This post has been modified from its original content. It’s important to remember that pterosaurs are lizards. They follow lizard-type growth patterns as reported by Maisano 2002 in which some lizards fuse bones and keep growing while others never fuse certain bones into old age. Pterosaurs also grow isometrically, with long-snouted, small eyed embryos known.
Traditional thinking follows the paradigm
that the unfused scapulocoracoid (s/c) in pterosaurs demonstrates immaturity. I tested this in a phylogenetic analysis. Turns out the patterns are not ontogenetic, but clearly phylogenetic. Scapulocoracoid fusion is on again, off again in patterns that are not the random pattern one would expect if ontogenetic in nature.
Figure 1. Click to enlarge. Pterosaur family tree (May 2013) highlighting scapulocoracoid fusion in pterosaurs (bright green) and lack of fusion (bright blue). Other taxa do not preserve the s/c. If ontogenetic we would expect a more scattered, randomized pattern. That’s not the case here as fusion patterns follow phylogeny, not maturity. Some taxa here do not preserve the scapula and coracoid. Not listed here, but related to Cearadactylus, Barbosania does not fuse the s/c. Some taxa have complete fusion. Others retain a line of fusion. Among the higher ornithocheiridae there is the greatest randomness in fusion.
Pterodaustro is known from embryos to fully mature individuals
Codornú et al. (2013) report on 300+ individual specimens from a single bone bed: “Interestingly, proxies for full skeletal maturation are thus far present only in isolated elements (i.e., all complete or semicomplete specimens belong to osteologically immature individuals). These proxies include the complete fusion (lack of any sutural evidence) between the extensor tendon process and the shaft of the first wing phalanx, the complete fusion between the tibia and the proximal tarsals, and the fused distal secondary ossification centers of the humerus.” Note they did not report fusion of the scapula and coracoid. That’s because Pterodaustro nests in a clade (Fig. 1) that does not fuse the scapulocoracoid.
So what’s the pattern?
Basal pterosaurs do not have a fused scapulocoracoid. Dimorphodon may have a fused s/c. Campyognathoides and basal Dorygnathus fuse the s/c. Basal Rhamphorhynchus specimens are smaller and lack fusion. Derived Rhamphorhynchus regain fusion. Dorygnathid pre-azhdarchids beginning with tiny TM 10341 lose fusion. Large azhdarchids regain fusion. No ctenochasmatid or dorygnathid pre-ctenochasmatid fuse the scapulocoracoid. Jianchangnathus and all subsequent scaphognathids lose fusion. Basal ornithocheirds, no matter how large their wings are do not fuse the s/c. Certain, but not all derived ornithocheirds regain fusion. On another branch of scaphognathids, certain germanodactylids regain fusion. Shenzhoupterids and basal tapejarids lose fusion. Derived tapejarids, the big ones, regain fusion. (Does anyone have a good dsungaripterid scapulocoracoid? I haven’t seen one yet.) Germanodactylids including Pteranodon have fusion (not sure about basal taxa because so many are known just by skulls), but eopteranodontids and nyctosaurs lack scapulocoracoid fusion.
A little pterosaur referred to Eudimorphodon, BsP 1994 has a fused s/c. Arthurdactylus a much larger, longer winged ornithocheirid, does nto fuse the s/c. So size is not the issue.
All known pterosaur embryos come from clades that do not fuse the scapulocoracoid. However, the juvenile Pteranodon has a fused s/c.
Addendum
Once a clade began to fuse the s/c, then lack of fusion generally accompanied phylogenetic size reductions. Among azhdarchids, only Quetzalcoatlus fuses the s/c. This includes a smaller Pteranodon YPM2525 which may also represent a size reduction shown here.
Among the derived ornithocheirds you do get a more randomized on-off-on-off pattern.
So there you have it. All results subject to change with injections of new valid data.
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
Maisano JA 2002. The potential utility of postnatal skeletal developmental patterns in squamate phylogenetics. Journal of Vertebrate Paleontology 22:82A.
