Identification and distribution of the Olympic Shrew (Eulipotyphla: Soricidae), Sorex rohweri Rausch et al., 2007 in Oregon and Washington, based on USNM specimens

In the mid-1980s, the Patuxent Wildlife Research Center's Biological Survey Unit (BSU) in the Mammal Division, National Museum of Natural History (USNM), Washington, D.C., accessioned several large collections of mammals from the northwestern coastal states of Oregon and Washington. These vouchers were generated by R. Bruce Bury and Paul Stephen Corn under the auspices of the Old-growth Wildlife Habitat Program of the U.S. Forest Service. This initiative was established to address a lack of information regarding vertebrate communities in old-growth forests in the Cascade and Coast Ranges and the effects of fire and logging on those species associations (Corn et al. 1988, Corn & Bury 1991). Among the specimens of shrews from this project were a number of individuals that were originally identified as—and most closely matched—the masked shrew, Sorex cinereus Kerr, 1792. Those from Washington were identified and installed in the USNM collection as such. This species is unknown in Oregon (Verts & Carraway 1998), however, and certain key characteristics of the specimens were at variance with all known populations of the masked shrew. For these two reasons, this group of specimens was not incorporated into the main collection.

We recently re-examined this collection, and, based on character differences and morphometric comparisons presented herein, we determined that the Oregon “S. cinereus”, as well other specimens from Washington traditionally identified as that species, are instead individuals of the Olympic Shrew, Sorex rohweri Rausch et al., 2007. Our identifications greatly expand the southern limit of the geographical distribution of S. rohweri, which previously was known only from western Washington and the southwestern corner of British Columbia south of the Frasier River (Rausch et al. 2007, Nagorsen & Panter 2009). These results additionally reduce the known geographical range of the species S. cinereus in Washington by effectively eliminating the subspecies S. c. streatori Merriam, 1895 from that state.

Herein, we present the results of our identifications and illustrate the distributions of S. rohweri and S. c. streatori as represented by USNM specimens from Alaska, British Columbia, Oregon, and Washington.

We inspected and reviewed the identifications of 1230 USNM specimens of Sorex from Oregon, Washington, and British Columbia (see Appendix), using qualitative and quantitative characters described by Junge & Hoffman (1981), Alexander (1996), Verts & Carraway (1998), Rausch et al. (2007), and Nagorsen & Panter (2009). Anatomical terminology follows Junge & Hoffmann (1981). Among the soricids of western Oregon, Washington, and British Columbia, the species most similar in size and general appearance to the specimens of interest here are S. cinereus, S. rohweri, and S. vagrans Baird, 1857, and we limit our discussion to those three species. There are two currently recognized subspecies of S. cinereus in this region: the generally smaller, paler S. c. cinereus occurs inland to the east of the Cascade Range, and the typically larger, darker S. cinereus streatori Merriam, 1895, ranges along the northern Pacific coast (Merriam 1895, Jackson 1928, Hall 1981). Because mean size may differ significantly between these two taxa, we treated them as separate entities in our quantitative analyses.

We recorded the following standard external measurements from skin tags (Table 1): Total length (TOT), tail length (TL), length of hind foot including the claw (HF), and weight (WT). We calculated head-and-body length (HB) by subtracting tail length from total length and proportional length of tail (TL%) by dividing tail length by head-and-body length. One of us (NW) used a handheld digital caliper or an ocular micrometer in a binocular dissecting scope to measure 14 cranio-mandibular variables to the nearest 0.1 mm as described by Junge & Hoffmann (1981), Neithammer & Krapp (1990), and Woodman & Timm (1993), unless otherwise indicated: greatest length of skull (GLS), condylobasal length (CBL), breadth of braincase (BB), breadth of zygomatic plate (ZP), postorbital breadth (PO), width across the upper second molars, from the buccal edge of the metastyles (M2B); postglenoid width (PGW); length of palate (PL); length of unicuspid toothrow (UTR); length of upper molariform toothrow, from the posterior of M3 to the anterior of P4 (MTR); dentary length along the buccal surface, from the superior articular condyle to the posterior cingulum of the lower first incisor (DL; new measure); height of the coronoid process (HCP); length of the mandibular toothrow, from the posterior edge of M3 to the anterior edge of P3 (TRM); breadth of the articular condyle (BAC). Although greatest length of skull and condylobasal length are strongly correlated and redundant, we considered it relevant to report both measurements for each taxon because each is used as a character in various authoritative keys and descriptions (e.g., Junge & Hoffmann 1981, Verts & Carraway 1998).

Multivariate analyses were carried out using Systat 11 (Cranes Software, Bangalore). We used 12 log₁₀-transformed skull variables (CBL, ZP, PO, M2B, PGW, PL, UTR, MTR, DL, HCP, TRM, BAC; excluding GLS, BB) in principal components analyses (PCA) and discriminant function analyses (DFA) to describe the four taxa morphometrically and to aid in identifying individual variables or suites of variables useful for distinguishing taxa. We ran PCAs and DFAs using both 4-taxon (i.e., S. cinereus cinereus, S. c. streatori, S. rohweri, S. vagrans) and 3-taxon (S. vagrans excluded) models.

To better understand the species associated with the Olympic Shrew, gauge its contribution to the small mammal communities in which it occurs, and understand its habitat distribution, we include results from Corn et al. (1988) and Corn & Bury (1991) and compiled data from field catalogs maintained in the USNM Mammal Division.

In Washington, Corn et al. (1988) sampled ten forest stands and two recent (<10 yrs) clear-cuts. The stands were classified by age of the dominant Douglas-Fir trees [Pseudotsuga menziesii (Mirb.)] and, in the case of old-growth forest, along a moisture gradient. These included one wet, 3 moderately moist, and one dry old-growth (195–450 yrs) stands; 3 mature (80–195 yrs) stands; and 2 closed-canopy young (30–80 yrs) stands. Two pitfall sets were established 25 m apart in each stand, each set including 12 pitfalls arrayed along 3 radiating, 5-m aluminum drift fences. Each stand was sampled for 180 days, beginning in May 1983 (Corn et al. 1988), for a total of 51,840 trapnights.

In western Oregon, Corn & Bury (1991) sampled 48 stands in Benton, Coos, Douglas, Lane, and Lincoln counties. These included 27 old-growth (150–525 yrs) stands; 10 mature (80–120 yrs) stands; 8 closed-canopy young (40–75 yrs) stands; and 3 recent (<10 yrs) clear-cuts. A square grid of 36 pitfall traps 15 m apart was set up in each stand. These were monitored continuously for 50 days in 1984 and 30 days in 1985 (Corn & Bury 1991), for a total of 138,240 trapnights.

We identified 44 specimens of Sorex rohweri from 6 counties in western Washington and 4 counties in western Oregon (Fig. 1) on the basis of qualitative morphological characters. These specimens include all USNM individuals previously identified as S. cinereus streatori from Washington. Based on USNM loan records, Rausch et al. (2007) inspected all USNM specimens of S. c. streatori from Washington, and in retrospect, they appear to indicate that some of these specimens may be S. rohweri. Explicit identifications of these specimens were not published, however, and two USNM localities are plotted only on the distribution map for S. rohweri (Rausch et al. 2007:Fig. 1). Moreover, the relevant taxonomic changes were not reported to USNM personnel (Mammal Division loan records). Our identification of these USNM specimens as S. rohweri (Appendix) removes evidence that supported the presence of S. c. streatori in Washington.

Although Rausch et al. (2007) provided few measurements to substantiate their estimates, they stated that S. rohweri is generally similar in size to S. vagrans and larger externally than either S. cinereus cinereus or S. c. streatori. They also described S. rohweri as larger in most cranial dimensions (particularly lengths) than S. c. streatori. In contrast, our samples indicate that S. rohweri and S. c. cinereus are more similar in external dimensions, and both taxa are considerably smaller than either S. c. streatori or S. vagrans (Table 1). Sorex vagrans possesses a relatively broader cranium than the other two species (Table 1), and S. rohweri is the smallest, or among the smallest, of the four taxa in most cranial dimensions we recorded (see also Nagorsen & Panter 2009:Table 2). Regardless, individual measurements overlap greatly among S. rohweri, S. cinereus, and S. vagrans in the Pacific Northwest and alone provide little assistance in differentiating these species.

Instead, a suite of cranio-dental and external characters is most useful in distinguishing these species (see Rausch et al. 2007, Nagorsen & Panter 2009). The two subspecies of the masked shrew in the Pacific Northwest typically possess the same qualitative characters, although S. c. streatori tends to have larger, more massive, more strongly pigmented dentition than S. c. cinereus; the bones of its skull are generally thicker; and the sagittal and lambdoidal sutures are more likely to be fused.

Viewed anteriorly, the medial edges of the upper first incisors (I1s) of S. rohweri tend to diverge greatly, forming an angle of up to 45° (Rausch et al. 2007). In S. cinereus and S. vagrans, however, the medial edges of I1s are typically straight and appear approximately parallel or diverge only slightly.

The medial tine on the I1 of S. rohweri is weak and often difficult to discern; pigmentation is pale or lacking. The medial tine is located proximal to the margin of pigmentation on the primary cusp of the tooth. In contrast, the medial tines of S. cinereus and S. vagrans are typically obvious and noticeably pigmented (strongly pigmented in S. vagrans). In S. cinereus, the medial tine extends well distal to the margin of pigmentation on the body of I1. In S. vagrans, the medial tine is positioned at or proximal to the margin of pigmentation.

In occlusal view, the unicuspid toothrows of S. rohweri and S. cinereus appear long, straight, and even, and the individual teeth generally diminish in size from the first unicuspid (U1) to the fifth (U5). In S. rohweri, U3 and U4 may appear equal in length, but the posterior edge of U3 is typically narrower than the neighboring anterior margin of U4. In S. c. cinereus, U3 is slightly longer and narrower than U4, whereas, in S. c. streatori, U3 is slightly longer, but typically as wide as, or wider than, U4. In contrast, the unicuspid toothrow of S. vagrans is relatively short and irregular, with U3 noticeably shorter and slightly narrower than U4.

Pigmentation of the dentition of S. rohweri is typically paler overall than in S. cinereus and S. vagrans. The lingual ridge, which extends from the main cusp to the lingual cingulum on U1–U4 (Junge & Hoffman 1981), is typically unpigmented in S. rohweri, palely pigmented in S. c. cinereus, and darkly pigmented in S. c. streatori and S. vagrans.

On the first lower incisor (i1) of S. rohweri and S. cinereus, the pigment on the distal end of the tooth is separated from the pigment on the two more proximal denticles, whereas, in S. vagrans, pigmentation is continuous from the distal tip to both denticles.

In Sorex rohweri, a large, obvious postmandibular foramen is typically present on one or both dentaries: among 43 specimens we inspected, 25 (58%) had the foramen on both dentaries, 11 (26%) had the foramen on one dentary, 7 (16%) lacked the foramen. Among S. cinereus, in contrast, a postmandibular foramen is atypical. Where a foramen is present, it is typically tiny. Among a sample of 63 S. c. cinereus from Oregon and Washington, 38 (60%) lacked a postmandibular foramen, 18 (29%) had the foramen on one dentary, and 7 (11%) had foramina on both dentaries. Among 110 S. c. streatori from British Columbia and Alaska, 75 (68%) lacked a foramen, 25 (23%) had a foramen on one dentary, and 10 (9%) had foramina on both dentaries. Our sample of S. c. streatori included a subsample of 14 specimens from the region of Juneau, Alaska, in which the incidence of a postmandibular foramen was much higher and the foramen was generally much larger: 3 (21%) had no foramen, 6 (43%) had the foramen on one dentary, and 5 (36%) on both dentaries. In S. vagrans, a small postmandibular foramen is typically present only on one dentary; in this species, the foramen may be separate or confluent with the mandibular foramen.

The plantar surface of pedal digits II–IV of S. rohweri each possesses five paired digital callosities (friction pads), whereas, there are six pairs in S. cinereus, and four or fewer in S. vagrans (van Zyll de Jong 1982, Nagorsen & Panter 2009).

Our 4-taxon PCA (not shown) clearly distinguished S. vagrans from the other three taxa based on its larger size (approximated by PC1; Table 1) and shorter palate and unicuspid toothrow (PC2). In general, neither the 4-taxon nor the 3-taxon PCA completely separated S. rohweri from the two subspecies of S. cinereus. Mean PC1 scores (± SD) from the 4-taxon PCA substantiate a difference in mean skull size, increasing from S. rohweri (−0.91 ± 0.44) to S. c. cinereus (−0.56 ± 0.50) to S. c. streatori (0.25 ± 0.43) to S. vagrans (1.40 ± 0.60) among our samples, but the SDs indicate the considerable degree of overlap among taxa.

The best 4-taxon DFA was a complete analysis that used all 12 variables and yielded an overall correct classification rate of 89% (jackknifed classification = 83%; Table 2). Most errors in classification occurred between the two subspecies of Sorex cinereus rather than among species. Sorex vagrans was correctly classified 100% of the time, and neither other species was mistaken for it. The first canonical variate (CV1) separates S. vagrans from the other three taxa with no overlap in multivariate space (Fig. 3A). The separation along this axis conveys the overall larger size of S. vagrans compared with the other taxa (Table 1), as reflected by the large positive loadings of nearly all measurements on CV1 (Table 2), in particular those that capture its broader cranium (M2B, PGW, PO) and robust mandible (BAC, HCP). In contrast, the significant negative loading of UTR on CV1 underscores the absolutely and proportionately smaller size of the unicuspid series in S. vagrans (Table 1). The second canonical variate (CV2) distinguishes S. vagrans and the two subspecies of S. cinereus from S. rohweri, reflecting its generally smaller cranial proportions (Table 1), most particularly in length measurements (e.g., CBL, PL). Our DFA does not, however, show the complete separation between S. cinereus and S. rohweri achieved by Nagorsen & Panter (2009:Fig. 1).

The best 3-taxon separation was achieved using a backward, stepwise DFA that incorporated eight variables (CBL, ZP, PGW, MTR, DL, HCP, TRM, BAC) and had an overall correct classification rate of 83% (Table 3). Most misclassifications were between the two subspecies of Sorex cinereus rather than between that species and S. rohweri. The first canonical variate (CV1) separates Sorex cinereus from S. rohweri with minimal overlap in multivariate space (Fig. 3B). The separation along CV1 communicates the generally longer cranial lengths (but not breadths) of S. cinereus (Table 1), as well as its higher coronoid process of the mandible (Table 3). The second canonical variate (CV2) distinguishes the two subspecies of S. cinereus and reflects the more robust mandible (TRM, DL, BAC) of S. c. streatori compared to that of S. c. cinereus, and its generally larger cranial proportions (Table 1).

In Skamania County, Washington, 19 S. rohweri were identified among 1867 individuals belonging to 19 species of small mammals captured during 1983 (overall trap success rate <4%). Soricids were abundant in all plots, constituting ≥50% of captures on each. The Olympic Shrew, however, was taken in only 7 of the 10 forest stands and 2 recent clear-cuts that were included in the study. Specifically, it was found in 2 of the 5 old-growth stands (one moderately moist; one dry), 2 of 3 mature stands, both of the young stands, and one of the clear-cuts. Species richness at the sites where S. rohweri occurred ranged from 9 to 14 species. The Olympic Shrew averaged 5^th in rank abundance (range 3–8) in these communities, with numbers of captures ranging from 1 to 5 individuals. In Washington, S. rohweri frequently occurs with Sorex monticola and Myodes californicus. It is also commonly associated with S. trowbridgii, S. vagrans, Neurotrichus gibbsii, Peromyscus keeni, and P. maniculatus, and infrequently with S. bendirii and Microtus oregoni.

In western Oregon, 19 S. rohweri were identified among 3047 small mammals representing 21 species captured in 45 forest stands and 3 recent clear-cuts during 1984–1985 (overall success rate >2%). As in Washington, soricids were abundant and comprised >50% of captures in each plot. The Olympic Shrew was less commonly captured here than in Washington, found in only 7 of the 48 study sites (15%). These included 2 of the 27 old-growth stands (7%), 3 of 10 mature stands (30%), and 2 of 8 young stands (25%), but in none of the three clear-cuts. The Olympic Shrew generally occurred in communities of small mammals with lower species richness (Table 4B) than in Washington (Table 4A), possibly as a consequence of different capture protocols and lower numbers of total captures at individual sites (Corn et al. 1988; Corn & Bury 1991). Where it occurred, S. rohweri was associated with 4 to 11 other species of small mammals (Table 4B). It again averaged 5^th in rank abundance (range 2–6) in these associations, although this typically corresponded to only 1 or 2 individuals per site. At one locality (site 8-62), where 10 individuals of the Olympic Shrew were recorded, it was the second most abundant species in a low-diversity young forest community of 6 species. The Olympic Shrew frequently occurs in western Oregon with Sorex trowbridgii and Myodes californicus, and less frequently with S. pacificus, Neurotrichus gibbsii, and Peromyscus maniculatus.

Only a single genus of Soricidae, Sorex, occurs in Oregon, yet statewide species diversity for this family is among the highest in North America. Our identification of Sorex rohweri from west of the Cascades Range in Oregon increases the known number of shrew species in the state to 11 (Verts & Carraway 1998). To put this number in context, the number of species in Oregon alone equals the total number of soricid species found in the 26 states east of the Mississippi River (i.e., 11 species, but in 3 genera; Whitaker & Hamilton 1998). The high species diversity is probably, in part, a reflection of the geographical variation and resulting number of distinct ecoregions in the state (Pater et al. undated). For example, the rain shadow of the north–south running Cascades Range provides a distinct transition between the moist coast and the drier interior. Six species of shrews—S. bairdi, S. bendirii, S. pacificus, S. rohweri, S. sonomae, S. trowbridgii—are mostly restricted to the western coastal region and western slopes of the Cascades; whereas, three species—S. merriami, S. palustris, S. preblei—occur in the eastern Cascades and/or the generally drier eastern two-thirds of the state.

Our re-identification of all USNM specimens of S. c. streatori in western Washington as S. rohweri confirms that the former taxon is unknown in the state. Based on this fact and the evidence provided by Nagorsen & Panter (2009), S. c. streatori is present only north of the Frasier River in western British Columbia and southern Alaska (Fig. 2).

Sorex rohweri occurs primarily in the Coastal Range and western Cascades ecoregions of Oregon and Washington, the Puget Lowland and North Cascades ecoregions of Washington, and their equivalents in southwesternmost British Columbia (Pater et al. undated). Its recorded elevational distribution is from near sea level to at least 1585 m. Nagorsen & Panter (2009) reported the occurrence of S. rohweri and the similarly-sized S. cinereus at approximately 1400 m elevation in the area of Liumchen Mountain, along the southern border of British Columbia in the western Cascades Range. Otherwise, the Olympic Shrew is generally allopatric with S. c. cinereus to the east and with the larger S. c. streatori to the north. The distributional limits of S. rohweri may be confined by mutual competitive exclusion with S. cinereus, by ecological constraints, or by some combination of the two. Despite generally similar external appearance, S. cinereus and S. rohweri are not sister taxa and are only distantly related to each other (Rausch et al. 2007, Hope et al. 2014).

Our analysis of inventory records indicates that Sorex rohweri is generally an uncommon member of small mammal communities in western Oregon and Washington. Despite this, it can occur in Douglas Fir forests of a wide variety of ages, from clear-cuts and young closed-canopy forests to old-growth forests. Nagorsen & Panter (2009) noted that much of the suitable forest habitat for S. rohweri in the Frasier River Valley of British Columbia has been fragmented or lost as a result of agricultural and urban development. They suggested that determination of its conservation status in Canada should be a priority given the severe loss of habitat and the potential isolation of populations in British Columbia from more extensive populations in Washington. The Olympic Shrew faces similar threats of range fragmentation and habitat loss in Washington and Oregon as a result of forest clearance and urban development (Rausch et al. 2007). Logging changes the structure and diversity of forest small mammal communities in western Oregon and Washington. Most small forest species, including most species of shrews, appear to be able to sustain viable populations in most successional stages if forests are permitted to regenerate naturally (Corn et al. 1988, Corn & Bury 1991). This appears to be the case for the Olympic Shrew. Many of the same species can even be found in clear-cut areas where stands of old-growth remain nearby, although community composition is greatly altered and abundances of most species are much lower, suggesting that clear cuts may be sink habitats for those species (Corn et al. 1988, Corn & Bury 1991). Coastal lowland populations within the tsunami inundation zone associated with the Cascadia subduction zone face an additional, albeit rare, threat. They are vulnerable to sudden flooding and severe erosion following land subsidence and tsunamis associated with a predicted magnitude 8.9–9.0 earthquake (Cascadia Region Earthquake Workgroup 2005; Schulz 2015).

Sorex rohweri is well represented in systematic collections (Appendix; also see Rausch et al. 2007, Nagorsen & Panter 2009). Many additional specimens are probably present, but misidentified, in other institutions. The difficulty of accurately identifying small, cryptic taxa in the field—and even in well-maintained museum collections—highlights the scientific value and importance of preserving high-quality voucher specimens at an institution where they are readily available for re-study. Only then can results of field studies be verified or changed as our understanding of the natural world increases.

Toward that goal, we provide the following dichotomous key using qualitative traits to identify the smaller (GLS typically <18.3 mm; Fig. 4), morphologically confusing Sorex that inhabit western Oregon and western Washington, including also S. cinereus streatori from British Columbia. We exclude S. bairdi, S. bendirii, S. merriami, S. navigator, S. pacificus, S. preblei, and S. sonomae.

• 1.

  U3 much shorter and narrower than U4; palate relatively broad (M2B/CBL ≥ 24%); interorbital region relatively broad (PO/CBL ≥ 19%); unicuspid toothrow relatively short (UTR/CBL ≤ 16%); pigment on i1 extends from the tip of the tooth and includes pigment on the denticles . . . 2

• 1′.

  U3 only slightly shorter to slightly longer than U4; palate relatively narrow (M2B/CBL ≤ 24%); interorbital region relatively narrow (PO/CBL ≤ 21%); unicuspid toothrow relatively long (UTR/CBL ≥ 15%); pigment on denticles of i1 is separated from that on the tip of the tooth . . . 4

• 2.

  Cranium relatively short (GLS ≤ 17.6 mm; CBL ≤ 17.1 mm); 4 or fewer paired digital callosities on pedal digits III and IV; tail relatively short, typically <88% of HB . . . S. vagrans.

• 2′.

  Cranium relatively long (GLS ≥ 17.5 mm; CBL ≥ 16.8 mm); 5–7 paired digital callosities on pedal digits III and IV; tail relatively long, typically >83% of HB . . . 3

• 3.

  Tail strongly bicolored; medial edges of I1s curved and strongly divergent; medial tines of I1s at or proximal to the proximal edge of the pigment on the primary cusp; lingual ridges of U1–U3 unpigmented; typically a large postmandibular foramen on each dentary, and typically confluent with mandibular foramen; 6–7 paired (or confluent) digital callosities on pedal digits III and IV . . . S. trowbridgii.

• 3′.

