A comprehensive review of crossbreeding studies in farm animals yielded 3,670 estimates of heterosis, which were classified into 6 farm species, 334 individual traits, 4 character groups, 193 breeds, 7 levels of a breed's contribution to the blood composition of crossbred progeny, and 66 countries where heterosis was reported. Results of the least square analysis revealed no significant differences (P>0.05) in heterosis estimates between farm species (S) and between breed's contribution to the blood composition of crossbred progeny as a measure of heterozygosity in different crossbreeding systems. Heterosis reports in the Philippines and other countries also do not vary significantly (P>0.05).
However, heterosis values were significantly influenced (P<0.01) by the type of formula used to estimate heterosis (F) and character group (C). Heterosis based on one parent only were overestimates of heterosis based on mid parent average. Growth and yield traits exhibited substantially higher heterosis than product quality and cost reduction traits. The high heterosis values in highly heritable traits may be attributed to the "additive x additive" epistatic term in the many loci model described by Willham and Pollak (1984) or a result of improving the average genetic merit of the foundation breeds used in the cross, as suggested by Warwick and Legates (1979). Significant interaction effects (P<0.01) were also found for FxS, FxC, and SxC. Caution must therefore be made when predicting heterosis depending on the farm species, type of formula used to estimate heterosis, and character group contributing to the breeding objective.