Phytoplankton community structure and variability at the landfast ice–water column interface in Terra Nova Bay, Ross Sea (Antarctica)

Landfast ice in Terra Nova Bay represents a dynamic ecological interface where sea-ice structure, meltwater inputs, and platelet-ice processes interact to shape microalgal communities across the ice–ocean boundary. During the austral spring–summer 2015/2016, we examined how seasonal transitions—from early-season to progressive melting and stratification—modulate the development of sympagic assemblages and their connection with the underlying water column. By integrating sea-ice thermodynamics, pigment signatures, nutrient variability, and water-column profiles, we describe how physical forcing and biogeochemical gradients interact to shape community composition and functional structure. Surface salinity decreased from a maximum of 36.67 to a minimum of 30.67, while surface chlorophyll a exceeded 40 µg L - ¹ during the early-December biomass maximum and subsequently declined to below 1 µg L - ¹ in late December and January. The pigment record showed changes in the relative contribution of fucoxanthin, 19′-hexanoyloxyfucoxanthin (19′HF), and chlorophyll c2, while the multivariate analysis identified a significant vertical effect (PERMANOVA, F = 1.7188, R² = 0.1629, p = 0.0003), with sampling date accounted for as a temporal blocking factor. Together, these findings highlight the meltwater-influenced layer beneath landfast ice as a biologically relevant interface linking sea-ice processes with phytoplankton dynamics in Antarctic coastal waters.

Sea ice is one of the most dynamic and threatened habitats on Earth, acting as a driver of biodiversity in polar regions and as a key regulator of climate and biogeochemical cycles at the global scale (Arrigo, 2014; Post et al., 2013; Thomas, 2017; Vancoppenolle et al., 2013). Although it may appear as a uniform white blanket over the oceans, sea ice encompasses highly diversified microenvironments with distinct chemical–physical characteristics. Among these, landfast ice represents the stationary sea ice attached to the coast that does not move horizontally and accounts for only 4–13% of total sea ice coverage (Eicken, 2003; Fraser et al., 2023; Thomas, 2017). A distinctive feature of landfast ice is the presence of the sub-ice platelet layer (SIPL), formed through the accumulation of platelet ice resulting from basal ablation of ice shelves and interactions with supercooled ocean water (Foldvik and Kvinge, 1974; Jacobs et al., 1985; Wongpan et al., 2021). This process generates large ice flakes that accumulate beneath the sea ice, creating a porous layer in which the combination of large ice crystals, high nutrient availability, and sufficient light fosters the growth and accumulation of substantial sympagic biomass (Arrigo et al., 2010). When interstitial water between crystals freezes, the SIPL consolidates forming an extremely productive habitat providing feeding grounds, refugia, and spawning sites for numerous species, thereby sustaining Antarctic biodiversity and playing a key role in Antarctic trophodynamics (Bluhm et al., 2017; Fraser et al., 2023; Gow et al., 1998; Jeffries et al., 1993; Smith et al., 2001). Sympagic diatoms (Bacillariophyceae) dominate primary production in platelet ice, with pennate benthic species prevailing in interstitial spaces and centric diatoms dominating the SIPL, where planktonic contributions are higher (Riaux-Gobin et al., 2003; Saggiomo et al., 2017; van Leeuwe et al., 2018). These microalgae are crucial for polar ecosystems, contributing an estimated 50–65% of total primary production in Antarctic coastal systems (McMinn et al., 2010a). Their growing season begins in spring, when light is no longer limiting and nutrient availability from seawater enables colonization of SIPL interstices, forming dense algal layers. With rising air temperatures and periodic intrusions of warm, nutrient-rich Circumpolar Deep Water (CDW), landfast ice undergoes progressive summer melting (Arrigo et al., 2012; McMinn et al., 2010b; Young et al., 2024). Mechanical breakup, driven by wave action and wind stress, destabilizes the ice cover, while freshwater inputs from ice-sheet melting—against a background of Antarctic ice-mass loss documented since 2002 (NASA, 2023)—may contribute to changes in coastal ecological dynamics. In addition, algal accumulation within sea ice enhances light absorption, paradoxically accelerating melting and releasing large amounts of biogenic carbon into the water column, much of which supports benthic food webs (Duncan et al., 2024; Kohlbach et al., 2016; Niemi et al., 2024). This self-reinforcing cycle highlights sea ice as a dynamic and ephemeral habitat where interactions between small- and large-scale processes drive complex energy flows. Sea ice also represents an important reservoir of macro- and micronutrients (e.g., iron), mediated by Fe-binding ligands (Genovese et al., 2022; Lannuzel et al., 2015; Smith, 2022). Seasonal melting releases freshwater that induces stratification in the water column, stabilizing upper layers while seeding them with nutrients and microalgae, thereby favoring planktonic blooms (Bolinesi et al., 2020; Daly and Smith, 1993; Garrison et al., 1987; Smith and Comiso, 2008). Stratification in coastal areas is typically associated with diatom dominance, linked to their superior photoprotection, higher half-saturation constants for iron, and high silica demand compared to other groups (e.g., haptophytes) (Person et al., 2020, 2021; Ryan-Keogh et al., 2017; Strzepek et al., 2012; Zhang et al., 2023). However, intense and unusual Phaeocystis antarctica blooms have been reported in Terra Nova Bay (Ross Sea) under strong stratification, raising questions about the relative importance of mesoscale dynamics versus intraspecific nutrient competition in shaping functional group distribution (Bolinesi et al., 2020; Lancelot et al., 2009; Mack et al., 2017; Mangoni et al., 2019). Despite early descriptions of sea-ice extent and landfast-ice dynamics in Terra Nova Bay dating back to 1910–1913 (Wright and Priestley, 1922), information on biological interactions in this area remains scarce, particularly in the context of ongoing climate change.

This study aims to elucidate the temporal evolution of microalgal community structure associated with landfast ice in Terra Nova Bay during the austral spring–summer of 2015/2016. Special attention is dedicated to the ice–ocean interface, tracking under-ice dynamics to assess how communities respond to freshwater inputs, how sympagic assemblages develop within the ice, and how shifts in the physico-chemical properties of the water column—driven by offshore intrusions, continental meltwater discharge, and sea-ice melting—reshape community composition. In this study, functional traits are considered as integrative ecological descriptors derived from pigment composition and photophysiological responses, providing a framework to interpret how phytoplankton functional characteristics vary in relation to environmental conditions and seasonal change. We specifically hypothesized that (i) seasonal freshening and the resulting near-surface stratification would progressively modify the coupling between the ice–water interface and deeper waters, and (ii) these changes would be accompanied by a shift in the relative contribution of diagnostic pigments, from a stronger fucoxanthin-associated signal toward greater contributions of 19′HF and other pigments associated with smaller phytoplankton groups. We further used the UV/iron incubation experiment to test whether iron availability and UV exposure modified phytoplankton biomass, photophysiological performance, and pigment composition, providing a complementary physiological perspective on the field observations.

