Among cartilaginous fishes, Holocephali represents the species-depauperate, morphologically conservative sister to sharks, rays, and skates and the last survivor of a once far greater Paleozoic and Mesozoic diversity. Currently, holocephalan diversity is concentrated in deep-sea species, suggesting that this lineage might contain relictual diversity that persists in the ocean depths. Yet, the relationships of living holocephalans to their extinct relatives and the timescale of diversification of living species remain unclear. Here, we reconstruct the evolutionary history of holocephalans using comprehensive morphological and DNA sequence datasets. Our results suggest that crown holocephalans entered and diversified in deep (>1000 m) ocean waters after the Cretaceous-Paleogene mass extinction, contrasting with the hypothesis that this ecosystem has acted as a refugium of ancient cartilaginous fishes. These invasions were decoupled from the evolution of key features of the holocephalan body plan, including crushing dentition, a single frontal clasper, and holostylic jaw suspension, during the Paleozoic Era. Yet, they considerably postdated the appearance of extant holocephalan families by 150 million years ago during a major period of biotic turnover in oceans termed the Mesozoic Marine Revolution. These results clarify the origins of living holocephalans as the recent diversification of a single surviving clade among numerous Paleozoic lineages.

Morphological Dataset Construction. In order to test the phylogenetic position of living holocephalans and Mesozoic species among the larger Paleozoic diversity of the total clade, we constructed a new taxon-character matrix by expanding the dataset of Frey et al. [8] using characters from Didier [17] and Patterson [46]. We also edited character states and added 14 living and extinct holocephalan taxa to thoroughly sample holocephalan diversity within and proximal to the crown clade. Newly added data include genus-level scorings for all five living holocephalan genera, as well as for species in the genera †Acanthorhina, †Chimaeropsis, †Elasmodectes, †Ischyodus, †Metopacanthus, †Myriacanthus, and †Squaloraja. Due to the lack of available information on their morphology and our focus on incorporating post-Palaeozoic taxa into the analysis, we did not include any representatives of the Paleozoic pan-holocephalan clades †Eugenodontia and †Petalodontiformes, which mostly include tooth taxa and for which only a handful of holomorphic specimens [10,11,47] of limited phylogenetic informativeness are known. Future discoveries, including the description of unpublished specimens [4], will be needed to resolve the relationships of these Carboniferous and Permian pan-holocephalans; we expect that the phylogenetic matrix employed in this study will contribute to this pursuit. The final matrix included 36 operational taxonomic units coded for 236 characters. Details of sources for character scorings, character additions, removals, state modifications, and taxon inclusion are included in the Supplementary Information.

DNA Sequence Dataset Assembly. In order to sample the species diversity of living holocephalans most fully, we targeted sequences for the cytb, COXI, ND2, and 16s mitochondrial loci on the NCBI repository Genbank. Using this approach, we were able to sample 100% of all living holocephalan genera and species complexes, including all three species in Callorhinchus, seven of the nine recognized species in Rhinochimaeridae, and 35 of 47 described species of Chimaeridae [40], plus two specimens of Chimaera sp. with mitogenomes that appear to diverge considerably from other species [48,49]. We concatenated these data into a single set of aligned sequences for subsequent phylogenetic analyses.