# Endangered and Threatened Wildlife and Plants; Determination of Threatened Status for the Polar Bear (Ursus maritimus) Throughout Its Range

> Briefs, arguments, decisions, and more.

URL: https://www.frixlaw.com/law-library/documents/fr%3AE8-11105

## Record

- **Collection:** Federal Register
- **Document type:** Rule
- **Published:** May 15, 2008
- **Citation:** 73 FR 28212

## Text

DEPARTMENT OF THE INTERIOR
Fish and Wildlife Service
50 CFR Part 17
[FWS-R7-ES-2008-0038; 1111 FY07 MO-B2]
RIN 1018-AV19
Endangered and Threatened Wildlife and Plants; Determination of Threatened Status for the Polar Bear (Ursus maritimus) Throughout Its Range

AGENCY:

Fish and Wildlife Service, Interior.

ACTION:

Final rule.

SUMMARY:

We, the U.S. Fish and Wildlife Service (Service), determine threatened status for the polar bear (
Ursus maritimus
) under the Endangered Species Act of 1973, as amended (Act) (16 U.S.C. 1531 et seq.). Polar bears evolved to utilize the Arctic sea ice niche and are distributed throughout most ice-covered seas of the Northern Hemisphere. We find, based upon the best available scientific and commercial information, that polar bear habitat—principally sea ice—is declining throughout the species' range, that this decline is expected to continue for the foreseeable future, and that this loss threatens the species throughout all of its range. Therefore, we find that the polar bear is likely to become an endangered species within the foreseeable future throughout all of its range. This final rule activates the consultation provisions of section 7 of the Act for the polar bear. The special rule for the polar bear, also published in today's edition of the
Federal Register
, sets out the prohibitions and exceptions that apply to this threatened species.

DATES:

This rule is effective May 15, 2008. The U.S. District Court order in
Center for Biological Diversity
v.
Kempthorne,
No. C 08-1339 CW (N.D. Cal., April 28, 2008) ordered that the 30-day notice period otherwise required by the Administrative Procedure Act be waived, pursuant to 5 U.S.C. 553(d)(3).

ADDRESSES:

Comments and materials received, as well as supporting scientific documentation used in the preparation of this rule, will be available for public inspection, by appointment, during normal business hours at: U.S. Fish and Wildlife Service, Marine Mammals Management Office, 1011 East Tudor Road, Anchorage, AK 99503. Copies of this final rule are also available on the Service's Marine Mammal website:
http://alaska.fws.gov/fisheries/mmm/polarbear/issues.htm.

FOR FURTHER INFORMATION CONTACT:

Scott Schliebe, Marine Mammals Management Office (see
ADDRESSES
section) (telephone 907-786-3800). Persons who use a telecommunications device for the deaf (TDD) may call the Federal Information Relay Service (FIRS) at 1-800-877-8339, 24 hours a day, 7 days a week.

SUPPLEMENTARY INFORMATION:

Background

Information in this section is summarized from the following sources: (1) The Polar Bear Status Review (Schliebe et al. 2006a); (2) information received from public comments in response to our proposal to list the polar bear as a threatened species published in the
Federal Register
on January 9, 2007 (72 FR 1064); (3) new information published since the proposed rule (72 FR 1064), including additional sea ice and climatological studies contained in the Intergovernmental Panel on Climate Change (IPCC)
Fourth Assessment Report
(AR4) and other published papers; and (4) scientific analyses conducted by the U.S. Geological Survey (USGS) and co-investigators at the request of the Secretary of the Department of the Interior specifically for this determination. For more detailed information on the biology of the polar bear, please consult the Status Review and additional references cited throughout this document.

Species Biology

Taxonomy and Evolution

Throughout the Arctic, polar bears are known by a variety of common names, including nanook, nanuq, ice bear, sea bear, isbjørn, white bears, and eisbär. Phipps (1774, p. 174) first proposed and described the polar bear as a species distinct from other bears and provided the scientific name
Ursus maritimus
. A number of alternative names followed, but Harington (1966, pp. 3-7), Manning (1971, p. 9), and Wilson (1976, p. 453) (all three references cited in Amstrup 2003, p. 587) subsequently promoted the name
Ursus maritimus
that has been used since.

The polar bear is usually considered a marine mammal since its primary habitat is the sea ice (Amstrup 2003, p. 587), and it is evolutionarily adapted to life on sea ice (see further discussion under General Description section). The polar bear is included on the list of species covered under the U.S. Marine Mammal Protection Act of 1972, as amended (16 U.S.C. 1361 et seq.) (MMPA).

Polar bears diverged from grizzly bears (
Ursus arctos
) somewhere between 200,000 and 400,000 years ago (Talbot and Shields 1996a, p. 490; Talbot and Shields 1996b, p. 574). However, fossil evidence of polar bears does not appear until after the Last Interglacial Period (115,000 to 140,000 years ago) (Kurten 1964, p. 25; Ingolfsson and Wiig 2007). Only in portions of northern Canada, Chukotka, Russia, and northern Alaska do the ranges of polar bears and grizzly bears overlap. Cross-breeding of grizzly bears and polar bears in captivity has produced reproductively viable offspring (Gray 1972, p. 56; Stirling 1988, p. 23). The first documented case of cross-breeding in the wild was reported in the spring of 2006, and Wildlife Genetics International confirmed the cross-breeding of a female polar bear and male grizzly bear (Paetkau, pers. comm. May 2006).

General Description

Polar bears are the largest of the living bear species (DeMaster and Stirling 1981, p. 1; Stirling and Derocher 1990, p. 190). They are characterized by large body size, a stocky form, and fur color that varies from white to yellow. They are sexually dimorphic; females weigh 181 to 317 kilograms (kg) (400 to 700 pounds (lbs)), and males up to 654 kg (1,440 lbs). Polar bears have a longer neck and a proportionally smaller head than other members of the bear family (Ursidae) and are missing the distinct shoulder hump common to grizzly bears. The nose, lips, and skin of polar bears are black (Demaster and Stirling 1981, p. 1; Amstrup 2003, p. 588).

Polar bears evolved in sea ice habitats and as a result are evolutionarily adapted to this habitat. Adaptations unique to polar bears in comparison to other Ursidae include: (1) White pelage with water-repellent guard hairs and dense underfur; (2) a short, furred snout; (3) small ears with reduced surface area; (4) teeth specialized for a carnivorous rather than an omnivorous diet; and (5) feet with tiny papillae on the underside, which increase traction on ice (Stirling 1988, p. 24). Additional adaptations include large, paddle-like feet (Stirling 1988, p. 24), and claws that are shorter and more strongly curved than those of grizzly bears, and larger and heavier than those of black bears (
Ursus americanus
) (Amstrup 2003, p. 589).

Distribution and Movements

Polar bears evolved to utilize the Arctic sea ice niche and are distributed throughout most ice-covered seas of the Northern Hemisphere. They occur throughout the East Siberian, Laptev, Kara, and Barents Seas of Russia; Fram Strait (the narrow strait between northern Greenland and Svalbard),

Greenland Sea and Barents Sea of northern Europe (Norway and Greenland (Denmark)); Baffin Bay, which separates Canada and Greenland, through most of the Canadian Arctic archipelago and the Canadian Beaufort Sea; and in the Chukchi and Beaufort Seas located west and north of Alaska.

Over most of their range, polar bears remain on the sea ice year-round or spend only short periods on land. However, some polar bear populations occur in seasonally ice-free environs and use land habitats for varying portions of the year. In the Chukchi Sea and Beaufort Sea areas of Alaska and northwestern Canada, for example, less than 10 percent of the polar bear locations obtained via radio telemetry were on land (Amstrup 2000, p. 137; Amstrup, USGS, unpublished data); the majority of land locations were bears occupying maternal dens during the winter. A similar pattern was found in East Greenland (Wiig et al. 2003, p. 511). In the absence of ice during the summer season, some populations of polar bears in eastern Canada and Hudson Bay remain on land for extended periods of time until ice again forms and provides a platform for them to move to sea. Similarly, in the Barents Sea, a portion of the population is spending greater amounts of time on land.

Although polar bears are generally limited to areas where the sea is ice-covered for much of the year, they are not evenly distributed throughout their range on sea ice. They show a preference for certain sea ice characteristics, concentrations, and specific sea ice features (Stirling et al. 1993, pp. 18-22; Arthur et al. 1996, p. 223; Ferguson et al. 2000a, p. 1,125; Ferguson et al. 2000b, pp. 770-771; Mauritzen et al. 2001, p. 1,711; Durner et al. 2004, pp. 18-19; Durner et al. 2006, p. pp. 34-35; Durner et al. 2007, pp. 17 and 19). Sea-ice habitat quality varies temporally as well as geographically (Ferguson et al. 1997, p. 1,592; Ferguson et al. 1998, pp. 1,088-1,089; Ferguson et al.2000a, p. 1,124; Ferguson et al.2000b, pp. 770-771; Amstrup et al. 2000b, p. 962). Polar bears show a preference for sea ice located over and near the continental shelf (Derocher et al. 2004, p. 164; Durner et al. 2004, p. 18-19; Durner et al. 2007, p. 19), likely due to higher biological productivity in these areas (Dunton et al. 2005, pp. 3,467-3,468) and greater accessibility to prey in near-shore shear zones and polynyas (areas of open sea surrounded by ice) compared to deep-water regions in the central polar basin (Stirling 1997, pp. 12-14). Bears are most abundant near the shore in shallow-water areas, and also in other areas where currents and ocean upwelling increase marine productivity and serve to keep the ice cover from becoming too consolidated in winter (Stirling and Smith 1975, p. 132; Stirling et al. 1981, p. 49; Amstrup and DeMaster 1988, p. 44; Stirling 1990, pp. 226-227; Stirling and Øritsland 1995, p. 2,607; Amstrup et al. 2000b, p. 960).

Polar bear distribution in most areas varies seasonally with the seasonal extent of sea ice cover and availability of prey. The seasonal movement patterns of polar bears emphasize the role of sea ice in their life cycle. In Alaska in the winter, sea ice may extend 400 kilometers (km) (248 miles (mi)) south of the Bering Strait, and polar bears will extend their range to the southernmost proximity of the ice (Ray 1971, p. 13). Sea ice disappears from the Bering Sea and is greatly reduced in the Chukchi Sea in the summer, and polar bears occupying these areas move as much as 1,000 km (621 mi) to stay with the pack ice (Garner et al. 1990, p. 222; Garner et al. 1994, pp. 407-408). Throughout the polar basin during the summer, polar bears generally concentrate along the edge of or into the adjacent persistent pack ice. Significant northerly and southerly movements of polar bears appear to depend on seasonal melting and refreezing of ice (Amstrup 2000, p. 142). In other areas, for example, when the sea ice melts in Hudson Bay, James Bay, Davis Strait, Baffin Bay, and some portions of the Barents Sea, polar bears remain on land for up to 4 or 5 months while they wait for winter and new ice to form (Jonkel et al. 1976, pp. 13-22; Schweinsburg 1979, pp. 165, 167; Prevett and Kolenosky 1982, pp. 934-935; Schweinsburg and Lee 1982, p. 510; Ferguson et al. 1997, p. 1,592; Lunn et al. 1997, p. 235; Mauritzen et al. 2001, p. 1,710).

In areas where sea ice cover and character are seasonally dynamic, a large multi-year home range, of which only a portion may be used in any one season or year, is an important part of the polar bear life history strategy. In other regions, where ice is less dynamic, home ranges are smaller and less variable (Ferguson et al. 2001, pp.51-52). Data from telemetry studies of adult female polar bears show that they do not wander aimlessly on the ice, nor are they carried passively with the ocean currents as previously thought (Pedersen 1945 cited in Amstrup 2003, p. 587). Results show strong fidelity to activity areas that are used over multiple years (Ferguson et al. 1997, p. 1,589). All areas within an activity area are not used each year.

The distribution patterns of some polar bear populations during the open water and early fall seasons have changed in recent years. In the Beaufort Sea, for example, greater numbers of polar bears are being found on shore than recorded at any previous time (Schliebe et al. 2006b, p. 559). In Baffin Bay, Davis Strait, western Hudson Bay and other areas of Canada, Inuit hunters are reporting an increase in the numbers of bears present on land during summer and fall (Dowsley and Taylor 2005, p. 2; Dowsley 2005, p. 2). The exact reasons for these changes may involve a number of factors, including changes in sea ice (Stirling and Parkinson 2006, p. 272).

Food Habits

Polar bears are carnivorous, and a top predator of the Arctic marine ecosystem. Polar bears prey heavily throughout their range on ice-dependent seals (frequently referred to as “ice seals”), principally ringed seals (
Phoca hispida
), and, to a lesser extent, bearded seals (
Erignathus barbatus)
. In some locales, other seal species are taken. On average, an adult polar bear needs approximately 2 kg (4.4 lbs) of seal fat per day to survive (Best 1985, p. 1035). Sufficient nutrition is critical and may be obtained and stored as fat when prey is abundant.

