Hunting the Denisovan belt

The other day I was having a long conversation about Denisovans and human origins. My friend suggested that "Denisovans" sound like some kind of Star Trek civilization. I have to admit, he had a point.

When I think about the present distribution of Denisovan ancestry in human populations, I'm thinking increasingly of Star Trek. No matter which remote planet the Enterprise orbits, Captain Kirk always seems to get along just fine with the alien women. If we visited this corner of the galaxy 50,000 years later, and discovered that human genes have been spread to far-off planets, we wouldn't be surprised. Kirk did it.

But if we visited 50,000 years from now and discovered that human genes are only on far-off planets, and never on any nearby planets, that would be pretty weird. Either Captain Kirk never met the aliens on nearby planets, or they didn't have women, or something else came along after the Enterprise and wiped out all the human genes.

This is sort of where we are with Denisovan ancestry in living human populations. David Reich and colleagues (2011) have showed that most island populations of Indonesia and Oceania east of Wallace's line have a measurable fraction of Denisovan ancestry, while they found no measurable fraction on the Asian mainland or anywhere that had been connected to the mainland during the last glaciation. Around the same time, Pontus Skoglund and Mattias Jakobsson (2011) generally agreed with these findings but additionally suggested that some people in South China have a substantial component of Denisovan ancestry. Neither study had the power to rule out very small levels (<1%) of ancestry, so this does not imply a total absence of all Denisovan introgression in other populations, just that the introgression is at most slight.

This present distribution of Denisovan genes is mysterious. Denisova Cave itself is in Russia, far from anywhere that demonstrably has Denisovan descendants today. Modern humans must have encountered Denisovans on their way to Wallacea, but where?

I've previously suggested the explanation of overprinting:

  1. One layer of modern human dispersal encountered Denisovans, mixed with them, then settled throughout Southeast Asia, Wallacea and Australia.

  2. Later, a second wave of population dispersal of modern humans brought new people into South Asia, Southeast Asia and China. These people were the majority ancestors of present groups like the Andaman Islanders and hunter-gatherers of Malaysia. They may have originated in West Asia, or have staged through Siberia and thence into China, or both.

  3. Later still, early agriculturalists dispersed from South China across much of Southeast Asia and Indonesia, probably in multiple events. There may have been additional layers of genetic dispersal before and after the initial dispersals of agriculture.

Today's genetic makeup of mainland Asia, Sumatra, Borneo, and Java comes mostly from the Holocene dispersals. This would be similar to Europe, where the Upper Paleolithic population had very little input into the agricultural Neolithic population. Today's relict hunter-gatherer populations come mostly from the second pre-agricultural dispersal. The first dispersal and mixture with Denisovans have been overprinted by these later events, which were certainly more complicated than three distinct stages that minimally happened.

This overprinting scenario leaves unresolved exactly where the Denisovans may have lived when modern humans came onto the scene. We know that some Denisovans lived in the Altai, but they were probably followed there by Neandertals, not modern humans. The rest of the Denisovans may have lived almost anywhere in South or Southeast Asia, or China. This would have been a "Denisovan belt". My favorite guess is India, which should have been home to the largest population of Middle Paleolithic people anywhere along the southern tier of Asia. But maybe India was inhabited by Neandertals, or by a mix of Neandertals and Denisovans, or by yet another unknown group.

I don't count it as unlikely that the Denisovans included the Shanidar people of northern Iraq, usually counted as Neandertals. We have no genetics yet from these skeletal remains, nor do we have genetic evidence from the "Neandertals" of the Levant. Could Amud and Kebara possibly be Denisovans?

The further west we place the Denisovans, the more problematic the overprinting scenario becomes. Deriving new modern humans from Africa, to push across the entire tier of southern Asia and largely supplant the previous early modern humans seems like a stretch.

Alan Cooper and Chris Stringer have an article in Science this week that makes an argument very different from mine. They propose that the mixture of modern humans and Denisovans did not happen on the mainland of Asia, or even in Sundaland. Instead, they propose that the Denisovans lived in Wallacea.

The apparent absence of Denisovan introgression in current mainland populations is most easily explained through overwriting by the DNA of incoming East Asian populations in areas other than Island Southeast Asia. However, analysis of indigenous negrito/hunter-gatherer populations on mainland Malaysia and the Andaman Islands revealed no Denisovan DNA introgression, even though the long-isolated Andaman Islanders show no admixture with other East Asian populations (3). Similarly, genomic analysis of an ancient modern human in China (Tianyuan, ?40,000 years old) detected no Denisovan DNA (12), arguing against the existence of a prehistoric interbreeding signal that has been overwritten.
Together, these observations argue against an ancient introgression of Denisovan DNA on the Asian mainland (3). Instead, the source of the Denisovan gene flow appears to have been east of Wallace's Line, with the lack of Denisovan DNA in mainland populations explained by Wallace's Line limiting the reverse dispersal of introgressed populations. Subsequent movements of East Asian/Neolithic modern humans appear to have diluted the Denisovan-introgressed populations outside Australia and New Guinea, and also carried the signal further throughout the area and across the Pacific (3).

Cooper and Stringer's strongest argument is that a wide range of aboriginal peoples of South and Southeast Asia lack the Denisovan ancestry signature, from Andaman Islanders to Semang and Semai. The overprinting hypothesis requires at least one pre-Holocene dispersal to have given rise to the present genetic makeup of these populations. The ancient DNA from Tianyuan (Fu et al. 2013) is somewhat weaker evidence, since this North Chinese specimen may have shared little with the ancestry of contemporaries in Southeast Asia.

Still, the island evidence itself shows the power of overprinting. Recent migration from China and Southeast Asia was enough to reduce the Denisovan ancestry fraction in many Wallacean populations down to 1 percent. This dilution, in at least its early stages, was accomplished by low-density non-agricultural peoples crossing in relatively small numbers by boat. Early modern humans arriving in South Asia and Southeast Asia before 60,000 years ago had to persist many times longer, in the face of spreading cultural and genetic adaptations carried by people who could walk in large numbers. The overprinting hypothesis predicts that 80% or more of the genetic variation across Southeast Asia may derive from later events than the initial layer of modern humans to have inhabited the region. That may not be unlikely.

What remains to throw a wrench in this problem is the possibility that Denisovan ancestry has been incorrectly estimated. The estimation of Denisovan ancestry for any individual requires an elaborate correction based on the assumption that Denisovan and Neandertal genomes can be cleanly separated; and that sub-Saharan Africans have neither. This procedure requires a population model that is probably wrong. Since Denisovan ancestry in most populations is very small, the error in population model may have a relatively large effect on that estimate.

Any clear demonstration of Denisovan ancestry on the Asian mainland will refute the Wallacea-only model of mixture. I suspect that is only a matter of time. The comparisons done by Skoglund and Jakobsson already suggest that the level of Denisovan ancestry in south China is nonzero. As many readers know, the Genographic Project is giving out "Denisovan ancestry" estimates that are nonzero for most people. Although these are probably overestimating the amount, I do not think we can assume that those estimates should be zero.

There may have been a Denisovan Enterprise spreading their genes into Wallacea, but I expect that wasn't the only way modern humans got their genes.


Cooper, A., Stringer, C. B., Oct. 2013. Did the denisovans cross wallace's line? Science 342 (6156), 321-323. URL

Fu, Q., Meyer, M., Gao, X., Stenzel, U., Burbano, H. A., Kelso, J., Pbo, S., Feb. 2013. DNA analysis of an early modern human from tianyuan cave, china. Proceedings of the National Academy of Sciences 110 (6), 2223-2227. URL

Reich D, Patterson N, Kircher M, Delfin F, Nandineni MR, Pugach I, Ko AM-S, Ko Y-C, Jinam TA, Phipps ME, et al. 2011. Denisova admixture and the first modern human dispersals into southeast Asia and oceania. American journal of human genetics. 89(4):516-28.

Skoglund P, Jakobsson M. 2011. Archaic human ancestry in East Asia. Proceedings of the National Academy of Sciences, U. S. A. 108(45):18301-18306.