After the retreat of Pleistocene glaciers, distributions of many high latitude organisms shifted resulting in new species assemblages and opportunities for genetic and ecological interactions [5–8, 18]. Historical genetic signatures of heterospecific mitochondrial genes may be preserved in hybrid zone populations that are no longer undergoing genetic exchange. Examination of these populations using population and coalescent methods should provide insight into the timing and dynamics of geographical shifts in species' ranges in response to climate change [8, 19].
Concordance of diagnostic characters
Identification of M. rutilus and M. gapperi in this transect based on the degree of closure of the post-palatal bridge was concordant 98.7% of the time with the distribution of the species-specific MYH6 alleles, with the exception of four individuals. The discordance observed in these individuals was a result of the character state of the post-palatal bridge being opposite of what was found in all other members of their respective population. We therefore reason that because there is a distinct change in the developmental state of the post-palatal bridge at the Stikine River area and this character distinguishes M. rutilus from M. gapperi. This abrupt change in frequency of the ossification of post-palatal bridge and distribution of the MYH6 diagnostic alleles coincides with Hall's  depiction of parapatry of M. rutilus and M. gapperi at the Stikine River area.
The distribution of highly differentiated mitochondrial cytochrome b haplotypes was not concordant with the morphological and nuclear characters. Our analysis revealed three groups: M. rutilus is characterized by an incomplete post-palatal bridge and a set of closely related MYH6 alleles and cytochrome b haplotypes; M. gapperi is characterized by a complete post-palatal bridge and a set of closely related MYH6 alleles and cytochrome b haplotypes that are highly differentiated from those of M. rutilus (1.9% and 7.2%, respectively); and an introgressant form that has the post palatal bridge and MYH6 alleles of M. gapperi but a set of cytochrome b haplotypes that is unique, yet clearly more closely related to M. rutilus haplotypes. Thus, across an 80 km expanse separating pure M. rutilus and M. gapperi populations, these introgressant red-backed voles are characterized by a combination of features of both M. rutilus and M. gapperi.
Colonization and hybridization dynamics
Estimates of expansion times into southeast Alaska obtained from the mismatch distribution (t = τ/2u) indicate that these species arrived post-glacially. Working under the assumption that rates of evolution are consistent among these three groups, estimates of time since expansion are similar in M. gapperi and the introgressants, dating to 9.46 Ka and 8.57 Ka, respectively. As introgressants largely reflect the genetic signature of the hybridizing M. rutilus, expansion of both species into southeast Alaska date back to the early Holocene [21, 22]. These estimates of expansion into southeast Alaska are notably earlier than the expansion time of those pure populations of M. rutilus that now exist north of the LeConte Glacier (4.31 Ka).
Given our refined view of the distribution of M. rutilus and M. gapperi and potential barriers, there appears to be no contemporary contact between these species in this transect. Consistent with that view, is the lack of MYH6 heterozygotes with alleles diagnostic of pure M. rutilus and M. gapperi that might suggest ongoing hybridization with M. rutilus . Likewise, inspection of the post-palatal bridge, presumably controlled by multiple nuclear loci, revealed no intermediate morphs.
In addition to not finding evidence for contemporary gene exchange between these two species, we did not find the parental species in sympatry. Furthermore, there was not extensive overlap of the introgressants with the pure parentals. M. gapperi and the introgressants occur in sympatry only at the southern edges of the zone of introgressants (localities 11, 12, 15), and the current tidewater position of the LeConte Glacier prevents contemporary contact of pure M. rutilus with voles in areas farther south that are now occupied by introgressants or, even farther south, by pure M. gapperi. As such, it would appear that the introgressants are self-sustaining populations and not hybrids that are continuously generated from pure parental crosses. An active contact zone does exist, however, at the southern, leading edge of the introgressant distribution and the northern edge of pure M. gapperi (at localities 11, 12, 15). The direction and degree of genetic exchange at the latter contact zone is the subject of an ongoing study (Runck et al., in prep). Future ecological studies should explore whether abiotic or biotic shifts are associated with the transition between these groups and/or whether direct competitive interactions limit their coexistence.
