The suite of ANTOSTRAT proposals for drilling the Antarctic margin adopted the dual strategy of sampling both the ice-transported diamicts of the continental shelf and slope for a direct but low-resolution record of ice sheet evolution and the derived fine-grained sediments of the continental rise drifts for an indirect but much more detailed record, combining the results from several areas by means of glaciological models of ice sheet development. The major contributions of Leg 178 were to test that strategy and to examine Antarctic Peninsula glacial history. Thus, a discussion of Leg 178 results must consider several topics: lessons for margin drilling, sedimentary processes, and Antarctic Peninsula glacial history. These topics are intertwined, in the sense that the experience of drilling at Leg 178 sites exposed some of the practical difficulties of drawing simple conclusions about paleoclimate, which differ depending upon the depositional environment, the information sought, and the focus of postcruise activity. Discussion is best organized first in terms of depositional environment, then as a consideration of glacial history.
The experience of Leg 178 on the continental shelf was daunting: recovery was very poor at shallow depth and in young sediments and improved only marginally with depth. Downhole logging was only partial, and data interpretation was impeded by poor hole conditions. A magnetostratigraphy was not possible because of the coarse-grained, multidomain nature of the main remanence carriers. Against these difficulties was the recognition that diatoms were incorporated into the tills to an extent sufficient to permit a crude stratigraphic determination, even with poor recovery. The stratigraphy was improved by additional postcruise studies. Most of these difficulties were anticipated and it was known that, even had recovery been perfect, the record of topset sediment accumulation on the outer continental shelf was most probably partial, involving only short periods of rapid deposition and being vulnerable to reerosion. It was clear that only low-resolution questions could be answered by drilling on the shelf.
The low recovery at shallow depth could be explained in terms of the inability of the soft diamict matrix to hold clasts in place during drilling. As the matrix became more consolidated recovery improved, but not to the extent we had hoped. The most likely explanation for low recovery of consolidated topsets on the shelf was a persistent ocean swell, which at times became sufficient to prevent drilling altogether. It was possible to reoccupy a hole, having pulled out for vessel heave or ice, but time was lost and repeated passage of the bottom-hole assembly (BHA) degraded the hole, contributing to premature abandonment and to logging difficulties. Recovery was not improved by varying rotation rate, water pressure, or length cored. The basic RCB drilling tool was used at all shelf sites, for fear of damage to the more delicate APC/XCB tools in these lithologies. The technical limitations, combined with the effects of ocean swell, made fine control of weight-on-bit impossible, which undoubtedly affected recovery. A constant and controllable weight-on-bit was probably a key factor in the high recovery in similar lithologies achieved by the Cape Roberts Project, which drilled in an area remote from ocean swell on fast sea ice using a converted land drill rig with riser (Cape Roberts Science Team, 1998, 1999, 2000). Recovery remote from land might be improved by selecting a time of year when less ocean swell is generated, by changes in drill bit technology, and possibly by active heave compensation. The quality and completeness of log data might be improved by logging while drilling (as undertaken at Site 1167 in Prydz Bay; O'Brien, Cooper, Richter, et al., 2001). It is certainly the case that only low-resolution questions should be asked of shelf drilling. However, answers to such questions could be valuable contributions towards an understanding of Antarctic glacial history.
One additional effect should be noted. ODP Leg 119 in Prydz Bay (Barron, Larsen, et al., 1989), and perhaps common sense, seemed to suggest that till topsets should be more consolidated and indurated than foresets at equivalent burial depth because the topsets would have experienced ice load (even if only an intermittent, tidally driven load by an almost-floating low-profile ice stream). In fact, the tomographic velocity measurements reported here (Tinivella et al., Chap. 16, this volume) suggest the opposite, in that foresets have higher velocities, even at shallow depth. This is potential reassurance for plans to drill foresets where the equivalent topsets are missing, as at the Wilkes Land margin (Fig. F1).