Field work activities were carried out in the frame of the Italian project for Antarctic Research (PNRA) CEFA (Coastal Ecosystem Functioning in a Changing Southern Ocean). Sampling activities were carried out between 9 November 2015 and 19 January 2016 after the installation of a research camp on the landfast ice at Station C (74°41.02′S, 164°07.85′E) (Figure 1). At the start of the sampling period, landfast-ice thickness was 260 cm, decreasing to approximately 250 cm by the end, while snow cover was absent. The station was located in a transitional area receiving, on one side, meltwater from glaciers discharging into Tethys Bay and, on the other, the influence of open waters from the Ross Sea continental shelf. A circular hole with a diameter of 1.5 m was drilled through the landfast sea ice, above which a fish hut was installed to provide shelter for sampling operations. At this station, vertical profiles of temperature, salinity, and fluorescence were obtained using a CTD profiler (SBE 19plus, Sea-Bird Electronics, Communication Technology, USA) equipped with a fluorometer. Discrete water samples were collected with a 10-L Niskin bottle (Ocean Test Equipment) at selected depths to complement the CTD measurements. Temperature, salinity, and fluorescence values were acquired along the entire water column, whereas discrete samples for phytoplankton community structure and nutrient concentration analyses were collected at the defined depths (0, 2, 5, 15, 30, 50 and 100 m). These predefined depths were selected to provide consistent temporal sampling of the immediate ice–water interface and progressively deeper water-column layers, allowing comparison of near-surface, intermediate, and deep conditions across the season. The discrete sampling was therefore designed as a standardized depth framework rather than as a CTD-feature-triggered sampling strategy. A PAR sensor (LI-190R, LI-COR Biosciences, Lincoln, NE, USA) was mounted on the roof of the fish-hut container to continuously monitor the incident irradiance. Measurements were recorded every 30 seconds, resulting in a continuous time series from 10 November to 12 December 2015. These data provided an accurate characterization of the light regime during the initial part of the sampling period. To investigate the sympagic communities at the landfast-ice bottom, ice cores were collected around the fish hut using a Kovacs corer (Kovacs Ice Coring System, Kovacs Enterprise, Roseburg, OR, USA) 0.1 m diameter. Each core was sectioned into 10-cm intervals starting from the ice–water interface. The basal 10 cm were designated as the bottom brown (BB) layer, whereas the overlying 10 cm were designated as the bottom white (BW) layer. Platelet Ice Water (PIW) refers to the water immediately beneath the basal ice within the platelet-ice layer and was sampled separately from the ice sections.

After melting in the dark at +4 °C, meltwater from each section was subsampled for analyses of total biomass, nutrient concentrations, and pigment composition. From the same boreholes used for ice coring, discrete water samples were additionally collected to characterize the phytoplankton community structure in the water column immediately beneath the ice. Sampling was performed at depths of 0, 2, and 5 m using a peristaltic pump, while supplementary samples were obtained from the main hole with a 10-L Niskin bottle (Ocean Test Equipment) at 15, 30, 50, and 100 m. This combined approach allowed the simultaneous assessment of sympagic assemblages at the ice–water interface and the phytoplankton communities in the underlying water column. At each of the 24 sampling dates, three sea-ice cores were collected at the sampling location. The three cores were assigned to different analytical components of the study: one core was used for zooplankton analyses, one for taxonomic analyses, and one for the analyses presented in this study. For the latter, each core was sectioned into 10-cm intervals starting from the ice–water interface, with the basal 10 cm designated as the bottom brown (BB) layer and the overlying 10 cm as the bottom white (BW) layer. Platelet Ice Water (PIW) was sampled separately from the water immediately beneath the basal ice. Samples from BB, BW, and PIW were processed and analyzed individually, without pooling among sampling dates or between compartments. Because one independent ice core was available per sampling date for the variables analyzed in the present study, the sea-ice time series were evaluated descriptively, and no inferential statistical tests were applied to individual sampling dates.

Landfast-ice temperature

Miniature thermocouples and resistance temperature detectors (RTDs; Thermo Electric Instrumentation, The Netherlands) were deployed within the ice column to monitor vertical temperature profiles. Sensors were inserted starting from the ice surface at depths of 5, 60, 120, 180, and 245 cm, the latter corresponding to the bottom of the ice. Temperature data were acquired continuously at 10-minute intervals throughout the day and night using the SOPHIA (Sensing On Pervasive Hard sItuAtion) sensor cluster and data acquisition system, a patented autonomous monitoring system developed by the Stazione Zoologica Anton Dohrn. In addition to storing measurements on an in situ datalogger, the system transmitted temperature data in near real time to Mario Zucchelli Station, enabling continuous remote monitoring of the sea-ice thermal regime throughout the field campaign. This configuration provided high-resolution records of thermal gradients across the ice during the sampling period.

Total phytoplankton biomass and pigmentary spectra composition

Total phytoplankton biomass was estimated using fluorometrically determined chlorophyll-a (Chl-a) concentration as a proxy. For fluorometric Chl-a determination, seawater samples (50–500 mL, depending on the amount of biomass in the water) were filtered onto Whatman GF/F glass-fiber filters (25 mm diameter, 0.7 µm pore size, Cytiva, USA). Filters were rapidly frozen at −80 °C and subsequently processed in Italy. Pigments were extracted in 90% acetone following Holm-Hansen et al. (1965), and Chl-a concentrations were determined fluorometrically using a Shimadzu RF-6000 spectrofluorometer (Shimadzu Corporation, Japan), calibrated daily against a Chl-a standard solution (Sigma-Aldrich).

Phytoplankton pigment composition was determined independently by HPLC on separate seawater aliquots. For HPLC analysis, seawater volumes ranging from 50 mL to 2 L were filtered onto Whatman GF/F glass-fiber filters (47 mm diameter, 0.7 µm pore size, Cytiva, USA), with larger volumes used when phytoplankton biomass was low. Filters were rapidly frozen at −80 °C and subsequently processed in Italy. Pigment extraction and HPLC analysis followed the method described by Vidussi et al. (1996), using methanol and methanol–ammonium acetate (70:30) as mobile phases and applying the chromatographic gradient described in that study. Pigment separation was performed using an Agilent 1100 HPLC system (Agilent Technologies, Santa Clara, CA, USA) equipped with an HP 1050 photodiode array detector and an HP 1046A fluorescence detector for the determination of chlorophyll degradation products. External pigment standards were provided by the International Agency for 14C Determination–VKI Water Quality Institute and were used for instrument calibration, pigment identification, and quantification. The pigments considered were chlorophyll c2 (Chl c2), Mg-2,4-divinyl pheoporphyrin (MGDVP), peridinin (Perid), prasinoxanthin (Prasino), fucoxanthin (Fuco), 19′-hexanoyloxyfucoxanthin (19′HF), violaxanthin (Viola), diadinoxanthin (Diadino), antheraxanthin (Anterax), alloxanthin (Allox), diatoxanthin (Diato), zeaxanthin (Zeax), lutein (Lut), chlorophyll b (Chl b), β-carotene (β-Car), and Chl-a. The HPLC method had a reported detection limit of approximately 1 ng L - ¹ (0.001 µg L - ¹), with a coefficient of variation below 3% among replicate injections of pigment standards. Thus, fluorometric Chl-a measurements were used as a proxy for total phytoplankton biomass, whereas HPLC-derived pigment concentrations from separate filtered samples were used to characterize pigment composition and relative pigment-associated phytoplankton group signatures.

