Hughes et al. 2018 ray-fin fish cladogram: genes vs. traits

Zoologists are increasingly relying on genomics (genes)
to reconstruct nature’s family trees, to their peril. We’ve already seen how genes fail to match phenomic (trait-based) tests, the only kind that can include and incorporate fossil taxa.

Hughes et al.  2018 report,
“Ray-finned fishes form the largest and most diverse group of vertebrates. Establishing their phylogenetic relationships is a critical step to explaining their diversity. We compiled the largest comparative genomic database of fishes that provides genome-scale support for previous phylogenetic results and used it to resolve further some contentious relationships in fish phylogeny. Our time-calibrated analysis suggests that most lineages of living fishes were already established in the Mesozoic Period, more than 65 million years ago.”

Hughes et al. employed no fossil taxa
and failed to confirm the tree topology recovered by the large reptile tree (LRT, 1553 taxa) which tests physical traits as well as large clades omitted or overlooked by Hetal.

Let’s compare trees.

1. Hughes et al. do not separate lungfish from other lobefins, but lump them all in the monophyletic outgroup Sarcopterygii. The LRT starts earlier, with jawless theolodonts. Some traditional ray-fins are lobefins. Some lobefins give rise to tetrapods. Others give rise to ray-fins (Fig. 1). 
2. Hughes et al. do not test members of the shark + ratfish clade.
3. Hughes et al. nest sturgeons (Pseudoscaphirhynchus, Acipenseriformes) between dissimilar bichirs (Polypterus) and bowfins (Amia). The LRT nests sturgeons with more similar ratfish.
4. Hughes et al. do not test members of the placoderm clade.
5. Hughes et al. nest catfish (Clarias, Siluriformes) with dissimilar knife fish (Gymnotus). (Gymnotiformes).  The LRT nests catfish as placoderm sisters, not far from the morphologically similar basal shark with similar tooth pads, Rhincodon. Thus the radiation of convergent ray fin fish was established in the Silurian, not the late Cretaceous.
6. Hughes et al. nest the bichir, Polypterus, as the basalmost ray fin fish. The LRT nests air breathing Polypterus with similar lungfish, and Polypterus is not the basalmost ray fin fish in the LRT (see 1–5 above). 
7. Hughes et al. nest garfish (Lepisosteiformes) with dissimilar Amiiformes. The LRT nests garfish with sticklebacks while Amiiformes (Amia) nest between tuna (Thunnus) and arowana (Osteoglossum) among living taxa. 
8. Hughes et al. nest dissimilar tarpons and eels together (clade: Elopomorpha). The LRT nests more similar carp (Cyprinus) with tarpons (Megalops) and eels nest in many places, but European eels (Anguilla) nest with similar cusk eels (Dicrolene), derived from fast-swimming rudder fish (Seriola zonata). 
9. Hughes et al. nest dissimilar goosefish (Lophius, Lophiformes) with pufferfish (Diodon) and triggerfish (Balistes, Tetraodontiformes). The LRT nests goosefish with more similar (when you look at their skulls) knife fish (Gymnotus) + electric eels (Electrophorus) 
10. …and so it continues with little consensus.

As we’ve seen before with birds, mammals and reptiles in general,
deep time genomic (gene) studies fail time and again to provide a tree topology in which sister taxa demonstrate a gradual accumulation of derived traits — which is the ultimate validation test of any cladogram. Gene studies, like Hughes et al., that a priori exclude taxa based on traits (= ‘pulling a Larry Martin’): assuming that no ray fin fish are convergent, or that all eels are related to one another, are also doomed to fail.

References
Hughes L et al. (20 co-authors) 2018. Comprehensive phylogeny of ray-finned fishes (Actinopterygii) based on transcriptomic and genomic data. PNAS 115(24):6249–6254.

pnas.org/cgi/doi/10.1073/pnas.1719358115