A general approach to modeling and analysis of species abundance data with extra zeros

Abstract

A general method for the analysis of ecological count data with extra zeros is presented using a Markov birth process representation of discrete distributions. The method uses a nonparametric formulation of the birth process to model the residual variation and therefore allows the data to play a greater role in determining an appropriate distribution. This enables a more critical assessment of covariate effects and more accurate predictions to be made. The approach is also presented as a useful diagnostic tool for suggesting appropriate parametric models or verifying standard models. As an illustrative example, data describing abundance of a species of possum from the montane ash forests of the central highlands of Victoria, southeast Australia, is considered.