The intelligent homosexual’s guide to natural selection and evolution, with a key to many complicating factors

San Francisco Pride, 2008. Photo by ingridtaylar.

This is a cross-posting of my latest contribution to the Scientific American guest blog. Since the original went up at SciAm, P.Z. Myers has pointed out a few more complicating factors. If you read one paper to follow up on what I’ve written here, I’d suggest Nathan Bailey and Marlene Zuk’s excellent 2009 review [PDF], which is posted in PDF format by none other than The Stranger.

ResearchBlogging.orgJune is Pride Month in the United States, and in communities across the country, lesbian, gay, bisexual, and transgendered Americans are celebrating with carnivals, parades, and marches. Pride is a rebuke to the shame and marginalization many LGBT people face growing up, and a celebration of the freedoms we’ve won since the days when our sexual orientations were considered psychological diseases and grounds for harrassment and arrest. It’s also a chance to acknowledge how far we still have to go, and to organize our efforts for a better future.

And, of course, it’s a great big party.

I’m looking forward to celebrating Pride for the first time in my new hometown of Minneapolis this weekend–but as an evolutionary biologist, I suspect I have a perspective on the life and history of sexual minorities that many of my fellow partiers don’t. In spite of the progress that LGBT folks have made, and seem likely to continue to make, towards legal equality, there’s a popular perception that we can never really achieve biological equality. This is because same-sex sexual activity is inherently not reproductive sex. To put it baldly, as the idea is usually expressed, natural selection should be against men who want to have sex with other men–because we aren’t interested in the kind of sex that makes babies. An oft-cited estimate from 1981 is that gay men have about 80 percent fewer children than straight men.

Focusing on the selective benefit or detriment associated with particular human traits and behaviors gets my scientific dander up, because it’s so easy for the discussion to slip from what is “selectively beneficial” to what is “right.” A superficial understanding of what natural selection favors or doesn’t favor is a horrible standard for making moral judgements. A man could leave behind a lot of children by being a thief, a rapist, and a murderer–but only a sociopath would consider that such behavior was justified by high reproductive fitness.

And yet, as an evolutionary biologist, I have to admit that my sexual orientation is a puzzle.

There’s reasonably good evidence for a genetic basis to human sexual orientation–although the search [$a] for a specific “gay gene” [$a] has had mixed results [$a]. Gene variants, or alleles, associated with an 80 percent decrease in reproductive fitness should be naturally selected out of the population pretty quickly. So why aren’t all humans heterosexual?

Straight people are in the overwhelming majority, but gay men, lesbians, bisexuals, and transgendered people account for a non-trivial minority–the most recent survey I’m aware of found 7 percent of women and 8 percent of men in the U.S. identify as L, G, B, or T. We don’t have remotely comparable historical data, but mention of same-sex sexuality goes back to the dawn of recorded history. If natural selection is homophobic, it’s not particularly good at it.

Before I get much farther, here’s a disclaimer: I’m going to consider how same-sex attraction might persist in human populations in the face of its apparent selective disadvantages. In the absence of direct data–such as systematic measures of the the total evolutionary fitness of gay men or lesbians in specific societal contexts–it’s easy to make up stories about natural selection, but much harder to determine which stories reflect reality. I’ll try to delineate which stories fit with what we know about how selection works, and with the little data we do have–but that’s the best I can do. If there’s one point I hope you take from all that follows, it’s that evolution is complicated, and human evolution doubly so.

Natural selection isn’t all-powerful

Human Rights Campaign banner from Portland Pride 2008. Photo by PDX Pixels.

Natural selection causes traits associated with having fewer children to become less common over time. But natural selection is not the only evolutionary process at work in natural populations. Mutation introduces new alleles even as natural selection removes them. Furthermore, the effects of random chance in small populations creates an effect called genetic drift, which can interfere with the expected operation of natural selection.

Evolutionary biology has developed an excellent understanding of how mutation, selection, and drift interact over time to shape the genetic diversity of populations. That understanding allows us to do some back-of-the-envelope calculations to see how selection might operate on a gene associated with same-sex attraction. In setting this up, I’m following the lead of the evolutionary biologist Joan Roughgarden, who makes a similar point in her book Evolution’s Rainbow. Brace yourself for some math!

