Optimizing predictions for applying genomic selection to texture in Apple

International audience; Although genomic selection promised to increase drastically breeding efficiency in perennial crops, it is not yet broadly applied in fruit trees. Among the several targets in apple breeding, fruit texture is an essential feature affecting both consumers’ appreciation and storage performance. Due to the fixation of alleles at the major genes underlying texture in modern breeding programs, the available genetic variability for this trait might rely on minor effect alleles, which can be targeted with genomic selection. In this study, we aimed to assess the feasibility of genomic selection for texture in a population of 537 individuals consisting in a collection of 259 individuals and 6 full-sib biparental families, all genotyped with 8,294 SNPs. We dissected the fruit texture complexity by measuring twelve acoustic and mechanical subtraits with a TA.XTplus texture analyzer. We applied different scenarios for the design of the training population with and without taking into account genetic parameters. The use of the entire collection to predict families was not the most efficient strategy, since families were best predicted using training populations composed of the 10 to 202 most related individuals. Training population optimization allowed increasing prediction accuracies by 0.17 on average, reaching a maximum accuracy of 0.78 for predicting the number of force peaks in the family ‘Gala x Pink Lady’. As the two first principal components of texture traits summarized firmness (PC1) and crispiness (PC2), we predicted these two synthetic traits and obtained maximum accuracies of 0.81 and 0.40, respectively. Furthermore, our results indicate that although genetic clustering may be correlated to texture profiles, genetic relatedness might already capture the clustering effect in predictions. Our work sheds light on the high potential for predicting texture in apple. The methods presented here also provide insights for efficiently designing genomic selection trials in fruit trees.