Deglaciation of the Sierra Nevada opened it up for reoccupation from the east by _Sorex vagrans_ of the Great Basin. In response to the montane environment the subspecies _obscuroides_, resembling the subspecies _obscurus_ of the Rockies, developed.

Desiccation of the intermontane parts of New Mexico, Arizona, and Chihuahua, left "marooned" populations of _Sorex vagrans_ on suitable mountain ranges. In this way _Sorex vagrans orizabae_ may have been isolated in southern Mexico. The isolated populations of Arizona and New Mexico differentiated _in situ_ into the subspecies _monticola_ and _neomexica.n.u.s_.

Western Canada and Alaska were populated by shrews which originated in the habitable parts of the Rocky Mountains and Colorado Plateau during Wisconsinan time (as opposed to shrews originating, as subspecies, in the Great Basin or on the Pacific Coast). These shrews differentiated into the currently recognized subspecies of the west coast and coastal islands of British Columbia and Alaska in response to the different environments in these places, many of which were isolated; the subspecies _isolatus_, _mixtus_, _setosus_, _longicauda_, _ela.s.sodon_, _prevostensis_, _malitiosus_, and _alaskensis_ are thought to have originated in this fashion after the areas now occupied by them were freed of Wisconsinan ice.

This group of shrews from the Rocky Mountains probably came into contact with the Pacific coastal segment of the species somewhere in northwestern Oregon. The clinal decrease in size from _S. v.

pacificus_ to _S. v. setosus_ seems steepest in this area. Upon the establishment of this contact reproductive continuity was resumed, probably because the temporal separation of the two stocks involved was not so great as, say, that between _S. v. vagrans_ and _S. v.

pacificus_, and in addition the morphological differentiation was not so great.

On the eastern side of the Rockies the montane stock moved northeastward, occupying suitable territory opened up by the dissolution of the Laurentide ice sheet. Still later changes in the character of the northern plains owing to desiccation divided the range of the species and isolated _S. v. soperi_ in Manitoba and central Saskatchewan and a population of _S. v. obscurus_, in the Cypress Hills. A number of semi-isolated stocks in central Montana became differentiated as a recognizable subspecies there.

A number of other boreal mammals have geographic ranges which resemble that of _Sorex vagrans_, except that the geographic ranges of subspecies do not overlap. Because of the general similarities of these geographic ranges, it is pertinent to examine the reasons suggested by students to account for the present geographic distributions of some of these other boreal species.

The red squirrel genus, _Tamiasciurus_, has a Rocky Mountain (and northern coniferous forest) species, _T. hudsonicus_, that occurs all along the Rocky Mountain chain and northward into Alaska. In the Cascade Mountains of Washington and British Columbia this species meets the range of a well marked western species, _T. douglasii_, with no evidence of intergradation. Dalquest (1948:86) attributes the divergence of the two species to separation in a glacial age but feels that the degree of difference between the two is too great to have all taken place during the Wisconsinan. Perhaps he has overemphasized the importance of the differences between the two, but, be that as it may, it seems that the two kinds differentiated during a glacial age when they were isolated, perhaps by ice on the Cascades into a coastal population and an inland population. One difference between the distribution of the red squirrels and vagrant shrew is that the squirrel of the Sierra Nevada is the species of the Pacific Coast, whereas the vagrant shrew of the Sierra Nevada was derived from the Great Basin population, which in turn was derived from the Rocky Mountain kind. Red squirrels do not occur on any of the boreal montane "islands" of Nevada. During the pluvial periods when hydrosere-loving shrews populated the Great Basin, that region may have been a treeless gra.s.sland. Vagrant shrews, then as now, probably depended on hydrosere communities, while red squirrels required trees. Therefore the shrews were able to traverse the Great Basin, while the Sierran red squirrels were of necessity derived from the coastal population.

