Is Bellubrunnus a hatchling or a juvenile?

It’s too bad that the new small pterosaur Bellubrunnus (Hone et al. 2012) was not admitted to a comprehensive phylogenetic analysis. The authors and others considered it a juvenile, if not a hatchling, due to a lack of fusion of various skeletal elements that are fused on other (but not all) pterosaurs. In the large pterosaur tree Bellubrunnus nests as an adult between other small adults.

Let’s imagine that Bellubrunnus is a hatchling or juvenile. 
What can we learn by scaling Bellubrunnus up to adult size?

ellubrunnus to scale

Figure 1. Click to enlarge. Bellubrunnus to scale, imagined as an adult of a hatchling and at other sizes in between. If a hatchling, Bellubrunnus would end up larger than the largest Rhamphorhynchus.

Currently Bellubrunnus is the size of sister taxa. However, if we consider it a hatchling, then at 8x larger an adult becomes the largest taxon in this clade. If we consider Bellubrunnus half grown or quarter grown, then intermediate sizes in blue (Fig. 1) indicate adult size, larger than sisters and more in line with the larger Rhamphs.

As evidenced by pterosaur embryos, pterosaurs hatched as virtual copies of the adult only 8x smaller. Not all small pterosaurs had unfused bones. Not all large pterosaurs had fused bones. Fusion or a lack thereof, follows phylogenetic patterns. This becomes visible when certain taxa are not excluded from analysis a priori. It is likely that Bellubrunnus is an adult because sister taxa are similar in size and shape. However, the adult would have been identical to a 8x smaller hatchling, not quite half as tall as the adult’s standing tibia.

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.

References
Hone DWE, Tischlinger H, Frey E and Röper M. 2012.
 A New Non-Pterodactyloid Pterosaur from the Late Jurassic of Southern Germany. PLoS ONE 7(7): e39312. doi:10.1371/journal.pone.0039312

What is happening between the legs of Sordes?

The hind limbs and soft tissues of Sordes.

Figure 1. Click to enlarge. The hind limbs and soft tissues of Sordes. Above, color-coded areas. Below the insitu fossil. Pay no attention to the little Sordes-like impression beneeath the right foot. It is likely an illusion. The disarticulated ulna and radius are not indicated here.

The Problem
The holotype of Sordes (Sharov 1971, Fig. 1) presented lots of interesting soft tissue preservation, including hair, wing membranes and uropatagia. Unwin and Bakhurina (1994) further described the various soft tissues, promoting a unique structure, the uropatagium, which they reported as a membrane stretched between the hind limbs and pedal digit 5 tips, without contacting the tail. They also reported the wing membrane had a deep chord with a trailing edge that contacted the ankle.  Unfortunately not much detail has been published on this all-important area to demonstrate these observations. Both published illustrations were mere small outlines with Sharov (1971) adding texture and Unwin and Bakhurina (1994) adding halftone dots.

What Are People Seeing Here?
In Figure 1 The soft tissues at the margins of these soft tissue structures are clear to see. Unfortunately, these soft tissues do not continue to contact the legs, either medially or laterally. In fact, there appears to be a large open space between both legs and the tail. Earlier workers graphically filled in these areas. The question is: is there more soft tissue impressed in the matrix in these open spaces? Or is that just the matrix itself?

An Alternate Proposal
Earlier I noted here and here that the right wing was largely complete and preserved a narrow chord wing, narrow at the elbow with a fuselage fillet. The ulna and radius of the left wing appear to be disarticulated and drifted posteriorly. The green area (Fig. 1) appears to be disarticulated wing membrane, twisted and folded, starting left of the left leg and continuing between the legs. No other pterosaur uropatagia are this substantial, this structured and this shape. So, what is the correct interpretation?

Still…
The alternate model has not been criticized, but it has also not been accepted. I’d like those who hold to the traditional model to produce some sort of interpretation of the various hair balls and membranes like the one above (Fig. 1). The Sordes holotype has colored so much thinking about pterosaurs that it would be GREAT if someone could translate this fossil in detail for a better understanding. OR, if someone could come up with just one more example of a wing membrane attached to the ankle with a single uropatagium spanning the hind limbs.

