Stratégies raisonnées d'échantillonnage pour capturer la diversité génétique et sa structuration dans les populations naturelles. Application aux mesures de gestion conservatoire

Both environmental and evolutionary factors determine the distribution of the genetic diversity among and within natural populations of forest tree species, resulting in genetically structured metapopulations. The existence of such distribution has major implications for resource conservation policies, as the shape and extent of in situ natural networks and the composition of ex situ collections of valuable forest tree species. A major and realistic issue we face is how to effectively sample natural variation both for optimal structure mirroring and for maximal genetic diversity capturing when designing in situ population networks and ex situ collections. Genetic differentiation descriptors like Fst and Qst represent the utmost visible consequences of underlying landscape and evolution factors acting on the metapopulation genetic diversity. The first question we address here is how to sample populations, individuals and marker loci knowing the neutral and selected genetic structure of a theoretically built metapopulation. Results of a Monte-Carlo allele-based simulation model are presented, where the sample of populations, individuals and marker loci was simultaneously modelled over ranges of Fst and Qst. Each sample strategy was not only evaluated in terms of genetic diversity potential yield relative to total diversity at the metapopulation level, but also in terms of bias and precision rendering in the estimation of Fst and Qst from the actual sample. Thus the ability of the sample to render both the diversity and its distribution over the populations was analyzed. The second question we address is whether present ex situ collections of two disseminated forest tree species, black poplar and wild cherry, represent a good sampling of natural populations. For that purpose, natural populations and ex situ collections were characterised with neutral markers and phenotypic characters. The simulations then serve to infer the degree of representativeness of existing ex situ collections for the two species based on the actual sample sizes and several differentiation estimates of Fst and Qst, and eventually guide the enrichment of conservation resources for these two species. / La diversité génétique d'une espèce se distribue entre populations et à l'intérieur des populations. Sa prise en compte à ces deux niveaux a une importance majeure pour la mise en place de modes de gestion conservatoires in situ et ex situ. La question principale qui se pose est : quel échantillonnage de la diversité faire, en terme de populations, d'individus et de gènes ? Nous abordons ici cette question, d'un point de vue théorique à travers le développement d'un modèle Monte-Carlo, puis d'un point de vue expérimental sur deux espèces forestières, le merisier et le peuplier noir, à travers une comparaison de la diversité présente en populations naturelles et en collection ex situ. Les deux approches sont ensuite mises en regard l'une de l'autre afin de donner des éléments sur les stratégies d'échantillonnage optimales de la diversité de ces deux espèces.