A 2-million-year-old ecosystem in Greenland uncovered by environmental DNA

The study investigates the composition and ecology of Arctic biotic communities during the Late Pliocene–Early Pleistocene, a time of strong polar amplification with mean annual temperatures far above today. Fossil records from the High Arctic are sparse, leaving a major gap in knowledge of the flora and fauna that inhabited these regions under warmer climates. The authors target the Kap København Formation in Peary Land, North Greenland—now a polar desert but during an interglacial of the Early Pleistocene an estuarine to nearshore setting receiving terrestrial input—to reconstruct past ecosystems using ancient environmental DNA. Prior macrofossil records from the site indicated a coniferous boreal forest and rich insect fauna with few vertebrate remains, and comparable High Arctic sites in Arctic Canada preserve mammals that could have colonized Greenland. The purpose is to use sedimentary eDNA alongside stratigraphic and geochronological constraints to reveal plant and animal assemblages, assess climate and habitat conditions, and test the preservation mechanisms allowing DNA survival over ~2 million years.

Decades of work at Kap København established stratigraphy and palaeoenvironments, including macrofossil plants, insects, and marine microfossils, and constrained age using palaeomagnetism, biostratigraphy, and allostratigraphy, narrowing deposition to an Early Pleistocene interglacial around ~2.4 Ma. Comparable Pliocene sites (e.g., Fyles Leaf Bed, Beaver Pond on Ellesmere Island) preserve mammals (e.g., Protarctos, Dipoides, Eucyon, camelines), raising questions about biogeographic barriers like the Nares Strait. Previous studies documented strong Pliocene warmth and polar amplification and used environmental DNA from younger sediments to reconstruct past environments. The oldest DNA prior to this work was ~1.2–1.7 Ma from mammoths. Pollen-macrofossil records at Kap København and elsewhere in the Arctic established mixed forest-tundra assemblages but left gaps in vertebrate representation and some plant taxa detection. DNA preservation literature suggests adsorption to mineral surfaces (e.g., clays, smectites, quartz) can protect DNA from degradation, and thermal and chemical decay kinetics constrain ancient DNA survival.

Site and stratigraphy: The Kap København Formation comprises Member A (offshore glaciomarine laminated muds) and Member B (nearshore/estuarine sandy and silty units B1–B3 with organic-rich beds). Mineralogy was characterized by X-ray diffraction; proximal B3 has lowest clay and highest quartz; clay minerals (including smectite) quantified by weight percent. Age constraints: New palaeomagnetic data show Member A reversed polarity and most of B2 normal polarity, consistent with Early Pleistocene subchrons. Biostratigraphic last appearance data for foraminifera, mammals, and molluscs indicate ~2.4 Ma. Cosmogenic 26Al/10Be burial dating at four sites yields a recommended maximum burial age 2.70 ± 0.46 Ma; continuous deposition across Members A–B rules out older scenario; remaining scenarios support ~1.9 or ~2.1 Ma. DNA preservation tests: Thermal age and depurination rates estimated using local mean annual temperature (−17 °C) indicated plausible survival of short fragments (~50 bp) if frozen. Adsorption experiments measured DNA binding capacities of minerals representative of the sediments. At a natural DNA concentration (4.9 ng ml−1), smectite adsorbed ~200× more DNA than quartz. Extraction tests using a sedimentary eDNA protocol retrieved ~5% of DNA from smectite, ~10% from other clays, and ~40% from quartz, implying mineralogy affects preservation and recovery. Sampling and lab procedures: DNA extracted from 41 organic-rich sediment samples across five sites, yielding 65 dual-indexed Illumina libraries. ddPCR screened 34 libraries for plastid genes (psbD; psbA for Poaceae); 31/34 were positive for plant plastid DNA. A subset of 34 libraries underwent mammalian mitochondrial DNA capture using Arctic PaleoChip 1.0. All libraries (initial and captured) were shotgun sequenced on Illumina HiSeq 4000 and NovaSeq 6000, generating 16,882,114,068 reads; after processing (adapter trimming, ≥30 bp length, Phred ≥30, duplicate removal), 2,873,998,429 reads remained. Bioinformatics and taxonomic assignment: Reads analyzed with multiset k-mer comparisons (simka), then competitively mapped using HOLI with an expanded plant genome skim reference (>1,500 Arctic and boreal taxa). For taxonomic classification (ngsLCA), initial similarity thresholds of 95–100% were used; for ancient chloroplast placements (e.g., Betula, Populus, Salix), thresholds lowered to 90% and reads merged by unit to improve coverage. Ancient DNA authenticity filtering used metaDMG (v0.14.0), estimating terminal damage (D-max) and likelihood ratio (Δ-LR). Thresholds: D-max ≥25%, Δ-LR ≥1.5; taxa below read-abundance thresholds (based on medians) and with fewer than three replicates were filtered; reads normalized to proportions. Pollen counts: Pollen counted in six samples from locality 119, unit B3, with percentages computed for four samples (pollen sums 71–225). Comparison across DNA, pollen, and macrofossil records assessed overlap and taphonomic biases. Phylogenetic placement and molecular dating: Chloroplast reads for Betula, Populus, Salix mapped to modern references; Betula reads, placed basally and with high coverage (mean 24.16× for Betula; 57.06× Populus; 27.04× Salix), were dated using BEAST v1.10.4, calibrating the Betula–Alnus split at 61.1 Ma; robust across priors and models, yielding a median age of 1.323 Ma (95% HPD 0.6786–2.0172 Ma). Animal mitochondrial reads for Elephantidae (including Mammut in NCBI taxonomy), Cervidae (Rangifer), Leporidae (near base of Eurasian hares), Cricetidae (Arvicolinae), and Anatidae (basal to Branta) were phylogenetically placed; mastodon mitogenome dated with BEAST using two approaches (radiocarbon-only and radiocarbon plus molecular clock estimates). Marine planktonic eukaryotes were detected by aligning to Tara Oceans SMAGs, assessing present-day oceanic distributions and DNA damage patterns.

- Ancient environmental DNA from Kap København sediments documents an open boreal forest ecosystem around 2 million years ago with mixed vegetation including poplar (Populus), birch (Betula), and thuja (Thuja), alongside diverse Arctic and boreal shrubs and herbs. Many taxa were not previously detected by macrofossils or pollen. - Animal DNA indicates the presence of hare (Leporidae), reindeer/caribou (Rangifer), rodents (Arvicolinae), geese (Anatidae, basal to Branta), and mastodon (Mammut), implying viable populations and higher habitat productivity and diversity than inferred from sparse vertebrate fossils. - Marine signals include Limulidae (horseshoe crab), a reef-building coral (Merulinidae), and multiple planktonic eukaryote SMAGs (Opisthokonta, Stramenopila, Archaeplastida), consistent with warmer sea surface conditions; horseshoe crab spawning suggests warmer waters than today at that latitude. - Plant diversity: Of 175 vascular plant genera native to Greenland today, 70 (40%) were detected; 102 plant genera identified in total, with 39% no longer occurring in Greenland (e.g., Picea, Populus, Crataegus, Taxus, Thuja, Filipendula). Dominant reads: Salix, Dryas, Vaccinium, Betula, Carex, Equisetum. - Cross-record concordance: 76% of taxa identified to genus/family by DNA also occurred in macrofossil and/or pollen assemblages from the same units. DNA detected 24 plant taxa absent from macrofossils/pollen at the site but known from other Arctic Pliocene/Pleistocene localities. - Sequencing output: 16.88 billion raw reads; 2.874 billion high-quality, deduplicated reads analyzed. Eukaryotic DNA showed high damage (mean D-max ~40.7%). - Geochronology: Palaeomagnetism supports Early Pleistocene subchrons; cosmogenic 26Al/10Be burial dating gives a maximum burial age 2.70 ± 0.46 Ma. Geological considerations reject the oldest scenario; remaining age scenarios indicate ~1.9 or ~2.1 Ma. Betula chloroplast molecular date: median 1.323 Ma (95% HPD 0.6786–2.0172 Ma). Mastodon mitogenome median age estimates: 1.2 Ma (95% HPD 0.191–3.27 Ma) using only radiocarbon-dated specimens; 5.2 Ma (95% HPD 1.64–10.1 Ma) when including molecular-dated specimens. - DNA preservation mechanism: Clay minerals, especially smectite (1.2–3.7 wt%), have much higher DNA adsorption capacity than non-clay minerals (59–75 wt%); smectite ~200× quartz at 4.9 ng ml−1 DNA. Extraction recovery differed by mineral (smectite ~5%, other clays ~10%, quartz ~40%), underscoring mineralogical control on preservation and retrieval. - Climate/ecology: Assemblage has no modern analogue; combined terrestrial and marine signals indicate warmer conditions and mixed boreal–Arctic communities at very high latitude.

