ONE SIGN OF HUMAN ARRIVAL on Pacific Islands is the presence and ecological impact of commensal plants and animals that are not capable of independent dispersal (Flenley 1989; Kirch and Ellison 1994; Kirch et al. 1991; McGlone et al. 1994). Recent research on the Pacific Rat, Rattus exulans, suggests that DNA-based phylogenies of extant populations can provide a model for prehistoric human mobility in the Pacific region (Matisoo-Smith 1996; Matisoo-Smith et al. 1998; Roberts 1991). Rattus exulans was transported intentionally by ancestral Polynesians, who valued it as a food source, and its remains are found in early archaeological layers throughout Polynesia. This rat cannot swim more than a few meters in open ocean (Spenneman and Rapp 1989), and its behavior and habitat preferences suggest it was an unlikely stowaway (Matisoo-Smith 1994; Williams 1973). Like other rodent species, R. exulans has a rapid generation turnover and, with little competition for resources, viable populations establish quickly and grow rapidly (Holdaway 1999). Although European contact and exploration in the Pacific resulted in the introduction of two more rat species (R. rattus and R. norvegicus), hybridization among these three species does not occur.
Though a R. exulans-based model of monitoring prehistoric Polynesian voyaging has been questioned on grounds of intentionality of transport and the possibility of dispersal and evolution of the rat without human intervention (Anderson 1996; Langdon 1995), testing the model depends not on a priori assumptions but on evaluation of the patterns of variation uncovered by genetic analyses. However, synchronic patterns of variation are always open to multiple interpretations as to the historical processes that produced them, as has been demonstrated in the human mtDNA out-of-Africa debate (Brown 1980; Cann et al. 1987; Vigilant et al. 1991). (For an in-depth discussion of mtDNA, see Cann  and for ancient DNA see Hagelberg  and Richards and Sykes ). One way to constrain the number of possible interpretations is to introduce a diachronic genetic perspective through the analysis of ancient DNA (Krings et al. 1997).
The study described in this article focuses on an analysis of genetic variation of 15 archaeological and 8 modern samples of R. exulans from the Chatham Islands. These samples are compared with 7 modern samples from Raoul Island, in the Kermadecs, and 50 modern samples from New Zealand. By combining the genetic perspectives afforded by ancient and modern samples, a clearer understanding of the population history of Chatham R. exulans can be obtained, which in turn can be interpreted in light of other information available on the prehistory of that island group. In addition, a more thorough understanding of the Chatham situation is useful in interpreting a contrasting genetic pattern observed in the Kermadec Islands and New Zealand.
The Chatham Islands lie approximately 950 km due east of the South Island of New Zealand ([Figure 1 ILLUSTRATION OMITTED] Fig. 1). The group consists of two main islands, Chatham Island and Pitt Island, and several smaller uninhabited islands. They lie in a westerly zone of subtropical and sub-Antarctic convergence and are regularly exposed to frontal low-pressure systems, making the islands a difficult and dangerous location to travel to and from (Levison et al. 1973). The date of Polynesian discovery of the Chathams is unknown, but estimates vary from A.D. 800-1000 (Sutton 1985) to around A.D. 1450 (Anderson 1994; Irwin 1992; McFadgen 1994; McGlone et al. 1994). Archaeological and linguistic evidence supports a New Zealand origin for Chatham Island colonists (Clark 1994; Sutton 1985).
[Figure 1 ILLUSTRATION OMITTED]
The Kermadec Islands lie approximately 1000 km northeast of New Zealand, nearly halfway between New Zealand and Tonga. The four islands were uninhabited when Europeans first arrived in 1788 (Johnson 1995), but there is clear evidence in the form of archaeological sites, and the presence of R. …