Effects of a Submesoscale Oceanographic Filament on Zooplankton Dynamics in the Arctic Marginal Ice Zone

Introduction

Mesoscale dynamics (horizontal scale of 10–100 km), including fronts and eddies, have been studied extensively and, even though still not fully understood, significant knowledge about underlying oceanographic mechanisms as well as their influence on biological and biogeochemical processes has been gained (McWilliams, 2008; McGillicuddy, 2016). Impacts can vary widely, including the support of new production by inducing vertical mixing and thus modulating the supply of nutrients into the euphotic layer (e.g., Lee and Williams, 2000; Mahadevan and Archer, 2000; McGillicuddy et al., 2003). Numerical studies investigating the mesoscale injection of nutrients found that estimated primary production increased by up to 30%, better matching empirical observations (Lévy et al., 2001). Besides enhancing vertical velocities, mesoscale dynamics can also act as vectors of transport for organisms in a horizontal direction. Fronts have been shown to concentrate organisms at the site of convergence, creating foraging areas, which are actively sought by top predators such as tunas, birds and whales (Acha et al., 2015). Several studies further demonstrate that mesoscale eddies can serve as a vector of transport, dispersing zooplankton and larvae to oceanic regions (Mackas and Galbraith, 2002; Batten and Crawford, 2005), or as a retention mechanism, keeping offspring close to their spawning site (Singh et al., 2018). However, while mesoscale dynamics play a vital role for such processes, the resolution of empirical studies still seems to be too coarse.

Finer, so-called submesoscale, dynamics have thus recently gained increasing interest among physical and biological oceanographers (Lévy et al., 2018). In contrast to mesoscale fronts and dynamics, submesoscale structures are rapidly evolving features with horizontal scales of 0.1–10 km. They are defined by a high Rossby number ⪆1 (mesoscale: ≤0.1), resulting in ageostrophic circulation and associated significant vertical velocities (D’Asaro et al., 2018; Zhabin and Andreev, 2019; Rossby number: horizontal velocity gradients divided by Coriolis frequency). Thus, meso- and submesoscale dynamics are different regimes and cannot be summarized as one process.

Although the existence of submesoscale motions has been acknowledged since the 1980s (McWilliams, 1985), their impact was considered as weak, and computational abilities were too limited to include them in numerical simulations (Marsac et al., 2014). However, expanding computational capacities during the last decade allowed the modeling of oceanographic structures on a higher resolution. Such models highlighted the intense vertical and horizontal velocities within submesoscale dynamics and hence indicated their underestimated potential in affecting pelagic fluxes and processes (Klein et al., 2008; Klein and Lapeyre, 2009).

Submesoscale structures are ubiquitous throughout the oceans and often appear between mesoscale eddies (Lévy et al., 2001, 2018) or in the presence of strong horizontal gradients, for instance in the marginal ice zone (MIZ) (von Appen et al., 2018). Due to their small spatial (0.1–10 km) and short temporal (hours to weeks) scales, in situ measurements, especially the assessment of their impact on biological processes, are difficult. Recent studies utilizing satellite data to analyze the effect of submesoscale fronts on phytoplankton revealed increased chlorophyll concentrations in such regions (Shulman et al., 2015; Liu and Levine, 2016; Guo et al., 2019). The main underlying cause is presumably due to the intense vertical velocities associated with small-scale dynamics, which lead to an enhanced upward transport of nutrients (Mahadevan and Archer, 2000; Mahadevan, 2016). Although submesoscale processes are limited in space and time, this nutrient injection has major implications for phytoplankton productivity, as the relevant time scales are similar to those of phytoplankton growth (Mahadevan, 2016). These studies emphasize the impact of submesoscale structures on phytoplankton, yet little is known about interactions between such small-scale dynamics and zooplankton ecology. Zooplankton research is still largely dependent on point sampling by net hauls from research vessels, with the distance between stations usually too coarse to resolve submesoscale features. Few studies reported elevated mesozooplankton abundances at submesoscale fronts (Ohman et al., 2012; Powell and Ohman, 2015; Trudnowska et al., 2016), yet the biological and physical mechanisms supporting such accumulations are still far from understood.

During the expedition PS107 of the German research icebreaker RV Polarstern to the Arctic marginal ice zone in Fram Strait in July/August 2017, a submesoscale filament was detected by in situ observation and satellite imagery (Schewe, 2018). A high-resolution ad hoc sampling campaign was designed, combining physical oceanographic measurements (published by von Appen et al., 2018) with depth-stratified zooplankton sampling. The objective of the present study is to link community structure, vertical distribution and dynamics of zooplankton to physical oceanographic properties of the submesoscale filament. Specifically, we demonstrate that (1) despite the short temporal and small spatial scale of the submesoscale feature, the convergence of surface water results in the accumulation of epipelagic zooplankton close to the filament center; (2) distinct water masses at different depths and/or on different sides of the filament are characterized by different zooplankton communities despite the small vertical and horizontal distances; (3) mesopelagic species emerge and occur at shallower depths close to the filament center due to doming of the isopycnals in the mesopelagic layer; and (4) the along-frontal jets on both sides of the filament have the potential to act as high-speed vectors for the transport of zooplankton organisms.

