Polar bear (Ursus maritimus) COSEWIC assessment and status report: chapter 4

4. Habitat

4.1 Habitat Requirements

The physical attributes of sea ice are the primary determinants of the quality of polar bear habitat. Changes in sea ice and associated snow cover affect light transmission and thermodynamic processes important to lower trophic levels of the arctic marine ecosystem (Welch et al. 1992; Barber et al.1995), which, combined with kinematic or topographic characteristics of sea ice, influence the distribution of ringed seals (Stirling and Lunn 1997; Barber and Iacozza 2004). In the Canadian Arctic, polar bear habitat is closely associated with that of the ringed seal (Stirling and Øritsland 1995) and includes areas of consolidated pack ice, areas immediately adjacent to pressure ridges, between multi-year and first-year ice floes, and at the floe edge between marginal and landfast sea ice (Stirlinget al. 1982; Kingsley et al. 1985; Stirling and Derocher 1993; Stirling et al. 1993; Ferguson et al. 2000a). Seals are hunted through breathing holes, in birth lairs, or when hauled out on ice (Stirling and Archibald 1977; Smith 1980). Bearded seals, harp seals (Pagophilus groenlandica), spotted seals (Pusa largha), hooded seals (Cystophora cristata), walrus (Odobenus rosmarus), beluga whales (Delphinapterus leucas), and narwhal (Monodon monoceros) also feature in the diet of polar bears (Stirling and Archibald 1977; Kiliaan and Stirling 1978; Fay 1982; Lowry et al. 1987; Calvert and Stirling 1990; Smith and Sjare 1990; Derocher et al. 2002); however, scientific knowledge and ATK suggest it is the young ringed seal hunted at its subnivian den that is most important to the majority of polar bears (Stirling and Archibald 1977; Smith 1980; McDonald et al. 1997). Ringed seals, which live exclusively in association with sea ice for at least part of the year (as do bearded and harp seals), have apparently been the principal prey of polar bears for much of their co-evolutionary history, and many ringed seal behaviours appear to be adaptations to avoid predation by polar bears (Stirling 1977; Amstrup 2003). Changes in the distribution of ice-dependent phocids in response to climate warming is certain to impact the distribution of polar bears (Stirling and Derocher 1993; Barber and Iacozza 2004; Derocher et al. 2004).

Because the sea ice provides access to their main prey species, the distribution of polar bears in most areas changes with the seasonal extent of sea-ice cover. Amstrup et al. (2007) and Durner et al. (2007) discuss the different types of ice used and preferred by polar bears, including ecoregions described by divergent, convergent, archipelago, and seasonal ice conditions. Throughout the polar basin and into the Canadian Arctic Archipelago, polar bears spend their summers concentrated along the edge of the persistent pack ice. Significant northerly and southerly movements appear to be dependent on seasonal melting and refreezing of ice near shore (Amstrup et al. 2000). In other areas (Hudson Bay, Foxe Basin, Baffin Bay, Davis Strait, Hudson Bay, James Bay, and portions of the Canadian High Arctic) polar bears are forced onto land (summer retreat areas) for several months during the open water season while they wait for new ice to form (Amstrup et al. 2007).

If forced on land for summer due to lack of sea ice (50–60% of the Canadian population), polar bears vary in their habitat selection, often by sex and age group, with males displacing females and cubs inland and away from the coast (Stirling et al. 2004). Food may not be consumed, and bears may rely entirely on fat reserves (Derocher and Stirling 1990). In some areas (e.g., northeast Manitoba, Derocher et al. 1993; Davis Strait, M.K. Taylor, Department of Environment, Government of Nunavut, pers. obs.), polar bears have been observed to feed on blueberries (Vaccinium uliginosum) and crowberries (Empetrum nigrum). On occasion, polar bears may also depredate nests of waterfowl (e.g., Smith and Hill 1996) and have been observed to kill caribou (e.g., Derocher et al. 2000; Brook and Richardson 2002). In Labrador, feeding on salmon by bears has also been observed (Brazil and Goudie 2006). Whale carcasses attact large numbers of bears during the open-water season (Kalxdorff 1997; Perham 2005). The attraction of bears to garbage during the ice-free season is of major concern to the management of polar bears and human safety in the Arctic (Lunn and Stirling 1985). As the ocean freezes again in late autumn, bears that were trapped on land redistribute themselves throughout subpopulation ranges, except for pregnant females, which excavate maternity dens (Section 5.1).

