Carnufex: a new Triassic proto-dinosaur (not a proto-croc)

Carnufex carolinensis (NCSM 21558, Zanno et al. 2015, Fig. 1) is a new Carnian (earliest Late Triassic, 231 mya), croc(?) based on fragments of a skull, a cervical neural spine, a dorsal neural spine and a humerus (Fig. 1). And yes, it was a likely biped, but it was a little closer to dinos than to crocs as it nests with the proto-dinos Pseudhesperosuchus and Junggarsuchus.

Figure 1. Carnufex is basically a giant Pseudhesperosuchus. Here they are compared to one another to scale and with skulls side by side. Dark gray areas are imagined on the original at bottom by Zanno et al. Click to enlarge. With a skull 4x larger than that of Pseudhesperosuchus, Carnufex was a likely 4.4 meter long bipedal killer. Note the smaller orbit and deeper jugal. Both neural arches are missing a centrum.

Figure 1. Carnufex is basically a giant Pseudhesperosuchus. Here they are compared to one another to scale and with skulls side by side. Dark gray areas are imagined on the original at bottom by Zanno et al. Click to enlarge. With a skull 4x larger than that of Pseudhesperosuchus, Carnufex was a likely 4.4 meter long bipedal killer. Note the smaller orbit and deeper jugal. Both neural arches are missing a centrum.

Unfortunately the Zanno tree did not include Pseudhesperosuchus
in their cladogram and that is the sister taxon to Carnufex in the large reptile tree. Unfortunately Zanno et al. used the very troubled matrix of Butler et al. 2014, derived from the equally troubled matrix of Nesbitt 2011 in which pterosaurs are included derived from parasuchians and sisters to ornithosuchids, aetosaurs and lagerpetonids. As you can tell, almost nothing could be worse and that cladogram has now been through at least four generations of papers. Earlier we looked at problems with papers based on Nesbitt et al. 2011 here and here.

Zanno et al. nest Carnufex between the giant rauisuchians Rauisuchus + Postosuchus and tiny bipedal crocs, reporting that Carnufex bridges the gap between them. Also missing from their tree is Decuriasuchus, which does bridge the gap between basal (not derived) rauisuchians like Vjushkovia (also missing from their tree) and basal poposaurs like Turfanosuchus, which gives rise to Gracilisuchus at the base of the Archosauria (dinos and birds + crocs).

In the Zanno et al. tree Turfanosuchus is derived from a sister to the armored Revueltosaurus and the odd Ornithosuchus, in turn the sister of tiny flying pterosaurs in that tree. I’ll stop there before it gets weirder.

Without Pseudhesperosuchus
the reconstruction of Carnufex goes a little bit astray (Fig. 1). Even so, you can readily see that that humerus is so small compared to the skull that Carnufex was a likely biped (to their credit Zanno et al. recognized this despite lacking hind limb material) just like Pseudhesperosuchus. 

Unfortunately the Zanno et al. tree did not nest
basal crocs  just a short phylogenetic step from basal dinos. Thus, with taxon exclusion they did not recognize that Carnufex is closer to basal dinos than to basal crocs and not related to derived rauisuchians. Hopefully someone out there will recognize this phylogenetic solution someday.

The origin of crocs + dinos still sits with Gracilisuchus. Taxon exclusion and inaccurate matrix scoring are keeping the pros away from replicating this very robust result.

Butler RJ. et al. 2014. New clade of enigmatic early archosaurs yields insights into early pseudosuchian phylogeny and the biogeography of the archosaur radiation. BMC Evol. Biol. 14, 128.
Nesbitt SJ 2011. The early evolution of archosaurs: relationship and the origin ofmajor clades. Bull. Amer. Mus. Nat. Hist. 352, 1–292.
Zanno LE, Drymala S, Nesbitt SJ and Schneider VP 2015. Early Crocodylomorph increases top tier predator diversity during rise of dinosaurs. Scientific Reports 5:9276 DOI: 10.1038/srep09276.

37 thoughts on “Carnufex: a new Triassic proto-dinosaur (not a proto-croc)

  1. I’ve said it before and I’ll say it again, the Nesbitt matrix and all those based on it do not show pterosaurs derived from parasuchians, and they most certainly do not show pterosaurs to be the sister to ornithosuchids and aetosaurs. You’re arguing a straw man. I am genuinely baffled as to how you can look at the phylogenies they produce and think that’s what they are saying

    • The sister to the pterosaurs in the Nesbitt tree and its expansions are the dinosauromophs (Dinosaurs and silesaurids and the suchlike).

      The clade containing both pterosaurs and dinosauromorphs is found to be the sister to the clade containing aetosaurs and ornithosuchids and various other bits and bobs (including crocodiles).

      The parasuchians are found to be the sister to the clade containing pterosaurs, dinosaurs and crocodiles. Not the ancestor to it; they are an independent lineage with more than 20 million years of separate evolution, much of it alongside pterosaurs.

