Résumé
The ability to document the effects of domestication from archaeological remains of animals and plants is essential for reconstructing the history of one of the most important transitions in human prehistory – the shift from hunting and gathering to farming. In mammals, teeth are well preserved in archaeological remains and are known to be taxonomically informative. In this study, we compare three sets of dental morphometric descriptors in wild and domestic pigs (Sus scrofa) – maximum length, size and shape variables from 2D geometric morphometrics – in order to assess which of the three provides the best ability to correctly distinguish current wild and domestic West Palaearctic pigs. For this purpose, we used predictive linear discriminant analysis with cross-validation taking into account potential bias due to heterogeneous sample sizes and important number of predictors. Classification accuracy of wild and domestic status ranged between 77.3 and 93% depending of the tooth and the descriptor analyzed. However, individual posterior probabilities of correct classification were appreciably smaller when using tooth length and centroid size compared to shape variables. Size appeared to be a poor indicator of wild and domestic status, contrary to shape which in addition provides a high degree of confidence in the wild versus domestic predictions. Our results indicate that geometric morphometrics offers an extremely powerful alternative to more traditional biometric approaches of length and width measurements to capture the elusive morphological changes induced by the domestication process in archaeological remains.
► Geometric morphometrics was used to accurately identify wild or domestic Suc scrofa. ► We compare and contrast the accuracy of 3 descriptors to recognize wild/domestic pigs. ► Molar size is an extremely poor indicator of the domestication status in pigs. ► Molar shape provides the best available technique to identify domestic pigs. ► Geometric morphometrics is a powerful alternative to traditional biometric techniques.