Under-ice observations by trawls and multi-frequency acoustics in the Central Arctic Ocean reveals abundance and composition of pelagic fauna

The rapid ongoing changes in the Central Arctic Ocean call for baseline information on the pelagic fauna. However, sampling for motile organisms which easily escape vertically towed nets is challenging. Here, we report the species composition and catch weight of pelagic fishes and larger zooplankton from 12 trawl hauls conducted in ice covered waters in the Central Arctic Ocean beyond the continental slopes in late summer. Combined trawl catches with acoustics data revealed low amounts of fish and zooplankton from the advective influenced slope region in the Nansen Basin in the south to the ice-covered deep Amundsen Basin in the north. Both arctic and subarctic-boreal species, including the ones considered as Atlantic expatriate species were found all the way to 87.5o N. We found three fish species (Boreogadus saida, Benthosema glaciale and Reinhardtius hippoglossoides), but the catch was limited to only seven individuals. Euphausiids, amphipods and gelatinous zooplankton dominated the catch weight in the Nansen Basin in the mesopelagic communities. Euphausiids were almost absent in the Amundsen Basin with copepods, amphipods, chaetognaths and gelatinous zooplankton dominating. We postulate asymmetric conditions in the pelagic ecosystems of the western and eastern Eurasian Basin caused by ice and ocean circulation regimes.


Results
Sea ice and oceanic conditions.. The trawl and acoustic sampling were conducted in a region spanning strong gradients in sea ice and oceanic conditions (Fig. 2). The sea ice concentration increased substantially just to the north of NB1, and NB2 and GR1 had sea ice concentrations between 70 and 90% (Fig. 2a). Sea ice concentration at the three northernmost locations exceeded 90%. The sea ice drift during sampling was only modest with mean drift mostly from east (northeast or southeast) to west except at location AB2 and between location NB2 and GR1 when it was reversed (Fig. 2b). www.nature.com/scientificreports/ Velocity, temperature, and salinity also changed rapidly to the north of NB1 (Fig. 2c-e). From about 83 o N, extremely cold (− 1.8 °C < T < − 1.5 °C) and fresh polar waters covered the upper 100 m. A pronounced surface salinity front appeared in the northern Nansen Basin-Gakkel Ridge region (Fig. 2e), and the vertical stratification strengthened going northwards due to decreasing upper layer salinities. Warmer water (T > 0 °C) of Atlantic origin was present below 125-170 m depth (Fig. 2d).
The vessel-mounted ADCP data revealed eastwards velocities reaching 25-30 cms -1 along the slope of the Nansen Basin (Figs. 1c and 2c). In this region, warm and saline Atlantic Water which enters through Fram Strait (Fig. 1a) flows eastwards into the Arctic Ocean 28 . Our southernmost location NB1 was slightly to the north of the slope and main core (located between the 300 and 1000 m isobaths 29,30 ) of the Atlantic Water current. Instead, the ADCP data showed rapidly varying horizontal flow near NB1 indicating mesoscale eddies (Figs. 2c and S1). To the north of NB1 (Figs. 1c and 2c), our mean 25-250 m ocean currents showed mostly weak, zonal flow, i.e., no direct flow from the southern part of the transect to the northern parts.
Organisms caught by trawls. Twelve trawl hauls were taken in pairs at the six locations along the route: eight hauls with a Harstad trawl targeting pelagic fish and four hauls with a macroplankton trawl for catching larger zooplankton (Table 1). Three hauls were towed in the epipelagic layer, six hauls in mesopelagic layers and three covering both layers.
The biological samples contained 26 taxa, with some organisms identified to species level, and others to family, order, or phylum levels ( Table 2). The trawl catches by the macroplankton and Harstad trawls confirmed the hypothesized ubiquitous occurrence but generally low abundance of organisms, including fish, in the CAO. All Harstad trawl catches were very small and varied between 3 and 325 g nm −1 (Fig. 3a, Table S2). The two first hauls in the Nansen Basin had the largest catches (NB1e and NB1m, 164 g nm −1 and 325 g nm −1 respectively), together with one haul in the Amundsen Basin (AB2em, 208 g nm −1 ). The Harstad trawl revealed very small catches at the Gakkel Ridge (GR1m and GR2m, 9 g nm −1 and 3 g nm −1 respectively). The four hauls with the finer-meshed macroplankton trawl generally yielded higher catches (137-696 g nm −1 ) than the Harstad trawl (Fig. 3b, Table S2).
