Early Cenozoic "greenhouse" climates have been identified as a high-priority interest of U.S. and International global change programs. Moreover, the ODP Extreme Climates Program Planning Group has also ranked the Paleocene–Eocene as a high-priority objective. Here, we outline questions specific to the nature and causes of these warm episodes that can be addressed by drilling in the southeast Atlantic. We also discuss in greater detail some other aspects of Paleogene paleoceanography that are relevant to this proposal.

The EECO represents the most recent episode of sustained global warmth. For 1–2 m.y. of the early Eocene, global climate was very warm and ice free (Fig. F7). In the oceans, paleontologic and isotopic proxies indicate that the high-latitude seas and bottom waters were as much as 8°C warmer than present (Miller et al., 1987; Shackleton and Boersma, 1981; Zachos et al., 1994; Stott and Kennett, 1990). On land, the biogeographic ranges of subtropical to temperate fauna and flora extended well into polar latitudes (Axelrod, 1984; Estes and Hutchison, 1980; Wolfe, 1980) and polar ice sheets were small or nonexistent.

Several critical issues concerning the EECO need to be addressed. The first concerns the nature of climate variability during this period. At present, not a single marine record details paleoceanographic/climatic variability of this interval on orbital timescales. As a result, we do not know the approximate duration of the EECO, nor how stable climate was during this period. The second issue concerns the origin of the EECO. Empirical and theoretical geochemical studies suggest that greenhouse gas levels were significantly higher, possibly six times the preindustrial level at the peak of the EECO (Berner et al., 1983; Pearson and Palmer, 2000). These estimates are supported to some extent by climate modeling in which the observed EECO meridional thermal gradients could only be attained in simulations with greenhouse gas levels six to eight times those at present (Sloan and Rea, 1996) and by a new boron isotope–based pH record that indicates unusually low levels for the Eocene ocean, consistent with high pCO₂ (Pearson and Palmer, 2000). The third issue concerns the underlying mechanism for greenhouse gas changes: why did CO₂ levels increase in the early Eocene? Were mantle outgassing rates higher than suggested by geochemical models? Regardless, several lines of evidence now indicate that the EECO was indeed a "greenhouse" climate and that other factors such as continental geography and oceanic gateways played a subordinate role in sustaining this extreme global warmth.

In terms of the rate and degree of warming, the PETM is unprecedented in Earth history. Isotope records suggest that at 55 Ma the deep-sea and high-latitude oceans warmed by 4° and 8°C, respectively, in a period of <10 k.y. (Fig. F8). This period of extreme warmth, which lasted <150 k.y., triggered profound changes in global precipitation and continental weathering patterns (e.g., Gibson et al., 1993; Kaiho et al., 1996; Robert and Kennett, 1994). The PETM also affected biota on a global scale, triggering both rapid turnover of benthic and planktonic organisms in the ocean (Kelly et al., 1996; Thomas and Shackleton, 1996; Thomas and Ward, 1990) and a sudden radiation of mammals on land (Clyde and Gingerich, 1998; Koch et al., 1992; Rea et al., 1990).

Several mechanisms have been proposed for the PETM, including massive outgassing associated with rifted margin volcanism (Eldholm and Thomas, 1993) and sudden dissociation of methane hydrates stored on continental shelves and slopes (Dickens et al., 1995, 1997). Both hypotheses were inspired, in part, by marine and terrestrial carbon isotope records that show an abrupt 3‰–4‰ decrease in the ocean/atmosphere carbon reservoirs in <10 k.y. (Bralower et al., 1995, 1997; Kennett and Stott, 1991; Koch et al., 1992, 1995). Such a large and rapid carbon isotope excursion (CIE) requires a sudden injection of a large volume of isotopically depleted carbon into the ocean/atmosphere system. In terms of fluxes and isotopic mass balances, the hydrate dissociation model is clearly more plausible. If estimates are correct, only a fraction of the total reservoir of methane hydrate stored on continental margins is sufficient to generate the observed isotopic excursion. In principle, with warming of deep waters, shelf and slope hydrates could become unstable, triggering a catastrophic release of CH₄ and immediate greenhouse warming.

