Eight large (~120 m sea-level lowerings), 100-kyr-scale ice ages over the past 800 kyr;
Sixty-two stages5 representing 31 ice-sheet advances on ~20-, 41-, and ~100-kyr-scale during the Pleistocene; and
Over 100 named stages and 50 glacial advances since the late Pliocene “initiation” of NHIS (Emiliani, 1955; Hays et al., 1976; Shackleton, 1967).
A pulse of ice-rafted detritus (IRD) into the northern North Atlantic ca. 2.6 Ma (late Pliocene) is associated with a major δ18O increase; this has been interpreted as the inception of NHISs (Shackleton et al., 1984). However, this inception reflects not initiation but an increase in the size of NHISs growth and decay (e.g., Larsen et al., 1994). Significant (at least Greenland-size) NHISs extend back at least to the middle Miocene (ca. 14 Ma; see summary in Wright and Miller, 1996) and recent data indicate that large NHISs may have existed since the middle Eocene (Eldrett et al., 2007; Moran et al., 2006).
The imperfect direct record of Antarctic glaciation has similarly led to the progressive extension of initiation of a continent-size ice sheet from 15 Ma (middle Miocene) back to 33.55 Ma (earliest Oligocene) (see summaries in Miller et al., 1991, 2005a,b; Zachos et al., 1996). In this contribution we suggest that continental ice sheets have been intermittently present on Antarctica through the Late Cretaceous, a time when Antarctica took up residence at the pole (http://www.ig.utexas.edu/research/projects/plates/).
Deep-sea isotope records have long been used to interpret Antarctic ice-sheet history. Based on deep-sea δ18O records, early studies of Shackleton and Kennett (1975) and Savin et al. (1975) assumed that a continent-size ice sheet first appeared in Antarctica in the middle Miocene (ca. 15 Ma), though they noted that glaciation (in the form of mountain glaciers and sea ice) probably occurred back through the Oligocene. Also using isotope data, Matthews and Poore (1980) suggested that large ice sheets existed in Antarctica since at least the earliest Oligocene (33.5 Ma). The differences in interpretation partly illustrate problems in using δ18O as an ice-volume proxy, because deep-sea δ18O values also reflect deep-water temperature changes that generally mimic high-latitude surface temperatures. Miller and Fairbanks (1983, 1985) and Miller et al. (1987, 1991) provided the strongest isotopic evidence for the presence of ice sheets during the Oligocene; high δ18O values measured in deep-sea cores (>1.8‰ in Cibicidoides spp. or >2.4‰ in Uvigerina spp.) require bottom-water temperatures colder than today if an ice-free world is assumed. Such low bottom-water temperatures are incompatible with an ice-free world; their isotopic synthesis (updated and presented in Figures 1 and 2) suggest at least three major periods of Oligocene glaciation.
A campaign of drilling near Antarctica by the Ocean Drilling Program (ODP) in the late 1980s returned firm evidence that supported the δ18O record for large ice sheets in the earliest Oligocene that included grounded tills and IRD at lower latitudes than today (see summaries by Miller et al., 1991; Zachos et al., 1992). We update the summary of the direct evidence for Eocene-Oligocene ice in the form of tills and glaciomarine sediments near the Antarctic (Figure 1), using more recent drilling by ODP (Cooper and O’Brien, 2004; Strand et al., 2003), the Cape Roberts drilling project (Barrett, 2007), plus studies that extend West Antarctic glaciation back through the early Oligocene (Seymour Island) (Ivany et al., 2006) and into the Eocene (King George Island; Birkenmajer et al., 2005; Troedson and Riding, 2002; Troedson and Smellie, 2002). The evidence for large, grounded ice sheets begins in the earliest Oligocene and continues through the Oligocene (Figure 1). Seismic stratigraphic studies summarized by Cooper et al. (forthcoming) also show intense glacial activity beginning in the Oligocene in both East and West Antarctica. There is excellent agreement among proxies that Antarctica was in fact an icehouse during the Oligocene and younger interval. Ice-volume changes have been firmly linked to global sea-level changes in the Oligocene and younger “icehouse world” of large, varying ice sheets (Miller et al., 1998); Pekar et al. (1996, 2002) recognized that the three to four major Oligocene glaciations of Miller et al. (1991) in fact reflected six myr-scale sea-level falls and attendant ice-growth events. The record of glaciomarine sediments documents that the ice sheets occurred in Antarctica (Figure 1), though an NHIS component cannot be precluded due to scarce Northern Hemisphere Oligocene records. Today 33.5 Ma is cited as the inception of the Antarctic ice sheet, though this supposition is now being challenged and pushed back into the Cretaceous (Miller et al., 1999, 2003, 2005a,b; Stoll and Schrag, 1996). Nevertheless, 33.55 Ma was probably the first time in the past 100 myr that the ice sheets reached the coast, allowing large icebergs to calve and reach distal locations such as the Kerguelen Plateau (Figure 1) (ODP Site 748) (Zachos et al., 1992).
There is evidence for glaciation in the older Antarctic record. Coring by ODP Legs 119 and 120 (Barron and Larsen, 1989; Breza and Wise, 1992) and seismic stratigraphic studies (Cooper et al., forthcoming) suggest the possibility of late Eocene (or even possibly middle Eocene) glaciers in Prydz Bay. Seismic stratigraphic studies (Cooper et al., forthcoming) also suggest the possibility of late Eocene glaciers in the Ross Sea. Other studies extend the record for West Antarctic glaciation back from 10 Ma to 45 Ma (Birkenmajer, 1991; Birkenmajer et al., 2005). Though Birkenmajer et al. (2005) interpreted the Eocene tills as evidence for mountain glaciers and not necessarily ice sheets, it points to the likelihood that the continental interior could have supported an ice sheet