Accuracy and responses of genomic selection on key traits in apple breeding

Abstract

The application of genomic selection in fruit tree crops is expected to enhance breeding efficiency by increasing prediction accuracy, increasing selection intensity and decreasing generation interval. The objectives of this study were to assess the accuracy of prediction and selection response in commercial apple breeding programmes for key traits. The training population comprised 977 individuals derived from 20 pedigreed full-sib families. Historic phenotypic data were available on 10 traits related to productivity and fruit external appearance and genotypic data for 7829 SNPs obtained with an Illumina 20K SNP array. From these data, a genome-wide prediction model was built and subsequently used to calculate genomic breeding values of five application full-sib families. The application families had genotypes at 364 SNPs from a dedicated 512 SNP array, and these genotypic data were extended to the high-density level by imputation. These five families were phenotyped for 1 year and their phenotypes were compared to the predicted breeding values. Accuracy of genomic prediction across the 10 traits reached a maximum value of 0.5 and had a median value of 0.19. The accuracies were strongly affected by the phenotypic distribution and heritability of traits. In the largest family, significant selection response was observed for traits with high heritability and symmetric phenotypic distribution. Traits that showed non-significant response often had reduced and skewed phenotypic variation or low heritability. Among the five application families the accuracies were uncorrelated to the degree of relatedness to the training population. The results underline the potential of genomic prediction to accelerate breeding progress in outbred fruit tree crops that still need to overcome long generation intervals and extensive phenotyping costs.

Combining Genome-Wide Information with a Functional Structural Plant Model to Simulate 1-Year-Old Apple Tree Architecture

“…Correlations between estimated and observed values from the genome wide prediction model were high, showing that the model was able to estimate genotypic differences from SNP polymorphisms. However, during the cross validation, the accuracy was much lower even though similar to previous studies on apple (Kumar et al, 2012; Muranty et al, 2015) or on other species (Rutkoski et al, 2013; Fodor et al, 2014). This highlights the limit of our genome wide prediction model to predict phenotypic values for genotypes that are not included in the training set.…”

“…This highlights the limit of our genome wide prediction model to predict phenotypic values for genotypes that are not included in the training set. The prediction model accuracy is influenced by the density of the genetic map (Kumar et al, 2012; Heslot et al, 2013), the population size (Brito et al, 2011), and the heritability of traits (Combs and Bernardo, 2013; Muranty et al, 2015). Among the arguments mentioned above, the population size is probably the most limiting factor in the present study.…”

Estimation of total genetic effects for survival time in crossbred laying hens showing cannibalism, using pedigree or genomic information

“…Muranty et al . () observed that genomic prediction accuracy is highly influenced by phenotypic distribution. The choice of model may therefore have affected the outcome of our results.…”

“…Shortcomings of this model are that censored records are considered as true survival time records and that the non-normality of survival time records violates the normality assumption of the linear mixed model. Muranty et al (2015) observed that genomic prediction accuracy is highly influenced by phenotypic distribution. The choice of model may therefore have affected the outcome of our results.…”

Regularized quantile regression applied to genome-enabled prediction of quantitative traits

“…Muranty et al (2015) verified that accuracy was strongly affected by phenotypic distribution. Specifically, traits that showed poor results often had skewed phenotypic distributions or low heritability.…”