Why we think giant pterosaurs could fly

Giant azhdarchid pterosaur in flight. Images like this virtually always trigger discussions about the validity of giant pterosaur flight hypotheses.

Every so often the idea of flightless giant pterosaurs circulates in the press or on social media. It doesn't take much to ignite these discussions: a new giant pterosaur fossil, a PR event from a museum, or simply artwork emphasising the size of giant flying reptiles will see someone, somewhere, questioning their flightworthiness. These suggestions are often made with strong conviction, to the extent of dismissing or even arguing with scientists who study pterosaur anatomy and biomechanics. After all, how can any sensible individual think that animals with 10 m wingspans and body masses hovering around 250 kg were capable of flight? At most they were gliders, or flighted as juveniles and flightless as adults, right?

Confession time: as someone actively involved in research and outreach on these animals, I often find these discussions frustrating, for two reasons. The first is that, among those who actually study pterosaur functional morphology – that is, those who make detailed observations and measurements of pterosaur fossils, compile biomechanical data and use computer modelling to objectively test their flight capacity – there is no controversy about the volant nature of these animals, even at their maximum size. Peer-reviewed claims that pterosaurs were flightless are genuinely rare (perhaps limited to Sato et al. 2009; Henderson 2010 and Prentice et al. 2011 in the last decade) and have a consistent record of being flawed on some critical anatomical or functional detail (Witton and Habib 2010). There is no debate about giant pterosaur flight among those of us who study their fossils: the press and social media fuss about the topic is a genuine palaeontological nontroversy.

The second source of frustration is that, away from technical literature, discussions of giant pterosaur flight frequently suffer from major cases of Dunning-Kruger effect, especially when parties have knowledge of planes. I've experienced this a lot in my career, and not just in the wilds of social media: many of my TV and film consultancy jobs have required defending basic tenets of pterosaur anatomy - even their basic, there-for-all-to-see proportions preserved in articulated fossils - to folks who just can't or won't believe what the fossils show. Having a casual understanding of engine-driven man-made flying machines does not equate to knowing all there is about everything that has ever flown, but you would not know this from some conversations.

Our controversial giant azhdarchid friends Arambourgiania philadelphiae (middle) and Hatzegopteryx thambema (right), compared to a record-breaking giraffe and the Disacknowledgement.

Whether through naivety of palaeontological theory or unwillingness to accept good data, the lack of accessible overviews of current thinking on giant pterosaur flight probably fuels this ongoing nontroversy. What's needed, it seems, is a synthesis of modern ideas on giant pterosaur flight and justification for why pterosaur experts don't challenge this idea. I've attempted this below, including sufficient methodological detail and references so that anyone wanting to understand these ideas will have a useful jumping-off point, and to establish what needs to be overturned to challenge the null hypothesis of giant pterosaur locomotion. Our focus will be on the largest of all pterosaurs, giant members of the clade Azhdarchidae, as they are the main subject of most flightless claims (above). They are also among the most familiar giant pterosaurs, their number including species such as Quetzalcoatlus and Hatzegopteryx. The general points made below pertain to all large pterosaurs, however.

Evidence from comparative anatomy

Giant azhdarchids are invariably known from scant remains, sometimes a handful of fragments representing bones from across the skeleton or, in the case of Quetzalcoatlus northropi, an incomplete left wing (e.g. Lawson 1975; Frey and Martill 1996; Buffetaut et al. 2002; Vremir 2010; Martill and Moser 2018). Ordinarily, fragmentary remains are a barrier to interpreting the locomotory strategies of extinct organisms but flighted lifestyles adapt animal bodies to such an extreme degree that just a few bones can betray volant habits. It’s evident that even the largest pterosaurs bore wing anatomy comparable to their smaller, incontrovertibly flightworthy relatives. Although no complete giant wings are known, our fragments indicate similar linear forelimb bone proportions to smaller azhdarchids. Their wing joints – including details of their elbows, wrists and wing finger knuckle – are well understood, and indicate typical properties of pterosaur wing motion and function. We can make a number of predictions concerning muscle extent for giant taxa, the most important being related to the presence of a huge deltopectoral crest on their humeri. This broad flange of bone, situated at the proximal end of the humerus, anchored many muscles running from the shoulder to the wing and powered flapping motions in flight (Bennett 2003), so is a clear correlate for powered flight in giant species. The seemingly-small deltopectoral crest on the Hatzegopteryx humerus is sometimes raised as evidence of reduced flight ability: it’s actually just badly preserved with lots of bone missing on all margins (Buffetaut et al. 2002).

A collection of giant azhdarchid bones: A, cervical vertebra of Arambourgiania philadelphiae; B, humerus of Quetzalcoatlus northropi, C-D, the rather broken proximal humerus of Hatzegopteryx thambema. Other than their huge size (scale bars represent 100 mm) and some details of robustness, these bones are identical to those of smaller, incontrovertibly flying pterosaurs.

The only significant difference between the wings of giant and smaller azhdarchids concerns bone robustness, especially that of their joints. We would predict expanded wing bone diameters in large fliers as they enhance resistance to bending, and as giant animals they are going to experience proportionally greater bending stresses. If they flew, giant pterosaurs should have very large wing bones indeed, and - as is evident from the adjacent images - this is exactly what we find. Witton and Habib (2010) noted that the humeral shafts of giant azhdarchids are comparable in diameter to those of giant mammals, like hippos, despite the pterosaurs being a fraction of their weight (Witton 2008; Henderson 2010). Giant azhdarchid wing bones perform exceptionally well in bending strength tests, being able to resist multiple body weights before failing (Witton and Habib 2010), and their expanded diameters maintaining relative failure levels comparable to those of small or mid-sized pterosaurs, despite their size (Witton et al., in prep).

Pterosaur humeral scaling: as pterosaurs got bigger, their wing bones and joints expanded disproportionately to accommodate greater stresses incurred in flight and launch. Image from Witton (2013).

But for all their expansion, giant azhdarchid wings retain exceptionally thin bone walls. Even in Hatzegopteryx, the most robust of the group, they’re only 4-7 mm thick. This is a high value for a pterosaur, but still (in relation to bone diameter) at the low end of the cortical thickness spectrum. Other giant species have cortices of just 2 mm or so, values only just slightly larger than those of mid-sized pterosaurs. Expanded but bone-lite wings are another feature of flying creatures, being common to most flying birds and all large pterosaurs, and offering their owners a lightweight but bending-resistant flight skeleton. However, what optimises these skeletons for flight compromises resistance to buckling forces, which is why most non-volant animals tend to have much thicker cortices. The correlation between thin bone walls and flight is not watertight (Hutchinson 2001) but it's a feature we would predict for any seriously large flying animal, and is thus consistent with volant habits in giant pterosaur species.

Flight models

It’s often asked how animals as large as the biggest azhdarchids could attain and sustain flight. It’s important to stress that no-one imagines giant azhdarchids as breezy fliers flitting around Cretaceous plains like busy songbirds. As animals operating close to the size limits of flight for the azhdarchoid bauplan (Marden 1994; Habib and Cunningham pers. comm. in Witton 2010; Habib 2013) we should assume a flight frequency comparable to our largest modern fliers – creatures like bustards, geese, swans, albatrosses and so forth. Though different in flight mechanics these birds are united in their relatively low launch frequencies, taking to the air when they must (such as to evade danger) or when they have long distances to travel. Launch is very energy-demanding because of their great body masses, and in some cases specific environmental conditions are needed (such as space for taxiing in albatross), limiting their options for frequent takeoff. We should assume the same was true for large azhdarchids: their functional morphology and trackways show strong terrestrial abilities (Hwang et al. 2002; Witton and Naish 2008, 2013) and they probably spent a lot of time grounded, only flying when harassed, or wanting to move far and fast.

When imagining giant pterosaurs flying, we need to have birds like the kori bustard in mind: large, powerful animals which are strong fliers, but unable to flit about the sky like small songbirds or bats. When these guys take off, they mean it. Photo by Arnstein Rønning, from Wikimedia, CC BY 3.0.

Indeed, in all likelihood giant pterosaurs couldn’t launch every few moments. Flying animals tend to allocate about 20-25% of their body mass to flight musculature, which gives our large azhdarchids 50 kg or so of flight muscle to use in launch and flight (Paul 2002; Marden 1994). Even so, models of muscle energy availability show that giant pterosaurs could not launch aerobically (that is, using muscle contractions supplied with oxygen) and they had to rely on stronger, but less endurable, anaerobic muscle contractions. Anaerobic muscle power is essential to launch in the largest birds and almost certainly played a role in extinct giant insect flight, too (Marden 1994, see graph below), so its inferred use in giant pterosaurs is quite plausible. This reliance on anaerobic muscle power would necessitate resting periods between launches (hence the inability to launch continuously like a small flyer) as well as after vigorous bouts of flapping. Witton and Habib (2010) predicted that the hard flapping window for a giant azhdarchid was about 90 seconds, after which a rest was needed. So, does that limit our giants to turkey-like burst flights?

Launch for giant azhdarchids - like Quetzalcoatlus northropi - would be no more challenging than it is for large birds. The dotted line on this graph represents the minimum muscle energy output needed for flight. Using the same mechanism of anaerobic muscle power as large living fliers, giant azhdarchids are on the right side of that line. From Marden (1994).

Probably not. One of the world's leading experts on animal flight, Mike Habib, found that Colin Pennycuick’s freeware Flight programme – software designed to model bird flight - can be easily modified to predict pterosaur gliding and soaring capabilities, even accounting for the differences between feathered and membranous wings (see Witton and Habib 2010 for this methodology). Using this software, Mike and I predicted that giant azhdarchids were supreme soarers, easily able to sustain long-distance gliding even at body masses of 180-250 kg (Witton and Habib 2010). Predicted giant flight velocities exceeded 90 kph and, in that 90 second flapping burst, giant azhdarchids would cover several kilometres - plenty of distance to seek areas of uplift such as deflected winds or thermals. Having located these, azhdarchids could easily adopt energy-saving soaring to recover their flight muscles, their glide ratios being consistent with those of large soaring birds such as storks, Procellariiformes and raptors. Mike has presented calculations that these giants would have sufficient on-board energy resources to travel the planet, their speed and flight range being sufficient to ignore most geographical barriers. Note that these models assume modern day parameters of atmospheric density and gravity: we do not need to modify these to keep giant azhdarchids airborne. Sure, if you did change these parameters you might make the job easier, but the giants already have very strong flight performance without it. If you don't buy this, remember that you can play around with Flight yourself: download the program, get the method and pterosaur parameters from our open access paper and go at it. None of the science behind these animals is mystical - the methods are entirely conventional and repeatable.

Too heavy to fly?

For all this talk of modelling pterosaur flight at quarter-tonne masses, two sets of authors have proposed that giant pterosaurs were simply too heavy to attain flight. Sato et al. (2009) based this on their understanding of procellariiform takeoff, modelling a maximum possible volant mass of 40 kg for these birds and assuming the same limit must apply to pterosaurs. The next year, Don Henderson (2010) compiled a series of volumetric estimates of pterosaur mass including a 450 kg Quetzalcoatlus. Don – probably correctly – assumed that such an animal would be too heavy to fly.

Mike and I addressed both these proposals in a 2010 publication about giant pterosaur flight. On Sato et al. (2009), we found numerous problems with the overt biomechanical links drawn between bird and pterosaur flight. Avian and pterosaur anatomy is comparable enough to assume some broad analogies in wing shape and flight styles (e.g. Hazlehurst and Rayner 1992), but the detailed kinematics of flight – including launch – are too distinct to assume that the size limits of one group apply to the other. There are reasons to think pterosaurs launched in a very different way to birds (see below) and were subject to a different set of scaling regimes and size limits (Habib 2008, 2013; Witton and Habib 2010). Sato et al. (2009) may have predicted a flight mass limit for long-winged, dynamically soaring birds, but the application of this limit to flying reptiles is not supported by our understanding of pterosaur and avian biomechanics.

Don Henderson's (2010) Quetzalcoatlus model compared to the articulated skeleton of the small, completely known azhdarchid Zhejiangopterus linhaiensis. Note the clear distinction in torso size, and the actual torso length of the fossil pterosaur compared to the humerus. Images from Henderson (2010) and Cai and Wei (1994).

We found a much simpler issue with Don Henderson’s half-tonne Quetzalcoatlus model: its body was simply too large. Don based his work on a silhouette in Wellnhofer’s (1991) pterosaur encyclopaedia, a reasonable decision given the paucity of reconstructions of this animal at the time, but ultimately a problematic one for making accurate mass estimations. Azhdarchoids were relatively poorly known in the early 1990s and Wellnhofer’s silhouette reflects this, being a mostly imaginary pterosaur only accurate in wingspan. Crucially, its body is monstrously oversized at 1.5 m long. Complete azhdarchoids discovered since this time, including that of the azhdarchid Zhejiangopterus, have shoulder-hip lengths only 30-50 % longer than their humeri (above) and - in lieu of giant pterosaur torso fossils - we have to assume this was true for the giants, too. The 544 mm long humerus of Q. northropi translates to a predicted torso length of just c. 750 mm – a fraction of the size used in Don’s estimate. Mike and I adjusted Don's calculations to a more reasonable body proportion and, presto, the predicted mass was in the more familiar quarter-tonne range, a value flight models are happy to see launching and soaring without difficulty (Witton and Habib 2010).

The key to everything: quad launch

A critical hypothesis for giant pterosaur flight concerns recent interpretations of their launch strategy. This idea is that pterosaurs – probably all of them – took off from a quadrupedal start, not a bird-like bipedal one. The quad-launch hypothesis has origins in technical literature dating back to 2008 (Habib 2008, 2013, Witton and Habib 2010) but has a longer history through Mike H’s and Jim Cunningham’s contributions to the Dinosaur Mailing List. In retrospect, quad launch can be seen as a unifying hypothesis in studies of giant pterosaur flight, the piece of the jigsaw that allowed us to see how data from comparative anatomy, body masses and relative bone strength fit together. Before quad-launch, pterosaur flight models struggled to transfer giant azhdarchids from the ground to the air and were forced to cap their body masses at unrealistically low values (e.g. 75 kg in Chatterjee and Templin 2004) in order to launch them like big birds. Other than the fact that masses of 75 kg are untenable for creatures the size of giraffes (they’d need to be something like 70-80% air; Witton 2008), bipedal launch models suffer from several biomechanical issues involving bone strength, limb bone scaling and muscle size, as well as inconsistencies concerning pterosaur gaits. It's these issues which Mike and Jim investigated in their studies, making them the first researchers to approach pterosaur launch with objectivity, rather than a priori assuming an avian launch model, and bending pterosaur palaeobiology until it fit.

Ratios of limb bone strength in birds and pterosaurs. Positive values trend towards strength in humeri vs. femora, while negative values skew towards stronger femora vs. humeri. I've left the caption on for greater explanation. From Habib (2008).

Let's unpack these points in a little more detail. Firstly, the main launch limbs of flying animals are - above body masses of 500 g - stronger than their non-launching counterparts, and scale with more pronounced positive allometry (Habib 2008). This reflects launch being the most demanding part of flight. Look closely at a launching animal (high speed video helps) and you'll see that flight does not begin with a flap, but a leap: something like 80-90 % of launch effort stems from a powerful jump initiated by the main launch limbs. This explains why birds have proportionally robust and strong hindlimb skeletons but relatively slender wing bones: as they increase in size, their legs must become proportionally stronger to initiate flight at greater masses (Habib 2008). Pterosaurs, in contrast, show the opposite condition: their forelimbs are larger and stronger than their legs, with this relationship increasingly pronounced in larger species. Mike's 2008 study quantified this distinction, showing that the section modulus - a value proportionate to the strength of a given cross section - is consistently larger in pterosaur humeri than femora, and vice versa in birds, and that these ratios are more extreme at larger body sizes (see diagram, above). We presented similar data highlighting distinction in limb bone scaling in our 2010 paper, as well as quantifying the relative weakness of azhdarchid femora - their neck vertebrae are actually stronger than this major limb bone. Their humeri, in contrast, were very strong even in giant taxa modelled at the upper limits of pterosaur mass estimates (below). This is already a strong sign of a forelimb-dominated launch strategy in giants, and there's more to consider yet.

Raw data on azhdarchid limb bone strength from Witton and Habib (2010). Note the 'avian expectation' column - pterosaur bones do not scale in the same manner as bird bones, indicating a different regime of biomechanical selection pressures, and thus different limits on parameters like size.

Secondly, the avian skeleton has two large girdles for limb muscles: an enlarged shoulder and chest region for flight muscles, and an enhanced pelvic region to anchor those powerful hindlimb launch muscles. Pterosaurs, in contrast, have only one large limb girdle - their shoulders, making this the de facto likely candidate for powering their launch cycles. Using volumetric modelling, Paul (2002) predicted that a giant azhdarchid would have space for 50 kg of muscle in their pectoral region, a value appropriate for initiating flight in 200-250 kg animals (Marden 1994). This strategy is a far more economical use of muscle mass because the same muscles that power flight can also initiate launch, thus allowing quad launchers to have smaller torsos - and thus lower masses - than bipedal launchers. For all their power, the moment birds have launched their legs are effectively useless - they're just dead weight to be hauled around until it's time to land.

Reconstructed skeletons of large and giant azhdarchids in quad-launch poses, from Naish and Witton (2017). Note how the nearly completely known Quetzalcoatlus sp. - D - E - lacks a large site for hindlimb muscles - that's typical of all pterosaurs, and an important argument in favour of quad launch.

This is an critical point for giant pterosaur flight as it allows us to make hypotheses about body size maximums related to launch strategy. Because quad launch is a mass-efficient route to flight we can hypothesise that quad launchers could attain much larger overall sizes and masses than bipedal launchers (Witton and Habib 2010; Habib 2013). As everyone knows, this is borne out in our fossil record of volant birds, which max out at 5-6 m wingspans and masses of 22-40 kg (Ksepka 2014), while giant azhdarchids attained wingspans of 10 m and 200-250 kg body masses. Mass-efficient launch mechanics is almost certainly a major factor in how azhdarchids became so big, especially combined with the exceptional azhdarchoid ability for skeletal pneumaticity (Claessens et al. 2009).

Fossil birds like Pelagornis sandersi are pretty big (extant bird with the greatest wingspan, the wandering albatross, shown top right), but they wouldn't be able to poke giraffes in the face when standing next to them. 5-6 m wingspans are the known size limit for bird flight, and their inefficient launch mechanism is probably the cause. From Ksepka (2014).

A further line of evidence for quad launch concerns pterosaur trackways. Habib (2008) also notes that launch in living tetrapod fliers correlates to terrestrial gait: the number of limbs used to locomote on the ground is the same as the number used to take-off. Birds walk and launch with two legs, while bats walk and launch using all four. An extensive record of pterosaur trackways shows that pterosaurs were quadrupedal animals like bats, and it stands to reason that they also launched from four limbs: they would contrast with our living fliers if they had to shift gaits to take off. Our pterosaur footprint record includes trackways of quadrupedal giant pterosaurs (Hwang et al. 2002), so we can comfortably extend this observation to them, too. Incidentally, the fact that several bats take off quadrupedally is often overlooked in discussions of pterosaur launch: bird-like bipedal launches dominate our consciousness only because we see them taking off every day, but they do not represent the only way tetrapods can become airborne.

Quad launch cycle in vampire bats Desmodus rotundus, traced from video footage: this is a real, proven launch mechanic folks, not something dreamt up by pterosaur workers desperate to prove giant azhdarchids could fly. Several other bats launch in this way, too. From Schutt et al. (1997).

These points - bone strength, concentrations of muscle bulk, limb bone scaling and trackway data - are the cornerstones of the pterosaur quad launch hypothesis, an idea which explains many independently observed features of pterosaur biomechanics, bone proportions and absolute size. Crucially, all of these points can be investigated for giant azhdarchids, and there are no red flags suggesting quad launch did not apply to these pterosaurs. We can thus assume that giant azhdarchids used the most efficient launch mechanism conceivable for a tetrapod, negating any need for unreasonably low mass estimates or cliff jumping to become airborne. They were simply 'extreme' versions of the pterosaur bauplaun, not evolutionary weirdos that take us back to the biomechanical drawing board.

Despite the sound scientific basis to quad-launch, it is sometimes dismissed out of hand, perhaps because many folks just can't imagine it working: this excellent video by Mike and Julia Molnar does a good job of showing the kinematics. My experience is that counter-arguments are made without knowledge of its supporting data as they focus on less knowable components of the launch cycle, such as the speed of wing action or intuitive ideas about how high pterosaurs could leap. These are poor arguments because they a) are largely speculative, and not based on measurable/observable phenomena like bone strength or trackways; and b) ignore the fact that any launch mechanic requires rapid deployment of wings or an ability to obtain good ground clearance. We have to assume giant pterosaurs could achieve these feats no matter what our preferred launch strategy is. Moreover, somewhat ironically, the elevated flight speeds necessitated by giant pterosaur mass actually minimises some of these concerns. Flapping amplitude scales negatively with animal size and flight speed, making ground clearance less of an issue for large fliers than smaller ones (Habib 2008).

Flying the gauntlet

Let's put all this together - congratulations if you've waded through this long, often technical post. Giant azhdarchids...

• are poorly known, but have anatomy consistent with volant habits in every known aspect.
• do not seem to have struggled with take-off energetics more than any other large flyer.
• are often anatomically mischaracterised, being overly compared to extant birds or modelled in ways which distort their likely flight parameters.
• evolved from animals with a fundamentally more efficient launch strategy than that of birds, which lifts their body mass ceiling well above that predicted for avians. Every tested aspect of giant azhdarchid anatomy points to retention of this launch strategy even at their huge sizes.
• have flight parameters which, when modelled using conventional animal flight software in modern-grade atmosphere and gravity, equate to excellent flight performance, analogous to that of large soaring birds.

The take-home message is that interpretations of giant azhdarchids as flying animals are based on numerous lines of investigation and hypotheses which support and predict one another. Moreover, the methods used in these studies are entirely conventional techniques of palaeontological inquiry, and to disregard or ignore them requires dismissal of entire scientific fields of study. Don't buy the limb bone strength studies? Fine, then you also don't buy beam theory or structural engineering. Don't believe the flight analyses? OK, but you're also challenging software written by noted experts in animal flight, using data measured from real flying animals and a deep understanding of aerodynamics.

This is not to say that we know all there is to know about giant pterosaur flight - far from it. They remain poorly known animals and we can only guess at their variation in flight performance. Who knows, maybe a flightless species will turn up one day - this is not a ridiculous concept, we just don't have any evidence for it yet. But, for now, anyone seriously wanting to challenge this interpretation needs to discredit a robust theoretical foundation of pterosaur flight mechanics and provide a superior interpretation of the many strands of evidence we've discussed. This seems like a tall order to me, but it's the gauntlet that anyone who says giant pterosaurs were 'too big to fly' or 'they needed different gravity' has to run. Such comments reflect an ignorance or unwillingness to engage with a growing body of sound technical research on these animals, and - unlike giant pterosaurs - these arguments just don't fly.

Enjoy monthly insights into palaeoart, fossil animal biology and occasional reviews of palaeo media? 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!

References

• Bennett, S. C. (2003). Morphological evolution of the pectoral girdle of pterosaurs: myology and function. Geological Society, London, Special Publications, 217(1), 191-215.
• Buffetaut, E., Grigorescu, D., & Csiki, Z. (2002). A new giant pterosaur with a robust skull from the latest Cretaceous of Romania. Naturwissenschaften, 89(4), 180-184.
• Chatterjee, S., & Templin, R. J. (2004). Posture, locomotion, and paleoecology of pterosaurs (Vol. 376). Geological Society of America.
• Claessens, L. P., O'Connor, P. M., & Unwin, D. M. (2009). Respiratory evolution facilitated the origin of pterosaur flight and aerial gigantism. PloS one, 4(2), e4497.
• Frey, E., & Martill, D. M. (1996). A reappraisal of Arambourgiania (Pterosauria, Pterodactyloidea): one of the world's largest flying animals. Neues Jahrbuch für Geologie und Paläontologie, 199, 221-247.
• Habib, M. B. (2008). Comparative evidence for quadrupedal launch in pterosaurs. Zitteliana, 159-166.
• Habib, M. (2013). Constraining the air giants: limits on size in flying animals as an example of constraint-based biomechanical theories of form. Biological Theory, 8(3), 245-252.
• Hazlehurst, G. A., & Rayner, J. M. (1992). Flight characteristics of Triassic and Jurassic Pterosauria: an appraisal based on wing shape. Paleobiology, 18(4), 447-463.
• Henderson, D. M. (2010). Pterosaur body mass estimates from three-dimensional mathematical slicing. Journal of Vertebrate Paleontology, 30(3), 768-785.
• Hutchinson, J.R., 2001b. The evolution of femoral osteology and soft tissues on the line to extant birds (Neornithes).Zoological Journal of the Linnean Society. 131, 169–197.
• Ksepka, D. T. (2014). Flight performance of the largest volant bird. Proceedings of the National Academy of Sciences, 111(29), 10624-10629.
• Hwang, K. G., Huh, M., Lockley, M. G., Unwin, D. M., & Wright, J. L. (2002). New pterosaur tracks (Pteraichnidae) from the Late Cretaceous Uhangri Formation, southwestern Korea. Geological Magazine, 139(4), 421-435.
• Lawson, D. A. (1975). Pterosaur from the latest Cretaceous of West Texas: discovery of the largest flying creature. Science, 187(4180), 947-948.
• Naish, D., & Witton, M. P. (2017). Neck biomechanics indicate that giant Transylvanian azhdarchid pterosaurs were short-necked arch predators. PeerJ, 5, e2908.
• Marden, J. H. (1994). From damselflies to pterosaurs: how burst and sustainable flight performance scale with size. American Journal of Physiology-Regulatory, Integrative and Comparative Physiology, 266(4), R1077-R1084.
• Martill, D. M., & Moser, M. (2018). Topotype specimens probably attributable to the giant azhdarchid pterosaur Arambourgiania philadelphiae (Arambourg 1959). Geological Society, London, Special Publications, 455(1), 159-169.
• Paul, G. S. (2002). Dinosaurs of the air: the evolution and loss of flight in dinosaurs and birds. JHU Press.
• Prentice, K. C., Ruta, M., & Benton, M. J. (2011). Evolution of morphological disparity in pterosaurs. Journal of Systematic Palaeontology, 9(3), 337-353.
• Sato, K., Sakamoto, K. Q., Watanuki, Y., Takahashi, A., Katsumata, N., Bost, C. A., & Weimerskirch, H. (2009). Scaling of soaring seabirds and implications for flight abilities of giant pterosaurs. PLoS One, 4(4), e5400.
• Vremir, M. (2010). New faunal elements from the Late Cretaceous (Maastrichtian) continental deposits of Sebeş area (Transylvania). Acta Musei Sabesiensis, 2, 635-684.
• Wellnhofer, P. (1991). The illustrated encyclopedia of pterosaurs. Crescent Books.
• Witton, M. P. (2010). Pteranodon and beyond: the history of giant pterosaurs from 1870 onwards. Geological Society, London, Special Publications, 343(1), 313-323.
• Witton, M. P. (2008). A new approach to determining pterosaur body mass and its implications for pterosaur flight. Zitteliana, 143-158.
• Witton, M. P., & Habib, M. B. (2010). On the size and flight diversity of giant pterosaurs, the use of birds as pterosaur analogues and comments on pterosaur flightlessness. PloS one, 5(11), e13982.
• Witton, M. P., & Naish, D. (2008). A reappraisal of azhdarchid pterosaur functional morphology and paleoecology. PLoS one, 3(5), e2271.
• Witton, M. P., & Naish, D. (2013). Azhdarchid pterosaurs: water-trawling pelican mimics or “terrestrial stalkers”?. Acta Palaeontologica Polonica, 60(3), 651-660.

Armoured theropod faces, rhino horns and pterosaur skin crests: how artists can predict elaborate skin structures in fossil animals

Abelisaurid Kryptops palaios recieves some TLC in Early Cretaceous Niger, while duck-faced Anatosuchus minor sneaks out of the water. But what's with the swollen, bulging look to the abelisaur's face? (Concept by Chidumebi Browne. Award yourself an extra biscuit if you spot the homage to one of my palaeoart heroes in this scene.)

A hot topic in modern palaeoart circles is the relationship between bone surface texture and soft-tissues. Specifically, artists are interested in what bone textures mean for skin composition and thickness, and whether it tells us anything about epidermal structures such as scales, feathers or hair. The idea that bone texture has a relationship to skin anatomy is not new, and palaeontologists have been linking details of fossil bones to beaks, horns, feathers and so on for many decades. Recent research on this matter is much more detailed and informed than previous efforts however, and uses careful comparisons of bone structure (both external and internal) in fossil and living species to make detailed predictions about the life appearance of long-extinct animals. Tobin Hieronymus and other individuals from the Witmer Lab (University of Ohio) have made some especially valuable contributions to this field, and their work adds to a growing literature that palaeoartists should consult to make credible restorations of past species. Palaeoart is long past the stage where we can doodle a rough outline around a skeleton and call it a day - more than ever, production of truly credible palaeoartwork is only possible after careful and thorough research.

One of the most interesting aspects of this recent work concerns a skin type that we rarely discuss for ancient animals: dense, stiff dermal tissues that forms thick armour in animals like hippos, mouse deer and pigs; and epidermal projections, such as horns and crests, in rhinos and birds. We might assume that these soft-tissue structures would leave little trace on bone and that we're ignorant of their presence in fossil animals until specimens preserving soft-tissues show us otherwise. However, this is not so. Work by Hieronymus (2009) and Hieronymus et al. (2006, 2009) shows that we can identify the presence of skin armour and epidermal projections without soft-tissue preservation. This has significant implications for how we might restore fossil animals, and artists should be on the lookout for features evidencing these structures when researching their reconstructions.

Reinforcing skin to make armour and skin projections

Before we get to the fossil examples, it will help to know how skin is armoured and epidermal projections are reinforced without the aid of bony material. For armoured skin, white rhinoceros hide provides a well-studied example. Here, the dermis is reinforced with densely packed collagen fibres criss-crossing one another in three perpendicular planes (Shadwick et al. 1992). This structure differs markedly from typical reptilian and mammalian skin (Hieronymus et al. 2010) and has correspondingly different skin mechanics. Not only is it considerably stiffer and highly resistant to tearing, but under compression it is stronger than cartilage (Shadwick 1992). White rhinoceros skin is, on average, 25 mm thick (though their belly skin is about half that measurement) and it serves them well at resisting damage during intraspecific bouts or when on the wrong end of a predatory act. Similar skin has convergently evolved in pigs, hippos, mouse deer and seals, these being species that engage in biting and stabbing fights and having obvious need for protective tissues.

Indian rhinoceros skin in all it's supercollagenous glory. Note the thick folding but otherwise sparse wrinkling, a consequence of poor elasticity in this skin type. From Wikimedia user Sanjay ach, CC BY-SA 3.0.

Soft-tissue crests, horns and other projecting structures can also be made of expanded, dense dermis (seen in comb ducks), or reflect enhancement of another skin tissue: the epidermis (Hieronymus et al. 2006, 2009). These epidermal elements anchor to underlying dermis and are formed of dense keratin matrices, producing ultra-tough cornified tissue not dissimilar in composition to beaks, claws, horn sheaths or baleen (Hieronymus et al. 2006). Some skin projections can incorporate non-keratinised components as well - rhino horns, for instance, have mineral and melanin components as additional stiffening agents (Hieronymus et al. 2006) - and the degree of keratinisation can vary, depending on the functional demands of the projection. The crests of white pelicans are a well-known example of these structures, and are noteworthy for their ephemeral nature. Unlike most epidermal outgrowths, pelicans shed and regrow these structures annually. That's food for thought for not only palaeoartists, but also those of us wondering if soft-tissue crests have significance to taxonomy. Would a fossil pelican with a rostral crest be considered a different species to one without? Quite possibly.

American white pelican with rostral crest, photographed by Travis Barfield. Photo from Wikimedia.

When either of these skin types overgrow bone, they leave clear traces on the bone surface texture and histology. The best places to look for these markers are animal skulls, as they are generally not separated from skin by layers of muscle and fat like postcranial bones, and they tend to be body parts that animals adorn with horns, crests and other epidermal outgrowths. Particularly good sites to check for skin-derived markers are the bones forming the cheek, those around the orbit, and along the forehead and snout, as these regions generally have the closest relationship between bone and skin.

Collagen-dense armoured dermis leaves relatively coarse (1-2 mm) rugose projections of bone beneath it, often of sufficient extent that they are discernible in photographs. You can readily see them on rhino skulls, for instance, as well as around the jaw tips of hippos and the rostral bosses of red river hogs. They have a corresponding histological signature, too: patches of obliquely-orientated metaplastically ossified dermal collagen fibres (Hieronymus 2009; Hieronymus et al. 2009). These patches of rugose bone cover large areas of the skull, and, in hippos, they even wrap into the mouth, betraying the presence of soft-tissue armour inside the jaws. This is something for palaeoartists to note as it shows dense, armoured skin can grow around complex structures, like gums and teeth, and also allows for lip-like structures that sheath teeth (so no, these tissues are not excuses for toothy prehistoric artwork). I assume the stiffness of armoured skin explains why the 'canine pocket' in the hippo upper jaw does not collapse when gaping, despite their lack of skeletal reinforcement.

Business end of a hippopotomus skull - note the rugose textures around the end of the snout, characteristic of collagen-dense armoured skin. Cropped detail from a CC BY-SA 3.0 photo by Wikimedia user ContinentalEurope. From Wikimedia.

Similarly coarse rugose projections or spicules exist underneath epidermally-derived horns and crests, but with an important distinction to those underlying armoured skin. Rather than leaving uniform patches, these structures leave ring-like rugosities that outline the circumference of the projecting structure. This is true for massive projections, like rhino horns, and also more delicate ones, like pelican crests (Hieronymus 2009; Hieronymus et al. 2009). It's thought that stresses inflicted on projecting structures explain their ring-shaped 'footprint'. Virtually any load placed on a crest or horn is transmitted to base of the opposing side, meaning the edges of these structures experience the greatest loading in life. It makes sense, therefore, that the outline of the structures have the deepest developmental scarring (Hieronymus 2009). A boss or other elevated bony region is sometimes associated with epidermal structures too, but this is not universal. Ring rugosities do not tell us much about the exact morphology of projecting structures but they do reveal something about the extent of the base and - from the size of the rugosities and spicules - we can predict the size of the compositional fibres. If we assume that bigger structures need larger fibres for reinforcement, which seems borne out in modern animals - rugosity dimensions might give us some clue of overall structure size (Hieronymus 2009). 

Where can we find these structures in the fossil record?

Spoilers: in species like this guy. Mmm... abelisaur fresh... 

Turning our attention to extinct creatures, these bony correlates are robust enough to withstand fossilisation and we can look for hints of thick, armoured skin or epidermal projections in any specimen with reasonable preservation. The skulls of fossil rhinocerotids are an obvious place to seek such structures and the results are quite fascinating. Evidence of dense, armoured skin appears in taxa from c. 40 million years ago, while their horns are a more recent development, from about 20 million years ago (Hieronymus 2009). The development of large tusks, rather than horns, seems to have spurred the development of skin armour in ancient, hornless rhinos, and we can note parallel correlations between large teeth and armoured skin in other lineages. When animals are routinely slashing, ripping and biting one another, armoured skin seems to be a common adaptive response (Hieronymus 2009).

Majungasaurus crenatissimus skull, showing extent of bone texture related to armoured skin (blue) and tough cornified skin (purple). Skull drawn from Sampson and Witmer (2007); distribution of bone textures after Hieronymus (2009).

It's not just mammals that get in on this act. The top and front of the skull of the abelisaurid theropod, Majungasaurus crenatissimus, matches osteological and histological criteria for dermal armour, and this is good reason to restore this species with thick, collagen-reinforced skin over its snout and braincase region (Hieronymus 2009). Adjacent skull areas - the sides of the jaws, the roof of the mouth, the orbital and cheek regions  - also show hints of a gnarly skin covering, these being marked with a bone texture characterised by deep pits and grooves. Among modern animals, this seems best correlated with thick, highly keratinised skin, such as cornified pads or beaks (Heironymus 2009; Hieronymus et al. 2009). When considered with the correlate for dermal armour, these textures suggest the face and oral cavity of Majungasaurus was covered in deep, reinforced skin tissue, and we have to wonder how much of the underlying skull structure was obvious in life. Abelisaurids are well known for their gnarly, pitted skull bones (e.g. Sereno et al. 2004), and it's likely that thick facial skin occurred in other members of the group (Hieronymus 2009). These are animals that science encourages artistic speculation with: what would the armoured face of an abelisaur look like? I've taken a punt at this concept with the Kryptops painting accompanying this post, but I'm sure there are other configurations that could be explored. It would be remiss not to mention that armoured skin on theropod faces aligns well with face-biting antagonistic behaviour predicted from their pathological bones (Tanke and Currie 1998; Hieronymus 2009), and that this again chimes with biting behaviour driving evolution of armoured skin.

Evidence of epidermal structures are common in a paleaoart mainstay: pterosaurs. By now, most of us will be familiar with the idea that many pterosaurs had soft-tissue headcrests thanks to well-publicised exceptionally preserved fossils (e.g. Bennett 2002; Frey et al. 2003), but can we predict them in species represented by bones alone? Thanks to bone textures, we can. Soft-tissue crests grow over low bars of bone projecting from pterosaur snouts, often with expanded anterior regions (Bennett 2002). Fine, curving striations and spicules are discernible on the top of these projections, contrasting with the smooth bone forming the base of the bony crest and the rest of the pterosaur skull. These rugosities mostly project vertically, or somewhat anteriorly at the front of the base structure. It is difficult to know if these rugose regions have a ring-like distribution given the flattened nature of most pterosaur fossils, but their presence around the top of a projecting bone bar implies a ring-distribution. Collectively, these components meet predictions for epidermal projections and their distribution points to a tall, narrow structure - a crest - rather than a horn or boss. We would likely see this as the most parsimonious take on pterosaur crest bases even without exceptional fossil preservation so, wherever you see these features on pterosaur skulls, it is reasonable to assume a large, prominent crest. I stress that you do not these features in all crested pterosaurs: some bony crests are completely smooth, and have no evidence for extensive soft-tissue elaboration. This is mainly seen in the ornithocheiroids (the group that includes taxa like Anhanguera, Pteranodon and Nyctosaurus).

Darwinopterus robustodens as a case study for pterosaur striated crests, and what they mean for soft-tissues. Yes, they were that daft - don't feel you need to be conservative when restoring them!

We are fortunate to have pterosaur specimens with preserved soft-tissues to help us gauge the size and shape of their crests. They are generally rounded, with the deepest portion posteriorly, and their size seems correlated with bony crest development (coarser rugosities and taller crests seem to indicate larger crests). Their crests were generally large, even in animals with modestly developed bony supports (Czerkas and Ji 2002), and they can grow to many times the area of the skull in species with strongly developed crest rugosities (Campos and Kellner 1997; Frey et al. 2003). Don't hold back when drawing these things, chaps: they were nuts (see diagram, above).

Applications to other species, and potential pitfalls

There are surely other animals that we could discuss with these features, but I think our point is made by now: with careful observation and comparison to modern species, we can detect the presence of body profile-altering skin structures in fossil animals, and these features should be on the radar of anyone trying to restore fossil tetrapods credibly. It should be stressed how phylogenetically widespread the examples given in this post are: as if it needs saying - pterosaurs, rhinos, abelisaurids, deer and so on are not closely related, and yet they share basic aspects of bone texture and histology related to skin structure. The take-home here is that skin is a highly plastic, adaptable tissue that we need to be especially open-minded about reconstructing. It is naive to assume fossil animals will only have skin types common to their closest extant relatives.

There are some caveats and pitfalls to be aware of about predicting tough dermis and epidermal projections. For example, there are a few cases where skin elaborations lack osteological correlates. Warthog warts, for instance, are prominent, permanent and conspicuous skin structures, but they leave no trace on the underlying bone. Likewise, the presence of armoured skin becomes difficult to predict beyond the skull because postcranial bones tend to be buried under other soft-tissues. We know from living animals that collagen-dense skin can be regionalised (mouse deer, for instance, tend to localise it on their dorsum and rumps - Dubost and Terrade 1970), so evidence of cranial armour is only a partial indicator for armouring across the body.

Cuspicephalus scarfi regrets sporting a hunk of tall, cornified cranial epidermis on a windy day.

Detection of bone rugosity type is also an issue, at least in cases where we are unable to see fossil material first-hand. Yes, the rugosities and structures discussed here can be seen in photos, but not always. Moreover, unless the photo is especially clear, it's easy to confuse them with other types of bone surface rugosity, of which there are several, all with different soft-tissue correlates (Hieronymus et al. 2009). So, before going nuts with armour, crests and horns on a fossil animal because they seem to have a rough surface somewhere on their skull, check out specimen descriptions, high-res photos, histological studies, quiz those consultants, and make sure the criteria for these elaborate skin structures are met.

That final point seems particularly relevant given the modern palaeoart fashion of speculating about fossil animal appearance. Long-time readers will know that I'm an advocate of this practise, but science of the kind discussed here puts an onus on artists to be careful when adorning extinct animals with elaborate skin structures. Yes, there are loopholes which can justify these outlandish reconstructions if we want to find them, but consider that some speculative structures included in modern palaeoartworks would be expected to leave osseous markers if they were present. Maybe this is a case where absence of evidence is actually evidence of absence and, if we cannot find these correlates, we should assume those structures were not present in our subject species.

Enjoy monthly insights into palaeoart and fossil animal biology? Consider supporting 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!

References

• Bennett, S. C. (2002). Soft tissue preservation of the cranial crest of the pterosaur Germanodactylus from Solnhofen. Journal of Vertebrate Paleontology, 22, 43-48.
• Campos, D.A. & Kellner, A.W.A. (1997). Short note on the first occurrence of Tapejaridae in the Crato Member (Aptian), Santana Formation, Araripe Basin, Northeast Brazil. Anais da Academia Brasileira Ciências, 69, 83–87.
• Czerkas, S. A., & Ji, Q. I. A. N. G. (2002). A new rhamphorhynchoid with a headcrest and complex integumentary structures. Feathered Dinosaurs and the Origin of Flight, 1, 15-41.
• Dubost, G., & Terrade, R. (1970). La transformation de la peau des Tragulidae en bouclier protecteur. Mammalia, 34, 505-513.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, 233-266.
• Hieronymus, T. L. (2009). Osteological Correlates of Cephalic Skin Structures in Amniota: Documenting the Evolution of Display and Feeding Structures with Fossil Data (Doctoral dissertation, Ohio University).
• Hieronymus, T. L., Witmer, L. M., & Ridgely, R. C. (2006). Structure of white rhinoceros (Ceratotherium simum) horn investigated by X‐ray computed tomography and histology with implications for growth and external form. Journal of Morphology, 267, 1172-1176.
• 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, 1370-1396.
• Sampson, S. D., & Witmer, L. M. (2007). Craniofacial anatomy of Majungasaurus crenatissimus (Theropoda: Abelisauridae) from the late Cretaceous of Madagascar. Journal of Vertebrate Paleontology, 27, 32-102.
• Sereno, P. C., Wilson, J. A., & Conrad, J. L. (2004). New dinosaurs link southern landmasses in the Mid–Cretaceous. Proceedings of the Royal Society of London B: Biological Sciences, 271, 1325-1330. 
• Shadwick, R. E., Russell, A. P., & Lauff, R. F. (1992). The structure and mechanical design of rhinoceros dermal armour. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 337, 419-428.
• Tanke, D. H., & Currie, P. J. (1998). Head-biting behavior in theropod dinosaurs: paleopathological evidence. Gaia, 15, 167-184.

New paper: pterosaur palaeoecology, as told by the fossil record

A female Pteranodon tries to explain the new Silverstone et al. (2017) paper on Pteranodon taxonomy to the Cretaceous shark Squalicorax. Unfortunately for her, the sharks quite liked the 'Dawndraco' hypothesis.

Last year I posted a couple of overviews of the better parts of the pterosaur palaeoecological record, discussing what we know was eaten by Rhamphorhynchus and azhdarchid pterosaurs, as well as what species ate them. These reviews were tied to a peer-reviewed paper on the same subject which, at the end of Febuary 2017, was published as part of an upcoming collection of pterosaur papers (Witton 2017). This collection, edited by David Hone, myself, and David Martill, is the proceedings of the Flugsaurier 2015 pterosaur meeting and will, when finished, contain over a dozen new insights into pterosaur research, with an emphasis on their palaeobiology. You can check out the existing content here - keep an eye on that site, as there are more papers to come.

With my paper now out (though sadly not open access, but I will eventually be able to post an unformatted version online next year) I thought it would be a good time to take a holistic look at direct fossil evidence of pterosaur lifestyles. What are some of the most interesting examples of pterosaurs interacting with other species? Which purported interactions stand up to scrutiny, and which ones are a little tenuous? And what do they tell us about the all important Big Picture of pterosaur palaeobiology?

Yes, some pterosaurs may well have been seabird mimics

A number of pterosaur specimens have been reported as being associated with the remains of their last meals. Several of these have been lost, found to be erroneously interpreted, or are simply too poorly preserved to interpret their gut content. However, examples of the Jurassic non-pterodactyloids Rhamphorhynchus and Scaphognathus, the Triassic Eudimorphodon, and the famous Cretaceous taxon Pteranodon show reliable insights into their dietary preferences (below). These are virtually all remains of aquatic animals - mostly fish - preserved in intimate association with pterosaur skeletons, either between their jaws, aligned with their throats or within the torso skeleton. One example of a coprolite is known, though it's difficult to say exactly what it contains.

