The Frozen Skin of the Southern Ocean

Stephen Nicol & Ian Allison. American Scientist. Volume 85, Issue 5. Sep/Oct 1997.

Antarctica has earned the nickname “the Frozen Continent.” Freshwater ice derived from millennia of accumulated snowfall blankets the continent in a layer averaging more than two kilometers in thickness. The South Pole stands 2,800 meters above sea level, but 2,722 of those meters are made of ice. Furthermore, about 50 percent of the coastline of the continent consists of floating extensions of this ice sheet-ice shelves that breed most of the icebergs calving from the continent. But Antarctica is actually surrounded by a vast ice band derived mainly from a source other than snow: direct freezing of ocean water.

The Antarctic pack-ice zone is one of the last places on the planet to be explored and scientifically investigated. Since the discovery of the seventh continent, the surrounding ice has been seen mainly as a hindrance to navigation-one that more than doubles the effective size of Antarctica from summer to winter and that varies in extent not only with the seasons but also from year to year. Only lately has this zone begun to be accorded the importance it deserves. Advances in technology and the development of icebreaking vessels for scientific work have allowed a reappraisal of the pack-ice zone, and the results are overturning many long-established beliefs about the physical and biological systems of the Antarctic.

Climatologists now consider the sea ice to be crucial in the regulation of the planet’s climate, and biologists believe it to be the driving force behind the biological productivity that once fed vast populations of whales each summer in the Southern Ocean. Very little plant biomass exists on Antarctica, and the lichens, mosses and algae that do grow there do not contribute to the food chain that supports the abundant animal life of the region. All vertebrate species in the Antarctic-whales, seals, birds and fish, whether resident in the region or migratory-depend on the productivity of the surrounding ocean, which for much of the year is covered in a layer of ice.

The Southern Ocean and the Antarctic pack ice, along with their associated atmosphere and biota, are part of a complex interactive system. The pack ice influences characteristics of the ocean, the atmosphere and the biosphere, yet the distribution and characteristics of the ice are themselves forced by atmospheric and oceanic variables-such as temperature, wind and salinity-and also may be influenced by organisms living in the ice. Several of the resulting feedback processes have the potential to play an important role in global climate change.

Extent and Seasonality

The sea ice on the Southern Ocean undergoes an annual change from a minimum area of about 4 million square kilometers in February to a maximum of 19 million square kilometers in September-one of the greatest seasonal changes of surface characteristics on earth. Sea ice stretches as much as 2,200 kilometers from the coast into the Weddell Sea during winter. But between the longitudes 120E and 135E, the winter ice edge may extend only 300 kilometers offshore. The total ice extent decays rapidly through November and December, and by February the only remaining large areas of perennial sea ice are in the eastern Ross Sea, the Bellingshausen and Amundsen Seas, and the western Weddell Sea. The region of the Southern Ocean that annually fluctuates between open water in summer and a winter ice-cover is known as the seasonal sea-ice zone.

Sea ice is far from monolithic. Only in winter, and only along some parts of the coastline, is a narrow band of sea ice attached to the coast as a solid, immobile and continuous sheet. This landfast ice, or fast ice, may extend some tens of kilometers from the coastline, but most of the Antarctic sea ice consists of pack ice, broken sheets moved by the wind and ocean currents. Pack ice constantly changes configuration as the individual pieces of ice, or floes, are bonded together by refreezing, broken by ocean swell and wind action, and deformed by collisions. There are usually some areas of open water within the pack in leads, or linear openings, and satellite-derived estimates indicate that during winter open water occupies as much as 20 percent of the total area. The outer 150-to-200-kilometer region of the pack, known as the marginal ice zone, lies close enough to the open-ocean boundary to be significantly affected by ocean waves and swell; the ice concentration (the percentage of ocean covered by ice) in this zone is lower than within the interior pack, and the floes have much greater freedom of movement.

The outer boundary of the Antarctic sea ice is unconstrained by land and subject to strong westerly circumpolar winds. These winds produce an average northerly stress component on the ice and a net divergence of ice within the pack. Thus the location of the outer edge of the ice is determined not only by in situ freezing but also by the advection of ice formed in other regions. The position of the ice edge can vary by tens of kilometers per day with the passage of weather systems and changing wind directions. Further within the sea-ice zone and south of the meandering oceanic boundary between the Southern Ocean circumpolar current and more local currents, known as the Antarctic Divergence, the ice is driven westward by the wind. But even here large-scale oceanic gyre systems in the Weddell Sea, the Ross Sea, Prydz Bay and other areas transport the ice from high latitudes, where it is formed, into the divergent westward flow to the north.

