Intelligent Design

You Won’t Believe This New Epicycle: Congruence Incongruence is a Powerful Phylogenetic Signal

Spread the love

Remember how evolution was confirmed by congruence and proven by parsimony? The idea was that different anatomical comparisons lead to the same evolutionary tree. Even at the genetic level, different genes told the same evolutionary story. Similar evolution trees are derived from completely different genes. Such congruence of independent data was predicted by evolution and evolutionists have consistently proclaimed it as a powerful confirmation of the fact of evolution. It is, as evolutionists like to say, a powerful phylogenetic signal. There’s only one problem: all of this is false. It is yet another example of evolution’s theory-laden science where the findings are dictated not by the data but by the doctrine. There is no powerful phylogenetic signal. That is a myth. For when evolutionists construct their phylogenies, they first filter out the anatomical comparisons that don’t cooperate. But that is not enough so after their first try they filter some more. As one evolutionist admitted, “We are trying to figure out the phylogenetic relationships of 1.8 million species and can’t even sort out 20 [types of] yeast.” And so it is good to see a new paper that admits that data are routinely filtered in order to satisfy stringent criteria so as to eliminate the possibility of incongruence.  Read more

49 Replies to “You Won’t Believe This New Epicycle: Congruence Incongruence is a Powerful Phylogenetic Signal

  1. 1
    Mung says:

    Lot’s of “junk DNA” gives a strong signal in proof of evolution. Very little “junk DNA” gives a strong signal in proof of evolution.

    How much “junk DNA” is “just right”?

  2. 2
    Alan Fox says:

    Ah mung the mereticious!

    As Professor Moran can’t post here, let me refer you to this post at his blog which may go some way to helping you.

  3. 3
    bornagain77 says:

    OT: Stephen Meyer continues his reply to Charles Marshall’s critique of Darwin’s Doubt:

    To Build New Animals, No New Genetic Information Needed? More in Reply in Charles Marshall – Stephen C. Meyer – October 7, 2013
    http://www.evolutionnews.org/2.....77541.html

  4. 4
    Mung says:

    Alan, I’ve heard “just look at the onion” too many times already.

  5. 5
    JGuy says:

    The Onion Test seems more like an argument from ignorance.

    “Can I explain why an onion needs about five times more non-coding DNA for this function than a human?”

    Maybe instead ask or explain first why an onion actually has five times more non-coding DNA.

    From ID, it could be that onion provides nutritional or health benefits to non onions. e.g. of a non onion: a human. 🙂

    related:
    http://www.evolutionnews.org/2.....52321.html

  6. 6
    JGuy says:

    A bit more on the onion test. I think it’s possible that the information content to build an organism, whether it leads to building blocks or instructions is still not clear. The notion that DNA is everything is still a popular idea or feeling – even for those that know better.

    So… another question…
    1) Let’s call the sum of ALL information for development in an organism is S. The value of which varies per organism {onion, human, rabbit, worm, amoeba..etc..}
    2) Let’s call the idea of a metric based on an arbitrarily small amount of genetic information as B. e.g. it can be set to equal 100,000 sequence of base pairs or some other number. So, if it was 100k bp, a human’s DNA would be approx. 30,000B (3Giga base pairs).
    3) Let’s call the sum of all the other information that is either (a)NOT contained in an organisms DNA OR (b) it is part of the known coding portion of the organisms DNA as E. i.e. This is all the information that isn’t called junk DNA; it is either known coding or nonDNA information involved in developmental information.

    So, if we assume B is equal to the size of all claimed non coding portions of human DNA, then this follows:

    For humans S = B + E.
    For onions S = 5*B + E.

    For the Onion Test to have any of it’s intended meaning, it needs to prove that E is not significantly greater for humans than for the E in onions.

    For example, it might be that the human S value is ten or twenty times larger than the onion’s S value. Leaving the the onion test in tears. 😛

    Even so, I still have an affinity for the idea that the extra DNA could benefit other organisms outside of the onion, or it could make the onion more adaptable to more environments.

  7. 7
    tjguy says:

    Another example of the unfalsifiability of Darwinian evolution.

    Opposite results are both claimed as supporting evidence. It’s like the convergent evolution argument.

    Heads I win. Tails you lose.

  8. 8
    wallstreeter43 says:

    I call it the Darwin of the gaps theory. All roads must lead to Darwin, no matter how logically contradicting they are.

  9. 9
    Silver Asiatic says:

    I didn’t understand this story and the articles related to it. There was little reaction from the usually-outspoken Darwinist commentators. But it seemed like it deserved a lot more attention.

    According to a new study partly focused on yeast, the conflicting picture from individual genes is even broader than scientists suspected. “They report that every single one of the 1,070 genes conflicts somewhat,” said Michael Donoghue, an evolutionary biologist at Yale who was not involved in the study. “We are trying to figure out the phylogenetic relationships of 1.8 million species and can’t even sort out 20 [types of] yeast,” he said.

    The only thing I was was an attempt to minimize by stressing the word “somewhat”. But what about the self-depreciating quote at the end? It seems like something a critic of evolution would say. (???)
    And does “somewhat” really minimize the fact that every one of the 1070 genes contradicted the expected tree and was incongruent with each other?

    The resulting tree was accurate according to standard statistical analysis. But given that similar methods have produced trees of life that are rife with contradiction, Rokas and Salichos decided to delve deeper. They built a series of phylogenetic trees using data from individual yeast genes and employed an algorithm derived from information theory to find the areas of greatest agreement among the trees. The result, published in Nature in May, was unexpected. Every gene they studied appeared to tell a slightly different story of evolution.

    “Rife with contradiction”? I guess the answer to that is, “Obviously, evolutionary theory predicts that there will be contradictory findings. That’s why it’s so powerful.”

    “Just about all the trees from individual genes were in conflict with the tree based on a concatenated data set,” says Hilu. “It’s a bit shocking.”

    I think the concatenated data set is the one built to model common descent, and “just about all the trees” conflicted with that. Really???

    Mr. Hilu then says “It’s a bit shocking”. So, again with the minimizing. I would think that on the scale of emotional values, “a bit shocking” is more than “extremely surprising”. Or let’s put it this way: he was shocked by what the data shows — but this is really not a problem, apparently. ???

    They concluded that if a number of genes support a specific architecture, it is probably accurate. But if different sets of genes support two different architectures equally, it is much less likely that either structure is accurate. Rokas and Salichos used a statistical method called bootstrap analysis to select the most informative genes.

    If the genes supported the model, then they accept them as accurate, if they don’t — then they exclude them. That seems like a very strange way to test results.

    In essence, “if you take just the strongly supported genes, then you recover the correct tree,” said Donoghue.

    Cornelius Hunter stated that they’re only taking the samples that match the prediction.

    I don’t get it.

  10. 10
    lifepsy says:

    If the genes supported the model, then they accept them as accurate, if they don’t — then they exclude them. That seems like a very strange way to test results.

    Well the researchers already *know* that evolution is true. When you start with your conclusion, it’s okay to fudge the data until it is in agreement.

  11. 11
    bornagain77 says:

    OT: Blood Vessel Cells Can Repair, Regenerate Organs – Oct. 8, 2013
    Excerpt: Damaged or diseased organs may someday be healed with an injection of blood vessel cells, eliminating the need for donated organs and transplants,,,
    the researchers show that endothelial cells — the cells that make up the structure of blood vessels — are powerful biological machines that drive regeneration in organ tissues by releasing beneficial, organ-specific molecules.,,,
    The researchers also found that organs dictate the structure and function of their own blood vessels, including the repair molecules they secrete.
    Together, the studies show that endothelial cells and the organs they are transplanted into work together to repair damage and restore function,,,
    “Our work suggests that that an infusion of engineered (immature) endothelial cells could engraft into injured tissue and acquire the capacity to repair the organ,”,,,
    http://www.sciencedaily.com/re.....152218.htm

  12. 12
    bornagain77 says:

    OT: Study shows alpine swift (bird) can stay aloft for 200 days – Oct. 8, 2013
    Excerpt: In analyzing the data captured by the sensors, the researchers found that the test birds stayed in the air at one point for 200 days, covering approximately 10,000 kilometers in the process. This, the researchers report, is the longest flight duration ever recorded by a bird, and is only equaled by some sea-going creatures who need only propel themselves forward—birds of course also have to keep themselves in the air, a process that consumes a lot of energy.
    Some of the most obvious questions that come to mind regarding the birds are: how do they eat and drink? When do they sleep? Prior research has an answer for the first, they eat what is known collectively as aerial plankton—a mix of fungus spores, small insects, seeds and even bacteria that float about in the sky. The water in their food is apparently enough to sustain the birds indefinitely. As for how and when they sleep, scientists are still divided. Data from the sensors in the study indicated slow-downs, or periods of reduced activity where the birds glided more than flapped, but that clearly isn’t enough evidence to prove that the birds were sleeping. Some suggest that the birds, like some other organisms, don’t have to sleep, or only do so during certain periods of their lifecycle, such as during mating season.
    http://phys.org/news/2013-10-a.....-days.html

  13. 13
    Silver Asiatic says:

    lifepsy — thanks for your reply. Your sounds sarcastic, but isn’t that exactly what they did — fudge the data? If so, that’s why I still can’t understand it. As I see this, the researchers blatantly admitted that they were going to bias the sample towards their pre-determined conclusion and disregard information that contradicted it (and such a huge amount was incongruous that they were “shocked”).
    I don’t understand how they could get away with that.
    I also don’t understand why their findings are not a massive problem for evolutionary theory and even for claims of common descent.

    I think I must be misreading it, or exaggerating something. I don’t understand how they could report things like:

    “Just about all the trees from individual genes were in conflict with the tree based on a concatenated data set,” says Hilu.

    similar methods have produced trees of life that are rife with contradiction

    The result, published in Nature in May, was unexpected. Every gene they studied appeared to tell a slightly different story of evolution.

    … and how could those statements say nothing at all about the validity or predictive power of evolutionary theory? This just remains just an ordinary bit of science research. And that is aside from the fact that they openly state that they kept only the results that matched their expectations.

  14. 14
    wd400 says:

    Cornelius Hunter stated that they’re only taking the samples that match the prediction.

    I don’t get it.

    Hunter doesn’t get it. The approach is not to match gene trees to a “favoured” topology, but to find the gene trees that have phylogenetic signal. This is a pretty standard practice in science – remove the noisey data and focus on the bits that have signal.

  15. 15
    Silver Asiatic says:

    You may be completely right – I’m not an expert, but it doesn’t seem like a question noise in the signal but rather whether there’s the correct tree structure in place. There was some expectation about the pattern that they were going to find — but were “shocked” that “trees from individual genes were in conflict with the tree based on a concatenated data set”.
    So, wouldn’t the “concatenated data set” be the target or “favoured” topology that they measured against?
    Elsewhere it says:

    … if different sets of genes support two different architectures equally, it is much less likely that either structure is accurate.

    It’s not that there is noisy signal from the genes, but that the architectures or structures are incorrect.
    When they remove the genes that conflict with their architectures, then you get “the correct tree”.
    That’s the way it seems to me, anyway.

  16. 16
    wd400 says:

    So, wouldn’t the “concatenated data set” be the target or “favoured” topology that they measured against

    No. The trees were not filtered on their topology, but on how much signal they have. As it happens, when you filter out the noisey data you get rid of some incgrunence. That’s hardly a surprise – noisey data makes for uncertain estimates.

    You other quote (I don’t know where it’s from, so can only guess context) is talking abut something different – conflicting by well supported topologies in gene trees. This is expected to happen, especially when speciation is recent or rapid. That’s because population genetic processes like incomplete lineage sorting make gene-trees different than species trees. New methods, like those reviewed in the TREE paper Hunter linked to, can take advantage of population genetic models to reconstruct species trees and even population genetic parameters (like population size) from ancient populations.

  17. 17
    bornagain77 says:

    semi related: Darwinists Grapple with Findings on Convergent Evolution – podcast
    http://intelligentdesign.podom.....6_36-07_00

  18. 18
    lifepsy says:

    wd400

    The approach is not to match gene trees to a “favoured” topology, but to find the gene trees that have phylogenetic signal.

    There’s no such thing as a “phylogenetic signal”. That’s only a theoretical construct that comes from inference.

    This is a pretty standard practice in science – remove the noisey data and focus on the bits that have signal.

    Now you have two things that don’t exist: “noise” and “signal” Which is which? What objective criteria are you using to determine there is a signal at all between two fundamentally different classes of organisms?

    http://www.pnas.org/content/10.....nsion.html
    This phylogeny from 2006 based on transposon data placed Humans closer to mice and rats than to dogs

    http://www.ncbi.nlm.nih.gov/pubmed/17206860
    This phylogeny from 2007 found that humans and dogs are more closely related than mice and rats.

    Please tell me which study was following the phylogenetic signal, and which study was hampered by noise.

    The trees were not filtered on their topology, but on how much signal they have. As it happens, when you filter out the noisey data you get rid of some incgrunence. That’s hardly a surprise – noisey data makes for uncertain estimates.

    That is an amusing statement, considering the “noise” is subjectively identified as the data causing the incongruence in the first place… and gee, wouldn’t you know when you remove the incongruence, the incongruence is removed. Now that’s science.

  19. 19
    wd400 says:

    ow you have two things that don’t exist: “noise” and “signal” Which is which?

    Really… If there is signal then of course there is lack of signal, that’s called noise.


    Which is which? What objective criteria are you using to determine there is a signal at all between two fundamentally different classes of organisms?

    I don’t know what “fundamentally different classes of organisms” have to do with it. But phylogenetics is a statistical process, there are various ways to estimate the uncertainty of a phylgenetic estimate. Genes that give well supported trees have signal, those that don’t do not.

    I’ve never seen the papers you link to, but not PLoS one uses whole genomes sequences with no filtering and very few taxa. I’m not very inclined to believe that sort of paper.

    That is an amusing statement, considering the “noise” is subjectively identified as the data causing the incongruence

    But, of course, that’s not how signal is measured. You really ought to try an understand the basis of the theory you are so opossed to…

  20. 20
    wd400 says:

    (I should say, in case you googled “phylogentic signal” and were led up a garden path, there are two definitions of the term floating around. I mean the power of particular DNA sequences to resolve a phylogenetic tree. The term is also used in comparative biology to describe the degree to which variance in a trait across species can be explain by evolutionary relationships among those species. That “phylogentic signal” does compare data to a given tree, but it’s not relevant to the studies discussed here)

  21. 21
    Mung says:

    wd400:

    Really… If there is signal then of course there is lack of signal, that’s called noise.

    sigh.

  22. 22
    Mung says:

    Anything that doesn’t confirm our preconceived notion of what we should see is obviously “noise.”

    Heck, there could be no phylogenetic signal whatsoever and that’s what we mean by noise, the non-existence of any signal.

    Or, it could mean that there is certainly a phylogenetic signal there, it must be there, after all, the theory demands it, but we can’t find it, it must be due to “noise.”

    And that’s science.

  23. 23
    wd400 says:

    Please try and read what I’ve said. Signal is not about matching a tree topology, it’s about the power of a dataset to resolve any tree with certainty.

  24. 24
    lifepsy says:

    I don’t know what “fundamentally different classes of organisms” have to do with it.

    Evolutionists attempt to use phylogenetic trees to map relatedness…. common descent.. you know, that thing Darwin wrote about?

    But phylogenetics is a statistical process, there are various ways to estimate the uncertainty of a phylgenetic estimate. Genes that give well supported trees have signal, those that don’t do not.

    Obviously a pattern will surface within a given data set. And a hierarchy of similarity is inevitable. The whole point of contention is whether there is consilience across multiple data sets, and there is not. And not just over minor points but major groupings.

    As you just stated the signal is a statistic inferred from a data set. So when two sets of data contradict each other, which one is the signal and which is noise?

    I’ve never seen the papers you link to, but not PLoS one uses whole genomes sequences with no filtering and very few taxa. I’m not very inclined to believe that sort of paper.

    So, that is your opinion. I’m not inclined to believe Evolution is true because it’s ridiculous. But these researchers feel their phylogeny is quite robust: “…our results based on full-genome analysis of the phylogenetic relationship between human, dog, and mouse yielded overwhelming support for a primate-carnivore clade with the exclusion of rodents.”

    While the other study, that you seem to prefer, resolves a closer human/rodent grouping apart from dogs.

    This illustrates my point. Individually, the two research groups discovered the so-called “signal” and eliminated the “noise” to resolve a phylogenetic tree. But when you compare both of the research groups’ data sets and trees, then you find completely contradictory results telling two different stories of Evolution. Which one is true?

  25. 25
    wd400 says:

    Obviously a pattern will surface within a given data set. And a hierarchy of similarity is inevitable. The whole point of contention is whether there is consilience across multiple data sets, and there is not. And not just over minor points but major groupings.

    IF you think there isn’t a huge degree of consilience among different genetic datasets, morphology and fossil data then you ought to read more. Even the yeast study in Hunter’s first post found a strong degree of phylogenetic congruence among the diferent gene trees.

    As you just stated the signal is a statistic inferred from a data set. So when two sets of data contradict each other, which one is the signal and which is noise?

    You’d have to measure phylogenetic signal to dind out, of course. It’s possibly to have genuinely conflicting datasets (i.e. strong signal for differing topologies), which point to interesting population genetic event like hypridisation or incomplete lineage soring (i.e. relatively rapid speciation).

    This illustrates my point. Individually, the two research groups discovered the so-called “signal” and eliminated the “noise” t

    No. The PLoS paper is based on whole genome sequences. All the noise is still in the dataset. As I say, using a huge concatenated dataset with no accounting for differing genetic histories, and with only a few “tips” in the tree is not good practice. So on theoretical grounds I like their paper less the the PNAS one, though I’ve literally spent 5 mins reading each.

  26. 26
    lifepsy says:

    IF you think there isn’t a huge degree of consilience among different genetic datasets, morphology and fossil data then you ought to read more.

    I would expect a great deal of consilience in one respect, because similar types of animals will generally group together and return strong patterns. The fudging comes in with the campfire stories about Common Descent and common relatedness of all life. That’s when evolutionists begin invoking all sorts of contortions to support their origins belief system.

    You’d have to measure phylogenetic signal to dind out, of course. It’s possibly to have genuinely conflicting datasets (i.e. strong signal for differing topologies), which point to interesting population genetic event like hypridisation or incomplete lineage soring (i.e. relatively rapid speciation).

    It also points to phylognetic tree models being unfalsifiable when you can make up imaginary past scenarios like those resulting in “incomplete lineage sorting” to explain conflicting data.

  27. 27
    wd400 says:

    I’m afraid you don’t know what you are talking about. Incomplete lineage sorting is an inevitable consequence of population genetics, not some ad hoc explanation for data people don’t like.

  28. 28
    lifepsy says:

    No. The PLoS paper is based on whole genome sequences. All the noise is still in the dataset. As I say, using a huge concatenated dataset with no accounting for differing genetic histories, and with only a few “tips” in the tree is not good practice. So on theoretical grounds I like their paper less the the PNAS one, though I’ve literally spent 5 mins reading each.

    I accept that you dislike the researchers’ approach. However they clearly feel their methodology was quite robust and that their conclusions are “overwhelmingly” supported. We are left with two very different stories of mammalian evolution being told.

    The other option that you aren’t considering is that such phylogeny is unresolvable because Evolution didn’t happen.

    In this case phylogenetics is a statistics program based on similarities and has nothing to do with evolutionary relatedness.

  29. 29
    lifepsy says:

    I’m afraid you don’t know what you are talking about. Incomplete lineage sorting is an inevitable consequence of population genetics, not some ad hoc explanation for data people don’t like.

    Nice try, but I never said ILS-causing events don’t occur. What I said is that it renders your phylogenetics model unfalsifiable since you have an explanation for any conflicting data, whether those speciation events really happened causing incomplete lineage sorting, or you’re just imagining they did. Your theory actually predicts a non-falsifiable scenario.

  30. 30
    wd400 says:

    This is hardly unique to evolutionary biology! Bad papers, and papers that with “wrong” conclusions are published all the time in all fields. Science advances through a synthesis of papers, and careful synthesizers think about the methods that lay behind results as well as the results themselves.

    The idea that phylogeny is merely about similarities is wrong, btw. The mostly widely used statistical methods are explicitly about modelling sequence evolution (i.e. changes in character state) through a tree.

  31. 31
    lifepsy says:

    This is hardly unique to evolutionary biology! Bad papers, and papers that with “wrong” conclusions are published all the time in all fields.

    Sure, just look how many papers tell silly just-so stories about culled genetic accidents building animals.

    Science advances through a synthesis of papers, and careful synthesizers think about the methods that lay behind results as well as the results themselves.

    They are so methodical, yet they refuse to question whether or not the premise of Evolution is true or not. So the research proceeds in a box as a philosophical program.

    The idea that phylogeny is merely about similarities is wrong, btw. The mostly widely used statistical methods are explicitly about modelling sequence evolution (i.e. changes in character state) through a tree.

    And this fails to support Evolution in any unambiguous or convincing way.

  32. 32
    Silver Asiatic says:

    That was an interesting exchange – thanks. There’s one aspect I noticed that answers my concern, somewhat.

    I had the same question about signal and noise. There are the PLoS papers with contradictory conclusions.

    I’m not very inclined to believe that sort of paper.

    But isn’t this doing exactly the same thing we were discussing? It seems to be ignoring the fact that there is a contradiction. If the work of scientists who have spent time and expertise working through analysis, and who claim that their findings have “overwhelming” support can be dismissed as being “bad papers” — on what basis do we determine that something is a “good paper”?

    But aside from that, this project seemed to be looking at the reason for the incongruence in the results. So, I’d think they have to look at the noise and not only (whatever was meant by) the signal.

    But given that similar methods have produced trees of life that are rife with contradiction, Rokas and Salichos decided to delve deeper. They built a series of phylogenetic trees using data from individual yeast genes and employed an algorithm derived from information theory to find the areas of greatest agreement among the trees. The result, published in Nature in May, was unexpected. Every gene they studied appeared to tell a slightly different story of evolution.

    So, they were clearly looking for “areas of greatest agreement”, and they found very little of that. This then was “shocking”.

    I did not want to draw a negative conclusion about the researchers. I’d have no basis to simply say that something was a “bad paper” (what does that say about peer review, the integrity of the scientists … how many bad papers are published making claims that support evolution???).

    But this comment makes sense:

    The other option that you aren’t considering is that such phylogeny is unresolvable because Evolution didn’t happen.

    The idea that the data does not support evolutionary theory may be something they never thought of — or they could never even imagine it. So, they determine that some biological evidence is “noise”.

    But I think they have to explain the noise also. What it seems like they’re doing is just stating that various trees are “rife with contradiction” and never going any farther with what that really means.

    It has to be a problem for evolutionary theory. Clearly, the results were not what they expected and they were “shocked”.

    These are some of the sources I looked at to try to understand this.

    http://www.ncbi.nlm.nih.gov/pubmed/22891620
    These drivers need to be uncovered because mismatches between evolutionary processes and phylogenetic models can lead to high confidence in incorrect hypotheses.

    https://www.simonsfoundation.org/quanta/20130604-a-new-approach-to-building-the-tree-of-life/

    http://darwins-god.blogspot.co.....oes-0.html

  33. 33
    Silver Asiatic says:

    They are so methodical, yet they refuse to question whether or not the premise of Evolution is true or not.

    As above, I finally started to understand this. To me, it’s incredibly obvious that they were bringing a bias into the sample. I couldn’t believe they would openly admit that. How could I see that so clearly and they don’t even notice it?

    But it must be that they can’t possibly imagine that evolution is not true. They never thought of that.

  34. 34
    Querius says:

    Silver Asiatic noticed:

    “Rife with contradiction”? I guess the answer to that is, “Obviously, evolutionary theory predicts that there will be contradictory findings. That’s why it’s so powerful.”

    Haha! Brilliant!

    That was an interesting exchange – thanks. There’s one aspect I noticed that answers my concern, somewhat.

    My feelings, as well.

    To me, everything interesting happens at the edges (paraphrasing Ansel Adams). What do you do with conflicting data?

    – You can investigate whether there’s a flaw in the experiment, and learn from it.

    – You can see whether there is a flaw in your hypothesis, and learn from it.

    – You can imagine new scenarios that reconcile the data, at least in part, and forge ahead.

    – You can marginalize the disparities, and learn nothing. This “filtering” can often be unconscious.

    To me, this type of data is precisely the most interesting and provocative part of scientific investigation.

    “Hmmmm. There’s something funny going on here . . .”

    Rather than force fit a solution, let’s “listen” to the data.

    “We are trying to figure out the phylogenetic relationships of 1.8 million species and can’t even sort out 20 [types of] yeast.”

    Instead, I’d suggest that the 20 types of yeast would be an excellent springboard for new ideas!

  35. 35
    wd400 says:


    But isn’t this doing exactly the same thing we were discussing? It seems to be ignoring the fact that there is a contradiction. If the work of scientists who have spent time and expertise working through analysis, and who claim that their findings have “overwhelming” support can be dismissed as being “bad papers” — on what basis do we determine that something is a “good paper”

    This is probably the hardest idea for non scientists (and unergrad scientists!) to come to terms with. Lots of results in published papers in all disciplines are “wrong”. That’s an inevitable result of the way statistics, and the publishing system, works. This is then compounded by the fact bad papers are published and methods improve (I can’t imagine that PLoS paper would be published today, for instance).

    So we can’t just accept published results on their face value. We have to understand all of the results in a given field, and understand the strengths an weaknesses of the studies contributing those results (this is why meta analyses are so crucial to medical science, for instance). In this case, many many papers find evidence for a (primate(rodent,rabbit-like mammals) relationship. Very few show evidence for a (primate, carinovara) relationship. On top of that, the methadology of the PLoS paper is bad. Few taxa, many genes treated as one and no attempt to measure conflicting signals or noisy data. At this point, a sensible person would discount the PLoS paper and accept the preponderance of evience.

    But it must be that they can’t possibly imagine that evolution is not true. They never thought of that.

    It would be bizzre to imagine scientists should ignore the overwhelming degree of congruence between phylogenetic trees estimated from different sequences and with morphological and fossil data when they hit a few groups that are harder to resolve. If you really imagine the tree of life is in dissaray then you need to reader a bit more widely.

  36. 36
    Querius says:

    wd400 opined:

    It would be bizzre to imagine scientists should ignore the overwhelming degree of congruence between phylogenetic trees estimated from different sequences and with morphological and fossil data when they hit a few groups that are harder to resolve. If you really imagine the tree of life is in dissaray then you need to reader a bit more widely.

    No one’s suggesting that scientists ignore any data. It’s actually the opposite: that they start paying more attention to the source of divergence from the expected results.

    By analogy, being mostly accurate with “a few groups that are harder to resolve” as outliers would have left Science stuck in Newtonian physics.

    For Darwinists to play whack-a-mole with dissonant data harms Science, delaying its progress. Darwinists who choose between morphological similarity in some circumstances, and reject it in others (i.e. convergent evolution) have far too much flexibility to maintain credibility and rigor.

