Double the pleasure, double the fun

The NY Times science question this week is, "Do twins always skip a generation?" This one is right up my alley, so let's see what Anahad O'Connor has to say:

Scientists have known for some time that there is a gene that can predispose women to hyperovulation, or releasing two or more eggs in a single menstrual cycle. When both eggs are fertilized, the resulting siblings are fraternal twins. Because this gene can be passed on, the tendency to have fraternal twins can in fact run in families.
Identical twins, on the other hand, result from one fertilized egg randomly splitting in two, creating two siblings with identical DNA. Because there is no known gene that influences this process, it is considered a mere coincidence when one extended family has multiple sets of identical twins.

The latter part is my category: identical (monozygotic) twins are generally considered not to run in families. Lichtenstein et al. (1996) found a slight increase in the rate of monozygotic twinning among women who were MZ twins themselves, suggesting some genetic influence, but this effect does not seem to have been confirmed elsewhere. The rate of MZ twinning in humans is very nearly the same in all human populations, and is very similar to the rate in chimpanzees (between 0.3% and 0.4% in humans, Pison and D'Addato 2006, 0.43% in chimpanzees, Ely et al. 2006).

But let's look at that "dominant gene" explanation for dizygotic twins a little more closely. Meulemans and colleagues (1996) looked at around 1400 pedigrees including twins, and discovered that dizygotic twinning appeared to be inherited, with a pattern consistent with a single dominant locus of large effect.

The question of whether dizygotic (DZ) twinning is inherited is a century old. In 1902, Weinberg (1902) [Yes, this is the Weinberg of Hardy-Weinberg fame] discovered higher DZ twinning frequencies among the offspring of female relatives of DZ twins. Wyshak and White (1965) observed DZ twinning rates of 1.45% for the offspring of a female DZ twin and of 1.65% for their female relatives. These frequencies exceeded the rate of 0.69% found among the offspring of male DZ twins. Bulmer (1970) showed that the relative risk of bearing DZ twins for daughters/mother and the sisters of women with DZ twins were 1.8 and 2.6 times higher than the general population, respectively. He postulated that a recessive gene with a low penetrance and a gene frequency of 50% caused the birth of spontaneous DZ twins. The underlying biological mechanism for DZ twinning is multiple ovulation (Milham, 1964). Animal models have confirmed a genetic influence for multiple ovulation. Montgomery et al. (1993) localized an autosomal codominant gene (FecB) that increases the ovulation rate in Booroola sheep to a region homologous to human chromosome 4. The existence of an X-linked gene, which influences multiple ovulation and polyzygous multiple births, was demonstrated through segregation analysis in Romney sheep (Davis et al. 1991) (Meulemans et al. 1996:258).

Later studies examined the chromosome 4 sheep twinning locus and found no effect in humans. Likewise, there is no X-linked factor linked to twinning in humans, and the follicle stimulating hormone receptor (FSHR) does not have polymorphisms linked to twinning.

Still, Meulemans et al. (1996) found that the model of a single dominant allele was the best fit to their pedigree data. They noted that the DZ twinning rate is a very imperfect reflection of the underlying gene (in other words, the allele has low penetrance on twinning) -- the real phenotype of interest is the rate of multiple ovulations, and DZ twinning is a conveniently observable consequence. They could conclude that the effect was consistent with an allele of dominant effect, with a population frequency around 3 percent, and a relatively low penetrance on twinning. But no locus has yet been identified.

Unlike MZ twinning, it is fairly widely known that different human populations vary by as much as an order of magnitude in the rate of DZ twinning. West Africans have a lot of twins: Yoruba in Nigeria have the highest twinning rate, at 53 twin births per 1000 live births (Knox and Morley 1960), the rate among other Nigerians and in Ghana is around 30/1000 (Mosuro et al. 2001), and the rate in Africa as a whole is 20/1000 (Pison 2000). Europeans have fewer twins, now around 10-20 twin births per 1000 live births, and East Asians fewer still -- averaging only around 7/1000 (Pison 2000). The rate of MZ twinning is virtually the same in all these populations: the entire difference comes from the DZ twinning rate. Does this mean they have different frequencies of the presumed gene?

There are environmental effects that we might consider.

The rate of dizygotic twinning and other multiple births has gone up lately (Pison and D'Addato 2006). Most people probably know that a lot of this recent change is attributable to reproductive assistance technologies like ovulation-inducing drugs and IVF. Additionally, an increase in the average maternal age has increased the rate of twinning -- mothers between 30 and 40 have twins at nearly double the rate of women in their early twenties. But the rate has fluctuated during the last 150 years in Europe, with a major increase beginning at around 1900 and peaking at 1920, then falling again. This pattern seems concordant with the history of the average childbearing age, which peaked and fell at the same times (Pison 2000), particularly during the 1960's as the fertility rate declined, so that women who had twins early tended to choose not to have further pregnancies. These dynamics are very explanatory for changes in twinning rates in Europe over time, but they cannot account for the pattern of variation among populations.

Steinman (2006) ties changes in the twinning rate to diet:

Cows can be selectively crossbred to enhance their spontaneous twinning rate. Concentrations in the serum of insulin-like growth factor (IGF) in such cows are typically 15-2-fold higher than average. Genomic loci that affect the twinning rate in Norwegian cattle have been detected in close proximity to the IGF gene on chromosome 5.6 In human beings, an autosomal dominant gene related to twinning has been identified.
During periods of malnutrition -- e.g., during World War 2 -- the twinning rates decrease, perhaps because of a fall in maternal concentrations of IGF. In vegans, in whom concentrations of IGF are typically 13% lower than in the general population, the twinning rate is a fifth of that in vegetarians and omnivores. A study in 15 European countries showed a direct correlation between average milk consumption and national dizygotic twinning rate.

Hmmm...does a body good, does it? Still, diet does not itself explain the interpopulation differences. But maybe part of it:

Black people have the highest and oriental people the lowest overall twinning rates in specific ethnic groups; the mean concentration of serum IGF follows the same trend. For unknown reasons, consumption of yams by members of the Yoruba tribe in Nigeria is believed to be the cause of their unusually high twinning rate; when they move from the countryside into the city, with a corresponding change in diet, their twinning rate drops significantly. Similarly, when Japanese people relocate to California, their twinning rate doubles.

Interestingly, there's also a substantial seasonal variation in the twinning rate. Women who give birth from March to June have historically been more likely to have multiples (Fellman and Eriksson 1999). This parallels a general increase in births during this time period, but exceeds it -- not only are women more likely to have babies during those months; they are more likely to have twins. This suggests some kind of environmental effect relating to seasonality, either dietary, activity-related, or sunlight-mediated, during the late summer or early autumn.


Ely JJ, Frels WI, Howell S, Izard MK, Keeling ME, Lee DR. 2006. Twinning and heteropaternity in chimpanzees (Pan troglodytes). Am J Phys Anthropol 130:96-102. doi:10.1002/ajpa.20310

Fellman J, Eriksson AW. 1999. Statistical analysis of the seasonal variation in the twinning rate. Twin Res 2:22-29. doi:10.1375/twin.2.1.22

Steinman G. 2006. Can the chance of having twins be modified by diet? Lancet 367:1461-1462. doi:10.1016/S0140-6736(06)68623-6

Knox G, Morley D 1960. Twinning in Yoruba women. J Obstet Gynaecol Br Emp 67:981-984

Lichtenstein P, Olausson PO, Kallen AJB. 1996. Twin births to mothers who are twins: a registry-based study. Br Med J 312:879-881.

Mosuro AA, Agyapong AN, Opuko-Fofie M, Deen S. 2001. Twinning rates in Ghana. Twin Res 4:238-241. doi:10.1375/twin.4.4.238

Pison G, D'Addato AV. 2006. Frequency of twin births in developed countries. Twin Res Hum Genet 9:250-259. doi:10.1375/twin.9.2.250