Category Archives: Placoderms

How Many Species of Titanichthys are there?

One thing I’ve been puzzled by ever since I first started learning about Titanichthys is that there are currently five named species from the Appalachian Basin alone (T. agassizi, T. clarki, T. attenatus, T. rectus, and T. hussakofi). There are also two more species described, one from Poland (T. kozlowskii) and one from Morocco (T. termieri). However, only three of the species T. agassizi, T. clarki, and T. termieri are described from even remotely complete remains with the others being nomen dubia from the early 20th century. While nobody has done any work to test the theory of filter-feeding in this genus from what I’ve seen all the evidence points towards that position so I’ll stick with that assumption of life habit for this post. While there are quite a few modern-day mega-planktivores (basking shark, whale shark, megamouth shark, all the baleen whales) it seems suspicious that there were five species of a single filter-feeding genus in the relatively small Appalachian Basin. Especially given that most modern mega-planktivores are very widespread and easily distinguished from gross morphology that I would think would be visible in the fossil record. So are those five species really valid and if not how many species of Titanichthys are there really?

Figure 1. Modern giant filter-feeders. Whale shark (left, CCBY-SA by Arturo de Frias Marques). Megamouth shark (right, CCBY-SA by FLMNH Ichthyology).

            The main difference between the two well-known species from the Appalachian Basin (T. agassizi and T. clarki) is the form of the lower jaw (inferognathal). The image below shows the two morphs from Newberry’s 1889 monograph quite well. Essentially, agassizi has a thin rod-like inferognathal without much of a ventral bend at the anterior edge. In contrast, clarki has a more robust inferognathal with a developed posterior blade and a distinct ventral bend at the anterior edge. While a complete inferognathal of T. termieri has not been published a casting company appears to have a complete specimen and the Wyoming Dinosaur Center also appears to have a specimen with jaws. Assuming these specimens do represent Moroccan specimens this species inferognathal corresponds strongly to the clarki morph.


Figure 2. Inferognathal of T. agassizi (top) and T. clarki (bottom). From Newberry 1889 (Public Domain). Note that the inferognathals are upside down in this plate because Newberry thought the anterior portion projected upward and the groove faced dorsally.

            But there is more variance in remains attributed to Titanichthys than the original descriptions imply. There are inferognathals that are simply long rods with no apparent anterior bend, essentially an extreme agassizi form (Fig. 3), which then grades towards a more robust form, the clarki type, and finally the most extreme morph has a well-developed posterior blade with an extremely exaggerated anterior curvature that does not correspond to any named or published material that I’ve dubbed the “tusk” morph (Fig.4). Beyond the inferognathals of the well-known species the rest of skeleton is not actually that different. There are some minor differences in the shapes of plates such as the pineal and paranuchal between species but the holotype and paratype of T. termieri have differently shaped paranuchals. And although we don’t have many (published) samples of North American taxa to look at intraspecific variability studies from the extremely impressive Gogo Reef fauna have shown lots of variability in plate shapes. The combination of gradation between inferognathal shapes, lack of other distinguishing characters, and the known degree of intraspecific variability are, in my opinion, a strong argument that Titanichthys has been taxonomically oversplit maybe representing only two or maybe even just one species.


Figure 3. Extreme T. agassizi rod-like morph, left inferognathal from Dunkle & Bungart 1942.


Figure 4. Extreme T. sp. tusk morph, CMNH 7412. Partial left inferognathal. CCBY-Image by author

If we accept that some of the variation in inferognathal shape is within species what might explain the range of forms? There are a couple of possibilities including 1) anagnetic evolution, 2) ontogeny, or 3) intraspecific variability/plasticity.

Anagenetic Evolution

By anagenetic evolution I mean that the range of inferognathal shapes could be a single lineage evolving over time with the T. agassizi form transitioning to the clarki form and then finally the ‘tusk’ form or vice versa. If this was true we should not find any of the different forms to overlap each other in time, at least we shouldn’t find extreme agassizi in the same strata as clarki morphs. Unfortunately, there has not been much published work on stratigraphy within the Cleveland Shale where most Titanichthys material is from so we aren’t able to resolve this as a possibility currently. I suspect people from the Cleveland Museum or other who have collected material from the Cleveland Shale would have some sense of this.



Another possibility is that the different morphs represent different stages of growth with the jaw changing shape as the animal grew. It wouldn’t be the first time several species of placoderm were found to be ontogenetic morphs. The famous Dunkleosteus used to contain half a dozen species, all in the Appalachian Basin based on inferoganthal shape, but were found to all fall into a simple growth curve and thus synonomized (Hlavin 1976). An ontogenetic explanation would predict that inferognathal forms would be correlated with size, small inferognathals would all be one shape and all large ones a different shape. Now this hypothesis, I think, can be rejected from the collections in the Cleveland Museum because all three morphs (agassizi, clarki, and ‘tusk’) occur at large sizes. It is possible that ontogeny could still explain some variation if there really are several valid species, but the inferognathals in collections really need to be measured to detect this.

Intraspecific Variability/Plasticity

I think this is the most likely explanation for much of the variation in inferognathals of Titanichthys. But this is also the hardest to test because there isn’t an obvious way to test it. Anagenetic evolution and ontogeny both have specific predictions but this explanation does not. To test this hypothesis we would need to have much better descriptions of the species of Titanichthys with many examples of each. Then examine if there are consistent differences between and gaps in morphology between specimens, and whether they correspond to previously designated species. Basically, you really need a monograph of the genus to sort this out.


So in the end I don’t know how many species of Titanichthys there really are but I have serious doubts that there are five (or six including the ‘tusk’ morph). Based on my time in the collections of the Cleveland Museum of Natural History I’d guess there are between one and three species. It might be that T. agassizi, T. clarki, and the ‘tusk’ morph are all valid but I’m very skeptical still. It is because of all this uncertainty in species identification that I left the recently described specimen of Titanichthys in open nomenclature as Titanichthys cf. clarki (Boyle & Ryan 2017). It is worth noting that the new specimen shares characteristics with all three well-known species, including the Moroccan one. In fact, Robert Carr (2009) has suggested that T. termieri occurs in the Appalachian Basin and conversely that Dunkleosteus terrelli occurs in Morocco. This is from a conference abstract several years ago, but it does suggest that material to resolve questions about Titanichthys are already available in museum collections and while I would love to get around to working on this issue myself it seems unlikely any time in the near future. So if anyone wants to jump on this go for it, placoderms need some more attention!



Boyle, J. and M.J. Ryan. 2017. New information on Titanichthys (Placodermi, Arthrodira) from the Cleveland Shale (Upper Devonian) of Ohio, USA. Journal of Paleontology 91:318-336.

Carr, R.K. 2009. A big fish story: new links between the Appalachian Basin and Morocco in the Late Devonian. Cincinatti Museum Center Scientific Contributions 3:204.

Dunkle, D.H. and P.A. Bungart. 1942a. The infero-gnathal plates of Titanichthys. Scientific Publication of the Cleveland Museum of Natural History 8:49-59.

Hlavin, W.J. 1976. Biostratigraphy of the Late Devonian black shales on the cratonal margin of the Appalachian geosyncline, Unpublished D. Phil. Thesis, Boston University.

Newberry, J.S. 1889. Paleozoic Fishes of North America. US Geological Survey: Monographs 16.

New Data on Titanichthys At Last

Since the last time I posted here I’ve managed to have two new papers published. One in Paleobiology on graptolite extinction risk (Boyle et al. 2017), which I’ll summarize in another post, and one in the Journal of Paleontology (Boyle & Ryan 2017) on the placoderm Titanichthys. The work on Titanichthys was part of my undergraduate capstone research that started waaaay back in 2010 with Dr. Ryan at the Cleveland Museum casually remarking that a new nearly complete skeleton had just been prepared and somebody needed to describe it. So today I’ll be summarizing the results of that paper and where I’d like to get more done on this taxon in the future.

To start, a little background on Titanichthys. It is an old taxon named in 1885 by John Strong Newberry (an interesting man for a number of reasons worthy of his own post) and also a very large animal. Ballpark estimates are a total length >5 meters, making it larger than its more famous contemporary Dunkleosteus and possibly the largest vertebrate animal that had evolved up to that time. The genus is confined to the latest Devonian (Frasnian?-Fammennian) making it one of the last arthrodire placoderms before the entire group perished at the end of the Devonian. The remains of Titanichthys are the second most common taxon after Dunkleosteus but are commonly isolated and fragmentary making its overall morphology somewhat mysterious. Of the 7 currently recognized species only 3 (T. agassiziT. clarki, and T. termieri) are known from at least partially complete specimens. While the description of T. termieri (Lehman 1956) is pretty good by modern standards the other two well-known species were described in the late 19th or earliest 20th century without many modern morphologic conventions and without designation of a holotype. In fact, the effective “holotype” of T. clarki described by Eastman (1907) is a composite of several individuals and has been partially restored with plaster making interpretation of the figures somewhat dubious.

So prior to this new paper the most recent publication on Titanichthys was over 50 years old and we lacked definitive reference material to work from. The new specimen of Titanichthys (CMNH50319) is unique because it is articulated and nearly complete, only lacking parts of the thoracic shield and the ventral shield (Fig. 1). This has allowed for the first description of five new plates (rostral, postmarginal, postsuborbital, submarginal, and posterior superognthal; see Placoderm primer for plate info) and updated descriptions for every other plate that’s been previously described for the genus, with exception of the anterior dorsal lateral. So now we have a very nice specimen to compare new material to and could start looking at the range of variation in morphology. The posterior superognathal was perhaps the most interesting because it is tiny relative to the lower jaw (inferognathal) especially when compared to other arthrodires. This, along with the animals overall large size, the flattened body shape, and elongated lower jaws without anything resembling teeth are all strongly convergent with large filter-feeding vertebrates that are known today such as the Whale Shark, Basking Shark, and Megamouth Shark as well as the fossil teleost fishes Pachycormids. The suggestion that Titanichthys was a filter-feeder wasn’t a novel idea but the new data does lend a lot more support to this hypothesis which could now be tested explicitly.

Figure 2_2col

Figure 1. Reconstruction of CMNH50319 (Titanichthys cf. clarki) in lateral view. Lightly shaded regions are portions of plates that are overlapped. Dark shaded regions are empty space. IG = Inferognathal. Scale Bar = 10cm.

One other important feature we were able to sort out from this specimen was the orientation of inferognathal, which has a deep groove and an anterior tip that bends noticeably from the rest of the jaw. Originally, the groove was thought to be the dorsal surface and to have held tooth plates, making the anterior tip projecting dorsally as well. However, Dunkle and Bungart (1942) presented evidence that the groove faced ventrally making the anterior tip project ventrally as well. When Lehman (1956) described T. termieri he favored the original orientation. Because CMNH50319 had the left inferognathal preserved and articulated with the cheek shield we can now definitively say that the Dunkle and Bungart’s hypothesis is correct. That ventrally projecting tip is odd among arthrodires and can be very strongly developed (Fig. 2) based on observations in the Cleveland Musueum collections. This ventral curve makes the most sense, in my opinion, as an adaptation to increase gape size for filter-feeding.

Tusk Jaw

Figure 2. Reconstruction of undescribed left inferognathal of Titanichthys from the Cleveland Museum of Natural History showing the extreme condition of the ventrally projecting anterior tip. I refer to this as the tusk morph.

Despite all the anatomical detail we were able to report, when it came to assigning CMNH50319 to a particular species we ran into a problem. It had a mix of features from the well-known species and because all the species are poorly known (sample size of 1 or 2) we don’t really have any idea whether those differences are variable within or between species. We did have the inferognathal which is the easiest way to distinguish between T. agassizi and T. clarki, CMNH 50319 was definitively closer to the T. clarki form (it’s worth noting that the inferognathal of T. termieri is not known). So in the end we decided to classify CMNH50319 as Titanichthys cf. clarki, the cf. stands for confers and means close to. We probably could have just called it T. clarki without any complaints but I thought that would mask the uncertainty and variation within the genus.

We were able to perform a phylogenetic analysis with the specimen expanding on previous analyses (Carr 1991, Trinajstic & Dennis-Bryan 2009, Carr & Hlavin 2010,  Zhu & Zhu 2013, and Zhu et al. 2015). We added CMNH50319, T. agassiziBungartius perissus, and Tafilalichthys lavocati to the analysis with 121 characters. The resulting consensus tree (Fig. 3) is not very robust in all honesty. Too many taxa for the number of characters and A LOT of missing data. However, Titanichthys was recovered as a basal aspinothoracid in a monophyletic clade with Tafilalichthys and Bungartius. It will be interesting to see if that clade holds up in the long run because Tafilalichthys and Bungartius have both been associated with taxa that have crushing dentition. That would be a lot of variation in jaw morphology within a single group. I’m convinced the key to resolving arthrodire relationships are the Wildungen taxa from west Germany that seem perfectly placed in time and space to bridge the phylogenetic gap that currently exist.


Figure 3. Strict consensus of 2769 trees from a PAUP analyis.

Looking ahead to more research in this area several questions are clear. First, we can explicitly test the hypothesis that Titanichthys was a filter-feeder. Second, I gleaned over the geographic distribution of the species but as of right now there were 5! species of Titanichthys coexisting in the Appalachian Basin of North America. For giant filter-feeding organisms that seems incredibly unlikely to me and I suspect that most of the species should actually be synonomized. Looking through the Cleveland Museum collections for this research revealed a large amount of variation in Titanichthys jaws and I wouldn’t be surprised if the different morphs grade into one another or are ontogenetic. There’s plenty of material to work with to explore this project and I’d like to get to it sometime in the future, but if somebody beats me to it I’d be happy just to know that it’s being worked on.


Boyle, J. & M.J. Ryan, 2017, New Information on Titanichthys (Placodermi, Arthodira) from the Cleveland Shale (Upper Devonian) of Ohio, USA. Journal of Paleontology, v. 91, p. 318-336.

Carr, R.K., 1991, Reanalysis of Heintzichthys gouldii, an aspinothoracid arthrodire: Zoological Journal of the Linnean Society, v. 103, p. 349–390.

Carr, R.K., and Hlavin, W.J., 2010, Two new species of Dunkleosteus Lehman, 1956, from the Ohio Shale Formation (U.S.A., Famennian) and the Kettle Point Formation (Canada, upper Devonian) and a cladistic analysis of the Eubrachythoraci (Placodermi, Arthrodira): Zoological Journal of the Linnean Society, v. 159, p. 195–222.

Dunkle, D.H., and Bungart, P.A., 1942, The infero-gnathal plates of Titanichthys: Scientific Publications of the Cleveland Museum of Natural History, v. 8, p. 49–59.

Eastman, C.R., 1907, Devonic fishes of the New York formations: New York State Museum Memoir, v. 10, p. 1–235.

Lehman, J.-P., 1956, Les Arthrodires du Dévonien superieur du Tafilalet (sud Marocain): Notes et Mémoires du Service Géologique du Maroc, v. 129, p. 1–114.

Trinajstic, K., and Dennis-Bryan, K., 2009, Phenotypic plasticity, polymorphism and phylogeny within placoderms: Acta Zoologica, v. 90, p. 83–102.

Zhu, Y.-A., and Zhu, M., 2013, A redescription of Kiangyousteus yohii (Arthrodira: Eubrachythoraci) from the Middle Devonian of China, with remarks on the systematics of the Eubrachythoraci: Zoological Journal of the Linnean Society, v. 169, p. 798–819.

Zhu, Y.-A., Zhu, M., and Wang, J,-Q., 2015, Redescription of Yinosteus major (Arthrodira: Heterostiidae) from the Lower Devonian of China, and the interrelationships of Brachythoraci: Zoological Journal of the Linnean Society, v. 176, p. 806–834.

Placoderm Nurseries

Many sharks and bony fishes make use of protected shallow habitats for their young to grow in today. Typically these habitats provide protection from predators and abundant resources that rush downstream from the eroding highlands. This allows the maximum number of offspring to grow quickly before reaching adult size where they are less vulnerable to predation and can migrate out to deeper waters. Given how common this is today it would be expected that extinct organisms would have done this as well. Unfortunately, even if they were common we still might not be likely to find evidence of nurseries in the fossil record for two reasons. First, nurseries tend to be in shallow environments such as rivers and estuaries which are likely to be eroded as sea level rises and falls or continents collide erasing the environment. Second, identifying a nursery requires finding a location that is dominated by juvenile organisms and in most nurseries the juveniles are either eaten (no fossils left) or successfully reach adulthood and leave the nursery. So it’s only in the rare circumstance where there is a mass death of juveniles due to some catastrophe (at least from the organisms’ perspective) and then up to luck that the sediments aren’t eroded away for millions of years before they are discovered and sampled by paleontologists.

Despite these unlikely circumstances there are a number of likely nursery site of chondrichthyans (sharks and their relatives) in Triassic rocks of Kyrgyzstan (Fischer et al. 2011) and Pennsylvanian rock of Illinois (Sallan & Coates 2014). The longer ago sediments were formed the more likely they will have been destroyed by geological processes so it was exciting news this past week when a new study in interpreted a site in Belgium as a placoderm nursery (figure 1) from the Late Devonian (bonus points for being placoderms!). I’ll detail some of the bits I find the most interesting here. The study (Olive et al. 2016) was published in PLoS One and thus is open-access so I encourage everybody to go read the full paper for themselves, it’s a light read at nine pages.


Figure 1. From Olive et al. 2016 showing a reconstruction of the Strud nursery site. Scale bar =2cm. Illustration by J. Jacquot Hameon (MNHN, Paris).


The locality in Strud, Belgium would have been along the edge of the paleocontinent of Laurussia/Euramerica, a combination of North America and northern/western Europe, in the Late Devonian. It most likely represents the river deposits on an alluvial plain, a relatively calm environment, except in the case of flooding events. The authors report the recovery of 105 fragments of placoderms comprised of three species all of which were relatively small and showed other morphological correlates of juveniles based on previous studies of closely related placoderms (Werdelin & Long 1986; Deaschler et al. 2003). So what happened to these young placoderms that they weren’t eaten and yet they didn’t survive into adulthood? At Strud the most likely explanation is that these juveniles were in a pond or small tributary that dried up and became isolated from the main channel. So bad luck for the placoderms, good luck for paleontologists. What’s even more interesting though is that the Strud locality is the second known placoderm nursery of the Late Devonian. The other is in Tioga County, Pennsylvania (Downs et al. 2011) with similar species. In that case the hundreds of specimens are more complete and tend to be oriented in the same direction. Again, the explanation for their remains is that they were isolated in a shrinking body of water that also slowly became anoxic (aiding preservation) leading to a mass kill.

Both of these placoderm nurseries had large numbers of similarly aged individuals, and in the case of the Pennsylvania site there’s strong evidence that they represent a life assemblage, rather than one which has been time-averaged. So these sites probably show us that these placoderms had large numbers of offspring, though how exactly is up for debate. Some placoderms are known to have had live offspring (Ptyctodonts; Long et al. 2008) while there is questionable evidence of egg-laying behavior in other (Ritchie 2005; figure 2). Because of the number of similarly aged juveniles both nurseries most strongly support an egg-laying behavior in the species found there (primarily antiarchs). Given the position of placoderms as the outgroup to the rest of the jawed vertebrates their reproductive strategies can help us chart the evolution of reproductive strategies in Earth’s early history. It will be exciting to see if more of these nursery sites appear in other parts of the world outside of Luarussia and even earlier in fossil record. There are also tantalizing hints that large placoderm species (e.g. Dunkleosteus) might have used nearshore habitats for their juveniles as well (Daeschler & Cressler 2011)!

Cowralepis egg

Figure 2. Possible egg case of the plcaoderm Cowralepis. From Ritchie 2005. 



Daeschler, E.B. and W.L. Cressler III. 2011. Late Devonian paleontology and paleoenvironments at Red Hill and other fossil sites in the Catskill Formation of north-central Pennsylvania. Geological Society of America Field Guide 20:1-16.

Daeschler, E.B., A.C. Frumes, and C.F. Mullison. 2003. Groenlandaspid placoderm fishes from the Late Devonian of North America. Records of the Australian Museum 55:45-60.

Downs, J.P., K.E. Criswell, and E.B. Daeschler. 2011. Mass mortality of juvenile antiarchs (Bothriolepis sp.) from the Catskill Formation (Upper Devonian, Famennian Stage), Tioga County, Pennsylvania. Proceedings of the National Academy of Science Philadelphia 161:191-203.

Fischer, J., S. Voigt, J.W. Schneider, M. Buchwitz, and S. Voigt. 2011. A selachian freshwater fauna from the Triassic of Kyrgyzstan and its implication for Mesozoic shark nurseries. Journal of Vertebrate Paleontology 31:937-953.

Long, J.A., K. Trinajstic, G.C. Young, and T. Senden. 2008. Live birth in the Devonian period. Nature 453:650-652.

Olive, S., G. Clement, E.B. Daeschler, and V. Dupret. 2016. Placoderm assemblage from the tetrapod-bearing locality of Strud (Belgium, Upper Famennian) provides evidence for a fish nursery. PLoS One 11:e0161540.

Ritchie, A. 2005. Cowralepis, a new genus of phyllolepid fish (Pisces, Placodermi) from the late Middle Devonian of New South Wales, Australia. Proceedings of the Linnean Society of New South Wales 126:215-259.

Sallan, L.C. and M.I. Coates. 2014. The long-rostrumed elasmobranch Bandringa Zangerl, 1969, and taphonomy within a Carboniferous shark nursery. Journal of Vertebrate Paleontology 34:22-33.

Werdelin, L. and J.A. Long. 1986. Allometry in the placoderm Bothriolepis canadensis and its significance to antiarch evolution. Lethaia 19:161-169.