My talk at ISEC 2014 (http://isec2014.sciencesconf.org/) on how to model occupancy data using hidden Markov models

This document discusses using hidden Markov models to analyze occupancy data. It describes how occupancy models can be formulated as hidden Markov models, with sites as hidden states that can be occupied or unoccupied over time. Both single-season and dynamic occupancy models are discussed. Modeling occupancy data as hidden Markov models provides a unified framework and links occupancy modeling to capture-recapture methods. Software called E-SURGE that was originally developed for capture-recapture analysis can also be used to fit occupancy models. An example case study uses E-SURGE to model Eurasian lynx occupancy data from France allowing for detection heterogeneity. Extensions to occupancy hidden Markov models including distribution mapping, accounting for lack of independence, and multistate models