Thursday, February 01, 2018

Sex isn't as beneficial as you might think

One of the most interesting topics in my molecular evolution class was the discussion over the importance of sex. Most students seem to think the problem is solved. They were taught that sex increases variation in a population and this gives sexual populations an evolutionary advantage. The fact that sex (recombination) breaks up as many linkages as it creates makes the explanation much less viable. The fact that there's very little evidence to support the claim comes as quite a surprise to my students. Sex is still one of the greatest mysterious in evolutionary biology [What did Joe Felsenstein say about sex?] [Everything you thought you knew about sex is probably wrong].

It is worth noting that Maynard Smith's argument invalidates the earliest genetic argument for the evolution of recombination, that advanced by East (1918). That argument is also the one commonly found in textbooks, which tend to be a bit out of date (in this case, by over 50 years). East argued that recombination creates new genotypes. So it does. An AB/ab parent will have among its gametes not only the two types that formed it, AB and ab, but also Ab and aB if there is recombination between the two loci. But if the population is in linkage equilibrium, then somewhere else an Ab/aB parent will be undergoing recombination, which will remove Ab and aB gametes and replace them by AB and ab. These two processes will exactly cancel each other if the two types of double heterozygote, coupling (AB/ab) and repulsion (Ab/aB) are equally frequent. This will happen precisely when the population is in linkage equilibrium. In that case no new genotypes arise by recombination....

We have that anomalous situation that a detailed population genetic analysis analysis reveals not only that the standard explanation for the evolution of recombination will not work, but also that there is a good evolutionary reason for believing that modifiers will be selected to eliminate recombination. [my emphasis LAM]

                        Joe Felsenstein (1988)
Richard Lenski decided to test the benefits of sex. He took the twelve cultures from the long-term evolution experiment and added sex to see if they would adapt faster (Maddamsetti and Lenski, 2018). Specifically, he added Escherichia coli strain K12 Hfr (high frequency recombination) to the cultures. This strain promotes conjugation and the exchange of genes.

The results are described in the abstract.
Maddamsetti, R., and Lenski, R. E. (2017) Analysis of bacterial genomes from an evolution experiment with horizontal gene transfer shows that recombination can sometimes overwhelm selection. PLOS Genetics, January 31, 2018. [doi: 10.1371/journal.pgen.1007199]


Few experimental studies have examined the role that sexual recombination plays in bacterial evolution, including the effects of horizontal gene transfer on genome structure. To address this limitation, we analyzed genomes from an experiment in which Escherichia coli K-12 Hfr (high frequency recombination) donors were periodically introduced into 12 evolving populations of E. coli B and allowed to conjugate repeatedly over the course of 1000 generations. Previous analyses of the evolved strains from this experiment showed that recombination did not accelerate adaptation, despite increasing genetic variation relative to asexual controls. However, the resolution in that previous work was limited to only a few genetic markers. We sought to clarify and understand these puzzling results by sequencing complete genomes from each population. The effects of recombination were highly variable: one lineage was mostly derived from the donors, while another acquired almost no donor DNA. In most lineages, some regions showed repeated introgression and others almost none. Regions with high introgression tended to be near the donors’ origin of transfer sites. To determine whether introgressed alleles imposed a genetic load, we extended the experiment for 200 generations without recombination and sequenced whole-population samples. Beneficial alleles in the recipient populations were occasionally driven extinct by maladaptive donor-derived alleles. On balance, our analyses indicate that the plasmid-mediated recombination was sufficiently frequent to drive donor alleles to fixation without providing much, if any, selective advantage.

Author summary

Bacteria often transfer genes encoding antibiotic resistance as well as other important traits, but the extent of intergenomic recombination—in effect, sex—is highly variable across bacterial species. Why? A better understanding of how and why bacteria exchange genes would help people combat the spread of infectious disease as well as shed light on the evolutionary origins of sex. Here, we sequenced genomes from an evolution experiment with Escherichia coli in which recombination was extensive but, unexpectedly, did not speed up the rate of adaptation. In this experiment, the effective rate of chromosomal recombination was much higher than previously inferred for natural E. coli populations. In fact, the rate was so high that introduced genes sometimes drove established beneficial alleles to extinction in the experimental populations. In effect, genes that were physically linked to the genes causing recombination had a strong transmission advantage, whether or not they provided any selective advantage to the recipient cells.

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. 74-86. [PDF]


  1. "In effect, genes that were physically linked to the genes causing recombination had a strong transmission advantage, whether or not they provided any selective advantage to the recipient cells."

    That almost sounds like sex is a selfish act of the alleles that recombine. As in sex is a sort of genetic parasite that can take over a host and even a population.

    Sex is itself a sexually transmitted disease. :)

    1. Isn’t it a confusion of issues, when the horizontal transfer of F-plasmids (and DNA from the bacterial chromosome hitch-hiking with them) is lumped together with meiotic recombination. Where meiosis involves a reduction in the replication-rate of genes, an F-plasmid that carries other bacterial DNA along is a case of boosting that DNA’s replication-rate. Can this boost in the replicator-function of that DNA not explain why it can sometimes fix despite being disadvantageous in its interactor-function? In fair meiosis, no gene gets such a boost in its replicator-function. On the contrary, the paradox of sex has been conceived by George C. Williams as the cost of meiosis for that reason. I think I'll post another historical comment below the ones on Maynard Smith pointing out how they cut the cake differently.


    This hypothesis dates back at least as early as Montgomery (1901),and has been eloquently argued for three decades by Carol and Harris Bernstein and colleagues (1). Their latest 2018 paper (2) is a gem!

    1. Bernstein C, Bernstein H. 1991. Aging, Sex and DNA Repair. San Diego: Academic Press.

    2. Bernstein H, Bernstein C, Michod RE. 2018. Sex in microbial pathogens. Infection, Genetics & Evolution 57: 8–25.

  3. Odd. Two comments in a row disappeared before posting.

  4. ... and in Drosophila males, how do they correct double-strand breaks without crossing-over?

  5. Is gene transfer the essence of sex? Or is it crossing over? If I understand correctly, crossing over tends to limit differences (which would be the Bernstein' error correction hypothesis in a different formulation.)

    As to the reproductive advantage of sex, looking at single-celled life, the relative biomass of asexual vs. non-sexual single celled organisms provide a measure? If you don't count all forms of gene transfer as "sex," that would seem to break down to prokaryotes and eukaryotes...and prokaryotes would seem to have the advantage?

    Does the apparent trend in plant evolution for the decrease in the sexual phases in alternation of generation (to the point where the gametophyte is reduced to an organ,) suggest an advantage to the asexual phase?

    Metazoans seem to have a much greater biomass than single-celled eukaryotes suggesting that it's not sex per se that is advantageous, but multicellularity. Could the fundamental advantage of sex be its use in achieving that?

    1. Multicellularity probably requires bottlenecking the population of cells, but that doesn't need to entail recombination (after all there are clonal multicellular organisms). But if you have multicellular organisms, then the only time you could reasonably have recombination is during that bottleneck.

    2. A small founder population in a previously colonial species would ensure sufficient genetic similarity. (I'm guessing that what "bottlenecking" means.) But I don't understand how such a small population could have much advantage in microbial ecology over other colonies.

      Recombination to ensure sufficient compatibility as portions of a previously colonial population begin to differentiate, so that the new specializations can still be in the newly multicellular organism's genome seems more likely?

    3. By bottlenecking I just mean that if you had a large number of cells to start with for offspring, then there's be a greater risk of cell level selection taking over. If you started each human with 10^5 cells, chances are we'd all have cancer before hitting our teens. Going to a single zygote means that the risk of it containing genes that are more fit on a cellular level, but less fit on an organismal level is minimized.

      The main advantage of recombination is that it reduces and eventually eliminates linkage.

      If you have multicellular organisms, you can only get recombination if you go to a single gamete.


      Another log on the fire.

