Angkor Wat is an icon of architecture. Arguably one can speak of it in the same breath as the pyramids of Giza or the Taj Mahal. Angkor Wat is a concrete manifestation of the apogee of Khmer civilization, which extended to the Chao Praya to the west, and the estuary of the Mekong to the east, at its height. As is evidenced by the fact that it began its life as a Hindu temple complex, Indian influences bled deeply into the high culture of much of mainland Southeast Asia during the centuries before 1000 A.D.. The king who initiated the building of Angkor Wat had the throne name Suryavarman. Its South Asian cadence is pretty unmistakable.
Today mainland Southeast Asia to the west of Vietnam is dominated by Theravada Buddhism, not Hinduism. But the Indian tincture persists. Theravada Buddhism is the dominant religion outside of Southeast Asia in only nation, Sri Lanka, in South Asia. Additionally, Southeast Asia harbors the world’s only native Hindu ethnic groups outside of the Indian subcontinent. Famously the Balinese of Indonesia, and less well known, the Hindu Cham of Vietnam, as well as various Javanese communities such as the Tenggerese (there are animist groups which are aligning with Hinduism in Indonesia, but that’s a recent phenomenon). The kings of Thailand strongly support the Theravada Buddhist religion, but their courts also sponsor the services of Hindu Brahmins. And the native scripts of Southeast Asia tend to have to South Indian origins.
This world of “Greater India” shattered in the centuries between 1000 and 1500 A.D. The rise of Islam along the Straits of Malacca in the century or so after 1000, and the eventual spread of the religion until it broke through to the interior of Java in the 16th century, is well known. But nearly contemporaneous with the rise of Islam in maritime Southeast Asia mainland Southeast Asia was subject to a massive migration of Tai warbands. Anyone curious about the whole story is recommended to read Strange Parallels: Southeast Asia in a Global Context, c. 800-1830. The author works through the slow process of nation-state formation in the 2,000 years between prehistory and modernity. The varied indigenous polities dealt with the challenge ways. In the west the Tai established a foothold, but could not overwhelm the Burman or Mon societies and political institutions. The Shan states were born. In the east, in Vietnam, the Tai were repelled or assimilated in totality. In fact, the Vietnamese absorbed Champa on the central coast, and began their long push toward the Mekong delta, expanding Sinic civilization at the expense of Indic (the Vietnamese emulated the Chinese model, and their popular religious cults were based on Mahayana Buddhism). In the center the Tai were victorious in near totality. Modern day Thailand, like modern day France, takes the name of its conquerors. And like the Franks the Thai absorbed most of their high culture from the Khmer and Mon whom they defeated. Unlike the Vietnamese, the Thai did not emulate Sinic forms of governance or promulgate Confucian ideology. Rather, the Tai warriors took on the mantles of the Khmer kings, and became sacral kings in an Indian sense, just as the long-haired Merovingians became bathed in Romanitas with their conversion to Catholic Chrisitanity. Though unlike the Franks, and like the Anglo-Saxons, the militarized bands maintained their linguistic identity, until their language superseded the Mon and Khmer dialects previous dominant. Modern day Cambodia only exists in large part because European colonialism sheltered it from total absorption into the Siamese Empire, which was digesting it in pieces when the French absorbed the Khmer monarchy.
Which finally brings me to the presumption of cultural diffusionism. The impact of Indian culture on Southeast Asia is undeniable. But that does not necessarily entail that Indians, as people, engaged in migration. The Chola invasion of Sumatra in 1025 is viewed as a somewhat singular event. Though there are legends of Indian ancestry of the Khmer kings, these can easily be chalked up to the sort of foundational myth-making which is normal for many cultures (e.g., descent from the House of David or Trojans were common for European noble houses). Though there are plenty of historical references to the ancestors of the Peranakan Chinese going back to Zheng He’s voyage, references to Indians are fewer, and must be gleaned through their cultural influence, both Dharmic and Islamic. After all, Christianity spread in Europe not through migration, but through conversion. Similarly, Islam also spread through the conversion of elites in Sub-Saharan Africa, and Central and Southeast Asia.
