Seasonal dynamics and life histories of three sympatric species of Pseudocalanus in two Svalbard fjords

Abstract Small copepods are the most diverse and numerous group in high-latitude zooplankton, yet our knowledge of important species remains poor because of the difficulties involved in correct species identification. In this study, we use a molecular method of identification, a species-specific polymerase chain reaction, to provide the first description of the seasonal dynamics and life histories of the important genus Pseudocalanus in two Svalbard fjords with contrasting environments. We conducted monthly investigations in the relatively warm and ice-free Adventfjorden, supplemented with seasonal samples from the colder, seasonally ice-covered Billefjorden. We found three species of Pseudocalanus (the Arctic P. acuspes and P. minutus, and the boreal P. moultoni). Pseudocalanus acuspes had a distinct annual life cycle and dominated during summer, when it actively reproduced. Surprisingly, the boreal P. moultoni was present year-round in both fjords and was the dominant species during winter; the presence of all life stages of this species throughout the year suggests a more continuous reproduction. The Arctic P. minutus was the rarest of the three species and was likely able to complete its life cycle in Billefjorden but not in Adventfjorden. Our study demonstrates that closely related species may have different life strategies and environmental preferences, which presumably make high-latitude zooplankton communities more resilient to climate change impacts on genus but not necessarily on species level.


INTRODUCTION
High-latitude pelagic systems are governed by a strong seasonality due to extreme changes in light and thus algal food availability (Leu et al., 2015). Zooplankton grazers living in these regions must tune their life cycles to capitalize on the short, intense primary production period to grow, reproduce, and to build lipid stores that allow them to survive long periods of darkness and low production. The Arctic is rapidly changing, with declining ice cover, rising wave and storm activity, higher sea temperatures and increasing inflow of sub-Arctic water masses into the Arctic (Meredith et al., 2019;Polyakov et al., 2020). Despite the drastic environmental changes, the sun's activity at high latitudes will continue to create extended periods of continuous light and darkness, limiting light-driven production, which will result in a changed environment for both resident and advected zooplankton species. It is the different species' ability to quickly adapt to this changing environment, or more specifically, the species' plasticity and robustness, that will determine the success the individual species and the ecosystem as a whole.
Although year-round observations are still relatively scarce in the Arctic, the life histories and life strategies of zooplankton have been well described for key members of the plankton such as the genus Calanus (e.g. Conover, 1988;Slagstad and Tande, 1990;Kosobokova, 1999;Freese et al., 2016), which dominate most Arctic and sub-Arctic zooplankton communities in terms of biomass. However, numerically zooplankton communities are dominated by small copepods, and seasonal cycles of these groups remain poorly studied. One of the most important groups in high-latitude zooplankton is the genus Pseudocalanus spp., which in many regions dominates numerically and can account for up to 5-25% of the total biomass (Hopcroft and Kosobokova, 2010;Ershova et al., 2017;Carstensen et al., 2019). The genus is composed of seven species (Frost, 1989), which share a very similar morphology, but are characterized by different, partially overlapping geographical distributions, with several members of the genus frequently co-occurring together in the same area. Two species are common in Arctic shelf seas: P. acuspes and P. minutus, but sub-Arctic species can also be transported into the Arctic with Pacific and Atlantic water currents. For example, in the Pacific Arctic, 'warm' years were shown to be characterized by a high presence of the temperate P. mimus and P. newmani, while during 'cold' years, the two Arctic species P. acuspes and P. minutes dominated (Ershova et al., 2017). In Svalbard, the Atlantic P. moultoni, which is remarkably similar morphologically to P. acuspes, was recorded for the first time in 2004 using genetic tools (Aarbakke et al., 2011). The even more temperate P. elongatus was recently observed in the north Norwegian Sea, well above the Arctic Circle (Ershova et al., submitted), and it is not unexpected that this species will also continue to expand northwards.
Svalbard is a unique region in the Arctic because it experiences typically high-Arctic light conditions, with nearly 4-month periods of continuous light and continuous darkness but is strongly influenced by Atlantic currents, which results in most fjords in western Spitsbergen remaining ice-free throughout the entire winter. In recent decades, extensive 'Atlantification' has been documented in Svalbard fjords across multiple trophic levels, with Atlantic taxa replacing their Arctic counterparts (Gluchowska et al., 2016;Vihtakari et al., 2018) The genus Pseudocalanus, which contains both Arctic and Atlantic members, can serve as a useful 'gauge' of the extent of Atlantification in Arctic systems. However, very few plankton studies identify this group at the species level due to very subtle morphological differences between species, which at the juvenile stages are practically indistinguishable. In recent years, the development of genetic tools has significantly facilitated organism identification, including for the genus Pseudocalanus (Grabbert et al., 2010;Aarbakke et al., 2014;Bucklin et al., 2015;Ershova et al., 2017;Ershova, 2020); however, no study thus far has examined the dynamics of these species over a seasonal cycle. In this study, we employ a species-specific polymerase chain reaction (ssPCR) protocol to routinely discriminate between the three species observed in two Svalbard fjords with contrasting oceanographic conditions, and to describe their population dynamics and life cycles over a seasonal cycle for the first time in an Arctic system.

