Understanding the effects of growth and size-at-age variation on the dynamics of fish populations

Abstract

Understanding drivers of populations is of tantamount importance across a broad scale of researchers, from theoretical ecologists to tactical resource managers. Drivers may be internal feedbacks (density-dependent) or external (density-independent) processes, such as changes in prey, predator, or competitor populations, or environmental stochasticity. In a closed population, these drivers affect populations by altering demographic rates (i.e. mortality, reproduction, somatic growth). Although there is increasing evidence that no demographic rates are static, at least in patchy and stochastic aquatic environments, it is an ongoing question to identify the most important types and scale of variation for population dynamics models. In this dissertation, I seek to quantify the magnitude and effect of growth and size-at-age variation on fish population dynamics using a variety of different modeling techniques. In the first chapter, I use a state-space statistical model to quantify the magnitude and type of temporal size-at-age variation experienced by a number of Pacific groundfish populations. In the second chapter, I use these estimates of growth variation, along with parameters taken from fisheries stock assessment models, to illustrate how both growth and recruitment variation may introduce fluctuations into simulated populations with otherwise static demographic rates. In the third chapter, I use an integrated analysis model to simulate and estimate patterns of growth variation in Petrale sole (Eopsetta jordani) to examine the effect of growth misspecification on estimates of population status. In the final chapter, I adapt a size-structured ecosystem model to Tonlé Sap Lake, Cambodia, and explore ways to validate model accuracy in a species-rich, data-poor ecosystem. This work highlights the importance of accounting for multiple types of demographic stochasticity across life history type and how appropriate model complexity scales with data quality and quantity.