ee eee —w ee a ~~ ee 6 es oe ae ee oe ee A ie By OE oe di a ce i ae Sr nr ee - ~ ’ eae as lati Teele te ee Pele Nee Fintan Ne Ba —— — = HARVARD UNIVERSITY EEL LIBRARY OF THE Museum of Comparative Zoology ee Orr 90 a a gah ign’ one, ech section a) = vine oe aa | . Nh, ry thane Liebides q yo J fi hy ’ a a! s Re Cae? « Po i a i ne Hid ae ny yy v6 an ae Me sist A et i af Pha ‘e JNIVERSITY OF KANSAS AUSEUM OF NATURAL HISTORY O- NA - LAWRENCE Boal Y-23-77 MUS. COMP. Zool: al LIBRARY APR 28 1973 i HARVARD Systematics of UNiversiqy, Three Species of Woodrats 4 (Genus Neotoma) in Central ~ North America aie es B i y } 2 nt .. _ Elmer C. Birney INIVERSITY OF KANSAS AWRENCE 1973 MISCELLANEOUS PUBLICATION No. 58 April 13, 1973 UNIVERSITY OF KANSAS PUBLICATIONS MUSEUM OF NATURAL HISTORY The University of Kansas Publications, Museum of Natural History, beginning) with volume 1 in 1946, was discontinued with volume 20 in 1971. Shorter research” papers formerly published in the above series are now published as Occasional Papers, Museum of Natural History. The Miscellaneous Publications, Museum of 7 Natural History, began with number 1 in 1946. Longer research papers are pub- lished in that series. Monographs of the Museum of Natural History were initiated in 1970. All manuscripts are subjected to critical review by intra- and extramural specialists; final acceptance is at the discretion of the publications committee. Institutional libraries interested in exchanging publications may obtain the Occa- sional Papers and Miscellaneous Publications by addressing the Exchange Librarian, University of Kansas Library, Lawrence, Kansas 66044. Individuals may purchase separate numbers of all series. Prices may be obtained upon request addressed to Publications Secretary, Museum of Natural History, University of Kansas, Lawrence, ‘ Kansas 66044. a UNIVERSITY OF KANSAS MusEuM OF NATURAL HISTORY MISCELLANEOUS PUBLICATION No. 58 April 13, 1973 Systematics of ‘Three Species of Woodrats (Genus Neotoma) in Central North America By ELMER C. BIRNEY James Ford Bell Museum of Natural History University of Minnesota Minneapolis, Minnesota A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy, The University of Kansas, 1970 UNIVERSITY OF KANSAS LAWRENCE 1973 UNIVERSITY OF KANSAS PUBLICATIONS, MUSEUM OF NATURAL HISTORY Editor: Linda Trueb Managing Editor: William E. Duellman MISCELLANEOUS PUBLICATION No. 58 pp. 1-173; 44 figures Published April 13, 1973 MusEuM oF NATuRAL History UNIVERSITY OF KANSAS LAWRENCE, Kansas 66044 WESzAe PRINTED BY UNIVERSITY OF KANSAS PRINTING SERVICE LAWRENCE, KANSAS CONTENTS ep E NCO) FD NO) Cry LON CDN a ee ar Bs 4 MATERIAES -AND* METHODS 2228 se Sif Let ON EERE, STs D INGKNOWELEDGMENTS 229007 ots 2 a eae eas, AO 7 AAS CONIC TTICGHIGT BE es Eva 0d Da EFL ff Sa a ey. SULT Pee ep Ae ae 8 PUISTORICATEE ACCOUNT Seewn® Dud 27 Seve eh a, oe 8 ACCOUNTS OF SPECIES AND SUBSPECIES 2°... SS ae ee ee ee 10 Western Subspecies of Neotoma floridana RIS ei a poo Rs BS 11 1 INCOLOTRAGIIUGTOPUS: = Mies OSA 1. Ae ee ee 2 | INC OLOMMERGNOUSTIPOLAUC Smsink sk ee ae S'S) COMPARATIVE MORPHOLOGICAL ANALYSES Ee eee 36 MMAGTERTATES ANID WiiHONGr es 22 Se 2 oo 4 8) a a ee ee F756 INGN-CEOCRAPHEGHW ARTATIONG = 0250s 28 ek ee 49 NaniatlOMunval het ACCre: ste 82.5) 5 6 ee oe eee 42, Secondany sexual) Variation, 0-52. 2 7 ee eee ee 48 erclivacl rae Walt Oise coe ee ee . 64 Variation Resulting from Captivity ETON SE MID A Mn 56 [SEOCEAPHIC AV ARPAMION = seeeee ei Ss Se ee Se ee eee 57 Rela cemNolissanduC@Olon gees. a! 5s Ee ee ee eee Sy @uatatives Grantalt Characters a2 = 2. . ee e 66 Bacttltnieeeesemeettere he. ae ae ee, ee eh eee ee ee 76 WnivariateAnalyses.of Mensural Characters) = ==) = eee 78 Multivariate Analyses of Mensural Characters _________-_________ 99 Multivariate Analyses of Size, Color, and Qualitative CramaleCharactenrs) stctit! sor. sae 2S) ree eee 109 Discriminant-hunction, Analyses... eee 120 Pe MORPHOLOGICALACHARACTERS 2222.2.55 2555) 2a ee 127 ROnMmPARATIVE PAmPRODUGTION. -) 9 i)... Ae ee ee 127 BORED AR AGES OE ROL OGY yee oe 0s 144 Starchs Gelstilectrophoresis of Hemoglobins — — =.) Se 144 mnnmoecleetrophoresissor Esterases: __.____. - ) ee eee 148 SO NAV ACTIVE KAR VOL OG Vee Sas ee ee 153 SeMMARY AND: ZOOGEOGRAPHIC CONSIDERATIONS = 157 SUCCESHIONS FORVADDITIONALSRESEARCH ...____.) a ee 160 POOCEOCRAPHIC# CONENENTS Be 22a 5: =... eee ee 161 TESTU Ria) a 08 De i eee nas 166 INTRODUCTION Two species of woodrats, Neotoma floridana and Neotoma micropus, have allopatric, but adjacent, distributions on the Great Plains; the species inhabit dis- tinctly different environments at most localities. Because N. floridana is brown and N. micropus is gray, the species are readily distinguishable. The geographic ranges of the two, as mapped by Hall and Kelson (1959:684), were known to abut in the central and southern Great Plains from southeastern Colorado and southwestern Kansas southward to the Gulf Coast of eastern Texas; but not a single locality of sympatry was known. However, it had been discovered (Dwight Spencer, pers. com.) that mem- bers of the two species would hybridize in the laboratory with the production of viable offspring. The polytypic characteristics of both species in the zone of potential sympatry and the existence of a geographically isolated subspecies of Neotoma floridana make the problem even more interest- ing from an evolutionary point of view. Neotoma floridana baileyi is restricted to the region of the Niobrara River in north- central Nebraska. Neotoma_ floridana campestris is a large pallid subspecies that lives in eastern Colorado, southwest- ern Nebraska and northwestern Kansas. Rats of this race may have been isolated in post-Wisconsin times from presently contiguous populations of the species to the east (Jones, 1964:26). The eastern parts of Kansas, Oklahoma, and northern Texas were inhabited by a dark brown race known as N. f. osagensis. In south- em Texas the range of N. f. attwateri abuts that of N. micropus; specimens of floridana from farther east in Texas have been referred to the subspecies rubida. When this study was begun, Neotoma micropus was divided into five nominal subspecies. Neotoma micropus micropus occupied roughly the eastern half of the range of the species from Tamaulipas to southern Kansas. Much of the western part of the range reportedly was occu- pied by N. m. canescens, allegedly a smaller and more pallid race. Two other subspecies, N. m. leucophea and N. m. planiceps, were known only from their respective type localities in New Mexico and San Luis Potosi. The fifth recognized subspecies, N. m. littoralis, was known only from a few localities in southern Tamaulipas. Initially, this study was centered in Nebraska, Colorado, Kansas, and Oklahoma, an area which includes the northern half of a zone in which the geographic range of Neotoma floridana approaches that of N. micropus. The study area was selected principally be- cause three subspecies of N. floridana ( baileyi, campestris, and osagensis) and two of N. micropus (canescens and mi- cropus ) occur within its boundaries. This facilitated comparison of a variety of parameters in order to determine rela- tionships among these five taxa. Al- though it had been supposed that the ranges of N. f. campestris and N. f. osag- ensis met in north-central Kansas, the exact zone of contact and the nature of the interaction had not been documented. Moreover, N. albigula generally is con- sidered to be closely related to both N. floridana and N. micropus (Burt, 1960; Hooper, 1960); this species occurs in southeastern Colorado and the western part of the Oklahoma panhandle. Finley (1958) suggested that N. albigula and N. micropus formed natural hybrids in this region. Finally, the area was selec- ted because of its accessibility to Law- rence, Kansas. This facilitated both field and laboratory investigations. Ultimately, all available specimens of both species from the initial region of study and from near the zone of potential contact in Texas were examined. All specimens of N. micropus in the Museum of Natural History of the University of Kansas were included, and selected spec- imens of N. micropus were examined in WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 5 other museums if it seemed likely that they might reveal information on the re- lationship of this species to N. floridana. Eastern subspecies of N. floridana were treated taxonomically by Schwartz and Odum (1957); with the exception of N. f. rubida, which occurs in extreme south- eastern Texas, the eastern subspecies were not considered in my investigation. One other species, Neotoma angus- tipalata, is considered herein. The affi- nities of this large woodrat, known only from Tamaulipas and San Luis Potosi, Mexico, have remained enigmatic since it was described by Baker (1951) as a member of the Neotoma mexicana spe- cies-group. Neotoma angustipalata since has been considered by different authors to be closely related to N. mexicana, N. micropus or N. albigula. Parameters studied in the field in- cluded habitat preference and utilization, exact distributional relationships in those areas where members of two taxa might come into contact, and seasonal repro- ductive patterns of natural populations. Parameters studied in the laboratory in- cluded data on the following: 1) control and experimental matings of members of each taxon to those of each other taxon; 2) mating success and fecundity of hy- brids; 3) growth and development of hybrids and nonhybrids; 4) karyological analyses; 5) serological studies involving hemoglobin — electrophoretic patterns (Birney and Perez, 1970) and immuno- electrophoretic reactions of esterases; 6) water balance physiology (Birney and Twomey, 1970); and 7) univariate and multivariate analyses of geographic varia- tion of mensural and qualitative morpho- logical characters. The primary purposes of my research have been to elucidate the systematic and evolutionary relationships, assess the zoogeographic history and _ re-evaluate the nomenclatorial arrangement of the woodrats studied. A secondary purpose has been to compare so-called classical taxonomic procedures with some of the newer methods of systematics and thereby evaluate the applicability of the various methods to systematic studies of closely related mammalian taxa. MATERIALS AND METHODS The general materials and methods that pertain to several facets of the study are discussed below. Specific materials and methods are related in detail pre- ceding results and discussion of the vari- ous topics covered. Initial efforts to collect live woodrats for laboratory studies were undertaken in September 1966. The last animals of the colony were sacrificed in October 1969. Specimens studied in the labora- tory were obtained either by dismantl- ing active dens and capturing the rats by hand as they fled, or by trapping them in live-traps set near active dens. Hava- hart traps (18 by 5 by 5 inches) were found to be highly successful, easily transported, and_ relatively durable. Woodrats are not difficult to trap and can be taken in practically any device large enough to permit entry and con- structed to prevent escape. In rocky habitats, usually it was necessary to use traps, but in other areas woodrats were captured more often by hand. To pre- vent undue destruction of available den- ning sites at two localities of special in- terest (Major County, Oklahoma, and Cherry County, Nebraska), woodrats were obtained only by trapping. Animals used in laboratory experi- ments were obtained during the months indicated at the following localities (here specified only to county; see lists of spec- imens examined for exact localities of record within these counties): NE- BRASKA: Cherry County (March and April 1967, August 1968); Rock County (August 1968). COLORADO: Baca County (April and May 1968); Prowers County (April 1968). KANSAS: Barber County (October 1966, March and July 1968); Douglas County (September, Oc- tober, and November 1966, March 1967, February and March 1968, March 1969); Ellis County (December 1966); Ells- worth County (September 1967, October 6 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY and December 1968); Finney County (September and October 1968); Hamil- ton County (September 1968); Haskell County (September and November 1966, June 1967, February, May, and August 1968, April 1969); Hodgeman County (September and November 1968); Logan County (August 1967); Meade County (November 1966, June 1967); Ness County (September 1968); Russell County (December 1968); Stevens County (August 1968). OKLAHOMA: Dewey County (June 1968); Major County (June 1968, January 1969). Woodrats were housed indoors and maintained on a daily regime of 14 hours of light and 10 hours of darkness from 1 February to 1 October. Illumination from several windows in the animal rooms was the only source of light dur- ing other months. An attempt was made to maintain a stable temperature of 20°C in the animal rooms, but temperatures ranged from as low as 13°C at times dur- ing winter to as high as 30°C on occa- sional summer afternoons. During one three-day period in June 1968, air con- ditioning failed and the temperature soared to at least 35°C and may have surpassed 38°C. No deaths were attrib- uted to extremes in temperature, but con- sumption of water increased noticeably as temperature increased. Relative hu- midity was not measured in the animal rooms during the summer or winter, but was measured regularly from March to June, 1968, when it ranged from 15 to 45 percent. Woodrats were caged individually ex- cept when two were placed together for breeding or when a female was rearing a litter. Litters were weaned at six weeks of age. Cages used to house woodrats were of three general types. One type was constructed of wood and %-inch mesh hardware cloth. Dimensions of these cages were 30 by 18 by 18 inches with 3.75 square feet of floor space. The other two types of cages used were all- metal commercially available cages with approximately four square feet of floor space. Both were satisfactory, but the type having a removable pan beneath a grated floor could be cleaned easily with- out disturbing the occupant. Females with unweaned litters were kept in spe- cial large metal cages that had eight square feet of floor space. Although woodrats can be maintained in smaller cages, I doubt that these would be satis- factory for maintenance of a breeding colony (see Wood, 1935:109). All cages were supplied with one gallon cardboard milk cartons or other disposable nest boxes of equivalent size. Cages having solid floors were covered with wood shavings; shredded newspaper for use as nesting material was available in all cages. Cages were cleaned at weekly intervals. Purina Laboratory Chow and water were available to woodrats in the labora- tory on an ad libitum regime. On occa- sion, especially in the early phases of the study, this diet was supplemented with lettuce leaves and whole-kernel corn; when it became evident that supple- mentary foods were unnecessary, this practice was discontinued. At times in- dividual rats would severely reduce food intake and begin to lose weight, usually indicating that the incisors were broken or maloccluding; rats with such teeth were removed from the colony. Occa- sionally the teeth of rats that were not feeding properly appeared to be normal; on these occasions feeding of laboratory chow was discontinued and the animals were given rolled oats ad libitum for a few days, then gradually returned to a diet of laboratory chow. While conduct- ing one experiment wherein it was neces- sary to limit protein intake, the experi- mental group was fed only corn (see Birney and Twomey, 1970). Diseases and ectoparasites caused lit- tle problem in maintaining a_ thriving woodrat colony. In the spring of 1967 several rats died after having had diar- rhea for two to five days, losing weight, and having an inactive, sickly appear- ance. Several sick rats and some that recently had died were taken to the Veterinary Diagnostic Laboratories afhil- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA U iated with Kansas State University at Manhattan, Kansas. Although the dis- ease was never diagnosed, the necropsy report (H. D. Anthony, pers. com.) stated that a “hemolytic E. [scherichia] coli was isolated from the spleen and in- testine of one of the specimens.” Anthony recommended immediate treatment of drinking water with nitrofurazone, fol- lowed by prophylactic doses every three months. This treatment prevented spread of the disease, and generally cured all but the sickest rats. On two later occasions when more than one rat evinced signs of the affliction, the entire colony was treated. Cages were cleaned after each occupancy and those that had housed sick woodrats were washed in a dilute lysol solution before being reused. To prevent ectoparasites from becom- ing a problem, each woodrat was dusted with commercial “flea powder” before being placed in the animal house. On three separate occasions individual wood- rats became infested with an unidentified species of mite; the entire colony subse- quently was dusted and the infested animals were dusted on two or three suc- cessive days. Although each of the indi- viduals survived, two females infested at the time of parturition abandoned their litters. Rats infested with mites tend to be lethargic, lose weight rapidly, and have matted, swollen eyes. In addition to study of live animals, a total of 2163 museum specimens, in- cluding seven holotypes or lectotypes, was examined by me. Several hundred additional specimens were examined. These include more than 300 laboratory- reared individuals prepared as museum specimens and deposited in The Museum of Natural History of the University of Kansas, specimens of eastern subspecies of Neotoma floridana not treated herein, specimens of other species of Neotoma (especially N. albigula and N. mexicana) examined for comparative purposes, and woodrats examined incidentally while searching for misidentified specimens of the species studied. ACKNOWLEDGMENTS I thank the following agencies for financial support for this investigation: 1) National Science Foundation Grant GB-4446X administered by the Commit- tee on Systematics and Evolutionary Bi- ology at The University of Kansas, pro- vided a Research Traineeship for my support from September 1966 until June 1969 and additional funds for cages and equipment to maintain woodrats in the laboratory; 2) Travel grants for field work and museum visitations were re- ceived from the Graduate School and the Museum of Natural History (Watkins Grant) at The University of Kansas; 3) Research grants from the Kansas Academy of Science supplied monies used to purchase Laboratory Chow; 4) Funds for computer time were provided by the Computation Center of The Uni- versity of Kansas; 5) Partial support for July and August 1969 was made avail- able through The University of Kansas General Research Fund (Biomedical Di- vision), under grant no. 3453-5038; 6) Equipment and supplies for serological experiments and rabbits used for prepa- ration of antisera were purchased with funds from the Public Health Service and research monies of the Department of Zoology at The University of Kansas. I am indebted to each of the persons listed below for permission to examine specimens of woodrats in their charge, which included both arrangement of loans and provision of working space in their respective institutions (Museum abbreviations used in the accounts be- yond are in parentheses): Sydney An- derson and Richard G. Van Gelder, American Museum of Natural History (AMNH); William J. Voss, Fort Worth Museum of Science and History (FWCM —formerly Fort Worth Children’s Mu- seum); John P. Farney, Kearney State College (KSC); Dwight L. Spencer, Kansas State Teachers College (KSTC); Robert S. Hoffmann and J. Knox Jones, Jr., Museum of Natural History of The University of Kansas (KU); George H. 8 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY Lowery, Jr., Museum of Natural Sciences of Louisiana State University (LSU); Eugene D. Fleharty, Museum of the High Plains of Fort Hays Kansas State College (MHP); Walter W. Dalquest, Midwestern University (MWU); Bryan P. Glass, Oklahoma State University (OSU); Edwin D. Michael, formerly at Stephen F. Austin State University (SFA); J. Keever Greer, Stovall Museum of Science and History of the University of Oklahoma (SM); Dilford C. Carter, formerly with the Texas Cooperative Wildlife Collection of Texas Agricultural and Mechanical University (TCWC); W. W. Newcomb, Texas Natural History Collection of the University of Texas (TNHC); Robert J. Baker and Robert L. Packard of Texas Tech University (TT); Bernardo Villa-R., Instituto de Biologia, Universidad Nacional Auton- oma de México (UNAM); Charles O. Handley, Jr., and R. H. Manville, United States National Museum, including the Biological Surveys Collection (USNM). Serological experiments were con- ducted in the Charles A. Leone Labora- tories at The University of Kansas. Per- sons granting permission to use these facilities or who were especially helpful include Jay D. Gerber, Richard D. Koehn, Charles A. Leone, Robert B. Merritt, and Julio E. Perez. Karyological studies were conducted in the botany laboratories at The University of Kansas and in the cell biology laboratories at Texas Tech University. Raymond C. Jackson at Kansas and Robert J. Baker at Texas Tech were especially generous with their supplies, equipment, and time. James Fraine and Sievert A. Rohwer as- sisted in the use of numerical taxonomy programs and other analyses of data, all TAXONOMIC TREATMENT HISTORICAL ACCOUNT Although woodrats were observed in North America by early naturalists such as John Bartram, no species was treated taxonomically until early in the nine- teenth century. Ord (1818:181) pub- of which were conducted at the Compu- tation Center of The University of Kansas. I am especially indebted to J. Knox Jones, Jr., formerly of the Museum of Natural History and Department of Sys- tematics and Ecology of The University of Kansas, for his excellent advice and general assistance throughout this study and for editorial advice in preparation of the manuscript. Frank B. Cross and Raymond C. Jackson also critically read the manuscript and provided sound edi- torial advice. Persons deserving of special thanks for assistance in the field include David M. Armstrong, Russell H. Birney, Jay D. Gerber, Thomas H. Kunz, D. Michael Mortimer, Robert R. Patterson, Duane A. Schlitter, Ronald W. Tumer, and Larry C. Watkins. Additionally, D. Michael Mortimer assisted by cleaning woodrat cages in the summer of 1968 and Hugh H. Genoways fed and watered the animals when it was necessary for me to be absent. Several of my colleagues at The Uni- versity of Kansas, especially David M. Armstrong, Hugh H. Genoways, Robert S. Hoffmann, and Carleton J. Phillips, were both stimulating and helpful by giving needed technical advice and will- ingly discussing topics germane to my research. Sydney Anderson, the Ameri- can Museum of Natural History, assisted this study in a similar manner. My wife, Marcia Bimey, has been both helpful and patient throughout this study and my graduate career. She pre- pared most of the figures used herein, typed drafts of the manuscript, and pro- vided much needed clerical assistance. lished a short description and figure of Mus floridanus, a woodrat from eastern Florida. Previously, Ord (1815:292) described the bushy-tailed woodrat as Mus cinereus. Having noted the distinc- tive dental characters of the New World WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 9 rats, Say and Ord (1825:345) diagnosed and named the genus Neotoma and des- ignated Mus floridanus as the type spe- cies. Later Baird (1855:333) named Neotoma micropus from the type locality of Charco Escondido, Tamaulipas (see remarks following synonymy of N. m. micropus ). Baird (1858:487-490, 492-495) treated Neotoma floridana and N. micropus as separate species, although he repeatedly commented on the similarities between the two. Having only the type of micro- pus with which to compare specimens of floridana, Coues (1877:15) considered micropus a synonym of floridana, a con- clusion based primarily on his interpreta- tion of the gray pelage of micropus as being that of an immature animal. With additional specimens of micropus from Tamaulipas, southern Texas, and from “the northwestern corner of the Indian Territory” (now western Oklahoma), Al- len (1891:282) recognized micropus as a species distinct from floridana. Allen (1891:285) applied a new subspecific name, Neotoma micropus canescens, to specimens from the Indian Territory pri- marily on the basis of their pallid colora- tion. Woodrats from a population at Valen- tine, Nebraska, were collected in June of 1888 by Vernon Bailey and later named as a new species, Neotoma baileyi, by Merriam (1894a:123). Merriam (1894b) published a synopsis of the known mem- bers of the genus Neotoma, including fossil relatives, and diagnosed the sub- family Neotominae. Allen (1894b:322) described Neotoma campestris, the flori- dana-like woodrat of northwestern Kan- sas and northeastern Colorado, on the basis of specimens from Pendennis, Kan- sas (type locality), and Fort Lyons, Colorado. In the same publication, Allen (1894b:323) noted that he considered the woodrats which he previously had allocated to Neotoma micropus canescens to be “inseparable from N. micropus.” Neotoma attwateri was named on the basis of a sample of woodrats from just east of the Edwards Plateau, near Kerr- ville, Texas (Mearns, 1897:721). Mearns (1897:722) suggested that ‘it is not im- probable [that N. attwateri, N. baileyi, and N. campestris] . . . may prove to be but geographic races of N. floridana.” Prior to the turn of the cen- tury, only one other name was applied to the woodrats considered here. Elliot (1899:279) assigned the name Neotoma macropus [sic] surberi to specimens from the vicinity of Alva, Oklahoma. Accord- ing to Elliot, surberi differed from both micropus and canescens in having a longer tail and darker pelage. Neotoma micropus littoralis, from Altamira, Tamaulipas, and Neotoma mi- cropus planiceps, based on a single spec- imen from Rio Verde, San Luis Potosi, were the last names applied (Goldman, 1905:31 and 32, respectively) to the woodrats treated in this study prior to the revision of the genus Neotoma by Goldman in 1910. In the latter work, Goldman recognized three subgenera and 28 species. Only two species, Neo- toma (Homodontomys) fuscipes and Neotoma (Teonoma) cinerea, were not considered members of the subgenus Neotoma. Woodrats presently included within the genus Neotoma, but which were considered as separate genera at that time are Neotoma (Hodomys) alleni and Neotoma (Teanopus) phenax (see Burt and Barkalow, 1942:296). Burt and Barkalow also placed Homodontomys in the synonymy of Neotoma. Goldman (1910:14) considered N. floridana and N. micropus to be separate but closely related species comprising the floridana species-group. Neotoma flori- dana baileyi was thought to occur in South Dakota, most of Nebraska and Kansas, and in eastern Colorado. The name campestris was treated as a junior subjective synonym of baileyi. Wood- rats of eastern Texas (except the extreme eastern part, where the name N. f. rubida Bangs was applied) and eastern Okla- homa, were assigned to N. f. attwateri. Goldman (1910:26-31) reinstated N. mi- cropus canescens as the best name for western populations of that species, 10 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY placed surberi in the synonymy of N. m. micropus, and recognized both N. m. littoralis and N. m. planiceps as distinct subspecies. Since Goldman’s revision, the nomen- clature of woodrats has remained rela- tively stable. Kellogg (1914:5) removed campestris from the synonymy of baileyi and recognized both as subspecies of floridana. Goldman (1933:472) gave a slightly paler (as compared to N. m. canescens) population of micropus the name Neotoma micropus leucophea. Blair (1939a:5) described woodrats from eastern Oklahoma, eastern Kansas, and adjacent parts of Missouri, Arkansas, and Texas as Neotoma floridana osagensis (type locality in Osage County, Okla- homa). Recognition of osagensis limited the distribution of baileyi to the Noibrara Valley of northern Nebraska. Burt and Barkalow (1942:290) considered N. mi- cropus to be intermediate between N. floridana and N. albigula. On that basis they created the micropus species-group thus removing micropus from the flori- dana group where it had been placed by Goldman (1910:14). Neotoma floridana and N. micropus were studied by Spen- cer (1968), who concluded that the two species are closely related and incom- pletely speciated. On the basis of two specimens from the Sierra Madre Oriental of southern Tamaulipas, Baker (1951:217) named the species Neotoma angustipalata. Hooper (1953:10) suggested that angus- tipalata may represent no more than a deeply pigmented population of micro- pus, and Hall (1955:329) thought it should be placed in the albigula species- group. ACCOUNTS OF SPECIES AND SUBSPECIES I regard the biological species con- cept (Wilson and Brown, 1953:97-99; Mayr, 1963, 1965, and 1969) as the best presently available concept of the species both for evolutionists and taxonomists, and subscribe to Tilden’s (1961:22) statement on the use of subspecies as follows: “In defense of the use of the subspecies concept, it may be mentioned that in our present system of classifica- tion the subspecies is the category ex- pressly provided for the treatment of populations less than species. That this tool is imperfect must be admitted. But to admit imperfection is not necessarily to reject the tool entirely. The point of view is held here, that the good results outweigh the objections that have been brought forward.” Because of the arbi- trary nature of the subspecies, it is neces- sary to state what “kind” of subspecies is to be recognized. Lidicker (1962:169) stated that “a subspecies is a relatively homogeneous and_ genetically distinct portion of a species which represents a separately evolving, or recently evolved, lineage with its own evolutionary ten- dencies, inhabits a definite geographical area, is usually at least partially isolated, and may intergrade gradually, although over a fairly narrow zone, with adjacent subspecies.” Further, it was noted by Lidicker (loc. cit.) that although most such subspecies will not become species, they are populations that have made ini- tial steps toward species formation and could form species under suitable isolat- ing conditions. This interpretation fo- cuses on the evolutionary process of spe- ciation rather than on individual geo- graphically variable characters. It is this sort of subspecies that I have attempted to recognize. Taxonomic decisions were made after intensive study and evaluation of the morphological, reproductive, serological, and karological data discussed beyond. Accounts of species and subspecies are presented first merely as a matter of con- venience so that the nomenclatorial ar- rangement proposed herein will obtain throughout subsequent discussions. Eight nominal taxa of woodrats, rep- resenting three species, are treated in the accounts that follow. The arrangement of species and that of subspecies within a species does not imply relationship or degree of specialization. Instead, taxa WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 11 are arranged with respect to geographic distribution, with more northern taxa treated before southern ones. Each ac- count includes a basic synonymy, which is followed by a relatively brief section of remarks. Remarks include comments on type specimens if such are appropri- ate, general comments on the variation within and between subspecies, and other comments that may be germane to the taxonomic status of the taxon under consideration. Records of occurrence fol- low remarks and include both specimens examined and additional records. The total number of specimens exam- ined is given for each taxon. This is fol- lowed by exact localities from which the specimens originated, the number of specimens examined from each locality, and the abbreviated designation for the museum(s) in which specimens are housed. In the account of N. m. canes- cens, specimens from localities in the United States are listed before those from México. States and counties within the United States are arranged alpha- betically. Localities within each county and Mexican state are arranged from north to south. If two localities are at the same latitude, the westernmost is given first. Locality data of some speci- mens examined were specific only to county; these are listed as “unspecified” after specific localities within the county. In a few instances, a group of specimens is from the same general locality with the exact localities of capture varying only slightly; these are totaled and listed col- lectively as within a certain radius of a single locality. Specimens judged to be “hybrids” of two species (see account of N. m. canescens) are listed with speci- mens of the taxon that they most resem- ble. Subspecific intergrades are included under the subspecies to which I consider them best assigned. Published citations to localities from which specimens have been collected but not examined by me are listed under “Additional records.” Also included in this category are reli- able, published observations of woodrats or their houses in areas known to be in- habited by a single species of Neotoma. Following records of occurrence, the distribution and habitat of each taxon are discussed. The latter topics are treated in especial detail if they directly con- cern the relationships of two species or two subspecies. The general ranges of all three species (floridana, micropus, and angustipalata) are shown in figure 1. Most locality records are plotted on re- gional maps, but some were not plotted to prevent undue crowding of symbols. Unplotted localities are set in italic type in lists of specimens examined and of additional records. Localities specified only to county were plotted only in the absence of any other county record; sym- bols for such records are square and placed near the center of the county. Localities of specimens bearing ques- tionable data are not plotted, but are dis- cussed in the account of N. m. canescens. Eastern subspecies of Neotoma flori- dana were reviewed by Schwartz and Odum (1957). With the exception of N. f. rubida, which I have treated only in eastern Texas, these races of the spe- cies are not considered here. Specimens of Neotoma micropus were — studied throughout the range of the species. However, because New Mexico is well to the west of the range of N. floridana and to the north of that of N. angustipalata, no attempt was made to examine all specimens from that state. Specimens from there were studied primarily to determine the best taxonomic position of the subspecies N. m. leucophea and to determine if New Mexican woodrats fit the pattern of variation of the species in a general way. In the accounts that fol- low, therefore, records of N. micropus from New Mexico are not plotted on a regional map and records from the pub- lished literature are limited to those that are distributionally marginal. Western Subspecies of Neotoma floridana Neotoma floridana baileyi Merriam Neotoma baileyi Merriam, 1894a:123 [Holotype 12 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY Fic. 1. Geographic distributions of Neotoma angustipalata, N. floridana, and N. micropus. Identi- 1) N. angustipalata; 2) N. f. attwateri; 3) N. f. baileyi; 4) N. f. campestris; 5) N. f. floridana; 6) N. f. haematoreia; 7) N. f. illinoensis; 8) N. f. magister; 9) N. f. rubida; 10) N. f. smalli; 11) N. m. canescens; 12) N. m. micropus; and 13) N. m. planiceps. The symbol in Oklahoma denotes the single known locality of sympatric occurrence of N. floridana and N. micropus. Distribution of eastern races of N. floridana follows Hall and Kels- fication of species and subspecies is as follows: son (1959:634). —USNM 4311/5034 from Valentine, Cherry County, Nebraska]. Neotoma floridana baileyi Bailey, 1905:109. Remarks.—Because of its present geographic isolation, Neotoma floridana baileyi assumes certain characteristics of an “insular” subspecies. Although mem- bers of this subspecies are distinctive in at least minor characteristics of every parameter studied, no results obtained in this study indicate that baileyi has evolved to a level warranting specific status. Records of occurrence.—Specimens exam- ined (56).—NEBRASKA: Cherry County: Val- entine, 6 (USNM); 4 mi E Valentine, 6 (KU); 6 mi E Valentine, 15 (KU); Clark’s Canyon, near Valentine, 12 (USNM); 3 mi SSE Valen- tine, 1 (KU); 10 mi S Cody, 5 (USNM); 22 mi SW Valentine, 3 (KU). Keya Paha County: 6 mi S, 8 mi E Springview, 1 (KU). Rock County: 11.5 mi N, 7.5 mi W Bassett, 7 (KU). Additional record—NEBRASKA: Brown County: Long Pine (Jones, 1964:218). Distribution and habitat—Selected localities of recorded occurrence of Neo- toma floridana baileyi are plotted in fig- ure 2. Goldman (1910:25) reported as WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 13 this subspecies a woodrat from 18 mi SE Rapid City, South Dakota, but Jones (1964:217) stated that the individual re- corded “is without question Neotoma cinerea rupicola.” I have searched ex- tensively for Neotoma floridana in Todd, Mellette, and Tripp counties, South Dakota, but found neither the woodrats nor their distinctive dwellings. Habitat that appeared suitable for woodrats was extensive along and near the Little White and Keya Paha rivers, and in the vicinity of Okreek, South Dakota. Although no specimens were taken in South Dakota, it seems likely that N. f. baileyi will be found there eventually. However, it is conceivable that dispersal farther north- ward is not possible even for a popula- tion long in the process of adapting to the inclement winters of northern Ne- braska. 100 __ 24-5B4E Fic. 2. Selected locality records for Neotoma floridana baileyi (symbols solid above) and N. f. campestris (symbols solid below) in Nebraska. In Cherry, Keya Paha, and Rock counties, Nebraska, N. f. baileyi occurs in three more or less distinguishable habitat types. On the Fort Niobrara Wildlife Refuge these rats were abun- dant in the heavily wooded floodplain of the Niobrara River in March and April 1967 and August 1968. Nests were constructed in and around fallen trees, inside hollow upright trees, at the bases of upright trees, and in piles of brush and treelimbs. In April, the bark and cambium layers of woody twigs appeared to serve as a primary source of food. Just north of the Snake River Falls in Cherry County (22 mi SW Valentine), three specimens were taken from nests con- structed in the steep, rocky, canyon walls bordering the Snake River. In Keya Paha County (6 mi S, 8 mi E Spring- view) a single specimen was trapped under a rocky ledge at the top of a deep canyon several miles north of the Nio- brara River; no additional nests or signs of woodrat activity were located as a result of searching similar ledges in the immediate area and lower in the same and adjacent canyons. Habitat of this 14 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY type apparently is marginal and not fre- quently utilized. Woodrats were common in aban- doned and little-used buildings at a ranch near Long Pine Creek (11.5 mi N, 7.5 mi W Bassett) in Rock County in 1968. At this locality I found no evi- dence that “natural” nest-site types (e.g. rock outcrops, trees, logs, brushpiles) harbored woodrats, although these sites probably are utilized occasionally. Near Long Pine Creek at Long Pine, Brown County, where this woodrat has been reported to occur (Jones, 1964:218), I found the habitat even more extensive than at the locality discussed above. A superficial search there for woodrats was unsuccessful, but I am confident they occur in the area and probably are locally common along the creek from Long Pine to the place where it empties into the Niobrara River. I did not revisit the locality of record 10 mi S Cody nor have I searched for N. f. baileyi along the Niobrara or elsewhere to the west of that locality. To the east of the easternmost locality of record (southwest of Spring- view), I have searched for these wood- rats as far as eastern Boyd and Holt counties. Apparent marginal habitat was found on Ponca Creek near Spencer (Boyd County) and at several localities along the Niobrara River. Another area worthy of further search was observed south of the Niobrara River just east of Midway in Holt County. I suspect that N. f. baileyi occurs farther to the east than present records indicate. Neotoma floridana campestris J. A. Allen Neotoma campestris J. A. Allen, 1894:322 [Holotype—AMNH _ 7765/6742 from Pen- dennis, Lane Co., Kansas]. Neotoma floridana campestris—Kellog, 1914:5. Remarks.—Although not recognized by Goldman (1910:24), Neotoma flori- dana campestris is distinctly paler in color than adjacent populations of N. f. attwateri, occupies relatively distinct types of habitat in comparison with other populations of the species, and tends generally to be larger than rats to the east. Although relatively narrow, the area of contact between campestris and attwateri in eastern Ellsworth and west- ern Russell counties, Kansas, forms an obvious zone of integradation. The arbi- trary line dividing the ranges of the two taxa is drawn on a north-south axis gen- erally corresponding to the county line separating Ellsworth and Russell coun- ties. A specimen (KU 119700) assigned to campestris from only one mile west of that county line might be equally well assigned to attwateri, but two specimens (KU 14001-02) from one mile east of the line clearly are best assigned to attwateri. The county line serves as a convenient line of demarcation and is as accurate as any other would be. Records of occurrence-—Specimens exam- ined (221)—COLORADO: Crowley County: 3 mi N Fowler, 4400 ft, 7 (KU); Olney (= Olney Springs), 12 (USNM). El Paso County: 1.5 mi SW Fountain, 5700 ft, 2 (KU); 2.5 mi SW Fountain, 5700 ft, 1 (KU); 3 mi S, 2 mi W Fountain, 5600 ft, 1 (KU). Kit Carson County: Tuttle, 2 (USNM). Yuma County: Wray, 5 (1 USNM, 4 AMNH); 1 mi S Wray, 3550 ft, 3 (KU); 2 mi W Hale, 1 (KU); 1 mi S, 3 mi W Hale, 1 (KU). KANSAS: Decatur County: 5 mi S, 8 mi W Oberlin, 1 (KU). Ellis County: 16 mi N Hays, 13 (KU). 13 mi N; 1 mi We Haysvel (MHP); SE % sec. 28, T. 11 S, R. 18 W (13 mi N Hays), 6 (MHP); NW # sec. 31, T. 13 S, R. 18 W (2 mi W Hays), 3 (MHP); Hays, 7 (USNM); 0.5 mi S, 3.5 mi W Hays, 1 (KU); 2 mi S Hays, 1 (MHP); 7 mi S, 10 mi W Hays, 3 (KU); NW X sec. 11, T. 15 S, R. 20 W (8 mi S, 10 mi W Hays), 2 (MHP); SW & sec. 16, T. 15 S, R. 19 W (9 mi S, 6 mi W Hays), 1 (MHP). Finney County: 19 mi S Dighton, it (KW): 23 mi S Dighton, 2) (KU) ys Gouve County: Castle Rock, 9 (KU). Hodgeman County: 4 mi S, 0.5 mi W Jetmore, 2 (KU). Lane County: 1 mi N Pendennis, 10 (KU); Pendennis, 29 (6 AMNH, 23 USNM); 12 mi SW Pendennis, 2 (KU); unspecified, 2 (KU). Logan County: NE i sec. 8, T. 13 S, R. 35 W (2.5 mi N, 2.5 mi W Russell Springs), 1 CMMRIR RINGS, Aaa Pie Th Is) S. Jae Sie WY (UI mi S, 1 mi W Russell Springs), 1 (MHP); 1 mi S Russell Springs, 6 (KU); 5 mi SE Elkader, 2 (KSTC); 5 mi S Elkader, 4 (KU); un- specified, 4 (KU). Ness County: 1 mi S, 16 mi W Ness City, 4 (KU). Rawlins County: 7 mi N, 16.5 mi W Atwood, 1 (KU). Rooks County: 1.5 mi S, 1 mi W Stockton, 6 (MHP); 3 mi S, 3 mi W Stockton, 1 (MHP); 6 mi SW Woodston, 3 (KU); 20 mi N Hays, 1 (MHP); WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 15 SW 4% sec. 34, T. 10 S, R. 17 W (7 mi S, 4.5 mi E Plainville), 1 (MHP). Russell County: ME msec, 34. tT. 12'S; R. 11 W (6 mi S,, 1 mi E Lucas), 1 (MHP); NW &% sec. 8, T. 13 S, R. 11 W (8 mi S, 2 mi W Lucas), 1 (MHP); me sec, 17. T. 13)S, BR. 11 W (9 miS; 1 mi W Lucas), 4 (MHP); SE % sec. 13, T. 13 S, R. 11 W (10 mi S, 3 mi E Lucas), 1 (MHP); 0.5 mi W Russell, 1 (KSTC); NW % sec. 34, T. 13 S, R. 12 W (11 mi E Russell), 1 (MHP); 2 mi W Wilson, 1 (KU); 6 mi S$, 4 mi E Russell, 18 (KU). Scott County: State Park, 1 (KU); 12 mi N, 3 mi W Scott City, 2 (MHP). Thomas County: unspecified, 1 (MHP). Trego County: sec. 29, T. 13 S, R. 25 W (Banner), 6 (KU); unspecified, 1 (USNM). Wallace County: Lacey Ranch (9 mi S, 4.5 mi E Wallace), m( KU). NEBRASKA: Dundy County: 5 mi N, 2 mi W Parks, 8 (KU); Haigler, 1 (USNM). Additional records —COLORADO (Finley, 1958:318, unless otherwise noted): Bent County: Fort Lyon. Elbert County: 8 mi NE Agate; Cedar Point, 6 mi NW Limon. Fl Paso County: 7 mi SSE Colorado Springs, 5900 ft; 10 mi S Colorado Springs; 16 mi W Wigwam (Armstrong, 1972). Kit Carson County: South Fork Republican River, near Flagler (Cary, 1911:114). Lincoln County: Big Sandy Creek, near Hugo (Cary, 1911:115). Pueblo County: N of Pinon (Warren, 1942:209); Chico Basin, 20 mi N Pueblo (ibid.); Pueblo. Yuma County: Dry Willow Creek, Boyce Ranch. NEBRASKA (Jones, 1964:219): Chase County: 5 mi S Imperial. Dawson County: 10 mi S Gothenburg. Frontier County: vicinity Curtis. Hays County: 0.5 mi S Hamlet. Lin- coln County: North Platte; sec. 10, T. 11 N, R. 27 W (5 mi S, 2.5 mi W Brady). Red Willow County: McCook. Distribution and habitat—Locality records for Neotoma floridana campestris are shown in figures 2, 3, and 4. Several localities listed by Finley (1958:318) and one reported by Jones (1964:219) are based on observations of dens by collectors and laymen. Of the undocu- mented reports I have traced, I accept the following: 5 mi S Imperial, Chase Co., Nebraska (Jones, 1964:215); near Flagler, Kit Carson Co., Colorado, and near Hugo, Lincoln Co., Colorado (Cary, 1911:114, 115); Chico Basin, 20 mi N Pueblo, Pueblo Co., Colorado, and N of Pinon, Pueblo Co., Colorado (Warren, 1942:209). I have disregarded the follow- ing records pending their documentation by specimens: 6 mi N, 12 mi W Pueblo, Pueblo Co., Colorado—this is a den rec- ord reported by Finley (1958:318) and, although the den most likely was con- structed by N. f. campestris, it is con- ceivable that both N. mexicana (disre- garded by Finley as constructing dens unlike the one he observed) and N. albigula occur at this locality; 10 mi N Arlington, Kiowa Co., Colorado—Cary (1911:114) entered this record on the strength of reports by “stockmen” that a few woodrats occurred in the area, which, if true, probably represented N. f. campestris but could have been N. micropus; along the Arkansas River, south of Chivington, in Prowers Co., Colorado—this is another “stockmen” re- port cited by Cary (loc. cit.), but because micropus now is known on the north side of the Arkansas River both east and west of this locality, it is more likely that the observed woodrat dens were those of that species; Arkansas River bottom, near Holly, Prowers Co., Colorado—originally reported by Warren (1910:112), this rec- ord of woodrat dens clearly should be removed from localities included in the distribution of campestris because micro- pus has been collected approximately six miles east in adjacent Kansas (Coolidge) and the general habitat near Holly is more like that of micropus than that of campestris. One other locality of record for campestris is worthy of comment. Finley (1958:318) discussed a specimen repre- sented only by a skull with incomplete data (USNM 6301) and Armstrong (1972) reported another (USNM 6320) of N. floridana from Denver, Colorado, both collected by E. Palmer. Both au- thors agree, as do I, that Denver is cer- tainly not within the present distribu- tional range of campestris and, although the identity of the skulls is not in ques- tion, they probably are from some lo- cality(ies) to the east of Denver. Considering only the records ac- cepted, the distribution of campestris in Colorado extends from just north of the Arkansas River (Fort Lyon, 3 mi N Fowler, and Olney Springs) to the foot- hills of the Rockies (several localities MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY 16 .-—- soe Ss oo FN ( eee 100 Miles Fic. 3. Selected locality records for Neotoma floridana campestris (symbols solid below) and N. micropus canescens (solid symbols) in Colorado. N. f. Circles rep- > ri (symbols solid above), and N. micropus canescens (solid symbols) in Kansas. resent records accompanied by specific locality data; squares denote records specific only to county. Fic. 4. Selected locality records for Neotoma floridana campestris (symbols solid below) attwate WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA Ly south and southeast of Colorado Springs), thence northeastward at least as far as Wray and in Nebraska as far as North Platte. In Kansas, the race occurs east to the line (discussed above) between Russell and Ellsworth counties, where campestris intergrades with attwateri. I have not attempted to determine the northern limit of range of campestris through field investigations, but Ander- sen and Fleharty (1967:39) did not find woodrats in Jewell County, Kansas, and neither Jones (1964) nor Choate and Genoways (1967) reported finding them in southern Nebraska east of Red Willow County. The first specimens to be asso- ciated with the name campestris in Ne- braska were reported by Jones (1954: 485), although Goldman (1910:25) earlier listed a specimen from Haigler under the name baileyi. The southern extension of the species in western Kansas seems to correlate well with the southern extent of the Ogallala limestone formations north of the Arkan- sas River. I have searched for woodrats throughout the area between the Arkan- sas River and the southernmost locality records for campestris. Specimens of campestris from the Pawnee River in northern Finney County are from only about 20 miles north of a locality (just north of the Arkansas River) where I have collected Neotoma micropus. The distributional status of floridana and mi- cropus in those areas where the two might come into contact is discussed in the account of N. m. canescens. A detailed and extensive ecological account of N. f. campestris in Colorado is given by Finley (1958:499-514). He found that this rat is one of the most versatile species of the genus in utilizing available materials, rocks, tree cactus, trees, or bushes for the construction of dens. Type of vegetation available is apparently of little importance, with the exception that unsculptured shortgrass prairie is insufficient for the fulfillment of denning requirements. In Kansas and Nebraska the situation is undoubtedly much the same. In Dundy County, Ne- braska, I have seen houses of campestris constructed in sagebrush in a manner that rendered them indistinguishable from houses of micropus in southwestern Kansas. At several localities in Kansas (e.g. 19 mi S Dighton, Finney County, and 4 mi S and 0.5 mi W Jetmore, Hodge- man County) I have collected this wood- rat from rock outcrops in otherwise open grasslands. In many cases the nearest outcrops or river systems were several miles distant. At two localities (16 mi N Hays, Ellis County, and 6 mi S and 4 mi E Russell, Russell County ) large pop- ulations of campestris were observed in eastern red cedar windbreaks not far from major rivers (the Saline River and the Smoky Hill River, respectively). Most nests in the windbreaks were on the ground at the bases of trees, partially shielded by overhanging boughs. A few nests, however, were two to about 10 feet above the ground, supported solely by the trees. Along the Pawnee River in Finney County and the Smoky Hill River in Logan County, woodrats were living in piles of flood debris. In some cases the debris served as a skeleton upon which the rats had amassed large superstruc- tures, but in others the only outward signs of the presence of woodrats were fecal droppings and faint runways. Throughout the range of N. f. cam- pestris, the distributional pattern is one of apparent disjunctness, with most pop- ulations almost certainly at least semi- isolated. Such a pattern probably re- sults from inability of these woodrats to permanently occupy shortgrass prairie between river systems, rock outcrops, and stands of trees. A study of long- range dispersal habits of the race would be enlightening in revealing how iso- lated populations are founded and how much (if any) interpopulational gene flow takes place (see Wiley, 1971). Neotoma floridana attwateri Mearns Neotoma attwateri Mearns, 1897:721 [Holotype —USNM 11964/10402 from Lacey’s Ranch, Turtle Creek, Kerr Co., Texas]. [Neotoma floridana] attwateri—Elliot, 1901: ee 18 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY 1939:5 Okesa, Neotoma floridana osagensis Blair, [Holotype—UNMZ 76070 from Osage Co., Oklahoma]. Remarks.—Relegation of the name Neotoma floridana osagensis to the syn- onymy of N. f. attwateri extends the dis- tribution of attwateri north to Kansas and Missouri. Total geographic varia- tion within the subspecies is increased as a result of this change. As discussed be- yond, the pattern of variation northward through Texas and in southern Oklahoma is clearly clinal, and lacking in abrupt changes that might indicate either re- stricted gene flow or secondary inter- gradation. I have not examined speci- mens from the area between Hill and Robertson counties, Texas, but according to Strecker (1924:16 and 1929:220) woodrats are common in that area, at least along the Brazos River, Tehuacana Creek, and White Rock Creek. These reports indicate that although docu- mentary specimens may be lacking, woodrats are more or less continuously distributed throughout the area in which the ranges of attwateri and osagensis were alleged to meet; therefore, no rea- son exists to suspect reduced gene flow. The pattern of variation at the zone of contact between attwateri and campes- tris is discussed in the remarks of the previous account. On the east, the range of attwateri meets that of N. f. illinoensis in Missouri and Arkansas and that of N. f. rubida in southeastern Texas. I have only superficially studied woodrats from Missouri and Arkansas, and cannot com- ment in detail on the relationships of illinoensis to attwateri. These two races seemingly resemble each other to a greater degree than either resembles rubida, but detailed study might not support this supposition. I have exam- ined specimens of rubida from Texas and found woodrats assignable to that name relatively distinct from those represent- ing attwateri. One specimen from Harris County, Texas (SFA 2312), appears to be an intergrade between the two sub- species, but herein is assigned to attwa- teri on the basis of its comparatively small size and the absence of reddish coloration typical of rubida. Records of occurrence.—Specimens examined (680).