Statistical modeling implications for coffee progenies selection

Abstract

A reliable phenotyping and a thorough investigation of the experimental data via accurate statistical methods are key requirements for attaining selection gain. Coffee bean yield data are provided from annual harvests. The data analysis is generally performed based on total phenotypic data of entire period or in biennia using a split-plot-in-time model. An essential aspect of these data is the covariance associated with some random factors of the statistical model. The aim of this work was to evaluate different covariance matrix structures in coffee progenies bean yield modeling and their implications for prediction accuracy of progenies genotypic values and selection under different harvest data grouping strategies. We evaluated 21 S0:1 Coffea arabica L. progenies during eight harvests. The analyses were conducted considering all the harvests (annual or biennia) and focusing only on the high yield or low yield years. In each case, we modeled the residual covariance matrix (R) and the genetic covariance matrix over harvests (G). We noticed that some models are more suitable in explaining the coffee yield pattern. There were alterations in parameter estimates, prediction error variance of genotypic values, rankings and coincidence index in selecting the best progenies. The model involving annual harvests gave more information regarding the coffee progenies yield behavior in comparison to biennia.