An enduring mystery (among many mysteries) is the existence of women exhibiting sexual attraction to other women. Unlike male homosexuality, where a likely explanation has been put forth (see Greg Cochran’s “Gay Germ” Hypothesis – An Exercise in the Power of Germs), female same-sex attraction remains the realm of speculation. My previous foray into the matter, The Evolution of Female Bisexuality, contains much of that speculation. I’ve learned quite a bit since then, so the time has come to revisit the matter. This time I will make use of one of the most powerful exploratory methods in social science, behavioral genetics.
The first thing to look at is the heritability. An analysis of large twin registry studies pegs the heritability of female same-sex attraction (SSA) as 33% (Whitehead, 2011). This is in contrast to the very low heritability of male SSA, 22%, as reported in the same study. However, these heritabilities were not significantly different.
Unfortunately, a key weakness here (as with virtually all sex research) is that this relies entirely on self-report – worse, with no form of corroboration from any other measurement. Hence, measurement error can be expected to be large.
Looking at the various studies examined by Whitehead, heritabilities were quite variable. A lot of this stems from the relative rarity of SSA, making samples of SSA individuals small even in large studies. (Another problem with that non-response was generally high in these studies, which may have biased heritabitlity estimates.) Only the largest population-based studies with good compliance can firmly pin down the heritability of SSA.
The low heritability and evolutionary contradiction of male homosexuality necessitates the pathogenic explanation. But, assuming the 0.33 heritability of female SSA is reliable, is an evolutionary explanation workable? It is quite possibly is.
A big component is the fitness impact of female SSA. It appears to be much more common than male non-heterosexuality (again, assuming self-reports are to be believed). A British survey (Mercer et al, 2013) finds that among the youngest cohorts of women (ages 16-34), as much as 19% claim to have sexual contact with another woman (the fraction maxes out at around 9-10% for men).
However, the racial composition of the sample changes considerably across age, going from 92% to only 82% White from the oldest to the youngest cohort. Hence it’s unclear how much demographic changes are driving this apparent generational change (more on that shortly).
Nonetheless, female SSA is quite common. Its historic fitness impact then would appear to be – at worst – not as deleterious male same sex attraction. According to the Add Health data (a nationally representative U.S. teen/young adult sample), the predominant (de facto) orientation of non-heterosexual women is some sort of bisexual, indeed “mostly heterosexual” (from Udry & Chantala, 2006):
(My own personal suspicion is that most of the men claiming to be “mostly heterosexual” are in fact gay.)
As a check, I looked at the General Social Survey (GSS) to see what the reported overall behavior of non-hetero woman was. I looked at women (all races) who reported 1 or more female sex partners. This is the number of male sex partners these women claim to have had:
As we can see, non-hetero women are apparently quite promiscuous; such women with no or only a few male sex partners are very much the exception. Indeed, a third or more have had more than eight male partners; ~15% or more claim to have had more than 20 male partners!
The second chart is the number of female sex partners women who report one or more female sex partners claim to have had (all races). As we see, the most common value is just the one. All these indicate that such women are indeed primarily attracted to men.
If this is representative of past potential inclination (not necessarily realized behavior]), then the fitness impact of female SSA couldn’t have been too negative, at worst. An interest in other women likely did not preclude marrying and having children for women historically, especially if they were primarily attracted to men.
Also, for the record, I checked the GSS to see if the apparently highish frequency of women reporting SSA was driven by racial differences in SSA. As we see, even when we look at Whites only, we see a noticeable generational rise in the fraction of women who report sexual experiences with other women. I also looked at racial differences, and the values are similar for other races, except for Hispanics, who have consistently reported a 10-13% female-female sex rate for all of these cohorts. Whites and Blacks have merely converged with the Hispanic rate as of late.
Could the fitness impact of female SSA have been positive? That is, was it specifically selected for? I suspect not. While it is common, its nonetheless minority status would entail some sort of balancing selection to remain at its low level. The most plausible type, frequency-dependent selection, implies that female SSA is beneficial when it is rare. I can’t see how this would be the case (especially since it would be tough for such a girl who can’t find other women who are interested in her).
No, I don’t think female SSA is an adaptation at all. Rather, if it is actually genetic in nature, then the most likely explanation appears to be that it is some sort of side effect of something else. I suspect that it may be due to sexual antagonistic selection. That is, its existence may be driven by selection on alleles that have positive effects in men.