Coastal polynyas, areas of open water or thin ice other than leads, persist or recur within the winter pack and are common features of the Antarctic coastline. Typical coastal polynyas range in size from hundreds to tens of thousands of square kilometers. Polynyas form where strong offshore winds from the continental ice sheet remove newly formed ice; they are usually found in the lee of northsouth protruding headlands, glacier tongues, ice shelves or grounded icebergs, which act as barriers to ice import from upwind. Although they occupy only a small fraction of the area of the overall sea-ice zone, polynyas are regionally important as areas of high heat flux between the ocean and atmosphere, as “ice factories” with production rates as much as 10 times higher than average, as areas of high salt flux into the ocean associated with enhanced ice growth and as possible oases of biological activity.

The sea ice of the Southern Ocean, and its associated snow cover, play a number of important roles in the ocean-atmosphere climate system, in ocean circulation and in structuring the marine ecosystem. Sea ice greatly reduces the exchange of heat, gas and momentum between the ocean and the atmosphere. Compared to open water, snow-covered ice is highly reflective to short-wave radiation and significantly reduces both the absorption of solar radiation at the surface and the amount of light available for photosynthesis under the ice.

The sea ice is also a very effective insulator, greatly restricting the loss of heat from the relatively warm ocean (-1.9 degrees Celsius) to the much colder atmosphere (which can be -30 degrees or colder). During winter the turbulent heat loss from leads can be up to two orders of magnitude greater than from ice-covered ocean. The ice also influences ocean structure and circulation since, during ice formation and growth, salt is rejected to the underlying ocean, increasing the density of the water and setting up deep convection that contributes to the upwelling of nutrients and to the overall thermohaline circulation, or water movement driven by salinity and temperature differences, of the global ocean. Conversely, when the ice melts in spring it stabilizes the upper ocean, and, since ice formation and melt occur in different regions, this causes an effective flux of low-salinity seawater to the north.

Ice Formation

Surface seawater around the coast of Antarctica typically has a salinity between 33.5 and 34.0 parts per thousand and a freezing point between -1.84 degrees and -1.87 degrees. Unlike freshwater, seawater continues to increase in density as its temperature decreases towards the freezing point. So seawater, as it cools from the surface by heat loss to the atmosphere, tends to sink and cause convection. That brings the deeper water, which is a little warmer, to the surface. Thus a several-meters-thick upper layer of the ocean must cool to the freezing point, or slightly below, before sea-ice formation begins. The first crystals to form are minute spheres of pure ice, which grow rapidly into thin discs and then, in calm conditions, to hexagonal dendritic stars.

In the open ocean, wave-induced turbulence during ice formation stirs the initial discoid crystals through a depth of several meters, giving the water a greasy, soupy appearance. The small crystals, called frazil, are herded by wind and wave action into long plumes parallel to the wind. These eventually consolidate into roughly circular pieces of new ice with upturned edges, known as pancake ice. The pancakes grow larger and thicken by bonding together and by rafting on top of each other, leaving areas between them that are relatively free of crystals and that act as source areas for new crystals.

Inside the ice boundary, frazil forms in relatively protected areas within leads, consolidating into continuous new flexible sheets, called nilas, 0.1 meter to 0.2 meter thick. These sheets buckle easily at the slightest disturbance, easily rafting over each other to form thicker floes and creating new open-water areas.

Although new ice forms most rapidly in open-water areas where the temperature difference between the air and water is large, ice also grows on the bottom of existing floes as heat is removed from the ice-water interface by conduction through the floe. In contrast to the small, randomly oriented grains of ice that form from aggregation of frazil, congelation ice grows on the underside of existing floes as long, columnar crystals.

New ice can also form when the weight of snow cover on a floe depresses the ice surface below sea level, saturating the near-surface snow with seawater and refreezing as “snow-ice.” Because ice is such an effective insulator, this is a much more efficient method of ice growth than bottom freezing. It also incorporates surface biological communities into the ice itself. Texturally this ice is similar to that formed from coarse-grained frazil, but it can be discriminated from frazil by stable-isotope-ratio (6180, or the oxygen-18 to oxygen-16 ratio) measurements.

The structure, temperature and salinity of the ice influence its mechanical and thermal properties and its suitability as a habitat for organisms. The ice texture also records a history of the different processes that formed the ice. Typical textural, salinity and delta^sup 18^O sections for a number of cores taken from floes in the Indian Ocean sector show repeated patterns of layers of frazil and congelation crystals and multiple “c-curves” in the salinity profile, indicating that much of the ice has been built up by repeated rafting of thin elements on top of each other. The profiles show that much of the ice greater than 0.3 meter thick has been rafted or ridged and suggest that increases in ice thickness beyond 0.4-0.5 meter are usually the result of these dynamic processes. Studies of ice cores suggest that, of the total ice mass in the Indian and western Pacific sectors of the Southern Ocean, around 48 percent forms by rapid freezing in open water, 39 percent by accretion on the underside of existing floes and 13 percent by the freezing of seawatersaturated snow. In the somewhat thicker pack in the Bellingshausen and Amundsen seas (60W-135oW), the snow-ice fraction may be as large as 25 percent.

