Be advised: rape, Republicans, and evolutionary psychology will be discussed herein. Also there is math.

So I didn’t think I’d be posting anything about that idiot who’s running for Senator from Missouri, Todd Akin, who said that it’s not necessary to ensure that abortion remain a safe and legal option for victims of rape because women’s bodies somehow kill or eject sperm from people they didn’t want to have sex with, or something. Which is just so obviously, stupidly, dangerously wrong that what could I possibly have to say about it that hasn’t already been said better elsewhere?

Then, on Monday, this happened:

Original tweet here.

That’s the immediate follow-up to a tweet linking to an article Bering wrote a couple weeks ago, in which he explains the reasoning behind an evolutionary psychology hypothesis that women’s bodies can, in some cases, do something very like what Todd Aikin thinks they can. In 2006, Jennifer Davis and Gordon Gallup (yes, that’s Gordon “adaptive homophobia” Gallup) noted in a book chapter [PDF] that there’s an interesting pattern associated with preeclampsia, a common, generally late-term complication of preganancy: pregnant women seem to be less likely to experience preeclampsia if they’ve been living with the father of the fetus for a longer time before becoming pregnant [$a]. The idea being, that if a woman is inseminated with “unfamiliar” sperm, her body is more likely to reject the resulting pregnancy.

If this sounds ridiculous to you, well, it did to me, too. I reacted right off the bat with Tweeted snark, and had some back-and-forth with Bering in which, I have to say, I didn’t acquit myself especially well. I try to do better than engaging in scientific debate while steam is coming out of my collar, and, once I’d cooled down a bit I attempted an apology (which recieved a perfectly polite response), and I then resolved to sit down and actually figure out the merit of Davis and Gallup’s hypothesis.

Spoiler alert: Further study did not make D&G’s hypothesis any more plausible. But my reasons for disbelieving it after doing the background reading aren’t what you might expect.

Preeclampsia and semen familiarity

First, a little more detail about preeclampsia. It’s a condition linked to pregnancy-induced hypertension—not so much a single complication as a cluster of symptoms that seem to be connected to an immune reaction against the “foreign” tissue of a fetus. As the name implies, it can escalate to eclampsia, which involves seizures and a coma, and can seriously endanger both the woman and her fetus; but preeclampsia can create dangerous complications [$a] even if it doesn’t get to that stage, causing premature delivery, restricted fetal growth, and loss of the pregnancy.

One of the first things I found out, following up on papers cited in Bering’s post and D&G’s original description of the hypothesis, is that the “semen familiarity” idea is actually quite widespread in the medical literature and, as far as I can tell, reasonably well supported. Various studies have found that preeclampsia is more common in first pregnancies; more common in women who have switch sexual partners between pregnancies [PDF]; and more common in women who become pregnant via artificial insemination from an unknown donor [$a] than in those who are artificially inseminated with their current partner’s sperm. That last study has a rather small sample size, but it’s about as close as you could get to an outright experimental test of the “semen familiarity” effect, I think.

However, if sperm familiarity is one factor contributing to the risk of preeclampsia, it’s not the only thing, and its role isn’t quite universally accepted. It’s a risk factor in the epidemiological sense, not a direct, clear-cut cause. Women who are overweight, who are diabetic, who smoke, or who have hypertension when they become pregnant have an elevated risk of preeclampsia independent of any effect of sperm familiarity. A relatively recent review article notes that there’s some support for an alternative hypothesis that longer waits between pregnancies is a stronger determinant; women who change partners often also tend to wait longer between pregnancies because, well, it can take a while to make that kind of switch. Studies that account for time between pregnancies have found that, in fact, switching partners can be associated with a somewhat reduced risk of preeclampsia.

Rolling the die against rape

But so if preeclampsia is indeed made more likely by unfamiliar semen, how much of a selective advantage could this tendency incur? (Assume, for the moment, that there’s a genetic underpinning to the tendency to respond to unfamiliar sperm by developing preeclampsia; natural selection can’t act on any trait that isn’t passed from parent to offspring with reasonable reliability, no matter how useful or detrimental that trait might be.) As I’ve noted in other contexts, a very basic result in population genetics is that, for natural selection to over come the effects of genetic drift and mutation [$a], it has to have some minimumn strength; any selective advantage at all isn’t enough for a gene variant to spread via natural selection.

To determine whether selection favoring preeclampsia as a response to unfamiliar sperm might be strong enough to overcome drift and mutation, we’ll have to do some back-of-the envelope calculations. Here, I can draw on some data from the medical literature, but this is all pretty crude, so grab your salt-cellar.

First, how much more likely does unfamiliar semen make preeclampsia? In that above-mentioned comparison of artificial insemination by unknown donors versus familiar partners, which was published in 1997, women recieiving donor sperm were about 1.85 times more likely to develop preeclampsia than those who recieved sperm from their partners; a much larger study from 1999 [PDF] reports that preganancies resulting from sperm donation had about 1.4 times the risk of preeclampsia seen in comparable natural pregnancies.

That baseline risk is about anywhere from 2 to 7 percent of pregnancies. So if we take the higher end of both estimates (the baseline probability of preeclampsia, and the factor by which unfamiliar sperm increases it), we’re talking about an effect that elevates the probability of developing preeclampsia up to about 13 percent. That’s not zero, but (to take Trisha Greenhalgh’s advice to heart) let’s try to think about that in concrete terms: it’s less than the probability of rolling a six with one toss of a die.

Then, consider that not all preeclampsia cases result in loss of the pregnancy. The current risk of fetal death associated with preeclampsia is about 1 to 2 percent of cases; so now figure that you have to toss that die more than enough times to roll six 100 times—more than six hundred tosses—to be reasonably sure of ending just two rape-related pregnancies this way. Put it another way: a (purely hypothetical) gene variant responsible for making a woman likely to develop preeclampsia when she encounters unfamiliar sperm would help her avoid carrying a rapist’s baby to term with something less than one chance in 300.

(One caveat: of course, I’m working from present-day risks of pregnancy loss due to preeclampsia, and of course preeclampsia would’ve been more likely to result in pregnancy loss—and also maternal death—before the advent of modern medicine. But I wasn’t able to find similarly precise estimates of those risks predating modern medicine, and in any event Davis and Gallup, and Bering, discuss the hypothesis in terms of its implications for modern society.)

Strong enough for selection?

Now, let’s compare that educated guess to the estimated strength of natural selection acting on two adaptations biologists have studied much more closely in humans: the capacity to survive in high-altitude, low-oxygen conditions, and the ability to digest milk sugars as adults. These are each cases where a useful genetic variant has spread through a population, which means selection overcame drift and mutation; although I don’t believe that either adaptive variant has “fixed,” or spread to the entire population.

In the first case, a gene variant found in people living on the high Tibetan plateau is associated with reduced risk of death for the children of women carrying the variant. A 2004 study of Tibetan women found that those without the variant gave birth to about 4.5 children, and an average of 2.5 of those children died before the age of 15; women carrying the beneficial gene variant had about the same number of live births, but only an average of 0.5 children who died. In other words, carrying the high-altitude gene variant meant they had twice as many children surivive to age 15.

In the second case, a 2009 study used population genetic data to estimate the strength of selection on the gene variant responsible for lactase persistance, the ability to digest milk sugars as an adult, in European populations that have historically raised cattle for milk. The estimated selective benefit of being able to digest milk was about 1.8 percent. That is, people in those European populations who couldn’t digest milk had about 98.2 children for every 100 children born to people who could digest milk.

Stack those selective effects alongside that proposed for preeclampsia as a response to rape: less than a one-in-six chance for a two percent chance of losing an unwanted pregnancy, or somewhat less than three out of a thousand rape-related pregnancies ended prematurely. And, as Kate Clancy notes in her excellent discussion of the Akins fiasco, preeclampsia characteristically occurs late in pregnancy—so, in the rare cases when it does end an unwanted pregnancy, it does so after a mother has already invested months of resources in supporting the fetus.

As Clancy points out, an adaptation to prevent pregnancy by rape would be much more effective if it caused miscarriage well before preeclampsia could even come into play—and, indeed, Davis and Gallup proposed, at the end of their book chapter, that earlier miscarriages might also be related to semen familiarity. They cite no data to test that hypothesis, and I haven’t found any published since their book chapter. But as Clancy describes quite clearly, we have reaonable evidence that rates of pregnancy from rape are similar to rates of pregnancy from consensual sex, and that would seem to close the book on the question of anti-rape defenses in early pregnancy.

