The Ross Sea holds immense ecological significance, encompassing diverse marine systems like polynyas, marginal ice zones, coastal and offshore areas. It is characterized by high biodiversity and is among the most productive regions in the Antarctic Ocean (Arrigo and McClain, 1994; Nelson et al., 1996; Smith and Gordon, 1997; Arrigo et al., 1998; Saggiomo et al., 2002; Vaillancourt et al., 2003; Arrigo et al., 2008; Smith and Kaufman, 2018; Smith, 2022), such that the Commission for the Conservation of Marine Living Resources (CCAMLR) in 2016 designated the Ross Sea a Marine Protected Area.

The investigation site, Terra Nova Bay (TNB) located along the Victoria Land coast between the Drygalsky Glacial tongue and Cape Washington ( Figure 1 ), is one of the most productive sectors in the Ross Sea (Lazzara et al., 2000; Saggiomo et al., 2002). An essential aspect of the Ross Sea ecology involves the extent and dynamics of ice, closely connected to water column stabilization due to the UML formation. This process significantly impacts the temporal variability of seasonal production (Innamorati et al., 2000; Mangoni et al., 2017).

The homogenization of the water column, resulting from Circumpolar Deep Water (CDW) upwelling, leads to a substantial rise in nutrients concentration in the surface layer during early spring. Typically, during this time, at TNB, concentrations of NO₂ ^-+ NO₃ ^-, PO₄ ^3- and Si(OH)₄ can reach values of 35, 2, 80 µM respectively, with ratios approximating the theoretical values (Redfield et al. 1963; Catalano et al., 1997, 2000; Innamorati et al., 2000).

In spring, increasing irradiance, temperatures and CDW upwelling trigger ice melting, causing a dilution of surface waters and the formation of a less dense surface layer (UML) separated from the underlying waters, this process being particularly pronounced in the Marginal Ice Zones (MIZ). This strong water column stabilization enables phytoplankton to better utilize available irradiance and nutrients; moreover, the presence of ice further increases nutrient concentrations, including iron, a limiting factor in Antarctic waters (Martin et al., 1990; Sedwick and DiTullio, 1997; Sedwick et al., 2000; Alderkamp et al., 2012). These conditions - strong summer irradiation, stable UML and nutrients availability - contribute to rapid phytoplankton growth between December and the end of January early February, with chlorophyll a reaching 8-10 mg m^-3 (Innamorati et al., 2000; Mangoni et al., 2004).

Phytoplankton assemblage dynamics in the Ross Sea alternate between diatoms and the Haptophyte Phaeocystis antarctica dominance, playing distinct role in terms of eco-physiology, biogeochemistry and food web functions (Smith et al., 2014). Biogeochemically, diatoms require silicon for frustules and assimilate N and P at lower ratio than that of Redfield (Arrigo et al., 2000; Sweeney et al., 2000; Dunbar et al., 2003). P. antarctica prevails in the offshore and southern waters (Arrigo et al., 2000; Liu and Smith, 2012), flourishing in the spring mixed water column and in polynya waters (Smith and Asper, 2001; Arrigo et al., 2003; Smith et al., 2010) due to its adaptation to low light conditions. In TNB, Phaeocystis dominance in the mixed water column under the UML has been observed (Goffart et al., 2000; Nuccio et al., 2000), while Mangoni et al. (2019) reported an unusual Phaeocystis bloom in the UML of TNB. The diatoms growth primarily occurs in the summer stratified waters of the MIZ (Garrison et al., 2005; Smith et al., 2014) and it is well documented in TNB by many authors (Goffart et al., 2000; Nuccio et al., 2000; Mangoni et al., 2004; Saggiomo et al., 2021). Despite interannual variability in ice melt and diatom bloom timing, there is a general consensus on the overall constancy of the diatoms bloom composition: centric diatoms (Thalassiosira, Chaetoceros, Dactyliosolen) dominate open waters (Leventer and Dunbar, 1996; Saggiomo et al., 2021), while pennate diatoms, particularly Fragilariopsis curta - known as “ice algal” species (Nelson and Smith, 1986) - predominate in the summer surface layer of TNB waters (Andreoli et al., 1995; Goffart et al., 2000; Nuccio et al., 2000). The relevant presence of this species has been observed into the annual sea-ice environment, during the spring-summer transition (Guglielmo et al., 2000, 2004; Lazzara et al., 2007).

The growth of diatoms leads to a substantial decrease in nutrients concentrations, in fact at TNB the concentrations of NO₃ ^- + NO₂ ^-, PO₄ ^3- and Si(OH)₄ in the UML fall down to values around 5, 0.05 and 15 µM respectively (Catalano et al., 2000; Innamorati et al., 2000; Saggiomo et al., 2002). These values remain high compared to other seas, also due to Iron deficiency acting as limiting factors for the phytoplankton growth, especially in the offshore Ross Sea areas (Martin et al., 1990; Sedwick and DiTullio, 1997). Additionally, nutrient ratios N/P, Si/N and Si/P undergo modifications due to differential nutrient utilization strategies, as distinct nutrient utilization strategies and biogeochemical consequences are linked to phytoplankton taxonomic composition. Ross Sea studies on inorganic elements removal ratios (ΔC/ΔN, ΔC/ΔP, ΔN/ΔP, ΔC/ΔN, ΔC/ΔSi/ΔN) reveal pronounced variations based on dominant species (Catalano et al., 1997; Arrigo et al., 2000; Sweeney et al., 2000; Smith and Asper, 2001; Saggiomo et al., 2002; Dunbar et al., 2003). Particularly, Phaeocystis antarctica-dominated assemblages exhibit significant higher ΔN/ΔP ratios (17.5-19.7) than Redfield, while diatom-dominated assemblages display lower ratio (10.1-15.2), underscoring the close connection between phytoplankton composition and biogeochemical properties in Ross Sea waters (Arrigo et al., 2000; Sweeney et al., 2000; Smith and Asper, 2001; Dunbar et al., 2003). A shift in the initial N/P ratio towards strong P deficiency during Fragilariopsis curta blooms is evident at TNB (Innamorati et al., 2000; Nuccio et al., 2000), showing that this phytoplankton assemblage assimilates greater quantities of P than N compared with the theoretical ones of Redfield (Catalano et al., 1997, 2000; Saggiomo et al., 2002).

This study is focused on the relationships between phytoplankton assemblage composition, studied through both pigment composition and optical microscopy, their nutrients assimilation preferences, and the resulting biogeochemical effects at TNB.

Analysis covers the temporal evolution of NO₃ ^-, NO₂ ^-, PO₄ ^3- and Si(OH)₄ concentrations in the UML, alongside phytoplankton biomass and taxonomic composition dynamics during austral springs and summers 2018-19 and 2019-20, from the ice melt until the end of January. The findings are compared with data from three ‘90s campaigns (1989-90, 1994-95, 1997-98), highlighting similarities, differences and the potential influence of climate change.

The area in front of the Italian base at TNB experienced significant ice-driven dynamics, classifying it as Marginal Ice Zone (MIZ) during both spring and summer. The upwelling of Antarctic Deep Water in spring brings high concentrations of nutrients and warmer water to the surface layers, which, along with ice melting and warming due to increased solar irradiance, leads to the segregation of the UML. In line with the findings reported by Mangoni et al. (2004), in the coastal sampling stations of the BioAPRoS campaigns, the UML was persistent and always very shallow, never exceeding the depth of 20 m ( Figure 2 ). UML thickness was at the minimum depth during the beginning of the sampling period (late November early December) and reached the maximum depths in mid-January, shortly before the maximum of Chla concentration observed within the UML ( Figure 3 ).

