree of materials (Chl a, BSi, NO₃⁻, PO₄³⁻, and Si(OH)₄) incorporated into newly formed sea ice using the enrichment index (EI) following Gradinger and Ikävalko (1998). This index is defined as:

where $CXIce$ and $CXSeawater$ are concentrations of the variables X (Chl a, BSi, NO₃⁻, PO₄³⁻, and Si(OH)₄) in sea ice and seawater, respectively. The terms S_Seawater and S_Ice denote the salinity of seawater and sea ice, respectively. We used Chl a, BSi, nitrate, phosphate, or silicate for the subscript X.

Based on the pictures of thin sections, the proportion of frazil ice (P_Frazil, %) was defined as the thickness of the frazil ice layer to the total ice thickness (Figure 2), calculated as:

where h_F and h_I are the thickness of the frazil ice-origin layer and ice thickness, respectively. As the frazil ice and columnar ice layers are not always uniform horizontally, the percentage of frazil ice was calculated at five points across the thin section width and their average was used. We excluded the edge of the thin ice for calculating P_Frazil if it was broken or diminished due to rafting and ridging.

Snow has a granular texture when it solidifies into snow ice, which could therefore resemble the frazil ice texture. Hence, the contribution of frazil ice in thin ice can be overestimated when there is heavy snow loading. To minimize this effect, we also calculated a corrected proportion of the frazil ice-origin layer (P′_Frazil) using the snow water fraction F_snow (Section 2.3). This corrected proportion was determined as:

where F_snow is the snow fraction and P_Frazil is the proportion of the frazil ice-origin layer (Equation 5).

All statistical analyses were performed using the R software. The Kruskal-Wallis test and post-hoc multiple comparison (the Steel-Dwass test in the “NSM3” package) were used to compare the differences in the biogeochemical properties and their enrichment indices in newly formed sea ice among the stations. These tests were chosen because the variance was not homogeneous among the stations, except for sea-ice salinity and EI_Nitrate, according to the Bartlett test (p < 0.05). The thin ice, old brash sea ice, and seawater samples were classified by hierarchical clustering to see the difference in algal species composition in discrete samples. For the pre-treatment, the cell abundance was normalized to the total cells to obtain the relative abundance of each taxon (%). Then, the Bray-Curtis dissimilarity (Bray and Curtis, 1957) in the “vegan” package was calculated among each sample. The cluster dendrogram was plotted by Ward’s method in the “vegan” package as well. Kendall’s rank correlation test (Kendall, 1938) in the “stats” package was used to gain the correlation among EI and environmental variables. This non-parametric test was chosen because the Shapiro-Wilk test found that, among the variables chosen (Figure 3), only EI_Nitrate showed a normal distribution (p > 0.05). The one rank sample test (Wilcoxon, 1945) in the “stats” package was used to test EI for significant difference from 1.0 to investigate if the parameters were enriched (>1.0) or depleted (<1.0) in newly formed sea ice.

Newly formed sea ice with different physical and biogeochemical properties (e.g., ice-algal concentration, snow fraction, and ice textures) was collected in the Dalton polynya. In Figure 3, we show the correlations between enrichment of Chl a, BSi, or macronutrients and various sea-ice properties as well as correlations between each factor. The following sections discuss the properties of the thin ice samples and their influences on the enrichment of biogenic particles (Chl a and BSi).

