Resolution of portions of the kangaroo phylogeny (Marsupialia: Macropodidae) using DNA hybridization

We generated a DNA hybridization matrix comparing eleven 'true' kangaroos (Macropodinae) and two outgroup marsupials, the rufous rat-kangaroo Aepyprymnus rufescens (Potoroinae) and the brush-tailed phalanger Trichosurus vulpecula (Phalangeridae). A small matrix included additional species of the genus Macropus (large kangaroos and wallabies). The results indicate that the New Guinean forest wallaby Dorcopsulus vanheurni, and the quokka Setonix brachyurus, represent successively closer sister-groups of other macropodines. The remaining taxa examined form two clades: the tree kangaroo Dendrolagus matschiei with die pademelons Thylogale and rock wallabies Petrogale, and Macropus including the swamp wallaby Wallabia bicolor. The smaller matrix of five Macropus species and Wallabia (with Dorcopsulus as an outgroup) pairs the red-necked wallaby M. rufogriseus and Parry's wallaby M. parryi, with the eastern grey kangaroo M. giganteus as their nearest relative; and associates the red kangaroo M. rufus and wallaroo M. robustus, with Wallabia as their sister-taxon. In the larger study, we found mat inclusion of both outgroups provided little resolution among the macropodines, judging by jackknife and bootstrap tests. When Aepyprymnus was deleted, the Dendrolagus-Thylogale-Petrogale association obtained; with Trichosurus eliminated instead, the Wallabia-Macropus group was recovered. Only analysis of the eleven ingroup taxa by themselves gave a topology which supported both major clades. Our findings suggest that, at least for DNA hybridization studies, when ingroup taxa are separated by very short internodes experimental error in outgroup-to-ingroup distances may seriously compromise determination of ingroup affinities as well as the position of the root. We recommend that in such cases separate analyses with the outgroups sequentially eliminated and rigorous validation of die topology at each step should be conducted.