The study of dental microwear allows the analysis of the microscopic marks that foods leave on the surface of tooth enamel during mastication. In palaeoanthropology, this methodology helps reconstruct the diet of fossil primates and hominins throughout human evolution. The microscopic striations in dental enamel are like microscopic archives that reveal whether the diet was rich in foods with softer or more abrasive components. Now, a study published in the journal Scientific Reports presents an innovative artificial intelligence (AI)-based methodology for identifying 3D wear patterns consistently and independently of the analyst.
These 3D wear patterns differ among primates inhabiting diverse ecosystems and following different diets. The study also identifies which variables are most informative for the classification of dental microwear and proposes an analytical framework open to the entire scientific community for studying this type of surface.
The study is led by Professor Laura M. Martínez, of the Faculty of Biology and the Institute of Archaeology (IAUB) at the University of Barcelona, a pioneering expert in the application of machine learning techniques in palaeoanthropology. The research also involves Ferran Estebaranz, member of the UB and of the Milà i Fontanals Institute for Research in Humanities (IMF-CSIC); Juan José Ibáñez (IMF-CSIC); Simón Rodríguez (Comillas Pontifical University), and Kristina Kit and David R. Insua, from the Institute of Mathematical Sciences (ICMAT-CSIC), who are leading researchers in the application of machine learning techniques to archaeological research.
Studying environmental changes through dental microwear
The study of dental microwear has a long history in research on the origin and evolution of the human lineage. “Until now, simpler wear measures, usually in 2D, had often been used, relying on conventional statistical techniques that established relatively direct relationships between these parameters and diet,” explains Laura M. Martínez of the Department of Evolutionary Biology, Ecology and Environmental Sciences at the UB.
In the current context, there is little precedent for applying AI technologies to study dental wear and the palaeoecological adaptations of hominins. Now, Laura M. Martínez is leading a project that uses AI models trained on 3D dental wear surfaces of primates with known diets, with the aim of applying them to the study of fossil primates from the African and Iberian Plio-Pleistocene.
“To reconstruct the diet of fossil primates and hominins, it is essential to have good comparative models based on living primates and hunter-gatherer populations with known diets,” adds Martínez. “With the incorporation of 3D techniques, it has been possible to generate a very large number of variables, which makes interpretation with conventional statistics difficult. In this context, AI facilitates the integration and compression of this complex information, thereby allowing the identification of patterns in 3D surfaces that are not easily interpretable directly,” the researcher explains.
The project focuses particularly on the cercopitheciids — an extensive family of primates present in various habitats — from northern, eastern and southern Africa, based on sites dated between 4 and 1 million years ago. During this period, significant climatic changes occurred that profoundly affected African ecosystems. The aim is to analyse, in relation to climate change, the evolution of diet in these primates, which also coexisted in time and space with the earliest hominins.
“Cercopithecids lived in the same ecosystems and during the same time as hominins, making them an excellent model for understanding how Plio-Pleistocene climate changes affected the diet and adaptation of these primates,” the researcher explains.
From this perspective, the study opens up a new scenario with models capable of distinguishing extant primates with diverse diets, thereby providing a reference framework to which fossil primates can be incorporated.
In the future, the team aims to significantly increase the sample size to improve the model’s accuracy and robustness. To this end, more samples from different species and diverse ecosystems, with well-characterized diets, as well as other ecological factors, are being incorporated to make the analysis more consistent.
“With a good primate reference model, we will be able to develop robust references for interpreting the diet of our ancestors in 3D, in an integrated way with other palaeoecological and climatic indicators,” concludes Laura M. Martínez.