The true source of the steppe ancestry in modern Indians

 

This post aims to clarify the source of the bronze age steppe ancestry in modern Indians. For this, I have chosen the following targets.

Modern targets and their details

I have not chosen NW Indian samples because that region possibly has had multiple genetic introgressions which could muddle the signals (I have modelled Kalash with qpAdm before in this post, and the results do not differ much from what I will show below). Such is likely not the case for east and south India where the steppe ancestry would have arrived from one or a maximum of two sources.

TOOLS AND METHODS

For each of the private samples, I extracted the SNPs common with Harvard 1240k data using Plink v1.9. Each of these samples was then merged with the Harvard v54.1 1240K dataset.

I will use the Admixtools 1 package on Linux. Relevant labels will be run through qpfstats() first with parameters 'allsnps: YES' and 'inbreed: NO', and the F4 stats output will be passed as input to qpadm(). I will record the qpadm() output - p-value for the model, the admixture weights and standard errors for each model tested for each target. Models with P-values > 0.05, all positive admixture weights and low std. errors will be accepted and marked green. Models which do not pass these criteria will fail and will be marked in red. If no models pass, then I will note the models with the highest p-values and all positive admixture weights.

I will run a rotating strategy (Harney et al, 2021) for each of the 11 targets. The following labels will be used as fixed references (will not be tested as a source).

Fixed References

List of sources to be tested

There are 4 steppe sources as alternatives. 2 SC Asian sources, 1 east Iranian source and 5 local Indian sources. Of the local sources - I will only use either Irula or Roopkund_PallanLike in a model because they are similar. It turns out that the Roopkund_PallanLike label fits North Indians better whereas Irula fits southern targets better. Both these sources have minuscule to no steppe ancestry and therefore are decent proxies for the underlying non-steppe ancestry of India. 

So, for each target, I will test combinations from these 11 sources - 4 steppe, 4 local (Irula or Roopkund_pallanLike but not both), 2 SC Asian, and 1 East Iranian. The remaining sources and the fixed references will be used in the reference list for that particular model.

It would be appropriate for me to say a few things about the RoopkundA_PallanLike label. The 2 samples under this label are the only 2 R1a found from Roopkund_A (the 850 CE skeleton cluster, B cluster is from ~1800 CE comprised of Mediterranean people) among 13 males but have ~0 steppe ancestry. None of the other 11 male samples with higher steppe ancestry are R1a. I have written about the Roopkund samples here. I have named them 'PallanLike' because on PCA they cluster with modern Pallans and some Irulas. However, they have slightly higher northern ancestry than Pallan or Irula (4-5% more Siberian ancestry) and they also make better models with northern targets. So they are likely to be northern locals rather than southern migrants to the north. A similar high AASI profile was also found from I10409, an IVC migrant at Gonur in Turkmenistan with ~50% AASI/Onge ancestry.

A few caveats before I proceed:

1. Modeling modern targets with ancient sources has issues because ancient samples have deamination damage whereas modern samples do not. This damage (C--->T and G--->A, especially at ends of the strand). Some of the ancient samples used here are not even UDG-treated. UDG treatment removes some of the deamination damage. Harney et al, 2021 show that good qpadm models might fail while modelling modern targets if the damage in the ancient DNA is high (because it will reduce affinity between the modern and ancient samples). However, this cannot make otherwise bad models pass the p-value threshold. 

2. 4 private samples have a relatively low SNP count (<150k). This is moderate but not too bad (<100k is considered very low). This has the effect of making some otherwise bad models also pass the p-value threshold. It also increases the std. errors of the admixture weight output. However, this cannot make good models fail.

3. The more the number of populations in the reference list of the qpadm model (10 fixed + 9 rotating = 19 in a 2 source model), the more it could suppress the p-value output of the model. 

4. There is a close to 100% chance that the 5 local sources do not cover the variance within the pre-steppe population of India. So, regardless of what the passing models say, with new data and more choices, the working models will keep changing. Models should not be taken too literally eg. Irula being chosen as a successful source does not literally mean that the Irula ancestors admixed with the other sources.

5. A rotating model strategy checks each source for fit in an unbiased manner. However, the list of sources chosen introduces bias. Since there is a limit to the number of sources I can test (due to very high program runtimes) there could possibly be some Labels not present in this analysis which could be the true source. I have tried to ensure that the most relevant theorized sources are included here.

#1 & #3 suppress p-value, whereas low SNP count in #2 pushes up p-value. With these factors in mind, a valid model with a p-value > 0.05 should be considered good. High p-values (>0.50) are often not seen at all while modelling modern targets with ancient samples.

Telugu Niyogi Brahmin

Niyogi Brahmin models

Let me explain the screenshot once. All feasible models have been arranged in descending order of the p-value. The models in green are the ones which have a p-value > 0.05. In this case, the models which pass are 

1) Irula + Turkmenistan_IA and 

2) Irula + Gonur_BA_1 (BMAC)

The admixture weights are 81.2% + 18.8% and 79% + 21% for each model respectively. The std errors for model 1 are 2.9% (ie Irula: 81.2 +- 2.9% and Turkmenistan_IA: 18.8% +- 2.9%)

This person has a low steppe admixture among Brahmins. This is also validated by his G25 modelling, and PCA using the G25 coords. He can also be modelled without steppe (Irula + Bmac, which may also be a proxy for a northern shifted local pre-steppe population without actual BMAC ancestry).

The closest person by G25 distance is Gujarati, also a low steppe population.

On G25, he shows 5% steppe ancestry. His Y hg is H-M69, which is local and mtDNA H2a which is near the eastern/steppe.

Brahmin.DG from Vizag

Model for Brahmins from Vizag 

All 2 source models fail.

Brahmin from Vizag -  3 source models

Multiple 3 source models pass. All four steppe sources pass in some combination, although Sintashta and OyDzhaylau only pass in combination with Sarazm ancestry. In contrast, Tkm_IA and Kangju pass with other local sources as well, including Shahr-Sokhta_BA2 (IVC).

The combination of Irula + 2nd (lower AASI) local source tells us that the steppe source admixed with a population which was more northern in ancestry than Irula.

On a side note, no three-source model with RoopkundA_PallanLike in the place of Irula has p>0.05.

Tamil Brahmins

Tamil Brahmins - All 2 source models with Irula fail

All 2 source models fail, but Irula + Tkm_IA have the highest p-value of 0.003. Next, I test 3 source models.

Tamil brahmins - 2 sources with RoopkundA_PallanLike

However, with RoopkundA_pallanLike instead of Irula, we have a passing 2 source model

~70% RoopkundA_PallanLike + ~30% Tkm_IA.

Tamil SC

Tamil SC 2 source

Tamil Scheduled Caste samples can be adequately modelled with Irula and a northern local source without steppe.

Kashmiri Pandit

Pandit 2 source models

All 2 source models fail, but Tkm_IA + RoopkundA_PallanLike has the highest p-value.

Pandit 3 source models

The only passing model for Pandit is Tkm_IA + Sarazm + RoopkundA_PallanLike

Tyagi UP

Tyagi 2 src models

The only passing model for Tyagi_UP is Tkm_IA + RoopkundA_PallanLike (40% + 60%).

UP Brahmins (Bhojpuri)

UP Brahmin 2 src models fail

All 2 source models fail. However, TKM_IA + R_PallanLike has the best p-value.

UP Brahmin 3 source models fail

All 3 source models fail as well. However, the Turkmenistan_IA model has the best p-value, followed by Kangju.

UP Scheduled Caste

UP SC 2 source models fail

All 2 source models fail ( so do 3 source models). The best model is Tkm_IA/Kangju + R_PallanLike

Saryuparin_Dwivedi

Dwivedi 2 source models

All 2 source models fail. The Tkm_IA model has the best p-value.

Dwivedi 3 source models

Only the model with Kangju as a source passes. Turkmenistan_IA ranks second.

Rajput from East UP

Rajput from East UP 2 source models

The only passing model is with Turkmenistan_IA.

Rajput from UP/MP

UP MP Rajput 2 source models

Only the model with Tkm_IA as a source passes.

DISCUSSION

The results solve a dilemma which I have been facing since the Narasimhan et al paper from 2019. They claimed that Indians received steppe ancestry in the bronze age but not BMAC-related ancestry. Did they somehow skip mixing with BMAC ancestry which was ubiquitous on the path from Steppe to India? If unmixed steppe people were present near NW India in the bronze age, why are there no archaeological settlements with Andronovo materials and Kurgans?

It turns out that the common sense logic was true, and scepticism was warranted. From the passing models of the above 11 targets, it appears that Turkmenistan_IA from 850 BCE is the best source of steppe ancestry for modern Indians. This sample is from the Yaz II culture and can be modelled as BMAC + Steppe_MLBA in a 43/57 ratio. No east Asian ancestry is detected in this sample. (Guarino-Vignon et al, 2021)

The older samples in Turkmenistan are from the Parkhai and Sumbar LBA sites, dating to ~1400 BCE. Those do not show elevated steppe ancestry. Only one of those 3 samples, I6667, shows some steppe ancestry to the tune of 10%. Similar samples cannot be the source of steppe in Indians.

LBA samples from Turkmenistan have little steppe ancestry and cannot be a source of the steppe for modern Indians.

Therefore it seems likely that TKM_IA-like ancestry took hold in Turkmenistan only post 1000BCE when Yaz II started. 

The amount of east Asian ancestry in Turkmenistan and Uzbekistan remains low even in the Iron age. For example, the ~100 BCE samples from Rabat and Sherkharakat in Uzbekistan only have additional 5-7% east Asian ancestry over BMAC (Kumar et al, 2021). The present-day Uzbek populations show additional east Asian and Siberian ancestries over the 100 BCE samples, proving a later entry of extra east Asian ancestry (Kumar et al, 2021). Modern Indians do not prefer a steppe source heavy in East Asian ancestry (like the Śāka from Kyrgyzstan) but minor east Asian ancestry is tolerated well by qpAdm (Narasimhan et al, 2019 model Indians successfully with Kangju; and the above targets work well with Kangju if Tkm_IA is removed from competition).

The Turkmenistan_IA ancestry survives even today in the Tajik Yaghnobi population. They are modelled as deriving 88-93% of their ancestry from a Turkmenistan_IA like population, the rest being East Asian (Guarino-Vignon et al, 2021). The Yaghnobis speak an Eastern Iranian dialect (related to Sogdian) whereas the Tajiks speak a dialect of Southwestern Iranian (Persian). 

Using DATES, Guarino-Vignon et al date the east Asian admixture in Yaghnobis to 1019+-447 years ago, ie ~100CE at the earliest (95% confidence = 2x of std error = ~900, ie ~1900 yrs ago)

Given this, it is safe to conclude that the admixture of this population into Indians likely occurred between 1000 BCE and 100 CE. 

Yaghnobi Men: From wiki

Evidence from Archaeology and Literature

Steppe ancestry is first seen at Swat valley around 1200 BCE (15-20%), but the ancestry probably trickled in post-1700 BCE (Narasimhan et al, 2019). However, as seen from the above models, Swat samples are not a good source of ancestry for modern Indians. Furthermore, there is no archaeological, literary or genetic evidence to suggest that steppe ancestry reached deeper into India in that period. The crude theory that 'Steppe Aryans' ransacked IVC settlements causing their downfall is not supported anymore. As per the latest research - the drying up of the Saraswati, drought and a shift in monsoonal patterns are said to be the main catalysts (Malik, 2020; Sengupta et al, 2019; Chatterjee et al, 2019). Despite the lack of evidence, Dr David Reich from Harvard in his 2019 book maintains:

But the lack of archaeological evidence does not mean that there were no major incursions from the outside. Between sixteen hundred and fifteen hundred years ago, the western Roman Empire collapsed under the pressure of the German expansions, with great political and economic blows dealt to the western Roman Empire when the Visigoths and the Vandals each sacked Rome and took political control of Roman provinces. However, there so far seems to be little archaeological evidence for destruction of Roman cities in this time, and if not for the detailed historical accounts, we might not know these pivotal events occurred.” It is possible that in the apparent depopulation of the Indus Valley, too, we might be limited by the difficulty archaeologists have in detecting sudden change. The patterns evident from archaeology may be obscuring more sudden triggering events.

This is a very flippant statement, but it does give me a peek into the worldview of these scholars. Visigoths and Vandals had no noticeable linguistic impact on the Romans, and their Germanic language is extinct today. This is in stark contrast to Reich's claim that the invisible steppe men who invaded IVC changed the whole linguistic landscape of Northern India, so much so that all native languages became extinct without leaving records. How is that even possible? On the other hand, invasions in India by Achaemenids, Śāka, Greeks, Kushans, and Turko-Mongols are well attested - none of them was able to impose their languages.

Leaving fantasy behind, let us look at some hard data.

Hemphill et al (1991) concluded a biological discontinuity between 800 BCE and 200 BCE at Harappan sites. It would seem that the steppe ancestry introgression into Indus valley sites post 800 BCE is supported by anthropometric data.

As for the question of biological continuity within the Indus Valley, two discontinuities appear to exist. The first occurs between 6000 and 4500 BC and is reflected by the strong separation in dental non-metric characters between neolithic and chalcolithic burials at Mehrgarh. The second occurs at some point after 800 BC but before 200 BC. In the intervening period, while there is dental non-metric, craniometric, and cranial non-metric evidence for a degree of internal biological continuity, statistical evaluation of cranial data reveals clear indications of interaction with the West and specifically with the Iranian Plateau.

I am inclined to accept this proposal. In the next section, I will propose some routes through which this ancestry could have come into the ancestors of modern Indians.

1. Śāka presence in Punjab in the Pre-Achaemenid Period

Petrie and Magee (2011) claim a connection between Yaz culture and the assemblages found in the Bannu basin of Pakistan between 900-600BCE. However, apart from those two scholars, I have not found confirmation of this connection. The Wikipedia article puts the Bannu ceramics as part of the Indian Northern Black Polished Ware culture.

The Bannu Archaeological Project excavations at the sites of Ter Kala Dheri and Akra between 1995 and 2001 revealed evidence for a regionally distinct cultural assemblage marked by what has been referred to as Bannu Black on Red Ware, which has been dated to the early-mid 1st millennium BC, thus making it pre-Achaemenid (c. 900-600 BC; Khan et al. 2000c: 81-100; Magee et al. 2005; Magee and Petrie 2010).

The two most characteristic ceramics are Bannu Black on Red Ware and globular spouted

vessels (labelled Assemblage 2). Neither of these ware types has been reported elsewhere in the western borderlands of the subcontinent, and the closest technological, morphological and stylistic parallels for Bannu Black on Red Ware are to be found in the early Iron Age (Yaz Depe I) cultures of south-west Central Asia (Magee et al. 2005; Magee and Petrie 2010).

I discovered something most interesting in VS Agarwala's 1953 book on Pāṇini. Pāṇini was a Gandharan Brahmin, expert philologist and Sanskrit grammarian who lived in the period between 600-400BCE. Since his era coincides with that of the Achaemenid rule, his works can shed light on the NW Indian region during his era. Panini gives interesting information regarding a place name suffix that was in use in Uslnara and VarNu (Bannu) during his time. He names various places names ending with 'kanthA' in the region of Punjab and Bannu.

'kanthA' - 'city' is a Śāka suffix whereas 'Kanda'- 'city' is a Sogdian suffix, as is informed to us by Kharoshthi inscriptions. SC Asia still abounds in 'kanthA' ending place names, as Agarwala informs us - Samar-kand, Panj-Kand, Tash-kent, etc. This is indicative of Śāka presence in Punjab long before Pāṇini's time and long before the later Indo-Śāka or Kushana rule.

The Connection made by Petrie and Magee between Bannu ware (900-600BCE) and Yaz culture is probably an answer to Agarwala's quandary. It would seem that Śāka were present in Punjab, not just in the Oxus in the pre-Achaemenid times.

It is possible that these Śāka comingled with the Hindu upper classes and mediated steppe ancestry in the ancestors of modern Indians. However, no other information about this period exists to confirm this.

2. During Achaemenid Rule

It is accepted that the Yaz III period (700-400BCE) saw the region come under the rule of the Achaemenid empire (Lhuillier et al, 2013; Basafa & Davari, 2021). From the Behistun inscription, it is known that the Achaemenid King Darius I ordered his Bactrian satrap to crush the Margiana revolt, aka 'Revolt of Frāda' in 522 BCE. 55,243 of Frāda’s followers were killed and 6,972 were captured (Encyclopædia Iranica).

Achaemenid empire under Cyrus and his successor Darius I (530BCE onwards) had parts of modern Afghanistan and Pakistan as its satrapies. This is clear in the Behistun, Persepolis and Naqsh-E-Rustam inscriptions. The satrapies were - Gandhara, Hindush and Sattagydia. These satrapies were also mentioned by Strabo and Herodotus.

Based on some material found from the Bannu basin dated between 600-300BCE, Petrie and Magee conclude that Akra (Bannu basin) might have been the Achaemenid capital of the Sattygydia satrapy.

Assemblage 1 includes examples of bowls with offset vertical rim, banded beakers, bowls with an s-shaped rim and tulip bowls, and typological comparisons with material from Iran and Afghanistan suggest a dating between 600 and 300 BC for this assemblage (Magee et al. 2005: 724-25). The presence of this material provides some archaeological confirmation for the argument, hitherto based on epigraphic and historical data, that Akra might be the capital of the Achaemenid satrapy of Thatagush

Per David Fleming, Bhir Mound at Taxila remains the best choice for the Achaemenid capital of India

However, it is noticeable that some of the early Bhir Mound pottery from Taxila, whose mid-to-late sixth century BCE date is not in dispute, bears a surprising resemblance to material from contemporaneous Kandahar and southern Iran, where an Achaemenid presence is certain. This would indicate that contacts of various sorts were maintained between the Kandahar region of southern Afghanistan (and hence to the Iranian plateau proper) and the upper Punjab by way of the Gumal Pass area. In geographical terms such a direct route between the fertile Punjab and the important nexus of trade routes meeting at Kandahar eliminated the need to negotiate the mountainous terrain between modern Ghazni, Gardez, Kabul, and the Peshawar vale.It also passed through solidly Achaemenid territory in Sattagydia and reiterates the importance to the empire of Sattagydia as a way station on the route to India.

Until and unless another contender is found, Taxila's Bhir Mound remains the most plausible candidate for the capital of Achaemenid India.

Indian soldiers were part of the Achaemenid army as well, and the 331 BCE battle between Alexander and Darius III is said to have included soldiers from India as well as Bactria and Sogdiana. This suggests that at least some Indian territories were under the rule of Achaemenids between 530 and 330 BCE.

What genetic changes occurred during this rule cannot be estimated, but some mixing between the Hindu elites and the Achaemenid elites is possible. Śāka were also part of the armies and administrators commanded by the Achaemenids.

