Group Effects and Computing Strategies for Models for Estimating Breeding Values

Abstract

No criterion to define group effects in models for estimating breeding values is universally accepted. In this paper it is argued that models that assign group effects to classes of unidentified animals allow group effects and relationships between animals to be combined in a coherent manner. Such models are discussed together with computing strategies to fit them to large amounts of data. In particular, a computing strategy is presented to estimate genetic merits of bulls and cows using a best linear unbiased prediction model, which uses all relationships and all lactations.