Genghis Khan: Super-Stud

Posted by Ed

This morning Arts and Letters Daily linked to a fascinating Independent article about Genghis Khan. The article is a charming read and I'd strongly recommend it.

What struck me most wasn't the review's discussion of medieval Mongolian carnage, its suggestion that the "Mongol Peace" allowed massive cultural cross-fertilization, or its use of atrocious puns like "steppes toward the future." Instead, the line that really sparked my interest was the following: "If recent research in Oxford's biochemistry department is to be believed, [Genghis Khan] was one of history's most philoprogenitive studs, with 16 million living descendents." The author of this article seems to think that it's surprising that a medieval Mongol emperor has so many descenfants alive today, but if anything, I'd be surprised if so few descendants of Genghis still walked the earth.

Consider this 2002 Atlantic Monthly article, which discusses the research of Yale statistician Joseph Chang:

In a 1999 paper titled "Recent Common Ancestors of All Present-Day Individuals," Chang showed how to reconcile the potentially huge number of our ancestors with the quantities of people who actually lived in the past. His model is a mathematical proof that relies on such abstractions as Poisson distributions and Markov chains, but it can readily be applied to the real world. Under the conditions laid out in his paper, the most recent common ancestor of every European today (except for recent immigrants to the Continent) was someone who lived in Europe in the surprisingly recent past—only about 600 years ago. In other words, all Europeans alive today have among their ancestors the same man or woman who lived around 1400. Before that date, according to Chang's model, the number of ancestors common to all Europeans today increased, until, about a thousand years ago, a peculiar situation prevailed: 20 percent of the adult Europeans alive in 1000 would turn out to be the ancestors of no one living today (that is, they had no children or all their descendants eventually died childless); each of the remaining 80 percent would turn out to be a direct ancestor of every European living today.

Chang's research (you can see an abstract of his article here) has other interesting consequences: given the prevalence of intermarriage in the Americas, almost everyone in the New World must be descended from English royalty--and everyone of European ancestry must descend from Muhammad. Moreover, The Atlantic writes,

Chang's model has even more dramatic implications. Because people are always migrating from continent to continent, networks of descent quickly interconnect. This means that the most recent common ancestor of all six billion people on earth today probably lived just a couple of thousand years ago. And not long before that the majority of the people on the planet were the direct ancestors of everyone alive today. Confucius, Nefertiti, and just about any other ancient historical figure who was even moderately prolific must today be counted among everyone's ancestors.

Which brings us back to the Oxford research I alluded to above. I was sufficiently curious about this research to do a quick web search, and found that the research (conducted by Christopher Tyler-Smith) may have been quoted incorrectly: it actually found that 16 million men in Central Asia were descended from Genghis Khan, which suggests that "one in every 200 men alive today is a male-line relative of the Mongol leader, who lived between 1162 and 1227." (I got this information from an Oxford University newsletter, and here's a Science News article on the research.)

Later on, I may read through some of the actual research, to get a better sense of what Chang and Tyler-Smith really argue. (One of the articles I've linked to says that there are 16 million Genghis Khan descendants in Asia, not Central Asia or the world...) I wouldn't be at all surprised if I've gotten some of Chang's and Tyler-Smith's findings wrong, since this is outside my field of research, I'm reading popular accounts of their work, and I haven't given either question much thought. It's fascinating stuff, though--and with consequences for how we view genealogy and history.

Update: One of my informants has sent me a link to Chang's paper (and some discussion of it.) He also provides the following commentary: "The interesting thing about it is that he gives a very precise analysis and for example computes exactly what fraction of people in the distinct past are common ancestors. However, the random mating assumption in the model is grossly unrealistic (and it's not the only simplification; he also assumes a fixed population size). Chang never intended it to predict anything precise about the real world, just to illustrate that one can precisely analyze a toy model and that the answers are quite different from the sorts of one-parent models used in studying mitochondrial Eve.

"The Atlantic Monthly article makes some pretty implausible claims. I think it's almost certain true that humanity's most recent common ancestor is more recent than the couple hundred thousand years to mitchondrial Eve. However, I can't imagine that the couple thousand year figure from the article is accurate. I suspect it's true for most people of European descent, but I'd be surprised if the time for the entire world is under ten thousand years."

The author of The Atlantic article responds to the point that Chang's model assumes a fixed population size in the letters section of a subsequent issue (scroll down to "Common Ancestors.") My sense is that my informant and the letter-writer are correct that Chang's model wasn't intended to lead to precise predictions about the real world, and I have no idea what consequences it has for our understanding of how many people are descended from Genghis Khan. (Maybe I'll spend some more time reading these papers...) Even so, it's fascinating stuff.

Posted by Ed at March 14, 2004 02:09 PM