The Caucasus as an Asymmetric Semipermeable Barrier to

Ancient Human Migrations

Bayazit Yunusbayev,,1,2,3 Mait Metspalu,,1 Mari Jarve,*,1,4 Ildus Kutuev,1,2,3 Siiri Rootsi,1
Ene Metspalu,4 Doron M. Behar,1 Kart Varendi,4 Hovhannes Sahakyan,1,5 Rita Khusainova,2,3
Levon Yepiskoposyan,5 Elza K. Khusnutdinova,2,3 Peter A. Underhill,6 Toomas Kivisild,1,4,7 and
Richard Villems1,4,8
1

Estonian Biocentre, Tartu, Estonia

Institute of Biochemistry and Genetics, Ufa Research Center, Russian Academy of Sciences, Ufa, Russia
3
Department of Genetics and Fundamental Medicine, Bashkir State University, Ufa, Russia
4
Department of Evolutionary Biology, University of Tartu, Tartu, Estonia
5
Human Genetics Group, Institute of Molecular Biology, Academy of Sciences of Armenia, Yerevan, Armenia
6
Department of Genetics, Stanford University School of Medicine
7
Department of Biological Anthropology, University of Cambridge, Cambridge, United Kingdom
8
Estonian Academy of Sciences, Tallinn, Estonia
These authors contributed equally to this work.
*Corresponding author: E-mail: mari.jarve@ut.ee.
Associate editor: Sohini Ramachandran
2

The Caucasus, inhabited by modern humans since the Early Upper Paleolithic and known for its linguistic diversity, is
considered to be important for understanding human dispersals and genetic diversity in Eurasia. We report a synthesis of
autosomal, Y chromosome, and mitochondrial DNA (mtDNA) variation in populations from all major subregions and
linguistic phyla of the area. Autosomal genome variation in the Caucasus reveals significant genetic uniformity among its
ethnically and linguistically diverse populations and is consistent with predominantly Near/Middle Eastern origin of the
Caucasians, with minor external impacts. In contrast to autosomal and mtDNA variation, signals of regional Y
chromosome founder effects distinguish the eastern from western North Caucasians. Genetic discontinuity between the
North Caucasus and the East European Plain contrasts with continuity through Anatolia and the Balkans, suggesting major
routes of ancient gene flows and admixture.
Key words: Caucasus, population genetics, human migration, autosomal markers, Y chromosome, mtDNA.

Introduction
The earliest evidence of the dispersal of the genus Homo
outside Africa comes from the Caucasus, a mountainous
region between the Black and Caspian Seas, linking the
Near/Middle East and the East European Plain (Lieberman
2007; Lordkipanidze et al. 2007). Anatomically modern
humans appeared there at least 42,000 years ago (Adler
et al. 2008). The early Upper Paleolithic (UP) sites in the
Caucasus (Bar-Yosef et al. 2006; Pinhasi et al. 2008) and
the adjacent East European Plain (Krause et al. 2010) are
nearly contemporary, whereas the lower Danube Basin
UP temporal estimates (Mellars 2006) predate the northern
Pontocaspian UP by several thousands of years. This makes
the two pathsacross the Caucasus and via Anatolia and
the Balkansequally plausible for the pioneer phase of
peopling of East Europe. Likewise, people from both the
Caucasus and the East European Last Glacial Maximum
(LGM) refugia (Adams and Faure 1997; Tarasov et al.
2000) may have been the source population for the repopulation of East Europe after the LGM, whereas the route

across the Caucasus may have been used by early farmers

to reach the northern areas. A second set of intriguing issues
is related to the linguistic diversity of the Caucasus. Although
the Abkhazian-Adyghe (Northwest [NW] Caucasian), NakhDagestanian (Northeast [NE] Caucasian), and Kartvelian
(South Caucasian) language families are indigenous for the
area, several additional languages of the Indo-European and
Turkic families have been brought to the Caucasus by people
migratingfromelsewhere,presumablymuchlater,andarespoken there at present (Comrie 2008) (fig. 1A). But the role of
the Caucasus in human dispersals in Eurasia has remained
obscure, and the extent to which the unique cultural and linguistic diversity of the Caucasus is reflected in the genetic diversity of the extant populations in the region has been studied
only from the uniparental perspective (Nasidze et al. 2003;
Nasidze et al. 2004; Roostalu et al. 2007; Underhill et al.
2010; Balanovsky et al. 2011; Myres et al. 2011). Accordingly,
here, we address several so far unsolved problems: Does the
present-day genetic structuring of Eurasian populations offer
an insight to the peopling of the Caucasus and to later

The Author 2011. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please
e-mail: journals.permissions@oup.com

Mol. Biol. Evol. 29(1):359365. 2012 doi:10.1093/molbev/msr221

Advance Access publication September 13, 2011

359

from http://mbe.oxfordjournals.org/ by guest on June 24, 2013

ResearchDownloaded
article

Abstract

Yunusbayev et al. doi:10.1093/molbev/msr221

MBE

Downloaded from http://mbe.oxfordjournals.org/ by guest on June 24, 2013

FIG. 1. (A) geographical map of the populations of the Caucasus included in this study. The language family affiliation of each population is
given. Adapted from Wikipedia. (B) Pairwise FST distances between the Caucasus and neighboring populations, ranging from red (low) to blue
(high), based on autosomal data. The populations (data from this study and the literature; Li et al. 2008; Behar et al. 2010) are divided into
regional groups. (C) Plot of the first and second components of the PCA of the Caucasus and neighboring populations based on autosomal

360

The Caucasus as a Barrier to Human Migrations doi:10.1093/molbev/msr221

prehistoric and historic migrations to and from the Caucasus?

How and to what extent is the present-day remarkable linguistic diversity of the region reflectedin itsautosomaland genderspecific genetic variation?

Materials and Methods

DNA samples from all the subregions and major language
groups of the Caucasus (fig. 1A) were analyzed using whole
genome, Y chromosome, and mitochondrial DNA (mtDNA)
markers. DNA samples were obtained from unrelated male
volunteers after getting informed consent in accordance
with the guidelines of the ethical committees of the institutions involved. DNA was purified from blood by the phenol/chloroform extraction method. DNA concentrations
were determined by spectrometry (NanoDrop products,
Wilmington, DE).

Autosomal Analyses

Genetic Clustering Analysis

We used ADMIXTURE (Alexander et al. 2009) implementing
a structure-like (Pritchard et al. 2000) model-based maximum likelihood clustering algorithm to assess population
structure. We used PLINK software 1.05 (Purcell et al.
2007) to filter the combined data set to include only SNPs
on the 22 autosomal chromosomes with minor allele frequency .1% and genotyping success .97%. Because background linkage disequilibrium (LD) can affect both principal
component analysis (PCA) and structure-like analysis, we
thinned the marker set by excluding SNPs in strong LD (pairwise genotypic correlation r2 . 0.4) in a window of 200 SNPs
(sliding the window by 25 SNPs at a time). The final data set
consisted of 210,575 SNPs and 1119 individuals that were
used in subsequent analyses.

PCA and FST

Since the PCA method (Patterson et al. 2006) assumes that
markers are unlinked, we thinned our marker set for this analysis with PLINK software 1.05 (Purcell et al. 2007) according to
the same parameters used for genetic clustering analysis in
order to mitigate background LD. However, LD pruning
was carried out after the exclusion of the populations from

Africa (except Egyptians), East Asia, and Siberia and also Hazaras, Kurds, Uygurs, and Altaians, resulting in a final data set
of 212,398 markers and 848 individuals. PCA was carried out in
the smartpca program (Patterson et al. 2006) using outlier
removal procedure (5 outliers were removed, leaving 843 individuals). Pairwise genetic differentiation indices (FST values)
were also estimated using smartpca software based on the
same thinned marker set. The Gplots R package (http://
cran.r-project.org/web/packages/gplots/index.html) was used
to graphically represent genetic similarities between populations by color-coding pairwise FST values on a heatmap.

Geodesic Distances between Populations

For each pair of populations, we calculated geodesic distances in kilometers using distonearth R function (Banerjee
2005). Geographic coordinates for populations in the
HGDP data set are given as ranges of longitude and latitude
in Cann et al. (2002). Geographic coordinates for HGDP
populations were computed as a central point in a range
of longitude and latitude values.

Multiple Regression on Distance Matrices

We used multiple regression on distance matrices (MRM)
(Manly 1986; Smouse et al. 1986; Legendre et al. 1994;
Goslee and Urban 2007) to explore various explanatory variables (genetic distance, barriers to gene flow), predicting
the genetic distances between populations. In this method,
a single dependent distance matrix Y is considered as
a function of multiple independent distance matrices Xi
(independent variables), and the statistical significance
of regression coefficients for each independent variable
Xi is tested based on matrix permutations (Legendre
et al. 1994). The corresponding permutation procedure
is described in Legendre et al. (1994) and implemented
in the ecodist R package (Goslee and Urban 2007).
In order to test whether factors other than geographic
distance can explain the observed variation in genetic distances between populations and whether the contribution
of each variable is statistically significant, we considered
a matrix of pairwise FST distances between populations as
a response matrix and included explanatory variables into
the regression model either separately or in combinations,
defining them as putative barriers (supplementary table
S2; supplementary note 1, Supplementary Material online).

Y Chromosome Analyses
A total of 1952 samples from 24 populations from the Caucasus were analyzed for Y chromosome markers. The samples

data, with the clustering of populations approximating geography. The thick lines denote probable directions of movements of people,
interrupted between the North Caucasus and Eastern Europe. Despite the genetic discontinuity between the Caucasus and the East European
Plain, some admixture (denoted by the thin two-pointed arrow) has occurred, apparent from the several samples occupying the gap between
the regions. For population abbreviations see supplementary table S1 (Supplementary Material online). (D) Population structure inferred by
ADMIXTURE analysis of the autosomal data at K 5 7. Each individual is represented by a vertical (100%) stacked column of ancestry
probabilities in the seven constructed ancestral populations. Populations introduced in this study and analyzed together with data from Li et al.
(2008), Behar et al. (2010) and Rasmussen et al. (2010) are labeled in red. Language families of the Caucasus populations are also denoted: AA,
Abkhazian-Adyghe; ND, Nakh-Dagestanian; KV, Kartvelian; IE, Indo-European; TU, Turkic.

361

Downloaded from http://mbe.oxfordjournals.org/ by guest on June 24, 2013

Two hundred and four samples from this study were genotyped with the Illumina 610 K single nucleotide polymorphism (SNP) array and used for whole-genome analysis
together with 929 samples from the literature (Li et al.
2008; Behar et al. 2010; Rasmussen et al. 2010) (supplementary tableS1, Supplementary Material online).

MBE

Yunusbayev et al. doi:10.1093/molbev/msr221

were typed for 52 Y chromosome SNP markers, analyzed by

PCR/AFLP, PCR/RFLP, or PCR/sequencing methods. The haplogroup designation in this study follows the most recent YCC
nomenclature presented in Karafet et al. (2008). Genotyping
results are presented in supplementary table S3 (Supplementary Material online).
In addition, a subset of the samples was analyzed for 19 Y
chromosome STR markers. The phylogenetic network of
the data obtained was constructed with the program Network 4.5.0.0 (Fluxus-Engineering), using the median-joining
algorithm (supplementary fig. S1, Supplementary Material
online). Spatial frequency maps (supplementary figs. S2S4,
Supplementary Material online) were drawn with the program Surfer 8 (Golden Software Inc., Cold Spring Harbor,
NY). Coalescence ages (supplementary table S4, Supplementary Material online) were calculated according to
the ASD0 method (Zhivotovsky et al. 2004).
The haplogroups of 2262 mtDNA samples from 24 Caucasus populations were determined by typing HVSI, HVSII,
and coding region markers (supplementary table S5, Supplementary Material online), analyzed by PCR/RFLP or
PCR/sequencing methods. The nomenclature of the Global
Human Mitochondrial DNA Phylogenetic Tree (http://
www.phylotree.org) was used. Genotyping results are presented in supplementary table S6 (Supplementary Material
online).
Both Y chromosome and mtDNA haplogroup frequency
data of the Caucasus populations together with data of neighboring populations from the literature were used for PCA using the POPSTR software (http://harpending.humanevo.
utah.edu/popstr/). The methods used are described in greater
detail in supplementary note 1 (Supplementary Material
online).