Macronutrient concentrations

Macronutrient concentrations (NO 3 -, NO 2 -, NH 4 +, Si(OH) 4, and PO 4 ³ -) were determined from seawater samples collected directly from Niskin bottles. Aliquots of 20 mL were transferred into low-density polyethylene (LDPE) vials and preserved at −20 °C until subsequent laboratory analysis. Nutrient determinations were carried out using a five-channel continuous-flow autoanalyzer (Technicon Autoanalyser II; Systea S.p.A. Analytical Technologies, Italy), following the protocol of Hansen and Grasshoff (1983) with adaptations to modern instrumentation. Analytical performance was assessed using certified reference material (CRM) MOOS-3, and measured nutrient concentrations did not differ significantly from the certified values (p > 0.05). As a reference for the analytical performance of the instrument, published detection limits for nutrient analyses performed using a Technicon Autoanalyser II in marine waters were 0.06 µmol L - ¹ for NO 3 - + NO 2 -, 0.02 µmol L - ¹ for PO 4 ³ -, 0.20 µmol L - ¹ for Si(OH) 4, 0.01 µmol L - ¹ for NO 2 -, and 0.05 µmol L - ¹ for NH 4 +. These values were determined from replicate artificial seawater blanks and are reported here as literature-based reference values rather than as detection limits specifically determined for the present dataset.

Iron experiment

To assess the effects of iron availability and UV radiation on phytoplankton growth and photophysiology, a 96-h incubation experiment was conducted using seawater collected at 10 m depth at Station C, in the absence of landfast ice. A total of 60 L of seawater was collected using a peristaltic pump and pre-filtered through a 200-μm mesh to remove potential zooplanktonic grazers. The seawater was subsequently distributed into twenty-four 2-L incubation bottles, consisting of 12 polycarbonate bottles, which excluded UV radiation, and 12 Teflon bottles, which allowed UV radiation to reach the samples. The bottles were incubated in tanks covered with a neutral-density filter to simulate approximately 15% of incident surface irradiance.

The experiment consisted of three iron treatments: Control (no addition; CTRL), Low Iron-DFB (100 nM desferrioxamine B [DFB]), and High Iron-dFe (3 nM FeNO 3; dFe). Each iron treatment was established in duplicate under each UV condition, resulting in a 3 × 2 factorial design, with three iron treatments (CTRL, DFB, and dFe) and two UV conditions (−UV and +UV). For each treatment combination, independent bottles were assigned to the two sampling times, T1 (48 h) and T2 (96 h), with each bottle sampled destructively at its assigned time point. Thus, T1 and T2 represented independent experimental units rather than repeated measurements of the same bottles. This sampling design applied to all variables measured during the incubation experiment, including fluorometric chlorophyll-a (Chl-a), maximum quantum efficiency of photosystem II (Fv/Fm), and HPLC pigment composition. Samples collected at T0 represented the initial condition before incubation, whereas samples collected at T1 and T2 represented independent experimental units. Chlorophyll a concentration was measured fluorometrically as an estimate of phytoplankton biomass, while Fv/Fm was measured as an indicator of the maximum quantum efficiency of photosystem II. For both variables, the effects of iron treatment (CTRL, DFB, and dFe), UV exposure (−UV and +UV), and their interaction were evaluated separately at T1 (48 h) and T2 (96 h) using two-way factorial ANOVA. Because bottles assigned to T1 and T2 were independent experimental units and were sampled destructively, the two time points were not treated as repeated measurements. Replicate bottles within each treatment combination were considered independent experimental units (n = 2). T0 measurements were considered baseline observations and were not included in the factorial analyses. Given the limited replication, the statistical results were interpreted cautiously. HPLC pigment analysis was used to characterize pigment composition and pigment-associated phytoplankton groups.

Maximum quantum efficiency of photosystem II (Fv/Fm) was measured using a PHYTO-PAM fluorometer (Walz, Germany) with Phyto-Win software. Samples were dark-adapted for ≥30 min prior to measurement. Baseline blanks were matched to sample salinity, and instrument gain was adjusted for each measurement run. Fv/Fm was calculated as (Fm − F0)/Fm from three consecutive saturating pulses. Fv/Fm measurements were also obtained from selected field samples collected between December and January using the same measurement protocol and were used for descriptive purposes only.

Statistical analysis

All pigment variables were expressed as pigment:chlorophyll a ratios to account for biomass differences among samples. These ratios were normalized using a min–max transformation to ensure comparable scaling across pigments. All variables, including pigment ratios and environmental parameters, were subsequently standardized using z-score transformation prior to multivariate analysis. The min–max step placed the pigment ratios on a common 0–1 scale, whereas subsequent z-score standardization centered and scaled the variables for multivariate analysis. Principal Component Analysis (PCA) was used to identify covariation among environmental conditions, phytoplankton biomass, and pigment-associated signals and to visualize how these multivariate profiles varied along the water-column depth gradient. The PCA was not intended to provide a direct taxonomic classification and was performed on the standardized matrix. For multivariate analyses, only sampling dates for which complete data were available for all variables included in the analysis were retained, resulting in 13 sampling dates. Because samples were repeatedly collected at the same station over time, sampling date was included as a blocking factor in the permutation scheme, with permutations restricted within sampling dates to account for the temporal structure of the dataset. Differences among depth layers were statistically tested using PERMANOVA based on Euclidean distance, applied to the same standardized multivariate matrix used for the PCA. The analysis included salinity, temperature, NH 4 +, NO 2 -, NO 3 -, PO 4 ³ -, Si(OH) 4, chlorophyll a, phaeopigments, pigment:chlorophyll a variables, diagnostic pigments (Chl c3, Chl c2, 19′HF, lutein, fucoxanthin), particulate organic nitrogen and carbon, and C:N ratio. A total of 9,999 permutations were used. To assess whether differences in multivariate dispersion could contribute to the PERMANOVA result, a PERMDISP analysis was also performed. For the incubation experiment, the effects of iron treatment, UV exposure, and their interaction on chlorophyll a and Fv/Fm were evaluated separately at T1 (48 h) and T2 (96 h) using two-way factorial ANOVA.

Sea ice

Temperature

Temperature increased over time at every depth within the sea-ice (Figure 2).

The shallowest layer showed strong fluctuations, likely due to atmospheric influences, while layers within the ice showed very little variation. At 5 cm depth, temperatures initially varied between the overall minimum of -5.53 °C and just below 0 °C. After November 29, fluctuations became more pronounced with peaks above 0 °C and troughs below -5.00 °C. The magnitude of variation increased significantly later in the sampling season, with several short-lived heating events. The most extreme occurred on December 10, when temperatures reached an overall maximum of 14.69 °C. Afterwards values varied between -3.28 °C and 4.41 °C for the rest of the sampling season. Temperatures at 60 cm and 120 cm depths were nearly identical at the start of the sampling period, and while both increased linearly over time, the rise was more rapid at 60 cm. At this depth, temperature began at -4.53 °C on November 23 and rose to a maximum of -1.31 °C by December 12. At 120 cm, temperatures started at -4.63 °C and reached -2.47 °C by the end of the sampling period. The initial temperature at 180 cm was slightly warmer at -3.97 °C, but the increase over time was more gradual, peaking at -2.56 °C on December 12, a value similar to that recorded at 120 cm on the same date. Temperatures at the 245 cm ice-water interface remained highly stable and increased ever so slightly from a minimum of -1.97 °C near the beginning of the sampling period on November 25, to a maximum of -1.75 °C on December 11.

Temporal evolution of chlorophyll-a and pigment:chl a ratios in bottom ice layers

Chlorophyll-a concentrations in the bottom brown (BB) layer were consistently higher than those in the bottom white (BW) layer throughout the sampling period (14 November 2015 to 10 January 2016). In the BB layer, chlorophyll-a peaked between 16 and 22 November, reaching maximum values close to 1800 µg L - ¹ (Figure 3).

