Estudo de características morfométricas em equinos Mangalarga Marchador por meio de modelo animal e componentes principais

Abstract

In this thesis, two studies were developed to evaluate the genetic evolution for morphometric traits, to determine which of these measures most explain the phenotypic variability and to verify how the use of these can affect the expected response to selection in Mangalarga Marchador equines. Data from 38,341 animals, born between 1987 and 2012, and a relationship matrix with 77,719 animals were used. In the first work, for the estimation of genetic parameters and values, 12 morphometric traits were evaluated, being height at withers (AC); height at croup (AG); head length (CCA); body length (CCO); shoulder length (CE); croup length (CG); neck length (CP); back and loin length (DL); head width (LC); croup width (LG); perimeter of cannon bone (PC); perimeter of thorax (PT). Genetic trends were estimated for each trait where the genetic value was regressed according to the year of birth. The heritabilities were high and ranged from 0.48 to 0.65. The additive coefficients of variation were low and ranged from 1.61% to 5.23%. The genetic trends of 11 of the 12 traits evaluated were significant, evidencing small gains for the traits during the evaluated period. The heritabilities for the morphometric traits for the Mangalarga Marchador breed were high, in contrast the variability and the genetic tendencies were low, indicating that the selection can be practiced for these traits, but the effective gains will be low. Since morphometric data are widely used in the phenotypic selection of equines, it is important to genetically evaluate these traits, which indicate the possibility of obtaining genetic gains by selecting them. In the second study the phenotypic traits were submitted to the analysis of main components and later the elimination of the variables with eigenvalues below 0.70, the genetic evaluation was performed by means of a multi-character model with six traits that were maintained due to their greater capacity to explain the variance being AG, CG, CP, DL, LC and LG. A second multivariate analysis in which the 12 traits were grouped two by two and analyzed through the 7

model was performed in order to validate the results of the multivariate analysis in selecting variables that, in fact, should be used as selection criteria in program designs improvement of equine Mangalarga Marchador. Through the multivariate analysis it was verified that six eigenvalues were less than 0.70, indicating that this is the number of variables to be discarded in the model. After excluding the variables, the traits that remained in the model were able to explain 80.04% of the total variability observed. The averages a posteriori of the heritabilities obtained through the multi-character model with six traits were equal to AG (0.48), CG (0.32), CP (0.29), DL (0.38), LC (0, 36) and LG (0.30). Genetic correlations estimated by this same model ranged from low to moderate and, in general, residual and phenotypic correlations ranged from low to moderate. The heritabilities obtained through the multi character model with two traits were low to moderate. Genetic correlations were high between AC and AG, between CCO and AC and between CCO and AG. It was verified that, through the analysis of main components, it is possible to select the most representative morphometric measurements for the Mangalarga Marchador breed. In this way the genetic evaluation of the animals does not need to be performed with all the traits and making the selection more efficient.