Comparative Assessment of Models for Genome Wide Association Studies in a New Sugarcane Diversity Panel

Aneu-polyploidy (x = 8-10) with varying chromosome number (2n ≈ 80-140) of sugarcane hybrids and the lack of a draft genome poses unique challenges to fine mapping and cloning of quantitative trait loci controlling agronomic traits of interest. However, high linkage disequilibrium among modern sugarcane cultivars offers better opportunity for genome wide association studies (GWAS) over bi-parental mapping to identify molecular markers associated with specific trait(s). In order to develop a diversity panel (DP) for GWAS of different agronomic traits, genotyping of 1,485 clones (1,236 clones from the world collection of sugarcane and related grasses and 249 clones from Louisiana breeding programs) was done with 11 SSR (simple sequence repeat) markers spanning sorghum chromosomes. A nearest neighbor-joining tree was constructed using 423 resultant SSR alleles. A maximum length subtree was used to inform panel selection resulting in a Louisiana breeder-preferred 300-clone diversity panel. To identify the best model for GWAS studies using the DP, different Q (structure)-K (kinship) combinations were compared. Four different Q matrices, Tassel and JMP’s PCA (principle component analysis), JMP’s MSD (minimum separable distance) and Structure’s Bayesian model; and, Tassel and JMP’s IBS (identity-by-state) K matrices were compared. A mixed population of 64 clones with the genotypic data of 1,041 SSR and 23 AFLP alleles, and 2 years of sucrose yield traits (brix, percentage sucrose and pol) was used in the study to identify markers associated with sucrose traits. Results of marker-trait association from different Q-K analyses were comparable with minor differences found only among the Q matrices.