Marine mammal hotspots across the circumpolar Arctic

Identify hotspots and areas of high species richness for Arctic marine mammals.


| INTRODUC TI ON
Climate change impacts have been documented in most ecosystems on Earth (e.g. IPCC, 2014;Scheffers et al., 2016). Shifts in species distributions and core habitats are ongoing and are predicted to continue throughout this century (Foote et al., 2013;Hazen et al., 2013;Poloczanska et al., 2016). These shifts are expected to alter trophic dynamics, cause mismatches between resource availability and consumers, alter species interactions and lead to species extirpations or extinctions (IPCC, 2014;Post et al., 2013).
Impacts are likely to be especially severe in the Arctic, where air temperatures are rising 2-3 times faster than the global average and sea-ice extent is declining precipitously; a seasonally sea-ice free Arctic is predicted within a few decades (Meredith et al., 2019;Wang & Overland, 2012). Ongoing environmental change, particularly the loss of sea ice, has already caused considerable change in Arctic marine ecosystems (Meredith et al., 2019;Post et al., 2013). Sea-ice-associated species may have limited ability to shift their ranges in response to the ongoing changes (IPCC, 2014).
Sea-ice loss represents direct habitat loss for ice-associated marine mammals, but such losses will likely also have many indirect effects through changes in their prey bases, increased presence of invasive species, temperate competitors and predators, altered disease risks and a variety of pressures associated with expanding human activities (IPCC, 2014;Meredith et al., 2019;Van Wormer et al., 2019). Due to low functional redundancy within Arctic ecosystems, impacts from species extirpations or range shifts are likely to be more severe in the Arctic than in more species-rich ecosystems (Post et al., 2009). Benchmark data on core habitats and migratory pathways are critical for detecting changes, performing risk assessments regarding impacts of human activities and informing spatial planning of protected areas.
Arctic marine mammals are dependent on sea ice for critical phases of their life cycles and their position near the top of the Arctic food web, in combination with being long-lived and slow to reproduce, means that they are sensitive to rapid changes in their environment (Kovacs et al., 2021;Reid et al., 2013;Tynan & DeMaster, 1997). Based on the boundaries defined by the and polar bear (Ursus maritimus) (Kovacs et al., 2011;Laidre & Regehr, 2018). Four species of ice-associated seals, the hooded seal (Cystophora cristata), harp seal (Pagophilus groenlandicus), ribbon seal (Histriophoca fasciata) and spotted seal (Phoca largha), use drifting sea ice as a resting, pupping, nursing and moulting platform during winter and spring, but are generally found in open water or in subarctic areas for the rest of the year (Kovacs et al., 2011;Laidre & Regehr, 2018). Additionally, the harbour seal (Phoca vitulina) and grey seal (Halichoerus grypus), which are generally considered temperate species, do have populations that reside year-round within the CAFF Arctic boundary.
One of the main challenges for evaluating environmental changes and their impacts in the Arctic marine ecosystem in a holistic manner is that the available data are spread across many sources and generally analysed at the species level using a wide range of methods. Therefore, large overall changes affecting multiple species may go undetected or are underrated. To address this data gap for marine mammals, biotelemetry data from 13 marine mammal species (collected by 33 scientific institutes) were synthesized to identify species hotspots and areas with high species richness across the circumpolar Arctic. Habitat features of the identified hotspots were also analysed to investigate the relative importance of different features and how these vary seasonally and across Arctic regions for individual species. Large-scale syntheses of biotelemetry data in other ecosystems have been successful in identifying multi-species hotspots, important ecosystem components and the magnitude of threats within these areas (e.g. Block et al., 2011;Hindell et al., 2020;Queiroz et al., 2019). This synthesis of Arctic marine mammal biotelemetry data provides information vital for: (1) determining the environmental and ecological drivers that shape Arctic marine ecosystems; (2) marine spatial planning for protected areas; (3) conducting environmental risk assessments that include cumulative effects for marine mammal populations (i.e. evaluating the overlap between important habitats and present or proposed human activities); and (4) identifying knowledge gaps.

| Study area and species
The study area comprised the circumpolar Arctic as defined by CAFF (Figure 1; CAFF, 2013). Biotelemetry data from all marine mammal species that spend most of their annual cycle within the CAFF Arctic boundary were included. Data came from 2115 biologging devices that were deployed on 13 species in the period 2005-2019 (Table 1). This time frame was chosen to represent recent species distributions during a period of rapid change in the Arctic.

