부산대학교 도서관

With reduced budgets allocated for international agricultural research, site rationalisation had become an important issue to consider when carrying out multilocational testing of promising selections. The aim of this paper was to determine the importance of the genotype-by-environment interaction in multilocational trials of plantains and bananas (Musa spp. L.) in selected sites of West Africa comprising the humid forest and the forest-savanna transition zones. A sample of plantain-banana hybrids, plantain landraces, exotic banana cultivars and diploid parental banana accessions were evaluated in three locations : Mbalmayo and Onne (humid forest) and Ibadan (forest-savanna transition). The experimental results of our research suggested that multilocational testing is more profitable than single site evaluation over several years in the Musa breeding station. Furthermore, based on correlated responses across environments for yield potential, we suggest that one of the selection sites in the humid forest (i. e., Mbalmayo) be dropped since selections in one site (Onne) may be well adapted to the other location in the same agroecozone. Conversely, the relatively poor performance of most genotypes in dry environments (e. g. Ibadan) reinforces the importance of early testing across a wide range of environments. In this way selections with broad or specific adaptation may be identified for further release to targeted farmers.