6 posts were merged into an existing topic: Comments on Sanford and Carter respond to PS participants
Dr Sanfordâs book is going to do the best job of explaining it. It boils down to what we can know about what the genome is (coded information), and what happens when you introduce undesigned changes into functional coded information.
More empirically, we know generally that most mutations are damaging because we can see what happens in mutagenesis experiments and we can look at mutational load over time. This was addressed in the joint article (OP), as well as at https://creation.com/fitness. Many would like to bait-and-switch, pretending that just because we cannot directly measure individual VSDMs, that somehow we know nothing about the distribution of most mutations in general (we do).
It is clearly you doing the bait-and-switch, saying that because we know something about mutations of large effect, we can extrapolate that to mutations of small effect.
2 Likes
And what we see both in models that implement the empirical phenomenon of epistasis, and in experimental evolution confirms that the DFE of mutations doesnât remain fixed in shape, and as organisms become more and more adapted, the DFE skews more to look biased towards deleterious in proportion to how adapted the organisms become. And conversely as we move further down in absolute fitness (using the proper population genetic definition of the word), the effects of mutations increase proportionally skewing the distribution back towards more equal.
1 Like
We already know what happens. Usually there is no change in functionality, sometimes the functionality lessens, sometime it improves. The cases where functionality lessens tend to get weeded out by natural selection. The cases where it improves gets rewarded by higher reproductive success and the changes get incorporated in the gene pool. The invisible to selection mutations donât secretly add up until one day all the members of a species suddenly keel over dead. 
I wonder if Creationists pushing GE even understands what the term invisible to selection means
2 Likes
The distribution doesnât matter if the cumulative effect of VSDMâs is tiny.
If we focused on a single human intron and compared it to many other species of varied relatedness, what would we see? We would see a spectrum of shared sequence, with the most distantly related species having almost no shared sequence outside of splice sites. The divergence from humans sequence follows no pattern other than that expected from random mutations. So at what point does this intron stop functioning? At what point is the information gone? At what point should it pass the threshold Sanford has drawn for species extinction?
4 Likes
Ok, to reiterate: genetic entropy absolutely implies the existence of âperfect genomesâ.
This is an aspect I do not think you (or indeed Sanford) fully appreciate. Your argument is that VSDMs must accumulate, because such mutations fall beneath the threshold of selection. You extent this to imply that a species accrues VSDMs continuously with the passing of generations until at some woefully unspecified threshold it becomes universally non-viable and goes extinct.
So let us turn the arrow of time backward. If the passing of generations must, necessarily make genomes worse due to accumulation of VSDMs, then ancestral genomes concomitantly must have fewer VSDMs than extant genomes. As we trace back through ancestry, genomes must (under GE proposals) get âbetterâ. And under GE, this is a finite process: there is a point at which a genome contains zero VSDMs. Under GE, genomes can only get worse, so clearly this genome cannot have any ancestors.
Now as you state, for GE to apply, life need not start at this precise âperfectâ state (though special creation of imperfect genomes seemsâŚodd), but this does not change the fact that GE necessarily implies that such a perfect state can exist.
Perfect genomes are not a requirement of evolutionary theory, so the problem does not arise there. Under evolutionary mechanisms, âcrap but robustâ is the sweet spot life iterates to (and life appears to dwell within such a sweet spot), and perfection is not only not predicted, but also not even conceptually rational.
Under GE, perfection is conceptually necessary: how can you say any genome is âworseâ when you do not have a reference standard?
I will note that this is not a conceptual issue restricted to creationists: many papers addressing mutational accumulation make this error, modelling decline of some reference genome assuming that that genome is âcorrectâ, rather than more prosaically acknowledging that the reference genome is itself âcrap but robustâ like everything else, and is already saturated with VSDMs accrued over billions of years.
So: genetic entropy states human genomes are degrading, necessarily implying that ancient genomes were better, and if you go far enough back: âbestâ.
I would like to know what traits @PDPrice thinks this âbestâ genome carried. I picked eye colour, skin tone and height because these are fairly obvious traits that everyone has, and which (today) carry advantages and disadvantages depending on environment. Under GE, all extant eye colours, skin tones and heights must be âdegradedâ variations on the ancestral âbestâ genome, so it seems fairly straightforward and reasonable to ask what the eye colour, skin tone and height were of the individual or individuals carrying this first, âbestâ genome.
For bonus points, I would love Paul or Sanford to explain the methods they used to obtain their answer.
1 Like
Welcome to Peaceful Science @Sweary_Biochemist! I didnât see you come in, but Iâm the guy who found you on Reddit. 
2 Likes
In a comments sub-thread, I noted that by definition, GE is adding variation in fitness to the population. Even if the initial population starts out perfect or uniform, it wonât stay that way for long. Sooner or later the difference between the most and least fit in the population will become large enough for selective pressures to act. There will minimally be purifying selection, removing the least fit, and increasing the average fitness back towards some equilibrium state (assuming a stable environment).
Genetic Entropy is a self-solving problem.
2 Likes
Iâm comparing this to your earlier reply here:
âŚand it just looks like the same post using different words. I do get the idea behind the hypothesis is that the VSDs outweigh the beneficial ones.
Iâm aware we can measure the effects of large (selectable) deleterious mutations and we can measure the effects of large beneficial mutations, but if we cannot measure the effects of VSDs, how can we then say VSDs that outweigh the beneficial mutations? I am still not clear on this at all. This is a really specific claim and Iâm not sure how we can get there without being able to measure it.
If all thatâs being claimed is that most mutations are bad, I can see why Dan Cardinale sees things the way he does, but if the claim is actually that itâs specifically the mutational load of VSDs that will cause extinction, Iâm not still understanding what measurements have led to that conclusion.
1 Like
Sorry, just not true. GE does add variation, but you must recall that individual variations are not selectable. Now if you had individuals with a big enough difference in the number of these variations, that would be selectable. But the way GE works (in the GE scenario, that is) is that the variations accumulate fairly uniformly in the population, with nobody differing sufficiently from anybody else, in the number of those variations in their genomes, to cause significant differences in fitness, and so they remain invisible to selection until the species goes extinct.
There are no such measurements.
The later is exactly what I mean.
If the population remains nearly uniform, then I would agree, but accumulating mutations imply increasing variation, therefore a non-uniform distribution.
Maybe if mutations are introduced slowly enough and become fixed before so many accumulate that selection kicks in, then a situation allowing GE might occur. This requires a ânarrowâ distribution of fitness in the first place which seems unreasonable to me (is that what âperfectâ means?). I will take a harder look at the math as see it there is a sweet spot for GE to persist.
We would not expect that to happen, for the reasons I mentioned.
It isnât necessary for them to be introduced slowly, given drift operating on lots of independently assorting sites. Why would you expect a large variance in the number of such alleles among individuals? Why would anything require a narrow distribution of fitness? Remember that each of these alleles individually contributes almost nothing to fitness.
No measurement points to that conclusion unless you assume:
- That the DFE of effectively neutral mutations (those with very small fitness effects) look similar to the measured DFE of stronger mutations in highly fit organisms.
and - That this DFE remains essentially constant no matter how low or high absolute fitness gets.
Not if a lot of the deleterious mutations fix in the population.
1 Like
One thing thatâs wrong with my model is that it assumes all VSDs have the same effect on fitness. In reality (that is, in the assumed reality of GE), they would have a range of effects, and this would contribute to the variance in fitness between individuals, and therefore to the ability of natural selection to prune these deleterious alleles.
What would the effect of this be? Iâll assume that the (negative) selection coefficient (s) could be as large as 1/30,000 and still be invisible to purifying selection (which is the right ballpark for an effective population size of ~15,000 for humans). Thatâs 100 times the mean s Iâve been assuming â which I want to leave unchanged so the current loss of fitness is too great from the alleles weâre already carrying. So I can allow fluctuations in s as large as 100x the mean (call the mean sel. coeff. âSâ). The maximum variance this can introduce between VSDs corresponds to a standard deviation of 10S.
Now we take ~3 million of the VSDs, with this variance in fitness effects, per person and determine their overall effect on fitness. Thatâs a large enough number that the central limit theorem applies in spades, so the distribution of overall fitness caused by this variation will be normally distributed with Ď = 10S/sqrt(3000000) = 0.006 x S. Thatâs roughly ten times the contribution from variation in the number of VSDs per individual.
Does these mean it has ten times the effect? Not on the number of VSDs eliminated by selection â it has little or no effect, because the variation in fitness isnât correlated here with the number VSDs. So varying selection coefficients will have cause some selection against the more deleterious VSDs, even though theyâre nominally invisible to selection, but not on the number that is accumulating. Exactly how much discrimination there will be will depend on the distribution of fitness effects â the more skewed, the more selection will be able to operate, because the situation more closely resembles a small number of genuinely deleterious alleles and a larger number of really neutral ones. I canât say I have any interest in pursuing this any further, since teasing out these tiny effects requires more work than itâs worth.
In summary, selection acting on bulk VSDs should have nonzero but small effects on the number and severity of the VSDs that accumulate, but I see no reason to think they would prevent the accumulation. Note that this is all under the assumption of a fixed effect size distribution, independent of absolute fitness.
Thatâs a point worth considering: as GE progresses and the population decreases in size, the selection coefficient of alleles that canât be eliminated by selection will become more negative, thus accelerating the GE.
If I understand correctly, youâre saying the so-called âzone of no selectionâ will widen to include mutations of stronger effects?
Years ago (it was 2013) when I was much newer at studying this, I asked Dr Sanford a similar question to what you have asked here. Perhaps his answer to me will help you as well.
âRegarding the actual shape of the distribution, there is strong consensus among population geneticists that the shape is basically exponential (the frequency of mutational effects increases exponentially as one approaches zero effect). ⌠Regarding experimental evidence for the distribution, it is hard to measure any genetic effect which is subtle - i.e. anything less than a few percentage point change in fitness. So the vast majority of mutations are clearly not measurable (not even âmother natureâ can âseeâ their effects). Since the mutation rate is high and we all carry thousands of inherited deleterious mutations (which no one doubts), and yet we are still viable - this strongly argues that the average mutationâs effect is tiny. With our default settings [in Mendelâs Accountant], the mean deleterious mutation effect is near .001 - which agrees well with published estimates.â
Regarding your question of how we can say they outweigh the beneficials, it is simply put because beneficial mutations are so rare as to be nearly non-existent, just like typos that add to the meaning of typed text in a useful way are nearly non-existent. One published estimate of the ratio is 1,000,000 to 1 (Gerrish and Lenski).
The only remotely conceivable way that deleterious mutations could not outweigh beneficials, would be if natural selection were close to 100% effective in removing the deleterious ones. Otherwise, the trend overall is going to be downward. Beneficials are simply too rare otherwise.
Another way you can infer the DFE of mutations is to look at ageing. One of the chief causes of ageing is the buildup of mutations over our lifespan in our somatic cells. Lynch says that by the time one is sixty years old, oneâs somatic cells have accumulated on average about 40,000 mutations (each cell has its own set).
So you can answer your question, in one sense, by just comparing old people to young people. If deleterious mutations did not outweigh beneficials, then we would either not age, or we would get healthier over time. Neither of these, unfortunately for us, describe reality.
1 Like
I think you mean 1,000,000 to 1, if youâre citing Gerrish and Lenski (1998). Iâm sure youâre aware that this is one of the highest ratios out there, and other published estimates go as low as 10,000 to 1 (see de Visser and Rozen, 2005) or even 100 to 1 (Perfeito et al. 2007).
6 Likes