Another route through which some foreign genetic material could have entered Indians is by incorporating non-Aryans into the Aryan fold. Agarwala cites Pāṇini's work which talks about Vrātyas [non-Aryans or Aryans who have lost the Aryan way] and the Vrātya-stoma rituals to bring them back to the Aryan fold. The simplified rituals are meant for 4 kinds of people - the young, the elderly, ones who are violent, and cultural leaders in the Vrātya society.  

Agarwala writes "It is possible that the converted Vrātya who had been admitted to the Brahmana or Kshatriya fold were spoken of as ब्राह्मण-कृताः and क्षत्रिय-कृताः ." 

Whether Śāka were converted into the Aryan fold through such rituals remains a matter of debate, it is not very clear who exactly were the subjects of this ritual.

3. Greek and Mauryan period

After Alexander won the 331 BCE war with the Achaemenids, he ventured east into the Indus river territory. Strabo notes that one of Alexander's companions (Aristobulus) saw Taxila's dead being fed to vultures - this indicates the presence of Zoroastrianism at Taxila (Vivero, 2020)

His achievements in the east (Battle vs Porus?) lasted only a few decades and by 317BCE the Greek satraps left behind were driven out by the Mauryan empire of Chandragupta Maurya. The 305 BCE battle by Seleucus to take back control resulted in a Greek defeat and a peace treaty was signed. As part of this diplomacy, Seleceus' daughter (?) married Chandragupta Maurya. Perhaps this also signifies the lack of aversion of Indian elites marrying foreigners, or perhaps this was just due to political expediency. The treaty resulted in mutually beneficial diplomatic relations between the Greeks and Mauryas. Perso-Hellenistic art inspired some of the Magadhan architecture (presumably along with the import of Iranian craftsmen). Terracotta figures excavated from Mauryan era Mathura and Sarnath also reveal the presence of foreigners (if not just depictions of foreigners).

a. Persian Nobleman Clad in Coat Dupatta Trouser and Turban - Circa 2nd Century BCE. From wiki

b. Foreigner with Śāka conical hat at Sarnath. From wiki

Ashoka's reign saw the Mauryan empire expand into the south of India. Perhaps it was during this rule that the steppe ancestry from the NW was mediated into the other parts of India?

After the death of Ashoka, the Maurya empire fell around 180 BCE, and the NW part of India was taken hold of by Graeco-Bactrians under Śāka administrators. 

4. Indo-Greek and Indo-Śāka rule

The Greek-born but Buddhist convert Milinda ruled between 160-130 BCE and had his capital at Sakala (Sialkot in Punjab). He campaigned as far as Mathura and Pataliputra against the Shungas (successors of Mauryas in Magadha). It is expected that there would be Śāka soldiers in the Indo-Greek army.

Post 100 BCE, a Śāka administrator of the Indo-Greeks named Maues became the king after marrying the widow - Greek Queen Machene. From the numismatic evidence, we see that Maues reversed the Hellenic ban of representing Hindu icons on the Indo-Greek coinage by the predecessors - presumably to appease his Hindu citizens and elites. The Indo-Śāka ruled with the help of Satraps and Maha-Satraps.

Rudradaman I, the Indo-Śāka king who ruled as a Maha-Kshatrapa from 150 CE waged various wars against the Andhra Satavahana dynasty but ended up giving his daughter's hand in marriage to the Brahmin Andhra king Vashishtiputra Satakarni (from Junagadh inscription).

The period saw the rise of the Kushana empire under Kanishka I. Kushanas had control over Northern India till around 370 CE when they were pushed out by the Gupta kings. Samudragupta's Prayagraj Pillar mentions that the last of the Kushanas offered their daughters' hand in marriage in an attempt to keep control over their territories.

5. Maga Brahmins

We have one confirmed instance of foreign priests (Iranian Mages in particular) being slowly accepted into the Hindu Brahmin fold. Maga Brahmins also known as Śākadvipiya Brahmins or Bhojaka Brahmins are thought to have instituted Solar deity worship (as vigraha/mUrti). They are mentioned in Buddhist sources (mostly negatively due to their foreign customs, especially relating to incestuous marriages) as well as some Purāṇas (Bronkhorst 2015). With the centuries of rule by the Achaemenids, Indo-Śāka and Graeco Bactrians; these foreign priests may have gained full Brahmin status in a society wherein Mlecchas (foreigners) were looked down upon.

Johannes Bronkhorst writes this in his 2015 book:

No longer do we need to find out why Indian Brahmins, contrary to their habits, were willing to accept in their midst immigrants from a neighbouring country in the northwest. The fact is that they did not do so, or if they did, not knowingly. They did not accept as Brahmins immigrants from a neighbouring country in the northwest, but immigrants from a remote continent, not reachable by ordinary travel. And these immigrants had always been Brahmins, for unlike the countries in the northwest, society in the remote continent of Śākadvīpa was organized according to Brahmanical principles.  But why should immigrants from Śākadvīpa be granted privileges that were not granted to visitors from neighbouring countries? The answer is simple and straightforward: Because quite independently of the arrival of the Magas, and presumably already before this event, there was a Brahmanical tradition that maintained that the remote continent of Śākadvīpa was inhabited by people who followed the Brahmanical order of society. This is clear from a passage in the Bhīṣmaparvan of the Mahābhārata, which says a great deal about Śākadvīpa and its inhabitants, but nothing whatsoever about migrating Magas.

Our reflections appear to justify the following picture. For some reason so far unknown there was a belief in Brahmanical circles according to which there was a remote continent called Śākadvīpa whose population consisted of Brahmins, Kṣatriyas, Vaiśyas and Śūdras. Independently of this tradition, and presumably at some later date, sun-priests from Persia settled in India. In order to be recognized as Brahmins in their new surroundings, they or their descendents made the claim that they were the Magas of Śākadvīpa, who had been called hither.

Bronkhorst believes that the Magi came to India during the Indo-Śāka or Kushana rule.

However, KC Srivastava (1968) believes that the Magi may have entered Indian society along with the Achaemenid invasion of 500 BCE.

It is not far-fetched to assume that a lot of the steppe ancestry in Brahmins came from intermarriage with the Maga Brahmins before caste endogamy set in.

Added on 7-Dec-2022

There is a curious passage in the Mahābhārata 6.12.33–37 which goes thus (translated by Bronkhorst 2015)

In that [continent] there are four meritorious countries, esteemed by the people: Maga, Maśaka, Mānasa and Mandaga. Maga is mainly inhabited by Brahmins who love their tasks. In Maśaka there are virtuous Kṣatriyas who are generous in accordance with the wishes of all. In Mānasa the Vaiśyas survive by their tasks; they are brave, devoted to the wishes of all, bent on dharma and artha. The Śūdras in Mandaga, for their part, are men constantly pious. There is neither king nor punishment, whether big or small. The [people] preserve dharma with regard to each other by [sticking to] their own dharma. This much can be said about that continent. This much you should hear about Śākadvīpa, full of splendor.

Later Greek sources mention a tribe called Massagétai, who probably lived around the Oxus region, somewhere to the east of the Caspian Sea (Schmitt 2018). Schmitt (2018) writes:

The name Massagétai mostly and most plausibly is explained as the plural form (containing the suffix East Ir. *-tā, reflected in Gk. -tai) of *Masi̯a-ka-, which can be understood as a regular derivation with Ir. *-ka- from *masi̯a- “fish”.

Masi̯a-ka- sounds like the same Maśaka from the Mahābhārata verse above.

Strabo also distinguishes four groups of Massagétai - living on the islands, the marshes, the mountains, and the plains (Schmitt 2018), similar to 'four meritorious countries' in MBh. Herodotus notes that they worship only the Sun God and sacrifice horses to him (Schmitt 2018)

Additional evidence comes from the Gadha(Jasdan) inscription of 200 CE during the Indo-Śāka rule of the Western Kshatrapas. 

 

Jasdan Inscription - Translation From Epigraphia Indica Vol 16 by FW Thomas

'Son of Pratā-Śāka, of the Mānasa gotra' is mentioned here. Mānasa is also the same country as mentioned in the MBh verse above and is associated with a Śāka (Pratā-Śāka). This is the region where the Vaishyas (trader varNa) of Śākadvīpa are supposed to reside.

The Massagétai are from the right location (Turkmenistan region), have the right name, speak the right language and worship the right God to be associated with the people that the Maga Brahmins came from. 

This is an informative article by Aryāṃśa on the Maga Brahmins.

False claims made by Narasimhan et al 2019 

This article is not complete without me reminding everyone how this paper has set the Indian population genetics research field back by a decade. Apart from getting very generic ancestry proportions of Indians right, it gets almost everything else wrong.

1. "People of the BMAC were not a major source of ancestry for South Asians" 

This can now be said to be untrue. Turkmenistan_IA is derived from BMAC and Steppe_MLBA populations, and the analysis above showed that it is the best steppe source in Indians. So either both are major sources of ancestry in South Asians, or none are.

2. "Steppe ancestry in modern South Asians is primarily from males and disproportionately high in Brahmin and Bhumihar groups." 

I have rebutted both of these claims in previous posts. Simply put, R1a-L657 (which makes up 70% of modern Indian R1a) is not from the steppe and is local to the Indian subcontinent. It is found in exactly 0 bronze age samples from the steppe and is absent in any meaningful way from modern steppe populations. The paper avoids this nuance and concludes wrongly after omitting this relevant data. The rebuttal can be read here.

Steppe ancestry is not disproportionately high in Brahmin groups. UP Rajput groups have similar ancestry as Brahmins, and the highest steppe groups in India are from NW India (Jats). This is covered here.

3. "By the end of the second millennium BCE, these people were joined by numerous outlier individuals with East Asian–related admixture that became ubiquitous in the region by the Iron Age. This East Asian–related admixture is also seen in later groups with known cultural impacts on South Asia, including Huns, Kushans, and Śāka, and is hardly present in the two primary ancestral populations of South Asia, suggesting that the Steppe ancestry widespread in South Asia derived from pre–Iron Age Central Asians."

The authors of the paper wanted to sell us the idea that steppe ancestry entered Indians before the late bronze age (roughly 1300 BCE) because of the presence of 'ubiquitous east Asian ancestry' in steppe regions post-LBA. This timing best fits their idea of the 'Aryan Invasion'.

This was already proven false by the data of their own paper. There was no east Asian ancestry in the 1100 BCE Kashkarchi samples from Uzbekistan. And there is no east Asian ancestry in the 850 BCE Turkmenistan_IA sample. What's more, Kangju (from 200 CE, which has minor east Asian ancestry) was a good fit as a source in their own models. East Asian ancestry is clearly not ubiquitous in the regions to the west and south of Kyrgyzstan in the iron age either (Kumar et al 2021; Guarino-Vignon et al, 2021).

From Narasimhan et al supplement pg 287 (emphasis mine)

We were unable to reject a single Iron Age population, Kazakhstan_Kangju.SG as a source, though their time period , ~200-300 CE, is much too late for them to be a viable source based on the fact that Late Bronze-Iron Age populations from South Asia from almost a millennia earlier already have substantial amounts of Steppe ancestry, and the fact that our admixture timing estimates on the modern individuals provide dates for the admixture of Steppe pastoralist-related and Iranian farmer-related ancestry in South Asia in the 2nd millennium BCE.

The presence of steppe ancestry in Swat is not proof that the ancestry spread further into India in the same burst. Swat iron age populations do not form a good source for Indians.

Their own dates of admixture in modern Indians do not show 2nd millennium BCE mixing. The error bars are wide and also support a post-1000BCE mixing date for most groups. (Graphed from Narasimhan et al 2019 supplement excel Table S5)

Error Bars represent a 95% confidence level around the Admixture date

4. In Shinde et al 2019, coauthored with Harvard geneticists, they conclude "First farmers of the Fertile Crescent contributed little to no ancestry to later South Asians"

This point specifically seems to have been emphasized to deny a western origin of the Indo-European languages in India, so that only a central Asia route via the steppe remains plausible. 

This is a false claim. Not only do the Indus Periphery samples need Anatolian Farmer or Levant PPN-related ancestry (Maier et al, 2022; my previous work), but so do the South Indian tribals like Irula and steppe-free samples such as RoopkundA_PallanLike.

CONCLUSION

All the available evidence tells us that the steppe ancestry in modern Indians is from an iron age South Central Asian source like Turkmenistan_IA. This also agrees with the archaeological and literary evidence. Since the probable source was already Iranian by language, it cannot explain the introduction of Indo-Aryan languages into the Indian subcontinent. Furthermore, a post-1000 BCE date for the entry of Indo-Aryans into North India is not supported by the Rig Veda and neither by common sense.

There is a lot of evidence to suggest contact between the Eastern Iranian Śāka and NW India, starting from 900 BCE till 350 CE. This contact seems to have been the most likely source of steppe ancestry in modern Indians.

SUPPLEMENTARY FOLDER

Rotating Model results & Sample Id list

REFERENCES

(2022). Maga Brahmins. vriitrahan. https://vriitrahan.wordpress.com/2021/07/28/maga-brahmins/

Agrawala, V. S. "India as Known to Pāṇini.[A Study of the Cultural Material in the Ashţādhyāyī] India as Known to Panini.[A Study of the Cultural Material in the Ashtadhyayi]." Journal of the American Oriental Society 84.4 (1964).

Albaladejo Vivero, Manuel. (2020). 'Aristobulus of Cassandreia: An Engineer at the Court of Alexander the Great', in E. Myrdal (ed.), South Asian Archaeology and Art 2014 Papers Presented at the Twenty-Second International Conference of the European Association for South Asian Archaeology and Art held at the Museum of Far Eastern Antiquities/National Museums of World Culture, Stockholm, Sweden, 30th of June to 4th of July 2014, New Delhi 2020. Pp: 99-114. ISBN: 978-93-87496-36-1.. 

Basafa, Hassan, and Mohhamadsadegh Davari. "Evaluation of components and distribution area of​​ Yaz-I-III culture in northeastern Iran with emphasis on Khorasan cultural area." Journal of Iran's Pre-Islamic Archaeological Essays 5.2 (2021): 45-62.

Bronkhorst, Johannes. “How the Brahmins Won,” 2016. doi:10.1163/9789004315518.

