In 2000, when the term “archaeogenetics” (1) was initially proposed to describe the study of the human past by using the techniques of molecular genetics, ancient DNA studies were in their infancy. At that time, DNA data samples provided by geneticists to archaeologists, anthropologists, and linguists for archaeogenetic studies were almost entirely derived from living populations, with a few ancient DNA samples mainly restricted to mitochondrial DNA (mtDNA, or mitogenome). However, it was already clear that ancient DNA could play a much greater role once a number of major technical problems were resolved. Two studies in this issue exemplify the rapid progress that has been made since then. On page 1028, Ebenesersdóttir et al. (2) report the genomes of the first Icelanders, and on page 1024, Scheib et al. (3) use ancient Native American genomes to reconstruct the first phases of the human spread in the Americas.
In the early years of this century, no one anticipated the extent and speed at which ancient genomes and new inferences concerning prehistoric demographic events and historical accounts would be published. Moreover, ancient genomes are no longer retrieved solely from geographic regions with climate conditions that allow DNA to be well preserved, but are now recovered from sites all over the world. Today, we are in a second phase of the discipline initially baptized by Renfrew and Boyle (1), a phase that could be more accurately defined as the era of archaeogenomics, as exemplified by the two papers in this issue (2, 3) and by other very recent studies (see the figure) (4, 5).
The Power of Genetic Drift
Ebenesersdóttir et al. analyzed 27 genomes from skeletal remains excavated at multiple sites in Iceland. This European island was peopled in the second half of the 9th century by Vikings and their slaves who were mostly from Norway and the British-Irish Isles. The skeletal remains were selected for DNA studies because archaeological and radiocarbon dates indicated that they were from the earliest settlers of Iceland. The genomes show that first-generation settlers were mainly unmixed, coming from either Norse or Gaelic sources (see the photo). This finding was supported by the study of strontium 87 and 86 isotope ratios in the settlers’ tooth enamel; these ratios reflect the place of childhood residence through diet.
Comparison with modern genomes yielded an unexpected finding. The ancient settlers’ genomes are more similar to those of the modern populations of Scandinavia and the British-Irish Isles than those of modern Icelanders. This observation highlights the power of genetic drift on a population that, for almost the entire first millennium after its foundation, rarely counted more than 50,000 people.
Ancient Genomes from Africa
In a recent study, van de Loosdrecht et al. (4) reported genomic data from seven individuals from Grotte des Pigeons near Taforalt in eastern Morocco. These individuals are associated with the Later Stone Age Iberomaurusian culture and lived between 15,100 and 13,900 years ago. Their genomes are the most ancient analyzed so far from Africa, a technical success that overcomes previous limitations in obtaining ancient DNA from tropical or subtropical regions. The most likely reason is that all Taforalt DNAs were extracted from petrous bones, where endogenous DNA is best preserved.
Comparison with genomes from other ancient samples and modern populations reveal three major ancestry components that best characterize the Taforalt genomes: early-Holocene Levantines (Natufians), West Africans, and East African hunter-gatherers (Hadza). The first component indicates an important pre-Neolithic genetic input from the Near East, whereas the other two indicate a substantial input from sub-Saharan Africa, a contribution that is much less evident in modern North Africans. No gene flow signals from Paleolithic Europeans were detected, directly and definitively refuting a previously proposed European origin for the Iberomaurusian culture.
As a single, maternally inherited locus, mtDNA often does not reflect the whole complexity of demographic events and is also particularly prone to genetic drift. However, after refinements in its evolution rate (6, 7) and the nesting relationships within its phylogeny, mtDNA can now provide rather narrow time boundaries for dating specific demographic events. All seven Taforalt specimens belong to two North African branches (haplogroups) of the mtDNA tree: Six belong to mitochondrial haplogroup U6, and one belongs to M1. These two haplogroups are those proposed 12 years ago, on the basis of modern DNA data, as markers of an Upper Paleolithic migration from the Levant to North Africa (8). At least for U6, this is a scenario that the ancient mitogenomes from Taforalt (and elsewhere) now confirm.
Migrations of the First Americans
The first ancient human genome sequence was from a ∼4000-year-old Paleo-Eskimo from Greenland (9). In 2015, Raghavan et al. reported 23 ancient genomes from North and South American individuals, dated to between 6000 and 200 years ago (10). Comparison with modern and other ancient genomes led to the conclusion that modern Amerindian and Athabaskan groups derive from a single source population that began to diversify ∼13,000 years ago, when it split into two branches: a northern one now restricted to North America (Athapaskans and northern Amerindians) and a southern one that is now dispersed across North and South America. The latter also includes the genome of Anzick-1, a Clovis-culture male infant from Montana dated at ∼12,600 years ago. The split was considered to have most likely happened in lower-latitude North America (south of the Cordilleran and Laurentide ice sheets) rather than in eastern Beringia (Alaska).
Recently, the same group reported the genome from another infant, in this case a female from Upward Sun River, Alaska, dated at 11,500 years ago (5). Her genome belonged to a Beringian-specific ancient population that had diverged from the ancestors of Amerindians and Athapaskans between 22,000 and 18,100 years ago, showing that the ancestors of Native Americans were already differentiating during their stay in Beringia (“Beringian standstill”). More importantly, this infant Alaskan genome suggests that the split between the southern and northern branches occurred 17,500 to 14,600 years ago, rather than ∼13,000 years ago as previously proposed. This date range is much more realistic in light of the age (≥14,500 years) of Monte Verde in southern Chile, the oldest known archaeological site in South America.
Scheib et al. have now sequenced an even larger number of genomes—91, including mitogenomes—from ancient Native American remains; only 27 of these remains were radiocarbon-dated, with ages ranging between 4800 and 200 years ago (3). They were mainly selected from the Channel Islands of California and from Southwestern Ontario. Both regions show signs of human occupation as early as 13,000 years ago, were never reached by the much more recent ancient Neo- and Paleo-Eskimo migrations, and were expected to provide a good representation of the southern and northern Native American branches.
Scheib et al. identified two distinct basal ancestries, ANC-A and ANC-B, which differentiated before the split between the northern and southern Native American branches. ANC-A is largely associated with the southern branch (and Anzick-1), whereas ANC-B reaches its maximum in the northern ancient Ontarians and modern Algonquians. However, both components are present and admixed in the ancient Californian populations as well as in present-day populations from Central and South America, including the Chilote and Huilliche, who show a very high ANC-B frequency despite living in Chile, not far from the Monte Verde site.
To explain these findings, the authors propose different models, which all assume an initial split between ANC-A and ANC-B south of the Cordilleran and Laurentide ice sheets (see the figure) and different possible merging points between the two derived groups either in North America (once or multiple times) or along the way to South America. However, they could not have begun to merge until a few thousand years after the split, taking into account that the differentiation of ANC-A and ANC-B was not a sudden process. Alternatively, the split may have occurred north of the ice sheets in Beringia, but in this case, ANC-A and ANC-B would have theoretically been susceptible to differential gene flow from other human groups in Beringia. Scheib et al. reject this possibility because the two ancestries do not differ significantly in their affinity to modern non-American populations, including Asians.
In their models, Scheib et al. do not exploit the mitogenome information that they obtained. For instance, the ancient Ontarians harbored mtDNA haplogroups X2a and C4c, which are currently found at high frequency among the Algonquian-speaking populations; they could thus be distinctive mitochondrial markers of ANC-B. In this view, the detection of an early offshoot of C4c in modern Colombians (11) is in full agreement with the scenario that ANC-B contributed to the shaping of South American genomes.
Given an upper boundary of 17,500 years ago for the split time between the southern and northern branches (5) and the requirement of a few thousand years of differentiation after the split (3), it may be time to reconsider an initial differentiation in Beringia. A reconsideration of the time and the place of the population split (see the figure) would also be compatible with the most recent archaeological findings in South America (12). However, a split in Beringia brings back the scenario of a dual entry into the American continent and thus the possibility that the two ancestries entered the Americas either at the same time or at different times, following the same or different routes (13–15).
Considering the rapid progression in the production of ancient genomes, their densities on geographical maps are expected to increase very rapidly. There is no doubt that these data will allow many long-lasting questions to be addressed, but it is likely that many others will arise.