After this peak, concentrations declined progressively, stabilizing below 400 µg L - ¹ by early January. In contrast, the BW layer exhibited lower chlorophyll-a levels, remaining below 600 µg L - ¹ throughout the time series. The highest values in BW were recorded in mid-November, reaching approximately 500 µg L - ¹, followed by a gradual decrease similar in trend to BB but with reduced magnitude.

Sea ice pigment signature

The evolution of the sea-ice pigment signature is shown in Figure 4.

The BW layer displayed greater temporal variability in pigment composition. Fucoxanthin remained the dominant pigment, with normalized values generally below ~0.30, but showing marked fluctuations across the sampling period, reflecting dynamic fucoxanthin-associated pigment signals near the ice–water interface. Chlorophyll c3 exhibited pronounced isolated peaks, including a normalized value of 1 on 18/11/2015, indicating short-lived chromophyte pulses. Chlorophyll c2 was consistently present at low-to-moderate normalized levels (typically <0.40), with occasional increases. Minor pigments were nearly absent: pheophorbide a appeared only once, with a low normalized value (~0.06), suggesting localized degradation processes. Lutein and 19’HF were not detected in the BW layer.

The BB layer exhibited a consistently fucoxanthin-associated pigment profile, with fucoxanthin showing moderate normalized values (generally between ~0.23 and ~0.46). Chlorophyll c2 and chlorophyll c3 were regularly present as accessory pigments, with Chl c2 typically ranging between ~0.25 and ~0.44, and Chl c3 between ~0.04 and ~0.28. These patterns are consistent with a stable contribution of fucoxanthin- and chlorophyll-c-associated pigment groups throughout the season. Minor pigments appeared sporadically: lutein was detected only on a few dates, with normalized values up to ~1, suggesting episodic inputs from chlorophytes or euglenophytes. Pheophorbide a was present exclusively on 23/12/2015, with a normalized value of ~0.12, indicating localized chlorophyll degradation. 19’HF was absent throughout the BB layer.

The PIW layer exhibited the highest overall normalized pigment intensities. Fucoxanthin frequently reached values close to 1 (e.g., 26/11/2015), showing a strong and sustained fucoxanthin-associated pigment signal in the upper ice strata. Chlorophyll c2 and chlorophyll c3 were both abundant, with normalized values often exceeding 0.7–0.8 on several dates, showing substantial chlorophyll c2 and c3 signals. Minor pigments showed sporadic but ecologically relevant signals: pheophorbide a reached normalized values up to ~0.07 on a few dates, consistent with episodic chlorophyll degradation. Lutein appeared only once, with a moderate normalized value (~0.24), consistent with a brief lutein-associated pigment signal. 19’HF was detected twice, with normalized values of 1 and ~0.33, consistent with occasional contributions from 19′HF-associated phytoplankton groups.

Water column

The thermohaline profile of the water column for the 0–5 m and 0–20 m layers is shown in Figure 5.

The temporal evolution of salinity varied notably with depth. Surface layers were strongly influenced by melting inputs, while deeper layers were characterized by almost constant values. These differences were absent before November 14, when salinity remained uniform (34.744 ± 0.0117) throughout the water column. At 0 m, there was an increase during the first part of the sampling period, likely due to the contributions of brine channels at the beginning of the melting phase, with values that reached a maximum of 36.6672 on December 9. After December 11, values rapidly decreased to a minimum of 30.6672 on December 30 and then fluctuated widely between 31.4432 and 33.9280 for the rest of the sampling period. A similar trend was observed at 2 m, although with less pronounced variations. Values ranged between 32.9022 and 35.4202, with a slight increase beginning November 30 and peaking on December 4. A shorter period of decline followed, reaching a minimum on December 14, almost two weeks before the surface layer. The same initial period of increase was observed at the 3 m depth, beginning November 30 and ending at the maximum value of 35.226 on December 4. The subsequent decrease was far less drastic compared to the overlying depths, ending at the overall minimum of 34.4118 on December 16, followed by variations between 35.6472 and 34.4268 for the rest of the sampling period.

The 4 m, 5 m and 6 m sampling depths showed similar profiles with salinity steadily declining from December 19 to a minimum on January 6. At 4 m, values ranged between 34.4981 and 35.0538; At 5 m, from 34.5126 and 34.9986; and at 6 m, from 34.5366 and 34.9403.

Salinity was more stable at deeper sampling depths. At 15 m, values averaged 34.72174 ± 0.0448 with a slight decline after December 19. At 30 m, values averaged 34.73336 ± 0.0410 with brief dips to 34.6051 on November 17 and 34.4985 on November 28. Salinity remained highly stable at 50m, 100m and 200m depths, with similar averages of 34.784 ± 0.0201, 34.754 ± 0.0156 and 34.758 ± 0.0157, respectively.

The evolution of temperature varied with depth, with a unique profile at 0 m, similar patterns from 2-6m, and distinct deeper layers. Temperature was relatively uniform from 2–200 m at the beginning of the sampling season, averaging -1.863 ± 0.068 °C from November 13 to November 24.

The surface layer showed a distinct profile compared to deeper depths. Temperatures rose gradually from November 13-30, ranging between -1.9059 °C and -1.2776 °C. Afterward, fluctuations became more pronounced, and the overall increase accelerated, reaching a maximum of -0.5821 °C on December 9. Although values continued to fluctuate, the upward trend stopped, stabilizing between -1.6062 °C and -0.6089 °C for the rest of the sampling period. At 2 m, temperatures followed a similar trend from November 13 to December 16 albeit with less pronounced variations. This was followed by a rapid rise until the end of the sampling season, peaking at 0.1379 °C on January 8.

The 3–6 m depths exhibited similar profiles, with a steady increase from November 13 to December 4, followed by a sharp rise that peaked at the end of the sampling season on January 8. Maximum values increased with depth. Temperatures ranged between -1.9049 °C and 0.3431 °C at 3 m, -1.9027 °C to 0.5313 °C at 4 m, -1.9026 °C to 0.6102 °C at 5 m, and -1.9029 °C to 0.6494 °C at 6 m.

At 15 m, temperatures increased steadily until December 11, followed by a slower rise through December 23. A sharp increase then occurred, peaking at 0.7242 °C on January 8, the highest temperature recorded at any depth. At 30 m, values rose from a minimum of -1.9065 °C on November 13, with a more rapid increase between December 23 and December 30. Unlike shallower depths, the increase then slowed, reaching a maximum of 0.2215 °C on January 8. Similarly, at 50 m, temperatures began at -1.070 °C on November 13, rising sharply from December 26 to a peak of 0.0229 °C on December 30. Afterward, values stabilized, fluctuating between -0.0606 °C and -0.4336 °C for the remainder of the sampling period.

At 100 m, temperatures increased overall but at a much slower rate than at shallower depths, starting at -1.9028 °C on November 13 and reaching a peak of -0.9265 °C on December 30. Values fluctuated between -1.7225 °C and -1.3686 °C for the remainder of the sampling season. At 200 m, temperatures remained the most stable overall, ranging from -1.9174 °C to -1.8577 °C between November 14 and November 30, followed by a slight rise to a maximum of -1.6617 °C on December 12. Unlike the 2–100 m depths, temperatures at 200 m declined overall for the rest of the season, varying between -1.9265 °C and -1.7572 °C.

Nutrient concentrations (NO 2 -, PO4 3 -, SiO 4, NO 3 -, NH 4 +)

Nitrite (NO 2 -) and phosphate (PO 4 ³ -) remained low and showed negligible temporal or vertical variability throughout the sampling period (Figure 6). Nitrite ranged between 0.04 and 0.08 µmol L - ¹ across all depths, while phosphate remained within ~1.2–1.5 µmol L - ¹.

The absence of distinct peaks indicates relatively limited temporal variability in these nutrients during the sampling period, although this pattern alone does not allow biological processing to be excluded. In contrast, dissolved silica (Si(OH) 4) and nitrate (NO 3 -) exhibited pronounced variability, with coherent fluctuations across depths.

In the 0–5 m layer, Si(OH) 4 and NO 3 - showed synchronous oscillations, characterized by early-December minima, mid-December maxima, and sharp declines in early January. At 0 m, Si(OH) 4 decreased from 43.4 to 32.8 µmol L - ¹ (4–6 Dec), peaked at 63.9 µmol L - ¹ (14 Dec), and reached a minimum of 28.1 µmol L - ¹ (6 Jan). Nitrate followed the same pattern, with a minimum of 12.5 µmol L - ¹ (6 Dec), a peak of 24.1 µmol L - ¹ (12 Dec), and a January minimum of 12.0 µmol L - ¹ (4 Jan). Similar dynamics were observed at 2 and 5 m, although the 5 m layer showed earlier and more abrupt peaks (e.g., Si(OH) 4 reaching 59.3 µmol L - ¹ on 10 Dec, followed by a rapid drop to 11.7 µmol L - ¹ on 12 Dec).

In 15–30 m layer, Si(OH) 4 and NO 3 - displayed the same sequence of mid-December maxima followed by sustained declines. Si(OH) 4 peaked at 63.1 µmol L - ¹ (12 Dec) at 15 m and at 71.4 µmol L - ¹ (16 Dec) at 30 m, the latter representing the highest value in the upper 30 m. Nitrate maxima occurred concurrently (25.0 µmol L - ¹ at 15 m and 30.5 µmol L - ¹ at 30 m), followed by sharp decreases to early-January minima (7.8 and 7.4 µmol L - ¹, respectively).

At 100 m, Si(OH) 4 and NO 3 - exhibited the strongest amplitude of variation. Si(OH) 4 increased from 57.4 µmol L - ¹ (12 Dec) to 99.2 µmol L - ¹ (16 Dec), the highest concentration recorded, before declining to 37.4 µmol L - ¹ (4 Jan). Nitrate showed a parallel rise to 36.4 µmol L - ¹ (16 Dec) and a subsequent decline to 8.8 µmol L - ¹ (19 Jan).

Ammonium concentrations were generally low but showed episodic increases that often coincided with Si(OH) 4 and NO 3 - minima. Surface NH 4 + peaked at 5.8 µmol L - ¹ (8 Dec), while deeper layers exhibited smaller but recurrent maxima (e.g., 3.7 µmol L - ¹ at 5 m on 19 Dec; 4.5 µmol L - ¹ at 30 m on 19 Jan). At 100 m, NH 4 + remained comparatively stable (1.2–2.6 µmol L - ¹), with only a modest peak on 14 Dec.

POC, PON and C:N ratio

In the 0–5 m layer, POC and PON showed the highest variability of the water column (Figure 7). At 0 m, both parameters were elevated and highly variable between 4–14 December, followed by sharp declines to minima on 23 December and subsequent stabilization. PON ranged from 65.4 to 292.1 µmol L - ¹, while POC varied from 379.5 to 2292.4 µmol L - ¹ before stabilizing at 230–477 µmol L - ¹. The C:N ratio fluctuated between 5.5 and 8.7 during early December and stabilized at ~5.7 in January.

At 2 m, POC and PON increased steadily throughout the sampling period (PON: 9.3–118.1 µmol L - ¹; POC: 57.4–669.7 µmol L - ¹), while the C:N ratio remained stable (5.3–6.4).

At 5 m, early-season values were stable, followed by increased variability after 12 December. PON ranged from 22.3 to 83.3 µmol L - ¹, and POC from 129.3 to 500.5 µmol L - ¹. The C:N ratio remained between 5.3 and 6.1 until a sharp peak on 23 December (10.38), then stabilized (5.9–6.1).

In the 15–30 m layer, POC and PON increased through mid-December and stabilized thereafter, with moderate C:N variability. At 15 m, PON increased from 18.8 to 97.1 µmol L - ¹ (30 Dec), then stabilized at 46.8–63.4 µmol L - ¹. POC rose from 107.2 to 556.5 µmol L - ¹, stabilizing at 288–382 µmol L - ¹. The C:N ratio increased slightly from 5.23 to 6.72. At 30 m, PON increased from 10.4 to 73.5 µmol L - ¹ (30 Dec), then stabilized at 43.2–57.5 µmol L - ¹. POC rose from 74.5 to 465.5 µmol L - ¹, stabilizing at 300–354 µmol L - ¹. The C:N ratio decreased from 7.14 to 5.12 (12 Dec), then gradually increased to 6.95 (Figure 7). At 40 m, only one sampling date was available (8 Dec), with values (PON = 33.8 µmol L - ¹; POC = 183.0 µmol L - ¹; C:N = 5.41) consistent with adjacent depths.

In the 50–100 m layer, POC and PON showed late-December increases followed by stabilization, with moderate C:N variability. At 50 m, PON remained stable (19.6–31.4 µmol L - ¹) until late December, then increased to 80.4 µmol L - ¹ (30 Dec) and 97.1 µmol L - ¹ (8 Jan). POC increased from 137–213 µmol L - ¹ to 618.4 µmol L - ¹ (30 Dec) and 720.3 µmol L - ¹ (8 Jan). The C:N ratio rose from 5.36 to 7.69 (30 Dec), stabilizing between 6.0 and 7.4. At 100 m, PON remained low and stable (16–20 µmol L - ¹) until late December, then peaked at 71.4 µmol L - ¹ (30 Dec) and fluctuated between 17 and 43 µmol L - ¹ thereafter. POC decreased from 225.8 to 128.6 µmol L - ¹ (26 Dec), then increased to 297.1 µmol L - ¹ (30 Dec). The C:N ratio increased from 6.93 to 10.30 (14 Dec), then stabilized between 6.1 and 7.8.

Chlorophyll a and pigment concentrations

Chlorophyll a concentrations showed strong vertical gradients and substantial temporal variability across the water column during the sampling period (Figure 8).

The highest values were consistently recorded at the surface (0 m), where chlorophyll a ranged from <1 µg L - ¹ during late December and January to peaks exceeding 40 µg L - ¹ in early December. This layer exhibited rapid fluctuations, with alternating phases of elevated and reduced biomass, reflecting the highly dynamic conditions immediately beneath the landfast ice. At 2 m depth, chlorophyll a concentrations were generally lower than at the surface but followed a similar temporal pattern. Values typically ranged between 0.03 and 6 µg L - ¹, with moderate increases occurring during the same early-summer period characterized by elevated surface biomass. Although the magnitude of variability was reduced compared to 0 m, the upper 2 m behaved as a coherent unit, with biomass pulses detectable throughout this layer.

At 5 m, chlorophyll a concentrations ranged from <0.05 to approximately 4.5 µg L - ¹. The temporal evolution resembled that of the shallower depths, with modest increases during the early-summer peak followed by a progressive decline toward uniformly low values in late December and January. Variability at this depth was less pronounced, indicating partial decoupling from the highly dynamic surface layer.

