Gene flow and local adaptation given assortative mating: An application to captive breeding management in salmon

Background/Question/Methods

The general paradigm that gene flow between populations experiencing differential selection impedes local adaptation can break down under specific conditions such as small population sizes. This potential underlies the expectation that captive breeding programs can benefit wild populations despite the potential for artificial selection in captivity to reduce fitness. A potentially important but underexplored dynamic related to this "migration load", and to unintended fitness consequences of captive breeding, is assortative mating where the trait(s) under selection also determine parental mating likelihood (e.g., phenological traits). To investigate the influence of assortative mating on the effects of gene flow, we develop a quantitative genetic model that follows evolution in a trait that experiences differential selection across two populations and influences interbreeding likelihood. To explore the implications for captive breeding management, we link the genetic dynamics to population dynamics for a generic salmon life cycle with interacting hatchery and wild populations. In the context of hatcheries, this model allows us to quantify the relative performance of alternative management strategies: 1) minimize selection in captivity, and therefore the adverse fitness effect of interbreeding, versus 2) intentionally select for divergent populations, which leads to larger fitness differentials but might reduce interbreeding.  

Results/Conclusions

Model results indicate that both similar and divergent selection between the hatchery and wild populations lead to lower negative impacts on the wild spawner population size than an intermediate strategy throughout much of the parameter range. Increasing the strength of assortative mating accentuates this bimodality. Increasing the similarity between hatchery and wild fish might be more difficult to achieve given unavoidable selection in captivity. However, similar selection in the hatchery to the wild does more effectively mitigate hatchery-driven reductions in fitness as well as the ocean adult population size and therefore harvest, and it becomes the clear optimal strategy in terms of wild spawner population size under some assumptions and parameters (e.g., weak selection, small hatchery populations, hatcheries that target hatchery-reared fish, weak density-dependence, high ocean survivorship, and natural selection occurring at later life history stages). These results indicate when the approach of selecting for a distinct population presents a viable alternative strategy to reducing unintended fitness consequences of hatcheries.  In addition, they inform a basic understanding of the conditions under which gene flow impedes local adaptation given assortative mating.