In an idealized population of infinite size, the balance between natural selection’s effect of removing disadvantageous alleles, and mutation’s effect of spontaneously re-creating them, means that the equilibrium frequency of the allele in the population should be about equal to the square root of the ratio between the mutation rate and the selective cost associated with carrying two copies of the disadvantageous allele.

A single base pair of human DNA has a chance of of mutating equal to about one in one hundred million [$a] every generation. Since there may be thousands of base pairs [$a] in a single gene, the probability of a mutation occurring somewhere in the gene is more like one in one hundred thousand. If we assume that it takes two copies of our hypothetical “gay allele” to make a person attracted to members of the same sex, and about five percent of people are attracted to members of the same sex, then mutation alone could balance a selective cost to being gay of 0.0002. That is to say, mutation-selection balance alone could explain the frequency of LGBT folks in the population if those attracted to the same sex had, on average, 0.9998 children for every child born to the average straight parent. That’s pretty weak selection.

This is where genetic drift enters the picture. Most natural populations don’t behave anything like the mathematical ideal assumed for the calculations in the preceding paragraph, because most natural populations are not infinite in size. In finite populations, randomness–“mere bad luck” in the words of pioneering biologist J.B.S. Haldane [$a]–can prevent selection from operating efficiently. Smaller populations are more prone to genetic drift–the relevant number is not necessarily the number of individuals in the population, but the effective population size. The classical estimate of the human effective population size is about ten thousand [$a], and more recent estimates have come up with even smaller numbers [$a]. This is because our population’s expansion to billions is a very recent phenomenon by evolutionary standards, and may reflect the fact that for much of our history we lived in smaller, isolated populations [$a]. For populations in this size range, selection may not operate efficiently.

Setting up an experiment that would take into account the effects of drift, mutation, and selection acting together on the human population is impossible both in practical and ethical terms. That leaves biologists two ways to approach the question of how a particular disadvantageous allele can persist in the human population: intensive study of the population’s genetics, and mathematical or computer-based modeling. Lacking easy access to massive amounts of human population genetic data, I’ve built a computer model.

The model is a script for the excellent open-source programming language R, and you can download it here if you’re interested. In an nutshell, it simulates the evolution of a population of critters that may have two copies, one copy, or no copies of a deleterious allele. Critters with two copies have their chances of reproducing reduced by a set amount, which is the selective cost of carrying two copies of the allele; critters with one or no copies experience no such cost.

Every generation, the critters who survive to mate pair with randomly-selected partners to form offspring, who then replace their parents to start the cycle all over again. This randomized mating allows for drift to occur–everyone who survives to mate has an equal chance to mate, but some are randomly paired more than once, and some miss out. Mutation occurs at the moment of reproduction, when the alleles passed on from parents to their offspring have a small chance of changing to the deleterious form, or back to the harmless form.

I set up the simulation with a population size of ten thousand, in which about five percent of the critters carry two copies of the deleterious allele. I set the cost of carrying two copies of the allele to 20 percent, and the probability of mutation each generation to one in one hundred thousand. Here’s what happens to the percentage of critters carrying two copies of the deleterious allele over fifty generations of sim-evolution. The blue line is the percentage of critters in the population carrying two copies of the deleterious allele; the dotted black line marks the starting percentage, for reference.

Image by jby.

You can see that selection wins out, as we’d expect–but it doesn’t do so immediately. In fact, there are periods where the percentage of critters with two disadvantageous alleles increases. That’s the randomness of drift in action.

To get a sense of how this drifting randomness plays out in general, we need to run the simulation many times, and see what tends to happen. This is like flipping a coin over and over to see whether it really does land heads side up fifty percent of the time. Here are ten replicate simulations, with graphs like the one above superimposed on each other for comparison:

Image by jby.

And here are a hundred simulations:

Image by jby.

So drift and mutation complicate things–sometimes, rarely, the disadvantageous allele persists for 50 generations. However, we can say from these simulations that selection removes the allele far more often than not. These simulations assume selection quite a bit weaker than the widely-cited cost of same-sex sexuality–20 percent lower fitness instead of 80 percent–so you would expect selection to be even more effective against an allele that makes men gay and women lesbian. However, as I noted above, same-sex sexuality has been present in human populations for considerably longer than 50 generations. That suggests my simulations don’t accurately reflect the real-world evolution of human sexual minorities–not because I’ve simulated weaker selection, but because I’ve simulated selection that works too well.

