Mike the Mad Biologist

I came across this excellent article by Jerry Coyne, which is part book review, part defense of natural selection. I recommend it highly. But, in reading the article, I wondered why people are so threatened by natural selection. Because that’s not the philosophically challenging part. Unless you’re a biblical ‘literalist’, the idea of a creator dude who acts through the mechanism of natural selection isn’t too theologically challenging. After all, traits that are beneficial (at least locally and in the short term) increase, while the deleterious ones decrease. Surely, this is the best of all possible worlds.

But what will bend most people’s noodle is the neutral theory, the implications of which are gloriously pessimistic.

The neutral theory was developed to make sense of molecular variation. Importantly, it asks a subtly different question about variation. Typically, when you encounter explanations of natural selection, it’s usually along the lines of: “There are a bunch of grey ones, and a bunch of brown ones. Under the right ecological circumstances, the grey ones replace the brown ones.”

Neutral theory starts from the (obvious) assumption that a novel allele (gene variant) initially occurs as a single copy in a population. So, if we consider a population of 1,000 diploid organisms (critters that, like us, have two copies of every gene), the initial frequency of that novel allele is one in two thousand. If the allele is neutral–does not confer a reproductive or survival advantage–it will become the dominant allele (‘sweep to fixation’) with a probability of one in two thousand (I’m leaving out some math). However, since it is rare, it will most likely go extinct due to genetic drift: the odds of that allele surviving more than a few generations are very slim (in fact, it has a 37% chance of disappearing by the next generation).

In this model, selection acts as a nudge that increases the probability that the now advantageous allele will not disappear. In the same population, if the allele has a two percent fitness advantage, which is quite large for most alleles in natural populations, the advantageous allele will likely disappear…over 96% of the time.

This is not Panglossian at all. But it gets worse. An allele with a weak negative effect on fitness–it lowers fitness–can also sweep to fixation. In the same population, an allele with a cost of 0.1% has a 0.4% chance of ‘beating out’ the ‘good’ allele. That doesn’t sound so bad, until we remember that humans have around 20,000 genes (as do many other eukaryotic organisms).

Oh dear.

Let’s throw in one more complication: effective population size. You’ll notice that I stipulated a population size of a thousand. Actually what I meant is the effective population size which considers that not all individuals reproduce, and that the breeding range of populations within a species can be quite small. For example, the effective population size of Finnish wolves is forty (even though the total number of wolves is around 250). To keep things simple, the smaller the population size, the stronger the advantage of a rare allele has to be to overcome removal by random chance (conversely, small population sizes also make it easier to fix ‘bad’ alleles). In other words, the fitness of an allele is contingent on those stupid fucking natural history facts. If volcanic eruptions subdivide a species, for instance, the likelihood that an adaptive allele will become predominant decreases–and has nothing to do with genetics.

So let’s move to the long arc of evolution. Unless selection is really strong, it will take a long time to purge a population of the less fit allele. In the case of an allele with a two percent advantage in a population with an effective size of 1,000, it will take around 1,500 generations to eliminate the less fit allele. That’s a long time for crap to hang around.

Let’s review:

1) More fit alleles have a high chance of going extinct.

2) On occasion, less fit alleles will replace more fit alleles.

3) Whether an allele is more fit in part depends on the population size. The fitness of an allele is contingent on those stupid fucking natural history facts.

4) It takes a really long time to purge less fit alleles. Consequently, until Glorious Equilibrium Day Draweth Nigh, populations are full of a lot of genetic flotsam.

But natural selection is supposedly the theologically challenging part?

Nothing I’ve described here is news to evolutionary biologists or geneticists: this is taught in introductory evolution courses. Yet most people are completely unaware of the neutral theory, even though it was a raging debate in population genetics for decades. Here are the potential explanations I came up with for its ‘popular’ absence:

1) Math is hard and icky[/snark]. Probably not good fodder for a newspaper article. On the other hand, this doesn’t explain why it’s not discussed in books.

2) Math is hard and icky, part deux. Maybe science popularizers don’t understand it very well?

3) Perhaps there’s a concern that explaining natural selection is hard enough, and that neutral theory would simply confuse the issue.

4) The scientists who popularize science fall into the ‘selectionist’ camp, and they’re not interested in writing about this.

