Discriminant analysis for the classification and clustering of robusta coffee genotypes
Aymbiré Francisco Almeida da Fonseca, Tocio Sediyama, Cosme Damião Cruz, Ney Sussumu Sakiyama, Romário Gava Ferrão, Maria Amélia Gava Ferrão, and Scheilla Marina Bragança
This study evaluated the adequacy of the composition of three clonal Coffea canephora varieties recommended for the State of Espírito Santo by a multivariate method designated discriminant analysis. This method consists in the establishment of functions that enable the classification of a given individual into one, among various distinct populations, reducing the probability of a misclassification. It simultaneously considers measures of several traits, in order to give the new variety homogeneity. The original classification of genotypes in the three studied varieties, based on agronomical criteria, maintained expressive concordance with the results of the discriminant analysis, with an apparent deviation rate of only 6.25%. Corrected discriminant functions were also proposed, capable of classifying a new genotype into one of the three clonal varieties to be used in improvement programs, eliminating the subjectivity of the clustering process.