Figure 3 shows the resulting mean conditional AFS for nine western lowland gorillas (excluding the three lowest-coverage samples as above), and comparable samples from three human populations whose ancestry derives from Africa, Asia and
Europe. In a population whose effective size has remained constant, the theoretical expectation for such a conditional AFS is a straight line of constant negative slope , shown by the dashed line in Figure 3. Compared to this, western lowland gorillas show a deficit of rare alleles, consistent with their having undergone genetic drift due to a bottleneck or other reduction in effective population size during their demographic history. The similar signal in non-African human populations has been attributed to population contraction associated with the out-of-Africa event . By contrast to the gorillas and the non-African humans, the African YRI population in Figure 3 shows an excess of rare alleles, consistent with population expansion and again similar to the signal seen in other African human data .
A population bottleneck culls rare alleles, and thus leads to a "dip" in the AFS on the left. An allele that occurs at, say, 10% frequency in a very small population is more likely to go extinct "by accident" than one which occurs at exactly the same frequency in a large population. You can think of this by taking the two extreme cases:
- if an allele exists in exactly one physical copy (i.e., 1 copy in 5 diploid individuals = 10% frequency), then its bearer must survive, must reproduce, and the allele must be inherited by one offspring in order for it survive.
- if the population has infinite size, then 10% frequency is still = infinite number of physical copies, hence the allele will survive no matter what [unless it always kills its unlucky bearer, but then how did it end up at 10% frequency in the first place?]
On the other hand, low-frequency alleles become more prevalent when the population expands in the recent past, because there is an ever growing number of bodies, an ever growing number of mutated alleles, but not enough time for these new alleles to grow in frequency.
A different mechanism whereby low-frequency alleles appear in excess in a population is admixture. African Americans, for example, have ~20% European ancestry, so any alleles that are present in Europeans and absent in Sub-Saharan Africans would tend to appear as low-frequency alleles in African Americans.
It is an open question to what degree modern human differences in the presence of low-frequency alleles are due to bottlenecks (such as the Out-of-Africa event) and to what degree they are due to admixture with other non-modern groups. A recent paper discovered a signal of Neandertal admixture when one considered alleles with 10% or less frequency in Europeans.
In the case of African Americans, we can tell that some of their low-frequency alleles were acquired by admixture with Europeans, because we have European samples; and, in the case of Europeans, we can tell that some of their low-frequency alleles were acquired by admixture with Neandertals, because we have a Neandertal genome. But, we don't have many of the genomes of archaic human groups that may have contributed variants to modern humans (and in some cases, like the Denisovans, we did not even know they existed in the first place!), so we must always keep in mind the possibility that such alleles may lurk on the left side of the AFS.
A genome-wide survey of genetic variation in gorillas using reduced representation sequencing
Aylwyn Scally et al.
All non-human great apes are endangered in the wild, and it is therefore important to gain an understanding of their demography and genetic diversity. To date, however, genetic studies within these species have largely been confined to mitochondrial DNA and a small number of other loci. Here, we present a genome-wide survey of genetic variation in gorillas using a reduced representation sequencing approach, focusing on the two lowland subspecies. We identify 3,274,491 polymorphic sites in 14 individuals: 12 western lowland gorillas (Gorilla gorilla gorilla) and 2 eastern lowland gorillas (Gorilla beringei graueri). We find that the two species are genetically distinct, based on levels of heterozygosity and patterns of allele sharing. Focusing on the western lowland population, we observe evidence for population substructure, and a deficit of rare genetic variants suggesting a recent episode of population contraction. In western lowland gorillas, there is an elevation of variation towards telomeres and centromeres on the chromosomal scale. On a finer scale, we find substantial variation in genetic diversity, including a marked reduction close to the major histocompatibility locus, perhaps indicative of recent strong selection there. These findings suggest that despite their maintaining an overall level of genetic diversity equal to or greater than that of humans, population decline, perhaps associated with disease, has been a significant factor in recent and long-term pressures on wild gorilla populations.