Meta-análise das estimativas de parâmetros em genótipos de arroz irrigado em Minas Gerais
Arquivos
Data
2018-07-31
Autores
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Federal de Viçosa
Resumo
A experimentação agrícola realizada para determinada cultura proporciona um banco de informações que precisam ser analisadas em conjunto para que se possa entender a amplitude das informações e a sua dinâmica. Uma maneira eficaz de realizar análise acurada dessa informação acumulada é por meio da meta-análise. Para isso, três modalidades de meta-análise podem ser realizadas: (a) Análise descritiva das estimativas; (b) Análise combinada temporal; e (c) Análise combinada ponderada. Assim, este trabalho foi realizado com o objetivo de avaliar o dinamismo do programa de melhoramento genético de arroz irrigado de Minas Gerais durante 12 safras agrícolas, a partir da meta-análise nas estimativas dos parâmetros das principais características avaliadas nos Ensaios Comparativos Preliminares (ECP’s) e Ensaios Comparativos Avançados (ECA’s ou VCU’s). O banco de dados utilizados neste trabalho foi constituído pelos ECP’s, conduzidos em Leopoldina e Lambari, e ECA’s (VCU’s), conduzidos em Nova Porteirinha, Leopoldina e Lambari, totalizando 378 linhagens e cultivares avaliadas. Realizou a análise exploratória dos dados por meio do gráfico Box-plot e verificou a presença de “outliers” para maioria dos parâmetros. Verificou que o programa vem trabalhando com nível de herdabilidade satisfatória, acima de 50%, mas que este valor tem decrescido ao longo do tempo. O coeficiente de variação para ECP está estabilizado em 16,02%, com pequena variação nos anos. Entretanto o ECA apresenta tendência de acréscimo no tempo. Na meta-análise temporal, não houve a sobreposição do intervalo de confiança entre os estudos para a variância genotípica, variância fenotípica e coeficiente de variação genético para dias para floração e coeficiente de variação genético para altura de plantas, assim não foi possível obter os resultados combinados para o modelo aleatório desses parâmetros. Para todos os parâmetros das três características, o modelo de regressão preferível para predição das estimativas dos parâmetros ao longo dos anos foi o modelo de regressão de efeito fixo. Foram verificadas as pressuposições de independência, normalidade e homogeneidade, requeridas para realização da meta-análise ponderada. A suposição de independência foi satisfeita devido às estimativas dos parâmetros terem sido obtidas de experimentos distintos. Verificou que todos os parâmetros avaliados seguem uma distribuição normal, com exceção da herdabilidade para altura de plantas, por isso não foi realizada a meta- análise ponderada para esse parâmetro. Na meta-análise ponderada, a partir do teste de homogeneidade, verificou que o coeficiente de variação foi o único parâmetro que adotou o modelo aleatório, enquanto que a média, variância fenotípica e variância genotípica assumiu o modelo fixo. Para o coeficiente de variação genotípico e variância ambiental, de produtividade de grãos, e herdabilidade, para dias para floração, foi adotado o modelo aleatório. A meta- análise, por considerar o efeito de estudo, mostra ser uma metodologia adequada para escolha do modelo de regressão, a fim de obter equações para predição das estimativas dos parâmetros ao longo do tempo. Além do mais, a meta-análise assegura à obtenção da estimativa combinada com um menor desvio padrão e maior precisão, para assim avaliar da forma correta o dinamismo em programas de melhoramento. Desta forma, o programa de melhoramento genético de arroz irrigado demonstrou ser dinâmico no período avaliado.
