- Founder effects in modern populations
The genomes of ancient humans can reveal patterns of early human migration (see the Perspective by Achilli et al.). Iceland has a genetically distinct population, despite relatively recent settlement (∼1100 years ago). Ebenesersdóttir et al. examined the genomes of ancient Icelandic people, dating to near the colonization of Iceland, and compared them with modernday Icelandic populations. The ancient DNA revealed that the founders had Gaelic and Norse origins. Genetic drift since the initial settlement has left modern Icelanders with allele frequencies that are distinctive, although still skewed toward those of their Norse founders. Scheib et al. sequenced ancient genomes from the Channel Islands of California, USA, and Ontario, Canada. The ancient Ontario population was similar to other ancient North Americans, as well as to modern Algonquian-speaking Native Americans. In contrast, the California individuals were more like groups that now live in Mexico and South America. It appears that a genetic split and population isolation likely occurred during the Ice Age, but the peoples remixed at a later date.
Abstract
Little is known regarding the first people to enter the Americas and their genetic legacy. Genomic analysis of the oldest human remains from the Americas showed a direct relationship between a Clovis-related ancestral population and all modern Central and South Americans as well as a deep split separating them from North Americans in Canada. We present 91 ancient human genomes from California and Southwestern Ontario and demonstrate the existence of two distinct ancestries in North America, which possibly split south of the ice sheets. A contribution from both of these ancestral populations is found in all modern Central and South Americans. The proportions of these two ancestries in ancient and modern populations are consistent with a coastal dispersal and multiple admixture events.
An increasing body of archaeological (1–3) evidence shows that the initial peopling of the Americas occurred at least a few thousand years prior to the spread of the Clovis cultural complex ~13,000 years ago (all dates are calibrated) (4), with a majority of well-supported Pre-Clovis sites clustered in coastal areas and around glacial edges (1, 3, 5). Studies of ancient and modern genomes have uncovered four distinct ancestry components within the Americas arriving in three hypothesized waves: the most recent Thule-related Neo-Eskimo ~2000 years ago, the Saqqaq/Dorset Paleo-Eskimo ~4500 years ago (both restricted to the Arctic region), and a “First American” dispersal prior to 13,000 years ago that split within North America into a northern and a southern branch (6–10). The northern branch is ancestral to populations including Algonquian, Na-Dené, Salishan, and Tsimshian speakers from Canada (NAM), whereas the southern branch includes the ancestors of the Clovis individual (Anzick-1) and all Mexicans, Central Americans (CAM), and South Americans (SAM) (9–12). Within the southern branch there is some localized evidence of early population structure, as a few modern Amazonian populations show an excess genetic affinity to Australasians (13, 14). The second oldest North American genome, The Ancient One (Kennewick Man, 8700 to 8400 years ago), is likely to have derived from the ancestral northern branch but is a poor proxy for modeling this ancestry, given its low sequencing depth (10).
Here, we investigated the ancestral relationship between the northern (NAM) and southern (Mexico, CAM, and SAM) branch populations. To do so, we sequenced 91 ancient whole genomes from North America, mainly from two geographic areas: the California Channel Islands in the west and Southwestern Ontario in the east, near modern Algonquian-speaking populations (Fig. 1A and table S1) (15). Both of these areas show evidence of occupation from at least 13,000 years ago (5, 16) and are geographically located south of the known distribution of the ancient Neo- and Paleo-Eskimo dispersals (6). We radiocarbon-dated 27 individuals (table S2) (15) to between ~4800 and ~200 years ago and sequenced all genomes to an average depth of 0.007 to 13.6× (tables S1, S3, and S4) (15). Mitochondrial DNA (mtDNA) haplotypes were recovered from all samples (tables S1 and S5) (15) and Y chromosome haplotypes from 34 of the male individuals (fig. S1 and data S1) (15). In addition, a set of modern whole mitochondrial genomes (n = 45) were resequenced from a previous study to explore sex-specific migration patterns on the west coast of Southern California (tables S3 and S5) (15, 17).
(A) Sites of newly sequenced ancient individuals are designated by colored triangles. Comparative modern populations and ancient individuals are designated by black circles and triangles, respectively. (B) PCA with ancient individuals projected onto modern Native American and Siberian variation. Inset: Ancient genomes projected onto modern worldwide data. (C) Visualization of model-based ancestry analysis at fivefold cross-validation–supported K = 9 ancestral components (15). Underlines denote new and ancient genomes; italics, published ancient genomes; single asterisks, masked data; double asterisks, Oceanian populations including Onge, Aeta, and Agta (15). (D) Probability area of radiocarbon dates grouped by population and calibrated with IntCal13 (colored area) and Marine13 (light gray area) (28) where appropriate (15). (E) A model that explains genetic diversity in the Late Southern Channel Island populations through three-way admixture.
All ancient Native American individuals clustered with modern Native Americans on a worldwide principal components analysis (PCA) (Fig. 1B) (15, 18). In a regional plot that includes Siberians, the northern and southern branches show clear distinction (Fig. 1B, inset). Ancient Californians cluster alongside southern branch populations near to Anzick-1, whereas the ancient Southwestern Ontario (ASO) population clusters with modern Algonquian speaking populations and The Ancient One (Fig. 1B). Modern and ancient Athabaskans map between ASO and Northeast Siberian populations (Chukchi/Koryak) (Fig. 1B). The Upper Sun River individuals (USR1 and USR2) map near to Shuká Káa, between Central Siberians and the northern branch populations (Fig. 1B).
Both PCA and ADMIXTURE (19) analysis of the ancient genomes within the context of a worldwide panel (Fig. 1, B and C, fig. S2, and data S2) (15) indicate that, relative to the ASO, extant NAM as well as ancient Pacific Northwest Coast (PNWC) and ancient Northern Athabaskans have up to 50% more Arctic-related ancestry, prominent in Greenland Inuits and also found in Siberian Eskimos (Fig. 1C) (15). These differences are further confirmed by significant Z (Z > 3) scores for D-statistics D(Mbuti,SiberianPop;Mixe,TestPop) (20) and f3-statistics f(ASO,TestPop;Mbuti) (21, 22) (table S6 and data S3) (15). We found no significant evidence of gene flow into the ASO from any non-American population (table S7 and data S3) (15).
All ancient Californian genomes clustered together with southern branch (Anzick-1, Mexico, CAM, SAM) populations on the regional PCA (Fig. 1B) (15). As suggested by the archaeological and osteological evidence (23–25), radiocarbon-dated individuals from San Nicolas Island from early (ESN) and late (LSN) periods (Fig. 1D) clustered into two distinct genetic populations in both the autosomal and uniparental markers with evidence of 11% autosomal continuity between the early and late populations (Fig. 1E) (15). The genome-wide diversity estimates (26) of the earliest-dated LSN individuals are greater than those of the ESN individuals (fig. S3) (15), which is consistent with higher effective population sizes in the LSN source population and/or recent admixture. Outgroup f3(Test,X;Mbuti) and D(Mbuti,Test;Mixe,X) tests show that the ESN shares more drift with SAM (Z > 3) than with geographically proximal populations (data S3) (15), which suggests that the ESN is related to a population that expanded into South America. By contrast, the LSN population shares more drift and alleles with geographically proximal populations (Z > 3) (data S3) (15).
We modeled the population history of the Americas using qpGraph (15, 21) and found that the ASO and Mexican (Pima) populations were consistently outgroups to sets of clades formed by Anzick-1, SAM (Surui), and ESN populations in analyses that did not involve admixture (fig. S4) (15, 21). Fit between the data and the tree could be significantly improved when modeling ancient Californian, modern Pima, and Surui populations through admixture of two basal ancestries that we call ANC-A and ANC-B (Fig. 2A) (15). The ESN, Northern Channel Islands and Santa Barbara (NCI/SB), and Surui populations share similar proportions of both components, while the Pima have a higher ANC-B component (Fig. 2A) (15). We used qpGraph to estimate the ANC-B contribution in modern CAM and SAM populations and found it to vary within a range of 42 to 71% (average 53%; table S8) (15). In SAM populations, the lower end of the spectrum of contributions of ANC-B are found in the Amazonian Equatorial Tucanoan-speaking groups (including Surui) (40 to 53%) and the highest in the Andeans (50 to 71%) (Fig. 2B and table S8) (15), particularly in the Chilote and Huilliche (~70%) from locations overlapping the Monte Verde site (~18,500 to 14,500 years ago) (Fig. 2B).
(A) A model with four admixture events that offers a good fit to the data (Z = 0.888) (15). (B) Scale of ANC-B ancestry from 0% in Anzick-1 to 100% in the ASO and modern Algonquian-speaking populations.
The clear separation of ANC-A and ANC-B ancestries is further supported by the sharing of unambiguous, derived haplotype segments in modern Surui and Pima populations (27) with both the ASO (CK-13) and Anzick-1 individuals (fig. S5) (15). The results of this analysis are consistent with ancient substructure and a separation of at least a few thousand years between the ANC-A and ANC-B populations prior to merging (fig. S6) (15). The summary of evidence presented here allows us to reject models of a panmictic “first wave” population from which the ASO diverged after the peopling of South America or in which solely the ANC-A population contributed to modern southern branch populations. Because populations vary in ANC-A and ANC-B proportions but do not differ significantly in their affinity to non-American populations (table S7) (15), it is possible that ANC-A and ANC-B split within America as opposed to Beringia where there would have been ongoing gene flow with Siberia.
Four possible models can explain the contribution of both branches to CAM and SAM populations: (i) an admixture event in North America prior to the peopling of South America (Fig. 3A); (ii) ANC-B–related ancestral population(s) dispersing into South America first, followed by a dispersal of ANC-A–related population(s) and admixture of the two branches occurring in South America (Fig. 3B); (iii) ANC-A–related ancestral population(s) dispersing into South America first, followed by a dispersal of ANC-B–related population(s) and admixture of the two branches occurring in South America (Fig. 3C); and (iv) multiple admixture events occurring in North America, with multiple dispersals into South America (Fig. 3D). Additional ancient DNA from terminal Pleistocene human remains within the Americas is needed to determine which model best describes the sequence of events constituting the complex population history of the Americas.
Red and blue indicate ANC-A and ANC-B, respectively; symbols denote admixture event(s). Locations of admixture events are hypothetical. (A) A model with one admixture event in North America. (B) A model in which an ANC-B population first reached South America, followed by an ANC-A population with multiple admixture events. (C) The same model as (B), but reversing the populations. (D) A model with multiple admixture events and dispersals.
