Luperosuchus: an erythrosuchid, not a rauisuchian

Luperosuchus fractus (Romer 1972, PULR 04) was considered a indistinct pseudosuchian originally and later a rauisuchian by Desojo and Arcucci (2009). The large reptile tree recovers it as an erythrosuchid and a sister to Shansisuchus, which had an even larger subnarial fenestra. Earlier we looked at the two Shansisuchus specimens, noting that the referred specimen was much larger than the holotype with a distinct morphology, more like Luperosuchus.

Figure 1. Luperosuchus restored based on Romer 1971. Above: original drawing by Romer. Below tracing based on photo in Romer 1971, specimen PULR 04. At right is referred specimen PULR 057. Although related, the referred specimen strikes me as generically different with the low placement of the naris and large postorbital.

Figure 1. Luperosuchus restored based on Romer 1971. Above: original drawing by Romer. Below tracing based on photo in Romer 1971, specimen PULR 04. Extension of the qj and a deeper max gives it more of a erythrosuchid look. At right is referred specimen PULR 057. Although related, the referred specimen strikes me as generically different with the low placement of the naris and large postorbital. Analysis on PULR 057 has not been done.

The reconstruction by Desojo and Arcucci (2009, Fig. 1, above) assumes a short quadratojugal, but a longer qj (Fig. 1, below) matches sister taxa.

This one is probably a rauisuchid
Another much smaller specimen (PULR 057, Fig. 1) was referred to Luperosuchus. That seems doubtful based on the lower placement of the naris, the straighter rostral profile, the larger antorbital fenestra, the deeper pmx/mx notch and the more robust postorbital. These traits appear to lead to Ticinosuchus and the aetosaurs as other archosauriformes retain a high naris. A second possibility leads toward the euparkeriid Osmolskina. A phylogenetic analysis was not attempted due to the small number of traits shown.

Desojo JB and Arcucci AB 2009. New material of Luperosuchus fractus (Archosauria: Crurotarsi) from the Middle Triassic of Argentina: the earliest known South American ‘Rauisuchian’. Journal of Vertebrate Paleontology 29(4): 1311-1315. 
Romer AS 1971. The Chañares (Argentina) Triassic reptile fauna. VIII. A fragmentary skull of a large thecodont, Luperosuchus fractus. Breviora 373:1-8.

Short note: has just passed a million hits for this year. Between 4.2 and 5.7 thousand unique visitors access the site every month.


7 thoughts on “Luperosuchus: an erythrosuchid, not a rauisuchian

  1. If you didn’t know, a redescription of Luperosuchus has been published by Nesbitt & Desojo (2017), which classifies it as a close relative of Prestosuchus and Saurosuchus. Here’s a link: This contrasts with your placement as an erythrosuchid, and I have determined that every one of the 16 characters you have used to justify this placement were incorrectly scored in Luperosuchus. Here are all the miscodings:

    Character 4 should be changed from 1: flat parietal table (like erythrosuchids) to 0: convex parietal table (as in Pres/Saur). The parietal has both ridges and depressions, but at the midline of the skull it is a convex ridge.

    Ch. 9 should be changed from 1: prefrontal-postfrontal contact (like erythrosuchids) to 0: prefrontal-postfrontal separated (as in Pres/Saur). They’re separated by a palpebral which is similar to that of Saurosuchus and fused to the postfrontal and frontal. You can read in the “palpebral” section the full rationale behind their identification of a palpebral.

    Ch. 11 (dorsal nasal shape) should be changed from 0: parallel sides (eryth) to 2: narrows towards naris (Pres/Saur). See the dorsal view in figure 2 of Nesbitt & Desojo (2017)

    Ch. 14 (premaxilla-maxilla notch) and 15 (premaxilla orientation) should each be scored as unknown, since the alveolar margin of the premaxilla is not preserved.

    Ch. 20 (orientation of naris) is a judgement call since the states are continuous, but it should be scored identically to Prestosuchus regardless.

    Ch. 22 should be 2: naris opening laterally (Pres/Saur) rather than 3: naris opening dorsolaterally (eryth). See the dorsal view in figure 2.

    Ch. 24 (width of the postnarial process of the premaxilla) should be 0: narrower than nares (Pres/Saur) rather than 1: not narrower (eryth). Admittedly it’s thicker than that of Prestosuchus, but not to the same extent as in erythrosuchids. I’d say it’s thinner than the widest part of the naris.

    The orbit might be larger than the lower temporal fenestra (Ch. 35:0, as in Shansisuchus), but the quadratojugal is only reconstructed and the shape of the jugal and squamosal suggest that the ltf may be larger than it’s currently reconstructed as. Best to keep it as unknown.

    I’m not sure what the states of ch. 41 (frontal-parietal suture) are saying, but the dorsal view in figure 2 more closely resembles Saurosuchus (scored as 0) rather than erythrosuchids (scored as 1). So Luperosuchus should probably have its score changed to 0, rather than 1.

    Ch. 48 should not be scored as 1: postparietals on occiput plane (eryth). Rather, it should be scored as unknown, since both postparietals and the occipital part of the skull are unknown.

    Ch. 53 is actually miscored for Prestosuchus and Saurosuchus, since their postorbitals tentatively contact the parietal (state 1, as in erythrosuchids and Luperosuchus) rather than having no contact (state 0). You may want to revise that score for multiple “rauisuchians”, I see it in Postosuchus as well.

    Ch. 59 (posterior parietal angle) should be scored as unknown, rather than 1: 20-40% posteriorly (eryth). The posterior portion of the parietal is not preserved. This is also the case with ch. 84, which focuses on how far the postorbital extends compared to the parietal, I believe.

    Another score that needs to be changed to unknown is 69 (squamosal-quadratojugal indentation), which in Shansisuchus was scored as 0: not indented. The quadratojugal area is not preserved, only reconstructed. Same thing with ch. 74, which talks about how much the quadrate is obscured in lateral view. The quadrate is unknown, so it should be unknown, rather than Shansicuhus’s score as 0: squamosal-quadratojugal majority coverage.

    • Neil, thank you for the data and your careful observations.

      Ch4: The parietal table is now scored unknown because the posterior parietal is unknown.
      Ch9: Okay, separated by a palpebral, contra Romer 1971 and the 2017 paper uses a dotted line.
      Ch11: No sister taxa have a pmx dorsal to the naris. Rescored as ‘hourglass’. Shansisuchus and Garjainia narrows toward naris.
      Chs14, 15: agreed, no score
      Ch20: Score remains 3-90º
      Ch22: All erythros scored lateral except Erythrosuchus: dorsolateral.
      Ch24: Luperosuchus narrower
      Ch35: Orbit stays larger. If smaller other erythros have a smaller orbit, too.
      Ch41: Suture not straight in Luperosuchus, Shansisuchus, Garjainia, Erythrosuchus. Note the oval depression in all of these.
      Ch48: agreed, no score
      Ch53: As you note: scores changed for Prestosuchus and Saurosuchus
      Ch59: agreed, no score. Was working from earlier restorations.
      Ch69: agreed for both taxa
      Ch74: agreed, no score

      Ran PAUP: No topology change. Force-moving Luperosuchus to Prestosuchus adds 19 steps. Note the extreme narrowness for such a tall skull, a trait found otherwise only among erythrosuchids.

      • I am practically certain that you are mistaken about the 19 steps. Experimenting in TNT, I found that it did take 19 steps to move Luperosuchus to Prestosuchus before any revisions. However, after your revisions to Luperosuchus and other taxa, as outlined above, it takes only 5 steps. Moving Luperosuchus one node crownward of Prestosuchus only takes 3 steps. This position seems to be the most optimal one among the “rauisuchians”. TNT and PAUP don’t count steps differently, and it would be an odd coincidence that the step difference found by you after revisions and myself before revisions would be identical.

        Going by all of my listed revisions to Luperosuchus (and your revisions to other taxa), it moves to this position crownward of Prestosuchus and takes 4 steps to move back to Shansisuchus. So the devil is in the details regarding the five primary differences between our revisions. Let’s discuss those.

        For character 4, I’ll give you the benefit of the doubt since I wasn’t sure how much of the parietal needed to be preserved to assess this character. So this character is neutral.

        For character 11, I mistakenly said that Luperosuchus should be scored as 2 (narrows over naris) rather than 7 (narrows towards naris, as in Pres/Saur). That being said, scoring Luperosuchus as 9 (hourglass) would be a neutral local autapomorphy anyways. I personally think it should be 7 since the dorsal process of the premaxilla (i.e. the portion medial to the naris) is not preserved well enough to clarify whether there is truly a “premaxilla dorsal to the naris”, at least compared to “rauisuchians”. I also don’t support your decision to score Shansisuchus and Garjainia as 7, since both have a straight nasal (Figure 3b in Wu et al. 2013 for the former and figures 3, 6, and 7 in Ezcurra et al. 2018 for the latter). So I lean towards this being a pro-“rauisuchian” character.

        For character 20, it is unclear how you score the tilt of an incomplete naris. The naris may look like it is angled without the front portion to clarify whether there is a more substantial horizontal component. Plus, the premaxilla has shifted a little relative to the maxilla, moving it back down reduces the slope. Finally, both Postosuchus (Weinbaum, 2011) and smaller specimens of Prestosuchus (like UFRGS-PV-0629-T described by Mastrantonia et al, 2019) have a similar slope. Perhaps you should score Prestosuchus as polymorphic. If you do, then this character is neutral.

        For character 35, it is unclear what you use to measure a “larger orbit”. In every metric (height, width, area) the ltf is unknown. In fact, it might be wider in anteroposterior width once you factor in the newly-found jugal fragment, which reconstructs a jugal posterior process much thinner and longer than that in your reconstruction. I also know that most erythrosuchids have a larger ltf, I was referring specifically to Luperosuchus in relation to Shansisuchus, which may have had a larger orbit (though they seem the same size in the 2013 specimen). We can score this as unknown, therefore neutral.

        I stand by what I said about character 41. Look at how the frontal/parietal suture is trapezoidal in figure 2 of Nesbitt & Desojo (2017), extending a short ways laterally before sloping posterolaterally towards the postorbital. This is similar to Saurosuchus, but different from Erythrosuchus (where it’s just a straight posterolateral slope) and Garjainia (where the suture has abrupt right angles). So Erythrosuchus should still be scored as straight, and the rest of the character distribution is neutral.

        Don’t forget to revise character 84. After you do, it’s unknown and neutral.

        So let’s recap on the character distribution and how it supports or refutes the two hypotheses. Your revision says that ch. 20, 35, and 84 support an erythrosuchid placement by 3 steps, 9 and 24 supporting a “rauisuchian” placement, and the rest being neutral or unknown. My second revision (circa this post) says that ch. 9, 11, and 24 support a “rauisuchian” placement by 3 steps, with the rest (including 20, 35, 41, and 84) being neutral or unknown.

        If you want to bring up a tall, narrow skull, I would be happy to bring up other traits not scored by the LRT. These include the presence of a palpebral (as in Saurosuchus and Postosuchus), a “roman nose” on the nasals (as in Batrachotomus and some Prestosuchus specimens), and an antorbital fossa which extends below the antorbital fenestra (lots of taxa). None of these are known in erythrosuchids. And all of these were coded in the phylogenetic analysis of Nesbitt & Desojo (2017).

      • re: mistaken about the 19 steps: we must be talking apples and oranges now. I would tell you if it was 5 steps, just as I told you about the 19 steps. With the new revisions (#84) and shifting Luperosuchus basal to the Terratosaurus/Postosuchus clade results in 12 steps added.
        re: Nesbitt Desojo 2017: I don’t see any erythrosuchids in their cladogram. Taxon exclusion?
        re: Ch35: the orbit is likely larger in L but it just doesn’t matter.
        re: Ch41: None of these taxa have a straight suture > the na/fr suture. Earlier I was too liberal in scoring this.
        re: Ch84: revised as requested.
        re: characters in general. Experience tells me, convergence is common-to-rampant in the LRT. Experience also tells me that any analyses based on Nesbitt 2011 bring about flawed results due to taxon exclusion and inappropriate inclusion, problems the LRT minimizes.
        If you say certain characters support a rauisuchian placement, you’re “pulling a Larry Martin” which is common in paleo. Better to just let the software figure things out and let the convergences happen.

    • Erythrosuchus is always the representative erythrosuchid in Nesbitt’s analyses, including Nesbitt & Desojo (2017). And I was only bringing up certain characters supporting a “rauisuchian” placement because you brought up one supporting an erythrosuchid placement. I did let your software decide, and it decided that Luperosuchus was a “rauisuchian” one node crownward of Prestosuchus. Luperosuchus shares no characters with erythrosuchids that are not also present in “rauisuchians”, and it shares several (9, 11, 24) which are present in “rauisuchians” but not erythrosuchids.

      Even if you don’t make the revisions to ch. 11, 20 and 35, Moving Luperosuchus from Shansisuchus to the Postosuchus/Teratosaurus clade takes only 2 steps, and moving it to one node crownward of Prestosuchus only takes 1. With 35 as unknown, 20 as polymorphic for Prestosuchus, and Shansisuchus and Garjainia back at 0 for ch. 11, Luperosuchus is one node crownward of Prestosuchus and takes 3 steps to move back to Shansisuchus. That sounds like the software figuring things out to me.

      Something is seriously fishy about the step differences you are getting. 12 is much larger than 2 in a phylogenetic context, and the latter definitely sounds more likely considering that only two LRT characters (20, 35) stand in the way of a “rauisuchian” placement, cut down to 0 after being revised. I’ll send you my TNT files so you can run them yourself.

      So, in summary, the traits supporting an erythrosuchid placement are 20: naris long axis tilted 30-90 degrees (only because Prestosuchus and Postosuchus aren’t rescored), 35: Orbit larger than the LTF (only if you disregard the posterior process of the jugal described in Nesbitt & Desojo 2017), and tall, narrow skull (not scored by the LRT anyways)

      Those supporting a “rauisuchian” placement are 9: prefrontal and postfrontal separated, 11: nasal narrows towards naris (Shansi and Garj have parallel sides), 24: thin posterodorsal process of the premaxilla, and several more not scored by the LRT, such as the presence of a palpebral (a discrete new ossification) and a “roman nose” of the nasal (a distinct ridge which doesn’t need a judgement call to assess).

      Which of these two placements look more likely. Nesbitt & Desojo (2017) use a phylogenetic analysis to support the latter. So does the LRT after the proper revisions. The software fully supports it, so what’s the problem?

      • Apples and oranges, as mentioned before. As I make the changes you suggest, I see errors made when these taxa first entered the LRT, ca. 2011 and I am changing them appropriately. Still not finished on this end. I appreciate your efforts.

Leave a Reply

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

You are commenting using your 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.