Comparison of Different Statistic Models in the Genetic Parameters Estimation, on Guava (psidium Guajava, Myrtaceae) Full Sibling Families

Guava, is a tropical fruit tree with great nutritional values, rich in vitamins and mineral salts. In perennial plants, repeated measurements during several crops in the same plants are made, therefore the data obtained from these experiments do not satisfy the assumptions of the general linear models. Generalized linear mixed models allow data to be analyzed with repeated measurements over time and extend its use to cases with predictors of distributions belonging to the exponential family. For its correct application, is necessary to carry out a previous study of characters and select model specifications. The objective of the present work is to compare the precision of different statistical models in the estimation of genetic parameters, in guava full sibling families. Ten quantitative traits were evaluated during three years, in three guava full sibling families in Cuba. Factorials analyses of variance (Model I and Model II) were made, using the GLM procedure. After that, GLIMMIX procedures were used, in which genotype and genotype-year interaction as random factors, year as repeated measure and covariance structure were specified. The estimation of statistical genetic parameters, with the restricted maximum likelihood and least squares methods, allowed to detect a high genetic variability in the three families, as well as a significant genotype-environment interaction in the evaluated traits. The estimates of heritability in the broad sense were obtained by both methods, very similar and showed mostly mean values. However, when they were obtained by the restricted maximum likelihood method they were more accurate. Citation: Bandera, E., Pérez, L., Valdés-Infante, J. & Velázquez, J.B. 2021. Comparación de diferentes modelos estadísticos en la estimación de parámetros genéticos, en familias de hermanos completos de guayabo (Psidium guajava, Myrtaceae). Revista Jard. Bot. Nac. Univ. Habana 42: 55-68. Received: 18 September 2020. Accepted: 19 December 2020. Online: 8 April 2021. Editor: José Angel García-Beltrán.