The ocean circulation in the southern hemisphere and its climatic impacts in the Eocene

Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9 – 28 www.elsevier.com/locate/palaeo

The ocean circulation in the southern hemisphere and its climatic impacts in the Eocene Matthew Huber a,b,*, Doron Nof a

c,d

Earth and Atmospheric Sciences Department, Purdue University, West Lafayette, Indiana, United States b Purdue Climate Change Research Center, West Lafayette, Indiana, United States c Department of Oceanography, Florida State University, Tallahassee, Florida, United States d Geophysical Fluid Dynamics Institute, Florida State University, Tallahassee, Florida, United States Received 15 April 2005; accepted 8 July 2005

Abstract Opening of Southern Ocean gateways may have had a controlling influence on the evolution of global climate. The largest step in the history of Cenozoic climate, from an Eocene bGreenhouseQ toward the modern glaciated state, is correlated with opening of a deep water connection between the Southern Indian and Southern Pacific Oceans. This study synthesizes results from previous work, assesses the possibility that ocean heat transport was much greater than modern during the globally warm Eocene, and explores the sensitivity of polar climate to large changes in ocean heat transport. The magnitudes of these sensitivities are compared with those due to greenhouse gas concentration changes. We find that ocean heat transport was unlikely to have been sufficiently greater than modern during the Eocene to explain the change in the climate. Specifically, in order to have significantly affected Antarctic climate, ocean heat transport changes would have had to change by an order of magnitude—much greater than the largest change produced in our simulations. In coupled climate model simulations, changes in CO2 concentrations(from 1120 to 560 ppm) supported by proxy records, lead to significant Antarctic responses in the absence of ocean heat transport changes. The results show that Antarctic snow accumulation is primarily controlled by changes in summer temperature, not precipitation, and that snow did not remain on Antarctica during summer months in Eocene even at low greenhouse gas concentrations. Greenhouse gas concentration changes and subsequent processes such as ice albedo and weathering feedbacks, were more likely to have been the primary cause of Antarctic glaciation than ocean heat transport changes. D 2005 Elsevier B.V. All rights reserved.

1. Introduction Among the least understood characteristics of climate change are the wild fluctuations of polar climate in the face of relative stability in the Tropics. One of the best illustrations of this behavior is found in the transformation of the Eocene bGreenhouseQ into the Oligo* Corresponding author. Purdue Climate Change Research Center, West Lafayette, Indiana, United States. E-mail address: [email protected] (M. Huber). 0031-0182/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.palaeo.2005.07.037

cene bIcehouseQ. During most of the Eocene, winters in mid-and-high latitude continental interiors were above freezing in the Northern Hemisphere (Greenwood and Wing, 1995; Sloan, 1994; Wolfe, 1994; Markwick, 1994, 1998), polar surface and deep ocean temperatures were ~10 8C warmer-than-modern, and both poles were free of significant permanent ice (Miller et al., 1987; Zachos et al., 1993, 1994). For much of the Oligocene, deep ocean temperatures were cooler, seasonality was more marked, and large ice glaciers existed on Antarctica (Miller et al., 1987; Mackensen and Ehrmann,

10

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

1992; Ivany et al., 2000; Lear et al., 2000; Elderfield, 2000; Zachos et al., 2001). The transition between these two climates involved stepwise cooling, terminated by the abrupt emplacement of a large ice sheet on Antarctica (Shackleton and Kennett, 1975; Miller et al., 1987; Zachos et al., 1996). This represents one of the most fundamental climate changes in Earth history. One possible explanation for Eocene warmth—increased greenhouse gas (GHG) concentrations—is supported by some data (Pearson and Palmer, 2000), but the concentrations necessary to sufficiently warm high latitudes and continental interiors are so great that climate models have difficulty matching tropical sea surface temperature reconstructions (Crowley and Zachos, 2000; Huber and Sloan, 2000; Sloan and Rea, 1996; Shellito et al., 2003). Many GHGs gases exist, and some, like methane, are more effective at trapping outgoing longwave radiation than others (e.g. CO2) on a per molecule basis. Nevertheless, changes in the concentrations of all well-mixed GHGs have the same climate impacts. Of the important GHGs the only one with a proxy for atmospheric concentration is CO2, so consequently we use changes in CO2 as a surrogate for the suite of potential GHG gases. In the experiments described below we vary CO2 well within the range predicted by proxies, with the understanding that important variations in other GHGs, such as methane (Sloan et al., 1992), may also have occurred. For this heuristic reason, explanations for Eocene climate commonly feature the ocean circulation in the leading role of maintaining high latitude warmth in conjunction with relatively cool tropical temperatures (e.g., Covey and Barron, 1998; Rind and Chandler, 1991; Lyle, 1997; Barron et al., 1995; Schmidt and Mysak, 1996 in the Cretaceous). Independent of these attempts to reverse-engineer the functioning of Eocene climate, extensive reorganizations of the ocean circulation during the Eocene have been proposed based on interpretations of climate and circulation proxies (Shackleton and Kennett, 1975; Kennett and Stott, 1991, 1995; Pak and Miller, 1992; Zachos et al., 1993). One strain that emerges from such studies is a dominant role for the tropical oceans as the main source of warm waters both to the deep ocean and to high latitudes. Warm, subtropical currents are postulated to have maintained Antarctic warmth during the Eocene, and that in the Oligocene, opening of a gateway between Australia and Antarctica (the Tasmanian Gateway) thermally isolated Antarctica from these currents, initiating glaciation (Kennett, 1977). Two lines of inquiry have thus led to the conjecture that Eocene ocean circulation and ocean heat transport were fundamentally different than they are today, and

that the transition to the Oligocene represented a step in the direction of modern oceanic conditions (Shackleton and Kennett, 1975; Kennett and Stott, 1991, 1995). In addition to gateway-mediated changes in ocean heat transport (OHT), other factors, such as changes in GHG concentrations and orbital parameters, and sea ice feedbacks, have also been proposed (cf. Zachos et al., 1996). However, the importance of these processes tends to be underplayed in most current discussions of Eocene–Oligocene climate deterioration. Recently, this consensus view has been challenged by DeConto and Pollard (2003) and Huber et al. (2004). We believe that the question, bwhat caused Oligocene cooling?Q may be more profitably approached by asking, bwhat made the Eocene warm, and why did it stop?Q In this study, we investigate the nature of the circulation in the Eocene Southern Ocean and the potential of changes in this circulation to impact climate. The aims of this paper are to: (1) Examine the magnitude of the climatic impacts on Antarctica that might have occurred if there were major changes in OHT and GHG concentrations; (2) Determine the degree to which the assumption of significant OHT change is realistic for the Paleogene; (3) Evaluate the likely changes in Southern Hemisphere climate due to GHG concentration changes. First, we briefly review the role of OHT in the modern day and its hypothesized role in maintaining polar warmth in the Eocene. Then we address the three main foci in turn. We conclude by summarizing the implications of the relative importance of ocean circulation changes and other possible climate forcing mechanisms in light of our results and those of other complementary studies. 2. Ocean heat transport 2.1. Implications of modern observations Even presently—when observations are best—the importance of OHT in climate is equivocal. The difficulty of estimating OHT is reflected in the spread of estimates, with error bars at high latitudes of around a factor of 2 (Ganachaud and Wunsch, 2000). Nevertheless, all present estimates make it increasingly apparent that oceanic heat transport is important mostly in cooling the Tropics, as such, heat transport makes very small contributions poleward of 308 latitude (Trenberth

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

and Caron, 2001). In fact, in present estimates, less than 5% of the total amount of heat transported past 608 North or South is attributable to the ocean (Peixoto and Oort, 1992; Trenberth and Caron, 2001). Nevertheless, ample evidence exists that even changes in this relatively small value can influence high latitude climates via strong ice albedo feedbacks (Seager et al., 2002). One of the most tangible, and most-cited example of the importance of OHT in modern climate is the ameliorating effect of the Gulf Stream and the North Atlantic Drift on Northern European winter temperatures. Interestingly, even the importance of OHT in this paradigmatic example is in question. Seager et al. (2002) suggest that most of this winter warming is driven by release of heat stored seasonally in the North Atlantic, not the heat transported by the bconveyor beltQ, and also that planetary waves set up by the Rockies play a crucial role in creating east–west temperature contrasts across the North Atlantic. While it is tempting to assume that high latitude climate may be insensitive to the OHT, it is likely that perturbations to the OHT may have a large, though perhaps indirect effect on high latitudes (by changing tropical temperatures, and by altering the amount of heat available for atmospheric transport). The effects of OHT changes simulated with climate models include local effects as large as 88–108 and regional effects over the ocean of ~4 8C (Manabe and Stouffer, 1998). Thus there is ample evidence for oceanic control on regional climate features. Furthermore, strong fluctuations in both the North Atlantic temperature and atmospheric temperature during the past 125,000 years are suggestive of an important role for the OHT and its alternating structure (see e.g., Bard, 2002; Broecker, 2003). Because of these considerations, it is difficult to a priori ascribe a leading role (in polar climate change) to high latitude OHT. Other processes such as changes in atmospheric heat transport or in the local (polar) radiative budget (e.g., albedo feedbacks) may also be important. There is, furthermore, a strong line of theoretical and modeling evidence that backs up the supposition that the Southern Ocean plays a dominant role in modern day ocean circulation (Toggweiler and Samuels, 1995) and that past changes in this region may have been especially important (Toggweiler and Bjornsson, 2000). Consequently, it is worth exploring whether Southern Ocean changes might have been important in the Paleogene. The important role of the Southern Ocean is due to the meridional overturning cell (MOC), which originates in the surface water of the Southern Ocean (Nof, 2003), flows northward as a surface current, and sinks

