1 Introduction

In the Arctic Ocean, high seasonal variation in sunlight and ice conditions regulates primary production both within and under the sea ice in the upper water column (Fortier et al. 2002; Leu et al. 2015). Sea ice algae growth starts at the bottom of the ice when incident solar radiation is sufficient, and ends with their release into the water column during snow and ice melt (Cota 1991; Juul-Pedersen et al. 2008; Campbell et al. 2015). Concurrently, the increased irradiance and the haline stratification resulting from snow and sea ice melt rapidly trigger the onset of a pelagic algal bloom, which may also take place under sea ice cover (Legendre et al. 1981; Strass and Nöthig 1996; Mundy et al. 2009; Arrigo et al. 2012). Sea ice and pelagic algae subsequently either transfer carbon to pelagic grazers or are rapidly exported toward the seafloor (Tremblay et al. 1989; Michel et al. 1996; Nadaï et al. 2021).

Sea ice algae generally dominate algal export and contribute to a large proportion of particulate organic carbon (POC) fluxes over the deep basins (Gosselin et al. 1997; Zernova et al. 2000; Boetius et al. 2013; Lalande et al. 2019). By contrast, sea ice algae do not contribute significantly to algal and POC fluxes on Arctic shelves (Juul-Pedersen et al. 2010). In addition to algal cells, POC fluxes are usually composed of zooplankton fecal pellets, organic aggregates, and ice-rafted particulate matter (Michel et al. 1996; Fortier et al. 2002). Importantly, zooplankton are important for the regulation of POC export through grazing on algal cells and production of fast-sinking fecal pellets, simultaneously attenuating and enhancing POC fluxes (Fortier et al. 1994). Therefore, spatial and temporal variations in algal and zooplankton production lead to large variations in the composition and magnitude of POC fluxes (Michel et al. 1996; Fortier et al. 2002). In addition to natural variability, the recent anthropogenically induced decline in sea ice and the resulting increase in incident light recently observed in the Arctic Ocean (Serreze et al. 2007; Comiso 2012; Nicolaus et al. 2012) may modify the magnitude, timing, and duration of algal growth and the seasonal development and composition of the zooplankton community, with poorly known implications for the biological carbon pump.

As a relative large portion of algal production and POC export in the Arctic Ocean occurs after snowmelt and before ice melt, and therefore, before possible detection by satellites (Mundy et al. 2009; Arrigo et al. 2012; Lalande et al. 2019; Nadaï et al. 2021), in situ under-ice measurements are critical to accurately estimate the potential importance of POC export to the biological carbon pump during sea ice melt. In this context, the objective of the present study was to evaluate spatial variations in the magnitude and composition of POC fluxes at a high spatial resolution under sea ice. To attain this objective, a total of 25 short-term drifting sediment traps deployments were completed at two ice camp sites visited during August 2018 over the rarely sampled Chukchi Plateau, a wide ridge extending from the Chukchi Sea shelf between the Mendeleev Ridge and the Northwind Ridge in the Pacific Arctic region (Fig. 1). In addition to POC fluxes, under-ice fluxes of chlorophyll a (chl a), an indicator of algal production, were also measured, and zooplankton collected in the sediment traps were quantified and identified to estimate the impact of grazing on under-ice fluxes. Overall, these measurements provide insights into the biological properties influencing the spatial variability of the magnitude of under-ice production and export in a rapidly changing but relatively little studied region of the Arctic Ocean.

Fig. 1
figure 1

a Location of the two ice camps visited over the Chukchi Plateau in the Pacific Arctic region, and b satellite-derived daily sea ice concentration on 19 August 2018 retrieved at a 3.125 km resolution from the AMSR2 satellite data archive of the University of Bremen (https://seaice.uni-bremen.de/data/amsr2)

Full size image

2 Materials and Methods

2.1 Study Sites

Ice camps were set up on two drifting first-year ice floes with extensive melt ponds visited at the northernmost sites reached over the Chukchi Plateau (CP) during the ARA09B expedition on board the Korean IBRV Araon to the Pacific Arctic region in August 2018 (Fig. 1). The first ice floe (CP1) visited from August 17 to 19 consisted of a mix of flat and distorted first-year ice covered by a thin snow layer and refrozen meltponds (Veyssière et al. 2022). The second ice floe (CP2) visited from August 20 to 22 was also composed of first-year ice covered by refrozen meltponds (Veyssière et al. 2022). While a thin, melting snow cover was observed at both sites, the CP1 ice floe was overall thicker (0.70–1.25 m) than the CP2 ice floe (0.40–0.88 m; Veyssière et al. 2022). Images obtained from the under-ice deployment of a remotely operated vehicle showed sea ice algae filaments at both sites (Fig. 2).

Fig. 2
figure 2

Sea ice algae images obtained using a digital imagery system (GoPro HERO 4) mounted on a remotely-operated vehicle deployed under ice (a–d) at CP1 and (e–f) at CP2 over the Chukchi Plateau during August 2018

Full size image

2.2 CTD-Rosette Measurements

Vertical profiles of water temperature, salinity, and fluorescence-derived chlorophyll were obtained at the stations sampled immediately after the completion of each ice camp through SBE 911plus CTD casts (Sea-Bird Scientific, USA) equipped with a fluorescence sensor. Seawater samples collected at discrete depths in the upper 100 m using a rosette system holding 10 L Niskin bottles were analyzed for nutrients and concentrations of suspended chl a and POC. Seawater samples were transferred into 50 mL conical tubes stored at 4 ºC prior to nutrient analyses. Nitrate (NO3−) + nitrite (NO2−), silica (SiO2), and phosphate (PO43−) concentrations were measured using standard colorimetric methods adapted for use with a four-channel continuous auto-analyzer (QuAAtro; Seal Analytical, USA). Seawater samples for POC measurements were collected into pre-rinsed amber polyethylene bottles. Subsamples (500 mL–1 L) were then filtered onto 47 mm diameter Whatman GF/F filters pre-combusted at 550 °C for 6 h and stored at − 80 °C until laboratory analyses. Seawater subsamples (300–500 mL) were also filtered onboard through a cascade connection filtration system containing a 20 μm nylon mesh, a 2 μm Whatman Nuclepore filter, and a 0.7 μm Whatman GF/F filter to determine the chl a concentration of microphytoplankton (> 20 μm), nanophytoplankton (2–20 μm), and picophytoplankton (< 2 μm) following Sieburth et al. (1978). Each filter was extracted in 90% acetone in the dark for 12 to 24 h before fluorescence was measured using a pre-calibrated Trilogy Turner Designs fluorometer to determine chl a concentrations following Parsons et al. (1984). The relation between total chl a concentrations obtained from seawater subsamples and fluorescence-derived chl a values obtained at the same depths was used to adjust the offset in CTD-derived fluorescence measurements.

2.3 Sediment Trap Deployments

Short-term sediment traps deployed at CP1 and CP2 consisted of cylindrical tubes (2.1 or 3.5 L) with height:diameter ratios larger than 8 to reduce turbulent mixing (Knauer and Asper 1989). A total of 25 sediment traps were deployed under ice or in open water at depths ranging from 5 to 30 m at CP1, and at depths ranging from 2 to 10 m at CP2 (Fig. 1; Table 1). Each sediment trap was attached to a single line and deployed in individual ice holes made with an auger at distances ranging from 1 to 18 m from each other for periods ranging from 30 to 54.5 h (Table 1). At CP2, three sediment traps were also attached on three distinct lines anchored to the Araon approximately 5 m away from the ice floe edge and deployed in open water at 5 m depth for 36.5 h (Table 1). All sediment traps were weighted to remain vertical in the water column and filled with a solution of filtered seawater, formalin, and borate to preserve samples. Upon recovery, sediment trap samples were transferred into plastic bags and stored at 4 °C. Prior to subsampling, recognizable zooplankton were removed with forceps from samples obtained at 2, 5, and 10 m and preserved in a formalin solution. Subsamples for POC and particulate nitrogen (PN) measurements (1.2–2 L) were filtered onto 47 mm Whatman GF/F filters pre-combusted at 550 °C for 6 h and stored at 80 °C until laboratory analyses. Subsamples (300 mL) were filtered onto Whatman GF/F filters and extracted in 90% acetone for 24 h in the dark prior to fluorescence measurement using the same pre-calibrated Trilogy Turner Designs fluorometer used for CTD-derived measurements.

