Digging into another cladogram, pt. 7 (finished!)

Posted on August 20, 2019 by davidpeters1954

Today’s blogpost
concludes a recent series in reducing and restoring a previously published cladogram focusing on Archosauriformes (Nesbitt et al. 2017, derived from Nesbitt 2011). Click the [previous] button above to see earlier steps in this study.

Here are several reasons
why derived rauisuchians, like Postosuchus (Fig. 1), nest with basal bipedal crocodylomorphs, like Hesperosuchus (Figs, 1, 2), in traditional cladograms, like Nesbitt et al. 2017. They really do look alike, in broad aspect and fine detail.

Figure 1. Postosuchus to scale with Hesperosuchus (Fig. 2). Except for size, these two share a long list of traits, but the list that separates them and nests them with others is even longer.

Postosuchus and Hesperosuchus converged
on a long list of traits, from the shape of the antorbital fenestra to the crocodile-like temporal architecture in Postosuchus, distinct from other rauisuchians. Except for size (Fig. 1), these two taxa share a long list of traits, leading to prior paradigms. Here the list that separates these two taxa and nests them with other taxa is even longer when rescored and when blank scores are scored for several taxa, including Hesperosuchus.

Figure 2. Hesperosuchus, a basal bipedal crocodylomorph from the Late Triassic. Note the many traits here shared with Postosuchus (Fig. 1).

The experiment on Nesbitt et al. 2017
is finished enough. I did not examine every single matrix box, but called it quits after the topology shifted from the traditional paradigm to the LRT topology. The reduced and rescored, and too often scored-for-the-first-time 400 traits vs. 62 taxa has resulted in a tree topology different than Nesbitt et al. 2017 recovered and quite similar to the large reptile tree (LRT, 1560 taxa). Here the key was not ‘taxon inclusion’ or the number of characters, but scoring and re-scoring more accurately. The .nex file is available now by request at info@reptileevolution.com.

I hope that young Sterling Nesbitt (2011)
was not unduly influenced or misguided in his scoring (or lack of scoring) decisions in order to please his mentors. Over the past eight years the LRT has been derided and suppressed because it differs from the traditional paradigm promoted by Nesbitt and others. When taxa, like Diandongosuchus and Chilesaurus, are tested without prior restraints, then they, too, nested as they did earlier in the LRT (and without citation). Details here and here.

Figure 1. Reduced and rescored cladogram from Nesbitt et al. 2017. This cladogram now greatly resembles the LRT.

The cladogram of Nesbitt et al. 2017
did not hold up to experimentation and testing. It followed Nesbitt 2011, which followed the paradigm of the day: that rauisuchians gave rise to crocodilians within a then popular clade, Pseudosuchia, based on ankle traits, for the last several years invalidated by the results of the LRT.

Along the way,
I learned more about Orthosuchus, Kayentasuchus and Yonghesuchus (which now nests closer to Saltopus), which lacks a skull. Two of these taxa are not in the LRT, but soon will be.

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: 292 pp.
Nesbitt S et al. 2017. The anatomy of Teleocrater Rhadinus, an early avemetatarsalian from the lower portion of the Lifua Member of the Manda Beds (Middle Triassic). Journal of Vertebrate Paleontology 142-177. https://doi.org/10.1080/02724634.2017.1396539

Digging into another cladogram, pt. 6

Posted on August 19, 2019 by davidpeters1954

I hate seeing this week-long experiment drag out like this…
but there is a lot of work here and slow, but steady progress. Still not finished. Thought you’d like to see the progress and transformation of the cladogram (Fig. 1). Click the [previous] button above to see earlier iterations of this work.

Re-scoring a reduced Nesbitt et al. 2017 matrix
A surprisingly large number of data matrix boxes in Nesbitt et al. 2017 were left unfilled, so I’m filling them. The resulting cladogram (Fig. 1) is getting to look more and more like the large reptile tree (LRT, 1560 taxa). Of those scores that needed to be re-scored, most were perplexingly obvious.

Figure 1. Getting close to the end cladogram reduced and rescored from Nesbitt et al. 2017.

The clade Phytodinosauria
now nests derived from the Herrerasaurus clade, as in the LRT. The Silesaurus clade now nests within the Poposauria, as in the LRT. The next step is to figure out what is attracting giant terminal taxa in the Rauisuchia, like Postosuchus, to some much smaller, not-quite-basal Crocodylomorpha (Fig. 1) in Nesbitt et al. 2017 (Fig. 1), contra the LRT. This traditional attraction, creating the invalidated clade Pseudosuchia, is the basis for including many more clades in the Archosauria than just the crocs + dinos, as recovered in the LRT.

Here’s a Nesbitt et al. 2017 character I thought oddly worded:
“Maxilla ventral portion: Mediolateral height greater than dorsoventrally length.” Typically mediolateral refers to width. Dorsoventral refers to height. Data entries like this I just left alone.