The ecological requirements of jumping mice, genus _Zapus_, and the subspecies of _Sorex vagrans_ that dwell in hydroseres are essentially similar. The species _Zapus princeps_ lives in the Rocky Mountains, the Great Basin, the Sierra Nevada, and north to Yukon (Krutzsch, 1954:395). Its geographic range is similar to that of the montane and basin segments of _S. vagrans_. The species _Z. trinotatus_ occurs along the Pacific coast and in the Cascades north to southwestern British Columbia. Its distribution thus coincides in general with that of the large red coastal subspecies of _S. vagrans_. Krutzsch (1954:368-369) thought that these two kinds of jumping mice were first separated by the formation of the Cascade Mountains and the Sierra Nevada and finally by Pleistocene glaciation. The Sierran jumping mouse (_Zapus princeps_), as is the Sierran vagrant shrew, is more closely related to the jumping mouse of the Great Basin and of the Rocky Mountains than it is to the jumping mouse (_Z. trinotatus_) of the Pacific Coast, just as the Sierran vagrant shrew is related to the shrew of the Great Basin and Rocky Mountains. The jumping mouse also is limited in its distribution by hydrosere communities, not by forests.

In western North America there are two species of water or marsh shrews: _Sorex pal.u.s.tris_ and _S. bendiri_. They have been placed in separate subgenera, but, as pointed out beyond, are closely related and here are placed in the same subgenus. The species _pal.u.s.tris_ is found throughout the Rocky Mountains, north into Alaska, across the Great Basin into the Sierra Nevada, and west to the Pacific coast in Washington. The species _bendiri_ is found from northwestern California north along the Pacific coast to southwestern British Columbia and east to the Cascades. Where the ranges of the two species overlap in western Washington they do not interbreed so far as is known, and are somewhat different in their ecology, _bendiri_ being a lowland, and _pal.u.s.tris_ being a montane, species. The two species probably were separated in a glacial period as seems to have been the case with the wandering shrews. Also, the water shrew of the Sierra Nevada is derived from that of the Great Basin and Rocky Mountains.

_Sorex pal.u.s.tris_ is tied closely in its distribution to hydrosere communities and is not dependent upon the presence of forests.

Red-backed mice, genus _Clethrionomys_, occur throughout the Rocky Mountains and west to the Cascades in Washington as the species _C.

gapperi_. The species _C. californicus_ is found along the Pacific Coast from California north to the Olympic Peninsula. Where the ranges of the two species meet in Washington they seem not to intergrade. In some glacial interval these two species may have evolved in the same manner as has been described for the species of _Zapus_ and those of _Tamiasciurus_. No _Clethrionomys_ are found in the Sierra Nevada, nor are red-backed mice found in the boreal islands of the Great Basin. It is not known why _Clethrionomys californicus_ does not occur in the Sierra Nevada. Some boreal birds have distributional patterns similar to those of the mammalian examples cited above. One kind of sapsucker, _Sphyrapicus varius nuchalis_, occurs in the Rocky Mountains north into British Columbia and west to the Cascades and Sierra Nevada. A related kind, _S. varius ruber_, occurs along the Pacific Coast from California north into British Columbia. Recently Howell (1952) has shown that some intergradation takes place between _ruber_ and _nuchalis_ in Washington and British Columbia, although they do not intergrade freely. Previously the two kinds were thought not to intergrade and were regarded as two species. The two kinds intergrade also in northeastern California, although in that state _S. v.

daggeti_, rather than _S. v. ruber_, is involved in the intergradation. Howell considered the two kinds to be conspecific with one another as well as with the eastern _S. varius_. He attributed a measure of the distinctness of _nuchalis_ and _ruber_ to their separation during a glacial period, but felt that the separation was much older than Wisconsinan. Whatever the time of separation, the pattern seems clear: _nuchalis_ and _ruber_ (as well as _varius_) were separated into montane, coastal, and eastern segments respectively, probably by glaciation (it seems to me in the Pleistocene), and have since re-established contact with one another.

The grouse genus _Dendrogapus_ is divided into a Great Basin species, _D. obscurus_, which extends northward into British Columbia, and a Rocky Mountain species, _D. fuliginosus_, that is found in the Sierra Nevada and northward along the coast and Cascades into British Columbia. Although the two kinds have at times been considered conspecific, they differ in voice, hooting mechanism, and characters of the downy young, and so far no actual intergradation between the two has been shown (Grinnell and Miller, 1944:113). These grouse thus seem to offer additional evidence for a Pleistocene, possibly Wisconsinan, separation of the boreal fauna into a Rocky Mountain and a Pacific coastal segment.

