Recently Li et al. (2012) announced the discovery of Diandongosuchus (Fig. 1), described as a “basal poposaurid” close to Qianosuchus. There seemed to be something wrong with this nesting (because poposaurids are all dinosaur herbivores derived from bipeds and Qianosuchus is a sort of piscivorous rauischid close to Ticinosuchus), so I tested those results in phylogenetic analysis. The new test recovered Diandongosuchus between Proterochampsa and the phytosaurs. It’s the basalmost phytosaur! Moving Diandongosuchus to the poposaurids added 51 steps, a substantial number.
Figure 1. Diandongosuchus nests as a basal phytosaur when choristoderes and basal younginoids are included, far from Qianosuchus, which also does not nest with poposaurs. The discovery of the basalmost phtyosaur is very exciting! Diandongosuchus provides clues to the timing of the acquisition of various phytosaurian traits. The small size of Diandongosuchus compared to other phytosaurs is also thought-provoking.
The Problems were Many and Basic
The problem with the Li et al. (2012) study is that they relied on the Nesbitt (2011) study on archosaurs, which included several taxa that do not belong (Vancleavea, pterosaurs and Mesosuchus) and several taxa that are needed but were not included (Youngina, choristoderes). More importantly, the Nesbitt (2012) study did not include taxa and character traits that could recover lineages with gradual accumulations of characters. Rather the Nesbitt (2012) study produced “strange bedfellows,” as discussed earlier over seven posts ending here. Finally the Nesbitt (2012) study was largely unresolved outside of the Diandongosuchus clade, which gave this new taxon many possible sisters and ancestors. Not good. The large reptile tree solves all these problems.
Figure 2. More strange bedfellows created by the Nesbitt (2011) study. None of these taxa belong together! Note the distinctly different pelvis that all save Diandongosuchus share. None have metatarsal 4 longer than mt3. Yet they nest as close relatives in the Li et al. (2012) study. Here Diandongosuchus appears twice, once in scale with the others (the smaller image) and once larger to see the details.
Here is what the Nesbitt (2012) study produces:
A real mixed bag (Fig. 2). In the large reptile tree none of these taxa nest together. While Qianosuchus has a skull that appears piscivorous (a sister, Ticinosuchus includes fish in the belly), like other euarchosauriforms, the skull was taller than wide, unlike the wider, flatter skull of Diandongosuchus and its true kin (Fig. 3),
Figure 3. The real sister taxa and close relatives of Diandongosuchus beginning with the Youngina specimen with the longest, lowest rostrum, BPI 2871, and moving forward toward the parasuchians. Diandongosuchus further separates the new putative Proterochampsa from the original. Note the deeply notched coracoid, a dead giveaway for parasuchians. No other reptiles share this trait.
If you’re going to look for the relatives of Diandongosuchus…
Look for the relatives that look like it. Diandongosuchus has a very crocodilian appearance down to the bony scutes on its back. The BPI 2871 specimen referred to Youngina (Fig. 3) has that look, but lacks an antorbital fenestra and mandibular fenestra. It shares with Diandongosuchus a rostrum longer than the dentary, dorsal nares and a wide, flat triangular (in dorsal view) skull. Like Proterochampsa, in Diandongosuchus the premaxilla was just starting to lengthen, which goes to extremes in parasuchians. Neither of these relatives to Diandongosuchus include post-crania. For those bones we’ll take a closer look at parasuchians.
For the four separate origins of the antorbital fenestra look here.
Parasuchians have long been considered analogs to modern crocodilians with the major difference in the location of the external nares, either back toward the eyes in the parasuchians or at the snout tip in crocs. Like all pararchosauriforms (except Champsosaurus and perhaps Doswellia) the nares in Diandongosuchus are displaced posteriorly, but not quite dorsal in location. The orbit is midway in relative size between BPI 2871 and Parasuchus.
The cervicals are robust in Diandongosuchus, taller than long, as in parasuchians. The leaf-like cervical ribs are closer to those seen in choristoderes, like Champsosaurus, than the narrow ribs of parasuchians. The dorsal ribs are more robust than those of parasuchians, again closer to those in choristoderes. The tail was broad proximally and narrow distally as in choristoderes and parasuchians.
