QUANTITATIVE APPROACHES FOR THE ANALYSIS OF MORTALITY IN MARINE FISH POPULATIONS

Abstract

Fisheries science has historically focused on recruitment variability and fishery landings as key drivers of dynamics in populations exposed to fishing. However, comprehending the persistence of these populations and successfully managing fishing impacts on them depends on understanding the magnitude, dynamics and causes of mortality from all sources. In this thesis, I develop a series of approaches for the analysis or estimation of discard and natural mortality, two key sources of mortality that have often previously been overlooked. In chapters two and three I present methods for estimating indicators of discard mortality (DM). These approaches provide simple and cost-effective means to evaluate the influence of factors affecting DM and case-specific risks posed by DM, and to improve formal DM estimates in directed studies. In chapters four and five, I present a case study for the estimation of fishing mortality for principally discarded species, skates in the southern Gulf of St. Lawrence (sGSL; Atlantic Canada). Estimating fishing mortality for these species of conservation concern poses several challenges, including catch reporting that is not disaggregated to species and that is somewhat sparse. I derive and apply a method to disaggregate the catches and an integrated approach, based in part on the methods from chapter two, to estimating landed and discarded catch amounts and DM rates. In chapter six, I develop and validate a model for data from DM experiments that is generalizable to estimate different components of mortality, including jointly estimating discard and natural mortality. Finally in chapter 7, I develop a simulation model to evaluate whether predation by a top predator is a scientifically defensible explanation for elevated natural mortality in three fish species of conservation concern in the sGSL. This approach is presented as a means of evaluating the possible consequences of predation in the absence of reliable estimates of predator diets. Together, the methods developed in this thesis represent new tools for the estimation of mortality components and for the attribution of their likely drivers. In turn, these tools should help inform management actions aimed at mitigating mortality or sustainably managing populations and communities in light of existing mortality.