Here we will go through a simple hierarchical model of species occupancy that accounts for imperfect detectability of a species. The main aims are (1) to show a variant of binomial-family regression, (2) to demonstrate hierarchical models with observation and process components, and (3) to introduce the concept of latent variables.
We will work with Marc Kery’s data from chapter 20 of his Introduction to WinBUGS for ecologists. The data describe detections of Gentianella germanica at 150 sites, each visited 3 times. We wish to estimate the relationship between \(P\) of occurence, and a humidity index. However, we also know that the detectability of the species depends on humidity itself.