The coracoid of Diandongosuchus is a close match to that of Parasuchus, with its deep anterior notch, a trait rarely–if ever–found elsewhere within the Reptilia The clavicles were quite long in Diandongosuchus, as in choristorderes. They were smaller in parasuchians. The ilium of Diandongosuchus was a simple posterior process, as in choristoderes, bearing no trace of an anterior process as in parasuchians. The shapes of the pubis and ischium are close matches to those found in Parasuchus. Both were oriented largely medially, creating a not-so-deep pelvic area. The sigmoidal femur of Diandongosuchus is a close match to that of Parasuchus. The fibula includes a long, low ridge-like trochanter in both taxa. The tarsus was similar in both taxa. Manus and pes proportions were similar in both taxa to the exclusion of all other candidates.
Figure 4. Chanaresuchus a taxon derived from a sister to Diandongosuchus. Note the similar long-legged proportions and long, low, wide skull shape. In the upper right is Gualosuchus.
While we’re on the subject…
Keep in mind that as relatives to parasuchians, proterochampsids like Gualosuchus and Chanaresuchus are not far from Diandongosuchus, as you’ll see for yourself (Fig. 4). The skulls are strikingly similar with similar limb to torso proportions.
The Li et al. (2012) study nested parasuchians (phytosaurs) basal to their “Archosauria” with so many unresolved derived branches (including pterosaurs) that really anything goes here, including Diandongosuchus and Qianosuchus as two possibilities among many (at least nine taxa).
The images above (Figs. 2, 3) should prove to be good guides. Diandongosuchus, in all regards, belongs with the taxa in figure 3 more parsimoniously than those in figure 2. The details await anyone caring to see the data matrix of the large reptile tree or to duplicate the taxon list in their own study.
As always, I encourage readers to see specimens, make observations and come to your own conclusions. Test. Test. And test again.
Evidence and support in the form of nexus, pdf and jpeg files will be sent to all who request additional data.
Reference
Li C, Wu X-C, Zhao L-J, Sato T and Wang LT 2012. A new archosaur (Diapsida, Archosauriformes) from the marine Triassic of China, Journal of Vertebrate Paleontology, 32:5, 1064-1081.
Nesbitt SJ 2011. The early evolution of archosaurs: relationships and the origin of major clades. Bulletin of the American Museum of Natural History 352: 292 pp.
The Pterosaur Heresies 
If the shape of the pelvis excludes it from the Poposauridae why should it be included with phytosaurs? The pelvis also doesn’t look like a phytosaur’s pelvis. Are there specific characters that lead you to this idea? I saw: “The ilium of Diandongosuchus was a simple posterior process, as in choristoderes, bearing no trace of an anterior process as in parasuchians. The shapes of the pubis and ischium are close matches to those found in Parasuchus. Both were oriented largely medially, creating a not-so-deep pelvic area. The sigmoidal femur of Diandongosuchus is a close match to that of Parasuchus”
But the illia of phytosaurs also have an anterior projection (as you illustrated above). The pubes of Parasuchus is almost vertical whereas the pubes of Diandongosuchus points anteriorly with an almost planar ventral surface (with a slight concavity near the pubic/illiac contact). You don’t illustrate the pelvis in enough detail to compare the medial deflection of the the pubes and illia, but given the other differences noted in these two bones I don’t see how you can consider them visually similar.
Furthermore the sigmoidal femur – yes they both posses a sigmoidal femur. This isn’t a terribly uncommon trait and it has popped up in many different clades. The actual structure of the femur outside of just a lateral view of its shape should bear close inspection. Compare the relatively modest femoral head (relative to shaft size) of Diandongosuchus to Parasuchus’ massive femoral head. The tibial and fibular articular surfaces on the femur of Parasuchus are greatly reduced compared to Diadongosuchus. Except bearing an S-curve there isn’t much in common with their femora. The femur of Chanaresuchus is likewise dissimilar – being comparatively more gracile despite being used for terrestrial locomotion.
This is not looking at any codings in a matrix, this is simply looking at the morphology that you’ve posted and discussed. I am not seeing what you are seeing.
Forgot to mention – of course poposaurids have different pelvic structures. They’re pretty much all bipedal. You need that larger illium to provide an anchor for your only locomotor muscles at that point.
My point exactly. Thank you for making it.
Don’t see how that makes your point. The pelvises are different. You say they look similar. My point is that you shouldn’t expect morphological similarity, even between organisms that are closely related, if their lifestyles/habitats/niches are radically different. Look at humans and gorillas. The pelvis of a gorilla is pretty different than ours. This isn’t because we aren’t related – it is because we do different things.
Agree there are differences. But it’s always degrees of differences and similarities. Would you dispute that give 100 mammalian or tetrapod pelves to compare phylogenetically that the higher primates would not form a clade? I think they would. That’s how great paleontologiests can tell by just a tooth or toe (sometimes) what the rest of the animal might look like, if similar specimens are known in their entirety. It’s the new ones that throw us a loop sometimes. But a large analysis, and mine is getting larger every week, reduces the chance for error by providing a wider gamut of opportunities and more possible nesting sites for novel forms than smaller studies. That’s why I’m here.
Simple. Diandongosuchus is not a phytosaur per se, but its closest known ancestral form. We’re close, not exact. This is evolution, one morphology morphs into another over time. The term “most parsimonious” cannot be forgotten. What I described in the blog is shorthand for what was recovered in the matrix and by visual interpretation. Do the the images of the dissimilar taxa in figure 2 give you more ‘faith’ in the Nesbitt process?
The illia of Diandongosuchus and Arizonasaurus certainly appear the most alike. Anyway – it is primitive for the clade AND a specialized aquatic predator. I wouldn’t expect it to look exactly like a much larger bipedal or quadrupedal organism. Lots of functional things have to be in place in order to be an adequate aquatic predator that do not increase fitness in a terrestrial carnivore.
Rob, never pick just one trait to hang your analysis on. There is just too much homoplasy within the Reptilia for that. A complete analysis with all traits and all taxa included must be the only way to approximate the model echoing actual evolutionary events. In Arizonsaurus BTW, the ilium has an anterior process and a downward oriented acetabulum, which makes it quite a bit different than phytosaurs and Diandongosuchus. Diandongosuchus MIGHT demonstrates that the anterior process in the phytosaur ilium might have been one of the last processes to develop prior to the origin of phytosaurs proper. I was surprised by its possible absence, actually since Chanaresuchus has such an ilium. The other thing to consider is: what is hiding beneath that femur over the ilium? Was there a process erupting there? Perhaps so? Either was it doesn’t matter in the analysis because ancestors have it and descendants don’t, but my money would be on some sort of process appearing there.
“…never pick just one trait to hang your analysis on. There is just too much homoplasy within the Reptilia for that.”
You are the one that mentioned the illium (along with other characters). I am just pointing out that some of your statements don’t agree with the specimens as they exist (and as you have illustrated). Look back at my first post on this topic to see that I haven’t just talked about various illia – they are just one of the characters that you describe/attribute similarity to wrongly.
“…if similar specimens are known in their entirety.”
Most fossil species are not known in their entirety – this is a straw man argument. Your data is not as complete as you are implying. See the whole Poposaurus skull discussion we are simultaneously having.
” In Arizonsaurus BTW, the ilium has an anterior process and a downward oriented acetabulum.”
Yes it does have a very small anterior process. Not nearly as large as in phytosaurs or in the bipedal poposaurs, but I agree it is there.
“Not sure how much can be gleaned from the drawing on p. 108 of Gauthier et al. I took nothing from it to the matrix. They reconstructed the skull like Allosaurus, which is also unlikely. Evidence does change things, but with the rest of the data weighing in otherwise, at this point I have to follow what the tree recovers.”
What we can say is that a premaxilla with teeth is known and that this premaxilla and teeth do not resemble Daemonosaurus. Simply compare the illustration to the premax of Daemonosaurus and see that is the case. I agree – don’t code from that drawing, but in terms of overall shape/size/orientation it gives us SOME information.
Maybe email Gauthier (or another author) and see if they have pics they could send?
Also: “because poposaurids are all dinosaur herbivores.”
Your reconstruction of Poposaurus as a herbivore seems to be in opposition to what little we know of the skull. There is nothing (NOTHING) in the preserved skeletal remains that I am aware of to suggest your Daemonosaurus-like skull ever sat on the cervicals of Poposaurus.
Send me your data. All I know is how they nest at present. My data agrees with yours. No postcrania for one. No crania for the other. Hence your statement is correct, but allows for possibilities. Let me know what you have and send it. We’ll see what happens.
Gauthier et al.’s paper on the GSENM Poposaurus makes brief mention of the skull parts but that is all the discussion I have found. They reconstruct the skull as that of a carnivorous animal based on the teeth and premaxilla and maxilla fragments. None are procumbant as in Daemonosaurus, but almost all the teeth are highlighted (but not figured in detailed) indicating that they are known. This should suggest that the two are not synonymous.
Not sure how much can be gleaned from the drawing on p. 108 of Gauthier et al. I took nothing from it to the matrix. They reconstructed the skull like Allosaurus, which is also unlikely. Evidence does change things, but with the rest of the data weighing in otherwise, at this point I have to follow what the tree recovers.
Seriously, Rob. Are you defending the nesting of Diandongsuchus with the bipedel poposaurs + Qianosuchus? (Given the differences there and the similarities with Youngina + Proterochampsids + Parasuchians? I can’t make it any clearer with words or pictures.
The bipedal forms didn’t just come from nowhere. And not all the poposaurs were bipedal (see Arizonasaurus again).
Generally though, I am not defending much – I am pointing out problems I see with your alternative. I am not saying “Option A is stronger because X, Y, and Z.” I am saying “Option B is weaker because X, Y, and Z.” That, I suppose, could be interpreted as de facto support for the Nesbitt position but that isn’t really what I am going after here.
Besides, it doesn’t matter what I believe to be correct. As you often point out: look at the evidence. Instead of assuming what I believe and asking what I support, why don’t we talk about the points I listed:
1. Phytosaur illia do not resemble the illium of Diandongosuchus in any meaningful way.
2. The pubes of Diandongosuchus do not resemble Parasuchus pubes in any meaningful way from a lateral view (again, lack of ventral and dorsal views may shed other light).
3. The sigmoidal femur is not meaningful due to its homoplastic nature across Reptilia.
These are direct, evidence-based criticisms of the traits that you identified as uniting Diandongosuchus with phytosaurs. The data does not back up your assertions about them. You also characterize the cervicals of Diandongosaurus as resembling phytosaurs by being taller than long – but you do not compare them with poposaurs – but I can leave this trait for discussion later since most of my refs are at work. Finally my last point, which was:
4. The skull of Poposaurus is poorly known but what little is known and illustrated does not resemble your reconstruction in any way.
Instead of worrying about what I “believe”, lets focus on those points.
1. The ilium of Diandongosuchus does not have to resemble the ilium of Parasuchus because Diandongosuchus is not a phytosaur. Rather (at present) it resembles that of ancestral taxa, like Youngina. The ilium was evidently one of the last parts to evolve in that direction — unless there’s an anterior process hidden beneath the femur. Which is very possible. But it doesn’t matter. Note the extreme changes in limb/torso proportion between Diandongosuchus and Parasuchus, along with the magnitude increase in size. These factors could be responsible for the changes noted.
2. As above.
3. Its just one more brick in the wall. Not unique. If you want to talk about characters restricted to clades, let’s talk about that calcaneum and that coracoid, both of which are close matches to Parasuchus.
4. Gauthier et al. report the skull is known from the premaxillae and a few partial tooth fragments in Poposaurus. Their illustration of the premaxilla indicates a robust tooth-bearing portion and a high naris on a gently curving margin, not beyond the bounds of Daemonosaurus or something close to it. It appears Gauthier et al. estimated tooth sizes in their illustration. So, I’m willing to change it, but not yet.
Thanks!
Unrelated: Have you read this article: Mario Bronzati, M., Montefeltro, F. C. & Langer, M. C. 2012. A species-level supertree of Crocodyliformes. Historical Biology DOI: 10.1080/08912963.2012.662680
Just saw a post on it on TetZoo and figured you might be interested if you hadn’t already taken a peek.