During sea-ice sampling operations, snow depth, ice thickness, and freeboard were measured at each ice-coring site. Over the cruise, ice thickness varied between 32 and 108 cm, while freeboard varied between 10 and -8 cm (top of the ice under water). Several cores were retrieved at each site using a 9 cm diameter Mark II ice corer (Kovacs Enterprises Inc., Roseburg, OR, USA). Each ice core was sliced into 10 cm segments (from the bottom) after temperature was measured. Salinity was assessed after thawing and filtration using a salinometer (Guildline Autosal 8400B, Guildline Instruments Ltd., Smith Falls, ON, Canada). Snow density and granulometry were assessed opportunistically (Eicken et al., 2009).

Hydrological conditions during the cruise were determined using several tools. A Moving Vessel Profiler (MVP) was deployed, oscillating between 0 and 300 m depth while towed at an average speed of 12 knots, rendering a very high spatial resolution (Fig. 8, bottom row, one profile every 2 km). Data obtained with the MVP matched the patterns observed from the rosette data acquired on sampling stations (Fig. 8, middle row). Profiles of conductivity, temperature, and pressure were collected using a Sea-Bird SBE 911plus CTD system rigged on the rosette. The data were post-processed according to the standard procedures recommended by the manufacturer and averaged over 0.2 m vertical bins. While there was a sharp transition in SIC at the ice edge along transect 300, the change in SIC was less steep in the more extensive MIZ of transect 500 (top row in Fig. 8). Nonetheless, both transects show similar patterns, with 100 % SIC and colder (below -1 ∘C) and fresher (salinity below 33.5 g kg-1) waters close to the surface on the western side, and 0 % SIC and saltier (above 33.6 g kg-1) and warmer (above 0 ∘C) waters within the first 50 m on the eastern side. These observations were consistent with the northward inflow of Atlantic-origin waters along the Greenland shelf break and the southward outflow of Arctic/Pacific-origin waters along the Baffin Island shelf break.

Currents in the water column were measured using a hull-mounted 150 kHz acoustic Doppler current profiler (ADCP, Teledyne RD Instruments Ocean Surveyor, California, USA) as well as two L-ADCP installed on the rosette structure (RDI, WHM300-1-UG304) in a master/slave configuration. Vertical profiles of water turbulence were measured at each FULL station using a Self-Contained Autonomous MicroProfiler (SCAMP, Precision Measurement Engineering, California, USA) deployed from the zodiac. A detailed description of the hydrological structures in the studied area during the Green Edge cruise is presented in Randelhoff et al. (2019).

An abundance of viruses and bacteria was determined on fresh and preserved (glutaraldehyde 4 % final concentration) water samples taken from the rosette at each FULL and BASIC station (10 depths) using two different flow cytometers. On board, fresh samples were counted using an Accuri^™ C6 and preserved samples were counted back in the lab using a FACSCanto (both machines from Becton Dickinson Biosciences, San Jose, CA, USA). Samples were processed according to Marie et al. (2001). Bacteria (and viruses) are ubiquitous in the oceans, and in Baffin Bay, we measured bacterial abundances up to 2.9×106 cells mL-1.

For bacterial diversity analysis, water samples were filtered sequentially onto 20 µm, 3 µm (both polycarbonate filters, Millipore), and 0.22 µm (Sterivex-GV, Millipore). The filters and Sterivex were stored at -80 ∘C with RNAlater (Qiagen) until analyzed. The DNA/RNA co-extraction was carried out using the AllPrep DNA/RNA kit (Qiagen). The V4–V5 hypervariable region of the 16S rRNA gene was amplified by PCR using primers 515F-Y and 926R, covering a broad spectrum of diversity, including Archaea and Bacteria (Parada et al., 2016). PCR, as well as sequencing settings and bioinformatics of sequence data, can be found in Dadaglio et al. (2018).

A comparison of the bacterial diversity as a function of geographic location and size fractions (free living bacteria, bacteria attached to particles smaller than 20 and larger than 20 µm) was made at a relatively broad taxonomic level (Fig. 10). The samples from the different groups were mainly dominated by Bacteroidetes and Proteobacteria. In general, the proportion of Proteobacteria decreased from ice stations to open-water stations featuring a more advanced stage of PSB, giving way to Bacteroidetes and, more specifically, Flavobacteriaceae.

