Home » Intelligent Design » Kimura and the Adriatic Lizards

Kimura and the Adriatic Lizards

Over at Panda’s Thumb, they are taking issue with the values for selection probabilities of neutral and advantageous mutations that Sal has taken from Kimura and Ohta’s “Theoretical Aspects of Population Genetics”. Since there was a link that provided a ‘look-see’ inside the book, I did so. Well, what I found was very fascinating.

Kimura and Ohta give a very brief overview of the entire field of population genetics up to the time of their writing (1971), distinctly admiring the pioneering work of R.A. Fisher, but not following it because it uses a more sohpisticated “branching process”, and because his model assumes an “infinite” population size. So they write the following:
“. . . [Haldane's] results allow us to make statements as ‘it takes about 1,000 generations until the gene frequency changes from 0.7% to 99.3% with selective advantage s= 0.01′. . . .

“More than 30 years after publication of Haldane’s paper (1927b), we have finally begun to understand more about the fate of individual mutant genes in terms of the powerful diffusion methiods based on the Kolmogorov forward and backward equation (cf. Kimura 1964). In particular, the average number of generations until extinction, and also the time until fixation of an individiual muatant gen in a finite population have been workd out (Kimura and Ohta 1969a,b).”

Kimura and Ohta go on to give the kinds of equation that they have developed over the prior 15 years when dealing with fixation. All of this brought to mind a thread I posted just a few weeks back about the amazing phenotypic changes that had taken place in lizards transplanted from one Adriatic island to another just 36 years ago.

Kimura and Ohta, in this highly regarded work on population genetics, first give Haldane’s number of 1,000 generations for the time of a new beneficial mutant gene to become fixed, then talk briefly of the advances made since Haldane’s estimates (we hear this in the quote above), and then, finally, give their own equation.

Their equation is: K=4Nes1v= the rate of gene substitution, where Ne=effective population, s1=selective factor for the beneficial allele, and v=nu/2N=mutation rate/gamete/generation. So, to demonstrate the improvement (i.e., the speedier rate for fixation) of their formulation over Haldane’s estimate, they work out an example. They use Ne- 10^4, s1=0.01, and v is worked out using the rate of deleterious mutations in Drosophila.

Hence, they write: “If we assume that advantageous genes occur by mutation only 1/1000 as oftne as lethal genes, i.e., v=1.5 x 10^-5, and still assuming Nsube=10^4 and ssub1= 10^-2, then we have K= 400 x 1.5 x 10^-5 = 6 x 10^-3 or about one substitution every 170 generations.” They conclude saying: “It is evident from these considerations that the formula for the probability of gene fixation has important applications in evolutionary theory.”

Well, the transplanted population size of the Adriatic lizards was 12. Obviously it grew over time. Let’s see how Kimura’s numbers work out. Let’s assume that Navg is 250. Ne is 0.8 x N = 0.8 x 250= 200. Let’s just assume that ONE mutant gene was fixed (although we know from the paper that many more fixed genes were involved), and let’s assume that it took the entire 36 years to become fixed. Then, using Kimura’s formula, let’s calculate what s1, the selection factor, is.

1/36=K x 36= ONE fixed mutant gene= 4Nes1v. Using v= 1.5 x 10-5, and Ne=200, then solving for s1 we have: s1=1/36 x 200 x 1.5 x 10^-5 =10^5/36 x 300= 100,000/10,800=9.3. But, of course, s cannot be greater than 1. In fact, it can’t even be 1 since that would mean the entire population died out. Thus, the simple math demonstrates that, using the Modern Synthesis, what was documented to have occurred with the transplanted lizard population cannot even begin to be explained using the Darwinian account.

Of course, this is not how Darwinists see things. They don’t look at the evidence, then look at the numbers, and then conclude that obviously something other than Darwinian mechanisms are at play. They just simply say, “Darwinism is a fact. It’s even more than a fact than gravity itself. Therefore, Darwinism is what caused this transformation. Now I’ll just simply look for some kind of facile explanation.” If you don’t believe me, then just watch some of the posts that we all know are coming!!!

It’s the inadequacy of Darwinism to explain matters such as these that led Kimura to his “Neutral Theory”. It also led Sir Fred Hoyle to completely dismiss Darwinian mechanisms. It was my exploration of the mathematics involved that convinced me that Darwinism has to go.

  • Delicious
  • Facebook
  • Reddit
  • StumbleUpon
  • Twitter
  • RSS Feed

48 Responses to Kimura and the Adriatic Lizards

  1. Glad to see you putting numbers to the problem.

    A couple of suggestions:

    1) Perhaps you can do some equation formatting to make it easier to follow your computation?

    2) I think people might take issue with your assumption of 250 for average population size. If this estimate is based on some empirical reasoning, you propbably should show why you arrived at that number. Does the original article hint at the current population size?

  2. PS By equation formatting I mean putting the equations on separate lines with spacing between each line, like this:

    y = x ^ 2

    2N = 180 * 1/p

    etc

  3. could you please make the following corrections to this post, based on data taken from the paper in question:

    the original population size was 10 lizards, not 12

    the final population size was 5,000 lizards. THis suggests that your estimate of the average population size of 250 is orders of magnitude too low. If you substitute in a more reasonable population size, say 5,000, the math works out perfectly fine. s = 1.39 x 10^-6.

  4. actually it looks like the real problem is that you mixed up 10^5 and 10^-5 somewhere..

  5. dmso74:

    First, thank you for picking out the numbers. However, if we assume a doubling of population each generation, let’s just say that it takes 8 years to get to an average population size of 4,000. Ne is 3200. Instead of 36 years, we need to use 28 years.

    The formula is:

    K = 4 x Ne x s x v

    v= mutation rate = 1.5 x 10^-5.