Study area
The study was conducted in Isfjorden, the largest fjord system on the west coast of Spitsbergen (Fig. 1). Adventfjorden (78.3 • N, 15.5 • E) is a small side fjord (3.5 km wide and 8 km long) of Isfjorden. This fjord is 60-120 m deep and has no sill. The fjord is mainly influenced by Atlantic water from the West Spitsbergen Current and remains ice-free throughout the winter. Sampling was conducted at the time series station Isfjorden-Adventfjorden (stn. IsA), situated at the fjord mouth at 78.26 • N, 15.53 • E, from March 2018 to February 2019. The second study site was located in Billefjorden (78 • 40 N, 16 • 40 E), another fjord arm in the Isfjorden system where a shallow sill prevent water exchange with the rest of Isfjorden and thus severely restricts advection of warm Atlantic water. Billefjorden is therefore primarily shaped by local processes resembling 'true' Arctic environmental conditions, with a pronounced seasonal ice cover from December to June, and biological communities dominated by Arctic species (Arnkvaern et al., 2005). Sampling in this location was conducted at the time series station in Billefjorden, Adolf Bukta (stn. BAB) in May 2019 during a period of ice cover, near the ice edge, and in open water in July, August and November 2019.

Hydrography
Sampling was conducted from larger research vessels (RV Helmer Hanssen and RV Dalnie Zelentsy) and small open boats (Supplementary Table I). The in situ conductivity, temperature, density and fluorescence were measured at both stations and all sampling dates (with the exception of fluorescence in IsA during June 2018). A handheld SD208 CTD (SAIV, Bergen) was used at all IsA stations and dates, except at IsA May 11 and August 18 when a SBE911+ CTD-Rosette water sampler system (Sea-Bird Electronics Inc.) was used onboard R/V Helmer Hanssen. The handheld CTD, programmed to make a measurement every second, was lowered to around 10 m above the sea floor with a speed of maximum 0.5 m s −1 , after leaving it 1 min just below the surface for acclimatization, and was raised with the same speed.

Zooplankton collection
Mesozooplankton was collected using a WP2 net (Hydro-Bios, Kiel) with a mouth opening of 0.25 m 2 . A mesh size of 60 μm was used for all months, except May 2018 in Adventfjorden during the spring bloom when a 200-μm net was used to avoid clogging of the finer mesh net. All Billefjorden samples were collected using a 200-μm net as part of the regular IMOS (Isfjorden Marine Observatory Svalbard) and Økokyst Svalbard sampling. The net was lowered to approximately 10 m above the seafloor and towed vertically with a speed 0.5 m s −1 . Minimum two hauls were taken at all stations; one was fixed with 4% buffered formalin and the other with 80-96% ethanol.

Zooplankton processing
Pseudocalanus spp. was counted quantitatively from either the formalin or ethanol samples (Supplementary Table  II). Samples were washed from the fixative and diluted with filtered seawater to a fixed volume. Five milliliter subsamples were taken using a 5-mL Finntip pipette until a minimum of 100 Pseudocalanus spp. individuals were counted under a Leica Stereomicroscope at 25-40× magnification. Counted individuals were separated by copepodite stage [C1-C5, adult females (AF), adult males (AM)]. Nauplii were also identified and counted, although nauplii counts should not be considered quantitative in the 200-μm mesh nets (e.g. from Billefjorden 2019 and station IsA in May 2018).

Molecular identification
Between 30 and 140 individuals (typically 80-100) from each sample were identified genetically. In most samples (see Supplementary Table I for list), prosome length of each processed individual was measured using the ZoopBiom digitizing system (Supplementary Table III) (Roff and Hopcroft, 1986). All prosome length measurements used in this study were thus obtained from ethanolpreserved samples, so if there was shrinkage due to preservation, the bias was consistent between samples. Each specimen was soaked in Milli-Q water for 30 min to remove traces of ethanol, after which DNA was extracted from each specimen using the HotSHOT method (Truett et al., 2000). This extraction method is rapid and inexpensive, requiring only 30 min of incubation time and no DNA cleaning steps. Identification was carried out using a ssPCR designed to separate three species that have been described from this area previously: P. acuspes, P. minutus and P. moultoni (Ershova, 2020). The species-specific primers were designed to attach to locations on the COI gene that were conserved within a species but variable between species, in order to amplify fragments with a size difference of 60-150 base pairs (bp). The primers were selected to have a similar melting temperature (within 3-4 • C) and were evaluated for primer dimer formation using the online tool Multiple Primer Analyzer (Thermo Fisher Scientific). Two forward primers (one for P. moultoni and one for P. acuspes and P. minutus) and three reverse primers (one for each species) were selected (Table I). A 10 μL polymerase chain reaction (PCR) was carried out, containing 5 μL of ToughMix polymerase master mix, 0.5 μL of each of the five species-specific primers (two forward and three reverse, Table I), 0.2 μL green dye, 1.8 μL sterile water and 0.5 μL extracted DNA. The PCR protocol was 5 min at 95 • C; 35 cycles of 40 s 94 • C, 40 s 62 • C, 50 s 72 • C, 7 min at 72 • C. The resulting amplicons were placed on a 2% agarose gel together with a 50 bp ladder; identification was carried out visually based on the length of each fragment (Fig. 2). A minimum of 10 and a maximum of 30 individuals of each stage were identified in this way (when present in the sample in sufficient numbers), with a total of 1366 individuals for all samples combined. Although the possibility remains that P. elongatus was also present in the samples, a simultaneously conducted metabarcoding study did not detect P. elongatus or other Pseudocalanus species than the three mentioned above in the Svalbard fjords (Ershova et al., submitted). For this reason, and due to the fact that a large portion of the analysis for this study has been carried out prior to the development of species-specific primers for P. elongatus, we did not include this species in the identification protocol.

Data processing
CTD measurements were quality controlled and interpolated at 1-m intervals. Salinity at each depth was calculated from conductivity using the package oce in R (Kelley, 2018). Water mass types were assigned at 10-m intervals following temperature/salinity thresholds defined in Svendsen et al. (2002) and Nilsen et al. (2008). The relative species proportions at each developmental stage, as determined by ssPCR, were multiplied by the quantitative counts of copepodite stages to estimate abundance and population structure of each species per sample. Biomass in dry weight (micrograms) was estimated from prosome length (micrometers) using a regression relationship described for this genus (log 10 DW = −7.62 + 2.85 * log 10 PL) (Liu and Hopcroft, 2008). Biomass at each station was estimated by multiplying mean weight of that species/stage with its abundance. As zooplankton experiences some shrinking in ethanol, these values may be slightly underestimated. For those samples where no measurements were taken, the average weight of each species/stage from all the samples was used to estimate biomass. Mean developmental stage of each species at each sampling date was calculated by multiplying the proportion (0-1) of each stage by 0-6 (0, nauplii; 1-5, C1-C5 stages; 6, AF/AM) and summing up the values for each stage together. For example, a population that consisted of 20% C1 stage, 70% C2 stage and 10% AF would have a mean stage of 2.2 (0.2 * 1 + 0.7 * 2 + 0.1 * 6).
The correlations between species distribution and population structure of the three Pseudocalanus species and the physical environment were investigated using canonical correspondence analysis (CCA). The CCAs were performed on square-root transformed abundance data and scaled physical parameters using the R package vegan (Oksanen et al., 2019). Each species was divided into four developmental stage groups: nauplii-C2, C3-C4, C5 and AF/AM. Examined physical variables included surface, bottom and integrated temperature and salinity, as well as maximum and depth-integrated fluorescence. The best model was selected via the ordistep function in the package vegan in both directions with 10 000 permutations (Blanchet et al., 2008). The significance of the resulting model and predictors were tested, with significance level set to P ≤ 0.05. Differences in body size of individuals of the same species/stage between months were tested using a one-way analysis of variance (ANOVA).

Physical environment
The water masses in Adventfjorden (Stn. IsA) comprised a mixture of local water (LW), locally formed winter cold water (WCW) and Arctic water (ArW) (as defined in Svendsen et al., 2002;Nilsen et al., 2008) from March to May, but never reached temperatures lower than −0.5 • C (Fig. 3). In June, a sharp shift in water mass properties was seen, with warmer (1-2 • C) and more saline transformed Atlantic water (TAW) appearing in the entire water column. Snow melt and increased river run off combined with solar heating led to a distinct upper fresher and warmer (up to 7 • C) surface water (SW) layer from July to September, with a mixture of SW and TAW below defined as intermediate water (IW). In October, the IW filled the entire water column. In November, TAW appeared below 30 m and filled nearly the entire water column by January 2019. By February, cooler LW dominated.
In Billefjorden, sea ice was still present in May 2019, and WCW dominated with low temperatures between −1.6 and −1.8 • C in the entire water column. In July and August, SW with relatively warm temperatures (>6 • C) and relatively low salinity (<32) was found in the upper 15-30 m, while below 50 m (and the threshold depth), WCW with temperatures between −1.8 and −1 • C still prevailed. In the period between September and November, cooling of the SW and winter convection led to gradually colder SW, while a warmer layer between 35 and 70 m depth remained present.
In Adventfjorden, the spring bloom began in May, with high chlorophyll found throughout the entire water column. Elevated chlorophyll values continued in the surface layers until July, when the Chl-a maximum occupied the upper 20 m. A second, smaller bloom appeared in October. In Billefjorden, an intense bloom was observed in May; measurable fluorescence was maintained through July, but in August and in November, estimated chlorophyll values approached zero (Fig. 3).

