The joy of finding mistakes: fewer stem dinosaurs

Finding mistakes is what I hope to do every day
in my own work, as well as that of others. Each time that happens, the data set improves. Lumping and splitting improves. The hypothetical topology of the large reptile tree (LRT, 1036 taxa) gets closer to echoing the topology of Nature itself. Science is a process of winnowing through the data and finding earlier mistakes.

Figure 1. Revision to the LRT with a focus on the Archosauria. Here taxa with a long carpus all nest within the Crocodylomorpha, following traditional thinking. Dinosaur outgroups are reduced. PVL 4597 is still the basalmost archosaur.

Figure 1. Revision to the LRT with a focus on the Archosauria. Here taxa with a long carpus all nest within the Crocodylomorpha, following traditional thinking. Dinosaur outgroups are reduced. PVL 4597 is still the basalmost archosaur.

Today
I discovered some scoring errors among former ‘stem dinosaurs’ that turned them into basal crocodylomorphs. That’s a small shift and it involved turning some ‘absent’ scores in pedal digit 5 to ‘unknown’. It’s noteworthy that some related taxa have two tiny phalanges on pedal digit 5. A related taxon, Gracilisuchu, was illustrated by Romer (1972, Fig. 3) as a combination or chimaera of separate specimens, something I just today realized and rescored. One of those specimens is the so-called Tucuman specimen (PVL 4597, Fig 1), which nests apart from the Gracilisuchus holotype (Fig. 2) in the LRT.

Figure 1. The PVL 4597 specimen attributed to Gracilisuchus by Lecuona et al. 2017, but nesting at the base of the Dinosauria in the LRT.

Figure 2. The PVL 4597 specimen attributed to Gracilisuchus by Lecuona et al. 2017, but nesting at the base of the Dinosauria in the LRT. That fibula flange turns out to be another important trait. 

The corrected results
resolve the long proximal carpal issue in crocodylomorphs very neatly. Now, as in traditional thinking, that trait is restricted to only the crocodylomorphs and it gives us a basalmost taxon with the trait, Junggarsuchus. You might think, and it would be reasonable to do so, that phylogenetic bracketing permitted the addition of a long carpus and long coracoids with more confidence to taxa that don’t preserve this, like Gracilisuchus and Saltopus. But another related basal crocodylomorph, Scleromochlus, has small round coracoids, evidently a reversal. The carpal length is not clearly documented in Scleromochlus (Fig. 4).

Gracilisuchus

Figure 3. A basal archosaur with a very similar nasal bone, Gracilisuchus. Note pedal digit 5 here. This is how Romer 1972 illustrated it. The actual data is shown in figure 2, the Tucuman specimen, PVL 4597. The coracoid is not known in the holotype. 

Despite the short round coracoids of Scleromochlu
and its apparently short carpals, enough traits remain to nest it as a basal crocodylomorph, following the rules of maximum parsimony.

Figure 1. Scleromochlus forequarters. The yellow area shows the hand enlarged in situ. The size of the Scleromochlus hand makes it the last possible sister to pterosaurs, famous for their very large hands.

Figure 4. Scleromochlus forequarters. The yellow area shows the hand enlarged in situ. Large carpals do not appear to be present and the coracoids are not elongated. 

On a more personal note
I found out my art and a short bio were included in a paleoart website:
http://paleoartistry.webs.com while looking for information on friend and paleoartist, Mark Hallett, (wikipage here) whose website is down and I worried about his health. No worries. Mark just let his website lapse.

The author of the paleoartistry page
had both kind words and controversy for me:
“After David Peters’ excellent paintings in Giants, and A Gallery of Dinosaurs and Other Early Reptiles, as well as his own calendar, it seemed he was on his way to becoming one of the most reliable paleoartists of the 1990s, if not of all time. However, very controversial theories on reconstructing pterosaurs led to some harsh critiques obscuring Peters’ artistic brilliance.” 

That’s okay.
“Very controversial” does not mean completely bonkers (or am I reading too little into this?). It just means it inspires a lot of chatter. Or… it could mean that the author of the post follows the invalidated observations of Elgin, Hone and Frey 2010, which are the traditional views (Unwin and Bakhurina 1994), still used in David Attenborough films. If so, that would be a shame. Science is usually black and white – is or isn’t, because you can observe and test (Fig. 5) and all tests, if done the same, should turn out the same.

And you don’t toss out data
that doesn’t agree with your preconception, like Elgin, Hone and Frey did. In reality, my “very controversial reconstructions” remain the only ones built with DGS, not freehand guesswork or crude cartoonish tracings (as in Elgin, Hone and Frey 2010). The membranes (brachiopatagia and uropatagia) were documented in precise detail in Peters 2002, 2009 and here online.

Click to animate. This is the Vienna specimen of Pterodactylus, which preserves twin uropatagia behind the knees.

Figure 5. Click to animate. This is the Vienna specimen of Pterodactylus, which preserves twin uropatagia behind the knees.

References
Elgin RA, Hone DWE and Frey E 2011. The extent of the pterosaur flight membrane. Acta Palaeontologica Polonica 56 (1), 2011: 99-111. doi: 10.4202/app.2009.0145
Peters D 2002. A New Model for the Evolution of the Pterosaur Wing – with a twist. – Historical Biology 15: 277–301.
Peters D 2009. A reinterpretation of pteroid articulation in pterosaurs. Journal of Vertebrate Paleontology 29:1327-1330.
Romer AS 1972. 
The Chañares (Argentina) Triassic reptile fauna. An early ornithosuchid pseudosuchian, Gracilisuchus stipanicicorum, gen. et sp. nov. Breviora 389:1-24.
Unwin DM and Bakhurina NN 1994. Sordes pilosus and the nature of the pterosaur flight apparatus. Nature 371: 62-64.

wiki/Gracilisuchus
paleoartistry.webs.com/1980s.htm

Advertisements

Saltopus preserves early archosaur skin and scales

Basal Crocodylomorpha

Figure 1. Basal Crocodylomorpha, including Gracilisuchus, Saltopus, Scleromochlus and Terrestrisuchus

The basal archosaur
(crocs + dinos) Saltopus (von Huene 1910; Late Triassic; ~210 mya, ±60 cm long; Figs. 1–3) poorly preserves bones, but also preserves some scaly skin.

Figure 2. Saltopus skin and scales surrounding the right pelvis.

Figure 2. Saltopus skin and scales surrounding the right pelvis. Not all bones nor all scales are traced here. Benton and Walker 2011 reported no evidence for osteoderms. The bones are hard to delineate and segregate from scales because here they are covered with fossilized desiccated skin. Photo from Benton and Walker 2011 who trace the femoral head extending beneath the pelvis. 

Saltopus nests well within
the Crocodylomorpha and, along with Scleromochlus (Fig. 1), present examples of basal archosaur skin and scales. Even more basal, Gracilisuchus (Fig. 1) had dorsal scutes derived from ancestors going back to the Early Triassic Euparkeria.

Figure 2. Saltopus pelvis latex peel from Benton and Walker 2011. They found two large sacrals. Using DGS I found four sacrals, the same length as the dorsals and causals. Sister taxa have four sacrals.

Figure 2. Saltopus pelvis latex peel from Benton and Walker 2011. They found two large sacrals. Using DGS I found four sacrals, the same length as the dorsals and causals. Sister taxa have four sacrals.

Gracilisuchus and basal dinosaurs
had only two sacral vertebrae, but basal bipedal crocs, like Scleromochlus, double that number. Benton and Walker 2011 traced two sacrals in a latex cast of the sacral area, but each sacral was twice as long as proximal dorsals and causals. Here four sacrals are tentatively identified in the latex peel, all about as long as proximal non-sacral vertebrae.

Dinosaur skin
can be scaly, or naked with feathers, or a combination of the two. Dinosaur scales may be different than croc or lizard scales in that at least some dinosaur scales, like those on the metatarsus of theropods appear to be derived from former feathers.

References
Benton MJ and Walker AD 2011. Saltopus, a dinosauriform from the Upper Triassic of Scotland. Earth and Environmental Science Transactions of the Royal Society of Edinburgh: 101 (Special Issue 3-4):285-299. DOI:10.1017/S1755691011020081
von Huene FR 1910. Ein primitiver Dinosaurier aus der mittleren Trias von Elgin. Geol. Pal. Abh. n. s., 8:315-322.

wiki/Saltopus

Pseudhesperosuchus fossil photos

Earlier I used
Greg Paul and José Bonaparte drawings of the basal bipedal croc Pseudhesperosuchus Bonaparted 1969) for data on this taxon. The specimen has some traits that lead toward the secondarily quadrupedal Trialestes. Together they are part of a clade that is closer to basal dinosaurs than traditional taxa paleontologists have been working with.

The drawings were great,
but I wondered what the real material looked like…and more importantly, what was real and what was not.

A recent request to
the curators at Miguel Lillo in Argentina was honored with a set of emailed jpegs from their museum drawers (Fig.1), for which I am very grateful. These were traced in line and color and reassembled with just a few unidentified parts left over (Fig. 2).

Figure 1. GIF movie of the skull of Pseudhesperosuchus showing the original drawing, the fossil and DGS tracings of the bones.

Figure 1. GIF movie of the skull of Pseudhesperosuchus showing the original drawing, the fossil and DGS tracings of the bones.

Pseudhesperosuchus jachaleri (Bonaparte 1969 Norian, Late Triassic ~210mya, ~1 m in length, was derived from a sister to Junggarsuchus and  Lewisuchus and was at the base of a clade that included Trialestes on one branch and the Dinosauria on the other branch.

Figue 1. A new reconstruction of the basal bipedal croc, Pseudhesperosuchus based on fossil tracings. Some original drawings pepper this image. Note the interclavicle, missing in dinosaurs and the very small ilium, only wide enough for two sacrals. The posterior dorsals are deeper than the anterior ones.

Figue 2. A new reconstruction of the basal bipedal croc, Click to enlarge. Pseudhesperosuchus based on fossil tracings. Some original drawings pepper this image. Note the interclavicle, missing in dinosaurs and the very small ilium, only wide enough for two sacrals. The posterior dorsals are deeper than the anterior ones.

Much larger and distinct from Lewisuchus,
the skull of Pseudhesperosuchus had a smaller antorbital fenestra, an arched lateral temporal fenestra, a deeper maxilla and a large mandibular fenestra. The seven cervicals were attended by robust ribs.

The scapula and coracoid were each rather slender and elongated. An straight interclavicle was present. The forelimbs were long and slender. The radiale and ulnare were elongated, a croc trait. Only three metacarpals and no digits are known.

The ilium was relatively small, but probably longer than tall and not perforated. The femur remained longer than the tibia. The tarsus, if that astragalus is identified correctly, included a simple hinge ankle joint. Only two conjoined partial metatarsals are known.

There is a small box
full of little sometimes interconnected squares among the Pseudhesperosuchus material (Fig. 2, aqua colored). I’m guessing that those are osteoderms, and if so, were probably located along the back. These would have helped keep that elevated backbone from sagging in this new biped.

The improvements in the Pseudhesperosuchus data
changed a few scores, but did no change the tree topology. The large reptile tree (LRT) can be seen here.

It’s good to see what Pseudhesperosuchyus really looked like,
— or at least get a little closer to that distant ideal. Size-wise and morphologically, this largely complete specimen is closer to the basal dinosaur outgroup than any other currently included in the LRT. And yet it is also distinctly different as it shares several traits with Trialestes unknown in any dinosaur. As a denizen of the Late Triassic, Pseudhesperosuchus represents a radiation that occurred tens of millions of years earlier, probably in the Middle Triassic. None of this clade survived into the Jurassic, as far as we know.

References
Bonaparte JF 1969. Dos nuevos “faunas” de reptiles triásicos de Argentina. Gondwana Stratigraphy. Paris: UNESCO. pp. 283–306.

Long carpals on crocodylomorphs = quadrupedal stance?

Figure 1. Terrestrisuchus is a bipedal basal crocodylomorph with elongate proximal carpals.

Figure 1. Terrestrisuchus is a bipedal basal crocodylomorph with elongate proximal carpals.

Long proximal carpals,
like the radiale and ulnare in Terrestrisuchus (Figs. 1, 2; Crush 1984), distinguish most crocodylomorphs from all basal dinosaurs (Fig. 2).

The question is:
why did long carpals develop? A recent comment from a reader suggested they enabled quadrupedal locomotion. But looking at the proportions of Terrestrisuchus does not inspire great confidence in that hypothesis. Terrestrisuchus has elongate carpals AND it seems to be comfortably bipedal with hands that only descend to the knees. And the pectoral girdle is relatively gracile.

Figure 2. Manus of several crocodylomorphs compared to the basal dinosaur, Herrerasaurus. Not sure what those two bones are on Junggarsuchus as they were cut off as shown when published.

Figure 2. Manus of several crocodylomorphs compared to the basal dinosaur, Herrerasaurus. Not sure how long those two proximal carpals are on Junggarsuchus. They were cut off as shown when published. Since the distal carpals are labeled (dc) I assume  proximal carpals are cut off below them. Oddly the radiale is much smaller than the ulnare if so, or rotated beneath it, unlike the other crocs.

Looking back toward more primitive taxa
provides only one clue as to when the proximal carpals first started elongating: with Terrestrisuchus. The following basal and often bipedal croc taxa unfortunately do not preserve carpals.

  1. Lewisuchus
  2. Gracilisuchus
  3. Saltopus
  4. Scleromochlus
  5. SMNS 12591
  6. Litargosuchus
  7. Erpetosuchus

Phylogenetic bracketing suggests that all
were bipedal or facultatively bipedal. Post-crania is missing or partly missing in several of these specimens.

Gracilisuchus

Figure 3. Gracilisuchus does not preserve the hands or carpals, but was possibly experimenting with bipedal locomotion based on its proximity to taxa that were obligate bipeds. Note the tiny pectoral girdle.

The distal carpals,
wherever preserved (Figs. 2, 3), appear to be small, scarce and flat, the opposite of a supple flexible wrist. So the proximal carpals of crocs comprise the great majority of the wrist, distinct from dinosaurs (Fig. 2).

Figure 3. Alligator carpals.

Figure 3 Alligator carpals. Of course, this is a quadruped that has inherited long carpals from bipedal ancestors in the Triassic.

So… what do other bipedal taxa do with their hands?
Cosesaurus, a bipedal ancestor to pterosaurs, probably flapped, based on the shape of its  stem-like coracoid and other traits. Herrerasaurus, a bipedal ancestor to dinosaurs had elongate raptorial unguals (claws) lacking in any basal crocodylomorph (Fig. 2). Such claws were probably used in grasping prey in dinos… not so much in crocs.

The elongate proximal carpals in crocodylomorphs
appear to extend the length of the slender antebrachium (forearm) of Terrestrisuchus for only one reason at present. The offset lengths of the shorter radius and longer ulna become subequal again with the addition of the longer radiale and shorter ulnare. So there is no simple hinge joint at the antebrachium/proximal carpal interface. So that joint was relatively immobile. The lack of deep distal carpals also suggests a lack of mobility at the metacarpal/distal carpal interface in basal taxa. However in extant crocs, that hinge appears to be more flexible.

Figure 5. Trialestes parts. Note the much larger ulna relative to the radius and the much longer forelimb relative to the bipedal basal crocs.

Figure 5. Trialestes parts. Note the much larger ulna relative to the radius and the much longer forelimb relative to the bipedal basal crocs.

In Trialestes
(Fig. 5) the elongate fore limbs more closely match the hind limbs. So the elongate carpals in Trialestes do appear to enhance a secondarily evolved quadrupedal stance.

Also take a look at
Hesperosuchus, Dromicosuchus, Protosuchus. Saltoposuchus, Dibrothrosuchus, Baurusuchus, Simosuchus, and Pseudhesperosuchus. After long carpals first appeared in Terreistrisuchus, they do not change much despite the many other changes in the morphology of derived taxa. Bipeds have them. Quadrupeds have them. Long-bodied taxa have them, Short-bodied taxa have them.