The findings demonstrate that sedimentary ancient DNA can preserve and resolve complex biotic communities from around 2 million years ago, addressing the long-standing gap in Arctic palaeoecology caused by scarce vertebrate fossils. The reconstructed Kap København ecosystem—an open boreal forest intermixed with Arctic taxa—reveals higher productivity and habitat heterogeneity than previously inferred, necessitating revisions to earlier temperature and permafrost reconstructions. The detection of mastodon implies suitable boreal browse and seasonal habitats at extreme latitude, while Rangifer, hares, rodents, and geese reflect herbivore-dominated biomass and taphonomic biases favoring herbivores over carnivores. Marine indicators, notably horseshoe crab and temperate–cosmopolitan planktonic lineages co-occurring with Arctic taxa, support warmer coastal waters consistent with Pliocene–Early Pleistocene warmth and polar amplification. Mineralogical analyses explain the exceptional DNA longevity via adsorption to clay surfaces, bolstering confidence in authenticity confirmed by elevated damage patterns. Integration of palaeomagnetism, cosmogenic burial ages, and molecular phylogenetics supports an age near 2 Ma, pushing the limits of ancient DNA-based ecosystem reconstructions and enabling phylogenetic placement and dating of lineages from environmental samples.

This study presents the oldest environmental DNA-based reconstruction to date, revealing a unique Early Pleistocene boreal–Arctic ecosystem in North Greenland with trees (Betula, Populus, Thuja), diverse shrubs and herbs, and vertebrates including mastodon and reindeer, alongside marine taxa indicative of warmer seas. The work demonstrates that ancient eDNA bound to minerals can survive for ~2 million years and can be used to recover, authenticate, and phylogenetically place taxa, substantially enriching palaeoenvironmental reconstructions beyond macrofossils and pollen. Similar deposits elsewhere may preserve comparably detailed records, offering opportunities to refine our understanding of climate variability, species biogeography, and ecosystem function during warm intervals of the Early Pleistocene across the High Arctic. Future research should expand reference genomes (especially for extinct and underrepresented lineages), target additional sites, and further quantify mineralogical controls on DNA preservation and recovery to improve taxonomic resolution and dating precision.

- Reference database limitations likely caused overclassification or underdetection for poorly represented groups (e.g., bryophytes, some marine metazoans) and may misplace extinct or unsampled lineages at higher taxonomic levels. - Vertebrate DNA was comparatively sparse, with limited mitochondrial coverage for several taxa; placements sometimes relied on few informative SNPs (e.g., mastodon placement based on two transitions), reducing confidence. - Molecular dating of the mastodon mitogenome yielded wide and method-dependent HPDs, reflecting uncertainties in calibration and lineage rate variation. - Taphonomic and depositional processes (alluvial concentration, differential decay resistance) can inflate representation of some taxa (e.g., Salix, Betula, Populus, Carex, Equisetum) and spores (ferns, club mosses), making read counts or pollen percentages imperfect proxies for biomass. - Mineral-dependent extraction biases (low recovery from clays) can skew observed community composition and reduce detectability of low-abundance taxa. - Age constraints, while convergent, retain scenario ambiguity (~1.9 vs ~2.1 Ma) and cosmogenic burial ages provide maximum ages only. - Rare taxa are challenging to detect across both DNA and macrofossil records, limiting completeness of assemblage reconstructions.