Key to this idea is the apparently near neutral past fitness impact of female SSA (its present impact is decidedly deleterious, as per the GSS). It was not rapidly selected out, unlike any alleles which would cause male homosexuality.
A clue here is the characteristics of non-hetero women. They are often masculinized relative to other women, with higher sex drives and, as we see, greater promiscuity.
Here again, twin studies are informative. One twin study (Burri, Spector, and Rahman, 2015) found that in female twins, masculinity (measured by childhood “gender non-conformity”) was genetically correlated both with female non-heterosexuality and number of sex partners, through a latent genetic factor. EDIT: added the genetic pathway image:
One idea that we can lay to rest here is the notion that female masculinization and SSA stem from prenatal exposure to male hormones. The way to test this idea is to look at females with a male fraternal twin (Fm) compared to women with female fraternal twins (Ff). If the Fm women were more masculine that Ff women, then it would imply prenatal hormones were at work. A massive review looking at such twin studies (Tapp, Maybery, and Whitehouse, 2011) found little support for this notion, particularly from the larger studies examined. Another recent twin study out of Denmark (Ahrenfeldt et al, 2015) found no such effect on women’s academic performance (Fm twins didn’t exhibit a more male-typical cognitive profile).
It also has been said that non-heterosexual women look different from straight women, the former appearing more masculine. Unfortunately, I haven’t been able to find much by way of good studies examining female appearance and sexual orientation. There are a few small and questionable ones, but nothing I’d take too seriously. That said, there’s another way to look indirectly at the matter.
Two twin studies by the same team (Mitchem et al, 2014 and Lee et al, 2014) looked at the heritability of facial attractiveness and facial masculinity-femininity in two sets of large and well-measured samples. They found that both are highly heritable (primarily driven by additive genetic factors), but what’s more, high facial masculinity in men led to high facial masculinity in their sisters, decreasing these women’s attractiveness.
This spillover of male traits on their female relatives can also be seen in “sociosexuality” (promiscuity). A largish twin study out of Australia (Baily et al, 2000) found sociosexuality in men predicted the same in their female co-twins. As well, there is the Zietsch et al (2008) study that found that masculinized women were more likely to be non-heterosexual, and that straight women with a non-heterosexual female twin reported somewhat more sexual partners.
Unfortunately, all these fall short of a more ideal of study, one that examines the morphological and psychological characteristics of women and their male relatives to pin down good familial predictors of female SSA.
Nevertheless, taken together, these point in the direction of masculinity being a distinct suite of traits and that appears related to female SSA. In terms of facial features, that masculinity is distinct from attractiveness in men is interesting. It speaks to perhaps a set of “masculinity promoting” genes that act somewhat orthogonal to those that lead to good looks. Research is mixed on the extent that women find masculine facial features attractive, but I’d say it’s safe to say that these alleles underwent positive selection in men. However, unfortunately, manly men beget manly daughters. Some of these alleles may serve to increase “gynephilia” (attraction to women), and they may have this effect in both sexes. This may be where SSA comes from in women. Unlike male SSA, female SSA appears to be continuously distributed – consistent with what we’d expect if it arose from the general load of various gynephilic alleles (resulting in pure lesbianism in the most extreme cases). The female SSA itself, being near selectively neutral, could reach high frequencies in women, just as masculinity in general has. However, the general suite of masculinized traits (e.g., sociosexuality, facial masculinity) was likely maladaptive in women, and was selected against, counteracting the positive selection in men. In other words, sexual antagonistic selection. A GWAS (Drabant et al, 2012) found no genetic hits to sexual orientation, indicating that female SSA – if primarily genetically driven – is caused by many genes of small effect, as we might expect.
Of course, much of this info comes from WEIRD countries (i.e., NW European and offshoots). As with much of social science in general, these investigations should be repeated with large samples from non-Western societies.
As for the male interest in girl-girl sex (*guilty*), I still suspect it’s a simple matter being able to have multiple women. Even men from highly conservative cultures may not express interest, but often such men harbor rather “colorful” sexual proclivities. Further research needed.
Now, is all of this compatible with a pathogenic source for female SSA? Absolutely, especially if the true heritability turns out to be truly low. See Peter Frost (Yes, Demons Do Exist). Perhaps the recent trend towards more women displaying SSA is due to the spread of a new pathogen. Time will tell. However, unlike male homosexuality, I don’t think we can retire the pure genetic explanation, yet.
The obvious theme song for this post: