Post by Admin on May 3, 2023 18:10:37 GMT
Males from Persia and females from Africa
We tested whether male and female ancestors contributed the same proportions of African-like, Persian-like and Indian-like ancestry to ancient individuals in the northern coastal sites and Kilwa (Table 1). To carry out this analysis, we used the fact that chromosomes 1–22, the X chromosome, mitochondrial DNA (mtDNA) and the Y chromosome are passed down to subsequent generations in different ways by males and females. We could not perform the same analysis at Songo Mnara because no individuals with high quality data fit the three-way model.
We first analysed mtDNA (Extended Data Table 1 and Table 1). Analysing 62 individuals with confidently determined mtDNA haplogroups (including relatives and individuals with low coverage genome-wide data), we find that 59 carry an L* haplogroup, which in the present day is almost entirely restricted to sub-Saharan Africans27. The exceptions are a pair of first- or second-degree relatives from Mtwapa carrying M30d1, which in the present day is largely restricted to South Asia27, and an individual with haplogroup R0+16189, which today is characteristic of Saudi Arabia and the Horn of Africa28. These results are consistent with female ancestry deriving overwhelmingly from African sources.
Analysing male-transmitted Y chromosome DNA, we find that two out of three non-first-degree related males from Manda carry haplogroup J2, and the third carries G2. Both haplogroups are characteristic of Southwest Asia (plausibly Persia) and are largely absent in sub-Saharan Africans29. The Kilwa individual also carries J2. Fourteen out of 19 males from Mtwapa have Y chromosome haplogroups in the J family, and two are of the R1a haplogroup, all considered typically non-African. Only 3 out of 19 Mtwapa males, along with the Faza male, are in the E1 family characteristic of sub-Saharan Africa.
We next compared chromosome X, which occurs as two copies in females and one in males and so reflects mostly female history, to the autosomes (chromosomes 1–22), which equally reflect female and male history (Methods and Supplementary Information). Chromosome X estimates of African ancestry are higher than on the autosomes at all sites, providing an independent line of evidence that African ancestry is primarily from females and Persian ancestry is primarily from males (Table 1). Assuming that the mixture occurred over just a few generations, we obtain quantitative estimates of the proportion of African ancestry from females as 100% at Manda, 69–97% at Mtwapa-Faza, and 69–100% at Kilwa (Methods and Table 1). We estimate Persian ancestry at Mtwapa-Faza and Kilwa to be 100%, and at Manda as 90–100%. If the mixture occurred over more generations, we cannot obtain a point estimate, but can nevertheless infer primarily African female and Persian male sources.
Together, these multiple lines of evidence show that the Southwest Asian ancestors of the Mtwapa and Faza, Manda and Kilwa individuals were almost entirely male, whereas the African ancestors were almost entirely female.
Mixing began by AD 1000
We estimated when mixture occurred on the basis of the sizes of stretches of ancestry inherited from the ancestral populations, which break up at a known rate every generation30. We calculated 95% confidence intervals for the inferred dates of AD 795–1085 for a pool of the northern Mtwapa, Faza and Manda individuals, and AD 708–1219 for a pool of the southern Kilwa and Songo Mnara I19550 individuals (Table 1 and Extended Data Fig. 3). The uncertainty intervals overlap from AD 795 to AD 1085. These estimates would be biased too old if there was a marine reservoir effect. The inferred dates also reflect an assumption that the mixture occurred all at once; however, mixture of Eurasians and Africans was certainly drawn out over a number of generations, and indeed the historical evidence and our genetic analysis that follows document continued incorporation of migrants from both inland Africa and Eurasia until the present. However, simulations show that mixture must have begun by the inferred date31, and thus we can be confident that already-mixed males with both Indian and Persian ancestry were present along the coast by around AD 1000, and began mixing with primarily female sub-Saharan Africans by that time.
Arabians and other migratory influences
Although almost all the coastal individuals we analysed had Asian ancestry, there were exceptions. Some early modern individuals at Lindi and Songo Mnara showed no evidence of recent Asian ancestry (I14001 and I7944) (Extended Data Fig. 2a, Extended Data Table 3, Supplementary Table 11 and Supplementary Information). We find possible Malagasy-associated ancestry in Songo Mnara (I19547) (Extended Data Fig. 2a, Extended Data Table 3 and Supplementary Information). Our finding of coastal individuals who differ from others from similar times or regions attest to continued exchange with people in the Indian Ocean trading network, although our sample size is too small to identify general patterns.
For some of the individuals in our study with Asian ancestry (from Manda, Kilwa and I19550 from Songo Mnara), there is only evidence of Persian or Persian–Indian ancestry. However, other individuals, particularly from Mtwapa, can only be modelled by sources with some Arabian-associated ancestry when using other Mtwapa or Manda individuals as a source (Supplementary Fig. 6, Supplementary Tables 12–15 and Supplementary Information). We were unable to determine the exact source of the Arabian-associated ancestry. However, we know that it is somewhere on the genetic gradient between Arabians and Persians. A proxy source that provides a fit for this Arabian-related ancestry at Mtwapa is present-day people who live on the shores of the Strait of Hormuz, which separates the Arabian Peninsula from Iran. The Strait of Hormuz and the Swahili coast were under Omani control by the end of the seventeenth century.
Direct genetic evidence for Arabian-associated migration comes from two individuals from Songo Mnara. Both date to the early modern period when contacts with Arabia are well documented, and they can only be modelled with Arabian-related ancestry in qpAdm (Fig. 2a). Analyses of present-day coastal populations also point to Arabian genetic influences. Whereas the Asian ancestry of some individuals can be modelled entirely as Persian–Indian as in medieval Manda, for other individuals, ancestry from the Strait of Hormuz is a better fit (coloured salmon in Supplementary Data File 7).
We tested whether male and female ancestors contributed the same proportions of African-like, Persian-like and Indian-like ancestry to ancient individuals in the northern coastal sites and Kilwa (Table 1). To carry out this analysis, we used the fact that chromosomes 1–22, the X chromosome, mitochondrial DNA (mtDNA) and the Y chromosome are passed down to subsequent generations in different ways by males and females. We could not perform the same analysis at Songo Mnara because no individuals with high quality data fit the three-way model.
We first analysed mtDNA (Extended Data Table 1 and Table 1). Analysing 62 individuals with confidently determined mtDNA haplogroups (including relatives and individuals with low coverage genome-wide data), we find that 59 carry an L* haplogroup, which in the present day is almost entirely restricted to sub-Saharan Africans27. The exceptions are a pair of first- or second-degree relatives from Mtwapa carrying M30d1, which in the present day is largely restricted to South Asia27, and an individual with haplogroup R0+16189, which today is characteristic of Saudi Arabia and the Horn of Africa28. These results are consistent with female ancestry deriving overwhelmingly from African sources.
Analysing male-transmitted Y chromosome DNA, we find that two out of three non-first-degree related males from Manda carry haplogroup J2, and the third carries G2. Both haplogroups are characteristic of Southwest Asia (plausibly Persia) and are largely absent in sub-Saharan Africans29. The Kilwa individual also carries J2. Fourteen out of 19 males from Mtwapa have Y chromosome haplogroups in the J family, and two are of the R1a haplogroup, all considered typically non-African. Only 3 out of 19 Mtwapa males, along with the Faza male, are in the E1 family characteristic of sub-Saharan Africa.
We next compared chromosome X, which occurs as two copies in females and one in males and so reflects mostly female history, to the autosomes (chromosomes 1–22), which equally reflect female and male history (Methods and Supplementary Information). Chromosome X estimates of African ancestry are higher than on the autosomes at all sites, providing an independent line of evidence that African ancestry is primarily from females and Persian ancestry is primarily from males (Table 1). Assuming that the mixture occurred over just a few generations, we obtain quantitative estimates of the proportion of African ancestry from females as 100% at Manda, 69–97% at Mtwapa-Faza, and 69–100% at Kilwa (Methods and Table 1). We estimate Persian ancestry at Mtwapa-Faza and Kilwa to be 100%, and at Manda as 90–100%. If the mixture occurred over more generations, we cannot obtain a point estimate, but can nevertheless infer primarily African female and Persian male sources.
Together, these multiple lines of evidence show that the Southwest Asian ancestors of the Mtwapa and Faza, Manda and Kilwa individuals were almost entirely male, whereas the African ancestors were almost entirely female.
Mixing began by AD 1000
We estimated when mixture occurred on the basis of the sizes of stretches of ancestry inherited from the ancestral populations, which break up at a known rate every generation30. We calculated 95% confidence intervals for the inferred dates of AD 795–1085 for a pool of the northern Mtwapa, Faza and Manda individuals, and AD 708–1219 for a pool of the southern Kilwa and Songo Mnara I19550 individuals (Table 1 and Extended Data Fig. 3). The uncertainty intervals overlap from AD 795 to AD 1085. These estimates would be biased too old if there was a marine reservoir effect. The inferred dates also reflect an assumption that the mixture occurred all at once; however, mixture of Eurasians and Africans was certainly drawn out over a number of generations, and indeed the historical evidence and our genetic analysis that follows document continued incorporation of migrants from both inland Africa and Eurasia until the present. However, simulations show that mixture must have begun by the inferred date31, and thus we can be confident that already-mixed males with both Indian and Persian ancestry were present along the coast by around AD 1000, and began mixing with primarily female sub-Saharan Africans by that time.
Arabians and other migratory influences
Although almost all the coastal individuals we analysed had Asian ancestry, there were exceptions. Some early modern individuals at Lindi and Songo Mnara showed no evidence of recent Asian ancestry (I14001 and I7944) (Extended Data Fig. 2a, Extended Data Table 3, Supplementary Table 11 and Supplementary Information). We find possible Malagasy-associated ancestry in Songo Mnara (I19547) (Extended Data Fig. 2a, Extended Data Table 3 and Supplementary Information). Our finding of coastal individuals who differ from others from similar times or regions attest to continued exchange with people in the Indian Ocean trading network, although our sample size is too small to identify general patterns.
For some of the individuals in our study with Asian ancestry (from Manda, Kilwa and I19550 from Songo Mnara), there is only evidence of Persian or Persian–Indian ancestry. However, other individuals, particularly from Mtwapa, can only be modelled by sources with some Arabian-associated ancestry when using other Mtwapa or Manda individuals as a source (Supplementary Fig. 6, Supplementary Tables 12–15 and Supplementary Information). We were unable to determine the exact source of the Arabian-associated ancestry. However, we know that it is somewhere on the genetic gradient between Arabians and Persians. A proxy source that provides a fit for this Arabian-related ancestry at Mtwapa is present-day people who live on the shores of the Strait of Hormuz, which separates the Arabian Peninsula from Iran. The Strait of Hormuz and the Swahili coast were under Omani control by the end of the seventeenth century.
Direct genetic evidence for Arabian-associated migration comes from two individuals from Songo Mnara. Both date to the early modern period when contacts with Arabia are well documented, and they can only be modelled with Arabian-related ancestry in qpAdm (Fig. 2a). Analyses of present-day coastal populations also point to Arabian genetic influences. Whereas the Asian ancestry of some individuals can be modelled entirely as Persian–Indian as in medieval Manda, for other individuals, ancestry from the Strait of Hormuz is a better fit (coloured salmon in Supplementary Data File 7).