The weird skull and affinities of Brachydectes

Before you read any further, check out Jason Pardo’s letter below. He’s the expert. I’m only a freshman when it comes to this very unusual taxon and its kin. 

AND > See Thursday, September 1 (based on the Int’l Date Line near New Zealand, 1pm CDT) for updates to the Brachydectes morphology. The phylogenetic placement did not change. 

Brachydectes newberryi (Cope 1868, AMNH 6941; latest Carboniferous; 300 mya; Fig. 1-4) was long considered a lysorophian amphibian with a tiny skull, an extremely long snake-like torso, vestigial limbs and a very short tail. You find them in eastern Kansas.

Figure 1. Brachydectes overall and skull in four views.

Figure 1. Brachydectes overall and skull in four views.

A recent PlosOne article
by Pardo and Anderson (2016) studied the skull of Brachydectes (Fig. 3) using micro CT scanning. They report, “Contra the proposals of some workers, we find no evidence of expected lissamphibian synapomorphies in the skull morphology in Brachydectes newberryi, and instead recognize a number of derived amniote characteristics within the braincase and suspensorium. Morphology previously considered indicative of taxonomic diversity within Lysorophia may reflect ontogenetic rather than taxonomic variation.” Later they wrote, “an expansive phylogenetic analysis is outside the scope of this study and will appear elsewhere.” 

Earlier
in the large reptile tree (LRT), Brachydectes nested between Adelospondylus and Eocaecilia, which also has a long snake-like torso, but composed of far fewer and individually much longer vertebrae and a distinct skull architecture. A large, but not exhaustive, selection of basal amniotes was tested and none attracted Brachydectes as much as the two lissamphibians listed above, given the prior data of a line drawing of the skull (Fig. 2) by Marjanovic and Laurin 2013 derived from Wellstead C F 1991.

Figure 1. Brachydectes skull data from a line drawing produced by Marjanović and Laurin 2013. Most leposponysls have a very narrow parasphenoid process and large interptyergoid vacuities, but eocacaecilians expanded this bone and reduced the vacuities like Brachydectes did. 

Figure 2. Brachydectes skull data from a line drawing produced by Marjanović and Laurin 2013. Most leposponysls have a very narrow parasphenoid process and large interptyergoid vacuities, but eocacaecilians expanded this bone and reduced the vacuities like Brachydectes did.

The new data 
(Figs 2,3 ) are not too far off from the Wellstead C F 1991 data. Notably the tabular no longer extends ventrally alongside the squamosal as it does in the larger specimen. Does this represent a break? or fusion? Or phylogenetic difference? Below (Fig. 3) is the new data on KUVP 49541, plus a reinterpretation of skull sutures based on the micro CT scans. The nesting of the new Brachydactes does not shift in the LRT. It is still a lissamphibian close to microsaurs and caecilians. That’s a broad range, indicative of a long list of yet to be found taxa.

Figure 3. Brachydectes skull morphology updated.

Figure 3. Brachydectes skull morphology updated. DGS tracing is more precise than freehand figures, as shown in figure 5.

Pardo and Anderson’s reconstruction
(Fig. 3) does not include the coronoid or lateral exposure of the splenial.  Pardo and Anderson note the single supraoccipital compares well with that of various basal reptiles, and indeed it does.  The occipital arch of other lissamphibians consists of only paired exoccipitals,.. until you include microsaurs.

More on supraoccipital homologies
According to Pardo and Anderson, “the presence of a well-developed median supraoccipital is restricted to the amniote crown and recumbirostran ‘microsaurs’. Although the supraoccipital of Brachydectes and ‘microsaurs’ has traditionally been considered convergent with the amniote supraoccipital, new data from μCT have demonstrated that the ‘microsaur’ supraoccipital shares a number of morphological details with early amniotes, and early eureptiles in particular, and is likely homologous with the amniote element. This homology does not extend far down the amniote stem, as seymouriamorphs lack a supraoccipital and ‘anthracosaurs’ generally exhibit paired elements within the synoptic tectum.” 

Noteworthy:
In the LRT, microsaurs are sisters to the clade that includes Adeospondylus, Brachydectes and Eocaeceila. That’s a great deal of phylogenetic distance, but not as great as any other pairing in the LRT. Perhaps more taxa will fill the apparent gaps someday.

Figure 4. Four sizes of Brachydectes in situ. Here, unfortunately, the authors have penned in the sutures, negating any possibility of any reviewer to judge whether they were drawn correctly or not.

Figure 4. Four sizes of Brachydectes in situ. Here, unfortunately, the authors have penned in the sutures, negating any possibility of any reviewer to judge whether they were drawn correctly or not.

Pardo and Anderson also report
“neurocranial morphology does not support a close relationship between Brachydectes and lissamphibians.” Admittedly, Brachydectes is indeed quite different from its sisters…yet it is not closer to other tested taxa in the LRT. If you look at various microsaurs and other lissamphibians, you get a wide range of morphologies at every node.

By noting various key features in contention with the traditional relationship. Pardo and Anderson essentially ‘put the cart before the horse.’ They waited to do the phylogenetic analysis, when they should have done that analysis before publishing. Homoplasy is rampant in tetrapods. I think they fell prey to yet another example. Only analysis, at present, settles all issues.

Pardo and Anderson then report, 
“Morphology of the braincase of Brachydectes suggests a close relationship with the brachystelechid ‘microsaurs’ Carrolla craddocki  and Quasicaecilia texana, within the Recumbirostra.” These two are new to me and untested in the LRT. Wikipedia nests them with Batropetes, which has long legs, and a horned-lizard type body, only distantly related to Brachydectes in the LRT. The skull of Quasicaecilia is shown here, but no post-crania is shown. Recumbirostran microsaurs, are considered the earliest known example of adaptation to head-first burrowing in the tetrapod fossil record. I wish the sister candidates offered by Pardo and Anderson were long and snake-like, but they are not. Deletion of post-cranial traits from the LRT does not shift the placement of Brachydectes within the LRT.

Figure 3. Original interpretation of Brachydectes, KUVP 49541, by Pardo and Anderson. Colors added for clarity and to match micro CT scan.

Figure 5. Original interpretation of Brachydectes, KUVP 49541, by Pardo and Anderson. Colors added for clarity and to match micro CT scan.

References
Carroll RL 1967. An Adelogyrinid Lepospondyl Amphibian from the Upper Carboniferous: Canadian Journal of Zoology 45(1):1-16.
Carroll RL and Gaskill P 1978. The order Microsauria. American Philosophical Society, Philadelphia, 211 pp.
Cope ED 1868. Synopsis of the extinct Batrachia of North America. Proc Acad Nat Sci 20: 208–221. doi: 10.5962/bhl.title.60482
Jenkins FA and Walsh M 1993. An Early Jurassic caecilian with limbs. Nature 365: 246–250.
Jenkins FA, Walsh DM and Carroll RL 2007. Anatomy of Eocaecilia micropodia, a limbed caecilian of the Early Jurassic. Bulletin of the Museum of Comparative Zoology 158(6): 285-366.
Marjanović D and Laurin M 2013. The origin(s) of extant amphibians: a review with emphasis on the “lepospondyl hypothesis”. Geodiversitas 35 (1): 207-272. http://dx.doi.org/10.5252/g2013n1a8
Pardo JD and Anderson JS 2016. Cranial Morphology of the Carboniferous-Permian Tetrapod Brachydectes newberryi (Lepospondyli, Lysorophia): New Data from µCT. PLoS ONE 11(8): e0161823. doi:10.1371/journal.pone.0161823. online here.
Wellstead C F 1991. Taxonomic revision of the Lysorophia, Permo-Carboniferous lepospondyl amphibians. Bulletin of the American Museum of Natural History 209: 1–90.

wiki/Eocaecilia
wiki/Brachydectes
wiki/Adelospondylus

19 thoughts on “The weird skull and affinities of Brachydectes

  1. Hi David,

    First, thank you for your interest in this paper.

    The drawing you refer to Marjanovic & Laurin 2013 is actually adapted from Wellstead (1991). Marjanovic & Laurin reinterpreted a few bones of the skull based on comparisons with Batropetes, but their interpretation is not based on a study of any fossil material, just intertaxon comparisons. As far as I understand, Marjanovic at least is now convinced of the interpretations of the skull bone homologies that we present here. The interpretation of the skull bone homologies presented by M&L2013 can be treated as defunct.

