Some numerical models suggest1,2 that melting of the Antarctic Ice Sheet will cause global sea level to rise by 3 metres or more by 2300. But such models differ in the exact magnitude of the rise, as well as in which parts of the ice sheet are likely to contribute and the mechanisms involved. One way of assessing how well models perform is to see whether they correctly ‘hindcast’ the ice sheet’s response to past changes — which requires independent geological evidence of what actually happened. On page 225, Gulick et al.3 present a record of sedimentation on the periphery of the Antarctic, beyond the Aurora sector that drains ice from the east of the continent. The record spans the whole period from ice-sheet inception to the present day, and reveals how the ice sheet has responded — or not — to warm climate periods in the past.
Since the inception of ice on Antarctica about 34 million years ago, there has been a general trend of global cooling. But superimposed on this trend have been periods of relative warmth4, during which temperatures and atmospheric carbon dioxide concentrations were comparable to those expected under future climate-warming scenarios. Geological records show that sea levels were relatively high during these periods, which means that ice equivalent to several metres of sea-level rise must have melted from Antarctica2,5.
Antarctic surface melting today had been thought to be negligible, but recent work6 shows that surface drainage is widespread around the periphery (Fig. 1). The fate of increased meltwater in a warming climate and its effect on ice flow are unclear. Meltwater could speed up ice flow by enhancing the slipperiness of the ground over which the ice flows, or cause water-filled crevasses to propagate through floating ice and drive its collapse. A higher meltwater volume could also increase the efficiency of meltwater export, avoiding either of those developments.
On the other hand, it is widely feared that ocean warming will destabilize marine margins of the ice sheet1,2, particularly where the ground deepens inward of the ice margin. This effect promises future sea-level rise, with deep basins in the interior of the ice sheet, such as the Aurora Subglacial Basin in East Antarctica, being especially vulnerable.
Continental shelves are repositories for sediments that have eroded from the continental interior and been brought to the margin. Gulick et al. used a technique called seismic reflection profiling across the Aurora continental shelf to reveal the subsurface stratigraphy of sediments approximately 1,300 metres deep, which encompass the whole history of the Antarctic Ice sheet in this sector. The authors identify three sediment groups (packages) of distinctly different character. The lowermost package is non-glacial, but, towards the top, isolated pebbles indicate the first arrival of icebergs loaded with glacial debris, drifting across the site.
The second and third packages reveal two contrasting glacial regimes. In the lower of these packages, 11 stacked sediment layers have rough surfaces, produced by erosion from an ice sheet that had expanded onto the continental shelf. The surfaces are dramatically incised by channels, indicating that liquid water flowed in large volumes along the bed of the ice sheet. These periods of abundant meltwater are interspersed with long periods of marine sedimentation (which occurred when the ice margin retreated hundreds of kilometres into the continental interior, before advancing again). By contrast, the uppermost package lacks channels or any substantial marine sedimentation. This indicates a marked switch to an ice-sheet mode characterized by the polar conditions of more-recent, large ice sheets, which retreat only briefly between glacial expansions.
Thanks to erosion of the sea floor in the most recent phase(s) of glaciation, parts of each of these three sediment packages can be found close to the sea floor. This allowed Gulick and co-authors to analyse marine microfossils and terrestrial pollen in sediment cores to determine the ages of the three sediment packages. They found that the first arrival of debris-carrying icebergs to the continental shelf occurred earlier than the climate transition 34 million years ago. The change from channel-rich, dynamic ice sheets to the current polar conditions occurred a minimum of 8.6 million to 4.78 million years ago. Importantly, this work provides a regional framework of ice-sheet events with some preliminary age constraints for a little-understood sector of the Antarctic Ice Sheet.
There are several unknowns in this picture. It is not possible to infer the distances to which grounded ice withdrew during any retreat episode, nor the rates at which it did so. Instead, the authors make an assumption based on different erosion patterns beneath the present-day ice sheet7. Reconstructing ice extents smaller than current ones is a considerable challenge for all scientists in this field — any surviving evidence is now buried.
The report of large channels is exciting. The authors link them to the production of surface meltwater by warm air temperatures — which is reasonable, because the channel (water) volumes are large and unlikely to be produced by melting due to pressure or friction at the bed, and because well-developed subglacial river networks are commonly fed by surface meltwater that infiltrates the ice body8. The drainage of subglacial lakes could also have formed the channels. But such lakes are active beneath the present-day ice sheet9, and the offshore sediment stratigraphy of recent glaciations3 is distinctly different from that of the earlier, water-rich glaciations discovered by Gulick and colleagues. A lake origin would therefore demand a different storage, routing and refilling regime from that observed today.
The authors’ discovery of a relatively stable Aurora sector for roughly the past 5 million to 8 million years is at odds with findings from the neighbouring Wilkes sector, in which ice underwent episodes of major retreat10 and advance11 about 5.3 million to 3.3 million years ago, during a warm period known as the Pliocene epoch. This disparity is not necessarily cause for concern. Alternative behaviours or rates of change in different ice-sheet sectors are well known from contemporary observations12 and from the most recent deglaciation13, and model simulations of the Pliocene echo the contrasting responses of the Wilkes and Aurora basins recorded by geological evidence5.
Recent experiments1414 show that different Antarctic sectors are probably influenced by different climate properties, and that the Aurora sector is indeed most sensitive to air-temperature changes, whereas others are vulnerable to ocean warming. Gulick and colleagues’ findings may constrain those results, by showing that an early period of warmth produced abundant meltwater and drove major changes in the size of the Aurora sector, but that this part of the ice sheet was basically unresponsive to the subsequent Pliocene. Ice-sheet sectors that dance to different tunes may be the norm, but that means that there is an urgent need for geological data that can constrain the magnitudes and scales of that variability if we are to better understand the different vulnerabilities of different sectors.
Nature 552, 183-184 (2017)