Genomic Diversity and Admixture Differs for Stone-Age Scandinavian Foragers and Farmers

See allHide authors and affiliations

Science  16 May 2014:
Vol. 344, Issue 6185, pp. 747-750
DOI: 10.1126/science.1253448

Hunters and Farmers

The Neolithic period in Europe saw the transition from a hunter-gatherer lifestyle to farming. Previous genetic analyses have suggested that hunter-gatherers were replaced by immigrant farmers. Skoglund et al. (p. 747, published online 24 April) sequenced one Mesolithic and nine Neolithic Swedish individuals to examine the transition from hunter-gatherers to farmers. Substantial genetic differentiation was observed between hunter-gatherers and farmers: There was lower genetic diversity within the hunter-gatherers and gene flow from the hunter-gatherers into the farmers but not vice versa.


Prehistoric population structure associated with the transition to an agricultural lifestyle in Europe remains a contentious idea. Population-genomic data from 11 Scandinavian Stone Age human remains suggest that hunter-gatherers had lower genetic diversity than that of farmers. Despite their close geographical proximity, the genetic differentiation between the two Stone Age groups was greater than that observed among extant European populations. Additionally, the Scandinavian Neolithic farmers exhibited a greater degree of hunter-gatherer–related admixture than that of the Tyrolean Iceman, who also originated from a farming context. In contrast, Scandinavian hunter-gatherers displayed no significant evidence of introgression from farmers. Our findings suggest that Stone Age foraging groups were historically in low numbers, likely owing to oscillating living conditions or restricted carrying capacity, and that they were partially incorporated into expanding farming groups.

The impact of the adoption of agriculture in Europe—the Neolithic transition—has long been debated in archaeology (1) and was one of the first historical questions to be addressed by population geneticists (2). There are contrary schools of thought of the relative importance of migration versus cultural diffusion as the Neolithic lifestyle dispersed across Europe. Farming appeared in the Near East at least 10,000 years before present (B.P.) and spread into Europe via the Balkan and Mediterranean regions, arriving in Scandinavia ~6000 years B.P. (3). Analyses of mitochondrial DNA (46) and genomic sequences (79) from Stone Age human remains provide evidence that farming and foraging groups were genetically distinct and that admixture between these groups played a role in forming the genetic landscape of modern-day Europe. However, most studies so far have focused on the comparison of ancient individuals to recent groups, and the degree of direct genetic relatedness between ancient groups, gene flow, and within-group genomic variation remains unaddressed.

We generated between 0.01- and 2.2-fold genome-wide coverage for six Neolithic hunter-gatherers from the Pitted Ware Culture context (PWC), four Neolithic farmers from the Funnel Beaker culture context (TRB), and one late Mesolithic hunter-gatherer (Table 1). All of the Neolithic individuals were excavated in Sweden and were dated either directly or contextually to ~5000 years B.P. The late Mesolithic hunter-gatherer individual, directly dated to ~7500 years B.P., was excavated in Sweden in close proximity to the Neolithic hunter-gatherers. All generated sequence data showed evidence of nucleotide misincorporations (figs. S1 to S3) characteristic of postmortem degradation.

Table 1 Ancient genomic sequence data.

Only sequences with mapping quality of at least 30 were used. mtDNA, mitochondrial DNA; hg, haplogroup; Contam., Contamination estimate; CI, confidence interval; cal years BP, calibrated years BP. Dashes indicate no reliable estimate.

View this table:

The mitochondrial genome was covered to an average depth of at least 23-fold for seven of the individuals, and estimates of contamination were low (Table 1). All six Neolithic hunter-gatherers and the late Mesolithic hunter-gatherer were assigned to haplogroups U or V (Table 1), similar to other Mesolithic and Paleolithic hunter-gatherers (4, 8, 10). The four TRB individuals were assigned to haplogroups H and K, similar to previous results (5) and as observed in other Neolithic farmers in Europe (6). The Y chromosome of the Ajvide58 individual (Neolithic hunter-gatherer) was assigned to haplogroup I2a1 (table S5), a specific subhaplogroup uncommon in current-day Scandinavia (11). Recent studies suggest that variants associated with phenotypes and immune system genes differ between Stone Age hunter-gatherers and farmers, often from ancestral to derived variants (8, 9). However, we found little evidence of differentiation for these variants (tables S6 to S10) between Neolithic Scandinavian hunter-gatherers and farmers, except for the SLC24A5 locus involved in pigmentation. Ajvide58, like the Mesolithic Iberian LaBrana1 (8), carried the ancestral allele, and both the Tyrolean Iceman (9) and Gökhem2 carried the derived allele (table S8), which is associated with light pigmentation and nearly fixed in current-day Europe and common in current-day South Asia (12).

To investigate the ancient individuals’ relationship to the modern-day genetic landscape, we used principal component (PC) analysis with a set of 57 Western Eurasian populations (table S11), including individuals across modern-day Sweden, and projected the 11 Scandinavian Stone Age individuals together with the Mesolithic Iberian LaBrana1 (8), the Paleolithic Siberian MA1 (13), and the Chalcolithic (~5300 years B.P.) Tyrolean Iceman (Fig. 1A) (9). All six Neolithic hunter-gatherers clustered outside the variation of modern populations, which is consistent with a previous study (7). The Mesolithic Scandinavian and the Mesolithic Iberian sample also clustered together in this PC space, which is consistent with population continuity from late Mesolithic groups to Neolithic Scandinavian hunter-gatherers. The Mesolithic and the Scandinavian Neolithic hunter-gatherers do not cluster particularly close to any individuals self-identifying as Swedish Saami (14) or indigenous populations from central Russia. All Scandinavian farmer individuals clustered among southern and central European populations.

Fig. 1 Population structure and genetic affinities of ancient Eurasians.

(A) Ancient individuals were projected on PCs computed on 57 modern-day Western Eurasian groups. (B) D-statistics assessing affinity to either Lithuanians or Sardinians show stratification between all sequenced ancient Eurasian individuals (circles), which is robust for when only sequences with clear evidence of degradation are used (squares). Error bars represent 1 SE on each side (total 2 SEs). (C) Shared genetic drift between Western Eurasian populations and Ajvide58, a Neolithic hunter-gatherer from Scandinavia. (D) Shared genetic drift between Western Eurasian populations and Gökhem2, an early farmer from Scandinavia.

To test whether the North-South difference between farmers and hunter-gatherers was consistent for different individuals from each cultural context, we examined which modern-day population had the greatest genetic affinity to particular ancient individuals by identifying the modern-day population with the greatest shared genetic drift with each ancient individual using an outgroup f3-statistic (13, 15). The greatest shared genetic drift with Ajvide58 was observed for Lithuanians, whereas the greatest shared genetic drift with Gökhem2 was observed for Sardinians (Fig. 1, C and D). Both of these higher-coverage Neolithic individuals show negligible levels of contamination (Table 1).

We also computed D-statistics (15) to test whether each single Stone Age individual was closer to Lithuanians or Sardinians and found statistically significant segregation between all Stone Age farmers and hunter-gatherers (Fig. 1B); we found the same results when we restricted the analyses to sequences that displayed clear evidence of postmortem damage (16). Focusing on modern-day Swedish groups, which are intermediate to the two Scandinavian Neolithic groups (7), direct tests indicate that modern-day Swedish populations are closer to Ajvide58 (table S14). In addition to Sardinians, many current-day groups from the Near East and southern Europe are significantly closer to Gökhem2 than to Ajvide58 (17).

Because both Stone Age farmers and hunter-gatherers contributed genetic material to modern-day European groups (7), we investigated whether this putative admixture between ancestral populations had already begun in the Neolithic. We tested symmetry between ancient genomes (table S13) and found that although the Tyrolean Iceman is consistent with forming a group with modern-day Sardinians to the exclusion of Stone Age hunter-gatherers (D-statistics, |Z| < 2), the Scandinavian early farmer Gökhem2 is significantly closer to hunter-gatherers (2 < |Z| < 8 for all comparisons with >100,000 loci). In contrast, the Scandinavian hunter-gatherer Ajvide58, which postdates the arrival of agriculture in Scandinavia, shows no significant evidence of admixture (|Z| < 2) from early farmers when contrasted with an Iberian Mesolithic individual [LaBrana1, which predates agriculture in Iberia (table S13)]. This suggests that the ancestors of the group represented by Gökhem2 admixed with individuals from hunter-gatherer groups during the northward expansion, whereas the Scandinavian Neolithic hunter-gatherers show no meaningful evidence of admixture with incoming farmers despite at least 40 generations of coexistence in Scandinavia.

We fitted an admixture graph population model (17) to all publicly available ancient genomes of at least onefold coverage [as well as a version including modern-day individuals from Sardinia and France (17)] and incorporated the gene-flow events identified in the D-tests above as well as in previous studies (Fig. 2A). Ancestry related to the Upper Paleolithic Siberian MA1 is estimated to ~15% in Ajvide58 (table S15), which is not found in the more southerly LaBrana1, suggesting gene-flow or shared ancestry among Northern Eurasian hunter-gatherer populations. Gökhem2 and the Tyrolean Iceman had ancestry from a basal group separated from all other ancient genomes (Fig. 2B) (17, 18). However, the early farmers also have a substantial amount of ancestry related to European hunter-gatherers, which is significantly higher in Gökhem2 (77.2 ± 6.1%) as compared with the Tyrolean Iceman (56.0 ± 3.0%), which is consistent with the D-statistics results above.

Fig. 2 Sample locations and population history model.

(A) Sample locations of ancient human remains that are included in the population history model. (B) Admixture graph of population history fitted to ancient genomes, showing more hunter-gatherer admixture in Neolithic Scandinavian farmers than in central European farmers (table S15).

A possible contact phase in Northern Europe between expanding farming populations and hunter-gatherers that may explain the admixture in Gökhem2 is the period between ~7500 and 6000 years B.P., when the hunter-gatherer Ertebølle culture coexisted for more than a millennium with the early farming Linearbandkeramik (LBK) complex (19, 20). Hunter-gatherers and farmers also coexisted during the Early Neolithic expansion of agriculture into Scandinavia, which may have involved admixture between groups, but little is known about the populations from this period because skeletal remains are scarce.

Comparison of relative levels of genetic diversity in populations can provide information on effective population sizes because smaller populations often have reduced levels of diversity. We computed “conditional nucleotide diversity” (a relative measure of genetic diversity) for the Neolithic Scandinavian groups based on positions that are variable in single African individuals sequenced to high coverage (21) in order to avoid effects of post mortem degradation and sequence errors (17). To avoid effects of potential inbreeding, we always compared variants from separate individuals (17). The Scandinavian Neolithic hunter-gatherer group had significantly lower conditional nucleotide diversity (0.181 ± 0.0015) as compared with that of the Scandinavian Neolithic farmer group (0.201 ± 0.0038) (Fig. 3A and fig. S9). Although the specific properties of ancient DNA may still affect comparisons with sequence data from modern-day individuals, the conditional nucleotide diversity in the hunter-gatherers was also lower than in any modern-day European and a Chinese population (22) analyzed by using the same approach as for the ancient groups. In contrast, the diversity of the TRB is more similar to the level observed in modern-day populations (Fig. 3A).

Fig. 3 Population genomic diversity and differentiation between Stone Age farmers and hunter-gatherers contrasted with modern-day groups.

(A) Nucleotide diversity estimated as the average pairwise difference at transversion single-nucleotide polymorphisms (SNPs) identified in an African Yoruba individual. (B) Wright’s FST estimated at transversion SNPs identified in African Yoruba and Mbuti individuals. Error bars represent 1 SE on each side (total 2 SEs), estimated by using a block jackknife procedure across the genome, with 5 Mb in each block. IBS, Spanish; TSI, Tuscan; CEU, Northern and Western European ancestry; FIN, Finnish; GRB, British; CHB, Chinese.