Although seals are their primary prey, polar bears occasionally take much larger animals such as walruses (
Odobenus rosmarus
), narwhal (
Monodon monoceros
), and belugas (
Delphinapterus leucas
) (Kiliaan and Stirling 1978, p. 199; Smith 1980, p. 2,206; Smith 1985, pp. 72-73; Lowry et al. 1987, p. 141; Calvert and Stirling 1990, p. 352; Smith and Sjare 1990, p. 99). In some areas and under some conditions, prey other than seals or carrion may be quite important to polar bear sustenance as short-term supplemental forms of nutrition. Stirling and Øritsland (1995, p. 2,609) suggested that in areas where ringed seal populations were reduced, other prey species were being substituted. Like other ursids, polar bears will eat human garbage (Lunn and Stirling 1985, p. 2,295), and when confined to land for long periods, they will consume coastal marine and terrestrial plants and other terrestrial foods (Russell 1975, p. 122; Derocher et al. 1993, p. 252); however the significance of such other terrestrial foods to the long-term welfare of polar bears may be limited (Lunn and Stirling 1985, p. 2,296; Ramsay and Hobson 1991, p. 600; Derocher et al. 2004, p. 169) as further expanded under the section entitled “Adaptation” below.

Reproduction

Polar bears are characterized by late sexual maturity, small litter sizes, and extended parental investment in raising young, all factors that contribute to a low reproductive rate (Amstrup 2003, pp. 599-600). Reproduction in the female polar bear is similar to that in other ursids. Females generally mature and breed for the first time at 4 or 5 years and give birth at 5 or 6 years of age. Litters of two cubs are most common, but litters of three cubs are seen sporadically across the Arctic (Amstrup 2003, p. 599). When foraging conditions are difficult, polar bears may “defer” reproduction in favor of survival (Derocher et al. 1992, p. 564).

Polar bears enter a prolonged estrus between March and June, when breeding occurs. Ovulation is induced by mating (Wimsatt 1963, p. 72), and implantation is delayed until autumn. The total gestation period is 195 to 265 days (Uspenski 1977, cited in Amstrup 2003, p. 599), although active development of the fetus is suspended during most of this period. The timing of implantation, and therefore the timing of birth, is likely dependent on body condition of the female, which depends on a variety of environmental factors. Pregnant females that spend the late summer on land prior to denning may not feed for 8 months (Watts and Hansen 1987, p. 627). This may be the longest period of food deprivation of any mammal, and it occurs at a time when the female gives birth to and then nourishes new cubs.

Newborn polar bears are helpless and have hair, but are blind and weigh only 0.6 kg (1.3 lb) (Blix and Lentfer 1979, p. 68). Cubs grow rapidly, and may weigh 10 to 12 kg (22 to 26 lbs) by the time they emerge from the den in the spring. Young bears will stay with their mothers until weaning, which occurs most commonly in early spring when the cubs are 2.3 years of age. Female polar bears are available to breed again after their cubs are weaned; thus the reproductive interval for polar bears is 3 years.

Polar bears are long-lived mammals not generally susceptible to disease, parasites, or injury. The oldest known female in the wild was 32 years of age and the oldest known male was 28, though few polar bears in the wild live to be older than 20 years (Stirling 1988, p. 139; Stirling 1990, p. 225). Due to extremely low reproductive rates, polar bears require a high survival rate to maintain population levels (Eberhardt 1985, p. 1,010; Amstrup and Durner 1995, pp. 1,313, 1,319). Survival rates increase up to a certain age, with cubs-of-the-year having the lowest rates and prime age adults (between 5 and 20 years of age) having survival rates that can exceed 90 percent. Amstrup and Durner (1995, p. 1,319) report that high survival rates (exceeding 90 percent for adult females) are essential to sustain populations.

Polar Bear—Sea Ice Habitat Relationships

Polar bears are distributed throughout the ice-covered waters of the circumpolar Arctic (Stirling 1988, p. 61), and rely on sea ice as their primary habitat (Amstrup 2003, p. 587). Polar bears depend on sea ice for a number of purposes, including as a platform from which to hunt and feed upon seals; as habitat on which to seek mates and breed; as a platform to move to terrestrial maternity denning areas, and sometimes for maternity denning; and as a substrate on which to make long-distance movements (Stirling and Derocher 1993, p. 241). Mauritzen et al. (2003b, p. 123) indicated that habitat use by polar bears during certain seasons may involve a trade-off between selecting habitats with abundant prey availability versus the use of safer retreat habitats (i.e., habitats where polar bears have lower probability of becoming separated from the main body of the pack ice) of higher ice concentrations with less prey. Their findings indicate that polar bear distribution may not be solely a reflection of prey availability, but other factors such as energetic costs or risk may be involved.

Stirling et al. (1993, p. 15) defined seven types of sea ice habitat and classified polar bear use of these ice types based on the presence of bears or bear tracks in order to determine habitat preferences. The seven types of sea ice are: (1) stable fast ice with drifts; (2) stable fast ice without drifts; (3) floe edge ice; (4) moving ice; (5) continuous stable pressure ridges; (6) coastal low level pressure ridges; and (7) fiords and bays. Polar bears were not evenly distributed over these sea ice habitats, but concentrated on the floe ice edge, on stable fast ice with drifts, and on areas of moving ice (Stirling 1990 p. 226; Stirling et al. 1993, p. 18). In another assessment, categories of ice types included pack ice, shore-fast ice, transition zone ice, polynyas, and leads (linear openings or cracks in the ice) (USFWS 1995, p. 9). Pack ice, which consists of annual and multi-year older ice in constant motion due to winds and currents, is the primary summer habitat for polar bears in Alaska. Shore-fast ice (also known as “fast ice”, it is defined by the
Arctic Climate Impact Assessment
(2005, p. 190) as ice that grows seaward from a coast and remains in place throughout the winter; typically it is stabilized by grounded pressure ridges at its outer edge) is used for feeding on seal pups, for movement, and occasionally for maternity denning. Open water at leads and polynyas attracts seals and other marine mammals and provides preferred hunting habitats during winter and spring. Durner et al. (2004, pp. 18-19; Durner et al. 2007, pp. 17-18) found that polar bears in the Arctic basin prefer sea ice concentrations greater than 50 percent located over the continental shelf with water depths less than 300 m (984 feet (ft)).

Polar bears must move throughout the year to adjust to the changing distribution of sea ice and seals (Stirling 1988, p. 63; USFWS 1995, p. 4). In some areas, such as Hudson Bay and James Bay, polar bears remain on land when the sea ice retreats in the spring and they fast for several months (up to 8 months for pregnant females) before fall freeze-up (Stirling 1988, p. 63; Derocher et al. 2004, p. 163; Amstrup et al. 2007, p. 4). Some populations unconstrained by land masses, such as those in the Barents, Chukchi, and Beaufort Seas, spend each summer on the multi-year ice of the polar basin (Derocher et al. 2004, p. 163; Amstrup et al. 2007, p. 4). In intermediate areas such as the Canadian Arctic, Svalbard, and Franz Josef Land archipelagos, bears stay on the sea ice most of the time, but in some years they may spend up to a few months on land (Mauritizen et al. 2001, p. 1,710). Most populations use terrestrial habitat partially or exclusively for maternity denning; therefore, females must adjust their movements in order to access land at the appropriate time (Stirling 1988, p. 64; Derocher et al. 2004, p. 166).

Sea ice changes between years in response to environmental factors may have consequences for the distribution and productivity of polar bears as well as their prey. In the southern Beaufort Sea, anomalous heavy sea ice conditions in the mid-1970s and mid-1980s (thought to be roughly in phase with a similar variation in runoff from the Mackenzie River) caused significant declines in productivity of ringed seals (Stirling 2002, p. 68). Each event lasted approximately 3 years and caused similar declines in the birth rate of polar bears and survival of subadults, after which reproductive success and survival of both species increased again.

Maternal Denning Habitat

Throughout the species' range, most pregnant female polar bears excavate

dens in snow located on land in the fall-early winter period (Harington 1968, p. 6; Lentfer and Hensel 1980, p. 102; Ramsay and Stirling 1990, p. 233; Amstrup and Gardner 1994, p. 5). The only known exceptions are in western and southern Hudson Bay, where polar bears first excavate earthen dens and later reposition into adjacent snow drifts (Jonkel et al. 1972, p. 146; Ramsay and Stirling 1990, p. 233), and in the southern Beaufort Sea, where a portion of the population dens in snow caves located on pack and shore-fast ice. Successful denning by polar bears requires accumulation of sufficient snow for den construction and maintenance. Adequate and timely snowfall combined with winds that cause snow accumulation leeward of topographic features create denning habitat (Harington 1968, p. 12).

A great amount of polar bear denning occurs in core areas (Harington 1968, pp. 7-8), which show high use over time (see Figure 8). In some portions of the species' range, polar bears den in a more diffuse pattern, with dens scattered over larger areas at lower density (Lentfer and Hensel 1980, p. 102; Stirling and Andriashek 1992, p. 363; Amstrup 1993, p. 247; Amstrup and Gardner 1994, p. 5; Messier et al. 1994, p. 425; Born 1995, p. 81; Ferguson et al. 2000a, p. 1125; Durner et al. 2001, p. 117; Durner et al. 2003, p. 57).

Habitat characteristics of denning areas vary substantially from the rugged mountains and fjordlands of the Svalbard archipelago and the large islands north of the Russian coast (Lønø 1970, p. 77; Uspenski and Kistchinski 1972, p. 182; Larsen 1985, pp. 321-322), to the relatively flat topography of areas such as the west coast of Hudson Bay (Ramsay and Andriashek 1986, p. 9; Ramsay and Stirling 1990, p. 233) and north slope of Alaska (Amstrup 1993, p. 247; Amstrup and Gardner 1994, p. 7; Durner et al. 2001, p. 119; Durner et al. 2003, p. 61), to offshore pack ice-pressure ridge habitat (Amstrup and Gardner 1994, p. 4; Fischbach et al. 2007, p. 1,400). The key characteristic of all denning habitat is topographic features that catch snow in the autumn and early winter (Durner et al. 2003, p. 61). Across the range, most polar bear dens occur relatively near the coast. The main exception to coastal denning occurs in the western Hudson Bay area, where bears den farther inland in traditional denning areas (Kolenosky and Prevett 1983, pp. 243-244; Stirling and Ramsay 1986, p. 349).

Current Population Status and Trend

The total number of polar bears worldwide is estimated to be 20,000-25,000 (Aars et al. 2006, p. 33). Polar bears are not evenly distributed throughout the Arctic, nor do they comprise a single nomadic cosmopolitan population, but rather occur in 19 relatively discrete populations (Aars et al. 2006, p. 33). The use of the term “relatively discrete population” in this context is not intended to equate to the Act's term “distinct population segments” (Figure 1). Boundaries of the 19 polar bear populations have evolved over time and are based on intensive study of movement patterns, tag returns from harvested animals, and, to a lesser degree, genetic analysis (Aars et al. 2006, pp. 33-47). The scientific studies regarding population bounds began in the early 1970s and continue today. Within this final rule we have adopted the use of the term “population” to describe polar bear management units consistent with their designation by the World Conservation Union-International Union for Conservation of Nature and Natural Resources (IUCN), Species Survival Commission (SSC) Polar Bear Specialist Group (PBSG) with information available as of October 2006 (Aars et al. 2006, p. 33), and to describe a combination of two or more of these populations into “ecoregions,” as discussed in following sections. Although movements of individual polar bears overlap extensively, telemetry studies demonstrate spatial segregation among groups or stocks of polar bears in different regions of their circumpolar range (Schweinsburg and Lee 1982, p. 509; Amstrup et al. 1986, p. 252; Amstrup et al., 2000b, pp. 957-958.; Garner et al. 1990, p. 224; Garner et al. 1994, pp.112-115; Amstrup and Gardner 1994, p. 7; Ferguson et al. 1999, pp. 313-314; Lunn et al. 2002, p. 41). These patterns, along with information obtained from survey and reconnaissance, marking and tagging studies, and traditional knowledge, have resulted in recognition of 19 relatively discrete polar bear populations (Aars et al. 2006, p. 33). Genetic analysis reinforces the boundaries between some designated populations (Paetkau et al. 1999, p. 1,571; Amstrup 2003, p. 590) while confirming the existence of overlap and mixing among others (Paetkau et al. 1999, p. 1,571; Cronin et al. 2006, p. 655). There is considerable overlap in areas occupied by members of these groups (Amstrup et al. 2004, p. 676; Amstrup et al. 2005, p. 252), and boundaries separating the groups are adjusted as new data are collected. These boundaries, however, are thought to be ecologically meaningful, and the 19 units they describe are managed as populations, with the exception of the Arctic Basin population where few bears are believed to be year-round residents.

ER15MY08.002

Population size estimates and qualitative categories of current trend and status for each of the 19 polar bear populations are discussed below. This discussion was derived from information presented at the IUCN/SSC PBSG meeting held in Seattle, Washington, in June 2005, and updated with results that became available in October 2006 (Aars et al. 2006, p. 33). The following narrative incorporates results from two recent publications (Stirling et al. 2007; Obbard et al. 2007). The remainder of the information on each population is based on the available status reports and revisions given by each nation, as reported in Aars et al. (2006).

Status categories include an assessment of whether a population is believed to be not reduced, reduced, or severely reduced from historic levels of abundance, or if insufficient data are available to estimate status. Trend categories include an assessment of whether the population is currently increasing, stable, or declining, or if insufficient data are available to estimate trend. In general, an assessment of trend requires a monitoring program or data to allow population size to be estimated at more than one point in time. Information on the date of the current population estimate and information on previous population estimates and the basis for

those estimates is detailed in Aars et al. (2006, pp. 34-35). In some instances a subjective assessment of trend has been provided in the absence of either a monitoring program or estimates of population size developed for more than one point in time. This status and trend analysis only reflects information about the past and present polar bear populations. Later in this final rule a discussion will be presented about the scientific information on threats that will affect the species within the foreseeable future. The Act establishes a five-factor analysis for using this information in making listing decisions.

