When a Year Is Not Enough: Further Study of the Seasonality of Planktonic Protist Communities Structure in an Ice-Free High Arctic Fjord (Adventfjorden, West Spitsbergen)

: As a contribution to understanding the ecological framework of protistan seasonal succession patterns, we present the weekly-to-monthly (January–October) light microscopy-based study of nano- and microplanktonic protist communities of Adventfjorden waters in 2013. In general, protist dynamics corresponded to the classic paradigm for the Arctic ice-free waters with extremely low abundance and diversity in winter, with the main abundance and chlorophyll- a peak in April-May, followed by a diverse but low abundant community during summer/autumn. However, the reference of the obtained data to the previously conducted year-round research in 2012 allows us to observe substantial variability in seasonal patterns between the two consecutive years. The most striking difference concerned the spring bloom composition and abundance, with clear domination of Phaeocystis pouchetii in Atlantiﬁed fjord waters in 2012 and Bacillariophyceae-dominated (mainly Fragilariopsis , Thalassiosira nordenskioeldii , and, in a lesser extent, also Pseudo-nitzschia seriata ) bloom in 2013 when local water prevailed. On the other hand, a surprisingly high share of spring bloom taxa persisted throughout the summer/autumn of 2013 when they co-occurred with typical summer taxa (dinoﬂagellates and other small ﬂagellates). Their extended growth could, at least in part, result from scarce Ciliophora throughout the season, which, in turn, can be attributed to the high grazing pressure of very numerous meroplankton and mesozooplankton. In light of this, our results may be relevant in discussions proposed for the West Spitsbergen waters link between the Atlantic water inﬂow and the spring bloom composition, as well as its further progression in the productive season. They also highlight the strong need for further high-resolution monitoring of annual plankton cycles and great caution when looking for phenological patterns within a single year or when interpreting short-term data.


Introduction
Protist communities, particularly at high latitudes, show strong seasonal shifts in qualitative and quantitative compositions associated with light conditions, ice cover, water masses, nutrient availability and grazing pressure. Protists are critically important for the functioning of marine ecosystems as primary producers and consumers of carbon and maintainers of biogeochemical cycles [1]. Despite their unquestionable importance, little is known about their seasonal dynamics in the Arctic waters [2]. This is primarily because most studies have been limited mainly to snapshots of the plankton ecosystem during either spring or summer sampling campaigns [3]. Thus, it is clear that more high-resolution

Adventfjorden
Adventfjorden is a small (approximately 8 km in length and 3 km in width), openended southeastern branch of Isfjorden ( Figure 1). Most of the Adventfjorden basin's depth exceeds 50 m, with depths greater than 100 m in the outer areas. In the summer and autumn (June-October), two rivers, Longyearelva and Adventelva (the latter one of the largest rivers of the Svalbard archipelago), supply large amounts of fresh water and sediment load to the fjord, affecting the optical properties of the seawater [25]. Due to climatedriven changes in wind stress patterns over the Fram Strait and increased transport of warm AW into Isfjorden, ice-free conditions throughout winter have tended to prevail over the last decade, with the fjord remaining mostly open since 2005 [22].

Data Collection
Samplings were conducted at weekly-to-monthly intervals at the mouth of Adventfjorden (an 85-m-deep IsA time-series station), close to Longyearbyen (78°16′ N, 15°32′ E; Figure 1, Table 1, sampling dates (n = 27)). During the bloom period (April-June), samples were collected with a higher frequency. The ice-free conditions allowed boatbased sampling throughout the campaign. A vertical conductivity-temperature-depth (CTD) profiler was obtained at each sampling event using an SD204 CTD probe (SAIV A/S, Bergen, Norway). Based on the CTD measurements, we calculated the seawater density (sigma-t) using the formula developed by Mamaev [26]. Light measurements were conducted with a cosine PAR (photosynthetically active radiation) sensor and an LI-1000 DataLogger device (Li-Cor, Inc., Lincoln, NE, USA). Seawater was sampled using a 10-L

Adventfjorden
Adventfjorden is a small (approximately 8 km in length and 3 km in width), openended southeastern branch of Isfjorden ( Figure 1). Most of the Adventfjorden basin's depth exceeds 50 m, with depths greater than 100 m in the outer areas. In the summer and autumn (June-October), two rivers, Longyearelva and Adventelva (the latter one of the largest rivers of the Svalbard archipelago), supply large amounts of fresh water and sediment load to the fjord, affecting the optical properties of the seawater [25]. Due to climate-driven changes in wind stress patterns over the Fram Strait and increased transport of warm AW into Isfjorden, ice-free conditions throughout winter have tended to prevail over the last decade, with the fjord remaining mostly open since 2005 [22].

