Endangered and Threatened Wildlife and Plants; 12-Month Finding on a Petition to List the Pacific Walrus as Endangered or Threatened

DEPARTMENT OF THE INTERIOR Fish and Wildlife Service 50 CFR Part 17 [Docket No. FWS-R7-ES-2009-0051; MO 92210-0-0008-B2] Endangered and Threatened Wildlife and Plants; 12-Month Finding on a Petition to List the Pacific Walrus as Endangered or Threatened AGENCY:

Fish and Wildlife Service, Interior.

ACTION:

Notice of 12-month petition finding.

SUMMARY:

We, the U.S. Fish and Wildlife Service, announce a 12-month finding on a petition to list the Pacific walrus ( Odobenus rosmarus divergens ) as endangered or threatened and to designate critical habitat under the Endangered Species Act of 1973, as amended. After review of all the available scientific and commercial information, we find that listing the Pacific walrus as endangered or threatened is warranted. Currently, however, listing the Pacific walrus is precluded by higher priority actions to amend the Lists of Endangered and Threatened Wildlife and Plants. Upon publication of this 12-month petition finding, we will add Pacific walrus to our candidate species list. We will develop a proposed rule to list the Pacific walrus as our priorities allow. We will make any determination on critical habitat during development of the proposed listing rule. Consistent with section 4(b)(3)(C)(iii) of the Endangered Species Act, we will review the status of the Pacific walrus through our annual Candidate Notice of Review.

DATES:

The finding announced in this document was made on February 10, 2011.

ADDRESSES:

This finding and supporting documentation are available on the Internet at http://www.regulations.gov at Docket Number FWS-R7-ES-2009-0051. A range map of the three walrus subspecies and a more detailed map of the Pacific walrus range are available at the following Web site: http://alaska.fws.gov/fisheries/mmm/walrus/wmain.htm. Supporting documentation we used in preparing this finding is available for public inspection, by appointment, during normal business hours at the U.S

//www.regulations.gov at Docket Number FWS-R7-ES-2009-0051. A range map of the three walrus subspecies and a more detailed map of the Pacific walrus range are available at the following Web site: http://alaska.fws.gov/fisheries/mmm/walrus/wmain.htm. Supporting documentation we used in preparing this finding is available for public inspection, by appointment, during normal business hours at the U.S. Fish and Wildlife Service, Alaska Regional Office, 1011 East Tudor Road, Anchorage, AK 99503. Please submit any new information, materials, comments, or questions concerning this finding to the above address.

FOR FURTHER INFORMATION CONTACT:

James MacCracken, Marine Mammals Management, Alaska Regional Office ( see ADDRESSES ); by telephone: 800-362-5148; or by facsimile: 907-786-3816. If you use a telecommunications device for the deaf (TDD), please call the Federal Information Relay Service (FIRS) at 800-877-8339.

SUPPLEMENTARY INFORMATION:

Background

Section 4(b)(3)(B) of the Endangered Species Act of 1973, as amended (Act) (16 U.S.C. 1531 et seq. ), requires that, for any petition to revise the Federal Lists of Endangered and Threatened Wildlife and Plants that contains substantial scientific or commercial information that listing the species may be warranted, we make a finding within 12 months of the date of receipt of the petition. In this finding, we will determine whether the petitioned action is: (a) Not warranted, (b) warranted, or (c) warranted, but the immediate proposal of a regulation implementing the petitioned action is precluded by other pending proposals to determine whether species are endangered or threatened, and expeditious progress is being made to add or remove qualified species from the Federal Lists of Endangered and Threatened Wildlife and Plants

the petitioned action is: (a) Not warranted, (b) warranted, or (c) warranted, but the immediate proposal of a regulation implementing the petitioned action is precluded by other pending proposals to determine whether species are endangered or threatened, and expeditious progress is being made to add or remove qualified species from the Federal Lists of Endangered and Threatened Wildlife and Plants. Section 4(b)(3)(C) of the Act requires that we treat a petition for which the requested action is found to be warranted but precluded as though resubmitted on the date of such finding, that is, requiring a subsequent finding to be made within 12 months. We must publish these 12-month findings in the Federal Register .

Previous Federal Actions

On February 8, 2008, we received a petition dated February 7, 2008, from the Center for Biological Diversity, requesting that the Pacific walrus be listed as endangered or threatened under the Act and that critical habitat be designated. The petition included supporting information regarding the species' ecology and habitat use patterns, and predicted changes in sea-ice habitats and ocean conditions that may impact the Pacific walrus. We acknowledged receipt of the petition in a letter to the Center for Biological Diversity, dated April 9, 2008. In that letter, we stated that an emergency listing was not warranted and that all remaining available funds in the listing program for Fiscal Year (FY) 2008 had already been allocated to the U.S. Fish and Wildlife Service's (Service) highest priority listing actions and that no listing funds were available to further evaluate the Pacific walrus petition in FY 2008.

On December 3, 2008, the Center for Biological Diversity filed a complaint in U.S. District Court for the District of Alaska for declaratory judgment and injunctive relief challenging the failure of the Service to make a 90-day finding on their petition to list the Pacific walrus, pursuant to section 4(b)(3) of the Endangered Species Act, 16 U.S.C

evaluate the Pacific walrus petition in FY 2008.

On December 3, 2008, the Center for Biological Diversity filed a complaint in U.S. District Court for the District of Alaska for declaratory judgment and injunctive relief challenging the failure of the Service to make a 90-day finding on their petition to list the Pacific walrus, pursuant to section 4(b)(3) of the Endangered Species Act, 16 U.S.C. 1533(b)(3), and the Administrative Procedure Act, 5 U.S.C. 706(1). On May 18, 2009, a settlement agreement was approved in the case of Center for Biological Diversity v. U.S. Fish and Wildlife Service, et al. (3:08-cv-00265-JWS), requiring us to submit our 90-day finding on the petition to the Federal Register by September 10, 2009. On September 10, 2009, we made our 90-day finding that the petition presented substantial scientific information indicating that listing the Pacific walrus may be warranted (74 FR 46548). On August 30, 2010, the Court approved an amended settlement agreement requiring us to submit our 12-month finding to the Federal Register by January 31, 2011. This notice constitutes the 12-month finding on the February 7, 2008, petition to list the Pacific walrus as endangered or threatened.

This 12-month finding is based on our consideration and evaluation of the best scientific and commercial information available. We reviewed the information provided in the petition submitted to the Service by the Center for Biological Diversity, information available in our files, and other available published and unpublished information. Additionally, in response to our Federal Register notice of September 10, 2009, requesting information from the public, as well as our September 10, 2010 press release, and other outreach efforts requesting new information from the public, we received roughly 30,000 submissions, which we have considered in making this finding, including information from the U.S

blished and unpublished information. Additionally, in response to our Federal Register notice of September 10, 2009, requesting information from the public, as well as our September 10, 2010 press release, and other outreach efforts requesting new information from the public, we received roughly 30,000 submissions, which we have considered in making this finding, including information from the U.S. Marine Mammal Commission, the State of Alaska, the Alaska North Slope Borough, the Eskimo Walrus Commission, the Humane Society of the United States, the Center for Biological Diversity, the American Petroleum Institute, and many interested citizens. We also consulted with recognized Pacific walrus experts and Federal, State, and Tribal agencies.

Species Information

Taxonomy and Species Delineation

The walrus ( Odobenus rosmarus ) is the only living representative of the family Odobenidae, a group of marine carnivores that was highly diversified in et al. 1999, p. 60; Harington 2008, p. 27).

Three modern subspecies of walruses are generally recognized (Wozencraft 2005, p. 525; Integrated Taxonomic Information System, 2010, p. 1): The Atlantic walrus ( O. r. rosmarus ), which ranges from the central Canadian Arctic eastward to the Kara Sea (Reeves 1978, pp. 2-20); the Pacific walrus ( O. r. divergens ), which ranges across the Bering and Chukchi Seas (Fay 1982, pp. 7-21); and the Laptev walrus ( O. r. laptevi ), which is represented by a small, geographically isolated population of walruses in the Laptev Sea (Heptner et al. 1976, p. 34; Vishnevskaia and Bychkov 1990, pp. 155-176; Andersen et al. 1998, p. 1323; Wozencraft 2005, p. 595; Jefferson et al. 2008, p. 376). Atlantic and Pacific walruses are genetically and morphologically distinct from each other (Cronin et al. 1994, p. 1035), likely as a result of range fragmentation and differentiation during glacial phases of extensive Arctic sea-ice cover (Harington 2008, p. 27)

34; Vishnevskaia and Bychkov 1990, pp. 155-176; Andersen et al. 1998, p. 1323; Wozencraft 2005, p. 595; Jefferson et al. 2008, p. 376). Atlantic and Pacific walruses are genetically and morphologically distinct from each other (Cronin et al. 1994, p. 1035), likely as a result of range fragmentation and differentiation during glacial phases of extensive Arctic sea-ice cover (Harington 2008, p. 27). Although geographically isolated and ecologically distinct, walruses from the Laptev Sea appear to be more closely related to Pacific walruses (Lindqvist et al. 2009, pp. 119-121).

Pacific walruses are ecologically distinct from other walrus populations, primarily because they undergo significant seasonal migrations between the Bering and the Chukchi Seas and rely principally on broken pack ice habitat to access offshore breeding and feeding areas (Fay 1982, p. 279) ( see Species Distribution, below). In contrast, Atlantic walruses, which are represented by several small discrete groups of animals distributed from the central Canadian Arctic eastward to the Kara Sea, exhibit smaller seasonal movements and feed primarily in coastal areas because the continental shelf is narrow over much of their range. The majority of productive feeding areas used by Atlantic walruses are accessible from the coast, and all age classes and gender groups use terrestrial haulouts during ice-free seasons (Born et al. 2003, p. 356; COSEWIC 2006, p. 15; Laidre et al. 2008, pp. S104, S115).

The Pacific walrus is generally considered a single population, although some heterogeneity has been documented. Jay et al. (2008, p. 938) found some differences in the ratio of trace elements in the teeth of Pacific walruses sampled in winter from two breeding areas (southeast Bering Sea and St. Lawrence Island), suggesting that the sampled animals had a history of feeding in different regions. Scribner et al. (1997, p. 180), however, found no difference in mitochondrial and nuclear DNA among Pacific walruses sampled from different breeding areas

ferences in the ratio of trace elements in the teeth of Pacific walruses sampled in winter from two breeding areas (southeast Bering Sea and St. Lawrence Island), suggesting that the sampled animals had a history of feeding in different regions. Scribner et al. (1997, p. 180), however, found no difference in mitochondrial and nuclear DNA among Pacific walruses sampled from different breeding areas. Pacific walruses are identified and managed in the United States and the Russian Federation (Russia) as a single population (Service 2010, p. 1).

Species Description

Walruses are readily distinguished from other Arctic pinnipeds (aquatic carnivorous mammals with all four limbs modified into flippers, this group includes seals, sea lions, and walruses) by their enlarged upper canine teeth, which form prominent tusks. The family name Odobenidae (tooth walker), is based on observations of walruses using their tusks to pull themselves out of the water. Males, which have relatively larger tusks than females, also tend to have broader skulls (Fay 1982, pp. 104-108). Walrus tusks are used as offensive and defensive weapons (Kastelein 2002, p. 1298). Adult males use their tusks in threat displays and fighting to establish dominance during mating (Fay et al. 1984, p. 93), and animals of both sexes use threat displays to establish and defend positions on land or ice haulouts (Fay 1982, pp. 134-138). Walruses also use their tusks to anchor themselves to ice floes when resting in the water during inclement weather (Fay 1982, pp. 134-138; Kastelein 2002, p. 1298).

The Pacific walrus is the largest pinniped species in the Arctic. At birth, calves are approximately 65 kilograms (kg) (143 pounds (lb)) and 113 centimeters (cm) (44.5 inches (in)) long (Fay 1982, p. 32). After the first 7 years of life, the growth rate of female walruses declines rapidly, and they reach a maximum body size by approximately 10 years of age. Adult females can reach lengths of up to 3 meters (m) (9.8 feet (ft)) and weigh up to 1,100 kg (2,425 lb)

At birth, calves are approximately 65 kilograms (kg) (143 pounds (lb)) and 113 centimeters (cm) (44.5 inches (in)) long (Fay 1982, p. 32). After the first 7 years of life, the growth rate of female walruses declines rapidly, and they reach a maximum body size by approximately 10 years of age. Adult females can reach lengths of up to 3 meters (m) (9.8 feet (ft)) and weigh up to 1,100 kg (2,425 lb). Male walrus tend to grow faster and for a longer period of time than females. They usually do not reach full adult body size until they are 15 to 16 years of age. Adult males can reach lengths of 3.5 m (11.5 ft) and can weigh more than 2,000 kg (4,409 lb) (Fay 1982, p. 33).

Behavior

Walruses are social and gregarious animals. They tend to travel in groups and haul out of the water to rest on ice or land in densely packed groups. On land or ice, in any season, walruses tend to lie in close physical contact with each other. Young animals often lie on top of adults. Group size can range from a few individuals up to several thousand animals (Gilbert 1999, p. 80; Kastelein 2002, p. 1298; Jefferson et al. 2008, p. 378). At any time of the year, when groups are disturbed, stampedes from a haulout can result in injuries and mortalities. Calves and young animals are particularly vulnerable to trampling injuries (Fay 1980, pp. 227-227; Fay and Kelly 1980, p. 226).

The reaction of walruses to disturbance ranges from no reaction to escape into the water, depending on the circumstances (Fay et al. 1984, pp. 13-14). Many factors play into the severity of the response, including the age and sex of the animals, the size and location of the group (on ice, in water, on land), their distance from the disturbance, and the nature and intensity of the disturbance (Fay et al. 1984, pp. 14, 114-119). Females with calves appear to be most sensitive to disturbance, and animals on shore are more sensitive than those on ice (Fay et al. 1984, p. 114)

severity of the response, including the age and sex of the animals, the size and location of the group (on ice, in water, on land), their distance from the disturbance, and the nature and intensity of the disturbance (Fay et al. 1984, pp. 14, 114-119). Females with calves appear to be most sensitive to disturbance, and animals on shore are more sensitive than those on ice (Fay et al. 1984, p. 114). A fright response caused by disturbance can cause stampedes on a haulout, resulting in injuries and mortalities (Fay and Kelly 1980, pp. 241-244).

Mating occurs primarily in January and February in broken pack ice habitat in the Bering Sea. Breeding bulls follow herds of females and compete for access to groups of females hauled out onto sea ice (Fay 1982, pp. 193-194). Males perform visual and acoustical displays in the water to attract females and defend a breeding territory. Subdominant males remain on the periphery of these aggregations and apparently do not display. Intruders into display areas are met with threat displays and physical attacks. Individual females leave the resting herd to join a male in the water where copulation occurs (Fay et al. 1984, pp. 89-99; Sjare and Stirling 1996, p. 900). Gestation lasts 15 to 16 months (Fay 1982, p. 197) and pregnancies are spaced at least 2 years apart (Fay 1982, p. 206). Calving occurs on sea ice, most typically in May, before the northward spring migration (Fay 1982, pp. 199-200). Mothers and newborn calves stay mostly on ice floes during the first few weeks of life (Fay et al. 1984, p. 12).

The social bond between the mother and calf is very strong, and it is unusual for a cow to become separated from her calf (Fay 1982, p. 203). The calf normally remains with its mother for at least 2 years, sometimes longer, if not supplanted by a new calf (Fay 1982, pp. 206-211)

200). Mothers and newborn calves stay mostly on ice floes during the first few weeks of life (Fay et al. 1984, p. 12).

The social bond between the mother and calf is very strong, and it is unusual for a cow to become separated from her calf (Fay 1982, p. 203). The calf normally remains with its mother for at least 2 years, sometimes longer, if not supplanted by a new calf (Fay 1982, pp. 206-211). After separation from their

Species Distribution

Pacific walruses range across the shallow continental shelf waters of the northern Bering Sea and Chukchi Sea, occasionally ranging into the East Siberian Sea and Beaufort Sea (Fay 1982, pp. 7-21; Figure 1 in Garlich-Miller et al. 2011). Waters deeper than 100 m (328 ft) and the extent of the pack ice are factors that limit distribution to the north (Fay 1982, p. 23). Walruses are rarely spotted south of the Alaska Peninsula and Aleutian archipelago; however, migrant animals (mostly males) are occasionally reported in the North Pacific (Service 2010, unpublished data).

Pacific walruses are highly mobile, and their distribution varies markedly in response to seasonal and interannual variations in sea-ice cover. During the January to March breeding season, walruses congregate in the Bering Sea pack ice in areas where open leads (fractures in sea ice caused by wind drift or ocean currents), polynyas (enclosed areas of unfrozen water surrounded by ice) or thin ice allow access to water (Fay 1982, p. 21; Fay et al. 1984, pp. 89-99). The specific location of winter breeding aggregations varies annually depending upon the distribution and extent of ice. Breeding aggregations have been reported southwest of St. Lawrence Island, Alaska; south of Nunivak Island, Alaska; and south of the Chukotka Peninsula in the Gulf of Anadyr, Russia (Fay 1982, p. 21; Mymrin et al. 1990, pp. 105-113; Figure 1 in Garlich-Miller et al. 2011)

9-99). The specific location of winter breeding aggregations varies annually depending upon the distribution and extent of ice. Breeding aggregations have been reported southwest of St. Lawrence Island, Alaska; south of Nunivak Island, Alaska; and south of the Chukotka Peninsula in the Gulf of Anadyr, Russia (Fay 1982, p. 21; Mymrin et al. 1990, pp. 105-113; Figure 1 in Garlich-Miller et al. 2011).

