Wednesday, August 29, 2007

Pterosaurs: What the hell ARE they?

Pterosaurs were, as far as anyone can tell, the first vertebrates to achieve powered flight. These little reptiles developed a unique method of flight via elongation of the fourth manual digit. This is unlike the case in bats, wherein digits II, III, IV, and V are elongated. The pterosaur wing was, by comparison, long and narrow. It must have worked out alright, though, because pterosaurs ruled the skies for the entirety of the Mesozoic. In fact, although the first unquestionably pterosaurian fossils are of Late Jurassic age, those unquestionable specimens are unquestionably advanced. There is no Archaeopteryx among pterosaurs. This means two things: First, pterosaurs were probably around for quite some time before they show up in the fossil record. Second, their ancestry is a matter of some debate. Traditionally, the Pterosauria has been seen as either an outgroup to either the Ornithodira or, in some cases, the Dinosaurmorpha. Other workers see them branching from the more basal archosaurs. Still others see the pterosaurs as closer to protorosaurs or even, dare I say it, the prolacertiforms.
Part of the problem, of course, is that the first pterosaur fossils known (Eudimorphodon, Dimorphodon) are already incredibly derived. Ironically, this quandry is the same as for bats, who just "appear" in their modern form. Molecular evidence points to a Canivora ancestry for bats, but there is no such evidence for pterosaurs. I suppose a good analogy would be whales: If you had to figure out where whales fit into the Mammalia, but the only whale you could look at was a modern baleen whale, you'd have some trouble. This is essentially the situation with pterosaurs, who, even at their earliest known appearance, are so derived and suited for a flying lifestyle that their ancestry is clouded by those very derivations.
One thing we can be sure of is that pterosaurs are reptiles. Better yet, they are diapsid reptiles. Of course, seeing as diapsid reptiles ruled the world for the better part of the Mesozoic, our search for a pterosaur ancestor still crosses a broad spectrum. Like I said before, the traditional model is that pterosaurs are, in some vague way, related to dinosaurs. But I'm hoping you'll see, after reading this post, that this hypothesis is in fact not very well supported.
Behold, Marasuchus, the iconic dinosaurmorph. Fully digigrade and bipedal with a tri-radiate pelvis, shortened forearms, a long, flexible tail, armor scutes, and a unique ornithodiran ankle structure allowing that specialized horizontal posture. Surely, pterosaurs even as "primitive" as Dimorphodon could not have come from as specialized a creature as this. Marasuchus stood on the very threshold of being called a dinosaur. Indeed, Greg Paul has said that "lagosuchids" were early dinosaurs. We know that pterosaurs were not dinosaurs, so perhaps Marasuchus is a bad example. We need something a little more basal, but still among the ornithodirs.

This big-headed cutie is Scleromochlus, a primitive ornithodir discovered in Scotland, and it has been suggested by a few scientists as being a pterosaur ancestor. This seems to be due to its basal nature, as it's usually considered an outgroup of the Ornithodira. So, if we want pterosaurs to be ornithodirs, then maybe Scleromochlus is a kind of ornithodiran "Euparkeria," providing a nice basal form from which several more specialized animals can branch from. However, both Marasuchus and Scleromochlus do nothing more than emphasize the problem with placing the Pterosauria within the Ornithodira. Specifically, the structure of the pelvis is completely different in pterosaurs compared to ornithodirans (I'll get to specific aspects of pterosaur anatomy in a bit). Whereas ornithodiran pelves are built with strut-like elements that point forward and back, and connect in the middle, pterosaur pelvises are shield-like when viewed laterally with no elements connecting in the middle. Moreover, pterosaurs were plantigrade like humans and bears--we walk on the soles of our feet. One of the highlights of being an ornithodir is that you walk on your tip-toes, thus increasing the length of your stride. Finally, ornithodir arms are short while the legs are long. In pterosaurs, the opposite is true. And perhaps most frustratingly, pterosaurs are unquestionably sprawling in their gait (rhamphyrhychids moreso than pterodacyloids), while Marasuchus, Scleromochlus, and all of their cousins were fully bipedal with parasaggital stances. In his 1998 redescription of Scleromochlus, Michael Benton had this to say:

"Padian (1997) named the postulated clade containing Scleromochlus and Pterosauria the Pterosauromorpha. The characters proposed to link Scleromochlus with either pterosaurs or dinosaurs are not wholly convincing. Sereno (1991, pp. 36-37) listed the following as potential synapomophies of Sclermochlus and Pterosauria: (33) Skull more than 50 percent presacral column length; (34) Length of scapula is less than 75 percent of humeral length; (35) Fourth trochanter absent; (36) Metatarsal 1 length is at least 85 percent of metatarsal 3 length.
"The first three of these could relate simply to the small size of the animals. Character (33) is true also for phytosaurs and many crocodylomorphs (although because of snout elongation). Character (34) is probably spurious because it compares what is genuinely a relatively short scapula in Scleromochlus with the relatively elongated forelimbs of pterosaurs (Bennet 1996, p. 279). Character (36) is present also in some basal crocodylomorphs, so it can evidently arise convergently."

So what, pray tell, is keeping pterosaurs rooted so firmly in the Ornithodira in the first place? Quite simply, the structure of the ankles share features with ornithodirans and dinosaurs which no other diapsid group (heck, reptile group) shares. This could, of course, be due to convergence. And I'm tempted to think so. After all, when the number of features that would keep an organism out of a particular group so heavily outnumbers the quota for keeping that animal in, something's up.

Archosauriforms, perhaps?

If Scleromochlus does not provide a basal enough form, than why not just start climbing down the diapsid family tree until we get somebody who fits the bill? Pterosaurs are almost always thought of as archosaurs based mainly on a single, perhaps unambiguous character: the antorbital fenestrae, that big skull window in front of the eyes shared by all archosauriforms (and secondarily lost in many crurotarsians). However, practicality puts a dent in even this idea. If it were true that pterosaurs, those most lightweight and long-armed of organisms, would sit among fat, heavy, short-limbed crocodile-like critters. Also of note is that pterosaurs lack a mandibular fenestrae, a hole in the jaw that is also diagnostic of the Archosauriformes. If pterosaurs are archosauriforms, then that means that they, for whatever reason, secondarily lost the mandibular fenestrae. This makes no sense at all, given that pterosaurs excavated their skeletons like crazy in an effort to become more lightweight.

Sharovipteryx cousins?

There are few gliding vertebrates stranger than Sharovipteryx. Aside from its strange anatomy, Sharovipteryx is unique because it was found in the same area (Kyrgyzastan) in the same year (1965) by the same person (Alexander Sharov) as the infamous Longisquama, the animal that because Alan Feduccia's posterchild for a non-dinosaurian avian ancestor. At any rate, Sharovipteryx has often been suggested as a pterosaur ancestor, and it does have one big thing going for it: gliding membranes. This little prolacertiform, who is also related to Tanystropheus, was clearly arboreal. The physical remnants of Sharovipteryx, however, provide little more than trace evidence for the creature. It's clear that the legs were extremely long, as was the tail, and the spine and parts of the skull are preserved. Aside from some questionable humeri shapes, the arms are unknown. Sharovipteryx's fossil, however, does preserve impressions of a skin membrane stretching from the toes of the feet to the tail. Sadly, the state of preservation above the pelvis is terrible, so whether Sharovipteryx actually had gliding membranes above the waist is unknown.

However, several studies involving paper and computer models have suggested that, in order to control aerial stability and extend the gliding period, Sharovipteryx must have had a sort of canard membrane originating on the forelimbs. Whether this membrane stretched from the front of the arm and attached to the neck, or the back of the arm to the body, or both, is unknown. A 2006 study by Dyke, et al. suggested that Sharovipteryx had a forelimb membrane like the one above, making it the first "Delta-wing glider," like a Vulcan bomber.

Tempting though Sharovipteryx is as a pterosaur ancestor, there are several problems even with it. Despite its aerodynamic capabilities, Sharovipteryx has extremely long legs and ridiculously small forelimbs (from what's known). Its arms are also simply unknown, and although Sharov suggested in 1971 that the animal's fourth finger may have been elongated, but there is no evidence of the animal's hands. So, although the gliding membranes provide some indirect evidence for a pterosaurian relationship, Sharovipteryx is simply too poorly preserved to answer our question.

So what ARE they?

What makes pterosaurs so hard to place, like I said before, is their level of derivation. And whether certain features are actually primitive or derived is hard to say. For example, pterosaurs were plantigrade--they walked on the soles of their feet like humans, bears, lizards, and crurotarsians. Pterosaur trackways and a well-preserved 3D fossil of a Dimorphodon foot have secured pterosaurs as plantigrade. However, is the plantigrade posture retained from an ancestor or was that ancestor digigrade, only to switch over to plantigrade due to the physical constraints of the connecting wing membrane? In discussing Dimorphodon weintraubi's plantigrade posture to the competing bipedal pterosaur model, Clark et al had this to say:

"Other features of digits I-IV of the D. weintraubi foot indicate a capacity for grasping that is consistent with an ability to climb but is unexpected in an obligate cursor. The claws are moderately curved (nearly as strongly as the claws of the manus); all phalanges except the most proximal have well developed flexor tubercles for the insertion of digital flexors (Fig. 2); and all of the IP joints allow for extensive flexion of the digits (as exhibited by digit IV; Fig. 2). Furthermore, the phalangeal proportions of the digits of Dimorphodon and other basal pterosaurs are similar to those of birds with grasping feet (that is, perching, climbing, and raptorial species) and unlike those of primarily ground-living birds, bipedal dinosaurs and the primitive dinosauromorphs Lagerpeton and Marasuchus."

Pterosaurs, or at least rhamphorhychoids, were scansorial and probably rarely ventured to the ground. In fact, according to David Unwin, no confirmed rhamphorhychoid trackways have ever been found. Perhaps, given the extremely limiting connections of the rhamphorhychoid wing membranes, being grounded meant getting eaten.

Pterosaurs pelves are also a bit of a mystery. Again, the old bipedal model depends on the structure and orientation of the femur socket, and a bipedal posture was proposed by Padian and Bennett at different times (Padian preferring a horizontal posture while Bennett suggested a rather comical human-like construction). For a dinosaurian bipedal posture to work, however, the legs would have to be able to be held underneath the body. This picture, as well as the next one, are from Wellnhofer's excellent study of an Anhanguera pelvis from Brazil. While this particular pterosaur is fairly large pterodactyloid, it is comparable to rhamphorhychoid pelves and so serves as a great model for pterosaurs in general. The first thing I notice upon looking at this is that the acetabula is completely closed. In ornithodirans, the acetabula is at least partially open, and in early dinosaurs it is mostly open if not completely so. The construction of the iliac blades are also wildly divergent from a supposed ornithodiran (or even archosaurian) ancestor in that they are thin, not flat. The pubis is wide, as is the ischium. In rhamphyrhynchoids, the pubis and ischium actually fuse to form a sort of D-shaped bone. Pterosaurs also developed "prepubes." A prepube looks a bit like a traditional theropod pubis, but chopped off right above the boot. The prepubes attach to the front of the pubis and connect to the belly ribs. They essentially anchor the pelvis to the sternum, making the body even more immobile than birds. Prepubes are "new" bones which, in addition to the pteroid wrist bones, are entirely unique to pterosaurs.

As you can see from a front view of Anhanguera's pelvis, the pubis and ischium do not connect in the middle. Rather, each side forms a sort of "pelvic shield" which widens and strengthens the body. In addition, the position of the acetabula prohibit a parasaggital posture. Notice that the femur sockets actually point out and up. Wellnhofer reflects on the diagram above:

"Putting the femur in its proper place and allowing some space for cartilage in the acetabulum, the most comfortable setting for the head would result in a horizontal position of the shaft (Figure 6, left). Then, the axis of the collum femoris coincides with the axis of the plane of the acetabular rim directed 35 degrees upward. A higher lifting of the femur was certainly possible. In the opposite direction, the femur could not be adducted too much (Figure 6, right). A maximal angle of the shaft 150 degrees downward would have been achieved at best, because the ventral rim of the acetabulum formed by the pubis was a natural stop. On the other hand, the head could no longer find support in the hip socket without being in danger of luxation. Therefore, the orientation of the femora during terrestrial locomotion was probably less extreme than shown here."

In other words, pterosaurs who tried a parasagittal posture would have dislocated their hips. More to the point this time, Wellnhofer continues:

"In any case, a parasagittal swing of the hind legs was absolutely impossible. The femora of pterosaurs were splayed out, their stance and gait was semi-erect. Consequently they could not have been bipedal animals. Furthermore, the morphology of the foot skeleton suggests that they were not digigrade."

That last comment makes me chuckle, because Wellnhofer realized that pterosaurs were plantigrade in 1988, ten years before the foot of Dimorphodon weintraubi was described.

Again, however, we can't be sure whether the strangely-directed acetabulum is a primitive or derived characteristic among pterosaurs. Perhaps the hip socket was directed upwards as a necessity for a smooth wing surface. When a pterosaur flew, it moved its legs into a full lateral position, which straightened the patagium but also moved the foot into a position perpendicular to the wing surface. According to Unwin, the toes of pterosaurs were webbed, not to aid in aquatic habits, but to control yaw during flight. Had the acetabula of pterosaurs been directed fully to the side, it is doubtful that the legs could have been of much use during flight. Thus, it is entirely possible that the seemingly primitive sprawling condition of the hind legs is a derivational necessity of the flight aparatus.

I cannot pretend to advance my own theory of pterosaur origins here, but I often find myself questioning the traditional viewpoints regarding their branching-off points. It seems obvious to me that pterosaur cannot be ornithodirs, and perhaps just as unlikely that they are archosaurs proper. A more basal archosauriform is more likely, in my opinion. Certainly the possibility of a relationship with Sharovipteryx and the greater Prolacertiformes cannot be ruled out. Like I find myself saying at the end of almost every paleo-related post, we need more specimens! And better ones! It would be fantastic to find the arms of Sharovipteryx, for example, or better yet, a pterosaur who could only glide instead of fly. That would make my day. Pterosaurs are my favorite flying vertebrates by far, and I hope you've enjoyed reading about their murky origins as much as I've enjoyed writing about them.

P.S. Much as I'd love to keep writing about how pterosaurs are incredibly unique--for example, I didn't even go into their bizarre forelimb anatomy--I've got to force myself to stop at some point, because otherwise, I'll write a book. And nobody wants that.


Unwin, D. M. (2006). The Pterosaurs from Deep Time. Pi Press, New York, NY.

Wellnhofer, P. (1996). The Illustrated Encyclopedia of Prehistoric Flying Reptiles. Barnes & Noble Books, London, UK

Wellnhofer, P. (1988). Terrestrial locomotion in pterosaurs. Historical Biology(1): 3-16.

Clark, J. M. et al (1998). Foot posture in a primitive pterosaur. Nature(391): 886-889.

Benton, M. J. (1999). Scleromochlus taylori and the origin of dinosaurs and pterosaurs. Phil. Trans. R. Soc. Lond. B(354): 1423-1446.

Dyke, G. J. et al (2006). Flight of Sharovipteryx mirabilis: the world's first delta-winged glider. Journal of Evolutionary Biology (published online).


Laelaps said...

Excellent work, Zach! A very informative post, indeed.

Will Baird said...

A bit late, but an excellent post, Zach.

Mike Walley said...

A well put together and most informative post. We also find a great deal of confusion surrounding the origin of Pterosauria and I like the analogy of comparing the lack of evidence of Pterosaurs with bats and baleen whales.

Anonymous said...


You omitted work by Peters 2000 that demonstrated Cosesaurus, Sharovipteryx and Longisquama were successively closer taxa to the Pterosauria. The first two have pterosaur feet (long fifth toe), a pterosaur pelvis and an attenuated tail. The last has a longer ulna than humerus, strap-like scapula and coracoid, a sternal complex and multi-cusped teeth. Cosesaurus and Longisquama have an antorbital fenestra. Yhe wings came last.

Anonymous said...

Great article.
Perhaps the ancestry of these amazing flying creatures is from the stem mammals.
The stem mammals are also called mammal-like reptiles as I expect you know.

Zachary Miller said...

Stem-mammals are not reptiles, though. They're synapsids. Mammals didn't evolve from reptiles.

Recall that somebody...the name escapes me now...back in the early days of pterosaur paleontology, suggested that pterosaurs represent marsupial bats. Hogwash, of course.

Pterosaurs most certainly diapsid reptiles, though: their skull anatomy alone confirms that. The question now is where they sit in relation to other reptiles. The most recent analyses have suggested a basal position among ornithodirans.

Anonymous said...

I have been investigating this for a long time. Do not be so quick to rule out a stem mammal ancestry. It resolves a lot of questions.
I am interested in discussing this in a friendly, scientific way if you are interested.

Zachary Miller said...

Sure, but understand that I'm not pterosaur expert. I merely pull from my available sources and the advice of my colleagues.

Anonymous said...

I would begin by saying that stem mammals are reptile-like mammals. So they still have a number of reptile-type characteristics from their reptile-like pre-amniote ancestors.
So the reptile-type characteristics that we see, especially in the basal pterosaurs, could have come from their stem mammal ancestors. Does that make sense so far?

Zachary Miller said...

Hmmm...let's move this to an email thread. Try sillysaur at gmail dot com. We'll start with this question: what features are shared between basal pterosaurs and basal synapsids?