Development of a Genome-Wide Selection Model to Sugarcane Breeding Using High Throughput Sequence Capture Genotyping

Sugarcane (Saccharum spp.) has one of the most complex genomes among crops. Modern cultivars are interspecific hybrids of two polyploid species, with a large genome (~10 Gb), in which different chromosome numbers for each homeologous groups are observed. The recent development of high-throughput SNP genotyping paved the way to the use of statistical models in breeding programs based on a great number of molecular markers widespread in the genome of many species. In order to evaluate the feasibility and potential of genome-wide selection (GWS) for sugarcane breeding, we developed genome prediction models using six different methods. Four hundred and eighty individuals were genotyped for 37,913 SNPs and phenotyped for six traits related to plant growth, biomass production and sugar content. The accuracy of each model was assessed through the ratio of the correlation between observed and predicted breeding values, estimated by ten-fold cross validation, and the square root of heritability. Our results showed that sugarcane breeding should greatly benefit from the use of GWS approach, since high accuracy values were estimated for most traits (>0.50). Models obtained by the six different methods showed similar accuracies, suggesting that no significant gains are expected when different methods of analysis are applied, corroborating with results described in the literature for other crops. Our results provide evidence that, if properly applied, the GWS strategy can increase the genetic gain, and reduce the time cycle and costs involved in sugarcane breeding programs.

Financial Support: Ridesa; Petrobras.