Melhor Predição Linear Imparcial (BLUP) no melhoramento vegetal: seleção entre famílias de meios-irmãos

• Ano de defesa: 2008
• Autor(a) principal: Faria, Vinícius Ribeiro
• Orientador(a): Viana, José Marcelo Soriano
• Banca de defesa: Barros, Willian Silva , Miranda, Glauco Vieira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Viçosa
• Programa de Pós-Graduação: Mestrado em Genética e Melhoramento
• Departamento: Genética animal; Genética molecular e de microrganismos; Genética quantitativa; Genética vegetal; Me
• País: BR
• Palavras-chave em Português: BLUP; Modelo de reprodutor; Seleção recorrente
• Palavras-chave em Inglês: BLUP; Beeder model; Recurrent selection
• Área do conhecimento CNPq: CNPQ::CIENCIAS BIOLOGICAS::GENETICA::GENETICA QUANTITATIVA
• Link de acesso: http://locus.ufv.br/handle/123456789/4776

The objectives of the present work were to install the REML/BLUP methodology in the recurrent populational improvement, using half-brothers families, to compare this methodology with the traditional analysis of the ordinary least squares and to present the SAS routines for the analyses of mixed models. The model used corresponds to the beeder model used in the animal improvement. The computer analyses were carried out using the computer software SAS® through proc mixed, proc inbreed, proc iml and proc gml procedures. As the theory support, data from three cycles of recurrent selection in the population of Voçosa por-corn, of the Programa de Melhoramento do Setor de Genética (Improvement Program of the Genetic Sector), do Departamento de biologia Gerald a Universidade Federal de Viçosa (Department of General Biology of the Viçosa Federal University) were used. The experiments were designed in single 14 x 14 lattice. In each block of the experiment, the production and the expansion ability (CE) in ml/g were evaluated. For each improvement cycle, the variance components were estimated the genotypic means were predicted, according to the REML/BLUP methodology. For each cycle the analysis by the ordinary Least Squares (OLS) was also made with the estimation of the variance components and of the genotypic means. Besides, the gains due to the selection in each cycle mere calculated. Moreover the ordination of the selected individuals was verified and the confidence intervals for the genetic values predicted and estimated were built. By the variance analysis, it could be verified that as the improvement cycles progressed, there was a sifnificant reduction in the variability for CE, and in the third selection cycle the variability was no more detected. This variability reduction can be explained by the high selection intensity, because the 20 families selected in the third cycle were from only mine plants of the first cycle of selection. Referring the genetic parameters estimated by the two methodologies, it was observed that both with the REML and the OLS methodology the estimate values for the variance components were similar. For the genotypic means, when there was no relationship among the individuals there was an accordance among the best individuals estimated and predicted, but the predicted values presented narrower confidence intervals, indicating that these values are more accurate than the estimated ones. When the relationship information was included, there was a reduction on the coincidence values between the selected ones and those of the Speerman co-relation indicating that the use of the prodictd genotypic means in the selection of superior individuals will provide greater gains due to the selection.