Sorting the flock: Quantitative identification of sheep and goat from isolated third lower molars and mandibles through geometric morphometrics

International audienceSheep and goat are often herded together and show morphological similarities in their skeleton. Being able to identify archaeological remains of these two taxa to species level is particularly important for understanding and characterising past herding practices. Discrete criteria are now available to identify a large number of their bones and teeth, and quantitative approaches have been developed for post-cranial elements but not for mandible and isolated teeth. In this paper we explore the discriminating potential of geometric morphometrics to identify modern sheep and goat third lower molar and mandible and its application on archaeological specimens. The size and shape of the mandible and the third lower molar of 143 modern specimens (101 sheep and 42 goats) were quantified using 2D-landmark and sliding semi-landmarks geometric morphometric approaches. The results show that sheep and goat differ in terms of the size, shape, and form (i.e. size and shape together) in both studied elements. Classification accuracy of the two species reaches 93.3% (CI: 90.0–95.7%) for third lower molar shape, 62.7% (CI 57.1–68.6%) for third lower molar size, 95.2% (CI: 92.0–97.4%) for mandible shape and 84.0% (CI 81.6–86.8%) for mandible size. Form does not provide better classification than shape alone. Sex and age appear to have little impact on the ability to differentiate between sheep and goat, despite the two species displaying distinct sexual dimorphism and changes through age. The same methodology was then applied on 32 Middle Ages third lower molars from Missignac-Saint Gilles le Vieux, Aimargues, France. The identifications obtained through geometric morphometrics were only partially congruent with the identifications based on visual observations calling for caution in the interpretation and further investigations. Further research should include molecular identification of the archaeological specimens to assess whether the geometric morphometric identification can be made with confidence for all periods and all geographic areas. Nevertheless, the results obtained with the newly developed geometric morphometric protocols represent an important contribution toward a better understanding of past livestock husbandry practice