The Milankovitch orbital hypothesis on the climate effects of orbital precession, obliquity, and eccentricity is now generally accepted as a primary pacing mechanism for late Pleistocene ice ages (e.g., Imbrie et al., 1989, 1992, 1993). Based on this finding, tuning of climate records to orbital variations provides a primary tool for chronology development over late Cenozoic time (e.g., Shackleton et al., 1995b). In an application of this method, Tiedemann et al. (this volume) developed internally consistent orbitally tuned timescales for the interval 6–2.5 Ma from four Leg 202 sites (1236, 1237, 1239, and 1241), providing an exceptional framework for further studies of regional climate change.

In spite of widespread agreement on the importance of Earth's orbit to climate change, the mechanisms that translate orbital changes in the global and regional climate changes remain under debate (e.g., Maasch and Salzman, 1990; Muller and MacDonald, 1997; Shackleton, 2000). In particular, it is not clear to what extent climate varies internally in a set of natural oscillations, or if on long timescales it responds directly to external forcing, or if it rapidly switches between discrete modes, given either random or systematic forcing. One likely path to illuminate such mechanisms is to examine the evolution of the orbital climate cycles as the sensitivity of response, relative to orbital forcing, rises and falls through time (Pisias et al., 1990).

A particular puzzle is the 100-k.y. climate cycle, which arose most recently in mid-Pleistocene time, ~1 m.y. ago (Pisias and Moore, 1981). The primary signature of this so-called Middle Pleistocene Transition (MPT) is a frequency change in global ice volume oscillations from a dominant 41-k.y. cycle to a 100-k.y. cycle recorded in benthic foraminiferal 18O. This change in the frequency of primary climate rhythms occurred in the absence of any significant changes in the pattern or amplitude of orbital forcing and thus required a fundamental change in processes internal to the Earth system. Most hypotheses for the MPT invoke a threshold response to a gradual long-term cooling, which is often attributed to an assumed secular decrease in atmospheric pCO2 (e.g., Raymo, 1997; Raymo et al., 1997; Mudelsee and Schulz, 1997; Paillard, 1998; Berger et al., 1999; Lea, 2004; Lisiecki and Raymo, 2005). Further rhythmic variations in CO2 provide a likely feedback mechanism for sustaining the 100-k.y. climate cycle as a response to greenhouse feedbacks (Pisias and Shackleton, 1984; Shackleton, 2000; EPICA Community Members, 2004).

McClymont and Rosell Melé (2005) demonstrated that the MPT was not simply a development within the Northern Hemisphere ice sheets and high-latitude climates but was preceded by a significant change in the tropical Pacific circulation system, thereby identifying the tropics as a potential driver of the MPT. They proposed that the development of the modern zonal temperature gradient in the equatorial Pacific (1.2–0.9 Ma) intensified Walker circulation and reduced heat flux but increased moisture transport to high latitudes, leading to the development of more extensive ice sheet growth and the shift toward the 100-k.y. world.

Other ideas for long-term change in the sensitivity to orbital forcing include mountain uplift (Ruddiman and Raymo, 1988) or erosion of topography and rise of plateaus (Berger and Jansen, 1994) that allowed larger glaciers to survive through smaller cycles of orbital precession (dominated by 23- and 19-k.y. cycles), eventually creating a 100-k.y. climate cycle (although this would also likely yield a 412-k.y. cycle, which is not observed) (Imbrie and Imbrie, 1980). Gildor and Tziperman (2001) also note long-term cooling of the deep sea at polar outcrops, which influenced sea ice distributions. Ruddiman (2006) suggests that the abrupt rise of the 100-k.y. cycle at the MPT results from greenhouse feedback and the interaction of forcing from the 23- and 41-k.y. orbital cycles (but note that nonlinear interaction of these two cycles yields sum and difference tones of 14 and 52 k.y., so additional mechanisms are still needed).

Clark and Pollard (1998) and Clark et al. (in press) propose an alternate mechanism for the origin of the 100-k.y. climate cycle, in which gradual erosional exposure of crystalline Precambrian Shield bedrock yielded a high-friction substrate that supported thicker ice sheets, with an attendant change in their response time to orbital forcing. In this scenario, 100-k.y. cycles arise as a natural consequence of the dynamics of thick and extensive ice sheets.

A puzzle in this regard, however, is that 100-k.y. cycles have been noted in various components of the Earth system at times prior to the MPT (e.g., Mix, 1987; Mix et al., 1995a; Groeneveld et al., this volume, Steph et al., this volume; Holbourn et al., 2005). Do these earlier 100-k.y. climate cycles arise from mechanisms similar to the Pleistocene cycles, or are they fundamentally different? Do they provide precursor "templates" in the climate system, which could then be entrained through feedback mechanisms to grow a Pleistocene 100-k.y. ice sheet cycle in a gradually cooling climate? Or are these other examples simply independent and unrelated features of the climate system?

Sites from Leg 202 that provide new opportunities for study of the long-term evolution of orbital-scale climate variability include Sites 1236 and 1237 in the subtropical South Pacific (both exceptionally long sequences sampled with the APC), Sites 1238–1240 in the eastern equatorial Pacific cold tongue (EEPCT), and Sites 1241 and 1242 in the eastern tropical north Pacific warm pool of the Panama Bight. At this stage of Leg 202 studies, orbital-scale records of upper ocean and deep-sea variability have focused on the time interval 2.5–5.0 Ma, including the so-called early Pliocene "golden age" of global warmth (Sarnthein and Fenner, 1988) that was characterized by relatively small polar ice sheets and warmer deep-sea temperatures (Sarnthein and Tiedemann, 1989; Shackleton et al., 1995b; Mix et al., 1995b), as well as the early stages of cooling toward the Pliocene–Pleistocene ice ages of the past ~2.5 m.y. (Fig. F6). This is a time generally found to be dominated by 41-k.y. climate cycles. In addition, other Leg 202 studies focus on the interval from 12.7 to 14.7 Ma, an analogous time of cooling following the so-called Miocene climatic optimum (Fig. F6).

Groeneveld et al. (this volume) combined Mg/Ca paleotemperature estimates with 18O measurements of the near-surface dwelling foraminifer G. sacculifer over the interval 4.8–2.4 Ma at Site 1241 in order to assess the sensitivity of orbital-scale climate variability in the eastern Pacific warm pool (EPWP) to Pliocene global warmth, closure of the Panama Isthmus, and global cooling during early development of the Pliocene–Pleistocene ice ages from ~3.7 to 2.4 Ma (Fig. F8). The beginning of this interval is noted by the development of a strong and shallow thermocline from ~4.8 to 4.0 Ma and to an increasing gradient of 13C from the intermediate waters (at Site 1236) to the thermocline (Neogloboquadrina dutertrei at Site 1241) developing strongly after 3.3 Ma. This likely reflects a change in the high-latitude source waters that feed the Equatorial Undercurrent (EUC) in response to polar cooling and the intensification of NHG (Steph et al., this volume). The shallowing thermocline has little or no apparent effect on SST (Steph et al., this volume; Groeneveld et al., this volume).

During the early Pliocene time interval of 4.8–3.7 Ma, benthic foraminiferal 18O is dominated by the well-known 41-k.y. (obliquity) cycle with minor contributions from the 23-k.y. (precession) effect. In contrast, Mg/Ca surface temperatures at Site 1241 are dominated by the 23-k.y. precession cycle, along with a ~100-k.y. cycle that is associated either with orbital eccentricity or a nonlinear response to precession (Groeneveld et al., this volume). The 18O of G. sacculifer at Site 1241 also includes strong variability in the precession and eccentricity bands, as does the calculated paleosalinity record, which is consistent with ideas of tropical origins of such cycles related to insolation and monsoon interactions with the continents (Crowley et al., 1992).

During the later interval of 3.7–2.4 Ma, Groeneveld et al. (this volume) find that benthic 18O (presumed to reflect mostly ice volume) and planktonic formaminiferal Mg/Ca paleotemperatures at Site 1241 are both dominated by ~41-k.y. cycles, reflecting the growing influence of high-latitude processes on tropical climates. The 18O of G. sacculifer and estimates of ice volume–corrected paleosalinity contain more long-period variability near the 100-k.y. cycle, however, suggesting continued variations in the tropical hydrologic cycle related to precession and eccentricity. Such a result is consistent with the finding of Mix et al. (1995a) of a persistent ~100-k.y. cycle in biogenic sedimentation at Site 846. At that site, the phase of the 100-k.y. cycle of sedimentation is synchronous with orbital eccentricity prior to the MPT (suggesting local or regional responses to insolation as suggested by Crowley et al., 1992). After the transition, however, the phase of the 100-k.y. cycle of sedimentation shifts and synchronizes with orbit-lagging ice volume and polar temperatures (benthic 18O).

The observed dominance of precession-paced SST variations in the EPWP prior to ~3.7 Ma (Groeneveld et al., this volume), however, contrasts the cyclic temperature variability in the EEPCT as suggested by a recently published alkenone-based paleotemperature record from Site 846 (Lawrence et al., 2006). The authors emphasize a dominant 41-k.y. cycle of SST variability throughout the past 5 m.y. at Site 846 and an in-phase relationship with benthic 18O wherever the obliquity-related signals are coherent. These alkenone temperature estimates from the SEC imply a tight coupling between SST variability in the equatorial Pacific cold tongue and high-latitude-driven climate processes associated with changes in global ice volume. In contrast, SST variability at Site 1241 in the EPWP prior to ~3.7 Ma, based on the Mg/Ca proxy, suggests a dominant response to low-latitude changes in seasonal insolation (and associated processes) associated with variations in precession. The threshold of the EPWP for a significant response to high-latitude obliquity forcing was probably low during the early Pliocene, when the volume of polar ice sheets and their variability were relatively small (Philander and Fedorov, 2003).

A number of mechanisms may explain the difference in cyclic SST variability between Sites 1241 and 846, including different dynamics in surface and subsurface currents, thermocline depth, wind field, and associated atmosphere-ocean couplings. Today, Site 1241 (~6°N) is positioned within the Intertropical Convergence Zone (ITCZ) region and reflects changes in surface water signatures of the NECC, which carries the return flow of relatively warm, low-salinity surface waters eastward out of the west Pacific warm pool. Site 846 (3°S) is positioned south of the ITCZ in the EEPCT, whereas SST depends on wind-driven (southeast trades) changes in upwelling and cold-water advection from the Humboldt Current. Lawrence et al. (2006) favor obliquity-driven changes in atmospheric CO2 concentrations rather than vertical movements of the thermocline as a possible mechanism to explain SST variations at Site 846. If so, the obliquity-related temperature variability at Sites 846 and 1241 should provide similar phasing and amplitudes, which is testable by cross-spectral analyses. As variations in atmospheric CO2 would synchronize tropical SST changes, other mechanisms must be responsible for the different dominance in cyclic SST variability between EPWP and EEPCT. One possible explanation relates to the different origin of surface water masses. Site 856 is directly influenced by the SEC, which sources Southern Ocean water masses. Site 1241 is influenced by surface water masses, which originate from the western Pacific warm pool. Modeling studies indicate that the western Pacific warm pool is not particularly sensitive to cooling of the EEPCT (e.g., Lee and Poulsen, 2005, 2006), mainly because of equatorial insolation. Therefore, the warm pool may serve as a buffer preventing Southern Hemisphere signals from appearing in the Northern Hemisphere at Site 1241.

Steph et al. (this volume) examined whether the difference in cyclic SST variability could be explained by local changes in thermocline depth. They demonstrate that planktonic Mg/Ca temperature and 18O of the deep-dwelling planktonic foraminifers are good indicators of changes in thermocline depth, as both records match each other. Given that the high-resolution 18O records of G. tumida from Sites 1241 (warm pool), 851 (warm pool, west of Site 1241) (Cannariato and Ravelo, 1997), and 1239 (1°S, upwelling region off Ecuador) (Steph, 2005) are almost identical in absolute values and orbital-scale variability during the early Pliocene, it is not very likely that changes in thermocline depth led to the observed differences in cyclic SST between the cold tongue and the warm pool in the tropical east Pacific. In this context, however, it is important to note that changes in SST and thermocline depth are not necessarily coupled. On a long timescale, Steph et al. (this volume) document that the early Pliocene shoaling of the thermocline (at, e.g., Site 1241) has little or no apparent effect on SST. On a shorter (late Pleistocene) timescale, Benway et al. (2006) present similar findings for decoupling of pycnocline intensity and SSTs. At this stage of Leg 202 data analyses and assessment, a possible explanation for the apparent difference in early Pliocene cyclic SST variability between Sites 1241 and 846 is that the EPWP more likely responded to local insolation changes, whereas the EEPCT was more directly linked to high-latitude processes, wind-driven thermocline dynamics, or possibly atmospheric CO2 variability. We expect that further information from the west Pacific warm pool and the eastern boundary current and upwelling system off South America will help to assess whether these differences reflect true regional differences in climate response.

