• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Phylogeography of Masked (Sorex Cinereus) and Smoky Shrews (Sorex Fumeus) in the Southern Appalachians

Posted on: Saturday, 13 November 2004, 03:00 CST

Sorex cinereus (masked shrew) and Sorex fumeus (smoky shrew) are syntopic species co-occurring in relict fragments of spruce-fir habitat on southern Appalachian mountaintops. We conducted phylogenetic and population genetic analyses of 20 high-elevation Sorex populations in 8 distinct boreal islands. Partial mitochondrial DNA sequences (cytochrome b and D-loop) were compared with amplified fragment length polymorphism markers obtained by restriction of whole genomic DNA. The 2 species, though similar in morphology and ecological niche, have dissimilar phylogeographic patterns. S. cinereus, despite its more limited present-day southern Appalachian range, exhibits markedly less population structure than S. fumeus. What structure is present among masked shrew populations is randomized geographically, in contrast to a distinct association between genetic partitioning and geographic location among smoky shrew populations. Disparity in post-Pleistocene population densities of these species might be implicated in the discrepant patterns of phylogeographic structuring evident in their genomes as a result of historical vicariance. Although the metapopulations of both species exhibit genetic signatures consistent with continuous historical expansion, we believe that a localized catastrophic event induced a severe genetic bottleneck in Sorex populations at Whitetop Mountain, Virginia. This study contributes to a better understanding of the repercussions of boreal habitat fragmentation on the population dynamics and genetic diversity of associated vertebrate species.

Key words: amplified fragment length polymorphism, control region, cytochrome b, D-loop, mitochondrial DNA, population structure, shrews

Small, isolated populations of plants and animals frequently are generated by the vicariant fragmentation of extensive tracts of habitat into patches. Such drastic reduction in population size can have a significant genetic effect through random lineage sorting, inbreeding, loss of heterozygosity, and increased susceptibility to genetic drift (Usher 1995). Affected populations typically experience a net loss of genetic variability over time in the absence of gene flow. This study focuses on the spruce-fir (Picea rubens and Abies fraseri) habitat of the southern Appalachian mountain range and the effect of its historical fragmentation on the population genetics of associated soricid fauna.

Appalachian red spruce-Fraser fir habitat is confined to a series of relictual mountaintop populations, primarily as the result of a post-Pleistocene vicariance event. These isolated habitat patches can be envisioned as an archipelago (MacArthur and Wilson 1967), the southernmost extension of which is to 3530'N latitude in western North Carolina. Total area of North American spruce-fir habitat is estimated to be 100,000 km^sup 2^; southern Appalachian patches comprise a mere 0.003% (266 km^sup 2^) of this total (Dull et al. 1988). Some expanses of Appalachian boreal habitat are connected by spruce-fir corridors along ridgelines; other patches resemble "sky islands" punctuated by deep river valleys and surrounded by wide "oceans" of alternative vegetation.

Much of the region east of the Mississippi River and north of 41 latitude was covered by the Laurentide ice sheet until approximately 18,000 years ago. Although glaciers did not reach the southeastern United States during the Quaternary, the climate of this region was affected by changes in global atmospheric circulation patterns that induced glacial advances and retreats (Nalepa et al. 2002). Landform and pollen records show that the southern Appalachians experienced harsh periglacial conditions during the Wisconsinian glacial interval (Delcourt et al. 1993; Braun 1989). Mixed-composition forests containing spruce and fir thrived at low elevations in a wide swath south of the glacial line (Delcourt and Delcourt 1985). During the postglacial era, red spruce and Fraser fir gradually invaded the coolest and most humid mountain regions. By 8,000 years ago, the southern spruce-fir forest had fragmented and was completely isolated from its contiguous northern expanse (White and Cogbill 1992).

Sorex cinereus and Sorex fumeux represent the dominant soricid fauna of these spruce-fir fragments (Steele and Powell 1999). A long- term study of soricid communities at 104 localities in the Blue Ridge and Upper Piedmont of the southern highlands (Laerm et al. 1999) provided insight into the Appalachian distribution of Sorex. S. cinereus has the narrower habitat range in this area and is common only in high elevation mesic habitats such as spruce-fir. Trapping density of this species had a positive correlation with elevation (r = 0.55); S. cincreiis was trapped only above 610 m, with 70% of captures above 915 m (Laerm et al. 1999). S. fumeus, a habitat generalist, has a more widespread Appalachian range and a greater abundance in high-elevation deciduous forests than in spruce- fir. This species exhibited a weaker positive correlation between density and elevation (r = 0.33), and specimens were taken at sites ranging from 244 to 1,525 m in elevation (Laerm et al. 1999).

Previous studies addressing the phylogeography of southern Appalachian vertebrates (Browne et al. 1999; Crespi et al. 2003; Ferree 1998; Reese et al. 2001 ) have indicated that the degree of genetic fragmentation experienced by a species is related directly to how closely it is linked, ecologically and climatically, to the boreal spruce-fir habitat. We hypothesized that the reliance of Sorex species on such a habitat should have resulted in significant substructuring of their nuclear and mitochondrial genomes. Furthermore, the more pronounced preference of S. cincreus for high- elevation habitat (compared with S. fumeus) should have resulted in greater isolation and subsequent genetic differentiation of S. cinereus populations. Because the ranges of these species have minimum elevational limits, depressed landscape features such as the French Broad River should present barriers to dispersal and gene flow. Due to the mountainous nature of the region, there should be an increasing number of barriers to dispersal as geographic distance increases. Reduction of gene flow between populations should thus be a function of the reduction in vagility of individuals, such that distant populations have the greatest genetic dissimilarity, although some populations likely are linked genetically through retention of ancestral lineages or ongoing migratory exchange along spruce-fir corridors.

MATERIALS AND METHODS

Sampling.-Twenty high-elevation (>910 m) trapping sites were designated throughout the southern Appalachians (Fig. 1). Many of these sites were linked by potential dispersal corridors along ridgelines; sites spanned 8 isolated patches of spruce-fir habitat of varying size (Table 1). Sampling took place during the months of May-October 2000 and 2001. Shrews were collected in pitfall traps with plastic buckets (14 mm deep 11.5 mm in diameter) partially filled with water. Traps were placed several meters apart in transects along natural runways, usually abulting logs or rocks, with fallen branches placed to create makeshift drift fences. Liver, heart, and/or blood tissue was obtained from 44 masked shrews and 47 smoky shrews. Total DNA extraction was performed on each specimen with the Puregene genomic DNA isolation kit (Gentra Systems, Minneapolis, Minnesota).

Mitochondrial DNA analysis.-A 726-base pair (bp) fragment at the 5' end of the cytochrome-b gene (Cytb) was amplified by the polymerase chain reaction (PCR) with the use of general mammalian primers L-14724 and H-15450 (Irwin et al. 1991). PCR products were purified via the QiaQuick PCR purification kit microcentrifuge protocol (Qiagen, Valencia, California). Approximately 500 bp of the fragment were sequenced with the forward primer using an ABI Prism 377 automated sequencer (Perkin-Elmer, Boston, Massachusetts). Sequences were aligned with Bioedit 5.0.9 (Hall 1999) and Clustal X 1.81 (Thompson et al. 1997) software. Cytb sequences were submitted to GenBank as accession numbers AY127953-AY127999 (S. fumeus) and AY128044-128087 (S. cinereus).

Amplification of a 500-bp fragment at the 5' end of the control region (D-loop) was achieved with a Sorex-specific L-primer located in the tRNA^sup Pro^ gene (Stewart and Baker 1994) and an internal H- primer located in a conserved sequence block (Southern et al. 1988; Wilkinson and Chapman 1991). The 79-bp last tandem repeat region, previously found to contribute significant information to phylogenetic analyses of shrew species (Stewart and Baker 1994, 1997), was sequenced with the forward primer and aligned as described above. Last tandem repeat sequences were submitted to GenBank as accession numbers AY127901-127947 (S. fumeus) and AY128000-AY128043 (S. cincreus). Sequences of PCR primers are presented in Table 2.

To evaluate patterns of mitochondrial DNA molecular diversity, Arlequin 2.000 (Schneider et al. 2000) was used to generate statistics, including nucleotide frequencies, number of polymorphic sites, and number of haplotypes for each permutation of gene region and species. Mea\n numbers and total variance of pairwise base pair differences between haplotypes (π) were calculated as in Tajima (1993). Nucleotide diversity (h), the probability that 2 randomly chosen homologous nucleotides were different, and its sampling variance were estimated by Nei's (1987) index. Gene diversity (H), the probability that 2 randomly chosen haplotypes were different, and its sampling variance were estimated (Nei 1987).

To select the models of DNA substitution that best fit the data, hierarchical likelihood ratio tests of 56 nested models of nucleotide substitution were run with ModelTest 3.06 (Posada and Crandall 1998) software. Phylogenetic relationships were reconstructed for individuals of each species under the model of evolution with the highest likelihood (HKY + I + G for combined mitochondrial DNA data in both species) by applying the neighbor- joining algorithm to Tajima-Nei genetic distance estimates (Tajima and Nei 1984) with PAUP* 4.0b10 (Swofford 2002) software. The Tajima- Nei model was selected to assist in comparing the amplified fragment length polymorphism data (below). Phylogenetic analyses were conducted separately for each gene region and for a combined 579-bp data set. Bootstrap values were obtained through 10,000 replicates under the highest likelihood model of evolution.

We employed Φ^sub ST^, an analog of Wright's (1921) fixation index (F^sub ST^), to quantify the inbreeding effect of population substructure. Pairwise Φ^sub ST^ values were calculated in Arlequin for all combinations of populations sampled. The null distribution of pairwise values under a hypothesis of no difference between populations was obtained by permuting haplotypes between populations. P values were determined as the proportion of permutations leading to Φ^sub ST^ values larger than or equal to the value observed.

Analyses of molecular variance permitted determination of the extent of geographic subdivision of mitochondrial haplotypes (Excoffier et al. 1992). We used Arlequin to generate Tajima-Nei distance estimates between all possible pairs of haplotypes. Population subdivision was analyzed for masked and smoky shrews between all populations; between populations north and south of the French Broad River, a presumptive migratory barrier; and between the 8 disjunct spruce-fir islands. The significance of the different variance components was tested by randomizing sequences among populations (1,000 permutations).

FIG. 1.-Geographic locations of 20 high-elevation trapping sites in North Carolina, Tennessee, Virginia, and West Virginia. Site labels correspond to information and localities in Table 1.

To determine whether the pattern of differentiation among shrew populations supported a hypothesis of isolation-by-distance, Mantel's test (Mantel 1967) was applied following the method of Hutchison and Tcmplcton (1999). This approach tested whether genetic distances between populations significantly increased with greater geographic distance between their sites of origin. Point-to-point geographic distances were measured from topographic maps of the study area. Genetic distances were calculated as mean pairwisc number of sequence differences between individuals collected from separate populations (Nei and Li 1979).

TABLE 1.-Sampling site locations, elevations, and number of individuals of each species of Sorex trapped at each site. Identical letters in the 2nd column indicate adjacent populations connected by spruce-fir corridors. NC is North Carolina, TN is Tennessee, VA is Virginia, WV is West Virginia.

TABLE 2.-Sequences of primers and adapters used for polymerase chain reaction amplification of mitochondrial DNA and amplified fragment length polymorphism of genomic DNA. Preselective and selective nucleotides are indicated in bold italics. The 3rd base pair of the general mammalian primer H-15450 was changed from adenine (Irwin et al. 1991) to thymine in order to more readily amplify Sorex sequences.