Within the Dalton polynya, we observed several types of newly formed sea ice (Table 1). At Stations 39 and 108, brash ice, which originally came from FYI/OI, was found around nilas and pancake ice (Figure 1j and l). At Station 43, streaks of grease ice were observed in open waters (Figure 1k). In contrast to Station 43, nilas with uniform ice thickness were found at Stations S138 and J59 (Figure 1h and i). The salinity of newly formed sea ice was variable, ranging from 5.2 to 23.0. The salinity median was lower at Stations 39, 108, and S138 (<12.0) relative to Stations 43 (19.4) and J59 (18.4; Figure 4a). The thickness of all sea ice ranged from 0.02 m to 0.15 m, with young ice (sea ice thicker than 0.1 m) sampled only at Stations 39 and 108 (Figure 4b). The range of brine volume fraction (V_b/V) was 12.2–61.4%, with the highest percentage in pancake ice at Station 43 (Figure 4c). The δ¹⁸O ranged from −0.2‰ to 1.5‰, with only 2 samples from Station 39 defined as snow ice (δ¹⁸O < 0; Lange, 1988). The snow fraction F_snow, as determined using δ¹⁸O, ranged from −1.7% to 9.2%. The variation of F_snow was generally higher at Stations 39 and 108, compared to Stations 43 and S138 (Figure 4d and e). The thin-section analysis revealed that sea-ice texture differed considerably among the stations (Figure 4f). Newly formed sea ice collected at Stations 43 and 108 (excepting sea-ice sample No. 108–5) had 100% of P_Frazil,, while at the other stations, both frazil and columnar ice layers were observed. The P_Frazil was the most variable at Station 39, ranging from 20% to 100%. At Stations S138 and J59, columnar ice texture was more present (and dominant at Station J59) and the median of P_Frazil was 41% and 67%, respectively. The F_snow-corrected proportion of the frazil ice-origin layer (P′_Frazil) ranged from 20% to 102%. The P′_Frazil above 100% was due to a negative F_snow value possibly caused by spatial variations in the δ¹⁸O seawater (Tamura et al., 2023). The increase in the minimum value at Station 39 was due to the lack of δ¹⁸O data in a sample. However, P′_Frazil generally showed a similar trend as P_Frazil (Figure 4g); values were higher at Stations 43 and 108 (76–102%) than Stations 39 and S138.

We observed several types of newly formed sea ice, which corresponded with the surface wind speed and turbulence in sea surface. Grease ice was only sampled at Station 43, where the highest wind speed (8.2 m s⁻¹) and low water temperature were recorded, which agrees with the results of Eicken and Lang (1989), who found that increased wind speed results in the formation of grease ice over thin ice such as nilas. Notably, the proportion of frazil ice (P′_Frazil) showed positive and negative correlation with wind speed and air temperature, respectively (Figure 3), stressing the significant role of turbulence and rapid cooling of seawater on frazil ice texture. In addition, the minimum thickness of pancake ice (0.02 m) and high salinity suggest that sea ice at Station 43 was likely at the earliest stage of sea-ice formation. In contrast, new ice at Station S138 was composed of nilas with uniform thickness and low P_Frazil (Figure 4b and f). The low wind speed (2.1 m s⁻¹) likely favored the congelation growth of sea ice. At the other stations, the presence of brash ice (Stations 39 and 108) and fast ice (Station J59) likely made ice textures more variable. The ranges of P_Frazil were greater at Stations 39, 108, and J59 than Stations 43 and S138, and some thin ice had high P_Frazil under moderate wind speed (4.8–6.1 m s⁻¹). These results are partly due to the presence of old ice floes which can mitigate waves and assist formation of nilas with uniform thickness (Zatko and Warren, 2015) or congelation ice (Toyota et al., 2020).

Chl a and BSi concentrations ranged over 0.2–11.6 µg L⁻¹ and 0.5–11.1 µmol L⁻¹, respectively, with the maximum value of 11.6 µg L⁻¹ found at Station 39 (Figure 5a and b). Chl a concentrations differed significantly between Stations 39 and S138 and Station 108, yet no difference was detected for BSi samples among any stations. Chl a concentrations in seawater ranged over 0.5–1.7 µg L⁻¹ and was the highest in Station 108 (Table 1). The maximum seawater BSi concentration was also found at Station 108, yet the concentration was more variable (up to 6-fold difference) among stations. NO₃⁻, PO₄³⁻, and Si(OH)₄ concentrations in newly formed sea ice ranged over 4.6–16.8 µmol L⁻¹, 0.2–1.1 µmol L⁻¹, and 2.5–21.9 µmol L⁻¹, respectively (Figure 5c–e). They generally followed sea-ice salinity and were higher at Stations 43 and J59.

