Debunking ReptileEvolution.com? More Bogus Evidence Incoming!

Preamble: I’ll admit that there have been errors on ReptileEvolution.com brought about by a variety of circumstances including inexperience and bad data.  Every effort has been made to fix those problems and dozens have already seen changes made. I found virtually all of them, but others have helped. None of these have affected the topology of the large reptile tree at the center of the site. None of my major issues have, so far, been reversed with evidence coming in from the outside. Even so, I’m still open to the possibility.

Can the traditional guys say the same? Hope so! Haven’t seen it yet, though.

Now Back to our Debunking Topic
Since Darren Naish’s blog urging web surfers, “Why the world has to ignore ReptileEvolution.com” other bloggers have joined in. As scientists we need to explore their arguments to see if they contain validity. It’s important to promote the truth in science. Anything less needs to be corrected. That’s why I’m interested in what these bloggers have to say. Given that, I’m –still– not seeing from them –any– specific arguments that hold water. See if you agree:

Pterosaur.Net wades in against ReptileEvolution.com
Pterosaur.Net is the work of a group of paleontologists who hold that pterosaurs are dinosaur kin, had deep chord wing membranes and were not able to rise to a bipedal configuration. Plus this group holds that pterosaur eggs were buried prior to hatching. and they took off by rebounding off their forelimbs. ReptileEvolution.com and PterosaurHeresies.com have provided more than adequate evidence against all of these bogus ideas. [note that I’m not saying don’t believe anything in Pterosaur.net. It’s important to make that distinction.]

Mark Witton, writing for Pterosaur.Net quoted Darren Naish when he wrote, ReptileEvolution.com does not represent a trustworthy source that people should consult or rely on. Students, amateur researchers and the lay public should be strongly advised to avoid or ignore it.”  Naish and now Witton risk appearing to be a little heavy-handed in their assessment by not recognizing something of value here. We looked at Naish’s bogus arguments earlier. Sorry to see Witton buy into that without critical thinking and good evidence.

Dr. Witton was kind enough to provide what he considered evidence for his arguments with regard to mistakes I made with pterosaurs. I’d like to say that he provided some insights to help correct errors, but alas, he did not. You know, I always ask for anyone who has evidence to the contrary to step forward.

Witton compared my Anurognathus reconstruction to a photograph of another, much smaller pterosaur purported to be Anurognathus (Bennett 2007, Fig. 1), but actually is a distinct genus, with a much flatter skull and many other distinctions. Here Witton could have shown a photograph of the specimen in question.  He chose not to. Here’s how different the two are from one another:

The flat-head pterosaur

Figure 1. The flat-head pterosaur, a private specimen (on the left) attributed by Bennett (2007) to Anurognathus ammoni (on the right).

But let’s go with Witton’s comparison anyway. Witton wrote, “Note the number of phalanges in the wing finger, shape of the skull and long, fibrous tail [in the flat head specimen].” Unfortunately, Witton failed to note that no other anurognathids have only three phalanges on the wing membrane. You can see samples here, here and here. He failed to note that the original Anurognathus preserves a crushed skull that portrays the lateral and dorsal views with palatal elements scattered about and that the flat head pterosaur is preserved strictly in dorsal view. Earlier we discussed and provided evidence for the mistaken interpretations Bennett (2003). By his own admission Bennett was unable to identify several bones. His big gaff: he misinterpreted a maxilla as a gigantic sclerotic ring preserved on edge (which never happens in other fossils).

Bennett’s reconstruction breaks several “rules” that apply to pterosaur morphology. The ReptileEvolution.com tracing identified all the bones. Left and right elements were symmetrical. In the reconstruction all the bones “fit” and none of them broke any morphological “rules.” Rather, every bone looked like bones in other anurognathids, which no one else had ventured or dared to attempt. The eyes were small. The maxillahad tiny teeth.  On the flat head anurognathid, three wing phalanges are indeed exposed on the surface, but as we looked at earlier, there is evidence for more phalanges still buried in the matrix. This problem could be readily solved with just a little more digging. That goes for the tiny, bead-like tail too.

