Distantly related lineages of the enigmatic giant rosette plants of tropical alpine environments provide classical examples of convergent adaptation. For the giant senecios (Dendrosenecio), the endemic landmarks of the East African sky islands, it has also been suggested that parallel adaptation has been important for within-lineage differentiation. To test this hypothesis and to address potential gene flow and hybridization among the isolated sky islands, we organized field expeditions to all major mountains. We sampled all currently accepted species and all but one subspecies and genotyped 460 plants representing 109 populations. We tested whether genetic structuring corresponds to geography, as predicted by a parallel adaptation hypothesis, or to altitudinal belt and habitat rather than mountains, as predicted by a hypothesis of a single origin of adaptations. Bayesian and Neighbor-Net analyses showed that the main genetic structure is shallow and largely corresponds to geography, supporting a hypothesis of recent, rapid radiation via parallel altitude/habitat adaptation on different mountains. We also found evidence for intermountain admixture, suggesting several long-distance dispersals by wind across vast areas of unsuitable habitat. The combination of parallel adaptation, secondary contact, and hybridization may explain the complex patterns of morphological variation and the contradicting taxonomic treatments of these rare enigmatic giants, supporting the use of wide taxonomic concepts. Notably, the within-population genetic diversity was very low and calls for increased conservation efforts.