Procedimentos multivariados no agrupamento de genótipos de maracujazeiro com base em matriz de distância conjunta e em separado para características quantitativas e categóricas.

Abstract

The identification of genotypes with important characteristics for genetic breeding programs is fundamental to the development of superior materials. However, in interspecific breeding, the knowledge of the genetic variability present in base populations aiming optimizing genetic gains is key. The objective of the present work was to characterize 24 accessions of the Passion-fruit Germplasm Bank at Embrapa Mandioca e Fruticultura Tropical, by cluster analysis using the Mahalanobis distance for quantitative data, the Cole-Rogers distance for the multicategoric data and Jaccard´s distance for the molecular data and combined analysis of the combined data using Gower´s algorithm. Among the separate analysis, the molecular data presented greater reliability for the grouping of the genotypes according to the cophenetic correlation value. The combined analysis of the data using Gower´s algorithm demonstrated to be efficient since it expressed the relationship of influence existing between the types of characteristics evaluated in clustering. The quantitative variables which most contributed to the dissimilarity between accessions were: fruit weight, peel weight and weight of pulp without seed.