Author Correction: Population genomics of the Viking world

Abstract

A Correction to this paper has been published: https://doi.org/10.1038/s41586-021-03328-2.

Full text

390 | Nature | Vol 585 | 17 September 2020
Article
Population genomics of the Viking world
Ashot Margaryan1,2,3,71, Daniel J. Lawson4,5,71 , Martin Sikora1,71, Fernando Racimo1,71,
Simon Rasmussen6, Ida Moltke7, Lara M. Cassidy8, Emil Jørsboe7,9, Andrés Ingason1,1 0,11 ,
Mikkel W. Pedersen1, Thorinn Korneliussen1,12, Helene Wilhelmson1 3,14, Magdalena M. Buś15,
Peter de Barros Damgaard1, Rui Martiniano16, Gabriel Renaud1,17, Claude Bhérer18,
J. Víctor Moreno-Mayar1,19, Anna K. Fotakis3, Marie Allen15, Raili Allmäe20, Martyna Molak21,
Enrico Cappellini3, Gabriele Scorrano3, Hugh McColl1, Alexandra Buzhilova22, Allison Fox23,
Anders Albrechtsen7, Berit Schütz24, Birgitte Skar25, Caroline Arcini26, Ceri Falys27,
Charlotte Hedenstierna Jonson28, Dariusz Błaszczyk29, Denis Pezhemsky22,
Gordon Turner-Walker30, Hildur Gestsdóttir31, Inge Lundstrøm3, Ingrid Gustin13,
Ingrid Mainland32, Inna Potekhina33, Italo M. Muntoni34, Jade Cheng1, Jesper Stenderup1,
Jilong Ma1, Julie Gibson32, Jüri Peets20, Jörgen Gustafsson35, Katrine H. Iversen6,17,
Linzi Simpson36, Lisa Strand25, Louise Loe37, Maeve Sikora38, Marek Florek39, Maria Vretemark40,
Mark Redknap41, Monika Bajka42, Tamara Pushkina43, Morten Søvsø44, Natalia Grigoreva45,
Tom Christensen46, Ole Kastholm47, Otto Uldum48, Pasquale Favia49, Per Holck50, Sabine Sten51,
Símun V. Arge52, Sturla Ellingvåg1, Vayacheslav Moiseyev53, Wiesław Bogdanowicz21,
Yvonne Magnusson54, Ludovic Orlando55, Peter Pentz46, Mads Dengsø Jessen46,
Anne Pedersen46, Mark Collard56, Daniel G. Bradley8, Marie Louise Jørkov57, Jette Arneborg46,58,
Niels Lynnerup57, Neil Price28, M. Thomas P. Gilbert3,59, Morten E. Allentoft1,60 , Jan Bill61,
Søren M. Sindbæk62, Lotte Hedeager63, Kristian Kristiansen64, Rasmus Nielsen1,65,66 ✉,
Thomas Werge1,10,11,67 ✉ & Eske Willerslev1,68,69,70 ✉
The maritime expansion of Scandinavian populations during the Viking Age (about
750–1050) was a far-ung transformation in world history1,2. Here we sequenced
the genomes of 442humans from archaeological sites across Europe and Greenland
(to a median depth of about 1×) to understand the global inuence of this expansion.
We nd the Viking period involved gene ow into Scandinavia from the south and east.
We observe genetic structure within Scandinavia, with diversity hotspots in the south
and restricted gene ow within Scandinavia. We nd evidence for a major inux of
Danish ancestry into England; a Swedish inux into the Baltic; and Norwegian inux
into Ireland, Iceland and Greenland. Additionally, we see substantial ancestry from
elsewhere in Europe entering Scandinavia during the Viking Age. Our ancient DNA
analysis also revealed that a Viking expedition included close family members. By
comparing with modern populations, we nd that pigmentation-associated loci have
undergone strong population dierentiation during the past millennium, and trace
positively selected loci—including the lactase-persistence allele of LCT and alleles of
ANKA that are associated with the immune response—in detail. We conclude that the
Viking diaspora was characterized by substantial transregional engagement: distinct
populations inuenced the genomic makeup of dierent regions of Europe, and
Scandinavia experienced increased contact with the rest of the continent.
The events of the Viking Age altered the political, cultural and demo-
graphic map of Europe in ways that are evident to this day. Scandinavian
diasporas established trade and settlements that stretched from the
American continent to the Asian steppe
1
. They exported ideas, tech-
nologies, language, beliefs and practices to these lands, developed
new socio-political structures and assimilated cultural influences2.
To explore the genomic history of the Viking Age, we shotgun-
sequenced DNA extracted from 442human remains from archaeo-
logical sites dating from the Bronze Age (about 2400) to the Early
Modern period (about 1600) (Fig.1, Extended Data Fig.1). The data
from these ancient individuals were analysed together with published
data from 3,855present-day individuals across two reference panels
(Supplementary Note6), and data from 1,118ancient individuals (Sup-
plementary Table3).
Scandinavian ancestry and Viking Age origins
Although Viking Age Scandinavian populations shared a common
cultural background, there was no common word for Scandinavian
identity at this time1. Rather than there being a single ‘Viking world’, a
series of interlinked Viking worlds emerged from rapidly growing mari-
time exploration, trade, war and settlement, following the adoption
https://doi.org/10.1038/s41586-020-2688-8
Received: 12 July 2019
Accepted: 21 May 2020
Published online: 16 September 2020
Check for updates
A list of aff iliations ap pears at the end o f the paper.

Nature | Vol 585 | 17 September 2020 | 391
of deep-sea navigation among coastal populations of Scandinavia and
the area around the Baltic Sea3,4. Thus, it is unclear to what extent the
Viking phenomenon refers to people with a recently shared genetic
background or how far population changes accompanied the transition
from the Iron Age (500–700) to the Viking Age in Scandinavia.
The Viking Age Scandinavian individuals of our study fall broadly
within the diversity of ancient European individuals from the Bronze
Age and later (Fig.2, Extended Data Figs.2, 3, Supplementary Note8),
but with subtle differences among the groups that indicate complex
fine-scale structure. For example, many Viking Age individuals from
the island of Gotland cluster with Bronze Age individuals from the
Baltic region, which indicates mobility across the Baltic Sea (Fig.2,
Extended Data Fig.3). Using f4-statistics to contrast genetic affinities
with steppe pastoralists and Neolithic farmers, we find that Viking Age
individuals from Norway are distributed in a manner similar to that of
earlier Iron Age individuals, whereas many Viking Age individuals from
Sweden and Denmark show a greater affinity to Neolithic farmers from
Anatolia (Extended Data Fig.4a). Using the qpAdm program, we find
that the majority of groups can be modelled as three-way mixtures of
hunter-gatherer, farmer and steppe-related ancestry. The three-way
model was rejected for some groups from Sweden, Norway and the
Baltic region, which could be fit using four-way models that addition-
ally included either Caucasus hunter-gatherer or East-Asian-related
ancestry (Extended Data Figs.4b, c)—the latter of which is consistent
with previously documented gene flow from Siberia5–7.
Investigating genetic continuity with Iron Age groups that are tem-
porally more proximate to the Viking Age Scandinavian populations,
we find that most Viking Age groups can be fit using a single Iron Age
source and broadly fall into two categories: (i) English Iron Age sources
(most of the Viking Age individuals from Denmark, as well as popula-
tions of the British Isles) and (ii) Scandinavian Iron Age sources (from
Norway, Sweden and the Baltic region) (Extended Data Fig.5a). Notable
exceptions are individuals from Kärda in southern Sweden, for whom
only early Medieval Longobard individuals from Hungary can be fit as
a single source group (P>0.01) (Extended Data Fig.5a). Groups with
poor one-way fits can be modelled by including either additional north
-
eastern ancestry (for example, Viking Age individuals from Ladoga) or
additional southeastern ancestry (for example, Viking Age individuals
from Jutland) (Extended Data Fig.5b). Overall, our analyses suggest that
the genetic makeup of Viking Age Scandinavian populations largely
derives from ancestry of the preceding Iron Age populations—but these
analyses also reveal subtle differences in ancestry and gene flow from
both the south and east. These observations are largely consistent with
archaeological findings8,9.
Viking Age genetic structure in Scandinavia
To elucidate the fine-scale population structure of Viking Age Scan-
dinavia, we performed genotype imputation on a subset of 298 indi-
viduals with sufficient (>0.5×) coverage (289 from this study, along
with 9 previously published individuals
10
) and inferred the genomic
segments they shared via identity-by-descent with a reference panel of
present-day European individuals (n=1,464) (Supplementary Notes6,
10, 11). Genetic clustering using multidimensional scaling and uniform
manifold approximation and projection (UMAP) shows that Viking Age
Scandinavian individuals cluster into three groups by geographical
origin, with close affinities to their respective present-day counterparts
(Fig.3a, Supplementary Fig.10.1). Some individuals—particularly those
from the island of Gotland in eastern Sweden—have strong affinities
with Eastern Europeans; this probably reflects individuals with Baltic
ancestry, as clustering with Bronze Age individuals from the Baltic
region is evident in the identity-by-state UMAP analysis (Fig.2b) and
through f4-statistics (Supplementary Fig.9.1).
We used ChromoPainter11 and a reference panel enriched with
Scandinavian individuals (n=1,464) (Supplementary Notes6, 11)
to identify long, shared haplotypes and detect subtle population
structure (Supplementary Figs.11.1–11.10). We find ancestry com-
ponents in Scandinavia with (inexact and indicative) affinities with
present-day populations (Supplementary Fig.11.11), which we refer to
as ‘Danish-like’, ‘Swedish-like’, ‘Norwegian-like’ and ‘North Atlantic-like’
(that is, possible individuals from the British Isles entering Scandinavia).
The sampling is heavily structured, so these complex results (Sup-
plementary Fig.11.12) are visualized over time and space (Fig.4) using
spatial interpolation12 to account for sampling locations and report
significant trends (Supplementary Table11.2) using linear regression
(Supplementary Notes11, 12).
Norwegian-like and Swedish-like components cluster in Norway
and Sweden, respectively, whereas Danish-like and North-Atlantic-like
components are widespread (Fig.4, Supplementary Fig. 11.12,
a
60°
N
40°
N
20° E 40° E20° W40° W
Atlantic Ocean
b
Ireland
Faroes
Iceland
Greenland
Norway
Denmark
Sweden
Poland
Russia
Ukraine
Italy
Bronze
Age
Iron
Age
Early
Viking Age
Viking
Age
Medieval and
Early Modern
Denmark LNBA
Norway LNBA
Denmark IA
Sweden IA
Norway IA
Zealand EVA
Öland EVA
Salme EVA
Dorset VA
Oxford VA
IoM VA
Wales VA
Ireland VA
Orkney VA/IA
Faroe VA
Iceland VA
Iceland2 VA
Early Norse E VA
Early Norse W VA
Late Norse E VA
Late Norse W VA
Norway S VA
Norway M VA
Norway N VA
Jutland VA
Funen VA
Langeland VA
Zealand VA
Malm
ö
VA
K
ä
rda VA
Skara VA
Öland VA
Gotland VA
Sigtuna VA
Uppsala VA
Bodzia VA
Cedynia VA
Sandomierz VA
Ladoga VA
Gnezdovo VA
Kurevanikha VA
Pskov VA
Ukraine VA
Foggia MED
Faroe EM
Iceland MED
Norway M MED
Poland MED
Ukraine MED
UK and
Isle of Man
Fig. 1 | Overv iew of the Viki ng Age genomi c dataset . a, Map of the Viking
World from eighth to eleventh centuries , showing geograph ical locatio n
and broad age c ategory of site s with ancien t samples newly re ported in thi s
study. Age catego ries of sites (circl es) are coloured-co ded as: dark gree n, LNBA
(2400–500); light green, Iron Age (50 0–700); yellow, Early Viking Age
(700–800); Viking Age (800–1100); Med ieval and Early Mod ern
(1100–1600). Red region, are a of Viking ori gins; green re gion, area of Vik ing
raids, set tlement and tr ade; dark blue regi on, area of pionee r Viking
colonization. b, All of the anc ient individua ls from this study (n=442), and
previously p ublished Vik ing Age samples f rom Sigtuna10 and Iceland18,
categori zed on the basis of t heir spatiote mporal origin . The ancient s amples
are divided i nto the following f ive broad cate gories: Bronze A ge (BA), Iron Age
(IA), Early Vik ing Age (EVA), Viking Age (VA), Medieva l (MED) and Early Moder n
(EM). Random jit ter has been a dded along the xaxi s in each categor y to aid
visualiz ation. LNBA, L ate Neolithic a nd Bronze Age; Nors e W, Norse western
settle ment; Norse E, N orse easter n settleme nt; Norway S, sou thern Norway ;
Norway N, northern Norway; Norway M, middle Norway.