Materials and Methods

Study Area and Physical Oceanographic Characterization of the Submesoscale Filament

Fram Strait, between Greenland and the Svalbard Archipelago, is the only deep-water connection between the North Atlantic and the Arctic Ocean. On the eastern side, the West Spitsbergen Current (WSC) transports warm, saline water masses from the Atlantic (‘Atlantic Water’ – AW) northwards (von Appen et al., 2016). On the western side, the East Greenland Current (EGC) exports cold and rather fresh Polar Surface Water (PSW) and sea ice from the Arctic Ocean into the Atlantic (de Steur et al., 2009). Part of the AW flows northward into the Arctic Ocean, while the rest turns westward to ‘recirculate’ and subduct below the PSW (Hattermann et al., 2016). The zone in between the WSC and EGC is thus characterized by a highly dynamic and turbulent regime of mesoscale features (Kawasaki and Hasumi, 2016; von Appen et al., 2018). This highly dynamic regime, together with the semi-permanent sea-ice edge (von Appen et al., 2016) and the large horizontal density gradients observed in the MIZ, may be beneficial for and contribute to the development of submesoscale features (von Appen et al., 2018).

During the expedition PS107 with the German research icebreaker RV Polarstern, a submesoscale filament was detected via satellite data (Figure 1A), due to the presence of a nearly straight sea ice streak of 500 m width and 50 km length extending in northeast-southwest direction in the MIZ of Fram Strait. Extensive in situ oceanographic as well as biological sampling of the filament was conducted in cross-frontal direction (Schewe, 2018; Figure 1A). The physical oceanographic structure of the filament was described in detail by von Appen et al. (2018). The observed ‘cyclonic filament’ is comparable to the theoretical description of a ‘dense filament’ by McWilliams et al. (2009). It can be regarded as two parallel fronts between light-dense and dense-light surface waters. The flow field is characterized by geostrophically balanced cyclonic along-frontal jets and an ageostrophic secondary circulation in cross-frontal direction. Specifically, cold polar water (≤0°C) prevailed at the surface at distances of 6–8 km on both sides of the filament, while in between warmer Atlantic water (≥1°C) dominated the uppermost layer (Figure 1B present study; Figures 2A, 3B in von Appen et al., 2018). The ageostrophic circulation resulted in convergence toward the center of the filament, accumulating and stabilizing the narrow streak of sea ice (Figures 1A,C). The convergence led to downwelling, subducting dense water which was evident in a depression of the 27 isopycnal at the center of the filament (Figure 1C). Vice versa, below 100 m depth, a doming, i.e., an upward deflection, of isopycnals was apparent, resulting in a characteristic hourglass shape of isopycnals at the filament center (with depressed isopycnals at the surface and doming ones below 100 m). Associated with this structure, strong geostrophic along-frontal jets with a maximum speed of >0.5 m s^–1 in 50–100 m depth at around 3 km distance from the center occurred on both sides of the filament (Figures 1B,C). On the eastern side, the along-filament flow was directed northward, whereas the jet on the western side flowed southward. The temporal persistence and frequency of such filaments in the MIZ is unknown. Our best guess for the time period over which the sampled submesoscale filament persisted is one to a few weeks.

Figure 1. (A) Sentinel 1A (©European Space Agency) radar reflectivity on July 26, 2017 including the position of the cross frontal transect with the four MultiNet stations. (B) CTD profiles of temperature and salinity as well as along-filament velocities at the four MultiNet stations. (C) Schematic illustration of the submesoscale filament in the Arctic marginal ice zone in cross-frontal direction (modified after von Appen et al., 2018). PW, polar water. Colored triangles indicate locations of MultiNet stations.

Zooplankton Sampling, Abundance, and Biomass Analysis

Four stations (stns) across the submesoscale filament were analyzed for zooplankton abundance, biomass and taxonomic composition. Stns 10 and 12 were located near the center, stns 14 and 16 further away from the center (Figure 1A). Two stations were situated on either side of the central convergence zone, i.e., stns 16 and 10 on the western side and stns 12 and 14 on the eastern side (Table 1 and Figures 1A,C). Sampling was conducted within 30 h (Table 1) with a multiple opening and closing net equipped with five nets for stratified vertical hauls (Hydrobios MultiNet Midi, mouth opening 0.25 m², mesh size 150 μm, sampling speed 0.5 m s^–1). Oceanographic observations indicated an extent of the filament to >250 m depth, thus, zooplankton sampling depths were adjusted accordingly, with five vertical sampling intervals covering 300–200–100–50–10–0 m. Immediately after the haul the samples were preserved in a 4% borax-buffered formaldehyde seawater solution. Prior to each MultiNet station, oceanographic data were obtained by a CTD attached to a rosette water sampler at the same stations.

Table 1. Geographic position, sampling date and time as well as the distance to the center of the filament of the four MultiNet Stations sampled across a submesoscale filament in Fram Strait.

For the microscopic analysis, zooplankton samples were split into subsamples (1/2 to 1/128, depending on the total abundance of the sample) using a Motoda plankton splitter (Motoda, 1959). Identification was performed to the lowest possible taxonomic level using a dissecting microscope Leica MZ12.5. Copepoda (calanoid, cyclopoid, and harpacticoid), Amphipoda, Euphausiacea, and Pteropoda were generally identified to genus or, if possible, to species level, including developmental stages for Copepoda. Individuals were counted until at least 80 specimens were reached within a complete subsample. Biomass values were calculated for abundant species by multiplying the abundance data with published individual dry mass values for the Greenland Sea (Richter, 1994 and references therein).

In the Arctic, three Calanus species co-occur and usually dominate zooplankton communities in terms of biomass and ecological importance (Mumm et al., 1998; Auel and Hagen, 2002; Kosobokova and Hirche, 2009). While Calanus hyperboreus is easily identified based on morphological characteristics and its larger body size, Calanus finmarchicus and Calanus glacialis are two sister species, which are morphologically not distinguishable. A differentiation between both species based on prosome length has been proposed and widely used, considering copepodite stages CV larger than 2.9–3.0 mm and adult females larger than 3.2 mm to be C. glacialis, whereas smaller individuals are assigned to C. finmarchicus (e.g., Unstad and Tande, 1991; Kwaśniewski et al., 2003). However, several recent studies question the applicability of prosome length as a valid characteristic for Calanus species identification, as the size ranges of both species can considerably overlap (e.g., Nielsen et al., 2014; Choquet et al., 2017, 2018). Thus, fixed size thresholds are not applicable for species identification. In this study, we measured the prosome lengths of all 2,661 C. finmarchicus and C. glacialis specimens to check the applicability of length-frequency distributions for species identification. Depth-specific size distributions generally revealed distinct, non-overlapping bimodal size distributions, based on which adult females and copepodids CV of these two species were distinguished from each other. Calanus copepodids CI–III were pooled as Calanus spp. CI–III. Calanus copepodids CIV were separated between C. hyperboreus CIV and C. finmarchicus/glacialis CIV.

Community Analysis and Impact of Environmental Factors

Statistical analyses were conducted with PRIMER6 (version 6.1.6; Clarke and Gorley, 2006) and R (version 3.5.3, R Core Team, 2019). To investigate if and how the stations differed in terms of species composition, averaged Bray–Curtis dissimilarities were calculated based on log(x + 1) transformed total abundances of all species and on depth-specific abundances at each station. A non-metric multidimensional scaling (nMDS) plot and dendrogram were generated with PRIMER6 to visualize the differences, respectively.

To further evaluate how environmental factors affected the distribution of zooplankton species across the filament, we applied canonical correspondence analysis (CCA) using the vegan package in R (version 2.5–6, Oksanen et al., 2019). CCA is a multivariate method to help explain the relationships between species assemblages and their environment (ter Braak and Verdonschot, 1995; Greenacre and Primicerio, 2013) and has found widespread use in aquatic science (e.g., Herman and Dahms, 1992; Pinto-Coelho et al., 2005; Sell and Kröncke, 2013). For the analysis, species abundances (ind. m^–3) for each of the five depth intervals were used with the mean depth and mean temperature of the respective depth interval as well as the distance of the four stns to the filament center as the abiotic factors. Distance was used as positive values. Thus, the analysis does not differentiate between east and west of the filament center, but rather indicates the distribution of species close to and far away from the center of the filament. Permutation tests were further performed in R with the vegan package using the anova.cca function (version 2.5–6, Oksanen et al., 2019).

Results

Zooplankton Abundance and Biomass

Total zooplankton abundance was by far the highest in the upper 50 m of stn 10 (0–10 m: 18,257 ind. m^–3, 10–50 m: 10,912 ind. m^–3), close to the center of the filament (Figure 2). Below 50 m numbers dropped to 1,152 ind. m^–3 (50–100 m) and less than 250 ind. m^–3 (100–300 m). Surface abundance at stn 12 was also high but approximately half the values of stn 10. The elevated numbers at both stns were largely due to Oithona spp. and copepod nauplii. Stns 14 and 16 had lowest total abundances in surface waters (<1,800 ind. m^–3), values in the upper 50 m comprised less than 16% of the numbers at stn 10. Between 100 and 300 m Microcalanus spp., Metridia spp., and Oncaea spp. became important components in terms of contribution to abundance. Total abundance below 100 m was similar at all stns.

