Cranial variability in North American domestic and wild canids

The identification of domestic dogs from North American archaeological assemblages is frequently accomplished using methods developed in Europe for differentiating dogs from their wolf ancestor. However, these methods, many of which are focused on the crania, were never designed to separate dogs from North American canid taxa like coyotes. We test several of these cranial osteometric methods and demonstrate that an index reflecting snout proportions, specifically palate width to length, is among the most reliable univariate methods available. We also find that a series of bilateral comparisons based on the methods we test effectively identify wild canids with over 95% accuracy, but do not necessarily identify dogs with the same efficacy.