Forensic Science International 119 (2001) 269272

The forensic DNA implications of genetic differentiation

between endogamous communities
Lev A. Zhivotovskya,b, Suhaib Ahmedc, Wei Wangb, Alan H. Bittlesb,*
a

N.I. Vavilov Institute of General Genetics, Moscow, Russia

Centre for Human Genetics, Edith Cowan University, Perth, WA 6027, Australia
c
Armed Forces Institute of Pathology, Rawalpindi, Pakistan

Received 4 July 2000; received in revised form 14 November 2000; accepted 14 November 2000

Abstract
In many indigenous minority populations, and among migrants from Asian and African populations now resident in western
Europe, North America and Australia, there is a strong tradition of endogamy and a preference for consanguineous unions.
These marriage practices can result in FST values greatly in excess of the maximum value (0.01) currently recommended for
forensic DNA purposes under guidelines established by the National Research Council (NRC) of the USA. To examine the
possible extent of deviation from this accepted norm, three co-resident Pakistani communities were studied using 10
autosomal dinucleotide markers and six tetranucleotide markers on the Y-chromosome. The mean population subdivision
coefficient (FST) value was 0.13 for the autosomal loci, and Y-chromosome loci exhibited even stronger differentiation with
unique alleles identified in all three communities. The data indicate that even when sub-populations are virtually
indistinguishable in terms of anthropology, geography, ethnicity or culture, they may still exhibit major genetic differentiation.
Where significant population stratification is known to exist, more detailed genetic databases should be developed for forensic
DNA purposes, based on reference data from each of the appropriate sub-populations and not on random or combined
samples. # 2001 Elsevier Science Ireland Ltd. All rights reserved.
Keywords: Forensic DNA; Population genetic structure; Endogamous communities; FST; STR markers

1. Introduction
During the last two decades the application of genomic
analysis to forensic investigation has resulted in levels of
certainty in individual identification and paternity testing
previously impossible with exclusion-based criteria. However, the extent and effects of genetic differentiation within
different populations is an important facet of these investigations that has yet to be satisfactorily addressed.
The degree of uncertainty surrounding this topic was
evident in the creation of the ceiling principle, formulated
by the National Research Council (NRC) of the USA, which
sought to establish a maximum threshold for human allele

Corresponding author. Tel.: 61-8-9400-5623;

fax: 61-8-9400-5851.
E-mail address: a.bittles@ecu.edu.au (A.H. Bittles).

frequency variation [1]. Subsequent discussion ([25], and

others) resulted in rejection of the ceiling principle as overly
conservative, and in its place the NRC adopted criteria that
would allow for a variable measure of population subdivision [6]. In the USA it has been strongly recommended
that a maximum value of 0.01 should be allocated to
the population subdivision coefficient (FST) for forensic
investigations conducted in Careasian, Black and Hispanic
populations, increasing to 0.03 in certain small isolated
groups and communities ([6], recommendation 4.1, p. 122).
Within these limits, a reference genetic database can be
developed and applied to any sub-population when estimating the statistical outcomes of DNA-based forensic investigations, including likelihood ratios, match probabilities
and paternity indices [7].
Significantly, the NRC recommendations were based on
studies carried out on random samples ([6], pp. 102106,
114116, 119). For some groups with few available data,

0379-0738/01/$ see front matter # 2001 Elsevier Science Ireland Ltd. All rights reserved.
PII: S 0 3 7 9 - 0 7 3 8 ( 0 0 ) 0 0 4 4 2 - 4

270

L.A. Zhivotovsky et al. / Forensic Science International 119 (2001) 269272

such as native American and Inuit tribes, it was recommended that a substitute database should be chosen, composed of other groups living in the same geographical region
and/or selected on the recommendation of appropriately
qualified physical anthropologists ([6], p. 123). While random sampling can provide credible estimates for large-scale,
inter-population differentiation, it effectively ignores intrapopulation subdivisions. But, since substantial genetic subdivision occurs in many human societies and recognised
population isolates, the FST values generated by these
internal differences may prove to be significantly larger
than the inter-population differentiation.

3. Results
As indicated in Table 1, for the autosomal dinucleotide
loci the mean FST value across all three communities
was 0.13, ranging from a minimum of 0.05 (D13S270
and D15S101) to a maximum of 0.21 (D15S108). The Ychromosome loci exhibited even greater differentiation, with
unique alleles identified in all three communities (Table 2).
Indeed, at the DYS390 locus the alleles were communityspecific, and so the estimate of FST matched the theoretical
maximum of 1.00. This is especially interesting from a forensic
perspective, given the extensive use of Y-chromosome allele
variability in forensic studies [14].
It could be argued that STR dinucleotide markers have
not been used in forensic settings because of potential errors
in genotyping, and the expectation that mutation rates may
be higher than at tetranucleotide loci [1517]. If these
caveats proved correct the net result could be greater allelic
variation and hence larger genetic differences between
diverged populations at dinucleotide than tetranucleotide
loci. Since FST is treated as a measure of differentiation
among populations that is relative to the total degree of
population diversity, there should however, be little difference between the FST values calculated for markers with di
or tetranucleotide repeats.
To test this supposition, the degree of FST variation found
with different types of STR markers was compared with RST
values (an analog of FST for microsatellites [18]) calculated
for previously reported global data [19,20]. The resultant
mean values in samples from comparable geographical
regions were 0.14 for dinucleotide loci and 0.12 for tetranucleotide loci, levels of differentiation which correspond
well with those reported for other genetic markers ([21],

2. Subject and methods

To investigate the potential degree of genetic differentiation within a major human population known to exhibit
significant subdivision on traditional social/occupational
grounds [8], DNA samples were collected from 165 members of three large co-resident Pakistani communities, the
Awan, Khattar and Rajpoot, drawn from extended families.
Ten autosomal dinucleotide markers on chromosomes 13 and
15 and six tetranucleotide markers on the Y-chromosome
were run on these samples [9].
The communities share the same religion, and they have
few obvious social or cultural differences. Inter-marriage is,
however, generally rare and within all Pakistani communities consanguineous marriage is favoured, most commonly
contracted between first cousins [10,11]. To calculate the
FST values at each locus in terms of the quantity y [12],
genotyping data for the autosomal markers were analysed
using the software package GDA [13].

Table 1
Mean values of genetic differentiation FST at autosomal DNA markers among three co-resident communities in the province of Punjab,
Pakistana,b
Loci

D13S192

D13S126

D13S133

D13S270

D15S11

D15S97

GABRB3

D15S101

D15S108

D15S98 Average

FST

0.16

0.09

0.11

0.05

0.10

0.17

0.17

0.05

0.21

0.20

0.13  0.019

FST values have been estimated as y [12] using the GDA software [13].
Samples from two communities (Khattar and Rajpoot) are identical to those from [9], the Awan sample used in this study differs from
that in [9].
b

Table 2
Allele content in the three Punjab communities at Y-chromosome tetranucleotide loci (data from [9])a
Community

Loci
DYS19

DYS389-I

DYS389-II

DYS390

DYS391

DYS393

Awan
Khattar
Rajpoot

202 (41)
194 (24)
186 (7), 190 (15)

253 (41)
253 (24)
249 (12), 253 (10)

373 (41)
373 (24)
365 (15), 369 (7)

223 (41)
215 (24)
211 (22)

287 (41)
283 (24)
283 (22)

124 (41)
124 (24)
116 (15), 124 (7)

a
Italics indicate the allele notation (size in bp); the number in parenthesis shows the total number of individuals with this allele; unique
alleles are underlined.

L.A. Zhivotovsky et al. / Forensic Science International 119 (2001) 269272

p. 117). On the basis of these results, significant deviation

between the FST values calculated for dinucleotide loci and
the tetranucleotide loci more widely used in forensic studies
would not be expected. Further investigation would however, be needed to determine the precise level of genetic
differentiation in endogamous communities at the commonly used autosomal forensic loci.

4. Discussion
In the present example, and in many other Asian and
African populations, community endogamy is the rule. In
addition, consanguinity acts as a major determinant of
genetic differentiation, and of pregnancy outcome [22].
For example, unions between first cousins (coefficient of
inbreeding, F 0:0625) currently account for 49.4% of all
marriages in Pakistan [10], and in the southern states of India
29.5% of marriages are contracted either between uncle and
niece (F 0:125) or first cousins [23]. What these general
statistics fail to indicate is that marriages are not merely
contracted within families, but also occur almost exclusively
within wider endogamous mating groups and hence separate
breeding pools, examples being castes in India, biradaris in
Pakistan, and tribal groupings in Arab populations. Therefore, in such communities the interpretation of DNA
evidence can be complicated both by a high level of
within-gene pool inbreeding and significant between-gene
pool differentiation [24,25].
The importance of including co-ancestry in the calculation of probability estimates has been emphasised [26], and
some doubt has been expressed ([27], p. 586) as to the
applicability of the FST maxima suggested by the NRC for
forensic purposes, i.e. F ST 0:01 to 0.03 [6]. The present
study clearly demonstrates that within specific communities,
endogamy reinforced by a preference for consanguineous
unions can result in FST values greatly in excess of these
maxima (Table 1). In fact even a value for FST of 0.05, which
was suggested to correct for the effects of population subdivision [28], could result in overstatement of forensic
indices and corresponding match probabilities by two orders
of magnitude based on an assumption of no differentiation
([26], p. 7; [29], Tables 4 and 5).
It should be emphasised that the estimates of genetic
differentiation refer to differences between extended
families (pedigrees) recruited from different communities.
Although the FST estimates formally include possible differences between families in each community, it is appropriate to regard the family and not the entire community as
the forensic population unit. Since in addition to customary
endogamy most marriages also are intra-familial, reflecting
the perceived social and economic benefits of such unions
[8,30].
Although, low values of FST were observed in databases
composed of random samples drawn from large US populations of Caucasian, Black or Hispanic origin ([29], Table 7;

271

[31], Table 1), the present findings show that differentiation

resulting from stratification by genetic relatedness can be
very pronounced in specific populations and thus be important in the evaluation of forensic DNA evidence. Moreover,
this phenomenon is much more widespread than has
been appreciated to date [32]. For example, as a result
of large-scale international migration during the last two
generations, there are now many communities in western
countries that originated in regions where consanguineous
marriage is widely preferential. Thus, in 1990 an estimated
19.7 million foreign-born persons were resident in the
USA, some 25% of whom were born in Asia, including
the Middle East [33]. Comparable figures for the UK were
3.0 million resident migrants (5.8% of the total population),
of whom 1.48 million had been born in the Indian subcontinent or were of Indian, Pakistani or Bangladeshi origin
[34].
In addition, there is a strong tradition of endogamy
and genetic isolation in many indigenous minority communities, such as the Roma (Gypsies) and Travellers, who
frequently favour close consanguineous marriages and are
estimated to number between 7.0 and 8.5 million in Europe
[35]. Likewise, the effects of genetic differentiation have
been recognised among native American tribes, with FST
estimates between 0.020 and 0.111 calculated for the
Apache, Navajo and Pima ([29], Table 8). Therefore,
hidden genetic differentiation that results from preferential
social association between biological relatives challenges
the concept of a reference forensic population, characterised only on grounds of ethnicity, anthropology and
geography.
Where population stratification is known to exist, more
detailed genetic databases may have to be developed for
forensic DNA purposes. In the case of a highly structured
population the associated forensic statistics, and the final
conclusions reached in any given DNA-based forensic
investigation, must be based on reference data from each
of the appropriate sub-populations and should not be reliant
on random or combined samples. If this information is not
readily available, as may currently be the case for many
migrant or itinerant communities, appropriate expansion of
assumed FST values may be essential to preclude legal
challenges based on the adoption of unrealistic DNA-based
probability estimates.

Acknowledgements
We are grateful to an anonymous reviewer for helpful
and constructive suggestions. The work was supported in
part by the Australian Research Council (grant A-350-629),
Edith Cowan University, the Russian Foundation of Basic
Research (grant 98-04-49292), the National Institutes of
Health (grant 1 R03 TW00491-01), and the Morrison
Institute for Population and Resource Studies, Stanford
University.

272

L.A. Zhivotovsky et al. / Forensic Science International 119 (2001) 269272

References
[1] National Research Council, DNA Technology in Forensic
Science, National Academy Press, Washington, DC, 1992,
pp. 7496.
[2] J. Cohen, The ceiling principle is not always conservative in
assigning genotype frequencies for forensic DNA testing,
Am. J. Hum. Genet. 51 (1992) 11651168.
[3] B.S. Weir, Forensic population genetics and the National
Research Council (NRC), Am. J. Hum. Genet. 52 (1993)
437440.
[4] I.W. Evett, J. Scranage, R. Pinchin, An illustration of the
advantages of efficient statistical methods for RFLP analysis
in forensic science, Am. J. Hum. Genet. 52 (1993) 498505.
[5] N.E. Morton, Genetic structure of forensic populations, Am.
J. Hum. Genet. 55 (1994) 587588.
[6] National Research Council, The Evaluation of Forensic DNA
Evidence, National Academy Press, Washington, DC, 1996,
pp. 89124.
[7] K.L. Monson, B. Budowle, Effect of reference database on
frequency estimates of polymerase chain reaction (PCR)based DNA profiles, J. Forensic Sci. 43 (1998) 483488.
[8] S.A. Shami, J.C. Grant, A.H. Bittles, Consanguineous
marriage within social/occupational boundaries in Pakistan,
J. Biosoc. Sci. 26 (1994) 9196.
[9] W. Wang, S.G. Sullivan, S. Ahmed, D. Chandler, L.A.
Zhivotovsky, A.H. Bittles, A genome-based study of
consanguinity in three co-resident endogamous Pakistan
communities, Ann. Hum. Genet. 64 (2000) 4149.
[10] J.C. Grant, A.H. Bittles, The comparative role of consanguinity in infant and child mortality in Pakistan, Ann. Hum.
Genet. 61 (1997) 143149.
[11] R. Hussain, A.H. Bittles, The prevalence and demographic
characteristics of consanguineous marriages in Pakistan, J.
Biosoc. Sci. 30 (1998) 261275.
[12] B.S. Weir, Genetic Data Analysis, II, Sinauer Associates,
Sunderland, MA, 1996, pp. 161201.
[13] P.O. Lewis, D. Zaykin, Genetic data analysis, Computer
Program for the Analysis of Allelic Data, Version 1.0, 1997,
http://chee.unm.edu/gda.
[14] M.A. Kayser, D. Caglia, N. Corach, C. Fretwell, G. Gehring,
Evaluation of Y-chromosomal STRs: a multi-centre study, Int.
J. Leg. Med. 110 (1997) 125133.
[15] R. Chakraborty, M. Kimmel, D.N. Stivers, L.J. Davison, R.
Deka, Relative mutation rates at di, tri and tetranucleotide
microsatellite loci, Proc. Natl. Acad. Sci. U.S.A. 94 (1997)
10411046.
[16] M.W. Feldman, J. Kumm, J.K. Pritchard, Mutation and
migration in models of microsatellite evolution. in:
D.G.Goldstein, C. Schlotterer (Eds.), Microsatellites: Evolution and Applications, Oxford University Press, Oxford,
1999, pp. 98115.
[17] L.A. Zhivotovsky, L. Bennett, A.M. Bowcock, M.W.
Feldman, Human population expansion and microsatellite
variation, Mol. Biol. Evol. 17 (2000) 757767.

[18] M. Slatkin, A measure of population subdivision based on

microsatellite allele frequencies, Genetics 139 (1995) 457462.
[19] A.M. Bowcock, A. Ruiz-Linares, J. Tomfohrde, E. Minch,
J.R. Kidd, L.L. Cavalli-Sforza, High resolution of human
evolutionary trees with polymorphic microsatellites, Nature
368 (1994) 455457.
[20] L.B. Jorde, A.R. Rogers, M. Bamshad, W.S. Watkins, P.
Krawowiak, et al., Microsatellite diversity and the demographic history of modern humans, Proc. Natl. Acad. Sci.
U.S.A. 94 (1997) 31003103.
[21] L.L. Cavalli-Sforza, P. Menozzi, A. Piazza, The History and
Geography of Human Genes, Princeton University Press,
Princeton, NJ, 1994, pp. 60157.
[22] A.H. Bittles, J.V. Neel, The costs of human inbreeding and
their implications for variation at the DNA level, Nature
Genet. 8 (1994) 117121.
[23] A.H. Bittles, W.M. Mason, J. Greene, N. Appaji Rao,
Reproductive behaviour and health in consanguineous
marriages, Science 252 (1991) 789794.
[24] K.L. Ayres, A.D.J. Overall, Allowing for within-sub-population inbreeding in forensic match probabilities, Forensic Sci.
Int. 103 (1999) 207216.
[25] J.M. Curran, C.M. Triggs, J. Buckleton, B.S. Weir, Interpreting DNA mixtures in structured populations, J. Forensic Sci.
44 (1999) 987995.
[26] D.J. Balding, R.A. Nichols, A method for quantifying
differentiation between populations at multi-allelic loci and
its implications for investigating identity and paternity,
Genetica 96 (1995) 312.
[27] D.J. Balding, R.A. Nichols, Significant genetic correlations
among Caucasians at forensic DNA loci, Heredity 78 (1997)
583589.
[28] D.J. Balding, R.A. Nichols, DNA profile match probability
calculation: how to allow for population stratification,
relatedness, database selection and single bands, Forensic
Sci. Int. 64 (1994) 125140.
[29] B.S. Weir, The effects of inbreeding on forensic calculations,
Ann. Rev. Genet. 28 (1994) 597621.
[30] A.H. Bittles, The role and significance of consanguinity as a
demographic variable, Pop. Dev. Rev. 20 (1994) 561584.
[31] B. Budowle, The effects of inbreeding on DNA profile
frequency estimates using PCR-based loci, Genetica 96
(1995) 2125.
[32] A.H. Bittles, Empirical Estimates of the Global Prevalence of
Consanguineous Marriage in Contemporary Societies, Morrison Institute for Population and Resource Studies, Stanford
University, Stanford, 1998, pp. 169.
[33] OECD. Trends in international migration. Organisation for
Economic Co-operation and Development, Paris, 1999, p. 38.
[34] HMSO. Social Focus on Ethnic Minorities, Her Majestys
Stationery Office, London, 1996, pp. 918.
[35] A.H. Bittles, Consanguineous marriage and its outcomes in
European populations. in: C. Susanne, E.B. Bodzsa r (Eds.),
Human Population Genetics in Europe, Eo tvo s University
Press, Budapest, 2000, pp. 135146.