  6. At it's core, the existence of sex tells us that two distinct individuals may bring more to the proverbial table than slightly different flavors of a shared genome. So freeing our self of the genetic variation group think, how about considering epigenetic variation as well? There is certainly an ever-growing body of evidence to suggest that epigenetic modifications can be a) environmentally acquired and b) vertically inherited. Perhaps such a concept is too Lamarckian for the liking of many

    1. For how many generations are epigenetic modifications inherited? Last I heard no more than about 3 generations. This is the point that is ignored when people wave their hands about and want to explain the differences between species as "epigenetic".

    2. @Joe Felsenstein There's this paper: "Transgenerational transmission of environmental information in C. elegans - Klosin et al" where epigenetic changes were inherited for 14 generations.

      But that doesn't really matter though. Protein evolution and the evolution of regulatory sequences are well documented as far as I am aware.

    3. Doesn't epigenetics change the outcomes of gene selection?
      It's not clear to me that three generations is entirely negligible. Though how important epigenetic changes are would be proportional to the importance of gene selection, I suppose.

    4. @Joe Felsenstein The argument of persistence ignores any effects of environmental reinforcement may have. If parental exposure can somehow (...and I realize that the "somehow" is not trivial) induce germline changes, this effect could be expected to persist as long as the generative environmental conditions persisted. As far as sex, giving both two parents the opportunity to respond (epigenetically) to environmental conditions independently could add one more avenue by which the F1s could increase fitness. More long term genome evolution could then result as those epigenetically primed loci have constitutively shifted exposure to natural selection.

    5. @Joe

      Speaking as a high school teacher participating on Biology Teacher forums - I have to express my frustration how instance of DNA Methylation is immediately identified by my peers as “epigenetic” and significant in evolutionary terms as “memory” passed from one generation to the next.

      My protests generally fall on deaf ears – which is what motivated me to assemble a collection of “experts” to offer gentle correction and guidance to Biology teachers on a closed site to avoid the interference of trolls and creationists… but I digress

      Parenthetically – that was a shameless plug for my idea hoping to elicit response from the many experts present.

      Back to Joe’s statement above:

      JF:For how many generations are epigenetic modifications inherited? Last I heard no more than about 3 generations. This is the point that is ignored when people wave their hands about and want to explain the differences between species as "epigenetic”.

      Primum non nocere – for the sake of my students - please correct me if I am wrong

      I ask my students to consider two possible default settings

      1 – “Frugal” metabolic setting

      advantage: maximizes efficiency of nutritional uptake during “famine” times
      disadvantage: maximized efficiency is detrimental during “feast” times leading to metabolic diseases such as diabetes

      2 – “Spendthrift” metabolic setting

      Advantage: reduced efficiency of nutritional uptake during “feast” times is less likely to lead to metabolic diseases such as diabetes
      Disadvantage: reduced efficiency of nutritional uptake is detrimental during “fast” times

      I then ask my students to imagine a “sticky toggle switch” of gene regulation. Offspring tend to be born in identical ongoing environmental circumstances as their parents. So, if the parents’ “toggle switch” has switched to “frugal” as a sustained response to famine; then the offspring would have a selective advantage if their default setting was also immediately “fugal” at birth.

      Environmental changes can be short-lived and any setting change of the “toggle-switch” would require more than one generation to reset the default setting.

      In other words – the epigenetic “memory response” is not significant over the multigenerational timeframe required for allele fixation, for example. What I am suggesting is that any putative “epigenetic memory response” to environmental change is, in of itself, a trait subject to selection.

      There are scenarios where species which have such a “sticky toggle switch” would have selective advantage over species which do not. Of course, there are also scenarios where the converse is true.

      Am I being overly naïve here?

      Should I stop telling my students this tale?

      Thanks in advance to any and all…

    6. @tom: In other words – the epigenetic “memory response” is not significant over the multigenerational timeframe required for allele fixation, for example. What I am suggesting is that any putative “epigenetic memory response” to environmental change is, in of itself, a trait subject to selection.

      Yes, that could happen. However, note that transgenerational epigenetic responses seen in humans tend to not be adaptive. The effect on your children of your exposure to a famine is not to make them more resistant to famines, but (say) to increase your probability of heart disease.

      People too-easily assume that most epigenetic responses are adaptive. And they fail to notice that they don't last for more than a few generations (maybe more in Caenorhabditis).

    7. Joe

      Perhaps, my example was too naïvely stated.

      I am thinking of some of the Swedish Överkalix data. I agree that the interval to reset the “sticky toggle switch” would prove immediately disadvantageous to the first generations no longer experiencing famine.

      I still wonder out loud whether the putative epigenetic memory response can still be regarded as adaptive in overall terms, even though it appears maladaptive for that brief interval.

      If I am understanding correctly – species selection would need to be occurring.

    8. Tom, did you start from the assumption of something like the following? If a locus for the persistence-time of methylation-patterns existed and alleles for, say, no persistence, three generations, and five generations occurred, the fact that most methylation patterns in a certain species persisted three generations would suggest this to be a selected outcome.
      Also, if heart disease struck at old age, then it might not matter for selection, if famines were frequently occurring, while old age-parenthood was very rare.

    9. Hi Joachim

      I am saying that differential methylation as a result of epigenetic memory which can persist say three generations - is itself a trait which can be adaptive in some scenarios and maladaptive in others.

      The epigenetic trait could remain adaptive even when the interval to reset the “sticky toggle switch” would prove immediately disadvantageous to the first generations no longer experiencing famine. Overall - that epigentic trait would prove adaptive in the grand scheme of things, but only under certain scenarios.

      FTR - I no longer believe that one one gene = one locus aka "one functional transcript". I believe Larry and Doolittle also part paths on this point as well, if I understand both correctly.

    10. Tom, me neither concerning bean-bag genetics. I only used that lingo for my lack of sophistication in genetics and as a short-hand for expressing the idea about some genetic variability for a trait being present in some form (not necessarily one coherent locus etc.).

  7. Cilliates must undergo meiotic processes or the lineage “ages” and dies. The classic is the zoology 101 conjugation of Paramecium. But many cilliates can also undergo autogamy (self fertilization) which rejuvinates the strain.

  8. @Dennis: Thanks for the information about C. elegans. As for protein evolution and regulatory sequence evolution, yes, an offspring might have "your beauty and my brains", but, as George Bernard Shaw once noted, it can also have "my beauty and your brains".

    It is less than obvious that producing AB gametes from an Ab/aB parent by recombination offsets producing also aB and Ab gametes from an AB/ab parent by recombination. For an explanation of the issue which I like, see the paper cited at the end of Larry's post above. However the page with the PDF linked to there is gone, you can also find it here:

    1. Damn it, which Reply button do I push to get my reply to Dennis to follow Dennis's most recent comment above? The Reply in David's comment put the replies here.

  9. Here's an interesting counter-example from eukaryotes (it's cited in Lenkski's paper) that does provide evidence that sex can improve the efficiency of adaptation:

    Sex speeds adaptation by altering the dynamics of molecular evolution

    "Sex and recombination are pervasive throughout nature despite their substantial costs1. Understanding the evolutionary forces that maintain these phenomena is a central challenge in biology. One longstanding hypothesis argues that sex is beneficial because recombination speeds adaptation. Theory has proposed several distinct population genetic mechanisms that could underlie this advantage. For example, sex can promote the fixation of beneficial mutations either by alleviating interference competition (the Fisher–Muller effect) or by separating them from deleterious load (the ruby in the rubbish effect). Previous experiments confirm that sex can increase the rate of adaptation, but these studies did not observe the evolutionary dynamics that drive this effect at the genomic level. Here we present the first, to our knowledge, comparison between the sequence-level dynamics of adaptation in experimental sexual and asexual Saccharomyces cerevisiae populations, which allows us to identify the specific mechanisms by which sex speeds adaptation. We find that sex alters the molecular signatures of evolution by changing the spectrum of mutations that fix, and confirm theoretical predictions that it does so by alleviating clonal interference. We also show that substantially deleterious mutations hitchhike to fixation in adapting asexual populations. In contrast, recombination prevents such mutations from fixing. Our results demonstrate that sex both speeds adaptation and alters its molecular signature by allowing natural selection to more efficiently sort beneficial from deleterious mutations."

    1. There are other examples that purport to show the benefits of sex. There are also many other papers that purport to show no benefit from sex.

      This is why the problem hasn't been solved. That's point I emphasized in class. If you think there's a single correct answer to why sex is important then you are wrong.

    2. Just to note that the statements by Rich Lenski that Dave Carlson quotes are supporting the well-known Fisher-Muller and Muller's Ratchet argument(s) for the evolution of "sex". They are not putting forward Yet Another Theory.

  10. Of course. Note that in my comment I never suggested that the problem was solved or that there was a single correct answer. That said, there seems to be some good evidence that increased efficiency of selection and alleviation of Muller's Ratchet are likely important contributors to the maintenance of sex and recombination.

    1. I see that I somehow managed to misplace my previous comment. It should be a reply to Larry's comment on the abstract I quoted.

    2. @Dave Carlson: That's OK, we all seem to do that. Some day someone will explain to us which Reply button to push.

  11. The bible says God created male and female. So reproduction simply was to be this way. why not? what's a better way? the benifit of sex is its simple and hardly without thought or intent.
    Saying sex evolved would be hard to figure out as to why!

    1. At the beginning of the world there was only darkness, void. Creation began when the dual Ometecuhtli (Lord of Duality) / Omecihuatl (Lady of Duality) created itself. This first god was good and bad, male and female, and gave birth to four other gods: Huizilopochtli, Quetzalcoatl, Tezcatlipoca and Xipe Totec . These gods created the world.

      The first things created by Quetzalcoatl and Huitzilopochtli were fire and a half sun. They then undertook the creation of humanity by sacrificing a god whose blood drops on a mass of ground-up bones produced the first man and woman, named Oxomoco and Cipactonal respectively. The birth of each took 4 days.

      Why not?

  12. I think John Maynard Smith conceived the maintenance of recombination as a problem of within-population selection between alleles that increase and others that decrease recombination rates. He accepted Williams's criticism of group-selection arguments for this issue (what he called the "balance argument" of Williams). He agreed that this problem requires an immediate individual-level explanation. But he also maintained that the competition between a sexual population and a genetically isolated asexual clone is a case of between-population selection. At this level, he did allow for long-term or group selection to play some (limited) role.

    Anyway, his distinction begins in the preface already:

    "I am under no illusion that I have solved all the problems which I raise. Indeed, on the most fundamental question - the nature of the forces responsible for the maintenance of sexual reproduction and genetic recombination - my mind is not made up. On sex, the relative importance of group and individual selection is not easy to decide. On recombination, group selection can hardly play a significant role, but it is not clear to me whether the short-term selective forces I discuss are sufficient to account for the facts, or whether models of a qualitatively different kind are needed." (Maynard Smith 1978, p. ix)

    "It may help to classify the various theories; first, according to the time scale on which selection is supposed to act, and then according to the 'unit of selection' - population, individual, or gene." (Maynard Smith 1978, p. 1)

    "I do not find it possible to give an unequivocal answer concerning the role of group selection in the maintenance of sexual reproduction. It has played some role, as evidenced by the taxonomic distribution of parthenogens; but it is not the only relevant force, as will be apparent from the review of the balance argument in Chapter 4, section E. But, whatever one may think of the role of group selection in the maintenance of sex, it cannot explain how it started, and it cannot explain the maintenance of high levels of genetic recombination within sexual populations." (Maynard Smith 1978, p. 6)

    And so throughout the book. Maynard Smith consistently distinguishes the maintenance of sexual reproduction from that of recombination, the former being an issue of selection between isolated populations and clones, the latter being one of selection between alleles within one population.

