More Recent Comments

Thursday, March 14, 2013

Misconceptions about Random Genetic Drift

Genetic Drift

Evolutionary change that occurs by random sampling of different alleles from one generation to the next. This causes nonadaptive evolutionary change.

Jerry Coyne
"Why Evolution Is True"
There seem to be two important themes in the current pedagogical literature on science education. One of them is about student-centered learning—a concept I think we should all adopt. The other is about student misconceptions and how to deal with them. Much of the literature suggests that misconceptions need to be confronted and corrected. They can't be corrected by simply presenting the "correct" information. You need to actually address the misconception and show why it is wrong. This is a form of "teach the controversy" and that's not going to sit well with many American supporters of evolution.

Here's an interesting paper on "Biology Undergraduates’ Misconceptions about Genetic Drift" (Andrews et al., 2012). The abstract covers all the important points.
This study explores biology undergraduates’ misconceptions about genetic drift. We use qualitative and quantitative methods to describe students’ definitions, identify common misconceptions, and examine differences before and after instruction on genetic drift. We identify and describe five overarching categories that include 16 distinct misconceptions about genetic drift. The accuracy of students’ conceptions ranges considerably, from responses indicating only superficial, if any, knowledge of any aspect of evolution to responses indicating knowledge of genetic drift but confusion about the nuances of genetic drift. After instruction, a significantly greater number of responses indicate some knowledge of genetic drift (p = 0.005), but 74.6% of responses still contain at least one misconception. We conclude by presenting a framework that organizes how students’ conceptions of genetic drift change with instruction. We also articulate three hypotheses regarding undergraduates’ conceptions of evolution in general and genetic drift in particular. We propose that: 1) students begin with undeveloped conceptions of evolution that do not recognize different mechanisms of change; 2) students develop more complex, but still inaccurate, conceptual frameworks that reflect experience with vocabulary but still lack deep understanding; and 3) some new misconceptions about genetic drift emerge as students comprehend more about evolution.
We all know that evolution is a key concept in biology and every student should take a course on evolution. That course should cover population genetics and the main mechanisms of evolution—including random genetic drift. Among experts on evolution, there's no disagreement on those points.

The problem, from an educational perspective, is that students aren't prepared to learn about random processes because they enter university with a deep-seated misconception that evolution is all about selection and purpose.
Indeed, the most tenacious misconception in biology may be the idea that all processes serve a purpose (Gregory, 2009; Kelemen and Rosset, 2009; Mead and Scott, 2010). This idea is so deep-seated that students fail to even consider random processes as responsible for biological patterns (Garvin-Doxas and Klymkowsky, 2008). The fact that random processes confound students is particularly worrisome, because random processes occur at every level of the biological world, from gene expression (Cai et al., 2006) to clade diversification and extinction (Raup et al., 1973).

Despite these obstacles, understanding random processes such as genetic drift is essential for a deep understanding of the theory of evolution. In contrast to natural selection, genetic drift is nonselective and therefore results in nonadaptive changes in populations (Beatty, 1992). Genetic drift occurs in any finite population and therefore occurs in every population all the time (Futuyma, 2005; Barton et al., 2007).
The authors asked 356 biology students to define genetic drift and only 12% of them could do so satisfactorily. This was part of an American study on developing effective criteria for teaching evolution to undergraduates.

The most common misconceptions involved confusing genetic drift with genetics, thinking that genetic drift was a form of natural selection, mixing up drift and mutation, and believing that genetic drift was about populations moving to new locations.

This was AFTER students had taken a course on evolution! When tested BEFORE taking an evolution course, only 1% of students had some knowledge of genetic drift.

