Ce mémoire porte sur les relations phylogénétiques, géographiques et historiques du genre afro-malgache Delonix qui contient onze espèces et des genres monospécifiques et endémiques Colvillea et Lemuropisum. Les relations intergénériques et interspécifiques entre les espèces de ces trois genres ne sont pas résolues ce qui limite la vérification d’hypothèses taxonomiques, mais également biogéographiques concernant la dispersion de plantes depuis ou vers Madagascar. Une meilleure compréhension des relations évolutives et biogéographiques entre ces espèces menacées d’extinction permettrait une plus grande efficacité quant à leur conservation. L’objectif de ce mémoire est de reconstruire la phylogénie des espèces à l’aide de régions moléculaires des génomes chloroplastique et nucléaire, d’identifier les temps de divergences entre les espèces et de reconstruire l’aire géographique ancestrale pour chacun des groupes. Ce projet démontre que le genre Delonix n’est pas soutenu comme étant monophylétique et qu’une révision taxonomique s’impose. Les relations intergénériques demeurent floues quant à la position phylogénétique de Colvillea et nos résultats suggèrent de l’hybridation ou un assortiment incomplet de cette lignée. Les espèces sont apparues et se sont diversifiées au Miocène à partir d’un ancêtre commun du sud de Madagascar. La phylogénie montre deux clades associés aux aires géographiques de répartition des espèces opposant les espèces largement répandues à celles majoritairement restreintes au fourré aride. Différentes hypothèses afin d’expliquer la dispersion des Delonix africains au Miocène à partir de Madagascar sont discutées. Un point de mire sur les interactions biotiques et abiotiques, passées et présentes, dans le fourré aride de Madagascar est recommandé en terme de conservation. / This thesis discusses the phylogenetic, geographic, and historic relationships of the eleven species of the afro-madagascan genus Delonix, as well as the endemic, monospecific genera Colvillea and Lemuropisum. The intergeneric and interspecific relationships of these taxa are unresolved which limits the evaluation of taxonomic changes suggested by some studies and of conflicting biogeographical hypotheses regarding plant dispersal around Madagascar. These species are all endangered, therefore a better understanding of their evolutionary pattern would benefit conservation efforts in Madagascar. The objective of this thesis is to reconstruct a phylogeny using multiple loci from two distinct genomes, and to use it to reconstruct the ancestral geographic range of species and estimate the times of divergence of the lineages. Results indicate that genus Delonix is not supported as monophyletic and a taxonomic revision is necessary. The exact position for Colvillea is ambiguous and suggests hybridization or incomplete lineage sorting for this taxon. Species evolved mostly during the Miocene from southern Madagascan ancestors. The phylogeny shows two clades associated with the geographic distribution of species, opposing the widespread species to the rare and mostly southern, endemic, spiny thicket species. The different hypotheses that could explain the Miocene long distance dispersal of the two African Delonix species from Madagascar are discussed. We recommend that conservation efforts should focus on past and present biotic and abiotic interactions in the Madagascar spiny thicket, considering that this unique habitat is an evolutionary cradle for at least three genera.

Many plant species are closely related to river biotopes or to biotopes influenced by rivers. River systems create important linear corridors in ecosystems and directly or non-directly influence spatial spread of species in these environments. This offer many questions about species spread in this system. We can answer these questions by using molecular methods. Using 10 microsatellite (SSRs) primers, 202 individuals of Phragmites australis from 60 populations were analysed. Those analyses allowed reveal kinship of individuals, obtain information about spatial spread of populations and about spreading of common reed. Phragmites australis creates both - monoclonal and polyclonal - populations. Dependence of rate of clonality on environment was revealed. Populations from river banks are more monoclonal, populations from pond shores are more polyclonal. Populations are isolated. The highest percentage of variability was explained on among-population level. This is common for anemochoric species. Communication between populations is present, more on shorter distances. Evidence for vegetative spread was found on short distance. Generative spread is much more common. Long-distance spreading is mediated by generative diaspores - seeds. This spreading includes within-river spread, among-river spread and spreading...