Some thoughts arise
when considering the first crcoc with elongate carpals, Terrestrisuchus.

  1. At some point in the day Terrestrisuchus probably rested on its elongate pubis bone (the first in this lineage), flexing its long hind limbs beneath itself to do so. In that pose elongate carpals may have been useful in steadying the animal as it balanced on the pubis tip and whenever it rose to a bipedal stance.
  2. A male Terrestrisuchus may have used its hands to steady itself while riding on the back of a female while mating. The carpals were elongated as part of the balancing act performed during this possibly awkward bipedal conjugation.
  3. Coincidentally, the coracoids in crocodylomorphs begin to elongate in this taxon. So freed from quadrupedal locomotion duties, basal crocs may have done some early form of flapping as part of a secondary sexual behavior, long since lost in extant taxa.

So, in summary
I think the elongate carpals developed in crocs with a really long pubis to steady it while resting. Very passive. Not sure what other explanation explains more.

Did I miss anything?
Has anyone else promoted similar or competing hypotheses?

References
Crush PJ 1984. A late upper Triassic sphenosuchid crocodilian from Wales. Palaeontology 27: 131-157.

wiki/Terrestrisuchus

Estimating dino/croc divergence times: Turner et al. 2017

This might have been yet another case
of scientists TRUSTING authority (= the work of others) rather than TESTING competing phylogenetic analyses. In this case, however, two of the three authors in Turner, Pritchard and Matze 2017 relied on their own flawed (= serious taxon exclusion problems) phylogenetic analyses and for good measure they threw in a third flawed (= more taxon exclusion problems) analysis (Nesbitt 2011) that we examined and reexamined in an 11-part series.

In any case, since none of the trees
in the new Turner et al. study  stand up to scrutiny (= do not agree with one another, do not produce gradual accumulations of traits in derived taxa and depend on long ghost lineaages), everything Turner et al. (2017) did afterwards has no credibility and no utility. So sadly, the entire paper is a waste of their time. Metaphorically, they built their house on sand.

On the other hand,
when you start with a study that provides a gradual accumulation of derived traits in all derived taxa, and minimizes the effect of taxon exclusion, like the large reptile tree (LRT (949 taxa) then you’ve metaphorically built your house on solid ground. And it’s much simpler to pinpoint the dino/croc divergence time because you are provided with a last common ancestor for these sister clades: Gracilisuchus (Figs. 1, 2). Crocs and dinos are sister taxa. None of the studies used by Turner et al. (Pritchard et al. 2015, Nesbit 2011, Turner 2015) recovered that tested relationship.

Figure 1. The origin of dinosaurs to scale. Gray arrows show the direction of evolution. This image includes Decuriasuchus, Turfanosuchus, Gracilisuchus, Lewisuchus, Pseudhesperosuchus, Herrerasaurus, Tawa and Eoraptor.

Figure 1. The origin of dinosaurs to scale. Gray arrows show the direction of evolution. This image includes Decuriasuchus, Turfanosuchus, Gracilisuchus, Lewisuchus, Pseudhesperosuchus, Herrerasaurus, Tawa and Eoraptor.

So when did dinos and crocs diverge?
Let’s look a the three most recent taxa both clades share in common in reverse chronological and phylogenetic order:

  1. Gracilisuchus = 230 mya.
  2. Turfanosuchus = 235 mya.
  3. Decuriasuchus = 240 mya.

So that narrows the divergence time pretty well…

And how did Turner et al. do?
They report,“The average ghost lineage for the group as sampled is 31 million years.” Their conclusion states no firm date or date range. Rather, their whole paper appears to be a long story on how they tested this that and the other without getting around to their headline topic. And without nailing down a last common ancestor or a croc/dino divergence time.

Figure 2. Basal crocs. Decuriasuchus and Gracilisuchus are found in both croc and dino lineages.

Figure 2. Basal crocs. Decuriasuchus and Gracilisuchus are found in both croc and dino lineages.

All the other taxa
and all the other testing performed by Turner et al. were for nought.

For more information
on any of the taxa employed by Turner et al, just look them up at ReptileEvolution.com.