The growth dynamics we have tried to represent with our sampling frequency (almost daily sampling), being higher or similar to phytoplankton growth rate (Smith et al, 1996), allows us to follow with a certain accuracy the temporal evolution of nutrient concentrations, phytoplankton biomass and its taxonomic composition in a largely segregated UML system (Catalano et al., 1997). UML spring summer phytoplankton growth is supported by favourable conditions of light, temperature and large nutrient availability ( Figure 5 ). Under these conditions, variations in nutrient concentration appear to be essentially regulated by phytoplankton consumption. Only at the end of the cycle, at the end of summer, when nutrients become scarce, some other processes can supply the surface layers with nutrients in relation to weakening of the UML or on-site degradation of accumulated particulate and dissolved organic matter (Smith and Kaufman, 2018). These processes can modify the dynamics of phytoplankton biomass within the UML as highlighted by Mangoni et al. (2017). Innamorati et al. (2000) especially associate weakening of the UML and replenishing of nutrients with the appearance of a second maximum of Chla, not seen in the two more recent campaigns. These contributions can alter and affect the measurements of phytoplankton consumption/utilization of nutrients. In these cases, given the sensitivity of the system, we should expect a positive response of phytoplankton dynamics, which however has not been recorded ( Figure 4 ).

Both BioAPRoS campaigns exhibited comparable temporal nutrient dynamics and concentrations within the UML. Maximum nutrient concentrations were detected as the ice cover vanished, followed by their minimum level during the Chla peaks ( Figures 4 , 5 ). Recent campaigns maximum and minimum nutrient concentrations mirror those reported for 1987/88, 1989/90 and 1994/95 at TNB by Innamorati et al. (2000) and Catalano et al. (2000). UML Nutrients dynamics was synchronously opposite to that of Chla highlighting a pronounced consumption of nutrients during the growth of phytoplankton biomass ( Figures 4 , 5 ). Although Chla values correspond to those reported by Innamorati et al. (2000) for ‘90s campaign, their temporal dynamics exhibited some deviations. The previous campaigns maximum occurred approximately a month earlier, between end of December-mid January, followed by a second rise in early February not observed in the BioAProS campaigns. Innamorati et al. (2000) outlined that phytoplankton biomass declines at TNB during summer primarily derive from reduced nutrients availability. As shown in Figure 8 , the ratio Chl/(NP)½ approaches 100 in both BioAPRoS and 90s campaigns. This threshold poses a limitation of nutrients exploitation at TNB, which is still lower than the more efficient nutrient utilization in other seas (Innamorati and Giovanardi, 1992). Thus, the presence of other limiting factors beyond general nutrient scarcity can be assumed. Iron deficiency leads to the failure to exploit the still high concentrations of macro-nutrient. Although iron limitation has been postulated (Martin et al., 1990) and verified (Sedwick et al., 2000; Feng et al., 2010; Smith and Kaufman, 2018) to be a limiting factor in The Ross Sea, it seems more pertinent to offshore Ross Sea areas than TNB coastal zones (Sedwick and DiTullio, 1997; Sedwick et al., 2000; Arrigo et al., 2003) strongly influenced by the melting of the ice. However, the role of iron in the UML of coastal TNB requires further investigation. In summer, following robust phytoplankton growth, together with the lower nutrient concentrations a strong imbalance in their ratios occurs, which has been defined as an important limiting factor at TNB (Innamorati et al., 2000). Notably, the N/P ratio undergoes significant changes during the spring-summer period. Initially and early spring UML, N/P ratio was near the theoretical Redfield value ( Figure 8A ). Over time, during the phytoplankton bloom, in BioAPRoS18 and particularly in the ‘90s campaigns ( Figure 8B ), this ratio rises to remarkably high levels, leading to a pronounced P deficiency (Catalano et al., 2000; Innamorati et al., 2000; Sweeney et al., 2000; Saggiomo et al., 2002). In BioAPRoS18 the N/P ratio rose up to 40 and higher values was reached in ‘90s ( Figures 8 , 6A ). Conversely, in the BioAPRoS19, N/P ratio fluctuated from the beginning of the season up to the Chla maximum, culminating at approximately 4 ( Figure 6B ), a condition of N deficiency unprecedented in previous observations at TNB. Modification of the in-water nutrients ratios in the UML likely arises from the distinct ways in which the phytoplankton assimilate these nutrients. A confirmation is evident in the dynamics of Si(OH)₄. The concentration of Si(OH)₄ remained constant during the short initial bloom of Phaeocystis in BioAPRoS18 ( Figures 5E , 9A , 10A ) and starts to decrease only when diatoms become dominant in the phytoplankton community. In BioAPRoS19, where the community has always been dominated by diatoms, Si(OH)₄ shows a continuous decrease ( Figure 5F ). This dynamics underscores the system sensitivity to changes: even a slight shift in assemblage composition over a few days results in an immediate alteration in UML nutrient concentration, confirming the prevalent role of the phytoplankton and its taxonomic composition in UML nutrient removal following their preferences.