Figure 2. Vertical distribution of total zooplankton abundance (ind. m^–3, A) and biomass (mg dry mass m^–3, B) across the submesoscale filament. Colors indicate the contributions of different taxa. Colored triangles indicate the location of the four stations (16, 10, 12, and 14) across the filament.

Similar to abundance, total biomass was highest in surface waters at the center of the filament (stn 10), followed by stn 12 (Figure 2). At the outer stations, biomass in the upper 50 m was 27% (stn 14) or even below 20% (stn 16) of the value at stn 10. This difference was again mainly due to the dominance of Oithona spp. at stns 10 and 12, but higher numbers of copepod nauplii, Pseudocalanus spp., Calanus CI–III, C. finmarchicus/glacialis CIV also contributed to the increased biomass. Due to their small size, Oithona spp. are usually not the main contributors to biomass. However, at stn 10, and partly also at stn 12, Oithona spp. appeared in such high densities (13,126 ind. m^–3 at stn 10), that it comprised up to 78% of total biomass in the upper 10 m, compared to only 30% at stns 14 and 16. Calanus spp. were main contributors to total biomass throughout the sampled water column, while Metridia longa contributed significantly below 100 m. Similar to abundance, the variability of total biomass between the stns decreased with depth.

In total, 35 different taxonomic categories were identified at the four stns (Table 2). With at least 27 species, copepods were the most diverse and abundant taxon, comprising about 80% of total abundance at all stns. While Calanoida were the most species-rich order, Cyclopoida (mainly Oithona spp.) dominated in terms of abundance with 59–93% of all copepods at all stns.

Table 2. Abundance (ind. m^–3) of zooplankton taxa at the four stations across the submesoscale filament in Fram Strait, July/August 2017.

Zooplankton Distribution Across the Filament

Visualization of Bray–Curtis dissimilarities between the four stations as a nMDS plot indicates that stns 14 and 16, even though they were the furthest apart from each other, were the most similar with regard to species composition (Figure 3A). Stn 10 deviated most from the other stations. Stn 12 was somewhere in between the central stn 10 and the outer stns 14 and 16. The hierarchical cluster analysis of depth-specific species composition at each station revealed that depth had a strong influence on differences between and within stations (Figure 3B). Two main clusters were formed, with surface waters (0–100 m, D1–3) in one and lower epipelagic to mesopelagic depths in the other cluster (100–300 m, D4–5), except for 50–100 m (D3) of stn 10 and 12 which was included in the cluster of surface samples. Similar to the nMDS plot, the hierarchical cluster analysis also indicated that stns 14 and 16 were more similar, as the individual depth intervals of both stns clustered together. Stns 10 and 12 also had lower distances to each other, except 0–10 m (D1) at stn 12, which grouped to stns 14 and 16.

Figure 3. (A) Non-metric multidimensional scaling (nMDS) plot of the four stations across the filament based on a Bray–Curtis dissimilarity matrix of log(x + 1) transformed total abundances of all species at each station. (B) Hierarchical cluster analysis based on a Bray–Curtis dissimilarity matrix of log(x + 1) transformed depth-specific species abundances at each station. D1: 0–10 m, D2: 10–50 m, D3: 50–100 m, D4: 100–200 m, D5: 200–300 m.

To evaluate distribution patterns across the filament, a canonical correspondence analysis (CCA) was performed, using depth, temperature and the absolute distance to the filament center as environmental variables. The two axes with highest explanatory factors were used to generate the ordination plot (Figure 4). The first axis explained 60% of total variance, the second axis added another 4%. Thus, 64% of total variance was explained. Depth was mainly correlated to CCA1, i.e., increasing toward the right. Temperature and distance to the filament center were associated with the second axis, but in opposite directions, i.e., lower temperatures with increasing distance to the center.

Figure 4. Canonical correspondence analysis (CCA) triplot of the most common zooplankton taxa, environmental variables (depth, temperature, and distance to the filament center) and individual depth-stratified samples across the submesoscale filament. Stn: Station. D1–D5 indicate depth intervals, D1: 0–10 m, D2: 10–50 m, D3: 50–100 m, D4: 100–200 m, D5: 200–300 m. The first axis (CCA1) explains 60%, the second axis (CCA2) explains 4% of the total variance. Colored circles indicate the association of taxa with the four distribution types, red: convergence and concentration of surface inhabitants close to the filament center, blue: polar taxa at outer stations, black: emergence of mesopelagic taxa beneath the filament, yellow: mesopelagic taxa, including travelers with along-frontal jets. Indicated in bold are the main contributors of respective distribution types. Species abbreviations: C. glac, Calanus glacialis; C. hyp, Calanus hyperboreus; C. fin, Calanus finmarchicus; P. glac, Paraeuchaeta glacialis; P. norvegica, Paraeuchaeta norvegica; M. norvegica, Microsetella norvegica; T. libellula, Themisto libellula; T. abyssorum, Themisto abyssorum; T. longicaudata, Thysanoessa longicaudata; L. helicina, Limacina helicina. CI to CVI, copepodite stage; f, female; m, male.

Based on the arrangement of the taxa in the CCA triplot, the following four distinct distribution patterns across the filament were identified:

Convergence and Concentration of Surface Inhabitants Close to the Filament Center

Taxa of this group showed an accumulation in the surface layer (upper 50 m) close to the center of the filament, while occurring at much lower concentrations at the outer stations. This group includes Oithona spp., copepod nauplii, Pseudocalanus spp., Calanus copepodids CI-III, C. finmarchicus/glacialis CIV, Microsetella norvegica and C. hyperboreus CIV (Figure 4, red circle). The distributions of the two most abundant representatives of this group (Oithona and nauplii) are visualized in Figure 5, Group A.

Figure 5. Distribution heatmaps of most abundant and/or ecologically important zooplankton taxa sorted according to the four distribution types (Groups A–D) across the submesoscale filament. Numbers inside the heatmaps show abundances (ind. m^–3). Darker shading indicates higher abundances.

Polar Taxa at the Outer Stations

In contrast to taxa of Group A with high abundances in surface waters close to the filament center, C. glacialis CV and adult females, C. hyperboreus CV and adult females, Paraeuchaeta glacialis CV–CVI, ophiopluteus larvae, the pteropod Limacina helicina, Bivalvia larvae, Mysidacea and Appendicularia were generally more abundant in the polar surface waters of the outer stns 14 and 16 (Figure 4, blue circle).

This distribution pattern was most prominent for ophiopluteus larvae (Figure 5, Group B), which were completely absent at stn 10 and only appeared in low numbers in the upper 100 m at stn 12 (51 ind m^–2), while exhibiting much higher densities at stn 14 (261 ind m^–2) and stn 16 (903 ind m^–2). Appendicularians were the most abundant representative of Group B (Figure 5), with high numbers in the upper 100 m at the outer stns (631 ind. m^–2 at stn 14; 1,222 ind. m^–2 at stn 16), but also exhibiting high abundances at stn 10 (952 ind m^–2). C. glacialis and C. hyperboreus copepodids CV and adults were most abundant in the Polar Surface Water at the outer stns (Figure 5, Group B). While C. glacialis occurred at both sides of the filament, C. hyperboreus was only present on the eastern side.

Emergence of Mesopelagic Taxa Beneath the Filament/Isopycnal Upwelling

Occurrences of Metridia spp., comprising mainly M. longa and only few Metridia lucens, and Heterorhabdus spp. (mainly Heterorhabdus norvegicus) were both highly correlated with depth (Figure 4, black circle). In contrast to the surface-inhabiting taxa of Groups A and B, their distribution was less influenced by the distance to the filament center.

The distribution pattern is depicted for Metridia spp. in Figure 5, Group C. Metridia was rare in the epipelagic at all stns and showed highest abundances below 200 m. At stn 10, Metridia spp. showed an elevated abundance between 100 and 200 m compared to the other stns, linked to the characteristic doming of the isopycnals at that depth close to the center of the filament.

Travelers With the Along-Frontal Jets

Group D includes Microcalanus spp., C. finmarchicus adult females (presented in Figure 5) and Ostracoda. While these taxa were rather abundant throughout the sampled water column below 50 m, patches of higher concentrations occurred at the shoulders of the filament in the areas of the frontal jets. This pattern was most distinct for Microcalanus spp. (Figure 5, Group D), with minimum abundances at stn 10, but maxima between 100 and 300 m at stns 12, 14, and 16. Adult females of C. finmarchicus showed a similar pattern, but in contrast to Microcalanus spp., they were more abundant on the eastern side of the filament at stns 12 and 14 between 50 and 200 m (Figure 5, Group D). In contrast, ostracods were most abundant on the western side of the filament at stn 16 between 100 and 300 m (Table 2).

Calanus finmarchicus copepodids CV occurred in the lower epipelagic to mesopelagic layers, but without emergence at the filament center. Similar to females, they were also found in the northward flowing along-front jet on the eastern side. Highest densities were determined between 200 and 300 m at stns 10 and 12 and between 50 and 200 m at stn 16 (Table 2).

