Introduction

Arctic phytoplankton are experiencing a transition from a polar to temperate regime due to rapid climate change and dramatic sea ice loss (Carmack et al. 2015; Polyakov et al. 2020). Correspondingly, phytoplankton biomass and net primary production have shown continued increases over the last two decades especially on the interior and inflow shelves in association with reduced sea-ice extent and a longer phytoplankton growing season in light of earlier retreat and later advance of sea ice (Arrigo and van Dijken 2015; Lewis et al. 2020). Regions with delayed sea ice development can display a second phytoplankton bloom in the autumn supported by wind-driven vertical mixing (Ardyna et al. 2014) and greater inflows of Atlantic waters (Orkney et al. 2022b). Massive under-ice phytoplankton blooms dominated by the pelagic genera of Chaetoceros, Thalassiosira and Fragilariopsis rather than ice algae can extended far into the ice pack because of the enhanced light penetration into the upper water column through the thinning sea ice (Arrigo et al. 2012; Horvat et al. 2017; Ardyna et al. 2020a). The gradual expansion of Atlantic-derived waters over the Barents Shelf (Oziel et al. 2016) and ‘Atlantification’ of the Eurasian Arctic Ocean (Polyakov et al. 2017) have led to an intrusion of temperate species into the high Arctic, co-existing with the boreal and Arctic assemblages and the sympagic (sea ice associated) algae (Hegseth and Sundfjord 2008). Such cases have been identified from the poleward advective expansion of the coccolithophore Emiliania huxleyi (Winter et al. 2014; Neukermans et al. 2018; Oziel et al. 2020), a sentinel species for temperate ecosystems. It is foreseeable that future borealization of the Arctic will continue shaping the phytoplankton diversity and biogeography of the region (Polyakov et al. 2020; Henson et al. 2021; Ingvaldsen et al. 2021), altering the productivity, phenology and assemblage composition (Ardyna and Arrigo 2020), with potential ramifications on the Arctic marine food web and biogeochemical cycles (Wassmann et al. 2006).

The Barents Sea is a favourable region for water mass transformations and physical-biological coupling processes (Carmack and Wassmann 2006), with a mean depth of about 230 m. Four major water masses: warm and saline Atlantic water (salinity > 34.8; temperature > 3 °C), cold and fresh Arctic water (< 34.7; <0 °C), dense Barents Sea water (> 34.8; ≤2 °C), as well as fresh Norwegian Coastal Current water (< 34.4; >3 °C), are characterized and constrained by shallow topography (Oziel et al. 2016). Consequently, a sharp contrast in temperature and salinity of seawater exists at the Polar Front, a crucial hydrographic feature in the Barents Sea, which is split eastward into two branches: the southern front with strong temperature gradients and the northern front with strong salinity gradients (Oziel et al. 2016). Although the Arctic Ocean is a net exporter of dissolved inorganic nutrients, the Barents Sea Opening is the most important gateway for the import of nitrate, phosphate and silicic acid to the Arctic (Torres-Valdés et al. 2013), and most phytoplankton new primary production in the Barents Sea is probably supported by Atlantic-derived nutrients (Tuerena et al. 2021). Given the Barents Sea is one of the most productive (23–45% of the pan-Arctic primary productivity) high-latitude seas (Carmack et al. 2006), whole-Arctic primary production cannot be fully understood without having a clear understanding of bottom-up controls on Barents Sea phytoplankton blooms, such as nutrients from inflowing Atlantic-derived waters. However, the current status of the nutrient budgets in the region is still under debate, particularly whether the Atlantic water inflow, enhanced mixing, benthic nitrogen cycling and land-ocean interaction can sustain the continuing increase of primary productivity in the Barents Sea (Tuerena et al. 2022). Any limitation of primary production by dissolved nutrients may lead to a knock-on effect on the Arctic ecosystems by altering the phytoplankton physiology (Lewis et al. 2019; Ko et al. 2020), primary productivity and energy flows to higher trophic levels (Wassmann et al. 2006; Hunt et al. Jr 2013).

The variability of phytoplankton taxonomic composition could greatly affect the ecosystem services of the Barents Sea by altered contributions to export of biogenic matter (Lalande et al. 2013), megabenthic secondary production (Degen et al. 2016), as well as trophic pathways (Vernet et al. 2017). Moreover, the borealization of the Barents Sea may fundamentally change the existing bottom-up controls on the phytoplankton community, leading to compositionally novel assemblages (Makarevich et al. 2023). The taxonomic composition, distribution and blooms of phytoplankton assemblages in the Barents Sea exhibit strong spatial and temporal variability (Rat’kova and Wassmann 2002; Oziel et al. 2017), because of their close relation with seasonal ice edge position, stratification (driven by ice-melting) and deep vertical mixing (driven by brine rejection and wind). Phytoplankton blooms, composed primarily of diatoms, usually occur at the boundaries of the marginal ice zone, where strong physical gradients create suitable niches (water stability, enhanced irradiation) for biomass accumulation in spring (Wassmann et al. 1999; Makarevich et al. 2022). However, the spring bloom can also commence in the absence of surface layer stratification due to the small changes in actively mixing layer depth (Koenig et al. 2024). The composition of the dominant diatom taxa within the marginal ice zone is a mixture of sympagic (Attheya, Fossulaphycus, Synedropsis), cryo-pelagic (Fragilariopsis) and pelagic (Chaetoceros, Porosira, Thalassiosira) populations and some genera (Navicula, Nitzschia) may have species of all three ecotypes (Hegseth and Von Quillfeldt 2022). By contrast, the abundance of dinoflagellate cells tends to fluctuate seasonally but exhibit a diverse taxonomic composition with the genera Gymnodinium, Gyrodinium, and Protoperidinium often dominant (Rat’kova and Wassmann 2002; Makarevich et al. 2021; Assmy et al. 2023; Kohlbach et al. 2023). It is noteworthy that some dinoflagellates may often be mixotrophic (Stoecker and Lavrentyev 2018) and tend to bloom later in the year, in comparison to autotrophic diatoms. Bio-optical investigations revealed the increasing contributions of prymnesiophytes to the Barents Sea phytoplankton community, such as the north-eastward expansion of coccolithophores and the increasing dominance of Phaeocystis (Oziel et al. 2017; Neukermans et al. 2018; Orkney et al. 2020). Satellite observations have shown that coccolithophore blooms in the Barents Sea commonly develop in summer (Signorini and McClain 2009), and are strongly related to the positive temperature and negative salinity anomalies (Smyth et al. 2004). Field in situ summer surveys have confirmed a low species diversity of the coccolithophore population and the overwhelming dominance of the species E. huxleyi, with highest bloom concentration up to 1.2 × 107 cells L−1 and close relations with well-stratified Atlantic waters (Giraudeau et al. 2016; Silkin et al. 2020a, b), as well as faster turnover rates and greater retention of phosphorus (Downes et al. 2021). Regardless of the future climate fluctuations in the Arctic, it is undoubted that the fast Atlantification of the Barents Sea will produce persistent influence on the marine ecosystem, by conspicuous and potential changes on the phytoplankton biogeography, taxonomic composition, timing of blooms, as well as its ecological importance in the Arctic food web (Dalpadado et al. 2020).

