Complete Mitochondrial Genomes of Ancient Canids Suggest a European Origin of Domestic Dogs

Abstract

Dog Domestication
The precise details of the domestication and origins of domestic dogs are unclear. Thalmann et al. (p. 871 ; see the cover) analyzed complete mitochondrial genomes from present-day dogs and wolves, as well as 18 fossil canids dating from 1000 to 36,000 years ago from the Old and New Worlds. The data suggest that an ancient, now extinct, central European population of wolves was directly ancestral to domestic dogs. Furthermore, several ancient dogs may represent failed domestication events.

Full text

DOI: 10.1126/science.1243650
, 871 (2013);342 Science et al.O. Thalmann
European Origin of Domestic Dogs
Complete Mitochondrial Genomes of Ancient Canids Suggest a
This copy is for your personal, non-commercial use only.
clicking here.colleagues, clients, or customers by , you can order high-quality copies for yourIf you wish to distribute this article to others
here.following the guidelines can be obtained byPermission to republish or repurpose articles or portions of articles
): November 14, 2013 www.sciencemag.org (this information is current as of
The following resources related to this article are available online at
http://www.sciencemag.org/content/342/6160/871.full.html
version of this article at: including high-resolution figures, can be found in the onlineUpdated information and services,
http://www.sciencemag.org/content/suppl/2013/11/14/342.6160.871.DC2.html
http://www.sciencemag.org/content/suppl/2013/11/13/342.6160.871.DC1.html
can be found at: Supporting Online Material
http://www.sciencemag.org/content/342/6160/871.full.html#related
found at: can berelated to this article A list of selected additional articles on the Science Web sites
http://www.sciencemag.org/content/342/6160/871.full.html#ref-list-1
, 32 of which can be accessed free:cites 72 articlesThis article
registered trademark of AAAS. is aScience2013 by the American Association for the Advancement of Science; all rights reserved. The title CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
on November 14, 2013www.sciencemag.orgDownloaded from on November 14, 2013www.sciencemag.orgDownloaded from on November 14, 2013www.sciencemag.orgDownloaded from on November 14, 2013www.sciencemag.orgDownloaded from on November 14, 2013www.sciencemag.orgDownloaded from

effect of the E1021K mutation on the PI3Kdac-
tivity may be cell-type or stimulus-specific, or it
may be compensated for by effects of other PI3K
isoforms or PTEN. Nevertheless, we cannot ex-
clude that a subtle defect in neutrophil function
may contribute to the disease pathogenesis in
these patients.
In summary, we have described a PID caused
by a recurrent autosomal-dominant germline
mutation E1021K in the PIK3CD gene that en-
codes p110d. We found it in 17 patients from
seven unrelated families, suggesting that it is
frequent among PID patients and may explain a
substantial fraction of patients with recurrent
respiratory infections and bronchiectasis. Our
rapid genotyping assay should facilitate screen-
ing for the E1021K mutation in existing PID and
bronchiectasis cohorts, as well as new patients.
The E1021K mutation was previously noted in
one Taiwanese patient with recurrent respiratory
infections and PID; however, its causative and
pathogenic role has not been demonstrated (23).
Here, we have shown that E1021K increases
PI3Kdactivity, augmenting the production of
PIP
3
and activating the downstream AKT protein
in lymphocytes. This leads to defects in T and B
cell function and inefficient immune responses to
bacterial pathogens, predisposing to recurrent res-
piratory infections and eventually to bronchiectasis.
We named this disorder activated PI3K-dsyn-
drome (APDS).
Activation of the PI3K pathway is associated
with malignant transformations, and it has been
shown that overexpression of p110dcan trans-
form cells (24). To date, only one of our APDS
patients, P13, has been diagnosed with lymphoma
(Table 1). Nonetheless, the oncogenic potential of
PI3K up-regulation can be enhanced by addition-
al mutations (25,26). Therefore, APDS patients
may be at increased risk of leukemia or lymphoma
if they acquire additional somatic mutations.
The APDS patients described here had been
treated with immunoglobulin replacement and
antibiotics. Despite this, there is evidence of con-
siderable airway damage in most cases. Because
of progressive severe disease after splenectomy,
patient P8 underwent allogeneic hematopoietic
stem cell transplantation (HSCT) at the age of
8 years. One year after HSCT, his clinical con-
dition had improved dramatically, suggesting that
HSCT may be a long-term treatment option for
young patients. Nevertheless, our results raise the
possibility that selective p110dinhibitors, such as
GS-1101, may be an alternative effective ther-
apeutic approach in APDS patients. GS-1101
(CAL-101 or Idelalisib) has been tested in phase
1 and 2 clinical trials for treatment of chronic
lymphocytic leukemia (www.clinicaltrials.gov).
The possibility of treating APDS patients with
p110dinhibitors should therefore be considered.
References and Notes
1. A. F. Barker, N. Engl. J. Med. 346, 1383–1393 (2002).
2. A. Durandy, S. Kracker, A. Fischer, Nat. Rev. Immunol.
13, 519–533 (2013).
3. W. Al-Herz et al., Front Immunol 2, 54 (2011).
4. M. J. Bamshad et al., Nat. Rev. Genet. 12,745–755 (2011).
5. Materials and methods are available as supplementary
materials on Science Online.
6. A. Hodgkinson, A. Eyre-Walker, Nat. Rev. Genet. 12,
756–766 (2011).
7. Z. A. Knight, M. E. Feldman, A. Balla, T. Balla,
K. M. Shokat, Nat. Protoc. 2, 2459–2466 (2007).
8. C. Sadhu, B. Masinovsky, K. Dick, C. G. Sowell,
D. E. Staunton, J. Immunol. 170, 2647–2654 (2003).
9. B. J. Lannutti et al., Blood 117, 591–594 (2011).
10. E. D. Scheeff, P. E. Bourne, PLOS Comput. Biol. 1, e49 (2005).
11. O. Vadas, J. E. Burke, X. Zhang, A. Berndt, R. L. Williams,
Sci. Signal. 4, re2 (2011).
12. D. Mandelker et al., Proc. Natl. Acad. Sci. U.S.A. 106,
16996–17001 (2009).
13. J. E. Burke, O. Perisic, G. R. Masson, O. Vadas,
R. L. Williams, Proc. Natl. Acad. Sci. U.S.A. 109,
15259–15264 (2012).
14. J. E. Burke et al., Structure 19, 1127–1137 (2011).
15. K. Okkenhaug, Annu. Rev. Immunol. 31, 675–704 (2013).
16. J. Clark et al., Nat. Methods 8, 267–272 (2011).
17. F. Sallusto, J. Geginat, A. Lanzavecchia, Annu. Rev.
Immunol. 22, 745–763 (2004).
18. A. N. Anzelon, H. Wu, R. C. Rickert, Nat. Immunol. 4,
287–294 (2003).
19. A. Suzuki et al., J. Exp. Med. 197, 657–667 (2003).
20. S. A. Omori et al., Immunity 25, 545–557 (2006).
21. M. L. Janas et al., J. Immunol. 180, 739–746 (2008).
22. A. M. Condliffe et al., Blood 106, 1432–1440 (2005).
23. S. T. Jou et al., Int. J. Immunogenet. 33, 361–369 (2006).
24. S. Kang, A. Denley, B. Vanhaesebroeck, P. K. Vogt,
Proc. Natl. Acad. Sci. U.S.A. 103, 1289–1294 (2006).
25. J. A. Engelman, Nat. Rev. Cancer 9, 550–562 (2009).
26. K. M. Kinross et al., J. Clin. Invest. 122, 553–557 (2012).
Acknowledgments: S.N. is a Wellcome Trust Senior Research
Fellow in Basic Biomedical Science (095198/Z/10/Z). S.N. is
also supported by the European Research Council (ERC)
Starting grant 260477 and the European Union (EU) FP7
collaborative grant 261441 (PEVNET project). S.N., A.C.,
D.K., and R.D. are supported by the National Institute for
Health Research (NIHR) Cambridge Biomedical Research
Centre. O.V. was supported by a Swiss National Science
Foundation fellowship (grant PA00P3_134202) and a European
Commission fellowship (FP7-PEOPLE-2010-IEF, no. 275880).
