Genetic Dissection And Breeding Application Of Kernel Traits In Maize Multi-parent Population

High yield is the one of the primary goals pursued by maize breeders.Kernel yield is the complex quantitative trait,and kernel traits would be to decompose into component traits such as length,width,thickness and test weight for kernel.In this study,the genetic basis of maize kernel traits were analyzed by genome-wide association studies(GWAS)in a 24-parent synthetic population(Complete-diallel design plus an Unbalanced Breeding-like Inter-Cross population,CUBIC).Through the analysis of QTL pleiotropism and interaction among the QTLand the environments,the breeding application value of the identified QTLs was analyzed.The main results are as follows:1.Genome wide association analysis based on the SNP(Single nucleotide polymorphism)markers(namely s GWAS).Based on the 11.8 M SNPs(MAF?0.02)for the whole genome,171 loci(namely s QTL,p < 1.23E-8)were detected to affect seven kernel traits.The loci interval ranged from 50.0 Kb to 15.5 Mb.The phenotypic variation explained by single locus varied from 0.13% to 12.11%.Of all the loci,54 loci explained phenotypic variation beyond 5%,while only 6 loci beyond 10%.By joining all the detected loci for each trait,they can explain phenotypic variance between 22.28% and 54.63%.Most loci had relatively weak additive effects,reflecting a typical polygenic nature of quantitative traits.According to local linkage disequilibrium(LD)of QTL territory and gene annotation,totally 137 candidate genes for kernel related traits were predicted.Gene Ontology(GO)enrichment analysis revealed that the 137 candidate genes were enriched on some functional proteins(52.55%)and enzymes(32.85%),with a fraction of transcriptional factor(5.84%).2.Genome wide association analysis(namely s GWAS)based on the identity-bydescent(IBD).A total of 128 loci(namely h QTL,LRT?6.8)were identified for seven kernel traits.Single locus explained phenotypic variation between 1.93% and 12.16%.Of all the loci,96 loci explained phenotypic variance beyond 5% while only 13 loci beyond 10%.The loci interval size varied from 1.03 Kb to 4.06 Mb,109 loci(85.16%)of which were mapped to a region less than 1 Mb.Further analysis found that,24 CUBIC parents are able to contribute to the optimal haplotype from 1 to 11 loci,respectively,among which,the parents E28 and Dan340 contributed to the most optimal haplotypes.3.Candidate gene analysis.A major locus on chromosome 10(147.7-149.6 Mb)controlled total kernel number per liter based on the two GWAS methods,which was further explored for the underlying candidate genes.In the locus interval,there were 87 nonsynonymous SNP mutations that probably cause amino acid change,spanning across 65 genes.Of which,only 5 SNPs exceeded GWAS significance threshold that involved 4 genes.Moreover,a candidate-gene association study within the locus region using In Del mutations revealed 3 significant In Dels influencing total kernel number per liter.In total,the gene GRMZM2G173636 was determined to be the putative candidate gene responsible for this QTL,named as Zm ACBP6.4.Analysis for pleiotropic QTL.Totally,for seven kernel related traits,we detected 171 s QTLs and 128 h QTLs,which were not evenly distributed across the genome,but enriched to some specific regions as QTL hotspots.Based on the h QTLs for different traits,a total of 26 pleiotropic QTLs were identified in this study,locating on all chromosomes except for the chromosome 9,the majority of which(19/26)merely influenced two kernel related traits.Combining with previous QTL results,a total of 8 kernel related pleiotropic QTLs were found to influence flowering time,plant architecture and ear traits.In the sense of the breeding aspect,these 8 pleiotropic traits have two implications: i)“Win-win linkage QTL”,a linkage phase of favorable alleles for different traits,which can be genetically improved simultaneously.The pleiotropic p QTL16 and p QTL17 is the case;ii)“Linkage drag QTL”,a linkage phase between favorable and unfavorable alleles of different traits,which need to be balanced selected in breeding.The pleiotropic p QTL1 is the case.5.The interaction between QTL and environments.Based on the co-localized 24 QTLs between two GWAS methods,the interaction between QTL and environments were analyzed.The coefficient of variation of QTL additive effects across five environments were calculated for each QTL,as a measurement for quantifying the interaction level.In summary,different QTLs interacted distinctly with environmental alternations,the Q×E level of which varied from 0.11 to 0.81(CV).Accordingly,the 24 QTL were categorized into three scenarios: i)strong interaction when 0.50< CV <0.81.It would be more visible and reliable to apply this type of QTL in molecular marker assisted selection(MAS)under specific environment;ii)weak interaction when 0.11<CV<0.20.This type of QTL can express stably across different environments,which can preferentially be useful for MAS programs;iii)median interaction when 0.20<CV<0.50.The type of QTL can be stably used in several environments while the caution is needed.