Historical demography of populations was investigated with Fu's Fs statistic (Fu 1997). We estimated θ by equating it with the mean number of observed pairwise differences. Significantly large negative F^sub S^ values, compared with a null distribution generated from a stationary population with estimated parameter θ, were interpreted as evidence of population expansion. Mismatch distributions of pairwise distances between haplotypes (Rogers and Harpending 1992) were plotted with DNAsp 3.53 software (Rozas and Rozas 1999). Mismatch distributions were compared with Poisson expectations (Slatkin and Hudson 1991), and associated raggedness indices were interpreted with the generalizable simulation results of Harpending (1994).

Amplified fragment length polymorphism marker analysis.-Total genomic DNA was digested with EcoRI (rare cutter) and Tru91 (frequent cutter) and ligated to EcoRI and Tru9I adapters. Digestion- ligation samples were incubated at 37C for 2 h and diluted 20-fold for use as template in the initial PCR reaction, which was performed with nonselective primers. This product was diluted 5-fold and used as template for the 2nd amplification with primers extended by 3 selective nucleotides. Sequences of amplified fragment length polymorphism adapters and primers are presented in Table 2. Amplified fragments were electrophoresed on 4.5% polyacrylamide gels with 10X Tris-borate-edctic acid buffer and visualized by silver staining with the Silver Sequence DNA sequencing system (Promega, Madison, Wisconsin). DNA ladders (Invitrogen, Carlsbad, California) were used as size standards on either side of gels to verify the homology of bands in separate lanes. Each of 5 primer combinations (Table 3) for each species was examined on a single gel. The 10 amplified fragment length polymorphism gels were dried on glass plates and digitally scanned for archival purposes.

The presence or absence of amplified fragment length polymorphism bands on each gel was scored visually and entered into a binary matrix. Only clearly visible bands within a size range of 100-500 bp were included. A combined raw data matrix consisting of markers scored from all 5 gels was analyzed. To estimate divergence among individuals, we used a distance calculated by the restriction site method of Nei and Li (1979), mathematically equivalent to the Dice (1945) ecological similarity index, to construct neighbor-joining trees. Confidence in the topology of trees was evaluated by bootstrap analysis with 10,000 replicates. For comparison with phylogeographic patterns inferred from mitochondria! DNA sequence analysis, we calculated Φ^sub ST^, analyses of molecular variance, and Mantel's statistic from amplified fragment length polymorphism data.

TABLE 3.-The 5 selective amplified fragment length polymorphism primer pair combinations used to amplify DNA fragments generated by double restriction digest and the number of markers generated by each primer pair for both species of Sorex. Primers are identified by the 3 selective terminal nucleoticles.

RESULTS

Variation in mitochondrial DNA sequences.-Cytochrome b and D- loop last tandem repeat sequences were aligned without ambiguity. Seven S. fumeus D-loop sequences had 12-bp deletions at the 5' end of the last tandem repeat. Five of these individuals were trapped at Whitetop Mountain and 1 each at Winespring Bald and Standing Indian. Fourteen S. fumeus sequences, including the 7 previously mentioned, had 1-bp deletions 30 bp from the 5' end of the repeat. The additional 7 individuals were collected from the Black Mountains, Mount Mitchell, and Roan Mountain populations. Last tandem repeat deletions were verified by repeated sequencing. There was no evidence of corresponding losses in flanking repeats for those individuals. To our knowledge, such deletions have not been reported elsewhere.

As expected given differences in functional constraints on the sequenced mitochondrial DNA regions (Irwin et al. 1991; Stewart and Baker 1994, 1997), D-loop last tandem repeat sequences for both species showed a greater amount of variability than Cytb sequences, as measured by π, H, and h (Table 4). Nucleotide substitutions were observed at 36 of 79 (45.6%) D-loop loci in S. cinereus and 52 of 79 (65.8%) loci in S. fumeus, compared with 29 of 500 (5.8%) Cytb loci in S. cinereus and 57 of 500 (11.4%) loci in S. fumeus. For the 44 S. cinereus individuals sampled, 21 Cytb and 31 D-loop haplotypes were noted; 47 S. fumeus individuals yielded 25 Cytb and 39 D-loop haplotypes.

Because the mitochondrial genome is inherited maternally as a unit and the topologies of trees constructed with sequence data from separate gene regions were not discordant, only those phylogenetic analyses of a combined 579-bp data set are presented. The combined S. cinereus mitochondrial DNA data set had 65 polymorphic sites (11.2%); S. fumeus had 109 (18.8%). This trend was evident in diversity indices calculated from 579-bp mitochondrial DNA sequences from each species (Table 4). Thirty-seven S. cinereus and 41 S. fumeus combined mitochondrial DNA haplotypes were observed for the 44 and 47 sequences, respectively.

TABLE 4.-Diversity indices SD calculated for individual mitochondrial DNA gene regions and a combined mitochondrial data set for 2 species of Sorex.

Phylogenetic analysis of mitochondrial DNA.-A relatively small amount of interhaplotypic sequence divergence was observed among S. cinereus sequences (X SD, 0.8 0.3% in Cytb, 6.7 3.0% in the D- loop repeat, 1.2 0.5% in the combined data set). The close similarity of S. cinereus haplotypes translated to poor support for relationships between haplotypes in a neighbor-joining analysis (not shown). There was no evidence of phylogeographic structuring of mitochondrial DNA haplotypes because individuals from the same trapping locality were scattered haphazardly through the topology of the tree.

Phylogenetic analysis \of S. fumeus mitochondrial DNA, in contrast, resolved distinct geographic clades (Fig. 2). Mean pairwise distances were greater among haplotypes of this species (1.5 1.8% in Cytb, 18.6 11.1% in the D-loop repeat, 3.4 2.0% in the combined data set). The S. fumeus phylogenetic topology corresponded well with individuals' sites of origin, with relatively good bootstrap support for several clades in close geographic proximity.

Phylogeographic structure of mitochondrial DNA.-Partitioning the total molecular variance in pairwise population comparisons into within-population and total variance components yielded pairwise Φ^sub ST^ values. Permutational significance tests of S. cinereus mitochondrial DNA indicated significant structure (P < 0.05) in 8 pairwise comparisons involving Whitetop Mountain, but in none of 145 other comparisons. Significant Φ^sub ST^ comparisons in S. fumeus were not as localized; structure was indicated for 22 of 105 permuted pairs, including populations from the Blue Ridge on both sides of the French Broad River and isolated populations at Roan, Grandfather, and Whitetop Mountains.

An analysis of molecular variance indicated that a significant amount (8.49%) of the total variance in the S. cinereus data set was partitioned to the between-population component, which corresponded to a Φ^sub ST^ of 0.085 and a P value of 0.008 (Table 5). Removal of the 6 Whitetop Mountain individuals from the analysis reduced the between-population component to a nonsignificant 2.79% (Φ^sub ST^ = 0.028, P = 0.213), reinforcing our conclusion that the substantial majority of phylogeographic structure within S. cinereus mitochondrial DNA in the southern Appalachians is associated with this locality. Regional groupings based on division at the French Broad River and the 8 disjunct spruce-fir regions also had nonsignificant values of Φ^sub CT^ (a statistic concerned with regional groupings relative to the metapopuiation and analogous to F^sub RT^).

A much greater amount of between-population variability (28.79%, Φ^sub CT^ = 0.288, P < 0.001) was observed in S. fumeus mitochondrial DNA (Table 5). Genetic variation of smoky shrew populations showed clear large-scale geographic structure, as evidenced by a north-south division at the French Broad River (12.31%, Φ^sub CT^ = 0.123, P = 0.018). A grouping reflecting the 7 spruce-fir islands in which smoky shrews were trapped also was significant (32.99%, Φ^sub CT^ = 0.330, P < 0.001). The degree of population structuring in 5. fumeus mitochondrial DNA indicates strong restriction of mitochondrial gene flow across the low- elevation French Broad River and between northern mountaintop populations.

Genetic distances based on S. cinereus mitochondrial DNA sequences did not increase significantly with greater geographic distance (1,000,000 permutations of Mantel's test, r = -0.006, P = 0.475). The geographic subdivision of S. fumeus mitochondrial DNA variation was, in contrast, adequately described by an isolation-by- distance model (1,000,000 permutations of Mantel's test, r = 0.565, P = 0.002).

Historical population demography.-Significantly negative values of F^sub S^ (Fu 1997) were associated with a demographic model inferred from mismatch distributions that implied sudden expansion. The F^sub S^ value obtained for S. cinereus mitochondrial DNA (F^sub S^ = -25.2) was significant (P < 0.001). The mismatch distribution for this data set showed no significant deviation from Poisson expectation (Kolmogorov-Smirnoff test, P = 0.960) and had the shape usually associated with expanding populations. The raggedness index was low (r = 0.005, P = 0.970) and well within the range indicating a smooth distribution (Harpending 1994). The F^sub S ^value obtained for S. fumeus mitochondrial DNA (F^sub S^ = -21.0) had an equally low probability value. The shape of this mismatch distribution was remarkably similar to that obtained for S. cinereus (Kolmogorov- Smirnoff test, P = 0.670), as was the smoothness of the curve (r = 0.003, P = 0.910), with only slight deviation from Poisson expectation.

Variation in amplified fragment length polymorphisms.-Four S. cinereus and 5 S. fumeus individuals were eliminated from the analysis because selective amplification failed for >1 primer pair. One S. fumeus sequence for which 1 primer pair failed to amplify was included in the analysis, with data from the unsuccessful amplification coded as missing. The number of bands recovered from 40 5. cinereus and 42 S. fumeus individuals was 664 and 650, respectively, with means of 133 and 130 bands for each primer combination respectively, (Table 3). Ninety-seven percent of S. cinereus markers were polymorphic compared with 93.1% of S. fumeus markers.

FIG. 2.-Neighbor-joining tree of 47 Sorex fumeus combined mitochondrial DNA sequences based on the HKY + I + G model of sequence divergence. The 1st number for each individual is a unique identifier; numbers in parentheses indicate site of capture as in Table 1. Numbers on branches indicate percentage bootstrap support on the basis of 10,000 bootstraps. Bootstrap values <50% are not shown. Branch lengths are proportional to the estimated number of substitutions per nucleotide site along each branch.

Phylogenetic analysis of amplified fragment length polymorphisms.- Bootstrap support was strong for the topology of neighbor-joining trees on the basis of Nei and Li (1979) distances calculated from both S. cinereus (Fig. 3) and S. fumeux (Fig. 4) amplified fragment length polymorphism data. Both analyses resolved numerous clades localized with respect to geography. Nei-Li distances between individuals were similar for masked and smoky shrews, with a mean SD of 0.169 0.048 and 0.166 0.042, respectively.

Phylogeographic structure of amplified fragment length polymorphisms.-Pairwise Φ^sub ST^ values implicated the Whitetop Mountain populations as the primary source of structure for both species of Sorex. Ten of 153 pairwise comparisons of S. cinereus populations were significant, all of which involved Whitetop Mountain; 9 of 12 significant pairwise comparisons of 5. fumeus populations also involved this locality.