Results and Discussion

Our autosomal data, in the form of a heatmap plot of FST
distances (fig. 1B) reveal three clusters of low genetic distance, encompassing geographically nearby populations:
the Near/Middle East, the Caucasus, and Europe. Note that
these clusters overlap partially in phylogeographically
informative ways. Europe is not as homogenous as the
other clusters, with the French Basques and Volga Basin
Turkic-speaking Chuvashes being clearly more distant from
their immediate neighbor populations, whereas geographically somewhat southern populations, from the Atlantic to
the Black Sea (French, Italians, Bulgarians, and Romanians),
exhibit particularly low interpopulation genetic distances.
The populations of the Caucasus, irrespective of their linguistic differences, show the lowest autosomal genetic distances to one another, followed closely by their distance to
the populations of the Near/Middle East, Turks in particular (fig. 1B). The Indo-European-speaking Armenians and
Ossetians follow the same pattern and do not show higher
genetic similarity to Indo-European-speaking populations
from Europe or the Near/Middle East. The smooth transition
from the Caucasus to Anatolia (Turks) and Iran and from the
362

latter to Syrians, Lebanese, Jordanians and further southward contrasts with the sharp border between the Caucasians and the Slavic, Finno-Ugric and Turkic-speaking
populations of the East European Plain. However, the same
heatmap plot reveals a range of low genetic distances starting from the populations of the Levant and Syria, extending
to the East European Plain. Importantly, this series proceeds
around the western flank of the Black Sea: from Turks to
Bulgarians to Romanians to Ukrainians to Russians, ending
with Mordvins, but does not extend through the Caucasus.
We suggest that this similarity gradient reflects most likely
a combination of several ancient human migrations, perhaps
also including elements of the spread of the Neolithic. A
previous study based on eight Alu insertion loci found the
Caucasus populations to exhibit high levels of betweenpopulation differentiation, with an average FST of 0.113,
a value which is almost as large as the FST of 0.157 for
worldwide populations (Nasidze et al. 2001). Our results,
based on genome-wide data, reveal instead that the populations of the Caucasus region show between population
differentiation (average FST 5 0.004) that is slightly lower
than that for the Near East (0.006) and Europe (0.006) and
are thus more consistent with the results of Bulayeva et al.
(2003).
Similarly to our FST analysis, PCA of the autosomal data
shows the within-region clustering of the Caucasian populations and reveals a noticeable gap between the Caucasus
and the East European Plain, contrasting the continuous
transition from the Near/Middle East to the Caucasus
(fig. 1C; supplementary fig. S5 and S6, Supplementary Material online). Geography rather than language-based clustering can also be observed in the PC analysis of Y
chromosome data (supplementary fig. S7, Supplementary
Material online). Indeed, we see a clustering of Georgians,
who belong to the Kartvelian language family, and Indo-Europeanspeaking Armenians from the South Caucasus, together with the peoples of the NW Caucasus, of which the
Karachays and Balkars speak Turkic, the Ossetians Indo-Iranian, and the remainder Abkhazian-Adyghe languages
(supplementary fig. S7, Supplementary Material online).
The Nakh and Dagestanian language speakers, however,
differ considerably both from each other and the rest of
the Caucasian populations due to the unique structure
of Y chromosome haplogroup frequencies in the NE Caucasus (supplementary note 2, Supplementary Material online), but this distinction is not supported by autosomal
or mtDNA data, consistent with the results of an earlier
study where highland Dagestanian populations were
found to have reduced Y chromosomal but not mitochondrial or autosomal genetic diversity (Marchani
et al. 2008). The Caucasian populations do not differ in
common mtDNA haplogroup frequencies to an extent
that would allow the discrimination of geographic subregions or language groups (supplementary fig. S8, Supplementary Material online).
We also analyzed the autosomal data of the Caucasus
and reference populations (Li et al. 2008; Behar et al.
2010; Rasmussen et al. 2010) using a structure-like

Downloaded from http://mbe.oxfordjournals.org/ by guest on June 24, 2013

mtDNA Analyses

MBE

The Caucasus as a Barrier to Human Migrations doi:10.1093/molbev/msr221

Equally importantly, these Y chromosome data also show

a genetic continuity between the NW Caucasus and South
Caucasus populations that sets them apart from the populations inhabiting the NE Caucasus (supplementary fig. S7,
Supplementary Material online).
Thus, several lines of evidence suggest a genetic discontinuity between the Caucasus and the East European Plain.
To test whether this pattern is statistically significant, we
used multiple regression on distance matrices (Legendre
et al. 1994; Goslee and Urban 2007) to model genetic distances based on geographic distances. We built a series of
distance matrices (see Materials and Methods for details),
incorporated additional factors known to increase genetic
dissimilarity as putative barriers (supplementary table S2;
supplementary note 1, Supplementary Material online),
and used multiple regression to estimate their relative importance in predicting genetic distances and statistical significance (Legendre et al. 1994; Goslee and Urban 2007).
Three of the putative barriers we tested first were geographicthe Caucasus barrier between North Caucasus and
Eastern Europe, the Balkans barrier between Anatolia and
Europe, and the South Asian barrier between South Asia
and the Near/Middle East. The South Asian barrier, a presumably nonexistent factor, was intentionally considered
as a control of the method, and only a slight increase in
the coefficient of determination (r2 increased by 0.1%)
was observed, most likely by chance since the regression
coefficient for this factor is not statistically significant (supplementary table S2, Supplementary Material online). The
other barriers tested separated populations that are known
isolates/outliers due to religion, language, or different originthe Jewish groups, Kuban Nogays, French Basques,
Druze, and Burushofrom their respective surrounding
populations. Geographic distance by itself explained only
43% (coefficient of determination r2 5 0.43) of the variation
in genetic distances between populations, measured by FST.
The putative Kuban Nogay, French Basque, Druze, Iraqi Jewish, Georgian Jewish, and Burusho barriers did not prove to be
statistically significant (supplementary table S2, Supplementary Material online). The largest improvement in the fit of
the observed FST distances between populations to the model
was achieved with the assumption of a Caucasus barrierr2
increased by 0.12 (supplementary table S2, Supplementary
Material online).
It should be noted that the effect of the Caucasus barrier is observed between the populations of the northern
flank of the High Caucasus Mountain Range and the East
European Plain, not at the mountain range itself, since the
populations of North and South Caucasus are autosomally
more similar to each other than either group is to Eastern
European populations. A putative post-LGM northward
progression of people (recolonization?) could have been
halted due to pre-existing inhabitation of the East European Plain, possibly at higher population densities than
in the mountains and may have been additionally inhibited by the Khvalynian transgression, a connection between the Black and Caspian Seas dated 1214,000 years
BP (Badyukova 2007; Svitoch 2009), which may have
363

Downloaded from http://mbe.oxfordjournals.org/ by guest on June 24, 2013

(Pritchard et al. 2000) clustering approach (Alexander et al.

2009). At K 5 7, the major ancestry component of the Caucasus populations (shown in blue) has comparable presence in the Near/Middle East but is almost absent
among the immediate northern neighbors of the Caucasusthe populations of the East European Plain (fig. 1D;
supplementary fig. S9, Supplementary Material online).
Similarly to the blue ancestry component, the green component is also ubiquitously present among the Caucasus
populations, irrespective of their linguistic affinities, but
at much lower frequencies than blue. The green component is most frequent in the Indus basin (Pakistan), extending to Central Asia and the Near/Middle East, but fading
away in Europe. Structure-like clustering cannot be readily
interpreted into (human) migrations. Thus, this pattern
might reflect shared ancestry or suggest a gene flow from
South Asia to the west and northwest. The ancestry component depicted with orange, ubiquitous in East Asia, is
almost absent in the South Caucasus and remains at very
low frequencies among the North Caucasian populations
as well, with an only slightly stronger presence among the
Turkic-speaking Balkars. Due to their distinct population
history (see below), the Kuban Nogays, living on the
northern slopes of the Caucasus, form an exception.
The only population in the Caucasus that shares the
other major Near/Middle Eastern, North, and East
African ancestry component, depicted in light blue
(fig. 1D), are the Armenians, indicating the introduction
of a genetic component novel to the region.
Nevertheless, the mountain range that divides the Caucasus into the North and South Caucasus has apparently
not been an impenetrable barrier for gene flows. This is
illustrated by the relative similarity of the ancestry component patterns of the Caucasus populations on either side of
the High Caucasus Mountain Range (fig. 1D). However, the
dark blue ancestry component, dominant among the Slavic-,
Turkic-, and Finnic-speaking East European Plain populations, reaches the North Caucasus (1020%), but just barely
(;5%) crosses the High Caucasus to the three linguistically
distinct South Caucasus populationsArmenians, Georgians, and Abkhazians (fig. 1D). Remarkably, the decrease
of Y chromosome haplogroup G and J1 frequencies toward
the Eastern European populations inhabiting the area adjacent to the North Caucasus, such as southern Russians and
Ukrainians (Balanovsky et al. 2008), forms an abrupt boundary (supplementary figs. S2 and S3, Supplementary Material
online), indicating that patrilineal gene flow from the Caucasus in the northern direction has been negligible. This sets
a frontier of genetic discontinuity not at the mountain range
itself but between the North Caucasus and East European
Plain. However, similarly to autosomal data, there is some
evidence of opposite gene flow from East Europe into the
populations of NW Caucasus, as demonstrated by the presence of Y chromosome haplogroups R1a1, including
the European-specific lineage R1a1-M458 (Underhill et al.
2010; Balanovsky et al. 2011) and I2a (supplementary
table S3, Supplementary Material online) and of certain sublineages of the mtDNA haplogroup H (Roostalu et al. 2007).

MBE

MBE

Yunusbayev et al. doi:10.1093/molbev/msr221

served as a natural barrier (see also supplementary note 2,

Supplementary Material online).
The Kuban Nogays and the Kara Nogays (fig. 1A) have
a special status among the Caucasian populations due to
their recent, late 18th to early 19th century arrival from the
Pontocaspian steppes (Kolga et al. 2001). It has been shown
earlier that the Nogays possess 40% of East Eurasian
mtDNA lineages (Bermisheva et al. 2004). Although the
Kara Nogays have more (;35%) typical eastern Y chromosome lineages than the Kuban Nogays (17%) (supplementary table S3, Supplementary Material online), it is perhaps
more interesting that both the Kuban Nogays and the Kara
Nogays preserve a certain combination of STR haplotypes
in the Y chromosome haplogroup C, the so-called Genghis
Khan modal haplotype (Zerjal et al. 2003). Because the historic Nogay Khan, a powerful late 13th century general of
the Golden Horde, was indeed a great-grandson of Genghis
Khan, such a coincidence is intriguing.

We conclude that irrespective of the Early UP presence of

anatomically modern humans both south and north of
the Caucasus (Mellars 2006; Adler et al. 2008; Krause
et al. 2010), the combined autosomal and gender-specific
genetic variation of the Caucasian populations testifies to
their predominantly southern, Near/Middle Eastern descent.
Y chromosomal variants under strong founder events, seen
in particular among populations inhabiting the northern
flank of the High Caucasus Mountain Range, appear to never
have expanded to the East European Plain, whereas the nomadic people of the latter, once settled down predominantly
on the northern slopes of the Caucasus, have likely preserved, to different extent, some of their earlier genetic heritage. In sum, though the Caucasus may well have served as
a corridor for numerous invasive expeditions in the past, this
has had only a minor influence on the largely sedentary core
populations of the region, characterized by much greater autosomal uniformity than that might be expected from a region of deep linguistic and cultural diversity. This suggests
that the core of the autosomal genetic structure of the Caucasian populations may have formed before its present-day
linguistic diversity, including the language families autochthonous for the region, arose.

Supplementary Material
Supplementary figures S1S9, tables S1S6, and notes 1
and 2 are available at Molecular Biology and Evolution online (http://www.mbe.oxfordjournals.org/).

Acknowledgments
We thank the individuals who provided DNA samples for
this study, and Mari Nelis, Georgi Hudjashov, and Viljo
Soo for conducting the autosomal genotyping. This work
was supported by the European Commission, DirectorateGeneral for Research (FP7 Ecogene grant number 205419
to R.V. and D.M.B.); the European Union Regional Development Fund through the Centre of Excellence in Genomics to
364

References
Adams JM, Faure H. 1997. Preliminary vegetation maps of the world
since the last glacial maximum: an aid to archaeological
understanding. J Archaeol Sci. 24:623647.
Adler DS, Bar-Yosef O, Belfer-Cohen A, Tushabramishvili N,
Boaretto E, Mercier N, Valladas H, Rink WJ. 2008. Dating the
demise: neandertal extinction and the establishment of modern
humans in the southern Caucasus. J Hum Evol. 55:817833.
Alexander DH, Novembre J, Lange K. 2009. Fast model-based estimation
of ancestry in unrelated individuals. Genome Res. 19:16551664.
Badyukova EN. 2007. Age of Khvalynian transgressions in the
Caspian Sea region. Oceanology 47:400405.
Balanovsky O, Dibirova K, Dybo A, et al. (23 co-authors). 2011.
Parallel evolution of genes and languages in the Caucasus region.
Mol Biol Evol. 28:29052920.
Balanovsky O, Rootsi S, Pshenichnov A, et al. (11 co-authors). 2008.
Two sources of the Russian patrilineal heritage in their Eurasian
context. Am J Hum Genet. 82:236250.
Banerjee S. 2005. On geodetic distance computations in spatial
modeling. Biometrics 61:617625.
Bar-Yosef O, Belfer-Cohen A, Adler DS. 2006. The implications of the
Middle-Upper Paleolithic chronological boundary in the Caucasus to Eurasian prehistory. Anthropologie XLIV:4960.
Behar DM, Yunusbayev B, Metspalu M, et al. (21 co-authors). 2010. The
genome-wide structure of the Jewish people. Nature 466:238242.
Bermisheva MA, Kutuev I, Korshunova T, Dubova NA, Villems R,
Khusnutdinova EK. 2004. Phylogeografic analysis of mitochondrial
DNA in the Nogays: the high level of mixture of maternal lineages
from Eastern and Western Eurasia. Mol Biol (Mosk). 38:617624.
Bulayeva K, Jorde LB, Ostler C, Watkins S, Bulayev O, Harpending H.
2003. Genetics and population history of Caucasus populations.
Hum Biol. 75:837853.
Cann HM, de Toma C, Cazes L, et al. (41 co-authors). 2002. A
human genome diversity cell line panel. Science 296:261262.
Comrie B. 2008. Linguistic diversity in the Caucasus. Annu Rev
Anthropol. 37:131143.
Goslee SC, Urban DL. 2007. The ecodist package for dissimilaritybased analysis of ecological data. J Stat Softw. 22:119.
Karafet TM, Mendez FL, Meilerman MB, Underhill PA, Zegura SL,
Hammer MF. 2008. New binary polymorphisms reshape and
increase resolution of the human Y chromosomal haplogroup
tree. Genome Res. 18:830838.
Kolga M, Tonurist I, Vaba L, Viikberg J. 2001. The red book of the
peoples of the Russian Empire. Tallinn (Estonia): NGO Red Book.
Krause J, Briggs AW, Kircher M, Maricic T, Zwyns N, Derevianko A,
Paabo S. 2010. A complete mtDNA genome of an early modern
human from Kostenki, Russia. Curr Biol. 20:231236.

Downloaded from http://mbe.oxfordjournals.org/ by guest on June 24, 2013

Conclusions

R.V.; the Estonian Ministry of Education and Research (Basic

Research grant numbers SF 0270177As08 to R.V. and SF
0270177Bs08 to E.M.); the Swedish Collegium for Advanced
Studies to R.V.; the Estonian Science Foundation (grant
numbers 7445 to S.R., 7858 to E.M., and 8973 to M.M.);
the Russian Academy of Sciences Program for Fundamental
Research Biodiversity and dynamics of gene pools to E.K.K.;
the Ministry of Education and Science of the Russian
Federation (state contracts P-325 and 02.740.11.07.01 to
E.K.K.); the Russian Foundation for Basic Research (grant
numbers 04-04-48678-a and 07-04-01016-a to E.K.K. and
08-06-97011 to I.K.); the President of the Russian Federation
(grant of state support for young Russian scientists number
MK-488.2006.4 to I.K.); and the Sorenson Molecular Genealogy Foundation to P.A.U.

The Caucasus as a Barrier to Human Migrations doi:10.1093/molbev/msr221

Pinhasi R, Gasparian B, Wilkinson K, et al. (13 co-authors). 2008.

Hovk 1 and the Middle and Upper Paleolithic of Armenia:
a preliminary framework. J Hum Evol. 55:803816.
Pritchard JK, Stephens M, Donnelly P. 2000. Inference of population
structure using multilocus genotype data. Genetics 155:945959.
Purcell S, Neale B, Todd-Brown K, et al. (11 co-authors). 2007. PLINK:
a tool set for whole-genome association and population-based
linkage analyses. Am J Hum Genet. 81:559575.
Rasmussen M, Li Y, Lindgreen S, et al. (52 co-authors). 2010. Ancient
human genome sequence of an extinct Palaeo-Eskimo. Nature
463:757762.
Roostalu U, Kutuev I, Loogvali E-L, Metspalu E, Tambets K, Reidla M,
Khusnutdinova EK, Usanga E, Kivisild T, Villems R. 2007. Origin
and expansion of haplogroup H, the dominant human
mitochondrial DNA lineage in West Eurasia: the Near Eastern
and Caucasian perspective. Mol Biol Evol. 24:436448.
Smouse PE, Long JC, Sokal RR. 1986. Multiple regression and
correlation extensions of the mantel test of matrix correspondence. Syst Zool. 35:627632.
Svitoch AA. 2009. Khvalynian transgression of the Caspian Sea was
not a result of water overflow from the Siberian Proglacial
lakes, nor a prototype of the Noachian flood. Quatern Int.
197:115125.
Tarasov PE, Volkova VS, Webb T 3rd, et al. (13 co-authors). 2000. Last
glacial maximum biomes reconstructed from pollen and plant
macrofossil data from northern Eurasia. J Biogeogr. 27:609620.
Underhill PA, Myres NM, Rootsi S, et al. (34 co-authors). 2010.
Separating the post-Glacial coancestry of European and Asian Y
chromosomes within haplogroup R1a. Eur J Hum Genet. 18:479484.
Zerjal T, Xue Y, Bertorelle G, et al. (23 co-authors). 2003. The genetic
legacy of the Mongols. Am J Hum Genet. 72:717721.
Zhivotovsky LA, Underhill PA, Cinnioglu C, et al. (17 co-authors).
2004. The effective mutation rate at Y chromosome short
tandem repeats, with application to human populationdivergence time. Am J Hum Genet. 74:5061.

365

Downloaded from http://mbe.oxfordjournals.org/ by guest on June 24, 2013

Legendre P, Lapointe F-J, Casgrain P. 1994. Modeling brain evolution

from behavior: a permutational regression approach. Evolution
48:14871499.
Li JZ, Absher DM, Tang H, et al. (11 co-authors). 2008. Worldwide
human relationships inferred from genome-wide patterns of
variation. Science 319:11001104.
Lieberman DE. 2007. Palaeoanthropology: homing in on early Homo.
Nature 449:291292.
Lordkipanidze D, Jashashvili T, Vekua A, et al. (18 co-authors). 2007.
Postcranial evidence from early Homo from Dmanisi, Georgia.
Nature 449:305310.
Manly BF. 1986. Randomization and regression methods for testing
for associations with geographical, environmental and biological
distances between populations. Res Popul Ecol. 28:201218.
Marchani EE, Watkins WS, Bulayeva K, Harpending HC, Jorde LB.
2008. Culture creates genetic structure in the Caucasus:
autosomal, mitochondrial, and Y-chromosomal variation in
Daghestan. BMC Genet. 9:47.
Mellars P. 2006. A new radiocarbon revolution and the dispersal of
modern humans in Eurasia. Nature 439:931935.
Myres NM, Rootsi S, Lin AA, et al. (20 co-authors). 2011. A major Ychromosome haplogroup R1b Holocene era founder effect in
Central and Western Europe. Eur J Hum Genet. 19:95101.
Nasidze I, Ling EY, Quinque D, et al. (17 co-authors). 2004.
Mitochondrial DNA and Y-chromosome variation in the
Caucasus. Ann Hum Genet. 68:205221.
Nasidze I, Risch GM, Robichaux M, Sherry ST, Batzer MA,
Stoneking M. 2001. Alu insertion polymorphisms and the
genetic structure of human populations from the Caucasus. Eur
J Hum Genet. 9:267272.
Nasidze I, Sarkisian T, Kerimov A, Stoneking M. 2003. Testing
hypotheses of language replacement in the Caucasus: evidence
from the Y-chromosome. Hum Genet. 112:255261.
Patterson N, Price AL, Reich D. 2006. Population structure and
eigenanalysis. PLoS Genet. 2:20742093.

MBE