Michael I. Coates

Discovery of an unusually intact and ancient fossil fish provides further evidence that the search for modern vertebrate origins requires breaking out of the Devonian and into the preceding period.

As a rule, the earliest fossils of living groups tend to be scrappy, and such fragments lend themselves to contentious interpretations. For ‘bony fishes’, Osteichthyes — the division of vertebrates that includes everything from humans to halibut — the record of articulated fossils peters out within the Lower Devonian¹, some 400 million years ago. Earlier stretches of osteichthyan history are littered with fossil detritus, such as isolated teeth and scales. In certain instances, bits and pieces have been reassembled into conjectural species^2–4, some of which have surprising combinations of anatomical features². On page 469 of this issue, Zhu et al.⁵ introduce a fresh — albeit long-dead — fish into this poorly resolved patch of vertebrate evolution. Crucially, this piscine offshoot of our own distant past is both unusually intact and exceptionally old.

So what kind of fish is it? A summary of vertebrate diversity helps to make sense of the answer. Of the 51,000 or more living species of vertebrates, 99.9% have jaws: these are the gnathostomes. Gnathostomes include the bony Osteichthyes and the cartilaginous Chondrichthyes. Chondrichthyes (sharks, rays and chimaeras) account for only 2% of gnathostome species, the Osteichthyes accounting for the other 98%. Around half of the Osteichthyes are Actinopterygii, or ‘ray-finned fishes’, and half are Sarcopterygii, or ‘lobe-finned fishes’. Actinopterygians include some 28,000 species, from zebrafish to bichirs, and living sarcopterygian fishes are limited to three genera of lungfishes and one coelacanth. Land-dwelling tetrapods constitute the remaining majority of sarcopterygians.

Thus far, the origins of these major divisions of today’s gnathostomes can be traced back to the Devonian, between 416 million and 359 million years ago, the Age of Fishes. Fossils that are clearly chondrichthyan are known from around 400 million to 405 million years ago⁶, but we have little idea as to whether these belong within the living radiation, the ‘crown group’, or represent side branches of their common ancestry, the ‘stem group’. As for osteichthyans, although it is agreed that fossils from the earliest Devonian^2,7 belong within the crown, osteichthyan fragments of less-certain affinity are also known from the Late Silurian³, 423 million to 416 million years ago.

But there’s more to this story, because the question of gnathostome origins also involves a pair of extinct groups of gnathostomes known to appear earlier in the geological record, the placoderms and acanthodians1. Importantly, recent analyses⁸ have begun to reveal new relationships between early vertebrates, in which acanthodians and placoderms are scattered among the early divisions of gnathostome evolution; acanthodians, in particular, are cropping up on chondrichthyan and osteichthyan stem groups. The straightforward message is that the origin of modern gnathostomes is not a Devonian phenomenon, after all. The basal divergence between osteichthyans and chondrichthyans occurred somewhat earlier.

This, then, is the context within which to place Guiyu oneiros, the new species of early osteichthyan named and described by Zhu et al.⁵. Preserved in 418-million-year-old limestone in what is now southern China, the fossils of Guiyu show the skeletal anatomy of a small sarcopterygian, around 33 centimetres long. The very fact that Guiyu can be identified as a sarcopterygian provides further and arguably clinching evidence that a whole series of major branching events within the gnathostome crown group must have taken place well before the end of the Silurian.

Like any other fossil, Guiyu is a mixture of primitive and advanced features. With regard to its anatomical completeness, Guiyu provides exceptional corroboration for the decidedly odd reconstruction of the early osteichthyan genus, Psarolepis². Cobbled together from a disparate set of fossils, the incongruent suite of features⁹ displayed by Psarolepis has been viewed with caution. Now, it turns out to be thoroughly plausible. Like Psarolepis and other sarcopterygian fishes (including Latimeria, the living coelacanth), the braincase of Guiyu is divided into separate front and rear units. Like Psarolepis, the cheek bones resemble those of early actinopterygians. Like Psarolepis and many other early gnathostomes¹, including at least one chondrichthyan⁶, the shoulder girdle bears a spine in front of the pectoral fin. Similarly, the dorsal-fin spine and anterior spinebearing plate of Guiyu are probably primitive. These are all widespread features of early gnathostomes, and seeing such characteristics in Guiyu provides a first glimpse of the sequential order of anatomical changes that resulted in the standard set of sarcopterygian traits.

The evolutionary tree proposed by Zhu et al.⁵ (see Fig. 5 on page 473) adds to a growing set of analyses of early osteichthyan and gnathostome interrelationships^8,10. Uncertainties still surround the branching pattern of non-osteichthyans, but the addition of Guiyu to the cast of early fishes does not change the basic pattern of interrelationships among early osteichthyans. Instead, it adds support to notable consistencies in the emerging pattern of sarcopterygian evolution, including the clustering of some of the earliest-known examples to form an as-yet unnamed group.

Finally, what does the conclusion that Guiyu is unequivocally sarcopterygian imply? On the whole, early fossils are thought to be unreliable as minimum-date markers of evolutionary branching events¹¹, because they are less complete and/or lack the full anatomical signature of the group to which they are assigned. Guiyu might be an exception that proves the rule, for it provides a new and exceptionally reliable earliest fossil marker for a major split in vertebrate evolution. By pushing a whole series of branching points in gnathostome evolution out of the Devonian and into the Silurian, the discovery of Guiyu also signals that a significant part of early vertebrate evolution is unknown (Fig. 1).

The new shape of the gnathostome tree shows that early sarcopterygians, as well as Actinopterygians and chondrichthyans, ought to be turning up in Silurian sediments. But where are they? Modern fish groups have Silurian roots, but these fishes are consistently absent from existing scenarios of Silurian life. The discovery of Guiyu should provoke a rash of new fieldwork and a fresh look at existing collections of pre-Devonian fossils.

Michael I. Coates is in the Department of Organismal Biology and Anatomy, University of Chicago, Chicago, Illinois 60637, USA.

1. Janvier, P. Early Vertebrates (Oxford Univ. Press, 1996).

2. Zhu, M. et al. Nature 397, 607–610 (1999).

3. Botella, H. et al. Nature 448, 583–586 (2007).

4. Basden, A. M. & Young, G. C. J. Vert. Paleontol. 21, 754–766 (2001).

5. Zhu, M. et al. Nature 458, 469–474 (2009).

6. Miller, R. F. et al. Nature 425, 501–504 (2003).

7. Zhu, M. et al. Nature 441, 77–80 (2006).

8. Brazeau, M. D. Nature 457, 305–308 (2009).

9. Ahlberg, P. E. Nature 397, 564–565 (1999).

10. Friedman, M. J. Syst. Palaeontol. 5, 289–343 (2007).

11. Donoghue, P. C. J. & Benton, M. J. Trends Ecol. Evol. 22, 424–431 (2007).

For details click ：The oldest articulated osteichthyan reveals mosaic gnathostome characters