OCCASIONAL PAPERS THE MUSEUM TEXAS TECH UNIVERSITY NUMBER 58 JUNE 1979 CHROMOSOMAL AND MORPHOLOGICAL VARIATION IN THE PLAINS POCKET GOPHER, GEOMYS BURSARIUS, IN TEXAS AND ADJACENT STATES Rodney L. Honeycutt and David J. Schmidly The plains pocket gopher, Geomys bursarius, has an extensive distribution in Texas, occurring throughout most of the state save for the southwestern region and South Texas south of the San Antonio River. This species demonstrates considerable geographic variation in Texas and its adjacent states as exemplified by the recognition of 13 subspecies (Hall and Kelson, 1959), eight of which occur in the eastern part of the state. Most taxonomic and distributional studies of G. bursarius have been based on morphological and osteological characters (Baird, 1854; Merriam, 1890, 1895; Bailey, 1905; Davis, 1938, 1940; Villa and Hall, 1947; Baker, 1950; Baker and Glass, 1951). Although such studies have contributed to a better understanding of geo¬ graphic variation, certain systematic problems have remained unsolved with respect to populations in Texas. As noted by Davis (1940), these problems concern the interpretation of the relation¬ ships, distribution, and differentiation of populations east of the Balcones Escarpment. Several aspects of the biology of pocket gophers make syste¬ matic studies of this group (especially those based on conven¬ tional morphological characters) difficult. These rodents form small local populations isolated from one another by topography and indurate soils (Davis, 1940; Kennerly, 1954; Miller, 1964; Vaughan, 1967; Turner et al., 1973). This pattern of distribution, defined by Wright (1943) as “island model’’ distribution, in com- 2 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY bination with the pocket gopher’s low vagility greatly restricts migration and gene flow between populations. Where their ranges come in contact, different species and subspecies of pocket gophers do not occur sympatrically, but rather are distributed parapatrically with respect to one another (Miller, 1964; Vaughan, 1967; Thaeler, 1968c), making interpretation of the degree of reproductive isolation difficult. Soil is the major factor influencing the distribution of pocket gophers (Davis, 1938, 1940; Kennerly, 1954; Miller, 1964; Downhower and Hall, 1966), and certain external and cranial dimensions have been correlated with soil characteristics (Davis, 1938; Kennerly, 1954, 1959; Miller, 1964; Hendrickson, 1972). For these reasons it is difficult to deter¬ mine whether morphological differences among populations are due to genetic factors or to environmentally related phenotypic responses. A combination of isolation and phenetic adaptation to local¬ ized conditions, then, probably serves to promote divergence and explains in part the considerable degree of morphological varia¬ tion recorded for Geomys bursarius. These factors also demon¬ strate the need to identify an index or population marker for the determination of genetically distinct populations. Morphological characters have not provided this index and in some cases have added to the already existing confusion. Several authors (Patton and Dingman, 1968; Davis and Baker, 1974; Baker et al., 1975) have noted the value of karyotypes as population markers in systematic studies. Karyotypic information has been used successfully to decipher systematic problems among other fossorial rodents, including Spalax (Wahrman et al., 1969; Raicu et al., 1968), Ctenomys (Reig and Kiblisky, 1969), Thomo- mys (Patton and Dingman, 1968, 1970; Thaeler, 1968a; Went¬ worth and Sutton, 1969; Berry and Baker, 1971; Patton, 1972), and Pappogeomys (Berry and Baker, 1972). Like other fossorial mammals, G. bursarius shows considerable variation in chromosome number and morphology. Several authors (Hart, 1971; Kim, 1972; Baker et al, 1973; Hart, 1978) have investigated karyotypic variation in the plains pocket gopher in Texas and have described eight karyotypically distinct populations (Fig. 1). These populations, referred to as chromo¬ somal races (Baker et al., 1973), can be arranged into three groups on the basis of similar karyotypic characteristics and geographic distribution. These are the lutescens group, chromosomal races A, B, C, D; the attwateri group, including chromosomal races F and G; and the breviceps group, consisting of chromosomal races E HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 3 Fig. 1 .— Distribution of chromosomal races A through H of Geomys bursarius in Texas and adjacent states. The diploid number is represented by 2N and the fundamental number by FN; morphology of the X-chromosome is represented by SM, submetacentric, and Ac, acrocentric. Stars indicate those races exhibiting chromosomal polymorphism. The lutescens chromosomal group is represented by chromosomal races A, B, C, and D; the attwateri group by chromosomal races F and G; and the breviceps group by chromosomal races E and H. and H. The karyotypic characteristics separating these groups involve differences in diploid number (2N), fundamental number (FN), morphology of the X-chromosome, and degree of chromo¬ somal polymorphism found within each region. The lutescens complex is distributed in northwestern Texas, eastern New Mexico, and western Oklahoma. It includes four rec¬ ognized subspecies, G. b. llanensis (race A), G. b. texensis (race A), G. b. knoxjonesi (races A and B), and G. b. major (races A, B, C, and D). Chromosomal variation in this complex involves dif¬ ferences in diploid and fundamental numbers, the former varying from 69 to 72 and the latter from 68 to 72. These differences result primarily from presence or absence of biarmed elements in the autosomal complement. Races A, C, and D exhibit chromosomal polymorphisms, but the sex chromosomes in all four races are constant and consist of an acrocentric X and Y. 4 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Races F and G of the attwateri group are present west of the Brazos River in central and southern Texas, have a diploid number of 70, and possess a fundamental number of either 74 (race F) or 72 (race G). The X-chromosome is always a submeta- centric and the Y a small acrocentric. Chromosomal polymor¬ phism has not been found within either race. The distribution of this group coincides with that of three recognized subspecies, G. b. bvazensis (races F and G), G. b. ammophilus (race G), and G. b. attwateri (race G). The breviceps assemblage appears east of the Brazos River throughout central and eastern Texas, western Louisiana, eastern Oklahoma, and western Arkansas. A diploid number of 74 and a fundamental number of either 72 (race E) or 74 (race H) charac¬ terize this group. Chromosomal polymorphism is unknown within either race; the sex chromosomes are a large submetacen- tric X and an acrocentric Y. Race H, described by Kim (1972), is limited to two localities in central Texas. Race E, on the other hand, has an extensive distribution that encompasses the range of seven currently recognized subspecies, G. b. brazensis (races E and H), G. b. sagittalis, G. b. pratincolus, G. b. terricolus, G. b. lude- mani, G. b. dutcheri, and G. b. breviceps. The designation of three major groups has an historic basis. Davis (1940), in his revision of the genus Geomys, considered populations of the plains pocket gopher as representing two dis¬ tinct species. These were G. lutescens in northwestern Texas, east¬ ern New Mexico, and western Oklahoma, and G. breviceps from central and eastern Texas, eastern Oklahoma, Arkansas, and Louisiana. These remained as full species until Villa and Hall (1947) and Baker and Glass (1951) relegated them to subspecies of G. bursarius on the basis of morphological intergradation at zones of contact. Pocket gophers of the lutescens chromosomal group (including chromosome races A, B, C, and D) in this study correspond to G. lutescens in Texas as defined by Davis (1940). Pocket gophers of the breviceps group (including chromosome races E and H) occur east of the Brazos River in eastern Texas and were referred to the species G. breviceps by Davis (1940); those occurring west of the Brazos River in central and southern Texas (including chromosome races F and G) were not regarded as a separate species by Davis (1940), but rather as distinct subspe¬ cies of G. breviceps. However, because of the karyotypic distinc¬ tions between chromosome races F and G and races E and H, the former two have been arranged in a separate chromosomal group (designated the attwateri group) in this study. HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 5 In order to investigate the relationships, distribution, and amount of differentiation within and among the three chromo¬ somal groups, a restricted study area consisting of 11 counties in eastern Texas was selected. Four chromosomal races are present in these counties; three (races E, F, and G) represent the attwateri and breviceps groups and the fourth (race D) represents one of the western Texas races of the lutescens group. Specifically, the objectives of this study are as follows: 1) to determine the extent of chromosomal variation within and among populations of G. bursarius in an 11-county study area of eastern Texas; 2) to document the distribution of the chromosome races in the study area; 3) to examine factors important to the dis¬ tribution of chromosomally distinct populations and to pocket gophers in general; and 4) to relate chromosomal variation to morphological variation in understanding the systematics of this species in Texas and adjacent states. Study Area The area chosen for intensive study represents 11 counties located in eastern Texas (Fig. 2) and was selected because the ranges of the three major chromosomal groups ( lutescens , attwateri, and breviceps) converge in this region. Soils in the study area can be divided into three types (alluvial, clay, and sandy soil). Alluvial soils are located along river chan¬ nels and two major associations occur along the Brazos River, the Miller-Norwood-Yahola and Miller-Norwood associations, which contain soils consisting of clay to fine sandy loam (Templin et al., 1958). Clay soil associations, ranging in texture from clay to clay loam, occupy large portions of Falls and McLennan counties and half of Milam County. Sandy soil associations are character¬ ized by textures ranging from sandy to sandy loam. Four rivers (Colorado, Brazos, Navasota, and San Jacinto), which represent potential barriers to pocket gopher dispersal, flow transversely across the study area. The Brazos, located in the center of the area, is the most extensive of these and has no less than 12 oxbows along its course. These oxbows represent local¬ ized areas where the river has changed course, as a result of shift¬ ing channels, and has transferred land from one bank to another. The entrance to an abandoned oxbow becomes separated from the new river channel by deposits of silt and, over time, the entire lake is replaced with sedimentation. Larger tracts of land also can be shifted from one bank to the other by the meandering of a 6 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Fig. 2.—Map showing the 11-county study area in eastern Texas. Stippled areas represent alluvial soil associations; diagonal lines, clay soil associations; and unshaded areas, sandy soil associations. Dots along the Brazos River identify the location of the 12 oxbows. The Old River in Burleson County represents a meander scar of the Brazos. river, a common occurrence in the flood plain of larger rivers (Cleland, 1929). Old River in Burleson County represents just such a meander scar (old river channel) of the Brazos. In this area, a considerable amount of land was transferred from what is now Brazos County to Burleson County when the river altered course. Methods and Materials A total of 464 specimens (386 for morphological analysis and 200 for karyotypic analysis) from Texas and Louisiana were uti¬ lized in this study. All were prepared as conventional museum study skins (skin with skull or skeleton only) and deposited in either the Texas Cooperative Wildlife Collection, Texas A&M University (TCWC) or The Museum, Texas Tech University (TTU). HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 7 Fig. 3. —Skull of Geomys bursarius showing cranial measurements used in sta¬ tistical studies. Measurements are as follows: greatest length of skull (GLS), A-A'; basal length (BL), B-B'; breadth of rostrum (BR), C-C'; zygomatic breadth (ZB), D- D'; interorbital breadth (IB), E-E'; breadth of braincase (BB), F-F'; mastoidal breadth (MB), G-G'; length of nasals (LN), A-H; length of rostrum (LR), A-I; length of maxillary toothrow (LTR), J-J'; palatal length (PL), B-K; and palato- frontal depth (PFD), L-L'. Karyotypic Analysis Pocket gophers used in the karyotypic analysis were live- trapped by means of a technique described by Baker and Williams (1972). The in vivo colchicine—hypotonic citrate sequence described by Patton (1967) and modified by Lee (1969) was used to prepare the karyotypes of metaphase chromosomes from bone marrow cells. The diploid number (2N) was determined by count¬ ing at least 10 spreads per slide. A representative spread was pho- 8 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Fig. 4.—Map showing the geographic location of 55 samples of Geomys bur- sarius used for statistical analysis of morphological variation. The enlarged area represents the location of samples from four counties (Robertson, Brazos, Milam, and Burleson) in the center of the 11-county study area in eastern Texas. tographed and a karyotype constructed on the basis of the number of biarmed and uniarmed autosomes and the morphology of sex chromosomes (Patton and Dingman, 1968). Metacentric, submetacentric, subtelocentric, and acrocentric chromosomes were described using the terminology of Patton (1967). The fundamen¬ tal number (FN) was defined as the number of major chromo¬ some arms in the autosomal complement. Morphological Analysis Twelve cranial measurements were taken, as illustrated in Fig. 3, by means of dial calipers and recorded in millimeters (mm.). Only adult animals were used in the morphological analysis. Animals were considered adults if the basioccipital—basisphenoid suture was completely ossified. Males and females were treated separately due to marked sexual dimorphism (Kennerly, 1958; Baker and Genoways, 1975). HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 9 Specimens from approximately 218 localities were grouped into 55 units (Fig. 4) to obtain larger samples for statistical analysis of morphological variation. Care was taken to group localities according to geographic and physiographic regions, known taxo¬ nomic boundaries, major waterways, and known karyotypes. Due to small sample sizes in males, only adult females were used in the analysis of morphological variation. Samples with only one individual were eliminated to mitigate errors resulting from improperly-sexed individuals. Various univariate and multivariate statistical techniques were employed to elucidate patterns of morphological variation among chromosomally distinct samples. Standard statistics (mean, range, standard deviation, standard error of the mean, variance, coeffi¬ cient of variation) were computed for each of the 55 samples with a program of the Statistical Analysis Systems (SAS) designed and implemented by Barr and Goodnight (Service, 1972). Several geo¬ graphic transects were constructed for selected characters from these statistics and displayed in the form of Dice-Leraas diagrams to illustrate the degree of clinal variation in size. Multivariate techniques were used to cluster samples according to phenetic similarity. These were performed with NT-SYS com¬ puter programs (Rohlf and Kishpaugh, 1972). Each sample was considered an operational taxonomic unit and means of cranial measurements were used as characters. Matrices of Pearson’s product—moment correlation and phenetic distance coefficients were generated from standardized character values. Cluster analy¬ ses were conducted by means of UPGMA (unweighted pair-group method using arithmetic averages) on both the correlation and distance matrices, and a phenogram was generated for each. Only the distance phenogram is illustrated here because it had a higher cophenetic correlation value. In order to assess the degree of divergence among samples, a multivariate analysis of variance (MANOVA) in SAS was used. Characteristic roots and vectors were then extracted and mean canonical variates computed for each locality. This analysis con¬ siders all characters simultaneously and provides weighted combi¬ nations of characters that maximize among-sample variance and minimize within-sample variance. For a detailed description of the mechanics of this particular analysis, see Schmidly and Hen¬ dricks (1976) and Yates and Schmidly (1977). 10 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Distributional Analysis Soil samples were collected from the walls of burrows in which pocket gophers were trapped and from areas completely void of gophers. Four soil parameters (clay, silt, sand, and moisture) were measured to investigate the influence of soil texture and moisture on the distribution of G. bursarius. Prior to analysis, all samples were air-dried and passed through a 2-mm. screen. The texture of the soil in each sample was ana¬ lyzed by the hydrometer method in order to assess the percentage of clay, silt, and sand present (Kilmer and Alexander, 1949). The field capacity of the soil for moisture was determined with a tech¬ nique described by Milford (1970). The location of all samples was plotted on a county soil map and the soil association corre¬ sponding to each location was determined. The importance of the Brazos River as a boundary between karyotypic races also was investigated. At several locations along that river, pocket gophers were trapped on both banks and karyo¬ typed to determine the distribution of chromosomal races. Oxbow lakes and other major land transfers along the river were located with the aid of aerial photographs and examined for their poten¬ tial as avenues for dispersal of pocket gophers across this water¬ way. Other rivers in eastern Texas also were examined to determine their effects on the distribution of the various chromo¬ somal races. Results Chromosomal Analyses Karyotypic Descriptions More than 183 specimens were karyotyped from the 11-county study area. Of these, 18 had the karyotype of race D; 93, of race E; 30, of race F; 25 of race G; and 17 were of hybrid origin. No polymorphism was found in any of the populations sampled. Representative karyotypes are shown in Figs. 5-7. Karyotypic descriptions follow: Chromosomal race D (Fig. 5A).—Diploid number, 72; fundamental number, 70. This karyotype consists of 35 pairs of large to small acrocentric autosomes, a large acrocentric X-chromosome, and a small acrocentric Y-chromosome. Chromosomal race E (Fig. 5B).—Diploid number, 74; fundamental number, 70. The autosomal complement contains 36 pairs of size-graded acrocentric chromo¬ somes. The X-chromosome is a medium-sized submetacentric and the Y- chromosome is a small acrocentric. Kim (1972) described chromosomal race H from Robertson County. However, specimens examined during this study from HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 11 All A A AA 0 * an AA nn aa a a aa ** * ft nn a n HA a a a a /v a n A m ** ft * Hi* nft #1- A A A A A ft* ft* ft ft «% Aft m ft •% ft: ■- • „ ft A X Y AA All • a aa an c c a* aa an A* ft ft ftft A — an ia aa a m ftft •# ft • a* Aft A -ft* #A ft ftfe a A 1 1 1 i a ft ft*. ft ft ft ft & * B X Y Fig. 5.— Karyotypes of male Geomys bursarius representing: A, chromosomal race D from McLennan County near Waco, Texas; B, chromosomal race E from Falls County near Cedar Springs, Texas. localities given by Kim possessed karyotypes identical to those found for chromo¬ somal race E. Chromosomal race F (Fig. 6A).—Diploid number, 70; fundamental number, 74. In the autosomal complement there are 31 pairs of large to small acrocentrics, one pair of large submetacentrics, one pair of medium submetacentrics, and a pair of small metacentrics. The X-chromosome is a medium submetacentric and the Y- chromosome a small acrocentric. Chromosomal race G (Fig. 6B).—Diploid number, 70; fundamental number, 72. The autosomal chromosomes are comprised of 32 pairs of large to small acrocen¬ trics, one pair of large submetacentrics, and a pair of small metacentrics. The X- chromosome is a medium submetacentric, the Y-chromosome, a small acrocentric. FXG hybrids (Fig. 7A).—Karyotypic differences between these two races center around the number of biarmed chromosomes present in the autosomal comple¬ ment, and F, hybrids can be detected due to this difference. All F, individuals have a diploid number of 70 and a fundamental number of 73. The autosomal chromosomes consist of 30 pairs of large to small acrocentrics, one unpaired acro¬ centric, one pair of large submetacentrics, an unpaired medium submetacentric, 12 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Ml A A a % Aft ftA of| ft* — ft A ft ft ftft ft ft *• lift ftft OA A ftr* ha AA O'* A A — — AH #% A m •*• * * ft « — K* A X X u * m nfc A« Aft lift Id AH *• HA OA A A ft A Aft ft „ A O A E F F G D D E G E E E E E E D F F F F F G G F F G E E G E E G E F F E F F E E E G E G E E Fig. 10.— Distance phenogram resulting from cluster analysis of 51 soil samples. These samples correspond to areas yielding the four chromosomal races and to the two major clay belts separating the chromosomal races. The cophenetic correla¬ tion coefficient is 0.90. The letters D, E, F, and G represent the four chromosomal races. Mi and Bu correspond to the clay belts located in Milam and Burleson counties, respectively. The sample numbers correspond to grouped localities pre¬ sented in Fig. 4. 20 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY maximum moisture content on the distribution of the various chromosomal races. A NT-SYS clustering analysis was used to determine if the dif¬ ferent chromosomal races had distinct soil preferences. Means of the four soil measurements for 51 soil samples (45 of which yielded one of the four chromosomal races and six of which rep¬ resented the two major clay belts separating the chromosomal races) were used in the analysis. A phenogram diagraming the phenetic relationships of all samples was computed by cluster analysis from distance matrices (Fig. 10). The samples correspond to those used in the morphological analysis (Fig. 4) and the karyotypes represent chromosomal races. The distance matrix divides the samples into four major clus¬ ters with a cophenetic correlation coefficient of 0.90. Cluster A separates the two major clay belts located in Milam and Burleson counties, where gophers were absent, from samples where gophers were present. There is considerable overlap of soil samples in clusters B, C, and D indicating little distinction among the chromosomal races with respect to the four soil parameters. In general, the four groups represented by clusters A to D demon¬ strate a clustering of samples that show a gradual decrease in the percentage of clay. Forty-nine of the 51 samples used in the clustering analysis were plotted on a textural triangle to determine in which textural class the various chromosomal races and areas void of gophers would occur. As can be seen in Fig. 11, most of the four chromo¬ somal races occupy sandy loam to sandy clay loam soils. The seven samples in loam and clay loam soils are pocket gophers taken from the Miller-Norwood soil association in Burleson County and are almost exclusively representative of race E gophers from localities near the Brazos River. Open squares at the top of the triangle are samples taken in heavy clay soils from areas void of pocket gophers in Milam and Burleson counties. The Milam County samples are from the belt of Trinity-Catalpa soils separating races E and F, whereas the Burleson County sam¬ ples are from the belt of Miller-Norwood soils that separate races E and G. The four soil measurements also were used to determine soil tolerances of the chromosomal races as well as to indicate the overall influence of each particular parameter on the presence or absence of gophers. As revealed in the clustering analysis and plots on the texture triangle, there is a large degree of overlap HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 21 Fig. 11.