It could be that I’ve modeled the genetics incorrectly. Most of what I’ve discussed up to now assumes that there’s only a single gene involved in determining sexual orientation. However, as the evolutionary biologist Douglas Futuyma has pointed out [$a] in his review of Roughgarden’s book for the journal Evolution, if multiple genes are involved, they would “share” the selective costs associated with same-sex attraction. Selection would then have proportionally less power to remove alleles for same-sex sexuality in the face of drift and mutation. Given that recent genome studies have not clearly identified a single gene region [$a] associated with sexual orientation, it seems likely that multiple genes are indeed involved.

The other possibility, however, is that selection against same-sex sexuality is not as strong as I’ve made it in my simulations. How is that possible? Well, simply put, the lives of gay men, lesbians, and transgendered people in Western societies over the last few decades might not be much like the lives we would live in other times and other places. And that could make all the difference.

Context matters

In the United States, the most well-known LGBT life story goes something like this: one discovers same-sex attraction in adolescence, and comes out of the closet before having much sexual interaction with members of the opposite sex. For committed same-sex partners, having biological children is possible via surrogacy or sperm donation, but it’s complicated by the lack of legal recognition and protection for the couple and for children they choose to have. All these factors tend to reduce the number of biological children gays, lesbians, and trans folks have–but they’re also all phenomena of our current historical and political moment. In different social contexts, LGBT fitness could very well be higher.

Before the gay rights movement, social expectations probably led many people who today would identify as gay or lesbian to enter into straight marriages and raise families. This kind of social pressure may explain why gay and lesbian couples are more likely to be raising children if they live in the conservative southern United States–not because adoption is more common in that region, but because in the South, gay men and lesbians are more likely to have heterosexual relationships, and children, before they come out.

On the other side of that coin, it’s possible to imagine that in societies in which same-sex relationships recieve the same legal recognition as straight marriages, gay men and lesbians might eventually have biological children at substantially higher rates than they do today. So if oppression might reduce the fitness cost to being gay, equality probably could, too.

Many non-Western societies, too, have accepted social roles for a “third gender” that can encompass identities approximating Western gays, lesbians, bisexuals, and transgendered folks. The roles and behaviors of third-gendered individuals vary considerably, but in some cases they may have biological children even if their primary relationships are with members of the same sex. As Christopher Ryan and Cacilda Jetha describe extensively in their book Sex at Dawn, pre-agricultural human societies may have been highly polygamous by modern standards, with children raised communally. In that context, Ryan and Jetha propose, same-sex sexual interactions may have provided social capital that could help to support children produced in relatively infrequent heterosexual couplings.

A bluegill sunfish. Photo by IcK9s [M.H. Stephens].

This is consistent with the picture Joan Roughgarden paints in Evolution’s Rainbow. Roughgarden’s book describes widespread same-sex activity that usually fosters social relationships in support of reproductive sex rather than instead of it. To pick just one example out of dozens, some male bluegill sunfish look and behave like females, and “court” more masculine males–then help attract a female and share in the resulting three-way reproductive opportunity by fertilizing some of her eggs. The wide range of non-reproductive sexual behavior in the broader animal kingdom suggests that human sexual minorities are just one manifestation of a phenomenon that could date back to the origins of sex itself.

Reproduction by proxy

It has also been suggested that LGBT folks might make up for a lack of biological children by boosting the reproductive success of their close relatives [PDF]. My brother shares half of my genetic material–so if I help him and his future wife support more children than they would have otherwise, those nephews and nieces “count” towards the children I’m not making myself.

This kind of indirect fitness might offset a lack of direct reproduction, but I doubt that it can cancel it out. Genetically, a nephew “counts” towards my fitness about half as much as a son does. Therefore, if I would otherwise have two children on my own, I have to help my brother to have four additional children to make up for them. I’d like to think I’ll make a good uncle, but I won’t be that good. And, in fact, one survey of gay men has found that they aren’t significantly more generous [$a] toward their nephews and nieces than straight men are. (One objection you might make to this study is that it addresses our current societal context, not that in which humans originally evolved.)