    So, s=1/[28 x 3200 x 1.5 x 10^-5]

    s=10^5/ 28 x 4800

    s=100,000/134,400

    s = .75

    If you insist on using 5,000,

    then s = 4/5 x .75 = 0.6

    This is an extremely high value for s. Nonetheless, this is what s must be for there to have been just ONE beneficial mutant becoming fixed in the number of years recorded.

    How do you respond?

    P.S. Atom, is the equation clear enough now? The 1/36, or the 1/28 numbers represent K. I calculate K reasoning this way. We have ONE mutant fixed. And the denominator represents the number of generations needed to have that ONE mutan fixed. (I’m assuming one breeding season a year.)

  6. Sorry, I left out the “4″. So, we’re looking at s = 0.15-0.19

  7. dmso74…..

    The formula for s is

    s = K / [4 x Ne x v].

    If 10^-5 is in the denominator, then you can very simply place it in the numerator and switch sign; so it becomes 10^5 in the numerator. Then the formula is

    s = K x 10^5 / [4 x Ne x 1.5]

    s = K x 10^5 / [6 x Ne]

  8. i would respond by saying that an improbably high s value is a lot different than an impossible s value.

  9. You have the units for K wrong. Check page 11.

  10. dmso74:

    Would you also like to say that the enlarged skull and bite, the cecal valves and the changed behavior are the result of ONE mutant?

  11. Yes PaV, much better. Thanks.

    And dmso74, thanks for answering my other question about average population size.

  12. 12
    dreamwalker007

    Something’s fishy about this. There’s a small sample of the book at:
    http://books.google.com/books?.....E#PPA27,M1
    (Or just google “Theoretical Aspects of Population Genetics”)

    The page with the formula that you referenced isn’t included in the preview (probably on page 11 or 12), but it was mentioned again on page 27.

    One quote stood out:
    “From the estimation made so far, K is at least 20″
    I can’t tell how without the full text, but It seems that you’ve misunderstood something.

    It doesn’t make sense the way you presented it anyway. For example, if we use a very small number for Ne, and say that all mutations are neutral (s1 = 0), then the rate of gene substitution should be similar to the mutation rate(v). But the formula just gives a value of 0.

    There’s a number of alarm bells going off for me on this post. Including the fact that most of the paper around where the formula was mentioned deals with gene fixation instead of rate that genes are replaced.

  13. of course not. but the whole point of your post was that the selective factor was impossibly high, and thus evolutionary theory could not even begin to explain the phenotypic changes. however, this is only true if you use a population size that is off by an order of magnitude or so from the real data. so your post is incorrect in its implications. agreed?

  14. dreamwalker007:

    It’s on page 12. On the same page Kimura makes a calculation of K. The value he arrives at is 6 x 10^-3—that is, .006.

    There’s a number of alarm bells going off for me on this post. Including the fact that most of the paper around where the formula was mentioned deals with gene fixation instead of rate that genes are replaced.

    The K we’re calculating is the time for a beneficial mutation to become fixed. K times a number equals 1. That number is the number of generations required to get the frequency of an allele up to 100%. But once you get above 60%, it’s not going to take that many more generations before it reaches 100%. I think this answers your objection. Again, we’re talking about ONE mutant. It should be quite obvious that more than ONE mutation is required. We’re talking about what, a one amino acid change in a protein?

  15. Look at Larry Moron’s post here:
    Visible Mutations and Evolutions

    Richard Lewontin uses the example of the Indian and African rhinoceros to focus the debate. The African rhinoceros has two horns while the Indian rhinoceros has only one. The question is whether this difference is due to natural selection—is two horns better than one in Africa? Or, is it just an accident of evolution that one species has two horns while the other has only one?

    Answer: probably not selection.

    I don’t understand why the adaptationist camp is so reluctant to admit that some visible characters can be fixed by random genetic drift. The idea that every feature of an organism has to be an adaptation seems so out of touch with our modern understanding of evolution that I’m really puzzled by the vehemence with which adaptationists defend their orthodoxy. It seems as though admitting that visible phenotypes might be non-adaptive is a major threat to their worldview.

    Larry Moron

  16. Sal, don’t you think this post is really about Haldane’s Dilemna? We discussed this before when it came to Biston betularia, the peppered moth, and I remain convinced that Haldane wrote the paper we discussed in response to the Kettlewell experiment since he had previously estimated that it would take a 1,000 generations to fix a mutant and the Kettlewell experiment was forcing him to rethink things since the phenotypic change had taken place in so few generations. But we saw the change of only one simple trait in the case of the peppered moths, whereas here we have a slew of morphological and behaviorial changes in an incredibley small period of time, with a very small initial–and final–population. This, indeed, is a dilemna for the Modern Synthesis. I’m suspect that’s why Allen MacNeil says that the Modern Synthesis is dead.

    But, of course, if the Modern Synthesis is the foundation of Darwinian theory, then if it fails, why, then, should we take Darwinian theory seriously any longer?

  17. By trying to point out that the modern synthesis fails, you are slaying a straw man. The current synthesis or whatever you want to call it contains many more things than simple gene selection, genetic drift and gene flow. Evolutionary biologists are looking at numerous other things that affect both genetic and morphological changes.

    None of which make selection suspect at all. Selection may not explain the immediate changes that take place but it will explain a lot of the long range changes that set in.

    I am far from an expert in any of this but basic common sense will tell you that there are wide variances within a populations and some of these will affect success in survival. Many of these variance are not visible as morphological differences but still affect survival. They could be enzyme availability or something similar.

    I fail to see what such analysis as this gets ID. So we indicate that certain formula are not perfect. So what. That is what science is about. None of this undermines the Darwinian paradigm or helps ID.