Agricultural experimentation conducted for a particular crop provides a database of information that needs to be analyzed together in order to understand the breadth of information and its dynamics. An effective way of performing accurate analysis of this accumulated information is through meta-analysis. For this, three modalities of meta- analysis can be performed: (a) Descriptive analysis of the estimates; (b) Combined temporal analysis; and (c) weighted combined analysis. Thus, this work was carried out to evaluate the dynamism of the genetic improvement program of irrigated rice of Minas Gerais during 12 agricultural crops, from the meta-analysis in the estimations of the parameters of the main characteristics evaluated in the Preliminary Comparative Trials (ECPs) and Advanced Comparative Trials (ECA's or VCU's). The database used in this work was ECP's, conducted in Leopoldina and Lambari, and ECA's (VCU's), conducted in Nova Porteirinha, Leopoldina and Lambari, totaling 378 lines and cultivars evaluated. He performed the exploratory analysis of the data through the Box-plot graph and verified the presence of outliers for most parameters. She verified that the program has been working with a satisfactory heritability level, above 50%, but that this value has decreased over time. The coefficient of variation for ECP is stabilized at 16.02%, with little variation in the years. However, ECA shows an upward trend in time. In the temporal meta-analysis, there was no overlap of the confidence interval between the studies for genotypic variance, phenotypic variance and coefficient of genetic variation for days for flowering and coefficient of genetic variation for plant height, so it was not possible to obtain the results combined for the random model of these parameters. For all parameters of the three characteristics, the preferred regression model for predicting the parameter estimates over the years was the fixed-effect regression model. The assumptions of independence, normality and homogeneity required to perform the weighted meta- analysis were verified. The assumption of independence was satisfied because the estimates of the parameters were obtained from different experiments. She verified that all evaluated parameters follow a normal distribution, except heritability for plant height, so the weighted meta-analysis for this parameter was not performed. In the weighted meta-analysis, from the homogeneity test, we verified that the coefficient of variation was the only parameter that adopted the random model, while the mean, variance phenotype and variance genotype assumed the fixed model. For the genotype variation coefficient and environmental variance, grain yield, and heritability, for days for flowering, the random model was adopted. The meta- analysis, considering the study effect, shows to be an adequate methodology for choosing the regression model, in order to obtain equations to predict the parameter estimates over time. Moreover, the meta-analysis ensures that the estimation is obtained in combination with a lower standard deviation and greater precision, in order to properly evaluate the dynamism in breeding programs. In this way, the irrigated rice genetic improvement program proved to be dynamic during the period evaluated.
Agricultural experimentation conducted for a particular crop provides a database of information that needs to be analyzed together in order to understand the breadth of information and its dynamics. An effective way of performing accurate analysis of this accumulated information is through meta-analysis. For this, three modalities of meta- analysis can be performed: (a) Descriptive analysis of the estimates; (b) Combined temporal analysis; and (c) weighted combined analysis. Thus, this work was carried out to evaluate the dynamism of the genetic improvement program of irrigated rice of Minas Gerais during 12 agricultural crops, from the meta-analysis in the estimations of the parameters of the main characteristics evaluated in the Preliminary Comparative Trials (ECPs) and Advanced Comparative Trials (ECA's or VCU's). The database used in this work was ECP's, conducted in Leopoldina and Lambari, and ECA's (VCU's), conducted in Nova Porteirinha, Leopoldina and Lambari, totaling 378 lines and cultivars evaluated. He performed the exploratory analysis of the data through the Box-plot graph and verified the presence of outliers for most parameters. She verified that the program has been working with a satisfactory heritability level, above 50%, but that this value has decreased over time. The coefficient of variation for ECP is stabilized at 16.02%, with little variation in the years. However, ECA shows an upward trend in time. In the temporal meta-analysis, there was no overlap of the confidence interval between the studies for genotypic variance, phenotypic variance and coefficient of genetic variation for days for flowering and coefficient of genetic variation for plant height, so it was not possible to obtain the results combined for the random model of these parameters. For all parameters of the three characteristics, the preferred regression model for predicting the parameter estimates over the years was the fixed-effect regression model. The assumptions of independence, normality and homogeneity required to perform the weighted meta- analysis were verified. The assumption of independence was satisfied because the estimates of the parameters were obtained from different experiments. She verified that all evaluated parameters follow a normal distribution, except heritability for plant height, so the weighted meta-analysis for this parameter was not performed. In the weighted meta-analysis, from the homogeneity test, we verified that the coefficient of variation was the only parameter that adopted the random model, while the mean, variance phenotype and variance genotype assumed the fixed model. For the genotype variation coefficient and environmental variance, grain yield, and heritability, for days for flowering, the random model was adopted. The meta- analysis, considering the study effect, shows to be an adequate methodology for choosing the regression model, in order to obtain equations to predict the parameter estimates over time. Moreover, the meta-analysis ensures that the estimation is obtained in combination with a lower standard deviation and greater precision, in order to properly evaluate the dynamism in breeding programs. In this way, the irrigated rice genetic improvement program proved to be dynamic during the period evaluated.
Descrição
Palavras-chave
Arroz - Melhoramento genético, Estimativas de parâmetros
Citação
COSTA, Weverton Gomes da. Meta-análise das estimativas de parâmetros em genótipos de arroz irrigado em Minas Gerais. 2018. 97 f. Dissertação (Mestrado em Genética e Melhoramento) - Universidade Federal de Viçosa, Viçosa. 2018.