I’ve been waiting for this since 1994. Let’s not be shy now. Words won’t cut it. Cartoon outlines won’t cut it. Bring out the hirez tracings with circles and arrows and let’s settle this issue.

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.

References
Elgin RA, Hone DWE and Frey E 2011. The extent of the pterosaur flight membrane. Acta Palaeonntologica Polonica 56(1): 99-111.
Peters D 2002. A New Model for the Evolution of the Pterosaur Wing – with a twist. Historical Biology 15: 277–301.
Sharov AG 1971. New flying reptiles from the Mesozoic of Kazakhstan and Kirghizia. – Transactions of the Paleontological Institute, Akademia Nauk, USSR, Moscow, 130: 104–113 [in Russian].
Unwin DM and Bakhurina NN 1994. Sordes pilosus and the nature of the pterosaur flight apparatus. Nature 371: 62-64.

Breaking Down (a Part of) The Large Reptile Tree

There are various tests one can administer to a tree. One of them is to permit one or more extra steps beyond the minimum to see where the weaknesses and strengths lay.

The following four illustrations show the Archosauria and their ancestors from fully resolved (Fig. 1) to what happens when 1, 2 and 3 extra steps are permitted (Figs 2-4).

segment of the large reptile tree.

Figure 1. A fully resolved segment of the large reptile tree. 1265 steps.

segment of the large reptile tree

Figure 2. Adding one extra step produces lack of resolution at weaker nodes.

Shortest tree plus two extra steps.

Figure 3. Shortest tree plus two extra steps. More breakdown occurs and this is expected, but could be repaired with more characters specific to these clades.

segment of the large reptile tree

Figure 4. The shortest tree with three extra steps permitted. Here the loss of resolution equals or exceeds the retained resolution. No further testing was attempted. Strong clades defined by unique characters retain resolution. Partial specimens also help break down trees. Many of these are skull-only taxa or skull-less taxa.

Some large clades are retained while closest kin lose resolution, as expected. More characters specific to certain clades would help them keep resolution. Note that the major clades retained their distinction from other major clades.

Only more taxa can increase the possibilities for novel nestings by offering more possible nesting sites. More characters cannot provide more nesting opportunities. More characters can refine nestings within a limited range and/or they can strengthen established nestings by finding more synapomorphies (shared traits).

Any other insights would be gratefully appreciated.

Notes on Character Lists

David Marjanovic asked the following questions:

1) Have you understood the difference between a phenetic matrix and a matrix for phylogenetic analysis yet? Please explain it here. I ask because you describe how to make a phenetic matrix. I don’t want to run around accusing you of not understanding things that you actually do understand, so please explain it.

I had to look up phenetics to answer this. My analysis finds plesiomorphies and apomorphies, so it is a phylogenetic analysis designed to explore phylogenetic relationships. I used many of the same characters used by other large phylogenetic relationships. I tried to avoid general terms like “large” or “small” but attempted to measure ratios between elements, as in “Pedal 2.1 larger than p2.2 or not.”

2) When people say your matrix contains too few characters, they mean three different things. 

– One is the abstract ratio of taxa to characters: you have lots of taxa, but relatively few characters for them. The chance is therefore high – statistically speaking – that adding more characters that carry phylogenetic signal would change the tree you get.

Having tested this numbers question, I find that I evidently have enough characters to fully resolve my tree. Despite your objections, resolution is important. Lack of resolution reveals no news about relationships, whereas complete resolution does.

How does one know if a phylogenetic signal is carried by a character? Perhaps by testing to see if so. Most of my characters appear in clades. However a few appear isolated randomly, by convergence. Of course, in such cases, I could make a clade with more taxa sporting that trait simply by adding taxa related to those outliers.

– Another is what Mickey has been demonstrating in detail: your matrix lacks certain specificcharacters that are known to have an effect. You have not tested dinosaur phylogeny, because almost all of the characters that support other people’s dinosaur topologies are not in your matrix!As long as you don’t add them, your matrix remains incapable – in principle as well as in practice – of testing the Phytodinosauria hypothesis. Yes, incapable. I do not exaggerate.