Populations are discussed in a counterclockwise order from Figure 1, beginning with East Greenland. There is no population size estimate for the East Greenland polar bear population because no population surveys have been conducted there. Thus, the status and trend of this population have not been determined. The Barents Sea population was estimated to comprise 3,000 animals based on the only population survey conducted in 2004. Because only one abundance estimate is available, the status and trend of this population cannot yet be determined. There is no population size estimate for the Kara Sea population because population surveys have not been conducted; thus status and trend of this population cannot yet be determined. The Laptev Sea population was estimated to comprise 800 to 1,200 animals, on the basis of an extrapolation of historical aerial den survey data (1993). Status and trend cannot yet be determined for this population.

The Chukchi Sea population is estimated to comprise 2,000 animals, based on extrapolation of aerial den surveys (2002). Status and trend cannot yet be determined for this population. The Southern Beaufort Sea population is comprised of 1,500 animals, based on a recent population inventory (2006). The predicted trend is declining (Aars et al. 2006, p.33), and the status is designated as reduced. The Northern Beaufort Sea population was estimated to number 1,200 animals (1986). The trend is designated as stable, and status is believed to be not reduced. Stirling et al. (2007, pp. 12-14) estimated long-term trends in population size for the Northern Beaufort Sea population. The model-averaged estimate of population size from 2004 to 2006 was 980 bears, and did not differ in a statistically significantly way from estimates for the periods of 1972 to 1975 (745 bears) and 1985 to 1987 (867 bears), and thus the trend is stable. Stirling et al. (2007, p. 13) indicated that, based on a number of indications and separate annual abundance estimates for the study period, the population estimate may be slightly biased low (i.e., might be an underestimate) due to sampling issues.

The Viscount Melville Sound population was estimated to number 215 animals (1992). The observed or predicted trend based on management action is listed as increasing (Aars et al. 2006, p. 33), although the status is designated as severely reduced from prior excessive harvest. The Norwegian Bay population estimate was 190 animals (1998); the trend, based on computer simulations, is noted as declining, while the status is listed as not reduced. The Lancaster Sound population estimate was 2,541 animals (1998); the trend is thought to be stable, and status is not reduced. The M'Clintock Channel population is estimated at 284 animals (2000); the observed or predicted trend based on management actions is listed as increasing although the status is severely reduced from excessive harvest. The Gulf of Boothia population estimate is 1,523 animals (2000); the trend is thought to be stable, and status is designated as not reduced. The Foxe Basin population was estimated to number 2,197 animals in 1994; the population trend is thought to be stable, and the status is not reduced. The Western Hudson Bay population estimate is 935 animals (2004); the trend is declining, and the status is reduced. The Southern Hudson Bay population was estimated to be 1,000 animals in 1988 (Aars et al. 2006, p. 35); the trend is thought to be stable, and status is not reduced. In a more recent analysis, Obbard et al. (2007) applied open population capture-recapture models to data collected from 1984-86 and 1999-2005 to estimate population size, trend, and survival for the Southern Hudson Bay population. Their results indicate that the size of the Southern Hudson Bay population appears to be unchanged from the mid-1980s. From 1984-1986, the population was estimated at 641 bears; from 2003-2005, the population was estimated at 681 bears. Thus, the trend for this population is stable. The Kane Basin population was estimated to be comprised of 164 animals (1998); its trend is declining, and status is reduced. The Baffin Bay population was estimated to be 2,074 animals (1998); the trend is declining, and status is reduced. The Davis Strait population was estimated to number 1,650 animals based on traditional ecological knowledge (TEK) (2004); data were unavailable to assess trends or status. Preliminary information from the second of a 3-year population assessment estimates the population number to be 2,375 bears (Peacock et al. 2007, p. 7). The Arctic Basin population estimate, trend, and status are unknown (Aars et al. 2006, p. 35).

On the basis of information presented above, two polar bear populations are designated as increasing (Viscount Melville Sound and M'Clintock Channel-both were severely reduced in the past and are recovering under conservative harvest limits); six populations are stable (Northern Beaufort Sea, Southern Hudson Bay, Davis Strait, Lancaster Sound, Gulf of Bothia, Foxe Basin); five populations are declining (Southern Beaufort Sea, Norwegian Bay, Western Hudson Bay, Kane Basin, Baffin Bay); and six populations are designated as data deficient (Barents Sea, Kara Sea, Laptev Sea, Chukchi Sea, Arctic Basin, East Greenland) with no estimate of trend. The two populations with the most extensive time series of data, Western Hudson Bay and Southern Beaufort Sea, are both considered to be declining.

As previously noted, scientific information assessing this species in the foreseeable future is provided later in this final rule.

Polar Bear Ecoregions

Amstrup et al. (2007, pp. 6-8) grouped the 19 IUCN-recognized polar bear populations (Aars et al. 2006, p. 33) into four physiographically different functional groups or “ecoregions” (Figure 2) in order to forecast future polar bear population status on the basis of current knowledge of polar bear populations, their relationships to sea ice habitat, and predicted changes in sea ice and other environmental variables. Amstrup et al. (2007, p. 7) defined the ecoregions “on the basis of observed temporal and spatial patterns of ice formation and ablation (melting or evaporation), observations of how polar bears respond to those patterns, and how general circulation models (GCMs) forecast future ice patterns.”

The
Seasonal Ice Ecoregion
includes the Western and Southern Hudson Bay populations, as well as the Foxe Basin, Baffin Bay, and Davis Strait populations. These 5 IUCN-recognized populations are thought to include a total of about 7,200 polar bears (Aars et al. 2006, p. 34-35). The 5 populations experience sea ice that melts entirely in summer, and bears spend extended periods of time on shore.

ER15MY08.003

The
Archipelago Ecoregion,
islands and channels of the Canadian Arctic, has approximately 5,000 polar bears representing 6 populations recognized by the IUCN (Aars et al. 2006, p. 34-35). These populations are Kane Basin, Norwegian Bay, Viscount Melville Sound, Lancaster Sound, M'Clintock Channel, and the Gulf of Boothia. Much of this region is characterized by heavy annual and multi-year ice that fills the inter-island channels year round and polar bears remain on the sea ice throughout the year.

The polar basin was split into a
Convergent Ecoregion
and a
Divergent Ecoregion,
based upon the different patterns of sea ice formation, loss (via melt and transport) (Rigor et al. 2002, p. 2,658; Rigor and Wallace 2004, p. 4; Maslanik et al. 2007, pp. 1-3; Meier et al. 2007, pp. 428-434; Ogi and Wallace 2007, pp. 2-3).

The
Divergent Ecoregion
is characterized by extensive formation of annual sea ice that is transported toward the Canadian Arctic islands and Greenland, or out of the polar basin through Fram Strait. The Divergent ecoregion includes the Southern Beaufort, Chukchi, Laptev, Kara, and Barents Seas populations, and is thought to contain up to 9,500 polar bears. In the Divergent Ecoregion, as in the Archipelago Ecoregion, polar bears mainly stay on the sea ice year-round.

The
Convergent Ecoregion,
composed of the Northern Beaufort Sea, Queen Elizabeth Islands (see below), and East Greenland populations, is thought to contain approximately 2,200 polar bears. Amstrup et al. (2007, p. 7) modified the IUCN-recognized population boundaries (Aars et al. 2006, pp. 33,36) of this ecoregion by redefining a Queen Elizabeth Islands population and extending the original boundary of that population to include northwestern Greenland (see Figure 2). The area contained within this boundary is characterized by heavy multi-year ice, except for a recurring lead system that runs along the Queen Elizabeth Islands from the northeastern Beaufort Sea to northern Greenland (Stirling 1980, pp. 307-308). The area may contain over 200 polar bears and some bears from other regions have been recorded moving through the area (Durner and Amstrup 1995, p. 339; Lunn et al. 1995, pp. 12-13). The Northern Beaufort Sea and Queen Elizabeth Islands populations occur in a region of the polar basin that accumulates ice (hence, the Convergent Ecoregion) as it is moved from the polar basin Divergent Ecoregion, while the East Greenland population occurs in area where ice is transported out of the polar basin through the Fram Strait (Comiso 2002, pp. 17-18; Rigor and Wallace 2004, p. 3; Belchansky et al. 2005, pp. 1-2; Holland et al. 2006, pp. 1-5; Durner et al. 2007, p. 3; Ogi and Wallace 2007, p. 2; Serreze et al. 2007, pp. 1,533-1536).

Amstrup et al. (2007) do not incorporate the central Arctic Basin population into an ecoregion. This population was defined by the IUCN in 2001 (Lunn et al. 2002, p.29) to recognize polar bears that may reside outside the territorial jurisdictions of the polar nations. The Arctic Basin region is characterized by very deep water, which is known to be unproductive (Pomeroy 1997, pp. 6-7). Available data indicate that polar bears prefer sea ice over shallow water (less then 300 m (984 ft) deep) (Amstrup et

al. 2000b, p. 962; Amstrup et al. 2004, p. 675; Durner et al. 2007, pp. 18-19), and it is thought that this preference reflects increased hunting opportunities over more productive waters. Also, tracking studies indicate that few if any bears are year-round residents of the central Arctic Basin, and therefore this relatively unpopulated portion of the Arctic was not designated as an ecoregion.

Sea Ice Environment

As described in detail in the “Species Biology” section of this rule, above, polar bears are evolutionarily adapted to life on sea ice (Stirling 1988, p. 24; Amstrup 2003, p. 587). They need sea ice as a platform for hunting, for seasonal movements, for travel to terrestrial denning areas, for resting, and for mating (Stirling and Derocher 1993, p. 241). Moore and Huntington (in press) classify the polar bear as an “ice-obligate” species because of its reliance on sea ice as a platform for resting, breeding, and hunting, while Laidre et al. (in press) similarly describe the polar bear as a species that principally relies on annual sea ice over the continental shelf and areas toward the southern edge of sea ice for foraging. Some polar bears use terrestrial habitats seasonally (e.g., for denning or for resting during open water periods). Open water is not considered to be an essential habitat type for polar bears, because life functions such as feeding, reproduction, or resting do not occur in open water. However, open water is a fundamental part of the marine system that supports seal species, the principal prey of polar bears, and seasonally refreezes to form the ice needed by the bears (see “Open Water Habitat” section for more information). Further, the open water interface with sea ice is an important habitat used to a great extent by polar bears. In addition, the extent of open water is important because vast areas of open water may limit a bear's ability to access sea ice or land (see “Open Water Swimming” section for more detail). Snow cover, both on land and on sea ice, is an important component of polar bear habitat in that it provides insulation and cover for young polar bears and ringed seals in snow dens or lairs (see “Maternal Denning Habitat” section for more detail).

Sea Ice Habitat

Overview of Arctic Sea Ice

According to the
Arctic Climate Impact Assessment
(ACIA 2005), approximately two-thirds of the Arctic is ocean, including the Arctic Ocean and its shelf seas plus the Nordic, Labrador, and Bering Seas (ACIA 2005, p. 454). Sea ice is the defining characteristic of the marine Arctic (ACIA 2005, p. 30). The Arctic sea ice environment is highly dynamic and follows annual patterns of expansion and contraction. Sea ice is typically at its maximum extent (the term “extent” is formally defined in the “Observed Changes in Arctic Sea Ice” section) in March and at its minimum extent in September (Parkinson et al. 1999, p. 20,840). The two primary forms of sea ice are seasonal (or first year) ice and perennial (or multi-year) ice (ACIA 2005, p. 30). Seasonal ice is in its first autumn/winter of growth or first spring/summer of melt (ACIA 2005, p. 30). It has been documented to vary in thickness from a few tenths of a meter near the southern margin of the sea ice to 2.5 m (8.2 ft) in the high Arctic at the end of winter (ACIA 2005, p. 30), with some ice also that is thinner and some limited amount of ice that can be much thicker, especially in areas with ridging (C. Parkinson, NASA, in litt. to the Service, November 2007). If first-year ice survives the summer melt, it becomes multi-year ice. This ice tends to develop a distinctive hummocky appearance through thermal weathering, becoming harder and almost salt-free over several years (ACIA 2005, p. 30). Sea ice near the shore thickens in shallow waters during the winter, and portions become grounded. Such ice is known as shore-fast ice, land-fast ice, or simply fast ice (ACIA 2005, p. 30). Fast ice is found along much of the Siberian coast, the White Sea (an inlet of the Barents Sea), north of Greenland, the Canadian Archipelago, Hudson Bay, and north of Alaska (ACIA 2005, p. 457).

Pack ice consists of seasonal (or first-year) and multi-year ice that is in constant motion caused by winds and currents (USFWS 1995, pp. 7-9). Pack ice is used by polar bears for traveling, feeding, and denning, and it is the primary summer habitat for polar bears, including the Southern Beaufort Sea and Chukchi Sea populations, as first year ice retreats and melts with the onset of spring (see “Polar Bear-Sea Ice Habitat Relationships” section for more detail on ice types used by polar bears). Movements of sea ice are related to winds, currents, and seasonal temperature fluctuations that in turn promote its formation and degradation. Ice flow in the Arctic often includes a clockwise circulation of sea ice within the Canada Basin and a transpolar drift stream that carries sea ice from the Siberian shelves to the Barents Sea and Fram Strait.

