Trusting then testing Colosteus

Another short one today,
as I’m learning that you can’t trust the flat-skull reconstruction(s) of Colosteus, another basal tetrapod (Figs. 1,2) traditionally considered a temnospondyl. 

Figure 1. Traditional figure of Colosteus as a flat-headed temnospondyl.

Figure 1. Traditional figure of Colosteus as a flat-headed temnospondyl.

A little digging into the literature
reveals the holotype had a different look (Fig. 2).

Figure 2. Colosteus holotype drawing of the fossil in situ from Hook 1983 compared to the closely related Osteolepis.

Figure 2. Colosteus holotype drawing of the fossil in situ from Hook 1983 compared to the closely related Osteolepis. The torso drawing includes part of the skull. The forelimb is tiny and pink here. Four fingers were present. Note the giant dentary fang.

At present
Colosteus is a sister to the pre-tetrapod, Osteolepis. in the large reptile tree (not yet updated). Which means if Colosteus is a tetrapod (it has a vestigial forelimbs with four fingers), then tetrapods had a second sterile origin and Colosteus is part of that short radiation, not the main line leading to amphibians, reptiles, birds and mammals that ran through Acanthostega and Ichthyostega.

Hook 1983 reports on the history of this specimen:
“The first report of fossil vertebrates from the Ohio Diamond Coal Company Mines at Linton was given by J. S. Newberry in 1856; it included a brief description of Pygopterus scutellatus, a supposedly new species of paleoniscoid fish. Although no figures were supplied or specific specimens adequately described, a general picture of a heavily scaled, elongate form with flattened head and pointed snout was established.” 

“In 1869, E. D. Cope erected the batrachian genus Colosteus on the basis of Linton material lent him by Newberry. Cope cited three species, C. crassiscutatus, C. marshii and C. foveatus, and provided measurements by which the type specimens can be identified today. However, Cope later admitted (1871 a, p. 41) that in describing the type species, C. crassiscutatus, he had “overlooked” Newberry’s original account of P. scutellatus, thereby implying that both taxa were inadvertently based on the same specimen. Since Pygopterus was (and is) a valid fish genus, and the material in question was certainly not piscine (Cope, 1873), Cope recognized the proper combination of Colosteus scutellatus (Newberry, 1856) for the type species.”

Here’s a little information
the fish genus, Pygopterus (Permian – Triassic, Fig. 3).

Figure 3. Pygopterus, the Permian-Triassic fish with which Colosteus was originally confused.

Figure 3. Pygopterus, the Permian-Triassic fish with which Colosteus was originally confused.

There is much more to this story in Hook 1983.
PDF online here. If there are fossils that demonstrate that some specimens of Colosteus were indeed flat-headed, let me know. The flat-head reconstruction has to have some basis in the fossil record.

Figure 6. Subset of the large reptile tree focusing on basal tetrapods, updated with Gerrothorax.Figure 6. Subset of the large reptile tree focusing on basal tetrapods, updated with Gerrothorax.

Figure 6. Subset of the large reptile tree focusing on basal tetrapods, updated with Gerrothorax.

Cope ED 1869. Synopsis of the extinct Batrachia, Reptilia, and Aves ofNorth America. Trans. Amer. Phil. Soc., vol. 14, pp. 1-252.
Hook RW 1983. Colosteus scutellatus (Newberry), a primitive temnospondyl amphbian from the Middle Pennsylvanian of Linton, Ohio. American Museum Novitates 2770:1-41. PDF online
Newberry JS 1856. Description of several new genera and species of fossil fishes from the Carboniferous strata of Ohio. Proc. Acad. Nat. Sci. Philadelphia, vol. 8, pp. 96-100.


12 thoughts on “Trusting then testing Colosteus

  1. The three-dimensional shape of the skull of Colosteus cannot be directly inferred because all the specimens are from cannel coals (Linton, Five Points) and are thus fully flattened. However, it is very similar in a number of ways to Greererpeton, Deltaherpeton, and other colosteids for which many 3D skulls are available. Colosteus of course is represented by dozens of specimens across several major museums, and Bob Hook’s reconstruction is pretty reliable.

