Gallery and print store

Friday, 26 January 2018

Did tyrannosaurs smile like crocodiles? A discussion of cranial epidermal correlates in tyrannosaurid dinosaurs

Brain 1: "Right, you need an image for your tyrannosaurid facial tissue post."
Brain 2: "OK, here're some Tyrannosaurus rex in a really dark and back-lit scene. Their faces are in shadow, and you can't really see the features."
Brain 1: "This is perfect. After all, only losers want to see the faces of animals in posts about facial tissues."
Brain 2: "Exactly. Hey, since when did I have two brains?"
Brain 3: "Beats me."
Discussing the craniofacial tissues of tyrannosaurid dinosaurs is the palaeointernet equivalent of lighting a match in a straw-filled barn - the slightest spark of opinion spawns a 100-strong comment field about extra-oral tissues, tooth exposure, rictal tissues, facial skin depth and a number of other topics. But despite this keen popular interest, there's been relatively little academic study into tyrannosaurid facial tissues, perhaps because their soft-tissues mostly remain unrepresented in the fossil record. Happily, close examination of tyrant skulls reveals a number of textures and rugosity profiles which were almost certainly created by bone-skin interaction, so we can form some idea of their life appearance even without soft-tissue specimens. The first detailed attempt at interpreting tyrant cranial rugosities was published last year by tyrannosaur expert Thomas Carr and colleagues (Carr et al. 2017 - you might also know Thomas by his super-comprehensive blog Tyrannosauroidea Central). This widely publicised paper proposed a number of hypotheses about the face of Daspletosaurus horneri: that the sides of the jaw were adorned with crocodile-like 'facial scales'; that various scales, dermal armour and cornified sheaths adorned the nasal and orbital region; and that it lacked lips (not explicitly stated in the paper, but restored as such in an illustration and touted in the paper's PR). The idea that tyrannosaurids may have had crocodylian-like facial tissues has since generated a lot of discussion online, some in favour, some against, and as someone increasingly looking at epidermal correlates for palaeoartistic purposes, I thought this topic was worthy of a blog article: are tyrannosaurid jaws really croc-like enough to assume comparable skin types?

(An important caveat before we start this discussion is that the following is based on tyrannosaurids generally, not D. horneri specifically, because the horneri study does not include photographs of its alleged epidermal correlates. The D. horneri paper describes them very well (see Carr et al. 2017, supplementary data), but it's difficult to evaluate them without images of the bone surfaces themselves. Dave "Tyrannosaur Chronicles" Hone needs a shout out here for sharing his expertise and extensive image library of tyrant fossils as I prepared this post - though I have some experience with tyrannosaur bones and their interpretation, this article has been considerably improved by his involvement.)

Tyrannosaurids and crocodylians: face off

An obvious place to begin this discussion is crocodylian facial structure. Crocodylian skulls are so familiar that it's easy to forget how distinctive they are among modern animals, and I don't think it's widely known that their skin plays a significant role in shaping their skull tissues. Crocodylian jaw bones have incredibly high numbers of foramina, with averages of 100 in each major jaw bone (premaxilla, maxilla and dentary) and over 1000 in each bone in some specimens (Morhardt 2009). These openings are the loci around which gnarly ridges and tubercles grow by a process of dermal ossification: tissues from the skin are turned to bone and build up the sculpting on the skull surface (Grigg and Kirshner 2015; de Buffrénil et al. 2015). Simultaneously, the bone immediately surrounding the foramina is resorbed, enhancing the rugosity pattern further and creating that highly distinctive, deeply pitted and grooved crocodylian skull texture (de Buffrénil et al. 2015). This restructuring can be extensive and, over ontogeny, crocodylian snout surface area can increase by as much as 20% (de Buffrénil et al. 2015). That's a major reworking of the superficial bone of the skull, and their skin has a major role in its development.

Skull of a mature American crocodile, Crocodylus acutus, demonstrating that classic crocodylian skull texture. Cropped from public domain Wikimedia image by Daderot.
Among living tetrapods, only some turtles and a couple of geckos show a comparable degree of sharply-defined cranial sculpting (Evans 2008; de Buffrénil et al. 2015) but, among extinct taxa, stem-tetrapods, temnospondyls, parareptiles and many crocodylomorphs present analogous cranial conditions (Witzmann et al. 2010; de Buffrénil et al. 2015). Studies show that temnospondyl skulls developed their sculpting via a similar mechanism of ossifying dermal tissues (Witzmann et al. 2010), perhaps indicating croc-like skin properties in these animals, too. Until recently it was thought that crocodylian facial skin was scaly, but new research shows that it is actually a sheet of toughened skin which cracks through growth, creating a scaly appearance, but not true epidermal scales akin to those seen in lizards (Milinkovitch et al. 2013). Regardless of whatever other conclusions are drawn here, this has to be a minor amendment to Carr et al.'s (2017) interpretation: if tyrannosaurids (or any other extinct animal) have croc-like textures on their jaw bones, we should be visualising tight, tough skin, not epidermal scales.

Juvenile alligator, Alligator mississippiensis, showing virtually crack-free facial skin - it's only adults that develop the extensively cracked, superficially 'scaly' faces. Photo by Joxerra Aihartza, from Wikimedia, FAL 1.3.
Whether tyrannosaurid jaws are truly crocodylian-like is open to question, however. Carr et al, (2017) are clear that they consider tyrants and crocodylians jaws as identical in superficial appearance ("The texture in crocodylians is identical to that of tyrannosaurids, except that the entire face of crocodylians is coarse in texture" - p. 21; Supplementary information to Carr et al. 2017) but I disagree: there are a number of ways in which they differ and, given the link between crocodylian skull development and dermal tissues, these differences may be critical to our considerations of facial anatomy. Many of these contrasts pertain to jaw foramina, which we know are important in defining crocodylian cranial sculpting (de Buffrénil et al. 2015) and may have a deeper relationship with jaw tissue properties (Morhardt et al. 2009; Hieronymus et al. 2009).

Firstly, although tyrannosaurids have elevated numbers of jaw foramina compared to other dinosaurs, their numbers are, on average, significantly lower than those of crocodylians (Morhardt 2009). No tyrannosaurid jaw bone reported by Morhardt (2009) exceeds 81 foramina, which is high for a dinosaur, but still short of the crocodylian average, and well below the 1000+ figure reported for some croc jaws. Interestingly, data in Morhardt (2009) suggests that foramina numbers weakly correlate to jaw size: the longer a jaw is, the more foramina it generally has. This trend is particularly well shown in her tyrannosaurid sample but seems true of other fossil and extant animal groups as well, and might also be reflected in ontogeny (smaller Tyrannosaurus have fewer foramina, on average, than large ones). The cause behind this trend seems to be elusive at present - might it reflect a change in tissue type with age (Morhardt 2009)? does it reflect demands of supplying an absolutely larger jaw with nervous and vascular tissues? - but whatever the reason, it implies that we should consider foramina frequency proportionate to jaw size when analysing rugosity profiles. Under this metric, foramina values in crocodylian jaws are even more impressive as, compared to some extinct animals, their skulls are of middling size. By contrast, the slightly above-average foramina counts of even the largest tyrannosaurines seem less significant because, even with extreme jaw size, they don't attain a value comparable to a much smaller alligator. If we remove size from our consideration by comparing similarly-sized tyrant and croc jaws, we find they are worlds apart in terms of jaw perforation. Indeed, the foramina values of smaller tyrants are nothing special - they are comparable to most other similarly-sized tetrapods (Morhardt 2009). Presumably, this explains why - as many internet conversations have pointed out - tyrannosaurid jaws simply don't have that same obvious, pitted surface as those of crocodylians.

Further differences might be noted in relative foramina sizes. Those foramina occurring high on tyrant snouts - such as at the top of the maxilla - are much smaller than the broader, obviously deep labial foramina paralleling the jaws (Brochu 2003; Carr et al. 2017). In crocodylians however, jaw foramina seem to have a lower size range. Foramina shape and size is an important consideration for facial tissues (Hieronymus et al. 2009) and this might imply different tyrant facial tissues over the side of the snout vs. those at the jawline, whereas the more uniform foramina sizes of crocodylians are entirely consistent with their homogeneous jaw skin.

Schematic drawing of Tyrannosaurus skull FMNH PR2081 (the specimen better known as 'Sue') showing the distribution and (somewhat conservatively) size differences in jaw foramina. This huge skull is said to be one of the most rugose Tyrannosaurus skulls known (Brochu 2003), but it fails to meet the high foramina numbers, sculpting extent and uniform foramina size of mature crocs. Image from Brochu (2003).
A related issue concerns a possible link between extra-oral tissues and foramina counts. Morhardt (2009) noted that, as a general rule, extant animals with average foramina counts below 50 in each jaw bone have tooth-covering extra-oral tissues; that those above 50 but below 100 have immobile facial tissues; and only those with 100 or more are reliably excluded from having lips or other means of tooth coverage. Average tyrant jaw foramina counts are well below that upper threshold for exposed teeth so, by this metric, they should have lips, and would not look like bipedal crocodiles. This might match what we're seeing with tyrant foramina size: perhaps those large labial foramina are something to do with nourishing and innervating extra-oral tissues, while those on the side of the snout need only access the overlying skin. There are some complications to Morhardt's data (if anyone is looking for a PhD project, a more extensive follow up would be terrific) but, at face value, her research does not support crocodile-like facial tissues for tyrannosaurids.

Finally, we can observe that the ontogeny of tyrannosaurid skull textures is not at all crocodylian-like. Tyrants do have some sculpting on their jaw bones and, as with most reptiles, these become better defined with maturity (e.g. Evans 2008; de Buffrénil et al. 2015). However, even the most rugose tyrannosaurid skulls do not match the complex and sharply pitted rugosity patterns of mature crocodylians (e.g. Osborn 1912; Carr et al. 1999; 2017; Brochu 2003; Hone et al. 2011). Given that ossifying facial skin is a direct factor in jaw bone sculpting in crocodylians, the lack of comparable development in tyrannosaurids is a blow to the idea that their faces bore the same dermal regime. Histological examination of tyrannosaurid jaw bones for might have further insight here, as the resorption/remodelling pattern might reveal details about bone/dermal interactions (Witzmann et al. 2010; de Buffrénil et al. 2015) but, for now, this inconsistency seems to be a big hole in the idea that tyrannosaurids had crocodylian-grade facial tissues.

Does the tyrannosaurid EPB help here?

Collectively, these points seem to suggest that tyrant jaws are not as croc-like as argued, and that it's not a given that the two groups had similar facial tissues. A counterargument to this is that crocodylians are the best tyrant analogues in their extant phylogenetic bracket (EPB), and thus give us our best, most phylogenetically informed insight into tyrannosaurid faces. Indeed, the croc-snouted tyrant hypothesis was informed primarily by comparisons with taxa from the tyrannosaurid EPB - specifically the skulls of birds and alligators (Carr et al. 2017) and, sure, crocs and tyrannosaurid jaws may not be exactly alike, but they're undeniably more similar to each other than either is to a bird. Might we concede that the comparisons aren't perfect, but that this is simply the best we can do without violating the tyrannosaurid EPB?

Our issue is that, while the EPB is a terrific method for predicting ancient anatomies, it really struggles with the complexity of archosaur facial tissue evolution, perhaps to the extent of being redundant. One major issue is that we can be near certain early archosaurs had neither croc- or bird-like facial tissues because no species representing the earliest phases of archosaur evolution have comparable skin-influenced jaw textures (see Nesbitt et al. 2013, and papers therein). Rather, we only see these features developing in relatively crownward archosaur groups, implying independent development of their respective facial anatomies well after the croc-bird split. This being the case, the common archosaur ancestor must have had a different set of facial tissues, and the facial anatomy of extant archosaurs may tell us little about the faces of Mesozoic dinosaurs.

Like crocodylians, birds have jaws with surface textures shaped by their overlying skin: networks of branching neurovascular canals and oblique foramina underlie cornified sheaths (their beaks). The prominence of such jaw rugosities in living archosaurs allows us to predict the facial condition of fossil archosaurs and stress test the tyrannosaur EPB, and it doesn't seem to hold up well. This skull is a marabou stork (Leptoptilos crumenifer), photo by me.
A second major issue is evidence that living archosaur faces don't reflect tissues known from their fossil cousins. In addition to tight facial dermis and cornified sheaths, a plethora of fossil evidence show that fossil archosaurs had faces with epidermal scales, projecting skin tissues (e.g. pterosaur crests) armoured dermis, and cornified pads (Frey et al. 2003; Hieronymus 2009; Hieronymus et al. 2009; Carr et al. 2017). These go well beyond the anatomical range implied by the EPB and show that fossil archosaur faces sometimes had more in common with non-archosaurs than their closest extant relatives. We must remember that the EPB is a predictive method which should be applied where no other data is forthcoming: in this case, we have enough fossil data to show that our EPB predictions are problematic, and that we can't rely on it for insight into tyrannosaurid faces. I'm hardly the first to suggest EPB approaches don't help discussion of non-avian dinosaur faces (e.g. Vickaryous et al. 2001; Knoll 2008), but these points are worth repeating in this context: I don't think the EPB is a compelling supporting argument for a croc-faced tyrannosaurid.

So, if not croc-like, what might be happening here?

If croc-skinned tyrant snouts are problematic, what are our other options? Our discussion above really only pertains to the maxillary region of tyrannosaurid snouts and, for the rest of the skull, I think Carr et al. (2017) nailed it: what I've seen of tyrannosaurid skulls suggests the orbital region and skull roofs were covered in cornified sheaths, armoured dermis and large scales. There seems to be quite a bit of variation in these tissues, with some taxa having more defined scale correlates over the nasals than others, as well as differences in elaborations of the hornlets above the eye. In all likelihood, different tyrant species would be highly recognisable in life by the development of scales, armour and horn across the top of their faces. These armoured tissues are entirely consistent with what we understand of tyrannosaurid behaviour: if you were being routinely bitten about the face by another tyrannosaur, you'd want some protection too (see opening image).

Dorsal view of the snout of a red river hog (Potamochoerus porcus). These pitted, grooved bone textures are fairly widespread across tetrapod skulls and don't seem to correlate to any one skin type, but might indicate the presence of tough, well-cornified skin (these hogs wrestle with their faces, so need protected snouts). Note the projecting rugosities on the side of the snout and on the ascending maxillary projection - these anchor vast skin projections in life. Red river hog skulls are awesome. Photo by me.
But what of that maxillary portion of the snout - the lateral region suggested as being crocodile-like? The surfaces of tyrannosaurid maxillae are pretty complex with a hierarchy of rugosity profiles (Carr et al. 2017). Very obvious features include many pits and short, branching neurovascular grooves: these might not necessarily indicate of particular tissue type in themselves, but are often associated with a well-cornified, tough epidermis (above). The high number of foramina in tyrant maxillae implies immobile facial tissues (Morhardt 2009), which I guess we probably expected in a reptile anyway.

Holotype maxilla of Zhuchengtyrannus magnus: check out that network of elliptical depressions bordered by raised regions. Note how they terminate about a few centimetres above the line of labial foramina - we'll come back to this in a moment. From Hone et al. (2011).
Underlying these pits and grooves are a series of large, elliptical shallow depressions surrounded by low ridges (above). These vertically-aligned structures are found in many tryannosaurids and are especially obvious in large tyrannosaurines like Tyrannosaurus, Tarbosaurus and Zhuchengtyrannus. You can see them easily in museum mounts, even from across the room. Some taxa have single rows of these structures below the antorbital fenestra (Tyrannosaurus), but others have tessellating networks of depressions and ridges that extend to the top of the maxilla (Tarbosaurus), terminating beneath the scaly region overlying the nasal bone. They're unusual structures which are almost certainly epidermal in origin: they're in a place where epidermal correlates often form; are more pronounced in mature individuals; are regularly and consistently arranged across the surface of the skull; and are not associated with any pneumatic or neurovascular openings. They broadly recall the 'hummocky' rugosity profile seen under epidermal scales (Hieronymus et al. 2009) and, if so, the convex, ridged areas probably underlay vertically aligned scales, or rows of scales. Some tyrant skulls, such as the especially rugose Tyrannosaurus skull AMNH 5027, have especially sharp and rugose ridges which, to me, recall the facial ridges of certain iguanine lizards: specifically, anoles, chameleons and basilisks. These are often quite rugose and sculpted, but smoother, more tyrannosaurid-like conditions exist in a number of species (I'm thinking of things like helmeted basilisks and smooth chameleons). Prominent, ornamental rows of relatively large and often colourful scales overly these structures in these iguanines and I wonder if the same was true for tyrants. Alternative hypotheses, such as scales sitting in the depressions between the ridges, aren't consistent with the relationship between scales and bone in living species, and there's no indication that other tissue types (e.g. cornified sheaths, armoured dermis) were present in these areas, so I think the ornamental ridge hypothesis is sensible (or, at least, not outrageously daft given the available data). I must admit to liking this hypothetical juxtaposition of fancy ornamental scales around the mouth and tough, reinforced tissues over the snout: perhaps tyrannosaurs weren't just big biting machines, but liked to look nice, too.

AMNH 5027 is just riddled with interesting surface textures that probably relate to epidermal features. To my reckoning, in addition to those depression/ridge pairings on the maxilla, the dorsal region of the lacrimal bears coarse projecting rugosity (armoured dermis); the top of the premaxilla and postorbital has a series of coarse hummocks (probable scales); and the ascending processes of the postorbital, lacrimal and maxilla are covered in a dense, anastomising network of neurovascular foramina (cornified sheath). What a neat looking animal Tyrannosaurus must've been - my take on this data is seen in the paintings accompanying this post. Image from Osborn (1912), in public domain.
Significantly, I can't find any tyrannosaurid skulls where these possible scale correlates extend right to the base of the maxilla (see photos, above). Rather, they terminate a few centimetres above the line of labial foramina, and this might have bearing on ongoing discussions about dinosaur lips. Scleroglossan lizards (the group that includes geckos, skinks, varanoids and amphisbaenians) frequently have osteoderms on their faces which cover their snouts (including the maxillae) except for a region around the labial foramina, which is smooth. This seems to relate to the presence of lip tissues displacing the scales from the skull and prohibiting formation of a epidermal correlate adjacent to the toothrow. Their maxillary juxtaposition of epidermal correlates is the same configuration that we see in tyrannosaurids as well as a number of other non-avian dinosaurs with maxillary epidermal correlates (e,g, pachycephalosaurids, ankylosaurids, some ceratopsids) and this has to be regarded consistent with hypotheses of extra-oral 'lips' in tyrannosaurids and other dinosaurs. If we add this to the evidence from foramina counts (Morhardt 2009, also see above) as well as other arguments for extra-oral tissues the case for crocodylian-like exposed teeth is looking increasingly doubtful. I must admit to thinking that proponents of exposed dinosaur teeth really need to start making better cases for this idea: most ways we can slice this particular debate suggests that extra-oral tissues are looking likely (and no, the common argument that their teeth were too big to be sheathed isn't valid: it's simply a speculation based on incredulity, not actual data from dinosaur skulls).


To sum up this long, detail-heavy post:
  1. Crocodylian skull textures are basically built by their skin, and we should expect any prehistoric animal with croc-like facial tissues to have a croc-like cranial rugosity profile. What we see in tyrannosaurs is a little croc-like, but only superficially. Differences between croc and tyrant skull tissues may be more significant than their similarities and seem to contradict the notion of croc-like facial tissues in tyrannosaurids.
  2. Attempts to ground discussions of dinosaur facial tissue in the EPB are problematic: a great deal of what we know about archosaur facial tissue refutes what the EPB predicts. Basic comparative anatomy, framed by a wide phylogenetic bracket, might be the way forward for understanding dinosaur faces.
  3. Tyrant faces - as largely predicted by Carr et al. (2017) - seem to have been adorned with scales, cornified sheaths and armoured dermis, but their jaw regions may have been covered in vertical (perhaps ornamental?) bands of epidermal scales, not croc-like skin. Distribution of epidermal correlates around the jaws of tyrannosaurids (and other dinosaurs) is suspiciously reminiscent of many lizard skulls, and may favour a lipped condition.
Tyrannosaurus rex portrait, based on my take of epidermal correlates of the AMNH 5027 skull. No, you tell it that its ornamental ridges look a bit silly.
Perhaps unsurprisingly, I couldn't research and write all this without wanting to draw my take on tyrannosaur facial anatomy. I'll leave you with my take on the face of AMNH 5027 (above): I'm sure it'll need modifications as more details on tyrannosaurid faces come to light, but I won't pretend it wasn't neat to draw a Tyrannosaurus based on relatively objective reading of available data. Palaeoart is at it's most exciting when we join dots between data rather than, as is so often the case, largely imagine huge swathes of our subject species. The duelling Tyrannosaurus that welcomed you to the post are based on the same model.

