Search

Horses and donkeys have had a far-reaching impact on human history, providing mechanical power for agriculture and transportation. Their F1-hybrids, especially mules, have also been of considerable importance due to their exceptional strength, endurance and resistance. The reconstruction of the respective role that horses, donkeys and mules played in past societies requires prior identification of their osseous elements in archaeological assemblages. This, however, remains difficult on the basis of morphological data alone and in the absence of complete skeletal elements. While DNA sequencing provides almost certain identification success, this approach requires dedicated infrastructure and sufficient ancient DNA (aDNA) preservation. Here, we assessed the performance of a cost-effective alternative approach based on geometric morphometric (GMM) analysis of the bony labyrinth, a structure carried within the petrosal bone. This extremely compact osseous structure provides good aDNA preservation and is frequently found in archaeological assemblages. To assess the GMM performance, we first used High-throughput DNA sequencing to identify 41 horses, 24 donkeys, 36 mules and one hinny from 11 archaeological sites from France and Turkey spanning different time periods. This provided a panel of 102 ancient equine remains for micro-computed tomography (microCT) and GMM assessment of the variation of the bony labyrinth shape, including the cochlea and the semicircular canals. Our new method shows good-to-excellent prediction rates (85.7%–95.2%) for the identification of species and hybrids when considering the cochlea and semicircular canals together. It provides a cheap, non-destructive alternative to aDNA for the taxonomic identification of past equine assemblages. • Cheap, non-destructive alternative approach to aDNA based on geometric morphometric. • 102 ancient equine remains genetically identify used for micro-computed tomography. • Analysis of the shape variation of the structures carried by the bony labyrinth. • Good-to-excellent prediction rates for the identification of species and hybrids. [ABSTRACT FROM AUTHOR]