Marine Plankton during the Polar Night: Environmental Predictors of Spatial Variability

Simple Summary Plankton are a key component of Arctic marine ecosystems, connecting all trophic levels and being involved in organic matter recycling. Our main purpose was to reveal differences in plankton characteristics in two Arctic sites that were strongly affected by the inflow of warm Atlantic waters during the period of polar night. We detected similar bacterial abundances in both regions, while other plankton parameters were significantly different. Temperature, salinity, and sampling depth shaped the plankton communities. Phytoplankton biomass also had a marked influence on microbial abundance and zooplankton assemblages. Changes in the proportions of boreal taxa suggested the Atlantification of pelagic communities, a phenomenon reported in the Arctic due to global warming. Abstract We studied the spatial patterns of the planktonic ecosystems at two Arctic sites strongly affected by Atlantic Inflow (FS, the Fram Strait; and BS, the Barents Sea). A high degree of similarity in the bacterial abundance (mean: 3.1 × 105 cells mL−1 in FS vs. 3.5 × 105 cells mL−1 in BS) was found, while other plankton characteristics were different. Bacterial biomass reached a maximum in BS (3.2–7.9 mg C m−3), while viral abundances tended to be higher in FS (2.0–5.7 × 106 particles mL−1). Larger bacterial cells were found in BS, suggesting the presence of different bacterial populations at both locations. The virus-to-bacteria ratio was significantly higher in FS than in BS (13.5 vs. 4.7). Chlorophyll a concentration was extremely low (<0.25 mg m−3). The highest zooplankton abundance was in the surface layer (919 individuals m−3 in FS vs. 602 ind. m−3 in BS). Zooplankton biomass strongly varied (1–39 mg C m−3), with the maximum in BS. High proportions of boreal taxa in the total zooplankton abundance indicate the Atlantification of pelagic ecosystems in the Arctic. Plankton indicators are correlated with temperature, salinity, and sampling depth. Strong intercorrelations were found between major plankton groups, suggesting tight links in the studied plankton ecosystems.


Introduction
The Arctic Ocean and adjacent marginal seas have been documented to be strongly affected by climate changes observed during the past few decades [1]. The global water temperature has demonstrated an increasing trend since the 1970s that caused gradual warming in the Arctic [2][3][4][5]. Sea ice extent has rapidly decreased in the Arctic seas, and earlier ice melting and the northward retreat of ice cover have greatly impacted marine environments and biota [6]. Total net primary production (NPP) and chlorophyll a values may be considered to be integral indicators of marine ecosystems in the Arctic, and these were reported to have increased in the last few decades [7][8][9]. Moreover, this trend is proposed to continue in the future [10,11]. Such climatic fluctuations have led to the alteration of pelagic assemblages and food web structures in the Arctic [12][13][14][15]. The most prominent consequence is the borealization of the Arctic biota, a process widely documented in recent reports [6,[16][17][18][19]. Many taxa were found to expand their ranges, and some boreal taxa were recorded in more northern regions, due to ocean warming [6,16]. The total biomass of krill and fishes has increased in the Barents Sea (BS) from the 1990s to the 2000s [15,16,20]. The boreal copepods and euphausiids have expanded northwards, while some Arctic species have declined in range and biomass, and have moved further north [9,20]. In the Fram Strait, the boreal copepod Calanus finmarchicus transported with Atlantic inflow began to develop and grow faster, owing to higher temperatures [21]. Zooplankton biomass was found to be negatively correlated with the temperature of the Atlantic Water, and it decreased as the abundances of small zooplankton taxa increased [5,22,23].
Plankton are a major component of any aquatic ecosystem, playing a crucial role in biogeochemical cycles in the World ocean [24,25]. In the Arctic, the phytoplankton growth season is short and is controlled by the annual cycle of light regime and nutrients, as well as by the sea ice dynamics [26][27][28]. Sea-ice melting triggers an annual spring phytoplankton outburst which is dominated mainly by diatoms [29,30]. Much of the primary production formed in the spring is sequestered in the benthos [31,32], supporting extremely rich benthic populations which are utilized by higher trophic-level taxa [27,28]. The summer stratification of the upper layer declines the vertical replenishment of surface nutrients, resulting in growing biomasses of small flagellates [29,30]. The Arctic plankton foodwebs are considered to be short (from diatoms to top predators) and very sensitive to environmental forcing [26]. Zooplankton are thought to be an important intermediate link connecting primary producers to higher trophic levels [9,24,25]. Microbial assemblages have a great contribution to the total productivity, being involved in the pelagic foodweb as producers (autotrophic microbes), consumers, and transformers of dissolved organic matter (heterotrophic protists and bacteria) [26,29,33]. Viruses also play a key role in controlling microbial populations, and they are responsible for a significant part of prokaryotic and phytoplankton mortality [34]. Virioplankton affect marine nutrient cycling and the transport of organic matter through the 'viral shunt' [35,36].
BS and FS are two of the most productive regions in the Arctic [10,27], having enhanced primary production relative to other Arctic sites. The Southern BS and the Western FS support rich pelagic and benthic assemblages [38][39][40][41][42][43]. These systems have been intensively investigated since the 1900s [26][27][28]37]. Despite significant research efforts in studying plankton in BS and FS, some important issues remain poorly explored. Plankton communities exhibit clear seasonal changes associated with fluctuations in environmental variables (nutrient concentration, light intensity, grazing impact, predation pressure, and hydrological conditions). A bulk of previous studies has been focused on the spring-autumn period [44][45][46][47][48][49][50][51][52][53][54][55][56][57][58][59][60] while the winter season is less well studied due to severe environmental conditions (the presence of ice cover, low temperatures, storms, and polar night). Recent publications have reported the activities of pelagic communities under such extreme conditions [45,[61][62][63][64][65]. To adequately evaluate the seasonal dynamics of the plankton in the Arctic, complex winter investigations focused on the main groups (from viruses to zooplankton) are strongly needed.
Here, we present a comparative analysis of the plankton collected in two Arctic regions during the period of the polar night. Our study aimed to reveal the spatial distributions of virio-, bacterio-, phyto-, and zooplankton in the Southern BS and in the Eastern FS. Considering the strong Atlantic influence at both sites, we hypothesized that the plankton abundance and biomass would demonstrate similar regional patterns. Twoway PERMANOVA was selected to test our hypothesis regarding the similarity of plankton abundances in the two Arctic regions. Another goal of our work was to evaluate the main factors driving the plankton abundance in the winter season. To date, there are no previous studies regarding marine viruses in FS during the polar night, and our study is the first report on this plankton group. Additionally, we obtain novel data regarding other major microplankton in the Arctic marine environment.