Chl a concentrations in newly formed sea ice observed in this study were on the lower side of the range reported by previous studies (Table 2). A few studies have reported BSi concentrations in sea ice (Niimura et al., 2000; Fripiat et al., 2007): unlike the FYI (7.0–443.0 µmol L⁻¹; Fripiat et al., 2007), our newly formed sea ice accumulated less BSi, possibly due to the short period since its formation. In this study, silicoflagellates (Dictyocha speculum) and silicic cysts of Achaeomonadaceae were present, but their relative abundance was quite low (<1% of total cells) compared to that of diatoms (60–99% of total cells). Hence, the BSi concentration was largely derived from diatoms.

The range in total cell abundance in newly formed sea ice was 1.7 × 10⁵–9.4 × 10⁶ cells L⁻¹ (Figure 6). Brash ice generally exhibited higher and more variable values (open diamonds in Figure 6, range of 1.4 × 10⁵–4.5 × 10⁸ cells L⁻¹). The clustering analysis classified seawater and sea-ice (newly formed sea ice and brash ice) samples into the two major clusters (Figure 6) based on species composition. One (Group A) was composed of all brash sea-ice samples and five newly formed sea-ice samples from Station 39. The other (Group B) included the rest of newly formed sea-ice and all seawater samples. The old brash ice was characterized by a higher contribution of the diatom Fragilariopsis cylindrus, accounting for 39.9–94.1% of the total cells. Contrastingly, the newly formed sea ice in Group B showed a generally higher contribution of F. curta (0.0–36.8%), Pseudo-nitzschia spp. (2.8–45.3%), and nanoflagellates (0.0–26.3%). The exception was Station J59, where Chaetoceros spp. were dominant (8.2–38.3%) in seawater and two sea-ice samples. These characteristics are evident from the averaged composition of algae (Figure 7), where algal composition in newly formed sea ice in Group A is similar to old brash ice, but clearly distinct from sea ice in Group B and from seawater.

While the relative abundance of nanoflagellate (<5 µm) in sea ice was lower than in seawater, diatoms (Fragilariopsis curta, Pseudo-nitzschia spp., and Chaetoceros spp.) were more dominant, suggesting that selective incorporation occurred during sea-ice formation. The difference in species composition between seawater and new ice has been reported in several studies (Różańska et al., 2008; Kauko et al., 2018). Różańska et al. (2008) found a higher proportion of larger (4 µm or greater) cells in new ice. Gradinger and Ikävalko (1998) also showed that the enrichment index is 15.7–49.1 times higher for diatoms than for Synechococcus-like picoalgae. The high stickiness of cells mediated by exopolymeric substances (EPS), especially associated with larger autotrophs, is suggested to be responsible for their higher enrichment in thin ice (Riedel et al., 2007). This selectivity likely promotes diatoms (>5 µm) in successfully colonizing thin sea ice.

The brash ice samples from FYI/OI were dominated overwhelmingly by Fragilariopsis cylindrus (Group A; Figure 6), suggesting the succession of species composition from thin ice to FYI/OI (Garrison et al., 1983; Gleitz and Thomas, 1993; Niimura et al., 2000; Louw et al., 2022). This proliferation of diatoms is suggested by Nomura et al. (2023), where a significant decrease of salinity-normalized Si(OH)₄ concentration was observed in the drift ice from the Indian sector of the Southern Ocean. During the sea-ice growth season, F. curta abundance increases from winter to spring in the East Antarctic sea ice (Fiala et al., 2006; McMinn et al., 2007). During the sea-ice melt season (late spring to summer), F. cylindrus may become predominant in drift ice (Takahashi et al., 2022). Our results corroborate that ice-algal community shifts from the mixed communities (nanoflagellates, Pseudo-nitzschia spp., Chaetoceros spp.) in thin ice to the diatom-dominated communities (F. cylindrus) in FYI/OI. The change in predominant diatom species starts as early as 2 weeks after ice formation (Gleitz and Thomas, 1993). Hence, species succession is assumed to be responsible for the clear separation between Groups A and B in our samples.