The Harstad trawl caught a total of 15 taxa, namely included Ctenophora (44%), Arthropoda (40%), Cnidaria (12%), Chaetognatha (2%), Mollusca (2%) and Teleostei (1%) by wet weight (Fig. 3a). Euphausiids (Meganyctiphanes norvegica, 3.6-4.0 cm,  Table S3) and periphyllids (Periphylla periphylla) contributed most to the catches in the Nansen Basin, while T. libellula (3.0-3.8 cm, Table S3) dominated the catches in the Amundsen Basin and at the Gakkel Ridge (Fig. 3). For P. periphylla, our record is the furthest north to our knowledge. Three fish families, each represented by one species, were found including Myctophidae (glacier lanternfish), Gadidae (polar cod), and Pleuronectidae (Greenland halibut Reinhardtius hippoglossoides). The few fish sampled included one 8 cm (total length) polar cod (at NB1e), one 4 cm Greenland halibut larva (at NB2e), three glacier lanternfishes at NB1m (4.5, 4.5 and 5.5 cm) and one at AB1m (5.0 cm). This is, as far as we know, the first time glacier lanternfish has been caught this far north, although the species has been tentatively identified by video recordings even further to the north in this region 13 . It is also to our knowledge the northernmost record of Greenland halibut, and the first time it has been observed north of the shelf break in the CAO.
As expected, the macroplankton trawl had higher catches of the smaller organisms (e.g., copepods, ctenophores, and chaetognaths) compared to the Harstad trawl (Fig. 3). Twelve plankton taxa were caught by the macroplankton trawl only (Table 2). Combining results from both trawls catches show that some taxa such as Ctenophora (Beroe cucumis) and hyperiid amphipods (Themisto libellula) were observed along the whole transect. Copepoda is another group that we expect to be represented at all stations, however, this was not evident as the gears we used are not suitable for catching mesozooplankton. In the macroplankton trawl, though, we recorded biomass of the larger Calanus hyperboreus (ca. 7 mm), these are most likely underestimated. Other taxa were taken at either only southern or northern locations. The amphipod Eusirus holmii and the decapod Hymenodora glacialis were found at the northernmost locations only. Polar cod and Greenland halibut were caught at southern locations only, while glacier lanternfish occurred at both southern and northern locations.
Multi-frequency acoustics. The acoustic data were categorized using sequential thresholding, frequency response (rf 18/38 ), and target strength (TS) analysis (Supplementary section 3). The sequential thresholding showed epi-and mesopelagic layers at all locations (Fig. 4), albeit with varying strength and much weaker than usually observed in the Norwegian and Barents Seas. In the Nansen Basin, a faint (maximum s A of 0.3-0.4 m 2 nmi -2 , where s A is the Nautical area scattering coefficient (NASC) 50 ) acoustic scattering layer was present at NB1 at 10-100 m in water with temperatures slightly below zero ( Fig. 4a and S16). The layer contained mostly weak . Station code includes labelling for areas where samples were taken: Nansen Basin (NB), Gakkel Ridge (GR) and Amundsen Basin (AB). The station code also includes a notation denoting the depth of the trawl (e, epipelagic; m, mesopelagic, and em, both layers). The trawls denoted with MT were macroplankton trawls, while the others were Harstad trawls. Additional information, including on the acoustic stations and the CTDs, can be found in Table S1.  www.nature.com/scientificreports/ detections were evident, and most acoustic targets had lower TS 38 and were probably plankton organisms, which also totally dominated the trawl catches (Fig. 3, Table S2). Compared to the other stations, a stronger mesopelagic scattering layer was present at NB1 in 300-500 m depth in Atlantic Water temperatures (1.5-2.0 °C) (Figs. 4a and S16). The TS analysis implied that the layer contained larger fish than caught by the trawl, like for instance, polar cod (Supplementary section 3). If so, the observed TS 38 mode would translate into fishes of lengths of about 7 cm 51 . In addition, one single echo from a target at 265 m stood out with an s A more than 10 times higher than the rest (Fig. S18). More than 20 TS 38 detections of this target were in the range − 20 to − 32 dB.
In the central Nansen Basin (NB2), faint echoes forming a scattering layer could be seen in the depth interval 100-200 m (Figs. 4b and S19). Very cold polar water (− 1.8 °C < T < − 1.5 °C) occupied the upper 100 m and the scattering appeared at depths where the temperature approached 0 °C (Fig. 4b). All targets had TS 38 weaker than − 50 dB, indicating that no larger fish were present. Consistently, the matching trawl catches consisted of a mixture of gelatinous and crustacean plankton and an armhook squid (Table S2). Additionally, one larval Greenland halibut (4 cm long) was caught, which would have a TS 38 far lower than − 50 dB since it does not have swim bladder. The scattering in the mesopelagic layer was weak (Fig. 4b), but TS 38 revealed a mode at − 52 dB (Fig. S17).
The echograms from the two locations at the Gakkel Ridge (GR1 and 2) showed a distinct epipelagic layer in 20-40 m depth, in polar waters with temperature below − 1.5 °C (Fig. 4c, d, Figs. S20, S21). A sharp peak of increased fluorescence in the upper water column overlapped with the increased acoustic scattering (Fig. S22), implying that the scattering stemmed from zooplankton gathered in a layer of phytoplankton. Consistently, the S75 and S80 categories at GR2 were higher than anywhere else in the transect (Fig. 4c, d), implying a different zooplankton composition or density. However, it is also possible that some of the backscatter originated from the sound velocity contrast in this pycnocline (Fig. S22). At greater depths there was a weak mesopelagic backscatter, with TS 38 values with a mode for − 54 dB (Fig. S17). The frequency response (rf 18  www.nature.com/scientificreports/ a mixture of organisms with and without gas inclusions, while the catches from the Harstad (50-550 m) and macroplankton (60-537 m) trawls were dominated by arctic hyperiid amphipods (Fig. 3, Table S2).
In the southern Amundsen Basin (AB1), the 25-30 m depth layer had a stronger acoustic signal than elsewhere in the study area, and with a different composition and/or density (a stronger S70 fraction, Fig. 4e). The echo registrations at deeper water were weak, consistent with the trawls catches being dominated by a mixture of amphipods, copepods, ctenophores and chaetognaths with a clear dominance of smaller organisms ( Fig. 3 and Table S2). However, the TS 38 registrations at depth resembled those at the more southern locations, although with a wider distribution peaking at − 58 dB (Fig. S17). Nevertheless, also at this location the TS 38 distribution at depth revealed numerous TS 38 registrations at − 54 dB. In the central Amundsen Basin (AB2), both the epi-and mesopelagic scattering was weak (Figs. 4f and S17).

Discussion
Epi-and mesopelagic fauna in the western Nansen and Amundsen Basins. The survey covered a gradient from an advective/mesoscale eddy regime strongly influenced by the Atlantic Water slope current in the southern Nansen Basin towards a gradually more pronounced Arctic regime from ~ 83 o N and northwards (Fig. 2). However, also the Arctic regime had waters of Atlantic origin (T > 0 °C) at depth (Fig. 2d), holding a distinct mesopelagic scattering layer (Fig. 4), which has been shown to be widely distributed in the CAO 23 . Our trawl catches show that the mesopelagic layer contains both subarctic-boreal and Arctic species all the way north to 87.5 o N (Fig. 5), consistent with earlier results 11 . Some of the larger zooplankton species we found are also present in the Amerasian Basin as assessed by ROV 52 while many of the smaller zooplankton species found there were not caught in our trawls, which likely is partly an effect of gear rather than biogeography.
Near-slope in the Nansen Basin. Higher catches (Fig. 3) and acoustic backscattering (Fig. 4a) in the near-slope region in the Nansen Basin (NB1) compared to the locations visited further north are consistent with the high inflow of zooplankton in the Atlantic Water flow at the slope of the southern Nansen Basin 33,35,[53][54][55] . Some advected taxa of boreal Atlantic origin, such as Meganyctiphanes norvegica, are not currently likely to reproduce in the Eurasian Basin 11 . However, if warming persists, these species may in future potentially complete their life cycles in these waters. These advective organisms play an important role as a substantial food source to consumers in the region 35 .
The epi-and mesopelagic fauna in the southern Nansen Basin was more diverse compared to further north (Fig. 5). It included well-documented subarctic-boreal ctenophores, both subarctic-boreal and arctic amphipods and euphausiids (consistent with a synthesis by Kosobokova 11 ), and glacier lanternfish and polar cod. The acoustic data revealed a strong echo target (Fig. S18) with TS 38 detections in a range implying a large (130-140 cm) Atlantic cod, or a marine mammal or a big fish without a swim bladder (Supplementary section 3). The most probable candidate is a Greenland shark (Somniosus microcephalus), which is commonly found near the slope in this area 56 . Juvenile Beaked redfish (Sebastes mentella), haddock (Melanogrammus aeglefinus), capelin (Mallotus villosus), polar cod, Atlantic cod and Greenland halibut can also be found in the meso-and epipelagic layers closer the slope slightly further west, yet some species appear to occur seasonally 19,21,57 . Also, adult redfish and Atlantic cod have been caught in the mesopelagic layer just off the slope slightly further west 21,58 . However, these reported catches were taken closer to the Atlantic Water slope current.
The central Nansen Basin. Backscatter declines strongly in the upper mesopelagic zone when crossing boundaries between subpolar and polar water masses 59 , and such a decline was also evident when moving from the near-slope (NB1) to the central (NB2) Nansen Basin (Figs. 2 and 4a, b). It is worth nothing, however, that our observed gradient can be biased by dial migration occurring in these regions 20 . Since the central Nansen Basin location (NB2) was sampled near midnight (Table S1), the weak scattering at depth in combination with enhanced scattering near 150 m (Fig. 4b) at least partly can be caused by organisms, which have moved upwards from the deeper mesopelagic layer toward the surface during night-time.
The mesopelagic layer carried much the same species as near the slope (Fig. 5), with the addition of armhook squid and one larva of the subarctic-boreal Greenland halibut. Armhook squid appear to be widely distributed in the Arctic 13,21,52 . The Greenland halibut larva, on the other hand, has not been observed this far north before, and was likely advected with the Atlantic Water branch flowing around the Yermak Plateau (Fig. 1) or by mesoscale eddies during the summer. Both mesoscale eddies 30,31 and recirculation of the Atlantic Water 32 appear to be common in this region. Such smaller scale and regional features can provide a means for bringing subarcticboreal organisms, advected with the Atlantic Water current 21,33-36 , off the slope and into the central Nansen Basin. Another possibility is that the Greenland halibut larva originates from the shelf-break on the Kara, Laptev Seas, or East Siberian Seas 60,61 .
The mesopelagic layer in the central Nansen Basin contained some crustacean plankton including subarcticboreal euphausiids and arctic amphipods (Fig. 5), but was by weight dominated by the subarctic-boreal gelatinous deep-water plankton Periphylla periphylla. Increased inflow of Atlantic Water may have provided more suitable conditions for this species to colonise the European Arctic in recent decades 48 . Yet our record appears to be the furthest north so far.
Gakkel Ridge. Despite heavy sea ice (80-90% concentration) and low (T < − 1.5 °C) temperatures ( Fig. 2a   and d), the acoustics revealed a distinct epipelagic layer in 20-40 m depth at the Gakkel Ridge (Fig. 4c, d), possibly consisting of organisms associated with a bloom situation. Ice algae blooms at the ice-water interface are common 62 , but also under-ice phytoplankton blooms occur at low magnitudes in this area and have been found to coincide with a shallowing of the pycnocline to 50 m 63 . Pelagic algal blooms generally occur after ice breakup, www.nature.com/scientificreports/ which often is as late as August-early September in these northern regions 64 . The trawls hauls do not overlap with the most intense backscattering in the epipelagic layer, and hence, cannot directly corroborate the acoustic data. A mesopelagic scattering layer was also evident in these regions, although very weak (Fig. 4c, d). The fine meshed macroplankton trawl showed copepods and euphausiids, together with some cnidarians at ca. 350 m depth and, to our knowledge, the northernmost record of Nematoscelis megalops (Table 2). However, both the macroplankton and the Harstad trawl catches revealed epi-and mesofauna dominated by arctic species (Table S2 and Fig. 5), which can possibly be related to the Transpolar Drift 27 (Fig. 1b). Earlier studies have hypothesised that both arctic amphipods (albeit the more sympagic species) 65 and polar cod 9 follow the ice drift from the east toward the western Nansen Basin.
