Saturday, March 8, 2014

Testing pterosaur origins in Ezcurra et al. (2014)

There's a new amniote analysis out from Ezcurra et al. (2014) emphasizing the lepidosauromorph-archosauromorph split.  Thev changed a ton of codings from the earlier version by Reisz et al. (2010), almost doubled the size of the matrix, properly ordered most characters, made sure inapplicable codings stopped redundancy in most cases, AND got rid of suprageneric OTUs.  Among the negative points are no parareptiles or aquatic diapsids (yet many synapsids) and a lack of many of the traditional series-based characters (e.g. number of phalanges on digit X, number of vertebrae in this segment).  Regardless, this seems to be better than most amniote analyses, and the lepidosauromorph-archosauromorph split is the perfect place to test pterosaur origins.

Among relevant taxa, the matrix has Peters' closest consensus lepidosaur to pterosaurs, Huehuecuetzpalli, plus two more squamates.  It has Macrocnemus, placed even closer to pterosaurs by him, plus three more protorosaurs.  Also Euparkeria and Erythrosuchus for the Bennett hypothesis.  I added two pterosaurs from the basalmost clade (Preondactylus and Eudimorphodon), Megalancosaurus, Longisquama and Eoraptor.  Note this is far from perfect, as incomplete basal dinosauromorphs (lagerpetonids, Marasuchus), more basal simiosaurs (Vallesaurus, Hypuronector), and more derived pterosaurs with well preserved braincases, palates and axial details would all be useful to add.  Not to mention Scleromochlus, Sharovipteryx, Cosesaurus and Atanassov's taxa.  But what does this preliminary test find?

The pterosaurs were closest to Megalancosaurus, but this clade was sister to Eoraptor within Archosauriformes.  No pseudosuchians were included, so it's uncertain if pterosaurs are strictly archosaurs.  What about Longisquama?  Either a tanystropheid protorosaur or a member of the megalancosaur-pterosaur clade, though note I did (conservatively) use Peters' (2000) illustration for the anatomy.  This may bias it toward those interpretations, though his interpretations were far less fanciful back then.

Forcing pterosaurs to be protorosaurs is 9 more steps, so rather unlikely.  Forcing them to be lepidosauromorphs is 17 more steps though, so probably wrong.

Forcing Megalancosaurus to be an archosauriform outside Ornithodira is only two steps longer, so about equally likely.   The same number of steps move it to Tanystropheidae, with Longisquama following, but not pterosaurs.

Forcing Longisquama to be a lepidosauromorph only takes two steps, and it becomes the most basal one.  Forcing it to be sister to Coelurosauravus results in both being basal lepidosauromorphs, which only takes 3 more steps.  Having this pairing outside Sauria as in Senter (2004) is 5 more steps, so less likely.  Adding Megalancosaurus to complete Senter's Avicephala is 9 more steps though, so rather unlikely.

Finally, enforcing Peters' versions of Lepidosauromorpha (with captorhinids, caseids, weigeltisaurids, trilophosaurs, rhynchosaurs, simiosaurs, Longisquama, tanystropheids and pterosaurs) and Archosauromorpha (with derived synapsids, Paleothyris, araeoscelids and younginiforms) takes a whopping 75 extra steps.  This is worse than any alternative phylogeny I've ever tested for anything.

Instead of asking what pterosaurs are, maybe we should be asking if simiosaurs belong in Archosauria.  Recall even in Nesbitt's (2011) huge analysis focusing on archosaur interrelationships, they could be placed in Avemetatarsalia with only five extra steps.  At the time I thought a broader analysis would be more useful for that question, but here they flit between Archosauriformes and Protorosauria with even less difficulty.  Longisquama is pretty clearly too poorly described to strongly favor any alternative, but it's interesting that Senter's placement for it comes out as badly as it does.  As for what pterosaurs are, while archosauriform beats out protorosaur and both beat out lepidosauromorph, I've seen taxa recover from a 9 step disadvantage often, and a 17 step disadvantage rarely, so both of the latter alternatives are still possible.  75 steps?  Not so much.

References- Senter, 2004. Phylogeny of Drepanosauridae (Reptilia: Diapsida). Journal of Systematic Palaeontology. 2, 257-268.

Reisz, Laurin and Marjanovic, 2010. Apsisaurus witteri from the Lower Permian of Texas: Yet another small varanopid synapsid, not a diapsid. Journal of Vertebrate Paleontology. 30, 1628-1631.

Ezcurra, Scheyer and Butler, 2014. The origin and early evolution of Sauria: Reassessing the Permian saurian fossil record and the timing of the crocodile-lizard divergence. PLoS ONE. 9(2), e89165.


  1. You've preceded me in doing these tests (although I did not plan to write a post)! :-)

  2. Thanks for testing this, Mickey. I definitely will be playing with their data in the future (-psst- I already am).

  3. Just found this post. Sorry for being tardy.
    Unfortunately for Ezcurra et al., (Fig. 2) taxon exclusion did them in. Their Amniota excludes Tseajaia. With more taxa, Tseajaia nests within the Amniota.

    With taxon exclusion you don’t get the lepidosauromorph/archosaurmorph (lizard/croc) split at the base of the Amniota. So Ezcurra nests caseids with synapsids because millerettids and kin are excluded. That also leads to morph mismatches with Captorhinus nesting between Tseajaia and Paleothyris + Petrolacosaurus.

    They note: "Our revision suggests a minimum fossil calibration date for the crocodile-lizard split of 254.7 Ma.” That’s about 90 million years too late.

    On their tree:
    Not sure how Prolacerta got separated from Protorosaurus.
    Not sure how Youngina got so separated from Euparkeria.
    Not sure how Prolacerta was derived from the Rhynchosauria or how the Rynchosauria was derived from Macrocnemus and Protorosaurus.

    One of the tenets of evolution is that sister taxa look like each other, demonstrating a gradual accumulation of character traits with no untenable gaps inbetween. You’ll find that gradual accumulation at in which rhynchosaurs gradually evolve from sphenodontids with trilophosaurs tossed in.

    Their Fig. 2 raises more red flags simply for mixing genera with suprageneric taxa. Again, many morph mismatches like the rib glider Coelurosauravus with the aquatic Thalattosauria + Ichthyoptergia.

    Really, Mickey, what kind of goggles are you wearing when you don’t see problems here?
    And please use the latest drawings of my work. They reflect the latest thinking.

    Best regards