—Soil textural triangle representing the percentage of clay, silt, and sand in samples taken from localities yielding pocket gophers as well as areas void of pocket gophers. Stars represent samples of chromosomal race D; triangles, race E; open circles, race F; closed circles, race G; and half circles, hybrids between races F and G. Open squares represent the two clay belts located in Milam and Burleson counties where pocket gophers were absent. Scales located on each side of the tri¬ angle indicate percentage of clay, silt, and sand. The triangle has been subdivided into 12 textural classes based on these percentages. among the four chromosomal races with regard to their soil toler¬ ances. As shown in Table 1, chromosomal race E inhabits a wider range of soil types than do races D, F, and G; however, if samples occurring in the Miller-Norwood soils (where a higher than aver¬ age percentage of clay, silt, and moisture are present) are excluded from this analysis, the tolerance of race E is similar to that of the others. Most pocket gophers were found in soils having a low content of clay (mean, 19; range, 12 to 39%) and a high content of sand (mean, 63; range, 19 to 22%). Conversely, pocket gophers are absent from soils having a high clay content (mean, 65; range, 56 to 72%) and low sand content (mean, 8; range, 1 to 16%). Moisture also seems to be correlated with the local presence or absence of pocket gophers. On the average, soils capable of containing a high moisture content (mean, 24; range, 23 to 33%) excluded 22 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Table 1 —Mean soil tolerances of chromosomal races D, E, F, and G of Geomys bursarius. Minimum and maximum values are given in parentheses. Chromosomal races % Clay % Sand % Silt % Moisture Race D 19.65 68.96 11.39 11.57 Race E (18.32-20.32) (64.96-70.96) (08.72-14.72) (09.56-12.63) 19.77 60.40 19.80 14.13 (with Miller-Norwood samples) (12.88-39.44) (19.84-81.84) (06.72-45.44) (08.75-23.15) Race E 16.55 67.62 17.41 12.64 (without Miller-Norwood samples) (12.16-26.16) (51.84-81.84) (06.00-33.24) (08.75-16.95) Rare E 28.26 44.64 28.52 17.42 (Miller-Norwood samples) (21.44-39.44) (19.84-68.56) (10.00-45.44) (12.48-23.15) Race F 18.78 63.98 17.24 14.20 Race G (12.88-26.88) (48.40-82.40) (04.00-30.00) (12.64-16.61) 20.88 60.62 18.58 15.25 (16.16-30.32) (24.96-78.40) (04.72-44.72) (12.24-18.51) gophers. Per cent of silt did not affect the distribution of gophers in the study area. Influence of Rivers on Distribution In the major floodplain of the Brazos River from southern Falls County to Washington County, chromosomal race E is limited primarily to the east bank and chromosomal races F and G to the west (Fig. 8). Other rivers considered in the course of this study did not appear to effectively limit the distribution of chromo¬ somal races. The Navasota and San Jacinto rivers located east of the Brazos River, for example, had chromosomal race E distrib¬ uted along both the east and west banks; likewise, chromosomal race G occupied both banks of the Colorado River in south- central Texas. Several areas reflected exceptions to the general distribution of chromosomal races along the Brazos. Outside the major flood- plain to the north in Falls and McLennan counties, chromosomal races D and E were found on both the east and west banks. Within the floodplain, chromosomal race E occurred on the west bank of the river in Milam and Burleson counties. Both of these locations represent isolated areas near the river. In an effort to explain the presence of race E in Milam and Burleson counties, we investigated certain actions of the Brazos River. Using aerial photographs and ground surveys, we found 12 oxbow lakes along the river (Fig. 2). Two of them were on the west bank in Falls and Milam counties and in the general vicinity of chromosomal race E in Milam County. In the region of Burle¬ son County, also occupied by race E, two more oxbow lakes were located as well as a major shift of the streambed; Old River HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 23 represents a meander scar or old river bed of the Brazos. We sug¬ gest that populations of pocket gophers belonging to race E and presently occupying the west bank of the Brazos historically occurred on the eastern bank. Shifts in the course of the river, as indicated by the meander scar, effectively isolated portions of the once contiguous distribution of race E and placed them on the western side of the river. Morphological Analyses The following morphological analyses were conducted to examine the degree of correlation between chromosomal and mor¬ phological variation in Geomys bursarius. Chromosomal races served as a population index in grouping specimens for morpho¬ logical consideration. Univariate Analysis Four cranial characters (greatest length of skull, palatal length, length of rostrum, and mastoidal breadth) were examined for cli- nal variation in size among populations of G. bursarius in Texas. Three geographical transects (north to south, east to west, and northeast to southwest) were constructed for each character. All four characters fit a single pattern with only minor deviations; therefore, only greatest length of skull is illustrated by Dice- Leraas diagrams in Fig. 12. Along the north-south transect, greatest length of skull shows a clinal decrease in size from the larger pocket gophers in northern Texas (samples 3, 7, and 17), representing the lutescens group, to the smaller gophers in southern Texas (samples 25, 24, 20, 22, and 21), which are representative of the attwateri group. Two samples from the Texas coast in Aransas County (21 and 22) show a significant increase in size, and the mean values are reminiscent of pocket gophers from farther to the north. Along the east-west transect from Jasper and Newton counties (sample 45) to Winkler County (sample 15), there is a smooth cli¬ nal increase in greatest length of skull. Three size groups can be recognized based on slight deviations or breaks along the dine. These include the small gophers of the breviceps group (samples 45, 44, 43, 32, and 33), the medium-sized gophers of the attwateri group (samples 51 and 53), and the large gophers of the lutescens group (samples 17, 16, 14, and 15). Sample 16 from Mason County also shows a shift in the dine and gophers from there 24 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Fig. 12.—Clinal variation in greatest length of skull as expressed by Dice-Leraas diagrams. A, a north to south cline; E, east to west; and C, northeast to southwest. Measurements in millimeters (mm) are given along the horizontal axis. For each diagram, the horizontal line represents the range; the vertical line, the mean; the unshaded part of the rectangle, one standard deviation; and the shaded part of the rectangle, two standard errors of the mean. The locality number is to the left of the diagram and the sample size is to the right. For a key to samples, see Fig. 4. average slightly smaller than those from samples to the east and west. Clinal trends along the northeast-southwest transect are some¬ what erratic with the exception of adjacent samples in rather localized areas. There is a gradual increase in size from specimens in Louisiana (sample 48) to sample 28 in central Texas. This situation is reversed and size decreases beginning with sample 28 HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 25 Fig. 13 .—Three-dimensional projections of the 44 samples of Geomys bursanus females onto the first three canonical vectors. Circles represent the lutescens chromosome group; squares, the attwateri group; and hexagons, the breviceps group. and continuing to sample 25; gophers in sample 19 from south¬ ern Texas are slightly larger than those represented by sample 25 from Guadalupe and Gonzales counties. Multivariate Analysis A multivariate analysis of variance (MANOVA) with projection onto canonical axes was used to describe the amount of variation among 44 samples by considering all characters simultaneously. Four different criteria (Wilk’s Criterion, Roy’s Maximum Root Criterion, Hotelling-Lawley’s Trace, and Pilla’s Trace) were used to test the hypothesis of no significant morphological difference among samples. All four tests produced F-values that were signif¬ icant at PC0.0001; therefore, significant morphological differences among samples were assumed to exist. The first three characteristic roots were significant at the PCO.OOOOl level and their canonical variates (vectors) accounted 26 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Table 2.—Variable coefficients for canonical variates I, II, and III with an estimate of the per cent influence of each character on each vector for 44 samples of Geomys bursarius. Vector i Vector ii Vector in Character Variable Coefficient Per cent Influence Variable Coefficient Per cent Influence Variable Coefficient Per cent Influence Greatest length of skull 0.0877 22.52 -0.0104 4.05 -0.0364 16.14 Basal length 0.0386 8.99 -0.0247 8.74 0.0250 10.06 Breadth of rostrum 0.0775 4.54 -0.0291 2.59 0.0956 9.64 Zygomatic breadth -0.0206 3.24 -0.0304 7.26 -0.0353 9.61 Interorbital breadth -0.0230 0.83 0.0712 3.90 -0.0862 5.37 Breadth of braincase -0.0553 6.06 0.0428 7.12 0.0309 5.85 Mastoidal breadth 0.0570 8.27 0.0589 12.97 -0.0534 13.38 Length of nasals 0.0564 4.69 -0.0087 1.10 0.0050 0.72 Length of rostrum -0.1453 15.33 -0.0584 9.36 0.0885 16.15 Length of maxillary toothrow -0.0090 0.47 0.0129 1.03 0.1032 9.41 Palatal length -0.1163 17.13 0.1403 31.38 -0.0043 1.09 Palatofrontal depth 0.0876 7.93 -0.0764 10.50 0.0166 2.60 for the greatest percentage (VI, 48.14%; VII, 16.11%; and VIII, 10.61%) of the total variance. A three-dimensional plot of the means of the first three canonical variates is shown in Fig. 13. The per cent contribution of each character to these three canoni¬ cal variates is given in Table 2. Vector I primarily separates samples of the lutescens group (represented by circles) from those of the attivaten and breviceps groups (represented by squares and hexagons, respectively). Char¬ acters having the highest per cent influence on this vector reflect size, especially in skull length. Greatest length of skull, length of rostrum, and palatal length all exert a heavy influence on this vector. In all these characters, gophers of the lutescens group average larger than those from either the attwateri or breviceps groups. No overlap occurs among these three chromosomal com¬ plexes along Vector I; however, samples 21 and 22 of the attwateri group are closely associated with the larger gophers of the lutes- HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 27 cens group. No distinct geographic trends with regard to taxo- nomically recognized subspecies are indicated along this vector. Vector II separates the attwateri group from the breviceps group. Characters with the highest influence on this vector are mastoidal breadth, palatal length, and palatofrontal depth. These measurements are wider and deeper in gophers of the attwateri group than those of the breviceps group. Overlap between these two groups along Vector II is evident in sample 47, which is asso¬ ciated with the attwateri group, although it has a breviceps karyotype. A slight geographic trend is evident among the lutes- cens group along this vector as illustrated by the separation of samples 8, 15 to 18, 35, and 36 from samples 3, 4, 7, 9, 11, 13, and 14. The former assemblage represents the recognized subspecies, G. b. knoxjonesi (samples 8 and 15), G. b. texensis (sample 16), G. b. llanensis (samples 17 and 18), and G. b. major (samples 35 and 36 from central Texas). The latter assemblage of samples represents G. b. major from northwestern Texas. Vector III separates samples within each major chromosomal group. The primary contributing characters are greatest length of skull, length of rostrum, and mastoidal breadth. Most of the sepa¬ ration occurs within the lutescens group with samples 16, 17, 18, 35, and 36 (representing gophers from the eastern edge of the Edwards Plateau) splitting away from other samples of the group. No other geographic trends are indicated along this vector. In order to compensate for some of the disadvantages of ordina¬ tion techniques, a NT-SYS clustering analysis was performed with the 12 canonical variates from the MANOVA for each of the 44 samples of females. A phenogram diagraming the phenetic relationships of the 44 samples computed by clustering from dis¬ tance matrices is presented in Fig. 14. Chromosomal and subspe¬ cific designations for each sample are also included. The distance phenogram is subdivided into three major clusters. Starting from the top, the first cluster includes samples 3, 4, 7 to 9, 11, 13 to 18, and 36, representing pocket gophers of the lutescens group. All of these samples, with the exception of sample 36 from McLennan and Hill counties in central Texas, represent specimens from the western part of the state. Within this first cluster, two subdivi¬ sions are evident. The first (samples 3, 13, 4, 7 to 9, 11, 14, and 15) represents two subspecies, G. b. major (chromosome race D) and G. b. knoxjonesi (chromosome race A), which are known from northwestern Texas. The second subcluster (samples 16 to 18 and 36) includes three subspecies, G. b. texensis (chromosome 28 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY & -S Ma Ma Ma Ma Ma Ma Kx Kx Ma Tx La La Ma At At Bz At Bz Bp Bz Bz At At At Am Bz At Te Bz Bz Bz Bz Bz Bz Lu Pr Pr Pr Bz Du Sa Pr At Ma Fig. 14.— Distance phenogram resulting from cluster analysis of 44 samples of female Geomys bursarius. The cophenetic correlation coefficient is 0.797. L, A, and B represent the lutescens, attwaten, and breviceps chromosome groups. Sub¬ species abbreviations are as follows: Ma, G. b. major ; Kx, G. b. knoxjonesi ; Tx, G. b. texensis; La, G. b. llanensis ; At, G. b. attwaten ; Bz, G. b. brazensis ; Bp, G. b. breviceps-, Am, G. b. ammophilus-, Te, T. b. terricolus ; Lu, G. b. ludemani; Pr, G. b. pratincola\ Du, G. b. dutcheri\ and Sa, G. b. sagittahs. HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 29 race A), G. b. llanensis (chromosome race A), and G. b. major (chromosome race D), which are known from the eastern edge of the Edwards Plateau and central Texas. The second major cluster contains samples 19 to 21, 23 to 34, 38, 41, and 43 to 54; these are pocket gophers from central and east Texas and Louisiana. Two main subclusters can be seen readily. One contains samples 19 to 21, 23 to 29, 47, and 51 to 53. All of these samples, with the exception of 47, belong to the attwateri group and are located in central and southern Texas. Sample 47 (from Mer Rouge, Louisiana) represents the subspe¬ cies, G. b. breviceps, and, although chromosomally identical to populations in eastern Texas with a breviceps group karyotype, it is morphologically similar to populations of larger gophers in central and southern Texas with an attwateri group karyotype. The arrangement of samples in this subcluster does not indicate that morphological differences exist between chromosomal races F and G or among the currently recognized subspecies occurring in this geographic area. Rather, it represents pocket gophers that are intermediate in size between western and eastern samples of G. bursarius, and that exhibit the attwateri group karyotype. The second subcluster is composed of samples 30 to 34, 38, 41, 43 to 46, 48 to 50, and 54 from central and eastern Texas that possess a breviceps group karyotype. A total of six described subspecies (G. b. pratincolus, G. b. terricolus, G. b. dutcheri, G. b. brazensis, G. b. saggitalis, and G. b. ludemani) are represented by these sam¬ ples and it is evident from the clustering that only minor patterns of geographic differentiation exist among them. The third major cluster contains two samples, 22 and 35, from diverse areas in the distribution of the species. Sample 22, of the attwateri chromosome group and the subspecies G. b. attwateri, is from Aransas County in southern Texas. Sample 35 is located in McLennan and Bosque counties in north-central Texas and represents the lutescens chromosome group and the subspecies G. b. major. The univariate analysis, as well as the MANOVA, shows a strong association of sample 22 with the samples of larger pocket gophers from the lutescens group. When considered independent of one another, samples 22 and 35 show similarities to samples possessing similar karyotypes. Sample 22 has an aver¬ age distance coefficient of 0.105 to samples of the attwateri group as compared to 0.137 from samples of the lutescens group. Sam¬ ple 35 has an average distance coefficient of 0.106 to samples of the attwateri group and 0.104 to samples of the lutescens group. 30 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY VECTOR 1 (61.44%) Fig. 15.—Projections of the first two canonical vectors for eight samples of Geomys bursanus females in Robertson, Brazos, Milam, and Burleson counties in east Texas. The numbers are positioned near the mean value for each sample in the character space; the ellipse surrounding each number represents one standard deviation around the mean. To investigate the extent to which patterns of morphological variation shown throughout the state compare to variation in a more localized area, eight samples (50 to 55, 33, and 34) from four counties in the east Texas study area were analyzed using a MAN- OVA (Fig. 15). The first two canonical variates account for 76.83 per cent (VI, 61.45%; and VII, 15.39%) of the total variation. Fig. 15 gives a two-dimensional plot of the means of the first two canonical var¬ iates. The ellipsoid curves represent one standard deviation to either side of the mean for each vector. The per cent infuence of each character is given in Table 3. Characters most influential along Vector I are palatal length, mastoidal breadth, breadth of braincase, and palatofrontal depth. Two main groups of gophers can be distinguished. The first con¬ sists of samples 50, 54, 55, 33, and 34; these represent the brevi- ceps group karyotype and the subspecies G. b. brazensis. The second group, samples 52, 53, and 51, represents the attwateri group karyotype and the subspecies G. b. brazensis. As can be HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 31 Table 3. —Variable coefficients for canonical variates I and II with an estimate of the per cent influence of each character on each vector for eight samples of Geomys bursarius in a four county area. Vector i Vector 11 Character Variable Coefficient Per cent Influence Variable Coefficient Per rent Influence Greatest length of skull 0.2597 18.53 -0.2513 30.73 Basal length -0.4289 27.81 0.0305 3.39 Breadth of rostrum -0.0007 0.01 -0.1982 5.53 Zygomatic breadth -0.1670 7.46 -0.0192 1.47 Interorbital breadth -0.1173 1.21 -0.2207 3.88 Breadth of braincase 0.0330 1.04 0.1768 9.49 Mastoidal breadth 0.1036 4.22 -0.0831 5.80 Length of nasals -0.1559 3.59 -0.1370 5.40 Length of rostrum -0.1838 5.42 0.4768 24.08 Length of maxillary toothrow 0.2200 3.29 0.1069 2.73 Palatal length 0.6087 25.10 0.0239 1.69 Palatofrontal depth -0.0915 2.32 0.1336 5.81 seen by the ellipses, considerable overlap occurs within each chromosomal group but none between the two groups. Although the degree of difference between these two complexes is rather small, samples of the attwateri group average larger than those of the breviceps group in measurements that exert the greatest influ¬ ence on this vector. The morphological variation along this vec¬ tor does not separate the two chromosomal races of the attwateri group karyotype, which are represented by samples 51 (chromo¬ some race F) and 53 (chromosome race G), nor does it indicate the intermediate size of the intergrades between these two races (represented by sample 52). Discussion Chromosomal Variation In light of the chromosomal variation found within Geomys bursarius, two questions concerning the systematics of this species 32 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY in Texas need to be answered. First, does the chromosomal diver¬ gence found indicate that more than one species is represented in the state? Second, how does this variation relate to currently rec¬ ognized subspecies of G. bursarius ? Recent studies of chromosomal variation throughout the range of G. bursarius have repeatedly pondered the matter of popula- tional relationships within this species (Hart, 1971; Kim, 1972; Hart, 1978). Both Hart and Kim concluded that some of the chromosomal divergence seen within G. bursarius could represent karyotypic differences among cryptic species. Studies of other fos- sorial animals also have indicated that karyotypic differences may be indicative of species-level differentiation (Patton and Ding- man, 1968; Patton, 1973; Reig and Kiblisky, 1969; Thaeler, 1968a, 19686, 1974; Wahrman et al, 1969). However, as revealed by recent investigations of contact zones between chromosomally dis¬ tinct populations, not all karyotypic differences warrant species recognition (see Baker et al., 1975, for a review). According to Baker et al. (1975), and Thaeler (1974), the true role of karyotypic variation in speciation can be determined only by examining interactions of chromosomal forms in zones of contact. To date three contact zones between chromosomal forms of G. bursarius have been studied (Hart, 1971; Pembleton and Baker, 1978). Two of these involve chromosomal races considered in this study. Hart (1971) examined a contact zone between chromosomal races D and E in Oklahoma. From this preliminary study, he concluded that some genetic exchange between these chromo¬ somal races was occurring; however, he did indicate that a more detailed analysis was needed. Pembleton and Baker (1978) sur¬ veyed a contact zone between chromosomal races B and C in New Mexico. They found a low level of hybrid production as well as reduced hybrid fertility but refrained from concluding that spe¬ cies level differences existed because of the presence of an appar¬ ently fertile Fi female and the inability to detect backcross individuals with the use of chromosomal differences. No contact zones between chromosomal races occurring in east¬ ern Texas have been examined prior to this study. Hart (1971) suggested that chromosomal differences between races G and E could be related to specific differences and that studies of areas of contact should elucidate this. The two contact zones found in eastern Texas during our work reveal two different types of inter¬ actions. The zone of contact between races F and G in Milam County involves chromosomal races that differ in fundamental HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 33 number but not in diploid number. This chromosomal variation involves the presence or absence of a pair of biarmed chromo¬ somes. Even though the two races maintain distinct distributions, F 1 hybrid formation is high (41 per cent), suggesting extensive interbreeding. Unfortunately, the degree of gene exchange cannot be determined because of an inability to recognize backcross indi¬ viduals. Chromosomal diversity between these two races could be indicative of subspecific differentiation, although an alternative interpretation is possible: the second arms of the two biarmed chromosomes in race F could be heterochromatic, in which case no meiotic penalties would result during interbreeding with race G. Baker and Genoways (1975) considered chromosomal differen¬ ces between races A and B to represent such a mechanism and, therefore, relegated them to the same subspecies (G. b. knox- jonesi). Further studies using electrophoresis and chromosome banding are needed before any conclusions can be drawn concern¬ ing the amount of gene flow between chromosomal races F and G. However, even if gene flow is reduced, two important ques¬ tions still remain to be answered. First, how did the chromosomal differences arise, and second, why do they persist in this particu¬ lar area of Texas? The second contact zone located during this study involves chromosomal races E, F, and G. However, only one F 1 hybrid with an intermediate karyotype between races E and G has been found, and none between races E and F. This indicates consider¬ able reduction in gene exchange between races F and G, which are found west of the Brazos, and race E, occurring primarily east of the Brazos River. This is expected because chromosomal differ¬ ences between these two major groups include differences in both diploid and fundamental numbers. Similar instances, involving this magnitude of chromosomal differentiation, have been reported for other pocket gopher populations (Patton, 1973; Thaeler, 1974). A taxonomic decision on the specific status of these two chromosomal races will be deferred until further studies can be conducted in this contact zone. Davis (1940) recognized nine subspecies throughout central and eastern Texas and western Louisiana; however, chromosomal variation in these areas does not correspond to the ranges of the subspecies defined by Davis. This is especially true for chromo¬ somal races E, F, and G. As indicated in the chromosomal analy¬ sis, race E persists throughout parts of central Texas, eastern Texas, eastern Oklahoma, and western Arkansas and Louisiana 34 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY with no apparent variation. This area spans the ranges of seven described subspecies. Likewise, the ranges of chromosomal races F and G correspond to three subspecies. Distribution of Chromosomal Races The distributional patterns of chromosomal races A, B, C, and D occurring in northwestern Texas have been investigated pre¬ viously by Baker et al. (1973). Hart (1971) and Kim (1972) con¬ ducted preliminary studies on the overall distribution of chromosomal races D, E, F, and G throughout Texas and the neighboring states; however, neither author described the details of this distribution, although they did indicate that all four chromosomal races inhabit eastern Texas. As a result, investiga¬ tions during this study were directed toward determining the dis¬ tributional patterns of chromosomal races D, E, F, and G (in an 11 county study area) in eastern Texas (Fig. 2). Several distributional patterns are evident within and among chromosomal races D, E, F, and G in the study area. Although local populations may conform to an island type distribution, the chromosomal races are, for the most part, continuously distrib¬ uted wherever suitable soils exist. At the edge of their ranges along the Brazos River, the races maintain a linear distribution throughout alluvial soils close to the water. In the case of race D in McLennan County and race E in Falls County, this distribu¬ tion is restricted to areas along both sides of the river and can best be described as a double dendritic distribution, as defined by Kennerly (1964). With few exceptions, the Brazos River represents a boundary that separates race E on the east bank from races F and G on the west bank. No localities have been found where the chromosomal races are sympatric, although the ranges of certain races approach one another at several localities and actually con¬ tact at two places. In all situations where chromosomal races are not in contact, either the Brazos River or indurate soils exist between them. Factors Affecting Distribution