AMS radiocarbon dating of pollen concentrates in a karstic lake system

In lake sediments where terrestrial macrofossils are rare or absent, AMS radiocarbon dating of pollen concentrates may represent an important alternative solution for developing a robust and high resolution chronology suitable for Bayesian modelling of age-depth relationships. Here we report an application of the heavy liquid density separation approach (Vandergoes and Prior, Radiocarbon 45:479-492, 2003) to Holocene lake sediments from karstic Lake Sidi Ali, Morocco. In common with many karstic lakes, a significant lake 14C reservoir effect of 450-900 yr is apparent, evidenced by paired dates on terrestrial macrofossils and either aquatic (ostracod) or bulk sediment samples. AMS dating of 23 pollen concentrates alongside laboratory standards (bituminous coal, anthracite, IAEA C5 wood) was undertaken. Concentrates were prepared using a series of sodium polytungstate (SPT) solutions of progressively decreasing density (1.9-1.15 g/cm3) accompanied by microscopic analysis of the resulting residues to allow quantification of the terrestrial pollen content. The best fractions (typically precipitating at 1.4-1.2 g/cm3) yielded dateable samples of 0.5-5 mg (from sediment samples of ∼15 g), with C content typically ∼50% by weight. Terrestrial pollen purity ranges from 29% to 88% (μ = 67%), reflecting the challenge of isolating pollen grains from common aquatic algae, e.g. Pediastrum and Botryococcus. A Poisson-process Bayesian depositional model incorporating radiocarbon (pollen and macrofossil) and 210Pb/137Cs data is employed. As all pollen samples incorporate some non-terrestrial organic matter, we assume an exponential outlier distribution treating each pollen concentrate datum as an old outlier and terminus post quem. This approach yields strong data-model agreement, and differences between the prior and posterior age distributions are furthermore consistent with theoretical offsets anticipated for the known reservoir ages and sample-specific terrestrial content. This application of the pollen concentrate dating approach reinforces the importance of microscopic inspection of the residues during the separation and sieving stages. Sample specific differences mean that the pollen concentrate preparation cannot be reduced to a simplistic “black box” protocol, and dating and subsequent age-model development must be supported by detailed analysis of the microfossil content of the sediments.