Very roughly the story goes like this (don't use any dates or figures from here):
Let's look at 7000 years before present. What do we have?
- attested in Iran by aDNA "Iranian farmer".
- in India, unattested by aDNA, but reasonably safe theoretical assumption, the AASI profile.
What else in India? What would the ancestry of someone in Punjab 7000 years before present be? In the absence of anything other information we'd have to say, it is some combo of Iranian farmer + AASI.
Suppose we found in Punjab aDNA from 7000 years ago. Then whatever it was, it would not be explainable in terms of some contemporary Iranian farmer; at best it would trace to some common ancestor with the contemporary Iranian farmer aDNA, with the theoretical AASI thrown in. The finding in BMAC of 3 persons with AASI but no steppe ancestry suggests that this Punjab aDNA would also show no recent common ancestor with steppe.
The other peoples in India of 7000 years ago would presumably be better modeled via this Punjabi aDNA + AASI rather than Iranian farmer + AASI. But we don't have aDNA from Punjab from 7000 years ago, and so the only choice is to model people in India in terms of Iranian farmer (ostensibly an outsider) and AASI. And of course, in the absence of aDNA there is nothing really to model. A principle of parsimony says that the people of north India of this time had (Iranian farmer + AASI) ancestry. Nothing more can be said.
The same goes for Indians from 6000 years ago, 5000 years ago, 4000 years ago. 3000 years ago, we get the first aDNA in the Indian subcontinent, that from Swat. It can only be modeled in terms of aDNA found so far - Iranian farmer, Siberian and Central Asian Hunter Gatherer, Steppe MLBA, EMBA) and theoretical populations like AASI. If we had aDNA from Punjab from 7000 years ago, for example, the best model of Swat aDNA might be rather different; just like the discovery of the aDNA of three Indus_Periphery persons turned the theoretical constructs ANI and ASI from being quite ancient (from more than 14000 years ago) to arising only post-Saraswati-Sindhu civilization.
One revelation of this paper (assuming that it is correct) is that genetic models seem to be not stable against small perturbations. Three samples of aDNA turned the origins of ANI and ASI upside down.
Similarly on the Eurogenes blog there is a post from about an year ago, that Indian populations are better modeled with Steppe EMBA than Steppe MLBA. The recent pre-print finds Steppe MLBA to be more relevant. That is, with "Indus_Periphery" aDNA thrown into the mix, now Steppe MLBA provides a better fit for Indian populations than Steppe EMBA. This is counter-intuitive to me, and IMO, is another indication of the instability of genetic models.
It may be a fair inference that further discoveries could destabilize the model that this pre-print proposes. Unfortunately it seems likely that little if any aDNA in the main part of India will ever be found.
Let's look at 7000 years before present. What do we have?
- attested in Iran by aDNA "Iranian farmer".
- in India, unattested by aDNA, but reasonably safe theoretical assumption, the AASI profile.
What else in India? What would the ancestry of someone in Punjab 7000 years before present be? In the absence of anything other information we'd have to say, it is some combo of Iranian farmer + AASI.
Suppose we found in Punjab aDNA from 7000 years ago. Then whatever it was, it would not be explainable in terms of some contemporary Iranian farmer; at best it would trace to some common ancestor with the contemporary Iranian farmer aDNA, with the theoretical AASI thrown in. The finding in BMAC of 3 persons with AASI but no steppe ancestry suggests that this Punjab aDNA would also show no recent common ancestor with steppe.
The other peoples in India of 7000 years ago would presumably be better modeled via this Punjabi aDNA + AASI rather than Iranian farmer + AASI. But we don't have aDNA from Punjab from 7000 years ago, and so the only choice is to model people in India in terms of Iranian farmer (ostensibly an outsider) and AASI. And of course, in the absence of aDNA there is nothing really to model. A principle of parsimony says that the people of north India of this time had (Iranian farmer + AASI) ancestry. Nothing more can be said.
The same goes for Indians from 6000 years ago, 5000 years ago, 4000 years ago. 3000 years ago, we get the first aDNA in the Indian subcontinent, that from Swat. It can only be modeled in terms of aDNA found so far - Iranian farmer, Siberian and Central Asian Hunter Gatherer, Steppe MLBA, EMBA) and theoretical populations like AASI. If we had aDNA from Punjab from 7000 years ago, for example, the best model of Swat aDNA might be rather different; just like the discovery of the aDNA of three Indus_Periphery persons turned the theoretical constructs ANI and ASI from being quite ancient (from more than 14000 years ago) to arising only post-Saraswati-Sindhu civilization.
One revelation of this paper (assuming that it is correct) is that genetic models seem to be not stable against small perturbations. Three samples of aDNA turned the origins of ANI and ASI upside down.
Similarly on the Eurogenes blog there is a post from about an year ago, that Indian populations are better modeled with Steppe EMBA than Steppe MLBA. The recent pre-print finds Steppe MLBA to be more relevant. That is, with "Indus_Periphery" aDNA thrown into the mix, now Steppe MLBA provides a better fit for Indian populations than Steppe EMBA. This is counter-intuitive to me, and IMO, is another indication of the instability of genetic models.
It may be a fair inference that further discoveries could destabilize the model that this pre-print proposes. Unfortunately it seems likely that little if any aDNA in the main part of India will ever be found.
