Quantitative climate reconstructions are fundamental to understand long-term trends in natural climate variability and to test climate models used to predict future climate change. Recent advances in molecular geochemistry have led to calibrations using glycerol dialkyl glycerol tetraethers (GDGTs), a group of temperature-sensitive membrane lipids found in Archaea and bacteria. GDGTs have been used to construct temperature indices for oceans (TEX₈₆ index) and soils (MBT/CBT index). The aim of this study is to examine GDGT-temperature relationships and assess the potential of constructing a GDGT-based palaeo-thermometer for lakes. We examine GDGT-temperature relationships using core top sediments from 90 lakes across a north–south transect from the Scandinavian Arctic to Antarctica including sites from Finland, Sweden, Siberia, the UK, Austria, Turkey, Ethiopia, Uganda, Chile, South Georgia and the Antarctic Peninsula. We examine a suite of 15 GDGTs, including compounds used in the TEX₈₆ and MBT/CBT indices and reflecting the broad range of GDGT inputs to small lake systems.

GDGTs are present in varying proportions in all lakes examined. The TEX₈₆ index is not applicable to our sites because of the large relative proportions of soil derived and methanogenic components. Similarly, the MBT/CBT index is also not applicable and predicts temperatures considerably lower than those measured. We examine relationships between individual GDGT compounds and temperature, pH, conductivity and water depth. Temperature accounts for a large and statistically independent fraction of variation in branched GDGT composition. We propose a GDGT-temperature regression model with high accuracy and precision (R² = 0.88; RMSE = 2.0 °C; RMSEP = 2.1 °C) for use in lakes based on a subset of branched GDGT compounds and highlight the potential of this new method for reconstructing past temperatures using lake sediments.