Sampling and Processing
Oceanographic data, microbial, nutrient, and zooplankton samples were collected during a multidisciplinary cruise on board the R/V Dalnie Zelentsy during the winter season (24 November to 18 December 2021) ( Figure 1, Table 1). Two transects were investigated in FS (3 stations) and in BS (5 stations).
under such extreme conditions [45,[61][62][63][64][65]. To adequately evaluate the seasonal dynamics of the plankton in the Arctic, complex winter investigations focused on the main groups (from viruses to zooplankton) are strongly needed.
Here, we present a comparative analysis of the plankton collected in two Arctic regions during the period of the polar night. Our study aimed to reveal the spatial distributions of virio-, bacterio-, phyto-, and zooplankton in the Southern BS and in the Eastern FS. Considering the strong Atlantic influence at both sites, we hypothesized that the plankton abundance and biomass would demonstrate similar regional patterns. Two-way PERMANOVA was selected to test our hypothesis regarding the similarity of plankton abundances in the two Arctic regions. Another goal of our work was to evaluate the main factors driving the plankton abundance in the winter season. To date, there are no previous studies regarding marine viruses in FS during the polar night, and our study is the first report on this plankton group. Additionally, we obtain novel data regarding other major microplankton in the Arctic marine environment.

Sampling and Processing
Oceanographic data, microbial, nutrient, and zooplankton samples were collected during a multidisciplinary cruise on board the R/V Dalnie Zelentsy during the winter season (24 November to 18 December 2021) ( Figure 1, Table 1). Two transects were investigated in FS (3 stations) and in BS (5 stations).  A CTD profiler, SEACAT SBE 19plus V2, was used to determine the oceanographic values and the water column structure. Sampling for microbial plankton, nutrients (nitrate, phosphate, silicate, and dissolved oxygen), and chlorophyll a (Chl-a) was carried out with 10 L Niskin bottles mounted on a Rosette system.
Five to seven fixed-depth layers (0 m, 10 m, 25 m, 50 m, 100 m, 200 m, and near the bottom), or less if the bottom was shallower, were sampled. Water samples for nutrient analyses (120 samples) were immediately filtered through a 0.45 µm polycarbonate membrane (FMPA, Vladisart, Vladimir, Russia), and kept frozen in liquid nitrogen in polyethylene 20 mL bottles and stored at −20 • C until their analysis in the Murmansk Marine Biological Institute laboratory. Chl-a samples (2-5 L for each sampling stratum) were filtered through 0.6 µm Vladiopore filters and then frozen (−20 • C) until the analysis. To estimate bacterial abundance and biomass, water samples were collected into plastic vials and preserved by adding a pre-filtered formaldehyde solution to a final concentration of 2%. Bacterial samples were kept in slide boxes at −20 • C. A total of 43 and 81 samples of Chl-a and marine microbes were collected, respectively (Table 1).
After the Rosette being taken onboard, vertical zooplankton hauls were made using a 50 cm mouth diameter closing WP-2 net with a mesh size of 180 µm. The water volume filtered through the net during sampling was estimated with a calibrated flowmeter (Hydrobios). Sampling was performed at two sampling layers (0-50 m, and 50 m or near the bottom) to evaluate the variability in the total abundance in the upper and sub-surface strata. The samples were preserved using a 4% formalin solution in seawater buffered with borax. In total, 16 zooplankton samples were obtained.

Laboratory Procedures
The determination of nutrients was performed in accordance with a standard manual [66]. Dissolved oxygen content was determined following the Winkler method using an automatic-endpoint-detection burette, Digital Burette VITLAB (63762 Grossostheim, Germany). The precisions of the hydrochemical analyses were 0.1 µM (nitrate and silicate) and 0.04 µM (phosphate). In the laboratory, the Chl-a filters were extracted in 90% acetone, kept in a freezer at 4 • C for 24 h, and centrifuged, following a standard procedure [67]. The fluorescence was determined with a Nicolett Evolution 500 spectrophotometer (Spectronic Unicam, Scotia, NY, USA), previously calibrated by the manufacturer. Chl-a content (mg m −3 ) was used as a measure of the total phytoplankton biomass.
The microscopic analysis of bacterioplankton samples was made as follows: subsamples were incubated with DAPI for 8-10 min and filtered through 0.2 µm Nuclepore filters at a vacuum pressure of 10-12 cm Hg [68]. The filter films were dried and then fixed to glass slides with fluorescence-free immersion oil and stored at −30 • C until analysis. An OlympusBX 53 epifluorescence microscope was used to examine the DAPI-stained samples. Approximately 400-500 bacterial cells were counted per film in 20 fields randomly selected for counting at a 1000× magnification. The bacterial abundance was expressed as the number of cells per 1 L. The dimensions of 30-50 bacterial cells in each sample were measured to calculate the average bacterial cell volume (ABV-average bacterial cell volume, µm 3 ). The carbon biomass (C, fg cell −1 ) was estimated according to Norland [69].
Virioplankton samples were stained with SYBR Green I fluorochrome (Molecular Probes, Eugene, OR, USA) [70]. A volume of 0.5-1.0 mL of seawater was filtered through 0.02 µm Anodisc aluminum oxide membrane filters (Whatman). For the counting of viruses, the films were examined under an OlympusBX 53 epifluorescence microscope at a magnification of 1000×. View fields were randomly selected and enumerated until the total counts exceeded 200 particles. The ratio of viral to bacterial abundance (VBR) was calculated for each sampling layer.
The zooplankton samples were divided in the laboratory using a Folsom splitter. Sub-samples (1/16-1/32 of the total sample) were examined using an MBS-10 stereomicroscope at a magnification of 32-56×. The organisms were identified to the lowest possible taxonomical levels, in accordance with our previous studies [22,23,[55][56][57][58][59]71]. Zooplankton abundance for each sampling layer was expressed as individuals m −3 . Biomass was the calculated length-mass or the published mean individual wet, dry, or carbon masses [72][73][74][75]. All values were then presented as mg carbon mass (DM) per 1 cubic meter according to the equations: 1 mg wet mass = 0.04 mg dry mass = 0.02 mg C for gelatinous zooplankton, and 1 mg wet mass = 0.2 mg dry weight = 0.1 mg C for other taxa [76].

