In Science, Cooper and Stringer discuss the distribution of Denisovan ancestry. Before I say anything else – their map is great.
We don’t see any Denisovan ancestry in populations living in mainland Southeast Asia, or Indonesia, or indeed in any population west of Wallace’s Line [which marks the eastern border of placental mammals, while Lydekker's line marks the western border of the marsupial fauna: Wallacea lies between]. It could be that later population movements have diluted the Denisovan taint, but it is apparently undetectable even in existing populations that are thought to represent earlier strata, such as the Andaman islanders. Nor has it been seen in ancient DNA from modern humans on the Asian mainland.
Cooper and Stringer suggest that Denisovans may have settled Wallacea before modern humans arrived, and that they may have accounted for a higher ancestry fraction there because the number of modern human settlers was very small: rafts and canoes can’t carry many. Which would mean that Denisovans had some seagoing technology. That tech had to be in a fairly narrow range: good enough to reach Wallacea (and the Philippines), but not good enough to reach Sahul (Australia/New Guinea). I wouldn’t say that is impossible, but it’s a bit unlikely.
If Denisovans had arrived in Australia, they would have left a mark – extinct giant tortoises, if nothing else. Considering that the entire Australian fauna seems to have gone to school on the short bus, even Neanderthal-class hunters would have caused a mass extinction. And in fact, the Custodians, whose technology was Neanderthal-class until a few thousand years ago, managed to do just that. So the Denisovans never made it that far.
Let me suggest another model. Back in the day, Indonesia was a big peninsula, called Sundaland. Let us assume that there were Denisovans there. In some way, Denisovan effective population density was higher there than in mainland Asia, or perhaps they were harder to displace. For example, some of Sundaland seems to have been tropical rainforest, otherwise known as the Green Hell – maybe there were some potent tropical diseases (vector-borne) that the Denisovans had developed resistance to, while the invading anatomical modern humans had not [Cooper and Stringer were also thinking about jungle pathogens] – the same reason that Europeans or Middle Easterners didn’t replace sub-Saharan African populations.
There may be genetic hints of this: Melanesians have picked up a couple of very ancient alleles of innate immune system genes.
So modern humans expand into Sundaland more slowly, in something like a range expansion. The further they moved into Sundaland, the more Denisovan genes they picked up. Again, like Arabs moving into Africa. Only not as much, because Denisovans were more different from AMH than any two existing human populations, because of much, much longer separation.
So the humans living at the eastern edge of Sundaland, particularly in Borneo, have more Denisovan ancestry in this scenario than anyone else in Eurasia – and they would be the people who take the next step, crossing the narrow seas to the Philippines, Wallacea, and then Sahul (Australia/New Guinea), virgin lands in which the hand of man had never set foot [with the possible exception of Flores]. A quite small, unusually Denisovanized coastal population of hunter-gatherers then undergoes a vast expansion, becoming as numerous as the stars in the sky.