Introduction to Integrated Population Models

Abstract

We introduce integrated population models (IPMs) in two different ways. First, we start from a matrix population model and progressively add more data sets along with likelihoods for estimating certain parameters from them as part of a single model fit with the JAGS program. We emphasize that the inclusion of population count (or similarly, population index) data is a crucial part of an IPM because the process model for these data allows integration with the other data sets. The second way to develop an IPM starts with the definition of a population model that links demographic rates to a vector of age- or stage-specific population sizes. It proceeds by defining the likelihoods for each of the available data sets and finally by developing the joint likelihood from which inference is obtained under the IPM. In the BUGS language, there is no need to explicitly write this joint likelihood in a single expression—integration of the different data sets is ensured by giving the same name to parameters that appear in different component data likelihoods. Finally, we introduce data simulation under an IPM and demonstrate that IPMs provide unbiased parameter estimates when the assumptions are met.