The island of Sri Lanka’s exceptional biodiversity and enigmatic biogeography begs investigation, as the island is key in understanding the evolution of the Asian tropical flora. Since the Jurassic, Sri Lanka has been subjected to remarkable tectonic changes, thus its flora could have been influenced by that of a number of nearby landmasses, as well giving Sri Lanka the potential to have played a wider role in the assemblage of floras elsewhere. Firstly, as Sri Lanka originated as a fragment of the supercontinent Gondwana, part of its flora may contain Gondwanan relict lineages. There is also the potential for immigration from Laurasia after the Deccan Plate collided with it 45-50 Mya. Further, Sri Lanka may harbour floristic elements from nearby land masses such as Africa and Southeast Asia as a result of long distance dispersals, and in situ speciation has the potential to have played an important role in enhancing the endemic Sri Lankan flora. I tested the relative contributions of the above hypotheses for the possible origins of the Sri Lankan flora using three representative families, Begoniaceae, Sapotaceae and Zingiberaceae. These families represent both herbaceous and woody elements, and have high diversity across the tropics. Dated molecular phylogenies were constructed for each family. I used recent analytical developments in geographic range evolution modelling and ancestral area reconstruction, incorporating a parameter J to test for founder event speciation. A fine scale area coding was used in order to obtain a better picture of the biogeography of continental Asia. Amongst all the models compared, a dispersal-extinction cladogenesis model incorporating founder event speciation proved to be the best fit for the data for all three families. The dates of origin for Sri Lankan lineages considerably post-date the Gondwanan break up, instead suggesting a geologically more recent entry followed by diversification of endemics within the island. The majority of Sri Lankan lineages have an origin in the Sunda Shelf (53%). Persistence of warm temperate and perhumid climate conditions in southwestern Sri Lanka resembling those of Peninsular Malaysia and Sumatra could have facilitated suitable habitats for these massive dispersals from the Sunda Shelf region. Some trans-oceanic long distance dispersals from Africa (11%) are also evidenced, again these are too young to accept a hypothesis of dispersal during the Deccan Plate’s migration close to the African coast during the late Cretaceous, but occurred later during the Miocene. Further, some lineages of Laurasian origin (20%) are evidenced in the Zingiberaceae with ancestral areas of China and Indochina, which is congruent with a post collision invasion. Among the families tested, dispersals have occurred stochastically, one during the Eocene, six during the Oligocene, seven during the Miocene, two during the Pliocene and one during the Pleistocene. The highest number of dispersals occurred during the Miocene when a warm climate was prevailing during the Miocene thermal maximum. My results confirm that in situ speciation is an important contributor to the Sri Lankan flora. More rapid radiation of endemics has occurred during Pliocene-Pleistocene; two endemics in Begoniaceae, ten endemics in Sapotaceae and ten endemics in Zingiberaceae have evolved in situ during this period. Sri Lanka will have been subjected to expansion and contraction of climatic and vegetation zones within the island during glacial and interglacial periods, potentially resulting in allopatric speciation. As a conclusion, long distance dispersals have played a prominent role in the evolution of the Sri Lankan flora. The young ages challenge the vicariant paradigm for the origin and current disjunct distributions of the world’s tropical lineages and provide strong evidence for a youthful tropics at the species level. The thesis contains six chapters; first two are introductory chapters, then there are three analytical chapters, one for each family, and finally a summary chapter is provided. Each analytical chapter is written as a stand-alone scientific publication, thus there is some repetition of relevant content in each.

