Sockeye salmon are a socially and economically important keystone species on the Pacific west coast. Understanding the migration patterns of spawning salmon is critical to maintaining a healthy commercial fishery. Representing a complex ecological system, the data collected on salmon migration has an abundance of characteristics that are difficult to address using standard statistical models. Data collection is also constrained due to limited time, financing, and the presence of competing social interests.

Our research journey begins with a discussion of our development of a multivariate conditional autoregressive model to address data challenges. These challenges inlcude sparse observations in data subgroups, restrictive sample sizes at various time points, and time points where sampling is missed altogether. Subsequently, we propose sampling methods that are complementary to our Bayesian estimation while still sensitive to the economic and social constraints that influence organizations responsible for data collection such as the Pacific Salmon Commission.

Keywords: Bayesian; Multivariate Conditional Autoregressive; Sampling

Biography: Carolyn Huston has a BSc. in biology with a minor in philosopy from the University of Alberta. Subsequently she completed a MSc. in biostatistics at the same institution. Currently she is completing a PhD. in statistics at Simon Fraser University.