    With respect to the phylogeny, I can assure you that a phylogenetic analysis of Brachydectes, ‘microsaurs’, and other ‘lepospondyls’ has been completed; this was a substantial portion of the thesis work that Pardo & Anderson 2016 stems from. This phylogenetic analysis is itself a pretty massive undertaking, with a large number of new neurocranial characters and introduction of new data from several important taxa which have not previously been treated in these analyses (particularly the aistopod Lethiscus). Publication of these data necessarily has to occur piecemeal, because of the extensive amount of new information we need to report. This work is all in the pipeline, but has to be finished before a phylogenetic analysis can be published.

    With respect to the similarities/differences between Brachydectes and other ‘lepospondyl’ taxa (we have serious questions about the monophyly of this group), the braincase of Brachydectes is extremely similar to that of Carrolla craddocki and Quasicaecilia texana. Descriptions of both taxa have been published recently from high-quality CT data (Maddin et al., 2011; and Pardo et al., 2015), and these taxa can be clearly shown to share morphology of the antotic foramina, the orbitosphenoid, the basisphenoid, and the occiput. While the overall body shape of Batropetes and Brachydectes is pretty different, similar lability of trunk length is seen in a lot of modern lizard groups (e.g. scincids), so the idea that Brachydectes and Batropetes might be differently adapted but closely-related (as many small scincids are) isn’t totally insane.

    With respect to your reinterpretation of the antorbital region in Brachydectes, I can assure you that our interpretation of the prefrontal is correct, based on tomographic data from all the specimens we examined as well as inspection of those specimens under the microscope. There is a distinct scarf joint construction of the suture between the prefrontal and the frontal and parietal, where the prefrontal partially overlaps the medial elements. Some of the sutures of the skull roof are difficult to make out in CT, but this one is really easy to see and can be repeatedly identified in both slice data as well as whole specimens. Similarly, the “suture” you draw between the frontal and parietal is a crack, not a suture; the frontal-parietal suture is deeply interdigitated, and can be seen relatively readily under the microscope and transiently in the micro-CT data. Similarly, the sutures between the postparietals and parietals are pretty clear, but are transient enough that it is difficult to segment these out as different bones for the entire length of the suture. Most of your lines seem to be associated with cracks in the specimens. I will note that, while the sutures in the skull roof of the Kansas and Nebraska material are pretty tightly-closed, there are a lot of specimens from the redbeds of Oklahoma and Texas which are preserved under slightly different taphonomic settings, and the sutures are even more apparent and agree closely with what we’re seeing in the Kansas-Nebraska material.

    As an aside, in your interpretation of the position of the lower jaw, you’ve disarticulated the quadrate-articular joint and displaced the articular posteriorly.

    Again, thank you for your interest. Brachydectes is an interesting animal, even though we disagree on why it’s interesting.

    • Now this is what I like to see. Emotion-free scientific discourse. Thank you, Jason. Will be even more interested to see where you nest Brachydectes if not among the caecilians and kin. These have always been among the most diverse tetrapods in the panoply. Not really like anything else out there.

      • Does this not make you question all of your other results, since yet again when an expert weighs in they find your reconstructions are inaccurate? You’ve heard this exact issue, where your ‘sutures’ are cracks, for ceratopsians, pterosaurs, theropods, etc.. How many times will it take before you consider that your technique doesn’t work well enough to code from? Surely there’s some limit to the amount of times any technique is shown to provide a false positive until it’s discontinued. So what is that limit for you?

        Also, Pardo stated what he thinks Brachydectes is probably closely related to- Batropetes. I know very little about lepospondyls, but the fact you find taxa with similar body shapes to be closely related is a red flag, just like Gauthier et al.’s squamate analyses that found all the legless taxa to group together. Your broad brush analysis is simply not designed to recover convergence.

  2. Pardo also mentioned amniote affinities. Mickey, I encourage you to look at the subject of statistics, which is the basis for phylogenetic analysis. One, a dozen or a hundred mistakes do not upset the results of tens of millions of data points. Mistakes will be corrected. This is the scientific process. Get used to it. No one has a perfect record.

    re: your attitude. You and I are metaphorically ‘swimming in the same pond’ with regard to our blogposts and you don’t like results that differ from yours, or traditional results, a subject that was covered a few blog posts ago regarding Darren Naish. Your response here is typical of primates competing for the same bananas and a higher rung on the status ladder. On that note, please do yourself a favor and drop Gauthier a note on your thoughts on his squamate analysis. Or is his primate status too high?