Permutation tests to analyze the importance of the environmental variables revealed that depth was highly significant in explaining the variance of zooplankton distribution across the filament (p = 0.001), whereas temperature (p = 0.112) and distance to the filament center (p = 0.449) were not significant. However, when only surface species (red and blue clusters; Groups A and B) were included in the analysis, temperature became a significant variable (p = 0.031), i.e., temperature significantly impacted epipelagic zooplankton distribution. The p-value of distance to the filament center also improved, but remained non-significant (p = 0.082).

Depth-Dependent Size-Frequency Distributions of C. finmarchicus and C. glacialis

Prosome lengths of C. finmarchicus and C. glacialis copepodids CV and adult females increased with depth at all stns (Figure 6). This trend was most pronounced in copepodids CV (Figure 6A), but was also observed in adult females (Figure 6B). Thus, fixed size thresholds are not applicable for the discrimination of the two co-occurring Calanus species. As individual size increases with depth, the threshold to differentiate between C. finmarchicus and C. glacialis has to increase as well, otherwise deeper occurring larger specimens of C. finmarchicus would be incorrectly identified as C. glacialis.

Figure 6. Density plot of size distributions of C. finmarchicus/glacialis CV (A) and adult females (B) prosome lengths for four distinct depth intervals at stations 10, 12, 14, and 16. Sampling intervals 0–10 m and 10–50 m were pooled due to low numbers of individuals. Second y-axes indicate kernel density estimations. Stations are arranged in cross-frontal direction. Each dot within the size-frequency distribution indicates one individual per respective size. Black lines show the thresholds for species separation of C. finmarchicus and C. glacialis, i.e., smaller individuals to the left of the respective line were identified as C. finmarchicus, larger specimens to the right as C. glacialis.

Several samples, in particular those from the outer stns dominated by Polar Surface Water, showed bimodal length-frequency distributions of Calanus with smaller secondary maxima at 3.2–3.5 mm in copepodids CV and >3.5 mm in adult females, respectively (Figure 6). For the present study, individuals from those secondary peaks were considered C. glacialis with a depth-dependent increase in the length threshold for species identification (black lines in Figures 6A,B).

Discussion

Calanus Depth-Dependent Size Distribution

Body size of C. finmarchicus and C. glacialis copepodids CV and adult females increased with increasing depth even on the relatively small vertical scale from 0 to 300 m (epipelagic to upper mesopelagic). Such a body size-dependent habitat partitioning within developmental stages of Calanus has to our knowledge not been published before. Calanus species are capable of diel vertical migrations (DVM), seeking shelter from predators in deeper layers during day and migrating to the surface to feed under the cover of darkness at night. However, several studies demonstrate that in polar regions DVM is only weakly pronounced during periods of midnight sun due to only little fluctuations in the light regime (Blachowiak-Samolyk et al., 2006; Cottier et al., 2006). The sampling campaign of the submesoscale filament was within the timespan of constant daylight. Further, local time of samplings varied over the day between the stations (9 pm, 5:30 am, 5 pm, and 2 am). Although sampling times differed, the size-depth distribution pattern stayed the same at the four stations. Hence, it is highly unlikely that DVM was responsible for the size-depth distribution of Calanus. Among other explanations of depth-dependent increases in body size or vertical segregation by size in zooplankton are predator avoidance with size-specific differences in predation risk (Hunt and Harrison, 1990; De Robertis et al., 2000) and niche selection (e.g., Laakmann et al., 2009, for vertical partitioning to avoid inter-specific competition; Kaiser et al., 2018, for cryptic species). In addition, variations in lipid content may affect buoyancy and the start of the ontogenetic descent to overwintering depths (Hirche, 1997; Melle et al., 2014). However, those mechanisms usually act on larger scales and between ontogenetic stages. Vertical sampling with high spatial resolution by optical methods (e.g., LOKI, Hirche et al., 2014) will be required to elucidate the reasons for the fine-scale vertical zonation by body size within ontogenetic stages of Calanus.

Ecological Roles of Submesoscale Filaments

The present study reveals several key effects of submesoscale structures on zooplankton distribution, ecology and dynamics.

Accumulation of Zooplankton Biomass

Convergence of surface water has the potential to increase abundance and biomass of planktonic organisms in frontal zones (Epstein and Beardsley, 2001; Strass et al., 2002; Ohman et al., 2012). Empirical studies often focus on phytoplankton, as the applicability of methods like remote sensing allows an investigation of the influence of small-scale structures in much easier and feasible ways compared to zooplankton research. With the aid of satellite data those studies identified increased chlorophyll concentrations associated with submesoscale fronts (Shulman et al., 2015; Liu and Levine, 2016; Guo et al., 2019).

Across the filament, extreme differences in densities of epipelagic zooplankton species were observed, with maximum abundance and biomass values in the upper 50 m at the stns closest to the center of the filament (stns 10 and 12). The concentration of epipelagic zooplankton in the center of the filament is likely the result of physical-biological interactions (Folt and Burns, 1999). Physically, the convergence of surface water associated with the submesoscale filament, which also led to the accumulation of sea ice (von Appen et al., 2018), concentrates epipelagic organisms at the site of downwelling. Among the zooplankton taxa which occurred across the filament, different swimming capabilities are found. Euphausiid species and the hyperiid amphipod Themisto are usually considered good and active swimmers (Kraft et al., 2012; Richerson et al., 2015) and thus may resist accumulation processes. Generally, however, planktonic organisms passively drift with horizontal ocean currents, but they can, to some extent, swim against vertical velocities (Genin et al., 2005). Several studies demonstrated a strong fidelity of zooplankton species to particular depth layers and associated environmental conditions (Ashjian and Wishner, 1993; Ashjian et al., 1994).

Active vertical counter-swimming leads to zooplankton accumulation at downwelling sites (Olson et al., 1994). In our case, an over 10- and 16-fold increase in abundance of epipelagic zooplankton close to the center of the filament in comparison to the outer stations supports our first statement.

Ohman et al. (2012) detected elevated mesozooplankton abundances at a submesoscale front in the California Current, including local maxima of calanoid copepods and Oithona. A higher ratio of nauplii to copepodids at the front suggested enhanced secondary production. In our study, the nauplii to copepodids ratio remained constant across the filament. However, the presence of Calanus males and eggs at stn 10 and 12 (Table 2, H. Auel, pers. obs.) indicated reproductive activities. The generation times of zooplankton usually exceed the persistence of small-scale dynamics. Nevertheless, the enhanced availability of phytoplankton at convergence zones creates beneficial feeding conditions and may stimulate secondary production. In turn, elevated concentrations of zooplankton at small-scale frontal zones represent foraging hotspots for top predators, as indicated by increased foraging success or time spend at such fronts for tunas (Snyder et al., 2017), seabirds (Hyrenbach et al., 2006; De Monte et al., 2012), elephant seals (Rivière et al., 2019; Siegelman et al., 2019) and whales (Davis et al., 2002).

Structuring the Pelagic Realm

Depth is usually the main factor structuring zooplankton communities, as demonstrated for the Arctic by Auel and Hagen (2002) and Kosobokova et al. (2010). However, the present study shows that submesoscale filaments can alter the vertical distribution of certain zooplankton species on very small horizontal and vertical scales. The doming of isopycnals in the upper mesopelagic at the filament center leads to an emergence of the mesopelagic copepods Metridia spp. and Heterorhabdus spp. Most likely, they passively followed the upwelling of the surrounding water body in order to stay in their preferred environmental conditions in terms of temperature and salinity (Ashjian and Wishner, 1993; Ashjian et al., 1994). In addition, enhanced food availability (phytoplankton and small copepods) caused by the accumulation through convergence in the center of the filament could have attracted these omnivorous copepods closer to the surface.

The center and intermediate depths of the filament were characterized by Atlantic Water and its associated species, such as C. finmarchicus and P. norvegica, but also carried more widespread taxa such as Microcalanus. In contrast, surface waters at the outer stns 14 and 16 were of polar origin and contained clearly different communities. Polar species, such as the copepods C. glacialis, C. hyperboreus, and P. glacialis, the pteropod L. helicina and ophiopluteus larvae, were considerably more abundant in those waters. The high spatial heterogeneity of zooplankton communities on small spatial scales matches physical oceanographic observations, which show that distinct unmixed water masses can occur close to each other in submesoscale filaments (von Appen et al., 2018).

The results support our second and third statements: different water masses associated with the filament are characterized by distinct zooplankton communities even on short spatial scales and the doming isopycnals at the filament center cause the emergence of mesopelagic species.

Vector for Biological Connectivity

Associated with the submesoscale filament, along-frontal jets with enhanced velocities at intermediate depths on both sides of the convergence zone were observed. Hancke et al. (2014) suggested that such fast-flowing along-frontal currents could act as vectors of biological connectivity. They demonstrated that drifters traveling within frontal jets crossed the Mozambique Channel significantly faster compared to drifters trapped in much slower propagating eddies. The transport in those jets represents a suitable time frame for the survival of planktonic larvae and thus may explain the biological connectivity between regions across the Mozambique Channel (Hancke et al., 2014; Marsac et al., 2014).