The primary productivity of the Barents Sea is subjected to various processes of seasonal light variability, formation and melting of sea ice, wind-induced mixing and exchange of heat, salt and nutrients with different water masses (Slagstad et al. 2015; Henley et al. 2020; Sandø et al. 2021), as well as grazing and infection by viruses (Suttle et al. 1990; Verity et al. 2002). However, it is still uncertain whether competing factors will result in increases or decreases in net primary production in the Barents Sea. Thus, continued monitoring of phytoplankton distributions and understanding of their governing mechanisms are of great importance. In a recent study in the western Barents Sea (Orkney et al. 2022a), multivariate statistical analyses were applied to explore how the representation of phytoplankton community structure influences the interpretation of environmental dependencies. The conclusion pointed out that, while pigment and absorption spectral datasets provide useful information about seasonal succession and the influence of Atlantic water masses, datasets of cell counts revealed the greatest insight into environmental dependencies. Here, we make use of a comprehensive dataset of microphytoplankton cell counts to determine the environmental controls on phytoplankton community structure in the Barents Sea. We seek to test whether previously reported environmental controls on community structure remain robust in the face of increasing Atlantification of hydrographic conditions and borealization of Arctic phytoplankton communities. To address this, we first examine the taxonomic composition, abundance and biomass of diatoms, dinoflagellates and coccolithophores in the western Barents Sea, followed by investigating the physical seawater changes and nutrient conditions in association with water masses and fronts. We then use multivariate statistical methods to explore the principal physico-chemical factors that are responsible for the spatial and seasonal distributions of these three groups of phytoplankton. This study illustrates the influence of water masses on spatial and seasonal distributions of phytoplankton composition and carbon biomass during two contrasting years of sea ice conditions in the western Barents Sea, with potential implications for biogeochemical modeling and projections of the Arctic marine ecosystems in a fluctuating climate.

Methods

Cruise track and field sampling

Four multidisciplinary cruises were carried out in the western Barents Sea (Fig. 1): in late summer 2017 (JR16006; 30th June to 8th August) and early summer 2018 (JR17006; 11th June to 6th July) by RRS James Clark Ross, and winter (HH180101; 4th to 18th January) and spring 2018 (HH180423; 23rd April to 5th May) by R/V Helmer Hanssen. Cruises operated across a strong latitudinal gradient in hydrographic conditions, extending from the perennially ice-free Barents Sea Opening to the quasi-permanently ice-covered Nansen Basin. Sampling stations were set along the transects of the Barents Sea Opening and 30°E meridian northward to the shelf break near the Nansen Basin for the two summer surveys. Most northerly stations were not visited during the winter and spring cruises due to the adverse sea ice conditions. However, sufficient sampling stations were occupied to ensure spatial and temporal coverage of various water masses, fronts, and the seasonal ice zone. This diverse range of environmental observations and in situ phytoplankton collections provide a valuable opportunity to study the relationship between phytoplankton community composition in a region on the front-line of climate change.

Fig. 1
figure 1

Map of circulation and sampling locations in the western Barents Sea (a: Diagrams of circulation after Oziel et al. (2016); Black dashed lines indicate the position of the Polar Front; b, c, d, e: Seasonal sampling stations; Blue dashed lines indicate the monthly averaged sea ice extent for that cruise; Green dashed lines indicate the potential boundaries of the two phytoplankton assemblages according to the results of cluster analysis)

Full size image

A rosette sampler, equipped with a Seabird SBE911plus CTD (conductivity, temperature and depth) package, a LI-COR PAR sensor, a Chelsea AquaTracka III Fluorometer and 12 Niskin bottles, was deployed for hydrographic profile observation and seawater sample collection. Temperature, salinity, depth, PAR (photosynthetically active radiation) and in vivo chlorophyll a fluorescence were recorded by the sensor array on the CTD sampler. Typically, discrete phytoplankton samples of surface, sub-surface chlorophyll maximum, and deep samples (where PAR attenuated to 1% of the surface value) were collected during the upward cast. Triplicate 50 mL subsamples, after filtering through 200-µm sieve, were taken from each depth for the analysis of dissolved nitrate, phosphate and silicic acid concentrations. Nutrient analysis was conducted onboard using a Lachat Quikchem 8500 flow injection analysis system standardised using international certified reference materials for nutrients in seawater (KANSO Ltd., Japan) (Henley et al. 2020).

Regarding phytoplankton sampling, 50 mL seawater from each depth was subsampled and preserved with Lugol’s solution to a final concentration of 1% in brown polyethylene bottles for the subsequent analysis of diatoms and dinoflagellates. A total of 50, 40 and 19 water samples at discrete depths were collected during the summer 2017, summer and spring 2018 cruises, respectively. Only six surface samples were collected during the winter 2018 cruise. Another 400 mL of seawater was collected and vacuum filtered at low-pressure through 0.8-µm pore size Whatman cellulose membrane with an additional-buffered Milli-Q solution rinse for the sampling of coccolithophores. After oven drying at 50 °C for at least 8 h, each membrane with filtered particles was then transferred and preserved in a plastic petri dish until analysis. A total of 50, 40, 11 and 5 filter samples were collected during summer 2017, summer, spring and winter 2018 cruises, respectively.