  Tail not strongly bicolored; medial edges of I1s straight; medial tines of I1s distal to the proximal edge of pigment on the primary cusp; lingual ridges of U1–U3 strongly pigmented; typically no postmandibular foramen on either dentary; 5–6 paired digital callosities on pedal digits III and IV . . . S. monticola setosus.

• 4.

  Tail strongly bicolored; medial edges of I1s curved and strongly divergent; medial tines of I1s typically small, unpigmented or weakly pigmented, positioned at or proximal to the proximal edge of the pigment on the primary cusp; lingual ridges of U1–U3 unpigmented or weakly pigmented; typically a large postmandibular foramen on each dentary, and typically confluent with mandibular foramen; paired digital callosities on pedal digits I–V: 4-5-5-5-3 . . . S. rohweri.

• 4′.

  Tail not strongly bicolored; medial edges of I1s straight; medial tines of I1s large and positioned distal to the proximal edge of pigment on the primary cusp; lingual ridge of U1–U3 pigmented; typically no postmandibular foramen on either dentary, or occasionally a postmandibular foramen on one dentary; paired digital callosities on pedal digits I–V: 5-6-6-6-3 (Sorex cinereus) . . . 5

• 5.

  Dorsal pelage medium brown to dark brown; averages smaller (HB typically ≤62 mm); cranium shorter (GLS typically ≤17.0 mm; CBL typically ≤16.2 mm); dentition typically finer, paler; bones of the cranium typically thinner and the sagittal and lambdoidal sutures less likely to be fused . . . S. c. cinereus.

• 5′.

  Dorsal pelage dark dusky brown; averages larger (HB typically ≥ 56 mm); cranium longer (GLS typically ≥ 16.7 mm; CBL typically ≥ 16.1 mm); dentition typically more massive, more darkly pigmented; bones of the cranium typically thicker and the sagittal and lambdoidal sutures more likely to be fused . . . S. c. streatori.

We thank R. Bruce Bury for the large series of specimens from Oregon and Washington that provided the impetus for this study and Paul Stephen Corn for providing forest stand ages in Oregon. Michael D. Carleton, Alfred L. Gardner, Link E. Olson, and an anonymous reviewer provided substantive comments on earlier drafts that improved our manuscript. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. government.

All specimen catalog numbers are from the Mammal Division, U. S. National Museum of Natural History (USNM), Smithsonian Institution, Washington, D.C.

on lake at head of Bad River (209375); Bear Lake, site of Fort Connolly (202803); Bennett (99312–99314, 130024, 130025, 130028); S branch of Big Salmon River, near Canyon (209412); Cariboo Lake, Near Kamloops (67434); Field (69219, 69220); Fort Grahame (170723); Glacier (69132, 69137–69140, 69217, 290715); Hazelton (202796–202798, 202805–202809); Klappan River Valley, Tset-Ee-Yeh River (170716); McDame Creek, Zuartz Creek (206100); McDame Post, Dease River (206101, 206105, 206106, 206108, 206109, 206113); Moose Lake (174410, 174413, 174415–174417); Moose Pass (174407); Moose River (174494, 174495); N fork of Moose River (174404–174406); head of Parsnip River (209373); Penticton , Okanagan Lake (113417); Puntchesakut Lake (290705, 290706); Sicamous (69320); N end of Tacla Lake, site of Bulkley House (202801); Babine Trail, 12 mi W of Tacla Lake (202810); head of Tatletuey Lake, 12 mi W of Thudade Lake (202804); Telegraph Creek (170743); Vermilion Crossing, Kootenay Park (290703, 290704); Yellowhead Lake (174418, 174420, 174422).

Chelan Co.: head of Lake Chelan (30629). Lewis Co.: Mount Rainier, Paradise Creek (89579). Yakima Co.: Yakima Indian Reservation, Signal Peak (226858). Okanogan Co.: Tunk Mountain (234874); Tungsten Mine, west end of Bauerman Ridge (235204). Ferry Co.: 5 mi W of Curlew (235461, 235463). Stevens Co.: Loon Lake (236140, 236141). Pend Oreille Co.: 9 mi N of Metaline (236443).

Alaska: Bartlett Cove, Glacier Bay (506969); Chichagof Island, Tenakee Inlet (130154–130164, 130232–130241, 130939); Haines (203066); Haines, Lynn Canal (97525, 203065); Horn Cliff (110397); Juneau (235776–235778, 235786–235789, 235792, 235793, 235795, 235799); Juneau, Eagle Creek, Douglas Island (290270, 290271); Mount Roberts, Near Juneau (567520); Kupreanof Island (127076, 127077, 127084, 130840–130842); Loring (74913, 74919, 74921, 74925, 74926); Martin River, 55 mi S Cordova (298419, 298420); Mitkof Island (130835, 130836); Petersburg, Mitkof Island (126846, 126851, 126856, 126857); Muir Inlet, Muir Cabin Camp (506968); Orca (97990); Port Wells, Cloghill Point; 85 mi NW Cordova (298421); Quadra Lake (217414); Skagway (97524, 97526–97529); Skagway, Lynn Canal (97523); Sitka (59739, 73793–73800, 73802–73807, 200286, 200287); Valdez, Sawmill Bay (298422–298424); Wrangel (74704, 74710, 74714–74717, 74719, 100583–100585, 217417, 526794, 526795, 526798); Yakutat (73535, 73537, 73540, 73542, 73546, 73548).

(8): head of River Inlet (90190, 90430, 90431, 92804); Fort Simpson (90204, 90205, 90220); Howe Sound, Gibson's Landing (90259).

Clallam Co.: 3 mi S of Soleduck River (241819). Jefferson Co.: headwaters of Dosewallips River(241818). King County: Enumclaw (234281). Okanogan Co.: West Fork of Pasayton River (235119, 235120). Pierce Co.: Mount Rainier, Longmire (233603–233606); Mount Rainier, Meslers Ranch; 1 mi W of Rainier Park (233607, 233608, 233610); Puyallup (238293). Skagit Co.: Rockport (234512). Skamania Co.: 4 mi N, 7 mi W of Carson (558143, 558155, 558159, 558165); 5 mi N, 6.3 mi W of Carson (558166); 8.5 mi N, 1.5 mi W of Carson (558207). Whatcom Co.: Glacier (234290, 234875).

Benton Co.: 2.4 km S, 15.5 km W of Philomath (561633–561635, 562066–562069, 563325–563327); 4.4 km N, 7.6 km W of Alsea (561636, 563328); 8 km S, 2.8 km E of Alsea (561637, 563329). Douglas Co.: 15.7 km S, 5.6 km W of Elkton (561639). Coos Co.: 7.4 km N, 4.2 km W of Sitkum (561643, 563084). Lincoln Co.: 4 km S, 3.2 km E of Yachats (562048); 0.6 km N, 16.9 km E of Waldport (563324).

Clallam Co.: Neah Bay (88486). Grays Harbor Co.: Cedarville (231007). Jefferson Co.: head N fork Quinault River, 1219 m (241977). Skamania Co.: 5 mi N, 6.3 mi W of Carson (558710–558712); 7 mi N, 7.4 mi W of Carson (558184); 7 mi N, 8.3 mi W of Carson (558723); 8.4 mi N, 7.5 mi W of Carson (558731, 558734, 558735, 558737); 8.5 mi N, 1.5 mi W of Carson (558193, 558449, 558453, 558730); 11.5 mi N, 10.25 mi W of Carson (558788); 12 mi N, 8.5 mi W of Carson (558321, 558327, 558781, 558784, 558785). Thurston Co.: Tenino (89163). Whatcom Co.: Glacier, 274 m (234293); 36 mi E of Glacier, 1585 m (234787).

Clallam Co.: Neah Bay (88507, 88518, 88520, 88531, 88534, 88746). Grays Harbor Co.: Aberdeen (17441, A24403). King Co.: 3 mi E of Kirkland (171167). Mason Co.: Olympic Mountains, 3 mi SE of Mount Ellinor (67625, 67633, 67634). Pierce Co.: Nisqually Valley, Ashford (89587); Steilacoom (17401, 31059, 130242). Skagit Co.: Mount Vernon (88807, 88826–88830). Thurston Co.: Tenino (31459).

Chilliwhack (186691); Cranbrook, 899–914 m (159418, 159419, 234082–234086, 234088–234091); Glacier (290717, 290718, 290729, 290730); Langley (89473); Nanaimo (177723–177725); Okanagan (102686); Okanagan, Coldstream (290707–290710); W side Okanagan Landing, 914 m (94337); 9 mi W Okanagan Landing, 1524 m (94338); W of Okanagan Landing, 762 m (94339); Point Gray (264599–264601); Port Moody (67050–67053, 75611, 88842, 88844–88847, 88852, 88854, 88860–88862, 88865); Revelstoke National Park, 701 m (290722); Similkameen River, 3 mi E of River; 5 mi N of U.S. Boundary (91040); Sumas (62443, 62444, 62449, 62450, 62454, 62457, 62985–62987, 62994, 62995, 62997, 88709, 88714, 102681, 102684, 102685); Colwood, Vancouver Island (290713); Cowichau Lake, Vancouver Island (290712); Departure Bay, Vancouver Island (290711); Vancouver (250625); Vancouver Island, Goldstream (71913); South Vancouver (264598).

Baker Co.: Anthony (153551, 153552); Bourne (208280, 208281, 210118, 210119); Cornucopia (208286–208293, 223355, 223356); 2.5 mi NE Cornucopia; East Pine Creek (209206); Homestead, 549 m (212441); 3 mi NE Huntington, 640 m (212285); McEwen (208285); Rock Creek (209205); Anthony [Lake] (154240). Benton Co.: 4.4 km N, 7.6 km W Alsea (561636); 8 km S, 2.8 km E Alsea (561637); Corvallis (242524, 242525, 262284, 262285); 0.4 km S, 12.5 km W Philomath (563148, 563668–563683, 561615–561632, 562049–562063); 1.5 km S, 11.3 km W Philomath (562064); 2.4 km S, 15.5 km W Philomath (561633–561635, 562065–562069); 5 mi SW Philomath (231654, 231655, 231660, 231661, 231663). Clackamas Co.: Canby (262300); W Slope Mount Hood, near Timberline (79966); Oregon City (56900). Clatsop Co.: Seaside (142661). Coos Co.: 7.3 km N 4 km W Sitkum (63150–63156, 63687, 63688, 61641, 61642, 62071–62073); Empire (166938, 166940–166943). Crook Co.: Maury Mountains (78604–78606). Curry Co.: Gold Beach (68159–68161, 166972); Port Orford (68162). Deschutes Co.: La Pine (204451–204453, 204913–204917); Paulina Lake (79779, 79780, 204918–204922). Deschutes/Lane Co.: Three Sisters, North Base, 1524 m (204469); Three Sisters, North Slope, 1829 m (204446, 204449). Douglas Co.: Diamond Lake (79781–79785, 80093, 265384); Drain (69807–69810, 166934); 0.8 km N, 2.2 km W Drain (561638, 563684, 563685); 10 km N, 10.5 km W Drain (560325); 4 km W Drain, 15 m from Elk Cr. (560094, 560326); 15.7 km S, 5.6 km W Elkton (561639); 9.7 km S, 16.2 km W Elkton (563686); Looking Glass (214072); Scottsburg (166935–166937); 2 km N, 25 km W Sutherlin (562070); 2.2 km N, 26.2 km W Sutherlin (563149); 3.6 km N, 11.9 km W Sutherlin (561640); 4.3 km N 12.1 km W Sutherlin (560327). Grant Co.: Austin (208279); 15 mi NW Austin, 1646 m (273108); Strawberry Mountains (208278, 209621–209631). Harney Co.: Burns (222236–222239); Diamond (215972). Hood River Co.: N slope Mount Hood, 853 m (231664, 231665); 2 mi W Parkdale, 457 m (231671). Jackson Co.: Ashland, 602 m (203608); Ashland, W slope Grizzly Peak, 1067 m (203609). Jefferson Co.: 20 mi W Warmsprings, Mill Creek (207016, 207017). Klamath Co.: Anna Creek, Mount Mazama, 1829 m (80421); Chemult Highway between Chemult and Sand Creek (275131); Crater Lake (79786, 79787, 79944–79949, 79951–79953, 80090–80092, 80418, 80419); Crater Lake near Glacier Peak (79956); Crater Lake National Park, Red Blanket Creek (274875); Crater Lake National Park, South Entrance (274876); Fort Klamath (15653, 15746–15749, 59672, 59673, 62228, 62229, 80422–80428, 80430–80436, 81108–81114, 186869, 186870, 303330, 303331); Klamath Falls (209474, 222877–222881, 526801, 526802); Upper Klamath Marsh (205261, 205262); Yamsay Mountains, Yamsay River, 1463 m (205263). Lake Co.: 10 mi SE Lakeview, 2012 m, (273110); Plush Lake (79289); West Silver Creek, 10 mi SW Silver Lake, 1417 m (205254–205256); West Silver Creek, Yamsay Mountains, 2134 m (205257–205260). Lane Co.: 5 km N, 7.6 km E Blue River, 850 m (557373); 6.8 km N, 7.6 km E Blue River, 520 m (557953); Eugene (204438, 204468); Florence (69446); Mapleton (205271); McKenzie Bridge (204442); 1.6 km S, 5.2 km W McKenzie Bridge, 460 m (557374–557377, 557955, 557956); 10 mi SE McKenzie Bridge, Lost Creek R. S. (204443, 204444); 2.6 km S McKenzie Bridge, 780 m (557958, 557959); 2.6 km S, 1.4 km E McKenzie Bridge, 670 m (557957); 2.6 km S McKenzie Bridge, 780 m (557378, 557379); 3.1 km N, 3 km W McKenzie Bridge, 1160 m (557369, 557370); 3.3 km N, 2.4 km W McKenzie Bridge, 1170 m (560446, 557371, 557372); 3.8 km S, 14.8 km E McKenzie Bridge, 1070 m (557392–557402, 557404–557430, 557965–557982); 4.4 km S, 14.3 km E McKenzie Bridge, 1050 m (557382–557391, 557961–557964); 5.4 km S, 8.75 km E McKenzie Bridge, 610 m (557380, 557960); 6 km N, 0.6 km E McKenzie Bridge, 790 m (557365); 6 km N, 7.5 km E McKenzie Bridge, 690 m (557368); 6.4 km N, 0.6 km E McKenzie Bridge, 760 m (557366, 557952); Vida (204439). Lincoln Co.: 4 km S, 3.2 km E Yachats (563667). Linn Co.: 10.25 km N, 0.8 km W McKenzie Bridge, 1040 m (557360–557364, 557950, 557951); 13.2 km N, 0.8 km E McKenzie Bridge, 1020 m (556808, 557350, 557351, 557352, 557944, 557945); 9.2 km N, 1.2 km W McKenzie Bridge, 810 m (557353–557359, 557946, 557948, 557949); Permilia Lake, west base Mount Jefferson (91060); Shelburn (91473). Malheur Co.: 8 mi W Jordan Valley (228078). Marion Co.: Salem (57112–57118, 186871, 255443, 255444). Multnomah Co.: Portland (142010, 142069–142072, 142681–142687, 204464–204467, 205266–205269, 250152–250155, 262281); Portland, Reed Canyon (291924); Portland, Switzler Lake (161487–161491); Reed College, orchard (291923). Tillamook Co.: Netarts (249299, 262302, 262304, 264883); Tillamook Alderbrook Golf Course (262286). Umatilla Co.: Meacham (209200–209202); 10 mi W Meacham (78147, 78149). Union Co.: Elgin (78144, 78145); Hot Lake (209203, 209204); Kamela (78146, 78148). Wallowa Co.: 25 mi N Enterprise at Sled Springs, 1402 m (231712–231715); Wallowa Lake, 1219 m (90720–90722, 96020, 96021, 231696, 231697, 231699–231710). Washington Co.: Beaverton (186868); Forest Grove (291925); Gales Creek (262301); Hillsboro (262282); Farmington, 5 mi SE Hillsboro (244328). Yamhill Co.: Sheridan (69781, 69782). County uncertain: Cascade Mountains (1259); Des Chutes River Farewell Bend (79962); Des Chutes E Fork (79960, 79961); Warner Creek (205264); Warner Mountains (79957–79959).

