Ancient DNA provides new insights into the origin of the Chinese domestic horse

Ancient DNA provides new insights into the origin of the Chinese
domestic horse

Dawei Cai

, Zhuowei Tang

, Lu Han

, Camilla F. Speller

, Dongya Y. Yang

, Xiaolin Ma

, Jian’en Cao

Hong Zhu

, Hui Zhou

Ancient DNA Laboratory, Research Center for Chinese Frontier Archaeology of Jilin University, Changchun 130012, China

College of Life Science, Jilin University, Jiefang Road 115, Changchun 130023, China

Ancient DNA Laboratory, Department of Archaeology, Simon Fraser University, 8888 University Drive, Burnaby BC, Canada V5A 1S6

Henan Provincial Institute of Cultural Relics and Archaeology Zhengzhou 450000, China

Inner Mongolian Institute of Cultural Relics and Archaeology Hohhot 010010, China

a r t i c l e

i n f o

Article history:

Received 21 August 2008
Accepted 11 November 2008

Keywords:

Ancient DNA
Domestic horse
Mitochondrial DNA
Haplogroup
Przewalski horse

a b s t r a c t

Domestic horses played a pivotal role in ancient China, but their exact origin remains controversial. To
investigate the origin of Chinese domestic horses, we analyzed mitochondrial DNA (mtDNA) from 35
horse remains, aged between 4000 and 2000 years, excavated from nine archaeological sites in northern
China. The Chinese ancient horses exhibited high matrilineal diversity, falling into all the seven hap-
logroups (A–G) observed in modern horses. These results suggest that several maternal lines were
introduced into the gene pool of Chinese horses in the past. Haplogroups A and F were more prevalent in
ancient horses than the other haplogroups. Interestingly, only haplogroups A and F were present in the
samples older than 4000 years, while the more recent horses (between 2000 and 3000 years BP) fell into
all seven haplogroups. Comparison with DNA data of present-day horses suggests that haplogroup F is
like to be an ancient haplogroup of East Asian origin. These analyses also suggest that the origin of
Chinese domestic horses is complex, and external mtDNA input occurred after initial domestication. Our
results indicate that the Chinese ancient horses are more related to the modern Mongolian horses. Lastly,
our results cannot support the previous hypothesis that early Chinese domestic horses were derived
from the Przewalski horse.

1. Introduction

Humans have had a profound inﬂuence on the horse since its

domestication in the late Neolithic. Used for transport, labour, food
and recreation, horses became important in many facets of our
society (

Mills and McDonnell, 2005

). Archaeological evidence

suggests that the horse was probably ﬁrst domesticated about 6000
years ago on the grassland steppes of Eurasia from Ukraine to
Turkestan, and the earliest domesticated horses subsequently
spread from this area (

Anthony, 1996; Bennett and Hoffmann, 1999;

Clutton-Brock, 1999

). However, mitochondrial DNA (mtDNA)

analysis of modern and ancient horses has revealed extensive
variation within and among breeds, with little congruence of
haplogroup distribution to breeds or geographic areas, suggesting
that separate and geographically diverse populations participated
in domestication process, and that domestic horses have multiple
origins (

Jansen et al., 2002; Lister et al., 1998; Vila et al., 2001

Domestic horses played a pivotal role in ancient China, being

used mainly in agriculture and transport, especially in military
purposes (

Chen, 1994

). However, the geographic origin of Chinese

domestic horses remains controversial. Zooarchaeological data
show that abundant domestic horse remains appeared suddenly in
China at the sites of the Shang Dynasty (ca. 3000 BP). Prior to the
Shang Dynasty, there are few records of domestic horses. Excava-
tion of thousands of Neolithic and early Bronze sites in China
yielded only a few sporadic fragments of horse tooth and bone at
a limited number of sites, and it is difﬁcult to determine whether
the remains are those of wild or domestic horses (

Chen, 1999

). The

absence of evidence for the early stages of domestication, coupled
with the sudden appearance of horses around 3000 BP, have led
scholars to suggest two hypotheses. First, that domestic horses
were imported into China from the Eurasian steppe via that Gansu
and Qinhai provinces (

Yuan and An, 1997

), or, second, that horses

underwent autochthonous domestication in China during the Later
Neolithic (ca. 4000 BP) or perhaps even earlier, and Przewalski
horses were proposed as the ancestors of Chinese domestic horses
(

Wang and Song, 2001; Zhang, 2004

). In order to discriminate

between these two hypotheses, we used ancient DNA techniques to

Corresponding author. College of Life Science, Jilin University, Jiefang Road 115,

Changchun 130023, China. Tel./fax:

86 431 8849 8031.

E-mail address:

zhouhui@mail.jlu.edu.cn

(H. Zhou).

Contents lists available at

ScienceDirect

Journal of Archaeological Science

j o u r n a l h o m e p a g e : h t t p : / / w w w . e l s e v i e r . c o m / l o c a t e / j a s

0305-4403/$ – see front matter

doi:10.1016/j.jas.2008.11.006

Journal of Archaeological Science 36 (2009) 835–842

investigate the genetic diversity and phylogenetic relationships of
Chinese ancient horses, as this approach has proved useful for the
study of the origins and spread of other livestock (

Cai et al.,

2007a,b; Leonard et al., 2002; McGahern et al., 2006; Troy et al.,
2001

We analyzed mitochondrial DNA (mtDNA) sequences from 26

ancient horses excavated from different archaeological sites in
Northern China, along with nine previously reported sequences
from Mongolia (

Cai et al., 2007a

). The sequences were compared

with previously published ancient, modern and wild horses stored
in the GenBank.

2. Materials and methods

2.1. Archaeological information

The ancient horse samples came from seven archaeological sites

in Northern China (

Fig. 1

): three from the Inner Mongolia Auton-

omous Region (two sites – Bancheng and Xiaoshuanggucheng – in
Liangcheng County, and Xindianzi in Helingeer County); two from
Henan Province (Maoyuan in Xinzheng City, Yin Ruins in Anyang
City); Qianzhangda site in Tengzhou City, Shandong Province;
Yujiazhuang site in Pengpu Village, Guyuan County, Ningxia Hui
Autonomous Region. The ages of these sites vary from ca. 3000 BP
(Shang Dynasty) to ca. 2500 BP (from the end of Spring–Autumn to
the start of the Warring States period) (

Table 1

2.2. Archaeological horse remains

Thirty-seven horse teeth and bones were retrieved from seven

archaeological sites [Inner Mongolia: Bancheng (

7), Xiaosh-

uanggucheng (

4), Xindianzi (

13); Henan Province: Maoyuan

(

8), Yin Ruins (

2); Qianzhangda (

1) and Yujiazhuang

(

2)] (

Table 1

). After morphological studies, the samples were

stored in sealed plastic bags placed in boxes in a store room. Prior to
their arrival at the ancient DNA laboratory, the remains had been
handled by less than ﬁve archaeozoologists. The samples were of
different ages, and exhibited different macroscopic preservation.
Most of the tooth samples from Inner Mongolia were well

preserved, with smooth and compact external surfaces. In contrast,
most of the horse bones from Henan Province were light-weight
and sandy coloured, with damaged external surfaces.

To increase the sample size, we included nine previously

reported sequences, (

DQ900922–DQ900930

) from nine horse

remains excavated from Dashaqian (

5) and Jinggouzi (

sites, Chifeng region, Inner Mongolia, China, dating from the Lower
Xiajiadian Culture (ca. 4000 BP) to the late Warring States period
(ca. 2000 BP) (

Cai et al., 2007a

2.3. DNA extraction, ampliﬁcation and sequencing at
Jilin University

Before DNA extraction, bone and tooth samples were prepared

using the procedure described by

Cai et al. (2007a)

. DNA was

extracted following a silica-spin column method (

Yang et al., 1998

Several DNA extractions were performed for each specimen as
shown in

Table 1

Fig. 1.

Geographical location of archaeological sites.

Bancheng and Xiaoshuang-

gucheng;

Xindianzi;

Dashanqian;

Jinggouzi;

Yin Ruins;

Maoyuan;

Qianzhangda; and

Yujiazhuang.

Table 1

Archaeological horse samples studied, with associated codes, skeletal element used,
dates and PCR results.

Lab
code

Archaeological
code

Skeletal
element

Date

No. of
extractions

Result

Bancheng site

LB01

03LBM4-1

Tooth

ca. 2500 BP

Reproducible

LB02

03LBM20-1

Tooth

ca. 2500 BP

Reproducible

LB03

03LBM26-1

Tooth

ca. 2500 BP

Reproducible

LB04

03LBM28-1

Tooth

ca. 2500 BP

Reproducible

LB05

03LBM31-1

Tooth

ca. 2500 BP

Reproducible

LB06

03LBM40-3

Tooth

ca. 2500 BP

Reproducible

LB07

03LBM67-2

Tooth

ca. 2500 BP

Reproducible

Xiaoshuanggucheng site

LS01

03LSM4-2

Tooth

ca. 2500 BP

Reproducible

LS02

03LSM7-4

Tooth

ca. 2500 BP

Reproducible

LS03

03LSM9-10

Tooth

ca. 2500 BP

Reproducible

LS04

03LSM11-1

Tooth

ca. 2500 BP

Reproducible

Xindianzi site

HX01

99HXM2-2

Tooth

ca. 2500 BP

Reproducible

HX02

99HXM2-3

Tooth

ca. 2500 BP

Reproducible

HX03

99HXM2-7

Tooth

ca. 2500 BP

Reproducible

HX04

99HXM5-2

Tooth

ca. 2500 BP

Reproducible

HX05

99HXM7-1

Tooth

ca. 2500 BP

Reproducible

HX06

99HXM20-1

Tooth

ca. 2500 BP

Reproducible

HX07

99HXM34-1

Tooth

ca. 2500 BP

Reproducible

HX08

99HXM37-7

Tooth

ca. 2500 BP

Reproducible

HX09

99HXM38-2

Tooth

ca. 2500 BP

Reproducible

HX10

99HXM43-2

Tooth

ca. 2500 BP

Reproducible

HX11

99HXM43-9

Tooth

ca. 2500 BP

PCR fail

HX12

99HXM43-10

Tooth

ca. 2500 BP

Reproducible

HX13

99HXM43-16

Tooth

ca. 2500 BP

Reproducible

Qianzhangda site

QZ01

Tooth

ca. 3000 BP

PCR fail

Yujiazhuang site

YJZ01

87YJZ11

Tooth

ca. 2500 BP

Reproducible

YJZ02 87YJZ20

Tooth

ca. 2500 BP

Reproducible

Maoyuan site

MY01 03ZHMY21

3rd phalange

ca. 2500 BP

PCR fail

MY02 03ZHMY22

3rd phalange

ca. 2500 BP

PCR fail

MY03 03ZHMY23

3rd phalange

ca. 2500 BP

PCR fail

MY04 03ZHMY24

3rd phalange

ca. 2500 BP

PCR fail

MY05 03ZHMY25

3rd phalange

ca. 2500 BP

Reproducible

MY06 03ZHMY26

3rd phalange

ca. 2500 BP

PCR fail

MY07 03ZHMY27

3rd phalange

ca. 2500 BP

PCR fail

MY08 03ZHMY28

3rd phalange

ca. 2500 BP

PCR fail

Yin Ruins

YX01

YXH433

2nd phalange

ca. 3000 BP

PCR fail

YX02

YXHF26

2nd phalange

ca. 3000 BP

Contradiction

D. Cai et al. / Journal of Archaeological Science 36 (2009) 835–842

836

A 300 bp fragment of the mtDNA D-loop region (nucleotides

15473–15772 based on reference sequences X79547) was ampliﬁed
by using two sets of overlapping primers described by

Cai et al.

(2007a)

: forward primer L15473 5

–CTTCCCCTAAACGACAACAA–3

and reverse primer H15692 5

–TTTGACTTGGATGGGGTATG–3

; and

forward primer L15571 5

–AATGGCCTATGTACGTCGTG–3

and

reverse

primer

H15772

–GGGAGGGTTGCTGATTTC–3

PCR

ampliﬁcations were performed in an Mastercycler

personal

Thermal Cycler (Eppendorf, Hamburg, Germany) in a 50

L reaction

volume containing 2.5 mM Mg

, 1

Buffer, 0.2 mM dNTPs, 1.6 mg/

ml BSA, 0.5

M of each primer, 1 U Taq polymerase (Promega, USA)

and 2

L DNA template. The PCR condition were as follows: pre-

denaturation at 95

C for 5 min, 8 cycles of 94

C for 1 min, 55

C for

1 min and 72

C for 1 min; followed by 28 cycles of 92

C for 1 min,

50–55

C for 1 min, 72

C for 1 min; and a ﬁnal elongation step of

C for 10 min.

The PCR products were isolated from 2% agarose gels and

puriﬁed by using QIAEX

GEL Extraction Kit (QIAGen, Germany).

Both DNA strands of PCR products were directly sequenced with
original primers on an ABI 310 automated DNA sequencer (Applied
Biosystems, USA). The obtained electropherograms were assem-
bled to examine any base pair ambiguities using Chromas 2.22
(

www.technelysium.com.au

2.4. DNA extraction, ampliﬁcation and sequencing at
Simon Fraser University

Three bone samples (LB04, LS02 and HX07) were extracted

using a similar DNA extraction method (

Yang et al., 1998

) in the

dedicated ancient DNA laboratory at Simon Fraser University. PCR
ampliﬁcations were conducted in a Mastercycler Personal in
a 30

L reaction volume containing 50 mM KCl, 10 mM Tris–HCl,

2.5 mM MgCl

, 0.2 mM dNTP, 1.0 mg/mL BSA, 0.3

M each primer,

3.0

L DNA sample and 1.5–3.0 U AmpliTaq Gold

(Applied

Biosystems).