In spring, as the Bering Sea pack ice deteriorates, most of the population migrates northward through the Bering Strait to summer feeding areas over the continental shelf in the Chukchi Sea. However, several thousand animals, primarily adult males, remain in the Bering Sea during the summer months, foraging from coastal haulouts in the Gulf of Anadyr, Russia, and in Bristol Bay, Alaska (Figure 1 in Garlich-Miller et al. 2011).

Summer distributions (both males and females) in the Chukchi Sea vary annually, depending upon the extent of sea ice. When broken sea ice is abundant, walruses are typically found in patchy aggregations over continental shelf waters. Individual groups may range from less than 10 to more than 1,000 animals (Gilbert 1999, pp. 75-84; Ray et al. 2006, p. 405). Summer concentrations have been reported in loose pack ice off the northwestern coast of Alaska, between Icy Cape and Point Barrow, and along the coast of Chukotka, Russia, as far west as Wrangel Island (Fay 1982, pp. 16-17; Gilbert et al. 1992, pp. 1-33; Belikov et al. 1996, pp. 267-269). In years of low ice concentrations in the Chukchi Sea, some animals range east of Point Barrow into the Beaufort Sea; walruses have also been observed in the Eastern Siberian Sea in late summer (Fay 1982, pp. 16-17; Belikov et al. 1996, pp. 267-269). The pack ice of the Chukchi Sea usually reaches its minimum extent in September

17; Gilbert et al. 1992, pp. 1-33; Belikov et al. 1996, pp. 267-269). In years of low ice concentrations in the Chukchi Sea, some animals range east of Point Barrow into the Beaufort Sea; walruses have also been observed in the Eastern Siberian Sea in late summer (Fay 1982, pp. 16-17; Belikov et al. 1996, pp. 267-269). The pack ice of the Chukchi Sea usually reaches its minimum extent in September. In years when the sea ice retreats north beyond the continental shelf, walruses congregate in large numbers (up to several tens of thousands of animals in some locations) at terrestrial haulouts on Wrangel Island and other sites along the northern coast of the Chukotka Peninsula, Russia, and northwestern Alaska (Fay 1982, p. 17; Belikov et al. 1996, pp. 267-269; Kochnev 2004, pp. 284-288; Ovsyanikov et al. 2007, pp. 1-4; Kavry et al. 2008, pp. 248-251).

In late September and October, walruses that summered in the Chukchi Sea typically begin moving south in advance of the developing sea ice. Satellite telemetry data indicate that male walruses that summered at coastal haulouts in the Bering Sea also begin to move northward towards winter breeding areas in November (Jay and Hills 2005, p. 197). The male walruses' northward movement appears to be driven primarily by the presence of females at that time of year (Freitas et al. 2009, pp. 248-260).

Foraging and Prey

Walruses consume mostly benthic (region at the bottom of a body of water) invertebrates and are highly adapted to obtain bivalves (Fay 1982, p. 139; Bowen and Siniff 1999, p. 457; Born et al. 2003, p. 348; Dehn et al. 2007, p. 176; Boveng et al. 2008, pp. 17-19; Sheffield and Grebmeier 2009, pp. 766-767). Fish and other vertebrates have occasionally been found in their stomachs (Fay 1982, p. 153; Sheffield and Grebmeier 2009, p. 767). Walruses root in the bottom sediment with their muzzles and use their whiskers to locate prey items. They use their fore-flippers, nose, and jets of water to extract prey buried up to 32 cm (12.6 in) (Fay 1982, p

pp. 17-19; Sheffield and Grebmeier 2009, pp. 766-767). Fish and other vertebrates have occasionally been found in their stomachs (Fay 1982, p. 153; Sheffield and Grebmeier 2009, p. 767). Walruses root in the bottom sediment with their muzzles and use their whiskers to locate prey items. They use their fore-flippers, nose, and jets of water to extract prey buried up to 32 cm (12.6 in) (Fay 1982, p. 163; Oliver et al. 1983, p. 504; Kastelein 2002, p. 1298; Levermann et al. 2003, p. 8). The foraging behavior of walruses is thought to have a major impact on benthic communities in the Bering and Chukchi Seas (Oliver et al. 1983, pp. 507-509; Klaus et al. 1990, p. 480). Ray et al. (2006, pp. 411-413) estimate that walruses consume approximately 3 million metric tons (3,307 tons) of benthic biomass annually, and that the area affected by walrus foraging is in the order of thousands of square kilometers (sq km) (thousands of square miles (sq mi)) annually. Consequently, walruses play a major role in benthic ecosystem structure and function, which Ray et al. (2006, p. 415) suggested increased nutrient flux and productivity.

The earliest studies of food habits were based on examination of stomachs from walruses killed by hunters. These reports indicated that walruses were primarily feeding on bivalves (clams), and that non-bivalve prey was only incidentally ingested (Fay 1982, p. 145; Sheffield et al. 2001, p. 311). However, these early studies did not take into account the differential rate of digestion of prey items (Sheffield et al. 2001, p. 311). Additional research indicates that stomach contents include over 100 taxa of benthic invertebrates from all major phyla (Fay 1982, p. 145; Sheffield and Grebmeier 2009, p. 764), and while bivalves remain the primary component, walruses are not adapted to a diet solely of clams. Other prey items have similar energetic benefits (Wacasey and Atkinson 1987, pp. 245-247)

Sheffield et al. 2001, p. 311). Additional research indicates that stomach contents include over 100 taxa of benthic invertebrates from all major phyla (Fay 1982, p. 145; Sheffield and Grebmeier 2009, p. 764), and while bivalves remain the primary component, walruses are not adapted to a diet solely of clams. Other prey items have similar energetic benefits (Wacasey and Atkinson 1987, pp. 245-247). Based on analysis of the contents from fresh stomachs of Pacific walruses collected between 1975 and 1985 in the Bering Sea and Chukchi Sea, prey consumption likely reflects benthic invertebrate composition (Sheffield and Grebmeier 2009, pp. 764-768). Of the large number of different types of prey, statistically significant differences between males and females from the Bering Sea were found in the occurrence of only two prey items, and there were no statistically significant differences in results for males and females from the Chukchi Sea (Sheffield and Grebmeier 2009, pp. 765). Although these data are for Pacific walrus stomachs collected 25-35 years ago, we have no reason to believe there has been a change in the general pattern of prey use described here.

Walruses typically swallow invertebrates without shells in their entirety (Fay 1982, p. 165). Walruses remove the soft parts of mollusks from their shells by suction, and discard the shells (Fay 1982, pp. 166-167). Born et al. (2003, p. 348) reported that Atlantic et al. 2006, p. 411), while lactating females may eat two to three times that of nonpregant, nonlactating females (Fay 1982, p.159). Calves up to 1 year of age depend primarily on their mother's milk (Fay 1982, p. 138) and are gradually weaned in their second year (Fisher and Stewart 1997, pp. 1165-1175).

Although walruses are capable of diving to depths of more than 250 m (820 ft) (Born et al. 2005, p. 30), they usually forage in waters of 80 m (262 ft) or less (Fay and Burns 1988, p. 239; Born et al. 2003, p. 348; Kovacs and Lydersen 2008, p

r of age depend primarily on their mother's milk (Fay 1982, p. 138) and are gradually weaned in their second year (Fisher and Stewart 1997, pp. 1165-1175).

Although walruses are capable of diving to depths of more than 250 m (820 ft) (Born et al. 2005, p. 30), they usually forage in waters of 80 m (262 ft) or less (Fay and Burns 1988, p. 239; Born et al. 2003, p. 348; Kovacs and Lydersen 2008, p. 138), presumably because of higher productivity of their benthic foods in shallow waters (Fay and Burns 1988, pp. 239-240; Carey 1991, p. 869; Jay et al. 2001, p. 621; Grebmeier et al. 2006b, pp. 334-346; Grebmeier et al. 2006a, p. 1461). Walruses make foraging trips from land or ice haulouts that range from a few hours up to several days and up to 100 kilometers (km) (60 miles (mi)) (Jay et al. 2001, p. 626; Born et al. 2003, p. 349; Ray et al. 2006, p. 406; Udevitz et al. 2009, p. 1122). Walruses tend to make shorter and more frequent foraging trips when sea ice is used as a foraging platform compared to terrestrial haulouts (Udevitz et al. 2009, p. 1122). Satellite telemetry data for walruses in the Bering Sea in April of 2004, 2005, and 2006 showed they spent an average of 46 hours in the water between resting bouts on ice, which averaged 9 hours (Udevitz et al. 2009, p. 1122). Because females and young travel with the retreating pack ice in the spring and summer, they are passively transported northward over feeding grounds across the continental shelves of the Bering and Chukchi Seas. Male walruses appear to have greater endurance than females, with foraging excursions from land haulouts that can last up to 142 hours (about 6 days) (Jay et al. 2001, p. 630).

Sea-Ice Habitats

The Pacific walrus is an ice-dependent species that relies on sea ice for many aspects of its life history. Unlike other pinnipeds, walruses are not adapted for a pelagic existence and must haul out on ice or land regularly. Floating pack ice serves as a substrate for resting between feeding bouts (Ray et al. 2006, p

st up to 142 hours (about 6 days) (Jay et al. 2001, p. 630).

Sea-Ice Habitats

The Pacific walrus is an ice-dependent species that relies on sea ice for many aspects of its life history. Unlike other pinnipeds, walruses are not adapted for a pelagic existence and must haul out on ice or land regularly. Floating pack ice serves as a substrate for resting between feeding bouts (Ray et al. 2006, p. 404), breeding behavior (Fay et al. 1984, pp. 89-99), giving birth (Fay 1982, p. 199), and nursing and care of young (Kelly 2001, pp. 43-55). Sea ice provides access to offshore feeding areas over the continental shelf of the Bering and Chukchi Seas, passive transportation to new feeding areas (Richard 1990, p. 21; Ray et al. 2006, pp. 403-419), and isolation from terrestrial predators (Richard 1990, p. 23; Kochnev 2004, p. 286; Ovsyanikov et al. 2007, pp. 1-4). Sea ice provides an extensive substrate upon which the risk of predation and hunting is greatly reduced (Kelly 2001, pp. 43-55; Fay 1982, p. 26).

Sea ice in the Northern Hemisphere is comprised of first-year sea ice that formed in the most recent autumn-winter period, and multi-year ice that has survived at least one summer melt season. Sea-ice habitats for walruses include openings or leads that provide access to the water and to food resources. Walruses generally do not use multi-year ice or highly compacted first-year ice in which there is an absence of persistent leads or polynyas (Richard 1990, p. 21). Expansive areas of heavy ice cover are thought to play a restrictive role in walrus distributions across the Arctic and serve as a barrier to the mixing of populations (Fay 1982, p. 23; Dyke et al. 1999, pp. 161-163; Harington 2008, p. 35). Walruses generally do not occur farther south than the maximum extent of the winter pack ice, possibly due to their reliance on sea ice for breeding and rearing young (Fay et al. 1984, pp. 89-99) and isolation from terrestrial predators (Kochnev 2004, p. 286; Ovsyanikov et al. 2007, pp

s a barrier to the mixing of populations (Fay 1982, p. 23; Dyke et al. 1999, pp. 161-163; Harington 2008, p. 35). Walruses generally do not occur farther south than the maximum extent of the winter pack ice, possibly due to their reliance on sea ice for breeding and rearing young (Fay et al. 1984, pp. 89-99) and isolation from terrestrial predators (Kochnev 2004, p. 286; Ovsyanikov et al. 2007, pp. 1-4), or because of the higher densities of benthic invertebrates in northern waters (Grebmeier et al. 2006a, pp. 1461-1463).

Walruses generally occupy first-year ice that is greater than 20 cm (7.9 in) thick and are not found in areas of extensive, unbroken ice (Fay 1982, pp. 21, 26; Richard 1990, p. 23). Thus, in winter they concentrate in areas of broken pack ice associated with divergent ice flow or along the margins of persistent polynyas (Burns et al. 1981, pp. 781-797; Fay et al. 1984, pp. 89-99; Richard 1990, p. 23) in areas with abundant food resources (Ray et al. 2006, p. 406). Females with young generally spend the summer months in pack ice habitats of the Chukchi Sea, where they feed intensively between bouts of resting and suckling their young. Some authors have suggested that the size and topography of individual ice floes are important features in the selection of ice haulouts, noting that some animals have been observed returning to the same ice floe between feeding bouts (Ray et al. 2006, p. 406). However, it has also been noted that walruses can and will exploit a fairly broad range of ice types and ice concentrations in order to stay in preferred foraging or breeding areas (Freitas et al. 2009, p. 247; Jay et al. 2010a, p. 300). Walruses tend to make shorter foraging excursions when they are using sea ice rather than land haulouts (Udevitz et al. 2009, p. 1122), presumably because it is more energetically efficient for them to haulout on ice near productive feeding areas than forage from shore. Fay (1982, p

ns in order to stay in preferred foraging or breeding areas (Freitas et al. 2009, p. 247; Jay et al. 2010a, p. 300). Walruses tend to make shorter foraging excursions when they are using sea ice rather than land haulouts (Udevitz et al. 2009, p. 1122), presumably because it is more energetically efficient for them to haulout on ice near productive feeding areas than forage from shore. Fay (1982, p. 25) notes that several authors reported that when walruses had the choice of ice or land for a resting place, ice was always selected.

Terrestrial Habitats (Coastal Haulouts)

When suitable sea ice is not available, walruses haul out on land to rest. A wide variety of substrates, ranging from sand to boulders, are used. Isolated islands, points, spits, and headlands are occupied most frequently. The primary consideration for a terrestrial haulout site appears to be isolation from disturbances and predators, although social factors, learned behavior, protection from strong winds and surf, and proximity to food resources also likely influence the choice of terrestrial haulout sites (Richard 1990, p. 23). Walruses tend to use established haulout sites repeatedly and exhibit some degree of fidelity to these sites (Jay and Hills 2005, pp. 192-202), although the use of coastal haulouts appears to fluctuate over time, possibly due to localized prey depletion (Garlich-Miller and Jay 2000, pp. 58-65). Human disturbance is also thought to influence the choice of haulout sites; many historic haulouts in the Bering Sea were abandoned in the early 1900s when the Pacific walrus population was subjected to high levels of exploitation (Fay 1982, p. 26; Fay et al. 1984, p. 231).

Adult male walruses use land-based haulouts more than females or young, and consequently, have a greater geographical distribution through the ice-free season

o influence the choice of haulout sites; many historic haulouts in the Bering Sea were abandoned in the early 1900s when the Pacific walrus population was subjected to high levels of exploitation (Fay 1982, p. 26; Fay et al. 1984, p. 231).

Adult male walruses use land-based haulouts more than females or young, and consequently, have a greater geographical distribution through the ice-free season. Many adult males remain in the Bering Sea throughout the ice-free season, making foraging trips from coastal haulouts in Bristol Bay, Alaska, and the Gulf of Anadyr, Russia (Figure 1 in Garlich-Miller et al. 2011), while females and juvenile animals generally stay with the drifting ice pack throughout the year (Fay 1982, pp. 8-19). Females with dependent young may prefer sea-ice habitats because coastal haulouts pose greater risk from trampling injuries and predation (Fay and Kelly 1980, pp. 226-245; Ovsyanikov et al. 1994, p. 80; Kochnev et al. 2007, pp. 1-4; Kavry et al. 2008, pp. 248-251; Mulcahy et al. 2009, p. 3). Females may also prefer sea-ice habitats because they may have difficulty nourishing themselves while caring for a young calf that has limited swimming range (Cooper et al. 2006, p. 101; Jay and Fischbach 2008, p. 1).

The numbers of male walruses using coastal haulouts in the Bering Sea during the summer months, and the relative uses of different coastal haulout sites in the Bering Sea have varied over the past century. Harvest records indicate that walrus herds were once common at coastal haulouts along the Alaska Peninsula and the islands of northern Bristol Bay (Fay et al. 1984, pp. 231-376). By the early 1950s, most of the traditional haulout areas in the Southern Bering Sea had been abandoned, presumably due to hunting pressure. During the 1950s and 1960s, Round Island was the only regularly used haulout in Bristol Bay, Alaska. In 1960, the State of Alaska established the Walrus Islands State Game Sanctuary, which closed Round Island to hunting

tol Bay (Fay et al. 1984, pp. 231-376). By the early 1950s, most of the traditional haulout areas in the Southern Bering Sea had been abandoned, presumably due to hunting pressure. During the 1950s and 1960s, Round Island was the only regularly used haulout in Bristol Bay, Alaska. In 1960, the State of Alaska established the Walrus Islands State Game Sanctuary, which closed Round Island to hunting. Peak counts of walruses at Round Island increased from 1,000-2,000 animals in the late 1950s (Frost et al. 1983, pp. 379) to more than 10,000 animals in the early 1980s (Sell and Weiss, p. 12), but subsequently declined to 2,000-5,000 over the past decade (Sell and Weiss 2010, p. 12). General observations indicate that declining walrus counts at Round Island may, in part, reflect a redistribution of animals to other coastal sites in the Bristol Bay region. For example, walruses have been observed increasingly regularly at the Cape Seniavin haulout on the Alaska Peninsula since the 1970s, and at Cape Peirce and Cape Newenham in northwest Bristol Bay since the early 1980s (Jay and Hills 2005, p. 193; Figure 1 in Garlich-Miller et al. 2011).