Enjoy monthly insights into palaeoart and fossil animal biology? Support this blog for $1 a month and get free stuff!

This blog is sponsored through Patreon, the site where you can help online content creators make a living. If you enjoy my content, please consider donating $1 a month to help fund my work. $1 might seem a meaningless amount, but if every reader pitched that amount I could work on these articles and their artwork full time. In return, you'll get access to my exclusive Patreon content: regular updates on research papers, books and paintings, including numerous advance previews of two palaeoart-heavy books (one of which is the first ever comprehensive guide to palaeoart processes). Plus, you get free stuff - prints, high quality images for printing, books, competitions - as my way of thanking you for your support. As always, huge thanks to everyone who already sponsors my work!


  • Brochu, C. A. (2003). Osteology of Tyrannosaurus rex: insights from a nearly complete skeleton and high-resolution computed tomographic analysis of the skull. Journal of Vertebrate Paleontology, 22, 1-138.
  • Carr, T. D. (1999). Craniofacial ontogeny in Tyrannosauridae (Dinosauria, Coelurosauria). Journal of vertebrate Paleontology, 19(3), 497-520.
  • Carr, T. D., Varricchio, D. J., Sedlmayr, J. C., Roberts, E. M., & Moore, J. R. (2017). A new tyrannosaur with evidence for anagenesis and crocodile-like facial sensory system. Scientific reports, 7, 44942.
  • De Buffrénil, V., Clarac, F., Fau, M., Martin, S., Martin, B., Pellé, E., & Laurin, M. (2015). Differentiation and growth of bone ornamentation in vertebrates: a comparative histological study among the Crocodylomorpha. Journal of morphology, 276(4), 425-445.
  • Evans, S. E. (2008). The skull of lizards and tuatara. Biology of the Reptilia, 20, 1-347.
  • Grigg, G. (2015). Biology and evolution of crocodylians. Csiro Publishing.
  • Hieronymus, T. L. (2009). Osteological correlates of cephalic skin structures in amniota: Documenting the evolution of display and feeding structures with fossil data. Ohio University.
  • Hieronymus, T. L., Witmer, L. M., Tanke, D. H., & Currie, P. J. (2009). The facial integument of centrosaurine ceratopsids: morphological and histological correlates of novel skin structures. The Anatomical Record, 292(9), 1370-1396.
  • Hone, D. W., Wang, K., Sullivan, C., Zhao, X., Chen, S., Li, D., ... & Xu, X. (2011). A new, large tyrannosaurine theropod from the Upper Cretaceous of China. Cretaceous Research, 32(4), 495-503.
  • Frey, E., Tischlinger, H., Buchy, M. C., & Martill, D. M. (2003). New specimens of Pterosauria (Reptilia) with soft parts with implications for pterosaurian anatomy and locomotion. Geological Society, London, Special Publications, 217(1), 233-266.
  • Knoll, F. (2008). Buccal soft anatomy in Lesothosaurus (Dinosauria: Ornithischia). Neues Jahrbuch für Geologie und Paläontologie-Abhandlungen, 248(3), 355-364.
  • Milinkovitch, M. C., Manukyan, L., Debry, A., Di-Poï, N., Martin, S., Singh, D., ... & Zwicker, M. (2013). Crocodile head scales are not developmental units but emerge from physical cracking. Science, 339(6115), 78-81.
  • Morhardt, A. C. (2009). Dinosaur smiles: Do the texture and morphology of the premaxilla, maxilla, and dentary bones of sauropsids provide osteological correlates for inferring extra-oral structures reliably in dinosaurs?. Western Illinois University.
  • Nesbitt, S. J., Desojo, J. B., & Irmis, R. B. (2013). Anatomy, phylogeny and palaeobiology of early archosaurs and their kin. Geological Society, London, Special Publications, 379(1).
  • Osborn, H. F. (1912). Crania of Tyrannosaurus and Allosaurus; Integument of the iguanodont dinosaur Trachodon. Memoirs of the AMNH; new ser., v. 1, pt. 1-2.
  • Vickaryous, M. K., A. P. Russell, and P. J. Currie. (2001). Cranial ornamentation of ankylosaurs (Ornithischia: Thyreophora): reappraisal of developmental hypotheses. In K. Carpenter (ed). The Armored Dinosaurs. 318–340. Indiana University Press.
  • Witzmann, F., Scholz, H., Mueller, J., & Kardjilov, N. (2010). Sculpture and vascularization of dermal bones, and the implications for the physiology of basal tetrapods. Zoological Journal of the Linnean Society, 160(2), 302-340.

Friday, 17 November 2017

Can we predict the horn shapes of fossil animals? A thought experiment starring Triceratops

Triceratops horridus with some crazy long and curving brow horns. Just speculation, right? Surprisingly, maybe not...
For palaeoartists, animals with flamboyant headgear are among the most rewarding to render, but it's not only the bony aspects of their cranial ornaments that we have to pay attention too. Animal headgear is covered with various amounts of soft-tissue that, in extreme cases, can dramatically augment the shape of the underlying bony features. The headgear of living species has a spectrum of soft-tissue coverings from nothing at all (mature deer antlers), to relatively thin dermal tissues (giraffe ossicones), through to hard keratin sheaths that can add significant depth and length to a horn or crest (most other animal horns). This excellent breakdown of a bighorn sheep face by Aaron Drake of Colorado State University (uploaded by Simpleware Software Solutions) gives a pretty good idea of how much tissue extreme keratin sheaths can add to the underlying skull.

Not all horns are augmented to the extent seen in bighorn sheep, but even modestly proportioned keratin sheaths can add a lot of bulk, length and characteristic geometry to horn tissues. Thus, anyone hoping to accurately predict the appearance of ancient horned animals should want to predict the shape of their horn sheaths along with understanding the skull geometry. This isn't easy because, though incredibly tough and resistant, keratin sheaths are still prone to decay and rarely fossilise.

Researching horn growth for an upcoming book project has made me wonder if horn sheath shape might be more predictable than we've traditionally thought, however. Horn sheath growth mechanics are relatively simple, closely related to bone shape, and constrained by the properties of heavily keratinised tissues. They're also fairly universal across across tetrapods - the same processes that make a goat horn will make the enormous keratin sheath of a skimmer jaw, for instance. These properties might allow insights into sheath shape in fossil species even when the sheath is not preserved. So what aspects of horn sheath growth might allow this, and how could we transfer them to fossil animals?

Growing horn sheaths in living animals

Keratin sheaths are dead tissue with their only living components being the cells that synthesise the keratin at the horn core/sheath interface (e.g. at the inner surface of the horn soft-tissues, see diagram, below). Because no living tissue reaches the outer horn surface, they cannot grow by adding tissue to the tip. Rather, they grow by internal accumulation of keratin layers, each new deposit displacing the older sheath from the bony core. This creates a stack of keratin cones, with new cones growing at the base and causing the horn tissues to lengthen. Continuous internal deposition and displacement of old material is what creates the soft-tissue horn extension, as each new keratin layer shoves the older material a little further from the bony tip. This makes the tip of a keratin horn the oldest part of the sheath, and in many bovids the tips are many years old. Conversely, the youngest part of the horn tissues are located at the base. As we discussed in a recent post about the horns of Arsinoitherium, this growth mechanism binds the internal horn tissues in the overlying sheaths, limiting their ability to change size or shape. Changes in size or curvature can only be achieved by displacing the older horn layers, but complicating the horn shape - say, by branching the tip - is impossible unless the sheath is shed, pronghorn-style. The sheath itself can't be modified after deposition either, on account of no living tissue reaching it. Thus, old sheaths permanently maintain the size and geometry they were created with.

Stylised bovid horn growth, heavily modified from Goss (2012).
This growth mechanic presents three important points relevant to predicting the shape of fossil horn sheaths. The first is that sheath tissues are synthesised directly over the horn core, effectively making the internal sheath margin a cast of the bone at the time it grew. The second is that the shape of new keratin layers are constrained by the keratin sheaths that preceded them. They can't deviate too radically from the overlying horn shape and the horn core of the emerging layer should mostly nestle into margins of the older one. The third is that horn extensions are not simply exaggerations of their contemporary horn core, but a keratinous record of the horn history. Geometry exhibited by the earliest growth stages is maintained in the extending sheath regardless of later changes to the horn core morphology, and only periodic shedding or heavy abrasion are likely to alter this.

This being the case, could ontogenetic changes in horn cores provide insight into the sheath shape of fossil animals? If bone shape translates to keratin sheath shape, and sheath shape dictates the horn extension profile, then a growth series of bony horn anatomy may allow us to reconstruct horn keratin accumulations that are otherwise lost to decay. Horn core profiles give us a 'cast' of the inner sheath margin for that growth stage, and we can fit these into the margin of the preceding sheath layer (which, of course, can be deduced by the shape of a ontogenetically preceding horn core). Building a stack of nestled horn core profiles creates something akin the bovid horn diagram above and tells us something of how keratin layers were accumulated for that horn shape. The very tip of the horn sheath is lost to time because we cannot predict external appearance from horn core casts (they only represent the internal structure) but if the youngest animal in a growth series is suitably juvenile, we probably aren't missing much.

As proof of concept, I've taken the horncore outlines from the schematic bovid horn above and attempted to recreate the horn shape. Stacking them was achieved by simply eyeballing the margins, trying to fit the horn core outlines together as tightly as possible without their margins overlapping. Here's how it turned out...

I don't think that's too bad. It's not perfect, but it gives a pretty good idea what's going on with the actual horn. This method is very simple, but - as outlined earlier - keratin horns are simple, so we might not need a particularly complex method to predict their shape. But you're not here to talk about ram horns: what happens when we apply this idea to a fossil animal with a well-known growth series, and how do the results compare to our conventional means of reconstructing horn sheaths in fossil taxa?

Step forward, Triceratops

Triceratops growth series from Horner and Goodwin (2006). Both species of Triceratops are included here, but the generalities of this growth sequence are thought to apply to both. Say, that brow horn curvature looks pretty changeable - what would that mean for horn sheath shape?

The super-famous horned dinosaur Triceratops is a great animal to explore this idea with. It's known from dozens of specimens representing a range of ontogenetic stages, from small juveniles to giant adults (above, Horner and Goodwin 2006 - and no, the adults in question here are not Torosaurus). Like the horns of other ceratopsids, Triceratops brow horns have well-developed epidermal correlates for keratinous sheaths (oblique foramina and anastomosing neurovascular channels - Horner and Marshall 2002; Hieronymus et al. 2009) and these textures are present in the smallest known skulls, indicating that most or all their life was spent with sheathed brow horns (Goodwin et al. 2006). Confirmation of a horn sheath comes from poorly-preserved soft-tissues found on some Triceratops horns (Farke 2004; Happ 2010).

Triceratops skulls underwent pretty major changes as they grew, including complete reorientation and allometric scaling of the brow horns. In juveniles these curve backwards, but in big adults they arc forwards (Horner and Goodwin 2006). Typically, artists have assumed that the keratin sheaths covering these horns changed shape with them. Even pros, such as Greg Paul (2016), who have stressed that the keratin sheath should extend the horn shape, render the sheaths as more-or-less reflecting the underlying horn core of a given growth stage, without any hangovers from a previous iteration of horn shape. Whether intentional or not, the implication here is that the horn sheath was dynamic - capable of changing as the animal grew.

....just like this. Note how the brow horns of this Triceratops group are clearly changing shape as the animals increase in size, but that the keratin sheaths don't reflect any earlier horn history. Hmm. Say, do you know this image is on the front of my 2018 calendar?
The model outlined above conflicts with this traditional take, however. If we assume that the horn extension was composed of a series of retained keratin sheaths, and using Horner and Goodwin's (2006) ontogenetic sequence as a basis, the resultant horn shape is pretty surprising. Stacking horn cores in the juveniles sees those recurved shapes pushed off the horn core to extend and extenuate the curve strongly, to the point where the horn tip even points posteriorly at one stage (below). As the horn base tips forward on the approach to adulthood, these arcing tips rotate with them, creating a long, elaborate set of horns which curved twice: once at the tip, and again, but inversely, at the base. If the Triceratops in this model retained the full history of their horn sheaths into adulthood, the result would be pretty fantastic: very long horns where the tips pointed 90° away from the point of the horn core. Yowsers - that's quite different from our traditional 'just make it pointier' approach.

Stacking Triceratops horn cores, mimicking how living animal keratin sheaths grow, suggest the keratinous extension of the brow horns was strongly curved even in adult animals. As in the mock bovid horn above, the horn cores were stacked simply by trying to make them fit as neatly as possible.
Which is more likely: twirly horn sheaths or the more conventional, 'dynamic' sheaths? Where morphing horn sheaths immediately lose points is their requirement for the inert keratin horn tissues to react to each horn core shape, as well as for the horn sheath history to continuously disappear. Modern horn sheaths just don't grow like this: their extensions only exist because the old keratin tissues hang around, and we have to ask how the extending sheaths are created in our 'dynamic' sheath model. There are perhaps two ways we could attain morphing sheaths: the first is through continuous eradication of old sheath material, allowing new keratin to grow over the horn core without being obscured by previous sheath layers. This might have been achieved by Triceratops shedding and regrowing sheath extensions, or by abrading outer sheath tissues away. The second is that the horns weren't covered in one sheath but several interlocking plates, like the beaks of some birds, which might allow for jimmying and reconfiguration of the horn tissues through growth without adding lots of material to the end.

Let's consider shedding first. It's possible that at least some layers of Triceratops horns were shed because exfoliation is common on keratin sheaths in living species. For instance, puffins shed the outer layer of their beaks annually, and bovids exfoliate outer layers of their horns once or twice in their lifetimes (O'Gara and Matson 1975; Goss 2012). The fact that only a superficial layer of tissue is lost prevents the sheath being significantly altered however: exfoliation alone would probably not give us particularly 'dynamic' horn sheaths.

Constant reshaping of horn tissues might be plausible if Triceratops could regularly shed and regrow the horn sheath, as performed by pronghorns. Unfortunately, these mammals show us that detecting this growth mechanic in fossil species is challenging, however. Despite their unusual habit of regrowing an entirely new sheath each year, pronghorn horn cores have similar textures to those of animals with permanent sheathing (Janis et al. 1998). There are some differences, but they're subtle. O'Gara (1990) reported that pronghorn horn cores have annually variable properties, alternating between a spongy, relatively rounded horn core when the sheath is growing, and a smooth-textured, sharper horn core at peak sheath hardness (O'Gara 1990). It's pretty well established that dinosaur skeletons grade from spongy, rounded bones to smoother, sharpened bones as they aged, so perhaps variation in texture and shape of Triceratops horns that broke this pattern could indicate horn shedding - provided these differences could be distinguished from ontogenetic or intraspecific factors. I'm not aware of any evidence of this kind, despite the frequency in which Triceratops skull bone texture is commented, but I also don't know that anyone has specifically looked for this variation yet.

Lovely, lovely epidermal correlates on the skull of Triceratops prorsus illustrated in Hatcher (1907). Note that there's no divide between the correlates on the brow horn and surrounding skull - might we expect some sort of dividing sulcus if the horn sheath was routinely cast? From Wikimedia, uploaded by Biodiversity Heritage Library, CC BY 2.0.
A more illuminating insight may be that the correlates for Triceratops horn keratin are continuous with the epidermal correlates of the face (above). Horner and Marshall (2002) noted that the horn correlates for keratin sheathing extend over virtually the entire face - including the back of the frill (this is why so many Triceratops reconstructions have smooth 'face shields' nowadays). However, what's not seen on Triceratops horns is a boundary dividing the face sheath and a hypothetical temporary horn sheath, as might be expected where two keratinous sheaths meet (I'm assuming that the entire face shield wasn't shed annually either (palaeoartists: exfoliating/shedding Triceratops face - go!) - that's not a discussion I want to get into here).

A last, more arm-wavy point against horn shedding is that it is not at all common among living animals, possibly not even being present in some close pronghorn relatives (Janis et al. 1998). If Triceratops did shed its horns, it would be part of a club with very few members. This isn't a particularly scientific argument, but we have to concede that permanent horn sheaths are - by some way - far more common than ephemeral ones, and probably the 'default' condition for horned animals. Maybe we should assume permanence until there's good reason to think otherwise?

Could wear and abrasion create our morphing, dynamic horn sheaths in Triceratops? It's certainly true that keratin horns can be worn down, sometimes considerably. Bighorn sheep, for instance, can wear away years of horn growth in a behaviour known as 'brooming', but the results do not look like our palaeoart - in other words, they don't look like these sheep stuck their horns in a pencil sharpener. Nor do they echo the shape of the underlying skeleton. Instead, the ends are blunt, frayed and fractured (below). Any Triceratops that removed horn keratin through abrasion would presumably adopt a similarly 'sawn-off' appearance, and lack neat, pointed tips.

File:Desert Bighorn Sheep (8981484583).jpg
The broomed horns of a bighorn sheep (Ovis canadensis) - notice that they're heavily and deliberately worn at the tips, but they aren't shaped into fine points. From Wikimedia, uploaded by Lake Mead NRA Public Affairs, CC BY-SA 2.0.
Might a compound horn sheath be a route to horn sheath dynamism for Triceratops? Some readers may recall that we discussed compound keratin sheath covers last month and that they typically have deep grooves between abutting sheets. We don't see grooves of this nature on Triceratops skulls despite the very obvious rugosity profile created by the epidermal tissues, so I think we have to reject this hypothesis outright. The coverage of Triceratops horn core epidermal rugosities are pretty near identical to what we see on the horns of animals like cattle or goats, and I think we have to assume they indicate a similar, all-encompassing sheath morphology.

If Triceratops horns couldn't be renewed or take advantage of a more complex sheath arrangement, the likelihood of dynamic Triceratops horn sheaths is probably low. But does this idea of continuous sheath growth and twirly horns fare better under scrutiny? It seems to pass some basic tests, at least. The Triceratops brow horn outlines fit together pretty well with only a little displacement of the preceding horn layer, which is just what in see in modern horn growth, and the fact that their horn profiles don't change suddenly is consistent with them being perpetually constrained by layers of hard tissue. The predicted Triceratops sheath profile it is unexpected, but not beyond anything we see in living animals. And it scores points generally for being a simple model that is grounded in a well-understood aspect of living animal biology, in not needing to explain the loss of sheath tissue, and for factoring data we know is relevant to horn growth in living animals. I'm not saying this model is correct, but I am thinking that explains and fits our available data better than the dynamic sheath concept.

Of course, there are still lots of caveats. Remember that the model here is rough, being based on a generic Triceratops dataset and not the growth regime of a single species. The growth series outlined by Horner and Goodwin (2006) is a good general illustration of Triceratops growth, but results might vary if we restricted the data to a single species. My illustrations do not assume any exfoliation or tip abrasion, and we still don't have any idea what the external sheath morphology - including the presence of absence of ridges, spirals and bosses - might have been like. My attempt to stack the horn core profiles has also assumed minimal sheath thickness. If the sheath was thicker, the arcs of the horn could be stretched out over longer distances. So if you're buying this concept, remember that the horn shape proposed is only a general one - it's more in keeping with our understanding of sheath grow in modern animals, but it's still quite sketchy.