Based on this awareness, nutrient removal ratios ΔN/ΔP, ΔSi/ΔN and ΔSi/ΔP were estimated. Specifically, ΔN/ΔP ratio deviated from its theoretical value, assuming a lower value of about 11.8 during BioAPRoS18, similarly to the average of 11.5 measured during the ‘90s campaigns ( Table 1 ; Figure 11 ). These values align with those reported by Catalano et al. (1997); Catalano et al. (2000); Saggiomo et al. (2002) for TNB, as well as by Arrigo et al. (2000); Sweeney et al. (2000); Smith and Asper (2001); Dunbar et al. (2003) for diatom dominated phytoplankton assemblage of the Ross Sea. This phenomenon, termed luxury consumption, signified that P removal surpasses theoretical N removal, a pattern often observed in diatoms. It is to be expected that the strategies of phytoplankton nutrients assimilation determine alteration in the original in-water N/P ratios. A ΔN/ΔP removal ratio of about 11.5 determines a P deficiency state over time. Conversely, the nutrients ratios dynamic during BioAPRoS19 displayed an opposite trend. The ΔN/ΔP ratio average was 17.6, surpassed the theoretic reference of 16 ( Table 1 ; Figure 11 ), indicating a greater preference for N over P utilization. This accounts for the occurrence of N deficiency, an unprecedented event in previous years in TNB.

In BioAPRoS19, along with the previously unobserved N deficiency, an equally unprecedented bloom of Dactyliosolen tenuijunctus was identified ( Figure 10B ). In both BioAPRoS campaigns diatoms dominated; however, the phytoplankton bloom described since the ‘90s in TNB was typically dominated by pennate diatoms, mainly Fragilariopsis curta, as reported by Nuccio et al. (2000); Mangoni et al. (2017); Saggiomo et al. (2021). The same phytoplankton composition was also observed during the Chla maximum in BioAPRoS18 ( Figure 10A ). In BioAPRoS19, alongside the pennate diatoms (mostly Pseudo-nitzschia spp.), the centric Dactyliosolen tenuijunctus dominanted during the peak phase. Dactyliosolen tenuijunctus was previously found in TNB as the prevailed species in the November sea-ice column (Saggiomo et al., 2017), in limited quantities during the ice melting period (Nuccio et al., 2000; Mangoni et al., 2017), together with other centric diatoms, primarily in offshore waters (Saggiomo et al., 2021). The dominance of Dactyliosolen tenuijunctus, along with a notable presence of Pseudo-nitzschia spp. and Chaetoceros spp., and the reduced presence of Fragilariopsis, confirm the peculiar nature of the blooming phytoplankton assemblage of BioAPRoS19. A similar Dactyliosolen tenuijunctus bloom was recently documented by Lee et al. (2022) in the neighboring Amundsen Sea during austral summer 2020. The authors classified this event as an extraordinary occurrence, possibly associated to the ongoing effects of climate change. It is plausible that the Dactyliosolen tenuijunctus bloom could be connected through horizontal transport between the Amundsen Sea and the Ross Sea. Various waters fluxes in the coastal area of TNB lead to the growth and accumulation of centric diatoms assemblage, more typical of offshore waters (Saggiomo et al., 2021).

