Elementary Human Genetics
Population Genetics
Physical Anthropology
The Central Asian Gene Pool
The Qaraqalpaq Gene Pool
Discussion and Conclusions

Elementary Human Genetics

Every human is defined by his or her library of genetic material, copies of which are stored in every cell of the body apart from the red blood cells. Cells are classified as somatic, meaning body cells, or gametic, the cells involved in reproduction, namely the sperm and the egg or ovum. The overwhelming majority of human genetic material is located within the small nucleus at the heart of each somatic cell. It is commonly referred to as the human genome. Within the nucleus it is distributed between 46 separate chromosomes, two of which are known as the sex chromosomes. The latter occur in two forms, designated X and Y. Chromosomes are generally arranged in pairs - a female has 22 pairs of autosome chromosomes plus one pair of X chromosomes, while a male has a similar arrangement apart from having a mixed pair of X and Y sex chromosomes.

A strand of DNA
A neutron crystallography cross-sectional image of a chromosome, showing the double strand of DNA wound around a protein core.
Image courtesy of the US Department of Energy Genomics Program

A single chromosome consists of just one DNA macromolecule composed of two separate DNA strands, each of which contains a different but complementary sequence of four different nucleotide bases - adenine (A), thymine (T), cytasine (C), and guanine (G). The two strands are aligned in the form of a double helix held together by hydrogen bonds, adenine always linking with thymine and cytasine always linking with guanine. Each such linkage between strands is known as a base pair. The total human genome contains about 3 billion such base pairs. As such it is an incredibly long molecule that could be from 3 cm to 6 cm long were it possible to straighten it. In reality the double helix is coiled around a core of structural proteins and this is then supercoiled to create the chromosome, 23 pairs of which reside within a cell nucleus with a diameter of just 0.0005 cm.

A gene is a segment of the DNA nucleotide sequence within the chromosome that can be chemically read to make one specific protein. Each gene is located at a certain point along the DNA strand, known as its locus. The 22 autosome chromosome pairs vary in size from 263 million base pairs in chromosome 1 (the longest) down to about 47 million base pairs in chromosome 21 (the shortest - chromosome 22 is the second shortest with 50 million base pairs), equivalent to from 3,000 down to 300 genes. The two sex genes are also very different, X having about 140 million base pairs and expressing 1,100 genes, Y having only 23 million base pairs and expressing a mere 78 genes. The total number of genes in the human genome is around 30,000.

A set of human chromosomes
A complete set of 23 human homologous chromosome pairs
Image courtesy of the National Human Genome Research Institute, Maryland

Each specific pair of chromosomes have their own distinct characteristics and can be identified under the microscope after staining with a dye and observing the resulting banding. With one exception the chromosome pairs are called homologous because they have the same length and the same sequence of genes. For example the 9th pair always contain the genes for melanin production and for ABO blood type, while the 14th pair has two genes critical to the body's immune response. Even so the individual chromosomes within each matching pair are not identical since each one is inherited from each parent. A certain gene at a particular locus in one chromosome may differ from the corresponding gene in the other chromosome, one being dominant and the other recessive. The one exception relates to the male sex chromosomes, a combination of X and Y, which are not the same length and are therefore not homologous.

A set of human chromasomes
A set of male human chromosomes showing typical banding

Various forms of the same gene (or of some other DNA sequence within the chromosome) are known as alleles. Differences in DNA sequences at a specific chromosome locus are known as genetic polymorphisms. They can be categorized into various types, the most simple being the difference in just a single nucleotide - a single nucleotide polymorphism.

When a normal somatic cell divides and replicates, the 23 homologous chromosome pairs (the genome) are duplicated through a complex process known as mitosis. The two strands of DNA within each chromosome unravel and unzip themselves in order to replicate, eventually producing a pair of sister chromatids - two brand new copies of the original single chromosome joined together. However because the two chromosomes within each homologous pair are slightly different (one being inherited from each parent) the two sister chromatids are divided in two. The two halves of each sister chromatid are allocated to each daughter cell, thus replicating the original homologous chromosome pair. Such cells are called diploid because they contain two (slightly different) sets of genetic information.

The production of gametic cells involves a quite different process. Sperm and eggs are called haploid cells, meaning single, because they contain only one set of genetic information - 22 single unpaired chromosomes and one sex chromosome. They are formed through another complex process known as meiosis. It involves a deliberate reshuffling of the parental genome in order to increase the genetic diversity within the resulting sperm or egg cells and consequently among any resulting offspring. As before each chromosome pair is replicated in the form of a pair of sister chromatids. This time however, each half of each chromatid embraces its opposite neighbour in a process called synapsis. An average of two or three segments of maternal and paternal DNA are randomly exchanged between chromatids by means of molecular rearrangements called crossover and genetic recombination. The new chromatid halves are not paired with their matching partners but are all separated to create four separate haploid cells, each containing one copy of the full set of 23 chromosomes, and each having its own unique random mix of maternal and paternal DNA.

In the male adult this process forms four separate sperm cells, but in the female only one of the four cells becomes an ovum, the other three forming small polar bodies that progressively decay.

During fertilization the two haploid cells - the sperm and the ovum or egg - interact to form a diploid zygote (zyg meaning symetrically arranged in pairs). In fact the only contribution that the sperm makes to the zygote is its haploid nucleus containing its set of 23 chromosomes. The sex of the offspring is determined by the sex chromosome within the sperm, which can be either X (female) or Y (male). Clearly the sex chromosome within the ovum has to be X. The X and the Y chromosomes are very different, the Y being only one third the size of the X. During meiosis in the male, the X chromosome recombines and exchanges DNA with the Y only at its ends. Most of the Y chromosome is therefore unaffected by crossover and recombination. This section is known as the non-recombining part of the Y chromosome and it is passed down the male line from father to son relatively unchanged.

A human X and Y chromosome
Scanning electron micrograph of an X and Y chromosome
Image courtesy of Indigo Instruments, Canada

Not all of the material within the human cell resides inside the nucleus. Both egg and sperm cells contain small energy-producing organelles within the cytoplasm called mitochondria that have their own genetic material for making several essential mitochondrial proteins. However the DNA content is tiny in comparison with that in the cell nucleus - it consists of several rings of DNA totalling about 16,500 base pairs, equivalent to just 13 genes. The genetic material in the nucleus is about 300,000 times larger. When additional mitochondria are produced inside the cell, the mitochondrial DNA is replicated and copies are transferred to the new mitochondria. The reason why mitochondrial DNA, mtDNA for short, is important is because during fertilization virtually no mitochondria from the male cell enters the egg and those that do are tagged and destroyed. Consequently the offspring only inherit the female mitochondria. mtDNA is therefore inherited through the female line.

Population Genetics

Population genetics is a branch of mathematics that attempts to link changes in the overall history of a population to changes in its genetic structure, a population being a group of interbreeding individuals of the same species sharing a common geographical area. By analysing the nature and diversity of DNA within and between different populations we can gain insights into their separate evolution and the extent to which they are or are not related to each other. We can gain insights into a population's level of reproductive isolation, the minimum time since it was founded, how marriage partners were selected, past geographical expansions, migrations, and mixings.

The science is based upon the property of the DNA molecule to occasionally randomly mutate during replication, creating the possibility that the sequence of nucleotides in the DNA of one generation may differ slightly in the following generation. The consequence of this is that individuals within a homogenous population will in time develop different DNA sequences, the characteristic that we have already identified as genetic polymorphism. Because mutations are random, two identical but isolated populations will tend to change in different directions over time. This property is known as random genetic drift and its effect is greater in smaller populations.

To study genetic polymorphisms, geneticists look for specific genetic markers. These are clearly recognizable mutations in the DNA whose frequency of incidence varies widely across populations from different geographical areas. In reality the vast majority of human genetic sequences are identical, only around 0.1% of them being affected by polymorphisms.

There are several types of genetic marker. The simplest are single nucleotide polymorphisms (SNPs), mentioned above, where just one nucleotide has been replaced with another (for example A replaces T or C replaces G). SNPs in combination along a stretch of DNA are called haplotypes, shorthand for haploid genotypes. These have turned out to be valuable markers because they are genetically relatively stable and are found at differing frequencies in many populations. Some are obviously evolutionarily related to each other and can be classified into haplogroups (Hg). Another type of polymorphism is where short strands of DNA have been randomly inserted into the genetic DNA. This results in so-called biallelic polymorphism, since the strand is either present or absent. These are useful markers because the individuals that have the mutant insert can be traced back to a single common ancestor, while those who do not have the insert represent the original ancestral state . Biallelic polymorphisms can be assigned to certain haplotypes. A final type of marker is based upon microsatellites, very short sequences of nucleotides, such as GATA, that are repeated in tandem numerous times. A polymorphism occurs if the number of repetitions increases or decreases. Microsatellite polymorphisms, sometimes also called short-tandem-repeat polymorphisms, occur more frequently over time, providing a different tool to study the rate of genetic change against time.

Of course the whole purpose of sexual reproduction is to deliberately scramble the DNA from both parents in order to create a brand new set of chromosome pairs for their offspring that are not just copies of the parental chromosomes. Studies show that about 85% of genetic variation in autosomal sequences occurs within rather than between populations. However it is the genetic variation between populations that is of the greatest interest when we wish to study their history. Because of this, population geneticists look for more stable pieces of DNA that are not disrupted by reproduction. These are of two radically different types, namely the non-recombining part of the Y chromosome and the mitochondrial DNA or mtDNA. A much higher 40% of the variations in the Y chromosome and 30% of the variations in mtDNA are found between populations. Each provides a different perspective on the genetic evolution of a particular population.

