Identifying the bone-breaker at the Navalmaillo Rock Shelter (Pinilla del Valle, Madrid) using machine learning algorithms

In recent years, reports on bone breakage at archaeological sites have become more common in the taphonomic literature. The present work tests a recently published method, based on the use of machine learning algorithms for analysing the processes involved in bone breakage, to identify the agent that broke the bones of medium-sized animals at the Mousterian Navalmaillo Rock Shelter (Pinilla del Valle, Madrid). This is the first time this method has been used in an archaeological setting. The results show that these bones were mostly broken by anthropic action, while some were slightly ravaged by carnivores, probably hyaenas. These findings agree very well with published interpretations of the site, and show the method used to be useful in taphonomic studies of archaeological materials with poorly preserved cortical surfaces.