Intermediate depths (15–30 m) exhibited chlorophyll a concentrations generally between 0.3 and 6 µg L - ¹. Although sampling frequency was lower than in the upper layers, these depths showed clear evidence of mid-December increases, followed by a decline toward the end of the month. The presence of these peaks indicates that biomass enhancements were not restricted to the surface but extended into the mid-water column. Multivariate analyses were used separately to evaluate covariation in pigment and environmental variables. The overall PERMANOVA detected a significant vertical effect in the multivariate dataset (F = 1.7188, R² = 0.1629, p = 0.0003), with sampling date included as a blocking factor in the permutation scheme. This result indicates significant vertical structuring of the combined environmental, biomass, pigment and elemental variables. The PERMANOVA does not by itself establish discrete taxonomic communities. At deeper levels (50–100 m), chlorophyll a concentrations were lower and less variable, typically ranging between 0.4 and 5 µg L - ¹. Despite the reduced amplitude of variation, these depths also displayed increases during mid-December, suggesting that deeper layers responded to the same seasonal signal observed in the upper water column. After this period, chlorophyll a values decreased and remained low through January.

Overall, chlorophyll a concentrations revealed a vertically structured system characterized by a pronounced early-summer biomass maximum in the upper 2–5 m, with attenuated but coherent signals extending to 30 m and detectable responses even at 50–100 m. The subsequent decline across all depths toward late December and January indicates a progressive reduction in phytoplankton biomass throughout the water column as the season advanced.

Normalized pigment concentrations revealed vertical structuring of the pigment profile from the surface down to 100 m (Figure 9). Among all pigments, fucoxanthin frequently exhibited the highest normalized values, indicating a sustained fucoxanthin-associated signal across the observation period. The strongest signal occurred at the surface (0 m), where fucoxanthin frequently approached its maximum normalized value, with pronounced peaks in mid-November and early December. At 2 and 5 m, fucoxanthin remained elevated but showed greater variability, consistent with changes in the hydrographic conditions near the ice–water interface. Between 15 and 30 m, normalized fucoxanthin values remained relatively high and stable, whereas at 50 m fucoxanthin decreased slightly and co-occurred more frequently with other accessory pigments. At 100 m, fucoxanthin maintained moderate normalized values, indicating that fucoxanthin-associated pigments remained an important component of the measured pigment pool at depth.

Although generally low, normalized concentrations of 19′-hexanoyloxyfucoxanthin (19′HF) were ecologically relevant. This pigment appeared intermittently at 0 m and more consistently between 2 and 30 m, with distinct increases during late December. These changes are consistent with episodic increases in 19′HF-associated phytoplankton groups during periods of lower surface salinity, but the pigment data alone do not establish a specific taxonomic contribution. Below 50 m, 19′HF was minimal or absent, indicating that this pigment-associated signal was concentrated in the upper water column.

Accessory chlorophylls supported this pattern. Chl c2 showed consistently high normalized values across all depths, with maxima at the surface and slightly lower but stable values between 10 and 50 m. Chl c3 exhibited episodic peaks at 0 and 2 m and moderate values down to 30 m, indicating changes in chlorophyll-c-associated pigment signals under stratified conditions. Minor pigments such as prasinoxanthin and lutein occurred sporadically within the upper 5 m, coinciding with periods of reduced salinity; these observations should be interpreted as pigment-associated signals rather than direct taxonomic identifications.

Multivariate analysis

To assess the overall multivariate structure of the water-column dataset, a PCA was performed on the standardized matrix including environmental, biomass, elemental, and pigment-associated variables. The PCA revealed a structured distribution of samples along the depth gradient, with PC1 and PC2 explaining 22.00% and 20.65% of the total variance, respectively (Figure 10). The corresponding PERMANOVA identified a significant effect of depth on the multivariate dataset (F = 1.7188, R² = 0.1629, p = 0.0003). Sampling date was accounted for as a temporal blocking factor in the permutation scheme. Multivariate dispersion did not differ significantly among depths (PERMDISP: F = 1.5682, p = 0.1747), indicating that the significant PERMANOVA result was not attributable to differences in within-group dispersion. These results indicate vertical structuring of the combined environmental, biomass, pigment, and elemental profiles, but do not by themselves establish discrete taxonomic communities.

Effect of UV and iron limitation experiment

Chlorophyll a concentration

The experiment was conducted over three time points: T0 (initial conditions), T1 (after 48 h), and T2 (after 96 h). At T0, chlorophyll a concentration was approximately 0.56 μg L−1 across the experimental treatments (Figure 11). At T1, mean chlorophyll a concentrations were 0.864 ± 0.033 μg L−1 (CTRL), 0.976 ± 0.019 μg L - ¹ (CTRL-UV), 1.131 ± 0.066 μg L−1 (dFe), 0.944 ± 0.130 μg L−1 (dFe-UV), 0.923 ± 0.013 μg L−1 (DFB), and 0.823 ± 0.002 μg L−1 (DFB-UV). The two-way ANOVA indicated a significant effect of iron treatment (F 2, 6 = 7.564, p = 0.0229) and a significant iron × UV interaction (F 2, 6 = 6.207, p = 0.0346), whereas the main effect of UV exposure was not significant (F 1, 6 = 2.701, p = 0.1514).

At T2, mean chlorophyll a concentrations were 1.542 ± 0.185 μg L−1 (CTRL), 1.261 ± 0.060 μg L−1 (CTRL-UV), 1.603 ± 0.080 μg L−1 (dFe), 1.506 ± 0.000 μg L−1 (dFe-UV), 1.240 ± 0.004 μg L−1 (DFB), and 1.054 ± 0.025 μg L−1 (DFB-UV). At this time point, the ANOVA showed significant main effects of iron treatment (F 2, 6 = 22.597, p = 0.0016) and UV exposure (F 1, 6 = 14.117, p = 0.0094), whereas the iron × UV interaction was not significant (F 2, 6 = 1.117, p = 0.3868). Thus, the statistical pattern differed between the two sampling times, with an iron × UV interaction detected at T1 but not at T2. The relatively low natural irradiance during the incubation period, due to persistent cloud cover, further warrants cautious interpretation of the UV-related responses.

Maximum quantum efficiency (Fv/Fm)

All treatments shared the same baseline Fv/Fm value at T0 (0.4305). At T1 (48 h), mean Fv/Fm values were 0.4583 in CTRL, 0.4398 in CTRL-UV, 0.5469 in DFB, 0.4498 in DFB-UV, 0.5247 in dFe, and 0.5357 in dFe-UV. At T2 (96 h), mean Fv/Fm values were 0.6122 in CTRL, 0.6248 in CTRL-UV, 0.5203 in DFB, 0.4752 in DFB-UV, 0.5367 in dFe, and 0.5934 in dFe-UV.

Two-way factorial ANOVA performed separately at T1 and T2 did not detect significant effects of iron treatment, UV exposure, or their interaction. At T1, the effects of iron treatment (F 2, 6 = 3.02, p = 0.124), UV exposure (F 1, 6 = 1.64, p = 0.247), and the iron × UV interaction (F 2, 6 = 1.41, p = 0.315) were not significant. At T2, the effects of iron treatment (F 2, 6 = 1.87, p = 0.234), UV exposure (F 1, 6 = 0.025, p = 0.880), and the iron × UV interaction (F 2, 6 = 0.33, p = 0.730) were likewise not significant.

Despite the absence of statistically significant effects, Fv/Fm showed different temporal trajectories among treatment combinations. Fv/Fm increased from T1 to T2 in CTRL, CTRL-UV, and dFe-UV, whereas DFB showed a decline and DFB-UV showed a smaller increase. At T2, the highest mean Fv/Fm was observed in CTRL-UV (0.6248), followed by CTRL (0.6122) and dFe-UV (0.5934). These treatment-specific patterns are descriptive, given the limited replication (n = 2 per treatment combination).

Our observations indicate that seasonal changes in hydrography and pigment composition were closely coupled at the ice–water interface. The most notable features were the development of a low-salinity surface layer, strong temporal variability in near-surface chlorophyll a, and changes in the relative contribution of diagnostic pigments. These patterns are consistent with the view that freshwater-driven stratification can modify the physical and biogeochemical environment experienced by under-ice phytoplankton (Bolinesi et al., 2020; Saggiomo et al., 2021). Antarctic ice mass loss is documented over the satellite-observation period, but the present study does not directly quantify its contribution to the local freshwater budget (NASA, 2023). Enhanced freshwater inputs may intensify stratification beneath landfast ice, potentially favoring phytoplankton responses that differ from the typically strong diatom-associated signal described for Antarctic coastal and ice-associated environments (Riaux-Gobin et al., 2003; Saggiomo et al., 2017; van Leeuwe et al., 2018; Zigic et al., 2026).

Rising temperatures may contribute to changes in ice porosity and permeability, which could influence the conditions experienced by ice-associated microalgae. At the same time, increased meltwater inputs may alter nutrient ratios, modify species composition, or change the physical properties of the sub-ice platelet layer, including its freezing point (Jeffries et al., 1993; Gow et al., 1998; Smith et al., 2001; Fraser et al., 2023). Such changes could profoundly affect energy transfer to higher trophic levels, interactions between ice and benthic communities, and competitive dynamics within a system characterized by short-term pulses of massive energy production and highly efficient carbon flow.

The temporal evolution of ice temperature highlights the role of landfast ice thickness in attenuating heat transfer from the surface to the water column, as evidenced by the damping of thermal curves (Figure 2). Incident irradiance increased steadily from mid-November to mid-December, reflecting the seasonal rise in solar elevation toward the austral summer solstice. Peak PAR values were recorded between 8 and 12 December 2015, approaching 1700 µmol photons m - ² s - ¹ (Supplementary Table S1). This pattern reflects both the expected summer trend and short-term variability driven by local atmospheric conditions. The absence of irradiance data after 12 December represents a limitation of this study, preventing full characterization of the light regime until the end of the sampling campaign (19 January 2016). Nevertheless, the available time series provides a robust description of early- and mid-summer light conditions, supporting the interpretation of hydrographic and biological patterns observed in the same period. Future campaigns should ensure uninterrupted irradiance monitoring to capture the complete seasonal variability and its ecological implications.

The intensification of melting processes led to an increase in salinity in the first decade of December 2015 within the upper 50 cm of the water column due to salt extrusion, followed by a marked decrease as a consequence of freshwater inputs (Figure 5). These changes were associated with the development of a cold, low-salinity surface layer, beneath which salinity values increased sharply. Although isotopic data are lacking to trace the precise origin of freshwater inputs, the persistence of landfast ice—whose thickness decreased by only 10 cm during the sampling period—and direct observations (e.g., freshwater streams flowing into Tethys Bay, less than 1 km from station C) are consistent with an important contribution of continental freshwater, although the present observations cannot quantitatively separate continental meltwater from sea-ice melt or other freshwater sources. During the initial phase of the sampling period, up to late November 2015, the water column exhibited homothermal and homohaline conditions. In this phase, total biomass concentrations (chl a) at the ice–water interface ranged from 4 to 15 µg L - ¹, while concentrations at 2 m and 5 m remained below 1 µg L - ¹. (Figure 8) In early December, a rise in salinity coincided with a sharp increase in biomass at the ice–water interface, exceeding 35 µg L - ¹, while deeper layers remained below 5 µg L - ¹. This was followed by a marked decline in surface biomass, with concentrations stabilizing and becoming comparable across all three depths until the end of the sampling period.

These variations were accompanied by a pronounced shift in pigment composition relative to chlorophyll a, with a clear increase in 19′-hexanoyloxyfucoxanthin (19′HF) and chlorophyll c2 across all three layers—most notably at the ice–water interface—alongside a sharp reduction in fucoxanthin. The increase in 19′HF, a pigment commonly associated with Phaeocystis antarctica and other small flagellated phytoplankton, together with the decline in fucoxanthin, which is strongly associated with diatoms, is consistent with a shift in the relative contribution of these phytoplankton groups. However, diagnostic pigments provide an indication of group-level contributions rather than direct taxonomic identification.

The PCA (Figure 10) explained 22.00% of the variance on PC1 and 20.65% on PC2, for 42.65% combined. The two axes therefore summarize a substantial but incomplete portion of the total variability. Their separation is consistent with covariation between pigment and environmental variables, but should not be interpreted as a direct demonstration of a transition between taxonomically defined communities. The PERMANOVA independently confirmed significant vertical structuring of the multivariate dataset (F = 1.7188, R² = 0.1629, p = 0.0003), with sampling date treated as a blocking factor. Importantly, the PERMDISP test was not significant (F = 1.5682, p = 0.1747), indicating no evidence of differences in multivariate dispersion among depth groups. Thus, the significant PERMANOVA result is not readily attributable to unequal within-group dispersion.

This pattern represents a deviation from the classical Antarctic paradigm, where diatoms typically dominate coastal and marginal ice zone (MIZ) areas during ice melt (Arrigo et al., 1999; 2000, 2008; DiTullio and Smith, 1996; Smith et al., 1998; 2014). Consistently, Mangoni et al. (2019) reported a bloom of Phaeocystis antarctica in the same region under conditions of strong stratification. The temporal evolution of nitrate and dissolved silica closely mirrored fluctuations in salinity within the surface layer, with maxima occurring during periods of higher salinity in early December and minima coinciding with the pronounced freshening at the end of the month. At intermediate depths, similar synchronous dynamics were observed, although with greater magnitudes, while deeper layers exhibited stable salinity yet still showed coherent nutrient oscillations, suggesting that processes other than surface freshwater input, including water-mass mixing and biological uptake/regeneration, may also have contributed to nutrient variability. Ammonium displayed episodic peaks often associated with shifts in nitrate and silica but without a consistent relationship to salinity, whereas phosphate and nitrite showed comparatively limited variability. Limited variability alone, however, does not demonstrate conservative behavior or the absence of biological processing. The stronger variability of nitrate and silicate is compatible with their involvement in phytoplankton uptake and regeneration, as well as with changes in water-mass contributions; the present dataset does not allow these processes to be quantitatively separated.

The hydrographic record is consistent with combined influences of offshore water and freshwater inputs, but the available measurements do not independently quantify horizontal advection or identify individual water masses using tracer-based criteria. Pigment-associated signatures differed across the three layers. In the present dataset, however, the HPLC pigment record provides group-associated pigment signals rather than independent taxonomic identification. This distinction underscores the importance of microscale dynamics beneath the ice for understanding ecological processes in these systems, where stratification induced by freshwater inputs may substantially alter Antarctic trophodynamics. Stratification creates a layer that can partially isolate the near-surface environment from deeper waters, potentially favoring organisms adapted to these selective conditions.

Initial observations indicated that the bottom ice layer became visibly pigmented during peak algal growth, reflecting high biomass accumulation. This biomass appeared to detach from the ice, but vertical export and subsequent seeding of deeper or benthic communities were not directly measured. In the following phase, despite the ice becoming more porous and spongy, the landfast ice failed to recolor to previous levels. One possible explanation is reduced access to nutrient-rich modified Circumpolar Deep Water (mCDW), although the present measurements do not directly demonstrate hydrographic isolation or nutrient limitation by mCDW.