Sea ice is an important component of the Arctic climate system (ACIA 2005, p. 456). It is an effective insulator between the oceans and the atmosphere. It also strongly reduces the ocean-atmosphere heat exchange and reduces wind stirring of the ocean. In contrast to the dark ocean, pond-free sea ice (i.e., sea ice that has no meltwater ponds on the surface) reflects most of the solar radiation back into space. Together with snow cover, sea ice greatly restricts the penetration of light into the sea, and it also provides a surface for particle and snow deposition (ACIA 2005, p. 456). Its effects can extend far south of the Arctic, perhaps globally, e.g., through impacting deepwater formation that influences global ocean circulation (ACIA 2005, p. 32).

Sea ice is also an important environmental factor in Arctic marine ecosystems. “Several physical factors combine to make arctic marine systems unique including: a very high proportion of continental shelves and shallow water; a dramatic seasonality and overall low level of sunlight; extremely low water temperatures; presence of extensive areas of multi-year and seasonal sea-ice cover; and a strong influence from freshwater, coming from rivers and ice melt” (ACIA 2005, p. 454). Ice cover is an important physical characteristic, affecting heat exchange between water and atmosphere, and light penetration to organisms in the water below. It also helps determine the depth of the mixed layer, and provides a biological habitat above, within, and beneath the ice. The marginal ice zone, at the edge of the pack ice, is important for plankton production and plankton-feeding fish (ACIA 2005, p. 456)

Observed Changes in Arctic Sea Ice

Sea ice is the defining physical characteristic of the marine Arctic environment and has a strong seasonal cycle (ACIA 2005, p. 30). There is considerable inter-annual variability both in the maximum and minimum extent of sea ice, but it is typically at its maximum extent in March and minimum extent in September (Parkinson et al. 1999, p. 20, 840). In addition, there are decadal and inter-decadal fluctuations to sea ice extent due to changes in atmospheric pressure patterns and their associated winds, river runoff, and influx of Atlantic and Pacific waters (Gloersen 1995, p. 505; Mysak and Manak 1989, p. 402; Kwok 2000, p. 776; Parkinson 2000b, p. 10; Polyakov et al. 2003, p. 2,080; Rigor et al. 2002, p. 2,660; Zakharov 1994, p. 42). Sea ice “extent” is normally defined as the area of the ocean with at least 15 percent ice coverage, and sea ice “area” is normally defined as the integral sum of areas actually covered by sea ice

(Parkinson et al. 1999). “Area” is a more precise measure of the areal extent of the ice itself, since it takes into account the fraction of leads (linear openings or cracks in the ice) within the ice, but “extent” is more reliably observed (Zhang and Walsh 2006). The following sections discuss specific aspects of observed sea ice changes of relevance to polar bears.

Summer Sea Ice

Summer sea ice area and sea ice extent are important factors for polar bear survival (see “Polar Bear-Sea Ice Habitat Relationships” section). Seasonal or first-year ice that remains at the end of the summer melt becomes multi-year (or perennial) ice. The amount and thickness of perennial ice is an important determinant of future sea ice conditions (i.e., gain or loss of ice) (Holland and Bitz 2003; Bitz and Roe 2004). Much of the following discussion focuses on summer sea ice extent (rather than area).

Prior to the early 1970s, ice extent was measured with visible-band satellite imagery and aircraft and ship reports. With the advent of passive microwave (PM) satellite observations, beginning in December 1972 with a single channel instrument and then more reliably in October 1978 with a multi-channel instrument, we have a more accurate, 3-decade record of changes in summer sea ice extent and area. Over the period since October 1978, successive papers have documented an overall downward trend in Arctic sea ice extent and area. For example, Parkinson et al. (1999) calculated Arctic sea ice extents, areas, and trends for late 1978 through the end of 1996, and documented a decrease in summer sea ice extent of 4.5 percent per decade. Comiso (2002) documented a decline of September minimum sea ice extent of 6.7 percent plus or minus 2.4 percent per decade from 1981 through 2000. Stroeve et al. (2005) analyzed data from 1978 through 2004, and calculated a decline in minimum sea ice extent of 7.7 percent plus or minus 3 percent per decade. Comiso (2006, p. 72) included observations for 2005, and calculated a per-decade decline in minimum sea ice extent of up to 9.8 percent plus or minus 1.5 percent. Most recently, Stroeve et al. (2007, pp. 1-5) estimated a 9.1 percent per-decade decline in September sea ice extent for 1979-2006, while Serreze et al. (2007, pp. 1,533-1,536) calculated a per-decade decline of 8.6 percent plus or minus 2.9 percent for the same parameter over the same time period. These estimates differ only because Serreze et al. (2007, pp. 1,533-1,536) normalized the trend by the 1979-2000 mean, in order to be consistent with how the National Snow and Ice Data Center
1

calculates its estimates (J. Stroeve, in litt. to the Service, November 2007). This decline translates to a decrease of 60,421 sq km (23,328 sq mi) per year (NSIDC Press Release, October 3, 2006).

1
The NSIDC is part of the University of Colorado Cooperative Institute for Research in Environmental Sciences (CIRES), is funded largely by the National Aeronautics and Space Administration (NASA), and is affiliated with the National Oceanic and Atmospheric Administration (NOAA) National Geophysical Data Center through a cooperative agreement. A large part of NSIDC is the Polar Distributed Active Archive Center, which is funded by NASA.

The rate of decrease in September sea ice extent appears to have accelerated in recent years, although the acceleration to date has not been shown to be statistically significant (C. Bitz, in litt. to the Service, November 2007). The years 2002 through 2007 all exceeded previous record lows (Stroeve et al. 2005; Comiso 2006; Stroeve et al. 2007, pp. 1-5; Serreze et al. 2007, pp. 1,533-1,536; NSIDC Press Release, October 1, 2007), and 2002, 2005, and 2007 had successively lower record-breaking minimum extent values (
http://www.nsidc.org
). The 2005 absolute minimum sea ice extent of 5.32 million sq km (2.05 million sq mi) for the entire Arctic Ocean was a 21 percent reduction compared to the mean for 1979 to 2000 (Serreze et al. 2007, pp. 1,533-1,536). Nghiem et al. (2006) documented an almost 50 percent reduction in perennial (multi-year) sea ice extent in the East Arctic Ocean (0 to 180 degrees east longitude) between 2004 and 2005, while the West Arctic Ocean (0 and 180 degrees west longitude) had a slight gain during the same period, followed by an almost 70 percent decline from October 2005 to April 2006. Nghiem et al. (2007) found that the extent of perennial sea ice was significantly reduced by 23 percent between March 2005 and March 2007 as observed by the QuikSCAT/SeaWinds satellite scatterometer. Nghiem et al. (2006) presaged the extensive decline in September sea ice extent in 2007 when they stated: “With the East Arctic Ocean dominated by seasonal ice, a strong summer melt may open a vast ice-free region with a possible record minimum ice extent largely confined to the West Arctic Ocean.”

Arctic sea ice declined rapidly to unprecedented low extents in summer 2007 (Stroeve et al. 2008). On August 16-17, 2007, Arctic sea ice surpassed the previous single-day (absolute minimum) record for the lowest extent ever measured by satellite (set in 2005), and the sea ice was still melting (NSIDC Arctic Sea Ice News, August 17, 2007). On September 16, 2007 (the end of the melt season), the 5-day running mean sea ice extent reported by NSIDC was 4.13 million sq km (1.59 million sq mi), an all-time record low. This was 23 percent lower than the previous record minimum reported in 2005 (see Figure 3) (Stroeve et al. 2008) and 39 percent below the long-term average from 1979 to 2000 (see Figure 4) (NSIDC Press Release, October 1, 2007). Arctic sea ice receded so much in 2007 that the so-called “Northwest Passage” through the straits of the Canadian Arctic Archipelago completely opened for the first time in recorded history (NSIDC Press Release, October 1, 2007). Based on a time-series of data from the Hadley Centre, extending back before the advent of the PM satellite era, sea ice extent in mid-September 2007 may have fallen by as much as 50 percent from the 1950s to 1970s (Stroeve et al. 2008). The minimum September Arctic sea ice extent since 1979 is now declining at a rate of approximately 10.7 percent per decade (Stroeve et al. 2008), or approximately 72,000 sq km (28,000 sq mi) per year (see Figure 3 below) (NSIDC Press Release, October 1, 2007).

ER15MY08.004

ER15MY08.005

In August 2007, Arctic sea ice area (recall that “area” is a different metric than “extent” used in the preceding paragraphs) also broke the record for the minimum Arctic sea ice area in the period since the satellite PM record began in the 1970s (University of Illinois Polar Research Group 2007 web site;
http://arctic.atmos.uiuc.edu/cryosphere/
). The new record was set a full month before the historic summer minimum typically occurs, and the record minimum continued to decrease over the next several weeks (University of Illinois Polar Research Group 2007 web site). The Arctic sea ice area reached an historic minimum of 2.92 million sq km (1.13 million sq mi) on September 16, 2007, which was 27 percent lower than the previous (2005) record Arctic ice minimum area (University of Illinois Polar Research Group 2007 web site). In previous record sea ice minimum years, ice area anomalies were confined to certain sectors (North Atlantic, Beaufort/Bering Sea, etc.), but the character of the 2007 summer sea ice melt was unique in that it was both dramatic and covered the entire Arctic Basin. Atlantic, Pacific, and the central Arctic sectors all showed large negative sea ice area anomalies (University of Illinois Polar Research Group 2007 web site).

Two key factors contributed to the September 2007 extreme sea ice minimum: thinning of the pack ice in recent decades and an unusual pattern of atmospheric circulation (Stroeve et al. 2008). Spring 2007 started out with less ice and thinner ice than normal. Ice thickness estimates from the ICESat satellite laser altimeter instrument indicated ice thicknesses over the Arctic Basin in March 2007 of only 1 to 2 m (3.3 to 6.6 ft) (J. Stroeve, in litt. to the Service, November 2007). Thinner ice takes less energy to melt than thicker ice, so the stage was set for low levels of sea ice in summer 2007 (J. Stroeve, quoted in NSIDC Press Release, October 1, 2007). In general, older sea ice is thicker than younger ice. Maslanik et al. (2007) used an ice-tracking computer algorithm to estimate changes in the distribution of multi-year sea ice of various ages. They estimated: that the area of sea ice at least 5 years old decreased by 56 percent between 1985 and 2007; that ice at least 7 years old decreased from 21 percent of the ice cover in 1988 to 5 percent in 2007; and that sea ice at least 9 years old essentially disappeared from the central Arctic Basin. Maslanik et al. (2007) attributed thinning in recent decades to both ocean-atmospheric circulation patterns and warmer temperatures. Loss of older ice in the late 1980s to mid-1990s was accentuated by the positive phase of the Arctic Oscillation during that period, leading to increased ice export through the Fram Strait (Stroeve et al. 2008). Another significant change since the late 1990s has been the role of the Beaufort Gyre, “the dominant wind and ice drift regime in the central Arctic” (Maslanik et al. 2007). “Since the late 1990s * * * ice typically has not survived the transit through the southern portion of the Beaufort Gyre,” thus not allowing the ice to circulate in its formerly typical clockwise pattern for years while it aged and thickened (Maslanik et al. 2007). Temperature changes in the Arctic are discussed in detail in the section entitled “Air and Sea Temperatures.”

Another factor that contributed to the sea ice loss in the summer of 2007 was an unusual atmospheric pattern, with persistent high atmospheric pressures over the central Arctic Ocean and lower pressures over Siberia (Stroeve et al. 2008). The skies were fairly clear under the high-pressure cell, promoting strong melt. At the same time, the pattern of winds pumped warm air into the region. While the warm winds fostered further melt, they also helped push ice away from the Siberian shore.

Winter Sea Ice

The maximum extent of Arctic winter sea ice cover, as documented with PM satellite data, has been declining at a lower rate than summer sea ice (Parkinson et al. 1999, p. 20,840; Richter-Menge et al. 2006, p. 16), but that rate appears to have accelerated in recent years. Parkinson and Cavalieri (2002, p. 441) reported that winter sea ice cover declined at a rate of 1.8 percent plus or minus 0.6 percent per decade for the period 1979 through 1999. More recently, Richter-Menge et al. (2006, p. 16) reported that March sea ice extent was declining at a rate of 2 percent per decade based on data from 1979-2005, Comiso (2006) calculated a decline of 1.9 plus or minus 0.5 percent per decade for 1979-2006, and J. Stroeve (in litt. to the Service, November 2007) calculated a decline of 2.5 percent per decade, also for 1979-2005.

In 2005 and 2006, winter maximum sea ice extent set record lows for the era of PM satellite monitoring (October 1978 to present). The 2005 record low winter maximum preceded the then-record low summer minimum during the same year, while winter sea ice extent in 2006 was even lower than that of 2005 (Comiso 2006). The winter 2007 Arctic sea ice maximum was the second-lowest in the satellite record, narrowly missing the March 2006 record (NSIDC Press Release, April 4, 2007). J. Stroeve (in litt. to the Service, November 2007) calculated a rate of decline of 3.0 plus or minus 0.8 percent per decade for 1979-2007.

