Deciphering sedimentary archives of Neanderthals and Early Anatomically Modern Humans - A micromorphological study at the key site of Fumane cave, Northern Italy
Prehistory and World Archaeology
Final Report Abstract
The sedimentary sequence of Fumane Cave represents an outstanding archive of environmental change and human occupation in Northern Italy during the Last Glacial, and in particular for the Middle to upper Palaeolithic transition. The DFG funded project offered the opportunity to analyse a large collection of sediment thin sections from this sequence. Based on this micromorphological approach, we aimed at elucidating the microstratigraphy, site formation processes and climate implications of the sequence. The second task was to improve quantitative estimates and selective mapping of sediment components in thin sections by automated image analyses. In the course of the project, several problems were encountered due to Covid 19 related delays and difficulties in accessing the thin sections and laboratories. However, objectives of both tasks could mostly be reached. The detailed micromorphological analyses show significant variation in granulometric composition, structural features of the groundmass and type and amount of pedofeatures. Most thin sections show sandy deposits with various admixtures of gravel and silt. The silt component is derived from dust inputs, but the spatial variation in silt contents overprints possible diachronic dust signal changes. Spatial variation is also observed in the amount of secondary carbonate related with shifts of the dripping zone. We observed a more pronounced expression of silt cappings during accumulation of Units A1, A2, A2R and A6, probably related with increased snow melt during colder periods. Concerning the second task, we elaborated a new procedure of automated image analyses using thin section flatbed scans. The scans are captured using three illumination techniques yielding nine bands per pixel (1200 dpi) as input for a supervised Random-forest classification using the open-source software QGIS and relevant plugins. The procedure allows to automatically detect rock or lithic fragments as well as charcoal, bone, coprolites or plant remnants. In addition, pedofeatures including silt cappings, clay coatings, calcite and hydromorphic nodules can be mapped. Banded or microlaminated fabrics can also be classified and iron depletion zones detected. The procedure allows to estimate the areal extent of these features and to selectively display them. However, misclassifications have to be expected and the analysis requires expert knowledge of a micromorphologist as well as careful checking under the microscope. Overall, the project provided an excellent frame to pursue the detailed micromorphological study at Fumane and to advance in the field of automated thin section analyses.