Amundsen Basin. Both trawl catches (Fig. 3) and acoustic registrations (Fig. 4e, f) revealed very low concentrations of zooplankton and fish when entering the extensively ice covered (> 90%), cold and highly stratified (Fig. 2d, e) southern Amundsen Basin (AB1). No catches were taken to corroborate the acoustic peak observed in 25-30 m depth in the southern Amundsen Basin (Fig. 4e), but the macroplankton trawl from the deeper layers revealed a high biomass of copepods in this region (Table S2).
The catch in the mesopelagic layer in the Amundsen Basin consisted of a mixture of gelatinous and crustacean plankton and chaetognaths, and the only fish caught were a few individuals of glacier lanternfish in the southern parts of this Basin (Figs. 3 and 5). However, the observed TS 38 peak at -55 to 54 dB (Fig. S17) were considerably stronger than the previous estimates of about − 62 dB for adult glacier lanternfish 66 . Since the TS data were filtered to be within 2° of the acoustic axis before analysis, this result should not be biased due to reduced signal-to-noise ratio in the outer parts of the beam. Furthermore, similar (− 55 to 52 dB) modes were evident at all locations south of AB1. If these stronger echoes came from for instance polar cod, our TS 38 data imply that polar cod of lengths 6-9 cm 51 were present at most locations, although in too low concentrations to be caught by the trawls.
Subarctic-boreal species such as the boreal Atlantic expatriate 11 euphausiid Meganyctiphanes norvegica were observed all the way to 87.5 o N in the Amundsen Basin (Table 2). These findings add support to our hypothesis that the pelagic composition is driven by dominating water mass origins, but with presence of subarctic-boreal species also in the northernmost parts. Moreover, since the strong topographic steering cause the Atlantic www.nature.com/scientificreports/ Water to mainly follow slopes and ridges rather than crossing the deep basins 26,28 (Fig. 1a, c), it is more likely that these subarctic-boreal species reached the northernmost regions by advection through the eastern Nansen and possibly Amundsen Basins. The long advection times from Fram Strait (2-5 years) 69 .
In the central Amundsen Basin (AB2) the catches revealed a strong presence of gelatinous plankton (ctenophores), again including subarctic-boreal species (Fig. 5). Overwintering individuals of three ctenophore species have been observed in the Chukchi Sea, leading to a proposed hypothesis that they can survive the winter under sea ice in the coastal Arctic by the continued availability of prey related to high productivity, including production by ice algae 70 . Ctenophores had in general a wide distribution in the Nansen and Amundsen basins and were observed at all locations (Fig. 5, Table 2). These observations are consistent with those from the upper 100 m of the Canada Basin 71 . The finer meshed macroplankton trawl also showed presence of Calanus spp., Themisto spp., Pseudosagitta and Eukrohnia spp. (Table S2). These comparatively small organisms are clearly poorly represented in the Harstad trawl which is designed to capture fish.

Regional differences within the European Arctic.
Our results show that the densities of both fish and their prey organisms in the Nansen and Amundsen Basins are very low compared to regions further south. The observed catch rates are two orders of magnitude lower than in the slope current north of Svalbard 19,21 . Also the acoustic backscattering is two orders of magnitude less than in northern Svalbard waters 22,58,72 , a region which in turn has backscattering an order of magnitude less than the Norwegian Sea 21 . This confirms our hypothesis that pelagic fishes and large zooplankton in the CAO have lower abundance compared to the south, although the abundances are even lower than we expected. Moreover, recent studies have shown lower densities in the western Nansen and Amundsen Basins as compared to their eastern parts 13 . Thus, despite being closer to the inflow region in Fram Strait, the density of fish and other pelagic organisms in the deep western Nansen Basin seem to be lower than further east. At the same time, the biogeographic composition reflects connectivity to the Atlantic inflow. It remains unclear if the addition of species not found earlier in the study region is a result of strengthened or more frequent inflow or increased research effort.
