出版社：Grupo de Pesquisa Metodologias em Ensino e Aprendizagem em Ciências
摘要：The present work aimed to predict the genetic variability in provenances of O.bacaba do Pará through multivariate analysis applied to morphoagronomic characters. Eighteen characters were recorded, three relating to the plant, six of bunch and nine of fruit, in 43 palms in the state of Pará. Data were identified by analysis of genetic divergence between those belonging to the Euclidean distance, and the groups formed through complete binding method, and additionally performed the importance and discards of characters by principal component analysis. The greatest divergence, dii’= 11.45, was found between the representatives TS-22 from Terra Santa and BAI-8 from the municipality of Baio. The grouping by the furthest neighbor method trained on the dissimilarity matrix training to 6 divergent groups. The weight of one hundred fruits and the number of rachiles per bunch are the characters with the greatest contribution to the divergence between the provenances. In the analysis of the main components, the first four components absorbed 63.40% of the accumulated variation, the fruit and bunch characters are those related by the variation among selected ones, being possible to discard as variables clump maturation and thickness of the pulp. The individuals of O. bacaba from the two provenances of Pará studied show considerable phenotypic variation for the morpho-agronomic characters studied, mainly fruit and bunch, with those from Terra Santa being more divergent.