Freshwater inputs, potentially including glacial meltwater, were associated with lower salinity and stratification of the upper water column, coinciding with changes in nutrient concentrations and reduced bottom-ice algal biomass. A distinct intermediate layer between 2 and 5 m exhibited elevated temperatures and signs of biological activity. This layer was influenced by intrusions of offshore waters and salinity gradients driven by meltwater input, forming a distinct hydrographic layer that may provide a different ecological setting for phytoplankton. Below 5 m, the water column transitioned into a deeper hydrographic regime with implications for vertical mixing and nutrient transport. At depths exceeding 100 m, the water column was characterized by relatively cold and stable conditions. These stratified dynamics, shaped by freshwater input, offshore water intrusion, and ice structure, underscore the sensitivity of ice-associated ecosystems to hydrographic shifts. Previous work from the same Terra Nova Bay expedition documented phytoplankton blooms beneath landfast ice during the melt season (Saggiomo et al., 2021) and, based on taxonomic observations of the same samples analyzed here, identified a prominent contribution of nanoflagellates, including Phaeocystis cf. antarctica and Ochromonas spp. The present pigment data provide a complementary perspective on these community changes. At the ice–water interface, the increase in 19′-hexanoyloxyfucoxanthin (19′HF), commonly associated with Phaeocystis and other small flagellated phytoplankton, together with the marked decline in fucoxanthin, strongly associated with diatoms, is consistent with a shift in the relative contribution of these groups during the seasonal transition. The agreement between the pigment patterns and the taxonomic observations reported by Saggiomo et al. (2021) supports the interpretation of a changing phytoplankton community beneath the melting landfast ice. Our observations further show that these changes occurred across both the ice and the immediately underlying water column, highlighting the close ecological coupling between the two habitats. The combined occurrence of warmer, saltier offshore water and freshwater-associated stratification may have created contrasting environmental conditions across the upper water column; whether horizontal transport acted as a biological inoculum cannot be established from the present observations. In situ Fv/Fm measurements, available at selected sampling dates in December and January, reached values of up to 0.67, indicating relatively high photochemical efficiency during these observations. Measurements were more frequent in December, whereas only a limited number of observations were available in January; therefore, the data are interpreted as complementary indicators of field photophysiological status rather than as evidence of a continuous temporal trend. The observed Fv/Fm values may reflect physiological acclimation to the prevailing light conditions and/or nutrient availability during the sampling period. However, these observations do not establish a causal relationship with concurrent hydrographic or nutrient changes, and a potential contribution of continental iron remains hypothetical because dissolved iron concentrations were not measured.

In parallel with these in situ observations, the UV and iron availability experiment showed that phytoplankton biomass responded differently to the experimental conditions at the two sampling times. At T1, chlorophyll a varied significantly among iron treatments, and the significant iron × UV interaction indicates that the response to UV exposure differed depending on iron availability. This pattern is consistent with the contrasting responses observed between the three iron conditions: UV exposure was associated with an increase in chlorophyll a under CTRL conditions, but with lower values under both dFe and DFB conditions. At T2, both iron treatment and UV exposure had significant effects on chlorophyll a, whereas their interaction was no longer significant. Thus, the relationship between iron availability and UV exposure was not constant throughout the incubation. These results suggest that the effects of UV radiation on phytoplankton biomass may depend on the prevailing iron conditions and on the duration of the incubation. However, the experiment was based on two independent bottles per treatment combination at each time point, and the statistical results should therefore be interpreted with caution. In particular, the significant effects detected by the ANOVA indicate differences among the experimental groups, but they do not by themselves identify the physiological mechanisms responsible for the observed changes in chlorophyll a. The response of Fv/Fm provides a complementary measure of photophysiological condition and showed a different pattern, with no significant effects of iron treatment, UV exposure, or their interaction at either T1 or T2. Taken together, the chlorophyll a and Fv/Fm results suggest that changes in phytoplankton biomass were more responsive to the experimental treatments than the maximum quantum efficiency of PSII under the conditions tested.

At Station C, the seasonal evolution of landfast ice was associated with the development of a distinct near-surface layer beneath the ice, directly influenced by ice melting and characterized by marked changes in salinity, stratification, nutrient conditions, and phytoplankton community structure. Surface salinity declined from 36.67 to 30.67, while surface chlorophyll a exceeded 40 µg L - ¹ during the early-December biomass maximum and subsequently declined to below 1 µg L - ¹. Together with the observed changes in nutrient concentrations and pigment composition, these patterns indicate that the water immediately beneath the ice was not simply part of the surrounding water column, but represented a rapidly changing environment in which the effects of melting were directly expressed in both physical and biological properties.

The pigment record further indicates that phytoplankton composition changed during the seasonal transition. Variations in fucoxanthin, 19′-hexanoyloxyfucoxanthin (19′HF), and chlorophyll c2 suggest changes in the relative contribution of pigment-associated groups, although diagnostic pigments alone cannot be used to demonstrate taxonomic dominance. The progressive development of stratification and freshening beneath the ice may have created conditions favoring different phytoplankton assemblages as the season progressed, linking changes in the physical environment to changes in the biological community.

The ecological relevance of this layer may extend beyond the seasonal conditions observed at Station C. As Antarctic coastal environments experience increasing ice loss and changes in the timing and intensity of melting, enhanced freshwater input may promote the formation and persistence of a low-salinity, stratified layer beneath landfast ice. A longer-lasting or thicker meltwater-influenced layer could modify the physical and biogeochemical conditions experienced by both sympagic communities associated with the ice and planktonic communities inhabiting the underlying water. Changes in the duration, thickness, and stability of this layer may therefore alter the connection between sea ice and the water column and influence the timing and development of primary producers at the ice–ocean interface. In this context, the layer beneath melting landfast ice represents a potentially important, but still poorly constrained, component of Antarctic coastal ecosystem responses to climate-driven changes in the cryosphere. The incubation experiment provides an additional indication that biological responses within this environment may also depend on resource availability. Phytoplankton biomass responded significantly to iron availability at both sampling times, while the rapid increase in Fv/Fm during the first 48 h, particularly under iron-enriched conditions, suggests an early physiological response to changes in iron availability. The subsequent variability in Fv/Fm indicates that this response may be influenced by other processes developing during the incubation. Although the experiment was limited in replication and was not designed to reproduce the full complexity of the natural system, it supports the view that changes in resource availability can contribute to the biological variability observed within the meltwater-influenced layer.

Overall, these observations highlight the ecological continuity between landfast ice and the underlying water, with melting ice creating a transient but biologically relevant environment at their interface. Rather than representing a simple transition between sympagic and pelagic habitats, this near-surface layer may function as an ecological space in its own right, where physical forcing and biogeochemical conditions interact to shape microbial and phytoplankton communities. The study represents a seasonal case study at a single station, and the spatial generality of these patterns cannot be established from the present observations alone. Nevertheless, the results identify the meltwater-influenced layer beneath landfast ice as a feature that deserves greater attention when assessing how Antarctic coastal ecosystems may respond to ongoing and future changes in sea-ice melting. Future studies combining repeated observations across sites and years with freshwater tracers, dissolved iron and nutrient measurements, and taxonomic and physiological approaches will be important for determining how the formation and persistence of this layer affect both sympagic and planktonic communities.