Data Collection
Samplings were conducted at weekly-to-monthly intervals at the mouth of Adventfjorden (an 85-m-deep IsA time-series station), close to Longyearbyen (78 • 16 N, 15 • 32 E; Figure 1, Table 1, sampling dates (n = 27)). During the bloom period (April-June), samples were collected with a higher frequency. The ice-free conditions allowed boat-based sampling throughout the campaign. A vertical conductivity-temperature-depth (CTD) profiler was obtained at each sampling event using an SD204 CTD probe (SAIV A/S, Bergen, Norway). Based on the CTD measurements, we calculated the seawater density (sigma-t) using the formula developed by Mamaev [26]. Light measurements were conducted with a cosine PAR (photosynthetically active radiation) sensor and an LI-1000 DataLogger device (Li-Cor, Inc., Lincoln, NE, USA). Seawater was sampled using a 10-L Niskin bottle (KC Denmark) at four Water 2021, 13,1990 4 of 18 standard depths (5 m, 15 m, 25 m and 60 m). Samples used for the protist community and chlorophyll-a analyses were collected from 10 January to 9 October, whereas samples for nutrient analyses were collected until mid-June.

Chlorophyll-a
Subsamples (250 mL-400 mL) were analysed in triplicate, as described by Holm-Hansen and Riemann [27]. They were filtered through Whatman GF/F filters (nominal pore size of 0.7 µm), folded, wrapped in aluminium foil and immediately frozen at −80 • C. Pigment extraction was performed in 10 mL of methanol for 20-24 h in the cold (4 • C) dark place no later than nine months after sampling. The chlorophyll-a concentration [mg m −3 ] was measured with a Turner Design AU-fluorometer (calibrated with pure chlorophyll-a, Sigma S6144).

Planktonic Protists
Subsamples of 200 mL were immediately fixed with an acidic Lugol's solution for 24 h and then with borax-buffered formaldehyde. Both fixatives were added to a final concentration of 2%. Qualitative and quantitative analyses were conducted using protocols described by Utermöhl [28] and modified by Edler [29] no later than six months after collection. A subsample of 10-50 mL was placed in a Utermöhl settling chamber for 24 h. Settled protists were counted under an inverted microscope equipped with phase and interference contrasts (Nikon Eclipse TE-300). Microplankton (>20 µm) was enumerated from the entire chamber surface at 100× magnification. Nanoplanktonic protists (3-20 µm) were counted at 400× magnification by moving the field of view along the length of three transverse transects. We counted up to 50 specimens for the most numerous taxa, and the number of fields of view was considered individually. The taxa were identified to the lowest possible taxonomic level following the World Register of Marine Species (WoRMS, http://www.marinespecies.org (accessed on 1 May 2013-30 June 2014)). Small (≤10 µm) indeterminate flagellates were classified as mono-or biflagellates and categorised as Flagellate indet.; microbial eukaryotes of size 0.45 to 10 µm were extensively studied during the IsA campaign and reported separately by Marquard et al. [11].

Data Analysis
All contouring of hydrographical, hydrochemical and biological data for vertical sections was done using SURFER (version 10.0, Golden Software) and Kriging interpolation, which estimates values throughout a specified region based on the available data (marked on the graphs). Water mass classifications were made using Ocean Data View (version 5.3.0) based on categories specific to Isfjorden [5]: surface waters (SW) = Sal < 34, T > 1 • C, intermediate waters (IW) = 34 < Sal < 34.7, T > 1 • C, Atlantic waters (AW) = Sal > 34.9, T > 3 • C), transformed Atlantic water (TAW) = Sal > 34.7, T > 1 • C, Arctic water (ArW) = 34.4 < Sal < 34.8, −1.5 > T < 1 • C, winter cooled water (WCW) = Sal > 34.74, T < −0.5 • C) and local water (LW) = T < 1 • C. The total chlorophyll-a concentrations and abundance (per cubic metre) were averaged for the 5-60 m water column. The relative abundances [%] were presented for the major protist groups (classes where possible), which constituted at least 10% of the total abundance per cubic metre in at least 10% of the samples, as adopted in Kubiszyn et al. [13]. We used R (version 3.4.3, [30]) and the Vegan 2.4-3 library Community Ecology Package for R [31] for all statistical analyses. Non-metric multidimensional scaling (NMDS) ordination analysis based on the taxonomic composition data of the lowest possible taxonomic level expressed as log10transformed abundance per square metre was used to explore the variability in the protist community structure (the time trajectory was added to the plot). Bray-Curtis dissimilarity was used as the distance metric. The associations of the protists to their seasonal changes are presented as a heat map of their contributions to the total abundance per square metre. Only taxa with a contribution higher than 10% in any of the samples were selected. A Spearman rank correlation matrix between each taxon was calculated and converted to values representing a mutual association "distance" as follows: d = 1 − rho (the most correlated species presented the lowest distance, whereas non-correlated species had a distance close to 1). Completelinkage hierarchical clustering was applied to the "distance" matrix, and the results were used to order the species plotted on the heat map. The strength and direction of association between environmental and biological variables were measured using Spearman's rank-order correlation analyses. The correlation matrix was visualised using the R package "corrplot".

