Mate selection: A useful approach to maximize genetic gain and control inbreeding in genomic and conventional oil palm (Elaeis guineensis Jacq.) hybrid breeding

Genomic selection (GS) is an effective method for the genetic improvement of complex traits in plants and animals. Optimization approaches could be used in conjunction with GS to further increase its efficiency and to limit inbreeding, which can increase faster with GS. Mate selection (MS) typically uses a metaheuristic optimization algorithm, simulated annealing, to optimize the selection of individuals and their matings. However, in species with long breeding cycles, this cannot be studied empirically. Here, we investigated this aspect with forward genetic simulations on a high-performance computing cluster and massively parallel computing, considering the oil palm hybrid breeding example. We compared MS and simple methods of inbreeding management (limitation of the number of individuals selected per family, prohibition of self-fertilization and combination of these two methods), in terms of parental inbreeding and genetic progress over four generations of genomic selection and phenotypic selection. The results showed that, compared to the conventional method without optimization, MS could lead to significant decreases in inbreeding and increases in annual genetic progress, with the magnitude of the effect depending on MS parameters and breeding scenarios. The optimal solution retained by MS differed by five breeding characteristics from the conventional solution: selected individuals covering a broader range of genetic values, fewer individuals selected per full-sib family, decreased percentage of selfings, selfings preferentially made on the best individuals and unbalanced number of crosses among selected individuals, with the better an individual, the higher the number of times he is mated. Stronger slowing-down in inbreeding could be achieved with other methods but they were associated with a decreased genetic progress. We recommend that breeders use MS, with preliminary analyses to identify the proper parameters to reach the goals of the breeding program in terms of inbreeding and genetic gain.Author summary: Oil palm is a tropical perennial plant and the world main source of vegetable fats. Oil palm breeding requires the estimation of the additive genetic value of the selection candidates to identify the best hybrid crosses, that will be released as commercial varieties. Genomic selection (GS) has the potential of boosting the rate of genetic progress for quantitative traits. However, GS could also be used in conjunction with approaches to optimize selection and mating, in order to maximize genetic progress while limiting inbreeding, which can increase faster with GS and can have negative consequences (inbreeding depression detrimental for seed production, loss of favorable alleles). This study compared several strategies to reach this goal. In particular, we considered mate selection (MS), a method that uses an algorithm optimizing the selection of individuals and their crosses. Our results showed that MS reduced inbreeding in parental populations and increased annual genetic progress. The originality of this work resides in the fact that, despite the high computational burden of MS, we applied it in a large simulated dataset and showed its efficiency in the context of genomic and conventional selection. Also, this is the first study to characterize, in terms of breeding rules, the optimized solution (i.e. the set of selected individuals and their mating design) retained by MS.