Pterosaurs and their last meals (shaded grey). A, torso of Eudimorphodon; B-D, various Rhamphorhynchus with gut content and coprolite (C), E, Scaphognathus; F, Ludodactylus; and G, Pteranodon. From Witton (2017).

Many of these specimens have been known for several decades, and their evidence of aquatic feeding probably played some part in the stereotyping of pterosaurs as seabird analogues (e.g. Wellnhofer 1991). Nowadays, we need to be a little more circumspect about what they tell us. Yes, they do show that some pterosaurs ate fish and other pelagic prey and, along with results from detailed studies into functional morphology, they help portray certain pterosaur species in the 'classic' seabird niche. Rhamphorhynchus, Scaphognathus, Eudimorphodon and Pteranodon have at least some adaptations consistent with foraging for pelagic prey, such as long wings ideal for marine soaring, 'fish-grab' jaws and adaptations for launching from aquatic settings, as well as occurrences in coastal or marine settings. It would be a little odd if these aquatic-adapted species weren't catching aquatic animals from time to time.

But we can't maintain the older view that these specimens, on their own, undermine the increasingly diverse and nuanced takes on pterosaur palaeoecology hinted at by form-function studies, biomechanics, and modern understandings of pterosaur habitats. We have thousands of pterosaur specimens in museums around the world, of which gut content is known from less than a dozen examples, and in four species. That's not even enough to demonstrate the full dietary range of the species in question, let alone tell us about the ecology of all pterosaurs. Indeed, the scarcity of pterosaur gut content agrees with some new predictions of pterosaur lifestyles in that non-aquatic food sources now suggested for pterosaurs - insects, wormy things, fruits, small tetrapods - have limited preservation potential, particularly outside of Lagerstätten. When factored against common agents of taphonomy and preservation, these hypotheses predict empty bellies in many pterosaur fossils, which is what we find virtually all of the time. It is, of course, difficult to be certain of anything concerning negative evidence, but it's nevertheless useful to note this predicted match between modern ideas and fossil data.

A selection of pterosaur foraging traces - beak tip impressions and scrape marks - from Jurassic and Cretaceous sites. The black-filled elements are the feeding traces, dark grey are manus prints, and light grey are footprints. From Witton (2017).

Evidence that not all pterosaurs were obtaining their food out to sea comes in the form of feeding traces - small, paired impressions and scratch marks created by beak tips (above). These were likely formed by pterosaurs wandering over water margins in pursuit of invertebrates and other small prey, much like extant shorebirds and waders. Indeed, if you walk across a mudflat on a falling tide you can find near identical traces made by living avians mimicking this pterosaur strategy. Somewhat frustratingly, the identities of the pterosaurs that made these tracks remain mysterious. That said, in my new paper, I have - finally - formalised a case for a Late Cretaceous Mexican set of tracks and possible feeding traces (panel D, above) having an azhdarchid trace maker.

Pterosaur feeding evidence: the 'close, but no biscuit' specimens

Inferring palaeoecological details from fossils can be tricky, and it is unsurprising that some purported insights into pterosaur diets and lifestyles are contentious. One of these is the famous and perhaps darkly comic circumstances surrounding the holotype skull and mandible of Ludodactylus sibbicki, a Cretaceous, likely fish-eating Brazilian ornithocheirid found with a sharp, pointed leaf between its lower jaw rami (panel F in the image above). Much of the 2003 description of this specimen (Frey et al. 2003) discusses this association and concludes that ingestion of these plant remains led to the death of the pterosaur. According to this story, the pterosaur accidentally scooped up the leaf, having mistaking it for its usual prey, stabbed the plant material on its throat tissues, frayed the end of the leaf trying to work it loose, but starved to death before it could dislodge it.

I must admit a little scepticism about this scenario. This is not because animals getting things stuck in their mouths is implausible, but because the story presented by Frey and colleagues is pretty presumptive. It infers a lot about pterosaur behaviour, foraging strategies, throat tissue strength and so on that we can't confirm at present. Moreover, the hyoid apparatus - the skeletal support for much of the throat and tongue tissue - is preserved lying on top of the leaf, despite the suggestion that the plant matter was deeply imbedded in the throat tissues. How did that work itself loose with the leaf fatally stabbed between the jaws? The answer to that question - as with a lot of questions about this association - would easily fall into speculation and special pleading about all manner of unknown quantities, and thus has little value to understanding fossil animal palaeobiology. Boring and po-faced as it is, I don't think the unusual Ludodactylus holotype provides enough information to tell us much about pterosaur behaviour, or how this unlikely fossil association came to be.

A similar observation might be made about insect specimens - a dragonfly and lacewing - from the Jurassic Solnhofen Limestone that have torn wings, allegedly from a pterosaur attack (Tischlinger 2000). The logic goes that these otherwise perfectly preserved insects cannot have been attacked by aquatic predators, or else they would have been eaten after their wings were damaged. Failed attack from an airborne predator that would not pursue the injured insects into water is suggested as more likely. Solnhofen deposits do hold pterosaurs that were almost certainly aerial insect hawkers - such as Anurognathus (below, see Bennett 2007 and Witton 2013) - and these might be ideal perpetrators in this scenario.

Anurognathus ammoni was an insect-hawking pterosaur that lived over the Solnhofen lagoon. Has it left feeding traces on fossil insect wings after a failed attack?

 As with Ludodactylus, this set of circumstances is quite elaborate to base purely on damaged insect wings. The extent of their wing damage is considerably greater than we might expect under general 'wear and tear' and foul play was probably involved, but whether it was a pterosaur, a conspecific, or even those disregarded aquatic predators is difficult to say. I appreciate the logic that aquatic predators would eat disabled insects after a failed strike, but animals are not predictable, logic-driven machines: they make mistakes, strike at things they have no intention of eating, get bored, distracted and so on. In all, other than the fact that these insects were almost certainly attacked by something, it might be difficult to say anything more substantial about their final moments.

Pterosaurs vs. dinosaurs, crocodyliforms and... the revenge of the fish

The fossil record gives us an insight on the question "did pterosaurs taste good?", and that answer seems to be "yes". Bite marks, embedded teeth and vomited pterosaur remains indicate that dinosaurs, crocodyliforms and fish all ate pterosaur flesh, at least on occasion (below). Among the more impressive examples of these interactions is a spinosaurid tooth, likely from the Brazilian spinosaurine Irritator challengeri, embedded in the cervical vertebra of an ornithocheirid (Buffetaut et al. 2004). Alas, no other evidence of their interaction was evident on the specimen (a series of pterosaur vertebrae) and it's not possible to ascertain much about circumstances that brought these species together.

Evidence of many, many things that ate pterosaurs. A, ornithocheirid cervical vertebrae with embedded spinosaurid tooth; B, azhdarchid tibia with tooth gouges and embedded dromaeosaur tooth; C, ornithocheiroid wing metacarpal with unidentified puncture marks; D, Quetzalcoatlus sp. skull with puncture marks; E, Eurazhdarcho langendorfensis cervical vertebrae with crocodyliform puncture marks; F, Pteranodon sp. cervical vertebra with intimately associated Cretoxyrhina mantelli tooth; G, Velociraptor mongoliensis torso with possible azhdarchid pterosaur gut content; H, probable fish gut regurgitate including Rhamphorhynchus bones; I, associated Rhamphorhynchus muensteri and Aspidorhynchus acutirostris skeletons. Images drawn and borrowed from many sources - see Witton 2017 for details.

