Processes controlling aggregate formation and distribution during the Arctic phytoplankton spring bloom in Baffin Bay

The thickness and areal extent of sea ice in the Arctic Ocean have changed dramatically in recent decades due to global warming (Comiso, 2002; Stroeve et al., 2007; Comiso et al., 2008; Kwok and Rothrock, 2009; Screen and Simmonds, 2010; Vihma, 2014). The consequences for marine ecosystems remain difficult to predict. Furthermore, in this context, the Arctic Ocean exerts a feedback on global climate (Masson-Delmotte et al., 2018). Sea ice is an essential component of global biogeochemical cycles in polar ecosystems and supports polar marine food webs (Vancoppenolle et al., 2013). Ice-algal and phytoplankton blooms begin when light intensity and nutrient concentrations are sufficient (e.g., Tremblay et al., 2006; Assmy et al., 2017). The resulting biomass is controlled by additional factors such as coupling with grazer life cycles (Søreide et al., 2010; Leu et al., 2011) and ocean surface temperature (Feng et al., 2016). Phytoplankton blooms typically occur after ice melt when the water column is stratified (Søreide et al., 2010; Randelhoff et al., 2019). The succession of ice-algal and phytoplankton blooms, within and beneath the sea ice, determines the reproduction and growth of grazers (Søreide et al., 2010: Leiknes et al., 2016). Ice algae are grazed by small planktonic crustaceans (e.g., krill and copepods), which in turn are preyed upon by higher trophic levels such as fish, birds, and mammals (Nielsen and Hansen, 1995; Hansen et al., 1996; Darnis et al., 2008; Darnis et al., 2012; Darnis et al., 2017). Unconsumed organic matter produced within and beneath the ice descends into the water column as senescent algae, aggregates, appendicular houses, or fecal pellets (Forest et al., 2018). Depending on the depth reached (Kellogg et al., 2011), this organic matter can either be consumed by benthic feeders, buried in the sediment, or sequestered over long periods of time in deep water (Leu et al., 2011; Wassmann et al., 2011; Forest et al., 2018).

With the decline of sea ice in recent decades, a decrease in the total amount of organic matter produced in the ice is expected, due to the loss of habitat for sympagic algae (primary producers associated with the ice), with a possible collapse of the entire food web and associated biogeochemical processes underneath the sea ice (Arrigo and Dijken, 2011; Arrigo, 2014; Babin, 2020). Early spring sea ice breakup will affect the timing of summer phytoplankton blooms and may result in a lag between ice algae and phytoplankton blooms, which may have a significant impact on copepod reproductive success and subsequent biomass (Søreide et al., 2008; Søreide et al., 2010; Søreide et al., 2013; Janout et al., 2016) and affect the entire pelagic food web after sea ice retreat (Tedesco et al., 2019).

Baffin Bay, located at the gateway to the Arctic Ocean, is a heterogeneous environment with two different currents flowing in opposite directions: a warm, salty current from the southeast and a cold current from the northern Arctic that flows along the west coast of Baffin Bay (Tang et al., 2004). The circulation affects the stratification regime and the biogeochemical environment of Baffin Bay, which controls phytoplankton community production and structure and, in the end, carbon transfer through the planktonic food web (Saint-Béat et al., 2020). The bay is completely ice-covered from December to March. During the Arctic spring, due to ice retreat, light passes through the water column and nutrient-rich waters flow into the bay. Nutrient-rich Atlantic waters from the Greenland Current (Torres-Valdés et al., 2013), nutrient-rich waters from the Canadian Archipelago (Coachman and Aagaard, 1974; Newton et al., 1974), and nutrient flux from depth, through remineralization and its vertical input due to mixing, provide nutrients to Baffin Bay (Tremblay et al., 2002; 2006). The combination of light, nutrients, and increased stratification creates a stable environment for phytoplankton growth in the water column around May (Massicotte et al., 2018; Massicotte et al., 2019; Oziel et al., 2019; Randelhoff et al., 2019). Diatoms and Phaeocystis spp. are the dominant components of the phytoplankton assemblage in the turbulent, nutrient-rich waters in spring (Smith et al., 1991; Rousseaux and Gregg, 2013). The contribution of Phaeocystis spp. to spring blooms may increase with global change, as nutrient limitations and inflow from Atlantic waters may increase (Marchant and Thomsen, 1994; Gabric et al., 2003; Schoemann et al., 2005). Isolated phytoplankton cells are generally remineralized by microbial activity or consumed by zooplankton at the ocean surface. Diatoms and Phaeocystis spp. can also aggregate into large particles that sink at the end of the phytoplankton bloom (Smith et al., 1991; Riebesell, 1993; Riebesell et al., 1995; Mari et al., 2005; Cavan et al., 2019; Henson et al., 2019; Prairie et al., 2019). Due to their rapid sedimentation rates, these aggregates typically drive vertical export (Moriceau et al., 2007; Turner, 2015). The succession of ice-algal and pelagic blooms and their fate through higher trophic levels and vertical transfer remain poorly understood.

