I promised John Harshman for several months that I would start a discussion about common design vs. common descent, and I’d like to keep my word to him as best as possible.
Strictly the speaking common design and common descent aren’t mutually exclusive, but if one invokes the possibility of recent special creation of all life, the two being mutually exclusive would be inevitable.
If one believes in a young fossil record (YFR) and thus likely believes life is young and therefore recently created, then one is a Young Life Creationist (YLC). YEC (young earth creationists) are automatically YLCs but there are a few YLCs who believe the Earth is old. So evidence in favor of YFR is evidence in favor of common design over common descent.
One can assume for the sake of argument the mainstream geological timelines of billions of years on planet Earth. If that is the case, special creation would have to happen likely in a progressive manner. I believe Stephen Meyer and many of the original ID proponents like Walter Bradley were progressive creationists.
Since I think there is promising evidence for YFR, I don’t think too much about common design vs. common descent. If the Earth is old, but the fossil record is young, as far as I’m concerned the nested hierarchical patterns of similarity are due to common design.
That said, for the sake of this discussion I will assume the fossil record is old. But even under that assumption, I don’t see how phylogenetics solves the problem of orphan features found distributed in the nested hierarchical patterns of similarity. I should point out, there is an important distinction between taxonomic nested hierarchies and phylogenetic nested hierarchies. The nested hierarchies I refer to are taxonomic, not phylogenetic. Phylogeneticsits insist the phylogenetic trees are good explanations for the taxonomic “trees”, but it doesn’t look that way to me at all. I find it revolting to think giraffes, apes, birds and turtles are under the Sarcopterygii clade (which looks more like a coelacanth).
Phylogeny is a nice superficial explanation for the pattern of taxonomic nested hierarchy in sets of proteins, DNA, whatever so long as a feature is actually shared among the creatures. That all breaks down however when we have orphan features that are not shared by sets of creatures.
The orphan features most evident to me are those associated with Eukaryotes. Phylogeny doesn’t do a good job of accounting for those. In fact, to assume common ancestry in that case, “poof” or some unknown mechanism is indicated. If the mechanism is unknown, then why claim universal common ancestry is a fact? Wouldn’t “we don’t know for sure, but we believe” be a more accurate statement of the state of affairs rather than saying “universal common ancestry is fact.”
So whenever orphan features sort of poof into existence, that suggests to me the patterns of nested hierarchy are explained better by common design. In fact there are lots of orphan features that define major groups of creatures. Off the top of my head, eukaryotes are divided into unicellular and multicellular creatures. There are vetebrates and a variety of invertebrates. Mammals have the orphan feature of mammary glands. The list could go on and on for orphan features and the groups they define. Now I use the phrase “orphan features” because I’m not comfortable using formal terms like autapomorphy or whatever. I actually don’t know what would be a good phrase.
So whenever I see an orphan feature that isn’t readily evolvable (like say a nervous system), I presume God did it, and therefore the similarities among creatures that have different orphan features is a the result of miraculous common design not ordinary common descent.
petrushka,
That is indeed the case. Taking an example of the amphipathic alpha helix I’ve mentioned before, there is an asymmetry around the axis of the helix of hydrophilic and hydrophobic residues, which leads to the helix having an affinity for membranes on the one side, cytosol on the other. Arrange a bundle of such helixes in a ring and you get a pore.
This has an enormous number of degrees of freedom for change, within that basic constraint. People who say different don’t know their protein chemistry.
Sal continues to play with his shiny new programs and continues to have no idea what he’s doing with them. But he’s sure that if he perseveres he will receive much cargo.
Point out to the reader please where I said anything otherwise!!!!!!!!!!!!!!!!!!!!!!!!!
Did you even friggin even look at the triangle diagram I provided and the distance matrix?
You will have to explain what your point was, because I don’t think anyone got it. What you think is obvious is seldom obvious to anyone else. Please exercise more care in posting. Try to imagine that your audience may be unlike you in some respects and may not follow vague hints in the same direction you would.