of Chromosomal Races Previous studies have shown that the distribution of Geomys bursarius is correlated with sandy soils and that clay soils act as an effective barrier to dispersal (Davis, 1938, 1940; Davis et al., 1938; Russell, 1953; Miller, 1964; Downhower and Hall, 1966). Major rivers also have been discussed as potential barriers to the distribution of this pocket gopher (Davis, 1940; Kennerly, 1954; HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 35 Miller, 1964). Both of these factors were found to influence the distribution of chromosomal races D, E, F, and G in eastern Texas. Overall distribution patterns of the four chromosomal races are influenced directly by certain soil associations (Fig. 9). Chromo¬ somal races were found consistently in soils ranging from loam to sandy loam; soils consisting of clay loam and clay appeared to act as effective barriers. Chromosomal races along the Brazos River were situated in isolated areas of well-drained alluvial soils or soils containing a high sand content. As indicated by Kennerly (1954, 1963), alluvial soils provide avenues for dispersal through areas of otherwise unfavorable soils. This is the case in McLen¬ nan and Falls counties where chromosomal races D and E occur almost exclusively in the Miller-Norwood-Yahola soil association. Dispersal of chromosomal race D into McLennan County proba¬ bly was from counties (Hill and Bosque) to the north by way of these soils. Dispersal of race E from Falls County along alluvial soils also could explain the occurrence of that chromosomal race on the west bank of the Brazos River in Milam County. Support for this idea can be shown by the fact that race E appears in an isolated area representing the southernmost extension of the Miller-Norwood-Yahola soil association in Milam County. Barri¬ ers between chromosomal races were represented by two soil asso¬ ciations. The Trinity-Catalpa association separates chromosomal races E and F in Milam County and the Miller-Norwood separ¬ ates races E and G in Burleson County. From studies of chromosomal races in areas of contact, some authors have indicated differences in soil preference between chromosomal races (Patton, 1973; Pembleton and Baker, 1978). No differences were found among chromosomal races D, E, F, and G. Both contact zones examined in the course of this study failed to reveal any chromosomal race having preference for a particular soil type. As indicated by the clustering analysis and plots on the soil triangle, the distribution of chromosomal races overlapped with respect to the four soil parameters measured (per cent of clay, silt, sand, and moisture). The only differences found involved samples of race E, which occurred in the Miller- Norwood association in Burleson County and parts of Robertson County. Miller (1964) noted that in Colorado G. bursarius was adapted to the narrowest range of soil conditions and was limited in dis¬ tribution by soil tolerances. Likewise, Downhower and Hall 36 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY (1966) indicated a direct influence of the per cent clay and sand on the distribution of G. bursarius in Kansas. The distribution of chromosomal races D, E, F, and G in eastern Texas is governed, we think, by three soil parameters, namely per cent sand, clay, and moisture. In general, chromosomal race E has a wider toler¬ ance to these parameters than the other three; it occupies soils with a higher than average percentage of clay, sand, and mois¬ ture. This wider tolerance is related to the presence of these pocket gophers in the Miller-Norwood association in Burleson County. With the exclusion of these samples, the general distribu¬ tion of all the chromosomal races corresponds to soils having a low clay content, low soil moisture, and high sand content (Table 1 ). The Brazos River is a significant factor affecting the distribu¬ tion of the four chromosomal races in eastern Texas. It not only provides an avenue for dispersal through alluvial soils along its banks for chromosomal races occurring in Falls and McLennan counties, but it also acts as an effective barrier to the distribution of chromosomal races E, F, and G in the floodplain. Two major questions may be posed regarding the importance of this river to the distribution of the plains pocket gopher. First, why is the Brazos River a barrier in the floodplain? Second, how does the Brazos River affect the dispersal of chromosomal races in the floodplain? As indicated by Davis (1940), rivers that cut transversely across the range of pocket gophers and that have a good flow of water at all seasons could act as effective barriers. As seen in Fig. 9, the Brazos River pursues such a course across the range of chromo¬ somal races E, F, and G. The Brazos River is also one of the major waterways in Texas. Average annual streamflow for the years 1940 to 1965 was seven million acre-feet (Texas Water Development Board, 1968). The Brazos has a maximum discharge and minimum flow equal to, if not slightly larger than, the Trin¬ ity, Nueces, and Sabine rivers in eastern Texas, and greatly exceeds that of the Colorado, San Jacinto, and Navasota rivers. Kennerly (1963) indicated that swimming probably represents the major means of dispersal across waterways, and that width of a river dictates, therefore, the amount of dispersal that can occur. Grinnel (1914) and Goldman (1931, 1935) studied the distribution of pocket gophers along the Colorado River in the Southwest and noted that it, in combination with its floodplain, constitutes a barrier to pocket gopher distribution. Both authors concluded HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 37 that transfers of land from one side to the other could account for the distribution of pocket gophers along the lower part of the Colorado. The dispersal of race E into Burleson County is best explained by this type of action. Oxbow lakes (which represent minor land transfers) cannot account for the wide distribution of race E; however, the major land transfer represented by Old River correlates well with this distributional pattern. Two alternative explanations may be considered for the dispersal of race E in Milam County. First, the oxbow lake in the vicinity of race E may have been a source for dispersal of pocket gophers from Robertson County. Second, dispersal could have occurred from the north in Falls County along the Miller-Norwood-Yahola soil association. The above information may explain the appearance of chromo¬ somal race E on the west bank of the Brazos; however, the factor or factors restricting races F and G to the west bank are not as obvious. Two possible explanations are: 1) the occurrence of races F and G into the vicinity of the Brazos may be the result of a rel¬ atively recent dispersal into the area when compared to race E; 2) races F and G may be limited in their distribution by race E. Morphological Variation As stated by Jackson (1971), changes in chromosome morphol¬ ogy do not necessarily indicate similar changes in external pheno¬ typic characters; however, the two may be partially correlated. This seems to be the case with Geomys bursarius in Texas where morphological differences, for the most part, correspond with chromosomal differences. This correlation is especially good when gross chromosomal changes, such as those represented by the lutescens, attwateri, and breviceps chromosomal groups, are considered. Single morphological characters, as shown in the analysis of clinal variation, cannot explain entirely the patterns of geo¬ graphic variation in G. bursarius. They reveal a clinal increase in size from east to west across Texas corresponding to the regions occupied by the three chromosomal groups. However, major mor¬ phological separation was not apparent in the univariate analy¬ sis. Multivariate statistical techniques provided a better means for analyzing the variation seen with G. bursarius. Four generalizations can be made from both the MANOVA and cluster analyses used to interpret morphological variation of G. bursarius in Texas. First, three main groups of pocket gophers 38 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY can be recognized on the basis of morphology and karyotype. These are the large pocket gophers from northwestern Texas with the lutescens group karyotype, the intermediate-sized gophers from central and southern Texas with the attwateri group karyo¬ type, and the small gophers from central and eastern Texas with the breviceps group karyotype. One important aspect to note when viewing this overall trend is that there is more morphologi¬ cal variation among samples with the lutescens group than there is between samples of the attwateri and breviceps complexes. This correlates with a greater amount of chromosomal variation and polymorphism among chromosomal races of the lutescens group. Second, within each chromosomal group, certain chromosomal races are not distinguished by morphological differences. Varia¬ tion with the lutescens group, for example, does not effectively separate chromosomal races A, B, C, and D into morphologically distinct units. Chromosomal races F and G of the attwateri group also are morphologically indistinguishable. Third, peripheral iso¬ lates of certain chromosomal races are morphologically distinc¬ tive. Sample 47 (race E) occurs at the easternmost part of the range of chromosomal race E in an isolated area of Louisiana. Individuals in this sample are morphologically larger in size than those in adjacent samples of race E, and are more like the intermediate-sized gophers of the attwateri group. Sample 22 is the southernmost extension of chromosomal race G. It occupies an isolated area along the coast in Aransas County. In size, it resembles the larger pocket gophers of the lutescens group. Sam¬ ples 16, 17, and 18 occur in isolated areas of the Edwards Plateau. Karyotypically, they are related to chromosomal race A; however, morphologically, they are nearer in size to pocket gophers of race D, as represented by samples 35 and 36 from the eastern edge of the Edwards Plateau. Samples 35 and 36 also are the southern¬ most extension of chromosomal race D in central Texas. These samples are the smallest pocket gophers of the lutescens group. Fourth, a large numer of subspecies recognized prior to our study in central and eastern Texas do not reveal either morphological or karyological differences. These include the breviceps chromo¬ some group representing the subspecies G. b. brazensis, G. b. ter- ricolus, G. b. sagittalis, G. b. ludemani, G. b. pratincolus, G. b. dutcheri. The only exception is sample 47, G. b. breviceps, which is noticeably larger in size. The subspecies G. b. ammophilus, which occurs in central Texas, is allied karyotypically with chromosomal race G and is morphologically similar to other samples represented by this race. HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 39 The morphological variation occurring throughout the state also is evident when a more localized area is considered. This is illustrated by the separation of the attwateri and breviceps groups into morphologically distinct units in the 11-county study area of eastern Texas. Although chromosomal variation does not coincide completely with morphological variation, the two do correspond in many respects. This supports use of the karyotype as another character for grouping individuals in morphological analyses, especially within species demonstrating chromosomal variation throughout their geographic range. The morphological and chromosomal concordance seen within G. bursarius indicates that, at least in central and eastern Texas, misinterpretations have been made in the recognition of subspecies. It also implies that the three chromosomal groups may be different genetic units. However, we have retained these as a single interbreeding unit (under the name G. bursarius ) until further studies from zones of contact can be conducted. Taxonomic Conclusions In recognizing subspecies, we have followed Mayr’s (1969) defi¬ nition that “a subspecies is an aggregate of phenotypically sim¬ ilar populations of a species, inhabiting a geographic subdivision of the range of a species.” We interpret the morphological and karyological data as revealing seven distinct sub units of Geomys bursarius present in Texas and adjacent states (Fig. 16). Table 4 gives some mean values for the major morphological characters distinguishing the seven subspecies. The first subspecies is characterized by a diploid number of 74 and a fundamental number of 70. It represents the largest individ¬ uals of chromosomal race E and occurs in Mer Rouge, Louisiana. To this group the trinomial name G. b. breviceps Baird applies. Reasons for retaining this subspecies have to do with its disjunct distribution and large size. Another subspecies, G. b. sagittalis Merriam, ranges from the Brazos River in central Texas, eastward through extreme eastern Texas and western