Wednesday, March 1, 2017

Basal Dinosauromorpha in the Lori tree

Hi everyone!  Sorry for being away so long.  As an interesting experiment, after adding every decently preserved theropod to the Lori matrix, I added every decently preserved basal dinosauromorph as well.  As you all know, the Lori matrix is basically every TWG analysis through 2012 combined together with a complete taxon sample.  This yields 700 characters designed to sort out coelurosaurian phylogeny, excluding crown Aves, tyrannosaurids (fixed with Brusatte's later additions), enantiornithines and caenagnathoids.  Only a few non-coelurosaurian theropods have ever been included in TWG analyses- Sinraptor dongi and Allosaurus as outgroups from the beginning, with a few authors adding Coelophysis, Dilophosaurus and Monolophosaurus as well.  It actually performs pretty well at approximating avepod phylogeny, but once we get to the base of Dinosauromorpha, this happens-

Initial majority rule consensus of the non-avepod section of the Lori tree.  Running a traditional TBR search from this shows that Nyasasaurus, Lagosuchus, Teyuwasu and Dromomeron gregorii all have uncertain positions, as does Pisanosaurus within Neornithischia.  Also the pairing of Guaibasaurus and Saturnalia within their basal phytodinosaur clade lowers to 97%.  TNT crashed before I could screen capture the reduced consensus however.

Notice how Peters-ian it is.  Not that it has a similar topology to Peters' Large Reptile Tree, though there are a few similarities (Phytodinosauria, ornithischian Chilesaurus, Buriolestes and Marasuchus as theropods...).  Indeed, our ornithischian topologies are completely different.  No, what I mean is that while there are some consensus clades recovered (in my case Ornithischia, Scelidosaurus+Huayangosaurus, the Agilisaurus+ clade, herrerasaurids, avepods), it's largely a novel tree with newly paraphyletic groups (e.g. Silesauridae) and misplaced taxa (e.g. Diodorus and Dromomeron romeri in Ornithischia; Eucoelophysis and Dromomeron gregorii in a clade of basal sauropodomorphs).  Like Peters' trees, mine actually has a bigger taxon sample than prior basal dinosauromorph analyses.  Superior to Peters' analyses, I use far more characters than prior analyses (700 vs. 228 in Peters, online 2016; 315 in Nesbitt et al., 2009; 256 in Cabreira et al., 2016; 139 in Martinez et al., 2011; 221 in Butler et al., 2008; 255 in Boyd, 2015; ~370 characters in the newest version of Yates' basal sauropodomorph analysis), and these are formed well and the taxa coded accurately.

Yet despite the latter facts, I wouldn't even think of claiming e.g. Tawa and Daemonosaurus aren't dinosaurs, or that sauropodomorphs are diphyletic and partly paraphyletic to ornithischians.  This is because I haven't included the characters supporting the consensus topology in most cases, so I haven't really tested those ideas.  Conversely, it may be that some TWG characters actually do overwealm those used in basal dinosauromorph analyses to support novel relationships.  But we'll never know that until we combine the character sets.

I do find it interesting that Chilesaurus falls out basally by ornithischians, and not all that far away from the Tawa+Daemonosaurus clade I proposed it belonged to in my last post on the taxon.  Certainly it never clades with basal tetanurans, and still emerges as a basal alvarezsaur if it's an avepod.  Of course in the Lori tree, it emerges within a paraphyletic derived Thyreophora which is itself the basalmost group of ornithischians within a paraphyletic Sauropodomorpha.  So the combination of character states there probably doesn't reflect the combination of character states in a traditional basal Ornithischia, where they're sister to Saurischia or various silesaurids.  Ideally without adding the hundreds of characters involves with this part of the tree, I could constrain a traditional topology here, but that's assuming the traditional topology is correct.  Honestly there's no easy way to test such a controversial taxon as Chilesaurus.

Finally, I'd like to mention the excellent new paper by Marjanovic and Laurin (2016), which is basically a published version of what I do on the Evaluating Phylogenetic Analyses portion of the Database, targeting an analysis of basal tetrapods.  It's incredibly detailed and finds "thousands of suboptimal scores" (I'd like to know the percentage changed, but it's not listed AFAIK) and a very different topology from the original.