      If you’re going to criticise other people’s phylogenies, you really need to know how to read a phylogeny. Firstly, this is a cladogram, so it can not show direct ancestors, it shows which taxa share the most recent common ancestors, represented by a node. So saying X is shown to be derived from Y is impossible, unless Y is a series of paraphyletic outgroups to X (this is definitely not the case in parasuchians and pterosaurs).

      Secondly, just because the names of two clades are written next to each other, that does not mean they are sisters. The way phylogenies are drawn can lead to two very distantly related taxa with millions of years of independent evolution may end up with their names written next to each other. This does not mean they are sisters. Clades are sisters when they share a more recent common ancestor with each other and with no other clades. The way to do this is not to look at where to find the node that links the branches leading to the two sets of tips. If this node leads to only the two clades of interest then they are sisters. If the node has other descendants, the two are not sisters. A phylogeny has only one axis, and that is distance from node to tip. The position of the labels relative to eachother is irrelevant.

      This is the point I’ve been trying to make in so many comments. You keep saying that these trees find pterosaurs and parasuchians together, and that proposing this is ridiculous because of how different they are. But no publication has ever suggested they are sisters or that one is derived from the other. No publication has ever suggested that the flying pterosaur evolved from an aquatic ancestor. The publications suggest that pterosaurs and parasuchians both belonging to a monophyletic group that includes , aquatic, semi-aquatic, terrestrial, cursorial, arboreal, flying, carnivores, herbivores, piscivores, omnivores, ovivores, small, large, super-large, bipeds, facultative bipeds and quadrupeds.

      • Oh, naive Neil. David knows this, but has his own definition of ‘sister’ he uses. To David, if you’re the first to branch off within a subclade, your ‘sister’ is the sister group of that clade (or the sister group’s most ‘basal’ member). Ignore the fact that branching order depends on how much you sample each subclade. So in a traditional phylogeny, the ‘sister’ to ornithischians is silesaurids. Certainly not saurischians. The sister to iguanodontids is camptosaurids, not hadrosauroids. Why Peters doesn’t use the same definition everyone else does, we’ll never know.

  2. You’re basing your argument on the hope that somewhere in the ghost lineages of parasuchians, Euparkeria and pterosaurs you’ll find a common ancestor. And you do. In this case it is a sister to Gephyrostegus bohemicus in the large reptile tree. Cladograms can show ancestor relationships or at least sister to that ancestor relationships — when they are done right. When done right you can trace the acquisition and loss of traits — and you don’t paint yourself into a corner by having vestigial lateral digits turn into hyper-elongated flying mechanisms. There is a lack of critical thinking involved in the Carnufex cladogram. Evolutionary patterns here too often don’t make sense. And you know that.

    In the large reptile tree pterosaur traits are built up in a steady progression, accelerating at the base of the Fenestrasauria in Cosesaurus. Turning a blind eye to better solutions is not good Science.

    To your earlier point, dinosaurs have their own ancestry apart from Lagerpeton and pterosaurs. The large reptile tree found that the Dinosauria and the basal bipedal Crocodylomorpha are sisters (the Archosauria), which is not recovered in the Carnufex tree.

    I found dozens of errors in the matrix for what served as the basis for the Carnufex tree. Chase after those errors and you’ll find that tree, with those repaired scores and a completely diffferent set of characters, will turn out to be very close to the topology of the large reptile tree.

    • You’re dodging the issue. I’m not making a point about what your tree shows and the character evolution and what not. You’re welcome to your opinions on topology and morphology. I wasn’t complaining about that

      What I’m objecting to is that you are misrepresenting the work of others by you warped definition of “sister” and “derived from.” If you’re going to write a post on why other phylogenies are wrong, you need to describe the phylogenies properly, not argue against relationships that have never been suggested.

      And no, cladograms can not show ancestor-decendant relationships. You can attempt to infer ancestor-descendant relationships from them, but the cladogram itself does not show these The cladogram is a picture showing branching orders. In the cladogram the ancestors are hypothetical. There are methods by which tips can be inferred to be ancestors, but the cladogram doesn’t show this.

  3. Let’s agree to disagree on this. I’ve got a cladogram with Icthyostega at the bottom and lots of extant taxa, including humans, chickens, gators, snakes, etc. at terminal nodes — and logical pathways from one to another. That tells me I’ve got a mirror or model of the actual evolutionary pathways here. I agree with you that parasuchians/chanaresuchians/euparkeria have never been promoted as pterosaur ancestors (except in pterosaur trees, where the latter often nests as an outgroup taxon), yet the Nesbitt 2011 tree and all tree based on it show this. That has to stop. It’s ridiculous.

    • But The Nesbitt tree and those based on it do not show that! They really don’t. I don’t know how I can explain this more plainly! I have detailed as clearly as I can what the Nesbitt tree does show and I have explained why it does not show what you claim it shows. How much simpler do I have to make it?

      Oh, and PS, outgroup does not equal ancestor

  4. I”m glad we agree that the Nesbitt tree does not show relationships that mirror actual evolutionary pathways. Outgroup SHOULD be the closest sister to the clade under study. As such it SHOULD be the sister to the unknown ideal ancestor. We’re arguing over semantics here. The main issue remains taxon exclusion and bad scores that recover trees that do not reflect evolutionary patterns.

    • Are you being deliberately obtuse? Or just a not bothering to read my comments.

      I’m not saying that at all. I’m saying that what you say the Nesbitt tree says is not what the Nesbitt tree says! I honestly can’t say it any more clearly. I’m not making a comment in evolutionary relationships, I’m commenting on your inability to read a phylogeny.

    • So you are making claims of truth. Not evidence-based conclusions in comparison to others, each with support that must withstand the test of time; rather, you are claiming an absolute true tree, and this is one only you’ve recovered? Yours, with its ridiculously limited, cherry-picked, and terribly qualified character sets and codings, and presumptions that typological and teleological arguments are sound, when none of this has been shown to be true.

      You misunderstand cladistics; you seem to want to read trees where taxa are ancestors, not individuals, as if a taxon can be ancestral to any damn thing (it cannot); you misunderstand branching orders and stepwise transformation, which is read into the cladogram after it’s been produced. Instead, you seem to have developed an analysis that results in your conclusion, reached beforehand. No mystery here. Mickey’s gone on at length on the fallacies inherent in your “systematics” and their understanding, and it boggles my mind how the preceding comment could occur from anyone serious about the study of cladistic phylogenetics. This misunderstanding smacks of a whole different order, but in league with, that fundamental error of thinking from Feduccia, Olson, and others.

      There is nothing to lose here but respect for continuing to espouse this ridiculously terrible idea of how cladograms work, and phylogenetic reconstruction operates; there is only respect to be gained for sitting back and learning how it all works. I suggest taking some remedial courses on this. Please, for all our sakes.

      • What illegal substances are you taking? Everything here is evidence based. If what I have is so wrong, please do us all a favor and let us know which of the two taxa listed in the large reptile tree should not be sisters, and where should they nest separated instead? For having such bad methods, the results seem to come out quite well with all sisters resembling one another. And thank you for your kind remarks.

  5. Then — you’re saying the undiscovered common ancestor of pterosaurs and ornithosuchians is closer to parasuchians than to any other included taxa, right? Because both branch off from an unknown ancestor to the parasuchians (their own terminal taxon) that nested somewhere between parasuchians and Euparkeria. Right? How else should I read this? Still difficult to imagine.

    When you say what the Nesbitt tree is not… instead of what it IS… you’ re following Feduccia logic. Put your arguments into a positive format. Tell me what it IS. What can I glean about pterosaur ancestry and the gradual accumulation of pterosaurian traits from the Nesbitt tree?

    In a similar vein, the closest sisters to pterosaurs in the large reptile tree are Longisquama and Sharovipteryx. Both have, to all appearances, diverged from the unknown main line pre-pterosaur, which probably looked like Cosesaurus with longer fingers. But we can’t be sure. Even so, Longisquama shares more traits with pterosaurs than Sharovipteryx and Cosesaurus do.

    By Nesbitt nesting pterosaurs where they do in that cladogram, he’s telling us that no other taxa on that tree share more traits with pterosaurs. And that may be true given the taxa and scores that were input. That is called a ‘by default’ nesting. It only occurs in the absence of accurate scoring and taxon exclusion. I’ve recognized and corrected similar mistakes when I added more basal snakes to the large reptile tree cladogram, which then pulled the odd burrowing forms away from the Heloderma clade and nested them with the non-burrowing snakes.

    With regard to ornithosuchian sisterhood, do this for me: on the Nesbitt/Sues/Zanno tree chop off all taxa more derived than pterosaurs on one branch and chop off all taxa more derived than ornithosuchians on the other, and you’ll get a node with pterosaurs nesting next to ornithosuchians. Or you should if the tree topology is correct. A good cladogram can handle such topology cuts and still make sense. Does that experiment help, a least a little?

    By default nestings can be interesting and silly. Back in the day, with 360 taxa, Proganochelys, the turtle, nested with pterosaurs at the base of the sauropterygia when all other taxa were removed from the lepidosauromorph side of the tree. That’s bizarre, but interesting. Today, with 508 taxa, and deleting all other lepidosauromorphs, the pterosaur now nests between Cathayornis and Struthio, the ostrich, + Gallus, the chicken. Proganochelys, the turtle, now nests with the frogs, between Doleserpeton and Gerobatrachus + Rana.

    Delete the amphibians and add Huehuecuetzpalli and the three lepidosauromorphs: the turtle, the lizard and the pterosaur, all nest together again in their own clade at the base of the Sauropterygia… in other words, nowhere near dinos, pre-dinos, parasuchians or Marasuchus. Without Huehuecuetzpalli and amphibians, Proganochelys nests with Henodus, as you might imagine, while the basal pterosaur bounces back to the birds.

    Bottom line: by including more and more taxa the large reptile tree provides more and more nesting sites, minimizing ‘by default’ nestings. The Nesbitt tree suffers from ‘by default nestings’ and several hundred bad scores.

    • Right. I’m going to have one last desperate effort to explain what I’m trying to get across to you. I honestly can not believe you are finding this so difficult.

      Let us first get one thing very clear. I am making no comment on your tree. I have made no reference you your tree, and it is entirely irrelevant to my comments.

      Secondly. I am not espousing any opinion on the Nesbitt tree. I actually don’t agree with a few things on the. Nesbitt tree, parasuchians ammong them. But again, that is not relevant to the point I am making. I am not making any comment on archosaurs relationships at all, about taxon inclusion or morphology

      All I am trying to do is tell you what the Nesbitt tree actually says. I am not trying to say he is right. I am not trying to compare his result to yours. I have made no reference to characters. I am simply trying to tell you what the cladogram is showing since you don’t seem to understand how a cladogram works.

      So, I have said his before and I will say it again.The cladogram shows pterosaurs to be the sister to dinosauromorphs, not aetosaurs and ornithosuchians. Before you get so defensive, remember, I have not said I think this is right or made any comment on your tree. I am merely telling you what the Nesbitt tree says.

      The Nesbitt tree says parasuchians are the sister to a clade containing ornithodirans and crurotarsans. Again, before you get defensive, I have espouses no opinion on the comparative merits of his and your phylogenies. I am merely saying what the Nesbitt tree shows and pointin out it is not what you say it shows.

      No I’m not saying that the undiscovered common ancestor of pterosaurs and ornithosuchians is closer to parasuchians than to any other included taxa. How do you get that from what I said?

      And finally the point on ancestry. I never said you can’t infer evolutionary histories and character evolution from a phylogeny. You can, of course you can. What I was trying to get across is that he cladogram on it’s own does not tell you these things. It requires post hoc analyses of the phylogeny, characters and stratigraphic ranges. The cladogram itself does not show any ancestors. You can infer ancestors from analyses after you have the cladogram, but that is an extra step.

      Do you understand what I’m saying yet? I’m not making comments on the merits of yours and nesbitts trees, I’m saying you’re misinterpreting nesbitt’s tree!

  6. You wrote: “The cladogram itself does not show any ancestors. You can infer ancestors from analyses after you have the cladogram, but that is an extra step. It requires post hoc analyses of the phylogeny, characters and stratigraphic ranges.”

    I ask you, where in the academic literature is this “extra step”, the “post hoc analyses”? Evidently I’m not taking that extra step…

    Does any new data enter this extra step?

    Stratigraphic range appears to be not an issue or a factor. Some taxa cross many chronological borders.

    Characters are a given, or are we looking for absence of a trait than cannot or should not return, like snake legs?

    Analyses of phylogeny: “the history of the evolution of a species or group, especially in reference to lines of descent and relationships among broad groups of organisms.”

    That gets kind of close to ‘ancestry’. I’m trying to understand, but you seem to be saying what I’m saying. The fine line, if there is one, is not clear based on your instruction.

    If you have trouble with my concept of a sister taxon, that changes as taxa are added until you get all the way down to two taxa with the same parents. That can also change, theoretically, if taxa are removed, so cousins and kin can become sisters by removing intervening taxa. Right?

  7. Ah, we’re getting somewhere. Yes, my problem is with your definition of a sister. A sister should be the closest relative. A lineage’s sister should be the lineage that shares it’s most recent common ancestor. What worries me about your definition is that purely aesthetic changes such as collapsing a lineage or spinning a node, which change the appearance of the cladogram without changing the relationships it shows, will lead to you inferring a different sister, and it shouldn’t!

    I’m not sure if you’re being sarcastic or disbelieving when you say “I ask you, where in the academic literature is this extra step, the post hoc analyses?”. There is an extensive literature on ancestral state reconstruction, and it is a complicated issue. You can do it just by finding the most parsimonious ancestral state, but that is problematic, even if you ignore the issue you raised (some character changes are more likely than others). For example, parsimony assumes a constant rate of change along all branches (well known to be untrue), does not take into account time (on a longer branch, there is more time for character change, and therefore a higher probability of convergence; this is why stratigraphic ranges and time calibration method is important) and other issues (see references below).

    For these and other reasons, ancestral state reconstruction by likelihood or Bayesian methods are generally considered better. These effectively model the evolution of a character over a tree multiple times, in order to deduce which set of ancestral states is most likely to have lead to the observed states we see in the tips. The good thing about this is that it not only tells you which ancestral state is most likely, but also how much more likely it is than other possible states. Anyway here are some references on all this.

    Cunningham, C. W. (1999). Some limitations of ancestral character-state reconstruction when testing evolutionary hypotheses. Systematic Biology, 665-674.

    Omland, K.E. 1999. The Assumptions and Challenges of Ancestral State Reconstructions. Symposium: Reconstructing Ancestral Character States: 604.

    Ronquist F. 2004. Bayesian inference of character evolution. Trends in Ecology & Evolution 19:9.

    Mooers A., Schluter D. 1999. Reconstructing Ancestor States with Maximum Likelihood: Support for One- and Two- Rate Models. Systematic Biology 48: 3.

    SCHLUTER, D. 1995. Uncertainty in ancient phylogenies. Nature 377:108–109.

    SCHLUTER, D., T. PRICE, A. Ø. MOOERS, AND D. LUDWIG. 1997. Likelihood of ancestor states in adaptive radiation. Evolution 51:1699–1711.

    As for inferring the ancestors themselves, this is a hideous issue and not one to be taken lightly. See Bapst (2013) for some of the problems we have with how our current phylogenetic methods end up resolving ancestors. Smith used a set of criteria by which a tip could be inferred to be ancestral to it’s sister; one of them is it should have no autapomorphies, since in theory an ancestral taxon should have either the states of the ancestral node or the derived state of it’s descendant. Even this is problematic.

    Anyway some more references:

    Bapst, D. W. (2013). When can clades be potentially resolved with morphology?. PloS one, 8(4), e62312.

    Smith, A.B. 1994. Systematics and the fossil record: documenting evolutionary patterns. Blackwell Scientific Publications, Oxford

  8. Neil, thank you for this reference list. I looked at all of them and note that many are theoretical pursuits, others are behavioral. All seem to cast doubt on the ability of a morphological cladogram to infer an undiscovered ancestral state. I note that most were written at the genesis of computer-assisted phylogenetic studies in the late 1990s. If one had anything to do with vertebrate morphology, I missed it.

    I am not inferring ancestors here. I am letting the tree show me that, knowing full well that what is known in the fossil record probably represents the greatest radiation of the taxon (which makes it easier to find) rather than the first or last of its kind. And knowing full well that what is known in the fossil record typically represents a morphologically distinct taxon, a cousin, from the undiscovered true ancestor, yet close enough to stand in as a “stunt double” with autapomorphies noted. We have to work with what we have and keep searching for the missing pieces of the puzzle.

    With every addition to the large reptile tree taxon list I am further minimizing the average phylogenetic and morphological differences between all of them. That reduces uncertainty because nested sisters more closely resemble one another. This is a never-ending quest as most of the puzzle pieces will never be found. Even so, with what has been gathered, an accurate picture is beginning to emerge. The tree can be tested by pruning taxa or traits. It is also tested with the addition of taxa (and traits were necessary to increase resolution.)

    With regard to stratigraphy, I’ll note that specimens of the basalmost amniote in the tree, Gephyrostegus bohemicus, were found in Westphalian strata 30 million years younger than more derived amniotes found in Viséan strata. I can infer that something close to G. bohemicus must have preceded those Viséan amniotes. That’s not much of a leap in faith or science. That’s all I’m able to do with regard to ancestral states.

    I think my work showcases an ability to produce a lineage for every included taxon and show a gradual accumulation of derived traits for each one of them, with the proviso, of course, that the true ancestors remain unknown and their cousins provide some, but not all, and sometimes too many of these accumulating traits.

  9. Neil, you’re trying way too hard. David, yes or no, do you understand that in Nesbitt’s (2011) and Zanno et al.’s (2015) trees, the sister group of pterosaurs is Dinosauromorpha? Your choices are-

    A. No, you’re inexcusably ignorant of phylogenetic terms, as you’ve been corrected numerous times on this exact topic.

    B. Yes, but you like how bad it sounds rhetorically to claim your opponents think e.g. Ornithosuchus is sister to pterosaurs.

    C. Yes, but you use a different definition of ‘sister’ than everyone else, and don’t care that you’re misleading readers by doing that.

    I’m pretty sure it’s C, as I don’t think you’re stupid, or that you would lie to malign your opponents, but do think you have the philosophy that once you start doing something your way, you just stick with it (e.g. are any of your characters ordered yet? probably not). So if you’d just answer C, then you can stop wasting Neil’s time.

  10. As you know, Mickey, at the fork in the road, the two tines are sisters. I’m asking you to see the two tines that occur more derived than parasuchians in the Zanno/Nesbitt tree. What are they? (sorry to be patronizing here, but you started it.) On one tine/leaf/branch: pterosaurs (and kin or pretending for the moment that dinosauromorpha never evolved). On the other tine/leaf/branch are the orinthosuchians (and kin or pretending for the moment that aeotosaurs, turfanosuchians, rauisuchians, crocs and kin never evolved). Your revulsion to this concept is completely natural and understandable. It is revolting. It doesn’t makes sense. It required tens of millions of years of ghost lineages, reversals, autapomorphies, etc.

    On a personal level, why is this revolting? Because you logically understand that the fork in the Zanno/Nesbitt tree involves evolutionary improbabilities by any stretch of the imagination. You understand that evolution works in small steps, not giant improbable leaps. Unfortunately, like Darren Naish, you are a human being. And you have loyalties. Your gut / world view / paradigm / loyalties tell you that professionals are correct and amateurs / loners are nuts. So now you are faced with an amateur who is telling you the professionals are in error. This inner conflict needs to find resolution… and so far you’re following your gut, not your brain. You’re a human. It happens.

    • “On one tine/leaf/branch: pterosaurs (and kin or pretending for the moment that dinosauromorpha never evolved). On the other tine/leaf/branch are the orinthosuchians (and kin or pretending for the moment that aeotosaurs, turfanosuchians, rauisuchians, crocs and kin never evolved).”

      A fun experiment for you. Do an ancestral state reconstruction for the five relevant nodes using the Nesbitt tree and character matrix. 1) the ancestor of the clade containing phytosaurs, crurotarsans and ornithodirans; 2) the phytosaur node; 3) the archosaur node (ornithodirans and crurotarsans); 4) the ornithodiran node; and 5) the crurotarsan node. By this I don’t mean choose a taxon from the tree that you consider to be representing the ancestral state at the nodes, and I don’t mean come up with a vague description of what you think from looking at the tree the ancestor will look like. I mean take the character matrix and optimise the characters over the tree using likelihood. Hell, do it using parsimony if you like. This will tell you what the most likely morphology of the ancestors was. And then see what characters are required to change to get from node 1 to node 2, from node 1 to node 3, and from node 3 to node 4, and from node 3 to node 5. Then see if you still think it is really so unlikely that a node immediately ancestral to phytosaurs could also be immediately ancestral to archosaurs.

      You can’t base your entire line of reasoning on comparing just three clades (parasuchians, pterosaurs and ornithosuchians) out of dozens descended form the relevant node. You have to look at all the taxa, as all the taxa will contribute to what we infer the ancestral state of the node to be (at least, they should, if you’re doing the ancestral state reconstruction properly). When Nesbitt suggests that phytosaurs are the sister to the two main archosaur clades, this does not imply that the ancestor to the two archosaur clades looks anything like a phytosaur. This is the problem I was trying to get across above: Your attempts at ancestral character state reconstruction just by looking at the cladogram simply don’t work, because you will only be able to imagine the ancestor being like one of the tips. We have no reason to suspect that ancestors cannot have a morphology outside the range of the descendants, and it is only by doing a proper ancestral state reconstruction (which, I’m sorry, requires an actual quantitative analysis, not just visually inspecting a phylogeny) that we can show this.

      The entire basis of the argument in your last post is that the relationships look silly if you pretend that about 90% of archosauromorph taxa didn’t exist. I’m not accepting that as a reasonable argument, and I find it difficult to believe that anyone who understands how phylogenetics works would accept it as a reasonable argument.

  11. Neil, I’ve done the work, discovered the trees that fulfill every expectation. You’re working in the realm of hope and imagination. In the large reptile tree I can prune and no problems emerge. In the Nesbitt tree taxa nest ‘by default’ because they don’t belong there. Taxon inclusion solves this problem.

  12. You haven’t done the work. You’ve done the sort of work that was acceptable in the 90s when phylogenetics was just getting started. Methods have advanced since then. I’ve explained very clearly, with references, why the sort of methods you are using just don’t work, and I’ve shown the sort of alternative methods that could help either confirm your hypothesis or show where you’ve gone wrong.

    I don’t understand why you wouldn’t want to try a new method. That’s what I love about palaeontology; almost every year someone publishes a new method, usually looking at something completely irrelevant to what I do, and I can think how this method originally used to examine adaptive radiations in milkworts can suddenly provide me with new info about evolution of synapsids during Olson’s extinction.

    I seem to remember under every blog post you used to write “Test, test and test again.” Try the test I’ve suggested

  13. I have done the work, Neil. The last common ancestor of the two ‘ornithodires’, Marasuchus and pterosaurs is the basalmost amniote, Gephyrostegus bohemicus. The Ornithodira is a junior synonym for the Amniota which is a junior synonym for the Reptilia. I don’t buy into wrong cladograms in which sister taxa do not resemble one another and massive taxon exclusion is present. Yesterday the data nested pterosaurs with turtles, Huehuecuetzpalli and basal sauropterygians (in the absence of all other lepidosauromorphs) when the pterosaur had every opportunity to nest with Marasuchus and basal dinos. But it didn’t. The data nested pterosaurs with birds when Huehuecuetzpalli was not present.

    I have done the work. I have done the tests.

    • Irrelevant to the issue I’m talking about. I’m trying to persuade you to use a proper definition of “sister taxon” and a modern method for inferring ancestral morphologies.

    • I’m actually speechless. This beggars belief. It really does. I’m not sure how much time I have apparently wasted trying to explain what is wrong with the means you used to get the ancestral states, and explain the rationale behind what you should be doing. And what is your response?

      “I don’t have to infer ancestral morphologies. I have them.”

      I have no words.

      • And now you know the Peters Timesink.

        Our point, David, is that “ornithosuchians” is not the same as “ornithosuchians and kin”. Just because ornithosuchids appear to be the first to branch off that side when a ton of taxa closer to crocodylians are included DOESN’T mean the croc-line ancestor looked like an ornithosuchid. Ornithosuchus had as much time to evolve from that ancestor as Poposaurus, Rauisuchus, Hesperosuchus, etc..

        This is why Neil is speechless. You don’t have ancestral morphologies because no OTU is likely to be an ancestral morphology. But you refuse to learn this and instead think the OTU with the shortest route to the base of the clade (on an evenly spaced cladogram) is closest to the ancestral type. Neil tried to show you the actual methods people use to infer the ancestral morphology, but you don’t care because you think you have the answers already. If you cared, you would see that in Nesbitt’s tree, the ancestral dromaeosuchian (Ornithosuchus + Crocodylus) only differs from the ancestral ornithodiran (pteros + dinos) in less than 20 characters of the 412 he used. But if you just compare Ornithosuchus to his two pterosaurs, there are 47 differences. Because these differences piled up when Ornithosuchus and pterosaurs both started evolving away from that ancestral morphotype.

        So “what are they?” is not Ornithosuchus and pterosaurs, with their 47 differences. They’re not even the ancestral dromaeosuchian and the ancestral ornithodiran with their <20 differences. "The two tines that occur more derived than parasuchians" at their bases are the ancestral pseudosuchian and ancestral avemetatarsalian, which started out with zero differences. THESE are the sister species, just as the clades that each evolved into are the sister taxa. Pseudosuchia and Avemetatarsalia, not Ornithosuchidae and Pterosauria.

  14. Sorry guys. There is such a thing as tree pruning. There is such a thing as taxon exclusion. And I judiciously avoid suprageneric taxa because it gets everyone in trouble. It permits vast clouds of imagination to form, as you’re experiencing now. Pseudosuchia is a junior synonym, so that taxon is invalid. Same with Avemetatarsalia. See earlier posts with these keywords. You’re fighting for a flat Earth and a Newtonian universe. Expand your taxon list.

    • You can’t answer every single criticism of your methods with “but my tree has a wider sample of taxa”. That solve one problem. Taxon inclusion. And everyone admits its a problem. But your matrix and methods have hideous problems that you are ignoring. Every time someone tries to point them out, you fall back on your one line of defence and refuse to budge. You’ve created a system where you can put in the minimum effort, get a very high output, and answer every criticism with one statement whilst never having to learn any theory, anything about evolution, or even more than one method.

      I, and many others have explained, clearly and simply, the problems with your analysis, which have accumulated into the mess you have created. I have explained the problems with your character list, both its inclusivity and the characters themselves. I have explained the problems with your tree search methods. I have explained the problems with your support values and your interpretations of them. I have explained the problems with your attempts to get the ancestor-descendant sequence, both in terms of taxa and morphology. I have explained the problems with your interpretation of other trees.

      And yet you have refuse to look into any of these these issues. You don’t even want to learn about new methods and advances in phylogenetics. You’re stuck in this rut of using 90s methods and an undergrad quality matrix to try and prove actual research wrong, whilst showing a complete lack of knowledge about any of the work you are criticising.

      I’m sorry, a phylogenetic analysis needs considerably more than a large taxon list. Your analysis has so many flaws and problems that it is effectively meaningless.

      • Neil, if my results are logical and all sister taxa resemble one another… if I can prune the tree at will and recover the same topology… if my characters, flawed though they may be, continue to accommodate more and more taxa… if every taxon ancestry provides a gradual accumulation of traits… if my tree is fully resolved… with high support values… Neil, that’s the ideal. There is no reason to discard it. I will continue to correct scores where necessary. Typically when I do they cement relationships. I appreciate you efforts, but this is working pretty well. I have found errors with the Nesbitt tree that, when corrected, bring it into accord with the large reptile tree. Neil, the Nesbitt matrix is not the ‘little angel’ everyone thinks it is. As an analogy, even though the Wright “Flyer” did not have ailerons or wheels and you had to lie on your belly to fly it, it still was the first controllable airplane to fly. This is much the same. No one has ever attempted such a wide gamut study of the reptiles. Whenever someone wants to add or subtract characters, they are free to build on this strong foundation.

      • “if my characters, flawed though they may be, continue to accommodate more and more taxa…”

        They don’t. Remember back when your characters supported a diphyletic origin of snakes? Now that you’ve added more taxa, you get monophyletic snakes. This is what happens when you add more and more taxa to a small character set- phylogeny changes more rapidly. Just where is your 9-29-2011 post on diphyletic snakes, btw? I get a “Page Not Found” error, as if you never advocated diphyletic snakes in the first place. This is more dishonesty like your changing figures in blog posts without notice.

        “I judiciously avoid suprageneric taxa because it gets everyone in trouble”

        Suprageneric taxa can be bad OTUs, but they can still be the correct answer to which taxa are sisters in a given tree. So again, do you agree that the sister taxon to Pterosauria in Nesbitt’s tree is Dinosauromorpha? Not Ornithosuchidae, not Parasuchia, but Dinosauromorpha?

      • This is a reply to MMs comments on March 28 (tech problems prevent a direct reply to those comments). Your hypothesis is not born out by the facts. Snakes continue to be diphyletic when certain taxa are pruned because the burrowers are attracted to burrowing lizard morphology. The earlier blogpost on diphyletic snakes was deleted because it is false based on the addition of taxa. This is not dishonesty. This is scientific progress. Why promote a bad hypothesis? There are kids watching!

        Re: “This is what happens when you add more and more taxa to a set character set.” No. You can prove this by deleting as many other taxa as you like, while retaining the dozen or so taxa in focus, dramatically raising the character/taxon ratio.

        re: Sister taxa in Nesbitt’s tree, if you’ll recall my earlier reply, the two tines in every fork/node are sister taxa. Pruning is always an option. Which generic taxa are sisters to the parasuchians listed?

      • “Your hypothesis is not born out by the facts.”

        It’s mathematically certain- if you add taxa without adding characters, eventually you’ll start getting increasing numbers of polytomies. And before this, the consensus trees will start changing because more similar taxa will have fewer characters distinguishing them, which usually exhibit homoplasy. My impression has been your Big Reptile Tree has had more changes over the past year than before, when you liked to brag adding taxa just rearranged a few poorly supported nodes sometimes.

        “The earlier blogpost on diphyletic snakes was deleted because it is false based on the addition of taxa. This is not dishonesty. This is scientific progress. Why promote a bad hypothesis?”

        That is dishonesty. It would be dishonest to delete your 2000 pterosaur ancestry paper now that you think they aren’t prolacertiforms, just as it would be for Holtz to delete his 1994 and 2000 theropod phylogeny papers now that his ideas on that have changed. Papers and posts are records of our past thoughts, not Wikipedia entries. You have for the latter purpose, just as I have The Theropod Database. Much as with your modified illustrations, you just don’t seem to get that. You never see Naish, or Headden, or Witton, or Carr, or Mallison, or any science blogger I know of delete posts because their content was obsolete. I’ve written posts that ended up being wrong (e.g. my 7-23-2012 post on Bhullar et al.’s archosaur morphometric paper that didn’t understand how their program worked), but they’re left online because I want my mistakes to be transparently public. Your behavior makes it appear you want your mistakes to be forgotten.

        “Which generic taxa are sisters to the parasuchians listed?”

        Archosauria is the sister taxon to Parasuchia in Nesbitt’s tree. There are no genus-level sister taxa to parasuchians in the tree, and you can’t just substitute ornithosuchids or pterosaurs because neither one is necessarily more similar to the basal archosaur than any other archosaur of similar age is.

        You have this weird idea of pruning which is based on your lack of understanding how cladograms work. Taxa merely seem basal because A: there are less included taxa in their clade; and/or B: they are less closely related to the interesting/living group. Ornithosuchus only seems basal in Pseudosuchia because we know of relatively few ornithosuchids. But it’s not less changed from the basal archosaur than other Carnian-Norian archosaurs are. Given living bird phylogeny, your method would place Struthio as the most basal bird, as there are less paleognaths than neognaths and Struthio is first to diverge of living taxa. But no one would say the basal avian/neognath was like an ostrich, just as the basal archosaur wasn’t like Ornithosuchus or pterosaurs. But I’ve explained this to you x+1 times, so I don’t expect you to get it now. You refuse to learn when it comes to methods. Which is why it’s futile to spend much time helping you. But maybe, just maybe, if Neil and I can get you to say the sister of pterosaurs in Nesbitt’s tree is Dinosauromorpha, our typing will not have been in vain.

  15. “If my characters, flawed though they may be, continue to accommodate more and more taxa”

    I have a suspicion on this front that I would like to test. Shoot me your most up-to-date matrix.

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.