To understand and predict carbon export in the Arctic Ocean, we need to understand the parameters that trigger aggregation in the Arctic spring phytoplankton bloom. The Green Edge project was developed to (1) study the timing and development of the Arctic spring phytoplankton bloom, (2) understand the fate of the organic matter produced, and (3) improve our understanding of the polar ecosystem in a changing climate. This study, focusing on the fate of the Arctic spring bloom, benefited from the large data set collected during the 2016 Green Edge expedition in Baffin Bay. In an effort to highlight the role of phytoplankton bloom dynamics and environmental biochemical conditions in macroaggregate (marine snow) formation and distribution, this large data set was linked to particle and zooplankton abundance profiles collected with an Underwater Video Profiler 5 (UVP5).

The Green Edge expedition onboard the icebreaker CCGS Amundsen was conducted from June 9 to July 10, 2016, in Baffin Bay. A total of 7 transects with 137 stations crossing an area from 55°W to 64°W and from 67°N to 71°N were visited (Figure 1).

Sea ice concentration (SIC) is an index that ranges from 0 (for complete open water) to 1 (for full ice cover). SIC data were extracted from AMSR2 radiometer daily data on a 3.125-km grid (Beitsch et al., 2014). At each station, the SIC was extracted from the closest pixel using the Euclidean distance between the ship position and the center of each pixel. In this study, a site was considered free of ice when the SIC was less than 50% (where 0% is ice-free open water, and 100% is totally ice-covered). For each sampling station and at each sampling moment, the number of open water days (OWD) was computed. OWD corresponds to the number of days since the site became ice-free (positive values) or before the ice melt (negative values).

For each station, the depth of the euphotic zone, Ze, was defined as the depth where a daily irradiance of photosynthetically available radiation reached 0.415 mol photons m^–2 d^–1 (Lafond et al., 2019; Massicotte et al., 2019; Randelhoff et al., 2019).

In Baffin Bay during the Green Edge 2016 expedition, the depth of the EML was shallow (17 to 30 m) depending on freshwater inputs due to ice melting (Randelhoff et al., 2019). This variation was neglected in our analysis, and surface data were averaged over a fixed EML of 30 m, the maximum mixing layer depth in Randelhoff et al. (2019).

Seawater was collected, using a rosette of Niskin bottles, at 7–14 different depths. At each hydrographic station, vertical profiles of water column properties were made using a conductivity–temperature–depth (CTD) sensor system (Seabird SBE-911 plus), installed onboard the CCGS Amundsen, which also included a Seapoint SCF Fluorometer for detection of chlorophyll a, and a WetLabs C-Star transmissometer to measure beam attenuation as an indicator of biomass. Water samples were collected as the Niskin bottles were brought up to depths determined using data collected by the CTD upon descent.

Nutrient (NO₃^–, HPO₄^3–, and H₄SiO₄) concentrations were measured at 44 stations covering the 7 transects (Table S1). Samples were taken from surface to depth with bathymetry ranging from 350 to 2000 m. Analyses were performed onboard on a Bran + Luebbe Autoanalyzer 3 using standard colorimetric methods (Grasshoff et al., 2009). Analytical detection limits were 0.03 µM for NO₃^–, 0.05 µM for PO₄^3–, and 0.1 µM for H₄SiO₄.

Biogenic silica (bSi) concentrations were measured at 28 stations covering the 7 transects (Table S1). Samples were taken at depths between 0 and 350 m with one deep profile going to a depth of 1,650 m. For each sample, 1 L of seawater was filtered onto 0.4 µm polycarbonate filters (Nucleopore Whatman). Filters were then stored in Eppendorf vials and dried for 24 h at 60°C before being stored at room temperature. bSi was measured by the hot NaOH digestion method of Paasche (1973) modified by Nelson et al. (1989), with an analytical detection limit of 0.005 μM.

