Genetic entropy, a rant
This is written in response to @TamiHoshiyama who (as a creationist with an apparently deeply-rooted anti-evolutionary stance) claims genetic entropy is
A) a thing that genuinely exists
B) responsible for a steady degradation of the human genome, with presumably concomitant effects on human health
As far as I can tell, this stance is selected more or less to support the usual run of “EVOLUTION CAN’T HAPPEN” claims. Aside from the fact that evolution can happen (as we’ve observed multiple instances of random mutations followed by natural selection leading to novel behaviour, substrate utilisation, environmental tolerances and outright generation of novel species), this does pose rather a puzzle from a purely theistic perspective:
1) Humans are too complex and amazing to have evolved, they must be designed
2) Humans are mutating themselves to death because they’re poorly designed
If the argument is that both 1) and 2) are correct, it doesn’t really paint this hypothetical designer in a very good light. Rather than being omnipotent and making perfect creations (a common argument), god instead makes a whole bunch of species that barely hold together for a few thousand years before imploding under cumulative mutational load due to not apparently bothering to design sufficient repair enzymes.
A further note is that a huge number of mutations are due to nucleotide choice: thymidine, for instance, is highly prone to UV-induced crosslinking. Anywhere in the genome where two thymidines are adjacent is vulnerable to thymidine dimer formation. UV hits on thymidine, it crosslinks to the other. These then have to be chopped out and replaced.
With four bases to choose from, getting TT is a 1/16 chance, so in a genome of human size (3x10^9) that’s just over 187 million places to go wrong every second of every day. Repair systems are busy.
Note, incidentally, that uracil provides exactly the same base-pairing properties as thymidine but doesn’t crosslink (A-T pairing and A-U pairing both work): why doesn’t life use uracil in DNA? The answer is because of the OTHER major source of DNA mutations: cytosine deamination.
Cytosine contains an amine group (NH2) which has a tendency to fall off, because hey: thermodynamics. What do you get when cytosine spontaneously deaminates? You get uracil.
If DNA used uracil rather than thymidine, it would be extremely difficult for the repair mechanisms to distinguish between a legitimate uracil and one resulting from cytosine deamination. In a U-G mismatch is the U wrong, or the G?
As it is, the repair mechanism (uracil DNA glycosidase, or UDG) simply scans for any uracils and chops them out to be replaced with cytosine, because that’s a far easier mechanism to develop. It does this millions of times a day, PER CELL.
Repair systems are BUSY.
So on one hand we have a divine creator who deliberately selected a range of nucleotides specifically to be really prone to mutating and necessitating a massive array of repair systems just to stay reasonably intact but apparently not enough to prevent massive eventual mutational collapse of his/her magnificent creation….
….or we have naturally formed early life that, over a billion of so years, evolved a system using the least detrimental combination of nucleotides that can form spontaneously, because that’s all that was available. And since this system cannot prevent mutation, genomes will change. Successive rounds of mutation and selection will produce huge varieties of life, always selecting for “good enough”, because that’s all you really need.
Anyway, on to genetic entropy.
So the idea here is that humans (or possibly all organisms) mutate too fast for natural selection to act, and there are various scientific papers claiming this, along various different lines of reasoning. Creationists love this, because they assume it implies that humans must be either
A) extinct, or
B) a relatively young species (ideally, created)
and that since A isn’t true, B must be.
Leaving aside the hilarious consequence that this means they’re fervently using “incompetence of god” as a justification of “existence of god”, we need to ask if the assumption is actually correct.
Creationists go apeshit crazy happy at the idea their god’s creation is decaying unstoppably because it’s poorly made:
Actual paper they’re using as their basis for this
And Kondrashov’s “OMG WHY R HUMAN NOT DED” paper (may be behind a paywall)
So first we have Kondrashov’s 1995 paper: Contamination of the Genome by Very Slightly Deleterious Mutations: Why Have We Not Died 100 Times Over?
I would warn anyone thinking of reading this paper: it is not very user-friendly. Frequent use of lots and lots of terminology that is defined only very vaguely, and then applied incredibly broadly, dotted with buckets of equations (because MATHS WOO).
General premise is that mutations that cause big problems are strongly selected against (all very straightforward, all very Darwinian) but that “very slightly deleterious mutations” or VSDMs are not deleterious enough to be negatively selected. As mutations are constantly occurring, in a stochastic (i.e. random) fashion, VDSMs will tend to accumulate, leading to (hypothetical) progressive loss of fitness.
So, already we have a very clear demonstration of why you don’t see propagation of really bad mutations: because really bad mutations kill the carrier before they can breed. And this is a spectrum phenomenon, as illustrated by the wonderful equations listed in the paper. Thus (paraphrasing):
• Mutations that lower breeding fitness totally (via death) will be selected against totally.
• Mutations that lower breeding fitness “quite a lot” will be selected against “quite a lot”
• Mutations that lower breeding fitness “only a bit” will be selected against only slightly
• Mutations that lower breeding fitness a tiny bit will not be selected against at all.
This spectrum is also related to population size, so larger effective populations will tend to exert greater stringency against accumulation of mildly deleterious mutations, because there are simply more individuals to dilute out the mutations for effective fitness testing (whereas in small populations traits can get fixed more easily just through chance). And the corollary of all this is that steady accumulation of very slightly deleterious mutations gradually leads to too many of them, at which point the combination of tiny problems behave as one “really quite deleterious mutation”, and the species dies.
He goes on to work out the maths and conclude that almost all vertebrate species have too small an effective population to stop the cumulative onslaught of VSDMs, and we should all be dead. Like, SO dead. OMG SO DEAD.
Now this is all a bit handwavy when applied to actual organisms, as in simulations and equations you can attach actual numerical values to things like “selection coefficient”, whereas in life you simply can’t.
How deleterious is a point mutation that lowers efficiency of liver lactate dehydrogenase by 3%? No fucking idea. Under some situations it’d even be beneficial.
You’d be effectively making up a number.
They are simulating hypothetical variables that you can’t actually measure, and which frankly might not even be appropriate: a mutation may be useful in one context and detrimental in another. Hell, it might even be useful in one ORGAN but detrimental in another. Biology is complicated.
Also note they also don’t include VSAMs (Very Slightly Advantageous Mutations) or MTDFAs (Mutations That Do Fuck-All). They claim they include VSAMs, but only in the context of REVERSING a previous VSDM.
It is assumed, for the sake of the model, that an organism starts with “the BEST genome”, one that is utterly and totally appropriate to its environment, and in an environment that does not change.
Thus all changes are various shades of detrimental, or reversions to the original best genome. “A beneficial mutation is possible only in a site occupied by a suboptimal nucleotide.”
THIS IS A FLAW. We will come back to this.
He concludes by suggesting various ways in which this apparent paradox (OMG WE SO DEAD, but we’re not dead) can be resolved.
1) The ticking timebomb: we are going to go extinct, we just haven’t yet. He argues that this could explain why many vertebrate lineages go extinct, but then seems to forget that as all vertebrate lineages have a common ancestor, this should apply universally. In fact, should apply to the first vertebrate, and there shouldn’t BE any vertebrate lineages.
2) He’s underestimated effective population sizes (effective and total populations can be very different numbers, the latter is easily measured, the former less so).
3) That mutations that are “deleterious enough to be bad when cumulative” but “not bad enough to be selected against directly” are rare. I would say this manages to further miss the point that has already been spectacularly missed: the relative effect of a given mutation on an individual (or indeed a population) is a HUGELY context-dependent phenomenon.
A better statement would be “mutations deleterious enough to be bad when cumulative but not bad enough to be selected against directly ARE ONLY REALLY A THING THAT CAN BE ASSERTED TO EXIST IF YOU USE OUR RIDICULOUSLY UNREALISTIC MODEL”
4) Epistasis: it’s not a matter of just adding up small bad things until you get one big bad thing, because genes interact with each other in novel and unpredictable ways which we can’t easily model (see model limitations, above).
5) Soft selection applies: this is basically “rather than mutations accumulating solidly and invisible until suddenly EVERYTHING DIES” (hard selection), it’s “mutations accumulate and gradually lower the fitness of individuals so that excessive accumulations never actually occur as those with excessive accumulations breed less effectively”. This seems to me to be fairly self-evident, and I kinda wonder why the hell they were treating this as a hard selection scenario in the first place.
So really, 1) is the silliest (but the one the creationists have pounced on), 2) is just another variable, 3) exposes the silliness of this model (which the creationists will ignore) while 4) and 5) are critically important to understanding this process.
In conclusion, if an organism (for whatever magical reason) pops into existence with a genomic makeup absolutely and perfectly suited to its environment, and that environment never, ever changes (while mutations still nevertheless occur), that organism will gradually lose fitness, and the species will eventually die out if we assume that only hard selection criteria apply (which of course, they don’t).
The answer to “why haven’t we died out?” is in essence, because the model they’re using is entirely unrealistic.
The second paper (linked above) is more recent, by Michael Lynch, and documents a fairly extensive analysis of human mutation rates and the expected consequence on fitness. The mutation rates they measured are all spontaneous mutations, i.e. how many new random things happen per genome, per generation.
By comparing these rates with the study above, they conclude that some of these mutations MUST be deleterious, and that some of these deleterious mutations MUST be “not quite deleterious enough to be selected against”, and thus the human genome MUST steadily acquire deleterious mutations over time, leading to a progressive loss of fitness.
The data is pretty good, and with a rate of 12.8x10^-9 per site per generation, you get an average of about 38 point mutations per genome per generation (and assorted other insertions, deletions, transpositions etc). This is relatively high when compared to other organisms, though I’m interested to note they don’t address genome size: Arabidopsis thaliana has a mutation rate of only 4.56x10^-9 per site per generation, but also a genome size of only 70x10^6, so would get about 0.5 mutations per generation.
Both humans and A.thaliana have about 25000 genes, but our genome of 3x10^9 is about 40x bigger than A.thaliana’s 70,000,000. We don’t have “genes 40x bigger”, we simply have more padding: bucketloads of repeats, huge introns and so on, all of which are relatively tolerant of mutation (because they don’t do anything). In terms of CODING mutations, therefore (i.e. mutations in actual protein coding regions) we actually have a mutation rate near enough identical to A.thaliana.
Humans get ca. 38 point mutations, of which maybe 0.5 are in a coding region.
Long story short, assessing mutation rates is tricky.
Anyway, the paper then goes on to apply this to humans with respect to fitness and future consequences. It stresses that these numbers are NOT trends, because they based on data from humans alive today, so we don’t actually have a clue whether the measured mutation rates are globally applicable to all humans ever, to modern humans only, to western-world humans only, or even to the humans used to build the dataset only.
(mind you, it’s LIKELY that basal mutation rates haven’t changed significantly, for our species, in thousands of years….though exposure to modern lifestyle almost certainly will increase the net level of mutation, what with all the environmental mutagens heavy industry generates, and stuff)
It then explains, in delightfully precise and impossible to miss terms, how modern western lifestyles have relaxed selection criteria. In other words, “fitness” as defined here, is declining, IN THE WESTERN WORLD ONLY, AND IN HUMANS ONLY, as a consequence of us having much more pampered lifestyles.
Have a beneficial mutation that allows you to survive with less food?
Irrelevant, food is plentiful.
Have a beneficial mutation that confers disease resistance?
Irrelevant, we have antibiotics and antivirals.
And of course also:
Have a detrimental mutation that makes you MORE dependent on food?
Irrelevant, food is plentiful.
Have a detrimental mutation that lowers disease resistance?
Irrelevant, we have antibiotics and antivirals.
In essence, many things that would constitute selection criteria in a less pampered lifestyle no longer act as such in the western world.
“For example, fetal mortality has declined by approximately 99% in England since the 1500s (52), and just since 1975, the mortality rate per diagnosed cancer has declined by approximately 20% in the United States population”
We may indeed be slowly losing “fitness” (again as nebulously defined here), but this is not because “genomes inevitably decay”, it’s because we’ve massively lowered the stringency of specific aspects of the selection process.
More people are alive that otherwise wouldn’t be (because MEDICINE).
Mutations that lower fitness in the developing world are still filtered there (because they lower fitness), whereas they are not, here. This “reduced fitness” is not observed in less industrialised nations.
Nothing to do with thermodynamics, entropy, original sin or any of that silliness. It’s just what you get when you take “mutation + natural selection” and remove most of the selection. It’s a consequence of recent technological and medical advances.
And there are caveats even here!
You’ll note that I’ve put quotes around “fitness” here, because definitions are important, and here they’re being used ….badly.
What is fitness? In terms of evolutionary selection, it’s more or less “being suited to your environment”. Note it doesn’t care what that environment is.
The assessments they use for “loss of fitness” following relaxed selection is to compare success of organisms back in their original environment. I.e. take some yeast growing slowly on really hard to metabolise sugars, then grow them on easily metabolised glucose for a few generations, and note that they’re less good at metabolising the harder sugars: they’ve “lost fitness”.
True, but they’ve also GAINED FITNESS when it comes to hoovering up glucose. They lose fitness vs. their previous environment because there’s no selective pressure to retain it, but now there’s definite selective pressure to be the fastest at turning glucose into more yeast.
With respect to human mutations, therefore, what they are doing here is taking the criteria for living in a non-industrialised world full of disease and famine, and applying those to assess the fitness of humans living in a pampered world full of antibiotics and WalMarts.
This is…well, stupid. We’re still selecting, we’re just selecting for genomes more suited to having lots of children in a world full of antibiotics and WalMarts. That is our “fittest” in the west. Resistance to deep vein thrombosis might increase, for example (sedentary lifestyles increase risk of DVT, so there’s pressure there).
In summary, this whole idea is fundamentally flawed. A central, fundamental, and vital part of this sort of analysis is the concept of an IDEAL GENOME. A further requirement is an UNCHANGING ENVIRONMENT.
And neither of these exist.
There is no such thing as an ideal genome which decays over time. All genomes change. All genomes are more or less appropriate for the current environment, and all environments change.
A given mutation can be good, or bad, or neutral, in the current environment. And if it’s good, it will be positively selected. If it’s bad, it will tend to be negatively selected. If it’s bad but not bad enough, then IT’S NOT A PROBLEM. The estimates they give for “probable number of VSDMs” are almost certainly real, but also apply to all genomes, and crucially, ALWAYS HAVE DONE.
No genome has ever been perfect, and the idea is inherently ridiculous.
What would a perfect genome even look like? Everything has consequences. What is the ideal height for humans? Short people have advantages sometimes, disadvantages at other times. Tall people? DITTO. There is no such thing as an ideal height; there is only (at best) the most optimal height for the current environment. Plus mutations that increase height also have pleiotropic effects, often reducing lifespan. Mutations that lower height also have pleiotropic effects. Everything is a balancing act; there is no “perfect”.
Genomes are ALL simply the result of multiple interacting genetic phenomena subject to constant selective pressure. What happens to species that accumulate too many minor deleterious mutations? Well, quite often they turn into different species, because what’s deleterious in one environment is positively advantageous in another. If the changes are too serious, then those individuals carrying them die, and those individuals that don’t carry them persist. En bloc gradual genetic decay is something that can only occur in simulations.
There is no genetic entropy while selection exists, and selection ALWAYS exists.