Postcruise work on the rise drift cores has tended to focus on the glacial cycle—not entirely, but the longer-term studies benefit from a better understanding of the shorter-term processes and variations, which we consider here. Before discussing the problems, it should be mentioned that at the rise sites recovery was very good, logging was not perfect but was extensive and complementary to drilling, and the magnetostratigraphy benefitted from the terrigenous, fine grain-sized remanence carriers, continuous deposition, and high paleolatitude (we had also anticipated higher microfossil abundance and better preservation, but that was perhaps unrealistic). The rarity of calcareous microfossils ruled out the determination of a detailed oxygen isotopic stratigraphy, but that was to be expected so far south of the Polar Front. Indeed, the preservation of (albeit limited) calcareous assemblages at the shallower sites was a surprise. In addition, it is clear that many studies of the properties of cored sediments from the rise drifts are at a preliminary stage, so additional insights may emerge.
The failure of spectral analysis of downhole log and core properties at the continental rise drift sites to show the clear dominance of orbital frequencies (Lauer-Leredde et al., Chap. 32, and Pudsey, Chap. 25, both this volume) has at present a wide range of possible explanations. One class of possibility is that the very uneven rates of sedimentation through the glacial cycle (glacial rates appear much higher than interglacial rates at the more proximal sites on the continental rise) significantly degraded or entirely destroyed evidence of the ice sheet's true orbital sensitivity. Another is that the shipboard stratigraphy, used in both studies published here, was sufficiently in error to mask an orbital sensitivity. A third is that for long periods of ice sheet history, ice sheet volume changes were autocyclic rather than orbitally driven. Enmeshed with these are other considerations: that the sedimentary processes or the ice sheet response may have changed with time, that the behavior of the Antarctic Peninsula ice sheet may have been different from that of the Antarctic ice sheet in general, or that some primary parameters may have been modified by diagenesis.
In considering these possibilities, it is important to distinguish between those properties of drift sedimentation that record ocean circulation and those that reflect the behavior of the ice sheet. It seems inherently likely that changes in ocean circulation, which extends simply and directly as far north as the Polar Front and through mixing processes very much farther, should be driven by orbital insolation changes, even if the ice sheet is not. Ocean temperature and (though perhaps not entirely) sea-ice cover (and thus biogenic production and bioturbation) should reflect this oceanic influence. A detailed oxygen isotopic study of the (sparsely) calcareous Pliocene-Pleistocene rise drift sediments (by using the ice volume effect) could examine in detail the time relation between Northern Hemisphere and Antarctic Peninsula cyclicities.
Autocyclic ice sheet behavior is not unreasonable. At present, the large variation in sea level caused by orbitally induced changes in Northern Hemisphere ice sheet volume through the glacial cycle is a major (perhaps the only significant) cause of Antarctic (and Antarctic Peninsula) ice volume change, involving ice sheet grounding line migration to the continental shelf edge during glacial maxima. If, as Huybrechts' (1992, 1993) glaciological model suggests, the volume of the present Antarctic ice sheet is not sensitive to temperature, and was so also for a few million years before the late Pliocene, when the Northern Hemisphere ice sheets were very small or nonexistent during glacial maxima, it is difficult to see how a strong orbital sensitivity could then have existed (this case is argued in more detail by Barker et al., 1999). In such a circumstance, the small Antarctic Peninsula ice sheet would be sensitive to sea level change generated by changes in the volume of the main Antarctic ice sheet, but such changes would not necessarily be orbitally driven. However, the Antarctic Peninsula ice sheet might also be sensitive to local changes in ocean and atmospheric temperature that could be orbitally driven. A possible outcome of this situation would be the presence of orbital frequencies (via the oceanic effect) without their dominance.
If the only cause (of the absence of dominant orbital frequencies in a spectral analysis) was autocyclic ice sheet behavior, then the analysis of late Pliocene and Pleistocene sedimentation (when Northern Hemisphere ice sheets were present as drivers of sea level change) should reveal an orbital sensitivity. Neither spectral study (Lauer-Leredde et al., Chap. 32, and Pudsey, Chap. 25, both this volume) showed this, despite clear nearby evidence to the contrary. Such evidence includes identification of the youngest glacial cycles in the sedimentary record (for example, Fig. F6), Cowan's recognition of orbital cyclicity after 1.9 Ma at Site 1101, and Guyodo et al.'s (2001) correlation between magnetic susceptibility and orbital cyclicity at the same site (which was not examined within either spectral study). The absence of clear evidence of dominant orbital cyclicity in late Pliocene and Pleistocene sedimentation on Drift 7 in the spectral analyses suggests that either the sedimentation rates used in these studies are incorrect or that diagenetic or sedimentary processes are distorting the record. Among such sedimentary processes might be (as generally suspected) a gross disparity in glacial and interglacial sedimentation rates (close to the continental slope) together with varying fractions of each cycle for which the grounding line was at the continental shelf edge, changes with ice temperature of the rate of terrigenous sediment supply to the outer slope, or processes such as channel switching, which are common in low-latitude terrigenous sediment transport systems. The existence of such sedimentary processes does not rule out autocyclic ice sheet behavior, but does make it more difficult to detect.
The simple, ideal circumstance is that both the Antarctic and the Antarctic Peninsula ice sheets have remained sensitive to orbital insolation. Such sensitivity could then be demonstrated by a repeated analysis, with a sufficiently improved stratigraphy or by seeking significant ratios of "orbital" frequencies (e.g., Fischer and Roberts, 1991), that would be independent of sedimentation rate. The same study would also test the third hypothesis, that of essentially autocyclic ice sheet behavior.
The direct evidence of Leg 178 drilling bears on the last 10 m.y. of Antarctic Peninsula glacial history. On the continental shelf, seismic sequence group S3 was penetrated at Sites 1097 and 1103 to 7-8 Ma. On the rise, Drift 7 was penetrated at Site 1095 to ~10 Ma. Both environments show glacial influence to the base of each hole; the onset of glaciation lies deeper in the record. Here, therefore, to provide an up-to-date context for Leg 178 results, we also examine evidence bearing on earlier Antarctic Peninsula glacial history and discuss how it relates to the glacial history of Antarctica as a whole.
It has long been assumed (e.g., Kennett and Barker, 1990; Wise et al., 1992; Barker and Camerlenghi, 1999) that changes in the glacial states of different parts of Antarctica, driven by the same global climate changes, have been essentially in phase but have started from different initial states. For example, Antarctic Peninsula glaciation would always have been less severe than that of the East Antarctic interior, and may have begun later, because of its more northerly location. It may be time for this simplistic view to be reassessed.
Modern Antarctic Peninsula and general Antarctic glacial history are coupled by two powerful factors: the dominantly external drive of sea level change by orbitally induced changes in Northern Hemisphere grounded ice volume and the general circumpolar equalization of ocean temperature and sea-ice cover by the Antarctic Circumpolar Current (ACC). There is some discussion about the time of onset of the ACC (see, for example, Lawver and Gahagan, 1998; Barker, 2001). All estimates make it considerably older than the time extent of recovered Leg 178 sediments (the past 9-10 m.y.), but it may have been younger than the onset of substantial Antarctic glaciation, now generally dated at ~34 Ma (e.g., Wise et al., 1992; Barrett, 1996; Barker et al., 1999; O'Brien, Cooper, Richter, et al., 2001; Zachos et al., 2001). Maximum Northern Hemisphere ice volumes during glacials appear to have grown since ~3 Ma and to have been very large since ~0.8 Ma; Northern Hemisphere ice volume changes seem to have been driven by orbitally induced (Milankovich) insolation changes (e.g., Hays et al., 1976; Ruddiman et al., 1989; Tiedemann et al., 1994). As already discussed, without such external sea level drive, the volume of the earliest Pliocene and late Miocene Antarctic ice sheet may not have been sensitive to orbital cyclicity. Without the thermal equalization effect of the ACC, the climates of different parts of the Antarctic margin may have been more different from each other than they are today.