Identification and quantitative assessment of phytoplankton communities were carried out on 43 samples from the surface (depth of 0–2 m) and the subsurface chlorophyll a maximum (SCM, depth of 15–44 m) at 24 stations (Table S1). The SCM was identified using a Seapoint SCF Fluorometer for detection of chlorophyll a. Aliquots of 25–50 mL were preserved with 0.8 mL Lugol iodine solution in dark glass bottles and then stored in the dark at 4°C until analysis. For each sample, microscopic counts and identifications were made in the laboratory following the Utermöhl settling method (1958). Counts were performed by inverted microscopy (Nikon Eclipse TS100; Utermöhl, 1958). A minimum of 400 phytoplankton cells per sample were counted at 100 magnification (see Lafond et al., 2019, for further details). As they constitute the major components of the Arctic ecosystem, diatoms (Baccilariophyceae) and Phaeocystis spp. (Prymnesiophyceae) were studied as the priority. Diatom carbon biomass (µg C L^–1) was estimated following the method provided by Lafond et al. (2019; their table S3). Concerning Phaeocystis spp., cell abundance (Green Edge Dataset, 2021) was converted to biomass (µg C L^–1) assuming a carbon cell content of 14.2 pg C cell^–1 (Rousseaux et al., 1990).

For pigments, 521 samples were analyzed from 52 stations covering 7 transects (Table S1). About 2.7-L seawater samples were filtered onto 0.7-µm pore size Whatman GF/F filters and stored in liquid nitrogen prior to analysis. Then, filters were extracted in 100% methanol, disrupted by sonication, and clarified by filtration (GF/F Whatman). Samples were analyzed the same day using high-performance liquid chromatography with an Agilent Technologies HPLC 1200 system with diode array and fluorescence detection (Ras et al., 2008).

The vertical distribution of large particulate matter (>600 µm to 2 cm) was studied using a nondestructive imaging system, the UVP5 (Picheral et al., 2010). The UVP5 was mounted on the CTD-rosette frame and positioned at the lowest location in order to profile minimally disturbed water during descent. The UVP5 deployments were undertaken on 137 hydrologic stations, from the surface to the bottom depth, and a total of 196 vertical profiles covering 7 transects were performed during the cruise (Table S1).

Images recorded by the UVP5 were processed with custom-developed image analysis software and extracted using the ZooProcess^® software (Gorsky et al., 2010). Using the ECOTAXA 2.0 software, for all images, the major and minor axes of each object were measured and converted into biovolume (mm³) assuming that the particle-projected shape was ellipsoid. A computer-assisted method was used to classify all organisms in groups of vignettes using the Random Forest algorithm. Machine learning classification was subsequently verified manually. In this study, the concentration of marine snow aggregates and distribution of copepods were analyzed along with the 196 profiles at a 5-m depth resolution.

Marine snow aggregates were identified and characterized by irregular shape, their grainy appearance, and their heterogeneity of shape. In this study, long stick-shaped particles were disregarded as more representative of fecal pellets. The aggregate equivalent spherical diameter was calculated from the biovolume assuming that aggregates are prolate spheroid in shape (Hillebrand et al., 1999).

Copepod images (copepod vignettes including copepod-like subvignettes) were validated by taxonomic expertise considering that the prosome shape (lengthy ovoid) is about twice as long as the urosome and that the presence of 2 long and thin antennae that at least represented twice the prosome length could be affiliated to copepod vignette. For the UVP images of copepods ranging from 0.7 to 7 mm, the resolution was not adequate to distinguish copepod species or developmental stages. In Arctic environments, zooplankton communities are dominated by only a few copepod species such as Calanus glacialis and Calanus finmarchicus, Calanus hyperboreus, Metridia longa, and Pseudocalanus spp. adults (Vilgrain et al., 2021).

To describe and relate biogeochemical parameters to aggregate abundance, we performed a PCA (Legendre and Legendre, 2012). We included the following parameters, all of which were available concurrently for 28 stations: the biogeochemical parameters of EML nitrate, phosphate, and silicate concentrations; the biological parameters of EML copepod abundance and bSi, chlorophyll a, and phaeopigment concentrations, surface and SCM biomass of diatoms and Phaeocystis spp., and maximum depth of copepod abundance; and the aggregate distributional characteristics of EML aggregate abundance, depth-integrated (30–200 m) aggregate abundance, and depth of maximum aggregate abundance. The 28 stations were clustered by hierarchical classification on the coordinates for the first 3 principal components, using Euclidean distance.

