Loss of historical immigration and the unsuccessful rehabilitation of extirpated salmon populations

Comprehensive evaluations of multiple genetic factors are rarely undertaken in rehabilitation attempts of extirpated populations, despite a growing need to address why some rehabilitation projects succeed and others fail. Using temporally-spaced samples of microsatellite DNA, we tested several genetic hypotheses that might explain an unsuccessful attempt to re-establish Atlantic salmon populations (Salmo salar) in two rivers of the inner Bay of Fundy, Canada. Census sizes (N) in both populations plummeted to near zero from initial increases after reintroduction/human-mediated recolonization occurred. Over the same period (1974-1996), both populations were characterized by low or relatively low effective sizes (N _e ) and temporally unstable genetic structuring, whereas neighbouring populations, known historically for their significant salmon production, were not. Despite evidence for genetic bottlenecking and continual linkage disequilibrium over time in both populations, neither exhibited detectable inbreeding or a significant loss of allelic diversity or heterozygosity relative to known donor/source populations. Ratios of N _e to N also increased with decreasing N in both populations, implying a buffering capacity against losses of genetic diversity at depressed abundances. Most significantly, multiple lines of evidence were consistent with the hypothesis that there has been substantial and recurrent asymmetric migration (migration rate, m) from neighbouring areas into both populations even after initial rehabilitation. This included migration from a historically productive population that became extirpated during the course of rehabilitation efforts, indicating that both populations might have naturally depended on immigration from neighbouring areas for persistence. Our results highlight the value of incorporating temporal genetic data beyond commonly used metrics of neutral genetic diversity (F _ST, allelic richness, heterozygosity) to evaluate rehabilitation successes or failures. They also illustrate how the joint evaluation of multiple genetic concerns in rehabilitation attempts, at spatial scales beyond donor and rehabilitated populations, is useful for focusing future rehabilitation efforts.