SNIFFING OUT RED HERRINGS: DOG SKULL SHAPE IS A POOR PROXY FOR SCENT-TRACKING ADVANTAGE

The many breeds of domestic dog we know and love today are tied to a substantial history regarding the tasks they were bred to accomplish and the observable characteristics that belong to a breed's exemplars. These standards are tracked and managed by large organizations such as the American Kennel Club (AKC) and the United Kennel Club (UKC) and are used to group and categorize dog breeds, largely based on the tasks for which the breed was historically tailored to perform. We explore the cranial shape alongside wild canids and foxes to test how well these associations map to morphological clustering.

We present a principal components analysis of forty cranial landmarks representing over 100 different specimens. We find the most distinguishing characters for these groups are the angle of repose between the snout and cranium and the ratio between the snout length and cranial length (aka craniofacial ratio). We find three main clusters of points when viewing these two characters which contrasts the task based groupings of both kennel clubs. We find only companion and guardian group dogs from the UKC classification map well onto one of these clusters, as does the toy group of the AKC which is analogous to the aforementioned companion group.

Finally, we discuss work on building computational systems around these skeletal elements, focusing on analyzing olfactory performance. Given the complexity of the internal morphology we discuss our methods for reducing the complexity of the system to make a large computational analysis tractable without sacrificing the explanatory potential of the work. Similar applications are increasingly possible, common, and relevant to biomechanics studies in the fossil record. In a time of increasing demand on computer resources we discuss this process as a template for performing low-complexity exploratory analyses prior to engaging in higher complexity, resource intensive, simulation experiments.