Perhaps the take-home message here is not, however, that Triceratops might have had loopy horns, but that there might be more to consider about fossil horn sheaths than we've assumed. Our discussion of dynamic horn sheaths does not just apply to Triceratops: artists take this approach with most horns and spikes in palaeoart, and it's clearly at odds with how most animals grow keratin sheaths today. But maybe this isn't just a topic for artists to ponder. There's potentially scope for a real study here and, seeing as fossil horn shape has a lot of functional significance, predicting sheath morphology would be a useful aid to predicting ancient behaviour. This needn't be restricted to horned dinosaurs, or even just horns, either: keratin sheaths on plates, spikes and so on grow in a similar way, and there's not reason this technique couldn't be used on other body parts, if validated. Moving this from food-for-thought-blog post to genuine science would require testing on modern species, perhaps through reconstructing living animal horns, to see how well it holds up. Recreating a schematic, 2D goat horn sheath using this method is fine, but real-world tests - especially using 3D horn casts, not just 2D drawings - might be more challenging. In the meantime, I'm curious to know what others think of all this - the comment field is open below...
"Hello, I'm Triceratops. I'll be your odd-looking concluding dinosaur reconstruction for this evening."

Enjoy monthly insights into palaeoart and fossil animal biology? Support this blog for $1 a month and get free stuff!

This blog is sponsored through Patreon, the site where you can help online content creators make a living. If you enjoy my content, please consider donating $1 a month to help fund my work. $1 might seem a meaningless amount, but if every reader pitched that amount I could work on these articles and their artwork full time. In return, you'll get access to my exclusive Patreon content: regular updates on research papers, books and paintings, including numerous advance previews of two palaeoart-heavy books (one of which is the first ever comprehensive guide to palaeoart processes). Plus, you get free stuff - prints, high quality images for printing, books, competitions - as my way of thanking you for your support. As always, huge thanks to everyone who already sponsors my work!


  • Farke, A. A. (2004). Horn use in Triceratops (Dinosauria: Ceratopsidae): testing behavioral hypotheses using scale models. Palaeontologia Electronica, 7(1), 1-10.
  • Goodwin, M. B., Clemens, W. A., Horner, J. R., & Padian, K. (2006). The smallest known Triceratops skull: new observations on ceratopsid cranial anatomy and ontogeny. Journal of Vertebrate Paleontology, 26(1), 103-112.
  • Goss, R. J. (2012). Deer antlers: regeneration, function and evolution. Academic Press.
  • Happ, J. W. (2010). New evidence regarding the structure and function of the horns in Triceratops (Dinosauris: Ceratopsidae). In: Ryan, M. H., Chinnery-Allgeier, B. J. & Eberth, D. A. (Eds.) New Perspectives on Horned Dinosaurs: The Royal Tyrrell Museum Ceratopsian Symposium. Indiana University Press. pp. 271-281.
  • Hieronymus, T. L., Witmer, L. M., Tanke, D. H., & Currie, P. J. (2009). The facial integument of centrosaurine ceratopsids: morphological and histological correlates of novel skin structures. The Anatomical Record, 292(9), 1370-1396.
  • Horner, J. R., & Goodwin, M. B. (2006). Major cranial changes during Triceratops ontogeny. Proceedings of the Royal Society of London B: Biological Sciences, 273(1602), 2757-2761.
  • Horner, J. R., & Marshall. C. (2002). Keratinous covered dinosaur skulls. Journal of Vertebrate Paleontology 22(3, Supplement):67A.
  • Janis, C. M., Manning, E., & Ahearn, M. E. (1998). Antilocapridae. In: Janis, C. M., Scott, K. M., & Jacobs, L. L. (Eds.). Evolution of tertiary mammals of North America: Volume 1, terrestrial carnivores, ungulates, and ungulate like mammals (Vol. 1). Cambridge University Press
  • O’Gara, B. W. (1990). The pronghorn (Antilocapra americana). In: Bubenik, G.A. & Bubenik, A. B. (Eds). Horns, pronghorns, and antlers: evolution, morphology, physiology, and social significance, Springer-Verlag. pp 231-264.
  • O'Gara, B. W., & Matson, G. (1975). Growth and casting of horns by pronghorns and exfoliation of horns by bovids. Journal of Mammalogy, 56(4), 829-846.
  • Paul, G. S. (2016). The Princeton field guide to dinosaurs. Princeton University Press.

Wednesday, 11 October 2017

The appearance and lifestyle of Thalassodromeus sethi, supercrested pterosaur

Thalassodromeus sethi, a juvenile Mirischia asymmetrica, and half a spinosaurid hang out in Cretaceous Brazil. The spinosaurid wants to go home.
One of my favourite pterosaurs is the Brazilian thalassodromid Thalassodromeus sethi: a large (4-5 m wingspan) Cretaceous azhdarchoid known only from a broken skull and cranial fragments of disputed affinity (Kellner and Campos 2002; Veldmeijer et al. 2005; Martill and Naish 2006). Characterised by an especially large bony cranial crest, toothless jaws and a robust skull construction, Thalassodromeus gained fame (and it's name, which translates to 'sea runner') from a presumed habit of skim-feeding. Long-time readers or pterosaur aficionados will know that multiple studies have suggested pterosaurian skim-feeding was unlikely on anatomical grounds (we discussed this most recently here and here) and was especially improbable for large species on account of the huge energy demands of ploughing large, blunt jaws through water (e.g. Humphries et al. 2007). A lack of skim-feeding habits does not make Thalassodromeus any less interesting however: it's a large, charismatic animal with a heavy dose of pterosaur weirdness, so there's still plenty to like. I recently had reason to overhaul the Thalassodromeus painting from my 2013 book (above) and took the opportunity to revisit my understanding of this animal's anatomy. The process had me fall for Thalassodromeus' cresty charms all over again, and I've taken this as impetus to share the love here.

The continuing puzzle of the Thalassodromeus skull

Thalassodromeus sethi skull elements as figured in Witton (2013). Note how the holotype skull is a giant jigsaw with well- and ill-fitting elements. The little (drawn) jaw to the left is no longer referred to Thalassodromeus, but is now the holotype of the dsungaripterid Banguela oberlii. This photo composite was created using photographs provided by the excellent Andre Veldmeijer and Erno Endenburg. 
The holotype skull of Thalassodromeus is pretty well preserved as pterosaur fossils go, but isn't quite as exceptional as it first appears (above). Though three dimensionally preserved and uncrushed, it's suffered damage in several areas and is broken into multiple pieces, some of which are ill-fitting with the rest of the skull or are missing entirely. It's a jigsaw puzzle which is complete enough to get the general picture of the skull shape, but some large areas remain open to interpretation. Pterosaur literature records that different bits of this specimen were once scattered across American research institutions and we have to hope that some of the last missing elements are still in a drawer somewhere, waiting to be reunited with the rest of the skull.

That the shape of the Thalassodromeus skull is somewhat ambiguous is evident by our history of T. sethi skull reconstructions (below). The first reconstruction - published in Kellner and Campos (2002) - is a little odd in that it shows a downturned, irregular upper jaw with a straight mandible. It also features 'classic' structures that we've come to know and love in this species: that badass 'V'-shaped chunk missing out of the back of the crest, a boss-like structure on the upper jaw, and a partly hooked mandibular tip. This reconstruction has always looked a little odd to me because I'm not sure how the animal is meant to close its mouth. A second reconstruction, which I presented in my 2013 book, was similar to the first except for showing both jaws as straight, without a downturned upper jaw. My logic was that Thalassodromeus should look something like the better known thalassodromid Tupuxuara, which has entirely straight jaws. Later, Headden and Campos (2015) presented a third interpretation, where the mandible was bent down at the base of the mandibular symphysis. Jaime Headden's (as far as I know unpublished) skull reconstruction hints at further differences from previous reconstructions, including a lack of that cool 'V' notch in the back of the crest.

Select T. sethi skull reconstructions, with my latest take at the bottom right. All three agree on some aspects of basic morphology, but there's not quite enough data to eliminate some possibilities of jaw and crest shape. Note that the 2017 skull outline is pretty conservative - the crest may have been longer and taller.
Which of these, if any, is correct? We await a comprehensive description of the skull to fully augment our understanding of T. sethi anatomy but, based on published information, it's likely that some of our earlier interpretations were erroneous. The gnarly crest shape drawn by Kellner and Campos (2002) probably takes damaged margins and missing elements too literally - this includes that awesome-looking V-shaped notch at the end, which is likely just another chunk of missing crest (this is certainly reported by colleagues who've examined the skull first hand). There's also no obvious reason why the mandible should be restored with an upturned tip. This interpretation was at least partly fuelled by an upturned jaw tip once referred to Thalassodromeus (Veldmeijer et al. 2005), but this specimen has since been considered a new genus of dsunagripterid pterosaur (Headden and Campos 2015).

It's also looking possible that - as indicated by Headden and Campos (2015) - both sets of Thalassodromeus jaws were downturned. It's difficult to be confident about any jaw reconstruction in this animal because these regions are not well represented in the holotype skull, but preserved elements of the upper and lower jaw margins imply a subtle downturn at the base of the rostrum and mandibular symphysis (and no, this isn't an effect of distortion or damage). Either Thalassodromeus had some sort of wibbly jaw shape or else it had a downturned jaw similar to azhdarchoids such as Tapejaridae* and Caupedactylus**. Whether this is convergence or further evidence of a close relationship between thalassodromids and tapejarids depends on your take on azhdarchoid interrelationships - this is still an area of disagreement that would benefit from dedicated investigation.

*of which thalassodromids - or thalassodromines - may, or may not be, a subdivision of. Ah, pterosaur phylogeny...

**I'm as confident as I can be that Caupedactylus is synonymous with my own "Tupuxuara" deliradamus. I should really write this up one day...

But hey, evidence for facial tissues and life appearance!

Thalassodromeus has some interesting features which allow us to reconstruct some aspects of its facial anatomy in detail, even in lieu of soft-tissue preservation. The crest of Thalassodromeus is marked by very conspicuous neurovascular grooves which were linked to a thermoregulatory function by Kellner and Campos (2002). They look pretty near identical to the sorts of branching grooves you find under bird beaks however (below), and my suspicion is that they're not a specialisation for controlling body temperature but simply a correlate for a keratinous sheath (Hieronymus et al. 2009). Similar grooves are seen on crestless parts of pterosaur jaws (the holotype of Serradraco sagittirostris has some especially obvious ones, for instance - see Rigal et al. 2017) as well as under the keratinous horns and beaks of animals everywhere. We don't need to imagine a unique function for these grooves just because they're on a big pterosaur crest, they're a standard variant of tetrapod skull anatomy.

Branching neurovascular networks on the Thalassodromeus crest - this is the region above the eye and posterior end of the nasoantorbital fenestra. Note the conspicuous groove crossing across the photo - this is the boundary between the premaxilla and underlying skull bones. From Kellner and Campos (2002).
Keratinous sheaths can have sharp margins which leave signature textures on the underlying skull. Bony steps or 'lips' can mark the transition to another tissue type, or a groove may form where one sheath plate abuts another. Both are evident in bird species which have beaks composed of multiple plates instead of a single keratinous covering (below), and we can look for similar features in fossil skulls to make predictions about life appearance. In Thalassodromeus we see a deep groove running along the boundary between the large premaxillary bone (the bone which makes up the jaw tip and top region of the entire crest) and the frontoparietal region (the base of the crest from the eye region backwards). Correlates for keratinous sheaths occur on both sides of this groove, so there's a chance that the crest covering was a compound structure composed of two abutting sheaths rather than one continuous one. If so, we might have been able to see this join on the live animal, just as we see the joins on the beaks of certain birds.

Gannet (Morus bassanus) skull with keratinous sheaths removed. Note the branching neurovascular impressions and deep grooves that mark the position of keratinous sheaths - we would predict a compound beak from these textures if we only knew gannets from fossils.
Can we test this idea? We could chop up our super-rare Thalassodromeus specimens to see if  histological data matches the surface texture interpretation (it's not only bone surface texture which records epidermal types - see Hieronymus et al. 2009) but I'll wager that most folks don't want the Thalassodromeus holotype carved up any more than it already is. Happily, there are other lines of data that might help us out. The first is the presence of the crest groove itself. Pterosaur skulls are normally devoid of sutures between bones because, in adults, they fuse so solidly that all trace of the original bone outlines is obliterated. Thus, the presence of a conspicuous groove in a mature Thalassodromeus specimen indicates that something unusual was happening, and influence from facial tissues is a well-known phenomenon that could explain this feature.

Schematic take on thalassodromid crest growth, from Martill and Naish (2006). The crest doesn't begin fully formed in juveniles, with the premaxillae (dark shading) having to overgrow the rest of the skull. Fun fact: my first ever PR palaeoart, now 11 years old, was to publicise this study.

A second line of support stems from studies into thalassodromid crest growth (Martill and Naish 2006). The "upper" (or premaxillary) component of the thalassodromid crest does not cover the skull in juveniles: rather, it has to overgrow the skull as the animal ages (above and below). Keratin sheaths are difficult to modify once formed because they're thick and inert (Goss 2012), so it's likely that parts of the premaxillary sheaths formed in juveniles migrated with the bone over the skull, meeting their counterparts at the skull posterior in later life. If the sheaths couldn't join once they met because they couldn't be modified or resorbed, they probably continued to grow as a compound cover, explaining the retention of an obvious groove between the two crest-forming bones. I find this idea pretty neat. Features like grooves on beaks or crests are nuances of animal appearance that are mostly lost to time but are important to characterising the appearance of living species. The idea that Thalassodromeus (and probably thalassodromids) had this feature makes them that little bit more real. Painting the images for this post certainly felt a little more like painting an animal than illustrating a hypothesis, just because of this detail.

Thalassodromid crest growth and compound keratinous sheathing, modelled by T. sethi. Note how the juvenile has an obvious 'two part' crest composition, and that the front/upper part (the premaxilla) sits on top of the posterior (frontoparietal) elements. With enough time, they form the monster-sized crest we know from big thalassodromid specimens. See Martill and Naish (2006) for more details.

Skull mechanics and lifestyle

It would be remiss to write about Thalassodromeus without mentioning its robust skull construction. The skull is proportionally wide, has especially deep jaws, a partly sealed orbit region, and the mandibular symphysis has a robust 'teardrop' cross section instead some flimsy crest. Its robustness is especially obvious when compared to the skull of the otherwise similar Tupuxuara (below), which has more typically open and airy pterosaurian cranial architecture. Thalassodromeus thus has a skull which looks like it could take a little more punishment than that of an average pterosaur, and this correlates nicely with observations that the regions for jaw adductor muscles are expanded on both the skull and lower jaw (Witton 2013; Pêgas and Kellner 2015). It's unsurprising that foraging hypotheses for Thalassodromeus have favoured forceful feeding habits such as skim-feeding (Kellner and Campos 2002) or being a predator of small-to-medium animals in terrestrial settings (Witton 2013).

Tupuxuara leonardii skull and mandible - looking pretty slender compared to the star of this post.
The possibility of downturned jaws in Thalassodromeus becomes especially interesting in light of its robust skull. Long, curving bones are a biomechanical paradox because they're weaker in compressive loading than a straight equivalent. This is, in part, because applying loads directly to both ends of a curved bone induces bending stresses even though the bone is not being bent in a traditional fashion. This is why big, slow animals tend to have straighter limb bones than smaller ones: they benefit from the increased strength of straight shafts, and they load their limbs in compression virtually all the time. From this perspective, the curved jaw of Thalassodromeus might seem like a disadvantage, being weaker under compression than that of a straight jawed animal. If striking violently at prey head on, the straight jawed species might be less likely to go home with a broken jaw.

However, curved bones are superior to straight bones at handling unpredictable, dynamic stresses. Curvature introduces predictability to stress distribution throughout a bone shaft, so they behave more reliably under a variety of loading regimes, be it compression, bending or twisting. A bone which responds to stress in the same way no matter how you deform it is easier to manage behaviourally, and to optimise mechanically, than a straight bone, and loss of raw strength created by bone curvature can be compensated for by modifying cross sections, shaft diameters and internal reinforcement (Bertram and Biewner 1988). These attributes have not been ignored by evolution and, in fact, most animal limb bones are curved to some degree to take advantage of these effects (Bertram and Biewner 1988). The superior compressive performance of a straight bone may not be as advantageous as the reliability and potential all-round stress resistance of a curved variant so, in simple terms, if you're planning some crazy stunts with your long bones, you want curved bone shafts, not straight ones.

A curved jaw thus complements the strong skull and jaw muscles of Thalassodromeus. If Thalassodromeus used foraging mechanics which were forceful or violent - such as catching big or powerful prey types, or using its beak to batter or tear at other animals - a curved beak may have served it well. This jaw shape - assuming we've interpreted it correctly, remember - could be further evidence of foraging habits at the more explosive and exciting end of the pterosaur ecological spectrum. Exactly what Thalassodromeus did for a living remains unknown, but it's hard not to compare these cranial features with other ideas of robust, terrestrial azhdarchoid predators - maybe this 'large pterosaur predator' niche has a longer roster than we've traditionally thought.

Hypothesis B: spinosaurids were allergic to curved jaws. Hey, it could happen.
Thalassodromids and their azhdarchoid kin are exceptionally interesting animals and we could probably talk about them all day, but we'll have to stop there. Coming soon: pterosaurs from the other end of the pterodactyloid spectrum, or a return to the world of extinct mammals. Probably.

Enjoy monthly insights into palaeoart and fossil animal biology? Support this blog for $1 a month and get free stuff!

This blog is sponsored through Patreon, the site where you can help online content creators make a living. If you enjoy my content, please consider donating $1 a month to help fund my work. $1 might seem a meaningless amount, but if every reader pitched that amount I could work on these articles and their artwork full time. In return, you'll get access to my exclusive Patreon content: regular updates on research papers, books and paintings, including numerous advance previews of two palaeoart-heavy books (one of which is the first ever comprehensive guide to palaeoart processes). Plus, you get free stuff - prints, high quality images for printing, books, competitions - as my way of thanking you for your support. As always, huge thanks to everyone who already sponsors my work!


  • Bertram, J. E., & Biewener, A. A. (1988). Bone curvature: sacrificing strength for load predictability?. Journal of Theoretical Biology, 131(1), 75-92.
  • Headden, J. A., & Campos, H. B. (2015). An unusual edentulous pterosaur from the Early Cretaceous Romualdo Formation of Brazil. Historical Biology, 27(7), 815-826.
  • Hieronymus, T. L., Witmer, L. M., Tanke, D. H., & Currie, P. J. (2009). The facial integument of centrosaurine ceratopsids: morphological and histological correlates of novel skin structures. The Anatomical Record, 292(9), 1370-1396.
  • Humphries, S., Bonser, R. H., Witton, M. P., & Martill, D. M. (2007). Did pterosaurs feed by skimming? Physical modelling and anatomical evaluation of an unusual feeding method. PLoS Biology, 5(8), e204.
  • Goss, R. J. (2012). Deer antlers: regeneration, function and evolution. Academic Press. 
  • Kellner, A. W., & de Almeida Campos, D. (2002). The function of the cranial crest and jaws of a unique pterosaur from the Early Cretaceous of Brazil. Science, 297(5580), 389-392.
  • Martill, D. M., & Naish, D. (2006). Cranial crest development in the azhdarchoid pterosaur Tupuxuara, with a review of the genus and tapejarid monophyly. Palaeontology, 49(4), 925-941.
  • Pêgas, R. V., & Kellner, A. W. (2015). Preliminary mandibular myological reconstruction of Thalassodromeus sethi (Pterodactyloidea: Tapejaridae). Flugsaurier 2015 Portsmouth, abstracts, 47-48.
  • Rigal, S., Martill, D. M., & Sweetman, S. C. (2017). A new pterosaur specimen from the Upper Tunbridge Wells Sand Formation (Cretaceous, Valanginian) of southern England and a review of Lonchodectes sagittirostris (Owen 1874). Geological Society, London, Special Publications, 455, SP455-5.
  • Veldmeijer, A. J., Signore, M., & Meijer, H. J. (2005). Description of two pterosaur (Pterodactyloidea) mandibles from the lower Cretaceous Santana Formation, Brazil. Deinsea, 11(1), 67-86.
  • Witton, M. P. (2013). Pterosaurs: natural history, evolution, anatomy. Princeton University Press.

Sunday, 24 September 2017

The horns of Arsinoitherium: covered in skin or augmented with keratin sheaths?

1.5 Arsinoitherium zitteli trotting about Eocene Egypt, looking a bit like they could be advertising farm products. But what's with those more elaborate than usual horns?
The horns of the giant, Egyptian, Oligocene afrotherian Arsinoitherium zitteli are probably a key factor in its status as one of the better known fossil mammals. Though perhaps not quite as popular as mammoths or sabre-toothed cats, this 3 m long, four-horned species has enough osteological charisma to warrant display in many museums as well as starring roles in books and films (including, cinema fans, narrowly missing out on an appearance in the 1933 King Kong). And unlike a fossil rhinocerotid (to which it is not at all related), Arsinoitherium doesn't need us to imagine the shape of its ornament in life: two enormous horns project over the end of the snout and another pair of smaller, sub-vertical horns grew above the eyes.

Recently, I painted a portrait of Arsinoitherium for an upcoming book project and, based on my understanding of epidermal osteological correlates, I threw a keratinous sheath over the entire horn set (below). This is not a typical reconstruction - Arsinoitherium has been reconstructed with 'regular' mammalian skin (perhaps better termed 'villose skin' - Hieronymus et al. 2009) on its horns for decades but, as we all know, popularity and longevity don't always equal 'credibility' when it comes to fossil animal reconstructions.

Arsinoitherium zitteli, sporting antelope-like horn sheaths.
Shortly after this image was shared online, Darren Naish, he of Tetrapod Zoology (and the upcoming TetZooCon meeting, which you should definitely attend if you're in the UK and reading this article), had a question: had I checked horns without keratinous sheaths, like deer antlers or giraffe ossicones? It turns out that these are the more typical artistic analogues for Arsinoitherium horns, and their reconstruction without a keratinous sheath reflects this interpretation. It wasn't a question I could easily answer because I'd zeroed in on a keratinous sheath quickly in my research for the image and, in a major palaeoart faux pas, hadn't given due consideration to other options. Simultaneously, neither of us could argue for any model of Arsinoitherium horn coverage confidently because no-one has looked into this in any detail. There are some ideas in the literature, but they are fleeting and conflicting (keratin sheaths - Anonymous 1903; Andrews 1906; Osborn 1907; or skin, Prothero and Schoch 2002; Rose 2006).

It's difficult to turn away a good palaeobiological mystery, and because I like to make sure my work is as credible as it can be, I followed this question up with more research. I reasoned that the structure, development and surface texture of the three major types of mammalian headgear - horns, ossicones and antlers - could be compared to Arsinoitherium horns to see which, if any, is the best match and indicator of life appearance. Looking into this has been very informative and might be of interest to fellow palaeoartists as well as those interested in cool fossil animals, so I thought I'd share my thoughts and process here. We'll start by looking at Arsinoitherium horns themselves, then move through modern potential analogues, and finally compare them at the end to see which model seems most apt.

Arsinoitherium horns: growth, structure and surface texture

PV M 8463, the most famous of all Arsinoitherium skulls, as illustrated in Andrews (1906). Note the dotted lines across the horns - they mark the end of the preserved skull and the start of reconstructed elements.
As noted above, Arsinoitherium has two pairs of horns: a larger anterior set, which grows out of the nasal bones and over the snout, and a smaller, second pair formed from the frontal bones, above the eyes. Both sets are highly conspicuous and dominate the skull, the weight of the anterior pair presumably accounting for the development of a bony bar between the nostrils in mature animals (Andrews 1906; Court 1992). Note that the Arsinoitherium horns we're used to seeing in museums are partly reconstructed and thus of limited use as reference material. Most exhibited skulls are based on NHMUK specimen PV M 8463 (above), a 'moderately sized' adult specimen (Osborn 1907) in which neither horn is complete (Anonymous 1903; Andrews 1906). This skull was among the earliest Arsinoitherium skulls collected from Egypt but was restored rapidly once it arrived back in London. A 1903 report describes how the skull was:

"...brought home by Dr. Andrews from Egypt, and after cleaning, strengthening, and the restoration of parts deficient on the left side by modelling from the right side, is now exhibited in the central hall of the Natural History Museum in Cromwell Road."
Anonymous, 1903, p. 530

The fact that some parts of the skull were in less than stellar shape is evident from this photo of PV M 8463 (from the NHM's data portal): note the variation in colour and texture, reflecting places of reconstruction against real bone. Thus, while the familiar Arsinoitherium museum skull is a useful reference for morphology, illustrations and descriptions in technical literature will be more informative for reconstructing their integument. I've based my assessment mostly on Charles Andrews (1906) monograph, as well as that of Court (1992).

Structure. Both horn pairs of Arsinoitherium are relatively simple in gross shape and maintain the same basic morphology throughout their lives (below), though the horns of mature animals are wider, taller and more pointed than those of juveniles. The figures presented in Andrews (1906) show an increase in anterior horn base length from 41.6% in the smallest specimen to over 56% in the largest. Both horn sets are hollow, with vast internal cavities being supported by sheets of trabecular bone. In some places the exterior bone walls are surprisingly thin, only 5-10 mm (Andrews 1906).

Arsinoitherium zitteli skull ontogeny. I wonder if the horns of the largest skull should be reconstructed as longer and taller, given their arcs in the completely known skulls and gentler tapering of other nasal horn specimens (e.g. Sanders et al. 2004). Skull drawn from Andrews (1906), skull measurements by me.
Surface texture. The base of the horns are marked by deep, broad and branching neurovascular channels running from the facial region onto the horns themselves. The horn shafts are rugose on account of many deep pits, grooves and branching channels aligned along their long axes (Andrews 1906; Sanders et al. 2004). The horn tips of young animals have an especially spongy texture at the tip, presumably reflecting growth of the horn core (Andrews 1906). These textures are not typical of the rest of the skull, which are of a more typical, smooth mammalian variety even in regions where skin was probably in close proximity to the bone (e.g. the zygomatic arch, over the braincase). This is an important distinction, implying that a different epidermal configuration - different skin types, in other words - was present on the horns compared to the rest of the skull.

Having learned something of the Arsinoitherium condition, let's take a look at how modern horns, antlers and ossicones compare...

Analogue 1. Bovid-style horns (keratinous sheaths over a bone core)

Bovid horns typify a widely used approach to cranial ornamentation and weaponry across Tetrapoda. They are perhaps the simplest approach to producing a sturdy cranial projection, being little more than a bony horn core covered in a hard keratinous sheath and are permanent feature in almost animals that bear them. The one exception is the pronghorn, which sheds its horn sheath annually (it also isn't a bovid). Biology, eh: can't we have one rule without an exception?

Bovid (bighorn sheep, Ovis canadensis) horn anatomy. From Drake et al. 2016.
Structure. Bovid-style horns are composed of a hollow bony core lined with trabeculae that strengthens an otherwise thin-walled structure (Drake et al. 2016). The bone portion only occupies the basal portion of the horn, anchoring ever-growing bands of keratin that grow from the bone-keratin interface, not at the horn tip (below). This means that the tip of the horn sheath is the oldest part of the structure and that the base of the sheath is the youngest. Because keratin sheaths are inert, dead and tough tissue, they cannot be remodelled once they are formed. This dictates that the growing bony core has to forever comply with the shape of the horn sheath and cannot change shape much over time. Size changes can be accommodated as wider and longer sheath layers can cover expanding horn cores, but it is not possible to form a more complex shape - say a branch or spur - at the tip of the horn. And before anyone mentions pronghorns: their horn branches are entirely soft-tissue: the bony core retains a simple shape.

Schematic bovid horn growth, adapted from an illustration in Goss (2012).
Surface texture. Deep, oblique foramina and branching neurovascular canals characterise the surface texture of bovid horn cores. This rugosity profile is most pronounced in younger animals, but is maintained to a lesser extent in adults - in many bovids, the horns never stop growing, they just slow down a great deal. This texture is not unique to horns but accompanies many structures with keratinous sheathing, including claws and beaks (e.g. Heironymus et al. 2009). A sharp lip and particularly deep rugosity can mark the transition from horn to facial skin.

Analogue 2. Giraffe ossicones (skin over ossified dermis)

Giraffes have awesome skulls with two - and often more - ossicones that are covered in the same skin as the rest of their faces (Davis 2011). Their approach to cranial ornamentation seems unique to giraffes and their fossil relatives but might be an apt model for aberrant extinct forms, so is worth reviewing here. Clive Spinage (1968) provides an excellent overview of ossicone structure and development: the following is taken from his work.

Structure. Ossicones are low humps or columnar protuberances, continuous with the surrounding skull anatomy but formed from dermal ossifications, not outgrowths of skull bones. They eventually fuse with the skull in adult life but, unlike the underlying skull bones, ossicones are solid and very dense - they are described as having 'ivory-like' in compactness and hardness by Spinage (1968). Mature specimens show increasingly complex shapes including development of swollen tips on the frontoparietal 'horns', as well as hornlets and bosses across the major 'forehead' ossicone. Having an adaptable, living integument is essential to this process, as the ossicone covering needs to change shape to reflect the changing size and complexity of the underlying bone.

Giraffe skulls are full of sinuses, but they do not extend to their ossicones, which are extremely dense. From Spinage (1968).
Surface texture. Generally smooth with oblique foramina in juveniles and young adults, but increasingly gnarly in mature animals (more so in males). The continued ossification of dermal tissues produces a conspicuous pitting and 'flaky' rugosity profile that overgrows the surrounding skull bones and obscures the textures from earlier growth stages. In mature males, this rugosity can overgrow the entire upper surface of the skull and enhance the height and ornamentation of the ossicones considerably.

Young adult male giraffe skull by Wikimedia user Nikkimaria, CC BY-SA 3.0. Note the flaky, irregular textures of the ossicones and their complex shape: they are much more intricate and developed than those of less mature animals. There's room for more irregularity and texture on this skull, too: the skulls of old males look like they have cathedral spires growing from their faces.

Analogue 3. Deer antlers (bony projections atop cranial pedicles)

The familiarity of deer antlers allows us to forget what remarkable and unusual structures they are. Present almost universally in male deer (and in female reindeer), these elaborate, sometimes enormous structures are cast and regrown each year using a regenerative process that is the source of much anatomical and medical interest - no other mammal can regenerate such a complex appendage in this way, and the speed of the regeneration process is remarkable. Antlers are so unusual that they are only partly useful to our discussion here: we are primarily interested in antlers when they are covered in their velvet (specialised antler skin), as this is most comparable to the likely Arsinoitherium condition. Antler skin itself is interesting as, although it is continuous with the skin of the underlying pedicle, it lacks sweat glands and arrector pili (the tiny muscles that pull hair up or give us goosebumps) (Li and Suttie 2000). The antler pedicle (the permanent bony base) in contrast, is covered in the same type of skin as elsewhere on the body (Li and Suttie 2000).

A happy-looking moose (Alces alces) with his fuzzy antlers. Note the visible blood vessels on the underside of each palm. Photo by AlbertHerring, in public domain.
Structure. Both antlers and pedicles are solid, and antlers can - by virtue of growing at their tips - become more complex as they grow, developing from single spurs into networks of brows, tines and palms. As with giraffes, antler skin needs to be living and adaptable to facilitate this: a covering of inert keratin would preclude this form of growth.

More Alces antlers, this time without velvet. Note the long, branching channels. By Wikimedia user Nkansahrexford, CC-BY-4.0.

Surface texture. Antlers have variably developed rugosities consisting of conspicuous, long and branching channels impressed into smooth bone or around prominences and tubercles. These grooves are the impressions of blood and nervous networks that facilitated rapid antler growth. These textures are easily discerned even from a distance, and thus contrast with the texture of the pedicles, which are smoother and lined with relatively shallow, narrow and long impressions of vascular networks. It is unusual for hairy skin to leave such a significant osteological scar on underlying bone: typically, this form of epidermis leaves little to no remnant on skull bones (Hieronymus et al. 2009).

Arsinoitherium vs the analogues

Having looked at three major types of cranial projection in living animals, which - if any - best match the condition in Arsinoitherium? Giraffe ossicones are incomparable to Arsinoitherium horns in several aspects, perhaps the most significant being their increasing complexity and development of flaking bone textures in later life. Furthermore, the development of giraffe ossicones from bony growths in dermal tissues suggests a fundamentally different relationship between skull and dermis than of Arsinoitherium, where the bony horn component represents skull bones alone. There's enough differences here to question whether giraffe ossicones are a good model for the life appearance of Arsinoitherium horns.

In being formed of polished, deeply vascularised bone, deer antlers are closer approximations of Arsinoitherium horns. However, there is so much weirdness associated with deer antler formation and tissues that they almost remove themselves from meaningful comparison to permanent skull horn cores. The fact that antler velvet, as hairy skin, is (to my knowledge) unique in leaving deep vascular channel impressions is a major issue here, implying that either antler bone is unusually susceptible to neurovascular imprinting (do they grow so fast that they grow around their blood vessels?) or that velvet is better at altering bone textures than other skin types. Both scenarios point to antlers having some endemic oddness about them, which complicates their use as a model for life appearance of non-antlered species.

All is not lost with the cervid data, however: antler pedicles are comparable to Arsinoitherium horns in being permanent outgrowths of bone, and they also have neurovascular impressions. However, these shallow grooves compare poorly to the deeper channels and pitting of Arsinoitherium horns. Indeed, there is little about antler pedicle texture to distinguish them from the surrounding skull bones, whereas the opposite is true for Arsinoitherium.

Our comparisons improve with the bovid horn condition, which seems to chime with the Arsinoitherium skull in many regards. Both are hollow outgrowths of skull bones supported by internal trabeculae; both have bone textures characterised by deep, bifurcating neurovascular channels as well as conspicuous longitudinal grooves and oblique foramina; and both maintain the same basic shape throughout growth - excepting some basic changes in base width and horn length. Further similarities include the development of particularly deep rugosties at the base of the horn cores, which is evident in at least large Arsinoitherium skulls (Andrews 1906). This interpretation is consistent with one of the longer (but still rather short, if we're honest) interpretations of the blood vessel impressions in Arsinoitherium:

"These channels evidently lodged blood-vessels which served for the conveyance of blood to or from the covering of the horn, and judging from the marked way in which both these vessels and those on the anterior face of the horns impress the bone, it seems probable that the covering was hard and of much, the same nature as that clothing the horn-cores of the cavicorn ruminants."
C. Andrews (1906), p. 7


Of the three models looked at here, it seems the basic structure and textural package of bovid-like horns best matches what we see in Arsinoitherium. Moreover, unlike the antler or ossicone models, there's no obvious mismatches with this configuration: pretty much everything we would correlate to a bovid-like horn anatomy seems present on or in the Arsinoitherium skull. The idea that a keratinous sheath might have existed in Arsinoitherium might seem odd, but it is not that outlandish given the apparent ease through which keratinous sheaths evolve. This is, after all, the tissue which has covered just about every claw, hoof, nail, horn, cranial dome and beak that has ever existed, whereas ossicones and antlers seem like specialised, clade-restricted approaches to cranial projections. The functionality of hollow Arsinoitherium horns is further reason to suspect a horn sheath. Studies of bovid horns suggest hollow cores and keratin sheaths compliment each other biomechanically, optimising the horns for for impact dissipation (Drake et al. 2016 and references therein). Stripped of a keratinous sheath, we find that hollow horn cores are great at transmitting energy but are brittle and prone to buckling and fracturing under heavy loading. It's only with a tough, fracture resistant keratin sheath that these structures can avoid breaking under heavy use so, if Arsinoitherium employed its horns for anything vaguely physically demanding, they probably needed a keratinous sheath.

It's possible, of course, that these structures were just for show, but they do look like they had a function beyond display. It occurs to me as I write this that this scene recalls the painting from Ghostbusters II. I guess we'll call this guy 'Vigo'. 
Putting all this together, I feel the case for a keratinous sheath over the Arsinoitherium horn sets is reasonable, at least so far as it can be made with publicly available data. Aspects of morphology, growth, surface texture and - perhaps - functionality seem fully consistent with a bovid-like horn configuration, whereas other potential models are less comparable. From an artistic perspective, this is exciting: horn sheaths can be extremely elaborate structures and exaggerate the size of the horn core considerably, so Arsinoitherium might have been far more extravagant in life than we have previously imagined. I've tried to hint at this with my reconstructions - remember, this animal wasn't just a funny-faced rhinoceros!

But - before we go crazy with this - do remember that the core of this analysis - the interpretation of Arsinoitherium headgear - is entirely literature based. I've not seen original specimens nor even modern, high-res imagery of an unreconstructed skull (this wasn't for lack of trying - the literature on these animals needs updating). Thus, while I've tried to be as thorough as I can with my observations, and as cautious as I can with my interpretations, I might be ignorant of some important detail. Take everything here with an appropriate pinch of salt, and please chime in below if you can provide superior insight. There's clearly scope for a more detailed study on this topic and, given how unique the horns of Arsinoitherium are, there might be some interesting functional findings to emerge from further investigation.

Enjoy monthly insights into palaeoart and fossil animal biology? Support this blog with a monthly micropayment, see bonus content, and get free stuff!

This blog is sponsored through Patreon, the site where you can help online content creators make a living. If you enjoy my content, please consider donating $1 a month to help fund my work. $1 might seem a meaningless amount, but if every reader pitched that amount I could work on these articles and their artwork full time. In return, you'll get access to my exclusive Patreon content: regular updates on research papers, books and paintings, including numerous advance previews of two palaeoart-heavy books (one of which is the first ever comprehensive guide to palaeoart processes). Plus, you get free stuff - prints, high quality images for printing, books, competitions - as my way of thanking you for your support. As always, huge thanks to everyone who already sponsors my work!


  • Andrews, C. W. (1906). A descriptive catalogue of the Tertiary Vertebrata of the Fayum. Publ. Brit. Mus. Nat. Hist. Land. XXXVII.
  • Anonymous. (1903). A New Egyptian Mammal (Arsinoitherium) from the Fayûm. (1903). Geological Magazine, 10(12), 529-532.
  • Court, N. (1992). The skull of Arsinoitherium (Mammalia, Embrithopoda) and the higher order interrelationships of ungulates. Palaeovertebrata, 22(1), 1-43.
  • Davis, E. B., Brakora, K. A., & Lee, A. H. (2011). Evolution of ruminant headgear: a review. Proceedings of the Royal Society of London B: Biological Sciences, 278(1720), 2857-2865.
  • Drake, A., Donahue, T. L. H., Stansloski, M., Fox, K., Wheatley, B. B., & Donahue, S. W. (2016). Horn and horn core trabecular bone of bighorn sheep rams absorbs impact energy and reduces brain cavity accelerations during high impact ramming of the skull. Acta biomaterialia, 44, 41-50.
  • Goss, R. J. (2012). Deer antlers: regeneration, function and evolution. Academic Press.
  • Hieronymus, T. L., Witmer, L. M., Tanke, D. H., & Currie, P. J. (2009). The facial integument of centrosaurine ceratopsids: morphological and histological correlates of novel skin structures. The Anatomical Record, 292(9), 1370-1396.
  • Li, C., & Suttie, J. M. (2000). Histological studies of pedicle skin formation and its transformation to antler velvet in red deer (Cervus elaphus). The Anatomical Record, 260(1), 62-71.
  • Osborn, H. F. (1907). Hunting the Ancestral Elephant in the Fayûm Desert: Discoveries of the Recent African Expeditions of the American Museum of Natural History. Century Company.
  • Prothero, D. R., & Schoch, R. M. (2002). Horns, tusks, and flippers: the evolution of hoofed mammals. JHU press.
  • Rose, K. D. (2006). The beginning of the age of mammals. JHU Press.
  • Sanders, W. J., Kappelman, J., & Rasmussen, D. T. (2004). New large-bodied mammals from the late Oligocene site of Chilga, Ethiopia. Acta Palaeontologica Polonica, 49(3), 365-392.
  • Spinage, C. A. (1968). Horns and other bony structures of the skull of the giraffe, and their functional significance. African Journal of Ecology, 6(1), 53-61.