| Locations
Data-handling and statistical analyses were conducted using R version 3.5.3 (R Core Team, 2019). Biologging devices provided ARGOS (CLS, 2016) or Fastloc GPS locations that were filtered to remove unlikely locations using the Douglas Argos-Filter (Douglas et al., 2012;walruses in East and West Greenland)   walk model (CTCRW model;crawl package;Johnson et al., 2008). Table 1 and in the Supporting Information provide details regarding capture methods, the specific type of biologging devices used, research permits and protocols for each deployment. A tag had to transmit locations for at least 14 days to be included in the analyses. Locations every 12 hours were predicted by the CTCRW models for each animal. Some biologging devices were duty-cycled (see Table 1); only days when the tag was transmitting were used in the analyses. CTCRW model-generated locations that were within transmission gaps greater than seven days were also removed from the analyses. Locations on land were moved to the closest in-water location in time for each species with the exception of polar bears.

References in
Given the grid size used in this analysis (30 x 30 km) and the large geographic extent (circumpolar), moving these locations had little or no effect on the hotspot locations.

| Marine mammal hotspots
Getis-Ord G i * hotspots (hereafter referred to as G i * hotspots/ statistic) were calculated for each species or species group (i.e. all pinnipeds, all cetaceans, and all species) to identify areas with higher use by marine mammals (Getis & Ord, 1992;Ord & Getis, 1995). The G i * statistic measures the concentration of a variable at a point by comparing the local sum of values (i.e. a point and its neighbours within a specified distance) to an expected sum (i.e. random permutations drawn without replacement from all points in the dataset). A statistically significant positive z-score is assigned if the calculated local sum is larger than the expected sum and the difference is too large to be the result of random chance (Getis & Ord, 1992;Ord & Getis, 1995). The G i * statistic was cal-   Table S1; Bivand & Wong, 2018;Ord & Getis, 1995 and current scientific knowledge) or where movement patterns and space use of animals was generally separate from animals tagged at adjacent tagging locations (see Table 1). Each individual, regardless of any existing finer stock structure, was given an equal weighting The amount of overlap between high and null hotspot levels (95 and 99%) and high and null levels of species richness index areas (≥4 species) were calculated.

| Hotspot habitat
To explore the similarities and differences in the habitat features within marine mammal hotspots across the circumpolar Arctic, principal component analysis (PCA) was conducted on standardized habitat variables extracted from the G i * hotspots (70-99% statistical significance) for each species (prcomp function). These variables included distance to the coast (km), distance to the nearest tidewater glacier front (km), water depth (m), sea-ice frequency (% of days during a month that an area had sea ice), sea surface temperature (SST; °C) and distance to the nearest polynya (km; winter only).

Sea-ice frequency and SST values were extracted from March and
September means (2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018)(2019). March and September correspond to the periods of maximum (winter) and minimum sea-ice extent (summer) throughout most of the circumpolar Arctic. It is important to note that using a 14-year mean of monthly sea-ice data may not reflect the sea ice available to the animals during the time frame of their respective biotelemetry device deployments. However, this index should be sufficient for identifying broad-scale patterns in sea-ice use. While these variables may not cover all possible environmental conditions of potential importance, they are highly relevant and have been used frequently in previous habitat modelling for marine mammals (e.g. Cameron et al., 2018;Laidre et al., 2015;Matthews et al., 2020). Because preliminary analysis showed that SST had very little influence on the PCA plots, this variable was excluded from further analyses. The range of each species was divided into three regions (where applicable) as data exploration showed large regional differences in hotspot habitats for some species. The regions were defined as: (1)

| RE SULTS
A total of 400,460 tracking days from 2115 biologging devices were available for 13 marine mammal species from 2005 to 2019. Data were available for most of the year for most species and regions, although exceptions did occur, notably for walruses and white whales because winter data were sparse (see Table 1). Discrepancies between the IUCN global range and hotspot locations also identify spatial data gaps for each species (Figures 2-5 and Table 1). Age and sex were not included in these analyses due to species and regional differences in individuals that were tagged. For example, the majority of tagged polar bears were female polar bears, all walruses in Svalbard and the Pechora Sea were male walruses while most of the walruses from the Bering-Chukchi-Beaufort region (BCB) were female walruses, the majority of bearded seals from the BCB were juveniles and all harp seals tagged in the Greenland Sea and all grey seals tagged in Iceland were pups.

| Ringed seals
Hotspots occurred in coastal and offshore areas around the circumpolar Arctic (Figures 2, S1, S9,

| Bearded seals
Hotspots occurred in coastal regions of northwest Svalbard and in the BCB (in the northern Bering Sea (including Norton Sound and F I G U R E 2 Getis-Ord G i * hotspots for ringed seals, bearded seals and walruses in the circumpolar Arctic during the summer (Jun-Nov) and winter (Dec-May) based on the number of individuals per grid cell. Increasing intensities of red indicate hotspots of increasing level of significance. The red dotted polygon shows the global range of the species (IUCN Red List) Bering Strait) and in the eastern Chukchi Sea (including Kotzebue Sound); Figures 2, S1, S9, Table 1). Winter hotspots were slightly further south and more offshore than summer hotspots in the BCB ( Figure 2). The null model hotspots occurred in the vicinity of where bearded seals were tagged and overlapped high hotspot levels (95 and 99%) by 32% (Figures S15, S16).

| Walruses
Hotspots occurred in shallow regions along the northern and south-

| Bowhead whales
Hotspots occurred in the Greenland and northern Barents Seas be-  further north (only Baffin Region) than winter hotspots (Figure 3).

| Polar bears
Summer hotspots occurred slightly further north than winter hotspots in most regions (Figure 4). Additional hotspot data were available from Yurkowski et al. (2019) for the Canadian Arctic (i.e. orange hotspots in Figures 4, S4, S11). Null model hotspots occurred in the vicinity of where polar bears were tagged and overlapped high hotspot levels (95 and 99%) by 58% (Figures S15, S18).

| Spotted seals
Hotspots in the BCB region were generally offshore in the central Kotzebue Sound, in summer (Figures 4, S5, S11, Table 1). Coastal hotspots also occurred along the Bering Strait region and the northwest Alaska coast in the Chukchi Sea ( Figure 6). Winter hotspots were generally further south and further offshore than summer hotspots ( Figure 4). Null model hotspots were found near tagging locations for spotted seals and overlapped high hotspot levels (95 and 99%) by 31% (Figures S19, S20).

| Ribbon seals
Hotspots occurred in the Bering Sea, including in the Gulf of Anadyr (Figures 4, S5, S11, Table 1). Hotspots were generally offshore, except for in the Gulf of Anadyr (Figures 4, S11). Null model hotspots occurred in the vicinity of where ribbon seals were tagged and overlapped high hotspot levels (95 and 99%) by 63% (Figures S19, S20). Hotspots based on the number of locations were also found in southern portions of the Norwegian Sea and southeast of Iceland ( Figure   S6). Summer hotspots generally occurred further north than winter hotspots ( Figure 5). Null model hotspots occurred in the vicinity of where hooded seals were tagged and overlapped high hotspot levels (95 and 99%) by 23% (Figures S19, S20).

| Harp seals
Hotspots occurred in the MIZ of the Greenland Sea, northern Barents Sea and around the Svalbard Archipelago ( Figures 5, S6, S12, Table 1). Hotspot maps based on the number of locations also occurred south of Svalbard and in fjords in western Svalbard ( Figure   S6). Summer hotspots were located further north than winter hotspots ( Figure 5). Null model hotspots occurred in the vicinity of where harp seals were tagged and overlapped high hotspot levels (95 and 99%) by 25% ( Figures S19, S20).
Summer and winter hotspots were located in similar areas in the North Atlantic ( Figure 5). Null model hotspots occurred in the vicinity of where harbour seals were tagged and overlapped high hotspot levels (95 and 99%) by 86% ( Figures S19, S21).
Summer and winter hotspots were located in similar areas around northern Iceland ( Figure 5). Null model hotspots occurred in the vicinity of where grey seals were tagged and overlapped high hotspot levels (95 and 99%) by 75% (Figures S19, S21).

| Species groups
Hotspots calculated for species groups (all pinnipeds, all cetaceans and all species) and the species richness index highlight regions across the circumpolar Arctic that are important for Arctic marine mammals (Figures 6, 7, S4, S8, S13). These included predominantly continental-shelf habitats in the Svalbard Archipelago, the MIZ in the northern Barents and Greenland Seas, East Greenland, regions around Baffin Island, Foxe Basin, much of the BCB, as well as areas within the Kara and Pechora Seas (Figures 6, 7, S13). Summer hotspots were generally found further north than winter hotspots in the BCB and in the Canadian Arctic Archipelago (Figures 6, 7). Winter data for some cetacean species were quite limited.
Particularly, high values of species richness were found in the "Arctic gateways" in the North Atlantic (Fram Strait) and North Pacific (Bering Strait) (Figures 7, S14). Overlap between high levels of species richness (≥4 species) and high hotspot levels (95 and 99%) was generally quite high, especially during the winter (Figures 7, S14).

| Hotspot habitat
Large regional differences in the habitat features within hotspots for marine mammals occurred across the circumpolar Arctic (Figures 8,   9, Tables S2, S3). Hotspots in Svalbard, Greenland and the eastern Canadian Arctic Archipelago were generally in areas with tidewater glacier fronts, whereas this habitat is absent in the BCB (Figures 8, 9, Table S2). Sea-ice frequency was higher in the winter hotspots than the summer hotspots for all species, although a large range of sea-ice frequencies were used during the winter period (Table S2). Hooded, harp, ribbon, harbour and grey seals were generally found in areas with less sea ice than the other marine mammal species during the winter months (Figures 8,9, (Figures 8, 9, Table S2). Positive water depths (i.e. land) reflected tight coastal distributions in some regions for some species (Table S2). Most marine mammal species were generally further away from polynyas in the winter in the Canada/West Greenland region than the other regions (Figures 8, 9, Table S2).

