Whether a species is present or absent at a site is often a question we are interested in. Inference is based on observations of sites, where detection or non-detection of the species is recorded. Since probability of detection when present is hardly ever 1, non-detection does not mean absence. Occupancy modelling enables us to estimate both probability of detection when present and the probability that a site with no detections is occupied.

Even the simplest occupancy model is in fact a hierarchical model. At the lowest level we have an **observation model**, which relates the detection/nondetection data to probability of detection when present and a latent presence-absence variable. On top of this sits a **biological model** or process model which describes the presence-absence variable.

We’ll see how this works for a basic occupancy model with no covariates, and then move on to more sophisticated models with covariates.