18O values from the early Eocene to early Oligocene (Fig. F8). Superimposed upon these long-term (tectonic scale) trends in the Cenozoic were quasi-periodic rhythms induced by orbital (Milankovitch) forcing (for review, see Zachos et al., 2001a).
One of the central paradigms of Cenozoic climate evolution in the Southern Hemisphere is that Antarctic cooling and ice sheet development was related to the opening of tectonic gateways (Tasmanian Seaway and Drake Passage), which permitted the unrestricted flow of the ACC (Kennett, 1977; Kennett and Barker, 1990). The traditional view is that the Tasmanian Seaway opened sometime near the E/O boundary (Shipboard Scientific Party, 2001) and the Drake Passage opened in the late Oligocene or earliest Miocene (Barker and Burrell, 1977); however, the timing of these events are only loosely constrained because of complex tectonics in both regions (Lawver et al., 1992; Cande et al., 2000). Together with the closure of equatorial gateways and declining atmospheric pCO2, the development and strengthening of the ACC played an important role in the transition of Earth's climate system from "greenhouse to "icehouse" mode. This transition is expressed in the long-term increase in benthic 18O values from the early Eocene to early Oligocene (Fig. F8). Superimposed upon these long-term (tectonic scale) trends in the Cenozoic were quasi-periodic rhythms induced by orbital (Milankovitch) forcing (for review, see Zachos et al., 2001a).
During Leg 177, Paleogene and lowermost Neogene sediments were recovered at only one site. A 330-m-thick section ranging in age from early Miocene to middle Eocene was drilled at Site 1090, located on the southern end of the Agulhas Ridge in the area of an elongated contourite drift body (Wildeboer Schut et al., in press). A clean geomagnetic signal and a verifiably complete composite section make this site an appealing target for paleoceanographic study.
The late Eocene (~35 Ma) has been recognized as a period when multiple meteorites may have impacted the Earth (Glass and Koeberl, 1999). For this reason, Kyte (Chap. 4, this volume) searched the well-dated upper Eocene section of Site 1090 for any evidence of an impact deposit. Concentrations of cosmogenic iridium peak at 950 pg/g in Sample 177-1090B-30X-5, 105-106 cm (291 mcd), providing concrete evidence of an impact event. In addition, both clear and dark-colored spherules were found in samples with the highest Ir and are believed to be microtektites and microkrystites. This event correlates with similar deposits found at ODP Leg 113 Site 689 on the Maud Rise (Glass and Koeberl, 1999) and elsewhere, thereby providing an important time-stratigraphic horizon for long-distance correlation.
Latimer and Filippelli (2002) and Diekmann et al. (submitted a [N1]) produced a large geochemical (bulk sediment parameters), sedimentological (grain size distribution), and clay mineralogical data set to chronicle changes in terrigenous and biogenic fluxes, decipher changes in paleoproductivity, and reconstruct changes in ocean circulation at Site 1090 between ~44 and 17 Ma. On the basis of elevated P/Al ratios, Latimer and Filippelli (2002) identified two intervals of enhanced export production in the South Atlantic during the middle Eocene (44-42 Ma) and across the Eocene-Oligocene transition (Fig. F9). A late Eocene-early Oligocene increase in biogenic export is also indicated by enhanced biogenic opal deposition between 37.5 and 33.5 Ma (Diekmann et al., submitted a [N1]). The latter part of this interval of high opal export was associated with the highest concentration of organic carbon recorded at Site 1090. The late Eocene opal event represents the first significant deposition of biogenic silica at Site 1090, and the timing was similar to increases in silica accumulation reported from the Tasman Rise and Falkland Plateau (Andrews et al., 1975; Ciesielski and Weaver, 1983). At more southerly locations, such as the Maud Rise (Site 689) and the Kerguelen Plateau (Sites 738 and 744), opal accumulation also began in the late Eocene and reached its maximum only after the Eocene/Oligocene boundary (Ehrmann and Mackensen, 1992; Diester-Haass and Zahn, 1996; Salamy and Zachos, 1999). Diekmann et al. (submitted a [N1]) attributed the changes in productivity regime at Site 1090 to paleogeographic changes in the tropical regions that led to increased meridional ocean circulation and the establishment of a more vigorous ACC, including the development of upwelling cells and oceanic frontal systems. They speculated that the late Eocene mass deposition of biogenic opal and organic carbon may have contributed to enhanced consumption of atmospheric CO2, which served as a positive feedback mechanism for further cooling.
Site 1090 terrigenous proxies indicate a major shift in the composition and style of sedimentation at ~32.8 Ma in the earliest Oligocene (Fig. F9), marking a change from continental to oceanic crustal sources. A possible cause of this shift in source area is a change in circulation that permitted the transport of basaltic material from the rifting region west of Site 1090 in response to the opening of the Drake Passage. The opening of the Tasmanian Gateway to deepwater flow also occurred in the earliest Oligocene (by ~33 Ma), and some geophysical evidence supports a near-synchronous opening of the Drake Passage and Tasmanian Gateway (Lawver et al., 1994; Lawver and Gahagan, 1998). This time was marked by a rapid reduction in siliciclastic sediment supply and the onset of carbonate ooze in Leg 189 cores (Shipboard Scientific Party, 2001). Similar decreases in siliciclastics and increases in biogenic sedimentation occurred near the E/O boundary on the Maud Rise in the South Atlantic (Diester-Haass and Zahn, 1996; Kennett and Barker, 1990; Salamy and Zachos, 1999). Major ice growth began on Antarctica in the earliest Oligocene (Fig. F8) (Zachos et al., 1992) and changed the style of Antarctic continental weathering (Robert and Kennett, 1997). At Site 1090, the clay mineral assemblage changes during the late Eocene to early Oligocene and is marked by an increase in pure terrigenous clay minerals such as illite, chlorite, and kaolinite at the expense of smectite, which appears to be largely of marine origin (Diekmann et al., submitted a [N1]). These changes in terrigenous proxies at Site 1090 are related to global cooling and glaciation that promoted stronger mechanical weathering of continental source areas.
Hopes of obtaining a continuous Paleogene oxygen isotope record at Site 1090 were thwarted by the low abundance and poor preservation of foraminifers, especially across the E/O boundary. Foraminifers were abundant enough, however, for isotope analysis in the upper Oligocene to lower Miocene section. Billups et al. (2002) generated a benthic isotope record from ~25 to 16 Ma at Site 1090, which they correlated to the astronomically tuned 18O signal from Site 929 (Fig. F5) (Shackleton et al., 1999). The significance of this correlation lies in the fact that Site 1090 has a geomagnetic polarity reversal stratigraphy, whereas Site 929 does not. Therefore, Site 1090 can potentially be used to fix the geomagnetic polarity timescale to the astronomical timescale (Channell et al., in press).
Billups et al. (2002) found that the 18O maximum (Mi 1), which is customarily associated with the Oligocene/Miocene (O/M) boundary (Fig. F8), is actually slightly older (23.86 Ma on the timescale of Cande and Kent, 1995) and falls in Subchron C6Cn.2r. The Mi 1 glaciation was probably caused by a "silent node" in Earth's orbital configuration, when low obliquity and low eccentricity resulted in a 200-k.y.-long period of low seasonality (Zachos et al., 2001b). In addition to Mi 1, the O/M boundary is also marked by a maximum in 13C values referred to as CM-OM (Hodell and Woodruff, 1994). Billups et al. (2002) found that CM-OM is present only 24 cm above the polarity subchron boundary C6Cn.2n/r at Site 1090 (Fig. F5). Because the O/M boundary is placed at this polarity reversal, the 13C maximum is very useful for recognizing the boundary.
Comparison of South Atlantic Site 1090 and tropical North Atlantic Site 929 isotope records reveals no carbon isotope gradient between these sites, whereas 18O values of Site 1090 are consistently higher than those at Site 929 (Fig. F5). In fact, there is a general lack of a carbon isotope gradient between the North Atlantic, Southern Ocean (Site 1090), and Pacific Ocean during the majority of the latest Oligocene and early Miocene (Billups et al., 2002). However, a strong oxygen isotopic gradient existed between the Southern Ocean and other ocean basins prior to 17 Ma, suggesting that the deep Southern Ocean was colder and/or more saline than the deep North Atlantic or the Pacific. The relatively cold Southern Ocean reflects a well-developed ACC and deep flow through the Drake Passage by ~26 Ma (Billups et al., 2002).
None of the Leg 177 sites contain a continuous section of early middle Miocene age. A hiatus of 14 m.y. duration occurs at Site 1090 at ~70 mcd, extending from the early Miocene to the early Pliocene, whereas basal sediment at Sites 1088 and 1092 date to the middle middle Miocene between ~13 and 14 Ma. Censarek and Gersonde (submitted [N2]) estimated middle to late Miocene thermal gradients across the ACC using diatoms preserved in a latitudinal transect of cores extending from Sites 689 and 690 on Maud Rise (~64.5°S), to Site 1092 on Meteor Rise (47°S), and to Site 1088 on the Agulhas Ridge (41°S). Relatively warm surface water conditions and low latitudinal differentiation persisted until ~13.5 Ma. This is followed by gradual cooling, culminating with the northward expansion of cold waters into the present subantarctic realm centered around 11 Ma, a period of lowest sea level in the Miocene, according to Haq et al. (1987). Although the magnitude of this sea level low stand, which in the absence of Northern Hemisphere ice sheets should indicate expansion of Antarctic ice, is under discussion (Kominz et al., 1998), the co-occurrence of cold surface water expansion and a drop in Southern Ocean bottom-water temperatures (Billups and Schrag, 2002) points to the establishment of the West Antarctic Ice Sheet (WAIS) around the middle/late Miocene boundary. This episode of cooling was followed by a resumption of warmer conditions and decrease of thermal differentiation. However, beginning at ~9 Ma, diatoms that are similar to modern sea ice taxa are present at the Maud Rise sites and then subsequently appear at Site 1092 (Meteor Rise) at ~8 Ma. This pattern may mark the establishment of a seasonal sea ice field and may coincide with the growth of the WAIS.
Billups (2002) produced a preliminary low-resolution benthic isotope record for the upper Miocene through Pliocene section at Site 1088. Although this section was drilled in a single hole only, recovery was better than 95% with gaps only existing at core breaks. Sediment accumulation was slow but apparently continuous from ~12 to 2.5 Ma. Carbon isotopic gradients between the North Atlantic, South Atlantic, and Pacific Oceans indicate that a nutrient-depleted water mass has existed in the South Atlantic since ~6 Ma. Intraocean 13C gradients suggest the contribution of Northern Component Water (NCW) was similar to today by 6.0 Ma and greater than today during the early Pliocene. In the late Miocene, benthic 18O values at Site 1088 increased in two steps at 7.4 and 6.9 Ma, indicating cooling of intermediate and deepwater masses formed in the Southern Ocean.
Diekmann et al. (submitted b [N3]) compared the sedimentologic histories of the middle Miocene-Pleistocene sections of Site 1088 (41°S) and Site 1092 (46°S) to infer past changes in surface water conditions near the present-day Subtropical Front (Site 1088) and Subantarctic Front (Site 1092), respectively (Fig. F1). Although sediments at both sites are dominantly carbonate (CaCO3 = 90 to 95 wt% at Site 1088 and 60 to 90 wt% at Site 1092), biogenic opal deposition began near the middle/late Miocene boundary at both sites. A switch from nannofossil to foraminiferal oozes and a small opal maximum is recorded at both sites in the latest Miocene and correlates with the global "biogenic bloom" across the Miocene/Pliocene boundary (Dickens and Owen, 1999; Hermoyian and Owen, 2001). Thereafter, the histories of opal deposition at the two sites diverge, indicating a latitudinal decoupling of opal deposition between the northern and southern parts of the Leg 177 transect. At Site 1088 (41°S), carbonate deposition prevailed near the modern Suptropical Front throughout the Pliocene-Pleistocene with an opal peak (up to 6 wt%) in the late Pliocene between 3.1 and 2.2 Ma. This opal maximum at Site 1088 correlates with the early Matuyama Diatom Maximum (MDM) observed to the north at ODP Leg 175 sites in the Namibia upwelling area (Lange et al., 1999).
Sediment at Site 1092 (47°S) contains a higher proportion of biogenic silica than at Site 1088 and is consistent with the formation of the circum-Antarctic opal belt since 2.5 Ma near the modern Polar Front (Diekmann et al., submitted b [N3]). Opal accumulation peaks in the early Pleistocene after 2 Ma at Site 1088 and coincides with an increase in weight percent opal at nearby Site 704 (Froelich et al., 1991). Although poorly recovered, high biogenic silica export during the early Pleistocene is also inferred from the numerous and thick sequences of diatom mats penetrated at Site 1093 (Shipboard Scientific Party, 1999; Pearce et al., submitted [N4]). It is noteworthy that the early Pleistocene diatom maximum between 46° and 50°S in the South Atlantic may have coincided with the end of the MDM off southwest Africa (Lange et al., 1999; Berger et al., 2002), although better chronologic control is needed at Sites 1091 and 1093 to calculate precise opal accumulation rates. These two regions may be linked by production of thermocline water by subduction in the subantarctic region and subsequent upwelling off southwest Africa.
Prior to Leg 177, Site 704 (Leg 114) on Meteor Rise was one of the few sites with sufficient stratigraphic continuity and carbonate content for high-resolution Neogene paleoceanographic studies in the Southern Ocean. Site 1092 was also drilled on Meteor Rise but in shallower water than Site 704 and in an area that was less likely to be affected by downslope transport (Gersonde, Hodell, Blum, et al., 1999). Two related studies collected stable isotope and ice-rafted debris (IRD) data from the lower and lower upper Pliocene section of Site 1092 and compared the results to those obtained at Site 704 (Andersson et al., 2002; Murphy et al., 2002).
The oxygen and carbon isotope record of Site 1092 is very similar but not identical to the isotope signals at Site 704 (Fig. F10) (Andersson et al., 2002). As first noted by Mix et al. (1995), benthic 18O values at Site 704 (and Site 1092) are greater than those at deep Pacific Site 849 between 3.5 and 2.7 Ma. This observation is enigmatic because it requires that deep Pacific waters (at a depth of 3850 m) were warmer or less saline than those at a depth of ~2000 to 2500 m in the subantarctic South Atlantic. Nonetheless, the 18O results from Site 1092 support those from Site 704 and demonstrate an offset from Pacific values that is often close to 0.3
-0.5 and as high as 0.7-0.8. Changes in the flux and/or salinity of North Atlantic Deep Water (NADW) do not offer a satisfactory explanation (Andersson et al., 2002), and other mechanisms must be sought to reconcile the Pliocene isotope records from the Southern Ocean and eastern Pacific.
Andersson et al. (2002) also compared benthic 13C values at Site 1092 with North Atlantic and deep Pacific end-members (Fig. F10). Prior to 3.6 Ma, 13C values at Site 1092 were lower, suggesting that the site was bathed by a more nutrient-rich water mass compared to NADW prior to 3.6 Ma. Taken at face value, this contradicts the results of Billups et al. (1998, 2002) suggesting that NADW was as strong or stronger during the early Pliocene than in the Holocene. Either 13C gradients between the subantarctic South Atlantic and Pacific Oceans evolved differently than the Atlantic-Pacific gradient (Billups et al., 1998), or benthic 13C changes in the Southern Ocean were overprinted by productivity effects (Mackensen et al., 1993).
Benthic 13C values at Site 1092 decreased abruptly at 2.75 Ma relative to records in the North Atlantic and Pacific Oceans (Fig. F11), which corroborates previous results from Site 704 (Hodell and Ciesielski, 1990; Hodell and Venz, 1992; Raymo et al., 1992). Prior to 2.75 Ma, Southern Ocean benthic 13C oscillated between those of the North Atlantic and Pacific Oceans. At ~2.75 Ma, Southern Ocean 13C values decreased abruptly, indicating a reduction of deepwater ventilation at the same time as expansion of Northern Hemisphere ice sheets. A progressive reduction of NADW over the past 3 m.y. is also supported by Nd and Pd isotopes in Atlantic ferromanganese crusts (Frank et al., 2002).
In the same set of Site 1092 samples analyzed for stable isotopes, Murphy et al. (2002) documented the occurrence of ice-rafted debris. IRD arrived frequently during the early and early late Pliocene, but only as "background rafting" (occasional grains per sample). The first identifiable IRD above background occurred at MIS KM4 (~3.18 Ma) (Shackleton et al., 1995). Successive IRD peaks become progressively larger thereafter, similar to the pattern noted at nearby Site 704 (Warnke et al., 1992). The greatest IRD peak at 2.8 Ma represents a lag deposit owing to a hiatus. The latest Gauss Chron (~2.8-2.5 Ma) was a time of pronounced change in the subantarctic region that included surface waters cooling, a northward shift of the Polar Front, and establishment of the modern circum-Antarctic opal belt (Fig. F12) (Hodell and Ciesielski, 1990; Froelich et al., 1991; Hodell and Venz, 1992). The pattern of IRD delivery to Meteor Rise is also similar to IRD records from the high-latitude North Atlantic (Kleiven et al., 2002), although detailed correlation is required to determine the exact phase relationship. At Site 1092, one interval in the Gauss Chron and several short intervals in the upper Gilbert Chron are devoid of IRD. Oxygen isotope values at these times were only 0.6 less than modern, however (Andersson et al., 2002). These data corroborate a slight warming of deep water during the mid-Pliocene and possible minor deglaciation of Antarctica but do not support speculation of a substantially reduced East Antarctic Ice Sheet.
One of the main objectives of Leg 177 was to recover expanded sections arrayed across the ACC that could be used to study Pleistocene paleoceanography at orbital and suborbital resolution. For convenience, we review the Pleistocene postcruise contributions under several themes that were elaborated in the original scientific objectives of Leg 177.
