This thesis has examined aspects of detection and utilisation of genetic marker technology in animal breeding systems. Models have been proposed to include genetic marker information into mixed model equations for estimating major gene, or Quantitative Trait Loci (QTL) effects. These involve building a Gametic Relationship Matrix (GRM) with probabilities of identity by descent of QTL alleles between gametes. Algorithms to build the GRM were compared and a further method was proposed to build the GRM based on simulated QTL information. The simulated QTL method was identified as an approximation of the true GRM, as it involved simulating all possible QTL-marker genotype combinations. Detection of linkage between genetic markers and QTL is usually based on data collected from animals in the field which are assumed to be unselected. The effect of this assumption on parameter estimates is unknown. Results from use of a restricted maximum likelihood detection method and a granddaughter design population, show that in the presence of within family (or sire) selection, large underestimates of variance occur and QTL position and effect estimates are inaccurate. However, selection of grandsires only has little effect on estimation of variance. |
|