Hydrography
The water column (down to a depth of 85 m) properties varied substantially throughout the investigated period ( Figure 2a). From January until late February, pulses of relatively warm (<1.78 • C) waters of Atlantic origin (TAW) were observed in Adventfjorden, alternating with local water throughout the entire winter ( Figure 2b). During March to late April, these waters were substantially colder, saltier and denser (−1.23 to −0.05 • C, 34.35 to 34.87, 27.64 to 28.41, respectively) due to low air temperatures. At this time, WCW developed by convection of very cold and dense water produced by cooling. ArW carried by the inshore coastal current contributed to a relatively homogeneous water column at the IsA station. These waters were slightly warmer in May (−1.04 to 1.90 • C), with the first signals of the presence of SW at the end of the month. During summer, the hydrographical parameters underwent intense and rapid changes. In June, the upper layer of the water column was predominated by warmer and fresher waters of SW and IW, which resulted in thermal stratification above depths of approximately 30 m. From July to September, the water column stratification was much stronger and deeper, with markedly higher temperature (4.25 to 7.28 • C) and lower salinity (<34.00) down to approximately 50 m depth. This layer of variable thickness and temperature was highly influenced by SW, formed from glacial melt and river runoff. Beneath this, mixing between SW and the underlying Atlantic water produced IW. By the end of the summer, the temperature gradually decreased in the upper water column due to surface cooling, and a loss of thermal stratification was observed. At the end of October, the presence of relatively warm and saline TAW (>4.80 • C, >34.70) below 40 m depth was noted in the fjord. developed by convection of very cold and dense water produced by cooling. ArW carried by the inshore coastal current contributed to a relatively homogeneous water column at the IsA station. These waters were slightly warmer in May (−1.04 to 1.90 °C), with the first signals of the presence of SW at the end of the month. During summer, the hydrographical parameters underwent intense and rapid changes. In June, the upper layer of the water column was predominated by warmer and fresher waters of SW and IW, which resulted in thermal stratification above depths of approximately 30 m. From July to September, the water column stratification was much stronger and deeper, with markedly higher temperature (4.25 to 7.28 °C) and lower salinity (<34.00) down to approximately 50 m depth. This layer of variable thickness and temperature was highly influenced by SW, formed from glacial melt and river runoff. Beneath this, mixing between SW and the underlying Atlantic water produced IW. By the end of the summer, the temperature gradually decreased in the upper water column due to surface cooling, and a loss of thermal stratification was observed. At the end of October, the presence of relatively warm and saline TAW (>4.80 °C, >34.70) below 40 m depth was noted in the fjord.

Nutrients
Strong seasonal fluctuations of hydrographic conditions were accompanied by high temporal variability in nutrients availability ( Figure 3). The PO 4 3− and Si(OH) 4  Water 2021, 13,1990

Nutrients
Strong seasonal fluctuations of hydrographic conditions were accompanied by temporal variability in nutrients availability (Figure 3). The PO4 3− and Si(OH)4 conce tions followed the same general pattern observed for NO3 − + NO2 − . The highest nu concentrations were recorded in winter and early spring, with the maxima on 19 Apr

Euphotic Zone Extension
Although the limited number of measurements of euphotic zone thickness prevented us from accurately tracking the temporal changes in the extent of the euphotic zone, the collected data indicated its significant seasonal variability (Figure 4a). The euphotic zone

Euphotic Zone Extension
Although the limited number of measurements of euphotic zone thickness prevented us from accurately tracking the temporal changes in the extent of the euphotic zone, the collected data indicated its significant seasonal variability (Figure 4a). The euphotic zone depths varied irregularly from 6 m to 45 m, with a maximum value on 12 June and a minimum on 5 June. A single measurement in early spring (1 March, 40 m depth) indicated a substantial thickness of the euphotic zone at this time. Concomitant with the spring, its extension decreased, ranging from 15 m on 24 April to 30 m on 15 May. In summer/autumn, the euphotic zone depth at the IsA station was about 20 m on average.

Protist Taxonomic Composition
In 107 samples, 158 planktonic protist taxa were detected, of which 107 were identified to the species level (a list of taxa with their shares [%] in the total abundance on a given day is available online as Supplementary File S1). The identified taxa belonged to 12 higher systematic groups, of which the vast majority (approximately 72%) represented Bacillariophyceae and Dinophyceae, followed by Ciliophora (17%) (Figure 4b). The lowest

Protist Taxonomic Composition
In 107 samples, 158 planktonic protist taxa were detected, of which 107 were identified to the species level (a list of taxa with their shares [%] in the total abundance on a given day is available online as Supplementary File S1). The identified taxa belonged to 12 higher systematic groups, of which the vast majority (approximately 72%) represented Bacillariophyceae and Dinophyceae, followed by Ciliophora (17%) (Figure 4b). The lowest taxonomic diversity was observed in winter and early spring (5-13 taxa per sample, mainly Bacillariophyceae and Dinophyceae). From April, the total number of taxa began to increase, and this trend with the predominant share of the two groups mentioned above continued until early June. The summer/autumn community constituted a mixture of different taxa with a substantial portion of Dinophyceae and Ciliophora. Although the number of taxa fluctuated throughout this period, a pronounced decline on 9 October was likely a symptom of approaching the end of the productive season. Additionally, besides the pool of identified taxa, indeterminate nanoplanktonic flagellates were common in the investigated material. Occasionally (mainly in winter and spring), resting stages of Chaetoceros furcillatus (Bacillariophyceae) and cysts of non-identified Chrysophyceae, Dinophyceae, and Ciliophora were observed.

Protist Abundance and Chlorophyll-a
After extremely low total abundance and low chlorophyll-a concentrations in winter and early spring (<0.05 × 10 8 cells m −3 (primarily Dinophyceae), <0.12 mg m −3 , respectively), we observed a gradual increase in their values in the first two weeks of April (Figure 4c). The spring bloom occurred in the second half of April and lasted until late May. Its onset was manifested by an abrupt increase in the number of cells (almost 10fold between 19 April and 24 April, from 0.12 × 10 8 cells m −3 to 1.10 × 10 8 cells m −3 , primarily due to intensive growth of Bacillariophyceae, Figure 4d), also reflected in high chlorophyll-a content (up to 13.42 mg m −3 ; a proxy for primary producers biomass). High protists abundance was maintained throughout May, even after a sudden drop of pigment concentration in early May (2.89 to 3.95 mg m −3 on 2 May and 10 May, respectively) and subsequent more gradual reduction to 0.89 mg m −3 at the end of the month. Low chlorophyll-a concentrations, comparable to those observed earlier in winter, persisted until the end of the observational period with a minimum value on 9 October (0.22 mg m −3 ). In turn, a significant decrease in the protists number was observed in late May and early June, when the optical properties of the water column deteriorated, possibly due to the river runoff with suspended solids usually observed at this time of the year (Figure 4a; turbidity of the water column was not measured during this study). From mid-June to late July, the total abundance considerably increased to almost the spring bloom values. It was mainly due to high numbers of fixation-sensitive indeterminate Prymnesiophyceae, most likely of genus Chrysochromulina. Since the end of July, the total abundance remained relatively constant with a low level, up to 0.31 × 10 8 cells m −3 . Initially, they were mainly flagellates (Dinophyceae and Cryptophyceae), followed by Bacillariophyceae from late August.

