Inferring heterotic patterns and the effect of incorporating dominance deviations for hybrid prediction: an example in tropical maize under nitrogen stress conditions

• Ano de defesa: 2020
• Autor(a) principal: Morosini, Julia Silva
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Zea mays; Capacidades de combinação; Combining abilities; Dominance; Estrutura de população; Grupos heteróticos; Heterotic groups; Hybrid breeding; Melhoramento; Nitrogen stress; Population structure
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/11/11137/tde-15072020-121959/

Hybrid breeding exploits heterosis, a phenomenon that relies on the development of heterotic groups (HG). Hybrid prediction incorporates relationship information in genetic-statistical analyses to increase the accuracy and to improve the assessment of individuals\' genetic values. Such relationship matrices, however, often neglect non-additive effects. Moreover, a critical factor in a hybrid breeding program is the genomic prediction being performed based on the same allelic effects for two heterotic groups. Finally, nitrogen (N) deficiency is a major constraint for maize productivity and may significantly confound inferences from genetic approaches. In this study, we considered models including additive and the combination of additive+dominance effects for the estimation of combining abilities, the determination of heterotic groups, and the impact of the differential modeling of marker effects on heterotic groups in a maize breeding population. We also investigated the N stress effect on these parameters. For that, 906 single crosses obtained from a diallel scheme of 49 inbred maize lines were genotyped in silico using 34,571 SNP and evaluated in four environments in State of São Paulo, Brazil, each with two N regimes: ideal (IN) and stress (LN). Three modeling scenarios were considered: pedigree-based (I), where no genomic relationship information was considered; additive (Ga), where an additive incidence matrix was assigned to lines; and additive+dominance (Ga+d), where the additive and the dominance effects were considered for lines and hybrids, respectively. HG were defined based on the specific combining ability (SCA) in each scenario. Prediction abilities (PA) were obtained using a 5-fold cross-validation approach. Our results indicate that the incorporation of both additive effects and dominance deviations allows us to discriminate the estimates of general and specific combining abilities better. SCA assigned distinct clustering of parents for each scenario. More considerable differences in heterotic pool composition occurred between scenarios Ga and Ga+d, although there was still a meaningful overlap. Incorporating HG into the prediction analysis provided a significantly increased SCA of the single crosses, and a noteworthy increase in PA. The genetic estimates were lower for LN compared to IN (significant at the 0.01 level), and the genetic distance between groups was higher for IN. Our findings reveal that additive effects are well assessed by using pedigree data when the population lacks structure. In addition, clustering HG based on SCA from the Ga+d scenario is a useful approach for reciprocal recurrent selection when no previous information is available about population genetic structure. We also observed that N stress hinders resolution in determining heterotic patterns, which compromises the exploration of heterosis. Ultimately, the incorporation of dominance provides relevant information for the determination of heterotic groups, and the differential modeling of marker effects for each HG is crucial to a sustainable breeding program.