The aggregate equivalent spherical diameter ranged from 1 to 5 mm (mean ± standard error of 1.3 ± 0.2 mm, n = 25,649) with corresponding biovolume ranging between 0.60 and 78.9 mm³ (mean of 1.04 ± 0.01 mm³). No variation in the aggregate size spectra among transects was detected, nor did the average aggregate size change with depth or with the progressive aging of the bloom from the ice edge to the open water stations. When plotted against OWD, the profiles of aggregate abundance illustrate the progressive deepening of the aggregates (Figure 2). Aggregate abundances were at a maximum at the surface when sea ice was still present (OWD < –10) and between 13 and 30 m for OWD ranging from –10 to 11. As OWD increased further, several profiles showed high aggregate abundances distributed between depths of 50 and 200 m with large abundances just below the maximum mixed layer depth. Higher abundances (>5,000 aggregates m^–3) were recorded below this depth after the ice melt (Figure 2).

Average nitrate, silicate, and phosphate concentrations in the EML (first 30 m) were higher in stations under the ice (OWD < 0). Nitrate concentrations varied between 2 and 6 µM, silicate concentrations between 2 and 10 µM, and phosphate concentrations between 0.2 and 0.8 µM. After ice melting, represented by OWD from 0 to 30, nitrate concentrations rapidly approached 0 µM in 2–10 days depending on the station, while concentrations of silicate dropped to 1 µM and phosphate to 0.2 µM (Table S1; Figure 3). After 10 days free of ice, most stations were totally depleted of nitrate; after 20 ice-free days, most stations were totally depleted of silicate. Figure 3 shows the progressive nutrient depletion from stations under the sea ice (OWD < 0) to open water stations (OWD > 0), with an intensification of the depletion around the ice edge, confirming that nutrient uptake by the phytoplankton community strongly depleted nutrient stocks in surface waters. Aggregate abundance in the EML was highest around OWD = 0. The highest aggregate abundances were recorded concurrently at stations where silicate concentrations were lower than 2 µM. At these stations, nitrate and phosphate concentrations were more variable, ranging between 0 and 5 µM for nitrate and 0.1 and 0.4 µM for phosphate. After more than 5 days free of ice, aggregate maxima appeared below the mixed layer. This evidence of aggregate sinking can explain why the aggregate abundance at the surface around the ice edge at these stations was as low as it was prior to detecting aggregate formation.

Phaeocystis spp. and diatoms constituted the bulk of phytoplankton stocks in both abundance and biomass (Table 1) at the two sampled depths: the surface (0–5 m) and the SCM. Concentrations of Phaeocystis spp. ranged from 2.8 × 10³ to 11.5 × 10⁶ cells L^–1 (with biomass ranging from 0.04 to 163 µg C L^–1), whereas diatom concentrations ranged from 1 × 10³ to 9 × 10⁵ cell L^–1 (with biomass ranging from 0.1 to 210 µg C L^–1). Cell abundance in the surface was higher for Phaeocystis spp. than for diatoms, but due to their small size, Phaeocystis spp. represented lower biomass (18 ± 21 µg C L^–1, n = 36, vs. 30 ± 36 µg C L^–1, n = 44, for diatoms). Cell concentrations of both diatoms and Phaeocystis spp. increased simultaneously and reached maxima around OWD = 0, when the ice was melting. Microscopy revealed that diatoms were the main silicifiers (Lafond et al., 2019), validating the use of bSi as a proxy for diatom abundance in this study.

High bSi concentrations and aggregate abundances were observed concurrently in the surface along transects T1, T3, and T5 (Figures 4 and 5). Along transect T3, a deep bSi concentration maximum (close to 1 µM) was observed below 100 m from West to East (Figure 4). At the same time, the maximum aggregate abundance was measured at 32.5-m depth at station G310 (7.2 × 10³ aggregates m^–3). Close to the eastern part of the transect, deep aggregate concentrations were observed at 102.5 m at station G3000 (11.3 × 10³ aggregates m^–3), and bSi concentrations were >0.5 µM (G3000 on T3; Figure 4). Within transect T1, we observed high aggregate abundance and low bSi concentration (<1 µM). Abundances higher than 1 × 10³ aggregates m^–3 were measured below a depth of 50 m and deepened from West to East (T1; Figure 4). While at some stations aggregate abundance may be associated with Phaeocystis spp. concentration, the highest aggregate abundances co-occurred with the highest bSi concentrations (up to 1.5 µM, at stations G204, G207, G309, G507, and G215; Figure 5). This trend was particularly evident during ice melting when the aggregates formed, with the exception of station G300 where bSi concentration and aggregate abundance remained high (0.66 µM and >2 × 10³ aggregates m^–3) 26 days after ice melting (Figure 5).

In Figure 5, the 7 stations with elevated aggregate abundances (G204, G207, G300, G309, G507, G512, and G707) plot above the other stations. Five of them were located at the ice edge (except G204 and G707), and all of them were located at sites with more than 0.6 µM of bSi. At these 7 stations, Phaeocystis spp. concentrations ranged from 8 to 74 × 10⁵ cell L^–1, and diatom concentrations varied between 2 and 6 × 10⁵ cell L^–1, except at station G300 where the diatom concentration was only 3.7 × 10⁴ cell L^–1 at the maximum aggregate concentration (at 55-m depth). At these high-aggregate stations, diatom communities were dominated by centric species (60%–92%) and specifically by Chaetoceros spp. and numerous Thalassiosira spp. Station G512 differed somewhat as pennate and centric species dominated the community equally. At this station, while Chaetoceros spp. were well represented, ice algal genera such as Melosira, Cylindrotheca, and Entomoneis were more abundant. Interestingly, at stations at the ice edge where aggregate abundance remained low at the surface (e.g., stations G605, G713, and G719), the concentrations of both Phaeocystis spp. and diatoms were one order of magnitude lower and the diatom community consisted mainly of pennate diatoms (87%–98%).

Mean chlorophyll a concentration in the EML ranged from 0.06 to 3.28 µg L^–1 across all stations, while mean EML phaeopigment concentration ranged from 0 to 0.1 µg L^–1 (Table S2; Green Edge Database, 2021). Locations of the surface chlorophyll a maximum generally coincided with the surface bSi concentration maximum. We observed a gradient in phaeopigment concentrations from West to East, with maximum values at stations G3000 and G300 within the EML. Globally we observed a positive gradient of phaeopigments from ice-covered stations (negative OWD) to open water stations (positive OWD; Figure 6).

Under the ice (OWD < 0), the highest copepod abundances (>500 ind m^–) were observed above the EML, with maximum values deepening progressively as OWD increased (Figure 7).

Results of the PCA (Figure 8) indicated that the stations and parameters examined fell into three groups. The first group, comprised of pre-phytoplankton bloom stations, was characterized by high nutrient concentrations, SIC close to 1, low bSi and chlorophyll a concentrations, and low aggregate abundance. The second group of phytoplankton bloom stations was characterized by high bSi and chlorophyll a concentrations, with both Phaeocystis spp. and diatoms contributing to the phytoplankton community in the EML. The third group, called the post-phytoplankton bloom stations, was where we observed high phaeopigment concentrations in the EML and high OWD.

Aggregation is an important process driving both vertical carbon export, through the formation of rapidly sinking particles (Turner, 2015), and carbon transfer to higher trophic levels (Leu et al., 2011; Wassmann et al., 2011; Leu et al., 2015; Forest et al., 2018). As such, aggregation represents the termination and the fate of phytoplankton blooms. The aggregation process and following massive sinking may be occasional and very fast (Alldredge and Gotschalk, 1989; Kiørboe et al., 1996; Prairie et al., 2019), and thus hard to observe and to predict. During the Green Edge campaign, we took advantage of the huge amount of data collected to try to identify what triggers aggregation of the spring phytoplankton bloom.

