are you saying that since there are so possible sequences that can perform the cytochrome c function, the chance of geting the same sequence/tree by convergent (design or convergent evolution) is extremly low, and thus it support common decent? is that your main point basically?
I am loss here! In what sense according to you is the trout more closely related to the kiwi than to the shark? Phylogenetically ? And in that case, what types of studies were used to determine this outcome? Gene and/or protein sequences studies? Others?
Yes. Thatâs what ârelatedâ means, you know.
Yes. All types of studies. There is zero controversy on this point. Feel free to look up any morphological or molecular study that includes at least one shark, one actinopterygian, and one tetrapod. Does this really come as a surprise to you?
2 Likes
Itâs the chance of getting the same tree as that inferred from another gene, or morphology, is extremely low yes. Presumably some designer would âfine tuneâ the gene sequence for their functions, not for the phylogenetic trees they produce. So since there are so many possible, functionally equivalent sequences even among the set that are similar, it is still extremely unlikely that fine-tuning towards function should end up yielding a tree that agrees with the tree from a different independent gene or the tree from morphology.
That means you would have to posit that the designer is deliberately picking the sequences that yield similar trees. But then the designer would be deliberately picking sequences that give the same evidence at that expected from common descent, so heâd be faking his data to give the deceptive appearance of having been produced by a process of common descent.
is extremly low, and thus it support common decent? is that your main point basically?
Yes, because common descent explains why the trees from different genes and from morphology are so highly similar. We expect this pattern from common descent, we donât expect it from design. To expect it on design, we have to posit a deceptive designer using a really weird criterion for picking the functional sequences that he does.
It all gets even worse when we consider nonfunctional sequences. Pseudogenes, retroviral sequences, degraded transposons, introns, and many other types of junk-DNA. Why should these nonfunctional stretches of DNA also yield the same trees when they are under almost zero sequence constraints(besides merely not being deleterious)?
2 Likes
actually some shark proteins are closer to human than to other fishes. but in general that is true.
human for instance has some genes that in birds are using for feathers development.
Once again, you seem to have forgotten that phylogeny is not decided by similarity. Even if these proteins are âcloserâ, phylogenetic analysis of them would recover the usual tree, since âclosenessâ is not the criterion of fit.
1 Like
not if they are closer to human than to bony fishes. in this case they will not produce the correct tree. here is again a phylogenetic tree base on cytochrome b phylogeny:
as we can see itâs actually present a wrong tree. unless we will check more genes/proteins and get the âcorrectâ tree .
(image from ncse site)
i showed here an example of convergent evolution of 44 amino acids. so according to theobald calculation the chance of that case can be as low as 20^44. by his own calculation it should be near impossible, and yet he believe that its happened somehow. so his main claim isnt true, and we must consider the possibility that there is indeed a strong correlation between function and sequence similarity.
Yes, that gene sequences in trout are closer to kiwi than to sharks comes somewhat as a surprise to me. But I guess thatâs the way it is! And In that case, I am afraid that my fine tuning hypothesis is in trouble. Thanks for your explanations.
Once again you have forgotten that phylogenetic analysis is not done by simple similarity. Please try to remember that simple point, because I hate having to repeat it every time.
How do you know itâs wrong? You would appear to be denying that there should be a tree at all, so there canât be a wrong tree any more than there can be a right tree.
Now, what that tree actually does is unclear without reference to the actual paper. Please provide a real reference when you post this sort of thing. Itâs impossible to discuss the tree without a reference, and you donât even tell me where I could find it at NCSE, much less the actual publication. But I suspect that it fails to resolve many of those nodes and that they would collapse in any test of support.
1 Like
Please stop talking about similarity. âCloserâ is not the criterion. You actually thanked me for my explanation of this, if you can recall.
Aha! I was able to find the NCSE reference. As I suspected, the numbers attached to the branches are bootstrap values. Anything below 70 is not significant, and anything below 50 can be ignored. This is an unresolved tree except for the Loris/Bushbaby and the three monkeys (counting the human). Cytochrome b has nothing to say about those relationships, at least as analyzed there.
2 Likes
arent chimp and human the closest ( and also similar) to each other than to any other creatures?
i never said there is no tree. im just saying that that tree can also be explained by design. under the design scenario a reptile protein is closer to other reptile protein than to a mammal protein because of similar morphology\physiology.
Yes, in most ways. You just canât make the assumption, in analyzing data, that those are the same thing. You will get many wrong answers under that assumption.
Yes, youâre saying that. But so far you have managed no justification.
But that isnât even true, is it? You have shown that some reptile proteins are closer to each other than to other reptile proteins. You have no clue how many âkindsâ there are, what they are, or how to recognize them. And you have no clue as to how common design would result in a branching tree.
1 Like
are you saying that if we will go only by similarity\dissimilarity we can get a wrong tree? if so can you give an example of that?
sure. what is true for groups is also true in general for sub groups. i gave that specific case because its clearly show functional reason for sequence similarity. also remember that even according to evolution we should find a correlation between morphology and sequence similarity. this is why all mammals group together.
Yes but that showed the opposite of what you wanted it to show. You wanted the gene trees to be similar to the morphology trees because, you said, similar morphology requires similar genetics. But the publication you references showed the opposite. In that publication, the morphology tree showed that species A and B were more closely related than to C. But that a host of mitochondrial genes in species A, were grouped closer to the C species instead of close to B. In other words, the trees were incongruent, arguing against your claim that the gene trees should be functionally constrained to match the morphology trees.
so according to theobald calculation the chance of that case can be as low as 20^44. by his own calculation it should be near impossible, and yet he believe that its happened somehow.
The explanation for that is common descent. You need to sit down and try to work out an example for yourself to see it, as it might not be immediately intuitively obvious to some why this is so. It wasnât to me at first. But if you work out an example for yourself you will see how it has to be that way. I recommend you do something like make up three different gene sequences to use as the âgenomeâ of some hypothetical common ancestor species, and then you try to evolve them independently through a process of common descent. I did a toy example like that a few years ago to show that it really does work. I made up a phylogeny by evolving 5 different sequences according to this plan:
Here I will make five gene sequences of a more realistic size (300 nucleotides pr. gene), and then I will evolve them by splitting and copying, and introduce random mutation (using a dice-roller to determine which nucleotide position to mutate, and another dice roller to determine how it mutates) until we have 13 âspeciesâ. Then I will âevolveâ those 13 species for a few generations with more random mutations. Then Iâm going to make trees for all five genes from the 13 species and see if they match, or if random mutations cause a common ancestral template to somehow magically destroy phylogenetic signal.
Okay, so here are the results. First of all, here is the overall phylogeny I generated:
It looks like that because itâs actually a zoomed out screenshot from excel where I saved the ancestral and descendant gene sequences. Subsequent generations were evolved by simply making two copies of the ancestor, then introducing between three and five random mutations in each gene. Then making two new copies of the mutated genes, and mutating them again by the same rules. That means there are two âlevelsâ of randomness that affect what type of mutation it is. The position in the 300 nucleotide gene where a mutation occurs is random, and the type of nucleotide it is changed into, is random. To keep things simple, I only allowed substitutions.
I then went on to infer the trees from each of the five genes shared by all 13 species and compared their topology to each other. They all matched exactly, completely identical trees.
Here are 3 of them side by side:
As you can see, they have identical topologies. They were inferred using the maximum likelihood algorithm.
You can do the same, make up some genes, then evolve them along some phylogeny. Then infer the trees and test them against each other.
so his main claim isnt true, and we must consider the possibility that there is indeed a strong correlation between function and sequence similarity.
No, the explanation is common descent. And you can demonstrate it to yourself by doing what I did.
4 Likes
Yes, thatâs exactly what Iâm saying, and Iâve been saying it very loudly for quite a long time now. To see an example, find any phylogenetic analysis that shows branch lengths scaled by number of changes and shows some taxon sticking way out from the rest. The first example I think of is the tinamous. Ostriches are more similar to emus than tinamous, and yet emus are more closely related to tinamous than to ostriches. You might look at Harshman J., Braun E.L., Braun M.J., Huddleston C.J., Bowie R.C.K., Chojnowski J.L., Hackett S.J., Han K.-L., Kimball R.T., Marks B.D., Miglia K.J., Moore W.S., Reddy S., Sheldon F.H., Steadman D.W., Steppan S.J., Witt C.C., Yuri T. Phylogenomic evidence for multiple losses of flight in ratite birds. Proceedings of the National Academy of Sciences 2008; 105:13462-13467.
What do you mean by âgroupsâ and âsub groupsâ? Why should there by any such thing under common design?
Yes, but the explanation for that is phylogeny, which you reject. Under common design you have no reason to suppose such a correlation.
1 Like
It was also my take, but I am not sure anymore now that John Harshman has argued to me that trouts are genetically more related to kiwi than to sharks despite the fact that trouts and sharks are arguably closer physiologically and morphologically. How would you handle this difficulty ?
Would it be wrong to say that at the morphological level phylogeny is determined by similarities?
If morphologies follow a nested hierarchy and if proteins are fine tuned according to morphological/physiological needs, it makes sense that protein sequences would also follow a nested hierarchy, doesnât it?