Nothofagus (‘southern beech’) is a major component in forests throughout the South Island of New Zealand. However all four species are disjunct across the central portion of the island. Hypotheses for the disjunction include the following explanations based on vicariance: (i) 'glacial refugia' hypothesis, where distributions have not recovered following elimination during the Pleistocene glacial period; (ii) 'environmental barriers' hypothesis, where edaphic, climatic and migration barriers hinder Nothofagus migration; and (iii) 'lateral plate shift' hypothesis, where populations have been rafted apart by lateral movement on the Australia-Pacific tectonic plate boundary since the early Miocene. Dispersal explanations imply nonsurvival south of the disjunction during glacial periods followed by seed dispersal during the Holocene. One dispersal hypothesis explains the presence of N fusca and N truncata in the south through long-distance pollen dispersal and hybridisation resulting in species 'reconstitution'. A phylogeographic approach using sequences and RFLPs of non-coding chloroplast DNA (cpDNA) regions in the three New Zealand Fuscospora species is used to infer historic patterns of distribution and gene flow in the South Island. Maternally inherited cpDNA is transmitted via seed, and is therefore appropriate for revealing variation at a geographically localised scale. Comparison of N fusca/N truncata and N solandri cpDNA haplotypes south of the disjunction is used to assess the validity of the long-distance pollen dispersal and hybridisation hypothesis. Evolutionary rates of cpDNA in Fuscospora was found to be very slow, in line with other studies. A single insertion in the trnL-trnF intergenic spacer was restricted to south of the disjunction. A very approximate estimate of its age was made, and it is unlikely to be more than 1-2 mya. It is unlikely to have occurred during the Holocene, indicating that Fuscospora populations have existed in isolation south of the disjunction throughout the last Pleistocene glacial period at least, and possibly much longer. Therefore a vicariance origin to the disjunction is favoured. Although no resolution can be made between the 'glacial refugia' and 'environmental barriers' hypotheses, the 'lateral plate shift' hypothesis is not supported. The southern South Island mutation is shared between all three species, suggesting that introgression of the chloroplast genome has occurred, possibly in the absence of nuclear introgression. Assuming that N solandri var. cliffortioides survived in the south during the last glacial period, this finding is compatible with scenarios of (i) N fusca/N truncata survival in the south; (ii) Holocene dispersal of N fusca/N truncata seed, and (iii) 'reconstitution' of N fusca/N truncata via long-distance pollen dispersal and hybridisation.