Baron 2017: 21 ‘unambiguous’ theropod/ornithischian synapomorphies don’t pan out

Yesterday we looked at Baron et al. 2017, who proposed uniting Ornithischia with Theropoda to the exclusion of Sauropodomorpha + Herrerasaurus and kin (Fig. 1), among several other relationships not recovered by the large reptile tree (LRT, 980 taxa). They did so by excluding dinosaur outgroup taxa recovered by the LRT, like Gracilisuchus and Pseudhesperosuchus, while including inappropriate outgroup taxa, like pterosaurs, Lagerpeton and kin, and poposaurs, like Silesaurus. In paleontology this is known as ‘cherry-picking’ and yesterday’s post showed how cherry-picking outgroup taxa, like the pterosaur Dimorphodon, can lead to having scansoriopterygid basal birds recovered as basal dinosaurs. Baron et al. did this by focusing on, and mis-scoring minute traits, not readily visible from an arm’s length of viewing. See below.

By contrast,
the LRT provides a very long list of candidate outgroup taxa going back to Devonian tetrapods and lets the computer decide the topology of the reptile family tree including the Dinosauria. It thereby minimizes a priori bias and subjective or traditional opinion in taxon selection. The LRT also employs more readily observable traits and few to no minutia. The LRT is fully resolved with high Bootstap scores, in contrast to the Baron et al. trees.

Today we’ll dive deeper into Baron et al. 2017
They start with a false premise by supporting the clade ‘Ornithodira‘, which is a junior synonym for Reptilia, since it includes pterosaurs. In the LRT pterosaurs share a last common ancestor with dinosaurs in the Devonian amniote Tulerpeton, the last (and only) known common ancestor of all reptiles.

Baron et al. report, “A formal hypothesis proposing dinosaur monophyly was proposed in 1974, and consolidated in the 1980s. As a direct result of these and other analyses, Ornithischia and Saurischia came to be regarded as monophyletic sister-taxa: this hypothesis of relationships has been universally accepted ever since.” Not in the LRT, which recovered evidence in 2011 to support a clade Phytodinosauria, uniting Sauropodomorpha with Ornithischia + several basal phytodinosaur genera.

Baron et al. report, “No studies on early dinosaur relationships have included an adequate sample of early ornithischians and the majority of studies have also excluded pivotal taxa from other major dinosaur and dinosauromorph (near dinosaur) lineages.” The LRT did so include more than an adequate sample of all pertinent taxa.

Baron et al. report, “In order to examine the possible effects of these biases on our understanding of dinosaur evolution, we carried out a phylogenetic analysis of basal Dinosauria and Dinosauromorpha and compiled, to our knowledge, the largest and most comprehensive dataset of these taxa to date.” No, the LRT is larger and more comprehensive. It is under the authority of the LRT that mistakes can be revealed in the Baron et al. study.

Baron et al. report,Although this study has drawn upon numerous previous studies, no prior assumptions were made about correlated patterns of character evolution or dinosaur interrelationships.” Not true. Their exclusion of appropriate and inclusion of inappropriate taxa demonstrates their assumptions. By this statement they appear to have fooled themselves as well, based on the taxon list of the the LRT.

Baron et al. report, “We analysed a wide range of dinosaurs and dinosauromorphs, including representatives of all known dinosauromorph clades.” Not true. They did not include dinosaur outgroup taxa recovered by the LRT (Fig. 2).

Figure 1. According to Baron et al. 2017 these taxa are related in this fashion.

Figure 1. According to Baron et al. 2017 these taxa are related in this fashion. The LRT does not recover these relationships.

Here is the ‘meat’ of todays post:
Baron et al. report, “The formation of the clade Ornithoscelida [Ornithischia + Theropoda] is strongly supported by 21 unambiguous synapomorphies including: [comments follow]

  1. an anterior premaxillary foramen located on the inside of the narial fossa [present in basal sauropodomorphs Leyesaurus and Pampadromaeus.]
  2. a sharp longitudinal ridge on the lateral surface of the maxilla [present in basal sauropodomorph Pantydraco.]
  3. a jugal that is excluded from the margin of the antorbital fenestra by the lacrimal–maxilla bone contact (this appears convergently in some ‘massospondylids’) [not excluded in Tawa or Coelophysis.]
  4. an anteroventrally oriented quadrate [seemingly all dinosaurs have this sort of quadrate orientation]
  5. short and deep (length of more than twice the dorsoventral height) par occipital processes [apparently a mistake because the figure 2 caption text lists, “elongate par occipital processes.”]
  6. a post-temporal foramen that is entirely enclosed within the par occipital process [I cannot check this minutia with available data]
  7. a supraoccipital that is taller than it is wide [I cannot check this minutia with available data]
  8. a well-developed ventral recess on the parabasisphenoid [I cannot check this minutia with available data]
  9. a surangular foramen positioned posterolaterally on the surangular [I cannot check this minutia with available data]
  10. an entirely posteriorly oriented retroarticularprocess, which lacks any substantial distal upturn [present in basal sauropodomorph Pantydraco.]
  11. at least one dorsosacral vertebra anterior to the primordial pair [I cannot check this with available data]
  12. neural spines of proximal caudals that occupy less than half the length of the neural arches (which are also present in some sauropodomorphs, but absent in Herrerasauridae, Guaibasaurus, and nearly all sauropodomorphs as or more derived than Plateosaurus [it doesn’t matter about derived taxa, we’re looking only at basal taxa, this is a variable trait not present on Scuttelosaurus, but present on Efraasia]
  13.  scapula blade more than three times the distal width (also found in Guaibasaurus) [also found in Herrerasaurus and Sajjuansaurus]
  14. humeral shaft that has an extensively expanded ventral portion of the proximal end, creating a distinct bowing (convergently acquired in plateosaurids and more derived sauropodomorphs) [sounds like a deltopectoral crest, If so, this is universal among Dinosauria]
  15. absence of a medioventral acetabular flange (which was also lost in plateosaurids and more derived sauropodomorphs) [unable to check this minutia with available data]
  16. a straight femur, without a sigmoidal profile (which was also acquired by more derived sauropodomorphs, but absent in basal forms such as Saturnalia and Pampadromaeus, and is also absent in Herrerasauridae) [also absent in Eoraptor, present in Pantydraco]
  17. a well-developed anterior trochanter that is broad and at least partly separated from the shaft of the femur [absent in Eodromaeus and otherwise difficult to check with available data]
  18. a strongly reduced fibular facet on the astragalus [unable to check this minutia with available data]
  19. a transversely compressed calcaneum with reduced posterior projection and medial process [unable to check this minutia with available data]
  20. a first metatarsal that does not reach the ankle joint, but that is instead attached ventrally to the shaft of metatarsal II [not in Tawa, Scelidosaurus or Haya]
  21. fusion of the distal tarsals to the proximal ends of the metatarsals.[not in Tawa, Scelidosaurus or Haya]

Note
several of these ‘traits’ are minutia. The LRT uses larger traits that one can see and measure from a greater viewing distance or with published figures.

According to Baron et al.
other shared features uniting Ornithischia with Theropoda included: [comments again follow]

  1. a diastema between the premaxillary and maxillary tooth rows of at least one tooth crown’s length [not in Eodromaeus, Emausaurus]
  2. an extended contact between the quadratojugal and the squamosal bones [not in a wide variety of ornithischians]
  3. an anterior tympanic recess (convergently acquired in Plateosaurus) [unable to check this minutia with available data]
  4. a fibular crest on the lateral side of the proximal portion of the tibia (described as present in Eoraptor, although we could not confirm its presence, which is also absent in Tawa [unable to check this minutia with available data]
  5. an oblique articular end of the tibia in which the outer malleolus extends further distally than the inner malleolus (although this appears to be absent in Pisanosaurus [unable to check this minutia with available data]
  6. fusion of the sacral neural spines [unable to check this minutia with available data, often hidden by the pelves]
  7. presence of an antitrochanter on the ilium [unable to check this minutia with available data]
  8. reduction of the distal end of the fibula [not in Buriolestes, Tawa, Scelidosaurus]
  9. fusion of the tibia, fibula and proximal tarsals into a tibiotarsus [not in BuriolestesTawaScelidosaurus]
  10. fusion of the metatarsals [not in BuriolestesTawaScelidosaurus]

Apparently Baron et al. were not
thorough enough in these assessments and again depended for the most part, on minute traits rather than large, readily observable ones, Apparently referees were likewise not thorough enough on their vetting of this manuscript. I imagine because it is difficult to do when all the data is not gathered into a single readily reference resource, like RepitleEvolution.com. The present vetting took only a few hours.

According to Baron et al. 
“20 additional steps would be needed to recover Saurischia as previously defined.” But that’s a false goal according to the LRT results that do not recover a clade Saurischia. And with such bad scoring (see above) this goal turns out to be a misstep, not a step.

Baron et al. report,
“in our hypothesis a fully carnivorous feeding strategy is not recovered as the plesiomorphic condition for Dinosauria and we are forced to interpret some of the anatomical similarities between herrerasaurids and theropods as convergences.” In the LRT, herrerasaurids are basal to all remaining dinosaurs, yet have certain autapomorphies that indicate an older, more plesiomorphic last common ancestor of all dinosaurs is awaiting discovery.

Baron et al. report, 
“Dinosauria is recovered in a polytomy with Silesauridae and the enigmatic Late Triassic British taxon Saltopus elginensis.” In the LRT, both of those outgroups are surrounded by other taxa that separate them from Dinosauria.

Figure 1. The origin of dinosaurs to scale. Gray arrows show the direction of evolution. This image includes Decuriasuchus, Turfanosuchus, Gracilisuchus, Lewisuchus, Pseudhesperosuchus, Herrerasaurus, Tawa and Eoraptor.

Figure 2. The origin of dinosaurs to scale according to the LRT.  Gray arrows show the direction of evolution. This image includes Decuriasuchus, Turfanosuchus, Gracilisuchus, Lewisuchus, Pseudhesperosuchus, Herrerasaurus, Tawa and Eoraptor.

Several years ago
the above (Fig. 2) was published online. It remains the best graphic portrayal of basal Dinosauria and their outgroups to date, based on a much larger number of outgroup taxa than has ever been published before. Unfortunately, the Baron et al. team did not take advantage of this readily available and thoroughly verified hypothesis.

References
Baron MG, Norman DB, Barrett PM 2017. A new hypothesis of dinosaur relationships and early dinosaur evolution. Nature  543:501–506.

 

 

Advertisements

7 thoughts on “Baron 2017: 21 ‘unambiguous’ theropod/ornithischian synapomorphies don’t pan out

  1. As I commented on the previous post, I agree that the outgroup sample is very small. On p. 11 of the supplementary information, the authors explain their taxon sample, and it reads as if they thought the default is to include no taxa at all, so that every exception to this rule needs a special justification…

    However, when they replaced one set of outgroups (Euparkeria, Postosuchus) by another (just Dimorphodon), they got practically the same ingroup topology. Evidently, the outgroup sample has just about no effect on this particular dataset. These kinds of surprises happen all the time, in case you’re wondering.

    and poposaurs, like Silesaurus

    As I commented on the previous post, there’s no way Silesaurus is a poposaur. Even the skull is completely different. Don’t even get me started on the ankle.

    Baron et al. did this by focusing on, and mis-scoring minute traits, not readily visible from an arm’s length of viewing.

    That’s a… thoroughly bizarre thing to say, considering how many specimens they studied firsthand, and how many you have ever seen in your life. If they did misscore any of the many traits you list that you can’t find in the literature, how would you even know?

    The LRT also employs more readily observable traits and few to no minutia.

    You keep acting as if that were some kind of feature.

    It’s a bug.

    No, the LRT is larger and more comprehensive.

