Mater of fact, pater of opinion

We're all familiar with patrilineal descents. The genealogies of biblical patriarchs, tribe and clan names from Scotland to Maharashtra to Somalia to China, rabbinic dynasties, royal families, and modern surnames illustrate the extent to which kinship, identity, and inheritance have been based on the relationship between father and child, particularly among the agricultural and pastoral peoples of Eurasia and Africa. Those tracing their family genealogy also tend to concentrate on surname lines (if not their own, their children's, their mother's, or their grandmothers') for two related reasons. First, their own surname and those of their close kin are more psychologically meaningful to people than names they have never heard, so these lines are an obvious place to focus research. Second, since people in previous generations thought the same way, male lines are easier to trace.

Genetic genealogy has added another advantage to the male line. The Y chromosome of two men with the same surname (men, because women don't have a Y chromosome) who don't know if they have a common male ancestor can be tested to find out the probability that they do. The Y chromosome is copied pretty much faithfully from father to son, but mutations that occur from time to time at a regular rate make it possible to tell approximately when the man who carried a version ancestral to both lived. Because people tend to reproduce with people who live near them, it is also possible to tell approximately where he lived (though there will be much caution as to the limits of this in this blog).

There is a problem though. As Mark Twain should have said, "Maternity is a matter of fact; paternity is a matter of opinion." Where p is the number of conceptions that had to occur for a common ancestor to produce two descendants and c is the rate of non-paternity (where the purported biological father is not the actual biological father), the likelihood that the two descendants will share a genetic patrilineal descent is (1-c)p. Even if c is very small, and the lowest estimate I've ever seen for non-paternity among men who have high paternity confidence is 1.9%, the likelihood drops below 50% for a common ancestor who lived 500 years ago, and the highest the non-paternity rate can be in order for two genealogically-related men with a common ancestor who lived 200 years ago to have a >50% probability of sharing the same Y chromosome is 4.83%. This is low enough to be very useful, and, indeed, common patrilineal ancestors of people with the same surname or clan identity have been found in the very distant past, but a warning is necessary. Supposedly, the common ancestor of men who carry the Cohen modal haplotype lived something like 3500 years ago, and about 50% of Kohanim carry it, necessitating a non-paternity rate of only 0.54% per generation. Whether this says more about the effectiveness of Jewish sexual law or the errors in Skorecki, et al.'s mutation rate calculation is open to question.

I find matrilineal descent more interesting. Everyone has mitochondria, and mitchondria are passed only from mother to child (well, usually). Mitochondrial DNA makes it possible to trace uterine, or matrilineal descent, which was much more commonly used as an identity structure in the past, but which is currently limited to clan groups among peoples who had been until recently relatively isolated from the Eastern Hemipshere global economy and in traces in the rituals and religious laws of many groups which are in most others ways patrilineal, such as in Judaism and in South India.

Matrilineal descents are particularly fun within a large genealogical database. The largest that exist are primarily data on the royal and noble families of Europe, and looking at them is instructive both for studies of kinship and of European history. Patrilineally, European dynasties, just like those in the rest of the world, tended to be drawn from the lineage of the more powerful military leaders during a key period of historical development. For the European dynasties that were in charge of states from the early medieval to the modern period, this was in the three or four centuries following the fall of the Western Roman Empire in the 5th century, and then even more rapidly with the expansion of feudal structures from the Carolingian Empire in the 8th through 10th centuries. Oldenburg, Capets/Bourbon/Aviz, Habsburg, Wittelsbach, Piast, all date to this time. What's interesting is that royal matrilineal descents also tend to trace to only a few 10th century women, perhaps even fewer than the male lines. After all, it's more difficult to get an unbroken line of daughters than an unbroken line of sons (men have more children). Also, because of the genetic tests used to identify the bodies of Nicholas and Alexandra in their grave in Yekaterinburg, we know what mitochondrial haplogroups they and their matrilineal relatives belonged to:

It's now been questioned whether or not the researchers actually did find Nicholas and Alexandra (the Russian government doesn't want to hear it though), but the known identity relatives they tested as a control make it still possible to trace those mitchondrial lineages. Prince Philip and Alix, both Haplogroup H, are descended through the female line from Queen Victoria, and through her to a 13th century Spanish noblewoman nicknamed La Palomilla. The rest of the current British Royal Family are not in this line, since the Queen, the Prince of Wales, and all the rest are descended in the female line from English gentry and commoners, not from royal families, but other descendants of La Palomilla include Marie de' Medici, Charles II, James II, and William III of England, Louis XIV of France, and Felipe III, Carlos II, and Felipe V of Spain. Marie Antoinette was also tested as Haplogroup H, but her earliest known matrilineal ancestor (and that of Pedro II of Brazil, Stanisław August Poniatowski, and many others) was a 12th century German noblewoman named Bertha von Putelendorf. Through the Tsar and his brother, Haplogroup T has another royal lineage that includes the Electress Sophia, Charles I of England, George V of Great Britain, and Charles X of Sweden.

The one I find most interesting though is not descended from a Western European source, but instead from Euphrosyne Dukaina Kamatera, the wife of Roman Emperor Alexios III and great-grandmother of Michael VIII, the Emperor who took Constantinople back from the Crusaders. One of her great-granddaughters married the Saljuq Sultan of Rum at Konya, one of her granddaughters, Maria Laskarina, married Béla IV of Hungary. The latter is a matrilineal ancestor of George IV and William IV of Great Britan, Emperor Maxmilian of Mexico, Kazimierz the Great of Poland, Alexander III of Russia, and Wenceslas II of Bohemia.

And now for the question that genealogists so rarely ask (sorry!): what's the point? That Charles I's head bled Haplogroup T blood all over his platform at Whitehall tells us nothing about the English Civil War. What matters is that royal genealogy databases are our biggest, and analyzing them in detail (even though highly mobile nobles were an exceptional group) tells us about how rapidly descent can diffuse geographically. Europeans who are Haplogroup O (most common in China) or European Americans who are descended from Genghis Khan show up all the time. They are not the norm, but neither are they so rare as many would like to think. It is even easier for descent to pass through all lines than just through males and females, and there are enough of documented entrance points of Arab, African, and Central Asian lines into royal databases to be worthy of looking at to provide real world examples that back up the mathematics that show our species' most recent common ancestors to be very recent indeed.

For a later post:

  • When did our most recent common ancestors live?
  • If it was as recent as mathematicians Chang and Rohde calculate (3rd to 1st millennium BC), why does it matter even though none of us likely carry a single gene from those ancestors? (short hint: kinship and identity do not correlate to genetics)

Links: Ysearch, Misattributed paternity rates and non-paternity rates, Roglo database, Famous DNA,
Reading recommendations: Sarah Franklin and Susan McKinnon, eds.. Relative values: reconfiguring kinship studies. Durham, NC: Duke University press, 2001; Patricia Buckley Ebrey and James L. Watson, eds. Kinship organization in late imperial China: 1000-1940. Berkeley: University of California Press, 1986.