iven a non-observable random vector u of breeding values jointly distributed with a vector y of observations, the best linear unbiased predictor (BLUP) of u can be obtained provided the (co)variance structure underlying the observations is known (Henderson, 1973). Often, in practical applications these components of (co)variance are unknown and estimates are substituted for the true parameters in order that the analysis proceed.

%B Proceedings of the World Congress on Genetics applied to Livestock Production. %V 7. Symposia (1) %P 134–138 %G eng