Reconstructing the genetic history of late Neanderthals

Abstract

Although it has previously been shown that Neanderthals contributed DNA to modern humans, not much is known about the genetic diversity of Neanderthals or the relationship between late Neanderthal populations at the time at which their last interactions with early modern humans occurred and before they eventually disappeared. Our ability to retrieve DNA from a larger number of Neanderthal individuals has been limited by poor preservation of endogenous DNA and contamination of Neanderthal skeletal remains by large amounts of microbial and present-day human DNA. Here we use hypochlorite treatment of as little as 9 mg of bone or tooth powder to generate between 1- and 2.7-fold genomic coverage of five Neanderthals who lived around 39,000 to 47,000 years ago (that is, late Neanderthals), thereby doubling the number of Neanderthals for which genome sequences are available. Genetic similarity among late Neanderthals is well predicted by their geographical location, and comparison to the genome of an older Neanderthal from the Caucasus indicates that a population turnover is likely to have occurred, either in the Caucasus or throughout Europe, towards the end of Neanderthal history. We find that the bulk of Neanderthal gene flow into early modern humans originated from one or more source populations that diverged from the Neanderthals that were studied here at least 70,000 years ago, but after they split from a previously sequenced Neanderthal from Siberia around 150,000 years ago. Although four of the Neanderthals studied here post-date the putative arrival of early modern humans into Europe, we do not detect any recent gene flow from early modern humans in their ancestry.

Extended Data Figure 1 |. Frequency of nucleotide substitutions at the beginning and the end of nuclear alignments for the final dataset of Les Cottés Z4–1514, Goyet Q56–1, Mezmaiskaya 2, Vindija 87 and Spy 94a.

Only fragments of at least 35 bp that mapped to the human reference genome with a mapping quality of at least 25 (MQ ≥ 25) were used for this analysis. Solid lines depict all fragments and dashed lines the fragments that have a C-to-T substitution at the opposing end (‘conditional’ C-to-T substitutions). All other types of substitutions are marked in grey.

Extended Data Figure 2 |. Fragment size distribution of fragments longer than 35 bp mapped to the human reference genome with MQ ≥ 25 for each of the five late Neanderthals.

All fragments are depicted in solid lines and fragments with C-to-T substitutions to the reference genome (putatively deaminated fragments) are depicted with dashed lines.

Extended Data Figure 3 |. Sex determination based on the number of fragments aligning to the X chromosome and the autosomes.

The expected ratios of X to (X + autosomal) fragments for a female and a male individual are depicted as dashed lines. The results were concordant for all fragments (in red) and for deaminated fragments only (in grey).

Extended Data Figure 4 |. Principal component analysis of the genomes of Vindija 33.19, Altai, the Denisovan individual, five late Neanderthals and Mezmaiskaya 1.

Genomes of the high-coverage archaics were used to estimate the eigenvectors of the genetic variation and low-coverage Neanderthals were projected onto the plane. Only transversion polymorphisms and bi-allelic sites were considered for the analysis, to a total of 1,010,417 sites as defined by the high-coverage genomes. PC, principal component.

Figure 1 |. Specimen information and the effects of 0.5% hypochlorite treatment.

a, Location and age of the five late Neanderthal specimens analysed in this study (new), and other sites for which genome-wide data of Neanderthal samples have previously been published (old). Map source: Vectormapcollection. b, Proportion of DNA fragments aligned to the human reference genome in untreated bone and tooth powder and in powder treated with 0.5% sodium hypochlorite (Supplementary Table 2.1). c, Proportion of present-day human contamination (with 95% binomial confidence intervals) inferred from mtDNA fragments in treated and untreated samples. Two-sided Pearson’s χ² tests were used to calculate significant differences (**α ≪ 0.001) (Supplementary Table 2.5).

Figure 2 |. Phylogenetic relationships of late Neanderthals.

a, Bayesian phylogenetic tree of mitochondrial genomes of 23 Neanderthals, 3 Denisovans, 64 modern humans and a hominin from Sima de los Huesos. The posterior probabilities for the branches are shown. b, Neighbour-joining tree of Y chromosome sequences of Mezmaiskaya 2, Spy 94a, 175 present-day humans and two present-day humans carrying the A00 haplogroup. The number of substitutions is shown above the branches. c, Neighbour-joining tree of nuclear genomes based on autosomal transversions among late Neanderthals, Vindija 33.19, Mezmaiskaya 1, Altai Neanderthal, Denisovan and 12 present-day humans. Bootstrap support values after 1,000 replications are shown.

Figure 3 |. Proximity to the introgressing Neanderthal populations in present-day and ancient humans calculated using D(Neanderthal1, Neanderthal2; non-African, African).

Three Mbuti individuals from the Simons Genome Diversity Project (SGDP) were used as an outgroup and standard errors were calculated using a weighted block jackknife (Supplementary Information 10). Shaded grey region corresponds to D < 0. a, All late Neanderthals and the older Mezmaiskaya 1 are significantly closer to the introgressing Neanderthal population(s) than the Altai Neanderthal. b, There is no significant difference between late Neanderthals, Mezmaiskaya 1 and Vindija 33.19 in their proximity to the introgressing Neanderthal population(s) in present-day and ancient humans.