harvard university Library of the Museum of Comparative Zoology BIOSYSTEMATICS OF THE YELLOW-FACED POCKET GOPHER, CRATOGEOMYS CASTANOPS (RODENTIA: GEOMYIDAE) IN THE UNITED STATES Robert R. Hollander Special Publications, The Museum TEXAS TECH UNIVERSITY NUMBER 33 BIOSYSTEMATICS OF THE YELLOW-FACED POCKET GOPHER, CRATOGEOMYS CASTANOPS (RODENTIA: GEOMYIDAE) IN THE UNITED STATES Robert R. Hollander Texas Tech University Press 1990 Special Publications, The Museum Texas Tech University Number 33 Series Editor J. Knox Jones, Jr. Published 15 August 1990 Copyright 1990 Texas Tech University Press All rights reserved. No portion of this book may be reproduced in any form or by any means, including electronic storage and retrieval systems, except by explicit, prior written permission of the publisher. Special Publications of The Museum are numbered serially and published on an irregular basis. Institutions interested in exchanging publications should address the Exchange Librarian at Texas Tech University. ISSN 0149-1768 ISBN 0-89672-229-5 Texas Tech University Press Lubbock, Texas CONTENTS Introduction . Historical Taxonomy . Fossil Record . Natural History . Reproduction . Molt . Habitat . Material and Methods .... Nongeogrphic Variation . . . Geographic Variation . Accounts of Subspecies .... Discussion . Biogeography and Evolution Acknowledgments . Literature Cited . . 5 . 5 . 7 . 8 9 . 8 9 .10 . 14 .24 . 36 . 56 56 .59 . 60 Introduction The problem addressed herein is the systematic and evolutionary relation¬ ships, from a morphometric perspective, of currently recognized races ol Cratogeomys in the United States. On the basis ol published findings to the end of 1987, nine subspecies of C. castanops currently are recognized as occurring in the United States. Eight ol these occur in Texas, with the possibility of the ninth occurring in the extreme northwestern corner ol the Texas Panhandle. This species also occurs in Colorado, Kansas, New Mexico, and Oklahoma, and southward into Mexico. Since Russell’s (1968/?) revision of the genus Pappogeomys (sensu lato), no taxonomic study has been done concerning more than two adjacent subspecies. Several criticisms ol Russell’s (1968/?) work can be made: his analysis was not of a statistical nature; specimens from many intermediate geographic areas were not available to him, and, in fact, his study was based on relatively few specimens from the overall distribution of the species. Moreover, Russell (1968/?) discussed two subspecies as “ . . . occurring sympatrically with no apparent intergradation. . .” in the Guadalupe Mountains ol Texas. Schmidly (1977) also reported two races from another locality in the Trans-Pecos region of Texas. Such situations likely do not exist unless more than one species is involved. Historical Taxonomy The genus Cratogeomys , which contains 10 species (Lee and Baker, 1987), occurs from southeastern Colorado south to southern Mexico (Hall, 1981). The only species that occurs north of Mexico isC. castanops. Baird (1852) described castanops (in the genus Pseudostoma ), with type locality on the prairie road to Bent’s Fort, near what is now the town of Las Animas, Bent County, Colorado. Later in the same year, Le Conte (1852) placed this taxon in the genus Geomys. Baird (1855) described Geomys clarkii as a species closely related to castanops from Presidio del Norte along the Rio Grande in Chihuahua. Coues (1875) later placed G. clarkii in synonymy with G. castanops. Merriam (1895), in his monographic revision of the family Geomyidae, described the genera Pappogeomys and Cratogeomys with G. merriami as the type of Cratogeomys. He assigned specimens of castanops from the United States to this new genus as well. He also agreed with Coues (1875) as to the status of clarkii and described an additional subspecies, Cratogeomys castanops goldmani , from central Mexico. Merriam (1895), however, examined only 30 specimens of the genus Cratogeomys from the United States. Nearly 40 years elapsed before the next revision of the genus was undertaken (Nelson and Goldman, 1934). In that period of time, the systematic arrangement of Merriam (1895) had remained unchanged. Nelson and Goldman (1934), based on many additional specimens (they listed 68 from the United States as examined plus those of C. c. castanops , 5 6 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY which they did not list), resurrected clarkii as a subspecies of castanops, and described 13 additional races of that species. Four of these were from the United States, bringing the number of recognized races from north of the Rio Grande to Five (they restricted the distribution of C. c. clarkii to south of the river). No additional taxa of C. castanops from the United States were described until Russell (19686) revLed the genus Pappogeomys, within which he recognized Cratogeomys as a subgenus. Until that time, most authors, such as Simpson (1945) and Wood (1955), followed Merriam (1895) in recognizing Cratogeomys as a genus distinct from Pappogeomys. Russell (19686 ) examined a total of 448 specimens of castanops from the United States from which he described four additional subspecies. He placed C. c. lacrimalis Nelson and Goldman in synonymy under C. c. perplanus Nelson and Goldman, and allocated specimens from north of the Rio Grande in the Big Bend area of Texas to the subspecies clarkii. This increased the number of recognized races of C. castanops from the United States to nine. Few studies of systematic importance dealing with Cratogeomys have appeared since Russell’s (19686 ) revision. Dowler and Genoways (1979) evaluated geographic variation of these gophers on the Llano Estacado of Texas and New Mexico, and found the subspecies C. c. simulans (Russell) from the eastern Llano to be indistinguishable from the earlier-named C. c. perplanus Nelson and Goldman from the western part of the Llano. In a study of the genic relationships of selected pocket gophers, Honeycutt and Williams (1982) found members of the subgenus Cratogeomys to be clearly distinct from those of the subgenus Pappogeomys , and suggested that the two subgenera were deserving of generic recognition. This systematic arrange¬ ment has been followed recently ( Jones et al., 1986; Lee and Baker, 1987). In a study of the chromosomal relationships of taxa of Cratogeomys , Berry and Baker (1972) found two distinct chromosomal types of castanops— a. northern group with 46 chromosomes and a southern group with 42. Lee and Baker (1987) suggested that the southern races of castanops should be recognized as specifically distinct from those to the north based upon analysis of differentially stained chromosomes. They noted that none of the southern types (with 42 chromosomes) occurs north of 25 degrees N latitude, thus none is found in the United States. Two major distributional records have been reported since Russell’s (19686) study. Birney etal. (1971) reported C. castanops from Kansas. They allocated populations north of the Arkansas River in western Kansas to the nominate subspecies castanops, based upon their cranial measurements. Cleveland (1977) reported a population of C. castanops from the Texas side of the lower Rio Grande Valley near Brownsville. He did not assign these specimens to any subspecies, but noted that the nearest records were from Tamaulipas, Mexico. HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNTI ED STATES 7 Fossil Record In his treatise on the classification of the Geomyinae, Russell (1968a) indicated that Pappogeomys (including Cratogeomys ) was not known from Pleistocene deposits older than Wisconsin times, but noted a pre-Pleistocene (Pliocene) occurrence in the Benson Beds of Arizona. Russell attributed this to a proposed southern distribution ol the genus, probably on the central Mexican Plateau, an area where few early to middle Pleistocene deposits have been found. In outlining the intraspecific population structuic of Cratogeomys castanops, Russell (1969) hypothesized the retreat of the genus from the southwestern United States during the Wisconsin pluvial cycle, with a subsequent postglacial reinvasion of the region. However, when depicting the proposed early Pleistocene geographic ranges of the geomyid genera (Russell, 1968 a : fig. 2), he mapped the genus Zygogeomys as occupying the southwestern United States (most of the area currently occupied by Cratogeomys ), with Pappogeomys (including Cratogeomys ) restricted to the southern Mexican Plateau. Harris (1985) questioned Russell’s (1969) hypothesized retreat of Cratogeomys from the southwestern United States during the Wisconsin by pointing out the occurrence of Cratogeomys at several stadial sites in the Guadalupe Mountains, indicating that populations occurred farther north during the Pleistocene than envisioned by Russell (1968a, 1969). Other remains of Cratogeomys were reported by Rinker (1941) from Meade County, Kansas, from a Recent terrace, and by Gilmore (1947) as common in Quaternary cave deposits near Cuatro Cienegas, Coahuila. Mooser and Dalquest (1975) reported P. cf. castanops remains from Pleistocene (probably Illinoian) deposits in Aguascalientes in central Mexico. Harris (1987) reported Pappogeomys sp. from mid-Wisconsin deposits in Dry Cave, Eddy County, New Mexico. Graham (1987), in a compilation of Quaternary mammalian faunas of the southwestern plains, discussed C. castanops remains from the following Texas localities; Cueva C^uebrada (late Pleistocene, ^ 14,000 years BP); Bonfire Shelter (approximately 10,200 BP); Baker Cave (Level IV, middle Holocene, 3000 to 6000 BP); Devil’s Mouth (undifferen¬ tiated Holocene); Deadman’s Shelter, Canyon City Club Cave (Level 1 to 5); Alibates 28, Roper, and Spring Canyon (all from late Holocene, 300 to 3000 BP). 8 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Natural History Little is known of the natural history of Cratogeomys castanops] most published information is anecdotal. Two recent papers (Davidow-Henry and Jones, 1988; Davidow-Henry et al., 1989) summarized the biology of this species, and are the primary sources for the sections that follow. Reproduction Reproduction in Texas was summarized by Davidow-Henry and Jones (1988). Their data, and the data from Ikenberry (1964) and Smolen et al. (1980), indicate that pregnant females have been taken in every month, with an average of 2.08 fetuses per female. Davidow-Henry and Jones (1988) recorded lactating females from January, February, March, May, July, August, October, and December. Smolen et al. (1980) reported lactating females from April as well. Davidow-Henry and Jones (1988) also reported the occurrence of juve¬ niles in all months except September and December. They concluded that C. castanops is reproductively active throughout the year in Texas; they found no unusually marked peaks of reproductive activity. They agreed with Smolen et al. (1980) that at least some individuals bear multiple litters annually. Northwardly, in Colorado and Kansas, the breeding season of C. castanops probably is limited to the warmer months of the year, but only indirect evidence (Birney et al . , 1971) is available. Molt Davidow-Henry and Jones (1988) described the juvenile pelage of C. castanops as straw-colored or grayish yellow, contrasting markedly with the generally darker pelage of adults. They stated that postjuvenile molt evidently begins on the head and proceeds posteriorly to the level of the eyes and ears while proceeding ventrally over the cheeks and upper throat. They opined that the venter molted rapidly because they observed few animals in the process of ventral molt. From the head region, molt progressed caudally on the dorsum with the middorsal areas molting in advance of the flanks. The areas at the base of the tail and on the posterior flanks were the last regions to molt. Birney et al. (1971) reported distinctive semiannual molts in adult C. castanops in Kansas. They found molt from winter to summer pelage early in spring and from summer to winter pelage in September and October. Ikenberry (1964), however, reported a single extended molting period, beginning in August and continuing through March in adult specimens from Lubbock County, Texas. HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNI TED STATES 9 Habitat Cratogeomys castanops frequently is found in deep, sandy soils that are relatively free of rocks (Schmidly, 1977; Davis, 1940). This pocket gopher generally inhabits valleys and avoids the hard soils ot arid mesas and upper slopes (Bailey, 1905, 1932). Findley (1987) noted that in eastern New Mexico, Cratogeomys was excluded from deep, sandy soils and forced to survive in shallower, rocky soils in the presence of Geomys, a situation similar to that reported in Kansas by Birney et al. (1971). However, in the presence of Thomomys , Cratogeomys seems to exclude the former from the more pliable soils and force these smaller gophers into thin, rocky soils (Findley, 1987, Hollander et al ., 1987). The displacement of Thomomys by Cratogeomys has been documented in Limpia Canyon, Jeff Davis County, Texas (Reichman and Baker, 1972; Williams and Baker, 1976), and is known elsewhere where ranges of the two broadly overlap. Schmidly (1977) discussed shifts in the distributions of Thomomys and Cratogeomys , in an area of the Trans-Pecos, that corresponded to the available moisture, and opined that Cratogeomys seemed to be favored as conditions became more xeric. Moulton et al. (1979) reported an area of sympatry between Cratogeomys and Thomomys in south¬ eastern Colorado. Moulton et al. (1983) discussed the ecological parameters that separated all three genera of gophers in southeastern Colorado. They found that where Cratogeomys and Thomomys occurred sympatrically, the latter had significantly shallower burrows. However, they concluded that the two genera were mutually exclusive competitors and the zone ol sympatry probably represented an area where one species was displacing the other. Where Cratogeomys and Geomys came into contact, they found the latter to occupy disturbed soils, whereas the former was primarily restricted to native shortgrass rangeland. From personal observations in the field, it is apparent that Cratogeomys occupies the most favorable soils that are available in an area. However, it is not restricted to such areas. For example, in the area of Independence Creek, in northern Terrell County, Texas, these gophers are common along the Pecos River and on a private golf course in the creek bottom. Both of these areas have relatively deep, pliable soils, but gophers also are common above the bottom land, in rocky, caliche type soils. A similar situation exists in western Upton and Crane counties and northeastern Pecos County, Texas, where gophers are abundant in the river bottom ol the Pecos but also occur in upland areas characterized by soils that are extremely hard and rocky. It seems apparent from these observations, and those cited above, that Cratogeomys castonops is a generalist and has the ability to survive in many different types ol habitat. 10 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Materials and Methods Three external measurements (total length, length of tail, and length of hind foot) were obtained from original specimen labels. However, variation among methods of individual preparators in measuring these external dimensions rendered them virtually useless for statistical analysis. They were excluded from further analyses in this study, but some external measure¬ ments (mm.) are listed in text. Fifteen cranial and mandibular measurements, similar to those used in other studies of geomyids (Davis and Buechner, 1946; Villa-R. and Hall, 1947; Youngman, 1958; Russell, 1968£; Hendricksen, 1973; Smith et al., 1983), were taken with Fowler digital calipers to the nearest 0.01 millimeters. Description of these measurements, depicted in Figure 1 (letters following measurement names correspond to the letters on the Figures), follows. Condylobasal length (A- A).— Shortest distance from posteriormost projection of occipital condyle to anteriormost projection of premaxilla. Zygomatic breadth (B-B). — Greatest distance parallel to long axis of skull across zygomatic arches. Mastoid breadth (C-C).— Greatest distance parallel to long axis of skull across mastoid process of squamosal bone. Occipital depth (D-D).— Shortest distance perpendicular to long axis of skull from ventralmost portions of auditory bullae to temporal ridge of squamosals. Breadth of rostrum (E-E). — Greatest width across rostrum anterior to zygomatic arches. Length of rostrum (F-F). — Distance from anteriormost projection of nasal to lateral junction of lacrimal and maxilla. Length of nasals (G-G).— Greatest distance from anteriormost to posteriormost point of nasals. Interorbital constriction (H-H). — Least width across frontals. Palatofrontal depth (I-I).— Shortest distance perpendicular to long axis of skull between frontals and palatine bones between molars. Alveolar length of maxillary toothrow ( J-J).— Distance from anterior lip of alveolus of P4 to posterior lip of alveolus of M3. Length of palate (K-K).— Shortest distance from anteriormost point on posterior border of palate to posterior lip of alveolus of incisors. Width of upper incisor (L-L).— Greatest width of incisor immediately distal to alveolus. Alveolar length of mandibular toothrow (M-M). — Distance from anterior lip of alveolus of p4 to posterior lip of alveolus of m3. Depth of ramus (N-N).— Shortest perpendicular distance from angle of mandible to dorsalmost point of coronoid process. Width of lower incisor (O-O).— Greatest width of incisor immediately distal to alveolus. Multivariate tests were employed to detect group differences before any univariate tests were used. Willig*/ al. (1986) and Willig and Owen (1987) have demonstrated that the results of multiple univariate tests (for example, 15 ANOVAs on 15 characters) do not emulate the results of a multivariate test (MANOVA on 15 characters); and when analyzing morphological variation in natural populations, it is a multivariate question that is asked. However, for an alternative view, see Corruccini (1987). All statistical tests used (both multivariate and univariate) are from programs available in the SPSSX statistical package (SPSS Inc., 1986). Specific tests utilized in particular phases of the analyses are discussed in HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 1 1 Fig. 1.— Dorsal, ventral, and lateral views of the skull and dorsal and lateral views of the mandible of an adult female Cratogeomys castanops (TTU 1524) from 8 mi. N Lubbock, Lubbock Co., Texas. Letters correspond to measurements described in text. 12 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY more detail in the following sections on nongeographic and geographic variation. The accounts of the subspecies begin with an abbreviated synonymy that includes: 1) the original description with citation and type locality; 2) citation to the first use of the current name combination (if any); and 3) any junior synonyms in chronological order with their respective citations and type localities. The synonymy is followed by a short section describing the geographic distribution of the subspecies. Next is a morphological descrip¬ tion of the taxon followed by a section on comparisons with adjacent taxa. This is followed by a remarks section and a list of specimens examined. Localities from which specimens were examined are mapped and listed in the specimens examined section following each account. Not all localities are plotted as undue crowding of symbols would have resulted and, for the same reason, some symbols are slightly offset on the figures. Those localities not mapped are in italic type in the lists of specimens examined. Localities for a given taxon are listed alphabetically by state and by county within a state. Within a county, localities are listed from north to south and west to east. In the course of plotting specimen localities, some were found to have specific localities that did not match the county designation on the label. In such cases, it was assumed that the specific locality was the correct site of capture. These specimens are listed under what is believed to be the correct county, with a note as to the county that was recorded on the original specimen label. I am indebted to the individuals from the following institutions, with accompanying acronyms corresponding to those used in the lists of specimens examined in the following accounts, for loan of material or for allowing me access to collections in their care. AMNH— American Museum of Natural History, Sydney Anderson and Guy G. Musser. ASU— Angelo State University, Mark D. Engstrom. CCSU— Corpus Christi State University, Brian R. Chapman. ENMU — Eastern New Mexico University, Antonio L. Gennaro. KU— Museum of Natural History, University of Kansas, Robert M. Timm. MHP— Museum of the High Plains, Ft. Hays State University, Jerry R. Choate. MSB— Museum of Southwest Biology, University of New Mexico, Terry L. Yates. MVZ— Museum of Vertebrate Zoology, University of California Berkley, James L. Patton. MWSU— Midwestern State University, Walter W. Dalquest and Fredrick B. Stangl, Jr. NMSU— New Mexico State University, Charles S. Thaeler, Jr. OMNH— Oklahoma Museum of Natural History, University of Oklahoma, Michael A. Mares. OSU— Oklahoma State University, David Edds. SRSU— Sul Ross State University, James F. Scudday. TAI— Texas A&I University, Allan H. Chaney. TCWC— Texas Cooperative Wildlife Collection, Texas A&M University, David J. Schmidly. TNHC— Texas Natural History Collection, University of Texas Austin, Robert F. Martin. HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 13 Fig. 2.— Lateral view of the lower toothrow of a typical juvenile geomyid showing the relationship of the deciduous premolar and permanent premolar (after Merriam, 1895). TTU— Texas Tech University, Robert J. Baker (specimens for which no acronym is listed are in this collection). TWC- Texas Wesleyan College, Arthur G. Clevland. UCM — University of Colorado Museum, David M. Armstrong. UIMNH— University of Illinois Museum of Natural History, M. Raymond Lee. USNM— National Museum of Natural History, Don E. Wilson and Robert Fisher. UTEP— University of Texas El Paso, Arthur H. Harris. WTSU— West Texas State University, Flavius C. Killebrew. 14 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Nongeographic Variation Specimens of Cratogeomys castanops were allocated to one of five age classes (one through five) based on criteria modified from Merriam (1895) and Russell (1968(6). To document the amount of morphological variation that is attributable to secondary sexual, age, or individual variation, a sample of 184 individuals from Lubbock County, Texas (the largest single sample available), was analyzed. Within this sample, both sexes and all the age classes (except age class one and females of age class five) were well represented (see Table 1 for sample sizes within each class). Juvenile ( age class 1).— Both upper and lower deciduous premolars present (see Fig. 2). Only one specimen from this age class was examined. Merriam (1895) reported that juveniles were extremely rare (he examined only four in the entire family, none of them Cratogeomys ). Russell (1968 6) did not mention any juveniles and Dowler and Genoways (1979) stated that they did not examine any among 473 individuals from the Llano Estacado of Texas and New Mexico. Young (age class 2).— Both upper and lower permanent premolars present. Exoccipital-supraoccipital suture unfused in both sexes. Few sutures of the skull are fused and the bones are quite porous. Skulls appear rounded when viewed laterally. Young individuals evidenced juvenile pelage as described by Davidow-Henry and Jones (1988). Subadult (age class 3).— Supraoccipital-exoccipital suture fused but the basioccipital-basisphenoid suture unfused in both sexes. Temporal ridges present but not in contact with each other in either sex, but more prominent in males. Many subadults (approximately 50 percent) exhibited adult pelage. Adult (age class 4).— Females: temporal ridges in contact or basioccipital- basisphenoid suture fused and obliterated. Only a few individuals that had the temporal ridges in contact did not have a fused basioccipital-basisphenoid suture, but an occasional individual with the suture fused did not have tem¬ poral ridges in contact. Males: temporal ridges in contact but basioccipital- basisphenoid suture unfused. Old adult (age class 5).— Occasional females with a pronounced sagittal crest and overall angular skull. In males, basioccipital-basisphenoid suture fused and obliterated, accompanied by large sagittal crest and much angularity to the skull. Many more old adult males were examined than females. Table 1 provides standard descriptive statistics (for combination of age class and sex) for the Lubbock County sample. A two-way multivariate analysis of variance (MANOVA), with sex and age as the main factors, was performed using program MANOVA from SPSSX (SPSS, Inc., 1986). Highly significant results (P <0.001) were obtained for each main effect, which indicated significant differences between sexes and between at least some of the age classes. The absence of a significant sex by age interaction indicated that males differed from 2 3 4 5 2 3 4 2 3 4 5 2 3 4 2 3 4 5 2 3 4 2 3 4 5 2 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Table 1. — Continued. N Mean SD SE Range 24 9.663 Breadth of Rostrum Males 0.541 0.110 8.82-10.79 17 11.004 0.499 0.121 10.28-12.23 33 12.472 1.295 0.225 10.31-14.72 8 13.839 0.833 0.295 12.60-15.31 22 9.527 Females 0.835 0.178 7.67-10.84 30 10.579 0.625 0.114 9.13-11.94 49 11.322 0.491 0.070 10.35-12.64 24 19.018 Length of Rostrum Males 1.353 0.276 16.76-21.87 17 22.721 0.899 0.218 21.43-24.68 32 24.885 2.056 0.363 21.11-27.63 8 27.543 1.009 0.357 25.95-28.90 21 18.601 Females 1.708 0.373 14.99-20.90 29 21.465 1.083 0.201 18.88-23.85 46 22.921 1.033 0.152 21.15-24.86 Length of Nasals Males 24 15.620 1.262 0.258 13.07-18.17 17 18.975 1.005 0.244 17.02-20.51 32 21.041 1.890 0.334 17.18-24.04 8 23.595 1.037 0.367 22.20-25.23 Females 21 15.313 1.586 0.346 12.08-17.06 29 18.105 1.131 0.210 15.08-20.26 46 19.390 0.864 0.127 17.84-21.09 Interorbital Constriction Males 24 6.752 0.282 0.058 6.23-7.26 17 6.788 0.554 0.134 5.82-7.85 33 6.899 0.496 0.086 6.11-7.95 8 6.515 0.340 0.120 6.07-7.18 Females 22 6.592 0.339 0.072 5.69-7.30 30 6.835 0.390 0.071 6.24-7.85 49 6.825 0.376 0.054 6.15-7.66 HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 17 Table 1. — Continued. Group N Mean SD SE Range 2 24 17.728 Palatofrontal Depth Males 0.850 0.174 16.19-19.16 3 17 19.772 0.603 0.146 18.96-20.98 4 33 21.502 1.571 0.273 18.09-24.03 5 8 23.649 0.882 0.312 22.42-24.97 2 22 17.406 Females 1.226 0.262 14.84-18.87 3 30 18.986 0.703 0.128 17.84-20.65 4 49 20.114 0.815 0.117 18.25-21.76 2 24 Length of Maxillary Toothrow Males 9.025 0.486 0.099 8.27-10.10 3 17 9.805 0.452 0.110 9.13-10.66 4 33 10.322 0.468 0.081 9.69-11.75 5 8 10.839 0.395 0.140 10.30-11.39 2 22 8.953 Females 0.568 0.121 7.70-10.05 3 30 9.682 0.396 0.072 9.00-10.53 4 49 9.992 0.416 0.059 9.02-11.39 2 24 23.793 Length of Palate Males 1.649 0.337 20.79-26.73 3 17 28.299 1.030 0.250 27.10-30.69 4 32 31.278 2.200 0.389 27.40-35.12 5 8 34.028 1.233 0.436 32.47-36.22 2 22 23.442 Females 2.314 0.493 18.52-26.98 3 30 27.072 1.281 0.234 24.07-29.32 4 49 28.875 1.182 0.169 26.62-31 .72 2 24 1 2.454 Width of Upper Incisor Males 0.215 0.044 2.08-2.96 3 17 2.991 0.108 0.026 2.79-3.17 4 33 3.393 0.289 0.050 2.89-3.98 5 8 3.678 0.198 0.070 3.37-3.93 2 22 2.468 Females 0.285 0.061 1.84-3.03 3 30 2.878 0.159 0.029 2.50-3.21 4 48 3.065 0.096 0.014 2.88-3.27 18 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Table 1. — Continued. Group N Mean SD SE Range 2 24 Length of Mandibular Toothrow Males 8.434 0.335 0.068 7.88-9.09 3 17 8.991 0.214 0.052 8.52-9.34 4 33 9.383 0.372 0.065 8.68-10.21 5 8 9.898 0.380 0.134 9.33-10.51 2 22 8.387 Females 0.492 0.105 7.40-9.68 3 30 8.928 0.373 0.068 8.22-9.57 4 49 9.207 0.304 0.044 8.63-10.34 2 24 15.768 Depth of Ramus Males 0.699 0.143 14.68-17.33 3 17 17.903 0.618 0.150 17.03-18.97 4 33 18.858 1.134 0.198 17.19-21.29 5 8 19.949 0.526 0.186 19.11-20.84 2 22 15.621 Females 1.072 0.229 13.35-17.18 3 30 17.093 0.835 0.152 15.63-19.36 4 49 17.698 0.694 0.099 16.30-19.13 2 24 2.315 Width of Lower Incisor Males 0.196 0.040 1.96-2.77 3 17 2.882 0.118 0.029 2.70-3.11 4 33 3.273 0.322 0.056 2.72-3.95 5 8 3.528 0.173 0.061 3.30-3.71 2 22 2.323 Females 0.310 0.066 1.57-2.87 3 30 2.755 0.168 0.031 2.42-3.08 4 49 2.924 0.120 0.017 2.71-3.21 females in a consistent fashion regardless of age. Because of the highly significant difference between sexes and the nonsignificant interaction, males and females were separated for all subsequent analyses. To ascertain which age classes were significantly different within each sex, one-way analysis of variance (ANOVA) was performed on each variable with age as the main factor. The results of these analyses are presented in Table 2. If significant results (P <0.05) were obtained due to the effects of age, three different a posteriori multiple range tests (Student-Newman- Keuls, Scheffe’s, and Duncan’s— Sokal and Rohlf, 1981) were employed to HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 19 Table 2. — Results of the one-way ANOVAs with age as the main factor for each sex for the sample of Cratogeomys castanops from Lubbock County, Texas. Given are source of variation (between groups, within groups, and total variation), degrees of freedom (DF), sum of squares (SS), mean square (MS), F ratio (F), and significance level (P). Source DF SS MS F P Between 3 Condylobasal Length Males 2410.209 803.403 99.938 < 0.001 Within 78 627.042 8.039 Total 81 3037.251 Between 2 Females 886.013 443.007 89.011 < 0.001 Within 94 467.839 4.977 Total 96 1353.852 Between 3 Zygomatic Breadth Males 1790.948 596.983 90.354 < 0.001 Within 76 502.145 6.607 Total 79 2293.094 Between 2 Females 549.752 274.876 78.702 < 0.001 Within 97 338.783 3.493 Total 99 888.536 Between 3 Mastiod Breadth Males 638.770 212.924 72.080 < 0.001 Within 76 224.502 2.954 Total 79 863.273 Between 2 Females 210.989 105.494 61.460 < 0.001 Within 94 161.349 1.717 Total 96 372.338 Between 3 Occipital Depth Males 183.061 61.020 67.308 < 0.001 Within 78 70.713 0.907 Total 81 253.774 Between 2 Females 69.677 34.838 71.302 < 0.001 Within 96 46.906 0.489 Total 98 116.582 20 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Table 2. — Continued. Source DF SS MS F P Between 3 Palatofrontal Depth Males 300.589 100.196 73.146 < 0.001 Within 78 106.845 1.370 Total 81 407.434 Between 2 Females 113.112 56.556 71.225 < 0.001 Within 98 77.817 0.794 Total 100 190.929 Between 3 Length of Maxillary Males 31.441 Toothrow 10.480 48.714 < 0.001 Within 78 16.781 0.215 Total 81 48.221 Between 2 Females 16.383 8.191 40.919 < 0.001 Within 98 19.618 0.200 Total 100 36.001 Between 3 Length of Palate Males 1020.393 340.131 109.009 00.001 Within 77 240.257 3.120 Total 80 1260.650 Between 2 Females 448.412 224.206 96.762 < 0.001 Within 98 227.074 2.317 Total 100 675.486 Between 3 Width of Upper Males 15.594 Incisor 5.198 96.675 < 0.001 Within 78 4.194 0.054 Total 81 19.788 Between 2 Females 5.373 2.687 90.633 < 0.001 Within 97 2.876 0.030 Total 99 8.249 HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 21 Table 2. — Continued. Source DF SS MS F P Between 3 Breadth of Rostrum Males 160.101 53.367 60.137 < 0.001 Within 78 69.219 0.887 Total 81 229.320 Between 2 Females 49.659 24.829 64.828 < 0.001 Within 98 37.534 0.383 Total 100 87.193 Between 3 Length of Rostrum Males 660.507 220.169 87.782 < 0.001 Within 77 193.127 2.508 Total 80 853.634 Between 2 Females 269.361 134.681 89.992 < 0.001 Within 93 139.183 1.497 Total 95 408.544 Between 3 Length of Nasals Males 571.126 190.375 85.710 < 0.001 Within 77 171.029 2.221 Total 80 742.154 Between 2 Females 239.671 119.836 93.078 < 0.001 Within 93 119.735 1.288 Total 95 359.407 Between 3 Interorbital Constriction Males 1.031 0.344 1.739 0.166 Within 78 15.412 0.198 Total 81 16.442 Between 2 Females 0.963 0.482 3.467 0.035 Within 98 13.611 0.139 Total 100 14.574 22 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Table 2. — Continued. Source DF ss MS F P Between 3 Length of Mandibular Males 18.527 Toothrow 6.176 55.081 < 0.001 Within 78 8.745 0.112 Total 81 27.272 Between 2 Females 10.225 5.112 36.963 < 0.001 Within 98 13.554 0.138 Total 100 23.779 Between 3 Depth of Ramus Males 172.838 57.613 74.329 < 0.001 Within 78 60.458 0.775 Total 81 233.296 Between 2 Females 65.547 32.773 47.613 < 0.001 Within 98 67.456 0.688 Total 100 133.003 Between 3 Width of Lower Incisor Males 15.871 5.290 89.163 < 0.001 Within 78 4.628 0.059 Total 81 20.499 Between 2 Females 5.481 2.741 76.065 < 0.001 Within 98 3.531 0.036 Total 100 9.012 identify nonsignificant subsets of the age classes, and the results were compared for congruence. The experiment-wise error rate for all tests was 0.05. Both Student-Newman-Keuls and Scheffe’s procedures are relatively conservative tests, whereas Duncan’s test is more sensitive (Sokal and Rohlf, 1981). Scheffe’s procedure is mathematically equivalent to sums-of-squares simultaneous testing procedure (SS-STP) that has been used in previous systematic analyses (Genoways, 1973). HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 23 In males, all of the characters save interorbital constriction were highly significant for variation due to age. For all significant variables, the a poster¬ iori tests identified four significantly different subsets that corresponded to the four age classes (two through five). This indicates that these four age classes are each morphometrically distinct and should not be pooled in systematic analysis of morphometric variation. For females, all of the characters were significant for variation due to age. Interorbital constriction was the only character that was not highly signifi¬ cant (P >0.001). Moreover, only two significantly different subsets were formed in the a posteriori tests for this variable. The Student-Newman-Keuls and Duncan's procedures each formed subsets of age class two and another of age classes four and three, whereas the Scheffe’s procedure produced subsets of age classes two and four and another of age classes four and three. For all other variables, three significantly different subsets that corresponded to the three age classes (two through four) were formed with the a posteriori tests. In previous studies (Russell, 1968/?; Dowler and Genoways, 1979), age classes four and five were combined in the comparisons of geographic groups. Based upon the results of this analysis, which is based on the largest single series of specimens available for study to date, the pooling of these two age classes could bias the results of the overall analysis of sexual or geographic variation. 24 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Geographic Variation For the analysis of geographic variation, localities were pooled to form groups of adequate sample size for statistical analyses, following the methodologies outlined by Genoways (1973). Care was taken not to cross potential biogeographic boundaries nor current taxonomic boundaries when forming groups. Specimens from peripheral or intermediate localities were not allocated to a priori groups, but subsequently were treated as unknowns. A general description of the resultant 22 groups (Fig. 3) follows, using abbreviated localities (see specimens examined in subspecies accounts beyond for complete localities). Group 1.— Finney, Ford, Gray, Hodgeman, and Lane counties, Kansas. Group 2.— Bacca, Bent, Huerfano, Las Animas, Otero, Prowers, and Pueblo counties, Colorado. Group 3.— Union County, New Mexico. Group 4. — Northwestern Cimarron County, Oklahoma, vicinity of Kenton. Group 5. — Ochiltree, Hansford, Sherman, and Moore counties, Texas, and Texas and Beaver counties, Oklahoma. Group 6. — Armstrong, Randall, and Potter counties, Texas. Group 7. — Deaf Smith and Parmer counties, Texas. Group 8. — Lamb, Hale, Floyd, Hockley, Lubbock, Terry, Lynn, Gaines, and Dawson counties, Texas. Group 9. — De Baca, Roosevelt, Chaves, Lea, and Eddy counties, New Mexico, and Loving, Winkler, Ward, and northern Reeves counties, Texas. Group 10. — Southern Howard, Glasscock, Sterling, Reagan, and Irion counties, Texas. Group 11. — Northern Terrell County, Texas, vicinity of Independence Creek. Group 12. — Southern Terrell County, Texas, vicinities of Sanderson and Dryden. Group 13. — Southeastern Culberson (vicinity of Kent), southwestern Reeves (vicinity of Balmorea), northern Brewster (vicinity of Alpine), and Jeff Davis counties, Texas. Group 14. — Southern Brewster County, vicinity of Big Bend National Park, Texas. Group 15. — Southern Presidio County, vicinity of Presidio, Texas. Group 16. — Southern Hudspeth County (vicinity of Sierra Blanca) and southwestern Culberson County (vicinity of Van Horn), Texas. Group 17. — Northwestern Hudspeth (vicinity of Hueco Tanks) and El Paso counties, Texas. Group 18.— Otero County, vicinity of White Sands National Monument, New Mexico. Group 19. — Sierra County, vicinity of Rhodes Pass, New Mexico. Group 20. — Lincoln County, vicinity of Carrizozo, New Mexico. Group 21.— Maverick County, vicinity of Eagle Pass, Texas. Group 22.— Cameron County, vicinity of Brownsville, Texas. The presence of significant geographic variation within sexes but among groups was tested by a MANOVA. For both males and females, highly significant {P <0.001) differences were obtained, indicating some morpho¬ metric differentiation between at least some a prion groups. The 15 cranial characters were then subjected to one-way ANOVAs with the a priori groups as the main effects for each sex. If significance was detected, indicating differences between a priori groups, the characters were then subjected to Student-Newman-Keuls, Scheffe’s, and Duncan’s multiple range tests. Although each of the 15 cranial characters were significant, for both males and females, the results of the multiple range tests indicated that many of HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 25 Fig. 3.— Map showing the 22 a priori groups (1-22) of C. castanops recognized from the United States and the six clusters (A-F) formed in the cluster analysis. 26 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY the a prion groupings were indistinct. Similar methodologies have been utilized with similar results in other morphometric analyses of mammalian species (Genoways, 1973; Willig, 1985; Riddle and Choate, 1986). The following heuristic methodologies and a priori rational were utilized to determine whether any of the a priori groups (described above) could be pooled in subsequent analyses. Groups 21 and 22 (which are geographically isolated from the main distribution of the species in the United States) had such small sample sizes that their inclusion in multivariate analyses was not appropriate. Each of the remaining 20 groups was treated as an operational taxonomic unit (OTU— Sneath and Sokal, 1973), and, as in previous studies (Genoways, 1973; Riddle and Choate, 1986), the characters for each OTU were the means of the characters for that group. However, only the females of each group were utilized in this procedure because most taxonomic work on this species, and geomyids in general, historically, has been based on females (Russell, 1968/? ; Hendricksen, 1973), which are much less variable within geographic groups than males. A cluster analysis (CA) based on a distance matrix was performed to ascertain which OTUs clustered together and thus resembled each other most closely morphologically. A principle component (PC) analysis also was performed and the first three components extracted. The PC scores on the first three components then were plotted to ascertain if there were phenetic relationships among OTUs that were comparable to the clusters found in the cluster analysis. If there was congruence between the results of the cluster analysis and the PCA, and if the OTUs that clustered together were geographically adjacent, then these were pooled in further analyses. The cluster analysis (program CLUSTER, SPSS Inc., 1986) performed on these data used the squared Euclidian distance matrix. An unweighted pair-group method using arithmetic averages (UPGMA) was the chosen algorithm of agglomeration. The results of the cluster analysis are illustrated in Figure 4. Based on these results, five geographic clusters were formed as follows: A— groups 1 and 2; B— groups 4, 5, 6, and 8; C— groups 11, 12, 13, 14, 15, and 16; D— groups 17, 18, 19, and 20; E— groups 3, 7, 9, and 10. Only one of the clusters (cluster E) contained groups that were not geographically adjacent. Group 10 was geographically separated from the others in this cluster by group 8 of cluster B and by unsuitable habitat. Due to this geographic separation, group 10 was treated as a distinct cluster (cluster E) and the other groups (3, 7, and 9) as cluster F (Fig. 3). The PCA (program FACTOR) was performed on the OTUs and the first three components were extracted. The PC scores on the first three axes for each OTU were plotted and the results demonstrated in Figure 5. Examination of Figure 5 illustrates that the phenetic relationships among OTUs indicated by the results of the PCA correspond well with the results of the CA with the exception of cluster F, which appears to be a heterogeneous cluster. HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 27 n - 1 - r — i - 1 - r 1.0 2.0 3.0 10 30 50 Fig. 4.— Phenogram depicting the results of the cluster analysis (see Fig. 3 for geographic location of samples). Note that the distance scale below the phenogram is nonlinear. A two-way MANOVA was performed on the raw data with sex and clusters as the main effects. Highly significant (P <0.001) differences among clusters within sexes and a nonsignificant interaction were obtained. These results indicated the presence of morphological differences between clusters that were consistent within the sexes. A discriminant function analysis (DFA— program DISCRIMINANT) then was employed to ascer¬ tain how well the individuals within these geographic clusters could be distinguished. In all of the clusters save F, individuals were correctly classified more than 60 percent of the time, whereas individuals within cluster F were correctly classified less than 30 percent of the time. The misclassified individuals of cluster F were not, however, assigned member¬ ship to any one of the other clusters more frequently than to any other. 28 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY PCA1 PCA2 □ 0 A GROUPS 4, 5, 6, 8 □ GROUPS 3, 7, 9 0 GROUPS 1, 2 ★ GROUP 10 ♦ GROUPS 11, 12, 13, 14, 15, 16 ■ GROUPS 17, 18, 19, 20 Fig. 5.— Three-dimensional plot of the first three principle components of the 20 groups used in the cluster analysis (see Fig. 3 for geographic location of samples). Thus, no strong morphological affinity with any of the other clusters was indicated. These results, along with the results of the PCA (see Fig. 5), indicate that cluster F is a heterogeneous unit. All individuals from this cluster were treated as unknowns in subsequent analyses, leaving Five clusters (A-E) to be tested (Fig. 3). Another DFA was performed on individuals using these Five clusters (A-E). The result was an overall correct classiFication rate of 80.7 percent for females (upon which the clusters were based) and 77.55 percent for males. HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 29 Table 3 Results of the discriminant Junction classification using the five clusters formed by cluster analyses. The columns are the clusters into which individuals were classified and the rows are the clusters to which the individuals belong. Groups A B C D E A 36 Males 4 l 0 0 B 10 58 l 0 2 C l 3 42 4 3 D 0 1 8 14 0 E 0 4 1 0 6 A 33 Females 6 3 0 3 B 5 no 5 0 4 C 0 4 91 6 6 D 0 0 9 20 0 E 1 5 9 0 26 Table 3 shows the number of individuals from each cluster that were misclassified and into which cluster they were assigned membership. The individuals of cluster F (groups 3, 7, and 9; see Fig. 3), which were treated as unknowns in this analysis, were assigned membership to one of the five clusters (A-E) with the following results: all individuals but one of group 3 were assigned to cluster B; all individuals of group 7 were assigned to cluster B; individuals of group 9 were assigned with approximately equal frequency to clusters A, B, C, and E, indicating no apparent morphological affinities with any one cluster. One-way ANOVAs (program ONEWAY) were performed on each of the 15 cranial characters for each sex separately with the geographic clusters as the main effect. For all characters that were significant for geographic variation, the three previously mentioned multiple range tests were em¬ ployed to identify maximally nonsignificant subsets of clusters. The results of these tests are shown in Table 4. Fourteen of the 15 characters were highly significant (P <0.001) for geographic variation in both males and females. Interorbital constriction was significant {P <0.006) in females but nonsignificant (P >0.05) in males. For four of the characters (mastoid breadth, length of rostrum, width of upper incisor, and depth of ramus), five significantly different subsets that corresponded to the five groups (A-E) were identified among females. The remaining characters, with the exception of interorbital constriction, formed four significantly different subsets in females but the groups that 30 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Table 4 .—Results of the one-way ANOVAs on each of the 15 cranial characters with the five clusters as the main effect for each sex. Also the results of Student- Newman- Keuls multiple range test (MRT) (Scheffes and Duncans gave similar results). Asterisks (*) in a column (MRT) indicate nonsignificant subsets of the clusters. Group MRT Mean Range P Condylobasal Length Males D * 49.66 45.88-53.65 < 0.001 C * 52.61 46.76-57.13 A * * 53.76 45.64-59.97 E * * 55.10 47.87-57.90 B * 55.82 49.03-61.14 Females D * 45.23 42.88-47.67 < 0.001 C * 47.85 43.97-54.69 A * 48.78 46.65-52.36 E * 49.26 45.69-52.95 B * 51.25 48.28-55.22 Zygomatic Breadth Males D * 32.64 27.97-35.54 < 0.001 C * 34.84 29.98-40.45 A * 35.87 31.57-41.10 B * 37.61 30.85-43.00 E * 37.71 32.32-40.35 Females D * 28.14 26.48-29.76 < 0.001 C * 30.56 27.26-38.82 A * 30.87 28.75-33.54 E * 31.47 28.43-33.86 B * 33.16 30.10-36.16 Mastoid Breadth Males D * 29.09 26.64-31.53 < 0.001 A * 30.34 27.11-34.23 C * 30.62 25.81-33.99 B * 31.84 27.80-35.85 E * 32.36 28.15-34.30 Females D * 26.21 24.25-29.10 < 0.001 A * 27.44 25.86-29.13 C ★ 27.93 24.36-31.05 E * 28.42 26.62-30.49 B ★ 29.13 26.82-39.91 HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 31 Table 4. — Continued. Group MRT Mean Range D ♦ Occipital Depth Males 16.30 15.57-17.32 C 4« 16.90 15.21-19.02 A 4c 17.36 16.02-19.73 E 4c 18.02 15.60-19.28 B 4c 18.07 15.40-21.27 D * Females 15.05 13.94-15.87 C 4c 15.83 14.04-17.91 A 4c 16.10 15.07-17.38 E 4c 16.17 15.00-17.29 B 4c 16.87 15.72-18.47 D * Breadth of Rostrum Males 11.60 9.80-12.88 A 4c 4c 11.95 10.75-13.60 G 4c 4t 12.10 10.72-13.90 E 4c 12.56 10.53-13.53 B 4c 12.58 10.31-14.72 D 4c Females 9.94 9.27-10.75 A 4c 10.38 9.64-11.32 C 4c 4c 10.55 9.40-12.25 E 4c 10.63 9.48-12.31 B 4c 11.26 10.16-12.64 D 4« Length of Rostrum Males 21.86 19.38-23.86 C 4c 22.98 19.47-26.49 A 24.26 22.22-26.86 E 4c 24.28 20.61-26.25 B 4c 25.02 21.11-28.66 D * Females 19.49 17.63-21.78 C 4c 20.51 18.40-23.79 E 4c 21.04 19.35-22.71 A 4c 21.67 20.16-27.03 B 4« 22.70 20.16-25.18 P < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 32 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Table 4. — Continued. Group MRT Mean Range P Length of Nasals Males D * 18.18 16.20-20.27 < 0.001 C * 19.00 16.40-21.33 E * 19.95 16.99-21.60 A * 20.04 17.67-22.24 B * 21.02 17.18-24.04 D * Females 16.02 13.85-17.25 < 0.001 C * 17.01 15.40-20.08 E * 17.15 15.19-19.29 A * 17.75 16.38-19.52 B * 18.95 16.99-21.09 D Interorbital Constriction Males 6.77 6.09-7.13 0.251 A 6.93 6.03-7.74 C 6.95 6.13-7.95 B 6.98 6.08-7.95 E 7.03 6.38-7.49 D * Females 6.57 5.91-7.23 0.005 C * 6.72 5.88-7.92 E * 6.74 5.78-7.29 A * 6.77 5.53-7.53 B * 6.83 5.94-7.70 D * Palatofrontal Depth Males 19.25 18.10-20.94 < 0.001 C * 20.00 17.41-22.14 A * 20.55 18.70-23.23 E * 21.16 18.01-22.01 B * 21.56 18.09-25.05 D * Females 17.74 16.48-18.95 < 0.001 C * 18.63 16.97-20.99 A * * 18.87 15.46-19.93 E * 19.09 17.67-20.28 B * 20.02 18.25-21.76 HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 33 Table 4. — Continued. Group MRT Mean Range P D ♦ Length of Maxillary Toothrow Males 9.61 8.97-10.47 < 0.001 C * 9.75 8.58-10.90 E * 9.81 8.57-10.32 A * 10.14 9.18-10.99 B ♦ 10.36 9.55-11.75 D * Females 9.20 8.59-9.84 < 0.001 C * 9.36 7.78-10.35 E * 9.46 8.75-10.28 A * 9.76 9.09-10.25 B * 9.93 8.77-11.39 D * Length of Palate Males 26.79 24.77-29.21 < 0.001 C * 28.50 25.32-31.34 A * 30.26 26.72-34.07 E * 30.30 25.65-32.61 B * 31.23 27.40-35.12 D * Females 24.29 22.52-25.62 < 0.001 C * 25.88 23.46-30.83 A * 27.20 25.51-29.42 E * 27.22 25.30-29.63 B * 28.59 26.39-31.72 D * Width of Upper Incisor Males 3.08 2.69-3.43 < 0.001 C * 3.21 2.67-3.67 E * 3.22 2.84-3.44 A ♦ 3.38 2.15-3.78 B ♦ 3.43 2.88-3.98 D ♦ Females 2.73 2.52-3.00 < 0.001 C * 2.81 2.47-3.42 E * 2.86 2.41-3.33 A ♦ 2.96 2.76-3.29 B ♦ 3.07 2.70-3.96 34 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Table 4. — Continued. Group MRT Mean Range P D * Length of Mandibular Toothrow Males 8.71 8.12-9.28 < 0.001 C * 8.92 7.83-9.78 A * 9.24 8.48-9.93 E * 9.25 8.26-10.29 B * 9.51 8.68-10.36 D * Females 8.47 7.88-9.01 < 0.001 C * 8.70 7.75-9.91 E * 9.04 8.22-9.83 A * 9.07 8.35-10.07 B * 9.30 8.14-10.34 D * Depth of Ramus Males 17.68 15.43-19.16 < 0.001 A * * 17.94 15.91-19.74 C * 18.26 16.37-19.61 E * 18.80 16.96-19.91 B * 18.82 16.75-21.29 D * Females 16.21 15.24-17.70 < 0.001 A * 16.55 15.41-17.35 C * 17.05 15.31-18.97 E * 17.36 15.48-18.53 B * 17.67 16.08-20.20 D * Width of Lower Incisor Males 2.95 2.48-3.26 < 0.001 C * 3.13 2.69-3.52 E * 3.14 2.78-3.41 A * 3.24 2.78-3.91 B * 3.27 2.41-3.95 D * Females 2.54 2.33-2.81 < 0.001 C * 2.65 2.33-3.36 E * 2.69 2.37-3.20 A * 2.75 2.51-3.02 B ♦ 2.88 2.51-3.21 HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 35 overlapped were not the same for each character. There was much more overlap in males, indicated by the failure to identify five subsets for any of the characters, and in only three of the characters (condylobasal length, length of nasals, length of mandibular toothrow) were four subsets identified. The results of my study of geographic variation of C. castanops in the United States lead me to recognize nine subspecies. The five morphologi¬ cally distinct clusters (A-E) and individuals of group 9 are here recognized as separate races. Three peripherally isolated populations are recognized as well. These are described in some detail in the accounts that follow. Subspecies are arranged alphabetically. 36 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Accounts of Subspecies Cratogeomys castanops (synonymy under subspecies) Distribution.— In the United States (Fig. 6), this species occurs from the Arkansas River drainage in eastern Colorado and western Kansas southward through the Oklahoma Panhandle, western Texas, and eastern New Mexico to the Rio Grande; isolated populations are known in the upper Rio Grande Valley (vicinity of Albuquerque, New Mexico), in Maverick County, Texas (vicinity of Eagle Pass), and near Brownsville, Cameron Co., Texas, at the mouth of the Rio Grande. In Mexico, this pocket gopher probably occurs south of the Rio Grande to southern Coahuila and northern Zacatecas, in parts of Nuevo Leon, and eastward along the south side of the Rio Grande to the Gulf Coast in Tamaulipas (see Davidow-Henry et al., 1989). Description.— Medium-sized for gophers of the genus Cratogeomys] skull with¬ out strong platycephalic specializations; squamosals unspecialized, expanded neither medially nor laterally; breadth across zygomata greater than breadth across squamosals; lambdoidal crest convex posteriorly, never sinuous; P4, Ml, and M2 lacking posterior enamel plate; outer surface of upper incisors with single median groove; diploid chromosome number 46 (Davidow-Henry et al. , 1989). Comparisons.— From Cratogeomys merriami , the only other member of the castanops species-group, C. castanops differs in being smaller both externally and cranially. Cratogeomys castanops shows less cranial and dental specializa¬ tion than C. merriami (Russell, 1968 6). From the gymnurus species-group, to which all other members of the genus are assigned, castanops differs in being generally smaller and lacking the strong platycephalic specializations characteristic of the gymnurus group (Russell, 19686; Hall, 1981). Remarks .—Cratogeomys castanops is the most wide-spread species of this genus and the only one not restricted to the southern mountainous region of the Mexican Plateau and the Neovolcanic belt (Russell 19686). Given the large geographic distribution and the diverse ecological conditions in which C. castanops exists, it is not surprising that it is the most variable of the species in the genus, with 19 currently recognized subspecies (Davidow- Henry et al ., 1989). The most subspecies in any of the other species of the genus is seven (Russell, 19686). Cratogeomys castanops angusticeps Nelson and Goldman Cratogeomys castanops angusticeps Nelson and Goldman, Proc. Biol. Soc. Washington, 47:139, 1934. Holotype from Eagle Pass, Maverick Co., Texas. Distribution.— Known only from vicinity of the type locality. See Figures 6 and 7. Description.— A small, pale race geographically isolated from main distribu¬ tion of species in the United States; size and cranial dimensions resembling HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 37 108 96 40 32 24 Fig. 6. — Map showing the distribution of the herein recognized subspecies of C. castanops in the United States. 38 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY C. c. perplanus B C. c. dalquesti 3 C. c. lacrimalis □ C. c. parviceps B C. c. clarkii B C. c. angustieeps B Q. c. tamaulipensis Fig. 7.— Map showing the distributional limits (shading) and localities for specimens examined (dots) of subspecies of C. castanops in Texas. those of C. c. parviceps from southwestern New Mexico; means of selected cranial measurements of two adult females and four adult males, respectively, are as follows: condylobasal length, 46.45, 50.27; mastoid breadth, 26.05, 27.89; occipital depth, 15.18, 15.78 (three specimens only); breadth of rostrum, 10.14, 10.76; length of rostrum, 18.85, 21.06 (three specimens only); palatofrontal depth, 17.81, 18.76 (three specimens only); length of mandibular toothrow, 8.80, 8.95. Comparisons.— From topotypic material of C. c. tamaulipensis , angustieeps differs in averaging smaller in all cranial dimensions, is paler, and lacks dark postauricular patches. From the population of tamaulipensis in Cameron County, Texas, angustieeps averages smaller in all cranial measurements except breadth of rostrum, length of maxillary toothrow, and length of mandibular toothrow (the means for these characters are approximately equal for the two races), and specimens of angustieeps are much paler (see HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 39 account of tamaulipensis for details on the Cameron County population). For comparison with C. c. clarkii, see account of that subspecies. C. c. clarku and C. c. tamaulipensis are the only subspecies occurring near the geographic range of angusticeps to the north of the Rio Grande. Remarks The sample size available for angusticeps was insufficient for inclusion in multivariate analyses. Russell (19686) referred specimens from northern Terrell County, Texas (west of the Pecos River), to angusticeps. Comparison of females from Eagle Pass with a series of females from Independence Creek, Terrell County, using one-way ANOVAs, revealed significant differences in two (length of rostrum and length of mandibular toothrow) of the 15 cranial characters analyzed. These results, combined with the apparent isolated nature of the population in the vicinity of Eagle Pass, warrant its tentative recognition as a distinct subspecies until additional material from the type locality and intervening areas can be obtained. The only gophers presently known from the area between Eagle Pass and the Pecos River are of the genera Geomys and Thomomys. For details on the Terrell County population, see the account of C. c. clarkii. Specimens examined..— Total of 13 as follows. Texas. Maverick Co. : 1 mi. W Seco Mines, 2; 1 mi. W Eagle Pass, 2; Eagle Pass, 9 (1 KU, 8 USNM). Cratogeomys castanops castanops (Baird) Pseudostoma castanops Baird, Exploration and survey of the valley of the Great Salt Lake of Utah, . . . . Lippincott, Grambo & Co., Philadelphia, p. 313, 1852. Holotype from “Prairie road to Bent’s Fort,” restricted to near the present town of Las Animas, Bent Co., Colorado by Nelson and Goldman (1934). Cratogeomys castanops, Merriam, N. Amer. Fauna, 8:159, 1895. Distribution.— Southeastern Colorado and eastward along the north side of the Arkansas River to Ford and Hodgeman counties, Kansas. See Figures 6, 8, and 9. Description.— Medium-sized race with relatively long, narrow skull; size and cranial dimensions somewhat intermediate between C. c. perplanus of Texas and Oklahoma panhandles and C. c. clarkii of Trans-Pecos area of Texas; color similar to that of perplanus and C. c. lacrimalis of southeastern New Mexico, but averaging darker dorsally. See Table 4, group A, for means and ranges of all cranial measurements. Comparisons.— From C. c. perplanus , the only adjacent subspecies (to the south), castanops differs in averaging smaller in all cranial dimensions for each sex (see Table 4); it is especially small in mastoid breadth and breadth of rostrum (in both sexes), with only C. c. parviceps, C. c. angusticeps , and C. c. tamaulipensis averaging smaller in these dimensions. From perplanus , the subspecies castanops also differs in having a greater number of dark-tipped hairs on the dorsum, imparting an overall darker appearance. Remarks.— Specimens of C. c. castanops most closely resemble, in size and color, populations from Glasscock and adjacent counties (group E, Table 4) in west-central Texas than gophers from any other sample. They are, 40 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Fig. 8.— Map showing the distributional limits (shading) and localities for specimens examined (dots) of C. c. castanops in Colorado. however, separated from this population by a substantial geographic area that is occupied by a larger race, C. c. perplanus. Gene flow between perplanus and castanops is evident in samples from southern Baca County, Colorado, and Cimarron County, Oklahoma. Russell (1968£ ) noted that specimens from northwestern Oklahoma appeared to be intermediate between perplanus and castanops but were best referred to the former. I agree with Russell and assign specimens from that area to C. c. perplanus. Specimens from southern Baca County, Colorado, although grading toward the larger perplanus , are best referred to castanops. Russell (1968£ ) assigned three specimens from northeastern New Mexico to castanops. The current availability of much more material from this area than was available to Russell has facilitated a more thorough analysis. Individuals from Union and Colfax counties, New Mexico, were treated as unknowns in discriminant function analysis (described in section on geo¬ graphic variation). Only one of these individuals was classified into group A ( castanops ). The remaining individuals (save one) were classified into group B {perplanus ). Based on these results, specimens from northeastern New Mexico are best assigned to perplanus and not to castanops. HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 41 Specimens examined. —Total of 205 as follows. Colorado. Baca Co.\ Gaume’s Ranch, NW corner of Baca County, 2 (UCM); 14 mi. N Springfield 2 (TCWC); Bear Creek bottom, Springfield, 6 (UCM); 15.25 mi. S, 8 mi. W Pritchett, 1 (MHP); 17 mi. S, 4 mi. W Pritchett, 3 (MHP); 18 mi. S, 4.25 mi. W Pritchett, 1 (MHP )■ Johnston’s Ranch, Monon, 2 (UCM). Bent Co. : Las Animas, 6 (USNM); 2.2 mi. S, 1.5 mi. W John Martin Dam. 2 (MHP); 5.5 mi. S, 1.5 mi. W John Martin Dam, 1 (MHP); 12 mi. E La Junta [recorded as from Otero County], 2 (KU). El Paso Co.: 8 mi. S, 3 mi. W Rush [recorded as from Lincoln County], 1 (MHP); 14 mi. S, 4 mi. W Ellicott, 1 (MHP); 16 mi. S, 2 mi. W Elhcoti, 1 (MHP); 17 mi. S, 4 mi. W Ellicott, 1 (MHP). Huerfano Co. : 3.5 mi. Ejct. hwys. I 25 and 10, 1 (MHP); 5.6 mi. S, 7.5 mi. Ejct. hwys. I 25 and 10, 1 (MHP). Las Animas Co.: 15.4 mi. Ejct. hwys. I 25 and 10, 1 (MHP); 7 mi. S, 13.6 mi. Ejct. hwys. I 25 and 10 [recorded as from Huerfano County], 1 (MHP); l' mi. S, 5.5 mi. E Kim, 1 (MHP); 1 mi. S, 6.5 mi. E Kim, 1 (MHP); Tecolote Mesa, 4 mi. S, 2 mi. E Kim, 1 (MHP); 1.5 mi. W Lone Butte, 1 (MHP); Mesa de Maya by Lone Butte, 5 (MHP); west end of Mesa de Maya, 6 (MHP); 9 mi. N, 11.5 mi. E Branson, 1 (MHP); 8.5 mi. S, 10 5 mi W Kim, 1 (MHP); 8.5 mi. S, 9 mi. W Kim, 10 (MHP); 11.5 mi. S, 7.25 mi. E Kim, 1 (MHP); 12.25 mi. S, 0.5 mi. E Kim, 1 (MHP); 12.5 mi. S, 7.5 mi. E Kim, 1 (MHP); 13 mi. S, 5 5 mi E Kim, 2 (MHP). Lincoln Co.: 11 mi. S, 9 mi. W Punkin Center, 1 (MHP); 10 mi. S, 3 mi W Karval, 1 (MHP); 12 mi. S, 3 mi. W Karval, 1 (MHP); 13 mi. S, 4 mi. W Karval, 1 (MHP). Otero Co.: 1 mi. S jet. hwys. 50 and 167, 1 (MHP); 4 mi. W Rocky Ford, 6 (3 KU, 3 TTUV 1 5 mi E Rocky Ford, 8 (NMSU); Apishapa River, 7 mi. S, 1 mi. W Fowler [recorded as from Huerfano County], 1 (MSB); La Junta, 3; jet. hwys. 167 and 10, 1 (MHP); 4 mi. N, 6 mi. W Timpas, 2 (MHP); 2 mi. N Timpas, 1 (MHP). Prowers Co. : Lamar, 1; 2 mi. S, 1 mi. E Lamar, 1. Pueblo Co.: 2.7 mi. S, 0.75 mi. E Avondale, 4 (MHP); 14.8 mi. W jet. hwys. 167 and 10, 1 (MHP)- 1 mi. N, 8.5 mi. W Goodnight, 1 (MHP). Kansas. Finney Co.: 19 mi. S Dighton, 3 (KU); 8.5-9 mi. N, 2 mi. W Kalvesta, 3 (MHP); 8 mi N, 2.5 mi. W Kalvesta, 1 (MHP); 6.5 mi. N, 4.5 mi. W Kalvesta, 4 (MHP); 5 mi. W Kalvesta, 1 (KU)- 3-4 mi. W Kalvesta, 3 (KU). Ford Co.: 7. 5-8. 5 mi. N, 6 mi. E Dodge City, 8 (KU); 42 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Ford County State Lake, 6.5 mi. N, 4 mi. E Dodge City, 30 (KU); Ford County Lake, 2 mi. N, 1-1.25 mi. W Wright, 4 (MHP); 5 mi. N, 1 mi. E Dodge City, 2 (KU); 5 mi. N, 2.5 mi. E Dodge City, 4 (KU). Gray Cor. 10 mi. N, 4.5 mi. E Cimarron, 1 (KU). Hamilton Cor. 5 mi. N, 1 mi. W Syracuse, 1 (MHP); 5 mi. N Syracuse, 1 (KU); 4 mi. N, 1.5 mi. W Syracuse, 3 (KU); 4 mi. N, 1 mi. W Syracuse, 5 (MHP); Hamilton County Lake, 2 (KU); 3 mi. N, 8 mi. W Syracuse, 2 (MHP); 2.3 mi. N, 0.5 mi. W Syracuse, 1 (KU); 0.5 mi. NW Syracuse, 1 (MHP). Hodgeman Cor. 12.7 mi. W Jetmore, 1 (KU); 9-10.4 mi. W Jetmore, 5 (KU); 8.8 mi. W Jetmore, 1 (KU); 2-2.75 mi. S, 3-3.5 mi. E Jetmore, 10 (MHP); 4 mi. S, 0.5 mi. W Jetrrtore, 1 (KU); 10 mi. S, 8 mi. W Jetmore, 1 (KU); 14 mi. S, 6 mi. E Jetmore, 1 (KU). Lane Cor. 14 mi. S, 6 mi. E Dighton, 1 (KU); 15 mi. S, 7.5 mi. E Dighton, 1 (KU). Cratogeomys castanops clarkii (Baird) Geomys clarku Baird, Proc. Acad. Nat. Sci. Philadelphia, 7:332, 1855. Holotype from Presidio del Norte, on the Rio Grande, at or near the present town of Ojinaga, Chihuahua. Cratogeomys castanops clarkii, Nelson and Goldman, Proc. Biol. Soc. Washington, 47:140, 1934. Pappogeomys castanops pratensis Russell, Univ. Kansas Publ., Mus. Nat. Hist., 16:653, 1968. Holotype from 8 mi. W and 3 mi. S Alpine, Brewster Co., Texas. Pappogeomys castanops torridus Russell, Univ. Kansas Publ., Mus. Nat. Hist., 16:665, 1968. Holotype from 3 mi. E Sierra Blanca, Hudspeth Co., Texas. Distribution.— Southern Trans-Pecos area of Texas from southern Hudspeth County eastward to Crane, Upton, and Val Verde counties, hence southward to the Rio Grande and across the river in the vicinity of Ojinaga, Chihuahua, and adjacent northeastern Coahuila. See Figures 6 and 7. Description. — Small race characterized by relatively short, wide skull. Skull averaging smaller in all characters than in C. c. lacrimalis from northern Pecos Valley (see Table 4, group C, for means and ranges of all cranial measurements). Pelage color in this race extremely variable and seemingly correlated with ecological conditions under which individuals exist; speci¬ mens from Davis Mountains, Texas, for example, average much darker than those from lower elevations along the Rio Grande. Comparisons.— From C. c. parviceps and C. c. angusticeps, the adjacent races on the west and east, respectively, clarkii differs in averaging larger in all cranial dimensions. Specimens of parviceps average much darker than most specimens of clarkii although, as noted above, some specimens of the latter from higher elevations are dark in color. Specimens of angusticeps typically are much paler than those of clarkii. From C. c. lacrimalis , the subspecies occurring to the north in the northern Pecos River Valley, clarkii differs in averaging smaller for most cranial measurements and has much smaller lacrimal bones. Color in specimens of clarkii from lower elevations is similar to that of typical specimens of lacrimalis. See account of C. c. angusticeps for comparison with specimens from the eastern range of clarkii. For comparison with C. c. perplanus, see account of that subspecies. Remarks.— Russell (19686 ) assigned specimens from the southern Trans- Pecos to four subspecies— angusticeps, clarkii , pratensis , and torridus. The latter two are here placed in synonymy under the older name clarkii. Specimens from the eastern part of the Trans-Pecos that were allocated to angusticeps by Russell are discussed in the account of that subspecies. HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 43 Specimens from northern Brewster and Jeff Davis counties, which Russell (19686 ) assigned to pratensis , and specimens from southern Hudspeth County, which he assigned to torndus, were found, based upon much larger samples, to be statistically indistinguishable from topotypic material of clarkii from the vicinity of Ojinaga, Chihuahua, and from specimens from just north of the Rio Grande near Presidio. Specimens from southern Hudspeth County in, and to the south of, the Sierra Diablo Mountains do average slightly smaller than typical clarkii , and this appears to be an area of intergradation between clarkii and the smaller parviceps to the northwest. These specimens are, however, clearly referable to clarku, whereas specimens from north of the Sierra Diablos in the Culberson Salt Flats are best assigned to the smaller parviceps. Three specimens, housed in the United States National Museum, are recorded from localities east of the Pecos River on the Edwards Plateau- near Ft. Lancaster and Howard Spring, Crockett County, and Juno, Val Verde County. All of these specimens were collected near or before the turn of the century. Recent field work at Ft. Lancaster and Juno in the summers of 1986 and 1987 failed to provide evidence of any extant pocket gopher populations in these areas. In an analysis of the present and past distributions of pocket gophers from cave deposits on the Edwards Plateau, Dalquest and Kilpatrick (1973) reported Thomomys and Geomys, but not Cratogeomys. These three specimens were not included in the multivariate analyses due to skull damage or because they were not of the appropriate age class. Thus, an appraisal of their subspecific status is tentative at best. Because recent efforts to obtain gophers from these areas has failed, and because Dalquest and Kilpatrick (1973) did not find Cratogeomys in the fossil material they examined from the Edwards Plateau, it is possible that these specimens represent immigrants from the nearest established populations, which are immediately to the west in the Trans-Pecos area and are heie assigned to clarkii. Russell (19686) reported two subspecies from the vicinity of Ojinaga, Chihuahua. I have found that specimens recorded from south of Ojinaga are not assignable to clarkii and probably are referable to C. c. consitus, a race that occupies northwestern Chihuahua. Additional material from western Chihuahua will be needed to resolve this problem. Specimens from northern Coahuila tentatively are assigned to clarkii on the basis of size. Analysis of additional material from Coahuila and other parts of northern Mexico will be needed before an accurate designation of these specimens can be made. Specimens examined. — Total ol 421 as follows. Chihuahua. 2 mi. WNW Ojinaga, 1 (AMNH); 1.5 mi. WNW Ojinaga, 1 (AMNH); Ojinaga, 5 (3 CoIhui’la. ^7 mi. S Dryden, Texas, on Rw Grande, 6 (KU); Villa Acuna, 5 (KU); Canyon del Cochino, 16 mi. N, 21 mi. E Piedra Blanca, 1 (KU); 11 mi. W Hidalgo San Miguel, 1 (KU). 44 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Texas. Brewster Co.: 18.6 mi. N, 1.2 mi. E Marathon, 5; 11 mi. N Alpine, 2 (MWSU); 10.2 mi. N Alpine, 1 (CCSU); 5 mi. N Alpine, 1 (SRSU); 4 mi. N Alpine, 1 (SRSU); 11.8 mi. N, 2 mi. E Marathon, 5; 11.5 mi. N, 2 mi. W Marathon, 4; 2 mi. NW Alpine, 1 (SRSU); 3 mi. W Alpine, 2 (SRSU); Alpine, 10 (6 SRSU, 3 TTU, 1 MWSU); Sul Ross Ranch, 1 (SRSU); Toronto Pass, 1 (SRSU); 2 mi. S, 6 mi. W Alpine, 3 (KU); 8 mi. N, 17 mi. W Marathon, 2; 3 mi. S, 10 mi. W Alpine, 2 (KU); 3 mi. S, 8 mi. W Alpine, 5 (KU); 4 mi. N, 10 mi. W Marathon, 1 (KU); 10 mi. S Alpine, 1 (SRSU); 12 mi. SE Alpine, 2 (SRSU); 12 mi. S Alpine, 1 (SRSU); Marathon County Park, 3 (TAI); 16 mi. S Alpine, 1 (MWSU); 4.5-5 mi. S Marathon, 4 (TAI); 6 mi. S Marathon, 1 (CCSU); 9 mi. S Marathon, 1 (TAI); 22 mi. SE Alpine, 1 (SRSU); 22 mi. S Alpine, 2 (SRSU); 15.4 mi. S Marathon, 2 (CCSU); 26 mi. S Alpine, 1 (UTEP); Black Gap, 3 (TNHC); 8 mi. N Terlingua, 3 (KU); Terlingua Creek, 4 mi. E Terlingua, 8 (KU); 4 mi. E Terlingua, 1 (TCWC); 5-6 mi. S Terlingua, 3 (KU); Lajitas, 4 (KU); 1 mi. E Lajitas, 1; 1 mi. S W Boquillas, Rio Grande, 3 (MVZ); Cottonwood Campground, BBNP, 1 (TCWC); 3 mi. W Rio Grange Village, BBNP, 1 (TCWC); 8 mi. SW Rio Grande Village, BBNP, 2 (TCWC); Big Bend of Rio Grande, 5 (MVZ); Brewster County only, 1 (SRSU). Crane Co.: 7 mi. SW McCamey, 1. Crockett Co.: Ft. Lancaster, 1 (USNM); 5 mi. S Howard Springs, 1 (USNM). Culberson Co.: 25 mi. N Van Horn, 1 (MWSU); 25 mi. NE Van Horn, 1 (MWSU); 20 mi. N Van Horn, 1 (MWSU); 21 mi. NNE Van Horn, 1 (MWSU); 6 mi. N Kent, 1 (MWSU); 6 mi. NW Kent, 1 (MWSU); 1.5 mi. N Kent, 2 (MWSU). Hudspeth Co.: Bat Cave, Sierra Diablos, 1 (TCWC); 12 mi. N Allamore, 1 (TCWC); 1 mi. N, 0.5 mi. E Sierra Blanca, 8 (UIMNH); 3 mi. W Sierra Blanca, 1 (TCWC); 0.25 mi. W Sierra Blanca, 4 (UIMNH); Sierra Blanca, 2 (1 UIMNH, 1 TTU); Methodist Churchyard, Sierra Blanca, 3 (UIMNH); 2 mi. E Sierra Blanca, 2 (KU); 3 mi. E Sierra Blanca, 1 (KU). Jeff Davis Co.: 10 mi. S Kent, 1 (MWSU); 4 mi. W Toyavale, 1 (MWSU); 12 mi. S Kent, 1 (MWSU); 12 mi. S Kent [recorded as from Culberson County], 1 (MWSU); 13 mi. S Kent, 1 (MWSU); 15 mi. S Kent [recorded as from Culberson County], 2 (MWSU); 20 mi. SSE Kent, 1 (MWSU); 11 mi. NE McDonald Observatory, 1 (MWSU); 10 mi. NE McDonald Observatory, 1 (MWSU); 3.6 mi. NNW Nunn Hill, Davis Mts., 1 (MWSU); 26 mi. NW Ft. Davis, 1 (MWSU); 16 mi. N Ft. Davis, 3 (TCWC); Madera Canyon only, 4 (TCWC); Frazier Canyon, 10 mi. N Ft. Davis, 2; Limpia Canyon, 10 mi. NE Ft. Davis, 1; 4.8 mi. WSW McDonald Observatory, 1 (MWSU); 9 mi. E Mt. Livermore, 3 (TCWC); 7.5 mi. E Mt. Livermore, 3 (TCWC); Limpia Canyon, 9.5 mi. NE Ft. Davis, 1; 9 mi. N Ft. Davis, 1 (CCSU); Limpia Canyon, 8. 8-9. 2 mi. NE Ft. Davis, 30; Limpia Canyon, 8 mi. NE Ft. Davis, 1 \ 7.5 mi. NW Ft. Davis, 1 (MWSU); Limpia Canyon, 6. 7 mi. NE Ft. Davis, 2 ; Limpia Canyon, 4 mi. NW Ft. Davis, 3 ; Limpia Canyon, 3 mi. NW Ft. Davis, 2; Limpia Canyon only, 2 (ASU); 2 mi. NW Ft. Davis, 1 (MWSU); Limpia Creek by Davis Mts. State Park, 1; 1 mi. N Ft. Davis, 8 (TCWC); 9 mi. W Ft. Davis, 5 (TCWC); 5 mi. W Ft. Davis, 12 (TCWC); 1 mi. SE Ft. Davis, 1; 2 mi. S Ft. Davis, 2; 3.8 mi. SE Ft. Davis, 1; 4.1 mi. SE Ft. Davis, 1; 5 mi. S Ft. Davis, 1 (SRSU); 15 mi. NW Alpine, 3 (SRSU); 12.6 mi. N Alpine [recorded as from Brewster County], 1 (CCSU). Pecos Co.: 20 mi. N Ft. Stockton, 1 (MWSU); 2 mi. N Girvin, 3; Ft. Stockton, 4 (KU); 30 mi. SE Ft. Stockton, 3 (SRSU); 35 mi. SE Ft. Stockton, 1 (SRSU); 33 mi. S Ft. Stockton, 2; 21.2 mi. N, 1.5 mi. W Marathon, 2. Presidio Co.: 11 mi. W Valentine, 10 (TNHC); 9 mi. W Valentine, 4 (CCSU); 9 mi. NE Marfa on hwy. 67, 2; 2 mi. S Paisano, 10 (TCWC); 36 mi. SE Marfa, 2; 37 mi. S Marfa, 2 (TCWC); 1 mi. W Plata, 1 (MWSU); 63 mi. S Marfa, 1 (TNHC); 3 mi. NW Presidio, 2 (AMNH); 2 mi. NW Presidio, 1 (TCWC); Presidio, 8 (KU); 0.5 mi. S Presidio, 1 (TCWC); 1 mi. S, 2 mi. E Presidio, 1 (KU); 1 mi. S, 4 mi. E Presidio, 2 (KU); 3 mi. S, 6 mi. E Presidio, 1 (KU); 7 mi. ESE Presidio, 2 (AMNH). Reeves Co.: 3 mi. WNW Toyavale, 3 (MWSU); 4 mi. W Toyavale, 4 (MWSU). Terrell Co. : 15-16 mi. S Sheffield, 10 (9 TNHC, 1 TTU); hwy. crossing at Independence Creek, 1 (SRSU); 16 mi. S, 6 mi. E Sheffield, 11; 19-20 mi. S Sheffield, 20 (TNHC); 24 mi. S Sheffield, 1; 1 mi. N Sanderson, 4 (MWSU); 2 mi. E Sanderson, 14 (12 TCWC, 2 KU); 3 mi. W Dryden, 9 (KU); 2 mi. W Dryden, 5 (KU); 1 mi. W Dry den, 2 (KU). Upton Co.: 4 mi. N, 5 mi. W McCamey, 1. Val Verde Co.: 20 mi. E Juno, 1 (USNM); Samuels, 19 mi. W Langtry, 1 (USNM); 8 mi. S Langtry, 1 (USNM); between Pecos and Rio Grande rivers, 1 (USNM). HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED S'l AT ES 45 Cratogeomys castanops dalquesti, new subspecies Holotype.- Female, adult, skin and skull, The Museum, Texas Tech Univer¬ sity no. 44458, from 1 mi. N and 4 mi. W Sterling City, Sterling Co., Texas; obtained on 10 June 1986 by Robert R. Hollander, original no. 1506. External and cranial measurements ol the holotype are as follows: total length, 268; length of tail vertebrae, 68; length ol hind foot, 38; length of ear, 7; weight, 256 grams; condylobasal length, 51.79; zygomatic breadth, 32.88; mastoid breadth, 28.86; occipital depth, 17.01; breadth ol rostrum, 10.59; length of rostrum, 21.64; length of nasals, 16.97; least interorbital con striction, 6.62; palatofrontal depth, 19.84; length ol maxillary toothrow, 9.02, length of palate, 28.50; width of upper incisor, 2.98; length of mandibular toothrow, 9.17; depth of ramus, 17.28; width of lower incisor, 2.57. Distribution. — West-central Texas north of the Edwards Plateau and south of the Llano Estacado from the upper Concho River Valley in Sterling and Glasscock counties westward to eastern Upton County. See Figures 6 and 7. Description . —Relatively large race having a long skull with relatively short rostrum and nasals, short toothrows relative to lengths of skull and palate, and large and conspicuous lacrimals; see Table 4, group E, lor means and ranges of all cranial measurements. Color similar to castanops in having a large number of dark-tipped hairs on the dorsum imparting a dark, grizzled appearance. Comparisons . — From female C. c. perplanus, the subspecies occurring to the north, female dalquesti differ in averaging smaller in all cranial dimensions. Males of dalquesti average smaller than males of perplanus in all cranial dimensions save zygomatic breadth, mastoid breadth, and interorbital constriction, which are similar. In both sexes, dalquesti averages darker dorsally than perplanus. The most distinguishing characteiistic sepaiating dalquesti from perplanus is the size and shape of the lacrimal bones. This qualitative character was not included in the multivariate analysis but is depicted in Figure 10. In this character, dalquesti demonstiates a closer relationship to C. c. lacrimalis and C. c. clarkii than with either perplanus or castanops. From C. c. clarkii, the race to the southwest, dalquesti differs in averaging larger in all cranial characters and in being much darker in color (with the exception of specimens of clarkii from the Davis Mountains). 4 he lacrimal bone of dalquesti is larger than that of clarkii. C c. dalquesti is geographically isolated from C. c. lacrimalis ol the northern Pecos River Valley of Texas and eastern New Mexico (see Fig. 7) but the two taxa closely resemble each other. From lacrimalis, dalquesti differs in averaging darker in color and having smaller lacrimal bones that protrude less into the orbit (see Fig. 10). Remarks.— In a study of geographic variation in Cratogeomys castanops on the Llano Estacado of Texas and New Mexico, Dowler and Genoways (1979) included specimens from Glasscock County in one ol their samples. They 46 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Fig. 10. — Partial view of the crania depicting the size, shape, and relative position of the lacrimal bone of A) C. c. lacrimalis, adult female (TTU 45535) from 2 mi. N, 5 mi. E Mentone, Loving Co., Texas; B) C. c. perplanus, adult female (TTU 26407) from 1.8 mi. N Littlefield, Lamb Co., Texas; and C) C. c. dalquesti, adult female (TTU 26047) from 2 mi S, 12 mi. W Garden City, Glasscock Co., Texas. noted that gophers in this sample (which they labeled 12) were distinctly smaller than specimens to the north and west and did not fit the pattern of clinal variation they observed. Other samples from the Llano indicated a north-south increase in size. They suggested that the Glasscock County sample might represent intergrades between C. c. perplanus and the smaller C. c. angusticeps to the south. Most of the specimens examined by them, along with many additional specimens from adjacent areas, are here recognized as distinct and the name C. c. dalquesti proposed to represent them. This population is geographically isolated from angusticeps by the Edwards Plateau and shows no affinities with that race. C. c. dalquesti may come into contact with C. c. perplanus along the southeastern edge of the Llano. A specimen from Big Spring is clearly referable to perplanus and probably is from above the caprock on the north side of that city, whereas specimens from Glasscock County to the south of Big Spring are clearly assignable to dalquesti and are from below the caprock. There is no evidence of hybridization in this area and it appears that the caprock may be a substantial barrier to movement of C. castanops and prevents these two races from coming into contact. A similar situation is present to the west in the area of Stanton. A specimen labeled as from Stanton and on top of the caprock is referable to perplanus , whereas those from southeast of Stanton in Glasscock County and off the caprock are referable to dalquesti. C. c. dalquesti is separated from C. c. clarkii in western Upton County by a fingerlike extension of the Edwards Plateau escarpment and a population of Thomomys bottae (Hollander et al., 1987). This extension of the escarpment HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 47 extends in a northwestern direction almost to the Monahans Sandhills in eastern Crane County, and is referred to locally as King and Castle mountains. Specimens from west of the escarpment are clearly assignable to clarkii, whereas specimens from east of the escarpment (vicinity of Rankin) are clearly referable to dalquesti. There is no evidence of intergradation in this area between clarkn and dalquesti. It appears that the population of Thomomys that occurs near and to the east of McCamey or the Edwards escarpment, or some combination of both, provide a substantial barrier to gene flow between clarkii and dalquesti in this area. Individuals of dalquesti show many morphological affinities with specimens of C. c. lacrimalis from the Pecos River Valley of western Texas and eastern New Mexico. Cranially, these races are similar although typical specimens of dalquesti are much darker in color than are those typical of lacrimalis. The two races are geographically separated by the Llano Estacado (which is occupied by C. c. perplanus ) and the vast area of the Monahans Sandhills to the south and west of the Llano (the only gopher reported from this sandy tract is Geomys bursanus knoxjonesi — Hollander et al., 1987). Extensive field work in this area of Texas has produced no populations of Cratogeomys that could link these two subspecies. Until additional material from intermediate areas is available to suggest otherwise, the two appear to be geographically isolated. Etymology.— It gives me great pleasure to name this subspecies in honor of Walter W. Dalquest of Midwestern State University. Dr. Dalquest has made numerous contributions to our knowledge of mammalian faunas of Texas, especially Pleistocene faunas. Through his efforts and those of his students, the largest series of this taxon available for study is housed in the mammal collection of Midwestern State University. Specimens examined. — Total of 95 as follows. Texas. Glasscock Co. : 20 mi. SSE Big Spring, 1 (MWSU); 15 mi. N Garden City, 2 (MWSU); 14.5 mi. N Garden City, 1 (MWSU); 12. 7 mi. N Garden City, 1 (MWSU); 12 mi. N Garden City, 1 (MWSU); 11 mi. N Garden City, 1 (MWSU); 10.9 mi. N Garden City, 1 (MWSU); 10. 7 mi. N Garden City, 1 (MWSU); 9.6-9. 7 mi. N Garden City, 5 (MWSU); 6 mi. N Garden City, 2 (ASU); 5. 7 mi. N Garden City, 2 (MWSU); 5.1 mi. N Garden City, 1 (MWSU); 5 mi. N Garden City, 1 (MWSU); 4.7 mi. N Garden City, 1 (MWSU); 3.3 mi. N Garden City, 1 (MWSU); 2 mi. N, 13.7 mi. W Garden City, 1; 2 mi. N, 13 mi. W Garden City, 1; 1.8-1. 9 mi. N, 12.7-12.8 mi. W Garden City, 5; 1.4 mi. N, 13.3 mi. W Garden City, 1; 19-20 mi. S Stanton, 3; 2.6 mi. S jet. hwys. 137 and 158 on 137, 1 (ASU); 1.1 mi. N Garden City, 1 (MWSU); 0.9 mi. N Garden City, 1 (MWSU); 0.6 mi. N Garden City, 2 (MWSU); 10.4-10.6 mi. W Garden City, 3 (MWSU); 0.8 mi. W Garden City, 1 (MWSU); O. 5 mi. S, 13 mi. W Garden City, 1; 0.7 mi. S, 12.4 mi. W Garden City, 2; 1 mi. S, 12.5 mi. W Garden City , 1; 1.3 mi. S, 12 mi. W Garden City, 2; 2 mi. S, 12 mi. W Garden City, 2; 2.4 mi. S, 11.8-12 mi. W Garden City, 4; 0.2 mi. W Sterling County line on hwy. 158, 1 (ASU); Glasscock County only, 4 (NMSU). Howard Co.: 11.2 mi. S Big Spring, 1 (MWSU); 13 mi. SSE Big Spring, 1 (MWSU). Irion Co.: 10.5 mi. Njct. hwys. 163 and 2469, 1 (ASU); 9.2 mi. N jet. hwys. 163 and 2469, 1 (ASU); 22 mi. N Barnhart, 1; 8.2 mi. Njct. hwys. 163 and 2469, 2 (ASU); 3 mi. N jet. hwys. 163 and 2469, 1 (ASU). Reagan Co.: 30 mi. S Garden City, 1; 19.1 mi. N Big Lake, 1 (ASU); Centralia Draw, 18.2 mi. N Big Lake, 1 (ASU); 3 mi. NE Stiles, 1. Sterling Co.: 7.3 mi. N jct. hwy. 163 and U.S. hwy. 87, 1 (MWSU); 6.7 mi. Njct. hwy. 163 and U.S. hwy. 87, 1 (MWSU); 6 mi. Njct. hwy. 163 and U.S. hwy. 87, 1 (MWSU); 9.7 mi. NW Sterling City, 1 (MWSU); 7.5 mi. NW Sterling City, 1 48 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY (MWSU); 2 mi. Njct. hwy. 163 and U.S. hwy. 87, 1 (MWSU); 6.4 mi. NW Sterling City, 2 (MWSU); 1.5 mi. Njct. hwy. 163 and U.S. hwy. 87, 1 (MWSU); 5. 5-5. 6 mi. NW Sterling City, 4 (MWSU); 0.7 mi. Njct. hwy. 163 and U.S. hwy. 87, 2 (MWSU); 4 mi. NW Sterling City, 2 (MWSU); 1 mi. N, 4 mi. W Sterling City, 1; 23.7 mi. E Garden City, 1 (ASU); 25 mi. E Garden City, 1 (ASU). Upton Co.: Rankin, 2; 2 mi. E Rankin, 1; 8 mi. E Rankin, 1. Cratogeomys castanops hirtus Nelson and Goldman Cratogeomys castanops hirtus Nelson and Goldman, Proc. Biol. Soc. Washington, 47:138, 1934. Holotype from Albuquerque, Bernalillo Co., New Mexico. Distribution.— Known only from vicinity of the type locality. See Figures 6 and 1 1 . Description.— Russell (1968 b) described this race as small for the species, but he included specimens of C. c. parviceps from the Tularosa Basin in the description. Nelson and Goldman (1934), in the original description of hirtus, stated it was a dark-colored subspecies closely allied to C. c. lacrimalis (of the Pecos Valley), and limited it to the upper Rio Grande Valley. Selected cranial measurements of an adult female topotype are as follows: condylobasal length, 47.31; zygomatic breadth, 29.84; breadth of rostrum, 10.66; palatofrontal depth, 18.18; length of maxillary toothrow, 10.04. Comparisons .— Only one adult specimen of this subspecies was examined by me. Russell (19686) included some specimens here referred to parviceps within this race and described it as being small. The adult female and a subadult female (the third specimen was a young individual) I examined both are larger than the average of parviceps examined for most cranial characters and approximate the size of C. c. clarkii of the Trans-Pecos. Remarks. — This taxon is poorly known and peripherally isolated from the main body of the species in the United States. Specimens from southwestern New Mexico that were assigned to this race by Russell (19686) are separated from the type locality by the large expanse of the Jornado del Muerto, an area from which no pocket gophers of any kind have been reported. Specimens from Rhodes Pass and El Paso are clearly referable to parviceps (see account of that subspecies for details). Of the four specimens of this subspecies (as here defined) that are known, three were obtained in 1894. The fourth was not collected until 1962, and none has been acquired since that time. Until additional material from the vicinity of the type locality is available for study, an accurate description of hirtus cannot be written nor can its status be ascertained with confidence. Specimens examined. — Total of 3 as follows. New Mexico. Bernalillo Co.: Albuquerque, 2 (USNM); South Valley, Albuquerque, 1 (MSB). HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 49 Fig. 11. — Map showing the distributional limits (shading) and localities for specimens examined (dots) of subspecies of C. castanops in New Mexico. Cratogeomys castanops lacrimalis Nelson and Goldman Cratogeomys castanops lacrimalis Nelson and Goldman, Proc. Biol. Soc. Washington, 47:137, 1934. Holotype from Roswell, Chaves Co., New Mexico. Distribution.— Pecos River Valley west of the Llano Estacado from Gua¬ dalupe County, New Mexico, southwardly to Reeves, Ward, and Winkler counties, Texas. See Figures 6, 7, and 11. 50 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Description.— Medium to large-sized subspecies characterized by large lacrimal bones that project posteriorly into the orbit (see Fig. 10). Size slightly smaller than in C. c. perplanus and slightly larger than in C. c. castanops and C. c. dalquesti. Comparisons From C. c. perplanus , the race occurring on the Llano Estacado to the east, lacrimalis differs in being smaller in most cranial dimensions and in having enlarged lacrimals. Size and shape of the lacrimal of C. c. lacrimalis , C. c. perplanus , and C. c. dalquesti are compared in Figure 10. For comparisons with C. c. clarkii, C. c. dalquesti, and C. c. parviceps, see accounts of those subspecies. Remarks Russell (1968/?) assigned specimens from the Pecos Valley of eastern New Mexico and northwestern Texas to perplanus, placing lacrimalis in synonymy. In discriminant function analysis, individuals here referred to lacrimalis (group 9, Fig. 3) were treated as unknowns so as to ascertain if they more closely resembled morphologically one of the a priori clusters (A-E). These specimens were classified almost equally into C. c. castanops (cluster A), C. c. perplanus (cluster B), C. c. clarkii (cluster C), and C. c. dalquesti (cluster E), indicating no stronger alliance with any one than the others. Specimens from above the western escarpment of the Llano Estacado (which is not as distinct as the eastern escarpment) were almost always classified as perplanus. These results and the characteristic size and shape of the lacrimal bone (discussed above) indicated the need for resurrecting the name lacrimalis for this distinct race inhabiting the Pecos Valley. There is an area of integration between lacrimalis and perplanus in the vicinity of Maljamar, Lea Co., New Mexico. This area is just above the caprock and, based on the size of gophers and variation in the shape of the lacrimal bone, gene flow between the two races takes place in this area. All specimens from there, however, are assignable to perplanus. Specimens examined. — Total of 106 as follows. New Mexico. Chaves Co.: 20 mi. W Roswell, 1 (NMSU); 2 mi. E Roswell, 1 (MSB); Roswell and 0.5 mi. S, 12 (11 KU, 1 ENMU). De Baca Co.: 2. 7-3.4 mi. S, 0.3 mi. E Taiban, 4 \ 5.8 mi. S, 0.3 mi. E Taiban, 2; 13 mi. S, 0.75 mi. W Taiban, 2. Eddy Co.: Artesia, 7 (NMSU); 3 mi. NW Carlsbad, 1 (MSB); Carlsbad, 16 (3 KU, 1 MHP, 12 NMSU); 1-2 mi. E Carlsbad, 10 (KU); 6 mi. S, 22 mi. E Carlsbad, 1 (ENMU); 5 mi. S, 1 mi. E Black River Village, 1 (KU); 2 mi. S, 1 mi. W White City, 1 (KU); Rattlesnake Springs, CCNP, 2 (KU); 0.8 mi. N, 8.5 mi. E jet. hwy. 128 and 31 on 128, 2 (MSB). Guadalupe Co.: 1 mi. S Santa Rosa, 4; Catfish Falls, Los Esteros Lake Site, 2 (MSB). Lea Co.: 2 mi. S, 7 mi. EJal, 1 (ENMU). Roosevelt Co.: 15.3-15.4 mi. W Floyd, 4; 11.8-13 mi. W Floyd, 17. Texas. Loving Co.: Red Bluff Lake Dam, 1 (SRSU); 2 mi. N, 5 mi. E Mentone, 1; 2 mi. W Mentone, 1; 1 mi. W Mentone, 3; 0.5 mi. W Mentone, 1. Reeves Co.: 15 mi. N Pecos, 2 (SRSU); 5 mi. S, 10 mi. E Pecos, 1. Ward Co.: 2 mi. W Barstow, 2. Winkler Co.: 5.5 mi. W Kermit, 1; Kermit, 2; 2.6 mi. N Wink, 1; 1.25 mi. N, 3 mi. W Wink, 1. HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 51 Cratogeomys castanops parviceps (Russell) Pappogeomys castanops parviceps Russell, Univ. Kansas Publ., Mus. Nat. Hist., 16:673, 1968. Holotype from 18 mi. SW Alamogordo, Otero Co., New Mexico. C[ratogeomys ]. c[astanops]. parviceps, Jones, Jones, and Schmidly, Occas. Papers Mus., Texas Tech Univ., 119:11, 1988. Distribution. — Tulorosa Basin of southwestern New Mexico and western Texas. See Figures 6, 7, and 11. Description.— A small subspecies with a short, narrow skull; pelage typi¬ cally dark both dorsally and ventrally, but some individuals from lower elevations are paler. See Table 4, group D, for means and ranges of cranial measurements. Comparisons.— From C. c. lacrimalis of the Pecos Valley to the east, parviceps differs in being much smaller in all cranial dimensions, and specimens of parviceps are typically much darker than those of lacrimalis. For comparisons with C. c. clarkii and C. c. hirtus, see accounts of those subspecies. Remarks .— Russell (19686 ) allocated C. c. parviceps to the subnubilus subspecies-group of C. castanops , a subspecies-group that he restricted primarily to Mexico; parviceps was the only member found in the United States. This subspecies-group was characterized by small size, both exter¬ nally and cranially. Although parviceps is one of the smallest races in the United States, it is no smaller than C. c. angusticeps and is similar in size to C. c. tamaulipensis. These latter two races were allocated to the larger excellsus subspecies-group by Russell (19686 ). In a recent study of the louse-host associations with C. castanops , Hellenthal and Price (1976) redefined the distribution of Russell’s excellsus- group and subnubilus -group based on the distribution of species of Geomydoecus . Their new distribution corresponded with the suggested distribution of Berry and Baker (1972), and later Lee and Baker (1987), of the two groups of gophers based on chromosome number. All these data suggest that the subnubilus -group (which Lee and Baker, 1987, believed is specifically distinct from C. castanops ) should be restricted to the southern Mexican Plateau, and that parviceps should be placed in the excellsus -group. The results of the analyses of morphological data in this study support these conclusions. Russell (19686 ) also reported an area of sympatry between parviceps and specimens of C. c. perplanus (here assigned to C. c. lacrimalis ) in the Guadalupe Mountains of western Texas. In a study of the mammals of this area, Genoways et al. (1979) reported only C. c. parviceps. They addressed the problem of sympatry between the two races reported by Russell (19686 ) and stated that they never obtained specimens of C. castanops from east of the mountains despite extensive efforts to do so. All specimens available to me from this area are clearly referable to parviceps ; lacrimalis seems to be restricted to lower elevations in the Pecos Valley. 52 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Specimens examined.— Total of 144 as follows. New Mexico. Donna Anna Co.: 1.8 mi. N, 11.7 mi. E Organ, 1 (NMSU); Parker Lake, E of Organ Mts., 4 (USNM). Lincoln Co.: Ancho, 1 (USNM); 1.4 mi. N, 0.8 mi. W Carrizozo, 1 (NMSU); 1.2 mi. N, 1 mi. W Carrizozo, 3 (NMSU); 1.2 mi. W Carrizozo, 5 (UTEP); 0.5 mi. S, 2.7 mi. E Carrizozo, 6 (NMSU); 0.4 mi. S, 1.6 mi. W Paton Mt., 2 (NMSU). Otero Co.: 5 mi. S Alamogordo, 1; 5 mi. SW Alamogordo, 1 (KU); Holloman AFB golf course, 3 (NMSU); 1.5 mi. N, 1.9 mi. W White Sands National Monument HQ, 2 (NMSU); White Sands National Monument, 3; 18 mi. SW Alamogordo, 6 (2 MSB, 3 KU, 1 MVZ); T20S, R12E, N 1/2 sec. 9, 1 (UTEP). Sierra Co.: 3.9 mi. S, 13.2 mi. W Salinas Peak, 33 (NMSU); 4.9 mi. S, 12.1 mi. W Salinas Peak, 13 (NMSU); 4.2 mi. N, 17.7 mi. E Engle, 1 (NMSU); 6.2 mi. S, 10.2 mi. W Salinas Peak, 1 (NMSU); 8.5 mi. S, 11.2 mi. W Salinas Peak, 1 (NMSU); 8.5 mi. S, 11.8 mi. W Salinas Peak, 2 (NMSU); Rhodes Pass, San Andres Mts., 5 (UTEP). Texas. Culberson Co.: 7 mi. N Pine Springs, GMNP, 1 (TCWC); Scott Canyon, GMNP, 2 (TCWC )\ Pine Springs Canyon, GMNP, 1 (TCWC ), Delaware Springs, 1 (SRSU); 2 mi. SSE El Capitan, 1 (MWSU). El Paso Co. : 1.1 mi. N, 1.7 mi. E Hueco Tanks State Park, 1 (UTEP); south side of Hueco Tanks State Park, 4 (1 SRSU, 3 UTEP); Hueco Poison, 1 (UTEP); 0.7-0. 8 mi. N hwy. 180/62 on FR 2775, 6 (UTEP); 11 mi. N, 12.5 mi. E El Paso on hwy. 180/62, 1 (UTEP); 11 mi. N, 13 mi. E El Paso on hwy. 180/62, 1 (UTEP); 10 mi. N, 11 mi. E El Paso, 1 (UTEP); 15.2 mi. E El Paso on hwy. 180/62, 1 (UTEP); 0.7 mi. S Carlsbad hwy. and 13.2 mi. E Cinema Park [ = El Paso], 1 (UTEP); 2 mi. S Carlsbad hwy. and 14 mi. E Cinema Park [ = El Paso], 1 (UTEP); El Paso, 1 (UTEP); Municipal golf course, El Paso, 1 (MVZ); 9. 3-9. 4 mi. NE hwy. I 10 on Fabens-Carlsbad Rd., 11 (UTEP); El Paso County only, 1 (UTEP). Hudspeth Co.: Lewis Well, GMNP, 1; 1 3/8 mi. N, 4.25 mi. W Guadalupe Peak, GMNP, 2; 3.5 mi. E Salt Flat, [recorded as from Culberson County] 1 (SRSU); southern Hueco Mts., 3 mi. E Horizon Lake, 1 (UTEP); 24 mi. W Cornudas, 1 ; 3.6 mi. W Cornudas, 1; 2 mi. W Cornudas, 1; 1 mi. E Cornudas, 1. Cratogeomys castanops perplanus Nelson and Goldman Cratogeomys castanops perplanus Nelson and Goldman, Proc. Biol. Soc. Washington, 47:136, 1934. Holotype from Tascosa, 3000 ft., Oldham Co., Texas. Pappogeomys castanops simulans Russell, Univ. Kansas Publ., Mus. Nat. Hist., 16:656, 1968. Holotype from 17 mi. SE Washburn, Armstrong Co., Texas. Distribution.— Northeastern New Mexico eastward through the Oklahoma Panhandle, and on the High Plains of the Texas Panhandle southwardly on the Llano Estacado at least as far as Martin County, Texas, and central Lea County, New Mexico. See Figures 6, 7, 11, and 12. Description.— Largest subspecies of C. castanops in the United States; color about as in C. c. lacrimalis\ lacrimal bone small and articulating more with maxilla than with frontal (Fig. 10). See Table 4, group B, for means and ranges of the cranial measurements. Comparisons.— From C. c. clarku, of the Trans-Pecos, perplanus is geographi¬ cally isolated by the Monahans Sandhills and the escarpment of the Llano Estacado. For comparisons with the subspecies castanops , dalquesti, and lacnmalis, see accounts of those taxa. Remarks.— Dowler and Genoways (1979) allocated specimens referred by Russell (19686 ) to C. c. simulans to C. c. perplanus and placed the name simulans in synonymy. Russell (19686) reported that specimens of simulans occurred east of the caprock (Llano Estacado escarpment) in the Texas Panhandle and that this was the barrier between simulans and perplanus. Dowler and Genoways (1979), however, documented that no specimens of HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 53 Fig 12.— Map showing the distributional limits (shading) and localities for specimens examined (dots) of C. c. perplanus in Oklahoma. C. castanops were known from east of the caprock, and demonstrated, with the much larger sample available to them, that simulans and perplanus were indistinct at the subspecific level. My findings agree with theirs. For details on situations where perplanus comes into contact with the subspecies castanops , dalquesti, and lacrimalis, see accounts of those taxa. Specimens examined. — Total of 654 as follows. New Mexico. Chaves Co.: 5.7 mi. N, 27.7 mi. E Hagerman, 1; 33 mi. E Lake Arthur, 1; 7 mi. N Maljamar, 4 (KU). Colfax Co.: Chico Springs, 2 (USNM). Lea Co.: 4.2 mi. W Crossroads, 1; 2. 7-3.5 mi. W Crossroads, 10; 0.7-1 mi. W Crossroads, 9; 7.2 mi. N, 2.1 mi. W Maljamar, 2; 5.8 mi. TV, 1.5 mi. W Maljamar, 1 ,3.5 mi. TV, 0.5 mi. W Maljamar, 3; 2 mi. N, 3 mi. E Maljamar, 1; 0.6-1. 4 mi. N, 0.5-0. 6 mi. E Maljamar, 8; 11 mi. E Maljamar, 1; 3 mi. N Hobbs, 4 (KU). Roosevelt Co. : Portales, 1 (ENMU); 0.9 mi. S, 1.3 mi. W Kenna, 1. San Miguel Co. : 1 mi. S, 2 mi. W Conchas Dam, 5 (KU). Union Co.: Pepper Ranch, 9 mi. N, 34 mi. E Folsom, 1 (ENMU); 29.5 mi. N, 0.9 mi. E Mt. Dora, 1 (ENMU); 6 mi. N, 2.6 mi. E Moses, 1 (ENMU); 3.3 mi. N, 2.2 mi. E Moses, 1 (ENMU); 3 mi ENE Seneca, 1 (MWSU); Rabbit Ears Mt., 2 (MWSU); 6.5 mi. N, 3.5 mi. E Clayton, 3 (ENMU); 4.1 mi. N, 9 mi. E Clayton, 3 (ENMU); 2.9 mi. TV, 1.7 mi. W Clayton, 1 (ENMU); 1.1 mi. TV, 2.3 mi. W Clayton, 1 (ENMU); 6.6 mi. S, 3.5 mi. E Mt. Dora, 2 (ENMU); 0.5 mi. E Clayton Lake, 1 (ENMU); 9.7 mi. S Clayton, 2 (ENMU); 4.5 mi. N, 5.2 mi. E Pasamonte, 1 (ENMU); 1 .3 mi. E Gladstone, 2 (ENMU); 2. 2 mi. E Gladstone, 1 (ENMU); 0.6 mi. N, 0.5 mi. W Amistad, 1 (ENMU). Oklahoma. Beaver Co.: 4 mi. E Elmwood Post Office, 4 (OSU). Cimarron Co.: 8.3 mi. N, 0.6 mi. E Kenton, 1 (MHP); 7 mi. N Kenton, 7 (KU); 5 mi. TV, 0.5 mi. W Kenton, 3 (ENMU); 5 mi. TV Kenton, 1 (MWSU); 4.5 mi. NNW Kenton, 2 (MWSU); 4.4 mi. TV, 4 mi. W Kenton, 1 (OMNH); 4.1 mi. N, 8.2 mi. E Kenton, 1 (MHP); 4 mi. TV Kenton, 3 (MWSU); 3 mi. TV, 1 mi. E Kenton, 3; near Kenton, 1 (OMNH); 1 mi. E Kenton, 3; 0.4 mi. S, 3.4 mi. E Kenton, 3 (ENMU); 1.5 mi. S, 3 mi. E Kenton, 3; 4-4.1 km. SE Kenton, 1 1 (OMNH); 3 mi. S, 3 mi. E Kenton, 1; 5 mi. E Kenton, 1 (MHP); 3.75 mi. S, 9.25 mi. E Kenton, 1 (OMNH); 25 mi. NW Boise City, 1 (OMNH). Texas Co.: Railroad right-of-way just W Hooker, 5 (OSU). 54 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Texas. Armstrong Co.: 8 mi. S, 7 mi. W Claude, 1 (KU); 17 mi. SE Washburn, 8 (TNHC). Bailey Co.: 5.5 mi. S, 2 mi. W Needmore, 1; 22 mi. S Muleshoe, 2. Cochran Co.: 0.5 mi. N, 2.7 mi. W Morton, 1; 1.2 mi. S, 1.5 mi. E Morton, 1. Dallam Co. : 12 mi. E Texline, 1; 1.7 mi. S, 0.3 mi. W Texline, 1 (NMSU); 2.4 mi. S, 0.3 mi. W Texline, 1 (NMSU). Dawson Co.: 11.1 mi. N, 3 mi. E Lamesa, 3 (MHP); 11.1 mi. TV, 4.3 mi. E Lamesa, 5 (MHP); Lamesa, 1 (MWSU); 10 mi. E Lamesa, 2 (TNHC); 2.2 mi. S, 0.3 mi. E Lamesa, 1; 2.3 mi. S Lamesa, 7; 2.8 mi. S, 0.6 mi. E Lamesa, 2; 2.9 mi. S Lamesa, 4; 3.3 mi. S, 4 mi. E Lamesa, 5; 4.6 mi. S, 4.3 mi. E Lamesa, 1; 5-5.2 mi. S, 4.5-5 mi. E Lamesa, 6; 6 mi. S, 5.7 mi. E Lamesa, 6; 6.8 mi. S, 1.3 mi. E Lamesa, 4; 22 mi. SW Lamesa, 5 (ASU); 1 mi. NNW Ackerly, 1. Deaf Smith Co.: 1 mi. TV, 18.3 mi. W Hereford, 15; 1 mi. N, 17.9 mi. W Hereford, 4; 1 mi. N, 16.4 mi. W Hereford, 7; 1 mi. TV, 15.5 mi. W Hereford, 1. Floyd Co.: 0.4 mi. TV, 1.8 mi. W Aiken, 1; 0.4 mi. N, 1.2 mi. W Aiken, 12; 0.4 mi. N, 1.8 mi. W Lockney, 7; 1.5 mi. W Lockney, 1; 1.3 mi. W Lockney, 2; 1 mi. W Lockney, 5; 0.5 mi. W Lockney, 1. Gaines Co. : 4.4 mi. TV, 9.3 mi. W Seminole, 19; 4.4 mi. N, 7.6 mi. W Seminole, 1; 4.4 mi. TV, 6. 2-6. 6 mi. W Seminole, 5; 0.8 mi. N, 6.3 mi. E Seminole, 1; 0.8 mi. S, 15 mi. E Seminole, 1; 3 mi. SW Seminole, 1. Hale Co.: Plainview, 7 (2 MWSU, 5 TTU); 3 mi. S Plainview, 2; 3 mi. N, 1 mi. E Abernathy, 5 (MHP). Hansford Co.: 5 mi. SW Gruver, 2 (MWSU); 6 mi. S, 3 mi. W Gruver, 1 (KU); 11 mi. SSW Gruver, 1 (MWSU). Hartley Co.: 1 mi. N, 8 mi. W Channing, 1. Hockley Co.: 6 mi. SE Anton, 6 (4 ASU, 2 TTU); Yellow House Ranch, 1 (SRSU); 1 mi. N, 4.3 mi. W Levelland, 3; 1 mi. TV, 1 mi. W Levelland, 1 ; 0.5 mi. TV, 3. 2-3. 5 mi. W Levelland, 20; 0.5 mi. TV Levelland, 1; Levelland, 1; 2 mi. E Smyer, 2 (ASU); 3 mi. SW Levelland, 1; 2 mi. S, 3.8 mi. W Levelland, 2; 2 mi. S, 2.8 mi. W Levelland, 7; 7 mi. S Levelland, 1; 0.5 mi. W Sundown, 1; Ropesville, 3. Howard Co.: Big Spring, 1 (USNM); 1 mi. from jet. I 10 and Cauble Rd., 1 (ASU). Lamb Co.: 1.3 mi. E Earth, 1; 4.8 mi. S, 0.3 mi. W Earth, 5; 1.8 mi. N Littlefield, 3; 1.5 mi. N, 1.5 mi. W Littlefield, 4; Littlefield, 1 (TNHC); 0.5 mi. S, 3 mi. W Littlefield, 5; 1.5 mi. S, 1.8 mi. E Littlefield, 5. Lipscomb Co.: 5 mi. S Booker, 4 (WTSU). Lubbock Co.: 10-10.1 mi. N Lubbock, 7; 7. 5-8. 5 mi. N Lubbock, 101 (1 MWSU, 100 TTU); Airport, 5-6.3 mi. N Lubbock, 26 (1 MWSU, 25 TTU); 5 mi. NW Lubbock, 1 (ASU); 4.4 mi. N, 2.5 mi. E Lubbock, 1; Mackenzie Park, 2-3 mi. NE Lubbock, 7; 1.3 mi. TV, 2.3 mi. W Lubbock, 1; 1 mi. N, 10 mi. W Lubbock, 1; 6.5-7 mi. W Lubbock, 4 (1 MHP, 3 TTU); 5-5.5 mi. W Lubbock, 13; 4 mi. W Lubbock, 1; Lubbock, 29 (1 TAI, 24 TTU, 4 MWSU); 6 mi. E Lubbock, 1 (MWSU); 1 mi. S, 7 mi. W Lubbock, 2; 2.5 mi. S, 4.5 mi. E Lubbock, 1; 4 mi. S, 5.7 mi. E Lubbock, 1. Lynn Co.: 1 mi. E West Point, 1; Tahoka, 1 (MWSU); 3 mi. S Tahoka, 1 (MHP). Martin Co.: Stanton, 1 (USNM). Moore Co.: 3 mi. S Dumas, 2. Ochiltree Co.: 11 mi. S, 4 mi. E Perryton, 1; 12 mi. S, 9 mi. E Perryton, 9. Oldham Co.: 3 mi. W Boy’s Ranch Headquarters, 1; Tascosa, 2 (USNM); 17 mi. N, 1 mi. W Adrian, 1; 20.2 mi. NW Vega, 2 (SRSU). Parmer Co.: 0.5 mi. N Friona, 1 (MHP). Potter Co.: 3.5 mi. W Amarillo, 1; 2 mi. E Amarillo, 2 (TCWC). Randall Co.: 1 mi. N, 4.8 mi. E Canyon, 4; 0.2 mi. TV, 6.5 mi. E Canyon, 2; 2-2.6 mi. E Canyon, 7; 3 mi. E Canyon, 6; 4.8 mi. E Canyon, 1; 5 mi. S Canyon, 1; Palo Duro Canyon, 1 (TWC). Sherman Co.: Stratford, 3; 8 mi. S, 2 mi. E Stratford, 1. Swisher Co.: 4.5 mi. S County Line (probably the small country store referred to locally as County Line approximately 19 mi. W Tulia), 1. Terry Co.: 1.7 mi. S, 0.5 mi. W Meadow, 1; 11.2 mi. W Brownfield, 1; 6 mi. W Brownfield, 3; Brownfield golf course, 4; near Brownfield, 5. Yoakum Co.: 1.6 mi. E Plains, 1; 10.7 mi. W Plains, 1. Cratogeomys castanops tamaulipensis Nelson and Goldman Cratogeomys castanops tamaulipensis Nelson and Goldman, Proc. Biol. Soc. Washington, 47:141, 1934. Holotype from Matamoros, Tamaulipas. Distribution.— Recorded from the United States only from the vicinity of Brownsville, Cameron Co., Texas; known also along the lower reaches of the Rio Grande in Tamaulipas. See Figures 6 and 7. Description.— Small subspecies that approaches the size of C. c. clarkii and C. c. castanops; characterized by dark postauricular patches. Geographically HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 55 isolated from other Cratogeomys in the United States by the lower Rio Grande Valley, which is occupied by Geomys per sonatus. Comparisons.— For comparison with C. c. angusticeps , see account of that subspecies. Remarks.— Cleveland (1977) first reported C. castanops from southern Texas and noted that the nearest record of occurrence for the species was from across the Rio Grande in Tamaulipas. He observed numerous burrows southeast of Brownsville in 1976, but stated that no burrows were observed at the same site in 1972. This suggests the possibility of recent invasion by these gophers from the south side of the river even though Hickmann (1977) reported that Cratogeomys was the poorest swimmer among the three genera of pocket gophers in the United States. Cleveland did not assign his specimens to subspecies. These and additional material from the same area were compared by me with topotypic material of tamaulipensis using one-way ANOVAs (sample size was too small for multivariate tests). The results of these univariate tests indicated no significant differences between the two populations for any of 15 cranial characters. Thus, until additional material is available for study, this population is best referred to tamaulipensis . Specimens examined . —Total of 19 as follows. Tamaulipas. 3 mi. SE Reynosa, 3 (KU); Matamoros, 8 (5 USNM, 3 TTU). Texas. Cameron Co.: 5 mi. SE Brownsville, 1 (TWC); 6 mi. SE Brownsville, 1 (TWC); 7.2 mi. SE Brownsville, 6 (TAI). 56 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Discussion Nine subspecies of C. castanops are here recognized as occurring in the United States. They are differentiated primarily by cranial size and some qualitative cranial characters. They can be divided into four general size categories based on cranial dimensions: smallest subspecies— parviceps and angusticeps\ small subspecies — clarkii, hirtus , and tamaulipensis\ medium-sized subspecies— castanops, dalquesti, and lacrimalis ; large subspecies —perplanus. None of the subspecies in any single size category is geographically adjacent to any of the others in that same category. It seems that size parameters that distinguish the races here recognized are morphological expressions of adaptations to particular environmental conditions under which the several populations exist. For example, the largest subspecies, perplanus , occupies the Llano Estacado and the High Plains regions of Texas, Oklahoma, and New Mexico. This area is characterized by relatively deep, rock-free soils (in most areas) as indicated by the vast amount of agriculture in the region. These conditions may favor the increased size of these gophers in that area. Particular conditions under which the other races exist, however, are not as readily apparent because most do not occur in as homogeneous an environ¬ ment as does perplanus. Biogeography and Evolution The earliest fossil representative of a lineage that includes C. castanops is a species of the closely related genus Pappogeomys from late Pliocene deposits in southern Arizona (Russell, 1968a). The presumed origin of castanops was prior to Rancholabrean times, probably in the late Blancan (Russell, 1968 b ), although it was noted that the first known fossils of Cratogeomys were from late Pleistocene (Wisconsin) deposits. Russell (1968a) pointed out that although one species ( castanops ) of Pappogeomys (including Cratogeomys as a subgenus) ranged into the southwestern United States in the late Pleistocene, the genus was essentially Mexican in distribution. Current fossil records of C. castanops (discussed in the introductory remarks and beyond) indicate that the species had occupied most of its current range in the United States by late Pleistocene and early Holocene times. The interpopulation structure of C. castanops was discussed in detail by Russell (1969). Using the then available fossil material as evidence for a southern distribution of the species, he proposed three disjunct populations of castanops during the pluvial maximum of the Wisconsin (Russell, 1969:fig. 1 5) . These populations gave rise to the ‘ ‘subspecies clusters’ ’ and subspecies- groups that he described (Russell, 1968 1969). The availability of additional extant and fossil material (north of his proposed Pleistocene distribution) of C. castanops from the United States demands a reevaluation of Russell’s conclusions. Harris (1985) discussed fossil remains of C. castanops from several stadial sites in the Guadalupe Mountains of southeastern New Mexico and western HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 57 Texas. He demonstrated that populations of castanops were maintained much farther north during the Pleistocene than envisioned by Russell (1968(3; 1968/?; 1969). The occurrence of C. castanops in the southwestern United states during the Pleistocene negates Russell’s (1969) hypothesized post-Wisconsin reinvasion of this region. Although I agree with Russell on the Mexican origin of the species, it is obvious that it has occupied most of its range in the United States, with perhaps the exception of the northern part, since at least mid-Pleistocene times. The validity of the two subspecies-groups (as described by Russell, 1968/? ) recently has been questioned (Berry and Baker, 1972; Hellenthal and Price, 1976; Lee and Baker, 1987). Berry and Baker (1972) described two distinct cytotypes of C. castanops— a. southern population with a diploid number of 42, and a northern population with a diploid number of 46. The distribution of these chromosomal forms did not correspond to the distribution of Russell’s (1968/? ) subspecies-groups. Hellenthal and Price (1976) described the distribution of species of Geomydoecus (lice parasitic on Cratogeomys ) that corresponded with the distribution of the two cytotypes of castanops described by Berry and Baker (1972). Lee and Baker (1987) analyzed the G-banded chromosomes of the two cytotypes and suggested that the two were specifically distinct. Russell (1968/? ) placed C. c. parviceps from southwestern New Mexico in his subnubilus subspecies-group of small-sized gophers. This was the only race of castanops outside of Mexico that he relegated to that group. Russell (1968/? ) distinguished the two subspecies-groups primarily on size of cranial dimensions. As has been pointed out in the preceding accounts, C. c. parviceps averages no smaller than do some of the other races of castanops in the United States, all of which Russell placed in the larger excellsus subspecies-group. The results of my study, based on morphology, together with the results of the previously mentioned studies based on chromosomes and lice, suggest that the subspecies-groups as described by Russell (1968/?: fig. 3) have little basis in fact and thus there is no reason for their continued recognition. Most of the subspecies recognized and discussed in the preceding accounts closely resemble geographically adjacent races. Limited interbreeding may occur between those races not separated by a biogeographic barrier. For example, C. c. castanops and C. c. perplanus appear to intergrade in the area where Colorado, Oklahoma, and New Mexico share a common border; C. c. perplanus and C. c. lacrimalis seem to interbreed in the area of western Lea County, New Mexico, where the western escarpment of the Llano Estacado does not appear to be of sufficient magnitude to preclude gene flow. However, along the southeastern edge of the Llano, in Howard and Martin counties, Texas, the distributions of C. c. perplanus and C. c. dalquesti approach each other geographically but there is no apparent gene flow between the two races, one on the Llano Estacado and one to the south of 58 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY it. A similar situation exists between C. c. dalquesti and C. c. clarku in the area of Upton and Crane counties in west-central Texas. Here, a north¬ western extension of the Edwards Plateau escarpment (which is occupied by another gopher, Thomomys bottae ) separates the two races and there is no apparent interbreeding. Cratogeomys castanops dalquesti is geographically isolated from the closely related C. c. lacrimalis by the Monahans Sandhills, which are occupied by Geomys bursarius, and the southern Llano Estacado, which is occupied by C. c. perplanus , and no interbreeding can take place in the absence of geographic contact. The Sierra Diablo Mountains of western Texas seem to limit gene flow between C. c. clarkii and C. c. parviceps to a degree, but hybridization does appear to occur. Although there is no apparent barrier to gene flow between C. c. clarkii and C. c. lacrimalis in the northern Pecos Valley of Texas, I have been unable to detect intergradation there. Additional material may reveal that hybridization does in fact take place. The three remaining subspecies of C. castanops in the United States are geographically isolated from the main population and from each other. Cratogeomys castanops hirtus is restricted to the northern Rio Grande Valley in the vicinity of Albuquerque, New Mexico. So few specimens of this taxon are known that an accurate determination of its status is not possible at this time. The other two races, C. c. angusticeps and C. c. tamaulipensis , are isolated along the lower Rio Grande; C. c. angusticeps is restricted to the vicinity of Eagle Pass, Maverick County, Texas, whereas C. c. tamaulipensis occurs in Tamaulipas and in the United States only in the vicinity of Brownsville, Cameron County, Texas. The historical center of distribution of C. castanops in the United States seems to be Trans-Pecos Texas and southeastern New Mexico. This is the area where some of the oldest fossils of the species have been found (Harris, 1985). The subspecies that occur in this area today are C. c. clarkii and C. c. lacrimalis , which closely resemble each other. Assuming that this area was the geographic center of the species in the United States in late Pleistocene times, the current distribution of the subspecies may have proceeded as follows. A population extended up the Pecos River Valley (currently C. c. lacrimalis ) and from there gophers migrated onto the Llano Estacado, moved up the Llano to the Canadian River Valley and onward into southeastern Colorado, finally occupying the Arkansas River Valley. They also crossed the Canadian onto the High Plains of the Texas and Oklahoma panhandles. Populations on the