ABSTRACT
In the last chapter we formed probability models for ecological data, and obtained estimates of the posterior distribution of the parameters. The different models fitted will typically be based on biological knowledge of the system. However, in many cases there may be several competing biological hypotheses represented by different underlying models. Typically in such circumstances, discriminating between the competing models is of particular scientific interest. For example, consider ring-recovery data, collected from animals that were marked as adults of unknown age, where there is simply a survival probability (common to all animals) and a recovery probability; this is the model C/C of Section 3.3.1. There are potentially four different models, dependent on whether the survival and/or recovery probability is time dependent or constant over time. Each of these may be biologically plausible, a priori, but describe different biological hypotheses, corresponding to constant survival/recovery probabilities versus time dependent probabilities. We have already encountered this situation in Examples 3.4 through 3.6.