I no longer think cultural diffusion is the only explanation for why Hinduism and Indian civilization struck such deep roots in Cambodia. Rather, I think it is clear that there was a migration of people whose ultimate origins were in South Asia. The arrival of South Indian motifs to Southeast Asia was accompanied by South Indian people, and in some areas lots of them (or at least enough that their outsized demographic impact left its mark). We need to move beyond an excessively strong null hypothesis of cultural diffusionism. Migrations can echo down the generations and across continents. The agro-pastoralism triggered by the expansion of farmers out of the Near East thousands of years ago eventually cascaded all the way down to the South African Cape, and the genetic signal of these Eurasians is present in the Khoe people (almost certainly mediated by Nilotic pastoralists who pushed in from East Africa).
The evidence of Indian genetic imprint on Cambodia has been staring us in the face for nearly 10 years. Above is a STRUCTURE bar plot from the paper which brought the HGDP-CEPH panel into the post-genomic era. They took their 650,000 SNPs, and allowed STRUCTURE to assign ancestral quanta to each individual given a particular value of K populations. One thing that caught my eye immediately was the Cambodian affinity to South Asians. The bar plot generated a component, light blue, which was modal in South Asians from Pakistan. The Cambodians had low levels of this. The other Southeast Asian groups did not exhibit any signal of this ancestral component.
When Joe Pickrell came out with TreeMix he saw this:
Two inferred edges were unexpected. First, perhaps the most surprising inference is that Cambodians trace about 16% of their ancestry to a population equally related to both Europeans and other East Asians (while the remaining 84% of their ancestry is related to other southeast Asians). This is partially consistent with clustering analyses, which indicate shared ancestry between Cambodians and central Asian populations…To confirm that the Cambodians are admixed, we turned to less parameterized models. The predicted admixture event implies that allele frequencies in Cambodia are more similar to those in African populations than would be expected based on their East Asian ancestry.
I asked Joe whether he thought that this admixture had something to do with “Ancestral South Indians” (ASI). He said it was plausible. But I didn’t follow up on this at the time.
Recently a friend asked me to project his genotype on a PCA of Southeast Asian and Northeast Asian populations. For kicks I decided to throw in people who were not of non-East Asian ancestry. Myself, and a few friends who were mostly European. What I did was place the individual genotypes on top of the variation of the target populations. If I didn’t do this then my white friends and myself would just hog up the first PC as non-East Asian vs East Asian. To my surprise all of us non-East Asians landed right on top of the Cambodian cluster (I was shifted toward other Southeast Asians though, which is reasonable given my genetic background). I had naively assumed that we’d just land on the (0,0) position, except for me, who would be somewhat shifted off, because most of us would be symmetrically related to all these East Asian groups.
What was happening though was that something in the Cambodians was capturing genetic variation in individuals with West Eurasian ancestry. We now know that groups in the STRUCTURE plot above which are nearly 100% “South Asian” are actually highly skewed toward “Ancestral North Indians” (ANI), which is a West Eurasian group. It seemed implausible that the Cambodian signal would be due purely to shared ASI-like ancestry which was present from the Indus to the South China Sea before the incursions of agriculturalists from the northwest and northeast.
An initial hypothesis is that these individuals are admixed with Indians who arrived during the period of European colonialism. But there are problems with this. Historically there were many Indians who arrived in Burma and Malaysia. But far fewer in eastern mainland Southeast Asia. Additionally, observe that the Indian admixture in the Cambodian samples is relatively similar in fraction across all the individuals. That’s a major tell that the admixture is old enough for it to have distributed itself across the population.
As I implied above I have a private data set of Cambodians, with a sample size larger than the HGDP. I also have Filipinos and Vietnamese. I combined them with 1000 Genome data sets, and produced the plot below:
Click the figure to see what’s going on, but it’s pretty straightforward. PC1 separates Eurasia on a east to west axis. My North Indian population of Punjabis is at the top right. Then at the far left are crammed Japanese, Koreans, and North Chinese. At the bottom are Papuans from the HGDP. What’s interesting to me, and suggestive, is that the Cambodians are the most West Eurasian shifted of the East Eurasian populations. The Filipinos also exhibit a cline, but they are shifted toward the Papuans. As you may know there are “Negritos” in the Philippines, and their non-Austronesian ancestry has affinities to that of Melanesians. Following the convention in archaeology I will refer to the Negrito people as “Austro-Melanesian.”
In a 2011 paper David Reich’s group posited that there are two major groups of Austro-Melanesians. One clade consists of the Andaman Islanders and the Negrito people of inland Malaya, and another component is found among the Philippine Negritos and the peoples of Near Oceania, the Melanesians and Australian Aboriginals. These results are not consensus. But, they suggest that the history of pre-agricultural Southeast Asia is likely to be complex.
Of course I ran TreeMix on the data (170,000 markers with very little missingness):
First, a minor point. I included the 1000 Genomes Bengali data set mostly as a positive control. I wanted to make sure TreeMix was behaving right. Well, I think I found out something new that pushed forward the conclusion I came to in my earlier post, the East Asian ancestry in Bengalis can’t just be due to old Austro-Asiatic admixture. I didn’t add any Chinese samples in the original analysis because it seemed clear to me that they wouldn’t be relevant. Well, if you look through these TreeMix results you see that the gene flow edge is actually often between the Southeast Asian groups and the Southern Chinese! I think this points to something of the composite nature of the admixture. It was probably a Tibeto-Burman group which had absorbed Austro-Asiatic elements along the way.
The second issue is that the Cambodians consistently have a gene flow arrow coming from a population that is related to the Indians. I added the Papuans specifically to pick up the Austro-Melanesian signal. The archaeology and mtDNA strongly indicates that a substantial minority of the ancestry of Southeast Asians derives from the pre-agricultural Austro-Melanesians. The Austro-Asiatic groups were the first to arrive, so it makes sense if they have the highest fraction of Austro-Melanesian ancestry. Stereotypically Cambodians are darker skinned and curlier haired than their Thai and Vietnamese neighbors.
That being said, how can we rule out the possibility that this gene flow isn’t due to ASI-like ancestry shared between Indians and Cambodians? After all the gene flow edge derives from near the root of where the Indian and Papuan clade diverge. I decided to add another population, which I will label “ASI-rich Indian.” I basically took the data from the Estonian Biocentre and yanked out individuals which didn’t have East Asian ancestry but had more ASI than my middle-caste South Indian data set. Most of these populations where South Indian Dalit or Tribes. Perhaps this population would be a better fit to the donor group for the Cambodians?
Well that didn’t help! Now the Cambodians are often just a basal Indian populations with a lot of Dai admixture. Adding the ASI-enriched group just made the tree more Indo-centric, so it put the Cambodians on the Indian clade. Let’s narrow the set of populations under consideration.
OK, that’s better. Two things:
1) The Indian-like gene flow to the Cambodians is clear. It’s more Indian-than-Papuan.
2) The Filipinos now have a Papuan-like gene flow. This could be because of the Austro-Melanesian connection between these two groups. Remember the Reich paper above indicated that they had shared ancestry which was distinctive from that of the Austro-Melanesians who contributed to the ancestry of mainland groups (and distantly to Indians).
Next I decided to add Australian Aboriginals and Taiwanese Aborigines into the analysis. These are related populations to some of the ones already in the data set.
OK, the patterns reoccur, and the relationship between the sister clades is what we’ed expect. Though TreeMix does not tell us time, the fact that the gene flow into the Filipinos from the Oceanians is below the node where Papuans and Australians diverge probably suggest that the Austro-Melanesians of maritime Southeast Asia diverged more than 10,000 years ago. That’s pretty reasonable if you follow the natural history.
Finally, I’m going to remove Indians, and put in a true West Eurasian population: Armenians. I select Armenians because I think they’re close to the ancestral population of ANI which resulted in the non-Brahmin populations of Southern Indian (and to a large extent Northern India). I believe that the Indo-Aryans were a secondary migration which had an impact on the upper castes and the Northwest. Since the cultural connections of Southeast Asian Indian culture are to southern, not northern, India, the West Eurasian affinity would ultimately be from this region. I don’t have Y or mtDNA data on most of the Cambodian samples, but I do have Y on three of them. And one of those is Y chromosomal lineage J2. That’s commonly found in West Asia, and among some groups in South Asia. Additionally, Y haplogroup group H is found among the Khmers. This haplogroup is modal among low caste and southern Indian populations.
It’s not consistent, but it does look like the gene flow into Cambodians has West Eurasian affinities. Remember, the PCA that I mention above already suggests this. ASI, of whom the Andaman Islanders are a cousin lineage, are not any closer to Europeans than East Asians. East Asians and Southern Eurasians seem to form a very deep clade as contrasted to West Eurasians. The affinities are deep in the Pleistocene, but they’re there (e.g., mtDNA haplogroup M instance). That’s why those South Asians with more ASI ancestry get pushed “east” along the PCA plane. Mind you, I think that really there are two admixture events in rapid sequence. First, the Austro-Asiatic farmers spread across Southeast Asia 3 to 4 thousand years ago, and mixed with Austro-Melanesian groups. These are sister lineages to ASI and Andaman Islanders. Then, 1 to 2 thousand years ago you have a gene flow event mediated by a large number of South Indians. The admixture is probably smaller in magnitude than that of Austro-Melanesians. That’s probably why the gene flow edge keeps getting dragged toward the Papuans: you are seeing a composite edge of two admixture events which occurred sequentially.
This is all rather strange, and I hope that those with some knowledge of Southeast Asian history can not take in the new evidence and wonder how it affects previous interpretations. Those Indian warlords in Khmer myth may not be so mythical after all!
When making the case for the monopoly of Christianity on the public space of late antiquity St. Ambrose famously declared to the pagan senator Symmachus that “There is no shame in passing to better things.” St. Ambrose more broadly was setting the precedent for religious intolerance of falsehood, and exclusive promulgation of one singular truth from on high. The Christians of that period were clearing away the organically accrued institutions and folkways of pagan antiquity, sweeping aside the detritus and making it clean in their eyes. They believed that history was on their side, and they did not look backward. How times change. Just as the most confident of the New Atheists declare that humanity’s childhood is nearing an end, and so the need for God, so the Christians of antiquity believed that in their religion the revelation of the prophets and insights of the philosophers came to maturation. It was time to put childish things behind us.
In some ways Thor Heyerdahl is one of those “childish things.” Heyerdahl was active in my childhood. And despite being an old man his imagination was still fertile. I recall reading some of his books, and laughing even then. It struck me that his bizarre theories were no better than speculations about Lemuria. Here was a Madame Blavatsky for our age!
To some extent I’m being unfair to Heyerdahl. Some of his ideas were genuinely crazy, but he was pushing the envelope of speculative possibilities. He was a thinker who would have been more at home in the late 19th century, though no doubt his life would have been at risk since he almost certainly would attempt to discover “Lost Civilizations” in the heart of Africa. Heyerdahl was attracted to those “blank spaces” on the map, and in his adulthood those zones were filled in. The eye of the satellite became the all-seeing eye of Sauron. So he fled to the past, and wove speculations from his amorphous diffusionism. He posited strange and implausible contacts leading to syntheses between diverse peoples.
Well, he had a point. I believe that perhaps it is time to turn time back. Just as the Christ of St. Ambrose is now receding from the public spaces of Western civilization, becoming another cult among cults, so the clean and singular vision of archaeology and prehistory after World War 2 needs a reformation. Here is a question: would we believe that the Malagasy could exist if they did not exist? That is, would we think it possible that a small group of Austronesians from southern Borneo would somehow be the first humans to settle the great island off the southeast coast of Africa? It’s so fantastical I doubt it. The clear cultural impact of Malayo-Polynesians on East Africa is only plausible because the Malagasy exist. Otherwise they’d have been explained away. In addition, there is now suggestive evidence of pre-Columbian Polynesian and South American contact.
On the edges of history, in the lacunae between where text speaks to us, there were occurrences which would surprise us. The genetics sends clear signals that our race of adventurers and travelers continued to mingle and find ways to interact even after the emergence of great agricultural civilizations.. We have always sought out the “blank spaces” on the map, and filled them in our imaginations.
Postscript: Well, this post took half my day. About half running the analyses and generating the input data sets. And the other half writing. It’s a pretty strange world we live in! Special thanks to Chris Chang, whose new version of Plink is so fast and has so much incredible functionality. It really speeds up the process of data exploration. And similarly, a shout out to Joe Pickrell, whose TreeMix is truly addictive.