The history of African gene flow into Southern Europeans, Levantines, and Jews

Abstract

Previous genetic studies have suggested a history of sub-Saharan African gene flow into some West Eurasian populations after the initial dispersal out of Africa that occurred at least 45,000 years ago. However, there has been no accurate characterization of the proportion of mixture, or of its date. We analyze genome-wide polymorphism data from about 40 West Eurasian groups to show that almost all Southern Europeans have inherited 1%-3% African ancestry with an average mixture date of around 55 generations ago, consistent with North African gene flow at the end of the Roman Empire and subsequent Arab migrations. Levantine groups harbor 4%-15% African ancestry with an average mixture date of about 32 generations ago, consistent with close political, economic, and cultural links with Egypt in the late middle ages. We also detect 3%-5% sub-Saharan African ancestry in all eight of the diverse Jewish populations that we analyzed. For the Jewish admixture, we obtain an average estimated date of about 72 generations. This may reflect descent of these groups from a common ancestral population that already had some African ancestry prior to the Jewish Diasporas.

Figure 1. PCA Projection.

PCA was performed using genome-wide SNP data from East Asians (HapMap3- CHB) and South Africans (HGDP-CEPH- San). All West Eurasians populations with samples sizes of n ≥ 5 were then projected onto these PCs. (a) The first panel presents data for all populations and (b) the second panel provides a higher resolution view of West Eurasians after removing sub-Saharan Africans. Each point on this graph indicates the mean value of the first PC for a projected population. West Eurasians populations are colored by 5 regional groupings—“Northwest Europe”, “East-Central Europe”, “Southern Europe”, “Levant”, “Jewish Groups” (the assignments of populations to groups is shown in Table 1). The grouping “Sub-Saharan Africa” refers to six populations from the HGDP-CEPH panel: Kenyan Bantu, South African Bantu, Mandenka, Mbuti Pygmy, Biaka Pygmy and Yoruba.

Figure 2. Estimation of African ancestry using f₄ Ancestry Estimation.

f₄ Ancestry Estimation computes the quantity [(San-Papuan).(X–CEU)/[(San-Papuan).(YRI-CEU)]; where X  =  any West Eurasian population. The denominator is proportional to the genetic drift m that occurred in the ancestors of West or East Africans since their divergence from San but prior to their divergence from West Eurasians (intersection of red and orange lines). The numerator is proportion to p*(Ancestral Africans-YRI) + (1-p)*(Ancestral Europeans-CEU). Since the branches connecting (San, Papuan) and (CEU, X) do not overlap each other, the quantity (1-p)*(X-CEU)  =  0 and hence the numerator is expected to equal pm. Thus, the ratio of the numerator and denominator is expected to equal p (Ancestral African mixture proportion). This figure is adapted from reference , where we first developed f₄ Ancestry Estimation, and where we reported computer simulations demonstrating its robustness.

Figure 3. Testing for LD due to African admixture in West Eurasians.

To generate these plots, we used the ROLLOFF software to calculate the LD between all pairs of markers in each population, weighted by their frequency difference between YRI and CEU to make the statistic sensitive to admixture LD. We plot the correlation as a function of genetic distance for Portuguese, Russians, Sephardic Greek Jews and Palestinians. We do not show inter-SNP intervals of <0.5cM since we have found that at this distance admixture LD begins to be confounded by background LD, and so inferences are not reliable (exponential curve fitting does not include inter-SNP intervals at this scale).

Figure 4. ROLLOFF simulation results.

We constructed 10 individuals of mixed African and European ancestry (where individuals had 20% European ancestry) for various time depths ranging from 10–300 generations (with intervals of 10 generations). We performed ROLLOFF analysis using another independent dataset of European Americans and Nigerian Yoruba individuals as reference populations. We plot the true time depth (that was used for the simulations) against the estimated time depth computed by ROLLOFF. The expected time depth is shown as a dotted grey line. Standard errors were calculated using the Weighted Block Jackknife described in the Materials and Methods.