LONG WINDED VERSION AT UD:
Darwin’s Delusion vs. Death of the Fittest
CONCISE VERSION AT TSZ
From Kimura and Mayurama’s paper The Mutational Load (eqn 1.4), Nachman and Crowell’s paper Esitmate of the Mutation Rate per Nucleotide in Humans (last paragraph), Eyre-Walker and Keightley’s paper High Genomic Deleterious Mutation rates in Homonids (2nd paragraph) we see that by using the Poisson distribution, it can be deduced that the probability P(0,U) of a child not getting a novel mutation is reasonably approximated as:
where 0 corresponds to the “no mutation” outcome, and U is the mutation rate expressed in mutations per individual per generation.
If the rate of slightly dysfunctional or slightly deleterious mutations is 6 per individual per generation (i.e. U=6), the above result suggests each generation is less “fit” than its parents’ generation since there is a 99.75% probability each offspring is slightly defective. Thus, “death of the fittest” could be a better description of how evolution works in the wild for species with relatively low reproductive rates such as humans.
One minor point, it’s a bit of a misnomer to say call it “death of the fittest”, since by definition, the fittest are those with the highest rate of reproductive succes.
The fittest would never preferentially die out, if they did, they wouldn’t be “the fittest”.
A mere quibble. The fittest are what we decide to call them, not what is more successful. 😀
Edit to add:
This is precisely why ID nakes no sense.
IDists wan to define fitness as something other than reproductive success. You can see the results of ID thinking at any dog or cat show. Carefully selected individuals that could not survive a week in the wild.
I applaud the concision, but a concise post should still be logically coherent. Why do you assume that all these mutations are deleterious? How are you measuring the quality of “defective,” if not by reproductive success?
Where Sal has a good point I think is that there is a certain circularity in the equations if we consider them as empirical tools: we define fitness as whether a trait becomes more prevalent or not, yet we also acknowledge that traits can become more prevalent for reasons that are completely orthogonal to anything that might promote survival. For example, a trait might become more prevalent simply because by chance no human in clunky boots happened to tread on that lineage, whereas a neighboring family of small critters that, in the absence of a human boot, would have done rather better, happened to get stamped (literally).
However, the conundrum is easily resolved if we remember that these equations are entirely theoretical, and so we can postulate an omniscient Experimenter who can run the history of the population several times, the mutations, and systematic environmental variables remaining identical, but irrelevant historical events differing. And we can actually do this with computer simulation, and can show that sometimes traits that we know to be systematically deleterious, i.e. traits that overall tend to reduce the probability of reproductive success over several runs of the experiment, can nonetheless go to fixation on some runs. If all we had was the run in which that trait was successful, we’d be unable to know whether the trait went to fixation because it was systematically advantageous, or because it just got lucky on that run.
But that doesn’t mean there is anything wrong with the concept. In fact, we don’t even have to run the thing many times. I can set up a simulation where I set the odds of reproduction of one variant at something like .999999999999999 (which I know, because I typed it in) and note that it goes to fixation, whereas one that I set at 1.000000000000000001 goes extinct. And we can do it just by math alone (if we are good enough at integrating lots of intersecting pdfs – I tend to run Monte Carlos because I’m not).
And what we find is that using such a model, as Kondrashov claims, if the population is small enough, and the genome big enough, “truly” disadvantageous variants that are only Very Slightly Deleterious (VSDMs: Very Slightly Deleterious Mutations), build up even though when measured in terms of their “success” they, are, by definition, “succeeding”.
However we know empirically this doesn’t happen. Which is raises an interesting question, but one that is fairly readily explained by a number of different mechanisms, including the fact that as a trait becomes more prevalent, the selection coefficient of that variant itself is likely to change, because the prevalence of the trait is itself part of the changing environment. A trait that is “slightly deleterious” when only a few individuals in a population have it can easily become “slightly beneficial” when shared by most, and vice versa. “S” in other words, is itself a function of the other parameters in the equations.
Another is that most phenotypic traits are polygeneic. I think myself, at least when considering sexually reproducing populations, it’s a mistake to think that a new mutation either has an instantly beneficial or an instantly deleterious effect. It’s far more likely that most mutations are near-neutral and that therefore there is a constant drip-feed of innocuous variants into every population, drifting through it, or not, simply by chance; however that richness is precisely what serves the population well when the environment changes, and the optimal mean values of each trait change, and thus the prevalence of the various alleles that contribute to that optimum also change, those organisms with the cocktails that deliver phenotypic values closes to the optimum surviving to pass on their cocktails more often than those dealt a less optimal hand.
Learned Hand,
If we assume human chimp split, based on sequence analysis, the number of mutations per individual per generation is around 56 to 160. On the extremely conservative estimate that 11% are bad, and using the lower mutation rate of 56, you can get U= (56*11%) = 6.
Muller showed U=0.5 would be enough to eventually terminate humanity, but the U=0.5 scenario is not as obvious as the situation as U=6. What U really is equal to is hotly debated. If we include non-coding regions as functional, then U is easily at least 6, hence the debate over “junk DNA” is pretty important to this thesis.
The problem is if one defines “fit” as reproductive success, then one can conclude blindness in cave fish is a “fit” trait. That’s all well and fine, but by doing so, one has decoupled the necessary relationship of functionality to fitness, and hence reproductive success does not necessarily imply increase in complexity as Dawkins argued in blind watchmaker.
1. some will argue silent mutations aren’t important, but that claim is in doubt.
2. knockout experiments that show no change in reproductive fitness can still disable functionality, I gave examples of the failure of knockout experiments in C. Elegans, Buchnera and Wigglesworthia to change reproductive fitness even though there was a change in functionality
http://tinyurl.com/2xea4g
If not by reproductive success, how do we define “fit”? One might try defining fit by function (as we do in medicine). Dawkins equivocates the two notions (reproductive fitness and functional fitness), and that is incorrect. Darwin does the same.
When you toss together near neutral mutations and pure luck in environmental variables, you get something other than pure “progress”.
I think IDists cannot toleration the idea that evolutionary change has no knowable direction.
sez stcordova:
Hold it. Your calculation assumes no beneficial mutations, Mr. C. Since beneficial mutations do, in fact, exist, your calculation is incomplete and does not reflect Reality As She Is Spoke. In this, you are following in the footsteps of your fellow ID-pusher Michael Behe, whose ‘proof’ that evolution cannot generate Irreducibly Complex systems is fatally marred by its refusal to acknowledge the existence of two fo the three general classes of evolutionary change. Behe acknowledged the “add a new bit” general class, but ignored the “remove an old bit” general class and the “modify an old bit” general class; you acknowledge deleterious mutations, but ignore beneficial mutations.
I saw the longer version posted at UD, and decided that I didn’t have the patience to read it.
I don’t see what you describe as a problem for evolution. It is perhaps a problem for Darwinism, but then I am not a Darwinist.
Huh. Seems to me that what’s wrong with Corova’s question is like the economists’ old joke:
We know how this works in practice, but does it work in theory?
It’s bizarre to prefer a math formula which “proves” that we must have all died out from mutations over the actual reality that we are very much alive and thriving.
Makes even less sense if we apply the same question for a species with relatively fast reproduction which have had orders of magnitude more generations to accumulate this hypothesized deleterious-mutation load.
Now, I recollect that Cordova is a YEC or at least leans that way, but positing only 10,000 (maximum) years of “each generation less fit that the previous” is still impossible to reconcile with the evidence that mice, fruit flies, brine shrimp – and bacteria! – are still among the living, not genetically degraded to extinction after hundreds of thousands of generations within that 10 millennia.
What’s wrong with your question, Cordova? Why doesn’t your question match what we know works in practice?
Oh absolutely, I’ve… complained about the somewhat counterintuitive way fitness is defined in population genetics before. Reproductive success is measured in relation to the entire population’s average. Those with “high fitness” simply have above-average reproductive success. Since there’s always going to be differential reproductive sucess(for whatever reason, whether selection or drift etc.), some will always be “more fit” than others, even in a population in overall (fitness) decline and possibly heading towards extinction.
And then there’s the great ironies about fitness and “beneficial mutations”. No amount of gatling-gunning out offspring with novel enzymes and regulatory elements is going to out-evolve the oh-shit-my-planet’s-atmosphere-and-crust-now-mostly-consists-of-molten-lava-active-volcanoes-and-sulfuric-acid-induced-by-asteroid-shower. 😛
This appears to be nothing more than an attempt to gussie-up “genetic entropy” with “impressive math;” and we already know genetic entropy is bogus.
It’s in the same genre as the tornadoes-in-a-junkyard calculation; it’s irrelevant.
Yeah; who knows what dinosaurs might have achieved if it hadn’t been for that inconvenient asteroid.
There is also a price to be paid for the increased complexity that comes with “higher” levels of evolution; and that is delicacy. The structures that allow for many more degrees of freedom and many more directions in which to evolve are also the very structures that are more subject to assaults by the environment.
For example, the narrower the temperature range in which an organism can survive, the more sensitive it becomes to mutations. Whether or not a mutation is deleterious depends on whether or not it continues to narrow the temperature range in which the descendents of the organism can survive.
Minor point, but that is NOT what Dawkins argued in The Blind Watchmaker. He did not suggest that reproductive success necessarily implies increase in complexity.He does provide compelling arguments that evolution (undirected mutation with natural selection) is sufficient to explain all the biological complexity we observe.
That’s quite a different case than your misstatement.
And – how are you defining “functionality”?
stcordova
.
Mutation rates of that order (50-160) are typically produced by the various methods which are used. Larry Moran had an interesting series of posts on this at Sandwalk.
One would take issue with the ‘extremely conservative estimate’ that 11% are bad, however. Ohno’s paper introducing the term “junk DNA” into the lexicon used the mutational load as compelling evidence that large swathes of our genome cannot be functional, in the sense that it could suffer a deleterious mutation, otherwise the lineage would be doomed by mutational meltdown. Yours is the very issue Ohno set out to address, and the conclusion that the genome is degrading is not the one generally reached, but instead, the genome is largely impervious to mutations – much is nonessential; ‘junk’ if you will.
My own view is that 95% of the genome is junk. It would still not be a justifiable assumption that every mutation to the important 5% is necessarily bad. The vast majority of sites in most proteins are substitutable, with greater or lesser latitude on chemical property depending on where it is.
Of couse, you can argue that my estimate is hopelessly inadequate, and that ENCODE (for example) has shown something that it has not – the ‘death of junk DNA’. But the mutational load argument still stands (with extensive backup from genomic investigation : it’s mostly dead transposon). We are not obviously in any state of genetic decline, despite our mutation rate.
cubist,
” you acknowledge deleterious mutations, but ignore beneficial mutations.”
Thanks for raising the objection.
For proteins, only a very small fraction (less than 0.1%) of mutations are conceivably beneficial because of the scarcity of functional proteins in the space of all possible amino acid polymers. The rest of the mutations are neutral to deleterious. The ratio of beneficial to non-beneficial is negligibly small. So as you can see, there is justification for not weighing in beneficial mutations.
Codon Bias suggests every nucleotide that codes for a protein has importance, and even supposing it doesn’t have importance, it then raises the equally difficult question of why it is there in the first place. Either interpretation of its existence (functional or non-functional) is equally difficult in terms of random mutation and natural selection.
Even though a silent mutation and sometimes a missense mutation might not immediately affect reproductive fitness, many such mutations that go unchecked, will compromise the organism like a slow oil leak eventually compromising an airplane.
To illustrate, losing a spare tire in a car or spare navigation device in an airplane may not be noticeable in terms of immediate performance, but the system is functionally compromised nonetheless. Similar analogies apply to functional systems in biology with respect to slightly deleterious mutations.
Notes:
Consider E. Coli Frail Hypothesis in Evolution
the deleterious/beneficial mutation ratio is assumed to be as high as four to five orders of magnitude, implying that E. coli’s genome is fully optimized with respect to single nucleotide substitutions. The deleterious mutation rate would be higher than 2×10−4 per genome replication, or about one tenth the mutation rate. On the other hand, “the proportion of mutations that are beneficial is roughly one in a million”
This is circular Sal: you only know if a protein is “functional” if you know whether it can be “beneficial”. So your ratio doesn’t mean anything. A protein can be functional under some circumstances, and when expressed in some tissues, and deleterious when expressed in other circumstances and/or in other tissues. And that’s complete ignoring mutations in genes that affect the circumstances under which a protein-coding gene is expressed.
I think this confusion arises from a lack of clear distinction between the advantageousness phenotypic trait (the effect of a cocktail of alleles on an actual organisms) such as “longer legs” or “faster reaction times” and the “advantageousness” of a specific allele. Because most traits are polygeneic, what matters for fitness (which is a property of an organism, not the property of an allele) is the net result of a whole range of alleles, including regulatory as well as protein coding genes.
There is no reason to conclude this. If a mutation doesn’t affect reproductive fitness, it doesn’t affect reproductive fitness. If it only affects reproductive fitness when inherited in combination with another allele, then both alleles will tend to drop in frequency, unless the other is beneficial alone, in which case the first will tend to drop in frequency as the second becomes more prevalent. Perhaps you are suggesting that some alleles only slightly affect reproductive fitness, but not enough to remove them from the gene pooll, but nonetheless build up. But again, you are ignoring the fact that an “deleteriousness” or “advantageousness” is not an property of an allele but of an allele in the context of a genome within an organism within an environment, and all of those are interacting variables. It’s just not “like” a “slow oil leak”. It’s a nice metaphor, but that doesn’t mean it’s a good description!
Another important point that you might like to consider, Sal, is that the ratio of “advantageous” to “deleterious” mutations will be higher when a population is not well-adapted to an environment (for example, if the environment has recently changed quite radically) than when it is extremely well-adapted. For a well adapted population, it will take a pretty extraordinary mutation to do better than what is currently available, while for a poorly adapted population, many alternatives may be better than what is on offer. Thus a top predator in a stable environment like a shark in the open ocean may remain mostly unchanged for millions of years, while a population that finds itself stranded on an island may change very rapidly to adapt to a new and specific set of threats and resources. Which of course was what Darwin observed in the Galapagos, although he had no idea just how finely poised those populations were, responding to really very short cycles of wet and dry weather.
GAs often stop making progress when near ‘optimal’ solutions, or at least local maxima.
Yes indeed! Right at the point when a mere human can “see” the optimal answer!
That is not a correct statement. Mutations in existing life are NOT exploring the space of all possible amino acid polymers. They are exploring the immediate next-door neighbors of already-existing functional proteins.
If it were true that less that 0.1% of mutations could be beneficial, it would not be because of anything to do with your ID confusion about “search space”.
If meltdown were a serious threat it would show up first in somethjng like HIV.
hotshoe wrote
That’s a point I’ve made in Sal’s presence more times than I care to remember, going way back to ARN days. The random search metaphor for evolution by natural selection is a snare and a deception. It snares the unwary, and is used deceptively by Dembski & Marks and their acolytes, Sal among them.
“That is not a correct statement. Mutations in existing life are NOT exploring the space of all possible amino acid polymers. They are exploring the immediate next-door neighbors of already-existing functional proteins.”
I didn’t say it explores all possible amino acid polymers, I said the space of functional proteins occupies only a small space of amino acid polymers. But I wasn’t being clear and I failed to actually engage your first objection. So I will try again.
Let me take you up on your suggestion. How about exploring the immediate neighbor? Say the neighbor only involves changing simultaneously a measly 3 amino-acids, That specificity will require 1 out of 20^3 amino acid combinations to hit the target, or 1 out of 8000 (it’s a bit more complicated than that because of codon degeneracy, but that’s a rough estimate). So, that’s a 0.01% chance which is less than 0.1% that I suggested. That’ also renders RBH objection moot as well since this is only a small step to the neighbor. 0.01% is pretty generous given the E. coli study gave a probability that is 0.0001% (1 in a million).
As far as ARN days, I was posting on origin-of-life. So it’s rather inappropriate to invoke evolution from a pre-exiting protein that doesn’t pre-exist.
A protein isn’t functional without being a part of larger system, like a key without a lock or password without a system to incorporate it. So even supposing you can make a key an infinite number of ways, there is still the probability associated with making a corresponding lock.
You can look at it like the probability of building a key and a corresponding lock via random matching. There might be an infinite space of possible lock and key combinations, but for a specific key you need specific lock. That corresponds to proper binding. So when one says functional protein, it really implies a functional protein system.
A given lock that enables a key to be functional with a specificity with a mere 3-amino acids will yield probabilities of the order suggested above.
The improbability of biology isn’t solved by saying we can build a lock-and-key system an infinite number of ways, it is the improbability of building lock-and-key systems with various levels of specificity for each other that we see in living organisms.
But perhaps to settle the argument, everyone is invited to provide their estimate of beneficial, neutral, and deleterious mutations.
Is the ratio of beneficials relative to all mutations:
0.01%
0.1%
1.0%
10%
But even granting we have new innovations, that doesn’t solve the problem of fixing broken parts. This is like getting a new fancy set of wipers but the battery starts to degrade. The addition of good doesn’t heal the broken parts.
Anyway, I appreciate everyone’s criticism of what I asserted. Thank you for taking the time to read and respond to what I said.
I apologize for being verbose, but I was trying to respond to multiple criticisms at once, and I think they are good criticisms. Thank you.
So many confusions, so little time. I’ll tackle the argument in the OP right now, and leave very slightly deleterious mutations until another time.
The equation in the OP correctly (in a reasonable model) describes the probability of an offspring maintaining an absolutely pristine genome, given a certain deleterious mutation rate. There are several reasons why this number is unlikely to present a major problem for species like humans, however.
First, stcordova’s estimate of the deleterious mutation rate is likely much too high. The best estimate of the fraction of the genome that is really functional comes from ENCODE, and that puts it in the range of 9% – 15%; within that fraction, many sites are still more or less permissive of mutation. For example, four-fold degenerate sites are in coding sequence, which is universally classified as functional, but they can experience any point substitution without deleterious effect. Regulatory regions tend to be more weakly constrained than coding sequence and thus can tolerate even more mutation. Thus, the total fraction of mutations that are deleterious is probably well under 10%. No one really knows the number, but something in the range of 2 – 3 per human birth is a reasonable guess.
Second, an unknown fraction of deleterious mutations are lethal to the embryo, to the zygote or to the gamete. These contribute little or nothing to the effects of genetic load, since they waste little reproductive capacity.
Third, the formula provides the probability of maintaining a pristine genome, but of course we don’t maintain a pristine genome. In fact we each carry probably thousands of deleterious variants. If new deleterious mutations are introduced into a pristine genome in every generation, the number of deleterious variants will increase until it reaches an equilibrium value; at this value, the number of deleterious variants removed each generation (by the preferential loss of the least fit, who have a great many more deleterious variants than the most fit) equals the number of new ones introduced by birth. (In the simplest model, this number can readily be calculated from the formula for selection-mutation balance.)
The real argument about genetic load is that the equilibrium number of deleterious alleles reduces the absolute fitness of the entire population to the point where it must go extinct, that is, the reproduction rate is reduced to below the replacement level. The reason you can’t conclude this from the Poisson formula, however, is that not all deleterious mutations actually reduce the absolute fitness of the population; in fact, some can increase it. That’s because some selection is “soft” (in Wallace’s term) rather than “hard”. That is, a deleterious mutation may make an organism less reproductively successful than competitors, without reducing the overall reproduction rate if the entire population has the mutation. Most variants that affect competition for mates, for example, or competition for space and light with plants, are likely to experience primarily soft selection. So the actual reduction in reproductive capacity, at the equilibrium number of deleterious mutations, is certainly less than that predicted from the simplistic calculation. How much lower it is is anybody’s guess, since there are very few estimates of the relative contributions of hard and soft selection. It may be quite a lot.
