Scratching my head! - Criticism of Narsimhan VM et al 2019 - Pt 1

In this post, I will start getting my thoughts about the seminal SC Asian paper organized in written form. I have had these stored in my head for 2 years. The data in the paper is absolutely fantastic, and we get the first aDna from South Asia ie Swat valley. But there are so many conclusions and data analysis choices made by the authors that left me scratching my head, so much so that this post got delayed because I just have so much to say but don't know where to begin.

Abdullah Al Ameriki on Flickr, CC BY 2.0, via Wikimedia Commons

So without getting ahead of myself, I will only address this claim made in the paper1 (bolded emphasis is mine). I will deal with the rest of the head scratches later in some other posts.

Groups that traditionally view themselves as being of priestly status in this and the preceding panel are shown in red (“Brahmin,” “Pandit,” and “Bhumihar” but excluding “Catholic Brahmins”), and tend to have a significantly higher ratio of Central_Steppe_MLBA to Indus_Periphery_Cline ancestry than other groups.

More generally, there is a notable enrichment in groups that consider themselves to be of traditionally priestly status: 5 of the 6 groups with Z less than −4.5 were Brahmins or Bhumihars even though they comprise only 7–11% of the 140 groups analyzed (p less than 10−12 by a χ2 test assuming all the groups evolved independently).

Nevertheless, the fact that traditional custodians of liturgy in Sanskrit (Brahmins) tend to have more Steppe ancestry than is predicted by a simple ASI-ANI mixture model provides an independent line of evidence, beyond the distinctive ancestry profile shared between South Asia and Bronze Eastern Europe mirroring the shared features of Balto-Slavic and Indo-Iranian languages (59), for a Steppe origin for South Asia’s Indo-European languages prior to ~2000 BCE.

The claim is quite simple and appealing - Steppe ancestry has a special relation to brahmin groups, but not other groups and therefore another line of evidence that steppe ancestry IS CAUSAL TO THE ETHNOGENESIS OF BRAHMIN CASTE AND THEREFORE VAIDIK RELIGION.

One has to look deep at the supplementary information in the paper to find out flaws in this data. I'll call them 'NOT SO MAJOR FLAWS' and will discuss them a bit later. But before that, I want to point out a basic problem with this whole approach. Which is

CORRELATION DOES NOT IMPLY CAUSATION, AND THE RESEARCHER'S CONFIRMATION BIAS NEEDS TO BE CHECKED

This is one of the MAJOR FLAWS of the authors' claim I am discussing. The steppe ancestry of modern Brahmins cannot tell us that the ancient steppe ancestors (males as claimed in the paper) married the ancestors of these soon-to-be Brahmins and Bhumihars of UP, Bihar, and Nepal, and because of this admixture the Brahmin caste was born. These steppe males also brought the proto Sanskrit language apparently. That is just a conjecture at this point.

I'll posit another scenario as to why this Steppe/IndusPeriphery ratio is high in these UP Bihar area modern brahmins. It is this

The high social status of Brahmins over the centuries allowed them plenty of choice in marriage, and they were able to marry women with higher West Eurasian and steppe ancestry. 

My reasoning is backed up by this 2016 paper2 about Uttarakhand castes.

When comparing with the previous analyses, our data showed significant (unpaired t-test P<0.0001) higher proportion of West Eurasian-specific lineages among traditionally higher caste (Brahmin and Kshatriya) populations for maternal lineages, although it was non-significant for the paternal ancestry (unpaired t-test P=0.5468).

The paper is basically saying that west Eurasian (Iranian, SC Asian, and steppe) ancestry in modern Uttarakhand brahmins and Kshatriyas is mediated through women. This is exactly the conclusion made by Narasimhan from the study of the Swat valley data as well.

THE NOT SO MAJOR FLAWS

The authors completely fail to mention the Brahmin groups which do not have an extreme z score for the Steppe/Indus Periphery Ratio. Looking into the supplementary file of the paper which has Tables 1-5 in excel format, there are several brahmin groups with non-significant z scores. What about them?

Filtered brahmin data list from table S5 from the supplement. AHG = Onge Andamanese

As you can see, none of the steppe/Indus Periphery ratio of the Brahmin shaded populations is significant (<-3). Why is this fact not mentioned? the data is taken straight from the supplement, none of the analysis in the above table is my own. I have only filtered the list and shaded the populations.

THE BIGGEST FLAW

What does Steppe/IndusPeriphery ratio even mean?? It would have meant something important if the model for modern Brahmins was just Steppe + Indus Periphery. But as you can see from the above table, it's Onge + Steppe + Indus Periphery. And there is no pure Onge-like population to be found in India around 2nd mill BCE. So this 3-way model is not a proximal model. Why assume that locals in Punjab and locals in Bihar would have the same amount of Onge component as each other before steppe ancestry came in? Surely there was more Onge component in the east. Then why is it considered separately? Just to get a good steppe/Indus periphery ratio for Brahmins?

Let me try to explain with an example.

Suppose a Khatri man lived in Punjab, and a Brahmin man lived in Bihar around 1000bce.

The ancestry of Khatri man is 100% Indus Periphery and 0% Onge.

The Bihari Brahmin being eastern has more Onge - So 75% Indus periphery and 25% Onge related.

Now suppose that 2 steppe women migrate and each marries one of the 2 respectively. The couples have 1 child each. What will be the % of the three components in both children? The table below shows the ancestries of the children. 

See how we can create a high Steppe/IndusPeriphery ratio in eastern Indian Brahmins.

Voila!! We created a high Steppe/Indus periphery ratio in one of the two even though they had the same external steppe input. Hence proved, this Steppe/IndusPeriphery ratio means nothing for a 3 source model. The paper has many such head-scratchers, I shall continue at another time.

1 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

2 Negi, N., Tamang, R., Pande, V. et al. The paternal ancestry of Uttarakhand does not imitate the classical caste system of India. J Hum Genet 61, 167–172 (2016). https://doi.org/10.1038/jhg.2015.121

Also, read

Brahmins from NW India have ~15% frequency of European mtDNA - New paper

Vote closed:)

Where did R-L657 Originate

India/South Asia

Central Asia

Russian Steppe

Iran

Middle East

Im here for the show with popcorn

 

pollcode.com free polls