11

to the bottom or mid-depth in the North Atlantic and the Pacific-Indian Ocean system. This water then returns southward (below the level of the peaks of ocean bottom’s topography) to the Southern Ocean where it upwells and returns to its origination point (De Boer and Nof, 2005). Whether such an MOC warms or cools the Antarctic depends on the difference between the temperature of the water entering the Southern Ocean from below and the temperature of the surface water leaving Southern Ocean. Normally, the bottom water will be cooler than the surface water implying that the MOC cools the Southern Ocean by bstealingQ its heat and transferring it northward. We shall see, however, that during the period prior to and after Australia’s detachment the mean vertical temperature gradient along the northern boundary of the Southern Ocean was minimal implying that the role of the MOC was minimal as well. 2.2. Proposed role of Paleogene ocean circulation change It is a fundamental tenet of paleoceanography that ocean gateways changes—such as the opening of Drake Passage and closing of the Panamanian Gateway—have acted as drivers of critical transitions in the evolution of global climate (Hay, 1996; Crowley, 1998; Lawver and Gahagan, 1998). This belief finds its support in the approximate correlation between major gateway changes and climatic change (Zachos et al., 2001). Recent improvements in age estimates by Stickley et al. (2004) for the opening of the gateway between Australia and Antarctica—previously conjectured to exactly coincide with the initiation of Antarctic glaciation (Murphy and Kennett, 1986)—suggest over a two million-year lag between gateway opening and glaciation. The revised chronology fundamentally challenges the correlation that has largely driven these arguments. We do not focus on the timing of gateway opening in this paper, rather, we focus on the mechanisms by which such opening might affect climate and the relative importance of these mechanisms over other climate change mechanisms. The mechanism that links gateways to climate change is usually presumed to be MOC and OHT modifications induced by shifts in ocean currents (we term the OHT-gateway mechanism). The impacts that such ocean current changes have had on ocean productivity and sediment burial and hence global inventories of atmospheric GHGs have received less attention in comparison to the OHT-gateway mechanism. In part, this is because records of the atmospheric GHG con-

12

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

centrations have tended to have poor temporal resolution and large uncertainties (Pearson and Palmer, 1999, 2000; Royer et al., 2001) associated with them, and important GHGs, such as methane, have no proxies. A number of modeling studies have demonstrated that the ocean’s overturning circulation, heat transport and resulting temperature distribution are altered by changes in ocean gateways and this has provided a justification for the OHT-gateway mechanism. The magnitude of the OHT and surface temperature changes produced by these simulations has not received as much attention as the changes themselves, however. In studies in which a simple representation of coupling of ocean–atmosphere exchanges have been included, opening and closing of Southern Ocean gateways results in Southern Ocean surface temperature changes from a low value of 0.88 (Mikolajewicz et al., 1993) to a high value of ~3.58 (Toggweiler and Bjornsson, 2000; Nong et al., 2000; Najjar et al., 2002; Sijp and England, 2004). The relatively small simulated changes (~2.78 range) led Mikolajewicz et al. (1993) and Najjar et al. (2002) to conclude that the major Cenozoic cooling may not have been due to ocean gateway changes. As noted in Toggweiler and Bjornsson (2000), creation of the Antarctic Circumpolar Current (ACC) leads to a cooling of the Southern Hemisphere, and yet a warming of the Northern Hemisphere, because, as described in Section 2.1, the flows associated with the ACC bstealQ heat across the equator. Since the opening of polar gateways essentially leads to a rearrangement of heat between hemispheres, it is difficult to understand how closing such gateways could cause Eocene warmth in both hemispheres and lead to temperatures so much warmer than today’s. None of those studies incorporated a full treatment of ocean, atmosphere, and sea ice dynamics however, and it is only with the inclusion of all of these processes that a reliable estimate of the sensitivity of polar climate to gateway changes can be made. As described in more depth below, and in Huber and Sloan (2001), when a full treatment is included, the resulting change within the Southern Ocean is commensurate with the results of the aforementioned studies. 2.3. Proxies for ocean heat transport A common assumption in paleoceanography is that evidence from proxies for warm sea surface temperatures at high latitudes is unequivocal evidence for increased OHT, or more generally, that knowledge of ocean temperatures provides much information about OHT. This assumption is weak because it is always

possible that ocean temperatures are warm because little heat is lost to the atmosphere. To see this, we note that OHT is calculated by integrating the product of poleward and equatorward ocean transport and temperatures throughout the ocean to calculate the integral of the ocean heat transport divergence. We can write this in a simple form as, F = VDT, where F is the poleward heat flux, V is the meridional transport rate (i.e., mass flux which may or may not be identical to the MOC), and DT, is the temperature gradient. From this relation we see that decreasing temperature gradients should be accompanied by decreasing heat transport (Sloan et al., 1995) in the absence of other changes (see Lyle, 1997 for one such change). Another method for calculating ocean heat transport is to integrate the net vertical energy flux through the surface of the ocean to obtain the so-called implied OHT (Peixoto and Oort, 1992; Huber and Sloan, 1999; Trenberth and Caron, 2001; Huber et al., 2003). In steady state, the heat lost or gained through the sea surface (the net sensible, latent heat fluxes and incoming and outgoing radiation at the ocean surface) must be balanced exactly by heat transported via ocean currents. Consider one thought experiment, we call the bcovered hot-tub paradoxQ, that illustrates a family of problems that proxy interpretations have in providing insights into bgreenhouseQ climates. In an ocean covered with an insulating, opaque and water-proof blidQ (like the cover on a hot-tub) no heat is transported by the ocean because no heat is lost or gained through the surface, regardless of the details of the internal temperature and velocity distribution. In a more realistic example, imagine an atmosphere so thick with GHGs that high latitude outgoing longwave radiation is much diminished from modern values, in which case the atmosphere draws less heat from the ocean. In this extreme case, let us even assume that the ocean currents transport water all the way to the pole. Nevertheless, because of the strong greenhouse effect, no heat is lost from the ocean to be radiated by the atmosphere to space. In this idealized case, the ocean surface temperatures are as warm at the poles as in low latitudes and might be interpreted as an extreme example of very strong OHT, when, in fact, the ocean has transported no heat. A related assumption, the concept that deep water formation in the tropics ( i.e. warm, salty, deep water (WSDW)) provides a mechanism for increased OHT (Brass et al., 1982), is weak for similar reasons. We note that, in the absence of further information, it makes no difference to the question of OHT values, where (or

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

at what temperature) deep water forms; it is the degree to which the water is cooled and warmed along its circulation path that pertains to OHT. Considering the dhot-tubT concept again, if WSDW with a temperature of 188 forms in the tropics/subtropics and flows to the poles, where it does not cool, no heat is transported, regardless of the fact that the polar waters are quite warm. Paleoenvironmental proxies must provide quantitative estimates of either of two quantities to complete an estimate of OHT. Either the velocity structure must be constrained (in addition to the temperature distribution) or alternatively, a proxy for the net surface energy flux must be found. It may be possible to construct a proxy for the latter because the diabatic processes that make up this energy flux have much in common with the processes that govern the fractionation of oxygen isotopes, but, to our knowledge, this has not been attempted. Thus the most likely source of information on changes in OHT will come from combining proxies for temperature and velocity. Interestingly, there have been few studies that have directly constrained the direction and magnitudes of ocean currents in the Eocene with an important and early exception being that of Kennett et al. (1972) who showed the sudden influence of a northward flowing deep/bottom current in the Southwest corner of the Pacific in the early Oligocene. Recent work using proxies for sites deep water formation (Thomas, 2004) are a step in the right direction, but unless they quantify changes in the rates of deep water formation (i.e. current velocities) as well, they do not provide key constraints on OHT. Without a substantial improvement in our understanding of the velocity structure (in the upper several kilometers of the ocean in the Paleogene), there is little that can be determined about OHT, since existing proxy records constrain mostly temperature. In the Plio–Pliestocene such hurdles have been partially overcome (e.g. Ravelo and Andreasen, 2000), because of the spatial greater coverage afforded by working in more recent times for which appropriate sediment records are more available. More geographic coverage and additional circulation proxies will be required to improve our understanding of the Paleogene. In fact, where attempts have been made to infer current directions and magnitudes from temperature distributions, the approach has been to invert the above relation and make suppositions about V from DT. This, of course, can only be done by assuming a value of F. In this way, many studies that attempt to prove the validity of the OHT-gateway mechanism tend to run into issues of circularity. Reconstructions of