Substantial warming, subsequent sea ice loss and increased influence of subarctic-boreal species have been documented in the western Nansen Basin slope region 2,34,53 . However, for the off-slope deep Nansen Basin, the largest changes in summer sea ice have occurred in the eastern parts 1 . Changes in summer sea ice in the deep western Nansen Basin are much smaller due to the Transpolar Drift continuously bringing sea ice westwards into the region and further towards Fram Strait 27 (Fig. 1b). Moreover, the eastern Nansen and Amundsen Basins are also influenced by the Atlantic inflow from the Barents Sea (Fig. 1a). Both the Barents Sea and the eastern Nansen Basin are currently experiencing a pronounced Atlantification with an associated borealization of the ecosystem 2,73 . Therefore, the eastern parts of these basins may be more productive and have better living conditions for subarctic-boreal species than their western parts.
Sampling in ice covered waters. Acoustic monitoring and net sampling in ice-covered waters pose challenges. Utilizing multi-frequency acoustic data from research vessels is a major step forward methodologically that will facilitate mapping and monitoring of biota in ice-covered waters. However, the high noise and blocking of the acoustic signal generated by the ice-breaking process cause limitations. When lying still, however, the instrumentation provided acoustic data with a signal to noise ratio acceptable for use at several frequencies in the upper water column, and with the lowest frequencies (18 and 38 kHz) down to at least 500 m. The same conclusion was drawn from using 18 kHz acoustic data from a CAO crossing with R/V Oden 23 . We note, however, that because of the reduction in signal to noise ratio with range, only the TS of larger organisms were detected at a few hundred meters.
Pelagic trawling in ice-covered waters with ordinary equipment is challenging because ice floes floating behind the vessel can easily destroy the net during deployment or retrieval. Under-ice trawl samples have previously been obtained with a Surface and Under Ice Trawl 9,74 , but pelagic fish trawls have so far not been used in the CAO due to its thick ice cover 23 . Our results show that only modest adaptations of an ordinary pelagic trawl (Harstad trawl) used routinely on Norwegian research vessels for abundance and biomass estimations of fish, functioned well in ice-covered waters. We used ice trawl gallows to move the tow blocks into the centre of the stern of the vessel, thereby closing the trawl before it reached the surface during retrieval and before sinking beneath the surface during setting. In addition, the trawl was slightly modified by reducing the flotation and adding weight to the cod-end. This made the trawl hang almost vertically from the stern of the vessel during deployment and retrieval, which further reduced the problem with sea ice entering the trawl. For the smaller macroplankton trawl, the only modification made was adding 40 kg to the cod-end.
Horizontal towing of a trawl has clear advantages compared to other sampling methods (like gillnets, fishing lines, vertical towed nets or video), particularly for extremely low densities as in the CAO. Towing one to two nautical miles, as during this survey, the Harstad trawl filtered approximately 500.000-1.000.000 m 3 of water. Nevertheless, with its vertical opening of 10-12 m (Table 1), and relatively short towing times, the trawl is still too small to obtain large catches at these densities, despite using the same towing speed as in ice-free waters. With maximum catch rates of about 325 and 696 g nm −1 (Harstad trawl and macroplankton trawl, respectively), there are uncertainties related to whether the catch composition reflects the actual composition of organisms in the sea. The different catchabilities of the Harstad and macroplankton trawls add further complexity to this issue. Using a larger trawl would undoubtedly be a more efficient approach for catching fish but would also www.nature.com/scientificreports/ result in severe complications regarding operations in ice-covered waters. For catching smaller organisms (e.g., copepods), smaller nets like WP2 or Multinet with 180 µm mesh are more efficient.
Baseline mapping of an ice-covered ecosystem. In 2021, a CAO Fisheries Agreement entered into force (www. fao. org/ faolex/ resul ts/ detai ls/ en/c/ LEX-FAOC1 99323). In addition to preventing fisheries in the high seas of the CAO at least up to 2037, the agreement calls for the establishment of a Joint Program of Scientific Research and Monitoring with the aim of improving the understanding of the ecosystems. A draft plan for such a program emphasised the need for baseline mapping, monitoring and research 8 . Using trawls and rigging modified to use in ice covered waters, in combination with state-of-the art multifrequency acoustics, our results directly contribute to the baseline mapping of the CAO ecosystem while it is even today still ice-covered during most of the year. Our results show that, at present, the densities of both ecologically and economically important fish and their prey organisms in the western Nansen and Amundsen Basins of the CAO are extremely low. Model studies also conclude that the production potential of the region is too low to warrant commercial fisheries in the future 2,7 .

Material and methods
Trawl specifications. Two pelagic trawls were used and modified to reduce damage from the ice at the surface. The standard Norwegian pelagic trawl for pelagic fish (capelin, herring and juveniles) in northern areas (Harstad 320 trawl 75 23 . Prior to each deployment, the upper wings of the trawl were mounted together with thin rope to prevent ice from entering the trawl during shooting. This rope broke when the trawl doors were deployed due to the horizontal spread forces. Vessel speed was reduced during shooting of the trawl and the door release to increase the descent rate of the trawl (i.e., the trawl was hanging nearly vertically behind the vessel). When the doors were at approximately 4-6 m depth, the vessel increased its speed and the warps were deployed as during ordinary pelagic trawling. The Harstad trawl was towed at 1.5 ms −1 while the macroplankton trawl was towed at 1.0 ms −1 (speed over ground-SOG) for ~ 30 min at different depths (Table 1), which is the same speeds as used in ice-free waters. As the trawl gear approached the surface during haul back, the vessel reduced its speed to allow the doors and trawl to come on deck at a vertical angle close to the stern of the vessel (i.e., the trawl was hanging nearly vertically behind the vessel). The vessel used Seaonics ice trawl gallows (automated system to move the tow blocks into the centre of the stern of the vessel) when the sea ice thickness and volume was high. The vertical opening of the trawl net (Harstad trawl only), trawl depth, door spread and door depth were measured with acoustic trawl instrumentation (SCANMAR AS, Åsgårdstand, Norway).
Biological data. Trawl catches were sorted immediately, and organisms were identified to species level when possible. Fish were weighed and length measured. Excess water was gently filtered from the macrozooplankton (euphausiids, amphipods, etc.) samples using a 1 mm sieve. Representative subsamples of 100 g were preserved in 4% hexamine-buffered formalin solution in 500 ml plastic bottles. The preserved samples were species identified length measured, when possible, at the IMR laboratory. Institute of Marine Research fully adheres to Norwegian laws relevant to Ethics in Science as well as Animal Welfare. Institute of Marine Research is a governmental research institute with given permission to perform research cruises including fish samplings by the Norwegian Government. All methods were carried out in accordance with relevant guidelines and regulations. All methods are reported in accordance with ARRIVE guidelines. No experiments were carried out on live organisms.
Acoustical data. Acoustical data were obtained from a Simrad EK80 echosounder mounted behind an ice window under the hull. The echosounder was calibrated according to standard procedures 78 on 19.01.2021. Only data from periods when the ship was stationary were used due to mechanical noise when the vessel was moving through sea ice. An hour of acoustic data without noise was selected from as near the trawl locations as possible (Fig. 1c, Table S1). Applying a threshold of − 90 dB gave echosounder data down to at least 500 m for the frequencies 18 and 38 kHz. The usable range was less for the higher frequencies (for 70 kHz about 400 m, for 120 kHz about 200 m, for 200 kHz about 100 m, and for 333 kHz less than 50 m).
The echo integrator threshold in terms of S v in dB was set at − 90 dB re 1 m −1 (ref. 50 ). The 38 kHz frequency was used to determine backscatter intensity. Analysis was done in the postprocessing tool LSSS v. 2.11.0 (ref. 79 ). The backscattering data output were in the form of s A , and all definitions and notations of the acoustic quantities www.nature.com/scientificreports/ were according to Maclennan 50 (see Supplementary section 3 for more details). The data were integrated over the time period and stored at a grid of 10 m depth and 1 h (3600 s).
To illustrate how the total acoustic backscatter was distributed along an axis from weak scatterers to strong scatterers, the s A was classified into five classes according to their s V by sequential thresholding at 38 kHz. The five categories used were S70, S75, S80, S85 and S90 (Supplementary section 3), where S70 represent the organisms with strongest target strength (TS) and/or the highest density, while S90 represent the organism with weakest TS and/or the lowest density.
Various types of organisms of the same size often have different acoustic properties at different frequencies, and the distribution of types of scattering organisms was assessed using frequency response in s V . The different frequencies were considered down to their acceptable ranges, and we give the relative frequency response between 18 and 38 kHz (rf 18/38 ) when feasible.
We identified type of organisms present by measurements of TS of individual scatterers. We mainly considered TS measurement from the 38 kHz, but TS from 18 kHz were also assessed when available. The higher frequencies were only considered when we recorded extraordinary registrations (like at NB1). The TS measurements were identified in LSSS using the following settings: Detector type SED (Single echo detector), Min TS − 70 dB, Pulse length determination Level: 6 dB, Minimum echo length: 0.5 (relative to pulse length), Maximum echo length 2.0 (relative to pulse length), Max gain compensation 6 dB, Phase Deviation Check: On. Max phase deviation: 8 steps. Thereafter, we filtered the data restricting them to be within 2° off axis to avoid TS distributions being biased towards higher values due to reduced signal-to-noise ratio in the outer parts of the beam.
Environmental data. Both sea ice concentration and drift during the cruise period were obtained from the EUMETSAT Ocean and Sea Ice Satellite Application Facility (OSI SAF), in collaboration with ESA CCI (https:// osi-saf. eumet sat. int/ produ cts/ sea-ice-produ cts). Daily sea ice concentration from passive microwave satellite data is from the interim climate data record (OSI-430-b 80 ) and is provided daily on a grid with a resolution of 25 km. Sea ice drift velocities are from the low-resolution sea ice drift product (OSI-405-c) and are daily at a resolution of 62.5 km.
Observations of ocean currents were obtained from the vessel mounted RDI 150 kHz Acoustic Doppler Current Profiler (ADCP) mounted behind the ice window under the hull. Post-processing of ADCP data was performed using the Common Oceanographic Data Access System 81 . Tides were removed using the AOTIM tidal model 82 . The vessel mounted ADCP returned data from 25 to 250-350 m depth. Relatively continuous ADCP data were recorded during transit up to about 82.5 o N. To the north of this, VM-ADCP only delivered data when the vessel was stationary. To keep the high resolution when crossing the Atlantic Water slope current, ADCP data from 80.5 o N to 82.5 o N were gridded to an along-track vector with 10 km horizontal resolution. To the north of this, ADCP data were extracted from each CTD station (within 0.5 km of the station), and then averaged to obtain a vertical profile of velocity at the CTD station. This returned velocity along a vector with 10 km horizontal resolution up to 82.5 o N and the resolution of the CTD stations to the north of this.
Temperature and salinity were measured using a Seabird 911plus CTD at and between the trawl stations (Figs. 1c, 4d, e). The CTD was equipped with a WET Labs Fluorometer and a rosette system for collecting water samples. The conductivity, temperature, depth, and oxygen sensors are serviced and calibrated once a year by the manufacturer (Seabird). In situ water samples for salinity calibration (conductivity sensor) were taken at every station at maximum depth.

Data availability
The cruise data analysed in the study are available at Norwegian Marine Data Centre. CTD data can be found at https:// doi. org/ 10. 21335/ NMDC-18141 68447, ADCP data at https:// doi. org/ 10. 21335/ NMDC-11755 79976, trawl catch data at https:// doi. org/ 10. 21335/ NMDC-19465 1742 and acoustic data at https:// doi. org/ 10. 21335/ NMDC-12759 35147. Sea ice data are available at https:// osi-saf. eumet sat. int/ produ cts/ sea-ice-produ cts. www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.