Traditional male summer haulouts along the Bering Sea coast of Russia include sites along the Kamchatka Peninsula, the Gulf of Anadyr (most notably Rudder and Meechkin spits), and Arakamchechen Island (Garlich-Miller and Jay 2000, pp. 58-65; Figure 1 in Garlich-Miller et al. 2011). Several of the southernmost haulouts along the coast of Kamchatka have not been occupied in recent years, and the number of animals in the Gulf of Anadyr has also declined in recent years (Kochnev 2005, p. 4). Factors influencing abundance at Bering Sea haulouts are poorly understood, but may include changes in prey densities near the haulouts, changes in population size, disturbance levels, and changing seasonal distributions (Jay and Hills 2005, p. 198) (presumably mediated by sea-ice coverage or temperature)

mber of animals in the Gulf of Anadyr has also declined in recent years (Kochnev 2005, p. 4). Factors influencing abundance at Bering Sea haulouts are poorly understood, but may include changes in prey densities near the haulouts, changes in population size, disturbance levels, and changing seasonal distributions (Jay and Hills 2005, p. 198) (presumably mediated by sea-ice coverage or temperature).

Historically, coastal haulouts along the Arctic (Chukchi Sea) coast have been used less consistently during the summer months than those in the Bering Sea because of the presence of pack ice (a preferred substrate) for much of the year in the Chukchi Sea. Since the mid-1990s, reductions of summer sea ice coincided with a marked increase in the use of coastal haulouts along the Chukchi sea coast of Russia during the summer months (Kochnev 2004, pp. 284-288; Kavry et al. 2008, pp. 248-251). Large, mixed (composed of various age and sex groups) herds of walruses, up to several tens of thousands of animals, began to use coastal haulouts on Wrangel Island, Russia in the early 1990s, and several coastal haulouts along the northern Chukotka coastline of Russia have emerged in recent years, likely as a result of reductions in summer sea ice in the Chukchi Sea (Kochnev 2004, pp. 284-288; Ovsyanikov et al. 2007, pp. 1-4; Kavry et al. 2008, p. 248-251; Figure 1 in Garlich-Miller et al. 2011).

In 2007, 2009, and 2010, walruses were also observed hauling out in large numbers with mixed sex and age groups along the Chukchi Sea coast of Alaska in late August, September, and October (Thomas et al. 2009, p. 1; Service 2010, unpublished data). Monitoring studies conducted in association with oil and gas exploration suggest that the use of coastal haulouts along the Arctic coast of Alaska during the summer months is dependent upon the availability of sea ice

n large numbers with mixed sex and age groups along the Chukchi Sea coast of Alaska in late August, September, and October (Thomas et al. 2009, p. 1; Service 2010, unpublished data). Monitoring studies conducted in association with oil and gas exploration suggest that the use of coastal haulouts along the Arctic coast of Alaska during the summer months is dependent upon the availability of sea ice. For example, in 2006 and 2008, walruses foraging off the Chukchi Sea coast of Alaska remained with the ice pack over the continental shelf during the months of August, September, and October. However in 2007, 2009, and 2010, the pack ice retreated beyond the continental shelf and large numbers of walruses hauled out on land at several locations between Point Barrow and Cape Lisburne, Alaska (Ireland et al. 2009, p. xvi; Thomas et al. 2009, p. 1; Service 2010, unpublished data; Figure 1 in Garlich-Miller et al. 2011).

Transitory coastal haulouts have also been reported in late fall (October-November) along the southern Chukchi Sea coast, coinciding with the southern migration. Mixed herds of walruses frequently come to shore to rest for a few days to weeks along the coast before continuing on their migration to the Bering Sea. Cape Lisburne, Alaska, and Capes Serdtse-Kamen' and Dezhnev, Russia, are the most consistently used haulouts in the Chukchi Sea at this time of year (Garlich-Miller and Jay 2000, pp. 58-67). Large mixed herds of walruses have also been reported in late fall and early winter at coastal haulouts in the northern Bering Sea at the Punuk Islands and Saint Lawrence Island, Alaska; Big Diomede Island, Russia; and King Island, Alaska, prior to the formation of sea ice in offshore breeding and feeding areas (Fay and Kelly 1980, p. 226; Garlich-Miller and Jay 2000, pp. 58-67; Figure 1 in Garlich-Miller et al. 2011).

Vital Rates

Walruses have the lowest rate of reproduction of any pinniped species (Fay 1982, pp. 172-209)

t the Punuk Islands and Saint Lawrence Island, Alaska; Big Diomede Island, Russia; and King Island, Alaska, prior to the formation of sea ice in offshore breeding and feeding areas (Fay and Kelly 1980, p. 226; Garlich-Miller and Jay 2000, pp. 58-67; Figure 1 in Garlich-Miller et al. 2011).

Vital Rates

Walruses have the lowest rate of reproduction of any pinniped species (Fay 1982, pp. 172-209). Although male walruses reach puberty at 6-7 years of age, they are unlikely to successfully compete for access to females until they reach full body size at 15 years of age or older (Fay 1982, p. 33; Fay et al. 1984, p. 96). Female walruses attain sexual maturity at 4-7 years of age (Fay 1982, pp. 172-209), and the median age of first birth ranges from approximately 8 to 10 years of age (Garlich-Miller et al. 2006, pp. 887-893). Because gestation lasts 15-16 months, it extends through the following breeding season and thus, the minimum interval between successful births is 2 years. Ovulation may also be suppressed until the calf is weaned, raising the birth interval to 3 years or more (Garlich-Miller and Stewart 1999, p. 188). The age of sexual maturity and birth rates may be density-dependent (Fay et al. 1989, pp. 1-16; Fay et al. 1997, pp. 537-565; Garlich-Miller et al. 2006, pp. 892-893).

The low birth rate of walruses is offset in part by considerable maternal investment in offspring (Fay et al. 1997, p. 550). Assumed survival rates through the first year of life range from 0.5 to 0.9 (Fay et al. 1997, p. 550). Survival rates for juveniles through adults ( i.e., 4-20 years old) have been assumed to be as high as 0.96 to 0.99 per cent (DeMaster 1984, p. 78; Fay et al. 1997, p. 544), declining to zero by 40 to 45 years (Chivers 1999, p. 240). Using published estimates of survival and reproduction, Chivers (1999, pp

rvival rates through the first year of life range from 0.5 to 0.9 (Fay et al. 1997, p. 550). Survival rates for juveniles through adults ( i.e., 4-20 years old) have been assumed to be as high as 0.96 to 0.99 per cent (DeMaster 1984, p. 78; Fay et al. 1997, p. 544), declining to zero by 40 to 45 years (Chivers 1999, p. 240). Using published estimates of survival and reproduction, Chivers (1999, pp. 239-247) developed an individual age-based model of the Pacific walrus population, which yielded a maximum population growth rate of 8 percent, but cautioned this should not be considered to be an estimate of the maximum growth rate (Chivers 1999, p. 239). Thus, the 8 percent figure remains theoretical because age-specific survival rates for free-ranging walruses are poorly known.

Abundance

Based on large sustained harvests in the 18th and 19th centuries, Fay (1982, p. 241) speculated that the pre- et al. 1997, p. 539). The population apparently increased rapidly in size during the 1960s and 1970s in response to harvest regulations that limited the take of females (Fay et al. 1989, p. 4). Between 1975 and 1990, visual aerial surveys jointly conducted by the United States and Russia at 5-year intervals produced population estimates ranging from 201,039 to 290,000. Efforts to survey the Pacific walrus population were suspended by both countries after 1990, due to unresolved problems with survey methods that produced population estimates with unknown bias and unknown—but presumably large—variances that severely limited their utility (Speckman et al. 2010, p. 3).

In 2006, a joint U.S.-Russian survey was conducted in the pack ice of the Bering Sea, using thermal imaging systems to detect walruses hauled out on sea ice and satellite transmitters to account for walruses in the water (Speckman et al. 2010, p. 4). The number of walruses within the surveyed area was estimated at 129,000, with 95-percent confidence intervals of 55,000 to 507,000 individuals

3).

In 2006, a joint U.S.-Russian survey was conducted in the pack ice of the Bering Sea, using thermal imaging systems to detect walruses hauled out on sea ice and satellite transmitters to account for walruses in the water (Speckman et al. 2010, p. 4). The number of walruses within the surveyed area was estimated at 129,000, with 95-percent confidence intervals of 55,000 to 507,000 individuals. This is a minimum estimate, as weather conditions forced termination of the survey before much of the southwest Bering Sea was surveyed; animals were observed in that region as the surveyors returned to Anchorage, Alaska. Table 1 provides a summary of survey results.

Table 1—Estimates of Pacific Walrus Population Size, 1975-2006. Year Population size (with range or confidence interval) a Reference 1975 214,687 (Udevitz et al. 2001, p. 614). 1980 250,000-290,000 (Johnson et al. 1982, p. 3; Fedoseev 1984, p. 58). 1985 242,366 (Udevitz et al. 2001, p. 614). 1990 201,039 (Gilbert et al. 1992, p. 28). 2006 129,000 (50,000-500,000) (Speckman et al. 2010). a Due to differences in methods, comparisons of estimates across years (population trends) are not possible. Most estimates did not provide a range or confidence interval. We acknowledge that these survey results suggest to some that the walrus population may be declining; however, we do not believe the survey methodologies support such a definitive conclusion. Resource managers in Russia have concluded that the population has declined, and accordingly, have reduced harvest quotas in recent years (Kochnev 2004, p. 284; Kochnev 2005, p. 4; Kochnev, 2010, pers. comm.), based in part on the lower abundance estimate generated from the 2006 survey results. However, past survey results are not directly comparable among years due to differences in survey methods, timing of surveys, segments of the population surveyed, and incomplete coverage of areas where walruses may have been present (Fay et al. 1997, p

004, p. 284; Kochnev 2005, p. 4; Kochnev, 2010, pers. comm.), based in part on the lower abundance estimate generated from the 2006 survey results. However, past survey results are not directly comparable among years due to differences in survey methods, timing of surveys, segments of the population surveyed, and incomplete coverage of areas where walruses may have been present (Fay et al. 1997, p. 537); thus, these results do not provide a basis for determining trends in population size (Hills and Gilbert 1994, p. 203; Gilbert 1999, pp. 75-84). Whether prior estimates are biased low or high is unknown, because of problems with detecting individual animals on ice or land, and in open water, and difficulties counting animals in large, dense groups (Speckman et al. 2010, p. 33). In addition, no survey has ever been completed within a timeframe that could account for the redistribution of individuals (leading to double counting or undercounting), or before weather conditions either delayed the effort or completely terminated the survey before the entire area of potentially occupied habitat had been covered (Speckman et al. 2010). Due to these general problems, as well as seasonal differences among surveys (fall or spring) and technological advancements that correct for some problems, we do not believe the survey results provide a reliable basis for estimating a population trend.

Changes in the walrus population have also been investigated by examining changes in biological parameters over time. Based on evidence of changes in abundance, distributions, condition indices, and life-history parameters, Fay et al. (1989, pp.1-16) and Fay et al. (1997, pp. 537-565) concluded that the Pacific walrus population increased greatly in size during the 1960s and 1970s, and postulated that the population was approaching, or had exceeded, the carrying capacity of its environment by the early 1980s

ime. Based on evidence of changes in abundance, distributions, condition indices, and life-history parameters, Fay et al. (1989, pp.1-16) and Fay et al. (1997, pp. 537-565) concluded that the Pacific walrus population increased greatly in size during the 1960s and 1970s, and postulated that the population was approaching, or had exceeded, the carrying capacity of its environment by the early 1980s. Harvest increased in the 1980s: changes in the size, composition, and productivity of the sampled walrus harvest in the Bering Strait Region of Alaska over this time frame are consistent with this hypothesis (Garlich-Miller et al. 2006, p. 892). Harvest levels declined sharply in the early 1990s, and increased reproductive rates and earlier maturation in females occurred, suggesting that density-dependent regulatory mechanisms had been relaxed and the population was likely below carrying capacity (Garlich-Miller et al. 2006, p. 893). However, Garlich-Miller et al. (2006, pp. 892-893) also noted that there are no data concerning the trend in abundance of the walrus population or the status of its prey to verify this hypothesis, and that whether density-dependent changes in life-history parameters might have been mediated by changes in population abundance or changes in the carrying capacity of the environment is unknown.

Summary of Information Pertaining to the Five Factors

Section 4 of the Act (16 U.S.C. 1533) and implementing regulations (50 CFR part 424) set forth the procedures for adding species to, removing species from, or reclassifying species on the Federal Lists of Endangered and Threatened Wildlife and Plants

in population abundance or changes in the carrying capacity of the environment is unknown.

Summary of Information Pertaining to the Five Factors

Section 4 of the Act (16 U.S.C. 1533) and implementing regulations (50 CFR part 424) set forth the procedures for adding species to, removing species from, or reclassifying species on the Federal Lists of Endangered and Threatened Wildlife and Plants. Under section 4(a)(1) of the Act, a species may be determined to be endangered or threatened based on any of the following five factors:

(A) The present or threatened destruction, modification, or curtailment of its habitat or range;

(B) Overutilization for commercial, recreational, scientific, or educational purposes;

(C) Disease or predation;

(D) The inadequacy of existing regulatory mechanisms; or

(E) Other natural or manmade factors affecting its continued existence.

In making this 12-month finding, we considered and evaluated the best available scientific and commercial information. Information pertaining to the Pacific walrus in relation to the five

In considering what factors might constitute threats to a species, we must look beyond the exposure of the species to a particular stressor to evaluate whether the species may respond to that stressor in a way that causes actual impacts to the species. If there is exposure to a stressor and the species responds negatively, the stressor may be a threat and we attempt to determine how significant a threat it is. The threat is significant if it drives, or contributes to, the risk of extinction of the species such that the species warrants listing as endangered or threatened as those terms are defined in the Act. However, the identification of stressors that could impact a species negatively may not be sufficient to compel a finding that the species warrants listing

determine how significant a threat it is. The threat is significant if it drives, or contributes to, the risk of extinction of the species such that the species warrants listing as endangered or threatened as those terms are defined in the Act. However, the identification of stressors that could impact a species negatively may not be sufficient to compel a finding that the species warrants listing. The information must include evidence sufficient to suggest that these stressors are operative threats that act on the species to the point that the species meets the definition of endangered or threatened under the Act. Also, because an individual stressor may not be a threat by itself, but could be in conjunction with one or more other stressors, our process includes considering the combined effects of stressors.

To inform our analysis of threats to the Pacific walrus, we also took into consideration the results of two Bayesian network modeling efforts; one conducted by the Service (Garlich-Miller et al. 2011), and the other conducted by the U.S. Geological Survey (USGS) (Jay et al. 2010b). Although quantitative, empirical data can be used in Bayesian networks, when primarily qualitative data are available, such as for the Pacific walrus, the models are well suited to formalizing and quantifying the opinions of experts (Marcot et al. 2006, p. 3063). Bayesian network models (also known as Bayesian belief networks, reflecting the importance of expert opinion) graphically display the relevant stressors, the interactions among stressors, and the cumulative impact of those stressors as they are integrated through the network. In general terms, the network is composed of input variables that represent key environmental correlates ( e.g., sea-ice loss, harvest, shipping) and response variables, ( e.g., population status). Although we did not rely on the results of the Bayesian models as the sole basis for our conclusions in this finding, the models corroborated the results of our threats analysis

ted through the network. In general terms, the network is composed of input variables that represent key environmental correlates ( e.g., sea-ice loss, harvest, shipping) and response variables, ( e.g., population status). Although we did not rely on the results of the Bayesian models as the sole basis for our conclusions in this finding, the models corroborated the results of our threats analysis. Results of the models are presented in the five-factor analysis below, where pertinent.

Factor A. The Present or Threatened Destruction, Modification, or Curtailment of Its Habitat or Range

The following potential stressors that may affect the habitat or range of the Pacific walrus are discussed in this section: (1) Loss of sea ice due to climate change; and (2) effects on prey species due to ocean warming and ocean acidification.

Effects of Global Climate Change on Sea-Ice Habitats

The Pacific walrus depends on sea ice for several aspects of its life history. This section describes recent observations and future projections of sea-ice conditions in the Bering and Chukchi Seas through the end of the 21st century. Following this presentation on the changing ice dynamics, we examine how these changing ice conditions may affect the Pacific walrus population.

The Arctic Ocean is covered primarily by a mix of multi-year sea ice, whereas more southerly regions, such as the Bering Sea, are seasonal ice zones where first-year ice is renewed every winter. The observed and projected effects of global warming vary in different parts of the world, and the Arctic and Antarctic regions are increasingly recognized as being extremely vulnerable to current and projected effects. For several decades, the surface air temperatures in the Arctic have warmed at approximately twice the global rate (Christensen et al. 2007, p. 904)

ar ice is renewed every winter. The observed and projected effects of global warming vary in different parts of the world, and the Arctic and Antarctic regions are increasingly recognized as being extremely vulnerable to current and projected effects. For several decades, the surface air temperatures in the Arctic have warmed at approximately twice the global rate (Christensen et al. 2007, p. 904). The observed and projected effects of climate change are most extreme during summer in northern high-latitude regions, in large part due to the ice-albedo (reflective property) feedback mechanism, in which melting of snow and sea ice lowers surface reflectivity, thereby further increasing surface warming from absorption of solar radiation.