—KANSAS: Allen County: 2 mi N, 0.5 mi W Neosho River Bridge, Humbolt, 2 (KU). Anderson County: 3.7 mi S Garnett, 1 (KU); 4.5 mi NNE Welda, 1 (KU). Chase County: 9 mi E Lincolnville, 1 (KU); 1.5 mi S Safford- ville, 1 (KSTC). Chautauqua County: Cedar Vale, 8 (USNM); 1 mi N, 2.5 mi W Elgin, 1 (KU). Cowley County: 3.75 mi S, 1.5 mi W Udall, 1 (KU); 6 mi N, 12 mi E Arkansas City, 29 (KU); 3 mi W Cedar Vale, 1 (KU); 86 mi E Arkansas City, 2 (KU); 3 mi SE Ar- kansas City, 2 (KU). Crawford County: Mul- berry, 1 (KU). Douglas County: 7 mi N Lawrence, 1 (KU); 5 mi N, 2 mi W Law- rence, 1 (KU); 6 mi NW Lawrence, 1 (KU); 1 mi N Midland, 2 (KU); 1 mi NW Midland, 5 (KU); 1 mi W Midland, 2 (KU); within 3 mi radius of Lawrence, 17 (KU); 1 mi N, 5 mi W Lawrence, 1 (KU); 7 mi W Lawrence, 2 (SM); 8 mi SW Lawrence, 1 (KU); 8.5 mi SW Lawrence, 2 (KU); 10 mi SW Lawrence, 3 (KU); Lone Star Lake, 6 (KU); 9 mi S} 9 mi W Lawrence, 1 (KU); 10 mi S, 9 mi W Lawrence, 1 (KU); 2 mi S, 2 mi W Pleasant Grove, 1 (KU); unspecified, 1 (KU). Elk County: 1.12 mi S, 1.75 mi W Moline, 1 (KU). Ellsworth County: 3 mi S Wilson, 2 (KU); 3.5 mi SE Ellsworth, 8 (KU); 5 mi SW Ellsworth, 3 (KU). Geary County: Fort Riley, 2 (USNM). Greenwood County: within 5 mi radius of Hamilton, 32 (KU); 15 mi W Hamil- ton, 9 (1 AMNH, 8 KU); 12 mi W Hamilton, 1 (AMNH); 4 mi S, 17 mi W Hamilton, 3 (1 AMNH, 2 KU); 4 mi S, 14 mi W Hamilton, 1 (KU); 12 mi SW Hamilton, 4 (KU); 8 mi SW Toronto, 7 (KU); 8.5 mi SW Toronto, 17 (KU); unspecified, 1 (AMNH). Jackson County: 6 mi S, 10 mi W Holton, 1 (KU). Jefferson County: 13 mi NE Lawrence, 2 (KU). Leavenworth County: 5 mi NE Lawrence, 1 (KSTC); unspecified, 10 (KU). Lyon County: Ross Natural History Reservation, 2 (KSTC); Emporia, 1 (KSTC); unspecified, 1 (KSTC). Marshall County: 1 mi W Vermillion, 2 (KU). Montgomery County: 17 mi NNE Sedan, 1 (KU). Morris County: 4.12 mi S, 6 mi W Council Grove, 1 (KU). Rice County: 2 mi N, 2 mi E Little River, 1 (KSTC). Riley County: 3.25 mi S, 2 mi E Randolph, 1 (MHP); Manhattan, 9 (4 AMNH, 5 USNM). Shawnee County: 1 mi N, 1 mi W Big Springs, 1 (KU). Wabaunsee County: 1 mi N Alma, 4 (KU). Wilson County: 2 mi N, 3 mi E Benedict, 1 (KSTC). Woodson County: State Lake, 2 (KSTC). OKLAHOMA: Adair County: 5 mi SE Flint, 1 (OSU); 3 mi NNE Chewey, 5 (SM); 7 mi W Stilwell, 1 (KU); Stilwell, 8 (USNM). Blaine County: Canton Public Hunting Area and Lake—vicinity of Longdale, 17 (1 FWCM, WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 19 11 OSU, 5 USNM); Salt Creek Canyon, 2 (KU); 2 mi N, 9 mi W Okeene, 1 (KSTC); 2 mi N, 2 mi W Okeene, 1 (KSTC); Roman Nose State Park, 9 (OSU); 2 mi W Watonga, 3 (OSU). Bryan County: 10 mi SE Benning- ton, 1 (TNHC); 5 mi N Colbert, 2 (TNHC); 0.55 mi NW Colbert, 3 (TNHC). Caddo County: 5 mi W Cogar, 3 (SM); Fort Cobb, 2 (USNM). Canadian County: Methodist Church Camp, sec. 18, T. 11 N, R. 10 W, 1 (SM); 5.5 mi SE Hinton, 1500 ft, 1 (KU). Carter County: 3 mi N Springer, 3 (USNM). Cherokee County: 2.5 mi NW Chewey, 2 (SM). Choctaw County: 1.5 mi N Hugo, 1 (OSU); 1 mi N, 3 mi E Hugo, 1 (OSU). Cleveland County: 9 mi N, 2 mi E Norman, 2 (SM); within 5 mi radius of Norman, 18 (6 SM. 1 USNM, 11 KU); 1 mi N, 9 mi E Norman, 1 (SM); 0.5 mi N, 6 mi E Norman, 1 (SM); 6 mi E Norman, 1 (SM); 8 mi E Norman, 1 (SM); 9 mi E Norman, 1 (SM); 3 mi NE Noble, 1 (SM); 1 mi N, 3 mi E Lexington, 1 (OSU); 4 mi SE Lexington, 1 (OSU); unspecified, 2 (SM). Comanche County: Wichita Mountains Wildlife Refuge, 94 (10 OSU, 6 SM, 8 USNM); 19 mi NW Cache, 1 (SM); Mt. Scott, 7 (USNM); 9 mi NW Cache, 1 (SM); Chattanooga, 1 (USNM); unspecified, 1 (SM). Cotton County: 5 mi SE Taylor, 3 (SM). Creek County: Sapulpa, 4 (KU). Dewey County: 2 mi N, 6 mi W Long- dale, 4 (KSTC); NE corner Canton Public Hunting Area, 1 (OSU); 2 mi S, 2 mi W Seiling, 1 (KU); 6 mi S, 2 mi W Seiling, 2 (KU); 5 mi W Canton, 2 (KU); 6.5 mi S, 3 mi W Seiling, 2 (KU); 7 mi S, 2.5 mi W Seiling, 3 (KU); 8 mi S, 5 mi W Seiling, 2 (KU). Garfield County: 1 mi S, 2 mi E Enid, 1 (OSU). Haskell County: 8.5 mi S Stigler, 1 (SM). Johnston County: Tishomingo Na- tional Wildlife Refuge, 1 (USNM). Kay County: 1 mi S, 7 mi E Ponca City, 1 (OSU); Ponca Agency, 1 (USNM). Latimer County: 5 mi N Wilburton, 1 (KU); 3.5 mi N Wilbur- ton, 8 (SM); Red Oak, 1 (USNM); Wilburton, 2 (OSU). Le Flore County: 5 mi S Wister, 1 (SM); 2 mi NE Zoe, 2 (SM); 0.75 mi N Zoe, 4 (SM). Lincoln County: 3.5 mi S Perkins, 1 (OSU); 5 mi W Stroud, 1 (OSU). Major County: 15 mi S Waynoka (see remarks in account of N. m. canescens), 1 (OSU); 1.5 mi N, 0.25 mi W Cleo Springs, 8 (KSTC); 5 mi S, 2.5 mi E Cleo Springs, 2 (KSTC); 3 mi S Chester [=1.5 mi N Seiling] (see remarks in account of N. m. canescens), 16 (8 KSTC, 8 KU); 3 mi S, 0.5 mi E Chester, 4 (KU); 3 mi S, 1 mi E Chester, 1 (KU). Marshall County: 6 mi N Willis, 1 (SM); 5 mi S, 1 mi W Shay, 1 (OSU); 0.5 mi E Willis, 1 (SM); 2 mi E Willis, 4 (MWU); 1 mi S, 2 mi W Willis, 1 (SM); University of Oklahoma Biolog- ical Station, including Engineering Tract and Paul’s Landing, 19 (2 OSU, 17 SM); unspec- ified, 4 (1 MWU, 2 OSU, 1 SM). Mayes County: 1 mi S Spavinaw, 6 (KU). McCurtain County: 2 mi N Smithville, 1 (SM); 2.5 mi W Smithville, 3 (SM); 2 mi W Smithville, 2 (SM); Beavers Bend State Park, 4 (2 SM, 2 KU); 15 mi SE Broken Bow, 1 (SM). Mur- ray County: Sulphur, 1 (OSU); 1.5 mi S Dougherty, 1 (MWU); unspecified, 1 (OSU). Muskogee County: 4 mi below [=SW] Fort Gibson Dam, 1 (OSU); 6 mi SE Fort Gibson, 1 (OSU). Okmulgee County: 3 mi S Okmul- gee, 1 (OSU). Osage County: 10 mi NE Pawhuska, 1 (TNHC); Osage Hills State Park, 6 (OSU); 10 mi WSW Fairfax, 3 (SM); McClintock Boy Scout Camp, 1 (USNM); Heartwood Mountain, 1 (OSU). Pawnee County: 7.5 mi N, 2.75 mi W Pawnee, 1 (KU). Payne County: 2 mi N, 15 mi W Stillwater, 1 (OSU); 1 mi N, 9 mi W Stillwater, 2 (OSU); vicinity of Lake Carl Blackwell, 36 (34 OSU, 2 USNM); 11 mi W Stillwater, 1 (OSU); 10 mi W Stillwater, 8 (2 OSU, 6 TT); 8 mi W Stillwater, 1 (OSU); 5.5 mi W Stillwater, 1 (OSU); 4 mi W Stillwater, 1 (OSU); Stillwater, 1 (OSU); 2 mi E Stillwater, 1 (OSU); 4 mi E Stillwater, 1 (OSU); 5 mi E Stillwater, 1 (OSU); 1 mi S, 3 mi W Stillwater, 2 (OSU); 4.25 mi SW Stillwater, 1 (OSU); 2.5 mi S, 0.25 mi W Stillwater, 1 (OSU); 1 mi S, 3 mi W Mehan, 1 (OSU); 10.5 mi S Stillwater, 2 (OSU); unspecified, 7 (OSU). Pittsburg County: 4 mi NW McAlester, 1 (OSU); 2 mi E McAlester, 1 (OSU); Savanna, 1 (USNM). Pontotoc County: 4 mi S Ada, 1 (OSU); un- specified, 2 (OSU). Pottawatomie County: 1 mi W Pink, 3 (SM); 7 mi SE Tecumseh, 2 (KU). Pushmataha County: 4 mi SE Clayton, 1 (SM); 1 mi S Nashoba, 4 (TNHC). Rogers County: 0.5 mi E Chelsea, 1 (OSU). Stephens County: unspecified, 1 (SM). Tulsa County: 8 mi W Red Fork, 2 (USNM); Red Fork, 2 (USNM); unspecified, 1 (SM). Woodward County: unspecified (see remarks in account of N. m. canescens), 2 (SM). TEXAS: Brazos County: 3 mi W Bryan, 1 (TCWC); 7 mi W College Station, 1 (TCWC); within 5 mi radius of College Sta- tion, 19 (TCWC); 7 mi SW College Station, 1 (TCWC); 10 mi SE College Station, 1 (TCWC). Collins County: 1 mi NE Wylie, 2 (MWU). Colorado County: Eagle Lake, 1 (TCWC). Cooke County: Marysville, 2 (USNM); 2 mi S Marysville, 2 (TCWC); 8 mi W Gainsville, 2 (MWU); Gainsville, 1 (USNM); unspecified, 2 (USNM). Gonzales County: Palmetto State Park, 1 (TNHC). Gregg County: 3.2 mi E Gladewater, 1 (TT). Grimes County: Navasota, 3 (USNM). Harris County: 22 mi N Houston, 1 (SFA). Hender- son County: 2 mi NE Malakoff, 1 (SFA); 12 mi SE Athens, 1 (SFA). Hill County: 4 mi N Blum, 2 (KU). Jack County: 7 mi SE Jacksboro, 1 (MWU). Kaufman County: 19 mi SE Dallas, 2 (TT). Kerr County: Lacey’s Ranch, Turtle Creek, 11 (5 AMNH, 1 TNHC, 20 5 USNM); Ingram, 9 (USNM). Lawaca County: 3 mi W Hallettsville, 1 (TCWC); 0.5 mi W Hallettsville, 1 (TCWC); 14 mi WSW Hallettsville, 1 (TCWC). Montague County: 2 mi W Nocona, 1 (MWU); 3 mi E Nocona, 1 (MWU); 4 mi E Stoneburg, 1 (MWU). Navarro County: Barry, 1 (SFA). Parker County: 8.9 mi S Aledo, 13 (FWCM). Rob- ertson County: 2 mi W Hearne, 2 (TCWC). Tarrant County: Lake Worth Area, 1 (FWCM); Fort Worth, 1 (FWCM); 6.5 mi S, 4 mi W Benbrook, 2 (FWCM). Victoria County: Victoria, 1 (USNM); unspecified, 1 (USNM). Williamson County: 3 mi N Mc- Neil, 1 (TNHC). Additional records: 1956): Dickinson County: (Pl. 9). Greenwood County: 7 mi E Eureka (Pl. 2, Fig. 2). Lyon County: 6 mi N Madison (Pl. 3, Fig. 1). Marshall County (p. 634): 2 mi S Marysville; 5.5 mi S Beattie. Riley County: 7 mi S Manhattan (p. 551). OKLAHOMA (Blair, 1939b:124, unless otherwise noted): Adair County: 5 mi S Kansas. Bryan County: 5 mi SW Colbert (Mc- Carley, 1952:108). Cleveland County: Noble. McClain County: 7 mi SW Norman (Hays, 1958:235, 238). Murray County: Dougherty. Osage County: Conway Springs; Okesa (Blair, 1939a:7); 2 mi SW Okesa. Rogers County (Blair, 1939a:7): 3 mi W Catoosa; Garnett. TEXAS: Cooke County (Russell, 1953: 461): 13 mi NE Gainsville; 7 mi N Gains- ville; 4 mi NNE Myra; 4 mi NE Rosston; 3 mi NE Leo. Hunt County: 5 mi N Greenville (Baker, 1942:343). McLennan County: vi- cinity of Waco (Strecker, 1929:220). Distribution and habitat.—Figures 4, KANSAS (Rainey, 15 mi E Talmage MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY 5, and 6 show localities of occurrence of Neotoma floridana attwateri in Kansas, Oklahoma, and Texas, respectively. This subspecies also is known from adjacent parts of Missouri and Arkansas. The dis- tribution of attwateri extends from north- central and northeastern Kansas and cen- tral Missouri southward through Okla- homa, western Arkansas, and eastern Texas to the Gulf of Mexico. In extreme southeastern Texas it is replaced by the larger, reddish-colored N. f. rubida. Goldman (1910:26) listed two speci- mens as attwateri from the Edwards Plateau at Rocksprings, Kerr Co., Texas. I examined two specimens (USNM 117552-53) from 7 mi S Rocksprings that were collected in 1902; I do not doubt that they are the same specimens studied by Goldman. The identity of these and other woodrats, and resultant ramifica- tions as regards distribution of several species will be discussed elsewhere; it is only necessary to indicate here that both specimens are best referred to Neotoma albigula. Neotoma floridana, therefore, is not known to have occurred on the Edwards Plateau in Recent times (see Dalquest et al., 1969:250). The south- westernmost locality of record for the species is Ingram, Kerr County. mee or MILES © 10 20 30 40 Fic. 5. Selected locality records for Neotoma floridana attwateri (symbols solid above) and N. micropus canescens (solid symbols) in Oklahoma. locality of sympatric occurrence of the two species; for explanation of symbols see figure 4. The encircled symbol denotes the single known Distributional relationships of N. f. attwateri at the western limit of its range, where it abuts the range of N. m. canes- cens, will be discussed in the account of the latter. Neotoma floridana attwateri has not been found in Nebraska; factors possibly limiting its range to the north are discussed by Rainey (1956:632-637 ) and Jones (1964:218). The literature is replete with ecolog- ical accounts of N. f. attwateri in Kansas and northern Oklahoma but the habits of this rat in southern Oklahoma and Texas are not well known. Rainey (1956: 549), in one of the better ecological studies of the species, summarized the habitat in eastern Kansas as_ follows: “habitat of the woodrat in eastern Kansas is divisible into two principal types; the osage orange, Maclura pomifera (Raf.), hedge-row habitat type, which is the more widespread, and the rock outcrop type of habitat. Stone fences, upland woods, wooded stream-courses, shrubby hillsides, and uninhabited buildings con- stitute habitat types of less importance.” Fitch and Rainey (1956) contributed significantly to a general understanding of the ecology of woodrats in eastern Kansas, but found them only in the habitats reported by Rainey (loc. cit.). My observations of attwateri in eastern Kansas also fit the pattern outlined by Rainey. In Ellsworth County, this wood- rat was trapped in two markedly differ- ent habitats. At a locality 3.5 mi SE Ellsworth, woodrats were living in crev- ices and around large boulders of a steep rocky hillside just south of the Smoky Hill River, but 5.5 mi SW Ells- worth I found a sparsely distributed, and apparently small, population of wood- rats occupying what appeared to be mar- ginal habitat along the banks of a dry creek. Most houses were in small piles of flood debris, but one small, active nest was on the ground partially concealed by the dead stems of annual weeds in the corner of a recently harvested milo field. A few scattered trees near the creek may have afforded some cover and building materials, but no dens were WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 21 observed directly associated with the trees: Murphy (1952:205) concluded that woodrats in north-central Oklahoma pre- fer postoak-blackjack oak ravines and fringed forest ravines. Goertz (1970:96- 98) studied attwateri in the same area and found them in various types of woodlands, brushy areas, riparian asso- ciations along stream banks, and to a lesser extent in small rock outcrops in relatively open prairie. Spencer (1968: 38) reported collecting attwateri in a mesquite-prickly pear cactus association in gypsum rock outcrops near the west- ern edge of the range in Oklahoma. A colony of woodrats in thick wood- land along the banks of a sandy gully in Brazos County, Texas, was studied by Lay and Baker (1938). Rats variously occupied surface dens in the wooded area, burrows and root crevices in the gully bank, and an underground burrow in a nearby pasture. Farther north in Texas, Strecker (1929:220) reported at- twateri common along rocky and wooded river banks. Bailey (1905:110) stated that “near Ingram, in the valley of the Guadalupe River, a few of these wood rats were caught in the cliffs and rocks bordering the river valley, but they were more common under the great heaps of driftwood and rubbish along the river bottoms.” As expressed or implied by most of the authors cited and as observed by me, the most important factor of the habitat of Neotoma floridana (including this subspecies and the two previously discussed ) is the availability of cover to- gether with materials and structural ele- ments for construction of dwellings. Neotoma floridana seems to be extremely versatile as regards food requirements and relatively opportunistic in terms of habitat selection so long as minimal re- quirements for cover and nesting ma- terials are available. Neotoma floridana rubida Bangs Neotoma floridana rubida Bangs, 1898:185 [ Holotype—collection of E. A. and O. Bangs to to 36}— 104 100 —=. Ze (eo aa S . a ee aes ee Nae L ee Sapha age Ara | | ee ee jw) a mt 32/— ee. ea —K'4--+§ ar) el. Vaca at Relay foe | ee! : ye hy ns = { oe == wer. ae a 28/- 104 100 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY 96 100 Miles 96 Fic. 6. Selected locality records for Neotoma floridana attwateri (symbols solid above), rubida (symbols solid below), and N. micropus canescens (solid symbols ) tion of symbols see figure 4. 2872 from Louisiana]. Gibson, Terrebonne Parish, Remarks.—The systematic status of Neotoma floridana rubida was not a pri- mary concern of this study, but was con- sidered to determine the eastern extent of the distribution of N. f. attwateri in southern Texas. My findings agree with those of McCarley (1959:411), except that specimens from Nacogdoches and Panola counties (although herein as- signed to rubida) are much less reddish than specimens of rubida from farther south in Texas and those (LSU) exam- ined from Louisiana. This may be an area of intergradation between N. f. rubida and N. f. attwateri or N. f. illi- N. f. in Texas. For explana- noensis (or among all three). Kelson (1952:236) referred a specimen, pre- viously assigned to rubida (Goldman, 1910:23), from Texarkana, Bowie County, to N. f. illinoensis, but indicated it resembled N. f. osagensis (= N. f. attwateri) in certain characters. I have not seen the specimen in question and therefore follow Kelson. Goldman (1910: 26) assigned two specimens from Kountze, Hardin County, to attwateri and 22 specimens from nearby Sourlake, Hardin County, to rubida (p. 23). That portion of Texas was mapped by him as within the range of rubida (p. 21). I regard the specimens from Kountze as rubida on geographic grounds. Records of occurrence.—Specimens exam- ined (25).—TEXAS: Anderson County: 5.5 mi SE Slocum, 1 (SFA). Angelina County: Diboll, 1 (SFA). Cherokee County: 0.5 mi N Forest, 1 (SFA); 3 mi W Forest, 1 (SFA). Nacogdoches County: 1 mi N Nacogdoches, 2 (TT); Nacogdoches, 3 (SFA); 5 mi S Nacog- doches, 1 (SFA); 8 mi S Nacogdoches, 1 (SFA). Panola County: 2 mi S, 5 mi W Carthage, 1 (SFA). Polk County: 14 mi N Camden, 4 (TCWC); 12 mi W Camden, 1 (TCWC). Trinity County: 1 mi E Trinity, 11 (TCWC). Walker County: 17 mi WNW Huntsville, 1 (TCWC); Huntsville, 2 (TCWC); 4 mi E Huntsville, 1 (TCWC); 2 mi SW Hunts- ville, 1 (TCWC); 6 mi S Huntsville, 1 (TCWC); 11 mi NW New Waverly, 1 |(TNHC). Additional records: TEXAS: Hardin County: Kountze (Goldman, 1910:26); Sour- lake (Goldman, 1910:23). Distribution and habitat—The dis- tribution of N. f. rubida in Texas is shown ‘in figure 6. I have not studied woodrats of this subspecies in the field, but Mc- ‘Carley (1959:411) indicated that their habits are not markedly different from those of N. f. attwateri in Texas. Davis -(1960:192) reported that in some areas of eastern Texas woodrats live in bur- rows and do not construct surface nests. If true, this would be a departure from the usual habits of N. floridana in north- ern and western portions of the range. Neotoma micropus Three nominal subspecies of Neo- toma micropus are recognized herein; this is a reduction by two in the number heretofore regarded as valid. In com- parison with most other species of wood- rats, the biological attributes (including geographic variation and taxonomy) of micropus have received relatively little attention from mammalogists. As indi- cated previously, I studied micropus throughout its geographic range, but less intensively in New Mexico than in other areas. Recent studies have shown that Neotoma micropus and Neotoma albigula are more closely related than previously was thought. The two species have been found together at many localities and at some of these they apparently hybridize. WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 23 Finley (1958) first reported specimens from Colorado that appeared to be pos- sible hybrids and Anderson (1969) re- searched this problem in Chihuahua and Coahuila. I have examined the material studied by Finley as well as additional material from Texas and New Mexico. Elucidation of the micropus-albigula problem eventually will affect the total understanding of the systematics of the woodrats of both of these species and of Neotoma floridana. In this report, how- ever, I have concentrated on micropus and floridana. The patterns of geographic variation in micropus are such that I considered three alternative schemes with regard to the assignment of subspecific names. Ir- respective of the alternatives, it seems clear that the woodrats from White Sands, New Mexico, are only slightly more pallid ecomorphs of populations of the species in adjacent areas and do not warrant recognition at the subspecific level. However, one of the alternatives discussed and rejected below would re- quire use of the name leucophea for all populations of the species in western Texas, New Mexico and adjacent Chi- huahua, Coahuila, and Nuevo Leon. The first alternative was to leave ex- isting names (with the exception of leucophea) unchanged and simply de- scribe geographic variation within that system. The only advantage seemed to be that of nomenclatural stability. Rec- ognition of a darker eastern subspecies (micropus) and a generally paler western subspecies (canescens) would obscure some trends in variation in color and in external and cranial sizes. This alterna- tive also would result in continued rec- ognition of the name N. m. littoralis, even though woodrats previously as- signed to that name differ appreciably from specimens of N. m. micropus from farther north in Tamaulipas only in be- ing somewhat more brownish. The second alternative involved rec- ognition of seven subspecies to denote each general kind of variant seen. Neotoma micropus littoralis and N. m. 24 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY planiceps would continue to be recog- nized. Neotoma micropus micropus would be restricted to the small, dark, coastal woodrats of central and northern Tamaulipas. A new name would be pro- posed for the large, intermediate-colored woodrats of the Texas Coast and adjacent inland areas. The available name N. m. surberi would be applied to the large, dark woodrats of south-central Kansas and western Oklahoma (not including the panhandle of that state). The name canescens would be restricted to the large, pale-colored woodrats of the Okla- homa Panhandle, southwestern Kansas, and southeastern Colorado; and_ the name leucophea would be applied to the small, pallid woodrats of New Mexico, western Texas, and México, exclusive of coastal Tamaulipas and the vicinity of Rio Verde, San Luis Potosi. The third alternative, and the one adopted, involves provisional retention of the name N. m. planiceps (pending ac- quisition of specimens that elucidate the distributional and morphological rela- tionships of this woodrat, which pres- ently is known only by the holotype). The name N. m. littoralis is placed in the synonymy of N. m. micropus, a subspe- cies of small, dark-colored (often brown- ish) woodrats of coastal Tamaulipas. All other populations of the species, includ- ing those previously known as N. m. micropus from localities other than coastal Tamaulipas, are referred to a single subspecies, N. m. canescens. Be- cause variation both in size and color is gradually clinal throughout the distribu- tion of canescens, no clearcut areas of demarcation separate local populations. However, to recognize only a_ single taxon results in assignment of woodrats that are quite different (especially from the extremes of the clines) to the same subspecies. Total geographic variation within the subspecies N. m. canescens as thus conceived exceeds that in any of the other races considered in this study. Neotoma micropus canescens J. A. Allen Neotoma micropus canescens J. A. Allen, 1891: 285 [Lectotype—AMNH _ 3030/2350 from North Beaver (=North Canadian River), Indian Territory (Cimarron Co., Okla- homa), near the boundary line between the Indian Territory and New Mexico]. Neotoma macropus [sic] surberi Elliot, 1899: 279 [Holotype—Field Mus. Nat. Hist. 6755 from 3 mi W Alva, Oklahoma Territory (Woods Co., Oklahoma) J. Neotoma micropus leucophea Goldman, 1933: 472 [Holotype—USNM 251057 from White Sands, 10 mi W Point of Sands, White Sands National Monument, Otero Co., New Mexico]. Remarks.—Contfusion exists concern- ing the proper lectotype of Neotoma micropus canescens. Although I would have resolved this question differently, I follow Finley (1958:310-312) in the above synonymy to avoid belaboring a controversial point. Considerations ger- mane to the taxonomic status of this sub- species are discussed above, including nomenclatorial alternatives that might be used to reflect patterns of geographic variation within the samples of woodrats here treated as N. m. canescens. Records of occurrence-—Specimens exam- ined (1102) —COLORADO: Baca County: 14 mi N, 4 mi E Springfield (Two Buttes Reser- voir), 2 (KU); 5 mi S, 2 mi W Pritchett, 22 (KU); 2 mi N, 7 mi W Regnier, 4 (KU). Bent County: 2 mi S, 2 mi E Hasty, 11 (KU). Las Animas County: 11 mi N, 8 mi E Branson, 5600 ft, 4 (KU). Prowers County: 16 mi N, 1 mi E Springfield, 1 (KU); 1 mi N Two Buttes Reservoir, 4350 ft, 1 (KU). KANSAS: Barber County: 5 mi S Sun City, 1 (KU); 15 mi W Medicine Lodge, 1 (KSTC); 10 mi W Medicine Lodge, 16 (KU); 8 mi SW Medicine Lodge, 2 (KU); 6 mi N Aetna, 1 (KU); 7 mi N, 7 mi W Kiowa, 2 (KSC); 7 mi N, 6 mi W Kiowa, 1 (KSC); 1 mi N Aetna, 1 (KSC); 1 mi SW Aetna, 1 (KU); 3 mis Aetna. I (KU) SlosmieSeAcinawl (KU); Marty Ranch, 5 (KU); unspecified, 2 (SM). Clark County: 7 mi S Kingsdown, 3 (KU); 11 mi S, 1 mi W Kingsdown, 3 (KU). Comanche County: 7 mi S Coldwater, 3 (KSTC); 11.5 mi S, 16 mi E Coldwater, 6 (KU); Cave Creek, 1 (KU). Finney County: 1 mi S Pierceville, 2 (KU); 15 mi S, 4 mi W Garden City, 1 (KU). Gray County: 2 mi NW Ingalls, 1 (KU). Hamilton County: Coolidge, 1 (KU); State Lake, 5 (2 MHP, 3 KU); 2.5 mi N Syracuse, 1 (KU); 1 mi N, 3 mi W Syracuse, 4 (KSTC); 1 mi S, 6 mi W Syracuse, 1 (MHP). Haskell County: 2 mi S, 4 mi W Satanta, 30 (KU); 3 mi S31 mi IW. Satantay CK) 5 sanieSS) 4 mini WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 25 Satanta, 8 (KU). Kiowa County: 5 mi N Belvidere, 31 (17 AMNH, 14 KU); Rezeau Ranch, N of Belvidere, 1 (KU). Meade County: 0.5 mi S, 4 mi E Fowler, 7 (KU); 11.5 mi E Meade, 2 (KU); 11 mi SW Meade, 2 (AMNH); vicinity of State Park (12-17 mi SW Meade), 67 (21 AMNH, 3 MHP, 43 KU); unspecified, 1 (KU). Morton County: 9 mi N, 3 mi E Elkhart, 13 (KU); 8 mi N Elkhart, 3 (KSTC); 7 mi N, 2 mi W Elkhart, 2 (KU); 7 mi N Elkhart, 2 (MHP); unspecified, 4 (3 KU, 1 UNAM). Seward County: unspecified, 1 (KSTC). Stanton County: 1 mi N, 8 mi W Manter, 5 (KU); 1 mi N, 7.5 mi W Manter, 29 (KU); 1 mi N, 6 mi W Manter, 2 (KU); 3 mi S, 14 mi W Johnson, 7 (MHP). Stevens County: 4 mi E Moscow, 4 (KU). | NEW MEXICO: Eddy County: 3.25 mi NE Carlsbad, 2 (LSU); 24 mi E Carlsbad, 3500 ft, 1 (KU); 5 mi SW Carlsbad, 1 (KU); 2 mi NE Black River Village, 1 (KU); Carls- bad Cavern, 1 (KU); Rattlesnake Spring, 30 mi SW Carlsbad, 1 (KU). Hidalgo County: 6 mi SSE Lordsburg, 4200 ft, 1 (KU). Luna County: 3 mi N Deming, 4300 ft, 2 (KU). Otero County: 13 mi W Tularosa, 1 (TNHC); 3 mi SW Alamogordo, 2 (TNHC); 3 mi S Alamogordo, 3 (TNHC); 10 mi W Point of Sands, White Sands National Monument, 3 (USNM). San Miguel County: 1 mi S, 2 mi W Conchas Dam, 2 (KU). Santa Fe County: 0.5 mi NW Santa Fe Municipal Airport, 1 (KU); 1 mi W Santa Fe Municipal Airport, 1 (KU); 8 mi SW Santa Fe, 4 (KU); Santa Fe Field Station, 1 (KU). OKLAHOMA: Beaver County: 21 mi S | Meade, Kansas, 1 (KU); 8 mi NE Gate, 1 (KU); 1.5 mi NE Beaver, 2 (KU); 3 mi NE Slapout, 1 (SM). Blaine County: Canton Reservoir (see remarks below), 1 (OSU). Cimarron County: Regnier, 4375 ft, 1 (KU); 3 mi SE Regnier, 4350 ft, 1 (KU); 6 mi N Kenton, 3 (OSU); 4 mi N Kenton, 1 (OSU); 1 mi S, 2 mi E Kenton, 1 (SM); 3 mi S, 2 mi E Kenton, 1 (OSU); 7.5 mi S, 10 mi W Boise City, 3 (KU); 9 mi W Griggs, 3900 ft, 1 (KU). Greer County: Granite, 1 (OSU); 5 mi NE Reed, 1700 ft, 1 (SM); 1 mi S, 1 mi W Reed, 1 (SM); 10 mi SE Mangrum, 1 (MWU). Harmon County: 3 mi W Reed, 6 (SM); 1 mi S, 6 mi E Vinson, 1700 ft, 9 (SM); 1 mi S, 2 mi W Madge, 1 (SM); 6.5 mi SE Vinson, 1 (SM); 13 mi N Hollis, 1 (OSU); 11 mi N Hollis, 2 (FWCM); 6 mi N, 2 mi W Hollis, 2 (OSU); 6 mi N Hollis, 13 (FWCM); 5.5 mi S Hollis, 5 (FWCM). Harper County: Beaver River, Southern Great Plains Experiment Range, 15 (OSU); 3.4 mi N Fort Supply, 2 (USNM); 3 mi N Fort Supply, 43 (USNM). Jackson County: 14 mi S Olustree, 4 (TNHC). Kiowa County: 2 mi S Lugert, 1 (OSU). Major County: 5.5 mi S Waynoka, 2 (SM); 6 mi S, 3 mi E Waynoka, 1 (OSU); 16 mi NW Orienta, 1 (OSU); 1 mi N, 7 mi W Orienta, 3 (KSTC); 16 mi W Orienta, 8 (OSU); 5 mi W Orienta, 4 (2 SM, 2 USNM); 3 mi W Orienta, 1 (OSU); 3 mi N, 9 mi W Fairview, 1 (OSU); 3 mi S Chester [=1.5 mi N Seiling]—(see remarks below), of (Se KSEE 29 KU) ea Siem San Oto) mi) aE Chester, 2 (KU); unspecified, 2 (OSU). Roger Mills County: 7 mi N Cheyenne, 2000 ft, 2 (SM). Texas County: 5.5 mi N Guymon, 3100 ft, 1 (SM); 4.5 mi N Guymon, 3100 ft, 3 (SM); 2 mi N Guymon, 3000 ft, 3 (SM). Tillman County: 5.5 mi S Grandfield, 1 (SM). Woods County: Alva, 7 (USNM); 2 mi W Edith, 6 (5 SM, 1 USNM); 4 mi S, 12 mi W Alva, 3 (KU); Waynoka Dunes, 1 (OSU); 3 mi S Waynoka, 2 (SM); 5 mi S Waynoka, 1400 ft, 6 (SM). Woodward County: Alabaster Cav- erns, 4 (3 OSU, 1 SM); Boiling Springs State Park, 6 (1 KSTC, 5 SM); 2 mi NNW Wood- ward, 1900 ft, 4 (SM); Woodward, 3 (USNM); unspecified, 3 (SM). TEXAS: Andrews County: 10 mi NW Andrews, 1 (TCWC); 14 mi S Andrews, 1 (OSU). Aransas County: Aransas National Wildlife Refuge, 3 (TCWC); 46 mi NE Rockport, 3 (TNHC); 6 mi W Rockport, 1 (TNHC); Rockport, 1 (TNHC). Archer County: 20 mi N Archer City, 1 (MWU); 6 mi W Holliday, 1 (MWU); within 5 mi radius of Holliday, 4 (MWU); within 4 mi radius of Mankins, 17 (MWU); 7 mi SW Wichita Falls, 1 (MWU); 6 mi S Wichita Falls, 1 (MWU); 11 mi SW Wichita Falls, 1 (MWU),; vicinity of Lake Kickavoo, 9 (MWU); 1 mi W Scotland, 2 (MWU); 12 mi N Archer City, 1 (MWU); 9 mi N Archer City, 1 (MWU); 14 mi S Holliday, 2 (MWU); 4 mi W Archer City, 1 (MWU); 4 mi S Archer City, 1 (MWU); 5 mi S Archer City, 1 (MWU); 7 mi S Archer City, 1 (MWU); 7 mi NE Olney, 1 (MWU). Atascosa County: 7 mi SE Lytle, 1 (TNHC); 7 mi SW Somerset, 1 (TNHC); 8 mi SW Somerset, 1 (TNHC); 12 mi W Floresville, 1 (TNHC). Baylor County: 12 mi NW Seymour, 1 (MWU); Bomarton, 1 (MWU). Bee County: 12.5 mi N Beeville, 2 (TNHC). Bexar County: San Antonio, 2 (TNHC); unspecified, 1 (KU). Borden County: 16 mi W Gail, 1 (MWU). Brewster County: 11 mi N Alpine, 1 (MWU); 2 mi W Alpine, 2 (AMNH_); 7.4 mi S Marathon, 1 (AMNH);5 miS Terlingua, 2 (KU); Tornillo Creek, 12 mi N Government Springs, 2700 ft, 1 (AMNH); Government Springs, 3950 ft, Chisos Mountains, 2 (AMNH); East base Chisos Mountains, 2 (USNM); Burnham Ranch, 3950 ft, 2 (AMNH). Callahan County: 30 mi SE Abilene, 1 (SFA). Cameron County: 10 mi E Rio Hondo, 4 (LSU); 8 mi NW Bay- side, 1 (LSU); 2 mi W Port Isabel, 1 (TNHC); Brownsville, 5 (2 KU, 1 TNHC, 2 USNM); 14.7 mi E Brownsville, 5 (KU). Childress County: 18 mi N Childress, 1 (MWU); 15 mi N Childress, 1 (MWU); 5 mi N Childress, 26 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY 1 (INHC); 5 mi S Childress, 1 (TNHC). Clay County: 7 mi SE Wichita Falls, 2 (MWU); 2 mi NE Bellevue, 1 (MWU). Cottle County: 13 mi N Paducah, 3 (TNHC). Crane County: 20 mi NNW Crane, 1 (MWU). Crosby County: Home Creek Canyon, 1 (AMNH); unspecified, 2 (AMNH). Culberson County: 25 mi N Van Horn, 1 (MWU); 16 mi E Van Horn, 5 (TCWC); 16 mi SE Van Horn, 8 (TCWC). Dawson County: 10 mi E Lamesa, 4 (TNHC); 12 mi NW Patricia, 1 (TNHC). Dickens County: 7 mi NE Dickens, 1 (MWU); 5 mi NW Spur, 2 (MWU). Eastland County: 9 mi S Ranger, 2 (TNHC). El Paso County: 15 mi N El Paso, 2 (USNM); 10 mi N El Paso, 7 (USNM); East El Paso, 1 (USNM); near E] Paso, 3 (USNM). Fisher County: 12 mi E Hermleigh, 6 (TNHC). Floyd County: 6 mi S, 2 mi W Quitaque, 1 (OSU). Foard County: 1 mi N, 12 mi E Crowell, 1 (MWU). Frio County: 2 mi N Dilley, 1 (TNHC). Garza County: 4 mi W Post, 1 (OSU). Goliad County: 3.5 mi N Goliad, 2 (TCWC). Hans- ford County: 6 mi S, 3 mi W Gruver, 1 (KU); 6 mi S, 2 mi W Gruwver, 1 (KU). Hardeman County: 3 mi N Quanah, 1 (MWU); 3 mi SE Lazare, 1 (MWU); 7 mi SW OQuanah, 2 CMW) 13:55 mi S (uanahy a ((MWU):. Hartley County: Romero, 5 (AMNH). Haskell County: 6 mi N, 11 mi E Haskell, 3 (MWU); 12 mi SW Haskell, 1 (MWU). Hemphill County: 6 mi E Canadian, 4 (TCWC); 9 mi NNE Miami, 1 (MWU). Hidalgo County: 4 mi WSW Hargill, 1 (LSU); 17 mi NW Edin- burg, 3 (TNHC); Alamo, 1 (LSU); 5 mi S Mission, 1 (LSU); 6 mi S McAllen, 7 (TNHC). Howard County: 7 mi E Vealmoor, 2 (TNHC); Big Spring, 1 (USNM). Hudspeth County: Fort Hancock, 2 (1 AMNH, 1 USNM); W slope Sierra Diablo, 1 (FWCM). Hutchinson County: 1 mi S, 10 mi E Pringle, 2 (KU); 9 mi E Stinnett, 14 (TNHC). Jeff Davis County: 7 mi NW Toyahvale, 2 (MWU); 16 mi NE Fort Davis, 3 (TCWC); Mouth of Madera Canyon, 1 (TCWC). Jim Hogg County: 20 mi S Hebbronville, 9 (TNHC). Jim Wells County: Alice, 1 (LSU). Karnes County: 2 mi SW Kenedy, 2 (TNHC). Knox County: 4 mi SE Vera, 1 (MWU); 5 mi NW Knox City, 1 (MWU). La Salle County: 2 mi S Wood- ward, 1 (TCWC); 8 mi NE Los Angeles, 5 (TCWC); 3 mi NE Los Angeles, 1 (TCWC); § mi W Cotulla, 2 (TCWC); 25 mi E Cotulla, 1 (KU); 8 mi E Encinal, 3 (TCWC). Lynn County: 2 mi W Tahoka, 1 (MWU). Martin County: Stanton, 4 (USNM). Maverick County: Eagle Pass, 1 (TCWC). McMullen County: 21 mi W Three Rivers, 3 (TNHC); 20 mi W Three Rivers, 1 (TNHC); 10 mi W Simmons, 2 (TNHC); 18 mi SE Tilden, 1 (LSU); 21 mi SW Three Rivers, 1 (TNHC). Medina County: 7 mi N Castorville, 3 (KU). Midland County: 9 mi S Stanton, 1 (TCWC). Mitchell County: Colorado City, 2 (USNM). Montague County: 5 mi S_ Ringgold, 1 (MWU); 3 mi N Stoneburg, 1 (MWU); 2 mi N Stoneburg, 1 (MWU). Motley County: 6 mi N Flomot, 2 (MWU). Nueces County: }) 1 mi S Bishop, 2 (TNHC). Palo Pinto County: | Brazos, 1 (USNM). Presidio County: 7 mi W Valentine, 7 (TNHC); 1.5 mi SE Buford Well, Miller Ranch, 1 (TNHC); wnspecified, 1 (USNM). Reeves County: 20 mi S Pecos, 4 (KU). Roberts County: 6 mi N Miami, 4 (MWU). San Patricio County: 8 mi NE Sinton, 4 (LSU). Scurry County: 4 mi SW Synder, 1 (MWU); 20 mi NW Colorado City, 1 (USNM). Starr County: Garciaville, 2 (MWU). Taylor County: 6 mi W View, 4 (MWU). Terrell County: Lozier, 1 (USNM). Terry County: 8 mi N Tokio, 1 (TNHC). Throckmorton County: 18 mi SW Throckmor- ton, 2 (TNHC); 20 mi SW Throckmorton [= Lambshead Ranch], 2 (TNHC). Uvalde County: Montell, 2 (KU); 3 mi N Sabinal, 2 (TNHC); 20 mi E Uvalde, 1 (TCWC). Val Verde County: Comstock, 1 (USNM); Del Rio, 2 (USNM). Ward County: 4 mi NW Royalty, 3 (TNHC). Webb County: 45 mi NW Laredo, 10 (KU); 40 mi NW Laredo, 1 (TNHC); 40 mi SW Catarina, 3 (TNHC); 15 mi NE Laredo, 1 (TNHC); Islitas, 10 mi NNW Laredo, 3 (KU). Wichita County: 4 mi SE Electra, 2 (MWU); within 6 mi radius of Iowa Park, 25 (MWU); within 2 mi radius of Wichita Falls, 11 (MWU); 5 mi NNW Wichita Falls, 1 (MWU):; 6 mi E Wichita Falls, 1 (MWU); 1.5 mi N Oliversion Lake, 1 (MWU); 0.5 mi W Lake Wells, 1 (MWU). Wilbarger County: 8 mi S, 2 mi W Vernon, 1 (MWU); 9 mi S Vernon, 1 (MWU); 15 mi S Vernon, 2 (INHC); 7 mi S Harrold, 2 (MWU); unspecified, 2 (MWU). Willacy County: 10 mi NW Raymondville, 1 (TNHC); 28 mi E Raymondville, 3 (1 KU, 2 TCWC). Young County: 7 mi SW Graham, 1 (MWU). Zapata County: 16 mi N San Ygnacio (=1.8 mi from Webb County line on highway 83), 3 (TNHC); 6 mi NW San Ygnacio, 1 (TCWC); 3 mi N Zapata, 1 (TNHC); 3.5 mi NE Zapata, 2 (TNHC); 5 mi E Zapata, 6 (TNHC). Zavala County: 29 mi S Uvalde, 4 (TNHC); 14 mi W Crystal City, 1 (KU); unspecified, 8) (CIN E(C)). CHIHUAHUA: San Isidro, 10 mi SE Zabagoza, 2 (KU); 7 mi W Porvenir, 1 (KU); 3.5 mi ESE Los Lamentos, 420 m, 1 (KU). COAHUILA: 1 mi S, 9 mi W Villa Acuna, 8 (6 KU, 2 UNAM); 10 mi SE Villa Acuna, 1 (TNHC); Canon del Cochino, 3200 ft, 16 mi N, 21 mi E Piedra Blanca, 1 (KU); 11 mi S Hacienda San Miguel, 2200 ft, 1 (KU); 15 mi N, 8 mi W Piedras Negras, 5 (KU); 2 mi S, 11 mi E Nava, 810 ft, 4 (KU); Ciudad Allende, 1 (TNHC); 10 mi SE Guerrero, 7 (TNHC); 29 mi N, 6 mi E Sabinas, 10 (KU); 10 mi E Hacienda La Mariposa, 3000 ft, 1 (KU); Mariposa Ranch, 1 mi E Nacimiento, 27 mi WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 27 NE Ciudad Muzquiz, 1 (TNHC); La Gacha [=La Concha], 1600 ft, 1 (KU); La Lajita, Rancho de Ja Golondrina, 13 mi NE Ciudad Muzquiz, 1 (TNHC); 10 mi N Ciudad Muz- quiz, 2 (TNHC); 2 mi S, 3 mi E San Juan de Sabinas, 1160 ft, 1 (KU); Sabinas, 4 wEeNHE): 1O'mi ESE Sabinas, 2 (KU); 9 mi NW Don Martin, 2 (KU); Don Martin, 800 ft, 2 (KU); Base of Don Martin Dam, 2 (KU); 8 mi N Hermanas, 1500 ft, 2 (KU); Hermanas, 1205 ft, 2 (KU); 1 mi S Hermanas, 1200 ft, 1 (KU); Cuatrociénegas, 1 (TNHC); 5 mi N, 2 mi W Monclova, 1 (KU); 0.5 mi E San Antonio de Jaral, 4400 ft, 1 (UNAM); 3 mi N, 5 mi W La Rosa, 3600 ft, 3 (KU). NUEVO LEON: 15 mi N, 2 mi W Anahuac [=Rodriguez], 1 (KU); 5 mi N, 3 mi W La Gloria, 1 (KU); 5 mi WSW [General] Zuazua, 1 (UNAM); Rancho 14 de Mayo, 1 km E Casa Principal, 1 (UNAM); 7 mi NW Provi- densia, 1 (KU). TAMAULIPAS: 4 mi SW Nuevo Laredo, 900 ft, 14 (KU); 4.5 mi S Nuevo Laredo, 1 (KU). Additional records: COLORADO (Finley, 1958:315, unless otherwise noted): Baca County: Monon, Bear Creek; Furnish Canyon; Craugh Ranch, Cimarron River. Otero County: 18 mi S La Junta. Prowers County: 15 mi S Lamar (Armstrong, 1972). KANSAS: Barber County: Sun _ [City] (Goldman, 1910:28). Grant County: 10 mi S, 8 mi E Ulysses (record of unoccupied wood- rat dwellings, see text beyond in this account). NEW MEXICO (marginal records only): Luna County: 8 mi E Deming (Goldman, 1910:29). Rio Arriba County: Rinconada (Goldman, 1910:29). Valencia County: 8 mi SE Grants ( Hooper, 1941:32). OKLAHOMA (Blair, 1939b:125): Harper County: 4 mi N Laverne. Woods County: 3 mi W Alva; White Horse Spring; Waynoka. TEXAS (Goldman, 1910:28, unless other- wise noted): Bee County: Beeville. Bexar County: Adams. Brewster County: Alpine; Altuda; Marathon. Cameron County: 11 mi E Brownsville (Baker and Mascarello, 1969: 196). Clay County: Henrietta. Concho County: unspecified. Culberson County: Kent. Dimmit County: Blocker Ranch. Duval County: San Diego. Ector County: 9 mi N Odessa (Baker and Mascarello, 1969:196). Garza County: Post (Baker and Mascarello, 1969:196). Hall County: Newlin. Hudspeth County: Sierra Blanca. Jeff Davis County: Valentine. Kinney County: Fort Clark. Lamb County: 3 mi N Fieldton. La Salle County: Cotulla. Lipscomb County: Lipscomb. Lynn County: 2.5 mi S, 1 mi W Tahoka. Mawerick County: Moras Creek; Pinto Creek. Nueces County: Corpus Christi; Nueces Bay. Presidio County: 7 mi NE Marfa, 4900 ft (Blair, 1940:32). Reeves County: Toyah; Toyahvale. Roberts County: Miami. San Patricio County: 7 mi NE Sinton (Raun, 1966:2). Starr County: Roma; Rio Grande City. Taylor County: Tebo. Terrell County: Dryden. Tom Green County: San Angelo. Ward County: Monohans. Webb County: Dos Hermanos; Laredo; Santo Tomas. Wheeler County: Mobeetie; 5.5 mi S, 2.5 mi W Old Mobeetie (Stickel and Stickel, 1948: 292). Wilbarger County: Vernon. Winkler County (Baker and Mascarello, 1969:196): 2 mi N Wink; 8 mi SSE Kermit. CHIHUAHUA: Monument 15, Boundary Line (Anderson, 1969:29). COAHUILA (Goldman, 1910:21, 29, un- less otherwise noted): 3 mi NW Cuatrociénegas (Baker, 1953:253); 7 mi E Las Vacas; Sabinas; Monclova (also see Anderson, 1969:43); Saltillo (probably N. albigula, see Anderson, 1969:43). NUEVO LEON (Goldman, 1910:28, unless otherwise noted): Rodriguez; 70 mi S Nuevo Laredo (Booth, 1957:15); Doctor Cos; 16 mi S China; Allende (Jiménez—G., 1966:187); Linares. TAMAULIPAS (Goldman, 1910:28, unless otherwise noted): Nuevo Laredo; 10 mi S Nuevo Laredo (Booth, 1957:15); Camargo. Distribution and habitat—Locality records for Neotoma micropus canescens are shown in figures 3, 4, 5, 6, and 7. On the west, the range of this primarily Lower Sonoran subspecies corresponds roughly with the foothills of the Rocky Mountain-Sierra Madre Oriental Cordil- lera and associated extensions. The Rio Grande and Canadian rivers have served as corridors into the Rio Grande and Pecos valleys of New Mexico (see Bailey, 1932:171) as far north as Rinconada and possibly as far as the San Jose River Valley near Grants. A single juvenile tentatively identified as this subspecies by Hooper (1941:32) extends the distri- bution to the latter valley. In northern México, N. m. canescens extends into the lower mountainous areas in the natural breaks between various mountain ranges (Baker, 1956:129). In the watershed of the Rio Salado, for example, canescens is known to occur as far west as Cuatro- cienegas. Distributional relationships of N. m. canescens and the geographically contiguous subspecies, N. m. micropus, are discussed beyond in the account of that subspecies. In southeastern Colorado, southwest- em Kansas, western Oklahoma, and east- central Texas, the range of N. micropus 28 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY 105 ee eee | eo. ee e. ey = e@ 30 \ 99 0. 5450 150 [sex © dh bee czeeteae) Miles Fic. 7. Selected locality records in México for Neotoma angustipalata (symbols solid right), N. micropus canescens (solid symbols), N. m. micropus (symbols solid above), and N. m. planiceps (symbol solid below). abuts that of N. floridana. Spencer (1968) reported the single locality of known sympatry, which is on the North Cana- dian River in extreme southwestern Ma- jor County, Oklahoma (Figs. 1 and 5). The distributional relationships of the two species along the zone of potential abutment, with special emphasis on the area of sympatry, is considered below following a review of habitat types used by N. m. canescens. In comparison with Neotoma_ flori- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 29 dana, N. micropus occurs in more xeric habitats, is associated less often with either trees or rocks, and usually occurs in areas marked by a high incidence of one or more species of cactus (genus Opuntia). In western Kansas (Haskell County), I have found canescens com- mon in overgrazed shortgrass pastures, especially where prickly pear cactus (O. polyacantha) is locally abundant. In pas- tures that also have soapweed (Yucca) or sagebrush (Artemisia), large dens of cactus stems and cow dung are con- structed in and around these plants. However, in pastures having abundant cactus and lacking soapweed and sage- brush, dens are built over clumps of cac- tus, and are characterized by an exten- sive underground system of tunnels and usually a low superstructure of cactus and cow dung. In this situation the woodrat dwellings appear to be small dome-shaped protrusions in an otherwise gently rolling sea of buffalo grass and cactus. On August 11, 1968, approxi- mately 30 houses of this type were exam- ined in a pasture 10 mi S and 8 mi E Ulysses in Grant County. The dwellings were in good repair, some even contained green cactus stems and well-formed grass nests, but none was occupied at that time. Farther east in Kansas_ (Barber County ), this woodrat utilizes the crev- ices and caves formed by gypsum out- crops and often constructs dens around trees and brush in wooded draws. In such habitats the dwellings are not un- like those of floridana; in fact, much of the habitat in southern Barber County seems to me more like that normally oc- cupied by floridana than by micropus. In Baca and Prowers counties, Colo- rado, dens of N. m. canescens often are constructed in tree cactus (Opuntia ar- borescens). Finley (1958:494) reports that tree cactus also is the most important food plant of the species in Colorado, but where other species of cactus and Yucca are available, tree cactus is not essential. In New Mexico the habitat of this woodrat is apparently similar to that in Colorado. Bailey (1932:171) stated that these rats “are abundant in open arid valleys where cactus abounds and are usually found associated with either cac- tus or some of the thorny desert shrubs. Their favorite location for a house is in the midst of a bed of large prickly-pears or thorny bushes... where an abundance of cactus can be found for building ma- terial.” Blair (1943a, 1943b) studied canescens in the Tularosa basin near Alamogordo, New Mexico, where it was common in a mesquite association in which cactus apparently was either ab- sent or scarce. Published reports of the habits of N. m. canescens in Oklahoma and northern and western Texas are scarce. Glass (1949:29) found these rats nesting in canyons along with N. mexicana and N. albigula in the Black Mesa region of extreme western Oklahoma. Blair (1939b: 125) stated that “its bulky nests of sticks, liberally augmented by the remains of prickly pear plants, often are built around mesquite or other thorny shrubs.” In Major County, Oklahoma, I have col- lected this rat along the North Canadian River where it comes into contact with Neotoma floridana. This unique area is discussed in more detail below. Blair (1954:252) studied canescens in north- ern Texas and adjacent Oklahoma, and reported that “these rats show a remark- able amount of variation in their ecolog- ical preference . . . at some stations they are found only on rock bluffs; at other stations they live on the level plains and away from rocks.” In the Davis Moun- tains of southwestern Texas, Blair (1940: 32) collected woodrats of this subspecies from nests constructed in the bases of thorny shrubs in shortgrass-yucca, mes- quite-cholla, and shortgrass-mesquite as- sociations. In similar habitat in south- western Texas, Blair and Miller (1949: 18) noted that in some cases old dens of Dipodomys spectabilus were utilized. In one of the more complete ecological studies of Neotoma micropus, Raun (1966) reported the species common in 30 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY a shortgrass-cactus-mesquite association in San Patricio County, Texas. Blair (1952:224) characterized micropus as one of the commonest and widely dis- tributed mammals in the southern part of Texas. In Coahuila, N. m. canescens has been reported (Baker, 1956:285) to avoid rocky areas and densely vegetated arroyos. Houses are constructed most commonly in prickly pear cactus. At one locality (vicinity of Nava) rats lived in oak thickets, and at another (La Rosa) near houses constructed in short vegeta- tion on desert flats. In the summer of 1969, N. m. canes- cens first was collected from north of the Arkansas River in Colorado. This lo- cality, near Hasty in Bent County, is only about 10 miles from the nearest locality record, (Fort Lyon) for N. flori- dana. (The single specimen from Fort Lyon was collected about 80 years ago and the species probably does not occur at that locality today.) Southeastern Colorado, especially in the area of the Arkansas River, is undoubtedly one of the critical areas in which members of the two species are likely to come into contact. If they are in contact at this time or should come together at some future time, study of the two together would be most interesting because sym- patry of N. f. campestris and N. m. canescens is not known now. Neotoma micropus canescens is known from four localities north of the Arkansas River in western Kansas ( Fig. 4), but floridana is known from no nearer than 20 miles to any of these record stations. I have searched the in- tervening areas intensively; most are presently either cultivated or shortgrass patsureland devoid of habitat likely to support either species. Farther east in Kansas, the hiatus between the ranges of the two species widens appreciably and in the south-central portion of that state is roughly 80 miles in width. In the upland areas of much of the hiatus, land utilization is primarily for cultivation of grain and the fields are not habitable for woodrats. Intervening lowland areas, however, are often at least sparsely wooded, planted hedgerows are common but discontinuous, and several small tributaries of the Arkansas River extend riparian communities into the zone. The watershed of the Arkansas River is heavily wooded along most of its course. In September of 1967, I searched for several days, walking long sections of the Arkansas River and many likely-looking hedgerows, but neither woodrats nor any evience of their presence was seen. I am not convinced that this area is devoid of woodrats, but, if present, they are un- common; thus, the chance of the two species occurring together in south-cen- tral Kansas seems negligible. In northern Oklahoma, the haitus be- tween the ranges of the two species nar- rows rapidly and specimens of both spe- cies are known from localities separated by less than three miles along the Cimar- ron River just west of Orienta. South and west of Orienta on the north side of the North Canadian River on either side of U.S. Highway 281, micropus and floridana occur together. This unique area of sympatry was reported by Spen- cer (1968) and subsequently was visited by me in June, 1968, and January, 1969. Specimens from this area in the collec- tion at Kansas State Teachers College (collected by Spencer) are labeled 1.5 mi N Seiling, in Major Co., Oklahoma, whereas those collected by me (all KU) are labeled 3 mi S Chester, Major Co., Oklahoma. West of highway 281, woodrats occur in two distinct areas. Immediately ad- jacent to the river, vegetation is dense and shrubby with occasional trees and fallen trees in an area varying in width from 30 to 80 feet. Superficially this ap- pears to be habitat typical of floridana. A cultivated field approximately an eighth to a quarter of a mile in width separates the river-edge habitat from an area roughly an eighth of a mile wide where yucca, cactus, sparse grass, and scattered large trees grow on semi-stabi- lized sand dunes. This area is super- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 31 ficially like that often inhabited by mi- cropus. East of the highway, the river edge (50 to 100 feet wide) was densely wooded in 1968 with a dense layer of leaf-litter and a canopy that permitted little ground vegetation. Woodrats were not found in this habitat. Bordering the dense trees to the south was a pasture of relatively stable, vegetated sand dunes. Small stands of large trees were scattered throughout the pasture, primarily at the bases of the dunes. Vegetation on the sides and tops of the dunes consisted of grass and cactus. The area extended ap- proximately three-eighths of a mile north and half a mile east. Farther to the east, the density and sizes of trees increased, density of cactus decreased, and the area became more typical of the habitat of floridana. In 1969, the dense woods bor- dering the river east of the highway were uprooted and piled in a huge windrow bordering the sand dunes on the south. At that time the windrow already har- bored several woodrat dens. Of eighteen woodrats collected at this locality in 1965 and 1966 by Spencer (1968), six were identified as N. flori- dana, three as N. micropus, and nine as probable hybrids or intergrades. I have examined 16 of these specimens and identified seven as floridana, three as mi- cropus, and six as being of mixed parent- age. Using my identifications, six flori- dana, one micropus, and four “hybrids” were collected east of the highway and one floridana, two micropus, and two “hybrids” were obtained west of the high- way. Four of seven animals trapped west of the highway in 1968 were identified as micropus and three were considered to be of mixed parentage; of 16 specimens taken at that time from the east side, six were identified as micropus, three as floridana, and seven as “hybrids.” No traps were set west of the highway in 1969, but traps set within half a mile east of the highway yielded four mi- cropus and eight “hybrids.” No wood- rats identified as N. floridana were taken from the place defined by Spencer (1968) as the area of sympatry. How- ever, another series of traps set in an area 100 to 200 yards farther east caught one floridana, one micropus, and four “hybrids.” Although these findings are inconclusive, it appears that the hybrid zone may have shifted eastward at least a quarter of a mile between 1965 and 1969, although remaining about three fourths of a mile in length. The destruc- tion of trees and formation of the wind- row may have significantly altered the ecological balance of the two species in the area of sympatry; if so, the alteration likely will favor N. floridana. A single specimen trapped one mile east of the highway in 1968 was clearly referable to floridana; it showed no characteristics of micropus or of hybrids. Four specimens from Major and Woodward counties are worthy of spe- cial comment with respect to the dis- tributional relationship of the two spe- cies) athe (first COSUR3S89I) )riseauskim and skull that is referable to floridana on the bases of pelage (color) and cranial characters. This specimen bears the following information on the data label: “15 miles south of Waynoka, Okla., southside of Cimmarron [sic] River, Ma- jor Co.” Fifteen miles south of Waynoka is no closer than nine miles to the Cimarron River and all other specimens examined by me from within 15 miles of Waynoka, in any direction, are N. micropus. Be- cause these two species hybridize at the one known locality of sympatry, and be- cause the locality in question here clearly is in error as stated above, I have not plotted this specimen on figure 5; I sus- pect it is from some locality east of Waynoka. Another specimen, OSU 4063, also is not plotted because of probable error in locality data; a skin alone, it is from “Canton Res. Blaine Co., Okla.” Canton Reservoir is approximately 15 miles east along the North Canadian River from the locality of sympatry and hybridization. This rat is gray in color like micropus, but slightly atypical in be- ing buffy on the shoulders and sides. However, the color variation is so slight that if this specimen was not otherwise (oy) bo in question, it readily would be identified as micropus. I have examined several specimens from near the Canton Reser- voir and all seem to be typical representa- tives of floridana. Three explanations seem possible: (1) both species occur at this locality, but floridana is commoner than micropus; (2) this is a hybrid pop- ulation most like floridana, but occa- sional genetic combinations result in in- dividuals colored as in micropus; (3) the locality data are incorrect and the specimen is from some more western locality. I regard the last alternative as the most plausible and the first as the least plausible. Both specimens appar- ently were collected by beginning stu- dents (field catalog numbers of both are below 10). OSU 3891 bears a collection date of 26 October 1958 and OSU 4063 is dated 28 October 1958. Although not prepared by the same student, it is pos- sible that the labels were somehow switched in preparation. The other two noteworthy specimens are SM 4980 and 4981, preserved only as skins. Both spec- imens clearly were prepared some years ago (date of collection not on specimen labels) and are typical dark brown rep- resentatives of floridana. The specimen labels read “Woodward Co., Oklahoma.” The southeastern corner of Woodward County is two miles west of the area of sympatry and all other specimens exam- ined from that county are micropus. If the zone of contact is shifting gradually eastward and has been doing so for many years, it is possible that floridana may have occurred in Woodward County in the not too distant past. It also is pos- sible that these specimens were not from the North Canadian River, but are from some locality elsewhere in Woodward County that once supported or still sup- ports a population of N. floridana. Farther south in Oklahoma, speci- mens of the two species are known from localities at a minimum of 20 miles from each other. The Red River, 5.5 mi S Grandfield, Tillman, Co., Oklahoma, is the locality of capture for SM 3602, a N. m. canescens. This locality is nearly MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY straight south of Chattagnooga, Coman- che County, where N. f. attwateri has been taken. It is west along the Red River from another floridana locality, 5 mi SE Taylor, Cotton County. South of Taylor, in adjacent Clay and Wichita counties, Texas, micropus is known from several localities. Dalquest (1968:19) stated that “the ranges of N. floridana and N. micropus meet in Clay and Montague counties [Texas] but the two species do not inter- breed.” In actuality, floridana is not known from Clay County and micropus is known only from the extreme western edge of Montague County. Neotoma micropus canescens has been collected 2 mi N Stoneburg and N. f. attwateri is known from a locality 4 mi E Stoneburg, a distance of 4.47 miles. Continued field work in northern Texas and along the Red River likely will result in the loca- tion of a zone of contact between the two species, but none presently is known in northern Texas. A specimen of attwateri (MWU 5256) obtained 7 mi SE Jacksboro, Jack County, is from west of known micropus localities both to the north and to the south, but N. m. canescens has not been taken at any nearby localities. Another area in Texas that would be worthy of additional field work is along the Brazos River in Parker and Hood counties. Neotoma micropus canescens is known from the Brazos River just west of Parker County in Palo Pinto County. A series of 13 N. f. attwateri from 8.9 mi S Aledo, Parker County, is suspiciously grayish- brown in color, but cranially members of the series are more or less typical of floridana. Possibly this population con- tains some introgressed genetic material derived from micropus, but more likely the color is the result of adaptation to the local environment on the western edge of the range of the species. There are few records of museum specimens of woodrats from the central portion of Texas. Strecker (1929:220) reported N. f. attwateri from near Waco. West of Waco, the nearest locality of WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 33 record for N. m. canescens is approxi- mately 175 miles distant in Concho County. On a north-south axis there is an apparent hiatus of some 200 miles in the distribution of micropus along the eastern edge of the range of the species in central Texas. The Colorado River and many smaller waterways traverse this area. Insofar as I am aware, there are no physiographic factors that would be expected to prevent woodrats from inhabiting this sizable portion of central Texas, which is surrounded by areas known to support floridana on the east and micropus on the north, west, and south. Museum specimens indicate that the two species are in close proximity in southeastern Texas adjacent to the Gulf Coast. However, no localities of sym- patry are known; the nearest localities are in Victoria County (N. f. attwateri) and Goliad County (N. m. canescens). Neotoma micropus micropus Baird Neotoma micropus Baird, 1855:333 [Lectotype —USNM 1676/554 from Charco Escondido, Tamaulipas]. Neotoma micropus littoralis Goldman, 1905:31 [Holotype—USNM 92952 from Altamira, Tamaulipas]. Remarks.—When Baird named Neo- toma micropus he had specimens from both Charco Escondido, Tamaulipas, and Santa Rosalia (=Ciudad Camargo ), Chi- huahua. Unfortunately no holotype was designated by Baird. Merriam (1894b: 244) pointed out that the specimen from Santa Rosalia is “somewhat aberrant” and that because “the original descrip- tion is based wholly on the Charco Es- condido specimen . . . [it] must be taken as the type of this species.” The speci- men from Santa Rosalia, not seen by me, is assignable on geographic grounds to Neotoma albigula (see Anderson, 1969), but Merriam’s designation of the Charco Escondido specimen as a lectotype firmly reserved the name for the woodrats to which it is applied. The decision to consider the name N. m. littoralis as a junior synonym of N. m. micropus was not an easy one. How- ever, when analyzed by both univariate and multivariate statistics, the samples of woodrats from northern and_ southern Tamaulipas are consistently more alike than either is to any other sample of the species. From north to south in Tamauli- pas, these rats tend to become less gray- ish and more brownish. One extreme in this trend is reached in southern Tamau- lipas to the south of the Sierra de Tamau- lipas; it is the woodrats at this end of the range that formerly were recognized as littoralis. Alvarez (1963) studied mi- cropus in Tamaulipas and chose to rec- ognize two subspecies within the state. He (p. 453) concluded that micropus and littoralis intergraded in the vicinity of Soto la Marina, and assigned speci- mens from that locality to N. m. micropus and those from localities farther south in Tamaulipas to N. m. littoralis. Woodrats throughout the coastal plain of Tamaulipas have relatively longer tails than specimens of micropus from other localities and tend to be some- what brownish in coloration (rather than grayish), especially on the hind legs where the dorsal coloration meets the “white” of the feet. Although specimens of micropus from southern Texas are ap- preciably larger than those from Tamau- lipas and are assignable to N. m. canes- cens, an occasional specimen from the general area of Brownsville resembles N.m. micropus. Woodrats from just south of the Rio Grande near Nuevo Laredo, Tamaulipas, are somewhat intermediate between mi- cropus and canescens but have been as- signed to the latter. Conversely, speci- mens from Matamoros, Tamaulipas, have some characteristics of woodrats from farther north and west, but I have in- cluded these with N. m. micropus. Specimens from the type locality of N. m. micropus, Charco Escondido, prob- ably are intergrades between the small, brownish, long-tailed coastal woodrats and the equally small, but grayish short- tailed woodrats in Nuevo Leon and Coahuila. It is always somewhat incon- 34 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY venient when specimens from a type locality appear to be intergrades. To consider specimens from Charco Escon- dido to be of the same subspecies as those from Nuevo Leén and Coahuila would result in all the woodrats herein referred to as N. m. canescens being ar- ranged as N. m. micropus, and those from coastal Tamaulipas being regarded as N. m. littoralis. Although specimens from Charco Escondido and adjacent localities in northwestern Tamaulipas could be placed about equally well with either the coastal population or with the inland and northern populations, I concluded that they best represent those nearer the coast. Specimens from Tamaulipas north to approximately Reynosa and south to the vicinities of Ciudad Victoria (on the southwest) and Altamira (on the coast) are here assigned to the subspecies N. m. micropus, with the type locality being Charco Escondido, Tamaulipas; the name N. m. littoralis thus becomes a sub- jective junior synonym of N. m. micropus. Records of occurrence.—Specimens exam- ined (66).—TAMAULIPAS: 3 mi SE Reynosa, 1 (KU); 3 mi S Matamoros, 2 (KU); Charco Escondido, 2 (1 UNAM, 1 USNM); 33 mi S Washington Beach, 1 (KU); San Fernando, 180 ft, 3 (KU); 7 km S, 2 km W San Fer- nando, 2 (KU); 12 mi NW San Carlos, 1300 ft, 4 (KU); 9.5 mi SW Padilla, 800 ft, 3 (KU); 3 mi NE Guemes, 1 (KU); 3 mi N Soto la Marina, 3 (KU); Soto la Marina, 500 ft, 13 (12 KU, 1 LSU); 4 mi N La Pesca, 3 (KU); - La Pesca, 2 (KU); 1 mi E La Pesca, 1 (KU); 7 mi NE Ciudad Victoria, 1 (KU); Ciudad Victoria, 6 (KU); Sierra de Tamaulipas, 2 mi S, 10 mi W Piedra, 1200 ft, 6 (KU); Manuel, 1 (AMNH); 6 mi W Altamira, 8 (KU); Altamira, 100 ft, 5 (USNM). Additional records: TAMAULIPAS (Gold- man, 1910:28, unless otherwise noted): Mata- moros; Bagdad; 40 mi S Matamoros (Hooper, 1953:9); Sierra San Carlos [=E] Malato, Tamaulipeca] (Dice, 1937:254); 16 km N Ciudad Victoria (Hsu and Benirschke, 1968); Forlén. Distribution and habitat—The dis- tribution of N. m. micropus is essentially the coastal plain of Tamaulipas, extend- ing north to the Rio Grande River and south to Altamira, Tamaulipas (Fig. 7). Possibly the subspecies occurs in north- ern coastal Veracruz, but specimens from that state are not presently available (see Hall and Dalquest. 1963). Distributional relationships of N. m. micropus and N. m. canescens in western Tamaulipas, eastern Nuevo Leén, and across the lower Rio Grande are discussed in re- marks above. With respect to ecological habits, N. m. micropus probably differs little from N. m. canescens. According to Alvarez (1963:453), the subspecies occurs throughout the Tamaulipas Biotic Proy- ince and is most common in brushy areas. Specimens have been obtained from the beach near La Pesca and in rocky areas on the Sierra de Tamaulipas. Neotoma micropus planiceps Goldman Neotoma micropus planiceps Goldman, 1905:32 [Holotype—USNM 82105 from Rio Verde, San Luis Potosi]. Remarks.—Dalquest (1953:158) re- ported that no specimens of this woodrat were collected during his investigation of the mammals of San Luis Potosi, but did not indicate if specimens were sought near Rio Verde. He suggested (loc. cit.) that the holotype of N. m. planiceps might be “an aberrant specimen, not fully adult, of Neotoma albigula leu- codon.” I have examined the holotype and concur that it is not fully adult but concluded unequivocally that it is not a Neotoma albigula. 1 think it possible that Neotoma angustipalata, discussed in the following account, may be the same taxon as N. m. planiceps; however, spec- imens are not presently available to re- solve this problem. Record of occurrence.—Specimen examined (1)—SAN LUIS POTOSI: Rio Verde, 1 (USNM). Distribution and habitat.——This sub- species is known only from the type lo- cality, which is shown in figure 7. Goldman (1905:32) did not contribute ecological comments in the original de- scription of N. m. planiceps, but presum- ably it is an inhabitant of the plains im- mediately west of the Sierra Madre Oriental. | Neotoma angustipalata Neotoma angustipalata is one of the least well-known members of the genus. The species was described in 1951, long after most species of Neotoma were at least moderately well studied. The dis- tributional relationship of N. angusti- palata and N. micropus might appear to be that of two subspecies; Hooper (1953) and Alvarez (1963) both suggested that angustipalata is probably no more than _a subspecies of micropus. However, re- sults of analyses presented beyond indi- cate that angustipalata is best regarded as a distinct species, albeit in the same _ species-group as floridana and micropus. Neotoma angustiplata Baker _Neotoma angustipalata Baker, 1951:217 [Holo- j type—KU 36976 from 70 km (by highway) S Ciudad Victoria and 6 km W of the (Pan- American) highway (at El Carrizo), Tamaulipas]. Remarks.—As pointed out previously, the systematic affinities of this woodrat _ are poorly known. They have been con- sidered to be with Neotoma mexicana (Baker, 1951:217), N. albigula (Hall, | 1955:329), and N. micropus (Hooper, 1953:10; Alvarez, 1963:453). I suggested in the previous account that N. angusti- palata may be identical to the rat that bears the name N. micropus planiceps. _ Had I synonymized the two, the name _ angustipalata would have been placed as a junior synonym of planiceps, and the latter would have been elevated to spec- ific status. I agree with Hooper and Alvarez that N. angustipalata is much like N. mi- cropus, but have found that it shares a nearly equal number of characters with N. floridana in addition to having some characters unique unto itself. These char- acteristics are treated in detail in the _ discussion of quantitative and qualitative morphological comparisons beyond. The three specimens from San Luis Potosi to which Dalquest (1951:363) gave the name WNeotoma_ ferruginea griseoventer (placed in the species mexi- cana by Hall, 1955:330) were examined WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 35 to determine if this woodrat and N. angustipalata also might represent the same taxon. The type locality of N. mexicana griseoventer is Xilitla, San Luis Potosi; two of the specimens (LSU 3193, 3194) are from the type locality and one (LSU 3191) is from El Salto, San Luis Potosi. The two specimens from Xilitla appear to be referable to N. mexicana, but LSU 3191 is indistinguishable from N. angustipalata and is best assigned to that species. The similarities between N. angustipalata and N. m. griseoventer are many and the two may yet prove to be synonymous. However, in all specimens of N. angustipalata (including LSU 3191), the vomer is solid beyond the leading edge of the palate, whereas all N. mexicana examined by me have a deep notch in the vomer anterior and dorsal to the palate; the vomers of the two specimens from Xilitla are distinctly notched. Specimens of both species have a deep anteroreentrant angle on MI, the character long used to distinguish N. mexicana from other species of Neotoma, but several authors have commented on the variability in depth of this angle both in N. mexicana and N. micropus (see especially Hooper, 1953:10). Records of occurrence.—Specimens exam- ined (12)—TAMAULIPAS: 70 km [by high- way] S Ciudad Victoria, 6 km W [Pan-Ameri- can] highway [at El Carrizo], 2 (KU); 10 km N, 8 km W El Encino, 400 ft, 1 (KU); 12 km S Ciudad Mante, 1 (UNAM); 2 km S Quintero, 250 m, 2 (UNAM); 4 km SSE Quintero, 2 (UNAM). SAN LUIS POTOSI: El Salto, 1 (LSU); 5 mi W El Naranjo, 1 (TT); 30 km W Valles, edge of plateau, 1 (MWU). Additional records——TAMAULIPAS: Ran- cho del Cielo, 1050 m [6 km NW Gomez Farias] (Hooper, 1953:9; Goodwin, 1954:14; Koopman and Martin, 1959:7); Infernillo (= Inferno), 1320 m [7 km W Gomez Farias] (Koopman and Martin, 1959:6); Paraiso, 420 m [13 km SW Gomez Farias] (ibid.); El Pachon (Hooper, loc. cit.; Goodwin, loc. cit.). Distribution and habitat—Booth (1957:15) first reported Neotoma an- gustiplata from San Luis Potosi; reas- signment of a specimen previously as- signed to the species N. mexicana (see remarks) and assignment of TT 9769 36 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY and MWU 3055 to this species further elucidate the known geographic range of angustiplata in the state. All locality records for N. angustiplata are either in or near the eastern slopes of the Sierra Madre Oriental (Fig. 7). Koopman and Miller (1959:2-3) de- scribed the localities from whence their material (owl pellets) probably origi- nated as Tropical Evergreen Forest and Cloud Forest. Goodwin (1954:2) de- scribed Rancho del Cielo as being “on the first ridge of the Sierra Madre Orien- tal at 1150 meters. Humid, oak and sweet gum, cloud forest (humid upper tropical life zone) surrounding the ranch has been thoroughly lumbered since 1952.” The specimens on which the name originally was based were trapped “in rocks and crevices at the base of a small hill in thick vegetation growing in deep humus” (Baker, 1951:218). All speci- mens reported by Hooper (1953:9) were collected in limestone caves. COMPARATIVE MORPHOLOGICAL ANALYSES In view of the ubiquity of woodrats in the United States and México, it is interesting that relatively little attention has been devoted to generic variation in Neotoma as compared with that given other cricetine genera such as Peromyscus (King, 1968). Hooper (1938 and 1940) studied geographic variation in N. fuscipes and N. cinerea. Hoffmeister and de la Torre (1960) assessed variation in N. stephensi, comparing it to N. lepida. The systematics of N. goldmani were considered by Rainey and Baker (1955). Size and physiological attributes of sev- eral species of Neotoma (not including N. floridana or N. micropus) were cor- related with selected environmental fac- tors by Brown and Lee (1969). Geographic variation in six eastern subspecies of Neotoma floridana was studied by Schwartz and Odum (1957), but it has not been assessed in western races of the species or in N. micropus. Cockrum (1952:188) shifted the sub- species boundary between N. m. mi- cropus (herein restricted to coastal Tamaulipas) and N. m. canescens east- ward in Kansas from that proposed by Goldman (1910:27), but Cockrum did not study patterns of variation in mi- cropus outside of Kansas. Baker (1956: 286) regarded all specimens of the spe- cies from Coahuila as N. m. micropus, whereas Goldman (loc. cit.) considered those from western localities in the state as N. m. canescens. Although the above studies either lacked statistical treatment of data or were limited to univariate analyses of morphological characters, | Anderson (1969) employed multivariate statistics in comparisons of Neotoma micropus with N. albigula from Chihuahua and Coahuila. Multivariate statistics have been used widely in studies of members of the genus Canis (Jolicoeur, 1959; Law- rence Bossert, 1967, 1969) and recently have been employed in studies of geo- graphic variation in bats (Smith, 1972), shrews (Choate, 1970), and spiny mice (Genoways, 1971). Geographic variation in western subspecies of N. floridana and all subspecies of N. micropus is consid- ered here by means of a combination of univariate and multivariate analyses. MATERIALS AND METHODS Age determination—Age of speci- mens examined was determined by use of a modification of the scheme devised by Hoffmeister and de la Torre (1960: 479) for Neotoma stephensi. They rec- ognized four age-groups based on degree of eruption of upper molars and subse- quent wear on these teeth. The oldest and youngest categories in my arrange- ment correspond in a general way to those two groups as defined by Hoff- meister and de la Torre. However, pre- liminary calculations indicated that vari- ance of mensural characters of rats in the intermediate groups exceeded that ex- pected. Age groups then were recon- sidered and eight age classes were rec- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 37 ognized as follow: Group I—immature rats in which M3 is not occlusal and often not erupted. Group I].—immature rats in which M3 is occlusal, but with the posterior loph of the tooth still iso- lated. Group III.—rats with the dentine of the posterior loph on M3 continuous with that of the anterior loph and with the labial reentrant angles of M2 and M3 continuing out of view into the alveolus; the proximal termination of the labial reentrant angles of M1 often is visible. Group IV.—rats characterized by visible, proximal terminations of reentrant angles on all upper molars; the reentrant angles of M1 are more than three-fourths as long as the exposed portion of that tooth. Group V.—young adults in which the reentrant angles of M1 are shorter than defined for group IV, but less than half as long as the height of M1. Group VI. adults with the reentrant angles of M1 between a third and a half as long as the height of the tooth. Group VII.— rats with visible reentrant angles on M1 that are less than a third as long as the height of the tooth. Group VIII.—old adults with no visible reentrant angles on M1 but often with short reentrant angles on M2 and M3. For reasons dis- cussed beyond (see variation with age), only specimens of age groups VI, VII, and VIII were included in studies of geographic variation when such speci- mens were available, and males and fe- males were treated separately (see sec- ondary sexual variation). In two _ in- stances it was necessary to include spec- imens of age group V (the holotypes of N. m. planiceps and N. m. leucophea both are males of this age and older in- dividuals were not available). Molars of woodrats that had been reared or maintained in the laboratory were less worn than those of woodrats that had not been in captivity. As a re- sult, aging criteria described above were not applicable in separating laboratory animals into age groups comparable to those of non-laboratory rats. Laboratory specimens known to be more than two years of age frequently were placed in groups IV and V. Results of growth and development studies, which will be pub- lished elsewhere, indicated that labora- tory woodrats essentially had ceased growth by 30 weeks of age; this age then was used as the critical age and only laboratory woodrats more than 30 weeks old were used in comparisons or treated statistically. As discussed beyond, it was found that laboratory woodrats were larger in some measurements than their non-laboratory counterparts from the same localities. Thus, woodrats that had been maintained in the laboratory in ex- cess of one month were not included in studies of geographic variation employ- ing univariate analyses or in multivariate analyses using CLSNT; a few were in- cluded in MULDIS when other speci- mens from critical areas were not avail- able. Pelage variation A Photovolt Photo- electric Reflection meter (Model 610), which yields values that are percentages of reflection of pure white, was employed to quantify color variation of woodrats. Readings made for each of three reflec- tions (red, green, and blue) were taken from the lumbar region of museum spec- imens of age groups VI-VIII character- ized by unworn or relatively unworn pelage. Analyses of adult molts and pelages were made on museum skins. The number and sequence of matura- tional molts were studied on live wood- rats in the laboratory; these data will be included elsewhere in a discussion of growth and development. Qualitative cranial characters.—Three cranial characters (Fig. 10) that have been used as “taxonomic characters” (Finley, 1958:248) were found to be more variable than previously recorded. The anterior palatal spine may be pointed and nonbifurcate or distally bifurcate. The presence and size of the bifurcation was scored from one to five, with “one” denoting absence of the fork. The posterior margin of the hard palate varies from having a relatively well-de- fined medial indentation to having a well-developed projecting medial con- 38 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY vexity. Variation in this character was scored one to eight, with one being as- signed to specimens with the deepest indentation and eight to those with the largest convexity. Size of sphenopalatine vacuities was found to vary from nearly closed in a few individuals to a large opening extending anterior beyond the posterior edge of the hard palate in others; variation in this character was scored one to six, from smallest to largest. None of these characters lends itself to precise measurements without more sophisticated equipment than usually is available. To insure consistency in scor- ing, exemplary skulls were selected and used for comparison in scoring other specimens. Bacular variation —Bacula of selected adult male woodrats in the Museum of Natural History of The University of Kansas were prepared and stained ac- cording to the method described by Lidicker (1960:496), and subsequently removed from phalli. Measurements of bacula were taken to the nearest tenth of a millimeter with the aid of a Wild Heerbrugg Stereomicroscope, graph pa- per, and dial calipers. External and cranial size variation.— External measurements (total length, length of the tail, length of the hind foot, and length of the ear from the notch) were recorded from data on specimen labels. These data were omitted if ob- viously erroneous as recorded. Ten cranial measurements were taken to the nearest tenth of a millimeter. Seven of the measurements were taken as illus- trated and described by Hooper (1952: 9-11); these include greatest length of skull, zygomatic breadth, least interor- bital constriction (interorbital breadth), length of rostrum, breadth of rostrum, alveolar length of maxillary toothrow (length of molar row), and length of palatal bridge (length of palate). The remaining three measurements were taken as follow: condylobasilar length— midline length of the skull from anterior- most extensions of the premaxillae to the posterior surface of the condyles; breadth at mastoids—the distance, perpendicular} to the longitudinal axis of the skull, from the most lateral extension on one mastoid to the same point of the other; and length of nasals—the distance from the anterior edge of the longest nasal to the most posterior extension of either nasal. Selection of samples.—Because there were so few adults of each sex for statis- tical treatment of specimens from indi- vidual localities it was necessary to in- clude those from adjacent localities in pooled samples. Decisions for grouping specific localities and establishing size of geographic areas to include in each sam- ple were based on several criteria. In no case were specimens of different nominal taxa, as recognized at the onset of the study, included together in a single sam- ple. In areas of suspected intergradation or possible contact between species and subspecies, an attempt was made to keep size of geographic areas as small as pos- sible. Whenever practical, localities were grouped so that at least three and pre- ferably no fewer than five adults of each sex were available. Whenever the above criteria could be met and there was no cause to suspect biologically-based rea- sons for doing otherwise, locality group- ings often were made with consideration to political boundaries merely to facili- tate menial tasks such as sorting of orig- inal data. The thirty-two aggregate localities (samples) and their identifying symbols are shown in figure 8 and briefly outlined below. When all available specimens of a species or subspecies are included in a single sample, the geographic area is not described (exact localities were listed previously under specimens examined). Grouped localities of Neotoma floridana were coded with numeric symbols and those of N. micropus and the single sam- ple of N. angustipalata were given alpha symbols. Names given below in paren- theses are those by which the woodrats previously were recognized. Sample 1—Neotoma floridana bai- leyi. Sample 2.—N. f. campestris from all WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 39 localities in Colorado and Nebraska. _ Sample 3.—N. f. campestris from all localities in Kansas west of a north-south line extended from the boundary be- tween Trego and Ellis counties. Sample 4.—N. f. campestris from all localities in Kansas east of the line de- Iscribed for sample 3 and west of a paral- lel line extended from the boundary be- tween Russell and Ellsworth counties. | Sample 5—N. f. attwateri (N. f. osagensis) from all localities in Kansas east of the line described for sample 4 and west of a parallel line extended from Fic. 8. Sketch map of region in which woodrats were studied showing general geo- graphic areas of grouped localities, indicated by identifying code symbols. All samples of Neo- toma floridana (1-13) are represented by nu- meric symbols and those of N. micropus (A-P) and N. angustipalata (R) are represented by alpha symbols. Solid lines separate the distri- butions of species and dashed lines the distribu- tions of subspecies. The solid dot labeled S in Oklahoma marks the single known locality of sympatry between two of the species. See text for precise definitions of the area included in each grouped locality. the boundary between Saline and Dick- inson counties. Sample 6.—N. f. attwateri (N. f. osagensis) from all localities in Kansas east of the line described for sample 5 and north of a perpendicular line ex- tended from the boundary between Lyon and Greenwood counties. Sample 7.—N. f. attwateri (N. f. osagensis) from all localities in Kansas south of the line described for sample 6. Sample 8—N. f. attwateri (N. f. osagensis) from all localities in Okla- homa west of a line extended from the boundary between Major and Garfield counties. Sample 9.—N. f. attwateri (N. f. osagensis) from all localities in Oklahoma east of the line described for sample 8 and north of a perpendicular line ex- tended from the boundary between Lincoln and Pottawatomie counties. Sample 10.—N. f. attwateri (N. f. osagensis) from all localities in Oklahoma east of the line described for sample 8 and south of the line described for sam- ple 9. Sample 11—N. f. attwateri (N. f. osagensis) from all localities in Texas north of a line extended from the boun- dary between Navarro and Limestone counties and west of a perpendicular line extended from the boundary between Harrison and Gregg counties. Sample 12.—N. f. attwateri from all localities in Texas south of the east-west line described for sample 11 and south- west of a line extended from the south- western border of Walker County where it abuts Montgomery County. Sample 13.—N. f. rubida from all lo- calities in Texas. Sample A.—N. micropus canescens from all localities in Colorado, and in Cimarron County, Oklahoma. Sample B.—N. m. canescens from all localities in Kansas northwest of U.S. highway 54, and in Beaver and Texas counties, Oklahoma. Sample C.