A significant amount of the total amplified fragment length polymorphism variation of each Sorex species was partitioned into an among-population component by analysis of molecular variance (Table 5). S. cinereus populations differed by 17.46% of total variance and S. fumeus populations by 15.57%, both amounts significant at the P < 0.001 level. Neither species showed a genetic division at the French Broad River, although S. fumeus approached significance for this scheme (1.44%, Φ^sub CT^ = 0.014, P = 0.097). Indeed, if the north-south division of smoky shrew populations was shifted southward from the river into the Great Smoky Mountains, such that populations 9-12 became part of a group comprised of the 9 northernmost populations, the analysis of molecular variance became significant (3.62%, Φ^sub CT^ = 0.036, P = 0.004). This indicates that a true north-south genetic division exists in S. fumeus populations, although the break is not presently aligned with the French Broad River. A design delineating the 8 disjunct spruce- fir patches was significant for both species (S. cinercux P = 0.006; S. fumeus P = 0.022).

TABLE 5.-Hierarchical analysis of molecular variance of different subdivisions of 18 Sorex cinereus and 15 Sorex fumeus populations for mitochondrial DNA (mtDNA) and amplified fragment length polymorphism (AFLP) data sets. The among-group (AG), among-site within-group (APWG), and within-site (WP) components of variation are presented. P values are based on 1,023 permutations. Negative percentage total variance and Φ^sub ST^ values indicate greater within-group than among-group variance and are functionally equivalent to 0. Values in bold are significant at the P < 0.05 level.

DISCUSSION

Comparative phylogeography of Sorex species.-Whereas variation in S. cinereus mitochondrial DNA appears insufficient for detection of significant phylogenetic structure, we were able to construct topologies with significant bootstrap support from S. fumeus mitochondrial DNA sequences and amplified fragment length polymorphisms for both species. The minimal population structure inferred from S. cinereus amplified fragment length polymorphisms was geographically haphazard when compared with the distinct geographic partitioning of S. fumeux populations. It has been suggested that population genetic analyses based on F^sub ST^ might be better suited to examining recent population subdivision than phylogeny-based inference (Buerkle 1999) because significant features of subdivision might be evident before population genetic parameters approach equilibrium values (Slatkin 1993). Nevertheless, we detected clear differences in the phylogeographic patterning of 5. cinereus and S. fumeus populations with distance-based phylogenetic analyses.

FIG. 3.-Neighbor-joining tree of 40 Sorex cinereus combined amplified fragment length polymorphism presence-absence matrices consisting of 664 unique bands based on the Nei and Li distance metric. Numbers as in Fig. 2.

FIG. 4.-Neighbor-joining tree of 42 Sorex fumeus combined amplified fragment length polymorphism presence-absence matrices consisting of 650 unique bands based on the Nei and Li distance metric. Numbers as in Fig. 2.

Comparisons based on F^sub ST^ provide further evidence that patterns of spatial genetic structuring among southern Appalachian populations of S. cinereus and S. fumeus are strikingly distinct; contrary to expectation, the latter show greater hierarchical subdivision than the former. The S. fumeus metapopulation consists of several genetically differentiated regions centered on isolated spruce-fir islands, within which local populations are relatively similar. Gene flow has homogenized genetic variability w\ithin islands, whereas structure among islands has been preserved by physical barriers and climatic effects on vegetation. For this species, a comparable amount of interregional differentiation was detected in mitochondrial and nuclear DNA. The barrier to migration at the French Broad River and the allopatric fragmentation of widespread spruce-fir habitat into small, isolated northeastern mountaintop islands apparently have been the most significant historical influences on the sorting of S. fumeus haplotypes. The S. cinereus metapopulation, conversely, has little mitochondrial DNA structure, with no evidence of large-scale subdivision. Amplified fragment length polymorphism analysis of S. cinereus nuclear DNA did detect significant structure among spruce-fir regions, but the pattern inferred differed from that of S. fumeus in that the subdivision of populations with maximum intergroup differentiation in an analysis of molecular variance had no discernable geographic pattern.

These observations belie our prediction of greater geographic structure in S. cinereus populations on the basis of the species' more restricted present-day Appalachian distribution. We hypothesized that barriers to dispersal would have a more significant effect, and hence that genetic drift would play a greater role in fixation of alternate alleles, in the masked shrew. A factor that we did not consider when making this prediction was the mean size of isolated populations. In this study, S. cinereus was trapped in numbers that indicated a density approximately 4 times that of S. fumeus. This observation is supported by previous estimates of soricid abundance (Laerm et al. 1999). Fixation of alternative neutral alleles in 2 populations by drift takes on average 4 times the effective population size generations, as predicted by the diffusion approximation of the Wright-Fisher model of stochastic genetic drift (Fisher 1930; Wright 1931). On the basis of the observed trapping densities, fixation of alleles between a pair of S. fumeus populations should take place approximately 4 times faster than between a pair of S. cinereus populations. Given the limited time since fragmentation of the spruce-fir habitat (approximately 8,000 years), isolated S. cinereus populations might have remained of sufficient size not to be measurably affected by genetic drift.

Demographic and phylogeographic history.-If genetic drift is responsible for shaping Sorex population structure, then the populations at Whitetop Mountain warrant special consideration. All significant pairwise Φ^sub ST^ comparisons of S. cinereus mitochondrial DNA involved this population, and its elimination from the analysis of molecular variance reduced the interpopulation component of variation to nonsignificance. Several smoky shrews captured at Whitetop Mountain possessed the large D-loop last tandem repeat deletion, providing an additional indication of the peculiarity of this population.

