QTL Mapping and Genomic Selection for Fusarium Ear Rot Resistance Using Two F2:3 Populations in Maize

Guo Zifeng,Wang Shanhong,Li Wen-Xue,Liu Jiacheng,Guo Wei,Xu Mingliang,Xu Yunbi

DOI: https://doi.org/10.1007/s10681-022-03083-z

IF: 2.1849

2022-01-01

Euphytica

Abstract:Fusarium ear rot (FER) is a common disease of maize that can seriously affect the yield and quality of maize. Resistance to FER is highly quantitative, largely under polygenic control with minor effects, and strongly influenced by environmental conditions. In this study, two F2:3 populations with 220 and 152 lines were developed from PHW43 × 8107 (PH81) and Zong31 × DTMA165 (ZODT) for QTL mapping and genomic selection (GS). Artificial inoculation was performed to evaluate FER resistance in four environments, Gongzhuling 2019, Shunyi 2019, Xinxiang 2019, and Xinxiang 2020. For both populations, we constructed a relatively high-density linkage map. Each F2:3 population showed a lower average FER severity than the natural population tested before. Significant genotypic variance, environment variance, and genotype × environment interaction variance were detected in both populations by combined ANOVA analysis. The combined broad-sense heritability of the FER severity was estimated as 0.60 and 0.67 in PH81 and ZODT, respectively. In PH81, seven QTL were identified on chromosomes 1, 4, 6, 7, and 10, four of which were detected in Xinxiang 2019 environment together explaining 24.15% of the total phenotypic variation. The QTL qPBLUE–7–1 on bin7.01 explaining the highest phenotypic variation (6.14–8.60%) was consistently identified across Xinxiang 2019 environment and the best linear unbiased estimators. Eight QTL were identified in ZOTD, explaining 7.11% (qZBLUE–6–1) to 13.29% (qZS19–10–1) of phenotypic variation. Two candidate genes were identified in bin 6.05, of which GRMZM2G066153 encodes an enzyme involving the defense of pathogens in many plants. Genomic best linear unbiased prediction (GBLUP) showed that 500 SNP markers were enough for F2:3 populations to achieve reasonably good prediction. The prediction accuracy in F2:3 populations could be improved by integrating QTL information into the GBLUP model and by using the extended GBLUP model with additive and dominance effects included.