Y Chromosome Polymorphisms

By definition the Y chromosome is only carried by the male line. Although smaller than the other chromosomes, the Y chromosome is still enormous compared to the mtDNA. The reason that it carries so few genes is because most of it is composed of "junk" DNA. As such it is relatively unaffected by natural selection. The non-recombining part of the Y chromosome is passed on from father to son with little change apart from the introduction of genetic polymorphisms as a result of random mutations. The only problem with using the Y chromosome to study inheritance has been the practical difficulty of identifying a wide range of polymorphisms within it, although the application of special HPLC techniques has overcome some of this limitation in recent years.

Y chromosome polymorphisms seem to be more affected by genetic drift and may give a better resolution between closely related populations where the time since their point of divergence has been relatively short.

mtDNA Polymorphisms

By contrast the mtDNA is carried by the female line. Although less than one thousandth the size of the DNA in the non-recombinant Y chromosome, polymorphisms are about 10 times more frequent in mtDNA than in autosome chromosomes.

Techniques and Applications

Population genetics is a highly statistical science and different numerical methods can be used to calculate the various properties of one or several populations. Our intention here is to cover the main analytical tools used in the published literature relating to Qaraqalpaq and the other Central Asian populations.

The genetic diversity of a population is the diversity of DNA sequences within its gene pool. It is calculated by a statistical method known as the analysis of molecular variance (AMOVA) in the DNA markers from that population. It is effectively a summation of the frequencies of individual polymorphisms found within the sample, mathematically normalized so that a diversity of 0 implies all the individuals in that population have identical DNA and a diversity of 1 implies that the DNA of every individual is different.

The genetic distance between two populations is a measure of the difference in their polymorphism frequencies. It is calculated statistically by comparing the pairwise differences between the markers identified for each population, to the pairwise differences within each of the two populations. This distance is a multi-dimensional not a linear measure. However it is normally illustrated graphically in two dimensions. New variables are identified by means of an angular transformation, the first two of which together account for the greatest proportion of the differences between the populations studied.

Another property that can be measured statistically is kinship - the extent to which members of a population are related to each other as a result of a common ancestor. Mathematically, a kinship coefficient is the probability that a randomly sampled sequence of DNA from a randomly selected locus is identical across all members of the same population. A coefficient of 1 implies everyone in the group is related while a coefficient of 0 implies no kinship at all.

By making assumptions about the manner in which genetic mutations occur and their frequency over time it is possible to work backwards and estimate how many generations (and therefore years) have elapsed from the most recent common ancestor, the individual to whom all the current members of the population are related by descent. This individual is not necessarily the founder of the population. For example if we follow the descent of the Y chromosome, this can only be passed down the male line from father to son. If a male has no sons his non-combining Y chromosome DNA is eliminated from his population for ever more. Over time, therefore, the Y chromosomes of the populations ancestors will be progressively lost. There may well have been ancestors older than the most recent common ancestor, even though we can find no signs for those ancestors in the Y chromosome DNA of the current population.

A similar situation arises with mtDNA in the female half of the population because some women do not have daughters.

Physical Anthropology

In 1977 the American anthropologist Gordon T. Bowles published an analysis of the anthropometric characteristics of 519 different populations from across Asia, including the Qaraqalpaqs and two regional groups of Uzbeks. Populations were characterized by 9 standard measurements, including stature and various dimensions of the head and face. A multivariate analysis was used to separate the different populations by their physical features.

Bowles categorized the populations across four regions of Asia (West, North, East, and South) into 19 geographical groups. He then analysed the biological distances between the populations within each group to identify clusters of biologically similar peoples.

Central Asia was divided into Group XVII encompassing Mongolia, Singkiang, and Kazakhstan and Group XVIII encompassing Turkestan and Tajikistan. Each Group was found to contain three population clusters:

Anthropological Cluster Analysis of Central Asia
    Group         Cluster     Regional Populations
XVII 1Eastern Qazaqs
Alai Valley Kyrgyz
Nivkhi (Gilyak)
2Aksu Rayon Uighur
Alma Ata Uighur
Steppe Qazaqs
Alma Ata Qazaqs
Ferghana Kyrgyz
Kalmuk Sarts
Khalka Mongols
Eastern Mongols
T'ien Shan Kyrgyz
Sayak Kyrgyz
Nanay (Goldi)
XVIII 1Qaraqalpaqs
Samarkand Uzbeks
Tashkent Uzbeks
2Pamiri Tajiks
Yagnobi Tajiks
Karategin Takijs
Surkhan Tajiks
Kipchak Kyrgyz
3Yomut Turkmen
Merv Turkmen
Total Turkmen            

Within geographical Group XVIII, the Qaraqalpaqs clustered with the Uzbeks of Tashkent and the Uzbeks of Samarkand. The members of this first cluster were much more heterogeneous than the other two clusters of neighbouring peoples. Conversely the Turkmen cluster had the lowest variance of any of the clusters in the North Asia region, showing that different Turkmen populations are closely related.

The results of this study were re-presented by Cavalli-Sforza in a more readily understandable graphical form. The coordinates used are artificial mathematical transformations of the original 9 morphological measurements, designed to identify the distances between different populations in a simple two-dimensional format. The first two principal coordinates identify a clear division between the Uzbek/Qaraqalpaqs, and the Turkmen and Iranians, but show similarities between the Uzbek/Qaraqalpaqs and the Tajiks, and also with the western Siberians. Though not so close there are some similarities between the Uzbek/Qaraqalpaqs and the Qazaqs, Kyrgyz, and Mongols:

Physical Anthropology of Asia
Physical Anthropology of Asia redrawn by David Richardson after Bowles 1977
First and Second Principal Coordinates

The second and third principal coordinates maintains the similarity between Uzbek/Qaraqalpaqs and Tajiks but emphasizes the more eastern features of the Qazaqs, Kyrgyz, and Mongols:

Physical Anthropology of Asia
Physical Anthropology of Asia redrawn by David Richardson after Bowles 1977
Second and Third Principal Coordinates

The basic average morphology of the Uzbeks and Qaraqalpaqs shows them to be of medium stature, with heads that have an average length but an above average breadth compared to the other populations of Asia. Their faces are broad and are of maximum height. Their noses are of average width but have the maximum length found in Asia.

Qazaqs have the same stature but have longer and broader heads. Their faces are shorter but broader, having the maximum breadth found in Asia, while their noses too are shorter and slightly broader.

Some of these differences in features were noted by some of the early Russian visitors, such as N. N. Karazin, who observed the differences between the Qaraqalpaqs and the Qazaqs (who at that time were called Kirghiz) when he first entered the northern Aral delta:

"In terms of type, the Karakalpak people themselves differ noticeably from the Kirghizs: flattened Mongolian noses are already a rarity here, cheek-bones do not stand out so, beards and eyebrows are considerably thicker - there is a noticeably strong predominance of the Turkish race."

The Central Asian Gene Pool

Western researchers tended to under represent Central Asian populations in many of the earlier studies of population genetics.

Cavalli-Sforza, Menozzi, and Piazza, 1994

In 1994 Cavalli-Sforza and two of his colleagues published a landmark study of the worldwide geographic distribution of human genes. In order to make global comparisons the study was forced to rely upon the most commonly available genetic markers, and analysed classical polymorphisms based on blood groups, plasma proteins, and red cell enzymes. Sadly no information was included for Qaraqalpaqs or Qazaqs.

Results were analysed continent by continent. The results for the different populations of Asia grouped the Uzbeks, Turkmen, and western Turks into a central cluster, located on the borderline between the Caucasian populations of the west and south and the populations of Northeast Asia and East Asia:

Principal component analysis of Asian populations
Principal Component Analysis of Asian Populations
Redrawn by David Richardson after Cavelli-Sforza et al, 1994

Comas, Calafell, Pérez-Lezaun et al, 1998

In 1993-94 another Italian team collected DNA samples from four different populations close to the Altai: Qazaq highlanders living close to Almaty, Uighur lowlanders in the same region, and two Kyrgyz communities - one in the southern highlands, the other in the northern lowlands of Kyrgyzstan.

The data was used in two studies, both published in 1998. In the first, by Comas et al, mtDNA polymorphisms in these four communities were compared with other Eurasian populations in the west (Europe, Middle East, and Turkey), centre (the Altai) and the east (Mongolia, China, and Korea). The four Central Asian populations all showed high levels of sequence diversity - in some cases the highest in Eurasia. At the same time they were tightly clustered together, almost exactly halfway between the western and the eastern populations, the exception being that the Mongolians occupied a position close to this central cluster. The results suggested that the Central Asian gene pool was an admixture of the western and eastern gene pools, formed after the western and eastern Eurasians had diverged. The authors suggested that this diversity had possibly been enhanced by human interaction along the Silk Road.

In the second, by Pérez-Lezaun et al, short-tandem-repeat polymorphisms in the Y chromosome were analysed for the four Central Asian populations alone. Each of the four was found to be highly heterogeneous yet very different from the other three, the latter finding appearing to contradict the mtDNA results. However the two highland groups had less genetic diversity because each had very high frequencies for one specific polymorphism:

Y chromosome
Y chromosome haplotype frequencies, with labels given to those shared by more than one population
From Pérez-Lezaun et al, 1998.

The researchers resolved the apparent contradiction between the two studies in terms of different migration patterns for men and women. All four groups practised a combination of exogamy and patrilocal marriage - in other words couples within the same clan could not marry and brides always moved from their own village to the village of the groom. Consequently the males, and their genes, were isolated and localized, while the females were mobile and there were more similarities in their genes. The high incidence of a single marker in each highland community was presumed to be a founder's effect, supported by evidence that the highland Qazaq community had only been established by lowland Qazaqs a few hundred years ago.

Zerjal, Spencer Wells, Yuldasheva, Ruzibakiev, and Tyler-Smith, 2002

In 2002 a joint Oxford University/Imperial Cancer Research Fund study was published, analysing Y chromosome polymorphisms in 15 different Central Asian populations, from the Caucasus to Mongolia. It included Uzbeks from the eastern viloyat of Kashkadarya, Qazaqs and Uighurs from eastern Kazakhstan, Tajiks, and Kyrgyz. Blood samples had been taken from 408 men, living mainly in villages, between 1993 and 1995. In the laboratory the Y chromosomes were initially typed with binary markers to identify 13 haplogroups. Following this, microsatellite variations were typed in order to define more detailed haplotypes.

Haplogroup frequencies were calculated for each population and were illustrated by means of the following chart:

Haplogroup frequencies across Central Asia.
Haplogroup frequencies across Central Asia
From Zerjal et al, 2002.

Many of the same haplogroups occurred across the 5,000 km expanse of Central Asia, although with large variations in frequency and with no obvious overall pattern. Haplogroups 1, 2, 3, 9, and 26 accounted for about 70% of the total sample.

Haplogroups (Hg) 1 and 3 were common in almost all populations, but the highest frequencies of Hg1 were found in Turkmen and Armenians, while the highest frequencies of Hg3 were found in Kyrgyz and Tajiks. Hg3 was more frequent in the eastern populations, but was only present at 3% in the Qazaqs. Hg3 is the equivalent of M17, which seems to originate from Russia and the Ukraine, a region not covered by this survey - see Spencer Wells et al, 2001 below. Hg9 was very frequent in the Middle East and declined in importance across Central Asia from west to east. However some eastern populations had a higher frequency - the Uzbeks, Uighurs, and Dungans.

Hg10 and its derivative Hg36 showed the opposite pattern, together accounting for 54% of haplogroups for the Mongolians and 73% for the Qazaqs. Hg26, which is most frequently found in Southeast Asia, occurs with the highest frequencies among the Dungans (26%), Uighurs (15%), Mongolians (13%), and Qazaqs (13%) in eastern Central Asia. Hg12 and Hg 16 are widespread in Siberia and northern Eurasia but are rare in Central Asia except for the Turkmen and Mongolians. Hg21 was restricted to the Caucasus region.

The most obvious observation is that virtually each population is quite distinct. As an example, the Uzbeks are quite different from the Turkmen, Qazaqs, or Mongolians. Only two populations, the Kyrgyz from central Kyrgyzstan and the Tajiks from Pendjikent, show any similarities.

The researchers measured the genetic diversity of each population using both haplogroup and microsatellite frequencies. Within Central Asia, the Uzbeks, Uighurs, Dungans, and Mongolians exhibited high genetic diversity, while the Qazaqs, Kyrgyz, Tajiks, and Turkmen showed low genetic diversity. These differences were explored by examining the haplotype variation within each haplogroup for each population. Among the Uzbeks, for example, many different haplotypes are widely dispersed across all chromosomes. Among the Qazaqs, however, the majority of the haplotypes are clustered together and many chromosomes share the same or related haplotypes.

Low diversity coupled with high frequencies of population-specific haplotype clusters are typical of populations that have experienced a bottleneck or a founder event. The most recent common ancestor of the Tajik population was estimated to date from the early part of the 1st millennium AD, while the most recent common ancestors of the Qazaq and Kyrgyz populations were placed in the period 1200 to 1500 AD. The authors suggested that bottlenecks might be a feature of societies like the Qazaqs and Kyrgyz with small, widely dispersed nomadic groups, especially if they had suffered massacres during the Mongol invasion. Of course these calculations have broad confidence intervals and must be interpreted with caution.

Microsatellite haplotype frequencies were used to investigate the genetic distances among the separate populations. The best two-dimentional fit produces a picture with no signs of general clustering on the basis of either geography or linguistics:

Genetic distances based on micosatellite haplotypes.
Genetic distances based on micosatellite haplotypes
From Zerjal et al, 2002.

The Kyrgyz (ethnically Turkic) do cluster next to the Tajiks (supposedly of Indo-Iranian origin), but both are well separated from the neighbouring Qazaqs. The Turkmen, Qazaqs, and Georgians tend to be isolated from the other groups, leaving the Uzbeks in a somewhat central position, clustered with the Uighurs and Dungans.

The authors attempted to interpret the results of their study in terms of the known history of the region. The apparently underlying graduation in haplogroup frequencies from west to east was put down to the eastward agricultural expansion out of the Middle East during the Neolithic, some of the haplogroup markers involved being more recent than the Palaeolithic. Meanwhile Hg3 (equivalent to M17 and Eu19), which is widespread in Central Asia, was attributed to the migration of the pastoral Indo-Iranian "kurgan culture" eastwards from the Ukraine in the late 3rd/early 2nd millennium BC. The mountainous Caucasus region seems to have been bypassed by this migration, which seems to have extended across Central Asia as far as the borders of Siberia and China.

Later events also appear to have left their mark. The presence of a high number of low-frequency haplotypes in Central Asian populations was associated with the spread of Middle Eastern genes, either through merchants associated with the early Silk Route or the later spread of Islam. Uighurs and Dungans show a relatively high Middle Eastern admixture, including higher frequencies of Hg9, which might indicate their ancestors migrated from the Middle East to China before moving into Central Asia.

High frequencies of Hg10 and its derivative Hg36 are found in the majority of Altaic-speaking populations, especially the Qazaqs, but also the Uzbeks and Kyrgyz. Yet its contribution west of Uzbekistan is low or undetectable. This feature is associated with the progressive migrations of nomadic groups from the east, from the Hsiung-Nu, to the Huns, the Turks, and the Mongols. Of course Central Asians have not only absorbed immigrants from elsewhere but have undergone expansions, colonizations and migrations of their own, contributing their DNA to surrounding populations. Hg1, the equivalent of M45 and its derivative markers, is believed to have originated in Central Asia and is found throughout the Caucasus and in Mongolia.

The Qaraqalpaq Gene Pool

Spencer Wells et al, 2001

The first examination of Qaraqalpaq DNA appeared as part of a widespread study of Eurasian Y chromosome diversity published by Spencer Wells et al in 2001. It included samples from 49 different Eurasian groups, ranging from western Europe, Russia, the Middle East, the Caucasus, Central Asia, South India, Siberia, and East Asia. Data on 12 other groups was taken from the literature. In addition to the Qaraqalpaqs, the Central Asian category included seven separate Uzbek populations selected from Ferghana to Khorezm, along with Turkmen from Ashgabat, Tajiks from Samarkand, and Qazaqs and Uighurs from Almaty. The study used biallelic markers that were then assigned to 23 different haplotypes. To illustrate the results the latter were condensed into 7 evolutionary-related groups.

The study found that the Uzbek, Qaraqalpaq, and Tajik populations had the highest haplotype diversity in Eurasia, the Qaraqalpaqs having the third highest diversity of all 49 groups. The Qazaqs and Kyrgyz had a significantly lower diversity.

This diversity is obvious from the chart comparing haplotype frequencies across Eurasia:

Y chromosome haplotypes
Distribution of Y chromosome haplotype lineages across various Eurasian populations
From Spencer Wells et al, 2001.

Uzbeks have a fairly balanced haplotype profile, while populations in the extreme west and east are dominated by one specific haplotype lineage - the M173 lineage in the extreme west and the M9 lineage in the extreme east and Siberia.

The Qaraqalpaqs are remarkably similar to the Uzbeks:

Distribution of Y chromosome haplotype lineages in Uzbeks and Qaraqalpaqs
Distribution of Y chromosome haplotype lineages in Uzbeks and Qaraqalpaqs
From Spencer Wells et al, 2001.

the main differences being that Qaraqalpaqs have a higher frequency of M9 and M130 and a lower frequency of M17 and M89 haplotype lineages. M9 is strongly linked to Chinese and other far-eastern peoples, while M130 is associated with Mongolians and Qazaqs. On the other hand, M17 is strong in Russia, the Ukraine, the Czech and Slovak Republics as well as in Kyrgyz populations, while M89 has a higher frequency in the west. It seems that compared to Uzbeks, the Qaraqalpaq gene pool has a somewhat higher frequency of haplotypes that are associated with eastern as opposed to western Eurasian populations.

In fact the differences between Qaraqalpaqs and Uzbeks are no more pronounced than between the Uzbeks themselves. Haplotype frequencies for the Qaraqalpaqs tend to be within the ranges measured across the different Uzbek populations:

Comparison of Qaraqalpaq haplotype lineage frequencies to other ethnic groups in Central Asia
  YAP    M130  M89     M9      M45   M173   M17     Total  
Qaraqalpaqs44 -2218257 91899
Uzbeks Average366 41227148 1124100
Uzbeks Range366 0 - 77-1819-345-214-11 6-2113-30100
Uzbeks Khorezm70 710191310 93098
Uzbeks Bukhara58 2934194 728103
Uzbeks Samarkand45 21831159 111399
Uzbeks Kashkadarya19 -1633511 2116102
Uzbeks Surkhandarya68 61219215 62998
Uzbeks Tashkent43 5730127 1228101
Uzbeks Ferghana63 613221610 1322102
Turkmen Ashgabat30 0-301313 377100
Qazaqs54 2662138 64101
Uighurs41 -1534217 02299
Kyrgyz52 -146122 26399
Tajiks Samarkand40 10823188 1025102
Tatars Crimea22 5937100 932102
Tatars Kazan38 3042245 324101
Iranians Tehran24 2106740 44100
Siberians Tuva42 01704917 21499
Mongolian24 4596250 0498

Statistically Qaraqalpaqs are genetically closest to the Uzbeks from Ferghana, followed by those from Surkhandarya, Samarkand, and finally Khorezm. They are furthest from the Uzbeks of Bukhara, Tashkent, and Kashkadarya.