Louisiana, and north¬ ward into eastern Oklahoma and Arkansas. This subspecies encompasses samples formerly referred to as G. b. brazensis, G. b. sagittalis, G. b. terricolus, G. b. ludemani, G. b. pratincolus, and G. b. dutcheri. Morphological and karyotypic analyses show these six taxa to be indistinguishable from one another. They all represent chromosomal race E and have a diploid number of 74 40 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY and a fundamental number of 70. In size, they are the smallest subspecies of G. bursarius. The trinomen G. b. sagittalis Merriam has priority over the other five names. The third subspecies, G. b. attwateri Merriam, occupies a range from southern Texas near the San Antonio River to central Texas along the west bank of the Brazos River. The distribution of this subspecies coincides with that formerly recognized for G. b. attwateri, G. b. ammophi- lus, and G. b. brazensis; the oldest available name is G. b. attwateri Merriam. G. b. attwateri, as defined here, is represented by chromosomal races F and G and has a diploid number of 70 and a fundamental number of either 70 or 72. It also is character¬ ized as being of intermediate size. The four remaining subspecies, G. b. major, G. b. llanensis, G. b. texensis, and G. b. knoxjonesi are the largest pocket gophers and occupy parts of central and western Texas, western Oklahoma, and eastern New Mexico. From McLennan County in central Texas, G. b. major Davis is characterized by a diploid number of 72 and a fundamental number of 70; however, some populations throughout its range exhibit chromosomal polymorphisms ranging from a diploid number of 70 to 72 with a fundamental number of 70. G. b. llan¬ ensis Bailey and G. b. texensis Merriam have limited distribu¬ tions along the eastern edge of the Edwards Plateau in Llano and Mason counties, respectively. These two karyotypically indistin¬ guishable subspecies show morphological differences but these are less pronounced than those seen among G. b. attwateri, G. b. sagittalis, and G. b. major from eastern Texas. At this time we retain these two subspecies with the hope that future studies will reveal more clearly the systematic relationships between them. The last subspecies, G. b. knoxjonesi Baker and Genoways, occurs in western Texas and eastern New Mexico. Baker and Genoways (1975) characterized this race as being both chromo- somally and morphologically more closely related to G. b. llanen¬ sis and G. b. texensis than to adjacent populations of G. b. major. Our study of both the MANOVA and NT-SYS clustering analyses indicate G. b. knoxjonesi to be morphologically similar to G. b. major and less similar to populations of G. b. llanensis and G. b. texensis. There are two explanations for this discrep¬ ancy. First, Baker and Genoways placed high value on length of tail in distinguishing this subspecies from G. b. major, but we did not include external measurements in our analysis. Second, our sample of G. b. knoxjonesi consisted of only six individuals from Texas. These variations in procedure could account for the HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 41 42 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Fig. 16.—Geographic distribution of subspecies of Geomys bursarius in Texas and adjacent states: 1) G. b. sagittalis ; 2) G. b. breviceps; 3) G. b. attwateri; 4) G. b. llanensis ; 5) G. b. texensis; 6) G. b. knoxjonesi; and 7) G. 5. major. difference in results and we, therefore, retain knoxjonesi as a dis¬ tinct subspecies. Subspecific recognition of sample 22, which represents a peri¬ pheral isolate of G. b. attwateri, does not seem appropriate at this time. Further studies need to be conducted in this area before any taxonomic conclusions can be made. Accounts of Subspecies Geomys bursarius breviceps Baird 1855. Geomys breviceps Baird, Proc. Acad. Nat. Sci. Philadelphia, 7:335. 1951. Geomys bursarius breviceps. Baker and Glass, Proc. Biol. Soc. Washington 64:57, 13 April. Type locality.— Prairie Mer Rouge, Louisiana. Distribution — Known only from the vicinity of Mer Rouge, Morehouse Parish, Louisiana. Comparisons. —Compared with G. b. sagittalis, G. b. breviceps is larger in most cranial measurements (see Table 4): greatest length of skull averaging 39.7 mm. as opposed to 38.2 mm.; ros¬ trum seldom less than 16.3 mm.; palatal length not less than 22.2 mm.; skull wider with mastoidal breadth not less than 21.8 mm. G. b. breviceps differs from G. b. attwateri in having: skull shor- HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 43 ter with greatest length seldom exceeding 40.5 mm.; length of ros¬ trum averaging 16.3 mm. as opposed to 16.9 mm.; width of skull not exceeding 22.9 mm.; shallower skull with palatofrontal depth not exceeding 14.3 mm.; a diploid number of 74 and a fundamen¬ tal number of 70 as opposed to a diploid number of 70 and fun¬ damental numbers of 70 or 72 for G. b. attwateri. Remarks. —G. b. breviceps represents a peripherally-isolated population of G. bursarius. Chromosomally, it is indistinguish¬ able from G. b. sagittalis, and morphologically, it is intermediate in size between G. b. sagittalis from eastern Texas, western Loui¬ siana, and Arkansas, and G. b. attwateri from central and south¬ ern Texas. Most individuals of this subspecies also have a melanistic coat color not found in the other subspecies. Specimens examined (12).— Louisiana: Morehouse Parish : 1 mi. W Mer Rouge, 3 (TCWC); 1 mi. S Mer Rouge, 7 (TCWC); 2.5 mi. S Mer Rouge, 1 (TCWC); 3.5 mi. S Mer Rouge, 1 (TCWC). Geomys bursarius sagittalis Merriam 1895. Geomys breviceps sagittalis Merriam, N. Amer. Fauna, 8:134, 31 January. 1938. Geomys breviceps brazensis Davis, J. Mamm., 19:489, 14 November. 1940. Geomys breviceps dutcheri Davis, Texas Agr. Exp. Sta. Bull., 590:12, 23 October. 1940. Geomys breviceps ludemam Davis, Texas Agr. Exp. Sta. Bull., 590:19, 23 October. 1940. Geomys breviceps pratincolus Davis, Texas Agr. Exp. Sta. Bull., 590:18, 23 October. 1940. Cxeomys breviceps terricolus Davis, Texas Agr. Exp. Sta. Bull., 590:17, 23 October. 1951. Geomys bursarius sagittalis. Baker and Glass, Proc. Biol. Soc. Washington, 64:57, 13 April. Type locality. —Clear Creek, Galveston Bay, Galveston County, Texas. Distribution. —From the Brazos River in central Texas through¬ out eastern and northern Texas, western Louisiana, southwestern Arkansas and eastern Oklahoma. Comparisons. —For a comparison of G. b. sagittalis with G. b. breviceps see account of the latter. G. b. sagittalis differs from G. b. attwateri as follows: smaller size; length of skull shorter, aver¬ aging 38.2 mm. as opposed to 40.6 mm.; width of skull narrower; depth of skull shallower; diploid number of 74 and fundamental number of 70 as opposed to a diploid number of 70 and a funda¬ mental number of either 70 or 72, for G. b. attwateri. From G. b. major, G. b. sagittalis differs as follows: smaller size; skull shorter in length; mastoidal breadth averaging 21.4 mm. as opposed to 44 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY 24.1 mm.; depth of skull not exceeding 15.0 mm.; diploid number of 74 with fundamental number of 70 as opposed to diploid numbers of 70, 71, or 72 and fundamental number of 70 for G. b. major; X-chromosome submetacentric as opposed to acrocentric. Remarks. — G. b. sagittalis is the smallest subspecies of G. bur- sarius in Texas. Generally, size in this species increases from east to west across Texas with the smallest pocket gophers occurring in eastern and central Texas. Davis (1938) recognized six subspe¬ cies within the range here attributed to G. b. sagittalis; however, our analyses revealed neither morphological nor chromosomal differences among any of the previously recognized subspecies. Davis (1938) also depicted the range of G. b. sagittalis as extend¬ ing into central Texas near the Colorado River. Both chromo¬ somal and morphological analyses indicate that the range of G. b. sagittalis does not extend west of the Brazos River except in localized areas in Burleson and Milam counties. Specimens examined (199).— Louisiana: Caddo Parish : 2 mi. SW Shreveport, 6 (TCWC). Webster Parish: 5 mi. E Minden, 2 (TCWC); 4 mi. E Minden, 1 (TCWC). Lincoln Parish: 4 mi. E Cloudrant, 1 (TCWC). Vernon Parish: 4.2 mi. NE Louisiana border, Fm. Rd. 692, 1 (TCWC); 2 mi. E Texas-Louisiana border, Fm. Rd. 692, 1 (TCWC). Beauregard Parish: De Ridder, 1 (TCWC). Texas: New¬ ton Co.: Newton, 5 (TCWC); 12 mi. NE Burksville, 1 (TCWC); 13 mi. NE Kirby- ville, 1 (TCWC); 3 mi. NE Kirbyville, 1 (TCWC). Jasper Co.: 2.2 mi. N Buna, 1 (TTU); 7 mi. SW Buna, 1 (TCWC); 15 mi. N Jasper, 1 (TCWC); Clark Farm Study Site, 1 (TCWC). Sabine Co.: 3 mi. N Bronson, 1 (TCWC); 2 mi. N Bronson, 1 (TCWC); 7 mi. W Hemphill, 1 (TCWC); 8 mi. W Hemphill, 1 (TCWC). San Augustine Co.: 8 mi. S San Augustine, 1 (TCWC). Shelby Co.: 7 mi. S Center, 1 (TCWC); 10 mi. S Center, 1 (TCWC). Panola Co.: 4 mi. NE Carthage, 2 (TCWC). Rusk Co.: 12 mi. S Hendrickson, 1 (TCWC). Nacogdoches Co.: 11 mi. SW Nacog¬ doches, 1 (TCWC); 5 mi. S Nacogdoches, 1 (TCWC). Tyler Co.: 17 mi. S Wood- ville, 3 (TCWC). Hardin Co.: 5 mi. N Kountze, 2 (TCWC); Silsbee, 1 (TTU). Polk Co.: 3 mi. W Livingston, 2 (TCWC). LJpsher Co.: 1 mi. W Gilmer, 1 (TCWC); 7 mi. W Gilmer, 2 (TCWC). Wood Co.: 4 mi. S Winnsboro, 2 (TCWC); Mineola, 1 (TCWC). Jefferson Co.: 7 mi. SW Fannett, 3 (TCWC). Chambers Co.: 4.7 mi. N Anahuac, 1 (TTU). Liberty Co.: 2 mi. E Liberty, 4 (TCWC). Anderson Co.: Pales¬ tine, 1 (TCWC); 1 mi. W Palestine, 1 (TCWC). Trinity Co.: 13.5 mi. S, 3.5 mi. W Grove, 1 (TTU); Trinity, 1 (TTU). San Jacinto Co.: 3 mi. SW Evergreen, 1 (TCWC); 2 mi. W Evergreen, l(TCWC); Shepherd, 4 (TTU); 2 mi. S Shepherd, 2 (TTU). Montgomery Co.: 2 mi. S Conroe, 3 (TCWC); 5 mi. W conroe, 1 (TCWC); 5 mi. S Conroe, 1 (TCWC); 1.6 mi. E Decker Prairie, 3 (TCWC); 2 mi. E Decker Prairie, 1 (TCWC); 7 mi. E Tomball, 1 (TCWC). Harris Co.: 4 mi. N Huffman, 1 (TCWC); 3 mi. N Mason’s Bay, Laporte, 1 (TCWC); 3 mi. NE Webster 1 (TCWC). Galveston Co.: 1 mi. N Texas City, 4 (TCWC); 2 mi. N Texas City, 2 (TCWC); 1.4 mi. S, 2.3 mi. W Hitchcock, 2 (TTU). Walker Co.: 6 mi. S Huntsville, 1 (TCWC); 17 mi. W on the Huntsville-Bedias Rd., 1 (TCWC). Grimes Co.: 2 mi. E Shiro, 1 (TCWC); 5 mi. E Kurten, 1 (TCWC); 2 mi. E Carlos, Hwy. 30, 1 (TCWC); 0.8 mi. E Carlos, 2 (TTU). Leon Co.: 13 mi. E Centerville, 1 (TCWC); 7 HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 45 mi. N Normangee, 2 (TCWC). Brazos Co.: TAMU, College Station, 6 (TCWC); 1 mi. S Hwy. 30-Hwy. 6 intersection. College Station, 2 (TCWC); 6.2 mi. S College Station, 1 (TCWC); 7 mi. S College Station, 1 (TCWC); 2 mi. SE College Station, 1 (TCWC); 4 mi. SE College Station, 1 (TCWC); 3 mi. SW College Station, 1 (TCWC); 6 mi. SW College Station, 3 (TCWC); 7 mi. SW College Station, 2 (TCWC); 3 mi. E Kurten, 1 (TCWC); Bryan, 1 (TCWC); 0.7 mi. E Brazos River Bridge, Hwy. 21 (TCWC); 0.2 mi. E Brazos River Bridge, Hwy. 21, 1 (TCWC). Robertson Co.: 1 mi. W Bremond, 2 (TCWC); Sander’s Turkey Farm, 1 (TCWC); 7 mi. NW Hearne, 1 (TCWC); 4 mi. NE Hearne, 1 (TCWC); 4 mi. W Hearne, 1 (TCWC); 5 mi. W Hearne, 1 (TCWC); 7 mi. N, 1 mi. W Hearne, l(TCWC); 0.5 mi. N Calvert, Hwy. 6, 2 (TCWC); 2.3 mi. N Calvert, Hwy. 6, 2 (TCWC); 2.3 mi. N, 1.6 mi. W Calvert, 1 (TTU); 3.4 mi. N, 2 mi. W Calvert, 7 (TTU); 5 mi. SW Calvert, 1 (TCWC); 7 mi. S Calvert, Fm. Rd. 1644, 1 (TCWC); 8 mi. S Calvert, Fm. Rd. 1644, 1 (TCWC); 3.6 mi. SW Bremond, 1 (TCWC); 0.3 mi. S Eloise, Fm. Rd. 1373, 1 (TCWC). Falls Co.: 2.8 mi. SW Marlin, Fm. Rd. 712, 1 (TCWC); 3.1 mi. SW Marlin, Fm. Rd. 712, 1 (TCWC); 12.3 mi. SE Marlin, Hwy. 6, 1 (TCWC); O. 3 mi. S Cedar Springs, Fm. Rd. 2027, 1 (TCWC); 1.7 mi. S Cedar Springs, Fm. Rd. 2027, 1 (TCWC); 4.4 mi. N Cedar Springs, Fm. Rd. 2027, 1 (TCWC); 1.9 mi. SE Reagen, Hwy. 6, 2 (TCWC); 0.5 mi. E Highbank, Fm. Rd. 413, 1 (TCWC); 1 mi. N Eloise, Fm. Rd. 1373, 1 (TCWC); 2.3 mi. E Wilderville, Fm. Rd. 413, 1 (TCWC); 3.8 mi. E Wilderville, Fm. Rd. 413, 1 (TCWC); 0.8 mi. N Wilderville, Fm. Rd. 2027, 1 (TCWC); 0.6 mi. S Wilderville, Fm. Rd. 2027, 1 (TCWC); 0.4 mi. E Chilton, Hwy. 7, 1 (TCWC); 2.5 mi. NE Chilton, 1 (TCWC). Freeston Co.: Fair- field, 1 (TCWC). Milam Co.: 2 mi. E Maysfield, Hwy. 190, 1 (TCWC); 6.8 mi. E Maysfield, 2 (TCWC); 1.2 mi. SE Branchville, Hwy. 190, 1 (TCWC); 2.1 mi. SE Branchville near Brazos River, 1 (TCWC); 3.8 mi. SE Branchville, Hwy. 190, 1 (TCWC); 2.4 mi. S Wilderville, Hwy. 413, 1 (TCWC). Burleson Co.: 2.4 mi. NE Clay, 1 (TCWC); 3.0 mi. NE Clay, 3 (TCWC); 3.2 mi. NE Clay, 1 (TCWC); 6.1 mi. N Clay near Brazos River, 1 (TCWC); 17 mi. E Caldwell on W bank of Brazos River near Koppes Bridge, 2 (TCWC); 0.1 mi. SE Grant, Hwy. 50, 1 (TCWC); 1.4 mi. SE Grant, Hwy. 50, 1 (TCWC); 3.1 mi. SE Grant, near Brazos River, 1 (TCWC); 3.4 mi. SE Grant, near Brazos River, 1 (TCWC); 3.8 mi. SE Grant, Hwy. 50, 1 (TCWC); 0.6 mi. NW Grant, Hwy. 50, 1 (TCWC); 0.7 mi. NW Grant, Hwy. 50, 1 (TCWC); 0.8 mi. NW Grant, Hwy. 50, 1 (TCWC); 2.1 mi. NW Grant, Hwy. 50, 1 (TCWC); 2.3 mi. NW Grant, Hwy. 50, 1 (TCWC); 2.5 mi. NW Grant, Hwy. 50, 1 (TCWC); 2.8 mi. NW Grant, Hwy. 50, 1 (TCWC); 3.6 mi. NW Grant, Hwy. 50, 6 (TCWC); 3.8 mi. NW Grant, Hwy. 50, 1 (TCWC). Geomys bursarius attwateri Merriam 1895. Geomys breviceps attwateri Merriam, N. Amer. Fauna, 8:135, 31 January. 