Résumé
Since its introduction in Europe five centuries ago, maize spread in Europe and numerous landraces have been cultivated. During the second half of the XXth century, large collections have been established to preserve this genetic diversity. The objectives of this paper are (i) to review recent results on the genetic structuration and the origin of European maize, (ii) to present the constitution of the representative core-collection of European maize landraces built in RESGEN CT96-088 project, and (iii) to study the methodology of use of these landraces in present breeding programs. Based on molecular markers, five studies found a high allelic richness in landraces from Mediterranean regions such as Spain, and (for two of them) a strong similarity between several populations from Southern Spain and a group of Caribbean populations. These studies also attest the originality of Northern Eastern Europe landraces, for which a similarity is observed with American Northern Flint landraces. Historical investigations confirm the hypothesis of introductions of maize from this origin in the North of Europe, only a few decades after introduction of tropical maize in Southern Spain by Colombus. Starting from a total of 2899 European landraces, we established with the Mstrat software a representative core collection of 96 maize accessions that maximizes allelic richness at molecular markers and best represents variation at phenotypic traits. This collection is characterized for traits of agronomical interest such as silage quality and insect tolerance. Regarding the transfer of relevant traits to elite material, comparison of F[2] versus backcross foundation populations showed that this last strategy leads to a higher population mean while not leading to a decrease in variance, therefore backcross method appears superior. Preliminary selection of superior material within a landrace did not increase average expected genetic gain but increased stability in variable environments. Molecular markers should prove helpful to extend this back-cross approach to the targeted transfer of donor interesting genomic regions.