Statistical Analyses
Descriptive statistics (ranges and means with standard deviations) were calculated for the oceanographic and biological datasets. The data normality was checked with a Kolmogorov-Smirnov test, and a modified Levene's test was used for checking the homogeneity of variances. Principal component analysis (PCA) and one-way PERMANOVA [77] were performed to identify the similarity in environmental conditions (temperature, salinity, and nutrient concentrations) between stations. The Pearson correlation was used to calculate the resemblance matrix for the environmental data in PCA [78]. Two-way PER-MANOVA with two fixed factors was applied to test for differences in plankton variables (microbial abundance and biomass, AVB, VBR, Chl-a concentration, and zooplankton abundance and biomass) between the two regions and among sampling layers. Pair comparisons of each environmental or biological variable were made with one-way ANOVA or Kruskal-Wallis tests (in the case of non-normal data distribution) [78]. Descriptive statistics, PCA, PERMANOVA, ANOVA, and Kruskal-Wallis tests were performed in PAST 3.22 [79].
The zooplankton community was analyzed using multivariate techniques. The Shannon-Wiener index (H') [80] and Pielou's evenness (J) [81] were calculated to estimate the diversity of the zooplankton assemblages.
The zooplankton dataset was lg(x+1)-transformed, and the Bray-Curtis similarity index was calculated for all stations [82]. Hierarchical cluster analysis (group-average linkage method) was used to explore patterns in the zooplankton community structure [78]. An analysis of similarity (ANOSIM) was applied to test for differences between groups of samples. The similarity percentage (SIMPER) routine was carried out to identify the percent contribution of zooplankton taxa to the cluster. All analyses were performed using Primer version 5.2.3 software [83].
The detrended correspondence analysis (DCA) was performed to select between canonical correspondence and redundancy analyses with Canoco version 4.5.6 software [84]. Rare species (occurrence in samples < 20%) were removed from the dataset to reduce the influence of outliers before using DCA. The longest gradients of the (DCA) ordination axis were 0.541-0.945, suggesting that redundancy analysis (RDA) was suitable for investigating the relationship between zooplankton and environmental variables [85]. To reduce the impact of double zeros, we used a lg(x+1)-transformation to the abundance of common zooplankton taxa prior to analysis. Significant environmental variables in the RDA were selected with the Monte Carlo permutation test (999 permutations) [78]. Collinearity between variables was assessed with the variance inflation factor (VIF), which estimates the influence of collinearity among the variables [85].
The relationships between biotic variables and the measured environmental parameters were examined with Statistica version 10.01011.0 software (GLZ-generalized linear models). Being less sensitive regarding the normality and homoscedasticity of data, GLZ is more suitable than GLM. All input variables were lg(x+1)-transformed to fit normality [86]. Therefore, a logarithm link function and normal distribution (f(z) = log(z)) were applied to estimate the effects of all independent variables [87]. The best models were selected based on the Akaike Information Criterion (AIC). To validate the predicted regression models, the Wald statistic was applied to check the significance of r-coefficients for input parameters. A likelihood ratio test was used to assess the statistical significance of including variables. The goodness of fit for each model was evaluated with deviance and Pearson χ2. Relationships between biotic variables were also tested with GLZ.
Contour plots of environmental and biological variables were made using the kriging as a gridding method in Surfer 8.0 (Golden Software Inc., Golden CO, USA).

Hydrology
Both regions had no ice cover during the study period. Sampling stations in FS were mainly affected by relatively warm AW (Figure 2a There was a spatial gradient in the surface water temperature; it tended to decrease from south to north (Figure 2a). The salinity demonstrated a slight variation in the whole water column, with lower values in the upper 10 m layer (Figure 2b). In BS, AW was also present at all sampling stations (Figure 3a,b). There was no clear vertical stratification according to the hydrological properties in the water column (Figure 3a,b). The water temperature increased towards the surface, while salinity showed an inverse pat-  There was a spatial gradient in the surface water temperature; it tended to decrease from south to north (Figure 2a). The salinity demonstrated a slight variation in the whole water column, with lower values in the upper 10 m layer ( Figure 2b). In BS, AW was also present at all sampling stations (Figure 3a,b). There was no clear vertical stratification according to the hydrological properties in the water column (Figure 3a,b). The water temperature increased towards the surface, while salinity showed an inverse pattern (Figure 3a,b). There were clear differences in the salinity between regions (ANOVA or Kruskal-Wallis test, p = 0.001-0.008) for each sampling layer. The temperature was similar among the regions (ANOVA or Kruskal-Wallis test, p = 0.160-0.680).

Dissolved Oxygen and Nutrients
In FS, the dissolved oxygen concentration was high below 100 m, ranging from 9 to 10.8 mL L −1 (Figure 2c In BS, the dissolved oxygen content (5.8-10.7 mL L −1 ) slightly increased from the north to the south, and this pattern was more evident for the upper 50 m (Figure 3c). Nutrients were concentrated mainly below the 100 m layer, except for the two north-

Dissolved Oxygen and Nutrients
In FS, the dissolved oxygen concentration was high below 100 m, ranging from 9 to 10.8 mL L −1 (Figure 2c In BS, the dissolved oxygen content (5.8-10.7 mL L −1 ) slightly increased from the north to the south, and this pattern was more evident for the upper 50 m (Figure 3c). Nutrients were concentrated mainly below the 100 m layer, except for the two northernmost stations (Figure 3d (Table S3) and demonstrated the same pattern as the abundance (Figure 4b). In BS, the total bacterial density and biomass were 2.3 × 10 5 -6.8 × 10 5 cells mL −1 and 3.2-7.9 mg C m −3 , respectively, increasing from the north to the south in the surface and intermediate layers (Figure 5a,b). In both regions, the abundance and biomass of bacteria declined towards the bottom (Figure 4a,b and Figure 5a,b). Significant regional differences in the total bacterial biomass for the bottom and 0-bottom layers were revealed (Table S3). Bacterial abundances were comparable in both regions (Table S3).
The virus-to-bacteria ratio (VBR) measured from the total bacterial and viral abundances varied significantly (Kruskal-Wallis test, p = 0.001) between the study regions, with a greater VBR (10.7-18.7, 13.5 ± 4.6) observed in FS. In BS, VBR fluctuated from 3.0 to 9.2, averaging 4.7 ± 2.2.  total bacterial density and biomass were 2.3 × 10 -6.8 × 10 cells mL and m −3 , respectively, increasing from the north to the south in the surface and layers (Figure 5a,b). In both regions, the abundance and biomass of bacte towards the bottom (Figures 4a,b and 5a,b). Significant regional difference bacterial biomass for the bottom and 0-bottom layers were revealed (Table  abundances were comparable in both regions (Table S3).

Zooplankton
Pronounced spatial changes were evident for the total zooplankton abundance and biomass in FS, with the highest values occurring in the south. In the 0-50 m layer, both measures varied over a wide range (Table S3), while rather comparable values were recorded for the 50 m-bottom layer (Table S3). In contrast, the zooplankton abundance and biomass tended to decrease from the north to the south in the 50 m-bottom layer, whereas there was no clear spatial pattern of these variables in the upper 50 m layer (Table S3). Significant regional differences were found in the case of the total zooplankton abundance for the 50 m-bottom layer (Table S3). The total zooplankton biomass in the whole water column was significantly higher in BS (Table S3).
A total of 61 zooplankton taxa were identified during the study, with higher species richness in FS relative to BS (54 vs. 39 taxa) ( Table 2). Despite strong differences in the number of taxa, diversity estimates were similar in both regions: the Shannon-Wiener index was 3.15 ± 0.07 and 2.95 ± 0.08, and Pielou's evenness was 0.58 ± 0.01 and 0.61 ± 0.02 in FS and in BS, respectively. Cluster analysis revealed four groups at 57% similarity ( Figure 6a, Table 2). These groups showed good correspondence to the regions and sampling layers ( Figure 6a, Table 2). Copepods were the only major group in the four clusters constituting >98% of the total zooplankton abundance and >85% of the total biomass. The SIMPER revealed that the dissimilarity between the regions was 33.9% and that the copepods Acartia longiremis, Metridia spp., Calanus spp., Centropages hamatus, Pseudocalanus spp., Microcalanus spp., Oithona spp., Triconia borealis, the chaetognath Parasagitta elegans, the gastropod Limacina helicina, and the appendicularian Oikopleura juv. altogether explained >80% of the total variation in the zooplankton communities. Small copepods contributed mostly to the total zooplankton abundance in the upper 50 m layer, while the large-sized taxa occurred in great numbers below 50 m in both regions (Table 2). However, there were clear differences in the contribution of the main taxa, as revealed by the SIMPER analysis. ANOSIM showed significant differences in the zooplankton community structure among the four clusters (Global R = 0.808, p = 0.001). Pairwise tests revealed significant differences in zooplankton abundances between regions and layers (R = 0.569-0.741, p = 0.008-0.018).   Small copepods contributed mostly to the total zooplankton abundance in the upper 50 m layer, while the large-sized taxa occurred in great numbers below 50 m in both regions (Table 2). However, there were clear differences in the contribution of the main taxa, as revealed by the SIMPER analysis. ANOSIM showed significant differences in the zooplankton community structure among the four clusters (Global R = 0.808, p = 0.001). Pairwise tests revealed significant differences in zooplankton abundances between regions and layers (R = 0.569-0.741, p = 0.008-0.018).

Spatial Separation of the Plankton
PCA indicated that stations sampled in FS and in BS were significantly different based on biotic variables ( Figure S2, and Tables S4 and S5). A two-way PERMANOVA based on the biotic variables (bacterial abundance and biomass, Chl-a concentration, and total zooplankton abundance and biomass) indicated significant differences between FS and BS, as well as between sampling layers (F = 5.14-30.47, p = 0.0001-0.0004). The post hoc PERMANOVA revealed that most of the pairwise comparisons of the surface and bottom sampling layers were significantly different (p = 0.0001-0.0320), whereas the comparisons of the intermediate and upper sampling layers as well as the layers below 200 m were not (p = 0.072-0.758). Table S6 indicates the results obtained by the generalized linear models (GLZ) from environmental data and biotic variables combined for both regions. Temperature, salinity, and Chl-a content were the factors most influencing spatial variations in the microbial parameters (Table S6). Phytoplankton biomass was strongly associated with hydrological variables, nutrient concentrations, and sampling depth (Table S6). There were some differences between regions in factors explaining variations in microbial parameters (Table S6). In particular, temperature and depth were the most important variables driving spatial changes in the bacterial abundance and biomass, as well as viral abundance in FS, while salinity and phytoplankton biomass explained a significant portion of spatial patterns in microbial variables in BS (Table S6). Nutrients were responsible for changes in Chl-a content in FS, while dissolved oxygen was positively correlated to phytoplankton biomass in BS (Table S6) The RDA extracted two significant ordination axes, explaining together 77.2% of the total variability of the zooplankton abundance (Figure 6b). The first axis was strongly correlated with hydrological parameters and depth, while the second axis demonstrated a close relation to the nutrient concentrations (Figure 6b).