Table 1 Short-term sediment trap deployment information, with asterisks indicating open water sediment trap deployments
Full size table

2.4 Laboratory Analyses

Zooplankton previously removed from the sediment trap samples were enumerated and identified to the lowest taxonomic level possible using a microscope. Filters for POC and PN measurements were freeze-dried before being exposed to hydrochloric acid (HCl) fumes during 24 h to remove inorganic carbon prior to measurements on a CHN elemental analyzer (vario Macro cube, Elementar, Germany). Acetanilide was used as a standard and the precision of these measurements was ± 4% (Jung et al. 2020). Student’s t-tests were used to evaluate the difference in POC and chl a fluxes between ice camp sites.

3 Results

3.1 Water Column Properties

Low water temperature (< − 1.2°C) and low salinity (< 29) were recorded in the upper water column above a thermocline and halocline measured at 40–45 m at CP1 and at 28–38 m at CP2 (Fig. 3a and b). Peaks in fluorescence-derived chlorophyll concentrations (> 0.6 mg m−3) were also observed at the pycnocline depths at CP1 and CP2 (Fig. 3c). Nutrients were depleted or had very low concentrations above the pycnocline at both sites (Fig. 3d–f). Suspended POC concentration profiles were similar at both sites, with peak POC concentration values > 0.5 μmol L−1 observed at 47 m at CP1 and at 36 m at CP2 (Fig. 3g).

Fig. 3
figure 3

a Water temperature, b salinity, and suspended concentrations of (c) fluorescence-derived chlorophyll, d nitrate (NO3−) + nitrite (NO2−), e silica (SiO2), f phosphate (PO43−) and g particulate organic carbon (POC) obtained at the completion of two ice camps over the Chukchi Plateau during August 2018. Dotted lines in (c–f) are used to fill gaps between the discrete data

Full size image

Peaks in suspended chl a concentrations (> 0.5 μg L−1) were observed at 44 m at CP1 and at 34 m at CP2, similar to the fluorescence measurements (Fig. 4). The relative proportion of microphytoplankton (> 20 μm) in the total chl a concentrations remained below 25% and gradually decreased from the surface to the subsurface chl a maximum (SCM) depth at both sites. More than 50% of the chl a measured at the SCM depth was attributed to picophytoplankton (< 2 μm) at both sites (Fig. 4).

Fig. 4
figure 4

Suspended chlorophyll a (chl a) concentrations and relative contribution of size-fractionated chl a concentrations obtained at the completion of two ice camps over the Chukchi Plateau during August 2018. Dotted lines are used to fill gaps between the discrete data

Full size image

3.2 Export Fluxes

Under-ice POC fluxes were the lowest at 30 m at CP1 (208 mg m−2 d−1) and the highest at 5 m at CP2 (1604 mg m−2 d−1; Fig. 5a). Except for the lowest POC fluxes recorded at one of the CP2 open water sites (259 mg m−2 d−1), POC fluxes at 5 m were higher at CP2 (838–1604 mg C m−2 d−1) than at CP1 (421–780 mg C m−2 d−1; p < 0.05). By contrast, under-ice POC fluxes at 10 m were lower at CP2 (269–527 mg m−2 d−1) than at CP1 (629–772 mg m−2 d−1; p < 0.05). The C:N ratio of the sinking particles was the lowest at 30 m at CP1 (3.1) and the highest at 10 m at CP2 (15.2; Fig. 5b). At CP1, C:N ratio values displayed larger variations at 5 m (4.8–8.2) than at 10 m (5.5–6.8), while at CP2, the largest variations in C:N ratio values were recorded at 10 m (5.4–15.3). Similar to POC fluxes, chl a fluxes at 5 m were higher at CP2 (28–102 μg m−2 d−1) than at CP1 (17–40 μg m−2 d−1; p < 0.05; Fig. 5c). Contrary to POC fluxes, chl a fluxes obtained at 10 m at CP1 (19–41 μg m−2 d−1) and CP2 (20–63 μg m−2 d−1) were not statistically different.