    Micro-evolution is with us and generates most of life’s changes. Let’s embrace the processes that account for it which include natural selection, genetic drift, gene flow and a whole host of other processes that affect off-spring. Let’s get ID out of a rut as sniping at meaningless inconsistencies.

  18. jerry:

    I fail to see what such analysis as this gets ID. So we indicate that certain formula are not perfect. So what. That is what science is about. None of this undermines the Darwinian paradigm or helps ID.

    Maybe you should open your eyes. Firstly, it points out the failure of Darwinism. Despite anything that MacNeil wants to say, the Modern Synthesis is THE argument for Darwinism. As I pointed out to Dr. MacNeil when we discussed this some time back, once the MS is destroyed, so too is the mathematical foundation for Darwinism. What is worse, any explanation for the the formation of information is also destroyed. If there is no Darwinian mechanism for explaining the presence of information in the genome, then how is it to be explained? Does this not open the door to ID, jerry?

    Micro-evolution is with us and generates most of life’s changes. Let’s embrace the processes that account for it which include natural selection, genetic drift, gene flow and a whole host of other processes that affect off-spring. Let’s get ID out of a rut as sniping at meaningless inconsistencies.

    jerry, maybe you would like to explain just exactly how “random drift”, “natural selection” and “gene flow” can explain this profound morphological change in a population that began just 36 years ago with an initial population of 10 lizards. Would you like to do that? Remember that cecal valves are found in only a tiny portion of lizards, and had never been seen before in these particular lizards.

  19. It is quite possible that all of the morphological changes that have occurred in this population over the past 36 years have been the result of changes in development (i.e. changes in the expression of genes) rather than changes in the frequency of the genes (or, more precisely, their alleles) themselves. To determine if this is the case would require identifying the relevant developmental pathways and determining what developmental mechanisms had caused the observed changes.

    Similar research into the rapid changes observed in the beaks of finches on Isla Daphne in the Galapagos over the past three decades have shown that such changes have resulted from modifications in the expression of a single homeotic gene (BMP4), rather than multiple changes in multiple genes affecting skull morphology.

    The surprising appearance of cecal valves in these lizards may very likely be the result of the “up-regulation” of a genetic program previously suppressed in these lizards. The fact that cecal valves have not been observed in their close relatives is no evidence at all for the hypothesis that the genes for such valves are present in these lizards, but have not been expressed until recently.

    In other words, this may very well be simply another example of developmental plasticity in evolutionary developmental biology (“evo-devo”). Until such investigations have shown conclusively that the observed changes cannot have happened as the result of modifications of developmental pathways via modification of expression of homeotic genes, then entertaining the possibility that an omnipotent supernatural intelligent designer with an inordinate fondness for herbivorous lizards would be somewhat premature.

  20. As to the relevance of all of this to the “modern evolutionary synthesis”, that particular set of theories has been superceded for at least forty years. As I have pointed out on multiple occasions, until ID supporters start engaging scientists on the level of current theory, ID will not be taken seriously by anyone not already committed to supernatural explanations for natural phenomena.

  21. PaV,

    Selection is alive and well in evolutionary biology. It is not the only thing. No one and I repeat no one is saying that selection is everything except some kooks. Selection is no threat to ID.

    I am off for the weekend. When I get back I will have more to say. But maybe you should point out the non naturalistic way this happened. Was this an ID event? If not what it the issue. You seem to be fighting a war from 40 years ago.

  22. Allen MacNeil: (20)

    As I have pointed out on multiple occasions, until ID supporters start engaging scientists on the level of current theory, ID will not be taken seriously by anyone not already committed to supernatural explanations for natural phenomena.

    The problem, as I see it, Allen, is that current theory—by which I suppose you mean evo-devo—relies on pre-existing information that simply gets shuffled around or that becomes regulated in a novel way. The Modern Synthesis proposed a way in which information could be built up over time. But it is no longer tenable. And so with it goes any possibility for a naturalistic method of creating biological information.

    This, to me, seems deeply problematic. It’s like saying I can make a space ship out of the Corvettes. Well maybe you can. But where did the Corvettes come from. I don’t see how you can avoid this problem.

  23. As a follow-up, if the Modern Synthesis was demolished 30 or 40 years ago, then with it should have gone population genetics. Aren’t they still teaching that? What about at Cornell, don’t they teach it there? (Let’s remember that Fisher’s “fundamental theorem of natural selection” and other population genetics, were the basis of the so-called “Modern Synthesis”)

  24. jerry:

    “Selection is alive and well in evolutionary biology.”

    Is that what Will Provine would say?

    But maybe you should point out the non naturalistic way this happened. Was this an ID event?

    It’s an ‘epigenetic’ effect—Lamarkian in character; that is, I propose that the new food ingested by these lizards brought about the noted changes. A simple experiment would perhaps substantiate this: put some lizards in a sequestered area where the “plants” have been transplanted to the original island, and watch what happens in time. You would also have to put up some netting over the area to keep out the bugs that are their normal diet.

  25. Allen wrote: “The fact that cecal valves have not been observed in their close relatives is no evidence at all for the hypothesis that the genes for such valves are present in these lizards, but have not been expressed until recently.”

    I don’t think there are any such genes whose expression is being turned on. The “specification” of cecal valves is almost certainly an emergent property of multiple genes whose expression is simply being changed slightly.

  26. PaV: “As a follow-up, if the Modern Synthesis was demolished 30 or 40 years ago,…”

    He didn’t say that. He said that it was superceded, which is an enormous difference.

  27. 27
    JunkyardTornado

    PaV wrote: …[Kimura et. al. write]: “If we assume that advantageous genes occur by mutation only 1/1000 as often as lethal genes, i.e., v=1.5 x 10^-5, and still assuming Nsube=10^4 and ssub1= 10^-2, then we have K= 400 x 1.5 x 10^-5 = 6 x 10^-3 or about one substitution every 170 generations.”…

    [Pav:]Let’s just assume that ONE mutant gene was fixed (although we know from the paper that many more fixed genes were involved), and let’s assume that it took the entire 36 years to become fixed. Then, using Kimura’s formula, let’s calculate what s1, the selection factor, is.

    1/36=K x 36= ONE fixed mutant gene= 4Nes1v. Using v= 1.5 x 10-5, and Ne=200, then solving for s1 we have: s1=1/36 x 200 x 1.5 x 10^-5 =10^5/36 x 300= 100,000/10,800=9.3.

    Regarding the Adriatic lizard research, I only had access to the abstract and the press release at phys.org that was posted on the previous thread. In neither of those is the word “mutation” mentioned once. Before even reading them I would not have assumed the researchers were crediting actual mutations at all as a factor in the changes taking place in a mere 36 years. It would seem evident to me that those changes had to do with genetic material that was alread latent in the previous population on the previous island. So, Pav could you please clarify here, was anyone associated with that project (assuming you have access to the actual paper) talking about mutations occuring in that time span to account for the changes? (And presumably they were all Darwinists.)

    On the new island the beneficial “mutation rate” would in fact have to be viewed as the rate at which new beneficial traits were introduced via reshuffling of genes already present in the species. As such, that “mutation rate”, (v) would be much higher than 1.5×10-5. Would it not?????

  28. Sal, don’t you think this post is really about Haldane’s Dilemna?

    Kimura appealed to Haldane’s dilemma (not by the phrase “Haldane’s Dilemma”, but by the repeated use of his famous paper on the cost of natural selection).

    Natural selection as conceived by Darwin is the idea there is inevitable progress. This is an inappropriate model of evolution. We have mutation followed by lots of random selection, and then some natural selection.

    Kimura put numbers on how much selection is attributable to:

    1. differential reproductive success due to inherent advantage(selection in the Darwinian view)

    2. random selection

    But random selection is statistically shown to dominate over natural selection.

    The mutationists like Bateson were in large part correct at least for how most of how evolution happens. Selection actually has to be frequently disengaged for innovation to happen.

    For the readers benefit, this latest firestorm begain with my post: Gambler’s Ruin is Darwin’s Ruin.

    I haven’t had time to grind through the numbers in the original post, but I am distressed how sparsely selection has actually been actually measured in the wild.

    Granted there is great difficulty in making the measurement, but in such case, perhaps it is a bit pre-mature to attribute all or even most of the aspects of biodiversity to natural-selection.

    If mutation and random selection are the principle mechanisms of change (an not natural selection) in evolution today, we should acknowledge this. To do otherwise is to do a disservice to the facts.

    What happened in the deep past is another story, but I don’t believe we are even accurately modeling evolution in the present.

    If the genetic entropy thesis is correct, we should be able to empirically affirm or falsify it as cheap sequencing technologies like Solexa arrive on the market. I think the jury is still out.

    Of course if genetic entropy is in evidence in the present day, I expect it will force a re-thinking of what mechanism brought us here. If we have to say, “I don’t know”, the so be it. That is far better than insisting on mechanisms, like neo-Darwinism, which are highly suspect.

  29. By the way Allen, does Alexei Kondrashov still work at Cornell?

    His paper : “Why aren’te we dead 100 times over”? Has some relevance to my area of personal research interest.

  30. Greetings, Sal:

    “Selection actually has to be frequently disengaged for innovation to happen.”

    I have been making this point for many years. Natural selection has the effect of decreasing variation in populations, but variation is absolutely necessary for all evolutionary mechanisms to operate. R. A. Fisher based his “Fundamental Theorem of Natural Selection” at least partly on this realization.

    Relaxation of selection (as happens among domesticated animals and plants) generally has the effect of allowing increased variation to occur and persist. It may even have the effect of “stimulating” variation, by means of various developmental and epigenetic mechanisms.

    One of the reasons I am encouraged by the weakening hold that the “modern evolutionary synthesis” has on evolutionary theory is that, as more examples of evolutionary change that are not explainable via pure genetic change accumulate, it will become easier for scientists not already irretrievably wedded to the “modern synthesis” to investigate new and possibly more fruitful mechanisms of evolutionary variation.

    “If mutation and random selection are the principle mechanisms of change (an not natural selection) in evolution today, we should acknowledge this. To do otherwise is to do a disservice to the facts.”

    I would reword this slightly: mutation and the other mechanisms listed at my blog at

    http://evolutionlist.blogspot......awman.html

    aren’t strictly “mechanisms of evolution”. Rather, they are sources of phenotypic (and genotypic) variation, which provide the raw material for the actual “engines” of evolutionary change.

    Those “engines” can best be described as any process by which particular genotypic and phenotypic variants increase in frequency and persist over time in populations, compared with alternative genotypic and phenotypic variants that decrease in frequency and eventually disappear from populations.

    Such mechanisms include (at a bare minimum) natural selection, sexual selection, and genetic drift (including both bottlenecks and founder effects). There may be other mechanisms as well (meiotic drive comes to mind), which empirical study may eventually illuminate.

    As I have pointed out before, much of the argumentation in this thread, along with John Sanford’s arguments for “genetic entropy” are based on the same fallacy that has undermined the “modern evolutionary sythesis”: that all evolutionary change is exclusively reducible to changes in allele frequencies in populations over time. This idea, which got it’s start around 1900, was an important stimulus to theoretical and empirical research into the mechanisms of evolution. It is, however, woefully out of date. Evolutionary biology has changed significantly in the past half century, and is currently entering what I suspect will eventually be recognized as its “expansionary phase”, similar to what happened to physics during the 20th century.