Fourth, estimating the reduced absolute fitness based on the total number of accumulated deleterious mutations is basically absurd, since the comparison is to a perfect genome that has never existed. It might be meaning to ask how much the genetic load of higher primates (which have fairly high deleterious mutation rates) reduces their reproductive success relative to the load in earlier primates; what matters here is the change in the number of deleterious alleles carried per individual. Abstract arguments based on the absolute number of deleterious alleles are meaningless, however, since the reproductive fitness of a genetically perfect human something that has any referent in reality.
Hi, Steve. Welcome to TSZ. Excellent comment.
Either every generation has less children or you went wrong somewhere, Mr Cordova.
Sal, did you read my comment above yours?
Also the one above that too:
Your question itself reveals the error you are making 🙂
Seconded!
I suspect Sal is hoping, like his hero John Sanford, that humans were created recently — very recently — and that there hasn’t been enough time for mutational meltdown to occur, though it surely will if the world doesn’t end soon.
Yabbut – what about mice?
Or anything that reproduces way faster than people? They should be gone by now!
Yes! Agreed!
The fitness criterion (actually plural criteria expressed in the phenotype from all simultaneously actively functional genetic loci) is a measure that tells us whether the organism’s phenotype is “fit to its environment”.
Given how difficult it is to measure this criterion for a population directly, let alone an individual, we fall back on a proxy measurement – fitness measured by reproductive success. This is a necessary and sufficient outcome of phenotypic fitness (absent overwhelming exogenous extinction events), but it is not the meaning of “fitness”.
Maybe the Designer gets in there and fixes their little genomes before they’re too far gone. That’s the great thing about having God as the Designer — He can do anything!
(including gap-filling)
This, too, is wrong. Codon bias only matters in extreme cases, so no – Codon bias does NOT “suggest that every nucleotide…has importance”. The existence of codon bias is in no way inconsistent with the observation that vast majority of synonymous mutations are phenotypically silent (and of the ones that aren’t silent, it’s usually not because of codon bias…).
Although interesting GWAS reports sometimes suggest slightly increased odds ratios for “silent” SNPs. Could be statistical artefact of course, but could be something to do with the efficiency of expression maybe?
I call bullshit, on the grounds that you have no fucking clue what the actual ratio of beneficial-to-deleterious mutations may be. You can of course assume any value you like for that ratio, when you’re cobbling up your spiffylicious equations. But if your assumed value for that ratio does not represent the facts on the ground of Reality As She Is Spoke, it doesn’t fucking matter how spiffylicious your equations are, because said equations will not be a valid reflection of how things work here in the RealWorld. Said equations will, instead, be a valid description of how things work in some random/arbitrary hypothetical world other than the RealWorld.
Consider the gene for sickle-cell anemia. If you’ve inherited one copy of that gene, from one of your parents, you get the trait of heightened resistance to malaria; if you’ve inherited two copies of the gene, one from each of your parents, you get the trait of sickle-cell anemia. So is the sickle-cell gene beneficial or deleterious? Answer: Both. How does your equation handle genes which can be both beneficial and deleterious, Mr. Cordova? And there are also genes which, while beneficial for critters in one particular environment, are also deleterious for critters in a different environment. Classic example: a mutation that puts white fur on a critter will provide that critter with effective camouflage in arctic regions, but will make a critter stand out like a sore thumb in any region that isn’t snow-covered 24/7/365.
Go ahead, Mr. Cordova: Tell us all about whether the sickle-cell gene is Deleterious or Beneficial, and whichever fork of that dichotomy you run with, justify why it isn’t the other. Tell us all the sickle-cell gene’s One True Measure Of Benificence.
Tell us all about the One True Measure Of Beneficence for a mutation which puts white fur on a critter.
Tell us all your shiny whizbang definitions for ‘deleterious’ and ‘beneficial’, and why a mutation which fits your definition for ‘deleterious’ cannot possibly, in any set of circumstances whatsoever, fit your definition of ‘beneficial’.
Or, you know, keep right on assuming your conclusion up front, just like you’ve done here.
Sal writes at UD:
The cartoon does not apply to sexually reproducing organisms, in which genes can propagate independently from each other. It conceivably applies to asexually reproducing organisms if the population is very small.
I’ll let someone else take this one.
See 1.
Not sure what this means. Biological fitness is a perfectly well defined concept and can be measured and used.
Well, it’s not even a problem, as I think even Dembski now acknowledges. That’s why he moved on to “Search for a Search”. But the same objection applies – there’s no reason to think that searches have a uniform probability distribution.
Not a theoretical problem (see Lenski and Avida), and not easy to demonstrate that it is a practical problem (how do you prove a negative?)
As you say, Sal, this is not an objection to Darwin’s theory.
I expect he has 🙂
And as I pointed out, the opposite position (codon bias doesn’t matter), posses other problems because we then have a feature (a non-random a bias) that should have been wiped out by random mutation. Hartl and Clarke’s book on population genetics points out this enigma. Which ever way you interpret it, the U-paradox suggests codon bias shouldn’t be there.
The U-paradox also says if deeply “conserved” regions aren’t functional , they should have gotten scrambled too, but since they aren’t, that’s a problem and that’s a similar enigma to codon bias.
But, since we are talking science here, there is a testable prediction with the U-paradox especially now that we have Solexa technology to make sequencing very cheap. The testable prediction is that we should see an increase in SNPs in real time in the conserved regions. This will take a lot of sequencing to detect, but that is a testable prediction if it is practical. Because there is a lot of interest in human genetic diseases, and SNPs are now starting to get actively tracked in gene databases, we might see observationally if what I’ve laid out in this thread is accurate.
I suggested analysis of deeply conserved regions to Dr. Sanford at the tail end of an 8-hour conversation, and unfortunately I had to leave before we could pursue the idea further. I mentioned it to Walter ReMine also, and both of their eyes lit up because such an analysis avoids having to deal with unstable, unusable and frequently un measurable metrics like “selective advantage”. As Kondrashov hinted, weakly deleterious mutations aren’t immediately detectable. But SNP’s are!
Btw, you just made me realize something, the above derivation can be used as a critique of neutral theory. The deeply “conserved” regions shouldn’t exist because of the U-paradox. So the paradox holds whether one argues for a selectionist view of evolution or a neutralist view. That jogs my memory of an article Hartl wrote that was critical of the neutralists…
Thanks for you criticism.
Sal, since meltdown does not actually occur which is wrong; the model that predicts it, or reality?
I could be wrong, but I think the point SC is making is that RM & NS are not explanatory wrt what we actually see.
The problem is that Sal’s model predicts something that never happens.
The cartoon can be extended to sexually reproducing species because it deals with NUMBER of mutations, not specific mutations that can drift out because of sexual recombination. If Mom has 10,000 bad nucleotides and Dad has 10,000 bad nucleotides, when they have junior, junior on average will get 5,000 from Mom, 5,000. Then junior will get 6 brand new ones (if U=6).
Even if mom and dad have the same mutation in the same loci on the same chromosome, that means junior will become homozygous in that defect. The defect doesn’t go away just because of recombination, in the case of homozygosity, junior now has a double dose of the defect, and the U-paradox still holds.
Mutation ACCUMULATION is a separate problem form Muller’s ratchet which is mutation FIXATION. Accumulation implies that even if most specific mutations are drifted out of the population, there are plenty of new bad mutations to take their place, so to speak.
The issue of fixation (as in Muller’s ratchet) is separate, we’re talking about the lack of purification for all bad mutations (fixed and non-fixed).
I think that’s the whole point. Sal’s model is based on RM & NS being true (sufficiently explanatory). Since we don’t see those things happening, RM & NS cannot be sufficiently explanatory.
cubist,
It seems to me that Darwinists only challenge the meaning of “deleterious” and “beneficial”, and the comparative rates thereof, when those metrics are used to demonstrate serious problems with their theory.
Are you telling me human’s don’t use binary fission? Eeeeewwwww
Sal, you could support your thesis simply by naming a species that went extinct due to accumulation of mutations.
It would seem most likely to be a fast breeding organism like a virus or microbe.
Sal,
Your argument, as I understand it, is that
EITHER:
OR:
Thus
You appear to be ignoring the possibility of something in between these two extremes. It’s a false dichotomy. Hence my point:
Codon bias only matters in extreme cases
Far more exciting is the fact that you have made a testable prediction:
but will the human race survive long enough to see this effect?
