Attribution: You must attribute the work in the manner specified by the author or licensor (but no in any way that suggests that they endorse you or your use of the work).Noncommercial ‒ you may not use this work for commercial purpose.No Derivative works ‒ You may not alter, transform, or build upon this work.Sharing ‒ Individual scientists are hereby granted permission, without fees or further requests to GSA, to use a single figure, a single table, and/or a brief paragraph of text in other subsequent works and to make unlimited photocopies of items in this journal for noncommercial use in classrooms to further education and science.

Geologists have long recognized (e.g., Lyell, 1830–1833) that during the Cenozoic, a world without significant polar ice sheets (Greenhouse World) was transformed into one with large ice sheets at both poles (Icehouse World; Miller et al., 1991), with concomitant major changes in land and ocean biota (e.g., Prothero et al., 2003). When and where cooling occurred, when the ice sheets started to form on each hemisphere, whether they persisted once formed, and the cause(s) of the cooling have long been debated (e.g., Thomas et al., 2006). This issue of Geology contains two contributions to this debate, one on northern hemispheric cooling (Schouten et al., 2008, p. 147 of this issue), the other on tropical climate and biotic changes (Pearson et al., 2008, p. 179 of this issue). Where do these contributions fit in the overall debate?

Over the last few decades, estimates of the timing of initiation of glaciation have been increasing. In the early 1970s, Antarctica was thought not to have been cold enough for continental ice sheets until the late Miocene (Denton et al., 1971). Since the mid-1970s (Kennett and Shackleton, 1976), oxygen isotopic records of benthic foraminifera, documenting a combination of global ice volume and deep-sea (thus high-latitude surface water) temperatures, have become available (Zachos et al., 2001; Fig. 1), documenting stepwise combined cooling and glaciation, with the most important steps 1) close to the Eocene/Oligocene boundary; 2) in the middle Miocene, and 3) at the beginning of the Plio-Pleistocene ice ages (Fig. 1). These records were interpreted as indicative of a cooling of ∼5 °C in the earliest Oligocene and Antarctic glaciation in the middle Miocene (Kennett and Shackleton, 1976), but after recovery of lowermost Oligocene ice-rafted material, a consensus developed in the 1990s that Antarctic ice sheets reached sea level in the earliest Oligocene, with Northern hemispheric ice sheets developing millions of years later, in the late Miocene (e.g.,Zachos et al., 2001). The existence of a world with only a south polar ice sheet was seen as in agreement with the hypothesis that the opening of the Southern Ocean gateways led to the development of the Antarctic Circumpolar Current (ACC), reducing meridional heat transport to Antarctica by isolating the continent within a ring of cold water, causing formation of a continental-scale ice sheet (with feedbacks) (e.g., Kennett, 1977).

During the past few years, these views have been questioned. First, it has been difficult to accurately date the opening of Southern Ocean gateways and the initiation of the ACC, recent estimates ranging from middle Eocene (Livermore et al., 2007) to late Oligocene (Lyle et al., 2007). Second, climate modeling results indicate that the change in meridional heat transport associated with ACC onset was insignificant (e.g., Huber and Nof, 2006), and that changes in ocean circulation alone do not lead to Antarctic glaciation (De Conto and Pollard, 2003). The latter concluded that an Antarctic ice sheet could begin to form only when the concentration of greenhouse gases dropped (Fig. 1; Pagani et al., 2005). There is a striking similarity between the modeled results of De Conto and Pollard (2003) and high-resolution data from a Pacific Ocean site (Coxall et al., 2005), both showing an orbitally modulated, two-step Oi-1 event. Additional data from Pacific Ocean sites led to re-examination of the premise that northern-hemisphere ice sheets post-dated the Antarctic ice sheet (Tripati et al., 2005), ocean drilling in the Arctic Ocean provided evidence of ice-rafting in the middle Eocene (Moran et al., 2006), and development of dinoflagellate biostratigraphy dated ice-rafting in the Norwegian-Greenland Sea as middle Eocene (Eldrett et al., 2007; Tripati et al., 2008).

Schouten et al. (p. 147 in this issue) present a new biomarker proxy for land temperatures, by using organic compounds from soil bacteria in the same cores in which middle Eocene ice rafting was recognized (Eldrett et al., 2007), thus providing a novel approach to estimate temperatures on land. The authors arrive at a mean annual temperature of 13–15 °C in the middle to late Eocene, declining by ∼5 °C into the earliest Oligocene, similar to megafossil- and isotope-derived temperatures for coeval sections in the southern hemisphere (14–15 °C for the middle Eocene, declining to ∼10 °C by the end of the Eocene; Dutton et al., 2002; Poole et al., 2005). As Schouten et al. point out, such temperatures might appear high for an environment where we observe ice rafting, but could represent summer rather than annual temperatures.

The article by Pearson et al. (p. 179 in this issue) presents data from Tanzanian sections, a veritable treasure trove of beautifully preserved fossil material. Pearson et al. (2007) documented the tropics remained warm until the end of the Eocene, and that long-term middle Eocene cooling as seen in the deep-sea record (Fig. 1) is limited to high latitudes. Their article in this issue confirms the two-step nature of Oi-1, and suggests that only part of this event reflects ice-volume increase. Interpretation of this event as wholly due to ice-volume increase thus probably is incorrect, and benthic Mg/Ca data are not reliable temperature proxies during changes in carbonate ion concentration (Lear et al., 2004).

Pearson et al. (this issue) also establish that extinctions in shallow-water organisms, as well as in planktic organisms, started ∼0.5 m.y. before the isotope event. In addition, they finally establish that the Eocene/Oligocene boundary, defined by the last occurrence of the planktonic foraminiferal genus Hantkenina, occurs between the two steps of oxygen isotope event Oi-1, so that this isotope event can be used to approximate that boundary.

Where does that leave us in the debate on Eocene/Oligocene global cooling? The cooling in the late Eocene and early Oligocene was truly a global event, affecting organisms from high to low latitudes, thus supporting the view that Cenozoic cooling was caused mainly by decreasing CO2 levels in the atmosphere, with subsequent processes such as ice albedo and weathering feedbacks. Proxy records for atmospheric CO2 values, however, show strong variability in the middle-late Eocene (Pearson and Palmer, 2000; Pagani et al., 2005), and no precise correlation with glaciation events (Fig. 1). We need more detailed and accurate proxy records of past CO2 levels, as well as more geographically diverse records of temperature changes across the late Eocene and Oligocene before we can truly understand when, how, and why the Earth descended into the Icehouse state.