Phytoplankton taxonomy, enumeration and biomass estimation

For the taxonomy and enumeration of diatoms and dinoflagellates, 25 mL subsamples were examined after sedimentation for at least 24 h in Utermöhl chambers by using an inverted Zeiss Axiovert S100 Microscope with 200× magnification and Ph2 contrast observing conditions (Utermöhl 1958). Cells were then counted and identified to genus or species level according to their morphological characteristics in consultation with references (Tomas et al. 1997; Throndsen et al. 2007). The geometric parameters of a maximum of 30 cells of each taxon were measured simultaneously with a calibrated eyepiece graticule for the calculation of biovolume (Sun and Liu 2003), which was then used to estimate the carbon biomass of diatoms and dinoflagellates incorporating correction factors for preservation effects (Menden-Deuer and Lessard 2000; Menden-Deuer et al. 2001).

The identification and enumeration of coccolithophores was carried out using a combined observation with light microscope (LM) and scanning electron microscope (SEM) according to the morphological differences in coccoliths and coccospheres (Young et al. 2003). In regard to the examining of two summer surface samples, a piece of ~ 1 cm2 membrane was cut and mounted with neutral balsam between the glass slide and cover slip, followed by the LM examination of at least 100 fields of view by a Zeiss Axioskop 2 plus Microscope with 1000× magnification and DIC III contrast observing conditions after addition of immersion oil. In regard to the examining of the rest of the samples, a piece of the filter (∼ 0.5 cm2) was cut and attached to a stub using conductive double-sided adhesive tape, followed by coating in a Polaron SC7620 sputter coater. Cell and coccolith counts were undertaken across a total of 500 randomly selected viewing fields in horizontal rows at 2000 × magnification by a JEOL JSM-6390LV SEM. Coccolith volume was estimated from the species-specific shape constant and the average maximum coccolith length. Calcium carbonate masses were then derived from the product of inferred coccolith volume and a representative estimate of calcite density (Young and Ziveri 2000).

Statistical analysis

Cluster analysis was performed at the lowest taxonomic level to quantify similarity between samples based on Euclidean distance with a complete linkage by using Minitab 15 software, after log10 (x + 1) transformation of the abundance dataset. Redundancy analysis (RDA), a constrained ordination model, was undertaken to estimate how much of the variation in the phytoplankton distribution (response variables) could be attributed to changes in the environment (explanatory variables). In order to minimize the weight bias that was caused by identified and unidentified species in each genus, we aggregated the abundance data to genus level. The dominant taxa, which were identified by product (> 0.0001 for diatoms; > 0.001 for dinoflagellates) of the frequency of occurrence and the percentage abundance in all samples (Dufrêne and Legendre 1997), were incorporated into the RDA for ordination analysis. Before the RDA, a detrended correspondence analysis (DCA) was conducted to assess the lengths of gradient in species data and to choose between unimodal and linear ordination (Šmilauer and Lepš 2014). PAR, temperature, salinity and nutrient concentrations were employed as explanatory variables for the analysis. Both DCA and RDA were undertaken in the Canoco 5 software, after log10 (x + 1) transformation of the datasets of PAR and cell abundances.

Results

Characteristics of seawater temperature and salinity

Contrasting spatial and seasonal distributions of seawater temperature and salinity are evident in the latitudinal section profiles along the Barents Sea Opening (between Bear Island and the northern extremity of Norway) and the 30°E meridian. In summer 2017, a cooler and less saline water mass (Fig. 2a, b), indicating mixing of Atlantic and shelf waters, was encountered near station B5 (74.4°N), where the water column above the Spitsbergen Bank was characterized by temperature of 3.15–3.36 °C and salinity of 34.65–34.74. Two water masses with high temperature and low salinity were found in the vicinity of the Polar Front south of Svalbard at station B7 (< 40 m; 6.04–7.01 °C and 34.84–34.93) and the near Norwegian Coastal Current at station B2 (< 30 m; 7.30–8.80 °C and 34.77–34.83). The polar front is clearly defined in the distribution of temperature and salinity along the 30°E transect. In summer 2017 (Fig. 2c, d), the warm (> 3 °C) Atlantic water encroached north of station B14 (76.5°N) and was present in the upper 30 m of the water column. The cold (< 0 °C) Arctic water occurred northward of station B15 (78.2°N), with a fresh melt-water layer (< 10 m) of 32.40–33.72 (mean of 33.23 ± 0.55) extending to station B17 (81.4°N). We observed relatively warm and saline water (> 0 °C, > 34.8) at depth (> 60 m) beneath the cold Arctic water in the far north of the 30°E transect, which we interpret as a retrograde flow deriving from the Atlantic water boundary current. In winter (Fig. 2e, f), the polar front outcropped near B14, and both Atlantic and Arctic water masses were vertically well mixed. In spring (Fig. 2g, h), the boundary of the two water masses moved southward to station HH63 (76.2°N), and a marked stratified layer (< 0 °C, < 34.8) existed in the Arctic water (< 40 m). In comparison with late summer 2017, the Atlantic water (> 3 °C, > 34.8) still occurred to the north of station B14 in early summer 2018 (Fig. 2i, j) but was vertically mixed to the south. Similarly, the Arctic water was stratified with a fresh layer (< 10 m) of 33.98–34.18 (mean of 34.09 ± 0.09) north of station B34 (77.5°N), and the occurrence of the subducted Atlantic water (> 0 °C, > 34.8) was still conspicuous in the deeper waters (> 50 m) at the shelf break near Nansen Basin.

Fig. 2
figure 2

Seasonal vertical profiles of seawater temperature and salinity along the Barents Sea Opening (BSO) and 30°E longitude (0 °C line indicates the upper temperature limit of the Arctic water; 3 °C line indicates the lower temperature limit of the Atlantic water; 34.8 line indicates the lower salinity limit of the Atlantic water)

Full size image

It is noteworthy that although recurrent evidence of the same suite of water masses was found, the geographic boundaries between these water masses were mobile on seasonal and interannual time-scales, which may have been transiently interrupted by eddies, or potentially modified by processes such as sea-ice melt and terrestrial run-off.

Characteristics of seawater nutrients

Seawater nutrient concentrations in the western Barents Sea showed strong seasonal variation on section profiles (Fig. 3). Nutrient utilisation in surface waters generated seasonal minima in the two summers. In summer 2017, low values of nitrate, phosphate and silicic acid were measured at 1.03 ± 0.54 µmol L−1, 0.19 ± 0.07 µmol L−1 and 0.76 ± 0.31 µmol L−1 in upper waters (< 20 m) from station B5 to station B7 (Fig. 3a, b, c) along the Barents Sea Opening, and 0.45 ± 0.18 µmol L−1, 0.10 ± 0.05 µmol L−1 and 0.20 ± 0.13 µmol L−1 in upper waters (< 20 m) from station B13 to station B15 (Fig. 3d, e, f) along the 30°E transect. In winter, nitrate, phosphate and silicic acid were all well mixed vertically in the water column (Fig. 3g, h, i), with high values of 10.23 ± 0.27 µmol L−1, 0.61 ± 0.05 µmol L−1 and 2.26 ± 0.60 µmol L−1 in the Atlantic water and relatively high values of 6.78 ± 1.34 µmol L−1, 0.60 ± 0.04 µmol L−1 and 0.92 ± 0.32 µmol L−1 in the Arctic water. In spring, nutrients were rich in upper water layers, with high averages of 10.46 µmol L−1 of nitrate, 0.66 µmol L−1 of phosphate and 4.51 µmol L−1 of silicic acid at station B13 (< 30 m). By contrast, lower concentrations occurred in the phytoplankton bloom locations (Fig. 3j, k, l), where the nutrients were drawn down to 2.68 µmol L−1 of nitrate at station B14 (15 m), 0.16 µmol L−1 of phosphate at station B14 (15 m) and 0.22 µmol L−1 of silicic acid at station HH65 (2 m). In summer 2018, the low nutrient upper waters (< 20 m) extended far into the northern Barents Sea from station B13 to station JR89 (Fig. 3m, n, o), with concentrations of 0.29 ± 0.18 µmol L−1, 0.03 ± 0.02 µmol L−1 and 0.14 ± 0.12 µmol L−1 of nitrate, phosphate and silicic acid, respectively.

Fig. 3
figure 3

Seasonal vertical profiles of seawater nitrate, phosphate and silicic acid (Solid black lines indicate the drawdown of nutrient concentrations in spring and summer seasons)

Full size image

Diatom composition, abundances and biomass

Diatoms in the western Barents Sea demonstrated strong seasonality in both taxonomic composition and abundances (Table 1). Mean abundances and biomass peaked in spring (1.8 ± 1.6 × 106 cells L−1, 119.5 ± 100.2 µg C L−1), but were considerably lower in summer 2018 (2.1 ± 3.1 × 105 cells L−1, 20.8 ± 31.9 µg C L−1), summer 2017 (7.4 ± 13.6 × 104 cells L−1, 13.8 ± 44.1 µg C L−1) and winter (60.0 ± 45.6 cells L−1, 0.005 ± 0.003 µg C L−1). In spring, the dominant taxa were Chaetoceros, Fragilariopsis, Thalassiosira and Fossulaphycus, which on average accounted for 43.9%, 34.2%, 6.6% and 6.5% of abundances, corresponding to 21.4%, 23.9%, 36.4% and 8.1% of biomass. The highest abundance and biomass of 5.2 × 106 cells L−1 and 270.2 µg C L−1 were recorded at 15 m depth of station HH65 (Fig. 4e), where the diatoms were predominantly Chaetoceros and Fragilariopsis (Fig. 5f, g), with a total contribution of 92.8% and 69.1% in terms of total diatom abundance and biomass.

Table 1 Seasonal changes of cell abundances of phytoplankton taxa in the western Barents sea
Full size table
Fig. 4
figure 4

Vertical and seasonal distributions of cell abundances of diatoms and dinoflagellates

Full size image
Fig. 5
figure 5

Horizontal and seasonal distributions of dominant taxa of diatoms and dinoflagellates (Diamonds display the maximum abundance that occurred in the sampling layers)

Full size image

In summer 2018, the dominant taxa were composed of Chaetoceros, Pseudo-nitzschia, Thalassiosira, Fragilariopsis and Navicula, which on average accounted for 37.8, 31.0, 8.5, 4.1 and 4.7% of abundances, corresponding to 35.1%, 4.9%, 33.3%, 2.8% and 4.9% of biomass. Two water parcels (Fig. 4g) with high abundances and biomass were recorded at station B16 (25 m; 1.1 × 106 cells L−1, 163.2 µg C L−1) and station JR85 (20 m; 1.0 × 106 cells L−1, 61.8 µg C L−1), where the diatoms were dominated by Fragilariopsis (35.2%) and Thalassiosira (21.9%), and Chaetoceros (84.0%) and Fragilariopsis (10.2%) on percentage abundances respectively. In summer 2017, the dominant taxa were composed of Pseudo-nitzschia, Chaetoceros, Thalassiosira, Fragilariopsis and Cylindrotheca, which on average accounted for 48.8%, 14.4%, 8.7%, 6.5% and 6.5% of abundances, corresponding to 18.4%, 18.9%, 27.6%, 5.9% and 2.9% of biomass. Two water parcels (Fig. 4a, c) with abundant diatoms were detected at station B17 (12 m; 6.9 × 105 cells L−1, 281.3 µg C L−1) and station B4 (37 m; 3.3 × 105 cells L−1, 3.2 µg C L−1). Station B17 was dominated by Chaetoceros, Pseudo-nitzschia and Thalassiosira with percentage abundances of 36.7%, 25.7% and 25.6%, whilst station B4 was characterized by Pseudo-nitzschia delicatissima with percentage abundance of 99.9%. Relatively high abundances were also found in sub-surface of station B14 (48 m; 1.2 × 105 cells L−1) and station B9 (30 m; 1.2 × 105 cells L−1), where the diatoms were dominated by Pseudo-nitzschia (58.7%) and Thalassiosira (21.1%), and Pseudo-nitzschia (63.6%) and Chaetoceros (21.7%) respectively. In winter, the diatoms were in very low abundances and biomass, and very few taxa occurred within the samples.

Dinoflagellate composition, abundances and biomass

The mean abundances and biomass of dinoflagellates were high in summer samples of 2018 (1.1 ± 3.5 × 104 cells L−1, 10.0 ± 12.7 µg C L−1) and 2017 (1.0 ± 1.3 × 104 cells L−1, 7.0 ± 7.3 µg C L−1), relatively high in spring samples (4.6 ± 2.4 × 103 cells L−1, 5.2 ± 4.2 µg C L−1), and lowest in winter samples (473 ± 196 cells L−1, 0.16 ± 0.07 µg C L−1). In summer 2018, the dinoflagellates were mainly composed of Gymnodinium, Protoperidinium, Gyrodinium and Katodinium, with mean percentage abundances of 29.3%, 21.6%, 16.4% and 8.2%, corresponding to 23.3%, 28.4%, 13.8% and 2.5% of biomass. The highest abundance and biomass were detected in surface water of station B7 (2.2 × 105 cells L−1, 78.2 µg C L−1) south of Svalbard (Fig. 5o), where the predominant taxa Gymnodinium and Gyrodinium accounted for 61.4% and 11.9% of the total abundance. Relatively high abundance and biomass were also documented in the surface water of station JR77 (2.1 × 104 cells L−1, 25.4 µg C L−1) west of Svalbard, as well as the surface waters (Fig. 4h) of station JR85 (1.9 × 104 cells L−1, 25.7 µg C L−1) and station HH51 (1.1 × 104 cells L−1, 17.9 µg C L−1).

