The recent Antarctic ice sheet plays an important role in the global climate system by influencing the atmospheric and oceanic circulation of our planet (e.g., Barker et al., 1998). Very little, however, is known about the development of the Neogene Antarctic ice sheet and its impact on the environment. Until now, the Neogene glacial and climatic history of Antarctica has been deduced mainly from proxy data obtained at low latitudes, such as oxygen isotopes, which were measured on benthic foraminifers recovered from well-dated and complete marine sedimentary sequences in the pelagic sea. However, the isotope data are ambiguous because they record both the volume of the ice sheets and the temperature of seawater (e.g., Abreu and Anderson, 1998; Barker et al., 1999).
The behavior of the Cenozoic Antarctic ice sheet can best be studied from sediments drilled in proximal glaciomarine settings, because there, direct evidence exists for ice advances and retreats through time. Such settings have been drilled in Prydz Bay and in the Ross Sea. However, core recovery on the Prydz Bay shelf was very poor (Barron, Larsen, et al., 1989; Hambrey et al., 1991). Much better core recovery was achieved by drilling on the continental shelf of McMurdo Sound in the Ross Sea. Cores CIROS-1, CIROS-2, and MSSTS-1 recovered upper Eocene to Quaternary sediments there (Barrett and Scientific Staff, 1985; Barrett, 1986, 1989). More recent drilling off Cape Roberts in McMurdo Sound also recovered upper Eocene to Quaternary sediments (Cape Roberts Science Team, 1998, 1999, 2000). However, dating was problematic for all cores from the shelf areas, and all proximal sequences show evidence for long and repeated hiatuses.
The main objectives of Ocean Drilling Program (ODP) Leg 178 were to recover and investigate complete high-resolution Neogene to Quaternary sedimentary sequences at the Antarctic continental margin and, based on these studies, to reconstruct the evolution of the Antarctic ice sheet and its influence on the global climate system over the past ~10 m.y. The target area was the Bellingshausen Sea, west of the Antarctic Peninsula (Fig. F1). This paper contributes to the understanding of the glacial and climatic history by focusing on the clay mineral assemblages in sediments recovered at ODP Sites 1095 and 1096 (Fig. F1). We use the clay minerals to reconstruct the provenance and transport paths of the sediments in response to glacial dynamics in the Antarctic Peninsula region during both the present and late Neogene to Quaternary times.
The clay mineral types and proportions of the individual clay minerals in marine sediments strongly depend on the climatic conditions on land and on the nature of the source rocks. Therefore, clay minerals are predestined to be indicators of modifications in the environment and contribute to the reconstruction of the climatic history as well as of sedimentary processes (e.g., Chamley, 1989; Weaver, 1989). Clay minerals in sediments from the Southern Ocean are predominantly of detrital origin (e.g., Piper and Pe, 1977; Barker, Kennett, et al., 1988; Setti et al., 1997, 1998). Recent studies of sediment cores from the Antarctic continental margin have demonstrated the value of clay mineral assemblages for deciphering sediment provenance (e.g., Ehrmann et al., 1992; Petschick et al., 1996; Ehrmann, 1998a, 1998b, in press). Within the Antarctic Ocean, clay mineral assemblages also are good tracers for reconstructing the paleoceanography during Neogene and Quaternary times (e.g., Robert and Maillot, 1990; Hambrey et al., 1991; Ehrmann and Mackensen, 1992; Ehrmann et al., 1992; Diekmann et al., 1996; Ehrmann, 1998a).
