Optimum recombination rates for genetic gains in simulated recurrent selection in empirical maize populations

Meiotic recombination creates new allelic combinations, but it also disrupts favorable parental haplotypes. Our objective was to assess if optimum recombination rates exist in elite maize (Zea mays L.) populations undergoing simulated short-term and long-term recurrent selection. Genomewide marker effects were calculated for eight populations genotyped with 3072 single nucleotide polymorphism markers and phenotyped in up to 18 environments for yield, moisture, test weight, and plant and ear height. The recombination rate was altered by multiplying the sizes of linkage maps by 0.25-10. The regression of genetic gain on the number of crossovers was often curvilinear, and an optimum number of crossovers (CO_Opt) was found in nearly 90% of the trait-population combinations. The CO_Opt was higher in long-term than in short-term selection, and all linear regression coefficients that were significant (p ≤ 0.05) and negative in short-term selection became positive or nonsignificant in long-term selection. This result indicated that increased recombination initially disrupted favorable parental haplotypes but later facilitated the accumulation of favorable haplotypes. The variation in CO_Opt was greater among populations than among traits within the same population, and CO_Opt also increased as the size of the linkage map increased. Overall, our results showed that a higher recombination rate is not always beneficial and that each trait-population combination has an optimum recombination rate that maximizes genetic gain. The practical application of these results awaits the development of methods to both suppress and boost genomewide recombination rates.