How to Cherry-Pick Data and Taxa

Dr. Mike Habib (2008) promoted the hypothesis that pterosaurs probably launched using their forelimbs by first noting the greater diameter of the humerus vs. the femur in pterosaurs as compared to birds (Fig. 1).The “forelimb launch” notion (not yet proven with any tracks) has become quite popular. Mark Witton’s upcoming book on pterosaurs will feature many such take-offs, judging by the preview viewable here. Earlier we examined forelimb launch problems here,  here, here and here. Today we’ll note problems with the data in Habib (2008).

Cherry-picking taxa
Habib (2008) portrayed the humerus  : femur ratios of 14 birds and just 3 pterosaurs (Fig. 1). The birds Habib (2008) chose to use ranged from the flightless rhea and penguin to the ultra-soaring albatross. Ratios for birds ranged 2.25 : 1 (femur : humerus) for the rhea to 0.9 : 1 (humerus : femur) for the tropic bird. All three of the pterosaurs that Habib (2008) cherry-picked for this data chart (Dorygnathus, Zhejiangopterus and Anhanguera) exceeded the tropic bird in their humerus : femur ratio (Fig.1).

from Habib 2008 l

Figure 1. Click to enlarge. Ghosted chart from Habib 2008 listing 3 pterosaurs and 24 birds according to humerus vs. femur diameter ratios. Overprinted are several bird skeletons and their common names, so you don’t have to look them up from the binomial names. Also overprinted are several other pterosaur humeri and femora that Habib (2008) did not include and did not have such a large humerus vs. femur ratio. These would have nested near the center of Habib’s 2008 chart, well within the bird range.

There is no doubt that the vast majority of pterosaurs had a humerus larger in diameter than the femur, even when scaled to a common length. In some cases, as in many ornithocheirids, the ratio is quite enormous with very thick humeri and spindly femora. See Arthurdactylus and Boreopterus for some prime examples. In some cases (Jeholopterus) the humerus was much longer than the femur. In others (Zhejiangopterus), just the reverse.

Cherry-picking works both ways
When I cherry-picked six other pterosaurs I found similar diameters in the humeri and femora. These would have had a humerus : femur ratio around 0.0 (to use Habib’s data points). Just take a look at the data for Raeticodactylus, Preondactylus, Scaphognathus, Germanodactylus (the JME Moe 12 specimen) and the small germanodactylid, No. 13 (in the Wellnhofer catalog, Fig. 1). All of these pterosaurs would have nested in the center of Habib’s chart, along with owls and hawks. Notably, all of these pterosaurs had close relatives with thicker humeri and thinner femora. So they are oddballs, not typical.

Are there phylogenetic patterns here?
Not really. However, the Pterodactylus clade all had slender humeri, some more slender than their femora, but they were all derived from tiny pterosaurs with a thicker humerus.

Eudimorphodon ranzii had a much thicker humerus, but the closely related Eudimorphodon cromptonellus did not.

Cherry-picking data got Habib lots of press and notoriety.
And no one noticed or griped for the last four years. Here we can show with “twice as many pterosaurs” that Habib’s data was skewed to say the least. I like Mike. But his methods involve skewing data and cheating morphologies (which we talked about earlier) when he artificially shortened metacarpals and fingers 1-3 of his ornithocheirid example so the extensor tendon of the flight finger would be pinched between the weight of the pterosaur and the ground, which is impossible. Here using Habib’s own methods permits us to also falsify this bad hypothesis by using more inclusive data that was not originally presented.

Pre-loading the forelimb is like jumping with a pogo stick
Habib (2008) envisioned a vampire bat-like launch for pterosaurs, but that method is fraught with problems (except for vampire bats). It is especially unlikely when the humerus is the shortest element of the forelimb, as in the azhdarchid pterosaur, Zhejiangopterus.

A typical pogo-stick will only raise the rider only a foot off the ground. It takes an extreme pogo-stick using an air spring to reach the heights required for forelimb launch in pterosaurs. This is a device designed to work under extreme pre-loading.

MPUM 6009, the Milan specimen, the most primitive known pterosaur

Figure 2. MPUM 6009, the Milan specimen attributed to Eudimorphodon as a juvenile and attributed to Carniadactylus. This is the most primitive known pterosaur. It shares more traits with Longisquama. Note the long legs and relatively short forelimbs, atypical for most pterosaurs. Click to enlarge and for more information.

Using the pogo-stick analogy, it is difficult to envision the evolution of such a behavior in basal pterosaurs, already so well adapted to leaping and running with their hind limbs. As the wings got stronger, they provided more of the thrust so the hind limbs could shrink, which they did in all derived pterosaurs.

So maybe the forelimb launch hypothesis is NOT the answer.
If the humerus vs. femur rules do not apply to all pterosaurs, we might ask, what do you get when you have an inflated humerus (as demonstrated by most, but not all pterosaurs)?

There is no doubt a big fat pterosaur humerus is full of air. That makes it more buoyant when sitting on the water. It is interesting to see how really inflated some forelimbs (includes the ulna, humerus and wing metacarpal) of Pteranodon got. Check them out here. The question is, do the big humeri match to marine pterosaurs, the ones most likely to use floating as a behavior? Maybe so. But tiny Nemicolopterus also had a big humerus. There might be something to this. It would be worth investigating beyond this blog. Most Germanodactylus specimens had a big humerus, but the one I imagined was built for diving had a very small diameter humerus. Hmm.

Giraffe skeleton

Figure 2. Giraffe skeleton. Here the forelimb proportions are similar to those of the large Quetzalcoatlus, from the short, robust humerus to the elongated forelimb and metacarpus. I don’t see any giraffe preloading its forelimbs for any sort of leap.

A big fat humerus also provides more area for muscle attachment. The big Quetzalcoatlus humerus is much more robust than the smaller one. So this makes sense on that scale. The question then becomes, is the big humerus necessary for flight? For leaping? Or for walking around on the ground, supporting that weight and the superstructure of the elongated neck and giant skull? The big Q. does mimic the giraffe in certain aspects (Fig. 2), with its short humerus, long forelimbs, long neck and relatively smaller torso and overall size. Maybe terrestrial locomotion and some floating made the big Q humerus so big.

This is what makes science fascinating. Exposing problems in various hypotheses are just part of the process. On the other hand, whenever someone comes up with a forelimb launch track site, I will be among the first to announce it. I’ll even get excited the first time a wing metacarpal gets impressed into the matrix.

I do hope these presentations help Mark Witton to change his book before publication. It would be a shame to promote this bad hypothesis as valid.

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.

Habib M 2008. Comparative evidence for quadrupedal launch in pterosaurs. Pp. 161-168 in Buffetaut E, and DWE Hone, eds. Wellnhofer Pterosaur Meeting: Zitteliana B28

12 thoughts on “How to Cherry-Pick Data and Taxa

  1. Dave,

    Two things here. Firstly, your critique of Mike’s paper suggest you have not read or understood it properly. The graphic you show does not merely compare the diaphysis widths of pterosaur and bird limb bones, but their structural strength ratios. In Mike’s words, these are “Ratios [of] the length-corrected section modulus of the humerus, divided by the same value for the femur, and then natural logarithm transformed”. This is a bit more involved that simply comparing two bone widths, which you suggest is the case here. In addition, you ignore the fact that quad launch is supported by the scaling of pterosaur limb bones, which is shown in both ontogenetic and evolutionary series. In pterosaurs/birds over 500g, the humerus is stronger than the femur, and increasingly so with body size. I would suggest you look at some pterosaur ontogenetic sequences to verify this, but I don’t think you recognise any (otherwise, Rhamphorhynchus and Sinopterus would be good places to start). I strongly recommend that your readers read Mike’s paper on quad launch to see the other points in support of this launch strategy that your critique fails to address (download it, for free, at

    Secondly, please stop moaning about my book before you’ve even seen it, as you have done so here and in other posts. Unsurprisingly, it doesn’t follow the vast majority of your ideas, because I have stuck to discussions of peer-reviewed, testable hypotheses and effective use of evidence, which I do think make up the content of your websites. Nevertheless, your ideas are discussed where relevant, and -brace yourself- I even agree with one point you have suggested. I cannot stop you from attempting to hack my work to pieces once its published, but your efforts to discredit and undermine my book before you have seen or read it are unwarranted and rude, and I am politely asking you to cut it out.

  2. Certainly more involved than a simple diameter comparison, but the fact remains that the samples were cherry-picked. While most pterosaurs exhibit the differences Habib notes, many do not. Secondly, by posting previews of your book, you expose your ideas to scrutiny. Without such previews, there would be no scrutiny. If its’ out there, it’s fair game. Please take the criticisms as opportunities to answer them with better graphics and captions that answer the problems raised.

  3. I disagree that Mike ‘cherry picked’ taxa. The animals he chose are not closely related and are typical of the limb sizes common to the majority of pterosaurs. I appreciate that there are one or two pterosaurs with absolutely thinner humeri than femora (Caviramus, Preondactylus), but their bone diameters alone do not necessarily translate to bone strength. Mike’s strength ratios consider bone diameter, length, cross sectional shape and bone wall thickness, whereas yours only consider one of these factors. The comparison is not a fair one, and I suggest Mike’s work is considerably more telling about the structural properties of these elements than your simple comparisons.

    Re. my book. I’ll stick with the captions, graphics and text that I have, thanks, and I still consider your statements about my book being flawed (e.g. “Witton (2013)… will continue promoting bad morphology and misbehavior steeped in tradition and lacking in evidence” – your words) to be slanderous attempts to undermine my work. You have run into issues with this attitude of yours at least once before, when it got you permanently banned from the DML. You are well aware that the ‘problems’ you raise about pterosaur science are not considered valid arguments outside of your own writings, and your continual suggestions that others should alter their work because of your opinions are arrogant, rude and irritating. I am continually amazed that you consider a few hours spent tracing photographs of specimens and your hunches over the functional morphology of animals a substitute for the days, weeks or years that generations of real scientists have spent analysing actual fossils and developing biomechanical models. Your audacity to suggest that scientific literature should be rewritten because of your ‘findings’ on this website is beyond belief.

    • Mark, I hear hyperbole. When the “years” of study translate into detailed imagery and explanations, rather than half-baked cartoons, as is so often the case, then I will gladly jump over to the other side. When we find impressions of the wing metacarpal in ichnites, I will gladly jump over to the other side. When you find a wing membrane that is connected to the ankle, rather than the elbow, I will gladly jump to the other side. When you find a series of taxa within the Archosauria that demonstrate a gradual accumulation of pterosaurian traits, I will gladly jump to the other side. If my work is not considered valid outside of my own writings, you’re hanging around the wrong people. My work on pteroids just got confirmation from the Kellner group. So you’ve been falsified. Test the rest and see what happens.

  4. Dave, I think Mark is being pretty reasonable here. Look at this way: aren’t you going to feel pretty silly if it turns out that you misinterpreted one of the preview figures and tried to hack apart something Mark doesn’t even argue?

    On a related note, I like you too, but if you’re going to be condescending and accuse me (essentially) of scientific misconduct, then I’m going to have to get pretty firm with you. Let’s get to brass tacks: the physics of animals is basically mathematics. To do it right, you have to work in numbers. You essentially refuse to do so, and have even said publicly that you are not able to work the calculations. Now, that’s fine if you aren’t trying to form a counter-argument, but since you are, I’m going to have to call you on this: provide section modulus calculations to support your ratio argument and do a fluid mechanics analysis to determine how biped launch would work around the flutter and angle of attack problems.

    Now, as for the “cherry-picking”: I was very specifically and clearly interested in the problem of launch at large sizes in derived pterosaurs. I therefore choose a set of, gasp in amazement, medium to large sized, derived pterosaurs from a range of distantly related clades. I have never said that pterosaurs could not have bipedal origins. I have never claimed that all pterosaurs absolutely had to be quad launchers (though I suspect the vast majority were, if not all). I assume they were quadrupedal because, well, the tracks show they were quadrupedal.

    Oh, and enough about the fingers already. Your very own preferred finger arrangement works *better* for quad launch than the one Julia and I originally used, which I have remarked upon here and elsewhere. I don’t know if you’re correct, mind you, but at least make your arguments internally consistent.

    Finally: the limb ratios are only a piece of the overall issue with a bipedal launch in pterosaurs. You have never addressed any of them properly, probably because you do not know how to do aerodynamic analysis or estimate leaping performance. Jumping in animals does not work like a pogo stick, nor does the power magnification dynamics in bats or (as I have suggested) pterosaurs.

  5. As a side note: just a took a look at your water launch page. I think you must have missed the floating position work by Henderson and Hone. Lifting the wings as you have initially would force the head underwater. They already float dangerously as it is. The wings are not particularly good pontoons.

    • Speaking of the water launch page, Witton’s pterosaur never does launch. That needs to be shown. I (and many others) find most of Henderson’s and Hone’s work invalid (refs in many places here in the blog) so no points scored there, but send the paper if you have it. Floating and flying hasn’t been completely worked out, but certainly webbed feet and getting a little wet at a distinct angle could have provided the necessary stability in many cases for a wings launch, as in pelicans. Each specimen needs to be looked at individually.

      • “…certainly webbed feet and getting a little wet at a distinct angle could have provided the necessary stability in many cases for a wings launch, as in pelicans.” So you have numbers to show this? Incidentally, pelicans do not launch using wing power. The launch using a series of pushes from the hindlimbs, which are essentially leaps. Their launch is basically saltatorial. Most of the power is provided by the final leap, as evidenced by the fact that the last “splash” is the largest – they displace more water, not less, as the feet push, so the wings are not picking up more power as they go. Animals do not launch like airplanes.

  6. Jeepers, Mike, I gave you the last word.

    Alright, so to open some old wounds, today’s blog included some bipedal tracks and references to others, so all pterosaurs were not quadrupedal all the time as shown by fossil evidence. And they were secondarily quadrupedal, as we all know by several lines of evidence.

    Certainly no pterosaurs were able to produce thrust with their forelimbs while walking, as shown by matching tracks to trackmakers, so it seems odd that a mighty pre-loaded pushup could provide the necessary airspace for wing-unfolding and the first flap, especially in pteros with a small humerus and a big thigh.

    Given that, to your point, in vampire bats, with a large humerus and a secondarily evolved quadrupedal gait, the forelimb leap is possible. So, work with me here, can we figure out, phylogenetically, when your forelimb leap had its genesis in pterosaurs, given that pterosaur ancestors were hind limb leapers in any and every analysis?

    What was cool about the vampire presentation (ref not at hand) was that the numbers and videos were translated into graphics of the limb bones that duplicated the bat maneuvers in two views. It would be great if you could do something similar. Take the animation to graph paper and show essentially what the animation showed in top, side and front views with all the frames or representative frames portrayed in one, two or three figures. It’s already in the computer. Should be easy to print out. That way we (representing the vast majority of people who cannot reproduce your numbers and formulas without several more years of education) can see your evidence and how it does not break any speed, momentum or gravitational rules. You have to write to your audience. I need to see your pterosaur leaping to an apex, then slightly falling while deploying its wings prior to first flap, just like the vampire, both victims of gravity and mass.

    Failure to launch lines should also be shown (weak, injured, old and young pterosaurs unable to perform the acrobatics) to see how close to failure each take-off was and when they would have been strong enough to be successful. Crashes are devastating. This will give us all a sense of “breathing room” that we are currently lacking.

    Also, when you find a manus only take-off track I’ll jump on your bandwagon. Until then, best wishes.

    I’ve seen too many instance where numbers falsify and gloss-over reality, as in Bennett’s work with statistics and Rhamphorhynchus. You may be right on the money, but we’ll never know it until you show it.

    Yours truly,
    The “calls’em as I sees’em” loyal opposition.

  7. Dave: yes, you gave me the last word on the other thread, but then you started a new one where you accused me of scientific misconduct. I don’t think you meant it that way, but that’s how it came across.

    That said, your last reply in this thread is really reasonable and *way* more collegial. Thank you; it is appreciated. In fact, I think your suggestion regarding the frame-by-frame figure is wonderful. I’m going to do it. I can’t promise it for the next paper, but I have a couple I’m working on with takeoff bits in them and I’ll try my best to get it into the second one.

    You had two really good questions in there, too, so I want to do you the appropriate honor of answering them as best I can:

    1) Pterosaur forelimbs *can* produce thrust while walking if one uses a more horizontal gait, and this can also be made to match tracks (also the hands don’t have to fall behind the glenoid to exert a ground reaction force, but more on that later). Now, that said, let’s imagine for a moment that you’re right and the hands don’t do much walking propulsion. I doubt it, but let’s run with it. What’s interesting there is that the quad launch still works, because of the vault phase in the cycle. That takes the power from the hindlimbs (which we both know is not trivial) and basically puts in the bank by preloading the forelimbs. Because you can get more tendon stretching off the forelimbs, the animal basically gets interest on the deposit when the forelimbs unload. There’s even a somewhat ricochetal version quad launch that I haven’t talked about yet (sneak preview here folks!) that works for animals with big forelimbs that don’t normally touch the ground. *If* you’re right about pterosaur gait evolution, that would probably be the intermediate.

    2) And speaking of that transition: yes, with enough specimens we can find the transition point. My prelim data suggests it’s way back in the tree (probably a lot further than you’d tend to prefer, because I know you have a soft spot for bipedality, but that’s not a big deal here). Assuming pterosaurs did have bipedal ancestors (and Kevin Padian still seems to think so, too) then I expect this transition to be a bit complicated and mosaic, rather than simple, but who knows. That’s why paleontology is fun, right?

    Oh, and regarding safety factors: I’ve built a MatLab script that compares the biped and quad version for any size and basic pterosaur morphology, and graphs out the safety factor. Biped safety factors are much smaller. I think I’ll try to link it up with the Schutt et al. style figure you suggested; that should have a lot more punch than the simple graph.

    Have a safe holiday season, and it may it be filled with pterosaur related goodness.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.