Polypterid fish were considered to be archaic outliers of the bony-fish grouping. Fossil analysis now places them at the heart of early ray-finned fishes, a radical change that transforms the timing of their evolution. See Letter p.265
A proverbial 'odd fish', the bichir (Polypterus) is so extraordinary that the palaeontologist Georges Cuvier claimed that its discovery alone justified Napoleon Bonaparte's expedition to Egypt1. Its unusual mixture of seemingly primitive and highly specialized features, perhaps best exemplified by the distinctive characteristics of its muscular pectoral fins2,3, has led to a lengthy and convoluted classification history. The settled consensus on the basis of genetic and fossil analysis4 is that polypterids (Polypterus and its sister genus, Erpetoichthys) are members of the ray-finned fishes (Actinopterygii), so-called because their fins are mostly supported by slender bony rays, rather than by the predominantly muscular (but also ray-bearing) appendages of lobe-finned fishes.
However, because nearly all of the earliest fossil actinopterygians (informally grouped under the name palaeoniscoids) are widely agreed to be more closely related to all other living members of the Actinopterygii5,6 than to polypterids, the polypterids have been excluded to the outer margins of the clade. Thus, polypterids are considered to have split off from the rest of the actinopterygians close to the start of their evolutionary tree. On page 265, Giles et al.7 overthrow this established view, with radical consequences for how we understand the early history of the entire actinopterygian clade. This is no trivial matter: the Actinopterygii comprise around half of all living vertebrate species.
Confined to sub-Saharan Africa and the Nile region, living polypterids have variously been considered either as intermediates between ray-finned and lobe-finned fishes or as 'living fossils', species that have survived from an earlier era with minimal evolutionary change1,8. Polypterids have also been used as models for investigating the kinds of change that might have occurred in the fins or limbs of the earliest vertebrates that spent a substantial part of their life out of the water9,10. However, polypterid-centred research is hobbled by the lack of a substantial fossil record. The scanty remains available date back only as far as the mid-Cretaceous period11 (around 100 million years ago). They therefore shed little light on primitive conditions, and require a credibility-stretching lineage extension of more than 280 million years to reach the implied age of their departure from the rest of the Actinopterygii in the middle Devonian period5 (around 390 million to 380 million years ago).
Giles et al. find a solution to this problematic scarcity of fossil polypterids by investigating a group of fishes from the early Mesozoic era (252 million to 201 million years ago), the Scanilepiformes. Although these were previously suggested12,13 to be ancestral polypterids, supporting data were limited. Like many fossil fishes, scanilepiformes are usually preserved as little more than flattened spreads of head plates, fins and scales, yielding none of the data-rich structures of the animal's interior. Thus, reports of three-dimensionally preserved skulls of the scanilepiform genus Fukangichthys6 from the middle Triassic period (247 million to 235 million years ago) prompted Giles and colleagues to investigate these samples using an imaging technique called high-resolution X-ray computed microtomography. The authors' results reveal structural details of the jaws, palate, braincase and gill arches, each showing distinctive polypterid-like characteristics. Importantly, although all the new data are from parts of the skull, these features constitute multiple lines of evidence for polypterid affinity because they are not from a single functionally or developmentally interlinked system.
The evolutionary tree derived from Giles and colleagues' analyses places polypterids firmly within the Scanilepiformes. Furthermore, this analysis locates nearly all of the considerable array of palaeoniscoids on the actinopterygian stem group (those branches of the evolutionary tree that split off before the last common ancestor of all living ray-finned fishes). These changes transform our view of the evolutionary origin of this major vertebrate division: polypterids now branch from the heart rather than the margins of the early evolutionary radiation of ray-finned fishes, and the Actinopterygii crown group (which includes all living representatives of the group, plus all extinct descendants of their most-recent common ancestor) emerges some 45 million years later than existing estimates had suggested (Fig. 1). This is the most radical revision of early actinopterygian evolution since the late 1980s5: it offers new data, evolutionary trees and timescales, and provides newly populated stem lineages where none existed before for polypterids and the Actinopterygii as a whole.
But there is more, because the authors also tested the strength of the branching pattern in their evolutionary tree, and identified a substantial problem. The early evolution of actinopterygians can be summarized as a lengthy fuse that extended throughout the Devonian period (from 419 million to 359 million years ago), surviving extinctions close to the end of the Devonian, to enter the early Carboniferous period (359 million to 323 million years ago) with a burst of evolutionary branching. Groups that arose in the early Carboniferous had unprecedented arrays of body, fin and head shapes and dentitions14. Moreover, all the major divisions of modern ray-finned fishes now seem to emerge from this post-Devonian radiation. The problem is that Giles et al. find that, with the notable exception of scanilepiforms and polypterids, the statistical support for proposed relationships between actinopterygians from the Carboniferous through to the early Triassic period, spanning 359 million to 247 million years ago (one-quarter of their known history), is much lower than the statistical support for the relationships between both earlier and more-recent forms of actinopterygians.
It seems probable that computed microtomography methods will provide part of the solution to this challenge, because they allow sampling from a greater range of data in both living and fossil taxa. Not only might such additional data sets sort signal from noise in groups from the Carboniferous to the early Triassic, but they might also reverse the effects of what is known as stemward slippage15, a phenomenon describing how poorly preserved fossil species might look more primitive than they actually are. Inevitably, there is a suspicion that 'victims' of such slippage now lurk in the actinopterygian stem group. Post-Devonian ray-finned fishes are probably the least-studied vertebrates in the fossil record. Some of these will harbour data of comparable quality to the Fukangichthys material; a further subset of these might belong to the actinopterygian crown group and join the still sparsely populated roots of modern clades.
As emphasized by Giles et al. and others16, the early Carboniferous is increasingly being recognized as a key episode in the history of modern vertebrate diversity. Alongside the origin of modern tetrapods, the root of the actinopterygian crown group probably lies somewhere in this interval, as does the first ascent of the ray-finned fishes towards ecological prominence. But polypterids as we now understand them emerge as a Mesozoic phenomenon, with their earlier ancestry perhaps barely distinguishable from that of other early members of the ray-finned clade. Polypterid shape looks increasingly specialized, including those remarkable — and now apparently deceptively primitive — pectoral fins2,3.
Appel, T. A. The Cuvier–Geoffroy Debate (Oxford Univ. Press, 1987).
Cuervo, R., Hernández-Martínez, R., Chimal-Monroy, J., Merchant-Larios, H. & Covarrubias, L. Proc. Natl Acad. Sci. USA 109, 3838–3843 (2012).
Wilhelm, B. C., Du, T. Y., Standen, E. M. & Larsson, H. C. E. J. Anat. 226, 511–522 (2015).
Near, T. J. et al. Evolution 68, 1014–1026 (2014).
Gardiner, B. G. & Schaeffer, B. Zool. J. Linn. Soc. 97, 135–187 (1989).
Xu, G.-H., Gao, K.-Q. & Finarelli, J. A. J. Vert. Paleontol. 34, 747–759 (2014).
Giles, S., Xu, G.-H., Near, T. J. & Friedman, M. Nature 549, 265–268 (2017).
Greenwood, P. H. in Living Fossils (eds Eldredge, N. & Stanley, S. M.) 143–147 (Springer, 1984).
Markey, M. J. & Marshall, C. R. Proc. Natl Acad. Sci. USA 104, 7134–7138 (2007).
Standen, E. M., Du, T. Y. & Larsson, H. C. E. Nature 513, 54–58 (2014).
Gayet, M., Meunier, F. J. & Werner, C. Palaeontology 45, 361–376 (2002).
Selezneva, A. A. Paleontol. J. 19, 1–6 (1985).
Sytchevskaya, E. K. in Mesozoic Fishes 2 (eds Arratia, G. & Schultze, H.-P.) 445–468 (Pfeil, 1999).
Sallan, L. C. Biol. Rev. Camb. Phil. Soc. 89, 950–971 (2014).
Sansom, R. S. & Wills, M. A. Sci. Rep. 3, 2545 (2013).
Clack, J. A. et al. Nat. Ecol. Evol. 1, 0002 (2016).
Coates, M. I. Phil. Trans. R. Soc. B 354, 435–462 (1999).