Chatterjee, Anirban, et al. "On the existence of a perennial river in the Harappan heartland." Scientific reports 9.1 (2019): 1-7.

FLEMING, DAVID. “Where Was Achaemenid India?” Bulletin of the Asia Institute, vol. 7, 1993, pp. 67–72. JSTOR, http://www.jstor.org/stable/24048427. Accessed 3 Dec. 2022.

Éadaoin Harney, Nick Patterson, David Reich, John Wakeley, Assessing the performance of qpAdm: a statistical tool for studying population admixture, Genetics, Volume 217, Issue 4, April 2021, iyaa045, https://doi.org/10.1093/genetics/iyaa045

Guarino-Vignon, P., Marchi, N., Bendezu-Sarmiento, J. et al. Genetic continuity of Indo-Iranian speakers since the Iron Age in southern Central Asia. Sci Rep 12, 733 (2022). https://doi.org/10.1038/s41598-021-04144-4

Hemphill, Brian E. "Biological adaptations and affinities of Bronze Age Harappans." Harappa Excavations 1986-1990: A Multidisciplinary Approach to Third Millennium Urbanism (Monographs in World Archaeology No. 3) (1991): 137-182.

https://en.wikipedia.org/wiki/Achaemenid_conquest_of_the_Indus_Valley. Accessed 3-Dec-2022

https://en.wikipedia.org/wiki/Indo-Scythians. Accessed 3-Dec-2022

https://en.wikipedia.org/wiki/Menander_I. Accessed 3 Dec 2022

https://en.wikipedia.org/wiki/Seleucid%E2%80%93Mauryan_war. Accessed 3 Dec 2022

Lhuillier, Johanna, et al. "The Middle Iron Age in Ulug-depe: A preliminary typo-chronological and technological study of the Yaz II ceramic complex." (2013).

Magee, Peter, et al. "The Achaemenid empire in South Asia and recent excavations in Akra in northwest Pakistan." American journal of archaeology 109.4 (2005): 711-741.

Malik N. Uncovering transitions in paleoclimate time series and the climate-driven demise of an ancient civilization. Chaos. 2020;30(8):083108. doi:10.1063/5.0012059

Metspalu M, Romero IG, Yunusbayev B, et al. Shared and unique components of human population structure and genome-wide signals of positive selection in South Asia [published correction appears in 

Am J Hum Genet. 2012 Feb 10;90(2):378-9]. Am J Hum Genet. 2011;89(6):731-744. doi:10.1016/j.ajhg.2011.11.010

Narasimhan VM, Patterson N, Moorjani P, et al. The formation of human populations in South and Central Asia. Science. 2019;365(6457):eaat7487. doi:10.1126/science.aat7487

Petrie, Cameron A. and Peter Magee. “The Achaemenid Expansion to the Indus and Alexander’s Invasion of North-West South Asia.” (2012).

Reich, David. Who we are and how we got here: Ancient DNA and the new science of the human past. Oxford University Press, 2018.

Robert Maier, Pavel Flegontov, Olga Flegontova, Piya Changmai, David Reich. "On the limits of fitting complex models of population history to genetic data" bioRxiv 2022.05.08.491072; doi: https://doi.org/10.1101/2022.05.08.491072

Rüdiger Schmitt, “MASSAGETAE,” Encyclopædia Iranica, online edition, 2018, available at http://www.iranicaonline.org/articles/massagetae (accessed on 7 Dec 2022).

Sengupta, Torsa, et al. "Did the Harappan settlement of Dholavira (India) collapse during the onset of Meghalayan stage drought?." Journal of Quaternary Science 35.3 (2020): 382-395.

Shinde, V., Narasimhan, V. M., Rohland, N., Mallick, S., Mah, M., Lipson, M., ... & Reich, D. (2019). An ancient Harappan genome lacks ancestry from steppe pastoralists or Iranian farmers. Cell, 179(3), 729-735.

Srivastava, V. C. “ANTIQUITY OF MAGAS IN ANCIENT INDIA.” Proceedings of the Indian History Congress, vol. 30, 1968, pp. 86–94. JSTOR, http://www.jstor.org/stable/44141456. Accessed 3 Dec. 2022.

Vikas Kumar, E Andrew Bennett, Dongyue Zhao, et al. Genetic Continuity of Bronze Age Ancestry with Increased Steppe-Related Ancestry in Late Iron Age Uzbekistan, Molecular Biology and Evolution, Volume 38, Issue 11, November 2021, Pages 4908–4917, https://doi.org/10.1093/molbev/msab216

Widemann, François. “Maues King of Taxila: An Indo-Greek Kingdom with a Saka King.” East and West, vol. 53, no. 1/4, 2003, pp. 95–125. JSTOR, http://www.jstor.org/stable/29757574. Accessed 3 Dec. 2022.