Asotin Co.: Bly, 305 m (232356); Rogersburg (232355). Chelan Co.: Entiat River, 20 mi from Mouth, 512 m (229885, 230182). Clallam Co.: Blyn (242388); Cape Flattery (1261); Forks (241811); La Push (89151, 89152, 89156, 89158, 215166); Neah Bay (88475–88485, 88488–88496, 88517, 88524–88526, 88528, 88533, 88743, 88744, 88747); Olympic Mountains, Soleduc [Sol Duc] River, 1219 m (90486); Sequim (235625–235628); Suez, 10 mi S Neah Bay (88471, 88473). Columbia Co.: Starbuck, 197m (234167–234169). Columbia/Garfield Cos.: 21 mi SE Dayton, Blue Mountains (232040). Grant Co.: Moses Lake, Tichacek Orchards, 305 m (242402); 10 mi S Moses Lake (273115, 273116). Grays Harbor Co.: Aberdeen (17417, 17418, 17423–17428, 17430–17437, 17439, 17440, 18788–18790); Grayland (273119–273123); Lake Quinault (89637, 89642, 89644, 89646, 89652, 89653, 89659, 242391, 242393); Oakville (231011, 291921, 291922); Westport (235629–235633). Island Co.: Whidbey Island, Nas, Taxiway C (568056); Whidby Island, 3 mi N Clinton (140266); Whidby Island, Greenbank (235623, 235624); north end of Whidby Island (231646–231649); Whidby Island, San De Fuca (231650–231652). Jefferson Co.: Duckabush (231634, 231635, 231637–231640); Port Townsend (5069, 5100, 16428). King Co.: Cottage Lake (273104); south end of Lake Washington, near Renton (90754, 90755); North Bend, 183 m (29875, 229883); Redmond, Sammamish River (171174, 171175); Seattle (273126, 273127); Vashon Island (231009, 231010); Enumclaw (234280); Kent (90482); 0.5 mi E Kirkland (171162, 171163); 1 mi E Kirkland (171173); 3 mi E Kirkland (171164–171166, 171170); Kirkland, Lake Washington (171176, 171177, 176667–176677). Kittitas Co.: 2 mi S Blewett Pass, 914 m (242398, 242399, 242401); Easton (29545, 29551, 583273); Ellensburg, 457 m (235638, 235639). Klickitat Co.: Goldendale (89426); Trout Lake, 591 m (230205–230207, 230209, 230210). Lewis Co.: 7 mi E Centralia (506677); Chehalis (230187, 230214, 230216); 8 mi W Chehalis (230215); Toledo (231013); Mount Rainier, Bear Prairie (233087). Lincoln Co.: Sylvan Lake, 6 mi E Odessa (229876–229879); Sprague Lake, Sprague P.O. (273125). Mason Co.: Hoodsport (231630); Lake Cushman (66179–66181, 66183–66185, 66187–66191); Shelton (230201, 230202); North Fork of Skokomish River, Olympic Mountains (67631). Okanogan Co.: E end of Bauerman Ridge, near head of Haig Creek, 1981 m (235206); Conconully (91041, 91042, 228052); Hidden Lakes, 1250 m (235121); Loomis, 396 m (235155); Oroville (264884); Oroville, Osoyoos Lake, 305 m (235388); near mouth of east fork Pasayton River, 1189 m (235067); Sheep Mountain, Park Mountain, 1981 m (235122, 235123, 235295); Twisp, 488 m (234514). Pacific Co.: Ilwaco (230193); 2 mi N Ilwaco (567521); Shoalwater [= Willapa] Bay (1675); Shoalwater Bay (630); Tokeland, Shoalwater Bay (230186, 230188, 230192, 230217). Pend Oreille Co.: 9 mi N Metaline, 792 m (236441, 236442); Newport (273091); Sullivan Lake, 914 m (236440). Pierce Co.: Fort Steilacoom (362, 1677); Pullman (88447, 140267); Puyallup (210639, 210640, 223839–223842, 238292, 238294, 245765–245768, 273128, 273129, 288600, 288601, 526804, 526806–526813); Roy (90480, 90481, 90483); Steilacoom (A24402, 31060); 6 mi S Tacoma (231012, 231015). San Juan Co.: Lopez Island (266520); Orcas Island, East Sound (234294–234296); Richardson, Lopez Island (234297–234302); San Juan Island, Friday Harbor (130243, 266514). Skagit Co.: Avon (30956–30958); Cypress Island (266516–266519); Hamilton (17405); Laconner (231641–231645); Mount Vernon (17403, 76489); Sauk (30955); Strawberry Bay, Cypress Island (266515). Skamania Co.: Carson (230198); 10 mi N, 1.3 mi W Carson (558245, 558248, 558249, 558251, 558516, 558739–558742, 558744, 558745); 10.5 mi N, 4.5 mi W Carson (558257, 558260, 558261, 558265, 558270, 558746–558748, 558750–558759, 558761–558776); 10.7 mi N, 8.7 mi W Carson (558278, 558777–558779); 11.5 mi N, 10.25 mi W Carson (558786, 558787); 12 mi N, 8.5 mi W Carson (558782); 4 mi N, 7 mi W Carson (558707–558709); 7 mi N, 7.4 mi W Carson (558183, 558713–558715); 7 mi N, 8.3 mi W Carson (558431, 558432, 558716–558722; 558724–558727; 558187–558189); 8.4 mi N, 7.5 mi W Carson (558732, 558733, 558736, 558738); 8.5 mi N, 1.5 mi W Carson (558465, 558728, 558729); Stevenson (230181); 45 mi SE Toledo, Cascade Mountains (90718, 90719); 15 mi NW White Salmon, Berry Creek (180200); 5 mi N Willard (506675, 506676). Snohomish Co.: Granville (91710); Oso, 168 m (234508, 234509). Spokane Co.: Marshall (24096–24100, 24285, 24286). Stevens Co.: 13 mi SE Colville (273113); 15 mi E Colville (273114); Marcus (90757). Thurston Co.: Nisqually (231003); Nisqually Flats (231004, 231014); Olympia (288591–288593, 288595, 288596); 4 mi S Olympia (231005); 5.5 mi S Olympia (288590); Tenino (89160, 89161, 89164, 89165); 2 mi SW Tenino (273133). Wahkiakum Co.: Cathlamet (230189); 1 mi S Cathlamet, Puget Island (230225, 230227). Whatcom Co.: Barron, 1524 m (234881); Beaver Creek, McMillan Ranch, 518 m (234788, 234789, 234880); Blaine (234286); Glacier, 274 m (234291); Lake Whatcom (234287); Simiahmoo, Puget Sound (5037, 5039). Whitman Co.: 5 mi NE Wawawai (74633). Yakima Co.: Mount Adams, Bird Lake (273061–273063); Yakima Indian Reservation, Signal Peak, 1219 m (226857); Wiley City, 10 mi W Yakima, 610 m (235634–235637). County unknown: Puget Sound (4809, 5027, 5041, 5066, 5094, 5096).

Puget Sound (?) (8372–8376).