A massive methane dissociation event should have profound effects on ocean chemistry. Dickens et al., (1997) used a box model to simulate the effects on ocean carbonate chemistry of releasing roughly 1.1 x 10³ gigatons of methane (immediately oxidized to carbon dioxide) directly into the atmosphere. The amount of carbon added to the system was determined from mass balance calculations assuming a ¹³C of –60‰ for bacterially produced methane. This exercise found several notable effects including a dramatic increase in weathering rates on land, a reduction in ocean pH, and a shoaling of the CCD and lysocline, all within several thousands of years (Fig. F9). The effects on the ocean were greatest in areas proximal to sources of deep water where the excess CO₂ reentered the deep ocean via convective processes. The CCD is eventually restored, although not to its original position, as the system appears to initially overcompensate before returning to a steady-state level. This is consistent with the PETM low carbonate or clay layers in deep-sea sites (Fig. F10).

The clathrate and other models can be tested with deep-sea drilling. A series of paleodepth transects in each of the major ocean basins would allow for characterization of the lysocline/CCD changes during this event. The initial shoaling should be expressed as a dissolution interval or hiatus in deeper sites coincident with the benthic foraminifer extinction and carbon isotope excursion. The model-predicted overshoot should be preserved as well. Existing PETM sequences show a reduction in carbonate content to varying degrees coincident with the excursion. However, without vertically offset sites, it is not yet possible to quantify the extent to which the reduction in carbonate content reflects changes in microfossil preservation or the depth range over which the carbonate changes occurred. With a series of well-placed vertical depth transects, it should be possible to separate the effects of preservation from those of production. Walvis Ridge appears to be well suited for this, as rates of sedimentation were fairly high and continuous through the late Paleocene–early Eocene over most of the ridge.

An interesting new development is the emergence of evidence of additional biotic events and CIEs. Faunal and isotope data from Hole 690B (Thomas et al., 2000) point toward other times with conditions potentially similar to but not as extreme as those of the PETM. If this second event has a similar origin as the first, we would expect a similar but smaller-scale response in ocean carbonate chemistry. Moreover, drilling at Blake Nose shows similar repetitions of the climatological and sedimentological conditions of the Paleocene/Eocene (P/E) boundary in the rock record. A long-term gamma ray record from Site 1051 shows a series of major peaks that represent a reduction in the carbonate content and/or increased terrigenous flux augmenting the clay content. Cyclicity of clay fluxes on the timescale of ~2 m.y. most likely reflects on climate cycles, although the precise nature of the climatic changes is unknown. One of the gamma ray maxima coincides with the carbon isotope event and peaks again near the top of Chron C24r (Norris et al., 1998). These cycles persist throughout the early Eocene, indicating that this warm greenhouse period was not stable. The long-term periodicity in these records most likely reflects on a long-period orbital cycle (2 m.y.). The role of orbital forcing in driving these and other early Paleogene climate changes still needs to be evaluated.

The primary transition from "greenhouse" to "ice-house" conditions occurred during the middle Eocene to early Oligocene. Although this encompasses 18 m.y., stable isotopic records reveal a steplike procession, with much of the change occurring in just a few relatively brief steps in the earliest–middle Eocene (~52 Ma) and earliest Oligocene (Fig. F11) (EOGM; ~33.4 Ma) (e.g., Miller et al., 1987; Kennett and Stott, 1991; Zachos et al., 1996). The first event marks the onset of ephemeral glacial activity (Browning et al., 1996; Hambrey et al., 1991), whereas the second and larger of the two steps, EOGM, represents the first appearance of permanent ice sheets on Antarctica (Barrera and Huber, 1991; Zachos et al., 1992). Furthermore, the highest-resolution deep-sea isotope records of this transition indicate that the final transition may have been modulated by orbital forcing (Diester-Haass and Zahn, 1996).

In principle, this climatic transition should have had a dramatic effect on ocean/atmospheric circulation patterns and continental weathering rates and, hence, ocean chemistry. For example, reconstructions of bottom-water isotope patterns hint at a brief pulse of a proto-North Atlantic Deep Water (NADW) coincident with the EOGM (Miller et al., 1991; Zachos et al., 1996). Also, carbonate sediment patterns suggest a significant deepening of the CCD at the Eocene/Oligocene (E/O) boundary (Peterson and Backman, 1990; van Andel, 1975). The latter is consistent with a sudden lowering of sea level and/or an increasing chemical weathering of continents, both of which would increase the flux of dissolved ions to the deep sea. However, the CCD on regional scales is also sensitive to changes in the rates and patterns of carbonate production patterns. A depth transect is required to separate these potential effects and more tightly constrain rates of change.