1940. Geomys breviceps ammophilus Davis, Texas Agr. Exp. Sta. Bull., 590:16, 23 October. 1951. Geomys bursarius attwateri, Baker and Glass, Proc. Biol. Soc. Washington, 64:57, 13 April. Type locality. —Rockport, Texas. Distribution .—From the Brazos River in central Texas to southern Texas near the San Antonio River and along the coast as far south as Rockport, Aransas County. 46 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Comparisons. —For comparisons with G. b. breviceps and G. b. sagittalis see accounts of same. G. b. attwateri differs from G. b. major as follows: smaller size; skull shorter with greatest length of skull averaging 40.6 mm. as opposed to 42.3 mm.; mastoidal breadth narrower; depth of skull shallower. Remarks. —In size G. b. attwateri is intermediate between G. b. sagittalis to the east and G. b. major to the north and west. This subspecies represents two chromosomal races (see text for infor¬ mation pertaining to them). Both races contact in central Texas near the Brazos River, and a zone of intergradation is formed; however, the degree of chromosomal difference separating the two is not thought to be representative of distinct subspecies. Although there is an overlap in diploid and fundamental numbers between G. b. attwateri and G. b. major, the two are dis¬ tinct in the morphology of the X-chromosome and the morphol¬ ogy of chromosomes in their autosomal complements. G. b. attwateri contacts G. b. sagittalis in Burleson County. To date only one chromosomally intermediate individual has been found in this area of contact. Davis (1938) indicated that G. b. attwateri was not present north of the Colorado River. Morphological and chromosomal data indicate that the range of this subspecies extends to the west bank of the Brazos River. Specimens examined (223).— Texas: Milam Co.: Cause, 1 (TCWC); 6 mi. SW Hearne, Hwy. 79, 2 (TCWC); 1.7 mi. E Gause, Hwy. 79, 1 (TCWC); 3 mi. E Gause, 1 (TCWC); 3.8 mi. E Gause, 1 (TCWC); 1 mi. W Gause, 1 (TCWC); 4.4 mi. W Gause, 1 (TCWC); 6.3 mi. W Gause, 1 (TCWC); 2.5 mi. S Gause, 1 (TCWC); 2.6 mi. S Gause, 1 (TCWC); 3.2 mi. S Gause, 1 (TCWC); 4.1 mi. S Gause, 1 (TCWC); 4.3 mi. S Gause, 1 (TCWC); 5 mi. S Gause, 1 (TCWC); 5.6 mi. S Gause, 1 (TCWC); 2.4 mi. NE Gause, 1 (TCWC); 2.6 mi. NE Gause, 1 (TCWC); 3 mi. NE Gause, 6 (TCWC); 2.2 mi. SE Gause on Two Mile Rd. 2 (TCWC); 2.7 mi. SE Gause, 1 (TCWC); 5.7 mi. SE Gause, 2 (TCWC); 6.3 mi. SE Gause, 1 (TCWC); 7 mi. SE Gause, 1 (TCWC); 8.7 mi. SE Gause, 1 (TCWC); 1.3 mi. N, 3 mi. E Milano, 2 (TTU); 1.1 mi. N, 2.5 mi. E Milano, 6 (TTU); 0.8 mi. N, 1.3 mi. E Milano, 3 (TTU); 1.3 mi. S, 3.3 mi. W Milano, 1 (TTU); 2.8 mi. S Milano, 1 (TCWC); 9.1 mi. SE Cameron, Hwy. 36, 1 (TCWC); 4.1 mi. SE Cameron, 1 (TCWC); 7.5 mi. S Rockdale, 1 (TCWC); 4.2 mi. SW Rockdale, Fm. Rd. 2116, 1 (TCWC); 1 mi. S Rockdale, 1 (TCWC); Burleson Co.: 1.3 mi. NE Clay, 1 (TCWC); 2.6 mi. W Tunis, Fm. Rd. 166, 1 (TCWC); 2.4 mi. NE Clay, 2 (TCWC); 0.2 mi. E Snook, 1 (TCWC); 13 mi. NE Caldwell, Hwy. 50, 2 (TCWC); 3 mi. N Caldwell, Hwy. 36, 2 (TCWC); 3.3 mi. N Caldwell, Hwy. 36, 1 (TCWC); 7.4 mi. N Caldwell, Hwy. 36, 1 (TCWC); 12 mi. E Caldwell, Rt. 3, 5 (TCWC); 3.6 mi. NW Gram, Hwy. 50, 5 (TCWC); 4 mi. NW Grant, Hwy. 50, 8 (TCWC); 4.1 mi. NW Grant, Hwy. 50, 1 (TCWC). Washington Co.: 13 mi. W Brenham, 1 (TCWC); 1 mi. W Brenham, 2 (TCWC); Brenham, 1 (TTU). Lee Co.: 2.7 mi. S, 1.6 mi. E Lexington, 1 (TCWC); 3.5 mi. S, 1.6 mi. E Lexington, 3 (TCWC). Austin Co.: 2 HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 47 mi. NE Bellville, 3 (TCWC). Colorado Co.: 6.2 mi. S Eagle Lake, 14 (TTU); 6.2 mi. S Eagle Lake, 1 (TTU). Bastrop Co.: 3 mi. N, 3.7 mi. W Bastrop, 3 (TTU); 3 mi. N, 3.7 mi. W Bastrop, 3 (TTU); 0.8 mi. N, 2.5 mi. E Bastrop, 1 (TTU); 5 mi. E Bastrop, 1 (TCWC); 1.7 mi. W Bastrop, 2 (TTU); Bastrop, 4 (TTU); 1.5 mi. E Elgin, 1 (TTU); 1.9 mi. W Elgin, 4 (TTU); 1.5 mi. W Elgin, 1 (TTU); 0.7 mi. N, 2.6 mi. W Butler, 2 (TTU); 0.5 mi. N, 2.1 mi. W Butler, 1 (TTU); 1 mi. S Butler, 2 (TTU). Travis Co.: 0.7 mi. N, 5.7 mi. E Manor, 2 (TTU). Fayette Co.: E LaGrange, 1 (TCWC); 6 mi. W LaGrange, 2 (TCWC). Caldwell Co.: Luling, 10 (TTU). Fort Bend Co.: 1 mi. S Beasley, 2 (TCWC). Lavaca Co.: 9.1 mi. S, 9.8 mi. E Yoakum, 2 (TTU); 1.1 mi. W Halletsville, 1 (TTU). DeWitt Co.: Hochheim, 5 (TTU); Cuero, 1 (TTU). Guadalupe Co.: 12.1 mi. S, 0.7 mi. E Seguin, 12 (TTU); 12 mi. S Seguin, 2 (TCWC). Gonzales Co.: 1 mi. W Nixon, 2 (TCWC); 2.2 mi. N Nixon, 3 (TTU). Victorio Co.: 6 mi. S Victorio, 1 (TCWC); 3 mi. SW Victorio, 4 (TCWC). Goliad Co.: 8 mi. W Goliad, 3 (TCWC); 3 mi. E Goliad, 2 (TCWC); 3.5 mi. N Goliad, 2 (TCWC); 8 mi. SW Goliad, 2 (TCWC). Wilson Co.: 1 mi. W Flo- resville, 2 (TCWC). Atascosa Co.: 2 mi. NW Campbellton, 3 (TCWC); 2 mi. N Pleasanton, 2 (TCWC). Frio Co.: 1 mi. N Moore, 3 (TCWC). Aransas Co.: Rock- port, 1 (TCWC); 8 mi. SW Rockport, 5 (TCWC); 10 mi. SE Austwell, 8 (TCWC); C.C.C. Camp, Aransas Refuge, 4 (TCWC). San Patricio Co.: between Aransas Pass and Ingleside, Hwy. 361, 5 (TTU). Geomys bursarius major Davis 1940. Geomys lutescens major Davis, Texas Agric. Exp. Sta. Bull., 590:32, 23 October. 1947. Geomys bursarius major, Villa and Hall, Univ. Kansas Publ., Mus. Nat. Hist., 1:229, 29 November. Type locality.—8 mi. W Clarendon, Donley County, Texas. Distribution. —Central Texas (McLennan County) north to western Oklahoma, west throughout northwestern Texas and into parts of eastern New Mexico. Comparisons.— For a comparison of G. b. major with G. b. sagittalis and G. b. attwateri see accounts of same. G. b. major differs from G. b. llanensis as follows: size larger, greatest length of skull averaging 42.3 mm. as opposed to 40.9 mm.; palatal length longer; skull wider. G. b. major differs from G. b. texensis and G. b. knoxjonesi in being larger for almost all cranial measurements. Remarks.— G. b. major represents the largest subspecies of G. bursarius in Texas and adjacent states. Chromosomally, certain populations throughout its range in western Texas demonstrate varying degrees of chromosomal polymorphism making it diffi¬ cult to compare distinct differences among G. b. major, G. b. tex¬ ensis, G. b. llanensis, and G. b. knoxjonesi (see Baker et al., 1973, for a discussion of this variation). Specimens examined (127).— Texas: McLennan Co.: E Waco, 4 (TCWC); 0.2 mi. S Waco, Fm. Rd. 434, 1 (TCWC); S Waco, 1 (TCWC); 1 mi. S Waco, 18 (TTU); 6 48 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY mi. S Waco, 5 (TCWC); 1 mi. SE Waco, Fm. Rd. 434, 1 (TCWC); 1.3 mi. E Bos- queville, 1 (TCWC); 4.5 mi. NE Bosqueville, 1 (TCWC); 3.3 mi. SE Bosqueville, 1 (TCWC); 2.5 mi. SE Downville, 1 (TCWC); 3 mi. NE Downville, 1 (TCWC); 0.4 mi. N Gholson, 1 (TCWC). Hill Co.: 21 mi. NW Waco, Fm. Rd. 933, 2 (TCWC); 5.8 mi. SW Aguilla, Willis Camp, 2 (TCWC). Bosque Co.: 1 mi. SE Smith’s Bend, Fm. Rd. 2114, 1 (TCWC). Cooke Co.: 5.5 mi. SE Gainsville, 1 (TTU). Montague Co.: 3.1 mi. E Jet. Texas 59-Fm. Rd. 1758, 2 (TTU). Hardeman Co.: % mi. W Chilli Exp. Sta. No. 12, Chillicothe, 1 (TCWC). Dickens Co.: 10 mi. E Dickens, 3 (TTU). Knox Co.: 5 mi. SE Benjamin, 1 (TCWC). Collingsworth Co.: 0.1 mi. W Wellington, 13 (TTU); 2 mi. E, 1.5 mi. N Wellington, 1 (TTU); 0.5 mi. N county courthouse in Wellington, 4 (TTU); 2 mi. N, 9 mi. W Wellington, 7 (TTU); 0.2 mi. W Wellington, 3 (TTU); 3 mi. E Wellington, Hwy. 83, 1 (TTU); Wellington, 1 (TTU). Hemphill Co.: 18 mi. E Canadian, 1 (TTU); 5 mi. NE Canadian, ! (TTU). Donley Co.: 1 mi. NW Clarendon, 1 (TTU); 11 mi. W Clarendon, 1 (TTU). Garza Co.: 4.5 mi. NW Post, Hwy. 84, 1 (TTU); 4 mi. E Justiceburg, 2 (TTU). Lubbock Co.: Lubbock, 1 (TTU); Jet. loop 289-Hwy. 84, Lubbock, 2 (TTU); 6.5 mi. S, 0.4 mi. E Lubbock, 1 (TTU); 5 mi. S, 4 mi. E Lubbock, 1 (TTU); 1 mi. E intersection of Hwy. 84-SE Drive, Lubbock, 1 (TTU); 6 mi. SE Lubbock, 2 (TTU); 4 mi. SE Lubbock, 6 (TTU); Slaton, 1 (TTU); 4 mi. N Slaton, 1 (TTU); 11 mi. S Idalou, 1 (TTU). Mitchell Co.: 3 mi. N, 0.3 mi. E Colorado City, 9 (TTU). Howard Co.: 2 mi. NE Big Spring, 2 (TTU); 2 mi. N Big Spring, 1 (TTU). Midland Co.: 3 mi. E Jet. 349 on Hwy. 80, Midland, 1 (TTU); 4 mi. E Jet. 349 on Hwy. 80, Midland, 1 (TTU); 3 mi. N Midland, 1 (TTU); 5 mi. S Stan¬ ton, 2 (TTU). Lamb Co.: 2 mi. N Fieldton, 1 (TTU); 2 mi. NW Sudan, 1 (TTU). Bailey Co.: 2 mi. SE Muleshoe, 2 (TTU); 22 mi. S Muleshoe, Wildlife Refuge, 2 (TTU). Dallam Co.: 2.5 mi. E Dalhart, 1 (TTU). Geomys bursarius llanensis Bailey 1905. Geomys breviceps llanensis V. Bailey, N. Amer. Fauna, 2:4, 1 February. 1947. Geomys bursarius llanensis, Villa and Hall, Univ. Kansas Publ., Mus. Nat. Hist., 1:234, 29 November. Type locality. —Llano, Llano Co., Texas. Distribution.— Eastern edge of the Edwards Plateau of Texas in Llano and Gillespie Counties. Comparisons. —For a comparison of G. b. llanensis to G. b. major see account of the latter. Individuals of G. b. llanensis average larger than those of G. b. texensis and G. b. knoxjonesi for most cranial measurements (see Table 4). These differences are as follow: greatest length of skull longer, averaging 40.9 mm.; length of rostrum not less than 17.1 mm.; mastoidal breadth wider, averaging 22.1 mm.; deeper in skull depth. Remarks.—G. b. llanensis, G. b. texensis, and G. b. knoxjonesi in Texas are chromosomally indistinguishable with diploid numbers of 70 and a fundamental number of 68 (Kim, 1972; Baker and Genoways, 1975). Hart (1971) described a diploid number of 71 and a fundamental number of 69 for G. b. texensis HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 49 and G. b. llanensis, which may indicate chromosomal polymor¬ phisms within certain populations. The morphological analyses used in this study show certain morphological distinctions; how¬ ever, they are less than those shown by other subspecies. If mean cranial measurements are considered, G. b. llanensis and G. b. texensis can be separated, based on size (Table 4). It is possible that these two subspecies are not taxonomically distinct, but the small size of samples prevents us from making a formal taxo¬ nomic change at this time. Specimens examined (31).— Texas: Llano Co.: Castell, 1 (TTU); 2.6 mi. N, 1.8 mi. E Castell, 5 (TTU); Oatman Creek, 3 mi. S Llano, 3 (TCWC); Drier Creek, 7 mi. E Llano, 9 (TCWC); 51.6 mi. W Austin, Hwy. 71, 1 (TTU); 0.2 mi. N, 8.7 mi. W Llano, 3 (TTU); 8 mi. S, 0.9 mi. W Kingsland, 4 (TTU); 9.2 mi. S, 1.1 mi. E Kingsland, 1 (TTU); 10 mi. S, 1.8 mi. E Kingsland, 2 (TTU). Gillispie Co.: 1 mi. N Fredricksburg, 2 (TTU). Geomys bursarins texensis Merriam 1895. Geomys texensis Merriam, N. Amer. Fauna, 8:137, 31 January. 1950. Geomys bursarius texensis. Baker, J. Mamm., 31:349, 21 August. Type locality.— Mason, Mason County, Texas. Distribution. —Mason County. Comparisons. —For a comparison of G. b. texensis to G. b. major and G. b. llanensis see accounts of same. G. b. texensis compared to G. b. knoxjonesi differs as follows: palatal length averaging 22.1 mm. as opposed to 21.4 mm.; mastoidal breadth narrower, averaging 22.1 mm. as opposed to 22.8 mm.; skull shal¬ lower with palatofrontal averaging 13.6 mm. as compared to 14.9 mm. Remarks. —Several mean cranial measurements of G. b. texensis and G. b. knoxjonesi are similar (Table 4). These similarities are best depicted by measurements of greatest length of skull and length of rostrum. Specimens examined (22).— Texas: Mason Co.: 1 mi. E Mason, 4 (TTU); 6.5 mi. E Mason, Hwy. 29, 1 (TTU); 9.4 mi. W Mason, U.S. 377, 1 (TTU); 0.3 mi. S, 1.5 mi. W Castell, 3 (TTU); 0.3 mi. S, 0.9 mi. W Castell, 1 (TTU); 0.3 mi. S, 0.8 mi. W Castell, 1 (TTU); 0.7 mi. S, 2.1 mi. W Castell, 3 (TTU); 1.0 mi. S, 2.3 mi. W Castell, 1 (TTU); 2.0 mi. S, 2.7 mi. W Castell, 1 (TTU); 2.6 mi. S, 3 mi. W Cas¬ tell, 1 (TTU); 3.6 mi. N, 1.5 mi. W Mason, 1 (TTU); 1 mi. N, 1.1 mi. W Mason, 4 (TTU). Geomys bursarius knoxjonesi Baker and Genoways 1975. Geomys bursarius knoxjonesi Baker and Genoways, Occas. Papers Mus., Texas Tech Univ., 29:1, 25 April. 50 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Type locality.— 4.1 mi. N, 5.1 mi. E Kermit, Winkler County, Texas. Distribution. —Southern Cochran, Yoakum, Terry, Gaines, northwestern Martin, Andrews, Winkler, and Ward counties in western Texas and Chavez, Eddy, and Lea counties in southeast¬ ern New Mexico. Comparisons.— For a comparison of G. b. knoxjonesi to G. b. major, G. b. llanensis, and G. b. texensis see accounts of same. Remarks. —Baker and Genoways (1975) described G. b. knox¬ jonesi as consisting of two chromosomal races with populations in Texas differing in fundamental number from those in New Mexico. These authors also indicated that G. b. knoxjonesi is morphologically more similar to G. b. texensis and G. b. llanen¬ sis than to G. b. major. Although multivariate analyses used in this study failed to substantiate this claim, if one examines only mean cranial measurements (as shown in Table 4) these indicate that it averages smaller than G. b. major for all measurements and is indeed more similar to G. b. texensis and G. b. llanensis. This discrepancy can be explained by the fact that the calculated values for pooled populations of G. b. major would tend to com¬ pensate for any small-sized populations, thus emphasizing its overall larger size. Specimens examined (34).— Texas: Cochran Co.: 1 mi. W Lehman, 5 mi. W Morton, 1 (TTU); I mi. N, 0.4 mi. W Whiteface, 1 (TTU); 1 mi. N, 0.9 mi. W Whiteface, 2 (TTU); 1 mi. W Morton, 1 (TTU). Gaines Co.: 0.8 mi. S, 15 mi. E Seminole, 11 (TTU); 5 mi. SW Seagraves, Hwy. 385, 1 (TTU); 3 mi. N Seminole, 3 (TTU); 0.6 mi. S, 14 mi. E Seminole, 1 (TTU). Andrews Co.: 2.5 mi. E Andrews, 1 (TTU). Winkler Co.: 0.3 mi. N, 2.5 mi. E Kermit, 6 (TTU); 4.1 mi. N, 5.1 mi. E Kermit, 4 (TTU); 8.5 mi. S, 3.3 mi. E Kermit, 1 (TTU). Acknowledgments This study was conducted by R. L. Honeycutt in partial fulfil¬ lment of the requirements for the Master of Science degree in Wildlife and Fisheries Sciences at Texas A&M University. Thanks are extended to Dr. Gilbert L. Schroeter and Dr. Keith A. Arnold for their editorial assistance and evaluation of this manuscript, and to Dr. William B. Davis for the use of his professional library and his extensive knowledge of the systematics of pocket gophers. The College of Agriculture at Texas A&M University provided computer time for the statistical analyses. We are grateful to Dr. Hugh H. Genoways, previously of The Museum, Texas Tech University, for allowing us to examine specimens. Dr. Robert J. Baker, Texas Tech University, provided HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 51 us with our first pocket gopher trap and allowed us the use of facilities in his laboratory for completion of this manuscript. Dr. Fred S. Hendricks of Texas AScM University gave invaluable advice concerning the statistical treatments of the data, and Terry Maxwell provided the skull drawing. Dr. M. H. Milford, of the Soil Science Department at Texas A&M University, loaned us the use of his laboratory to analyze soil samples. Thanks are due also to Dr. A. Dean Stock, previously of the Cell Biology Section, The University of Texas at Houston, M. D. Anderson Hospital and Tumor Institute, for advice on karyotyping techniques. Two graduate students, Neil Gayden and Duke Rogers, assisted in col¬ lecting and karyotyping specimens and Mr. Marvin Porter, Mr. Irvin Worthington, and Mr. Henry Moehlman generously granted permission to collect specimens on their property. A special thanks is given also to Dierdre Honeycutt for typing this manuscript. Literature Cited Bailey, V. 1905. Biological survey of Texas. U.S. Dept. Agric., Bur. Biol. Surv., N. Amer. Fauna, no. 25:1-222. Baird, S. F. 1854. Characteristics of some new species of North American Mam¬ malia. Proc. Acad. Nat. Sci. Philadelphia, 7:333-337. Baker, R. H. 1950. The taxonomic status of Geomys breviceps texensis Merriam and Geomys bursarius llanensis Bailey. J. Mamm., 31:348-349. Baker, R. H., and B. P. Glass. 1951. The taxonomic status of the pocket gophers, Geomys bursarius and Geomys breviceps. Proc. Biol. Soc. Washington., 64:55-58. Baker, R. J., and S. L. Williams. 1972. A live trap for pocket gophers. J. Wildlife Mgt., 36:1320-1322. Baker, R. J., S. L. Williams, and J. C. Patton. 1973. Chromosomal variation in the plains pocket gopher, Geomys bursarius major. J. Mamm., 54:765-769. Baker, R. J., W. J. Bleier, and W. R. Atchley. 1975. A contact zone between karyotypically characterized taxa of Uroderma bilobatum (Mammalia: Chiroptera). Syst. Zool., 24:133-142. Baker, R. J., and H. H. Genoways. 1975. A new subspecies of Geomys bursarius (Mammalia: Geomyidae) from Texas and New Mexico. Occas. Papers Mus., Texas Tech Univ., 29:1-18. Berry, D. L., and R. J. Baker. 1971. Apparent convergence of karyotypes in two species of pocket gophers of the genus Thomomys (Mammalia, Rodentia). Cytogenetics, 10:1-9. -. 1972. Chromosomes of pocket gophers of the genus Pappogeomys, subgenus Cratogeomys. J. Mamm., 53:303-309. Cleland, H. F. 1929. Geology physical and historical. American Book Company, New York, 718 pp. 52 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Davis, B. L., and R. J. Baker. 1974. Morphometries, evolution, and cyto- taxonomy of mainland bats of the genus Macrotus (Chiroptera: Phyllostomatidae). Syst. Zool., 23:26-39. Davis, W. B. 1938. Critical notes on pocket gophers from Texas. J. Mamm., 19:488-490. -. 1940. Distribution and variation of pocket gophers (genus Geomys) in the southwestern United States. Texas A&M College, Texas Agr. Exp. Sta. Bull., 590:1-38. Davis, W. B., R. R. Ramsey, and J. M. Arendale, Jr. 1938. The distribution of pocket gophers (Geomys breviceps) in relation to soils. J. Mamm., 19:412-418. Downhower, J. F., and E. R. Hall. 1966. The pocket gopher in Kansas. Misc. Publ., Univ. Kans. Mus. Nat. Hist., 44:1-32. Goldman, E. A. 1931. New pocket gophers from Arizona and Utah. J. Washington Acad. Sci., 21:416-426. -. 1935. Pocket gophers of the Thomomys bottae group in the United States. Proc. Biol. Soc. Washington, 48:153-158. Grinnel, J. 1914. An account of the mammals and birds of the lower Colorado Valley, with especial reference to the distributional problems presented. Univ. California Publ. Zool., 12:51-294. Hall, E. R., and K. R. Kelson. 1959. The mammals of North America. Ronald Press, New York, l:xxx+546 pp.+79. Hart, E. B. 1971. Karyology and evolution of the plains pocket gopher, Geomys bursarius. Unpublished Ph.D. dissertation, Univ. Oklahoma, Norman, 111 pp. -. 1978. Karyology and evolution of the plains pocket gopher, Geomys bursarius. Occas. Papers, Univ. Kansas Mus. Nat. Hist., 71:1-20. Hendricksen, R. L. 1972. Variation in the plains pocket gopher (Geomys bursarius) along a transect across Kansas and eastern Colorado. Kansas Acad. Sci., 75:322-368. Jackson, R. C. 1971. The karyotype in systematics. Ann. Rev. Ecol. Syst., 2:327-368. Kennerly, T. E., Jr. 1954. Local differentiation in the pocket gopher (Geomys personatus ) in southern Texas. Texas J. Sci., 6:297-329. -. 1958. Comparisons of morphology and life history of two species of pocket gophers. Texas J. Sci., 10:137-146. -. 1959. Contact between the ranges of two allopatric species of pocket gophers. Evolution, 13:247-263. -. 1963. Gene flow pattern and swimming ability of the pocket gopher. Southwestern Nat., 8:85-88. -. 1964. Microenvironmental conditions of the pocket gopher burrow. Texas J. Sci., 16:395-441. Kilmer, V. J., and L. T. Alexander. 1949. Methods of making mechanical analysis of soils. Soil Sci., 68:15-24. Kim, Y. J. 1972. Studies of biochemical genetics and karyotypes in pocket gophers (family Geomyidae). Unpublished Ph.D. dissertation, Univ. Texas, Austin, 112 pp. Lee, M. R. 1969. A widely applicable technique for direct processing of bone marrow for chromosomes of vertebrates. Stain Technol., 44:155-158. Mayr, E. 1969. Principles of systematic zoology. McGraw-Hill, New York, 428 pp. HONEYCUTT AND SCHMIDLY—PLAINS POCKET GOPHER 53 Merriam, C. H. 1890. Description of a new pocket gopher of the genus Geomys, from western Nebraska. N. Amer. Fauna, 4:51. -. 1895. Monographic revision of the pocket gophers family Geomyidae (exclusive of the species of Thomomys). N. Amer. Fauna, 8:1-258. Milford, M. H. 1970. Introduction to soils and soil science (Laboratory Exercises). Kendall and Hunt Pub. Co., Dubuque, Iowa, 111 pp. Miller, R. S. 1964. Ecology and distribution of pocket gophers (Geomyidae) in Colorado. Ecology, 45:256-272. Patton, J. L. 1967. Chromosomal studies of certain pocket mice, genus Perognathus (Rodentia: Heteromyidae). J. Mamm., 48:27-37. -. 1972. Patterns of geographic variation in karyotype in the pocket gopher, Thomomys bottae (Eydoux and Gervais). Evolution, 26:574-586. -. 1973. An analysis of natural hybridization between the pocket gophers, Thomomys bottae and Thomomys umbrinus in Arizona. J. Mamm., 54:561-584. Patton, J. L., and R. E. Dingman. 1968. Chromosome studies of pocket gophers, genus Thomomys. I. The specific status of Thomomys umbrinus (Richardson) in Arizona. J. Mamm., 49:1-13. -. 1970. Chromosomal studies of pocket gophers genus Thomomys. II. Variation in T. bottae in the American southwest. Cytogenetics, 9:139-151. Pembleton, E. F., and R. J. Baker. 1978. Studies on a contact zone between chromosomally characterized populations of Geomys bursarius. J. Mamm., 59:233-242. Raicu, P., S. Bratosin, and M. Hamar. 1968. Study of the karyotype of Spalax leucodon Nordmann and S. microphthalmus Guld. Caryologia, 21:127-135. Reig, O. A., and P. Kiblisky. 1969. Chromosome multiformity in the genus Ctenomys (Rodentia, Octodontidae). Chromosoma, 28:211-244. Rohlf, F. J., and J. Kishpaugh. 1972. Numerical taxonomy system of multi¬ variate statistical programs. The State University of New York at Stony Brook, Stony Brook, New York. Russell, R. J. 1953. Mammals from Cooke County, Texas. Texas J. Sci., 4:454-464. Schmidly, D. J., and F. S. Hendricks. 1976. Systematics of the southern races of Ords kangaroo rat, Dipodomys ordii. Bull. S. California Acad. Sci., 75:225-237. Service, J. 1972. A users guide to the statistical analysis system. Institute of Statistics, Raleigh Division, North Carolina State University, 260 pp. Soil Conservation Service and Texas Agricultural Experiment Station. 1960. Generalized soil map of Burleson County, Texas. -. 1961. Generalized soil map of Brazos County, Texas. -. 1961. Generalized soil map of Falls County, Texas. -. 1961. Generalized soil map of McLennan County, Texas. -. 1961. Generalized soil map of Milam County, Texas. -. 1961. Generalized soil map of Robertson County, Texas. Templin, E. H., A. L. Nabors, T. R. Atkins, A. W. Crain, I. C. Mowery, D. T. Horton, J. S. Williams, and R. M. Voigtel. 1958. Soil survey of McLennan County, Texas. Soil Conservation Service (U.S.D.A) and Texas Agricultural Experiment Station, Series 1942, No. 17:1-124. 54 OCCASIONAL PAPERS MUSEUM TEXAS TECH UNIVERSITY Texas Water Development Board. 1968. The Texas water plan. Austin, Texas, 214 pp. Thaeler, C. S., Jr. 1968a. Karyotypes of sixteen populations of the Thomomys talpoides complex of pocket gophers (Rodentia-Geomyidae). Chro¬ mosoma, 25:172-183. -. 19685. An analysis of three hybrid populations of pocket gophers (genus Thomomys). Evolution, 22:543-555. -. 1968c. An analysis of the distribution of pocket gopher species in northeastern California. Univ. California Publ. Zool., 86:1-46. -. 1974. Four contacts between ranges of different chromosome forms of the Thomomys talpoides complex (Rodentia: Geomyidae). Syst. Zool., 23:343-354. Turner, G. T., R. M. Hansen, V. H. Reid, H. P. Tietjen, and A. L. Ward. 1973. Pocket gophers and Colorado mountain-rangeland. Colorado State Exp. Sta. Bull., 554S:l-90. Vaughan, T. A. 1967. Two parapatric species of pocket gophers. Evolution, 21:148-158. Villa, R. B., and E. R. Hall. 1947. Subspeciation in pocket gophers of Kansas. Univ. Kansas Publ., Mus. Nat., 1:219-236. Wahrman, J., R. Goitein, and E. Nevo. 1969. Mole rat Spalax; evolutionary significance of chromosome variation. Science, 164:82-84. Wentworth, G., and D. A. Sutton. 1969. Chromosomes of the Townsend pocket gopher, Thomomys toumsendii. Southwestern Nat., 14:157-161. Wright, S. 1943. Isolation by distance. Genetics, 28:114-138. Yates, T. L., and D. J. Schmidly. 1977. Systematics of Scalopus aquaticus (Linnaeus) in Texas and adjacent states. Occas. Papers Mus., Texas Tech Univ., 45:1-36. Addresses of Authors: R. L. Honeycutt, Department of Biological Sciences, Texas Tech University, Lubbock, Texas 79409; D. J. Schmidly, Department of Wildlife and Fisheries Sciences, Texas A&M University, College Station, Texas 77843. Received 31 October, accepted 30 November 1978. PUBLICATIONS OF THE MUSEUM TEXAS TECH UNIVERSITY Two publications of The Museum of Texas Tech University are issued under the auspices of the Dean of The Graduate School and Director of Academic Publications, and in cooperation with the International Center for Arid and Semi-Arid Land Studies. Shorter research papers are published as Occasional Papers, whereas longer contribu¬ tions appear as Special Publications. Both are numbered separately and published on an irregular basis. Institutional libraries interested in exchanging publica¬ tions may obtain the Occasional Papers and Special Publications by addressing the Exchange Librarian, Texas Tech University, Lubbock, Texas 79409. Individuals may purchase separate numbers of the Occasional Papers for $1.00 each from Texas Tech Press, Texas Tech University. Remittance must be enclosed with request. Institutional subscriptions also are available through Texas Tech Press.