A method for validating stochastic models of population viability: a case study of the mountain pygmy‐possum (Burramys parvus)

1. A method of validating stochastic models of population viability is proposed, based on assessing the mean and variance of the predicted population size. 2. The method is illustrated with a model of the population dynamics of the mountain pygmy‐possum (Burramys parvus Broom 1895), based on annual census data collected from a single population in the Snowy Mountains of New South Wales, Australia between 1986 and 1997. The model incorporates density‐dependence in survivorship and recruitment, and demographic and environmental stochasticity. 3. The model appeared to make reasonable predictions for the three populations that were used for validation, provided the equilibrium population size was estimated accurately. This may require that differences in habitat quality between populations be taken into account. 4. Following validation, the model was given new parameters using the additional data from the three populations, and the risk of population decline within the next 100 years was assessed. Although populations as small as 15 females are predicted to be relatively safe from extinction caused by stochastic processes, B. parvus appears vulnerable to loss of habitat and reductions in the population growth rate. 5. The approach used in this paper is one of few attempts to validate a model of population viability using field data, and demonstrates that some aspects of stochastic population models can be tested.