Abstract Initial physical anthropology studies into ethnic

diversity were largely dependent on comparative whole body

and craniometric measurements, and through time assessments
of ethnic diversity based on these measures exhibited
increasing statistical sophistication. Since the 1990s, in Asia as
elsewhere in the world, human diversity studies have
increasingly utilized DNA-based analyses, with Y-chromosome
and mtDNA markers providing complementary perspectives on
the origins and gene pool structures of different ethnic groups.
This approach is illustrated in a study of population genetic
structure in PR China, in which DNA samples from the Han
majority and eight ethnic minorities were analyzed. The Y-
chromosome and mtDNA data showed multiple paternal
geographical and ethnic origins but restricted maternal
ancestries. However, interpretive problems were apparent in
the denition of a number of the ethnic study populations,
which appear to reect political as well as genetic inuences.
In all anthropological studies, whether based on anthropometry
or genomic analysis, unambiguous and appropriate community
identication is a prerequisite. J Physiol Anthropol 26(2): 77
82, 2007 http://www.jstage.jst.go.jp/browse/jpa2
[DOI: 10.2114/jpa2.26.77]
Keywords: Anthropometry, Y-chromosome, mtDNA, India,
China
Introduction
The impetus for much of the early research in physical
anthropology was based on the premise that characteristic
differences between human races could be demonstrated
using anthropometric and craniometric measurements. An
early attempt at such an approach was made by Francis Galton,
who derived numerical formulae to describe facial proles,
with the three-fold aim of measuring: i) hereditary likenesses
or differences, ii) racial likenesses or differences, and iii)
ascertaining whether special types of physiognomy were
correlated with denite mental or moral characteristics
(Pearson, 1924). The methodology initiated by Galton was
adopted and extended by Pearson and colleagues in the Galton
Laboratory at University College, London. This culminated in
the derivation of the Coefcient of racial likeness (Pearson,
1926, 1928), which was applied to the quantication of
physical differences between racial groups, e.g., using
craniometric analysis to produce a classication of Asiatic
races (Woo and Morant, 1932).
The Coefcient of racial likeness gained considerable
popularity and, for example, it was employed in the 1931
Census of India, with the 29 anthropometric and craniometric
measures listed in Table 1, plus data on skin, hair, and eye
colour, used to compare and categorize individuals drawn from
a range of Indian populations. The application of the method to
individuals sampled from the four southern states of India,
modern-day Andhra Pradesh (Telegu), Karnataka (Kanarese),
Kerala (Malayali), and Tamil Nadu (Tamil) is illustrated is
Figure 1, with the calculated Coefcients of racial likeness
given as mean values (Guha, 1935).
Despite stringent critiques of the methodology by eminent
contemporaries, e.g., Fisher (1936) and Seltzer (1937), the
Coefcient of racial likeness continued to be employed as the
method of choice in differentiating between the races of man
although with a decline in usage following the abuses of the
discriminatory racial identication legislation and practices in
Germany in the 1930s and early 1940s. Today, comparative
anthropometric and craniometric analysis remains in use as an
investigative tool, e.g., in studies on caste and tribal groups in
India (Sirajuddin et al., 1994; Bharati et al., 2004), and the
Coefcient of racial likeness was adopted in a recent study of
body mass index (Adak et al., 2006).
During the 1940s and 1950s, blood group assays, mainly
based on serotyping techniques, were introduced for inter-
ethnic comparisons, and by the late 1950s and 1960s protein
electrophoresis was in widespread use for the study of
polymorphic variation. The introduction of genomic analysis
in the 1990s has not quelled the controversy as to whether or
Physical Anthropology and Ethnicity in Asia: The Transition from
Anthropometry to Genome-based Studies
A. H. Bittles
1,3)
, M. L. Black
1,2)
and W. Wang
2,1)
1) Centre for Human Genetics, Edith Cowan University, Perth, Australia
2) Capital Medical University, Beijing, PR China
3) Centre for Comparative Genomics, Murdoch University, Perth, Australia
not inter-ethnic genetic differences are meaningful, although in
recent years the debate has largely concentrated on the
question of ethnicity in biomedical studies and clinical practice
(Cooper et al., 2003; Gonslez Burchard et al., 2003; Jorde and
Wooding, 2004). However, there is considerable disquiet
surrounding the relevance and applicability of ethnic
differences in disciplines such as human biology and biological
anthropology.