😉
Then perhaps Sal should catch up on the last 60 years of research. Unless he thinks the seven or so mutations that have maintained the virulence of HIV in the last 50 years are the result of desigher intervention.
So, William, if a mathematical model does not completely comport with observed reality, does it make more sense to do some research to discover why, or more sense to assume that invisible daemons are jiggering the atoms.
Newton and Laplace redeaux.
Well, it may seem that way to you, William, but I put it to you that what is in fact that case is that we know perfectly well what we mean by those terms, and only “challenge the meaning” of them when they are used erroneously by people who have misunderstood the terms, and, as a result, think they have found “serious problems with their theory”!
So let me be as unambiguous as I can:
A “deleterious” variant is one that tends to reduce the probability that its bearer will reproduce successfully in the current environment relative (usually) to its peers who do not carry it, but occasionally relative to its parent variant. It is important to specify which, because if we are talking about a brand new variant (a “mutation”) we may want to compare it to the variant it mutated from, whereas if we are comparing polymorphisms that may share a common ancestral sequence, we would only be interested in comparing one polymorphism with another.
And a “beneficial” variant is one that tends to increase the probability yadda yadda.
A “neutral” variant is one that make no difference to the probability.
Clearly whether a variant sequence is advantageous, therefore, depends on its context – both its genomic context (what else is on the genome) and the environment, both of which are variable, and both of which are a function of the prevalence of the variant itself in the population. For example in a population of animals with 4 inch legs, an allele that tends to result in 4.5 inch legs might be advantageous, resulting in a population of animals with mostly 4.5 inch legs, whereupon it may no longer be advantageous, especially if by now there is an variant that results in 5 inch legs.
What strikes me over and over again about posts by IDists about evolution, and Sal’s is no exception, is that they really do reveal a profound lack of understanding of the theory they are attempting to critique.
Which is not to disrespect Sal – I am very glad he has posted here, and look forward to his response to these counter-critiques.
But the fact that he can cite that Gingerbread Man cartoon as some kind of argument is, as KF would say “sadly revealing”. Well, revealing, anyway! Of exactly the problem specifically YECs, including Sanford, have with understanding evolution.
Sanford, and the Gingerbread Man cartoon, both present a scenario in which there is some Edenic pre-lapsarian perfect [Gingerbread] Man, for which there is no direction to go but down. And while any change can happen in evolution, the chances of that exact change reversing, is very small.
And if 99.9999999 percent of changes to a perfect Gingerbread Man result in a less perfect Gingerbread man (the remainder either reversing a previous downward move, or making an equally good Gingerbread man, perhaps a Gingerbread Woman) then of course, all that will happen over time is that the Gingerbread Man lineage will degenerate (especially if there is no sex).
However, that is not the theory of evolution. The theory of evolution posits that the first proto-life forms were very poorly adapted, but just good enough that the population persisted over time – and so with lots of ways in which to be better reproducers (not just one standard of perfect Gingerbread Manness). And different lineages found different environmental niches to adapt to, and these themselves changed over time, not least as a result of other lineages sharing the same resources, and members of their own lineage competing for resources.
In that context a great many variants will be advantageous – sometimes. Far more will be neutral, and also many disadvantageous. The advantageous ones will, by definition, become more prevalent, and the neutral ones will stick around, becoming advantageous or disadvantageous in their turn.
This is easily demonstrable by a simple evolutionary algorithm – features that are initially advantageous, and become highly prevalent, are later outcompeted by something even better, at which point they have become disadvantageous.
The population models that Sanford and Sal cite hugely simplified, with limited range (for example, s is effectively treated as a constant in many models, which it clearly is not). Which is one reason I prefer to understand population dynamics through computer models than mathematical models – evolution is profoundly non-linear and chaotic, and I just don’t think neat equations are up to the job.
I could be wrong.
Sal: define a “bad nucleotide”.
Seriously, that’s where you are going wrong (or one of the ways).
What is your criterion for “bad”?