ABSTRACT
This chapter presents the most common types of statistical models formulated as generalized linear models (GLM), many designed specifically for ecological and environmental data. The development of GLMs more explicitly deals with non-Gaussian data and has been tremendously useful in many areas of science. The chapter describes the GLM specifications for several different types of common ecological questions and data, beginning with binary data. Ecological and environmental data are often binary or integer-valued. The binary data example involving observed presence or absence of the forest understory plant Desmodium glutinosum to demonstrate the Markov chain Monte Carlo algorithm. The auxiliary variables actually lead to a hierarchical binary regression model specification that yields the same inference as the non-hierarchical model. However, a more typical approach is to specify hierarchical binary data models that have a latent Gaussian process.