Quoting myself from a previous comment:
Somehow that caused Rumraket to insinuate I said something I never said:
Clearly you can see then if I said “species 10 vs 24: 47.6%” and “species 21 vs 10 : 4.2%” then that means the magnitude of difference of “species 10 vs 24: 47.6%” is 11.33 times bigger than the difference of “species 21 vs 10 : 4.2%.” Which means species 10 and 21 are more closely related than 10 and 24.
Rumraket insinuates I said something otherwise. I didn’t.
Is it the evolutionary “poofs” that are being mimicked or the evolutionary “it just happened, no reason, that’s all”?
Taht’s right. No level of dissimilarity is enough to warrant an inference of no common ancestry. Similarity indicates common ancestry, but absence of similarity does not mean absence of common ancestry. How convenient.
stcordova,
Again, what point were you trying to make? Were you trying to say that all evolution involves very good molecular clocks?
Look, if you didn’t mean to imply what I took you to, then I’m sorry. But then it isn’t clear what point, if any, you were making at all.
At the very least it looks like you’re trying to say there’s something inherently wrong with phylogenetics, so since your post talked about distance measures, what other point should one take from it?
Mung,
The pattern of common descent, of course. You know what this thread’s about?
First off I confirmed your phylogeny for Gene1. There are two separate diagrams generated by the same algorithm, just different ways of rendering the branches:
The unrooted version:
http://theskepticalzone.com/wp/wp-content/uploads/2017/11/rumraket_1.png
The classical tree version:
http://theskepticalzone.com/wp/wp-content/uploads/2017/11/mid_rooted_rumraket1.png
They appear to agree with your tree and approximate the actual evolutionary history. Is that right?
Yes, I agree.
Thanks. Sorry I lost my temper earlier.
But to the point I was making.
John Harshman said:
To which I strenuosly diagreed. He apparently objected to my use of distance measures. Distance measure would be better termed as “dis-similarity measure”. Every algorithm may or may not compute dis-similarity the same way, most notably the problem of handling in-dels. Some substitutions may be weighted more than others, etc. But there is generally the idea something resembles something else minus some dis-similarity, and the dis-similarity is meant to approximate how much something has evolved differently from the something else that is also evolving.
Now, there are variety of techniques to build phylogenetic trees. Even with Maximum Likelyhood trees, the default in MEGA 6.0 is to start with a Neighbor Joining tree. This is a description of a Neighbor Joining tree that disagrees with what John Harshman said:
https://en.wikipedia.org/wiki/Neighbor_joining
I’m complaining that John Harshman is making the most uncharitable interpreation of what I’ve said.
So, now in light of the Neighbor Joining algorithm which is used as a default in Mega 6.0 (and other algorithms) as a starting point for Maximum Likelihood, it appears distance matrices are rather important in phylogeny, and they are not irrelevant as John claims. I could be nasty, and suggest he revisit some of his training in phylogenetics in light of his apparently un-informed looking comments, but I’ll play nice for now.
I provided the distance matrix earlier for your Gene1:
http://theskepticalzone.com/wp/wp-content/uploads/2017/11/rumraket1_distances.txt
Sal, your arrogance is still a problem. I never said that distance measures were improper. I said that your determination of relationships by raw distance measures was improper, because it requires the assumption of a very good molecular clock. UPGMA assumes such a clock, but neighbor joining does not. Neighbor joining produces a fairly good approximation of a least-squares best fit tree. Now in fact I would argue that a maximum likelihood algorithm is better than a least-squares algorithm, and a least-squares algorithm is better than neighbor joining, but that’s a separate question. The important thing is that the incomprehension is yours alone. You don’t know what you’re doing and you don’t understand criticisms of what you’re doing, which is cargo cult science. By the way, announcing that you could have not played nice isn’t playing nice; it’s a fig leaf to hide an insult.
Oh, and you still haven’t explained what message you meant to convey with your triangle thingie.
So we take a hypothetical tree and see how well it agrees with the data. We can, in principle find a tree that best fits the data. It’s a curve fit of sorts.
Which ever way one describes the process of building such a tree, it’s absurd to suggest, as John did, the distances are irrelevant. John is just going into “I’ll disagree with everything Sal says, even if he agrees with me” mode.
Your cluelessness prevents you from realizing that you are indulging in a quote mine of me. And after I explained what I meant!
Hey, Joe cites me in that book. See if you can find the reference.
All different sorts of algorithms and good and bad reasons for all of them. Cargo cult science indeed.
I think it’s about the evidence that we normally take as evidence of common ancestry really should be interpreted as evidence for common design. But I haven’t really seen much in the way of argument to that effect. I’ve often wondered if I am in the right thread.
ETA: but if common descent can produce this pattern why do the designers need to mimic anything?
That’s like saying it’s cargo cult science because there are many different types of microscopes and good and bad reasons for using any of them for a particular job.
No, that doens’t make it cargo cult science. You’re proving you don’t even know what the “cargo cult science” term even means.
Which is faster, a clock, or freedom?
The distance measures I provided are not “raw”, they were the ones computed according the requirements of the particular tree being built. If the ML tree was constructed starting with an NJ tree with the Jones-Taylor-Thornton model, I provided the Jones-Taylor-Thornton distance matrix for discussion.
The point I was trying to show with Rumraket’s Gene1 is that species with low distance differences are assigned the same clades to the exclusion of other clades. The problem with the Lungfish-Coelecanth-Tetrpod clade is the distance measures are all wrong. Assume for the sake of argument, common descent is true, the prevailing views of the Sarcopterygiian clade does not agree with the distance matrices. The only way to make it “agree” is to make ad hoc rationalizations and some dubious manipulation of adjustable parameters to arrive at a foregone conclusion. I showed how easy it is to get different phylogenetic tree by simply adjusting the adjustable parmaters.
On the other hand, there is a clocking issue that needs even more examination and which Hoyle and Denton clearly saw.
Cicada’s reproduce once every 17 years, fruitflies manytimes a year. Bacteria reproduce even faster compared to sturgeons and humans. The patterns in the distance matrix do not accord with random mutations unless the mutations are finely tuned to adjust the mutation rates with reproduction rates.
So again, Universal Common Ancestry needs a miracle of fine tuning to make it agree with the extant data. If evolution needs miracles to make the theory work, then it’s really not fundamentally different from creation, it only pretends to be.
That’s quite a loaded statement out of mere frustration Sal. For one, universal common ancestry and evolution are not one and the same. Universal common ancestry is what we would expect if all life today (at least the life forms in your sample), ultimately evolved from a single common ancestor (or population of common ancestors). It would be part of the evolutionary history of the life forms we examine, but not a necessity for evolutionary theory. If universal common ancestry was false, evolutionary theory would stand as strong, because the theory is about the phenomenon of evolution, not about the particular history of our planet (with two or more lineages or just one).
For another, your frustration with the parameters that you might need to take into account doesn’t mean that evolution as a theory itself, requires a miracle. It only means that the construction of trees and discovery of relative phylogenetic relationships can be quite a challenging task. The challenge might vary in strength depending on the kind of data you have available. if the data contains highly conserved proteins, for example, the number of informative position in the data will be pretty small. Thus, it won’t work unless you’re analyzing a very small number of organisms, far enough from each other that the number of sites gives you enough information about their phylogenetic relationships. Etc.
I doubt that the necessity of making scientifically informed choices and being challenged by the data and the methods makes evolution into a myth.
Have a nice weekend.
Mung,
You’re not the only one to observe that this nettle has not been grasped by defenders of ‘common design’.
Two basic approaches have been
1) I would not expect multiple nested hierarchies from a process of descent.