Origin of mitochondrial signature in introgressants
The introgressant form is characterized by a monophyletic group of mitochondrial haplotypes nested within haplotypes otherwise characteristic of M. rutilus. Nonetheless, the introgressed group possesses a distinct mitochondrial genetic signature from M. rutilus populations across this region. Notably, not only do the introgressants not share any haplotypes with M. rutilus, they also have significantly greater mitochondrial variability than the donor species.
Three plausible scenarios could lead to this distinct genetic signature of novel, but closely related, haplotypes and overall higher mtDNA diversity that characterize the introgressants. First, increased mitochondrial diversity in the introgressants may simply reflect differences in population history when compared to M. rutilus in southeast Alaska. The higher estimates of mitochondrial diversity in the introgressants are similar to those for other populations of M. rutilus outside of southeast Alaska. For 54 M. rutilus found across a much greater geographic sampling area in northwestern Canada and interior Alaska, estimates are comparable to the level of diversity seen in the introgressants, with nucleotide diversity of 0.003, 19 segregating sites, and τ = 1.63 (unpublished data). Therefore, it is possible that contemporary populations of M. rutilus in southeast Alaska previously had higher levels of diversity, but lost diversity (e.g., through bottleneck events) after hybridizing with M. gapperi. This hypothesis does not explain the lack of shared haplotypes between the introgressants and M. rutilus, however, so southeast Alaskan M. rutilus subsequently must have lost all the haplotypes now found exclusively in the introgressants.
A possible alternative is that southeast Alaska may have served as a glacial refugium during the Last Glacial Maximum for red-backed voles. Under this scenario, a more diverse population of M. rutilus originally hybridized with M. gapperi thus creating the introgressants, which captured and sustained M. rutilus haplotype diversity. As glaciers retreated, genetic diversity was lost from coastal populations of pure M. rutilus as they expanded northward (reflected in low mitochondrial variation in contemporary populations of M. rutilus). Although southeast Alaska has been proposed as a coastal refugium for vertebrates during Pleistocene glacial advances [24, 25], we do not believe this is the case for northern red-backed voles again, because no introgressant haplotypes are shared with contemporary populations of M. rutilus.
A hypothesis that seems most consistent with the available data implicates multiple waves of colonization of M. rutilus into this coastal region. Multiple colonization events of M. rutilus into the region could also lead to this pattern of distinct haplotypes in the introgressants. Under this scenario, gene exchange between the species occurred first with an early colonizing population of M. rutilus, and the genetic signature of the introgressants now reflects this ancestral exchange along with the subsequent accumulation of new haplotypes through time. Later, a second colonization event would have given rise to extant M. rutilus in the region. If these extant-pure M. rutilus are indeed the result from a second, more recent colonization event, as the estimate of expansion suggests (~4.3 Ka), these voles would have been isolated from introgressant populations south of Jap Creek (locality 7) due to the advancement of the LeConte Glacier to tidewater around 5,000 ybp. Therefore, genetic exchange between the introgressants and contemporary M. rutilus would have been prevented and would account for the lack of shared cytochrome b haplotypes and lack of MYH6 heterozygotes. One would have to posit that the earlier colonizing wave of M. rutilus was extirpated from the region or was not sampled in this study.
Regional extinctions and multiple colonizations have been documented in southeast Alaska during times of climate oscillations  as the North Pacific Coast underwent repeated climatic fluctuations during the Pleistocene and Holocene. Recent advances correspond to the Younger Dryas around 10.6-9.9 Ka , with three additional advances around 5–6 Ka, 3.5-2.5 Ka, and 200-100 YBP [28–30]. During these cooling events, alpine glaciers advanced into lower elevations, reducing species ranges. Due to the extensive glacial coverage of the northern part of southeast Alaska, M. rutilus may have been susceptible to extirpation or displacement during periods of cooling and advancing glaciers.
Marginal support for the multiple waves of colonization of M. rutilus hypothesis is reflected in the relationship of the introgressant mtDNA relative to pure M. rutilus in the minimum spanning network and likelihood tree. In both analyses, introgressants form a subgroup, and are not intermixed with the remaining M. rutilus mitotypes suggesting that these two mitotypes were not part of a panmictic population. The genetic footprint of an earlier hybridization event supports the hypothesis of multiple colonizations of the northern red-backed vole along the coast.
