Multivariate techniques for analyzing morphological characteristics of cactus pear genotypes in semiarid Brazil

Abstract

The objective of this study was to identify variables with better discriminatory power between treatments and reduce the dimensions of the original data set. Multivariate principal component analysis (PCA) and dendrogram clustering techniques were used on a data set of morphological characteristics of cactus pear genotypes. The palm genotypes of the genera Nopalea and Opuntia were arranged in a completely randomized design, with 30 treatments and six replications. After 180 days of cultivation, morphological characteristics were evaluated, including plant height (PH) and width (PW), total number of plant cladodes (NCP), length (CL), width (CW), perimeter (CP), cladode thickness (CT), cladode area (CA), active photosynthetic area (APA) and cladode area index (CAI). Varimax rotation with Kaiser normalization was used to prevent variables with greater explanatory capacity from affecting the rotation. Measurements were obtained as part of cluster model generation, using a full set of morphological variables for subsequent analysis of principal component and cluster analyses. The main components obtained were efficient in reducing the total variation accumulated in 30 original variables correlated to two linear combinations, which explained 81.04% of the total variation present in the original variables. Grouping using the Ward hierarchical method formed 3 groups, namely, group I consisted of genotypes F27, F48, G1, G14, G15, G20, G3, PC2, PCD, PM, T72, V07 and V20; group II was composed of the genotypes BA, F16, F34, G4, G5, OEA, OEM, POEA, T35, T80 and V19 and; group III was composed of genotypes F24, PC1, PC3, PC4, PC5 and V17. PCA demonstrated that APA, CAI, CA, NCP, PH, PW and CT are morphological variables that determine the best palm characteristics. The association of cactus morphological variables varied according to the genotypes, within each genus Nopalea and Opuntia. The genotypes with the best morphological characteristics were F27, F48, G1, G14, G15, G20, G3, PC2, PCD, PM, T72, V07 and V20.