Sea-Ice Salinity

As the ice grows it develops a characteristic cellular substructure of evenly spaced platelets. Most of the salt within the sea ice is contained in brine inclusions along the platelet boundaries. The platelet width, or brine-layer spacing, is inversely proportional to the freezing rate, and the closer together the brine layers are, the more salt is included within the ice. The average salinity of even rapidly growing new ice, however, seldom exceeds 12 parts per thousand, and this percentage decreases as the growth rate slows.

The bulk salinity for cores sampled from east Antarctica averages 3.9 parts per thousand, considerably lower than that of seawater. The average salinity generally decreases with increasing ice thickness since the ice continues to desalinate with age. Thus there is a salt flux to the ocean over a period of time, not just during the initial growth phase. Sea salt in sea ice is contained within small, vertically elongated pockets of high-salinity brine along the platelet boundaries. These brine cells range in cross-sectional area from thousandths of square millimeters to a few square millimeters: At temperatures of -20 degrees they have a typical mean area of 0.01-0.02 square millimeter and a length of 0.1 to 0.2 millimeter. But as the ice temperature rises the brine cells become larger and the brine in them less concentrated. At warm enough temperatures (around -5 degrees) the brine cells can link up to form brine channels, large vertical tubular structures and associated radial tributary tubes. These channels are the major route for brine drainage from the ice. They allow seawater to percolate upward through the ice and saturate the surface snow during snow-ice formation, and they allow vertical replenishment of nutrient-rich seawater to algal communities within the ice. The salinity variation with thickness in Antarctic ice is similar to that in Arctic sea ice, despite structural differences.

Morphology of the Pack

In the summer months the pack ice consists mostly of remnants of the previous season’s winter ice and is composed of floes of generally uniform appearance and of thickness greater than 0.6 meters. In contrast, the winter- and springtime pack in the Antarctic is a diverse mix of ice of different thicknesses and of open water in leads. In mid- to late spring, as much as 30 percent of the total area of some parts of the Indian Ocean sector of the Antarctic ocean may consist of new ice less than 0.3 meter thick, with no or very little snow cover and with up to 30 percent open water. In winter and early spring there is less open water, but a substantial fraction of new, thin ice forms in leads opened by the constant motion of the pack.

The size and shape of the floes varies markedly with location and with time. One of the main mechanisms affecting their size is the penetration of swell into the pack. Longwavelength swell generated by storms in the Southern Ocean can penetrate hundreds of kilometers into the pack and break large floes into smaller rectilinear pieces. These become more rounded with time as they move and collide, creating broken brash ice between the floes. In some instances, however, other forces such as shear or compression may be equally important to determining flow size. Floes of any size can be found at all distances from the ice edge, including vast (greater than 2,000 meters) floes near the ice edge and much smaller floes in the central and southern pack. Floe break-up by swell effectively creates small areas of open water and enhances oceanatmosphere heat flux.

The Thickness of Antarctic Sea Ice The thickness of sea ice and the snow cover on it control both the surface reflectivity and the insulating effect of the pack, thus controlling oceanatmosphere energy exchange. The energy exchange by turbulent transfer and radiation from areas of snow-free thin ice and open water dominate the total surface heat budget within the sea-ice zone. But most of the ice mass is contained in the thick floes and in ridges, and it is the total mass of ice formed that determines salt input to the ocean. Consequently, to fully understand the influence exerted by pack ice, it is necessary to define its full thickness distribution. But until recently there were few systematic data available for Antarctic pack ice.

A number of studies based on measurements and observations from research icebreakers have been made over the past decade in the eastern Weddell Sea, in the Bellingshausen and Amundsen Seas, and in the Indian Ocean sector. These show a generally consistent picture: The sea-ice zone in these regions includes ice of many different thicknesses, but a large areal percentage is thinner than 0.3 meter, and the area-averaged thickness of unridged ice floes (including open water) is as low as 0.3 to 0.5 meter. There is also a trend of increasing average ice thickness with distance south from the ice edge. A typical early spring ice-thickness distribution for the Indian Ocean sector has a modal ice thickness of between 0.5 and 0.6 meter, and a second peak of very thin ice occurs because of rapid ice formation in new leads. This thin ice increases quickly in thickness by both rafting and thermodynamic growth to a half-meter or a little more, but further growth is inhibited by the resistance of thicker floes to rafting and by the replacement of heat lost by conduction through the ice with heat from the ocean. The heat flux from the deep ocean to the near-surface mixed layer can be 15 watts per square meter or more, enough to melt a half-centimeter of ice per day.