A notable sidelight on these data is the frequency with which species in the Sierra Nevada have their closest relatives in the Rocky Mountains, rather than in the geographically nearer Cascades or coastal areas. This similarity in fauna of the Sierra Nevada and the Rockies was noted long ago by Merriam (1899:86).

RELATIONSHIPS WITH OTHER SPECIES

During the Sangamonian interval, isolated segments of the once widespread ancestral _Sorex vagrans_ quite possibly persisted in such places as the Sierra Nevada, coastal southern California, the mountains of Arizona, New Mexico, and southern Mexico, and in the Black Hills (see fig. 6). One might expect that by Wisconsinan time these populations would have become reproductively isolated from their parent stock. They would therefore have remained specifically distinct when Wisconsinan _Sorex vagrans_, reoccupied these outlying areas, and may still be found isolated in places peripheral to the range of the ancestral species.

[Ill.u.s.tration: FIG. 10. Probable distribution of _S. veraepacis_, _S. longirostris_, and the _S. ornatus_ group (stipple) and of their Wisconsinan ancestors (lines). Heavy line indicates limits of Wisconsinan glaciation.]

In fact, we do find species closely related to _Sorex vagrans_ in just such places today (fig. 10). Probably _Sorex ornatus_, including members of the _ornatus_ group such as _S. trigonirostris_, _S.

sinuosus_, _S. willeti_, _S. tenellus_, and _S. na.n.u.s_, and also _S.

veraepacis_, arose by separation from the ancestral _vagrans_ stock in Sangamonian time. Probably the eastern _S. longirostris_ arose in a like manner. The ancestor of _S. ornatus_ may have been isolated in southwestern California during Sangamonian time, spread north and south during the Wisconsinan age, and afterward given rise to _S.

trigonirostris_ and the modern _S. ornatus_ complex of California and Baja California. In at least one place reproductive isolation between _ornatus_ and the invading _S. vagrans_ has broken down (Rudd, 1953); the place is a salt marsh along San Pablo Bay, where a hybrid population between _S. vagrans_ and _S. sinuosus_, an _ornatus_ derivative, has formed. _Sorex tenellus_ may have been isolated in the Sierra Nevada in the Sangamonian interval, moved into the valleys east of the mountains during the Wisconsinan age, and become restricted to its present range since the retreat of the last ice.

_Sorex na.n.u.s_ may have occurred in the Black Hills and isolated mountains of Arizona and New Mexico during the Sangamonian interval and remained in these general areas during the Wisconsinan age. Its present range is peripheral to the main body of the Rockies and the Colorado Plateau.

The eastern species _Sorex longirostris_ has many similarities with shrews of the _ornatus-vagrans_ stock. _S. l. longirostris_ is close in many ways to _S. na.n.u.s_. Indeed, the differences between the species _S. na.n.u.s_, _S. ornatus_, and _S. longirostris_ seem to me to be of the same magnitude and indicate a similar period of differentiation from a common ancestor. The ancestor of _S.

longirostris_ may have gained access to the eastern United States in the Illinoian Age _via_ the northern Great Plains south of the glacial boundary (fig. 7). The ancestor of _Sorex veraepacis_ of southern Mexico probably reached that area in Illinoian time as part of the ancestral _vagrans_ stock and probably attained its differentiation during the Sangamonian interval.

All the kinds of shrews so far discussed, including the _S. vagrans_ complex, might thus be thought of as having had a common ancestor in the Illinoian Age. This entire group of shrews has the third unicuspid smaller than the fourth, a pigmented ridge from the apex to the cingulum of each upper unicuspid, and, in most individuals, lacks a post-mandibular foramen in the lower jaw (Findley, 1953:636-637). The pigment is not always prominent in _S. longirostris_.