I will get to the comments
from the past week once this project reaches a conclusion. Sorry for the delay. I’d rather not answer to anything posted until the entire experiment is finished and all the pertinent images are uploaded (apparently server issues prevent this currently).

Digging into another cladogram, pt. 5

Posted on August 18, 2019 by davidpeters1954

Short note on progress today,
adding missing scores, repairing wrong scores, ignoring scores I can’t see. This is a good exercise in basic research, checking facts, learning new traits, becoming re-familiarized with taxa I have not looked at in years.

Over the past few days
here, here and here, we’ve been dissecting, reducing and re-scoring the cladogram from Nesbitt et al. 2017, which disagreed in many respects from the large reptile tree (LRT, 1560 taxa) and used a completely different set of some 400 characters to do so.

Figure 1. Midpoint cladogram reduced and rescored from Nesbitt et al. 2017. More work needs to be done.

Here (Fig. 1)
the dino mimics among the poposaurs, Silesaurus and Sacisaurus, are now nesting with other poposaurs, as in the LRT, still closer to dinos than the crocs are, distant from results recovered in the LRT.

The basal phytodinosaur,
Eoraptor, now nests closer to the other phytodinosaurs, but not within Herrarasaurus, as in the LRT.

More work needs to be done
with rauisuchians and crocs before I add relevant taxa (as determined by the LRT) until the pertinent scores are all repaired.

Digging into another cladogram, pt. 4

Posted on August 17, 2019 by davidpeters1954

Updated August 19, 2019
after a reader’s comment that the lizard with the impressive gular sac (Fig. 3) was not Anolis but Sitana. Thank you for the correction.

Figure 1. Procompsognathus, Marasuchus and Segisaurus nest together in the Theropoda despite their many differences. Among these three, only Marasuchus is tested in Nesbitt et al. 2017.

Today we look at Marasuchus and kin,
which traditional topologies nest outside the Dinosauria and the LRT nests inside the Theropoda, between Tawa and Coelophysis in the broadest sense.

I note that Nesbitt et al. 2017
did not score any traits for the skull of Marasuchus, even though the maxilla and occiput are known. Some characters in Nesbitt et al. 2017 are ‘greater than‘ versus ‘less than‘ dichotomies. Of course this makes no allowance for ‘sub-equal to‘ or ‘similar to.‘ Better to state a character as a true dichotomy as in: “greater than“, vs. “not greater than.”

Many of the characters scored for Marasuchus
were centered around the cervicals, hind limb and ankle. Some scores, like the relative lengths of metatarsals 2 and 4, were easy to repair. Others took calipers and time. Not sure why, but many readily scored matrix boxes were left unscored. I gained the impression this cladogram was a little rushed-through. I hope the tree topology was not determined in advance in order that Nesbitt’s work would fall in line with the traditional pseudosuchian hypothesis based on ankle traits. Satisfying the expectations of one’s mentor/professor is something that can happen, as we’ve seen in one prior similar incident involving a newly minted PhD.

Neotony
It is worth noting how much smaller Marasuchus is compared to its LRT sisters (Fig. 1) and to basal theropods in general. This may be the reason why there is no deep groove between the distal condyles of the femur; why there are fewer and smaller processes on the hind limb elements, etc.. Even so…

Figure 1. Midpoint cladogram reduced and rescored from Nesbitt et al. 2017. The higher number of MPTs happens when traits are rescored incompletely.

Figure 2. Midpoint cladogram reduced and rescored from Nesbitt et al. 2017. The higher number of MPTs happens when traits are rescored incompletely. Scleromochlus nested with Gracilisuchus a few days ago.

Results
Adding skull data and re-scoring Marasuchus and basal theropods shifted Marasuchus into the basal theropods. There is still much more work to be done.

On a different subject—
The following photo (Fig. 3) was posted on Facebook. Here a typically quadrupedal agamid (genus: Sitana) is showing off the largest and most colorful gular sac I have ever seen… and it is about to go bipedal.

Figure 2. Nearly bipedal anole (Anolis) and its gular sac on display.

Figure 3 Nearly bipedal agamid (Sitana) and its gular sac on display.

This is analogous
to what was happening in the Middle Triassic with lepidosaur pterosaur precursors like Cosesaurus, which were flashing and flapping their (probably colorful) fringed forelimbs while occasionally traveling bipedally (Rotodactylus tracks match its feet). Just a different way to impress the lady lepidosaurs.

Digging into another cladogram, pt. 3

Posted on August 15, 2019 by davidpeters1954

Continuing this experiment…
yesterday and the day before I opened up and pruned a cladogram from Nesbitt et al. 2017, wondering how the tree topology could be so different from the LRT. The two competing analyses share very few traits in common. Nesbitt et al. used nearly twice as many characters. Many are tiny traits found in ankles, braincases and vertebral articulation patterns. In this review based on published data I cannot deal with such minutia, so those scores have to stand. Otherwise taxon inclusion, exclusion and scoring problems were found yesterday.

Today
I looked at scores for two more sets of taxa that nest together in the large reptile tree (LRT, 1560 taxa), but not in Nesbitt et al. 2017. The first set includes Scleromochlus and Gracilisuchus (Fig. 1), two basal crocodylomorphs in the LRT. In counterpoint, in the Nesbitt et al. tree these two nest far from other crocs and far from each other.

Figure 1. The ancestry of Scleromochlus going back to Lewisuchus, Saltoposuchus, Terrestrisuchus, SMNS 12591 and Gracilisuchus.

After re-scoring,
Scleromochlus nests with Gracilisuchus in the reduced and rescored Nesbitt et al. cladogram (Fig. 2). A variety of other small bipedal crocs (Fig. 2) nest with them in the LRT.

The second set of taxa
include Turfanosuchus, Qianosuchus and Yarasuchus (Fig. 3). In the original Nesbitt et al. tree these three nest far from each other. They nest together after re-scoring.

Figure 2. Revised cladogram from Nesbitt et al. 2017. This experiment continues.

Overall the similarity
to other clade members is striking in both restored clades.

Figure 3. Yarasuchus, Qianosuchus and Turfanosuchus nest together in Nesbitt et al. 2017 after rescoring.

We’re not quite at the mid-point of this experiment,
so there is no expectation that other clades will remain inviolate in the rescored version of Nesbitt et al. 2017 (Fig. 2) as more taxa enter the testing phase. I am not surprised that so many wrong scores were originally entered because we talked about the many ‘odd bedfellows’ generated by Nesbitt 2011, examined in a nine-part series ending here.

Digging into another cladogram, pt. 2

Posted on August 14, 2019 by davidpeters1954

Yesterday we opened up a cladogram
by Nesbitt et al. 2017, for examination and taxon addition.

The first driving issue here
is one we looked at earlier and even earlier: the origin of the Aetosauria. Nesbitt 2011 and Nesbitt et al. 2017 considered similarly armored Revueltosaurus the basalmost taxon in the aetosaur branch. Barely armored Ticinosuchus nests nearby, one node crown ward (Fig. 1) and 8 steps removed.

Figure 1. Reduced cladogram from Nesbitt et al. 2017

By contrast,
the large reptile tree (LRT, 1560 taxa) nests Ticinosuchus as the proximal outgroup taxon to the Aetosauria. Revueltosaurus nests with Fugusuchus and Tasmaniosaurus, together forming a previously overlooked sister clade to traditional erythrosuchids, like Erythrosuchus.

A second driving issue is the simple addition of taxa
to another, larger, character list created by someone else and published in an academic journal. This is an experiment in doing so.

A third driving issue is character scoring,
checking Nesbitt et al. 2017 for errors and omissions, and if repairable, do so.

The MorphoBank Nesbitt character list needs a proofreader.
Several characters are not labeled and neither are their states, other than “State 0″, State 1”, etc. Those were deleted from my review of their cladogram since they contained no useful data. Before publishing, it’s best to take one last look around to see if everything is neat and tidy.

I realize that no character list is perfect,
but some in Nesbitt’s work look like they should be divided in two because a pair of dichotomies are presented. For instance:

Nasal

Does not possess a posterolateral process that envelops part of the anterior rams of the lacrimal.
Possess a posterolateral process that envelops part of the anterior ramis of the lacrimal.
Does not form part of the dorsal border of the antorbital fossa.
Forms part of the dorsal border of the antorbital fossa.

In this case
I pretended choices 3 and 4 did not exist. I may be wrong in doing so. You decide. IMHO this one needs to go back to the shop to be split in two even though the matrix, as a whole, is still good enough to run.

A minor point, but worth mentioning:
Verbose characters with many letters tend to run off the edges of the columns provided by MacClade, even when set quite wide. Characters do this too often in Nesbitt 2017 (example above). If I may offer a suggestion? In the above example, #2 could have been stated as is (or shortened by deleting the grammatical articles), while #1 could have been shortened to: “does not.” In the LRT characters are really pared down to make them easy to read in relatively narrower MacClade columns.

For the skull of Ticinosuchus
Nesbitt and Nesbitt et al. scored very few characters, perhaps because they did not understand the roadkill-crushed skull enough make a reconstruction. This is where DGS works wonders. The addition of Ticinosuchus skull traits brought this taxon within one step of the Aetosauria, compared to its original nesting. Corrections to the post-crania shifted Ticinosuchus to the base of the Aetosauria, as in the LRT, with 9 steps needed to move it back to its Nesbitt nesting.

With its emphasis on ankle traits
and hind limb long bone end shapes, the Nesbitt et al. cladogram has its roots in a traits-based era when archosauriforms were divided according to ankle traits. This era is when the traditionally beloved, but since invalidated clade ‘Pseudosuchia‘ originated. Since the concavity or flatness of certain ankle elements cannot be determined or confirmed from published data, I had to ignore all such traits and rely on others from the list of 400 or so. The remainder still left plenty to work with.

Several characters in Nesbitt et al. 2017
are related to braincase foramina, sutures and processes. These are not employed by the LRT because they, too, are not readily visible in published data. Palatal and occipital traits, whenever visible, are scored. Unfortunately palate and occipital reconstructions for Ticinosuchus have not been generated and those for Revueltosaurus have not been published. Bottom line: omitting all the tiny and hard-to-see traits lets all the large and easy-to-see traits determine relationships resulting in a cladogram in which sister taxa indeed do like each other.

Results
The addition of Fugusuchus (still need to add Tasmaniosaurus) and the re-scoring of errors and omissions moved Revueltosaurus to the Fugusuchus – Erythrosuchus clade away from Aetosauria. Ticinosuchus moved to the base of the Aetosauria (Fig. 2). Both of these are echoed in the LRT. Most of the rest of the Nesbitt et al. tree remains unchanged despite its difference from the LRT. Note: Euparkeria has moved one node root-ward, as in the LRT. Further testing is planned for several other taxa.

Figure 2. Reduced and partly rescored cladogram from Nesbitt et al. 2017. Colors focus on taxa discussed in today’s blog post. Revueltosaurus moves to Fugusuchus with one extra step.

The resulting .nex file is available now by request,
but I would postpone asking for it until more housecleaning has been done to it.

Digging into another cladogram

Posted on August 13, 2019 by davidpeters1954

Updated August 13, 2019
with a repair to a mistaken citation: Nesbitt et al. 2017, not Nesbitt and Desojo 2017.

A reader (NP) kindly supplied me with a cladogram
from Nesbitt et al. 2017 in regard to the placement of Revueltosaurus and Luperosuchus. This gave me a chance to reexamine scoring for some of the earliest taxa to enter the LRT and mistakes were found, but none to upset the tree topology.

The first step is to open the file.
I did this in MacClade, rather than Mesquite.

The next step is to omit all taxa outside the Archosauriforms
(as determined by the LRT), all taxa that are largely incomplete and all taxa absent from the LRT at present. 65 taxa remain to be tested along with 419 characters (almost twice as many as in the LRT).

Step 3 is to run the original data on PAUP.
This generates the tree topology (Fig. 1), based on a strict consensus of 260 trees. 373 traits are parsimony informative with regard to the shortened taxon list.

Figure 1. Reduced cladogram from Nesbitt et al. 2017

Step 4 is to order the 65 taxa in MacClade according to the PAUP results
because no tree file was saved by Nesbitt et al.

Step 5 is to give the results a cursory look.

Many clades, like Dinosauria, Rauisuchia and Crocodylomorpha have similar memberships in the LRT and in Fig. 1, with notable exceptions.
Arising from Euparkeria, the long-necked quadruped Yarasuchus gives rise to bipedal Scleromochlus and terminal Erpetosuchus. By contrast in the LRT Yarasuchus and the long-necked quadruped Qianosuchus are closely related.
Members of the Crocodylomorpha in the LRT are split into 5 clades in Fig. 1.
Members of the Poposauria in the LRT are split into 3 clades in Fig. 1.
Members of the Ticinosuchus clade in the LRT are split into 3 closely related clades and one distantly related clade in Fig. 1.
373 traits were not enough to separate the members of the Ornithischia.
Other issues I will postpone for the present.

The differences between the LRT and Nesbitt et al. 2017
are rather large. Why? We talked about Nesbitt 2011 in a 9-part series ending here.

With regard to Revueltosaurus,
Tasmaniosaurus and Fugusuchus are missing in Nesbitt et al. We looked at these three taxa earlier here. I will add them Nesbitt et al. to see what happens.

References
Desojo JB and Arcucci AB 2009. New material of Luperosuchus fractus (Archosauria: Crurotarsi) from the Middle Triassic of Argentina: the earliest known South American ‘Rauisuchian’. Journal of Vertebrate Paleontology 29(4): 1311-1315.
Nesbitt S et al. 2017. The anatomy of Teleocrater Rhadinus, an early avemetatarsalian from the lower portion of the Lifua Member of the Manda Beds (Middle Triassic). Journal of Vertebrate Paleontology 142-177. https://doi.org/10.1080/02724634.2017.1396539

The Pterosaur Heresies

There's something very wrong with our pterosaurs.