Genome-based Studies on the Relationship between
Chinese Ethnic Minorities
To study genetic differences between the majority Han
population of PR China and eight representative ethnic
minority populations, nger-prick blood samples or DNA
specimens were obtained from unrelated males belonging to
the majority Han population and the Muslim Hui community
of Liaoning Province; the Muslim Bonan, Dongxiang, and Sala
communities of Gansu Province; the Kucong, Miao, and Yao
communities of Yunnan Province; and Tibetans resident in the
Tibetan Autonomous Province. The samples were analysed for
eight Y-chromosome binary markers (M1 (YAP), M216/M130
(RPS4Y), M89/M213, M9, M175/M214, M122, M45, and
M17) and ve Y-chromosome microsatellite markers ( DYS19,
DYS388, DYS 389a, DYS389b, DYS393), via PCR-RFLP and
uorescent genotyping respectively (Black et al., 2006). The
mtDNA hypervariable region I (HV-I) was amplied and
sequenced using ABI Prism dye primer kits running on 6%
denaturing polyacrylamide gels, with sequence data analysed
using ABI DNA analysis software and SEQUENCER (Gene
Codes). The subjects also were screened for the mtDNA 9-bp
deletion in the COII/tRNA
Lys
intergenic region.
To identify the origins and genetic relatedness of the nine
study communities, the Y-chromosome and mtDNA data were
separately analyzed by summary statistic and phylogenetic
methods (Wang et al., 2003; Black et al., 2006). Broad
haplogroups were inferred from Y-chromosome binary
polymorphisms according to the Y Chromosome Consortium
(YCC 2002), while mtDNA haplogroups were inferred from
HV1 sequences (Yao and Zhang, 2002). Genetic distances
were calculated from Y-chromosome microsatellite data (dm
2
distance) with the software package MICROSAT, while
mtDNA genetic distances (TamuraNei distance) were
calculated with MEGA3 software. Neighbour-joining trees
were then generated from these distance matrices using
MEGA3.
The Y-chromosome haplogroup frequencies indicated a
range of dispersed male origins (Table 2). The Muslim Bonan,
Dongxiang, and Sala displayed a combination of Middle
Eastern, Central Asian, and East Asian male ancestries,
indicated by the presence of haplogroups P*(xR1), R1a1*, and
O3 respectively. The Muslim Hui additionally showed a
signicant frequency of haplogroup C, indicative of a North
Asian, and possibly Mongolian, male ancestral component.
The Han and Miao had predominantly post-Neolithic East
Asian male ancestries, comprising haplogroups O* and O3,
whereas the Tibetan and Yao populations showed older pre-
Neolithic East Asian origins with a high frequency of
haplogroup D ancestries, a relatively rare haplogroup indicated
by the presence of the YAP insertion which also is found at
high frequency in Japanese populations (Hammer et al., 2006).
The Kucong contained a pre-Neolithic male ancestry of
southern Asian origin, which was possibly older than the other
study populations (Black et al., 2006).
The neighbourjoining tree derived from the Y-chromosome
microsatellite data (Fig. 2a) showed no immediately obvious
grouping of populations equivalent to the genetic ancestries
found with the binary data, and the genetic distance measure
conrmed the high levels of male diversity. Examples of
seemingly incongruous relationships include the Hui grouping
closely with the Yao, and the co-clustering of the Bonan and
the Miao. On initial inspection these groupings could be
78 Anthropology and ethnicity in Asia
Table 1 Anthropometric and craniometric measurements used to calculate the Coefcient of racial likeness.
Stature Nasal length Breadth height ratio
Auricular height Nasal breadth Trans fronto-parietal index
Maximum head length Nasal height/depth Orbitonasal index
Maximum head breadth Upper facial length Nasal index
Minimum frontal breadth Total facial length Nasal elevation index
Maximum bizygomatic breadth Horizontal head circumference Upper facial index
Bigonial breadth Sagittal arc Total facial index
Inter-orbital breadth Transverse arc Trans cephalo-facial index
Orbitonasal breadth Length breadth index Vertical cephalo-facial index
Orbitonasal arc Length height index
Fig. 1 Coefcients of racial likeness for four South Indian populations.
interpreted as inferring close ancestral ties, but the histories of
both pairs of populations are notably different (Du and Yip,
1993), and would not support any such relationships. It,
therefore, seems probable that the clustering displayed by the
neighbouringjoining tree linked populations on the basis of
their similar internal population structures, rather than their
ancestral origins.
By comparison, the equivalent mtDNA haplogroup
frequency (Table 2) and the neighbour-joining tree (Fig. 2b)
plotted from the mtDNA data strongly suggested that the major
Bittles, AH et al. J Physiol Anthropol, 26: 7782, 2007 79
Table 2 Y-chromosome and mtDNA haplogroup frequencies of the nine Chinese study populations.
Province/region Liaoning Yunnan Xizang Gansu
Population Han Hui Kucong Miao Yao Tibet Bonan Dongxiang Sala
Y-chromosome haplogroup frequency
Haplogroup
DE 0.190 0.571 0.550 0.056 0.065 0.062
C 0.049 0.227 0.053 0.028 0.062
F*(xK) 0.146 0.227 0.524 0.132 0.071 0.100 0.167 0.152 0.094
O 0.227 0.090 0.282 0.071 0.250 0.056 0.065 0.218
O3 0.480 0.136 0.286 0.533 0.286 0.250 0.196 0.312
K*(xO,P) 0.098 0.136 0.100 0.111 0.109
R1a1 0.181 0.25 0.321 0.218
P*(xR) 0.056 0.092 0.032
mtDNA haplogroup frequency
Haplogroup
D 0.500 0.385 0.111 0.273 0.533 0.091 0.111 0.455
M* 0.167 0.231 0.083 0.111 0.091 0.200 0.111 0.182
C 0.083 0.167 0.273
Z 0.231 0.222
G 0.077
A 0.077 0.167 0.133 0.455
W 0.111
B 0.167 0.250 0.222 0.091 0.067 0.222 0.273
F 0.083 0.333 0.182 0.182 0.111
R* 0.083 0.25 0.222 0.273 0.067 0.111 0.091
Adapted from Black et al. (2006)
Fig. 2 (a) Unrooted neighbourjoining tree based on the genetic distances from Y-chromosome microsatellite haplotypes of nine Chinese
ethnic populations. The scale bar indicates branch length in dm
2
genetic distance units. (b) Unrooted neighbourjoining tree based on the
genetic distances from mtDNA HV1 sequences of nine Chinese ethnic populations. The scale bar indicates branch length in TamuraNei
genetic distance units.
female contribution in each of the ethnic communities was
provided by East Asian women, with considerable homogeneity
indicated by the TamuraNei genetic distance measure of just
0.002 units. Despite the fact that, since the foundation of
communities such as the Muslim Bonan, Dongxiang, Hui, and
Sala from the 8
th
century onwards (Lipman, 1997; Gladney,
1998), each community has been endogamous (Du and Yip,
1993; Black et al., 2006). The two complementary genomic
data sets clearly indicate that, while the male origins of the
non-Han ethnic minorities were drawn from multiple Chinese
and non-Chinese populations, the females of all of the ethnic
minorities showed an almost exclusively Chinese ancestry,
ndings which are in accord with historical records.
Assessing the Results of Anthropogenetic Studies
There are obvious problems in seeking to compare results
obtained via anthropometric measurements to those
determined on the basis of blood group or protein
polymorphisms, and to an even greater extent to studies based
on genomic analysis. Anthropometric measures, such as the
Coefcient of racial likeness, are based on variation that
derives from combinations of genes variously expressed from
conception to adulthood and interacting with environmental
factors that inuence their expression. In contrast, blood group
and protein polymorphisms are mainly single gene measures
which may be selective or neutral in effect, with partial
neutrality the most commonly observed outcome. The single
tandem repeat Y-chromosome loci and the HV-1 mtDNA
markers used in the present study of the Chinese ethnic
communities (Table 2) provide a selectively neutral perspective
on former generations, which can be used to identify the
geographical origins and approximate dates of past migrational
events. As such, the level and precision of the genetic
information they provide is much greater than that available
from blood group or protein polymorphisms, and incomparably
more specic than the genetic content of the anthropometric
analyses.
However, what each of these measures has in common is
absolute reliance on: i) adequate and representative sample
sizes; ii) the accuracy and relevance of the denitional
information available for their study populations; and iii) the
application of appropriate statistical methods. In a large
number of studies, the sample sizes are unrealistically small to
provide a representative picture of the population. Thus, in the
Census of India of 1931 (Guha, 1935), the all-India Coefcient
of racial likeness was calculated from a total of 2,163 males
and 348 females, at a time when the population of the country
totalled 279 million (Dyson, 2001). Of equal importance, in
the South Indian comparative study (Fig. 1), the four different
ethnicities were based on individuals drawn from the
Brahmin caste (varna). Yet the Brahmin varna is notably sub-
divided and, for example, in Karnakata, the home state of the
Kanarese, 101 different and mainly endogamous Brahmin
sub-communities were identied and enumerated in the 1901
Census (Ananda Row, 1901). In this example, the question
arises as to whether each of the four ethnic groups tested was
sampled from a single sub-community or from multiple
endogamous Brahmin communities. Further, the 1931 Census
of India drew specic attention to the fact that, In India, due
to the social prestige and privileges enjoyed by upper castes,
there is always a tendency for individuals to surreptitiously
pass off as members of castes higher than the ones to which
they rightly belong (Guha, 1935, p. iv). Under these
circumstances, unless the selection of each study population
has been conducted with absolute rigour, the results obtained
may be of questionable validity or value.
Despite the demonstrated differences between the Han and
the eight ethnic minority communities (Table 1), as already
indicated, the apparent genetic relationships suggested by the
neighbour-joining trees derived from the Y-chromosome data
(Figs. 2a, 2b) merit some scepticism. During the course of the
study it also became apparent that problems existed in the
identication of the Chinese study communities (Black
et al., 2006). The starting-point for the study was to select
representative ethnic communities, and for this purpose a
number of specic minzu (ethnic populations) were chosen. In
China, 56 minzu received governmental recognition following
the establishment of the Peoples Republic and, besides the
Han, seven of the communities sampled, the Bonan,
Dongxiang, Sala, Hui, Miao, Yao, and Tibetans were
recognized minzu. At the time of the establishment of the
Peoples Republic, the Kucong followed a hunter-gatherer
lifestyle in the tropical forests on the Yunnan-Burma border
and they were not accorded ofcial recognition as a minzu.
On subsequent investigation, it became apparent that at least
two of the ofcially recognized minzu could have been notably
heterogeneous. The Yao are resident in six different provinces
of PR China, and over thirty different zhixi (sub-groups) have
been identied within the ofcial Yao minzu structure
(Litzinger, 1995). Their cultural diversity is further demonstrated
by the worship of separate deities in different Yao
communities, and by particular sub-community beliefs in
mythological origins and religious festivities, all of which were
allegedly unknown to other Yao communities before the
establishment of the Yao minzu in 1949 (Litzinger, 1995). The
Miao also present classicatory difculties, and it has been
claimed that early use of the term Miao referred to most of the
disparate indigenous peoples across southwest China rather
than to a single ethnic group (Diamond, 1995). The present-
day Miao can be sub-divided into three broad groups, the
Hmong in Yunnan, the Hmu in Guizhou, and the Xioob in
Guizhou and Hunan, which exhibit cultural and linguistic
differences (Diamond, 1995).
From an anthropogenetic perspective, historical sources
would strongly suggest that it would be more realistic and
prudent to conduct subject selection in the Yao and the Miao
on a sub-community by sub-community basis. This contention
also applies to the 9.8 million Hui Muslims who are resident
across all of the provinces of PR China (National Bureau of
80 Anthropology and ethnicity in Asia
Statistics of China, 2002). Apart from their shared religion, the
Hui do not appear to have any other unifying cultural or
language criteria that would convincingly permit their analysis
as a single ethnic minority.
The population of Asia is now estimated at 3,921 million,
60.5% of the world total (PRB 2006), and in a large country
such as India the population of 1,104 million is sub-divided
into 50,00060,000 discrete endogamous communities (Bittles,
2005). Since some 93% of the population of PR China
are Han, the level of genetic differentiation is unlikely to match
that of India. But it still seems improbable that 1,304
million people can be convincingly sub-divided into just 56
discrete ethnicities, or that 1,213 million Han are genetically
homogeneous. The original decision to create recognized
ethnic minzu, therefore, appears to have been as much a matter
of political convenience as a recognition of longstanding
anthropological divisions, e.g., based on cultural, religious, and
language diversity (Black et al., 2006).
Most Asian countries, large and small, are characterized by
high levels of clan and tribal sub-division. While community
knowledge of family lineages and clan afliations usually
appears to be credible, genomic studies in Central Asia have
shown that the wider tribal origins claimed by communities
may more mythical than factual (Chaix et al., 2004). Whatever
the method of analysis adopted, in all anthropological studies,
appropriate care is essential in the sampling of study
populations. Otherwise, in the absence of stringent selectional
criteria, the outcomes obtained and hence the conclusions
drawn may be seriously impaired, misleading, or simply
erroneous.
Acknowledgements The cooperation of the nine Chinese
study communities in the provision of samples is
acknowledged with gratitude. Valuable assistance in sample
collection and transport was provided by the Institute of
Genetics, Chinese Academy of Sciences.
References
Adak DK, Gautum RK, Bharati S, Gharami AK, Pal M,
Bharati P (2006) Body mass index and chronic energy
deciency of adult males of Central Indian populations.
Hum Biol 78: 161178
Ananda Row T (1901) Census of India 1901, Vol XXIV,
Mysore Part 1 Report and Part 3. Government Printing
Press, Bangalore
Bharati S, Demarchi DA, Mukherji D, Vasulu TS, Bharati P
(2004) Spatial patterns of anthropometric variation in India
with reference to geographic, climatic, ethnic and linguistic
backgrounds. Ann Hum Biol 32: 407444
Bittles AH (2005) Population stratication and genetic
association studies in South Asia. J Mol Genet Med 1:
4348
Black ML, Wise CA, Wang W, Bittles AH (2006) Genetics and
population history in the study of ethnic diversity in PR
China. Hum Biol 78: 277293
Chaix R, Austerlitz F, Khegay T, Jacquesson S, Hammer MF,
Heyer E et al. (2004)The genetic or mythical ancestry of
descent groups: lessons from the Y chromosome. Amer J
Hum Genet 75: 11131116
Cooper RS, Kaufman JS, Ward R (2003) Race and genomics.
Science 348: 11661170
Diamond N (1995) Dening the Miao: Ming, Qing and
contemporary views. In Harrell S ed. Cultural encounters on
Chinas ethnic frontiers. University of Washington Press,
Seattle
Du R, Yip VF (1993) Ethnic groups in China. Science Press,
Beijing and New York
Dyson T (2001) The preliminary demography of the 2001
Census of India. Pop Dev Rev 27: 341356
Fisher RA (1936) The coefcient of racial likeness and the
future of craniometry. J Roy Anthropol Inst 66: 5763
Gladney DC (1998) Ethnic identity in China: the making of a
Muslim minority nationality. Harcourt Brace & Company,
Fort Worth, Texas
Gonslez Burchard E, Ziv E, Coyle N, Gomez SL, Tan H,
Kartner AJ et al. (2003) The importance of race and ethnic
background in biomedical research and clinical practice.
Science 348: 11701175
Guha BS (1935) Census of India 1931, Vol 1India, Part III
Ethnological, A, Racial afnities of the peoples of India.
Government of India Press, Simla
Hammer MF, Karafet TM, Park H, Omoto K, Harihara S,
Stoneking M et al. (2006) Dual origins of the Japanese:
common ground for hunter-gatherer and farmer Y
chromosomes. J Hum Genet 51: 4758
Jorde LB, Woding SP (2004) Genetic variation, classication
and race Nature Genet 36: S2833
Lipman JN (1997) Familiar strangers: a history of Muslims in
Northwest China. Hong Kong University Press, Hong Kong
University.
Litzinger RA (1995) Making histories: contending conceptions
of the Yao past. In Harrell S ed. Cultural encounters on
Chinas ethnic frontiers. University of Washington Press,
Seattle
National Bureau of Statistics of China (2002) China
Population Statistics Handbook. China Statistics Press,
Beijing
Pearson K (1924) The life, letters and labours of Francis
Galton. vol. 2, London, Cambridge University Press, 324
325
Pearson K (1926) On the coefcient of racial likeness.
Biometrika 18: 105117
Pearson K (1928) The application of racial likeness to test the
characters of samples. Biometrika 20B: 294300
PRB (2006) 2006 World population data sheet, Washington
DC, Population Reference Bureau
Seltzer CC (1937) A critique of the coefcient of racial
likeness. Amer J Phy Anthropol 23: 101109
Sirajuddin SM, Duggirala R, Crawford MH (1994) Population
Bittles, AH et al. J Physiol Anthropol, 26: 7782, 2007 81
structure of the Chenchu and other south Indian tribal
groups: relationships between genetic, anthropometric,
dermatoglyphic, geographic and linguistic differences. Hum
Biol 66: 865884.
Wang W, Wise C, Baric T, Black ML, Bittles AH (2003) The
origins and genetic structure of three co-resident Chinese
Muslim populations: the Salar, Bonan and Dongxiang. Hum
Genet 113, 244252
Woo TL, Morant GM (1932) A preliminary classication of
Asiatic races based on craniometric measurements.
Biometrika 24: 108134
Yao YG, Zhang YP (2002) Phylogeographic analysis of
mtDNA variation in four ethnic populations from Yunnan
Province: new data and a reappraisal. J Hum Genet 47:
311318
Y Chromosome Consortium (2002) A nomenclature system
for the tree of human Y-chromosomal binary haplogroups.
Genome Res 12: 339348
This article was presented at the 8th International Congress of
Physiological Anthropology, 2006 (ICPA 2006), in Kamakura,
Japan.
Received: September 25, 2006
Accepted: October 31, 2006
Correspondence to: Prof. A.H. Bittles, Room 19.387, Edith
Cowan University, 100 Joondalup Drive, Perth WA 6015,
Australia
Phone: 61863045623
Fax: 61863045851
e-mail: a.bittles@ecu.edu.au
82 Anthropology and ethnicity in Asia