Norway spruce postglacial recolonization of Fennoscandia

Norway spruce (Picea abies) is a key boreal tree in northern Europe, yet its postglacial colonization of Fennoscandia remains debated. Two main hypotheses exist: (1) a late Holocene migration from the east via Finland, as inferred from pollen records that show spruce arriving in central Sweden 2–3 cal. kyr BP; and (2) an earlier establishment from western microrefugia or nunataks shortly after deglaciation, based on sparse Lateglacial pollen, megafossils, and some ancient DNA. However, pollen evidence is often scarce and ambiguous for local presence, and there is no consensus on viable colonization routes given the persistence of the Scandinavian Ice Sheet into the early Holocene. Clonal, millennia-old spruce trees found above the current treeline in the central Scandes suggest early Holocene establishment but their origin is unresolved. The study aims to clarify when and from where Norway spruce colonized Fennoscandia after the last glaciation by combining ancient sedimentary DNA (sedDNA) with modern nuclear and mitochondrial DNA from clonal and nearby forest spruce populations.

Prior work shows Norway spruce has two genetically divergent European domains (northern and southern) with ancient divergence (~15 Mya). Pollen syntheses indicate a late Holocene east-to-west expansion into central Sweden, whereas some studies report low-abundance Lateglacial spruce pollen in southern Sweden and near the SIS margin, often dismissed as long-distance transport. Megafossils and ancient DNA suggest early presence in the Scandes (~13 cal. kyr BP), prompting hypotheses of western microrefugia (nunataks) or west-coast survival. This has been challenged due to limited pollen support and ecological constraints for spruce persistence west of the SIS. Modern genomic studies rule out substantial contributions from southern Alpine/Carpathian refugia to Fennoscandian recolonization, and indicate a complex spread with major eastern sources and a contact zone within central Scandinavia. Calls have been made to use multi-site ancient DNA to resolve the timing and routes of spruce recolonization.

Ancient sedDNA: Sediment and peat records from 15 sites across Fennoscandia and Russia were analyzed, including six lakes in the central Scandes (Sweden), two sites in southern Sweden (covering the Lateglacial–Early Holocene), four in Finland (including a site dating to ~43 cal. kyr BP), and two in Russia. Clean-room protocols were used for subsampling and DNA extraction (modified DNeasy PowerSoil protocol with bead-beating, proteinase K and DTT additions, overnight incubation; duplicate elutions). Inhibition tests were performed using spiked synthetic oligos in qPCR to guide dilution. Species-specific detection used a qPCR melting-curve assay targeting the mitochondrial mh05 locus with short amplicons (64/85 bp) distinguishing haplotype A (species-specific to P. abies) from haplotype B (shared with P. abies and Pinus sylvestris). Each sample was run in eight qPCR replicates with multiple negative controls; subsets were Sanger sequenced for validation. Potential reworking at basal contacts was avoided.

Modern DNA sampling: Needles were collected from 7 central Scandes P. abies populations, including clonal stands above treeline and adjacent forest at lower elevations in some sites. In total, 264 trees were genotyped (135 clonal, 129 forest), with 7 clonal trees previously radiocarbon-dated (4.4–9.5 cal. kyr BP). External reference samples included P. obovata and P. abies from distinct European clusters.

Nuclear genotyping: Genome-wide SNPs were generated via MIG-seq (two-step PCR; modified annealing; Illumina MiSeq paired-end 150). Reads were mapped to the P. abies v1.0 genome; SNPs were called with GATK and filtered for quality, minor allele frequency, and linkage disequilibrium, resulting in 23,130 SNPs (14,969 unlinked used). Population structure was analyzed by PCA (Eigensoft) and Admixture; pairwise FST (StAMPP) and isolation-by-distance were assessed versus geographic distances (geosphere), with visualizations in ggplot2. Chloroplast-assigned loci were removed; analyses were repeated under varying missing data thresholds.

Modern mitochondrial genotyping: The same mh05 qPCR melting-curve assay was applied to modern samples (with Sanger confirmation for ambiguous peaks) to estimate haplotype A/B frequencies across clonal and forest populations.

• Ancient presence: Norway spruce was present in southern Fennoscandia by 14.7 ± 0.1 cal. kyr BP, pushing back prior earliest records by ~1.7 kyr.
• Central Scandes: Haplotype A (P. abies-specific) was detected throughout the Holocene in all six central Scandes lakes (except the very oldest ~10.3 cal. kyr BP levels where only haplotype B was recovered), indicating long-term local presence despite low pollen.
• Wider region: Haplotype A was detected in southern Finland (~5–7 cal. kyr BP) and east Finnish Lapland (9.3 cal. kyr BP), and notably in northeast Finland at ~43 cal. kyr BP (MIS 3), but absent in the two Russian sites sampled. This shows haplotype A was not exclusive to a western refugium.
• Detection statistics: Across 1,176 qPCRs on sedDNA, there were 54 haplotype A positives and 72 haplotype B positives; haplotype A was significantly rarer in negative controls (1/208 extraction negatives; 1/272 qPCR negatives), supporting authenticity.
• Modern nuclear genomics: PCA and Admixture revealed two main ancestral clusters aligned with latitude; strong isolation by distance was observed (pairwise FST correlated with geographic distance, Pearson r > 0.8). Clonal trees were genetically indistinguishable from nearby forest trees, indicating shared origins; only one clonal individual (Lsk-19) was an outlier on PC2.
• Modern mitochondrial frequencies: Haplotype A frequencies were nearly fixed in two Härjedalen populations, high in Fulufjället (0.88), moderate in Dalarna (0.44–0.65), and lower in two of three Jämtland populations (0.25 and 0.45). Clonal and forest populations had nearly identical haplotype frequencies.
• Dated clones: Four of seven dated clonal trees carried haplotype A (e.g., Old Tjikko 9.5 cal. kyr BP; GS Spruce 6.3 cal. kyr BP; others 4.8–4.4 cal. kyr BP); three carried haplotype B (6.4–7.9 cal. kyr BP).

The sedDNA evidence demonstrates an early presence of Picea abies in southern Sweden during the Lateglacial and continuous Holocene presence in the central Scandes, consistent with early macrofossil finds but not readily visible in pollen records due to likely small, sparse, and clonally reproducing populations producing limited pollen. The occurrence of mitochondrial haplotype A in MIS 3 sediments in northeast Finland and early–mid Holocene Finland indicates that this haplotype was not confined to a western refugium, weakening the hypothesis of a purely western source. Modern nuclear and mitochondrial data show that ancient clonal trees and nearby forests share the same major eastern-derived genetic clusters, with strong isolation by distance and no distinct signature for a western refugium. The most parsimonious scenario is that early colonizers reached central Fennoscandia from the east as ice retreated—potentially via unusual long-distance dispersal mechanisms (e.g., seeds transported on ice-crusted snow or by water)—establishing small outposts that persisted for millennia, followed by a later, more extensive east-to-west expansion in the late Holocene. While survival west of the SIS cannot be entirely excluded, any western contribution appears to have had negligible genetic impact on contemporary populations.

This study integrates ancient sedimentary DNA with modern nuclear and mitochondrial markers to reconstruct Norway spruce recolonization of Fennoscandia. It shows spruce present in southern Fennoscandia by 14.7 ± 0.1 cal. kyr BP and sustained Holocene presence in the central Scandes, with clonal trees sharing genetic ancestry with surrounding forests derived from eastern sources. Haplotype A’s presence in MIS 3 northeast Finland and early Holocene Finland indicates it was not unique to a western refugium. The authors propose a refined colonization scenario: initial Lateglacial–early Holocene establishment from the east (possibly aided by snow/ice or water-mediated seed dispersal) forming small, persistent populations, followed by a late Holocene east-to-west expansion. Ancient sedDNA emerges as a powerful complement to pollen and macrofossils for detecting low-abundance, locally established taxa. Future work should target additional Lateglacial and MIS 3 sites around SIS margins, seek macrofossil corroboration in southern Scandinavia and the Baltic–Doggerland region, and refine temporal continuity with higher-resolution sedDNA sampling.

• Temporal gaps in sediment cores (chronological grey zones) limit inference of continuous presence, particularly in southern Sweden.
• Potential reworking of older DNA near basal contacts necessitated exclusion of basal sediments; deeper archives remain uncertain.
• Low detection rates and small sample sizes at some sites reduce power; sedDNA inherently captures sparse signals for rare taxa.
• Absence of macrofossil/megafossil corroboration for proposed southern SIS-margin survival (e.g., Denmark/Doggerland) leaves this aspect speculative.
• Western refugial survival cannot be entirely ruled out, though modern genetic data suggest minimal impact.
• Geographic coverage, especially in Russia and the Baltic region, was limited relative to the breadth of potential source areas.