Glacial sedimentation is preserved on the outer continental shelf and slope as tills and diamict within the prograded wedge, on the upper continental rise within hemipelagic sediment drifts, and on the abyssal plain as turbidite sands. The aims of Leg 178 were to sample the first two of these environments as well as a pocket of postglacial sediment on the inner shelf.
The inner continental shelf is a region of rugged topography, underlain by hard rock. Some of the islands exceed 2000 m in height. Water depths in the channels and fjords exceed 1000 m in places, and the rugged submarine topography results from glacial overdeepening of lines of weakness, structurally controlled. Radiocarbon ages suggest that the inner shelf was uncovered by 5-6 ka (Clapperton and Sugden, 1982; Harden et al., 1992) or earlier. Sidescan sonar records from the inner shelf show mostly bare, jointed rock, with thin sediment ponds confined to local troughs (Pudsey et al., 1994). In Palmer Deep, the sediment cover is significantly thicker and may well be older than 6 ka (Leventer et al., 1996; Rebesco et al., 1998). Rapidly deposited modern proglacial sediment occupies the floor of some inshore fjords (Domack and McClennen, 1996).
The middle and outer shelves, separated by the MSH, are much smoother than the inner shelf but exceed 600 m water depth in places (Fig. F5); some sidescan sonar records from the middle shelf show an erosional topography reflecting preglacial sedimentary geology. Side-scan sonar and 3.5-kHz profiles on the middle and outer shelf west of Anvers Island (Pudsey et al., 1994) show two fabrics. An older set of striations is interpreted as glacial flutes, lying along glacial troughs emerging from the main deep exits from the inner shelf, which were the loci of ice streams. On these, in the shallower outer shelf area, is superimposed a later set of more randomly oriented furrows, caused by icebergs moving under the influence of tides and shelf currents following the retreat of the ice-shelf front. The outer shelf is generally too deep (300-500 m) for any icebergs now produced in this region to ground there. Overlying these features, and concealing them inshore, is a thin (<5 m offshore and <20 m inshore) layer of pelagic diatomaceous mud above a proximal glaciomarine diamict. Interglacial deposition appears greater inshore than at the shelf edge, presumably as a result of current maxima around the shelf edge, an inshore source of suspended terrigenous fines, and perhaps more stable water mass stratification inshore, which would encourage diatom blooms. Interglacial deposition on the upper continental slope is probably slow for similar reasons. Cores from the middle and outer shelf (Pope and Anderson, 1992; Pudsey et al., 1994) give about 11-ka radiocarbon ages for horizons close to the base of the diatomaceous facies, dating the onset of open marine conditions. Retreat across the middle and outer shelf appears to have been rapid.
The middle and outer shelf of the Antarctic Peninsula is typical of Antarctic continental shelves, in deepening inshore from a shallow shelf edge and in having a thin, unconsolidated glaciomarine surface sediment layer overlying a diamict. In other places, too, similar radiocarbon ages have been determined (e.g., Prydz Bay [Domack et al., 1991a]). These features all reflect the retreat since glacial maximum of an ice sheet grounded almost everywhere to the continental shelf edge.
In general terms, the ice sheet covering the continental shelf during glacial periods is assumed to have been subject to the same range of constraints as modern ice sheets. Ice-stream flow was governed by the surface slope and the basal boundary condition. Water and thus, in most cases, low-strength dilated till, were available to enable fast flow. We assume that the ice sheet was generally thick and slow moving on the inner shelf, where basal sediment supply would have been low, requiring erosion of hard rock. Much of the seabed of the middle shelf is softer and more erodable, particularly during early advance when interglacial biogenic muds are available, but ultimately the till supply must come from the inner shelf or from middle-shelf bedrock erosion. In contrast, the outer shelf is essentially depositional and floored by thick tills. We see the outer shelf as the site of low-profile, fast-flowing ice streams because of the abundant till supply. These ice streams transported till to the continental shelf edge, building the progradational lobes. During the present interglacial, an ice-covered environment resembling the glacial-age outer shelf is difficult to find; perhaps, however, the "lightly grounded" area of the Ross Sea described by Shabtaie and Bentley (1987) is analogous.