The ice-thickness distribution changes seasonally, with the greatest intra-annual changes taking place in open water and in thin ice. In March, at the beginning of the ice-growth season, most of the ice is new and less than 0.4 meter thick. In the depth of winter, in August, leads are quickly refrozen, and the newly forming ice grows very quickly so there is only a small fraction less than 0.4 meter thick. By October, air temperatures and solar radiation start to increase, leads do not refreeze as quickly, and thin ice thickens at a slower rate. As a result, there is more open water and a more uniform distribution across all ice-thickness categories below 0.7 meter. The increased open water allows more solar radiation to be absorbed in the ocean, providing a positive feedback and even further reducing new ice formation. By December all thin ice has melted, the ice concentration is as low as 50 percent, and, for the most part, only ice thicker than 0.6 meter remains. The greatest area-averaged thickness of the nonridged ice is probably less than 0.6 meter, even in midwinter.

These thickness estimates apply only to the relatively thin and undeformed, nonridged component of the pack. Although rafting is the dominant dynamical mechanism by which floes reach half-meter thickness, continued convergence beyond that results in breaking and stacking of floes to form pressure ridges. The sail height, or the height above water level, of Antarctic ridges infrequently exceeds one meter, but the ridged areas still contain a disproportionately high fraction of the total ice mass within the pack. When the effect of ridging is included, the average thickness of sea ice in the Indian Ocean sector in September is estimated to be about 0.9 meter, an increase by a factor of 1.8 over the area’s mean nonridged thickness.

The average thickness of snow on the ice remains relatively constant throughout the year, at between 0.10 and 0.15 meter, despite continual snowfall. With an average snow density of 350 kilograms per cubic meter, compared to an ice density of 900 kilograms per cubic meter, more than about 0.15 meter of snow on 0.6-meterthick ice will depress the surface below water level and convert to “snow-ice.”

Ice Drift and Deformation

Antarctic sea ice is easily moved by winds and ocean currents, and the dominant processes in the sea-ice zone are dynamic ones associated with the drift and deformation of the ice. Drift of the pack is important not only for determining the extent of the ice but also because relative motion between floes opens new leads and deforms the ice by rafting and ridge building. Cycles of convergence and divergence, and of freezing, follow the passage of storm systems. Alternately thin new ice forms in leads, followed by thickening by deformation. Ice of 0.5 to 0.6 meter thickness forms the basic “building blocks” that make up most thicker Antarctic pack ice. Drift of the ice also transports plants and animals that live, feed or breed on and within the ice, which may contribute to the circumpolar continuity of Antarctic marine ecosystems.

Satellite-tracked data buoys, deployed either on ice floes or in the water between them, have been used to investigate ice drift and to collect surface meteorological data. Most of the data available are from the Weddell Sea and from the sector of eastern Antarctica between 40(deg)E and 160(deg)E.

Wind stress turns out to be the most important force driving the ice and causing its high day-to-day variability of motion-particularly in the relatively low-concentration ice of eastern Antarctica, where the force between colliding floes is small. Under freedrift conditions the ice typically moves at 2-3 percent of the geostrophic wind speed (a balance between pressure gradient and Coriolis forces), with a turning angle of 20-30 degrees to the left of the wind direction (in the Southern Hemisphere). Thus in the zone of westerlies to the north of the Antarctic Divergence average drift is toward the open ocean to the north, and there is an average net divergence within the pack. This divergence is important in determining the location of the ice edge and in creating new open-water areas where ice can grow rapidly. But differential motion between floes, resulting from variations in wind drag caused by surface roughness and ice thickness, also causes local divergence and convergence.

In the Weddell Sea the ice drifts in a cyclonic gyre with drift speeds averaged over 30 days varying from 0.05 meter per second (4 kilometers per day) in the southwest Weddell to almost 0.2 meter per second in the northeast around 62(deg)S 10(deg)W. No significant seasonal variability in the drift has been found, and the total residence time of ice in the gyre does not in general exceed two years. The day-to-day variability of this ice drift is high, and in the central part of the Weddell there are also significant interannual differences.

In the Indian Ocean and western Pacific sectors the ice is generally a little less compact than in the Weddell Sea, and drift speeds are even higher, with measured mean velocities as high as 0.9 meter per second (78 kilometers per day). The dominant features of the ice motion in this region are westward drift parallel to the bathymetry (bottom profile) along the Antarctic continent, a cyclonic gyre in Prydz Bay and eastward drift to the north of the Antarctic Divergence. South of the divergence, the westward daily mean velocity in the Indian Ocean sector (0.23 meter per second) is greater than in either the Western Pacific (0.17 meter per second) or Prydz Bay (0.11 meter per second), whereas the eastward-drifting ice to the north moves slightly slower and more uniformly with longitude.

The meridional component of ice drift is not as uniform as the zonal flow, and in general the pack meanders both north and south with the passage of synoptic weather systems. On the continental shelf persistent regions of northward or southward ice motion are associated with coastline form or bathymetry, but north of the divergence there is little systematic pattern in meridional transport. The gyre circulation in Prydz Bay, however, produces strong ice outflux from the coast (typically around 0.2 meter per second mean northward velocity).

