Field of Science

Time For Teeth (Taxon of the Week: Polygnathus)

Lindström's (1974) hypothetical reconstruction of the then-unknown conodont animal as a barrel-shaped floater, with radially arranged conodont elements providing protection from predators dorsally and support for feeding tentacles ventrally.

Conodonts are among the iconic fossils of the Palaeozoic. Minute (in the millimetre size range) but extremely abundant, conodont elements* are tooth-like in appearance. The earliest forms were simple and fang-like; later forms were often blade-like with a median row of teeth. Their abundance and variety mean that conodonts are widely used in biostratigraphy, but for many years the identity of the animal they came from was unknown - whatever it was, it appeared to possess no other hard parts that would normally be preserved. It wasn't until the 1980s that the first unequivocal conodonts with preserved soft parts were discovered, revealing them to be stem- or basal vertebrates** (Sweet & Donoghue, 2001). Each of the conodont animals had a number of conodont elements arranged around the mouth and pharynx. Slender-pointed elements towards the front of the mouth would have seized or filtered prey, while many conodonts also possessed more robust elements further back in the pharynx to grind up their food. The figure below from Dzik (1991) gives a good idea of how it would have all worked, even if the result does look a bit like a carnivorous sock puppet (Dzik's arrangement of the elements has also since been superceeded - see Purnell & Donoghue, 1997, for details). Those full-body fossils of conodonts that have been identified to date are eel- or lamprey-like, but it is worth keeping in mind that only two species of this very speciose lineage are known from such remains and we may not be seeing a proper representation of conodont diversity.

*Before the nature of conodonts was understood, most authors restricted the name to the fossils themselves; the then-hypothetical animal that produced these structures was referred to as a "conodontophore". Since the current identification of conodonts has been accepted, this distinction has been abandoned.

**Conodonts had been found in association with soft body parts before, but the animals concerned are now agreed to have been predators or scavengers of conodonts (with conodont elements in their gut as a result) rather than the conodont animals themselves.

Conodont head in retroventral view as reconstructed by Dzik (1991), with mouth open to show the grasping elements in front and back of head removed to show the grinding elements in back.

Polygnathus has been recognised as one of the largest of conodont genera - some 545 Early Devonian to Early Carboniferous species and subspecies have been assigned to it over the years (Weddige, 2005). Polygnathus belonged to the conodont order Ozarkodinida, and would have had an apparatus of toothed elements similar to that shown below (not Polygnathus, but another ozarkodinidan genus). The lower saw-like S elements at the front of the mouth would have been the initial graspers; the act of opening the mouth would have rotated the curved upper M elements forward, and their rotating back as the mouth closed would have probably drawn the prey in; and the two pairs of large P elements in the back would have sliced and diced the prey.

Reconstructed model of the apparatus of the ozarkodinid Idiognathodus in lateral view, from Purnell & Donoghue (1997).

The Early Devonian members of Polygnathus were recently revised by Bardashev et al. (2002) in what I can only describe as one of the most taxonomically incredible papers it has ever been my misfortune to read. In the early days of conodont taxonomy, working purely from dissociated elements, different elements were treated as taxonomically separate entities. As the recognition developed that a single individual conodont animal would have possessed a number of differently formed elements (something that happened even before the discovery of conodont soft-body fossils as researchers noted that certain element types were always found in association, while specimens were occasionally found in which normally separate elements had become fused together), the older independent element taxonomy was replaced by a multi-element taxonomy based on the apparatus as a whole*. Bardashev et al. (2002), however, base their classification solely on the Pa or P1 element, the large posteriormost element in the model above. All other elements, they seem to claim, are useless for distinguishing taxa (which could be a problem for dealing with basal conodonts that don't have P elements).

*At least ideally. In practice, of course, there are still a large number of cases in which the correct element associations cannot yet be reliably identified.

On the basis of Pa morphology, Bardashev et al. divide species of Polygnathus between six genera in two families - and this is where things really start to go down the rabbit hole. Members of the family Polygnathidae are divided between the temporally successive families Eognathodidae, Eopolygnathidae and Polygnathidae. Eopolygnathidae are derived from Eognathodidae and Polygnathidae from Eopolygnathidae. Now, the use of paraphyletic taxa is nothing unusual in micropalaeontology. But explicitly polyphyletic taxa? In the phylogeny presented by Bardashev et al., Eognathodidae gave rise to Eopolygnathidae on two separate occasions, with Eoctenopolygnathus descended from a separate group of eognathodids from Eocostapolygnathus and Eolinguipolygnathus (note also that there is no genus 'Eopolygnathus', so 'Eopolygnathidae' is an invalid name under the ICZN). After that, Polygnathidae derives from 'Eopolygnathidae' eleven times - two separate origins of Ctenopolygnathus within Eoctenopolygnathus, four origins of Costapolygnathus from Eocostapolygnathus, five of Linguipolygnathus from Eolinguipolygnathus (the authors refer to these multiple origins as representing common 'trends' between the lineages). Bardashev et al. also name the type species of the new genus Costapolygnathus as Polygnathus dubius, which happens to be the type species of Polygnathus (a point that Bardashev et al. had commented upon themselves earlier in the paper). There are also cases where the type specimens of 'undiagnostic' species are assigned to new species named by Bardashev et al. - surely, if you can identify them to a species, they can't be undiagnostic?

Posterior and anterior views of the Pa element of Polygnathus costatus partitus. Photo from

Bardashev et al.'s (2002) reclassification was criticised and rejected by Mawson & Talent (2003), who maintained that because it only covered Early Devonian taxa, it created a strong disconnect in apparent diversity between Early and Late Devonian. This criticism, I must say, is unfair - all revisions have to start somewhere, and to demand an 'all or nothing at all' approach in such cases would be to effectively prevent much possibility of large taxonomic groups being revised at all. Potentially more problematic (but unfortunately not supported with specific examples) is Mawson & Talent's implication that some of the new 'species' recognised by Bardashev et al. are in fact variants of other species and not phylogenetically distinct entities.

Bardashev & Weddige (2003) published a brief note in which they corrected the objective synonymy of Polygnathus and Costapolygnathus by publishing a new genus Eucostapolygnathus that they said "includes the same species as Costapolygnathus - except the species dubius". In a reply to Mawson & Talent's comments, Weddige (2005) defended Bardashev et al.'s (2002) use of a high number of taxa on the basis that the latter had been a 'pure form-taxonomic study'. Or, more extensively:

The genus subdivision proposed by BARDASHEV, WEDDIGE & ZIEGLER (2002) might be regarded as a subgeneric subdivision. In form-taxonomy, however, and the paper represents a pure form-taxonomic study, subgenera are not in usage. Because of the pure form-taxonomy, moreover, resp. because of a more or less subgeneric level of the proposed subdivision, a multielement reference, e. g. by suspect statistics, is not needed, for the first. Thus, a distinctive serious discussion has to focus on (form-) taxonomic characters, i. e. the valuation and order of the diagnostic characters as they are used for the generic subdivision by BARDASHEV et al.. Admittedly, a broadly splitted form spectrum, often including revolutionary ideas, is a hard diet. On the other hand, a well known unchanged form spectrum is a usual and therefore easy diet that, moreover, becomes much easier to digest when the spectrum, or parts of it, is furthermore lumped. The differentiation in “splitters” and “lumpers” is an inadequate simplification -- since the study by BARDASHEV et al. is not only a splitting because of different new taxa, it has rather more the character of a synthesis because of its search for phylogenetic lines by which single species were “lumped”). Thus, the study is a lumping on a quality level, higher than a taxonomic lumping that resigns to differentiate and searches for a conservative comfortable easy diet. Conservatives bloc progress, that is their job – and it would be a total misunderstanding that a SDS commission or a Working Party is entitled to condemn per joint decision (that could not be the target of a discussion!).

So in reply to accusation of being splitters, Weddige replies that no, they were lumpers, but his definition of 'lumping' can only be described as an Inigo Montoya moment. There is also the problem that Bardashev et al. was self-evidently not a purely form-taxonomy study. Form taxa are those based on morphological distinctions only that cannot be confirmed as phylogenetically distinct units - but Bardashev et al. (2002) presented their readers with no less than nine representations of preferred phylogenetic hypotheses, as well as specifically commenting on the descent of every one of the taxa they described. If these were only 'form taxa', then those 'lineages' are completely meaningless, and you, my friend, have just been treated to seventy-seven pages of intellectual masturbation.


Bardashev, I., & K. Weddige. 2003. The invalid genus name Costapolygnathus Bardashev, Weddige & Ziegler 2002 and the new conodont genus Eucostapolygnathus. Senckenbergiana Lethaea 83 (1-2): 1-2.

Bardashev, I. A., K. Weddige & W. Ziegler. 2002. The phylomorphogenesis of some Early Devonian platform conodonts. Senckenbergiana Lethaea 82 (2): 375-451.

Dzik, J. 1991. Evolution of oral apparatuses in the conodont chordates. Acta Palaeontologica Polonica 36 (3): 265-323.

Lindström, M. 1974. The conodont apparatus as a food-gathering mechanism. Palaeontology 17 (4): 729-744.

Mawson, R., & J. A. Talent. 2003. Conodont faunas from sequences on or marginal to the Anakie Inlier (Central Queensland, Australia) in relation to Devonian transgressions. Bulletin of Geosciences 78 (4): 335-358.

Purnell, M. A., & P. C. J. Donoghue. 1997. Architecture and functional morphology of the skeletal apparatus of ozarkodinid conodonts. Philosophical Transactions of the Royal Society of London B 352: 1545-1564.

Sweet, W. C., & P. C. J. Donoghue. 2001. Conodonts: past, present, future. Journal of Paleontology 75 (6): 1174-1184.

Weddige, K. 2005. Contra Ruth Mawson’s critizising Bardashev, Weddige & Ziegler 2002, e.g. in SDS Newsletters 20 (2004). Subcommission on Devonian Stratigraphy Newsletter 21: 51-52.


  1. 3WTF?!

    (That's one WTF for the animals and two for the taxonomic treatment thereof.)

  2. (nose bleeding)

    Look what you did! Now my brain is coming out! Seriously, that was a waltz through taxonomic hell. I had to read a few paragraphs twice to make sure I understood what was going on.

    On a somewhat related topic, what DID conodonts look like?

  3. Conodonts were around for something like 300 my (Cambrian to Jurassic). So, probably, their all-time number of species was... in the thousands? Of which two are known from good body fossils? (Any idea what proportion are known from natural assemblages of elements?)
    I think this is a situation in which it might make sense to detach taxonomy from phylogenetic hypotheses.
    If I were made dictator of the ICZN, I think I would put out a ukase to the effect that (i) all uses of binomials (and higher level group names) be preceded by a distinctive symbol, (ii) that the usual rules of nomnclatural etiquette are suspended, and, in particular, (ii-a) no publication of a phylogenetic hypothesis will justify ANY change in established nomenclature, and (iii) this is to continue until such time as the Politburo of the Pander Society (= Conodont Specialists' Organization, named for a 19th C conodontologist) decides that the evidential situation has improved to the point that the normal rules can sensibly be applied.

  4. Palaeos has another view on the apparatuses (apparati?) Which I think is a rotated version of your third image, with colors.

    I always thought conodonts were kind of cool, with mouths like Predators but more evil.

  5. Zach, as said in the post, those conodonts that we do have body fossils for looked rather like small lampreys. But that's only two species out of, as Allen said, thousands of species over hundreds of millions of years.

    I don't know how many natural assemblages are known, but again I suspect that it's very low compared to the overall number of species. Also, I expect that 'natural assemblages' would run the gamut from loose associations of element types within a single formation to actual in situ nodulated apparatuses. Even relatively non-disarticulated apparatuses would not be perfectly easy to interpret, because the original three-dimensional arrangement of elements is bound to have collapsed somewhat. What's even more of a problem is that body fossils and natural assemblages are not evenly spread phylogenetically - a number of conodont orders (particularly the basal families) are probably not known from natural assemblages at all.

    The reconstruction at is the Idiognathodus apparatus, though it's shown more splayed out than in the model I showed. Note that, contrary to the Dzik reconstruction I reproduced, the mouth wouldn't have opened along the underside, because of the presence of the median S0 element at the front of the ventral midline (Dzik had gotten the position of the S0 reversed, and placed it at the back rather than the front of the S elements; that 0 should be in subscript, by the way, but Blogger comments apparently don't allow that). Even considering the outdated positions of the elements, I'd still recommend looking through the reconstructions in Dzik (1991) to get an idea of the diversity of conodont apparatuses and the implied ecological diversity of conodonts. To my mind, the ozarkodinid apparatus like that found in Idiognathodus and Polygnathus looks more suited for active grasping of prey than that found in other orders with more slender, simple elements which probably captured prey by encagement behind the teeth instead. The really nasty looking jobs are chirognathids with crescent-shaped elements armed with massive spikes along the broad edge.


Markup Key:
- <b>bold</b> = bold
- <i>italic</i> = italic
- <a href="">FoS</a> = FoS