Age modelling for Pleistocene lake sediments: A comparison of methods from the Andean Fúquene Basin (Colombia) case study

• Lake sediments from Colombia are investigated over 2 glacial–interglacial cycles.
• The temperature record shows a 60-year temporal resolution.
• Age models, developed on the basis of 4 different methodologies, are compared.
• Radiocarbon dating of tropical lake sediments is discussed.
• Implications for developing age models for long continental records are discussed.

Challenges and pitfalls for developing age models for long lacustrine sedimentary records are discussed and a comparison is made between radiocarbon dating, visual curve matching, and frequency analysis in the depth domain in combination with cyclostratigraphy. A core section of the high resolution 284-ka long temperature record developed from Lake Fúquene in the Northern Andes is used to explore four different age models (a–d). (a) A model based on 46 AMS 14C dates of bulk sediment is hampered by low concentrations of organic carbon. (b) A model based on the comparison of the radiocarbon dated pollen record to the well-established record from Cariaco Basin using curve matching and visual tie points. For the upper 26 m of the core this approach yields an age interval of 28–59.5 ka. (c) Another age model is based on curve matching and the Intcal09 radiocarbon calibration curve, yielding an age range of 22.5–80.4 ka for the same core interval. (d) Finally, a model is developed based on spectral analysis in the depth domain of the temperature-related altitudinal migrations of the upper forest line. This method identifies periodicities without a pre-conceived idea of age. The main frequency of 9.07 m appears to reflect the 41-kyr orbital signal of obliquity, which is tuned to the filtered 41-kyr temperature signal from the well-constrained LR04 marine benthic ∂18O stack record (Lisiecki and Raymo, 2005). Using this last age-modelling approach, the upper 26 m of core Fq-9C yields a temporal interval of 27–133 ka. Problems arising from radiocarbon dating carbon poor sediments from a large lake are addressed and the visual curve matching approach is compared to the analysis of cyclic changes in sediment records in developing an age model. We conclude that the frequency analysis and cyclostratigraphy model is the most reliable one of the four approaches. These results show that cyclostratigraphy may provide a useful method for developing an age model for long terrestrial records including multiple orbital cycles.