The Antarctic Peninsula continental shelf has been the object of intense study, using a dense network of single-channel and multichannel seismic reflection profiles largely within the ANTOSTRAT framework. (A composite track chart is depicted in Fig. F10, and details of the origins of data at or close to sites are given in the "Explanatory Notes" chapter. See also "Appendix," and Fig. AF1, both in the "Leg 178 Summary" chapter) The data show that the outer shelf is underlain by a thick prograded wedge of sediments, containing deeper reflectors that possess the same characteristics as the present seafloor (inward slope, high topset reflectivity, and high acoustic velocity). These characteristics, together with a steep continental paleoslope and a gradually prograding paleoshelf break that nowhere shows signs of migration of the effective level of erosion from the present ice-stream base to a nonglacial (wave base) level, imply a glacial origin (Larter and Barker, 1989). The glacial-age sediments have been described and interpreted in many papers (e.g., Larter and Barker, 1989, 1991b; Anderson et al., 1990, 1991; Larter and Cunningham, 1993; Barker, 1995a; Bart and Anderson, 1995; Vanneste and Larter, 1995; Larter et al., 1997).
The prograded sediments along the Antarctic Peninsula outer shelf are focused into four depositional lobes (Figs. F8, F11), where both progradation of the paleoshelf edge and depression of the outer paleoshelf are maximal. Four shelf sediment sequence groups have been identified (Larter and Barker, 1989, 1991b; Larter and Cunningham, 1993; Larter et al., 1997; Fig. F12). S1 and S2 are clearly glacial. The MCS data do not resolve individual till units, so subdivision at the level of a single glacial cycle is impossible. Also, subsequent glacial erosion tends to remove topset beds, leading to the apparent merger of deposits corresponding to a single cycle into a larger sequence. Nevertheless, several sequences were identified within both S1 and S2. At the S2/S1 boundary, erosion has removed S2 topsets and truncated foresets. This characteristic relationship between S1 and S2 can be clearly seen along the entire margin to at least 105ºW in the Amundsen Sea (Nitsche et al., 1997), even though S1 is not always continuous between depositional lobes. Progradation in S1 and S2 is greatest within the lobes, but at least some progradation of these two sequence groups is seen everywhere along the margin. Isostatic response to loading by S1 and S2 topsets and foresets appears fairly local along the Antarctic Peninsula margin, as would be expected for relatively young oceanic lithosphere; the S3/S2 boundary, therefore, lies far deeper beneath the lobes than in an interlobe area, where it is more accessible to drilling.
Deep-tow boomer data collected on the outer shelf off Anvers Island (Vanneste and Larter, 1995) provide more detail of the uppermost units of S1, both along the axis of a lobe and in an interlobe area (between L1 and L2). Till units appear to be 10 to 50 m thick and to vary in thickness laterally. They are bounded by reflectors much smoother than the present seabed, which is disturbed by iceberg plow marks 5-10 m deep and as much as 50 m across. It seems likely that in many areas the present rough seabed topography will be smoothed or planed off during the next ice advance. The deep-tow boomer data identify an area where the youngest till body appears not to have reached the continental shelf edge (Fig. F13). This body has a sloping front, dipping seaward at >2º and showing faint internal reflectors that dip similarly. Vanneste and Larter (1995) suggest that this slope, at least 40 m high and 1 km long (its upper extent is confused by iceberg plowing), might be the normal prograding foreset slope of the till, a glacial continental shelf environment where the products of gravitational instability might develop within a till, close to the grounding line.
The continental slope is very steep, both on and between the lobes, reaching 17º in places. Deep-tow boomer profiles of the upper slope (Vanneste and Larter, 1995) show no penetration but are able to help define slope morphology. They show an uppermost region full of small-scale headwall scarps and slumped blocks, suggesting slope failure. Lower down are seen small gullies, 40-80 m deep and 400-1000 m apart. These do not appear to merge or enlarge downslope, and seismic data suggest that the lower slope is smoother than the upper. We suggest that small till-laden debris flows originate in the failure of the uppermost slope, erode the slope gullies, then continue as turbidity currents onto the continental rise (see below), depositing coarse debris on the lower slope. Their small scale is evidenced by the scale of slumping and gullying, and their likely energetic nature is suggested by the steep slope and massive initial load. We observe the slope profile as it is now, some 10-20 k.y. into an interglacial, when this margin is effectively sediment-starved. We assume, however, following the seismic evidence, that there is net progradation of the slope during glacials, when the upper slope is being loaded. Importantly, there is no evidence of large-scale slope failure within the prograded wedge along this margin.
Sequence S3 underlies S2, conformably in most places, and (mostly unconformably) overlies S4. S4 is considered precollision (see "Geology and Tectonics"), undeformed sediment deposited on the outer slope of a subducting margin, overlying an accretionary prism. Its truncation at the S4/S3 boundary is considered to have been caused by postcollision (mostly or entirely preglacial) uplift of the MSH and erosional truncation at wave base. In the appropriate margin segment, the S4/S3 boundary has been traced to the level at DSDP Site 325 associated with the loss of terrigenous deposition on the continental rise. The boundary is in fact diachronous along the margin, following collision.
Sequence 3 has been described as "preglacial." Its nature was unknown before drilling, but it lacks the clear progradational form of S1 and S2 as well as their focusing into depositional lobes. S3 resembles the Type IIA geometry attributed by Cooper et al. (1991) to "early glacial to nonglacial processes on normal water-depth shelves" and often seen beneath obviously glacial sequences. It appears to be of fairly uniform thickness along the margin, continuing to at least 105ºW (Nitsche et al., 1997), and in most sections has a partly conformable boundary with the overlying S2. Given the present effective glacial erosion level, several hundred meters below wave base, it seemed from this conformity that either S3 was completely preglacial or an early glaciation stage. If S3 was preglacial, the depositional environment we were seeing was on a continental slope, deeper than the glacial erosion level. If S3 was glacial and its different geometry is a reflection of an earlier stage of glaciation that does not involve marginal progradation or the creation of a sharp paleoshelf break, then we were seeing a glacial shelf environment.
As already described, unsorted terrigenous sediment is supplied to the uppermost continental slope--abundantly along a broad front (mainly but not entirely from within progradational lobes) during glacial parts of the climate cycle but hardly at all during interglacials. The slope is extremely steep (up to 17º in places) and has steepened with progradation. The slope shows no sign of large-scale slumping, but a part of the upper-slope glacial sediment is transported to the lower slope and rise by debris flow and turbidity current. Deep-tow boomer data from the upper slope, pointing to small-scale instabilities, have already been described (Vanneste and Larter, 1995).
Parts of the upper continental rise have been mapped by GLORIA sidescan survey (Tomlinson et al., 1992), seismic and 3.5-kHz profiles (Rebesco et al., 1996, 1997), and short piston cores (Camerlenghi et al., 1997b; Pudsey and Camerlenghi, 1998). The GLORIA survey of the northeastern part of the rise shows a dendritic channel network at the base of the slope, feeding a small number of substantial channels that continue northwest, beyond GLORIA coverage. By means of this network, each main channel "drains" a section of the slope. The main channels appear to have a longer life span than a single glacial cycle, but some of the small feeder channels may not have. The series of upper-slope gullies may not connect with the channel network on the upper rise. There is no a priori evidence of the balance of turbidity current creation on the slope between glacials (when the uppermost slope is being loaded) and interglacials (when it is not), but the preservation of a glacial-interglacial cyclicity on the rise would suggest that the mean residence time of unstable sediments on the upper slope is short.
