Seascape as an organizing principle for evaluating walrus and seal sea-ice habitat in Beringia



[1] The term “seascape”, as used here, relates the natural history of ice-dependent pinnipeds to their sea-ice environments at different spatial scales, following concepts of landscape ecology. Habitats are characterized by heterogeneous but repeatable structures of sea ice. As an example, multiple mesoscale (3–50 km) seascapes present conditions for different ecological preferences of five Beringian ice-dependent pinnipeds, as observed during 2006–2009 winter-spring icebreaker cruises. Seascape partitioning concepts are important for understanding and projecting species' responses to change under climate-change scenarios.

1. Introduction

[2] Five Beringian pinnipeds share the need for sea ice as habitat for reproduction, nursing, molt, and rest: Pacific walrus (Odobenus rosmarus divergens), ribbon seal (Histriophoca fasciata), spotted seal (Phoca largha), bearded seal (Erignathus barbatus), and ringed seal (Phoca hispida). Naturalists have long agreed that these species exhibit floe preferences and partition habitats during their winter-spring reproduction periods according to the character of the pack [e.g., Burns, 1970; Burns, 1981a; Fay, 1974; Braham et al., 1984; Lentfer, 1988; Ray and Hufford, 1989]. Substantial new observations, as reported on in this paper, have been obtained from four winter-spring cruises of the icebreaker USCGC Healy 2006–2009, further to describe the association of pinniped natural history with sea-ice. Concurrently, natural scaling properties of sea-ice have become better understood, and closely match the interacting scales that determine pinniped sea-ice habitat [e.g., McNutt and Overland, 2003].

2. Physical Processes That Define the Seascape

[3] Sea-ice formation in the Bering Sea may be conceived as a conveyor belt [Pease, 1980; Niebauer et al., 1999]. Ice is first formed along northern Bering Sea coasts in late October. Surface atmospheric circulation patterns dominated by cold northeasterly winds normally advect sea ice southward over the shelf where it melts in the warmer ocean to the south [Overland, 1981]. By mid- to late-March, sea ice is most extensive and normally covers about 75% of the continental shelf, the remainder of the area being dominated by large polynyas bordering land masses. By April, northeasterly winds decrease and temperatures begin to rise causing the ice to break up and melt. Ocean surface currents and solar insolation then play larger roles, and along with winds, melt and force sea ice slowly northward. Seasonal sea ice extends over much of the shelf from February through May. Only remnant ice remains in the Bering Sea by the end of June.

[4] Scaling is critical for understanding seascape formation and dynamics. Regional, mesoscale, and local scales reflect specific sea-ice properties [McNutt and Overland, 2003]. At local scales, pinniped/sea-ice associations are responsive to individual ice floes, characterized by such adjectives as “closed” or “open” pack, or “new”, “frazil”, “ridged”, or “heavy” ice. Most neglected has been the mesoscale scale (∼3-50 km), which we argue is most relevant to pinniped life histories. At and above the lower end of the mesoscale, the emergent scale, sea ice behaves more as a plastic continuum than as individual floes, and is governed primarily by fracture mechanics. On regional scales (>50 km), sea ice responds to external atmosphere and ocean forcing on weekly and longer time scales and is characterized by extent, concentrations, roughness, and mean thickness. An important physical transition occurs at the regional scale, the coherent scale, where sea-ice dynamics best match large-scale wind and ocean forcing.

[5] Formation of different mesoscale seascape types (Table 1 and Figure 1), lying between emergent and coherent transitions, is explained by interactions among winds, currents, waves, and tides within the constrained Bering Sea geomorphology. Ray and Hufford [1989] innovated the names used below for these mesoscale types, as the sea-ice nomenclature of that time did not seem suited for habitat description. The same situation remains today. For example, the WMO “egg code” ( has been developed for transportation and navigation and addresses stages of sea-ice development and floes by size, but not shape; aggregated seascape features specific to habitat are not included. Broken pack (Figure 1) occurs in the central Bering Sea where sea ice is relatively unconstrained by basin morphology, such that leads and polynyas are more frequent than elsewhere, winds and currents can freely disperse floes, and thick continuous pack is broken when oceanic swells penetrate far into the pack. Loose pack [2] is particularly affected by oceanic conditions at the sea-ice margin, and pancake ice is frequent. The northern Bering Sea is colder than the southern Bering and the western Bering is colder than the eastern Bering, partially explaining the formation of thick pack-ice-with-leads [3] to the northwest and into Gulf of Anadyr, characterized by parallel leads. Rounded pack [4] forms when northward-moving currents confront southward-moving ice, a condition unique to the eastern Bering Sea. Continuous pack [5] occurs near the Bering Strait region where the narrow Strait causes continuous stresses. Large polynyas [6] occur both within the pack and adjacent to land masses, according to wind conditions. Short-term spatial shifts in seascape types are to be expected, up to 100 km day−1. Also, two to three storms per month during winter and decreasing in number in spring tend to propagate eastward and northward across the Bering Sea, mixing the upper ocean, advecting sea ice floes, and causing seascape deformation and considerable sea-ice variability in extent and cover among and between years. Increases in sea-ice cover and extent have occurred recently in early winter in the Bering Sea. However, these increases have been accompanied by thin ice subject to frequent freeze-thaw cycles. Also, increases in southerly extent have come at the expense of overall cover, particularly during April and May; this situation arises when northerly winds create large polynyas that dominate much of the northerly shelf. Despite this variability, the mesoscale habitats considered here are spatially and temporally repetitive and predictable, as first noted by Burns [1981].

Figure 1.

Six (numbered) seascapes are defined in the text. Place names are: GA = Gulf of Anadyr; CN = Cape Navarin; SL = St. Lawrence Island; SM = St. Matthew Island; NU = Nunivak Island. Top: Pacific walrus on floes within broken pack southwest of St. Lawrence Island, 20 April 2007; area enclosed by dashed line was determined by helicopter flights and interpretation of MODIS imagery. On both the satellite image and the photograph, the sharp-edged character of individual floes and the intersecting leads separating floes are discernable. The thick, concentrated, rounded pack, extending from the Bering Strait region to west of Nunivak, is also obvious. Photograph © G. Carleton Ray. Bottom: Adult ringed seal on flat, continuous ice southwest of St. Lawrence Island, 20 April 2007. The wet area at the seal's location surrounds a hole made by the seal. MODIS image shows flat ice sequestered within an area of broken pack, as often occurs where the pack remains relatively coherent. Photograph courtesy of Elizabeth Labunski.

Table 1. Attributes of Mid-winter Seascape Types of the Bering Sea Ice Types From Ray and Hufford [1989]
Seascape TypeSpecies Natural HistoriesSeascape Distribution
Broken PackDominant species: walrus, bearded seal. Attributes: benthic feeders; require dispersed floes for rest over relatively shallow water; open water, or thin ice for access to benthic food supply.Central Bering Sea; Small scattered floes to almost continuous pack, broken into very large, angular floes; ice dispersed such that open water or thin ice is continuously available.
Loose packDominant species: ribbon seal mostly westward; spotted seal mostly eastward. Attributes: small seals that cannot break ice and require easy access to open water; seek ice where polar bears do not generally occur.Southernmost extent of the pack; small to moderate-sized floes dispersed from interactions with wind, waves, and oceanic forces; floes often rounded from collisions.
Pack ice with leadsDominant species: walrus. Attributes: as for broken pack.Gulf of Anadyr and Koryak coast; small to large congealed floes; leads perpendicular to wind direction.
Rounded packNo or very few pinnipeds due to lack of open water.Eastern Bering Sea; thick, ridged, often congealed floes with rounded edges.
Continuous packDominant species: Ringed seal. Attributes: make breathing holes in shorefast or continuous ice up to 2m thick; may also occur widely in the pack.Bering Strait region; ice semi-continuous and fractured in all directions.
PolynyasFew pinnipeds on isolated floes, remnant ice, or swimming.Occur south or north of land masses when winds force sea ice away from coasts.

