Fifteen reproductive, 22 vegetative and 21 agronomic descriptors were evaluated, the agronomic ones comprised seven descriptors for annual, winter and summer evaluations, respectively, with the objective to select the most appropriate descriptors for characterization of Paspalum guenoarum and Paspalum plicatulum. Principal components analysis was used to discard redundante or non-discriminative descriptors. The principal component analysis allowed a reduction of 53, 68 and 43%, respectively, in the number of initial descriptors considered. By means of this technique, it was possible to select eight descriptors, considered the most important descriptors to identify the genetic variability in this germplasm collection.
accessions; germplasm; principal components analysis; selection descriptors
