New insights into the ornithopod manus

Duckbills,
like Edmontosaurus, and their kin are the ornithopod ornithischian dinosaurs, a clade I have been ignoring until now. Wikipedia reports, “[they] started out as small, bipedal running grazers, and grew in size and numbers until they became one of the most successful groups of herbivores in the Cretaceous world, and dominated the North American landscape.” 

Dryosaurus, Camptosaurus, Iguanodon and Edmontosaurus are genera within this clade and each has an interesting manus (Fig. 1). When one works in phylogenetic analysis it is imperative to compare homologous digits (apples to apples). In ornithopods, those homologies appear to be masked and perhaps misinterpreted by the appearances of new phalanges and the disappearances of old phalanges. Putting them all in one image (Fig.1) clarifies all issues (even without traveling to visit the fossils firsthand!). Hopefully the data are accurate to start with.

This all started with a phylogenetic analysis
that appeared to indicate that Edmontosaurus had a manual digit 1 with an extra digit that made it look like manual digit 2. Comparisons to other ornithopods ensued. A quick look through the Internet brought B. Switek’s article (see below) to the fore.

Figure 1. Ornithopod manus. Here the hands of Dryosaurus, Camptosaurus, Iguanodon and Edmontosaurus are compared. Note the turquoise metatarsal homologies and the digit identification based on that.

Figure 1. Ornithopod manus. Here the hands of Dryosaurus, Camptosaurus, Iguanodon and Edmontosaurus are compared. Note the turquoise metatarsal homologies and the digit identifications based on that.

Science writer Brian Switek 
writing for Smithsonian.com reports,

  1. “…the great herbivore Iguanodon had prominent thumb spikes.
  2. “The peculiar false thumb of Iguanodon was originally thought to set into the dinosaur’s nose.”
  3. “But why should Iguanodon have a hand spike? “
  4. “Though my own suggestion is not any better than those I have been disappointed by, I wonder if the Iguanodon spike is a Mesozoic equivalent of another false thumb seen among animals today—the enlarged wrist bones of red and giant pandas…  the Iguanodon spike was rigid.” Unfortunately that’s as far as journalist Switek has allowed himself to go, rather than proposing the homologies and comparisons demonstrated here.

Giving credit where credit is due,
Switek may be the first to suggest the spike was not a digit. I don’t know and was not able to find out the history of the spike. Given the text from his blogpost, you can see Switek’s choice of words actually evolves from “thumb spikes” to “false thumb” to “hand spike” to “enlarged wrist bone”. Like Brian, I also lack a PhD, but that doesn’t stop us from making contributions. If I’m duplicating earlier academic efforts, please let me know so credit can be given.

Here we’ll show
that the spike is indeed a wrist element… that digit 1 in Iguanodon and related taxa have one more phalanx, making it look like digit 2.

We’ll start with
the right manus of Dryosaurus, a basal ornithopod (at least in the large reptile tree it is, where only one other ornithopod, Edmontosaurus, is currently represented). During the course of this, I want you to focus on the the homologies of metatarsals 2 and 3 (colored in turquoise). These, I think, will guide us to correct interpretations of the other elements of the various ornithopod manus.

Now back to the manus of Dryosaurus:

  1. Data comes form loose bones in a photo formed in the shape of a hand, not an in-situ articulated hand. Thus I do not know the identification or placement of the carpals
  2. Five metatarsals are present.
  3. Mt3 is the longest. Slightly shorter is mt2.
  4. Phalangeal formula is 2-3-4-3-2, but digit 1 does not appear to be tipped with a sharp ungual. Is it missing? If so, that adds a phalanx to the formula 3-3-4-3-2.
  5. Digit 3 is the longest. Slightly shorter is digit 2.
  6. Unguals are lost in digits 4 and 5.