    Maynard Smith's subject-to-provision support for "some role" of long-term or group selection in the maintenance of sex (not recombination) was also defending his earlier publication from 1958 (The Theory of Evolution, Penguin Books, pp. 138-139). It is often forgotten in potted histories about the paradox of sex, that Maynard Smith did already clearly state the cost of males in this early pop-science writing and also embraced the long-term group-selection explanation of the maintenance of sex. See Dagg (2016, DOI: 10.3998/ptb.6959004.0008.003) for the relevant quotes.

    By the way, Ghiselin (1988, p. 16, same book as the Felsenstein-1988 quote above), reminisced an instance of Williams reviewing one of his papers and telling him about the twofold cost of sex and that he [Williams] had found it in a book by Maynard Smith (1966), which must have been the second edition of the above quoted Penguin book by Maynard Smith (see also Dagg 2016).

    1. The fact that John Maynard Smith never changed his mind about his hedged support for some role of group (between-population) selection in the maintenance of sex is clear from an interview of Richard Dawkins with John Maynard Smith in 1997 (deposited at the Web of Stories in 2008,

    2. As promised above (see comment below Mikkel Rumraket Rasmussen's), my take on Williams's take on the maintenance of sex:

      George C. Williams conceptualized the maintenance of sexual reproduction as a problem of selection within one species. He began by considering organisms with complex life-cycles that include both sexual and asexual modes of reproduction, for example, aphids, rotifers, strawberries, corals. He conceived the cost of sex as the cost of meiosis, that is, the cost of reducing the relatedness with the own offspring from r = 1 to 0.5, when these organisms meet the time or conditions for switching from asexual to sexual reproduction. This kin-selection conception takes the maintenance of sexual reproduction as a problem of selection within a population. Herein, he disagreed with Maynard Smith:

      “I think that the primary disadvantage of sexual reproduction in relation to asexual is most fruitfully formulated as a paradox of kin selection—an organism devotes resources to the production and care of a more distant (r = 0:5) rather than a close (r = 1) relative. This formulation provides a number of advantages. In its focus on genes identical by descent, kin selection is genetically explicit and relates directly to evolution. Maynard Smith’s economic argument (resources wasted on males) makes it easy to overlook the fundamental distinction between (1) the evolutionary problem of sexual and asexual reproduction as alternative character states in a population, and (2) the purely ecological question of competition between a clone and a Mendelian population.” (Williams 1978, ‘A review of The Evolution of Sex by John Maynard Smith.’ Quarterly Review of Biology
      53: 287–289. Page 298)

      “I believe that understanding has been hampered by failure to distinguish the ecological from the evolutionary problem of sexuality. In important ways, insights gained from conceptual or experimental comparisons of sexual populations and competing clones (the ecological problem) may mislead in relation to sexual and clonal reproduction as alternative processes in a population (the evolutionary question with which I am concerned here).” (Williams 1980, ‘Kin selection and the paradox of sexuality.’ In Sociobiology: Beyond nature/nurture? Ed. by GW Barlow and J Silverberg. Boulder, CO: Westview: 371–384. Page 372)

      The fact that William and Maynard Smith cut the cake differently gets obvious from the way in which Williams treated the maintenance of recombination as not the problem he was at all concerned with:

      “I assume that observed chromosome numbers and crossover rates reflect the optimum compromise between maximizing whatever benefits there are in recombination, and minimizing recombinational load. Tighter linkage must reduce recombinational load, but it does nothing to alleviate o alleviate the cost of meiosis.” (Williams 1975, Sex and Evolution, Princeton Univ. Press, p. 108)

      That is, reducing replication-rate by fusing gametes is not alleviated by assuming, for example, a species with a genome consisting of one homologous pair of a giant chromosome and no crossing-over between this homologous pair. That would exclude recombination through segregating heterologous chromosomes as well as through crossing over between homologous chromosomes, but it would not pay the cost of reducing r from 1 to 0.5, or the cost of males, or the cost of fusing gametes, or whatever you conceive the cost of sex to be.

  13. There is now a very serious global invasion of crayfish clones that have been traced to a ~30 years ago pet store aquarium speciation event:

    This Mutant Crayfish Clones Itself, and It’s Taking Over Europe