Abundance and biomass
Overall Pseudocalanus abundance ranged from 20 to 570 ind m −3 , with the highest value observed in Adventfjorden in September and the lowest values in Billefjorden in April and Adventfjorden in May. Although Pseudocalanus abundance in Adventfjorden tended to be low during winter/spring from December to June, and generally higher during the summer/fall months (July to November), there was significant month-to-month variability. For example, March was characterized by relatively high Pseudocalanus abundance (190 ind m −3 ), comparable to summer values, while abundance in October was much lower than that observed during both September and November. The overall abundance of Pseudocalanus in Billefjorden was very low in May (20 ind m −3 ), but by July, increased to 230 ind m −3 , and remained high during August and November. Overall, abundance was higher in Billefjorden than in Adventfjorden during all months except May. The overall biomass ranged between 0.1 and 1.6 mg DW m −3 and, contrary to abundance trends, was highest in Billefjorden during August and November (Fig. 3B). Overall, biomass values were less variable month-to-month than abundance with less distinct seasonal peaks and no pronounced decline in biomass observed during winter.
All three Pseudocalanus species co-occurred in both fjord locations (Stns IsA and BaB) at most sampling dates, but the relative species composition varied between the two locations and between seasons (Fig. 4A). Pseudocalanus acuspes was the dominant species in Adventfjorden from May to October, reaching up to 400 ind m −3 in September. However, abundance of this species dropped dramatically during the winter months, reaching a minimum in April (10-50 ind m −3 ). Pseudocalanus moultoni in Adventfjorden had less pronounced seasonal dynamics in abundance, ranging from 10 to 150 ind m −3 , with higher monthto-month variability and peak numbers observed during March, September and November. During the winter months (November to April), the relative contribution of P. moultoni was particularly high, comprising 45-65% of the total Pseudocalanus abundance. Pseudocalanus minutus was only present in Adventfjorden in measurable quantities during the summer months, from June to September, but the abundance was low (<50 ind m −3 ). During the other months, this species was detected as single individuals and represented less than 1% of total Pseudocalanus numbers. In Billefjorden, P. acuspes was the dominant species during all sampling months. Pseudocalanus moultoni was present in much lower numbers than in Adventfjorden, with the lowest values (<3 ind m −3 ) observed in May. In July to November, however, it was present in significant quantities (50-75 ind m −3 ), comprising 20-25% of the total Pseudocalanus abundance. Similar to Adventfjorden, P. minutus was the least abundant species in Billefjorden, but unlike the former, it was observed in Billefjorden during all sampling months (∼20 ind m −3 ). Biomass of the three species generally followed slightly different trends from those of abundance, accounting for the larger body sizes of older developmental stages. As such, P. acuspes had biomass peaks in May and September, while P. moultoni peaked in March and November. In terms of biomass, the larger-bodied P. minutus played a relatively larger role at both locations and dominated the biomass in Adventfjorden in June and August.

Population structure
For P. acuspes, the distribution of developmental stages displayed a strong seasonality in Adventfjorden, with nauplii and C1-C2 stages appearing from June to November and being absent during the other months (note that the May sample was collected with a 200-μm net, so nauplii may have appeared in the plankton earlier) (Figs 4C and 5A). A distinct shift from a predominance of late life stages occurred between May and June (going down from mean stage 5.5 to 1.5) and the population comprised primarily of early life stages C1-C3 until November. During the winter months (December to April), almost the entire population was composed of C3-C4 stages (mean stage 3.8). Stage C5 peaked in abundance in May, and during the remaining months of the year, C5 copepodites were present in very low numbers (<3%). AF showed a strong abundance peak in May and re-appeared in the plankton again in August to September. During the other months, females of P. acuspes were practically absent. AM were observed in May and July. In Billefjorden, P. acuspes had an even more pronounced seasonal cycle than in Adventfjorden, with nauplii appearing in May and absent during the other months. C1 stages appeared in July and disappeared by August, with a clear succession of the population to the overwintering stages C3-C4. In contrast with Adventfjorden, however, AF and AM were present during all months, except November when no males were observed.
For P. moultoni, a shift from older stages to younger stages was also apparent in May in Adventfjorden, but the drop was less pronounced (from 4.9 to 2.2) than in P. acuspes (Figs 4C and 5A). Although nauplii were detected only during the summer months, just as P. acuspes, young copepodite stages of P. moultoni (C1-C2) were also present in low numbers (<10 ind m −3 ) during several of the winter months. C5 copepodites and AF of this species were relatively numerous (>10 ind m −3 ) year-round, and the overwintering population was composed of stages C3 through adult. In Billefjorden, no nauplii of P. moultoni were observed, but it is likely that they merely failed to be captured with the 200-μm net, since the overall abundance of this species was low and life stages C2-C5 were recorded during all months. In Billefjorden, the presence of P. moultoni AF was extremely low in May and not detected in July, but their numbers increased substantially in August and November (Fig. 5B).
The presence of P. minutus in Adventfjorden was only confirmed between June and August, and they comprised almost entirely of later developmental stages with only a few nauplii and C1's recorded (<0.5 ind m −3 ). In Billefjorden, nauplii of P. minutus appeared in May and C1 stage in July.