The manus of Camptosaurus

  1. Is reduced (stumpy) by comparison to Dryosaurus
  2. Mt 1 is a disk. M1.1 is a disk
  3. M3.2 appears to fuse with m3.3
  4. m4.3 and m5.2 are lost
  5. The new phalangeal formula is 2-3-3-2-1

The manus of Iguanodon

  1. is more robust and highly modified by comparison to Dryosaurus
  2. Two wrist elements fill the wrist. Two others extend medially.
  3. Digit 1 is longer and now sports an ungual
  4. Ungual 1 is not sharp
  5. Ungual 2 is a round hoof
  6. Ungual 3 (m3.4) is lost along with m3.3
  7. Mt4 is shorter. Two tiny phalanges are added.
  8. Digit 5 is absent.
  9. The new phalangeal formula is 3-3-2-4-0

The manus of Edmontosaurus 

  1. is long and gracile by comparison to Dryosaurus.
  2. Again, digit 1 has 3 phalanges, matching digits 2–4.
  3. Digit 4 is a vestige
  4. Mt 5 is again absent
  5. As in Iguanodon, ungual 1 is not sharp and ungual 2 is a hoof
  6. The new phalangeal formula is 3-3-3-3-0.

Always interesting to 
uncover little paradigm busters like these. Now back to phylogenetic analysis…

Do ceratopsid juveniles (phylogenetically) nest together?

The discovery of a second juvenile ceratopsid
(Currie et al. 2016) raised an interesting point: “In phylogenetic analysis, if all characters are coded as seen, the two juvenile ceratopsids (a partial Triceratops skull and the UALVP 52613 juvenile, Fig. 1) nest together. However, when size or age dependent characters are [not scored], the new juvenile (Chasmosaurus) specimen groups with other adult Chasmosaurus specimens.”

Figure 1. Chasmosaurus juvenile UALVP 52613 specimen.

Figure 1. Chasmosaurus juvenile UALVP 52613 specimen lacking forelimbs due to  taphoniomic loss down a nearby sinkhole.

So, does phylogenetic analysis fail us?
The new UALVP juvenile was recognized/identified as being closer to Chasmosaurus, just as the juvenile Triceratops was recognized as being closer to Triceratops, both on the basis of character traits and prior to analysis. But the Currie et al. unedited analysis takes us in another direction…

From the introduction
“The specimen comprises a nearly complete skeleton lying on its left side, lacking only the front limbs and girdle, which were lost many years ago into a large sinkhole….”

“The juvenile nature of this specimen is based on several lines of reasoning. At approximately 1.5 min total length, it is the smallest articulated ceratopsid skeleton that has ever been recovered. Immature bone textures on cranial bones (Brown et al., 2009), open neurocentral sutures throughout most of the vertebral column, incomplete fusion of sacral vertebrae, lack of fusion between caudal ribs and vertebrae, poorly formed articulations between limb bones, and many other characters confirm that this is an immature ceratopsid….”

“Of all the chasmosaurines from Dinosaur Park, it is most similar to Chasmosaurus belli and C. russelli.”

This interpretation
was made by expert and experienced assessment. The question is, why would the unedited Currie et al. analysis separate the juveniles from the adults and nest the juveniles together? They’re not exactly tadpoles or caterpillars, but they do change somewhat during maturation, following basic archosauromorph (including synapsid/mammal) growth strategies, that lepidosauromorphs (including pterosaurs) are less likely to follow.

When an adult Chasmosaurus
and the juvenile Chasmosaurus are added to the large reptile tree, using a character list NOT specific to ceratoposids, the juveniles nest with their respective adults, not with each other. And this happens despite the very few bones that represent the juvenile Triceratops (posterior face and shield only). Notably there are no other competing ceratopsid candidates in the present taxon list. All data was gleaned from online images. The adult data may be  represented by chimaera mounts and chimaera drawings. If the Currie et al analysis was restricted to just an adult and juvenile Triceratops and just an adult and juvenile Chasmosaurus, would adults nest with juveniles as they do in the large reptile tree? We don’t know because that test was not run.

Here’s how the large reptile tree divides
the Chasmosaurus adult and juvenile from the Triceratops adult and juvenile (posterior skull traits only). Please feel free to provide better data or more precise readings for any of these interpretations. Some were difficult to figure from available sources. At present I do not include traits for parietal fontanelles or horn lengths, which are the easiest two traits that most commonly separate Chasmosaurus from Triceratops and are reflected in their juveniles.

  1. skull table: C: depressed terrace, medial and lateral crests; T: convex
  2. snout in dorsal view: C: not constricted; T: constricted
  3. orbit positon: C: postorbital > preorbital; T: subequal
  4. lateral rostral shape: C: convex, smooth curve; T: double convex
  5. nasals/frontals: C: nasals >; T: subequal
  6. antorbital fenestra: C: absent; T: without mx fossa
  7. orbit/upper temporal fenestra: C: orbit not > T: orbit >
  8. orbit position/skull: C: anterior half of skull; T: not
  9. orbit shape: C: round to square: T: taller than wide
  10. upper temporal fenestrae: C: not closed or slit-like; T: closed or slit-like
  11. frontal shape: C: not wider posteriorly; T: wider posteriorly
  12. frontal shape 2: C: without posterior processes; T: with posterior processes
  13. posterior rim of parietal: C: transverse; T: anteriorly oriented or curved.
  14. parietal skull table: C: forms a sagittal crest: T: broad
  15. squamosal descent: C: mid level; T: ventral skull (ventral maxilla)
  16. skull roof fusion: C: parietal fusion only; T: frontal fusion and parietal fusion
  17. jaw joint orientation: C: descends from ventral mx; T: in line with ventral mx, after jugal arch.
  18. last maxillary tooth: C: posterior orbit; T: mid orbit
  19. mandible ventrally: C: 2-tier convex; T: straight
  20. 2nd sacral rib: C: not: T: double wide laterally
  21. manus/pes: C: subequal: T: manus smaller
  22. ilium: C: posterior process >; T: not
  23. metatarsal 1:4 ratio: C: 1 not > than half: 4 T: 1> half of 4
  24. metatarsals 2-4: C: < than half the tibia; T: not
  25. pedal 3.1 vs p2.1: C: not > T: 3.1>
  26. metatarsals 2 and 3: C: aligns with mt1; T: aligns with pedal 1.1
  27. pedal 4 length: C: subequal to mt 4; T: > mt4
  28. pedal digit 3 vs 4: C: 4 narrower than 3; T: 4 is not narrower

Shifting the juvenile Triceratops
to the juvenile Chasmosaurus adds 12 steps. Doing the opposite adds 21 steps. Bootstrap scores are over 99-100 for the three nodes represented by the four taxa. I have not reviewed the scores or data in the Currie et al study, which obviously adds more ceratopsid traits.

Added < 24 hours after original publication Below is a new reconstruction of the Triceratops juvenile based on text measurements and an adult skull compared to the original reconstruction that does not appear to have correctly scaled the mandible to the skull elements.

Figure 4. A new reconstruction of the Triceratops juvenile with the mandible and squamosal scaled to text measurements and shaped to adult elements compared to the original (Goodwin et al.) reconstruction which appears to have shortened the mandible.

Figure 4. A new reconstruction of the Triceratops juvenile with the mandible and squamosal scaled to text measurements and shaped to adult elements compared to the original (Goodwin et al.) reconstruction which appears to have shortened the mandible.

A YouTube video, Dinosaurs Decoded, shows Mark Goodwin reassembling the juvenile Triceratops skull. Click here to watch.

_______________________

Short notes for readers and critics
“Criticism of a writer is absolutely inevitable.” — Malcolm Gladwell.
Gladwell is one of the most respected and best-selling authors in current decades. Nevertheless, this interview on YouTube quotes several critics, many with scathing barbs. So, this give and take between writers and their critics is universal and ‘inevitable.’

On the other hand,
in Science, one either can or cannot duplicate experiments and observations. It should be cut and dried, but with errors and egos on both sides, it rarely is. Even so, most people think it is better to try/experiment with/refute alternate hypotheses. Aaaaaat least that’s the editorial policy at ReptileEvolution.com where occasional lack of talent and insight is sometimes overcome by tenacity, huge blocks of data and the ability to update online blunders.

References
Currie PJ,  Holmes RB, Ryan MJ and Coy C. 2016. A juvenile chasmosaurine ceratopsid (Dinosauria, Ornithischia) from the Dinosaur Park Formation, Alberta, Canada. Journal of Vertebrate Paleontology. DOI: 10.1080/02724634.2015.1048348.

 

 

Nesting Triceratops and its juvenile

Updated May 26 with suggestions from C. Collinson on skull sutures.
Updated again with a new reconstruction of the missing juvenile Triceratops face. 

No surprises here. 

Figure 1. Triceratops mount from an auction house. Pectoral girdle repaired. Skull colorized. Dorsal view comes from another specimen - always a dangerous proposition.

Figure 1. Triceratops mount from an auction house. Pectoral girdle repaired. Skull colorized. Dorsal view comes from another specimen – always a dangerous proposition.

Triceratops (Fig. 1, Marsh 1889) and its juvenile (Fig. 2) nest together with Yinlong downsi (Xu et al. 2006) Late Jurassic ~150 mya, ~1.2 m in length; Fig. 3) a primitive bipedal hornless pro-ceratopsian ornithischian, dinosaur, archosaur, archosauriform, archosauromorph, reptile. The large reptile tree is now up to 678 taxa.

Figure 2. Juvenile Triceratops compared to subadult Triceratops (in shadow).

Figure 2. Juvenile Triceratops compared to subadult Triceratops (in shadow).

A YouTube video, Dinosaurs Decoded, shows Mark Goodwin reassembling the juvenile Triceratops skull. Click here to watch.

Figure 2b. Original figure from Goodwin et al. of juvenile Triceratops, but mandible and squamosal scale bars don't match. Then compared to an adult. Then reconstructed based on new mandible/squamosal proportions based on text measurements. Evidently the juvenile Trike had a longer rostrum than Goodwin thought.

Figure 2b. Original figure from Goodwin et al. of juvenile Triceratops, but mandible and squamosal scale bars don’t match. Then compared to an adult. Then reconstructed based on new mandible/squamosal proportions based on text measurements. Evidently the juvenile Trike had a longer rostrum than Goodwin thought.

Liike all ornithischians, 
ceratopsians fuse the postfrontal to the frontal. However, in Yinlong, cracks (sutures?) appear where the postfrontal would have appeared and where the orbital horns ultimately appeared. So are the postorbital horns actually derived from postfrontal buds? We won’t know until we can determine a suture from a crack in the ontogenetically youngest and phylogenetically most primiitive specimens. It is also possible that, like the nasal horn, the orbital horns arose from novel ossificatiions that ultimately fused to the underlying bone.

Figure 3. Yinlong skull showing possible postfrontal in the position of the future orbit horns.

Figure 3. Yinlong skull showing possible postfrontal in the position of the future orbit horns.

Another juvenile nests with its adult counterpart!
Several workers and readers have pointed to studies (sorry, I don’t have the reference here) in which juveniles did NOT nest with adults in morphological analysis. Notably these samples  (as I recall…) came from taxa that metamorphosed during ontogeny, like caterpillars > butterflies and tadpoles > frogs.

In another argument, perhaps reflecting a majority view, a peerJ reviewer expressed concern/fear/trepidation that: – “Finally, I don’t know that a phylogenetic analysis including juvenile specimens alongside adult specimens is going to give you a particularly trustworthy result.“ citing no references, but noting that juvenile hadrosaurs have distinct characters in the skull from adults, which we all know.

Such arguments have been raised whenever I suggested workers include tiny Solnhofen pterosaurs in phylogenetic analyses, especially so since we KNOW that hatchling pterosaurs were virtual copies of adults. Not so with dinosaurs in which the rostrum is shorter and the orbits are larger than in adults. Even with that handicap, the differences, at least in this one case, were not enough to separate adult from juvenile Triceratops, given the present taxon list, which, frankly has no other ceratopsians.

