News & Views | Published:

Catching the first fish

Nature volume 402, pages 2122 (04 November 1999) | Download Citation


Most major animal groups appear suddenly in the fossil record 550 million years ago, but vertebrates have been absent from this ‘Big Bang’ of life. Two fish-like animals from Early Cambrian rocks now fill this gap.

The search for the deep past of the vertebrates has been a pet subject of palaeontologists during the past decade. Traditionally, the fossil record of fishes and their descendants is good from the present to the Early Silurian period, about 430 million years (Myr) ago, then very poor in the Ordovician period (until about 480 Myr ago), and totally lacking or controversial before that, in the earliest Ordovician and Cambrian periods (from 480 to 550 Myr ago) and earlier. On page 42 of this issue, Shu et al.1 report the discovery of two fish-like fossils from Chengjiang in Yunnan, China, that are probably the long-awaited Early Cambrian vertebrates.

British and Australian palaeontologists have found vertebrate remains from the beginning of the Ordovician2. Ordovician vertebrates have been shown to be far more diverse than was previously thought, consisting of a mixture of primitive, jawless groups and comparatively advanced groups with jaws3. However, all the Cambrian fossils thought to be vertebrates are from the Late Cambrian, and are controversial. They are chiefly the euconodonts (often called conodonts), a group long known from isolated denticles (small, tooth-like fossils) of debated derivation, but now known from complete, eel-like animals that are probably vertebrates4. Some palaeontologists dispute this, although they accept that euconodonts may be related to vertebrates5,6.

Other putative Cambrian vertebrates are represented by small carapace fragments, whose microstructure recalls a group of jawless vertebrates with bony armour — the so-called ostracoderms — that lived from the Ordovician to the Devonian (Fig. 1). Again, some palaeontologists regard these fragments as belonging to true vertebrates7, whereas others think they are carapace fragments from arthropods related to horseshoe crabs8. The exploration of the deep past of the vertebrates, before the Silurian period, is not for the faint-hearted, requiring the painstaking study of sketchy remains, with the constant danger of misinterpretation.

Figure 1: The current phylogenetic tree of the major living and fossil vertebrate groups.
Figure 1

This tree suggests that bone appeared relatively late, after the divergence of hagfishes and lampreys — the two living groups of jawless vertebrates — because the fossil, armoured, jawless vertebrates known as ostracoderms (in red), are more closely related to jawed vertebrates than to hagfishes and lampreys. This implies that, during a long portion of the Early Cambrian or before, vertebrates had no mineralized tissues and could only be fossilized in exceptional circumstances. The two new fossils (asterisks) discovered by Shu et al.1 in Chengjiang, China, display such exquisitely preserved details of their soft anatomy that there is little doubt that they are vertebrates, although their phylogenetic position remains speculative. bp, before present.

In fact, the bone-like fragments or denticles from the Cambrian might not lead to the ‘root’ of the vertebrates. The currently accepted phylogenetic trees showing the evolutionary relationships of fossil and living vertebrates suggest that mineralized tissues are a comparatively recent invention, and that the jawless ostracoderms are more closely related to the jawed vertebrates than to either lampreys or hagfish, the two living, jawless vertebrate groups, which have no bones or teeth (Fig. 1)9. This means that the earliest history of the vertebrates can only be documented by fossils formed under certain conditions, where the imprint of soft tissues has been preserved. For the Cambrian period, there are two famous sites where an exceptional preservation of the soft tissues has occurred. One is Burgess, in Canada, and the other is Chengjiang. Fossils resembling chordates (the large animal group that includes the vertebrates) or vertebrates have been recorded from both, such as Pikaia in Burgess, and Yunnanozoon and Cathaymyrus in Chengjiang10,11,12. Although probably close relatives of the vertebrates, none of these fossils looks entirely familiar to a vertebrate specialist.