13

biogeographic, and carbon isotopic, benthic oxygenation, or trace element patterns can help to unravel these difficult issues, and often can establish at least the directions of ocean currents (see McGowran et al., 1997; Carter et al., 1996; Huber et al., 2004, for some attempts), but the magnitude of the currents is usually subject to substantial uncertainty. 2.4. Sensitivity of terrestrial climate in previous studies A lesser-noted result of studies in which OHT is altered is that such changes alone have little impact on temperatures within continental interiors (Barron et al., 1995; Sloan et al., 1995), and consequently this mechanism does not appear to explain, by itself, Eocene extratropical continental winter warmth. Even assuming an infinite efficient ocean heat transport and infinite ocean heat capacity—sufficient to eliminate the ocean meridional temperature gradient—does not prevent continental winter temperatures from falling below freezing (Schneider et al., 1985) unless GHG concentrations are much higher than modern values. A parsimonious approach to the Eocene climate problem assumes that the major climate transitions in the Cenozoic may have been due to GHG changes and not due to OHT changes, unless there is either clear evidence (from the paleoclimate record or from theory or models) that the opposite is the case. 3. Terrestrial impacts of implied OHT and GHG changes 3.1. Fixed sea surface temperature experiments It is useful to ascertain the sensitivity of Southern Hemisphere terrestrial climate to a large, even implausible, range of forcings. By mapping out the predicted responses on Antarctica to extreme forcings we can evaluate the degree to which different mechanisms for past climate change in the region are likely or even possible. To do this, we have opted to explore the range of sea surface temperature values that have been proposed for the Eocene from such sources as Zachos et al. (1994) and Pearson et al. (2001). These reconstructions have an approximate spread for the Eocene of approximately 98 in the tropics (between 348 and 25 8C) and 48 (between 108 and 6 8C) at high latitudes. This range approximates the spread between different reconstructions for the same interval as well as the temporal variability over the Eocene. Atmosphere and ocean dynamics are sensitive to temperature gradients, hence we take the SST configurations that correspond to the

14

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

highest and lowest temperature gradients, to explore the implications of changes between these two extremes in this section. By driving an atmospheric (A)GCM with specified SST distributions, as shown in Fig. 1, we can calculate the effects on the atmosphere and on land of these changes. Changes in the OHT, in equilibrium with, or implied by the specified SSTs can also be calculated, thus the fixed SST experiments carried out here can be considered surrogates for experiments in which OHT is varied. If polar temperatures were 10 8C and tropical temperatures were 25 8C, Huber et al. (2003) calculate that the implied OHT in equilibrium with those temperatures is substantially greater than modern at all latitudes. The specific value of the increase depends on which latitude is chosen to make the comparison, but at most latitudes approximately a three-fold OHT increase over modern values is required to maintain such a small equator-to-pole temperature gradient. On the other hand, if tropical temperatures are 34 8C and polar temperatures are 6 8C, OHT does not need to be significantly elevated over modern values. LOGRAD2x and HIGRAD2x are the names for the experiments corresponding to the small and large specified SST gradient cases, respectively. The LOGRAD2x and HIGRAD2x simulations are calculated using CO2 concentrations fixed at 560 ppm. We also alter the concentration of CO2 used in the simulations to compare the relative effects as the CO2 concentration is increased to 1120 ppm, while holding the SSTs constant

with the same values as in the LOGRAD2x experiment; this is denoted the LOGRAD4x experiment. These experiments explicitly do not include mutual interaction between the ocean and atmosphere, therefore the GHG experiments provide an indication of the effect of GHG changes in the absence of feedbacks from the ocean. The comparison is intended as a sensitivity test to investigate the relative importance of changes in SSTs, and hence implied OHT, (by comparing LOGRAD2x to HIGRAD2x) with GHG-induced changes (by comparing LOGRAD2x to LOGRAD4x) on the temperature, precipitation and snow thickness in the Antarctic interior. We use the atmospheric general circulation model (GCM) Community Climate Model version 3.6 (CCM) developed by the National Center for Atmospheric Research (NCAR). This model has been extensively described and used in a variety of modern (Kiehl et al., 1998; Hack et al., 1998; Hurrell et al., 1998; Raphael, 1998) and paleo-contexts, both in stand-alone mode and coupled to other components (Huber and Sloan, 2001; Otto-Bliesner et al., 2002; Shellito et al., 2003; Huber and Caballero, 2003). In these simulations, we use CCM with a horizontal spectral resolution of T31 (~3.758  3.758), 18 vertical levels, and with the Eocene topographic, bathymetric, SST, and vegetation boundary conditions described in Huber et al. (2003), Sloan et al. (2001) and Sewall et al. (2000). Results shown are the average of the last ten years of equilibrated runs and significance of results has been determined using a students-t test. Only results significant at 99% level or above are discussed. 3.2. Mean annual temperature and precipitation

Fig. 1. Surface temperature averaged zonally and annually from the 5 cases considered in this study. In black are the three fixed sea surface temperature cases, HIGRAD2x (solid line), and LOGRAD2x (broken line) and LOGRAD4x (indiscernible behind LOGRAD2x). In grey are the COUPLED2x (thick solid line) and COUPLED4x (thin dashed line) cases.

Comparing LOGRAD2x to LOGRAD4x in Fig. 2, changes in CO2 concentrations have a limited effect on mean annual surface temperature (b28) over a limited area. Comparing LOGRAD2x to HIGRAD2x shows that changes in SSTs and, hence, implied OHT have an effect that encompasses most of Antarctica, but is less than a 38 impact for an imposed factor of three change in implied OHT. Variations in precipitation have also been conjectured to be a critical driver of Antarctic glacial initiation (Bartek et al., 1992) and have been linked to the substantial record of changes in weathering in the coastal geologic record of the Southern Ocean (Exon et al., 2002). As shown in Fig. 3, alterations in CO2 concentration that occur in the absence of SST changes have no significant impact on mean annual precipitation patterns in the Southern Hemisphere.

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

15

Fig. 2. Mean annual surface temperature derived from specified SST AGCM experiments. Cases as described in the text are indicated on figure labels (top) with contours at 255, 260, 265, 270, 273, 275, 280, 283, 285, 290, 295 K. Values less than 273 are stippled. Differences between cases are shown on the bottom of the figure with contours at 3, 1, 0, 1, 3, K. Note: Specify only those contours used values less than 3 are stippled and greater than 3 are crosshatched.

Comparing the LOGRAD2x and HIGRAD2x cases reveals that the high latitude sea surface warming associated with the LOGRAD2x case drives significant (0.5 to 1.0 m/yr) increases in precipitation in the easternmost Australo-Antarctic Gulf and through other large portions of the Southern Ocean. Decreases in precipitation occur over Australia. There is, however, no significant impact on precipitation values anywhere in Antarctica. 3.3. Snow accumulation Interestingly, although substantial snow accumulates on Antarctica in the winter (~15 cm; Fig. 4), snow does not persist through the summer anywhere in Antarctica in any of our model simulations. Hence, initiation of Antarctic glaciation is not possible in Eocene, based on the parameters of our models. If we were to use a much higher resolution model capable of resolving the full relief of Antarctic moun-

tain ranges (e.g. DeConto and Pollard, 2003), then we might find that some winter snowfall persists yearround in high mountains areas. The crucial control of summer temperatures on snow accumulation in this simulation is consistent with models (DeConto and Pollard, 2003) and data (Coxall et al., 2005) that emphasize the importance of cool summer orbits for Antarctic initiation. Changes in CO2 concentrations in our models have no significant impact on winter snow accumulation. The SST and implied OHT differences between LOGRAD2x and HIGRAD2x lead to an interesting pattern of snowfall with greater sensitivity along the Transantarctic Mountains and in the coastal parts Marie Byrd Land, and little change within the interior. 3.4. Summary of sensitivity Within the artificial restrictions imposed by our experimental design, the results indicate that Antarc-

16

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

Fig. 3. Mean daily precipitation (mm/day) produced by specific SST sea surface temperature AGCM experiments. Cases as described in the text are indicated on figure labels (top) with contours at .3, .5, .7, 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5, 6, 7, 8 mm/day. Differences between cases are shown on the bottom with contours at 4, 3, 2, 1.5, 1, .5, .3, 0, .3, .5, 1, 1.5, 2, 3, 4 mm/day. Values less than .5 mm/day are stippled and greater than .3 mm/day are crosshatched.

tica is relatively insensitive to changes in CO2 concentrations in the absence of feedbacks from the ocean surface and sea ice. In other words, if GHG changes did not also warm the ocean, their impact within continental interiors would be 38 or less for a doubling of CO2. When modifications are made to the SST pattern that encompasses a 3-fold range of implied OHT terrestrial temperature sensitivity is ~28–38. In none of the experiments does precipitation change significantly. The direct response of snow accumulation due to CO2 changes is insignificant, whereas SST perturbations generate alterations in winter snow accumulation in a thin band along the Transantarctic Mountains. The cool mean annual temperature and the presence of continent-wide below freezing temperatures (not shown) in the LOGRAD2x and HIGRAD2x simulations differ from most interpretations of this region’s climate during the Eocene. In the next section we assess the likelihood that changes of the order of a factor of 3 in OHT occurred during the Eocene–Oligocene transition.