3. Use of Seascape Types by Beringian Pinnipeds

[6] Figures 1234 illustrate representative sea-ice associations of each of the five species, drawn from in situ observations of thousands of walruses, hundreds of ribbon and spotted seals, and tens of bearded and ringed seals during 2006–2009, the vast majority of which were on sea ice consistent with the different seascape types described above. All photographs and accompanying MODIS imagery were separated in time by a maximum of 8 hours. The approximate extents of each seascape type indicated on the imagery recognize that species are patchily distributed throughout these potential habitats. Table 1 reviews associations among seascapes and species natural histories.

Figure 2.

Ribbon seal on a heavy, snow-covered floe of remnant loose pack southwest of St. Lawrence Island, 3 June 2006. The dashed line includes the area of remnant ice on which this individual, with several dozen others, was observed. Image-recognition, tracking technology indicated that this remnant ice drifted from near the mouth of the Anadyr River, 9 May, south to where the photo was taken by 3 June, and disintegrated near St. Lawrence Island, 11 June (G. L. Hufford, personal communication, 2009). Conditions during 2006 were such that the pack was not extensive during April–May, and loose pack was mixed with pack-ice-with-leads. Photograph courtesy of G. Carleton Ray.

Figure 3.

Spotted seal “family” (male, female, and pup) in loose pack near the outer sea-ice front between St. Matthew Island and Cape Navarin, 21 April 2007. Loose pack is clearly distinguished from broken pack to the north. In the photograph, floes show evidence of collision. The area of loose pack enclosed by the dashed line was sampled extensively in situ and is readily discerned in MODIS imagery, although in this case clouds obscure the eastern portion. Photograph courtesy of G. Carleton Ray.

Figure 4.

Newborn bearded seal pup near the edge of a heavy angular floe of broken pack southeast of St. Matthew Island, 8 April 2009. The approximate area of semi-continuous broken back enclosed by a dashed line was sampled extensively in situ and determined from MODIS imagery. The photograph shows evidence of some collisions among floes, but the floes remain angular and intersected with leads. MODIS imagery shows the parallel leads characteristic of pack-ice-with-leads in the Gulf of Anadyr; the dense, rounded floes of rounded pack to the east are also readily observed. Photograph courtesy of Elizabeth Labunski.

[7] The best-known Beringian pinniped is the gregarious Pacific walrus [Fay, 1982]. Herds may number in the thousands and cover areas of >100 km2 [Ray et al., 2006]. Subpopulations occur in broken pack southwest of St. Lawrence Island to the Gulf of Anadyr and in the northern, outer reaches of Bristol Bay. Walrus prefer thick, often ridged, and moderate-size floes separated by leads and polynyas (Figure 1); new ice among continuous floes is tolerated, as walruses are able to break ice up to 20 cm thick. During February-March, walrus gather in large aggregations in “arenas” within this seascape type to court and mate [Fay et al., 1984]. Calving occurs in May during northward advection of this ice into the Bering Strait region. By June, most females, juveniles, and young occupy marginal Chukchi Sea ice; contrarily, males move to land haulout sites along Bering and Chukchi Sea shores.

[8] Two species, too small to break through sea ice, occupy loose pack from St. Matthew and westward into the Gulf of Anadyr and south along the Koryak coast. The ribbon seal is pelagic for most of the year, but in winter-spring occupies floes of the inner loose pack that vary in thickness, concentration, shape, and size; ridged, snow-covered floes seem to be preferred [Burns, 1981b; Fedoseev, 2002]. In spring, ribbon seals frequently occur on heavy floes of remnant ice where the seascape shows evidence of wave action and collisions (Figure 2). Pups are born mostly in April and rarely enter the water until nursing and molting are complete by early July. Spotted seals often occur with ribbon seals, but tend to occupy floes of the outer portion of loose pack nearer the ice edge where floes are thinner and more dispersed (Figure 3). They remain in shelf waters and resort to coastal lands after the retreat of seasonal ice, hence are most abundant eastward. Pups are born with thick, white lanugo coats in late March to mid-April in lairs among pressure ridges, and rarely enter the water until the white coat is shed 4–6 weeks after birth.

[9] The last two species are less strictly associated with seascape type. The largest of Beringian seals, the bearded seal, is a benthic feeder that occurs widely in the pack during winter-spring, being large enough to break new or very thin ice [Burns, 1981c] and small enough to occupy less substantial floes than walruses. It is most frequent in a broken pack (Figure 4) and avoids regions of continuous, thick, shorefast ice. In winter-spring, males “sing” in reproductive arenas where floes surround leads and small polynyas [Ray et al., 1969]. On the contrary, the smallest of Beringian seals, the ringed seal, occurs across the arctic and is common in a heavy continuous sea ice seascape, as it is uniquely able to maintain breathing holes in ice of up to 2 m thick (Figure 1). It also creates subnivean birth lairs within pressure ridges that provide optimum protection from polar bears [Kelly, 1988]. Its greatest winter density is in heavy, ridged shorefast ice; its greatest numbers appear to be in stable, coherent pack, where it takes advantage of large floes of flat ice with cracks and pressure ridges where they create breathing holes.

4. Discussion and Conclusions

[10] Landscape ecology, as initially proposed by Troll [1939] and later modified by Forman and Godron [1986] draws attention to the physical-biological relationships that govern the different spatial units of a region. As such, this concept ties life histories of Beringian pinnipeds directly to different sea-ice properties, in terms of how different species use specific habitats. However, as important as the seascape may be in indicating where species are likely to occur, it is also useful for determining where they are not. Notably, pinnipeds are rare or not likely to occur in rounded pack, and have been absent in large expanses of young ice between St. Lawrence and St. Matthew during all years of our study. In this sense, some seascape types represent critical habitats, whereas other types do not, information that can be useful for making population estimates and for estimating sensitivities to change.

[11] The importance of the landscape/seascape concept lies in its connections to ecology and conservation. The application of the seascape concept to other regions than the Bering Sea will depend on particular conditions and species associations. But in all cases, lack of understanding of species' habitat preferences based on multiscaled, spatially explicit, environmental relationships can result in serious errors [Wu, 2003]. For example, landscape population models suggest that loss and structural change of habitat increases the proportion of time that the population spends in the portion of the habitat where reproduction may not be possible and where mortality is higher, thereby leading to a downward spiral to extinction [Fahrig, 2007]. For Beringian pinnipeds, loss and structural changes of sea-ice habitat, such as seems now to be occurring, theoretically have the potential to result in a similar downward spiral. Consequences for broader ecosystem function seem inevitable, as ice-dependent pinnipeds are significant consumers and have considerable potential to influence biotic community structures and transfers of energy throughout food webs [e.g., Ray et al., 2006].

[12] Finally, we conclude that similarities between the hierarchical approach of McNutt and Overland [2003] and the scaled habitat associations of pinniped habitat are not accidental, but are functional and evolutionary. Textural sea-ice analysis, using higher-resolution images, together with extended pinniped observations, will be needed in the future to determine more fully the nature of this relationship. At present, we hypothesize that high inter- and intra-annual variability of sea-ice conditions at multiple scales, and a trend towards reduction and structural changes of sea-ice, play reinforcing roles in shifting marine mammal habitats. Consequently, we predict that the variable conditions of the past decade [Walsh, 2008] will become more pronounced in the future, with implications for pinnipeds, other ice-dependent biota, and the ecology of polar regions.


[13] We thank the Captains, Chief Scientists, crews, and technical personnel for assistance recording pinnipeds from the icebreaker USGC Healy during 2006–2009 winter–spring cruises. Special thanks are due to Kathy Kuletz and Elizabeth Labunski of the Fish and Wildlife Service, Alaska, for marine mammal observations and photographs, and also to Michael Cameron and colleagues of NOAA's National Marine Mammal Laboratory and Chadwick Jay and Anthony Fischback of the U.S. Geological Survey for their observations. Thanks also to anonymous reviewers for their constructive suggestions. Steve Roberts of the National Center for Atmospheric Research, Boulder, Colorado, provided ship positions and access to satellite imagery aboard the Healy. Robert L. Smith, Charlottesville, Virginia, formatted the illustrations. NOAA's National Weather Service–Alaska Region, the U.S. Marine Mammal Commission, and the Global Biodiversity Fund of the University of Virginia provided support for this paper. PMEL contribution no. 3614.