4.2 Trends in Habitat

Trends in habitat as they relate to the status of polar bears focus on impacts of climate warming, particularly spatial and temporal trends in the types and extent of sea ice, including length of the open-water season. Climate-change-related trends in conditions of terrestrial habitat, including denning habitat (e.g., Obbard and Walton 2004), must also be taken into consideration; however, effects of climate warming on conditions of sea ice are most important to the status of the species. Climate change is modifying the dynamics of sea ice formation and distribution in the Arctic, and it is expected that amounts of multi-year sea ice will be reduced and that ice will continue to trend toward a predominance of thinner, annual ice formations. These changes are widely documented by both western science and through ATK. Several sources highlighting ATK of polar bears report Inuit concerns about deterioration of sea ice conditions and their impact on polar bears (Atatahak and Banci 2001; Dowsley 2005; Keith et al. 2005; NTI 2005; Nirlungayuk 2008). These conditions include the disappearance of multi-year ice and icebergs, which polar bears use as feeding and resting platforms. Other changes include thinner ice, more rough ice, and earlier spring break-up, which may reduce polar bear hunting efficiency.

The literature on climate change and loss of sea ice is constantly being updated, and this report presents only the most relevant summary results at the time of writing. The largest and most recent effort to summarize scientific observations of changes in sea ice in the Arctic was conducted by Lemke et al.(2007) in their collaborative chapter as part of the 2007 report of the Intergovernmental Panel on Climate Change (IPCC). Briefly, satellite data indicate a 2.7 ± 0.6% per decade decline in annual mean arctic sea ice extent observed since 1978. The decline for summer extent is larger than for winter in the Arctic, with the summer minimum declining at a rate of 7.4 ± 2.4% per decade since 1979. Some data indicate that the summer decline began around 1970. Submarine-derived data for the central Arctic indicate that the average thickness of sea ice in the central Arctic has very likely decreased by up to 1 m from 1987 to 1997. Model-based reconstructions support this, suggesting an arctic-wide reduction of 0.6 to 0.9 m over the same period. Large-scale trends prior to 1987 are ambiguous. In Western Hudson Bay, where variables of climate warming have recently been used to explain variation in survival rates of polar bears (Regehr et al. 2007a; see also below and Sections 6.1 and section7.10), analyses of regional climate data have shown that between 1950 and 2000, mean air temperatures in April, May, and June have warmed at a rate of 0.3–0.8°C per decade (Skinner et al. 1998; Gough et al. 2004; Ferguson et al. 2005; Gagnon and Gough 2005a,b). For example, April–May temperatures increased from a mean of -12.4°C in 1962 to -9.8°C in 2000 (Ferguson et al. 2005). From 1979 to 2004, spring break-up as measured from ice concentrations (50% ice: 50% water) shifted from late June to late May, an average change of -0.75 ± 0.25 (mean ± 1 SE) days earlier each year (Stirling and Parkinson 2006). Stirling and Parkinson (2006) demonstrated similar trends in earlier timing of break-up for Foxe Basin (-0.58 ± 0.19 days/year), Baffin Bay (-0.66 ± 0.20 days/year), and Davis Strait (-0.64 ± 0.69 days/year). In the above areas, almost all sea ice normally disappears during summer (Figure 4). For areas where ice persists in concentrations detectable from satellite imagery throughout the entire year (Figure 4), changes in ice concentrations (measured as minimum ice concentrations in summer) have been greatest in the Beaufort Sea and Gulf of Boothia and least in the central Arctic Archipelago (Parkinson and Cavalieri 2002; Comiso and Parkinson 2004).

