It is uncertain if native Americans knew of Crystal Ball Cave, for no ancient human artifacts were found in this study. The cave was discovered by the late George Sims of Gandy in February, 1956. He found the original 3 foot (1 meter) diameter entrance that leads into a large chamber (see figure 3). The original east entrance was enlarged, the north entrance was blasted out through a soil-filled passage at the other end of the cave (see figure 3), and other improvements were made by Cecil R. and Jerald C. Bates of Gandy, Utah and Thomas E. Sims of Elko, Nevada (J. C. Bates 1983 personal communication).

Herbert H. Gerisch and Robert Patterson collected bones from Site 1 (see figure 3) in 1956 that they donated to the Los Angeles County Museum (H. H. Gerisch 1983 personal communication). Later Michael Stokes of the Los Angeles County Museum collected additional specimens from Site 1. These early collections consisted of float only and included bones of extinct horses and camels. On at least one of these early expeditions, some specimens were also collected from Gandy Mountain Cave, a smaller cave that lies about 1/4 mile (0.4 km) south of and 100 feet (30 meters) higher than Crystal Ball Cave. Specimens from these two caves are indistinguishable in the Los Angeles County Museum collection because the cave in which each specimen was recovered was not recorded. I dug several test pits in Gandy Mountain Cave in 1981 and found preservation to be poor and specimens to be few and probably all Recent. So although some specimens were collected from Gandy Mountain Cave, they are not considered in this study, except some which may be among the Los Angeles County Museum collection.

The first extensive collecting in Crystal Ball Cave was done in 1977 by Wade E. Miller and his students from Brigham Young University who used fine screens to obtain thousands of specimens (all from Site 1). Miller (1982) described this investigation and listed the genera identified in a report on vertebrate fossils from Lake Bonneville deposits. Wade E. Miller and I operated similar collecting projects in 1981 and 1982 (Sites 1, 2, and 3), and I wrote a preliminary report on this study (Heaton 1984). Crystal Ball Cave is Los Angeles County Museum locality 4534 and Brigham Young University vertebrate paleontology locality 772; the specimens from the cave are cataloged as LACM 123655-123711 and BYUVP 5300-8888, 8911-8933. Taxa recovered are listed in table 1.

The Crystal Ball Cave assemblage is the first late Wisconsinan age fauna to be described from the state of Utah. Although Utah has extensive Pleistocene deposits from Lake Bonneville, surprisingly few vertebrate fossils have been recovered from them (Miller, 1982). The only other Pleistocene assemblage that has been described from Utah is the Silver Creek fauna of north-central Utah, 14 miles (22 km) east of and 1,200 feet (360 meters) above Lake Bonneville's highest level and of late Sangamon to early Wisconsinan age (Miller 1976).

The nearest described Pleistocene vertebrate localities are four shelters located in Smith Creek Canyon, White Pine County, Nevada, 9 miles (14 km) south of Crystal Ball Cave. New species of mountain goat (Stock 1936), eagle (Howard 1935), and gigantic vulture (Howard 1952) were described from Smith Creek Cave, the primary site. Literature on the Smith Creek Canyon sites includes a description of the avifauna by Howard (1952), the micromammalian fauna by Goodrich (1965), the herpetofauna by Brattstrom (1976), the whole fossil assemblage by Miller (1979) and Mead et al. (1982), and the archaeology by Bryan (1979), Harrington (1934), and others. Although the Crystal Ball Cave fauna is chronologically and geographically close to that of Smith Creek Canyon, it differs in having its fossils deep in the cave, and this has resulted in significant differences between these assemblages. Crystal Ball Cave, for example, has more abundant mammal fossils but less abundant bird fossils than the Smith Creek Canyon sites.

The Crystal Ball Cave assemblage contains only small bones with a maximum length of about 10 cm and maximum weight of about 50 grams. This has caused some problems in identifying the large species since only the smallest elements, which have rarely been considered in other studies, are represented. The assemblage, however, is very large and is important since few assemblages of late Pleistocene age have been reported from the region. The size of this assemblage and the time restraints upon the project have limited the depth to which each taxon could be studied. For taxa with large numbers of specimens, only the best specimens were considered. Additional work could turn up more species, and statistical studies on the more abundant taxa could yield much additional information.

I have recognized four distinct stages of the cave's history: 1) a period of dissolution of limestone to form the cave, 2) a period of precipitation forming a layer of large calcite crystals ("nail head" spar) up to 1 foot (0.3 meters) thick on the cave walls, ceiling, and floor, 3) a period of partial dissolution of these crystals in the upper portions of the cave, the appearance of joints that cut the large calcite crystals, and the dislodging of breakdown from the ceiling of the large entrance room, and 4) the formation of vadose calcite speleothems and influx of sediment and fossils from outside the cave.

The beginning of stage 1, the dissolution of the cave, is of uncertain date. Davis (1930) demonstrated that limestone dissolution to form caves occurs predominantly in a thin zone just below the water table which is rich in carbon dioxide from groundwater percolating down from the surface. Once it reaches the water table, this groundwater dissolves the rock as it moves very slowly down the water table slope (Davies 1960). This appears to be the case in Crystal Ball Cave, since no scalloped or stream-cut passages are present to suggest the presence of fast-moving water expected in an above water table origin (Malott 1938, Myers 1969). The cave tends to parallel the strike of the beds and is relatively horizontal, as would be expected if the cave were formed at the water table. Green (1961) cited evidence that some caves in western Utah predate the tilting associated with the Basin and Range uplift. The fact that Crystal Ball Cave is roughly horizontal and parallel to the strike of the beds suggests that it postdates this tilting. But since the cave is high in a small isolated hill, considerable uplift and/or erosion must have taken place since the cave was at the water table. The cave does not parallel the present land surface as the water table tends to do (Myers 1969), and this further suggests that much overburden has been removed since the original dissolution of the cave.

Stage 2, the precipitation of large calcite crystals, represents a different groundwater environment than the preceding dissolutional stage. It is generally agreed that such "nail head" spar forms in still, calcite-saturated water where nucleation centers are free to grow into large euhedral crystals (Hill 1976). This shift from dissolution to precipitation does not necessarily represent a significant change in the level of the water table, but it does represent the drastic reduction in the carbon dioxide content of the groundwater necessary for calcite precipitation (Moore and Nicholas 1964). Several vertical cavities (domes) extend upward into the cave ceiling, and these predate the calcite precipitation because they are partly filled by it. Moore and Nicholas (1964) cited evidence that such domes form late in cave dissolution and provide more direct water and air connections to the surface. Myers (1969) stated that they are of vadose (above water table) origin and caused by vertical seepage. Perhaps the formation of these domes allowed gas exchange between the groundwater in the cave and the surface permitting carbon dioxide to escape and calcite to be precipitated.

Stage 3 includes several events which are not chronologically separable. Some of the calcite crystals in the roof of the cave are completely dissolved, and locally some of the limestone bedrock underneath it is also. This is especially evident in the aforementioned domes. Joints and breakdown, both of which cut the previously formed calcite crystals, probably represent one or several earthquakes. If any uplift postdates the cave's origin, it probably occurred during this stage. These cracks and breakdown blocks were later filled and covered by the speleothems of stage 4, showing their chronologic relationship.

Stage 4 postdates the loss of voluminous standing water in the cave and the opening of entrances large enough to allow considerable gas exchange and sediment into the cave. Vadose speleothems such as the stalactites, columns, and rimstone pools found in Crystal Ball Cave form subaerially in caves having enough gas exchange with the surface to allow carbon dioxide to escape from the dripping groundwater (Moore and Nicholas 1964). Near the east entrance of the cave some small columns have formed upon and been partly covered by sediment coming in from the entrance, showing the concurrence of these events. The vertebrate material under study entered the cave during stage 4 when cave opening(s) were sufficiently large to allow their entry and dry conditions allowed their preservation; therefore cave stages 1 through 3 predate the oldest fossils.

Twelve sediment samples were collected at sites throughout the cave and screened to determine the degree of sorting. All samples are poorly sorted, but samples farthest from known entrances tend to have a higher percentage of fine particles. Particles under 0.0024 inch (0.061 mm) in diameter make up over half the weight percent of three such samples. Samples were placed in hydrochloric acid to remove all calcite. Sediment from Site 1 (see figure 3) is composed of about 80% calcite and 20% very fine but poorly sorted clastic grains, namely quartz, mica, and an unidentified ferromagnesian mineral. Larger clastic grains were found in samples closer to each entrance and comprised greater portions of the sediment.