Fig. 5
figure 5

Under-ice (a) particulate organic carbon (POC) flux, b C:N ratios of the sinking particles, and c chlorophyll a (chl a) flux obtained at two ice camps over the Chukchi Plateau during August 2018

Full size image

3.3 Zooplankton

Nearly all the zooplankton collected in the sediment traps were copepods (Fig. 6). The average number of copepods collected at CP1 at 10 m was at least 5 times higher than the average numbers of copepods collected at CP1 at 5 m and at all depths sampled at CP2. In all sediment traps, the majority of copepods collected consisted of the copepod Calanus glacialis/marshallae (Fig. 6). As juveniles (copepodite stages) and adults of the Arctic C. glacialis and the Pacific C. marshallae are difficult to differentiate, these individuals were combined and identified as C. glacialis/marshallae (Hopcroft et al. 2010; Questel et al. 2013; Ashjian et al. 2017). Among the C. glacialis/marshallae copepods, individuals of the copepodite stage CIV were dominant at both sites and at all depths sampled (Fig. 6). The other zooplankton collected at CP1 consisted of the copepods Paraeucheta glacialis and Oithona similis, and the amphipod Themisto libellula. At CP2, the other zooplankton collected at 5 and 10 m mostly consisted of the copepods Calanus hyperboreus and P. glacialis (Fig. 6).

Fig. 6
figure 6

Zooplankton and relative abundance of zooplankton collected in the sediment traps at two ice camps over the Chukchi Plateau during August 2018

Full size image

4 Discussion

4.1 Export Fluxes Associated with Sea Ice Algae Release

Sea ice properties at the CP1 and CP2 first-year ice floes displayed low salinity, low chl a concentrations, and very low nutrient concentrations, indicating that both ice floes were nutrient-limited and in advanced stages of melting and algal growth (Veyssière et al. 2022). Accordingly, the fresh and cold surface water layer, the nearly depleted nutrients, and the low suspended chl a and POC concentrations observed in the upper water column at CP1 and CP2 further reflected ongoing ice melt and indicated post-bloom conditions at both ice camps. Despite post-bloom conditions, high chl a and POC fluxes and low C:N ratios of the sinking particles collected at 5 m under both ice floes suggested the export of freshly produced material, such as sea ice algae. Although sediment trap samples were not microscopically examined to confirm and quantify the export of sea ice algae, the observation of large algal aggregates (Fig. 2) may have led to the relatively high chl a and POC fluxes recorded at both sites. Moreover, a greater abundance of sea ice algae filaments observed underneath the ice at CP2, where chl a and POC fluxes were on average twice larger than at CP1, supported sea ice algae strands as an important source of chl a and POC at that site. Such algal strands are typically composed of the exclusively sympagic centric diatom Melosira arctica that forms long brownish filaments commonly observed attached to the underside of the ice, as free‐floating filaments in the meltwater layer, or as deposited aggregates on the seafloor of the deep central basins (e.g.,Gran 1899; Booth and Horner 1997; Lee et al. 2011; Boetius et al. 2013). Whereas sea ice algae export typically begins at the onset of snowmelt during spring (Nadaï et al. 2021), a delay in the export of M. arctica has previously been observed over the Lomonosov Ridge, East Siberian Sea, and over the deep Nansen and Amundsen basins (Lalande et al. 2019). This delay in export may result from the formation of gas bubbles produced by photosynthetic oxygen production within the mucous matrix, regulating the buoyancy of the M. arctica aggregates even at low ambient nutrient concentrations (Fernández-Méndez et al. 2014). The distinct properties of this widespread sea ice algae make it a likely source of chl a and POC later in the productive season. Therefore, we hypothesize that the patchy under-ice distribution of the apparent M. arctica aggregates led to the spatial variations in chl a and POC fluxes observed at CP1 and CP2. Furthermore, the lower chl a and POC flux values observed in open water than under-ice at CP2 suggest a reduced collection of M. arctica due to a lower density of the sea ice algae in open water compared to under ice.

Small algal cells (nano-sized chl a; 2–20 µm) dominated the low suspended chl a concentrations at both ice camp sites during summer, similar to previous observations in the Pacific Arctic region and in the Central Arctic Ocean (Booth and Horner 1997; Gosselin et al. 1997; Lee et al. 2019). However, small algal cells, such as flagellates, dinoflagellates, and silicoflagellates, typically only contribute to a low fraction of the algal fluxes. Instead, diatoms (micro-sized chl a; > 20 µm) consistently dominate algal flux in the Arctic Ocean due to their high sinking rates associated with their heavy silicate frustules and formation of aggregates (Lalande et al. 2019; Dezutter et al. 2021; Nadaï et al. 2021), supporting a similar dominance of diatoms, presumably M. arctica, in chl a fluxes at CP1 and CP2.

4.2 Export Fluxes Associated with Particulate Matter Release

Alternatively, the higher POC fluxes observed at 5 m under a thinner sea ice cover at CP2 may also reflect an enhanced release of particulate matter from the melting ice at that site. Indeed, a substantial amount of resuspended particulate matter is commonly incorporated into sea ice during the freezing period over the shallow shelves and transported over the basins (Eicken et al. 2000; Eicken 2004; Wegner et al. 2005; Lalande et al. 2014). Exceptionally high concentrations of particulate matter have previously been observed into first-year sea ice over the broad Siberian shelves (Eicken et al. 2000), and high sediment loads may also be widespread in the Chukchi Sea (Eicken 2004). As sediment-laden ice displays a patchy distribution as it drifts toward the basin (Eicken et al. 2000; Eicken 2004), the higher under-ice POC fluxes observed at CP2 may reflect a larger release of POC during sea ice melt at that site. Although the contribution of the released POC to the under-ice POC fluxes has not been quantified, it should not be ignored as a potentially significant source of POC in this region.

4.3 Impact of Grazing Pressure on Export Fluxes

Compared to fluxes at 5 m, averaged chl a and POC fluxes obtained at 10 m were slightly higher at CP1 and lower at CP2. This discrepancy between both sites may be due to the varying grazing pressure exerted by the copepod C. glacialis/marshallae, dominant at both sites but more abundant at CP1 at 10 m. Indeed, higher POC fluxes at CP1 at 10 m than at 5 m may indicate an enhanced export of fecal pellets due to a larger grazing pressure resulting from the much larger presence of C. glacialis/marshallae at that site and depth. Conversely, lower POC fluxes at 10 m than at 5 m at CP2 possibly reflected reduced fecal pellet carbon export due to a lower grazing pressure at that site. In the Arctic Ocean, C. glacialis feeds on ice algae to reproduce and grow (Daase et al. 2013). Specifically, C. glacialis adult females (AF) migrate toward the surface before or at the onset of sea ice algae production to feed underneath the ice (Søreide et al. 2010). Several weeks later, C. glacialis copepodite stage CIV dominates and forms the overwintering population (Søreide et al. 2010; Daase et al. 2013). Interestingly, C. glacialis can rapidly develop from CIV to AF under favorable conditions, even potentially spawn late in the season (Wold et al. 2011). The larger relative proportion of C. glacialis/marshallae AF in the copepods collected late in the season at CP2 compared to CP1 may therefore reflect sustained food supply, potentially provided by the abundant sea ice algae observed at CP2.