In the current study, patches of higher abundances of certain species were detected within the along-frontal jets. It is important to note that abundances of those species were different on both sides of the filament. Thus, the northward and southward jets should not be seen as compensating each other with zero net transport. For instance, abundance of C. finmarchicus was higher in the northward jet, indicating a northward net transport. The along-frontal jets exhibited a maximum speed exceeding 0.5 m s^–1 (von Appen et al., 2018), which is equivalent to 43 km per day. In comparison, the mean current velocity of the core of the West Spitsbergen Current is around 0.15 m s^–1 with a maximum of >0.2 m s^–1. Velocities of the offshore West Spitsbergen Current branch range between 0 and 0.15 m s^–1 (Beszczynska-Möller et al., 2012). Similar and even higher values for along-frontal jets, up to 1 m s^–1, were reported by Hernández-Hernández et al. (2020) for a submesoscale frontal zone south of the Canary Islands. A drifting sediment trap, which was deployed during our filament survey close to the northward frontal jet, traversed 26 km in 21 h, further emphasizing the impact of such filament-associated flows (von Appen et al., 2018).

The length of the sea-ice streak suggests a horizontal extension of the filament of at least 50 km. Thus, zooplankton traveling within the fast-flowing jets may only have a short residence time of one to a few days in the filament. However, Hancke et al. (2014) demonstrated the connectivity between fronts by drifters moving from one frontal zone to another, enabling the coverage of large distances in a short time. Especially in highly dynamic systems like the MIZ, submesoscale dynamics may be omnipresent and could thus (involuntarily) be utilized by zooplankton as ‘transportation highways’. Hence, submesoscale dynamics could not only play a role in plankton patchiness and structuring the pelagic realm, but the associated along-frontal jets may also be relevant for biological connectivity and species distribution, supporting our fourth statement.

The elevated abundance of CV and females of C. finmarchicus, an expatriate species from the boreal-Atlantic, in the northward flowing eastern jet could particularly be of importance for the Arctic marine ecosystem. Due to ongoing climate change and associated rising temperatures in the Arctic, shifts in the biogeographical distribution of pelagic species can be observed, with Atlantic species extending their ranges northward (Beaugrand et al., 2009). Further, due to rising temperatures, the width of the MIZ is constantly increasing during summer (Strong and Rigor, 2013), possibly providing an extending area for submesoscale dynamics and associated fast-flowing jets, which may accelerate ‘Atlantification’ processes in the Arctic, although further supportive evidence is needed.

Atlantic zooplankton species tend to be smaller, less lipid-rich, i.e., less nutritious, and have different life-cycle strategies than their Arctic counterparts (Hagen, 1999; Scott et al., 2000; Auel et al., 2009). Shifts in species distribution can have major implications for polar ecosystems (Węsławski et al., 2009). For instance, several studies demonstrate the strong dependency of the Arctic planktivorous little auk (Alle alle) on the larger Arctic Calanus species (Kwaśniewski et al., 2010). The advection of different water masses can rapidly change the composition of zooplankton communities in Arctic fjords (Willis et al., 2006). When Atlantic water carrying C. finmarchicus intrudes, little auks accept longer foraging trips in order to find Arctic water masses with their preferred prey (Karnovsky et al., 2010; Kwaśniewski et al., 2010).

Figure 7 summarizes the different effects of the submesoscale filament on zooplankton dynamics, integrating underlying oceanographic mechanisms with biological implications: (A) convergence and associated accumulation of epipelagic zooplankton at the filament center; (B) distinct water masses in close proximity leading to changes in zooplankton community structure over short distances; (C) emergence of mesopelagic species at the filament center coinciding with the doming of isopycnals below 100 m depth; (D) along-frontal jets as high-speed transport vectors for certain zooplankton species.

Figure 7. Schematic illustration of the effects of a submesoscale filament on zooplankton dynamics in the Arctic marginal ice zone. Colors of copepods represent different distribution patterns, i.e., red: convergence of surface species at the filament center (Group A), blue: concentration of Arctic species in Polar Surface Water at the outer stations (Group B), black: emergence of mesopelagic species by isopycnal doming/upwelling (Group C), orange: species in along-frontal jets (Group D).

Data Availability Statement

The original contributions generated for this study are included in the article, further inquiries can be directed to the corresponding author.

Author Contributions

HA, WJvA, BN, and NH conducted the fieldwork. PK analyzed the samples, conducted the data analyses, and drafted the manuscript. PK, WH, WJvA, BN, NH, and HA significantly contributed to improving the final manuscript. All authors approved the submitted version.

Funding

Ship time was provided under grant AWI_PS107_10. NH was supported by the Russian–German Research Cooperation QUARCCS funded by the German Ministry for Education and Research (BMBF) under grant 03F0777A.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

We would like to thank the captain and crew of RV Polarstern during PS107 for their skillful support during the cruise.

References