Protist Community Development
The NMDS ordination confirmed high temporal variability of the protist community composition throughout the investigated period ( Figure 5). The Spearman rank test among environmental and biological data revealed that protists occurrence was negatively correlated with salinity and nutrients, suggesting that variations of both parameters were primarily responsible for the temporal changes of communities structure ( Figure 6). The protist communities' composition demonstrated seasonal patterns (Figure 7). The winter community (10 January-19 February) was composed primarily of nanoplanktonic representatives of the Gymnodinium genus (Gymnodinium galeatum, Gymnodinium spp. 5-20 µm, Gymnodinium wulffii). Indeterminate biflagellates of size 3-7 µm were numerous at that time. In early spring (1 March-11 April), the community shifted toward domination of Bacillariophyceae (Thalassiosira spp. 10-30 µm, Navicula pelagica, Navicula transitans var. derasa and Gyrosigma fasciola, Pleurosigma sp.). The spring bloom community comprised ribbon-forming Bacillariophyceae of Fragilariopsis genus (F. cylindrus, F. cf. cylindrus, and F. oceanica), Pseudo-nitzschia seriata, as well as colonyforming prymnesiophyte Phaeocystis pouchetii. Both these groups were strongly correlated with chlorophyll-a concentration ( Figure 6). The summer/autumn community observed from 5 June to 9 October was principally formed by indeterminate Prymnesiophyceae and P. pouchetii, Cryptophyceae (Plagioselmis prolonga and Teleaulax sp.), various Dinophyceae of Gymnodinium genus (species observed in winter), and Bacillariophyceae of taxa constituting the residue of the previous season (Figure 7). The increasing share of Dinobryon balticum genus (species observed in winter), and Bacillariophyceae of taxa constituting the residue of the previous season (Figure 7). The increasing share of Dinobryon balticum (Chrysophyceae), Dactyliosolen fragilissimus and Proboscia alata (both Bacillariophyceae) from 22 August indicated the transition of communities towards the autumn type. A few flagellates (Gymnodinium spp. 5-20 µm, Gymnodinium galeatum, and biflagellates 3-7 µm) constituted an almost year-round component of the protistan plankton.   genus (species observed in winter), and Bacillariophyceae of taxa constituting the residue of the previous season (Figure 7). The increasing share of Dinobryon balticum (Chrysophyceae), Dactyliosolen fragilissimus and Proboscia alata (both Bacillariophyceae) from 22 August indicated the transition of communities towards the autumn type. A few flagellates (Gymnodinium spp. 5-20 µm, Gymnodinium galeatum, and biflagellates 3-7 µm) constituted an almost year-round component of the protistan plankton.

Discussion
High-resolution monitoring of planktonic protist annual cycles in the Arctic are scant, and thus our ability to link protistan phenology to environmental variability is limited [10,33]. Our studies of protists time series in Adventfjorden (a 2012-based study published in Kubiszyn et al. [13]; 2013 is presented herein) helped fill some of the identified knowledge gaps in the protistan plankton dynamics of the high Arctic fjord, as well as showed the fundamental variability in phenological events between the two consecutive years.
The winter community was similar to that in 2012, with low nano-and microplanktonic protist richness and abundance, as well as extremely low chlorophyll-a concentrations (<0.05 × 10 8 cells m −3 , 5-13 taxa per sample, <0.12 mg m −3 , respectively), which can generally be taken as a common plankton feature for all West Spitsbergen fjords during the Polar Night ( [34,35], own observations). Nanoplanktonic flagellates (especially Dinophyceae of the genus Gymnodinium) dominated during this period, which corresponds to previous studies in the waters of West and North Spitsbergen [18,[36][37][38][39][40][41]. As is often the

Discussion
High-resolution monitoring of planktonic protist annual cycles in the Arctic are scant, and thus our ability to link protistan phenology to environmental variability is limited [10,33]. Our studies of protists time series in Adventfjorden (a 2012-based study published in Kubiszyn et al. [13]; 2013 is presented herein) helped fill some of the identified knowledge gaps in the protistan plankton dynamics of the high Arctic fjord, as well as showed the fundamental variability in phenological events between the two consecutive years.
The winter community was similar to that in 2012, with low nano-and microplanktonic protist richness and abundance, as well as extremely low chlorophyll-a concentrations (<0.05 × 10 8 cells m −3 , 5-13 taxa per sample, <0.12 mg m −3 , respectively), which can generally be taken as a common plankton feature for all West Spitsbergen fjords during the Polar Night ( [34,35], own observations). Nanoplanktonic flagellates (especially Dinophyceae of the genus Gymnodinium) dominated during this period, which corresponds to previous studies in the waters of West and North Spitsbergen [18,[36][37][38][39][40][41]. As is often the case with unarmoured cells, due to lack of characteristic morphological features and susceptibility to distortion by fixatives, many Gymnodinium taxa were identified only to the genus level with additional size class distinction [42]. This made it difficult to determine how many species overwinter in the water column. Certainly, due to the practically year-round presence of these dinoflagellates in the studied region and a significant share in the total protistan abundance, and potentially also in primary production, this genus deserves further research with the simultaneous use of molecular and microscopic tech-niques. Additionally, and in line with other studies, our results demonstrated that a very few primary producers (primarily Bacillariophyceae-Thalassiosira, Navicula) were present in Adventfjorden throughout the winter, most likely accidentally introduced into the water column from the bottom sediments as a result of strong vertical mixing (as evidenced by the presence of highly-siliceous benthic diatoms of size 70-100 µm). Although primary production during winter/early spring in the area is estimated to be close to zero [10,43], also reflected in the very low chlorophyll-a values (Figure 4c), phototrophic protists in the water column at this time is by no means a novel finding. Recent reports indicated that diatoms are generally well-adapted to the survival of the Polar Night (at the latitude of this study area (78 • N), the sun stays below the horizon for~4 months, Figure 4a) by sustaining an active photosynthetic apparatus for months in darkness, with taxon-specific survival traits [41,44]. This could allow them to become immediately active upon exposure to light and recover total photosynthetic activity within a short period. Although it is believed that this strategy gives Arctic diatoms a competitive advantage during the early spring, it does not necessarily imply that these survivalists dominate the final stages of the spring bloom [34,44]. Our two-year study clearly demonstrated that the late spring bloom stage was taxonomically different from the Bacillariophyceae population in the winter/early spring, indicating that these cells were likely not responsible for initiating blooms [13]. Additionally, the lack of typical spring bloom species during winter and early spring seems to exclude the possibility that bloom-initiating cells overwinter in the water column. Instead, our results support the overwintering of the spring bloom inoculum in a surface layer of marine sediments, as previously suggested, among others, by Hegseth et al. [45] and Brown et al. [46].
In Adventfjorden, the spring bloom initiated in both years in the second half of April (i.e., at the same time as in ice-free Kongsfjorden, [3]), indicating light availability (as a function of daylength and solar elevation) as the primary trigger initiating the spring blooms. As in 2012, the spring bloom progressed in an unstratified water column, promoting cell resuspension from the sea bottom to the surface. As a result, chlorophyll-a and abundances showed exponential growth from late April to the end of May (a 100-fold increase in pigment content and order of magnitude higher numbers in relation to winter/early spring values). The mismatch between the highest chlorophyll-a concentration (observed at the turn of April and May) and the highest abundance of nano-and microplanktonic protists (lasting throughout May) (Figure 4c), despite the relatively constant taxonomic composition during spring (Figure 4d), could have two reasons for causing this effect alone or combined. First, it could result from the domination of the pico-and smallest nanoplanktonic phytoflagellates in the early bloom, contributing to the total pigment content but not observable in microscopic examination or underestimated in numbers. The corresponding transition of pre-bloom communities toward autotrophic microbial eukaryotes of sizes 0.45 to 10 µm, dominated by the Arctic Micromonas ecotype (strain; 1-3 µm), was previously reported for the IsA station by Marquard et al. [11]. Secondly, a high chlorophyll-a concentration coinciding in time with lower protist numbers may indicate that the observed pigment content, at least in part, originated from macroalgae debris. Since the IsA station represents a highly dynamic and shallow ecosystem, the wind action may lead to the advection of macroalgal residues from ice-covered Isfjorden bays, where shallow water kelp forest zones are heavily affected by the ice scouring in winter and early spring [35]. Additionally, the wind-driven resuspension may cause an exchange of material between the sediment and the water column, including entrainment of benthic algae, which occasionally account for most of the chlorophyll found in the pelagic zone [47,48]. The spring community was dominated by Bacillariophyceae (mainly chain-formed Fragilariopsis, Thalassiosira nordenskioeldii, and, to a lesser extent, also Pseudo-nitzschia seriata) and haptophyte Phaeocystis pouchetii, commonly occurring at this time of the year in West Spitsbergen waters [13,38,49] and other Arctic regions [50]. Noteworthy, the strict dominance of colonial P. pouchetii over Bacillariophyceae during the 2012 bloom was not seen in 2013. The possible explanation for this may be differences in win-tertime and spring AW inflow events to Adventfjorden between the two years. Previous research indicated that shifts in bloom timing and composition characterise years with intensive AW advection (as in the case of 2012), with an increased relative abundance of the P. pouchetii (possibly due to introduction of haptophytes by Atlantic advection; [15,16,51]). Even though Atlantic advection also took place in winter 2013, it was somewhat episodic (January-February) compared to the long-term AW dominance at the IsA station in 2012 (end of February to the end of April) and well before the spring bloom onset. In light of this, our results may be relevant to support the already proposed link between the timing and magnitude of AW inflow and the spring bloom composition. Naturally, in addition to biological advection along with the AW