Million-year-old DNA sheds light on the genomic history of mammoths

Abstract

Temporal genomic data hold great potential for studying evolutionary processes such as speciation. However, sampling across speciation events would, in many cases, require genomic time series that stretch well back into the Early Pleistocene subepoch. Although theoretical models suggest that DNA should survive on this timescale¹, the oldest genomic data recovered so far are from a horse specimen dated to 780-560 thousand years ago². Here we report the recovery of genome-wide data from three mammoth specimens dating to the Early and Middle Pleistocene subepochs, two of which are more than one million years old. We find that two distinct mammoth lineages were present in eastern Siberia during the Early Pleistocene. One of these lineages gave rise to the woolly mammoth and the other represents a previously unrecognized lineage that was ancestral to the first mammoths to colonize North America. Our analyses reveal that the Columbian mammoth of North America traces its ancestry to a Middle Pleistocene hybridization between these two lineages, with roughly equal admixture proportions. Finally, we show that the majority of protein-coding changes associated with cold adaptation in woolly mammoths were already present one million years ago. These findings highlight the potential of deep-time palaeogenomics to expand our understanding of speciation and long-term adaptive evolution.

Extended Data Fig. 1. Mammoth molars and morphometric comparisons.

a-b, upper third molars in lateral and cross-sectional views; c, partial lower third molar in lateral and occlusal views. a, Chukochya (PIN-3341-737); b, Krestovka (PIN-3491-3) flipped horizontally; c, Adycha (PIN-3723-511), occlusal view flipped horizontally. Note the more closely-spaced lamellae and thinner enamel in a (primigenius-like) than b and c (trogontherii-like). d, Hypsodonty index vs lamellar length index of upper M3s; e, Enamel thickness index vs basal lamellar length index of lower M3s. Olyorian specimens yielding DNA are labelled by site name. Green dashed line: convex hull summarising Early to early Middle Pleistocene (ca. 1.5-0.5 Ma) North American Mammuthus samples (data points not shown). Green and blue squares: Early and Late Olyorian North-East Siberian samples, respectively; red and green circles: European M. meridionalis and M. trogontherii, respectively; blue circles, M. primigenius from North-East Siberia and Alaska. Note (i) similarity of Krestovka and Adycha to other Early Olyorian molars and to European steppe mammoths (M. trogontherii), (ii) similarity of early North American mammoths to these (Early Olyorian in particular), (iii) similarity of Chukochya to M. primigenius. For site details, measurement definitions and data, see Supplementary Section 1.

Extended Data Fig. 2. Sample age based on biostratigraphy, paleomagnetic reversals and genomic data.

Chart shows the stratigraphic position of the Kutuyakhian fauna, Phenacomys complex, Early Olyorian and Late Olyorian faunas in relation to important European, northwest Asian and northern North American stratigraphic benchmarks. ELMA - European Land Mammal Ages (small mammals), LMA - Land Mammal Ages (large mammals), MN/MQ - European Small Mammal Biozones, EEBU – East European biochronological units. Biostratigraphic and palaeomagnetic based chronological constraints for the specimens are provided, in comparison with the DNA-based age estimations.

Extended Data Fig. 3. DNA fragment length distributions for nine mammoths.

Reads are aligned to the LoxAfr4 autosomes. For the three Early-Middle Pleistocene samples (Krestovka, Adycha, Chukochya), reads of 25-200 bp length are shown, whereas 30-200 bp reads are shown for the remaining samples. Ultrashort reads (<35 bp) are denoted in red and were shown to be enriched for spurious alignments and therefore excluded from downstream analyses (Supplementary Section 4). The mean read lengths (μ) were calculated using only the retained reads (≥35 bp).

Extended Data Fig. 4. Post-mortem cytosine deamination damage profiles at CpG sites.

The most ancient samples (Krestovka, Adycha, Chukochya) carry a greater frequency of cytosine deamination compared to younger permafrost preserved woolly mammoth samples (Oimyakon and Wrangel) and the Columbian mammoth (M. columbi) specimen.

Extended Data Fig. 5. F(A|B) statistics.

The statistics reflect relative divergence between the genomes on the left and the right side. Lower values indicate reduced derived allele sharing between the sample indicated on the left and the right of the graph, at sites for which the genome on the right panel is heterozygous. The lower the value, the more drift has occurred between the genomes and thus the older their genetic divergence.

Extended Data Fig. 6. qpGraph model.

The most parsimonious graph model (highest Bayes Factor) of the phylogenetic relationships among mammoths lineages augmented with one admixture event. Branch lengths are given in f-statistic units multiplied by 1,000. Discontinuous lines show admixture events between lineages, with percentages representing admixture proportions.

Extended Data Fig. 7. Ghost introgression analysis of the Columbian mammoth genome.

a, The number of private alleles per 1000 bp within genomic regions identified as woolly mammoth (M. primigenius) ancestry or ghost ancestry. b, Maximum-likelihood phylogenies for those genomic regions identified as ghost ancestry in the Colombian mammoth (M. columbi) genome. c, Maximum-likelihood phylogenies for regions identified as un-admixed ancestry.

Fig. 1. DNA-based phylogenies and specimen age estimates.

a, Geographic origin of the mammoth genomes analysed in this study. b, Phylogenetic tree built in FASTME based on pairwise genetic distances, assuming balanced minimum evolution using all nuclear sites as well as 100 resampling replicates based on 100,000 sites each. c, Bayesian reconstruction of the mitochondrial tree, with the molecular clock calibrated using radiocarbon dates of ancient samples for which a finite radiocarbon date was available, as well as assuming a lognormal prior on the divergence between the African savannah elephant (not shown in the tree) and mammoths with a mean of 5.3 Ma. Blue bars reflect 95% highest posterior densities. Circles depict the position of the newly sequenced genomes. d, Densities for age estimates of samples Adycha and Chukochya based on autosomal divergence to African savannah elephant (L. africana) and e, Densities for age estimates of samples Krestovka, Adycha and Chukochya based on mitochondrial genomes as inferred from the Bayesian mitochondrial reconstruction.

Fig. 2. Inferred genomic history of mammoths.

a, D-statistics where each dot reflects a comparison involving one woolly mammoth genome and one genome depicted on the right side of the panel (where L. africana = African savannah elephant, P. antiquus = straight-tusked elephant, Mammuthus sp. = all mammoth specimens in this study, M. columbi = Columbian mammoth, and M. primigenius = woolly mammoth), iterating through all possible sample combinations using the mastodon (Mammut americanum) as an outgroup. No elevated allele sharing between any of the mammoth genomes and the reference (African savannah elephant) is observed, suggesting no pronounced reference biases in the Early/Middle Pleistocene genomes. A strong affinity between Columbian mammoths and sample Krestovka is observed, as well as a relationship between the North American woolly mammoth (Wyoming) and the Columbian mammoth. b, Best fitting admixture graph model for one admixture event, suggesting a hybrid origin for the Columbian mammoth. c, Hypothesized evolutionary history of mammoths during the last 3 Ma, based on currently available genomic data. Brown dots represent mammoth specimens for which genomic data has been analysed in this study, with error bars representing 95% highest posterior density intervals from the mitogenome-based age estimates obtained for the three Early and Middle Pleistocene specimens. Arrows depict gene flow events identified from the autosomal genomic data. The European steppe mammoth (M. trogontherii) survived well into the later stages of the Middle Pleistocene, and we hypothesize that it most likely branched off from a common ancestor shared with the woolly mammoth at ~1 Ma.