Beruflich Dokumente
Kultur Dokumente
Department of Biological Sciences, Middle East Technical University, 06531 Ankara, Turkey
ABSTRACT In the evolutionary history of modern by multidimensional scaling method. Analysis suggested
humans, Anatolia acted as a bridge between the Cauca- that, genetically, Anatolia is more closely related with the
sus, the Near East, and Europe. Because of its geographi- Balkan populations than to the Central Asian popula-
cal location, Anatolia was subject to migrations from tions. Central Asian contribution to Anatolia with respect
multiple different regions throughout time. The last, well- to the Balkans was quantied with an admixture analy-
known migration was the movement of Turkic speaking, sis. Furthermore, the association between the Central
nomadic groups from Central Asia. They invaded Anatolia Asian contribution and the language replacement episode
and then the language of the region was gradually was examined by comparative analysis of the Central
replaced by the Turkic language. In the present study, Asian contribution to Anatolia, Azerbaijan (another
insertion frequencies of 10 Alu loci (A25 5 0.07, APO 5 Turkic speaking country) and their neighbors. In the
0.96, TPA25 5 0.44, ACE 5 0.37, B65 5 0.57, PV92 5 present study, the Central Asian contribution to Anatolia
0.18, FXIIIB 5 0.52, D1 5 0.40, HS4.32 5 0.66, and was estimated as 13%. This was the lowest value among
HS4.69 5 0.30) have been determined in the Anatolian the populations analyzed. This observation may be
population. Together with the data compiled from other explained by Anatolia having the lowest migrant/resident
databases, the similarity of the Anatolian population to ratio at the time of migrations. Am J Phys Anthropol
that of the Balkans and Central Asia has been visualized 136:1118, 2008. V 2007 Wiley-Liss, Inc.
C
Anatolia (the Asian segment of modern day Turkey) lier movement was a facilitator for further Turkic tribe
has been occupied by modern humans since the lower migrations (Benedetto et al., 2001) because of the
Paleolithic times (Kuhn, 2002). Geographically, Anatolia Turkic language spoken in the region. Turkic tribal
is at the junction between the Balkans, the Near East migration into the Anatolian region continued for more
and the Caucasus, it has acted as a bridge and a reser- than two centuries after the Seljuks (Toynbee, 1970;
voir for numerous movements of modern human beings Roux, 1997). Hence, genetic contribution from the
since very early times. It provided a suitable refuge dur- tribes was expected to be relatively high within the
ing the Last Glacial Maximum where modern humans region compared with other neighboring, non-Turkic
could survive (Cinnioglu et al., 2004) and it was an im- speaking regions.
portant passageway for the emergence and spread of the Before Seljuks, Anatolia was under the rule of Eastern
farming industry (Renfrew, 2000; Cinnioglu et al., 2004). Romans but was mainly inhabited by people of Greek ori-
Excavations on Catal Hoyuk, one of the oldest settle- gin for nearly two millennia (Toynbee, 1970). The process
ment areas in Anatolia, revealed the presence of devel- of change of language and religion by the Seljuks that is
oped agricultural communities residing in this area as assimilation of the residents but not the invaders in Ana-
early as 9000 BCE (Before Common Era) (Hodder, 2006). tolia, was one of the puzzles of history (Toynbee, 1970).
Anatolia was populated by various civilizations, such As the part of puzzle, estimation of the relative size of
as the Hattians, the Hurries, the Hittites, the Phrygians, arriving nomads was the concern of many studies.
the Lydians, the Urartians, the Medes, the Romans, the In earlier studies different molecular markers such
Sassanids, the Byzantines, the Seljuk Turks, and the as proteins (Brega et al., 1998), mitochondrial DNA
Ottomans (Tambets et al., 2000). As a bridge between
the West and the East, Anatolia continued to be subject
to migrations from multiple regions throughout time.
Grant sponsor: Middle East Technical University Research Fund;
The most recent migration took place with the Turkic Grant numbers: BAP-2003-07-02-00-39 and BAP-2004-07-04-02.
speaking nomadic groups, the part of Oghuz Turks.
Starting from the 10th Century CE, they spread away *Correspondence to: Ceren Caner Berkman, Department of Bio-
from their homeland on the north bank of Syr Darya logical Sciences, Middle East Technical University, 06531 Ankara,
(Cahen, 1968). Conventionally, the date of their arrival Turkey. E-mail: cerenberkman@gmail.com
to Anatolia was 1071 and was formally referred to as
Seljuks (Toynbee, 1970; Lewis, 1995). Because of these Received 31 May 2007; accepted 5 November 2007
migrations, the language of the region started to be
replaced by the Turkic language. Furthermore, migra- DOI 10.1002/ajpa.20772
tions of Turkic tribes to and around Anatolia did not Published online 27 December 2007 in Wiley InterScience
cease after the arrival of the Seljuks. Perhaps, the ear- (www.interscience.wiley.com).
C 2007
V WILEY-LISS, INC.
12 C.C. BERKMAN ET AL.
Fig. 1. Map of Anatolia with the locations of the studied samples marked.
(mtDNA) (Calafell et al., 1996; Comas et al., 1996, 1998), determine the relative similarity of the Turkish popula-
and Y-chromosomal markers (Wells et al., 2001; Cinnioglu tion to that of the Balkans and Central Asia. To calcu-
et al., 2004) were used to compare Anatolian population late the proportion of Anatolians that can be traced to
with populations from Europe and Asia. Results sug- Central Asians, Alu insertion polymorphisms, multiple
gested that Anatolians are more closely related to Euro- independent autosomal loci, were employed jointly thus
pean populations than Central Asian populations, sug- providing a general view of the genomic contribution of
gesting that migrations stemming from Central Asia had the Turkic populations. Because the Alu markers repre-
little genetic effect on the current day Turkish gene pool. senting unique insertions (Batzer et al., 1991) and being
Some of the newly developed computational methods under the effect of pure drift (Garcia-Obregon et al.,
allowed scientists to investigate past population proc- 2007) satised the assumptions of Chikhi et al.s (2001)
esses more deeply. For example, contribution by migra- admixture method. Thus Chikhi et al.s (2001) method
tions to the gene pool of populations can be partitioned was used to estimate the proportion Finally, to obtain a
by analyzing the current pattern of genetic variation more detailed perspective about the contribution of
with an admixture analysis. Many statistical methods Turkic speaking tribes within the Central AsianBalkan
[ex: Roberts and Hiorns (1965), Longs (1991), Chakra- region, as well as Anatolia, Azerbaijan (another Turkic
bortys (1992), Bertorelle and Excofers (1998), and Chi- speaking population) and some other non-Turkic speak-
khi et al.s (2001) methods] have been developed to esti- ing neighboring populations were also examined using
mate admixture proportions from genetic data (Jobling Chikhi et al.s (2001) method.
et al., 2004). The only prior study that used an admix-
ture analysis method to quantify Central Asian contribu-
tion to Anatolia, with respect to Balkans, was that of MATERIALS AND METHODS
Benedetto et al. (2001). Samples from Anatolia, DNA isolation, and Alu
In the attempts to quantify the Central Asian contri- insertion loci
bution to the Anatolian gene pool, generally unilinear
markers: maternally inherited mitochondrial DNA In the present study, 10 Alu insertion polymorphisms,
(mtDNA) (Rolf et al., 1999; Benedetto et al., 2001) and namely A25, APO, TPA25, ACE, B65, PV92, FXIIIB, D1,
paternally inherited Y-chromosomal markers (Rolf et al., HS4.32, and HS4.69 were typed in 202 unrelated indi-
1999; Benedetto et al., 2001; Cinnioglu et al., 2004) viduals from different regions of Anatolia (Fig. 1). Blood
were analyzed. Estimates being between \9% (Cin- or buccal samples were obtained from unrelated volun-
nioglu et al., 2004) and 30% (Benedetto et al., 2001) teer donors with their consents. The allele frequencies of
were in a wide range. Although these markers provide 102 of the subjects had already been published (Dinc
some insights about the relative contributions of differ- and Togan, 2005).
ent sexes they are haploid in nature therefore suffer Total genomic DNA from the blood samples was iso-
from genetic drift four times more than that of the lated in accordance with the phenol-chloroform-isoamy-
autosomal loci (Jobling et al., 2004). Furthermore, lalcohol method (Sambrook et al., 1989). For the buccal
migration being a population process needs to be ana- samples, epithelial cells were collected from inside of
lyzed by employing many autosomal loci to give a more both of the cheeks by scraping with a sterile brush. The
reliable genome wide estimate. brush was placed into a 2X lysis buffer, the cells were
In the present study, the genetic diversity of Anatolia digested overnight by proteinase-K (20 mg/ml) and the
based on Alu insertion polymorphisms was determined. solution was extracted twice by phenol-chloroform-iso-
Then, together with the data compiled from databases, amylalcohol solution (25:24:1) by centrifuging at 10,000
multidimensional scaling (MDS) method was applied to rpm for 2 min at room temperature. One more extraction
was performed by a chloroform-isoamylalcohol solution to be so after the arrival, hence, second parental popula-
(24:1), 3 M Sodium Acetate (NaAc) together with isopro- tion was accepted as the Balkans. The composite popula-
panol were added and incubated overnight at 2208C. tion of the Balkans was contained populations from pres-
The DNA was precipitated by centrifuging at 10,000 rpm ent day Greece, Albania, and Romania. Because no
for 10 min. genetic barrier was observed between these populations
The obtained DNA was used to genotype 10 Alu inser- (Comas et al., 2004) it seemed sensible to pool them.
tion polymorphisms (A25, APO, TPA25, ACE, B65, PV92, Admixed populations represented present day Anatolia,
FXIIIB, D1, HS4.32, and HS4.69) as described in the Azerbaijan, Armenia, Georgia, Syrian, and the Northern
previous studies (Arcot et al., 1995; Batzer et al., 1996). Caucasus (which include Ingushetia, Kabardino-Balkaria,
During amplication of the loci, hot start was applied to Dagestan, Cherkessia).
improve the accuracy of primer annealing (Ueda et al., Comparative studies were performed using seven of
1996). the Alu insertion loci, which were common within the
populations under consideration, namely: A25, APO,
Compiled data for the other populations and the TPA25, ACE, B65, PV92, and FXIIIB polymorphisms.
loci used in comparative studies
Statistical analysis
Data for those populations from the Balkans, Central
Asia, the Near East, and the Caucasus were compiled The allele frequencies for the 10 Alu insertion poly-
using the allele frequency database, ALFRED (Rajeevan morphisms (A25, APO, TPA25, ACE, B65, PV92, FXIIIB,
et al., 2005). Populations together with their sample D1, HS4.32, and HS4.69) genotyped in Anatolia were
sizes are listed in Table 1. calculated by direct counting. The Hardy-Weinberg equi-
In this study, as well as individual populations such as librium was assessed by the Fishers exact test to calcu-
the one from Syria, regional, composite populations late the P value using the Markov-Chain Monte Carlo
(such as the populations of the Northern Caucasus) were method (Guo and Thompson, 1992) provided by the Arle-
considered (Table 1). In admixture analysis, Central Asia quin program (Excofer et al., 2005).
was considered as one of the parental populations. The To measure the degree of genetic differentiation among
Central Asian parental population was composed of sam- populations, GST values for seven Alu insertion polymor-
ples from present day Kazakhstan, Kyrgyzstan, Uighur, phisms (A25, APO, TPA25, ACE, B65, PV92, and
and Uzbekistan. These countries are located where long- FXIIIB) were calculated using the DISPAN package pro-
range spread of the Seljuks originated and/or these are gram (Ota, 1993). The GST value estimates the propor-
Turkic-speaking countries. Before the arrival of the Sel- tion of total variance due to differences among popula-
juks, Anatolia had a majority of Greek resident for more tions. Therefore, higher GST values indicate higher
than two millennia. Seljuk Turks arrived and settled in genetic differentiation among populations (Nei, 1987).
Anatolia (Toynbee, 1970). Therefore, Balkans repre- Furthermore, to visualize the genetic relatedness of the
sented before the Turks state of Anatolia and remained examined populations a multidimensional scaling (MDS)
analysis (Kruskal, 1964) based on an R matrix, the nor- To compare the genetic similarity of the Anatolian pop-
malized covariance matrix of allele frequencies across ulation with that of the Central Asian populations and
populations (Harpending and Jenkins, 1973), was per- the Balkan populations seven of the 10 Alu insertion
formed using SPSS (version 13.0, SPSS, Inc., Chicago, polymorphisms (A25, APO, TPA25, ACE, B65, PV92, and
IL) statistical package. FXIIIB) which were common among the populations
In the presented study, Chikhi et al.s (2001) admix- were analyzed. For these seven loci, the frequencies of
ture estimation method implemented in LEA package the allele with Alu insertion in all of the studied popula-
program (Langella et al., 2001) was used. This method tions were given in the Table 3.
assumes that P1 and P2 are two independent parental The heterozygosity values of Anatolia (0.35 6 0.07),
populations, of size N1 and N2. These two parental popu- the Balkans (0.35 6 0.07), and the Southern Caucasus
lations contributed proportions p1 and p2 of the genes of (0.36 6 0.06) were very close to each other and lower
a third admixed population, H of size Nh, T generations than that of the Central Asian heterozygosity value
ago. The method further assumes that the parental and (0.39 6 0.06). The Northern Caucasus heterozygosity
admixed populations continued to evolve independently values with a mean of 0.31 6 0.06 were the lowest val-
after hybridization. The method estimates the admixture ues determined.
proportion (p1) of one of the parents (P1) by applying a The highest GST values were obtained for FXIIIB and
Bayesian and a coalescent-based approach. In the pres- PV92 loci (Table 3). This suggests that these two loci
ent study, P1 represents the Central Asian population have the highest power to differentiate populations/
and calculated admixture estimates (p1) based on the regions. In the present study, when populations were
posterior distribution curves correspond to the Central compared, the GST values ranged from 0.20 FXIIIB to
Asian genetic contribution to the admixed population. P2 0.02 for A25 and APO loci. On the other hand, the range
and H represent the Balkan and admixed populations of GST values of the regions (0.12 with FXIIIB0.01 with
respectively. Furthermore, for both the parental popula- A25) is narrower than that of populations. The GST aver-
tions and the admixed population, the amount of drift age over all loci between populations was 0.09 whereas
since admixture, (t1, t2, th) which is determined by the the GST average over all loci between regions was 0.06
time (T) scaled by the effective population size (Ni), was
obtained from the Chikhi et al. (2001) admixture method
and the posterior distributions were given. The highest TABLE 2. Alu insertion frequencies in Anatolia
posterior density (HPD) intervals and likelihood curves The frequency of
were obtained by using the R language (R Development Locus 2Na the Alu insert
Core Team, 2007).
A25 396 0.068
APO 394 0.957
RESULTS TPA25 402 0.435
Alu insertion frequencies ACE 404 0.371
B65 370 0.568
The Alu insertion frequencies for 10 Alu insertion PV92 372 0.183
polymorphisms in the Anatolian population are listed in FXIIIB 380 0.516
D1 366 0.396
Table 2. As can be noted, all 10 loci were polymorphic HS4.32 336 0.658
with APO (0.957) being the closest to xation. The popu- HS4.69 192 0.302
lation was found to be in Hardy-Weinberg equilibrium
a
for all analyzed loci. Sample size in number of chromosomes typed for each locus.
TABLE 3. The Alu insertion frequencies, associated heterozygosity (H) and the GST values for populations and regions
Population A25 APO TPA25 ACE B65 PV92 FXIIIB Ha
Albania 0.08 1.00 0.56 0.47 0.67 0.20 0.60 0.35 6 0.08
Romania 0.03 0.92 0.58 0.47 0.57 0.20 0.41 0.36 6 0.07
Greece 0.04 0.96 0.56 0.32 0.65 0.10 0.50 0.32 6 0.08
Balkans 0.06 0.96 0.55 0.40 0.62 0.16 0.50 0.35 6 0.07
Kazakhstan 0.07 0.92 0.53 0.60 0.35 0.55 0.53 0.39 6 0.07
Uighur 0.08 0.95 0.53 0.57 0.43 0.55 0.78 0.37 6 0.07
Uzbekistan 0.10 0.87 0.51 0.60 0.59 0.47 0.55 0.42 6 0.06
Kyrgyzstan 0.09 0.90 0.47 0.63 0.55 0.54 0.81 0.38 6 0.06
Central Asia 0.08 0.92 0.51 0.60 0.47 0.53 0.69 0.39 6 0.06
Azerbaijan 0.00 0.94 0.51 0.22 0.69 0.38 0.10 0.29 6 0.07
Armenia 0.06 0.87 0.43 0.48 0.45 0.01 0.34 0.33 6 0.08
Georgia 0.09 0.93 0.49 0.35 0.73 0.25 0.61 0.36 6 0.06
Southern Caucasus 0.06 0.92 0.48 0.35 0.64 0.22 0.43 0.36 6 0.06
Kabardino-Balkaria 0.11 0.93 0.29 0.27 0.43 0.15 0.14 0.31 6 0.05
Cherkessia 0.05 0.94 0.45 0.30 0.68 0.27 0.25 0.35 6 0.07
Dagestan 0.03 0.87 0.37 0.16 0.32 0.16 0.15 0.29 6 0.05
Ingushetia 0.07 0.94 0.22 0.34 0.21 0.13 0.00 0.23 6 0.06
Northern Caucasus 0.06 0.93 0.35 0.27 0.49 0.18 0.17 0.31 6 0.06
Syria 0.00 0.93 0.51 0.40 0.31 0.18 0.28 0.32 6 0.07
Anatolia 0.06 0.97 0.43 0.38 0.59 0.20 0.51 0.35 6 0.07
GST based on populations 0.02 0.02 0.04 0.07 0.09 0.14 0.20 Average 0.09
GST based on regions 0.01 0.01 0.02 0.04 0.05 0.09 0.12 Average 0.06
a
Average heterozygosity 6 standard error.
Multidimensional scaling
To assess the population relationships, multidimen-
sional scaling (MDS) analysis was carried out. Popula-
tion relationships based on an R matrix (not shown)
were visualized on a two-dimensional genetic map shown
in Figure 2.
The two dimensional genetic map accounted for 94.8%
of the total variance. Central Asian populations (from
present day Kazakhstan, Kyrgyzstan, Uzbekistan, and
Uighur) were differentiated from the other populations.
The rst dimension separated the Central Asian popula-
tions from the populations of the Northern Caucasus.
Furthermore, the second dimension separated the Bal-
kan populations (from present day Greece, Romania, and Fig. 3. The posterior distribution curves of the ti 5 T/Ni
Albania) together with populations from Anatolia, Azer- distribution. (A) Central Asia, (B) Balkans, (C) admixed
baijan, Georgia, and Cherkessia from populations from populations.
Armenia, Central Asia, the Near Easter, and the North-
ern Caucasus.
mates (p1) for each of the admixed populations. Results
Admixture estimates indicated that the Central Asian contribution was 13%
for the Turkish population. Furthermore, it was observed
Table 4 presents the summary statistics for the poste- that the Central Asian contributions in the neighboring
rior distributions of the Central Asian admixture esti- regions were higher than that of Anatolia and ranged
between 20% and 33%. The Central Asian contribution lia. To nd the Central Asian contribution to Anatolia,
was highest in Syria and Azerbaijan. Moreover, the esti- an admixture analysis method has only been used in a
mates were approximately the same for Armenia, Geor- study by Benedetto et al. (2001). Benedetto et al. (2001)
gia, and the Northern Caucasus. analyzed mtDNA HVRI sequences, ve Y-STR (DYS19,
DYS390, DYS391, DYS392, DYS393) loci, and one auto-
somal microsatellite locus (TH01). On the basis of Ber-
Genetic drift
torelle and Excofers (1998) admixture method, which
Figure 3 displays the posterior distributions of the considers the effect of mutation as the main evolutionary
genetic drift since admixture for the parental and force, they estimated the Central Asian contribution to
admixed populations. The same parent populations (Cen- Anatolia as approximately 30%.
tral Asia and Balkans) were used for each admixed pop- Alu insertions are stable and selectively neutral
ulation. The method of Chikhi et al. (2001) gives three markers (Batzer et al., 1996). Without affects from muta-
estimates for drift corresponding to two parental popula- tion and selection, Alu insertions reect the processes of
tions and the admixed population. In the present study, interaction between gene ow and genetic drift within
seven admixed populations were analyzed and hence the past (Garcia-Obregon et al., 2007). Therefore, in the
seven posterior distribution curves were obtained for present study, to quantify the Central Asian genetic con-
each of the parents. Therefore, for the parental popula- tribution to Anatolia instead of using the Bertorelle and
tions (Fig. 3A,B) each curve corresponds to an estimate Excofers (1998) admixture method the Chikhi et al.s
of genetic drift obtained from the analysis of a particular (2001) admixture method was used. This method consid-
admixed population whereas the posterior distribution ers the effect of genetic drift as the main evolutionary
curves for the admixed populations are represented sep- force. Through this method, the posterior distribution
arately in Figure 3C. curves for genetic drift were obtained. If curves for a
Almost identical and narrow curves are accepted as a population are narrow and almost identical, then it is
sign of limited genetic drift since the time of an admix- suggested that the population experienced limited drift
ture event and thus indicate a large population size over since admixture. This suggests that the population has
a long period of time (Chikhi et al., 2001). In the present rather large long-term population size (Chikhi et al.,
study, the posterior distribution curves for Central Asia 2001, 2002). For parent populations which were repre-
were wider (Fig. 3A) than the posterior distribution sented with pooled, composite populations (with 2N 5
curves of the Balkans (Fig. 3B). 620 for Central Asia and 2N 5 462 for the Balkans),
In Figure 3C, each curve corresponds to the ti distribu- wide and variable posterior distribution curves were
tion of an admixed population. When the distributions obtained (Fig. 3A,B). Furthermore, much wider curves
for the admixed populations were considered, the nar- were obtained for Central Asia than the Balkans sug-
rowest posterior distribution curve was obtained for Ana- gesting that Central Asia exhibited signs of higher
tolia, suggesting that Anatolia had a large population genetic drift than the Balkans between the time of
size and therefore was not greatly affected by drift (Fig. admixture and sampling. Previously, observation of low
3C). Excluding the distribution of the Anatolians, all of diversity and high population specic Y-chromosomal
the admixed populations distributions were wide (Fig. clusters in Central Asian populations lead Zerjal et al.
3C). (2002) to a similar conclusion. For all of the examined
populations, heterozygosity levels (Table 3) were corre-
lated to the sample sizes of the populations (Table 1).
DISCUSSION However, despite the relatively large sample size (aver-
Previous studies suggested that Anatolia had a step- age 2N 5 373) the lowest heterozygosity was observed in
ping stone position between Asian and European popula- the Northern Caucasus. Also, among the composite
tions and that it is closer to the European populations populations, the Northern Caucasus displayed the high-
(Calafell et al., 1996; Comas et al., 1996, 1998; Brega est effect of drift (Fig. 3). In parallel to the conclusion of
et al., 1998; Wells et al., 2001; Cinnioglu et al., 2004). In a previous study by Nasidze et al. (2001), it can be con-
the present study, relative position of Anatolians was cluded that isolation and genetic drift in the Northern
addressed by employing Alu insertion polymorphisms. Caucasus are the causes of low levels of average hetero-
The MDS plot depicted that the Anatolian population is zygosity. Furthermore, excluding Anatolia, all of the
closest to the Balkan populations and hence results of admixed populations exhibited effects from drift (Fig.
the present study supported the previous observations. 3C). Observing the effect of genetic drift in both parent
Furthermore, there was a clear differentiation of Central populations and the admixed populations supported the
Asian populations from Anatolian and all other analyzed decision to use Chikhi et al.s (2001) admixture method
populations. in the present study.
In the literature, there are attempts to quantify the The Central Asian contribution to Anatolia, with
Asian genetic contribution in the Anatolian gene pool. In respect to the Balkans, was determined as 13% with a
these studies, Cinnioglu et al. (2004) and Rolf et al. HPD interval of 0.0000.474. The calculated Central
(1999) used the frequencies of Asian specic lineages to Asian contribution was considerably lower than the esti-
quantify the Central Asian contribution to Anatolia. mate (30%) made by Benedetto et al. (2001). At this
Cinnioglu et al. (2004) determined the Central Asian point of discussion, it must be emphasized that in addi-
contribution as lower than 9% by comparing the frequen- tion to the use of a different admixture method, composi-
cies of Asian specic Y-chromosomal haplogroups C and tion of the parent populations in Benedetto et al.s (2001)
O3 in Asia and Anatolia. Based on mtDNA haplogroups study and that of the present study were different. On
(H, U, K, T, and M) and allele frequencies of Y-chromo- the other hand, although the estimation methods and
somal microsatellite (DXY156 13 repeat and DYS389 analyzed molecular markers were different, the calcu-
5 repeat) Rolf et al. (1999) also determined that there is lated admixture estimate was parallel to that of Cinnio-
approximately 10% Central Asian contribution in Anato- glu et al. (2004) and Rolf et al. (1999).