inflows, oceanic water may also provide an extra pool of nutrients to the West Spitsbergen fjords [52,53] that could influence the succession of phytoplankton species during the spring bloom. According to Fragoso et al. [54], the relative presence of diatoms and Phaeocystis in the spring bloom of the Labrador Sea may be influenced by the pre-bloom Si* concentrations (Si* = Si(OH) 4 − NO 3 − ), associated with the input of riverine and glacial meltwater enriched with silica. Since our dataset showed only slightly lower Si* values in early spring 2013 than in the corresponding time in 2012, it is difficult to say whether these differences contributed to the Bacillariophyceae-dominated bloom presented in this study. In line with the expectations that a diatom-poor bloom in 2012 would probably also affect the following year's inoculum since fewer spores would be produced, Bacillariophyceae-dominated spring bloom in 2013 was in modest intensity (up to 1.49 × 10 8 cells m −3 , [3]). The development of spring bloom species was slightly different than in the previous year, starting with a mixture of T. nordenskioeldii and P. pouchetii, followed by Fragilariopsis cylindrus and F. oceanica, and P. seriata. It is worth to mention that the scarce portion of Chaetoceros during the bloom in 2013, to setae of which Phaeocystis usually attaches, should not have a negative effect on the Phaeocystis bloom since it has been proven that chains of Fragilariopsis (and pennate diatoms as general) also may serve as a substrate for the colonies development [55]. These communities exhausted surface nutrient concentrations by the end of May, followed by their almost complete depletion in the water column by June.
After the spring bloom, a change in the species composition of the fjord marked the entrance to the summer season. The nutrient-limited post-bloom period (June-October) was reflected both by the change in the qualitative structure and in the gradual decrease in total protist abundance, which was most likely the effect of nutrient depletion, glacial melting associated with high water column turbidity and stratification, as well as intensive grazing by zooplankton. The major contributors to the total abundance were Gymnodinium spp., Heterocapsa rotundata (both Dinophyceae), Plagioselmis prolonga, Teleaulax sp. (both Cryptophyceae), and other free-living, small cryptophytes and undetermined nanoflagellates, typical for summer West Spitsbergen water [4,14,56,57]. Commonly observed biflagellates (3-7 µm in size), at least some of which are motile morphotypes of P. pouchetii, characterised the termination of a Phaeocystis bloom ([58], Figure 7). Noteworthy, ciliates (mostly Strombidium genus, especially Strombidium sp. 2 (30-40 µm; sensu Bérard-Therriault et al. [59]) and, to a lesser extent, also Laboea strobila, Mesodinium rubrum and Lohmanniella oviformis) were unexpectedly scarce at this time. Their little importance, both in terms of share in the total abundance (Supplementary File S1) and the period of occurrence (mostly at the turn of spring and summer), may have several causes. First and most likely, it can be attributed to the high predation by very numerous meroplankton and mesozooplankton [4]. Furthermore, it could result from high turbidity since filter-feeding ciliates may be deterred by the high loads of fine sediment associated with summer glacier melting. Mineral-organic aggregates can also attach to lorica (shell-like protective outer covering), thus increasing the specific gravity of cells, making it difficult to actively graze and increase the loss of cells from the euphotic zone/or mixed layer depth. The selective, negative effect of high sediment content on ciliates with a simultaneous high share of dinoflagellates during summer (analogous to the research presented herein) has already been demonstrated for Kongsfjorden waters, inter alia, by Kubiszyn et al. [55]. Both the previous and current results, therefore, indicate a high resistance of dinoflagellates to waters siltation, which probably results from a variety of their feeding behaviour and mechanisms, along with the widespread hetero-and mixotrophy (modes of nutrition of species identified in the study are included in the open-access Nordic Microalgae database, http://nordicmicroalgae.org/ (accessed on 29 June 2021)). Another feature that distinguishes this season was the relatively high portion of the spring taxa in the community. The observed temporal resistance of spring taxa (previously reported in Isfjorden by Węsławski et al. [60]), associated with the commonly observed taxa in the West Svalbard fjords during summer (Kubiszyn et al. [13] and references therein), could be caused by several factors, i.e., (1) a strongly stratified water column that prevented cell sedimentation, (2) low microzooplankton pressure on primary producers that generally had more of a direct control on phytoplankton loss rates than mesozooplankton grazing [61], (3) resuspension of cells by currents and waves from a subtidal coastal site [62], or (4) a supply of cells from outside the fjord along with the advection of oceanic waters and/or sea ice.
Interestingly, on a short-term scale (two samplings in June and July with three weeks apart), the summer community was practically 100% dominated by indeterminate Prymnesiophyceae, most likely of genus Chrysochromulina, damaged or severely deformed by fixation. The massive growth of this taxon occurred soon after a significant reduction of the euphotic zone thickness in early June, which suggests that the fjord could experience substantial suspension matter loads with freshwater input from glacial rivers at this time that could bring Chrysochromulina-like species into the fjord. The latter return of the community to the state as it had presented before (in terms of species composition and abundance) additionally speaks for some disturbance factor that rapidly, temporarily rebuilt the structure of the community. Our results highlight salinity as a strong determinant of protist community composition while also demonstrating the importance of nutrient supply, where abundance generally decreased with increasing salinity and increased with nutrient depletion. However, it is certain that other abiotic factors are also important and can interact with or stand behind changes in salinity to influence protists communities (e.g., seawater turbidity; unfortunately, not included in the study). In other words, as Delpech et al. [63] documented for the structure of summer bacterial and archaeal communities of Isfjorden, reorganisation of protists community composition and functioning coinciding with gradients in salinity and nutrients (especially in melting season) could illustrate the significance of seasonal changes in terrestrial inputs.
Unfortunately, due to the shortened research period compared to the year-round 2012 study, we could not observe the entire phenological cycle with a return to the initial state by the end of the year ( Figure 5). Nevertheless, even with less frequent sampling throughout the season, we managed to capture the transition into the autumn community from late August onwards (consisting of Dinobryon balticum, Dactyliosolen fragilissimus, and Proboscia alata), concomitant with the gradual decrease in light availability.