It’s just too bad that anurognathids have not been precisely traced by any other paleontologists. Here’s an example of a good tracing. Other pterosaur workers seem to be content tracing coarse cartoony outlines of only the most readily observable features, as shown here.

Witton continues, “the resultant [phylogenetic] trees are understandably completely incongruous with anything seen in ‘mainstream’ literature.” Of course this makes ReptileEvolution.com look bad. — AND–  exploring untested taxa together is the whole reason for ReptileEvolution.com. To create a tree large enough to test the assumptions of smaller trees. That’s why it is completely incongruous with anything the the mainstream literature.

You want to test the large reptile tree? Go ahead and take out all the controversial reconstructions and you still recover the same tree. I know it works.

What Witton doesn’t tell you is the new pterosaur tree includes many times more specimens than ‘mainstream’ trees, many of which suffer from poor resolution, by their own admission. In other trees, sister taxa do not look like each other (mixing toothy and toothless taxa, for instance). By contrast, the ReptileEvolution.com trees all recover sister taxa that all look like one another, gradual producing derived taxa. After all, isn’t that what we expect from a model echoing the course of evolution?

Witton continues, “he [Peters] retains his ideas in the face of overwhelming evidence to the contrary (i.e. the inability to see the structures he claims to find on actual specimens despite microscopic, and UV observation, and CT scanning). As such, the work portrayed on his website and blog has to be pseudoscience, at best.” Well, here is where Dr. Witton could be most helpful. If he indeed had evidence to the contrary, I would hope that he would present it. Withholding it then complaining about it strikes me as unfair. If he’s referring to published works. To be fair, though, I’ve been just as tough on the guys at Pterosaur.Net, not disbelieving them, but dismantling and disproving their various weird contentions. The pseudoscience is rampant over there! No wonder they’re pointing fingers.

Witton references Hone et al. 2009, a paper that attempted to bash Parallel Interphalangeal Lines (PILs) without success. The rebuttal (Peters 2010) noted how many times Hone et al. (2009) reported that in many taxa PILs could indeed be drawn. In their arguments Hone et al. (2009) did not provide any examples with five or four toes, which is where the best PIL sets are to be found. Even their prime example, a three-toed theropod (Ornithomimus) with flexed toes retained PILs (*see below for comment). Hone et al. 2009 went so far as to provide a set of bogus computer-generated cat toes to demonstrate that PILs could not be drawn between the phalanges. Sensing pseudoscience, Peters (2010) countered with tracings of actual cat feet and paw prints to show complete sets of PILs, despite the extreme flexion and extension in the very derived cat toes. Truly a worst-case scenario that came out smelling like a rose.

pterosaur wing ungual

Figure 2. Pterosaur wing ungual almost articulated to manual 4.4 from an unidentified Chinese pterosaur.

Witton republished a fourth and fifth wing phalanx from an unknown pterosaur and claimed that I invented the ungual. Seems pretty clear to me. More to the point, as I described earlier, I was — not– the first to see the odd curved, hooked tip. That honor goes to David Hone! …another member of Pterosaur.Net! Hone described the oddity as a malformation in his blog here without understanding its significance. Contra Witton’s criticism, wing unguals are actually quite common, if not universal, in pterosaurs. Unfortunately, so often the fragile tip remains buried in the matrix or the ungual becomes dislocated. A careful search usually results in a discovery.

Dr. Witton concludes, “Hence, I urge you to read Darren’s discourse if you have not already done so and, if you are concerned about the accurate portrayal of palaeontological science online, then blog, tweet and discuss this issue as much as you see fit. As may be expected, Peters has started a rebuttal of the piece across a number of blogposts, which begins here.”

So here we are, again, much as in Darren Naish’s blog, with a refusal to “see” what has been described by others, to “see” what is very clear as evidence, and to use disassociated evidence to argue against my work. Heavy-handed? You be the judge.

Argument Styles
Lastly, I want you to note the difference between my approach and that of Witton and Naish. I don’t say, “don’t look at or believe anything on their sites.” I say. “take a closer look at their evidence in specific cases and here’s a reinterpretation based on precise tracings and more expansive analyses.” I know they’re sincere in what they say, but when direct scientific comparisons are not made their arguments come off as prejudice and propoganda, which is not their intention. They don’t even know how bad they sound. I would hope that someday both Witton and Naish would draw attention to a specific problem in ReptileEvolution.com and provide circles and arrows that highlight their arguments. Then we will have something to discuss.

Simply disrespecting an entire website, as if it had no value whatsoever is absurd. Painting me as a lunatic tells us more about them, ironically. Picking apart bogus arguments with facts and evidence is more my cup of tea. Wish it was theirs’.

* Regarding the Ornithomimus pes presented in Hone et al. 2009, I wrote (Peters 2010), “On a similar note, the caption beneath an illustration of a very similar Ornithomimus pes (Hone et al. 2009, figure 4) complains that ‘only three possible hinge lines can be drawn’, and this is despite the fact that the toes are shown unnaturally parallel (recovery stroke configuration) rather than radiating in use as theropod ichnites otherwise indicate. Fewer toes mean fewer lines. That was spelled out in Peters (2000a). Still, it is a testimony to the HLH (=”Hinge Line Hypothesis according to Hone et al. 2009) that Hone et al. (2009) were able to find hinge lines despite deforming a three-toed foot. It is also a testimony to the HLH that Hone et al. (2009) were unable to show any manus or pes in which PILs could not be drawn, no matter  the number of fingers and toes. If they were to someday attempt another attack on Peters (2000a), such evidence would be required.”

References
Bennett SC 2007. A second specimen of the pterosaur Anurognathus ammoni. Paläontologische Zeitschrift 81(4):376-398.
Hone DWE, Sullivan C and Bennett SC 2009. Interpreting the autopodia of tetrapods: interphalangeal lines hinge on too many assumptions. Historical Biology, iFirst article, 2009, 1–11, doi: 10.1080/08912960903154503.
Peters D 2010. In defence of parallel interphalangeal lines. Historical Biology iFirst article, 2010, 1–6 DOI: 10.1080/08912961003663500

10 thoughts on “Debunking ReptileEvolution.com? More Bogus Evidence Incoming!

  1. If you want to give yourself a little more solid ammunition in favour of your tree, how about some bootstrap values? Or a character consitency index? We know the characters you have selected support the tree, but HOW STRONGLY do they support it? I suspect that due to the character list being quite small, the support will actually be quite low. But if I’m wrong, you give yourself a considerably stronger case then if you just say ‘here’s a tree’. You will be able to say ‘here’s a strongly supported tree.’ (PS I emailed a while ago asking for the data marix for your big tree and the one for your modificaton to Amson & Laurin’s tree. any chance of you emailing them to me?)

      • Neil, I’m partitioning as I write. Also, note that all sister taxa look alike, primitive taxa have a gradual accumulation of derived traits, there are no bizarre “strange bedfellows.”

      • Ok I’ve had a look at the dataset David kindly provided. I have noticed a couple of issues with the character scoring, but I’ll communicate them to him personally in a few days after I’ve had a more detailed look. What I wanted to talk about here was what I mentioned earlier about support, and not just vague support like ‘sisters look alike’ and ‘no strange bedfellows’ (I would be very carfeul before you say that) but actual statistical support, for which I have done a few analyses using your dataset unmodified.

        First, I plugged your dataset into TNT, and used their new technology search to see if any new topologies came out (these should be more thorough than a heuristic search). None did, so there’s a good start. However from then, unfortunatly, things start to go downhill. Using the Bremer decay index I measured support for specific nodes, I found some well supported, but several extremely poorly supported (values of 1 or 2 were common, when ideally you would want at least 3, and preferably 5, before you can say a clade is well supported). A couple of examples (mostly from Synapsids and parareptiles as these are my areas of interest): your linking of Caseasaurs and Parareptiles has a support value of 2; Ophiacodontids as a sister taxon of therapsids is 1, your clade containing Heleosaurus and other Varanopids as well as many other Diapsids: 2. In fact high level relationships are very poor in general: collapse all nodes with a support of 2 or less, and enormous plytomies appear, particularly around the Synapsid parareptile area.