The suite of ANTOSTRAT drilling proposals used the model of ice sheet evolution developed by Huybrechts (1992, 1993), which examined the effect on ice sheet volume of uniformly increasing the present-day mean annual temperature at sea level around the continent (see also Barker et al., 1999). This model predicts a late-stage development of an Antarctic Peninsula ice sheet, appropriate to its more northerly situation. However, at present the Antarctic Peninsula experiences much greater than the average Antarctic precipitation, perhaps because its 2000-m elevation disrupts tropospheric circulation (Drewry and Morris, 1992). Also, East Antarctic cooling is conveyed to it within the clockwise Weddell Gyre (Reynolds, 1981). It may therefore, while these conditions held, have been more easily glaciated than its latitude would suggest.
There are several suggestions in the onshore regional geology that this is so. More recent studies (Dingle et al., 1997; Dingle and Lavelle, 1998; Troedson and Smellie, 2002; Troedson and Riding, 2002) have greatly refined the interpretation of early glaciations recorded onshore in the South Shetland Islands, off the northwestern Antarctic Peninsula (e.g., Birkenmajer, 1991), but evidence of middle to late Oligocene (26-30 Ma) and earliest Miocene (22-23 Ma) glaciations is confirmed on King George Island. In each case a single glacial episode is proposed, involving subglacial to distal glacial marine deposition, and an intervening late Oligocene (26-24 Ma) interglacial is inferred from associated marine sequences. The presence in these sediments of clasts of rocks outcropping mainly in the Transantarctic and Ellsworth Mountains, beyond the southernmost Weddell Sea (Fig. F1), is interpreted in terms of rafting on or within icebergs originating in a glaciated East Antarctica and circulating within an ancestral Weddell Gyre (see also Larter et al., 1997). The South Shetland Islands onshore sediments represent a shallow continental shelf environment and were subaerially exposed recently by rift-shoulder uplift associated with Bransfield Strait opening. Although it is difficult to compare onshore and offshore records, these sediments reveal a level of local glaciation similar in many ways to that of sequence group S3, in which glacial ice is grounded below sea level. Whether there really were very long glacials and interglacials at that time, or they were much shorter and more numerous but with sediments mostly not preserved, is uncertain. Dingle and Lavelle (1998) suggest an age of 30 Ma for glacial onset in the South Shetland Islands region, based on Sr and Ar isotopic ages for the oldest known glacial deposits exposed onshore. Troedson and Riding (2002) speculatively correlate the second glaciation with the isolated high oxygen isotopic event Mi-1 of the earliest Miocene (Miller et al., 1991; Paul et al., 2000). A third glaciation evident in the onshore South Shetland Islands record, the Legru glaciation (Birkenmajer, 1991), is of uncertain age but is considered younger than the other two (J. Smellie, pers. comm., 2001); otherwise, onshore evidence of paleoclimate is missing until the late Miocene. The only other information relevant to the onset of Antarctic Peninsula glaciation comes from onshore exposures of nonglacial sediment of late Eocene age on Seymour Island on the Weddell Sea margin of the northern Antarctic Peninsula (e.g., Francis, 1991; Dingle and Lavelle, 1998) and a temperate setting for shallow marine sediment of Eocene or early Oligocene age sampled during ODP Leg 113 at Site 696 on the South Orkney microcontinent farther east (e.g., Mohr, 1990). Thus, it is possible that East Antarctica became glaciated earlier than the Antarctic Peninsula, but not certain.
The offshore evidence of glacial onset comes mainly from the distribution of IRD at sites drilled during DSDP Leg 35 (Hollister, Craddock, et al., 1976). Tucholke et al. (1976) noted the earliest evidence of ice rafting in the late early to middle Miocene (15-17 Ma) at DSDP Site 325. The clast petrology accords with a subduction-related origin, thus at a Pacific margin, probably the Antarctic Peninsula. Leaving aside the possibility of downhole contamination by dropstones, this is the earliest offshore evidence of glaciation of any kind for the Antarctic Peninsula. Less directly, working on the seismic reflection character of the continental rise sediment drifts, Rebesco et al. (1997) concluded that the base of their seismic Unit M4, the onset of drift formation that they associated with the onset of glaciation, had an age of ~15 Ma. This and other interpreted ages were based on stratigraphy at DSDP Site 325 and on magnetic anomaly ages along the margin. Results of subsequent Leg 178 drilling on the drifts, while not reaching the base of Unit M4, coincided closely with the estimated ages of overlying units; the base of their Unit M2, at ~650 m close to Site 1096, was provisionally dated at 5 Ma, whereas Hole 1096C reached ~4.7 Ma at 608 mbsf. Similarly, Hole 1095B, which reached ~10 Ma at 570 mbsf, was considered to have just reached the Unit M4/M3 boundary, dated by Rebesco et al. (1997) at 8 Ma. Given these coincidences, a time of glacial onset of ~15 Ma does not seem unreasonable.