4. Theory and simple ocean models Studies using ocean (O)GCMs are susceptible to many uncertainties, especially when past climates are the focus (e.g., Barron and Peterson, 1991; Bice et al., 1997; Brady et al., 1998), thus we complement the coupled ocean-atmosphere-sea ice GCM study described in Section 5 with simpler methods to elucidate aspects of the ocean circulation. These simple methods take advantage of our advanced understanding of the horizontal, gyre circulations in the ocean, and the vertical, meridional overturning circulation. 4.1. Wind-driven gyre theory The Sverdrup transport—the gyre circulations driven by the curl of the wind stress—corresponds to the vertically averaged (barotropic) mode of the ocean general circulation (Pedlosky, 1996). For a given wind field it is possible to estimate the strength, shape, and disposition of gyre circulations as well as the western bound-

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

17

Fig. 4. Winter season (June, July, August) means of snow accumulation produced by specified SST AGCM experiments. Cases as described in the text are indicated on figure labels (top) with contours at 1, 3, 5, 7, 10, 15 cm. Anomalies between cases are shown on the bottom with contours at 10, 7, 5, 3, 0, 3, 5, 7, 10. Values less than 5 cm are stippled and values greater than 3 cm are crosshatched.

ary currents in equilibrium with them. With further assumptions, these calculations allow testable predictions to be made about current strengths, circulation trajectories, and ocean heat transport (Huber and Sloan, 1999; Klinger and Marotzke, 2000; Liu and Philander, 1995; Held, 2001). A basic characterization of the wind-driven ocean circulation may be obtained via the Ekman and Sverdrup relations without formulating a numerical ocean model. This approach has many limitations, including a lack of representation of important nonlinear dynamical processes as well as lack of interaction between the wind-driven ocean circulation and the atmospheric circulation. Nevertheless, in modern oceanographic studies, the Sverdrup and western boundary currents relations provide a clear constraint on the horizontal and vertical circulations produced by an OGCM driven explicitly by AGCM wind fields, and a first approximation to the behavior of a coupled GCM (Danabasoglu, 1998; England et al., 1992). In a paleoclimate context, these theoretical tools have proven useful for making testable

predictions of patterns in the paleoceanographic proxy data record from AGCM output. We estimate winddriven ocean gyre transport in equilibrium with the AGCM-derived winds by calculating the Sverdrup transport driven by the wind stress curl. Danabasoglu (1998) demonstrated the utility of using Sverdrup calculations derived from an uncoupled AGCM as a predictor of the ocean’s barotropic response in a fully coupled model. There are more advanced methods for calculating the wind-driven horizontal volume transports that relax some of the constraints of the classic Sverdrup calculations or allow for more realistic dynamics (e.g., Godfrey, 1989 and see a comparison of methods in Nof, 2000a); nevertheless, as argued above, the fundamental aspects of the wind-driven gyre circulation are captured by the methods used here. A comparison of predictions using this simple method and the barotropic mode of a fully coupled climate model simulation of the Eocene clearly show the strengths of the Sverdrup calculation (Huber et al., 2004).

18

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

The relative proportion of competing influences of changes in basin size and changes in wind stress are teased apart more fully in Huber (2001), but briefly, the results shown in Fig. 5 indicate that: i. Changes in the basin configuration tend to lead to the formation of larger gyres in the Pacific and

smaller gyres in the Atlantic. The increased size of the gyres due to changed land-sea distributions creates the potential for stronger flows in the Pacific. ii. On the other hand, weaker wind stress curl caused by reduced meridional temperature gradients reduces transport relative to what it would be with modern wind stress patterns. iii. There is little tendency for expansion of the poleward extent of the subtropical gyres regardless of the temperatures that those water masses would have. iv. An important distinction needs to be made that the term subpolar or subtropical gyre refers to the flow direction (clockwise or counterclockwise) and not to its temperature. 4.2. Changes in the overturning circulation

Fig. 5. Sverdrup transport calculated as described in the text, for the HIGRAD2x (a), LOGRAD2x (b), and LOGRAD4x (c) cases. Solid contours indicate counterclockwise gyres, dashed contours indicate clockwise gyres, and the contours are spaced every 10 Sverdrups.

Using a process-oriented layer-and-a-half numerical model we show how the meridional overturning circulation of the ocean (corresponding to the north–south and vertical component of the ocean bconveyor beltQ) changes as Australia is detached from Antarctica. The methodology behind these numerical experiments is the same as that in Nof (2000b, 2002, 2003). The main point is that a model (with the Southern Ocean geography) whose frictional forces are primarily along western boundary currents (WBC) at lower latitudes will not reach a steady state without sources and sinks, as a meridional transport must be allowed. That is to say, as shown by De Boer and Nof (2005), the total meridional transport is determined solely by the wind. The role of buoyancy forcing (not explicitly presented in these runs) is merely to determine the stratification and the exact regions where sinking and upwelling occur. Here, the MOC is determined by the wind field alone but the OHT does depend on the thermodynamics. The above situation is true only when the Drake Passage is open so that the sea level is continuous along a latitudinal belt running along the southern tips of the continents. Integration of the linearized momentum equation along this latitudinal belt shows that the meridional transport is determined solely by the wind, i.e., as just mentioned, thermohaline processes determine where the upper meter will be converted to lower water but the amount is determined by the wind. When the Drake Passage is closed, the sea level is no longer (zonally) continuous and, consequently, the above integration cannot be done. This implies that, in the closed passage case, both the mass flux and the location of upper-to-lower-water conversion are determined by the

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

thermodynamics as well as the wind. This case is so complicated that the advantage of using process-oriented experiments (of the kind used here) is in doubt. It turns out, however, that a partially open passage behaves very much the same as a completely open passage (even if the sill blocks 90% of the upper layer). Hence, we shall focus on an open and partially open Drake Passage. Instead of using complicated models, we used a simple model that does not contain complicated geography. Specifically, we considered a layer-and-a-half application of the breduced gravityQ version of the isopycnic model developed by Bleck and Boudra (1981, 1986), and later improved by Bleck and Smith (1990). The model is virtually non-diffusive and conserves mass. For these reasons, the model is the most

19

suitable model for our problem. This numerical model has no thermodynamics, nor does it have enough layers to accurately represent Antarctic Circumpolar Current (ACC) dynamics. Neither of the two is essential at this stage and, in order to represent the overturning, we introduced sources in the south and sinks in the north. The mass flux associated with these sources and sinks is not a priori specified but all sources and sinks are required to handle an identical mount of water (Fig. 6). The model has no explicit thermohaline processes in it, but the processes are represented in an implicit manner. This is because, as long as the Drake Passage is open, the meridional transport (MOC) is determined primarily by the wind. We performed a total of 19 runs for the attached Australia scenario with the length of the continent vary-

Fig. 6. A schematic of numerical models for simulating meridonial overturning of ocean currents before (lower model) and after (upper model) the separation of Australia from Antarctica. The reference simulation described in Section 4 corresponds to the state before separation.

20

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

ing from zero to the case where Australia’s tip is just north of the equator. The mass flux exiting the three sources and entering the three sinks represents the MOC. It was not specified a priori. Rather, we let the wind blow, measured the established transport and then required the sinks and sources to accommodate the measured amount. The procedure was then repeated continuously and the sources’ mass flux was continuously adjusted until a steady state was reached. The modelpredicted meridional overturning transports are shown in Fig. 7. We see that the MOC with an attached Australia is

in general weaker than that with an open Tasmanian Gateway and the MOC weakens as the northern tip of Australia moves equatorward. Fig. 7 also shows that the results are not very sensitive to the presence of a sill in the Drake Passage even if the sill protrudes well into the upper layer (reaching a depth of 50 meters or less). Note that the sill was implemented using the Nof and Van Gorder (2003) parameterization, i.e., by merely specifying the depth above the sill (at the gap) in the continuity equation. While details of the above prediction may be sensitive to the specific formulation of this model, the fact that the overall change associated with detachment of Australia is relatively small (a factor of two), is probably a robust feature. 4.3. Summary of theory and simple models Ocean gyres in the Eocene were likely to be very much like they are today with the exception that the current system that now makes up the ACC was a series of vigorous subpolar gyres arrayed around Antarctica. Importantly, the subpolar gyre in the South Pacific must have transported water northward along the East Coast of Australia rather than southward as required by some hypotheses for maintaining Antarctic warmth. This current system is also produced by the recent, and completely independent simulations of Omta and Dijkstra (2003) and Huber et al. (2004), which we believe is strong evidence of the validity of our approach. The existence of such subpolar gyres is supported by many proxies from the Cretaceous (Street and Brown, 2000) through the middle Paleogene (Wei and Wise, 1990; McGowran et al., 1997; Oleinik, 2001; Craig, 2002). Of course, there is nothing about the direction of the Fig. 7. In (A) we show simulations with a detached Australia and attached Australia with varying length (see Fig. 6). Solid circles indicate the experiment number. The solid horizontal line is the analytical calculation based on the integration of the linearized equations along the zonal contour connecting the southernmost tips of the continents [i.e., line BE in Fig. 6 (top), Nof, 2003]. Note that the MOC of the attached Australia is smaller than the present day MOC. However, at the most, the effect of detachment is to increase the MOC by a factor of approximately two (compare exp. 1 and exp. 19). Physical constants: gV = 0.015 m s 2; H = 600 m; b = 11.5  10 11 m 1 s 1; m = 16,000 m2 s 1; K = 20  10 7 s 1; f max = 1.518  10 4 s 1; R d = 37.5 km; Dx = D5 = 15 km; Dt = 360 s. In (B) the nondimensional transport as a function of a partial Drake Passage closure for the detached Australia case (top) and an attached Australia (bottom). Note that, in the attached Australia case (i.e., the left-most triangle), even a sill occupying 97% of the upper layer thickness makes very little difference to the transport. In the detached case, transports with sills occupying more than 90% of the undisturbed upper layer thickness deviate significantly from the no-sill values. Overall, the effect of the sill is very limited in our simulations.

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

gyres that determines the amount of heat they transport, and vigorous gyres were capable of significant OHT. In the Southern Hemisphere we expect the warming would have been along the eastern boundaries of the oceans, not the western boundaries (analogous to the warmth of coastal Northern Europe and Northwest North America today, which of course lies along the eastern side of the ocean). The behavior of the meridional overturning circulation as Australia moved away from Antarctica is somewhat more surprising. While there are many grounds for criticizing this estimate as simplistic, it demonstrates at the very least, that there are no strong theoretical or conceptual grounds for thinking that OHT by the MOC was greater when Australia was connected to Antarctica. What is responsible for this lack of sensitivity? As first made clear by Toggweiler and Samuels (1995), wind fields in the Southern Ocean have strong controlling influence on the strength of the MOC in general.

21

Given an Eocene temperature gradient (DT) 50% or less than present day values, and a meridional overturning circulation (V) approximately 100% to 50% of today’s values, we expect from these very simple considerations that ocean heat transport was less than half of what it is today. 5. Fully coupled GCM simulations: OHT and GHG sensitivity 5.1. Coupled experiments The previous sections have focused on simple mechanistic models and theory. Here we focus on the consequences of considering a fully interactive, dynamical, coupled ocean-atmosphere-sea ice-terrestrial model, NCAR’s Community Climate System Model (CCSM 1.4). Using the methods (including acceleration techniques) and boundary conditions described in Huber et al.

Fig. 8. Annual mean surface temperature (top) and daily precipitation rate (bottom). The two fully coupled cases as described in the text are shown on the left and the differences of the two cases are shown on the right. In addition to the model results, proxy reconstructions of surface temperature from a variety of sources (Zachos et al., 1994; Buening et al., 1998; Dutton et al., 2002; Shackleton and Kennett, 1975; Kamp et al., 1990) are superimposed (in ovals) on results from the COUPLED4x case (upper left). Paleolocations of these records are based on the original references and on Nelson and Cooke (2001). Conventions as in Figs. 2 and 3.

22

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

(2003), Huber and Sloan (2001), and Huber and Caballero (2003), we have produced a suite of simulations of Eocene climate. Here, we compare two Eocene simulations, one integrated with 1120 ppm CO2 (COUPLED4x), and the other integrated with 560 ppm CO2 (COUPLED2x; Fig. 8). The latter simulation is an extension of the Huber and Sloan (2001) case with an additional 4000 accelerated and N 200 unaccelerated years and the climate and circulation is essentially identical to that achieved at the end of the much shorter Huber and Sloan (2001) integration, demonstrating the ability of such methods to reach an equilibrium solution. The meridional overturning streamfunction in the COUPLED2x case is similar to that described in Huber and Sloan (2001), i.e. the southern ocean overturning is weaker than modern, in agreement with the results in Section 4. As more fully described in Huber and Caballero (2003), the COUPLED4x case was integrated to equilibrium using a different acceleration technique and has equilibrated on a climate and circulation very similar to the COUPLED2x case, but substantially warmer at high latitudes and in the deep ocean. The overturning in the COUPLED4x case is similar to that in the COUPLED2x case (not shown). Heat transport by the ocean in these simulations is close to modern values as described in Huber et al. (2003, 2004). The important change is an increase of ~.6 PW of heat transport (out of a maximum of ~2.0 PW) into the Southern Hemisphere and a commensurate decrease in the Northern Hemisphere in the Eocene ocean heat transport results relative to modern (consistent with the predictions of Toggweiler and Bjornsson (2000)). Results are averaged over the last ten simulated years of the COUPLED4x case and the last twenty simulated years of the COUPLED2x case and significance of results has been determined using a students-t test (b1%). 5.2. Mean annual temperature and precipitation In fully coupled model simulations, surface temperatures and OHT can change as CO2 concentrations change. As shown in Fig. 8, mean annual temperature increases everywhere roughly equally (28–38) as CO2 changes from 560 to 1120 ppm. In COUPLED2x, only the narrow fringe around Antarctica is above freezing in annual average. In COUPLED4x, this fringe is approximately one model grid cell wider and encompasses Marie Byrd Land and the Antarctic Peninsula. A comparison with surface temperatures compiled from proxy reconstructions (Fig. 8) reveals that the COUPLED4x simulation compares favorably with the data. Precipitation rates in both cases are quite high (Fig. 8). Ant-

arctic coastal values of ~1.9 mm/day are equivalent to ~.7 m per year, and values in New Zealand and Southern Australia are up to twice that. Comparing the difference in precipitation between COUPLED2x and COUPLED4x reveals that precipitation’s response to GHG changes is much more localized than the temperature response, although it has a significant magnitude (~0.1 m per year) and bears a spatial pattern to the response seen to fixed SST changes in Section 3.2. 5.3. Snow accumulation Snow does not persist perennially in any of the coupled simulations. The mechanism for preventing glaciation—as it was in the fixed SST experiments— is the warmth of Southern Hemisphere summer, which is sufficient in this model to melt all snow. Again, we examine the sensitivity of winter season snow accumulation to forcing in Fig. 9. Interestingly, the pattern of snow accumulation is similar to that previously in Section 3.3—warming of the coast of Antarctica associated with the COUPLED4x case leads to a decrease in snow depth along the Transantarctic mountain range, in both West and East Antarctica, but not within the center of the continent, compared to the COUPLED2x case. Interestingly, the warmer case has slightly more winter snow accumulation deep within Antarctic. 5.4. Summary of coupled experiments In these simulations, changes in GHG concentrations have direct radiative impacts on Antarctic terrestrial climate, as well as indirect effects engendered by the remote responses to SST changes. By comparing to the results in Section 3 we can tease apart the degree to which the differences in Antarctic surface temperature between COUPLED2x and COUPLED4x are due to the direct radiative impacts of GHGs versus driven indirectly by the remote (teleconnected) impacts of ocean temperature changes. Section 3 showed that changing GHG concentrations in the absence of accompanying SST changes had little impact on Antarctic temperatures, hence we conclude that the majority of temperature change in Antarctica is due to the effects that the GHG changes had on the ocean temperatures rather than due to the direct radiative effects. The same result appears to be supported by the precipitation results, but the pattern of change in that variable is much more complex. Snow accumulation patterns are also determined by the sea surface temperature changes engendered by GHG variation. Intriguingly, summer warmth seems to be the limiting factor in the lack of

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

23

Fig. 9. Winter season (June, July, August) means of snow accumulation (in cm) produced in fully coupled GCM experiments. Cases as described in the text are indicated on figure labels (top), and the differences between cases are shown at the bottom of the figure. Conventions as in Fig. 4.

Antarctic ice sheet growth in these simulations. This points to orbitally induced surface energy balance (i.e. a cold summer orbit configuration) as a potentially crucial element in the initiation of Antarctic glaciation. Ocean temperatures produced in the COUPLED4x case appear to be a close match to late Eocene proxy temperatures. Terrestrial temperatures may be too cold, but there is no data from the continental interior to truly refute the model. The sensitivity of these temperatures to the imposed GHG forcing is sufficient to hypothesize that further increases (above 1120 ppm), still within the plausible range (Pearson and Palmer, 2000) would warm the region still further. 6. Conclusions and implications Extensive analysis of climate model results do not support the idea that either the atmosphere (Pierrehumbert, 2002) or the oceans (Huber and Sloan, 2001; Huber et al., 2003, 2004) transported substantially more heat in the Eocene than occurs in the present day. The near

constancy of total heat transport, the existence of strong compensation between oceanic and atmospheric components of total heat transport, and even compensations within each component, are in agreement with some climate dynamics predictions (e.g., Stone, 1978; Saravanan and McWilliams, 1995; Kamenkovich et al., 2000; Caballero and Langen, 2005). Nevertheless, it should be noted that this near-constancy of transport remains largely unexplained, and in this study we explored, using a hierarchy of models, the question, what processes control ocean heat transport? In Section 3, we showed that both AGCM simulations with two different fixed, warm polar SSTs—including one that implied massive OHT increases above modern values—produce cold mean annual continental interior temperatures. We find no support, either with fixed sea surface temperatures or in the fully coupled model, for significant changes in Antarctic precipitation as being the likely cause of Antarctic glacial initiation. Instead, temperatures, particularly summer temperatures, appear crucial. Given that it is feasible that

24

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

relatively small changes in terrestrial temperatures might initiate Antarctic glaciation (Barker et al., 1999; DeConto and Pollard, 2003) we contemplate two options. Either (a) OHT changes associated with the Eocene–Oligocene transition were greater than 3 than modern values, (b) or OHT was not the key driver. Results from fully coupled simulations performed with 560 and 1120 ppm CO2, discussed in Section 5, show that Antarctic climate is mildly sensitive to GHG changes. The sensitivity to plausible changes in GHG concentration was shown to be as great as the sensitivity to a 3-fold change in OHT. Distinguishing between options (a) and (b) becomes an issue that hinges upon the relative plausibility of the two. Were ocean currents in the Eocene constituted in such a different fashion from the present, that they might transport vastly more heat? We found no evidence from simple models or theory that this might have been the case. When taken in conjunction with the coupled GCM results, a more convincing case for nearmodern OHT in the Eocene is made. We have also argued that temperature proxies that indicate warm high latitude temperatures can not also be used as for increased OHT. Could GHG changes have caused Antarctic climate change? Given the wide range of CO2 estimates (Pearson and Palmer, 2000; Royer et al., 2001) and the lack of a proxy for methane and other GHG concentrations we believe that this mechanism is viable. It has been clearly established that major changes in productivity, as expressed in a variety of records, occurred during the Eocene–Oligocene transition (van Andel et al., 1975; Salamy and Zachos, 1999; Diester-Haass and Zahn, 2001; Coxall et al., 2005) and we believe that this is strong evidence for a GHG control on the evolution of climate through this interval. The effects of relatively small GHG changes in the coupled climate model cases created an Antarctic climate response at least as great as the response to a 3fold increase in OHT and the results do not support an increase in OHT to any significant degree. Therefore, we argue that it is more likely that GHG concentration changes triggered Oligocene Antarctic glaciation than OHT changes. We have not addressed other potential forcing factors in this study (e.g. those of Otto-Bliesner and Upchurch, 1997; Sloan and Pollard, 1998; KirkDavidoff et al., 2002), so we can neither support nor deny their importance.

Our results are in agreement with those in the recent study of DeConto and Pollard (2003) that reached an identical conclusion, but from a different starting point. In their study, which included a highresolution terrestrial ice model, GHG concentration was found to be the critical driver of Antarctic glaciation, more so than orbital configuration or OHT changes (which were smaller than in this study). In this study, as in theirs, the main regions of sensitivity were in mountain ranges, which is suggestive of the importance of GHG, climate, and uplift feedbacks in determining the evolution of Antarctic glaciers. Robert and Chamley (1993) suggested that the fact that Southern Ocean clay records display striking change during the Eocene–Oligocene transition, whereas Northern Hemisphere records do not, constitutes evidence against GHG-related hypotheses (because of the global nature of GHG responses). We argue instead, that Southern Ocean clay signals reflect the physical weathering associated with glaciation, which is already understood to be unipolar. We find the dominance of the temperature signal over precipitation sensitivity in our results to be consistent with clay records that show physical weathering-mediated, rather than precipitationrelated, changes in Southern Hemisphere clays during the Eocene–Oligocene transition (Robert and Chamley, 1993). The most widely accepted explanation for the dramatic climatic deterioration at the beginning of the Oligocene is a dramatic decrease in the ability of the ocean to transport heat into the Southern Hemisphere induced by opening of the Tasmanian Gateway. This study confirms and extends that of DeConto and Pollard (2003) and Huber et al. (2004) by demonstrating that it is unlikely that there were major OHT changes when the Tasmanian Gateway opened, and that implausibly large changes in OHT would be required to alter Antarctic continental climate even as much as plausible changes in GHG concentrations. We note that such large OHT values, even if they did occur, would likely demand an even greater role for increased GHG concentrations to maintain Northern Hemisphere warmth in the Eocene (Toggweiler and Bjornsson, 2000). Given that GHG changes are required to explain the pattern of Paleogene climate, regardless of the magnitude of OHT, and given that the magnitude of OHT changes were unlikely to be as large as ~3, it is more parsimonious to conclude that GHG changes played the primary role in the Eocene–Oligocene climate transition.

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

Acknowledgments We appreciate input from many people: H. Brinkhuis, J. Kennett, L.C. Sloan, G. Shaffer; funding from NSF ATM9810799 and the D. and L. Packard foundation to L.C. Sloan; and from NSF, NASA and ONR to Nof. We also wish to acknowledge NCAR, where all the computing and data analysis were performed (using NCAR’s NCL). We thank the perceptive and helpful reviews of T. Crowley, R. Toggweiler, J. Kennett, N. Exon, an anonymous reviewer on a previous draft. We are also grateful to the editors of this volume, especially A. Cooper for editorial excellence. References Bard, E., 2002. Climate shock: abrupt changes over millennial time scales. Phys. Today 55 (12), 32 – 38. Barker, P.F., Barrett, P.J., Cooper, A.K., Huybrechts, P., 1999. Antarctic glacial history from numerical models and continental margin sediments. Palaeogeogr. Palaeoclimatol. Palaeoecol. 150, 247 – 267. Barron, E.J., Peterson, W.H., 1991. The Cenozoic ocean circulation based on ocean general circulation model results. Palaeogeogr. Palaeoclimatol. Palaeoecol. 83, 1 – 28. Barron, E.J., Fawcett, P.J., Peterson, W.H., Pollard, D., Thompson, S.L., 1995. A simulation of Midcretaceous climate. Paleoceanography 10, 953 – 962. Bartek, L.R., Sloan, L.C., Anderson, J.B., Ross, M.I., 1992. Evidence from the Antarctic continental margin of Late Paleogene ice sheets: a manifestation of plate reorganization and synchronous changes in atmospheric circulation over the emerging Southern Ocean? In: Prothero, D.R., Berggren, W.A. (Eds.), Eocene–Oligocene Climatic and Biotic Evolution. Princeton University Press, Princeton, N.J., pp. 131 – 159. Bice, K.L., Barron, E.J., Peterson, W.H., 1997. Continental runoff and early Cenozoic bottom-water sources. Geology 25, 951 – 954. Bleck, R., Boudra, D., 1981. Initial testing of a numerical ocean circulation model using a hybrid, quasi-isopycnic vertical coordinate. J. Phys. Oceanogr. 11, 744 – 770. Bleck, R., Boudra, D., 1986. Wind-driven spin-up in eddy-resolving ocean models formulated in isopycnic and isobaric coordinates. J. Geophys. Res. 91, 7611 – 7621. Bleck, R., Smith, L.T., 1990. A wind-driven isopycnic coordinate model of the north and equatorial Atlantic Ocean: 1. Model development and supporting experiments. J. Geophys. Res. 95, 3273 – 3285. Brady, E.C., DeConto, R.M., Thompson, S.L., 1998. Deep water formation and poleward ocean heat transport in the warm climate extreme of the Cretaceous (80 Ma). Geophys. Res. Lett. 25, 4205 – 4208. Brass, G.W., Southam, J.R., Peterson, W.H., 1982. Warm saline bottom water in the ancient ocean. Nature 296, 620 – 623. Broecker, W.S., 2003. Does the trigger for abrupt climate change reside in the ocean or in the atmosphere? Science 300 (5625), 1519 – 1522. Buening, N., Carlson, S.J., Spero, H.J., Lee, D.E., 1998. Evidence for the Early Oligocene formation of a proto-subtropical convergence from oxygen isotope records of New Zealand Paleogene brachiopods. Palaeogeogr. Palaeoclimatol. Palaeoecol. 138, 43 – 68.

25

Caballero, R., Langen, P., 2005. The dynamic range of poleward energy transport in an atmospheric general circulation model. Geophys. Res. Lett. 32, L02705. doi:10.1029/2004GL021581. Carter, R.M., Carter, L., McCave, I.N., 1996. Current controlled sediment deposition from the shelf to the deep ocean: the Cenozoic evolution of circulation through the SW Pacific gateway. Geol. Rundsch., 438 – 451. Coxall, H.K., Wilson, P.A., Palike, H., Lear, C.H., Backman, J., 2005. Rapid stepwise onset of Antarctic glaciation and deeper calcite compensation in the Pacific Ocean. Nature 433, 53–53. doi:10.1038/nature03135. Covey, C., Barron, E., 1998. The role of ocean heat transport in climatic change. Earth-Sci. Rev. 24, 429 – 445. Craig, R.S., 2002. The palaeobiogeography of Late Cretaceous and Cenozoic brachiopods from Western Australia. Palaeogeogr. Palaeoclimatol. Palaeoecol. 179, 267 – 292. Crowley, T.J., 1998. Significance of tectonic boundary conditions for paleoclimate simulations. In: Crowley, T.J., Burke, K.C. (Eds.), Tectonic Boundary Conditions for Climate Reconstructions. Oxford University Press, New York, pp. 212 – 226. Crowley, T.J., Zachos, J.C., 2000. Comparison of zonal temperature profiles for past warm time periods. In: Huber, B.T., MacLeod, K.G., Wing, S.L. (Eds.), Warm Climates in Earth History. Cambridge University Press, pp. 50 – 76. Danabasoglu, G., 1998. On the wind-driven circulation of the uncoupled and coupled NCAR climate system ocean model. J. Climate 11, 1442 – 1454. De Boer, A.M., Nof, D., 2005. The island–wind buoyancy connection. Tellus 57A, 783 – 797. DeConto, R.M., Pollard, D., 2003. Rapid Cenozoic glaciation of Antarctica induced by declining atmospheric CO2. Nature 421, 245 – 249. Diester-Haass, L., Zahn, R., 2001. Paleoproductivity increase at the Eocene–Oligocene climatic transition: ODP/DSDP sites 763 and 592. Palaeogeogr. Palaeoclimatol. Palaeoecol. 172, 153 – 170. Dutton, A.L., Lohmann, K.C., Zinsmeister, W.J., 2002. Stable isotope and minor element proxies for Eocene climate of Seymour Island. Antarctica 17. doi:10.1029/2000PA000593. Elderfield, H., 2000. A world in transition. . . Nature 407, 851 – 852. England, M.H., Tomczak, M., Godfrey, J.S., 1992. Water-mass formation and Sverdrup dynamics — a comparison between climatology and a coupled ocean atmosphere model. J. Mar. Syst. 3, 279 – 306. Exon, N., et al., 2002. Drilling reveals climatic consequence of Tasmanian Gateway opening. Eos 83, 258 – 259. Ganachaud, A., Wunsch, C., 2000. Improved estimates of global ocean circulation, heat transport and mixing from hydrological data. Nature 408, 453 – 457. Godfrey, J.S., 1989. A Sverdrup model of the depth-integrated flow for the world ocean allowing for island circulations. Geophys. Astrophys. Fluid Dyn. 45, 89 – 112. Greenwood, D., Wing, S., 1995. Eocene continental climates and latitudinal temperature gradients. Geology 23, 1044 – 1048. Hack, H.H., Kiehl, H.T., Hurrell, J.W., 1998. The hydrologic and thermodynamic characteristics of the CCM3. J. Climate 11, 1179 – 1206. Hay, W.W., 1996. Techtonics and climate. Geol. Rundsch. 85, 409 – 437. Held, I.M., 2001. The partitioning of the poleward energy transport between the tropical ocean and atmosphere. J. Atmos. Sci. 58, 943 – 948.

26

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

Huber, M., 2001. Modeling early Paleogene climate: From the top of the atmosphere to the bottom of the ocean. Ph. D. Dissertation, Univ. of Calif., Santa Cruz. Huber, M., Caballero, R., 2003. Eocene El Nin˜o: evidence for robust tropical dynamics in the bhothouseQ. Science 299, 877 – 881. Huber, M., Sloan, L.C., 1999. Warm climate transitions: a general circulation modeling study of the Late Paleocene thermal maximum (~56 Ma). J. Geophys. Res. 104, 16633 – 16655. Huber, M., Sloan, L.C., 2000. Climatic responses to tropical sea surface temperature changes on a bgreenhouseQ Earth. Paleoceanography 15, 443 – 450. Huber, M., Sloan, L.C., 2001. Heat transport, deep waters, and thermal gradients: coupled simulation of an Eocene bgreenhouseQ Q climate. Geophys. Res. Lett. 28, 3481 – 3484. Huber, M., Sloan, L.C., Shellito, C., 2003. Early Paleogene oceans and climate: a fully coupled modelling approach using NCAR’s CCSM. In: Wing, S.L., Gingerich, P.D., Schmitz, B., Thomas, E. (Eds.), Causes and Consequences of Globally Warm Climates in the Early Paleogene, Geological Society of America Special Paper vol. 369, pp. 25 – 47. Huber, M., Brinkhuis, H., Stickley, C.E., Doos, K., Sluijs, A., Warnaar, J., Williams, G.L., Schellenberg, S.A., 2004. Eocene circulation of the Southern Ocean: was Antarctica kept warm by subtropical waters? Paleoceanography, A4026. doi:10.1029/ 2004PA001014. Hurrell, J.W., Hack, J.J., Boville, B.A., Williamson, D.L., Kiehl, J.T., 1998. The dynamical simulation of the NCAR Community Climate Model Version 3 (CCM3). J. Climate 11, 1207 – 1236. Ivany, L.C., Patterson, W.P., Lohmann, K.C., 2000. Cooler winters as a possible cause of mass extinctions at the Eocene/Oligocene boundary. Nature 407, 887 – 890. Kamenkovich, I., Marotzke, J., Stone, P.H., 2000. Factors affecting heat transport in an ocean general circulation model. J. Phys. Oceanogr. 30, 175 – 194. Kamp, P.J.J., Waghorn, D.B., Nelson, C.S., 1990. Late Eocene– early Oligocene integrated isotope stratigraphy and biostratigraphy for paleoshelf sequences in southern Australia: paleoceanographic implications. Palaeogeogr. Palaeoclimatol. Palaeoecol. 80, 311 – 323. Kennett, J.P., 1977. Cenozoic evolution of Antarctic glaciations, the circum-Antarctic ocean and their impact on global paleoceanography. J. Geophys. Res. 82, 3843 – 3860. Kennett, J.P., Stott, L.D., 1991. Abrupt deep-sea warming, palaeoceanographic changes and benthic extinctions at the end of the Paleocene. Nature 353, 225 – 229. Kennett, J.P., Stott, L.D., 1995. Terminal Paleocene mass extinction in the deep sea. Effects of Past Global Change on Life. National Academy Press, p. 250. Kennett, J.P., et al., 1972. Australian-Antarctic continental drift, palaeocirculation changes and Oligocene deep-sea erosion. Nature 239, 51 – 55. Kiehl, J.T., Hack, J.J., Bonan, G.B., Boville, B.A., Wiliamson, D.L., Rasch, P.J., 1998. The national center for atmospheric research community climate model: CCM3. J. Climate 11, 1131 – 1149. Kirk-Davidoff, D.B., Schrag, D.P., Anderson, J.G., 2002. On the feedback of stratospheric clouds on polar climate. Geophys. Res. Lett. 29 (art. no. 1556). Klinger, B.A., Marotzke, J., 2000. Meridional heat transport by the subtropical cell. J. Phys. Oceanogr. 30, 696 – 705. Lawver, L.A., Gahagan, L.M., 1998. Opening of drake passage and its impact on Cenozoic ocean circulation. In: Crowley, T.J., Burke,

K.C. (Eds.), Tectonic Boundary Conditions for Climate Reconstructions. Oxford University Press, New York, pp. 212 – 226. Lear, C.H., Elderfield, H., Wilson, P.A., 2000. Cenozoic deep-sea temperature sand global ice volumes from Mg/Ca in benthic foraminiferal calcite. Science 287, 269 – 272. Liu, Z.Y., Philander, S.G.H., 1995. How different wind stress patterns affect the tropical subtropical circulations of the upper ocean. J. Phys. Oceanogr. 25, 449 – 462. Lyle, M., 1997. Could early Cenozoic thermohaline circulation have warmed the poles? Paleoceanography 12, 161 – 167. Mackensen, A., Ehrmann, W.U., 1992. Middle Eocene through Early Oligocene climate history and paleoceanography in the Southern Ocean: stable oxygen and carbon isotopes from ODP sites on Maud Rise and Kerguelen Plateau. Mar. Geol. 108, 1 – 27. Manabe, S., Stouffer, R.J., 1998. Two stable equilibria of a coupled ocean-atmosphere model. J. Clim. 1, 841 – 866. Markwick, P.J., 1994. Equability, continentality, and tertiary climate: the crocodilian perspective. Geology 22, 613 – 616. Markwick, P.J., 1998. Fossil crocodilians as indicators of Late Cretaceous and Cenozoic climates: implications for using palaeontological data in reconstructing palaeoclimate. Palaeogeogr. Palaeoclimatol. Palaeoecol. 137, 205 – 271. McGowran, B., Li, Q., Cann, J., Padley, D., McKirdy, D.M., Shafik, S., 1997. Biogeographic impact of the Leeuwin Current in southern Australia since the late middle Eocene. Palaeogeogr. Palaeoclimatol. Palaeoecol. 136, 19 – 40. Mikolajewicz, U., Maierreimer, E., Crowley, T.J., Kim, K.Y., 1993. Effect of Drake and Panamanian gateways on the circulation of an ocean model. Paleoceanography 8, 409 – 426. Miller, K.G., Fairbanks, R.G., Mountain, G.S., 1987. Tertiary oxygen isotope synthesis, sea-level history and continental margin erosion. Paleoceanography 2, 1 – 19. Murphy, M.G., Kennett, J.P., 1986. Development of latitudinal thermal gradients during the Oligocene: oxygen-isotope evidence from the southwest Pacific. In: Kennett, J.P., von der Borch, C.C., et al., (Eds.), Init. Repts. DSDP, 90. U.S. Govt. Printing Office, Washington, pp. 1347 – 1360. Najjar, R.G., Nong, G.T., Seidov, D., Peterson, W.H., 2002. Modeling geographic impacts on early Eocene ocean temperature. Geophys. Res. Lett. 29. doi:10.1029/2001GL014-438. Nelson, C.S., Cooke, P.J., 2001. History of oceanic front development in the New Zealand sector of the southern ocean during the Cenozoic—a synthesis. N.Z. J. Geol. Geophys. 44, 535 – 553. Nof, D., 2000a. Why much of the Atlantic circulation enters the Caribbean Sea and very little of the Pacific circulation enters the Sea of Japan. Prog. Oceanogr. 45, 39 – 67. Nof, D., 2000b. Does the wind control the import and export of the South Atlantic? J. Phys. Oceanogr. 30, 2650 – 2667. Nof, D., 2002. Is there a meridional overturning cell in the Pacific and Indian Oceans? J. Phys. Oceanogr. 32, 1947 – 1959. Nof, D., 2003. The Southern Ocean’s grip on the northward meridional flow. Prog. Oceanogr. 56 (2), 223 – 247. Nof, D., Van Gorder, S., 2003. Did an open Panama isthmus correspond to an invasion of Pacific water into the Atlantic? J. Phys. Oceanogr. 33 (7), 1324 – 1336. Nong, G.T., Najjar, R.G., Seidov, D., Peterson, W.H., 2000. Simulation of ocean temperature change due to the opening of the Drake Passage. Geophys. Res. Lett. 27, 2689 – 2692. Oleinik, A.E., 2001. Eocene gastropods of western Kamchatka— implications for high latitude north Pacific biostratigraphy and

