Exploring methods by which to quantify confidence in canid species determination

Abigail E. Fisher and Ian Jorgeson

Abstract: Due to their ubiquity across North America, dogs have the potential to inform on a variety of human behaviors. Dog skeletal remains, however, are particularly difficult to identify in archaeological assemblages due to their fragmentary nature (a product of life history and taphonomic processes) and similarity in morphology and size to wild canids such as coyotes and wolves. This research uses geometric morphometrics to better classify canid remains. While most methods of species differentiation result in an absolutist identification, we are developing a probabilistic model relating morphological variation to species likelihood. As a probabilistic model, our approach will both identify the most likely species represented by a sample, and quantify the uncertainty of that classification.

Problem: Dogs remains have the potential to provide insight on past people,including human-animal interaction, ritual behavior, and diet. However, they must first be identified in the archaeological record, which is sometimes regionally specific and often subjective.

Objective: Gather data from images of known dog, coyote, and wolf mandibles from different regions in North America to build a probabilistic model capable of assigning species likelihoods to unknown canid mandible specimens.

We need your help!

Please submit photographs of known species using the links below:

View our Poster here

To download instructions, click here

To download more submission forms, click here

To upload photos, click here

Questions? Having Trouble? Contact us at info@caniXproject.org