Résumé
Molecular markers have become an essential tool for genetic analysis. Marker-aided selection (MAS) has demonstrated its efficiency for introgression of major genes. However, most agronomically important traits like tolerance or resistance to biotic and abiotic constraints are quantitative. Quantitative traits loci (QTLs) can be mapped on the genome using statistical procedures and many mapping results are now available. Few results, however, are available for utilisation of this information for introgression of QTLs. Two separate studies were conducted with the aim of introgressing QTLs using markeraided selection, evaluating the selected progenies and assessing the efficiency of marker-aided selection. In both cases, the population used for QTL detection was the reference DH population derived from the cross between IR64 and Azucena. The interesting alleles were coming from the Azucena parent and IR64 was used as the recurrent parent in three cycles of marker-assisted backcrosses. Several QTLs were manipulated at once. The first study, conducted in the Philippines, was directed toward the introgression of alleles for root depth, in order to improve water-use efficiency. The second study conducted in Mali and in France, was directed toward the improvement of partial resistance to rice yellow mottle virus, an increasingly important disease in Africa. The genotype of the BC3F2 progenies was determined to assess the presence of Azucena alleles at the QTLs and the return to the recurrent parent. The phenotypic evaluation of the BC3F3 families showed that the introgressed QTLs were expressed in the new genetic background. It showed also the importance of initial QTL analysis methods. The comparison of our experimental process of MAS with the recommendations of analytic and simulation studies pointed out the methodological flexibility of MAS for QTLs. These were to our knowledge the first examples of markeraided selection for QTLs in rice. This was done on a model cross following QTL detection. For future work, advanced backcross methods may be a way to perform at the same time QTL detection and introgression in an agronomically interesting background. (Résumé d'auteur)