March 16, 2006

The good and bad of population genetics in Slate

The Good: John Hawks on DNA genealogy tests

The Bad: Science journalist Steve Olson on "Why We're All Jesus' Children: Go back a few millenniums, and we've all got the same ancestors." The quality of Olson's thinking can be guessed from his decision to aim his article at fans of The Da Vinci Code.

I reviewed Olson's National Book Award-nominated Mapping Human History, a popularization of the work of population geneticists like L.L. Cavalli-Sforza, on

In his book, Olson stops every few pages to tell you that there are no races that have been absolutely isolated genetically since the beginning of time because—you will be shocked, shocked to learn this—humans have been known to outbreed. (The reality of course is that for any human racial group, the inbreeding glass is both part empty and part full.) This makes Mapping Human History rather like a geology book that repeatedly admonishes the reader that the Earth is not flat...

Another curious feature that Olson's book shares with many other contemporary writings about population genetics is the author's apparent longing for the abolition of his own subject matter via universal random interbreeding. Although animal and plant biodiversity is routinely celebrated as a supreme good, the conclusions of books on human biodiversity tend to treat it as a temporary evil that will soon be gone, and good riddance to it. It's as if that geology textbook ended with an ode to the blessed day when the Earth will plunge into the Sun, thus happily eliminating the need for a science of geology.

Olson's new article is about his theory that everybody alive today is descended from everybody who ever lived who has descendents alive today. Okay, maybe, but the more functionally important question is where most of your ancestors came from. And for that, you need to think about the degree of inbreeding, which Olson refuses to mention explicitly.

He writes:

Imagine that you could identify all of your great-great-great-great- … grandparents 20 generations back—from about the time Columbus stepped ashore in the New World. (You would never be able to, of course, because no paper records connect you to virtually any of those people, but pretend that God handed you a perfect genealogical record.) Assuming typical human mating patterns, your direct ancestors 20 generations ago consisted of somewhere between 600,000 and 1,000,000 different people. Taking the lower figure, perhaps 480,000 of the ancestors of the average African-American were living in Africa in the year 1492, and approximately 120,000 were living in Europe, the Americas, and Asia. For the average European-American, more than a half-million ancestors were living in Europe, with the rest scattered through Africa, Asia, and the Americas.

Let's do the math. Ten generations back (about 250 years at 25 years per generation), your family tree has 1,024 open slots for ancestors (2 to the 10th power). Twenty generations back (500 years), your family tree has 1024 times 1024 or 1,048,576 or one meg of open slots.

Olson claims, " your direct ancestors 20 generations ago consisted of somewhere between 600,000 and 1,000,000 different people," but the higher end figure is obviously absurd. Your ancestors would have had to have been for each of the last 20 generations as outbreeding as Tiger Wood's immediate ancestors for virtually all the 1,048,576 slots in your family tree to be filled by different individuals. Olson's lower bound of 600,000 also seems absurdly high too.

Thirty generations ago, your family tree had one gig of open slots. Let's round these down from now on. So 40 generations or back around 1000 AD, you had one trillion open slots.

You can see where this is going. Back in about the year 1 AD, eighty generations ago, you had roughly 1,000,000,000,000,000,000,000,000 openings for ancestors. There weren't quite that many people alive back then, so the inbreeding coefficient (one minus the number of unique individuals in your family tree at that point divided by the number of slots to be filled) mathematically had to be over 99.99%.

So, inbreeding matters.

Now, it's theoretically possible that in your family tree just 120 generations ago, like Olson claimed, you number ever single living person in 1000 BC who has living descendents today. I doubt that's true -- there was just too much isolation of Andaman Islanders, Tasmanians, Tierra del Fuegans, New Guinea Highlanders, and so forth. But, overall, even if it were true, it's just a talking point. What counts is the distribution of genes, and those are driven by inbreeding.

And, as I've been pointing out for years, inbreeding is what turns an extended family into a racial group: a racial group is a partly inbred extended family.

My published articles are archived at -- Steve Sailer

No comments: