Chapter I: Part 1

==================================================================
UNIVERSITY OF KANSAS PUBLICATIONS
MUSEUM OF NATURAL HISTORY

Volume 9, No. 1, pp. 1-68, figures 1-18
-------------------- December 10, 1955 ---------------------

Speciation of the Wandering Shrew

BY
JAMES S. FINDLEY

UNIVERSITY OF KANSAS
LAWRENCE
1955

UNIVERSITY OF KANSAS PUBLICATIONS, MUSEUM OF NATURAL HISTORY

Editors: E. Raymond Hall, Chairman, A. Byron Leonard,
Robert W. Wilson

Volume 9, No. 1, pp. 1-68, figures 1-18
Published December 10, 1955

UNIVERSITY OF KANSAS
Lawrence, Kansas

PRINTED BY
FERD VOILAND, JR., STATE PRINTER
TOPEKA, KANSAS
1955

25-7903

Speciation of the Wandering Shrew

BY

JAMES S. FINDLEY

CONTENTS

PAGE

INTRODUCTION 4

MATERIALS METHODS AND ACKNOWLEDGMENTS 4

NON-GEOGRAPHIC VARIATION 7

CHARACTERS OF TAXONOMIC WORTH 8

PELAGE CHANGE 9

GEOGRAPHIC DISTRIBUTION AND VARIATION 9
Pacific Coastal Section 9
Inland Montane Section 11
Great Basin and Columbia Plateau Section 12
Summary of Geographic Variation 13

ORIGIN OF THE _Sorex vagrans_ RASSENKREIS 16

RELATIONSHIPS WITH OTHER SPECIES 26

CONCLUSIONS 60

TABLE OF MEASUREMENTS 62

LITERATURE CITED 66

FIGURES

FIGS. 1-2.--CRANIAL MEASUREMENTS 5

FIG. 3.--GRAPH ILLUSTRATING WEAR OF TEETH 8

FIG. 4.--GRAPH ILLUSTRATING HETEROGONIC GROWTH OF ROSTRUM 10

FIG. 5.--PRESENT GEOGRAPHIC DISTRIBUTION OF _Sorex vagrans_ 15

FIG. 6.--SKULLS OF _Sorex vagrans_ 17

FIGS. 7-10.--PAST GEOGRAPHIC DISTRIBUTION OF SHREWS 19-20-22-27

FIGS. 11, 12.--MEDIAL VIEW OF LOWER JAWS OF TWO SHREWS 30

FIGS. 13, 14.--SECOND UNICUSPID TEETH OF SHREWS 30

FIG. 15.--DIAGRAM OF PROBABLE PHYLOGENY OF SHREWS 32

FIGS. 16-18.--GEOGRAPHIC DISTRIBUTION OF SUBSPECIES 33-40-53

INTRODUCTION

The purpose of this report is to make clear the biological relationships between the shrews of the _Sorex vagrans-obscurus_ "species group." This group as defined by H. H. T. Jackson (1928:101) included the species _Sorex vagrans_, _S. obscurus_, _S. pacificus_, _S. yaquinae_, and _S. durangae_. The last mentioned species has been shown (Findley, 1955:617) to belong to another species group. _Sorex milleri_, also assigned to this group by Jackson (1947:131), seems to have its affinities with the _cinereus_ group as will be explained beyond. The position of the _vagrans_ group in relationship to other members of the genus will be discussed.

Of this group, the species that was named first was _Sorex vagrans_ Baird, 1858. Subsequently many other names were based on members of the group and these names were excellently organized by Jackson in his 1928 revision of the genus. Subsequent students of western mammals, nevertheless, have been puzzled by such problems as the relationship of (1) _Sorex vagrans monticola_ to _Sorex obscurus obscurus_ in the Rocky Mountains, (2) _Sorex pacificus_, _S. yaquinae_, and _S. obscurus_ to one another on the Pacific Coast, and (3) _S. o. obscurus_ to _S. v. amoenus_ in California. Few studies have been made of these relationships. Clothier (1950) studied _S. v. monticola_ and _S. o. obscurus_ in western Montana and concluded that the two supposed kinds actually were not separable in that area. Durrant (1952:33) was able to separate the two kinds in Utah as was Hall (1946:119, 122) in Nevada. Other mammalogists who worked within the range of the _vagrans-obscurus_ groups have avoided the problems in one way or another. Recently Rudd (1953) has examined the relationships of _S. vagrans_ to _S. ornatus_.

