Short-term selection for recombination among mutually antagonistic species

Abstract

The recombination of genes is clearly disadvantageous in a uniform and predictable environment where a single multilocus genotype is optimal. But heterogeneity or unpredictability do not necessarily imply a short-term advantage for recombination, which will be created only if the correlation between adaptively important features of the environment frequently changes sign¹. Another way of expressing this conclusion is to say that genes causing non-zero rates of recombination will tend to spread only if selection often favours a change in the sign of linkage (gametic phase) disequilibrium². Because it is difficult to imagine that the physical environ-ment often behaves in this fashion, recent theoretical work on the maintenance of sex, recombination and outcrossing has speculated that biotic interactions may be important^3–5. May and Anderson⁶ analysed a two-locus haploid model of the coevolution of host and parasite, similar to that considered here, and showed that the mean fitness of a population with recombination is typically higher than that of a population without, although they did not explicitly consider changes in the frequency of genes affecting recombination. Hutson and Law⁷ and Bell⁸ suggested that the crucial feature of antagonistic interactions between species is the effect on the current fitness of a genotype of its frequency in previous generations; such time-lagged frequency-dependent selection tends to destabilize both genotype frequencies and linkage disequilibrium, whose oscillation has been shown to fuel the spread of high-recombination alleles in explicit genetic models⁸. In this paper, we shall show how the mutual antagonism of two species can lead to a cyclical game in which high-recombination alleles can have a large short-term selective advantage in a fully defined genetic model.