Please use this identifier to cite or link to this item: http://repositorio.ufla.br/jspui/handle/1/12639
Title: Avaliação visual em braquiária: número de avaliadores e modelos de análise
Other Titles: Visual evaluation in brachiaria: number of evaluators and models analysis
Authors: Nunes, José Airton Rodrigues
Souza Sobrinho, Fausto
Silva Filho, João Luiz da
Bueno Filho, Júlio Silvio de Souza
Keywords: Brachiaria - Avaliação visual
Acurácia seletiva
Forrageiras - Melhoramento genético
Modelo linear misto generalizado
Transformação Box-Cox
Inferência bayesiana
Brachiaria - Visual assessment
Selective accuracy
Forage - Genetic improvement
Generalized linear mixed model
Box-Cox transformation
Bayesian Inference
Issue Date: 3-Apr-2017
Publisher: Universidade Federal de Lavras
Citation: FONSECA, J. M. O. F. Avaliação visual em braquiária: número de avaliadores e modelos de análise. 2017. 83 p. Dissertação (Mestrado em Genética e Melhoramento de Plantas)-Universidade Federal de Lavras, Lavras, 2017.
Abstract: Urochloa ruziziensis (R. Germ. & CM Evrard) Crins. (sin. Brachiaria ruziziensis) breeding programs have adopted visual sensitivity as an evaluation parameter of genotypes. Due to the lack of knowledge regarding the adequate number of evaluators in the field and the way of performing the analysis of grades attributed to genotypes, the objective of this study was to determine an optimized number of evaluators and present different alternatives for analyzing grades. In order to do that, 2204 U. ruziziensis genotypes were evaluated in augmented block design using two controls: Marandu (U. brizantha) and Basilisk (U. decumbens). Green mass production was measured and six evaluator attributed vigor grades based on visual inspection. To determine the number of evaluators, data were analyzed using mixed models approach with recovery of interblock and intergenotypic information considering grades given by one evaluator and by the average of 2, 3, 4, 5 and 6 evaluator combinations. In order to present the different alternatives for grade analysis, it was used approaches based on linear mixed models in original score scale, linear mixed models in BoxCox transformed scale, generalized linear mixed models and Bayesian generalized linear mixed models. To validate the analysis, both cases considered genetic and experimental parameters used in breeding programs assisted by visual evaluation. Based on the results, it was possible to conclude that indirect selection of U. ruziziensis genotypes using grades is valid and that the optimal number of evaluators to carry out visual evaluations is three. Moreover, it was not possible to detect significant differences between the alternatives grade analysis, but it was possible to observe the adequacy offered by the usage of models that consider original grade scale.
URI: http://repositorio.ufla.br/jspui/handle/1/12639
Appears in Collections:Genética e Melhoramento de Plantas - Mestrado (Dissertações)



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.