Two other species of North American shrews,_ Sorex pal.u.s.tris_, the water shrew, and _Sorex bendiri_, the marsh shrew, show these three characters to a greater or lesser degree, and it seems that these two species and the _vagrans-ornatus-veraepacis_ group had a common ancestor, probably before Illinoian time for reasons stated beyond. I judge, however, that far from being subgenerically distinct as they have been considered to be, _S. pal.u.s.tris_ and _S. bendiri_ are actually closely related species of the same subgenus and may have differentiated from one another because of separation into eastern (_pal.u.s.tris_) and western (_bendiri_) segments in the Sangamonian interval, much as has been postulated concerning the eastern and western stocks of _Sorex vagrans_. Indeed, Jackson (1928:192) has noted that in the Pacific northwest the characters of the two kinds approach one another and become differences of degree only.

The widespread species _Sorex cinereus_ resembles all the foregoing species in the ridges on the unicuspid teeth and in the lack of a post-mandibular foramen, but differs from those other species in having the third upper unicuspid larger than the fourth. The subspecies _S. cinereus ohionensis_, however, often has the sizes of these teeth reversed. With _S. cinereus_ I include _S. preblei_ (eastern Oregon) and _S. lyelli_ (Sierra Nevada), both obviously closely related to _cinereus_ as Jackson (1928:37) recognized when he included them in the _cinereus_ group. _Sorex milleri_ (Coahuila and central western Nuevo Leon) seems to me to resemble _S. cinereus_ more than it does other species of North American _Sorex_, and I judge that it also belongs to the _cinereus_ group. _Sorex cinereus_ and its close relatives seem more closely related to the species which have thus far been discussed than they do to such other North American species as _S. arcticus_, _S. fumeus_, _S. trowbridgii_, _S. merriami_, and the members of the _S. saussurei_ group; most of these five species last mentioned possess a post-mandibular foramen, lack pigmented unicuspid ridges, and have the third unicuspid larger than the fourth. Because of the morphological resemblances mentioned above, it seems likely to me that _S. cinereus_ and the _vagrans-ornatus-veraepacis-pal.u.s.tris_ complex had a common ancestor in early Pleistocene time. _Sorex cinereus_ has recently been considered to be conspecific with the Old World_ S. caecutiens_ Laxmann (Van den Brink, 1953) which name, being the older, would apply to the circ.u.mpolar species.

Hibbard (1944:719) recovered _S. cinereus_ and a species of _Neosorex_ (a name formerly applied to the water shrew) from the Pleistocene (late Kansan) Cudahy Fauna. This indicates that the ancestors of the modern _S. cinereus_ and of the water shrew had diverged from one another before that time. Brown (1908:172) recorded _S. cinereus_ and _S. obscurus_ from the Conard Fissure in Arkansas. These materials were deposited probably at a later time than was the Cudahy Fauna. The _S. obscurus_ from Conard Fissure probably represents the ancestral _S. vagrans_ stock which I think reached eastern United States in Illinoian time and gave rise to _S. longirostris_. The Conard Fissure material was deposited at a time (Illinoian?) when northern faunas extended farther south than they do today.

All of the species mentioned as having structural characters in common with _S. vagrans_ seem to have arisen from a common ancestor which had already differentiated from the ancestor of such species as _S.

arcticus_, _S. saussurei_, and others. Consequently all are here included in a single subgenus. The oldest generic name applied to a shrew of this group, other than the name _Sorex_, is _Otisorex_ DeKay, 1842, type species _Otisorex platyrhinus_ DeKay, a synonym of _Sorex cinereus_. The subgenus can be characterized as follows.

Subgenus =Otisorex= DeKay

1842. _Otisorex_ DeKay, Zoology of New York, pt. 1, Mammalia, p. 22, and pl. 5, fig. 1. Type, _Otisorex platyrhinus_ DeKay (= _Sorex cinereus_ Kerr).

Third unicuspid usually smaller than fourth; upper unicuspids usually with pigmented ridge extending from apices medially to cingula, uninterrupted by antero-posterior groove; post-mandibular foramen usually absent. Includes the species _S. cinereus, S. longirostris, S.

vagrans, S. ornatus, S. tenellus, S. trigonirostris, S. na.n.u.s, S.

juncensis, S. willeti, S. sinuosus, S. veraepacis, S. pal.u.s.tris, S.

bendiri, S. alaska.n.u.s_, and _S. pribilofensis_.

[Ill.u.s.tration: FIGS. 11-14. Characters of the subgenera _Sorex_ and _Otisorex_.

FIG. 11. Medial view of right ramus of _Sorex (Otisorex) vagrans_. 14.

FIG. 12. Medial view of right ramus of _Sorex (Sorex) arcticus_. 14.

FIG. 13. Anterior view of left second upper unicuspid of _Sorex (Otisorex) vagrans_. 45.

FIG. 14. Anterior view of left second upper unicuspid of _Sorex (Sorex) arcticus_. 45.]

Other species of _Sorex_ now occurring in North America differ from _Otisorex_ in having the 3rd unicuspid usually larger than 4th, in lacking a pigmented ridge from the apices to the cingula of the upper unicuspids, and in usually possessing a well-developed post-mandibular foramen. Exceptions to the last mentioned character are _S. fumeus_ and _S. dispar_. The subgenus _Sorex_ in North America should include only the following species: _S. jacksoni_, _S. tundrensis_, _S.

arcticus_, _S. gaspensis_, _S. dispar_, _S. fumeus_, _S. trowbridgii_, _S. merriami_, and all the members of the Mexican _S. saussurei_ group.

The subgenera _Otisorex_ and _Sorex_ probably separated in early Pleistocene or late Pliocene. _Sorex_ is unknown in North America earlier than the late Pliocene (Simpson, 1945:51).

In the genus _Microsorex_ the characters of the subgenus _Otisorex_ are carried to an extreme; the unicuspid ridges are prominent and end in distinct cusplets, and the 3rd unicuspid is not merely smaller than the 4th, but is reduced almost to the vanishing point. In addition, the post-mandibular foramen is absent. Although it is closer structurally to _Otisorex_ than to _Sorex_, the recognition of _Microsorex_ as a distinct genus seems warranted.

Figure 15 is intended to represent graphically some of the relationships discussed above. It must be re-emphasized that much of it is purely speculative, especially as regards actual time when various separations took place. It will be noted that I have indicated most separations as having taken place in interglacial ages. They are generally regarded as periods of warmth and aridity and, therefore, probably are times of segmentation of the ranges of boreal mammals and hence times exceptionally favorable to the process of speciation.

Glacial ages, characterized by extensive and continuous areas of boreal habitat, probably were times of relatively unrestricted gene flow between many populations of boreal mammals and hence not favorable to rapid speciation.

=Sorex vagrans=

Wandering Shrew

The size of the wandering shrew varies from small in the subspecies _monticola_ and _vagrans_ to large in the subspecies _pacificus_. The tail makes up from a little more than a third to almost half of the total length. The color pattern ranges from tricolored through bicolored to almost monocolored. Color ranges from reddish (Sayal or Snuff Brown) to grayish in summer pelage and from black to light gray in winter. Diagnostic dental characters include: 3rd upper unicuspid smaller than 4th, and unicuspids, except 5th, with a pigmented ridge extending from near apex of each tooth medially to cingulum and sometimes ending as internal cusplet. _S. vagrans_ differs from members of the _ornatus_ group in less flattened skull, and in more ventrally situated foramen magnum that encroaches more on the basioccipital and less on the supraoccipital. The wandering shrew differs from _S. trowbridgii_ and _S. saussurei_ in the dental characters mentioned above. These dental characters also serve to distinguish _S. vagrans_ readily from _S. cinereus_, _S. merriami_, and _S. arcticus_ which may occur with _vagrans_. The large marsh shrew and water shrew, _S. pal.u.s.tris_ and _S. bendiri_, can be distinguished at a glance from _S. vagrans_ by larger size and darker color.

[Ill.u.s.tration: FIG. 15. Diagrammatic representation of the probable phylogeny of _Sorex vagrans_ and its near relatives.]

In the following treatment of the 29 subspecies of _Sorex vagrans_, the subspecies are arranged in geographic sequence, beginning with the southernmost large subspecies on the California coast and proceeding clockwise, north, east, south, and then west back to the starting point.