Skip to Main content Skip to Navigation

Optimisation du schéma de sélection chez le blé tendre : apport des prédictions génomiques et des caractères corrélés

Abstract : Breeding consists in creating new varieties which combine qualities for several traits of agronomic interest to answer to the market demand. The objective of the phD was to propose strategies using genomic predictions to optimize bread wheat breeding programs in terms of genetic gain under economic constraint. In a first chapter, we tested methods aiming at improving genomic prediction accuracy of a trait that is expensive to measure using a correlated cheap trait, without increasing the budget allocated to phenotyping. We used a multi-trait genomic prediction models. We also developed an index called CDmulti to optimize the choice of a subset of lines to phenotype for two different correlated traits. We showed that multi-trait genomic predictions could be particularly interesting when lines of the validation set, or at least part of them, could be phenotyped for dough strength, which is correlated to bread-making quality and which is cheaper to phenotype. Indeed, this approach allowed to reduce the budget allocated to phenotyping without decreasing the genomic prediction accuracy of bread-making quality. In a second chapter, we developed a stochastic simulation pipeline to compare breeding scheme produce in silico, using genotyping and phenotyping of a reference population. One cycle lasts five years, including one year for crossing, one year for double haploids production, one year for seed multiplication, one year of selection based on either phenotypic value (PS strategy) or genomic predictions (GPS strategy), and one last year of phenotypic selection. For GPS strategy, we can mate the best lines of previous cycle at random or optimise mating using genomic predictions. We showed that GPS strategy with mating optimization is systematically significantly superior to other strategies for all tested parameters (trait heritability, budget, relative intensity of selection at two key steps). The efficiency of GPS strategy without mating optimization was similar to PS. However, the loss of genetic diversity was more intense for GPS strategies, with or without mating optimization. Some complementary modules will be added to this decision tool to simulate more realistic breeding schemes.
Document type :
Complete list of metadata
Contributor : Abes Star :  Contact Connect in order to contact the contributor
Submitted on : Monday, October 4, 2021 - 12:59:16 PM
Last modification on : Thursday, October 7, 2021 - 4:40:26 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03364017, version 1



Sarah Ben Sadoun. Optimisation du schéma de sélection chez le blé tendre : apport des prédictions génomiques et des caractères corrélés. Sciences agricoles. Université Clermont Auvergne, 2020. Français. ⟨NNT : 2020CLFAC014⟩. ⟨tel-03364017⟩



Record views


Files downloads