|Abstract||There are striking geographical and temporal patterns of juvenile chinook migrations that are not well understood. Chinook populations that migrate to the sea during their first year of life-ocean-type populations-predominate in latitudes north of 56 N, while those migrating later-stream-type populations-predominate south of the Columbia River. In rivers where these two life history types are sympatric, ocean-types are typically distributed more coastally than stream-types. Controversy exists over whether these patterns are the result of postglacial dispersal, or a result of geographical gradients in natural selection.
In this dissertation, I explore the selective pressures on migration timing of chinook salmon by using increasingly complex dynamic optimization models. The optimization models predict that migration distance can strongly influence migration timing, but other biological and physical quantities are also important. Models also suggest that geographical patterns of "growth opportunity" per se do not drive age at migration patterns observed.
A dynamic optimization model reveals two types of optimal behavior that, when appropriately pieced together, produce an optimal migration strategy. One behavior is characterized as "feeding and predator avoidance" and the other, "active migration." The optimal behavior is determined by the signs of two model-derived "switching functions," and the value of the maximum current velocity relative to the swimming speed that maximizes growth. Optimal strategies determined numerically show that behavior often switches from initial feeding and predator avoidance, to active downstream migration as the fish develops and/or during changes in environmental conditions. |