More Recent Comments

Saturday, October 03, 2020

On the importance of random genetic drift in modern evolutionary theory

The latest issue of New Scientist has a number of articles on evolution. All of them are focused on extending and improving the current theory of evolution, which is described as Darwin's version of natural selection [New Scientist doesn't understand modern evolutionary theory].

Most of the criticisms come from a group who want to extend the evolutionary synthesis (EES proponents). Their main goal is to advertise mechanisms that are presumed to enhance adaptation but that weren't explicitly included in the Modern Synthesis that was put together in the late 1940s.

One of the articles addresses random genetic drift [see Survival of the ... luckiest]. The emphasis in this short article is on the effects of drift in small populations and it gives examples of reduced genetic diversity in small populations.

These findings have big implications, because populations lose their ability to adapt and thrive if they lack genetic diversity for natural selection to work on.

The idea here is that drift is bad because it's an impediment to natural selection, but there's a lot more to random genetic drift than this. In fact, drift is a fundamental and important part of evolution.

Jerry Coyne has been criticising the "intellectual vacuity" of the New Scientist articles and one of his posts addressed random genetic drift [The intellectual vacuity of New Scientist’s evolution issue: 4. The supposed importance of genetic drift in evolution]. Jerry is not a big fan of Neutral Theory and drift but at least he asked the right question ...

Genetic drift certainly operates in populations, for it must given that populations are finite and alleles assort randomly when sperm (or pollen) and eggs are formed. The question that evolutionists have been most concerned with is this: “How important is genetic drift in evolution?”

I'm going to try and answer that question but first here's a short introduction to drift for those of you who need to refresh your memory: Random Genetic Drift.

Nearly-neutral alleles can be fixed in a population

Back in the 1960s, scientists began to realize that most populations had far more genetic diversity than you would expect if natural selection were the only significant force in evolution. This led to the development of Neutral Theory, which is the idea that much of this variation consists of alleles that are neutral with respect to selection so their frequency in the population is determined solely by drift. It explains the large amount of genetic diversity because the fixation and elimination of alleles by random genetic drift is a slow process, [Genetic variation in the human population].

Later, this idea of neutral alleles was extended to include alleles that can be slightly deleterious but are almost invisible to the effects of natural selection under some circumstances. This is Nearly-Neutral Theory [Celebrating 50 years of Neutral Theory].

Neutral and slightly deleterious alleles can be fixed in a population by random genetic drift. This is the important point that you must grasp if you are going to understand drift. It means that changes in the frequenceies of alleles in a population can be due to drift and not just selection.

The drift-barrier hypothesis

The competition between drift and selection is most often decided in favor of drift. That's why the probabilty of fixation of a beneficial allele is approximataly equal to 2s where s is the selection coeficient. Consider a newly-formed allele with a relatively high selection coefficient of 0.01; it will only be fixed 2% of the time in a given population. It will be lost by drift 98% of the time.

This competition is the key component of the drift-barrier hypothesis, which is an important part of modern evolutionary theory. The concept is extended to include population size because the larger the population the greater the power of natural selection. This leads to a variable probablity of winning the competition with drift dominating in small populations and selection dominating in large populations [Learning about modern evolutionary theory: the drift-barrier hypothesis].

Lots of evolutionary biologists think that much of multicellular eukaryotic evolution has been characterized by relatively small populations and that's why selection has not been terribly effective and why deleterious alleles are frequenctly fixed in a population. This explains why mammalian genomes are full of junk DNA but it also explains a lot of other things that aren't very obvious if you don't think drift is important [The frequency of splicing errors reflects the balance between selection and drift].

The molecular clock

Several workers started to compare the sequences of proteins from different species in the early 1960s. They discovered that you could build a tree of these sequences and the tree corresponded roughly to what biologists had already deduced about the evolution of the species. The remarkable think about the molecular trees was that most of the branches were about the same length suggesting that the evolution of protein sequences was proceeding at a relatively constant rate over time.

The explanation is that most of the amino acid substitutions are neutral and the alleles are being fixed by random genetic drift. Since the rate of fixation by drift is equal to the mutation rate, and since mutation rates are relatively constant in different species, this gives rise to the molecular clock. A result that's widely used today [The Modern Molecular Clock].

The role of chance in evolution

An appreciation of the importance of random genetic drift leads to a better understanding of the role of chance in evolution. This can radically transform your worldview if you were previously convinced that natural selection is by far the dominant force in evolution. [Evolution by Accident] [The role of chance in evolution] [Historical evolution is determined by chance events]

Constructive neutral evolution

Constructive neutral evolution is a mechanism for evolving complex structures by non-adaptive mechanisms. It relies on the creation of neutral alleles by mutation and their subsequent retention and fixation by random genetic drift. Let me give you a simple example. Imagine an enzyme (A) that catalyzes a metabolic reaction. Now imagine that another protein (B) just happens to acquire a mutation that causes it to bind to A from time to time.

Enzyme A could acquire a deleterious mutation that makes the free protein much less active but it retains activity when bound to protein B. The deleterious mutation could become fixed by random genetic drift because it is effectively neutral in the context of transient binding of A by B. Now you have a situation where only the complex A/B is active but there was never any direct selection for this increase in complexity.

I don't have time to explain constructive neutral evolution in more detail so I recommend that you check out these posts: [Constructive Neutral Evolution (CNE)] [Rube Goldberg’s 131st Birthday: Irremediable Complexity by Constructive Neutral Evolution]. The important point is that random genetic drift can play a major role in increasing complexity and this complexity is more accidental than adaptive.

The null hypothesis

The study of any biological features, including genomic sequences, typically revolves around the question: what is this for? However, population genetic theory, combined with the data of comparative genomics, clearly indicates that such a “pan-adaptationist” approach is a fallacy. The proper question is: how has this sequence evolved? And the proper null hypothesis posits that it is a result of neutral evolution: that is, it survives by sheer chance provided that it is not deleterious enough to be efficiently purged by purifying selection. To claim adaptation, the neutral null has to be falsified. The adaptationist fallacy can be costly, inducing biologists to relentlessly seek function where there is none. Koonin (2016)

If you don't think that drift is important then you are in danger of becoming the type of adapationist that Gould and Lewontin criticized in the Spandrels paper. Those adapationists fail to recognize the importance of the null hpothesis as described above by Koonin [see: You MUST read this paper if you are interested in evolution].

Large populations evolve by drift

Natural selection is just one of several evolutionary mechanisms, and the failure to realize this is probably the most significant impediment to a fruitful integration of evolutionary theory with molecular, cellular, and developmental biology.

Michael Lynch

How many times have you heard biolgists say that drift is only imporant in small populations? This is a common myth but it's time to set the record straight. The rate of fixation of neutral alleles in a population is equal to the mutation rate and it is independant of population size. Large populations fix alleles by random genetic drift and so do small populations [Random Genetic Drift and Population Size].

It's true that some alleles can be fixed (or lost) rapidly in small populations and it's true that isolated small populations may, by chance, have different allele frequencies than the parent population (founder effect) but that doesn't mean that random genetic drift only occurs in small popluations.

Phenotpic changes can be fixed by drift

The second most common way of dismissing drift is to say that it only applies at the molecular level. "Real" evolution—the kind you can seen with your naked eye—can only occur by natural selection. This is a common adaptationist claim. Sometimes it is applied to speciation events by claiming that drift plays almost no role in speciation. Here's how Richard Dawkins puts it [Richard Dawkins' View of Random Genetic Drift].

As geological time goes by, the genome is subjected to a rain of attrition in the form of mutations. In that small portion of the genome where the mutations really matter for survival, natural selection soon gets rid of the bad ones and favors the good ones. The neutral mutations, on the other hand, simply pile up, unpunished and unnoticed—except by molecular geneticists.

It's simply not true that molecular geneticists are the only one who care about neutral alleles. Look around you the next time you are near a crowd of people. Notice that each and every one of them has distinctive morphological features such as eye color, hair color, shape of the ear and nose, height etc. Most of those features have clear genetic components—that's why children resemble their parents more than they resemble strangers. It's extremely likely that many of these morphological features have no effect on the fitness of the individuals who exhibit them. They are due to neutral alleles that are segregating in the population by random genetic drift. Some of them can reach appreciable frequencies in isolated human populations, which is why natives of Japan look very different than natives of Sri Lanka.

Take a favorite example of Richard Lewontin who points to the fact that the African rhinoceros has two horns while the Indian rino has only one? Is this an example of natural selection? Are two horns better than one in Africa or is the null hypothesis just as likely? What about the difference between red pine, with two needles per cluster, and white pine, with five? Are those adaptations or are they just due to neutral alleles that have become fixed in the two species?

Don't dismiss random genetic drift because you think that all mutations affecting phenotypic change must be affecting fitness [Visible Mutations and Evolution by Natural Selection].

Random genetic drift is by far the most common mechanism of evolution

Jerry Coyne asked “How important is genetic drift in evolution?” and we can now answer that question. If you consider the fixation of alleles in a population as one measure of evolution then the vast majority of those alleles are either neutral or deleterious. Those alleles were fixed by random genetic drift so that makes drift the dominant mechanism of evolution. Adaptationists will try and avoid this conclusion by defining evolution as adaptation or phenotypic change but that's not all there is to evolution. We can safely say that almost all adaptations are due, in part, to natural selection but that's also mispleading since drift even plays a role in adaptations.

There's no avoiding the conclusion that random genetic drift is extremely important in evolution.

Koonin, E.V. (2016) Splendor and misery of adaptation, or the importance of neutral null for understanding evolution. BMC biology, 14:114. [doi: 10.1186/s12915-016-0338-2]


  1. Thanks for your post Professor Moran. I am eagerly waiting for your book.

  2. This is a very good summary of this subject. I've bookmarked it for future references.

  3. Hi Laurence A. Moran,

    some criticism on this:

    (i) A hypothesis cannot be part of a theory, but there may be a hypothetical extension of the theory in question. Example: The Copenhagen Interpretation of Quantum Mechanics. This relativizes the statement that the drift-barrier hypothesis is an important part of the modern evolutionary theory enormously.

    (ii) The statement “selection coefficient of 0.01” is structurally identical to the statement “it will only be fixed 2% of the time in a given population”. Of course, this is intended to provide a vivid representation. But what is the reason for this selection coefficient in this model? Also the axis labeling “performence”/”perfection”...

    (iii) What is the problem if the Sewall Wright effect is not perceived as the result of an adaptation? One reason may be that this cannot be resolved theoretically. From an evolutionary point of view only local optima exist (= survive).

    (iv) »Constructive neutral evolution is a mechanism for evolving complex structures by non-adaptive mechanisms. It relies on the creation of neutral alleles by mutation and their subsequent retention and fixation by random genetic drift.« Please show. Perhaps using the example of Pax 6?

    (v) Of course I know doi: 10.1186/s12915-016-0338-2. False dilemma: But if you carelessly state genetic drift is the null hypothesis of adaptation, then this explains nothing. Because then also adaptation is the null hypothesis of genetic drift. Of course: A suitable null hypothesis would be for example drift vs. no drift...

    (vi) Imagine a gene G1 that transcribes the protein P1. Now imagine that another gene G2 ensures that P1 is broken down in tissue T2. If P1 is located in T2, it should be unfavorable for the fitness of the organism concerned. Furthermore, the lack of P1 in the other tissues of T1 should also be detrimental to the fitness of the organism. First of all, a causal course of development can be deduced from this. Then some relations of the mentioned matrix are optional. But in the resulting population dynamics it says then: At the end of the day the selection works.

    (vii) About Lewontin's rhino: Please compare the relative skull length and the number of horns between adult and juvenile Ceratotherium simum.