I am finding association of 3 genotypes with one of trait using linear mixed model using animal as random effect. Which test is meaningful to test differences between genotypes.
The attached paper may give you a more modern view of such methods. These are all available in the bioconductor package..see the first page for the computer requirements.Best wishes David Booth
Use Tukey's procedure. It controls the FWER in a strong sense across all possible pair-wise comparisons. Duncan' procideúre is indirect. They don't give the same results.