Environmental Influences on Plankton and Biotic Interactions
According to the Monte Carlo permutations test, the variables that had a significant impact on the zooplankton structure were depth and temperature (Table S7). The large copepod species (Calanus spp., Metridia longa) tended to be positively associated with salinity and depth (Figure 6b). Small copepods (Oithona spp. and Pseudocalanus spp.) were positively correlated with Chl-a content (Figure 6b). Acartia longiremis and Metridia lucens demonstrated increasing abundances with temperature ( Figure 6b).
Our GLZ models used to assess the biological interactions among plankton groups revealed that bacterial abundance was closely related to Chl-a biomass, and was negatively associated with ABV in both regions (Table S8). The bacterial biomass was positively correlated with ABV. Viral abundance tended to increase with increasing VBR (Table S8). ABV was negatively related to bacterial abundance and Chl-a content, and positively to bacterial biomass (Table S8). Chl-a biomass was associated with most of the biotic variables (Table S8). Fluctuations in zooplankton abundance were best explained via Chl-a biomass (Table S8). In contrast, the total zooplankton biomass was negatively correlated with Chl-a concentration. In addition, it was associated with bacterial abundance and biomass, as well as ABV (Table S8). The responses of the plankton groups differed between regions, with a greater role of Chl-a as a structuring factor of microbial communities and zooplankton in BS (Table S8). In FS, bacterial abundance and biomass were positively correlated with viral abundance, whereas in BS, it had no significant impact on the marine bacteria (Table S8).

Environmental Conditions
Marked climatic fluctuations have been observed in the Arctic in recent years [4,17]. Sea ice cover has dramatically reduced since 1981 in the entire BS [1,28]. The highest decline of sea ice occurred in the Northern BS, and in the eastern Svalbard waters [3]. A stronger warming of the seawater was detected in the regions which are influenced by AW [2,5]. BS has become warmer in the last decade [1,6,12]. The intensity of Atlantic flow into BS and FS is reported to be clearly associated with atmospheric variability over the North Atlantic [2, 5,17,88]. Moreover, this trend is proposed to continue into the future [11]. The North Atlantic Oscillation (NAO) is a measure to estimate the atmospheric influence of the climatic conditions in the northern hemisphere [28]. Positive NAO phases are related to the stronger heat transport into BS, increased mean annual water temperature, and the subsequent decline of ice extent [29]. The winter and annual NAO indices demonstrated clear fluctuations during the period of 2000-2021, and the long-term trend was close to zero. The period of anomalously positive events can be noticed in 2014-2020 [5,11]. Moreover, warming processes caused an increase in the total area of open water and led to a prolonged duration of the open water season [8,9,12,17]. Considering a strong impact of water temperature on plankton assemblages, a general warming of the marine environment in the Arctic would have a great influence on the main plankton populations, and we revealed some responses of the plankton to the temperature variations.
Our study covered regions which can be considered as the areas where these warming impacts were the most pronounced. We revealed a similarity in water temperature at both locations. Comparisons of water temperature with previous multi-annual observations [2, 37,88] showed greater winter values in both study regions, suggesting a remarkable warming of the whole water column. Moreover, the annual average temperature of AW in the Kola Section (69 • 30 -77 • 30 N, 33 • 30 E) was typical of warm years, and exceeded the multi-year mean (1981-2010) by 0.4 • C [5]. In contrast, the salinity in BS and adjacent waters has shown minor changes over the past decade [2,5]. Our study indicated a significantly higher salinity in FS relative to BS. Such differences can partly be explained by a greater freshwater impact from the Kola Peninsula. In particular, the nearshore waters could be affected by less saline waters from Kola Bay toward the open sea. The distribution of water masses in FS and in BS showed a good correspondence to recent observations [2, 5,38], with AW occurring in the whole water column.
In the Arctic waters, there are regular seasonal cycles in major nutrients [7][8][9]26,29]. The maximum values of nitrate, phosphate, and silicate concentrations can be detected in the winter season as a result of a reduced photosynthetic activity. A considerable reduction in nutrients occurs in the spring period due to phytoplankton bloom [30,89,90]. During the summer period, phosphate, nitrate, and silicate remain low, and this is associated with seasonal phytoplankton succession in the Arctic marine environment [24,38]. The autumn season is the period when nutrient concentrations tend to increase [26,29,30]. Our study revealed the enhanced contents of nutrients in seawater relative to the summer and autumn observations [29,51,[89][90][91][92]. This pattern can be explained by low phytoplankton biomasses recorded in both regions, and the effective regeneration of the nutrients in the water column. Moreover, our measures for nutrients were higher compared to data recorded in FS during the winter of 2014 when nitrate and phosphate accounted for 7.5-10.0 µM and 0.6-0.8 µM, respectively [38]. It is more likely that interannual variations in the intensity of AW influx in the Arctic may be responsible for such differences. Recent observations suggest a stronger inflow of nutrient-rich AW into BS and adjacent waters [5,9]. Vertical distributions of the nutrients in FS in the winter period of 2021 were rather homogenous, indicating a strong impact of vertical mixing. In contrast, in BS, there were maxima of nutrients in the intermediate layers at the nearshore stations. The possible reasons for the observed pattern are the inflow of less saline waters from Kola Bay and adjacent coastal sites in the upper layers, and local circulations in the coastal zone [28,59].

Microbial Parameters
Our study provided novel data regarding the abundance of marine bacteria in two Arctic regions during the period of the polar night. Our estimates in FS (2-4 × 10 5 cells mL −1 ) were lower than the summer measures (6-17 × 10 5 cells mL −1 ) reported previously for the West Spitsbergen Current, but comparable with the abundance of 3 × 10 5 cells mL −1 in the East Greenland Current [92]. Other studies have detected that bacterial abundances in AW of FS in summer and autumn were greater compared to our values [93][94][95][96][97][98]. In the Southern BS, summer bacterial abundances fluctuate over a wide range, but they usually exceed 10 × 10 5 cells mL −1 [99]. Our data suggest the winter bacterial abundance and biomass to be lower than in the most productive seasons (spring and summer). However, these (2-4 × 10 5 cells mL −1 ) were comparable with earlier winter data recorded in the Murmansk coastal waters (1-7 × 10 5 cells mL −1 ) [26,100]. A comparison of our data with winter estimates from the central Arctic Ocean revealed higher winter abundances and biomasses in FS and in BS (2.5-6.8 × 10 5 cells mL −1 vs. 2.5-6.8 × 10 5 cells mL −1 and 2.8-7.9 mg C m −3 vs. 2.6-5.9 mg C m −3 ) [101]. Regional variability in the total bacterial number and biomass is a well-documented phenomenon for the Arctic and other marine environments, and it is thought to be associated with differences in resource availability or predation [33,65,96,102]. In general, we established that the distribution of bacterial abundance and biomass was rather homogenous in the whole water column, with a slight tendency to increase in the upper 50 m layer in both regions. The nearshore waters in BS also demonstrated enhanced bacterial abundance in the intermediate layers. Our study documented a clear regional variability in the ABV, with larger bacterial cells in BS. This discrepancy suggests spatial differences in microbial communities in the two Arctic locations. Recent studies have reported a clear seasonal succession of the microbial composition in FS [65,[96][97][98]. It should be noted that the plankton samples in FS and in BS were collected in late November and mid-December, respectively. Therefore, differences in microbial communities between the regions are expected and are in line with the mentioned observations.
The analysis of virioplankton in FS and in BS provided new insights into the current knowledge of the microbial plankton during the period of the polar night. Previous data regarding viral abundance in the Arctic are scarce. In the coastal waters of BS, the winter abundance of virus-like particles was estimated to be 2.5 × 10 6 viruses mL −1 [100], which is comparable with our estimates. At the same time, the concentrations of viruses in the summer and autumn seasons have been reported to be higher, relative to our data [60,99,103]. Stocks of marine viruses were located in the upper 50 m layer in FS, whereas they concentrated in the intermediate 100 m layer in BS. The most prominent difference was the significantly higher viral abundance in FS, where it was 1.4-3.5 times greater. Our study revealed a higher VBR in FS, most likely suggesting lower burst sizes and rates of virally mediated mortality in BS [34,36]. Another possible reason is the dominance of small bacterial cells in FS, providing a greater relative surface for hosting [34,36].

Phytoplankton
Phytoplankton abundance and productivity in BS and other Arctic sites fluctuate considerably throughout the year, with peaks during the spring-summer season [7][8][9]13,30]. Day duration, light regime, nutrient availability, oceanic currents, hydrological factors, grazing impact, and sea ice cover are recognized as the main factors controlling the growth and development of Arctic phytoplankton [7,9,24,29]. The winter season is a less productive period at high latitudes [61,62].
Many previous works have reported extremely low phytoplankton abundance and biomass during the period of the polar night in the BS, FS, and Svalbard waters [11,[61][62][63]100,104]. In the present study, we used Chl-a content as a proxy to estimate the total phytoplankton biomass. Our research detected negligible Chl-a concentrations (0.01-0. 25  Recent studies conducted in more productive seasons (spring and summer) have found generally higher Chl-a biomasses in the Arctic seas [38,46,51]. Our study revealed a stable vertical pattern in the phytoplankton biomass in the water column, with slightly higher concentrations in the upper layers. This was due to the strong vertical mixing that led to homogenous distributions of Chl-a.

Zooplankton
Seasonality in the zooplankton composition in Arctic ecosystems is a well-documented phenomenon connected with environmental fluctuations, food supply, predator pressure, advection, and local circulation [25]. Low zooplankton abundance and biomass have been documented in previous studies during the winter season in the Arctic marine environments [23,45,62,63,104]. Comparisons with other winter studies revealed higher biomass in FS in 2021 relative to the data collected in Svalbard fjords in January 2012 [104]. Our estimates for BS were comparable with the data collected in the same area in November 2012 [23].
Our investigation also indicated low estimates of the total zooplankton abundance and biomass in both regions. The winter zooplankton abundances were 2-15 times and 3-8 times lower than the spring-summer values previously reported in FS and in BS [15,44,49,53,55,105,106]. Despite the dominance of several common taxa, species richness and the diversity of zooplankton assemblages during the period of the polar night may be high and comparable with the estimates in other seasons [21,22,27,37,47,49,50,53,56,57,105]. We also found relatively high species richness, as well as the Shannon-Wiener diversity and the Pielou evenness in both regions. However, the total number of taxa was greater in the FS region. Some possible reasons could explain the higher species richness and zooplankton diversity in the Greenland Sea. First, the plankton fauna of FS included some deepwater species inhabiting waters below 200 m (Aetideopsis armatus, Chiridius obtusifrons, Gaetanus tenuispinus, Gaidius brevispinus, Heterorhabdus norvegicus, and Scolecithricella minor). Second, relative contributions of major taxa were higher in BS, leading to lower H'.
Copepods were found to be the most important members of the zooplankton during the polar night. In our study, they constituted an essential part of the total zooplankton abundance and biomass, showing a good agreement with earlier reports [23,45,62,63,104]. However, there were contrasting patterns in the vertical distributions between taxa with small or large body sizes. The upper 50 m layer was numerically dominated by small-and medium-sized taxa (Oithona spp., Microcalanus spp., Microsetella norvegica, Pseudocalanus spp.) while larger species (Calanus spp., Paraeuchaeta spp., and Metridia longa) occurred below 50 m. Differences in the depth distributions can be explained by taking into account the life strategies of the mentioned copepod species. The herbivorous copepods of the genus Calanus spp. overwinter at greater depths in diapause during the winter seasons [61,62]. They use lipids to survive severe environmental conditions (mainly the absence of phytoplankton). Omnivorous/detritivores taxa (Metridia longa, Oithona similis, Pseudocalanus spp., and Microcalanus spp.) have been found to be active throughout the year [23,45,48,59,63,64,104,107]. Smaller taxa prefer the surface layer, and Metridia longa usually occurs in the intermediate and deep water layers [62,64,104,107]. The larger Paraeuchaeta spp. are carnivorous copepods which inhabit great depth [37,61,62,108].
Our study detected clear regional differences in the total zooplankton biomass, with greater values in BS. The zooplankton abundance was significantly different only in the 50 m-bottom layer, with lower estimations in FS. It is more likely that bathymetry of the regions could play a role in determining such discrepancies. Deep water regions usually had lower abundances and biomasses compared to the shallower sites, as in the case of the coastal waters in BS.
Recent studies have emphasized the Atlantification of the Arctic regions as the process altering the structures of pelagic communities [1,6]. A poleward shift in the range of marine taxa has been proposed, and some boreal taxa could be found in more northern regions. Our study revealed a rather high abundance of the copepod Metridia lucens in FS and in BS, suggesting a more intensive advective impact of AW during recent years. Proportions of other Atlantic taxa were also slightly higher than previously reported [23,45,104]. These findings confirm general observations regarding the borealization process in Arctic pelagic ecosystems.