As not all phytoplankton species have the same ability to aggregate, we investigated the role played by the phytoplankton assemblage. According to the classical phytoplankton succession, Phaeocystis spp. blooms often follow diatom blooms, thriving when silicate concentrations drop below 2 µM (Smith et al., 1991; Smith, 1994; Smetacek, 1999; Wassmann et al., 1999; Reigstad and Wassmann, 2007). Our results, however, demonstrated the co-occurrence of Phaeocystis spp. and diatoms throughout the sampling period. The occurrence of Phaeocystis spp. blooms is expected to increase with climate change (Marchant and Thomsen, 1994; Gabric et al., 2003; Schoemann et al., 2005), and the co-occurrence of both diatoms and Phaeocystis spp. is becoming increasingly common (Dybwad et al., 2021). Moreover, in the eastern central Arctic Basin, Phaeocystis spp. grow at high silicate concentrations (>2 µM; Smetacek, 1999; Wassmann et al., 1999; Gieskes et al., 2007; Degerlund and Eilertsen, 2010). Both diatoms and Phaeocystis spp. are generally associated with strong aggregation events (Riebesell, 1993; Wassmann et al., 1999). The distribution patterns of diatoms and Phaeocystis spp. in our study globally matched the aggregate distribution observed along the sampling transects; however, the analysis in Figure 5 revealed that the maximum aggregate concentrations correspond to maximum bSi concentrations, which are attributed to diatom biomass. More specifically, our results revealed maximum aggregate abundance at the surface when centric diatoms, mainly Chaetoceros spp., dominated the diatom community compared to stations dominated by pennate diatoms.

Many studies have shown that aggregation occurs more often during the late stage of the phytoplankton bloom, as stickiness generally increases when senescent phytoplankton produce sticky organic compounds such as extracellular polysaccharide substances (EPS), Coomassie stainable particles, and transparent exopolymeric particles (TEP; Passow et al., 1994; Waite et al., 1995; Passow, 2000, 2002; Cisternas-Novoa et al., 2015; Cisternas-Novoa et al., 2019a; Cisternas-Novoa et al., 2019b).

In the Arctic, development of the phytoplankton bloom is strongly controlled by the ice cover and its melting. The senescence of phytoplankton blooms usually starts as nutrients become limiting and may be revealed by a change in pigment composition (Henriksen et al., 2002; Lewis et al., 2019). Nutrient limitation as discussed by Moore et al. (2013) may take the form of (1) a phytoplankton biomass threshold due to growth rate being limited by low nutrient concentrations (Blackman and Liebing theory); (2) physiological stress and deficiency of one element, for example, when the N:P or Si:N stoichiometric ratio is unbalanced and nitrogen becomes limiting, although the relationship between growth rate and stress due to nitrogen limitation can vary, and “the most deficient nutrient will not become limiting if all remain replete” (Moore et al., 2013); or (3) co-limitation, when two or more elements have been drawn down simultaneously to below the physiological level that allows for phytoplankton growth. Regarding values of half-saturation constants for nutrient uptake, the global average K_s for diatoms has been reported as 3.9 ± 5.0 µM (n = 25) for silicon, 1.6 ± 1.9 µM (n = 35) for nitrogen, and 0.24 ± 0.29 µM (n = 14) for phosphorus (Sarthou et al., 2005). For Arctic planktonic ecosystems, K_s is lower than 2 µM for silicate (1.15 µM from model parameters used in Zhang et al., 2010; 2 µM from in situ measurements in Krause et al., 2018) or even lower (e.g., 0.5 µM for nitrate and silicate for the Barents Sea; Slagstad and Støle-Hansen, 1991). From the K_s perspective, silicate concentrations under 2 µM and/or nitrate concentrations under 1.5 µM may be expected to limit diatom growth. Assuming these K_s values pertained to most of our stations, nitrate started to be limiting 2–10 days after ice melting, and silicate, 2 days before the ice melt (Figure 3). The stations with the highest aggregate abundance in the EML all had nitrate concentrations higher than the assumed K_s for nitrate (about 0.5–4.5 µM), but all had silicate concentrations around the value of the K_s measured for silicate in the Arctic by Krause et al. (2018; about 2 µM, Figure 3). Both nitrogen and silicon limitations could trigger aggregate formation in the EML, as they increase the mucus production that enhances stickiness (e.g., EPS, TEP; Passow et al., 1994; Waite et al., 1995; Passow, 2000, 2002; Cisternas-Novoa et al., 2015; Cisternas-Novoa et al., 2019a; Cisternas-Novoa et al., 2019b). As in Aumack and Juhl (2015), silicon limitation as defined by K_s values better explains high aggregate abundance in the EML than nitrate limitation in our study.

A senescing phytoplankton bloom may be characterized by high phaeopigment concentrations (Tremblay et al., 1989; Riebesell, 1991; Barlow et al., 1993). However, in our data set, chlorophyll a was not degraded and phaeopigments were not concentrated at the ice edge. Hence, aggregates occurred when phytoplankton were still viable. Considering K_s values, aggregates appear to have formed when diatoms became limited by silicate, suggesting that nutrient limitation plays a major role in aggregation (Kaltenbock and Herndl, 1992; Magaletti et al., 2004; Beauvais et al., 2006; Aumack and Juhl, 2015). Other signals of stress were reported by Amiraux et al. (2017) during this same field campaign, including evidence for salinity stress of the bacterial communities in sea ice before it had completely melted and photo-oxidative stress of pelagic algae after the ice had melted. These authors suggested that EPS might protect the algae and their associated bacterial communities against abiotic stresses, as EPS production may be enhanced at these points.

In the PCA, which summarizes both the chemical and biological variables implicated in causing differences between stations (Figure 8), low phytoplankton growth (pre-phytoplankton bloom stations) was not associated with cell aggregation. As presented in Figure 3, the decrease in all nutrient concentrations in the EML (nitrate, silicate, and phosphate) appeared to correlate with the increase in OWD, which corresponds to the progressive utilization of nutrients as the phytoplankton bloom developed. At the phytoplankton bloom stations, the PCA shows a strong covariance between bSi concentration and aggregate abundance in the EML (red arrows in Figure 8), suggesting that nutrient-limited diatoms may be more efficient at aggregating in this ecosystem than Phaeocystis spp. even if their respective biomasses were not discriminated in the PCA (Figure 8). At the postbloom stations, high values of integrated aggregate abundance and depth of maximum aggregate abundance are congruent with phaeopigment concentration in the EML and depth of the maximum copepod abundance. Phaeopigment concentrations in the EML could be associated with a declining phytoplankton bloom and altered pigments (Tremblay et al., 1989; Riebesell, 1991; Barlow et al., 1993).

In Arctic environments, zooplankton communities are dominated by only a few copepod species. Only C. glacialis and C. finmarchicus adults were counted in the eastern, Atlantic-influenced part of the Bay where they appeared in slightly higher numbers. C. hyperboreus, M. longa, and Pseudocalanus spp. adults were spread homogeneously among the stations (Vilgrain et al., 2021, on the same cruise). As the maximum aggregate abundance deepened with OWD, the same pattern was seen with depth of the maximum copepod abundance (Figure 2). When phytoplankton in the surface began to degrade, as evidenced by higher phaeopigment concentrations (Sanmartin et al., 2011), copepods seemed to migrate below the EML following the aggregates. As diatom cells can stay viable for long periods inside aggregates (Garvey et al., 2007, Agustí et al., 2015), sinking aggregates may be a better food source for copepods than degraded phytoplankton cells in surface waters (Helling and Baars, 1985; Barlow et al., 1993). In the post-phytoplankton bloom conditions, copepods seemed to forage deeper for food and switch their feeding behavior, as suggested in other studies (Dagg, 1993; Lampitt et al., 1993; Steinberg, 1995; Dilling et al., 1998; Kiørboe, 2000; Dilling and Brzezinski, 2004; Koski et al., 2017; Toullec et al., 2019). Copepods are the dominant component of the zooplankton communities that may actively participate in particle dynamics by breaking apart aggregates, as previously suggested from laboratory experiments (Toullec et al., 2019), mesocosm experiments (Moriceau et al., 2018; Taucher et al., 2018), and in situ studies (Dilling and Alldredge, 2000), or by favoring the aggregation process of free phytoplankton cells (Taucher et al., 2018; Toullec et al., 2019). At the ice edge, maximum copepod and aggregate abundances were found at similarly shallow depths; however, copepods were much less numerous than aggregates (1:1,000 to 1:100), and the average aggregate size was not affected by fragmentation due to copepod feeding activities. Further from the ice edge, the maximum copepod and aggregate abundances were deeper but again at similar depths. Although copepods may have gone deeper to escape predation from fishes and seabirds (Lampert, 1989; Fortier et al., 2001; Karnovsky et al., 2003; Schmid and Fortier, 2019), they may have also descended to feed on phytoplankton aggregates (Bochdansky and Herndl, 1992; Bochdansky et al., 1995; Green and Dagg, 1997; Shanks and Walters, 1997; Goldthwait et al., 2004; Koski et al., 2017; Toullec et al., 2019). Low food quality, illustrated by high phaeopigment concentrations, in the surface at stations where both aggregate and copepod maxima were deep suggests that copepods were following the aggregates to feed on them because they contained less degraded algal cells (Garvey et al., 2007).

The purpose of this study was to consider the importance of hydrobiological dynamics on the formation and distribution of macroaggregates (marine snow) in Baffin Bay during spring 2016. Here, we combined hydrological parameters and advanced optical particle analysis (UVP5 deployment and particle quality based on automatic determination) to describe the distribution and fate of the aggregates in an Arctic ecosystem.

Ice melting drives the hydrological context, and thus the phytoplankton phenology in Baffin Bay. Contrary to previous expectations of bloom succession, we observed the simultaneous occurrence of diatoms and Phaeocystis spp. blooming at the ice edge. Climate change has been identified as responsible for this phenological pattern, which is expected to become increasingly common (Marchant and Thomsen, 1994; Gabric et al., 2003; Schoemann et al., 2005). Indeed, the hydrodynamic context would be affected by climate change mainly by a reduction in the EML, leading to a decrease in the vertical input of nutrients (Tremblay et al., 2002; 2015), an inflow modification that will tend to increase nutrient limitation in Baffin Bay. During this study in Baffin Bay, diatoms still dominated phytoplankton groups during the spring bloom, with their aggregation beginning once silicate concentrations dropped to 2 µM, close to the physiologically limiting value at this latitude. However, nitrate is the first limiting nutrient in the Arctic and was observed to be limiting later in our study. Aggregation may be triggered by the first nutrient that reaches a limiting concentration that stresses the diatoms, suggesting that aggregation may be enhanced by increased stratification, regardless of the nutrient involved. Haline stress resulting from ice melting possibly also contributes to the aggregation process. If primary production increases in open waters in this region, as hypothesized in the context of climate change (Babin, 2020; Lewis et al., 2020), our study suggests a concomitant increase in aggregate formation and thus a high potential for increased vertical export through the gravity pathway. As copepods go deeper, switching their feeding target from free phytoplankton cells to sinking aggregates, the active transport of matter due to zooplankton migration may also increase.

All data are accessible at the Green Edge Database (2021), available at http://www.obs-vlfr.fr/proof/php/GREENEDGE/greenedge.php, and will be made public prior to publication.

The supplemental files for this article can be found as follows:

Table S1. List of parameters per sampling stations.

Table S2. Pigment concentrations in the EML.

The project is conducted under the scientific coordination of the Canada Excellence Research Chair on remote sensing of Canada’s new Arctic frontier and the CNRS and Université Laval Takuvik Joint International Laboratory (UMI3376). We thank officers and crew of CCGS Amundsen, Marie-Hélène Forget, and Joannie Ferland for organizing the fieldwork and all other scientists and technicians involved in the Green Edge campaign for their contributions to fieldwork and data collection. We also thank the reviewers and editors who brought considerable improvements to the article.

The Green Edge project is funded by the following French and Canadian programs and agencies: ANR (Contract #111112), CNES (Project #131425), IPEV (Project #1164), CSA, Fondation Total, ArcticNet, LEFE, and the French Arctic Initiative (Green Edge project). J. Toullec’s PhD is half funded by ANR BIOPSIS project, grant ANR-16-CE-0002-01 of the French Agence Nationale de la Recherche, and half funded by a grant from LabexMer axe 2.

The authors have no competing interests to declare.

Contributed to conception and design: JT, BM.

Contributed to acquisition of data: JT, BM, MB.

Contributed to analysis and interpretation of data: JT, BM, DV.

Drafted and/or revised the article: JT, BM, LG, DV, AL, MB.

Approved the submitted version for publication: JT, BM, DV, LG, AL, MB.

How to cite this article: Toullec, J, Moriceau, B, Vincent, D, Guidi, L, Lafond, A, Babin, M. 2021. Processes controlling aggregate formation and distribution during the Arctic phytoplankton spring bloom in Baffin Bay. Elementa: Science of the Anthropocene 9(1). DOI: https://doi.org/10.1525/elementa.2021.00001

Domain Editor-in-Chief: Jody W. Deming, University of Washington, Seattle, WA, USA

Associate Editor: Julie E. Keister, University of Washington, Seattle, WA, USA

Knowledge Domain: Ocean Science

Part of an Elementa Special Feature: Green Edge