Although zooplankton collected in the sediment traps do not reflect their abundances in the water column, they may reflect the composition of the zooplankton community at each trap depth. In this context, the nearly exclusive presence of C. hyperboreus copepodite stages CIII and older at CP2 suggests a greater exchange with the adjacent deep basin, as C. hyperboreus is the most abundant copepod residing in the Arctic basin (Campbell et al. 2009). Despite its low numbers, the presence of this large copepod at CP2 may have contributed to the attenuation and/or to the increase of the POC flux through grazing and the production of fecal pellets. Also, the amphipod T. libellula, a species associated with polar water masses that forms swarms in the upper water column (Kraft et al. 2011), was exclusively present at 5 m at CP1 and likely contributed to the enhanced POC fluxes through the production of fecal pellets. Overall, the unexplained large number of copepods collected at 10 m and the exclusive presence of T. libellula at 5 m highlight a high spatial heterogeneity in the distribution of copepods and amphipods in the region, likely contributing to the high spatial variability in export fluxes.

4.4 High Spatial Variability in Export Fluxes

Results obtained under the drifting ice floes CP1 and CP2 over the Chukchi Plateau showed high spatial variations in the magnitude of chl a and POC fluxes within small areas. The high flux variability may reflect the spatial heterogeneity in the distribution of the M. arctica aggregates, potentially influencing the spatial distribution of copepods and amphipods feeding on these aggregates. The heterogeneous grazing pressure exerted by zooplankton is likely to have further contributed to the spatial variability in under-ice chl a and POC fluxes at both CP1 and CP2.

Previous under-ice export fluxes measured using drifting sediment traps across the Arctic Ocean were mostly obtained before or at the onset of ice melt and/or below the euphotic zone. North of Svalbard, the large spatial variability in the magnitude of under-ice export fluxes observed over a relatively small area during spring was attributed to variations in the composition of the phytoplankton assemblages, the grazing pressure from large grazers, the distance to the open water, and the advection of Atlantic water (Dybwad et al. 2021). In the Canadian Arctic, chl a fluxes obtained between 0.5 and 25 m under ice consistently reached values > 1 mg m−2 d−1 following the release of sea ice algae during June (Michel et al. 1996; Fortier et al. 2002; Juul-Pedersen et al. 2008). Whereas chl a fluxes observed early in the productive season in the Canadian Arctic (> 1 mg m−2 d−1) were much larger than chl a fluxes observed during summer over the Chukchi Plateau (< 0.1 mg m−2 d−1), POC fluxes at these sites were typically lower (< 400 mg m−2 d−1) than those obtained at CP1 and CP2 (200–1600 mg m−2 d−1). By contrast, under-ice chl a fluxes measured at 2 and 5 m in the Central Arctic Ocean during late summer largely remained < 0.1 mg m−2 d−1 (Lalande et al. 2014), similar to the chl a fluxes at CP1 and CP2. However, the POC fluxes associated to these chl a fluxes were also lower in the Central Arctic Ocean than in the present study over the Chukchi Plateau (< 400 mg m−2 d−1; Lalande et al. 2014), emphasizing the remarkably high POC fluxes measured at CP1 and CP2 in August 2018.

5 Conclusions and Implications

Under-ice export flux measurements obtained at two ice camp sites over the Chukchi Plateau in August 2018 showed wide variations in chl a and POC fluxes. This high spatial variability in the magnitude of under-ice export fluxes was apparently due to the heterogeneous distribution of sea ice algae aggregates, possibly associated with variation in snow and ice cover thickness, the only distinctive physical feature between both ice camp sites. Interestingly, variations in the magnitude of under-ice fluxes were larger within the few meters sampled under a drifting ice floe than over three years at a single sample site influenced by different meteorological events and zooplankton assemblages (Fortier et al. 2002). This high spatial variability reflects the considerable patchiness of the Arctic marine ecosystem and should be considered when reporting on export fluxes. Finally, these results highlighted the potential importance of sea ice algae aggregates as a source of carbon for pelagic consumers during summer, beyond the period of maximal algal export commonly observed at the onset of snow and sea ice melt. This implies that the rapid decline in sea ice may significantly reduce this source of carbon and affect the biological carbon pump, emphasizing the necessity of monitoring carbon fluxes at a high spatial and temporal resolution in the Pacific Arctic region.