The major component of sedimentation on the upper rise is a series of mounds interpreted as hemipelagic drifts (Rebesco et al. 1994, 1996; McGinnis and Hayes, 1994, 1995; Camerlenghi et al., 1997b). Eight drifts, labeled D1-D8, are marked schematically on Figure F11 and appear also on Figure F5. The two largest (D6 and D7) are seen on MCS profile IT92-109, southwest of the Tula FZ and 100-170 km from the base of the continental slope (Fig. F14). These are elongated perpendicular to the margin, are 50 km across, and rise almost 1 km above their surroundings. The maximum sediment thickness beneath the crest of D6 may reach 3 km. D6 and D7 appear as subdued gravity highs on the map of satellite-derived gravity (Fig. F8).
The drifts are asymmetric in section (parallel to the margin), which we believe reflects the influence of a slow southwest boundary current (above). We speculate that the drift sediments are derived from turbidity currents flowing from the continental slope along the channels between the drifts, by downcurrent transport of suspended fines (Fig. F3). D3 and part of D4 lie within the GLORIA coverage (Fig. F15), so that their relationship to the channel pattern is clear: D4 lies downcurrent of Channel IV; D3, between II and III. D3 is low and thin and may be accessible to turbidite overbank deposition. The smaller Drift D2 is confined between Channels I and II and appears on MCS profile AMG 845-04 (Larter and Cunningham, 1993). Beyond the GLORIA coverage, the relationship of drifts to slope is more conjectural. The great majority, however, appear on MCS profiles to be similarly isolated from direct turbidite deposition and maintained by pelagic/hemipelagic deposition alone.
There is some uncertainty about the relative importance in a particular drift of sediment from the "local" margin channel system and from more distal sources. There is a close correspondence between the seismic sections within adjacent drifts (see Rebesco et al., 1996), which may be attributed to either a common distal source or a shared continental glacial climatic history (or both). The drifts become larger to the southwest, which could reflect either a growing distal contribution or growth in the proximal glacial source (as the Antarctic Peninsula becomes broader).
Six seismic sequences have been recognized within the two southwesterly drifts (D6 and D7 [Fig. F8]), which lie on ocean floor aged 35-40 Ma, within the 30-Ma collision zone. Sequence M6 drapes and fills basement topography and thickens from 0.5 s two-way traveltime to 1 s at the base of slope. The upper boundary may be the "uplift unconformity" at that collision segment of the margin. M6 just overlies 25-m.y.-old basement in the next youngest collision segment to the northeast. Reflectors within Sequence M5 onlap the M6/M5 boundary and appear to have been planar and horizontal on deposition, suggesting a turbidite origin. Sequence M4 is of variable thickness and shows the earliest signs of the mounding that characterizes the drifts (though the early asymmetry appears opposite to that of the present day). The upper parts of both M5 and M4 show erosion in places near the base of slope.
Sequence M3 shows the main phase of drift growth (Rebesco et al., 1997). Reflectors in its lower part downlap onto the M4/M3 boundary, suggesting very little deposition away from the drifts, but deposition in its upper part was more even. Correlation with DSDP Site 325 suggests Sequence M3 is younger than 8 Ma. Sequences M2 and M1 show even deposition, except on the steep southwest scarp slopes: there are minor differences between the two sequences in the strengths of concordant reflectors. GLORIA data suggest that the scarp slopes are dissected by gullies.
Three of the drifts (D5-D7) have been described by McGinnis and Hayes (1994, 1995) on the basis of a different set of MCS profiles. They conceive of an originally more continuous, less dissected upper rise than is envisaged here, and yet a more dissected modern margin; they place great emphasis on submarine canyon cutting as a sign of and as a mechanism for its later dissection. McGinnis and Hayes invoke overbank deposition (instead of nepheloid transport) to form the drifts, and they see contour current erosion within the drifts and along the modern margin where we do not. We see the present topography as less extreme, and (on the basis of GLORIA evidence [Fig. F15]) we invoke dissection by small-scale turbidity currents to produce and maintain it. However, part of the disagreement concerns the earlier history of the drifts, which will not be accessible to the planned drilling. The remainder, although of great sedimentological interest, does not affect the value of the drifts as a high-resolution record of glacial history.
The February 1995 site-survey cruise aboard OGS-Explora recovered nine short gravity cores along strike and dip transects of Drift 7 (Camerlenghi et al., 1997b). Alternations of brown bioturbated diatom- and foraminifer-bearing mud with gray laminated barren mud extending across the entire gentle northeast slope of the drift are interpreted as having been produced during glacial cycles. The diatoms indicate a late Quaternary (Thalassiosira lentiginosa zone) age, and the Ba/Al ratio (a proxy for paleoproductivity [Shimmield et al., 1993]) peaks at the core top (= Stage 1) and in a second foraminifer-bearing brown unit downcore assumed to be Stage 5e (Pudsey and Camerlenghi, 1998). This preliminary time scale gives sedimentation rates from 3 to 6 cm/k.y.
The brown bioturbated mud occurs at each core top and in one or more units downcore. It is fine-grained with 0%-10% sand, 20%-30% silt, and 65%-80% clay. The sand fraction includes unsorted terrigenous grains of ice-rafted origin, with calcareous foraminifers and less-common radiolarians and volcanic glass shards. At the core tops there is a diverse radiolarian assemblage, and in the inferred Stage 5 sediment, planktonic and benthic foraminifers are abundant. Diatoms form as much as 30% of the fine fraction.
The gray laminated mud is barren of microfossils except for a trace of diatoms and is not bioturbated. Parallel lamination occurs extensively, with some wavy lamination in the upper parts of the cores, and the mud is very fine grained (<1% sand, 20%-30% silt, and 70%-80% clay). Both sediment types show remarkable lateral continuity across the drift. Magnetic susceptibility profiles and visible features in the cores permit the identification of 25 datum levels in the top few meters, which can be traced over distances of as much as 100 km across the drift.
The dip- and strike-section continuity across the drift shows that the lateral continuity of seismic and 3.5-kHz reflection profiles extends to very high resolution, at least within a single drift. It also appears to verify that terrigenous sediment from the continental slope is supplied in numerous small-scale turbidity currents. The fine grain size supports the idea of slow bottom currents. Similarly, the glacial-interglacial lithologic variation supports the view that instability on the upper slope follows quickly after load, so that the residence time of unstable sediments on the slope is short. This is an important characteristic of the slope-rise transport system because it promises a high-resolution record of glacial-interglacial variation. The biogenic component of interglacial sediments is diverse, magnetic susceptibilities correlate precisely between cores, and remanence appears stable.
These cores apart, there are very few known sediment cores from the upper continental rise. 3.5-kHz profiles show 50-100 m penetration across the drifts, usually a good indicator of fine grain size. British Antarctic Survey Cores PC054 and 055 from close to the margin, in essentially a proximal channel overbank area (also showing good 3.5-kHz penetration) of Drift D3, have a thin biogenic hemipelagic mud overlying laminated silty and clayey mud with silt laminae. The Eltanin cores from the continental rise (Goodell, 1964) are described as silty clays and clayey silts with varying numbers of silt and sand laminae. Diatomaceous clay is described from the top of some, and some cores contain foraminiferal layers. It is impossible to tell which, if any, of the layers are turbidites. Clear turbidite sands are absent from the descriptions; we would expect turbidite sands to flow on past the drifts and to settle on the Bellingshausen abyssal plain to the north.
The importance of the drifts is that they provide a continuous hemipelagic sedimentary section intimately related to the glacial history of the adjacent continent. Sedimentation is likely to have been more continuous than on shelf or slope and also rapid (perhaps exceeding 10 cm/k.y.). Magneto- and biostratigraphic control seem assured, and there is significant glacial-interglacial lithologic variation, which is continuous at high resolution across the drift. The drifts seem likely to contain a high-resolution, drillable record of Antarctic glacial history that can be correlated to the more direct record of the shelf and slope.