Discrimination of forage pea seed lots by means of multivariate techniques

Resumo

Multivariate techniques allow to understand the structural dependence contained in the variables, as well as to characterize groups of seed lots according to specific standards. Thus, this study analyzes the efficiency of multi­variate exploratory techniques in discriminating forage pea seed lots as a function of the physiological potential of seeds. We evaluated ten seed lots of forage pea in a completely randomized design, considering the following variables: thousand seed weight, germination, first germination count, electrical conductivity, and accelerated aging. Moreover, seedling emergence, first count of seedlings in the field, and seedling emergence speed index in the field were added to randomized blocks with four replications per lot. Initially, the data obtained in each test were analyzed separately by means of analysis of variance, and the means of the treatments were compared by the Scott Knott test at 5% probability. Exploratory multivariate statistical techniques were applied by means of Cluster Analysis and Principal Components Analysis to discriminate seed lots with better physiological quality and to characterize the variables responsible for the differentiation between them. Multivariate analysis of principal components is efficient in discriminating vigor and seed germination tests in Pisum sativum subsp. Arvense, which help in identifying lots of superior performance in the field.