| DISCUSS ION
Arctic marine mammal hotspots and areas of high species richness occurred across the Arctic continental-shelf seas and in the MIZ, in regions previously identified as important habitats for Arctic marine mammals (Citta, Lowry, et al., 2018;Hamilton et al., 2020;Yurkowski et al., 2019). The seasonal presence of sea ice is a defining feature for most hotspot areas identified, with most Arctic marine mammal hotspots having >10% sea-ice frequency during both the summer and winter periods. Sea ice serves a variety of functions for Arctic marine mammals: it is a pupping, nursing and moulting platform for seals; a hunting platform and transport corridor between hunting and maternity denning areas for polar bears; and a resting platform, near benthic foraging areas, for walruses. It also offers protection from storm events and aquatic predators and ice edges are important foraging areas for all species (Kovacs et al., 2011;Laidre et al., 2008). Seasonally sea-ice-covered regions have high primary production from ice-algae and phytoplankton blooms that begin under the ice in spring, providing nutritional support for zooplankton, benthic fauna and fish upon which marine mammals depend (Ardyna et al., 2020;Kovacs et al., 2011;Sakshaug, 1997 For example, the winter period includes over-wintering behaviour, birth and nursing (either entirely for most pinnipeds or a proportion for cetaceans and polar bears) for most species; different areas and habitats are likely favoured for different life-history events. Another caveat is that in some regions, multiple stocks exist for some species and biotelemetry data are usually not evenly split among the different stocks (e.g. white whales in the BCB region). In these cases, hotspots will be biased towards the areas that are used by the stocks with the most tracking data. Logistical challenges when undertaking fieldwork, and the variable ecologies of the ages and sexes in many marine mammal species, also means that tracking datasets usually do not contain an equal representation of the different age classes and sexes. In extreme cases, only one age class or one sex has been tagged (e.g. Greenland Sea harp seal data includes only pups, the vast majority of tagged polar bears are females). Thus, extrapolating results to the species level should be done with caution.
Areas with high species richness were generally found within high hotspot levels for all species (overlap >80%). Because each species-region combination received an equal weight in the all species analysis, some regions (i.e. western Hudson Bay, Kara Sea), had high hotspots levels but low species richness. These regions are known to be important marine mammal areas, but biotelemetry data were lacking for them. Similarly, there were a few regions with high species richness that did not have high hotspots levels. This discrepancy is likely the result of areas that are used by multiple species for transit or only used intensively by subsets of the tagged populations.
The species richness and hotspot results should be used in tandem for conservation and management purposes.
The two "Arctic gateways" in the north Atlantic and Pacific oceans were especially species rich and they generally had high hotspot levels (also see CAFF 2017). These regions transport heat, nutrients and plankton into the Arctic Ocean (Basedow et al., 2018) and are also regions with rich fish stocks (Christiansen et al., 2014).
Both regions are experiencing warming trends and exhibiting reductions in Arctic ecosystem components (Fossheim et al., 2015;Huntington et al., 2020). Regional differences in hotspot habitats were also found in many cases, likely reflecting not only the different environmental features present but also the ecosystem differences F I G U R E 8 PCA plots for habitat variables in Getis-Ord G i * hotspots (70%-99%) for nine pinniped species (bearded seal, ringed seal, walrus, grey seal, harbour seal, harp seal, hooded seal, ribbon seal and spotted seal) in the circumpolar Arctic during the summer (Jun-Nov) and winter (Dec-May). The circumpolar range of each species was split into regions (BCB: Bering-Chukchi-Beaufort region; C: Canadian Arctic Archipelago and West Greenland; GS: East Greenland and Barents Sea region) across the circumpolar Arctic (e.g. Bluhm et al., 2015;Stenson et al., 2020). This indicates that environmental changes might have different impacts on species' distribution and behaviour across the circumpolar Arctic. The long-term impacts of these ecological trends on Arctic marine mammals, concomitant with other threats including increased presence of temperate marine mammal species, increased levels of human activities and changing trophic interactions are concerning.
Recurrent polynyas overlapped many marine mammal hotspots in the winter period ( Figure 6). Polynyas can be important for marine mammals as over-wintering and foraging areas; at times thousands of marine mammals from multiple species occupy these areas (Born & Knutsen, 1992;Kovacs et al., 2020;Laidre et al., 2008). The hotspots of some species, such as bearded seals, bowhead whales, white whales and spotted seals, were closer to polynyas than hotspots of other species explored herein, although regional differences in polynya use were apparent. Several recurrent polynyas did not overlap with identified hotspots, despite these areas being known to be im- Tidewater glacier front habitat is important for several Arctic marine mammal species, but is prevalent only in some Arctic regions, including Greenland, Svalbard, Franz Josef Land and Novaya Zemlya in the Russian Arctic (Laidre et al., 2016;Lydersen et al., 2014).
Correspondingly, distance to tidewater glacier fronts was a defining feature of regional differences in hotspot habitat. Close association with tidewater glacier fronts was identified in the hotspots of many Tidewater glacier fronts are also important pupping and denning areas for ringed seals and polar bears (in some locations) and provide hunting areas for polar bears in the spring (Freitas et al., 2012;Laidre & Stirling, 2020;Lydersen et al., 2014). In the BCB, where there are no tidewater glaciers and also in other Arctic regions, ringed seals use shore-fast ice for pupping and polar bears use sea ice, barrier islands and land for denning (Crawford et al., 2012).
The  Sea-ice declines and associated environmental changes are likely the largest current threat to Arctic marine mammals. These threats may operate directly through the loss of birthing, nursing and resting areas or transport corridors, or indirectly through changes in space use, prey composition, abundance and distribution, or the presence of interspecific competitors and predators (Kovacs et al., 2011;Laidre et al., 2008;Matthews et al., 2020;Reid et al., 2013;Stenson et al., 2016). Large-scale sea-ice declines and ecosystem changes are underway in many of the hotspot areas identified in this study (e.g. Fossheim et al., 2015;Huntington et al., 2020;Vihtakari et al., 2018;Yurkowski et al., 2018). Arctic marine mammals have reacted to past glacial and interglacial periods by changing their distributions (Foote et al., 2013;Harington, 2008;Louis et al., 2020). However, there is a limit to how far Arctic marine mammals can shift their distribution northward into the Arctic Ocean Basin given their strong affiliation with the highly productive continental-shelf habitats demonstrated by our hotspots analyses. The potential for northward shifts is likely variable among species and between regions. Productivity in the Arctic Ocean Basin is lower than the Arctic continental-shelf seas and it is unknown how productivity and fish distribution will change as the Arctic continues to warm.

