Source & Sourceability: Towards a probabilistic framework for dendroprovenance based on hypothesis testing and Bayesian inference

Long-distance procurement of timber was necessary for the construction of Ancestral Pueblo Great Houses in Chaco Canyon, New Mexico. A number of higher-altitude tree sources were available within 30-70 km, though some isolated trees may have been acquired more locally. Highly regional tree ring variations enable matching some construction timbers to their source. Here, a method is developed which 1) develops a rejection criteria for ruling out sources for a tree ring sequence, 2) quantifies the relative spatial representation of a given source sequence, and 3) applies Bayes theorem to calculate posterior probabilities of source attribution. The application of this method in part supports past sourcing work, but indicates that the majority (59-64%) of timbers cannot be ascribed with even low confidence to the most common high-altitude sources. This analysis supports a model of diverse tree acquisition from a number of different sources, though with high uncertainty for a majority of timbers used in the present study.