5) This is a population genetics approach to evolution, and that is underrepresented in popular treatments of evolution.

Other ideas? And how do we make it accessible to non-scientists? Cuz it’s kinda important.



  1. #1 Scented Nectar
    April 29, 2010

    ‘And how do we make it accessible to non-scientists?’
    I think you just did! :) I’m not a scientist, but found the article interesting and have a new aspect of this branch of science to think about. Thanks.

  2. #2 darwinsdog
    April 29, 2010

    And how do we make it accessible to non-scientists? Cuz it’s kinda important.

    Why is it our responsibility to make pop gen accessible to non-scientists? I had to take a pop gen course in grad school, in which the topics of effective population size, selective sweeps, the number of generations it takes for an allele to go to fixation given a certain selective advantage & population size, etc., were all discussed with mathematical rigor. Anyone who is interested in or needs to know this information can likewise take a course. Unless I am getting paid to teach pop gen, I feel no obligation to make these concepts and the math behind them, accessible to the public. The public can remain ignorant for all I care, unless they care enough to enroll in a course and learn what they want or need to know for themselves.

  3. #3 xebecs
    April 29, 2010

    I’m not sure I get your point. Advantageous alleles will usually win out, but usually is not always. Why is this philosophically difficult?

  4. #4 Coriolis
    April 29, 2010

    I also don’t understand your point.

    Advantageous alleles have a higher chance to win out, and disadvantageous alleles have a lower chance to win out, but the difference in these chances is not that big so the good guys don’t always win – is that the problem?

    This is a general problem of human perception of chances – I remember a recent Sid Meier talk on the civilization games where he explained that players considered that if the odds of a battle were 3 to 1 in their favor they should always win – and thought that the computer was cheating if it won such a battle. Of course if the odds of the battle were 3 to 1 against them they had no problem with winning. I don’t think there’s a solution to this perception problem, apart from not being stupid :).

    I don’t know why you’d call this “neutral theory” though unless I’m misunderstanding your point…

  5. #5 Joshua
    April 29, 2010

    Looking back on pop sci books I’ve read about evolution, I think I’m going to pick #5 as the most likely explanation. Most of the ones I have personal familiarity with have focussed on either paleontology or development or, to generalise, issues of evidence for common descent. There may be pop sci books about population genetics or the more general issue of the mechanisms of evolution that I’m not aware of, I guess, and by my hypothesis I’d expect them to go into depth about neutral theory and the selectionist debates.

  6. #6 Tantalus Prime
    April 29, 2010

    I go with option 4. People like to describe natural selection in terms of what does happen, and neutral theory explains things in terms of what doesn’t happen. And that isn’t nearly as sexy.

    Plus, any creationist can tell you that mutations are universally harmful, and therefore any suggestion that mutations are for the most part neutral is clearly insane.

  7. #7 Scented Nectar
    April 29, 2010

    I find this interesting because it shows the greyness in what is sometimes not considered to be a grey area. There is the one extreme view against evolution that sees the fittest surviving, and therefore always surviving, and therefore meaning some sort of creator-intent so it can’t be natural, and the other view that thinks evolution means that everything is random so selection doesn’t make sense to them and can’t be true. Those extremes get argued against as strawmen.
    The neutral concept shows how the gradual most of the time fittest traits surviving, and also shows that it coexists with a few of the not-so-fit traits surviving. It can’t be knocked down as a strawman because it’s true.

  8. #8 Josh
    April 29, 2010

    The weird thing is that neutral theory doesn’t actually fit the data. The even weirder thing is that while numerous papers have shown that most segregating variants are at least slightly deleterious, if you look in the right place (e.g. Drosophila) most of the fixed differences between sister species seem to be driven by positive selection.

  9. #9 Daniel
    April 29, 2010

    I think this idea (which is essentially “contingency”) has been “popularized” ad nauseum (e.g. Gould’s “Wonderful Life” on the Burgess Shale). I think alot of people just find it less interesting – and there tends to be not a whole lot to say about it.

    A prominent Drosophila developmental biologist recently and flippantly told me that evodevo is “just molecular natural history and not real science.” …Implying that the stochastic and contingent occurences that are often the true origins of our alleles, traits, and existence simply aren’t that interesting and no one in their right mind should study it. The bastard.

  10. #10 razib
    April 29, 2010

    i think part of the issue is that neutral forms are most dominant at the molecular level. i bet in terms of functional salient phenotypes which people are interested in in the public selective dynamics, whether positive selection, or constraint, stabilizing selection, etc., are more relevant. and as others have noted, gould’s spandrel stuff takes care of that at the coarse phenotypic/natural historic scale.

  11. #11 Janne
    April 29, 2010

    This description is a little disingenious in part. You take an example with a tiny population of a macroscopic apex predator, when almost all of evolution happened – and happens – with single-cell organisms and small, numerous species.

    If you take a bacterial colony the local population is easily in the millions, not a thousand (or forty), and 1500 generations is what, a month of time or so. Plenty fast enough to track many environmental changes. In fact, that may be a nice illustration as to why bacteria and other small organisms are so dominant in the number of species, in total mass and in range of lifestyles.

  12. #12 noddin0ff
    April 30, 2010

    Question: What’s the frequency that we correctly identify a trait as advantageous or disadvantageous assuming that we never really fully comprehend the complexity of biological interactions enough to unambiguously assess fitness?

  13. #13 Mokele
    April 30, 2010

    It’s pretty simple, actually, and well summed up in a quote I heard and have often repeated:

    “Natural & sexual selection aren’t the only parts of evolution, but they’re the only parts that are interesting.”

    That’s somewhat tongue-in-cheek, but it’s also very true of the public – they’d much rather hear about form-function relationships that lead to awesome/cool behaviors and abilities than similarly important but less attention-grabbing stuff about gene frequencies.

  14. #14 Bob
    April 30, 2010

    Aren’t you really discussing how genetic drift can often be a more powerful influence on allele frequencies than selection, particularly in small populations? Hence, a harmful allele can be fixed or an advantageous allele can be lost.

  15. #15 Min
    April 30, 2010

    “And how do we make it accessible to non-scientists?”

    Computer graphics.

    No, really. At 10 generations per second you get 1500 generations in 2 1/2 minutes.

    Seeing is believing.

  16. #16 QQQQ
    April 30, 2010

    I think the difference between the ‘Classical’ theory and neutrality theory is as follows.

    In the classical theory we let the population size N go to infinity, therefore the law of large numbers applies (in some sense) and the distribution of genes in the population is deterministic and follows some differential equation.

    In neutrality theory, N is finite so the distribution of genes in the population follows a Markov process, and when we observe a given population we are sampling from that Markov process. In particular, any allele will appear with proportion X, with some probability P(X).

    If what I described isn’t neutrality theory, then that’s what neutrality theory should be.

  17. #17 BaldApe
    May 1, 2010

    In one of my graduate courses we had to figure out, generation by generation, the fate of an allele which was always fatal in the homozygote, but harmless in the heterozygote. It was a real eye opener.

    Once it got rare, it hangs around a really, really long time. In fact, my impression was that since it is invisible to selection at low frequency, selection just can’t get rid of it.

    On the bright side, it’s an excellent argument against eugenics.

  18. #18 hoary puccoon
    May 5, 2010

    Markov process? What’s a Markov process?

    People who are out there in the trenches dealing with the public’s doubts about evolution are having to explain that fatal mutations aren’t passed down to offspring.

    Go look at the comments to the same Jerry Coyne review on Evolution Blog. A guy is seriously arguing about whether pigs could grow wings. And you think you could straighten him out by saying, “oh, it’s a Markov process.”?

  19. #19 Bobington
    June 13, 2010

    We had a huge argument the other day about whether neutral theory contradicts Dawkins’ claim that Evolution isn’t random.

  20. #20 Hypocee
    June 13, 2010

    Sooo…selection still takes place, but is some linear factor slower than it would otherwise be. So? Why is this cause to cry conspiracy in 4?

  21. #21 ScottFree
    June 14, 2010

    This whole post assumes a static view of dis/advantageous. In the real world the environment changes, predator and prey change and populations migrate. Total elimination of an allele is only a good thing in very simple and static situations. It is not surprising that evolution evolved the way it did given the complexity in the real world.