ABSTRACT
Mixed model analysis of canola breeding MET datasets in Beeck et al. (2010), based on factor analytic (FA) models introduced by Smith et al. (2001) and developed by Oakey et al. (2006, 2007) for plant breeding MET data, included terms for both genetic and non-genetic effects. Non-genetic effects included fi xed effects for different environments, random terms accounting for trial design, and additional fi xed and random terms for modeling spatial heterogeneity within trials. MET/FA analysis of canola breeding trials across southern Australia was divided into two phases (Beeck et al. 2010). In phase 1, signifi cant spatial trends (both fi xed and random) were identifi ed for seed oil and grain yield within trials in the MET analysis. In phase 2, genetic variance across sites was modeled. A series of FA models were fi tted to the data for both seed oil and grain yield, and for both traits the addition of pedigree information substantially improved fi t compared with the corresponding model excluding pedigree information (Beeck et al. 2010). The average proportion of additive variance at sites for grain yield was 0.561, compared to an average of 0.904 for seed oil (Beeck et al. 2010). This refl ects the higher heritability of seed oil compared with grain yield.