Cumulative Annual Sea Ice

Parkinson et al. (1999) documented that Arctic sea ice extent for all seasons (i.e., annual sea ice extent) declined at a rate of 2.8 percent per decade for the period November 1978 through December 1996, with considerable regional variation (the greatest absolute declines were documented for the Kara and Barents Sea, followed by the Seas of Okhotsk and Japan, the Arctic Ocean, Greenland Sea, Hudson Bay, and Canadian Archipelago; percentage declines were greatest in the Seas of Okhotsk and Japan, at 20.1 percent per decade, and the Kara and Barents Seas, at 10.5 percent per decade). More recently, Comiso and Nishio (2008) utilized satellite data gathered from late 1978 into 2006, and estimated an annual rate decline of 3.4 percent plus or minus 0.2 percent per decade. They also found regions where higher negative trends were apparent, including the Greenland Sea (8.0 percent per decade), the Kara/Barents Seas (7.2 percent per decade), the Okhotsk Sea (8.7 percent per decade), and Baffin Bay/Labrador Sea (8.6 percent per decade). Comiso et al. (2008) included satellite data from 1979 through early September 2007 in their analyses. They found that the trend of the entire sea ice cover (seasonal and perennial sea ice) has accelerated from a decline of about 3 percent per decade in 1979-1996 to a decline of about 10 percent per decade in the last 10 years. Statistically significant negative trends in Arctic sea ice extent now occur n all calendar months (Serreze et al. 2007, pp. 1,533-1,536).

Sea Ice Thickness

Sea ice thickness is an important element of the Arctic climate system. The sea ice thickness distribution influences the sea ice mass budget and ice/ocean/atmosphere exchange (Holland et al. 2006a). Sea ice thickness has primarily been measured with upward-looking sonar on submarines and on moored buoys; this sonar provides information on ice draft, the component of the total ice thickness (about 90 percent) that projects below the water surface (Serreze et al. 2007, pp. 1,533-1,536). Rothrock et al. (1999, p. 3,469) compared sea-ice draft data acquired on submarine cruises between 1993 and 1997 with similar data acquired between 1958 and 1976, and concluded that the mean sea-ice draft at

the end of the melt season (i.e., perennial or multi-year ice) had decreased by about 1.3 m (4.3 ft) in most of the deep water portion of the Arctic Ocean. One limitation of submarine sonar data is sparse sampling, which complicates interpretation of the results (Serreze et al. 2007, pp. 1,533-1,536). Holloway and Sou (2002) noted concerns regarding the temporal and spatial sampling of ice thickness data used in Rothrock et al. (1999), and concluded from their modeling exercise that “a robust characterization over the half-century time series consists of increasing volume to the mid-1960s, decadal variability without significant trend from the mid-1960s to the mid-1980s, then a loss of volume from the mid-1980s to the mid-1990s.” Rothrock et al. (2003, p. 28) conducted further analysis of the submarine-acquired data in conjunction with model simulations and review of other modeling studies, and concluded that all models agree that sea ice thickness decreased between 0.6 and 0.9 m (2 and 3 ft) from 1987 to 1996. Their model showed a modest recovery in thickness from 1996 to 1999. Yu et al. (2004, p. 11) further analyzed submarine sonar data and concluded that total ice volume decreased by 32 percent from the 1960s and 1970s to the 1990s in the central Arctic Basin.

Fowler et al. (2004) utilized a new technique for combining remotely-sensed sea ice motion and sea ice extent to “track” the evolution of sea ice in the Arctic region from October 1978 through March 2003. Their analysis revealed that the area of the oldest sea ice (i.e., sea ice older than 4 years) was decreasing in the Arctic Basin and being replaced by younger (first-year) ice. The extent of the older ice was retreating to a relatively small area north of the Canadian Archipelago, with narrow bands spreading out across the central Arctic (Fowler et al. 2004, pp. 71-74). More recently, Maslanik et al. (2007) documented a substantial decline in the percent coverage of old ice within the central Arctic Basin. In 1987, 57 percent of the ice pack in this area was 5 or more years old, with 25 percent of this ice at least 9 years old. By 2007, only 7 percent of the ice pack in this area was 5 or more years old, and ice at least 9 years old had completely disappeared. This is significant because older ice is thicker than younger ice, and therefore requires more energy to melt. The reduction in the older ice types in the Arctic Basin translates into a reduction in mean ice thickness from 2.6 m in March 1987 to 2.0 m in March 2007 (Stroeve et al. 2008).

Kwok (2007, p. 1) studied six annual cycles of perennial (multi-year) Arctic sea ice coverage, from 2000 to 2006, and found that after the 2005 summer melt, only about four percent of the thin, first-year ice that formed the previous winter survived to replenish the multi-year sea ice area (NASA/JPL News Release, April 3, 2007). That was the smallest amount of multi-year ice replenishment documented in the study, and resulted in perennial ice coverage in January 2006 that was 14 percent smaller than in January 2005. Kwok (2007, p. 1) attributed the decline to unusually high amounts of ice exported from the Arctic in the summer of 2005, and also to an unusually warm winter and summer prior to September 2005.

Length of the Melt Period

The length of the melt period (or season) affects sea ice cover (extent and area) and sea ice thickness (Hakkinen and Mellor 1990; Laxon et al. 2003). In general terms, earlier onset of melt and lengthening of the melt season result in decreased total sea ice cover at the end of summer (i.e., the end of the melt season) (Stroeve et al. 2005, p. 3). Belchansky et al. (2004, p. 1) found that changes in multi-year ice area measured in January were significantly correlated with duration of the intervening melt season. Kwok found a correlation between the number of freezing and melting temperature days and area of multi-year sea ice replenished in a year (NASA/JPL News Release, April 3, 2007).

Comiso (2003, p. 3,506), using data for the period 1981-2001, calculated that the Arctic sea ice melt season was increasing at a rate of 10 to 17 days per decade during that period. Including additional years in his analyses, Comiso (2005, p. 50) subsequently found that the length of the melt season was increasing at a rate of approximately 13.1 days per decade. Stroeve et al. (2006 pp. 367-374) analyzed melt season duration and melt onset and freeze-up dates from satellite passive microwave data for the period 1979 through 2005, and found that the Arctic is experiencing an overall lengthening of the melt season at a rate of about 2 weeks per decade.

The NSIDC documented a trend of earlier onset of the melt season for the years 2002 through 2005; the melt season arrived earliest in 2005, occurring approximately 17 days before the mean date of onset of the melt season (NSIDC 2005, p. 6). In 2007, in addition to the record-breaking September minimum sea ice extent, NSIDC scientists noted that the date of the lowest sea ice extent shifted to later in the year (NSIDC Press Release, October 1, 2007). The minimum sea ice extent occurred on September 16, 2007; from 1979 to 2000, the minimum usually occurred on September 12. This is consistent with a lengthening of the melt season.

Parkinson (2000) documented a clear decrease in the length of the sea ice season throughout the Greenland Sea, Kara and Barents Seas, Sea of Okhotsk, and most of the central Arctic Basin. On the basis of observational data, Stirling et al. (cited in Derocher et al. 2004) calculated that break-up of the annual ice in Western Hudson Bay is occurring approximately 2.5 weeks earlier than it did 30 years ago. Consistent with these results, Stirling and Parkinson (2006) analyzed satellite data for Western Hudson Bay for November 1978 through 2004 and found that, on average, ice break-up has been occurring about 7 to 8 days earlier per decade. Stirling and Parkinson (2006) also investigated ice break-up in Foxe Basin, Baffin Bay, Davis Strait, and Eastern Hudson Bay in Canada. They found that ice break-up in Foxe Basin has been occurring about 6 days earlier each decade and ice break-up in Baffin Bay has been occurring 6 to 7 days earlier per decade. Long-term results from Davis Strait were not conclusive, particularly because the maximum percentage of ice cover in Davis Strait varies considerably more between years than in western Hudson Bay, Foxe Basin, or Baffin Bay. Conversely, Stirling and Parkinson (2006) documented a negative short-term trend from 1991 to 2004 in Davis Strait. In eastern Hudson Bay, there was not a statistically significant trend toward earlier break-up.

Understanding Observed Declines in Arctic Sea Ice

The observed declines in the extent of Arctic sea ice are well documented, and more pronounced in the summer than in the winter. There is also evidence that the rate of sea ice decline is increasing. This decline in sea ice is of great importance to our determination regarding the status of the polar bear. Understanding the causes of the decline is also of great importance in assessing what the future might hold for Arctic sea ice, and, thus, considerable effort has been devoted to enhancing our understanding. This understanding will inform our determination regarding the status of the polar bear within the foreseeable future as determined in this rule.

In general terms, sea ice declines can be attributed to three conflated factors: warming, atmospheric changes (including circulation and clouds), and changes in oceanic circulation (Stroeve and Maslowski 2007). Serreze et al.

(2007, pp. 1,533-1,536) characterize the decline of sea ice as a conflation of thermodynamic and dynamic processes: “Thermodynamic processes involve changes in surface air temperature (SAT), radiative fluxes, and ocean conditions. Dynamic processes involve changes in ice circulation in response to winds and ocean currents.” In the following paragraphs we discuss warming, changes in the atmosphere, and changes in oceanic circulation, followed by a synthesis. It is critically important that we understand the dynamic forces that govern all aspects of sea ice given the polar bear's almost exclusive reliance on this habitat.

Air and Sea Temperatures

Estimated rates of change in surface air temperature (SAT) over the Arctic Ocean over the past 100 or more years vary depending on the time period, season, and data source used (Serreze et al. 2007, pp. 1,533-1,536). Serreze et al. (2007, pp. 1,533-1,536) note that, although natural variability plays a large role in SAT variations, the overall pattern has been one of recent warming.

Polyakov et al. (2003) compiled SAT trends for the maritime Arctic for the period 1875 through 2000 (as measured by coastal land stations, drifting ice stations, and Russian North Pole stations) and found that, since 1875, the Arctic has warmed by 1.2 degrees Celsius (C), an average warming of 0.095 degree C per decade over the entire period, and an average warming of 0.05 ± 0.04 degree C per decade during the 20th century. The increases were greatest in winter and spring, and there were two relative maxima during the century (the late 1930s and the 1990s). The ACIA analyzed land-surface air temperature trends as recorded in the Global Historical Climatology Network (GHCN) database, and documented a statistically significant warming trend of 0.09 degree C per decade during the period 1900-2003 (ACIA 2005, p. 35). For periods since 1950, the rate of temperature increase in the marine Arctic documented in the GHCN (ACIA 2005, p. 35) is similar to the increase noted by Polyakov et al. (2003).

Rigor et al. (2000) documented positive trends in SAT for 1979 to 1997; the trends were greatest and most widespread in spring. Comiso (2006) analyzed data from the Advanced Very High Resolution Radiometer (AVHRR) for 1981 to 2005, and documented an overall warming trend of 0.54 ± 0.11 degrees C per decade over sea ice. Comiso noted that “it is apparent that significant warming has been occurring in the Arctic but not uniformly from one region to another.” The Serreze et al. (2007, pp. 1,533-1,536) assessment of data sets from the National Centers for Environmental Prediction and the National Center for Atmospheric Research indicated strong surface and low-level warming for the period 2000 to 2006 relative to 1979 to 1999, consistent with the observed sea ice losses.

Stroeve and Maslowski (2007) noted that anomalously high temperatures have been consistent throughout the Arctic since 2002. Further support for warming comes from studies indicating earlier onset of spring melt and lengthening of the melt season (e.g., Stroeve et al. 2006, pp. 367-374), and data that point to increased downward radiation toward the surface, which is linked to increased cloud cover and water vapor (Francis and Hunter 2006, cited in Serreze et al. 2007, pp. 1,533-1,536).

According to the IPCC AR4 (IPCC 2007, p. 36), 11 of 12 years from 1995 to 2006 (the exception being 1996) were among the 12 warmest years on record since 1850; 2005 and 1998 were the warmest two years in the instrumental global surface air temperature record since 1850. Surface temperatures in 1998 were enhanced by the major 1997-1998 El Niño but no such large-scale atmospheric anomaly was present in 2005. The IPCC AR4 concludes that the “warming in the last 30 years is widespread over the globe, and is greatest at higher northern latitudes (IPCC 2007, p. 37).” Further, the IPCC AR4 states that greatest warming has occurred in the northern hemisphere winter (December, January, February) and spring (March, April, May). Average Arctic temperatures have been increasing at almost twice the rate of the rest of the world in the past 100 years. However, Arctic temperatures are highly variable. A slightly longer Arctic warm period, almost as warm as the present, was observed from 1925 to 1945, but its geographical distribution appears to have been different from the recent warming since its extent was not global.

Finally, Comiso (2005, p. 43) determined that for each 1 degree C increase in surface temperature (global average) there is a corresponding decrease in perennial sea ice cover of about 1.48 million sq km (0.57 million sq mi).

Changes in Atmospheric Circulation

Links have also been established between sea ice loss and changes in sea ice circulation associated with the behavior of key atmospheric patterns, including the Arctic Oscillation (AO; also called the Northern Annular Mode (NAM)) (e.g., Thompson and Wallace 2000; Limpasuvan and Hartmann 2000) and the more regional, but closely related North Atlantic Oscillation (NAO; e.g., Hurrell 1995). First described in 1998 by atmospheric scientists David Thompson and John Wallace, the Arctic Oscillation is a measure of air-pressure and wind patterns in the Arctic. In the so-called “positive phase” (or high phase), air pressure over the Arctic is lower than normal and strong westerly winds occur in the upper atmosphere at high latitudes. In the so-called “negative phase” (or low phase), air pressure over the Arctic is higher than normal, and the westerly winds are weaker.

