RESUMEN
This study was conducted to investigate the relationship between heterosis and gene effects estimated by the generation mean analysis. Nine traits with 74 cases of combinations cross-sit and cross- abiotic or biotic stress levels were assessed in three crops (durum wheat, pepper, and oat) and evaluated by lines crosses analysis. Trait performances of the F1 hybrid showed evident mid-parent heterosis varying from 0.6% to 89% for the 74 cases investigated. Results of Generation mean analysis revealed that the additive-dominance model was demonstrated adequate in 7 cases. Therefore the epistatic model was found appropriate in 67 cases. Analysis of correlations between gene effects estimated by the generation mean analysis revealed that heterosis was not correlated to additive, dominance or epistasis effects. Therefore, the majority of geneticists considered the non-additives effects as the genetic basis of heterosis. Thus, the lower correlations obtained between heterosis and non-additives effects were due to the bias of the classical approach’s models of genetic quantitative. In fact, many assumptions were proposed to develop this model. To conclude, non-additives results are apparently of great importance in the inheritance of quantitative traits and their roles in the heterosis expression are not to discuss. However, the quantitative genetic interpretation of mid-parent heterosis as a function of genetic effects was not possible basing on the model of line crosses analyses.