There is thus a large discrepancy between the time of onset of Antarctic Peninsula glaciation (defined, arbitrarily, as the onset of sequence group S3-type conditions in which ice is grounded below sea level) as postulated onshore and offshore. One possible explanation for this is that there was an earlier onset throughout the Antarctic Peninsula, at some time around 30 Ma perhaps, and then a return to warmer (nonglacial) conditions immediately after event Mi-1 (in the early Miocene, perhaps at 22 Ma). We need to bear in mind the widespread history of subduction and ridge-crest collision along the Antarctic Peninsula margin between 30 and 10 Ma. Glacial onset might be particularly difficult to identify at an actively subducting margin; oceanic sediments are subducted or deformed within the accretionary prism, and the fate of continental shelf sediments is highly uncertain. The offshore evidence for mid-Miocene glacial onset and glacial conditions (DSDP Site 325, most of the sediment drifts, and the outer continental shelf drilled during ODP Leg 178) comes from parts of the margin that have undergone subduction, culminating in ridge-crest collision, since 30 Ma. This notion of a separate, earlier glacial onset and then a later one is compatible with a simple correspondence to the modern benthic oxygen isotopic record (e.g., Zachos et al., 2001) (Fig. F9), which suggests a warmer intervening early Miocene and/or a smaller ice sheet. Offshore evidence for the earlier onset of Antarctic Peninsula glaciation may possibly be seen in sedimentary geometries on seismic reflection profiles crossing the continental margin farther south (Nitsche, 1997, 2000) where ridge crest subduction was earlier, although in those papers the authors identify only sequence groups that correspond closely to those seen (e.g., Fig. F3) farther north.
In passing, we should note that the onset of glaciation at a particular margin is one of the targets of the ANTOSTRAT proposal suite, in that its unifying glaciological model (Huybrechts, 1992, 1993; Barker et al., 1999) predicted how large an ice sheet would have to be in order to reach a particular part of the Antarctic margin. Leg 178 did not sample glacial onset, but the unusually high precipitation and high elevation of the Antarctic Peninsula make such information less generally applicable and the small size of the Antarctic Peninsula ice sheet, compared with that of East Antarctica, makes it unimportant in global terms. This target was reached by drilling during ODP Leg 188 at Prydz Bay, for example (O'Brien, Cooper, Richter, et al., 2001), where it is much more important.
We come now to the younger part of Antarctic ice sheet evolution, where the evidence from Leg 178 drilling is direct. The record of biogenic opal on the continental rise (Hillenbrand and Fütterer, Chap. 23, this volume) (Fig. F8) shows a moderately cool late Miocene, a warm early Pliocene, and cooling through the late Pliocene toward Pleistocene conditions similar to today's. Throughout this period, IRD (Cowan, Chap. 10; Hassler and Cowan, Chap. 11; Pudsey, Chap. 25, all this volume), clay mineralogy (Hillenbrand and Ehrmann, Chap. 8, this volume) (Fig. F7), and similar studies on the continental rise show the persistence of an Antarctic Peninsula ice sheet over the past 10 m.y., even through the warm early Pliocene, sufficiently large for the grounding line to reach regularly to the shelf edge. Neither the drilling results nor the disposition of glacial sediments on the shelf and slope revealed by seismic reflection profiles shows any sign of prolonged deglaciation during this period. Both on the shelf (where the progradational sequence group S2 was deposited) and on the continental rise drifts, terrigenous sedimentation was greater or more rapid (see the drift site sedimentation rate diagrams of Iwai et al., Chap. 36, this volume; Fig. F5) during the early Pliocene than during the colder late Pliocene and Pleistocene. This supports the observations and interpretation of Barker (1995) that terrigenous deposition is faster because the ice sheet budget is greater; warmer air at the ice margin, holding more water, leads to greater precipitation, which is balanced by the faster flow of warmer, weaker ice. Additional contributary factors might be a reduced area of basal melting (of colder ice) in colder times, and a greater sea-ice extent, making evaporation from the open sea more distant and reducing precipitation on the ice sheet. Such a temperature dependence of terrigenous deposition suggests further that if the Antarctic continent becomes even colder in the future, the margin would become more sediment starved (Barker et al., in press). It supports the suggestion (Barker et al., 1999) that during the late Miocene and Pliocene-Pleistocene, the volume of the Antarctic ice sheet remained essentially insensitive to temperature change.