The Cretaceous–Tertiary (K-T) impact and mass extinction caused a number of long-term changes in oceanic properties. These include (1) a drastic decrease in the organic flux to deep water, as indicated by decreased carbon isotopic gradients (Hsü et al., 1982; D'Hondt et al., 1998; Stott and Kennett, 1989; Zachos et al., 1989) and lower barium accumulation rates, and (2) a drastic decrease in deep-sea carbonate accumulation (D'Hondt and Keller, 1991; Zachos and Arthur, 1986), and, at least in the South Atlantic, enhanced 100-k.y. oscillations in deep-sea sedimentation (D'Hondt et al., 1996). All of these long-term changes were sustained for a million years or more. The final recovery from several of these changes is poorly constrained. For example, although we know that deep-sea carbonate accumulation did not recover for >2 m.y. after the mass extinction (Zachos and Arthur, 1986), the final recovery of deep-sea carbonate accumulation has only been identified at a single Caribbean site (ODP Site 1001), where it appears that carbonate accumulation did not recover for 4 m.y. after the K-T event (D'Hondt et al., 1998). Poor core recovery has precluded successful identification of this final recovery in previously drilled South Atlantic sites. However, the improved recovery made possible by modern ODP techniques should allow its successful identification at the proposed drill sites.

Documenting the timing of these recoveries and their relationships to other paleoceanographic properties would provide a critical test of the coupling between oceans, climate, and biota. The K-T mass extinction effectively changed the state of the global ecosystem. However, there is little reason to believe that any physical consequences of the K-T impact could have lingered for more than a few thousand years. Consequently, the Maastrichtian through Paleocene record of deep-sea sediments provides an ideal opportunity for testing the sensitivity of the global environment to biological disaster. For example, the long-term decrease in deep-sea carbonate accumulation might have changed oceanic alkalinity and atmospheric CO₂ concentrations. If so, the interval of decreased carbonate accumulation should be marked by changes in various paleoceanographic proxies, such as increased foraminiferal preservation and migration of the lysocline or CCD. Similarly, the 100-k.y. oscillations in early Paleocene deep-sea sediments may have resulted from a decreased ability of post-extinction biota to buffer seasonal and Milankovitch-scale climate change (D'Hondt et al., 1996). If so, these oscillations should correspond to similar oscillations in various paleoceanographic proxies, such as carbon isotopic differences between planktonic and benthic foraminifers.

In principle, changes in either the meridional thermal gradient or precipitation patterns can dramatically alter the mode of ocean circulation, thermohaline or otherwise (Broecker, 1997). Warming of polar regions coupled with increased precipitation, for example, would tend to lower the density of high-latitude surface waters, thereby inhibiting sinking. This, in turn, might be balanced by increased convection elsewhere, possibly in subtropical regions where high rates of evaporation raise seawater salinity and density. Given the extreme thermal gradients and precipitation patterns, its likely that deep convection during the PETM, and possibly during the EECO, was not occurring at high latitudes. If true, there should be some obvious evidence for this in deepwater chemical gradients as inferred from stable carbon isotopes and the distribution of carbonates on the seafloor.

A shift in the source of waters bathing Walvis Ridge should be reflected in carbon isotope, carbonate dissolution, and benthic assemblage patterns, as well as in other isotopes with short residence times (i.e., Nd). Several studies have shown that through much of the late Paleocene and early Eocene the most negative deep-ocean carbon isotope values were consistently recorded by benthic foraminifers from the Pacific Ocean (Kennett and Stott, 1990; Corfield and Norris, 1996; Corfield and Cartlidge, 1992; Pak and Miller, 1992). Although this pattern is of the same sign as modern, the gradient is much smaller, indicating that Pacific deep waters were only slightly more aged than Atlantic deep waters. As such, one would predict a similar CCD in the two basins. Low-resolution records of carbonate accumulation on Walvis Ridge for the Cenozoic show features suggestive of large-scale changes in the CCD at the P/E and E/O boundaries, although these "events" are not well defined (Moore, Rabinowitz, et al., 1984).