Résumé
Selection shapes genetic diversity around target mutations, yet little is known about how selection onspecific loci affects the genetic trajectories of populations, including their genome-wide patterns of diversityand demographic responses. Here we study the patterns of genetic variation and geographic structure in aneotropical butterfly,Heliconius numata, and its closely related allies in the so-called melpomene-silvaniformclade. H. numata is known to have evolved an inversion supergene which controls variation in wing patternsinvolved in mimicry associations with distinct groups of co-mimics whereas it is associated to disassortativemate preferences and heterozygote advantage at this locus.We contrasted patterns of genetic diversity andstructure 1) among extant polymorphic and monomorphic populations of H. numata, 2) between H. numataand its close relatives, and 3) between ancestral lineages.We show that H. numata populations which carrythe inversions as a balanced polymorphism show markedly distinct patterns of diversity compared to allother taxa. They show the highest genetic diversity and effective population size estimates in the entireclade, aswell as a lowlevel of geographic structure and isolation by distance across the entire Amazon basin.By contrast, monomorphic populations of H. numata as well as its sister species and their ancestral lineagesall show lower effective population sizes and genetic diversity, and higher levels of geographical structureacross the continent. One hypothesis is that the large effective population size of polymorphic populationscould be caused by the shift to a regime of balancing selection due to the genetic load and disassortativepreferences associated with inversions. Testing this hypothesis with forward simulations supported theobservation of increased diversity in populations with the supergene. Our results are consistent with thehypothesis that the formation of the supergene triggered a change in gene flow, causing a general increasein genetic diversity and the homogenisation of genomes at the continental scale.