Onshore evidence of late Miocene to latest Pleistocene Antarctic Peninsula paleoclimate is sparse and widely scattered in occurrence and discontinuous in section, comprising estimates of ice thickness above volcanic effusions and glacial sediments most probably preserved by overlying volcanics (Pirrie et al., 1997; Smellie, 1999) and is essentially compatible with the evidence from drilling.
The decoupling between climate and ice sheet behavior demonstrated by results from the continental rise casts doubt on the shipboard interpretation of the upper Miocene sequence group S3 sampled on the shelf as indicating a "less glacial" environment than the overlying, probably lower Pliocene progradational sequence group S2. The late Miocene interglacials, from the opal evidence (Hillenbrand and Fütterer, Chap. 23, this volume), were cooler than those of the succeeding early Pliocene. Certainly, "less-glacial" depositional environments were among those recorded by sediments recovered from sequence group S3 compared with sequence group S2, but recovery of both sequence groups was extremely low and the wider range of sequence group S3 sediments recovered does not extend outside the range of sediments produced within a modern glacial cycle. A change in the range of recorded environments and the striking difference in the geometries of sequence groups S3 and S2 shown in seismic reflection profiles could also have been affected by nonclimatic factors, such as changes in subglacial topography and accommodation space or sediment facies (source lithology and grain-size distribution), as glacial erosion of the Antarctic Peninsula proceeded, or by a change in sediment consolidation resulting solely from nonhydrostatic load. The longer-term decoupling of climate and ice sheet volume that we observe also supports the possibility of decoupling at the higher frequencies of the orbitally induced (Milankovich) insolation variation or equivalent autocyclic variation, as discussed above, which may have affected what was preserved by changing the relative lengths of periods of glacial and interglacial deposition.
The Antarctic Peninsula climate history summarized and hypothesized above is compared in Figure F9 with the oxygen isotopic data collated and displayed by Zachos et al. (2001). The initial correspondence is good; the early (Oligocene and earliest Miocene) glacial episodes recorded onshore are reflected in general terms in the isotopic variation, whether this variation is a result of changes in temperature or ice sheet volume (for a possible resolution of this ambiguity, see also Lear et al., 2000). This suggests a strong coupling of Antarctic Peninsula glaciation to global climate and probably therefore (e.g., Barker et al., 1999) a sensitivity to orbital insolation changes also through this period. Subsequently, renewed middle Miocene cooling, as shown by IRD onset and drift formation, matches the increase in isotopic ratio evident in Figure F9. However, it is difficult to detect in Figure F9 any isotopic evidence for a "warm" early Pliocene, and the volume of the Antarctic Peninsula ice sheet seems to have remained essentially independent of global climate change over the past 10 m.y. This comparison suggests that the behavior of the entire Antarctic ice sheet may at times have departed from a straightforward volume dependence on global climate change and shows the value and significance of information about the intervening early Miocene Antarctic Peninsula paleoclimate, which may have been warmer, possibly nonglacial. The inner shelf basins of the Pacific margin are a possible source of middle and lower Miocene sediments.