Addressing non-stationary fishery dynamics and demographic complexity in integrated stock assessment models

Abstract

Integrated stock assessment models are complex non-linear statistical models that fit a variety of fishery-dependent and independent data sources to estimate the status and trends of fish populations, providing recommendations for sustainable harvest levels. Given their complexity, simplifying assumptions are necessary in stock assessment models to balance model parsimony versus complexity, while also considering data limitations. In this dissertation, I investigated considerations for addressing non-stationary fishery dynamics and evaluated the consequences of oversimplified assumptions of fishery dynamics and sex-specific demography in stock assessment models. Using Alaska sablefish (Anoplopoma fimbria) as a case-study, I developed a generalized framework to standardize fishery-dependent abundance indices, while incorporating various data sources and catch records from multiple gear types to address rapid shifts in fishery fleet structure (Chapter 2). Building on this foundation, I then explored the implications of accounting for, or ignoring, complex temporal changes in fishery fleet structure in stock assessment models by comparing multi-fleet and single-fleet models (Chapter 3). Here, I found that the treatment of fleet structure generally had minimal impacts on quantities of management interest, but selectivity assumptions had large impacts on recommended harvest levels. I then employed a generalized simulation-estimation framework to evaluate the performance of different stock assessment approaches for addressing changes in fleet structure (Chapter 4), which suggested that single-fleet models with time-varying selectivity are adequate for accounting for changes in fleet structure. To understand the consequences of oversimplified assumptions of sex-specific demography in stock assessment models, I developed a simulation­ estimation framework (Chapter 5), which revealed that such simplifications led to suboptimal management advice. Collectively, this dissertation underscores the various approaches available for addressing non-stationary fishery dynamics, the importance of biologically motivated models that adequately reflect a population's demographic characteristics, and the necessity of expert judgment in stock assessment models given the constraints of data limitations.