In summer 2017, the dominant taxa were Gymnodinium, Protoperidinium, Prorocentrum, Gyrodinium and Scrippsiella, which on average accounted for 31.5%, 17.8%, 10.9%, 10.3% and 8.4% of abundances, corresponding to 32.7%, 25.2%, 4.2%, 9.1% and 6.7% of biomass. High abundances and biomass were recorded southwest of Svalbard in the surface waters (Fig. 4b) of station B10 (7.9 × 104 cells L−1, 24.5 µg C L−1) and station B9 (5.3 × 104 cells L−1, 30.1 µg C L−1), where the abundance contributions of major taxa were Gymnodinium (49.9%) and Prorocentrum (44.7%), and Protoperidinium (52.1%) and Gymnodinium (29.1%), respectively. Relatively high values (2.2 × 104 cells L−1, 14.4 µg C L−1) also occurred in the surface water of station B16 (Fig. 4d), with 81.2% of the abundance being contributed by dominant taxa of Protoperidinium, Gymnodinium and Scrippsiella. In spring, the dominant taxa were composed of Protoperidinium, Gymnodinium and Gyrodinium, which on average accounted for 35.1%, 18.8% and 12.1% of abundances, corresponding to 38.5%, 21.6% and 13.6% of biomass. No extreme high values of abundances and biomass were detected from the spring samples. In winter, the dinoflagellates declined to the lowest seasonal abundances and biomass, with the presence of very few taxa in the samples.

Distribution of coccolithophores

The distribution of coccolithophores demonstrated strong spatial variation and seasonality. High abundances of intact coccospheres were only recorded in summer 2017 and 2018, with means of 3.2 × 103 cells L−1 and 1.6 × 104 cells L−1, corresponding to calcite mass (CaCO3) of 0.1 µg C L−1 and 0.2 µg C L−1. The taxonomic composition was fairly low in number of species, and only the common species of E. huxleyi (mainly type B) and Coccolithus pelagicus (and its HOL type) were recorded from the two summer samples. High cell abundances and coccolith calcite were measured in summer 2018 in the surface and sub-surface waters of station B13 (Fig. 6), where the intact coccospheres reached up to 3.3 × 105 cells L−1 and 3.2 × 105 cells L−1, corresponding to 4.7 µg C L−1 and 3.4 µg C L−1 of calcite. Abundant coccolithophores were also found in summer 2017 in the surface waters of station B9 (5.4 × 104 cells L−1, 0.7 µg C L−1) and station B10 (1.6 × 104 cells L−1, 0.3 µg C L−1). On average, E. huxleyi overwhelmingly accounted for 95.2% of abundance and 69.8% of calcite in the two summer samples where it occurred.

Fig. 6
figure 6

Vertical distributions of cell abundances of coccolithophores (SEM images show intact cells of Emiliania huxleyi and Coccolithus pelagicus)

Full size image

Biogeographic distributions revealed by cluster analysis

To clarify the spatial distributions of diatoms and dinoflagellates, a cluster analysis was applied to the two summer datasets of taxonomic composition and abundances. The dendrograms clearly separated two distinct phytoplankton assemblages (Fig. 7), the Atlantic water assemblage and the Arctic water assemblage, with percent similarities lower than 11.9% (summer 2017) and 19.9% (summer 2018), showing potential phytoplankton boundaries in the survey region (Fig. 2).

Fig. 7
figure 7

Dendrograms of the cluster analysis according to the percent similarities of diatoms and dinoflagellates (Red cluster indicates the Atlantic water assemblage; Blue cluster indicates the Arctic water assemblage)

Full size image

Contrasting patterns of major taxa, abundances and biomass existed in the two assemblages (Table 2), not only in terms of the spatial distributions but also with respect to interannual variation. In summer 2017, diatoms dominated in the Atlantic water assemblage with means of 1.1 × 105 cells L−1 of abundance and 23.0 µg C L−1 of biomass, which were over 3 and 10 times higher than that of the Arctic water assemblage. We observed Pseudo-nitzschia and Thalassiosira to be the major diatom taxa of the Atlantic water assemblage that on average contributed to the total abundances of 51.6% and 11.0%, and the total biomass of 16.1% and 29.0%. However, the case was different in summer 2018, when the diatom abundance and biomass of the Arctic water assemblage reached up to 4.5 × 105 cells L−1 and 44.2 µg C L−1 on average, respectively, and were over 15 and 12 times higher than that of the Atlantic water assemblage. In our analyses, species of the genus Chaetoceros predominated in the Arctic water assemblage with high average percentage contributions of abundance (71.1%) and biomass (51.8%) of the total for diatoms. In regard to dinoflagellates, high values of abundances and biomass were all detected in the Atlantic water assemblage from the two summer surveys, with means of 1.5 × 104 cells L−1 and 7.7 µg C L−1 in summer 2017, and 1.6 × 104 cells L−1 and 12.1 µg C L−1 in summer 2018. Gymnodinium dominated and on average accounted for 36.7% of abundance and 38.8% of biomass in summer 2017, and 28.8% of abundance and 23.3% of biomass in summer 2018 of the total for dinoflagellates.

