California Sea Grant College Program

Frequently wildlife managers must decide how to allocate limited resources amongst a plurality of threatened salmon stocks. In the absence of adequate abundance data, knowledge of stocks life histories might be used to rank risk of extinction thereby allowing more efficient allocation of resources. In Chapter one I assess how differences in life histories contribute to relative risk of extinction using Pacific salmon as an example. Using simulations of coho and chinook salmon. I find increased spawning at multiple ages causes the fate of each cohort to be linked to the success of other cohorts. This novel effect I refer to as cohort coupling.

Since many assume that density dependence is a stabilizing force. Chapter two examines this assumption in terms of extinction risk for coho salmon. Here again I find that life history plays a role in determining risk. along with environmental variation and quasi-extinction threshold. Key also is how density dependence is introduced into the model. I find the widely held view that density dependence is stabilizing is of limited value since density dependence can either increase or decrease probability of extinction.

Chapter three explores the use of a diffusion approximation as a tool for understanding population dynamics of Pacific salmon. Diffusion approximations have a long history in ecology as models of population genetics and population growth. Modeling salmon using a diffusion approximation requires special consideration because salmon are both semelparous and anadromous. Diffusion estimates of probability of extinction are accurate only for a limited range of life histories.

Since salmon exhibit environmentally influenced variability in spawning age. Chapter four explores the effect variability in spawning age has on probability of extinction. I find that when variability is confined to fraction spawning at a single age. the probability of extinction is proportional to the integrated elasticity of A. for realized fecundity at that age. Additionally simulations indicated that variability in fraction spawning is most important when it affects the dominant age class of a cohort.