Search

International audience; Current research in genomics and ecophysiology strive to improve genetophenotype predictions beyond methods based solely on genetic information such as whole genome prediction (WGP). Predictive approaches coupling quantitative genetics and crop modeling are designed, either with an emphasis on ecophysiology or on statistics, according to the modeling scale of traits: from molecular to crop level (BustosKorts et al., 2016). Genetophenotype predictions are generally carried to predict the breeding values from genomewide information using statistical models. The difficulty with this method lies in the prediction of nonadditive gene effect and genebyenvironment interactions. Crop simulation models, by representing dynamic responses of crop and soil processes to the environment, are successfully used to predict phenotypic plasticity for agronomical traits (Chenu et al., 2017). Therefore, coupling quantitative genetics and processbased modeling should improve the accuracy of breeding value prediction, with different options depending on the phenotypic distance, as illustrated by Hammer et al. (2016). This study aims to predict the performance of a panel of sunflower hybrids in different farming environments, and therefore to identify the most promising genotypes and key physiological traits controlling yield stability and adaptation to future climatic scenarios. Our multiscale modeling strategy proceeded in two steps: (1) building whole genome prediction models to predict a set of component traits as a function of allelic combination of genotypes and (2) predicting the hybrid yield as a function of component traits, environmental variables and management practices using a crop simulation model (Casadebaig et al., 2016). We evaluated the accuracy of this multiscale model on a multienvironmental network (13 environments) where the training population was phenotyped and on independent trials with nonobserved genotypes and environments (9 environments). Over these 22 environments, we found that the overall accuracy of our multiscale model was worse than the accuracy of a single genomic prediction model fitted for crop yield. The WGP model showed a good accuracy on the training environments and on the test environments including hybrids selected for their yield range. However, the multiscale model showed a better accuracy than the WGP model in 4 out 9 test environments, specifically when tested hybrids were not selected for their extreme yield levels. Overall, we identified new bottlenecks related to the design of combined crop and genetic models, such as the definition of component traits articulating genomic and ecophysiological models and the environment(s) used to train the genomic models. …/…

International audience; Prediction of hybrid performance using incomplete factorial mating designs is widely used in breeding programs including different heterotic groups. Based on the general combining ability (GCA) of the parents, predictions are accurate only if the genetic variance resulting from the specific combining ability is small and both parents have phenotyped descendants. Genomic selection (GS) can predict performance using a model trained on both phenotyped and genotyped hybrids that do not necessarily include all hybrid parents. Therefore. GS could overcome the issue of unknown parent GCA. Here, we compared the accuracy of classical GCA-based and genomic predictions for oil content of sunflower seeds using several GS models. Our study involved 452 sunflower hybrids from an incomplete factorial design of 36 female and 36 male lines. Re-sequencing of parental lines allowed to identify 468,194 non-redundant SNPs and to infer the hybrid genotypes. Oil content was observed in a multi-environment trial (MET) over 3 years, leading to nine different environments. We compared GCA-based model to different GS models including female and male genomic kinships with the addition of the female-by-male interaction genomic kinship, the use of functional knowledge as SNPs in genes of oil metabolic pathways, and with epistasis modeling. When both parents have descendants in the training set, the predictive ability was high even for GCA-based prediction, with an average MET value of 0.782. GS performed slightly better (+0.2%). Neither the inclusion of the female-by-male interaction, nor functional knowledge of oil metabolism, nor epistasis modeling improved the GS accuracy. GS greatly improved predictive ability when one or both parents were untested in the training set, increasing GCA-based predictive ability by 10.4% from 0.575 to 0.635 in the MET. In this scenario, performing GS only considering SNPs in oil metabolic pathways did not improve whole genome GS prediction but increased GCA-based prediction ability by 6.4%. Our results show that GS is a major improvement to breeding efficiency compared to the classical GCA modeling when either one or both parents are not well-characterized. This finding could therefore accelerate breeding through reducing phenotyping efforts and more effectively targeting for the most promising crosses.