ESTIMATING VARIANCE-COMPONENTS FROM FULLSIB GROUP MEANS

A method is presented to estimate variance components in cases where no individual records but only fullsib group averages are available. A model is assumed which accounts for additive genetic and dominance relationships and non-genetic fullsib group effects. A Minimum Variance Quadratic Unbiased Estimation algorithm (LAMOTTE, 1971) is proposed and general rules are given to set up the dispersion matrices for fullsib group averages. Estimation from individual records and from fullsib group means is compared in a small application on fish data, which demonstrates that the loss of information due to averaging is almost negligible. Therefore, the method may also be used in some cases to reduce computational requirements when fullsib data are analyzed.