Table 2 Comparison of cell abundances and biomass of phytoplankton taxa between the Atlantic water assemblage and the Arctic water assemblage
Full size table

It is noteworthy that some species in the genera Thalassiosira and Chaetoceros may possess different ecological and biogeographic features in terms of ecotypes, such as T. nordenskioeldii (Arctic species) and C. socialis (Atlantic species). In this study, however, the contributions of these species to the total abundance of genus were quite small at the opposite side of the Polar Front. The average percentages of T. nordenskioeldii in Thalassiosira were zero for summer 2017 and only 10.4% for summer 2018 in the Atlantic waters. The highest abundances of T. nordenskioeldii were all recorded from the Arctic waters, with 2.3 × 104 cells L−1 (48 m, B14) in summer 2017 and 2.2 × 105 cells L−1 (25 m, B16) in summer 2018, respectively. Similarly, the high abundances of C. socialis were only recorded from the Arctic area greatly influenced by the Atlantic water, with 8.6 × 105 cells L−1 (20 m, JR85), 7.7 × 105 cells L−1 (1.7 m, JR85) and 7.3 × 105 cells L−1 (10 m, JR85) in summer 2018. Warm water species of the genus Tripos (T. furca, T. fusus, T. muelleri) was also recorded from the Atlantic waters (B2 and B13) in two summers, but with low abundances (maximum of 120 cells L−1).

Multivariate ordinations on diatoms and dinoflagellates with environmental conditions

To clarify the distributions of diatoms and dinoflagellates in association with the ambient environmental conditions, multivariate ordinations were applied to distinguish how much of the variation in the taxonomic composition could be attributed to changes in the environmental factors. We used six available environmental parameters of PAR, temperature, salinity, nitrate, phosphate and silicic acid, representative of the light conditions, physical seawater properties and nutrient concentrations, as the explanatory variables. The linear ordination of RDA was chosen for the subsequent analysis on diatoms and dinoflagellates based on the initial DCA results (Table 3), of which the longest lengths of gradient in the composition data were all less than 4.

Table 3 Multivariate analysis results of DCA and RDA (Bold texts highlight that the longest lengths of gradient are all less than 4 in the DCA)
Full size table

Regarding diatoms in summer 2017, spring and summer 2018, the first two canonical axes collectively explained 71.4%, 96.4% and 86.8% of the species-environment relationship, and 20.1%, 75.9% and 48.9% of the percentage variance in the taxonomic composition data, of which the first axis accounted for 13.3%, 63.7% and 41.4%, respectively. According to the Monte Carlo permutation tests of significance, temperature was a significant (p < 0.01) explanatory variable for the diatoms, explaining 10.1%, 50.3% and 38.6% (contributing 35.6%, 63.9% and 68.6%) of the taxonomic composition variance in summer 2017, spring and summer 2018, respectively. The dominant diatom taxa, such as Chaetoceros and Thalassiosira, showed clear and strong negative correlations with the temperature gradient in the biplots (Fig. 8a, c, e). In contrast with the high weight allocation to temperature independently, the nutrients of nitrate, phosphate and silicic acid accounted for a combined contribution of 36.4%, 12.4% and 19.4% in summer 2017, spring and summer 2018, respectively.

Fig. 8
figure 8

Biplots of RDA ordinations on major taxa of diatoms and dinoflagellates along the gradients of physico-chemical factors (Percentage values in parentheses refer to the contributions of explained variance in taxa composition by environmental variables; p values refer to the significance of environmental variables by Monte Carlo permutation tests; Environmental vectors in blue colour show close correlation with dominant taxa of each season)

Full size image

With respect to dinoflagellates in summer 2017, spring and summer 2018, the first two canonical axes explained 77.3%, 73.1% and 66.5% of the overall species-environment relationship, and 25.5%, 36.9% and 22.3% of the variances in the taxonomic composition data, of which the first axis accounted for 20.2%, 20.3% and 14.2%, respectively. Although the biplots depict that silicic acid accounted for 37.0% (summer 2017), phosphate accounted for 36.4% (spring) and salinity accounted for 32.4% (summer 2018) in the explanations of the overall distribution of dinoflagellates, it is conspicuous that the predominant taxa of Gymnodinium, Gyrodinium, Prorocentrum, Protoperidinium showed strong positive correlations with PAR in the analysis (Fig. 8b, d, f).

Discussion

Influence of physical seawater conditions on phytoplankton distribution

It is well recognized that, phytoplankton blooms and annual productivity in the Arctic are closely associated with the marginal ice zone (Wassmann et al. 1999; Arrigo et al. 2012; Makarevich et al. 2022), particularly near the ice edge (Perrette et al. 2011). Phytoplankton productivity benefits from favourable conditions such as the sunlit surface, greater stability of the underlying water column, as well as the nutrient inventory from winter accumulation, vertical mixing and regional upwelling. This primary biological process may thus greatly influence on the secondary production (Scott et al. 1999) and transfer of energy to high trophic levels of marine food webs (Wassmann et al. 2006). However, the marginal ice zone can be either abrupt or be extended over hundreds of kilometers by physical forcing (Sundfjord et al. 2007), and is thus highly variable in space and time and ice characteristics, both seasonally and interannually. In summer 2017 and 2018, the sea ice conditions in the sampled region in the north of the Barents Sea were quite different (Fig. 1b, e), with ice coverage north of station B14 in summer 2017 and mostly ice-free water in summer 2018. According to the source data (NSIDC 2023), the monthly averaged sea ice extent in the Barents Sea in summer 2018 was only 58.1% (June), 25.7% (July) and 0.1% (August) of that in summer 2017. Although the contrasting ice conditions and increased exposure to open waters could be one of the factors for the increases of phytoplankton abundance in summer 2018, we managed to capture an ice-edge bloom from spring declining and then transforming into a sub-surface maximum in early summer 2018. Most of the stations east of Svalbard were still ice-covered two weeks before we visited and the northerly stations of JR80 (81.9°N) and JR85 (82.6°N) were at the ice edge (Cottier 2019). So, this close link with the marginal ice zone in summer 2018, is likely to be the overarching driver of why the mean abundances of diatoms, dinoflagellates and coccolithophores were 2.8, 1.1 and 5.2 times higher than those in summer 2017.

In spring 2018, the diatom bloom was observed near the southern border of the marginal ice zone (Figs. 1d and 5) with a peak density of over 5 × 106 cells L−1. It is noteworthy that, in addition to the common bloom taxa of Chaetoceros, Fragilariopsis and Thalassiosira in Arctic seas (Wassmann et al. 1999; Arrigo et al. 2012; Makarevich et al. 2021), we found that the sympagic algae, Fossulaphycus, was also abundant in this bloom with a maximum abundance of 3.4 × 105 cells L−1 and biomass of 31.6 µg C L−1. However, this genus was only detected in spring, and was not recorded in other seasons, including the early summer 2018, suggesting that this ice algae does not seed the spring-summer phytoplankton transition. Interestingly, the pelagic taxa of Chaetoceros (Fig. 5f, k), Fragilariopsis (Fig. 5g, l) and Thalassiosira (Fig. 5i, n), were observed to follow the retreat of the sea ice in the Arctic water to the north at the continental slope near the Nansen Basin through the summer. Their vertical distributions were characterized by marked sub-surface maxima, implying a northward and deep propagation of these dominant spring bloom taxa. An Arctic-like space-for-time seasonal variability in chlorophyll a was also found in the south-north direction (deep maxima to the south and shallow to the north), with the summer situation progressing northwards following sea ice retreat (Koenig et al. 2024). This bloom propagation in the northern shelf of the Barents Sea in summer 2018, triggered high abundances and biomass of diatoms in the Arctic water assemblage (Table 2), which were on average 13.3 and 19.9 times higher than those in summer 2017 (in a later stage of phytoplankton seasonal progression).

Temperature and light are important bottom-up factors in controlling the growth of phytoplankton (Fernández‑González and Marañón 2021), magnitude of blooms (Popova et al. 2010), as well as their physiological adaptation (Lewis et al. 2019). Phytoplankton niches estimated from field data have also established that temperature and mean irradiance are important predictors of individual species presence during the seasonal succession of phytoplankton community; diatoms tend to be found in habitats of cooler waters and lower mean irradiance, while dinoflagellates tend to favour warmer waters with higher mean irradiance (Irwin et al. 2012; Brun et al. 2015). Concurrent with increased heat fluxes from the Atlantic water masses northwards through the Barents Sea (Årthun et al. 2012), more and more areas in the Arctic are exposed to sunlight for longer periods each year because of the receding sea ice. Light has thus become less of a limiting resource, and field evidence illustrates that Arctic phytoplankton can efficiently exploit the irradiance of high-light in summer and low-light in spring by regulating light absorption and photosynthetic capacity through photo-acclimation (Gallegos et al. 1983; Lewis et al. 2019), even in the polar night during winter (Randelhoff et al. 2020). In this study, RDA biplots clearly demonstrated the contrasting ecological niches for the seasonal succession of phytoplankton taxonomic composition (Fig. 8). Temperature significantly (p < 0.01) explained a greater proportion of variance in diatom composition than the other explanatory variables, suggesting that it was one of the most important factors in driving the seasonal patterns of diatoms, particularly in spring (explained variance of 50.3%, independent contribution of 63.9%) when diatom blooms occurred in the stratified waters of the Arctic water mass (Fig. 4e). Moreover, the predominant taxa of diatoms, in combination with most of the other diatoms, showed clear and strong negative correlations with the temperature gradients in the spring and the two summers. Meanwhile, the dominant taxa of dinoflagellates such as Gymnodinium and Protoperidinium, showed unique strong positive relation to PAR (a proxy for ice-free, stratified and oligotrophic conditions) in the ordination biplots (Fig. 8), and tended to flourish in summer surface waters (Fig. 4), indicating the knock-on effect of enhanced light intensity on shaping the distributions of these taxa. Interestingly, some dinoflagellates, such as Prorocentrum minimum, which was also abundant in the samples, may behave mixotrophically (Stoecker and Lavrentyev 2018), and could become an important component of the phytoplankton community at the end of the bloom in late summer (highest abundance of 3.5 × 104 cells L−1 in late summer 2017), when the diatom cells began to die and sink out of the surface water, allowing greater amounts of lights through. The different realized ecological niches of these two groups of phytoplankton could be separated by evolutionary selection, in terms of adaptive strategies, stress tolerance, as well as fundamental trait-based (morphological, functional, life-history) ecophysiological characteristics (Reynolds 2006; Fragoso et al. 2018).

Influence of seawater nutrient conditions on phytoplankton distribution

Regardless of the physical conditions governing large-scale patterns of phytoplankton biogeography (Brun et al. 2015), nutrient availability is important in supporting primary production, especially in regulating the longevity and total magnitude of the phytoplankton bloom (Henley et al. 2020). Nutrient replenishment in the euphotic zone of the Barents Sea is greatly dependent on the advection of Atlantic water inflows, vertical mixing (wind induced, tidal forcing, thermal convection), as well as wintertime shelf break upwelling (Randelhoff et al. 2018; Randelhoff and Sundfjord 2018), with the drawdown of the nutrient pools being governed by the magnitude of spring blooms and the timespan of the subsequent growing season through the summer. Our field observations showed that, in the waters (15 m) with highest abundance and biomass of diatoms on 27th April 2018 (Fig. 4e), the nitrate, phosphate and silicic acid had already been drawn down to relatively low concentrations of 3.39 µmol L−1, 0.30 µmol L−1 and 0.26 µmol L−1, suggesting their fast utilization by the spring bloom. Then, following the northward and deep propagation of phytoplankton, the nutrient inventory in the upper water column continued to decrease through the summer (Fig. 3), in consistent with a meridional observation for dissolved inorganic nutrients in the northwestern Barents Sea shelf (Koenig et al. 2024), as well as a longer oligotrophic summer period (Kohlbach et al. 2023). A conspicuous sub-surface maximum of diatoms was found to occur over the north shelf from station HH51 to station M East (Fig. 4g), where the concentrations of nitrate, phosphate and silicic acid in the upper waters (< 30 m) were measured to be only 0.34 ± 0.24 µmol L−1, 0.03 ± 0.03 µmol L−1 and 0.05 ± 0.03 µmol L−1 respectively. This matched with the time-series findings from a nearby mooring (Henley et al. 2020), with a rapid overall nitrate drawdown rate of 0.67 ± 0.03 µmol L−1 d−1 during the first half of May. In contrast with the almost depleted nutrient conditions to the east of Svalbard in summer 2018, the station M West (Fig. 1e) to the north of Svalbard, was still relatively rich in nutrients (< 30 m), with nitrate, phosphate and silicic acid concentrations of 1.38 ± 0.53 µmol L−1, 0.19 ± 0.10 µmol L−1 and 1.85 ± 0.09 µmol L−1, indicating episodic mixing during summer at the shelf break. However, in comparison with the well stratified waters around station B16, the abundance and biomass of diatoms at station M West were extremely low, with only 1.0% and 0.4% of that at station B16, potentially because of the unstable waters at station M West driven by Atlantic water advection along the northern slope from the west.

