Genetic parameters of liveweight of Zebu males in performance test under Cuban grazing conditions, estimated by univariate, multi-trait and random regression models

The individual results of the liveweight (LWi and i = 7, 8 ... 18 months of age) from 1091 young purebred Zebu males (13092 estimations of LWi) were studied by univariate (UAM), multi-trait (MAM) and random regression model (RRM), with the objective of estimating the genetic (co)var throughout the age trajectory. The animals were progenies of 107 sires and 893 mothers and were subject to a performance test in Cuban grazing conditions for 12 months. The genetic variances manifested a low homogeneous pattern, with high and low estimates, when applying the UAM and the MAM. On the contrary, the results of the RRM showed a harmonious trend throughout the time span. The values of heritability (h(2)) were very similar in the three procedures, although higher coherence was manifested in the RRM. In general, there is slight decrease in h(2) as age advances, with average values of h(2) = 0.27 for LW7 and h(2) = 0.23 for LW18. The genetic correlations were R-g > 0.9 between LWi at adjacent ages and decreased as the age difference increases between the estimations. The estimates at LW13 or more manifested R-g > 0.8 with LW18, suggesting the possibility of an early selection of the animals. A polynomial of order 5 for the matrix of the random regression coefficients (K-a) and of order 6 for the permanent environment effects due to the repetitions of the dependent variable in each animal was selected as the best RRM. The 71.8 % and 24.1 % of the variation was explained by the first two eigenvalues of the matrix K-a, suggesting possible changes in the form of the growth curve through selection and breeding. This possibility was noted when selecting the best 2 % of the animals under study according to their breeding value for LW18. In this group of elite sires, two subsets of animals were identified, one of stable performance throughout the study, while the other manifested a totally different and dynamic form of growth curve. The study proves the possibilities of use of the random regression models.