Modelos mistos na avaliação e ordenação de genótipos de cana-de-açúcar, com e sem efeitos de competição com parcelas vizinhas
Carregando...
Data
2009-07-26
Autores
Candido, Liliam Silvia [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
Nos programas de melhoramento de cana-de-açúcar grande número de clones são avaliados todos os anos, em experimentos realizados em diferentes safras, épocas e regiões. E como é cada vez mais difícil selecionar os melhores genótipos fenotipicamente, o uso de métodos precisos de análise estatística é necessário, a fim de garantir maior confiabilidade ao processo seletivo. A avaliação genética com base em modelos mistos do tipo REML/BLUP (máxima verossimilhança restrita/ melhor predição linear não-viesada) tem merecido atenção especial dos pesquisadores. Outra questão importante durante o processo de seleção é a avaliação de possíveis efeitos da alocompetição sofrida pelos genótipos nos experimentos, pois, um genótipo em alocompetição pode ter comportamento diferente quando plantado comercialmente, apenas em autocompetição. Por essa razão, estratégias de análise que modelam a dependência espacial na forma de covariância, têm sido utilizadas no melhoramento com o objetivo de neutralizar os efeitos da competição entre parcelas vizinhas, além de aumentar a precisão experimental. Para o estudo desses fatores, foram utilizados dados de 1º e 3º corte, do atributo tonelada de cana/ha (TCH), da rede de ensaios estaduais do programa de melhoramento do Centro de Cana–IAC, instalados no ano de 2002. O delineamento experimental foi o de blocos casualizados com 3 repetições, sendo que a casualização em cada bloco foi feita uma única vez, de forma que em todos os experimentos, os genótipos tivessem sempre os mesmos vizinhos. As parcelas experimentais foram constituídas de cinco linhas de 8m espaçadas a 1,50m. De maneira geral, a análise por modelo misto e a análise tradicional apresentaram poucas diferenças em relação aos parâmetros avaliados e ao ordenamento dos genótipos, assim...
In the programs for sugarcane improvement large number of clones are evaluated each year, in experiments realized in different crops, seasons and regions. And it is increasingly difficult to select the best genotypes phenotypically, the use of precise methods of statistical analysis is needed to ensure greater reliability in the selection process. The genetic evaluation based on mixed models of the type REML / BLUP (restricted maximum likelihood / best linear unbiased prediction) has deserved special attention from researchers. Another important issue during the selection process is the evaluation of possible effects of allocompetition suffered by the genotypes in the experiments, because a genotype in allocompetition may have different behavior when planted commercially, only in autocompetition. Therefore, strategies of analysis to model the spatial dependence in the form of covariance, have been used in improvement with the objective of neutralizing the effects of competition between neighboring plots, besides increasing the experimental precision. For the study of these factors, we used data from 1st and 3rd cuts, of the attribute tons of cane per hectare (TCH), than the network experiments state test of the improvement program of the Centro de Cana-IAC, Brazil, installed in 2002. The experimental design was a randomized block with 3 replications, with each block randomization was performed only once, so that in all experiments, the genotypes have the same neighbors. The experimental plots consisted of five lines of 8 m spaced to 1.50 m. In general, the mixed model analysis and traditional analysis showed little difference with respect to the parameters evaluated and the ranking of genotypes, as well as the use of different methodologies for the neighborhood analysis contributed little to increase the experimental precision. The effects of... (Complete abstract, click electonic access below)
In the programs for sugarcane improvement large number of clones are evaluated each year, in experiments realized in different crops, seasons and regions. And it is increasingly difficult to select the best genotypes phenotypically, the use of precise methods of statistical analysis is needed to ensure greater reliability in the selection process. The genetic evaluation based on mixed models of the type REML / BLUP (restricted maximum likelihood / best linear unbiased prediction) has deserved special attention from researchers. Another important issue during the selection process is the evaluation of possible effects of allocompetition suffered by the genotypes in the experiments, because a genotype in allocompetition may have different behavior when planted commercially, only in autocompetition. Therefore, strategies of analysis to model the spatial dependence in the form of covariance, have been used in improvement with the objective of neutralizing the effects of competition between neighboring plots, besides increasing the experimental precision. For the study of these factors, we used data from 1st and 3rd cuts, of the attribute tons of cane per hectare (TCH), than the network experiments state test of the improvement program of the Centro de Cana-IAC, Brazil, installed in 2002. The experimental design was a randomized block with 3 replications, with each block randomization was performed only once, so that in all experiments, the genotypes have the same neighbors. The experimental plots consisted of five lines of 8 m spaced to 1.50 m. In general, the mixed model analysis and traditional analysis showed little difference with respect to the parameters evaluated and the ranking of genotypes, as well as the use of different methodologies for the neighborhood analysis contributed little to increase the experimental precision. The effects of... (Complete abstract, click electonic access below)
Descrição
Palavras-chave
Cana-de-açúcar, Analise de covariancia, Alocompetição, Precisão experimental, Saccharum spp., Allocompetition, Covariance, experimental precision
Como citar
CANDIDO, Liliam Silvia. Modelos mistos na avaliação e ordenação de genótipos de cana-de-açúcar, com e sem efeitos de competição com parcelas vizinhas. 2009. xvi, 76 f. Tese (doutorado) - Universidade Estadual Paulista, Faculdade de Ciências Agrárias e Veterinárias, 2009.