The objective of this study was to evaluate the viability of using partial records of milk yield instead of the production accumulated at 305 days of lactation to obtain the predicted breeding values (BV). Sires and cows ranked by BV using accumulated milk yield records at 90(T90), 150(T150) and 210(T210) days of lactation were compared with BV estimated by using production up to 305 days (T305). An animal model was used to estimate the animals BV including herd-year, season and age of calving, and genetic group as fixed effects and the animal, permanent environment and residual as random effects. Heritability and repeatability estimates for T90, T150, T210 and T305 were 0.21±0.03 and 0.47±0.03; 0.23±0.03 and 0.50±0.03; 0.24±0.03 and 0.50±0.03; and 0.23±0.03 and 0.49±0,03, respectively. Genetic correlations between T305 and T90, T150 and T210 were 0.92, 0.96 and 0.98, respectively. Rank correlations between BV305 and BV90, BV150 and BV210 ranged from 0.83 to 0.97 and were smaller as production periods and percentage of selected or culled animals based on their BV were reduced. Selecting the top 20% animals by BV305, 65, 76 and 87% of those animals would be selected if BV90, BV150 and BV210 were used, respectively. Culling the bottom 20% animals by BV305, only 53, 68 and 83% would be culled if BV90, BV150 or BV210 were used, respectively. The results were similar between sires and cows. Therefore, under a technical point of view, the use of partial records of milk yield, in a short period, could result in wrong decisions in the selection of the best or in the culling of the worst animals in the herd.
animal model; genetic parameter; Gyr breed; partial lactation; rank correlation; selection