These results also show the distance between the Qaraqalpaqs and the other peoples of Central Asia and its neighbouring regions. Next to the Uzbeks, the Qaraqalpaqs are genetically closest to the Tatars and Uighurs. However they are quite distant from the Turkmen, Qazaqs, Kyrgyz, Siberians, and Iranians.

The researchers produced a "neighbour-joining" tree, which clustered the studied populations into eight categories according to the genetic distances between them. The Qaraqalpaqs were classified into cluster VIII along with Uzbeks, Tatars, and Uighurs - the populations with the highest genetic diversity. They appear sandwiched between the peoples of Russian and the Ukraine and the Mongolians and Qazaqs.

Genetic tree of Eurasian populations
Neighbour-joining tree of 61 Eurasian Populations
Qaraqalpaqs are included in cluster VIII along with Uzbeks, Tatars, and Uighurs
From Spencer Wells et al, 2001.

Spencer Wells and his colleagues did not attempt to explain why the Qaraqalpaq gene pool is similar to Uzbek but is different from the Qazaq, a surprising finding given that the Qaraqalpaqs lived in the same region as the Qazaqs of the Lesser Horde before migrating into Khorezm. Instead they suggested that the high diversity in Central Asia might indicate that its population is among the oldest in Eurasia. M45 is the ancestor of haplotypes M173, the predominant group found in Western Europe, and is thought to have arisen in Central Asia about 40,000 years ago. M173 occurred about 30,000 years old, just as modern humans began their migration from Central Asia into Europe during the Upper Palaeolithic. M17 (also known as the Eu19 lineage) has its origins in eastern Europe and the Ukraine and may have been initially introduced into Central Asia following the last Ice Age and re-introduced later by the south-eastern migration of the Indo-Iranian "kurgan" culture.

Comas et al, 2004

At the beginning of 2004 a complementary study was published by David Comas, based on the analysis of mtDNA haplogroups from 12 Central Asian and neighbouring populations, including Qaraqalpaqs, Uzbeks, and Qazaqs. Sample size was only 20, dropping to 16 for Dungans and Uighurs, so that errors in the results for individual populations could be high.

The study reconfirmed the high genetic diversity within Central Asian populations. However a high proportion of sequences originated elsewhere, suggesting that the region had experienced "intense gene flow" in the past.

The haplogroups were divided into three types according to their origins: West Eurasian, East Asian, and India. Populations showed a graduation from the west to the east with the Qaraqalpaqs occupying the middle ground, with half of their haplogroups having a western origin and the other half having an eastern origin. Uzbek populations contained a small Indian component.
Mixture of western and eastern mtDNA haplogroups across Central Asia
PopulationWest Eurasian   East Asian           Indian                Total       
Crimean Tatars10000100
Khorezmian Uzbeks542817100
Bukharan Arabs70300100

The researchers found that two of the haplogroups of East Asian origin (D4c and G2a) not only occurred at higher frequencies in Central Asia than in neighbouring populations but appeared in many related but diverse forms. These may have originated as founder mutations some 25,000 to 30,000 years ago, expanded as a result of genetic drift and subsequently become dispersed into the neighbouring populations. Their incidence was highest in the Qazaqs, and second highest in the Turkmen and Qaraqalpaqs.

The majority of the other lineages separate into two types with either a western or an eastern origin. They do not overlap, suggesting that they were already differentiated before they came together in Central Asia. Furthermore the eastern group contains both south-eastern and north-eastern components. One explanation for their admixture in Central Asia is that the region was originally inhabited by Western people, who were then partially replaced by the arrival of Eastern people. There is genetic evidence from archaeological sites in eastern China of a drastic shift, between 2,500 and 2,000 years ago, from a European-like population to the present-day East Asian population.

The presence of ancient Central Asian sequences suggests it is more likely that the people of Central Asia are a mixture of two differentiated groups of peoples who originated in west and east Eurasia respectively.

Chaix and Heyer et al, 2004

The most interesting study of Qaraqalpaq DNA so far was published by a team of French workers in the autumn of 2004. It was based on blood samples taken during two separate expeditions to Qaraqalpaqstan in 2001 and 2002, organized with the assistance of IFEAC, the Institut Français d'Etudes sur l'Asie Centrale, based in Tashkent. The samples consisted of males belonging to five different ethnic groups: Qon'ırat Qaraqalpaqs (sample size 53), On To'rt Urıw Qaraqalpaqs (53), Qazaqs (50), Khorezmian Uzbeks (40), and Turkmen (51). The study was based on the analysis of Y chromosome haplotypes from DNA extracted from white blood cells. In addition to providing samples for DNA analysis, participants were also interviewed to gather information on their paternal lineages and tribal and clan affiliations.

Unfortunately the published results only focused on the genetic relationships between the tribes, clans and lineages of these five ethnic groups. However before reviewing these important findings it is worth looking at the more general aspects that emerged from the five samples. These were summarized by Professor Evelyne Heyer and Dr R Chaix at a workshop on languages and genes held in France in 2005, where the results from Qaraqalpaqstan were compared with the results from similar expeditions to Kyrgyzstan, the Bukhara, Samarkand, and Ferghana Valley regions of Uzbekistan, and Tajikistan as well as with some results published by other research teams. In some cases comparisons were limited by the fact that the genetic analysis of samples from different regions was not always done according to the same protocols.

The first outcome was the reconfirmation of the high genetic diversity among Qaraqalpaqs and Uzbeks:
Y Chromosome Diversity across Central Asia
PopulationRegionSample Size   Diversity   
Qaraqalpaq Qon'ıratQaraqalpaqstan540.97
Qaraqalpaq On To'rt UrıwQaraqalpaqstan540.89
Uzbek KhorezmQaraqalpaqstan540.97
Qazaq KhorezmQaraqalpaqstan490.84
Turkmen KhorezmQaraqalpaqstan510.84
Uzbek KashkadaryaUzbekistan281.00
Uighur AlmatyKazakhstan330.98
Qazaq AlmatyKazakhstan140.86
Turkmen AshgabatTurkmenistan210.94
Tajik PendjikentTajikistan220.87
Tajik KamangaronFerghana Valley300.98
Tajik RichtanFerghana Valley290.98
Kyrgyz CentralKyrgyzstan370.91
Kyrgyz AndijanUzbek Ferghana Valley460.82
Kyrgyz JankatalabUzbek Ferghana Valley200.78
Kyrgyz DobolooUzbek Ferghana Valley220.70

The high diversities found in Uighur and Tajik communities also agreed with earlier findings. Qon'ırat Qaraqalpaqs had somewhat greater genetic diversity than On To'rt Urıw Qaraqalpaqs. Some of these figures are extremely high. A diversity of zero implies a population where every individual is identical. A diversity of one implies the opposite, the haplotypes of every individual being different.

The second more important finding concerned the Y chromosome genetic distances among different Central Asian populations. As usual this was presented in two dimensions:

Y chromosome distances between Central Asian populations
Genetic distances between ethnic populations in Qaraqalpaqstan and the Ferghana Valley
From Chaix and Heyer et al, 2004.

The researchers concluded that Y chromosome genetic distances were strongly correlated to geographic distances. Not only are Qon'ırat and On To'rt Urıw populations genetically close, both are also close to the neighbouring Khorezmian Uzbeks. Together they give the appearance of a single population that has only relatively recently fragmented into three separate groups. Clearly this situation is mirrored with the two Tajik populations living in the Ferghana Valley and also with two of the three Kyrgyz populations from the same region. Although close to the local Uzbeks, the two Qaraqalpaq populations have a slight bias towards the local Qazaqs.

The study of the Y chromosome was repeated for the mitichondrial DNA, to provide a similar picture for the female half of the same populations. The results were compared to other studies conducted on other groups of Central Asians. We have redrawn the chart showing genetic distances among populations, categorizing different ethnic groups by colour to facilitate comparisons:

mtDNA distances between Central Asian populations
Genetic distances among ethnic populations in Central Asia
Based on mitochondrial DNA polymorphisms
From Heyer, 2005.

The French team concluded that, in this case, genetic distances were not related to either geographical distances or to linguistics. However this is not entirely true because there is some general clustering among populations of the same ethnic group, although by no means as strong as that observed from the Y chromosome data. The three Qaraqalpaq populations highlighted in red consist of the On To'rt Urıw (far right), the Qon'ırat (centre), and the Qaraqalpaq sample used in the Comas 2004 study (left). The Uzbeks are shown in green and those from Qaraqalpaqstan are the second from the extreme left, the latter being the Uzbeks from Samarkand. A nearby group of Uzbeks from Urgench in Khorezm viloyati appear extreme left. There is some relationship between the mtDNA of the Qaraqalpaq and Uzbek populations of the Aral delta therefore, but it is much weaker than the relationship between their Y chromosome DNA. On the other hand the Qazaqs of Qaraqalpaqstan, the uppermost yellow square, are very closely related to the Qaraqalpaq Qon'ırat according to their mtDNA.

These results are similar to those that emerged from the Italian studies of Qazaq, Uighur, and Kyrgyz Y chromosome and mitochondrial DNA. Ethnic Turkic populations are generally exogamous. Consequently the male DNA is relatively isolated and immobile because men traditionally stay in the same village from birth until death. They had to select their wives from other geographic regions and sometimes married women from other ethnic groups. The female DNA within these groups is consequently more diversified. The results suggest that in the delta, some Qon'ırat men have married Qazaq women and/or some Qazaq men have married Qon'ırat women.

Let us now turn to the primary focus of the Chaix and Heyer paper. Are the tribes and clans of the Qaraqalpaqs and other ethnic groups living within the Aral delta linked by kinship? Y chromosome polymorphisms were analysed for each separate lineage, clan, tribe, and ethnic group using single tandem repeats. The resulting haplotypes were used to calculate a kinship coefficient at each respective level.

Within the two Qaraqalpaq samples the Qon'ırat were all Shu'llik and came from several clans, only three of which permitted the computation of kinship: the Qoldawlı, Qıyat, and Ashamaylı clans. However none of these clans had recognized lineages. The Khorezmian Uzbeks have also long ago abandoned their tradition of preserving genealogical lineages.

The On To'rt Urıw were composed of four tribes, four clans, and four lineages:

                    - Qıtay tribe
                    - Qıpshaq tribe, Basar clan
                    - Keneges tribe, Omır and No'kis clans
                    - Man'g'ıt tribe, Qarasıraq clan

The Qazaq and the Turkmen groups were also structured along tribal, clan, and lineage lines.

The results of the study showed that lineages, where they were still maintained, exhibited high levels of kinship, the On To'rt Urıw having by far the highest. People belonging to the same lineage were therefore significantly more related to each other than people selected at random in the overall global population. Put another way, they share a common ancestor who is far more recent than the common ancestor for the population as a whole:

Mean genetic kinship coefficient for 5 ethnic populations
Kinship coefficients for five different ethnic populations, including the Qon'ırat and the On To'rt Urıw.
From Chaix and Heyer et al, 2004.

The kinship coefficients at the clan level were lower, but were still significant in three groups - the Qaraqalpaq Qon'ırat, the Qazaqs, and the Turkmen. However for the Qaraqalpaq On To'rt Urıw and the Uzbeks, men from the same clan were only fractionally more related to each other than were men selected randomly from the population at large. When we reach the tribal level we find that the men in all five ethnic groups show no genetic kinship whatsoever.

In these societies the male members of some but not all tribal clans are partially related to varying degrees, in the sense that they are the descendants of a common male ancestor. Depending on the clan concerned this kinship can be strong, weak, or non-existant. However the members of different clans within the same tribe show no such interrelationship at all. In other words, tribes are conglomerations of clans that have no genetic links with each other apart from those occurring between randomly chosen populations. It suggests that such tribes were formed politically, as confederations of unrelated clans, and not organically as a result of the expansion and sub-division of an initially genetically homogenous extended family group.

By assuming a constant rate of genetic mutation over time and a generation time of 30 years, the researchers were able to calculate the number of generations (and therefore years) that have elapsed since the existence of the single common ancestor. This was essentially the minimum age of the descent group and was computed for each lineage and clan. However the estimated ages computed were very high. For example, the age of the Qon'ırat clans was estimated at about 460 generations or 14,000 years (late Ice Age), while the age of the On To'rt Urıw lineages was estimated at around 200 generations or 6,000 years (early Neolithic). Clearly these results are ridiculous. The explanation is that each group included immigrants or outsiders who were clearly unrelated to the core population.

The calculation was therefore modified, restricting the sample to those individuals who belonged to the modal haplogroup of the descent group. This excluded about 17% of the men in the initial sample. Results were excluded for those descent groups that contained less than three individuals:

Descent GroupPopulationNumber of
Descent Groups
Number of
Age in years95% Confidence
Clan      Qaraqalpaq
2   351,058454 - 3,704
Clan      Qazaq1   20    595255 - 2,083
Clan      Turkmen4102 3,0511,307 - 10,677
On To'rt Urıw
4   13    397170 - 1,389
LineageQazaq2  14     415178 - 1,451
LineageTurkmen3  17     516221 - 1,806

The age of the On To'rt Urıw and other lineages averaged about 15 generations, equivalent to about 400 to 500 years. The age of the clans varied more widely, from 20 generations for the Qazaqs, to 35 generations for the Qon'ırat, and to 102 generations for the Turkmen. This dates the oldest common ancestor of the Qazaq and Qon'ırat clans to a time some 600 to 1,200 years ago. However the common ancestor of the Turkmen clans is some 3,000 years old. The high ages of the Turkmen clans was the result of the occurrence of a significantly mutated haplotype within the modal haplogroup. It was difficult to judge whether these individuals were genuinely related to the other clan members or were themselves recent immigrants.

These figures must be interpreted with considerable caution. Clearly the age of a clan's common ancestor is not the same as the age of the clan itself, since that ancestor may have had ancestors of his own, whose lines of descent have become extinct over time. The calculated ages therefore give us a minimum limit for the age of the clan and not the age of the clan itself.

In reality however, the uncertainty in the assumed rate of genetic mutation gives rise to extremely wide 95% confidence intervals. The knowledge that certain Qaraqalpaq Qon'ırat clans are most likely older than a time ranging from 450 to 3,700 years is of little practical use to us. Clearly more accurate models are required.

Chaix, R.; Quintana-Murci, L.; Hegay, T.; Hammer, M. F.; Mobasher, Z.; Austerlitz, F.; and Heyer, E., 2007

The latest analysis of Qaraqalpaq DNA comes from a study examining the genetic differences between various pastoral and farming populations in Central Asia. In this region these two fundamentally different economies are organized according to quite separate social traditions:
  • pastoral populations are classified into what their members claim to be descent groups (tribes, clans, and lineages), practice exogamous marriage (where men must marry women from clans that are different to their own), and are organized on a patrilineal basis (children being affiliated to the descent group of the father, not the mother).
  • farmer populations are organized into nuclear and extended families rather than tribes and often practise endogamous marriage (where men marry women from within the same clan, often their cousins).
The study aims to identify differences in the genetic diversity of the two groups as a result of these two different lifestyles. It examines the genetic diversity of:
  • maternally inherited mitochondrial DNA in 12 pastoral and 9 farmer populations, and
  • paternally inherited Y-chromosomes in 11 pastoral and 7 farmer populations.
The diversity of mtDNA was examined by investigating one of two short segments, known as hypervariable segment number 1 or HVS-1. This and HVS-2 have been found to contain the highest density of neutral polymorphic variations between individuals.

The diversity of the Y chromosome was examined by investigating 6 short tandem repeats (STRs) in the non-recombining region of the chromosome.

This particular study sampled mtDNA from 5 different populations from Qaraqalpaqstan: On To'rt Urıw Qaraqalpaqs, Qon'ırat Qaraqalpaqs, Qazaqs, Turkmen, and Uzbeks. Samples collected as part of other earlier studies were used to provide mtDNA data on 16 further populations (one of which was a general group of Qaraqalpaqs) and Y chromosome data on 20 populations (two of which were On To'rt Urıw and Qon'ırat Qaraqalpaqs sampled in 2001 and 2002). The sample size for each population ranged from 16 to 65 individuals.

Both Qaraqalpaq arıs were classified as pastoral, along with Qazaqs, Kyrgyz, and Turkmen. Uzbeks were classified as farmers, along with Tajiks, Uighurs, Kurds, and Dungans.

Results of the mtDNA Analysis

The results of the mtDNA analysis are given in Table 1, copied from the paper.
Table 1. Sample Descriptions and Estimators of Genetic Diversity from the mtDNA Sequence
Population n Location Long Lat H π D pD Ps C
Qaraqalpaqs 20 Uzbekistan 58 43 0.99 5.29 -1.95 0.01 0.90 1.05
Qaraqalpaqs (On To'rt Urıw) 53 Uzbekistan/Turkmenistan border 60 42 0.99 5.98 -1.92 0.01 0.70 1.20
Qaraqalpaqs (Qon'ırat) 55 Qaraqalpaqstan 59 43 0.99 5.37 -2.01 0.01 0.82 1.15
Qazaqs 50 Qaraqalpaqstan 63 44 0.99 5.23 -1.97 0.01 0.88 1.11
Qazaqs 55 Kazakhstan 80 45 0.99 5.66 -1.87 0.01 0.69 1.25
Qazaqs 20   68 42 1.00 5.17 -1.52 0.05 1.00 1.00
Kyrgyz 20 Kyrgyzstan 74 41 0.97 5.29 -1.38 0.06 0.55 1.33
Kyrgyz (Sary-Tash) 47 South Kyrgyzstan, Pamirs 73 40 0.97 5.24 -1.95 0.01 0.49 1.52
Kyrgyz (Talas) 48 North Kyrgyzstan 72 42 0.99 5.77 -1.65 0.02 0.77 1.14
Turkmen 51 Uzbekistan/Turkmenistan border 59 42 0.98 5.48 -1.59 0.04 0.53 1.42
Turkmen 41 Turkmenistan 60 39 0.99 5.20 -2.07 0.00 0.73 1.21
Turkmen 20   59 40 0.98 5.28 -1.71 0.02 0.75 1.18
Dungans 16 Kyrgyzstan 78 41 0.94 5.27 -1.23 0.12 0.31 1.60
Kurds 32 Turkmenistan 59 39 0.97 5.61 -1.35 0.05 0.41 1.52
Uighurs 55 Kazakhstan 82 47 0.99 5.11 -1.91 0.01 0.62 1.28
Uighurs 16 Kyrgyzstan 79 42 0.98 4.67 -1.06 0.15 0.63 1.23
Uzbeks (North) 40 Qaraqalpaqstan 60 43 0.99 5.49 -2.03 0.00 0.68 1.21
Uzbeks (South) 42 Surkhandarya, Uzbekistan 67 38 0.99 5.07 -1.96 0.01 0.81 1.14
Uzbeks (South) 20 Uzbekistan 66 40 0.99 5.33 -1.82 0.02 0.90 1.05
Uzbeks (Khorezm) 20 Khorezm, Uzbekistan 61 42 0.98 5.32 -1.62 0.04 0.70 1.18
Tajiks (Yagnobi) 20   71 39 0.99 5.98 -1.76 0.02 0.90 1.05

Key: the pastoral populations are in the grey area; the farmer populations are in the white area.

The table includes the following parameters:
  • sample size, n, the number of individuals sampled in each population. Individuals had to be unrelated to any other member of the same sample for at least two generations.
  • the geographical longitude and latitude of the population sampled.
  • heterozygosity, H, the proportion of different alleles occupying the same position in each mtDNA sequence. It measures the frequency of heterozygotes for a particular locus in the genetic sequence and is one of several statistics indicating the level of genetic variation or polymorphism within a population. When H=0, all alleles are the same and when H=1, all alleles are different.
  • the mean number of pairwise differences, π, measures the average number of nucleotide differences between all pairs of HVS-1 sequences. This is another statistic indicating the level of genetic variation within a population, in this case measuring the level of mismatch between nucleotides.
  • Tajima’s D, D, measures the frequency distribution of alleles in a nucleotide sequence and is based on the difference between two estimations of the population mutation rate. It is often used to distinguish between a DNA sequence that has evolved randomly (D=0) and one that has experienced directional selection favouring a single allele. It is consequently used as a test for natural selection. However it is also influenced by population history and negative values of D can indicate high rates of population growth.
  • the probability that D is significantly different from zero, pD.
  • the proportion of singletons, Ps, measures the relative number of unique polymorphisms in the sample. The higher the proportion of singletons, the greater the population has been affected by inward migration.
  • the mean number of individuals carrying the same mtDNA sequence, C, is an inverse measure of diversity. The more individuals with the same sequence, the less diversity within the population and the higher proportion of individuals who are closely related.

The table shows surprisingly little differentiation between pastoral and farmer populations. Both show high levels of within population genetic diversity (for both groups, median H=0.99 and π is around 5.3). Further calculations of genetic distance between populations, Fst, ( not presented in the table but given graphically in the online reference below) showed a corresponding low level of genetic differentiation among pastoral populations as well as among farmer populations.

Both groups of populations also showed a significantly negative Tajima’s D, which the authors attribute to a high rate of demographic growth in neutrally evolving populations.

Supplementary data made available online showed a weak correlation between genetic distance, Fst, and geographic distance for both pastoral and farmer populations. Click here for redirection to the relevant online site.

Results of the Y chromosome Analysis

The results of the Y chromosome analysis are given in Table 2, also copied from the paper:
Table 1. Sample Descriptions and Estimators of Genetic Diversity from the Y chromosome STRs
Population n Location Long Lat H π r Ps C
Qaraqalpaqs (On To'rt Urıw) 54 Uzbekistan/Turkmenistan border 60 42 0.86 3.40 1.002 0.24 2.84
Qaraqalpaqs (Qon'ırat) 54 Qaraqalpaqstan 59 43 0.91 3.17 1.003 0.28 2.35
Qazaqs 50 Qaraqalpaqstan 63 44 0.85 2.36 1.004 0.16 2.78
Qazaqs 38 Almaty, KatonKaragay, Karatutuk,
    Rachmanovsky Kluchi, Kazakhstan
68 42 0.78 2.86 1.004 0.26 2.71
Qazaqs 49 South-east Kazakhstan 77 40 0.69 1.56 1.012 0.22 3.06
Kyrgyz 41 Central Kyrgyzstan (Mixed) 74 41 0.88 2.47 1.004 0.41 1.86
Kyrgyz (Sary-Tash) 43 South Kyrgyzstan, Pamirs 73 40 0.45 1.30 1.003 0.12 4.78
Kyrgyz (Talas) 41 North Kyrgyzstan 72 42 0.94 3.21 1.002 0.39 1.78
Mongolians 65 Ulaanbaatar, Mongolia 90 49 0.96 3.37 1.009 0.38 1.81
Turkmen 51 Uzbekistan/Turkmenistan border 59 42 0.67 1.84 1.006 0.27 3.00
Turkmen 21 Ashgabat, Turkmenistan 59 40 0.89 3.34 1.006 0.48 1.62
Dungans 22 Alexandrovka and Osh, Kyrgyzstan 78 41 0.99 4.13 1.005 0.82 1.10
Kurds 20 Bagyr, Turkmenistan 59 39 0.99 3.59 1.009 0.80 1.11
Uighurs 33 Almaty and Lavar, Kazakhstan 79 42 0.99 3.72 1.007 0.67 1.22
Uighurs 39 South East Kazakhstan 79 43 0.99 3.79 1.008 0.77 1.15
Uzbeks (North) 40 Qaraqalpaqstan 60 43 0.96 3.42 1.005 0.48 1.54
Uzbeks (South) 28 Kashkadarya, Uzbekistan 66 40 1.00 3.53 1.008 0.93 1.04
Tajiks (Yagnobi) 22 Penjikent, Tajikistan 71 39 0.87 2.69 1.012 0.45 1.69

Key: the pastoral populations are in the grey area; the farmer populations are in the white area.

This table also includes the sample size, n, and longitude and latitude of the population sampled, as well as the heterozygosity, H, the mean number of pairwise differences, π, the proportion of singletons, Ps, and the mean number of individuals carrying the same Y STR haplotype, C. In addition it includes a statistical computation of the demographic growth rate, r.

In contrast to the results obtained from the mtDNA analysis, both the heterozyosity and the mean pairwise differences computed from the Y chromosome STRs were significantly lower in the pastoral populations than in the farmer populations. Thus Y chromosome diversity has been lost in the pastoral populations.

Conversely calculations of the genetic distance, Rst, between each of the two groups of populations showed that pastoral populations were more highly differentiated than farmer populations. The supplemental data given online demonstrates that this is not as a result of geographic distance, there being no perceived correlation between genetic and geographic distance in both population groups.

Finally the rate of demographic growth was found to be lower in pastoral than in farmer populations.


At first sight the results are counter-intuitive. One would expect that the diversity of mtDNA in pastoral societies would be higher than in farming societies, because the men in those societies are marrying brides who contribute mtDNA from clans other than their own.

Similarly one would expect no great difference in Y chromosome diversity between pastoralists and farmers because both societies are patrilinear. Leaving aside the matter of immigration, the males who contribute the Y chromosome are always selected from the local sampled population.

To understand the results, Chaix et al investigated the distribution of genetic diversity within individual populations using a statistical technique called multi-dimensional scaling analysis or MDS. This attempts to sort or resolve a sample into its different component parts, illustrating the results in two dimensions.

The example chosen in the paper focuses on the Qaraqalpaq On To'rt Urıw arıs. The MDS analysis of the Y chromosome data resolves the sample of 54 individuals into clusters, each of whom have exactly the same STR haplotypes:

Table 1 from Chaix et al
Multidimensional Scaling Analysis based on the Matrix of Distance between Y STR Haplotypes
in a Specific Pastoral Population: the Qaraqalpaq On To'rt Urıw.

Thus the sample contains 13 individuals from the O'mir clan of the Keneges tribe with the same haplotype (shown by the large cross), 10 individuals of the Qarasıyraq clan of the Man'g'ıt tribe with the same haplotype (large diamond), and 10 individuals from the No'kis clan of the Keneges tribe with the same haplotype (large triangle). Other members of the same clans have different haplotypes, as shown on the chart. Those close to the so-called "identity core" group may have arisen by mutation. Those further afield might represent immigrants or adoptions.

No such clustering is observed following the MDS analysis of the mtDNA data for the same On To'rt Urıw arıs:

Table 1 from Chaix et al
Multidimensional Scaling Analysis based on the Number of Differences between the Mitochondrial Sequence
in the Same Pastoral Population: the Qaraqalpaq On To'rt Urıw.

Every individual in the sample, including those from the same clan, has a different HVS-1 sequence.

Similar MDS analyses of the different farmer populations apparently showed very few "identity cores" in the Y chromosome data and a total absence of clustering in the mtDNA data, just as in the case of the On To'rt Urıw.

The overall conclusion was that the existence of "identity cores" was specific to the Y chromosome data and was mainly restricted to the pastoral populations. This is reflected in the tables above, where we can see that the mean number of individuals carrying the same mtDNA sequence ranges from about 1 to 1½ and shows no difference between pastoral and farming populations. On the other hand the mean number of individuals carrying the same STR haplotype is low for farming populations but ranges from 1½ up to almost 5 for the pastoralists. Pastoral populations also have a lower number of Y chromosome singletons.

Chaix et al point to three reinforcing factors to explain the existence of "identity cores" in pastoral as opposed to farming populations:

  • pastoral lineages frequently split and divide with closely related men remaining in the same sub-group, thereby reducing Y chromosome diversity,
  • small populations segmented into lineages can experience strong genetic drift, creating high frequencies of specific haplotypes, and
  • random demographic uncertainty in small lineage groups can lead to the extinction of some haplotypes, also reducing diversity.
Together these factors reduce overall Y chromosome diversity.

To explain the similar levels of mtDNA diversity in pastoral and farmer populations, Chaix et al point to the complex rules connected with exogamy. Qazaq men for example must marry a bride who has not had an ancestor belonging to the husband's own lineage for at least 7 generations, while Qaraqalpaq men must marry a bride from another clan, although she can belong to the same tribe. Each pastoral clan, therefore, is gaining brides (and mtDNA) from external clans but is losing daughters (and mtDNA) to external clans. Such continuous and intense migration reduces mtDNA genetic drift within the clan. This in turn lowers diversity to a level similar to that observed in farmer populations, which is in any event already high. The process of two-way female migration effectively isolates the mtDNA structure of pastoral societies from their social structure.

One aspect overlooked by the study is that, until recent times, Qaraqalpaq clans were geographically isolated in villages located in specific parts of the Aral delta and therefore tended to always intermarry with one of their adjacent neighbouring clans.

In effect, the two neighbouring clans behaved like a single population, with females moving between clans in every generation. How such social behaviour affected genetic structure was not investigated.


The Uzbeks were traditionally nomadic pastoralists and progressively became settled agricultural communities from the 16th century onwards. The survey provided an opportunity to investigate the effect of this transition in lifestyle on the genetic structure of the Uzbek Y chromosome.

Table 2 above shows that the genetic diversity found among Uzbeks, as measured by heterozygosity and the mean number of pairwise differences, was similar to that of the other farmer populations, as was the proportion of singleton haplotypes. Equally the mean number of individuals carrying the same Y STR haplotype was low (1 to 1½), indicating an absence of the haplotype clustering (or "identity cores") observed in pastoral populations. The pastoral "genetic signature" must have been rapidly eroded, especially in the case of the northern Uzbeks from Qaraqalpaqstan, who only settled from the 17th century onwards.

Two reasons are proposed for this rapid transformation. Firstly the early collapse and integration of the Uzbek descent groups following their initial settlement and secondly their mixing with traditional Khorezmian farming populations, which led to the creation of genetic admixtures of the two groups.


Of course the Qaraqalpaq On To'rt Urıw have been settled farmers for just as long as many Khorezmian Uzbeks and cannot in any way be strictly described as pastoralists. Indeed the majority of Qaraqalpaq Qon'ırats have also been settled for much of the 20th century. However both have strictly maintained their traditional pastoralist clan structure and associated system of exogamous marriage. So although their lifestyles have changed radically , their social behaviour to date has not.

Discussion and Conclusions

The Qaraqalpaqs and their Uzbek and Qazaq neighbours have no comprehensive recorded history, just occasional historical reports coupled with oral legends which may or may not relate to certain historical events in their past. We therefore have no record of where or when the Qaraqalpaq confederation emerged and for what political or other reasons.

In the absence of solid archaeological or historical evidence, many theories have been advanced to explain the origin of the Qaraqalpaqs. Their official history, as taught in Qaraqalpaq colleges and schools today, claims that the Qaraqalpaqs are the descendants of the original endemic nomadic population of the Khorezm oasis, most of whom were forced to leave as a result of the Mongol invasion in 1221 and the subsequent dessication of the Aral delta following the devastation of Khorezm by Timur in the late 14th century, only returning in significant numbers during the 18th century. We fundamentally disagree with this simplistic picture, which uncritically endures with high- ranking support because it purports to establish an ancient Qaraqalpaq origin and justifies tenure of the current homeland.

While population genetics cannot unravel the full tribal history of the Qaraqalpaqs per se, it can give us important clues to their formation and can eliminate some of the less likely theories that have been proposed.

The two arıs of the Qaraqalpaqs, the Qon'ırat and the On To'rt Urıw, are very similar to each other genetically, especially in the male line. Both are equally close to the Khorezmian Uzbeks, their southern neighbours. Indeed the genetic distances between the different populations of Uzbeks scattered across Uzbekistan is no greater than the distance between many of them and the Qaraqalpaqs. This suggests that Qaraqalpaqs and Uzbeks have very similar origins. If we want to find out about the formation of the Qaraqalpaqs we should look towards the emergence of the Uzbek (Shaybani) Horde and its eastwards migration under the leadership of Abu'l Khayr, who united much of the Uzbek confederation between 1428 and 1468.

Like the Uzbeks, the Qaraqalpaqs are extremely diverse genetically. One only has to spend time with them to realize that some look European, some look Caucasian, and some look typically Mongolian. Their DNA turns out to be an admixture, roughly balanced between eastern and western populations. Two of their main genetic markers have far-eastern origins, M9 being strongly linked to Chinese and other Far Eastern peoples and M130 being linked to the Mongolians and Qazaqs. On the other hand, M17 is strong in Russia, the Ukraine, and Eastern Europe, while M89 is strong in the Middle East, the Caucasus, and Russia. M173 is strong in Western Europe and M45 is believed to have originated in Central Asia, showing that some of their ancestry goes back to the earliest inhabitants of that region. In fact the main difference between the Qaraqalpaqs and the Uzbeks is a slight difference in the mix of the same markers. Qaraqalpaqs have a somewhat greater bias towards the eastern markers. One possible cause could be the inter-marriage between Qaraqalpaqs and Qazaqs over the past 400 years, a theory that gains some support from the close similarities in the mitochondrial DNA of the neighbouring female Qaraqalpaq Qon'ırat and Qazaqs of the Aral delta.

After the Uzbeks, Qaraqalpaqs are next closest to the Uighurs, the Crimean Tatars, and the Kazan Tatars, at least in the male line. However in the female line the Qaraqalpaqs are quite different from the Uighurs and Crimean Tatars (and possibly from the Kazan Tatars as well). There is clearly a genetic link with the Tatars of the lower Volga through the male line. Of course the Volga region has been closely linked through communications and trade with Khorezm from the earliest days.

The Qaraqalpaqs are genetically distant from the Qazaqs and the Turkmen, and even more so from the Kyrgyz and the Tajiks. We know that the Qaraqalpaqs were geographically, politically, and culturally very close to the Qazaqs of the Lesser Horde prior to their migration into the Aral delta and were even once ruled by Qazaq tribal leaders. From their history, therefore, one might have speculated that the Qaraqalpaqs may have been no more than another tribal group within the overall Qazaq confederation. This is clearly not so. The Qazaqs have a quite different genetic history, being far more homogenous and genetically closer to the Mongolians of East Asia. However as we have seen, the proximity of the Qazaqs and Qaraqalpaqs undoubtedly led to intermarriage and therefore some level of genetic exchange.

Qaraqalpaq Y chromosome polymorphisms show different patterns from mtDNA polymorphisms in a similar manner to that identified in certain other Central Asian populations. This seems to be associated with the Turkic traditions of exogamy and so-called patrilocal marriage. Marriage is generally not permissible between couples belonging to the same clan, so men must marry women from other clans, or tribes, or in a few cases even different ethnic groups. After the marriage the groom stays in his home village and his bride moves from her village to his. The result is that the male non-recombining part of the Y chromosome becomes localized as a result of its geographical isolation, whereas the female mtDNA benefits from genetic mixing as a result of the albeit short range migration of young brides from different clans and tribes.

One of the most important conclusions is the finding that clans within the same tribe show no sign of genetic kinship, whether the tribe concerned is Qaraqalpaq, Uzbek, Qazaq, or Turkmen. Indeed among the most settled ethnic groups, the Uzbeks and Qaraqalpaq On To'rt Urıw, there is very little kinship even at clan level. It seems that settled agricultural communities soon lose their strong tribal identity and become more openminded to intermarriage with different neighbouring ethnic groups. Indeed the same populations place less importance on their geneaology and no longer maintain any identity according to lineage.

It has generally been assumed that most Turkic tribal groups like the Uzbeks were formed as confederations of separate tribes and this is confirmed by the recent genetic study of ethnic groups from Qaraqalpaqstan. We now see that this extends to the tribes themselves, with an absence of any genetic link between clans belonging to the same tribe. Clearly they too are merely associations of disparate groups, formed because of some historical reason other than descent. Possible causes for such an association of clans could be geographic or economic, such as common land use or shared water rights; military, such as a common defence pact or the construction of a shared qala; or perhaps political, such as common allegiance to a strong tribal leader.

The history of Central Asia revolves around migrations and conflicts and the formation, dissolution, and reformation of tribal confederations, from the Saka Massagetae and the Sarmatians, to the Oghuz and Pechenegs, the Qimek, Qipchaq, and Karluk, the Mongols and Tatars, the White and Golden Hordes, the Shaybanid and Noghay Hordes, and finally the Uzbek, Qazaq, and Qaraqalpaq confederations. Like making cocktails from cocktails, the gene pool of Central Asia was constantly being scrambled, more so on the female line as a result of exogamy and patrilineal marriage.

The same tribal and clan names occur over and over again throughout the different ethnic Qipchaq-speaking populations of Central Asia, but in different combinations and associations. Many of the names predate the formation of the confederations to which they now belong, relating to earlier Turkic and Mongol tribal factions. Clearly tribal structures are fluid over time, with some groups withering or being absorbed by others, while new groups emerge or are added.

When Abu'l Khayr Sultan became khan of the Uzbeks in 1428-29, their confederation consisted of at least 24 tribes, many with smaller subivisions. The names of 6 of those tribes occur among the modern Qaraqalpaqs. A 16th century list, based on an earlier document, gives the names of 92 nomadic Uzbek tribes, at least 20 of which were shared by the later breakaway Qazaqs. 13 of the 92 names also occur among the modern Qaraqalpaqs.

Shortly after his enthronement as the Khan of Khorezm in 1644-45, Abu'l Ghazi Khan reorganized the tribal structure of the local Uzbeks into four tüpe:

TüpeMain TribesSecondary Tribes Qaraqalpaq
On Tort UrughOn To'rt UrıwQan'glı
Durman, Yüz, Ming
Shaykhs, Burlaqs, Arabs

8 out of the 11 tribal names associated with the first three tüpe are also found within the Qaraqalpaq tribal structure. Clearly there is greater overlap between the Qaraqalpaq tribes and the local Khorezmian Uzbek tribes than in the Uzbek tribes in general. The question is whether these similarities pre-dated the Qaraqalpaq migration into the Aral delta or are a result of later Uzbek influences?

We know that the Qon'ırat were a powerful tribe in Khorezm for Uzbeks and Qaraqalpaqs alike. They were mentioned as one of the Qaraqalpaq "clans" on the Kuvan Darya [Quwan Darya] by Gladyshev in 1741 along with the Kitay, Qipchaq, Kiyat, Kinyagaz-Mangot (Keneges-Man'g'ıt), Djabin, Miton, and Usyun. Munis recorded that Qaraqalpaq Qon'ırat, Keneges, and Qıtay troops supported Muhammad Amin Inaq against the Turkmen in 1769. Thanks to Sha'rigu'l Payzullaeva we have a comparison of the Qon'ırat tribal structure in the Aral Qaraqalpaqs, the Surkhandarya Qaraqalpaqs, and the Khorezmian Uzbeks, derived from genealogical records:

The different status of the same Qon'ırat tribal groups among the Aral and Surkhandarya Qaraqalpaqs and the Khorezmian Uzbeks
Tribal GroupAral
Bu'gejeyli clanclan
Qostamg'alıclanbranch of tribe 
Qanjıg'alıtiırebranch of tribetube
Shu'llikdivision of arısclan 
Tartıwlıtiırebranch of tribeclan
Sıyraqclanbranch of clan 
Qaramoyıntribebranch of clan 
A tube is a branch of a tribe among the Khorezmian Uzbeks and a tiıre is a branch of a clan among the Aral Qaraqalpaqs.

The Qaraqalpaq enclave in Surkhandarya was already established in the first half of the 18th century, some Qaraqalpaqs fleeing to Samarkand and beyond following the devastating Jungar attack of 1723. Indeed it may even be older - the Qon'ırat have a legend that they came to Khorezm from the country of Zhideli Baysun in Surkhandarya. This suggests that some Qaraqalpaqs had originally travelled south with factions from the Shaybani Horde in the early 16th century. The fact that the Qaraqalpaq Qon'ırats remaining in that region have a similar tribal structure to the Khorezmian Uzbeks is powerful evidence that the tribal structure of the Aral Qaraqalpaqs had broadly crystallized prior to their migration into the Aral delta.

The Russian ethnographer Tatyana Zhdanko was the first academic to make an in-depth study of Qaraqalpaq tribal structure. She not only uncovered the similarities between the tribal structures of the Uzbek and Qaraqalpaq Qon'ırats in Khorezm but also the closeness of their respective customs and material and spiritual cultures. She concluded that one should not only view the similarity between the Uzbek and Qaraqalpaq Qon'ırats in a historical sense, but should also see the commonality of their present- day ethnic relationships. B. F. Choriyev added that "this kind of similarity should not only be sought amongst the Qaraqalpaq and the Khorezmian Qon'ırats but also amongst the Surkhandarya Qon'ırats. They all have the same ethnic history."

Such ethnographic studies provide support to the findings that have emerged from the recent studies of Central Asian genetics. Together they point towards a common origin of the Qaraqalpaq and Uzbek confederations. They suggest that each was formed out of the same melange of tribes and clans inhabiting the Dasht-i Qipchaq following the collapse of the Golden Horde, a vast expanse ranging northwards from the Black Sea coast to western Siberia and then eastwards to the steppes surrounding the lower and middle Syr Darya, encompassing the whole of the Aral region along the way.

Of course the study of the genetics of present-day populations gives us the cumulative outcome of hundreds of thousands of years of complex human history and interaction. We now need to establish a timeline, tracking genetic changes in past populations using the human skeletal remains retrieved from Saka, Sarmatian, Turkic, Tatar, and early Uzbek and Qaraqalpaq archaeological burial sites. Such studies might pinpoint the approximate dates when important stages of genetic intermixing occurred.


Sha'rigu'l Payzullaeva recalls an interesting encounter at the Regional Studies Museum in No'kis during the month of August 1988. Thirty-eight elderly men turned up together to visit the Museum. Each wore a different kind of headdress, some with different sorts of taqıya, others with their heads wrapped in a double kerchief. They introduced themselves as Qaraqalpaqs from Jarqorghan rayon in Surkhandarya viloyati, just north of the Afghan border. One of them said "Oh daughter, we are getting old now. We decided to come here to see our homeland before we die."

During their visit to the Museum they said that they would travel to Qon'ırat rayon the following day. Sha'rigu'l was curious to know why they specifically wanted to visit Qon'ırat. They explained that it was because most of the men were from the Qon'ırat clan.

One of the men introduced himself to Sha'rigu'l: "My name is Mirzayusup Khaliyarov, the name of my clan is Qoldawlı. After discovering that Sha'rigu'l was also Qoldawlı his eyes filled with tears and he kissed her on the forehead.


Bowles, G. T., The People of Asia, Weidenfeld and Nicolson, London, 1977.

Comas, D., Calafell, F., Mateu, E., Pérez-Lezaun, A., Bosch, E., Martínez-Arias, R., Clarimon, J., Facchini, F., Fiori, G., Luiselli, D., Pettener, D., and Bertranpetit, J., Trading Genes along the Silk Road: mtDNA Sequences and the Origin of Central Asian Populations, American Journal of Human Genetics, 63, pages 1824 to 1838, 1998.

Cavalli-Sforza, L. L., Menozzi, P., and Piazza, A., The History and Geography of Human Genes, Princeton University Press, Princeton, 1994.

Chaix, R., Austerlitz, F., Khegay, T., Jacquesson, S., Hammer, M. F., Heyer, E., and Quintana-Murci, L., The Genetic or Mythical Ancestry of Descent Groups: Lessons from the Y Chromosome, American Journal of Human Genetics, Volume 75, pages 1113 to 1116, 2004.

Chaix, R., Quintana-Murci, L., Hegay, T., Hammer, M. F., Mobasher, Z., Austerlitz, F., and Heyer, E., From Social to Genetic Structures in Central Asia, Current Biology, Volume 17, Issue 1, pages 43 to 48, 9 January 2007.

Comas, D., Plaza, S., Spencer Wells, R., Yuldaseva, N., Lao, O., Calafell, F., and Bertranpetit, J., Admixture, migrations, and dispersals in Central Asia: evidence from maternal DNA lineages, European Journal of Human Genetics, pages 1 to 10, 2004.

Heyer, E., Central Asia: A common inquiry in genetics, linguistics and anthropology, Presentation given at the conference entitled "Origin of Man, Language and Languages", Aussois, France, 22-25 September, 2005.

Heyer, E., Private communications to the authors, 14 February and 17 April, 2006.

Krader, L., Peoples of Central Asia, The Uralic and Altaic Series, Volume 26, Indiana University, Bloomington, 1971.

Passarino, G., Semino, O., Magri, C., Al-Zahery, N., Benuzzi, G., Quintana-Murci, L., Andellnovic, S., Bullc-Jakus, F., Liu, A., Arslan, A., and Santachiara-Benerecetti, A., The 49a,f Haplotype 11 is a New Marker of the EU19 Lineage that Traces Migrations from Northern Regions of the Black Sea, Human Immunology, Volume 62, pages 922 to 932, 2001.

Payzullaeva, Sh., Numerous Qaraqalpaqs, many of them! [in Qaraqalpaq], Qaraqalpaqstan Publishing, No'kis, 1995.

Pérez-Lezaun, A., Calafell, F., Comas, D., Mateu, E., Bosch, E., Martínez-Arias, R., Clarimón, J., Fiori, G., Luiselli, D., Facchini, F., Pettener, D., and Bertranpetit, J., Sex-Specific Migration Patterns in Central Asian Populations, Revealed by Analysis of Y-Chromosome Short Tandem Repeats and mtDNA, American Journal of Human Genetics, Volume 65, pages 208 to 219, 1999.

Spencer Wells, R., The Journey of Man, A Genetic Odyssey, Allen Lane, London, 2002.

Spencer Wells, R., et al, The Eurasian Heartland: A continental perspective on Y-chromosome diversity, Proceedings of the National Academy of Science, Volume 98, pages 10244 to 10249, USA, 28 August 2001.

Underwood, J. H., Human Variation and Human Micro-Evolution, Prentice-Hall Inc., New Jersey, 1979.

Underwood, P. A., et al, Detection of Numerous Y Chromosome Biallelic Polymorphisms by Denaturing High-Performance Liquid Chromatography, Genome Research, Volume 7, pages 996 to 1005, 1997.

Zerjal, T., Spencer Wells, R., Yuldasheva, N., Ruzibakiev, R., and Tyler-Smith, C., A Genetic Landscape Reshaped by Recent Events: Y Chromosome Insights into Central Asia, American Journal of Human Genetics, Volume 71, pages 466 to 482, 2002.

Return to top of page

Home Page