Mathematical models of most recent common ancestry hide relevant historical details. Actual human phylogenetics is considerably more sparse. For instance (ten seconds of googling):

http://austhrutime.com/genetic_evidence.htm

"They conclude that the ancestors of this Aboriginal man, and possibly all Aboriginals, are of similar distance from Africans as are other Eurasians, and that at about 62,000-75,000 BP the Aboriginal ancestors split from the gene pool that gave rise to all other populations of modern humans. The authors say their study supports the model of human evolution according to which the modern Australian Aboriginal people descended from an early wave of expansion into Asia about 62,000-75,000 BP. Their data also supports the substantial population admixing and replacement of populations of the first wave by the 2nd expansion wave, predicted by this model, though a few populations are descendants of the early dispersal, such as those in Australia and the highlands of Papua New Guinea and Aeta. According to the authors this is compatible with data from mtDNA that indicate they derived from the same few founder haplogroups shared by all populations outside Africa, though all haplogroups observed in Australia are unique to this area. The data also support the the suggestion that modern Aboriginal Australians have descended from the first humans that entered Australia at least about 50,000 BP."

Which is to say, it is easy to identify pairs of population groups which have had essentially zero genetic interaction with each other for tens if not hundreds of thousands of years, despite the existence of genetic clines that link their intermediary populations over shorter timescales.