Saturday, June 13, 2020

The Unecessary Death of Steneosaurus

Not a dinosaur, but a new paper on the classic crocodylomorph Steneosaurus exemplifies a troubling trend in recent vertebrate taxonomy.  Johnson et al. (2020) reexamine the original material of Steneosaurus, an aquatic croc from the Jurassic of France.  It hadn't been seriously looked at since the 1860s, so this is one of my favorite kinds of paleontology papers- restudying a fragmentary old specimen in a modern light.  What do they find?

We first get a detailed recount of its history, with two decades as Cuvier's "tête à museau plus allongé" (= head with a more elongated snout; I have to praise the authors for translating all the French to English, even in our spoiled era of Google Translate it saves time), before it was named Steneosaurus rostro-major by Geoffroy Saint-Hilaire in 1825.  Eudes Deslongchamps and son tackled it in the 1860s, where they viewed the specimen as too poorly preserved and so "stated that the taxon to represent the genus Steneosaurus should be either ‘Steneosaurus’ megistorhynchus Eudes-Deslongchamps, 1866, or ‘Steneosaurus’ edwardsi Eudes-Deslongchamps, 1868c."  Ha!  You don't get to just take somebody's genus and affix your new species as its type.  They were the last to examine the specimen in detail however, making that a pretty bad note to end on.

Johnson et al. then reexamine the type snout of Steneosaurus, correcting the species name by eliminating the hyphen, officially making it the lectotype, noting Steel had determined the posterior skull to be Metriorhynchus, and illustrating and redescribing the specimen.  Excellent work and very well done.  After eliminating Mycterosuchus nasutus, 'Steneosaurus' leedsi, 'S.' heberti and Lemmysuchus and other machimosaurins based on numerous dissimilar characters, the authors come to the contemporaneous 'Steneosaurus' edwardsi.

"As mentioned before, this was a second species that Eudes-Deslongchamps (1867–69) considered identical to S. rostromajor. These two taxa share a combination of features including:
1. A subcircular, moderately interdigitating premaxilla-maxilla suture.
2. Maxillae ornamented with irregular grooves.
3. A shallower mediolateral compression of the posterior maxillae, as opposed to ‘S.’ heberti (MNHN.F 1890-13).
4. Horizontally flat posterior premaxilla in lateral view.
5. Deep anterior and mid-maxillary reception pits that gradually become shallower towards the posterior maxilla.
6. Subcircular to circular alveoli that remain relatively the same size throughout the maxilla.
7. Teeth with well-pronounced enamel ridges at the base."

Well how cool is that?  They put in the hard work, found the matching more complete specimens, and now we have Steneosaurus edwardsi as a junior synonym of S. rostromajor, giving us a good look at what Steneosaurus really was after two hundred years.

Lectotype of Steneosaurus rostromajor (MNHN.RJN 134c-d) in dorsal (A, B) and ventral (C, D) views. (after Johnson et al., 2020).

But no.

Johnson et al. immediately say "it is important to note that many of these characters may, in fact, be related to sexual dimorphism, ontogeny and intraspecific variation."  True, but that could be said for basically every character supposed to diagnose Mesozoic croc genera, or theropod genera, pterosaur genera, etc..  Unless you have some specific example like 'enamel ridges have been shown to develop with age and both S. rostromajor and S. edwardsi are larger than S? leedsi or S? heberti with weak ridges', then it's just hand-waving.  And no, Johnson et al. never develop such an argument for one of those characters, let alone all seven.

Next, we get "In addition to the sexual dimorphism/ontogeny problem, one of the critical issues about MNHN.RJN 134c-d is that it is poorly preserved."  Sure, but you were still able to perform many comparisons.  Again, the authors never say any of their seven characters are taphonomic, so it's another objection without substance.

Yet the worst rationale for rejecting Steneosaurus is "in reality, the name Steneosaurus is extremely impractical. It was used for many metriorhynchid specimens (e.g. ‘Steneosaurus’ gracilis, ‘Steneosaurus’ palpebrosus and ‘Steneosaurus’ manselii) during much of the 19th century, largely in part due to Cuvier’s metriorhynchid skull region (MNHN. RJN 134a-b) being attributed to the teleosauroid rostral section (MNHN.RJN 134c-d). Indeed, the concise, classical definition of ‘Steneosaurus’ as we interpret it today was not given until the work of both Eudes-Deslongchampses (1868c, 1867–69)"

Substitute Megalosaurus in there to see how ridiculous it is.  That has had over 45 species assigned to it, and was named in the 1820s but didn't have a modern concept associated with it until the 1980s.  When Johnson et al. lament that for Steneosaurus "rather than comparing characters outright, comparison is by process of elimination (or the question of ‘what features does this specimen lack?’)", that perfectly describes the Megalosaurus paralectotype dentary.

"After the Eudes-Deslongchampses’ treatment, what was left was an undiagnostic, chimeric type specimen for S. rostromajor (MNHN.RJN 134) and the genus Steneosaurus was redefined using a new type species that was not accepted by some researchers. In addition, since the Eudes-Deslongchampses, there has been no attempt to rectify this taxonomic nightmare;"

You just showed it was diagnostic, Steel long ago got rid of the chimaeric portion, Eudes-Deslongchamps' stupid attempts to name new type species have no relevance, and you have done the work to finally rectify this taxonomic nightmare.

"Due to these three significant factors (uncertainty of variable characters, poor preservation and
unreasonable name), we have concluded that S. rostromajor, and therefore ‘Steneosaurus’ (MNHN.RJN 134c-d), cannot be confidently assigned to an existing teleosauroid species."

Nope, you just showed it can be assigned to the same species as S. edwardsi.

Actually, I correct myself.  THIS is the worst rationale for rejecting Steneosaurus- "In addition, MNHN.RJN 134c-d was initially diagnosed based on significant orbital and temporal characteristics (from the metriorhynchid MNHN.RJN 134a-b), along with generic rostral ones.  Because the skull material is now known to be from a metriorhynchid, this ‘hybrid type specimen’ factor adds to the doubtful validity of Steneosaurus. According to Article 23.8 of the ICZN Code, ‘a species-group name established for an animal later found to be a hybrid (Art. 17) must not be used as the valid name for either of the parental species (even if it is older than all other available names for them)’ (this also signifies that the species name rostromajor is itself invalid). As such, MNHN.RJN 134c-d serves as an undiagnostic specimen; we, therefore, consider MNHN.RJN 134c-d to be a nomen dubium and, as such, Steneosaurus is treated as an undiagnostic genus."

If the term "parental species" didn't tip you off, Article 23.8 applies to hybrid individuals (those resulting from different species interbreeding), not type specimens chimaerically combined from multiple species.  The Article doesn't even say what Johnson et al. think- it says a name for a hybrid can't be used for either of the species that bred to make it, so that e.g. even if  a mule's scientific name was erected prior to that of horse's or ass's, it can't be the name for horse or ass.  And indeed even the cited Article 17 says that hybrids and chimaeras can be the basis of valid names- "The availability of a name is not affected even if 17.1. it is found that the original description or name-bearing type specimen(s) relates to more than one taxon, or to parts of animals belonging to more than one taxon; or 17.2. it is applied to a taxon known, or later found, to be of hybrid origin..."

If Johnson et al.'s interpretation were right, there goes Gojirasaurus, Protoavis, Chuandongocoelurus, Chilantaisaurus, Fukuiraptor, Coelurus, Alectrosaurus, Dakotaraptor, etc..

Before the big reveal, we have in the Conclusion what can only be described as a lie- "Through character comparison-and-elimination, the only taxon with which MHNH.RJN 134c-d could hypothetically be referred to is ‘S.’ edwardsi, but the two do not share any clear autapomorphic characters or a unique combination of characters."

What are your seven listed characters if not "a unique combination of characters"?  Does any other teleosaurid have them?  If not, they are unique.  In any case, we get the motivation for dumping Steneosaurus twice at the end of the paper-

"We believe that establishing teleosauroid taxonomy from the beginning with a series of ‘clean’ type species/specimens, with every nomenclatural act correctly formulated, is the best course of action, which we will highlight in a forthcoming paper (Johnson, 2019)."

"We believe that establishing teleosauroid taxonomy from the beginning with a series of ‘clean’ type species/specimens, with every nomenclatural act correctly formulated, is the best course of action. This will necessitate a revised teleosauroid taxonomy, in which species previously referred to the genus Steneosaurus are given new generic names. This work will be published by us in a separate contribution, based on the comprehensive teleosauroid phylogenetic analysis in Johnson’s PhD thesis (2019)."

Basically everything I hate about a current trend in vertebrate paleontology- just throw out old specimens and dishonor their authors who correctly reported what was new at the time to come up with your own names.  At least dumping Stegosaurus armatus or designating a neotype for Allosaurus fragilis could be claimed to save time and effort actually analysing the types, if you don't want to do the science to figure out if armatus is actually different from stenops or if fragilis can be distinguished from Saurophaganax.  But Johnson et al. already did all the hard work and found Steneosaurus edwardsi was S. rostromajor, they would just rather use Johnson's new genus name for the taxon.

And their reasons are just grasping at straws.  'Sure we identified these seven charactesrs uniquely shared by Steneosaurus rostromajor and S. edwardsi, but uhh.. could be sexually dimorphic?  Or anything could be individual variation.  Or ontogenetic?  Lots of things turn out to be ontogenetic.  Plus it's broken.  Sooo broken.  Sure we could evaluate characters, but who wants a taxon whose holotype isn't pristine?  Plus a lot of people had stupid ideas about Steneosaurus over the past two hundred years.  What do us scientists do when we have a complicated situation to resolve that was only partially understood historically?  Trash their names and give yourselves credit for new genera.'  Thus Steneosaurus gets the eternal identity of "all evidence points to it being Johnsonosaurus edwardsi, but ehhh... we just sort of ignore it now as Teleosauridae indet. and it's forgotten."

To conclude, Steneosaurus is really outside my wheelhouse.  But if Johnson et al.'s philosophy spreads, we're in danger of losing a lot of historical taxonomy and deserved credit to lazy or selfish authors.  Just look at Microraptor for example, whose holotype of M. zhaoianus lacks a decent skull.  Some decades down the line, what if cranial differences support various Jiufotang species and someone's like 'the postcranial proportions are unique between the M. zhaoianus type and M. hanqingi, but I want a complete type specimen, so Microraptor is an invalid undiagnostic nomen dubium, and instead I propose Mybetterraptorgenus hanqingi and M. gui.'  Just hope they don't pull a Wilson and Upchurch and claim 'Microraptor is invalid and co-ordinate suprageneric Linnean taxa must likewise be abandoned' and replace Microraptorinae with Mybetterraptorgenusinae.

References- Johnson, 2019. The taxonomy, systematics and ecomorphological diversity of Teleosauroidea (Crocodylomorpha, Thalattosuchia), and the evaluation of the genus 'Steneosaurus'. PhD Thesis, University of Edinburgh. 1062 pp.

Johnson, Young and Brusatte, 2020. Emptying the wastebasket: A historical and taxonomic revision of the Jurassic crocodylomorph Steneosaurus. Zoological Journal of the Linnean Society. 189(2), 428-448.

17 comments:

  1. Johnson and colleagues agreed with what I had known for years: that Eudes-Deslongchamps treating ‘Steneosaurus’ megistorhynchus Eudes-Deslongchamps, 1866 as the Steneosaurus type species was erroneous. Some species formerly placed in Steneosaurus have lately been renamed or had available generic names resurrected, including Aeolodon, Sericodon, and Mystriosaurus, and having read Johnson's thesis, you'll note that she resurrects Macrospondylus von Meyer, 1830 for 'Steneosaurus' bollensis (the thesis also coins new genera for some putative Steneosaurus species, including megistorhynchus, but I'm not mentioning them until Chapter 4 of the thesis is published).

    ReplyDelete
  2. I can see both sides of this. On the one hand, I can see what Johnson et al. are aiming at here. They want to essentially 'reboot' teleosaurid taxonomy, to make it more stable and resilient going forward. To that end, Steneosaurus has been sacrificed.

    On the other hand, as you say, I don't think there is sufficient evidence to declare Steneosaurus a nomen dubium. Just synomize S. rostromajor under S. edwardsi, which are united by a unique combination of characters, and the hypodigm is diagnostic. So Steneosaurus is worth preserving.

    I interpret your mention of Wilson and Upchurch as an allusion to the example of Titanosaurus indicus, which was peremptorily discarded as a nomen dubium. Personally, I don't support nomina dubia being used as nominative taxa for coordinated family-level taxa. So if Titanosaurus is a nomen dubium, I don't support continued usage of Titanosauridae. Ditto for other family-level taxa founded on what are now considered nomina dubia (including Ceratopsidae and Troodontidae)... but we can agree to disagree on that. But to back up, I don't believe Titanosaurus should necessarily be treated as a nomen dubium in the first place. T. indicus has been shabbily treated. Regrettably, the same is true for S. rostromajor.

    ReplyDelete
    Replies
    1. Sorry, I meant to write "Just synonymize S. edwardsi under S. rostromajor"

      Delete
    2. It wouldn't have been so bad if Johnson et al. just said that S. rostromajor was probably S. edwardsi but they wanted a more complete specimen for the type, even though that goes against the spirit of the ICZN. But the arguments about character variation and specimen preservation came off as post hoc attempts to justify a decision already made, especially claiming there are no uniquely shared characters when they explicitly wrote such a list on the previous page. And I don't see how no reviewer called them out on that weird interpretation of ICZN Article 23.8. I mean, if it said what they claim there could never be a lectotype for composite specimens because any taxon based on a composite would be invalid.

      Similarly, I can understand the reason for not wanting clades to be named for nomina dubia, even though I disagree with it. What annoys me about Wilson and Upchurch's paper is that it presents this as if it were some ICZN rule that must be followed- "the genus Titanosaurus is invalid and co-ordinate suprageneric Linnean taxa must likewise be abandoned" and "the genus Titanosaurus and its co-ordinated rank-taxa (e.g. Titanosaurinae, Titanosauridae, Titanosauroidea) must be abandoned. The unranked taxon Titanosauria, however, remains valid." Must be, but only for the Linnaean taxa? Who says?

      Delete
    3. Steneosaurus was historically used to encompass not just rostromajor and edwardsi, but also Steneosaurus-like taxa and Machimosaurus-like machimosaurins. For example, bollensis (for which Macrospondylus von Meyer, 1830 is resurrected by Johnson in her 2019 Ph.D thesis) is the putative Steneosaurus species known from the most specimens.

      Johnson and colleagues knew rostromajor wasn't going to be congeneric with any Bathonian teleosauroid historically referred to Steneosaurus.

      Delete
    4. They want to essentially 'reboot' teleosaurid taxonomy, to make it more stable and resilient going forward. To that end, Steneosaurus has been sacrificed.

      It has not, because it can't be; the ICZN simply doesn't allow for such reboots. Specimens cannot cease to be type specimens of available names that compete for priority if they happen to be deemed undiagnostic, or indeed for any other reason other than suppression by the Commission.

      Delete
  3. "Similarly, I can understand the reason for not wanting clades to be named for nomina dubia, even though I disagree with it. "

    I wish that not naming co-ordinated taxa after nomina dubia was an ICZN rule - but as you say, it's not a rule. Though I still think it's bad practice.

    "The unranked taxon Titanosauria, however, remains valid. Must be, but only for the Linnaean taxa? Who says?"

    As you know, family-level and co-ordinated ranks (superfamily, subfamily, tribe, etc) are subject to ICZN rules. So Titanosauridae is subject to ICZN rules, Titanosauria is not. I think this is what Wilson and Upchurch had in mind. Personally, I would prefer that ICZN only deal with genera and species, and leave all suprageneric clades alone.

    ReplyDelete
    Replies
    1. At least having formalized rules for co-ordinated family-level taxa probably helps avoid some additional Archosaurus/Archosauria situations, where the genus is counter-intuitively not a member of the similarly-named suprageneric clade.

      Delete
    2. Yes, the ICZN Code has no say on whether Archosauria includes Archosaurus, or Ornithosuchia includes Ornithosuchus, Titanosauria includes Titanosaurus, and so on.

      You're correct that the ICZN Code stipulates that Archosauridae must include Archosaurus, Ornithosuchidae must include Ornithosuchus, Titanosauridae must include Titanosaurus, and so on. But phylogenetic nomenclature insists that nominative taxa be included as specifiers for clades. So I don't think we need the ICZN to regulate family-level taxa. Phylogenetic nomenclature does just fine without it.

      Delete
    3. "So Titanosauridae is subject to ICZN rules, Titanosauria is not. I think this is what Wilson and Upchurch had in mind."

      See, this is what annoys me. Even you seemed to fall for it right here. Wilson and Upchurch phrase things as if there is an ICZN rule about nomina dubia and family level taxa, but there isn't. The only place the ICZN involves nomina dubia is Article 75.5, specifying a case for establishing a neotype. Search in vain for another Article or even Recommendation that uses 'undiagnostic', 'indeterminate', etc.. Instead the Code is only concerned with nomenclatural validity, i.e. whether a name was established under the proper rules.

      Delete
    4. That's because whether you feel that a name is a nomen dubium is a matter of taxonomy, not of nomenclature, and therefore simply outside the purview of the International Code of Zoological Nomenclature, which grants taxonomic freedom.

      Delete
    5. the ICZN Code stipulates that Archosauridae must include Archosaurus, Ornithosuchidae must include Ornithosuchus, Titanosauridae must include Titanosaurus, and so on

      Only if these are in fact the type genera. Caeciliidae must include Caecilia, because that's the genus it is named for, but it need not include the beetle Caecilius; Cacopinae must include Cacops but need not extend to Cacopus...

      Delete
    6. Mentioning 'Archosauria & Archosaurus' along side 'Titanosaurus & Titanosauria' might be confusing the discussion a little. After all, Archosauria was originally named without reference to Archosaurus (which came into existence only a mere 100 years later!).

      The ICZN only regulates family group names (to be based on a stem of an exsiting genus name). Its why Pachycephalosauridae/Troodontidae got the nod over a long prior-existing name for the same content (Lambe's 1918 Psalisauridae), and why proposed junior synonymns like Sereno's 1986 Tholocephalidae are automatically invalid.

      Regarding Titanosaurus itself, this process makes Titanosauria easy enough to use as a phylogenetic label for a clade (it doesn't rely on Titanosaurus itself, and in its application by pretending that the genus Titanosaurus isn't real, the name Titanosauria can be treated in a way that is similar to other clade names that are not founded on a stem genus - e.g., Gravisauria, Archosauria or Elasmaria).

      But insisting on applying Titanosauridae in modern studies is unbalances things too much towards strict adherence of nomenclature to the detriment actual ever-evolving scientific communication. Let's assume for argument sake that the type of Titanosaurus indicus is undiagnostic among Lameta Formation sauropods, and that Titanosaurus can placed only at the level equivalent of the clades Titanosauriformes or Lithostrotia or Titanosauria...what would be the scientific/communicative value of having a label (Titanosauridae) with the same effect content/scope as any of these other three phylogenetically-definable clade labels? It would be a hamstrung process for the science going forward, if one had to use a group label Titanosauridae (fixed on a specificer Titanosaurus) over Titanosauria simply because the type of Titanosaurus could only be assigned that broadly (rather than finely within Titanosauria). In this scenario, if the type of Titanosaurus is undiagnostic, and therefore Titanosaurus dubious, what exactly is the Group Titanosauridae, and does it serve any purpose?
      Emphasis on 'group' here, because it isn't really a biologically identifiable clade.

      Delete
    7. "in its application by pretending that the genus Titanosaurus isn't real, the name Titanosauria can be treated in a way that is similar to other clade names that are not founded on a stem genus - e.g., Gravisauria, Archosauria or Elasmaria"

      That's simply not its history though, so I would not support that.

      "In this scenario, if the type of Titanosaurus is undiagnostic, and therefore Titanosaurus dubious, what exactly is the Group Titanosauridae, and does it serve any purpose?"

      I think you're viewing things backwards. I'm not saying we need to use a family Titanosauridae, merely that our ability to do so doesn't depend on Titanosaurus being diagnostic. If we can only narrow Titanosaurus' placement down to Lithostrotia, even if it's diagnostic, it's fine to have e.g. Saltasauridae, Nemegtosauridae, and such in Lithostrotia and have Titanosaurus free float. Titanosauridae would only have to come into play if another family clade encompassed Titanosaurus.

      Delete
  4. No, I didn't fall for it. This is Wilson and Upchurch's interpretation. I don't know if it was an honest mistake on their part, or they deliberately misconstrued the Code regarding the validity of family-level taxa.

    I agree with the spirit of what W & U did, but not the means (i.e. citing ICZN rules that don't exist). If Titanosaurus indicus is a nomen dubium (see below), I don't think it's sensible to continue with Titanosauridae. This is because I think it's bad policy to have nomina dubia as nominative taxa for family-level clades.

    BTW, I'm not wholly convinced that T. indicus is a nomen dubium in the first place. I think W & U were a bit trigger-happy in dismissing T. indicus as a nomen dubium. Personally, I think there's sufficient evidence to explore the option of synonymizing Antarctosaurus septentrionalis under T. indicus, and ditching the name Jainosaurus. But that's just my opinion.

    ReplyDelete
    Replies
    1. The reason the ICZN has no such rule is that if a specimen isn't diagnostic at the genus level, it can still be diagnostic at the family level. AFAIK, it is quite clear that Titanosaurus is a titanosaurid, Ceratops is a ceratopsid, and so on. That's enough to keep the diagnoses of the families reproducible.

      Note that ranks are not necessary for this argument.

      Delete
  5. From reading the abstract, I was wondering if this was a huge failure of peer review. Looks like it is.

    The big shocker isn't in the abstract. Seriously, did all of the authors, the reviewers and the editor not know the basic meaning of "hybrid", only the metaphorical one?!?

    Ha! You don't get to just take somebody's genus and affix your new species as its type.

    The modern concept of type species was still only being developed back then.

    "We believe that establishing teleosauroid taxonomy from the beginning with a series of ‘clean’ type species/specimens, with every nomenclatural act correctly formulated, is the best course of action."

    But you can't do that, because the ICZN doesn't let you!

    The only way to really get rid of a name is to get the Commission drunk on its plenary power. That does not happen very often, and requires a lot more than just publishing a paper; and when it happens, the resulting suppressions are often partial or conditional.

    ReplyDelete