MATERIALS METHODS AND ACKNOWLEDGMENTS

Approximately 3,465 museum study skins and skulls were studied. Most of these were assembled at the University of Kansas Museum of Natural History, but some were examined in other institutions.

Specimens were grouped by geographic origin, age, and sex. Studies of the role of age and sex in variation were made. Because it was discovered that secondary sexual variation was negligible, both males and females, if of like age and pelage, were used in comparisons designed to reveal geographic variation.

External measurements used were total length, length of tail, and length of hind foot. After studying a number of cranial dimensions I chose those listed below as the most useful in showing differences in size and proportions of the skull. Figures 1 and 2 show the points between which those measurements were taken.

_Condylobasal length._--From anteriormost projection of the premaxillae to posteriormost projection of the occipital condyles (a to a').

_Maxillary tooth-row._--From posteriormost extension of M3 to anteriormost extension of first unicuspid (b to b').

_Palatal length._--From anteriormost projection of premaxillae to posteriormost part of bony palate (c to c').

_Cranial breadth._--Greatest lateral diameter of braincase (d to d').

_Least interorbital breadth._--Distance between medialmost superior edges of orbital fossae, measured between points immediately above and behind posterior openings of infraorbital foramina (e to e').

_Maxillary breadth._--Distance between lateral tips of maxillary processes (f to f').

, 28545 KU.)]

In descriptions of color, capitalized terms refer to those in Ridgway (1912). In addition the numerical and alphabetical designations of these terms are given since a knowledge of the arrangements of these designations enables one quickly to evaluate differences between stated colors. Color terms which are not capitalized do not refer to any precise standard of color nomenclature.

In the accounts of subspecies, descriptions, unless otherwise noted, are of first year animals as herein defined. Descriptions of color are based on fresh pelages.

Unless otherwise indicated, specimens are in the University of Kansas Museum of Natural History. Those in other collections are identified by the following abbreviations:

AMNH American Museum of Natural History CM Carnegie Museum ChM Chicago Museum of Natural History CMNH Cleveland Museum of Natural History FC Collection of James S. Findley HC Collection of Robert Holdenreid SGJ Collection of Stanley G. Jewett CDS Collection of Charles D. Snow AW Collection of Alex Walker NMC National Museum of Canada OSC Oregon State College PMBC British Columbia Provincial Museum of Natural History SD San Diego Natural History Museum BS United States Biological Surveys Collection USNM United States National Museum UM University of Michigan Museum of Zoology OU University of Oregon Museum of Natural History UU University of Utah Museum of Zoology WSC Washington State College, Charles R. Conner Museum

In nature, the subspecies of _Sorex vagrans_ form a cline and are distributed geographically in a chain which is bent back upon itself. The subspecies in the following accounts are listed in order from the southwestern end of the chain clockwise back to the zone of overlap.

The synonymy of each subspecies includes the earliest available name and other names in chronological order. These include the first usage of the name combination employed by me and other name combinations that have been applied to the subspecies concerned.

In the lists of specimens examined, localities are arranged first by state or province. These are listed in tiers from north to south and in any given tier from west to east. Within a given state, localities are grouped by counties, which are listed in the same geographic sequence as were the states and provinces (N to S and W to E). Within a given county, localities are arranged from north to south. If two or more localities are at the same latitude the westernmost is listed first. Marginal localities are listed in a separate paragraph at the end of each account. The northernmost marginal locality is listed first and the rest follow in clockwise order. Those records followed by a citation to an authority are of specimens which I have not personally examined. Marginal records are shown by dots on the range maps. Marginal records which cannot be shown on the maps because of undue crowding are listed in Italic type.