Steph et al. (2006) compare the planktonic foraminiferal 18O and Mg/Ca records from Pacific Site 1241 with Caribbean Site 1000 and note the development of strong 23-k.y. cycles in the Caribbean related to orbital precession, in the interval 4.4–3.0 Ma, as the two oceans start to become isolated by the rise of the Central American Isthmus. These authors argue that precessional forcing of El Niño events, as predicted for the late Pleistocene by Clement et al. (1999), may have modulated flows of water between the Pacific and Atlantic, leading to a strong surface-ocean signal on the Caribbean side of the isthmus. Full closure of this isthmus eliminated the sensitivity to this effect. In the latest Pleistocene, Benway et al. (2006) examine the sensitivity of surface-ocean salinity to change at Site 1242, and although they find substantial millennial-scale changes, they find no significant changes from glacial to interglacial time, supporting the idea that the presence of an restricted but open CAS can change the sensitivity of low-latitude systems to orbital-scale changes.

Holbourn et al. (2005) examine the record of climate further back in time, during the middle Miocene global cooling event (14.7–12.7 Ma) (Fig. F6), to examine global climate changes in the orbital bands, pairing a new record from Site 1237 with a similar record from Site 1146. Both records were continuously cored and provide unprecedented resolution of the evolution in orbital-scale climate changes, with stable isotopes analyzed at 4- to 5-k.y. intervals and XRF scanned at ~1-k.y. intervals throughout the sequence. An orbitally tuned age model assumes that 18O varies in phase with both orbital precession and obliquity, based on the orbital solution of Laskar et al. (2004). Within this interval, a rapid increase in benthic foraminiferal 18O from 13.91 to 13.84 Ma records an increase in continental ice volume (constrained also by benthic foraminiferal Mg/Ca paleotemperature estimates at other sites [Lear et al., 2000]). Associated with the growth of continental ice sheets, a striking transition from a high-amplitude cycle in the 41-k.y. band to one in the 100-k.y. band occurred between 14.1 and 13.8 Ma. The strong 100-k.y. cycle in 18O is also recorded in the XRF record of iron at Site 1237, although it is not clear whether this is related to iron input, dilution by varying biogenic production, or carbonate dissolution cycles.

Although qualitatively similar to the better-known development of 100-k.y. climate cycles in the MPT, the Miocene 100-k.y. cycles are of lower amplitude and precede the development of the ice sheets by as much as a few 100 k.y., whereas the Pleistocene 100-k.y. cycles develop after a long interval of gradual cooling ice sheet expansion. Further, lowering of marine carbon isotope values indicates a rapid transfer of organic carbon to inorganic carbon in the ocean system during the MPT (Raymo et al., 1997; Clark et al., in press), just the opposite of the observed increase in 13C across the mid-Miocene transition (Holbourn et al., 2005). In both the Pleistocene and Miocene transitions, a significant ~400-k.y. cycle is found in benthic 13C, which may be related to eccentricity or precession control of the carbon cycle (Mix et al., 1995b; Holbourn et al., 2005). Under the assumption that a positive shift in 13C implies increased burial of 13C-depleted organic matter, Holbourn et al. (2005) infer that the growing 100-k.y. climate cycles of the Miocene were "primed" by decreasing atmospheric pCO2 and triggered by increases in the amplitude of orbital eccentricity (noting that better proxies for past CO2 variability are sorely needed). Thus, the late Pleistocene 100-k.y. climate cycles are not unique in Earth's history, and although the examples from the Miocene and Pleistocene are both associated with climate cooling, they occur under significantly different global boundary conditions, different global mean temperatures, and with different relationships to the carbon cycle. This important study issues a challenge to climatologists to understand multiple origins of 100-k.y. climate cycles that are now well documented in the geologic record.