References- Butler, Upchurch and Norman, 2008. The phylogeny of the ornithischian dinosaurs. Journal of Systematic Palaeontology, 6(1), 1-40.

Nesbitt, Smith, Irmis, Turner, Downs and Norell, 2009. A complete skeleton of a Late Triassic saurischian and the early evolution of dinosaurs. Science. 326, 1530-1533.

Martinez, Sereno, Alcober, Columbi, Renne, Montanez and Currie, 2011. A basal dinosaur from the dawn of the dinosaur era in Southwestern Pangaea. Science. 331, 206-210.

Boyd, 2015. The systematic relationships and biogeographic history of ornithischian dinosaurs. PeerJ. 3:e1523.

Cabreira, Kellner, Dias-da-Silva, da Silva, Bronzati, Marsola, Muller, Bittencourt, Batista, Raugust, Carrilho, Brodt and Langer, 2016. A unique Late Triassic dinosauromorph assemblage reveals dinosaur ancestral anatomy and diet. Current Biology. 26(22), 3090-3095.

Marjanovic and Laurin, 2016. Reevaluation of the largest published morphological data matrix for phylogenetic analysis of Paleozoic limbed vertebrates. PeerJ Preprints. 4:e1596v2.


  1. the excellent new paper

    It's the preprint version of the first revision a manuscript submitted to PeerJ. I'm working on the second revision right now! Other than clarity of presentation, not much more is going to change.

    I haven't counted the changes because I don't trust my ability to count to 20,000. They're color-coded in Appendix 2.

    Evaluating Phylogenetic Analyses

    Oh. That's fascinating. I knew about similar DML posts of yours, of course, but not about that site. Probably I should actually cite it. What year should I cite it as? (Not mentioned even in the source code.)

    Some of these should have been published!

    1. Not 20,000 – "just" 4,200. I found a way to count them, and that's going to be part of the next and hopefully final submission.

  2. I suspect the topology will result closer to the "standard model" once characters relevant for non-coelurosaurs are added.
    In my matrix (current version: 1773 characters vs >460 taxa, including non-dinosaurian dinosauromorphs, basal ornithischians and sauropodomorphs), Chilesaurus results a basal tetanuran.

    PS: for a consensus not crashing the TNT run, just use the agreement subtree option. TNT 1.5 has nice implementations for it.

  3. David- The years vary from 2010-2014. I really should put those on the pages themselves to reflect which year's available data I'm using.

    Andrea- I suspect the same, though I still have a hard time understanding how Chilesaurus ends up as a basal tetanurine. Did you use any of Novas et al.'s matrix codings when coding your Chilesaurus OTU? Because I think my post on it argues pretty well that it's heavily miscoded in all of their matrices. That's why in my tree it's "Chilesaurus_verified"- it's only what I could verify from the pictures and text (and Salgado et al., 2008). How many more steps does it take to move it to Tawa/Daemonosaurus or Ornithischia?

    1. I really should put those on the pages themselves to reflect which year's available data I'm using.

      Please do!

  4. I agree in not trusting all of the scores published. In fact, even my Chilesaurus OTU is scored exclusively according to what is visible in the photos published, or mentioned in the main paper or in the JVP paper by Salgado et al.
    Enforcing Chilesaurus in Ornithischia (note that in my dataset I have just Pisanosaurus, Heterodontosaurus and Tianyulong included): +13 steps.
    Enforcing Chilesaurus in the Tawa-Daemonosaurus clade: +19 steps.
    The basal tetanuran status for Chilesaurus is supported by 12 unambiguous synapomorphies.

  5. Well... You might be right about Ornithischian and Theropods clustering close together and Herrerasaurids being an outgroup...

    1. No doubt in reference to the new Baron et al. (2017) paper, this Lori tree doesn't actually show the topology you state. In the tree above, the clade of ornithischians plus derived sauropodomorphs branches off first, then herrerasaurids (and basal sauropodomorphs among other taxa) are theropods. The Baron et al. tree on the other hand, has herrerasaurids as part of a Sauropodomorpha which branches off first, then ornithischians sister to theropods. I'm writing a blog post on it now...

    2. The comment wasn't really meant to be serious. But I'm really looking forward to your next blogpost.