A subtle twist on the indirect fitness idea that I haven’t seen in the scientific literature could be to consider not the fitness of gay men and lesbians, but their mothers. In order for my mother to have as many grandchildren as she would with two straight sons, I only need to help my brother and his wife support two additional children. That seems more achievable, and might work out in the context of the polyamorous, mutually-helping tribes described in Sex at Dawn. It’s also consistent with the observation that men are more likely to be gay if they have older brothers–a woman who has already had several straight sons might, conceivably, have more surviving grandchildren by giving them a helpful gay uncle.

Finally, there is some evidence that genes associated with same-sex attraction in men might provide a fitness boost in women. Since any given gene has about a 50 percent chance of ending up in one sex or the other, a gene that makes men more likely to be gay but makes women more fertile might, on average, have no selective advantage or disadvantage. A 2004 study found that women related to gay men have more children, which supports this scenario. The biologists Sergey Gavrilets and William Rice used a mathematical model of selection on same-sex sexuality to consider this hypothesis in a 2006 paper, and found that a female fertility boost could indeed allow male same-sex sexuality to persist.

In the end, however, this is mostly storytelling–lots of possibilities, but much less hard data. What we need to test many of these ideas is detailed records of the total reproductive fitness of sexual minorities in specific social contexts–especially societies approximating the ones formed by our earliest human ancestors. The best we can say without this is that many societal contexts could have made the apparent fitness cost to same-sex attraction smaller than it appears at first glance.

Who cares what natural selection thinks, anyway? Photo source unknown, presumed public domain.

So where does all of this leave the evolutionarily-aware gay man, lesbian, or transgendered person? As I noted at the start, figuring out the exact nature of our tenuous relationship with natural selection doesn’t tell us much about our moral stature, our value to society, or the best way to live our lives. It does, however, offer to answer the question that evolutionary biology can potentially answer for all human beings, regardless of orientation, gender, or race: how did we come to be what we are?

The best answer we have so far is complicated–it may be that we’re children of history and chance, not a clear-cut adaptive path. But easy lives and clear-cut answers aren’t, I think, what we celebrate in the history of our LGBT forerunners, or remember at Pride rallies. If we queer folk live our lives in the tail of a probability distribution, the the good news is that the company here is pretty good.

Thanks (or blame) for this article is owed to Steve Silberman, who got me thinking about this topic in an e-mail conversation last year; and to Chris Smith, who gave me some invaluable suggestions on an earlier draft. I haven’t seen Tony Kushner’s newest play yet, but I’d happily trade a fraction of my lifetime reproductive fitness for a couple tickets.


Bogaert, A. (2006). Biological versus nonbiological older brothers and men’s sexual orientation Proc. Nat. Acad. Sciences USA, 103 (28), 10771-4 DOI: 10.1073/pnas.0511152103

Bailey, N., & Zuk, M. (2009). Same-sex sexual behavior and evolution. Trends in Ecology & Evolution, 24 (8), 439-46 DOI: 10.1016/j.tree.2009.03.014

Camperio-Ciani, A., Corna, F., & Capiluppi, C. (2004). Evidence for maternally inherited factors favouring male homosexuality and promoting female fecundity. Proc. Royal Soc. B, 271, 2217-21 DOI: 10.1098/rspb.2004.2872

Eller, E., Hawks, J., & Relethford, J.H. (2010). Local extinction and recolonization, species effective population size, and modern human origins. Human Biology, 81, 805-24 DOI: 10.1353/hub.2005.0006

Futuyma, D.J. (2005). Celebrating diversity in sexuality and gender. Evolution, 59, 1156-9 DOI: 10.1111/j.0014-3820.2005.tb01052.x

Gavrilets, S., & Rice, W.R. (2006). Genetic models of homosexuality: generating testable predictions. Proc. Royal Soc. B, 273, 3031-8 DOI: 10.1098/rspb.2006.3684

Haldane, J.B.S. (2008). A mathematical theory of natural and artificial selection. Part V: Selection and mutation. Mathematical Proc. Cambridge Phil. Soc., 23 (07) DOI: 10.1017/S0305004100015644

Hamer, D.H., Hu, S., Magnuson, V.L., Hu, N., & Pattatucci, A.M. (1993). A linkage between DNA markers on the X chromosome and male sexual orientation. Science, 261, 321-7 DOI: 10.1126/science.8332896

Herbenick D., Reece M., Schick V., Sanders S.A., Dodge B., & Fortenberry J.D. (2010). Sexual behavior in the United States: results from a national probability sample of men and women ages 14-94. The journal of sexual medicine, 7 Suppl 5, 255-65 PMID: 21029383; Available online for free via Indiana University.