There is evidence of seasonal variability in the ice-drift velocity in the Indian Ocean sector. The drift rates are greater early in the season when the ice is thinner and its concentration lower. Once the pack reaches winter thickness and concentration, the average velocity decreases by as much as a factor of three. But in late spring, in the seasonal sea-ice zone north of the divergence, the average velocity again increases as the edge of the pack starts to retreat and as new ice growth in leads slows, decreasing ice concentration. There is also an increase in average speed in the marginal ice zone, the outer 150-200 kilometers of the pack where ice concentration is lower and floe sizes are smaller.

The Sea-Ice Zone Ecosystem

Until very recently, Antarctica’s oceanic food chain had been viewed as a simple system: Upwelling around the coast was thought to bring nutrient-rich waters into the near-perpetual sunlight of the Antarctic summer, which initiated blooms of diatoms that were held in the surface waters by the lowered salinity caused by the melting pack ice. Krill-small, shrimplike crustaceans-fed on these extensive phytoplankton blooms and in turn became the staple diet of the vertebrates and many of the invertebrates of the region. In autumn, it was supposed, intense storms mixed the surface and deep waters. Then, in winter a layer of ice isolated the ocean from the atmosphere, and, with the lower light levels, productivity decreased and the phytoplankton became dormant until spring.

More recent studies utilizing research icebreakers and sophisticated, yearround scientific stations have allowed scientists to examine the biology of the sea-ice zone in seasons other than high summer, and the results of these continuing studies have forced a reappraisal of some of the established concepts of Antarctic biology. We now know that the simple diatoms-krill-whales linkage grossly oversimplifies the food web of the Southern Ocean. To be sure, this link does exist. But there are many others, and this one may not be the most important in terms of the flow of energy and carbon through the Southern Ocean system. We now think it likely that a series of communities, rather than a single ecosystem, vary in space and time in the waters between the Antarctic convergence and the continent. Significantly, we are beginning to understand that pack ice, and to a certain extent fast ice, play a number of important biological roles.

The most obvious ecosystem division is between the area seasonally covered by pack ice and that which remains ice free. There are also coastal areas and embayments such as the Ross and Weddell Seas and Prydz Bay that extend far to the south and experience quite different physical conditions from the rest of coastal Antarctica. Biological communities in these regions can have a degree of temporal stability but some are highly seasonal in their extent and location.

The ecosystems of the Southern Ocean that are affected by the pack ice can be further subdivided by region into the seasonal pack-ice zone, the coastal zone, the perennial pack ice and the marginal-ice zone. Of these, the last turns out to be the source of much of the region’s biological productivity. The realization that the seasonal movement of the sea ice and the productivity of the Antarctic region are intimately linked constitutes one of the major paradigm shifts in Antarctic biology. To better understand the influence of ice on the food chain it will be helpful to start from the top, with the vertebrates, and work downward.

Even the earliest studies recognized ice as a platform on which animals breed and rest. Because little of the Antarctic coastline has exposed rockand what it has lies isolated from the open ocean, and the main food supply, by the sea ice for much of the yearbirds such as the Emperor penguin utilize the fast ice as a stable, flat breeding ground. Their habit of laying eggs on the fast ice close to the continent at the height of Antarctic winter isolates them from their prey species by variable areas of fast ice. And there are suggestions that at least some Emperor penguin colonies are located in the vicinity of near-permanent coastal polynyas, which provide close access to the pelagic food supply even in deepest winter.

Other species that utilize the ice as a breeding ground include the pack-ice seals-the crabeater (the most numerous species of seal on the planet), Weddell, Ross and leopard seals. For these species, the pack ice provides a floating platform immediately above their food supply, allowing them to exploit their preferred species of prey even while they are weaning their pups and unable to travel far from them. Seals can cut their way through ice to gain water access and thus need not find leads or the ice edge to enter the water to feed.

In contrast, Adelie penguin colonies breed onshore on the few rock areas around the continent, so the fast ice, in spring, forms a barrier to their access to food. Because Antarctic penguins are at the mercy of fluctuations in the extent of the fast ice, it is now thought that the fast-ice extent can radically affect their breeding success. On the Antarctic Peninsula, where Adelies, rock hopper and macaroni penguins coexist, patterns are emerging in the relations between long-term changes in ice cover and the relative abilities of the three species to cope with these changes.

For flighted birds, the sea-ice coverage is less crucial. They can forage more widely than penguins and have a greater ability to exploit leads and polynyas. Pelagic animals, such as whales, are thought to remain offshore of the ice edge for the most part, although some species such as orcas and minke whales are known to inhabit the pack-ice zone. There is also emerging evidence that some proportion of the population may stay in the pack ice right through the winter. For species such as orcas that can prey on seals and penguins in the pack ice, there are obvious advantages to being able to live among the floes. Yet to invert the problem, why would many of the krill-feeding vertebrates leave the region during winter when their food supply remains? It is still unclear how the distribution of the krill-eating species of whales-the baleen whales: blue, minke, fin and humpback-relates to the distribution of krill in seasons other than summer.