    And finally, sometimes legless taxa DO nest together. Sometimes they do not.

    • Pardo mentioned amniote affinities because he agrees with most recent studies that find lysorophians to be closer to amniotes than to lissamphibians. So my point stands.

      So again you say that no matter how many times your interpretations are proven wrong, you’ll stand by your method? That’s been your response to versions of this question multiple times now, but surely must not actually be true. If hypothetically every single novel interpretation of yours was shown to be wrong by an expert in that taxonomic group, you’d give up, right? It would be irrational to argue otherwise. So there is SOME bound to how wrong your results can be before you reject your methods. I’m just asking what that boundary is. (btw, your matrix only has ~160 thousand data points, not tens of millions)

      As for Gauthier, I did criticize him on the Dinosaur Mailing List when that publication came out. If he had a blog, I would do so there, but he does not.

  3. Mickey, I urge you not to have students, not to have children, not to volunteer wherever you might interact with others wherever creativity and learning is involved. You have no tolerance for people making errors. You would have not been welcome at the Edison labs where they were testing light bulb filaments. You also have no tolerance for overturning untenable scientific traditions. Your emotions lead the way. Take a lesson from Pardo, please. And remember, it’s not the technique, it’s the mindset, the preconceptions, what happens between the ears that is the limiting factor in DGS.

    On the other hand, you are welcome to list all errors you find in my reconstructions and matrix. That’s why I made them.

    Not every aspect of every taxon is included in the matrix scores, that’s why I estimate that every point and line on every illustration represents tens of millions of additional unscored data points.

    You are also welcome to list all taxa within the tree that should not nest together. Find the taxa that do not demonstrate a gradual accumulation of traits and suggest a new nesting. On that point, you missed several opportunities with Protictis, Jesairosaurus and others.

    “Now, release your anger. Only your hatred can destroy me.” — Darth Vader. : )

  4. So… you didn’t answer my question, and instead just insulted me? I’m serious. It’s a standard scientific question- how would you know you are wrong? So far you haven’t given an objective answer. As for Protictis etc., I’m not saying you ARE wrong, I’m asking when you will think you are uninformed to answer the question.

    To put it basically, if I were adding taxa to a matrix and found ceratopsid workers saying I didn’t get the anatomy right, that would be an issue. Then I tried theropods, and their workers said I misinterpreted that. Then I coded a temnospondyl, but according to its describer I got that wrong, I would stop thinking I had a useful handle on that group. But you’ll just go on with it, assuming you’re more right than not. How would you defend that position? (and NO, saying your tree roughly matches morphology doesn’t matter because no one else cares about that).

  5. Mickey, apparently you don’t realize that a large phylogenetic analysis like the LRT is like a sponge or a coral in that you can chop into parts and each part will still survive and thrive. The LRT is also like a city, like London during the Blitz. Kill or deport several citizens and the city continues without them. You also fail to notice that error correction typically cements a phylogenetic nesting. And when we talk about getting something wrong, that doesn’t mean we got -everything- wrong. It means we got 96/100 or 98/100 right and a few things wrong. Finally, you fail to appreciate how many errors I correct without shouting about them.

    I would not have made the errors you found if I had done more homework before publishing. I regret them, but correct them. Nut cases and ego-psychos don’t correct their errors. Scientists do.

    Your hopeful hypothesis of several mistakes bringing down the house reminds me of Creationist thinking in that if Adam and Eve did not exist, if Mary wasn’t a virgin, then the whole of Christianity is up for grabs! Some of them really think like that. Don’t think like that. Allow a margin of error. It’s going to happen. It happens all the time to every scientist.

    The LRT tree topology has not changed from 260 to 790 taxa. That means you can delete 500 taxa and the tree topology will not change. I think I once cherry-picked it back to 60 or so taxa and it still had the same topology.

    • And you still didn’t answer my question, which is about your reconstructions, not even the resulting tree.

      Your LRT is like a sponge because you almost entirely use big obvious characters and never add new characters. So you’ll never have the case where e.g. a hundred added braincase characters change the topology. Similarly, your “error correction typically cements a phylogenetic nesting” because much of what you see as error correction is actually you finding homoplasy and reinterpreting the taxon to incorrectly fix that (e.g. Herrerasaurus’ premaxillary subnarial process in 2012). Thus these aren’t arguments in favor of your tree’s accuracy.

      Your topology _has_ changed numerous times since the intervening period. Back when your tree had 300 taxa, poposaurs were phytodinosaurs, Herrerasaurus was a theropod, the Saltopus clade were pseudosuchians, Lewisuchus was outside Archosauria, erythrosuchids further from archosaurs than Euparkeria, plus a ton of rearranging between many genera, etc.. So I hope you don’t use the argument your topology hasn’t changed again, since it’s untrue.

      Your argument re: religion is dissimilar in the sense that theology isn’t bound to reality. Thus someone can add unlimited ad hoc explanations so that Christianity could still exist even if his followers think Jesus and the supernatural never existed and that the New Testament is complete fiction, but they like some of their own interpretations of the scripture. There are near certainly people like that calling themselves Christians today.

      I think this actually may be a previously hidden factor into why you disagree with basically everyone else. To most scientists, the process is more important than the results. If the process is flawed, that can really mess up the results, and using a flawed process is untrustworthy even if the results look great. Whereas you dismiss everyone’s critiques of your process (your characters are based on a flawed tracing methodology, you don’t have enough of them for the taxon number or include suggested competing characters, you don’t order any, many are correlated or composite, etc.) because your results look good to you.

      Similarly, most scientists would say that as science disproves some aspects of Christianity only supported by scripture (the Genesis creation, Noachian flood, virgin birth, etc.) that the rest of that scripture-supported information (whatever else Matthew or Luke wrote, Yahweh/El, the reality of Jesus, etc.) loses trustworthiness too. The results of the reality of Christianity really do depend on the process of interpreting its data. There are Christian scientists who accommodate, but they are the minority and I doubt you’d want to equate their explanations of why they are still Christian (generally non-scientific or appealing to areas outside their expertise) with why you still believe in your topology.

  6. If you think the process is flawed, the results SHOULD be flawed. Whenever the process was flawed the results WERE flawed and, when discovered, the flaws were corrected, as you listed. The basic tree topology and most of its large branches has not changed since the inception. And it still remains unchallenged and unique. I have been transparent in announcing the smaller branch changes. The current character list is sufficient to lump and split all included taxa. ALL workers have some degree of flaw in their process and methods, especially if they choose to use traditional taxon inclusion lists rather than relying on a tested large gamut study to pull a branch from.

    You make a good lawyer, Mickey. Now, be a good scientist and please point to the flawed current results and we’ll move on from there.

  7. PS You’ve also adopted the Darren Naish attitude of once I have made a mistake, I can never be trusted again. That’s why Dr. Naish rummaged through the trash of my discarded ideas to critique the website, which contained none of those concepts. Of course this removes all opportunity for redemption, learning and the eventual gaining of expertise and insight. And it removes readers and referees from having to judge current work based on the bias that it cannot be any good because past flaws were found. On the other hand the bias enables you two to cast a blind eye to any PhD who insists that turtles are sauropterygians, pterosaurs are archosaurs, diadectids are pre-amniotes or caseids are pelycosaurs, etc. etc. Those mistakes are to be ignored in the Naish-Mortimer attitude, perhaps because they are popular and widespread. Please, Mickey, put on your thinking cap. Don’t listen to your reptile brain. These are ideas, not threats.

    • When you say “the basic tree topology and most of its large branches has not changed since the inception”, I can say the same about the coelurosaur-centered tree I’m working on as I add taxa but not characters. It doesn’t mean it’s likely to be correct, it means given only those characters, taxa are generally going to fall out in a certain range of relationships. That’s why adding characters is an important part of testing relationships. More difficult than adding taxa, but important.

