[Click on the image to enlarge.] For example, the cluster labeled "2a" is a mixture of New Guinea highland and coastal groups, along with a single representative of the Nasioi speakers of Bougainville, in the Solomon Islands. Since Papuan (i.e., non-Austronesian) speakers can be found in both the highlands and the coast, and since Nasioi is a Papuan language, this cluster could be seen, hypothetically, as representative of population group 2, as defined above, i.e., descendants of the original "Negrito" settlers. Since all but one of the groups represented in cluster "1b" are from the highlands, and thus in all likelihood also Papuan speakers, this cluster could also be assigned to group 2. Clusters "1c" and "1d," on the other hand, represent a mix of New Guinea highlanders and Australians, suggesting that the New Guineans in these samples might have originated in group 3, i.e., highland populations in New Guinea formerly centered along the coast, who are not "Negrito" descendants, but originated as Australoid immigrants. There are in addition several unhighlighted clusters (in block 1 only) representing purely Australian groups, which can unproblematically be assigned to population group 1, i.e., Australian Aborigines descended from the original Australoid immigrants. Finally, cluster "1a," a mix of Polynesian and New Guinean coastal populations, most likely belong to group 4, i.e., Austronesian speaking Neolithic farmers, living for the most part along the northern coast of New Guinea, but biologically and culturally most closely associated with Southeast Asia and Polynesia. The above picture, in addition to being highly speculative, is clouded by the authors' failure to identify the specific groups sampled (with the exception of the Nasioi), and also the tendency of Melanesian peoples generally to borrow cultural elements from one another, including not only tools and farming methods, but also rituals, musical instruments and even, on occasion, specific musical practices, at least to some extent. So clearly much more research will be necessary before the hypothesis in question can be adequately tested.
*As you may recall, this tree is a bit unusual, as it's based on the mtDNA coding region rather than the noncoding "neutral markers" of the "D-loop," which has usually been the focus of this sort of research in the past. For this reason, the authors chose to forego the usual haplogroup terminology (L, M, N, etc.), which could be misleading. Nevertheless, the clades grouped under the labels "1" and "2" are to be understood as essentially equivalent to mtDNA superhaplogroups "N" and "M" respectively.