References

Goodwin MB, Clemens WA, Horner JR and Padian K 2006. The smallest known Triceratops skull: new observations on ceratopsid cranial anatomy and ontogeny. Journal of Vertebrate Paleontology 26(1): 103-112.Lambe LM 1902. New genera and species from the Belly River Series (mid-Cretaceous), Geological Survey of Canada Contributions to Canadian Palaeontology 3(2):25-81
Marsh OC 1898. New species of Ceratopsia. Am J Sci, series 4 6: 92.
Xu X, Forster CA, Clark J M and Mo J 2006. A basal ceratopsian with transitional features from the Late Jurassic of northwestern China. Proceedings of the Royal Society B: Biological Sciences. First Cite Early Online Publishing. online pdf

 

wiki/Yinlong 
wiki/Triceratops

 

 

 

The large French Compsognathus specimen

Updated May 23, 2016 with a new mandible. M. Mortimer pointed out correctly that I had traced two coincident mandibles as one. 

The less well-known
French specimen of Compsognathus corallestris (Bidar et al. 1972b; Peyer 2006; CNJ79) is a bit larger with a different morphology (Fig. 1) than the coeval smaller Bavarian Solnhofen specimen, Compsognathus longipes (Fig. 1 right). Dr. Peyer considers these two Late Jurassic theropods conspecific and representative of ontogenic rather than phylogenetic variation.

Figure 1. The large (from Peyer 2006) and small Compsognathus specimens to scale. Several different traits nest these next to one another, but at the bases of two sister clades. Note the differences in the forelimb and skull reconstructions here. There may be an external mandibular fenestra. Hard to tell with the medial view and shifting bones.

Figure 1. The large (from Peyer 2006) and small Compsognathus specimens to scale. Several different traits nest these next to one another, but at the bases of two sister clades. Note the differences in the forelimb and skull reconstructions here. There may be an external mandibular fenestra. Hard to tell with the medial view and shifting bones.

From the Peyer abstract:
“The absence of an external mandibular fenestra, dorsally fan-shaped dorsal neural spines with hook-shaped ligament attachments, and a  very short McI and a PhI-1, which is stouter than the radius distinguish compsognathids from other coelurosaurs. Anatomical and morphological characters of the Bavarian specimen of Compsognathus are nearly identical to those of the French specimen. The differences are related to ontogenetic or within-species variation or are caused by preservational factors. Therefore this study proposes that C. corallestris is a subjective junior synonym of Compsognathus longipes from Bavaria.”

You’ll note that “compsognathids” sensu Peyer are scattered throughout this large reptile tree subset of the Theropoda (Fig. 2). Sinocalliopteryx and Juravenator are widely considered compsognathids, yet both nest far from one another here.

I tested the ontogenetic hypothesis of Peyer
in the large reptile tree. Indeed, the two Compsognathus specimens do nest next to one another, but at the bases of two different clades.

The smaller Compsognathus specimen
nested with Struthiomimus, Ornitholestes, Microraptor and T-rex, among others.

The large Compsognathus specimen
nested with the oviraptorid, Khaan, Limusaurus, therizinosaurs, Sinosauropteryx and others. More derived clades include Eotyrannosaurus and other paravians such as dromareosaurids, troodontids and birds.

Figure 2. Compsognathus corrallensis nests close to the holotype smaller specimen, but at the base of the next clade, which includes oviraptors, therizinosaurs, Juravenator and Sinosauropteryx.

Figure 2. Compsognathus corrallensis nests close to the holotype smaller specimen, but at the base of the next clade, which includes oviraptors, therizinosaurs, Juravenator and Sinosauropteryx. That means it is not the adult version of the smaller specimen.

The new reconstruction
of the large Compsognathus skull is relatively shorter. Both the premaxilla and the dentary tip are oriented slightly down. The bones of the mandible slid apart during taphonomy. Put them back together to match the skull length and you might get a mandibular fenestra, as also seen in the smaller Compsognathus. The new skull reconstruction (Fig. 1) was created using DGS, not freehand as in the Peyer reconstruction.

Figure 3. DGS tracing of large French Compsognathus skull. These parts were used to make the reconstruction in figure 1. Only the left side and top elements were colorized.

Figure 3. DGS tracing of large French Compsognathus skull. These parts were used to make the reconstruction in figure 1. Only the left side and top elements were colorized.

Current traditional compsognathids include the following taxa

  1. Compsognathus
  2. Sinocalliopteryx
  3. Juravenator (some say yes, others say no)
  4. Sinornithosaurus
  5. Huaxiagnathus

In the large reptile tree the clade that includes Compsognathus now include the following taxa

  1. Compsognathus
  2. all ornithomimids, including Struthiomimus

References
Bidar AL, Demay L and Thomel G 1972b. Compsognathus corallestris,
une nouvelle espèce de dinosaurien théropode du Portlandien de Canjuers (Sud-Est de la France). Annales du Muséum d’Histoire Naturelle de Nice 1:9-40.
Ostrom JH 1978. T
he osteology of Compsognathus longipes. Zitteliana 4: 73–118.
Peyer K 2006.
A reconsideration of Compsognathus from the upper Tithonian of Canjuers, southeastern France, Journal of Vertebrate Paleontology, 26:4, 879-896,
Wagner JA 1859. Über einige im lithographischen Schiefer neu aufgefundene Schildkröten und Saurier. Gelehrte Anzeigen der Bayerischen Akademie der Wissenschaften 49: 553.

wiki/Compsognathus

Marching Dinosaurs Video

Figure 1. Marching dinosaurs video. Click to view.

Figure 1. Marching dinosaurs video. Click to view.

Make sure you see this one.
Click on the pic or here to view. I am very impressed by the accuracy and quantity shown in this video.

And, of course,
I’m a big fan of humans and dinosaurs to scale AND walking videos!

A few days ago
I showed an early term pregnant pterosaur. Just a reminder (because I forgot, too!), that was not the first pterosaur with extra bones inside. Here’s the other one.

Nasal crests on theropods… insight from Sinosaurus

The latest addition 
to the large reptile tree is the crested theropod, Sinosaurus (Young 1948, Hu 1993; Figs. 1,2, Early Jurassic). So far (provisionally), It nests basal to the Allosaurus + Yutyrannus clade and is a sister to the Proceratosaurus+ Spinosaurus clade that also gave rise Deinocheirus (Fig. 1).

Figure 1. Here's where the crested theropod, Sinosaurusn nests in this subset of the large reptile tree.

Figure 1. Here’s where the crested theropod, Sinosaurusn nests in this subset of the large reptile tree. Note that other crested theropods, like Dilong and Gaunlong nest nearby. Proceratosaurus might have had a crest, larger than the tiny remnant that was preserved.

 

Figure 2. Sinosaurus cast with premaxilla in yellow.

Figure 2. Sinosaurus cast with premaxilla in yellow.

Two nasal crests
top the rostrum in Sinosaurus, but they are very close together. Because Sinosaurus does not nest with Dilophosaurus, and crestless taxa separate them, those two taxa developed similar crests by convergence.

Not so oddly
those two crests on Sinosaurus are barely a crest width apart from one another. Moreover, those those two crests include a healthy portion of premaxillae — and that’s unexpected because in most dinosaurs the ascending process of the premaxilla generally does not extend beyond the naris. But it does in this clade, as we saw earlier with Deinocheirus.

Figure 3. Sinosaurus skull model along with tracing and line art from Hu 1993 then stretched to fit the model.

Figure 3. Sinosaurus skull model along with tracing and line art from Hu 1993 then stretched to fit the model. Note the extent of the premaxilla. Hu considered this another Dilophosaurus, but other workers corrected that hypothesis.

