ESR dating of optically bleached quartz grains from Plio-Pleistocene to Holocene coastal dune deposits (Wilderness-Knysna area, South Africa): a comparison with luminescence

Five Plio-Pleistocene to Holocene aeolian quartz samples from the coastal dune deposits of the Wilderness-Knysna area (South Africa) previously dated by OSL were selected for ESR dating. Samples were processed following the Multiple Centre approach and using the Multiple Aliquot Additive dose method. Aluminium (Al) and Titanium (Ti) signals were systematically measured in all samples. Our study shows that ESR results obtained for Middle Pleistocene to Holocene samples may be strongly impacted by (i) the presence of a significant high frequency noise in the ESR spectra acquired for the Ti signals and (ii) the choice of the fitting function employed. In particular, if not taken into account, very noisy spectra can lead to a significant overestimation of the true ESR intensity measured for the Ti–H signal. These sources of uncertainty are however not sufficient to remove the ESR age overestimations. Consequently, our results indicate that the Al and Ti ESR signals of these quartz samples have not been fully reset during their aeolian transport. While this work contributes to improve our understanding of the ESR method applied to quartz grains, and especially of the potential and limitations of the Ti signals, it also provides additional baseline data to illustrate the existing variability among quartz samples of different origins or sedimentary context. Our results are consistent with previous studies by confirming that the Ti–H signal shows the best potential for the evaluation of low dose values (<100 Gy for these samples), whereas it becomes inappropriate for the higher dose range, and the Ti–Li–H (option D sensu Duval and Guilarte, 2015) should be used instead. Beyond the methodological outcome, this ESR dating study also provides a useful addition to the existing chronology of the aeolian deposits in this region. In particular, new (and possibly) finite numerical age results were obtained for the two oldest samples, constraining the aeolianite landward barrier dune and the coversand formations to the MIS 10-8 and Pliocene, respectively.