    Finally, Alexei Kondrashov, who used to teach both the majors’ evolution course and guest lecture in our non-major’s evolution course, moved on to the University of Michigan about a decade ago. I recommend checking out his biography on Wikipedia, which includes a brief list of his publications, along with his lab website at the University of Michigan:

    http://141.211.144.29/facultyr.....kondrashov

  31. JunkyardTornado:

    In neither of those is the word “mutation” mentioned once. Before even reading them I would not have assumed the researchers were crediting actual mutations at all as a factor in the changes taking place in a mere 36 years. It would seem evident to me that those changes had to do with genetic material that was alread latent in the previous population on the previous island.

    A few points:
    (1) From the news item:
    “Our data shows that evolution of novel structures can occur on extremely short time scales.” (2) They’re talking about “novel” structures because the cecal valves are found in only 1% of reptiles/lizards. (3) If it’s novel, that is new, then it had to come about someway. In other words, where did the genetic information for the cecal valves come from? How else does “novel” information come about except through mutations? Either you say that Darwinism can’t explain this, or you say that a non-Darwinian process explains it. Either way, there’s a big problem.

  32. 32
    JunkyardTornado

    Our data shows that evolution of novel structures can occur on extremely short time scales.” (2) They’re talking about “novel” structures because the cecal valves are found in only 1% of reptiles/lizards. (3) If it’s novel, that is new, then it had to come about someway. In other words, where did the genetic information for the cecal valves come from? How else does “novel” information come about except through mutations? Either you say that Darwinism can’t explain this, or you say that a non-Darwinian process explains it. Either way, there’s a big problem.

    If you’re still assuming the researchers thought that the lizards came to the new island and then mutations started occuring to account for the changes in 36 years, I think you are almost certainly wrong.

    If the new traits were present in the previous population, they could have been split-up – dispersed among other functionality or hidden altogether in nuetral configurations or whatever. No one’s claiming it only took 36 years for all that functionality to be created.

    You plugged 36 years and a beneficial mutation rate of 1.5*10-5 into your formula which I believe was completely erroneous.

    Just to repeat, new beneficial traits were emerging on the new island at a rate much higher than 1.5*10-5, because the ratio of beneficial to harmful genetic info already in the species was much, much higher than the ratio of occurence of beneficial mutations to harmful ones.

  33. Thank you Allen for your clarifications.

    One thing that should be evident from the graphs in the Excel Spreadsheet that I made available here is that evolution of the accumulated winnings of a statistically-advantaged gambler is extremely diffucult to estimate after the fact.

    For example, he could win 60 out of 100 hands, and it would be erroneous to suggest his inherent advantage is 60% vs. 40% or 20% (60%-40%). It may only be 1%, and worse yet he could be totally unskilled and his “advantage” is actually -2%.

    We have comparable challenges in statistically estimating inherent selective advantage in populations without the luxury of repeated trials (like the luxuries of repeated trials in determining the selective advantage of bacteria that have anti-biotic resistance).

    To settle the question of the Adriatic Lizards empirically, one would need to repeat the experiment probably a few times to see if the dramatic change can be repeated.

    If it is not repeatable, then we know random selection was responsible, not the inherent advantage of the phenotype. Unfortunately we don’t have the luxury of repeating the experiment as we do in other experiments like the evolution of anti-biotic resistance in bacteria.

    PaV,

    I’ll be meeting with Walter ReMine in Minnesota in month or so. I’ll try to visit this issue with him since he is more skilled at these topics than I.

    I’ll also try to run Allen MacNeill’s concerns about Sanford’s work by Walter. I think his concerns shouldn’t be ignored.

  34. JT:

    No one’s claiming it only took 36 years for all that functionality to be created.

    I’ve used classical neo-Darwinist equations for fixation of a mutant allele in a population. According to this equation, assuming a selective advantage that is quite high, ONE mutant allele (= trait) will be fixed. Now which trait was fixed? Was it the larger skull, or the larger jaws, or the cecal valves, or the changed behavior? The classic equation I’ve used simply means that you’re only going to get at most ONE phenotypic change in this time period. This, in turn, means that rapid phenotypic change has occurred in a non-neo-Darwinistic manner. So why should we believe in neo-Darwinism?

  35. Allen wrote:

    Such mechanisms include (at a bare minimum) natural selection, sexual selection, and genetic drift (including both bottlenecks and founder effects). There may be other mechanisms as well (meiotic drive comes to mind), which empirical study may eventually illuminate.

    This leads to an interesting discussion for which I leave open at this time. Is it legitimate to lump all of the bolded mechanisms into a radom variable, where this random variable represents the behavior of “random selection.”

    Analogously we can model all of the numerous complex mechanisms and factors in the trajectory of a die over craps table (i.e. air viscocity, materials characteristics of the die and the table, the throwing mechanisms, the landing zone characteristics, die instability defined by tensor analysis, etc. etc.) — but from an operational stand point, and as far as the casinos are concerned, the outcome is that each die has 1/6 probability of landing a particular number. The minutia of the physics involved does not figure into their bottom line….

    It might be disconcerting to lump all of the bolded mechanisms into a random variable, but I’m not sure it is totally illegitimate, unless of course more analysis of those mechanisms leads to insights into the appearance of design, or the paradox of purpose.

    Whether more insight into the mechanisms can actually explain organic change is an open question, but the ID proponents argue that if these mechanisms themselves are inherently stochastic, the net effect of all of them will also be stochastic (modeled by a random variable).

    It is a topic worth exploring…unfortunately I’ll have to save it for another day…

  36. Allen:

    Let me be brief. (1) If there are on-going “mechanisms of evolution” and so many “sources of variation”, then how do you explain stasis? (2) ID is based, essentially, on information theory. Certainly Wm. Dembski uses this very area to mount his argument in favor of ID. When you push aside “alleles”, and RM + NS, then we are left with NO source of the information contained in these “alleles”. To me, maintaining such a position sounds the death-knell for any Darwinistic explanation of evolution.

    (I’ll just add that I’ve looked over all 47 ‘sources of variation’, and save for the first three, everything else looks like already present information being switched around, or simply be activated or unactivated. But, of course, the first three represent classic neo-Darwinism.)

  37. 37
    JunkyardTornado

    I’ve used classical neo-Darwinist equations for fixation of a mutant allele in a population. According to this equation, assuming a selective advantage that is quite high, ONE mutant allele (= trait) will be fixed. Now which trait was fixed?

    I’ll have to plead a degree of ignorance about specifically what you’re referring to here, and I how it relates to my point above.

    But cannot dog breeders fix a whole set of traits in a few generations? How would your point apply to that? Of course, they’re killing off everything that doesn’t adhere to the traits they’re looking for. But even if the selection was not that extreme on the new island, isn’t the same principle at work.

    Also:

    Belyaev and his colleagues thus selected the foxes that exhibited the least fear/shyness of people for their breeding program; their aim was to selectively breed for the tame trait. With successive generations of selective breeding the foxes became tamer and tamer such that by the eighteenth generation they had bred a fox that exhibited all the characteristics of a domestic dog. The foxes would actually approach people, clamber over them, roll over to get their bellies tickled and even answered to their names.

    Perhaps the most surprising aspect of the selective breeding program was that the foxes no longer resembled foxes but looked and acted like dogs. The coats of the foxes were no longer the characteristic silver fur much sought after by the fur industry but were black and white piebald instead. What is more, the foxes’ tails were curly and upward turned, their ears were floppy and to crown it all the animals even barked like dogs! Dmitri Belyaev had not sought nor bred for these characteristics but all the same they still manifested!

    Of course in the above situation, only one trait was selected – tameness. But it resulted in a whole array of other traits being selected as well. But dog breeders in general I believe do select for multiple traits.

  38. The surprising appearance of cecal valves in these lizards may very likely be the result of the “up-regulation” of a genetic program previously suppressed in these lizards. The fact that cecal valves have not been observed in their close relatives is no evidence at all for the hypothesis that the genes for such valves are present in these lizards, but have not been expressed until recently.

    I’m glad I read through the comments first. I was about to express the same thought, except with a different interpretation. Engineers will often design functionality that goes unused unless particular stimuli causes a triggered event (a function that is generally unexpressed except under certain conditions triggered by other functions or changes in input/system). I believe that such observations could be an avenue for ID-oriented research: looking for foresighted mechanisms.

    Just the other day I was discussing this very subject with a friend. A good Designer would program biology to be proactive, to respond to an ever-changing environment. I used the example of Pseudomonas aeruginosa and it’s nylon-eating capabilities within 9 days, which some ID proponents have inferred may implicate foresighted mechanisms. I do not believe that hypothesis has been adequately explored yet. Merely saying that the processes of the modern synthesis and chance cannot account for it is not enough in my estimation. I would prefer that someone try and determine exactly what is triggering the change. It may also be that the design of the system itself allows such rapid evolution and not externally-triggered mechanisms (sort of like how the designed shape of legos allows many configurations). In any case, the modern synthesis proves useless.

    These Adriatic Lizards may be a similar avenue for ID research, although obviously a more difficult route. I think it’d be easier to observe bacteria.

    entertaining the possibility that an omnipotent supernatural intelligent designer with an inordinate fondness for herbivorous lizards would be somewhat premature.

    I don’t even consider that a serious option. Although I obviously believe there is a potential investigation for ID proponents, I don’t see why it must be limited to assuming direct intervention.

  39. JunkyardTornado,

    That’s artificial selection. Obviously selection is not toothless, else GAs would not work at all (see here). But they rely on intelligent funneling, also known as active information. The selection filter must be balanced properly in order to work (the search be funneled toward the target).

    Darwinists tend to over-focus on when NS does indeed work and not when it does not. The major issue is that natural selection is apparently not operating with regard to SOME traits. As in, it’s not operating uniformly but with only limited cases.

    The reason I think this is an issue is since natural selection usually relies on environmental factors. While some factors are generalized, some factors must be very specific in order for the funneling effect to work. What if the factors are very rare or don’t even exist? That means that in order for the modern synthesis to work not only does CSI have to emerge it must be paired with a rare event that offers selective pressure. As in, to provide positive evidence that Darwinism works as advertised you must provide a reasonable set of factors that would be enough for directional selection to work. Resorting to “deep time” and some rare lucky set of events seems a cop out. But then Darwinists would say that “I lack imagination”.

    EDIT:

    To save time, see this link:

    http://www.uncommondescent.com.....says-prof/

  40. I just picked this up at Michael Behe’s blog at Amazon.com.

    An interesting paper appeared recently in the New England Journal of Medicine. (1) The workers there discovered some new mutations which confer some resistance to malaria on human blood cells in the lab. (Their usefulness in nature has not yet been nailed down.) The relevance to my analysis in The Edge of Evolution is that, like other mutations that help with malaria, these mutations, too, are ones which degrade the function of a normally very useful protein, called pyruvate kinase. As the workers note:

    “[H]eterozygosity for partial or complete loss-of-function alleles . . . may have little negative effect on overall fitness (including transmission of mutant alleles), while providing a modest but significant protective effect against malaria. Although speculative, this situation would be similar to that proposed for hemoglobinopathies (sickle cell and –thalassemia and —thalassemia) and G6PD deficiency. . .”

