Multivariate analysis applied to phenotypic characterization of goats

Phenotypic characterization has been used in the creation of profiles in order to guide specific actions for each of them. The profiles are defined according to the traits under investigation, previously selected according to the problem. The aim of this study was to characterize 9 breeds of goats to evaluate the phenotypic diversity by the following morphometric measurements: height at withers, croup and breast, body length, thoracic circumference and length of the ear. A total of 308 goats of the Gra?na, Anglo Nubian, Alpine Brown, Canind?, Moxot?, Marota, Blue, Repartida and Boer breeds from a farm located in Elesb?o Veloso, Piau?, Brazil, were used. The data were submitted to principal component analysis (PCA) and cluster analysis (CA). The PCA was determined based on the correlation matrix and for CA the Euclidean distance was adopted as a measure of similarity between the elements by the Unweighted Pair Group Method with Arithmetic Mean (UPGMA). The R software version 3.2.1 was used. For PCA, from the 6 principal components, 4 of them had eigenvalue lower than 0.7. The first two principal components were selected explaining more than 71.7% of the total variation. In PCA, Boer goats (exotic breed) were positioned in the upper left quadrant, the Anglo Nubian and Alpine Brown breeds (both with milk aptitude) were in the lower left quadrant. Repartida, Gra?na and Marota were located in the upper right quadrant, and the Blue, Moxot? and Canind? breeds in the right lower quadrant. In CA, a dendrogram was generated where the bootstrap values above 50% expressed confidence in the cluster. There was the formation of four groups, the first consisting of the Canind?, Repartida, Gra?na and Marota breeds (94.0%), the second with the Blue and Moxot? breeds (96.0%), the third isolating the Boer breed (82, 0%), and the fourth with the Anglo Nubian and Alpine Brown breeds (57.0%). The PCA has confirmed the result of UPGMA at isolating the Boer breed and clustering the Anglo Nubian and Alpine Brown breeds. The UPGMA was more close to the biological traits, with a satisfactory result due to groups with a high internal homogeneity. The methods successfully portrayed the diversity among the breeds evaluated in this study.