The summer distribution of krill on a gross scale has long been established as the area between the Antarctic Convergence and the Antarctic coastline-a range of some 35 million square kilometers. The details of the summer distribution have become clearer over the last 20 years, however, through intensive research and the activities of the krill-fishing fleet. Krill abundance is greatest in the seasonal pack-ice zone. North of this there are a few areas-such as South Georgia Island in the South Atlantic-that sustain large populations of krill, but the majority of the krill is found much closer to the continent, particularly in a band along the edge of the continental shelf. With better understanding of the summer krill distribution, attention has shifted to their distribution and behavior during the other three seasons.

Krill are mainly herbivorous-at least in summer-and the vast swarms around Antarctica consume a significant proportion-between 5 and 10 percent-of the primary productivity of this region. The cycle and fate of primary production have been intensively studied over the past 10 years, and our understanding of the dynamics of the biology of the pack-ice zone has probably changed more than any other aspect of Antarctic knowledge.

The cycle of northward pack-ice advance in autumn and retreat toward the continent in spring appears to be one of the driving mechanisms for the productivity of the ocean. The melting of sea ice in spring has two effects. The first, a physical effect, is the release of fresher, less dense water into the surface layer of the ocean at the edge of, and in leads within, the disintegrating pack. A lens of fresher water forms, separated from the more saline, denser water below by a density gradient-a pycnocline. The second effect is biological. Within and on the underside of the pack-ice floes lives a community of microorganisms-bacteria and microscopic plants and animals-that are released into the lens of surface water by the dissolving ice. Trapped in the nearsurface layer by the pycnocline-in nutrient-rich water with abundant light-the unicellular plants begin to divide rapidly, forming ice-edge blooms. These blooms may be responsible for 60 percent of the productivity in areas of the ocean that are ice-covered in winter-with productivity on the order of 60-120 grams of carbon per square meter per year. The annual production of biogenic carbon in this whole area has been calculated to be around 0.29 gigaton.

These blooms are then exploited by the grazers of the ocean-but not predominantly krill, as was initially thought. Many of the microorganisms of the water column are herbivorous, and they consume a large fraction (up to 20 percent) of the primary production and are in turn consumed by other microorganisms and by larger planktonic organisms. Much of the primary productivity may not stay within the cells of the phytoplankton but may leach out into the water, feeding bacteria that remineralize the nutrients. This whole chain is known as the “microbial loop,” and it may be carbon’s major route from the atmosphere to the ocean in a dissolved form-completely bypassing the larger animals.

The larger herbivores such as copepods and krill still retain their importance as consumers of phytoplankton, particularly as a route for carbon and energy from the primary producers to the larger animals such as whales and birds. And it turns out that their role in biogeochemical cycles may be more pronounced in winter than in spring and summer.

The cycle of primary productivity in the Southern Ocean is highly light dependent and is also affected by the stability of the waters in the open ocean. During fall, winter and spring, the storms of the open ocean mix the surface waters deeply-well beyond the lighted layer, or euphotic zonethus keeping primary biological production at a low level. In summer in the open ocean, upper primary productivity remains low (around 16 grams of carbon per square meter per year in the water column) despite relatively high nutrient levels.

In the pack-ice zone, however, there is little wind-driven deep mixing during most of the year, because the ocean is cut off from the atmosphere by the presence of sea ice. The ice also tends to restrict the available light, with levels under ice and snow being 50-1,000 times less than those in an uncovered water column at similar depths. Living within the ice environment does, however, have its drawbacks: The temperature can be very low-down to -20 degrees-and the salinity can be very high-up to 300 parts per thousand putting considerable stress on the organisms that live there. As the temperature drops and more water freezes, the salinity within the brine cells in the ice rises, thus exacerbating the stresses. At the same time, the brine volume decreases, and this reduces space available for colonization by aquatic organisms. Nonetheless, in contrast to the open water below the ice, the internal spaces in the ice form an environment in which plant growth can occur. Additionally, the underside of ice provides a surface onto which thick mats of phytoplankton and other organisms can attach and be maintained in a high-nutrient, stable environment. The ice thus develops and incorporates its own biological community, which grows slowly over winter, increases pace during spring and then is released into the water when the ice melts.