Since 1979 (the beginning of the satellite record of sea-ice conditions), there has been an overall reduction in the extent of Arctic sea ice (Parkinson et al. 1999, p. 20837; Comiso 2002, p. 1956; Stroeve et al. 2005, pp. 1-4; Comiso 2006, pp. 1-3; Meier et al. 2007, p. 428; Stroeve et al. 2007, p. 1; Comiso et al. 2008, p. 1; Stroeve et al. 2008, p. 13). Although the decline is a year-round trend, far greater reductions have been noted in summer sea ice than in winter sea ice. For example, from 1979 to 2009, the extent of September sea ice seen Arctic wide has declined 11 percent per decade (Polyak et al. 2010, p. 1797). In recent years, the trend in Arctic sea-ice loss has accelerated (Comiso et al. 2008, p. 1). In September 2007, the extent of Arctic Ocean sea ice reached a record low, approximately 50 percent lower than conditions in the 1950s through the 1970s, and 23 percent below the previous record set in 2005 (Stroeve et al. 2008, p. 13). Minimum sea-ice extent in 2010 was the third lowest in the satellite record, behind 2007 and 2008 (second lowest), and most of this loss occurred on the Pacific side of the Arctic Ocean.

Of long-term significance is the loss of over 40 percent of Arctic multi-year sea ice over the last 5 years (Kwok et al. 2009, p. 1)

23 percent below the previous record set in 2005 (Stroeve et al. 2008, p. 13). Minimum sea-ice extent in 2010 was the third lowest in the satellite record, behind 2007 and 2008 (second lowest), and most of this loss occurred on the Pacific side of the Arctic Ocean.

Of long-term significance is the loss of over 40 percent of Arctic multi-year sea ice over the last 5 years (Kwok et al. 2009, p. 1). Since 2004, there has been a reversal in the volumetric and areal contributions between first-year ice and multi-year ice in regards to the total volume and area of the Arctic Ocean that they cover, with first-year ice now predominating (Kwok et al. 2009, p. 16). Export of ice through Fram Strait, together with the decline in multi-year ice coverage, suggests that recently there has been near-zero replenishment of multi-year ice (Kwok et al. 2009, p. 16). The area of the Arctic Ocean covered by ice predominantly older than 5 years decreased by 56 percent between 1982 and 2007 (Polyak et al. 2010, p. 1759). Within the central Arctic Ocean, old ice has declined by 88 percent, and ice that is at least 9 years old has essentially disappeared (Markus et al. 2009, p. 13: Polyak et al. 2010, p. 1759). In addition, from 2005 to 2008 there was a thinning of 0.6 m (1.9 ft) in multi-year ice thickness. It is likely that the rapid decline of sea ice in 2007 was in part the result of thinner and lower coverage, of the multi-year ice (Comiso et al. 2008, p. 6). It would take many years to restore the ice thickness through annual growth, and the loss of multi-year ice makes it unlikely that the age and thickness composition of the ice pack will return to previous climatological conditions with continued global warming. Further loss of sea ice will be a major driver of changes across the Arctic over the next decades, especially in late summer and autumn (NOAA 2010, p. 77503).

Due to asymmetric geography of the Arctic and the scale of weather patterns, there is considerable regional variability in sea-ice cover (Meier et al

ice pack will return to previous climatological conditions with continued global warming. Further loss of sea ice will be a major driver of changes across the Arctic over the next decades, especially in late summer and autumn (NOAA 2010, p. 77503).

Due to asymmetric geography of the Arctic and the scale of weather patterns, there is considerable regional variability in sea-ice cover (Meier et al. 2007, p. 430), and although the early loss of summer sea ice and volumetric ice loss in the Arctic applies directly to the Chukchi Sea, it cannot be directly extrapolated to the seasonal ice zone of the Bering Sea (NOAA 2010, p. 77503). The contrasts between the two are dramatic: The Bering Sea is one of the most stable in terms of sea ice, especially in the winter, and the Chukchi Sea has had some of the most dramatic losses of summer sea ice et al., p. 431). Below, we describe the sea-ice conditions in the Bering and Chukchi Seas as they occur presently, as well as recent trends and projections for the future.

In March and April, at maximal sea-ice extent, the Chukchi Sea is typically completely frozen, and ice cover in the Bering Sea extends southward to a latitude of approximately 58-60 degrees north (Boveng et al. 2008, pp. 33-52). The Bering Sea spans the marginal sea-ice zone, where ice gives way to water at the southern edge, and around the peripheries of persistent polynyas. Sea ice in the Bering Sea is highly dynamic and largely a wind-driven system (Sasaki and Minobe 2005, pp. 1-2). Ice cover is comprised of a variety of first-year ice thicknesses, from young, very thin ice to first-year floes that may be upwards of 1.0-m (3.3-ft) thick (Burns et al. 1980, p. 100; Zhang et al. 2010, p. 1729). Depending on wind patterns, a variable (but relatively minor) fraction of ice that drifts south through the Bering Strait could be comprised of some thicker ice floes that originated in the Chukchi and Beaufort Seas (Kozo et al. 1987, pp. 193-195)

sses, from young, very thin ice to first-year floes that may be upwards of 1.0-m (3.3-ft) thick (Burns et al. 1980, p. 100; Zhang et al. 2010, p. 1729). Depending on wind patterns, a variable (but relatively minor) fraction of ice that drifts south through the Bering Strait could be comprised of some thicker ice floes that originated in the Chukchi and Beaufort Seas (Kozo et al. 1987, pp. 193-195).

Ice melt in the Bering Sea usually begins in late April and accelerates in May, with the edge of the ice moving northward until it passes through the Bering Strait, typically in June. The Bering Sea remains ice free for the duration of the summer. Ice continues to retreat northward through the Chukchi Sea until September, when minimal sea-ice extent is reached.

Freeze-up begins in October, with the ice edge progressing southward across the Chukchi Sea. The ice edge usually reaches the Bering Strait in November and advances through the Strait in December. The ice edge continues to move southward across the Bering Sea until its maximal extent is reached in March. There is considerable year-to-year variation in the timing and extent of ice retreat and formation (Boveng et al. 2008, p. 37; Douglas 2010, p. 19).

Within various regions of the Arctic, there is substantial variation in the monthly trends of sea ice (Meier et al. 2007, p. 431). In the Bering Sea, statistically significant monthly reductions in the extent of sea ice over the period 1979-2005 were documented for March (−4.8 percent), October (−42.9 percent), and November (−20.3 percent), although the overall annual decline (−1.9 percent) is not statistically significant (Meier et al. 2007, p. 431). The Bering Sea declines were greatest in October and November, the period of early freeze-up

, statistically significant monthly reductions in the extent of sea ice over the period 1979-2005 were documented for March (−4.8 percent), October (−42.9 percent), and November (−20.3 percent), although the overall annual decline (−1.9 percent) is not statistically significant (Meier et al. 2007, p. 431). The Bering Sea declines were greatest in October and November, the period of early freeze-up. In the Chukchi Sea, statistically significant monthly reductions were also documented for 1979 to 2005 for May (−0.19 percent), June (−4.3 percent), July (−6.7 percent), August (−15.4 percent), September (−26.3 percent), October (−18.6 percent), and November (−8.0 percent): The overall annual reduction (−4.9 percent) is statistically significant (Meier et al. 2007, p. 431). In essence, the Chukchi Sea has shown declines in all months when it is not completely ice-covered, with greatest declines in months of maximal melt and early freeze-up (August, September, and October).

During the period 1979-2006, the September sea-ice extent in the Chukchi Sea decreased by 26 percent per decade (Douglas 2010, p. 2). In recent years, sea ice typically has retreated from continental shelf regions of the Chukchi Sea in August or September, with open water conditions persisting over much of the continental shelf through late October. In contrast, during the preceding 20 years (1979-1998), broken sea-ice habitat persisted over continental shelf areas of the Chukchi Sea through the entire summer (Jay and Fischbach 2008, p. 1).

From 1979 to 2007, there was a general trend toward earlier onset of ice melt and later onset of freeze-up in 9 of 10 Arctic regions analyzed by Markus et al. (2009, pp. 1-14), the exception being the Sea of Okhotsk. For the entire Arctic, the melt season length has increased by about 20 days over the last 30 years, due to the combined earlier melt and later freeze-up. The largest increases, of over 10 days per decade, have been seen for Hudson Bay, the East Greenland Sea, and the Laptev/East Siberian Seas

of 10 Arctic regions analyzed by Markus et al. (2009, pp. 1-14), the exception being the Sea of Okhotsk. For the entire Arctic, the melt season length has increased by about 20 days over the last 30 years, due to the combined earlier melt and later freeze-up. The largest increases, of over 10 days per decade, have been seen for Hudson Bay, the East Greenland Sea, and the Laptev/East Siberian Seas. From 1979 to 2007, there was a general trend toward earlier onset of ice melt and later onset of freeze-up in both the Bering and Chukchi Seas: For the Bering Sea, the onset of ice melt occurred 1.0 day earlier per decade, while in the Chukchi/Beaufort Seas ice melt occurred 3.5 days earlier per decade. The onset of freeze-up in the Bering Sea occurred 1.0 day later per decade, while freeze-up in the Chukchi/Beaufort Seas occurred 6.9 days later per decade (Markus et al. 2009, p. 11).

Later freeze-up in the Arctic does not necessarily mean that less seasonal sea ice forms by winter's end in the peripheral seas, such as the Bering and Chukchi Seas (Boveng et al. 2008, p. 35). For example, in 2007 (the year when the record minimal Arctic summer sea-ice extent was recorded), the Chukchi Sea did not freeze until early December and the Bering Sea remained largely ice-free until the middle of December (Boveng et al. 2008, p. 35). However, rapid cooling and advancing of sea ice in late December and early January resulted in most of the eastern Bering Sea shelf being ice-covered by mid-January, an advance of 900 km (559 mi), or 30 km per day (19 mi per day). Maximum ice extent occurred in late March, with ice covering much of the shelf, resulting in a near record maximum ice extent. Ice then slowly retreated, and the Bering Sea was not ice-free until almost July. Therefore, winter ice conditions are not necessarily related to the summer-fall ice conditions of the previous year

d-January, an advance of 900 km (559 mi), or 30 km per day (19 mi per day). Maximum ice extent occurred in late March, with ice covering much of the shelf, resulting in a near record maximum ice extent. Ice then slowly retreated, and the Bering Sea was not ice-free until almost July. Therefore, winter ice conditions are not necessarily related to the summer-fall ice conditions of the previous year.

Model Projections of Future Sea Ice

The analysis and synthesis of information presented by the Intergovernmental Panel on Climate Change (IPCC) in its Fourth Assessment Report (AR4) in 2007 represents the scientific consensus view on the causes and future of climate change. The IPCC AR4 used state-of-the-art Atmosphere-Ocean General Circulation Models (GCMs) and a range of possible future greenhouse gas (GHG) emission scenarios to project plausible outcomes globally and regionally, including projections of temperature and Arctic sea-ice conditions through the 21st century.

The GCMs use the laws of physics to simulate the main components of the climate system (the atmosphere, ocean, land surface, and sea ice) and to make projections as to the response of these components to future emissions of GHGs. The IPCC used simulations from about 2 dozen GCMs developed by 17 international modeling centers as the basis for the AR4 (Randall et al. 2007, pp. 596-599). The GCM results are archived as part of the Coupled Model Intercomparison Project-Phase 3 (CMIP3) at the Program for Climate Model Diagnosis and Intercomparison (PCMDI). The CMIP3 GCMs provide projections of future effects that could result from climate change, because they are built on well-known dynamical and physical principles, and they plausibly simulate many large-scale aspects of present-day conditions. However, the coarse resolution of most current climate models dictates careful application on smaller spatial scales in heterogeneous regions.

The IPCC AR4 used six “marker” scenarios from the Special Report on Emissions Scenarios (SRES) (Carter et al

se they are built on well-known dynamical and physical principles, and they plausibly simulate many large-scale aspects of present-day conditions. However, the coarse resolution of most current climate models dictates careful application on smaller spatial scales in heterogeneous regions.

The IPCC AR4 used six “marker” scenarios from the Special Report on Emissions Scenarios (SRES) (Carter et al. 2007, p. 160) to develop climate projections spanning a broad range of GHG emissions through the end of the 21st century under clearly stated assumptions about socioeconomic factors that could influence the emissions. The six “marker” scenarios are classified according to their emissions as “high” (A1F1, A2), et al. 2007, p. 160). The IPCC focused on three of the marker scenarios—B1, A1B, and A2—for its synthesis of the climate modeling efforts, because they represented “low,” “medium,” and “high,” scenarios; this choice stemmed from the constraints of available computer resources that precluded realizations of all six scenarios by all modeling centers (Meehl et al. 2007, p. 753). With regard to these three emissions scenarios, the IPCC Working Group I report noted: “Qualitative conclusions derived from these three scenarios are in most cases also valid for other SRES scenarios” (Meehl et al. 2007, p. 761). It is important to note that the SRES scenarios do not contain additional climate initiatives ( e.g., implementation of the United Nations Framework Convention on Climate Change or the emissions targets of the Kyoto Protocol) beyond current mitigation policies (IPCC 2007, p. 22). The SRES scenarios do, however, have built-in emissions reductions that are substantial, based on assumptions that a certain amount of technological change and reduction of emissions would occur in the absence of climate policies; recent analysis shows that two-thirds or more of all the energy efficiency improvements and decarbonization of energy supply needed to stabilize GHGs is built into the IPCC reference scenarios (Pielke et al

built-in emissions reductions that are substantial, based on assumptions that a certain amount of technological change and reduction of emissions would occur in the absence of climate policies; recent analysis shows that two-thirds or more of all the energy efficiency improvements and decarbonization of energy supply needed to stabilize GHGs is built into the IPCC reference scenarios (Pielke et al. 2008, p. 531).

There are three main contributors to divergence in GCM climate projections: Large natural variations, across-model differences, and the range-in-emissions scenarios (Hawkins and Sutton 2009, p. 1096). The first of these, variability from natural variation, can be incorporated by averaging the projections over decades, or, preferably, by forming ensemble averages from several runs of the same model.

The second source of variation is model to model differences in the way that physical processes are incorporated into the various GCMs. Because of these differences, projections of future climate conditions depend, to a certain extent, on the choice of GCMs used. Uncertainty in the amount of warming out to mid-century is primarily a function of these model-to-model differences. The most common approach to address the uncertainty and biases inherent in individual models is to use the median or mean outcome of several predictive models (a multi-model ensemble) for inference. Excluding models that poorly simulate observational data is also a common approach to reducing the spread of uncertainty among projections from multi-model ensembles.

The third source of variation arises from the range in plausible GHG emissions scenarios. Conditions such as surface air temperature and sea-ice area are linked in the IPCC climate models to GHG emissions by the physics of radiation processes. When CO 2 is added to the atmosphere, it has a long residence time and is only slowly removed by ocean absorption and other processes

lti-model ensembles.

The third source of variation arises from the range in plausible GHG emissions scenarios. Conditions such as surface air temperature and sea-ice area are linked in the IPCC climate models to GHG emissions by the physics of radiation processes. When CO 2 is added to the atmosphere, it has a long residence time and is only slowly removed by ocean absorption and other processes. Based on IPCC AR4 climate models, expected global warming—defined as the change in global mean surface air temperature (SAT)—by the year 2100 depends strongly on the assumed emissions of CO 2 and other GHGs. By contrast, warming out to about 2040-2050 will be largely due to emissions that have already occurred and those that will occur over the next decade (Meehl 2007, p. 749). Thus, conditions projected to mid-century are less sensitive to assumed future emission scenarios. For the second half of the 21st century, however, and especially by 2100, the choice of the emission scenario becomes the major source of variation among climate projections and dominates over natural variability and model-to-model differences (IPCC 2007, pp. 44-46).

Because the SRES group and the IPCC made no judgment on the likelihood of any of the scenarios, and the scenarios were not assigned probabilities of occurrence, one option for representing the full range of variability in potential outcomes, would be to evaluate projections from all models under all marker scenarios for which sea-ice projections are available to the scientific community—A2, A1B, and B1. Another typical procedure for projecting future outcomes is to use an intermediate scenario, such as A1B, to predict changes, or one intermediate and one high scenario ( e.g., A1B and A2) to capture a range of variability.

Several factors suggest that the A1B scenario may be a particularly appropriate choice of scenario to use for projections of sea-ice declines in the Arctic and its marginal seas

other typical procedure for projecting future outcomes is to use an intermediate scenario, such as A1B, to predict changes, or one intermediate and one high scenario ( e.g., A1B and A2) to capture a range of variability.