To persons in charge of the collections listed above I am deeply indebted. Without their generous cooperation in allowing me to examine specimens in their care this study would not have been possible. Appreciated suggestions in the course of the work have been received from Professors Rollin H. Baker, A. Byron Leonard, R. C. Moore, Robert W. Wilson, and H. B. Tordoff, and many of my fellow students. Mr. Victor Hogg gave helpful suggestions on the preparation of the illustrations. My wife, Muriel Findley, devoted many hours to secretarial work and typing of manuscript. Finally I am grateful to Professor E. Raymond Hall for guidance in the study and for assistance in preparing the manuscript. During the course of the study I received support from the University of Kansas Endowment Association, from the Office of Naval Research, and from the National Science Foundation.

NON-GEOGRAPHIC VARIATION

Non-geographic variation, that is to say, variation within a single population of shrews, consists of variation owing to age and normal individual variation. In _Sorex_ I have detected no significant secondary sexual differences between males and females; accordingly the two sexes are here considered together.

Variation with age must be considered in order to assemble comparable samples of these shrews. Increased age results in wear on all teeth and in particularly striking changes in the size and shape of the first incisors. Skulls of older shrews develop sagittal and lambdoidal ridges, and further differ from skulls of young animals in being slightly broader and shorter, and in developing thicker bone, particularly on the rostrum which thus seems to be, but is not always in fact, more robust. Pruitt has recently (1954) noted these same cranial differences in specimens of _Sorex cinereus_ of different ages.

Several students of American shrews, notably Pearson (1945) on _Blarina_, Hamilton (1940) on _Sorex fumeus_, and Conaway (1952) on _Sorex palustris_, have shown that young are born in spring and summer, usually reach sexual maturity the following spring, and rarely survive through, or even to, a second winter. The result is that collections made, as most of them are, in spring and summer, contain two age classes, first year and second year animals. These two age classes are readily separable on the basis of differences in the skull as well as on the decreased pubescence of the tail and the increased weight of second year animals. My own examination of hundreds of museum specimens confirms this for the _Sorex vagrans_ group. Separation of the two age classes in an August-taken series of _Sorex vagrans_ from coastal Washington is shown in figure 3, in which two tooth-measurements that are dependent upon wear are plotted against one another.

First year animals are more abundant in collections than are second year animals. Within the first year, that is to say from spring to late fall, animals vary but little. Dental characters are best studied in first year shews. For this reason I have used them as the basis for the study of geographic variation, and descriptions are based on first year animals unless otherwise noted.

CHARACTERS OF TAXONOMIC WORTH

Within the _Sorex vagrans_ complex, the only characters of taxonomic significance that I have detected are in size and color. It is true that cranial proportions, such as relative size of rostrum, may change from population to population, but these proportions seem to me to be dependent upon actual size of the individual shrew as I shall elsewhere point out. Of the cranial measurements here employed, palatal length and least interorbital breadth are the most significant and useful. Color in the _S. vagrans_ group seems to be in Orange and Cadmium Yellow, colors 15 and 17 of Ridgway (1912). No specimens actually possess these pure colors, but most colors in these shrews are seen to be derived from the two mentioned by admixture of black and/or neutral gray. In color designations an increase in neutral gray is indicated by an increased number of prime signs ('), whereas increase in black is indicated by progressive characters of the Roman alphabet (_i_, _k_, _m_). Thus, 17''_k_ is grayer than 17'_k_ and 17''_m_ is blacker than 17''_k_. In subspecific diagnoses in this report, color and size, and sometimes relative size, are the characters usually mentioned.

PELAGE CHANGE

In general, winter pelage is darker than summer pelage in these shrews. Winter pelage comes in first on the rump and spreads caudad and ventrad. The growth line of incoming hair is easily detected on the fur side of the skin. Throughout the winter the color of the pelage changes, often becoming somewhat browner, although no actual molt takes place. This was noted by Dalquest (1944) who assumed that the color change resulted from molt although he was unable to detect actual replacement of hairs. Summer pelage usually comes in first on the back or head and moves posteriorly and laterally. Time of molt depends on latitude and altitude. Summer pelage may appear fairly late in the season and may account for the anomalous midsummer molt noted by Dalquest. Fresh pelages of summer and winter are best seen in first year animals and are less variable than are worn pelages and hence are used as the basis of color descriptions.

GEOGRAPHIC DISTRIBUTION AND VARIATION

Pacific Coastal Section

The largest shrews of the _vagrans_ group (large in all dimensions) occur in the coastal forests of northern California and of Oregon. Those shrews are reddish, large-skulled, large-toothed, and have rostra that are large in proportion to the size of the skull as a whole. The very largest of these shrews live along the coast of northwestern California. To the southward they are somewhat smaller, and at successively more northern localities, to as far as southwestern British Columbia, they are likewise progressively smaller and also somewhat less reddish. The relative size of the rostrum decreases with the decrease in size of the skull; consequently smaller shrews have relatively smaller rostra (see fig. 4). In addition the zygomatic ridge of the squamosal decreases in relative size with decrease in actual size of the skull. Thus, these features change in a clinal fashion as one proceeds from, say, Humboldt County, California, northward to Astoria, Oregon.

Turning our attention now farther inland to the Cascade Mountains of northern Oregon, the shrews there also are smaller and less reddish (more brownish) than in northwestern California, and the trend to smaller and darker shrews culminates in the northern Cascades of Washington. Shrews from there, and from the southwestern coast of British Columbia, compared with those from northwestern California, are much smaller and have so great a suffusion of black that they appear brown rather than red. At places along the coast successively farther north of southwestern British Columbia the shrews become larger again, the largest individuals being those from near Wrangell, Alaska. From that place northwesterly along the coast of Alaska, size decreases again.

The shrews so far discussed inhabit forests in a region of high rainfall and a minimum of seasonal fluctuation in temperature. Such a habitat seems to be the optimum for shrews of the _vagrans_ group since the largest individuals are found there. In addition, shrews seem to be as common, or commoner, in this coastal belt, than they are in other places.

The large shrews of the _vagrans_ group on the Pacific coast were divided into three species by H. H. T. Jackson in his revision of the North American _Sorex_ in 1928. The large reddish shrews of the coast of California and southern Oregon were called _S. pacificus_. The somewhat smaller ones from the coast of central Oregon were called _S. yaquinae_. Still smaller shrews from northwestern Oregon and from the rest of the Pacific coast north into Alaska were called _S. obscurus_. I find these kinds to intergrade continuously one with the next in the manner described and conclude that all are of a single species.

Inland Montane Section

Inland from the coasts of British Columbia and Alaska the size of the _vagrans_ shrew decreases rapidly. Specimens from western Alaska, central Alaska, and the interior of British Columbia are uniformly smaller than coastal specimens. In addition the red of the hair is masked more by neutral gray than by black with the result that the pelage is grayish rather than brownish or reddish. Shrews of this general appearance are found southward through the Rocky Mountain chain to Colorado and New Mexico. On the more or less isolated mountain ranges of Montana east of the continental divide the _vagrans_ shrew is somewhat smaller still. On the Sacramento Mountains of southeastern New Mexico the shrew is somewhat larger and slightly darker. Southwestward from the Colorado Rockies this shrew becomes smaller and slightly more reddish (less grayish).

All of these montane populations of the _vagrans_ shrew are commonest in hydrosere communities, that is to say, streamsides and marshy areas where the predominant vegetation is grass, sedges, willows, and alders. Since these animals are less common within the montane forests, hydrosere communities, rather than the actual forest, seem to be the positive feature important for the shrews.

The shrews of the montane region just described were regarded by Jackson as belonging to two species: _Sorex obscurus_, occupying all the Rocky Mountains south to, and including, the Sacramento Mountains; _S. vagrans_, made up of small individuals from various places in Wyoming, Montana, and Colorado, and all the shrews of western New Mexico and all of Arizona. My study of these animals has led me to conclude that the smaller shrews of Arizona and New Mexico intergrade in a clinal fashion with the shrews of Colorado and in fact represent but one species. Since some individuals from Colorado are as small as larger individuals from this southwestern population of small animals, I conclude that such specimens are the basis for reports of _S. vagrans_ from Colorado. The shrews of the Sacramento Mountains resemble those of the Colorado Rockies more than they do the smaller shrews of western New Mexico and Arizona, possibly because the climate is similar in the Sacramento Mountains and the higher Colorado Rockies. There is less precipitation in the more western mountain ranges in New Mexico and in Arizona in April, May, and June than in the Colorado Rockies. These months are critical for the reproduction and growth of shrews.

As mentioned above, the shrews from east of the continental divide in Montana are smaller than those of the other mountains of the state, and it is upon such small animals that the name _Sorex vagrans_ has been based in this area. It is clear, however, that these smaller animals intergrade with the larger shrews of the more western mountains. The small size might be an adaptation to the lesser precipitation and harsher continental climate east of the continental divide in Montana.

Great Basin and Columbia Plateau Section

The vagrant shrews of the Great Basin and adjoining Columbia Plateau and Snake River Plains are smaller than their relatives in the Rocky Mountains and, by virtue of less gray in their pelage, are reddish in summer and blackish rather than grayish in winter. There is little significant geographic variation in shrews throughout this region, although owing to their restriction to the vicinity of water, the populations of shrews are more or less isolated from one another and each is somewhat different from the next. Those from nearest the Rockies are sometimes slightly larger and those from some places in Nevada are slightly paler than the average. This small reddish shrew is found all the way to the Pacific coast of California, Oregon, and Washington. In these coastal areas it is somewhat darker and sometimes a trifle larger than elsewhere. It intergrades with a somewhat larger, grayer shrew in the Sierra Nevada of California. Along the Wasatch front in Utah, this Great Basin shrew intergrades with the larger, grayer shrew of the Rockies. Owing to the abrupt change in elevation, the zone of intergradation is rather narrow horizontally. In the latitude of Salt Lake City, populations of intergrades occur at between 8,700 and 9,000 feet elevation. The lowland shrew occurs in the eastern part of the Snake River Plains, and along the valleys of the Bear and Salt rivers into Wyoming. Along the northern edge of the Snake River Plains and on the western edge of the mountains of central Idaho the transition from lowland to montane habitats is abrupt and in consequence the zone of contact between small and large shrews is narrow. In northern Idaho and northwestern Montana the transition from lowland to highland is more gradual. Tributaries of the Columbia River system, especially the Clark Fork, provide a path for movement of lowland forms into intermontane basins of western Montana. In addition, the vegetational zones are found at lower elevations, and there are boreal forests in the lowlands rather than only in the mountains as is the case in Utah and Colorado. In this area, therefore, the zone of intergradation between the smaller lowland shrew and the larger montane shrew is more gradual and gradually intergrading populations are found over a relatively large area. This has been well demonstrated for northwestern Montana by Clothier (1950). In southern British Columbia and northern Washington this shrew in the mountains is large and in the intermontane valleys is small. There is extensive interdigitation of valleys and mountain ranges, and, consequently, of life-zones in this region. In a few places, recognizably distinct populations of the vagrant shrew occur within a few miles of one another, but in other places there are populations of intergrades. West of the Cascades no evidence of intergradation has been found and the two kinds occur almost side by side and maintain their distinctness.

These Great Basin shrews dwell in hydrosere communities as do their Rocky Mountain counterparts. In this arid region such a habitat obviously is the only one habitable for a shrew of the _vagrans_ group. These shrews often maintain their predilection for such habitats when they reach the Pacific coast, and are commonly found in such places as coastal marshes, marshy meadows, and streamsides, while the woodlands are inhabited by other species.

These small shrews of the Great Basin and the small vagrant shrews of the Pacific Coast were called _Sorex vagrans_ by Jackson.

Summary of Geographic Variation

Large reddish shrews of the coast of California and southwestern Oregon become smaller and darker to the north. From southwestern British Columbia they again become larger as one proceeds northward along the coast to Wrangell, Alaska, and north of that they again become smaller. Moving inland from the coast the shrews become markedly smaller in Alaska and British Columbia. The smaller inland and montane form occurs south through the Rocky Mountains, becoming slightly smaller in central Montana, slightly larger in southeastern New Mexico, and slightly smaller in western New Mexico and in Arizona. This montane form intergrades with a smaller more reddish Great Basin shrew, the zone of intergradation roughly following the western slope of the Rocky Mountains. The Great Basin shrew occurs westward to the Pacific Coast; there the Great Basin shrew occurs with, although in part it is ecologically separated from, the large reddish coastal shrews.

There seems to be an intergrading chain of subspecies of one species, the end members of which (the small Great Basin form and the large coastal form) are so different in size and ecological niche that they are able to coexist without interbreeding. In southern British Columbia the morphological differences are not so marked as farther south along the Pacific Coast. There, in British Columbia, reproductive isolation is not complete and occasional populations of intergrades occur. In Montana extensive intergradation occurs in a broad zone of transitional habitat. Along the western edge of the Rockies from Idaho south to Utah the zone of transition from montane to basin habitat is sharp and the zone of intergradation, although present, is fairly narrow, perhaps because there is little intermediate habitat which logically might be expected to be most suitable for intergrading populations.

The oldest name applied to a shrew of the group under consideration is _Sorex vagrans_ Baird, 1858, the type locality of which is Willapa Bay, Pacific County, Washington. The name applies to the small vagrant shrew of this area, rather than to the larger forest dweller which has been known as _Sorex obscurus_. The name _S. vagrans_, in the specific sense, must therefore apply to all the shrews discussed which have heretofore been known by the names _S. pacificus_, _S. yaquinae_, _S. obscurus_, and _S. vagrans_.

A situation such as the one here described where well differentiated end members of a chain of subspecies overlap over an extensive geographic range throughout the year without interbreeding--thus reacting toward one another as do full species--so far as I know has not previously been found to exist in mammals. The overlapping end-members of the chain of subspecies of _Sorex vagrans_ really do coexist; specimens of the overlapping subspecies have been taken together at the same localities from California to British Columbia. I have taken a specimen of _S. v. vagrans_ and several of _S. v. setosus_ in the same woodlot at Fort Lewis, Pierce County, Washington. Two subspecies of deer, _Odocoileus hemionus_, in the Sierra Nevada of California, occur together over a sizeable area but for only a part of each year that does not include the breeding season (Cowan, 1936:156-157). In the deer mouse, _Peromyscus maniculatus_, the geographic ranges of several pairs of subspecies meet at certain places without intergradation of the two kinds. In these instances well marked ecological differences exist between the subspecies involved. In western Washington, for example, the geographic range of the lowland subspecies, _P. m. austerus_, interdigitates to the east and west with the range of the montane and coniferous forest-inhabiting subspecies, _P. m. oreas_, and the two kinds have not been shown to intergrade. _Peromyscus maniculatus artemesiae_ and _P. m. osgoodi_ come together without interbreeding in Glacier National Park, Montana. _P. m. artemesiae_ is almost entirely a forest-dwelling subspecies, whereas _osgoodi_ is an inhabitant of open country. The two kinds do not actually occur together ecologically although they occur together in buildings at the edge of the woods (A. Murie, 1933:4-5).

Cases of sympatric existence of two subspecies of one species are known in birds and in reptiles. Notable examples are in the gull, _Larus argentatus_ (Mayr, 1940), in the Old World warbler, _Phylloscopus trochiloides_ (Ticehurst, 1938), and in the great titmouse, _Parus major_ (Rensch, 1933), of the Old World. In the first species the two end-members, the herring gull and the lesser black-backed gull, occur together over an extensive region from northern Europe and the British Isles throughout Fennoscandia. Fitch (1940) described a rassenkreis with overlapping subspecies in the garter snake _Thamnophis ordinoides_.

The geographic distribution of the species _Sorex vagrans_ is shown in figure 5. The geographic range of the Great Basin subspecies is shown by a different pattern of lines than the other subspecies of _S. vagrans_. In the region in which the geographic range of the Great Basin subspecies overlaps those of the subspecies of the Pacific Coast, the pattern of shading for the Great Basin subspecies is superimposed on the patterns for the other subspecies.

ORIGIN OF THE _SOREX VAGRANS_ RASSENKREIS

The distribution of the species _Sorex vagrans_ and that of its immediate ancestors obviously has not always been the same; during glacial ages much of the present range of the species in Canada and in some of the higher mountains of the United States was covered with ice and not available to the shrew. Furthermore, large areas that are now too hot and dry to permit the existence of_ S. vagrans_ were at one time habitable. If we are to speculate on the manner in which the _Sorex vagrans_ rassenkreis originated we must inquire into the nature and extent of these climatic changes.

The most recent epoch of geological time, the Pleistocene, is known to have been divided into a series of alternating glacial and interglacial ages. During the glacial ages continental and montane glaciers are judged to have covered much of Canada and the northern United States. Concurrently the major storm tracks of the west probably were shifted southward; in any event much of the now arid intermontane west was much better watered than it is today.

The increased precipitation, and probably glacial meltwater, formed large lakes in the closed basins of the Great Basin. There were boreal forests at lower elevations than there are today in comparable latitudes and continuous boreal habitat probably connected many of the isolated mountain ranges of the southwest. That probability is supported by the presence of boreal animals and plants on many of these isolated ranges today. A boreal tree squirrel, such as _Tamiasciurus_, could hardly be suspected of crossing a treeless, intermontane desert valley, miles wide.

Interglacial ages were characterized by warmth and aridity as compared to the glacial ages. Glaciers retreated or disappeared, boreal forests became montane in much of the United States, and the lakes in the Great Basin were reduced or disappeared. One can envision that during such times boreal mammals were isolated, their geographic ranges were restricted, and Sonoran mammals expanded their ranges.

Evidence is more extensive concerning the number and extent of glacial ages in the eastern than in the western part of North America. This evidence suggests a division of the Pleistocene into four glacial ages and four interglacial ages, the fourth interglacial age corresponding to the present time. More information is available about the Wisconsinan, or last, glacial age, than about the earlier ones, because the last glaciation in many montane areas destroyed evidence of earlier glaciations. The names of currently recognized glacial and interglacial ages of the Pleistocene are listed below. The names of interglacial ages are in Italic type.

Wisconsinan
_Sangamonian_
Illinoian
_Yarmouthian_
Kansan
_Aftonian_
Nebraskan

We may think of these ages as an alternating series of cool moist and warm dry periods during which boreal mammals, and other organisms, alternately moved southward (disappearing in the glaciated regions) and northward into previously glaciated areas (while disappearing from southern areas except on isolated mountain ranges). _Sorex vagrans_ probably followed this pattern of movement and now is restricted to forested or well-watered places.

One possible series of events culminating in the formation of the _Sorex vagrans_ rassenkreis may be thought of as having begun during the Illinoian age. With much of Canada, and perhaps also many areas in the Rockies, Cascades, and the Sierra Nevada covered with glacial ice, the shrew-stock ancestral to _Sorex vagrans_ may well have occupied a more or less continuous range over the Colorado Plateau, the Columbian Plateau, the Great Basin, and in the forests of the Pacific Coast (as well as over part of eastern United States, as will be explained beyond; see fig. 7). At that time the species probably was a continuously interbreeding unit.

In the ensuing Sangamonian interglacial age all glaciers retreated or disappeared thereby opening up extensive areas in the north and in the higher mountains which were occupied by a boreal fauna, including _S. vagrans_. Concurrently the Great Basin, and probably also much of the Columbian Plateau, became dry, and desert conditions developed, perhaps much as they are today. Increasing aridity eliminated shrew habitat in most places between the Rocky Mountains and the Sierra Nevada-Cascade mountain chain with the result that the geographic range of the species resembled an inverted "U", one arm lying along the Rocky Mountains and the other along the Cascade-Sierra Nevada axis; the connection between the two arms was in British Columbia (see fig. 7). At present _Sorex vagrans_ does occur in isolated places in the Great Basin, but its existence there is tenuous and seemingly dependent upon the occurrence of permanent water such as Ruby Lake and Reese River. With such an arrangement as this it can readily be seen that gene flow between the eastern and western arms of the "U" would be greatly reduced by distance; consequently differentiation between the two might be expected.

Wisconsinan glaciation again rendered Canada uninhabitable, and it is quite possible that extensive areas in the Rocky Mountains, the Cascades and the Sierra Nevada were heavily glaciated. With the elimination of the northern part of the "U", the eastern and western arms became isolated, if not by the width of the Columbian Plateau at least by the glaciated Cascade Mountains. At the same time extensive areas on the Colorado Plateau and much of the area south to the Mexican highlands were again occupied by the species. Finally the Great Basin, again being well-watered, provided suitable habitat for, and was reoccupied by, _Sorex vagrans_ (see fig. 8). This reoccupation of the Great Basin took place probably from the Colorado Plateau and mountains of Arizona and Utah, since the present day shrews of the species _S. vagrans_ in the Great Basin closely resemble Rocky Mountain shrews but differ markedly from the large endemic subspecies of the Pacific Coast.

Finally, with the waning of Wisconsinan ice, the species again was able to occupy northern and montane areas as it had during Sangamonian times. Again dessication of the Great Basin caused drastic restriction of shrew habitat. The small, marsh-dwelling kind of wandering shrew which had developed there around the lakes of Wisconsinan time occupied suitable habitat all the way to the Pacific coast where its range came into contact with that of the western arm of the Sangamonian "U."-pattern of shrew distribution (see fig. 9). The animals of this western segment and the new arrivals from the east were by this time so different from one another that the two kinds lived in the same areas without interbreeding. The descendants of the original western arm now are known as _Sorex vagrans sonomae_, _S. v. pacificus_, _S. v. yaquinae_, and _S. v. bairdi_. The newcomers from the east are known as _S. v. vagrans_, _S. v. halicoetes_, _S. v. paludivagus_ and _S. v. vancouverensis_.

In addition to occupying the Pacific Coast from San Francisco Bay north to the Fraser Delta, the Great Basin subspecies populated the Columbia Plateau and the western foothills of the central and northern Rockies. By so doing that subspecies came into secondary contact with its own parent stock with which it was still in reproductive continuity in Utah. In some places in British Columbia differentiation between the two kinds had proceeded to such an extent that some reproductive isolation was effected, but in many other places the two interbred. The Rocky Mountain form spread north and west and occupied the Cascades and coastal lowlands in southwestern British Columbia and in Washington. Here the differentiation between the Rocky Mountain subspecies and the Great Basin subspecies was great enough to cause complete reproductive isolation.

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 _neomexicanus_.

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_, _elassodon_, _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 grassland. 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 palustris_ 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 _palustris_ 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 _palustris_ 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 palustris_ 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.

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. nanus_, 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 nanus_ 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. nanus_. Indeed, the differences between the species _S. nanus_, _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 palustris_, 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. palustris_ and _S. bendiri_ are actually closely related species of the same subgenus and may have differentiated from one another because of separation into eastern (_palustris_) 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-palustris_ 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 circumpolar 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. nanus, S. juncensis, S. willeti, S. sinuosus, S. veraepacis, S. palustris, S. bendiri, S. alaskanus_, and _S. pribilofensis_.

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

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

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

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