Résumé
Banana and plantain production is severely threatened by several pests and diseases. Furthermore, wind damage severely affects yield. Apart from the Cavendish bananas produced for exportation by multinational companies, the other dessert bananas and most plantains and cooking bananas are mostly cultivated by small-scale farmers for home consumption and for sale in local and regional markets. Since these farmers cannot afford pesticides, resistant varieties appear as an appropriate and environmentally friendly solution for sustainably improving yield. Cultivated bananas are mostly triploid and parthenocarpic. Being highly sterile, they are difficult to breed and little knowledge on Musa genetics and cytogenetics is available. Despite these constraints, progress has been made in conventional genetic improvement of Musa in recent years and some new varieties derived from popular regional cultivars are now becoming available from major conventional breeding programmes. Based on the large number of accessions collected in the primary and secondary zones of banana diversity, the initial breeding approach was focused on triploid x diploid crosses (3x/2x) aimed at producing disease-resistant tetraploid hybrids from partially fertile triploid cultivars. Using resistant diploid clones as male parents and exploiting non-reduced 3x female gametes produced by partially fertile female triploid cultivars, has produced, through embryo rescue, the production of several disease-resistant tetraploid hybrids bearing traits from the mother parent. Improved diploid parents from extensive 2x/2x crosses have also been used to optimize the production of superior hybrids. Moreover, this 3x/2x approach led to the creation of diploid hybrids that are used to develop elite diploids, among other breeding strategies. Previous efforts, namely at the former Jamaican banana breeding programme and later at FHIA, have produced improved dessert-type diploid parents with interesting resistances and agronomic performance. More recently, cooking and plantain-type diploids have been developed and are now available from CARBAP, IITA and EMBRAPA. Based on these improved diploids, FHIA, CARBAP, EMBRAPA and IITA are focusing their strategies on developing triploids by combining them with primary or secondary tetraploids. In vitro tissue culture and mutagenesis have allowed the emergence of another 4x/2x strategy, aiming at re-creating triploid hybrids from ancestral diploid material. Extensively used by CIRAD to develop triploid dessert banana, this strategy relies on a very good knowledge of banana genetic resources and of phylogenetic links between ancestral and present varieties. The induced tetraploid parent is obtained through colchicine treatment of a diploid one. Using a higher range of fertile genitors, this strategy has led to the creation of large hybrid populations of AAA and AAB triploids. This strategy could benefit from widening of the genetic base of the crosses and from marker-assisted selection. Both these strategies are however limited by the problem of BSV activable sequences in the B genome, which prevents their use in breeding. Some major breeding programmes are targeting the creation of high-yielding small stature or dwarf/early hybrids and the first interesting products are already produced. This overview indicates that much is possible with conventional breeding, although not everything. Conventional breeding is just one part of an integrated improvement banana scheme. Recent developments in breeding strategies and new hybrids from popular varieties will be presented, and approaches to reinforce and enhance the promising results of conventional breeding will be discussed. (Texte intégral)