Llano Estacado and adjacent High Plains, in the presence of optimal ecological conditions, differentiated into the large gopher here recognized as C. c. perplanus. The population in southeastern Colorado, in the presence of less favorable conditions and possibly more intergeneric competition, differentiated much in the same way as did perplanus but maintained a smaller size. This population, here recognized asC. c. castanops , moved down the Arkansas River into Kansas as well. HOLLANDER-CRATOGEOMYS CASTANOPS IN THE UNITED STATES 59 The original stock also migrated to the south and east. Gophers that became isolated east of the Pecos River, between the Llano Estacado and the Edwards Plateau, are now geographically separated from those in the Pecos Valley of southeastern New Mexico (yet the two populations closely resemble each other), and are here recognized as C. c. dalquesti. The southern population differentiated by becoming smaller, probably as an adaptation to the more xeric conditions of the Chihuahuan Desert, and is here referred as C. c. clarkii. This subspecies probably has recently crossed the Rio Grande into Mexico (Russell, 1968/?). A western extension of this population migrated into the Tularosa Basin of southwestern New Mexico and western Texas and differentiated into one of the smallest races in the United States, C. c. parviceps. The paucity of specimens from the isolated population in the northern Rio Grande Valley make an accurate determination of its affinities difficult. It may represent a relic of an earlier northward surge of gophers up the Rio Grande Valley. In any event, it represents a distinctive race (C. c. hirtus ) that, based on size alone, probably is more closely related to C. c. lacrimalis of the Pecos River Valley to the east. The populations of gophers along the southern Rio Grande (C. c. angusticeps and C. c. tamaulipensis ), because of their isolation from the main populations in the United States, probably are more closely-related to, and originated from (by breaching the Rio Grande), populations of C. castanops in Mexico as suggested by Russell (1969). Acknowledgments This research was submitted as partial requirements for the degree of Doctor of Philosophy to Texas Tech University. I am indebted to my committee members, J. Knox Jones, Jr., Clyde Jones, Sankar Chatterjee, Raymond C. Jackson, and Michael R. Willig, for their advice at various stages of the work and review of the resultant manuscript. This study would not have been possible without the efforts of numerous past collectors of pocket gophers throughout the range of Cratogeomys castanops. To all of them I am grateful. For directly assisting me in the field, I also thank the following individuals: Darryl K. Burton, Kirk Eddelman, Christopher Holder, Douglas Holder, Roland Holder, Rebecca S. Hollander, Clyde Jones, J. Knox Jones, Jr., and Richard W. Manning. I am also indebted to all the landowners who allowed me to collect material on their properties. I am most grateful to all of the curators who allowed me to examine specimens in their care or made specimens available on loan. Each of them is listed in the text. For assistance with gathering the mensural data, the following individuals are acknowledged: Darryl K. Burton, Rebecca S. Hollander, Robert Huber, Fredrick B. Stangl, Jr., and Larry L. Choate. Field work was supported in part by grants from The Graduate School, Texas Tech University, the National Institutes of Health, and the Theodore Roosevelt Memorial Fund of the American Museum of Natural History. 60 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Literature Cited Bailey, V. 1905. Biological survey of Texas. N. Amer. Fauna, 25:1-222. . 1932. Mammals of New Mexico. N. Amer. Fauna, 53:1-412. Baird, S. F. 1852. Mammals. Pp. 309-313, in Exploration and survey of the valley of the Great Salt Lake of Utah, .... Lippincott, Grambo & Co., Philadelphia, 2 vols. . 1855. Characteristics of some new species of Mammalia collected by the U. S. and Mexican boundary survey, Major W. H. Emory, U. S. A., commissioner. Proc. Acad. Nat. Sci. Philadelphia, 7:331-333. Berry, D. L., and R. J. Baker. 1972. Chromosomes of pocket gophers of the genus Pappogeomys , subgenus Cratogeomys. J. Mamm., 53:303-309. Birney, E. C., J. K. Jones, Jr., and D. M. Mortimer. 1971. The yellow-faced pocket gopher, Pappogeomys castanops, in Kansas. Trans. Kansas Acad. Sci., 73:368-375. Cleveland, A. G. 1977. First South Texas records of Pappogeomys castanops. Texas J. Sci., 24:299. Corruccini, R. S. 1987. Univariate versus multivariate morphometric variation: an alternate view-point. Syst. Zook, 36:397-398. Coues, E. 1875. Abstract of results of a study of the genera Geomys and Thomomys: with addenda on the osteology of Geomyidae, and on the habits of Geomys tuza. Rept. Maj. J. W. Powell’s Explor. Colorado River . . . and its tributaries, 1869, 1870, 1871, and 1872, pp. 215-291. Dalquest, W. W., and W. Kilpatrick. 1973. Dynamics of pocket gopher distribution in the Edwards Plateau of Texas. Southwestern Nat., 18:1-9. Davidow-Henry, B. R., and J. K. Jones, Jr. 1988. Notes on reproduction and postjuvenile molt in two genera of pocket gophers, Cratogeomys and Thomomys , in Texas. Texas J. Sci., 40:459-461. Davidow-Henry, B. R., J. K. Jones, Jr., and R. R. Hollander. 1989. Cratogeomys castanops. Mamm. Species, 338:1-6. Davis, W. B. 1940. A new Thomomys from Texas. J. Mamm., 21:204-205. Davis, W. B., and H. K. Buechner. 1946. Pocket gophers ( Thomomys ) of the Davis Mountains, Texas. J. Mamm., 27:265-271. Dowler, R. C., and H. H. Genoways. 1979. Variation in Pappogeomys castanops (Geomyidae) on the Llano Estacado of Texas and New Mexico. Southwestern Nat., 24:577-602. Findley, J. S. 1987. The natural history of New Mexican mammals. Univ. New Mexico Press, Albuquerque, x + 164 pp. Genoways, H. H. 1973. Systematics and evolutionary relationships of spiny pocket mice, genus Liomys. Spec. Publ. Mus., Texas Tech Univ., 5:1-368. Genoways, H. H. , R. J. Baker, andJ. E. Cornely. 1979. Mammals of the Guadalupe Mountains National Park, Texas. Pp. 271-332, in Biological investigations in the Guadalupe Mountains National Park, Texas (H. H. Genoways, and R. J. Baker, eds.), Proc. Trans. Ser., Nat. Park Serv., 4:xvii + 1-442. Gilmore, R. M. 1947. Bones from an archaeological cave site in Coahuila. J. Mamm., 28:147-165. Graham, R. W. 1987. Late Quaternary mammalian faunas and paleoenvironments of the southwestern plains of the United States. Pp. 24-45, in Late Quaternary mammalian biogeography and the environments of the Great Plains and prairies (R. W. Graham, H. A. Semken, Jr., and M. A. Graham, eds.), Illinois State Mus., Springfield, xiv + 491 pp. Hall, E. R. 1981. The mammals of North America. John Wiley & Sons, New York, 2nded., 1 : xv + 1-600 + 90. Harris, A. H. 1985. Late Pleistocene vertebrate paleoecology of the West. Univ. Texas Press, Austin, vii + 293 pp. - . 1987. Reconstruction of mid-Wisconsin environments in southern New Mexico. Nat. Geogr. Res., 3:142-151. HOLLANDER— CRATOGEOMYS CASTANOPS IN THE UNITED STATES 61 Hellenthal, R. A., and R. D. Price. 1976. Louse-host associations of Geomydoecus (Mallophaga: Trichodectidae) with the yellow-faced pocket gopher, Pappogeomys castanops (Rodentia: Geomyidae). J. Med. Ent., 13:331-336. Hendricksen, R. L. 1973. Variation in the plains pocket gopher (Geomys bursarius) along a transect across Kansas and eastern Colorado. Trans. Kansas Acad. Sci., 75:322-368. Hickmann, G. C. 1977. Swimming behavior in representative species of the three genera of North American geomyids. Southwestern Nat., 21:531-538. Hollander, R. R., C. Jones, R. W. Manning, andJ. K. Jones, Jr. 1987. Distributional notes on some mammals from the Edwards Plateau and adjacent areas of south-central Texas. Occas. Papers Mus., Texas Tech Univ., 110:1-10. Honeycutt, R. L., and S. L. Williams 1982. Genic differentiation in pocket gophers of the genus Pappogeomys, with comments on intergeneric relationships in the subfamily Geomyinae. J. Mamm., 63:208-217. Ikenberry, R. D. 1964. Reproductive studies of the Mexican pocket gopher, Cratogeomys castanops perplanus. Unpublished M.S. thesis, Texas Tech Univ., Lubbock, v + 47 pp. Jones, J. K. , Jr. , D. C. Carter, H. H. Genoways, R. S. Hoffmann, D. W. Rice, and C. Jones 1986. Revised checklist of North American mammals north of Mexico, 1986. Occas. Papers Mus., Texas Tech Univ., 107:1-22. Le Conte, J. 1852. [Title unknown]. Proc. Acad. Nat. Sci. Philadelphia, 6:163. Lee, H. Y., and R. J. Baker. 1987. Cladistical analysis of chromosomal evolution in pocket gophers of the Cratogeomys castanops complex (Rodentia: Geomyidae). Occas. Papers Mus., Texas Tech Univ., 114:1-15. Merriam, C. H. 1895. Monographic revision of the pocket gophers, family Geomyidae (exclusive of the species of Thomomys ). N. Amer. Fauna, 8:1-258. Mooser, O., and W. W. Dalquest. 1975. Pleistocene mammals from Aguascalientes, central Mexico. J. Mamm., 56:781-820. Moulton, M. P., J. R. Choate, and S. J. Bissell. 1979. Sympatry of pocket gophers on Mesa de Maya, Colorado. Trans. Kansas Acad. Sci., 82:194-195. - . 1983. Biogeographic relationships of pocket gophers in southeastern Colorado. Southwestern Nat., 28:53-60. Nelson, E. W., and E. A. Goldman. 1934. Revision of the pocket gophers of the genus Cratogeomys. Proc. Biol. Soc. Washington, 47:135-154. Reichman, O. J., and R. J. Baker 1972. Distribution and movement of two species of pocket gophers (Geomyidae) in an area of sympatry in the Davis Mountains, Texas. J. Mamm., 53:21-33. Riddle, B. R., and J. R. Choate. 1986. Systematics and biogeography of Onychomys leucogaster in western North America. J. Mamm., 67:233-255. Rinker, G. C. 1941. Cratogeomys castanops from a Recent terrace in southwestern Kansas. J. Mamm., 22:88. Russell, R. J. 1968a. Evolution and classification of the pocket gophers of the subfamily Geomyinae. Univ. Kansas Publ., Mus. Nat. Hist., 16:473-579. - . 1 968/6 . Revision of pocket gophers of the genus Pappogeomys. Univ. Kansas Publ., Mus. Nat. Hist., 16:581-776. - . 1969. Intraspecific population structure of the species Pappogeomys castanops. Misc. Publ. Mus. Nat. Hist., Univ. Kansas, 51:337-371. Schmidly, D. J. 1977. The mammals of Trans-Pecos Texas: including Big Bend National Park and Guadalupe Mountains National Park. Texas A&M Univ. Press, College Station, xiii + 225 pp. Simpson, G. G. 1945. The principles of classification and a classification of mammals. Bull. Amer. Mus. Nat. Hist., 85:1-350. Smith, M. F., J. L. Patton, J. C. Hafner, and D. J. Hafner. 1983. Thomomys bottae pocket gophers of the central Rio Grande Valley, New Mexico: local differentiation, gene flow, and historical biogeography. Occas. Papers Mus. Southwestern Biol., Univ. New Mexico, 2:1-16. 62 SPECIAL PUBLICATIONS MUSEUM TEXAS TECH UNIVERSITY Smolen, M. J., H. H. Genoways, and R. J. Baker. 1980. Demographic and reproductive parameters of the yellow-cheeked pocket gopher (Pappogeomys castanops ). J. Mamm., ' 61:224-236. Sneath, R. H. A., and R. R. Sokal. 1973. Numerical taxonomy: the principles and practice of numerical classification. W. H. Freeman and Co., San Francisco, xv + 573 pp. Sokal, R. R., and F. J. Rohlf. 1981. Biometry: the principles and practice of statistics in biological research. W. H. Freeman and Co., San Francisco, 2nd ed., xviii + 859 pp. SPSS Inc. 1986. User’s Guide. McGraw-Hill Book Co., 2nd ed., xiv + 988 pp. Villa-R., B., and E. R. Hall. 1947. Subspeciation in pocket gophers of Kansas. Univ. Kansas Publ., Mus. Nat. Hist., 1:217-236. Williams, S. L., and R. J. Baker. 1976. Vagility and local movement of pocket gophers (Geomyidae: Rodentia). Amer. Midland Nat., 96:303-316. Willig, M. R. 1985. Ecology, reproductive biology, and systematics of Neoplatymops mattogrossensis (Chiroptera: Molossidae). J. Mamm., 66:618-628. Willig, M. R., and R. D. Owen. 1987. Univariate analyses of morphometric variation do not emulate the results of multivariate analyses. Syst. Zool., 36:398-400. Willig, M. R., R. D. Owen, and R. L. Colbert. 1986. Assessment of morphometric variation in natural populations: the inadequacy of the univariate approach. Syst. Zool., 35:195-203. Wood, A. E. 1955. A revised classification of rodents. J. Mamm., 36:165-187. Youngman, P. M. 1958. Geographic variation in the pocket gopher, Thomomys bottae, in Colorado. Univ. Kansas Publ., Mus. Nat. Hist., 9:363-384. Address of author: The Museum and Department of Biological Sciences, Texas Tech University, Lubbock, TX 79409. Present address: Department of Biology, Sul Ross State University, Alpine, TX 79832. Received 13 July 1989, accepted 17 July 1989. TEXAS TECH UNIVERSITY PRESS Selected Titles in Mammalogy Carraway, L. N. 1990. A morphologic and morphometric analysis of the “Sorex vagrans species complex” in the Pacific Coast region. Spec. Publ. Mus., Texas Tech University, 32:1-78. $14.00. Dalquest, W. W., F. B. Stangl, Jr., andj. K. Jones, Jr. 1990. Mammalian zoogeography of a Rocky Mountain-Great Plains interface in New Mexico, Oklahoma, and Texas. Spec. Publ. Mus., Texas Tech University, 34:1-78. $14.00. Hollander, R. R. 1990. Biosystematics of the yellow-faced pocket gopher, Cratogeomys castanops (Rodentia: Geomyidae) in the United States. Spec. Publ. Mus., Texas Tech University, 33:1-62. $14.00. Kirkland, G. L., and J. N. Layne, eds. 1989. Advances in the study of Peromyscus (Rodentia). vi + 367 pp. $35.00 (cloth), $22.00 (paper). Martin, R. E., and B. R. Chapman. 1984. Contributions in mammalogy in honor of Robert L. Packard. Spec. Publ. Mus., Texas Tech University, 22:1-234. $30.00 (cloth), $20.00 (paper). Morris, D. W., Z. Abramsky, B. J. Fox, and M. R. Willig, eds. 1989. Patterns in the structure of mammalian communities. Spec. Publ. Mus., Texas Tech University, 28:iv + 1-266. $30.00 (cloth), $20.00 (paper). Occasional Papers, The Museum This series is available by purchase of individual titles ($2.00 each) or by subscription ($16.00 per year for individuals, $19.00 for institutions; add $3.00 for mailing outside North America). 29. Baker, R. J., and H. H. Genoways. 1975. A new subspecies of Geomys bursarius (Mammalia: Geomyidae) from Texas and New Mexico. 18 pp. 58. Honeycutt, R. L., and D. J. Schmidly. 1979. Chromosomal and morphological variation in the plains pocket gopher, Geomys bursarius, in Texas and adjacent states. 54 pp. 105. Armstrong, D. M., J. R. Choate, and J. K. Jones, Jr. 1986. Distributional patterns of mammals in the Plains States. 27 pp. 107. Jones, J. K., Jr., D. C. Carter, H. H. Genoways, R. S. Hoffmann, D. W. Rice, and C. Jones. 1986. Revised checklist of North American mammals north of Mexico, 1986. 22 pp. 110. Hollander, R. R. , C. Jones, R. W. Manning, and J. K. Jones, Jr. 1987. Distributional notes on some mammals from the Edwards Plateau and adjacent areas of south-central Texas. 10 pp. 111. Jones, J. K., Jr., R. W. Manning, R. R. Hollander, and C. Jones. 1987. Annotated check¬ list of Recent mammals of northwestern Texas. 14 pp. 114. Lee, H. K., and R. J. Baker. 1987. Cladistical analysis of chromosomal evolution in pocket gophers of the Cratogeomys castanops complex (Rodentia: Geomyidae). 15 pp. 119. Jones, J. K., Jr., C. Jones, and D. J. Schmidly. 1988. Annotated checklist of Recent land mammals of Texas. 25 pp. 126. Jones, J. K., Jr., R. W. Manning, C. Jones, and R. R. Hollander. 1988. Mammals of the northern Texas Panhandle. 54 pp. Order Information Orders must be accompanied by remittance in U.S. currency in the form of a check, money order, bank draft drawn on a U.S. bank, or MasterCard or Visa. Please include postage ($2.00 first title, $0.75 each thereafter). The Press pays domestic postage on prepaid orders of five or more books. Foreign postage is $3.00 per title. Texas residents please add 7.5% sales tax. Please direct orders or requests for a complete list of titles to: Texas Tech University Press, Sales Office, Lubbock, Texas 79409-1037, USA, or call: 1-800-832-4042. EXAS TECH U N I V H R S 1 T Y MCZ _ OCT 2 2 1990 HARVARD MAMMALIAN ZOOGEOGRAPHY ^ '» OF A ROCKY MOUNTAIN-GREAT PLAINS INTERFACE IN NEW MEXICO, OKLAHOMA, AND TEXAS Walter W. Dalquest, Frederick B. Stangl, Jr., and J. Knox Jones, Jr. Special Publications, The Museum TEXAS TECH UNIVERSITY NUMBER 34