Genomic breeding approaches for East African bananas

The polyploidy nature of banana is a limiting factor in the implementation of strategies such as marker assisted selection (MAS) or genome wide association mapping (GWAS). The triploid nature of cultivated varieties complicates conventional breeding strategies and improved varieties can take up to 20 years before they can be released to the public, which necessitates the use of efficient molecular tools to more rapidly respond to abiotic and biotic stresses and to address the needs of growers and consumers. In addition, the high cost of phenotyping perennial large-stature plants such as banana, and the rapidly decreasing cost of genotyping, makes the use of predictive genomic selection models using single nucleotide polymorphism (SNP) markers extremely attractive to banana breeders. A Genomic Selection (GS) training population consisting of 307 banana genotypes was developed for initial analysis with ploidy levels of the plant material ranging from diploids to tetraploids. Plants were genotyped using the genotyping by sequencing (GBS) approach (Elshire et al., 2011) with PstI as the sole restriction enzyme. Sequence data was processed through a bioinformatics workflow and single nucleotide polymorphisms (SNPs) were called using the genomic analysis tool kit (GATK). Data was filtered for quality and for loci with >50% missing data. Phenotypic data for 25 traits are being collected from two locations since 2012. Yield-related traits (fruit pulp diameter, bunch weight, number of suckers, etc.) are collected at flowering and harvest Analysis of GBS data resulted in 11201 SNP loci. The results of multiple prediction models are discussed and compared.