The above sets of primers (the Jilin lab) were used for PCR

ampliﬁcations. PCR was run for 60 cycles at 94

C for 30 s

(denaturing), 52–55

C for 30 s (annealing), and 72

C extension

for 40 s, with an initial 12 min denaturing period at 95

C. Five

microliters of PCR product were visualized via electrophoresis
on a 2% agarose gel using SYBR Green

staining. PCR products

were puriﬁed using QIAGEN’s MinElute

puriﬁcation kits and

were subjected to direct sequencing through Macrogen (

www.

macrogen.com

). PCR products were sequenced using both

reverse and forward primers, and electropherograms for the same
DNA sample were assembled and cross-examined using Chromas
Pro (

www.technelysium.com.au

2.5. Comparison sequences

Ancient horse sequences were truncated to 247 bp (15494–

15740) for phylogenetic analysis, and compared with 1053 modern
horse mtDNA sequences from GenBank, including 17 populations
from ﬁve broad geographical regions (

Table S1

): East Asia: Tuva

(

11), Cheju (

34), Mongolian (

33), Guan Mountain (

10),

Tibetan (

16); North Asia: Vyatskaya (

18), yakut (

20);

Central Asia: Akhal Teke (

24), Orlov (

18); Middle East and

Africa: Turkey (

27), Arabia (

87), Africa (

20); Europe:

Iberia (

251), British Isles (

223), Mainland Europe (

218),

Northern Europe [Mesenskaya (

18), Scandinavia (

25)].

Finally, we added the wild Przewalski horse (AF014409,

AF072994–

AF072995

AJ413830–AJ413832

and AF326635) and ancient horse

sequences

from

Ireland

(DQ327848,

DQ327850–DQ327851

Kazakhstan (

AJ876883–AJ876890

), Russia (

AJ876891–AJ876892

Korea (AF049720) and northern Europe (

AF326676–AF326679

), and

late Pleistocene sequences from Alaska (

AF326668–AF326675

2.6. Phylogenetic analysis

Multiple alignments were performed using the Clustal X 1.83

program (available at

ftp://ftp-igbmc.u-strasbg.fr/pub/ClustalX/

)

(

Thompson et al., 1997

). The median-joining (MJ) network (

Bandelt

et al., 1999

) was drawn using the program NETWORK4.1.1.2

(available at

www.ﬂuxus-engineering.com/sharenet.htm

). Haplo-

type diversity, nucleotide diversity, mismatch analysis, Fu’s

values and Linearised

distances were computed using ARLE-

QUIN 2.0 software (available at

http://cmpg.unibe.ch/software/

arlequin/

) (

Schneider and Excofﬁer, 2000

). The statistical signiﬁ-

cance of the values was estimated by permutation analysis using
1000 replications.

3. Results

3.1. Sequence variation and haplogroups assignment

Ancient DNA sequences were recovered from 26 of the 37

samples. Ten samples (HX11 from Xindianzi site, MY01-04 and
MY06-08 from Maoyuan site, QZ01 from Qianzhangda site and YX01
from Yin Ruins) did not yield PCR products. In addition, one sample
YX02 from Yin Ruins was excluded because sequences obtained from
several extractions were different. The samples from Inner Mongolia
showed the higher rate of success than those from other regions,
especially Henan, which can be attributed to the cold and dry envi-
ronment of Inner Mongolia which is more suited to DNA survival.
Ancient horse sequences data in this paper have been submitted to
GenBank with accession numbers:

EU931584–EU931609

The 35 ancient horses (including the nine sequences reported in

Cai et al., 2007a

and 26 sequences reported here) yielded 35 vari-

able positions when compared with the reference sequence
(GenBank X79547). All the substitutions were transitions, and
deﬁned 25 different haplotypes (H1–H25,

Table 2

). Eighteen

haplotypes were singletons, while seven haplotypes were detected
more than once, of which three haplotypes (H6, H8 and H13) were
shared by samples from the same archaeological site, and four
haplotypes (H3, H5, H9 and H19) between different sites. The 25
haplotypes were assigned to all of the seven haplogroups A–G in
modern horses, identiﬁed by

Jansen et al. (2002)

and in the skel-

eton network described by

McGahern et al. (2006)

(

Table 1

3.2. Haplogroup frequency

Haplogroups A and F occurred most frequently in our study

(42.8% and 31.4%, respectively). Haplogroups B, C, D, E and G were
found at lower frequencies (2.9%, 5.7%, 2.9%, 8.6% and 5.7%). We
compared our results with the frequency of the seven haplogroups
in 17 modern horse populations and Przewalski horses (

Table 3

and

Fig. 2

). Haplogroup A is the most common haplogroup in most

modern horse populations, with the highest frequency (66.6%) in
Orlov to the lowest (22%) in Mesenskaya. Interestingly, all analyzed
Przewalski horses belong to haplogroup A. Haplogroup F has the
highest frequency in Mongolian (36.4%) and Cheju horses (32.4%),
and the lowest frequency in Iberian (2.8%) and British Isles horses
(3.2%).