Scientific projections of effects of climate change on sea ice in the Arctic vary--sometimes widely--and so we recommend that model-averaged projections, such as those presented by the 2007 report of the IPCC and 2004 Arctic Climate Impact Assessment (ACIA), be used to anticipate effects of climate change on the distribution and abundance of polar bears. Projected changes in sea ice in the Arctic as outlined by the IPCC are presented in the chapter of Christensen et al. (2007). In summary, the Arctic is very likely to continue to warm during this century in most areas, and the annual mean warming is very likely to exceed the global mean warming. There will be an increase of 5°C in annual temperature from now to the end of the 21st century (as estimated by the MMD-A1B ensemble mean projection of the IPCC); however, there is a considerable across-model range of 2.8°C to 7.8°C. Warming is projected to be greatest in winter and smallest in summer.Annual arctic precipitation is also very likely to increase in winter. Arctic sea ice is very likely to continue to decrease in extent and thickness, but it is uncertain how circulation patterns in the Arctic Ocean might change.

Figure 4. Ice concentrations at their minimum extents in summer recorded over a 24-year span.
Panels a and b display the average minima recorded during the years 1979−90 and 1991−2002, respectively.
Panel c shows the minimum concentration recorded in 2003. Ice concentrations are measurable down to 8%, below which it is impossible to discriminate between open water and ice-covered areas.
Panel d, the difference between the first 2 panels (b − a), reveals the changes between the 2 periods. The average size of the ice pack in 1979−90 is greater than that in 1991−2002 by about 12%.
Source: reproduced from Comiso and Parkinson (2004) and © 2004 the American Institute of Physics.

Figure 4. Ice concentrations at their minimum extents in summer recorded over a 24-year span. Source: reproduced from Comiso and Parkinson (2004) and © 2004 the American Institute of Physics.

Model-averaged projections of change in extent of sea ice are most often presented as anticipated changes in the minimum extent of sea ice that occurs in late summer (September) in the Arctic. Projected changes in the extent of summer sea ice are presented in Figure 5, which are model-averaged projected changes reproduced from ACIA (2004). Changes in extent of sea ice have varied and will continue to vary regionally in Canada, and it is anticipated that the monthly extents of sea ice will show the least rates of change within the Arctic Archipelago and the greatest rates of change in Hudson Bay, Foxe Basin, Baffin Bay, Davis Strait, and the Beaufort Sea. Nonetheless, ACIA model-averaging indicates that by 2090 it is likely that almost all sea ice within Canada will form only as annual (winter) sea ice (Figure 5).

Figure 5. Current and model-averaged projected decreases in extent of sea ice in September as presented by ACIA (2004).
Source: ACIA (2004) and © Arctic Climate Impact Assessment.

Figure 5. Current and model-averaged projected decreases in extent of sea ice in September as presented by ACIA (2004). Source: ACIA (2004) and © Arctic Climate Impact Assessment.

Higher temperatures and loss of sea ice in the Arctic do not bode well for the future of polar bears. However, quantitative data on what trends in habitat mean for the future distribution and abundance of polar bears are limited. In particular, there is a lack of data on how the dependent variables of projection models produced by bodies like the ACIA and IPCC (e.g., temperature, precipitation, summer extent of sea ice) relate as predictors of survival and reproduction (and thus abundance and distribution) of polar bears. As vital rates will likely relate to variables of climate change in a non-linear manner, depending not only on location (northern subpopulations may initially benefit from reductions in less-productive, multi-year sea ice) but also important factors such as density of prey species, it is difficult to objectively predict population trends into the future from data on climate warming alone. For example, inferring demographic implications for polar bears from changes in extent of summer sea ice is problematic because total or near total melting of sea ice that forces bears onshore in summer is the normal situation faced by approximately 50–60% of the polar bear population in Canada.