Several factors suggest that the A1B scenario may be a particularly appropriate choice of scenario to use for projections of sea-ice declines in the Arctic and its marginal seas. First, the A1B scenario is widely used in modeling because it is a “medium” emissions scenario characterized by a future world of very rapid economic growth, global population that peaks in mid-century and declines thereafter, rapid introduction of new and more efficient technologies, and development of energy technologies that are balanced across energy sources, and it contains no assumption of mitigation policies that may or not be realized. Thus, there are a number of studies in the published sea-ice literature that use the A1B scenario and can, therefore, be used for comparative purposes ( e.g., Overland and Wang 2007; Holland et al. 2010; Wang et al. 2010). Second, both the A1B and A2 scenarios project similar declines in hemispheric sea-ice extent out to 2100 (Meehl et al. 2007, Figure 10.13, p. 771); thus, little new understanding is gained by using projections from both scenarios (see discussion of Douglas 2010 in subsequent paragraphs). Third, model projections based on the B1 scenario appear to be overly conservative (Meehl et al. 2007, Figure 10.13, p. 771), in that sea ice is declining even faster than the decline forecasted by the A1B scenario (see discussion at end of this section). Fourth, current global carbon emissions appear to be tracking slightly above (Raupach et al. 2007, Figure 1, p. 10289; LeQuere et al. 2009, Figure 1a, p. 2; Global Carbon Project 2010 at http://www.globalcarbonproject.org/carbonbudget/09/files/GCP2010_CarbonBudget2009_29November2010.pdf ) or slightly below (Manning et al. 2010, Figure 1, p. 377) the A1B trajectory at this point in time

s section). Fourth, current global carbon emissions appear to be tracking slightly above (Raupach et al. 2007, Figure 1, p. 10289; LeQuere et al. 2009, Figure 1a, p. 2; Global Carbon Project 2010 at http://www.globalcarbonproject.org/carbonbudget/09/files/GCP2010_CarbonBudget2009_29November2010.pdf ) or slightly below (Manning et al. 2010, Figure 1, p. 377) the A1B trajectory at this point in time. It may be reasonable to project this or a higher trend in global carbon emissions into the near future (Garnaut et al. 2008, Figure 5, p. 392; Sheehan 2008, Figure 2, p. 220; but see caveat by van Vuuren et al. 2010). Fifth, there is a growing body of opinion that stabilizing GHG emissions at levels well below the A1B scenario ( e.g., at 450 parts per million (ppm), equivalent to a 2 degree Celsius increase in temperature) will be difficult in the absence of substantial policy-mandated mitigation ( e.g., Garnaut et al. 2007, p. 398; den Elzen and Höhne 2008, p. 250; Pielke et al. 2008, pp. 531-532; Macintosh 2009, p. 3; den Elzen et al. 2010, p. 314; Tomassini et al. 2010, p. 418; Anderson and Bows 2011, p. 20), largely as a result of continuing high emissions in certain developed countries, and recent and projected growth in the economies and energy demands of rapidly developing countries ( e.g., Garnaut et al. 2008, p. 392; Auffhammer and Carson 2008, p. 1; Pielke et al. 2008, p. 532; U.S. Energy Information Administration 2010, pp. 123-124, 128). Because of these factors, we conclude that sea-ice projections developed by using the A1B forcing scenario provide an appropriate basis for evaluating potential impacts to habitat and related impacts to the Pacific walrus population in the future.

Our analysis of sea-ice response to global warming within the range of the

Douglas (2010, pp

rgy Information Administration 2010, pp. 123-124, 128). Because of these factors, we conclude that sea-ice projections developed by using the A1B forcing scenario provide an appropriate basis for evaluating potential impacts to habitat and related impacts to the Pacific walrus population in the future.

Our analysis of sea-ice response to global warming within the range of the

Douglas (2010, pp. 4-5) quantified sea-ice projections (from the A2 and A1B scenarios) by 18 CMIP3 GCM models prepared for the IPCC fourth reporting period, as well as 2 GCM subsets which excluded models that poorly simulated the 1979-2008 satellite record of Bering and Chukchi sea-ice conditions. Analyses focused on the annual cycle of sea-ice extent within the range of the Pacific walrus population, specifically the continental shelf waters of the Bering and Chukchi Seas. Models were selected for the two subsets, respectively, when their simulated mean ice extent and seasonality during 1979-2008 were within two standard deviations (SD2) and one standard deviation (SD1) of the observed means. In consideration of observations of ice-free conditions across the Chukchi Sea in recent years in late summer, any models that failed to simulate at least 1 ice-free month in the Chukchi Sea were also excluded from the Chukchi Sea subset ensembles. Ice observations and the projections of individual GCMs were pooled over 10-year periods to integrate natural variability (Douglas 2010, p. 5).

To quantify projected changes in monthly sea-ice extent, Douglas (2010, p. 31) compared future monthly sea-ice projections for the Bering and Chukchi Seas at mid-century (2045-2054) and late-century (2090-2099) with two decades from the observational record (1979-1988 and 1999-2008). The earliest observational period (1979-1988), which coincides with a timeframe during which the Pacific walrus population was considered to be occupying most of its historical range (Fay 1982, pp

mpared future monthly sea-ice projections for the Bering and Chukchi Seas at mid-century (2045-2054) and late-century (2090-2099) with two decades from the observational record (1979-1988 and 1999-2008). The earliest observational period (1979-1988), which coincides with a timeframe during which the Pacific walrus population was considered to be occupying most of its historical range (Fay 1982, pp. 7-21), provides a useful baseline for examining projected changes in sea-ice habitats.

Douglas (2010, p. 7) found that projected median sea-ice extents under both the A1B and A2 forcing scenarios are qualitatively similar in the Bering and Chukchi Seas in all seasons throughout the 21st century. This finding is consistent with the generally similar declines in hemispheric sea-ice extent between the A1B and A2 scenarios out to 2100 (Meehl et al. 2007, Figure 10.13, p. 771). Thus, our decision to focus on ice projections by the A1B forcing scenario (as described above) is further substantiated, as there would be little insight gained by considering the A2 scenario.

The analysis of Douglas (2010, pp. 24, 31) yields mid-century projections that indicate sea-ice extent in the Bering Sea will decline for all months when sea ice has historically been present, i.e., for October through June. The most pronounced reductions in Bering Sea ice extent at mid-century in terms of the percent change from baseline conditions are expected in the months of June and November, which reflects an increasingly early onset of ice-free or nearly ice-free conditions in the early summer and later onset of sea-ice development in the fall. In June, the projected extent of sea ice is −63 percent of the 1979-1988 baseline level, while the projected extent for November is approximately is −88 percent of the baseline level. By late century, substantial declines in Bering Sea ice extent are projected for all months, with losses ranging from 57 percent in April, to 100 percent loss of sea ice in November (Douglas 2010, p. 31)

in the fall. In June, the projected extent of sea ice is −63 percent of the 1979-1988 baseline level, while the projected extent for November is approximately is −88 percent of the baseline level. By late century, substantial declines in Bering Sea ice extent are projected for all months, with losses ranging from 57 percent in April, to 100 percent loss of sea ice in November (Douglas 2010, p. 31). The onset of substantial freezing in the Bering Sea is projected to be delayed until January by late century, with little or no ice projected to remain in May by the end of the century (Douglas 2010, pp. 8, 24, 31).

Historically, sea-ice cover has persisted, to at least some extent, over continental shelf waters of the Chukchi Sea all 12 months of the year, although the extent of sea ice has varied by month. For example, for the 1979-1988 period, the median extent of sea ice varied from about 50 percent in September to essentially 100 percent from late November through early May (Douglas 2010, p. 19). A pattern of extensive sea-ice cover (approaching 100 percent) in late winter and early spring (February-April) is expected to persist through the end of the century.

Projections of sea-ice loss during June in the Chukchi Sea are relatively modest; however, the sea ice is projected to retreat rapidly during the month of July (Douglas 2010, p. 12). Model subset medians project a 2-month ice-free season at mid-century and a 4-month ice-free season at the end of the century, centered around the month of September (Douglas 2010, pp. 8, 22, 24), with some models showing up to 5 months ice-free by end of the century (Douglas 2010, pp. 12, 22, 24). In the most recent observational decade (1999-2008), the southern extent of the Arctic ice pack has retreated and advanced through the Bering Strait in the months of June and November, respectively. By the end of the century, these transition months may shift to May (1 month earlier) and January (2 months later), respectively (Douglas 2010, pp. 12, 25-26)

he century (Douglas 2010, pp. 12, 22, 24). In the most recent observational decade (1999-2008), the southern extent of the Arctic ice pack has retreated and advanced through the Bering Strait in the months of June and November, respectively. By the end of the century, these transition months may shift to May (1 month earlier) and January (2 months later), respectively (Douglas 2010, pp. 12, 25-26).

The projected loss of sea ice involves uncertainty. In discussing this, Douglas 2010 (p. 11) states, in part: “Ice-free conditions in the Chukchi Sea are attained for a 3-month period (August-October) at the end of the century (fig 7) with almost complete agreement among models of the SD2 subset (fig 12). Consequently, a higher degree of confidence can accompany hypotheses or decisions premised on this outcome and timeframe.” Douglas also notes there is greater confidence in projections that the Chukchi Sea will continue to be completely ice covered during February-April at the end of century, and that large uncertainties are prevalent during the melt and freeze seasons, particularly June, November, and December (Douglas 2010, p. 11).

Several other investigations have analyzed model projections of sea-ice change in the Bering and Chukchi Seas and reported results that are consistent with those of Douglas (2010). Wang et al. (2010, p. 258) investigated sea-ice projections to mid-century for the Bering Sea using a subset of models selected on the basis of their ability to simulate sea-ice area in the late 20th century. Their projections show an average decrease in March-April sea-ice coverage of 43 percent by the decade centered on 2050, with a reasonable degree of consistency among models. Boveng et al. (2008, pp. 39-40) analyzed a subset of IPCC AR4 GCM models (selected for accuracy in simulating observed ice conditions) to evaluate spring (April-June) conditions in the Bering Sea out to 2050

ate 20th century. Their projections show an average decrease in March-April sea-ice coverage of 43 percent by the decade centered on 2050, with a reasonable degree of consistency among models. Boveng et al. (2008, pp. 39-40) analyzed a subset of IPCC AR4 GCM models (selected for accuracy in simulating observed ice conditions) to evaluate spring (April-June) conditions in the Bering Sea out to 2050. Their analysis suggested that by mid-century, a modest decrease in the extent of sea ice in the Bering Sea is expected during the month of April, and that ice cover in May will remain variable, with some years having considerably reduced ice cover. June sea-ice cover in the Bering Sea since the 1970s has been consistently low or absent. Their models project that by 2050, ice cover in the Bering Sea will essentially disappear in June, with only a rare year when the ice cover exceeds 0.05 million sq km (0.03 million sq mi) (Boveng et al. 2008, pp. 39-40), a projection similar to that reported by Douglas (2010, p. 24).

Boveng et al. (2009, pp. 44-54) used a subset of IPCC AR4 models to further investigate sea-ice coverage in the eastern Bering Sea (the area of greatest walrus distribution in the Bering Sea), Bering Strait, and the Chukchi Sea out to 2070. For the eastern Bering Sea, they projected that sea-ice coverage will decline in the spring and fall, with fall declines exceeding those of spring. By 2050, average sea-ice extent in November and December would be et al. (2009, pp. 49-50) reported a projected reduction in sea-ice extent for November by 2050, a slight decline for June by 2070, and a clear reduction for November and December by 2070.

Several authors note that sea-ice extent in the Arctic is decreasing at a rate faster than projected by most IPCC-recognized GCMs (Stroeve et al. 2007, p. 1; Overland and Wang 2007, p. 1; Wang and Overland 2009, p. 1; Wang et al. 2010, p. 258), suggesting that GCM projections of 21st century sea-ice losses may be conservative (Douglas 2010, p

a clear reduction for November and December by 2070.

Several authors note that sea-ice extent in the Arctic is decreasing at a rate faster than projected by most IPCC-recognized GCMs (Stroeve et al. 2007, p. 1; Overland and Wang 2007, p. 1; Wang and Overland 2009, p. 1; Wang et al. 2010, p. 258), suggesting that GCM projections of 21st century sea-ice losses may be conservative (Douglas 2010, p. 11, and citations therein) and that ice-free conditions in September in the Arctic may likely be achieved sooner than projected by most models using the A1B forcing scenario. In describing the “faster than forecast” situation, Douglas notes that the minimum ice extents in the Arctic for the summers of 2007-2009 were well below the previous record set in 2005, and concurs that serious consideration must be given to the possibility that the CMIP3 GCM projections collectively yield conservative time frames for sea-ice losses in this century (Douglas 2010, p. 11); i.e., the projected changes he reports for the range of the Pacific walrus may occur sooner than the model projections indicate.

In conclusion, the actual loss of sea ice in recent years in the Arctic has been faster than previously forecast, current GHG emissions are at or above those expected under the A1B scenario that we (and most scientists studying Arctic sea ice) relied on, models converge in predicting the extended absence of sea ice in the Chukchi Sea at the end of the century (Douglas 2010, pp. 12, 29), and there has been a marked loss of sea ice over the Chukchi Sea in the past decade. The best scientific information available gives us a high level of confidence that despite some uncertainty among the models, the projections are generally consistent and provide a reliable basis for us to conclude that sea-ice loss in the range of the Pacific walrus has a high likelihood of continuing.

Effects of Changing Sea-Ice Conditions on Pacific Walruses

The Pacific walrus is an ice-dependent species

tific information available gives us a high level of confidence that despite some uncertainty among the models, the projections are generally consistent and provide a reliable basis for us to conclude that sea-ice loss in the range of the Pacific walrus has a high likelihood of continuing.

Effects of Changing Sea-Ice Conditions on Pacific Walruses

The Pacific walrus is an ice-dependent species. Walruses are poorly adapted to life in the open ocean and must periodically haul out to rest. Floating pack ice creates habitat from which breeding behavior is staged (Fay et al. 1984, p. 81), and it provides a platform for calving (Fay 1982, p. 199), access to offshore feeding areas over the continental shelf of the Bering and Chukchi Seas, passive transportation among feeding areas (Ray et al. 2006, pp. 404-407), and isolation from terrestrial predators and hunters. In this section, we first analyze the effects of sea-ice loss on breeding and calving, because these are essential life-history events that depend on ice in specific seasons. In the second part of this section, we analyze how the anticipated increasing use of coastal haulouts due to the loss of sea-ice habitat may cause localized prey depletion and affect walrus foraging, as well as increase their susceptibility to trampling, predation, and hunting.

Effects of Sea-Ice Loss on Breeding and Calving

Breeding

During the January-to-March breeding season, walruses congregate in the Bering Sea pack ice (Fay 1982, pp. 8-11, 193; Fay et al. 1984, pp. 89-99), where the ice creates the stage for breeding. Females congregate in herds on the ice and the bulls station themselves in the water alongside the herd and perform visual and acoustical displays (Fay 1982, p. 193). Breeding aggregations have been reported southwest of St. Lawrence Island, Alaska, south of Nunivak Island, Alaska, and south of the Chukotka Peninsula in the Gulf of Anadyr, Russia (Fay 1982, p. 21; Mymrin et al. 1990, pp. 105-113)

Females congregate in herds on the ice and the bulls station themselves in the water alongside the herd and perform visual and acoustical displays (Fay 1982, p. 193). Breeding aggregations have been reported southwest of St. Lawrence Island, Alaska, south of Nunivak Island, Alaska, and south of the Chukotka Peninsula in the Gulf of Anadyr, Russia (Fay 1982, p. 21; Mymrin et al. 1990, pp. 105-113). It is unlikely that breeding is tied to a specific geographic location, because of the large seasonal and inter-annual variability in sea-ice cover in the Bering Sea at this time of year. Fay et al. (1984, p. 80) indicate probable changes in the locations of breeding aggregations based on differing amounts of sea ice. We anticipate that seasonal pack ice will continue to form across large areas of the northern Bering Sea, primarily in January-March, and will persist in most years through April (Douglas 2010, p. 25).

The distribution of walruses during the winter breeding season will likely shift in the future in response to changing patterns of sea-ice development. Core areas of winter abundance south of Saint Lawrence Island and the Gulf of Anadyr will likely continue to have adequate ice cover to support breeding aggregations through mid-century, as the extent of sea ice will still be relatively substantial, although slightly diminished from the current extent (Douglas 2010, p. 25). Walruses currently wintering in Northern Bristol Bay will likely shift their distribution northward in response to the projected loss of seasonal pack ice in this region (Douglas 2010, p. 25). By the end of the century, winter sea-ice extent across the Bering Sea is expected to be greatly reduced, and the median sea-ice edge is projected to be farther to the north (Douglas 2010, p. 25). Based on these projections, core areas of winter abundance and breeding aggregations will likely shift farther north

ponse to the projected loss of seasonal pack ice in this region (Douglas 2010, p. 25). By the end of the century, winter sea-ice extent across the Bering Sea is expected to be greatly reduced, and the median sea-ice edge is projected to be farther to the north (Douglas 2010, p. 25). Based on these projections, core areas of winter abundance and breeding aggregations will likely shift farther north. Potentially, the breeding aggregations may shift into areas north of the Bering Strait in the southern Chukchi Sea in some years by the end of the century (Douglas 2010, pp. 24, 28).

Although the location of winter breeding aggregations will likely shift in response to projected reductions in sea-ice extent, sea-ice platforms for herds of females will persist during the breeding season; therefore, we conclude that suitable conditions for breeding will likely persist into the foreseeable future. We have no information that indicates that the specific location of the ice is important, and sea ice is expected to remain over shallow, food-rich areas. Therefore, we do not consider changes in sea-ice extent during the winter breeding season to be a threat now or in the foreseeable future.

Calving

Female walruses typically give birth to a single calf in May on sea ice, shortly before or during the northward spring migration through the Bering Strait. By mid-century, ice extent in the Bering Strait Region is projected to be reduced during the May calving season, and by end of century, the Bering Sea is projected to be largely sea-ice-free during the month of May (Douglas 2010, p. 25). As is the case with breeding, the birth of a calf and the natal period in the weeks that follow are probably not tied to specific geographic locations. It is reasonable to assume that suitable ice conditions for calving and post-calving activity on sea ice will persist into the foreseeable future, even though the location of favorable ice conditions is likely to shift further to the north over time

s is the case with breeding, the birth of a calf and the natal period in the weeks that follow are probably not tied to specific geographic locations. It is reasonable to assume that suitable ice conditions for calving and post-calving activity on sea ice will persist into the foreseeable future, even though the location of favorable ice conditions is likely to shift further to the north over time.