    However, more importantly, this is a situation where your results are very clearly incorrect, and underscores the inappropriateness of your character sample for resolving relationships of tetrapods along the tetrapod and amphibian stems. Colosteids show a plethora of important tetrapod synapomorphies placing them well within limbed tetrapods and crownward of animals like Ichthyostega and Acanthostega. This includes but is not limited to the presence of a stapes integrating a portion of the otic capsule as a footplate, the specific geometry of the stapes within the spiracular/stapedial canal, the nature of the attachments for the jaw adductor musculature, three coronoids, the absence of a sphenotic joint, closure of the buccohypophyseal canal, laterally projecting basipterygoid processes contributing to the basicranial joint, a parasphenoid that extends posterior to floor the otic region, pineal foramen posterior to the orbits, loss of the posttemporal/extrascapular bridge between the skull and pectoral girdle, absence of opercular and gular bones, presence of true digits, pentadactyly, integration of the radius into the zeugopod, a large interclavicle, a well-developed pelvis that articulates directly to sacral ribs, absence of supraneural bones in the caudal skeleton, gastal scalation, well-developed pleurocentra and intercentra, and a bunch of other really important tetrapod things that osteolepids lack.

    There is a reason your analysis fails to produce a credible phylogeny of early tetrapods.

  2. How many impossible things do you believe before breakfast?

    It’s really amazing how your analysis gives you a big red flashing warning light and it the only thing that occurs to you is to say “isn’t it pretty?”.

    In the very first line I note once again how sporadic and random your knowledge of the literature is. Colosteids haven’t been considered temnospondyls in, like, twenty years. They’re way rootward of them.

    On the reliability of specifically the snout roof reconstruction of Colosteus, have you read the Deltaherpeton paper?

    I bet Jason wrote his list just off the top of his head, without looking anything up. That’s why it’s so short! :-) Additions off the top of my head: absence of the preopercular (the whatcheeriids still have it, as does everything rootward of them), absence of pit lines on the postsplenial and the surangular (two separate losses that happen early in tetrapod evolution), presence of a very large entepicondyle, absence of denticles on most of the hyobranchial apparatus, rostral margin of the parietals caudal to orbit midlength (a feature of Carboniferous and later tetrapods), lateral-line canals mostly in open grooves (still completely enclosed in Devonian tetrapods), infraorbital canal entering the maxilla and interrupted by the nostril, identifiable frontals, small postparietals…

    I don’t believe this Jarvik-style detachment of a piece of otic capsule and its complete and perfect fusion with the stapes; I rather suppose the lateral wall of the otic capsule is incompletely ossified because the stapes is there to protect that place anyway. And the stapes of Colosteus doesn’t seem to be known; I can’t find a mention of it in Hook’s description and can’t remember any mention elsewhere. However, the stapes of its closest known relative, Greererpeton, has been described in detail. Tetrapod characters off the top of my head: lack of opercular process, lack of (ossified?) ceratohyal process, lack of mobility, lack of space between the two proximal processes (or perhaps absence of one of them) – it’s just a buttress between the otic capsule and the epipterygoid. (The whole spiracular canal may have been absent in colosteids generally, judging from the lack of a squamosal embayment, but of course speculation about soft tissues has its limits.)

    Colosteus is unique among known tetrapods (I need to check my photos of one undescribed “microsaur”, though) in having dorsal scales that look like ventral scales: both have the plesiomorphic rhombic shape that is rare even for ventral scales (gastralia) among tetrapods. Even so, the lateral scales are different, namely round, as expected from the dorsal and lateral scales of tetrapods.

    • I don’t believe this Jarvik-style detachment of a piece of otic capsule and its complete and perfect fusion with the stapes; I rather suppose the lateral wall of the otic capsule is incompletely ossified because the stapes is there to protect that place anyway.

      Thompson et al. 2012 The origin of the stapes and relationship to the otic capsule and oval window. Dev Dynamics 241: 1396-404.

      • I’ve read the paper now. It does contain the sentence “Part of the stapes is, therefore, derived from the otic capsule.”, but that reads like some kind of later insertion – several statements and even figures in the text strongly indicate otherwise as far as I can tell.

        (Possibly it was a later insertion requested by a reviewer – such things happen after all –, but obviously I have no way of telling.)

        The paper establishes beyond any doubt that the stapes in, well, amniotes (mice and apparently chickens) has a dual origin: the outer rim of the circular footplate is mesodermal like the annular ligament around it and the relevant parts of the otic capsule, while the rest of the stapes, including the center of the footplate, are of neural-crest origin like the rest of the hyoid arch (stapes = hyomandibula = epihyal, right?).

        First of all, though, I can’t see how we could tell in Greererpeton whether the area homologous to the rim of the mouse footplate is part of the stapes or the otic capsule or perhaps absent altogether.

        Second, the paper says: “In mice, retinoic acid treatment has been shown to lead to formation of a complete stapes, at a distance from the otic capsule, indicating again an independent origin (Louryan et al., 2003).” A complete stapes, including the outer rim of the footplate, I have to assume.

        Third: “From a study of 51 ears with a missing oval window, only 2 showed a normal stapes (Jahrsdoerfer, 1977).”

        So there are two human ears known to science in which the complete stapedial footplate is there, and the area that normally forms the oval window is also there, separately. Right?

        Fourth, even in the earliest stages (fig. 3 of the paper), the neural-crest-derived part of the stapes doesn’t just touch the otic capsule from the outside; it already indents that area deeply. As the text puts it: “At E12.5, the neural crest-derived (Wnt1cre positive) part of the stapedial footplate could be seen protruding into a neural crest negative region at E12.5 (arrow in Fig. 3A). At this stage, Sox9 was expressed in the developing chondrocytes of the stapes head, footplate, and otic capsule, in addition to staining the vestibule of the inner ear (Fig. 3A′). When these images were combined, it is clear the neural crest-derived footplate was already embedded in the otic capsule at this early stage and surrounded by the Sox9-positive non-neural crest-derived tissue, which will form the mesodermal part of the footplate and otic capsule (Fig. 3A”).”

        Fifth: “If the neural crest part of the stapes interacts with the mesodermal part, then specific loss of the neural crest-derived tissue might result in a defect in the mesodermal footplate. Alternatively, if the two parts develop in isolation then loss of the neural crest might result in the mesodermal portion of the footplate being suspended within the oval window surrounded by annular ligament (Fig. 4C,D). In wild-type mice, the stapes can be seen positioned in the oval window held in place by the annular ligament at E18.5 (Fig. 4C,E). In contrast, in the Wnt1cre/Dicer mutants, a break in the otic capsule in the position of the presumptive oval window was observed at E18.5, despite the lack of neural crest-derived stapes (Fig. 4D′,F). This diminutive “oval window” looked similar to the forming annular ligament seen in the wild-type (Fig. 4E,F), and did not contain any cartilage structure. A similar rudimentary oval window was observed at E14.5 and E16.5 in the mutants. The mesodermal part of the stapedial footplate, therefore, appears not to form in the absence of the main body of the stapes.” If the mesodermal part of the footplate were a piece of otic capsule that somehow detached from the rest of the otic capsule and attached to the stapes, shouldn’t we – in the mutants in question, which lack neural-crest-derived cartilage – expect something like fig. 4D, which has a big red X through it? (Interestingly, that’s almost what the batrachian/lissamphibian opercular looks like.)