What is needed are data that link demographic rates to changes in sea ice. At the time of writing this report, four empirical studies have attempted to correlate annual ice conditions with survival rates of polar bears: studies in the Western Hudson Bay (Regehr et al. 2007a; Section 6.1 and section7.10); Southern Hudson Bay (Obbard et al. 2007; Section 7.11); Southern Beaufort Sea (Regehr et al. 2006, 2007b; Section 7.2); and Northern Beaufort Sea (Stirling et al. 2007; Section 7.3). The analyses for Western Hudson Bay and the Southern Beaufort Sea report links between survival of polar bears and conditions of sea ice; the studies for the Northern Beaufort Sea and Southern Hudson Bay found no environmental or body condition correlates of interannual variation in polar bear survival.

Although there is a general lack of data on the subject (in part due to the lengthy periods of study needed to build accurate models), there have been efforts (Amstrup et al. 2007; Durner et al. 2007) to forecast polar bear abundances based on projected changes in sea ice. Due to lack of data on carrying capacity in relation to ice conditions these projections are preliminary, but noteworthy because they are alarming: Amstrupet al. (2007) predict the loss of 2/3 of the world’s polar bears in 45 years (for Canada, complete extirpation or severe depletion of polar bears from Baffin Bay, Davis Strait, Foxe Basin, Western Hudson Bay, Southern Hudson Bay, and the Southern Beaufort Sea). Amstrup et al. (2007) is a Bayesian network model combining “empirical data, interpretations of data, and [Amstrup’s] professional judgment into a probabilistic framework.” Durner et al. (2007) models the projected disappearance of preferred polar bear habitat (using well-constructed resource selection functions) in the polar basin (in Canada, affected populations would include polar bears of the Southern and Northern Beaufort Sea). Although “less available habitat will likely reduce polar bear populations, exact relationships between habitat losses and population demographics remain unknown (Durner et al. 2007).”Like Population Viability Analysis (PVA; Section 7), the outputs of these models depend on inputs and assumptions. We further discuss the importance of trends in habitat (climate warming) to the status of polar bears in Sections 5, section6, section7 and section9.

4.3 Habitat Protection

Canada's Oceans Act of 1996 allows for the establishment of Marine Protected Areas to conserve marine habitat for polar bears; however, there are no National Marine Conservation Areas in the Arctic at present. Hence, there is no formal protection of the vast majority of polar bear habitat in Canada. Some protection of terrestrial habitat important to polar bears is afforded through Canada’s national parks and Ontario’s provincial parks, and National Wildlife Areas. The north shore of Ivvavik National Park (9,750 km²) offers protection of denning and onshore habitat for polar bears of the Southern Beaufort Sea. Tuktut Nogait National Park (16,340 km²) offers limited protection to habitat of bears of the Northern Beaufort Sea because it is largely removed (by approximately 20 km) from the coast of Amundsen Gulf. Large protected areas in the Canadian Arctic Archipelago include Aulavik National Park (12,274 km²) in the Northwest Territories and Auyuittuq (19,707 km²), Sirmilik (22,200 km²), and Quttinirpaaq (37,775 km²) National Parks in Nunavut, and the National Wildlife Areas of Polar Bear Pass (2,624 km²) on Bathurst Island and Nirjutiqavvik (1,650 km²) off southern Ellesmere Island (Coburg Island). The newly announced Torngat Mountains National Park Reserve (9,600 km²) in Labrador will protect a limited amount of terrestrial habitat for polar bears of Davis Strait. Parks specifically designated to protect denning and onshore habitat of polar bears in Hudson Bay include Wapusk National Park in Manitoba (11,475 km²) and Polar Bear Provincial Park (23,552 km²) in Ontario. Maternity denning occurs in the Cape Tatnam and Cape Churchill Wildlife Management Areas, Manitoba, and both of these regions have management plans under development that will control access to maternity denning areas. All formally protected areas within the range of the polar bear in Canada encompass approximately 2.9% of the area of occupancy of the species (Figure 3).

Page details

Date modified: