PerspectiveClimate

How Antarctica got its ice

See allHide authors and affiliations

Science  01 Apr 2016:
Vol. 352, Issue 6281, pp. 34-35
DOI: 10.1126/science.aad6284

Ice sheets such as those on Greenland and Antarctica today not only respond to changing climate but can also cause climate to change. Their sizes have fluctuated substantially in the past. In particular, Antarctica was effectively ice-free until its ice cover began to expand rapidly at the Eocene-Oligocene boundary around 34 million years ago (see the figure). Recent research, including a report by Galeotti et al. on page 76 of this issue (1), helps to identify the mechanisms that led to this rapid ice sheet growth.

The extent of Antarctica's sea ice and ice sheet was long thought to be driven by tectonic forces. According to Kennett's influential “gateway hypothesis,” the widening and deepening of Drake Passage that led to the development of the Antarctic Circumpolar Current caused Antarctica to become thermally isolated, promoting ice sheet growth (2). However, this hypothesis has been out of favor since DeConto and Pollard's pioneering modeling work (3) instead implicated a fall in atmospheric carbon dioxide as the main driver, with the resulting growth of the Antarctic ice sheet accelerated by positive ice sheet-climate feedbacks. Observational evidence has been mounting for such a drop in atmospheric CO2 at the Eocene-Oligocene boundary (4, 5). But Kennett's gateway hypothesis may yet prove to have been correct after all, albeit via more complex mechanisms than the one he originally imagined.

A green Antarctica.

At the time of the Eocene-Oligocene boundary, about 34 million years ago, the Antarctic ice sheet was still small but about to expand. The paleogeography in the image is based on the reconstructions in (15). Vegetation and cloud cover are based on (11), the Antarctic ice caps are based on (3), and the phytoplankton blooms indicating increased productivity are based on the ideas in (6, 7).

ILLUSTRATION: ALAN KENNEDY/UNIVERSITY OF BRISTOL

What caused the drop in CO2 at the Eocene-Oligocene transition? Recently, there have been some tantalizing indications that tectonically driven gateway changes may have caused the CO2 changes. In early work, Zachos et al. (6) found evidence for an increase in ocean productivity at the Eocene-Oligocene boundary, attributed to changes in ocean circulation and upwelling associated with the opening of the Tasman Seaway (7). Based on increased silicic acid use by diatoms in the Late Eocene, Egan et al. (8) have suggested that deepening and/or widening of the Southern Ocean gateways increased the strength of the Antarctic circumpolar current and led to increased upwelling south of the polar front. The resulting increase in surface-water nutrient concentration could have led to diatom proliferation, more organic carbon burial, and hence CO2 drawdown.

A recent modeling study (9) has also indicated that gateway changes could be important for the global carbon cycle. Using an Earth system climate model, Fyke et al. showed that the opening of Drake Passage can lead to an increase in Atlantic overturning circulation and a decrease in Pacific overturning circulation. As a result, the characteristic residence times in the two basins change, such that the Atlantic reservoir of dissolved inorganic carbon becomes smaller and that of the Pacific grows. The net effect is a global increase in the ocean carbon reservoir, and hence an increase in CO2 drawdown from the atmosphere.

Galeotti et al. now present sedimentological evidence from Antarctica that calls for a reinterpretation of the ice sheet history across the Eocene-Oligocene transition. Geochemical proxy records previously suggested that the early ice sheet was large and transient, perhaps reflecting an “overshoot” of the climate system in response to a rapid forcing (10). The ice sheet was thought to have retreated after this “Early Oligocene Glacial Maximum,” as climate feedback processes resulted in a new steady state with a smaller ice sheet (10). Instead, Galeotti et al. suggest that the early ice sheet did not reach the coast at the Ross Sea. Rather than retreating, it subsequently advanced, reaching the Ross Sea continental margin 32.8 million years ago. Galeotti et al. propose that the initial, smaller ice sheet was able to respond dynamically to local variations in insolation on the comparatively short time scales of orbital precession and obliquity changes (tens of thousands of years). Once it reached the continental margin, it became relatively insensitive to local insolation forcing, instead fluctuating in size on the longer eccentricity time scale (hundreds of thousands of years), in conjunction with other components of the global climate system.

Climate itself is influenced by changes in ice sheets, both directly through changing albedo and surface height, which lead to local cooling and atmospheric circulation changes, and indirectly through changes to the carbon cycle. Several recent model studies have explored the response of the climate system to an expansion of the Antarctic ice sheet (1113). All models agree that an ice sheet causes substantial local cooling caused by the higher altitude and more reflective surface. However, beyond Antarctica, the models provide different predictions for which regions of the Southern Ocean warm or cool due to the ice sheet. It therefore remains unclear whether growth and retreat of the Antarctic ice sheet is dominated by positive or negative feedbacks associated with ocean temperatures at its marine margin.

However, very recent work indicates that the sign of the feedback could depend on the configuration of the Drake Passage and Tasman Seaway. Kennedy et al. (11) modeled the response of the climate system under two different gateway configurations. Before 34 million years ago, the expansion of the Antarctic ice sheet led to a strong warming (and therefore a potential negative feedback on ice growth) in the Pacific and central Indian sectors of the Southern Ocean. After 34 million years ago, it led to a cooling in large parts of these regions (and thus a potential positive feedback on ice growth). Thus, the timing of glaciation could be determined, in part, by paleogeographic and gateway changes.

Ice sheets likely affect the global carbon cycle through numerous processes operating on different time scales. These effects include changes to ocean productivity and carbon burial, deep-water formation and carbon storage, continental weathering and CO2 drawdown, sea-level change and carbonate deposition, and permafrost area and terrestrial carbon storage. Galeotti et al. emphasize the difference between the dynamic behavior of the initial, smaller Antarctic ice sheet and the increased stability of the subsequent larger ice sheet. Meanwhile, Middle Miocene geochemical records suggest that different regions of the Antarctic may have different sensitivities to pCO2 (14).

The work by Galeotti et al. is consistent with the notion that the Antarctic ice sheet formed when a key carbon dioxide threshold was passed near the Eocene-Oligocene boundary. The authors conclude that CO2 concentrations also control the sensitivity of the ice sheet to orbital forcing. However, the results of this study and that of Greenop et al. (14) raise questions about how the ice sheet itself affects the carbon cycle. For example, to what extent do the size and stability of the Antarctic ice sheet affect CO2 variability on orbital time scales?

To understand climate-ice sheet interactions on geological time scales, we must consider the entire coupled climate, ice sheet, and carbon cycle system, including both tectonic and orbital forcings. Important advances toward this goal have already been made by both observational and modeling communities. However, on its own, each approach has its limitations. State-of-the-art Earth system models that include the necessary climate-carbon-ice sheet processes are often designed to be run for hundreds, not millions, of years; thus, modeling studies must use approximations that need to be considered when interpreting their results. Observational studies are often carried out at single locations and focus on measuring one part of the system. It can be challenging to correctly interpret these records in a global and system-wide perspective. However, combining insight from both modeling and data together allows the strengths of the two approaches to be maximized—the veracity of the data with the system-wide interpretation from the modeling. Only an integrated model and data approach will allow the complex web of interconnected processes associated with Antarctic ice sheet growth and decay to be untangled.

References

View Abstract

Stay Connected to Science

Navigate This Article