3.3. Genetic diversity

Within-species genetic diversity is thought to reﬂect population

size, history, ecology, and the ability to adapt, and can be assessed
by calculating the haplotype diversity and nucleotide diversity
within a population. In this study, the haplotype diversity and
nucleotide diversity of Chinese ancient horses were 0.978

0.012

and 0.025

0.013, respectively. Interestingly, the genetic diversity

was very similar in the modern horses (

Table 3

). The Przewalskis

D. Cai et al. / Journal of Archaeological Science 36 (2009) 835–842

837

had the lowest values (0.476

0.171 and 0.0019

0.0017, respec-

tively) reﬂecting the severe genetic bottleneck suffered by these
animals.

3.4. Phylogenetic network construction

In order to investigate the phylogenetic relationship of the

haplotypes, we constructed a median-joining (MJ) network with
1121 sequences of ancient, modern and wild horses (

Fig. 3

). Seven

major haplogroups and 17 clusters (A1–A7, B1–B2, C1–C2, D1–D3

and F1–F3) were evident in the star-like network. The 25 haplo-
types of our study fell into all of the seven haplogroups, and in
several of the clusters. Some haplotypes, for example H5 and H25,
shared the founder haplotype of a cluster with a variety of breeds
from different regions. The network showed that breeds of different
geographic regions overlapped, suggesting extensive gene ﬂow had
occurred among different breeds. Interestingly, some haplogroups
corresponded to particular breeds and regions. For example, hap-
logroup F clustered with breeds from East Asia and Middle East,
while D corresponded with many Iberian horses. Several ancient

Table 3

Diversity indices and frequency of mtDNA lineages in 19 horse populations.

Breed/population

Code

Number of
individuals

Number of
haplotypes

Haplotype
diversity (SE)

Nucleotide diversity (SE)

Lineage frequency (%)

Ancient horses

Anc

0.9782

0.0120

0.0246

0.0133

42.8

2.9

5.7

2.9

8.6

31.4

5.7

Mongolian

Mon

0.9564

0.0205

0.0293

0.0157

24.2

6.1

12.1

21.2

–

36.4

–

Guan Mountain

Gua

0.9556

0.0594

0.0258

0.0151

40.0

–

30.0

–

20.0

10.0

Tibetan

Tib

0.9833

0.0278

0.0273

0.0152

50.0

–

18.8

18.7

12.5

–

Korea Cheju

Kor

0.9554

0.0182

0.0250

0.0136

38.2

17.6

5.9

–

5.9

32.4

–

Tuva

Tuv

1.000

0.0388

0.0325

0.0185

45.4

9.1

–

18.2

–

18.2

9.1

Yakut

Yak

0.9421

0.0295

0.0244

0.0135

55.0

–

25.0

–

20.0

–

Vyatskaya

Vya

0.9281

0.0373

0.0211

0.0120

61.1

–

33.3

–

5.6

–

Mesenskaya

Mes

0.9150

0.0500

0.0225

0.0127

22.2

27.8

22.2

5.6

11.1

–

Orlov

Orl

0.8693

0.0610

0.0209

0.0119

66.6

16.7

11.1

5.6

–

Arabia

Ara

0.9767

0.0050

0.0259

0.0137

41.4

16.1

4.6

17.2

–

19.5

1.2

Akha-teke

Akh

0.9312

0.0326

0.0278

0.0151

29.2

4.2

12.5

29.2

8.2

12.5

4.2

Turkey

Tur

0.9687

0.0172

0.0232

0.0127

29.6

7.4

18.5

29.6

–

14.8

–

Africa

Afr

0.9789

0.0245

0.0300

0.0164

50.0

5.0

–

40.0

–

5.0

–

Iberia

Ibe

251

0.9419

0.0082

0.0274

0.0143

38.6

6.4

8.0

41.8

0.8

2.8

1.6

British Isles

Bri

223

0.9721

0.0035

0.0286

0.0149

23.8

8.5

20.6

34.5

9.4

3.2

–

Scandinavia

Sca

0.9233

0.0300

0.0241

0.0133

24.0

4.0

36.0

12.0

4.0

20.0

–

Maniland Europe

Eur

218

0.9712

0.0041

0.0297

0.0154

32.5

7.8

10.6

31.2

0.5

7.8

9.6

Przewalskii

Prz

0.4762

0.1713

0.0019

0.0017

100

–

Table 2

Variable nucleotides and lineage assignment of 35 ancient horses.

Haplotypes

Variable nucleotide positions

Lineages

Samples

X79547

DQ900927

DQ900928

DQ900929, HX10

DQ900930

DQ900924,
HX07, LB07

DQ900922,
DQ900923,
DQ900925

DQ900926

HX01, HX03

HX02, HX06, LB01

H10

HX04

H11

HX05

H12

HX08

H13

HX09, HX13

H14

HX12

H15

LB02

H16

LB03

H17

LB04

H18

LB05

H19

LB06, MY05

H20

LS01

H21

LS02

H22

LS03

H23

LS04

H24

YJZ01

H25

YJZ02

D. Cai et al. / Journal of Archaeological Science 36 (2009) 835–842

838

1995

). In this study, the genetic distances between 19 populations

were assessed by linearised

values (

Table 4

). The Chinese

ancient horses showed the lowest values, with horses from East
Asia (Tuva, Cheju and Mongolian horses), while the highest values
were found in the Przewalski (0.5822,

0.05). Generally, the

genetic distances between Chinese ancient horses and the 17
modern populations increased from East to West. These results
suggest that the Chinese ancient horses have an afﬁnity to East
Asian horses.

4. Discussion

4.1. Authenticity of ancient DNA sequences

A major concern in studying DNA recovered from archaeological

material is the authenticity of the results. One advantage of
working with animal rather than human bone is that species-
speciﬁc primers, which exclude the ampliﬁcation of human DNA,
can be used for PCR (

Barnes and Young, 2000

). Therefore, the most

likely forms of contaminations are cross-contamination between
ancient samples, or PCR product carry-over. In this study, we took
the following precautions to ensure that our results were genuine
and not the result of contamination: (1) all experiments were
performed in a dedicated ancient DNA laboratory in which modern
horse DNA had not been previously isolated or analyzed; (2) strict
contamination controls were exercised throughout all steps
according to commonly accepted recommendations (

Cooper and

Poinar, 2000

), and no contamination was detected in the negative

controls of DNA extraction or PCR ampliﬁcation; (3) sequences
were replicated for each sample from multiple independent
extractions (

Table 1

). Furthermore, several samples were validated

independently at a separate ancient DNA laboratory, located in the
Department of Archaeology at Simon Fraser University, Canada. (4)
To verify the accuracy of direct sequencing, PCR products of several
samples (HX03, LB07, LS03 and YJZ01) were cloned using the
PGemT Easy cloning kit (Promega) according to the handbook. Five
clones were picked for sequencing. The consensus sequence from
the clones was compared with that from direct sequencing, with
both types of sequences having the same sequence, indicating that
the sequences generated by direct sequencing were authentic.

4.2. The origins of Chinese horses

Modern horses exhibit high mtDNA diversity, indicating that the

maternal lineages have a widespread origin. In the present study,
we also observed high diversity in the Chinese ancient horses. We
found 25 haplotypes belonging to all seven haplogroups A–G found
in modern horses. Some of the ancient horses also shared the
founder haplotype of a cluster with a variety of breeds of different
geographical regions. This suggests that several different maternal
lines were introduced into the gene pool of the Chinese horses in
the past. To determine the origins of Chinese domestic horses, we
need to learn the time and place of the origin of the gene pool.
Haplogroups A and F are more abundant than the other hap-
logroups. Interestingly, haplogroups A and F were the only hap-
logroups detected in samples (

DQ90022–DQ90024

Cai et al.,

2007a

) aged more than 4000 years, while the samples aged

between 2000 and 3000 years had all haplogroups. These results
indicate a multiple origin scenario, where horses belonging to
haplogroups A and F were ﬁrst domesticated in China more than
4000 years ago, after which time the horses of the other hap-
logroups were introduced, consistent with a population expansion.

To shed light on the origin of the diverse haplogroups, we

studied the distribution of the seven haplogroups in 17 modern
populations, and showed that only F and D have a signiﬁcant
geographical pattern (

Table 3

and

Fig. 2

). Haplogroup F is prevalent

able

Linearised

diatance

(belo

diag

onal)

and

the

signiﬁcant

alues

betw

een

populations.

Anc

Mon

Gua

Tib

Mes

Akh

Afr

Ibe

Bri

Sca

Eur

Prz

Anc

þþ

þþþþ

Mon

0.0336

Gua

0.0756

0.0

Tib

0.04

0.0

0.02

0.0409

0.0634

0.0342

þþþþ

0.0

0.05

0.0

083

0.0

0.069

0.0

458

0.0387

0.02

0.0806

369

0.0

0.0387

Mes

0.054

0.0433

0.087

0.045

0.0

0.0982

365

þþþ

259

0.07

396

0.0908

0.04

0.0896

0.06

þþþþ

0.04

0.0

032

0.032

0.02

0.03

0.0

0.0224

0.0382

0.0

0.0402

þþþ

Akh

0.06

0.0

055

0.0

0.02

0.032

0.0463

0.0890

0.0

0.0459

0.0964

0.0

0.0332

0.0

0.0858

0.081

0.0294

0.0

Afr

0.0334

0.0

0.03

0.0

0.0222

0.0

0.0389

0.0

086

0.0

Ibe

0.04

0.0389

0.0525

545

0.0

0.039

0.024

0.04

0.0

0.0269

0.0

þþ

Bri

552

0.0522

0.0

0.06

308

0.0222

0.0509

0.0

0.0907

0.09

0.0636

0.0

0.024

0.0425

þþ

Sca

0.04

0.0

0.03

0.0456

0.0

0.04

0.07

0.0405

0.082

0.0296

0.0

093

0.02

0.0795

0.0886

0.04

þþ

Eur

0.083

0.0

0.0346

0.0904

0.0

0.024

0.0

0.0605

0.0838

0.0235

0.0

086

0.0

0.02

0.0306

Prz

0.5822

0.5639

0.65

0.6546

0.5

0.6955

0.8939

0.8645

0.8254

0.505

0.6607

0.8235

0.592

0.4940

0.6407

0.6875

0.5

D. Cai et al. / Journal of Archaeological Science 36 (2009) 835–842

840

in East Eurasian populations and its frequencies decline from east
to west, consistent with the results of previous studies (

McGahern

et al., 2006

). Based on our ﬁndings in Chinese horses older than

4000 years, we propose that haplogroup F is an ancient haplogroup
of East Asian origin. Haplogroup D is abundant in Western Eurasian
populations, especially Iberia (42%) and its frequencies decline from
west to east. Considering the abundant archaeological evidence
from the Iberia Peninsula and previous studies, we suggest that
haplogroup D originated in the Iberian Peninsula. We did not ﬁnd
a clear geographic speciﬁcity for haplogroups A, B, C, E and G,
although previous studies suggest that E and C tend to correspond
to ponies from the western and northern fringes of Europe
(

McGahern et al., 2006; Jansen et al., 2002

). Additional data are

required to reveal the geographical origins of these haplogroups.

In view of the proposed origin of haplogroups F and D, and the

presence of diverse maternal lineages, the population expansion of
Chinese ancient horses and the spread of the knowledge of horse
breeding about 1250 BC (

Levine, 1999

), we suggest that the origin of

Chinese domestic horses is more complex than previously thought,
and that both indigenous breeds and introduced maternal lineages
were involved in the process of domestication. Thus, our data fail to
support either of the two previous hypotheses on the domestica-
tion of horses in China.

Generally, the genetic diversity (haplotype diversity and

nucleotide diversity) in Chinese ancient horses was similar to that
observed modern horse populations (

Table 3

), suggesting that

Chinese ancient horses have an afﬁnity to modern horse pop-
ulations. Based on the present-day distribution of haplogroup F and
genetic distance analysis, the Chinese ancient horses appear to be
most closely related to the Mongolian horses and Cheju horses
(

Tables 3 and 4

). The Cheju horses are thought to be of Mongolian

origin as historical records indicate that the Mongols ﬁrst intro-
duced 160 horses to Cheju Island in 1276 (

Nam, 1969

). In conclu-

sion, Chinese ancient horses are closely related to the Mongolian
horses.

4.3. Ancient horses and Przewalski horses

Przewalski horses were once widespread in the Eurasian steppe.

As the Earth warmed following the last Ice Age, they lost ground
and were conﬁned to small areas in northern China and southern
Mongolia. Many Przewalski fossils have been found in the late
Pleistocene faunas and human settlements in northern China, from
Xinjiang in the west, to the Taiwan Strait (

Deng, 1999

). Due to their

widespread distribution, some scholars suggested that early
Chinese domestic horses were derived from the Przewalski horse
(

Deng, 2000; Zhang, 2004

Despite having a different chromosome number, Przewalski

horses (

66) can interbreed with domestic horses (

64) to

produce fertile offspring (

Ryder et al., 1978; Trommerhausen-Smith

et al., 1979

). However, the phylogenetic analysis of Przewalski and

domestic horses based on mtDNA and Y chromosome markers
revealed two different genetic patterns. Using mtDNA sequences,
Przewalski and domestic horses do not form monophyletic clades
in a phylogenetic tree (

Ishida et al., 1995; Oakenfull et al., 2000;

Oakenfull and Ryder, 1998

), while based on the Y chromosome,

there were two ﬁxed nucleotide differences between Przewalski
horse and domestic horse (

Wallner et al., 2003

). These studies

suggested that female mediated gene ﬂow occurred between
domestic horse and the Przewalski horse.

In our study, we detected no maternal connection between the

domestic horse and the Przewalski. Phylogenetic analysis revealed
that the Przewalski fell into cluster A2, distinct from Chinese
ancient horses and modern horses. Genetic distance analysis indi-
cated that the Przewalski’s were far from the ancient horses, sug-
gesting that the Przewalski horse and the domestic horse should be

considered sister taxa with a common ancestor, and that the
Przewalski horse is not the wild ancestor to the Chinese domestic
horse. However, this conclusion should be viewed with caution, as
the Przewalski horse underwent a severe genetic bottleneck: all
present individuals descend from only 13 survivors, with only four
maternal lineages (

Volf and Kus, 1991

). A large-scale analysis of the

ancient Przewalski horse DNA would permit the detection of
potential extinct maternal lineages.

5. Conclusion

In conclusion, our results reveal the genetic diversity and

phylogenetic structure of ancient Chinese domestic horses. This
research provides valuable new insights into the origin of Chinese
domestic horse: (1) haplogroup F is most likely to be an ancient
haplogroup of East Asian origin; (2) the origin of Chinese domestic
horses is complex, and external mtDNA input occurred during
domestication; and (3) Chinese ancient horses were more related
to the modern Mongolian horses. As a result, our results do not
support the previous hypothesis that early Chinese domestic horses
were derived from the Przewalski horse.

Acknowledgements

The work was supported by the National Science Fund for

Fostering Talents in Basic Research (grant no. J0530184). We are
grateful to Dr Erika Hagelberg for correcting grammar and helpful
comments on the manuscript. We also thank Henan Provincial and
Inner Mongolian Institute of Cultural Relics and Archaeology for
providing samples for analysis.

Supplementary information

Supplementary data associated with this article can be found, in

the online version, at

doi:10.1016/j.jas.2008.11.006

References

Anthony, D.W., 1996. In: Olsen, S.L. (Ed.), Horses Through Time. Roberts Rinehart for

Carnegie Museum of Natural History, Boulder, CO, pp. 57–82.

Bandelt, H.J., Forster, P., Rohl, A., 1999. Median joining networks for inferring

intraspeciﬁc phylogenies. Molecular Biology and Evolution 16, 37–48.

Barnes, I., Young, J.P.W., 2000. DNA-based identiﬁcation of goose species from two

archaeological sites in Lincolnshire. Journal of Archaeological Science 27, 91–100.

Bennett, D., Hoffmann, R.S., 1999.

Equus caballus

. Mammalian Species 628, 1–14.

Cai, D.W., Han, L., Xie, C.Z., Li, S.N., Zhou, H., Zhu, H., 2007a. Mitochondrial DNA

analysis of Bronze Age horses recovered from Chifeng region, Inner Mongolia,
China. Progress in Natural Science 17, 544–550.

Cai, D.W., Han, L., Zhang, X.L., Zhou, H., Zhu, H., 2007b. DNA analysis of archaeological

sheep remains from China. Journal of Archaeological Science 34, 1347–1355.

Chen, W.H., 1994. Agricultural Archaeology Catalogue of China. Jiangxi Scientiﬁc &

Technical Publisher, Jiangxi (in Chinese).

Chen, X.C., 1999. The origins of Chinese horses. Chinese Cultural Relic Newspaper

(in Chinese).

Chen, S.Y., Duan, Z.Y., Sha, T., Xiangyu, J., Wu, S.F., Zhang, Y.P., 2006. Origin, genetic

diversity, and population structure of Chinese domestic sheep. Gene 376,
216–223.

Clutton-Brock, J., 1999. A Natural History of Domesticated Mammals, second ed.

Cambridge University Press, Cambridge.

Cooper, A., Poinar, H.N., 2000. Ancient DNA: do it right or not at all. Science 289,

1139.

Deng, T., 1999. Historical distrbution of

Equus przewalskii

and its control by climate

ﬂuctuation. Acta Theriologica Sinica 19, 10–16.

Deng, T., 2000. Phylogenetic relationship of the Chinese miniature pony to

Equus

przewalskii

(Perissodacatyla, Equidae). Acta Veterinaria et Zootechnica Sinica 31,

28–33 (in Chinese).

Ishida, N., Oyunsuren, T., Mashima, S., Mukoyama, H., Saitou, N., 1995. Mitochon-

drial DNA sequences of various species of the genus Equus with special refer-
ence to the phylogenetic relationship between Przewalski’s wild horse and
domestic horse. Journal of Molecular Evolution 41, 180–188.

Jansen, T., Forster, P., Levine, M.A., Oelke, H., Hurles, M., Renfrew, C., Weber, J.,

Olek, K., 2002. Mitochondrial DNA and the origins of the domestic horse.
Proceedings of the National Academy of Sciences of the United States of
America 99, 10905–10910.

D. Cai et al. / Journal of Archaeological Science 36 (2009) 835–842

841

Document Outline