Considering the increase of world demand for wood, improving wood quality to better fit industrial requirements becomes a major objective. Hitherto, breeding for wood traits has been hampered by the cost of traditional assays and the need to wait until the trees are nearly mature to be evaluated. The recent development of molecular tools for genomic analyses of woody species makes it possible to identify genes controlling wood traits. The genus Eucalyptus includes the most widely used tree species for industrial plantation, therefore is one of the world's main source of biomass and the first hardwood tree used for pulp making. The objective of our project joining academic and industrial partners is to identify and map putative candidate genes (CG) and QTLs responsible for variations in wood properties, in order to implement Allele Aided Selection strategies in Eucalyptus breeding programs. With this aim in view, we have developed a functional genomics-based approach that does not require an a priori knowledge of the factors controlling these traits to search for pertinent "expressional" CG. Our hypothesis is that genes controlling wood properties would be specifically or at least preferentially expressed in differentiating xylem. To this end, Suppression Subtractive Hybridization (SSH) libraries have been constructed and cDNA clones have been spotted onto nylon membranes to make high density filters. The results of both sequence analysis and transcript profiling experiments, assist us in the selection of the most appropriate CG to be mapped on Eucalyptus genetic linkage maps, and allow us to gain a better understanding of the mechanisms involved in one of the most fascinating example of differentiation in plants, i.e. secondary xylem formation. (Texte intégral)