Maisano JA 2002. Terminal fusions of skeletal elements as indicators of maturity in squamates. Journal of Vertebrae Paleontology 22: 268–275.
The Pterosaur Heresies 
I don’t think this pattern passes the Occam’s Razor test. Multiple reversals of a trait that is well-supported as an ontogenetic character in other clades? Or actually being an ontongenic characteristic? I know which one seems more likely. Applying the idea of adaptive advantage to this scenario, I cannot see an adaptive advantage to having the flight-critial scapula and coracoid unfused where a fused unit would be structurally stronger without gaining weight.
Rob, remember that pterosaurs grow isometrically. We learn that from embryo pterosaurs that are virtually identical to adults. What are the odds that only the green areas represent immature taxa? [Very high.] Please see Maisano 2002. She reports, lizards grow with fused bones or never fuse their bones into old age. They grow differently than archosaurs. One more reason (as if we needed one) to understand that pterosaurs are indeed lizards.
I disagree that pterosaurs exhibit isometric growth, but that is neither here nor there. My question rests on the adaptive advantage of a flight-critical component being less rigid than would be expected AND the idea that this reversal occurred so many times within related clades.
Is there a corresponding phylogenetic signal using the traditional taxonomy?
If you disagree that pterosaurs exhibit isometric growth, please provide two related pterosaurs separated only by ontogeny that demonstrate what you propose. The embryo Pterodaustro, the JZMP embryo (ornithocheirid) and the juvenile Pteranodon all have adult proportions, especially in the skull. The IVPP embryo is an anurognathid, not an ornithocheirid as originally thought. That’s why it has a short snout.
I’ve just updated this post. Some small pterosaurs (BsP 1994) fuse the s/c while other much larger ones (Arthurdactylus) do not. With regard to traditional taxonomy (analyses by other workers) clade -wide fusion patterns have not been published to my knowledge. Nevertheless traditional workers support Bennett’s idea that lack of fusion demonstrates immaturity, an idea that is tested and falsified here in this post. It’s a paradigm, a tradition, a preconception. You can test it yourself.
Of the taxa you list here, 2 unfused s/c joints fall outside of pterosauria, 23 fall as basal to their clade and the remaining 42 unfused s/c fall out as derived. The 81 fused s/c joints outweigh the 65 total unfused in overall numbers. 42 of those are basal and 39 are derived. Running a Fisher’s Exact Test on this yields a P=0.0650, which is not statistically significant. You can’t support your conclusion based on your data presented here, even if your phylogeny is correct.
Isometric growth – I don’t think your reconstructions are as accurate as you think they are. I don’t think the data we have allows us to draw much in the way of conclusions for growth in pterosaurs. Some lepidosaurs show both allometric and isometric growth within a species (or individual, depending on the compared elements), some show only isometric. A sample of 4 pterosaur embryos (which may not be accurately portrayed) is not enough to draw a sweeping conclusion on the life history of every taxa within a clade.
My question about fusion patterns basically asks you to overlay your data on fusion on a traditional tree to see what the distribution would look like. Would you see the same scattershot pattern or would there be something else there?
You still haven’t addressed the point about adaptive advantage. Both birds and bats (huge phylogenetic difference) fuse the flight-critical pectoral girdle. Why would pterosaurs not fuse theirs?
Rob, as you know, it’s difficult to answer “why-why not” questions. For this mystery I can only answer “is-is not” -type questions.
It’s widely accepted that pterosaur hatchlings flew from the earliest moments or shortly thereafter. So, we have pterosaur flyers among all sizes of pterosaurs without fused scap/coracoids, including forms with the largest wing/torso ratios, like nyctosaurs and basal ornithocheirids.
Which traditional tree would you like to use? Ctenochasmatids, anurognathids, scaphognathids, cycnorhamphids and eopteranodontids and nyctosaurs, among many other clades will still have non-fused scap/coracoids no matter which tree or taxon list is used.
Remember also, pterosaurs come from a completely extinct clade of lepidosaurs outside the Squamata. So modern forms provide insight, but direct analogs may be asking too much. Huehuecuetzpalli is among the basal forms in this extinct clade in which two ontogenetic sizes are known and Reynoso (1989), who described them, found no allometric differences between the two. I know of no other juniors in this lineage. The two sizes of Tanystropheus and the several specimens in several sizes of Macrocnemus we know are not conspecific. I hope this helps. Flight critical or not doesn’t seem to matter with regard to scap/coracoid fusion in small or large pteros no matter how logical it might be to do so.
The “what is the adaptive advantage” question can certainly be asked and an answer attempted. It can even be framed as an “is-is not” question. “The adaptive advantage is _______,” and then search for evidence that would disprove your hypothesis.
I am not denying that some of these animals flew with unfused s/c – they must have unless even large pterosaur specimens we have now were flightless (I guess that’s a possibility). I am saying, however, that a fused s/c makes more sense for an adult animal and that, all other things being correct, what you see here is preservational bias and immature individuals. You stated, “Scapulocoracoid fusion is on again, off again in patterns that are not the random pattern one would expect if ontogenetic in nature.” I tested this hypothesis statistically with the Fisher Exact Test – it said that the distribution was not significant – essentially it was random, which is what you predicted for an ontongenic pattern. These data don’t pass your own “is-is not” test on the limited criteria you have set out, regardless of any other critique or questions I have.
re: traditional phylogeny. Obviously the fused/unfused forms wouldn’t change their fusion. I am curious if this character would follow the same scattershot pattern across the clade or if it would be more restricted.
re: lepidosaur growth. I don’t know enough about non-archosaurs to be able to address this cogently, but based on sections I am familiar with I think there are problems with the tree (which I have addressed elsewhere, as have others). Because of this I do not accept your hypothesis that pterosaurs are lepidosaurs, so I don’t think using fossil members of this clade to inform us about pterosaurs is useful.
Then which archosaur is closest to pterosaurs? I’ve seen parasuchians. I’ve seen Scleromochlus, both of which are easily dismissed (covered earlier). Your Fischer test appears to test the clade as a whole and gets an average for the Pterosauria. Now look at individual pterosaur clades (listed above). Take a look at the taxa preceding the Pterosauria in ReptileEvolution.com and you’ll see a gradual accumulation of pterosaurian traits. They did not appear de novo. Their ancestry can be traced back as far as you like.
The Fisher Exact Test checks to see if there is a correlation between two states in two groups. It can be used with very small groups as well as with larger two-state conditions.
I realized that I misinterpreted your graph. I thought green highlighted clades lacked fusion, not had fused s/c joints. I’d like to rerun the test. It would also be useful to highlight taxa that lack preserved scapulacoracoids.
Yes, there’s some slop in the graphic as some taxa do not preserve the s/c, but are known from skulls only. I’ll see what I can do to clean that up.
Okay, I reran the test using only applicable data. Here’s what I did:
I excluded embryonic (n=3) and private (n=2) specimens for a total of 137 pterosaurs for which we have data on scapulacoracoid fusion. Some sections of the tree that were not broken into clades seemed inconsistent with areas where similar chunks of taxa were broken into named/colored clades. I “created” three new clades to be able to better classify a given taxon as basal or derived. The three new clades were No13 + Nyctosaurus KJ2, Nemicolopterus + Tupuxura Goshura, and Coloborynchus + Anhanguera.
I then collapsed monophyletic taxa with the same fused/unfused AND basal/derived states. This resulted in the collapse of Pteranodon, Nyctosaurus, and Scaphognathus to one data point each. Pterodactylus was collapsed to one point except the outlying (on your tree) P. macronyx. Ramphorynchus collapsed to two data points to reflect the two fusion states. This resulted in 109 organisms. Breaking it down further there were 79 unfused taxa to 30 fused taxa. The unfused taxa had a distribution of 30 basal and 48 derived. The fused taxa were split 50/50.
Running these data through a Fisher Exact Test I got a P value of 0.2861. This is even LESS significant than before.
In conclusion, the distribution of s/c fusion across the pterosaurs in your tree is not correlated with phylogenetic position. It follows that s/c fusion cannot be used to support this tree, regardless of what ancestor they arose from. The distribution of these traits is random across this tree, a condition you yourself stated would invalidate your hypothesis on the utility of this character.
Rob, On the face of it your results don’t make sense. You can see large swathes of taxa without fusion and other large swathes with fusion. If there were sprinkles of mature and immature specimens in those areas, they would appear like pepper on salt, but they don’t, except in the ornithocheirids. Do you see the little boxes at certain nodes? Those are points of change. There are very few of them. Your using numbers as numbers without seeing the patterns. It could not be more simple. You’re reducing a color wheel to the gray all the colors combined make. Collapsing various specimens into single data points that are arbitrarily defined by names that often don’t make sense (Eosipterus in the midst of many Germanodactylus, for instance) also removes data to further your preconceived ends. Just look at the image. See the fusion here and the non-fusion there. It’s not that hard.
“Collapsing various specimens into single data points that are arbitrarily defined by names that often don’t make sense (Eosipterus in the midst of many Germanodactylus, for instance) also removes data to further your preconceived ends”
I’m not sure where you got the idea that I collapsed Germanodactylus. I eliminated the private Germanodactylus from my data and didn’t collapse the remaining two because Eosipterus was in between. I’m not here to judge what taxa are valid and which ones aren’t. If your tree shows one genus split by a second genus it didn’t get collapsed. The names may be arbitrary but I was actually trying to preserve as much information as possible while eliminating “noise” (in this case, particularly well-represented taxa being scored on an individual basis and giving each individual the same weight as a taxon known from only one total individual). If someone saw fit to distinguish between that organisms and the dozens of previously known that is for them. My quibble here isn’t with your phylogeny per se, it is with your conclusion that there is a phylogenetic signal in the fusion of the scapulacoracoid in the tree you have laid out. Which brings me to…
“Your using numbers as numbers without seeing the patterns. It could not be more simple…Just look at the image. See the fusion here and the non-fusion there. It’s not that hard.”
Actually it IS hard and the point of the numbers IS to find patterns. That’s what this statistical test does – finds if the correlation between two states in a data set is significant (and I use that term in the scientific sense). If the statistical significance is high then the “pattern” one sees probably is a real pattern. If the significance is low then the “pattern” is probably not an actual pattern but the mind playing “connect the dots” between data points. You say you saw a pattern there. I used a well-established tool (FET) to test this pattern. The test came back negative. Seeing something with your eyes is, by definition, a qualitative measurement. The numbers are quantitative. Our eyes (and brain) have evolved to pick out patterns, even when none exist. That’s why test like this were invented. Which leads to…
“points of change”
Yes, but I think you have misplaced some (will double check) and with so many incomplete specimens your number of reversals can easily double. Finally…
Methodological critique
“You’re reducing a color wheel to the gray all the colors combined make…also removes data to further your preconceived ends.”
I appreciate this point and you are correct that I do have preconceived notions. Everyone does. I have stated, from the outset, that I disagree with your interpretation of this character distribution. I don’t let this color my methodology. I collapsed both unfused derived and fused basal groups based on prexisting taxonomy, not if I thought something was valid. My three “new” clades that I used to define derived vs. basal were a mix. One clade was completely unfused, one was mainly fused (where data was known), and one was about 50/50 (where data existed). I eliminated privately-held specimens (since they aren’t available for review by other scientists) – and in a peer-reviewed published paper they wouldn’t be in there anyway. I don’t see how any of that skews the data “my way.” I eliminated the embryos since I think you have not reconstructed them correctly (by interpreting based on too little evidence). I don’t necessarily think that is biased (since it isn’t out there in the open literature, just like the private specimens) but when n(embryo)=3 and n(tested)=137 or 109 the skew isn’t going to be very large anyway even if it is flagrant bias.
To that point I began redoing the FET based on just what you have listed. I kind of suspected you would have that critique. I don’t necessarily agree with it but I also don’t mind seeing if I am wrong. So, no collapsed taxa, no new clades, all specimens with recorded data (per your tree). I didn’t get finished recoding the data when I have to leave but I’ll post again tomorrow when I finish that analysis with n=142.
I appreciate your efforts, Rob, but I see things and judge things on appearances. I see pictures. I don’t see a checkerboard of fusion/nonfusion on my tree (except in the derived ornithocheirids). With regard to numbers, my only experience is I’ve seen Bennett’s statistical studies from 20 years ago and they miss the point. I’m sorry, I can’t judge your work adequately.
“I see and judge things on appearence” is not good scientific method. Thats the whole point of statistics. Just eyballing something will lead to you either seeing things that arn’t there or missing things that are. Many is the time when I’ve seen curves that look correlated but have actually been shown not to be. Numbers are important. They can’t be ignored. It is the point I was trying to make when I suggested you include Bremer support of bootstrapping.
And then there’s that old cliche, “lies, damn lies and statistics.” Nothing personal. All due respect. Neil (and Rob), you might try your numbers on just the anurognathids or just the ctenochasmatids to see what you get. And then you’ll see my point. There is also the pattern that often, but not always develops, that phylogenetic size reduction is the point at which non-fusion takes hold (the origin of Rhamphorhynchus, Scaphognathus, the eopteranodon/nyctosaur clade, to name a few). And fusion kicks back in with size increase (larger Rhamphs, Quetz., Campylognathoides, Anhanguera, etc.), but not always (Arthurdactylus, Zhejiangopterus, etc. ). With regard to Bremer support, it’s been awhile, but the numbers >were< strong in all my trees with weaknesses developing when skull only meet skull-less taxa and other very incomplete specimens.
If you’re going to dismiss statistics like that, then how are you going to answer your critic when s/he says “I just don’t see the pattern you say you see.” In cases like that, the answer “I see it so it must be there” isn’t going to be accepted. The burden of proof is on the one who says s/he sees the pattern. Statistics aren’t perfect, but statistical support is better than the argument of “I see it.” (btw I’ll leave the analyses to Rob; I’m no expert on pterosaurs, and I actually wasn’t arguing against your post, more your dismissal of the statistical evidence).
oh and your Bremer support was most certainly not strong in many areas (although, to be fair and balanced, I will say the pterosaur portion was surprisingly well supported, as were the sphenodon/rhynchocephalian and the turtle bit!). I endeavoured to demonstrate that by collapsing all the nodes of low support. In your post you dismissed me doing that, saying it achieved nothing beyond loss of resolution. But it did achieve something, and the loss of resolution was the point. When you collapse nodes of low support (1 or 2), you get nothing more than a bush over much of the tree, particularly the areas of parareptile and synapsid relationships. This is standard practice (at least it used to and should be; I’m disappointed to see many authors aren’t doing it any more); you should only speak with certainty on nodes that are well supported; those that are not you should collapse, or at the very least temper your statements with caution, and say more work/characters/specimens are needed.
Neil, tell me the best way to collapse a node of low support using PAUP or MacClade, the two programs I have. Right now I simply let PAUP find the best tree. And, to be fair, we’re talking about the pterosaur tree, not the reptile tree at the moment.
With regard to bushy trees, it’s my job to find traits that divide a bush into discrete branches. Bushy trees are not illuminating. If I find loss of resolution, it’s often due to a mistake in scoring judgement. Corrections often add resolution. (But see below).
What I have done, in the past, is eliminate a few to many dozen taxa to see how the rest of the tree holds together. That’s how I was able to nest turtles to pterosaurs and both with early sauropterygians in an earlier post, just to demonstrate the point that pterosaurs, having lost over 150 closest sisters will -still- not nest with archosaurs.
Granted the lepisauromorph basal portion of the tree has certain areas of weak support around the clade that includes caseasaurs and millerettids. On the flip side, there is no other place these taxa nest AND they do nest together, just the exact order is under question right now. More basal taxa and more post-crania would help. Their interrelationships are what causes the low Bremer scores. They are, perhaps, too much like each other and too often too few parts are known. As an example, Colobomycter is known from a partial skull and it causes loss of resolution where it nests, next to a skull-less taxon, Adelosaurus. So please, don’t take loss of resolution as an indictment without looking into the cause for the loss of resolution.
I trust you have also put the iron to other tree makers who have results in the range of 100+ to 500,000+ MPTs based on more complete specimens, some of which include taxa that do not belong and exclude other taxa that need to be in there (which is my #1 rant and the reason for producing ReptileEvoution.com).
Unfotunately I’ve never used either of them, I’m a TNT user. Although you should be able to do bootstrapping with paup.
All you said about the reasons for loss of resolution are true, but the point is these comments need to be added in your posts as words of caution whenever you suggest a poorly supported relationship. I for one have no objection to the suggestion of a new controvertial set of relationships, I find it quite exciting, and I even think some those you propose are definitely worth further examination e.g. fluffy sharovipteryx, turtles with Stephanospondylus. Its more the tone I object to, it lacks caution which should always accompany a new idea. But it is your blog, so I suppose it’s your right.
And I would put the iron on them if they had blogs ;) the perils of taking your ideas to the interweb. Well for taxon exclusion at least, I think some taxon lists and outgroup taxa used are ridiculous (character lists as well, which is my main problem with your tree). I see no objection to presenting a poorly resolved tree, as you say it depends on the reasons for resolution; as Benson (2012) showed, hundreds of most parsimonious trees can be created just by 3 unstable taxa. remove them and its pretty well resolved. A poorly resolved tree is not wrong, it just needs more work.
Update on the analysis. I have 142 organisms scored as either basal or derived and fused or unfused. I did not collapse any taxa.
Unfused organisms: 66 derived, 33 basal
Fused organisms: 27 derived, 16 basal
The results become even less significant with P=.7027. I think the significance decreased since some of the things I did to reduce noise were taken out. Anyway, no matter how you cut it there doesn’t seem to be a phylogenetic pattern to fusion in your pterosaur tree. In this case I would argue (again) that this character does not support your phylogeny.
That isn’t to say that I am right about ontogeny being the driving force behind the fusion. I have had a couple people suggest to me that I might be wrong in that regard (and I’ll admit that). If fusion does not have an ontonogenic signal so be it. If pterosaurs are lepidosaurs and they grew isometrically without fusing their shoulders (or grew with fully fused elements), so be it. That is decoupled from what the data is showing about the claim of a phylogenetic signal.
What would be a significant number?
Getting back to patterns, Rob, Let’s take a trip throught the chart, which has been slightly updated since you last saw it, as I dived back into the data making sure it was as correct as possible:
OFF
1. 20 Basal lizards and basal pterosaurs in the chart, unfused s/c.
That’s an unbroken chain of 16 pre-pterosaurs and 4 pterosaurs. Does that mean those four basal pterosaurs were immature?
ON
2. 3 derived drepanosaurs: fused s/c. Does that mean only those three forms were mature?
ON
3. 10 specimens of Dimorphodon/ Preondactylus/ Peteinosaurus/ Eudimorphodon/ Campylognathoides: fused s/c. Are all ten in two rows mature?
OFF
4. 12 pre-anurognathids and anurothathids: unfused s/c. Are all twelve in a row immature? (Certainly the IVPP embryo is.)
OFF
5. 8 basal Rhamphorhynchus: unfused s/c. Are all 8 in a row immature?
ON
6. 11 derived Rhamphorhynchus: fused s/c. Are all 11 in a row mature?
ON/OFF
7. Then we come to three small pterosaurs (BSP 1994, Changchengopterus, Sordes), some with fused, others unfused s/c. Apparently a transtional phase as all three were smaller than E. ranzii.
ON
8. 8 basal Dorygnathus, including 4 pterorhynchids (aka: darwinopterids): fused s/c. These may be continued from the first wave of fused pterosaurs listed above in #3. Are all eight mature?
OFF
9. At least 11 derived dorygnathids/pre-azhdarchids: unfused s/c. Are all eleven in a row immature?
OFF
10. At least 13 derived dorygnathids/ctenochasmatids: unfused s/c. Are all thirteen in a row immature? We have a complete life history of Pterodaustro, from embryo to the biggest, 300 specimens, without a single fused s/c. Are the biggest specimens in your opinion still immature?
OFF
11. 35 in two rows: At least 17 taxa from Jianchangnathus to Arthurdactylus AND at least 18 others from Sos4006 to Eosipterus all with unfused s/c: Are all 35 in two rows immature?
ON/OFF
12. It gets more random between Coloborhynchus and Anhanguera with half the taxa known from crania only.
ON
13. In at least 9 taxa fusion returns from the BMM Germanodactylus to Pteranodon.
OFF
14. In 5 taxa from Nemicolopterus to Huaxiapterus: unfused s/c.
ON
15. In at least the next 2 taxa, both Tupuxuara, fusion returns.
OFF
16. In 10 eopteranodontid/nyctosaurs: unfused s/c. Are all these ten immature?
Don’t you think it’s odd that so many clades are represented only by immature or only by mature pterosaurs? Don’t you think it’s odd that out of 200 pterosaurs, the fusion switch only goes on about 8 times?
And finally, as a reminder, the immature Pteranodon is only a quarter the size of the adult, yet has a fused s/c, so what do we make of that? This is the phylogenetic pattern. I’m sorry you don’t see it. I don’t see this as random, but I do see this as heavily skewed toward adult forms, no matter their size or fusion patterns.
“What would be a significant number?”
Anything smaller than 0.05 would be significant. Anything smaller than 0.01 would be considered to be very significant. The closer to 1.0 the number gets, the less correlation there is.
“Getting back to patterns, Rob, Let’s take a trip throught the chart, which has been slightly updated since you last saw it, as I dived back into the data making sure it was as correct as possible:”
What changes did you make? I can update my codes on the computer here. I’d prefer not to have to recode everything again by hand.
“Don’t you think it’s odd that so many clades are represented only by immature or only by mature pterosaurs?…This is the phylogenetic pattern. I’m sorry you don’t see it. I don’t see this as random, but I do see this as heavily skewed toward adult forms, no matter their size or fusion patterns.”
Again, it is possible that I am wrong about fusion=maturity. If so, I admit that. I am not arguing against your data (now) because I think that I am right about maturity and fusion. I am looking at your data from the viewpoint of testing your hypothesis. Your hypothesis was that scapulacoracoid fusion was phylogenetically important – it carried a signal of relationships. I understand your point, even – there are runs of fusion or nonfusion that appear to be patterns. A rigorous analysis (“test, test, test again”) disproves this hypothesis. I think you are mistaking these fused/unfused runs for something other than just that – runs. Flip a penny 142 times and you will get runs of heads and tails, but the runs are meaningless.
You are arguing that the fusion is a shared trait among groups but the occurrence in your tree, based on the statistical analysis, is virtually certain to be random. As we increased resolution our P value got closer to 1.0 (though I think we also increased our noise as well).
Please note, and this is important for me to say, this has 0 bearing on whether your tree is correct or whether fusion carries an ontonogenic signal. That is not what I am testing. I am not looking to see if I am “right.” I am looking to see if you are right about this one trait (s/c fusion) having phylogenetic significance.
Related note: You have claimed that many traditional pterosaur/archosaur/phylogeneticist workers reject your work for a variety of reasons and that you have been blacklisted from publishing some of this stuff (and that’s why you have the site and blog, to spread the word). I hope this doesn’t come off the wrong way but your refusal to understand the significance of a well-established biostatistical test (first published in 1922) on your data doesn’t reflect well on you. The fact that you aren’t doing tests like this yourself on your own data, to make sure your conclusions are supported, is not likely to help your case either, and may be why some workers don’t take this blog (or your site) seriously.
Then there’s nothing more to say. Thank you for your thoughts and experiments.