Neanderthals and humans interbred in Europe for 5,400 years

Discussion
The African introgression hypothesis suggests that Late Pleistocene Neanderthal mtDNAs originated through gene flow from an African source8, which we constrain taking place more than ∼270 ka (Table 1). Our analytical calculations (Supplementary Table 8 and Supplementary Note 5) show that this event is plausible even if the introgressing lineage represented a minimal proportion of the initial gene pool. This scenario reconciles the discrepancy in the nDNA and mtDNA phylogenies of archaic hominins and the inconsistency of the modern human–Neanderthal population split time estimated from nDNA and mtDNA (Fig. 1d). Under this demographical model, the Denisovan mtDNA type was common among early Neanderthals in Eurasia (for example, Sima de los Huesos) and was then largely replaced by an introgressing African mtDNA that evolved into the Late Pleistocene Neanderthal mtDNA type. While the upper bound for the time of this putative gene flow event would be the divergence time between Neanderthal and modern human mtDNAs, here dated to 413 ka (95% HPD 468–360 ka), the lower temporal limit was represented so far by the ∼160 ka TMRCA of all published Neanderthal mtDNAs (Table 1). However, the finding of the deeply diverged HST lineage splitting from the Altai branch, ∼270 ka, sets an older lower boundary for the time of this admixture event. An alternative but less parsimonious scenario is that both HST and Altai mtDNA lineages reached Eurasia independently after diverging inside Africa. In that case the suggested introgression event might have occurred later but most likely before 160 ka, our estimated date for the start of the Altai branch diversification (Fig. 1c and Table 1).

The presence of modern human admixture into archaic humans has already been detected in the high coverage Neanderthal genome from the Altai region but not in sequences of chromosome 21 of two Neanderthals from Spain and Croatia14. The authors therefore suggested that a genomic contribution estimated between 0.1 and 2.1% occurred after the divergence of Altai from other late Neanderthals. However, there is a high level of uncertainty around the time of the inferred gene flow event since only one high coverage Neanderthal nuclear genome has been analysed so far. Moreover, the divergence time of the introgressing African population was estimated to date before or right after the TMRCA of modern-day humans (∼200 ka)14, while the mtDNA coalescence time between Neanderthals and modern humans is calculated at least twice as old (∼400 ka). The evolutionary scenario responsible for providing the mtDNA to the Late Pleistocene Neanderthals might have been an even earlier Middle Pleistocene gene flow from Africa, occurring in a time interval that we date between 413 and 268 ka (460–219 ka including upper and lower 95% HPD). It should be highlighted that this additional genomic contribution might have already been accounted for in ref. 14, which effectively measures the total amount of African introgression into Neanderthals after their split from Denisovans (473–381 ka; ref. 5).



The phylogenetic branch length of mtDNA sequences from 10 non-dated Neanderthal individuals was considered in BEAST, to assess individual molecular ages spanning from 130 to 40 ka. Although it is not known if the mtDNA mutation rate in modern humans is comparable to that of Neanderthals (Supplementary Note 4), molecular dating can at least be used to provide relative ages when the radiocarbon absolute chronometric method is not applicable. After the Altai mtDNA, HST is estimated to be the second oldest mtDNA with an age of 124 ka (95% HPD 183–62 ka). This wide temporal interval largely overlaps with the Marine Isotope Stage 5 (MIS 5: ∼130–73 ka)37. After its initial interglacial period (MIS 5e), central Europe was characterized by climatic fluctuations resulting in forestation cycles (MIS 5c/5a) alternated with the development of steppe-tundra biomass (MIS 5d/b)38. The stable isotopic δ13C and δ15N values of the archaic femur collagen and associated faunal remains support a more temperate, forested rather than a colder, steppe environment and is therefore consistent with an ecological context during the early warm phases of the last glaciation17.

Despite having only a single complete mtDNA on the HST lineage, the two highly differentiated Neanderthal mtDNA branches suggest higher mtDNA diversity during the Middle Pleistocene, which then declined during the Late Pleistocene (Supplementary Table 4). This observation is also supported by the steady decline in mtDNA effective population size displayed in the skyline plot before a steep growth in late Neanderthal population sizes (Supplementary Fig. 7). Studies focusing on the demographic patterns of late Neanderthals who overlapped with the earliest modern humans in Europe are of key importance to understand population dynamics and interactions between archaic and modern humans.



In conclusion, the HST mtDNA provided insights into the mtDNA diversity of Neanderthal populations through the Middle and Late Pleistocene. Its deep divergence time allowed us to further constrain the lower boundary for the time of the proposed African mtDNA gene flow into Neanderthal populations. The temporal corridor for this introgression event between 460 ka and 219 ka is compatible with the evidence of archaeological similarities between Africa and western Eurasia during the Lower to Middle Paleolithic transition39 and potentially may explain the dissimilarities in Middle Paleolithic industries between eastern and western Eurasia. Environmental changes across this time span might have facilitated a hominin expansion out of Africa and potentially spread cultural innovations such as the Levallois technology into Eurasia40. Alternatively, other scenarios such as multiple inventions of similar technologies by various hominin groups may explain the complex tapestry of technological variability during the late Middle Pleistocene.

Nuclear data from the HST femur would be pivotal in assessing its genomic relationships with Neanderthals, Denisovans and modern humans. However, the scarce preservation of HST endogenous DNA in combination with high level of modern human contamination challenge the retrieval of its complete genome. Analyses of high-quality nDNA from more than one well-preserved Neanderthal individual are necessary to detect the consequences of African admixture into archaic human populations.

Nature Communications 8, Article number: 16046 (2017)