Hey, my analysis can’t be all things to all people. You’ll notice I put no effort at into the phylogeny of elephants or swans. At the fringes, and dinos are definitely at the fringes, it’s easier to make node and leaf changes. I also note that other people’s trees lack certain taxa that affect the topology in my tree. So, there are problems on both sides. It’s not a Mexican stand-off. It’s an opportunity to try something new for both parties. If you think my tree is the final word, or that I think it is, you are mistaken. It is a new avenue to pursue, something interesting to look at, nothing more. Perhaps more right than prior studies. Only a competing study of similar size and scope can test it.

– The third is that the 228 characters in your matrix are not 228 independent informative characters. Several are correlated with each other; that’s like having fewer characters, some of which are weighted more than others. Others, on the other hand, may not even carry phylogenetic signal (there are tests for that).

No doubt true. Even so, an earlier test to randomly get rid 40% of the characters (which should have gotten rid of a larger percentage of correlated characters), still came up with the same basic tree. I tell you it is impossible to avoid some correlated characters as blunt teeth are usually correlated to a large torso. Wings are correlated to hollow bones and stem-like coracoids (or clavicles in bats). Lots more dorsal vertebrae is correlated to reduction and loss of limbs. So we follow the greatest parsimony and accept or work with certain faults.

I appreciate your questions and input. I will continue, over time, to add more strength and reliability to my tree, even if it takes chopping off some branches and moving them elsewhere, as the data and results indicate.

Characters: More or Less

The large reptile tree has been criticized for having too few characters, 228, to adequately work with 305 taxa. Despite these objections one tree was recovered (full resolution).

To test the supposition that the ratio of characters to taxa should change the tree topology I undertook the following exercise which can be repeated by anyone using the present matrix, available by request.

1. The number of taxa was reduced (Fig. 1) by excluding all taxa more primitive than Thadeosaurus and all of the new Lepidosauromorpha (the other half of the entire Reptilia). That reduced the number of taxa from 305 to 85 and effectively increased the ratio of characters to taxa by 268%. To no one’s surprise, a single tree was recovered from this segment with no change in tree topology.

2. The number of characters was reduced by 10% by deleting every 10th character. Three trees were recovered with no change in tree topology other than a minor loss of resolution.

3. The number of characters was reduced by another 10% by deleting every 10th character. 27 trees were recovered with no change in tree topology other than a minor loss of resolution.

4. The number of characters was reduced by another 10% by deleting every 10th character. 1200+ trees were recovered when I stopped the process. There was a great loss of resolution at the base of the Archosauria due to the inclusion of the very incomplete taxa Trialestes and SMNS 12352.

5. Trialestes and SMNS 12352 were deleted and the test rerun. 165 trees were recovered with no change in tree topology other than a minor loss of resolution.

6. The number of characters was reduced by another 10% (for a total of 40%) by deleting every 10th character. 1260 trees were recovered with 136 characters. The tree topology remained quite similar to the original tree (see Fig. 1) with loss of resolution at the weaker nodes.

tree topology after exclusion of 40% of the original 228 characters

Figure 1. Test segment from the large reptile tree showing loss of resolution but no change in tree topology after exclusion of 40% of the original 228 characters and deletion of two very incomplete taxa. Only 136 characters recovers this tree.

You be the judge
Is this still a good tree? Or would a further increase in the number of characters substantially change the tree topology? Most of the traditional clades have been recovered here (dinosaurs, protorosaurs, parasuchia, ornithischianas, etc). The only novelty occurs when certain taxa not previously included are present. That is the key to this segment of the large reptile tree. If more characters pertinent to specific leaves were included, I’ll grant you that may shift one leaf with another. If anything substantial changes, please let me know about it.

GIGO
David Marjanovic is fond of saying GIGO (Garbage In, Garbage Out). Well, the present test got rid of 40% of “the garbage” and still recovered the same tree. On the same note, with 40% of the characters deleted that should have removed many of the so-called correlated characters. The purported power of those arguments has been reduced with these tests.

The key to the success of the large reptile tree is its size, specifically the large number of taxa. More taxa provide more nesting opportunities. The large reptile tree provides nearly a full gamut of opportunities, greatly reducing the possibility of a “by default” nesting. By comparison, the ratio of characters to taxa has little effect on the topology in this series of tests.

I hope this settles the issue. If not, please send your results or make them known.

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.

Hyoid? Quadrate? Prearticular? Disagreements in the Skull of Bellubrunnus

The recent PlosOne paper on the small rhamphorhynchid, Bellubrunnus (Hone et al. 2012), presented readers with a ventral view of a crushed skull (Fig. 1). Hone et al. (2012) reported, The skull is especially difficult to interpret.” Alternate identifications were made here earlier (and in Figs. 2-5). Today we’ll take a closer look at several of the disagreements regarding skull bone identification.

I was going to let this pass (one blog is usually enough), but in communication with paleo artist Jaime Headden, I realized that something needed to be said. Jaime agreed with the identification of the hyoids here (in Fig. 1), but considered the bone labeled “ar” (=articular) as a prearticular, which in life was found inside the mandible below the cheek.

[*Prepublication Update Today from Jaime Headden:]
“After a third look at the UV photo in the Hone et al. (2012) paper, I am not convinced that the “hyoids” you identify are not even “detached” from the medial mandibular rami to which they are applied, but are continuous with the retroarticular process. There is no rostral taper on the element to the right (left mandibular ramus), and there is a continuous and non-jointed region on the element to the left (right mandibular ramus) in which the “elements” thus merely become part of the jaw. So my arguing they might even be disarticulated prearticulars? Thrown out the window. These elements are fused and locked with the jaws in question.”

Bellubrunnus skull

Figure 1. Bellubrunnus skull with identifications made by Hone et al. (2012). Disagreements in identification are colorized here. See Figure 5 for re-identified bone identifications. Click to enlarge.

Disagreements
Earlier and below (Figs. 2-5) I laid out three images of the skull with color overlays keyed to a color reconstruction of the Bellubrunnus skull in several views (Fig. 5). All of the bones fit like a puzzle and they greatly resembled those of a sister taxon, Rhamphorhynchus intermedius (no. 28 in the Wellnhofer 1975 catalog) and the most primitive Rhamphorhynchus species I have found in the large pterosaur tree.

In figure 1 from Hone et al. (2012), I think the “hyoids” (hy) were quadrates. Both are robust rods that lie very close to their articular points at the posterior jaws and occiput. Primitive quadrates like those in R. intermedius are much more similar to these than in more derived rhamphs, like the one at Digimorph.org. Moreover, large quadrates should be visible in this ventral view and on top of the other exposed bones. The bone identified by Hone et al. (2012) as a quadrate is a fragile gracile bone. It is the postorbital.

In figure one I think the “quadrate” (qd) was a quadratojugal. I think the “quadratojugal” (qj) was a postorbital. I think the “opisthotics” (?op) were a pair of supraoccipitals. I think the “articulars” (ar, in blue) were the hyoids. Jaime Headden thinks those were prearticulars (then changed his mind).  I think the “pterygoid”(pt) is the vomer pair. The “palatine” (pl) label is pointing to an indistinct area without delineation. I think the “nasals” (ns) were lacrimals. I think the “maxilla” (mx) was a nasal. Not colored in figure one, the left “squamosal” (sq) is a left opisthotic. The right squamosal (sq, on the left on this inverted skull) was correctly identified.

In the following figures the segregated images are colorized without labels, but a skull reconstruction of pasted images (Fig. 5) retains those colors. So here (Fig. 2) and in the reconstruction (Fig. 5) the quadrate is bright red.

Skull elements of Bellubrunnus.

Figure 2. Skull elements of Bellubrunnus. Whereas Hone et al. (2012) identified the red elements as large hyoids, here they appear to be quadrates.

The mandible and scleral rings of Bellubrunnus.

