Résumé
More efficient methods are required to breed oil palm (Elaeis guineensis Jacq.) for yield maximization, in order to meet the increased demand for palm oil while limiting environmental impacts. Today, genomic selection (GS) appears to be a disruptive improvement that can speed up breeding schemes by avoiding field trials in some cycles and increase selection intensity by the application of selection to a larger number of candidates than with the current methods. Oil palm is becoming a model species for GS, as it is one of the perennial crops with the largest number of published articles. GS was evaluated in oil palm for the prediction of parental general combining abilities and performances of hybrid crosses and clones. In all cases, GS accuracies high enough to allow selection were obtained for some traits. Best accuracies were obtained when training and validation populations were highly related, such as full-sibs or progenies. Array-based SNPs and GBS-derived SNPs allowed cost effective GS predictions, with densities of a few thousand markers being sufficient. Widely used statistical methods of GS predictions GBLUP and rrBLUP appeared efficient, and could be optimized by SNP filtering methods. Approaches to limit the increase in the rate of inbreeding associated with GS were identified. Evaluations of the annual genetic progress showed that GS should bring it to an unprecedented level. Further studies remain required for the optimal application of GS in oil palm. They should focus in particular on the optimization of training populations, the improvement of prediction models, the variation of GS accuracy between families, the use of multi-omics data (transcriptomics, proteomics, etc.), the modeling of G × E interactions and inter-specific selection.