        Sixth: “The oval window region does not form chondrocytes, suggesting that either it is intrinsically unable to do so, or inhibited from doing so by the surrounding tissue. In cases of otosclerosis, the oval window does ossify, indicating that it does have the potential to form skeletal tissue. An inhibition of chondrogenesis, therefore, appears likely.” So, the oval window underlying the stapes ossifies, and the complete footplate of the stapes is also present?

        Seventh: “However, our results are similar to those described in the turtle [Chelydra serpentina] after complete ablation of the neural crest, where the columella is lost and the oval window has the appearance of an irregular fissure (Toerien, 1965), indicating that neural crest is not necessary for oval window initiation. […] The position of the oval window, therefore, appears not to be dependent on the stapes, or neural crest-derived tissue. […] The signals responsible for the initial position of the oval window are, therefore, independent of interactions with the stapes. The correct development of the oval window after initiation, however, is dependent on the presence of the neural crest part of the stapes.” This conforms to the hypothesis that the oval window is a modified remnant of the lateral otic fissure, and contradicts the hypothesis that it’s a hole left behind by a detaching part of the otic capsule that attaches to the stapes.

        To me it seems that the proximal end of the neural-crest-derived part of the stapes induces ambient mesoderm to form the rim of the footplate. The rest of the mesoderm, or part of it, then forms the otic capsule. Later, the neural-crest-derived part of the stapes induces part of the lateral otic fissure not to chondrify; that results in the oval window.

        It would be great to have more comparisons available. Alas, the lungfish have completely lost the hyomandibula and the lateral otic fissure, and the embryonic development of coelacanths is not going to become accessible anytime soon. I’m surprised that no such research appears to have been done on frogs or salamanders, though.

      • I think you’re overthinking it. Another way of thinking about it is that the mesoderm recommits from an otic capsule fate to a stapes fate. It doesn’t have to have chondrified first….the mesodermal precursors are already there, and that’s what’s important.

      • But again, it’s not tissue that’s missing from the otic capsule later: you can have a complete stapes together with a nearly closed otic capsule.

    • Add characters to your matrix first, or you’ll just continue to be subject to the law of garbage in, garbage out.

      Also, read the Deltaherpeton paper before any description of Greererpeton.

    • Results, sure. But it’s important to understand that these illustrations were produced through careful study of dozens of specimens over years and years of work. There is a degree of interpretation involved, and you do need to consult the specimens themselves, because sutures and cracks can look superficially similar if you’re not careful. My experience studying these specimens and others is that Bob Hook did an excellent job of interpreting the skull of these animals and that his reconstructions align closely with the anatomy of not only to the specimens he described but also to new specimens that have appeared since his description. Cranial descriptions themselves are the results of substantial research effort, and should be treated as such. Reinterpreting the anatomy of these animals without first studying the specimens will certainly give you different phylogenetic results, but that does not necessarily mean anything.

      • Hook’s reconstruction of Colosteus does look extremely flat; I wonder if he based it to some extent on Smithson’s (1982) reconstruction of Greererpeton, which incorrectly assumed a temnospondyl-style occiput and therefore was a bit too flat (Clack 2003, description of Kyrinion).

        That said, the skull of Colosteus can’t have been much taller. In the mediolaterally flattened specimens, like the one shown in fig. 2 of this post, you can see that the quadratojugal and squamosal simply aren’t tall enough to allow a skull as tall as that of Osteolepis even if they were completely vertical.

  3. One more! Osteolepis retains a median dorsal and a median ventral scale row. Panderichthys still has them, too, and so does Tiktaalik. Colosteus does not; no known tetrapod does.

    Another: fucking cosmine. Plesiomorphically, osteichthyan scales (and also the other dermal bones in sarcopterygians!) are covered by enamel, followed underneath by dentine; that’s retained by Osteolepis. Eusthenopteron and everything crownward of it, including Colosteus, have lost the whole tooth-tissue layer and retain only the bone-tissue one. If you find a shiny scale in Carboniferous or Permian rock, it may be megalichthyid or actinopterygian, but it’s not tetrapod.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.