My model of the caste system basically envisages a phenomenon derived from purity-pollution rules of West Asian priestly caste (Check Origins of Indians: Version 6.x). With this, I think I can move away from the skin colour based caste system model proposed by the self-loathing Europeans and try to understand whether skin colour without any technological advantage had any positive selection in India in the remote past.
My first assumption: Europe, North Africa, West Asia and South Asia were isolated regions until the appearance of light skin colour. This assumption isn't from thin air and strongly derives from non-overlapping male lineages of these regions.
Origin and spread of skin colour:
The genetic studies on skin colour are still in progress. I think the present knowledge is that skin colour arose in European population (unclear which part of Europe) around 10000-20000 years back and swept across North Africa and West Asia. Another assumption here is that I consider Europe, North Africa and West Asia are 100% light skinned regions.
Second Assumption: Europe is the region where light skin colour gene mutations first appeared.
Note: The light skin of East Asia and SE Asia is due to different genetic convergence for that trait. The population movement to South Asia is mostly from West and Central Asia, therefore, I won't consider East and SE Asia populations.
Skin colour gene and Haplogroup association:
The present topic requires the clear association of genes responsible for lighter skin (SLC24A5, SLC45A2) with mtDNA or Y-haplogroups. Since there are no clear studies on this, I can make educated guesses based on predominant haplogroups in Europe, West Asia and northern Africa.
Region -- Y-Haplogroup -- mtDNA
Europe -- R1b, R1a, I -- H, U
North Africa -- E1b1b, J -- H, U
West Asia -- J, E1b1b -- H, U
Since the colour genes are present on autosomes, the mutations' first appearance on Y-haplogroup or mtDNA isn't important. But if the first occurrence is in Europe, it's most likely in any of the R1b or R1a or I or mtDNA H or mtDNA U haplogroups.
Third Assumption: Y-haplogrups R1a, R1b, I, mtDNA Haplogrups H and U are the root lineages for the spread of light skin.
Spread to North Africa and West Asia:
If you consider haplogroup frequencies of R1b and R1a in North Africa and West Asia, it's most likely around 10-15% comparable to darker Dravidian regions but probably lesser than equally darker East Indian regions.
In this situation spread of lighter skin to these regions could only be explained by female specific sexual selection. The early development of agrarian society probably helped these regions to get lighter skinned females from European lands without much changing male genetic composition.
The situation is different in India. Here similar male Y-haplogroup frequencies haven't resulted in wider spread of lighter skin for South and East Indian population. In north-west India probably lighter skinned female presence in the incoming population (during IVC and later during IE migrations and invasions) has raised the lighter skin population. It should be noted here that north-west Indian share of root light skin haplogroup (R1a) is higher (around 30-40%). Almost twice that of North Africa or West Asia. But north-west India isn't as light skinned as the other two regions. That plausibly mean male centered sexual selection didn't happen.
Therefore, light skin of north-west India is a phenomenon similar to West Asia and North Africa. Female centered sexual selection for light skin.
From the look of it, lack of root light skin females and lack of male centered sexual selection for light skin appear to be the reason for darker Dravidian regions. But even natural selection could be one of the reasons. If you see some of the tribes like Todas(J2 and R1a) who are lighter than most of the caste population, geographical regions might have made the difference too. The strongest evidence comes from Soejima et al. (2007) study.
Here is an interesting explanation given in Soejima et al.(2006) about the spread of allele SLC45A2 in South Asians.
There are several scenarios to explain the differences in allele frequency of the coding SNPs between the SLC24A5 and SLC45A2 genes in South Asian populations. One of them is that 374L has a selective advantage against harmful ultraviolet radiation (UVB), and the frequency of the 374F allele migrating from Central Asian populations might have been reduced by purifying selection in South Asian populations. This may mean that the functional constraint is stronger on SLC45A2 than on SLC24A5. Sequence analysis of the SLC24A5 gene flanking the p.A111T polymorphism and functional analysis of the coding SNPs will help us understand the evolution of these
two pigmentation genes.
But noticeable feature is that SLC45A2 is almost absent from South Indian population and it occurs at 0.1 to 0.25 in North Indian population (According to Razib at Gene Expression).
As such from where the present studies stand I make the following conclusions:
1. Light skin didn't give any advantage to European males in North Africa and West Asia. On the contrary, during Neolithic time advanced dark skinned population of these regions took up light skinned European females and lightened the whole region.
2. South Indian dark skin is misleading from the point of view of spread of light skinned males. Here, rather than artificial sexual selection(that may not have taken place due to lack of light skinned females), natural selection had a higher say.
1. Skin Colour posts at Razib's Gene Expression.