    Maybe the data will demonstrate that there’s no single tree, or maybe the data will reveal an entirely different topology! For example, what if ancient organisms had a profoundly greater capacity to adapt than modern ones?

  37. 37
    wd400 says:

    I find it very hard to relate what you are saying to phylogenetics as it is practiced. I mean, a statement like

    Darwinists who choose between morphological similarity in some circumstances, and reject it in others (i.e. convergent evolution) have far too much flexibility to maintain credibility and rigor.

    Is just unrelated to how phylogentics actually works.
    I don’t know why ancient organisms having a greater level of adaptability would change much about phylogenetics (molecular phylogenetics just as well for junk DNA as functional…). In any case, I was replying to the suggestion that discordance in gene trees should make us think life didn’t evolve. To entertain that idea you really would have to ignore a huge amount of data.

  38. 38
    Querius says:

    wd400,

    You’re the one who brought up “morphological and fossil data.” So yes, exactly. I agree with you that it’s not how phylogenetics actually works.

    So, you believe that adaptability has little effect on phylogenetics? Hmmm. Give it some thought.

    While I’m extremely skeptical of macroevolution, I understand your point that “discordance in gene trees should make us believe that life did not evolve.” However, to me it’s obvious that the discordance is important, and is likely to reveal a more effective, natural mechanism (which still might be the product of design) for adaptation than mutation.

    My impression is that you’re marginalizing this discordant data on the strength of “the preponderance of evidence,” a mistake in my opinion.

  39. 39
    Silver Asiatic says:

    To me, everything interesting happens at the edges (paraphrasing Ansel Adams). What do you do with conflicting data?

    I enjoyed that list you provided. The problem, as I’ve seen it, is that there was so much contradictory data that the researchers were “shocked” by it. I don’t think they explained the cause of those conflicts.

  40. 40
    Silver Asiatic says:

    This is probably the hardest idea for non scientists (and unergrad scientists!) to come to terms with. Lots of results in published papers in all disciplines are “wrong”. That’s an inevitable result of the way statistics, and the publishing system, works. This is then compounded by the fact bad papers are published and methods improve (I can’t imagine that PLoS paper would be published today, for instance).

    That was helpful, thanks. I still don’t understand how there can be claims of such a high degree of certainty about evolutionary theory when at the same time “lots of results” published in peer-reviewed journals are “wrong”.
    I think a valid theory has to explain all the data and not have to resort to “epicycles” or ad hoc explanations.
    It seems also that the statistical analysis has to begin with the assumption that things did evolve from a common ancestor. But correlation doesn’t mean causation so it doesn’t seem to be as conclusive as many scientists claim it to be.

    It would be bizzre to imagine scientists should ignore the overwhelming degree of congruence between phylogenetic trees estimated from different sequences and with morphological and fossil data when they hit a few groups that are harder to resolve. If you really imagine the tree of life is in dissaray then you need to reader a bit more widely.

    Yes, I see the term “overwhelming” quite a lot, but then I noticed the following quote:

    “We are trying to figure out the phylogenetic relationships of 1.8 million species and can’t even sort out 20 [types of] yeast.”

    The different between 20 and 1.8 million is pretty overwhelming in itself. After that, I don’t see how one could conclude that the tree of life is not in disarray.

    So there’s a contradiction, as I see it. Between the certainty you express and the quote I cited above.

    So we can’t just accept published results on their face value. We have to understand all of the results in a given field, and understand the strengths an weaknesses of the studies contributing those results (this is why meta analyses are so crucial to medical science, for instance). In this case, many many papers find evidence for a (primate(rodent,rabbit-like mammals) relationship. Very few show evidence for a (primate, carinovara) relationship. On top of that, the methadology of the PLoS paper is bad. Few taxa, many genes treated as one and no attempt to measure conflicting signals or noisy data. At this point, a sensible person would discount the PLoS paper and accept the preponderance of evience.

  41. 41
    Silver Asiatic says:

    Sorry – the bottom of my reply above was unintentionally pasted from wd400’s.

  42. 42
    lifepsy says:

    wd400:

    In any case, I was replying to the suggestion that discordance in gene trees should make us think life didn’t evolve.

    There’s no accordance in phylogenetic trees that should make you think life did evolve. You have overall genetic similarity where it is generally expected to be found based on similar phenotypes, and a lot of weight fudging and unfalsifiable storytelling when you attempt to draw a cohesive picture of common ancestry between major groups. It’s a type of scenario we would expect to find if Evolution was not true.

    What gives the whole game away is that all the researchers constructing these trees will not allow for the possibility that the tree does not exist. It *must* exist, and their job is to resolve it. That makes phylogeny little more than a philosophical statistics program.

