The estimation of survival parameters is of particular interest within ecological systems for obtaining underlying biological information and animal mark-recapture (capture-recapture) data are often collected for survival parameter estimation studies (King, 2012). However, survival parameters and some other parameters that researchers might be interested (for example, recapture probabilities) are often subject to individual heterogeneity and affected by environmental effects and observation errors. Besides that, recent interest has included additional complexities such as individual covariance and random effects within the statistical framework. To meet the requirements of summarizing unbiased biological information from complex mark-recapture or mark-recapture-recovery data, novel Bayesian fitting state-space models provide a practical tool by coupling a model of mechanistic movement properties (known as process model) with a model of the observation methods (known as observation model).

In this thesis, the Cormack–Jolly–Seber (CJS) model and the multi-state models are developed within Bayesian state-space framework were applied to analyze mark-recapture Dolly Varden data collected from five river systems in Canadian Arctic area. In model fitting process, the Markov Chain Monte Carlo (MCMC) method is being applied to in the estimation process to draw the samples to mimic the joint posteriror distribution since the necessary integration for posterior distribution is intractable. The Bayesian latent variable approach is also used to improve the performance of MCMC algorithm. Meaningful biological information is summarized from the Dooly Varden data which can be helpful to build a environment friendly and sustainable fishing community.