We believe that a localized historical event induced severe bottlenecking in shrew populations at Whitetop Mountain. Although the exact nature of this event is uncertain (be it a forest fire, climatic change, or a biological agent), the catastrophe apparently reduced Sorex numbers at Whitetop drastically by restricting local spruce habitat and extirpating Fraser fir from this mountaintop. Following this event, the spruce population increased to its present numbers, and associated fauna repopulated the area. Given a significant reduction in the size of Sorex populations, fixation of alleles would have been rapid. We would expect other boreal species native to this site to exhibit genetic uniqueness from surrounding populations if they too were affected. Indeed, a comparably great genetic distance between the Whitetop Mountain population of red- backed voles and nearby populations was noted by Reese et al. (2001).

A possible clue to the history of this region might be that spruce alone dominates the Whitetop forest at elevations at which spruce and fir codominate on nearby Mount Rogers (Ware 1999). To explain this pattern, Rheinhardt and Ware (1984) invoked the warmer temperatures of the post-Wisconsinian Hypsithermal Interval ~5,000 years ago and the associated shifting of ecotone boundaries. Relative elevations of these peaks are such that during the warm, dry Hypsithermal Maximum, the lower elevational limit for fir was too high for its survival on Whitetop (1,682 m), but not on the peak of Mount Rogers (1,746 m). The minimum elevation limit of spruce was somewhat lower, so it was able to survive on both peaks. Although during subsequent cooling, fir extended its range down the slopes of Mount Rogers to its present distributional limit 182 m below the height of Whitetop's summit (Rheinhardt and Ware 1984), the gap between the peaks (1,347 m in elevation) was too low for fir to subsequently invade Whitetop Mountain. This hypothesis is supported by palynological evidence that spruce and fir significantly decreased in abundance during the Hypsithermal Interval (Delcourt and Delcourt 1985) and could account for the genetic uniqueness of populations of boreal fauna on Whitetop Mountain, which would have been significantly reduced and genetically bottlenecked during the climatic "squeeze" of the Hypsithermal Interval, especially in light of the limited 150-ha present-day extent of spruce-fir at this site (Rheinhardt 1984).

Fu's F^sub S^ statistic and histograms of pairwise differences generated from mitochondrial DNA sequence data give us additional insight into the recent history of these species. Despite a precipitous historical decline in the extent of southern Appalachian spruce-fir habitat, significantly negative values of Fu's test and the shape of mismatch distributions for both Sorex species are indicative of abrupt historical population expansions. Population bottlenecks will erase the signs of past population expansion as detected by either of these methods (Excoffier and Schneider 1999), so it is reasonable to conclude that neither species has experienced a prolonged bottleneck. One might expect widespread bottlenecking of Sorex populations to have occurred, given the climatic history of the region; more boreal species such as the northern flying squirrel certainly show strong evidence of a bottleneck (Arbogast 1999; Browne et al. 1999). The pattern inferred from these analyses is, however, consistent with the fact that the ranges of both species of shrew, unlike those of many boreal vertebrates, extend to elevations below the spruce-fir ecotone.

Because techniques available for assigning a dale to these expansions are unreliable and quite sensitive to departures from specific model parameters (Schneider and Excoffier 1999), their exact timing is unclear. Expansion might have coincided with migration into formerly glaciated areas following the retreat of the Laurentide ice sheet. During the glacial advance, the tree line in the southern Appalachians was depressed to ~1,000 m and mountain summits were mantled by extensive areas of alpine tundra (Nalepa et al. 2002). When the glaciers began their fluctuating retreat, the warming climate allowed expansion out of southern refugia, and forests spread to previously inhospitable mountaintops (Delcourt et al. 1993). Given the extensive glacial history of the region, an alternative explanation might be that the genetic signal is convoluted with another of the many mini-ice ages of the Pleistocene induced by the Milankovitch cycle. A further possibility is that exponential growth of Sorex populations was more recent and coincided with the extension of spruce-fir habitat to lower elevations from localized mountaintop refuges during the cool period following the Hypsithermal Interval (Delcourt and Delcourt 1985).

Comparative phylogeography of Appalachian vertebrates.-Collation of the genetic patterning observed in this study with previous phylogeographic studies of Appalachian vertebrates is informative. Deer mice (Peromyscus maniculalus) show no evidence of population structuring in the southern Appalachians. This species generally occurs at relatively low elevations (>800 m) and inhabits largely contiguous regions with few barriers to migration (Browne et al. 1999). Because the red-backed vole (Clethrionomys gapperi) is more closely tied to the boreal spruce-fir habitat, one would expect migratory barriers between mountaintops for this species to be less porous. Indeed, sequence analysis of D-loop mitochondrial DNA revealed significant genetic structuring between Appalachian C. gapperi populations (Ferree 1998), and 2 microsatellite loci provided similar evidence in relict Virginia populations (Reese et al. 2001). Southeastern populations of the northern flying squirrel (Glaucomys sahrinus), which reside almost exclusively in spruce-fir patches at the southern extreme of their range, have <10% of the heterozygosity of northern and western populations (Browne et al. 1999). In a survey of allozymes and mitochondrial DNA, Crespi et al. (2003) detailed significant structure among populations of the endemic Appalachian salamander Desmognathus wrighti, which has a distribution very closely paralleling the spruce-fir habitat.

Our findings extend the comparative hierarchical approach espoused by Browne et al. (1999) of simultaneously examining genetic data for a variety of vertebrate species along a gradient of habitat specificity with different levels of restriction to the spruce-fir ecosystem. As in other southern Appalachian species, the degree of insularity imposed on shrew populations by reliance on boreal habitat largely dictates the genetic distance between them. Sorex species exhibit intermediate levels of structuring compared with other Appalachian vertebrates, being neither as vagile as P. maniculatus nor as insular as G. sabrinus, and the phylogeographic patterns we detected are consistent with fragmentationfrom large, widespread metapopulations at the glacial maximum.

ACKNOWLEDGMENTS

We thank K. Kron, D. Anderson, C. Zeyl, M. Steele, L. Carraway, and several anonymous reviewers for their comments on earlier drafts of this manuscript and K. Kron and D. Anderson for the loan of laboratory equipment and facilities. Thanks to E. Jung for technical advice on sequencing and the staff of the University of Florida Interdisciplinary Center for Biotechnology Research for their amplified fragment length polymorphism workshop. This project was funded in part by the following grants to T. W. Sipe: Lindsay S. Olive and Thelma Howell Memorial Scholarships from Highlands Biological Station, the Vecellio Fund for Graduate Research from the Wake Forest Biology Department, the Theodore Roosevelt Memorial Fund from the American Museum of Natural History, and a Grant-in-Aid of Research from the American Society of Mammalogists. The procedures of this study were approved by the Wake Forest University Animal Care and Use Committee and were in accordance with American Society of Mammalogists guidelines (Animal Care and Use Committee 1998).

LITERATURE CITED

ANIMAL CARE AND USE COMMITTEE. 1998. Guidelines for the capture, handling, and care of mammals as approved by the American Society of Mammalogists. Journal of Mammalogy 79:1416-1431.

ARBOGAST, B. 1999. Mitochondrial DNA phylogeography of the New World flying squirrels (Glaucomys): implications for Pleistocene biogeography. Journal of Mammalogy 80:142-155.

BRAUN, D. D. 1989. Glacial and periglacial erosion of the Appalachians. Geomorphology 2:233-256.

BROWNE, R. A., P. D. WEIGL, E. EAGLESON, AND J. KELLY. 1999. Mountaintops as islands: I. Genetic variation among southern Appalachian populations of the endangered northern flying squirrel, Ghiucomys sabrinus (Mammalia: Sciuridae). Pp. 205-214 in Proceedings of the Appalachian biogeography symposium (R. P. Eckerlin, ed.). Special Publication 7, Virginia Museum of Natural History, Martinsville, Virginia.

BUERKLE, C. A. 1999. The historical pattern of gene flow among migratory and nonmigratory populations of prairie warblers (Aves: Parulinae). Evolution 53:1915-1924.

CRESPI, E. J., L. J. RISSLER, AND R. A. BROWNE. 2003. Testing Pleistocene refugia theory: phylogeographical analysis of Desmognathus wrighti, a high-elevation salamander in the southern Appalachians. Molecular Ecology 12:969-984.

DELCOURT, H. R., AND P. A. DELCOURT. 1985. Quaternary palynology and vegetational history of the southeastern United States. Pp. 1- 37 in Pollen records of late-quaternary North American sediments (V. Bryant, Jr., and R. Holloway, eds.). American Association of Stratigraphic Palynologists Foundation, Austin, Texas.

DELCOURT, P. A., H. R. DELCOURT, D. F. MORSE, AND P. A. MORSE. 1993. History, evolution, and organization of vegetation and human culture. Pp. 47-79 in Biodiversity of the southeastern United States lowland terrestrial communities (W. H. Martin, S. G. Boyce, and A. C. Echternacht, eds.). John Wiley & Sons, Inc., New York.

DICE, L. R. 1945. Measures of the amount of ecologic association between species. Ecology 26:297-302.

DULL, C. W., J. E. WARD, H. D. BROWN, G. W. RYAN, W. H. CLERKE, AND R. J. UHLER. 1988. Evaluation of spruce and fir mortality in the southern Appalachian Mountains. Protection report R8-PR 13, United States Department of Agriculture, Forest Service, Atlanta, Georgia.

EXCOFFIER, L., AND S. SCHNEIDER. 1999. Why hunter-gatherer populations do not show signs of Pleistocene demographic expansions. Proceedings of the National Academy of Sciences 96:10597-10602.

EXCOFFIER, L., P. E. SMOUSE, AND J. M. QUATTRO. 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131:479-491.

FERREE, P. M. 1998. Gene flow between geographically isolated populations of the redback vole in the southern Appalachians: a coalescence-based study. M.S. thesis, Wake Forest University, Winston-Salem, North Carolina.

FISHER, R. A. 1930. The genetical theory of natural selection. Clarendon, Oxford, United Kingdom.

FU, Y.-X. 1997. Statistical tests of neutrality of mutations against population growth, hitchhiking, and background selection. Genetics 147:915-925.

HALL, T. A. 1999. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows. Nucleic Acids Symposium Series 41:95-98.

HARPENDING, H. 1994. Signature of ancient population growth in a low-resolution mitochondrial DNA mismatch distribution. Human Biology 66:591-600.

HUTCHISON, D. W., AND A. R. TEMPLETON. 1999. Correlation of pairwise genetic and geographic distance measures: inferring the relative influences of gene flow and drift on the distribution of genetic variability. Evolution 53:1898-1914.

IRWIN, D. M., T. D. KOCHER, AND A. C. WILSON. 1991. Evolution of the cytochrome b gene of mammals. Journal of Molecular Evolution 32:128-144.

LAERM, J., W. M. FORD, T. S. MCCAY, M. A. MENZEL, L. T. LEPARDO, AND J. L. BOONE. 1999. Soricid communities in the southern Appalachians. Pp. 177-193 in Proceedings of the Appalachian biogeography symposium (R. P. Eckerlin, ed.). Special Publication 7, Virginia Museum of Natural History, Martinsville, Virginia.

MACARTHUR, R. H., AND E. O. WILSON. 1967. The theory of island biogeography. Princeton University Press, Princeton, New Jersey.

MANTEL, N. 1967. The detection of disease clustering and a generalized regression approach. Cancer Research 27:209-220.

NALEPA, C. A., P. LUYKX, K.-D. KLASS, AND L. L. DEITZ. 2002. Distribution of karyotypes of the Cryptocercus punctulatus species complex (Dictyoptera: Cryptocercidae) in the southern Appalachians: relation to habitat and history. Annals of the Entomological Society of America 95:276-287.

