Genetic Drift for dummies

How, when and where does drift happen?

I don’t know much about the Kahn Academy but their Wikipedia entry suggests they don’t have any political or religious axe to grind. Here is their description of genetic drift:

Genetic drift is change in allele frequencies in a population from generation to generation that occurs due to chance events. To be more exact, genetic drift is change due to “sampling error” in selecting the alleles for the next generation from the gene pool of the current generation. Although genetic drift happens in populations of all sizes, its effects tend to be stronger in small populations.

So drift happens all the time but its effects are more noticeable in small populations and two ways that populations become small have their own names: bottleneck effect and founder effect. Bottlenecks occur when some catastrophe reduces a population to a small number of individuals, a partial extinction. The result is that individuals carrying alleles (alternative copies of genes) occurring at low frequency may be eliminated altogether or so reduced in number that those alleles are lost for good in the next generation or two and thus genetic diversity is lost. Founder effect is where a small sample of a large population (or the extreme example of a single individual) becomes permanently isolated from the main population, such as by being transported to a new environment devoid of the main population. Again, alleles present at a low level can be lost completely in the isolated small population.

I don’t think there is controversy that genetic drift is a real phenomenon. What I have had difficulty in seeing is how genetic drift contributes to evolution as a whole. Being, by definition, a random process and one that, in small populations, reduces diversity, I think I’m correct that drift does not contribute to selective adaptation. But perhaps, where small subsets of populations go through a bottleneck, it creates a consequent springboard for subsequent adaptation in the surviving population.

But I place myself in the dummy category regarding drift and appreciate any corrections as necessary in my text above. This isn’t the first time I’ve expressed skepticism as to the importance that drift should be accorded so my apologies for being a slow learner and I look forward to input from those more informed than me.

I should credit Berkeley University University of California, Berkeley for the diagram above.

165 thoughts on “Genetic Drift for dummies

  1. : Alan Fox: Jewishness is both an ethnicity and a religion

    Alan Fox: You should also be aware that suggesting that Jewishness is genetic is racist in my view.

    Alan Fox: Ethnicity does not depend on genetic make up but on the culture you you absorb and are immersed in. There are certainly genetic markers indicating biological ancestry.

    These three posts taken together belong in some kind of comedy hall of the absurd.

  2. Alan Fox: If you want to make jokes about the oddness of the Jewish religion, or indeed about any religion

    In your wildest distortion of the truth, you can’t possibly claim that it has not been a consistent theme of posters here to not just make fun of Christianity, but to likewise make fun of people who believe in Christianity (And ID, and Creationism, etc…).

    So I wonder what makes you pretend otherwise. There is no consistency in your position Alan. You can claim your motives are authentic, but the evidence shows otherwise. Its just a question of it being willful bias or just blind devotion to your viewpoints.

  3. phoodoo: but the evidence shows otherwise. Its just a question of it being willful bias or just blind devotion to your viewpoints.

  4. Alan Fox:
    phoodoo,

    So is that supporting your claim that I’m a racist or an admission that you can’t?

    Alan, what’s wrong with you? You suggested that it is racist to call being Jewish genetic (which by itself would be a totally bizarre use of both logic and English). You then said that being Jewish is an ethnicity, with genetic components. In fact, ethnicity is often considered to be inherited (If the ethnicity is only a choice, or a culture, than how can it be racist to make fun of a culture rather than a race?-Again you won’t understand any of this, but-balloons!).

    So, if you prefer not to use English, and instead wish to use some version of telly-tubbies babble talk, then perhaps my reply to you should be “Admission, what do carnivals have to do with rainbows, although rainbows are nice, I never said rainbows aren’t. Cotton candy!”

  5. phoodoo: In fact, ethnicity is often considered to be inherited (If the ethnicity is only a choice, or a culture, than how can it be racist to make fun of a culture rather than a race?-Again you won’t understand any of this, but-balloons!).

    narrator: Nobody was surprised by what happened next.

    https://www.npr.org/2019/12/07/785804791/uighurs-and-genetic-surveillance-in-china

    And in 2016, 2017, blood samples from essentially the entire population, people 12 to 65 in Xinjiang, was collected and potentially put in that database. And it can be part of a broader system of what we call total surveillance.

    SIMON: So can this technology be used, for example, to survey a crowd and pick out Uighurs or, let’s say, Tibetans, Miao people, Yao people, other minorities or their relatives?

    MOREAU: So defining ethnicity is extremely, extremely messy. So actually, ethnicities is first a social and cultural concept. And now suddenly, we’re talking about genes. However, it makes it possible to tomorrow decide that someone does belong or does not belong to a certain population.

    I’m extremely concerned about this because in history, actually, if you look back in the first half of the 20th century, German and then Belgian colonists in Rwanda and Burundi actually went there, and they were using pseudoscientific ideas about race and assigned people to a particular ethnicity. That actually was a significant factor in genocides. And the risk for this in the midterm is actually really worrying.

    Who’s a good lapdog for his authortaionam masters? phoodoo, is! good phoodoo, good phoodoo.

  6. Alan Fox:
    phoodoo,

    Jokes are a good example of how insidious racial stereotyping can be. Advertising too. None of which has anything to do with genetic drift.

    https://en.m.wikipedia.org/wiki/Ethnic_stereotype

    A lot of people consider it racist to use the terms jokes and advertising together. I am hesitant to even write this, except as a teaching example.

    In fact the word stereotype derives from solid printing presses. Why do you have to go down this road Alan?

    Alan Fox: None of which

    Nun of witch? Niche? Are you making fun of Cockneys.

  7. phoodoo: A lot of people consider it racist to use the terms jokes and advertising together.

    Name one. That’d be, you know, supporting a claim. Something you don’t seem able to do. A common pattern it appears.

  8. OMagain,

    I’m extremely concerned about this because in history, actually, if you look back in the first half of the 20th century, German and then Belgian colonists in Rwanda and Burundi actually went there, and they were using pseudoscientific ideas about race and assigned people to a particular ethnicity. That actually was a significant factor in genocides. And the risk for this in the midterm is actually really worrying.

    I find it really worrying too.

  9. Too lazy to read all the previous comments, so my apologies if this has already been mentioned. . .

    Neutral drift contributes to selective advantages through epistasis. A neutral mutation can later become advantageous when it interacts with a new mutation. Increased variation through neutral drift increases the chances for beneficial epistatic interactions.

  10. T_aquaticus:
    Too lazy to read all the previous comments, so my apologies if this has already been mentioned. . .

    All substantive comments welcome.

    Neutral drift contributes to selective advantages through epistasis. A neutral mutation can later become advantageous when it interacts with a new mutation.Increased variation through neutral drift increases the chances for beneficial epistatic interactions.

    Everything makes sense to me except “Neutral drift contributes…”. As I understand it, drift results in reduction in diversity. How does this contribute to variation?

  11. Alan Fox:

    Everything makes sense to me except “Neutral drift contributes…”. As I understand it, drift results in reduction in diversity. How does this contribute to variation?

    It only reduces diversity at the limit – if it promotes an allele toward fixation, or eliminates one (same thing, in fact). Which is the same as selection.

    If one imagines a set of neutral mutations occurring and being increased by drift – often only temporarily – we would expect an increase in variation from these random factors alone, compared to a scenario with a constraint of selective advantage.

    In fact, without randomness, most variants, even beneficial ones, would not get off the ground. While rare, random factors would dominate, even with an advantage. If the advantage is (say) an extra offspring per thousand, a significant selection coefficient, it would take a lot of dice rolls before that bias had any significant effect over background.

  12. Thanks for being gentle with me, Allan. Perhaps my lack of imagination is preventing me from doing this:

    If one imagines a set of neutral mutations occurring and being increased by drift…

    Is this what I am missing? Eliminating alleles randomly is a loss of diversity but that loss is balanced by the increasing frequency of the allele moving to fixation, bringing it to a level where it will be much more visible to future selection. Rare alleles (and new combinations of genes) are not picked up by selective processes and drift facilitates the fixation of fewer numbers rather than many alleles at low levels which might be supposed to occur if drift were not unavoidable.

    Not convincing myself here.

  13. Allan Miller: It only reduces diversity at the limit – if it promotes an allele toward fixation, or eliminates one (same thing, in fact). Which is the same as selection.

    Ha, if an allele is being promoted towards fixation, then by the very definition its not neutral.

    Have you forgotten how you calculate fitness? Remember when I was telling you that its circular? Its this very reason which insures that no definition of neutral could ever make sense, if it increases the frequency of an allele.

    You are in a fight with your own concepts.

  14. phoodoo: Ha, if an allele is being promoted towards fixation, then by the very definition its not neutral.

    Have you forgotten about the niche again. If alleles competing for a locus are neutral (no selective bias) in a particular niche, one will eventually fix randomly. This is what random genetic drift is, by definition.

    Have you forgotten how you calculate fitness?

    It’s relative fitness. It is normally done by counting offspring.

    Remember when I was telling you that its circular?

    I don’t think many folks worry about your assertions.

    Its this very reason which insures that no definition of neutral could ever make sense, if it increases the frequency of an allele.

    Nonetheless, the usual definition for relative fitness will carry on regardless. Unless you can come up with a better one.

    You are in a fight with your own concepts.

    He wrote, while thinking wishfully.

  15. Alan Fox: If alleles competing for a locus are neutral (no selective bias) in a particular niche, one will eventually fix randomly. This is what random genetic drift is, by definition.

    That’s also what natural selection is.

    Will you ever see how stupid this is?

  16. phoodoo:
    Alan Fox: If alleles competing for a locus are neutral (no selective bias) in a particular niche, one will eventually fix randomly. This is what random genetic drift is, by definition.

    phoodoo: That’s also what natural selection is.

    Nope, it isn’t. I added some emphasis to the quote to help you out.

    phoodoo:
    Will you ever see how stupid this is?

    I see that something is stupid, but I doubt we’re talking about the same thing.

  17. phoodoo: phoodoo on February 3, 2021 at 4:18 pm said:

    Alan Fox: If alleles competing for a locus are neutral (no selective bias) in a particular niche, one will eventually fix randomly. This is what random genetic drift is, by definition.

    That’s also what natural selection is.

    Oh dear phoodoo. Selection (whether natural, artificial or sexual) is change in allele frequency due to the bias of the niche. Random genetic drift is random change in allele frequency where there is no selective bias on allele frequency and one allele eventually fixes, well, randomly.

    Will you ever see how stupid this is?

    Remember the aphorism: first understand, then criticise.

    And the niche, phoodoo. remember the niche.

  18. phoodoo: Ha, if an allele is being promoted towards fixation, then by the very definition its not neutral.

    That is not the case. Even a mildly deleterious allele can be fixed a proportion of the time. Because – as has been explained to you ad nauseam – fitness is the expected value, not the specific outcome of a particular trial. Your argument is equivalent to 4 heads in 10 throws proving that the coin is not fair, even though this can easily happen with a fair coin. The expected value is 5 apiece, but it does not always happen, without any appeal to tautology.

    Have you forgotten how you calculate fitness?Remember when I was telling you that its circular?Its this very reason which insures that no definition of neutral could ever make sense, if it increases the frequency of an allele.

    It’s not neutrality that increases the frequency of a neutral allele, it is randomness. If there is no bias whatsoever, one allele will still become dominant in a neutral situation, as admirably demonstrated by OM’s admirable simulation of the random process.

    You are in a fight with your own concepts.

    Rather, with your misunderstood version of these concepts.

  19. Alan Fox:
    Thanks for being gentle with me, Allan. Perhaps my lack of imagination is preventing me from doing this:

    If one imagines a set of neutral mutations occurring and being increased by drift…

    Is this what I am missing? Eliminating alleles randomly is a loss of diversity but that loss is balanced by the increasing frequency of the allele moving to fixation, bringing it to a level where it will be much more visible to future selection.

    I shouldn’t get too fixated on fixation. It is a special case: the moment for an allele A at which the last allele not-A is lost. Exactly the same happens when a brand new allele never gets past that one copy. If A is the resident, a not-A mutant flickers briefly before A returns to fixation, 100%.

    But the broader reality is that novel alleles are being chucked up all the time. While rare, or recessive, the predominant determinant of their increase is drift. This makes them a-bit-less-rare, which among other things increases the chance that a beneficial combination between two such alleles will arise. This would not be possible if alleles could only become common through selection.

    Having become common (through drift) many such neutral alleles will then be lost (through drift) if they offer no subsequent advantage, or fail to pair in an advantageous manner. But such speculative punts are being offered up continually, increasing the chances that a beneficial allele or combination from this random foam will catch the selective wind and sail towards fixation by more than drift alone. And, increasing the standing variation in the population.

  20. Genetic drift on average reduces diversity in a population. But when there are two or more populations, it also increases diversity between populations, since different alleles often increase in different populations. If there is no gene flow between the populations, ultimately all the diversity within the populations gets converted to diversity between populations.

    If there are, say, several alleles in a population, and there is no natural selection, no further mutation, and no gene flow, one of the alleles will ultimately fix in the population. The probability of an allele fixing is just its initial gene frequency. This is the case because a random copy at the locus ends up being the progenitor of the all the copies in the population after about 4N generations. And of course the probability that this copy is of allele k is the initial gene frequency of that allele.

  21. Joe Felsenstein,

    But surely when mutation is not turned off, neutral drift will result in more variation than selection? That is, while a given ancestral locus must ultimately become fixed as the progenitor of all descendant sequences, mutation+drift would ensure that variants of that founder sequence persist for longer on the average than if selection was in operation, with the number of such variants dependent on mutation frequency and so (partly) on population size.

  22. Allan Miller: Because – as has been explained to you ad nauseam – fitness is the expected value, not the specific outcome of a particular trial

    Oh, the expected value, not the actual value, right I remember now. And the expected value is based on… Not actual values? Haha, good one.

    I like Alan’s take. Selection is when there is a bias and alleles out compete others and drift is when there isn’t a bias and alleles outcompete. That is so cool. Because sometimes it’s the niche that causes the disparity and sometimes it’s not the niche that causes the disparity. This is getting brilliant!

    You know how you can tell the difference Allan? The expected niche!

  23. phoodoo: Oh,the expected value,not the actual value,right I remember now. And the expected value is based on… Not actual values?Haha,good one.

    I like Alan’s take. Selection is when there is a bias and alleles out compete others and drift is when there isn’t a bias and alleles outcompete. That is so cool.Because sometimes it’s the niche that causes the disparity and sometimes it’s not the niche that causes the disparity.This is getting brilliant!

    You know how you can tell the difference Allan?The expected niche!

    It’s very closely analogous to a small piece of carbon ink in water, undergoing Brownian motion (analogous to genetic drift) and gravitation (analogous to selection). Is that so hard to understand?

  24. Allan Miller: But surely when mutation is not turned off, neutral drift will result in more variation than selection?

    But mutation, and other sources of variation, occur randomly (caveats regarding mutagens etc) prior to drift’s influence. Where is the connection?

  25. phoodoo: Because sometimes it’s the niche that causes the disparity and sometimes it’s not the niche that causes the disparity. This is getting brilliant!

    No, you’re not getting it. Toss a fair coin enough times and the average result is 50% H. Use a weighted coin and the result will be biased away from 50%. The niche is analogous to weighting the coin. Or see what Joe wrote.

  26. phoodoo: Oh,the expected value,not the actual value,right I remember now. And the expected value is based on… Not actual values?Haha,good one.

    This is just basic random-variable stats (as has also been explained to you ad nauseam). One does not necessarily know what the ‘expected’ value is. If you make a biased die, it has an expected value before you have even rolled it once. If you have a problem with that, it is with the entirety of probability and sampling, not merely with evolution.

    I like Alan’s take. Selection is when there is a bias and alleles out compete others and drift is when there isn’t a bias and alleles outcompete.

    So basically you accept that evolution occurs, due to both drift and selection; it just annoys you.

    That is so cool.Because sometimes it’s the niche that causes the disparity and sometimes it’s not the niche that causes the disparity.This is getting brilliant!

    Yes, I know, you excitedly think you have undone a century of mathematical work at a stroke. It is both cute and tragic at the same time.

  27. Alan Fox: But mutation, and other sources of variation, occur randomly (caveats regarding mutagens etc) prior to drift’s influence. Where is the connection?

    Those mutations subsequently change in frequency due to random effects, too. If drift were not a thing, a new mutation would (I imagine; it’s hard to visualise a realistic example) just exist in single copy, neither increasing nor decreasing. It might reoccur as a mutation I suppose, but far more likely is increase by drift.

    You don’t get pockets of relative variation from mutation alone, nor from selection alone. For recessives (for example), drift is the only means by which they can increase to a level at which they are freed of dominance effects – to allow selection to ‘see’ the allele in homozygotes.

  28. Allan Miller,

    Did I misunderstand? If you are not saying drift affects mutation rates then no problem. Surely new alleles proliferate by reproduction?

  29. Alan Fox:
    Allan Miller,

    Did I misunderstand? If you are not saying drift affects mutation rates then no problem. Surely new alleles proliferate by reproduction?

    I think you must have misunderstood, yes. Drift doesn’t affect mutation rates; mutation and drift are both ‘random’.

    ‘Proliferation by reproduction’ is drift, for a neutral allele.

  30. Alan Fox,

    What I was trying to emphasise is that drift takes longer to fix an allele than selection, and so more variants can be sustained under a drift scenario than under selection.

    If the ancestral sequence AAAAAA was advantageous, the comparative speed of selection means that it becomes fixed as an ancestor more rapidly. ‘Fixed as an ancestor’ meaning that the entire population descends from that sequence at that locus, even if they vary in detail – a handful of AAAAAB’s, AAAAAC’s etc (assuming they are equally as fit as AAAAAA) may be found; AAAAAA has not yet been fixed as a sequence.

    But with drift’s more leisurely progress, by the time AAAAAA has become ‘fixed as an ancestor’, it has had time to spawn more variants than in the ‘selective sweep’ scenario. Descendants of AAAAAA will be ‘more polymorphic’.

    Therefore drift has the capacity to increase a population’s standing variation over a selection-only scenario.

  31. Allan Miller,

    Oh this is such bullshit Allan. There is ZERO, absolutely zero method for claiming one type of fixation is based on random drift, and another on “selection.” In fact the very idea of selection (and on Alan’s omnipotent, omniscient, omnipresent niche) makes it impossible to separate events into random and biased. The measurement of bias is simply the result. There is no other metric.

    So when you start making these claims about one type of inheritance being based on neutral happenstance, and another on unknown factors of preference and “great” survival, and the different speeds at which these can occur, you are basing it on nothing other than desire and assumption. And we have explained this to you a thousand times. You just don’t want to learn!

  32. phoodoo,

    Yes, phoodoo. It’s me that doesn’t want to learn. 🙄

    Still, glad to hear you now accept evolution, and are simply concerned about a Measurement Problem in determining, after the fact, whether a bias was in operation or not.

  33. phoodoo: There is ZERO, absolutely zero method for claiming one type of fixation is based on random drift, and another on “selection.”

    ahem

    Am I right in supposing that you put ZERO, absolutely zero effort into verifying that your claim wasn’t actually total bollocks?

  34. I wonder if Purgatory will consist of me arguing phitness with phoodoo for all eternity? (Well, I can get why I might be subject to that, miserable sinner that I am, but I don’t know what he’s doing there! 🤔)

  35. phoodoo: You mean like ID?

    No, like populations reproducing and this having an unavoidable effect on allele frequencies, with or without biases, to the point of fixation at the limit.

  36. Allan Miller: Well, I can get why I might be subject to that, miserable sinner that I am, but I don’t know what he’s doing there!

    I guess that would be heaven for phoodoo.

  37. Corneel,

    Great, can you tell me were eyebrows the result of selection or drift? Can you just plug that in and tell me, would you mind?

  38. phoodoo: Great, can you tell me were eyebrows the result of selection or drift? Can you just plug that in and tell me, would you mind?

    Would your pride survive if you were to acknowledge that your claim:

    There is ZERO, absolutely zero method for claiming one type of fixation is based on random drift, and another on “selection.”

    was a tiny bit exaggerated?

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.