Genomic Selection For Yield-related Traits In Maize

Maize?Zea mays L.?originates from Central America and is adapted to temperate environment through domestication and artificial selection.As one of the most important sources for food,forage,bio-energy and wine-making industry,maize has a pivotal role in the development of national economy and ensurance of food security.It is important to use molecular breeding methods to improve maize production and accelerate breeding process.As a valid molecular breeding strategy,genomic selection?GS?has been widely practiced in plant breeding,and it has remarkable superiority in enhancing genetic gain,reducing breeding time,and accelerating breeding process.In this study,natural and multiple bi-parental populations were used to evaluate the factors affecting prediction accuracy in GS with a series of suggestions proposed for breeding programs.The trait-related markers identified by genome-wide association study?GWAS?and QTL mapping have been used to improve the prediction accuracy in GS.A better prediction accuracy could be obtained when an early generation bi-parental population was utilized as training population to predict the performance of its advanced-generation,providing guidance for plant breeding.1.Factors affecting prediction accuracy of GS and their application in breedingIn this study the factors affecting prediction accuracy(r _MG)in GS were evaluated systematically.The factors examined included marker density,population size,heritability,statistical model,population relationships and the ratio of population size between the training and testing sets.Prediction accuracy continuously increased as marker density and population size increased until it reaches plateau,and it was positively correlated with heritability;r_MG showed a slight gain when the training set increased to three times as large as the testing set.Low predictive performance between unrelated populations could be attributed to different allele frequencies,and predictive ability and prediction accuracy could be improved by including more related lines in the training population.A similar prediction ability was achieved with different statistical models,and non-additive effects should be taken into account when the GS is implemented in heterozygous populations.2.Genomic selection with trait-related markerGWAS and QTL mapping were used to identify trait-related markers using natural and bi-parental populations,and these trait-related marker loci were used to implement GS for evaluating prediction accuracy.The prediction accuracy was improved when the trait-related markers were used in GS.In addition,the prediction accuracy was improved for most traits when the marker loci closely related to the target trait were considered as fixed effects in statistical model.Therefore,the trait-relevant markers identified in previous genetics studies can be used to improve prediction ability and thus the breeding efficiency in breeding programs.3.Genomic selection with GCA?general combining ability?-related markersGWAS and QTL mapping identified trait-related marker loci across the whole genome for GCA effects on yield-related traits.Then these GCA-relevant markers and randomly selected markers were used to implement 10-fold cross-validation.Using GCA-related markers in GS could efficiently improve prediction accuracy,compared with using random markers.Meanwhile prediction accuracy could be largely enhanced for the traits with low GCA heritability,and much larger than the traits with high GCA heritability.4.Genomic selection using bi-parental populations of different generationsThe bi-parental populations of different generations derived from same parents were used to perform GS for yield-related traits and their GCA effects.Using the F_2:3 population to predict RIL&DH and F₅ populations and using F₅ population to predict RIL&DH population,a relatively high prediction accuracy was obtained for most traits.Besides,a moderate prediction accuracy for most yield-related traits was obtained when different testcross populations were used in GS.Therefore,selection of candidate breeding lines having heterosis with the testers can be achieved by constructing suitable training and testcross populations,and thus it can be used to enhance breeding efficiency and accelerate breeding process.