    This conclusion supports several strong themes of The Edge of Evolution which reviewers have shied away from. First, that even beneficial mutations are very often degradative mutations. Second, it’s a lot faster to get a beneficial effect (if one is available to be had) by degrading a gene than by making specific changes in genes. The reason is that there are generally hundreds or thousands of ways to break a gene, but just a few to alter it beneficially without degrading it. And third, that random mutation plus natural selection is incoherent. That is, separate mutations are often scattered; they do not add up in a systematic way to give new, interacting molecular machinery.

    Behe is making the same point I’m making here: RM + NS cannot give you information.

    Using the equations of the RM + NS approach to evolution cannot really provide an answer as to how such large-scale change took place in such short a period of time.

  41. Allen MacNeil (19):

    The surprising appearance of cecal valves in these lizards may very likely be the result of the “up-regulation” of a genetic program previously suppressed in these lizards.

    I agree with you 100% Now, what “caused” this “up-regulation”? Did the lizards know that they were on a new island? Did they notice that there weren’t as many of them as before?

    I’m being facetious. But the point is is that there must have been some kind of environmental trigger. I propose it was the changed diet. I further propose that it has to do with the proteins and protein by-products of the newly found plant life.

    However, it would seem that all of this presupposes the “front-loading” of both necessary genes and necessary developmental programs.

  42. Patrick (37)

    I’m glad I read through the comments first. I was about to express the same thought, except with a different interpretation. Engineers will often design functionality that goes unused unless particular stimuli causes a triggered event (a function that is generally unexpressed except under certain conditions triggered by other functions or changes in input/system). I believe that such observations could be an avenue for ID-oriented research: looking for foresighted mechanisms.

    This has been pointed out before at this blog. An intelligent designer would likely design with certain redundancies (duplicated genes) to prevent fatal losses of information, would make the genetic program as efficient as he could (homeotic genes), and would envision adaptation to changing environments (highly regulated genetic networks).

    As to research, how do you do research when they run you out of labs and universities and deny you funding?

  43. A full article on the fox domestication program:

    Current Biology: Selection for tameness has changed brain gene expression…

  44. Thanks, Atom, for the link. I find it fascinating that you can get foxes that look like dogs in such a short period of time.

  45. how do you do research when they run you out of labs and universities and deny you funding?

    I’m fully aware of the problems faced. Family comes first. Thus most ID proponents focus on what pays the bills, with ID mostly being a hobby (and a dangerous one, at that). Still, I wonder if the entire ID community pooled it’s resources it may be possible to fund undeniably ID-focused research.

    But, really, in today’s environment I think the only way such research would be funded if any discovered foresighted mechanisms were glibly written to be a product of evolution (along the same lines as modularity, due its beneficial nature) via a disclaimer sentence.

  46. 46
    JunkyardTornado

    Patrick wrote: But, really, in today’s environment I think the only way such research would be funded if any discovered foresighted mechanisms were glibly written to be a product of evolution (along the same lines as modularity, due its beneficial nature) via a disclaimer sentence.

    [My response to the above is actually in the last paragraph]

    I’m hoping that most people in ID or otherwise would view something like a dog as a mechanism. So for example if we were trying to explicate what a dog did in a given circumstance we could elaborate it in terms of stimuli, the dog’s basic instincts, pertinent attributes of the dog’s immediate environment, the processing capacity of a dog’s brain – how much and how detailed information he was capable of storing in his brain and for how long, and so on, and quite literally we could give a detailed physical explanation of the why the dog did what he did. And yet, certainly a dog has foresight as well, so that a dog can have an objective, either innate or conditioned, that guides his behavior towards some goal. But all that foresighted behavior can still be elaborated and explained purely in terms of a mechanism.

    We could look at a complex weather system as well, and as complex as it is, potentially explain its behavior in terms of a mechanism. Certainly even a weather system could have a complex saved state inherent within itself that dictated how it interacted with its environment (say the weather system we’re talking about is a tornado). A dog also has a complex saved state, most of it stored in his brain. But a tornado as well could have a complex state that its in that would determine how it interacts with its environment and deteriming what it ultimately does. Someone might say the difference is that a dog has “awareness”, meaning that he has sensory organs that are activated via light or sound and send chemical signals to his brain. But a tornado as well possesses some margin at which point whatever is external to it impinges on it, with the results of that interaction ultimately reverberating into the heart of the tornado itself, so that stuff happening outside of the tornado is reflected inside of it. Stuff that happened a long distance from a tornado could ultimately change the internal state it was in, just like things happening a long distance from a dog can ultimately change the internal state it is in and alter its behavior.

    However, purely as a matter of convention we do not attribute things like foresight to the weather. As a matter of convention, we attribute foresight to biological things. But from strictly an objective standpoint there is no reason why we could not or should not attribute foresight to evolution, the weather, or anything else. So a research project to identify mechanisms with foresight could identify any mechanism at all, as long as there were a nonrandom correlation between the behavior of that mechanism and the thing we were trying to explain. The less degree of randomness in that correlation, the higher the degree of design and foresight.

  47. 47
    JunkyardTornado

    the following might strike some as a little naive at times, but I would like to go over it all to help clarify my position on things. (If I may.)