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28 biogeography. Palaeogeogr. Palaeoclimatol. Palaeoecol. 166, 121 – 140. Omta, A.W., Dijkstra, H.A., 2003. A physical mechanism for the Atlantic-Pacific flow reversal in the early Miocene. Glob. Planet. Change 788, 1 – 12. Otto-Bliesner, B.L., Upchurch, G.R., 1997. Vegetation-induced warming of high-latitude regions during the late Cretaceous period. Nature 385, 804 – 807. Otto-Bliesner, B.L., Brady, E.C., Shields, C., 2002. Late Cretaceous ocean: coupled simulations with the National Center for atmospheric research climate system model. J. Geophys. Res. 107 (art. no. 4019). Pak, D.K., Miller, K., 1992. Paleocene to Eocene benthic foraminiferal isotopes and assemblages: implications for deepwater circulation. Paleoceanography 7, 405 – 422. Pearson, P.N., Palmer, M.R., 1999. Middle Eocene seawater pH and atmospheric carbon dioxide concentrations. Science 284, 1824 – 1826. Pearson, P.N., Palmer, M.R., 2000. Atmospheric carbon dioxide concentrations over the past 60 million year. Nature 406, 695 – 699. Pearson, P.N., et al., 2001. Warm tropical sea surface temperatures in the Late Cretaceous and Eocene epochs. Nature 413, 481 – 487. Pedlosky, J., 1996. Ocean Circulation Theory. Springer. 453 pp. Peixoto, J.P., Oort, A.H., 1992. Physics of Climate. American Institute of Physics, p. 520. Pierrehumbert, R.T., 2002. The hydrologic cycle in deep-time climate problems. Nature 419, 191 – 198. Raphael, M.N., 1998. Quasi-stationary waves in the Southern Hemisphere: an examination of their simulation by the NCAR Climate System Model, with and without an interactive ocean. J. Climate 11, 1405 – 1418. Ravelo, A.C., Andreasen, D.H., 2000. Enhanced circulation during a warm period. Geophys. Res. Lett. 27, 1001 – 1004. Rind, D., Chandler, M., 1991. Increased ocean heat transports and warmer climate. J. Geophys. Res. 96, 7437 – 7461. Robert, C., Chamley, H., 1993. Late Eocene–Early Oligocene evolution of climate and marine circulation: deep sea clay mineral evidence. In: Kennett, J.P., Warnke, D.A. (Eds.), The Antarctic Paleoenvironment: A Perspective on Global Change, Antarct. Res. Ser. vol. 56, pp. 97 – 117. Royer, D.L., Berner, R.A., Beerling, D.J., 2001. Phanerozoic atmospheric CO2 change: evaluating geochemical and paleobiological approaches. Earth-Sci. Rev. 54, 349 – 392. Salamy, K.A., Zachos, J.C., 1999. Latest Eocene–Early Oligocene climate change and Southern Ocean fertility: inferences from sediment accumulation and stable isotope data. Palaeogeogr. Palaeoclimatol. Palaeoecol. 145, 61 – 77. Saravanan, R., McWilliams, J.C., 1995. Multiple equilibria, natural variability, and climate transitions in an idealized ocean-atmosphere model. J. Climate 10, 2296 – 2323. Schmidt, G.A., Mysak, L.A., 1996. Can increased poleward oceanic heat flux explain the warm Cretaceous climate? Paleoceanography 11, 579 – 593. Schneider, E.K., Thompson, S.L., Barron, E., 1985. Mid-Cretaceous continental surface temperatures; are high carbon dioxide concentrations need to simulate above freezing conditions? The Carbon Cycle and Atmospheric CO2: Natural Variations Archean to Present. Geophys. Mono. Amer. Geophys. Union, Washington, D. C., pp. 554 – 560. Seager, R., Battisti, D.S., Yin, J., Gordon, N., Naik, N., Clement, A.C., Cane, M.A., 2002. Is the Gulf stream responsible for Europe’s mild winters? Q. J. R. Meteorol. Soc. 128, 22563 – 22586.

27

Sewall, J.O., Sloan, L.C., Huber, M., Wing, S., 2000. Climate sensitivity to changes in land surface characteristics. Glob. Planet. Change 26, 445 – 465. Shackleton, N.J., Kennett, J.P., 1975. Paleotemperature history of the Cenozoic and the initiation of Antarctic glaciation: oxygen and carbon isotope analyses in DSDP sites 277, 279, and 281. Deep Sea Drill. Proj., Init. Rep. 29, 743 – 755. Shellito, C., Sloan, L.C., Huber, M., 2003. Climate model constraints on atmospheric CO2 levels in the early–middle Palaeogene. Palaeogeogr. Palaeoclimatol. Palaeoecol. 193, 113 – 123. Sijp, W.P., England, M.H., 2004. Effect of the Drake Passage throughflow on global climate. J. Phys. Oceanogr. 34, 1254 – 1266. Sloan, L.C., 1994. Equable climates during the early Eocene: significance of regional paleogeography for North American climate. Geology 22, 881 – 884. Sloan, L.C., Pollard, D., 1998. Polar stratospheric clouds: a high latitude warming mechanism in an ancient greenhouse world. Geophys. Res. Lett. 25, 3517 – 3520. Sloan, L.C., Rea, D.K., 1996. Atmospheric carbon dioxide and early Eocene climate: a general circulation modeling sensitivity study. Palaeogeogr. Palaeoclimatol. Palaeoecol. 119, 275 – 292. Sloan, L. Cirbus, Walker, J.C.G., Moore Jr., T.C., Rea, D.K., Zachos, J.C., 1992. Possible methane-induced polar warming in the early Eocene. Nature 357, 320 – 322. Sloan, L.C., Walker, J.C.G., Moore, T.C., 1995. Possible role of oceanic heat transport in Early Eocene climate. Paleoceanography 10, 347 – 356. Sloan, L. Cirbus, Huber, M., Crowley, T.J., Sewall, J.O., Baum, S., 2001. Effect of sea surface temperature configuration on model simulations of bequableQ climate in the early Eocene. Palaeogeogr. Palaeoclimatol. Palaeoecol. 167, 321 – 335. Stickley, C.E., Brinkhuis, H., Schellenberg, S.A., Sluijs, A., Rohl, U., Fuller, M., Grauert, M., Huber, M., Williams, G.L., 2004. Timing and nature of the deepening of the Tasmanian Gateway. Paleoceanography, A4027. doi:10.1029/2004PA001022. Stone, P.H., 1978. Constraints on dynamical transport of energy on a spherical planet. Dyn. Atmos. Ocean. 2, 123 – 139. Street, C., Brown, P.R., 2000. Palaeobiogeography of Early Cretaceous (Berriasian–Barremian) calcareous nanno-plankton. Mar. Micropaleontol. 39, 265 – 291. Thomas, D.J., 2004. Evidence for deep-water production in the North Pacific Ocean during the early Cenozoic warm interval. Nature 430, 65 – 68. doi:10.1038/nature02639. Toggweiler, J.R., Bjornsson, H., 2000. Drake Passage and paleoclimate. J. Quat. Sci. 15, 238 – 319. Toggweiler, J.R., Samuels, B., 1995. Effect of Drake Passage on the global thermohaline circulation. Deep-Sea Res. 42, 477 – 500. Trenberth, K.E., Caron, J.M., 2001. Estimates of meridional atmosphere and ocean heat transports. J. Climate 14, 3433–3433. van Andel, T.H., Heath, G.R., Moore, T.C., 1975. Cenozoic history and paleoceanography of the central equatorial Pacific Ocean: a regional synthesis of deep sea drilling project data. Mem. Geol. Soc. 143. Wei, W., Wise, S.W., 1990. Middle Eocene to Pleistocene calcareous nannofossils recovered by ocean drilling program leg 113 in the Weddell Sea. Proc. Ocean Drill. Program Sci. Results 113, 639 – 666. Wolfe, J.A., 1994. Tertiary climate changes at middle latitudes of western North America. Palaeogeogr. Palaeoclimatol. Palaeoecol. 108, 195 – 205.

28

M. Huber, D. Nof / Palaeogeography, Palaeoclimatology, Palaeoecology 231 (2006) 9–28

Zachos, J.C., Lohmann, K.C., Walker, J.C.G., Wise, S.W., 1993. Abrupt climate change and transient climates during the Paleogene — a marine perspective. J. Geol. 101, 191 – 213. Zachos, J.C., Stott, L.D., Lohmann, K.C., 1994. Evolution of early Cenozoic temperatures. Paleoceanography 9, 353 – 387. Zachos, J.C., Quinn, T.M., Salamy, K.A., 1996. High-resolution (104 years) deep-sea foraminiferal stable isotope records of the

Eocene–Oligocene climate transition. Paleoceanography 11, 251 – 266. Zachos, J., Pagani, M., Sloan, L., Thomas, E., Billups, K., 2001. Trends, rhythms, and aberrations in global climate 65 Ma to present. Science 292, 686 – 694.