With the diallel mating design we can decompose total genetic variance into additive genetic variance (explained by the GCA effect) and dominance genetic variance (explained by the SCA effect). In the model below, the term cross defines SCA. In the data set, the levels of cross should code for a particular combination of parents without respect to which parent is male or female. In other words, the level for cross should be identical for reciprocals. SCA could also be defined as an interaction between female and male effects, but only if we are not treating male and female as pedigree-associated factors (since the pedigree information forces a specific variance-covariance matrix on the main effects that cannot be simply generalized to their interaction). Furthermore, to be sure we treat the effect of female l x male k as identical to female k x male l, we need to use the and() function again. Since there is no field defined as male.female, we would have to first define it in the model without fitting it, writing the part of the model coding for SCA as: