Abstract
The genus Eupelmus Dalman (Chalcidoidea: Eupelmidae: Eupelminae) includes ecto-parasitoids attacking mostly nymphal and larval stages of various holometabolous insects. So far, the systematics of the Eupelmus genus remained poorly resolved due to limited data restricted to morphological information and the absence of recent, reliable and comprehensive taxonomic revisions. In this context, many questions arise concerning (i) the relevance of the current sub-generic classification of the Eupelmus genus; (ii) the availability of the taxonomic status of some species described; (iii) the reliability of ecologic information such as the host range and the geographical distribution; and therefore, (iv) the understanding of the ecological specialization's processes and the potential role of certain species of Eupelmus as auxiliaries of biological control. This thesis work has therefore aimed to address all of these questions by using, for the purposes of phylogeny and taxonomy, an integrative approach combining molecular (nuclear and mitochondrial DNA) and morphological data.The results obtained concerning the sub-generic phylogenetic relationships show that the supposed structuration of Eupelmus genus into three subgenera (Eupelmus, Episolindelia and Macroneura), can not be retained and that this genus would be rather structured, at the retained geographical scale, in a dozen of species groups. In addition, the study of taxonomy of two complexes (sets of morphologically similar species), the complex “urozonus” and “vesicularis”, overall highlights unsuspected diversity in the Euro-Mediterranean area and more than ten species have been discovered and described as new species on the occasion of this work. Generally, the morphological characteristics, nuclear markers and mitochondrial markers have been relatively consistent except within the vesicularis complex which exhibits more marked divergence in the mitochondrial DNA.In the particular case of the urozonus species group, this thesis work has also allowed us to study the evolution of the host specificity in relation to a relatively well-resolved multi-locus molecular phylogeny and an estimate of the length of ovipositor, a morphological character that could explain the ability to parasitize protected hosts. In general, the comparative analyses show that the host specificity is not constrained by the phylogeny. We indeed observe highly contrasted hosts ranges between closely phylogenetically related species. Similarly, the length of ovipositor, which seems a very labile character at this scale, does not seem to determine the host range. All the results obtained have finally been used to better understand the potential role of some Eupelmus species on two insect pests, the fruit olive fly Bactrocera oleae and the chestnut gall wasp Dryocosmus kuriphilus.