ABSTRACT
As we have seen in Chapters 2 and 3, demographic parameters of interest often exhibit temporal variability. For instance, for the analysis of the dippers data in Example 6.1 there was evidence that the survival probability depended on the year of the study (flood versus non-flood years). Modelling the survival probabilities with an arbitrary time dependence, however, does not provide any information as to an underlying cause for the temporal heterogeneity. In addition, for long-term studies, over a period of 20 or 30 years, say, this creates a potentially very large number of parameters within the model, possibly creating over-fitting. An alternative approach that addresses these two issues is to consider the use of covariates to explain the temporal variation in the demographic parameters. For example, for the dippers example, we considered two different survival probabilities: one for the years that surrounded a flood (φf ), and a different survival probability for the other years (φn). Here, the covariate that explains the temporal variability is simply an environmental factor indicating whether there was a flood or not.
