Analysis of Fraser River sockeye salmon coastal migration route variation using bayesian estimation methods and individual-based modelling

• Main Author: Kolody, Dale Shawn
• Language: English
• Published: 2009
• Online Access: http://hdl.handle.net/2429/10117

Sockeye salmon (Oncorhynchus nerka) can return from the Gulf of Alaska to the Fraser River by migrating around the north or south end of Vancouver Island and the proportion of fish using each route varies considerably among years. This thesis consists of three separate studies that contribute to the overall objective of understanding the migration route variation. The first two components are concerned with the estimation of migration routes from fisheries data, and potentially have additional practical implications for fisheries management. The third component examines some explicit interactions between individual behaviour and oceanographic variability that could potentially affect migration route selection. The first study investigated how standard methods of estimating salmon fishery harvest rates introduce substantial errors because of violations to migration dynamics assumptions. This involved: 1) defining plausible stochastic salmon migration rate variability scenarios from observations of salmon migratory timing distributions and tagging studies, and 2) using Monte Carlo simulations of fisheries to examine how this variability affects harvest rate estimation when uniform migration rates are assumed during the process of run-reconstruction. A unique migration dynamics scenario could not be defined. Within the migration constraints that were identified, the simulations suggested that, in general, harvest rate estimates can be expected to be beta-distributed, and there is a potential underestimation bias for high harvest rates. The magnitude of the error variance and bias are dependent upon the migration dynamics, fishery temporal and spatial structure, and run-reconstruction method. The second study presents methods for estimating harvest rates and migration routes in a multiple approach salmon fishery. The Bayesian approach explicitly admits uncertainty from 1) the confounded relationship between harvest rates and migration routes, 2) escapement estimation error that arises from run-reconstruction, and 3) the unknown relationship between harvest rates and effort. The resulting parameter uncertainty is reflected in posterior probability distributions. Potential advantages and disadvantages of adopting this method for the Fraser River sockeye fishery are examined. The third study involved simulating adult sockeye salmon migration routes from the Gulf of Alaska to the coastal approaches of the Fraser River. A spatially-explicit individual-based model was used to explore potential mechanisms that could explain the observed interannual variation in migration routes. Assuming that sockeye are initially distributed throughout the central Gulf of Alaska and orient on a compass bearing, the following mechanisms were simulated to produce migration route variations among years in a physical environment described by dynamic surface temperatures and currents: 1) the distribution prior to homing was constrained by southern thermal limits, 2) sockeye were advected by currents during open ocean migration, and 3) sockeye tended to avoid high water temperatures. Optimization of the behavioral component of the model for a least squares fit between hindcast and observed coastal migration routes suggested that thermal limits and offshore-currents could not explain very much coastal migration route variability. Avoidance of high water temperature explained about 33% of the interannual variability and suggested that coastal processes (although poorly resolved in the model) could be more important than the offshore processes examined.