ANÁLISE DE COMPONENTES PRINCIPAIS PARA ESTUDO DA ASSOCIAÇÃO ENTRE CARACTERÍSTICAS MORFOLÓGICAS EM PERDIZES

Abstract

The work was carried out with the objective of studying the genetic relationships between morphometric and behavioral traits measured in partridges, using principal components analysis and hierarchical clustering. The following traits were considered for the study: tarsus length, beak length, wing length, head width, beak width, adult weight, weight at 90 days and tonic immobility. After data consistency, genetic parameters and breeding values (EBV) were estimated for all the characteristics under study. Using the EBVs of all animals, the principal components analysis will be performed to get a smaller number of orthogonal (independent) variables. In the final step, the selected principal components will be used to group the animals into clusters. Obtaining four clusters that were significant, with the exception of nozzle length in cluster two and tarsus length in cluster three, compared to cluster one and four were antagonistic to each other, each cluster had a higher weight characteristic within the group, and weight had influence in all groups.In this study, we verified the selection of principal components, representing 82.65% of the total variance. Using the first two principal components, the formation of 4 groups is verified. Approximately 11.61, 58.29, 19.64 and 10.46% of the animals were classified in Cluster 1, 2, 3 and 4, respectively. By the results obtained we can conclude that the technique of main components was efficient for the reduction of the dimensionality, originating two interesting components for the use in the technique of grouping and; the growth characteristics are more important in the definition of grouping of individuals, defining their importance as a selection criterion.