      “You’ve also adopted the Darren Naish attitude of once I have made a mistake, I can never be trusted again.”
      If I see you making those same mistakes constantly, then yes. Paul Sereno for instance makes phylogenetic analyses which don’t include many characters that contradict his published tree. Thus his results aren’t trustworthy examples of what the data we know says. It’s not “blackwashing” Sereno to say then I don’t think his future trees are accurate if he hasn’t changed his methods. You’re the same way. You have a list of issues (as I noted above and have detailed before) which make your results unreliable. I’ve not seen your theropod tracings improve in accuracy, you have yet to change your character definitions, order any characters, etc., so my belief in your accuracy remains low. You COULD redeem yourself and actually take some of my high level advice (for character ordering and separation, etc.), but you refuse to do so after years of critique. PAUP just uses algorithms, it’s not magic. The numbers you’re feeding it aren’t structured in a way to let it use its assumptions correctly, so you’re getting flawed results. It’s as if you’re coding whether the pterygoid is (0) blue, (1) over a foot long, (2) no, or (3) over a meter long. PAUP’s gonna spit out something, but it’s not going to be great. Until you value the process of character formulation over the result of thinking your tree looks good, things aren’t going to improve.

  8. Let it go, Mickey… or tell me which current taxa are not nested correctly as sisters. Which, in your opinion, or after your testing, do not make sense? The system is working. A gradual accumulation of traits is apparent in all derived taxa.

    • Don’t you notice that you just fell right into what I’m claiming your flaw is? You like your results, and thus your process is fine to you (or as you would say “the system is working”). But I care more about the process, and your process is inadequate. This is what we always fall back to. You always want a critique of your result (“tell me which current taxa are not nested correctly as sisters”), but really the issue is your bad process. I don’t know why you don’t get that, since you even admitted “If you think the process is flawed, the results SHOULD be flawed.”

      Your oft repeated defense that your tree is good because “a gradual accumulation of traits is apparent in all derived taxa” doesn’t even make sense. Why expect the accumulation of traits to be gradual, when the consensus among evolutionary biologists is that it actually varies between groups and times? And have you even measured how gradually changes occur in your tree compared to others’ trees? I doubt it.

  9. Mickey, You know and I know, no matter how many traits and processes I add or correct they will never be enough for you. I just found two taxa that will be nested elsewhere based on added attractive taxa. Thus the process and the data are sufficient to echo evolutionary events = the process is adequate. I don’t know why you don’t get that.

    A gradual accumulation of traits competes only with a shocking or radical appearance of traits found in competing results.

    • “You know and I know, no matter how many traits and processes I add or correct they will never be enough for you.”

      Not true. You’ve never added any characters, but if you did, that would be an improvement. That would be partially fixing your process, which would partially fix your result.

      “I just found two taxa that will be nested elsewhere based on added attractive taxa. Thus the process and the data are sufficient to echo evolutionary events = the process is adequate. I don’t know why you don’t get that.”

      I don’t get that because your statement doesn’t make sense. Because of the result that ‘taxa change position when new taxa are added’, that somehow validates the process? That not only contradicts one of your central claims (the topology is stable thus your analysis is good), but it’s also yet another example of you valuing the result instead of the process. The result can never validate the process (unless it agrees with some independent result with a good process, like molecular analyses or biogeography, which yours do not).

      “A gradual accumulation of traits competes only with a shocking or radical appearance of traits found in competing results.”

      I’ll repeat myself- “Why expect the accumulation of traits to be gradual, when the consensus among evolutionary biologists is that it actually varies between groups and times? And have you even measured how gradually changes occur in your tree compared to others’ trees?”

  10. For the record:

    The drawing you refer to Marjanovic & Laurin 2013 is actually adapted from Wellstead (1991). Marjanovic & Laurin reinterpreted a few bones of the skull based on comparisons with Batropetes, but their interpretation is not based on a study of any fossil material, just intertaxon comparisons. As far as I understand, Marjanovic at least is now convinced of the interpretations of the skull bone homologies that we present here. The interpretation of the skull bone homologies presented by M&L2013 can be treated as defunct.

    Every word of this is true. We – well, I – took Wellstead’s reconstruction for granted and proposed different interpretations of the identities of some bones. Pardo & Anderson, and also to some extent my personal observations of specimens soon after the 2013 paper came out, have shown very clearly that Wellstead’s reconstruction is itself wrong in a few crucial aspects. That makes my reinterpretation obsolete. For my upcoming paper I have discarded it and instead scored Brachydectes after Pardo & Anderson (2016).

    One of the details will likely interest you. Wellstead reconstructed a posttemporal foramen, but it wasn’t surrounded by the usual bones. My reinterpretation of some of those bones restored the usual homology. In reality (Pardo & Anderson 2016), the posttemporal foramen is simply not there; Wellstead must have been misled by damage or by an attempt to reconstruct the 3D shape of a crushed skull.

    On to your post:

    Most leposponysls have a very narrow parasphenoid process and large interptyergoid vacuities

    Temno-, not lepospondyls. Very few lepospondyls have large interpterygoid vacuities.

    Earlier
    in the large reptile tree (LRT), Brachydectes nested between Adelospondylus and Eocaecilia, which also has a long snake-like torso

    That’s actually a red flag right there. And this is because body elongation is correlated to a long and incompletely understood list of characters. Many matrices barely make an attempt to take this into account; they just contain these characters as separate characters. This means that what should be a single character is counted five or ten times, and its influence on the tree is inflated by the same amount.

    As Mickey pointed out, you’re in very good company here. The Gauthier team got the snakes, the amphisbaenians, the dibamids, and the most thoroughly limbless anguid (Anniella) nested within a clade of limbless skinks, itself nested within the limbed skinks. I really don’t think any one of the authors actually believes this result.

    From your comments:

    On the other hand, you are welcome to list all errors you find in my reconstructions and matrix. That’s why I made them.

    But who would pay Mickey’s costs of living in the meantime?

    Applying for a grant to look for mistakes in a matrix that isn’t even published is hopeless.

    Mickey, apparently you don’t realize that a large phylogenetic analysis like the LRT is like a sponge or a coral in that you can chop into parts and each part will still survive and thrive. The LRT is also like a city, like London during the Blitz. Kill or deport several citizens and the city continues without them. You also fail to notice that error correction typically cements a phylogenetic nesting.

    In all of these aspects, your matrix is very, very different from the ones I’ve worked on. I’ve found the consequences of adding taxa or correcting errors to be completely unpredictable again (2008b with Laurin) and again (2009 with Laurin: supp. inf.) and again (2015 with Witzmann) and again (preprint 2015/forthcoming with Laurin).

    Specifically, in that last matrix, correcting Brachydectes after Pardo & Anderson (2016) had the effect of restoring it as the lissamphibian sister-group in analyses of the original sample of 102 taxa. However, increasing that sample to 150, so that it includes Carrolla and Quasicaecilia, but not adding any characters, completely prevents this from happening; instead, the sister-group of Lissamphibia (or sometimes Lissamphibia + Albanerpetidae – those are a bit jumpy) is a clade composed of Brachydectes and the three brachystelechid “microsaurs” Batropetes, Carrolla and Quasicaecilia. Stay tuned, we’ll submit the next version soon.

    The mentioned addition of taxa also has the effect of highlighting what a huge mess temnospondyl phylogeny is. With fewer taxa it looks neat and tidy, but that’s all deceptive.

    BTW, the postcranium of Quasicaecilia remains unknown; Carrolla and Quasicaecilia are known from one isolated skull each.

    If you think the process is flawed, the results SHOULD be flawed.

    Not every single time. Statistically, we expect the results to be correct by chance occasionally.

    However, if the process is flawed, then the results won’t be reliable. This is why it’s so important to be right for the right reasons. If you’re right for the wrong reasons (or for no reasons, just random chance), you’ll probably be wrong next time!

  11. re: interpterygoid vacuiities – consider Gerobratachus as an example.
    re: long snake-like torso – can your provide a better sister, even one not on the current taxon list?
    re: errors – I find errors all the time (recent post on moles and mongooses). I’m not asking for a concerted effort, just if one or a dozen jump into view, let me know of them.
    re: unpredictable results from adding taxa – these are ideas that lead to discussions and solutions. Aren’t scientists supposed to look for optional answers to vexing problems? We’ll figure out the right answers together as the wrong answers are cast aside. And it’s interesting that adding taxa is adding problems in your matrix. Amphibians, with their ontogeneic changes, may be more difficult than other clades.
    re: not every single time – so true, but not due to chance, which is random, but by bias and error as happened, again, with moles and mongooses (rarely mongoose). In such cases another look at the results usually finds a ‘fly in the ointment.’ BTW, I’m not relying on random chance for my scores, which should be obvious, and hopefully pops your bubble of hope that the LRT cannot possibly provide any value.

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s