References
Nesbitt SJ 2011. The early evolution of archosaurs: relationships and the origin of major clades. Bulletin of the American Museum of Natural History;352:1–292.
Pritchard AC, Turner AH, Nesbitt SJ, Irmis RB and Smith ND 2015. Late Triassic tanystropheids (Reptilia, Archosauromorpha) from Northern New Mexico (Petrified Forest Member, Chinle Formation) and the biogeography, functional morphology, and evolution of Tanystropheidae. Journal of Vertebrate Paleontology. ;e911186.
Turner AH 2015. A Review of Shamosuchus and Paralligator (Crocodyliformes, Neosuchia) from the Cretaceous of Asia. PLoS ONE. 2015;10(2):e0118116. doi: 10.1371/journal.pone.0118116. pmid:25714338
Turner AH, Pritchard AC and Matzke NJ 2017. Empirical and Bayesian approaches to fossil-only divergence times: A study across three reptile clades. PLoS ONE 12(2): e0169885. doi:10.1371/journal.pone.0169885

 

Trialestes rises again!!

Lecuona et al. 2016
redescribe in greater detail Trialestes (Reig 1963; Figs. 1, 2), more than 50 years after its original discovery and publication. Glad to see this! Data used to nest Trialestes in the LRT as the proximal outgroup for the Dinosauria consisted of a few old drawings (Fig. 1), nothing more. The new data do not move the nesting much.

Figure 1. Tracings from old drawings are the data used to create this reconstruction of Trialestes, which nested it basal to the Dinosauria.

Figure 1. Tracings from old drawings are the data used to create this reconstruction of Trialestes, which nested it basal to the Dinosauria. New data from Lecuona et al. 2016 greatly reduce the guesswork here. 

From the Lecuona et al. abstract:
“Here, we describe in detail all the material assignable to the species and test its phylogenetic relationships using a comprehensive data matrix focused on early archosaurs. We support the referral of PVL 3889 to Trialestes and reject the presence of a mesotarsal ankle joint in this specimen. We recovered Trialestes within Crocodylomorpha, closer to Crocodyliformes than Pseudhesperosuchus, Hesperosuchus, Dromicosuchus and Sphenosuchus. Therefore, Trialestes represents the most completely known of the earliest non-crocodyliform crocodylomorph taxa known to date.”

They report, “In contrast, Reig’s third taxon, ‘Triassolestes’ romeri has received relatively little attention. In his original publication, Reig (1963) interpreted this taxon as a theropod dinosaur of the family Podokesauridae, a group now considered roughly equivalent to Coelophysoidea (Holtz 1994). Because of this interpretation, Reig considered only part of PVL 2561 as the holotype of the dinosaur ‘Triassolestes’, including an incomplete cranium and mandible, four cervical vertebrae, and 16 caudal vertebrae. Other postcranial remains associated with PVL 2561 were excluded from this genus, including a scapula, humerus, ulna, radius, carpus and proximal metacarpus, which were interpreted by Reig (1963, p. 15) as crocodilian elements because of the presence of an elongated radiale and ulnare.”

FIgure 2. Assembly of the many Trialestes parts featured in Lecuona et al. 2016.  This is an odd combination of robust cervicals and gracile limbs and girdles.

FIgure 2. Assembly of the many Trialestes parts featured in Lecuona et al. 2016. This is an odd combination of robust cervicals and gracile limbs and girdles.

Those proximal wrist elements
(radiale and ulnare) are never elongated in dinosaurs, but they are elongated in dinosaur ancestors like Trialestes. The name ‘Triassolestes’ was preoccupied by an Australian Triassic dragonfly, hence the change we know today.

Lecuona et al. continue: “Clark (in Benton & Clark 1988, fig. 8.6) hypothesized Trialestes as the sister taxon of Crocodylomorpha, but expressed some caution given the structure of the ankle and the poor knowledge of the specimens.”

“Novas (1989) recognized that the referred specimen PVL 2559 contained elements belonging to two individuals of different sizes (Reig 1963, fig. 4b part.) and assigned both of them to Herrerasauridae indet. (Novas 1989, 1993); presently, they can only be assigned to an indeterminate saurischian dinosaur.”

“Clark et al. (2000) questioned the referral of PVL 3889 to Trialestes romeri, suggesting that this specimen was more likely assignable to a basal dinosaur. Nevertheless, these authors could not reject the alternative explanation that PVL 2561 and PVL 3889 belong to one taxon with a combination of crocodylomorph and dinosaurian character states.”

“In addition, Trialestes has not been included in a quantitative phylogenetic analysis so far and, thus its affinities remain untested using modern methodologies.”

Well, the large reptile tree did that several years ago. But let’s keep an open mind moving forward.

Under materials and methods, Lecuona et al report,
“Herein we study the two specimens assigned to Trialestes romeri, the holotype PVL 2561 and PVL 3889; we exclude PVL 2559 because we agree with previous authors that it represents an indeterminate saurischian.”

Unfortunately
Lecuona et al. based their analysis on a cladogram originated by Nesbitt (2011) which had several problems listed here. They also included data from Butler et al. (2014) and added Carnufex, a related taxon.

The Trialestes fossils have a pristine beauty. The authors did not create a reconstruction, so I attempt one here (Fig. 2).

A little background data
Trialestes romeri (Bonaparte 1982)= Triassolestes (Reig, 1963/Tillyard, 1918) Carnian, Late Triassic ~235 mya is known from scattered parts. Clark, Sues and Berman (2000) redescribed the known parts and admitted the possibility that this taxon combined dinosaurian and crocodylomorph characters. As it nests here, Trialestes was derived from a sister to Carnufex. This clade phylogenetically preceded Herrerasaurus and the Dinosauria. We looked at this heretical relationship several years ago here.

Distinct from Pseudhesperosuchus,
the skull of Trialestes had a larger antorbital fenestra and a deeper rostrum. The mandibular fenestrae (yes, there are two!) were smaller.

The vertebral centra had excavated lateral surfaces, for bird-like air sacs. The radius was longer than the humerus, a character otherwise known only in dinosaurs. The radiale was smaller than the ulnare, matching the radius and ulna. The fingers were tiny.

The pelvis was semi-perforated, as in basal dinosaurs, with a well-developed supraacetabular crest. The dorsal pelvis was straight, as in Gracilisuchus. The femoral head was not inturned, suggesting a variable posture, promoted by that really long forearm. The ankle joint had a crocodile normal configuration and a functionally pentadactyl pes. Most crocs lose pedal digit 5, but not those basal to dinos, like PVL4597.

References
Bonaparte JF 1982. Classification of the Thecodontia. Geobios Mem. Spec. 6, 99-112
Clark JM, Sues H-D and Berman DS 2000. A new specimen of Hesperosuchus agilis from the Upper Triassic of New Mexico and the interrelationships of basal crocodylomorph archosaurs. Journal of Vertebrate Paleontology 20(4):683-704.
deFranca MAG, Bittencourt JdS and Langer MC 2013. Reavaliação taxonomica de Barberenasuchus brasiliensis (Archosauriformes), Ladiniado do Rio Grande do Sul (Zona-Assembleia de Dinodontosaurus). Palaenotogia em Destaque Edição Especial Octubro 2013: 230.|
Irmis RB, Nesbitt SJ and Sues H-D 2013. Early Crocodylomorpha. Pp. 275–302 in Nesbitt, Desojo and Irmis (eds). Anatomy, phylogeny and palaeobiology of early archosaurs and their kin. The Geological Society of London. doi:10.1144/SP379.24.
Kischlat EE 2000. Tecodôncios: a aurora dos arcossáurios no Triássico. Pp. 273–316 in Holz and De Ros (eds.). Paleontologia do Rio Grande do Sul. Porto Alegre: CIGO/UFRGS.
Lecuona A, Ezcurra MD and Irmis RB 2016. Revision of the early crocodylomorph Trialestes romeri (Archosauria, Suchia) from the lower Upper Triassic Ischigualasto Formation of Argentina: one of the oldest-known crocodylomorphs. Papers in Palaeontology (advance online publication). DOI: 10.1002/spp2.1056
Nesbitt S 2011. The early evolution of archosaurs: relationships and the origin of major clades. Bulletin of the American Museum of Natural History 352: 292 pp.

Reig, OA 1963. La presencia de dinosaurios saurisquios en los “Estratos de Ischigualasto” (Mesotriásico Superior) de las provincias de San Juan y La Rioja (República Argentina). Ameghiniana 3: 3-20.

Riff D et al. 2012. Crocodilomorfos: a maior diversidade de répteis fósseis do Brasil. TERRÆ 9: 12-40, 2012.

‘Origin of Dinosaurs’ video. 

wiki/Trialestes

When bipedal archosaurs become aquatic and croc-like

It happened twice.
Some bipedal theropods and bipedal theropod-like archosaurs evolved to become quadrupedal crocs and quadrupedal croc-like theropods.

Most famously and most recently,
Spinosaurus (Fig. 1), long suspected as being a fish-eater, was reconstructed with shorter than expected hind limbs, thus forsaking any ability to walk on its hind limbs alone. Some workers think this chimaera reconstruction is bogus. Others are more accepting. Spinosaurus was derived from bipedal theropods like Suchomimus and Sinocalliopteryx. Deinocheirus was giant frill back bipedal theropod related to Spinosaurus.

Figure 1. Derived from bipedal sisters, giant Spinosaurus had such short hind limbs that it could no longer rise to a bipedal configuration. Not only did it have a croc-like head, it had something approaching a croc-like post-crania (sans the sail, of course).

Figure 1. Derived from bipedal sisters, giant Spinosaurus had such short hind limbs that it could no longer rise to a bipedal configuration. Not only did it have a croc-like head, it had something approaching a croc-like post-crania (sans the sail, of course).

Basal bipedal crocs evolved to become extant quadrupedal crocs
Basal crocs were bipeds (Fig. 2), only later shortening the hind limbs to become quadrupedal, like the theropod dinosaur, Spinosaurus (Fig. 1).

Figure 2. The evolution of extant crocs from primitive bipeds and transitional quadrupeds. One branch led to Protosuchus, the other, along with several side branches, to extant species.

Figure 2. The evolution of extant crocs from primitive bipeds and transitional quadrupeds. One branch led to Protosuchus, the other, along with several side branches, to extant species. Interesting to see that Sphenosuchus has taller dorsal spines than either predecessors or successors.

It’s of mild interest
to note that the transitional croc, Sphenosuchushad taller dorsal spines than either more primitive or more derived taxa (Fig. 2).

It’s of even milder interest
to note the quadrupedal poposaur, Lotosaurus, is derived from a sister to the bipedal poposaur, Poposaurus.

Figure 1. Lotosaurus, a finback poposaur.

Figure 3. Lotosaurus, a finback poposaur.

Contra this pattern,
the finback Arizonasaurus is a likely biped (based on its deep pelvis) derived from quadrupedal, shallow-pelvis, basal rauisuchians, like Vjushkovia. Even so, the closest relatives of Arizonasaurus include croc-like Yarasuchus and Qianosuchus both of whom have semi-tall spines.

Figure 1. Arizonasaurus configured as a biped. The depth of the pubis suggests a similar length for the femur and tibia. The gracile pectoral girdle suggests a gracile forelimb. The long deep tail is based on the related Yarasuchus.

Figure 4. Arizonasaurus configured as a biped. The depth of the pubis suggests a similar length for the femur and tibia. The gracile pectoral girdle suggests a gracile forelimb. The long deep tail is based on the related Yarasuchus.

A relative of Arizonasaurus
by analogy, not homology, is Qianosuchus (Fig. 5). It shares many traits with Spinosaurus, sans the frill.

Figure 4. Qianosuchus shares quite a few traits with Spinosaurus, sans the frill. Qianosuchus has similarly-sized limbs.

Figure 5. Previously unnoticed, the derived rauischian, Qianosuchus, shares many traits with Spinosaurus, sans the frill. Qianosuchus has similarly-sized limbs and a similar long rostrum and neck.

References
Ibrahim N et al. 2014. Semiaquatic adaptations in a giant predatory dinosaur. Science 345 (6204): 1613–6.