Résumé
The Amazonian forest has long been considered "virgin" from any ancient human impact. However, evidence contradicting this preconceived idea has been accumulating over the past few decades, including vast expanses of terra preta. These are soils that are highly enriched in nutrients, very fertile and whose anthropogenic origin is now beyond doubt. They are proof that pre-Columbian societies have had a considerable impact on their environment. Many studies have focused on the central and western Amazon where terra preta areas are numerous and host very diversified vegetation. The Guiana Shield has not benefited from the same interest. However, in the Guianan forests, there are many signs of ancient human occupation, including in the vegetation. As on terra preta sites, many useful species are found in some places, in greater abundance than can be observed in the case of natural dispersion. It is in this context that this thesis, which focuses on the impact of pre-Columbian populations on the forest vegetation of the Nouragues Reserve in French Guiana, is based. This work seeks to characterize this impact through an anthracological approach - the identification of charcoal, a method rarely used in tropical ecosystems. As a result, a significant part of the thesis is devoted to methodology. As a first step, an identification key specifically adapted to the anatomical characteristics observed on charcoal and targeting Guianan species was implemented to facilitate the identification process. This key is freely downloadable online and is editable, so that new descriptions and taxa can be added. Then, different methods of charcoal sampling - soil profile sampling, auger sampling and opportunistic sampling on uprooting mounds or during digging - were compared. The results show that the first two methods allow to recover the same amount of charcoal but not the same taxonomic diversity. In most of the studied sites, sampling along a soil profile provides more diversity than augers because of the often monospecific assemblages provided by the latter. Nevertheless, almost every new auger sampling allows to recover a new taxon. Opportunistic sampling also increases the assemblage diversity. The combination of the three methods therefore makes it possible to obtain more diversified and more representative assemblages. A sampling protocol combining several pits, augers and opportunistic sampling is proposed to ensure a better representation of anthracological assemblages and to overcome this spatial heterogeneity as much as possible. Finally, the dates carried out on the charcoals collected during the thesis indicate that the Nouragues experienced two main periods of occupation ranging from 1300-1000 cal years BP to 600-400 cal years BP. Anthracological assemblages show that these occupations have induced significant vegetation changes at some sites. Indeed, in two sites, the forest has given way to very different plant formations, dominated by bambusiform poaceae or lianas. The results suggest that one of the bamboo formations may have developed after the abandonment of a semi-permanent cultivation plot, thus joining hypotheses formulated in previous historical ecological studies. The results of anthracological analyses also show the presence of useful and secondary vegetation taxa, thus reflecting the impact of ancient populations on forest environment. Finally, the selection of firewood seems to have been adapted according to the local vegetation. The type of activity that causes these vegetation changes remains to be defined and the understanding of pre-Columbian sites in French Guiana, as elsewhere in the Amazon, should be improved through the complementary contributions of other disciplines.