  43. 43
    Mung says:

    There are as many trees as there are …

  44. 44
    littlejohn says:

    Religions?

    My questions for WD400 (or anyone else) are: how do the trees account for the unknown quantity and nature of character states of extinct species, or species that were never found? And, how do the trees account for the unknown nature and quantity of character states in known species?

  45. 45
    wd400 says:

    Littlejohn,

    You can’t include unsampled species. All we can do is measure the effect of “taxon sampling” on trees estiamted form simulated or experimental data, where the true tree is known.

    The way phylogenetics works accounts for ancestral species, by calculating the likelihood aeach ancestral node in the tree (i.e. common ancestor) would have had each possible chracter state. Those likelihoods all contribute to the over-all likelihood for the tree, whichi is then used to search through the space of possible trees for the best-fitting one.

    I’m not sure exactly what you meant about the ” unknown nature and quantity of character states in known species”. Any analysis in any field is restricted to the data that you have to hand. Ambiguous data cab be included, with teh likelihood just spread out across all possible states.

  46. 46
    Mung says:

    wd400,

    What’s your opinion of Elliott Sober’s Reconstructing the Past?

  47. 47
    Mung says:

    …we cannot ignore a major conclusion of the previous chapters: No model, no inference. One cannot study the legitimacy of these conflicting claims about the evidential significance of character distributions without assuming, however provisionally, a concrete model of the evolutionary process.

    – Sober, Elliott. Reconstructing the Past: Parsimony, Evolution, and Inference

  48. 48
    littlejohn says:

    WD-400

    The vast majority of data is missing, so how can there be any other outcome than discordance, under these experimental constraints?

    From the layman’s perspective, the available data appears to be extremely limited in most known extinct species (such as trilobites- we have mostly hard parts and very little molecular data).

    The same problem appears to exist with most extant species, so that the full nature and quantity of states or data appear to be inaccessible.

    For these reasons, I am skeptical, and feel the modeling sample sizes are much too minute to be useful for developing phylogenies with any certainty, leading me to believe that the methodology is an exercise of endless speculation.

    Perhaps the book you recommended, “Tree Building”, will help me to better understand your confidence in the paradigm.

  49. 49
    Silver Asiatic says:

    All we can do is measure the effect of “taxon sampling” on trees estiamted form simulated or experimental data, where the true tree is known.

    This was helpful again, thanks. In the above, would it be correct to day that “the true tree” is the one in which the selected sample data is most correlated to?

    The way phylogenetics works accounts for ancestral species, by calculating the likelihood aeach ancestral node in the tree (i.e. common ancestor) would have had each possible chracter state.

    I find it hard to imagine how they could calculate the probabilities that convergent evolution caused separate, non-ancestral lineages.

    It also doesn’t seem like this statistical analysis says anything about the mechanisms that caused the supposed ancestry.

    From the article:

    They built a series of phylogenetic trees using data from individual yeast genes and employed an algorithm derived from information theory to find the areas of greatest agreement among the trees. The result, published in Nature in May, was unexpected. Every gene they studied appeared to tell a slightly different story of evolution.

    There are multiple trees — so why couldn’t that mean multiple origins? Why not consider that all these various trees and discordance are evidence that it is wrong to try to synthesize them into one evolutionary story. As above, there are different, supposed, stories of evolution. Why not conclude that there is no real story of evolution at all? The fact that data shows similar patterns does not seem like strong evidence of common ancestry, given the above.

    “Just about all the trees from individual genes were in conflict with the tree based on a concatenated data set,” says Hilu. “It’s a bit shocking.”

    I keep returning to the above quote from the researcher. To say something like “we often see discordant results, and that’s just noise from the data” or “there are multiple trees and there is overwhelming evidence that genetic data fill fit them” — that contradicts the idea that these findings were “shocking”.

    Obviously, they didn’t expect to find that kind of incongruence. So, it doesn’t seem right to explain it as part of the ordinary process. That’s where I have a problem with it. When the researcher says it’s shocking, but then makes no (or a very lame) attempt to explain it, there’s a big gap in the storyline.

    This seems to happen a lot, from my reading. Evolutionary results are not what was expected, as in this case. But then nothing is done to explain why. More importantly, it’s only very rare when someone offers the suggestion that the evolutionary story itself may be wrong, even though the findings seem to indicate that very thing.

    As it stands, given the very soft evidence of statistical probabilities, I don’t see that this particular problem has been resolved.

    I think the best answer so far is that there is an “overwhelming” amount of supportive evidence, so we shouldn’t pay much attention to this.

    But that doesn’t seem to be dealing with the fact that the evidence, in this case, was “shocking”.

Leave a Reply