Improving the accuracy of genomic predictions in an outcrossing species with hybrid cultivars between heterozygote parents: a case study of oil palm (Elaeis guineensis Jacq.).

Genomic selection (GS) is a method of marker-assisted selection revolutionizing crop improvement, but it can still be optimized. For hybrid breeding between heterozygote parents of different populations or species, specific aspects can be considered to increase GS accuracy: (1) training population genotyping, i.e., only genotyping the hybrid parents or also a sample of hybrid individuals, and (2) marker effects modeling, i.e., using population-specific effects of single nucleotide polymorphism alleles model (PSAM) or across-population SNP genotype model (ASGM). Here, this was investigated empirically for the prediction of the performances of oil palm hybrids for yield traits. The GS model was trained on 352 hybrid crosses and validated on 213 independent hybrid crosses. The training and validation hybrid parents and 399 training hybrid individuals were genotyping by sequencing. Despite the small proportion of hybrid individuals genotyped and low parental heterozygosity, GS prediction accuracy increased on average by 5% (range 1.4-31.3%, depending on trait and model) when training was done using genomic data on hybrids and parents compared with only parental genomic data. With ASGM, GS prediction accuracy increased on average by 3% (- 10.2 to 40%, depending on trait and genotyping strategy) compared with PSAM. We conclude that the best GS strategy for oil palm is to aggregate genomic data of parents and hybrid individuals and to ignore the parental origin of marker alleles (ASGM). To gain a better insight into these results, future studies should examine the respective effect of capturing genetic variability within crosses and taking segregation distortion into account when genotyping hybrid individuals, and investigate the factors controlling the relative performances of ASGM and PSAM in hybrid crops.

Three founding ancestral genomes involved in the origin of sugarcane.

BACKGROUND AND AIMS: Modern sugarcane cultivars (Saccharum spp.) are high polyploids, aneuploids (2n = ~12x = ~120) derived from interspecific hybridizations between the domesticated sweet species Saccharum officinarum and the wild species S. spontaneum. METHODS: To analyse the architecture and origin of such a complex genome, we analysed the sequences of all 12 hom(oe)ologous haplotypes (BAC clones) from two distinct genomic regions of a typical modern cultivar, as well as the corresponding sequence in Miscanthus sinense and Sorghum bicolor, and monitored their distribution among representatives of the Saccharum genus. KEY RESULTS: The diversity observed among haplotypes suggested the existence of three founding genomes (A, B, C) in modern cultivars, which diverged between 0.8 and 1.3 Mya. Two genomes (A, B) were contributed by S. officinarum; these were also found in its wild presumed ancestor S. robustum, and one genome (C) was contributed by S. spontaneum. These results suggest that S. officinarum and S. robustum are derived from interspecific hybridization between two unknown ancestors (A and B genomes). The A genome contributed most haplotypes (nine or ten) while the B and C genomes contributed one or two haplotypes in the regions analysed of this typical modern cultivar. Interspecific hybridizations likely involved accessions or gametes with distinct ploidy levels and/or were followed by a series of backcrosses with the A genome. The three founding genomes were found in all S. barberi, S. sinense and modern cultivars analysed. None of the analysed accessions contained only the A genome or the B genome, suggesting that representatives of these founding genomes remain to be discovered. CONCLUSIONS: This evolutionary model, which combines interspecificity and high polyploidy, can explain the variable chromosome pairing affinity observed in Saccharum. It represents a major revision of the understanding of Saccharum diversity.

Genomic predictions improve clonal selection in oil palm (Elaeis guineensis Jacq.) hybrids.

The prediction of clonal genetic value for yield is challenging in oil palm (Elaeis guineensis Jacq.). Currently, clonal selection involves two stages of phenotypic selection (PS): ortet preselection on traits with sufficient heritability among a small number of individuals in the best crosses in progeny tests, and final selection on performance in clonal trials. The present study evaluated the efficiency of genomic selection (GS) for clonal selection. The training set comprised almost 300 Deli × La Mé crosses phenotyped for eight palm oil yield components and the validation set 42 Deli × La Mé ortets. Genotyping-by-sequencing (GBS) revealed 15,054 single nucleotide polymorphisms (SNP). The effects of the SNP dataset (density and percentage of missing data) and two GS modeling approaches, ignoring (ASGM) and considering (PSAM) the parental origin of alleles, were assessed. The results showed prediction accuracies ranging from 0.08 to 0.70 for ortet candidates without data records, depending on trait, SNP dataset and modeling. ASGM was better (on average slightly more accurate, less sensitive to SNP dataset and simpler), although PSAM appeared interesting for a few traits. With ASGM, the number of SNPs had to reach 7,000, while the percentage of missing data per SNP was of secondary importance, and GS prediction accuracies were higher than those of PS for most of the traits. Finally, this makes possible two practical applications of GS, that will increase genetic progress by improving ortet preselection before clonal trials: (1) preselection at the mature stage on all yield components jointly using ortet genotypes and phenotypes, and (2) genomic preselection on more yield components than PS, among a large population of the best possible crosses at nursery stage.