Abstract
Vase life is a key trait for carnation cut flowers. A better understanding of the genetic basis of vase life is needed to implement selection methods like marker-assisted selection or genomic selection. Our objective was to investigate the genetic architecture of vase life and evaluate the potential of including known QTL in genomic prediction models to improve prediction accuracy. We constructed a linkage map from two segregating F1 carnation populations based on 5,412 SNP markers. Quantitative trait loci analysis detected one QTL for each of the two populations. The QTL were located on chromosomes 10 for population 1 and chromosome 11 for population 2 and accounted for 2.84 and 5.09% phenotypic variation, respectively. A genome-wide association study revealed potential genomic regions of interest for vase life. All detected markers accounted individually for less than 6% of phenotypic variation and were spread across ten chromosomes. This suggests that vase life is a polygenic trait. We conducted a cross-validation study in which 1 through 120 top scoring SNP markers were fitted as fixed effects to the baseline rigde-regression best unbiased linear prediction model. Fitting 1 to 50 top scoring SNP markers improved prediction accuracy compared to the baseline model by 0.016 on average. Fitting the 25 top scoring SNP markers resulted in the greatest prediction accuracy improvement, from 0.75 to 0.78. These findings extend the knowledge on the genetic basis of vase life and attest to the potential of genomic selection for breeding cut carnation with longer vase life.