Estimating genetic parameters using an animal model with imaginary effects

Abstract

Summary - The estimation of additive genetic variance by maximum likelihood is discussed. The extension of reduced animal models, when parents are not inbred to allow the use of existing algorithms for maximum likelihood estimation of additive genetic variance, is described. This involves the introduction of imaginary effects with negative variance, and leads to computation using complex arithmetic. Methods are developed to allow the computation to be carried out using only real arithmetic. This method has computational advantages when only a small proportion of animals have offspring. genetic parameter / animal model / estimation / maximum likelihood