Once again, Gil neatly falsifies the claim that “both sides are interpreting the same evidence differently” by offering an analogy that could only apply if one ignores piles of real-time evidence on mutations.
5 Likes
The only examples of strictly deleterious mutations that are still non-selectable have such infinitesimally small effects that billions are required to have even a noticeable impact on fitness, and would likely require trillions in order to drive extinction. Since this would take more generations than suggested by any evolutionary model I’m familiar with, this sort of mutation can be safely ignored.
Of course, most mutations are not strictly deleterious. Generally, mutations do not affect fitness directly but indirectly through changes to some trait. If the trait is relevant to fitness and the change is away from the fitness peak for that trait, then the mutation is deleterious. Since most populations are well-adapted to their environments, most mutations are deleterious because most traits will already be close to their respective fitness peaks. But peaks are ‘peaks’ because moving in either direction is going downhill, so a given deleterious mutation affecting a particular trait is as likely to increase the trait as decrease it. So the accumulation of such mutations would have no cumulative effect on fitness, because their effects on the trait itself would cancel out.
1 Like
My emphasis added.
@Giltil This repeats a point I was making in an earlier iteration of this topic. In the absence of selection (bias) mutations add variability to fitness (literally increases the statistical variance of the mean fitness). If there was not enough variability in the population for selection to “see”, then there soon will be. As mutations accumulate some part of the population will have a large enough fitness different from some other part of the population that selection will occur.
Proof of this is in the derivation of the Central Limit Theorem (a.k.a. the Law of Large Numbers) which is the heart of statistical theory.
4 Likes
Now there you go. That’s a fine objection to GE.
It might be something as simple as an increase in replication time. One might suppose that there would be some genomic size that would have significant replication cost, for example, even though no organisms except bacteria seem to care much.
There will be an equilibrium point at which deleterious and favorable mutations are equally likely, but the question is whether this point is reached before or after the population becomes extinct and how long it takes to reach that equilibrium. Empirical data tell us that GE isn’t a thing, but not necessarily why.
3 Likes
Right, as I said…
I’m honestly not sure if that is true for something like increased genome size, but to the extent it isn’t it would take billions of generations to matter (if it ever matters).
I think your mention of bacteria is illustrative here. Even the smallest possible fitness effects are evidently selectable in bacteria and viruses, whereas we know empirically that billions of the same sorts of mutations can accumulate in eukaryotes without any impact on fitness. In other words, in the only organisms where they aren’t selectable, they accumulate too slowly to matter. So clearly if GE is to work, it needs to use mutations with larger fitness effects than genome size differences and the like. But so far as I can tell, all such differences fit into what I described in my second paragraph of that comment, and are thus equally irrelevant.
So what GE proponents must do is describe a class of mutation with fitness effects large enough to matter in a relevant timescale while still smaller than the selection threshold and still strictly deleterious. Of course, this will only make GE ‘possible’. They’ll still have the substantial task of squaring it with the fact that, as you say, empirical data tells us it isn’t happening.
3 Likes
Even if that were true, I mean even if it was the case that at some point a part of the population had a higher fitness than another part so that selection could favor the former population, the fitness of the former population would still be inferior to the one of the starting population. IOW, GE would still occur.
Interesting. Examples of these known strictly deleterious mutations?
I am trying to understand what you meant by “billions are required”. Do you mean a billion people need to have these strictly deleterious mutations or VSDMs before we see its effects?
I think everything you’re saying is the point of GE. I honestly don’t get why you’re saying it’s not. Maybe the words we’d be contending over would be “to matter.”
What empirical data in humans tell us it isn’t happening? How else are we degrading? Sanford is arguing that the selection threshold is much different than what other population geneticists are saying. Of course selection has happened in the past and is slightly relaxed now, but the point of the book is that it’s not a good enough explanation for the problem we face.
For one, comparative genomics is telling us that the accumulation of nearly neutral mutations really doesn’t matter. The >90% of the human genome that is not under selection has very little similarity to the mouse genome. Mice and humans are doing just fine despite all of these mutations. This is in comparison to shared regions that are under selection which have up to 85% similarity between humans and mice.
3 Likes
Larger genome because of lots of junk, for instance.
Getting that much extra junk would require lots of mutations.
No, I’m saying you’d need billions of generations to see the effects.
Great, then you agree GE is impossible.
I’m not saying it isn’t the point of GE, I’m saying it makes GE impossible.
If GE can only happen in higher organisms with modest population sizes and even then would require billions of generations, and these organisms have generation times of >6 months, then seeing any effect at all would take half a billion years, and extinction would take 10-50x that. In other words, it can’t happen fast enough to have had an appreciable effect in evolutionary timescales, so it doesn’t matter.
Limiting it to humans is special pleading. If GE is a real thing, it needs apply evenly to all organisms. That doesn’t necessarily mean that it needs to matter in all organisms (bacteria might not be affected because of their massive population sizes, for instance), but the math must still be applicable. We have empirical data from many species and none show what Sanford says should be happening in all (or most), so GE must explain this. And the empirical data from humans isn’t consistent with GE anyway.
Relaxed selection.
The math that other population geneticists use works, so Sanford would need to provide a replacement mathematical framework able to explain all of the data currently explained better than the current math. He has not done this. In fact, his math explains less of the data. Honestly, it explains none of the data.
Then the point of the book is wrong, because it is a sufficient explanation.
7 Likes
You’re missing the point. His point is humans share a common ancestor with humans. We don’t share a common ancestor with mice. Random mutation + selection can’t create.
I never said it doesn’t apply to all organisms - neither does Sanford - however, what he does mention interfering with selection in humans will not apply evenly to all organisms. And we’ve already agreed bacteria might not be affected. Yes, the math should be applicable.
And I really can’t argue with your generalizations. It seems you don’t want to be specific with examples or data. That is OK; we can be done with this conversation.
Invisibly? I take the point to be that if the mutational effects are small enough, even though there is going to be variance in the number of mutations, or their relative effects, they are still so small as to not amount to a difference selection can act on. I think that’s the most charitable thing I can say about the idea of GE.
I do have to wonder, though, if this also results in a rate of fitness decline to be so low that it might still take many millions of years to drive any population to extinction.
Those are just assertions. We have mountains of evidence that humans and mice do share a common ancestor, so we are going to use that evidence to test Sanford’s claims.
Even if you don’t accept this evidence, you still need to explain why there are regions in both genomes that have 85% similarity between humans and mice and regions that have almost no similarity. If changes to these regions are as bad as Sanford claims then it doesn’t matter if mice and humans share a common ancestor. All of those changes should still cause massive problems according to Sanford, and yet they aren’t causing problems.
Empty assertions are meaningless.
4 Likes
Great. It isn’t so GE is false.
Models are generalizations, so the only way to defend a model is to argue with generalizations. That’s how that works. Saying you can’t argue with generalizations is saying you can’t defend GE.
If the model is structurally invalid, then specifics aren’t needed. Then again, every species is an example where we don’t see GE, because we don’t see GE in any species. Which means GE doesn’t apply, which means it is wrong.
5 Likes
Genotype AA, Aa, and aa each have an absolute fitness which is the product of viability to reproductive age and number of offspring produced.
How do you drop the absolute fitness of AA without changing the relative fitness of AA/Aa?
Drift is very slow in large populations.
I don’t think so. This comes back to his pointing out that the mutational and evolutionary mechanisms we observe in real time account for these differences.
You said 11 days ago that “the evidence isn’t actually consistent with that.”
Two days ago, you wrote, “I definitely could have provided data in the examples you mentioned.”
This is a variation on one of those examples I mentioned. I don’t see any reason why you could have provided data 11 days ago, but are unable to today.
Evidence?
You’re failing.
Today is the future relative to two days ago. How about providing any data, much less more? You’ve made multiple assertions here with zero evidence.
Yet the evidence shows that those two mechanisms create all sorts of incredibly specific antibodies in your body in less than two weeks. If that isn’t true, how does the SARS-CoV-2 vaccine work in the YEC universe?
I don’t know if “you” was singular or plural, but Taq opened with specific examples and data:
That is evidence, Valerie. You don’t offer any.
3 Likes
You are mistaken.
