Feathers and fangs: What is Hesperornithoides?

Answer:
Hesperornithoides miessleri (Figs. 1, 2; Late Jurassic, Wyoming, USA; Hartman et al. 2019; WYDICE-DML-001 (formerly WDC DML-001)) is the newest fanged anchiornithid theropod dinosaur to be described, compared and nested (Figs. 3, 4).

From the Hartman et al. abstract
“Limb proportions firmly establish Hesperornithoides as occupying a terrestrial, non-volant lifestyle. Our phylogenetic analysis emphasizes extensive taxonomic sampling and robust character construction, recovering the new taxon most parsimoniously as a troodontid close to Daliansaurus, Xixiasaurus, and Sinusonasus.” [see Figure 3, note: Xixiasaurus is not listed in their cladogram].

“All parsimonious results support the hypothesis that each early paravian clade was plesiomorphically flightless, raising the possibility that avian flight originated as late as the Late Jurassic or Early Cretaceous.” [this is an old hypothesis dating back to the discovery of Late Jurassic Archaeopteryx in the 1860s and it remains a well-established paradigm.]

Figure 1. Published reconstruction of Hesperornithes from Hartman et al. 2019, to scale with Caihong, a similar, though smaller, taxon and Sinusonasus, another sister based on very few bones, but look at that canine fang!

Figure 1. Published reconstruction of Hesperornithes from Hartman et al. 2019, to scale with Caihong, a similar, though smaller, taxon preserved with a complete set of bird-like feathers, and Sinusonasus, another sister based on very few bones, but look at that canine fang!

The cladogram by Hartman et al. 2017
(Fig. 3) is similar to one published by Lefevre et al. 2017 in nesting birds (Avialae) as outgroups to the Dromaeosauridae + Troodontidae, the opposite of the large reptile tree (LRT, 1540 taxa, subset Fig. 4).

Today
we’ll compare the Hartman et al. nesting (Fig. 3) to the one recovered by the LRT (Fig. 4).

Figure 2. Tentative restoration of the skull of Hesperornithes alongside to scale skull of Caihong. The maxillae are similar and both have a distinct fang.

Figure 2. Tentative restoration of the skull of Hesperornithes alongside to scale skull of Caihong. The maxillae are similar and both have a distinct fang.

The Hartman et al. cladogram
(Fig. 3) nested Hesperornithoides with Sinusonasus (IVPP V 11527, Xu and Wang 2004; Early Cretacaceous, Fig. 1), as in the LRT (Fig. 4).

The Hartman et al. cladogram included several taxa not previously included in LRT, 1540 taxa, subset Fig. 4), so I added five to the LRT.

  1. Hesperornithoides (Fig. 1) – sister to Sinusonasus in both cladograms
  2. Sinusonasus (Fig. 1) – sister to Hesperornithoides in both cladograms
  3. Daliansaurus (Fig. 5) – nearby outgroup taxon in both cladograms
  4. Alma (Fig. 6) – more distant outgroup taxon in both cladograms
  5. Protarchaeopteryx (Fig. 7) – primitive oviraptorid in both cladograms
Figure 3. Cladogram published by Hartman et al. 2019, colors added to more or less match those in the subset of the LRT (Fig. 4), a distinctly different topology. Here birds and troodontids/anchirornithids are polypheletic.

Figure 3. Cladogram published by Hartman et al. 2019, colors added to more or less match those in the subset of the LRT (Fig. 4), a distinctly different topology. Here birds and troodontids/anchirornithids are polypheletic.

Issues arise in the Hartman et al. cladogram

  1. Birds arise from the proximal outgroup, Oviraptorosauria
  2. Archaeopteryx is not in the lineage of modern and Cretaceous birds
  3. Anchiornithid troodontids are scattered about
  4. Balaur nests with birds
  5. Microraptors and basal tyrannosaurs nest with dromaeosaurids
  6. The outgroup taxon in figure 3 is: Compsognathus; in the SuppData: Dilophosaurus. Neither is a Triassic theropod.
  7. Running the .nex file results in thousands of MPTs (most parsimonious trees), even when pruned down to well-known, largely articulated taxa. Their phylogenetic analysis included 700 characters (and that means hundreds of less-than-complete taxa) tested against 501 taxa. Changing the outgroup taxon to Sinocalliopteryx resulted in far fewer MPTs, but see here for more validated outgroup taxa. Hartman et al. reported, “The analysis resulted in >99999 most parsimonious trees.” Essentially useless… and they knew that attempting to publish their report.
Figure 4. Subset of the LRT focusing on the theropod-bird transition, distinctly different than in Hartman et al. 2019. Here in a fully resolved cladogram, birds and anchiornithids are monophyletic. Taxon inclusion resolves cladistic issues raised by Hartman et al.

Figure 4. Subset of the LRT focusing on the theropod-bird transition, distinctly different than in Hartman et al. 2019. Here in a fully resolved cladogram, birds and anchiornithids are monophyletic. Taxon inclusion resolves cladistic issues raised by Hartman et al.

By contrast,
in the LRT (Fig. 4):

  1. The cladogram is fully resolved (1 MPT).
  2. Birds, including Archaeopteryx and 12 other Solnhofen bird-like taxa arise from anchiornithids, which arise from troodontids (including dromaeosaurids), which arise from Ornitholestes and kin, which arise from the CNJ79 specimen attributed to Compsognathus and kin (including therzinosaurs + oviraptorids), which arises from the holotype Compsognathus and kin (including ornithomimosaurs and tyrannosaurs).
  3. Double killler-clawed Balaur nests with Velociraptor, not with birds.
  4. The outgroup taxa in the LRT include the Triassic dinosaurs, Herrerasaurus, Tawa and a long list going back to Silurian jawless fish.
  5. Hesperornithoides (Fig. 1) and Sinusonasus (Fig. 1) nest with another anchiornithid with fewer teeth and one elongated canine, Caihong (Fig. 1) and a long list of other shared traits. Caihong has a full set of bird-like feathers, so less well-preserved Hesperornithoides likely shared this trait. Caihong nests closer to Archaeopteryx in the Hartman et al. cladogram.
Figure 6. Daliansaurus reconstructed from the original tracing.

Figure 5. Daliansaurus reconstructed from the original tracing. In the Hartman et al. cladogram, this taxon nests close to Hesperornithoides. In the LRT it nests at the base of the Hesperornithes clade.

A few suggestions for Hartman et al. 2019

  1. Build your tree with fewer, but more complete taxa in order to achieve full resolution
  2. Choose a plesiomorphic Triassic theropod or dinosaur outgroup for your outgroup
  3. Practice more precision in your reconstructions. Do not freehand anything. Do not add bones where bones are not known.
  4. Try not to borrow cladograms (like the TWiG dataset) from others, but build your own, especially when the results are so demonstrably poor (>99,999 MPTs)
  5. Include both Compsognathus specimens. They are different from one another and, apparently, key to understanding interrelationships.
  6. Include as many of the 13 Solnhofen birds and pre-birds that you can and show reconstructions so we know you understand the materials. Checking individual scores is like going to Indiana Jones’ government warehouse. Note how the Solnhofen birds split apart and nest at the bases of all the derived bird clades in the LRT (Fig. 4).
FIgure 5. Alma reconstructed and restored (gray).

FIgure 6. Alma reconstructed and restored (gray).

Hartman et al. report, 
“We follow the advice of Jenner (2004) that authors should attempt to include all previously proposed characters and terminal taxa, while explicitly justifying omissions. To this end we have attempted to include every character from all TWiG papers published through 2012, with the goal to continually add characters.”

As their results demonstrate, such efforts are a waste of time.
Pertinent taxa and suitable outgroup taxa were overlooked. The goal is full resolution and understanding. If incomplete taxa and too many characters prevent you from reaching this goal, start pruning, or start digging into the data. There is only one tree topology in Deep Time. Our job is to find it.

Figure 9. Protarchaeopteryx traced in situ, reconstructed a bit and the skull of Incisivosaurus for comparison.

Figure 7. Protarchaeopteryx traced in situ, reconstructed a bit and the skull of Incisivosaurus for comparison. This taxon nests with oviraptorids in both cladograms, basal to Archaeopteryx and birds in Hartman et al. 2019. Not sure if that is all the tail there is, or if more is buried or missing. Probably the latter, according to phylogenetic bracketing.

I sincerely hope this review of Hartman et al. 2019
is helpful. The LRT confirms their nesting of Hesperornithoides with Sinusonasus. Outside of that the two cladograms diverge radically and only one of these two competing cladograms is fully resolved with a gradual accumulation of traits at every node.

The above video tour of the Wyoming Dinosaur Center in Thermopolis
from Wyoming PBS spends a fair amount of time with Hesperornithoides. The conclusions mentioned by the various narrators are not supported by the LRT.


References
Hartman S, Mortimer M, Wahl WR, Lomax DR, Lippincott J and Lovelace DM 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
Lefèvre U, Cau A, Cincotta A,  Hu D-Y, Chinsamy A,Escuillié F and Godefroit P 2017. A new Jurassic theropod from China documents a transitional step in the macrostructure of feathers. The Science of Nature, 104: 74 (advance online publication). doi:10.1007/s00114-017-1496-y
Xu X and Wang X-l 2004. A New Troodontid (Theropoda: Troodontidae) from the Lower Cretaceous Yixian Formation of Western Liaoning, China”. Acta Geologica Sinica 78(1): 22-26.

wiki/Sinusonasus
wiki/Troodontidae
wiki/Hesperornithoides
wiki/Xixiasaurus
wiki/Anchiornthidae
wiki/Origin_of_birds

4 thoughts on “Feathers and fangs: What is Hesperornithoides?

  1. As the Hesperornithoides coauthor responsible for the Hartman et al. phylogenetic analysis, I’ll comment here.

    “note: Xixiasaurus is not listed in their cladogram”

    That’s just a reduced phylogram. The cladogram you want is figure S2 in the supplementary information, which of course has Xixiasaurus.

    “The cladogram by Hartman et al. 2017
    (Fig. 3) is similar to one published by Lefevre et al. 2017 in nesting birds (Avialae) as outgroups to the Dromaeosauridae + Troodontidae, the opposite of the large reptile tree”

    No, your LRT has consensus dromaeosaurids Velociraptor and Balaur sister to troodontids. In any case, as I describe in the text, our tree doesn’t strongly favor this over some other hypotheses, like troodontids being closer to birds. Either option is about the same length.

    Your Figure 2 cranial reconstruction does not match the material. Your premaxilla is just matrix, your broken lacrimal is not the lacrimal (both of which are exposed on the other side of the block), your olive bone overlapping your postorbital bar is just matrix, your braincase are bone fragments that just happen to be arranged in a vaguely basisphenoid-basioccipital shape in that view, etc..

    “Protarchaeopteryx (Fig. 7) – primitive oviraptorid in both cladograms”

    Nope, in our cladogram it is sister to Pennaraptora, though it only takes two steps to become a basal oviraptorosaur, so that’s quite possible.

    “Birds arise from the proximal outgroup, Oviraptorosauria”

    No, that’s you not knowing how to read a cladogram.

    “Archaeopteryx is not in the lineage of modern and Cretaceous birds”

    It can be in trees just one step longer, so that’s still very possible.

    “The outgroup taxon in figure 3 is: Compsognathus; in the SuppData: Dilophosaurus. Neither is a Triassic theropod.”

    Including Triassic theropods, sauropodomorphs, ornithischians, etc. etc. doesn’t change the coelurosaurian part of the tree. We excluded them from the published matrix because we didn’t include characters meant to resolve them, so weren’t testing that part of the tree. Just like you aren’t actually testing theropod phylogeny since you don’t include characters meant for theropods.

    “Changing the outgroup taxon to Sinocalliopteryx resulted in far fewer MPTs”

    I’m skeptical you found any MPTs in your several days with the matrix. It’s not so easy to run.

    ““The analysis resulted in >99999 most parsimonious trees.” Essentially useless… and they knew that attempting to publish their report.”

    No, a high number of MPTs being bad is something you came up with. It’s not something phylogenetecists believe. That you keep harping on any paper that reports many MPTs is just another indication you don’t understand phylogenetics. And *ahem* we succeeded in publishing, we didn’t merely attempt it.

    Ergo, “The cladogram is fully resolved (1 MPT).”

    That doesn’t matter.

    “Birds, including Archaeopteryx and 12 other Solnhofen bird-like taxa arise from anchiornithids, which arise from troodontids (including dromaeosaurids), which arise from Ornitholestes and kin, which arise from the CNJ79 specimen attributed to Compsognathus and kin (including therzinosaurs + oviraptorids), which arises from the holotype Compsognathus and kin (including ornithomimosaurs and tyrannosaurs).”

    No, that’s you not knowing how to read a cladogram.

    “Double killler-clawed Balaur nests with Velociraptor, not with birds.”

    Takes 8 more steps in my matrix, so possible but that likely.

    “Caihong nests closer to Archaeopteryx in the Hartman et al. cladogram.”

    It can nest in Dromaeosauridae with Hesperornithoides with only two more steps, or Hesperornithoides can move to be next to it in Archaeopterygidae in two more steps. So I think it’s quite likely Caihong and Hesperornithoides are closely related.

    to be continued…

    • This reminds me of the Stanford experts in their attempt at building the most accurate flying pterosaur. Rather than starting with a simple ornithopter that could fly, then adding bits and pieces to make it look more accurate, they pushed and rushed their project with hair, eyeballs, unproven mechanics and television coverage, all to show a failed result in spectacular fashion.

      In your own efforts, please consider the option of starting with fewer, better known taxa in order to establish relationships and topologies. Then add the lesser known incomplete taxa as ‘ornaments’ on your tree.

  2. “Build your tree with fewer, but more complete taxa in order to achieve full resolution”

    You can just delete taxa a posteriori until you get full resolution. This still lets the fragmentary taxa help with character distribution. The pruned tree in figure S2 is a lot less than 99999 MPTs for instance, though it’s not 1 because I left a lot of trichotomies in to better illustrate relationships.

    “Do not add bones where bones are not known.”

    You’re the one doing that with Hesperornithoides’ fictional premaxilla, olive green bone, etc..

    “Try not to borrow cladograms (like the TWiG dataset) from others, but build your own”

    This is one of your biggest problems in that you need to include others’ characters if you’re going to test their ideas. It’s why you’re not actually testing anyone else’s ideas on theropod (or mammal, or fish, etc.) phylogeny. Your own advice is hindering you.

    “Include both Compsognathus specimens.”

    It’s on the list of things to do. They differ from each other in 16 characters, so are not that different, but we’ll see what happens.

    “Include as many of the 13 Solnhofen birds and pre-birds that you can and show reconstructions so we know you understand the materials. Checking individual scores is like going to Indiana Jones’ government warehouse. Note how the Solnhofen birds split apart and nest at the bases of all the derived bird clades in the LRT”

    I did this after submission (well, all except the chicken wing and headless Maxburg specimen, which still need to be added). As I replied to your PeerJ question about this, all the Archaeopteryx specimens ended up grouped together within ‘anchiornithines’, while Ostromia and Alcmonavis were separate. So I didn’t even have to try to include characters proposed to join together Archaeopteryx specimens and they still emerged monophyletic. With almost three times your characters and many many more taxa. Sounds like that should disprove your hypothesis. I’ll send you the specimen scores if you want.

    “Pertinent taxa and suitable outgroup taxa were overlooked. ”

    No, I’ve included both pertinent taxa (= Archaeopteryx individual specimens) and every Triassic theropod and basal dinosaur taxon in unpublished versions of the matrix, but they don’t change the maniraptoromorph portion of my results. So nope, they didn’t matter.

    “If incomplete taxa and too many characters prevent you from reaching this goal”

    There’s not such a thing as “too many characters” and incomplete taxa can only help and then get pruned a posteriori for resolution.

    Your Protarchaeopteryx illustration features many imaginary parts, btw. I saw the specimen and have numerous photos. For instance the real tiny furcula is up by your proximal humerus outline, the real tail is more robust and curves up over the ilium, the only portion of the ischium is a triangular fragment missed by you, and of course pedal digit V is just a metatarsal (which is actually preserved in the foot on the left, but not in the position of your drawn digit V). So again your tracing technique doesn’t work and you’re scoring nonexistent animals.

    “I sincerely hope this review of Hartman et al. 2019
    is helpful.”

    Nope. It just shows again you don’t understand phylogenetics, and that your tracing fails to find the actual bones. If I had additional advice for you, it would be to stop looking at a cladogram as a single “new best tree” and rather as a network of more or less probable alternative topologies. So saying my study is bad because the mpt found archaeopterygids in Deinonychosauria is wrong because they switch to Avialae in just one step. That’s not something my data have much of a say about. Whereas Balaur being a bird is actually an appreciable difference, at eight steps.

    But what’s really funny about this post is that for the non-maniraptoromorph theropod portion of my tree (stemward of Ornitholestes), I did exactly what you do but better. It’s more taxon inclusion than you, almost three times as many characters, and scored mostly correctly unlike yours, with well formed characters unlike yours. And I get a crap tree like you do. Like yours it has some good parts, since some signal breaks through. But it also has weird stuff, but different weird stuff from yours. Megalosauroids are ceratosaurs, carnosaurs are paraphyletic, etc.. And under your hypothesis of “full taxon inclusion and 230ish characters will get you the real tree”, this makes no sense. I should have gotten your tree if that’s all you need. But I didn’t, not even close. Because both of us would need to include characters proposed for that part of the tree.

    • Often well considered, Mickey. The bottom line is this: in deep time there is a completely resolved tree of interrelationships. Our job is to find it. If you think a completely resolved tree topology in which all sister taxa document a gradual accumulation of traits is a ‘crap tree’ (to quote your dive out of professional etiquette), as opposed to your reported, 99,999+ trees, then the world has indeed turned topsy turvy.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.