    Consider the most human-like robot that man will be ever even conceivably be capable of achieving. Now consider the program for that robot – a big humongous binary string – call it y. Is it not relevant that there are an infinite number of program-input combinations f(x) that can output y? What difference does it make whether f’(x’) has foresight according to some vague philosophical standard and f”(x”) does not? Note also that y might be encoded according to any one of a number of different standards and it would still be y. In the same sense f(x), whatever it is, just another encoding for y itself, because it outputs y. Here’s the explanation: The computer that f(x) is running on is irrelevant, because a computer at its core is an extremely simple device and all the relevant information is in the program itself. So if C is the computer, then C(f(x)) = y. But C is simple. What’s the difference between C and some program that just translates y between one arbitrary encoding scheme and another. So any preexisting conditions x and mechanism f that you are able to discover in nature that together output y ((where let’s say now that y is an actual human), then that f(x) will just be an alternate encoding for y itself, so that means it takes something equating to y to output y.

    Someone could say, “But those prexisiting conditions and natural laws couldn’t think and act. What we have today can.” But if you could look at Einstein as a baby and say, “A long time from now he’s going to revolutionize science.” If that observation is true, you’re in fact saying something about Einstein as a baby as well, even though he did not accomplish all that as a baby. The same thing occurs if you look at some set of conditions a zillion years ago and some mechanism of nature that acted on it. That f(x) is equivalent to what we have today. Of course with evolution, x (the mutations) are spread out over a period of time, but its the exact same principle.

    So as far as evolution, if someone says, “This trival set of laws worked over a period of time to transform some input x (i.e. some mutations) into the world we have today.” Its easy to see that if those laws are extremely trivial, how really they equate to a simplistic program like a decoding program, and what they decoded really already had all the information, (so IOW, its almost the same as the random mutations without any sort of processing by natural selection just output the biological world on their own.) Of course f, the natural laws, could be more complex. But whether f is more or less complex, if f(x) outputs y, then f(x) equates to y.

    That all being said, things in the physical universe tend to flow predictably from previous conditions and natural laws. In spite of quantum theory, thats a fundamental assumption. We look for things existing in the past and the operation of various processes to account for what we see today. So even though the fact that if f(x) outputs y then f(x) equates to y implies you’re just constantly pushing back what needs to be explained, its nevertheless really the only legtimate activity of science – to explain conditions today on the basis of preexisting conditions and identified mechanisms. And also it defies comprehension to suppose the vast amount of energy in the unverse was not directly exploited in the creation of mankind. Furthermore whatever the preexisting conditions and natural laws were that led to us, they had to have looked very different than what we have today (in fact its self-evident since we didn’t exist back then.)

    In some ways the above is similar to The Design Inference, except with TDI, there is a huge emphasis on the idea that intelligence is not a mechanism. But that notion completes obscures the entire analysis in a number of ways. You have this mysterious “thing” that cannot be described that is just all over the place supposedly, but you can’t pin down where with any certainty, or account for the fact that mechanisms can output CSI as well, and so on. OTOH, if intelligence is not a physical mechanism, it seems obvious that it could not originate through physical mechanisms and nothing more really needs to be said. But if intelligence is not a mechanism then it can’t be described at all, and then you’re attributing all these things to something that can’t be described, and what’s the point of that.

  48. 48
    JunkyardTornado

    And just to close the loop, and repeat some ideas that I’ve maybe repeated two or three times before in the five months or so I’ve been here. (Its off-topic, but I’m probably just writing for myself at this point, and some repeat their major ideas in nearly every post):

    On the subject of design, clearly that is not something in the exclusive purview of some nonmaterial mind. Mechanisms do it all the time. Design is a synonym for Search. Design is searching a solution space for a solution that adheres to some set of criteria. The search could involve iterating through physical configurations of an item by physical manipulation, followed by testing. It could involve a simulation of such a process in a different medium (e.g. a brain) where the tradeoff would be speed for accuracy, wherein the alternate medium allowed for more rapid iteration. However actual design nearly always involves building numerous actual physical prototypes and then testing them. A search also equatable to design could alternatively entail an actual physical traversal of a region looking for a complete or nearly complete solution already extant in another locale. (Thus the latter half of the “make or buy” decision.) All of these are search. All are design. But clearly cannot a computer be told, “Go find some solution that adheres to such and such criteria.” If someone does not understand that computers do this all the time they understand very little.

    As far as the solution criteria, there is no requirement (nor is it hardly ever the case) that such criteria emanate mysteriously and inexplicably from the designing entity itself. Given that design of any consequence nearly always entails resources (i.e. energy) that exceed the capacity of any one individual, the design criteria, or at least a substantial portion of them, are generally imposed on an individual from outside sources who also have a vested interest. Thus an engineer is told, “We need a tool with such and such properties and capabilities and for this amount of money.”

    But design can also proceed incrementally and haphazardly by several individuals not working together at the same place or even the same time or even towards the same goal. Because civilization can and does record for posterity the achievements of a generation, their designs can be refined upon, and even drastically altered in subsequent generations, in ways that were never originally envisioned or imagined. Many artifacts could not have emerged apart from the work of hundreds if not thousand of individuals working completely independently over thousands of years. The capability of civilization itself to design in this way is directly tied to the ability to write, and thus a body of knowledge continually grows and grows and can be exploited by subsequent generations. For a person of our day to look at his designs and marvel over his seemingly godlike capability, and yet be oblivious to the extent that he is merely exploiting and in essence taking credit for a body of knowledge that in fact accumulated over millennia, is the most extreme example of shortsighted self-delusion imaginable.

    But design requires a civilization to be functioning in a very specific way. If every individual must spend all waking hours searching for food, then it severely limits culture and the possibility of any significant design at all. Certain peculiarities of geography as well as the availability of certain types of plants (wheat, barley) and domesticated animals (large group-oriented herbivores with relatively short gestation) are what make possible the sort of food surpluses necessary to free up a portion of society to direct their energies elsewhere, and start society in an upward spiral (A lot of this is common knowledge I guess.)

Leave a Reply