Monday, September 21, 2015

The Out of Africa Founding Population Size And Gender Imbalances

It takes a village with five men for every woman to settle the world (except Africa).  Specifically, about 330 men and 65 women of reproductive age.  So says another notable ASHG 2015 paper:

M. H. Quiver, et al., "Selective constraint and sex-biased demography of human populations from X chromosome-autosome comparisons"
Because the number of X chromosomes differs for men and women, comparisons between sex-linked and autosomal genetic loci reveal sex-biased patterns of human demography. Using 44 high-coverage whole genomes from a diverse global set of 11 human populations we quantified the strength of selective constraint on different chromosomes, found evidence of sex-biased colonization, and determined whether recent migrations are matrilocal or patrilocal. Relative amounts of genic and intergenic diversity were similar across all studied populations regardless of subsistence pattern or geography. The strength of selective constraint on genes was greater for X-linked loci compared to autosomal loci – a pattern that is consistent with selection against deleterious recessive alleles. The ratio of X chromosome to autosome diversity (Q) was greater than the null expectation of 0.75 for African populations and less than 0.75 for non-African populations, with lower values of Q for populations located farther from Africa. 
This pattern is consistent with a male-biased serial founder effect model, and computer simulations suggest a plausible out-of-Africa bottleneck size of 320-340 males and 60-70 females. 
Using PSMC, we found evidence of large historic population sizes for West African Pygmies, but not Hadza or Sandawe populations. Genetic distances revealed female-biased gene flow between Hadza and Sandawe hunter-gatherers, between Maasai pastoralists and African farmers, and between Chinese and Japanese populations. We found evidence of male-biased gene flow between African farmers and hunter-gatherers, and between different African farmer populations. This calls into question the idea that patrilocality is coupled with the emergence of agriculture.
Another paper from the conference using the same methods and researching the same issues has a less bold abstract, but appears to each essentially the same conclusions.

Arbiza and Keinan,"The relative effective population size of chromosome X and the autosomes along distinct branches of the human population tree."
In recent years, many studies have focused on the effective population size of chromosome X relative to the autosomes. This comparison can be useful to reveal past demographic processes, differences in the histories of males and females, and the action of natural selection. We have recently shown how the ratio of nucleotide diversity between the two (X-to-Autosome ratio; X/A), when compared between pairs of populations (relative X/A), can be used to uncover sex-biased processes in human history. While this strategy serves to alleviate the response of genetic diversity to the influence of events in a time range that largely predates the split of the studied populations, a different and more natural approach to capture recent changes occurring after populations split can be formulated based on the differentiation of allele frequencies between populations, as commonly summarized by the F ST statistic. Here, we consider population differentiation in humans, and extend beyond simple pairwise comparisons, using allele frequency differences across several populations to learn about the ratio of X-to-autosomal effective population size along distinct branches in the tree of human populations. We then test these for differences from the expectation of equal female-to-male breeding ratios, as well as differences between different branches. Using coalescent simulations of a variety of previously published human demographic models, we show that our approach is able to capture the ratio of interest and is more accurate than estimates based only on pairwise F ST across all pairs of populations. We then turn to the latest data from the 1000 Genomes Project, controlling for the effect of uncertainty associated with low coverage sequencing, as well as the influence of linked selection (background selection or hitchhiking), all of which differentially affect the X chromosome and the autosomes. 
Estimating the X-to-autosomal effective population size ratio for branches leading to different 1000 Genomes populations, as well as for internal branches in the population tree, points to a higher female effective population size in African-specific population history, but not in non-Africans. More interestingly, we localize previously-debated observations to a significant increase in male effective population size on the branch leading to all non-African populations, suggesting male-biased processes associated to the Out-of-Africa event.
Yet another ASHG 2015 paper looks at similar issues with Y-DNA:

F. L. Mendez, et al., "Estimation of growth rates for populations and haplogroups using full Y chromosome sequences."
Evolutionary processes affecting a population influence gene genealogies across the genome. Coalescent theory provides the mathematical framework to connect realized genealogies to the underlying evolutionary processes. However, in most cases, information about the genealogies is obtained only indirectly through the observation of genetic variation. Therefore, in general, very limited information about any individual locus is available. As the longest non-recombining portion of the human genome, the Y chromosome accumulates mutations relatively quickly. When large amounts of sequence are used, the Y chromosome provides an unparalleled ability to resolve the structure and coalescence times of its genealogy. Because patterns of variation in the Y chromosome are only influenced by processes affecting men, they can be used to study both demographic and social phenomena. The 1000 Genomes Project includes whole Y-chromosome data from more than 1000 men and has an extensive representation of most lineages that have experienced recent massive expansions in size. Though the dynamics of population growth have likely changed over time, we are more interested in the growth rates at the times of these rapid expansions than on an average effect. To study this, we have developed a new method that takes advantage of the temporal resolution provided by Y-chromosome data and of historical data, while accounting for the uncertainties associated with the coalescent and mutational processes. 
We estimate the growth rates for several branches of the Y-chromosome tree, including those in Europe, sub-Saharan Africa and South Asia. We estimate that several lineages within the European R1b, sub-Saharan African E1b, and South Asian R1a haplogroups experienced growth rates of at least 20-60% per generation at the onset of their massive expansions, some 3-5 thousand years ago. These high growth rates are comparable to those experienced by human populations during the 20th century. However, we find that most observed genealogies are unlikely to be the result of whole population expansion or of natural selection.
A fourth conference paper thinks that it sees two distinct Out of Africa waves (something previously suggested by the divergent geographical distributions of Y-DNA D and Y-DNA F and its descendants), but previously genetic studies have shown the timing of the waves to be nearly degenerate within the accuracy of the available genome based dating techniques.  I'm skeptical that there methods can really be as definitive as they claim.

Metspalu, et al., "Demographic inferences from 447 complete human genome sequences from 148 populations worldwide."
Complete high coverage individual genome sequences carry the maximum amount of information for reconstructing the evolutionary past of a species in the interplay between random genetic drift and natural selection. Here we use a novel dataset of 447 human genomes sequenced at 40X on the same platform (Complete Genomics) and uniform bioinformatic pipelines. Based on SNP-chip data we generally chose three samples to represent each population of interest. We cover a wide range of mostly Eurasian populations with additional populations from Oceania, South America and Africa.Here we describe the dataset in terms data quality and new recovered genetic variation that originates predominantly from previously subsampled continental regions. 
Using MSMC, D-statistics and Finestructure we have shown that peopling of the World from Africa is best explained by at least two migration waves (See Lawson et al abstract …). Here we expand on these conclusions by investigating short IBD segment sharing patterns using diCal, Hapfabia etc. We also disentangle split times involving the two migrations out of Africa (OoA), by running MSMC separately on genome chunks derived from OoA1 and OoA2. We also present detailed regional population histories in reconstructions of past dynamics of effective population size and population split times.
Roger Blench, et al, has a 2006 paper (pdf) that explores a similar two migrations out of Africa hypothesis. The abstract from that paper states:
Recent hypotheses on the early expansion of early modern humans out of Africa have emphasised the coastal route, crossing the Red Sea, following the coast of Arabia, India and eventually reaching insular SE Asia and Australia. Given the ca. 50,000 BP dates for these sites, a date of ca. 80,000 BP has been proposed for the preliminary move out of Africa. Indeed, this has been linked with the explosion of Mount Toba at this period. However, there is a striking lack of direct archaeological evidence for the greater part of this route; explanations for this lacuna are varied but none are wholly satisfactory. 
Nonetheless, there seems to be an array of linguistic, cultural and genetic evidence that links together relic populations throughout this area. The paper proposes that the Malagasy Vazimba, the Sri Lankan Vedda, the Andamanese, perhaps the Shom Pen of the Nicobar Islands, the Negritos of SE Asia all provide evidence for this early expansion. 
Recent proposals for features common to the languages of Africa and the Pacific will be considered in the light of this model.

No comments: