From the paper:
Although we cannot specify how many different migrations have colonized Sahul since the first settlement approximately 50,000 years ago, our results indicate ancient splits into seven major plausible groups: TNG, South-Papuan, North-West Papuan, North-East Papuan, West-Papuan, PN, and non-PN. The wide-spread families (TNG and PN) on both sides of the Torres Strait divide (~9,000 BP) are the result of more recent expansions of two of those groups, in the case of TNG probably linked to the development of agriculture, ~9,000 to 6,000 years ago, see ,.
The AN expansion is much more recent and has only had effects in eastern Indonesia, along the north coast of New Guinea and the islands east of the New Guinea mainland. We know on the basis of the comparative method correlated with archaeological data that approximately 3,200 years ago the Oceanic subgroup dispersed from its homeland on New Britain in three directions : (1) back along the north coast, (2) around the eastern tip of New Guinea along the south coast, and (3) much further into the Pacific. The results of our analysis capture some of the impact of this great expansion on the languages that were already in the region. We find that in the eastern islands there are clearly distinct AN and non-AN groups, with good evidence of a deep structural phylogenetic signal, albeit with some admixture . In the western islands however there is considerably more typological convergence between AN and non-AN languages (see also ). The linguistic population identified as Red appears to have members along the north coast (Mairasi, I'saka, and Kamasau) and on New Britain, where again both AN (Mangseng) and Papuan languages (Kol and Sulka) have contributions from the same cluster. This finding suggests an area of millennia of contact between AN and Papuan non-TNG speaking groups.
Gene Expression has more.
PLoS Biology doi:10.1371/journal.pbio.1000241
Explaining the Linguistic Diversity of Sahul Using Population Models
Ger Reesink et al.
The region of the ancient Sahul continent (present day Australia and New Guinea, and surrounding islands) is home to extreme linguistic diversity. Even apart from the huge Austronesian language family, which spread into the area after the breakup of the Sahul continent in the Holocene, there are hundreds of languages from many apparently unrelated families. On each of the subcontinents, the generally accepted classification recognizes one large, widespread family and a number of unrelatable smaller families. If these language families are related to each other, it is at a depth which is inaccessible to standard linguistic methods. We have inferred the history of structural characteristics of these languages under an admixture model, using a Bayesian algorithm originally developed to discover populations on the basis of recombining genetic markers. This analysis identifies 10 ancestral language populations, some of which can be identified with clearly defined phylogenetic groups. The results also show traces of early dispersals, including hints at ancient connections between Australian languages and some Papuan groups (long hypothesized, never before demonstrated). Systematic language contact effects between members of big phylogenetic groups are also detected, which can in some cases be identified with a diffusional or substrate signal. Most interestingly, however, there remains striking evidence of a phylogenetic signal, with many languages showing negligible amounts of admixture.