Tom Bird – PhD confirmation seminar
Date: Friday 27 January, 4pm
Venue:  IMAS Seminar Room, Sandy Bay

Abstract:
From their origins in estimating the size of closed populations, Capture-Mark-Recapture analyses have evolved to focus on population parameters such as survival and recruitment rates, migration and individual growth. Increasingly, CMR studies are incorporating data from automated tagging technology to allow the near-perfect observation of a subset of the tagged population, allowing inference on processes that are not well estimated with standard passive tag data. In spite of the increasing prevalence of automated tags, general methods for their incorporation into standard CMR studies remain underdeveloped. My thesis aims to create a conceptually simple yet robust statistical framework to allow the incorporation of automated tagging data for use in a wide variety of sampling situations.

Many CMR models focus on estimating rates at which individuals transition between states, some of which are unobservable. In recognition of this, we present CMR models as class of Hidden-Markov (HM) processes. By doing so in a Bayesian framework, we are able to incorporate the automated tagging data as information on the state of individuals. We also make use of a forwards-backwards Gibbs sampler to improve model convergence. We show how the HM modeling framework can be used to address a number of modeling situations, using a large CMR dataset on native fish in the Murray River, Victoria. We present preliminary results indicating that our approach may provide robust results in a wide range of contexts.