INTRODUCTION
Species persistence through the Pleistocene in multiple, allopatric ice-free glacial refugia followed by population expansion and contact has left a legacy of shared centers of endemism and geographically clustered hybrid zones (Hultén 1937; Heusser 1989; Heaton & Grady 2003; Swenson & Howard 2005). During glacial advances in the Northern Hemisphere in the late Pleistocene, terrestrial species were primarily isolated south of the major ice sheets or in a high-latitude, Beringian refugium (Hultén 1937). Multiple additional, smaller refugia are hypothesized within southeast Alaska’s contemporary Alexander Archipelago (e.g., Baranof, Chichagof, Dall, Heceta, and Prince of Wales islands; Foster 1965; Fedorov & Stenseth 2002; Carrara et al.2003, 2007; Ager 2019), British Columbia’s Haida Gwaii Archipelago, and surrounding areas (Heusser 1989; Mathewes & Clague 2017). Coastal refugia may have also played an integral role in human colonization of the Americas by opening a maritime migration corridor. New evidence shows the Pacific Northwest was inhabited by humans as early as > 15-16 thousand years ago (kya; Devièse et al.2018; Davis et al. 2019) and the Alexander Archipelago at least > 10 kya, but likely > 13 kya (Dixon et al. 2014; Carlson & Baichtal 2015; Lesnek et al. 2018; Mackieet al. 2018; McLaren et al. 2018), potentially predating the opening of an ice-free migration corridor through central Alberta, Canada (<=14.8 kya; Margold et al. 2019). An incomplete fossil record and cosmogenic 10Be exposure dating, however, have created uncertainty over the existence of hypothesized coastal refugia. Unlike humans, with access to rudimentary sea-faring technologies (Erlandson et al. 2007), other coastal mammals that persisted in refugia would have been isolated from mainland populations, leading to a cessation in gene flow and divergence over time (Hewitt 2000). As such, the genomes of refugial descendants can provide clues to the history of refugial isolation. Predicted variation includes high genetic differentiation from other refugial populations, endemic or ancestral alleles, overall lower genetic diversity as a consequence of small population sizes and genetic drift (Hewitt 2000).
Early reconstructions of coastal refugia paleoenvironments identify a transition from wet sedge tundra to relatively treeless, lowland freshwater environments following the LGM (Barrie et al. 1993; Mann & Hamilton 1995; Hansen & Engstrom 1996; Ager 2007). As such, most forest-associated taxa are hypothesized to have persisted in southern continental refugia and subsequently colonized northward following the most recent glacial recession (Mann & Hamilton 1995; Conroy & Cook 2000; Stone et al. 2002; Hope et al. 2016). However, recent evidence (Ager 2019) documenting the rapid re-establishment of forest communities along the western-most edge of the North Pacific suggests their potential persistence in small refugia (Ager 2019), which also enabled the persistence of forest-associated communities. Around 10.5 kya exposed areas of continental shelf were rapidly submerged, following the collapse of the isostatic forebulge (Barrie et al. 1993). This event separated insular taxa from continental populations through the formation of marine barriers and led to extended allopatric divergence.
In contrast to glacial isolation, post-glacial expansion from multiple refugia during warm interglacial periods can lead to secondary contact and the formation of hybrid zones between previously isolated taxa (Hewitt 2000; Swenson & Howard 2005; Latch et al. 2009; Slageret al. 2020). The rapid climatic oscillations of the Pleistocene (Williams 1998) led to recurrent opportunities for contact and gene flow between incompletely diverged taxa (Hewitt 2000, 2003; Shafer et al . 2010; Acevedo et al. 2015; Slager et al. 2020). The geography and timing of secondary contact can inform the sequence and duration of paleogeographic events. The consequences of genetic exchange are complex and range from homogenization to hybrid speciation (Arnold 1997; Harrison & Harrison 1993; Genovart 2009; Abbott et al.2016), depending on the level of differentiation, but they can now be examined in detail using genome level analyses (Twyford & Ennos 2012). Further, genomic reevaluation of forest-associated taxa along the NPC is critical to unraveling the location, timing and duration, and composition of coastal paleo-communities.
North American martens are relatively small meso-carnivores thought to be a single species until recently (Dawson et al. 2014; Colellaet al. 2018a,b). Now two species are hypothesized to have diverged in at least two independent glacial refugia south of North American ice sheets (Stone et al. 2002): one refugium east of the Rocky Mountains or Mississippi River drainage giving rise to American pine martens (Martes americana , Turton 1806) and another to the west, presumably the cradle for Pacific martens (Martes caurina , Merriam 1890). The widespread coast-to-coast boreal distribution (Fig. 1) of M. americana directly contrasts with the fragmented distribution of M. caurina, found along the Pacific coast (CA, OR, WA), mountaintops of the American Southwest (NM, CO, UT), northward into Montana and Idaho, and on four islands within the putative refugial archipelagos of the NPC: Graham, Moresby, Kuiu, and Admiralty islands. However, the disjunct contemporary range of Pacific martens and detection of two natural hybrid zones between these species\(\ \) one occurring on near-coastal islands (Kuiu and Kupreanof, AK) along the NPC and another in the northern Rocky Mountains (Fig. 1)\(\ \) suggest Pacific martens may have a deep and complex evolutionary history along the NPC.
In addition to natural hybrid zones, a series of intentional wildlife translocations in the mid-1900s introduced M. americana to multiple NPC islands without prior knowledge of the native marten species in the region (Powell et al. 2012). While these introductions may complicate the interpretation of genomic signals from this region, they also provide a framework for interpreting the consequence of natural versus anthropogenically-mediated gene flow on the evolution and persistence of species. Low genetic variation in certain insular populations of M. caurina (Stone et al.2002; Small et al. 2003) has led to further concerns over their persistence, which may be exacerbated by ongoing harvest of forests and extraction of minerals on those islands (Durbin 1999; USDA 2018).
We use whole-genome resequencing data from NPC martens to test the Coastal Refugium Hypothesis (CRH; Heusser 1989; Scudder & Gessler 1989; Demboski et al. 1999; Sawyer et al. 2018) and ultimately refine our understanding of the dynamic biogeographic history and potential paleoenvironments of this coastal corridor. With increased molecular resolution, we also explore the evolutionary consequences of introgression in New World Martes to inform natural resource management.