The calcite portion of the sediment is composed of both crystal fragments, probably derived from broken "nail-head spar," and cryptocrystalline caliche-like crust associated with clastic fragments, almost certainly precipitated in the cave. The clastic fragments could have been washed in, blown in, brought in by animals, or been released from the cave walls as impurities in the limestone. The sediments at Site 1 show no sign of ever having been wet except in some areas where they have been cemented with calcite. But water runs in through the east entrance during storms filling the large entrance chamber with mud. Wind gusts can be quite strong through the cave during storms, but only because the north entrance was opened by man. The importance of these factors is difficult to determine, but the fact that the bulk of the sediment far inside the cave is calcite demonstrates that the sediments are mostly derived from within the cave by weathering of the limestone and calcite crystals rather than from outside sources.

On early collecting trips most of the field time was spent digging through the sediments and collecting specimens by hand. Sediment was also taken to the lab in bags and screened in order to recover smaller bones and teeth. After using this method for several trips, the collection had overwhelming numbers of rodent and lagomorph fossils, but bones of larger mammals were few. So on the last collecting trip large volumes of cave sediment were screened inside the cave with a coarse screen, and the number of larger bones in the collection was thereby more than doubled.

Little laboratory preparation was necessary with the larger Crystal Ball Cave fossils. A few required removal of hardened dirt or calcite. All were washed to remove dust. Considerable time was spent manually separating small bones and teeth from cave sediments. This was done in the laboratory with forceps after the sediments had been washed through a fine screen and allowed to dry. Approximately 35 cubic feet (1 cubic meter) of cave sediment was prepared in this manner, and virtually all the bone was removed.

Because of the great abundance of small mammal fossils recovered, only the skulls and jaws were studied. All identifiable material was used for larger mammals because they were not as well represented and because few dental elements were recovered. Identification was made by comparison to Recent specimens housed at the Brigham Young University Monte L. Bean Museum, fossil specimens housed at the Brigham Young University Vertebrate Paleontology Laboratory, and by extensive use of the literature.

Small living mammals were captured inside Crystal Ball Cave and around Gandy Spring at the base of Gandy Mountain. This trapping was not extensive, but it did indicate what species are abundant in and around the cave. The species trapped are recorded in table 1. Jerald C. and Marlene Bates of Gandy (1983 personal communication, 1984 personal communication) were interviewed for additional information about the modern local fauna and recent history of the cave including modification by man.

Four bone samples were sent to Geochron Laboratories, Cambridge, Massachusetts, for carbon-14 dating. Because of the small size of bones in the assemblage, these samples (which included some of the largest recovered) were just over 25 grams, the minimum weight suggested for dating. Two were of small extinct horses: a thoracic vertebra (BYUVP 7687) and a distal metapodial epiphysis (BYUVP 7568); and two were fragments of unidentified limb bones of large mammals. Geochron Laboratories cleaned and washed the four samples in acetic acid to remove adhering materials then crushed them and soaked them in agitated acetic acid for 24 hours to remove normal carbonates. The samples were then hydrolized under vacuum with hydrochloric acid to dissolve bone apatite and evolve its carbon dioxide for collection. The carbon dioxide samples were converted to methane and counted in a low background beta counter (with C-13 correction), and dates were based on the Libby half life (5570 years). The ages reported are listed in table 2. The oldest date of "over 23,000 Y.B.P." was given because no C-14 was detected in that sample.

The oldest date of 23,000+ Y.B.P. gives a minimum age for the time fossils first entered the cave. The youngest horse bone date of about 19,000 Y.B.P. gives a maximum age for the loss of that species from the area, although other studies have shown that small horses lived beyond 10,370 Y.B.P. in Idaho and until about 8,000 Y.B.P. in Arizona and Alberta, Canada (Kurten and Anderson 1972, 1980, Martin 1967). Otherwise the four dates give only a general age for the assemblage and provide no information about the antiquity of individual taxa. The fact that all four dates are over 12,000 Y.B.P. suggests that bones, at least of large mammals, may have been deposited more frequently during the late Pleistocene than during the Recent. If so, this could be due to a greater abundance of the animals themselves, a change in what animals (or other processes) deposited the fossils, or the former presence of larger or additional entrances.

Thirty radiometric dates have been reported from the Smith Creek Canyon sites (Thompson 1979, Thompson and Mead 1982, Valastro et al. 1977), and they demonstrate that accumulation of fossils there was concurrent with fossil deposition at Crystal Ball Cave. The two oldest Smith Creek dates are 28,650 Y.B.P. (Smith Creek Cave) and 27,280 Y.B.P. (Ladder Cave), which correlate well with the date of "over 23,000 Y.B.P." from Crystal Ball Cave. The other 28 Smith Creek Canyon dates are younger than 18,000 Y.B.P. with the majority being from 10,000 to 13,000 Y.B.P. The mean age of the four dated Crystal Ball Cave specimens is considerably older than the mean age of dated specimens from any of the Smith Creek Canyon sites, suggesting that its major period of fossil deposition was earlier; but a sample size of four dates is not statistically significant enough to demonstrate this.

In addition to Lake Bonneville, many pluvial lakes filled the valleys of Nevada including one just west of the Snake Range, 18 miles (30 km) west of Crystal Ball Cave (Mifflin and Wheat 1979). Based on studies of temperature and precipitation correlation, Mifflin and Wheat (1979) estimated that development of pluvial lakes in the area involved a temperature decrease of 5 F (3 C). Lower temperatures and higher annual precipitation caused floral boundaries to move lower in elevation and latitude during the Wisconsinan glacial (Thompson and Mead 1982, Wells 1983). This shift had a dramatic effect on small boreal mammals in the Great Basin because it allowed them to disperse between ranges, whereas now the intermontane basins act as absolute barriers (Brown 1971, 1978, Harper et al. 1978). Brown (1971, 1978) demonstrated that distribution of small boreal mammals is relectual from the Wisconsinan glacial and not a case of colonization-extinction equilibrium. The Crystal Ball Cave fauna shows what taxa have been extirpated from the Snake Range since the Wisconsinan glacial and documents northward shifts in the ranges of several species at the close of the Pleistocene.

Another striking feature of the late Pleistocene is the well-documented megafaunal extinction. At the end of the Wisconsinan glacial 41 species of large mammals went extinct--3 times more than at the end of any of the other Pleistocene glacials (Kurten and Anderson 1980). Different workers have attributed this to the rapid post-glacial climatic shift (Martin and Neuner 1978, Webb 1969) and to overkill by Early Man (Martin 1967, Mosimann and Martin 1975). The Crystal Ball Cave assemblage contains several of these extinct taxa, but the fact that it lacks human association and stratigraphic control makes it unable to provide any substantial data to resolve this controversy.

Consideration needs to be given to the role Crystal Ball Cave played as a shelter and the way fossils got into the cave. When the cave was discovered in 1956, the east entrance was a 3-foot (1-meter) diameter opening in solid rock, half filled with soft soil, which sloped downward into the large entrance chamber (see figure 3). Several 1-foot (0.3-meter) diameter entrances (which are often filled with woodrat nesting material) also exist just north of the east entrance. The north entrance was completely filled with debris, which if removed could make it 8 feet (2.5 meters) high and 20 feet (6 meters) wide. It could have been a large important entrance when the earlier bones were being deposited, but several factors preclude this. First, there are very few fossils in the deep dry sediments of the north half of the cave; the rich bone deposits are in the south half. Second, the fossil assemblage provides no evidence that there ever was a large entrance since large mammals are represented only by their smallest elements. If the north entrance ever was large it was probably prior to deposition of the fossils under study.

Neotoma lepida, Peromyscus maniculatus, and Plecotus townsendii were captured live inside the cave, so their presence in the assemblage is easy to explain. Other small mammals could also have lived in the cave or used it as a shelter. Small carnivores and scavengers could have brought their prey into the cave to eat. The presence of only the smallest isolated elements of large mammals suggests that these bones were brought into the cave individually after the carcasses deteriorated. Small carnivores could have contributed to this, but it is my opinion that these bones were taken into the cave primarily by wood rats, since they are known to take materials into the cave now and since all the bones in the assemblage are small enough for a wood rat to transport. Because the cave has small entrances and because the bones are found far within the cave, it is very unlikely that birds transported prey inside. There is also no evidence that prehistoric humans brought material into Crystal Ball Cave. This suggests that the species found in the assemblage lived and died in or near the cave and were not transported long distances, as could have occurred at Smith Creek Cave (see Bryan 1979, Harrington 1934).

It is unusual for caves to have their richest bone deposits far inside the cave rather than near an entrance. The east entrance takes in water during storms, and other areas are damp from seepage. Sites 1 and 2, which contain the richest bone deposits, are in one of the driest areas of the cave and are just outside the zone of total darkness when the sun shines through the east entrance. North of Sites 1 and 2 the passage constricts and enters total darkness but remains dry. Wood rat nests are particularly common at Sites 1 and 2, which helps explain why rich bone deposits are present if wood rats play an important role in getting them there. The extremely dry conditions at Sites 1 and 2 and their proximity to the east entrance, which I consider the primary entrance, are probably the reason why these sites have been so productive. Rarity of fossils nearer the east entrance is probably due to poorer preservational conditions and poorer sites for wood rat dwellings. Lack of rich bone deposits in the northern half of the cave is probably due to constricted passages and greater distance from a late Pleistocene entrance.

Discussion--Eutamius has two premolars in each maxilla whereas Tamius has only one. E. minimus is the smallest species of Eutamius and has a narrower and squarer P/4 than E. dorsalis or E. umbrinus. All the specimens listed above match E. minimus with respect to the P/4 and are smaller than the other species. E. minimus and E. dorsalis live in the region of Crystal Ball Cave and E. umbrinus lives higher in the Snake Range and westward into Nevada (Hall 1981). E. minimus was also recovered from Smith Creek Cave (Mead et al. 1982). E. minimus inhabits diverse habitats from deserts to forests, so its presence in the assemblage is not surprising.

Material--Right dentary with P/4, M/1,/2,/3 (BYUVP 6233), right dentary with P/4, M/1 (BYUVP 6257), 2 right dentaries with M/1 (BYUVP 5974, 6304), 2 left dentaries with M/1 (BYUVP 6129, 6134). Three partial right maxillae with M1/ (BYUVP 6064, 6288, 6295) and a partial left maxilla with M1/ (BYUVP 6000) also compare favorably with this species.

Discussion--E. dorsalis is distinctly larger than E. minimus and slightly larger than E. umbrinus (Mead et al. 1982). It has a distinct isolated mesoconid on M/1 that is lacking in E. umbrinus and is part of an ectolophid in E. minimus (Miller 1976). The M/1's of the six dentaries listed above match E. dorsalis in this character, and the four maxillae listed above match best in size with E. dorsalis but cannot be positively distinguished from E. umbrinus. Of the larger chipmunks, only E. cf. umbrinus was reported from Smith Creek Cave (Mead et al. 1982), and I have found only E. dorsalis in Crystal Ball Cave. Their present ranges may account for this difference since E. umbrinus only inhabits the Snake Range west of Crystal Ball Cave while E. dorsalis inhabits the entire range (Hall 1981). Their ranges show that E. umbrinus is more isolated in areas of high elevation and more commonly absent from the areas once covered by Lake Bonneville.

Material--Anterior portion of skull with both I/'s (BYUVP 6656), anterior portion of skull with left I/ (LACM 123672), right dentary with /I, P/4, M/1 (BYUVP 6657), left dentary with P/4 (BYUVP 8283). Four palates without teeth (BYUVP 6653-6654, 6664-6665), 4 right dentaries without cheek teeth (BYUVP 6660, 6663, 6666, 8281), and 8 left dentaries without cheek teeth (BYUVP 6655, 6658-6659, 6662, 6681, 7009-7010, 8282) also compare favorably with this species.

Discussion--Thomomys is distinguished from other North American geomyids by the absence of a superficial groove on the anterior face of the upper incisors (illustrations in Hall 1981), and none of the I/'s listed above have this groove. T. umbrinus differs from T. talpoides and T. monticola, the only other species of Thomomys living in Nevada, Utah, or surrounding areas, by having a sphenoidal fissure, by not having the palatine foramina fully anterior to the anterior openings of the infraorbital canals (Durrant 1952), and by the absence of a lingual indentation in the anterior lobe of P/4 (Hall 1946). The two Thomomys skulls from Crystal Ball Cave have the sphenoidal fissure, and their palatine foramina are fully anterior to the infraorbital canals. The two P/4's also lack the lingual indentation as in T. umbrinus. My observations and also figures 308-321 in Hall (1946) indicate that T. umbrinus has a larger mean size than the other two species mentioned (contrary to Bergman's rule), and all the Crystal Ball Cave specimens compare best in size with the larger T. umbrinus.

T. umbrinus is the only geomyid currently inhabiting the Snake Range, and it is a southern species extending from Nevada and Utah southward into Mexico (Hall 1981). T. bottae and T. townsendii are now considered as subspecies of T. umbrinus (Hall 1981). T. talpoides, which inhabits mountain ranges to the east and west of the Snake Range, has Nevada and Utah as almost its southern boundary and extends northward into Canada. T. talpoides tends to inhabit higher elevations than T. umbrinus as well as higher latitudes. T. cf. talpoides is the only geomyid reported from Smith Creek Cave (Goodrich 1965).

Hall (1946) pointed out that although T. umbrinus is usually a lower elevation species than T. talpoides, T. umbrinus is the only geomyid in the Snake Range and occurs at all elevations (but is less abundant at higher elevations than T. talpoides is at similar elevations in other ranges). Hall (1946) attributed this to antiquity of occupancy and proposed that T. umbrinus, having no competitors in the Snake Range, developed populations adapted to higher elevations. Since T. umbrinus was the species best adapted to the valleys surrounding the Snake Range, no species which were better adapted to higher elevations could pass through to their favorable habitat. This could explain why the Crystal Ball Cave assemblage suggests no northward range shift for species of Thomomys as it does for other groups such as lagomorphs. If Hall (1946) is right, the tentative assignment of the Smith Creek Cave specimen to T. talpoides (Goodrich 1965) must be in error. Another possibility is that predatory birds transported the specimen to the cave, but this seems unlikely since T. talpoides occurs only as close as 45 miles (75 km) to the northwest and 108 miles (180 km) to the east of Smith Creek Cave. Hall (1946) also pointed out that geomyids, as individuals, are extremely sedentary, and this could be the cause of their slow invasion and northward retreat compared to other mammals.

Material--Partial right maxilla with P4/ (BYUVP 6682), 2 right dentaries with P/4 (BYUVP 6859, 6879), 2 right dentaries with M/2 (BYUVP 6711, 6856), left dentary with all teeth (BYUVP 6697), left dentary with P/4, M/1 (BYUVP 6786), left dentary with P/4, M/2 (BYUVP 6115).

Discussion--P. longimembris, P. parvus, and P. formosus now inhabit the Crystal Ball Cave area, and the closest other species range more than 150 miles (250 km) to the east and south (Hall 1981). Of the three local species, P. longimembris can be ruled out because its M/3 is distinctly smaller than its P/4 (Hall 1981), and BYUVP 6697 has the opposite condition. P. parvus and P. formosus are very similar dentally, and the Crystal Ball Cave specimens match well with both of them. P. formosus has a larger mean size than P. parvus, and the Crystal Ball Cave specimens compare best in size with P. formosus although P. parvus and several other western species cannot be ruled out. Miller (1979) referred all the Perognathus specimens found at Smith Creek Cave to P. cf. parvus, but since the identification was tentative at both caves, it does not seem wise to speculate about a possible difference between the two assemblages.

Material--Right maxilla with P4/, M1/,2/ (BYUVP 6695), right maxilla with P4/, M2/ (BYUVP 6781), right maxilla with M1/ (BYUVP 6709). Three partial right maxillae with P4/ (BYUVP 6669, 6674, 6797), a partial right maxilla with a partial M1/ (BYUVP 6759), a right dentary with /I, P/4, M/1 (BYUVP 6693), 2 right dentaries with P/4 (BYUVP 6795, 6860), and a left dentary with P/4 (BYUVP 6708) are of Microdipodops and compare favorably with M. megacephalus.

Discussion--Microdipodops is most similar to Perognathus but can be distinguished dentally by the molars having a single enamel loop as opposed to the biloph nature of Perognathus molars. The P/4's are also distinct in being more hypsodont and having a straight posterolabial border as opposed to the round and symmetrical nature of the Perognathus P/4's. M. megacephalus ranges throughout most of Nevada and into neighboring states including Utah, and it is currently found around Crystal Ball Cave (Hall 1981). M. pallidus, the only other species, occurs along the southern Nevada-California border more than 200 miles (320 km) southwest of Crystal Ball Cave (Hall 1981). M. megacephalus can be distinguished from M. pallidus by the latter possessing a small notch in the labial side of M1/, and all the Crystal Ball Cave specimens possessing the M1/ are clearly M. megacephalus. M. cf. megacephalus was reported at Smith Creek Cave (Miller 1979), and all heteromyid taxa recovered were low in abundance as at Crystal Ball Cave. This low abundance is probably due to a low density in life since even now they are rarely seen in the area.

Material--Two right dentaries with /I (BYUVP 6672 and 8284), left dentary fragment with P/4 (BYUVP 6676). Nine maxillae without teeth (BYUVP 5593, 6667-6668, 6670, 6675, 6677-6680) and 2 right dentaries without teeth (BYUVP 6673, 6683) also compare favorably with this taxon.

Discussion--Dipodomys is distinctly larger than other heteromyid genera. D. microps is distinguished from other species of Dipodomys by having chisel-shaped lower incisors (anterior face flat) rather than awl-shaped lower incisors (anterior surface round), and the incisors of BYUVP 6672 and 8284 are chisel-shaped. P/4 is also distinct in having a larger and more isolated anterior loph than D. ordii or D. merriami but not a complete separation of lophs as in D. deserti, and the P/4 of BYUVP 6676 clearly matches D. microps. The referred specimens also match perfectly with Recent D. microps, but lack the diagnostic teeth. Of the four species of Dipodomys presently living in Utah and Nevada, D. microps and D. ordii are found in the Snake Range while D. merriami and D. deserti occur more than 125 miles (200 km) to the south and west (Hall 1981). D. microps has a much smaller range than D. ordii, occurring only in Nevada and parts of surrounding states (Hall 1981). The Dipodomys specimens recovered from Smith Creek Cave (Miller 1979) were referred to D. ordii because they had awl-shaped lower incisors. This difference between the two assemblages is difficult to explain because the range differences between these species do not suggest distinct differences in habitat preference.

Material--Right maxilla fragment with M1/,2/ (BYUVP 6703), left maxilla with M1/,2/,3/ (BYUVP 6782), left maxilla fragment with M1/,2/ (BYUVP 6771). Thirty nine Peromyscus dentaries containing one or more molars compare favorably with P. maniculatus and P. crinitus.

Discussion--Of the six species of Peromyscus that inhabit Utah and Nevada, only P. maniculatus, P. truei, and P. crinitus currently live around Crystal Ball Cave (Hall 1981). P. maniculatus was captured live inside the cave by the author in 1982 and 1983. Peromyscus fossils from Smith Creek Cave were not identified to the species level (Goodrich 1965, Mead et al. 1982, Miller 1979). Dental characters which distinguish species of Peromyscus are few and not always reliable. P. maniculatus and P. truei belong to the subgenus Peromyscus, which has accessory tubercles or enamel loops on the labial side of M1/ and M2/; P. crinitus belongs to the subgenus Haplomyomys which lacks these features (Hall 1981). I found this character to be quite reliable, and the specimens listed above all have prominent cusps on M1/ and M2/. In further refinement of this character, Miller (1971, 1976) was able to separate P. maniculatus from all other western species of Peromyscus by the presence of an anteroconule on M1/ with direct attachment to the anterocone rather than being joined to it by a distinct loph as in P. truei. Specimens listed above fit P. maniculatus in this respect. Species of the subgenus Haplomyomys usually lack the anteroconule entirely (Hall 1981, Miller 1971, 1976). Unfortunately, excessive wear on the teeth erases this character.

Of the 40 Peromyscus dentaries containing one or more molars, 39 compare best in size with the smaller P. maniculatus and P. crinitus, but no character could be found to separate these species based on dentaries. Miller (1976) found the P/3's of P. maniculatus, P. crinitus, and P. eremicus to be relatively more reduced than P. boylii and P. truei. The 8 Crystal Ball Cave Peromyscus dentaries containing M/3 tend to have M/3 relatively reduced as in P. maniculatus, P. crinitus, and P. eremicus, and in size all the 39 dentaries listed above compare best in size with the smaller P. maniculatus and P. crinitus.

Material--Right maxilla with M1/,2/ (BYUVP 6780), left maxilla with M1/,2/ (BYUVP 6769), left maxilla with M1/ (BYUVP 6715). Thirty nine Peromyscus dentaries containing one or more molars compare favorably with P. maniculatus and P. crinitus.

Discussion--These specimens lack accessory tubercles and enamel loops on the 2 principle outer angles of M1/ and M2/, so they probably belong to the subgenus Haplomyomys (Hall 1981). Of the two species of Haplomyomys found in Utah, P. crinitus and P. eremicus, the Crystal Ball Cave specimens compare better in size with the smaller P. crinitus (although there is considerable overlap). Some of the 39 dentaries discussed under P. maniculatus (above) could also belong to this species since no character was found to distinguish them based on dentaries. P. crinitus is presently found around the cave while P. eremicus only ranges as close as 135 miles (225 km) to the south (Hall 1981), and this further suggests that these specimens are P. crinitus.

Material--Left dentary with M/1 (BYUVP 6718).

Discussion--P. truei is the largest species of Peromyscus living in Utah and Nevada (Durrant 1952, Hall 1981), and the M/1 listed above compares well in size with this species and is larger than the mean size of P. eremicus and P. boylii and distinctly larger than any P. maniculatus or P. crinitus M/1's examined. Identification is based only on size since no other character could be found to distinguish M/1's of Peromyscus. This species is found throughout the Great Basin, so its presence in the assemblage is not surprising.

Material--Partial skull without teeth (LACM 123671), 2 partial right maxillae with M1/ (BYUVP 7045, 7065), 2 left maxillae with M1/ (BYUVP 7154), partial left maxilla with M1/ (BYUVP 7246).

Discussion--N. lepida and N. cinerea are the only species of Neotoma that presently inhabit the Snake Range (Hall 1946, 1981). Of three wood rats that I trapped in Crystal Ball Cave and two elsewhere on Gandy Mountain in 1982 and 1983, all were N. lepida. I did trap a N. cinerea in another cave in the Snake Range 22 miles (35 km) south of Crystal Ball Cave, so they are known to inhabit caves in the area. Miller (1979) reported both N. lepida and N. cinerea from Smith Creek Cave but did not comment on their relative abundance. Of these two species, N. cinerea is much more boreal than N. lepida, having a more northern range and being found at higher elevations (Finley 1958, Hall 1946, 1981). Durrant (1952) and Hall (1981) also reported N. albigula, N. mexicana, and N. stephensi living in Utah but far south and east of Crystal Ball Cave.

Neotoma cinerea is usually distinguishable from N. lepida by its larger size and deeper anterolabial reentrant angle on M1/ (Finley 1958). According to Hall (1946), the maxillary alveolar length is always 8.8 mm or less in N. lepida and 9.1 mm or more in N. cinerea for the Nevada subspecies, and Finley (1958) reported only a slight overlap for the Colorado subspecies. The three other Utah species of Neotoma are intermediate in size between N. lepida and N. cinerea, and N. albigula has the M1/ pattern of N. lepida while N. mexicana and N. stephensi have the M1/ pattern of N. cinerea (Finley 1958). Because these are the most diagnostic characters, only maxillae with M1/ and/or a measurable alveolar length were considered.

The Crystal Ball Cave specimens listed above compare best in size with N. lepida, the only species of Neotoma known to presently inhabit the cave. Maxillary alveolar lengths of Neotoma specimens from the cave show a strongly bimodal distribution, suggesting that N. albigula, N. mexicana, and N. stephensi are not represented since they are intermediate in size between N. lepida and N. cinerea. The shallow anterolabial reentrant angle of the M1/'s also compares favorably with N. lepida. The scarcity of N. lepida specimens in the assemblage suggests that this species probably has not always inhabited the cave as it does now.

Material--Anterior portion of skull with both I/, M1/,2/ (BYUVP 7384), maxilla pair with all teeth except left I/ (BYUVP 7281), maxilla pair with right M1/,2/,3/, left M1/,2/ (BYUVP 7282), maxilla pair with both M1/,2/ (BYUVP 7067), maxilla pair with right M/1,/2 (BYUVP 7015), maxilla pair with left M/1 (BYUVP 7213), 9 right maxillae with M1/,2/,3/ (BYUVP 7136, 7149, 7158, 7167, 7214, 7248, 7254, 7314, 7320), 3 right maxillae with M1/ (BYUVP 7273, 7316, 7330), 25 partial right maxillae with M1/ (BYUVP 7014, 7018, 7024, 7038, 7046, 7104, 7114, 7125, 7134, 7138, 7147, 7170, 7177, 7180, 7182, 7197, 7204, 7216, 7242, 7247, 7249, 7276, 7344, 7348, 7349), 10 right maxillae without teeth (BYUVP 7255, 7343, 7353, 7367, 7377, 8286-8290), 7 left maxillae with M1/,2/,3/ (BYUVP 7095, 7212, 7250, 7257, 7274, 7376, 7379), 4 partial left maxillae with M1/,2/ (BYUVP 7101, 7174, 7179, 7324), partial left maxilla with M1/,2/ (BYUVP 7017), 34 partial left maxillae with M1/ (BYUVP 7021, 7061, 7062, 7072, 7073, 7087, 7099, 7106, 7133, 7140, 7142, 7144, 7145, 7151, 7162-7164, 7172, 7175, 7183, 7189, 7200, 7205, 7217, 7220, 7225, 7267, 7300, 7317, 7318, 7322, 7351, 7362, 7371), 6 left maxillae without teeth (BYUVP 7171, 7346, 8291-8294). Another approximately 200 maxillae, 200 dentaries, and 2,000 isolated molars compare best with this species.

Discussion--Neotoma cinerea is recognized by its large size and deep anterolabial reentrant angle on M1/ as discussed above. N. cinerea has the largest mean size of any species of Neotoma, and all the specimens listed above match Recent N. cinerea in size and have he deep anterolabial reentrant angle on M1/ when this tooth is present. This makes N. cinerea the second best represented species in the Crystal Ball Cave assemblage after Spermophilus townsendii. The fact that N. cinerea is abundant in the assemblage, but not found in the cave now, while N. lepida is rare in the assemblage, but now the only wood rat living in the cave, suggests that a replacement of N. cinerea by N. lepida has recently taken place in the area. The great abundance of N. cinerea remains at Sites 1 and 2 of Crystal Ball Cave also helps substantiate my hypothesis that wood rats were the primary means of transporting fossils, especially of large mammals, into the cave. The dominance of N. cinerea over N. lepida in the assemblage suggests that N. cinerea was the primary species involved in this transport.

The ecological differences between N. cinerea and N. lepida have significance both to the replacement of the former species by the latter and to the accumulation of fossils in the cave. Finley (1958), in his detailed study of Colorado wood rats, found den sites to be the most limited resource for all species. Since all wood rats prefer the same basic types of den sites, namely rocky crags and caves, multiple species are rarely found coexisting (Finley 1958). This suggests that when conditions at Crystal Ball Cave reached a threshold where they favored N. lepida over N. cinerea, the replacement took place quickly. N. cinerea prefers higher elevations and latitudes than N. lepida, and hot summers in arid regions seem to be a limiting factor for this species (Finley 1958, Hall 1981). The changing conditions that led to the replacement of N. cinerea by N. lepida may have been the increase in temperature and decrease in moisture at the close of the Pleistocene, the shift in vegetation caused by it, or both. Regarding food, Finley (1958) stated that N. cinerea prefers soft-leaved shrubs, forbs, and montane conifers, whereas N. lepida prefers xerophytic shrubs, forbs, cacti, and shrubby trees.

Species of Neotoma differ somewhat in den preferences and collecting habits. Finley (1958) stated: "Dens of N. cinerea are usually in high vertical crevices in cliffs or caves, whereas those of . . . N. lepida are usually in low horizontal crevices or under boulders or large fallen blocks. Dens of [N.] cinerea usually contain larger accumulations of sticks and bones." That N. cinerea collects more material, especially bone, is very significant since I consider wood rats as the primary mechanism of fossil deposition at Crystal Ball Cave. This suggests that the rate of bone deposition decreased when N. lepida replaced N. cinerea, and it helps explain why many elements of the present local fauna are so poorly represented and why all the dated fossils were late Pleistocene rather than Recent in age.

A replacement of N. cinerea by N. lepida parallels the replacement of Sylvilagus nuttallii by S. audubonii and Lepus townsendii by L. californicus (discussed above) and helps confirm that a warming trend took place in the recent past. Although N. cinerea still lives in the area, it seems to have been driven to higher elevations in the Snake Range.

Material--Palate without teeth (BYUVP 7383), partial right dentary with anterior 2/3 of M/1 (BYUVP 7391).

Discussion--Ondatra is easily distinguished from other microtine rodents by its large size combined with rooted molars. O. zibethicus is now considered the only extant species of Ondatra (Hall 1981), and the Crystal Ball Cave specimens are indistinguishable from this species. A number of fossil species have been named, but there is considerable confusion about their status (Miller 1976). All the extinct species considered valid by Semken (1966) and Nelson and Semken (1970) are smaller than O. zibethicus. The Crystal Ball Cave dentary is almost as large as the largest O. zibethicus to which it was compared. The M/1 is 7.9 mm long and 2.5 mm wide which best matches measurements taken from Wisconsinan-age O. zibethicus specimens (Nelson and Semken 1970). The palate is slightly smaller than the mean of O. zibethicus but well within its range of variation.

O. zibethicus is not presently found around Gandy but occurs as close as 100 miles (170 km) to the north, east, and south (Hall 1981). Since Ondatra is a reliable indicator of permanent water (Nelson and Semken 1970), the retreat of Lake Bonneville and loss of perennial streams in the area probably lead to its extirpation from the Snake Range.

Material--Two left M3/'s (BYUVP 6940 and 6981), 7 right M/3's (BYUVP 8220-8226), 15 left M/3's (BYUVP 7002, 8227-8241). Numerous other partial jaws and isolated molars cannot be distinguished from Lagurus but lack characters that would assign them to other species of Microtus, some of which are likely Microtus since over a third of the microtine M/3's belong to Microtus. Among these are a partial skull with both M1/,2/ and the posterior incisive foramina (BYUVP 8285) and a right maxilla with M1/,2/ (BYUVP 6943).

Discussion--Microtus differs from Lagurus, the only other microtine of its size with rootless molars, in having 3 transverse loops on M/3 rather than 4 prisms, some of which are closed triangles, and in having a large semicircular posterior loop on M3/ rather than a simple elongate loop (Hall 1981). The 2 M3/'s and 22 M/3's from Crystal Ball Cave listed above clearly match Microtus in this respect. There are many species of Microtus, some of which have distinct dental characteristics and some of which do not.

The only two species of Microtus now inhabiting the Snake Range are M. longicaudus and M. montanus, and no character could be found to distinguish them dentally. The incisive foramina of M. longicaudus are not constricted posteriorly as are those of M. montanus, but they differ from those of Lagurus only in having slightly curved rather than straight external margins. Since only the posterior end of the incisive foramina are found on skulls that could be Microtus from Crystal Ball Cave, skulls of M. longicaudus in the collection are indistinguishable from Lagurus. Of 13 skulls containing incisive foramina which may be Microtus, 2 have constricted incisive foramina as in M. montanus (listed below), and 11 compare well with M. longicaudus and Lagurus.

Three other species of Microtus presently occur in Utah but not in the Snake Range: M. pennsylvanicus and M. richardsonii in the central mountain ranges and M. mexicanus in the southwestern corner of the state. M. pennsylvanicus has a posterior loop on M2/ not found in other species, and this character was only found on one specimen (listed below). M. richardsonii is distinctly larger than the other species described here, and none of the microtine specimens from Crystal Ball Cave are large enough to compare with it. M. mexicanus is dentally indistinguishable from M. montanus and M. longicaudus, and its incisive foramina are identical to Lagurus and similar to M. longicaudus.

The specimens listed above are identical to Recent specimens of M. longicaudus, M. mexicanus, and more distant ranging species. But since M. longicaudus presently occurs at Crystal Ball Cave, while M. mexicanus occurs more than 250 miles (400 km) to the southeast (Hall 1981), and because the general trend in the region is for range boundaries to be migrating northward, the Crystal Ball Cave specimens (except the few discussed below) are referred to M. longicaudus.

Material--Two partial palates without teeth which include the posterior end of the incisive foramina (BYUVP 8218, 8219).

Discussion--M. montanus is the only microtine of its size presently occurring in Utah or Nevada with incisive foramina that abruptly constrict posteriorly and are narrower posteriorly than anteriorly. The posterior ends of the incisive foramina in these two specimens are too narrow to be M. longicaudus, M. pennsylvanicus, M. mexicanus, or Lagurus curtatus. M. townsendii and M. oregoni also have incisive foramina like M. montanus, but they both occur only along the pacific coast from northern California to southern British Columbia. Since M. montanus presently occurs in the Snake Range (Hall 1981), the Crystal Ball Cave specimens are referred to it. M. montanus tends to occur at higher elevations than other species of Microtus in Utah (Durrant 1952), so its presence in the assemblage suggests that conditions at the cave during the Late Pleistocene may have been like those of higher elevations in the Snake Range now.

Material--Partial skull with right M1/,2/ (BYUVP 6973).

Discussion--M. pennsylvanicus is unique in having a rounded posterior loop behind the 4 closed angular sections of M2/. This single specimen from the assemblage has this posterior loop, but the loop is not completely closed off from the preceding triangle as in the Recent specimens to which it was compared. Since the distinguishing character is not fully developed, the specimen is only referred to M. pennsylvanicus. This species is not presently found in the Snake Range, but it occurs 114 miles (190 km) east of Crystal Ball Cave in the mountains of central Utah and is a northern species (Hall 1981). Considering the climatic shifts since the recession of Lake Bonneville, it is not unlikely that it could have inhabited the Snake Range during the Late Pleistocene.

Material--Skull with right I/, M2/,3/, left I/, M1/,2/ (BYUVP 6899), left dentary with M/1,/2,/3 (BYUVP 6977), left dentary with M/2,/3 (BYUVP 6986), 28 right M/3's (BYUVP 8242-8270), 9 left M/3's (BYUVP 8271-8280). Numerous partial jaws and isolated molars may be L. curtatus but cannot be distinguished from Microtus longicaudus (as discussed above).

Discussion--The differences between Lagurus and Microtus are discussed above. L. curtatus, the only North American species of Lagurus, is distinguished from Old World representatives by having 4 instead of 5 closed triangles on M/3 and cement present in the reentrant angles of the molars (Hall 1981). This species presently occurs in the Snake Range and northward into Canada (Hall 1981). Lagurus specimens are nearly twice as abundant as those of Microtus in the assemblage, but since no information on their Recent relative abundance or habitat differences could be found, it is difficult to know the reason for this.

Discussion--Identification of these canid fossils is based on their size, being substantially larger than C. latrans but considerably less robust than C. dirus. They do, however, fit within the large size range of C. familiaris, so the identification must be tentative. Goodrich (1965) reported C. lupus from Smith Creek Cave but did not describe the material. C. lupus has been reported living in the Snake Range in Recent times (Hall 1981) although Man has now reduced its range and numbers considerably.

Material--Skull with right P1/,2/,4/, left P4/, M2/ (BYUVP 8299), posterior portion of right maxilla with M1/,2/ (BYUVP 7466), partial left maxilla with M1/,2/ (BYUVP 7467), two right C1/'s (BYUVP 7468, 7470), left C1/ (BYUVP 7469), right P4/ (BYUVP 7474), left P4/ (BYUVP 7471), right dentary with M/2 (BYUVP 7461), posterior portion of right dentary with P/4, M/1,/2 (BYUVP 7463), left dentary with M/1,/2 (BYUVP 7464), anterior portion of left dentary with M/1,/2 (BYUVP 7462), right P/4 (BYUVP 7475), left P/4 (BYUVP 7472). An anterior fragment of a right dentary without teeth (BYUVP 7465) and an anterior fragment of a left dentary without teeth (BYUVP 7476) also compare favorably with this species.

Discussion--Vulpes is distinguished from Urocyon by the configuration of the crest on top of the skull and the lack of a prominent "step" on the posteroventral margin of the dentary. The ventral margin of the dentary of Vulpes curves upward posteriorly beginning at the posterior end of the tooth row while in Urocyon it remains uncurved well behind the tooth row all the way to the "step." Urocyon, which ranges from the cave site southward and throughout North America, is intermediate in size between V. vulpes and V. velox. Four of the Crystal Ball Cave specimens include the posterior dentary and lack the "step" characteristic of Urocyon, and all the Crystal Ball Cave specimens are larger than the largest Urocyon specimen examined but compare well in size and shape to V. vulpes.

V. vulpes does not presently occur around Crystal Ball Cave but V. velox and U. cinereoargenteus do (J. C. Bates 1983 personal communication, Hall 1981). The presence of the more northern V. vulpes but not the more southern U. cinereoargenteus in the cave assemblage represents a northward shift of the boundary between these two species. The ranges of V. vulpes and U. cinereoargenteus do overlap to a degree now, but in the western United States the overlap is not great, and where it does occur V. vulpes favors the higher elevations and U. cinereoargenteus the lower elevations (Hall 1981). Based on range maps in Hall (1981), the range of V. vulpes in the western United States is quite scattered, suggesting that it is relectual and that this species is diminishing in numbers there. U. cinereoargenteus has a distinct northern boundary across Utah and Nevada with no remnant populations, suggesting that this species has been making a northward invasion. The Crystal Ball Cave assemblage confirms that U. cinereoargenteus has been expanding its range at the expense of V. vulpes.

Material--Left dentary with P/3 and partial M/1 (BYUVP 7477), posterior portion of left M/1 (BYUVP 7479). A partial left dentary with M/2 (BYUVP 7478) also compares favorably with this species.

Discussion--V. velox and V. macrotis are now considered conspecific (Hall 1981). The dentary (BYUVP 7477) lacks the "step" of Urocyon, and the M/1 lacks a small cuspule found on the posterolabial margin of the main cusp of all the Urocyon specimens but none of the Vulpes specimens examined. The Crystal Ball Cave specimens listed above are smaller than U. cinereoargenteus but may be similar in size to the smaller U. littoralis, which is known only from islands along the coast of southern California (Miller 1971).

Since V. velox still lives around Crystal Ball Cave (J. C. Bates 1983 personal communication), its presence in the assemblage is not surprising. Its low frequency compared to the now-extirpated V. vulpes suggests that it may not have always inhabited the area, may have inhabited it in much smaller numbers, or may have had a different microhabitat causing it to frequent the cave area less than V. vulpes. The ranges of V. vulpes and V. velox presently overlap to a degree, especially in the midwest, but in the western United States this overlap is small (Hall 1981). Although V. velox occurs in the Snake Range now, it is a more southern species than V. vulpes so its northern range extensions may be of Recent age.

Material--Left M1/ (BYUVP 7487), right dentary with P/2,/3, M/1,/2 (BYUVP 7483), partial right dentary without teeth (BYUVP 7484), left dentary with M/1,/2 (BYUVP 7488), left dentary with M/1 (BYUVP 7485), partial left dentary with M/1,/2 (BYUVP 7486).

Discussion--All these Mustela specimens compare best in size with M. frenata, which presently lives around Crystal Ball Cave. The size range of M. frenata is overlapped by the smaller but more variable M. erminea (Kurten and Anderson 1980) which also ranges in the cave area (Hall 1981). The specimens could belong to M. erminea since this species is dentally similar to M. frenata. M. rixosa is always smaller and M. nigripes and M. vison are always considerably larger than the Crystal Ball Cave specimens. M. frenata was the most abundant mustelid at Smith Creek Cave, but M. erminea was also present (Miller 1979). Since all the Crystal Ball Cave specimens fall in the narrow size range of M. frenata, they are referred to this species.

Material--Left M1/ (BYUVP 7482). A juvenile left dentary without teeth (BYUVP 7491) also compares well with this species.

Discussion--This isolated tooth was compared with a variety of Recent mustelids and other small carnivores and found most similar to M. vison. This is North America's largest extant species of Mustela (although the extinct sea mink, M. macrodon, was larger) and is distinctly larger than but similar in shape to M. frenata (described above). M. vison was recovered from Smith Creek Cave (Miller 1979) and presently occurs 100 miles (160 km) north and east of Crystal Ball Cave (Hall 1981), but it does not currently live in the Snake Range. This species requires lakes or streams to survive (Hall 1946), so its extirpated nature in the Snake Range may have been due to the recession of Lake Bonneville and/or loss of perennial streams in the area at the end of the Pleistocene.

M. vison is sometimes confused with M. nigripes since both are of similar size (Kurten and Anderson 1980), and no distinction in isolated molars could be found in the literature. M. nigripes is currently endangered, and no comparative material was available. It has never been reported from western Utah or Nevada, so the Crystal Ball Cave specimens are referred to M. vison, which is known to have lived in the area.

Material--Left dentary with M/1 (BYUVP 7480), left M/1 (BYUVP 7481). The anterior portion of a right dentary without teeth (BYUVP 7489) and the posterior portion of a right dentary without teeth (7523) probably also belong to this taxon.

Discussion--Anderson (1970), in her systematic review of the genus Martes, considered M. nobilis (found in four caves in Wyoming, Idaho, and northern California) to be a distinct species from M. americana. Of the two, M. nobilis is larger, and its lower carnasial has a relatively shorter trigonid. The lower canines of M. nobilis sometimes have faint grooves on the external surface not found in M. americana (Anderson 1970). The only other species of Martes presently living in Utah is M. pennanti, the fisher. It is considerably larger than M. americana, M. nobilis, and the Crystal Ball Cave specimens. Neither M. americana nor M. pennanti currently live in the Snake Range, but both occur in the mountains of central Utah and northward.

BYUVP 7480 is as large as the largest M. americana specimen to which it was compared, and judging from the incisor socket, its incisor was slightly larger. The other specimens are the same size as most Recent M. americana specimens. Both lower carnasials match perfectly in shape with M. americana and do not show a relatively shorter trigonid, so they are assigned to M. americana. A right M1/ of M. nobilis was recovered from Smith Creek Cave (Miller 1979) but M. americana has never been reported. The ecological and chronological separation of these two species in the Snake Range is, therefore, problematic. Brown (1971) listed M. americana as one of eight species of boreal mammals that presently range in the Sierra Nevada and the Rocky Mountains but on none of the isolated Great Basin ranges in between. This citing demonstrates that M. americana did range at least as far east in the Great Basin as the Snake Range before becoming extirpated.

Type--Anterior portion of skull including a complete palate except the most posterior (smallest) socket of each M1/ and extending posteriorly to include the entire anterior wall of the braincase (BYUVP 7490, see figure 5). Only the right P4/ was found in situ, but a right and left M1/ (previously cataloged as BYUVP 7492 and 8298 respectively) fit perfectly into the sockets of the type specimen where they have been permanently mounted. The type specimen is of a young adult based on complete fusion of the premaxillae, maxillae, nasals, and frontals and on lack of significant tooth wear. Both the skull and isolated M1/'s were recovered from Site 1, Channel A, Crystal Ball Cave, Millard County, Utah (see figures 1 and 3) by Wade E. Miller and party on March 19, 1977. The type specimen is housed at the Brigham Young University Vertebrate Paleontology Laboratory.

Diagnosis--Brachyprotoma brevimala has a short face and a maxillary tooth formula of I3/-3/, C1/-1/, P2/-2/, M1/-1/ as in B. obtusata. Face and maxillary dental measurements average 15% smaller than those of B. obtusata. B. brevimala is distinguished from B. obtusata by P4/ being transversely narrower and having a more posteriorly directed lingual cusp, and by M1/ being more reduced and distinctly shorter anteroposteriorly. In other known characters B. brevimala is equivalent to B. obtusata. B. brevimala has the most reduced P4/ and M1/ of any known skunk.

Description--The maxillary dental formula of I3/-3/, C1/-1/, P2/-2/, M1/-1/ is known among the mustelids only in two genera of skunks, Conepatus and Brachyprotoma (although I found one abnormal Recent Spilogale putorius specimen with this formula). The Crystal Ball Cave specimen is clearly a skunk (subfamily Mephitinae) based on the presence of only 2 pairs of upper premolars (mephitines have 2 or 3, all other mustelids have 3 or 4), the small size (only the subfamilies Mustelinae and Mephitinae have such small adult individuals), the lingual cusp of P4/ extending from the middle of the tooth (as opposed to the more anterior extension in the mustelines), M1/ being anteroposteriorly shorter labially than lingually (mustelines have the opposite condition), and the internal nares extending almost as far anteriorly as the posterior end of the tooth row (they are much more posterior in mustelines).

Compared to extant mephitines the Crystal Ball Cave specimen represents an individual of similar size to Spilogale but much smaller than Conepatus and Mephitis. The palate is shorter and wider than that of Spilogale putorius, but the interorbital breadth shows that the type specimen represents a larger individual than the average S. putorius. The P4/ is similar to Spilogale, differing only in having the lingual cusp slightly more posterior, but it is proportionally much narrower than the P4/ of Mephitis and Conepatus. The M1/'s are proximo-distally shorter than any of the living mephitines (especially Conepatus and Mephitis that have large square M1/'s) and are closest to Spilogale in shape and cusp pattern. The rostrum of the type specimen is shorter than that of Spilogale, matching that of Conepatus in proportions. The external nare is steep as in Conepatus, but it is relatively small and round as in Spilogale. Both infraorbital canals are single rather than double or triple, a species-diagnostic character in Conepatus (Hall 1981) but variable in Mephitis and Spilogale.

In addition to the three extant genera, three fossil genera have been named: Buisnictis, Brachyprotoma, and Osmotherium (Kurten and Anderson 1980). Osmotherium can be ruled out since it is large and very similar to Mephitis (Kurten and Anderson 1980), the living skunk genus that is most distinct from the Crystal Ball Cave specimen. Both Buisnictis and Brachyprotoma are small and have proportionally short jaws like the Crystal Ball Cave specimen. Buisnictis has been recovered from Late Pliocene deposits of southwestern Idaho (Bjork 1970) and Middle Pliocene to Early Pleistocene deposits of Kansas and Oklahoma (Hibbard 1941, 1950, 1954), but it has no record in the Late Pleistocene or Recent. Buisnictis has a short jaw with crowded premolars, but it differs from the Crystal Ball Cave specimen in having 3 pairs of upper premolars instead of 2 (Kurten and Anderson 1980). Based on an illustration by Hibbard (1954), the P4/ of Buisnictis has its lingual cusp extending from the anterior part of the tooth as in the mustelines, and the M1/ is distinctly longer than that of the Crystal Ball Cave specimen. These morphologic and age differences show that the Crystal Ball Cave specimen is distinct from Buisnictis.

The Crystal Ball Cave specimen matches the genus Brachyprotoma in having short jaws, only 2 pairs of upper premolars, P4/ and M1/ similar in shape and cusp pattern to Spilogale, and in the age of deposits in which they have been recovered. Until recently Brachyprotoma was only known from a few early Pleistocene to early Recent age ave deposits in the eastern United States. But during the period of this study P. M. Youngman (1984 personal communication) recovered several Brachyprotoma specimens from two cave deposits in the Yukon Territory of Canada. Although no previous Brachyprotoma specimens have been reported closer than 1130 miles (1880 km) from Crystal Ball Cave, morphology clearly allies the Crystal Ball Cave specimen with this genus. But there are specific differences between the Crystal Ball Cave specimen and other skulls which have been assigned to the genus Brachyprotoma. To test the amount of variation to be expected within a species of skunk, I measured 73 specimens of Recent Spilogale putorius, 60 from the Harvard University Museum of Comparative Zoology and 13 from the Brigham Young University Monte L. Bean Museum. Spilogale putorius makes a good standard for the expected individual variation in species of Brachyprotoma, both because Spilogale is probably the most closely related extant genus to Brachyprotoma and because S. putorius borders on being divisible into multiple species (although most workers presently consider it a single species). Based on the great amount of variation seen between the Crystal Ball Cave specimen and other skulls assigned to the genus Brachyprotoma compared with the amount of variation seen among individuals of S. putorius, I believe the Crystal Ball Cave specimen warrants the status of a new species.

The B. brevimala type is smaller than specimens of B. obtusata in most measured characters, averaging about 15% smaller (see table 3). The greatest differences occur in P4/ and M1/, which are the most varied maxillary teeth between skunk taxa. The mean length of P4/ in B. obtusata is only 7% greater than in B. brevimala while the mean width is 22% greater. The lingual cusp of P4/ in B. brevimala also points more posteriorly than in B. obtusata, being nearer M1/ at its lingual tip rather than closer at its base or parallel as in B. obtusata. The M1/ of B. obtusata is 16% transversely wider on the average, but the labial anteroposterior length is 30% greater and the lingual anteroposterior length is 59% greater than in the B. brevimala type on the average. Since there is only minor variation in these characters among B. obtusata skulls (see table 3) but distinct difference between them and the Crystal Ball Cave specimen, and because the differences between the B. brevimala type and specimens of B. obtusata are far greater than would be expected within a species (based on the variation found among 73 individuals of Spilogale putorius, the most closely related extant species), erection of a new species for the Crystal Ball Cave specimen is clearly justified.

Discussion--Brachyprotoma specimens have been previously recovered from the following deposits of Early Pleistocene to Early recent age: Port Kennedy Cave and Frankstown Cave, Pennsylvania (Cope 1899, Peterson 1926), Cumberland Cave, Maryland (Gidley and Gazin 1938), Crankshaft Cave and Brynjulfson Cave, Missouri (Oesch 1967, Parmalee and Oesch 1972, Parmalee et al. 1969), Connard Fissure, Arkansas (Brown 1908), and two caves in northern Yukon Territory, Canada (P. M. Youngman 1984 personal communication). Most of these specimens are lower jaws, and the only 7 skulls (or skull fragments) that have been previously reported are Carnegie Museum 11057A and 20233, American Museum of Natural History 11772 and 12426, U.S. National Museum 8155 and 11960, and a specimen identified as Carnegie Museum 308 by Oesch (1967) but which does not correspond to that number in the Carnegie Museum catalogs (M. R. Dawson 1984 personal communication). Parmalee et al. (1969) illustrated this latter specimen but did not identify it by catalog number.

Cope (1899) named Mephitis (Spilogale) obtusatus, for a single small dentary from Port Kennedy Cave, but E. D. Cope died before the completion of this paper, and a footnote stated that "none of the specimens labelled by Prof. Cope bear this name." Brown (1908) named the genus Brachyprotoma, and he considered M. obtusatus to belong to this genus as well as M. fossidens and M. leptops, two species named by Cope previous to the naming of M. obtusatus. From Connard Fissure Brown (1908) reported B. fossidens, B. leptops, and B. obtusatus based on dentaries, and he named B. pristina based on two partial skulls and three dentaries (the skull cataloged as American Museum of Natural History 12426 is the type for the genus and species) and B. spelaea based on one dentary. The dentaries Brown (1908) identified as B. fossidens and B. leptops are far too large to belong to the same genus as the small specimens he identified as B. obtusatus, B. pristina, and B. spelaea, and no one since has considered these two species as belonging to the genus Brachyprotoma. Later Hay (1923) named B. putorius from Frankstown Cave. Peterson (1926) identified material from Frankstown Cave as B. obtusata, correcting the specific ending to match the gender of the generic name.

The naming of multiple species of Brachyprotoma in the early publications listed above has been widely criticized by later workers because the variation among specimens is less than that seen within living species. Hall (1936) and Kurten and Anderson (1980) considered the genus Brachyprotoma to be clearly monotypic with the only valid species being B. obtusata, the earliest named species that can be applied to the genus. The Brachyprotoma skull from Crystal Ball Cave is the first specimen of Brachyprotoma distinct enough from B. obtusata to warrant the erection of an additional species of this genus.

Concerning the paleoecology of Brachyprotoma, Kurten and Anderson (1980) stated that this genus has always been associated with boreal faunas although other skunk genera were also recognized at each site. This matches the "more boreal than present" nature of the Crystal Ball Cave assemblage and suggests that the post-Pleistocene climatic shift may have lead, directly or indirectly, to the extinction of Brachyprotoma. Since fossils of Brachyprotoma are only found in a few deposits and even then are few in number, this genus probably never had a high density of individuals in life.

The Brachyprotoma brevimala type was first misidentified as Spilogale (Heaton 1984), the most similar living genus. Miller (1982) reported cf. Spilogale from Crystal Ball Cave, possibly based on this same specimen. Mephitis was also mentioned in my preliminary report (Heaton 1984), but further examination proved that the anterior right dentary (BYUVP 7489) upon which the identification was based was equally referable to Martes americana, which is represented by additional material. Although both Mephitis mephitis and Spilogale putorius (=gracilis) now inhabit the Snake Range (Hall 1981), and Spilogale has been recovered from deposits over 12,000 years old in Smith Creek Cave (Mead et al. 1982), their presence is unconfirmed in the Crystal Ball Cave assemblage.

Since Brachyprotoma seems to have lived contemporaneously with other skunk genera, it is interesting to speculate about how their niches varied. All living skunks tend to be nocturnal and omnivorous, so they are rarely tied to specific foods or habitats. Minor niche differences do occur between living North American genera: Spilogale is the most carnivorous, Mephitis the most herbivorous, and Conepatus the most insectivorous. Spilogale has narrow sharp teeth, Conepatus at the other extreme has very broad teeth, and Mephitis is intermediate but has the longest tooth rows. Brachyprotoma (especially B. brevimala) has pushed the narrowing of the teeth seen in Spilogale to an extreme, converging on the carnivorous genus Mustela. This suggests that Brachyprotoma was more carnivorous than any of the living skunks.

Why Brachyprotoma lost P2/ and shortened its tooth rows, paralleling the genus Conepatus, is a mystery. Members of the genus Mustela have longer tooth rows than skunks, so in that respect Brachyprotoma diverged from Mustela. Brachyprotoma was trending in a direction that is difficult to explain. Brachyprotoma also did not survive the post-Pleistocene changes as did the aforementioned genera (although some species were lost and ranges altered). I propose that these two facts are correlated. Brachyprotoma was probably adapting to a specialized niche that existed during the Pleistocene but disappeared during the Recent. I also propose that this specialization was a feeding habit and/or preference for a particular prey item since the specializations discussed are all dental. No postcranial material has been reported to document additional specializations, and skunks' most diagnostic characters, scent glands and color patterns, are in the soft anatomy which is obviously unavailable. With such limited data (about 27 specimens from 9 sites), further speculation would be unwarranted. All that can be concluded is that Brachyprotoma was restricted to boreal conditions, was widespread in North America, was probably low in density, and did not survive the post-Pleistocene changes.

The evolution of the genus Brachyprotoma has been discussed by Kurten and Anderson (1980). They stated that it seems most closely related to Spilogale, but both were probably derived from the Mio-Pliocene genus Promephitis. No intermediate forms are available to show the exact phylogeny, however. Some speculation can be made about the relationship of B. brevimala to B. obtusata. B. brevimala has gone to a greater extreme in the characters that differentiate Brachyprotoma from other skunks (shorter face and narrower teeth) and is therefore more specialized. Since specialists almost always evolve from generalists, B. brevimala probably evolved from B. obtusata. The fact that B. obtusata has been found in deposits from early Pleistocene to early Recent age (Kurten and Anderson 1980) while B. brevimala is known only from a late Pleistocene to Recent deposit also supports this conclusion.

Material--Partial left ectocuneiform (BYUVP 7530), claw (BYUVP 7497). Miller (1982) reported cf. Smilodon from Crystal Ball Cave based on a single vertebra (W. E. Miller 1983 personal communication), but this specimen is apparently lost (possibly due to an explosion that affected the collection).

Discussion--The ectocuneiform is dense, worn, and coated with a calcite crust. The claw is missing the outer plates but is otherwise in good condition. The specimens were compared with Smilodon and Felis atrox, the only two late Pleistocene cats large enough to be considered, and both compare best with Smilodon (W. E. Miller 1984 personal communication). The ectocuneiform was previously referred to Panthera atrox (Heaton 1984), but comparison with actual specimens rather than casts shows that it was clearly Smilodon. The only previous citing of Smilodon in Utah is from the Silver Creek fauna of north-central Utah (Miller 1976), but it has been found in Pleistocene assemblages throughout North America.

Kurten and Anderson (1980) considered S. fatalis to be the only valid species of late Pleistocene Smilodon in North America, but it has been known by many other names. Based on this synonymy, the Crystal Ball Cave specimens are referred to S. fatalis although they are doubtfully species specific.

Material-First right metacarpal (BYUVP 7502), 4 claws (BYUVP 7498-7501).

Discussion-F. concolor is the only cat of its size presently living in North America, but similar-sized species of Acinonyx and Homotherium existed during the Pleistocene. Lynx and other species of Felis (disregarding those often placed in the genus Panthera) are distinctly smaller than F. concolor, and Smilodon, Panthera atrox, and P. onca are distinctly larger. The Crystal Ball Cave specimens were compared with material of Felis species, Acinonyx, and Homotherium at the Los Angeles County Museum and found to match perfectly in size and shape with F. concolor but to clearly differ from the other felids mentioned. F. concolor presently lives throughout the Snake Range (Hall 1981), and J. C. Bates (1983 personal communication) reported a citing in the Snake Valley near Gandy as well as many higher in the mountains.

Material--Right C1/ (BYUVP 7494), right P/4 (BYUVP 7496). The anterior portion of a right maxilla without teeth (BYUVP 7495) is probably also referable to Lynx.

Discussion--L. rufus currently inhabits the area of Crystal Ball Cave (J. C. Bates 1983 personal communication) while L. canadensis ranges only as close as central Utah and northward and prefers colder climates (Hall 1981). L. canadensis is slightly larger than L. rufus and has considerably larger feet (Ingles 1965). The specimens recovered fall in the size range of both L. rufus and L. canadensis, but they tend to be closer in size to L. rufus. None of the claws recovered could be referred to this genus, so the difference in foot size was not helpful. Since L. rufus presently lives around the cave, the specimens are referred to it.