Flow cytometry was used (same protocols as for viruses and bacteria, Sect. 4.5.1) to count and differentiate the smallest cells (picophytoplankton, nanophytoplankton, cryptophytes, and Synechococcus) according to their fluorescence and scattering properties at each FULL station. At the surface, picophytoplankton and nanophytoplankton cell concentrations could reach 60 000 (station 719 – transect 7, station 19) and 9000 (station 515 – transect 5, station 15) cell mL-1, respectively, while cryptophytes were always below 360 cell mL-1. Synechococcus cyanobacteria were never observed.

Some samples were used to start phytoplankton cultures, which were taken back to the Roscoff laboratory for purification using flow cytometry sorting, serial dilution, and single-cell pipetting. Pure cultures were characterized by microscopy and 18S rRNA gene sequencing. Most cultures isolated during the cruise belonged to diatoms, especially to the genera Attheya and Chaetoceros (Gérikas Ribeiro et al., 2020). All cultures were deposited in the Roscoff Culture Collection and are available for distribution (http://www.roscoff-culture-collection.org/strains/shortlists/cruises/green-edge, last access: 19 September 2022).

To study the phytoplankton community composition, an Imaging FlowCytobot (IFCB, McLane Research Laboratories Inc., East Falmouth, MA, USA) was used during Leg 1B. The IFCB is best used for the study and identification of cells between 1 and 150 µm. Fresh samples (5 mL) taken from the rosette at each FULL station (all depths) and some BASIC and NUT stations (2 to 7 depths) were analyzed; samples from the 2 bottom-most slices of ice cores were also analyzed once melted. The IFCB takes pictures at a resolution of around 3.4 pixels per µm. Image descriptors and features were extracted with Matlab® using scripts developed by Heidi Sosik (Sosik and Olson, 2007). Taxonomic determination was achieved using Ecotaxa (Picheral et al., 2017; http://ecotaxa.obs-vlfr.fr, last access: 12 March 2022). Random forest algorithms were used for automatic classification. Reference sets and the validation of predictions were both done manually. Examples of specimens observed during the Green Edge campaigns can be found in Massicotte et al. (2020).

A total of 203 underwater vertical profiles were acquired using an Underwater Vision Profiler (UVP, model 5-DEEP, Hydroptics, France) installed on the frame of the rosette carousel. The UVP5 collects in-focus images in the small seawater volume lit by its light emitting diodes (LEDs) as it is lowered in the water column. An automated computer system (https://ecotaxa.obs-vlfr.fr/, last access: 12 March 2022) was used to sub-sample images of individual objects and sort them into the appropriate category (marine snow or various taxa of zooplankton). The UVP has been developed mainly to count and identify particles larger than 100 µm. Figure 11 (left panels) show the average vertical profiles of particle concentration (mL-1) over the top 350 m of the water column for open-water (top) and under-ice (bottom) stations. The number of particles close to the surface in open waters coincides with the larger phytoplankton biomass observed there compared with ice-covered stations, consistent with lower primary and secondary production under sea ice. Deeper in the water column, around 300 m, the large particle concentrations observed at open-water stations likely reflect resuspension of bottom sediments, because these observations were mostly made in the eastern part of Baffin Bay over the continental shelf, whereas under-ice stations were mostly located on the deeper Canadian side of the Bay (see the bathymetry in Fig. 1).

Samples for taxonomic analyses of micro-algae by microscopy were taken at each FULL and BASIC station at 10 sampling depths. Half a litre of seawater was preserved with Lugol and kept at 4 ∘C until it was analyzed in the laboratory. Visual observation and taxonomic determination were done using an inverted microscope (Eclipse TS100, Nikon Instrument Inc.) according to the Utermöhl method (Utermöhl, 1958) using 25 or 50 mL columns. Three transects of 26 mm at 400× were systematically observed for identification and counting of Bacillariophyceae, Dinophyceae, flagellates, and ciliates. Larger phytoplankton cells and colonies were observed in all chambers at 100×. Diatoms (Bacillariophyceae) were found at every station, primarily at the surface of the water column, along with flagellates (both at the surface and in the subsurface chlorophyll maximum, SCM) (Fig. 12). The most striking feature was the dominating presence of a Phaeocystis sp. (Prymnesiophyceae, blue bars in Fig. 12) at the SCM, reaching 60 to 90 % of the cell counts (and, to a lesser extent, at the surface) across a wide range of ice cover conditions (OWD values between -12 and 12 d).