According to the RDA results (Fig. 8), most of the variances of diatom composition in spring and two summers were explained by water column temperature, the key factor in determining the timing of the initiation of the spring bloom through forming haline stratification by sea ice melt, and the summer sub-surface maxima through developing thermal stratification by solar radiation. Although the aggregated explanations of variances by nutrients were relatively low in comparison with that of the temperature, it is undoubted that, the accumulation of vast biomass of phytoplankton in growing seasons relies greatly on the fueling of nutrients (Fig. 3), which are then converted into proteins, fats, and carbohydrates by photosynthesis. Thus, we support the previous long established paradigm in the Arctic (Harrison and Cota 1991; Reigstad et al. 2002; Ardyna et al. 2020b), that the role of macronutrients is pivotal in sustaining the spring-summer growing seasons, rather than controlling the timing of the blooms.

Influence of water masses on phytoplankton distribution

Water masses, generally, control all aspects of the physical, chemical and biological factors for realized niches of phytoplankton. Water mass transportation, transformation and potential climate-related changes, could thus fundamentally influence the biogeography of various functional groups, such as diatoms, dinoflagellates and coccolithophores. In the western Barents Sea, there are several distinct water masses including modified and intermediate waters (Oziel et al. 2016; Sundfjord et al. 2020), of which the Atlantic and Arctic waters occupied a greater proportion of the volume during our study. In this study, two distinct phytoplankton assemblages, closely linked with the Atlantic water and Arctic water, were geographically separated by the Polar Front (Fig. 1), and were clearly displayed in the cluster analysis dendrograms on a percent similarity of lower than 11.9% (Fig. 7). Diatoms flourished in a habitat of cold and fresh Arctic waters, whilst dinoflagellates thrived in the warm and saline Atlantic waters, suggesting a governing effect of the two water masses on large-scale distributions of phytoplankton in the western Barents Sea. Water mass transportation by the Atlantic water inflow, originates from the south along the isopycnal surface above the pycnocline onto the shelf of the Barents Sea. These Atlantic waters carries heat, salt, nutrients, as well as the temperate bio-components, such as coccolithophores (Neukermans et al. 2018; Orkney et al. 2020; Oziel et al. 2020), zooplankton (Gluchowska et al. 2017) and jellyfish (Mańko et al. 2022), through bio-advection (Vernet et al. 2019), a crucial process for the movement of temperate biota far into the north. There are three main branches of Atlantic water inflow; the first one is the West Spitsbergen Current via the Fram Strait, the second one flows through the Barents Sea Opening and the third one is Svalbard Branch towards Kvitøya (White Island). In this study, we found that three regions were habitats of abundant coccolithophores (Fig. 6), which could be used as an Atlantic water tracer. The first region was the waters surrounding station B9 southwest of Svalbard, where the West Spitsbergen Current mixes with the Arctic current (Fig. 2a, b). The second region was located in the southern Barents Sea waters adjacent to station B13, where the North Cape Current approaches to the southern Polar Front at station B14 (Fig. 2i, j). The third region was the waters adjacent to Kvitøya. Thus, the patchy distributions of coccolithophores was highly related to the three branches of the Atlantic water mass. It is interesting that, although in very low numbers, we recorded intact cells of E. huxleyi at station B17 (12 m) and even at station B16 (15 m) in summer, probably caused by transport of the Svalbard Branch (Figs. 1a and 2c, d) through the Kvitøya trough (Lundesgaard et al. 2022), further confirming the bio-advection of coccolithophores by Atlantic water inflow.

The European Arctic has been experiencing the process of Atlantification as part of borealization of the Arctic (Polyakov et al. 2020), by the advection of Atlantic water inflow along the continental slope north of Svalbard and over the shelf of the Barents Sea from the south. Accompanied by the sea ice withdrawal (Carmack et al. 2015; Polyakov et al. 2017), a series of physical, chemical and biological changes have occurred, including enhanced surface irradiation, salinification and decreased stratification (Lind et al. 2018; Polyakov et al. 2020), increased nutrient availability (Henley et al. 2020), as well as elevated primary production (Slagstad et al. 2015; Ardyna and Arrigo 2020), particularly in the northern Barents Sea. Consequently, these processes will accelerate the transition of the Arctic marine ecosystem towards a more Atlantic state. One piece of evidence of this has been provided by the indicator species E. huxleyi, whose surface bloom area has nearly doubled between 1998 and 2016 in the Barents Sea (Oziel et al. 2020). In this study, we recorded high abundance (exceeding 3 × 105 cells L−1) of E. huxleyi in waters around 75°N, which was approaching the previous satellite-derived maximum northern boundary (76°N) of its areal extent. Although the bloom centre (> 106 cells L−1) is generally located in the southern Barents Sea (Silkin et al. 2020a, b), this high abundance of E. huxleyi far into the north, in areas that were previously occupied by the Arctic water phytoplankton, may alter the carbon export and biogeochemical cycles by an enhanced calcite ballast effect (Klaas and Archer 2002). Moreover, we found that the population of E. huxleyi was predominated by type B (Fig. 6e), a morphotype which was first designated from seawaters of the North Sea (Van Bleijswijk et al. 1991), suggesting its ability to adapt and move northward in the Barents Sea, through polymorphism into an alternative phenotype. Such a pronounced poleward expansion of boreal species ranges (Fossheim et al. 2015; Gluchowska et al. 2017; Oziel et al. 2020) may also alter spatial distributions and seasonal rhythms of local communities, which will exert further stress on the whole Arctic food web in the context of regional climate change (Lannuzel et al. 2020).

Conclusion

This study provides present-day field observational evidence of the spatial and seasonal distributions of diatoms, dinoflagellates and coccolithophores and their association with different water masses in the western Barents Sea. Two distinct phytoplankton assemblages, the Atlantic water assemblage and the Arctic water assemblage, were shaped by these two water masses and separated by the Polar Front. Physical seawater conditions showed a governing effect on seasonal patterns of phytoplankton biogeography, while nutrients were pivotal in supporting the growing seasons of phytoplankton production. As one of the most productive inflow shelf seas in the Arctic Ocean, the Barents Sea is undergoing fast Atlantification. Characterisation of the distributions of local biota and bio-components being advected from temperate waters is thus crucial in understanding the changing Arctic marine ecosystem. More long-term and multi-seasonal observational campaigns such as this need to be undertaken in the future. These are necessary to support modeling efforts on the ecological importance of changing phytoplankton communities, their impacts on the biogeochemical cycles, as well as their cascading effects throughout the wider Arctic marine ecosystem in a fluctuating climate.