        The most damming assessment is the Character Consistency Index (a measure of how well an entire character list supports an entire tree): 10%. Thats TEN out of 100. Ok there’s a lot of taxa, so perhaps one shouldn’t expect values of 70 or 80, but TEN???? Klassen et al 1991 were getting better results than that with RANDOM datasets

    • Indeed, which is why I cast a skeptical eye at them too ;)

      On a more serious note, there are 3 major points I would like to make in response to that statement.

      Firstly, yes, there are several nodes within the current amniote tree which are poorly supported. However, there are several nodes, which are well supported, and occur again and again and again in the recent literature, always well supported, such as a dichotomy between Synapsids and Sauropsids, or Sphenacodontids as the sister to Therapsids. It is these nodes which are being argued against in this blog. I’m not saying the tree presented is wrong. I am merely saying that if you are going to go against the order established by countless morphological studies, as well as molecular in some cases, you need a much better argument than a node in a tree with a Bremer suppot of 1.

      Secondly, yes this is not the only tree that this criticism could be levelled against. I am equally skeptical of relationships in other phylogenies which have Bremer support of 1 or 2. But studies which contain these nodes either point out themselves when their relationships are weak, or they at least put the Bremer or Bootstraps on the tree so that the reader can see. The big tree does not. Relationships are shown, given to us in a blog post, and then left at that. I do not think that is right. If Dave Peters wants to say that he disagrees with the ‘canon’, that is fine, but he should at least acknowledge when the support for the new relationship is not great.

      Finally, I wanted to put more emphasis on the low consistency index than the Bremer support. And yes this can be low for other studies, but 10%??? Even Amson & Laurin’s 2011 phylogeny, which contains much I would like to criticise, had a CI of more than 50%.
      I did intend to add another paragraph to my previous post, but pressed send too soon. It is this. The low CI does not necessarily indicate the tree is wrong. IT INDICATES THAT THE CHARACTER LIST IS TOO SMALL TO MAKE A VALID JUDGEMENT! Add more characters, the CI will grow. Having a character list smaller than the number of taxa may get you a phylogeny, but it will never get you a well supported phylogeny. That was going to be the main thrust of my last post. But I forgot to say it.

      • The CI is extremely low because so many reptiles converge on one another = high homoplasy. The CI number is lower than random because only a relatively few traits work best and these keep popping up in evolution. The character list is large enough to produce complete resolution. The character list was tested earlier by removing every 10th character times and just recently by deleting post-cranial characters. What is interesting is that with just this list you can separate every single reptile of the 315 taxa listed. Bremer numbers could be helped by finding more characters, but also many of the basal taxa are woefully incomplete, providing relatively fewer traits to separate them from one another.

      • You’re falling into the trap of assuming that because you got one tree it means the tree is good. The number of trees you get is irrelevant to support. You can have a single tree with poorly supported relationships, or a poorly resolved tree but the nodes that are resolved are well supported. I would believe the latter over the former. This is why people put support on their phylogenies in publications. It tells the reader what nodes are reliable. The node being resolved does not mean it is right.
        Not sure I agree with your comments about the CI. The whole point of a phylogenetic analysis is to minimise homoplasy. If there’s a lot of homoplasy, it means that the characters you have selected are not sufficient to reliably resolve the relationships (see above, resolution isn’t enough, good resolution is needed).

        Anyway, as I said above the low support does not mean you are wrong, it just means more work is needed. support is a statistical problem, it can be questioned by words, but only dealt with by actions. I might, if I get a free weekend, experiment by adding some characters to your matrix. In the menatime I will email in a bit about some issues with the character scorings

  2. @neil brocklehurst comment #2443

    To be fair, those complaints could easily be levelled against most current analyses of amniote or diapsid relationships (e.g. poor Bremer support for most nodes and so on).

Leave a Reply

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

WordPress.com Logo

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

Google photo

You are commenting using your Google 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.