Understanding flower diversity requires on one hand the study of the molecular and developmental origin of floral architecture, and on the other the study of the functional and ecological consequences of flower morphology. A great deal of that diversity can be found at the perianth level which comprises the sepals and petals, sterile and versatile organs that play a major role in the reproductive success of animal pollinated flowering plants through their attractive characteristics.This thesis is the result of a multidisciplinary effort to understand the genetic and molecular origin as well as the evolutionary significance of perianth diversity, using the Nigella damascena L. as a model. This Ranunculaceae species presents a rare naturally occurring floral dimorphism affecting perianth architecture. The putatively ancestral form found in natural populations has a well differentiated bipartite perianth composed of five petaloid sepals and eight nectariferous petals, while the perianth in the alternative apetalous mutant, cultivated for horticultural purpose, has no petals and but is instead composed of numerous organs showing a continuum of forms from outer sepal-like to inner stamen-like.The first part of this thesis was dedicated to the study of the developmental, genetic and molecular origin of this dimorphism, via a detailed characterization of floral morphology and development in both morphs, which laid a foundation for the interpretation of the results of a candidate gene approach. Using expression analysis and functional validation we showed that NdAP3-3 is fully responsible for the complex N. damascena floral dimorphism, suggesting that it plays a role not only in petal identity but also in meristem patterning, possibly through the regulation of perianth organ number and perianth-stamen boundary.The second half of this thesis focused on the impact of the floral dimorphism on the reproduction mode and evolutionary maintenance of the two morphs. We assessed reproduction strategies and reproductive success in the two morphs by studying a polymorphic experimental population in natural conditions. The absence of petals in the mutant form was associated with a qualitative drop in pollinator visitation which resulted in a shift towards selfing. The study of their progeny suggests that selfing had a negative effect on the descendant’s vigor via inbreeding depression. Additionally, in our material, the allele responsible for the apetalous phenotype seems to be linked to a favorable allele increasing fitness. We discuss the mechanisms of the dimorphism maintenance in light of these results.