The fossil record's most common purveyors of pterosaur murder, however, are not dinosaurs or crocodyliforms, but fish. Apparently out for revenge after learning of all that fishy pterosaur gut content, we've got evidence of fish eating and spitting out pterosaurs, of pterosaurs getting entangled with piscine predators, and even fish bite marks on pterosaur bones. A lot of these pertain to specimens of Rhamphorhynchus and you can read more about them in this post - some of the specimens are exceptional and there's lots to say about them. One of the more famous examples of piscine-pterosaur consumption -  an Italian, Triassic pellet composed of alleged pterosaur bones (Dalla Vecchia et al. 1989) - has recently been reappraised. It's now more reliably interpreted as vomit ball made of bones from the tanystropheid Langobardisaurus (Holgado et al. 2015).

Lesser known, but pretty darned awesome examples of fishes eating pterosaurs are Pteranodon specimens that found themselves at the wrong end of Cretaceous sharks. Several Pteranodon bones reveal bite marks and even embedded teeth from two genera of sharks, the 2-3 m long 'crow shark' Squalicorax and the larger, 6 m long 'ginsu shark', Cretoxyrhina. The former seems to have eaten Pteranodon flesh on several occasions, while evidence of the latter is only currently known from a tooth closely associated with a cervical vertebra (panel F, above). Further work on the latter specimen is currently underway.

 Feeding traces from these sharks are common in Western Interior Seaway fossils and those of Squalicorax are particularly abundant and taxonomically indiscriminate. Given that even giant marine reptiles are among the species consumed by this mid-sized shark, it's often assumed that this animal was a scavenger, biting into whatever free meat floated about America's continental sea. However, it is less certain that Pteranodon was scavenged by Squalicorax, as even a 2 m long specimen would vastly outweigh the largest Pteranodon. It is not inconceivable that an unwary Pteranodon could be grabbed and killed by a stealthy Squalicorax, though I stress this scenario is no better supported than the shark simply chancing across a Pteranodon carcass. Whatever the scenario, it's somewhat grounding to think of a weird extinct creature like a pterosaur being devoured by a fairly conventional-looking shark. It's a reminder, perhaps, that Mesozoic life was not a pantomime of exotic, giant reptiles and weirdo evolutionary experiments, and that much of our modern ecosystem was in place many millions of years ago.

The big picture

Looking at the pterosaur palaeoecological record holistically, what patterns emerge? If we look at where the record focuses phylogenetically (below), it's obvious that our records are significantly biased towards certain taxa - Pteranodon, Rhamphorhynchus, and azhdarchids. Even their close relatives, with similar anatomy and adaptations, preservational conditions and so on, don't get much of a look in. There's a few data points scattered here and there, but tumbleweeds run though the palaeoecological data stores for the majority of the group.

Attempting to make sense of the pterosaur palaeoecological record in a holistic way mainly shows how paltry this record remains. It's improved a lot in recent years, but we await evidence of diet and consumer-consumed relationships in virtually all major pterosaur clades. The images at the bottom of this figure are takes on known examples of pterosaur ecology: Rhamphorhynchus ingesting fish, and azhdarchids being devoured by dromaeosaurs. From Witton (2017).

We wouldn't be scientists if we didn't ask ourselves why this is. I don't think it's simply a sampling issue. The pterosaur record is not great, but we are talking about several thousand specimens now - enough that we might start looking at what we don't have as well as what we do. So why does Rhamphorhynchus show 10 palaeoecologically-relevant fossils, but other Solnhofen species only preserve one confirmed piece of gut content? Why do azhdarchids, which are never found in sites of exception preservation and are generally only known from bits and pieces, have a better record than those lineages which are abundant, represented by dozens of complete skeletons, and often found in sites of exceptional preservation? Interestingly, there's no obvious correlation between factors like abundance, preservation quality and palaeoecological data. Several lineages - the ctenochasmatoids (wading pterodactyloids), the rhamphorhynchids (excluding Rhamphorhynchus) and ornithocheiroids (excluding Pteranodon) - have everything going for them in terms of abundant fossils, occurrences in sites of exceptional preservation, and yet they turn up very little in the way of gut content, or evidence of being consumed by other Mesozoic animals.

My take on all this is that there must other factors at play here. We don't get evidence of pterosaur palaeoecology just by throwing more fossils, or better quality fossils, into the mix. I'm sure these factors have some role, but perhaps only in concert with special traits of certain pterosaur groups - maybe behaviours and anatomies - that allow them to have good records. We might have a good record of azhdarchids being consumed by dinosaurs and crocs, for instance, because their bones are often quite big and allow predators to bite them without destroying them. Perhaps we have good palaeoecological insights for Rhamphorhynchus and Pternanodon because of their habits and behaviour - both have strong aquatic adaptations (see this blog post for ideas on that), and there is a bias towards preservation of aquatic animals in the fossil record. Perhaps this aids preservation of not only palaeoecological data, but also explains why these taxa are our most abundant pterosaurs (>100 Rhamphorhynchus fossils are known, >1000 Pteranodon).

The pterosaur palaeoecological record, then, is perhaps in a transformative state. Though vastly improved over its condition a few decades ago, it requires further augmentation to provide us with significant insights into pterosaur lifestyles, and to explain its biased nature. However, we should not be too pessimistic about the insight it offers into pterosaur palaeobiology: it still provides useful datapoints that can shape our interpretation of flying reptile ecology for several species. Cliched as it is, the take-home message of this project is that any palaeoecologically-relevant pterosaur fossils are worth putting on record. We still have a lot to learn about how these animals lived and behaved, and direct insights are the most reliable ways to do that.

If you enjoyed this post, you might enjoy my exclusive Patreon content

This blog is sponsored via Patreon, the site where you can donate a few dollars to your favourite online content creators to help them make a living. If you like what I do here, please consider becoming a patron for as little as $1 a month. In return, you'll get regular updates on projects I'm working on, including research, papers, books and paintings. Currently, this includes two palaeoart-heavy books, and my Patreon pages are filling with advance previews of their content, discussions of the animals concerned and that sort of thing. Plus, you get free stuff - prints, high quality images for printing, books, competitions - for sponsoring my work. As always, thanks to everyone who already sponsors my work!

References

• Bennett, S. C. (2007). A second specimen of the pterosaur Anurognathus ammoni. Paläontologische Zeitschrift, 81(4), 376-398.
• Buffetaut, E., Martill, D., & Escuillié, F. (2004). Pterosaurs as part of a spinosaur diet. Nature, 430(6995), 33-33.
• Dalla Vecchia, F. M., Muscio, G., & Wild, R. (1989). Pterosaur remains in a gastric pellet from the Upper Triassic (Norian) of Rio Seazza valley (Udine, Italy). Gortania, 10(1988), 121-132.
• Frey, E., Martill, D. M., & Buchy, M. C. (2003). A new crested ornithocheirid from the Lower Cretaceous of northeastern Brazil and the unusual death of an unusual pterosaur. Geological Society, London, Special Publications, 217(1), 55-63.
• Holgado, B., Dalla Vecchia, F. M., Fortuny, J., Bernardini, F., & Tuniz, C. (2015). A reappraisal of the purported gastric pellet with pterosaurian bones from the Upper Triassic of Italy. PloS one, 10(11), e0141275.
• Martin-Silverstone, E., Glasier, J. R. N., Acorn, J. H., Mohr, S. & Currie, P. J. (2017). Reassesment of Dawndraco kanzai Kellner, 2010 and reassignment of the type specimen to Pteranodon sternbergi Harksen, 1966. Vertebrate Anatomy Morphology Palaeontology, 3, 47–59.
• Tischlinger, H. (2001). Bemerkungen zur Insekten-Taphonomie der Solnhofen Plattenkalke. Archaeopteryx, 19, 29-44.
• Wellnhofer, P. (1991). The illustrated encyclopedia of pterosaurs. Salamander Books.
• Witton, M. P. (2013). Pterosaurs: natural history, evolution, anatomy. Princeton University Press.
• Witton, M. P. (2017). Pterosaurs in Mesozoic food webs: a review of fossil evidence. Geological Society, London, Special Publications, 455, SP455-3.