Conclusions
In the presented study, we generated a data set enabling us to accurately trace the protists phenology of Adventfjoren in the seasonal cycle (winter-autumn 2013) and referred the obtained results to analogous studies conducted throughout 2012. Although the general trend followed the classic annual paradigm [3], protists dynamics differed significantly between the years, as was proposed before for the more northerly located Kongsfjorden [15][16][17]. The most spectacular differences concerned the spring bloom which accounts for a substantial part of annual production, supports pelagic and benthic secondary production, and influences biogeochemical cycles, and, consequently, the summer/autumn communities. In contrast to the virtually completely Phaeocystis-dominated spring bloom in 2012, when AW dominated in the fjord, spring bloom in 2013 when local water prevailed was mainly composed of Bacillariophyceae (especially Fragilariopsis and Thalassiosira, and, in a lesser extent, also Pseudo-nitzschia). A high share of them persisted throughout the summer/autumn when they co-occurred with typical summer taxa (Dinophyceae and Cryptophyceae). Although the results do not allow us to prove mechanistic links between environmental variables and protists phenology, they suggest that the observed inter-annual variability in protist communities can be partially attributed to differences in AW inflow to the fjord, thus increasing the reliability of previously proposed scenarios for Kongsfjorden waters. Since the West Spitsbergen fjords' hydrography leads to a high level of unpredictability in recent years [64], with a pronounced trend of increasingly frequent winter and spring infusions of AW [65], we, therefore, concluded that it might be accompanied by increased inter-annual variability in protists phenology. While spring bloom's timing seems to be of light control in the ice-free waters and relatively constant between years (mid-April), these fluctuations will mainly manifest in species composition and abundance. Therefore, we would like to emphasise that each year should be treated individually in light of such high volatility, and any phenological patterns should be determined with great care. Similarly, short-term (usually summer) data should be interpreted in relation to traditional seasonal dynamics with high caution. Because our research clearly demonstrated that one or even two years of high-resolution monitoring is not enough to resolve the temporal variations of protist communities against the changing Arctic conditions, examination of annual cycles over many years is strongly required.