The Leg 178 sites were drilled along the continental margin in the Bellingshausen Sea. This marginal sea in the Pacific sector of the Southern Ocean is located between 60°W and 100°W and is bounded to the southeast by the Antarctic Peninsula (Fig. F1). Graham Land and Palmer Land, which constitute the mainland of the Antarctic Peninsula, are covered with an ice sheet of variable thickness, whereas the islands west of the Antarctic Peninsula carry local ice caps. The ice drains through valley glaciers and ice streams flowing perpendicular to the Antarctic Peninsula axis and away from the islands' highlands, respectively. The glaciers and ice streams terminate at the coasts as floating glacier tongues, ice cliffs, or small ice shelves (Keys, 1990). Icebergs calved into the Bellingshausen Sea drift with coastal currents across the shelf and slope before they are transferred into the clockwise-flowing Antarctic Circumpolar Current (ACC). The continental shelf west of the Antarctic Peninsula is up to 150 km wide (Fig. F1) and exhibits the typical Antarctic shelf morphology characterized by general landward sloping and overdeepening to an average water depth of 450 m. The shelf can be divided into three physiographic provinces: a mostly shallow inner-shelf province at ~200 m water depth, a mostly sediment-buried "mid-shelf high" (Fig. F1), and an extensive prograded outer shelf (Larter and Barker, 1989; Pudsey et al., 1994; Larter et al., 1997; Rebesco et al., 1998). Progradation is focused in discrete depositional centers ("lobes"). Broad troughs, like Marguerite Trough (Fig. F1), cross the inner shelf and the mid-shelf high and terminate between the lobes. The troughs are supposed to represent the paleo-pathways of thick, grounded ice streams (Pope and Anderson, 1992; Pudsey et al., 1994; Rebesco et al., 1998). The genesis and morphology of the shelf structures are explained by a complex interaction of climatically induced erosion and sedimentation processes with isostatic and tectonic subsidence during Neogene and Quaternary times (Larter and Barker, 1991; Larter et al., 1997; Rebesco et al., 1998).
The upper continental slope is very steep beyond the lobes (up to 17°) but more gentle beyond the glacial trough mouths (Vanneste and Larter, 1995; Rebesco et al., 1998). The lower slope is smooth (Larter and Cunningham, 1993). Dendritic tributary gullies, up to 1 km wide and 75 m deep, start at the foot of the slope below the lobes and converge into up to 5-km-wide and 150-m-deep meandering channels (Tomlinson et al., 1992; Rebesco et al., 1998). These deeply incised main channels are oriented perpendicular to the continental slope and cross the continental rise towards the Bellingshausen abyssal plain. On the upper continental rise, they flow around nine mound-shaped sediment bodies interpreted as asymmetric drifts (Fig. F1) (Rebesco et al., 1996, 1997, 1998). The drifts are 100-300 km long, 50-100 km wide, and have an elevation of up to 900 m above the seafloor. Sediment waves, dunes, and levees associated with the main channels were reported from the lower continental rise (Tucholke, 1977). Deep-sea fans with abyssal hills along their northern boundary cover the Bellingshausen abyssal plain (Vanney and Johnson, 1976; Tucholke, 1977; Wright et al., 1983; Anderson, 1990).
Sites 1095 and 1096 are located just within the Antarctic Zone of the Southern Ocean. The Antarctic Zone is bounded by the Polar Front in the north and by the southern boundary of the ACC in the south (Orsi et al., 1995). Because of seasonal sea ice coverage, open-water conditions prevail for only ~3 mo during spring and summer at Sites 1095 and 1096 (Parkinson, 1994). Warm Circumpolar Deep Water (CDW) upwells in the Bellingshausen Sea near the continental slope of the Antarctic Peninsula and intrudes onto the shelf underneath Antarctic Surface Water (AASW) (Talbot, 1988; Hofmann et al., 1996).
The circulation pattern on the shelf west of the Antarctic Peninsula is quite complex, with surface currents mainly toward the southwest. Hofmann et al. (1996) and Smith et al. (1999) assume the existence of weak cyclonic gyres located in the Bransfield Strait, off Anvers Island, and north of Marguerite Bay. A slow southwest current adjacent to the coast of the Antarctic Peninsula and a northeast countercurrent farther offshore are associated with these gyres. Northwest of the South Shetland Islands, information regarding surface- and deep-water circulation along the continental slope and rise is contradictory. Stationary hydrographic data by Orsi et al. (1995) suggest southwestward flow only extending from the shelf edge down to ~2400 m water depth, whereas current measurements by Nowlin and Zenk (1988) indicate persistently southwestward near-bottom flow along the continental margin down to ~4000 m water depth. No evidence for a slope current was found west of ~63°W (Hofmann et al., 1996; Whitworth III et al., 1998).
On the flanks of Drift 7 near Sites 1095 and 1096, Camerlenghi et al. (1997) detected a weak bottom-water flow over the upper continental rise using two moored current meters. The measured current speed 8 m above the seabed was an average of 6.2 cm/s and never exceeded 20 cm/s during the 10-mo measuring period. The direction of bottom-water flow followed the bathymetric contours. Thus, the bottom current was directed to the west at the drift's gentle northeastern side and toward the continental slope at its southwestern flank. The potential temperature of the bottom water is comparable to deep-water temperatures measured at the South Shetland margin. The bottom water in this area is derived from Weddell Sea Deep Water (WSDW), which leaves the Weddell Sea through topographic gaps in the Scotia Ridge (Nowlin and Zenk, 1988).
The Antarctic Peninsula predominantly consists of calc-alkaline plutonic and volcanic rocks of a deeply eroded magmatic arc belonging to the Andean Orogen. The magmatic arc formed from Late Jurassic to late Tertiary as a result of the subduction of proto-Pacific oceanic lithosphere beneath the continental margin of Gondwana (Thomson et al., 1983; Davey, 1990; Elliot, 1997). Along the Pacific side of the Antarctic Peninsula, subduction ceased progressively northeastward with time, as offset sections of spreading centers collided with the trench (e.g., Larter and Barker, 1991; Larter et al., 1997).
Magmatic activity associated with the Andean Orogeny started during the middle Triassic with vast intrusions of felsic to mafic plutonic rocks ("Antarctic Peninsula Batholith") and continued until the Neogene. Thereby, centers of magmatic activity shifted both to the north and in a westerly direction (Thomson and Pankhurst, 1983; Pirrie, 1991). Calc-alkaline volcanic rocks ("Antarctic Peninsula Volcanic Group") were extruded from the middle Jurassic to the Pliocene (Thomson and Pankhurst, 1983; Smellie, 1990). Miocene to Holocene volcanism concentrated on offshore islands west of the Antarctic Peninsula as well as east of northern Graham Land, thereby exhibiting alkaline affinities (Elliot, 1997; Smellie, 1999). Still-active volcanism occurs in the area of the South Shetland Islands in response to the slow subduction of Pacific lithosphere beneath the northern Antarctic Peninsula at the South Shetland Trench and to the extension-related opening of the Bransfield Strait as a backarc basin (Davey, 1990; Elliot, 1997; Smellie, 1999).
The Andean Orogeny led to the development of a Jurassic to Cretaceous accretionary complex exposed on Alexander Island (Macdonald and Butterworth, 1990). Mesozoic forearc basins, which developed on the Pacific flank of the Antarctic Peninsula, are exposed on several islands adjacent to Graham Land (Macdonald and Butterworth, 1990; Elliot, 1997). Forearc basin strata are also assumed to constitute much of the shelf (Larter et al., 1997). Middle Jurassic to Eocene backarc basins formed in eastern Palmer Land and on islands east of Graham Land (Macdonald and Butterworth, 1990; Pirrie, 1991; Elliot, 1997). The forearc and backarc basins were mainly filled with thick clastic and volcanogenic marine to terrestrial sequences, which predominantly contain debris supplied from the magmatic arc rocks.
The pre-Jurassic basement of the magmatic arc includes deformed metasedimentary and crystalline rocks of an older Paleozoic to Triassic consuming plate margin (Thomson and Pankhurst, 1983; Elliot, 1997; Loske et al., 1998). The basement comprises upper Paleozoic to lower Mesozoic very low grade to greenschist facies sandstones and shales that are mainly exposed in northern Graham Land. Lower Paleozoic granitoids and gneisses, which were affected by late Paleozoic/early Mesozoic amphibolite facies metamorphic events, occur in scattered outcrops in southern Graham Land and Palmer Land (Elliot, 1997; Loske et al., 1998).