My map of R1a in West, Southwest, Central and South Asia

[Passa]

Here it is.

[Passa]

Artmar made me notice that the Gulf area is not well defined. I fixed the map.

[Passa]

I'm also disappointed because the uploaded image's quality is much worse than on my (and on yours I guess) PC.

[parasar]

Here it is.

For the SE corner of India - specifically the state of Tamil Nadu - for a sample set of 1680 covering 31 populations the overall %age of R1a was found to be 12.74%.
http://journals.plos.org/plosone/art...e.0050269.t002

[Megalophias]

When people make these kind of maps do they weight by population of the subgroups sampled?

In that study of Tamil Nadu 570/1680 (34%) of samples were hill tribes. According to the 2011 census 1.1% of the population was Scheduled Tribes. Likewise 154/1680 (9.2%) were Scheduled Castes in the study, but in 2011 were 20% of the population. Another source says that 13% of the population were upper caste, 7% Brahmins, so the last is about right (135/1680 = 8%). I have no idea how to weight the other categories.

If we weight the results by these figures and put the rest of the population at the average rate of the middle/upper caste groups, then the result is 16% R1a. In this case it is not far off. If you were looking at some rarer haplogroup, like F*, the study average would be 16%, but the weighted amount only 5%.

If you take the figures for East India (drawn from Jharkand, Orissa, and West Bengal) from Trivedi, 50% are Austro-Asiatic-speaking tribal people, but Scheduled Tribes make up only 26% of the population of Jharkand, 22% of Orissa, and 6% of West Bengal. So if you took the raw average from Trivedi you would colour East India 33% O2, when the true figure should be considerably less than half of that.

[parasar]

When people make these kind of maps do they weight by population of the subgroups sampled?

In that study of Tamil Nadu 570/1680 (34%) of samples were hill tribes. According to the 2011 census 1.1% of the population was Scheduled Tribes. Likewise 154/1680 (9.2%) were Scheduled Castes in the study, but in 2011 were 20% of the population. Another source says that 13% of the population were upper caste, 7% Brahmins, so the last is about right (135/1680 = 8%). I have no idea how to weight the other categories.

If we weight the results by these figures and put the rest of the population at the average rate of the middle/upper caste groups, then the result is 16% R1a. In this case it is not far off. If you were looking at some rarer haplogroup, like F*, the study average would be 16%, but the weighted amount only 5%.

If you take the figures for East India (drawn from Jharkand, Orissa, and West Bengal) from Trivedi, 50% are Austro-Asiatic-speaking tribal people, but Scheduled Tribes make up only 26% of the population of Jharkand, 22% of Orissa, and 6% of West Bengal. So if you took the raw average from Trivedi you would colour East India 33% O2, when the true figure should be considerably less than half of that.

Unfortunately weighting is not done.
Thankfully for a relatively balanced large data-set like Arunkumar's TN one it comes up about right.

For eastern India if you add in Bihar, it gets even more skewed. Bihar populations are almost all R (R1a, R2) and H, with some C, J, G, and L, but almost no O.

[newtoboard]

I think including the Tarim would be helpful too.

[Humanist]

Here it is.

What are the sources for your frequencies?

[parasar]

Artmar made me notice that the Gulf area is not well defined. I fixed the map.

I would also add that the Sindh region has consistently come up with ~50% R1a1.
Sengupta et al., http://www.ncbi.nlm.nih.gov/pmc/arti...230/table/TB3/
Qamar et al., HG3 http://www.ncbi.nlm.nih.gov/pmc/arti...589/table/TB2/
Firasat: "The highest frequency of R1a1* was observed in the Mohanna (71.4%) [of Sindh] and lowest in the Parsi (7.8%). Other populations with appreciable (>50%) frequency of R1a1* included the Kashmiri (58.3%), Punjabi caste (56.7%), and Sindhi (51.4%)." http://prr.hec.gov.pk/Thesis/1026S.pdf

In fact up and down the Indo-Gangetic plains the levels are in the neighborhood of 50%.

[MonkeyDLuffy]

What is the likely origin of R1a1a in South asia? I am R1a1a and I know lots of Punjabis on 23&me with same Ydna, looks like it is really common among us.