Do you honestly know the difference between haploid and diploid?
So what? He’s not producing evidence to support that argument. Even when he looks like he is to a layperson, he’s cherry-picking evidence when he’s not objectively misrepresenting the evidence, as in misrepresenting influenza reassortments as mutations.
1 Like
Cosign.
Cosign.
Framing it this way assumes, among other problems, an optimal starting state. In other words, assuming special creation. In reality, the population would land at an equilibrium where inviable genomes are removed while viable genomes persist. The thing is…this is where life has always been, right on that knife’s edge. Sanford starts from an assumption of some perfect starting state following by constant, inevitable decline. He begs the question.
8 Likes
More than true, this is math. It’s true that fitness could decrease by some degree, but the more decrease the more selective pressure to increase.
The key misunderstanding here seems to be thinking the population are genetic clones, with no variation in fitness. I agree such a population, presuming it is already near peak fitness, will show some average decrease in fitness due to mutations. At this point the population are no longer clones and the decrease will begin to be countered by selection.
2 Likes
This is plain wrong, as can be clearly seen from Sanford’s paper below.
https://www.worldscientific.com/doi/pdf/10.1142/9789814508728_0010
Below are some quotes taken from the discussion that falsify you claim (knowing that STd is the selection threshold, I.e., the value of mutational fitness effect for which the accumulation fraction is 0.5, indicating that half as many mutations have accumulated as would be expected under complete neutrality)
« For a typical mammalian model population (e.g. 10,000 individuals, genome size of 3 billion), our estimate for the lower limit of STd is in the range of 10–4 to 10–5. Thus even with minimal levels of biological noise interfering with the phenotypic expression of the genotype, those deleterious mutations which reduce fitness by less than 10–4 to 10–5 will largely escape purifying selection and will accumulate linearly. We show that three important sources of noise which substantially increase the value of the selection threshold in large populations are: (1) selection interference between mutations; (2) environmental variance; and (3) any significant degree of probability selection (in contrast to truncation selection, which never occurs in nature). Our experiments show that depending on these variables, STd values for mammalian species may be as high as 10–3 or higher. Given Mendel’s default fitness effect distribution, STd values in the range of 10–3 to 10–5 results in 82–97% of all deleterious mutations becoming effectively unselectable ».
« Our simulations indicate that the on-going accumulation of low-impact mutations results in continuous fitness loss. Consistent with the findings of others, our analyses reveal that the greatest contributor to this fitness loss is not the entirely unselectable mutations (having negligible fitness effects even in large numbers), but rather the accumulation of mutations with effects near the selection threshold. We observe that mutations in this zone accumulate more slowly than if there was no selection, yet still accumulate continuously and in large numbers. This transition zone between mutations that are entirely selectable and entirely unselectable is often at least two orders of magnitude wide and typically encompasses fitness effects on the order of 0.001 to 0.00001. Accumulating alleles within this transition zone are primarily responsible for the reduction in fitness »
« These extensive investigations have indicated that mutation rate, environmental variance, selection mode, and time are important variables that affect STd — in addition to population size. In populations of 1000 or more these factors are often more important than population size. For this reason we focused this paper on those specific variables, exploring the full range of their potential effects. In so doing we consistently find that the majority of deleterious mutations are not selectable, except within small and extremely unrealistic slivers of parameter space (e.g., the combination of less than 1 mutation per individual, no environmental variance, and full truncation selection). In this light, our conclusion that most deleterious mutations are beyond the reach of natural selection appears to be robust »
« It has been speculated by Lynch [24] and others that greater fecundity and more difficult living conditions in the past resulted in enhanced natural selection which may have been powerful enough to stop deleterious mutation accumula- tion. In order to test that hypothesis, simulations were conducted with 12 off- spring per female, no random death, and a mutation rate of 3. These settings result in ten of every twelve offspring being selectively removed. This very extreme form of selection slowed mutation accumulation and the rate of fitness decline, but did not stop it. After 10,000 generations, fitness declined to 0.22 with prob- ability selection, and 0.57 with partial truncation. In both cases, mutations of non-trivial effect were still accumulating and fitness was still declining when the runs ended »