COASTAL REFUGIA AND POSTGLACIAL RECOLONIZATION ROUTES: A REPLY TO DEMBOSKI, STONE, AND COOK.

Abstract

For the past fifty years, there has been considerable speculation that a coastal refugium existed in the Pacific Northwest during the last glaciation (Wisconsin glaciation). Initially these speculations, which were based on surveys of both faunal (McCabe and Cowan 1945; Foster 1965) and floral (Calder and Taylor 1968; Ogilvie and Roemer 1984; Ogilvie 1989; Schofield 1989) endemism and disjunct plant distributions (Schofield 1965, 1984; Calder and Taylor 1968; Ogilvie and Roemer 1984; Ogilvie 1989, 1997), pointed to Haida Gwaii (previously known as the Queen Charlotte Islands) as the site of this refugium. However, geological data directly contradicted such speculations because abundant eractics, cirques, and striations on Haida Gwaii indicate that this archipelago experienced substantial ice cover during the late Wisconsin (Sutherland Brown and Nasmith 1962). Investigations of the endemic populations of stickleback (Gasterosteus) on the islands strongly indicate local adaptation to site-specific habitats (Reimchen 1994) as well as strong suggestion for the postglacial origin of the divergent populations (Moodie and Reimchen 1976). In response to this controversy, we initiated a broad molecular survey of some of Haida Gwaii's putative glacial relicts to determine whether mtDNA phylogeography could shed some light on this long-standing problem. In 1997, we reported the existence of two black bear (Ursus americanus) lineages, which we designated as coastal and continental. The coastal lineage was largely restricted to the coast (Haida Gwaii, Vancouver Island, coastal mainland of British Columbia, and the Olympic Peninsula), whereas the continental lineage was widely distributed throughout North America. From this phylogeographic pattern we suggested that (1) this pattern in western North America was the result of postglacial recolonization from two source areas: a refugium south of the ice front and coastal refugia on the continental shelf possibly adjacent to Haida Gwaii, which is consistent with previous predictions; and (2) the morphological attributes that differentiate black bear subspecies from coastal British Columbia (carlottae, kermodei, altifrontalis, and vancouveri) arose from a common ancestor sometime after the Wisconsin glaciation. Demboski et al. have argued that the phylogeographic pattern that we uncovered in the North American black bear is an artifact of incomplete sampling. As such, they argue that our suggestion that this pattern may have been influenced by dispersal from a coastal refugium is premature. Their argument is based on their finding that the coastal haplotypes of marten (Martes americana) and vagrant shrew (Sorex vagrans) also occur in continental regions. They also cite similar findings within the brown bear (U. arctos), where a divergent haplotype originally identified on the Admiralty, Baranof, and Chicagof Islands in Alaska (Talbot and Sheilds 1996) has been found on the mainland. We also noted (Byun et al. 1997; Byun 1999) that the coastal black bear haplotype may also be present in California (Wooding and Ward 1997) and Montana (Cronin et al. 1991; Paetkau and Strobeck 1996). It is possible that the mtDNA phylogeographic pattern identified in black bears that comprises a coastal and a continental lineage has nothing to do with coastal refugia. Limited sampling is a typical problem in phylogeographic studies and increasing sampling area always provides a more detailed picture of distribution. However, our suggestion of a coastal refugium was based both on molecular evidence as well as information outside of the field of molecular phylogeography. We will briefly outline this evidence. Point 1.-There are currently 18 disjunct hepatics and 12 disjunct mosses found in the Western Hemisphere only on Haida Gwaii, the Pacific Coast of British Columbia, and adjacent Alaska. In addition to this, seven disjunct bryophytes are found only in North America on Haida Gwaii. Strong affinity of many of these disjunct bryophytes with bryophytes found in western Europe or southeastern Asia is suggestive that they may be relicts of ancient flora, possibly dating back to the Tertiary. Persistence of these suspected relicts on Haida Gwaii was suggested by Schofield (1984) to be evidence that suitable habitat continued to exist during multiple Pleistocene glacial advances. Although the current distribution of these bryophytes might be the result of postglacial dispersal from mainland refugia, it was considered unlikely because of the lack of readily dispersible diaspores and absence of significant asexual reproduction, which limits these species to local populations. In addition to bryophytes, at least nine vascular plants are known to be disjunctly distributed. Ogilvie and Roemer (1984) and Calder and Taylor (1968) proposed that Haida Gwaii was likely to have been a late Wisconsin refugium for these plant species. Point 2.-While local ice sheets on Haida Gwaii began to recede about 16,000 years ago, surrounding areas were experiencing a glacial maximum. Pollen profiles taken from the