It is well known that grazing by krill and Copepods plays an important role in Antarctica and in the Ross Sea, through which about half of the phytoplankton primary production is transferred to higher trophic levels (Smith et al., 2017). Their intervention on the phytoplankton community is always very important on the biomass present, but it can also be important on the taxonomic composition due to the diet selectivity (Hopkins, 1987, Crescenti et al., 2000). This selectivity is essentially expressed on a dimensional basis and with the general preference of the diatoms over the Phaeocystis (Meyer and ElSayed, 1987, Hopkins 1987, Haberman et al., 2003; Nejstgaard et al., 2007; Smith et al., 2017). As far as it is known regarding grazing selectivity among diatoms, Meyer and El Sayed (1987) report that krill have a higher filtration efficiency for Nitzschia and Chaetoceros than for Fragilariopsis. On the contrary, we could not find information about the selectivity of grazing on Dactyliosolen tenuijunctus, therefore, it is very difficult to evaluate the role of grazing in selecting species dominance in TNB diatom assemblages. However, considering the size dominance of grazing process, it does not seem that this could be a strong discriminant between the dominance of Dactyliosolen tenuijunctus or Fragilariopsis curta.

By merging the described nutrients and phytoplankton dynamics, it can be assumed that the blooms of Fragilariopsis curta were associated with a condition of P deficiency due to their reliance of N and P with a ΔN/ΔP ratio within the 10.8-12.3 range, indicating a preferential assimilation of P over N. The BioAPRoS19 campaign in TNB stands out, not only for the different taxonomic composition of phytoplankton highlighted by the unexpected Dactyliosolen tenuijunctus bloom but also for its distinctive nutritional dynamics. Differing from all previous campaigns in TNB, the N/P ratio, at the Chla peak, showed a state of N deficiency, driven by the ΔN/ΔP ratio exceeding 17.5. This highlights elevated N disappearance respect to P, an opposite condition compared to the Fragilariopsis curta-dominated assemblage, indicating different nutrients assimilation modalities. An analogous increase in the ΔN/ΔP ratio was also noted in waters dominated by Phaeocystis antarctica, typically offshore and along the southern part of the Ross Ice Shelf in the Ross Sea. In these waters ΔN/ΔP ratio can rise beyond Redfield value, reaching approximately 19 due to the nutritional preferences of Phaeocystis antarctica (Arrigo et al., 2000; Sweeney et al., 2000; Smith and Asper, 2001; Dunbar et al., 2003). However, no Phaeocystis blooms were detected during BioAPRoS19, except for a very slight increase in its abundance in the final days of sampling. Instead, a short Phaeocystis antarctica bloom emerged between mid- December and the end of December during BioAPRoS18.

Typically, at TNB, in relation to the diatoms dominance, ΔSi/ΔN ratio consistently exceeds 1.5, often surpassing 3, particularly during the BioAPRoS campaigns. The Si concentrations in the two campaigns at blooming time are different, higher in BioAPRoS18 than in BioAPRoS19 ( Figures 5E, F ; 10 ). Also Si/P dynamics differ markedly between the two BioAPRoS campaigns ( Figure 6 ): in BioAPRoS18 it increased in correspondence with the peak of Chla, whereas in contrast, it decreased in BioAPRoS19. Notably, in latter campaign, ΔSi/ΔP values surged to over 50, the highest recorded. Petrou et al. (2019) described changes in Si utilization by Antarctic diatoms in relation to the acidification of the Southern Ocean.

The assemblage dominated by Dactyliosolen tenuijunctus exhibited not only an elevated ΔN/ΔP ratio but also grater ΔSi/ΔP ratio ( Figure 11 ). This can be explained by a greater N requirement, lower need of P by Dactyliosolen tenuijuncus unlike Fragilariopsis curta. Consequently, biogeochemical differentiation emerges between the two BioAPRoS campaigns which involves the concentrations and the relationships not only between N, P but also with Si, of which, most probably, the phytoplankton assemblage is the main protagonist. The series of campaigns of the ‘90s and further sampling of BioAPRoS18 exhibit a general stability in phytoplankton assemblage composition and biogeochemical characteristics (Catalano et al., 2000; Innamorati et al., 2000; Nuccio et al., 2000; Mangoni et al., 2017; Bolinesi et al., 2020; Saggiomo et al., 2021). The variations that arise from the comparison of this series of campaigns include a slight increasing trend of ΔSi/ΔP from 1990 to 2018, and lower ΔSi/ΔN in 1990, as well as the temporal differences in the dynamics of the phytoplankton biomass. The temporal discrepancies appear to correlate with the period of ice cover melting and disappearance, which, in turn, are connected to the annual weather variability. Conversely, the results of BioAPRoS19 show distinct conditions compared to previous year and entirety of the ‘90s. Specifically, in contrast to previous campaigns with the dominance of Fragilariopsis curta, where N/P reached hight values ( Figure 8A ), the appearance of a different phytoplankton composition with distinct nutrient requirements resulted in a state of N deficiency related to higher ΔN/ΔP. Literature has long highlighted the differing nutritional needs of the Ross Sea pennate diatoms assemblage and those dominated by Phaeocystis antarctica. This study, enabled by microscopic analyzes of the phytoplankton assemblage composition, further distinguished the nutritional requirements between the two assemblages dominated by Fragilariopsis curta and Dactyliosolen tenuijunctus.

These findings suggest at least two considerations. The first refers to the importance of deepening the study of the photo-physiological, ecological and biogeochemical behaviour of the different species, made possible by the recent development of bio-optical models for the estimation of pelagic primary production (Bonamano et al., 2020). The second refers to the conditions of the TNB ecosystems. The Ross Sea and TNB are among the planet most significant marine ecosystems, exerting a profound global influence, in fact these regions support high primary production and play a crucial role in generating the conveyor belt system that fuels ocean currents (Rahmstorf, 2006). TNB ecosystems, which remained fairly stable since the ‘90s, could potentially undergo shifts in their biological and biogeochemical characteristics, echoing concerns raised by Mangoni et al. (2017) and by Smith et al. (2012) for the entire Ross Sea. The insights gained so far underscore the need for in-depth research and ongoing monitoring into the context of planetary climate change.

The results of recent BioAPRoS18 and BioAPRoS19 campaigns compared with those carried out in the ‘90s at BTN show many similarities and some important differences. UML formation modalities and characteristics, nutrient concentrations and their ratios (like Redfield) in spring-early summer and their strong consumption associated with phytoplankton growth in summer are similar. At summer Chla maxima, reached with some temporal deviations between recent and ‘90s campaigns, nutrient concentrations were at their minima. At this stage in BioAPRoS18, a recurrent condition of P deficiency already observed at TNB, and elsewhere in the Ross Sea, was detected due to preferential assimilation of P over N (ΔN/ΔP about 11.5). Associated with this condition, the recurrent bloom of pennate diatoms like Fragilariopsis curta, was present. Conversely, the BioAPRoS19 campaign exhibited different nutritional dynamics, marked by N deficiency (ΔN/ΔP = 17.6) and a bloom of centric Dactyliosolen tenuijunctus, which shows different nutrient requirements among diatoms species and which was never detected in this area. Furthermore, during the bloom of the two phytoplankton assemblages differences also emerged in the Chla/Si and Si/P ratios, probably linked to their preferences in the utilization of Si.

These results highlight the importance of studying ecological and biogeochemical role of the different phytoplankton species, the potential for shifts in the TNB ecological and biogeochemical characteristics. Given their global influence on ocean currents, primary production, and carbon storage, understanding these ecosystems is vital in the context of planetary climate change. The study offers insights into the complex interactions between nutrient availability, phytoplankton dynamics, and the effects of changing ice cover, contributing to our understanding of the intricate relationships within these critical marine environments.