NEI, M. 1987. Molecular evolutionary genetics. Columbia University Press, New York.

NEI, M., AND W.-H. LI. 1979. Mathematical model for studying genetic variation in terms of restriction endonucleases. Proceedings of the National Academy of Sciences 76:5269-5273.

POSADA, D., AND K. A. CRANDALL. 1998. ModelTest: testing the model of DNA substitution. Bioinformatics 14:817-818.

REESE, C. L., J. M. WATERS, J. F. PAGEI.S, AND B. L. BROWN. 2001. Genetic structuring of relict populations of Gapper's red-backed vole (Clethrionomys gapperi). Journal of Mammalogy 82:289-301.

RHEINHARDT, R. D. 1984. Comparative study of composition and distribution patterns of subalpine forests in the Balsam Mountains of southwest Virginia and the Great Smoky Mountains. Bulletin of the Torrey Botanical Club 1 11:489-493.

RHEINHARDT, R. D., AND S. WARE. 1984. The vegetation of the Balsam Mountains of southwest Virginia: a phytosociological study. Bulletin of the Torrey Botanical Club 111:287-300.

ROGERS, A. R., AND H. C. HARPENDING. 1992. Population growth makes waves in the distribution of pairwise genetic differences. Molecular Biology and Evolution 9:552-569.

ROZAS, J., AND R. ROZAS. 1999. DNAsp Version 3: an integrated program for molecular population genetics and molecular evolution analysis. Bioinformatics 15:174-175.

SCHNEIDER, S., AND L. EXCOFFIER. 1999. Estimation of past demographic parameters from the distribution of pairwise differences when the mutation rates vary among sites: application to human mitochondrial DNA. Genetics 152:1079-1089.

SCHNEIDER, S., D. ROESSLI, AND L. EXCOFFIER. 2000. Arlequin version 2.000: A software for population genetics data analysis. Genetics and Biometry Laboratory, University of Geneva, Geneva, Switzerland.

SLATKIN, M. 1993. Isolation by distance in equilibrium and non- equilibrium populations. Evolution 47:264-279.

SLATKIN, M., AND R. R. HUDSON. 1991. Pairwise comparisons of mitochondrial DNA sequences in stable and exponentially growing populations. Genetics 129:555-562.

SOUTHERN, S. O., P. J. SOUTHERN, AND A. E. DIZON. 1988. Molecular characterization of a cloned dolphin mitochondrial genome. Journal of Molecular Evolution 28:32-42.

STEELE, M. A., AND R. A. POWELL. 1999. Biogeography of small mammals in the southern Appalachians: patterns of local and regional abundance and distribution. Pp. 155-165 in Proceedings of the Appalachian biogeography symposium (R. P. Eckerlin, ed.). Special Publication 7, Virginia Museum of Natural History, Martinsville, Virginia.

STEWART, D. T., AND A. J. BAKER. 1994. Patterns of sequence variation in the mitochondrial D-loop region of shrews. Molecular Biology and Evolution 11:9-21.

STEWART, D. T., AND A. J. BAKER. 1997. A phytogeny of some taxa of masked shrews (Sorex cinereus) based on mitochondrial DNA D-loop sequences. Journal of Mammalogy 78:361-376.

SWOFFORD, D. 2002. PAUP* 4.0b10. Phylogenctic analysis using parsimony and other methods. Sinauer Associates, Inc., Publishers, Sunderland, Massachusetts.

TAJIMA, F. 1993. Measurement of DNA polymorphism. Pp. 37-59 in Mechanisms of molecular evolution (N. Takahata and A. G. Clark, eds.). Sinauer Associates, Inc., Publishers, Sunderland, Massachusetts.

TAJIMA, F., AND M. NHI. 1984. Estimation of evolutionary distance between nucleotide sequences. Molecular Biology and Evolution 1:269- 285.

THOMPSON, J. D., T. J. GIBSON, F. PLEWNIAK, F. JEANMOUGIN, AND D. G. HIGOINS. 1997. The Clustal X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Research 24:4876-4882.

USHER, M. B. 1995. Small populations: fragmentation, population dynamics, and population genetics. Pp. 1 1-21 in The role of genetics in conserving small populations (T. E. Tew, T. J. Crawford, J. W. Spencer, D. P. Stevens, M. B. Usher, and J. Warren, eds.). Joint Nature Conservation Committee, Petersborough, United Kingdom.

WARE, S. 1999. Paleoclimatic causes of altitudinal and geographical distribution of high elevation woody species in Virginia. Pp. 45-57 in Proceedings of the Appalachian biogeography symposium (R. P. Eckerlin, ed.). Special \Publication 7, Virginia Museum of Natural History, Martinsville, Virginia.

WHITE, P. S., AND C. V. COGBILL. 1992. Spruce-fir forests of eastern North America. Pp. 2-39 in Ecology and decline of red spruce in the Eastern United States (C. Eager and M. B. Adams, eds.). Springer-Verlag, New York.

WILKINSON, G. S., AND A. M. CHAPMAN. 1991. Length and sequence variation in evening bat D-loop mtDNA. Genetics 128:607-617.

WRIGHT, S. 1921. Systems of mating. Genetics 6:111-178.

WRIGHT, S. 1931. Evolution in Mendelian populations. Genetics 16:97-159.

Submitted 9 October 2003. Accepted 27 October 2003.

Associate Editor was Robert D. Bradley.

TAVIS W. SIPE* AND ROBERT A. BROWNE

Department of Biology, Wake Forest University, Winston-Salem, NC 27109, USA

* Correspondent: geneticdrift@att.net

2004 American Society of Mammalogists

www.mammalogy.org

Copyright American Society of Mammalogists Oct 2004

Source: Journal of Mammalogy

More News in this Category