Population genetics: Female migration rate might not be greater than male rate

In a recent paper, which appeared in Nature Genetics, Jason Wilder¹ and colleagues at the University of Arizona, showed that measures of differentiation among populations for male and female lineages are much more similar than commonly thought.

Until now, the generally accepted view was that there is substantially more geographic differentiation in Y-chromosomal genetic variation than mtDNA variation. Since the Y-chromosomal DNA is inherited only by sons from their fathers and conversely mtDNA is only transmitted through the female lineage (males also have mtDNA, but they do not pass it on to their children) these differences have been taken to reflect sex-specific differences in population structure and/or history. There have been a number of explanations for these higher levels of geographic diversity (which Wilder et al estimated as Φ_ST, a haploid analog to the diploid F_ST statistic), namely greater reproductive variability for men relative to women and higher rates of migration for women relative to men. If these new results hold up, the reasons that earlier analyses gave a different answer should be carefully considered and future studies should be adjusted accordingly.

One of the major reasons that perhaps we should favor the conclusions of this new work over previous studies is that Wilder and colleagues have made a fundamental improvement in their study design compared to these previous efforts. Specifically they were able to sequence completely all of the 389 subjects they studied for the Y-chromosome regions under investigation. The Y-chromosome typically has very low levels of genetic diversity, so it has not generally proven feasible to design studies where all subjects are sequenced. The norm has been to genotype markers previously shown to be polymorphic and this fact alone introduces a form of ascertainment bias leading to higher levels of differentiation (Φ_ST). Markers that are discovered in small panels of individuals are likely to be common in one or more of the populations in the discovery panel and even if all populations are represented in this smaller panel, the differentiation will be higher than if all subjects are sequenced. This ascertainment bias effect is even greater if the discovered polymorphisms are used to study populations not represented in the discovery panel.

To avoid such ascertainment bias, Wilder et al hit upon the ingenious solution of focusing their sequencing on a special selection of recently inserted human retrotransposons, Y family Alu elements, which are hypermutable by virtue of a higher proportion of CpG dinucleotides. In short, this approach allowed them to use complete sequencing to collect enough data to test whether there was more or less geographic differentiation in the Y-chromosome compared to the mtDNA, for which a segment was also sequenced in each of the 389 subjects. The subjects under investigation were sampled from 10 geographically diverse populations; four African, two Asian, two European, and two Oceanic. With such a diverse selection of populations the authors were able to test for differences between these two loci, and the sexes they represent, across a long span of evolutionary time and geographic space, tens of thousands of years and thousands of miles. Overall they found that the levels of mtDNA differentiation (Φ_ST=0.382) were greater than the levels of differentiation on the Y-chromosome (Φ_ST=0.334).

One of the reasons for all of the interest in these patterns of mtDNA and Y-chromosomal diversity is because they can provide clues to the important demographic parameters of populations through time. Have women generally moved to the house and village of the men who fathered their children? Has variability in reproductive success been much higher for men compared to women, with some men having many children and others few or none? Has the generation time for women been substantially shorter, with women becoming mothers younger than when men become fathers (see Helgason et al²)?

As each of these factors, migration rate, reproductive variability, and generation time, would affect mtDNA and Y-chromosomal relative differentiation levels, it is clearly a complex issue. Wilder and colleagues have shown that ascertainment bias, caused by genotyping polymorphic sites and not sequencing each subject, can have an important effect on studies such as these. It is critically important to understand the types of experimental and statistical biases that can affect results, and to take these into account. These authors are to be commended on this account and workers in evolutionary genomics should heed this warning, as it is unlikely that these are the only statistics to be affected by such ascertainment bias.

Nonetheless, given the number of demographic forces that affect patterns of genetic differentiation, the implications of these results should be interpreted cautiously. The authors conclude, ‘that the role of female migration is no more important than that of males at the continental- and global scales.’ And, while this explanation is clearly consistent with their data, if generation times are on average shorter for females (ie mothers are generally younger than the babies fathers), which would imply a smaller global effective population size for females compared to males, then greater female migration or greater male reproductive variability might also be consistent with this pattern▪