Body length of adult females
The body size of adult females (AF) of P. acuspes varied among months, with individuals observed in May in both locations being distinctly larger (by about 200 μm) than those observed during late summer-fall (ANOVA, P < 0.01) (Fig. 6). The late-summer females of P. acuspes were the same size or even smaller than the typically smaller-bodied P. moultoni. For P. moultoni, the females were on average slightly smaller in late summer to autumn than in spring, but the differences were much less pronounced and not statistically significant.

Relationship between population structure and physical parameters
The best CCA model included mean temperature (P = 0.01) and maximum fluorescence (P = 0.03) as significant predictors for species and stage composition (Fig. 7). The model accounted for 54% of total inertia in the data, with the first two axes accounting for 83% of the explained variance. The ordination showed a clear seasonal pattern, with a distinct Adventfjorden winter group (November to April), dominated by P. acuspes C3-C4 and P. moultoni C3-C5 stages. Interestingly, August and November in Billefjorden were characterized by very similar species composition and hydrology and grouped with the Adventfjorden winter group. A distinct summer group included both fjord stations between July and October and was associated with warmer water and was dominated by early life stages of P. acuspes and P. moultoni and sub-adults of P. minutus. The third group, which was the most distant from the rest, included May samples from both fjords, which were distinguished by high fluorescence

Pseudocalanus distribution in Svalbard fjords
Although most zooplankton studies carried out in Svalbard recognize Pseudocalanus as an important component of the zooplankton community (Weslawski et al., 1991;Walkusz et al., 2003Walkusz et al., , 2009Daase and Eiane, 2007;Weydmann et al., 2013;Gluchowska et al., 2016;Hop et al., 2019a, b), these studies generally list the genus as a single group without attempting to differentiate between species. The few studies that describe Pseudocalanus at the species level have only focused on AF (Aarbakke et al., 2011(Aarbakke et al., , 2017, or merely made the assumption that most the individuals belonged to one species (Lischka and Hagen, 2005). Nonetheless, some results of these earlier studies describe broad patterns that are comparable to the present work. The highest abundances of this genus have been typically observed in the more 'Arctic-type' fjords (Cottier et al., 2010) in western (Hornsund, Weslawski et al., 1991), northern (Rijpfjorden, Weydmann et al., 2013) and eastern Svalbard (Storfjorden, Werner, 2005;Hirche and Kosobokova, 2011). This is also the case in our study, with the higher abundance of Pseudocalanus observed in the Arctic-type Billefjorden than in the warmer Atlantic-influenced Adventfjorden during all months except May. This becomes even more apparent when looking at the absolute values (per unit area), which were nearly three times higher in the deeper Billefjorden (190 m) than in the shallow Adventfjorden (70-90 m). Adventfjorden is ice free with water temperatures above freezing year-round during most years, resulting in more boreal conditions for the fauna living there. Advection of Atlantic water into the west coast Svalbard fjords occurs primarily during the summer months, and during winter, they generally contain locally formed winter water (Cottier et al., 2010), but wind-driven advective events can also occur in winter, causing upwelling of warm Atlantic water onto the West Spitzbergen shelves and into the fjords (Skogseth et al., 2020). The relatively warm temperatures that we observed in Adventfjorden during March to April 2018 (>0 • C) together with high month-to-month variability in Pseudocalanus abundance-e.g. the anomalously high abundance of Pseudocalanus observed during March-may have been due to such advective event(s). Billefjorden is less influenced by advection, resulting in more stable environmental conditions and less month-tomonth variability in the zooplankton communities. The very low abundances of Pseudocalanus that we observed in May, while the fjord was still ice-covered, likely reflects an annual 'reset' of the populations, marking the onset of a new generation (see section on life cycles below).
Previous studies that used genetic methods for identification have found P. minutus to be the dominant species in a number of Svalbard fjords, including Kongsfjorden (Lischka and Hagen, 2005) and Billefjorden (Aarbakke et al., 2011(Aarbakke et al., , 2017. However, these studies generally identified only AF, and our work strongly highlights the importance of taking into account the entire population. Fox example, if we had included only AF in our analysis, we would have come to the erroneous conclusion that P. moultoni was the only species present in Adventfjorden during the majority of the year. When the entire population was accounted for, P. acuspes was found to be the dominant Pseudocalanus species in both fjords, and P. minutus the least abundant, which is a somewhat surprising result, given previous records. The fact that earlier works focused only on AF explains, however, only part of the discrepancy, as the highest proportion of AF in our study was maximum 25% for P. minutus, in Billefjorden during May. In Adventfjorden and during other months in Billefjorden, its contribution was even lower, with generally <5-10% females belonging to P. minutus. Elsewhere, P. minutus dominates in seasonally ice-covered Arctic shelf seas (Melnikov et al., 2005;Persson et al., 2012; (Werner, 2005;Weydmann et al., 2013), but it is also numerous in the deep waters of the North Atlantic to the east of Jan Mayen (Wiborg, 1955;Aarbakke et al., 2017). This is an interesting example how the same species can occupy two very different habitat types, and it remains to be determined whether they represent isolated populations, subspecies or even two cryptic species. In general, P. minutus seems to prefer deeper, colder waters than the other Pseudocalanus species (Ershova et al., 2016), which would explain its low numbers in Adventfjorden. The low numbers that we observed in Billefjorden compared to previous studies could be due to our less frequent sampling here and thus missing the time window when this species peaks in abundance-at least for AF, which previously have been found to peak in abundance during March to May (Weydmann et al., 2013) and in Billefjorden during March (Aarbakke et al., 2017). However, it could also be due to improper identification in the earlier works or due to recent possible climate-induced shifts towards more boreal species (see below).

2019), including waters around Svalbard
Another unexpected result of our study was the yearround presence of the boreal P. moultoni in both examined fjords, where it composed up to 65% of the bulk Pseudocalanus abundance for several months of the year. Historically, P. moultoni was likely misidentified as P. acuspes, as these two species are virtually indistinguishable morphologically (Frost, 1989), and only P. acuspes was known to occur in the European Arctic (Frost, 1989). Genetically, however, these two species are among the most divergent Pseudocalanus species (Aarbakke et al., 2017). The distribution of P. moultoni was recently described using molecular identification, but these studies were only based on AF abundances (Aarbakke et al., 2014(Aarbakke et al., , 2017. Pseudocalanus moultoni has been observed in the North Atlantic Ocean, along the Norwegian coast, and in several fjords in Svalbard, including 'Arctic' ones such as the seasonally ice-covered Rijpfjorden and Billefjorden (Aarbakke et al., 2017). From these studies, it was unclear if P. moultoni was an advected Atlantic expatriate, such as Oithona atlantica or Limacina retroversa, or if Svalbard was part of its central distribution range. Only a few details are known about the ecology of P. moultoni, and those were mainly described prior to the development of molecular tools and thus before correct species identification could be proved (e.g. McLaren et al., 1989).

Life history strategies of Pseudocalanus in Svalbard
Arctic marine organisms have evolved life cycles adapted to the extreme seasonal environment in Polar regions (Hagen and Auel, 2001), which also determine prey availability for higher trophic levels. Life cycles of Arctic zooplankton have been well described for the genus Calanus (e.g. Falk-Petersen et al., 2009 and references therein), but small copepods, such as the numerous genus Pseudocalanus, have so far received little attention. Most Pseudocalanus life cycles were described before it was recognized that multiple species co-occur in the same Arctic and sub-Arctic locations (Pertsova, 1981;McLaren et al., 1989;Norrbin, 1991;Lischka and Hagen, 2005). Particularly the early life stages were easily misidentified, which may have skewed the results of these earlier works. The few recent studies that have used genetics to discriminate between species (Aarbakke et al., 2014(Aarbakke et al., , 2017Bucklin et al., 2015) have so far focused on AF only, which during most times of the year represent less than 5% of the population, or were studies restricted to single seasons (Ershova et al., 2017) not taking into account seasonal dynamics of the populations.
Polar copepod species generally differ from their lowlatitude counterparts in that they synchronize their reproduction to take advantage of the brief spurts of growth during the short Arctic summer, resulting in distinct cohorts, or generations, appearing in sequence. Species from lower latitudes, on the other hand, often exhibit continuous reproduction, such that all developmental stages are present in the plankton simultaneously. Of the three examined Pseudocalanus species, the life cycle of P. acuspes was the most typical for the Arctic (Fig. 8). Based on the seasonal dynamics of life stages, this species had a distinctive annual cycle in both sampling locations with a shift in generations in May to June and a prolonged period of reproduction between May and November. The earliest recruits produced in spring likely reached adulthood and maturity by August to September, resulting in a second summer generation, which explains the distinctly smaller females observed during that time compared to spring. Their offspring, as well as the remainder of the spring population, overwintered at the C3-C4 stages, suggesting that these are dedicated resting stages and that this species will not molt further in the absence of abundant food. During the peak phytoplankton production in May, these C3-C4 stages rapidly developed through the C5 stage and reached maturity shown as a distinct increase in adults in May after the onset of the spring bloom with first appearance of nauplii also occurring at this time.
