Wednesday, August 17, 2005

Cousin be perty, part n   posted by Razib @ 8/17/2005 09:24:00 PM

When I was in Bangladesh last year an ignorant distant cousin of mine was defending her choice of marrying another cousin (not someone I was related to) because she stated that unless "blood groups are the same" it isn't really much of a problem.1 Of course, that's bullshit. I happened to have brought Principles of Population Genetics with me and opened up to figure 13 in the chapter on inbreeding and showed her the graph which plotted "Frequency of deleterious recessive allele (q)" vs. "Percent of affected children whose parents were first cousins." The qualitative observation was clear: though few offspring whose parents were first cousins had recessive disease X, an enormous percentage of individuals in a breeding population who had recessive disease X had parents who were first cousins. The equation used to plot this graph is (K being the dependent variable):

K = c(1 + 15q)/(c + 16q - cq)

The proportion of first cousin matings being c within the population, and q being the frequency of the recessive allele in question.

Here are some statistics from an old post:

Condition - % of affected children whose parents were first cousins
Total color blindness - 15
Albinism - 21
Xeroderma Pigmentosum - 23
Ichthyosis Congenita - 35
Tay Sachs - 40

(the numbers are for the United States where c was assumed to be ~1%)

So, Lei is correct that when she states that "children of non-related couples have a 2-3% risk of birth defects, as opposed to first cousins having a 4-6% risk," but looking at this issue on a population wide level one can see that the cost vs. benefit analysis might lead to being less sanguine. Now, Lei brings up that genetic testing can obviate this issue, but remember the easiest implementation of the solution in the case of a positive match: abortion (I don't know how one would test sperm and not damage it, and it seems removing eggs and testing them to reimplant them would also be laborious and expensive). This isn't a palatable option for many couples. I remember watching a documentary where they highlighted the case of a Roman Catholic couple who had had 2 children, one with cystic fibrosis, and they decided to risk it and try for another child in the hopes that they would be lucky. They made it clear that they opposed abortion on principle. The documentary ended before they found out their results for the third pregnancy.

Additionally, Lei is talking specifically in relation to the situation of Arabs in Israel. I happen to have a friend who is an Israeli Arab and he tells me that it is common not just to marry cousins, but to marry within a clan. Lei notes that the heightened risk from second cousin marriages is minimal, but my concern is that these Arab clans are extremely inbred to begin with.

The equation for inbreeding coefficient is:

FI = sum over all common ancestors[(1/2)i * (1 + FA)]2

Here, you are summing over the paths to each common ancestor, FA, with i being the number of individuals in each path (obviously inbreeding via a common great-great-great grandparent is weighted far less than that via a grandparent). In the United States if you marry a cousin (outside of really rural areas) it is not likely that aside from those implied by the common grandparents you have many recent common ancestors. In other words, families that your parents do not share in common are not likely to be related. In the situation above with Arab clans, this may not be so, there could be a tight and interlinked network of common ancestries up all possible branches of the lineage.3

In Saudi Arabia this is a big medical issue. The reason you don't hear about it much from other Muslim countries is that I suspect that unlike Saudi Arabia they don't have a comprehensive health care system (Israel does, so that's probably why it is making the news). But the point I'm trying to make is that using American first cousin marriages as a proxy is probably not good anyway because I would not be surprised if the inbreeding coefficients were way higher in many parts of the Arab world than one would expect based simply on the reported familial relationship.

Related: Consanguinity prevents Middle Eastern political development.

Updated: offers a thorough PDF that has a nation by nation (broken up by studies and sample ethnic group or location) breakdown. Be no more ignorant about cousin-love the world over!

Here is a map for the visually inclined:

1 - Though consanguinity is not as common in Bangladesh from what I can gather as it is in the Arab world, or even Pakistan, it is not unknown. My distant cousin was a topic of conversation precisely because for some reason (never elucidated) my family tends to avoid these sort of unions, and in fact, my own personal background is not one of similarity but diversity (ie; my paternal grandfather was of admixed South Asian and West Asian origin, his wife was from a recently converted Hindu family, etc.). I suspect that empirically over the generations my family found that expanding social networks rather than firming up the ones they already had was a successful strategy, and, frankly their relative prosperity might have meant they were secure enough to not take what seems to be perceived by many as the "safe" strategy and marry within the family. I have come to find out recently that there is a long history of relative geographic spread in terms of spousal choice that I suspect was not typical in most of Bengal in the early 20th century. Even with those who remain in Bangladesh I note that one of my uncles married a women of Bihari (Urdu speaking migrants from India pre-1947) origin without any comment (I only know because I heard her speaking Urdu on the phone with her mother and was confused for a moment until another relative told me what her background was).

2 - If i starts getting big, well, no point, right? You judge when to stop counting the common ancestors.

3 - A page that eludicates the coefficient with some diagrams.