The early response of sea-surface temperatures (SSTs) in the Southern Ocean on orbital and suborbital timescales (Imbrie et al., 1992; Charles et al., 1996; Pichon et al., 1992) and the close link between Southern Hemisphere temperature and atmospheric CO2 (Cuffey and Vimeux, 2001) implicate this region as a potential driver of global climate change. One of the Leg 177 objectives, therefore, was to document Pleistocene SST changes along a north-south transect across the ACC. Summer sea-surface temperatures (SSST) were reconstructed using the modern analog technique (MAT) or paleoecological transfer functions applied to planktonic foraminifers, diatoms, and radiolarian assemblages at four Leg 177 sites (1089, 1090, 1093, and 1094). Changes in surface water density structure were examined by 18O analysis of depth-stratified planktonic foraminifers.
Becquey and Gersonde (2002) estimated SSST using planktonic foraminiferal assemblages in core PS2989-2 for the past 550 k.y., and their study was extended to the base of the Pleistocene at nearby subantarctic Site 1090 at millennial resolution (Fig. F13) (Becquey and Gersonde, in press). During the early Pleistocene (1.8-0.9 Ma), relatively cool SSSTs prevailed at 43°S as isotherms in the South Atlantic shifted north by ~7° latitude. This is approximately the same time when thick diatom mats were being deposited farther south at Site 1093 (50°S) (Pearce et al., submitted [N4]). A distinct change in spectral properties began at ~1.2 Ma when the power of the 100-k.y. cycle began to increase. The rather uniform and cold conditions of the early Pleistocene were followed by a transitional period between 0.87 and 0.43 Ma that was marked by increasing SSTs during interglacials. The last 0.4 m.y. was marked by strong variability in SSST with glacial-to-interglacial contrasts up to 8°C. Only during the upper portion of MIS 15 and the earliest parts of MISs 11, 9, 7, 5, and 1 ("climatic optima") do SSSTs exceed those at Site 1090 today. Carbonate dissolution can affect SSST estimates by increasing the relative number of cold-water dwellers (e.g., Neogloboquadrina pachyderma) during glacials and warm-water taxa resistant to dissolution (e.g., Globorotalia inflata) during warm periods. The latter has been observed during early MIS 11, resulting in an overestimated SSST (Fig. F13).
Cortese and Abelmann (2002) report a 160-k.y.-long record of SSST at centennial resolution derived from a radiolarian-based transfer function at Site 1089 in the northern Subantarctic Zone (Fig. F14). Results indicate a maximum temperature change of up to 7°C in surface waters for the last two glacial terminations (I and II). The rise in SSST precedes changes in oxygen isotopes at both terminations, supporting an early response of Southern Ocean SSST relative to Northern Hemisphere ice volume. Both deglaciations were also interrupted by sudden cooling episodes. Distinct millennial-scale oscillations in temperature occurred during MIS 3 and 4, and SSST changes were almost as great as those observed at terminations. These large SSST variations bear a strong resemblance to planktonic 18O changes in the same core (Ninnemann et al., 1999) but differ from the low-amplitude SSST signal (1°-3°C) and warming trend seen after 45 ka in the SSST record estimated by alkenone undersaturation ratios (Sachs et al., 2001). Millennial-scale SSST variations during MISs 3 and 4 at Site 1089 may have been related to rapid movements of the Subtropical Front associated with Dansgaard-Oeschger events. Spectral analysis of SSSTs shows significant power at frequencies observed in the Vostok ice core and North Atlantic sediment cores.
Mortyn et al. (2002) examined changes in surface water column structure over the last two terminations at ODP Sites 1089 and 1093 by measuring the isotopic differences between shallow- and deep-dwelling planktonic foraminifers. During glacial periods, surface water was more stratified south of the Polar Front Zone (PFZ) (Shemesh et al., 2002) and less stratified to the north (Mortyn et al., 2002). Pronounced planktonic 13C minima occur on Terminations I and II in deep-dwelling foraminifers at Sites 1089 and 1093. Spero and Lea (2002) have traced carbon isotope minimum events at the last glacial termination to the equatorial Pacific and interpreted them to represent the breakdown of surface stratification in Antarctic surface waters, renewed upwelling of Circumpolar Deep Water (CDW) in the Southern Ocean, and advection of low 13C waters to the Subantarctic Front where Antarctic Intermediate Water and Subantarctic Mode Water are formed.
Complementary to the radiolarian-based study of SSST south of the Subtropical Front at Site 1089, Bianchi and Gersonde (2002) reconstructed SSST and sea ice variations across Termination II in Antarctic surface waters, south of the present-day Polar Front. This study is based on diatom records from a longitudinal transect of six piston cores extending from the western Indian Ocean sector into the Scotia Sea and includes the late MIS 6 to MIS 5 sequence recovered at Site 1094. During late MIS 6, cold water from the Weddell Sea extended into the central and eastern South Atlantic and lowered SSST in the present Antarctic Zone to ~0°C. This permitted the expansion of the winter sea ice edge to the position of the modern Polar Front. According to their chronology (Martinson et al., 1987), warming at Termination II started at ~132-131 ka and temperatures increased by ~4°-5°C within a ~3-k.y. period and reached maximum values during earliest MIS 5, lasting for 2-3 k.y. Similar to the record at Site 1089 (Cortese and Abelmann, 2002), the warming at Termination II was punctuated by a major cold reversal. Additional short-lived SSST oscillations occur at 200- to 300-yr intervals and are possibly triggered by short-term meltwater discharges. The SSST pattern during MIS 5e is marked by several fluctuations with amplitudes of 1°-2°C and does not support severe climate instability during MIS 5e as originally proposed on the basis of the Greenland Ice Core Project (GRIP) record (Dansgaard et al., 1993). Applying an age model based on the radiometric dating of the Termination II midpoint at 135 ± 2.5 ka (Henderson and Slowey, 2000), Bianchi and Gersonde (2002) suggested that the onset of warming at Termination II was triggered by precessional changes influencing southern high-latitude summer insolation. Rapid reduction of the sea ice field and increased SSST may have acted as a positive feedback mechanism by reducing albedo and enhancing ocean-atmosphere gas exchange, thereby releasing CO2 to the atmosphere. Further reinforcement of the SSST during the termination may be related to changes in global circulation affected by a collapse or strong reduction of NADW production, resulting in additional warming of the southern latitudes.
Kunz-Pirrung et al. (2002) present similar diatom-based studies at and around MIS 11 (Fig. F15), which may have been the warmest and/or longest interglacial of the late Pleistocene. During glacial Stages 10 and 12, the Antarctic Zone, which is currently ice free, was seasonally covered by sea ice. The SSST record obtained at Sites 1093 and 1094 show distinct climate variability at millennial timescales during both cold and warm periods. Termination V is especially noteworthy at Site 1093 because it consists of an 8-m expanded section containing laminated diatom mats (Grigorov et al., 2002). At Termination V (MIS 12/11), SSSTs increased by 4°-6°C but the warming was punctuated by two distinct cooling events at Site 1094 that might be related to meltwater discharges. Maximum temperatures occurred during late Termination V and exceeded modern values by 2°C for a period of 8 k.y. This pattern supports a very early response of SSST in the Southern Ocean to orbital (Milankovitch) forcing. The maximum SSST during the MIS 11 climatic optimum do not exceed values obtained during other interglacial periods such as early MIS 5 or MIS 1, but the total duration of warmth was longer than other interglacials. These results from high-sedimentation-rate cores support the findings of a lower-resolution study of South Atlantic cores by Hodell et al. (2000).
Sea ice is a fast-changing variable with strong albedo feedback that profoundly affects the physical, chemical, and biological properties of Antarctic surface water and atmospheric circulation. Variations in sea ice also affect the air-sea exchange of gases and may play an important role in glacial-to-interglacial variations in deepwater ventilation and atmospheric pCO2 (Stephens and Keeling, 2000; Sigman and Boyle, 2000). As such, an important Leg 177 goal was to reconstruct the history, distribution, and seasonal variation of sea ice. One of the few sea ice proxies available is the analysis of sea ice diagnostic taxa of diatoms (Gersonde and Zielinski, 2000).
Shemesh et al. (2002) reconstructed sea ice extent over the last deglaciation in piston core TTN057-13 (Site 1094). High relative abundance of sea ice diatoms during the last glaciation argue for enhanced sea ice cover at this site. Sea ice is the first parameter to change at Termination I, followed by nutrient proxies and SSST. Sea ice decreased abruptly (in ~1000 yr) from 19 to 18 ka and led the increase in atmospheric pCO2 by ~2000 yr. The early response of Southern Ocean sea ice supports models calling for a role of sea ice in glacial-to-interglacial variations in atmospheric pCO2 (Moore et al., 2000; Stephens and Keeling, 2000); however, a delay mechanism of a few thousand years may be needed to explain the observed sequence of events.
Kunz-Pirrung et al. (2002) and Bianchi and Gersonde (2002) used diatoms to reconstruct sea ice extent in the Antarctic Zone (including Site 1094) from MIS 6 to MIS 5 and at Sites 1093 and 1094 from MIS 12 to MIS 10, respectively. During glacial Stages 6, 10, and 12, winter sea ice covered the area corresponding to the modern Antarctic Zone and Polar Front. This is similar to sea ice extent reported for the last glaciation by Crosta et al. (1998) and Gersonde et al. (in press). Cold-water diatom taxa and low biogenic sedimentation rates point to permanent sea ice cover in the area south of 54°S of the Atlantic sector during MIS 6. The winter sea ice edge began to retreat during Terminations II and V at ~132-131 ka and 432 ka, respectively, but advanced again during major cold reversals on both terminations. The central Atlantic sector of the Southern Ocean was ice free throughout the years between ~130 and 110 ka during MIS 5 and between ~425 and 390 ka during MIS 11. Spectral analysis of the diatom sea ice signals reveals distinct millennial-scale cycles including a persistent 3-k.y. periodicity during the mid-Brunhes Chron (Kunz-Pirrung et al., 2002).
Documentation of IRD, including Heinrich Events, in the North Atlantic has contributed greatly to our understanding of Laurentide Ice Sheet dynamics. Is there evidence for millennial-scale variability in Antarctic Ice Sheet similar to that observed for the North Atlantic? To address this question, Kanfoush et al. (2000, 2002) used piston and/or ODP cores at 41°, 47°, and 53°S to reconstruct the distribution of IRD across the PFZ. IRD in the South Atlantic is composed dominantly of ash and quartz with minor amounts of fine-grained volcanics, coarse-crystalline rock fragments, and mica. The source of the volcanic ash is believed to be the South Sandwich Islands in the Scotia arc, with minor contribution from nearby Bouvet Island (Smith et al., 1983). The ash settles on ice shelves and seasonal sea ice and is transported north by tabular icebergs and the seasonal advance of sea ice. Quartz and other nonash lithics are derived mainly from iceberg calving from ice shelves such as those found in the Weddell Sea region.
Kanfoush et al. (2000) found discrete episodes of IRD deposition on millennial timescales throughout the last glaciation (MIS 2 to 4). Prominent IRD layers correlate across the PFZ, suggesting episodes of Antarctic Ice Sheet instability. South Atlantic IRD peaks are associated with times of warming (interstadials) and increased NADW production in the North Atlantic. This interhemispheric linkage may be a manifestation of antiphase climate behavior between these regions (Broecker, 1998; Manabe and Stouffer, 1997). Alternatively, the linking mechanism may have been sea level rise associated with melting of the Laurentide Ice Sheet during strong interstadial events that unpinned grounded Antarctic ice shelves, releasing armadas of icebergs to the South Atlantic. Partial support for this mechanism comes from reported ages of sea level highstands from the Huon Peninsula, New Guinea, and Barbados during MIS 3 (Chappell, 2002) that match the South Atlantic IRD events within chronological uncertainty (Yokoyama et al., 2001; Kanfoush, 2002).
Kanfoush et al. (2002) extended the study of IRD delivery to the South Atlantic at Site 1094 (54°S) over the last four climate cycles (Fig. F16). They found that most, but not all, of the IRD variability is captured by whole-core physical properties such as magnetic susceptibility and gamma ray attenuation bulk density. Each of the last four glacial periods was marked by high IRD abundance and millennial-scale variability that may reflect instability of ice shelves in the Weddell Sea region.
In contrast, the early part of each interglacial period was nearly devoid of IRD, yet its abundance increased during the latter part of each interglacial, marking the onset of a neoglaciation. For the Holocene, Hodell et al. (2001) reported that IRD abundance was very low in piston core TTN057-13 (Site 1094) from 10 to 5.5 ka, coinciding with the early Holocene Hypsithermal. At ~5.5 ka, the delivery of IRD increased abruptly, heralding the onset of neoglacial conditions. This timing is in agreement with an early neoglacial advance of mountain glaciers in South America and New Zealand between 5.4 and 4.9 ka (Porter, 2000). The neoglaciations of MIS 7 and 9 were associated with a substantial increase in planktonic 18O, whereas the neoglaciation of MIS 5 began prior to the end of the Eemian and was accompanied by only a modest increase in planktonic 18O (Kanfoush et al., 2002). Using the SPECMAP age model at Site 1094, the duration of the "IRD-free period" during MIS 11 was longer than any other interglacial of the late Pleistocene (Kanfoush et al., 2002; Hodell et al., in press).
Kleiven and Jansen (Chap. 12, this volume) examined IRD in the early-middle Pleistocene from 0.73 to 1 Ma at Site 1094. This period includes the mid-Pleistocene Transition (MPT) when the dominant power of climate variability shifted from 41- to 100-k.y. cycles. Suborbital variability is present in IRD concentration throughout the middle Pleistocene, and the pacing is similar to that observed for the last glacial period (Kanfoush et al., 2000). This study demonstrates that millennial-scale variability in the South Atlantic was not restricted to the 100-k.y. world but also occurred under glacial conditions with reduced global ice volume. The IRD record at Site 1094 in the early-middle Pleistocene is similar to IRD variability at Site 983 in the North Atlantic, suggesting that the amplitude and pacing of iceberg discharge was similar between the two hemispheres.
The South Atlantic sector of the Southern Ocean is a key area for studies of global thermohaline circulation because it represents the insertion point of NADW into Antarctic circumpolar flow (Fig. F2). Deep water reflects a mixture of the outflows from all ocean basins. In addition, the process of intermediate and deepwater formation in the Southern Ocean links the atmosphere to the deep sea and the geochemical fingerprint of high-latitude surface waters is transmitted throughout the world's deep ocean. Consequently, changes in the ventilation of deep basins from the Antarctic region could be an important mechanism for atmospheric pCO2 variation (Toggweiler, 1999; Sigman and Boyle, 2000; Keeling and Stephens, 2001).
The thermohaline "conveyor belt" has undergone large changes during interglacial-to-glacial transitions as production of NADW was reduced and Southern Component Water (SCW) filled the deep Atlantic basins. In the North Atlantic, NADW was replaced by an intermediate water mass (referred to as Glacial North Atlantic Intermediate Water or upper NADW) that extended as far south as 28°S in the western South Atlantic during the last glaciation (Oppo and Horowitz, 2000). How these glacial-to-interglacial changes in North Atlantic circulation affected the Southern Ocean is uncertain because of a fundamental discrepancy in the interpretation of Cd and carbon isotope data in the South Atlantic (for review, see Boyle and Rosenthal, 1996). The Cd content of CDW remained unchanged between glacial and interglacial periods, indicating that thermohaline deepwater circulation in the Southern Ocean was much the same as today. In contrast, glacial benthic 13C values in cores from the South Atlantic cores currently bathed by lower CDW were lower than those in the deep Pacific. Taken at face value, this would imply that the oldest deep water in the ocean was located in the deep South Atlantic and/or the 13C in CDW source areas was much lower than today (due to reduced productivity or air-sea exchange). Neither Cd/Ca nor 13C is without complication, however. Benthic 13C may suffer from productivity overprints (for review, see Mackensen and Bickert, 1999), and Cd is potentially confounded by depth-dependent dissolution (McCorkle et al., 1995).
Although Leg 177 postcruise science has not resolved the "13C-Cd controversy," it has made several important contributions toward our understanding of glacial-to-interglacial changes in middepth and deepwater circulation. Venz and Hodell (2002) studied changes in deepwater circulation using benthic 13C at Site 1090. This site is near the interface of lower CDW and NADW today. The benthic 13C signal at Site 1090 is nearly identical to records at Site 704 and piston core RC13-229, all of which are bathed by CDW. At ~1.55 Ma (MIS 52), glacial 13C values at Site 1090 dropped below those in the Pacific, thereby establishing the pattern that persisted throughout the late Pleistocene (Fig. F11). Venz and Hodell (2002) speculate that the onset of lower-than-Pacific 13C values at Site 1090 may have been related to expansion of the Antarctic sea ice field and reduced ventilation of deep water during glacial periods after 1.55 Ma. This time also marked the onset of a strong 41-k.y. power in all surface and deepwater parameters at Site 704 and tight interhemispheric coupling between the high-latitude North and South Atlantic (Hodell and Venz, 1992).
Hodell et al. (2002) studied Pleistocene vertical carbon isotope gradients at Sites 1088, 1089, and 1090, which form a depth transect down the northern flank of the Agulhas Ridge (Fig. F2). The two deepest Sites 1089 and 1090 (below 3700 m) exhibit large glacial-to-interglacial variations in benthic 13C, whereas the amplitude of the 13C signal at Site 1088 (~2100 m water depth) is much smaller (Fig. F17). At no time during the Pleistocene are benthic 13C values at Site 1088 lower than the Pacific, indicating that the carbon isotopic signal of middepth waters evolved differently from deep waters in the South Atlantic. Reconstruction of vertical 13C gradients for the last glaciation support the existence of a sharp chemocline between 2100 and 2700 m that separated nutrient-depleted middepth waters above 2100 m from poorly ventilated deepwater masses below (Ninnemann and Charles, 2002) (Fig. F18). This intermediate-to-deep 13C gradient (
13CI-D) was pronounced for each glacial stage of the last 1.1 m.y. (Hodell et al., 2002), supporting the existence of a chemical divide in the glacial Southern Ocean separating low-CO2 middepth water above from high-CO2 deep water below. Comparison of late Pleistocene variations in 13CI-D and Vostok pCO2 lends support to the model of Toggweiler (1999) whereby glacial-to-interglacial changes in vertical chemical gradients in the Southern Ocean influence atmospheric pCO2 (Fig. F19).
