Archaeological sites often contain accumulations of remains derived from different independent events produced by different agents. Thus, in Palaeolithic sites, it is normal to find alternating occupations between humans and carnivores. The faunal assemblages at these sites usually include hundreds or thousands of bone fragments, which are very difficult to associate them to specific individuals since there are no currently available techniques able to do it in a straightforward and cost-effective way. In this work we present a methodology that allows us to characterise the anatomical remains of a bone accumulation and relate them all back to the specific individuals to which they belong. In order to provide a real world application, we have used a selection of animal bones from different individuals belonging to deer and sheep (fed in a controlled way using the same diet). On the other hand, fossilized faunal remains have also been analysed to verify if these fossilized bones keep some of the fingerprinting of the animal from which they come from. For this purpose, we have developed a protocol using Laser Induced Breakdown Spectroscopy (LIBS) together with Neural Networks (NN) implemented here to discriminate and reassemble deer and sheep bones from different individuals, which we subsequently applied for these proposes to fossilized material. To the best of our knowledge, this is the first time that this technique has been applied for individual fingerprinting to actuality and fossil samples. The elemental composition of bones provides enough information to get a correct discrimination of different individuals. The spectral correlation has exceeded 95 %. and all individuals were correctly classified to the individual from which they come from. There have been no instances of false positives or false negatives in our tests or applications.
Keywords: Archaeology; Bone discrimination; Laser induced breakdown spectroscopy; Neural networks.
Copyright © 2021 Elsevier B.V. All rights reserved.