Cladistic methods were used to infer the Mesozoic and Cenozoic history of the major southern hemisphere land areas. The study included reduced area cladograms for ten putative "Gondwanan" groups, including the angiosperm taxa Winteraceae and Proteaceae: Macadamieae. Brooks parsimony analysis produced a single most parsimonious tree that resolves the area relationships (southeastern Asia (India (Africa, Madagascar) (South America, Australia))). A proposed Tertiary land connection between Africa and Madagascar may help to explain their inferred sister relationship. Coding of Madagascar as a composite area also yielded a single tree differing only by the addition of Madagascar 2 as sister to India. Hovenkamp's (1997) vicariance analysis failed to resolve a general area cladogram because of conflicting area relationships and occurrence of sympatric (redundant) or widespread taxa in the source cladograms. The area relationships inferred using Brooks parsimony are more-or-less congruent with a model of Gondwanaland fragmentation derived from independent geological evidence. Brooks parsimony under a topological constraint corresponding to a hierarchical earth history model yielded a tree five steps longer than in the unconstrained analysis. Similarly constrained analyses of multiple randomizations of the source data failed to produce any trees as short as the "real" one, which evidently lies on the tail of a null distribution of tree lengths. The biogeographic pattern in the phylogenetic data examined is thus consistent with a vicariance hypothesis.

Key words: Gondwanaland, historical biogeography, Proteaceae, vicariance biogeography, Winteraceae