ABSTRACT
This chapter focuses on the implementation of dynamic autoregressive models and point out a few important aspects of constructing Markov chain Monte Carlo algorithms to fit these models to time series data. There is an important connection between conditionally specified dynamic time series models and jointly specified time series models. Higher order model specifications can result in smoother overall time series. The chapter also focuses on continuous-valued time series because that is what the classical methods are designed for. However, many data sets in wildlife ecology are comprised of discrete-valued observations. In time series analyses, measurement error is often accommodated using hierarchical models, which are sometimes called state-space models in the temporal context. However, the real power of dynamic modeling comes into play with multivariate temporal processes. For example, many spatio-temporal models are specifically designed to account for dynamics of evolving spatial processes over time.