Environmental Control of Plankton and Biotic Interactions
Our study revealed that hydrological factors and sampling depth have a measurable impact on the plankton groups. However, the importance of each factor was different at a regional scale. The bacterial abundance and biomass were found to covary with water temperature, salinity, and sampling layer, but this influence was stronger in FS. All these factors had a negative influence on the microbial abundance, and the bacterial number tended to decrease with depth and with increasing temperature and salinity. The effect of depth on bacterial abundance has been reported in FS, where rich bacterial assemblages were found in the upper layers [92].
Temperature has been reported as one of the main drivers determining bacterial growth, production, biomass, and abundance in the Arctic seas [96,102]. In our study, the salinity had a greater effect than that of temperature. This may be connected with the low temperature variations in both regions and wider fluctuations in salinity in BS, the region where the lower bacterial abundance and biomass were recorded. Marine bacteria during the polar night were concentrated mainly in the upper 50 m layer where there were more favorable conditions for their growth and development. An inverse relationship between the abundance of bacterioplankton and depth supports our conclusion. Moreover, colder temperatures might have an indirect impact on the bacterial parameters through the availability of resources for heterotrophic prokaryotes [33,101,102]. Lower temperatures may enhance potential sources for bacterioplankton growth, leading to a higher total bacterial abundance [109,110].
The influence of Chl-a on the bacterial abundance was significant in our study, although a strong positive impact on the bacterial biomass was found only for BS. There are some previous evidences that phytoplankton biomass may explain a substantial part of regional variations in bacterial density and production [60,93,94,101]. In contrast, nutrients had no significant influence on bacterial abundance and biomass. It is more likely that the significance of nutrients is higher in the productive seasons, while in the winter, these had lesser impacts.
The size of bacterial cells was one of the main variables explaining the spatial pattern in the total abundance and biomass, while the number of bacteria was correlated negatively with ABV, and the bacterial biomass increased with ABV. This is an expected result because the small-sized bacteria form a more numerous population compared to larger bacteria. In FS, the bacterial abundance and biomass were positively associated with the density of the virioplankton while there were no direct relationships between these variables in BS. Given the prevalence of small bacterial cells in FS, we may propose higher host availabilities for viruses that could cause a rise in their numbers relative to BS where larger bacteria dominated.
ABV was negatively correlated with salinity and temperature, and the larger bacterial cells were concentrated in the intermediate and bottom layers. Similar results have been noted in the Northwestern and Central BS during the autumn-winter periods [60,100]. Three reasons can be proposed to explain this pattern. First, colder temperatures might be favorable for larger bacterial populations. Second, the upper and bottom layers were different in the compositions of bacterial assemblages, with the dominance of smaller groups occurring in the surface layer. Third, large bacteria might sink from the surface to the bottom layers, forming aggregations near the seafloor. Negative correlations between AVB and Chl-a concentrations give additional evidence regarding the higher occurrence of small-sized bacteria in the upper layers where phytoplankton biomass reached maximum values.
Viral abundance and VBR were significantly related to each other and to the biomass of microalgae being greater in the layers, with relatively high Chl-a concentrations. In FS, depth was the most explanatory variable controlling vertical distributions of the virioplankton counts. The mechanism explaining the link between viral abundance, VBR, and Chl-a is not clear. Some studies report no direct correlations between these parameters, while other works provide significant correlations [35,36,60,103]. We may speculate that the higher Chl-a measures reflected enhanced concentrations of microalgae and cyanobacteria that could be favorable for viruses, leading to a higher viral abundance and VBR in the chlorophyll-rich layers.
We found that phytoplankton biomass reckoned as the Chl-a content tended to be increased with temperature, salinity, dissolved oxygen, and nitrate concentrations. Oceanographic conditions and nutrients are among the factors having a significant influence on the growth and development of marine phytoplankton [34,[89][90][91]. The effects of temperature and salinity on microalgae have been studied in the spring and summer periods [29,46,51]. These factors are thought to indirectly affect the phytoplankton in the Arctic, causing the stabilization of the water column [7][8][9]11,13]. In the productive seasons, the light regime plays an important role. Temperature probably would have a greater role in the period of darkness, determining the rates of growth, respiration, and other physiological processes. However, special studies focused on the biochemistry and physiology of the Arctic phytoplankton are needed to support our statement. We also revealed an inverse relationship between depth and Chl-a, suggesting that the phytoplankton was located in the upper layers. This result is in accordance with other studies dealing with Arctic microalgae [11,26,30], and suggests that the upper layers are more suitable for phytoplankton growth.
Zooplankton abundance and biomass were positively associated with Chl-a, bacterial abundance, and the size of bacteria in our study. In BS, these were also positively correlated with viral abundance. Food availability can be considered as an important factor affecting zooplankton growth and survival during periods when potential food items are scarce, because environmental variables may have a greater significance under food-saturated conditions [24,25,107]. Therefore, the positive relationship with phytoplankton biomass during the polar night is an expected finding, due to the extremely low Chl-a concentrations. Bacterial aggregations can be used as food resources by many small zooplankton taxa [111]. Small copepods were found to be the most numerous group in our study. They concentrated in the upper layers, with high bacterial abundances. It is more likely that bacterial cells may be utilized by these copepods supporting the direct correlation between zooplankton and bacterial densities [60].
The present study showed that the spatial variation in the zooplankton community was controlled by a set of environmental variables, with temperature and depth being the most significant. Common species demonstrated different responses to environmental factors. For instance, small copepods were positively related to phytoplankton biomass and tended to be located at the surface, while the larger species showed an opposite pattern. Coastal species, Acartia longiremis and the Atlantic copepod Metridia lucens, were strongly correlated with temperature, while Calanus finmarchicus was positively related to salinity, indicating a strong association of these taxa with warm AW. Temperature, being an indicator of intensity of AW inflow into the Arctic, is an important factor determining the abundance and biomass of zooplankton taxa throughout the year, as has been revealed in many previous reports [14,15,33,37,[55][56][57][58][59]62,106].
Finally, it should be noted that in the winter periods, other factors could have a significant influence on the plankton. In particular, dissolved organic matter provides an important source for microbial assemblages in any aquatic environment [29,102]. Bacterivorous heterotrophic eukaryotes (may control the abundances and biomass of marine prokaryotes through grazing [102], and this impact may be significant in the polar night. On the other hand, heterotrophic protists can be an important food source for larger zooplankton (copepods, appendicularians, and Limacina spp.) supporting their populations during the winter season [63,104,111].

Conclusions
The Arctic marine environments have been strongly changed over the past few decades due to strong climatic perturbations. Pelagic communities are good indicators of environmental forcing, and they may quickly respond to variations in natural conditions. The polar night is a unique phenomenon in the Arctic and Antarctic. Despite recent advances in marine ecology, many aspects regarding the structure and functioning of plankton assemblages during the period of the polar night remain unexplored. Our study presents novel data on the distribution of important plankton groups in the Fram Strait and in the Barents Sea, two major pathways connecting the Atlantic and Arctic oceans. We initially hypothesized that both regions should be similar in environmental and plankton characteristics. However, despite the similarities in temperature, bacterial abundance, and chlorophyll a, the two localities were significantly different, based on the whole dataset. Vertical patterns of the total bacterial abundance and biomass demonstrated a homogenous distribution in the water column, although the upper layers were colonized with more abundant populations. The mean size of bacterial cells was significantly higher in the Southern Barents Sea, suggesting potential regional differences in the composition of marine bacteria. Viral abundance and the virus-to-bacteria ratios tended to be higher in the Fram Strait. Phytoplankton biomass was found to be extremely low during the study period, but it had a considerable impact on the plankton. Zooplankton assemblages were dominated by copepods, demonstrating contrasting distributions in relation to individual body size, so that small taxa were common inhabitants in the upper layers, while larger taxa occurred mainly below 50 m. Temperature, salinity, and the depth of sampling were the most important drivers determining spatial fluctuations in the plankton variables, while nutrients played a minor role. Biotic parameters were correlated with each other, demonstrating strong links between all components of the plankton, from viruses to copepods. Our study expands the current knowledge regarding the structure and ecology of Arctic marine plankton during the period of the polar night, and it may be used as a baseline for future research dealing with the functioning of pelagic food webs.
Supplementary Materials: The following are available online at: https://www.mdpi.com/article/ 10.3390/biology12030368/s1, Figure S1: Results of principal component analysis (PCA): ordination diagram indicating separations of sampling stations based on hydrological variables (water temperature and salinity) and hydrochemical parameters (dissolved oxygen, nitrate, phosphate, and silicate) in the Barents Sea and the Fram Strait, winter 2021. Table S1: Factor scores extracted with principal component analysis (PCA) based on hydrological variables (water temperature and salinity) and hydrochemical parameters (dissolved oxygen, nitrate, phosphate, and silicate) in the Barents Sea and the Fram Strait, winter 2021. Table S2: Factor loadings extracted with principal component analysis (PCA) based on hydrological variables (water temperature and salinity) and hydrochemical parameters (dissolved oxygen, nitrate, phosphate, and silicate) in the Barents Sea and the Fram Strait, winter 2021. Table S3: Plankton abundance and biomass (range and mean ± SE) in the Fram Strait and in the Southern Barents Sea, winter 2021. Comparisons were performed using one-way ANOVA or Kruskal-Wallis test, significant differences at p < 0.05. n/a-no analysis. Figure S2: Results of principal component analysis (PCA): ordination diagram indicating separations of sampling stations based on biotic variables (bacterial abundance and biomass, viral abundance, chlorophyll a concentration, and zooplankton abundance and biomass) in the Barents Sea and the Fram Strait, winter 2021. Table S4: Factor scores extracted with principal component analysis (PCA) based on biotic variables (bacterial abundance and biomass, viral abundance, chlorophyll a concentration, and zooplankton abundance and biomass) in the Barents Sea and the Fram Strait, winter 2021. Table  S5: Factor loadings extracted with principal component analysis (PCA) based on biotic variables (bacterial abundance and biomass, viral abundance, chlorophyll a concentration, and zooplankton abundance and biomass) in the Barents Sea and the Fram Strait, winter 2021. Table S6: Results of 18 GLZ models with microplankton characteristics and chlorophyll a concentration as the dependent variables, and environmental parameters as independent variables in the Fram Strait and in the Southern Barents Sea, winter 2021. Only significant explaining variables (p < 0.05) are indicated. Table  S7: Ranking of environmental variables influenced zooplankton assemblages in the Fram Strait and in the Southern Barents Sea, winter 2021 (Monte Carlo permutation test, 999 permutations). Significant differences was set at p < 0.05. Table S8