Organisms of the Ice Floes

Although phytoplankton are dominated by pennate diatoms, a wide variety of heterotrophic microorganisms are found in sea-ice communities: bacteria, fungi, ciliates, flagellates, foramanifera, nematodes, turbellarians, polychaetes, amphipods, and cyclopoid and harpactecoid copepods. The sea-ice biota, in fact, partly resemble benthic (bottom living) rather than open-ocean fauna. How the sea-ice community forms is poorly understood. Many of the microorganisms may be filtered from the water by mats of newly formed ice crystals when water is pumped through by wave action. In spring, much of the biota may form aggregates and sink to the sea bed as the ice melts, and some may seed the phytoplankton blooms in the ice-edge zone. Ice algae are adapted to dim conditions, living and growing at light levels as low as 0.1 percent of the total surface irradiation. They are tolerant of a wide range of salinities and low temperatures found within the pack ice. Although it is uncertain to what extreme they can survive, we know they can tolerate temperatures below -6 degrees-lower than any fully aquatic organism would ever encounter. The biomass of algae in sea ice can be extremely large; chlorophyll levels of up to 2,000 micrograms per liter have been measured in the lower levels of ice cores (compared to water-column levels of 0.1-10 micrograms per liter). In platelet ice, levels can be five orders of magnitude higher than in the adjacent water column. Estimates of the overall production by the sea-ice biota are around 20 percent of the total productivity of the Southern Ocean.

Biotic communities are located throughout sea ice-on the surface, inside and on the bottom. The surface or infiltration community may form when seawater containing organisms ponds on the surface of the snow, and the greater light, nutrient and temperature levels found there favor their growth. Ponds can also form when the sea ice is deflected below water level by pressure ridges, trapping pools of seawater on the surface of the floes. The third type of surface community takes hold when melt pools form; these can be fresh, brackish or saline water, depending on the source of the melt water. The communities present in these melt pools reflect the salinities-the freshwater ones contain terrestrial algae and the more saline ones, marine species.

Interior habitats develop when air temperatures at or slightly below the freezing point initiate, but do not complete, brine drainage. Brine channels are the most common ice habitat, and they allow vertical movement of water and organisms through the interior of the ice. The freeboard habitat becomes populated when brine drains from the upper layers because of warming, and algal growth responds to the increased temperature.

Band communities form by the accretion of new ice under a previously formed bottom layer of organisms, by the incorporation of cells at the time of first freezing of surface waters or by ice rafting. The interstitial community occurs in the bottom layer of the ice where ice crystals are generally small. Platelet-ice communities can form when a layer of this ice scavenges organisms rising from deeper water and accumulates under the existing ice floes. They are usually found near ice shelves and can be areas of high primary productivity. The under-ice community may be attached to the ice in the form of mats or filaments, or may be floating or swimming just below the underside.

The under-ice community is highly variable and very patchy. The patchiness may be related to the physical structure of the ice-the extent of ridging and snow cover, both of which affect the amount of light that penetrates, and the type of ice. It may also reflect the productivity of the area in which the ice was formed. The communities themselves may also exert some control over their habitat. Pigmented organisms in the ice may affect the rate of ice breakup in spring through their effect on heat absorption and on the porosity of the ice and hence its structural strength.

The larger pelagic herbivores, such as krill, can congregate, feed and grow under the ice during the late winter and spring. The ice forms a concentrated layer of organisms near the sea surface that can be exploited by grazers such as krill, and they in turn can be preyed on by animals such as crabeater seals and penguins, which have been observed hunting in the complex structure of the under-ice zone.

The behavior of krill during winter has long been a mystery. Where do they go? And how do the vast swarms of these active, constantly swimming crustaceans find enough energy to overwinter? The revelation of the importance and scope of the sea-ice microbial community provides some clues to the answers to these questions. Studies of spring activity have revealed that krill scrape the layer of organisms off the bottom of floesthey can clear 100 square centimeters of algae from ice in about 5 minutes and reports of the presence of krill in this surface layer during spring have now become commonplace. Krill do, however, have to survive several months when the sea-ice community is not well established, and they are also found in areas such as South Georgia, which is usually ice free. So reliance on the sea-ice community cannot be the only strategy they employ. It seems likely that larval krill, which have very little ability to withstand starvation, depend to a large extent on the ice algal community to survive winter. Adult krill, on the other hand, are known to be able to withstand long periods of food deprivation and overwinter more opportunistically; they may reduce their metabolism, become carnivorous or detritivorous, shrink or utilize ice algae, depending on the circumstances.

The complex structure of the underside of the ice also provides something unique in the true pelagic zone-somewhere to hide. Herbivores can use the complex features found under floes, as well as the channels formed in decaying ice, to retreat from predators too large to pursue them there. The strategy is not always successful, however. There are reports of crabeater seals using their long snouts to penetrate crevices and suck krill out of the ice.

In the horizontal plane, the ice also provides the krill with seasonal protection from fishing fleets. The only winter fishery for Antarctic krill is in the waters around ice-free South Georgia. The fishery in the South Atlantic follows the retreating ice in spring and moves north again with ice growth in autumn. And because of perennially difficult ice conditions, certain areas where krill are abundant are unlikely to be fished.

Climate Links

Large-scale climate models predict that increased global temperatures would have considerable impact on the Antarctic sea-ice zone and lead to a decrease in ice extent. Changes to the ice would affect the local physical and biological environment, and, because of the role this zone plays in world climate, there are potential feedbacks on a global scale. Are there, then, any recent changes in the sea-ice zone around Antarctica that could be attributed to global warming?