| Future research needs and recommendations
Global climate predictions suggest that many high-use areas identified herein will likely become less favourable habitat during this century, which is likely to result in distributional shifts and to impact the long-term persistence of Arctic marine mammals (e.g. Durner et al., 2009;Øigård et al., 2014;Reimer et al., 2019). Ideally, biotelemetry studies should be expanded across the distributional range of species with sample sizes that will document movement and behaviour patterns of various age and sex groups in each region to ensure that important areas for critical life stages are identified (e.g. Fortune et al., 2020;.
A gap analysis is implicit in the identification of Arctic marine mammal hotspots, because discrepancies between hotspots and IUCN ranges highlight where, and for which species, more research is required. This comparison also highlights regions where IUCN species ranges need to be revised because biotelemetry data extend beyond the IUCN borders of species occupancy in some cases. More biotelemetry data are needed particularly for the Russian Arctic.
For example, identification of the western Chukchi Sea and East Siberian Sea as low use areas for polar bears is likely due to limited tagging effort. Data are also lacking for some species in East and West Greenland, the eastern Svalbard Archipelago and for some areas within the Canadian Arctic Archipelago. Additionally, data are lacking from some stocks in the larger regional areas (e.g. eastern Chukchi belugas in the BCB), influencing the identification of hotspot locations in these regions. Numerous species also require more tagging effort; limited data are available for bearded seals, white whales, walruses, narwhals and harp seals. Additional tagging efforts should also try to attain greater seasonal coverage for some species (see Table 1). For species that generally have had short attachment durations (e.g. whales, walruses in some regions), portions of the year with sparse or no data limits the detection of seasonal hotspots. Location data are also generally lacking for the moulting periods of seals, an energetically costly period, when seals may be particularly vulnerable. Addressing these gaps will require innovative new attachment methods. Efforts should also be made to continue time series (including re-tagging individuals when possible) to quantify changes in behaviour, distribution and habitat use due to climate change. Integration of behavioural data (e.g. dive and activity data) from the tags in future large-scale studies along with complementary data from many sources (e.g. diet, body condition, prey availability, passive acoustic data) will undoubtedly improve our understanding of marine mammal habitat use and how it is likely to be impacted by climate change and concomitant changes in human activities. In some Arctic regions, Indigenous Knowledge will provide information to help fill identified knowledge gaps (Gryba et al., 2021;Huntington et al., 2017;Loseto et al., 2018).
Effective conservation and management measures for Arctic marine mammals rely on comprehensive data from biotagging initiatives and syntheses across species to identify important areas. Minimizing man-made stressors may reduce the overall impacts and allow some species time to adjust to environmental changes. Protected areas, including Marine Protected Areas, nature reserves and national parks, need to be expanded and protecting areas that extend across international boundaries should be considered (Hussey et al., 2016).
Numerous examples exist of where biotelemetry data and identified hotspot regions have influenced policy and management decisions, including defining protected areas, reducing the risk of vessel strikes and in risk assessments for proposed human activities (e.g. Hays et al., 2019). Greater use of movement data is needed to help protect species against the variety of threats they face, including complications arising from transboundary management (e.g. Hays et al., 2016Hays et al., , 2019Titley et al., 2021). Given the current rate of change in the Arctic, spatially and temporally dynamic protected areas should be evaluated (D'Aloia et al., 2019;Hyrenbach et al., 2000). Specific examples where dynamic protected areas might be most effective include avoidance of disturbance to pinniped pupping, nursing and moulting areas during these life-history events, polynyas during winter and spatial adjustments relative to large intra-and inter-annual variations in the MIZ. Monitoring marine mammal use of hotspots through biotelemetry in near-real time could support dynamic boundary delimitations and active management (Hobday et al., 2011;.

| CON CLUS IONS
The Arctic continental-shelf seas and MIZ were identified as regions with a high density of hotspots and high species richness.
The "Arctic gateways" of the North Pacific and North Atlantic were particularly species rich. Hotspots differed by species, but some common areas were identified, such as coastal areas around the Svalbard Archipelago, the East Greenland continental shelf, waters surrounding Baffin Island and coastal and continental-shelf areas throughout the BCB region. Habitat features of marine mammal hotspots differed seasonally, regionally, within and among species.
Environmental changes associated with sea-ice declines and increases in human activity are currently taking place in many of the identified hotspots. Biotelemetry research is needed in regions and on populations where data are lacking. Efforts also need to be directed to continuing telemetry time series and incorporating behavioural data and Indigenous Knowledge where applicable. Although there are regional, sex and age gaps in the data at hand, the hotspots identified in this study do represent key marine mammal areas that can serve as a benchmark for spatial management to mitigate anthropogenic disturbances and reduce stress on marine mammal populations. Arctic marine mammals are currently facing multiple threats and the findings of this study can inform management efforts to help mitigate pressures related to these threats and assist this species group in adjusting to future environmental changes.