392 | Nature | Vol 585 | 17 September 2020
Article
Supplementary Table6). Unexpectedly, Viking Age individuals from
Jutland (Denmark) lack Swedish-like and Norwegian-like genetic
components (Supplementary Fig.11.12). We also find that gene flow
within Scandinavia was broadly from south to north, dominated by
movement from Denmark into Norway and Sweden (Supplementary
Table11.2).
We identified two ancient individuals from northern Norway (des
-
ignated VK518 and VK519) with affinities to present-day Saami popula-
tions in Norway and Sweden. The VK519 individual probably also had
Norwegian-like ancestors, which indicates genetic contacts between
Saami groups and other Scandinavian populations.
The genetic data are structured by topographical boundaries rather
than by the borders of present-day countries. Thus, the southwestern
part of Sweden in the Viking Age is genetically more similar to Viking
Age populations of Denmark than to those of central mainland Sweden,
probably owing to geographical barriers that prevented gene flow.
We quantified genetic diversity using two measures: conditional
nucleotide diversity (Supplementary Note9) and variation in inferred
ancestry on the basis of ChromoPainter results (Extended Data Fig.6,
Supplementary Note11, Supplementary Fig.11.13). We also visualized
this diversity as the spread of individuals on a multidimensional scal-
ing plot based on a pairwise identity-by-state sharing matrix (Fig.3b).
Diversity varies markedly from the more-homogeneous inland and
northern parts of Scandinavia to the diverse Kattegat (eastern Denmark
and western Sweden) and Baltic Sea regions, which suggests an important
role for these maritime regions in interaction and trade during the Viking
Age. On Gotland, there are many more Danish-like and North-Atlantic-like
genetic components (as well as an additional ‘Finnish-like’ ancestry com-
ponent) than Swedish-like components, which indicates extensive mari-
time contacts for Gotland during the Viking Age.
Our results for Gotland and Öland agree with archaeological indi-
cations that these were important maritime communities from the
Roman period (1–400) onwards13,14. A similar pattern is observed
on the central Danish islands (such as Langeland) but at a lower level.
The data indicate that genetic diversity on the islands increased from
the early (about eighth century ) to the late Viking Age (about tenth
to eleventh centuries ), which suggests increasing interregional
interaction over time. Evidence for genetic structure within Viking
Age Scandinavia
2,4,15–17
—with diversity in cosmopolitan centres such
as Skara, and trade-oriented islands such as Gotland—highlight the
importance of sea routes during this period.
Viking migrations
Our fine-scale ancestry analyses of genomic data are consistent with
patterns documented by historians and archaeologists (Figs.3, 4,
Supplementary Fig.11.12): eastward movements mainly involved
Swedish-like ancestry, whereas individuals with Norwegian-like
ancestry travelled to Iceland, Greenland, Ireland and the Isle of Man.
The first settlement in Iceland and Greenland also included individ-
uals with North-Atlantic-like ancestry18,19. A Danish-like ancestry is
seen in present-day England, in accordance with historical records
20
,
place names
21
, surnames
22
and modern genetics
23,24
, but Viking Age
Danish-like ancestry in the British Isles cannot be distinguished from
that of the Angles and Saxons, who migrated in the fifth to sixth cen-
turies  from Jutland and northern Germany.
Viking Age execution sites in Dorset and Oxford (England) contain
North-Atlantic-like ancestry, as well as Danish-like and Norwegian-like
ancestries. If these sites represent Viking raiding parties that were
defeated and captured
25,26
, then these raids were composed of indi-
viduals of different origins. This pattern is also suggested by isotopic
data from a warrior cemetery in Trelleborg (Denmark)27. Similarly, the
presence of Danish-like ancestry in an ancient sample from Gnezdovo in
present-day Russia indicates that eastern migrations were not entirely
composed of Viking individuals from Sweden.
Our results show that ‘Viking’ identity was not limited to individu-
als of Scandinavian genetic ancestry. Two individuals from Orkney
who were buried in Scandinavian fashion are genetically similar to
present-day Irish and Scottish populations, and are probably the first
Pictish genomes published (see ‘Evidence for Pictish genomes’ in Sup-
plementary Note11, Supplementary Figs.11.3, 11.12, 11.14, Supplemen-
tary Table6). Two other individuals from Orkney had 50% Scandinavian
MDS 1
MDS 2
UMAP 1
UMAP 2
VK518
VK519
VK531
VK518
VK531
Farmers
Hunter–gatherers
Steppe
East
Asia
Europe BA
and later
Western
Scandinavian
Pleistocene HG
Anatolia
British
Isles
Hunter–gatherers
Farmers
Steppe HG
Steppe
pastoralist
East
Asia
Anatolia
BA
Baltic
BA
Europe
early BA
Europe
early
Europe
late
Beaker
complex
Sarmatian
Saami
Baikal
HG
Denmark LNBA
Norway LNBA
Denmark IA
Sweden IA
Norway IA
Zealand EVA
Öland EVA
Salme EVA
Dorset VA
Oxford VA
IoM VA
Wales VA
Ireland VA
Orkney VA/IA
Faroe VA
Iceland VA
Iceland 2 VA
Early Norse E VA
Early Norse W VA
Late Norse E VA
Late Norse W VA
Norway S VA
Norway M VA
Norway N VA
Jutland VA
Funen VA
Langeland VA
Zealand VA
Malmö VA
Kärda VA
Skara VA
Öland VA
Gotland VA
Sigtuna VA
Uppsala VA
Bodzia VA
Cedynia VA
Sandomierz VA
Ladoga VA
Gnezdovo VA
Kurevanikha VA
Pskov VA
Ukraine VA
Foggia MED
Faroe EM
Iceland MED
Norway M MED
Poland MED
Ukraine MED
ab
Fig. 2 | Gene tic struc ture of Vikin g Age sample s. a, Multidimensional scaling
(MDS) of n=1,305 anc ient genomes, o n the basis of a pair wise identi ty-by-state
sharing mat rix of the Viki ng Age and other anc ient samples (Su pplementar y
Table3). Outlier individual s with hunter-gathe rer (VK531) or S aami-related
ancestr y (VK51 8 and VK519) are highli ghted. b, UMA P analysis of the sam e
datase t as in a, with fin e-scale ance stry group s highlighted . HG,
hunter-gatherer.

Nature | Vol 585 | 17 September 2020 | 393
ancestry, and five such individuals were found in Scandinavia. This
suggests that Pictish populations may have been integrated into Scan-
dinavian culture by the Viking Age.
Viking Age gene flow into Scandinavia
Non-Scandinavian ancestry in samples from Denmark, Norway and
Sweden agrees with known trading routes (Supplementary Notes11, 12):
for example, Finnish and Baltic ancestry reached modern Sweden
(including Gotland), but is absent in most individuals from Denmark
and Norway. By contrast, western regions of Scandinavia received
ancestry from the British Isles (Supplementary Notes11, 12). The first
evidence of South European ancestry (>50%) in Scandinavia is during
the Viking Age in Denmark (for example, individuals VK365 and VK286
from Bogøvej) and southern Sweden (for example, VK442 and VK350
from Öland, and VK265 from Kärda) (Fig.4, Supplementary Table6).
Disappearance from Greenland
From around  980 to 1440, southwest Greenland was settled by peo-
ple of Scandinavian ancestry (probably from Iceland)28,29. The fate of
these populations in Greenland remains debated, but probable causes
of their disappearance are social or economic processes in Europe
(for example, political relations within Scandinavia and changed trad-
ing systems) and natural processes, including climatic change29–31.
According to our data, the Greenland Norse populations were an
admixture between Scandinavians (mostly from Norway) and individu-
als from the British Isles, similar to the first settlers of Iceland
18
. We see
no evidence of long-term inbreeding in the genomes of Greenlandic
Norse individuals, although we have only one high-coverage genome
from the later period of occupation of the island (Supplementary
Note10, Supplementary Figs.10.2, 10.3). This result could favour
a relatively brief depopulation scenario, consistent with previous
demographic models
32
and archaeological findings. We also find no
evidence of ancestry from other populations (Palaeo-Eskimo, Inuit or
Native American) in the Greenlandic Norse genomes (Supplementary
Fig.9.4), which accords with the skeletal remains
32
. This suggests that
sexual interaction between the Greenland Norse populations and
these other groups was absent, or occurred only on a very small scale.
Genetic composition of earliest Viking voyage
Although maritime raiding has been a constant of seafaring cultures for
millennia, the Viking Age is partly defined by this activity33. However, the
exact nature and composition of Viking war parties is unknown5. One
raiding or diplomatic expedition has left direct archaeological traces:
at Salme in Estonia, 41men from Sweden who died violently were buried
in two boats, accompanied by high-status weaponry
34,35
. Importantly,
the Salme boat burial predates the first textually documented raid (on
Lindisfarne (England) in 793) by nearly half a century.
Kinship analysis of the genomes of 34 individuals from the Salme
burial reveals 4brothers buried side by side and a third-degree relative
of 1 of the 4brothers (Supplementary Note4). The ancestry profiles of
the Salme individuals were similar to one another when compared to the
profiles of other burials of the Viking Age (Supplementary Notes10, 11),
which suggests a relatively genetically homogeneous group of people
of high status (including close kin).
The five Salme relatives are not the only kin in our dataset; we also
identified two pairs of kin in which the related individuals were exca-
vated hundreds of kilometres apart from each other, which markedly
illustrates the mobility of individuals during the Viking Age.
Positive selection in northern Europe
We looked for single-nucleotide polymorphisms (SNPs) with allele frequen-
cies that have changed significantly in the last 10,000years
36,37
, beyond
what can be explained by temporal changes in ancestry alone (Supplemen-
tary Note14). Extended Data Figure8a shows the likelihood ratio scores in
favour of selection in the entire 10,000-year period (the general scan), the
period up to 4,000years before present (the ancient scan) and the period
from 4,000years before present up to today (the recent scan).
As expected38,39, the strongest candidates for selection are SNPs
near LCT, the frequency of which increased after the Bronze Age
40,41
.
Our dataset traces the frequency of the lactase-persistence allele
(rs4988235) and its evolution since the Bronze Age. Extended Data
Figure8b shows that Viking Age groups had very similar allele frequen-
cies at the LCT lactase-persistence SNP to those of present-day northern
European populations. Conversely, Bronze Age Scandinavian individu-
als, as well as individuals from central Europe associated with Corded
Ware and Bell Beaker assemblages, have a low frequency of this SNP
despite evidence for milk consumption. Our Iron Age samples have
intermediate frequencies, which suggests a rise in lactase persistence
during this period. The frequency is higher in the Bronze Age of the
Baltic Sea region than in Bronze Age Scandinavia, which is consist-
ent with gene flow between the two regions explaining the increasing
frequency of lactase persistence in Scandinavia.
Other candidates for selection include previously identified regions,
including the one containing the TLR1, TLR6 and TLR10 genes, the HLA
–5
0
5
–10–505
UMAP 1
UMAP 2
Finland
Denmark
Norway
Sweden
UK
Poland
Italy
a
Jutland VA
Funen VA
Langeland VA
Zealand VA
Malmö VA
Kärda VA
Skara VA
Öland VA
Gotland VA
Sigtuna VA
Uppsala VA
Bodzia VA
Cedynia VA
Sandomierz VA
Ladoga VA
Gnezdovo VA
Kurevanikha VA
Pskov VA
Ukraine VA
Foggia MED
Faroe EM
Iceland MED
Norwa
y M MED
Poland MED
Ukraine MED
Denmark LNBA
Norway LNBA
Denmark IA
Sweden IA
Norway IA
Zealand EVA
Öland EVA
Salme EVA
Dorset VA
Oxford VA
IoM VA
Wales VA
Ireland VA
Orkney VA/IA
Faroe VA
Iceland VA
Iceland 2 VA
Early Norse E VA
Early Norse W VA
Late Norse E VA
Late Norse W VA
Norway S VA
Norway M VA
Norway N VA
Salme
Gotland
Öland
Skara
Jutland
Zealand
Norway
Funen and Langeland
b
Fig. 3 | Gene tic struc ture and divers ity of ancie nt samples . a, UMAP analysi s
of n=1,624ancient and mode rn Scandinav ian individual s, on the basis of the
first 10dime nsions of MDS us ing identit y-by-descent se gments of im puted
individua ls. Large symb ols indicate m edian coordinat es for each group.
b, Genetic di versity in major p opulations of th e Scandinavia n Viking Age. Plo ts
next to the map s how MDS analysis on t he basis of a pair wise identi ty-by-state
sharing mat rix. Norway d enotes all the si tes from Norw ay. The scale is ident ical
for all the plots .

