How did pterosaurs take off?
Unlike bats and birds, pterosaurs, the first flying vertebrates, are all extinct. We can only imagine what it was like to see them fly, land and take off. It used to be thought that pterosaurs hung by their toes from cliff faces, like bats, before becoming airborne. Illustrations have shown pterosaurs clinging inverted to tree branches ready to drop off. The idea of bipedal pterosaur running downhill into a headwind was floated by Chatterjee and Templin (2004), but had few takers. Bennett (1997) imagined the first pterosaurs leaped with their hind limbs into flight. Padian (1981) imagined they ran first.
Recent studies by PhD candidate, Mike Habib (2008 pdf), on the dynamics of the pterosaur launch proposed a novel launch method (Figures 1 and 2.) based on the technique of Desmodus rotundus, the vampire bat (Schutt et al. 1999). The vampire launches itself quadrupedally, principally with its muscular forelimbs, opening its wings at the peak of its launch. Habib documented the great disparity in the comparative sizes and strengths of pterosaur forelimbs vs. the hindlimb and how these compared to bird limbs. Vampire bats are similar. In birds, Habib reported, the hind limbs provide the primary source of propulsion for launch, either by running or leaping, immediately before the wings take over. Habib reported that pterosaurs have stronger forelimbs than hindlimbs, just the opposite of birds. Thus, Habib reasoned, since all pterosaurs were quadrupeds (this is questionable, see below), they must have employed their strongest limbs, their wings (this is also questionable, see below), to take off. Habib reported, “…expected flapping forces alone seem insufficient, at present, to explain the structural strength ratios of pterosaurs; it is simply implausible that pterosaurian lift coefficients would be several times higher than in birds…”
Dr. David Hone agreed with this hypothesis, “If pterosaurs actually launched like birds, with the hindlimbs producing most of the force, then the ratio of their bone strength values should be about the same as birds, and big pterosaurs should have slender wing bone elements (especially the proximal elements, like the humerus) and big, robust hindlimb bones. Well, it turns out that the opposite is actually the case: pterosaurs have whopping huge forelimb bones and very slender hindlimb bones, and this gets more exaggerated in big pterosaurs.” Dr. Mark Witton is in agreement here. Julia Molnar created a video to demonstrate the Habib hypothesis here. Other images can be seen here and here. The forelimb launch seems to have become the widely accepted view.
Of course, evidence for either the Habib (2008) hypothesis or the heretical bird-style take-off would have to come in the form of take-off tracks. Unfortunately we don’t know of any such tracks. We only know of one landing trace (Mazin et al. 2009, Figure 3), which came in feet first. So let’s examine each of the precepts of Habib’s hypothesis and see if any problems arise.
Problem 1. Did all pterosaurs have stronger forelimbs than hindlimbs? While most pterosaurs had a larger diameter humerus than femur, not all did. Basal pterosaurs up to, but not including, Carniadactylus and Eudimorphodon had a humerus no thicker than the femur. The same held true for a few Rhamphorhynchus specimens, Sordes, some Pterodactylus specimens and at least one germanodactylid. The flightless pterosaur had a tiny humerus. Some pterosaurs, such as Zhejiangopterus and Fenghuangopterus had a much shorter humerus than femur. Thus the femur in these taxa would have had much more leverage, travel and associated muscles to launch the pterosaur further and higher. Why did some pterosaurs have a large diameter or longer humerus? I don’t know. I can’t see a clear phylogenetic pattern yet.
Problem 2. Were the proximal forelimb elements long enough and strong enough to produce the required height and airspeed during a forelimb launch to deploy the very long folded wings and initiate the first flap?
Habib (2008) reports the forelimb bones were more than strong enough. Were they long enough? The answer seems doubtful if one compares the skeleton of our greatest living leaper, the kangaroo with that of a pterosaur (Figure 4). Despite having five muscle groups from pelvis to toe contracting in a coordinated series, kangaroo initial leaps raise the toes only to the heights of the ankles. By contrast, pterosaurs had only their elbows and wrists to extend and they could extend their elbows a relatively shorter distance, not counting the effect of the propatagium, which in birds and bats prevents exactly this sort of overextension of the elbow. Even the vampire bat leaves 15-20 degrees of flex at the elbow during takeoff.
Problem 3. The evolutionary pathway to forelimb launching.
Pterosaur predecessors were bipeds that ran and/or launched themselves with their hind limbs. Basalmost pterosaurs were also bipeds, unable to touch the substrate with their hands. Nevertheless, the vast majority of pterosaurs had forelimbs long enough to touch the substrate without bending over. Thus quadrupedal locomotion was secondarily derived. That’s why pterosaur finger tracks point laterally and posteriorly instead of anteriorly, as in other tetrapods. At this point the forelimb launch pattern could have evolved. Meanwhile, the forelimbs/wings of pterosaurs were used for flapping, flying, clinging to tree trunks and, in beachcombing genera forelimbs supported the anterior torso while walking quadrupedally. However, in beachcombers, the forelimbs produced no forward propulsion vectors. The planted hands were never behind the elbows and shoulders. Pterosaur forelimbs acted more like crutches than traditional walking forelimbs (see walking pterosaur movie here).
So were does the impetus for a forelimb launch come from? There doesn’t appear to be an evolutionary sequence demonstrating the gradual acquisition of a rapid extension of the forearm at the magnitude needed to launch a pterosaur high enough to extend the wings and initiate flapping. Nor does there appear to be any change in pterosaur shoulder and elbow morphology that would signal such a change in behaviour. Moreover, flexion of the elbow would be constrained by the long pteroid in a few taxa and extension would be constrained by the propatagium in all taxa. The question of why most pterosaurs have a thicker, stronger humerus than femur remains unanswered. No skeletal correlates have been documented yet to gauge a large humerus or a large antebrachium with large claws, a large head or any other skeletal character. Such characters change one way or the other within genera such as Pteranodon and Rhamphorhynchus.
Problem 4. The margin of error.
Avoiding the narrow margin of error and calamity a pterosaur experiences every time it would have to snap open its (sometimes huge) wings during a forelimb launch (thousands of times during a lifetime) without ever striking the ground seems to tempt the odds. Moreover, the power and coordination needed to complete the vault and extend the wings in this fly-or-crash scenario gives little room for a learning curve in younger weaker pterosaurs or in the evolution of such a maneuver from predecessors that were already adept at performing the standard hind limb leap.
Here are three animations of Pteranodon (Figures 5-7) in the process of taking off. Two portray the widely accepted wing launch model (one successful and one not successful) and one portrays the heretical bird-style launch from a standing pose. Click each one to see the animation.
Problem 5. Opening time for the big wing finger.The vampire bat is able to snap open its wings in an instant at the acme of its leap because the finger bones are mere splints with little mass and therefore little momentum and drag. Moreover they can curl like human fingers and extend at least as quickly. Not so pterosaur wing fingers. They are long, stiff and massive by comparison. In many pterosaurs, the wing phalanges are the largest bones in the specimen, and some can exceed the skull in length. That means, relative to the vampire bat, the largest pterosaur wings would have to take longer to accelerate to opening speed, and then promptly decelerate to a stop at the acme of the leap. The individual phalanges didn’t bend much at all, so the entire wing finger acted like a very long rod or pipe. Imagine swinging a 1-2 meter length of PVC pipe through 180 degrees and stopping it precisely. That’s what a big pterosaur wing would have to do if it snapped open in the moments after lift off in the Habib/Molnar model. By contrast, in the bird-style model, a less snappy pterosaur can take as much time as a bird does to open its wings.
Problem 6. Incorrect reconstruction.
The Habib/Molnar reconstruction of the Anhanguera manus (Figure 9) had a few inaccuracies. The three free fingers were way to short. All pterosaurs have subequal metacarpals 3 and 4. Thus finger three (and usuallly two and one) always extended beyond the big wing metacarpolphalangeal joint. Ichnites confirm this. Only digits 1-3 ever appear in fossil handprints. The folded knuckle of finger 4 never made an impression. If knuckle 4 never touched the substrate then it could not be used to pinch the extensor tendon during the vault/preload phase of the quadrupedal launch, which was supposed to store energy prior to a sudden release of impulse power. Also, the metacarpals of all pterosaurs actually lined up anteriorly, as in all tetrapods, including lizards and fenestrasaurs. The Habib/Molnar reconstruction incorrectly placed metacarpals 1-3 on the dorsal (in flight) surface of metacarpal 4.
The Habib/Molnar model illustrated the extensor tendon running over the extensor tendon process (which would have enabled it to be pinched during launch. However, if pterosaurs followed lizards and other tetrapods (as they probably did), the extensor tendon would have split at the knuckle, inserting on both sides of the first phalanx a short distance away from the knuckle. Such a split would have prevented the sort of tendon pinching called for in the Habib/Molnar model. Flexors were similar and a more distal insertion point permitted complete wing folding (shown in gray below, and more on that subject later) than would have been impossible with the Habib/Molnar model with insertion at the flexor process tip. That sort of engineering would only have permitted folding the wing at right angles before running out of pull.
Problem 7. Did pterosaurs really have weak hind limbs?
On pterosaur.net Habib reported, “For all of the large pterosaurs, insufficient strength was present in the hindlimbs to initiative a bird-like launch. In addition, for most species, a bipedal launch position would have placed the wings at an inappropriately high angle of attack.” Of course, modern stilts, flamingoes and storks do very well in their bird-style launches, even with their “stilt”-like leg bones. Bending forward at the hip would have positioned the wings at the appropriate angle of attack (see below).
As descendants of bipedal lizards with sprawling femora (see below), pterosaurs would have run like highly improved lizards (see video), likely during take-off without a headwind. The Bruce Jayne lab in Cincinnati, Ohio, documented a zebra-tailed lizard, Callisaurus draconoides, running quadrupedally and later in the video, bipedally (Irschick and Jayne 1999). Note, the heels of the lizard never touch the ground. Speeds reach 5 m/sec or 11 mph for a 10 cm (snout/vent) length. Note that footfalls occur every three body lengths. That’s fast! What could a better-equipped pterosaur do? We’ll have to wait for running/takeoff footprints to find out, but here is an animation of a running Quetzalcoatlus to fire-up your imagination (Figure 10). Successive footfalls occur at a snout to vent length, which is quite a distance!
It is important to note that bird legs and pterosaur legs had one fundamental difference: bird femora are not only tucked close to the body they are constrained by torso skin from moving much. So bird strides really begin at the knee. Birds elongate their metatarsals to produce yet another flexible leg section (the so-called “backwards knee”). By contrast, pterosaur femora swing from the hip and they all had relatively short metatarsals. Thus any mathematical comparisons, like those performed by Habib (2008) between the two types of flyers are going to be affected by this basic difference.
We end with a video (Figure 11) of several albatross taking off in a no headwind situation. With headwind an albatross need only unfold and lift its wings to become airborne.
I respect Mike Habib’s mathematical studies on pterosaurs, but I remain his loyal opposition for the reasons listed above. We have corresponded on this problem without swaying each other. Look for more studies from him in the near future. We’re both looking for those elusive pterosaur take-off tracks that will prove our hypotheses!
Every week since it was originally posted, this report on pterosaur wing launching has been a very popular post. There’s more to this story in a more recent blog here.
As always, I encourage readers to see specimens, make observations and come to your own conclusions. Test. Test. And test again.
Evidence and support in the form of nexus, pdf and jpeg files will be sent to all who request additional data.
Bennett SC 1997. The arboreal leaping theory of the origin of pterosaur flight. Historical Biology 123(4): 265-290.
Chatterjee S and Templin RJ 2004. Posture, locomotion, and paleoecologyof pterosaurs. The Geological Society of America Special Paper, 376:1–63.
Habib MB 2008. Comparative evidence for quadrupedal launch in pterosaurs. Zitteliana B28:159-166.
Mazin J-M, Billon-Bruyat J-P and Padian K 2009. First record of a pterosaur landing trackway. Proceedings of the Royal Society B doi: 10.1098/rspb.2009.1161 online paper
Padian K 1984. The origin of pterosaurs. In: Reif W-E. & F.Westphal, Eds. Proceedings of the Third Symposium on Mesozoic Terrestrial Ecosystems, Short Papers. Tübingen, Attempto Verlag: 163-168.
Schutt WA Jr, Altenbach JS, Chang YH, Cullinane DM, Hermanson JW, Muradali F and Bertram JEA 1997. The dynamics of flight-initiating jumps in the common vampire bat Desmodus rotundus. The Journal of Experimental Biology 200: 3003-3012.