The authors don't propose any easy solutions to the problem but they do mention one goal. They say that part of the problem arises from teleological thinking, or the idea that evolution is driven by a need for change.
If this single (albeit tenacious) misconception is affecting students’ ability to learn concepts throughout biology, instruction specifically designed to help students think critically about this sort of reasoning could have an impressive impact on student learning. Future research can explicitly focus on determining the pervasiveness of the idea that need is a rationale for change in biological systems and on effective strategies for changing this misconception to a scientifically accurate explanation.
You can't argue with that! Don't we all want an education system that teaches critical thinking?

My general impression from this paper is that the authors have identified a real problem but they seem to think it can be solved by developing better curricula and guidelines for teaching evolution. What they fail to emphasize is that most teachers of evolution probably wouldn't have done any better on the questions! If we want to improve evolution instruction then the first thing we need to do is teach the teachers—this includes, unfortunately, many professors whose main research focus is evolutionary biology.

[Image Credits: (top) Ask a Geneticist: The Tech Museum of Innovation, (bottom) John Hawks: Some genetic drift graphs with Mathematica]

Andrews, T., et al. (2012) "Biology Undergraduates’ Misconceptions about Genetic Drift." CBE-Life Sciences Education 11:248-259. [doi: 10.1187/cbe.11-12-0107]


  1. Isn't the phrase 'random genetic drift' redundant?

    1. No, I don't think so. Drifting isn't necessarily random.

      I like to include the word "random" for emphasis. Most people just refer to "drift" or "genetic drift" but I'm sticking with the old-fashioned terminology.

    2. The phrase "random genetic drift" was used by Crow and Kimura in their 1970 book, and it was used because "genetic drift" could be misleading to people who had been exposed to physics. In the physics literature "drift" is often used for nonrandom processes. So the "random" was added as a clarification. Here is an example of the usage of "drift" in physics.

      So 1% of students could define genetic drift correctly before any evolution course, and 12% could do it after such a course? This makes the case for them to be required to take 7 more evolution courses before they graduate.

  2. I don't remember what my early ideas about evolution were like, nor when they adapted.

    But for a long time, I've been thinking about evolution as "it's simply applied statistics". Even rudimentary knowledge of statistics then means that stuff like random genetic drift isn't particularly surprising. (Even though I can't really parse the definition in the sidebar.)

    I think the last time a mechanism in evolution really surprised me was the concept that viruses can transfer genetic material between otherwise independent species - and that was a long time ago, and even then, it was one of these "afterwards it's obvious" things.

    Of course, there are at least as many misunderstandings about statistics as there are about evolution ...

  3. I agree with khms: understanding evolution requires a rudimentary understanding of random processes. Such an understanding is taught at an early undergraduate stage to students in physics, engineering, and computer science (I'm not talking about the more detailed understanding taught to statistics majors). But bizarrely, biologists (even evolutionary biologists) don't seem to think the same concepts need to be taught to biology students. And then they complain when their students don't understand drift.

  4. Larry,
    I notice that you don't use the AB0 blood group system as an example of genetic drift any more. I also used that example for teaching, but thought about changing it in the upcoming course, because there has come some evidence for long-term balancing selection (Segurel et al. 2012 PNAS). Did you use eye color as a new example for the same reason, or is that a coincidence?

  5. Personally, I would favour starting with the neutral theory. The effect of sampling and resampling on any finite population, irrespective of selection, is a powerful, and strangely counterintuitive, notion. Selective differentials superimpose upon this background process, changing rates and skewing probabilities. While I would describe myself as still instinctively 'adaptationist' via high school/university biology and pop-science, understanding the role of randomness in the whole process is illuminating.

    1. This makes a strong case for a thorough grounding in statistics.

      Which would be a good addition to the sceptical tool kit in the battle against woo in any field.

      I'm just finishing reading "Bully for Brontosaurus" by Stephen Jay Gould and this is a theme he keeps revisiting, notably in his "The Streak of Streaks" article, on how poorly we are equipped (most likely by evolution) to understand random processes.

      As he put it, If we knew Lady Luck better, Las Vegas might still be a roadstop in the desert.

    2. Of course at the root of the problem is the teaching goal of having as many students (or slightly interested members of the general public) as possible understand at least a little bit about a given topic. Therefore, the simplest and easiest to grasp concepts are focused on. In the case of evolution, that would be selection, which most people should be able to grasp in an intuitive sense as a process that will be inevitable.

      This is not unique to the teaching of evolution. Many scientific concepts are approached in this fashion and lead to varying degrees of miscomprehesion.

  6. The importance of random genetic drift wasn't really covered in much detail during my undergrad Biology/Archaeology degree. Being a reader of this blog for the past 4 years has taught me a lot about the importance of genetic drift. As well you've provided me with all sorts of references that I've used to learn more about how evolution works. Now that I'm a graduate student studying Evolutionary Anthropology I'm surprised at the amount of fellow grad students who don't understand what genetic drift is (let alone how important it is). Anyway, thanks for filling in the holes in my education Larry!

  7. This comment has been removed by the author.

  8. Hi. I came across a while ago with a very interesting paper about the evolution of features that have neutral or deleterious intermediate steps by M. Lynch – “The Rate of Establishment of Complex Adaptations” ( That has expanded a little bit my basic-level knowledge about how evolution works. But a few days ago, a creationist started an internet discussion and threw this at me: “The Limits of Complex Adaptation: An Analysis Based on a Simple Model of Structured Bacterial Populations”, Douglas D. Axe ( I took a look at that and one thing I found was that he claimed that Lynch was wrong about the study I mentioned early and that he established a limit (six mutations) concerning the needed mutations to acquire a complex adaptation. His arguments are over my pay grade, but I think that if is right and Lynch is wrong he would go throug the peer-review process. If Someone here knows the papers, can you tell me why Axe is wrong and Lynch is right?

    1. Peer review in Biocomplexity, and most aspects of the "journal" Biocomplexity are a bit questionable.

    2. Yes, that's why i can't give credit to an article that is published in that journal. If Axe is right (and Lynch is wrong), Axe would have submited the paper for peer-review.

    3. I'd be wary of extrapolations from bacterial studies to the whole of life. Recombination in meiosis, a feature either present in or ancestral to anything most people would regard as 'complex', provides a whole new mode to the probabilities and tempo of multi-site evolution. And before that, endosymbiosis: whole-genome combinations. Both eukaryotic features, and both preceded by about 2 billion years of nothing much happening. Axe may be confirming what was already known - even given 4 billion years, bacteria don't get particularly complex by mutation, or their own modes of gene combination.

    4. Allan Miller says,

      Recombination in meiosis, a feature either present in or ancestral to anything most people would regard as 'complex', provides a whole new mode to the probabilities and tempo of multi-site evolution.

      I'm not exactly sure what you mean here. Recombination is common in bacteria and recombination in eukaryotes occurs all the time, not just during meiosis.

      What sort of "whole new mode" are you talking about? There's very little evidence to suggest that recombination affects either the probabilities or the tempo of evolution. That's why the advantages of sex are still a mystery.

    5. I know that under some assumptions a the results may be one thing and that from there people should explore things under other assumptions. But what makes me stumped is the fact that he says lynch is wrong. But if that was true I think Axe would have submited his paper to peer-review.

    6. What sort of "whole new mode" are you talking about? There's very little evidence to suggest that recombination affects either the probabilities or the tempo of evolution. That's why the advantages of sex are still a mystery.

      Generational homologous recombination between non-sister chromatids in sexual species generates a mechanism by which populations can effectively work upon elements of a ‘problem’, and different ‘problems’, independently, and combine the better solutions and purge genomes of the worst. It’s an analogue of distributed processing, and enables a set of organisms to explore ‘solution space’ much more effectively than as whole undivided genomes. That’s the ‘tempo’ part. The probability part derives from an extra set of chances for a beneficial A-B to arise - clonally, it can only happen serially. With meiosis, it can happen serially PLUS by combination.

      The effects can be clearly seen in GAs. Turn crossover off, and your populations take much longer and tend to get stuck on lower peaks. Turn it on, and ‘exploration’ of the space is much more rapid and wide-ranging.

      I’m surprised to see you talk in terms of the ‘advantages’ of sex. “Twofold cost” thinking dominates, but this is very specific to gendered organisms, which themselves are specific to multicellular eukaryotic forms (I'd avoid confusing unicellular mating types with gender, just as I would avoid confusing analogous bacterial ‘sexual’ mechanisms with eukaryote ones). As far as the early evolution of sex is concerned, in isogamous unicells, it could be close to neutral, a cyclic alternation of haploid and diploid states for non-recombinational reasons. Reciprocal recombination could be a later refinement, with as mild an immediate benefit as bivalent stabilisation or resolution of genetic conflict. Crossovers still resolve with exact 50/50 impartiality between recombinant and nonrecombinant products, and neither outcome makes much difference to the cell. The ‘purpose’ is not to access wider benefit, but a more immediate cellular need, which just happens to have the side-effects we note today – much greater variation, speed of adaptive response, purging of detrimental mutations and stabilisation of multicellularity through the haploid bottleneck, which remains largely ‘impartial’ for any given allele.

    7. @Allan Miller

      This isn't the place for a discussion of the selective advantages of recombination/sex. However, if you think that mixis confers a selective advantage on individuals in a sexual population then I suggest you read the classic paper by Joe Felsenstein.

      Felsenstein, J. (1988) Sex and the Evolution of Recombination, in The Evolution of Sex: An Examination of Current Ideas. R.E. Michod, and B.R. Levin eds.

      It's true that you can construct models where breaking linkage disequilibrium can be an advantage but the circumstances where this works are very special. It's not likely to confer a selective advantage on individuals under normal conditions.

      Keep in mind that the number of genes in linkage disequilibrium over the long term is quite small and the probability that both of them will have significantly beneficial alleles is even smaller.

      The fact that this works in GAs is irrelevant. GAs are not good models of biological evolution by any stretch of the imagination.

      Reciprocal recombination could be a later refinement, with as mild an immediate benefit as bivalent stabilisation or resolution of genetic conflict.

      Recombination is ancient and it almost certainly arose as a repair mechanism. All bacteria can carry out homologous recombination. So can all eukaryotic cells even when they are dividing by mitosis. (Division isn't necessary for recombination.)

    8. This isn't the place for a discussion of the selective advantages of recombination/sex.

      Fair enough, but you asked what I meant. You did misunderstand my intended clarification, however. I am emphatically NOT arguing that meiotic recombination arose because of its impact on linkage. The primary benefits I suggested were bivalent stabilisation and resolution of genetic conflicts between nonidentical partners in dividing a diploid phase (ie mitosis is not relevant).

      The fact that this works in GAs is irrelevant. GAs are not good models of biological evolution by any stretch of the imagination.

      I think Joe Felsenstein would beg to differ. I use the term in its most general sense, as any computer system that applies 'biological' processes to strings in a computational environment, and this includes models of evolution itself.

      Note finally that the recombination I am talking about is specifically the reciprocal exchange of non-sister chromosomes, and that is unique to diploid organisms and meiosis. I don't doubt that many of the components involved were co-opted from pre-existing enzymes involved in very ancient DNA management. The central novelty is Spo11 (itself descended from a topoisomerase), which initiates random DSB's which kick the repair pathways into action. The long-term consequences are dramatic and wide-ranging, but the cellular benefit purely local, and almost certainly nothing to do with linkage. Only eukaryotes cyclically recombine whole chromosomes, whatever similar things other organisms do. This process has a dynamic unique to this process, precisely because of its symmetry.

  9. A few years ago I taught an intro class in biology with about 10 teaching assistants. Several of the TA's had biology degrees, and some had taken their intro bio or genetics from big famous professors so there was a high degree of confidence among the staff in the level of understanding of simple, basic, concepts, so when an argument broke out about random genetic drift it got a bit heated.

    I went through all of the textbooks in my office that had glossary or sidebar definitions of "genetic drift" and assembled them into a handout. We went over them together and discovered that in many cases, if you started with a particular definition and then examined other definitions, there was conflict. These were current and widely used textbooks. It may have been the case that none of the definitions was technically wrong but they differentially made reference to key corollaries of genetic drift. They were not formal definitions, but explanatory definitions designed to be pedagogically palatable.

    There may be an intermediate area of confusion other than just the teachers and the students.

    1. Nobody said that key concepts in biology have to be easy! :-)

      I routinely survey my second year students right after they have taken a first year course on evolution. Year after year I get the same result. Over 90% of the students cannot spontaneously name genetic drift as a mechanism of evolution when asked. Over 80% DEFINE evolution as natural selection.

      Yes, it's true that there are several areas of confusion but the first step is to make sure that instructors in evolutionary biology are at least aware of the fundamental concepts they are supposed to teach! I'm glad you were.

    2. @LM: attended a teaching talk recently by candidate for faculty position. The candidate implied that the stability of double helix only owes to hydrogen bonding between bases and also described the central dogma on one of his slides in the usual incorrect way - all in one talk. Couldn't help but think of you sitting in the audience and choking on your coffee :)

  10. In working with teachers, in workshops on evolution, etc., I am often alarmed to hear teachers say things like "Evolution is really a simple concept. All you have to do to teach it is ..." then they have some sort of advice they throw in there.

    I think there are concepts in evolution, like random genetic drift and natural selection, that actually do appear quiet simple when first described, and it is easy to go to that place where people actually think it is simple. But it isn't. Maybe saying "it is simple" is a pedagogical technique for some, but it has costs.

    Also, in the US anyway, biology courses usually have a section on "evolution" which is labeled distinctly from population genetics and "dna" etc, and that is often taught as a mixture of palaentology and and natural selection with a bit of Darwin biography, rather than having evolution integrated throughout much of the course. One reason for this might be that evolution is a hot button issue. Having it all in one module or section allows one to swallow the bitter pill as quickly as possible, or even in some cases, skip past it or, worse, allow some students to do "alternative curriculum" for that period of time. For these reasons the "evolution" section is usually shoved into one week.

    Those are all issues about the evolution-creationism problem and not so much about teaching about random genetic drift. But random genetic drift is not really part of what students come to view as evolution if the related concepts aren't included in the "evolution" module, and the role of random genetic drift in evolution is lost.

    My own course has always been some form of "human evolution" so the concept of evolution is always there. I also spend a lot of time on emphasizing that the term "evolution" refers to a collection of related concepts with Natural Selection being only one of them. Within the context of natural selection, I like the idea of examining numerous case studies where the majority of the cases conclude that natural selection is either not the best explanation for the observed changes over time, or could be but can't be demonstrated conclusively. Also, my first lecture after introducing the course runs for a couple of sections and is entirely about falsehoods. My wife, who teaches high school AP biology also has an extensive falsehoods routine early on in the course.

    (Not too related to the discussion but I wanted to throw this in for the teachers in the room: If you follow a rule that all examples have to be real, and not horribly modified for convenience, then there will be a better result. There are a number of advantages to this including simply becoming a person who has more actual knowledge about real research under your belt. But it also allows you to have examples where the research is inconclusive about natural selection more organically, as it were. Made up examples that result in counterintuitive results tend to bother and enrage students. Real life examples that result in counterintuitive or ambiguous results promote critical thinking.)

  11. Larry, have you ever heard of Andreas Wagner? I'm currently reading his book "Life Finds a Way". Genetic Drift is a major theme. I think you'd probably enjoy reading (at least the first five chapters).

    Wagner is also the autor of "The Arrival of the Fittest".