“I wanna be your dog”: Evaluating the efficacy of univariate and multivariate methods for differentiating domestic and wild canids in North America

The domestic dog (Canis familiaris) holds a unique place in human cultures as the first species to be domesticated and has been adapted to a greater diversity of functions requiring far greater morphological variation than any other domesticate. Because of this variability in morphology and pronounced skeletal similarities with other canid species, dog remains are frequently challenging to identify in the archaeological record. Analysts have attempted to overcome these challenges by proposing a diverse array of methods for identifying dog remains. Unfortunately, recent analyses have quantitatively tested and critiqued the effectiveness of several methods widely used for identifying dog remains since the mid‐1900s. In addition, many methods were developed specifically for differentiating dogs from their progenitor species, the grey wolf (Canis lupus) and analysts working in many regions of the world have frequently assumed, without testing; they will remain effective in differentiating dog remains from other canid species. Using data collected on 538 North American dog, wolf, coyote and fox mandibles, we test the effectiveness of several methods for differentiating dogs from an array of North American canids. Our results reveal that no single method is universally effective and that differentiating dogs from mid‐sized canids, like the North American coyote, remains a significant challenge for archaeologists working in North America and likely other parts of the world.