Elicitation of expert knowledge of phylogenies in the form of rooted trees

This thesis develops a methodology for eliciting expert knowledge on phylogenies focusing on a rooted tree. It is motivated by the necessity of incorporating the expert knowledge such as that obtained from morphology, fossil studies, etc. with molecular data in Bayesian inference of phylogenies. Based on the three point condition, expert opinion on phylogenies is elicited by asking an expert to introduce species one by one. Probability distributions are obtained on both tree topologies and branch lengths. Evaluation work is done by separately evaluating elicitation on branch length and on topology. The elicitation methodology is implemented in R to create an user-interface program to provide instant graphical feedback as well as to guide the elicitation. The use of the methodology, technique and software is demonstrated through six case studies on various species. The results of one case study are applied to Bayesian analysis by serving as the informative prior distribution. Effect of introduction order is first studied by a case study with three species, then explained by theoretical analysis.