Genetics and archaeogenetics of South Asia
Genetics and archaeogenetics of South Asia is the study of the genetics and archaeogenetics of the ethnic groups of South Asia. It aims at uncovering these groups' genetic history. The geographic position of South Asia makes its biodiversity important for the study of the early dispersal of anatomically modern humans across Asia.
Genomic studies have described the genetic landscape of South Asia as a composite of West Eurasian and East Asian exogenous components that mixed with the indigenous South Asian groups to create modern-day South Asians. Studies based on Mitochondrial DNA variations have reported genetic unity across various South Asian sub–populations. Conclusions of studies based on Y Chromosome variation and Autosomal DNA variation have been varied, although many researchers argue that most of the ancestral nodes of the phylogenetic tree of all the mtDNA types originated in South Asia. Recent genome studies appear to show that most South Asians are descendants of two major ancestral components, one restricted to South Asia and the other component derived from IVC-people and Steppe-people, making it more closely related to those in Central Asia, West Asia and Europe.
It has been found that the ancestral node of the phylogenetic tree of all the mtDNA types typically found in Central Asia, the West Asia and Europe are also to be found in South Asia at relatively high frequencies. The inferred divergence of this common ancestral node is estimated to have occurred slightly less than 50,000 years ago. In India, the major maternal lineages are various M subclades, followed by R and U sublineages. These mitochondrial haplogroups' coalescence times have been approximated to date to 50,000 BP.
The major paternal lineages represented by Y chromosomes are haplogroups R1a1, R2, H, L and Haplogroup J2. Some researchers have argued that Y-DNA Haplogroup R1a1 is of autochthonous South Asian origin. However, proposals for a Central Asian origin for R1a1 are also quite common.
Overview
All the mtDNA and Y-chromosome lineages outside Africa descend from three founder lineages:All these six founder haplogroups can be found in the present day populations of South Asia. Moreover, the mtDNA haplogroup M and the Y-chromosome haplogroups C and D are restricted to the area east of South Asia. All the West Eurasian populations derive from the N and R haplogroups of mtDNA and the F haplogroup of the Y-chromosome.
Endicott et al. state that these facts are consistent with the hypothesis of a single exodus from East Africa 65,000 years ago via a southern coastal route, with the West Eurasian lineages separating from the South Asian lineages somewhere between East/Northeast Africa and South Asia.
The predominant majority genome markers of South Asians are all closely related to West Eurasians and may have either originated in Western Asia or South Asia itself.
mtDNA
The most frequent mtDNA haplogroups in South Asia are M, R and U.Arguing for the longer term "rival Y-Chromosome model", Stephen Oppenheimer believes that it is highly suggestive that India is the origin of the Eurasian mtDNA haplogroups which he calls the "Eurasian Eves". According to Oppenheimer it is highly probable that nearly all human maternal lineages in Central Asia, the Middle East and Europe descended from only four mtDNA lines that originated in South Asia 50,000–100,000 years ago.
Macrohaplogroup M
The macrohaplogroup M, which is considered as a cluster of the proto-Asian maternal lineages, represents more than 60% of South Asian MtDNA.The M macrohaplotype in India includes many subgroups that differ profoundly from other sublineages in East Asia especially Mongoloid populations. The deep roots of M phylogeny clearly ascertain the relic of South Asian lineages as compared to other M sub lineages suggesting 'in-situ' origin of these sub-haplogroups in South Asia, most likely in India. These deep rooting lineages are not language specific and spread over all the language groups in India.
Virtually all modern Central Asian MtDNA M lineages seem to belong to the Eastern Eurasian rather than the South Asian subtypes of haplogroup M, which indicates that no large-scale migration from the present Turkic-speaking populations of Central Asia occurred to India. The absence of haplogroup M in Europeans, compared to its equally high frequency among South Asians, East Asians and in some Central Asian populations contrasts with the Western Eurasian leanings of South Asian paternal lineages.
Most of the extant mtDNA boundaries in South and Southwest Asia were likely shaped during the initial settlement of Eurasia by anatomically modern humans.
| Haplogroup | Important Sub clades | Populations |
| M2 | M2a, M2b | Throughout the continent with low presence in Northwest Peaking in Bangladesh, Andhra Pradesh, coastal Tamil Nadu and Sri Lanka |
| M3 | M3a | Concentrated into northwestern India Highest amongst the Parsees of Mumbai |
| M4 | M4a | Peaks in Pakistan, Kashmir and Andhra Pradesh |
| M6 | M6a, M6b | Kashmir and near the coasts of the Bay of Bengal, Sri Lanka |
| M18 | Throughout South Asia Peaking at Rajasthan and Andhra Pradesh | |
| M25 | Moderately frequent in Kerala and Maharashtra but rather infrequent elsewhere in India |
Macrohaplogroup R
The macrohaplogroup R is also widely represented and accounts for the other 40% of South Asian MtDNA. A very old and most important subdivision of it is haplogroup U that, while also present in West Eurasia, has several subclades specific to South Asia.Most important South Asian haplogroups within R:
| Haplogroup | Populations |
| R2 | Distributed widely across the sub continent |
| R5 | widely distributed by most of India. Peaks in coastal SW India |
| R6 | widespread at low rates across India. Peaks among Tamils and Kashmiris |
| W | Found in northwestern states. Peaks in Gujarat, Punjab and Kashmir, frequency is low elsewhere. |
Haplogroup U
is a sub-haplogroup of macrohaplogroup R. The distribution of haplogroup U is a mirror image of that for haplogroup M: the former has not been described so far among eastern Asians but is frequent in European populations as well as among South Asians. South Asian U lineages differ substantially from those in Europe and their coalescence to a common ancestor also dates back to about 50,000 years.| Haplogroup | Populations |
| U2* | is sparsely distributed specially in the northern half of the South Asia. It is also found in SW Arabia. |
| U2a | shows relatively high density in Pakistan and NW India but also in Karnataka, where it reaches its higher density. |
| U2b | has highest concentration in Uttar Pradesh but is also found in many other places, specially in Kerala and Sri Lanka. It is also found in Oman. |
| U2c | is specially important in Bangladesh and West Bengal. |
| U2l | is maybe the most important numerically among U subclades in South Asia, reaching specially high concentrations in Uttar Pradesh, Sri Lanka, Sindh and parts of Karnataka. It also has some importance in Oman. mtDNA haplogroup U2i is dubbed "Western Eurasian" in Bamshad et al. study but "Eastern Eurasian " in Kivisild et al. study. |
| U7 | this haplogroup has a significant presence in Gujarat, Punjab and Pakistan. The possible homeland of this haplogroup spans Gujarat and Iran because from there its frequency declines steeply both to the east and to the west. |
Y chromosome
The major South Asian Y-chromosome DNA haplogroups are H, J2, L, R1a1 and R2. Their geographical origins are listed as follows, according to the latest scholarship:
| Major South Asian Y-chromosomal lineages: | H | J2 | L | R1a | R2 |
| Basu et al. | no comment | no comment | no comment | Central Asia | no comment |
| Kivisild et al. | India | Western Asia | India | Southern and Western Asia | South-Central Asia |
| Cordaux et al. | India | West or Central Asia | Middle Eastern | Central Asia | South-Central Asia |
| Sengupta et al. | India | The Middle East and Central Asia | South India | North India | North India |
| Thanseem et al. | India | The Levant | The Middle East | Southern and Central Asia | Southern and Central Asia |
| Sahoo et al. | South Asia | The Near East | South Asia | South or West Asia | South Asia |
| Mirabal et al. | no comment | no comment | no comment | Northwestern India or Central Asia | no comment |
| Zhao et al. | India | The Middle East | The Middle East | Central Asia or West Eurasia | Central Asia or West Eurasia |
| Sharma et al. | no comment | no comment | no comment | South Asia | no comment |
| Thangaraj et al. | South Asia | The Near East | The Near East | South Asia | South Asia |
Haplogroup H
is found at a high frequency in South Asia. H is today rarely found outside of the South Asia but is common among the Romanis, particularly the H-M82 subgroup. H was also quite common in ancient samples of Europe and is still found today at a low frequency in Europeans and Arabs of the Levant. Haplogroup H is frequently found among populations of India, Sri Lanka, Nepal, Pakistan and the Maldives. All three branches of Haplogroup H are found in South Asia.It is a branch of Haplogroup F and descends from GHIJK family. Haplogroup H is believed to have arisen in South Asia between 30,000 and 40,000 years ago. Its probable site of introduction is South Asia, since it is concentrated there. It seems to represent the main Y-Chromosome haplogroup of the paleolithic inhabitants of South Asia. Some individuals in South Asia have also been shown to belong to the much rarer subclade H3. Haplogroup H is by no means restricted to specific populations. For example, H is possessed by about 28.8% of Indo-Aryan castes. and in tribals about 25–35%.
Haplogroup J2
Haplogroup J2 has been present in South Asia mostly as J2a-M410 and J2b-M102, since neolithic times. J2 clades attain peak frequencies in the North-West and South India and is found at 19% within South Indian castes, 11% in North Indian castes and 12% in Pakistan. In South India, the presence of J2 is higher among middle castes at 21%, followed by upper castes at 18.6% and lower castes at 14%. Among caste groups, the highest frequency of J2-M172 is observed among Tamil Vellalar's of South India, at 38.7%. J2 is present in tribals too and has a frequency of 11% in Austro-Asiatic tribals. Among the Austro-Asiatic tribals, the predominant J2 occurs in the Lodha. J2 is also present in the South Indian hill tribe Toda at 38.46%, in the Andh tribe of Telangana at 35.19% and in the Kol tribe of Uttar Pradesh at a frequency of 33.34%. Haplogroup J-P209 was found to be more common in India's Shia Muslims, of which 28.7% belong to haplogroup J, with 13.7% in J-M410, 10.6% in J-M267 and 4.4% in J2b.In Pakistan, the highest frequencies of J2-M172 were observed among the Parsis at 38.89%, the Dravidian speaking Brahuis at 28.18% and the Makrani Balochs at 24%. It also occurs at 18.18% in Makrani Siddis and at 3% in Karnataka Siddis.
J2-M172 is found at an overall frequency of 10.3% among the Sinhalese people of Sri Lanka. In Maldives, 20.6% of Maldivian population were found to be haplogroup J2 positive.
Haplogroup L
According to Dr. Spencer Wells, L-M20 originated in the Pamir Knot region in Tajikistan and migrated into Pakistan and India ca. 30,000 years ago. However, most other studies have proposed a West Asian origin for L-M20 and associated its expansion in the Indus valley to neolithic farmers. There are three subbranches of haplogroup L: L1-M76, L2-M317 and L3-M357, found at varying levels in South Asia.India
shows time of neolithic expansion. The clade is present in the Indian population at an overall frequency of ca. 7–15%. Haplogroup L has higher frequency among south Indian castes and reaches up to 68% in some castes in Karnataka but is somewhat rarer in north Indian castes. The presence of haplogroup L is quite rare among tribal groups, however a moderate, 14.6% has been observed among the Chenchus.Among regional and social groups, moderate to high frequencies have been observed in Konkanastha Brahmins, Punjabis, Gujaratis, Lambadis, Jats
Pakistan
In Pakistan, L1-M76 and L3-M357 subclades of L-M20 reaches overall frequencies of 5.1% and 6.8%, respectively.Haplogroup L3 is found frequently among Burusho and Pashtuns. Its highest frequency can be found in south western Balochistan province along the Makran coast to Indus River delta. L3a is found in approximately 23% of Nuristani in northwest Pakistan.
The clade is present in moderate distribution among the general Pakistani population.
Sri Lanka
In one study, 16% of the Sinhalese were found to be Haplogroup L-M20 positive. In another study 18% were found to belong to L1.Haplogroup R1a1
In South Asia, R1a1 has been observed often with high frequency in a number of demographic groups, as well as with highest STR diversity which lead some to see it as the locus of origin.While R1a originated ca. 22,000 to 25,000 years ago, its subclade M417 diversified ca. 5,800 years ago. The distribution of M417-subclades R1-Z282 in Central- and Eastern Europe and R1-Z93 in Asia suggests that R1a1a diversified within the Eurasian Steppes or the Middle East and Caucasus region. The place of origin of these subclades plays a role in the debate about the origins of Indo-Europeans.
India
In India, high percentage of this haplogroup is observed in West Bengal Brahmins to the east, Gujarat Lohanas to the west, Khatris in north, Iyengar Brahmins in the south. It has also been found in several South Indian Dravidian-speaking tribals including the Kotas of Tamil Nadu Chenchu and Valmikis of Andhra Pradesh as well as the Yadav and Kallar of Tamil Nadu suggesting that M17 is widespread in these Southern Indians tribes. Besides these, studies show high percentages in regionally diverse groups such as Manipuris to the extreme North East and in among Punjabis to the extreme North West.Pakistan
In Pakistan, it is found at 71% among the Mohanna of Sindh Province to the south and 46% among the Baltis of Gilgit-Baltistan to the north.Sri Lanka
23% of the Sinhalese people out of a sample of 87 subjects were found to be R1a1a positive according to a 2003 research.Maldives
In Maldives, 23.8% of the Maldivian people were found to be R1a1a positive.Nepal
People in Terai Region, Nepal show R1a1a at 69%.Haplogroup R2
In South Asia, the frequency of R2 and R2a lineage is around 10–15% in India and Sri Lanka and 7–8% in Pakistan. At least 90% of R-M124 individuals are located in South Asia. It is also reported in Caucasus and Central Asia at lower frequency. A genetic study by Mondal et al. 2017 concluded that haplogroup Haplogroup R2 originated in northern India and was already present before the Steppe migration.India
Among regional groups, it is found among West Bengalis, New Delhi Hindus, Punjabis and Gujaratis. Among tribal groups, Karmalis of West Bengal showed highest at 100% followed by Lodhas to the east, while Bhil of Gujarat in the west were at 18%, Tharus of north showed it at 17%, Chenchu and Pallan of south were at 20% and 14% respectively. Among caste groups, high percentages are shown by Jaunpur Kshatriyas, kamma, Bihar Yadav, Khandayat and Kallar.It is also significantly high in many Brahmin groups including Punjabi Brahmins, Bengali Brahmins, Konkanastha Brahmins, Chaturvedis, Bhargavas, Kashmiri Pandits and Lingayat Brahmins.
North Indian Muslims have a frequency of 19% and 13%, while Dawoodi Bohra Muslim in the western state of Gujarat have a frequency of 16% and Mappila Muslims of South India have a frequency of 5%.
Pakistan
The R2 haplogroup is found in 14% of the Burusho people. Among the Hunza people it is found at 18% while the Parsis show it at 20%. It is also found in the northeastern part of Afghanistan.Sri Lanka
38% of the Sinhalese of Sri Lanka were found to be R2 positive according to a 2003 research.Maldives
12% of the Maldivian people of Maldives are found to have R2.Nepal
In Nepal, R2 percentages range from 2% to 26% within different groups under various studies. Newars show a significantly high frequency of 26% while people of Kathmandu show it at 10%.Reconstructing South Asian population history
The, divides the population of South Asia into four ethnolinguistic groups: Indo-European, Dravidian, Tibeto-Burman and Austro-Asiatic. The molecular anthropology studies use three different type of markers: Mitochondrial DNA variation which is maternally inherited and highly polymorphic, Y Chromosome variation which involves uniparental transmission along the male lines, and Autosomal DNA variation.mtDNA variation
Most of the studies based on mtDNA variation have reported genetic unity of South Asian populations across language, caste and tribal groups. It is likely that haplogroup M was brought to Asia from East Africa along the southern route by earliest migration wave 78,000 years ago.According to Kivisild et al., "Minor overlaps with lineages described in other Eurasian populations clearly demonstrate that recent immigrations have had very little impact on the innate structure of the maternal gene pool of South Asians. Despite the variations found within India, these populations stem from a limited number of founder lineages. These lineages were most likely introduced to South Asia during the Middle Palaeolithic, before the peopling of Europe 48,000 years ago and perhaps the Old World in general." Basu et al. also emphasises underlying unity of female lineages in India.
Y Chromosome variation
Conclusions based on Y Chromosome variation have been more varied than those based on mtDNA variation. While proposes an ancient and shared genetic heritage of male lineages in South Asia, Bamshad et al. suggests an affinity between South Asian male lineages and west Eurasians proportionate to upper caste rank and places upper caste populations of southern Indian states closer to East Europeans.Basu et al. concludes that Austro–Asiatic tribal populations entered India first from the Northwest corridor and much later some of them through Northeastern corridor. Whereas, Kumar et al. analysed 25 South Asian Austro-Asiatic tribes and found strong paternal genetic link among the sub-linguistic groups of the South Asian Austro-Asiatic populations. Mukherjee et al. places Pakistanis and North Indians between west Asian and Central Asian populations, whereas Cordaux et al. argues that the Indian caste populations are closer to Central Asian populations. Sahoo et al. and Sengupata et al. suggest that Indian caste populations have not been subject to any recent admixtures. Sanghamitra Sahoo concludes his study with:
Closest neighbor analysis done by Mondal et al. 2017 concluded that Indian Y-lineages are close to southern European populations and the time of divergence between the two predated Steppe migration.":
Autosomal DNA variation
AASI-ANI-ASI
Results of studies based upon autosomal DNA variation have also been varied. In a major study using over 500,000 biallelic autosomal markers, Reich hypothesized that the modern South Asian population was the result of admixture between two genetically divergent ancestral populations dating from the post-Holocene era. These two "reconstructed" ancient populations he termed "Ancestral South Indians" and "Ancestral North Indians". According to Reich: "ANI ancestry is significantly higher in Indo-European than Dravidian speakers, suggesting that the ancestral ASI may have spoken a Dravidian language before mixing with the ANI." While the ANI is genetically close to Middle Easterners, Central Asians and Europeans, the ASI is not closely related to groups outside of the subcontinent. As no "ASI" ancient DNA is available, the indigenous Andamanese Onge are used as an proxy for ASI. According to Reich et al., both ANI and ASI ancestry are found all over the subcontinent in varying proportions, and that “ANI ancestry ranges from 39-71% in India, and is higher in traditionally upper caste and Indo-European speakers."Moorjani et al. 2013 state that the ASI, though not closely related to any living group, are "related to indigenous Andaman Islanders." Moorjani et al. also suggest possible gene flow into the Andamanese from a population related to the ASI. The study concluded that “almost all groups speaking Indo-European or Dravidian languages lie along a gradient of varying relatedness to West Eurasians in PCA ”.
A 2013 study using the single-nucleotide polymorphism, shows that the genome of Andamanese people is closer to those of other Oceanic Negrito groups than to that of South Asians.
According to Basu et al. 2016, further analysis revealed that the genomic structure of mainland Indian populations is best explained by contributions from four ancestral components. In addition to the ANI and ASI, Basu et. al identified two ancestral components in mainland India that are major for the Austro-Asiatic-speaking tribals and the Tibeto-Burman speakers, which they denoted as AAA and ATB respectively. The study also infers that the populations of the Andaman Islands archipelago form a distinct ancestry, which "was found to be coancestral to Oceanic populations".
The cline of admixture between the ANI and ASI lineages is dated to the period of c. 4.2–1.9 kya by Moorjani et al., corresponding to the Indian Bronze Age, and associated by the authors with the process of deurbanisation of the Indus Valley Civilization and the population shift to the Gangetic system in the incipient Indian Iron Age. Basu et al. suggests that "Dravidian tribals were possibly widespread throughout India before the arrival of the Indo-European-speaking nomads" and that "formation of populations by fission that resulted in founder and drift effects have left their imprints on the genetic structures of contemporary populations". The geneticist PP Majumder has recently argued that the findings of Reich et al. are in remarkable concordance with previous research using mtDNA and Y-DNA:
Lazaridis et al. notes "The demographic impact of steppe related populations on South Asia was substantial, as the Mala, a south Indian Dalit population with minimal Ancestral North Indian along the 'Indian Cline' of such ancestry is inferred to have ~ 18 % steppe-related ancestry, while the Kalash of Pakistan are inferred to have ~ 50 % steppe-related ancestry." Lazaridis et al.'s 2016 study estimated steppe related admixture in South Asians. Lazaridis et al. further notes that "A useful direction of future research is a more comprehensive sampling of ancient DNA from steppe populations, as well as populations of central Asia, which may reveal more proximate sources of the ANI than the ones considered here, and of South Asia to determine the trajectory of population change in the area directly.
Pathak et al. 2018 concluded that the Indo-European speakers of Gangetic Plains and the Dravidian speakers have significant Yamnaya Early-Middle Bronze Age ancestry but no Middle-Late Bronze Age Steppe ancestry. On the other hand, the "North-Western Indian and Pakistani" populations showed significant Steppe_MLBA ancestry along with Yamnaya ancestry. The study also noted that ancient South Asian samples had significantly higher Steppe_MLBA than Steppe_EMBA. The study also suggested that the Rors could be used as a proxy for the ANI.
David Reich in his 2018 book Who We Are and How We Got Here states that the 2016 analyses found the ASI to have significant amounts of an ancestry component deriving from Iranian farmers, with the remaining 75% of their ancestry deriving from native South Asian hunter-gatherers. He adds that ASI were unlikely the local hunter-gatherers of South Asia as previously established, but a population responsible for spreading agriculture throughout South Asia. In the case of the ANI, the Iranian farmer ancestry is 50%, with the rest being from steppe groups related to the Yamnaya.
, similarly, conclude that ANI and ASI were formed in the 2nd millennium BCE. They were preceded by a mixture of AASI ; and Iranian agriculturalists who arrived in India ca. 4700–3000 BCE, and "must have reached the Indus Valley by the 4th millennium BCE". According to Narasimhan et al., this mixed population, which probably was native to the Indus Valley Civilisation, "contributed in large proportions to both the ANI and ASI", which took shape during the 2nd millennium BCE. ANI formed out of a mixture of "Indus Periphery-related groups" and migrants from the steppe, while ASI was formed out of "Indus Periphery-related groups" who moved south and mixed further with local hunter-gatherers. The ancestry of the ASI population is suggested to have averaged about 73% from the AASI and 27% from Iranian-related farmers. Narasimhan et al. observe that samples from the Indus periphery group are always mixes of the same two proximal sources of AASI and Iranian agriculturalist-related ancestry; with "one of the Indus Periphery individuals having ~42% AASI ancestry and the other two individuals having ~14-18% AASI ancestry".
A genetic study by Yelmen et al. shows that modern South Asian populations are generally closest to West-Eurasians. They concluded that modern South Asians are basically a mixture of a native South Asian genetic component and a later-arriving West-Eurasian component. The authors also argue that the native South Asian genetic component is distinct from the Andamanese, and that the Andamanese are thus an imperfect proxy. This component was not detected in the northern Indian Gujarati, and thus it is suggested that the South Indian tribal Paniya people would serve as a better proxy than the Andamanese for the "native South Asian" component in modern South Asians, as the Paniya are directly derived from the natives of South Asia.
Two genetic studies analysing remains from the Indus Valley civilisation, found them to have a mixture of ancestry: Shinde et al. found their samples to have about 50-98% of their genome from peoples related to early Iranian farmers, and from 2-50% of their genome from native South Asian hunter-gatherers sharing a common ancestry with the Andamanese, with the Iranian-related ancestry being on average predominant. The samples analyzed by Narasimhan et al. had 45–82% Iranian farmer-related ancestry and 11–50% AASI. The analysed samples of both studies have little to none of the "Steppe ancestry" component associated with later Indo-European migrations into India. The authors found that the respective amounts of those ancestries varied significantly between individuals, and concluded that more samples are needed to get the full picture of Indian population history.
Genetic distance between caste groups and tribes
Studies by Watkins et al. and Kivisild et al. based on autosomal markers conclude that Indian caste and tribal populations have a common ancestry. Reddy et al. found fairly uniform allele frequency distributions across caste groups of southern Andhra Pradesh, but significantly larger genetic distance between caste groups and tribes indicating genetic isolation of the tribes and castes.Viswanathan et al. in a study on genetic structure and affinities among tribal populations of southern India concludes, "Genetic differentiation was high and genetic distances were not significantly correlated with geographic distances. Genetic drift therefore probably played a significant role in shaping the patterns of genetic variation observed in southern Indian tribal populations. Otherwise, analyses of population relationships showed that all Indian and South Asian populations are still similar to one another, regardless of phenotypic characteristics, and do not show any particular affinities to Africans. We conclude that the phenotypic similarities of some Indian groups to Africans do not reflect a close relationship between these groups, but are better explained by convergence."
A 2011 study published in the American Journal of Human Genetics indicates that Indian ancestral components are the result of a more complex demographic history than was previously thought. According to the researchers, South Asia harbours two major ancestral components, one of which is spread at comparable frequency and genetic diversity in populations of Central Asia, West Asia and Europe; the other component is more restricted to South Asia. However, if one were to rule out the possibility of a large-scale Indo-Aryan migration, these findings suggest that the genetic affinities of both Indian ancestral components are the result of multiple gene flows over the course of thousands of years.