394 | Nature | Vol 585 | 17 September 2020
Article
region, and the genes SLC45A2 and SLC22A441. We also find additional
candidate regions for selection that have associated trajectories that
start before the Viking Age, which suggests shared phenotypes between
ancient Viking and present-day Scandinavian populations (Supplemen-
tary Note14). These regions include one that overlaps DCC and that is
implicated in colorectal cancer42, as well as one that overlaps AKNA and
is involved in the secondary immune response43.
Evolution of complex traits in Scandinavia
To search for signals of recent population differentiation at SNP markers
associated with complex traits, we compared genotypes of Viking Age
individuals with those of apanel of present-day Danish individuals
44
. We
obtained summary statistics from 16 well-powered genome-wide associa-
tion studies through the GWAS ATLAS
45
and tested for a difference in the
distribution of polygenic scores between the two groups (Supplementary
Note15). The polygenic scores of Viking Age individuals and present-day
Danish individuals differed for three traits: black hair colour (P=0.00089),
standing height (P=0.019) and schizophrenia (P=0.0096), although the
latter two were not significant after accounting for the number of tests
(Extended Data Fig.7). Currently, we cannot conclude whether the observed
differences in allele frequencies are due to selection acting on these alleles
between the Viking Age and the present time or to some other factors (such
as more ethnic diversity in the Viking Age sample). A binomial test of the
number of black hair colour risk alleles at higher frequency in the Viking Age
sample and the present-day sample was also significant (65/41; P=0.025),
which suggests that the signal is not entirely driven by a few large-effect loci.
Viking genetic legacy in populations today
To test whether present-day Scandinavian populations share increased
ancestry with their respective counterparts in the Viking Age, we first
Iron AgeEarly Viking Age Viking Age
Southern
European
Southern
European
Southern
European
Danish-
like
Norwegian-
like
0
0.25
0.50
0.75
1.00
0
0.25
0.50
0.75
1.00
0
0.25
0.50
0.75
1.00
0
0.25
0.50
0.75
1.00
0
0.25
0.50
0.75
1.00
0
0.25
0.50
0.75
1.00
0
0.25
0.50
0.75
1.00
0
0.25
0.50
0.75
1.00
0
0.25
0.50
0.75
1.00
0
0.25
0.50
0.75
1.00
Danish-
like
Norwegian-
like
North
Atlantic
Swedish-
like
North
Atlantic
Fig. 4 | Spati otemporal p atterns of V iking and no n-Viking ance stry in
Europe dur ing the Iron Ag e, Early Vikin g Age and Viki ng Age. We perfor med
inverse distance-weighting interpolation of the ancestry painting proportions
of each indiv idual genome on a d ense grid of po ints covering the E uropean
continent, to better visualize the distribution of ancestry paintings at different
periods (Su pplementar y Note12). Top, distinct sphe res of influ ence in the
Viking world . Middle, Danish V iking ances try in southe rn Britain, N orwegian
Viking anc estry in Irel and and Isle of Man an d non-Scandin avian (‘North
Atlantic’ ) ancestry i n Orkney, Ireland and sou thern Brit ain. Bottom, L ate
southern European ancestry in southern Scandinavia. The Swedish-like
ancestr y is the highe st in present-day Esto nia owing to the anc ient samples
from the Salm e ship burial, whic h originated f rom the Mälaren vall ey of Sweden
(according to arch aeological s ources). n=289genomes use d for interpolati on.

Nature | Vol 585 | 17 September 2020 | 395
computed D-statistics of the form D(Yoruba (YRI), ancient; present-day
population1, present-day population2), which measure whether an
ancient test individual shares more alleles with present-day popula-
tion1 or with present-day population2. Viking Age individuals shift
subtly from Scandinavia towards their present-day counterparts in the
distributions of these statistics (Extended Data Fig.5c, Supplementary
Figs.9.2, 9.3).
We further examined ancient ancestry in present-day populations
using fineSTRUCTURE (Supplementary Note11, Supplementary
Fig.11.14). Within Scandinavia, most present-day populations resemble
their Viking Age counterparts. The exception is Swedish-like ances-
try, which is present at only 15–30% within Sweden today: one cluster
from Sweden is closer to ancient Finnish populations, and a second is
more closely related to Danish and Norwegian populations. Danish-like
ancestry is now high across the whole region.
Outside of Scandinavia, the genetic legacy of Viking Age populations
is consistent—although limited. A small Scandinavian ancestry compo-
nent is present in Poland (up to 5%). Within the British Isles, it is difficult
to assess how much of the Danish-like ancestry is due to pre-existing
Anglo-Saxon ancestry, but the Viking Age contribution does not exceed
6% in England (Supplementary Note11). The genetic effects are stronger
in the other direction. Although some North-Atlantic-like individuals
in Orkney became culturally Scandinavian, others found themselves
in Iceland, Norway and beyond, leaving a genetic legacy that persists
today. Present-day Norwegian individuals vary between 12 and 25%
in North-Atlantic-like ancestry; this ancestry is more uniformly 10%
in Sweden.
Discussion
Our genomic analyses shed light on long-standing questions raised by
historical sources and archaeological evidence from the Viking Age. We
largely confirm the long-argued movements of Vikings outside Scandi-
navia: Vikings from present-day Denmark, Norway, and Sweden going
to Britain, the islands of the North Atlantic, and sailing east towards the
Baltic region and beyond, respectively. However, we also see ancient
Swedish-like and Finnish-like ancestry in the westernmost fringes of
Europe, and Danish-like ancestry in the east, defying modern histori-
cal groupings. It is likely that many such individuals were from com-
munities with mixed ancestries, thrown together by complex trading,
raiding and settling networks that crossed cultures and the continent.
During the Viking Age, different parts of Scandinavia were not evenly
connected, leading to clear genetic structure in the region. Scandinavia
probably comprised a limited number of transport zones and maritime
enclaves
46
with active external contacts, and limited external gene flow
into the rest of the Scandinavian landmass. Some Viking Age Scandina-
vian locations are relatively homogeneous—particularly mid-Norway,
Jutland and the Atlantic settlements. This contrasts with the strong
genetic variation of populous coastal and southern trading communi-
ties such as in the islands of Gotland and Öland47–49. The high genetic
heterogeneity in coastal communities implies increased population
size, extending a previously proposed10 urbanization model for the Late
Viking Age city of Sigtuna (which suggested that more-cosmopolitan
trading centres were already present at the end of the Viking Age in
Northern Europe) both spatially and further back in time. The formation
of large-scale trading and cultural networks that spread people, goods
and warfare took time to affect the heartlands of Scandinavia, which
retained pre-existing genetic differences into the Medieval period.
Finally, our findings show that Vikings were not simply a direct contin-
uation of Scandinavian Iron Age groups. Instead, we observe gene flow
from the south and east into Scandinavia, starting in the Iron Age and
continuing throughout the duration of the Viking Age, from an increas
-
ing number of sources. Many Viking Age individuals—both within and
outside Scandinavia—have high levels of non-Scandinavian ancestry,
which suggests ongoing gene flow across Europe.
Online content
Any methods, additional references, Nature Research reporting sum-
maries, source data, extended data, supplementary information,
acknowledgements, peer review information; details of author con-
tributions and competing interests; and statements of data and code
availability are available at https://doi.org/10.1038/s41586-020-2688-8.
1. Brink, S. & Price, N. (eds) The Viking World (Routledge, 2008).
2. Jesch, J. The Viking Diaspora (Routledge, 2015).
3. Eriksen, M. H., Pedersen, U., Rundberget, B. & Axelsen, I. Viking Worlds: Things, Spaces
and Movement (Oxbow Books, 2014).
4. Sindbæk, S. M. & Trakadas, A. The World in the Viking Age (Viking Ship Museum in
Roskilde, 2014).
5. Sikora, M. etal. The population history of northeastern Siberia since the Pleistocene.
Nature 570, 182–188 (2019).
6. Lamnidis, T. C. etal. Ancient Fennoscandian genomes reveal origin and spread of
Siberian ancestry in Europe. Nat. Commun. 9, 5018 (2018).
7. Saag, L. etal. The arrival of Siberian ancestry connecting the eastern Baltic to Uralic
speakers further east. Curr. Biol. 29, 1701–1711.e16 (2019).
8. Hedeager, L. in The Viking World (eds Brink, S. & Price, N.) 35–46 (Routledge, 2008).
9. Hedeager, L. Iron Age Myth and Materiality: An Archaeology of Scandinavia AD 400–1000
(Routledge, 2011).
10. Krzewińska, M. etal. Genomic and strontium isotope variation reveal immigration
patterns in a Viking Age town. Curr. Biol. 28, 2730–2738.e10 (2018).
11. Lawson, D. J., Hellenthal, G., Myers, S. & Falush, D. Inference of population structure using
dense haplotype data. PLoS Genet. 8, e1002453 (2012).
12. Shepard, D. A Two-dimensional interpolation function for irregularly-spaced data. In
Proceedings of the 1968 23rd ACM National Conference (eds Blue, R. B. & Rosenberg, A. M.)
517–524 (ACM, 1968).
13. Hansen, U. L. Römischer Import im Norden: Warenaustausch zwischen dem Römischen
Reich und dem freien Germanien während der Kaiserzeit unter besonderer
Berücksichtigung Nordeuropas (Det Kongelige Nordiske Oldskriftselskab, 1987).
14. Andersson, K. I Skuggan av Rom: Romersk Kulturpåverkan i Norden (Atlantis, (2013).
15. Bill, J. in The Viking World (eds Brink, S. & Price, N.) 170–180 (Routledge, 2008).
16. Sindbæk, S. M. The small world of the Vikings: networks in early Medieval communication
and exchange. Norw. Archaeol. Rev. 40, 59–74 (2007).
17. Hilberg, V. & Kalmring, S. in Viking Archaeology in Iceland: Mosfell Archaeological Project
(eds Zori, D. & Byock, J.) 221–245 (Brepols, 2014).
18. Ebenesersdóttir, S. S. etal. Ancient genomes from Iceland reveal the making of a human
population. Science 360, 1028–1032 (2018).
19. Helgason, A. etal. mtDna and the islands of the North Atlantic: estimating the proportions
of Norse and Gaelic ancestry. Am. J. Hum. Genet. 68, 723–737 (2001).
20. Downham, C. Viking ethnicities: a historiographic overview. Hist. Compass 10, 1–12
(2012).
21. Fellows-Jensen, G. in The Viking World (eds Brink, S. & Price, N.) 391–400 (Routledge,
2008).
22. Bowden, G. R. etal. Excavating past population structures by surname-based sampling:
the genetic legacy of the Vikings in northwest England. Mol. Biol. Evol. 25, 301–309
(2008).
23. Leslie, S. etal. The ine-scale genetic structure of the British population. Nature 519,
309–314 (2015).
24. Athanasiadis, G. etal. Nationwide genomic study in Denmark reveals remarkable
population homogeneity. Genetics 204, 711–722 (2016).
25. Loe, L., Boyle, A., Webb, H. & Score, D. ‘Given to the Ground’: A Viking Age Mass Grave on
Ridgeway Hill, Weymouth (Dorset Natural History and Archaeological Society, 2014).
26. Wallis, S. The Oxford Henge and Late Saxon Massacre: With Medieval and Later
Occupation at St John’s College, Oxford (Thames Valley Archaeological Services Limited,
2014).
27. Douglas Price, T., Frei, K. M., Dobat, A. S., Lynnerup, N. & Bennike, P. Who was in Harold
Bluetooth’s army? Strontium isotope investigation of the cemetery at the Viking Age
fortress at Trelleborg, Denmark. Antiquity 85, 476–489 (2011).
28. Price, T. D. & Arneborg, J. The peopling of the North Atlantic: isotopic results from
Greenland. Journal of the North Atlantic 7, 164–185 (2014).
29. Arneborg, J. in The Viking World (eds Brink, S. & Price, N.) 588–603 (Routledge, 2008).
30. Dugmore, A. J. etal. Cultural adaptation, compounding vulnerabilities and conjunctures
in Norse Greenland. Proc. Natl Acad. Sci. USA 109, 3658–3663 (2012).
31. Arneborg, J. in Medieval Archaeology in Scandinavia and Beyond: History, Trends and
Tomorrow (eds Kristiansen, M. S. etal.) 247–271 (Aarhus Universitetsforlag, 2015).
32. Lynnerup, N. The Greenland Norse: A Biological–Anthropological Study (Museum
Tusculanum, 1998).
33. Sindbæk, S. M. in Routledge Handbook of Archaeology and Globalization (ed. Hodos, T.)
553–565 (Routledge, 2016).
34. Peets, J. etal. Research results of the Salme ship burials in 2011–2012. Archaeological
Fieldwork in Estonia 2012, 43–60 (2012).
35. Douglas Price, T., Peets, J., Allmäe, R., Maldre, L. & Oras, E. Isotopic provenancing of the
Salme ship burials in pre-Viking Age Estonia. Antiquity 90, 1022–1037 (2016).
36. Cheng, J. Y., Racimo, F. & Nielsen, R. Ohana: detecting selection in multiple populations by
modelling ancestral admixture components. Preprint at https://doi.org/10.1101/546408
(2019).
37. Alves, J. M. etal. Parallel adaptation of rabbit populations to myxoma virus. Science 363,
1319–1326 (2019).
38. Enattah, N. S. etal. Identiication of a variant associated with adult-type hypolactasia.
Nat. Genet. 30, 233–237 (2002).

396 | Nature | Vol 585 | 17 September 2020
Article
39. Bersaglieri, T. etal. Genetic signatures of strong recent positive selection at the lactase
gene. Am. J. Hum. Genet. 74, 1111–1120 (2004).
40. Allentoft, M. E. etal. Population genomics of Bronze Age Eurasia. Nature 522, 167–172
(2015).
41. Mathieson, I. etal. Genome-wide patterns of selection in 230 ancient Eurasians. Nature
528, 499–503 (2015).
42. Fearon, E. R. etal. Identiication of a chromosome 18q gene that is altered in colorectal
cancers. Science 247, 49–56 (1990).
43. Siddiqa, A. etal. Regulation of CD40 and CD40 ligand by the AT-hook transcription factor
AKNA. Nature 410, 383–387 (2001).
44. Pedersen, C. B. etal. The iPSYCH2012 case–cohort sample: new directions for unravelling
genetic and environmental architectures of severe mental disorders. Mol. Psychiatry 23,
6–14 (2018).
45. Watanabe, K. etal. A global view of pleiotropy and genetic architecture in complex traits.
Nat. Genet. 51, 1339–1348 (2019).
46. Westerdahl, C. The maritime cultural landscape. Int. J. Naut. Archaeol. 21, 5–14 (1992).
47. Hyenstrand, Å. Ancient Monuments and Prehistoric Society (Central Board of National
Antiquities, 1979).
48. Callmer, J. in Regions and Relections: In honour of Märta Strömberg (eds Jennbert, K.
etal.) 257–273 (1991).
49. Jakobsen, J. G. G. & Dam, P. Atlas Over Danmark: Historisk-Geograisk Atlas (Det Kongelige
Danske Geograiske Selskab, 2008).
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional afiliations.
© The Author(s), under exclusive licence to Springer Nature Limited 2020
1Lundbeck Foundation GeoGenetics Centre, GLOBE Institute, University of Copenhagen,
Copenhagen, Denmark. 2Institute of Molecular Biology, National Academy of Sciences,
Yerevan, Armenia. 3Section for Evolutionary Genomics, GLOBE Institute, University of
Copenhagen, Copenhagen, Denmark. 4MRC Integrative Epidemiology Unit, University of
Bristol, Bristol, UK. 5School of Statistical Sciences, University of Bristol, Bristol, UK. 6Novo
Nordisk Foundation Center for Protein Research, Faculty of Health and Medical Sciences,
University of Copenhagen, Copenhagen, Denmark. 7The Bioinformatics Centre, Department
of Biology, University of Copenhagen, Copenhagen, Denmark. 8Smurit Institute of
Genetics, Trinity College Dublin, Dublin, Ireland. 9Novo Nordisk Foundation Center for Basic
Metabolic Research, Faculty of Health and Medical Sciences, University of Copenhagen,
Copenhagen, Denmark. 10Department of Clinical Medicine, University of Copenhagen,
Copenhagen, Denmark. 11Institute of Biological Psychiatry, Mental Health Services
Copenhagen, Copenhagen, Denmark. 12HSE University, Russian Federation National
Research University Higher School of Economics, Moscow, Russia. 13Department of
Archaeology and Ancient History, Lund University, Lund, Sweden. 14Sydsvensk Arkeologi
AB, Kristianstad, Sweden. 15Department of Immunology, Genetics and Pathology, Uppsala
University, Uppsala, Sweden. 16Department of Genetics, University of Cambridge,
Cambridge, UK. 17Department of Health Technology, Section for Bioinformatics, Technical
University of Denmark (DTU), Copenhagen, Denmark. 18Department of Human Genetics,
McGill University, Montréal, Quebec, Canada. 19National Institute of Genomic Medicine
(INMEGEN), Mexico City, Mexico. 20Archaeological Research Collection, Tallinn University,
Tallinn, Estonia. 21Museum and Institute of Zoology, Polish Academy of Sciences, Warsaw,
Poland. 22Anuchin Research Institute and Museum of Anthropology, Moscow State
University, Moscow, Russia. 23Manx National Heritage, Douglas, Isle of Man.
24Upplandsmuseet, Uppsala, Sweden. 25NTNU University Museum, Department of
Archaeology and Cultural History, Trondheim, Norway. 26The Archaeologists, National
Historical Museums, Stockholm, Sweden. 27Thames Valley Archaeological Services (TVAS),
Reading, UK. 28Department of Archaeology and Ancient History, Uppsala University,
Uppsala, Sweden. 29Institute of Archaeology, University of Warsaw, Warsaw, Poland.
30Department of Cultural Heritage Conservation, National Yunlin University of Science and
Technology, Douliou, Taiwan. 31Institute of Archaeology, Reykjavík, Iceland. 32UHI
Archaeology Institute, University of the Highlands and Islands, Kirkwall, UK. 33Department
of Bioarchaeology, Institute of Archaeology of National Academy of Sciences of Ukraine,
Kiev, Ukraine. 34Soprintendenza Archeologia, Belle Arti e Paesaggio per le Province di
Barletta, Andria, Trani e Foggia, Foggia, Italy. 35Jönköping County Museum, Jönköping,
Sweden. 36Trinity College Dublin, Dublin, Ireland. 37Heritage Burial Services, Oxford
Archaeology, Oxford, UK. 38National Museum of Ireland, Dublin, Ireland. 39Institute of
Archaeology, Maria Curie-Sklodowska University in Lublin, Lublin, Poland. 40Västergötlands
Museum, Skara, Sweden. 41Department of History and Archaeology, Amgueddfa Cymru–
National Museum Wales, Cardiff, UK. 42Trzy Epoki Archaeological Service, Klimontów,
Poland. 43Historical Faculty, Moscow State University, Moscow, Russia. 44Museum of
Southwest Jutland, Ribe, Denmark. 45Department of Slavic–Finnish Archaeology, Institute
for the History of Material Culture, Russian Academy of Sciences, Saint Petersburg, Russia.
46National Museum of Denmark, Copenhagen, Denmark. 47Department of Research and
Heritage, Roskilde Museum, Roskilde, Denmark. 48Langelands Museum, Langeland,
Denmark. 49Department of Humanities, University of Foggia, Foggia, Italy. 50Department of
Molecular Medicine, Faculty of Medicine, University of Oslo, Oslo, Norway. 51Department of
Archaeology and Ancient History, Uppsala University Campus Gotland, Visby, Sweden.
52Tjóðsavnið – Faroe Islands National Museum, Tórshavn, Faroe Islands. 53Peter the Great
Museum of Anthropology and Ethnography (Kunstkamera), Russian Academy of Science,
St Petersburg, Russia. 54Malmö Museum, Malmö, Sweden. 55Laboratoire d’Anthropobiologie
Moléculaire et d’Imagerie de Synthèse, CNRS UMR 5288, Université de Toulouse, Université
Paul Sabatier, Toulouse, France. 56Department of Archaeology, Simon Fraser University,
Burnaby, British Colombia, Canada. 57Department of Forensic Medicine, University of
Copenhagen, Copenhagen, Denmark. 58School of GeoSciences, University of Edinburgh,
Edinburgh, UK. 59Department of Natural History, Norwegian University of Science and
Technology (NTNU), Trondheim, Norway. 60Trace and Environmental DNA (TrEnD)
Laboratory, School of Molecular and Life Sciences, Curtin University, Perth, Western
Australia, Australia. 61Museum of Cultural History, University of Oslo, Oslo, Norway. 62Centre
for Urban Network Evolutions (UrbNet), School of Culture and Society, Aarhus University,
Højbjerg, Denmark. 63Institute of Archaeology, Conservation and History, Oslo, Norway.
64Department of Historical Studies, University of Gothenburg, Gothenburg, Sweden.
65Department of Integrative Biology, UC Berkeley, Berkeley, CA, USA. 66Department of
Statistics, UC Berkeley, Berkeley, CA, USA. 67The Lundbeck Foundation Initiative for
Integrative Psychiatric Research, iPSYCH, Copenhagen, Denmark. 68Department of Zoology,
University of Cambridge, Cambridge, UK. 69The Danish Institute for Advanced Study,
University of Southern Denmark, Odense, Denmark. 70The Wellcome Trust Sanger Institute,
Cambridge, UK. 71These authors contributed equally: Ashot Margaryan, Daniel J. Lawson,
Martin Sikora, Fernando Racimo. ✉e-mail: rasmus_nielsen@berkeley.edu; Thomas.Werge@
regionh.dk; ew482@cam.ac.uk

Methods
No statistical methods were used to predetermine sample size. The
experiments were not randomized and investigators were not blinded
to allocation during experiments and outcome assessment.
Laboratory work
Laboratory work was conducted in the dedicated ancient DNA
clean-room facilities at the Globe Institute (University of Copenhagen),
according to strict ancient DNA standards50,51. The overwhelming major-
ity of ancient samples were petrous bones and teeth (Supplementary
Table1). The details of DNA extraction can be found in Supplementary
Note2. Double-stranded blunt-end DNA libraries were prepared using
Illumina-specific adapters and NEBNext DNA Sample Pre Master Mix
Set 2 (E6070) kit. We used an Agilent Bioanalyzer 2100 to quantify the
amount of the purified DNA libraries. The libraries were sequenced
80-bp single-read chemistry on Illumina HiSeq 2500 machines at the
Danish National High-throughput DNA Sequencing Centre.
Bioinformatics analysis and quality assessment
We used AdapterRemoval v.2.1.352 for removing Illumina adaptor
sequences, keeping only sequences with a minimum length of 30 bp.
Adaptor-free sequences were mapped against the human reference
genome build 37 using BWA v.0.7.10 aligner53 with the seed (-l parameter)
disabled for higher sensitivity of ancient DNA reads
54
. DNA sequences
were processed with samtools v.1.3.153, and only sequences with mapping
quality≥30 were kept. Picard v.1.127 (http://broadinstitute.github.io/
picard) was used to sort the reads and remove duplicates. DNA libraries
were combined at the sample level and realigned using GATK v.3.3.055
with Mills and 1000G gold-standard insertions and deletions (indels). At
the end, realigned .bam files had the md-tag updated and extended base
alignment qualities calculated using samtools calmd. Read depth and
coverage were determined using pysam (http://code.google.com/p/
pysam/) and BEDtools
56
. The mapping statistics for the ancient samples
are summarized in Supplementary Table2.
We used mapDamage v.2.0 to obtain approximate Bayesian estimates
of damage parameters57. Data authenticity was assessed by estimating
the rate of mismatches to the consensus mitochondrial sequence using
contamMix
58
and Schmutzi
59
, as well as the excess of heterozygous
positions in male haploid X chromosomes using ANGSD60. The sex of
ancient individuals was determined by calculating the Rγ parameter
61
.
Uniparental haplogroup determination and kinship analysis
The mitochondrial haplogroups of the ancient individuals were
assigned using haplogrep62. To get the mtDNA consensus sequences,
we aligned the trimmed reads of ancient samples to the human mito-
chondrial reference genome: revised Cambridge Reference Genome
(rCRS). Base quality≥20 and mapping quality≥30 filtering options
were applied. Only SNPs at sites≥3× coverage were considered for
consensus calling using samtools mpileup/bcftools v.1.3.153.
We identified male Y chromosome lineages using the pathPhynder
workflow (https://github.com/ruidlpm/pathPhynder) and Yleaf v.263.
For the latter, the analysis was restricted to 26,083 biallelic SNPs from
the International Society of Genetic Genealogy (ISOGG) 2019 database
(https://isogg.org/tree/ISOGG_YDNA_SNP_Index.html).
We used NgsRelate
64
to detect family relationships between all pairs
of individuals. NgsRelate is a maximum-likelihood based program that—
for a pair of individuals based on genotype likelihoods—estimates the
three coefficients, k0, k1 and k2, which denote the proportions of the
genome in which the pair of analysed individuals share 0, 1 and 2 alleles
identical-by-descent, respectively. We only included the 376 samples
with sequencing depth above 0.1× for the analysis. From these, we esti-
mated genotype likelihoods and allele frequencies with ANGSD60 using
the SAMtools genotype likelihood model (-gl 1) including reads with
mapping quality≥30 and bases with base quality≥20. We estimated
genotype likelihoods and allele frequencies only for the autosomal
transversion sites for which the 1000 Genomes CEU population (Utah
residents with northern and western European ancestry) has a minor
allele frequency of 0.05, resulting in 1,752,719 sites. READ
65
was used
to confirm the degree of relatedness between pairs of individuals.
The pedigree reconstructions on the basis of the kinship coefficients
were conducted using Pedigree Reconstruction and Identification of
a Maximum Unrelated Set (PRIMUS)66.
Imputation
We imputed the genotypes of 298 ancient samples (289 from this
study, and 9 from a previous study
10
) that had a sequencing depth
greater than 0.5×. We used Beagle v.4.1
67
for imputations based on
the genotype likelihood data, which was first estimated by GATK
v.3.7.0 UnifiedGenotyper. To generate the genotype data, we called
only biallelic sites present in the 1000 Genomes dataset, and only the
observed alleles (--genotyping_mode GENOTYPE_GIVEN_ALLELES).
The resulting .vcf files were filtered by setting genotype likelihoods
to 0 for all three genotypes (for example, hom ref, het and hom alt)
for sites with potential deamination (C>T and G>A), as described in
a previous study68. Following this, the per-individual .vcf files were
merged using bcftoolsv.1.3.1. The combined .vcf files were then split
into 15,000 markers each and imputed separately using Beagle 4.0
using the 1000 Genomes phase-3 map included with Beagle (*.phase3.
v5a.snps.vcf.gz and plink.chr*.GRCh37.map) with input through the
genotype likelihood option. Run time for imputing using Beagle was
approximately 280,000 core hours.
Merge with existing panels
Scandinavian panel. To assess the genetic relationships of various
Viking Age groups with their present-day counterparts, we constructed
a reference panel enriched with Scandinavian populations on the basis
of published datasets: the EGAD00010000632 data set from a previ-
ous publicaton
23
(UK dataset) and the EGAD00000000120 dataset
from The International Multiple Sclerosis Genetics Consortium and
The Wellcome Trust Case Control Consortium2 (ref. 69) (EU dataset)
(see Supplementary Note6 for details). The seven most relevant
populations from Denmark, Sweden, Norway, Finland, Poland, UK
and Italy were considered (n=1,464) with a total number of 414,264
SNPs. The Han Chinese (CHB) and Yoruba (YRI) populations from the
1000 Genomes project phase-3 database were merged to this panel
as outgroups.
The 1000 Genomes panel. We used a set of 1,995 individuals from 20
populations (excluding individuals from the AMR super-population,
as well as admixed ASW and ACB populations) of the 1000 Genom-
es project phase-3 release 5 (ftp.1000genomes.ebi.ac.uk/vol1/ftp/
release/20130502/). We restricted the dataset to a set of 12,731,663
biallelic transversion SNPs located within the strict mappability mask
regions (ftp.1000genomes.ebi.ac.uk/vol1/ftp/release/20130502/sup-
porting/accessible_genome_masks/).
Analyses of phenotype associated SNPs were carried out using five
European-ancestry populations: Spanish (IBS), Tuscan (TSI), CEU, Brit-
ish (GBR) and Finnish (FIN), along with CHB and YRI as outliers. These
were used to assess genome-wide allele frequencies for various SNPs
associated with pigmentation phenotypes and lactose intolerance.
Ancient panels. We constructed datasets for population genetic analy-
ses by merging the newly sequenced Viking Age individuals as well as
other previously published ancient individuals40,41,68,70–96 with the two
modern reference panels. Ancient individuals were represented with
pseudohaploid genotypes, by using mpileup command of samtools
and randomly sampling an allele passing filters (mapping quality≥30
and base quality≥30), further requiring that it matched one of the
two alleles observed in the reference panel (Supplementary Table3).

Article
Clustering analyses
On the basis of the pseudohaploid individuals from the ancient panels,
we ran ADMIXTURE97 by thinning the dataset for linkage disequilibrium
using plink with recommended settings (--indep-pairwise 50 10 0.1).
This dataset contained 1,324 individuals for 151,235 markers for the
autosomal chromosomes. Only samples with >20,000 SNPs overlap-
ping with the Human Origins panel were kept in the analysis, resulting
in 378 samples from this study. We did 50 replicates with different seeds
for k=2 to k=10. We used pong
98
to identify the best run for each k and
similar components between different k values.
The large number of ancient individuals included in the analysis
panels facilitates genetic clustering using the ancient individuals them-
selves, rather than projecting them on axes of variation inferred from
modern populations. We carried this out using MDS on a distance matrix
obtained from pairwise identity-by-state sharing between individu-
als, using the cmdscale function in R. We performed the main genetic
clustering on a set of 1,306 ancient Eurasian individuals with >50,000
SNPs with genotype data, restricting to the batch-corrected SNP set
described in Supplementary Note8. Results from the batch-corrected
MDS were combined with further dimensionality reduction using
UMAP, implemented in the uwot package in R.
Population genetics
We used f
4
statistics to investigate allele-sharing between sets of test
individuals and different modern and ancient groups (Supplementary
Note9). To characterize the deep ancestry relationship of the study
individuals we calculated f
4
(YRI, test individual; Barcin_EN.SG, Yam-
naya_EBA.SG) for all ancient Europeans from the Bronze Age onwards
(1000 Genomes panel merge). This statistic contrasts genetic affinities
of the test individuals with two major ancestry groups that contributed
to the gene pool of ancient Europeans from the Bronze Age onwards:
Anatolian farmers and Steppe pastoralists. Genetic continuity with
Scandinavian Iron Age groups was investigated using f4(YRI, test group;
test individual, Scandinavia Iron Age group) (1000 Genomes panel
merge). This statistic measures whether a test individual is consistent
with forming a clade with Scandinavian Iron Age groups to the exclusion
of a test group from outside of Scandinavia. Genetic affinities between
ancient groups and present-day populations were investigated using
f
4
(YRI, test individual; present-day test population, present-day refer-
ence population) (Scandinavian panel).
Ancestry modelling using qpAdm
We estimated ancestry proportions of Viking Age groups using
qpAdm70, which is based on f4-statistics of the from f4(X,O1;O2,O3), in
which X is either the source or target population, and O1, O2 and O3
are triplets of outgroups to the source and target groups. To minimize
batch effects and/or biases due to ancient DNA damage or SNP ascer-
tainment, we used a set of 1,800,038 transversion-only sites that were
found polymorphic with minor allele frequency≥0.5% and missing
genotype rate of≤15% in the 1000 Genomes panel merge.
Genetic diversity
The genetic diversity of ancient groups was assessed using conditional
nucleotide diversity, as previously described73. For this analysis, pair-
wise differences between individuals were calculated using SNPs
polymorphic in an outgroup population (YRI) and with a minor allele
count≥5 in the 1000 Genomes merge.
Identity-by-descent analysis
The imputed genotypes of 298 individuals were used to infer genomic
segments shared via identity-by-descent within the context of a refer-
ence panel of 1,464 present-day Europeans, using IBDseq
99
(version
r1206) with default parameters. We conducted genetic clustering by
MDS on a distance matrix obtained from pairwise identity-by-descent
sharing and UMAP to reveal fine-scale population structure among
Viking Age individuals.
Painting
To assess the fine-scale variation in genetic ancestry proportions of
Viking Age individuals we used Chromosome Painting
11
. The following
describes the general workflow of the Chromosome Painting analysis
(see Supplementary Note11 for details).
First, we created a modern reference panel using 1,675 modern
individuals sampled from northern Europe, using the standard Fin-
eSTRUCTURE pipeline. We applied ChromoPainter to paint all modern
individuals using the remaining individuals as donors using fs2.0.8.
Related individuals were identified through increased haplotype simi-
larity, and admixed individuals were identified by their FineSTRUCTURE
clustering. These were removed, leading to 1,554 unrelated individu-
als who were re-painted. These individuals were then clustered using
FineSTRUCTURE, resulting in 40 populations. After removal of small
populations and merging of the CHB and YRI subpopulations, this
resulted in 23 modern populations with geographical meaning. We
named the resulting clustering the modern reference panel, which
consists of 23 modern surrogate populations and 23 modern donor
populations (Supplementary Fig. 11.2).
Second, we created an ancient reference panel using the modern
reference panel, by applying ChromoPainter to paint all ancient indi-
viduals using the modern population palette (Supplementary Fig. 11.3).
We then created a supervised ancient population palette consisting
of 14 populations which either (a) represent a modern ancestry direc-
tion or (b) are best associated with a modern ancestry direction. The
paintings consider the average per-individual donor rate to each of
the seven modern populations, normalizing each donor label to have
a mean of 1 (Supplementary Fig. 11.4). The individuals that contribute
most to a population represent it (above a threshold amount chosen by
identifying a change point). The remaining individuals are assigned to
the population that they are best associated with. We create an ancient
population surrogate for each modern population, consisting of the
individuals that represent each modern population. For k=7 modern
populations, this results in a matrix of k=7 rows (surrogate populations)
and 2K=14 columns (donor palette populations), which captures the
ancient population structure (Supplementary Fig. 11.6).
Third, we inferred ancestry by learning about population structure in
modern individuals or ancient individuals, painting them with respect
to the ancient population panel and fitting them as a mixture using the
ancient population surrogates, using the non-negative least squares
implemented in GLOBETROTTER100 (Supplementary Information sec-
tion 11) with uncertainty estimated using 100 bootstrap replicates.
All samples were analysed by leaving out one individual per donor
population so that modern and ancient individuals are exchangeable
(as the ancient individual is itself excluded from its own ancient donor
population). We report this in a number of ways. The inferred ancestry
results (Supplementary Table6) are summarized by taking the mean
across inferred populations inSupplementary Fig. 11.11; Supplemen-
tary Fig. 11.12 shows the means over sample information labels. We
performed a spatiotemporal regression (Supplementary Table11.2)
using the model aik=αjkti + βjkxi + γjkyi + εijk in which aik is the amount of
ancestry individual i possesses from population k in regional analysis j,
t
i
is the age category of the individual (1=Iron Age, 2=Early Viking Age,
3=Viking Age, 4=Medieval) and xi and yi are the longitude and latitude
of the burial location of the individual, respectively. The modern ances-
try results are estimated using the spatial median instead of the mean,
to account for ancestry being constrained in a k-dimensional simplex
(Supplementary Fig. 11.14), with uncertainty quantified by bootstrap
resampling of individuals (Supplementary Fig. 11.15).
Fourthly, we performed sensitivity analyses to ensure that the infer-
ence procedure performed as expected. We checked that sequence
depth was not associated with cluster membership (Supplementary

Fig. 11.7), and that sequence depth did not significantly affect inferred
ancestry (Supplementary Fig. 11.8) by downsampling individuals with
high-depth data available, rephasing, re-imputing and repainting them,
and assigning ancestry using the above procedure. Results 2× and above
were extremely similar, whereas at 1× there was some loss of precision
but the broad structure remained clear.
Finally, we ran a principal components analysis of the ancient +
modern populations painted against our donor populations (Supple-
mentary Fig. 11.9) as well as an all-versus-all ChromoPainter analysis
including modern and ancient individuals (Supplementary Fig. 11.10).
Ancestry diversity measure
We wish to quantify diversity in ancestry for a population of individuals,
with diverse meaning a large deviation of individual ancestry estimates
from the average ancestry in that population. We compute the aver-
age Kullback–Leibler divergence for each individual label from the
average of that label:
∑
DA nKL Ap()=1(||)
l
l
i
n
ill()
=1
() ()
l
in which A(l) is the nl by K matrix of ancestry estimates in label l, p(l) is
the length K vector of average ancestries in that label, and
()
KL
QP q(||)=∑ log
k
K
k
q
p
=1 2
k
k
. We performed a simulation study to validate
this measure (Supplementary Information section 11, Supplementary
Fig. 11.13), which allowed us to calibrate the expected diversity as a
function of sample size.
Spatiotemporal patterns
To visualize the migration patterns of the Vikings, we used inverse dis-
tance weighting interpolation—implemented in the function idw of the
R package gstat—to interpolate the proportion of each ancient genome
that was attributed by our fineSTRUCTURE analysis (Supplementary
Table6) to one of the predefined ancestry groups: UK, Denmark, Nor-
way, Sweden, Italy, Poland and Finland. We used the Shepard method
of interpolation
12,101
with the weight for a given interpolation location x
equal to 1/(d(x,v)
2
), in which v is the location of an observed sample and
d(a,b) is the distance between two points a and b. For plotting maps,
we used a Mercator projection and downloaded coastal contours at
1:50-m scale from Natural Earth (https://www.naturalearthdata.com/).
Lactase persistence and pigmentation SNPs
For ancient populations we estimated the derived A allele frequency
of the SNP rs4988235, known to affect expression of the lactase (LCT)
gene. The ancestral G allele is responsible for lactase intolerance in
adult Europeans39. We used ANGSD60 to estimate the allele frequen-
cies of the ancient population on the basis of the genotype likelihood
data. We used the five European populations (CEU, FIN, GBR, TSI and
IBS) and two outgroups (YRI and CHB) from the 1000 Genomes Pro-
ject as comparative groups. We also included the present-day Danish
population from the IPSYCH case–cohort study44 and geographically
proximate Iron and Bronze Age populations to trace frequency shifts
of SNP rs4988235 through time. We also used ANGSD60 to estimate
the frequencies of 22 SNPs (HIrisPlex
102
) with strongest influence on
human pigmentation phenotypes in the Viking Age and Early Viking
Age Scandinavian population.
Signatures of selection
We aimed to find SNPs with allele frequencies that changed significantly
in the last 10,000 years, using our ancient human genomes to look at
the frequencies of alleles in the past. We combined our Viking Age
and Iron Age genomes with previously published present-day, Bronze
Age, Neolithic and Mesolithic sequence data typed at the Human Ori-
gins array (Supplementary Note6). We filtered for genomes that were
younger than 8000  and that were located within a bounding box
encompassing the European continent: 30°< latitude < 75° and −15°<
longitude < 45°. We then used neoscan in Ohana
36,103
to scan for variants
with allele frequencies that were strongly associated with time, after
controlling for genome-wide changes in ancestry that might have also
occurred over time. We analysed only sites with a minor allele frequency
>1% (Supplementary Note14).
Tracking the evolution of complex traits in Scandinavia
We wanted to examine whether we could identify signals of recent
population differentiation of complex traits by comparing genotypes of
Viking Age samples excavated in Scandinavia (that is, Denmark, Sweden
and Norway) with those of a present-day Scandinavian population. For
the latter, we used imputed genotypes from subjects born in Denmark
between 1981 and 2011 from the IPSYCH case–cohort study
44
. We down-
loaded summary statistics from the genome-wide association study
ATLAS webpage (https://atlas.ctglab.nl)45, from studies of 16 disease-
and anthropometric traits (excluding those related to cognition) pub-
lished in 2017 or later with SNP heritability estimated at >0.1, sample
size of >100,000 and >100 identified genome-wide significant loci. We
calculated polygenic risk scores based on independent (R
2
<0.1 within
10-Mb range) genome-wide significant allelic effects and standardized
them to a unit representing the standard deviation of the mean of their
distribution. We then removed outliers (anyone with a value for any of
the 25 principal components falling more than 4 standard deviations
away from the group mean) reiteratively from within each ancestry
group (treating the Scandinavian Viking age samples as one ancestry
group), and subsequently tested for difference in polygenic risk score
distribution between Viking Age samples and Danish-ancestry IPSYCH
random population samples using a linear regression model correcting
for sex and the 25 principal components.
Reporting summary
Further information on research design is available in theNature
Research Reporting Summary linked to this paper.
Data availability
Sequence data are available at the European Nucleotide Archive under
accession number PRJEB37976.
Code availability
Functions for calculating f-statistics are available as an R package at
GitHub (https://github.com/martinsikora/admixr).
50. Willerslev, E. & Cooper, A. Ancient DNA. Proc. R. Soc. Lond. B 272, 3–16 (2005).
51. Gilbert, M. T. P., Bandelt, H.-J., Hofreiter, M. & Barnes, I. Assessing ancient DNA studies.
Trends Ecol. Evol. 20, 541–544 (2005).
52. Schubert, M., Lindgreen, S. & Orlando, L. AdapterRemoval v2: rapid adapter trimming,
identiication, and read merging. BMC Res. Notes 9, 88 (2016).
53. Li, H. & Durbin, R. Fast and accurate short read alignment with Burrows–Wheeler
transform. Bioinformatics 25, 1754–1760 (2009).
54. Schubert, M. etal. Improving ancient DNA read mapping against modern reference
genomes. BMC Genomics 13, 178 (2012).
55. DePristo, M. A. etal. A framework for variation discovery and genotyping using
next-generation DNA sequencing data. Nat. Genet. 43, 491–498 (2011).
56. Quinlan, A. R. & Hall, I. M. BEDTools: a lexible suite of utilities for comparing genomic
features. Bioinformatics 26, 841–842 (2010).
57. Jónsson, H., Ginolhac, A., Schubert, M., Johnson, P. L. F. & Orlando, L. mapDamage2.0:
fast approximate Bayesian estimates of ancient DNA damage parameters. Bioinformatics
29, 1682–1684 (2013).
58. Fu, Q. etal. A revised timescale for human evolution based on ancient mitochondrial
genomes. Curr. Biol. 23, 553–559 (2013).
59. Renaud, G., Slon, V., Duggan, A. T. & Kelso, J. Schmutzi: estimation of contamination and
endogenous mitochondrial consensus calling for ancient DNA. Genome Biol. 16, 224 (2015).
60. Korneliussen, T. S., Albrechtsen, A. & Nielsen, R. ANGSD: analysis of next generation
sequencing data. BMC Bioinformatics 15, 356 (2014).
61. Skoglund, P., Storå, J., Götherström, A. & Jakobsson, M. Accurate sex identiication of ancient
human remains using DNA shotgun sequencing. J. Archaeol. Sci. 40, 4477–4482 (2013).
62. Weissensteiner, H. etal. HaploGrep 2: mitochondrial haplogroup classiication in the era
of high-throughput sequencing. Nucleic Acids Res. 44, W58–W63 (2016).

Article
63. Ralf, A., Montiel González, D., Zhong, K. & Kayser, M. Yleaf: software for human
Y-chromosomal haplogroup inference from next-generation sequencing data. Mol. Biol.
Evol. 35, 1291–1294 (2018).
64. Korneliussen, T. S. & Moltke, I. NgsRelate: a software tool for estimating pairwise
relatedness from next-generation sequencing data. Bioinformatics 31, 4009–4011 (2015).
65. Monroy Kuhn, J. M., Jakobsson, M. & Günther, T. Estimating genetic kin relationships in
prehistoric populations. PLoS ONE 13, e0195491 (2018).
66. Staples, J., Nickerson, D. A. & Below, J. E. Utilizing graph theory to select the largest set of
unrelated individuals for genetic analysis. Genet. Epidemiol. 37, 136–141 (2013).
67. Browning, S. R. & Browning, B. L. Rapid and accurate haplotype phasing and missing-data
inference for whole-genome association studies by use of localized haplotype clustering.
Am. J. Hum. Genet. 81, 1084–1097 (2007).
68. Martiniano, R. etal. The population genomics of archaeological transition in west Iberia:
investigation of ancient substructure using imputation and haplotype-based methods.
PLoS Genet. 13, e1006852 (2017).
69. International Multiple Sclerosis Genetics Consortium & The Wellcome Trust Case Control
Consortium 2. Genetic risk and a primary role for cell-mediated immune mechanisms in
multiple sclerosis. Nature 476, 214–219 (2011).
70. Haak, W. etal. Massive migration from the steppe was a source for Indo-European
languages in Europe. Nature 522, 207–211 (2015).
71. Gamba, C. etal. Genome lux and stasis in a ive millennium transect of European
prehistory. Nat. Commun. 5, 5257 (2014).
72. Jones, E. R. etal. Upper Palaeolithic genomes reveal deep roots of modern Eurasians. Nat.
Commun. 6, 8912 (2015).
73. Skoglund, P. etal. Genomic diversity and admixture differs for Stone-Age Scandinavian
foragers and farmers. Science 344, 747–750 (2014).
74. Schiffels, S. etal. Iron Age and Anglo-Saxon genomes from east England reveal British
migration history. Nat. Commun. 7, 10408 (2016).
75. Olalde, I. etal. Derived immune and ancestral pigmentation alleles in a 7,000-year-old
Mesolithic European. Nature 507, 225–228 (2014).
76. Sikora, M. etal. Ancient genomes show social and reproductive behavior of early Upper
Paleolithic foragers. Science 358, 659–662 (2017).
77. Fu, Q. etal. The genetic history of Ice Age Europe. Nature 534, 200–205 (2016).
78. Jones, E. R. etal. The Neolithic transition in the Baltic was not driven by admixture with
early European farmers. Curr. Biol. 27, 576–582 (2017).
79. Seguin-Orlando, A. etal. Genomic structure in Europeans dating back at least 36,200
years. Science 346, 1113–1118 (2014).
80. Raghavan, M. etal. Upper Palaeolithic Siberian genome reveals dual ancestry of Native
Americans. Nature 505, 87–91 (2014).
81. Hofmanová, Z. etal. Early farmers from across Europe directly descended from Neolithic
Aegeans. Proc. Natl Acad. Sci. USA 113, 6886–6891 (2016).
82. Damgaard, P. B. etal. 137 ancient human genomes from across the Eurasian steppes.
Nature 557, 369–374 (2018).
83. Günther, T. etal. Population genomics of Mesolithic Scandinavia: investigating early
postglacial migration routes and high-latitude adaptation. PLoS Biol. 16, e2003703 (2018).
84. Mittnik, A. etal. The genetic prehistory of the Baltic Sea region. Nat. Commun. 9, 442 (2018).
85. Kılınç, G. M. etal. The demographic development of the irst farmers in Anatolia.
Curr. Biol. 26, 2659–2666 (2016).
86. Lazaridis, I. etal. Genetic origins of the Minoans and Mycenaeans. Nature 548, 214–218
(2017).
87. de Barros Damgaard, P. etal. The irst horse herders and the impact of early Bronze Age
steppe expansions into Asia. Science 360, eaar7711 (2018).
88. Valdiosera, C. etal. Four millennia of Iberian biomolecular prehistory illustrate the
impact of prehistoric migrations at the far end of Eurasia. Proc. Natl Acad. Sci. USA 115,
3428–3433 (2018).
89. Martiniano, R. etal. Genomic signals of migration and continuity in Britain before the
Anglo-Saxons. Nat. Commun. 7, 10326 (2016).
90. Mathieson, I. etal. The genomic history of southeastern Europe. Nature 555, 197–203
(2018).
91. Gallego Llorente, M. etal. Ancient Ethiopian genome reveals extensive Eurasian
admixture throughout the African continent. Science 350, 820–822 (2015).
92. Broushaki, F. etal. Early Neolithic genomes from the eastern Fertile Crescent. Science
353, 499–503 (2016).
93. Veeramah, K. R. etal. Population genomic analysis of elongated skulls reveals extensive
female-biased immigration in Early Medieval Bavaria. Proc. Natl Acad. Sci. USA 115,
3494–3499 (2018).
94. Amorim, C. E. G. etal. Understanding 6th-century barbarian social organization and
migration through paleogenomics. Nat. Commun. 9, 3547 (2018).
95. Olalde, I. etal. A Common genetic origin for early farmers from Mediterranean Cardial
and Central European LBK cultures. Mol. Biol. Evol. 32, 3132–3142 (2015).
96. Olalde, I. etal. The Beaker phenomenon and the genomic transformation of northwest
Europe. Nature 555, 190–196 (2018).
97. Alexander, D. H., Novembre, J. & Lange, K. Fast model-based estimation of ancestry in
unrelated individuals. Genome Res. 19, 1655–1664 (2009).
98. Behr, A. A., Liu, K. Z., Liu-Fang, G., Nakka, P. & Ramachandran, S. pong: fast analysis and
visualization of latent clusters in population genetic data. Bioinformatics 32, 2817–2823
(2016).
99. Browning, B. L. & Browning, S. R. Detecting identity by descent and estimating genotype
error rates in sequence data. Am. J. Hum. Genet. 93, 840–851 (2013).
100. Hellenthal, G. etal. A genetic atlas of human admixture history. Science 343, 747–751 (2014).
101. Fortin, M.-J. & Dale, M. R. T. Spatial Analysis: A Guide for Ecologists (Cambridge Univ.
Press, 2005).
102. Walsh, S. etal. The HIrisPlex system for simultaneous prediction of hair and eye colour
from DNA. Forensic Sci. Int. Genet. 7, 98–115 (2013).
103. Cheng, J. Y., Mailund, T. & Nielsen, R. Fast admixture analysis and population tree
estimation for SNP and NGS data. Bioinformatics 33, 2148–2155 (2017).
Acknowledgements This work was supported by the Mærsk Foundation, the Lundbeck
Foundation, the Novo Nordisk Foundation, the Danish National Research Foundation,
University of Copenhagen (KU2016) and the Wellcome Trust (grant no. WT104125MA). E.W.
thanks St John’s College, Cambridge for providing an excellent environment for scientiic
thoughts and collaborations. S.R. was supported by the Novo Nordisk Foundation
(NNF14CC0001). F.R. was supported by a Villum Fonden Young Investigator Award (project no.
00025300). G.S. and E.C. were supported by a Marie Skłodowska-Curie Individual Fellowship
‘PALAEO-ENEO’, a project funded by the European Union EU Framework Programme for
Research and Innovation Horizon 2020 (grant agreement number 751349). R.M. was supported
by an EMBO Long-Term Fellowship (ALTF 133-2017). M.C. is supported by the Canada Research
Chairs Program (231256), the Canada Foundation for Innovation (36801) and the British
Columbia Knowledge Development Fund (962-805808). I. Moltke was supported by a YDUN
grant from Independent Research Fund Denmark (DFF-4090-00244) and a Villum Fonden
Young Investigator Award (project no. 19114). N.P. and C.H.J. are supported by the Swedish
Research Council (2015-00466).N.G. was supported by the Program of Fundamental Scientiic
Research of the State Academies of Sciences, Russian Federation, state assignment no. 0184-
2019-0006. D.G.B. and L.M.C. were supported by Science Foundation Ireland/Health Research
Board/Wellcome Trust award no. 205072.We thank the iPSYCH Initiative, funded by the
Lundbeck Foundation (grant numbers R102-A9118 and R155-2014-1724), for supplying SNP
frequency estimates from the present-day Danish population for comparison with Viking Age
samples; M. Jakobsson and A. Götherström for providing preliminary access to the sequencing
data of 23 Viking Age samples from Sigtuna; M. Corrente for providing access to the skeletal
remains from Cancarro; and N. M. Mangialardi and M. Maruotti for the useful suggestions;
Greenland National Museum and Archives, as well as the Gotland Museum, for permission
to sample their skeletons; J. Kavanagh for providing information on his excavation, and
L. Buckley, D. Keating and B. Ó Donnabháin for analysing the remains; R. Breward and
J. Murden from the Dorset County Museum for allowing access to their assemblage for DNA
sampling;J. Hansen and M. B. Henriksen at Odense Bys Museer for allowing sampling of
skeletal material from Hessum and Galjedil; Moesgaard Museum for allowing sampling of
skeletal material from Hesselbjerg; C. Bertilsson, P. Lingström, B. Lundberg, K. Lidén and J.
Andersson for their help in sampling the ancient human remains; L. Drenzel for permission to
sample the human remains; C. Ödman for suggesting relevant material for this study;
Ł. Stanaszek, M. Zaitz and the Regional Museum in Cedynia for providing samples; L. Vinner,
A. Seguin-Orlando, K. Magnussen, L. Petersen, C. Mortensen and M. J. Jacobsen at the Danish
National Sequencing Centre for producing the analysed sequences; P. S. Olsen and T. Brand
for technical assistance in the laboratories; R. M. Durbin and J. H. Barrett for comments and
suggestions; and J. Wilson, J. Jesch, E. Harlitz-Kern and F. Martín Racimo for their feedback.
Author contributions E.W. initiated and led the study. E.W., A.M., D.J.L., Martin Sikora, F.R., R.N.,
K.K., L.H., S.M.S., J.B., N.P., T.W., A.I., M.E.A., M.W.P., N.L., J.A., I. Moltke and A.A. designed the
study. A.M., P.d.B.D., L.M.C., M.M.B., A.K.F., I.L. and J.S. produced the data. A.M., D.J.L., Martin
Sikora, F.R., S.R., I. Moltke, R.N., T.W., L.M.C., E.J., A.I., M.W.P., T.K., R.M., G.R., C.B., J.V.M.-M.,
H.M., A.A., J.C., K.H.I. and M.E.A. analysed or assisted in analysis of data. E.W., A .M., D.J.L.,
Martin Sikora, F.R., S.M.S., K.K., L.H., R.N., M.C.and A.I. interpreted the results with
considerable input from I. Moltke, M.E.A., M.W.P., T.K., H.W., R.M., G.R., T.W., C.H.J., J.A., N.L.,
N.P., J.B., A.A., M.T.P.G., L.O. and other authors. E.W., A.M., D.J.L., Martin Sikora, F.R., S.M.S., K.K.
and L.H. wrote the manuscript with considerable input from M.C., J.B., N.P., I. Moltke, N.L., A.I.,
R.M., E.J., J.A., M.L.J., C.H.J., M.W.P., M.E.A ., G.R. and M.M., with contributions from all authors.
A.M., L.M.C., M.W.P., H.W., M.M.B., P.d.B.D., A.K.F., M.A., R.A., M.M., E.C., G.S., A.B., A.F.,
B. Schütz, B. Skar, C.A., C.F., D.B., D.P., G.T.-W., H.G., I.L., I.G., I. Mainland, I.P., I.M.M., J.M.,
J. Gibson, J.P., J. Gustafsson, L. Simpson, L. Strand, L.L., Maeve Sikora, M.F., M.V., M.R., M.B.,
T.P., M. Søvso, N.G., T.C., O.K., O.U., P.F., P.H., S.S., S.V.A., S.E., V.M., W.B., Y.M., P.P., M.D.J., A.P.,
D.G.B., M.L.J., J.A., N.L., N.P., M.T.P.G., M.E.A., J.B. and E.W. excavated, curated, sampled and/or
described analysed skeletons.
Competing interests The authors declare no competing interests.
Additional information
Supplementary information is available for this paper at https://doi.org/10.1038/s41586-020-
2688-8.
Correspondence and requests for materials should be addressed to R.N., T.W. or E.W.
Peer review information Nature thanks James Barrett, Wolfgang Haak and Pontus Skoglund
their contribution to the peer review of this work.
Reprints and permissions information is available at http://www.nature.com/reprints.

Extende d Data Fig. 1 | Viking Age archaeological sites. Ex amples of a few
archaeolo gical Vikin g Age sites and sa mples used in t his study. a, SalmeII
ship burial si te of the Early Vik ing Age, excavated i n present-day Eston ia:
schemat ic of skeletons (top lef t) and aerial imag es of skeletons (top ri ght, and
bot tom). b, Ridgeway Hill mas s grave dated to the t enth or eleventh c entury
, located on the cr est of Ridgeway H ill near Weymouth, o n the south coa st
of England (repro duced with p ermission f rom Dorset Cou nty Council /Oxford
Archaeology). Around 50predominantly young adult male individuals were
excavate d. c, The site of Ballad oole, around 900, a Vikin g was buried in an
oak ship at Balla doole (Arbor y) in the south ea st of the Isle of Ma n. d, Viking
Age archaeol ogical site in Var nhem, in Skara mu nicipality (Swede n). Schematic
map of the churc h foundation (le ft) and the excavate d graves (red marking s) at
the early Chri stian cemet ery in Varnhem; foun dations of the V iking Age stone
church in Varnhe m (middle) and the remain s of a 182-cm-long mal e individual
(no. 17) buried i n a lime stone cof fin close to t he church foundat ions (right).

Article
Extende d Data Fig. 2 | Model-based clustering analysis. Admixture p lot (K=2
to K=5) for 567 ancient individual s, spanning 71 p opulations . This fig ure is a
subset of th e most relevant in dividuals and p opulations f romSupplementa ry
Fig. 7.2; see Supple mentary N ote7 for further det ails. This p lot consists
of 378ancient sam ples from this st udy; Viking Age s amples from Si gtuna
(Sweden)10 (n=2 1), I cel and 18 (n=22) and other ancient co mparative group s
(n=146).

Extende d Data Fig. 3 | Fin e-scale po pulation st ructure. T he point cloud at
the top cent re shows an alterna tive view of the UM AP result from Fi g.2b, with
all ancient i ndividuals colo ured on the basis o f analysis group. T he framed
panels surrounding the point cloud highlight particular ancestry clusters
(as indicated ), with labels and lar ger symbols co rresponding t o the median
coordinate s for the respec tive group. Simila rly, the larger bottom pa nel shows
median gro up coordinates for t he large central p oint cloud, which i ncludes the
vast major ity of European i ndividuals fro m the Bronze Age onward s.

Article
Extended Data Fig. 4 | Ancestry modelling for distal sources. a, Contrast ing
allele-sha ring betwee n Anatolian far mers (Barcin_ EN) and Stepp e pastoralist s
(Yamnaya_ EBA) for European indi viduals from th e Bronze Age and later. Viol in
plots showing distributions of statistics f4(YRI,test individual;Barcin_
EN,Yamnaya_EBA) for n=515 in dividuals wit h a minimum of 1,00 0,000 SNPs
with genot ypes and gro ups with at lea st 2such individual s. b, Ancestry
proport ions of analysis g roups from the Bron ze Age and later infer red using
qpAdm. Target groups we re modelled usin g three distal s ources repres enting
European hunter-gatherer (Loschbour_M), Anatolian farmer (Barcin_EN) and
Steppe pas toralist (Yamnaya _EBA) ances try. Sample sizes for t arget groups c an
be found in Suppl ementary Table10. Error bar s indicate st andard error
obtaine d from qpAdm. c, Anc estry propo rtions of anal ysis groups for whic h
the three- source model wa s rejected us ing qpAdm (P<0.05). Target groups
were modelle d including one ad ditional dist al source repres enting eithe r
Steppe hunter-gatherer (Botai_EBA), Caucasus hunter-gatherer
(CaucasusHG_M) or East-Asian-related (XiongNu_IA) ancestry.

Extende d Data Fig. 5 | Se e next page for capti on.

Article
Extended Data Fig. 5 | Ancestry modelling for proximate sources. a, Testing
for continuit y between Eu ropean Iron Age an d later Viking Age a nd Medieval
groups. Co loured squares d epict wheth er a particular t arget group (row) can
be modelle d using a single so urce group (column). Pvalues for f4 rank of 0
(correspond ing to a single sour ce group) were obtain ed using qpAdm wit h a set
of 15 outg roups, which incl uded European Br onze Age groups th at preceded
the source g roups. Sample s izes for target gr oups can be found in
Supplementary Table12. b, Two-way admixture ance stry propor tions of tar get
groups for whic h a single source wa s rejected (P≤0.05). Target group s were
modelled u sing additiona l proximate Bronze an d Iron Age sources . Sample
sizes for tar get groups can b e found in Supplemen tary Table13. For both a and
b, only ancient g roups contai ning at least 3indiv iduals with a min imum of
1,000,000 S NPs with genot ypes are plot ted. c, Contrasting allele-sharing
betwee n populations o f present-day Denmar k and other popula tions. Violin
plots showing distributions of statistics f4(YRI,test individual;panel
population,Denmark) for n=489 indiv iduals with a mini mum of 50,000 SNPs
with genot ypes and gro ups with at lea st 2 such individu als. Median valu es for
distributions are indicated with horizontal lines.

Extende d Data Fig. 6 | An cestry di versity of dif ferent pop ulation gro ups.
Diversit y of different lab els (that is, sample lo cations com bined with his torical
age) are shown as a fu nction of thei r sample size. T he diversity me asure is the
Kullback–Leibler dive rgence from the la bel means, c apturing the dive rsity of a
group with r espect to th e average of that group (se e Supplement ary Note11 for
details). Large r values are more diver se, although a de pendence on s ample size
is expec ted. The simula tion expect ation for the bes t fit to the dat a (0=0.2) is
shown.

Article
Extende d Data Fig. 7 | Polyg enic risk sc ores. Polygen ic risk scores (PR S) for 16
complex huma n traits in 148 Vik ing Age samples f rom Denmark, Swe den and
Norway, compare d against a referen ce sample of 20, 551 Danish-anc estry
individua ls randomly drawn f rom all individu als born in Denm ark in 1981–2005.
The PRS is i n each case bas ed on allelic ef fects for >100 indep endent
genome-wide significant SNPs from recent genome-wide association studies
of the respe ctive trait s and standardis ed to a mean of 0 and st andard deviat ion
of 1 in the entire s ample. Diffe rence in PRS was e stimated in a lin ear regress ion
correct ing for sex and 25 pr incipal compo nents of overall gen etic struct ure.
The plott ed BETA indicates t he coeff icient for the tes t-group (Viking Age
sample) PRS co mpared to that of th e Danish compari son sample, w ith error
bars indica ting the 95% conf idence inte rval of BETA, and P indi cating the
two-tailed Pvalue of the corre sponding t-test (not c orrected for nu mber of
tests). Only PR S for black hair colour i s signific antly differen t between the
groups af ter taking ac count of multiple t esting.

Extende d Data Fig. 8 | Pos itive selec tion in Europ e. a, Manhattan p lots of the
likelihood ra tio scores in favour o f selection l ooking at the en tire 10,000-year
period (top, gene ral scan), the perio d up to 4,000 years befor e present (middle,
ancient sc an) and the perio d from 4,000 years b efore present up to th e present
day (botto m, recent sca n). The highlighte d SNPs have a score large r than the
99.9% quantile of the em pirical distr ibution of log-likeli hood ratios, a nd have at
least two n eighbourin g SNPs (± 500 kb) with a s core larger than the s ame
quantile. n=1,185 ge nomes are used i n the select ion scan. b, Frequenc ies of the
derived A alle le rs498823 5 SNP responsi ble for lactase p ersistenc e in humans
for different V iking Age group s, present-day popula tions from the 10 00
Genome s Project as well a s relevant Bronze A ge population pan els. The
numbers at t he top of the bars de note the sample si ze on which the allel e
frequenc y estimates a re based.

1
nature research | reporting summary October 2018
Corresponding author(s): Eske Willerslev
Last updated by author(s): May 3, 2020
Reporting Summary
Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency
in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist.
Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.
n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement
A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly
The statistical test(s) used AND whether they are one- or two-sided
Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested
A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons
A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient)
AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)
For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted
Give P values as exact values whenever suitable.
For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes
Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated
Our web collection on statistics for biologists contains articles on many of the points above.
Software and code
Policy information about availability of computer code
Data collection No specific software was used for data collection. All software used in this study is listed below.
Data analysis All software used in this work is publicly available. Corresponding publications are cited in the main text and supplementary material. List
of software and respective versions:
CASAVA v1.8.2
AdapterRemoval v2.1.3
bwa v0.7.10
bwa mem 0.7.10
picard tools v1.127
bamUtil v1.0.14
samtools v1.3.1
GATK v3.3.0
pysam 0.7.4 (python module)
bedtools 2.27.1
mapDamage2.0
contamMix v1.0-5
SHRiMP 2.2.3
YFitter v0.2
Haplogrep 2.0
FineSTRUCTURE
ANGSD v0.915
IBDseq v.r1206
PRANK v.150803
pathPhynder
BEASTv1.8.2

2
nature research | reporting summary October 2018
schmutzi v.1.5.4
Yleaf v2
admixtools v4.1
NGSrelate v.1
GLOBETROTTER
READ
PRIMUS v1.9
plink and v1.9
ADMIXTURE v1.3
RAxML-8.1.15
SnpEff
SPAdes-3.9.0
R 3.2.3
Ohana
Beagle v4.1
python 2.7.12
perl v5.22.1
CALIB
FigTree v.1.4.4
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers.
We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.
Data
Policy information about availability of data
All manuscripts must include a data availability statement. This statement should provide the following information, where applicable:
- Accession codes, unique identifiers, or web links for publicly available datasets
- A list of figures that have associated raw data
- A description of any restrictions on data availability
Sequence data are available at the European Nucleotide Archive under accession number PRJEB37976.
Field-specific reporting
Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.
Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf
Life sciences study design
All studies must disclose on these points even when the disclosure is negative.
Sample size We did not rely on statistical methods to predetermine sample sizes. Sample sizes in ancient population genetic studies are limited by the
number of samples yielding endogenous DNA proportions amenable to whole genome sequencing.
Data exclusions We selected 442 samples for whole-genome sequencing, out of all (n=528) screened samples, based on their endogenous content and low
contamination estimates. These criteria are described in detail in Supplementary Notes 2 and 3. Furthermore, closely related individuals were
excluded from analyses requiring population allele frequencies.
Replication We did not attempt to specifically replicate experimental findings. But we note that samples from the same population carry similar genetic
signatures. Moreover, genome-wide data allows for the analysis of multiple realisations of the sample history, by studying hundreds of
thousands of SNP sites.
Randomization We did not implement any randomization as no experimental groups or effect sizes were measured in this study.
Blinding No blinding techniques were implemented, as experimental group assignment is not relevant for population genetic studies of this kind.
Reporting for specific materials, systems and methods
We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material,
system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.

3
nature research | reporting summary October 2018
Materials & experimental systems
n/a Involved in the study
Antibodies
Eukaryotic cell lines
Palaeontology
Animals and other organisms
Human research participants
Clinical data
Methods
n/a Involved in the study
ChIP-seq
Flow cytometry
MRI-based neuroimaging