Diekmann and Kuhn (2002) inferred changes in deepwater circulation using variations in sediment composition and clay mineralogy at Site 1090 over the MPT. The paleoceanographic literature is somewhat contradictory in terms of when the MPT occurred and whether the transition was gradual or abrupt. At Site 1090, spectral changes are first observed at ~1.2 Ma (Fig. F20). Clay mineralogical changes indicate that Circumpolar Deep Water expanded farther north during glacials after 1.2 Ma (Diekmann and Kuhn, 2002). This is also supported by isotopic data of Venz and Hodell (2002), who report an increase in the carbon isotopic gradient between Sites 982 and 607 at 1.2 Ma, indicating increased glacial suppression of NCW and farther northward penetration of CDW into the deep North Atlantic. The first prominent 100-k.y. cycle in benthic 18O records is often taken to be MIS 22 at ~0.9 Ma. Several important changes occurred at Site 1090 at this time. A change in illite chemistry toward more iron-rich varieties at 0.9 Ma indicates more arid conditions in South African source areas (Diekmann and Kuhn, 2002). At MIS 22, the magnitude and pacing of glacial-to-interglacial variability in benthic 13C increased throughout the Atlantic, marking the start of strong 100-k.y. cycles consisting of well-ventilated deep water during interglacials and poorly ventilated deep water during glacials (Venz and Hodell, 2002). The power of the 100-k.y. cycle also increases progressively in the 13C gradient between middepth and deep waters in the South Atlantic between 1.2 and 0.6 Ma (Hodell et al., 2002). The emergence of a strong 100-k.y. cycle in 13C during the mid-Pleistocene is consistent with possible CO2 forcing of this climate transition.
Kuhn and Diekmann (2002) studied clay minerals at Site 1089 to infer changes in thermohaline circulation on millennial and orbital timescales. During glacial stages, substages, and stadials of the last 600 k.y., they document a fundamental reorganization of the global conveyor system that included a reduction of NADW input to the ACC, as well as a weakening of the return flow of surface waters from the Indian Ocean to the South Atlantic via the Agulhas Current. Grain-size variations indicate stronger contour-current activity during interglacials than during glacial periods in the southern Cape Basin, which may be related to glacial-to-interglacial changes in the production rate of bottom-water masses in source areas within the Weddell Sea.
During Leg 177, we used variations in percent color reflectance as a proxy for weight percent CaCO3 and assigned preliminary marine isotope stages assuming that carbonate concentrations were high during interglacial stages and low during glacial stages. This relationship is typical of cores from the Atlantic Ocean. When oxygen isotope measurements were completed postcruise, the "Atlantic-type" carbonate stratigraphy held for all Leg 177 sites except Site 1089, located at 4600 m water depth in the southernmost Cape Basin. Hodell et al. (2001) found the opposite pattern typical instead of Indo-Pacific Ocean cores (i.e., high carbonate during glacial stages and low carbonate during interglacials) (Fig. F21). Comparison of weight percent carbonate and foraminiferal fragmentation in the same samples suggests that the carbonate record is controlled mostly by dissolution (Fig. F22). The Site 1089 dissolution signal is nearly identical to cores from the Indo-Pacific Ocean but has much greater fidelity because it is free from many of the complications that limit other records (low sedimentation rates, blurring by chemical erosion, bioturbation, etc.). As such, it represents a qualitative, high-resolution record of the temporal evolution of the carbonate saturation state of the deep sea.
Weight percent carbonate lags changes in benthic 18O by an average of ~7.6 k.y., and carbonate variations are in phase with the rate of change (first derivative) of benthic 18O (Fig. F21). Maximum dissolution occurs at the transition from interglacial to glacial periods, and increased preservation occurs during deglaciations. The lagged response of carbonate to 18O reflects a steady-state mass balance process whereby the lysocline adjusts to maintain alkalinity balance between riverine input and marine burial. The Site 1089 carbonate signal is remarkably similar to inferred changes in the Sr/Ca of seawater for the past 250 k.y. (Martin et al., 1999), suggesting that both carbonate dissolution and seawater Sr/Ca may be controlled by sea level-induced changes in the location of carbonate deposition (shelf-basin fractionation) during glacial-to-interglacial cycles (Berger, 1982). The transient change in preservation during the transitions into and out of glacial stages may reflect a response of the carbonate system to a redistribution of alkalinity and dissolved inorganic carbon (DIC) in the ocean (i.e., so-called carbonate compensation by Broecker and Peng, 1987). Comparison of the Site 1089 carbonate and Vostok pCO2 records suggests a role of deep-sea CO32- variations for governing at least some second-order features of the atmospheric pCO2 signal.
Hodell et al. (in press) studied the mid-Brunhes dissolution cycle (MBDC) at Sites 1089 and 1090 using stable isotopes and dissolution indices. The MBDC is part of a long-period oscillation that is expressed in dissolution indices and planktonic 13C that reach a maximum during MIS 13 and MIS 11. A high correlation between Vostok pCO2 and percent foraminiferal fragmentation signals between 450 and 300 ka suggests a tight coupling of the marine carbonate system and atmospheric pCO2 during the mid-Brunhes Chron. MIS 11 is often suggested as a potential analog for Holocene climate change because the Earth's orbital geometry was similar to that of today. Hodell et al. (2002) caution, however, that the peak in dissolution and 13C during the mid-Brunhes Chron indicates that the marine carbonate-carbon cycle was fundamentally different than today. As such, MIS 11 and the Holocene are not entirely analogous.
Ice cores offer the most detailed records available for reconstructing changes in climate and atmospheric composition in the latest Pleistocene. However, many of the mechanisms that control atmospheric composition and climate are rooted in the oceans. The solution to Pleistocene climate problems therefore requires a coupled ocean-atmosphere approach where ice core data are integrated with marine sediment cores. One of the objectives of Leg 177 was to recover expanded upper Pleistocene sections that could be correlated to ice cores from Greenland and Antarctica. A north-south transect of expanded Pleistocene sections was recovered at Sites 1089 (41°S), 1091 (47°S), 1093 (50°S), and 1094 (53°S) that have suitable resolution and continuity for marine sediment-ice core correlation. Postcruise studies have thus far focused mainly on the end-members of the transect (i.e., Sites 1089 and 1094).