Figure 3. The mandible and scleral rings of Bellubrunnus. Getting the easy things out of the way first.

Some of the palate and occipital elements of Bellubrunnus.

Figure 4. Some of the palate and occipital elements of Bellubrunnus.

Unfortunately Hone et al. (2012) did not test their work with a reconstruction as I did here (and below in Fig. 5).

The second step of DGS, copying and pasting the traced elements into a reconstruction.

Figure 5. The second step of DGS, copying and pasting the traced elements into a reconstruction. The Digimorph.org Rhamphorhynchus makes a good guide for this.

Well, perhaps now you can see why there is so much disagreement among pterosaur workers! We can’t even agree which bones are which! I think the bones are readily identifiable. Hone et al. (2012) did not think so and (I think) they made several mistakes, as demonstrated by the reconstruction (Fig. 5). Hone et al. (2012) won’t appreciate that I’m doing this having never seen the fossil up close (I glanced at it once and took a photograph), while they had full access.

UV light image of a fully articulated Rhamphorhynchus

Figure 6. UV light image of a fully articulated Rhamphorhynchus skull from Frey and Tischlinger (2012, the one entangled with the fish). Note the slender rod beneath the posterior mandible. I think that is a hyoid. Earlier Jaime considered that a displaced prearticular, then chucked that ID. Note this hyoid is closely appressed to the mandible and bears no taper at either end.  In Bellubrunnus the hyoids were shorter, but identically located, and colored in blue.

Jaime’s Thoughts
In an effort to support my evidence, I sent Jaime Figure 6. (Earlier) Jaime thought the rod beneath the posterior mandible in figure six was not a hyoid but a displaced prearticular, a bone that in life would have been inside the skull. I think it is a hyoid. If that’s a hyoid, then the similar bones in Bellubrunnus are also hyoids and the large bones in Bellubrunnus can be quadrates.

Rhamphorhynchus intermedius, a sister to Bellubrunnus.

Figure 7. Rhamphorhynchus intermedius, a sister to Bellubrunnus. Note the simple shape of the quadrate.

Closest Sister Taxon Clue
Rhamphorhynchus intermedius (no 28 in the Wellnhofer catalog) nests as a sister to Bellebrunnus in the large pterosaur tree. Here (Fig. 7) the stem-like quadrate is shown in articulation.

Summary
Anyway, this why there are problems and (too often) ill will among paleontologists. Not only do some of us make mistakes, we don’t like it when others point out our errors. The pterosaur palate has been particularly vexing for most paleontologists. In 2000 I pointed out the fusion of the ectopterygoid with the palatine and that the secondary palate of pterosaurs was created by the maxilla, not the palatine. Furthermore I re-identified palatal elements in anurognathids.

If I’m wrong, I’m wrong with regard to Bellubrunnus, but at present this is my evidence and these are my convictions. Time for more tests and reconstructions.

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.

ADDENDUM
Here is the new lateral view reconstruction with the posterior jugal descending, as in closest kin.

The jugal here descends posteriorly,

Addendum Figure: The jugal here descends posteriorly, matching other rhamphs and campys. This smooths out the skull roof to match. Thanks to D. Marjanovic for pointing out the problem with the quadrates. 

References
Hone DWE, Tischlinger H, Frey E and Röper M. 2012. A New Non-Pterodactyloid Pterosaur from the Late Jurassic of Southern Germany. PLoS ONE 7(7): e39312. doi:10.1371/journal.pone.0039312

Wellnhofer P 1975a-c. Teil I. Die Rhamphorhynchoidea (Pterosauria) der Oberjura-Plattenkalke Süddeutschlands. Allgemeine Skelettmorphologie. Paleontographica A 148: 1-33.Teil II. Systematische Beschreibung. Paleontographica A 148: 132-186. Teil III. Paläokolgie und Stammesgeschichte. Palaeontographica 149: 1-30.

The strange teeth of Aerosaurus

Figure 1. Aerosaurus in situ, tracing of same and lateral view. Note the size disparity in the upper and lower teeth -- combined with the strangely curled (hyper recurved) teeth in the uppers -- combined with the very long tooth row extending far behind the orbit. What was this reptile eating?

Figure 1. Aerosaurus in situ, tracing of same and lateral view. Note the size disparity in the upper and lower teeth — combined with the strangely curled (hyper recurved) teeth in the uppers — combined with the very long tooth row extending far behind the orbit. What was this reptile eating?

Just took another look at Aerosaurus. Langston and Reisz (1981) reconstructed them a little more straightened out (see wiki illustration).

Figure 2. Aerosaurus skull reconstructed in dorsal and lateral views. Plus manus and pes and complete specimen in lateral view. This is one of the basalmost synapsids.

Figure 2. Aerosaurus skull reconstructed in dorsal and lateral views. Plus manus and pes and complete specimen in lateral view. Note the rotation of the postorbital from the in situ specimen. This is one of the basalmost synapsids.

The teeth were laterally compressed and >strongly< recurved. The teeth were so highly curved that they seem unable to penetrate flesh and the tooth row extends far behind the orbit. The lower teeth were relatively tiny. Such an arrangement suggests something other than a meat-eating diet, perhaps some sort of filtering or plant-eating. The parasphenoid is uniquely expanded laterally and posteriorly, and bears rows of teeth on ridges. The quadratojugal curved up posteriorly as did the slender mandible. A strange skull indeed.

References
Langston W Jr and Reisz RR 1981. Aerosaurus wellesi, new species, a varanopseid mammal-like reptile (Synapsida: Pelycosauria) from the Lower Permian of New Mexico. Journal of Vertebrate Paleontology 1:73–96.

Aerosaurus/wiki

A Game of Twenty Questions

Everyone knows the game of twenty questions.
Questioners start with general topics like animal, mineral or vegetable. You may end up with the winner of the best actor category in 1967 long before you get to question #20.

In phylogenetic analysis one asks whether a skull was longer than the neck, or not. Whether pedal 2.1 was longer than 2.2, or not. Whether the mandible was longer than the femur, or not. Combined with dozens to hundreds of other such questions one builds a model family tree of several to several hundred taxa based on the scores (answers) to the list of questions (characters).

Too Few Characters?
Recently several researchers (Mickey Mortimer and Darren Naish among them) have mentioned that my large reptile tree includes too few characters (228) for the number of taxa (305) and therefore was unreliable, thus giving themselves the free pass to ignore the fact that all sister taxa resemble one another. Actually, if you think about it, every taxon received 228 queries, quite a bit, especially when several had multiple answers.

Divide It!
Starting with a large taxon set (which I have), or a large character set (which they have) it is extremely easy to divide and reduce these sets to test and assess matrices and relationships. For instance, earlier I removed all the new lepidosauromorph taxa except the turtles and the basal pterosaur MPUM 6009 to see how these would nest without any other close sister taxa present. Both nested with basal sauropterygians, not archosaurs. In effect that doubled the characters/taxon ratio with no change on the topology for the remaining taxa.

A Friendly Challenge
In like manner I have challenged Mickey and Darren to divide their character lists in some sort of random fashion (only even characters, every fifth character, first quarter of all characters, etc.) to see if they recover the same tree at 800 characters as they do at 400, 200, 100 and 50 characters. If one deletes all skull characters, all skull only taxa need to be deleted also, obviously. I don’t need to know any more details than that. The test is simply to test the validity of analyses with various percentages of the character list excluded. If 100 characters out of 800 deliver 90% of the tree topology that should tell us something. If 700 characters deliver 90% of the tree topology that will tell us something else. If the percentages are much lower, we’ll all know the value of adding more characters, as they suggest.

It would be great if we can start with complete resolution, even if the taxon list was cherry-picked to do so.

I look forward to hearing these results.

On pterosaurs and dorsal frills [dangerous reading material ahead]

Dr. Darren Naish recently based an entire premise (“…ignore ReptileEvolution.com“) on observations I made several years ago with regard to seeing dorsal frills on pterosaurs. It should be noted that other than the basalmost pterosaur, MPUM6009, no such frills made their way into ReptileEvolution.com, with the exception of Pterorhynchus, which has clear fossil evidence on the tail, skull and gular sac for unusual extradermal membranes. Nothing present on the dorsal or cervical series though.

The search for dorsal frills followed the recovery of Longisquama at the base of the Pterosauria. Longisquama is famous for its plume-like frills, homologous with the dorsal spines seen in Huehuecuetzpalli (Reynoso 1998) and Sphenodon, the tuatara. If these spines made it from the Triassic to the present (in Sphenodon), it just made sense that such ornaments might linger on Longisquama sister/descendants, following the pattern of vestigial limbs or an appendix, so I looked for them.

Rhamphorhynchus specimen with soft tissue impressions

Figure 1. Click to enlarge. Rhamphorhynchus specimen with soft tissue impressions. Wing membranes, tail vane and uropatagia are well known in pterosaurs. a dorsal membrane occurs in some taxa (Pterorhynchus, Tapejara) over the skull, but is not traditionally known in Rhamphorhynchus. Here something mimicking a dorsal frill impression is indicated that extends from rostrum to proximal tail. Likely an artifact in the matrix or preparator marks, but it gives about the same impression as the commonly recognized wings and tail vane impressions. (Hopefully those were not also added by preparators! – but you never know…)  In any case the impressions are traced here for further study and identification. The left pectoral and forelimb elements are in green for clarity of presentation. The sternal complex cristospine is in deep red. PILS, parallel interphalangeal lines are indicated on the pes. Pedal 5.2/3 is atypical. Photo from Adam Hems. Please pass on the museum number if you have it.

This above specimen was featured in the Archaeopteryx exhibit at the Houston Museum of Natural Science in 2010. Unless they were added later, faint soft tissue impressions are preserved in the matrix of wing membranes, a tail vane and a uropatagium. In addition, equally faint parallel lines extend from the vertebrae dorsally and also from the skull. Are these matrix artifacts? Preparator marks? Perhaps so.

Dendrites forming around plume-like structures in Rhamphorhynchus.

Figure 2. Dendrites forming around plume-like shapes on the darkwing Rhamphorhynchus. There is a one-to-one correspondence with sacral and dorsal vertebrae. When the dorsal series turns down anteriorly, the dendrites no longer form. Note that dendrites rim much of the wing material. No other large plumes of dendrites are found around the fossil. Pay no attention to the parallel lines emanating from the skull as those are likely matrix artifacts, as in figure 1.

UV dendrites

Figure 4. Under UV light (Helmut Tischlinger) the dorsal dendrites are black as are the wing membranes.

The famous darkwing Rhamphorhynchus includes some dendrites and the largest of these form a dorsal series of dendrite plumes. Either they have formed or formed around plume-like structures with a one-to-one correspondence with the sacral and dorsal vertebrae. In UV the dendrites are black, as are the wings. Interesting that dendrites are minimal elsewhere on the fossil and no such plumes form elsewhere. Just an observation. Let’s leave this so-called “evidence/illusion” on the shelf for now. If you notice such things, you run the risk of becoming ostracized by paleontologists.

Damn That Longisquama!
Without that plume-laden pterosaur sister, Longisquama, in the lineage none of this dorsal plume-stuff would have ever crept into the imagination.

Interesting Side Note, Though…
More often than finding fossil bones of Longisquama, we’ve found disassociated plumes. If these plumes were indeed temporary structures, shed like deer antlers after the breeding season, then their apparent presence on just a few pterosaur specimens creates something of a scenario. If present they would have been no more substantial than dragonfly wings. But that’s IF they were present.

Warning
Do not attempt to add dorsal plumes to pterosaur reconstructions, except, perhaps, to very basal Triassic specimens, and then keep them small. At present these anomalies are best described as dendrites and matrix artifacts.

More heresies later.

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.

References
Reynoso V-H 1998. Huehuecuetzpalli mixtecus gen. et sp. nov: a basal squamate (Reptilia) from the Early Cretaceous of Tepexi de Rodríguez, Central México. Philosophical Transactions of the Royal Society, London B 353:477-500.