    It has a larger taxon sample. Conversely, its character sample is pretty small, and it’s a disgusting tangle of redundancies (as I explained at length in a comment to an earlier post of yours). Baron et al. used 457 characters, and their long explanation for why they ordered some of them (pp. 4–9 of the supp. inf.) shows that they put plenty of thought into them.

    present in basal sauropodomorphs Leyesaurus and Pampadromaeus

    Yeah. Now look where they are in the tree. Are you aware that “unambiguous synapomorphy” is a technical term with a meaning that isn’t immediately obvious?

    This holds for several other characters, too.

    Note
    several of these ‘traits’ are minutia. The LRT uses larger traits that one can see and measure from a greater viewing distance or with published figures.

    So? Do you believe “larger traits” are more reliable, or something???

    According to Baron et al.
    other shared features uniting Ornithischia with Theropoda included:

    Those are ambiguous synapomorphies: whether they are optimized as synapomorphies or not depends on ACCTRAN vs. DELTRAN. Do you know what I’m talking about?

    …apparently not, because it makes your comments on those characters moot.

    and again depended for the most part, on minute traits rather than large, readily observable ones

    Do you really believe every character that carries phylogenetic signal must be visible in a published line drawing?

    The present vetting took only a few hours.

    Yeah, because you simply skipped everything you can’t see in the illustrations you have.

    That doesn’t make it go away, you know.

    According to Baron et al.
    “20 additional steps would be needed to recover Saurischia as previously defined.” But that’s a false goal according to the LRT results that do not recover a clade Saurischia. And with such bad scoring (see above) this goal turns out to be a misstep, not a step.

    …That’s not a goal. It’s a test!

    It’s a test of how much better supported their topology is than the traditional Saurischia-Ornithischia dichotomy. Answer: it’s 20 steps more parsimonious.

    (That’s pretty impressive, BTW.)

    And with such bad scoring

    What bad scoring? You didn’t even look at the scoring.

    The matrix is right there on pp. 59–70 of the supp. inf.. Copy it, paste it into a text editor, make a TNT or a NEXUS file of it, and then you’ll be able to compare the scoring to the character list (which, in case you haven’t seen it, makes up most of the supp. inf.).

    Figure 2. The origin of dinosaurs to scale according to the LRT. Gray arrows show the direction of evolution.

    Why do you keep refusing to learn how to read your own trees?

    The direction of evolution in a phylogenetic tree isn’t from leaf to leaf. It’s from the root to the leaves, all of them.

    thoroughly verified hypothesis

    I know words, I have the best words, bl-lieve me.

  2. Oh, unrelatedly: one minutium (never used), two minutia. And “paroccipital” is one word; when two vowels collide (para- -occipital), it’s common for the first to drop out in Latin and Greek.

  3. You say here, “It is under the authority of the LRT that mistakes can be revealed in the Baron et al. study.”

    What happened to “just trying to learn about paleontology”?

      • David, if “all you do” is learn about paleontology, my advice is to stop claiming that the LRT has any sort of “authority” and stop complaining that published papers don’t acknowledge the LRT or recover the same results. Expecting the professional establishment to accept the work of a self-proclaimed stumbling learner is, to be generous, unrealistic. Listen to the (very) few remaining working scientists still willing to engage with you, and learn.

      • Thanks, Squid, you make good points that I have made before.

        No one has to acknowledge the LRT as it stands. It covers too much ground for focused work. But if others want to consider which taxa to include in their own works, and which to exclude, the LRT is a good authority for that. In other words, when you’re talking about dinosaurs, do not include pterosaurs. When you’re talking about synapsids, do not include caseasaurs, etc. etc.

        We’re all helping each other to avoid stumbling.

  4. Ooh, check this out. Only 404 of the 457 characters are parsimony-informative, the number of steps needed to recover Theropoda and Sauropodomorpha as sister-groups is slightly smaller than reported, and there are lots of issues with individual characters, some of which actually put you in good company, Dave. We’re living in interesting times!

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