Abstract Premise Species of Deuterocohnia (17 spp.) show extraordinary variation in elevation (0–3900 m a.s.l.) and growth forms, and many have narrow geographic distributions in the west‐central Andes and the Peru‐Chile coast. Previous research using few plastid and nuclear loci failed to produce well‐resolved or supported phylogenies. Here we sequenced 1815 single‐copy nuclear genes and whole plastomes to infer relationships, screen for reticulation, reconstruct evolution of vegetative and floral characters, and evaluate species groups and their historical biogeography. Methods We developed the Bromeliad1815 bait set to capture low‐copy nuclear genes across Bromeliaceae, producing nuclear and plastome phylogenies for Deuterocohnia and outgroups in six bromeliad subfamilies using maximum likelihood, ASTRAL, and network analyses; test for cytonuclear conflict and its potential causes; and evaluate evolution of morphological characters in relation to each other and elevation using phylogenetic PCA and phylogenetic regression. Results We produced fully resolved, strongly supported nuclear and plastome phylogenies for Deuterocohnia , with crown ages of 5.5 and 8.0 Mya, respectively. Cytonuclear conflict appears driven mainly by hybridization/introgression, consistent with several species co‐occurring in small areas. Vegetative organs and growth form become increasingly compact with elevation, reflecting adaptation to desiccation, wind exposure, and cold soils. Deuterocohnia arose in southeastern Bolivia and repeatedly evolved up‐ and downslope into other habitats from Andean Yungas at mid‐elevation. Conclusions Our results imply rapid adaptive divergence (e.g., in strobilifera‐chrysantha ), convergent evolution (two origins of the cushion growth‐form), phylogeny consistent with form in some cases (e.g., seramisiana‐brevispicata‐meziana ) and recurrent effects of the Rio Pilcomayo barrier on speciation and chloroplast capture.