We conclude that changes in sea ice during the spring calving season (April-May) are not a threat now or in the foreseeable future. We have no

Summary of Effects of Sea-Ice Loss on Breeding and Calving

Breeding and calving activities utilize ice as a platform in the months of January through May. Based on our current understanding of these activities, the specific location of the ice is not important. Although sea-ice extent is projected to move northward over time, sea ice is expected to persist in these months and be available for these life history functions. Therefore, we do not consider changes in sea-ice extent to be a threat to breeding or calving activities now or in the foreseeable future.

Effects of Increasing Dependence on Coastal Haulouts Due to Sea-Ice Loss

We begin this discussion with a summary of sea-ice loss projections and recent observations. We follow with an analysis of the potential effects to Pacific walrus from an increasing dependence on coastal haulouts, particularly in the Chukchi Sea, and examine the use of coastal haulouts by Atlantic walrus as a potential analog for Pacific walrus coastal haulout use. We analyze potential effects of increased dependency on coastal haulouts resulting from the loss of sea-ice habitats. Some of the effects to Pacific walrus that we have identified as a result of increasing dependence on coastal haulouts (i.e., trampling, predation, and hunting) would typically be discussed under other Factors

tic walrus as a potential analog for Pacific walrus coastal haulout use. We analyze potential effects of increased dependency on coastal haulouts resulting from the loss of sea-ice habitats. Some of the effects to Pacific walrus that we have identified as a result of increasing dependence on coastal haulouts (i.e., trampling, predation, and hunting) would typically be discussed under other Factors. These effects are discussed in this section in the context of responses to declining sea ice; however, it should be noted that we also discuss predation under Factor C ( Disease or Predation ), and hunting under Factor B ( Overutilization for Commercial, Recreational, Scientific, or Educational Purposes ) and Factor D ( The Inadequacy of Existing Regulatory Mechanisms ).

Summary of Sea-Ice Loss Projections

Sea ice has historically persisted over continental shelf regions of the Chukchi Sea through the entire melt season. Over the past decade, sea ice has begun to retreat beyond shallow continental shelf waters in late summer. The recent trend of rapid ice loss from continental shelf regions of the Chukchi Sea in July and August is projected to persist, and will likely accelerate in the future (Douglas 2010, p. 12). The onset of ice formation in the fall over continental shelf regions in the Chukchi and Bering Seas is expected to be delayed, and by mid-century (2045-2054), ice-free conditions over most continental shelf regions of the Chukchi Sea are projected to persist for 2 months (August-September). By late century, ice-free (or nearly sea-ice-free) conditions may persist for 3 months, and extend to 4 to 5 months in some years (Douglas 2010, pp. 8, 12, 22, 27). The average number of ice-free months in the Bering Sea is projected to increase from the approximately 5.5 months currently, to approximately 6.5 and 8.5 months at mid- and end of century, respectively (Douglas 2010, pp. 12, 27)

ber). By late century, ice-free (or nearly sea-ice-free) conditions may persist for 3 months, and extend to 4 to 5 months in some years (Douglas 2010, pp. 8, 12, 22, 27). The average number of ice-free months in the Bering Sea is projected to increase from the approximately 5.5 months currently, to approximately 6.5 and 8.5 months at mid- and end of century, respectively (Douglas 2010, pp. 12, 27).

Observed and Expected Responses of Pacific Walruses to Declining Sea-Ice Habitats

Adult male walruses make greater use of coastal haulouts during ice-free seasons than do females and dependent young, and consequently, have a broader distribution during ice-free seasons. Several thousand bulls remain in the Bering Sea through the ice-free summer months, where they make foraging excursions from coastal haulouts in Bristol Bay, Alaska and the Gulf of Anadyr, Russia. The size of these haulouts has changed over time; for example, at Round Island, the number of hauled out walruses grew from about 3,000 animals in the late 1950s to about 12,000 in the early 1980s (Jay and Hills 2005, p. 193), and has subsequently declined to 2,000-5,000 animals in the past decade (Sell and Weiss 2010, p. 12). The reasons for changes in walrus haulout use in the Bering Sea are poorly understood. Factors that could affect use of haulouts include; prey abundance and distribution, walrus density, and physical alteration or chronic disturbance at the haulouts (Jay and Hills 2005, p. 198). Tagged males traveled up to 130 km (81 mi) to feed from haulout sites in Bristol Bay (Jay and Hills 2005, p. 198). Because the benthic densities are poorly documented, it is not possible to link the changes in haulout use by males to prey depletion. However, non-use of areas with shallow depths closer to the haulouts suggests prey was not adequate for effective foraging (Jay and Hills 2005, p. 198)

agged males traveled up to 130 km (81 mi) to feed from haulout sites in Bristol Bay (Jay and Hills 2005, p. 198). Because the benthic densities are poorly documented, it is not possible to link the changes in haulout use by males to prey depletion. However, non-use of areas with shallow depths closer to the haulouts suggests prey was not adequate for effective foraging (Jay and Hills 2005, p. 198). Males have an advantage over females in that they are bigger and stronger and have no responsibilities related to the care of calves, and thus, can travel as far as necessary to locate food. Currently, males utilize terrestrial haulouts for 5 months or more (Jay and Hills 2005, p. 198). It is unlikely that the projected increase in ice-free months in the Bering Sea will alter male behavior or survival rates at terrestrial haulouts because the adult males that utilize Bering Sea haulouts do not rely on sea ice as a foraging platform. Indirect effects of global climate change on walrus prey species in this region are considered separately below in the section: Effects of Global Climate Change on Pacific Walrus Prey Species.

Most of the Pacific walrus population (adult females, calves, juveniles, and males that have not remained at coastal haulouts in the Bering Sea) migrate northward in spring following the retreating pack ice through the Bering Strait to summer feeding areas over the continental shelf in the Chukchi Sea. Historically, sufficient pack-ice habitat has persisted over continental shelf regions of the Chukchi Sea through the summer months such that walruses in the Chukchi Sea did not rely on coastal haulouts with great frequency or in large numbers. Over the past decade, however, sea ice has begun to retreat north beyond shallow continental shelf waters of the Chukchi Sea in late summer. This has caused walruses to relocate to coastal haulouts, which they use as sites for resting between foraging excursions

hrough the summer months such that walruses in the Chukchi Sea did not rely on coastal haulouts with great frequency or in large numbers. Over the past decade, however, sea ice has begun to retreat north beyond shallow continental shelf waters of the Chukchi Sea in late summer. This has caused walruses to relocate to coastal haulouts, which they use as sites for resting between foraging excursions. The number of walruses using land-based haulouts along the Chukchi Sea coast during the summer months, and the duration of haulout use, has increased substantially over the past decade, with up to several tens of thousands of animals hauling out at some locations along the coast of Russia during ice-free periods (Ovsyanikov et al. 2007, pp. 1-2; Kochnev 2008, p. 17-20, Kavry et al. 2008, p. 248-251). Coastal haulouts have also begun to form along the Arctic coast of Alaska in recent years (2007, 2009, and 2010) when sea ice retreated north of the continental shelf in late summer (Service 2010, unpublished data). The occupation of terrestrial haulouts along the Chukchi Sea coast for extended periods of time in late summer and fall represents a relatively new and significant change from traditional habitat use patterns. The consequences of this observed and projected shift in habitat use patterns is the primary focus of our analysis.

As sea ice withdraws from offshore feeding areas over the continental shelf of the Chukchi Sea, walruses are expected to become increasingly dependent on coastal haulouts as a foraging base during the summer months. With a delay the onset of ice formation in the fall, and in the absence of sea-ice cover in the southern Chukchi Sea and northern Bering Sea in the summer, walruses will likely remain at coastal haulouts for longer periods of time until sea ice reforms in the fall or early winter

walruses are expected to become increasingly dependent on coastal haulouts as a foraging base during the summer months. With a delay the onset of ice formation in the fall, and in the absence of sea-ice cover in the southern Chukchi Sea and northern Bering Sea in the summer, walruses will likely remain at coastal haulouts for longer periods of time until sea ice reforms in the fall or early winter. By the end of the century, dependence on Chukchi Sea coastal haulouts by mixed groups of walruses

Increased dependence on coastal haulouts creates the following potential impacts for walruses: Changes in foraging patterns and prey depletion; increased vulnerability to mortality or injury due to trampling, especially for calves, juveniles, and females; greater vulnerability to mortality or injury from predation; and greater vulnerability to mortality due to hunting. Each is discussed in detail below.

Changes in Foraging Patterns and Prey Depletion

The loss of seasonal pack ice from continental shelf areas of the Chukchi Sea is expected to reduce access to traditional foraging areas across the continental shelf and increase competition among individuals for food resources in areas close to haulouts. Information regarding the density of walrus prey items accessible from coastal haulouts is limited; however, some haulouts have supported sizable concentrations of animals (up to several tens of thousands of animals) for periods of up to 4 months in recent years (Kochnev 2010, pers. comm.). Many walrus prey species are slow growing and potentially vulnerable to overexploitation, and intensive foraging from coastal haulouts by large numbers of walruses may eventually result in localized prey depletion (Ray et al. 2006, p. 412). A walrus requires approximately 29 to 74 kg (64 to 174 lbs) of food per day (Fay 1982, p. 160), and may consume 4,000 to 6,000 clams in one feeding bout (Ray et al. 2006, pp

ey species are slow growing and potentially vulnerable to overexploitation, and intensive foraging from coastal haulouts by large numbers of walruses may eventually result in localized prey depletion (Ray et al. 2006, p. 412). A walrus requires approximately 29 to 74 kg (64 to 174 lbs) of food per day (Fay 1982, p. 160), and may consume 4,000 to 6,000 clams in one feeding bout (Ray et al. 2006, pp. 408, 412); therefore, when large numbers of walruses are concentrated on coastal haulouts, a large amount of prey (whether clams or other types of prey) must be available to support them.

The presence of large numbers of walruses at a coastal haulout over an extended time period could eventually lead to localized prey depletion. The most likely response to localized prey depletion will be for walruses to seek out and colonize other terrestrial haulouts that have suitable foraging areas (Jay and Hills 2005, p. 198). However, prey densities along the Arctic coast are not uniform (Grebmeier et al. 1989, p. 257; Feder et al. 1994, pp. 176-177; Grebmeier et al. 2006b, p. 346), and many coastal areas which provide the physical features of a suitable haulout, may not have sufficient food sources. A visual comparison of areas of high benthic production ( e.g., Springer et al. 1996, p. 209; Dunton et al. 2005, p. 3468; Grebmeier et al. 2006b, p. 346) and areas that have supported large terrestrial haulouts of walruses ( e.g., Cape Inkigur, Cape Serdtse-Kamen) indicates that walruses have historically selected sites near areas of very high benthic productivity. Benthic productivity along part of the western shore of Alaska (i.e., along the eastern edge of the Chukchi Sea) is low because of the nutrient-poor waters of the Alaska Coastal Current, especially for instance, in the Kotzebue Sound (Dunton et al. 2005, p. 3468; Dunton et al. 2006, p. 369; Grebmeier et al. 2006b, p. 346). Consequently, the number of sites with adequate food resources to support large aggregations of walruses is likely limited

shore of Alaska (i.e., along the eastern edge of the Chukchi Sea) is low because of the nutrient-poor waters of the Alaska Coastal Current, especially for instance, in the Kotzebue Sound (Dunton et al. 2005, p. 3468; Dunton et al. 2006, p. 369; Grebmeier et al. 2006b, p. 346). Consequently, the number of sites with adequate food resources to support large aggregations of walruses is likely limited.

A consequence of prey depletion could be an increased energetic cost to locate sufficient food resources (Sheffield and Grebmeier 2009, p. 770; Jay et al. 2010b, pp. 9-10). Energetic costs to walruses will increase if they have to travel greater distances to locate prey, or foraging efficiency is reduced as a consequence of lower prey densities (Sheffield and Grebmeier 2009, p. 770; Jay et al. 2010b, pp. 9-10). Observations by Russian scientists at haulouts along the coast of Chukotka (along the western side of the Chukchi Sea) in recent years suggest that rates of calf mortality and poor body condition of adult females are inversely related to the persistence of sea ice over offshore feeding areas and the length of time that animals occupy coastal haulouts (Nikiforov et al. 2007, pp. 1-2; Ovsyanikov et al. 2007, pp. 1-3; Kochnev 2008, pp. 17-20; Kochnev et al. 2008, p. 265). Over time, poor body condition could lead to lower reproductive rates, greater susceptibility to disease or predation, and ultimately higher mortality rates (Kochnev 2004, pp. 285-286; Kochnev et al. 2008, p. 265; Sheffield and Grebmeier 2009, p. 770).

The energetic cost of swimming a long distance is demonstrated by the observations made in the summer of 2007, when the melt season in the Chukchi Sea began slowly, and then sea-ice retreat accelerated rapidly in July and August. The continental shelf of the Chukchi Sea was sea-ice-free by mid-August; the ice edge eventually retreated hundreds of miles north of the shelf, and ice did not re-form over the continental shelf until late October (National Snow and Ice Data Center, 2007)

e in the summer of 2007, when the melt season in the Chukchi Sea began slowly, and then sea-ice retreat accelerated rapidly in July and August. The continental shelf of the Chukchi Sea was sea-ice-free by mid-August; the ice edge eventually retreated hundreds of miles north of the shelf, and ice did not re-form over the continental shelf until late October (National Snow and Ice Data Center, 2007). Ovsyanikov et al. (2007, pp. 2-3) reported that many of the walruses arriving at Wrangel Island, Russia, in August 2007 were emaciated and weak, some too exhausted to flee or defend themselves from polar bears patrolling the coast. The authors attributed the poor condition of these animals to the rapid retreat of sea ice off of the shelf in July to waters too deep for walrus to feed. They also noted that the exhausted walruses could not find enough food near the island for recovery (Ovsyanikov et al. 2007, p. 3).

Females with dependent young are likely to be disproportionally affected by prey depletion and increased reliance on coastal haulouts as a foraging base. Females with dependent young require two to three times the amount of food needed by nonlactating females (Fay 1982, p. 159). Over the past decade, females and dependent calves have responded to the loss of sea ice in late summer by occupying coastal haulouts along the coast of Chukotka, Russia, and more recently (2007-2010) haulouts along the coast of Alaska. Females typically nurse their calves between short foraging forays from sea-ice platforms situated over productive forage areas (Ray et al. 2006, pp. 404-407). Drifting ice provides walrus passive transport and access to new foraging areas with minimal effort. In 2007, radio-tagged females traveled on average, 30.7 km (19 mi) on foraging trips from several haulouts located along the Chukotka coastline (Kochnev et al. 2008, p. 265)

es between short foraging forays from sea-ice platforms situated over productive forage areas (Ray et al. 2006, pp. 404-407). Drifting ice provides walrus passive transport and access to new foraging areas with minimal effort. In 2007, radio-tagged females traveled on average, 30.7 km (19 mi) on foraging trips from several haulouts located along the Chukotka coastline (Kochnev et al. 2008, p. 265). Although we do not know the average distance of foraging trips taken from an ice platform, in general, we would expect them to be relatively short, because when the ice is over productive prey areas, the female only has to dive to the bottom and back up to the ice (Ray et al. 2006, pp. 406-407). Because calves do not have the swimming endurance of adults, if sufficient prey is not located within the swimming distance of the calf, the female either may not be able to obtain adequate nutrition or the calf may be abandoned when the female travels to locations beyond the swimming capability of the calf (Cooper et al. 2006, pp. 98-102). Lack of adequate prey for females could eventually lead to reduced body condition, lower reproductive success, and potentially death. Abandoned calves could face increased mortality from drowning, starvation, or predation.

In summary, by the end of the 21st century, ice-free conditions are expected to persist across the continental shelf of the Chukchi Sea for a period of up to several months (Douglas 2010). Based

The degree to which depletion of food resources near coastal haulouts will limit population size will depend on a variety of factors, including: The location of coastal walrus haulouts, the number of animals utilizing the haulouts, the duration of time walruses occupy the haulouts, and the robustness of the prey base within range of those haulouts. However, it is highly unlikely that the current population can be sustained from coastal haulouts alone

oastal haulouts will limit population size will depend on a variety of factors, including: The location of coastal walrus haulouts, the number of animals utilizing the haulouts, the duration of time walruses occupy the haulouts, and the robustness of the prey base within range of those haulouts. However, it is highly unlikely that the current population can be sustained from coastal haulouts alone. In particular, females and their calves will be susceptible to the increased energetic demands of foraging from coastal haulouts. We do not anticipate effects to males using coastal haulouts in the Bering Sea, because their current behavior can continue unaltered into the future. We do not have evidence that prey depletion is currently having a population-level effect on the Pacific walrus. Our concern is based on projections of continued and more extensive sea-ice loss that will force the animals onto land. Therefore, we conclude that loss of sea-ice habitat, leading to dependence on coastal haulouts and localized prey depletion, will contribute to other negative impacts associated with sea-ice loss, and is a threat to the Pacific walrus in the foreseeable future.

Increased Vulnerability to Disturbances and Trampling

Another consequence of greater reliance on coastal haulouts is increased levels of disturbances and increased rates of mortalities and injuries associated with trampling. Walruses often flee land or ice haulouts in response to disturbances. Disturbance can come from a variety of sources, either anthropogenic ( e.g., hunters, airplanes, ships) or natural ( e.g., predators) (Fay et al. 1984, pp. 114-118, Kochnev 2004, p. 286). Haulout abandonment represents an increase in energy expenditure and stress, and disturbance events at densely packed coastal haulouts can result in intra-specific trauma and mortalities (COSEWIC 2006, pp. 25-26). Although disturbance-related mortalities at all-male haulouts in the Bering Sea are relatively uncommon (Fay and Kelly 1980, p. 244; Kochnev 2004, p

, pp. 114-118, Kochnev 2004, p. 286). Haulout abandonment represents an increase in energy expenditure and stress, and disturbance events at densely packed coastal haulouts can result in intra-specific trauma and mortalities (COSEWIC 2006, pp. 25-26). Although disturbance-related mortalities at all-male haulouts in the Bering Sea are relatively uncommon (Fay and Kelly 1980, p. 244; Kochnev 2004, p. 285), the situation at mixed haulouts is different; because of their smaller size, calves, juveniles, and females are more susceptible to trampling injuries and mortalities (Fay and Kelly 1980, pp. 226, 244). Females likely avoid using terrestrial haulouts because their offspring are vulnerable to predation and trampling (Nikiforov et al. 2007, pp. 1-2; Ovsyanikov et al. 2007, pp. 1-3; Kochnev 2008, pp. 17-20; Kochnev et al. 2008, p. 265).

When walruses are disturbed on ice floes, escape into the water is relatively easy because fewer animals are concentrated in one area. In comparison, aggregations of walruses on land are often very large in number, densely packed, and “layered” several animals deep (Nikiforov et al. 2007, p. 2). The presence of some large males in groups using Chukchi Sea coastal haulouts increases the danger to calves, juveniles, and females. Consequently, the probability of direct mortality or injury due to trampling during stampedes is greater at terrestrial haulouts than it is on pack ice (USFWS 1994, p. 12). Also, whether on ice or land, calves may be abandoned as a result of disturbance to a haulout (Fay et al. 1984, p. 118).

In addition, sources of disturbance are expected to be greater at terrestrial haulouts than in offshore pack ice habitats, because the level of human activity such as hunting, fishing, boating, and air traffic is far greater along the coast. Haulout abandonment has been documented from these sources (Fay et al. 1984; p. 114; Kochnev 2004, pp. 285-286). There is also a greater chance of disturbance from terrestrial animals (Kochnev 2004, p. 286)

to be greater at terrestrial haulouts than in offshore pack ice habitats, because the level of human activity such as hunting, fishing, boating, and air traffic is far greater along the coast. Haulout abandonment has been documented from these sources (Fay et al. 1984; p. 114; Kochnev 2004, pp. 285-286). There is also a greater chance of disturbance from terrestrial animals (Kochnev 2004, p. 286). As sea ice declines, and both polar bears and walruses are increasingly forced onto land bordering the Chukchi Sea, we anticipate that there will be greater interaction between the two species, especially during the summer. We expect that one outcome of increased interactions will be increased walrus mortality due to predation (discussed below). Of equal, or more importance than predation is the disturbance caused at a haulout through the arrival or presence of a polar bear, which can cause stampeding. Repeated stampeding also increases energy expenditure and stress levels, and may cause walruses to abandon the haulout (COSEWIC 2006, p. 25).

Losses that can occur when large numbers of walruses use terrestrial haulouts are illustrated by observations in 2007, along the coast of Chukotka, Russia. In response to summer sea-ice loss in 2007, walruses began to arrive at coastal haulouts in July, a month earlier than previously recorded (Kochnev 2008, pp. 17-20). Coastal aggregations ranged in size from 4,500 up to 40,000 animals (Ovsyanikov et al. 2007, pp. 1-2; Kochnev 2008, p. 17-20, Kavry et al. 2008, p. 248-251). Hunters from the Russian coastal villages of Vankarem and Ryrkaipii reported more than 1,000 walrus carcasses (mostly calves of the year and aborted fetuses) at coastal haulouts near the communities in September 2007 (Nikiforov et al. 2007, p. 1; Kochnev 2008, pp. 17-20). Noting the near absence of calves amongst the remaining animals, Kochnev (2008, pp. 17-20) estimated that most of the 2007 cohort using the site had been lost

l villages of Vankarem and Ryrkaipii reported more than 1,000 walrus carcasses (mostly calves of the year and aborted fetuses) at coastal haulouts near the communities in September 2007 (Nikiforov et al. 2007, p. 1; Kochnev 2008, pp. 17-20). Noting the near absence of calves amongst the remaining animals, Kochnev (2008, pp. 17-20) estimated that most of the 2007 cohort using the site had been lost. Approximately 1,500 walrus carcasses (predominately adult females) were also reported near Cape Dezhnev in late October (Kochnev 2007, pers. comm.). Russian investigators estimate that between 3,000 and 10,000 animals died along the Chukotka coastline during the summer and fall of 2007, primarily from trampling associated with disturbance events at the haulouts (Kochnev 2010, pers. comm.).

Relatively few large mortality events at coastal haulouts have been documented in the past, but they have occurred (Fay 1982, p. 226). For

We recognize that Atlantic walruses (including females and calves) utilize coastal haulouts to a greater extent than Pacific walruses, foraging from shore along a relatively narrow coastal shelf; a situation that is similar to what Pacific walrus may experience in the future during ice-free months in the Chukchi Sea. However, Atlantic walrus occupy an area with abundant remote islands that are free or nearly free from disturbance from humans or terrestrial mammals. In essence, their insular habitats function in a manner analogous to the pack ice of the Pacific walrus, providing a refugium from disturbance. In contrast, when Pacific walruses are restricted to terrestrial haulouts, they face disturbance from a variety of terrestrial predators and scavengers, including bears, wolverines, wolves, and feral dogs, and higher levels of anthropogenic disturbances, because their haulouts are at the edge of continental land masses and there are very few islands in the Bering and Chukchi Seas

from disturbance. In contrast, when Pacific walruses are restricted to terrestrial haulouts, they face disturbance from a variety of terrestrial predators and scavengers, including bears, wolverines, wolves, and feral dogs, and higher levels of anthropogenic disturbances, because their haulouts are at the edge of continental land masses and there are very few islands in the Bering and Chukchi Seas. Sea ice, which has typically acted as a refugium from disturbance for Pacific walruses, particularly for females and young in the Chukchi Sea, will be lost entirely, or almost entirely, for increasingly long time periods annually in the foreseeable future. Therefore, although use of coastal haulouts is a form of adaptability available to Pacific walruses, it comes with negative impacts that are not associated with coastal haulouts for Atlantic walruses.

In summary, we anticipate that Pacific walruses will become increasingly dependent on coastal haulouts as sea ice retreats earlier off the continental shelf and the Bering and Chukchi Seas become ice-free for increasingly longer periods of time. The protection normally provided to females and calves by the dispersal of smaller groups of animals across a wide expanse of sea ice will be lost during periods of ice-free or nearly ice-free conditions. Significant mortality events from trampling have been documented at large haulouts, and we anticipate that they will continue with much greater frequency into the foreseeable future, resulting in increased mortality, particularly of calves and females. Therefore, we conclude that disturbances and trampling at haulouts is a threat to the Pacific walrus now and in the foreseeable future.

Increased Vulnerability to Predation and Hunting

As Pacific walruses become more dependent on coastal haulouts, they will become more susceptible to predation and hunting (Kochnev 2004, p. 286)

ulting in increased mortality, particularly of calves and females. Therefore, we conclude that disturbances and trampling at haulouts is a threat to the Pacific walrus now and in the foreseeable future.

Increased Vulnerability to Predation and Hunting

As Pacific walruses become more dependent on coastal haulouts, they will become more susceptible to predation and hunting (Kochnev 2004, p. 286). Although hunting and predation are discussed separately below (see Factors B and C, respectively), we also consider them here due to their relationship to increased loss of sea-ice habitat.

Because of their large size and tusks, adult walruses are much less susceptible to predation than are young animals or females. Females likely avoid using terrestrial haulouts because their offspring are vulnerable to predation (Kochnev 2004, p. 286; Ovsyanikov et al. 2007, pp. 1-4; Kelly 2009, p. 302). Apparently, some polar bear routinely rush herds to cause a stampede, expecting that some calves will be left behind (Nikulin 1941; Popove 1958, 1960; as cited in Fay et al. 1984, p. 119). As sea ice declines in the foreseeable future, increased use of terrestrial habitats by both polar bears and walruses will likely lead to increased interaction between them, and most likely an increase in mortality, particularly of calves. We conclude that loss of sea ice, which will force increased overlap between these two species, will increase mortality from polar bears through direct take or indirect take due to trampling during stampedes. See the section on predation in Factor C below, for further information.

Large concentrations of walruses on shore for longer periods of time could result in increased harvest levels if the terrestrial haulouts form near coastal villages and environmental conditions allow access to haulouts. Kochnev (2004, pp

rom polar bears through direct take or indirect take due to trampling during stampedes. See the section on predation in Factor C below, for further information.

Large concentrations of walruses on shore for longer periods of time could result in increased harvest levels if the terrestrial haulouts form near coastal villages and environmental conditions allow access to haulouts. Kochnev (2004, pp. 285-286) notes that many of the haulouts along the Chukotka coast are situated near coastal villages, and hunting activities at the haulouts can result in stampedes and cause movements from one haulout to another. Some communities in Chukotka situated in close proximity to the new haulouts have responded by developing hunting restrictions to limit disturbances to resting animals (Patrol 2008, p. 1; Kavry 2010, pers. comm.; Kochnev 2010 pers. comm.). See the section on Subsistence Hunting in Factor B below, for further information.

Summary of the Effects of Sea-Ice Loss on Pacific Walruses

The Pacific walrus is an ice-dependent species. Changes in the extent, volume, and timing of the sea-ice melt and onset of freezing in the Bering and Chukchi Seas have been documented and described earlier in this finding, there are reliable projections that more extensive changes will occur in the foreseeable future. We expect these changes in sea ice will cause significant changes in the distribution and habitat-use patterns of Pacific walruses. At this time we anticipate that breeding behavior in winter and calving in the early spring will not be impacted by expected changes to sea-ice conditions, although the locations where these events occur will most likely change as the location of available sea ice shifts to the north.

With the loss of summer sea ice, the most obvious change, which has already been observed, will be a greater dependence on terrestrial haulouts by both sexes and all age groups

calving in the early spring will not be impacted by expected changes to sea-ice conditions, although the locations where these events occur will most likely change as the location of available sea ice shifts to the north.

With the loss of summer sea ice, the most obvious change, which has already been observed, will be a greater dependence on terrestrial haulouts by both sexes and all age groups. Although walruses of both sexes are capable of using terrestrial haulouts, historically, adult males have used terrestrial haulouts, particularly in the Bering Sea, to a much greater extent than females, calves, and juveniles. The loss of summer sea ice means that walruses of both sexes, but females and their young in particular, will be using coastal haulouts for longer periods of time. This change is particularly notable in the Chukchi Sea, which has historically had sufficient sea ice in the summer so that females and calves could remain over the shallow continental shelf throughout the summer. Since approximately 2005, the Chukchi Sea has become ice-free or nearly so during part of the summer. This condition is projected to increase over time, and may occur faster than forecast. The consequences of this shift from sea ice to increasing use of land include: Risk of localized prey depletion; increased energetic costs to reach prey, resulting in decreased body condition; calf abandonment; increased mortality from stampedes, especially to females, juveniles, and calves; and potentially increased exposure to predation and hunting. These events are expected to reduce survivorship.

As large numbers of animals are concentrated at coastal haulouts, prey

Periodic ice-free conditions, as are currently occurring, are expected to lead to higher mortality rates, primarily through trampling at haulouts when walruses congregate in large numbers. Although of concern, if these events happen sporadically, as has been the case in the past, the population may be able to recover between harsh years

mbers of animals are concentrated at coastal haulouts, prey

Periodic ice-free conditions, as are currently occurring, are expected to lead to higher mortality rates, primarily through trampling at haulouts when walruses congregate in large numbers. Although of concern, if these events happen sporadically, as has been the case in the past, the population may be able to recover between harsh years. Although trampling mortalities have been documented in the past, increasing use of terrestrial haulouts, the higher probability of disturbance occurring at these haulouts, and in the near-term, the very large numbers of animals using particular haulouts, increases the probability that mortality from trampling will become a more regular event.

The increasing reliance of both polar bears and walruses on terrestrial environments during ice free periods will likely result in increased interactions between these two species. Polar bear predation and associated disturbances at densely crowded coastal haulouts will likely contribute to increased mortality levels, particularly of calves, and may displace animals from preferred feeding areas. Hunting activity at coastal haulouts does not appear to be a significant source of mortality at the present time, but may become more of a factor in the future. Local hunting restrictions at coastal haulouts have been established in some communities in Chukotka to reduce disturbance-related mortalities. The efficacy of efforts to mitigate sources of anthropogenic disturbances at coastal walrus haulouts (including hunting, boating and air traffic) will influence the degree to which these factors will affect the Pacific walrus population. See Factors B and C for further discussion on harvest and predation.

In conclusion, the loss of sea-ice habitat creates several stressors on the Pacific walrus population

cy of efforts to mitigate sources of anthropogenic disturbances at coastal walrus haulouts (including hunting, boating and air traffic) will influence the degree to which these factors will affect the Pacific walrus population. See Factors B and C for further discussion on harvest and predation.

In conclusion, the loss of sea-ice habitat creates several stressors on the Pacific walrus population. These stressors include: localized prey depletion; increased energetic costs to reach prey, resulting in decreased body condition; calf abandonment; increased mortality from stampedes, especially to females, juveniles, and calves; and increased exposure to predation and hunting. Because the Pacific walrus range is large, and the animals are not all in the same place at the same time, not all stressors are likely to affect the entire population in a given year. However, all stressors represent potential sources of increased mortality over the current condition, in which these stressors occur infrequently. In the foreseeable future, as the frequency of sea-ice loss in the summer and fall over the continental shelves increases to a near-annual event and the length of time ice is absent over the continental shelf increases from 1 to up to 5 months, we expect the effects on walruses to be heightened and a greater percentage of the population to be affected. Increased direct and indirect mortality, particularly of calves, juveniles, and females, will result in a declining population over time. Consequently, we conclude that the destruction, modification, and curtailment of sea-ice habitat is a threat to the Pacific walrus.

Outcome of Bayesian Network Analyses

Both the Service and USGS Bayesian network analyses (Garlich-Miller et al. 2011; Jay et al. 2010b) considered changes in sea ice projected through the 21st century. In both cases, the results indicate that expected loss of sea ice is an important risk factor for Pacific walrus population status over time

of sea-ice habitat is a threat to the Pacific walrus.

Outcome of Bayesian Network Analyses

Both the Service and USGS Bayesian network analyses (Garlich-Miller et al. 2011; Jay et al. 2010b) considered changes in sea ice projected through the 21st century. In both cases, the results indicate that expected loss of sea ice is an important risk factor for Pacific walrus population status over time. The USGS analysis deals more directly with projected outcomes of the Pacific walrus population, including the influence of sea-ice loss under different potential conditions (Jay et al. 2010b, p. 40). For the normative sea ice run (see Jay et al. 2010b for details), the probability of Pacific walruses becoming vulnerable, rare, or extirpated increases over time, from approximately 22 percent in 2050, to about 35 percent by 2075, and 40 percent in 2095 (Jay et al. 2010b, p. 40). A “worst case” influence run was also evaluated. For the worst case, model outputs were selected that have both the greatest number of ice-free months and the least ice extent for the Bering and Chukchi Seas and, therefore, represent the worst possible situation. The outcome for the worst case influence run for sea ice indicated that the probability of Pacific walruses becoming vulnerable, rare, or extirpated approximately doubles at mid-century to 40 percent, and reaches approximately 45 percent at 2075 (Jay et al. 2010b, p. 40). At the end of 21st century, the probability of Pacific walruses becoming vulnerable, rare, or extirpated in both the worst case scenario and the normative run are essentially equal, at about 40 percent; an outcome that is due to the projected amount of sea-ice loss being basically the same under the worst case and normative case by the end of the century. We note, however, that the models and emissions scenarios used by the IPCC in 2007 were the basis for this analysis

vulnerable, rare, or extirpated in both the worst case scenario and the normative run are essentially equal, at about 40 percent; an outcome that is due to the projected amount of sea-ice loss being basically the same under the worst case and normative case by the end of the century. We note, however, that the models and emissions scenarios used by the IPCC in 2007 were the basis for this analysis. Thus, it is possible that the “worst case scenario” reflects the “faster than forecast” loss of sea ice that may be realized if sea-ice loss continues on the current downward trend that began in 1979 (National Snow and Ice Data Center, 2010). Regardless of which trajectory will actually occur, the modeling efforts show that the future status of the Pacific walrus is linked to sea ice, which already is declining substantially, and more rapidly than previously projected.

Effects of Global Climate Change on Pacific Walrus Prey Species

The shallow, ice-covered waters of the Bering and Chukchi Seas provide habitat that supports some of the highest benthic biomass in the world (Grebmeier et al. 2006a, p. 1461; Ray et al. 2006, p. 404). Sea-ice algae, pelagic (open ocean) primary productivity, and the benthos (organisms that live on or in the sea floor) are tightly linked through the sedimentation of organic particles (Grebmeier et al. 2006b, p. 339). Sea-ice algae provide a highly concentrated and high-quality food source for plankton food webs in the spring, which translates to high-quality food for the benthos such as clams (Grebmeier et al. 2006b, p. 339; McMahon et al. 2006, pp. 2-11; Gradinger 2009, p. 1211). Because zooplankton, which also feed on the algae, have correspondingly low populations at this time in the spring, much of the primary productivity of algae falls to the sea floor, where it is available to the benthic invertebrates (Grebmeier et al. 2006b, p. 339)

h-quality food for the benthos such as clams (Grebmeier et al. 2006b, p. 339; McMahon et al. 2006, pp. 2-11; Gradinger 2009, p. 1211). Because zooplankton, which also feed on the algae, have correspondingly low populations at this time in the spring, much of the primary productivity of algae falls to the sea floor, where it is available to the benthic invertebrates (Grebmeier et al. 2006b, p. 339).

