Monday, August 19, 2019

Alvarezsaurs in the Lori matrix

This time our topology is-

I provide a new definition for Alvarezsauroidea that adds Therizinosaurus as an external specifier since I find it most parsimonious for Therizinosauria to be its sister group, and uses Alvarezsaurus as the internal specifier unlike Sereno's that uses Shuvuuia- (Alvarezsaurus calvoi < - Ornithomimus velox, Therizinosaurus cheloniformis, Passer domesticus) .  Alvarezsauroids have had a controversial phylogenetic placement, with the Lori matrix recovering them as basal maniraptorans sister to therizinosaurs.  Yet they can be outside therizinosaurs plus pennaraptorans in 3 steps, become arctometatarsalians in 4 steps (can bring therizinosaurs or not), non-maniraptoriforms in 6 steps (they bring therizinosaurs), closer to pennaraptorans than therizinosaurs in 6 steps, paravians in 11 steps (therizinosaurs move with), closer to Compsognathus than to birds in 15 steps, closer to birds than deinonychosaurs in 27 steps, and closer to Archaeopteryx and other birds than to dromaeosaurids and troodontids in 30 steps.

Fukuivenator is an odd taxon, recovered here as the basalmost alvarezsauroid.  But it can be a therizinosaurian in only two steps, and outside Maniraptoriformes in 4 steps (it emerges in Coeluridae).  One thing I don't think it is is a dromaeosaurid, as that takes 27 more steps, and getting it into Paraves or Pennaraptora requires 11 and 7 steps respectively.  Still, I wouldn't be surprised to see this taxon work its way around the base of Maniraptoriformes once an osteology comes out.

Shuvuuia deserti IGM 100/975 axial elements in ventral view and pelvis in dorsal view (courtesy AMNH).
Nqwebasaurus was recently redescribed by Sereno (2017), which I incorporated into its scorings.  Choiniere et al. (2012) recovered it in Ornithomimosauria, but note most of the characters they list to support that are also said to be present in alvarezsauroids.  Even they could place it in Alvarezsauroidea with only 4 steps.  The Lori matrix needs 6 steps to place it in Ornithomimosauria, which I think is higher partially due to it finding Pelecanimimus to be an alvarezsauroid too.  So similarities between the two like their teeth being in a common groove and maxillary teeth being confined to the anterior third of the bone are no longer ornithomimosaur-like.  As recently noted by Cerroni et al. (2019), this makes more sense biogeographically as well.  Oh, and note that the Lori matrix found Afromimus to be a ceratosaur as in that paper.  In any case, Nqwebasaurus takes 10 steps to move to Compsognathidae, and 7 steps to move sister to Pennaraptora.

As for Pelecanimimus itself, it seems plausibly alvarezsauroid if you think about it.  The skull is famously similar to Shuvuuia, the posterior tympanic recess is in the otic recess, ossified sterna are otherwise unknown for ornithomimosaurs, the long manual digit I was always out of place compared to Harpymimus, and Europe makes more sense for otherwise Gondwanan clades in the Cretaceous.  Now if only someone would release Perez-Moreno's thesis describing it in detail...

Shuvuuia deserti IGM 100/975 pectoral and forelimb elements. Note the tiny phalanx from digit II or III at the bottom (courtesy AMNH).
Patagonykus and Bonapartenykus are usually closer to parvicursorines than Alvarezsaurus and Achillesaurus, but the Lori matrix found them just outside Alvarezsauridae instead.  Interestingly, Xu et al. (2018) recovered the same results.  It takes 3 steps to move Patagonykus closer to parvicursorines, and 4 steps to join Alvarezsaurus and Patagonykus to the exclusion of parvicursorines as in Alifanov and Barsbold (2009).  Xu et al. recover these in 5 and 7 steps respectively, and the most recent version of Longrich and Currie's alvarezsaurid matrix (Lu et al., 2018) recovers a basal Patagonykus and a basal Parvicursorinae in 3 steps each.

One odd result is that the newly described Xiyunykus and Bannykus fall in Patagonykinae too.  Yet only 2 steps move them outside the Patagonykus plus Parvicursorinae clade, where they form a clade.  Another step breaks that up to place Xiyunykus more basal as in Xu et al..  Them being basal certainly fits better stratigraphically, and Xu et al. use several characters designed for alvarezsauroids that the Lori matrix didn't include yet.  Hopefully full osteologies will be published as well.

Mononykus olecranus cast YPM 56693 (of holotype) pes in plantar view (courtesy of Senter).
A patagonykine Achillesaurus as suggested by Agnolin et al. (2012) takes 7 additional steps in the Lori matrix where it instead emerges just closer to parvicursorines than Alvarezsaurus.  On the other hand, only a single step joins it with Alvarezsaurus as in Longrich and Currie (2009) and only 2 steps makes it just further from parvicursorines than Alvarezsaurus as in Xu et al. (2018).

Alnashetri is known from type hindlimb material, but now also from MPCA 377, a nearly complete specimen with interesting characters like flat and unfused sternal plates.  Makovicky et al. (2016) used this data to recover it as the sister group to Alvarezsauridae, and while the few published details left it more derived in the Lori tree, it can go to a more basal position with only two steps.  It should be interesting to compare to e.g. Bannykus once it is published.

Mononykus olecranus cast YPM 56693 (of holotype) (courtesy of Senter).
The arctometatarsal clade has a unique topology, but no other analysis has included nearly as many characters or all of these taxa, with Lu et al. omitting Albinykus and Ceratonykus among non-fragmentary specimens, and Xu et al. omitting the more recently described Qiupanykus.  Enforcing the Lori topology in Lu et al.'s matrix is only 5 steps longer, and doing so in Xu et al.'s matrix is only 6 steps longer.  On the other hand, Xu et al.'s topology is so unresolved at this level, the only difference in mine is placing the Albinykus plus Xixianykus clade basally near Albertonykus, which takes 5 steps to do in the Lori matrix.

It should be noted that Lu et al.'s illustrated topology (their Figure 3) is not their matrix's real result, as they did not fully analyze tree space.  Instead of 20 trees, there are 214 trees.  These differ in that Albertonykus, YPM 1049 and undescribed 41HIII-0104 can fall out anywhere more derived than Patagonykus, and that Parvicursor, the Tugriken Shireh taxon, Shuvuuia and Mononykus form an unresolved polytomy.  This leaves Linhenykus, Qiupanykus and Xixianykus unresolved between that polytomy and Patagonykus, which is perfectly compatible with the Lori topology.  This may also show that the small alvarezsauroid-specific matrix of Longrich and Currie is insufficient given all the new taxa described since 2009.  YPM 1049 was far too fragmentary to include (distal metatarsal III) but I tried testing undescribed Quipa specimen 41HIII-0104.  Didn't make it into the publication, but here's its scorings-

'41HIII0104'                        ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ???????-?? ?????????1 ?10??????? ?????????? ?????????? ???0?????? ????1????? ???1?????{01} ?????????? ?????????? ?????????? ????????0? ????????3? ?????????? ?????????? ?????????? ?????????? ?????????1 ?????????1 1????????? 1????????? ?????????? ?????????? ?????????? ???1?????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ????????0? ?????????? ?????????? ?????????? ?????????? ?????????? ???1?????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????????? ?????{123}???? ?????????? ?????????? ?????????? ?????????? ?????????? ??1???0???

Tugriken Shireh parvicursorine (IGM 100/99) vertebrae and ilia in ventral view, forelimb and fibula in lower right (courtesy AMNH).
Interestingly, Agnolin et al., Xu et al. and the Lori analysis all recovered Albinykus sister to Xixianykus outside Parvicursorinae.  Wonder if that's a real signal?  Unfortunately, the only attempt to name this clade was Agnolin et al. who also recovered Ceratonykus in there and called it Ceratonykini.  Xu et al. place Ceratonykus closer to parvicursorines, while I found it more basal than either, sister to Qiupanykus which neither of the other studies used.  Forcing Ceratonykus sister to Albonykus plus Xixianykus takes 3 more steps in the Lori matrix.  Forcing Ceratonykus sister to Mononykus as in its original description (with or without Qiupanykus) takes 5 more steps.  As stated in the paper, we were the first analysis to include Hateg tibiotarsi Bradycneme and Heptasteornis.  While the former can fall into many positions in Maniraptora, the latter was resolved as an alvarezsaurid as proposed by Naish and Dyke (2004).  Note this used only the tibiotarsus and not alvarezsaurid-like distal femur FGGUB R.1957.  A single step moves Heptasteornis to Troodontidae.

We also provide an updated definition for Parvicursorinae (Mononykus olecranus + Parvicursor remotus), like Choiniere et al.'s (2010) but using species.   One accident of our definitional and discovery history is that all these newer arctometatarsal alvarezsaurids (Xixianykus, Albertonykus, Albinykus, Linhenykus, Qiupanykus, Ceratonykus, etc.) emerge outside the originally discovered and defined Parvicursorinae.  We could really use some clade defining taxa closer to Mononykus than Patagonykus, Alvarezsaurus or Achillesaurus.  In any case, I got a lot of experience with parvicursorine specimens, examining Shuvuuia and the Tugriken Shireh specimen IGM 100/99 in person, and having photos of high quality casts of Mononykus thanks to Senter.  I found the Tugriken Shireh taxon closer to Shuvuuia, but moving it closer to Parvicursor as in Longrich and Currie is just 1 step longer.

Tugriken Shireh parvicursorine (IGM 100/99) vertebrae and ilia in dorsal view, forelimb and fibula in lower right (courtesy AMNH).
Next time, therizinosaurs...

References-  Naish and Dyke, 2004. Heptasteornis was no ornithomimid, troodontid, dromaeosaurid or owl: The first alvarezsaurid (Dinosauria: Theropoda) from Europe. Neus Jahrbuch für Geologie und Paläontologie. 7, 385-401.

Alifanov and Barsbold, 2009. Ceratonykus oculatus gen. et sp. nov., a new dinosaur (?Theropoda, Alvarezsauria) from the Late Cretaceous of Mongolia. Paleontological Journal. 43(1), 94-106.

Longrich and Currie, 2009. Albertonykus borealis, a new alvarezsaur (Dinosauria: Theropoda) from the Early Maastrichtian of Alberta, Canada: Implications for the systematics and ecology of the Alvarezsauridae. Cretaceous Research. 30(1), 239-252.

Choiniere, Xu, Clark, Forster, Guo and Han, 2010. A basal alvarezsauroid theropod from the early Late Jurassic of Xinjiang, China. Science. 327, 571-574. 

Agnolin, Powell, Novas and Kundrat, 2012. New alvarezsaurid (Dinosauria, Theropoda) from uppermost Cretaceous of north-western Patagonia with associated eggs. Cretaceous Research. 35, 33-56.

Makovicky, Apesteguia and Gianechini, 2016. A new, almost complete specimen of Alnashetri cerropoliciensis (Dinosauria: Theropoda) impacts our understanding of alvarezsauroid evolution. XXX Jornadas Argentinas de Paleontologia de Vertebrados. Libro de resumenes, 74.

Sereno, 2017. Early Cretaceous ornithomimosaurs (Dinosauria: Coelurosauria) from Africa. Ameghiniana. 54, 576-616.

Lu, Xu, Chang, Jia, Zhang, Gao, Zhang, Zhang and Ding, 2018. A new alvarezsaurid dinosaur from the Late Cretaceous Qiupa Formation of Luanchuan, Henan Province, central China. China Geology. 1, 28-35.

Xu, Choiniere, Tan, Benson, Clark, Sullivan, Zhao, Han, Ma, He, Wang, Xing and Tan, 2018. Two Early Cretaceous fossils document transitional stages in alvarezsaurian dinosaur evolution. Current Biology. 28, 1-8. DOI: 10.1016/j.cub.2018.07.057

Cerroni, Agnolin, Egli and Novas, 2019. The phylogenetic position of Afromimus tenerensis Sereno, 2017 and its paleobiogeographical implications. Journal of African Earth Sciences. DOI: 10.1016/j.jafrearsci.2019.103572

Hartman, Mortimer, Wahl, Lomax, Lippincott and Lovelace, 2019. A new paravian dinosaur from the Late Jurassic of North America supports a late acquisition of avian flight. PeerJ. 7:e7247. DOI: 10.7717/peerj.7247


  1. In general, with _Nqwebasaurus_ and _Pelecanimimus_ at the base of the Alvarezsauroidea, what does this suggest for the extreme transformation of the forelimb/manus in alvarezsauroids ?

    "As for Pelecanimimus itself, it seems plausibly alvarezsauroid if you think about it. "

    Yep, it makes a lot of sense - especially the _Shuvuuia_-like skull and the ossified sternal plates.

    Is the _Shuvuuia_-like skull of _Pelecimimus_ a product of convergence, given that the two are quite distantly related within Alvarezsauroidea?

    Is there any evidence that _Shuvuuia_ have a gular pouch? Any other alvarezsaur? I suspect the answer is no, but just checking.

    1. I'd say the forelimb transformation didn't start to really take place until Bannykus.

      Regarding skull similarities, Mononykus, Xiyunykus and Bannykus are too fragmentary to tell, and Ceratonykus lacks most of its snout. But the new Alnashetri specimen is said to only lack much of the tail, so should have a skull that can be checked.

