Autoria: MORAIS JÚNIOR, O. P.; DUARTE, J. B.; BRESEGHELLO, F.; COELHO, A. S. G.; BORBA, T. C. O.; AGUIAR, J. T.; NEVES, P. C. F.; MORAIS, O. P.

Resumo: In genomic recurrent selection programs of self-pollinated crops, additive genetic effects (breeding values) are effectively relevant for selection of superior progenies as new parents. However, considering additive and nonadditive genetic effects can improve the prediction of genome-enhanced breeding values (GEBV) of progenies, for quantitative traits. In this study, we assessed the magnitude of additive and nonadditive genetic variances for eight key traits in a rice population under recurrent selection, using marker-based relationship matrices. We then assessed the goodness-to-fit, bias, stability and accuracy of prediction for breeding values and total (additive plus nonadditive) genetic values, in five genomic best linear unbiased prediction (GBLUP) models, ignoring or not nonadditive genetic effects. The models were compared using 6174 single nucleotide polymorphisms (SNP) markers from 174 S1:3 progenies evaluated in field yield trial. We found dominance effects accounting for a substantial proportion of the total genetic variance for the key traits in rice, especially for days to flowering. In average of the traits, the component of variance additive, dominance, and epistatic contributed to about 34%, 14% and 9% for phenotypic variance. Additive genomic models, ignoring nonadditive genetic effects, showed better fit to the data and lower bias, in addition to greater stability and accuracy for predict GEBV of progenies. These results improve our understanding of the genetic architecture of the key traits in rice, evaluated in early-generation testing. Clearly, this study highlighted the advantages of additive models using genome-wide information, for genomic prediction applied to recurrent selection in a self-pollinated crop.

Ano de publicação: 2017

Tipo de publicação: Artigo de periódico

Unidade: Embrapa Arroz e Feijão

Palavras-chave: Arroz, GBLUP models, Genetic architecture, Melhoramento genético vegetal, Oryza sativa, Predictive accuracy, Seleção recorrente, Variance components