Spatial distribution and abundance in biomass in benthic habitat across the Bering and Chukchi Seas is influenced by a variety of ecological, oceanographic, and geomorphic features. In the subarctic region of the Bering Sea (from the Bering Strait south to latitude 50 degrees), benthic organisms are preyed upon by demersal fish (living near the bottom of the water column) and epifaunal invertebrates (those organisms living on top of the sea floor rather than in it), whose distribution is limited to the north by cold water (less th an 0 °C (32 °F)) et al. 2006b, pp. 1461-1463).

Within the Arctic region of the Bering Sea, marginal sea-ice zones and areas of polynyas appear to be “hot spots” of high benthic diversity and productivity (Grebmeier and Cooper 1995, p. 4439). Benthic biomass is particularly high in the northern Bering Sea, the southern Chukchi Sea, and the Gulf of Anadyr. However, the high diversity and productivity of the benthic communities is not seen in the Southern Beaufort Sea shelf and areas of the eastern Chukchi Sea, which are influenced by the nutrient-poor Alaska coastal current (Fay et al. 1977, p. 12; Grebmeier et al. 1989, p. 261; Feder et al. 1994, p. 176; Smith et al. 1995, p. 243; Grebmeier et al. 2006b, p. 346; Bluhm and Gradinger 2008, p. 2).

Ocean Warming

For the last several decades, surface air temperatures throughout the Arctic, over both land and water, have warmed at a rate that exceeds the global average, and they are projected to continue on that path (Comiso and Parkinson 2004, pp. 38-39; Christensen et al. 2007, p. 904; Lawrence et al. 2008, p. 1; Serreze et al. 2009, pp

al. 2006b, p. 346; Bluhm and Gradinger 2008, p. 2).

Ocean Warming

For the last several decades, surface air temperatures throughout the Arctic, over both land and water, have warmed at a rate that exceeds the global average, and they are projected to continue on that path (Comiso and Parkinson 2004, pp. 38-39; Christensen et al. 2007, p. 904; Lawrence et al. 2008, p. 1; Serreze et al. 2009, pp. 11-12). In addition, the subsurface and surface waters of the Arctic Ocean and surrounding seas, including the Bering and Chukchi Seas have warmed (Steele and Boyd 1998, p. 10419; Zhang et al. 1998, p. 1745; Overland and Stabeno 2004, p. 309; Stabeno et al. 2007, pp. 2607-2608; Steele et al. 2008, p. 1; Mueter et al. 2009, p. 96). There are several mechanisms working in concert to cause these increases in ocean temperature, including: Warmer air temperatures (Comiso and Parkinson 2004, pp. 38-39; Overland and Stabeno 2004, p. 310), an increase in the heat carried by currents entering the Arctic from both the Atlantic (Drinkwater et al., p. 25; Zhang et al. 1998, p. 1745) and Pacific Oceans (Stabeno et al. 2007, p. 2599; Woodgate et al. 2010, p. 1-5), and a shorter ice season, which decreases the albedo (reflective property) of ice and snow (Comiso and Parkinson 2004, p. 43; Moline et al. 2008, p. 271; Markus et al. 2009, p. 13). Due to their biological characteristics which include tolerance of considerable variations in temperature, direct effects to walrus are not anticipated with warmer ocean temperatures. Nevertheless, changes in the thermal dynamics of ocean conditions may affect walrus indirectly through impacts to their prey base. Changes to density, abundance, distribution, food quality, and species of benthic invertebrates may occur primarily through changes in habitat related to sea ice.

Walruses are the top predator of a relatively simple food web in which the primary constituents are bacteria, sea-ice algae, phytoplankton (tiny floating plants), and benthic invertebrates (Horner 1976, p

gh impacts to their prey base. Changes to density, abundance, distribution, food quality, and species of benthic invertebrates may occur primarily through changes in habitat related to sea ice.

Walruses are the top predator of a relatively simple food web in which the primary constituents are bacteria, sea-ice algae, phytoplankton (tiny floating plants), and benthic invertebrates (Horner 1976, p. 179; Lowry and Frost 1981, p. 820; Grebmeier and Dunton 2000, p. 65; Dunton et al. 2006, p. 370; Aydin and Mueter 2007, p. 2507). Sea ice is important to the Arctic food webs because: (1) It is a substrate for ice algae (Horner 1976, pp. 168-171; Kern and Carey Jr. 1983, p. 161; Grainger et al. 1985, pp. 25-27; Melnikov 2000, pp. 79-81; Gradinger 2009, p. 1201); (2) it influences nutrient supply and phytoplankton bloom dynamics (Lovvorn et al. 2005, p. 136); and (3) it determines the extent of the cold-water pool on the southern Bering shelf (Aydin and Mueter 2007, p. 2503; Coyle et al. 2007, p. 2900; Stabeno et al. 2007, p. 2615; Mueter and Litzow 2008, p. 309).

In the spring, ice algae form up to a 1-cm- (0.4-in-) thick layer on the underside of the ice, but are also found at the ice surface and throughout the ice matrix (Horner 1976, pp. 168-171; Cota and Horne 1989, p. 111; Gradinger et al. 2005, p. 176; Gradinger 2009, p. 1207). Ice algae can be released into the water through water turbulence below the ice, through brine drainage through the ice, or when the algal mats are sloughed as the ice melts (Cota and Horne 1989, p. 117; Renaud et al. 2007, p. 7). As noted above, sea-ice algae provide a highly concentrated food source for the benthos and the plankton (organisms that float or drift in the water) food web that is initiated once the ice melts (Grebmeier et al. 2006b, p.339; McMahon et al. 2006, pp. 1-2; Renaud et al. 2007, pp. 8-9; Gradinger 2009, p. 1211). Areas of high primary productivity support areas of high invertebrate mass, which is food for walruses (Grebmeier and McRoy 1989, p

ly concentrated food source for the benthos and the plankton (organisms that float or drift in the water) food web that is initiated once the ice melts (Grebmeier et al. 2006b, p.339; McMahon et al. 2006, pp. 1-2; Renaud et al. 2007, pp. 8-9; Gradinger 2009, p. 1211). Areas of high primary productivity support areas of high invertebrate mass, which is food for walruses (Grebmeier and McRoy 1989, p. 87; Grebmeier et al. 2006b, p. 332; Bluhm and Gradinger 2008, p. S87).

Spring ice melt plays an important role in the timing, amount, and fate of primary production over the Bering Sea shelf, with late melting (as occurs now) leading to greater delivery of food from primary production to the benthos and earlier melting (as is projected to occur in the future) contributing food primarily to the pelagic system (Aydin and Mueter 2007, p. 2505; Coyle et al. 2007, p. 2901). When ice is present from late March to May (as occurs now), cold surface temperatures, thinning ice, and low-salinity melt water suppress wind mixing, and cause the water column to stratify, creating conditions that promote a phytoplankton bloom. The burst of phytoplankton, seeded in part by ice algae, persists until ocean nutrients are drawn down. Because it is early in the season and water temperatures are cold, zooplankton populations are still low. Consequently, the pulse of phytoplankton production is not consumed by zooplankton, but instead sinks to the sea floor, where it provides abundant food for the benthos (Coyle and Cooney 1988, p. 177; Coyle and Pinchuk 2002, p. 177; Hunt and Stabeno 2002, p. 11; Lovvorn et al. 2005, p. 136; Renaud et al. 2007, p. 9). Blooms form a 20- to 50-km- (12-31 mi-) wide belt off the ice edge and progress north as the ice melts, creating a zone of high productivity. In colder years in the Bering Sea, when the ice extends to the shelf edge, there is greater nutrient resupply through shelf-edge eddies and tidal mixing, creating a longer spring bloom (Tynan and DeMaster 1997, pp. 314-315)

p. 136; Renaud et al. 2007, p. 9). Blooms form a 20- to 50-km- (12-31 mi-) wide belt off the ice edge and progress north as the ice melts, creating a zone of high productivity. In colder years in the Bering Sea, when the ice extends to the shelf edge, there is greater nutrient resupply through shelf-edge eddies and tidal mixing, creating a longer spring bloom (Tynan and DeMaster 1997, pp. 314-315).

The blooms that occur near the ice edge make up approximately 50 to 65 percent of the total primary production in Arctic waters (Coyle and Pinchuk 2002, p. 188; Bluhm and Gradinger 2008, p. S84). High benthic abundance and biomass correspond to areas with high deposition of phytodetritus (dead algae) (Grebmeier et al. 1989, pp. 253-254; Grebmeier and McRoy 1989, p. 79; Tynan and DeMaster 1997, p. 315). Regions with the highest masses of benthic invertebrates occur in the northern Bering Sea southwest of St. Lawrence Island, Alaska; in the central Gulf of Anadyr, Russia, north and south of the Bering Strait; at a few offshore sites in the East Siberian Sea; and in the northeast sector of the Chukchi Sea (Grebmeier and Dunton 2000, p. 61; Dunton et al. 2005, pp. 3468, 3472; Carmack et al. 2006, p. 165; Grebmeier et al. 2006b, pp. 346-351; Aydin and Mueter 2007, pp. 2505-2506; Bluhm and Gradinger 2008, p. S86). As noted above, the biomass of benthic invertebrates is much less in the eastern Chukchi Sea, which is under the influence of the nutrient-poor Alaska Coastal Current (Dunton et al. 2006, p. 369).

When the ice melts early (before mid-March, as projected for the future), conditions that promote the phytoplankton bloom do not occur until late May or June (Stabeno et al. 2007, p. 2612). The difference in timing is important, because when the bloom Theragra chalcogramma ) and other planktivorous fishes (Hunt and Stabeno 2002, pp. 14-15). Consequently, when zooplankton populations are high, instead of the primary production being transmitted to the benthos, it becomes tied up in pelagic food webs

hytoplankton bloom do not occur until late May or June (Stabeno et al. 2007, p. 2612). The difference in timing is important, because when the bloom Theragra chalcogramma ) and other planktivorous fishes (Hunt and Stabeno 2002, pp. 14-15). Consequently, when zooplankton populations are high, instead of the primary production being transmitted to the benthos, it becomes tied up in pelagic food webs. While this may be beneficial for fish-eating mammals, it reduces the amount of food delivered to the benthos and, thus, may reduce the amount of prey available to walrus (Tynan and DeMaster 1997, p.316; Carmack et al. 2006, p. 169; Grebmeier et al. 2006a, p. 1462). Most models project that sea-ice melt in the Bering Sea will occur increasingly early in the future, and will be 1 month earlier by the end of the century (Douglas 2010, p. 12). This is consistent with recent trends over the past two decades, and particularly in the past few years. Based on our current understanding of food web dynamics in the Bering Sea, this shift in timing would favor a shift to pelagic food webs over benthic production, consequently reducing the amount of prey available to walrus.

The importance of ice algae is not only in its role in seeding the spring phytoplankton bloom, but also in its nutritional value. As food supply to the benthos is highly seasonal, synchrony of reproduction with algal inputs insures adequate high-quality food for developing larvae or juveniles of benthic organisms (Renaud et al. 2007, p. 9). Ice algae have high concentrations of essential fatty acids, some of which cannot be synthesized by benthic invertebrates and, therefore, must be ingested in their diet (Arrigo and Thomas 2004, p. 477; Klein Breteler et al. 2005, pp. 125-126; McMahon et al. 2006, pp. 2, 5). Fatty acids in marine fauna play an integral role in physiological processes, including reproduction (Klein Breteler et al. 2005, p. 126)

ve high concentrations of essential fatty acids, some of which cannot be synthesized by benthic invertebrates and, therefore, must be ingested in their diet (Arrigo and Thomas 2004, p. 477; Klein Breteler et al. 2005, pp. 125-126; McMahon et al. 2006, pp. 2, 5). Fatty acids in marine fauna play an integral role in physiological processes, including reproduction (Klein Breteler et al. 2005, p. 126). Because ice algae are a much better source of essential fatty acids than phytoplankton, a loss in sea ice could change the quality of food supplied to areas that currently support high levels of benthic biomass. These changes may affect the success of invertebrate reproduction and recruitment, which, in turn, may affect the quantity and quality of food available to walrus (Witbaard et al. 2003, p. 81; McMahon et al. 2006, pp. 10-12). By the end of the century, the March (winter maximum) extent of sea ice is projected to be approximately half of contemporary conditions (Douglas 2010, p. 8). We expect ice algae will persist where ice is present; however, because of the reduced ice extent, current areas of high benthic productivity may be reduced or shift northward.

The eastern and western Bering Sea shelves are fueled by nutrient-rich water supplied from the deep water of the Bering Sea (Sambrotto et al. 1984, pp. 1148-1149; Springer et al. 1996, p. 205). Concentrations of nitrate, phosphate, and silicate are among the highest recorded in the world's oceans and contribute to the high benthic productivity (Sambrotto et al. 1984, p. 1148; Grebmeier et al. 2006a, p. 1461; Aydin and Mueter 2007, p. 2504). High productivity on the northern Bering-Chukchi shelf is supported by the delivery of nutrient-rich water via the Anadyr Current that flows along the western edge of the Bering Sea and through the Bering Strait (Springer et al. 1996, p. 206; Aydin and Mueter 2007, p. 2504)

high benthic productivity (Sambrotto et al. 1984, p. 1148; Grebmeier et al. 2006a, p. 1461; Aydin and Mueter 2007, p. 2504). High productivity on the northern Bering-Chukchi shelf is supported by the delivery of nutrient-rich water via the Anadyr Current that flows along the western edge of the Bering Sea and through the Bering Strait (Springer et al. 1996, p. 206; Aydin and Mueter 2007, p. 2504). Thus, the movement of highly productive water onto the northern Bering Sea shelf supports persistent hot spots of high benthic productivity, which in turn support large populations of benthic-feeding birds, walrus, and gray whales (Aydin and Mueter 2007, p. 2506). This contrasts with the southern subarctic region of the Bering Sea, which is south of the current range of the Pacific walrus, where the benthic mass is largely consumed by upper tropic-level demersal fish and epifaunal invertebrates whose northern distribution is limited by a pool of cold, near-freezing water in the northern region of the Bering Sea.

Benthic productivity on the northern Bering Sea shelf has decreased over the last two decades, coincident with a reduction of northward flow of the Anadyr current through the Bering Strait (Grebmeier et al. 2006a, p. 1462). Because of recent warming trends, the northern Bering Sea shelf may be undergoing a transition from an Arctic to a more subarctic ecosystem with a reduction in benthic prey populations and an increase in fish populations (Overland and Stabeno 2004, p. 310; Grebmeier et al. 2006a, pp. 1462-1463). The Bering Sea is a transition area between Arctic and subarctic ecosystems, with the boundary between the two loosely concurrent with the extent of the winter sea-ice cover (Overland and Stabeno 2004, p. 309). In the eastern Bering Sea, reductions in sea ice have been responsible for shrinking a large subsurface pool of cold water with water temperatures less than 2 °C (3.6 °F) (Stabeno et al. 2007, p. 2605; Mueter and Litzow 2008, p. 313)

Arctic and subarctic ecosystems, with the boundary between the two loosely concurrent with the extent of the winter sea-ice cover (Overland and Stabeno 2004, p. 309). In the eastern Bering Sea, reductions in sea ice have been responsible for shrinking a large subsurface pool of cold water with water temperatures less than 2 °C (3.6 °F) (Stabeno et al. 2007, p. 2605; Mueter and Litzow 2008, p. 313). The southern edge of the cold pool, which defines the boundary region between the Arctic and subarctic communities, has retreated approximately 230 km (143 mi) north since the early 1980s (Mueter and Litzow 2008, p. 316).

The northward expansion of warmer water has resulted in an increase in pelagic species as subarctic fauna have colonized newly favorable habitats (Overland and Stabeno 2004, p. 309; Mueter and Litzow 2008, pp. 316-317). Walleye pollock, a species common in the subarctic, which avoid temperatures less than 2° C (3.6 °F), have now moved northward into the former Arctic zone. Arctic cod ( Boreogadus saida ), which prefer cold temperatures, have also moved north to remain in colder temperatures (Stabeno et al. 2007, p. 2605). Because of the redistribution of these species, benthic fauna will be facing a new set of predators (Coyle et al. 2007, pp. 2901-2902). The evidence suggests that warming on the Bering Sea shelf could alter patterns of energy flow and food web relationships in the benthic invertebrate community, leading to overall reductions in biomass of benthic invertebrates (Coyle et al. 2007, p. 2902).

Continued changes in the extent, thickness, and timing of the melt of sea ice are expected to create shifts in production and species distributions (Overland and Stabeno 2004, p. 316). Because some residents of the benthos are very long lived, it may take many years of monitoring to observe change (Coyle et al. 2007, p. 2902)

ions in biomass of benthic invertebrates (Coyle et al. 2007, p. 2902).

Continued changes in the extent, thickness, and timing of the melt of sea ice are expected to create shifts in production and speci

This text is long and has been trimmed here. Open the source document for the complete record.