Evolution of an introgressant contact zone
Although genetic exchange may have been extensive, our data suggest that a very closely related set of haplotypes (or a single haplotype that subsequently mutated) was captured and maintained in the introgressants. The unimodal distribution of pairwise comparisons and position of the introgressant haplotypes in the network and phylogeny support the hypothesis of a single hybridization event instead of multiple temporally discrete events. Initial genetic exchange may also have been bidirectional, but we only have evidence thus far of the mitochondrion of M. rutilus being maintained on the morphological and presumably nuclear background of M. gapperi. Nonetheless, a more complete view of the nuclear composition of the introgressant form relative to the pure parental forms will provide insight to the extent of backcrossing that occurred in this system.
Once established, the novel mito-nuclear combination of the introgressants either diffused neutrally or was selected for and expanded their distribution while displacing the pure parentals. While we cannot determine which scenario is responsible for the 80 km zone of introgressants, it is notable that all individuals in this region are introgressants. Moreover, several of our findings are consistent with predictions of the bounded superiority hypothesis . The introgressants and pure parentals do not overlap extensively and the introgressants are self-sustaining and are not the result of continual hybridization. Also, consistent with the bounded superiority hypothesis is that the introgressants occupy a limited area of southeast Alaska, and have not established populations west of localities 11 and 12 on the Cleveland Peninsula.
The degree of genetic admixture that may have occurred while the introgressant voles were fairly uncommon relative to pure parentals awaits our more complete sampling of the nuclear genome of this group. Similar long-term persistence and spatial expansion of introgressant or hybrid forms has been documented in snails (genus Cerion), resulting from an ancient hybridization event between a now extinct fossil species and an extant species . Hybrids are hypothesized to have persisted due to the novel genetic combinations that enhanced survival during the time that one of the parental species was eliminated .
With the advance of the LeConte Glacier to tidewater approximately 5,000 years ago, any potential for gene flow between northern M. rutilus and the introgressants ceased. However at the southern edge, we do find localities (11, 12, and 15) where introgressants overlap spatially with pure M. gapperi suggesting the potential for ongoing gene flow. Ongoing analyses of microsatellite genetic variation should identify not only the degree of overall distinction between the introgressed form and pure M. gapperi but also the amount of gene flow that characterizes their current contact zone. Likewise, ecological studies may help identify the degree to which these groups compete either directly or indirectly with one another and whether the spatial distribution of each is currently stable or actively shifting.
A North Pacific Coast suture zone
The North Pacific Coast has been documented as post-glacial contact zone for several mammalian lineages from the high latitude refugium called Beringia and from multiple refugia that existed south of the continental ice sheets . Within species, independent colonizations into this recently deglaciated region have occurred by at least two divergent lineages of several species such as dusky shrew (Sorex monticolus) , long-tailed vole (Microtus longicaudus) , black bear (Ursus americanus) , and marten (Martes americana) [35, 36]. Ermine (Mustela erminea) are represented by three divergent lineages in southeast Alaska; one is hypothesized to be endemic to the region, perhaps surviving in the North Pacific Coast during the Pleistocene .
Introgression along the coast has been documented from divergent lineages of marten , black bears , and now red-backed voles. Contact zones are likely for divergent clades of dusky shrews and long-tailed voles, therefore the narrow strip of mainland along the southeast Alaska coast may be a suture zone, whereby several formerly isolated species have entered the region by discrete colonization routes, and have subsequently come into contact in the same geographic area [38–40]. The North Pacific Coast, and in particular, southeast Alaska, possess characteristics  commonly tied to suture zones, such as being located between Pleistocene glacial refugia, and nearby low mountain passes acting as corridors for dispersal [13, 41] during warm periods.
A renewed interest has emerged in testing the validity of Remington's thirteen North American suture zones through phylogeographic studies [39, 40, 42]. The North Pacific Coast was not originally identified as a suture zone by Remington , but phylogeographic studies repeatedly demonstrate the existence of multiple lineages within species (e.g., shrews, voles) in this region. Populations representing divergent lineages are now in contact there following postglacial expansion . Most of these studies, however, have limited ability to detect hybridization because only mitochondrial genes were assessed. Future phylogeographic studies of these species should employ multiple independent characters to more rigorously assess the influence of geologic and climatic events on structuring diversity along the North Pacific Coast.