Many people are under the false impression that evolution is always beneficial, but it can actually be quite harmful. This is especially true for evolution via genetic drift. Like natural selection, genetic drift removes variation from a population, but unlike selection, it is random and it can remove beneficial traits. Indeed, we often become so focused on selection that we assume that all of the features that we see in organisms were selected because of some beneficial role, but that is not true. Some of those features arose via the random process of genetic drift, and at times this process can even swamp natural selection and cause harmful traits to rise to prominence. In other words, evolution via genetic drift can actually be very detrimental to populations. So in this post, I want to talk about what this mechanism is, how it works, and how it interacts with the other evolutionary mechanisms.
What is genetic drift?
In its simplest terms genetic drift is just a random change in the allele frequencies of a population over time. This is in stark contrast to natural selection, in which the change in allele frequencies is not random (it is a common misconception that natural selection is random). Remember, evolution is simply any change in a population’s allele frequencies over time, so the key defining feature of genetic drift is that this change is random. The obvious question is then, what causes allele frequencies to change randomly? There are several answers to that question, but the classic answer (and indeed the one that is constantly at play) is the random nature of independent assortment.
If you think back to high school biology, you have two copies of your chromosomes, one of which you received from your mom, and one of which you received from your dad. Further, these chromosomes are arranged into pairs (i.e., you got one copy of chromosome #1 from mom, and one copy of chromosome #1 from dad). Additionally, you have two copies of each gene (each copy is known as an allele), and each copy is on a different chromosome in the pair (one from mom and one from dad). So, for example, if you have the blood type AB and your mom is AA and your dad is BB, then that means that you received a chromosome with the A allele from your mom, while the other chromosome in that pair came from your dad and contained the B allele. Now, every time that you produce an egg or a sperm cell, you will only pass on half of your chromosomes (one chromosome from each pair). So each individual sperm or egg will receive the A allele or the B allele, but not both. Importantly, which allele you pass to a given egg/sperm is completely random and is not affected by which other chromosomes you pass on. This is what we call independent assortment, and it is a key source of genetic drift. (Technical note: if you want to be more precise, independent assortment refers the random arrangement of homologous chromosomes along the metaphase plate during meiosis I, but the key point is simply that which member of a chromosome pair gets passed on is random)
Because which chromosomes you pass on is completely random, the allele frequencies can shift overtime if, just by chance, one chromosome happens to get passed more than the other. I’m going to use a simulator to illustrate this in a minute, but for now, let me illustrate with a coin (if you have one handy, please grab a piece of paper and try this yourself). First, assume that you have two individuals, each of which is a heterozygote, meaning that they have one copy of each allele (which in this case will be heads and tails). Now, to make the first individual, flip the coin twice, and whatever it lands on, that will be your first offspring. The first coin flip represents the allele from parent 1 (i.e., there is a 50% chance of passing H and a 50% chance of passing T, just like in independent assortment), and the second flip represents the allele from parent 2. Now, do it again to make a second offspring. There are only six possible outcomes: both TT, both HH, both HT, one HH one TT, one HT one TT, one HT one HH (TH and HT are the same). Write down your first two offspring, then “mate” them. If both of your offspring are HT, then the procedure is identical to what you did before; however, if one of them is a homozygote (i.e., only has one type of allele), then it can only pass on that allele (i.e., a TT can only pass T and HH can only pass H, thus if your offspring are TT and HH then both grandchildren will be TH). Make two more children in this manner (e.g., if you have an HH and a HT, then each of your two new individuals will receive an H from the HH parent and either an H or a T from the HT parent, flip the coin to decide which). Keep flipping your coin and making new generations like this, and you will pretty quickly get to a point where one of your alleles disappears and all that you have is TT or HH. At that point, we say that the allele is “fixed,” meaning that one of the alleles has been lost, and every individual in the population is now homozygous for the other allele (i.e., has two copies of it). The first time that I tried this, it only took five generations for one allele to become fixed, but if you do it numerous times, you’ll notice that the number of generations until fixation varies, and if you do it enough times, 50% of the time heads will become fixed and 50% of the time tails will become fixed.
What you have just done is simulate genetic drift. That’s really all that it is. Random chance produces slight variations in allele frequencies until one allele eventually becomes fixed. The image on the left illustrates the situation more clearly. It shows genetic drift in a population of 100 individuals. As you can see, the allele frequency randomly oscillates up and down until eventually the dominant allele becomes fixed.
Effects of population size
If you think about the math behind genetic drift, you would intuitively expect genetic drift to be more severe when population sizes are small, and indeed that is exactly what happens in nature. In fact, at very large population sizes, we expect genetic drift to have a relatively small effect. To illustrate this, I used a simulator to simulate four different scenarios (illustrated below). In each situation, I simulated 10 populations that consisted of 2, 10, 100, or 1000 individuals, and each population started with even allele frequencies for the gene being simulated (i.e. half of the alleles were dominant and half were recessive). As you can see, when there were only 2 or 10 individuals in the populations, alleles became fixed very rapidly (the simulations with 2 individuals are the same thing as what you did with a coin). When the population size jumps up to 100, however, things are more stable, but by the end of the 50 generations shown, you can see that some populations are trending towards having fixed alleles, and indeed after a few hundred generations all of the populations became fixed (not shown). Finally, the populations with 1000 individuals retained fairly stable allele frequencies, but even in those cases, alleles will eventually become fixed unless they are acted upon by another evolutionary force, and, in fact, in natural populations, those other mechanisms (particularly gene flow) do often interact with genetic drift and prevent the fixation of alleles in large populations.
Note: the next post in this series will be entirely devoted to gene flow, so I will talk about how it interacts with genetic drift in detail there.
Interactions with selection
When talking about genetic drift, we are often talking about neutral alleles (i.e., alleles that are neither beneficial nor harmful to individuals); however, it can occur for alleles that are not selectively neutral. Imagine a situation where half of the alleles in a population for a given trait are dominant (A) and half are recessive (a); however, the environment changes, and as a result, individuals with a dominant phenotype (i.e., they have at least one dominant allele, so they are either AA or Aa) survive to a reproductive age 100% of the time, whereas individuals with a recessive phenotype (aa) only survive to a reproductive age 90% of the time. In other words, there is selection against the recessive allele, because it reduces an individual’s ability to reproduce when two copies of it are present. Now, you should intuitively expect that natural selection will act on this situation and remove the receive allele from the population, and when the population is large enough, you would be correct. However, when the population is small then, depending on the strength of selection, genetic drift can actually overpower selection.
To illustrate this, consider the figure on the right. I once again used the simulator to simulate 10 populations with 10 individuals and 10 populations with 100 individuals, but this time I set a selection differential so that all of the AA and Aa individuals would survive to a reproductive age, but individuals who had two recessive alleles (aa) only had a 90% chance of surviving. As you can see, in all 10 populations of 100 individuals, selection removed the harmful allele and fixed the beneficial one. In four of the populations with only 10 individuals, however, the populations were so small that genetic drift overpowered selection and actually caused the recessive allele to become fixed! That is a really bad situation because, barring any gene flow or mutations, those populations are now stuck with the harmful allele, and the overall survival of those populations is 10% lower than the survival of populations that managed to rid themselves of the harmful allele.
As I’m sure you can imagine, this has extremely important implications for wildlife conservation efforts, and it is one of the key reasons that conservationists are so concerned with maintaining large numbers of individuals. When populations are small, genetic drift can cause harmful alleles to rise to prominence or even become fixed, and that is a very bad thing for the survival of those populations.
Genetic drift removes variation
It should now be clear that genetic drift removes variation from populations. Indeed, in many of the simulations that I have illustrated you can see that one of the alleles became fixed while the other was lost from the population. This is important because it means that, just like selection, genetic drift is constantly causing populations to become less diverse. As a result, populations are heavily reliant on gene flow from neighboring populations (which can restock their genepool with alleles that they had lost) as well as mutations (which are the only mechanism that is capable of making new variation).
Even when the alleles that are lost are neutral, this is often a serious problem for populations in the long run. Remember, selection simply adapts organisms to their current environment, so although an allele may be neutral at the moment, it may become very important if the environment changes. Indeed, a high level of genetic variation is one of the key factors for determining whether or not a population will survive changes in the environment, the introduction of a new disease or predator, etc.
Bottlenecks and founder events
Now that you understand the basics of genetic drift, I want to introduce you to two final concepts. The first of these is a genetic bottleneck. Bottlenecks are often considered to be a type of genetic drift, but they act a bit differently from the type of genetic drift that we have been talking about so for. In bottlenecks, a large number of individuals is rapidly lost, and as a result, the genetic variation is reduced to a small subset of what it was before. Image, for example, that there is a large population of frogs living at the base of a volcano, and 80% of the frogs contain an allele for green pigment while 20% contain an allele for brown pigment. Then, the volcano erupts and kills off 90% of the frogs. Further, just by chance, all of the frogs with the allele for brown pigment were killed during the eruption. This is, therefore, clearly a case of instantaneous evolution by genetic drift because evolution is a change in allele frequencies, and the allele frequencies change from 20% and 80% to 0% and 100%. Additionally, this would likely represent a genetic bottleneck because it is likely that many other alleles were lost as well. Indeed, one of the characteristics of a bottleneck is the loss of many rare alleles. Finally, the effects of a bottleneck are largely determined by how many generations it lasts for (i.e., how long the population remains small), because as you recall, small populations have greater genetic drift. So even alleles that survived the volcanic eruption may quickly be lost due to genetic drift, if the population does not grow rapidly.
At this point you may be wondering when it is appropriate to talk about a bottleneck as a type of genetic drift, and that is honestly something of a grey area, with some people/books preferring to entirely separate the two, while others lump them together. I personally think that the best way to think about this is to remember than genetic drift is a random change in allele frequencies. So, if the bottleneck killed individuals randomly (i.e., no individuals had alleles that made them more likely to survive the eruption) then it is ok to talk about it as a type of genetic drift. However, if something kills many individuals but does so by selecting its victims, then it should really be thought of as a natural selection event. For example, if a disease outbreak kills off 90% of individuals, and only the 10% of individuals that had alleles that made them resistant to the disease survived, then that likely would be a genetic bottleneck, but it would not be genetic drift because the survivors were selected rather than being chosen randomly. Genetic drift could, however, come in to play in the following generations if the populations do not recover rapidly enough (i.e., they remain small).
On a side note, the recovery of populations from disease outbreaks and whether or not the outbreaks caused genetic bottlenecks is actually a key focus of my current research.
A founder effect is really just a special type of a bottleneck, and it occurs when a new population is formed from a subset of the original population (i.e., the new population only contains a small portion of the genetic variation found in the original population). For example, let’s go back to our population of brown and green frogs, but this time, instead of a volcano, imagine that they live on the edge of a lake, and one day a storm blows several of them out to a previously uninhabited island. However, all 10 of the individuals that form the new population on the island contained only the allele for brown pigment. We would describe this as a founder event, because the new population is limited to the genetic material contained in the individuals that founded it (e.g., the green allele is not present in this new population), but it often also presents a bottleneck, because these new populations usually only contain a small portion of the variation that was in the source population.
In summary, genetic drift is simply an evolutionary mechanism that causes random changes in allele frequencies over time. It is most powerful when population sizes are small, and in some situations it can actually cause harmful alleles to become fixed in a population. As a result, it is a major concern for conservation efforts, and it is one of the reasons that conservation biologists place a high priority on maintaining large populations.
Other posts in this series
- Evolutionary mechanisms part 1: What is evolution?
- Evolutionary mechanisms part 2: Simulating evolution
- Evolutionary mechanisms part 3: The benefits of mutations
- Evolutionary mechanisms part 4: Natural selection
- Evolutionary mechanisms part 5: Sexual selection
- Evolutionary mechanisms part 7 : Gene flow