Rigor et al. (2002, cited in Stroeve and Maslowski 2007) showed that when the AO is positive in winter, altered wind patterns result in more offshore ice motion and ice divergence along the Siberian and Alaskan coastlines; this leads to the production of more extensive areas of thinner, first-year ice that requires less energy to melt. Rigor and Wallace (2004, cited in Deweaver 2007) suggested that the recent reduction in September ice extent is a delayed reaction to the export of multi-year ice during the high-AO winters of 1989 through 1995. They estimated that the recovery of sea ice to its normal extent should take between 10 and 15 years. However, Rigor and Wallace (2004) estimated that the combined winter and summer AO-indices can explain less than 20 percent of the variance in summer sea ice extent in the western Arctic Ocean where most of the recent reductions in sea ice cover have occurred. The notion that AO-related export of multi-year ice from the Arctic is the principal cause of observed declines in Arctic sea ice extent has been questioned by several authors, including Overland and Wang (2005), Comiso (2006), Stroeve and Maslowski (2007), Serreze et al. (2007, pp. 1,533-1,536), and Stroeve et al. (2008) who note that sea ice extent has not recovered despite the return of the AO to a more neutral state since the late 1990s. Overland and Wang (2005) noted that the return of the AO to a more neutral state was accompanied by southerly wind anomalies from 2000-2005 which contributed to reducing the ice cover over time and “conditioning” the Arctic for the extensive summer sea ice reduction in 2007 (J. Overland NOAA, pers. comm. to FWS, 2007). Maslanik et al. (2007) reached a similar conclusion that despite the return of the AO to a more neutral state, wind and ice transport patterns that favor reduced ice cover in the western and central Arctic continued to play a role in the loss of sea ice in those regions. Maslanik et al.

(2007) believe that circulation patterns such as the Beaufort Gyre, which in the past helped to maintain old ice in the Arctic Basin, are now acting to export ice, as the multi-year ice is no longer surviving the transport through the Chukchi and East Siberian Seas.

According to DeWeaver (2007): “Recognizing the need to incorporate AO variability into considerations of recent sea ice decline, Lindsay and Zhang (2005) used an ocean-sea ice model to reconstruct the sea ice behavior of the satellite era and identify separate contributions from ice motion and thermodynamics. Similar experiments with similar results were also reported by Rothrock and Zhang (2005) and Koberle and Gerdes (2003).” Rothrock and Zhang (2005, cited in Serreze et al. 2007, pp. 1,533-1,536), using a coupled ice-ocean model, argued that although wind forcing was the dominant driver of declining ice thickness and volume from the late 1980s through the mid-1990s, the ice response to generally rising air temperatures was more steadily downward over the study period (1948 to 1999). “In other words, without wind forcing, there would still have been a downward trend in ice extent, albeit smaller than that observed” (Serreze et al. 2007, pp. 1,533-1,536). Lindsay and Zhang (2005, cited in Serreze et al. 2007, pp. 1,533-1,536) came to similar conclusions in their modeling study: “Rising air temperature reduced ice thickness, but changes in circulation also flushed some of the thicker ice out of the Arctic, leading to more open water in summer and stronger absorption of solar radiation in the upper (shallower depths of the) ocean. With more heat in the ocean, thinner ice grows in autumn and winter.”

Changes in Oceanic Circulation

According to Serreze et al. (2007, pp. 1,533-1,536), it appears that changes in ocean heat transport have played a role in declining Arctic sea ice extent in recent years. Warm Atlantic waters enter the Arctic Ocean through the Fram Strait and Barents Sea (Serreze et al. 2007, pp. 1,533-1,536). This water is denser than colder, fresher (less dense) Arctic surface waters, and sinks (subducts) to form an intermediate layer between depths of 100 and 800 m (328 and 2,624 ft) (Quadfasel et al. 1991) with a core temperature significantly above freezing (DeWeaver 2007; Serreze et al. 2007, pp. 1,533-1,536). Hydrographic data show increased import of Atlantic-derived waters in the early to mid-1990s and warming of this inflow (Dickson et al. 2000; Visbeck et al. 2002). This trend has continued, characterized by pronounced pulses of warm inflow (Serreze et al. 2007, pp. 1,533-1,536). For example, strong ocean warming in the Eurasian Basin of the Arctic Ocean in 2004 can be traced to a pulse entering the Norwegian Sea in 1997-1998 and passing through Fram Strait in 1999 (Polyakov et al. 2007). The anomaly found in 2004 was tracked through the Arctic system and took about 1.5 years to travel from the Norwegian Sea to the Fram Strait region, and an additional 4.5-5 years to reach the Laptev Sea slope (Polyakov et al. 2007).

Polyakov et al. (2007) reported that mooring-based records and oceanographic surveys suggest that a new pulse of anomalously warm water entered the Arctic Ocean in 2004. Further Polyakov et al. (2007) stated that: “combined with data from the previous warm anomaly * * * this information provides evidence that the Nansen Basin of the Arctic Ocean entered a new warm state. These two warm anomalies are progressing towards the Arctic Ocean interior * * * but still have not reached the North Pole observational site. Thus, observations suggest that the new anomalies will soon enter the central Arctic Ocean, leading to further warming of the polar basin. More recent data, from summer 2005, showed another warm anomaly set to enter the Arctic Ocean through the Fram Strait (Walczowski and Piechura 2006). These inflows may promote ice melt and discourage ice growth along the Atlantic ice margin (Serreze et al. 2007, pp. 1,533-1,536).

Once Atlantic water enters the Arctic Ocean, the cold halocline layer (CHL) separating the Atlantic and surface waters largely insulates the ice from the heat of the Atlantic layer. Observations suggest a retreat of the CHL in the Eurasian basin in the 1990s (Steele and Boyd 1998, cited in Serreze et al. 2007, pp. 1,533-1,536). This likely increased Atlantic layer heat loss and ice-ocean heat exchange (Serreze et al. 2007, pp. 1,533-1,536), which would serve to erode the edge of the sea ice on a year-round basis (C. Bitz, in litt. to the Service, November 2007). Partial recovery of the CHL has been observed since 1998 (Boyd et al. 2002, cited in Serreze et al. 2007, pp. 1,533-1,536), and future behavior of the CHL is an uncertainty in projections of future sea ice loss (Serreze et al. 2007, pp. 1,533-1,536).

Synthesis

From the previous discussion, surface air temperature warming, changes in atmospheric circulation, and changes in oceanic circulation have all played a role in observed declines of Arctic sea ice extent in recent years.

According to DeWeaver (2007): “Lindsay and Zhang (2005) propose a three-part explanation of sea ice decline,” which incorporates both natural AO variability and warming climate. In their explanation, a warming climate preconditions the ice for decline as warmer winters thin the ice, but the loss of ice extent is triggered by natural variability such as flushing by the AO. Sea ice loss continues after the flushing because of the sea-ice albedo feedback mechanism which warms the sea even further. In recent years, flushing of sea ice has continued through other mechanisms despite a relaxation of the AO since the late 1990s. The sea-ice albedo feedback effect is the result of a reduction in the extent of brighter, more reflective sea ice or snow, which reflects solar energy back into the atmosphere, and a corresponding increase in the extent of darker, more absorbing water or land that absorbs more of the sun's energy. This greater absorption of energy causes faster melting, which in turn causes more warming, and thus creates a self-reinforcing cycle or feedback loop that becomes amplified and accelerates with time. Lindsay and Zhang (2005, p. 4,892) suggest that the sea-ice albedo feedback mechanism caused a tipping point in Arctic sea ice thinning in the late 1980s, sustaining a continual decline in sea ice cover that cannot easily be reversed. DeWeaver (2007) believes that the work of Lindsay and Zhang (2005) suggests that the observed record of sea ice decline is best interpreted as a combination of internal variability and external forcing (via GHGs), and raises the possibility that the two factors may act in concert rather than as independent agents.

Evidence that warming resulting from GHG forcing has contributed to sea ice declines comes largely from model simulations of the late 20th century climate. Serreze et al. (2007, pp. 1,533-1,536) summarized results from Holland et al. (2006, pp. 1-5) and Stroeve et al. (2007, pp. 1-5), and concluded that the qualitative agreement between model results and actual observations of sea ice declines over the PM satellite era is strong evidence that there is a forced component to the decline. This is because each of these models would be in its own phase of natural variability and thus could show an increase or decrease in sea ice, but the fact that they all show a decrease indicates that more than natural variability is involved, i.e., that external forcing by GHGs is a factor. In addition, the model results do not show a decline if they are not forced with the observed GHGs. Serreze et al.

(2007, pp. 1,533-1,536) concluded: “These results provide strong evidence that, despite prominent contributions of natural variability in the observed record, GHG loading has played a role.”

Hegerl et al. (2007) used a new approach to reconstruct and attribute a 1,500-year temperature record for the Northern Hemisphere. Based on their analysis to detect and attribute temperature change over that period, they estimated that about a third of the warming in the first half of the 20th century can be attributed to anthropogenic GHG emissions. In addition, they estimated that the magnitude of the anthropogenic signal is consistent with most of the warming in the second half of the 20th century being anthropogenic.

Observed Changes in Other Key Parameters

Snow Cover on Ice

Northern Hemisphere snow cover, as documented by satellite over the 1966 to 2005 period, decreased in every month except November and December, with a step like drop of 5 percent in the annual mean in the late 1980s (IPCC 2007, p. 43). April snow cover extent in the Northern Hemisphere is strongly correlated with temperature in the region between 40 and 60 degrees N Latitude; this reflects the feedback between snow and temperature (IPCC 2007, p. 43).

The presence of snow on sea ice plays an important role in the Arctic climate system (Powell et al. 2006). Arctic sea ice is covered by snow most of the year, except when the ice first forms and during the summer after the snow has melted (Sturm et al. 2006). Warren et al. (1999, cited in IPCC 2007 Chapter 4) analyzed 37 years (1954-1991) of snow depth and density measurements made at Soviet drifting stations on multi-year Arctic sea ice. They found a weak negative trend for all months, with the largest being a decrease of 8 cm (3.2 in) (23 percent) in May.

Precipitation

The Arctic Climate Impact Assessment (2005) concluded that “overall, it is probable that there was an increase in arctic precipitation over the past century.” An analysis of data in the Global Historical Climatology Network (GHCN) database indicated a significant positive trend of 1.4 percent per decade (ACIA 2005) for the period 1900 through 2003. New et al. (2001, cited in ACIA 2005)) used uncorrected records and found that terrestrial precipitation averaged over the 60 degree to 80 degree N latitude band exhibited an increase of 0.8 percent per decade over the period from 1900 to 1998. In general, the greatest increases were observed in autumn and winter (Serreze et al. 2000). According to the ACIA (2005) calculations: (1) during the Arctic warming in the first half of the 20th century (1900-1945), precipitation increased by about 2 percent per decade, with significant positive trends in Alaska and the Nordic region; (2) during the two decades of Arctic cooling (1946-1965), the high-latitude precipitation increase was roughly 1 percent per decade, but there were large regional contrasts with strongly decreasing values in western Alaska, the North Atlantic region, and parts of Russia; and (3) since 1966, annual precipitation has increased at about the same rate as during the first half of the 20th century. The ACIA report (2005) notes that these trends are in general agreement with results from a number of regional studies (e.g., Karl et al. 1993; Mekis and Hogg 1999; Groisman and Rankova 2001; Hanssen-Bauer et al. 1997; Førland et al. 1997; Hanssen-Bauer and Førland 1998). In addition to the increase, changes in the characteristics of precipitation have also been observed (ACIA 2005). Much of the precipitation increase appears to be coming as rain, mostly in winter and to a lesser extent in autumn and spring. The increasing winter rains, which fall on top of existing snow, cause faster snowmelt. Increased rain in late winter and early spring could affect the thermal properties of polar bear dens (Derocher et al. 2004), thereby negatively impacting cub survival. Increased rain in late winter and early spring may even cause den collapse (Stirling and Smith 2004).

According to the IPCC AR4 (2007, pp. 256-258), distinct upward trends in precipitation are evident in many regions at higher latitudes, especially from 30 to 85 degrees N latitude. Winter precipitation has increased at high latitudes, although uncertainties exist because of changes in undercatch, especially as snow changes to rain (IPCC 2007, p. 258). Annual precipitation for the circumpolar region north of 50 degrees N has increased during the past 50 years by approximately 4 percent but this increase has not been homogeneous in time and space (Groisman et al. 2003, 2005, both cited in IPCC 2007, p. 258). According to the IPCC AR4: “Statistically significant increases were documented over Fennoscandia, coastal regions of northern North America (Groisman et al. 2005), most of Canada (particularly northern regions) up until at least 1995 when the analysis ended (Stone et al. 2000), the permafrost-free zone of Russia (Groisman and Rankova 2001) and the entire Great Russian Plain (Groisman et al. 2005, 2007).” That these trends are real, extending from North America to Europe across the North Atlantic, is also supported by evidence of ocean freshening caused by increased freshwater run-off (IPCC 2007, p. 258).