Pillard RC, & Bailey JM (1998). Human sexual orientation has a heritable component. Human Biology, 70 (2), 347-65 PMID: 9549243

Rahman, Q., & Hull, M.S. (2005). An empirical test of the kin selection hypothesis for male homosexuality. Archives of Sexual Behavior, 34, 461-7 DOI: 10.1007/s10508-005-4345-6

Ramagopalan, S., Dyment, D.A., Handunnetthi, L., Rice, G.P., & Ebers, G.C. (2010). A genome-wide scan of male sexual orientation. Journal of Human Genetics, 55, 131-2 DOI: 10.1038/jhg.2009.135

Roach, J., Glusman, G., Smit, A., Huff, C., Hubley, R., Shannon, P., Rowen, L., Pant, K., Goodman, N., Bamshad, M., Shendure, J., Drmanac, R., Jorde, L., Hood, L., & Galas, D. (2010). Analysis of genetic inheritance in a family quartet by whole-genome sequencing. Science, 328 (5978), 636-9 DOI: 10.1126/science.1186802

Rice, G. (1999). Male homosexuality: Absence of linkage to microsatellite markers at Xq28. Science, 284 (5414), 665-7 DOI: 10.1126/science.284.5414.665

Roughgarden, J. (2004). Evolution’s Rainbow. University of California Press, Berkeley. Google Books.

Takahata, N. (1993). Allelic genealogy and human evolution. Molecular Biology and Evolution, 10, 2-22 : 8450756

Tenesa, A., Navarro, P., Hayes, B.J., Duffy, D.L., Clarke, G.M., Goddard, M.E., & Visscher, P.M. (2007). Recent human effective population size estimated from linkage disequilibrium. Genome Research, 17, 520-6 DOI: 10.1101/gr.6023607

Xu, L., Chen, H., Hu, X., Zhang, R., Zhang, Z., & Luo, Z.W. (2006). Average gene length is highly conserved in prokaryotes and eukaryotes and diverges only between the two kingdoms. Molecular Biology and Evolution, 23, 1107-8 DOI: 10.1093/molbev/msk019

  • Just read the comments on Sci Am…

    … Umm… I warned you.

    • Would you believe I haven’t looked at the comments on SciAm yet? And now I’m not sure I want to.

      • Aaaand now I have, and would you believe I don’t find any of it particularly shocking? I’m not going to wade in myself, but that’s really a pretty benign collection of crankery, offset by some folks who care enough to call bullshit on it.

  • Re: Reproduction by proxy

    First time I came across this notion was in one of E.O. Wilson’s books. I think it was Naturalist

  • (Sorry, no ref but) isn’t is known that the chance of a son being homosexual doubles with each successive son.  It’s been postulated that the reason is that with each pregnancy the mother is more likely to have developed antibodies to some of the fetal androgens that presumably prime the brain for sexual preference.  I think there is some similar but different biochemistry in lesbians with high maternal androgen levels.

    If this holds up the question is why doesn’t the fetus resist the mother’s antibodies, or the mother recognise the androgens as friendly.  The likely option is that they can’t easily.  After this we get into more complex social adaptive ideas like those you covered.  Or a combination.

    • Jim, I believe the reference you want is Bogaert (2006), which is freely available for download from PNAS. Short answer: yep, birth order seems to be a factor.

  • JustAGuy

    Jeremy, all that matters in this world is what PZ Myers thinks. He seems to agree with most of what you said. You win.


  • Nice summary of the literature on this.  I really enjoyed this post, and I’m glad I discovered your blog.

  • Brilliant article. I very much enjoyed reading it. As someone interested in evolution, statistics and being bisexual, it is interesting for me to think about how we (LGBT group) came to be what we are.

  • Michael McDarby

    It seems likely that we’re discussing a multiple-gene trait, where certain combinations are reproductive dead-ends but other combinations are strongly selected.  Kind of like schizophrenia – a Danish study suggested that a remixing of schizophrenic alleles were correlated with imagination and possibly entreprenureal risk-taking.