Friday, June 1, 2018

Basal theropods in the Lori analysis

In lieu of my original post plan, I just got results from Scott on the most recent analysis of the Lori matrix.  This includes the Archaeopteryx, Ichthyornis and confuciusornithid skull updates unlike the version in the manuscript, plus some added taxa.  The good news is that basic maniraptoromorph topology and Lori's position stayed the same, but you won't be getting any of that today.  Previously, I looked at the non-theropod Lori matrix topology to demonstrate how using maniraptoromorph characters for basal dinosauromorphs will fail even given several hundred accurately scored characters with complete taxonomic coverage.  Well, here's the non-tyrannoraptoran results of the full taxonomic coverage analysis.  It's a majority rule tree, and you can see that fragmentary taxa can fall out in the wrong place due to both a lack of certain characters and the vagaries of tree generation.  So no, I don't think Archaeornithomimus bissektensis' holotype femur is abelisaurian, and the analysis presumably found it to fall out in numerous positions within Avepoda in the most parsimonious trees.  But this gives you a good overall view of what the Lori characters have to say, and it's pretty interesting in some areas.  I would never publish this tree without including characters relevant to carnosaurian, megalosauroid, ceratosaurian, etc. phylogeny, so I don't mind sharing it here.



It's largely good.  The weirdest part is tetanurine coelophysoids.  Also odd is that Sciurumimus and Chuandongocoelurus don't fall out close to their published positions.  Gualicho is a ceratosaur here, so take take from that what you will.  Valdoraptor emerged as a ceratosaur, but the Angeac taxon came out as an ornithomimosaur and Thecocoelurus as a therizinosaur.  None of these taxa are in the Lori manuscript matrix since they had such unstable positions and so little of the Angeac taxon has been published (based on the mounted specimen, it looks very ceratosaurian).  The coelophysoid grade section actually looks pretty good.  Gasosaurus as a piatnitzkysaurid?  Reminds me of PDW.  Xuanhanosaurus and the contemporaneous Kaijiangosaurus sisters?  Synonyms?  I'm rather surprised at the retention of Megalosauroidea, which I usually think is a tenuous grouping.  I don't trust the structure in it, though there is a Spinosauridae (with probably wrongly placed Magnosaurus).  We get a paraphyletic sinraptorid base instead of a big Carnosauria.  Yangchuanosaur Bicentenaria might be worth looking into.  Australovenator and Fukuiraptor are allosaurids/carnosaurs unlike other megaraptorans.  Congrats, Molnar. ;)  Kelmayisaurus as a carcharodontosaur is surprisingly in line with Brusatte et al.'s results given how fragmentary the material is, and Labocania as one is certainly interesting.  Megaraptorans sister to Tyrannoraptora has been my pet hypothesis for a while and shows up again.  Hmm.

It all goes to show you that you can't just take one character list and expect it to give you the right results.  Topology is highly dependent on both character and taxon inclusion, and even if a matrix has more taxa and/or characters than another, you should never trust it to be better unless it includes all the competing data. 

5 comments:

  1. Gasosaurus as a piatnitzkysaurid? Reminds me of PDW.”

    If you don’t mind me asking, do you still think Gasosaurus is a coelurosaur?

    Xuanhanosaurus and the contemporaneous Kaijiangosaurus sisters? Synonyms?”

    Which name would take priority? Both were named in 1984.

    “Yangchuanosaur Bicentenaria might be worth looking into.”

    Something I’ve been wondering since your post on testing Cau’s matrix… could Bicentenaria be a basal megaraptoran?

    Australovenator and Fukuiraptor are allosaurids/carnosaurs unlike other megaraptorans. Congrats, Molnar. ;)”

    Is this a reference to The Dinosauria or something else?

    ReplyDelete
  2. "If you don’t mind me asking, do you still think Gasosaurus is a coelurosaur?"

    Nope.

    "Which name would take priority? Both were named in 1984."

    Kaijiangosaurus, since it was published in June, while Xuanhanosaurus was published in July.

    "Something I’ve been wondering since your post on testing Cau’s matrix… could Bicentenaria be a basal megaraptoran?"

    That takes 5 more steps in Cau's matrix, so seems easily possible.

    "Is this a reference to The Dinosauria or something else?"

    A reference to "Allosaurus" "robustus" astragalus described by Molnar et al. (1981, 1985) as allosaurid, and later identified as a megaraptoran similar to Fukuiraptor and Australovenator.

    ReplyDelete
  3. Interesting stuff. The 'coelophysoid grade section' has a few quirks. Timurlengia braincase as a coelophysid? And here Podokesaurus isn't a coelophysoid (or even a theropod) at all - but has a much, much more basal position.

    ReplyDelete
    Replies
    1. Timurlengia is an example of "fragmentary taxa [that] can fall out in the wrong place due to both a lack of certain characters and the vagaries of tree generation", as TWG matrices through 2012 aren't great at placing braincases. Especially partial braincases like Timurlengia's. Its authors used Brusatte et al.'s (2014) version of the TWG matrix that included 255 of Brusatte's characters designed to sort tyrannosauroids. The Lori matrix uses 58 of those independently but most of Brusatte's only vary within tyrannosaurids, and we only included maniraptoromorphs in the manuscript analysis. But tree generation is also important, since with so many fragmentary taxa that have multiple most parsimonious positions, even 99999 trees are just a tiny fraction of the true number. So what probably happened is that Timurlengia's braincase is Avepoda indet. as far as the TWG characters through 2012 are concerned, it fell out as a coelophysid in the first trees sampled and most of the 99998 permutations were moving other taxa elsewhere to test those OTUs.

      The case of Podokesaurus is probably similar since TWG matrices don't include most proposed avepod, eusaurischian, theropod, saurischian, etc. characters. Plus it's fragmentary with poor photos and illustrations of questionable accuracy, and not described well or in a modern context either. Incidentally, I think Yandangornis is a similar case, albeit more complete. But based on Lori results under different variables, I think it's likely the drawings and description are inaccurate, and the photos are also of poor quality. I hope it gets redescribed soon and doesn't become the Longisquama (or Sharovipteryx) of theropods.

      Delete
  4. Thanks for your response Mickey. Good luck with the Lori manuscript.

    Yandangornis is a sadly neglected theropod.

    ReplyDelete