Generation Means Analysis Using Mixed Models

This paper shows how to perform generation means analysis, including estimation of genetic variance components, by a mixed model package. The approach is exemplified with a small dataset, which is analyzed by the MIXED procedure of the SAS System and by ASReml. The proposed analysis accounts for between-plot variation, which is often ignored. Also, heterogeneity of standard errors of mean contrasts due to unequal sample size per plot and nonconstant genetic variance can be properly reflected.