Rain-on-snow events have increased across much of the Arctic. For example, over the past 50 years in western Russia, rain-on-snow events have increased by 50 percent (ACIA 2005). Groisman et al. (2003) considered rain-on-snow trends over a 50-year period (1950-2000) in high latitudes in the northern hemisphere and found an increasing trend in western Russia and decreases in western Canada (the decreasing Canadian trend was attributed to decreasing snow pack). Putkonen and Roe (2003), working on Spitsbergen Island, where the occurrence of winter rain-on-snow events is controlled by the North Atlantic Oscillation, demonstrated that these events are capable of influencing mean winter soil temperatures and affecting ungulate survival. These authors include the results of a climate modeling effort (using the earlier-generation Geophysical Fluid Dynamics Laboratory climate model and a 1 percent per year increase in CO
2
forcing scenario) that predicted a 40 percent increase in the worldwide area of land affected by rain-on-snow events from 1980-1989 to 2080-2089. Rennert et al. (2008) discussed the significance of rain-on-snow events to ungulate survival in the Arctic, and used the dataset European Center for Medium-range Weather Forecasting (ECMWF) European 40 Year (ERA40) Reanalysis (Uppala et al. 2005) to create a climatology of rain-on-snow events for thresholds that impact ungulate populations and permafrost. In addition to contributing to increased incidence of polar bear den collapse, increased rain-on-snow events during the late winter or early spring could also damage or eliminate snow-covered pupping lairs of ringed seals (the polar bear's principal prey), thereby increasing pup exposure and the risk of hypothermia, and facilitating predation by polar bears and Arctic foxes. This could negatively impact ringed seal recruitment.

Projected Changes in Arctic Sea Ice

Background

To make projections about future ecosystem effects that could result from climate change, one must first make projections of changes in physical

climate parameters based on changes in external factors that can affect the physical climate (ACIA 2005). Climate models use the laws of physics to simulate the main components of the climate system (the atmosphere, ocean, land surface, and sea ice) (DeWeaver 2007), and make projections of future climate scenarios-plausible representations of future climate-that are consistent with assumptions about future emissions of GHGs and other pollutants (these assumptions are called “emissions scenarios”) and with present understanding of the effects of increased atmospheric concentrations of these components on the climate (ACIA 2005).

Virtually all climate models use emissions scenarios developed as part of the IPCC effort; specifically the IPCC's
Special Report on Emissions Scenarios
(SRES) (IPCC 2000) details a number of plausible future emissions scenarios based on assumptions on how societies, economies, and energy technologies are likely to evolve. The SRES emissions scenarios were built around four narrative storylines that describe the possible evolution of the world in the 21st century (ACIA 2005, p.119). Around these four narrative storylines the SRES constructed six scenario groups and 40 different emissions scenarios. Six scenarios (A1B, A1T, A1FI, A2, B1, and B2) were then chosen as illustrative “marker” scenarios. These scenarios have been used to estimate a range of future GHG emissions that affect the climate. The scenarios are described on page 18 of the
AR4 Working Group I: Summary for Policymakers
(IPCC 2007), and in greater detail in the SRES Report (IPCC 2000).

The most commonly-used scenarios for current-generation climate modeling are the B1, A1B, and A2 scenarios. In the B1 scenario, CO
2
concentration is around 549 parts per million (ppm) by 2100; this is often termed a ‘low’ scenario. In the A1B scenario, CO
2
concentration is around 717 ppm by the end of the century; this is a 'medium' or ‘middle-of-the-road’ scenario. In the A2 scenario, CO
2
concentration is around 856 ppm at the end of the 21st century; this is considered a ‘high’ scenario with respect to GHG concentrations. It is important to note that the SRES scenarios include no additional mitigation initiatives, which means that no scenarios are included that explicitly assume the implementation of the United Nations Framework Convention on Climate Change (UNFCC) or the emission targets of the Kyoto Protocol.

Of the various types of climate models, the Atmosphere-Ocean General Circulation Models (AOGCMs, also known as General Circulation Models (GCMs)) are acknowledged as the principal and most rapidly-developing tools for simulating the response of the global climate system to various GHG and aerosol emission scenarios. The climates simulated by these models have been verified against observations in several model intercomparison programs (e.g., Achuta Rao et al. 2004; Randall et al. 2007) and have been found to be generally realistic (DeWeaver 2007). Additional confidence in model simulations comes from experiments with a hierarchy of simpler models, in which the dominant processes represented by climate models (e.g., heat and momentum transport by mid-latitude weather systems) can be isolated and studied (DeWeaver 2007).

For projected changes in climate and Arctic sea ice conditions, our proposed rule (72 FR 1064) relied primarily on results in the IPCC's
Third Assessment Report
(TAR) (IPCC 2001b), the
Arctic Climate Impact Assessment
(ACIA 2005, p. 99), and selected peer-reviewed papers (e.g., Johannessen et al. 2004; Holland et al. 2006, pp. 1-5). The IPCC TAR used results derived from 9-AOGCM ensemble (i.e, averaged results from 9 AOGCMs) and three SRES emissions scenarios (A2, B2, and IS92a). The ACIA (2005, p. 99) used a 5-AOGCM ensemble under two SRES emissions scenarios (A2 and B2); however, the B2 emissions scenario was chosen as the primary scenario for use in ACIA analyses (ACIA 2005). These reports relied on ensembles rather than single models, because “no one model can be chosen as 'best' and it is important to use results from a range of models” (IPCC 2001, Chapter 8). The other peer-reviewed papers used in the proposed rule (72 FR 1064) tend to report more-detailed results from a one or two model simulations using one SRES scenario.

After the proposed rule was published (72 FR 1064), the IPCC released its
Fourth Assessment Report
(AR4) (IPCC 2007), a detailed assessment of current and predicted future climates around the globe. Projected changes in climate and Arctic sea ice conditions presented in the IPCC AR4 have been used extensively in this final rule. The IPCC AR4 used results from state-of-the-art climate models that have been substantially improved over the models used in the IPCC TAR and ACIA reports (M. Holland, NCAR, in litt. to the Service, 2007; DeWeaver 2007). In addition, the IPCC AR4 used results from a greater number of models (23) than either the IPCC TAR or ACIA reports. “This larger number of models running the same experiments allows better quantification of the multi-model signal as well as uncertainty regarding spread across the models, and also points the way to probabilistic estimates of future climate change” (IPCC 2007, p. 761). Finally, the IPCC AR4 used a greater number of emissions scenarios (4) than either the IPCC TAR or ACIA reports. The emission scenarios considered in the AR4 include A2, A1B, and B1, as well as a “year 2000 constant concentration” scenario; this choice was made solely due to the limited computational resources for multi-model simulations using comprehensive AOGCMs, and “does not imply any preference or qualification of these three scenarios over the others” (IPCC 2007, p.761). For all of these reasons, there is considerable confidence that the AOGCMs used in the IPCC AR4 provide credible quantitative estimates of future climate change, particularly at continental scales and above (IPCC 2007, p. 591), and we have determined that these results are rightly included in the category of best available scientific information upon which to base a listing decision for the polar bear.

In addition to the IPCC AR4 results, this final rule utilizes results from a large number of peer-reviewed papers (e.g., Parkinson et al. 2006; Zhang and Walsh 2006; Arzel et al. 2006; Stroeve et al. 2007, pp. 1-5; Holland et al. 2006, pp. 1-5; Wang et al. 2007, pp. 1,093-1,107; Overland and Wang 2007a, pp. 1-7; Chapman and Walsh 2007) that provide more detailed information on climate change projections for the Arctic.

Uncertainty in Climate Models

The fundamental physical laws reflected in climate models are well established, and the models are broadly successful in simulating present-day climate and recent climate change (IPCC 2007, cited in DeWeaver 2007). For Arctic sea ice, model simulations unanimously project declines in areal coverage and thickness due to increased GHG concentrations (DeWeaver 2007). They also agree that GHG-induced warming will be largest in the high northern latitudes and that the loss of sea ice will be much larger in summer than in winter (Meehl et al. 2007, cited in DeWeaver 2007). However, despite the qualitative agreement among climate model projections, individual model results for Arctic sea ice decline span a considerable range (DeWeaver 2007). Thus, projections from models are often expressed in terms of the typical

behavior of a group (ensemble) of simulations (e.g., Arzel et al. 2006; Flato et al. 2004; Holland et al. 2006, pp. 1-5).

DeWeaver (2007) presents a detailed analysis of uncertainty associated with climate models and their projections for Arctic sea ice conditions. He concludes that two main sources of uncertainty should be considered in assessing Arctic sea ice simulations: uncertainties in the construction of climate models and unpredictable natural variability of the climate system. DeWeaver (2007) states that while most aspects of climate simulations have some degree of uncertainty, projections of Arctic climate change have relatively higher uncertainty. This higher level of uncertainty is, to some extent, a consequence of the smaller spatial scale of the Arctic, since climate simulations are believed to be more reliable at continental and larger scales (Meehl et al. 2007, IPCC 2007, both cited in DeWeaver 2007). The uncertainty is also a consequence of the complex processes that control the sea ice, and the difficulty of representing these processes in climate models. The same processes which make Arctic sea ice highly sensitive to climate change, the ice-albedo feedback in particular, also make sea ice simulations sensitive to any uncertainties in model physics (e.g., the representation of Arctic clouds) (DeWeaver 2007).

DeWeaver (2007) also discusses natural variability of the climate system. He states that the atmosphere, ocean, and sea ice comprise a “nonlinear chaotic system” with a high level of natural variability unrelated to external climate forcing. Thus, even if climate models perfectly represented all climate system physics and dynamics, inherent climate unpredictability would limit our ability to issue highly, detailed forecasts of climate change, particularly at regional and local spatial scales, into the middle and distant future (DeWeaver 2007).

DeWeaver (2007) states that the uncertainty in model simulations should be assessed through detailed model-to-model and model-to-observation comparisons of sea ice properties like thickness and coverage. In principle, inter-model sea ice variations are attributable to differences in model construction, but attempts to relate simulation differences to specific model differences generally have not been successful (e.g., Flato et al. 2004, cited in DeWeaver 2007). A practical consequence of uncertainty in climate model simulations of sea ice is that a mean and spread of an ensemble of simulations should be considered in deciding the likely fate of Arctic sea ice. Some model-to-model variation (or spread) in future sea ice behaviors is expected even among high-quality simulations due to natural variability, but spread that is a consequence of poor simulation quality should be avoided. Thus, it is desirable to define a selection criterion for membership in the ensemble, so that only those models that demonstrate sufficient credibility in present-day sea ice simulation are included. Fidelity in sea ice hindcasts (i.e., the ability of models to accurately simulate past to present-day sea ice conditions) is an important consideration. This same perspective is shared by other researchers, including Overland and Wang (2007a, p. 1), who state: “Our experience (Overland and Wang 2007b) as well as others (Knutti et al. 2006) suggest that one method to increase confidence in climate projections is to constrain the number of models by removal of major outliers through validating historical simulations against observations. This requirement is especially important for the Arctic.”

Projection Results in the IPCC TAR and ACIA

This section briefly summarizes the climate model projections of the IPCC TAR and the ACIA, the principal reports used in the proposed rule (72 FR 1064), while the following section presents detailed results published subsequent to those reports, including in the IPCC AR4.

All models in the IPCC TAR predicted continued Arctic warming and continued decreases in the Arctic sea ice cover in the 21st century due to increasing global temperatures, although the level of increase varied between models. The TAR projected a global mean temperature increase of 1.4 degree C by the mid-21st century compared to the present climate for both the A2 and B2 scenarios (IPCC 2001b). Toward the end of the 21st century (2071 to 2100), the mean change in global average surface air temperature, relative to the period 1961-1990, was projected to be 3.0 degrees C (with a range of 1.3 to 4.5 degrees C) for the A2 scenario, and 2.2 degrees C (with a range of 0.9 to 3.4 degrees C) for the B2 scenario. Relative to glacier and sea ice change, the TAR reported that “The representation of sea-ice processes continues to improve, with several climate models now incorporating physically based treatments of ice dynamics * * *. Glaciers and ice caps will continue their widespread retreat during the 21st century and Northern Hemisphere snow cover and sea ice are projected to decrease further.”

The ACIA concluded that, for both the A2 and B2 emissions scenarios, models projected mean temperature increases of 2.5 degrees C for the region north of 60 degrees N latitude by the mid-21st century (ACIA 2005, p. 100). By the end of the 21st century, Arctic temperature increases were projected to be 7 degrees C and 5 degrees C for the A2 and B2 scenarios, respectively, compared to the present climate (ACIA 2005, p. 100). Greater warming was projected for the autumn and winter than for the summer (ACIA 2005, p. 100).

The ACIA utilized projections from the five ACIA-designated AOGCMs to evaluate changes in sea ice conditions for three points in time (2020, 2050, and 2080) relative to the climatological baseline (2000) (ACIA 2005, p. 192). In 2020, the duration of the sea ice freezing period was projected to be shorter by 10 days; winter sea ice extent was expected to decline by 6 to 10 percent from baseline conditions; summer sea ice extent was expected to decline such that continental shelves were likely to be ice free; and there would be some reduction in multi-year ice, especially on shelves (ACIA 2005, Table 9.4). In 2050, the duration of the sea ice freezing period was projected to be shorter by 15 to 20 days; winter sea ice extent was expected to decline by 15 to 20 percent; summer sea ice extent was expected to decline 30 to 50 percent from baseline conditions; and there would be significant loss of multi-year ice, with no multi-year ice on shelves. In 2080, the duration of the sea ice freezing period was projected to be shorter by 20 to 30 days; winter sea ice extent was expected to decline such that there probably would be open areas in the high Arctic (Barents Sea and possibly Nansen Basin); summer sea ice extent was expected to decline 50 to 100 percent from baseline conditions; and there would be little or no multi-year ice.

According to ACIA (2005, p. 193), one model indicated an ice-free Arctic during September by the mid-21st century, but this model simulated less than half of the observed September sea-ice extent at the start of the 21st century. None of the other models projected ice-free summers in the Arctic by 2100, although the sea-ice extent projected by two models decreased to about one-third of initial (2000) and observed September values by 2100.

Projection Results in the IPCC AR4 and Additional Projections

The IPCC AR4, released a few months after publication of our proposed listing

rule for the polar bear (72 FR 1064), presents results from state-of-the-art climate models that are substantially improved over models used in the IPCC TAR and ACIA reports (M. Holland, NCAR, in litt. to the Service FWS, 2007; DeWeaver 2007). Results of the AR4 are presented in this section, followed by discussion of several key, peer-reviewed articles that discuss results presented in the AR4 in greater detail or use AR4 simulations to conduct additional, in-depth analyses.

In regard to surface air temperature changes, the IPCC AR4 states that the range of expected globally averaged surface air temperature warming shows limited sensitivity to the choice of SRES emissions scenarios for the early 21st century (between 0.64 and 0.69 degrees C for 2011 to 2030 compared to 1980 to 1999, a range of only 0.05 °C), largely due to climate change that is already committed (IPCC 2007, p. 749). By the mid-21st century (2046-2065), the choice of SRES scenario becomes more important for globally averaged surface air temperature warming (with increases of 1.3 degree C for the B1 scenario, 1.8 degree C for A1B, and 1.7 degree C for A2). During this time period, about a third of that warming is projected to be due to climate change that is already committed (IPCC 2007, p. 749).

The “limited sensitivity” of the results is because the state-of-the-art climate models used in the AR4 have known physics in connecting increases in GHGs to temperature increases through radiation processes (Overland and Wang 2007a, pp. 1-7, cited in J. Overland, NOAA, in litt. to the Service, 2007), and the GHG levels used in the SRES emissions scenarios are relatively similar until around 2040-2050 (see Figure 5). Because increases in GHGs have lag effects on climate and projections of GHG emissions can be extrapolated with greater confidence over the next few decades, model results projecting out for the next 40 to 50 years (near-term climate change estimates) have greater credibility than results projected much further into the future (long-term climate change) (J. Overland, NOAA, in litt. to the Service, 2007). Thus, the uncertainty associated with emissions is relatively smaller for the 45-year “foreseeable future” for the polar bear listing. After 2050, uncertainty associated with various climate mechanisms and policy/societal changes begins to increase, as reflected in the larger confidence intervals around the trend lines in Figure 5 beyond 2050.

ER15MY08.006

However, even if GHG emissions had stabilized at 2000 levels, the global climate system would already be committed to a warming trend of about 0.1 degree C per decade over the next two decades, in the absence of large changes in volcanic or solar forcing. Meehl et al. (2006) conducted climate change scenario simulations using the Community Climate System Model, version 3 (CCSM3, National Center for Atmospheric Research), with all GHG emissions stabilized at 2000 levels, and found that the global climate system would already be committed to 0.40 degree C more warming by the end of the 21st century.

With respect to warming in the Arctic itself, the AR4 concludes: “At the end of the 21st century, the projected annual warming in the Arctic is 5 degrees C, estimated by the multi-model A1B ensemble mean projection” (see IPCC 2007, p. 908, Fig. 11.21). The across-model range for the A1B scenario varied from 2.8 to 7.8 degrees C. Larger mean warming was found for the A2 scenario (5.9 degrees C), and smaller mean warming was found for the B1 scenario (3.4 degrees C); both with proportional across-model ranges. Chapman and Walsh (2007, cited IPCC 2007, p. 904) concluded that the across-model and across-scenario variability in the projected temperatures are both considerable and of comparable amplitude.

In regard to changes in sea ice, the IPCC AR4 concludes that, under the A1B, A2, and B1 SRES emissions scenarios, large parts of the Arctic Ocean are expected to be seasonally ice free by the end of the 21st century (IPCC 2007, p. 73). Some projections using the A2 and A1B scenarios achieve a seasonally ice-free Arctic by as early as 2080-2090 (IPCC 2007, p.771, Figure 10.13a, b). Sea ice reductions are greater in summer than winter, thus it is summer sea ice cover that is projected to be lost in some models by 2080-2090, not winter sea ice cover. The reduction in sea ice cover is accelerated by positive feedbacks in the climate system, including the ice-albedo feedback (which allows open water to receive more heat from the sun during summer, the insulating effect of sea ice is reduced and the increase in ocean heat transport to the Arctic further reduces ice cover) (IPCC 2007, p. 73).

While the conclusions of the IPCC TAR and AR4 are similar with respect to the Arctic, the confidence level associated with independent reviews of AR4 is greater, owing to improvements in the models used and the greater number of models and emissions scenarios considered (J. Overland, NOAA, in litt. to the Service, 2007). Climate models still have challenges modeling some of the regional differences caused by changing decadal climate patterns (e.g., Arctic Oscillation). To help improve the models further, the evaluation of AR4 models has been on-going both for how well they represent conditions in the 20th century and how their predicted results for the 21st century compare (Parkinson et al. 2006; Zhang and Walsh 2006; Arzel et al. 2006; Stroeve et al. 2007, pp. 1-5; Holland et al. 2006, pp. 1-5; Wang et al. 2007, pp. 1,093-1,107; Chapman and Walsh 2007).

Arzel et al. (2006) and Zhang and Walsh (2006) evaluate the sea ice results from the IPCC AR4 models in more detail. Arzel et al. (2006) investigated projected changes in sea ice extent and volume simulated by 13 AOGCMs (also known as GCMs) driven by the SRES A1B emissions scenario. They found that the models projected an average relative decrease in sea ice extent of 15.4 percent in March, 61.7 percent in September, and 27.7 percent on an annual basis when comparing the periods 1981-2000 and 2081-2100; the average relative decrease in sea ice volume was 47.8 percent in March, 78.9 percent in September, and 58.8 percent on an annual basis when comparing the periods 1981-2000 and 2081-2100. More than half the models (7 of 13) reach ice-free September conditions by 2100, as reported in some previous studies (Gregory et al. 2002, Johannessen et al. 2004, both cited in Arzel et al. 2006).

Zhang and Walsh (2006) investigated changes in sea ice area simulated by 14 AOGCMs driven by the SRES A1B, A2, and B1 emissions scenarios. They found that the annual mean sea ice area during the period 2080-2100 would be decreased by 31.1 percent in the A1B scenario, 33.4 percent in the A2 scenario, and 21.6 percent in the B1 scenario relative to the observed sea ice area during the period 1979-1999. They further determined that the area of multi-year sea ice during the period 2080-2100 would be decreased by 59.7 percent in the A1B scenario, 65.0 percent in the A2 scenario, and 45.8 percent in the B1 scenario relative to the ensemble mean multi-year sea ice area during the period 1979-1999.

Dumas et al. (2006) generated projections of future landfast ice thickness and duration for nine sites in the Canadian Arctic and one site on the Labrador coast using the Canadian Centre for Climate Modelling and Analysis global climate model (CGCM2). For the Canadian Arctic sites the mean maximum ice thickness is projected to decrease by roughly 30 cm (11.8 in) from 1970-1989 to 2041-2060 and by roughly 50-55 cm (19.7-21.7 in) from 1970-1989 to 2081-2100. Further, they projected a reduction in the duration of sea ice cover of 1 and 2 months by 2041-2060 and 2081-2100, respectively, from the baseline period of 1970-1989. In addition simulated changes in freeze-up and break-up revealed a 52-day later freeze-up and 30-day earlier break-up by 2081-2100.

Holland et al. (2006, pp. 1-5) analyzed an ensemble of seven projections of Arctic summer sea ice from the Community Climate System Model, version 3 (CCSM3; National Center for Atmospheric Research, USA) utilizing the SRES A1B emissions scenario. CCSM3 is the model that performed best in simulating the actual observations for Arctic ice extent over the PM satellite era (Stroeve et al. 2007, pp. 1-5). Holland et al. (2006, pp. 1-5) found that the CCSM3 simulations compared well to actual observations for Arctic ice extent over the PM satellite era, including the rate of its recent retreat. They also found that the simulations did not project that sea ice retreat would continue at a constant rate into the future. Instead, the CCSM3 simulations indicate abrupt shifts in the ice cover, with one CCSM3 simulation showing an abrupt transition starting around 2024 with continued rapid retreat for around 5 years. Every CCSM3 run had at least one abrupt event (an abrupt event being defined as a time when a 5-year running mean exceeded three times the 2001-2005 observed retreat) in the 21st century, indicating that near ice-free Septembers could be reached within 30-50 years from now.

Holland et al. (2006, pp. 1-5) also discussed results from 15 additional models used in the IPCC AR4, and concluded that 6 of 15 other models “exhibit abrupt September ice retreat in the A1B scenario runs.” The length of the transition varied from 3 to 8 years among the models. Thus, in these model simulations, it was found that once the Arctic ice pack thins to a vulnerable state, natural variability can trigger an abrupt loss of the ice cover so that seasonally ice-free conditions can happen within a decade's time (J. Stroeve, in litt. to the Service, November 2007).

Finally, Holland et al. (2006, pp. 1-5) noted that the emissions scenario used in the model affected the likelihood of future abrupt transitions. In models using the SRES B1 scenario (i.e., with GHG levels increasing at a slower rate), only 3 of 15 models show abrupt declines lasting from 3 to 5 years. In models using the A2 scenario (i.e., with

GHG levels increasing at a faster rate), 7 of 11 models with available data obtain an abrupt retreat in the ice cover; the abrupt events last from 3 to 10 years (Holland et al. 2006, pp. 1-5).

In order to increase confidence in climate model projections, several studies have sought to constrain the number of models used by validating climate change in the models simulations against actual observations (Knutti et al. 2006; Hall and Ou 2006). The concept is to create a shorter list of “higher confidence” models by removing outlier model projections that do not perform well when compared to 20th century observational data (Overland and Wang 2007a, pp. 1-7). This has been done for temperatures (Wang et al. 2007, pp. 1,093-1,107), sea ice (Overland and Wang 2007a, pp. 1-7; Stroeve et al. 2007, pp. 1-5), and sea level pressure (SLP; defined as atmospheric pressure at sea level) and precipitation (Walsh and Chapman, pers. comm. with J. Overland, NOAA, cited in litt. to the Service, 2007).

Overland and Wang (2007a, pp. 1-7) investigated future regional reductions in September sea ice area utilizing a subset of AR4 models that closely simulate observed regional ice concentrations for 1979-1999 and were driven by the A1B emissions scenario. They used a selection criterion, similar to Stroeve et al. (2007, pp. 1-5), to constrain the number of models used by removing outliers so as to increase confidence in the projections used. Out of an initial set of 20 potential models, 11 models were retained for the Arctic-wide area, 4 were retained for the Kara/Laptev Sea area, 8 were retained for the East Siberian/Chukchi Sea, and 11 were retained for the Beaufort Sea (Overland and Wang 2007a, pp. 1-7). Using these constrained subsets, Overland and Wang (2007a, pp. 1-7) found that there is: “considerable evidence for loss of sea ice area of greater than 40 percent by 2050 in summer for the marginal seas of the Arctic basin. This conclusion is supported by consistency in the selection of the same models across different regions, and the importance of thinning ice and increased open water at mid-century to the rate of ice loss.” More specifically, Overland and Wang (2007a, pp. 1-7) found that “By 2050, 7 of 11 models estimate a loss of 40 percent or greater of summer Arctic ice area. Six of 8 models show a greater than 40 percent ice loss in the East Siberian/Chukchi Seas and 7 of 11 models show this loss for the Beaufort Sea. The percentage of models with major ice loss could be considered higher, as two of the models that retain sea ice are from the same Canadian source and thus cannot be considered to be completely independent. These results present a consistent picture: there is a substantial loss of sea ice for most models and regions by 2050” (see Figure 6). With less confidence, they found that the Bering, Okhotsk, and Barents seas have a similar 40 percent loss of sea ice area by 2050 in winter; Baffin Bay/Labrador shows little change compared to current conditions (Overland and Wang 2007a, pp. 1-7). Overland and Wang (2007a, pp. 1-7) also note that the CCSM3 model (Holland et al. 2006, pp. 1-5) is one of the models with the most rapid ice loss in the 21st century; this model is also one of the best at simulating historical 20th century observations (also see Figure 12 in DeWeaver (2007)).

ER15MY08.007

DeWeaver (2007), applying a similar conceptual approach as Overland and Wang (2007a, pp. 1-7) and Stroeve et al. (2007, pp. 1-5), used a selection criterion to construct an ensemble of 10 climate models that most accurately depicted sea-ice extent, from the 20 models that contributed sea ice data to the AR4. This 10-model ensemble was used by the USGS for assessing potential polar bear habitat loss (Durner et al. 2007). DeWeaver's selection criterion was to include only those models for which the mean 1953-1995 simulated September sea ice extent is within 20 percent of its actual observed value (as taken from the Hadley Center Sea Ice and Sea Surface Temperature (HadISST) data set (Raynor et

[Text truncated at 120,000 characters. The full text is on the page linked above.]

---

Source: Frix Law Library, https://www.frixlaw.com/law-library/documents/fr%3AE8-11105. Public record. Not legal advice.