On the evolution of recombination in haploids and diploids: II. Stochastic models

In this article, we continue an investigation into the evolution of modifiers of recombination, comparing haploid and diploid models begun in Vol. 1, Issue 1, of Complexity. Here, we examine selection schemes that have been used recently in numerical studies of finite diploid populations and ask how recombination evolves in haploid versions of these models. Although the analysis keeps track of the recombination controlling locus rather than the time until a desired bit‐string appears, our result may be of use to the practitioners of genetic algorithms (GA's). We find that as a rule high recombination evolves more easily when selection is on haploids than it does in the diploid case. This is especially true of Gaussian selection schemes with high recombination recessive to low recombination. When the fitness regime is more jagged, however, the results depend on the level of jaggedness, with high recombination favored under smoother regimes. We also find that the direction of mutation and dominance relationships among the modifying alleles affect the results. Although there remains much to be done in reconciling population genetic theory with the properties of genetic algorithms, many new and interesting questions have emerged from, and will continue to be stimulated by, interactions between practitioners of each approach.