The recognition of ecological speciation and hybridisation as key components of speciation has led to a major shift in evolutionary biology over the last decade. The mimetic Heliconius butterflies of the Neotropics have served as a prominent example of both, although the vast majority of studies have focused exclusively on very recent divergences and on colour pattern adaptation, neglecting deeper timescales and patterns across the rich diversity of the adaptive radiation. The relative importance across adaptive radiations of allopatry, changing ecological pressures, adaptive morphology and introgression promoted by natural and sexual selection remains unknown. I combine phylogenetics, genomics and comparative approaches to elucidate the patterns and identify the key drivers of diversification in the continental-scale radiation of Heliconius and nine related genera. I present the first comprehensive, multilocus and time-calibrated phylogeny of the group and find that shifts in diversification rate cannot be unequivocally attributed to a single environmental factor. The potential role of coevolution with the obligatory host plants Passiflora is examined with the aid of a new phylogeny of the passion vines. Evidence is found for diffuse coevolution, as the diet of most Heliconiini is not predicted by their phylogeny and varies at short timescales. Although passion vine butterflies are the leading example of speciation by hybridisation, this process has been described in only one subgenus. I utilise whole exome data to examine the morphologically suggestive case of a putative hybrid from another clade and find no evidence of introgression. The data is further used to answer long-standing questions about the origins of the most phenotypically diverse species. In the final chapter whole genome data are applied to characterise the patterns of divergence and gene flow across the entire genera Heliconius and Eueides, characterising the patterns of conflicting signal and comparing the performance of philosophically distinct approaches to reflect the heterogeneity across the genome. I find that the phylogeny is unstable due to a combination if incomplete lineage sorting and introgression and may never be fully resolved, perhaps necessitating a network representation. Genomic admixture is a unique property of just one clade comprising a quarter of all species, and involves primarily the adaptive wing pattern loci. Surprisingly, the sex-linked Z chromosome shows a different order of speciation events. Altogether my results show unexpectedly limited role of allopatry, geoclimatic variables and host plant adaptation in the diversification of a major insect radiation, thus confirming the importance of ecological speciation driven by selection on wing patterns. However, I also demonstrate that introgression may be less important in this group than previously thought.