The scientific aims of the expedition were to investigate environmental changes at South Georgia since the Last Glacial Maximum (around 20,000 years ago) and to place these changes in the wider global context of climate change in Antarctica and the Southern Hemisphere. A better understanding of the nature, timing and environmental impact of past changes will allow better predictions of the impacts that might arise from rapid climate change in the future. The 10 members of the team, supported by the ocean-going ketch Ocean Tramp, spent 3-4 weeks at the island in January-February 2003 and visited a number of key locations on the north coast between Possession Bay and St Andrews Bay (figure 1).

The remote location of South Georgia in the Antarctic realm of the South Atlantic provides a unique opportunity to obtain high-resolution terrestrial paleoclimate records, in a region where such information is sparse. South Georgia is uniquely located to record changes in ocean and atmosphere circulation, as it is situated close to key global climatic boundaries: south of the present-day sub-Antarctic front, ~350 km south of the Antarctic convergence where warm temperate and cold Antarctic waters meet, and ~5º north of the average winter sea ice limit (figure 1). Changes in the position of these boundaries affect the climate on South Georgia profoundly (Hansom and Gordon, 1998).

A key question in the Antarctic is the extent of the ice and the timing of the Last Glacial Maximum and the transition to the Holocene (the warmer interval of the last 11,500 years). Variations in the timing of ice expansion and deglaciation reflect climatic gradients across the region and provide information on the mechanisms of climate change, particularly relevant given the current debate on the relative timing of Northern and Southern Hemisphere climate change (e.g. Blunier and Brook, 2001; Broecker, 2000; Stocker, 2000). The timing and pattern of deglaciation also provide constraints on the timing of relative sea level changes and glacier fluctuations. Contrary to earlier ideas, recent work has suggested that the last ice sheet on South Georgia was of relatively limited extent and that substantial deglaciation may have occurred as early as 18,600 years ago (Rosqvist et al., 1999). This suggests that the island was responding to ‘early’ warming recorded in Antarctica, rather than the later warming shown by deglaciation and vegetation response in southern South America (Bennet et al., 2000).

We are using a combination of approaches to test these ideas and to gain a better understanding of past environmental change at South Georgia. One team recovered sediment cores from lakes on the ice-free peninsulas of the north coast and will apply a range of laboratory techniques to reconstruct in detail the environmental history of the last 20,000 years (Principal investigator: Rosqvist). Another team investigated the controls on glacier behaviour, using historical records of glacier change, meteorological records and computer modeling of glacier responses to climate change. A third team carried out mapping of former glacier limits and marine limits and will date these using a variety of techniques as a means of reconstructing past glacier and sea-level changes.

The aim of the lake coring program is to establish the pattern of climate and other environmental changes by investigating a variety of environmental records preserved in lake sediments. In doing so we are building on previous palaeolimnological investigations (Birnie, 1990; Rosqvist, 1992; Rosqvist et al., 1999; Rosqvist and Schuber, 2003; Rosqvist and Wasell, 1991). Specific objectives of the work were:

• To obtain a new sediment sequence from lakes on the Tönsberg Peninsula, where Rosqvist et al. (1999) obtained a sediment record extending back over the last 18,600 years. It was particularly important to capture the youngest sediments, as these were not covered by the original study.
• To obtain sediment records extending back into glacial sediments from two new sites (Barff and Thatcher Peninsulas), to provide information on the timing of general ice retreat from the Cumberland Bay area.
• To obtain a detailed sediment record covering the last 1 000 years from the pro-glacial Block Lake (Rosqvist and Schuber, 2003).
• To establish relationships between past glacier fluctuations and sea-level changes.

Fieldwork focused on four locations: the Tönsberg Peninsula, the Husvik hinterland (Block Lake), the Barff Peninsula (between Coral Bay and Sandebugten) and Thatcher Peninsula (Maiviken). Coring equipment was deployed from a raft constructed from two inflatable boats and a wooden platform.

In addition to recovering lake sediment cores, a survey of water bodies was carried out to determine basic chemical properties (pH, temperature, electrical conductivity) and isotopic composition (oxygen and hydrogen). Information about local hydrological conditions is essential for effective interpretation of the palaeoenvironmental records. Modern diatom samples were also collected from a range of habitats to provide information on the present-day ecological distribution of diatoms.

All samples have been transported to Sweden and to the UK and are being kept in cold storage prior to analysis. The visual stratigraphy of sediment cores will be described and photographed to provide a permanent record prior to sub-sampling. Multi-proxy analysis of core material will include loss-on-ignition, magnetic susceptibility, grey-scale density, diatoms and pollen analysis. Stable isotope analysis of diatom silica in the cores will form the principal means of establishing a detailed record of past changes in isotopic composition of the lake water. This information will allow the reconstruction of past changes in atmospheric circulation patterns. This technique has previously successfully been applied to lake sediments from northern Scandinavia and South Georgia (Rosqvist et al., 1999; Rosqvist et al., in press; Shemesh et al., 2001). An application to the Natural Environment Research Council (NERC) Isotope Geoscience Steering Committee for financial support for this work has been successful. Modern isotope samples are currently being analysed at the NERC Isotope Geoscience Laboratory. Chronological control for the palaeoenvironmental records will be established using tephrochronology and radiocarbon dating.

The graphs in figure 2 show γ density variations in two lake sediment cores retrieved from Tönsberg peninsula and Barff peninsula, South Georgia. Density was measured at 0.5 cm resolution in a Multi Sensor Core logger (GEOTEK) at Stockholm University, Sweden. Data are plotted against a relative depth scale. Curves reveal transition from high-density grey silty glacial sediments to more organic rich sediments.  This transition occurs between 230 and 220 cm depth in the Tönsberg core and between 120 and 110 cm depth in the Barff core. The Tönsberg lake data represents the lowermost 225 cm of a total of 527 cm retrieved. The Barff core represent the whole postglacial sequence obtained. ¹⁴C (AMS) dating of macrofossils found just above these transitions will be used to determine the timing of ice retreat from the Cumberland East Bay and Stromness Bay. Previous work has shown that the transition at the Tönsberg site occurred prior to 18,600 calendar years BP (Rosqvist et al., 1999).