 

That sent me back looking at other clade members. 
And I found premaxillae extending beyond the naris in basal clade members. The derived forms have their own story to tell with the fused nasal boss overlying whatever premaxillary ascending process might have remained below it.

Figure 3. Yutyrannus with revised premaxilla extending beyond the naris.

Figure 3. Yutyrannus with revised premaxilla extending beyond the naris. Sinovenator and Majungasaurus both overlap the ascending process of the premaxilla with the nasal boss, so it appears much shorter. 

Is the crest of Ceratosaurus 
an emerging ascending process of the premaxilla? We’d have to look at the fossil itself, but this colorized version of an old illustration is a tantalizing clue to that possibility. In Allosaurus (Fig. 4) and Acrocanthosaurus the ascending process of the premaxilla is reduced to a slender rod that looks like a medial ridge of the nasal. I think this has been overlooked by prior workers.

Figure 4. Ceratosaurus with extended premaxilla. It is the horn sticking up between the nasals.

Figure 4. Ceratosaurus with extended premaxilla. It is the horn sticking up between the nasals.

What we’re seeing here is 
the counterintutiive phenomenon that crests came first in this clade, with convergent reduction, or coverup. in derived larger taxa.

Figure 5. Allosaurus skull with bones colorized. Note the extent of the premaxillary ascending processes over the nasals.

Figure 5. Allosaurus skull with bones colorized. Note the extent of the premaxillary ascending processes over the nasals.

With present taxa as data
the twin, barely separated crests of Sinosaurus apparently evolved from the similar morphology in its sister, Proceratosaurus, (Figs. 6, 7).

Figure 6. Proceratorsaurus skull with premaxilla (yellow) and nasal (pink) demonstrating how twin nasal crests could have originated with the loss of the premaxillary ascending process.

Figure 6. Proceratorsaurus skull with premaxilla (yellow) and nasal (pink) demonstrating how twin nasal crests could have originated with the reduction of the premaxillary ascending process.

A closeup in anterior view
of Proceratosaurus (Fig. 7) shows the early stages of splitting of the premaxilla and nasals to form two crests.

FIgure 7. Proceratorsaurus premaxilla. Note the tentative split at the top.

FIgure 7. Proceratorsaurus premaxilla. Note the tentative split at the top of the premaxilla. 

References
Hu S-J 1993. Short Report on the Occurrence of Dilophosaurus from Jinning County, Yunnan Province. Vertebrata PalAsiatica 31(1):65-79. Translated by Will Downs 1998.
Young CC 1948. On two new saurischians from Lufeng, Yunnan. Bulletin of the Geological Society of China (Acta Geologica Sinca) 28 (1–2): 75–90. doi:10.1111/j.1755-6724.1948.mp281-2007.x.

Those Dilophosaurus crests…

Revised February 15, 2016 with the deletion of the putative sister taxa, Megapnosaurus, which did not nest with Dilophosaurus when finally put to the test in the large reptile tree today. 

Sometimes it is easy to see the sutures on a skull. 
Sometimes, not so easy. That’s the case with Dilophosaurus (Welles 19454) an early theropod dinosaur with a partial skull that is cracked in so many places, the skull sutures that create the crest are camouflaged. So some alternates are proposed here (Fig. 1) without seeing the original paper, unfortunately.

Figure 1. Dilophosaurus skull with alternates for the maxilla (green), nasal (pink), premaxilla (gold) and lacrimal (tan) colorized. I have not yet seen the paper, so I don't know where Welles identified sutures.

Figure 1. Dilophosaurus skull with alternates for the maxilla (green), nasal (pink), premaxilla (gold) and lacrimal (tan) colorized. I have not yet seen the paper, so I don’t know where
Welles identified sutures.

 

If anyone has more valid details on this problem
please send them here.

References
Welles SP 1954. New Jurassic dinosaur from the Kayenta formation of Arizona”. Bulletin of the Geological Society of America 65 (6): 591–598. doi:10.1130/0016-7606(1954)65[591:NJDFTK]2.0.CO;2.