Quantifying expert opinion for modelling fauna habitat distributions (Q880899)

The paper reports results of an experiment conduced with a group of ecologists in Australia who were questioned about relationships between a variety of environmental attributes and the fauna habitant distributions of a number of species. A Bayesian approach is used to reach a consensus between the sample estimates and the experts opinions. The sampling distribution is described by a generalized additive model. The experts' knowledge is incorporated in a multivariate normal prior. The posterior distribution is evaluated via Markov chain Monte Carlo priors.