Abstract
The erosional landscape of the Jojosi dongas in KwaZulu-Natal, South Africa, exposes accretionary slope deposits that preserve important geological and archaeological information. This landscape was occupied by modern humans during the Stone Age for many thousands of years, as evidenced by the presence of numerous stone artefacts on the surface and interbedded within at least three phases of gully cut-and-fill deposits. A contextualisation of the artefacts and their role for human evolution in southern Africa, along with developing an understanding of the environmental conditions that shaped this inhabited landscape, is only made possible by establishing a robust chronological framework. Here, we use luminescence dating of feldspars to constrain the geochronological framework for the sequence of accretionary hillslope deposition at Jojosi at three sampling locations. Measurements of feldspar single grains show low luminescence sensitivity of the individual grains and a variable proportion of grains in saturation. Summing the luminescence signal of individual grains and creating synthetic aliquots enables us to study the effect of signal averaging on the luminescence sensitivity, signal saturation, and dose distributions. We then compare the results from individual grain measurements and synthetic aliquots to true multi-grain aliquots. To allow a quantification of the results, we apply four different dose models, including the Central Age Model (CAM), the Average Dose Model (ADM), BayLum, and a standardised growth curve (SGC) approach, using an averaged Ln/Tn value interpolated onto the SGC. Doses calculated for the different samples range from ∼ 80 to ∼ 800 Gy and contain 6 % to 67 % saturated grains. We evaluate the performance of the different dose models over this range in equivalent doses and relative number of saturated grains. On average, we find good agreement between the results obtained using the different dose models but observe that samples with a large number of saturated grains impact the consistency of the result. Overall, all dose models and data sets give consistent results below a saturated grain threshold of ∼ 15 %, corresponding to a dose of ∼ 120 Gy in this study. Finally, we favour BayLum for age calculations of the single-grain and multi-grain aliquot data sets, representing the opportunity to refine the chronology by including stratigraphic information in the age calculations. We are able to establish a chronology for the three sampled sections within the Jojosi dongas, constraining erosional and depositional processes from ∼ 100 to ∼ 700 ka, and human occupation of the area spanning from early marine isotope stage (MIS) 5 to late MIS 6.