SIMULATION ANALYSIS WITH BLUP METHODOLOGY OF DIFFERENT DATA-STRUCTURES IN GOAT SELECTION SCHEMES IN SPAIN

Drawbacks of data structure on reliability of estimated breeding value s were studied with data from a simulated population of 6748 does (dis tributed in 100 herds), They were bred for 20 generations under differ ent degrees of connectedness among herds and under different selection schemes. Different models and levels of genealogical information were used to estimate breeding values. Results showed that connectedness a nd selection increased bias while genealogical information reduced it. Prediction error was reduced when connectedness was increased under t he same conditions relative to the number of herds being selected. In the absence of connectedness, herd effect, when included, made predict ion worse. It must be included only when all flocks are connected. It was concluded that under the present state of data structure in record ing schemes of Murciano-Granadina dairy goats in the South of Spain, h erd effect should not be included in models for animal ranking.