Thursday, May 17, 2012

Testing alternative tetanurine topologies in Carrano et al. 2012

In my last post, I described Carrano et al.'s superb new tetanurine analysis.  One of my biggest complaints was that very few alternative topologies were tested, so while the cladogram had plenty of novel and interesting ideas, the paper doesn't tell us just how well they're supported.  Note Carrano et al.'s reported "branch support" (Bremer?) is flawed because 1) fragmentary taxa lead to nodes collapsing easily even if the node is well supported for complete taxa in it, 2) distant taxa will often share characters (which show up as homoplasies or autapomorphies in the most parsimonious tree), so that moving a taxon across three nodes will not simply cost the number of Bremer support steps in each of those nodes added together, and 3) rearrangements change character state distributions, which will change support in adjacent nodes.

Luckily, another wonderful thing about Carrano et al.s' paper is that it comes with a fully constructed NEXUS file.  Not only that, the NEXUS includes character descriptions for NDE and even ran without error the first time.  Future authors please follow suit!  So, with this I was able to test alternbative topologies myself.  I must say I was surprised by some of the results.  What follows are the results, arranged in order of extra steps needed, with my subjective interpretation of liklihood...


Basically ambiguous

1 step- (stated in paper) Cryolophosaurus outside Neotheropoda sensu Bakker, as in Smith et al. (2007).  Though this isn't a good test without more basal taxa.

1 step- (stated in paper) Monolophosaurus as a basal megalosauroid, as in Benson (2010).

1 step- Xuanhanosaurus as a piatnitzkysaurid, as in Benson (2010).

1 step- (stated in paper) Poekilopleuron as a piatnitzkysaurid, allosaur or other kind of megalosaurid.

1 step- Eustreptospondylus sister to Magnosaurus.  This was proposed by Rauhut (2003), who made the former a species of that latter.  Magnosaurus moves out of Afrovenatorinae to join Eustreptospondylus easily.

1 step- Duriavenator as an afrovenatorine.

1 step- Monophyletic Sinraptor.  Though the characters traditionally used to combine them are cranial ones compared to Yangchuanosaurus, and their pelves are somewhat different, so how Siamotyrannus relates really isn't obvious.

1 step- (stated in paper) Chilantaisaurus as a carcharodontosaurid sensu stricto, not a neovenatorine.

2 steps- Megalosauroidea as carnosaurs, as Huene originally proposed, and Rauhut (2003) and Cau more recently find.  This rearranges the whole tree so that megalosauroid relationships are less resolved, and Piatnitzkysauridae and Lourinhanosaurus are successively closer to sinraptorids+allosaurs than megalosauroids are.  The fact the backbone of the tree crashes down so easily is disconcerting.

2 steps- Streptospondylus as an afrovenatorine (or really Afrovenator as a streptospondyline).

2 steps- (stated in paper) Lourinhanosaurus as a basal carnosaur (as in Holtz et al., 2004) or sister to Avetheropoda.


Quite likely to be true

3 steps- "Dilophosaurus" sinensis outside Neotheropoda sensu Bakker, as in Smith et al. (2007).  Cryolophosaurus follows.  As with Cryolophosaurus above, we'd really need more basal taxa included to test this.

3 steps- Piatnitzkysaurus outside Orionides, as in Rauhut (2003). Other piatnitzkysaurids follow.

3 steps- Xuanhanosaurus outside Orionides, as in Holtz et al. (2004).  It ends up by Monolophosaurus.

3 steps- Eustreptospondylus as an afrovenatorine (or really Afrovenator as a eustreptospondyline), as in Allain (2002).

3 steps- Afrovenator as a carnosaur, as in Rauhut (2003).  Even though I constrained Afrovenator to be closer to Allosaurus than to Eustreptospondylus and Torvosaurus as in Rauhut's tree, this makes all megalosauroids carnosaurs.  The carnosaur phylogeny is (Strepto(Megalosauridae((Afro+Leshan)(Spino(Piatnitzky(Lourinhano(Sinrap,Allosauria))))))).  So megalosauroids end up being paraphyletic to carnosaurs.

4 steps- Monolophosaurus as a carnosaur, as in Sereno et al. (1996).  This also makes megalosauroids carnosaurs, with the topology (Megalosauria(Mono(Piat(Lour(Sinrap+Allo))))).

4 steps- Poekilopleuron as a sinraptorid, after Benson et al. (2010).

5 steps- Monolophosaurus sister to Avetheropoda, as in Smith et al. (2007).  Piatnitzkysaurids are even closer to avetheropods in this phylogeny.

5 steps-  Torvosaurus and/or Eustreptospondylus being closer to spinosaurids than Afrovenator is, as in Sereno et al. (1994).

5 steps- Siamotyrannus as a coelurosaur, as in Buffetaut et al. (1996).  It's closer to other coelurosaurs than Lourinhanosaurus is.

5 steps- Sinraptorids outside Avetheropoda, as in Paul (1988).  Poekilopleuron becomes a coelurosaur.


Somewhat possible

6 steps- Piatnitzkysaurus as an afrovenatorine, as in Holtz et al. (2004).  Megalosauridae gets shuffled around a lot.

6 steps- Sinraptorids and carcharodontosaurids as sister taxa, exclusive of Allosaurus, as in Allain (2002).  This also makes megalosauroids carnosaurs, with the topology (Megalosauria(Poekilo,Lour(Saurophag(Allo(Sinrap,Carch))))).

7 steps- Eustreptospondylus  closer to spinosaurids than Torvosaurus or Afrovenator are, as in Smith et al. (2007).

7 steps- Chilantaisaurus as a spinosaurid, as in Rauhut (2003).

7 steps- Neovenator as an allosaurid, as in Hutt et al. (2001).  This also makes megalosauroids carnosaurs, with the topology (Megalosauria(Piat(Lour(Sinrap+Allo)))), and neovenatorines become allosaurids.


Unlikely-

9 steps- Cryolophosaurus as a coelophysoid, though again we'd like more basal taxa.

10 steps- "Dilophosaurus" sinensis as a coelophysoid, and same as above.

11 steps- "Szechuanosaurus" zigongensis outside Orionides, as in Rauhut (2003).  It ends up by Monolophosaurus.  Odd that so many more basic rearrangements are more likely, as described above.

11 steps- Siamotyrannus and Fukuiraptor as sister taxa, as in Holtz et al. (2004).  The pair end up sister to Allosauria.

12 steps- Megaraptor as a carcharodontosaurine, as in Smith et al. (2007).

12 steps- Acrocanthosaurus as an allosaurid, as in Stovall and Langston (1950).  This really changes things.  It makes megalosauroids carnosaurs, with the topology (Megalosauria(Piat(Lour(Sinrap(Conca(Carch(Eocarch((Acro+Poek)(Neoven(Allo,Saurophag)))))))))).  Again, surprising.

15 steps- Megaraptor as a megalosauroid, as in Calvo et al. (2004).  Ends up sister to Megalosauria.

15 steps- Megaraptora as coelurosaurs, my pet hypothesis (but also partially Longrich 2001 and Novas et al. 2008).  Oh noes!  Even "worse", it's more like in this topology, "coelurosaurs become neovenatorids".  But as I said in my last post, the analysis needs more coelurosaurs.  Also remember the corrected TWG matrix needed 15 more steps to get a monophyletic Deinonychosauria, and that's standard now.  So any of the above hypotheses may end up being true.  Forcing megaraptorans to be coelurosaurs, but carnosaurs to be monophyletic otherwise needs 21 more steps, btw.  So it's on par with Bullatosauria.  Er.. *cough*


I guess the take home message is tetanurine basic relationships are highly uncertain.

No comments:

Post a Comment