Thursday, July 11, 2019

Phylogeny of the Lori analysis 1 - Philosophy

Now that Lori is published, as Hesperornithoides, I can finally discuss the topology I recovered for its phylogenetic analysis.  But first, here's the tedious part where I describe why I did what I did.  I mean, I could have just added Lori to Brusatte et al. (2014) and Reviewer 1 would have been happy.  But no, published Theropod Working Group (TwiG) analyses are flawed, and anyone using them is going to get a flawed result.  It's a great framework to start with though, and by standing on the shoulders of giants I've published the best version to date.  For non-avian maniraptoromorphs.

My first lesson was that this analysis turned out to be as difficult to run as it was to create.  Due to the huge number of taxa (501 Operational Taxonomic Units, or OTUs) and characters (700), and the large number of polymorphic scorings caused both by variation within an OTU and partial uncertainty of incomplete remains, you can't just load up TNT on a New Technology search and get the Most Parsimonious Trees (MPTs).  Note many analyses ignore polymorphies and just score the taxon a certain state or unknown, and that Cau in particular structured his character list to be bistate only, with additional characters scoring for each additional state.  I think the large amount of homoplasy might also contribute, since one of my takeaways was that certain areas of the tree are much less well resolved than common knowledge or consensus would have one believe.  In any case, about 70 hours on my new beast of a PC will get you within 10 steps of the MPTs, then you have to run further analyses to find shorter trees, and even then I found trees a step or two shorter using constraint analyses.  I wouldn't be surprised if there are trees a step or so shorter than what I found, but this is a problem we'll need to face as analyses become ever larger.  The fact none of the peer reviewers mentioned how unwieldy the analysis is makes me think none of them tried to run it, which is a flaw in reviewing I've called out before.  Like when Baron and Barrett added Chilesaurus to their Ornithoscelida analysis but included the wrong matrix, so that Mueller et al. found it fell out as a sauropodomorph, then Baron and Barrett's "corrected" matrix deleted all the dashes denoting inapplicable characters so that it couldn't possibly be run or checked for accuracy..  *sigh*  In any case, I included a saved shortest tree in the TNT file so even those with slow computers can access the MPTs.

Geneology of Theropod Working Group analyses, with indented references being expanded from the preceding reference.  All Mesozoic maniraptoromorphs from these analyses have been used here, and each bolded reference has had its character list completely utilized.  Number of characters informative within Maniraptoromorpha yet to be analyzed in parentheses.  After Hartman et al. (2019).

I described my phylogenetic philosophy in detail in the paper, but the basic gist is to take the TWiG analysis, add all usefully scorable taxa, and redefine the characters to reflect the modern sensibilities of Jenner (2004) and Sereno (2007).  Thus, I aimed to eliminate correlated characters, composite characters, eliminate "the use of “absent” as a state in a transformational character (Sereno, 2007:582-584), character traits constructed with discontinuous quantified states (so that certain ranges of values aren’t covered by any state) and those that include a state merely scoring for any condition except those specified by the other states (Jenner, 2004:301-302). We have begun the process of resolving these issues by quantifying 163 characters, isolating 240 composite states into single variables (often using the other variables to form new characters), and excluding 36 correlated characters (see Excluded Characters in the Supplementary Information). Our character list includes details of how each character has been changed from previously published versions. When possible, newly quantified character states have been formulated to best match the taxon distribution for each originally subjective character. All characters have been rewritten in the logical structure advocated by Sereno (2007) to reduce ambiguity and variability between analyses."

The first draft limited the taxa to maniraptoromorphs, but we expanded to all neotheropods in order to test Rauhut's idea Lori could be Coelurus or Ornitholestes.  Of course the TWiG character list was not intended to sort non-coelurosaurs, so that part of the tree is rather Peters-ian in recovering some standard clades but also getting some odd results.  It's possible taxa around the compsognathid grade would organize differently given a comprehensive carnosaur and/or tyrannosauroid character list, but our concern was paravian phylogeny and that's separated by several nodes.  The same philosophy occurs in the crown clade Aves, where I didn't recover Galloanseres or the topology of recent detailed avian analyses because TWiG wasn't designed for that.  This is a concept that's very rarely articulated in published analyses- barring extraordinary circumstances, there are parts of your tree that didn't sample a comprehensive range of suggested characters for that group.  I tried to communicate this in my discussion, that certain groups like therizinosaurs, non-ornithothoracine paravians and non-avian fake-ornithuromorphs have extremely well covered character evidence, while others like ornithomimosaurs, caenagnathoids and enantiornithines should not necessarily be trusted more than recent studies concentrating on those clades.

Holotype of Dalianraptor cuhe (D2139), a probable composite of jeholornithid skull, confuciusornithiform manus and enantiornithine feet.  Due to its chimaerical nature, it was excluded from the Lori analysis (after Gao and Liu, 2005).

So what taxa did I include?  Basically everything.  I "scored almost every named Mesozoic maniraptoromorph known from more than single elements or teeth (the seven exceptions are noted under Excluded Taxa in the Supplementary Information), as well as twenty-eight unnamed specimens. Five recent examples of Aves were included, the palaeognath Struthio and the neognaths Chauna, Anas, Meleagris and Columba. The Tertiary Lithornis and Qinornis were also included as both have been suggested to be outside Aves by some authors, as were Palaeotis, Anatalavis, Presbyornis, Sylviornis, Gallinuloides, Paraortygoides and Foro as basal representatives of modern clades."  "All named Mesozoic maniraptoromorphs described through 2018 and known from more than teeth or single elements were included with few exceptions.  Testing indicated Valdoraptor, Unquillosaurus, Canadaga and Gallornis each had spurious positions due to their fragmentary remains and the current character sample, although the addition of new characters could change this in future iterations.  "Ornithomimus" minutus is only known from a paragraph of text due to loss of the holotype and absence of illustration, so that possible hyperarctometatarsaly (219:2/3) is the only scorable character.  ... Finally, the chimaerical Bagaraatan, Beipiaognathus and Dalianraptor were not included pending detailed reanalysis of their types."

And what characters did I include?  "Characters are designed to incorporate all of those previously used in matrices using the Theropod Working Group (TWiG) as their base through 2012 with the exception of Senter (2011) a baraminology paper which was recognized too late in the coding cycle to be fully incorporated.  Functionally, this led to all proposed TWiG maniraptoromorph characters through mid 2018 being used except 20 from Senter (2011), 102 from Brusatte et al. (2014) and 23 from eight other published analysis (see Fig. S1)."  83% of relevant TWiG characters isn't too bad, especially as I rechecked every one using the latest data and reformatted the characters as noted above.  As I stated, "Ten characters are parsimony-uninformative among our 389 maniraptoromorphs (excluding the possibly tyrannosauroid megaraptorans, coelurids and proceratosaurids in this and the following taxon totals, to ensure similar content). These are retained pending future expansions of the analysis, leaving 690 parsimony-informative characters among our taxon sample of maniraptoromorphs. This makes it the second largest character sample and the largest taxonomic sample in a TWiG analysis of maniraptoromorphs to date, compared to other recent iterations of each TWiG lineage- Gianechini et al. (2018) (700 parsimony-informative characters for their 135 maniraptoromorph OTUs), Foth and Rauhut (2017) (534 such characters and 120 such OTUs), Brusatte et al. (2014) (666 such characters and 127 such OTUs), Agnolin and Novas (2013) (405 such characters and 80 such OTUs) and Senter et al. (2012) (367 such characters and 98 such OTUs)."

Skull of Sciurumimus albersdoerferi (BMMS BK 11).  This juvenile got a lot of N scores for ontogenetically variable characters, plus an entire section in the supplemnentary information.  After Rauhut et al. (2012).

Finally, "Several characters are known to vary ontogenetically among Mesozoic theropods.  These are noted under their descriptions and have been scored with 'N' in the NEXUS file if only young specimens can be coded.  This prevents juveniles from being analyzed as adults and indicates the OTU was not merely left uncoded by accident."  Thus taxa were scored rather conservatively, with e.g. Archaeopteryx scored N for pygostyle presence.  If you think any morphology reflects the adult condition, you can change the N score and test it.

Next up, the basic topology I recovered...

References- Jenner, 2004. The scientific status of metazoan cladistics: Why current research practice must change. Zoologica Scripta. 33, 293-310. DOI: 10.1111/j.0300-3256.2004.00153.x

Gao and Liu, 2005. A new avian taxon from Lower Cretaceous Jiufotang Formation of western Liaoning. Global Geology. 24(4), 313-316.

Sereno, 2007. Logical basis for morphological characters in phylogenetics. Cladistics. 23, 565-587. DOI: 10.1111/j.1096-0031.2007.00161.x

Senter, P. 2011. Using creation science to demonstrate evolution 2: Morphological continuity within Dinosauria. Journal of Evolutionary Biology, 24, 2197-2216. DOI: 10.1111/j.1420-9101.2011.02349.x

Rauhut, Foth, Tischlinger and Norell, 2012. Exceptionally preserved juvenile megalosauroid theropod dinosaur with filamentous integument from the Late Jurassic of Germany. Proceedings of the National Academy of Sciences. 109(29), 11746-11751. DOI: 10.1073/pnas.1203238109

Senter, Kirkland, DeBlieux, Madsen and Toth, 2012. New dromaeosaurids (Dinosauria: Theropoda) from the Lower Cretaceous of Utah, and the evolution of the dromaeosaurid tail. PLoS ONE. 7(5), e36790. DOI: 10.1371/journal.pone.0036790

Agnolin and Novas, 2013. Avian ancestors: A review of the phylogenetic relationships of the theropods Unenlagiidae, Microraptoria, Anchiornis and Scansoriopterygidae. Springer Netherlands. 96 pp. DOI: 10.1007/978-94-007-5637-3_1

Brusatte, S. L., Lloyd, G. T., Wang, S. C., & Norell, M. A. 2014. Gradual assembly of avian body plan culminated in rapid rates of evolution across the dinosaur-bird transition. Current Biology, 24, 2386-2392. DOI: 10.1016/j.cub.2014.08.034

Foth and Rauhut, 2017. Re-evaluation of the Haarlem Archaeopteryx and the radiation of maniraptoran theropod dinosaurs. BMC Evolutionary Biology. 17:236. DOI: 10.1186/s12862-017-1076-y

Gianechini, Makovicky, Apesteguía and Cerda, 2018. Postcranial skeletal anatomy of the holotype and referred specimens of Buitreraptor gonzalezorum Makovicky, Apesteguía and Agnolín 2005 (Theropoda, Dromaeosauridae), from the Late Cretaceous of Patagonia. PeerJ. 6:e4558. DOI: 10.7717/peerj.4558

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. The fact none of the peer reviewers mentioned how unwieldy the analysis is makes me think none of them tried to run it, which is a flaw in reviewing I've called out before.

    It is a flaw, and it is also normal. Without even trying to find out who the reviewers were, I guarantee none of them ran the analysis. I didn't when I reviewed this paper for Nature, even though I did read the character list and asked a lot of clarifying questions, which is already highly unusual. I didn't have a day of computation time at my disposal.

    "Gene[a]logy of Theropod Working Group analyses, with indented references being expanded from the preceding reference."

    Fucking scary.

    "Number of characters informative within Maniraptoromorpha yet to be analyzed in parentheses."

    102 in Brusatte et al. (2014)? Why so many?

    1. To be fair, simply running the analysis is the easiest part, so of course reviewers tend to trust authors on it, even though this trust has not always been deserved. (There's a paper from 2012 where the authors didn't find any MPTs at all, as it turns out, and found only a very small subset of the trees of the length they reported, which is one step longer than the MPTs. Reportedly they didn't know what the addition-sequence replications in a heuristic search are for, and did only one.) The character list and the actual scores are the non-trivial part.

  2. fake-ornithuromorphs

    Unfortunately, you do call that Ornithuromorpha in the paper despite finding Vorona highly nested in Enantiornithes.

    Euornithes Sereno, 1998, forever (not to be confused with the long-forgotten homonyms by Stejneger, 1885, Cope, 1889, Dementjev, 1940, or even Sanz & Buscalioni, 1992)... though, yes, it should be emended to using Enantiornis instead of Sinornis.

    1. Not true. I say "Bellulornis is the sister taxon to Ornithuromorpha" and define it in the supp info as "Ornithuromorpha - (Patagopteryx deferrariisi + Passer domesticus) (modified from Chiappe, 2002)".

    2. Ah. I still haven't read the supp. inf.. Interesting that you went for the 2002 and not the 2001 version, though.

    3. Because I find the 2001 version's inclusion of Vorona is harmful since its placement is controversial. In the Lori analysis it takes 6 more steps to move from Enantiornithes to something closer to Aves, which would make Ornithuromorpha a junior synonym of Ornithothoraces.

  3. How much should I read into Confuciusornis jianchangensis falling outside of Confuciusornithidae? Wang et al. 2018 synonymized it with C. sanctus - as you know, because you discussed it last year on this very blog. :-)

    1. I scored jianchangensis based on all available data, including Wang et al.. As a lumper, read a lot into it I suppose...

  4. I think Patagopteryx is a poor choice to anchor the clade Ornithuromorpha (despite Chiappe,2002). It's secondarily flightless, and so has lost many informative flight-related characters. It tends to jump around in phylogenies. Archaeorhynchus or Hongshanornis would be better choices, IMHO.

    Euornithes could be revived as a more inclusive clade (Passer domesticus <-- Enantiornis leali), such that the respective contents of Euornithes and Enantiornithes are mutually exclusive.

    1. While on the topic of definitions...
      Jeholornithidae - (Jeholornis prima + Jeholornis palmapenis + Jeholornis curvipes)
      This could lead to the possibility that genus Jeholornis is more inclusive than family Jeholornithidae (if a species of Jeholornis is described that is basal to the these three species.

      Am I correct to assume there is no name for the clade that includes Sapeornis, Jeholornis, Confuciusornis and crown birds (Aves) to the exclusion of Scansoriopterygidae? (Maybe call it 'Ornithes'?)

    2. Poor or not, that's the history we have. I agree I don't think a clade name in the songlingornithid/homngshanornithid range is that useful. And yes the definition of Jeholornithidae is bad, but I think J. curvipes is actually Jixiangornis so it works in the Lori topology. And you're correct, there's no name for that clade. Martynuik's Ornithes is based on Archaeopteryx, so would be Paraves in my topology.

    3. Euornithes could be revived as a more inclusive clade (Passer domesticus <-- Enantiornis leali), such that the respective contents of Euornithes and Enantiornithes are mutually exclusive.

      That was Sereno's (1998) explicit intent, except he used his pet Sinornis instead of Enantiornis as usual.