R.L.W. was supported by the Medical Research Council (file
reference U105184308). T.C. is supported by the National
Children’s Research Centre, Our Lady’s Children’s Hospital,
Crumlin, Dublin, Ireland. E.B.-H. is supported by a Wellcome
Trust Translational Medicine and Therapeutics award. A.C. is
supported by the Medical Research Council UK and the British
Lung Foundation. K.O. is supported by a strategic grant from
the Biotechnology and Biological Sciences Research Council
and a New Investigator Award from the Wellcome Trust. P.H.
and L.S. are funded by an Institute Programme grant from the
Biotechnology and Biological Sciences Research Council
(BB/J004456/1). S.K. is a Centre National de la Recherche
Scientifique (CNRS) researcher. A.D., A.F., and S.K. are funded
by Institut National de la Santé et de la Recherche Médicale;
A.D. is supported by the EU FP7 EUROPAD contract 201549,
Association Contre Le Cancer, and Agence Nationale de la
Recherche (grant 2010-CSRD). A.F. is supported by the EU FP7
ERC PIDIMMUNE grant 249816. G.B.-M. was supported by a
sabbatical grant from PASPA-DGAPA-UNAM. E.C. is a paid
consultant for GlaxoSmithKline, Roche, and Novartis; A.C. is a
paid consultant for GlaxoSmithKline; P.H. and L.S are paid
consultants for GlaxoSmithKline and Karus Therapeutics Ltd;
K.O. is a paid consultant for GlaxoSmithKline. Requests for
DNA of individual patients will require informed consent from
the patients and samples will be available under a material
transfer agreement. The p110dknockout mice are available
from the Babraham Institute under a material transfer
agreement. The mutation has been submitted to the ClinVar
database; accession no. SCV000083058.
Supplementary Materials
www.sciencemag.org/content/342/6160/866/suppl/DC1
Materials and Methods
Supplementary Text
Figs. S1 to S9
Tables S1 and S2
References (27–37)
16 July 2013; accepted 23 September 2013
Published online 17 October 2013;
10.1126/science.1243292
Complete Mitochondrial Genomes of
Ancient Canids Suggest a European
Origin of Domestic Dogs
O. Thalmann,
1
*B. Shapiro,
2
P. Cui,
3
V. J. Schuenemann,
4
S. K. Sawyer,
3
D. L. Greenfield,
5
M. B. Germonpré,
6
M. V. Sablin,
7
F. López-Giráldez,
8
X. Domingo-Roura,
9
†H. Napierala,
10
H-P. Uerpmann,
4
D. M. Loponte,
11
A. A. Acosta,
11
L. Giemsch,
12,13
R. W. Schmitz,
12
B. Worthington,
14
J. E. Buikstra,
15
A. Druzhkova,
16
A. S. Graphodatsky,
16
N. D. Ovodov,
17
N. Wahlberg,
1
A. H. Freedman,
5
R. M. Schweizer,
5
K.-P. Koepfli,
18
J. A. Leonard,
19
M. Meyer,
3
J. Krause,
4
S. Pääbo,
3
R. E. Green,
20
R. K. Wayne
5
*
The geographic and temporal origins of the domestic dog remain controversial, as genetic data suggest a
domestication process in East Asia beginning 15,000 years ago, whereas the oldest doglike fossils are
found in Europe and Siberia and date to >30,000 years ago. We analyzed the mitochondrial genomes
of 18 prehistoric canids from Eurasia and the New World, along with a comprehensive panel of modern
dogs and wolves. The mitochondrial genomes of all modern dogs are phylogenetically most closely related
to either ancient or modern canids of Europe. Molecular dating suggests an onset of domestication
there 18,800 to 32,100 years ago. These findings imply that domestic dogs are the culmination of a
process that initiated with European hunter-gatherers and the canids with whom they interacted.
Dogs are one of the best known examples
of domestication, the process of species
modification over time by human-induced
selection (1). Domestication often leads to in-
creased phenotypic variation and a geographic
distribution that can be heavily influenced by hu-
man dispersal. The extensive phenotypic variation
among dog breeds hinders a simple inference of
dog origins based on the presence of traits shared
between dogs and any specific population of the
www.sciencemag.org SCIENCE VOL 342 15 NOVEMBER 2013 871
REPORTS

gray wolf (Canis lupus) from which dogs derive
(2–4). Furthermore, inferences from genetic data
are confounded by a long history of trade and ad-
mixture among dogs from disparate geographic
areas, ancient and ongoing local admixture with
wolves, intense inbreeding within some lineages,
and the stochastic effects of incomplete lineage
sorting. Nevertheless, centers of dog origins from
genetic data have been proposed, including the
Middle East and East Asia (5–7). However, the
oldest putative dog remains are found in West-
ern Europe and Siberia and date from 15,000 to
36,000 years ago (2,8), although the classifica-
tion of these specimens remains contentious (9).
The earliest putative dog remains from the Mid-
dle East and East Asia are no older than about
13,000 years ago [see table S3 (10)].
DNA extracted from the earliest canids showing
phenotypic evidence of domestication (2,8,11–14)
can potentially be used to test hypotheses about
the origin of modern dogs. We generated com-
plete and partial mitochondrial genomes from
18 prehistoric canids and 20 modern wolves of
Eurasian and American origin (Table 1 and table
S2) by performing DNA capture followed by high-
throughput sequencing (15). The DNA fragments
recovered from these samples show patterns ex-
pected of ancient DNA, including a correlation
between sequence length and sample age (fig.
S1) and deamination patterns typical of ancient
DNA (15) (fig. S2). After filtering, iterative as-
sembly, and exclusion of mitochondrial genomes
with less than 50% of the length recovered, we
obtained a median 12-fold (1.9 to 625.7) cov-
erage of the 18 ancient genomes, with on average
15,014 (8667 to 16,415) nucleotides supported
by at least twofold coverage. These mtDNA as-
semblies from ancient canids were compared with
complete mitochondrial genome sequences from
49 wolves; 77 dogs, including divergent dog breeds
such as Basenji and Dingo; three recently published
Chinese indigenous dogs (7); and four coyotes
totaling 148 mitochondrial genomes.
Phylogenetic analyses of the mitochondrial
genome data using maximum likelihood, coales-
cence and Bayesian approaches all reveal a well-
resolved phylogeny (Fig. 1). Although dogs and
wolves are not reciprocally monophyletic, all mod-
ern dogs and many wolf populations fall within
one of several well-supported clades (Fig. 1 and
fig. S9). Within this tree topology, dogs fall with-
in one of four clades (Dog A to D) (Fig. 1 and fig.
S9), with clade A containing the majority of dog
sequences (64%). Three haplotypes from ancient
Belgian canids form the most deeply diverging
group in the tree. Although the cranial morphology
of one of these, the Goyet dog (Belgium 36,000)
(Table 1 and table S1) has been interpreted as dog-
like (2), its mtDNA relation to other canids places
it as an ancient sister-group to all modern dogs
and wolves rather than a direct ancestor of dogs.
One of the Belgian specimens (Belgium 26,000)
has been found to be uniquely large (2) and could
be related to a genetically and morphologically
distinct form of wolves from Late Pleistocene de-
posits of the High Arctic permafrost (16). However,
none of the sequences from the three northerly
permafrost wolves (Alaska 28,000,Alaska21,000,
and Alaska 20,800) (Fig. 1) fall within or are sister
to this clade. Given their mitochondrial distinctive-
ness, the Belgian canids, including the Goyet dog,
may represent an aborted domestication episode
or a phenotypically distinct, and not previously
recognized, population of gray wolf.
Dog clades A, C, and D, which make up 78%
of dog sequences in our study, are each sister to
one or more ancient canids of Europe. The most
diverse of these groups is clade A, which includes
divergent breeds, such as Basenji and Dingo, and
two of the Chinese indigenous dogs (7). Moreover,
three pre-Colombian New World dogs, ranging in
age from 1000 to 8500 years ago, fall within dog
clade A (Table 1 and table S1). The calculated time
to the most recent common ancestor (MRCA) of
dog clade A and ancient New World dog sequences
is ~18,800 years ago [95% highest posterior den-
sity (HPD): 15,100 to 22,600] (fig. S10), which
supports the hypothesis that pre-Colombian dogs
in the New World share ancestry with modern dogs.
Thus, these dogs likely arrived with the first humans
in the New World (17,18). The clade comprising
these ancient New World dogs and modern dog
clade A is most closely related to an ancient wolf
sequence from the Kesslerloch cave in Switzerland
(Switzerland 2 14,500) with a MRCA that existed
~32,100 years ago (95% HPD: 27,500 to 36,700).
The lowest diversity dog clade (D) contains
only sequences from two Scandinavian breeds
and is sister to an ancient wolflike canid from
Switzerland with a common ancestor that existed
~18,300 years ago (95% HPD: 15,300 to 21,900).
This grouping is most closely related to another
sequence from ancient European wolves, as well
as extant wolves from Poland and Italy, and is
rooted with the sequence from a putative early
dog from the Altai Mountains in Russia (13). The
grouping of clade D with ancient wolf lineages
and the association of the Altai specimen with this
clade do not support recent common ancestry of
the Altai specimen lineage with the great major-
ity of modern dogs. However, clade D dog hap-
lotypes could have been captured as a result of
interactions between ancient wolves and early
humans that migrated into Scandinavia (19).
The closest sister group for dogs in clade C,
which makes up 12% (9 of 77) of modern dog se-
quences, are two morphologically distinct ancient
dogs from Bonn-Oberkassel (12) and the Kartstein
cave in Germany (14)(Germany14,700 and Ger-
many 12,500, respectively) having a MRCA that
existed ~16,000 to 24,000 years ago (95% HPD:
13,500 to 28,100). Last, dog clade B, which con-
tains 22% (17 of 77) of dog sequences has the
closest phylogenetic associations with sequences
from modern wolves from Sweden and the Ukraine
and shares a MRCA with them some ~9200 years
ago (95% HPD: 6500 to 12,300).
The association of sequences from modern dogs
in clades A, C, and D with ancient European canid
specimens and of modern dogs from clade B with
European wolves suggests an origin of dogs in
Europe, rather than the Middle East or East Asia, as
previously suggested (5–7). Critically, none of the
modern wolf sequences from other putative centers
of origins such as the Middle East (Saudi Arabia,
Oman, Israel, Iran, and India) or East Asia (China,
Japan, and Mongolia) show close affinity with mod-
ern dog clades. Bayesian analysis of divergence
times implies a European origin of the domestic
dog dating to as much as 18,800 to 32,100 years
ago, given an upper limit of the MRCA of an an-
cient wolf sequence and dogs clustered in clade A
and the MRCA of the most diverse dog clade as a
lower limit (Fig. 1). Consequently, our results sup-
port the hypothesis that dog domestication preceded
the emergence of agriculture (20) and occurred in
the context of European hunter-gatherer cultures.
Previous research suggested that modern dogs
experienced a two-phase bottleneck. The first was
at the origin of the domestication process, and the
second was more recent during breed formation
over the past several hundred years (21). To inves-
tigate the demographic history of dogs, we used a
Bayesian Skygrid analysis (22) applied to dog clade
A and the closely related pre-Columbian dogs. We
find a continuous population size increase from the
1
Department of Biology, Section of Genetics and Physiology,
University of Turku, Itäinen Pitkäkatu 4, 20014 Turku, Finland.
2
Department of Ecology and Evolutionary Biology, University of
California Santa Cruz, 1156 High Street, Santa Cruz, CA 9506 4,
USA.
3
Max Planck Institute for Evolutionary Anthropology,
Deutscher Platz 6, 04103 Leipzig, Germany.
4
Institute for Ar-
chaeological Sciences, University of Tübingen, Rümelinstrasse
23, Tübingen, Germany.
5
Department of Ecology and Evolu-
tionary Biology, University of California Los Angeles, 2149
Terasaki Life Science Building, Los Angeles, CA 90095, USA.
6
Operational Direction “Earth and History of Life,”Royal Belgian
Institute of Natural Sciences, Vautierstraat 29, 1000 Brussels,
Belgium.
7
Zoological Institute, Russian Academy of Sciences,
Universitetskaya nab. 1, 199034 Saint Petersburg, Russia.
8
Yale
Center for Genome Analysis, Yale University, West Haven, CT
06516, USA.
9
Genètica de la Conservació, Institut de Recerca i
Tecnologia Agroalimentàries (IRTA), Carretera de Cabrils km 2,
08348, Cabrils, Barcelona, Spain.
10
Institute of Palaeoanatomy
and History of Veterinary Medicine, Ludwig-Maximilian-University
Munich, and ArchaeoBioCenter LMU, Kaulbachstrasse 37, 80539
Munich, Germany.
11
Instituto Nacional de Antropología y Pensamiento
Latinoamericano, Consejo Nacional de Investigaciones Científicas
y Técnicas (CONICET), 3 de Febrero 1378, CJN1429 Buenos Aires,
Argentina.
12
Landschaftsverband Rheinland (LVR)–Landesmuseum
Bonn, Bachstrasse 5-9, D-53115 Bonn, Germany.
13
Department of
Prehistoric and Protohistoric Archaeology, Institute for Archaeology
and Cultural Anthropology, University of Bonn, Regina-Pacis-Weg 7,
53113 Bonn, Germany.
14
Southeastern Archaeological Research,
Inc., 315 Northwest 138th Terrace, Newberry, FL 32669, USA.
15
Center for Bioarchaeological Research, School of Human Evolution
and Social Change, Arizona State University, Tempe, AZ 85287–
2402, USA.
16
Department of Genomic Diversity and Evolution,
Institute of Molecular and Cellular Biology, Siberian Branch of the
Russian Academy of Sciences, Novosibirsk, Russia.
17
Institute of
Archaeology and Ethnography, Siberian Branch of the Russian
Academy of Sciences, Novosibirsk, Russia.
18
Theodosius Dobzhansky
Center for Genome Bioinformatics, Saint Petersburg State Univer-
sity, 41A Sredniy Prospekt, Saint Petersburg 199034, Russia.
19
Estación Biológica de Doñana, Conservation and Evolutionary
Genetics Group (EBD-CSIC), Avenida Américo Vespucio s/n,
41093 Seville, Spain.
20
Department of Biomolecular Engineer-
ing, University of California Santa Cruz, 1156 High Street, Santa
Cruz, CA 95064, USA.
*Corresponding author. E-mail: olatha@utu.fi (O.T.); rwayne@
eeb.ucla.edu (R.K.W.)
†Deceased.
15 NOVEMBER 2013 VOL 342 SCIENCE www.sciencemag.org872
REPORTS

time of the MRCA to about 5000 years ago, which
may be attributable to the earliest domestication
phase (Fig. 2). A more recent decline occurred be-
tween 5000 and 2500 years ago and was followed
by a sharp increase in population size (Fig. 2). This
increase parallels the trajectory of human population
size (23), which suggests demographic dependence
of dogs on human populations. In contrast, wolf
numbers declined during this period, consistent
with the emergence of agrarian cultures and the loss
of vital wolf habitat and wild game.
Our findings support the conclusion that
the mitochondrial legacy of dogs derives from
wolves of European origin. Past mitochondrial
and Y chromosome analyses that suggested a
non-European location for the onset of domesti-
cation were more limited in sampling of modern or
ancient wolves or prehistoric dogs and had weak
statistical support for phylogenetic branching
points (4,6,24). The modern dog clades A to D
are well-supported in our tree of complete mtDNA
sequence. We find that the sequence diversity
that exists today in dogs can all be found in
ancient (clades A, C, and D) or modern (clade
B) European canids. The inferred recent divergence
of clade B from wolves now found in Sweden
and the Ukraine implies that it might represent a
mitochondrial genome introgressed from wolves
rather than one established by domestication, be-
cause dogs were clearly domesticated by this
time (8,12,14).
Notably, our ancient panel does not contain
specimens from the Middle East or China, two
proposed centers of origin (5,6). In fact, no ancient
dog remains older than ~13,000 years are known
from these regions (10). However, ancient wolf
and dog remains from these areas would need to
be rooted more closely to the four dominant dog
clades than any ancient or modern European
canids to contradict our primary conclusions. We
consider this scenario unlikely as it would require
a common recent coalescence of these ancestral
wolf and dog sequences from geographically dis-
parate areas. Nevertheless, a more complete and
nuanced picture of dog domestication will likely
emerge with the addition of ancient canine mtDNA
data from the Middle East and Asia. A further
caveat to our conclusions is that although the
mtDNA sequence tree is well supported, it repre-
sents a single genetic locus. The rapid coalescence
of mtDNA genomes and the lack of recombi-
nation are important advantages; however, both
mitochondrial and nuclear genomes suffer from
incomplete lineage sorting, which, given the re-
cent divergence of dogs and wolves, can potentially
confound evolutionary inference. The availabil-
ity of multiple independent loci in the nuclear
genome potentially offers more power to resolve
phylogenetic relations. We attempted to capture
multiple nuclear loci using a densely tiled cap-
ture array, but were not able to obtain sufficient
coverage to call genotypes confidently in any of
the ancient specimens, which reflects their poor
state of DNA preservation (15). Nonetheless, our
mtDNA genome tree shows that three of four
India
Italy
Dog C (9)
Alaska 20,800
Israel 1
Belgium 36,000
Ukraine
Oman
USA 8,500
Poland 1
USA 1,000
Dog B (17)
Russia 2
Saudi Arabia 2
Poland 2
Russia 22,000
Belgium 30,000
Argentina 1,000
Iran
Sweden 1
Ger m an y 12,500
Belgium 26,000
Spain
Russia 15,000
Israel 2
China 4
Switzerland 2 14,500
North America (10)
Russia 1
Russia 3
Finland
Mexico 2
Alaska 28,000
Russia 18,000
Switzerland 1 14,500
Japan
Alaska 21,000
Dog A (49)
Mongolia
China 3
Saudi Arabia 1
Mexico 1
Sweden 2
Sweden 3
Switzerland 3 14,500
Russia 33,500
Dog D (2)
Ger m an y 14,700
Cr o at i a
North America (11)
10,00020,00030,00040,00050,00060,00070,00080,000 0
Years before
p
resent
Fig. 1. Phylogenetic arrangement of modern and ancient dog (blue) and wolf sequences
(orange) as obtained from coalescence-based, maximum likelihood, and Bayesian methods. The
outgroup (four coyotes) and two Chinese wolf sequences were excluded [see (15) for more details]. Ancient
specimens are labeled with the respective country of origin and their approximate reported age (italicized; in
years before present). Fossil specimens with ambiguous taxonomic classification are indicated by a gray
color. Whenever modern canid sequences form a monophyletic cluster, the number of sequences in the
cluster is indicated in brackets. Asterisks highlight statistical support whenever both bootstrap values are
>90% and posterior support is >0.9 for the maximum likelihood and Bayesian analyses, respectively.
www.sciencemag.org SCIENCE VOL 342 15 NOVEMBER 2013 873
REPORTS

modern dog clades are more closely related to
sequences from ancient European rather than
extant wolves. Further, analysis of coalescence
times support a dog-wolf divergence time of
>15,000 years ago. An evolutionary scenario con-
sistent with these results is that dog domestication
was initiated close to the Last Glacial Maximum
when hunter-gathers preyed on megafauna (25).
Conceivably, proto-dogs might have taken advan-
tage of carcasses left on site by early hunters, as-
sisted in the capture of prey, or provided defense
from large competing predators at kills. Finally, our
results imply that some of the earliest putative dog
remains, such as the Goyet dog from Belgium (2)
or Altai Mountain specimen from Russia (13),
may represent aborted domestication episodes.
If true, this suggests that the conditions for dog
domestication were not unique to one place or time
and adds a role for serendipity to the process that
led to the early and singular domestication of a large
and dangerous carnivore.
References and Notes
1. G. Larson, J. Burger, Trends Genet. 29, 197–205 (2013).
2. M. Germonpré et al., J. Archaeol. Sci. 36, 473–490 (2009).
3. S. J. Olsen, J. W. Olsen, Science 197, 533–535 (1977).
4. C. Vilà et al., Science 276, 1687–1689 (1997).
5. B. M. vonHoldt et al., Nature 464, 898–902 (2010).
6. P. Savolainen, Y. P. Zhang, J. Luo, J. Lundeberg,
T. Leitner, Science 298, 1610–1613 (2002).
7. G. D. Wang et al., Nat Commun 4, 1860 (2013).
8. M. Sablin, G. Khlopachev, Curr. Anthropol. 43, 795–799
(2002).
9. S. J. Crockford, Y. V. Kuzmin, J. Archaeol. Sci. 39,
2797–2801 (2012).
10. G. Larson et al., Proc. Natl. Acad. Sci. U.S.A. 109,
8878–8883 (2012).
11. H. Napierala, H.-P. Uerpmann, Int. J. Osteoarchaeol. 22,
127–137 (2012).
12. G. Nobis, Umschau 79, 610 (1979).
13. N. D. Ovodov et al., PLOS ONE 6, e22821 (2011).
14. M. Baales, Mongraphien RGZM, Mainz 38, 106 (1996).
15. Supplementary materials are available on Science Online.
16. J. A. Leonard et al., Curr. Biol. 17, 1146–1150 (2007).
17. J. A. Leonard et al., Science 298, 1613–1616 (2002).
18. B. van Asch et al., Proc. Biol. Soc. 280, 20131142 (2013).
19. H. Malmström et al., BMC Evol. Biol. 8, 71 (2008).
20. E. Axelsson et al., Nature 495, 360–364 (2013).
21. K. Lindblad-Toh et al., Nature 438, 803–819 (2005).
22. M. S. Gill et al., Mol. Biol. Evol. 30, 713–724 (2013).
23. J. A. Tennessen et al., Broad GO, Seattle GO, on behalf
of the NHLBI Exome Sequencing Project, Science 337,
64–69 (2012).
24. Z. L. Ding et al., Heredity 108, 507–514 (2012).
25. J. Alroy, Science 292, 1893–1896 (2001).
Acknowledgments: Mitochondrial sequences have been
deposited at the NCBI database with the accession numbers
(KF661036 to KF661096), and a complete alignment is
available as a supplementary file. We would like to thank all
colleagues who provided samples for this study, the Illinois
State Museum and the Center for American Archeology for
allowing us to sample the material from the Koster site,
and the American Museum of Natural History, New York.
O.T. is grateful to D. Wegmann and D. Schwochow-Thalmann
for helpful discussions and comments on the manuscript;
M. Bruneaux for help with R; and A. v. Haeseler for helpful
advice with TREE-PUZZLE and IQ-TREE. We thank D. Ward,
M. Riesenberg, J. Beier, I. Bergmann, H. Mutlu, and S. Bealek
for assistance with ancient DNA extractions. R.E.G. is president
of Dovetail Genomics. Financial support for this study was
provided from the E. Aaltonen foundation and the Turun
Yliopistosäätiö to O.T.; Molecular and Cellular Biology, Siberian
Branch of the Russian Academy of Sciences, and Russian
Foundation for Basic Research grants to A.S.G.; and NSF
support to R.K.W. and B.S. (OPP 9617068, EF-1021387). J.K.
and V.J.S. were supported by the Carl Zeiss Foundation. This
work was further supported by the Max Planck Society.
O.T. is financed by a Marie Curie Intra European Fellowship
within the 7th European Community Framework Program. The
authors declare no conflict of interest.
Supplementary Materials
www.sciencemag.org/content/342/6160/871/suppl/DC1
Materials and Methods
Supplementary Text
Figs. S1 to S9
Tables S1 to S5
References (26–78)
23 July 2013; accepted 3 October 2013
10.1126/science.1243650
Table 1. Ancient specimens used and summary of sequencing statistics.
(A) Ancient specimens captured using custom designed capture arrays. (B)
Specimens enriched for mtDNA using long range PCR-products and custom
designed biotinylated adapters (15). Morphological classification and approx-
imate age are from the respective references (see table S1). Ancient specimens
with ambiguous morphological classification are shown in italic font. Nu-
cleotides were retained with a minimum of two reads per base. Further
information on filtering parameters is available (15).
Identification Origin
Morphological
classification
Approximate
age (years B.P.)
Average
mt-genome coverage
Retained
nucleotides
A
Belgium 26,000 Belgium, Trou des Nutons Wolflike 26,000 8.3 16,170
Belgium 36,000 Belgium, Goyet niveau 4 Doglike 36,000 4.1 12,020
Belgium 30,000 Belgium, Goyet niveau 4 Wolflike 30,000 20.4 16,348
Russia 18,000 Russia, Medvezya cave Wolflike 18,000 137.7 16,414
Russia 15,000 Russia, Eliseevichi Doglike 15,000 6.0 14,340
USA 8500 USA; Koster site, Illinois Doglike 8500 7.9 16,154
Argentina 1000 Argentina, Cerro Lutz Doglike 1000 27.8 16,369
Russia 22,000 Russia, Kostenki Wolflike 22,000 21.5 16,397
USA 1000 USA, Florida Doglike 1000 53.7 16,414
B
Switzerland 1 14,500 Switzerland, Kesslerloch cave Wolflike 14,500 14.7 16,357
Alaska 28,000 Alaska, Eastern Beringia Wolflike 28,000 90.1 16,415
Alaska 21,000 Alaska, Eastern Beringia Wolflike 21,000 2.1 9073
Alaska 20,800 Alaska, Eastern Beringia Wolflike 20,800 625.7 16,412
Switzerland 2 14,500 Switzerland, Kesslerloch cave Wolflike 14,500 4.2 13,965
Russia 33,500 Russia, Razboinichya cave Doglike 33,500 100.8 16,411
Germany 14,700 Germany, Bonn-Oberkassel Doglike 14,700 1.9 8667
Germany 12,500 Germany, Kartstein cave Doglike 12,500 8.6 16,239
Switzerland 3 14,500 Switzerland, Kesslerloch cave Wolflike 14,500 9.2 16,089
0 5,000 10,000 15,000 20,000
234567
Years Before Present
log Effective Population Size
Fig. 2. Bayesian Skygrid plot depicting the demographic trajectory of dog clade A and closely
related pre-Columbian dogs. Times are given in years before present and the effective population
size is indicated in median logN
e
(solid line) with the accompanying 95% HPD interval.
15 NOVEMBER 2013 VOL 342 SCIENCE www.sciencemag.org
874
REPORTS