After final models were determined, the main and interaction effects of the type of forage grazed during the breeding season were investigated as classification variables. This necessitated exclusion of 87 records (n¼579 included in analyses) because the pasture grazed during the breeding season was in Florida before shipment (n¼80), or was missing (n¼5) for the cows that had those records. Contrasts of least squares means were used to estimate genetic effects (Dickerson, 1973). Heterosis was estimated as the difference between the F1 averages and purebred average for traits: ½ [(RAþAR)(AAþRR)], where pairs of letters indicate cow breed group means: the first and second letters (first initial of each breed) represent the breed of sire and dam of cows in the group. The direct Romosinuano breed effects were estimated as the difference of the straightbred means minus the Romosinuano maternal effect: RR½ (ARRA)AA. Romosinuano maternal breed effects were estimated as the difference between the reciprocal crossbred groups ½ (ARRA). When forage type or its interaction with other main effects was determined to be significant, estimates of these genetic effects were again obtained from those with the modeled effect of forage included, in addition to the final model without the modeled effect. Cal