TYPE CLASSIFICATION IN ZEBU COWS - GENERAL RESULTS

With the purpose of evaluating the results of the type classification in Zebu cows the results of 16192 evaluations between January 1982 and June 1995 were studied. At the same time 1523 linear classification r ecords were analyzed to increase the objectivity of this evaluation. I n this new procedure every part of the animal was examined, data are e xpressed in a score from 1 (for the poorest or undesirable) to 6 ( for the best or excellent). Twenty seven characteristics component of the 5 traditional classes are evaluated. Different linear models were app lied to each group of data and it was defined that the herd-year-seaso n of classification was the most important factor that affected all ty pe traits. On the other hand, age as well as calving-classification in terval (CCI) were also of great importance. Coefficients of determinat ion were between 26.9 and 31.1%. Adjustment factors were developed for CCI and are shown for final classification and general appraisal (GA) , while for body capacity (BC), rump (Rp), legs and hoofs (LH) and bee fing capacity (BfC) are alsb calculated. An analysis of principal comp onents allowed to reduce the original 34 dependent variables to 5 prin cipal components (PCi...PC5). They accounted for 76.6% of the original variance. The use of PC1 allowed to compare animals for GA; BC, Rp, L H and BfC, whereas PQ accounted for the changes in form between animal s with high LH and BfC which were smaller, with less BC and less BfC. The vectors of PC1 and PC2 showed a greater effect on the regression a nalysis suggesting their use in a type index such as: [GRAPHICS] where ai represents the vectors in PCi for i = 1...p traits and Zi the stan dardized results for each trait in each individual. The APC can analys e a great number of data with complex (co) variance structures.