The conditional approach also allowed us to estimate population genetic differentiation, Wright’s FST, to 0.057 ± 0.017 between TRB and PWC, and to 0.035 ± 0.0026 between pan-European hunter-gatherers (Ajvide52 and LaBrana1) and farmers (Gökhem2 and Iceman). Both of these values are larger than the maximum observed pairwise FST between European populations in the 1000 Genomes project, which was 0.013 ± 0.00059 between Finnish and Italian individuals (Fig. 3B). These estimates suggest greater genetic stratification among Stone Age European groups as compared with current-day groups of European ancestry (7) but that this stratification in Stone Age Europe was correlated with the mode of subsistence instead of geography, as in current-day Europe (23).

The distinct features of the two Neolithic Scandinavian groups—nonsymmetric gene-flow into farmers, low level of diversity among hunter-gatherers, and strong differentiation between groups—have important implications for our understanding of the demographic histories of these groups. The greater diversity in the farmer population may have been influenced by gene flow from hunter-gatherers. However, the low level of genetic diversity in Neolithic hunter-gatherers likely has a demographic explanation, similar to that of the Iberian Mesolithic individual (8). Although we cannot exclude that this low diversity is a feature restricted to the Gotland island hunter-gatherer population, the low diversity may be due to the fact that hunter-gatherer ancestors resided in ice-free refugia in Europe during the Last Glacial Maximum (LGM), potentially causing population bottlenecks. Climatic changes and occasional population crashes also likely affected the population sizes of hunter-gatherers (24, 25). Furthermore, mobility may have decreased among late hunter-gatherer groups, especially when settling in coastal areas (20). Meanwhile, the population ancestral to the Neolithic farmers, that later were to expand across Europe, resided in warmer areas that could sustain larger population sizes during the LGM. Although it is possible that climate also affected populations in southern Eurasia, it may have been in a different manner (26), and farming economies are associated with greater carrying capacity than those of hunter-gatherer economies. It is likely that several factors contributed to the different levels of genetic diversity, and disentangling these processes and assessing their generality in prehistoric Europe may be possible as more genomic data from a wider geographic and temporal range becomes available.

Supplementary Materials

Materials and Methods

Supplementary References

Figs. S1 to S9

Tables S1 to S16

References (27101)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: We thank L. Gattepaille, M. Metspalu, T. Naidoo, M. Rasmussen, S. Rasmussen, D. Reich, E. Salmela, and P. Wallin for technical assistance and discussions. The SNP data for the modern-day Swedish individuals is available from P. Hall at the Karolinska Institute under a materials transfer agreement. We thank the late L. Beckman for contributing the Saami samples; K.T. ( retains governance over the Saami and the Mari samples. This project was supported by grants from the Nilsson-Ehle Foundation (P.S.), Helge Ax:son Foundation (P.S.), Royal Swedish Academy of Science (FOA12H-177 to P.S.), Danish National Research Foundation (E.W. and M.R.), Swedish Research Council (J.A. and J.S.), Berit Wallenberg Foundation (J.A. and J.S.), Wenner-Gren Foundations (T.G.) and European Research Council (M.J.). P.S., H.M., A.G., and M.J. conceived and designed the study. E.W., J.S., A.G., and M.J. supervised the study. H.M., A.O., M.R., and C.V. generated DNA sequence data from ancient human remains. P.S. processed and prepared the data. P.S., H.M., and T.G. analyzed the genetic data, supervised by M.J. with the following contributions: mtDNA and Y-chromosome (H.M. and P.S.); functional SNPs (T.G. and P.S.); and contamination, biological sex, and population genetic analyses (P.S.). K.-G.S., J.S., and J.A. provided archaeological information and interpretation. P.H., K.T., and J.P. contributed samples. P.S., H.M., J.S., A.G., and M.J. wrote the manuscript with input from all authors. Data are available from the European Nucleotide Archive under accession no. PRJEB6090, and data aligned to the human reference genome are available at The authors declare no competing interests.
View Abstract

Navigate This Article