2) I would expect multiple nested hierarchies from a process of design.
Which is nice, but I would and wouldn’t respectively ***
If they use descent, inserting little changes as if they were – ahem – ‘random’ mutations, that’s possible. The stance I mock is that there is a discontinuity – that organisms exist, quite recently derived in the taxonomic scale, with no parents. Standard Creationism, in other words. Because this broad-scale, whole-genome discontinuity is not manifest in genomes; the less controversially Common-Descent pattern of the recent is actually there at all scales.
Therefore, the Designer has mocked up Common Descent, in the genomes He designed from scratch. A bit like creating light in mid transit. Or First Couples with Ne’s in the thousands.
*** [eta – just thought of a #3 – whatever it might look like, it can’t be that because that’s impossible.]
stcordova,
I think the bigger problem at this point is that you are making sweeping statements like that on the basis of one mitochondrial gene under strong purifying selection and a period of hundreds of millions of years. There are numerous confounding factors that could influence analysis. More care required.
Allan Miller,
I need to learn to be as succinct as you.
Entropy,
Haha! It is a gift that visits only occasionally!
Thanks. You too.
You still misunderstand my meaning. By “raw” I mean distances independent of phylogenetic analysis. You can’t, in other words, read phylogeny directly off distances.
Finally!
That’s true only if you assume a molecular clock, which you shouldn’t do.
No, you showed that one mitochondrial gene, Cox I, with a small and dubious taxon sample, analyzed in a dubious manner, gives weak and dubious results.
You will have to explain what you mean by that, because none of what you said above is clear. What “clocking issue”? What is a “clocking issue”?
ERRATA:
Regarding Jones-Taylor-Thornton and Dayoff distances in MEGA6.0 they aren’t percent differences, they are some other number indicating difference. MEGA6.0 usually give a reference to a paper describing what the distance number represent. So when I referred to Jones-Taylor-Thornton or Dayhoff as a percent difference, that was a mistake on my part.
Nevertheless, the greater the distance number the greater the dis-similarity. Imho, the word “dis-similarity” might have been a better choice of word than “distance”.
In fact, one of the MEGA6.0 distance metrics is “number of differences”!
Nothing materially changed however when I said the composition of the Sarcopterygii clade doesn’t agree with the sequence dis-similarities. The problem I pointed out remains, and Rumraket’s example helps illustrate the problem of forming clades that don’t agree with patterns of gene dis-similarity.
Did you even read any of my posts? When you make claims of this sort, you are implicitly assuming a molecular clock. What is your evidence? And I have previously pointed out the actual problem: you are taking a small and poorly designed taxon sample with a mitochondrially-encoded protein.
Less and less of them because you are more and more put on my ignore list. Aside from correcting me on Cox1, your input has been of increasing lower value relative say to reading things like what Joe Felsenstein says in his book and related papers.
Sorry. If you want to write for the sake of other readers, that’s fine, but don’t expect you’ll get too much attention from me. I’m finding I can learn better from other sources.
Nothing personal, good luck and Happy Thanksgiving. Thank you nevertheless for participating in my discussion.
Shorter Sal: “LA LA LA I DON’T WANNA SEE THE EVIDENCE!!”
Much easier to choose willful ignorance over data he’s been lying about to support his YEC stupidity.
Is the clock running?
stcordova,
Yes, many people are uncomfortable hearing unpleasant truths about themselves. I understand.
That would explain why John has me on ignore. 🙂
Depends on the observer.
And so this thread ends, all without ever getting to the title subject. RIP.
Allan Miller suggested, and something I agree with, to examine non-mitochondrial proteins. I’m considering two classes of them: bone morphogenetic proteins and topoisomerases. I don’t have time for those two right now, but it’s on my to do list.
I think the molecular clocking issue that Hoyle and Denton pointed out are inconsistent with random common descent because the mutation rates would have to be fine-tuned with reproduction time to get the distances — it is an unsolved problem with the apparent molecular clock.
As my protein set is expanded out, hopefully I can revisit this issue because common descent doesn’t explain such patterns except by the requisite fine tuning Denton and Hoyle pointed out. But if there is fine tuning of the clocks, this is evidence of common design.
Phylogenetic trees are constructed to minimize the number of changes in the evolutionary process. This is very evident for Gene trees.
The degree of changes is correlated with the distance metric. The distance metric is used to construct distance matrices.
There is one unfortunate circularity in seeing the patterns of similarity and diversity, the assumption of common descent in the first place. The nested hierarchical structure actually may be clearer without the evolutionary assumption. Some of the distance matrices have evolutionary weighting such as the Dayhoff and Jones-Taylor-Thornton model. Joe Felsenstein’s book discusses these.
But there is also a simple measure of differences. In MEGA 6.0 there is p-distance and count of distances.
Here is an interesting Matrix of Cox1 where the differences are counted simply. It highlights the astonishing pattern discussed by Denton and Hoyle.
These were the number of differences from Ciona Cox1 with a variety of other species.
The difference was always around
482.94 with a standard deviation of a mere 2.41!
Ciona_robusta_(tunicate)_YP_009357192.1
Neoceratodus_forsteri_(Australian_lungfish)_NP_387474.1 479
Lepidosiren_paradoxa_(South_American_lungfish)_NP_542458.1 482
Protopterus_annectens_(West_African_lungfish)_YP_006884099.1 481
Latimeria_chalumnae_(coelacanth)_BAF43538.1 486
Latimeria_menadoensis_(Menado_coelacanth)_YP_220643.2 487
Leucoraja_erinacea_(little_skate)_YP_004935520.1 483
Scoliodon_macrorhynchos_(Pacific_spadenose_shark)_YP_006460407.1 482
Xenopus_tropicalis_(tropical_clawed_frog)_YP_203372.1 486
Xenopus_victorianus_(Lake_Victoria_clawed_frog)_YP_006883487.1 486
Homo_sapiens_(human)_AEG23663.1 485
Macropus_giganteus_(eastern_gray_kangaroo)_YP_009154379.1 484
Thunnus_obesus_(bigeye_tuna)_YP_003587610.1 481
Katsuwonus_pelamis_(skipjack_tuna)_ADA69876.1 481
Paradactylodon_mustersi_(Paghman_mountain_salamander)_YP_626701.1 482
Acipenser_gueldenstaedtii_(Russian_sturgeon)_YP_002808661.1 481
Salmo_salar_(Atlantic_salmon)_NP_008447.1 480
Lethenteron_camtschaticum_(Arctic_Lamrpey)_AIR11970.1 484
Mean 482.9411764706
Stdev 2.4101501052
No, what you have described is what a parsimony or minimum evolution algorithm does. Other algorithms do different things.
What I believe that shows is saturation. These are not true distances, just a limit at which sequences are largely randomized with respect to Ciona. This is a problem using mitochondrial proteins at such great divergence.
By the way, that isn’t a distance matrix, just a list of distances, or a single row in a distance matrix.
Sal’s back to technobabbling-for-the-sake-of-pretending-there’s-a-genuine-debate.
Sal, one more thing: you still haven’t figured out what the subject of this thread is, which is ironic since you started it. Then again, it was intended as a response to a challenge I made, so perhaps I am better able to tell when that challenge hasn’t been met, or even attempted.
Even if the pattern you claim to see in COX I distances of vertebrates from Cialis were correct, and even if it required intelligently designed fine-tuning of evoutionary rates, itself a dubious explanation even if the phenomenon were real, this would not be evidence against common descent. It would be evidence against unguided evolution. You have consistently conflated guided evolution and separate creation.
The point of this thread was supposed to be how common design in the context of separate creation of “kinds” is able to explain the nested hierarchy of life. Could we see something on that?
