ABSTRACT
The early papers by Jolly (1965) and Seber (1965) each distinguish between the separate activities of the survival of animals and the reencounters of them. The same is true, for instance, of occupancy data, with an observation process superimposed upon a process which describes whether or not sites are occupied, as seen in Section 6.1. It is convenient to have this common perspective for several of the models that we have considered, including recovery, recapture, recovery and recapture, and occupancy. The development of two linked stochastic processes, separately modelling how a system changes over time, and the observations on that system, can be described by state-space models, which also provide a natural way to model time-series of counts in ecology. Annual time-series in ecology are typically far shorter than time-series that arise in other areas, such as economics for example. As a consequence, although it is possible to use state-space models to describe relatively short ecological time series, resulting parameter estimators may lack precision. However, statespace models for ecological time series provide key likelihoods for integrated population modelling, the subject of the next chapter. For human populations we associate a census with a complete enumeration of the members of a population under consideration. However in statistical ecology, following Darroch (1958), the word census is used more generally, to indicate a count. Time series of count data may arise from national censuses, for example, as well as from smaller-scale ecological studies, and in some cases censuses of entire populations are in fact taken, as we have already seen in Chapter 7.