Our results demonstrate that P. moultoni successfully overwinters in Adventfjorden and is able to complete its life cycle there. Although its overall dynamics in stage composition resembled that of P. acuspes, there were some notable differences. Unlike P. acuspes, P. moultoni had no pronounced seasonal abundance peaks in Adventfjorden and reached its highest abundance in the winter months, when abundance of P. acuspes dropped dramatically. Although reproduction of P. moultoni peaked during the summer, similar to P. acuspes, the nearly year-round presence of both early (C1-C3) and late (C5-AF) stages, including AM, suggest more continuous reproduction, in line with a more 'temperate' life history strategy (Fig. 8). The lack of a dominant overwintering stage combined with the presence of young life stages year-round suggests that this species is less seasonally restricted than the other two Pseudocalanus species. Similarly, the absence of change in body size of females in spring versus summer-autumn also indicates a less seasonally dependent reproductive strategy. It is likely that the unique 'temperate' conditions in Isfjorden allow this species to thrive there. Interestingly, Pseudocalanus acuspes had a distinct annual cycle, with peak reproduction observed during June to September, and a second, smaller generation appearing at the end of the summer period. Pseudocalanus moultoni had less pronounced peaks in abundance and stage composition and is presumed to have a more continuous reproduction, with an unknown number of generations per year. Pseudocalanus minutus presumably has an annual cycle in Billefjorden, timing its reproduction to the ice bloom, but does not complete its life cycle in Adventfjorden.
although present in much lower numbers, this species was also important in the Arctic Billefjorden. Their very low numbers in May relative to the other two species, however, suggests low survival over the winter, but by early autumn, their stocks are replenished either through local production or advection.
Pseudocalanus minutus was only observed in Adventfjorden as late stage (C4-C5) individuals, suggesting that this species does not complete its life cycle within this fjord (Fig. 8). The presence of this species coincided with the inflow of modified Atlantic water, so it is likely that it was carried there with currents from the deeper Atlantic waters, representing the deep Atlantic part, and not the Arctic part of this species distribution range (Wiborg, 1955;Aarbakke et al., 2017). The simultaneous presence of another indicator of deep Atlantic water masses, O. atlantica (Nyeggen, 2019), further supports advection to play a major role here. The only location where P. minutus nauplii were observed in significant numbers, indicating recent reproduction, was in Billefjorden in May, near the ice edge. Previous studies reported aggregations of P. minutus under the Arctic sea ice and evidence of direct feeding on ice algae (Conover et al., 1986;Runge and Ingram, 1991), suggesting that this is an important adaptation of this species to life in the Arctic. Pseudocalanus minutus reaches the largest body size of the examined species and has very efficient lipid turnover (Boissonnot et al., 2016), comparable to that of Calanus, with more than half of lipids deposited as wax esters. Although our dataset for Billefjorden includes only 4 months and does not cover the winter period, we deduct that this species also has a distinct annual cycle, with a shorter and earlier reproduction period than P. acuspes, fueled by ice algae or lipid reserves. Aarbakke et al. (2017) reported that 95% of AF of P. minutus found in March in Billefjorden belonged to P. minutus. Similarly, peak abundances of P. minutus females were observed in Rijpfjorden during the spring transition period, under sea ice containing high algal biomass (Weydmann et al., 2013). This likely represents the period when maturity is reached and reproduction takes place, resulting in a peak of nauplii ∼45 days later, in May. This allows the early life stages to take advantage of the later spring phytoplankton bloom and reach the main diapausing stages (C4-C5) earlier in the summer, as supported by a dominance of late stages during the remainder of the year (Fig. 8). As such, P. minutus is likely to be a capital breeder, employing a very different strategy than P. acuspes and P. moultoni, which reproduce opportunistically as food becomes available and can continue to reproduce throughout the season.

Pseudocalanus in a future Arctic
Svalbard has experienced dramatic changes in recent decades, with increasing influence of Atlantic water in the fjords carrying sub-Arctic fauna and with many previously ice-covered fjords remaining ice-free throughout the winter (Muckenhuber et al., 2016;Hop et al., 2019b). There is increasing evidence of communities across all trophic levels shifting from an Arctic to a more Atlantic character (Gluchowska et al., 2017;Vihtakari et al., 2018;Hop et al., 2019b). Within the zooplankton, recent shifts in species composition were observed in Kongsfjorden within the genus Calanus from the Arctic C. glacialis to the temperate C. finmarchicus (Hop et al., 2019b). It is fully expected that similar changes will occur, or have already occurred, for Pseudocalanus spp., which also contains both Arctic and temperate members. For example, the low numbers of P. minutus observed during our study, despite this species being reported as the dominant one in Svalbard previously, may represent anecdotal evidence that a species shift may already have taken place. Pseudocalanus minutus thrives in seasonally ice-covered waters, timing its reproduction to sea ice algae blooms (Norrbin, 1991;Weydmann et al., 2013), and several fjords on the west coast of Svalbard have been transitioning from more 'Arctic type' to more 'Atlantic type', without a winter ice cover (Cottier et al., 2010;Hop et al., 2019b). This may result in a decline of this species. However, this result may also have been due to inaccurate species identification in previous studies. Similarly, the higher contribution of the temperate P. moultoni than reported by previous studies may represent an increase in this species' relative importance or simply being a result of an increased effort in proper molecular identification. Regardless, this study provides an important baseline to monitor the absolute and relative contribution of these species in the future. Additionally, this study did not include species-specific primers for P. elongatus, another North Atlantic species that has not yet been recorded in Svalbard or the Arctic but that can also be potentially advected from temporal regions. We cannot exclude that some of the failed PCRs in our study belonged to this species, and this will be up to future studies to resolve.
The coexistence of three very similar species with slightly different life histories and environmental tolerances ensures that despite possible short-or longterm shifts in their relative numbers, e.g. if P. minutus are replaced with P. moultoni, the overall abundance and productivity of the genus may largely remain the same, ensuring their availability for higher trophic levels.
Although P. minutus is larger and more lipid rich than the other species (Boissonnot et al., 2016), these differences in energy content may be offset by higher growth and turnover rates of the smaller species, as is predicted for the Calanus complex, with C. finmarchicus replacing the larger, more lipid-rich C. glacialis (Renaud et al., 2018). In fact, these authors predict that the energy transfer between trophic levels will become more efficient under these future scenarios. As such, the coexistence of the Pseudocalanus species complex similarly represents a mechanism of resilience of Arctic marine systems to climate change.

CONCLUSION
Our study highlights the importance of seasonal, speciesspecific investigations that focus on the entire population of an organism spanning an entire seasonal cycle, not just their adult stage. This can lead to very different conclusions, as evidenced by previous studies that found P. minutus to be the dominant species in Svalbard. Furthermore, we demonstrate that closely related species can differ significantly in their life cycle strategies and contribution to the biomass of the ecosystem. P. acuspes had a distinct annual life cycle in Isfjorden, being the dominant species during the summer months and peaking in reproductive activity and the presence of early life stages. Surprisingly, the boreal P. moultoni was present year-round in both fjords and was equally important or even the most numerous Pseudocalanus species in winter. This species had a more even distribution of life stages throughout the year, suggesting it to have a continuous reproduction. The Arctic P. minutus was the least abundant of the three species in both fjords. It was able to complete its life cycle in the Arctic Billefjorden but not in the Atlantic-influenced Adventfjorden. Since the different species have different environmental adaptations, further climate-related changes in the Arctic may shift their distributions and relative importance in zooplankton communities. The cooccurrence of several closely related species, however, can compensate in the individual species population success, thus ensuring high-latitude zooplankton communities to be more resilient to climate change impacts at the genus level but not necessarily at the species level.
conditions. We also sincerely thank Kim Praebel (UiT The Arctic University of Norway) and Tatiana Neretina (Moscow State University) for providing molecular lab facilities and logistical support. We also thank two anonymous reviewers for their constructive feedback that helped improve the manuscript.

FUNDING
This work is part of the seasonal time series studies in Isfjorden IMOS led by UNIS with financial support from Økokyst Svalbard, Fram Centre in Tromsø (Arctic Ocean Flagship; project FADE) and the 2017-2018 Belmont Forum and BiodivERsA joint call for research proposals, under the BioDivScen ERANet COFUND programme (ACCES, Project No. 296836). The study was also performed in the framework of the state assignment of IO RAS (theme No. 0149-2019-0008) with additional financial support by the Russian Foundation for Basic Research in the framework of Project No. 18-35-00341 mol_a.