Worthless Scientific Reporting

Here's what you get when you have the scientifically illiterate reportng the news: the headline blares "Most Ashkenazi Jews From Four Women", and then goes on to disclose the following:

Some 3.5 million of today's Ashkenazi Jews about 40 percent of the total Ashkenazi population are descended from just four women, a genetic study indicates.

Those women apparently lived somewhere in Europe within the last 2,000 years, but not necessarily in the same place or even the same century, said lead author Dr. Doron Behar of the Rambam Medical Center in Haifa, Israel.

It's bad enough that "40 percent" somehow gets translated into "most", but just how newsworthy is it to learn that some researchers have been able to find 4 women who are ancestors to said 40 percent of Ashkenazis? While it might seem like some sort of awe-inspiring revelation to the uninformed, the fact of the matter is that it is just about as unsurprising a discovery as one could possibly make: take any group of geographically promixal individuals selected in any fashion you please and you'll find that most of them share ancestors in common within the last 2,000 years as well. In fact, one can do better and say, for instance that virtually Europeans are descended from Charlemagne, or that nearly every single human being alive today is a descendant of Confucius, Ramses II, or some other person who lived far back enough in the past: the point is that beyond a certain limit, either people who were alive at a point in time are ancestors to none of us or to all of us - a straightforward consequence* of the mathematics of the coalescent process - and all these researchers have really done is located 4 nodes out of several million possibilities for which this process happens not yet to have reached completion.

That any large group of humans shares common ancestry a few thousand years in the past is not news, as we all do, and what separates us from each other is the size of the contribution made to our genes by the same shared pool of recent ancestors we all share in common. To throw out screeching copy about "most" [sic] Ashkenazi Jews sharing 4 maternal ancestors all of 2,000 years ago, as if it were some miraculous or anomalous finding, is to do no more provide pseudoscientific fodder for all sorts of cranks.

*See this paper by Joseph Chang and this one by Douglas Rohde for details.

Comments

[Ken Hirsch]

Your point is almost correct, but the studies you reference are considering bilineal descent, whereas mitochondrial or Y chromosome studies just track matrilineal or patrilineal descent.

The math is quite different. The number of ancestors by generations back is: 1,1,1,1,... instead of 2,4,8,16,32,...

Surnames are similar to the Y chromosomes, so you can compare the dispersion of surnames. In Korea, 3 surnames account for 45% of the population (similar to the Ashkenazi mitochondria), but in England, the top 10 surnames accounted for only 6.3% of the population in the 1851 census.

So, there's nothing tautological about it at all.

[Andrew]

"Surnames are similar to the Y chromosomes, so you can compare the dispersion of surnames. In Korea, 3 surnames account for 45% of the population (similar to the Ashkenazi mitochondria)"

Eh? What stops two unrelated men from adopting the same surname? Surely we don't expect that the reason Smith is the most common English surname is that Mr Smith's descendents were very successful at producing male offspring, rather than that being a blacksmith was a pretty common occupation to be named after back in the Middle Ages?

[Abiola]

"the studies you reference are considering bilineal descent, whereas mitochondrial or Y chromosome studies just track matrilineal or patrilineal descent."

While I appreciate what you're getting at here, the following statements of yours actually serve to undermine the notion that this result is as significant as all that:

"In Korea, 3 surnames account for 45% of the population (similar to the Ashkenazi mitochondria)"

This hardly supports the idea that what we're seeing with the Ashkenazim is somehow newsworthy. Given a Markov process of the kind we see in relatively closed systems such as with surnames in Korea or mtDNA in Ashkenazim, 2,000 years is *plenty* of time for most lines to go extinct: assuming a 98% chance of propagation per generation and a generation time of 25 years, after 2,000 years only 44% of extant mtDNA lines should survive, while using a 95% survival rate reduces that number to less than 13%.

"in England, the top 10 surnames accounted for only 6.3% of the population in the 1851 census."

This has more to do with a later date of surname adoption, a longer history of immigration - it is a known fact that the English are the most outbred of European populations - and an irregular orthography which encourages variations in surname spellings than it does with England being somehow a more typical case than Korea.

"So, there's nothing tautological about it at all."

I never said the report was "tautological", I said it wasn't newsworthy, which I still maintain it isn't. Nothing I've seen anywhere on ancestral mtDNA distributions in relatively reproductively isolated populations leads me to believe that finding 4 women who account for 40% of extant mtDNA lines is particularly worth paying attention to. Going by the abstract available here

http://www.journals.uchicago.edu/AJHG/journal/issues/v78n3/43026/brief/43026.abstract.html

the one thing about this research which is even mildly interesting is precisely that which is underemphasized: that these mtDNA lines are only found in low frequencies amongst non-Ashkenazim, and are possibly of Middle Eastern origin, which is hardly all that shocking a revelation if true.

[David B]

Abiola should find this more interesting. If these findings prove that Ashkenazi Jews have gone through a severe population bottleneck in the last 1000 years or so, this tends to weaken the Cochrane-Harpending theory of Ashkenazi IQ.