The mode of transport within and the thermal regime of a glacier are thought to be important in determining pebble shape. Boulton (1978) examined pebbles from both temperate and polythermal glaciers and from different zones within glaciers. Fields for supraglacial/englacial, and basal debris, and lodgement till were defined by circumscribing the area in which 90% of the clasts in each category plotted on the sphericity vs. roundness graph. Boulton (1978) found that clasts that had fallen onto the glacier and were transported near the surface as supraglacial/englacial debris tend to retain their original shape and roundness because they are not subject to the abrasion and rotation that occurs at the bed of the glacier. Sphericity values for pebbles transported supraglacially are varied, but they are very angular, that is, they have low visual roundness (Boulton, 1978). Clasts eroded from the bed of the glacier and dragged along at the glacier sole (basal debris) tend to have distinctive surface features, intermediate roundness values, and slightly higher sphericity than supraglacial and englacial debris (Fig. F8). Clasts in the bed of the glacier that are deposited as lodgement till, not picked up and transported by the ice, tend to be abraded into faceted, more streamlined shapes with higher roundness values (Boulton, 1978).
Glacial thermal regime may also affect the shape of clasts (Hambrey, 1994; Kuhn et al., 1993). Kuhn et al. (1993) found that glaciers in very cold regions such as the Weddell Sea and the Lazarev Sea produce more angular clasts and cautioned against comparing temperate glaciers in the Northern Hemisphere with the extremely cold glaciers of Antarctica. Different climatic regimes influence the thermal characteristics of glaciers in a given region, which, in turn, may affect the amount of pebbles in each zone of transport (Hambrey, 1994). In the case of warm or temperate glaciers, meltwater flows to the bed of the glacier and facilitates sliding and erosion at the bed. The water also carries basal sediments away from the glacier. In the case of cold glaciers, if meltwater exists, it is surficial and usually seasonal (Hambrey, 1994). Movement is slower for the colder glacier, and there is less erosion of the bed. The basal layer of sediment, however, is thicker than that of the temperate glaciers (Hambrey, 1994). In Boulton's (1978) study, the Breidamerkurjökull in southeastern Iceland and the Søre Buchananisen in Spitsbergen correspond to the temperate and cold glaciers, respectively. Currently, the Antarctic Peninsula is warmer than the rest of the continent and has a moderate diversity of glacial types, including valley glaciers, glacier tongues, and outlet glaciers (Anderson et al., 1991). The northern Antarctic Peninsula is considered to be subpolar, whereas the rest of the continent is polar.
Icebergs calved from different types of glaciers may deposit sediments from each of the three zones of transport in different proportions (Domack et al., 1980). Ice sheets contribute small amounts of basal debris and only trace amounts of supraglacial/englacial debris (Hambrey, 1994). Basal debris is the first to be melted out of the ice when it enters the water; thus by the time icebergs calve from a large ice shelf, they may have very little sediment left (Anderson, 1999). Outlet and valley glaciers release a larger volume of sediment because they move faster and carry more debris, particularly supraglacial debris, than do ice sheets. These smaller glaciers carry a substantial amount of basal debris and varied amounts of supraglacial and englacial debris. However, their icebergs may not travel as far from the continent as large tabular icebergs from ice shelves.
The Antarctic Peninsula is the most likely source for dropstones found at Sites 1095, 1096, and 1101 based on the lithologies present and the proximity of the drill sites to the peninsula. The Scotia Arc region including the Antarctic Peninsula has a complex history of accretion and was an active plate margin both before and after the breakup of Gondwana (Barker et al., 1991). Greenschist to amphibolite facies metasedimentary and metavolcanic rocks, including but not limited to deformed greywacke, conglomerate, greenschist, local blueschist, mafic lava, and undivided metasedimentary rocks on Alexander Island, represent a subduction complex accreted onto the peninsula before the breakup of Gondwana (Barker et al., 1991). Accreted forearc marine and volcaniclastic sedimentary units are also found on Alexander Island. Magmatic arc and subduction related lithologies such as calc-alkaline igneous rocks, rhyolite, andesite, and minor amounts of basalt and volcaniclastics are found on the peninsula (Barker et al., 1991). An increasing percentage of volcanic and calc-alkaline intrusive lithologies from Units 2 to 1, and a decreasing percentage of metamorphic lithologies, indicate a change in sediment provenance. The changes in lithology may reflect a lateral shift of source region, from the deformed accretionary prism (map Unit 3c; Barker et al., 1991), which now crops out extensively on Alexander Island, to younger intrusive complexes (map Units 9, 10a, 10b, and 17d), which now crop out on the peninsula, or a change in the level of erosion.
The overall shape of pebbles in our study has not changed significantly over the last 3 m.y. This probably reflects the fact that the major pebble lithologies identified in this study are fairly uniform over time. Foliated, bedded, or jointed rocks do not roll at the glacial bed as easily as massive rocks (Boulton, 1978); instead they tend to slide along the bed and form disc or blade shapes. Granite, granitic gneiss, and unfoliated volcanic tuffs tend to form rods and blades. Other lithologies tend to have foliation, bedding, or flow banding and are more likely to form flattened disc shapes. Metamorphic lithologies, particularly metavolcanics, were the most common rock type in each unit, which is reflected in the fact that discs are the dominant shape over time.
Sphericity and roundness of pebbles transported by icebergs to the continental rise sites over the last 3.0 m.y. show variability in the mode of glacial transport on the Antarctic Peninsula. There is a shift from a mixed population of pebbles representing transport in supraglacial, englacial, and basal debris zones within Unit 2 to a more restricted population that represents basal transport within Unit 1. Within Unit 2, pebbles are abundant and they represent diverse lithologies within mostly metamorphic and volcanic rock types. Within Unit 1, the total number of ice-rafted pebbles is smaller, as is the IRD MAR at Site 1101, based on the coarse sand fraction (Cowan, Chap 10, this volume). The lithology of pebbles is more restricted in Unit 1. There is an increase in the presence of basalt pebbles between Units 3 and 2, from 1.3% to 10.5%. The transition between Units 2 and 1 is present at 0.76 Ma and is marked by several other changes within the drift sediments. For example, at Site 1095 rounded quartz sand grains are absent from 0.78 to 0.2 Ma, even though the sand accumulation rate is similar in magnitude (Wolf-Welling et al., Chap 15, this volume). At Site 1101, foraminifers are present within the interglacials until 0.76 Ma, after which they are replaced by diatoms (Barker, Camerlenghi, Acton, et al., 1999).
Often interpretations of ice-rafting records and pebble shape data are not clear cut and involve uncertainties (Anderson et al., 1980; Barrett, 1975, 1980; Domack et al., 1980; Kuhn et al., 1993). Several possibilities could account for the shift in the pebble population at the continental rise sites of this study: inundation of the continent by ice so that supraglacial debris is no longer available for transport; change in the glacial regime to more polar conditions, which would reduce the amount of debris and change the size and type of icebergs; long-term changes in iceberg drift tracks around the Antarctic Peninsula; and changes in oceanic conditions that would affect iceberg melting. Several of these conditions may have acted together to produce the shift in the observed variables. For example, rounded quartz sand grains may be produced by subglacial meltwater that is believed to have been more extensive under ice streams that advanced to the shelf edge in the past (Rebesco et al., 1998; Pudsey, 2000). The absence of rounded grains may indicate the absence of meltwater and colder glacial conditions between 0.78 and 0.2 Ma. Models of ice thickness during the late Wisconsinan maximum predict glacial ice up to 2000 m thick on the continental shelf near Marguerite Bay (Fig. F1) (Anderson et al., 1991). This thickness of ice, if extended inland over the continent, would be effective in inundating the topography and cutting off the supply of supraglacial debris from the mountain peaks and valley walls. Both thicker and colder ice would result in a reduced supply of debris available for rafting. Large tabular icebergs also survive greater transport distances and may bypass sites closer to the continent with their debris. We propose the hypothesis that climate conditions on the Antarctic Peninsula cooled and the ice sheet built up to great thickness sometime around 0.76 Ma, shortly after the mid-Pleistocene Climate Transition and the shift to 100-k.y. glacial-interglacial cyclicity (Cowan, Chap. 10, this volume). Although we favor topographic inundation and colder glacial conditions on the continent as the reason for the change in pebble population, we do acknowledge that oceanic changes have also occurred in the vicinity of the drift sites. Today, the Polar Front is close to the Antarctic Peninsula near Drift 4 and Site 1101 (Pudsey, 2000). It was probably closer to the continent between 2.2 and 0.76 Ma, when foraminifer-bearing mud was deposited during the interglacials at Site 1101. Because the pebble results at all three drift sites are similar and previous work in Antarctica did not yield systematic differences in the pebble shape between nearshore and distal sediments (Kuhn et al., 1993), we believe that oceanic influences on iceberg melt rates are not the major factor controlling the pebble shape variations at the drift sites.