Determining the diagenetic paths of archaeofaunal assemblages and their palaeoecology through artificial intelligence: an application to Oldowan sites from Olduvai Gorge (Tanzania)

Abstract

The implementation of deep-learning methods to the taphonomic analysis of the microscopic modification of bone-surface modifications exposed to different chemical diagenetic pathways can effectively discriminate between acidic and alkaline soil properties, indirectly reflecting different ecological conditions. Here we use this novel method to assess the sedimentary conditions of two of the oldest Oldowan archaeofaunal records (DS and PTK, Bed I) from Olduvai Gorge Bed I in Tanzania. We show how the results support different diagenetic conditions for both penecontemporaneous sites, which are appropriate for their respective locations on the palaeolandscape to which they belonged. We also show how geochemical analyses of the clay deposit that embedded both sites indicate a similar soil pH divergence. PTK was formed on an alluvial sloping surface affected by rills but draining efficiently, which resulted in alkaline soil conditions, that optimised bone-surface preservation. DS occurred in a more depressed area that underwent intermittent flooding, affecting soil chemistry by creating more acidic conditions. This impacted on bone surfaces by dynamically modifying mark morphology. This deep-learning approach has relevance for the interpretation of the local palaeoecological conditions of both assemblages and their respective depositional loci. The results presented here open a new window to the incremental information gain through the use of artificial intelligence methods in taphonomic and palaeoecological research.