Phytoplankton and ice-algae pigments were measured to derive indices of micro-algae biomass and taxonomic composition and to get information on processes such as photoacclimation, senescence, and grazing activities (Roy et al., 2011). Rosette water samples were filtered onto GF/F filters (Whatman^™, GE Healthcare Life Sciences) and quickly frozen in liquid nitrogen. Back in the land-based laboratory, samples were thawed and extracted in 100 % methanol, separated, and identified by HPLC, as described by Ras et al. (2008). A total of 25 individual pigments or groups of pigments were identified and quantified at each FULL and BASIC station (10 depths sampled each time). Figure 13 shows the distribution of total chlorophyll a and phaeophorbide concentrations along transects 300 and 500. The high chlorophyll a concentrations close to the surface in the MIZ and deepening towards open waters in the east confirmed the evolution of the PSB from an under-ice bloom to a SCM. Note that highest concentrations of phaeophorbide were systematically found underneath the accumulation of chlorophyll a, indicating the sinking of degrading phytoplanktonic material.

Zooplankton represents the second level of the food chain. The UVP5 and the Imaging FlowCytobot (see previous sections) both rendered valuable information on small zooplankton specimens (below 150 µm). Figure 11 (right panels) show copepod volumetric fraction (cm3 m-3) over the top 350 m of the water column for open-water (top panel) and under-ice (bottom panel) stations. The total copepod volumetric fraction calculated based on automatic identification made using the Ecotaxa web application (http://ecotaxa.obs-vlfr.fr, last access: 12 March 2022) shows subsurface peaks around 20 m at both open-water and ice-covered stations. A secondary peak right at the surface is present at ice-covered stations, where a subpopulation of copepods may stay close to the bottom of sea ice to feed on sympagic microalgae, small animals, and related detritus.

For bigger specimens, a series of vertical nets and trawls were deployed at each FULL station (no trawling operations took place when sea ice was present). An assembly of 4 nets of 3 different mesh sizes (50, 200 and 500 µm) coupled with a Lightframe On-sight Key species Investigation (LOKI) system rendered high-resolution pictures of individuals sampled along the water column together with actual specimens. The multi-net plankton sampler (Hydro-bios, Altenholz, Germany) uses a different sampling strategy. Composed of 9 identical nets (200 µm mesh size), it is hauled vertically in the water column, with the nets opening sequentially at different depths, each net collecting a slice of the water column displaying the vertical distribution of the species sampled. Figure 13c shows the abundance (ind m-3) of zooplankton sampled using the vertical 200 µm mesh net along transects 300 and 500. Copepods represent the main zooplankton class observed during the cruise, and they were present at every sampling site. The particularly high abundance at station 507 (situated under the ice where a phytoplankton bloom had already disappeared) might be due to the relatively high abundance of copepod nauplii (25 % of all copepod individuals compared to the usual 4 % at other stations).

Ichthyoplankton were sampled using a double square net towed obliquely from the side of the ship at a speed of ca. 2–3 knots to a maximum depth of 90 m. A Star-Oddi^® mini-CTD attached to the frame and flowmeters determined the real depth and volume of sampling. For fish sampling, an echo sounder (EK60, Simrad, Kongsberg Maritime, Norway) mounted on the hull was used to locate and determine the depth of fish aggregations along the ship track during the entire cruise. When pelagic juveniles and adult fish were present at the sampling stations, an Isaac–Kidd Midwater Trawl (IKMT, Filmar and Québec-Océan, Québec, Canada) was towed for 20 min at a speed of 2–3 knots. When demersal fish were detected, a benthic beam trawl (BBT, Filmar and Québec-Océan, Québec, Canada) was used to sample bigger specimens (between 10 and 32 mm mesh size) living on the bottom sediment.

