I was thinking again about that Denisovan allele of EPAS1 that plays a big role in Tibetan altitude adaptation. Considering modern humans, it has only been found in Tibetans (high frequency) and in the Chinese (couple of percent). The preferred model in the paper is that it entered the common ancestors of Tibetans and Han, rising to high frequency among the Tibetans because of its advantage. I doubt this: the authors are clinging to a claim of a recent split in a previous publication of theirs – but the idea that the modern Tibetans are a fusion of a Han-like population with a long-established group of Tibetan hunter-gatherers seems more likely to me. So the few copies of the high-altitude EPAS1 allele among the Chinese are probably a result of recent gene flow, possibly from the Tibetan empire (618-841) that controlled parts of China, or from ethnic Tibetans identifying as Chinese.
This allele has some pretty powerful effects on the hypoxia response, which is there for a reason. The usual evolutionary rule is that change is bad: even though the Denisovan allele confers a big advantage at high altitude, the odds are that it is disadvantageous at low altitude. This would explain why it is rare in China and apparently unknown in Japan. This would also explain why it never made it to the Andes - even though there might have been a copy or two in the long-ago East Asian ancestors of the Amerindians, who have a bit of Denisovan admixture admixture (at least, I think they do – interesting if that isn’t the case) , it would most likely have been lost in Beringia. Along the same lines, altitude adaptations probably never managed to travel from Ethiopia to Tibet, which is why they have different approaches to altitude adaptation today.
It is therefore no surprise that this EPAS1 allele does not exist in Melanesians, even though they have 25 times as much Denisovan ancestry as mainland East Asians.
At low altitude, it was likely disadvantageous in Denisovans as well. Probably it was only found in Denisovans who lived at fairly high altitude, or that had recently migrated from a high-altitude area. Which suggests that Denisovans in Denisova Cave had recent ancestors living at high altitude.
The introgression was most likely to be successful in an area where it conferred advantage, which suggest that admixture probably occurred in or near the Tibetan plateau. When you think about it, anatomically modern humans would have more trouble than average displacing archaics from areas for which they had special adaptations not possessed by AMH – so Denisovans may have lingered longer in Tibet and neighboring high-altitude areas, affording greater opportunities for gene flow. This was especially the case when the local environmental challenge could not be solved by ancient technology – you can deal with cold through better clothes, but people weren’t quite up to oxygen masks 40,000 years ago.
As I have said before, Denisovans might have persisted longer in Sundaland because of resistance to local infectious disease – same principle.