Abstract
The development of the minimal spanning tree method of panbiogeographic analysis is outlined and its validity and utility discussed. A rationale for geographic parsimony is suggested and problems with the application of spanning trees, including incongruent track orientation, missing cladistic data, and the significance of minimality, are identified. The concepts of graph theory are used to develop criteria for recognition of nodes and to formulate statistical tests of biogeographic patterns. The algorithms and tests described are illustrated with examples from the biogeography of New Zealand Trichoptera.