Samples of zooplankton were all processed in the same way. Swimmers (fish larvae and juveniles) were sorted out, measured, identified, and preserved in a mix of 95 % ethanol and 1 % glycerol (final concentrations) for later analysis, while zooplankton samples were preserved in 4 % formaldehyde solution. Zooplankton abundance and diversity were determined using binoculars back at the laboratory. A few samples (hydro-bios) were analyzed using the ZooScan and the Ecotaxa identification tools (https://ecotaxa.obs-vlfr.fr/prj/802, last access: 12 March 2022). Fish from IKMT and BBT sampling were sorted, counted, identified, and measured before preservation in a -20 ∘C freezer in case further analyses are needed in the future. Acoustic data from the EK60 were analyzed in Echoview^® (see Geoffroy et al., 2016 for details).

Benthos sampling was performed to test if (i) sea-ice cover is the primary environmental driver of the contribution and geographic distribution of sympagic carbon on the seabed, (ii) sympagic carbon is the most important baseline food source supporting benthic consumers during spring and summer in areas close to the MIZ, and (iii) deep benthic food web dynamics and structural variability are directly linked to both depth and availability of food sources (Yunda-Guarin et al., 2020). The sampling was achieved using two different strategies; a total of 16 Agassiz trawling and 34 box coring operations (some down to more than 2000 m depth) were carried out during the cruise. The Agassiz trawl (KC Denmark a/s Research Equipment) is a medium-size dredge trawled behind the ship, allowing sampling of the macrofauna living on the sediment surface. Once brought back on board, the contents of the net were immediately rinsed with seawater, manually sorted, and identified to the lowest taxonomic level possible. Samples were brought back to the laboratory to be identified under a dissecting microscope when onboard identification was impossible. Figure 14 shows an example of the diversity of benthic organisms. More than 220 species of macrofauna were identified during the Green Edge cruise from more than 25 classes (Grant, C. and Yunda-Guarin, G., unpublished data). A box corer was used to sample sediment from each FULL station. The sediment samples were then divided among the research teams for diverse analyses (see Table 3 for a complete list), including (but not limited to) identification of organisms living inside the sediment, incubations for respiration, and nutrient utilization or chemical analysis.

During the entire cruise, a systematic bird and marine mammals survey was carried out from the ship's wheelhouse (see LeBlanc et al., 2019 for detailed methodology). A total of 20 different bird species and 8 different mammal species were identified. Northern fulmar, thick-billed murre, and little auk were the most common bird species observed. Ringed seal, hooded seal, and harp seal were the most common seals. The long-finned pilot whale was the most common whale species observed. A total of 10 polar bears were observed.

A total of 123 seabirds from 7 species were also collected from a zodiac deployed from the CCGS Amundsen in Greenland waters between 10 June and 8 July. This includes black-legged kittiwakes (n=8), glaucous gulls (n=6), great black-backed gull (n=1), little auks (n=19), northern fulmars (n=42), and thick-billed murres (n=36). Sampled birds were frozen at -20 ∘C until laboratory analyses. A first study aimed to investigate the co-distribution of seabirds and their fish prey along the MIZ (LeBlanc et al., 2019). To this end, stomach contents were examined for 74 birds (35 murres, 30 fulmars, and 9 kittiwakes) under a dissecting microscope. Otoliths were retrieved and used to identify fish species, age, and size. A second focus was the recording of plastic in the stomachs. Plastic data are used in OSPAR monitoring and AMAP working groups on plastic pollution. A third study aimed to determine the birds' association to sea ice and ice-derived resources by the combination of different trophic markers. Hence, liver, muscle, and blood (from the cardiac clot) samples were collected from a total of 52 bird carcasses (27 murres, 14 auks, 3 kittiwakes, and 8 fulmars), on which highly branched isoprenoids (HBIs), carbon and nitrogen stable isotopes, and fatty acids were measured. Finally, a fourth study was looking at stable isotope data together with mercury (Hg) data for both muscle and liver.