La Béringie, un vaste territoire qui s’étend de la Sibérie orientale au Territoire du Yukon, est perçue comme le point d’entrée des populations humaines en Amérique. A la fin du Pléistocène, ce territoire déglacé aurait constitué un refuge aux premières populations préhistoriques se dispersant hors d’Asie. Selon les données génétiques et paléo-génétiques, la Béringie fut occupée au cours du Dernier Maximum Glaciaire (19 000-23 000 cal BP, années calibrées Before Present) par une population humaine qui demeura génétiquement isolée durant près de 8000 à 9000 ans, donnant ainsi naissance à la lignée des Natifs Américains qui allaient se disperser, plus tard, au sud des masses glaciaires nordaméricaines et jusqu’en Amérique du Sud. Cette « Beringian standstill hypothesis », toutefois, ne trouva aucun soutien dans le registre archéologique : en Sibérie orientale, le plus ancien site est daté à 32 000 cal BP, tandis qu’en Alaska et au Yukon, la présence humaine ne remonte pas au-delà de 14 000 cal BP. Dans les années 70-80’s, le site des Grottes du Poisson-Bleu (Yukon) livra des outils en pierre et des ossements supposés modifiés par les humains, enfouis dans un dépôt loessique pléistocène ; les découvertes encouragèrent les archéologues J. Cinq-Mars et R. Morlan à évoquer l’hypothèse d’une occupation humaine sporadique dans le nord du Yukon entre 11 000 et 30 000 cal BP environ. La nature anthropogénique des échantillons osseux soumis aux datations radiocarbones ainsi que l’intégrité de la stratigraphie furent toutefois remises en question par une majorité d’archéologues. La présente dissertation propose une analyse archéozoologique et taphonomique rigoureuse et systématique des assemblages fauniques de mammifères des Grottes I et II dans le but d’appréhender les facteurs responsables de l’accumulation et de la modification du matériel osseux. De nouvelles datations radiocarbones effectuées par le laboratoire Oxford Radiocarbon Accelerator Unit sur des ossements portant des traces indéniablement culturelles permettent une datation précise de l’occupation humaine du site. Les résultats illustrent plusieurs traces de découpe sur des os de cheval, caribou, wapiti et possiblement bison et mouflon, tandis que des ossements de mammouth pourraient avoir été collectés pour l’industrie osseuse. Les nouvelles datations AMS suggèrent que les Grottes du Poisson-Bleu étaient occupées de façon sporadique entre 12 000 et 24 000 cal BP, soit pendant et après le Dernier Maximum Glaciaire. Le site offre ainsi un soutien archéologique à l’hypothèse de l’isolation génétique des populations béringiennes à l’origine des premières dispersions en Amérique. L’histoire taphonomique des Grottes du Poisson-Bleu rejoint celle des sites karstiques béringiens qui illustrent des occupations interspécifiques alternées entre carnivores et des fréquentations humaines de courte durée pour des activités de chasse. En outre, les altérations anthropiques sur des os de cheval des Grottes I et II ravivent le débat sur les extinctions de la mégafaune à la fin du Pléistocène (ca. 14 000 cal BP). Le site souligne l’incomplétude du registre archéologique et invite à multiplier les efforts de recherche en Béringie si l’on veut être à même de comprendre la préhistoire du peuplement des Amériques. / Beringia, a vast landscape stretching from eastern Siberia to the Yukon Territory, is thought to be the initial entry point of humans into North and South America. At the end of the Pleistocene, this unglaciated region constituted a refugium for the first prehistoric populations dispersing out of Asia. According to genetic and palaeogenetic data, Beringia was occupied during the Last Glacial Maximum (19 000-23 000 cal BP, calibrated years before present) by a human population that remained genetically isolated for about 8000 to 9000 years, leading to the divergence of the Native American lineage that would eventually disperse south of the ice-sheets into North and South America. The « Beringian standstill hypothesis » is not well supported in the archaeological record, however: in eastern Siberia, the oldest archaeological site is dated to 32 000 cal BP while in Alaska and the Yukon, evidence for a human presence doesn’t exceed 14 000 cal BP. Excavated in the 70s-80s, the Bluefish Caves site (Yukon) yielded stone tools and bone remains thought to have been culturally modified, buried in a Pleistocene loess deposit; the discovery encouraged archaeologists J. Cinq-Mars and R. Morlan to propose that humans occupied the caves sporadically between about 11 000 and 30 000 cal BP. The anthropogenic nature of the bone samples submitted for radiocarbon analysis and the stratigraphic integrity of the site didn’t convince the scientific community, however. The current dissertation proposes a rigorous archaeozoological and taphonomic analysis of the mammal bone assemblages of Caves I and II in order to identify the agents responsible for the accumulation and modification of the bone material. The results show several cut marks on bone specimens belonging to horse, caribou, wapiti and possibly bison and Dall sheep, while mammoth skeletal remains may have been collected for bone industry. New radiocarbon dates obtained by an Oxford laboratory (Oxford Radiocarbon Accelerator Unit) on bone bearing indisputable evidence of cultural modification allow the precise dating of the human occupation at the site. The AMS dates suggest that the Bluefish Caves were occupied sporadically between 12 000 to 24 000 cal BP, i.e., during and after the Last Glacial Maximum. The site, therefore, offers archaeological support for the Beringian standstill hypothesis. The taphonomic history of the Bluefish Caves, as well as other Beringian karstic sites, shows use of the caves by various carnivores and short-term human occupations for hunting activities. Moreover, cultural modifications on horse bone from Caves I and II enhance the debate surrounding the megafaunal extinctions at the end of the Pleistocene (ca. 14 000 cal BP). The site underlines the incompleteness of the archaeological record and invites us to expand research efforts in Beringia if we are to understand the prehistory of the first people of the Americas.