One good way to investigate global processes is to use large-scale remotesensing techniques. Satellite data, however, show no statistically significant change in the total ice extent over a period that saw a recorded atmospheric temperature increase. However, significant interannual variations in the spatial distribution of the ice extent around the hemisphere have been observed at a periodicity of 4-5 years, and these are coupled to anomalies in the distribution of atmospheric pressure, wind stress and temperature in the Southern Ocean. There is also evidence of some regional decrease in ice cover during summer. But the satellite passive-microwave data have coarse resolution (25 kilometers and greater) and provide little or no information on many important ice characteristics, especially ice thickness. Changes in the total ice volume, for example, could result from decreased thickness without there being significant retreat of the ice edge or reduction of the mean ice concentration as monitored by the satellite data.

There is also the problem of obtaining data prior to 1973, when passivemicrowave satellite data first became available. A recent study used data from the whaling fleets that plied the Southern Ocean for most of the first 70 years of this century. The whaling fleets tracked the ice edge as they followed the whales that congregated there and moved south with the edge as the season progressed. Data on the position of the whaling ships from the middle years of the century suggest that, although the sea-ice extent has been relatively stable since 1973, there was a major change in average ice extent from the mid-1950s to the early 1970s. This change amounts to some 5.56 million square kilometersequivalent to a decline in area of approximately 25 percent.

What effect might such a dramatic decline in sea ice area have had, both locally and globally? Sea ice is postulated to be one of the geophysical parameters most sensitive to climate change, and this sensitivity makes the inter-annual variation in sea ice a prime variable for detecting systematic global trends. It is probable that any response of sea ice to climatic change will be seen in the total sea-ice mass, not just in changes in ice extent and concentration.

Processes within the sea-ice zone are also linked to a number of other globally important systems, and these, too, are likely to be affected by changes in sea-ice production. For example, the sea-ice region plays a major role in the formation of not only Antarctic bottom water but also of other Antarctic water masses that are important in the circulation and mixing of the world oceans. Salt rejection from freezing sea ice, together with supercooling beneath ice shelves, drives deep mixing and thermohaline convection, which transports cold water into the deep ocean. The exchange of gases between the atmosphere and the ocean is also modified by the presence of ice, and thermohaline convection is important for ventilation of the deep ocean. Sea ice, floating ice shelves and continental ice sheets are coupled through the effects of the sea-ice cover on evaporation from the ocean, and thus on atmospheric moisture and snowfall rates on the continental ice sheet. Ice shelves may also be coupled to sea-ice growth by way of vertical circulation caused by salt rejection.

In terms of the biota, there are a number of lines of evidence that point both to an intimate relation between the extent of sea ice and the productivity of the region and to changes that may be related to global warming. Although some of these changes may be mediated through the cycle of ice-related production outlined above, there may be more subtle effects caused by shifts in energy flow to vertebrate predators because of changing ice conditions. Other recent results indicate that krill are more abundant and productive following years of above-average ice cover. But in years when there is above-average open water, salps dominate. Salps are gelatinous zooplankton-tunicates, related to sea squirts-that are thought to be of little use as food for birds, seals, whales or fish. In years when salps are abundant and krill are scarce, landbreeding vertebrates have poor reproductive years because they are unable to find enough food to rear their offspring. Off the Antarctic Peninsula there is some controversial evidence that years in which krill are scarce are becoming more frequent, and this is correlated with more frequent years with low ice cover during winter. Data from this region suggest that there may have been an overall tenfold decline in krill abundance and that the driving force behind this change is the reduction in overwinter ice cover. Associated with these changes are shifts in the abundances of some of the penguin species. These findings have stimulated major research programs into current abundance levels of krill and the relationship between biological productivity and climaterelated cycles.

Obviously, changes in the physical environment that cause a diminished abundance of krill have major implications for the health of the populations of animals that depend on them-and for the fishery that focuses on the Antarctic Peninsula area during summer. Less obvious is the effect that such an ecological shift would have on carbon flow. In years when salps are abundant, it has been estimated that 19 percent of the carbon fixed during primary production in the Peninsula area is consumed by salps, and, because salps are not heavily preyed on, this carbon may end up sinking out of the surface layers in fecal pellets and in dead bodies. This loss of organic material to deep water may have further feedback effects in the ecosystem.

The Uncertain Future

Our growing understanding of the pack-ice zone comes at a time of increasing recognition of the importance of this zone in global climatic processes and in the regulation of the biological cycles of the region. Much has been accomplished, but recent evidence of changes that may be linked to the increase in global temperatures add urgency to the work yet to be done. The next round of research into the Antarctic and its systemsphysical and biological-may well reveal that what we have learned thus far is actually the harbinger for dramatic change in the world’s climate system.