There was an argument over the distance matrix vs. phylogenetic distances.
Referring to Joe Felsenstein’s book, we have the notion of Branch lengths or the number of evolutionary changes in a hypothetical phylogeny. The issue, as I pointed out many times, and which John doesn’t seem to appreciate, is the number of evolutionary changes is hypothetical and is dependent on what one believes is the correct tree. This is therefore subject to uncertainty.
So as has often been the case, John mis-represented what I was saying in order to justify the accusation that I don’t understand or was confused. Not so. In the process, he embarrasses himself over the issue of the relevance of the distance matrix.
In contrast to the supposed phylogenetic branch lengths, that are model dependent, a simple count of differences is very much more objective. If we are dealing with comparing sequences of exact length, then this is very objective. There are slightly more gray areas when we have to insert gaps to align sequences of different lengths, but this is still very highly objective, and less subject to uncertainty.
For example the sequence:
MNPSSI
is only one amino acid letter different than
MNPSSL
It’s not that hard then to arrange things approximately according to the degree of similarity if the data make such arrangements amenable.
So, revisiting Rumraket’s example, one can see the phylogenetic tree in his example approximately lumping of similar species with one another. Not all examples can lump so nicely where the distance matrix so nicely agrees with hypothetical phylogenies, but one can’t totally run away from the idea that it is desirable to put similar species in the same clade.
So much for John Harshman’s dismissiveness of distance matrices!
There is an important distinction between Gene trees and Species trees, but since Common Design doesn’t require Gene trees to agree with Species trees, there is no problem. We can simply construct nested hierarchical trees based on similarity and diversity and not require any sort of forced agreement between gene trees. In fact one may see interesting patterns as a result, for example the gene trees between Marsupials and Placental mammals….
And as can be seen from the distance matrices in the previous comment, the distances are almost clocklike if one chooses a real out group like Ciona vs. a shark.
Yes, there will be inaccuracies if the pattern was created by common descent, but it’s not a problem if the pattern is by common design. The way things will lump together is sufficient to build the nested hierarchy since there is no need to assume the hierarchy as created by common descent.
Now one thing I’m going to keep an eye out for is comparing proteins from Marsupial and Placental of convergent forms.
Here are the differences based on simple counting of amino acid mismatches:
John Harshman,
Once you allow divine intervention into the door you have put common descent in doubt. It becomes only one of many possible explanations of how diversity arose because the door is open for non material explanations.
No, there wasn’t. There was an argument over your use of distances, all by themselves with no use of any algorithm, to infer relationships.
No argument there. But of course all science is subject to uncertainty. This isn’t an argument we’re actually having, just your misunderstanding of the argument we are having.
We weren’t talking about a distance matrix, but about a set of a few distances not organized into a matrix and not analyzed by any method.
Yes, simple matching distances are easy. The problem with them is that they don’t account for multiple hits, and they don’t account for a difference in rate among different transformations at a site. There are a number of distance measures that do attempt to account for these. If divergence is slight, simple matching distance can be adequate; as divergence increases, simple matching distance is subject to increasing error.
You consistently ignore the fact that your notions require the assumption of a molecular clock.
Which, I repeat, I never did.
Gene trees mostly agree with each other and with species trees. How do you explain that?
On what basis have you determined that Ciona is a “real out group”? According to you, aren’t these all different kinds with no actual reationship to each other? And, once more, what you see is likely to be saturation, which ought to be apparent given that the distances are almost equal to the sequence length. Randomized sequences would be about that similar.
All you are saying here is that since there is no expectation from common design, anything is compatible with it. But why is there a nested hierarchy? Can you explain that?
What gene? What lesson do you draw from those three sequences?
Anyway, depicted below is the unprejudiced “phylogeny” built from the data. And lest Rumraket complain, this was pretty much the computational method applied to build the phylogeny for his species. To dispute the results below, one must throw in numerous ad hoc rationalizations to pick an atypical phylogenetic tree that can be constructed from sequence data.
Now, in light of the comments previously, it can be seen that similarities should count for something in building a hypothetical phylogenetic tree. But even by the assumption of common descent, tetrapods, humans then have no business being lumped in with the Sacropterygians (if by Sarcopterygiians means a clade composed of Lungfish and Coelecanth to the exclusion of Actinopterygians!).
Get a load of this Neighbor-Joining tree employing the distance matrix I spoke of and constructed by the least squares minimization. Qualitatively, it simply lumps the most similar things together. It’s shouldn’t be that outrageously hard to understand what is being done, the rest are mathematical details.
Note, the humans fit nowhere in Sacropterygiians, just as I have claimed. The vertebrates all lump into the vertabrates, similar to the way the creationist Linnaeus had arranged them long before Darwin. I even drew Venn Diagrams with colors to drive home the point.
There is a problem putting shark as an outgroup. This relies on circular reasoning about the evolutionary pathway, not to mention it really doesn’t agree with the distance matrices. Ciona is a better more legitimate outgroup of the vertebrates because it is a non-vertebrate chordate. I don’t see why a vertebrate like a shark should be the outgroup of sister vertebrates.
Click to Enlarge Image
http://theskepticalzone.com/wp/wp-content/uploads/2017/11/nj_differnces_circled2-1.png
If that’s true, doesn’t it put all of science in doubt? If a divine explanation for mutation is allowed, and that suggests separate creation, doesn’t it also suggest that theories of gravity, brownian motion, the weather, the circulation of the blood, and everything else are equally in doubt? If not, what stops the contagion from one thing to another?
No, the origin of variation is a separate question from common descent. Common descent explains the nested hierarchy of life. Nothing else does. Go ahead, try something else.
Once you put any made-up cause in the door you have put any effect in doubt. That’s why you don’t get to just make up whatever cause you want, you need evidence that a cause exists and evidence for its actions.
Yes, and justice would be utterly perverted by allowing junk “causes” in the door as well. That’s why we don’t do it.
Glen Davidson
That was perhaps the most naive thing you have said.
No, one must realize that COX I is perhaps not the best gene for this purpose, and simple matching distances perhaps not the best measure of distance, and whatever unstated algorithm you used to perform the analysis perhaps not the best algorithm. And there seems good evidence that the distances between outgroup and ingroup are saturated.
And if one assumes that this analysis makes any sense in the first place.
Those are two different methods. I will assume you actually used neighbor-joining, which is in many cases a reasonable approximation of a least squares tree, though it isn’t a least squares tree.
No, that isn’t at all what a neighbor joining tree does. Neighbor joining (and least squares as well) is capable of dealing with large differences in evolutionary rates among taxa.
Note that this tree contradicts the tree constructed from any random set of nuclear genes. Why do you think that might be? Why is there such a thing as “vertebrate”? Why is there a nested hierarchy of life?
Exactly: you don’t see. Why do you think there’s such a thing as “vertebrate” for Ciona to be the outgroup to, when you don’t think there’s such a thing as “Osteichthyes” for the shark to be the outgroup to? You are making the same assumption you accuse others of; you just don’t realize it.
Rumraket,
John Harshman is just spewing. He does of course have PAUP* and the data. He can build a diagram for you where tetrapods are lumped WITHIN the same clade as the Sarcopterygiian to the exclusion of the Actinopterygiian. That means you can’t have a phylogeny that looks like the one I constructed where humans are outside Sarcoptyerigii, Actinopterygiia, Teleostomi and Gnathostomata.
He can put up or shut up. Seriously. You and I are the only ones actually making any effort to work with data. Until he shows some effort to actually wrestle with the actual data vs. his armchair complaining, he’ll remain on my ignore list.
I don’t think it’s too much to ask of a professional phylogeneticist to build a gene tree. Sheesh.
He could settle the issue rather fast rather than just complain and nitpick trivialities.