The two new fossils described by Shu et al.1 from Chengjiang are the most convincing Early Cambrian vertebrates ever found. With their zigzag-shaped muscle blocks, relatively complex and presumably cartilaginous skull, gill arches, heart, and fin supports, these fish resemble the larvae of living lampreys (Box 1, overleaf). One detail, however, is at odds with what we know about vertebrates: the peculiar, forward tilt of the endoskeletal rays of the dorsal fin, which contrasts with the backward tilt of these fin rays in most fishes. Whether this is due to distortion or not awaits confirmation. Surprisingly, these animals seem to have long, ribbon-shaped, paired fins, which were thought to have appeared later in vertebrate history9, after the divergence of lampreys.

Box 1: Box 1: How do you recognize a vertebrate?

All living vertebrates share a relatively small number of characters found in no other group, which are assumed to have been inherited from a common vertebrate ancestor. All these characters are made of initially non-mineralized tissues (including cartilage) and are thus unlikely to be preserved in very early fossil vertebrates. The two Early Cambrian fossils from China display some vertebrate characters directly, whereas others can be inferred. Both show a dense imprint of what is probably a cartilaginous skull, one of the main vertebrate features. The gill skeleton may be indirect evidence for a neural crest, the main vertebrate character. The fin rays (radials) in one of the two forms and a large heart, lying behind the gills and possibly enclosed in a pericard, are also unique vertebrate characters. Chevron-shaped muscle blocks also occur in cephalochordates (the closest living relatives of vertebrates), but have a more complex zigzag shape in vertebrates, as can be clearly seen in at least one of the fossils.

Shu et al. have analysed the phylogenetic position of these two Cambrian fossils in the framework of previous analyses of vertebrate phylogeny. Strangely, one of the two species seems to be more closely related to lampreys than to any other vertebrate group, whereas the other appears as the sister-group to all other vertebrates but hagfishes (Fig. 1). The support for this result is admittedly tenuous; morphology-based data matrices for fossil vertebrate groups include a large number of question marks, because of missing data in fossils and uncertainties as to the interpretation of some of the soft-tissue characters seen in imprints. This is true for the two Cambrian fish. Nevertheless, the result makes broad sense, because it suggests that lampreys and their fossil relatives had diverged already in the Cambrian, as predicted by previous phylogenies.

Lacking fossils, palaeontologists and anatomists have often tried to imagine the earliest vertebrates. The two new fossils from China resemble some of these imaginary reconstructions, but are completely at odds with others. Were theoretical reconstructions of the ancestral vertebrate partly accurate, or do we pay special attention to fossils that resemble the reconstructions in our minds?


  1. 1.

    et al. Nature 402, 42–46 (1999).

  2. 2.

    J. Vert. Paleontol. 17, 1–25 (1997).

  3. 3.

    , & Nature 379, 628–630 (1996).

  4. 4.

    , , , & Phil. Trans. R. Soc. Lond. B 340, 405–421 (1993).

  5. 5.

    , & Lethaia 29, 317–328 (1997).

  6. 6.

    Mod. Geol. 20, 275–285 (1996).

  7. 7.

    , & Nature 380, 702–704 (1996).

  8. 8.

    Rapp. Grønland Geol. Undersøg. 91, 111–115 (1979).

  9. 9.

    Palaeontology 39, 259–287 (1996).

  10. 10.

    in Atlas of the Burgess Shale (ed. Conway Morris, S.) 26 (Palaeontol. Assoc., London, 1982).

  11. 11.

    et al. Nature 377, 720–722 (1995).

  12. 12.

    , & Nature 385, 865–868 (1996).

Download references

Author information


  1. Philippe Janvier is at the Laboratoire de Paléontologie (UMR 8569 du CNRS), Muséum National d'Histoire Naturelle, 8 rue Buffon, 75005 Paris, France.

    • Philippe Janvier


  1. Search for Philippe Janvier in:

About this article

Publication history



Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing