Abstract
The Neotropical genus Cecropia Loefl. includes pioneer trees that colonize open gaps and disturbed environments. It is distributed from northern Argentina to southern Mexico and the West Indies. Cecropia constitutes a model for study in a wide variety of disciplinary fields (e.g.: ecological role in forest succession, mutualistic relationships with ants, structurefunction relationships in trees, invasive dynamics...). In spite of the volume and diversity of research carried out on this genus, it is paradoxical that the core ability to identify Cecropia species remains problematic and impedes progress. The monograph by Berg & Franco- Rosselli (2005) describe 61 species and 2 subspecies whereas 215 taxa have been described since the description of the genus in 1758. The proposed circumscription is essentially based on morphological characters. For the last 30 years, the advent of digital technology and advances in molecular biology have made it possible to approach taxonomy in a much more integrative manner. Approaches in numerical taxonomy can thus be confronted with other points of view such as the coherence of the ecological niche, phenology or genetic structure of populations. With the disappearance of the last two specialists of the genus and authors of the monograph, the objective of this thesis is to organize a framework through a series of software and analytical tools to (i) extract the "substantial marrow" of the monograph by a numerical approach and (ii) to review the circumscription of species by an integrative taxonomy approach coupling numerical taxonomy and niche models on a continental scale.A study of the demography of the populations of two species on the Paracou experimental station complements this study to discuss the importance of scales in the characterization of the ecological niche of the species. Our results show that the monograph is coherent, with taxa that are morphologically well delineated, but with a massive synonymy that is not argued. Our analysis allows us to identify the most discriminating characters to be used in a numerical taxonomy approach, on the one hand, and allowed us to build a multi-entry identification key for the genus, on the other hand. At the specimen level, we find well delineated morphotypes that are consistent with the description of the species in the monograph. On the other hand, the species within the "peltata" infrageneric group forms not very coherent subsets. It seems that C.C. Berg and P. Franco-Rosselli favoured geographical coherence to the detriment of morphological coherence within the group. Nevertheless, our results do not allow us to reconcile these two aspects and we remain at an impasse. The modelling of the climatic niche shows a strong overlap between the species and no climatic identity really specific to one of them. The identification of aberrant specimens in the geographical space or in the climatic niche space nevertheless helps to identify poor determinations or heterogeneous collection intensities within the potential distribution range of the species. We also identify several species with highly invasive potential. The analysis of population dynamics on the Paracou experimental device shows subtle edaphic preferences of species ata local scale. This work calls for the integration of molecular biology approaches that seem indispensable to help circumscribe species and better understand the evolutionary dynamics of what appears to be a species complex, where frequent hybridization generates a morphological continuum. This work makes it possible to generate "roadmaps" for designating geographical locations to be studied as a priority where species recognized in the monograph cohabit, but whose morphological delimitations remain fragile.