EI_Chla and EI_BSi were significantly higher than 1.0 (one rank sample test, p < 0.05) with ranges of 1.3–70.3 and 1.3–27.0, respectively (Figure 8a and b). The median of EI_Chla differed between stations and was significantly higher at Station 39 (Steel-Dwass test, p < 0.05) than Stations 108 and S138 (1.3–4.0). Unlike EI_Chla, the median of EI_BSi was the highest at Station S138 (8.3), and there was no significant difference among the stations. Even at the same station (Station S138), EI_BSi was 4.6–6.7-fold higher than that of EI_Chla.

The EI_Chla and EI_BSi values significantly higher than 1.0 suggest that newly formed sea ice concentrated biogenic particles within the ice. This concentration could have been caused by algal stickiness (Riebesell et al., 1991), where ice algae are thought to attach to EPS, as indicated by high concentrations of EPS in new ice and older FYI (Krembs et al., 2002; Meiners et al., 2004; Riedel et al., 2007; Becquevort et al., 2009). van der Merwe et al. (2009) and Becquevort et al. (2009) found that algal concentration in brines (directly sampled using the sack-hole technique) exhibited values lower than those observed in sea-ice cores. This retention of ice algae within the brine network of the ice would increase the EI of ice algae as sea ice grows and becomes desalinated. Previous studies found that EI is higher in thin FYI and pancake ice than in nilas and grease ice (Table 2). Positive relationships were found in Arctic sea ice between EI_Chla and ice thickness (Gradinger and Ikävalko, 1998; Riedel et al., 2007). We did not find any significant correlation between ice thickness and EI_Chla or EI_BSi, likely due to the lower range of ice thickness (0.02–0.15 m) than in the previous studies (<0.01–0.48 m).

The ranges of macronutrient EIs were 0.8–1.4 for EI_Nitrate, 0.7–3.0 for EI_Phosphate, and 0.3–1.1 for EI_Silicate (Figure 8c–e). The median EI was 1.2, 1.1, and 0.8 for EI_Nitrate, EI_Phosphate, and EI_Silicate, respectively, and significantly higher than 1.0 for EI_Phosphate whereas the median for EI_Silicate was lower than 1.0 (one rank sample test, p < 0.05). These findings suggest that nitrate and phosphate were enriched in thin ice while silicate concentration was diminished, although these results fall within the ranges reported in previous studies (Gradinger and Ikävalko, 1998; Riedel et al., 2007; Janssens et al., 2016). Although nitrate concentration in snow is an order of magnitude lower than in sea ice (Nomura et al., 2023), snow may be a slight source of nitrogen to newly formed sea ice. Other sources of NO₃⁻ are the regeneration of NH₄⁺ from organic matter and its nitrification to NO₃⁻, which is reported to lead to nitrate enrichment in sea ice (Priscu et al., 1990; Riedel et al., 2007). The regeneration of NH₄⁺ occurs even in thin ice (0.001–0.03 m thickness) at rates equivalent to those in open water (Riedel et al., 2007). Such heterotrophic activity could also enhance the phosphate concentration in sea ice. Elevated phosphate concentration with high algal biomass has been reported from sea ice (Takahashi et al., 2022), which stems from the selective assimilation of phosphorus by diatoms and remineralization to PO₄³⁻ through algal-derived organic matter (Arrigo et al., 1999; Fripiat et al., 2017). A positive relationship between phosphate and algal biomass is also supported in this study, as maximum EI_Phosphate (sample No. 39–4) also had the highest Chl a concentration and was dominated by diatoms (98% of total cells). On the other hand, the regeneration of Si(OH)₄ from diatom frustules is much slower than NO₃⁻ and usually observed in MYI (Sahashi et al., 2022) leading to lower EI_Silicate than EI_Nitrate and EI_Phosphate.

Although the uptake of silicate by ice diatoms is suggested in our thin ice samples, no relationship was observed for EI_Nitrate or EI_Silicate with EI_Chla or EI_BSi (Figure 3). Nutrient and Chl a concentrations in sea ice do not always show negative relationships even in FYI/OI due to supply from seawater (Corkill et al., 2023; Henley et al., 2023). However, Garrison et al. (1983) reported that algal Chl a concentration can reach 27 µg L⁻¹ without nutrient assimilation (via only physical incorporation of cells) in thin ice. The lack of relationships between algal enrichment and EI_Nitrate or EI_Silicate suggests that growth within the sea ice, as determined by nutrient uptake, does not explain the variability in ice-algal concentration.

On Antarctic sea ice, snow is generally considered low in macronutrients, biologically active algae (Chl a), and salt (Sahashi et al., 2022); as a result theoretically snowfall should not affect EI (Equation 4). However, we found that F_snow correlated positively with EI_Chla (Figure 3). One possible explanation is that high snow thickness depressed newly formed ice under the freeboard, increasing sea-ice porosity and temperature and facilitating the growth of ice algae, as observed in thicker winter young ice and FYI (Tison et al., 2017). In our study, however, the estimated snow thickness was a maximum of 0.8 cm (from F_snow and ice thickness), which would only depress ice with a thickness <3.2 cm below sea level to allow flooding (Arndt et al., 2017). Sea-ice temperature was close to that of seawater, ranging from −3.8°C to −1.6°C, and V_b/V exceeded 5% (12.1–61.4%). Under these conditions, snowfall would not create a significantly mitigating environment for ice algae or a replenishment of seawater (carrying nutrients) compared to observations by Tison et al. (2017) who found that some young ice exhibited low permeability (V_b/V < 5%) or an extreme environment for ice algae (<−6°C or salinity above 50; Arrigo and Sullivan, 1992).

As discussed in Nomura et al. (2023), the uncertainties in calculating F_snow may help explain the observed correlation between F_snow and EI_Chla. Our sampling sites were located adjacent to the Moscow University Ice Shelf and the Totten Ice Shelf (Figure 1b), and the inflow of glacial meltwater (Moreau et al., 2019; Hirano et al., 2023) can influence the seawater δ¹⁸O. Input of meltwater does not affect the snow δ¹⁸O; however, F_snow could be overestimated if the seawater had lower δ¹⁸O due to mixing with glacial meltwater with lower δ¹⁸O (Equations 1 and 2). After removing the two southern-most stations closest to the ice shelves (Stations 43 and 108; Figure 1b), the relationship between F_snow and EI_Chla showed no significance (τ = 0.33, p = 0.16). Using the mean for seawater end-member had thus resulted in an overestimation of F_snow at these stations and strengthened the positive relationship between EI_Chla and F_snow.

Our results for EI_Chla and EI_BSi (Figure 9) revealed that the EI was highly variable and that neither EI_Chla nor EI_BSi showed a positive correlation with P_Frazil or P′_Frazil. Frazil ice contains only about 1% of algal cells within its structure through nucleation, compared to the higher proportion of algae trapped in the gaps between clustered frazil ice crystals; therefore, interactions between frazil ice and algae are essential to capturing an optimal amount of biogenic particles (Ackley, 1982). Because this flocculation process takes 30 min or more (Garrison et al., 1989), frazil ice crystals must remain suspended in the water column before consolidating at the surface. The likelihood of contact between frazil ice and particles depends on the depth at which frazil ice forms, which can reach 100 m (Ito et al., 2021). In our study, the seawater temperature at 8 m depth was above the freezing point (Table 1) despite air temperature being low (−13.3 to −8.2°C, except at Station J59). Ohshima et al. (2022) reported that frazil ice can penetrate to approximately 80 m depth under strong and persistent winds (>15 m s⁻¹) in an Antarctic coastal polynya during winter. However, we recorded only 2.1–8.4 m s⁻¹ at our observation sites, which suggests that the supercooling of seawater and the subsequent sea-ice formation took place at 0–8 m depth. The moderate conditions may prevent frazil ice suspension in the water column; hence, the time to flocculate and contact phytoplankton is likely limited. This hypothesis is supported by an experiment that used suspended sediment particles (Smedsrud, 2001), where increased particle concentration (higher possibility of contact with ice crystals) resulted in a higher degree of particle incorporation into frazil ice.

Although Chl a and BSi concentrations in sea ice correlated positively with each other (Kendall’s rank correlation test, p < 0.05), their enrichment versus ice texture showed different trends. EI_BSi significantly decreased with higher P_Frazil (P′_Frazil), while no relationship was discerned for EI_Chla (Figure 9). The decoupling of Chl a and BSi enrichment was possibly caused by differing algal retention due to algal species and size and the microstructures of frazil ice and columnar ice. The predominant non-silicified algae were nanoflagellates (presumably Phaeocystis sp.), which accounted for up to 66% of the total cells in the seawater (Figure 6). As discussed earlier (Section 3.3), small algae (<5 µm) were less enriched than diatoms owing to their smaller cell size and lower stickiness (as inferred from EPS concentration; Riedel et al., 2007; Różańska et al., 2008). In this study, the surface area of the nanoflagellates (3.2 µm in diameter) was 32 µm², resulting in a smaller contact area with ice crystals than the smallest dominant diatom, Fragilariopsis cylindrus (6.0 µm in length, surface area 68 µm²) (Takahashi, unpublished data). Passow (2002) found that Phaeocystis spp. have a lower exopolymer concentration per Chl a than diatoms, suggesting that nanoflagellates are more susceptible to movement during brine exchange or release, whereas diatoms (BSi) tend to remain in thin ice. The lower retention of nanoflagellates in thin ice may become more pronounced as the newly formed ice develops via congelation growth. In our study, V_b/V exceeded 5% (22.3% on average), indicating that an exchange with the underlying seawater occurred (Golden et al., 1998). Krembs et al. (2000) reported that the surface area of the brine tubes within columnar ice was several times larger than that of granular ice despite the fact that V_b/V was almost the same between the two ice textures (approximately 20%). Nanoflagellates with poor attachment properties are thought to be prevented from entrapment in newly formed ice by seawater exchange (V_b/V > 5%), especially in columnar ice, where a greater surface area is in contact with seawater. This process leads to a decrease in EI_Chla in congelation ice (low P_Frazil or P′_Frazil samples); hence, unlike BSi, no negative correlation was detected between EI_Chla and sea-ice texture.

In addition to frazil ice production, the presence of FYI/OI can influence EI and the community composition of newly formed sea ice. Algal cells incorporated into newly formed sea ice come from either seawater (phytoplankton), sediment (benthic algae), or FYI/OI (released ice algae on melting), and each contribution may differ regionally (Różańska et al., 2008). Water depths >80 m and high seawater temperatures (Table 1) rule out incorporation from sediments. Kauko et al. (2018) suggested MYI (released ice algae) as a source of algal cells incorporated into thin ice formed in cracks and leads north of Svalbard. Such release from and re-incorporation into sea ice of sympagic algae has been reported elsewhere in the northern hemisphere FYI/OI (Niimura et al., 2000; Olsen et al., 2017). In this study, the clustering analysis (Figure 6) indicated that several newly formed sea-ice samples (39–4 to 39–8, belonging to Group A) were influenced by the release of algal material from these older sea-ice floes because of the high similarity in species composition. Using this result, relationships between the “source” of algae in thin ice and P_Frazil (P′_Frazil) as well as EI (EI_Chla and EI_BSi) were further investigated by highlighting the older ice-influenced thin ice samples (red squares in Figure 9). The 5 samples (including the maximum No. 39–5) belonging to group A had relatively high EI (8.5–70.3 for EI_Chla and 1.3–27.0 for EI_BSi; Figure 9).

The released ice algae from FYI/OI could be one of the reasons for the wide range of Chl a concentration and EI_Chla (Table 2) as well as high EI in columnar ice in this study. Even in late summer, the release of ice algae can take place via melting or ablation of sea ice (Wilson et al., 1986; Wright et al., 2010), as indicated by the similarity in algal species or pigments from sea ice and the water column. Chl a concentration in FYI/OI is known to be two orders of magnitude higher than in newly formed sea ice (Arrigo, 2017), while salinity is lower due to the desalination during sea-ice growth and surface melting (Petrich and Eicken, 2017; Nomura et al., 2023). The freeze-up of low salinity meltwater with high algal concentration can thus result in high EI, which may account for extremely high EI_Chla in nilas (EI_Chla = 140, Table 2) formed under calm conditions. Considering that there is no relationship between the proportion of frazil ice and EI_Chla (Figure 9a and b), the process may enhance algal concentration/incorporation in thin ice in late summer–early autumn without high production of frazil ice (DeJong et al., 2018).

The phytoplankton composition was different from that of ice algae in older ice (Figure 7), suggesting that seawater collected in our study was not generally influenced by FYI/OI. This general lack of influence may stem from the patchy oceanographic features in the region at this time of year. In thin ice at Station 39, 5 of 8 thin ice samples were found to be influenced by FYI/OI (clustering based on the species composition; Figure 6), suggesting that release from FYI/OI and re-incorporation into thin ice occurred near the surface of the water column (above 8 m depth) or around older ice. For instance, under ice floes with advanced melting, there is a very low salinity and high Chl a sea-ice meltwater at the surface (Mundy et al., 2011) possibly due to its low density. In this study, we only sampled from 8 m depth and did not conduct high-resolution water sampling vertically or horizontally, so the actual salinity and algal concentration of the (melt) water remains unknown. However, as Takahashi et al. (2022) reported that phytoplankton composition in seawater can be quite similar to ice under lower seawater temperature or close proximity to sea ice, we infer that the release of ice algae occurred on a smaller scale (such as around individual pieces of broken ice). Sampling seawater at a higher vertical and horizontal resolution would provide a further test of our hypothesis of release and re-incorporation of algae from FYI/OI to newly formed sea ice.

FYI/OI can also influence the texture of surrounding newly formed sea ice. Toyota et al. (2020) and Janssens et al. (2016) observed vertical textures of thin ice formed in holes or pools within FYI. They showed that columnar ice grows earlier (from the top few millimeters) than the surrounding ice floes. Field observations have also reported the formation of nilas near ice sheets even at high wind speeds (Zatko and Warren, 2015). Therefore, when older ice is still present during new ice formation, frazil ice formation by wave action may be limited and congelation growth could be the main process driving ice formation. Its close proximity to older ice also suggests that newly formed sea ice is more susceptible to meltwater and associated release of ice algae from older ice (Niimura et al., 2000). Therefore, when ice forms around broken ice or old ice floes, thin columnar ice with a high EI is more likely to form. We conclude that EI_BSi and P_Frazil (P′_Frazil) show a negative relationship because of the contribution of old sea ice and the low enrichment of biogenic particles in frazil ice (Section 3.7).

Physical and biogeochemical properties of newly formed East Antarctic sea ice were analyzed at multiple sites covering different sea-ice conditions. Contrary to the “suspension freezing” hypothesis for particle incorporation by frazil ice, frazil ice formation did not result in higher incorporation of algae because of the limited turbulence and very early stage of ice formation. Based on the algal composition of thin ice, seawater and older sea ice (FYI/OI), we suggest that some newly formed sea ice can re-incorporate ice algae released from FYI/OI. FYI/OI in polynyas not only provide ice algae but can also promote congelation growth of thin ice by mitigating the turbulence of seawater. The combined effect of the incorporation of released ice algae and columnar ice formation around older sea ice may explain the higher enrichment of diatoms (BSi) in our columnar ice. This process could be most prominent around the edges of polynyas (Kashiwase et al., 2021), where FYI/OI and new ice coexist. Recently, an annual decline in the circum-Antarctic sea-ice area and fast ice areas from several sectors have been reported (Eayrs et al., 2021; Fraser et al., 2021). Such shifts in ice cover may change algal enrichment in newly formed sea ice as well as total sea-ice production (Tamura et al., 2016). To better understand the seasonal and annual changes in Chl a and organic carbon concentration in the seasonal sea ice zones, environmental conditions in polynyas (e.g., sea surface temperature, wind speed, and presence of older sea ice) should be addressed and paired with biogeochemical variables.

Chl a and macronutrient concentration data are available in the data repository of the National Institute of Polar Research (Tokyo, Japan) for 2018 (http://doi.org/10.17592/001.2020070703) and 2020 (http://doi.org/10.17592/001.2021040701). The other data used in this study are stored in the data repository (https://doi.org/10.17592/001.2024091101).

We are grateful to the captain, crew, and summer and winter members onboard the icebreaker Shirase during the 59th and 61st Japanese Antarctic Research Expeditions for assisting the observations. Our appreciation goes to Dr. Tsuneo Odate who led the Antarctic research for years. We appreciate Dr. Kay I. Ohshima, Dr. Takenobu Toyota, Masahiro Hirasawa, Kyoko Kitagawa (Hokkaido University) for supporting the vertical section analysis of sea ice at the Institute of Low Temperature Science, Hokkaido University. We wish to thank Dr. Jean-Louis Tison (Université Libre de Bruxelles), Dr. Delphine Lannuzel (University of Tasmania), and the anonymous reviewer, as well as Dr. Jody W. Deming (editor), for their helpful comments, which greatly improved the manuscript.

This study was supported by JSPS KAKENHI Grant Nos. 21J14914, 24KJ0006, and 24K17997 to K.D. Takahashi, Nos. 16J04868, 20K23370, and 22K18027 to M. Ito, No. 18F18794 to P. Wongpan, No. JP17H06319 to M. Moteki, No. 20H04313 to R. Makabe, and No. 20H05707 to K.I. Ohshima. This study was also supported by Research Project Funds of the National Institute of Polar Research (No. KP-308 to T. Odate), Japanese Antarctic Research Expedition (Nos. AP-0923 and AP-0939 to M. Moteki), and the Grant for Joint Research Program of the Institute of Low Temperature Science, Hokkaido University (22S012). P. Wongpan was also supported by the Australian Government as part of the Antarctic Science Collaborative Initiative program and contributes to Project 6 (Sea Ice) of the Australian Antarctic Program Partnership (project ID ASCI000002).

The authors declare that there are no competing financial or personal interests in relation to the work described.

Conception and design: KDT, RM, TT.

Acquisition of data: KDT, MI, NN, MS, MYK, PW, TT, RM.

Analysis and interpretation of data: KDT, MI, NN, DN, SA, PW, GDW, MYK, RM.

Drafted the article: KDT.

Revised and approved the submitted version for publication: KDT, MI, NN, DN, MS, SA, PW, GDW, MYK, TT, MM, TH, RM.

How to cite this article: Takahashi, KD, Ito, M, Nojiro, N, Nomura, D, Sano, M, Aoki, S, Wongpan, P, Williams, GD, Yamamoto-Kawai, M, Tamura, T, Moteki, M, Hirawake, T, Makabe, R. 2025. Algal enrichment process for newly formed sea ice in the Dalton polynya off East Antarctica during the late summer–early autumn. Elementa: Science of the Anthropocene 13(1). DOI: https://doi.org/10.1525/elementa.2024.00038

Domain Editor-in-Chief: Jody W. Deming, University of Washington, Seattle, WA, USA

Guest Editor: Delphine Lannuzel, University of Tasmania, Hobart, Tasmania, Australia

Knowledge Domain: Ocean Science

Part of an Elementa Special Feature: Understanding the Trajectory and Implication of a Changing Southern Ocean: The Need for an Integrated Observing System