The Darwinian Sweet Spot: 3rd Cousin Marriages?

It's long been understood theoretically that there must exist a Darwinian fitness trade-off between too much inbreeding and too much outbreeding, but nobody knew where that was. If you marry your first cousin, you are likely to suffer a 30% higher infant mortality rate. But if you marry somebody too genetically dissimilar, you can start running into various reproductive problems as well.

Now, deCODE Genetics of Iceland, who foisted upon the world the most likely fallacious claim that James D. Watson is 25% nonwhite, is claiming that the Darwinian fitness sweetspot is 3rd cousin marriage:

In a paper published today deCODE scientists establish a substantial and consistent positive correlation between the kinship of couples and the number of children and grandchildren they have. The study, which analyzes more than 200 years of deCODE’s comprehensive genalogical data on the population of Iceland, shows that couples related at the level of third cousins have the greatest number of offspring. For example, for women born between 1800 and 1824, those with a mate related at the level of a third cousin had an average of 4.04 children and 9.17 grandchildren, while those related to their mates as eighth cousins or more distantly had 3.34 children and 7.31 grandchildren. For women born in the period 1925-1949 with mates related at the degree of third cousins, the average number of children and grandchildren were 3.27 and 6.64, compared to 2.45 and 4.86 for those with mates who were eighth cousins or more distantly related.

The findings hold for every 25-year interval studied, beginning with those born in the year 1800 up to the present day. Because of the strength and consistency of the association, even between couples with very subtle differences in kinship, the authors conclude that the effect very likely has a biological basis, one which has yet to be elucidated. The paper, ‘An association between the kinship and fertility of human couples,’ is published online in Science magazine at www.sciencemag.org .

deCODE has access to the amazing Icelandic national family tree, in which most Icelanders who ever lived over the last 1000+ years are enrolled. Genealogy is easier in Iceland because there hasn't been much immigration for the last 1000 years, and because of the surname system: for example, the PR lady who wrote this press release is named Berglind Olafsdottir -- i.e., she is "Olaf's daughter."

Icelanders are of Scandinavian and Celtic descent.

The odds of genetic problems due to inheriting two deleterious recessive genes falls off pretty fast as you move from first cousin outward. I believe at the third cousin marriage level, it's only 1/16th as high as at the first cousin marriage level, but don't quote me when proposing marriage to somebody you met at Great Aunt Meg's 90th birthday party. Still, I'm not sure how much faith I should put in these findings.

I could imagine some non-biochemical reasons for this, such as that 3rd cousins might have tended to marry at younger ages -- in early modern England, as Gregory Clark pointed out in A Farewell to Alms, age of marriage is the main determinant of fertility. Or perhaps healthy people tended to quickly find spouses within their social circles, who tended to be related to them, while sickly people had to wander further afield to find somebody who would marry them.

John Hawks notes an even likelier reason: people who are descended from highly fertile people will have more third cousins to marry. That could be biological or cultural or both.

Some of it could be purely mathematical -- the chance of falling in love with your third cousin depends in part on the number of third cousins you have.

And the number of cousins of any type you have is wildly dependent upon typical family size in your family tree. To simplify genealogical calculations, assume that every person in Family Tree A for the last four generations has had only one child, every person in Family Tree B has had exactly two children, and so forth. Here's what you would face in terms of number of relatives of your own generation:

kids/family	Siblings	1st Cousins	2nd Cousins	3rd Cousins
1	0	0	0	0
2	1	4	16	64
3	2	12	72	432
4	2	24	196	1536
5	4	40	400	4000

Thus, if everybody has had exactly one child for the last four generations, you would have no siblings, no cousins, no 2nd cousins, and no 3rd cousins. At your family reunion, you'd be assured of getting a big slice of the pie, but you'd be pretty lonely.

But if your ancestors had have a nice stable two surviving/breeding children per person, then you would have 1 sibling, 4 cousins, 16 2nd cousins, and 64 3rd cousins.

Yet, if your ancestors averaged five children surviving to reproduce, you'd have 4,000 third cousins!

Of course, humans do not breed in an evolutionarily stable manner. We've taken over this planet by having more than two children each. So, most people are descended, on average, from people who had more surviving children than the average.

It rural Iceland, if you came from "good stock," it might have been hard to avoid marrying your third cousin.

Anyway, I haven't seen the paper yet, so I can't tell if the the deCODE people have been able to deal with these objections. They certainly have a lot of data to work with.

My published articles are archived at iSteve.com -- Steve Sailer