      Finally, no other alvarezsauroid preserves skin impressions, but even Pelecanimimus was only said to have "either a throat pouch or dewlap" by Briggs et al. (1997). Honestly given what we've learned about integumentary preservation in the past two decades, I wouldn't be surprised if this and the supposed soft parietal crest were taphonomic spreading of originally tighter tissue.

    2. Briggs arrived at the conclusion that the throat poach or dewlap, and the soft occipital crest, were real, not artifacts.

    3. arrived at the conclusion

      Sure, but that was 22 years ago, so let me repeat: "Honestly[,] given what we've learned about integumentary preservation in the past two decades, I wouldn't be surprised if this and the supposed soft parietal crest were taphonomic spreading of originally tighter tissue."

  2. I'll be interested to see where newbie _Shishugounykus_ shakes out when it's added. It's description found it at the base of Alvarezsauroidea, in trichotomy with _Aorun_ and the rest of the group, with _Haplocheirus_ and _Tugulusaurus_ as more derived members of the clade. But all of those other three are Compsognathids in the Lori matrix, so I wonder if _Shishugounykus_ may be one as well.
    Also, the new topology on the Phylogeny of Theropoda section is different from the published result (i.e. _Xiyunykus_ and _Bannykus_ are more basal than Patagonykinae); is that because you're deferring to Xu et al. ("Xu et al. use several characters designed for alvarezsauroids that the Lori matrix didn't include yet"), or because you tested the relationships with updated information and got a different result (like with Ornithomimosauria)?

    1. Also also…
      Alvarezsauroids "become avemetatarsalians [sic] in 4 steps (can bring therizinosaurs or not)". When you refer to Arctometatarsalia, do you mean simply Ornithomimosauria+Alvarezsauroidea, or a broader group closer to what Holtz originally intended (e.g. including tyrannosauroids)?

      _Fukuivenator_ "can be a therizinosaurian in only two steps, and outside Maniraptoriformes in 4 steps". What about it being the most basal member of Alvarezsauroidea+Therizinosauria, branching off before the two groups split from each other? Or being the basalmost maniraptoran?

    2. Shishugounykus comes out sister to Haplocheirus in Compsognathidae once added.

      And yeah, the Database topology isn't going to be copied directly from the Lori results. The Lori cladogram is just a set of hypotheses based on the data used, so if I know of additional data I'll take that into account.

    3. By Arctometatarsalia I mean taxa closer to Ornithomimus than to birds.

      Constraining Fukuivenator as a basal member of the Alv+Ther clade takes 6 steps, forcing it to be the basalmost maniraptoran takes 10 steps.

    4. The cladistic analysis of the Shishugounykus paper places Compsognathidae as a basal maniraptoran clade (Ornitholestes is recovered as sister to all compsognathid taxa). But it doesn't place Nqwebasaurus and Pelecanimimus in Alvarezsauroidea. Remember, alvarezsaurs were first thought to be flightless birds before being recognized as falling outside Avialae. It's possible that compsognathids could be early alvarezsaurs, because only in this decade have we started understanding the early evolution of alvarezsaurs.

    5. Like Wandering Eyes, I'm also just a long-time enthusiast, not a pro (if you couldn't already tell by the questions I ask). Anyway, I know in the Basal Coelurosaurs post from a couple of weeks ago, you said it took 15 more steps to move Compsognathidae to Alvarezsauroidea, but I'm wondering how you did that. Did you move the whole clade (Compsognathidae) at once, or each of the 7 OTUs one-at-a-time until they were all in Alvarezsauroidea? Would it make a difference either way? that is, does moving the whole clade at once constrain the internal topology of Compsognathidae when moved to Alvarezsauroidea, or would the computer program still have the ability to assess the possibility that Compsognathids are a paraphyletic assemblage of basal Alvarezsauroids?

      Also, as a layman, my initial assumption would be that: usually as an OTU gets further from its most parsimonious position, the number of steps required for the move would increase, but that's obviously not so. Case in point, Fukuivenator. It moves outside of Maniraptoriformes in only 4 steps, but moving to the base of Alv+Ther takes 6 steps, and basalmost Maniraptoran in 10. Why is it that further distance from the most parsimonious position sometimes requires fewer steps, while less distance from the most parsimonious position takes more steps?

    6. Good question! Maybe I'll do a post on how to constrain taxa in the Lori analysis. The short answer is that I gave TNT the minimum amount of rules to follow- that Shuvuuia had to be closer to Compsognathus than either one was to Columba, Segnosaurus or Dromiceiomimus. Then TNT figures out where the other taxa go best, whether that involves a paraphyletic Compsognathidae or whatever. In the present case, Shishugounykus actually decreased the 15 steps to 9 steps. So it helped the hypothesis, even though it's still not that great. The best arrangement TNT found was a clade of Aorun, Scipionyx and Sciurumimus just crownward of Huaxiagnathus. Then a paraphyletic Compsognathidae of- Compsognathus plus Juravenator; Sinosauropteryx, Santanaraptor, Aniksosaurus and Tugulusaurus; Haplocheirus; Nqwebasaurus; Shiushugounykus; Xiunykus; Bannykus; patagonykines and alvarezsaurids. Fukuivenator is sister to Maniraptoriformes, Pelecanimimus is an ornithomimosaur, and ornithomimosaurs are sister to therizinosaurs.

      For your second question, it's because taxa can share characters with multiple groups, but if you place the taxon between those groups, then those characters it shares might not be present at that part of the cladogram. In this case, Fukuivenator apparently shares some characters with coelurids, which is where it goes if excluded from Maniraptoriformes. And we know it shares characters with alvarezsauroids. But if you force it to the base of Maniraptora, then it has to evolve the coelurid characters convergently AND the alvarezsauroid characters convergently.

  3. Hi Mickey, recently started reading you blog and I've been meaning to ask you this. First, I'm not a pro by any means. I'm just your average paleo enthusiast. What I want to know is, what is this Matrix that you're talking about? I've tried looking it up to get a better understanding of it, but google doesn't help since I end up getting info on rock matrix, digital matrix, or the Keanu Reeves Matrix. From what I can gather reading all the blog, it's use to determine an animal's phylogenetic placement, but beyond that I like a little better understanding than just that.

    Any chance you can point me in the right direction, or maybe even a blog post from you explaining it?

    1. Hi there. When you want to use a computer program like PAUP or TNT to figure out phylogeny, part of the text file you feed the program is called the data matrix. This is basically a giant spread sheet with different species listed vertically on the left, then each column represents a different characteristic it might have or not. Then you go through and fill out the spread sheet for each species, with (simplistically) a 0 usually indicating it doesn't have that characteristic and a 1 indicating it does have that characteristic. If you don't know, give it a ?. It gets a bit more complicated than that, but your grid of 0s, 1s and ?s is then your data matrix. It gets tedious making big matrices in Notepad, so usually other programs with fancier interfaces like Nexus Data Editor or Mesquite are used to generate the text file.

      Honestly, the graphical interface of NEXUS Data Editor makes it easy to create a file with a data matrix, whereas most online tutorials or guides are pretty technical for a beginner since they involve the code in the text file itself. Many papers include entire NEXUS files to download which you can then open and change in NEXUS Data Editor, then run in PAUP or TNT once you know how to do that.

      If there's enough interest in how to make and run your own phylogenetic analysis, that's a potential future post.

  4. Oh, so it's a computer program! I knew computers were used in these types of analysis, but I didn't know the matrix literally refers to the data matrix you enter into the program. That clears things up a lot. Thank you.

  5. Recently I happened to come across a conference abstract that suggested that Fukuivenator was a basal ornithomimid instead of an alvarezsaurian, directly referencing your team's paper for the latter result. It is available in this booklet:
    Curious if you may have any thoughts on it (of course, only preliminary data so far).

    1. I'd say that's a nice reinforcement of its general placement, remembering alvarezsaurs themselves can become sister to ornithomimosaurs in only three steps in the Hartman et al. matrix.

      Of their listed ornithomimosaur characters...
      - "maxillary process posteriorly extended separating maxilla from nasal posteriorly to external nares, number of caudal vertebrae between 35-40," were not shown in its original description, and the caudal number isn't known to be different in alvarezsaurs. Of course, I also found Pelecanimimus to be an alvarezsaur, and it has the elongate premaxillary process too, so that would fit in my topology.
      - "coracoid with shallow ventral blade and elongate posteroventral process, metacarpal III approximately equal or greater than 50% compared to anteroposterior diameter of metacarpal II, manuals phalanges with flexor processes on proximal portion of ventral surfaces" are also present in alvarezsaurs.
      - "medial side of metacarpal II not expanded" is untrue (Azuma et al., 2016: fig. 7e).

      So I guess consider me unmoved by their abstract.

  6. An osteology of Fukuivenator was released by Hattori et al. (2021), and they found it to be a basal therizinosaur.

    1. It's a great paper, and as I said that result is only two steps longer in the Lori matrix. What does the totality of new evidence suggest? This, new Nqwebasaurus skull info, new Ornitholestes CT scan info, Pelecanimimus postcrania info, etc.? Upcoming post coming up soon...