Site 1089 is a continuous, 780-k.y.-long sedimentary sequence taken at the same location as piston core RC11-83, which is a benchmark core for climate studies of the high-latitude South Atlantic over the last 80 k.y. Charles et al. (1996) demonstrated that the 18O of planktonic foraminifers at this location mimics the D in the Vostok ice. In addition, benthic 13C variations in the same core resemble the oxygen isotope record of Greenland ice cores. The sediment record of Site 1089 can be correlated, therefore, to two polar ice cores located in opposite hemispheres.
By comparing planktonic 18O and benthic 13C in the same core, Charles et al. (1996) were able to determine the phase relationship between surface ocean temperature changes in the Southern Ocean and deep-ocean circulation changes controlled by changes in the high-latitude North Atlantic. They found that Northern Hemisphere climate fluctuations lagged those of the Southern Ocean by 1.5 k.y. over the past 80 k.y. This finding was validated by Blunier et al. (1998), who came to nearly the same conclusion by correlating Greenland and Antarctic ice cores using methane fluctuations.
Ninnemann et al. (1999) extended the study of Charles et al. (1996) through the last glacial cycle using ODP Site 1089. They demonstrated that millennial-scale variability in benthic 13C could be correlated to 18O of the Greenland ice cores, and millennial-scale variability in planktonic 18O could be tied to the D signal in Vostok for the last 120 k.y. (Fig. F23). Because expressions of both Greenland and Vostok ice cores are contained in Site 1089 signals, the site offers the opportunity of studying the relative phasing of interhemispheric climate change. Models suggest that SST between the high-latitude North and South Atlantic should be antiphase (Manabe and Stouffer, 1997) because the South Atlantic transports water and heat to the North Atlantic across the equator to balance the water lost due to NADW formation. Broecker (1998) coined this response the "bipolar seesaw." Ninnemann et al. (1999) tested the bipolar seesaw hypothesis using Site 1089. They found a greater poleward extent of warmer surface waters in the South Atlantic with reduced NADW, but it's not straightforward to assess whether the millennial-scale SSTs were actually antiphase. The Site 1089 data are in agreement with at least the sense of the modeling results; however, the phase relationship between thermohaline circulation changes and Southern Hemisphere climate varies at different timescales (e.g., millennial vs. glacial-to-interglacial). Benthic 13C at Site 1089 indicates that millennial-scale variability in thermohaline circulation has been a persistent feature of the climate system even during the last interglacial period (MIS 5). Several prominent reductions in NADW occurred during MIS 5, consistent with findings from the North Atlantic (Oppo et al., 2001; Bianchi et al., 2001). This suggests that large ice sheets and meltwater are not always necessary to trigger abrupt changes in the ocean's deep conveyor.
The long Vostok ice core extends back to ~450 ka through the last four climatic cycles (Petit et al., 1999). By correlating millennial-scale oscillation in D in the Vostok ice core with the 18O records of Globigerina bulloides and Neogloboquadrina pachyderma at Site 1089, Mortyn et al. (submitted [N5]) built a marine sediment analog to the Vostok record for the past 400 k.y. In doing so, they developed a new timescale for Vostok that is tied to the orbitally tuned chronology of Site 1089. The chronology is similar to that derived independently by Shackleton (2000) for the Vostok ice core on the basis of tuning the 18O of atmospheric oxygen to insolation forcing. The Site 1089-Vostok correlation is used to test the phasing of ocean and atmospheric variables at glacial terminations. As observed in other studies, the minimum values in planktonic 18O (warmest surface water temperatures) during terminations led the minimum in benthic 18O by several thousands of years, confirming the "lead" of Southern Ocean temperature with respect to ice volume (Imbrie et al., 1992; Charles et al., 1996; Pichon et al., 1992). Over the last four terminations, abrupt changes in the chemistry and temperature of the deep Southern Ocean were synchronous with changes in atmospheric pCO2 and polar air temperatures. This supports a physical rather than biological mechanism for glacial-to-interglacial pCO2 variations and is consistent with recent models that emphasize the role of sea ice and deep ocean ventilation in controlling atmospheric pCO2 (Toggweiler, 1999; Stephens and Keeling, 2000; Keeling and Stephens, 2001).
Kanfoush et al. (2002) proposed that Site 1094 could be correlated to Vostok by matching percent ash in South Atlantic sediments with sodium concentrations in Vostok (Fig. F16). Both parameters are thought to be influenced by the areal extent of sea ice in the South Atlantic. The correlation yields close agreement between changes in Vostok inferred temperature and the marine 18O record on both orbital and millennial timescales. Natural gamma radiation (NGR) at Site 1094 exhibits similarities with dust concentration in Vostok, which originate from the Patagonian region of South America. This may indicate that the NGR signal is controlled by fine-grained terrigenous influx (rich in U and Th) that is transported to Site 1094 by winds.
Kunz-Pirrung et al. (2002) and Bianchi and Gersonde (2002) compared the diatom-based SSST records from Site 1093 (MIS 11 and 10) and Site 1094 (MIS 6 and 5) to Vostok estimated temperature at the level of inversion (Ti) (Fig. F15). The two signals are very similar in both shape and amplitude, suggesting that the seasonally ice-free areas south of the Polar Front were a source region of water vapor to the Vostok ice core. SSST estimates during MIS 5 at Site 1094 are in good agreement with Vostok temperature change after correction for deuterium excess (Cuffey and Vimeux, 2001). Ice disturbance near the base of the Vostok ice core has led paleoclimatologists to question whether temperature and pCO2 estimates are reliable for MIS 11. Correlation of Sites 1093 and 1094 (Kunz-Pirrung et al., 2002), as well as Site 1089 (Hodell et al., in press), to the Vostok ice core suggests that peak warmth conditions of MIS 11 were captured near the base of the ice core. Peak warmth in both South Atlantic surface waters and Vostok temperatures during MIS 11 were similar to the Holocene, indicating that temperature in the high-latitude Southern Hemisphere during MIS 11 was not substantially warmer than other interglacials of the late Pleistocene (Hodell et al., 2000).