In other words, if women have evolved some sort of physiological adaptation to avoid getting pregnant as a result of rape—whether via elevated risk of preeclampsia or another means—the actual benefits conferred by such an adaptation are so miniscule as to stretch the definition of “adaptive” to meaninglessness. But I can think of another well-known adaptation that does allow women to end unwanted pregnancies with a high degree of reliability: human intelligence. Women have been using abortifacients and other means to end pregnancies, sometimes well before preeclampsia typically occurs, since the dawn of recorded history, and modern medical technology from hormonal birth control to emergency contraception to, yes, abortion itself makes this simpler and safer than it’s ever been.

Contrary to Jesse Bering’s quippy title, Darwin’s morning after pill isn’t some mysterious power of a woman’s reproductive tract; it’s the big brain that millions of generations of evolutionary history gave her.◼

As noted in the main text, all calculations herein are back-of-the-envelope estimates, and subject to the foibles of my limited numerical skills; if you see something wrong with them, let me know in the comments!

References

Beall, C. M., K. Song, R. C. Elston, and M. C. Goldstein. 2004. “Higher offspring survival among Tibetan women with high oxygen saturation genotypes residing at 4,000 m.” Proc. Nat. Academy Sci. U.S.A. 101:14300. DOI: 10.1073/pnas.0405949101.

Davis J.A., and G.G. Gallup Jr. 2006. “Preeclampsia and other pregnancy complications as an adaptive response to unfamiliar semen.” in Female Infidelity and Paternal Uncertainty: Evolutionary Perspectives on Male Anti-Cuckolding Tactics. SM Platek and TK Shackleford, eds. Pages 191-204. Full text PDF.

Dekker, G., and P.-Y. Robillard. 2007. “Pre-eclampsia: Is the immune maladaptation hypothesis still standing?: An epidemiological update.” Journal of Reproductive Immunology 76:8-16. DOI: 10.1016/j.jri.2007.03.015.

Gerbault, P., C. l. Moret, M. Currat, and A. Sanchez-Mazas. 2009. “Impact of selection and demography on the diffusion of lactase persistence.” PLoS ONE 4:e6369. DOI: 10.1371/journal.pone.0006369.

Haldane, J. B. S. 1927. “A mathematical theory of natural and artificial selection. Part V: Selection and mutation.” Proceedings of the Cambridge Philosophical Society 23:838-844. DOI: 10.1017/S0305004100015644.

Hoy, J., A. Venn, J. Halliday, G. Kovacs, and K. Waalwyk. 1999. “Perinatal and obstetric outcomes of donor insemination using cryopreserved semen in Victoria, Australia.” Human Reproduction 14:1760-1764. DOI: 10.1093/humrep/14.7.1760.

Li, D.-K., and S. Wi. 2000. “Changing paternity and the risk of preeclampsia/eclampsia in the subsequent pregnancy.” American Journal of Epidemiology 151:57-62. Full text PDF.

MacKay, A. P., C. J. Berg, and H. K. Atrash. 2001. “Pregnancy-related mortality from preeclampsia and eclampsia.” Obstetrics & Gynecology 97:533. Full text PDF.

Robillard, P.-Y., and T. Hulsey. 1996. “Association of pregnancy-induced-hypertension, pre-eclampsia, and eclampsia with duration of sexual cohabitation before conception.” The Lancet 347:619 DOI: 10.1016/S0140-6736(94)91638-1.

Sibai, B., G. Dekker, and M. Kupferminc. 2005. “Pre-eclampsia.” The Lancet 365:785-799. DOI: 10.1016/S0140-6736(05)17987-2.

Skjaerven, R., A. J. Wilcox, and R. T. Lie. 2002. “The interval between pregnancies and the risk of preeclampsia.” New England Journal of Medicine 346:33-38. DOI: 10.1056/NEJMoa011379.

Smith, G. N., M. Walker, J. L. Tessier, and K. G. Millar. 1997. “Increased incidence of preeclampsia in women conceiving by intrauterine insemination with donor versus partner sperm for treatment of primary infertility.” American Journal of Obstetrics and Gynecology 177:455-458. DOI: 10.1016/S0002-9378(97)70215-1.