A deep dive into the coelacanth phylogeny.

The discovery in 1938 of a living coelacanth, Latimeria chalumnae, triggered much research and discussion on the evolutionary history and phylogeny of these peculiar sarcopterygian fishes. Indeed, coelacanths were thought to represent the 'missing link' between fishes and tetrapods, a phylogenetic position which is now dismissed. Since the first analyses using a phylogenetic approach were carried out three decades ago, a relatively similar data matrix has been consistently used by researchers for running analyses, with no significant changes aside from the addition of new taxa and characters, and minor corrections to the states' definition and scorings. Here, we investigate the phylogeny of Actinistia with an updated data matrix based on a list of partially new or modified characters. From the initial list of characters available in the most recent studies, we removed 16 characters, modified 16 other characters' definition and added 18 new characters, resulting in a list of 112 characters. We also revised the data matrix by correcting 171 miscoding found for 37 taxa. Based on the new phylogeny, we propose a new classification of coelacanths including 46 coelacanth genera, part of them allocated within nine families and four sub-families. Most of these groups were already named but were not recognised as clades, or poorly or not diagnosed in previous phylogenetic analyses. We provide several new or emended diagnoses for each clade. For the first time, a set of Palaeozoic coelacanth genera are found gathered within a clade, namely the Diplocercidae. All Mesozoic coelacanths, including extant Latimeria, are resolved as members of the order Coelacanthiformes, a clade that arose in the Permian, with Coelacanthus diverging first. We also found that most Mesozoic coelacanths are gathered into a clade, the Latimerioidei, itself divided into the Latimeriidae and the Mawsoniidae, each of which is divided into two subfamilies. Although these important changes, the new phylogeny of the Actinistia shows no significant alteration, and it remains relatively similar compared to previous studies. This demonstrates that the coelacanth phylogeny is now rather stable despite the weak support for most nodes in the phylogeny, and despite the difficulty of defining relevant morphological characters to score in this relatively slowly evolving lineage.