—N. m. canescens from all localities in Meade and Clark counties, 40 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY Kansas, and in Harper County, Okla- homa. Sample D.—N. m. canescens (N. m. micropus) from all localities in Barber, Comanche, and Kiowa counties, Kansas. Sample E.—N. m. canescens from all localities in New Mexico except the White Sands National Monument. Sample F.—N. m. canescens from all localities in Texas north of a line ex- tended from the boundary between An- derson and Winkler counties and west of a perpendicular line extended from the boundary between Fisher and Curry counties. Sample G.—N. m. canescens (N. m. micropus) from all localities in Oklahoma north of the South Canadian River, ex- clusive of localities in samples A, B, and C. Sample H.—N. m. canescens (N. m. micropus) from all localities in Oklahoma south of the South Canadian River. Sample I—N. m. canescens (N. m. micropus) from all localities in Texas north of the east-west line described for sample F, and east of the north-south line described for that sample. Sample J.—N. m. canescens from all localities in Chihuahua and those in Texas south of the east-west line de- scribed for sample F and west of a line extended from the boundary between Reagan and Irion counties. Sample K.—N. m. canescens (N. m. micropus) from all localities in Texas east of the line described for sample J, south of the east-west line described for sample F, west of a line extended from the boundary between Medina and Bexar counties, and north of the Webb-Zapata County boundary. Sample L.—N. m. canescens (N. m. micropus) from all localities in Texas south of the east-west line described for sample F and the Webb-Zapata County boundary, and east of a north-south line extended south to the southern Webb County boundary from the boundary be- tween Medina and Bexar counties. Sample M.—N. m. canescens (N. m. micropus) from all localities in Coahuila and Nuevo Leon and those in Tamaulipas north of an east-west line passing through Reynosa. Sample N.—N. m. micropus from all localities in Tamaulipas south of the line’ described for sample M and north of 23° 30’ N latitude. Sample O.—N. m. canescens (N. m. leucophea) from White Sands National Monument in New Mexico. Sample P.—N. m. micropus (N. m. littoralis) from all localities in Tamaulipas south of 23° 30’ N latitude. Sample Q.—N. m. planiceps. Sample R.—N. angustipalata. Sample $.—Neotoma_ floridana, N. micropus, and their natural hybrids from 3 mi S Chester, Major Co., Oklahoma. Statistical analyses.—Statistical anal- yses were selected for their appropriate-_ ness, ease of interpretation, and avail- ability at The University of Kansas Com- putation Center. Standard _ statistics (mean, range, standard deviation, stand- ard error of the mean, variance, and coefficient of variation) were calculated, after which group-means were simul- taneously tested for significant differ- ences at the 0.95 level of confidence (0.05 level of significance) by single classifica- tion analysis of variance (univariate ANOVA). If significant variation was present among the group-means and if more than two samples were being com- pared, the Sums of Squares Simultaneous Testing Procedure (SS-STP) described by Gabriel (1964) was employed to de- termine maximal non-significant subsets. Calculations involved in the SS-STP were outlined by Sokal and Rohlf (1969:582), and use of the procedure in studies of geographic variation was considered by Gabriel and Sokal (1969). All of the above calculations were computed by Power's UNIVAR program (Power, 1970). Univariate analyses first were con- ducted to compare males and females by selecting samples of woodrats considered to be adults that were from the same geo- graphic areas (ideally, animals from a single locality would be used, but no suf- ficiently large samples were available). WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 41 When it was clear that significant sexual dimorphism existed, samples of the pre- viously described age groups were com- pared separately within each sex. Re- sults indicated that animals of age groups V and younger frequently were signifi- cantly smaller than those in groups VI or older. Although animals in group VI occasionally were significantly smaller than those of groups VII and VIII, these three age groups were considered to- gether with sexes being treated separ- ately in analysis of geographic variation. So-called Dice-grams (Dice and Leraas, 1936) have been employed frequently to illustrate a general array of variation. For reasons discussed by Sokal and Rinkel (1963), Dice-grams are not ap- propriate for determination of statistical significance when more than two samples are being compared; therefore, all deter- minations of significance or the absence thereof were based on SS-STP tests. Because the sample of specimens from each locality usually exhibited various subset relationships with samples from other localities when different characters were considered, it was necessary to use multivariate analyses to determine rela- tionships based on all characters exam- ined. This was accomplished by means of two programs (CLSNT-Version 2, and MULDIS) available in the Numerical Taxonomy System at the Computation Center of The University of Kansas. CLSNT was used to compare samples of populations for geographic variation by considering the sample of specimens from each aggregate locality as an Oper- ational Taxonomic Unit (OTU) and sample means as characters. When only a single individual was available, as for Neotoma micropus planiceps, the char- acters of that specimen were treated as means. Discriminant function analysis (MULDIS) was employed to compare individuals from various samples and to analyze for hybridization and intergrada- tion. Among other sets of values, CLSNT and optional subroutines as employed by me computed matrices of Pearson’s prod- uct moment correlations and matrices of taxonomic distance coefficients (see Sokal, 1961, and Sokal and Sneath, 1963). Each matrix was then subjected to clus- ter analyses using UPGMA (unweighted pair group method using arithmetic aver- ages), and a two-dimensional phenogram was generated from each. A coefficient of cophenetic correlation (Sokal and Sneath, 1963) was computed to express the reliability of the phenogram based on comparisons with the respective matrices. Moss (1968) discussed rela- tionships of the two phenograms and ex- perimentally studied general types of variation affecting each. I have consid- ered both matrices and both phenograms in all analyses, but because coefficients of cophenetic correlation usually were higher between the distance matrix and phenogram, these have been given greatest consideration. All computations for CLSNT were conducted on stand- ardized data, which was derived by con- verting the mean for each character to zero and the variance and standard devi- ation to one so that the value for each character was expressed in terms of standardized deviates (see Sokal and Rohlf, 1969:109 ). A principal components analysis also was conducted on the among-characters correlation matrix to “condense” or “com- press” the variation in the characters con- sidered into a smaller number of “new” characters, the first few principal com- ponents. At least the first three com- ponents were extracted in all cases and the first five components frequently were considered, especially when the number of characters for each OTU was large. The percentage of the total variation ac- counted for by each principal component also was calculated. OTU’s were pro- jected onto the principal components, and bivariate scatter diagrams were made by plotting projections of OTU’s on all combinations of components. Projection of OTU’s into three-dimensional draw- ings was accomplished with a PROJ-3D program whereby the first three prin- cipal component scores of each OTU A2 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY were transferred to a magnetic tape from which a Benson-Lehner incremental plot- ter made the perspective drawings. The shortest minimally connected network between OTU’s was computed from the matrix of distance coefficients and in- cluded on the three-dimensional models. Discriminant function analysis (MULDIS) employs variance-covariance mathematics to differentially weight each character relative to the variance within and between groups of that character when two reference samples are con- sidered. A discriminant multiplier ( dis- criminant function) was calculated for each variable; then each discriminant multiplier was multiplied by the value of its respective variable and summed for all variables to yield a discriminant score for each OTU. The discriminant scores then were plotted on a frequency histo- gram to compare the individuals of two populations, with or without additional comparisons of a test sample of geo- graphic intermediates or suspected and known (laboratory-bred ) hybrids. NON-GEOGRAPHIC VARIATION In order to undertake a meaningful assessment of geographic variation, it is necessary to understand the non-geo- graphic variation that exists in popula- tions of the organisms to be compared. I have analyzed variation with respect to age, secondary sexual characteristics, individual differences, and effects of hav- ing been held in captivity. Variation in pelages resulting from age and molts was considered and was found to vary sea- sonally. Because seasonal timing of molt and certain characteristics of pelage in woodrats vary geographically, these as- pects are considered beyond with dis- cussion of geographic variation in color. Variation with Age Variation in size correlated with age differences was analyzed for samples of Neotoma floridana campestris (Table 1), N. f. baileyi, N. f. attwateri, and N. mi- cropus canescens (Birney, 1970). Be- cause age-groups and sexes were sepa- rated sample size frequently was small, especially for baileyi. As a result, some age-groups were strikingly different in size, but the differences were not always statistically significant. Dimensions of two external charac- ters (length of hind foot and length of ear) and those of two cranial characters (least interorbital constriction and alve- olar length of the maxillary toothrow) were influenced less by age than were those of other characters. Growth and development studies of laboratory wood- rats demonstrate that the hind foot and ear grow at a disproportionately faster rate than the body and tail. Also, varia- tion in external measurements of museum specimens is high (see discussion of indi- vidual variation) because of inconsis- tencies in techniques used to measure these characters by various collectors. Alveolar length of the maxillary tooth- row does not increase with age after the molars have become occlusal, because individual teeth of woodrats do not in- crease in diameter after eruption. There is a tendency for the molar row to be slightly longer in rats of age-groups III and IV than in older animals. The alve- olar tissue in cleaned skulls of older woodrats usually is separated slightly from the base of the teeth and in some senile rats the molar roots extend to the alveolus. These factors tend to result in smaller measurements, and a decrease in accuracy of the measurement. Least interorbital constriction shows a general increase in size up to age-groups III and IV. Sequence of means varied noticeably in the older age-groups, indicating that the constriction changes little with age after the early period of rapid growth. Dimensions of other cranial charac- ters indicated that animals in age-groups VII and VIII do not differ significantly in size. Although the mean of a sample in age-group VI frequently was less than that of older woodrats, this difference seldom was significant. Highest F, values and the greatest number of non- significant subsets frequently were com- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA TABLE 1. Variation with age in 14 external and cranial measurements of Neotoma floridana campestris. Fs; was calculated by single classification analysis of variance. are at the P 0.05 level of significance; ns indicates no significant difference within a group of means. Tabular F values Nonsignificant subsets (as calculated by the Sums of Squares Simultaneous Testing Procedure ) of significantly different groups of means are shown in the last column. Measurement, sex, and F, age class N Mean se SIR Range CV F SS=Sie Total length Females VIII 15 370.8 8.70 (340.0-402.0) 4.54 22.88 I VI 16 369.6 8.88 (340.0-409.0) 4.81 PPA) | I VII 10 268.7 11.39 (344.0-395.0) 4.88 I V 20 353.5 6.72 (325.0-377.0) 4.95 il IV 14 337.0 8.91 (303.0-365.0) 4.94 ll Ill 13 314.2 1s) (291.0-374.0) Ugaie: Ib AL II 2 284.0 24.00 (272.0-296.0) 5.98 I Males VIII ul 399.7 15.56 (371.0-434.0) 5.15 30.90 I VI 1 382.8 9.84 (350.0-408.0) 4.26 25 I V 117 377.6 13.19 (325.0-42.4.0) 7.20 eel VII 9 373.9 10.30 (341.0-395.0) 4.13 owe IV 13 347.8 12.54 (307.0-383.0) 6.50 Ik 1 Ill 7 302.0 21.00 (288.0-365.0) 8.38 I I 3 264.7 SMILE (232.0-286.0) 10.85 I II 6 264.3 17.82 (240.0-287.0) 8.26 I Length of tail vertebrae Females VI 16 156.6 4.54 (144.0-172.0) 5.80 13.65 I VIII 15 154.5 4.85 (136.0-175.0) 6.07 SH I VII 10 154.5 7.46 (138.0-172.0) meOo Me i V 20 148.4 4.86 (131.0-167.0) Weoo ie IV 14 142.2 St3D (133.0-152.0) 4A] ial Ill 13 130.8 5.48 (115.0-154.0) 7.56 if at Il 2 122.5 25.00 (110.0-135.0) 14.43 I Males VIII 7 164.4 6.38 (@USIRO=175!0) Bale 17.39 I VI 11 161.0 8.33 (130.0-178.0) 8.58 25 I Vv 17 155.9 6.53 (132.0-177.0) 8.64 I VII 9 151.6 9.17 (120.0-164.0) 9.07 ou IV 13 146.8 6.08 (129.0-168.0) TAT if al Ill i IB PATE 9.77 (107.0-147.0) 9.73 Idi I 3 113.0 21.94 (92.0-129.0) 16.81 I II 6 108.5 8.53 (98.0-124.0) 9.63 I Length of hindfoot Females IV 14 39.9 1.10 (36.0-43.0) Salle ee 100 ns VI 16 39.7 0.60 (38.0-42.0) 3.01 2.21 V 20 39.4 0.69 (36.0-41.0) 3.90 VIII 16 39.2 0.71 (36.0-41.0) 3.64 VII 10 39.1 1.28 (36.0-42.0) 5.18 Ill 13 38.8 1ek3 (35.0-41.0) 5.24 II 2, 38.0 4.00 (36.0-40.0) 7.44 43 44 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 1.—Continued. Measurement, sex, and F, age class N Mean =+ 2SE Range CV F SS-STP Males VII 4 40.8 0.88 (39.0-43.0) 3.43 4.84 I VIII 8 40.8 1.50 (36.0-44.0) 6.35 2.14 I V 10 40.6 0.90 (36.0-44.0) 4.73 if VI 5 40.4 1.24 (37.0-44.0) Sell I IV of 40.0 1.01 (35.0-42.0) 4.56 dt III 5 38.9 1.34 (37.0-41.0) 4.56 I I II 5 36.7 B52: (35.0-40.0) 5.08 I I 3 36.7 0.67 (36.0-37.0) Sir I Length of ear Females VIII 8 28.0 1eOT (26.0-30.0) 5.40 1.36 ns VI 12 oS 1.18 (25.0-32.0) sey? 27S) VII 5 Die, 1.47 (25.0-29.0) 6.04 IV 9 272, 1.19 (25.0-30.0) 6.57 V 6 27.0 0.89 (25.0-28.0) 4.06 Ill 9 26.4 eat (25.0-30.0) 6.30 II 2 25.0 4.00 (23.0-27.0) 1ST Males VI 5 29.0 2.28 (26.0-33.0) 8.79 1.70 ns IV 7 28.9 1.60 (26.0-32.0) Us 2.26 VII 8 28.6 2.28 (26.0-33.0) 8.79 VII 4 28.5 0.58 (28.0-29.0) 2.03 V 8 28.5 1.00 (26.0-30.0) 4.96 Ill 5 26.0 1.41 (24.0-28.0) 6.08 Il 5 25.6 1.36 (24.0-28.0) 5.92 I 3 25.0 ISIS) (24.0-26.0) 4.00 Greatest length of skull Females VIII 14 49.9 0.69 (48.3-52.1) sy 24.01 I VI 13 49.4 0.87 (47.0-53.3) 3.18 8) I VII 8 49.3 0.86 (47.5-51.8) 9,45 Jha V 19 47.3 4.19 (45.8-49.5) 1.93 It) dL IV 14 46.7 0.57 (44.4-48.1) 2.28 I III 10 44.9 fay) L (42.5-48.3) roll II 2 41.8 2.90 (40.4-43.3) 4.90 Males VIII 10 51.8 ALS (49.5-55.2) 3.44 32.64 I VII 6 51.6 1.30 (49.0-53.4) 3.08 De lil I V 20 49.8 0.76 (47.3-53.7) 3.40 I VI 9 49.7 1.52 (46.0-52.7) 4.60 eo IV 13 46.8 0.95 (43.3-49.6) 3.66 Pel Ill 4 44.4 2.25 (42.3-47.4) 5.06 I I 1 40.9 oat (40.9-40.9) a I Il 5 39.2 2.19 (37.1-42.7) 6.26 Condylobasilar length Females VIII 15 48.4 0.75 (46.9-51.6) 2.98 36.40 I VII 8 48.3 0.90 (46.4-50.9) 2.63 228 I VI 10 47.3 0.67 (45.2-48.9) D2 eT V 20 45.7 0.45 (44.0-48.2) DON ive i IV 14 44.6 0.57 (42.7-46.3) DBM I Ill 12 42.4 127 (39.9-45.6) 5.19 Il 2 39.3 2.60 (38.0-40.6) 4.68 WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA TABLE 1—Continued. Measurement, sex, and Fy age class N Mean + 2SE Range CV F SS-STP Males VIII 12 50.6 1.07 (48.1-54.3) 3.66 36.44 I VII ll 49.6 1.63 (46.6-52.3) 4.32 OIL I VI 9 48.3 EAN (44.5-52.3) 4.57 I V 20 48.2 0.82 (45.3-52.8) 3.82 I IV 13 44.7 0.95 (41.1-46.8) 3.81 I III 4 42.0 2.94 (39.6-46.1) 6.99 I I 1 38.6 a (38.6-38.6) ets if at II 5 36.7 2.09 (34.8-40.1) 6.38 I Zygomatic breadth Females VIII 14 27.4 0.52 (26.0-29.6) 3.54 34.92 I VII 6 26.8 0.54 (26.0-27.7) 2.46 OF8) I VI 14 26.3 0.41 (25.1-28.0) 2.88 Io V 19 25.6 0.36 (24.5-27 1) 3.09 eT IV 14 24.7 0.33 (23.6-25.9) 9.54. if J Ill 1} DN 0.54 (22.6-25.5) 4.15 I II 2 ODT I5) 0.60 (22,.2-22.8) 1.89 Males VIII 11 28.2 0.51 (27.0-29.9) 2.99 49.56 II VII 7 HT/83 0.72 (26.1-28.5) 3.47 Dy I VI 8 26.7 0.54 (25.4-28.1) 2.85 if I V 19 26.6 0.44 (25.3-28.7) By aM/ I IV 13 24.8 0.49 (23.4-26.4) 3.56 I Ill 6 24.3 1.13 (22.9-26.1) 5.69 I I 3 20.6 1.64 (19.0-21.7) 6.88 I II 5 20.5 1.30 (18.9-22.6) 7.07 I Least interorbital constriction Females VI 15 6.8 0.19 (6.1-7.5) 5.55 01 ns VII 9 6.7 0.17 (6.2-7.0) 3.78 Reon VIII 16 6.6 0.13 (6.2-7.1) 3.92 V 20 6.6 0.12 (6.1-7.1) 4.07 IV 14 6.6 0.12 (6.1-6.9) 3.49 Ill 14 6.5 0.18 (6.0-7.3) Bulli II 2 6.2 0.10 (6.1-6.2) eas Males VII 10 7.0 0.23 (6.5-7.7) 5.19 5.97 I VIII 1b 6.8 0.13 (6.5-7.1) 3.32 2.14 el VI 10 6.8 0.18 (6.4-7.1) 4.20 1 fe ay Dee | V 20 6.7 0.11 (6.2-7.2) 3.60 | ae! Lee) aL Ill 0 6.6 0.20 (6.3-6.9) 3.97 LPS is V1 IV 13 6.5 0.18 (GET) 4.90 Jy toe a Il 6 6.4 0.31 (5.9-6.9) 5.87 eral I 3 6.2 0.2 (6.1-6.4) 2.79 I Breadth at mastoids Females VIII 15 19.4 0.36 (18.3-21.0) 3200 7.96 I VI JUL 19.3 0.40 (18.1-20.2) 3.42 depo} I VII 10 19.2 0.38 (18.1-20.0) Bip 1 Ib ou V 19 18.7 0.20 (17.9-19.7) 2.32 Thy ace I IV 13 18.7 0.20 (17.9-19.2) 1.91 Liet Ill 1 18.3 0.30 (17.7-19.5) 2.82 I II 2 17.9 0.60 (17.6-18.2) ROT I 45 46 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 1.—Continued. Measurement, sex, and F, age class N Mean + 2SE Range GN. F SS-STP Males VIII ill 20.5 0.51 (19.3-22.4) 4.15 16.91 I VII 10 19.6 OSS (18.7-20.2) 2.66 Dall AL V 20 19.6 0.37 (18.3-20.8) 4,20 It lt VI 8 19.1 0.73 (17.4-20.3) 5.43 eel: IV 13 18.7 O27 (17.6-19.6) 2.59 100 Ill 5 18.6 0.36 (18.1-19.0) OR AUTl LoL II 6 17.0 0.48 (16.3-17.7) 3.45 I I 1 16.9 (16.9-16.9) z I Length of rostrum Females VII 10 19.5 0.51 (18.2-21.0) 4,11 19.25 I VIII 15s 19.4 0.29 (18.6-20.5) 2.92 OPAL I VI 16 19.0 0.44 (17.3-21.1) 4.62 1 Gate V 19 18.3 0.26 (17.5-19.8) 3.14 I IV 15 18.1 0.36 (16.7-19.4) 3.84 eel Ill 12 ie, 0.61 (15.9-18.6) 6.11 1 II 2, DES 0.90 (15.3-16.2) 4.04 Males VIII 11 20.5 0.64 (19.4-22.9) 5.19 oil I VII 8 20.4 0.44 (19.3-21.2) 3.08 Asi a! I V 20 19.7 0.43 (18.0-21.7) 4.92 I VI iat 19.2 0.51 (18.0-20.9) 4.38 Teer IV 133 18.2 0.49 (16.4-19.6) 4.88 I Ill 6 ies 0.95 (16.1-18.8) 6.66 i I 3 14.7 1.92 (12.8-15.9) ES. I II 6 14.4 0.94 (13.2-15.6) 7.99 I Breadth of rostrum Females VIII 16 8.5 0.17 (8.0-9.4) 4.10 8.62 I VII 9 8.5 Q:211! (8.0-9.0) 3.64 DADA I at VI 15 8.2 0.12 (7.7-8.5) 2.84 1 ed | Vv 20 8.1 0.15 (iz3=8:0) 4.1] ge da cif IV 15 7.9 1.16 (7.4-8.4) 3.89 Ihe dl Ill 13 7.8 0.23 (7.3-8.5) 5.20 I II 1 7.6 (7.6-7.6) Es I Males VIII 12 8.9 0.14 (8.6-9.3) 2.64 iyeay I VII 9 8.6 0.24 (8.2-9.2) 4.10 2.14 lope | V 20 8.4 0.17 (7.8-9.4) 4.54 iT, ie VI 11 8.2 0.46 (6.1-8.8) 9.35 it IV 13 8.1 0.20 (7.5-9.0) 4.58 Teele ll III 7 Voll 0.23 (7.2-8.1) 3.90 Pes “1 II 6 7.0 0.44 (6.3-7.6) 7.79 ek I 3) 7.0 0.46 (6.6-7.4) thal I Alveolar length of maxillary toothrow Females IV 15 9.9 0.17 (9.3-10.5) 3.31 2.32 I VII 10 9.8 0.17 (9.4-10.3) STIS: DO ie V 20 9.7 0.14 (9.1-10.3) SHS if Ut VI 16 9.7 0.17 (9.3-10.5) 3.49 lel Ill 13 9.7 0.13 (9.4-10.1) 2.41 1b II 9) 9.6 0.50 (9.3-9.8) 3.70 Lael: VIII 16 9.5 0.18 (8.9-10.0) 3.82 I WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA AT TABLE 1.—Concluded. Measurement, sex, and Fy age class N Mean + 2SE Range GY F SS-STP Males V 20 9.9 0.12 (9.2-10.4) 2.64 2.13 ns | IV 13 9.9 0.19 (9.3-10.6) 3.43 2.14 Ill il 9.8 0.19 (9.6-10.3) 2255 VIII 2, 9.7 0.26 (9.0-10.4) 4.68 VII 10 9.7 0.26 (8.9-10.2) 4,22, VI 11 9.6 0.23 (9.1-10.3) 4.05 I 3 9.5 0.18 (9.3-9.6) 1.61 II 6 9.4 0.28 (8.9-9.8) 3.66 Length of palatal bridge Females VII 10 8.6 0.20 (8.0-9.1) OHhe 4.91 I VIII 16 8.5 0.25 (7.8-9.4) 5.96 SHOAL lie VI 16 8.2 On (7.5-8.8) 4.13 Hk Al Vv 20 8.2 0.15 (7.4-8.7) 4.15 I I IV T5 8.1 0.19 (7.3-8.6) 4,44 |e | Ill 13 8.1 0.16 (7.6-8.8) 3.67 I II 2; 7.6 0.10 (7.6-7.7) 0.92 I Males VII 9 9.0 0.44 (7.9-9.8) TAT 14.27 I VIII 12 9.0 0.21 (8.5-9.7) 4.05 2.14 Ti VI 11 8.5 0.25 (7.6-9.0) 4.81 1 Oa gay V 20 8.4 0.23 (7.4-9.5) 6.15 le eel III 6 8.1 0.33 (7.6-8.7) 4.96 ee el IV 13 7.9 0.18 (7.4-8.4) 4.07 1s ent I 2 7.4 0.60 (7.1-7.7) Bo eat II 6 wal 0.41 (6.5-7.9) 7.02 I Length of nasals Females Vill 15 19.4 0.30 (18.7-20.4) 3.03 20.54 I VII 10 19.3 0.36 (18.4-20.4) 2.96 PAPAL I VI 16 18.9 0.44 (17.6-20.3) 4.70 they ae V 19 18.1 0.25 (17.2-19.3) 3.04 ie IV 15) 17.9 0.42 (16.2-19.2) 4,55 I al III 12 16.8 0.61 (15.4-18.5) 6.26 eyl II 2 16.0 2.00 (15.0-17.0) 8.84 I Males VII 8 20.3 0.42 (19.1-21.0) 2.94 29.19 I VIII 11 20.2 0.83 (18.3-23.3) 6.77 2.14 I V 20 19.5 0.43 (18.3-21.5) 4.90 | (ta VI ma 19.0 0.60 (17.7-20.5) p20 leer ey IV 13 18.1 0.52 (16.0-19.5) 5.21 jlo | III 6 17.0 IB? (15.5-19.0) 8.06 IF I 3 14.8 2.07 (12.8-16.2) 12.10 lel II 6 14.4 0.91 (1333-1557) 7.69 I puted for measurements involving di- mensions of the anterior portions of skulls. In studies of Neotoma micropus, Allen (1894a:240) noticed that relative growth of the preorbital region exceeded that of the postorbital area in post- partum development. Hall (1926:396) observed similar relative rates of growth for Spermophilus beecheyi and con- sidered them a common feature of mam- malian development. Therefore, it was expected that measurements such as length of rostrum and length of nasals would be most critical in terms of group- 48 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY ing animals of different ages for other analyses. Animals of age-group V_ usu- ally were significantly different from animals of older groups, thus indicating that specimens of age-group V_ should not be included with older woodrats. Al- though this resulted in reduction of the sizes of available samples, comparisons of smaller, more homogeneous samples are more reliable than those involving heterogeneous samples. Secondary Sexual Variation Although no detailed analytical tests comparing woodrats of different sexes from the same geographic areas are avail- able, males and females generally have been treated separately in studies of geographic variation (Hooper, 1938, 1940; Hoffmeister and de la Torre, 1960). However, Schwartz and Odum (1957) apparently treated both sexes of N. flori- dana in the same samples. Using only specimens of age groups VI, VII, and VIII, males and females were tested by single classification ANOVA to deter- mine if secondary sexual variation was present in external and cranial dimen- sions of N. f. baileyi, N. f. campestris, N. f. attwateri, N. m. micropus, and N. m. canescens (Table 2). In one sample, N. f. campestris, size variation attributable to sex was ob- served in most measurements. Samples of campestris generally included more individuals than other samples. Means for males were larger at the 0.01 level of significance (P < 0.01) in eight of the 14 measurements, and significantly larger (P <0.05) in two of the remaining six. Only in length of ear did the mean for females exceed that for males. In N. m. micropus, the taxon represented by few- est individuals, no significant differences in means were detected, but means of measurements for males were larger than those for females in nine characters. This suggests that non-significant results were a function of the small samples rather than absence of real differences between sexes. Males of baileyi were found to be larger than females in only six measure- ments and the difference was significant only in one (least interorbital constric- tion). Size of females exceeded that of males in three characters, but the dif- ference was significant in none. Means accurate to the nearest tenth of a milli- meter were identical in five dimensions considered. Although larger samples might alter the results, it appears that baileyi has less secondary sexual varia- tion than other taxa considered. Samples of canescens and attwateri were rela- tively large. Seven significant differences (one at the 0.01 level and six at the 0.05 level) apparently resulting from secon- dary sexual variation were observed for attwateri, whereas only four (one at the 0.01 level and three at the 0.05 level) were exhibited by canescens. Significant secondary sexual variation was not demonstrated for length of tail vertebrae, length of ear, breadth of ros- trum, and alveolar length of maxillary toothrow. Total length, condylobasilar length, and length of nasals varied sig- nificantly between the sexes in all of the taxa having samples of more than 10 individuals of each sex. When all taxa and characters are considered, it is seen that sufficient sec- ondary sexual variation exists to discour- age treatment of males and females as a single sample. Therefore, the sexes were treated separately in geographic considerations of mensural data. Spec- imens of both sexes were treated as a single sample in only one set of analyses that included mensural data (discrim- inant function analysis). Because dis- criminant function analysis was used in comparisons of individuals and not in comparisons of sample means, the sex of each individual could be considered when interpreting results. As discussed beyond (comparative reproduction), it was observed in the laboratory that males capable of phys- ically dominating females in breeding cages were the more successful breeders and, conversely, females that were phys- ically subordinate were more successful breeders than large dominant females. Thus secondary sexual variation in size WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 49 TABLE 2. Secondary sexual variation in 14 external and cranial measurements of selected samples of adult Neotoma floridana and N. micropus. Fs was calculated by single classification analysis of variance. Tabular F values are given at the level of significance or at P<0.05 if not significant. One asterisk and two asterices indicate significance at P<0.05 and P<0.01, re- spectively, whereas ns indicates no significant difference. Measurements F and sex N Mean a= PSF Range CV F Sample 1 (Neotoma floridana baileyi) Total length Females 9 374.4 9.87 (350.0-393.0) 3.95 < 1.00 Males ui 381.3 10.04 (361.0-398.0) 3.48 4.60 ns Length of tail vertebrae Females 9 161.7 9.00 (136.0-180.0) 8.35 <1.00 Males it 159.7 10.20 (138.0-176.0) 8.44 4.60 ns Length of hind foot Females iil 39.1 0.51 (38.0-41.0) 2.14 3.00 Males 8 39.8 0.60 (38.0-41.0) Dal2 4.45 ns Length of ear Females 5 26.6 1.20 (25.0-28.0) 5.04 <1.00 Males 2 205 3.00 (26.0-29.0) Tata 6.61 ns Greatest length of skull Females ll 48.8 0.44 (47.5-49.7) 15m! < 1.00 Males 7 48.4 iil (46.5-50.7) 3.03 4.49 ns Condylobasilar length Females fat. 47.4 0.55 (46.0-48.9) 1.92 <1.00 Males i 47.4 Lay, (45.7-49.8) SHOE 4.49 ns Zygomatic breadth Females 11 26.1 0.27 (25.4-26.6) 2; < 1.00 Males of 25.9 0.61 (24.8-27.1) 3.10 4.49 ns Least interorbital constriction Females ia 6.7 0.15 (6.3-7.0) 3.63 6.34 Males 9 6.9 0.17 (6.6-7.4) 3.68 4,41* Breadth at mastoids Females 11 19.0 0.13 (18.2-19.6) 230 <1.00 Males 8 19.0 0.51 (18.0-20.4) 3.79 4,45 ns Length of rostrum Females Tet 18.8 0.32 (17.5-19.4) 2.86 <1.00 Males 8 18.9 0.43 (17.8-19.6) Sep 4.45 ns Breadth of rostrum Females eu 7.9 0.13 (7.5-8.2) 2.76 < 1.00 Males 9 7.9 On (7.5-8.2) 3.26 4.41 ns Alveolar length of maxillary toothrow Females 1 9.4 O17 (8.8-9.9) 3.07 < 1.00 Males 9 9.5 0.20 (9.2-10.0) 3.10 4.41 ns Length of palatal bridge Females ii 8.7 0.34 (7.3-9.2) 6.42 < 1.00 Males 9 8.7 0.33 (8.2-9.6) 5.66 441 ns Length of nasals Females 11 18.7 0.33 (18.0-20.2) 2.96 <1.00 Males 8 18.7 0.43 (17.6-19.4) oe22) 4.45 ns 50 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 2.—Continued. Measurements F, and sex N Mean Se) Os) By Range GV F Samples 2, 3, and 4 (Neotoma floridana campestris) Total length Females 4] 369.8 roo (340.0-409.0) 4.61 10.42 Males 27 384.2 7.47 (341.0-434.0) 5.05 7.04 ** Length of tail vertebrae Females Al 1553 3.03 (136.0-175.0) 6.24 1.48 Males 27 158.7 ball? (120.0-178.0) 8.37 3.99 ns Length of hind foot Females 42, 39.4 0.46 (36.0-42.0) 3.80 9.71 Males 33 40.6 0.72 (36.0-44.0) 5.07 70s Length of ear Females 25 27.8 0.71 (25.0-32.0) 6.35 3.02 Males 17 28.7 0.82 (26.0-33.0) 5.88 4.08 ns Greatest length of skull | Females 5) 49.6 0.46 (47.0-53.3) 2.76 10.02 | Males 25 51.0 0.84 (46.0-55.2) 4.12 Tally Condylobasilar length Females 33 48.1 0.47 (45.2-51.6) 2.82 11.84 Males 28 49.6 0.83 (44.5-54.3) 4.43 Te Zygomatic breadth Females 34 26.8 0.32 (25.1-29.6) 3.53 6.53 Males 26 27.5 0.42 (25.4-29.9) 3.85 4:02) Least interorbital constriction Females 40 6.7 0.09 (6.1-7.5) 4.56 7.44 Males 32 6.9 0.10 (6.4-7.7) 4,29 Ol. = Breadth at mastoids Females 36 19.3 0.22 (18.1-21.0) 3:00 5.96 Males 29 19.8 0.37 (17.4-22.4) 4.97 4:00 = Length of rostrum Females Al 19.3 0.24 (17.3-21.1) 3.98 11.40 Males 30 20.0 0.38 (18.0-22.9) 5.26 1.047" Breadth of rostrum Females 40 8.4 0.10 (7.7-9.4) 3.97 3.35 Males 32 8.6 0.20 (6.1-9.3) 6.77 3.98 ns Alveolar length of maxillary toothrow Females 42, 9.6 0.11 (8.9-10.5) 3.60 < 1.00 Males 33 9.7 0.14 (8.9-10.4) 4.21 3.98 ns Length of palatal bridge Females 42, 8.4 0.13 (7.5-9.4) 5.06 12.48 Males 32 8.8 0.15 (7.6-9.8) 5.83 HAO Length of nasals Females 4] 19.2 0.23 (17.6-20.4) 3.84 7.91 Males 30 19.8 0.44 (17.7-23.3) 6.03 OA = Samples 5, 6, and 7 (Neotoma floridana attwateri) Total length Females 20 364.6 8.89 (329.0-397.0) 5.45 8.28 Males 18 386.9 13.02 (345.0-450.0) 7.14 Moo HS WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA TABLE 2.—Continued. Measurements F, and sex N Mean ae OAT Range CV F Length of tail vertebrae Females 20 1572 4,23 (142.0-170.0) 6.02 < 1.00 Males 18 160.1 4.18 (139.0-175.0) 5.54 4.11 ns Length of hind foot Females 21 38.1 0.96 (34.0-42.0) 5.75 4.51 Males 19 39.4 0.69 (36.0-42.0) 3.81 ANOe Length of ear Females 15 Dill 1.79 (23.0-38.0) 12.76 <1.00 Males 15 26.7 0.93 (25.0-30.0) 6.70 4.20 ns Greatest length of skull Females 21 49.4 0.70 (47.0-52.2) O25 5.49 Males 18 50.7 0.80 (47.4-53.5) 3.36 Ae Condylobasilar length Females 21 48.1 OFT. (45.6-51.7) 3.68 5.90 Males 18 49.6 0.87 (46.1-52.2) 3.74 Asli Zygomatic breadth Females 22 26.9 0.39 (25.6-29.1) 3.42 5.61 Males 17 WiLal 0.55 (25.7-29.2) 4,12 ALY * Least interorbital constriction Females 23 6.5 0.13 (6.1-7.2) 4.81 2.67 Males 20 6.7 0.18 (6.0-7.8) 6.01 4.08 ns Breadth at mastoids Females 23 19.2 0.31 (17.8-20.4) 3.84 5.98 Males 18 19.9 0.49 (16.9-21.0) 5.23 ANNO! = Length of rostrum Females al 19.2 0.32 (18.1-20.6) SRrlT Bl y/ Males 20 19.7 0.43 (17.9-21.5) 4.85 4.10 ns Breadth of rostrum Females 21 8.1 0.11 (7.7-8.5) 3.24 Sok Males 19 8.3 0.21 (7.5-9.1) 5.45 4.10 ns Alveolar length of maxillary toothrow Females 23 9.4 1.54 (8.7-10.1) 3.95 2.63 Males 20 9.6 1.62 (9.0-10.2) 3.78 4.08 ns Length of palatal bridge Females P23) 8.5 1.60 (7.6-9.3) 4.50 SAT Males 20 8.7 1.98 (7.9-9.6) 5.08 4.08 ns Length of nasals Females 20 19.1 0.36 (17.8-21.3) 4.17 6.03 Males 19 19.8 0.37 (18.0-21.6) 4,12 A Samples B and C (Neotoma micropus canescens) Total length Females 31 355.8 5.97 (310.0-382.0) 4.67 7.11 Males 23 370.0 9.46 (334.0-411.0) 6.13 4.03 * Length of tail vertebrae Females 31 147.1 3.70 (130.0-165.0) 7.01 3.22 Males 23 152.6 5.10 (131.0-175.0) 8.01 4.03 ns Length of hind foot Females 30 38.4 0.54 (36.0-41.0) 3.85 1.95 Males 25 39.2 1.01 (35.0-45.0) 6.46 4.03 ns 51 52 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 2.—Continued. Measurements F, and sex N Mean + 2SE Range CV F Length of ear Females 24 wile 0.56 (25.0-30.0) 5.10 <1.00 Males 16 PAfall 0.72 (25.0-29.0) 5.31 4.10 ns Greatest length of skull Females 2G 48.8 0.70 (44.2-51.8) 3.10 1.89 Males 25 49.5 0.63 (46.4-52.9) 3.17 4.03 ns Condylobasilar Jength Females 29 47.0 0.58 (42.8-50.0) 3.34 8.72 Males 24 48.3 0.66 (44.6-50.9) 3.33 Tale Zygomatic breadth Females 30 26.5 0.39 (24.7-29.1) 4.06 < 1.00 Males 26 26.7 0.36 (25.1-28.8) 3.47 4.03 ns Least interorbital constriction Females 32 6.3 0.10 (5.8-7.0) A472 < 1.00 Males OM 6.3 0.11 (5.8-6.9) 4,39 4.02 ns Breadth at mastoids Females 27 19.1 0.23 (17.9-20.3) 3.10 1.86 Males 24 19.3 0.28 (18.0-20.8) 3.58 4.04 ns Length of rostrum Females 30 18.9 0.27 (17.2-20.2) 3.98 (Gls) Males 26 19.4 0.28 (17.8-20.7) 3.67 4035" Breadth of rostrum Females 32 8.3 0.14 (7.2-9.3) 4,92 < 1.00 Males ON 8.3 0.14 (7.5-9.2) 4.41 4.02 ns Alveolar length of maxillary toothrow Females 32 9.4 0.14 (8.5-10.1) 4.35 <1.00 Males 27 9.3 0.12 (8.7-10.1) 3.47 4.02 ns Length of palatal bridge Females 31 7.9 0.19 (7.1-9.5) 6.56 <1.00 Males 26 8.1 0.18 (6.8-8.9) ate 4.02 ns Length of nasals Females 30 19.2 0.35 (16.7-21.2) 5.04 6.03 Males 26 19.8 0.32 (18.0-21.1) 4.11 ALORS Sample P (Neotoma micropus micropus) Total length Females 4 354.5 24.84 (333.0-377.0) 7.01 < 1.00 Males 3 364.3 24.04 (362.0-366.0) b.71 6.61 ns Length of tail vertebrae Females 4 1735 19.77 (155.0-193.0) 11.40 < 1.00 Males 3 169.7 4.06 (166.0-173.0) 2.07 6.61 ns Length of hind foot Females 4 36.5 1.29 (35.0-38.0) 3.54 2.76 Males 3 38.0 1 ss (37.0-39.0) 2.63 6.61 ns Length of ear Females 4 28.2 RO? (25.0-30.0) 7.85 < 1.00 Males 2 28.0 2.00 (27.0-29.0) 5.05 let ms Greatest length of skull Females 4 45.8 1.49 (43.9-47.1) Bab) 1.21 Males 3 46.6 0.42 (46.2-46.9) 0.77 6.61 ns WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 53 TABLE 2.—Concluded. Measurements By and sex N Mean + 2SE Range CV F Condylobasilar length Females 4 43,2 1.44 (41.7-44.5) 3.32 1.24 Males 3 44,2 0.75 (43.6-44.9) 1.47 6.61 ns Zygomatic breadth Females 4 24.0 0.38 (23.5-24.4) 1.57 3.88 Males 3 25.0 WLeIbe (24.4-26.2) 4.04 6.61 ns Least interorbital constriction | Females 4 6.3 0.36 (6.0-6.8) 5.68 1.42 {| Males 3 6.0 0.29 (5.8-6.3) A417 6.61 ns _ Breadth at mastoids | Females 4 18.2 0.48 (17.6-18.6) 2.64 <1.00 Males 3 18.3 0.44 (18.0-18.7) 2.07 6.61 ns Length of rostrum Females 4 17.6 0.91 (16.4-18.5) 5.19 < 1.00 Males 3 17.4 0.47 (17.0-17.8) 3.33 6.61 ns Breadth of rostrum Females 4 7.4 0.25 (7.2-7.8) 3.38 < 1.00 Males 3 led 0.12 (7.4-7.6) 1.33 6.61 ns Alveolar length of maxillary toothrow Females 4 8.9 0.34 (8.6-9.4) 3.82 < 1.00 Males 3 9.1 0.41 (8.8-9.5) 3.84 6.61 ns Length of palatal bridge Females 4 7.8 0.42 (7.5-8.4) 5.39 < 1.00 Males 3 ted 0.81 (6.8-8.2) 9.33 6.61 ns Length of nasals Females 4 eget 0.80 (16.3-18.2) 4.70 2.42 Males 3 17.9 0.35 (17.6-18.2) 1.68 6.61 ns (with males being larger than females ) might convey a selective advantage. However, the laboratory breeding cage was clearly an unnatural situation. In the natural environment, domination of females by males may not be important if females tolerate males only during estrus but are willing to accept any male at that time. Considering the habit of solitary occupancy of dens and competi- tion for den sites during times of high population density (see Fitch and Rainey, 1956:517), females that are large enough to protect choice dens for maternity pur- poses may rear more young than less robust females. On the other hand, most adult male N. micropus collected in west- ern Kansas in June, 1967, had what ap- peared to be fresh wounds in the region of the lower back, but no “battle scars” were noted for females at that time (see Fitch and Rainey, 1956:521, for similar comments pertaining to N. floridana). Perhaps in the natural environment, physical competition and fighting is most common between males, which would cause selection for large size to be more intense in males than females. Brown (1968) and Brown and Lee (1969) studied various physiological and morphological responses of woodrats to differing thermal regimes. It was found that body size was related to tempera- ture, and although no comparisons of sexes were reported, it is clear that selec- tive responses to temperature and other environmental factors play a major role in determination of size of woodrats. Such factors probably act similarly on animals of both sexes, tending to reduce 54 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY secondary sexual variation. The magni- tude of secondary sexual variation prob- ably is controlled by several interacting forces, some of which favor secondary sexual differences while others operate to minimize such differences. The result is as seen in Table 2; perceptible differences exist, but these appear to be more pro- nounced in some taxa (campestris, for example) than in others (such as baileyi). Individual Variation Of more than 2000 woodrats exam- ined during this investigation, only five had obviously atypical coloration that I assumed to be genetically based. Three of these, Neotoma micropus canescens that were previously reported by Baker (1956:286), are from near Sabinas, Coahuila; all were collected on the same day. One is a young adult female that was lactating and the other two are juveniles nearing completion of the post- juvenal molt. Most hairs of the venter are white to the base on all three, and white hairs extend onto the sides and lower rump. The female probably is the mother of the two juveniles. The status of the color pattern in the population from which these specimens originated would be of interest. The other two abnormally colored specimens are Neo- toma floridana attwateri. One (OSU 4541) is a young adult female from northeastern Dewey County, Oklahoma; dorsal coloration is a uniform creamy tan and the ears are nearly white. The other (KU 18682), from Anderson County, Kansas, is an albino in fresh winter pelage. The ears, plantar surfaces of the feet, hair, and underlying skin are de- void of pigment; written on the data label are the words “eyes pink.” A specimen of Neotoma micropus obtained in western Kansas in June 1967 and another from Prowers County, Colo- rado captured in April 1968 were dis- tinctly reddish dorsally, at the time of capture. Both were in old pelage and the one from Colorado was still in reddish pelage when sacrificed about six weeks after capture. The other was molting at the time of capture and eventually com- pleted molt into a gray pelage lacking the reddish coloration. This reddish coloration is not considered to be ge- netically based, but probably was the result of chemical alterations of pigment in old pelage and likely caused by ex- trinsic factors such as high concentra- tions of ammonia in the nest. Because coefficients of variation re- flect the ratio of the standard deviation to the mean, the statistic is useful in comparing the degree of variation be- tween populations of a single species or between populations of different taxa. The coefficient of variation can be used also to compare the relative reliability of different measurements of a single sample. Long (1968, 1969) recently sum- marized patterns of individual variation and comparative variation of measure- ments commonly used in taxonomic in- vestigations. Coefficients of variation for each sex of three samples of floridana and three samples of micropus are shown in figure 9. The coefficients are superimposed on Dice-grams illustrating the trends of vari- ation among the measurements taken. Size of all samples except the two (one male, one female) for canescens from Coahuila (M) are given in table 2. Sam- ple M consisted of 12 females and 13 males. Length of tail vertebrae and length of ear are the most variable di- mensions considered, and also are two of the four recorded from specimen labels. Length of hind foot and _ total length were recorded from specimen labels. The former shows about average variability, whereas total length is more variable than all except one cranial mea- surement, but less variable than lengths of tail and ear. Of the 10 cranial measurements, length of the palatal bridge is the most variable. This measurement varies in part with the shape of the posterior mar- gin of the palate (see beyond under geographic variation of qualitative cra- nial characters ), which is relatively vari- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 55 i 2 § 12 Sra 8 3 3 gueey i 4 sus 10 ) = s eo E PD eo ie s wn oO o 28 10 ats 45 112 6 2 uu 1 Le 4 6 1 9 10 5 2 . Ss aS ESE + + 12 5.0 6.0 70 8.0 9.0 Coefficients of Variation Fic. 9. Coefficients of variation for 14 external and cranial measurements showing variability of that statistic in each. Measurements are as follows: A—total length; B—length of tail vertebrae; C—length of hind foot; D—length of ear; E—greatest length of skull; F—condylobasilar length; G—zygomatic breadth; H—least interorbital constriction; I—breadth at mastoids; J—length of ros- trum; K—breadth of rostrum; L— alveolar length of maxillary toothrow; M—length of palatal bridge; N—length of nasals. Numbers plotted on the horizontal lines are individual coefficients of variation for each sample of adult woodrats, as defined below; odd numbers are males, and even numbers are females (see figure 8 for geographic areas included within the coded localities indicated in paren- theses): 1 and 2, Neotoma floridana baileyi (1); 3 and 4, N. f. campestris (2, 3, and 4); 5 and 6, N. f. attwateri (5, 6, and 7); 7 and 8, N. micropus canescens (B and C); 9 and 10, N. m. canescens (M); 11 and 12, N. m. micropus (P). The apex of the darkened triangle is the arithmetic mean of the coefficients of variation of the 12 samples, and the thick horizontal bar is plus and minus two standard errors of the mean; the thin horizontal bar is plus and minus one standard deviation of the mean, and the horizontal line is the range. able even at the intra populational level. All cranial measurements were recorded only to the nearest tenth of a millimeter and length of palatal bridge is one of the smaller dimensions; thus, the pre- cision of data recorded relative to size of the character measured would be only about one-fifth of that for, say, greatest length of skull. Least interorbital con- striction was the next most variable cra- nial measurement and also is one of the smaller dimensions; however, other char- acters measured having means of less than 10 (alveolar length of maxillary toothrow and breadth of rostrum) dem- onstrate near average variability. Great- est length of skull, condylobasilar length, zygomatic breadth, and mastoid breadth are the least variable characters mea- sured; all having average coefficients of variation less than 4.0. To compare the relative variation of the 14 characters simultaneously between the 12 groups, three simple tallies were made. When only extreme coefficients were considered, one or the other of the two samples of baileyi is least variable in six of the 14 measurements and one or the other of the samples of attwateri is most variable in five instances. This tally also shows that a sample of males is most variable in 10 of the 14 characters con- sidered. Each pair of samples from each locality next was compared for all mea- surements to determine if males were in fact more highly variable than females. Of the 84 comparisons (six localities times 14 measurements ) made, males are 56 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY more variable than females in 49 in- stances and less variable in 35, a differ- ence not significant at P < 0.05 as tested by Chi-square. Lastly, the coefficients of variation for each character were as- signed values on a rank-order basis so that the lowest coefficient for a character accrued one unit and the highest accrued 12 units to the respective taxa. These scores then were totaled for each sample by summing the 14 rank-order scores; thus a low total implies less variation and a high total more variation. As deter- mined by this crude technique, the sam- ple of baileyi females is least variable (41.5) and the sample of campestris males is most variable (134.0). Males were shown to be more variable than their female counterparts at four lo- calities (N. m. micropus from locality P, and N. m. canescens from combined lo- calities B and C being the two excep- tions); they have a rank-order total of 587 compared to 504 for females ( differ- ent at P < 0.05, as tested by Chi square). When scores of the sexes were summed for each locality, the sample of baileyi is least variable (106.5) and that of attwateri is most variable (215.5). Coef- ficients of variation in the sample of micropus is intermediate (158.0) be- tween baileyi and the four samples of widely distributed taxa (201.5-215.5); but in micropus the distribution of coef- ficients is erratic, probably reflecting the small sample size available. The tendency for samples of males to be more variable than those of females is indicated by all three methods of anal- ysis that I employed. Long (1969:298) found males of domestic mammals more variable than conspecific females, but indicated that no basis presently exists for attributing greater variation to males. The apparent presence of relatively less individual variation in the isolated sub- species, baileyi, as compared to widely distributed taxa is not surprising; small, isolated populations are prone to loss of variation by chance or “drift.” Addition- ally, they lack one of the most important means of acquiring “new” genetic varia- tion, i.e. immigration. Mayr (1963:177) suggested that a reasonable estimate of “new genes normally acquired by a local population through immigration is at least 90 percent and possibly exceeds 99 percent. It was expected a priori that individual variation in the samples of campestris would be more pronounced than in other taxa. In part the prediction was correct; the sample of campestris males is more variable than other samples, but only baileyi females and micropus males are less variable than campestris females. Possibly selection acting on populations that live in similar environments main- tains the observed degree of homogeneity (Ehrlich and Raven, 1969), or perhaps there is more interpopulational gene flow in campestris than my observations have indicated. Variation Resulting from Captivity The most striking differences ob- served between a specimen that had been reared, or at least maintained, in the laboratory for an extended period of time and one that had been killed at the time of capture were in the teeth. The cheekteeth of woodrats fed on laboratory chow did not wear at a rate comparable to that in natural populations. The molars of cleaned skulls of laboratory rats often are as much as a third or a half longer than those of non-laboratory animals. Furthermore, the reentrant angles of lab- oratory-reared woodrats extend much nearer to the alveolus than do those of comparably aged non-laboratory rats. Al- though this may be the result of reduced tooth growth to compensate reduced wear, the stimulus that stops or slows growth is unknown. In some laboratory specimens, alveolar tissue near the base of the molars appeared reduced and slightly porous. If the alveolar tissue of laboratory rats grows abnormally slow or if it is resorbed, the molar may undergo normal growth but have higher crowns. The incisors of woodrats living in the laboratory frequently are broken, result- ing in abnormal occlusion and the ab- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 57 sence of wear on the opposing incisor. The frequency of abnormal growth of incisors is relatively high in laboratory animals, whereas woodrats living in the natural environment and having maloc- cluding incisors probably are destined to early death. Comparisons of size of woodrats reared in the laboratory with non-labora- tory animals from the same geographic areas are shown in table 3. Only speci- mens at least 30 weeks of age that either were born in the laboratory or were cap- tured before they had completed the postjuvenal molt were included in lab- oratory samples. After specimens meet- ing these criteria were separated by sex, only three samples, Neotoma micropus canescens (localities B and C) males and females, and Neotoma floridana cam- pestris (localities 3 and 4) females, con- tained enough specimens (10 or more) for conducting the tests. Non-laboratory comparative samples included all avail- able specimens of age groups VI, VII, and VIII from the grouped localities in- dicated above. Differences in two measurements, total length and alveolar length of maxil- lary toothrow, are highly significant (P <0.01) between the two samples of N. f. campestris females; no highly sig- nificant differences were observed for N. m. canescens. Significant differences (P <0.05) were observed for two other characters (length of hind foot and length of nasals) in floridana, one mea- surement (total length) of micropus fe- males, and five measurements (length of hind foot, length of ear, greatest length of skull, condylobasilar length, and breadth of rostrum) for N. m. canescens males. Total length of micropus females is the only dimension significantly larger in the non-laboratory or “wild” sample. On the average, laboratory samples are slightly larger in most measurements that are not significantly different. Increased size of laboratory animals may be the result of a more nutritious diet or it may reflect differences in age. Many animals in the laboratory samples were near two years of age. It is doubt- ful that the average age of non-labora- tory animals equals that of the laboratory sample. GEOGRAPHIC VARIATION In the discussion beyond, I will in- terpret patterns of both qualitative and quantitative variation of morphological characteristics primarily from an evolu- tionary point of view in an attempt to elucidate the relationships of woodrats. If it can be determined whether patterns of variation are concordant or discordant, and clinal or abrupt, one can surmise which patterns have resulted from pri- mary intergradation, secondary intergra- dation, or from present restrictions to gene flow. Also, I will attempt to ascer- tain if natural hybridization is introgres- sive in N. floridana and N. micropus. Pelage, Molt, and Color Finley (1958:232) described the suc- cession of molts and pelages of woodrats as juvenal pelage, postjuvenal molt, sub- adult pelage, second molt, first autumn pelage, third molt, first winter pelage, annual molt. My observations agree in a general way with this scheme. Animals born late in summer or early in autumn, however, do not undergo the complete sequence, but spend the first winter in either the subadult pelage or first autumn pelage. Remarkably little published informa- tion pertaining to molt in adult Neotoma is available. Goldman (1910:12) sum- marized his understanding of molt on adults as follows: “The molting season is somewhat irregular, especially in the southern part of the range of the group. The northern species molt once a year, toward the end of summer or fall. The southern forms usually molt in early winter, but individuals in worn and in fresh pelage may often be seen together.” Linsdale and Tevis (1951:450-458) de- scribed and discussed molt in Neotoma fuscipes, and Finley (1958) studied it in those species of woodrats that occur in 58 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 3. Size comparisons of woodrats reared in the laboratory and those killed at the time of initial capture (wild). Statistics given are sample size, mean, two standard errors of the mean, range, coefficient of variation, Fs (F value calculated by single classification ANOVA), and F (tabular F value at level of significance or at P<0.05 if not significant). One asterisk and two asterices indicate significance at the 0.05 and 0.01 levels, respectively, whereas ns indicates no significant difference. Measurement F, and treatment N Mean a= PND, Range (GW F Neotoma floridana campestris females (Samples 3 and 4) Total length Wild 33 370.8 6.23 (340.0-409.0) 4,83 9.86 Laboratory 10 394.2 17.94 (331.0-421.0) 7.20 (gl Length of tail vertebrae Wild 30 154.9 3.59 (136.0-175.0) 6.66 3.70 Laboratory 10 162.7 8.93 (129.0-178.0) 8.68 4.07 ns Length of hind foot Wild 34 39.3 0.55 (36.0-42.0) 4.09 5.94 Laboratory 10 40.7 0.45 (39.0-43.0) 3.48 4.07 * Length of ear Wild 19 2.8 0.72 (25.0-32.0) 5.53 < 1.00 Laboratory 10 2.9 0.13 (24.0-31.0) 7.01 4.21 ns Greatest length of skull Wild 29 5.0 0.52 (47.0-53.3) DiS <1.00 Laboratory 12 5.0 il li/ (46.2-52.4) 4.05 4.10 ns Condylobasilar length Wild 26 48.2 0.56 (45.2-51.6) 2.97 2.83 Laboratory 12 49.1 0.99 (45.2-51.2) 3.49 4.11 ns Zygomatic breadth Wild oe 27.0 0.38 (25.1-29.6) 3.68 <1.00 Laboratory 12 27.2 0.63 (25.7-29.0) 4.03 4.11 ns Least interorbital constriction Wild oz 6.7 0.11 (6.1-7.5) 4.72, < 1.00 Laboratory 12 6.7 0.21 (6.2-7.5) oul 4.07 ns Breadth at mastoids Wild 29 19.4 0.24 (18.1-21.0) Sol <1.00 Laboratory 12 19.6 0.35 (18.5-20.5) Sale 4.10 ns Length of rostrum Wild 33 19.4 0.25 (i8=21)) 3.70 1.85 Laboratory 12 19.8 0.72 (17.4-21.2) 6.35 4.07 ns Breadth of rostrum Wild 32 8.5 0.12 (7.9-9.4) 3.85 1.34 Laboratory 11 8.6 0.25 (7.8-9.2) 4.81 4.08 ns Alveolar length of maxillary toothrow Wild 34 9.7 0.09 (9.1-10.3) 2.84 NPAT Laboratory 12 10.0 0.09 (9.6-10.3) 1.61 ems Length of palatal bridge Wild 34 8.5 0.14 (7.8-9.4) 4.72 < 1.00 Laboratory 12 8.6 0.25 (8.2-9.6) 5.09 4.06 ns Length of nasals Wild 38} 19.2 0.26 (17.6-20.4) 3.93 5.30 Laboratory) 12 19.8 0.57 (18.0-21.1) 4.97 4.07 * WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 59 TABLE 3. Continued. Measurement F,; and treatment N Mean == 2SE Range CV F Neotoma micropus canescens females (Samples B and C) Total length Wild 31 355.8 5.97 (310.0-382.0) 4.67 4.47 Laboratory 11 351.8 11.25 (326.0-382.0) 5.30 4.08 * Length of tail vertebrae Wild 31 147.1 3.70 (130.0-165.0) 7.01 < 1.00 Laboratory iil 146.2 8.32 (126.0-171.0) 9.43 4.08 ns Length of hind foot Wild 30 38.4 0.54 (36.0-41.0) 3.85 2.90 Laboratory 13 39.2 0.69 (37.0-41.0) Bills 4.08 ns Length of ear Wild 24 27 0.56 (25.0-30.0) 5.10 4.00 Laboratory 13 28.1 0.86 (25.0-30.0) Bibs 4.l3)us Greatest length of skull Wild aT 48.8 0.70 (44.2-51.8) 3.75 <1.00 Laboratory 14 48.9 0.86 (46.2-51.4) Soll 4.10 ns Condylobasilar length Wild 29 47.0 0.58 (42.8-50.0) 3.34 < 1.00 Laboratory 15 47.4 0.71 (45.0-49.4) 2.92 4.07 ns Zygomatic breadth Wild 30 26.5 0.39 (24.7-29.1) 4.06 < 1.00 Laboratory 14 26.7 0.43 (25.2-30.0) 3.04 4.07 ns Least interorbital constriction Wild 32 6.3 0.11 (5.8-7.0) 4.72 2.80 Laboratory 15 6.2 0.15 (5.7-6.6) 4.73 4.06 ns Breadth at mastoids Wild HH if yal 0.23 (17.9-20.3) 3.10 Sea Laboratory 14 19.4 0.17 (18.9-19.9) 1.65 4.08 ns Length of rostrum Wild 30 18.9 0.27 (17.2-20.2) 3.98 < 1.00 Laboratory 14 18.8 0.39 (17.8-20.2) 3.93 4.07 ns Breadth of rostrum Wild 32 8.3 0.14 (7.2-9.3) 4,92 < 1.00 Laboratory 15 8.4 0.21 (7.9-9.3) 4.87 4.06 ns Alveolar length of maxillary toothrow Wild 32 9.4 0.14 (8.5-10.1) 4,35 < 1.00 Laboratory 15 9.3 0.19 (8.4-9.9) 3.92 4.06 ns Length of palatal bridge Wild 31 8.0 0.19 (7.1-9.5) 6.56 < 1.00 Laboratory 15 8.0 0.18 (7.4-8.5) 4.38 4.06 ns Length of nasals Wild 30 19.2 0.35 (16.7-21.2) 5.04 < 1.00 Laboratory 14 19.4 0.43 (18.0-20.9) 4.15 4.07 ns Neotoma micropus canescens males (Samples B and C) Total length Wild 23 370.1 9.46 (334.0-411.0) 6.13 PA Laboratory 8 383.0 11.84 (354.0-398.0) 4.37 4.18 ns 60 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 3.—Concluded. Measurement Fs and treatment N Mean as USD Range CV F Length of tail vertebrae Wild 23 152.6 5.10 (131.0-175.0) 8.01 < 1.00 Laboratory 8 154.8 (hey (142.0-172.0) 6.87 4.18 ns Length of hind foot Wild 5: 39.2 1.01 (35.0-45.0) 6.46 6.18 Laboratory 10 41.3 0.79 (39.0-43.0) 3.03 ANG Length of ear Wild 16 Weal 0.72 (25.0-29.0) 5.31 7.39 Laboratory 11 28.7 1.05 (27.0-32.0) 6.05 4.94 * Greatest length of skull Wild 25 49.5 0.63 (46.4-52.9) Bh ly 4.79 Laboratory 10 50.6 0.43 (49.4-51.9) 1.35 AL lis) Condylobasilar length Wild 24 48.3 0.66 (44.6-50.9) 3:08 5.24 Laboratory 10 49.5 0.44 (48.3-50.6) 1.39 ANS Zygomatic breadth Wild 26 26.7 0.36 (25.1-28.8) 3.47 < 1.00 Laboratory TR 26.9 0.39 (26.1-28.2) 2.42 4.13 ns Least interorbital constriction Wild Nil 6.3 0.11 (5.8-6.9) 4.39 < 1.00 Laboratory 11 6.4 0.13 (6.1-6.9) 3.38 4.11 ns Breadth at mastoids Wild 24 19.3 0.28 (18.0-20.8) 3.58 < 1.00 Laboratory ll 19,4 0.10 (19.2-19.8) 0.87 4.15 ns Length of rostrum Wild 26 19.4 0.28 (17.8-20.7) 3.67 < 1.00 Laboratory 10 19.5 0.40 (18.3-20.4) 3.23 4.13 ns Breadth of rostrum Wild 27 8.4 0.14 (7.5-9.2) 4.4] 4.86 Laboratory 11 8.6 0.22 (8.0-9.2) 0.22 ALL Alveolar length of maxillary toothrow Wild 27 9.3 0.12 (8.7-10.1) 3.47 2.19 Laboratory 11 9.5 0.22 (9.0-10.0) 3.81 411 ns Length of palatal bridge Wild 26 8.1 0.18 (6.8-8.9) 5.73 3.37 Laboratory 10 8.4 0.13 (7.9-8.6) 2.54 4.13 ns Length of nasals Wild 26 19.8 0.32 (18.0-21.1) 4.11 < 1.00 Laboratory 10 19.9 0.35 (19.0-20.7) 2.79 4.13 ns Colorado. In both studies it was con- cluded that only a single annual molt occurs in adult woodrats. Seasonal occurrence of molt in se- lected samples of N. floridana and N. micropus is shown in table 4. Specimens of age-groups V-VIII were included in these tabulations. With one exception, animals were considered to be molting if new pelage appeared to have been re- placing old pelage regardless of whether the replacement was symmetrical or in- volved a complete replacement of hair. Many woodrats have a varying number of tiny spots of actively growing hair. These probably are areas in which hair WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA lost while fighting or in other ways not directly associated with seasonal molt is replaced. Specimens with such “spots” that were not molting elsewhere were not considered to be molting. _ As can be seen in table 4, some wood- rats obtained in every month were molt- ‘ing. It was thought that possibly only animals of age-groups V and VI (which still might have been in some stage of maturational molt) were molting at times other than late summer and au- tumn as has been reported previously. However, when only specimens of age- groups VII and VIII were considered, the seasonal array of molting and non- molting individuals remained approxi- mately the same. In northern populations of the two species studied, adults are almost invari- ably in a dense, luxuriant winter pelage by late November or early December. Beginning in late winter or early spring, the winter pelage of many individuals begins to deteriorate; it becomes thinner, less luxuriant, has many broken tips, and appears “scruffy.” In other rats, the )winter pelage seems to be well main- /tained into late May or early June. When the pelage begins to deteriorate, it gen- erally is replaced erratically over the body, usually beginning in those areas where the winter pelage is thinnest or 61 most worn. This “molt” has little or no symmetry of pattern. In occasional indi- viduals, the winter pelage is nearly or completely replaced by a shorter, usually darker “summer” pelage. Some rats have only localized spots of this pelage and others show no sign of replacement until July or August. At this time they appar- ently begin the “annual molt” and molt the old winter pelage directly into a new winter pelage. Those individuals that re- placed some or all of the pelage earlier also molt into a new winter pelage in late summer and autumn. The molting sequence in southern populations of both species is less clear. Even the new winter pelage of southern woodrats is shorter than “summer pelage” of those from northern populations, and the “annual molt” (molt into winter pel- age) is only weakly synchronized among animals of the same population. Usually this molt occurs anytime from June to October, and generally is complete by November. Whether one wishes to consider the replacement of worn winter pelage prior to the attainment of new winter pelage as a “vernal molt” and the resultant pel- age as a “summer pelage” is primarily a question of semantics. Almost certainly the only molt that is consistently com- plete and common to all adults is the TABLE 4. Seasonal distribution of molt in selected samples of Neotoma floridana and N. micropus. See figure 8 for geographic areas included in coded localities. Jan. Feb. Mar. April May June July Aug. Sept. Oct. Nov. Dec. Sample 1 (Neotoma floridana baileyi) Females Molting = dat a = 0 = 1 2 = = = = Examined __ : & 1 _— i 3 ae pi ae ses Males Molting = z ee es 1 S pases 2 si 2 BE as Examined __ <= Ss — i om mat 2 28 2; = a Samples 2, 3, and 4 (Neotoma floridana campestris) Females Molting re fe re 3 0 2, 2 2 5 2 1 0 Examined - tal 3 il 3 oe 3 5 38 2 9 Males Molting <4 = 1 id 5 Z 1 3 c 0 1 Examined _ i = 5 z 1 3 E 2; To 62 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 4.—Concluded. Jan. Feb. Mar. April May June July Aug. Sept. Oct. Nov. Dec Samples 5, 6, and 7 (Neotoma floridana attwateri) Females Molting 0 5 2 1 js 1 1 1 6 1 Examined 2 Ta 12 2 : as 1 1 2 Tal 12 Males Molting 0 1 0 0 sde i = 3 1 2 5 0 Examined 4 ®) 8 4 ne =a 3 1 2 9 6 Samples 8, 9, 10 (Neotoma floridana attwateri) Females Molting 2 0 2 7a i 0 ss 1 4 3 2 Examined 9 1 9} mah 3 1 sats 2 6 6 9 Males Molting i 0 0 0 Sis 3 std 2a 5 3 1 Examined 8 9) 6 3 3 =a Pa 5 4 6 Samples 11 and 12 (Neotoma floridana attwateri) Females Molting 0 1 _ 0 2 5 3 Examined 3 1 ell, eee ed I 3 6 5 Males Molting 1 0 Ad ae se Boe Ae 2 2 5 1 Examined 2, 2 a ae en aie a 2 3 5 2 Samples A, B, C, and F (Neotoma micropus canescens) Females Molting 0 0 il 8 3 i 2 2 0 = Examined 2 i 1 10 i 6 9) 2 3 Males Molting 1 1 0 0 3 2 6 fone 3 1 0 Examined = 1 1 1 1 5 4 6 ae 4 5 4 Samples D, G, H, and I (Neotoma micropus canescens) Females Molting 0 om 0 1 6 zs 1 3 0 0 Examined 2 es 1 3 us at 1 3 1 9 Males Molting 0 0 a ees 3 2 a is pall rs 0 Examined 3 2 aes oe 3 6} ee EN fe tte 10 Samples J, K, L, and M (Neotoma micropus canescens) Females Molting 0 0 3 0 1 0 1 1 4 0 Examined 1 2 4 1 9) 1 2 i 12 i Males Molting 0 1 4 0 1 i = i = 10 * Examined I 1 6 1 il 3 be i as 16 2 Samples N and P (Neotoma micropus micropus) Females Molting 0 2 1 1 2 hse 2 0 Examined a 4 4 1 1 2 10 ] Males Molting ) 3 2 2, 4 3 0 Examined 1 4 2 4 5 9 2) WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 63 autumn molt or in Finley’s (1958) ter- minology, the annual molt. The “vernal molt” apparently is primarily a mechan- ism to maintain the pelage and may be complete, abbreviated, or absent. For ‘many years it was thought that members of the genus Peromyscus molted only ‘once a year (see Layne, 1968:141, for review), but recent studies have shown ‘that at least in some species molting also ‘occurs at other times of the year ( Brown, 11963: Lawlor, 1965). Possibly the sea- sonal molting regimes of members of the two genera are similar, but characterized y much more individual and geographic ariation than previously has been hought. An attempt was made to correlate re- roductive data from specimen labels with molt in females. Pregnant or lactat- ing females that were collected in spring or early summer usually were in the old winter pelage; however, pregnant and lactating females collected in late sum- mer often were actively molting. An adult female N. f. campestris (KU 120844) was captured in December in a relatively new winter pelage. She was maintained in the laboratory until late March without having been placed with amale. At that time she was undergoing what probably would have been a nearly complete molt from her typical winter pelage, which was in remarkably good repair, to a new shorter darker “summer pelage.” On 20 March she was placed with a male and on 26 April she gave birth to a litter. Insofar as I could deter- mine, the molt had progressed very little between those dates and remained with- out change until she was killed on 10 June. The specimen shows no line of “current” activity between the two pel- ages, which are markedly different in length, color, and general texture. Ap- parently molt of woodrats is influenced by the hormones of reproduction and shortly after this female became preg- nant the molt was arrested. Other factors that probably influence the timing and degree of completeness of the “vernal molt” include age, condi- tion of health, and condition of the ex- isting pelage. Although exchange of pel- ages clearly is necessary, especially in northern woodrats preparing for winter, molt may be one of the body processes that is under a relatively loose genetic control and easily altered when it is physiologically advantageous for an in- dividual to divert energy or reserves elsewhere. The variation in molts and pelages of woodrats in spring and summer re- sulted in some difficulty selecting speci- mens for color measurements. Ideally, only adults in fresh winter pelage would be considered in analyses of geographic variation in color. However, sufficient samples from the various aggregate lo- calities were not available when samples were thus limited. It was necessary to include all specimens whose pelage was in relatively good repair, regardless of season. The effect of including animals col- lected at different times of the year was tested for each of the three reflectance readings and for the total (value ob- tained by summing the three individual reflectance readings for each individual) by pooling readings of animals from lo- calities 5 through 11 (all Neotoma flori- dana attwateri) and separating the indi- viduals into four seasonal samples. Each sample included animals killed during a three-month period so that four samples corresponding roughly to winter (De- cember-February ), spring (March-May), summer (June-August), and autumn (September-November) were available. Seasonal variation is not significantly dif- ferent in reflectance of blue or green, but is significantly different (0.05 > P > 0.01) for reflectance of red and for total reflectance. These data were separated by sex to test males against females. Males were significantly paler (0.05 > P > 0.01) as indicated by reflectance of red, but F, values were less than unity for blue and green, and less than F for total reflectance. Explanation of the slight seasonal variation is commensurate with the above 64 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY discussion of molt and pelages. Most woodrats are darkest in fresh winter pel- age and palest in old pelage just before undergoing molt into winter pelage, which is the only more or less synchro- nous molt of adults. This molt usually occurs in September, October, or early November; thus most animals in the au- tumn sample recently had molted or were molting. The high coefficients of variation for this sample are attributable to the fact that all specimens were not in the same pelage. The three seasonal sam- ples involving mostly animals in winter pelage were not significantly different in color, and even the sample composed mostly of summer specimens was not sig- nificantly different from the spring or winter samples. Specimens of both sexes and from all seasons were pooled for each locality for studies of geographic variation. Re- sults of univariate analyses of intra- specific color variation in Neotoma flori- dana are shown in table 5. Highly significant (P < 0.01) differences exist in comparisons of group-means for all re- flectance readings. In no case, however, are animals from localities 5-13 (all sam- ples of N. f. attwateri and the single sample of N. f. rubida) significantly dif- ferent from each other with respect to color. Specimens from localities 12 (southern Texas), 6 (northeastern Kan- sas), and 11 (northern Texas) tend to be slightly paler in color than other spec- imens. Samples 5 (north-central Kan- sas), 10 (southeastern Oklahoma), and 13 (N. f. rubida) generally are darkest. Within the subspecies N. f. campestris, a rather clear trend exists from paler animals in the west (localities 2 and 3) to darker ones in the east (locality 4) where the range of campestris meets that of attwateri. In no case is the difference between animals from localities 2 and 3 significant, but those from locality 4 are significantly darker than those from 2 in all reflectance readings. Specimens in sample 4 also are significantly paler in all readings from those of attwateri from adjacent locality 5. Only in reflectance of blue are differences between samples 3 and 4 shown to be significantly differ- ent. The darker color of specimens from locality 4 as compared to those in sam- ples 2 and 3 probably has resulted from intergradation with the darker attwateri population to the east. With respect to color, the zone of intergradation would appear to have been assigned largely to campestris, although in size (see be- yond) animals from locality 5 are more like those from locality 4 than from locality 6. Neotoma floridana baileyi (locality 1) is paler than all samples of attwateri, and significantly darker than campestris (with the exception of the sample of campestris from the narrow zone where campestris intergrades with attwateri at locality 4). The pale coloration of baileyi is less tannish than that of campestris, and although most specimens of baileyi are distinctly paler than specimens of attwateri, their coloration more closely resembles that of attwateri, than that of campestris. The habitat in which baileyi occurs resembles that where attwateri is found, but most adjacent habitat types in northern Nebraska, which thus far have not been found to support wood- rats, are mostly shortgrass pasture (more like the habitat of campestris). The pale color seen in baileyi and campestris probably signifies convergence, which has resulted from adaptation to an arid environment, from a darker common an- cestor. Table 6 contains results of univariate analyses of intraspecific color variation for Neotoma micropus. As opposed to the pattern of color variation in floridana, that in micropus is clearly clinal and lacks noticeable steps. Specimens from locality E (New Mexico) are paler on the average than those from White Sands National Monument (locality O, pre- viously N. m. leucophea). Therefore, there seems to be no sound reason for recognizing the name leucophea. (If larger samples separable by season had been available, the White Sands popula- tion might have averaged paler, but cer- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 65 tainly the differences are slight.) When all samples of micropus are considered, specimens from localities in New Mexico (E and O) and western Texas (F and J) generally are palest in color as evinced by higher reflectance readings. Those from localities D, G, H, K, L, N, and P generally are darker than those from A, B, C, I, and M, which tend to be intermediate. Darkest populations gen- erally occur at localities in the eastern parts of the range of the species, and palest populations are from the more arid western localities. As shown by the TABLE 5. Geographic variation in color of selected samples of Neotoma floridana. F, was calculated by single classification analysis of variance. Tabular F values are at the P<0.05 level of significance; ns indicates no significant difference within a group of means. Nonsig- nificant subsets (as calculated by the Sums of Squares Simultaneous Testing Procedure) of significantly different groups of means are shown in the last column. See figure 8 for geographic areas included within each coded locality. Color reflectance measured, and His coded localities N Mean + 2SE Range CV F SS-STP Red 2 2 19.2 0.50 (19.0-19.5 ) 1.84 12.78 I 3 19 17.0 1.06 (13.0-20.5 ) 13.66 1.83 I 1 23 15.6 0.88 (12.0-20.0) 13.56 til 4 16 15.5 0.97 (12.0-19.5) 1252: ay Tne 12 16 14.1 0.92 (11.5-17.0) 12.96 Gemma 9 7 13.1 1.18 (12.0-16.5 ) 11.98 ee ok 6 12 13.0 0.85 (10.0-14.5 ) 133 Ie ll TST il ONT 0.69 (11.5-14.0) Tells} if Jl 0 28 12.7 0.53 (10.0-17.0) 11.07 I 13 5 12.6 1.66 (10.5-15.0) 14.69 I 8 4 D5 0.82 (11.5-13.5) 6.53 I 10 5 12.2 0.93 (10.5-13.0) 8.50 I 5 6 ED 1.09 (10.0-13.5) 11.90 I Blue 2 2 10.5 2.00 (9.5-11.5) 13.47 41.62 I 3 19 9.6 0.45 (7.5-11.5) 10.20 ESS I 4 16 8.3 0.42 (7.0-10.5 ) 9.96 I 1 23 8.0 0.32 (6.5-9.5 ) 9.61 I 12 16 6.6 0.28 (5.5-7.5) 8.49 I 5 6 6.5 0.52 (6.0-7.5) 9.73 I 11 ii 6.5 0.65 (5.5-8.0) 13/32 I 8 4 6.5 0.00 (6.5-6.5) 0.00 I 6 12 6.4 0.45 (5.0-7.5 ) Qa, I 9 1 6.3 0.37 (5.5-7.0) 7.76 I 7 28 6.0 0.19 (5.5-7.0) 8.41 I 13 5 5.6 0.37 (5.0-6.0) TAT I 10 5 Dro 0.75 (4.0-6.0) 15.79 I Green 2 2, 12.0 2.00 (11.0-13.0) 11.79 39.62 I 3 19 10.5 0.52 (8.5-12.5) 10.85 1.83 | Lea 4 16 9.4 0.49 (8.0-11.5) 10.40 ae 1 23 8.3 0.28 (7.5-9.5) 8.10 MeL 6 102 les 0.40 (6.5-8.5 ) 9.35 IE JI 8 4 UP 0.65 (6.5-8.0) 8.90 iy, Ll 5 6 Mp2, 0.43 (6.5-8.0 ) E23 Ill 12 16 Ux 0.36 (6.0-9.0 ) 10.11 I 11 il UP 0.61 (6.5-9.0) PR22; I a 28 6.9 0.20 (5.5-8.0) 7.65 I 9 1 6.9 0.29 (6.5-7.5) 5.51 I 13 5 6.5 0.32 (6.0-7.0) 5.44 I 10 5 6.2 0.68 (5.0-7.0) 12.23 I 66 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 5.—Concluded. Color reflectance measured, and Firs coded localities N Mean =+ 2SE Range CV F SS-STP Total 2 2, 41.8 4.50 (39.5-44.0) 7.62 29.51 3 19 BH/all 1.90 (29.0-44.5 ) 11.19 1.83 I 4 16 33.2 1.66 (28.0-41.5) 9.96 oe 1 23 Sei, Lea (26.0-35.5 ) 9.14 eel 12 16 28.0 1.40 (23.0-33.5 ) 10.00 HI 6 12, 26.8 1.44 (2275=30)5)) 9.31 I 11 a 26.4 ei (23.5-31.0) 8.84 I 8 4 26.2 0.87 (25.5-27.5) 3.30 I 9 a 26.2 1.41 (24.0-30.0) 7.12 I i 28 25.7 0.79 92,.0-31.5) 8.18 I 5 6 24.9 1.85 22,.5-29.0) 9.10 I 13 5 24.7 2.14: 22,.0-28.0) 9.67 I 10 5 VT 2.20 19.5-25.5) 10.40 I sequence of means and arrangement of maximal non-significant subsets in table 6, however, the trends in color variation in N. micropus take the form of gradual clines. Qualitative Cranial Characters Finley (1958:248-252) discussed sev- eral cranial features of woodrats that vary among taxa. Three cranial charac- ters can be employed to distinguish skulls of Neotoma angustipalata, N. floridana, and N. micropus, although none is diag- nostic. A fourth varies greatly with sex and age but is useful in skull identifica- tion. The three most useful characters, the anterior palatal spine, the posterior margin of the bony palate, and the sphenopalatine vacuities (Fig. 10), are discussed individually below and anal- yzed geographically. The fourth, shape of the interorbital region, was assessed in the measurement of least interorbital constriction. In micropus, the supraor- bital region tends to be narrower and more ridged than in floridana. Espe- cially in mature micropus males (less frequently in females ), ridging is so pro- nounced that a structure resembling a postorbital shelf is formed. In floridana such a “shelf” is never present and the interorbital region usually is nearly level. Ridging between the orbits and presence of the “shelf” generally are distinctive to micropus, but because of variation with sex and age, absence of these characters is not distinctive to floridana. The num- ber of adult N. angustipalata available for analysis of normal variation in the interorbital region is small; however, specimens examined tend to be more like micropus than floridana in this character. Berry and Searle (1963) discussed occurrence and frequency in several rodent species of characters similar to the three considered below. Hedges (1969) studied such characters inter- specifically and geographically in two species of Apodemus. These authors, and others, referred to such characters as “epigenetic characters.” Berry and Searle (1963:607) stated that “many genes are concerned in the determination of each character, while environmental factors are also very important, so that the ef- fects of individual genes cannot be iso- lated.” I am not presently prepared to comment on the relative genetic versus environmental control of the germane characters, but planned study of these through several generations of labora- tory-bred woodrats should be elucidat- ing. Although the term “epigenetic” may well apply to characters such as these, I prefer to avoid use of the term until more is known about their developmental control and functional importance. How- ever, “qualitative characters” is also WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 67 TABLE 6. Geographic variation in color of selected samples of Neotoma micropus. Fs; was calculated by single classification analysis of variance. Tabular F values are at the P<0.05 level of significance; ns indicates no significant difference within a group of means. Non- significant subsets (as calculated by the Sums of Squares Simultaneous Testing Procedure) of significantly different groups of means are shown in the last column. See figure 8 for geographic areas included within each coded locality. Color reflectance measured, and F, coded localities N Mean + 2SE Range CV F SS-STP Red E 5 yall 0.97 (16.5-19.0) 6.34 5.44 I O 2 16.8 0.50 (16.5-17.0) 2.11 V6 II J 3) 16.8 2.50 (15.5-18.0) 10.55 en F 2 16.8 0.50 (16.5-17.0) 2 Aa if ah 1 B 18 16.2 0.98 (13.0-19.5) 12.87 It 1k C 17 15.6 0.88 (13.5-19.5) 11.56 Vit M 32 15.1 0.78 (10.5-20.5 ) 14.69 1h J fai A 5 14.9 0.73 (14.0-16.0 ) 5.51 oe It wt I 3 14.5 0.58 (14.0-15.0) 3.45 ele teon G 2 14.2 2.50 (13.0-15.5) 12.41 If Je Wt VE H 18 13.9 0.72 (5 = 17.0) 10.94 Ie Te I L 3 13.7 2.40 (12.0-16.0) 15.18 Je al 12) 10 13.4 1.09 (11.5-17.0) 12.81 If lt N 23 Sal 0.66 (11.0-16.5) eds I D 12 Ls! 0.81 (10.0-14.5) 10.79 I K 13 13.0 1.08 (8.5-15.5) 14.97 I Blue F 2 ORD, 3.50 (10.5-14.0) 20.20 AKG: I E 5 10.5 1.18 (9.0-12.5) 12.60 1.76 ie O 2 10.2 1.50 (9.5-11.0) 10.35 II J 2 100 2.00 (9.0-11.0) 14.14 ie B 18 10.0 0.56 (8.0-12.5) 11.88 1 C yy 9.4 0.42 (8.0-11.0) 9.2) IE JEU A 5 9.1 0.66 (8.0-10.0) 8.15 if Ae I M 32 8.9 0.52 (7.0-13.0) 16.69 i it i I 3 8.8 0.67 (8.5-9.5) 6.54 it G 2 8.2 1.50 (7.5-9.0 ) 12.86 eae It al H 18 8.2 0.42 (7.0-10.0) 10.88 It Jt dt D 12 7.9 0.46 (7.0-9.5) 10.02 1G It L 3 7.5 1.00 (7.0-8.5) E55 WAT K 13 eo 0.56 (5.0-9.0) 13.84 Tek N 23 eS 0.34 (6.0-9.5 ) 11.26 I P 10 6.8 0.44 (5.5-7.5) 10.35 I Green F » 15.0 1.00 (14.5-15.5) 4.71 11.44 I J 2 12.8 7.50 (9.0-16.5) 41.59 1.76 ey E 5 10.7 ONT (9.5-13.0) 12.63 eter B 18 10.4 0.68 (8.5-13.5) 13.83 leer O ® 10.0 1.00 (9.5-10.5) 7.07 ee a Cc 17 9.6 0.44 (8.0-11.0) 9.41 Ie Ut We Te at A 5 9.5 0.63 (8.5-10.5) 7.44 ME Te M 32 9.2 0.57 (6.5-13.5) 17.45 IL JG a Wh I H 18 8.7 0.47 (7.0-11.0) 11.53 ML ee a a G 2 8.5 1.00 (8.0-9.0 ) 8.32 Teele Vest I 3 8.5 1.00 (8.0-9.5) 10.19 We A L 3 8.3 1.20 (7.5-9.5) 12.49 Me a D 12 8.2 0.57 (7.0-10.5 ) 12.06 eel K 13 Hell 0.60 (5.5-9.5 ) 14.18 lial N 3} 7.6 0.35 (6.0-9.5 ) 10.86 I 1p 10 Ue 0.58 (6.0-8.5 ) 12.76 I 68 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 6—Concluded. Color reflectance measured, and F; coded localities N Mean + 2SE Range CV F SS-STP Total F 2 44.0 2.00 (43.0-45.0 ) PAL 8.93 I J 2, 39.5 12.00 (33.5-45.5 ) 21.48 1.76 II E 5 38.3 3.20 (35.5-44.5 ) 9.35 I ll O 2 37.0 3.00 (35.5-38.5 ) 5.73 It it 3 B 18 36.0 PAPAL (27.5-43.5 ) 13.03 Mh Jt I Cc 17 34.6 1.53 (29.5-40.5 ) 9.11 ETL A A 5 3o.0 1.70 (30.5-35.5 ) 5.68 1G Ct 3 M 32 Soul: Wee (24.5-46.5 ) Way ils} YL Wu I 3 31.8 E33} (30.5-32.5 ) 3.63 Ie Ie i G 2, 31.0 5.00 (28.5-33.5) 11.40 ye I ae H 18 30.8 1.47 (26.5-36.5 ) 10.10 oe Ie 1G; 3 29.5 1.53 (28.0-30.5 ) 4,48 it at dt D 12 29.2 1.67 (25.0-34.5 ) 9.92 ML gl N 2S 28.0 ore, (24.0-34.0) 10.41 II K lis} 28.0 2.16 (19.0-33.5 ) 13.90 I 12 10 Hao) 1.74 (23.5-33.0) 10.00 I somewhat of a misnomer because the variation is nearly continuous and, as — y noted by Berry and Searle (loc. cit.), must be treated by statistics rather than Mendelian methods. Another important characteristic of woodrat skulls involves the relative de- velopment of the vomer within the narial passage and the resultant absence or presence and relative size of a maxillovo- merine notch (Finley, 1958:249). In four species of Neotoma, the three dis- cussed herein and N. palatina (Hall and Genoways, 1970), the vomerine septum is solid anterior to the palate (see also Anderson, 1969:47, Fig. 7). None of the specimens of N. angustipalata, N. flori- dana, or N. micropus from the Central Ces Plains examined by me exhibit a maxil- lovomerine notch, but all specimens of N. f. magister in the Museum of Natural History of The University of Kansas have Fic. 10. Semidiagrammatic drawing of a skull of Neotoma floridana showing: A—ante- rior palatal spine; B—posterior margin of the bony palate; and C—sphenopalatine vacuities. Enlargements of A-C (scale at bottom applies to all) to the right illustrate the range of vari- ation seen in these characters in N. floridana and N. micropus. Numbers represent sequential scoring values assigned to each character. KU numbers of skulls from which enlargements were drawn are: A—(1) 53908, (2) 3094, (3) 117335, (4) 117325, (5) 119707; B—(1) 119615, (2) 117786, (3) 117783, (4) 117324, (5) 119797, (6) 119796, (7) 119804, (8) 53906; C—(1) 16111, (2) 117774, (3) 117786, (4) 117324, (5) 119804, (6) 38922. a deep notch and also lack a fork on the anterior palatal spine discussed below (but see Schwartz and Odum, 1957). Anterior Palatal Spine-—Geographic variation in the morphology of the an- terior palatal spine is shown graphically in figure 11 and in percent frequency of occurrence of the five categories (Fig. 10) in table 7. Calculations for making the pie diagrams (Fig. 11) were as fol- low: 1) the percent frequency of each WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 69 category was multiplied by the score shown (Table 7) for that category; 2) these values were summed for each grouped locality; 3) the lowest total was subtracted from each total; and 4) this value then was divided by the largest remaining value and converted to per cent development of the fork relative to the sample having the largest fork. This percentage value is represented by the darkened areas in the symbols of figure 11. Thus, circles that are most darkened represent samples having a greater fre- quency of occurrence and larger size of the fork on the palatal spine than samples from localities with more open circles. Morphology of the anterior palatal spine varies geographically in both N. floridana and N. micropus. In N. f. baileyi, the bifurcate condition is ob- served in more than 91 percent of the specimens examined and in more than 40 percent of these the spine is classified as “large.” As a result, baileyi represents 100 percent development for this charac- ter and the percent development for all other samples is relative to this situation in baileyi. Of the two specimens of N. m. canescens from White Sands National Monument, New Mexico (locality O, previously N. m. leucophea), examined for this character, neither has a fork; thus, that sample represents zero percent relative to the condition seen in baileyi. TABLE 7. Percent frequency of occurrence of five morphological categories of the anterior palatal spine in 31 grouped samples of Neotoma floridana, N. micropus, and N. angustipalata. See figure 10 for illustrations of morphological categories and figure 8 for geographic areas included within each coded locality. Spatu- Small Medium Large Locality Pointed late fork fork fork code N (1) (2) (3) (4) (5) 1 49 8.16 8.16 40.82 42,86 2, 43 9.30 = 55.82 25.58 9.30 3 79 10.13 5.06 48.10 26.58 10.13 4 69 8.70 11.60 Bo00 31.88 14.49 5 12 4 =e 58.33 33.34 8.33 6 67 13.43 4,48 55S 19.40 7.46 i 63 SLAM Take aL 34.92 31.75 ILA 8 73 24.66 4.11 41.09 17.81 12.33 9 93 6.45 13.98 47.31 32.26 10 63 17.46 2a 28.57 33.33 20.64 J ON 14.82 14.82 33.33 33.33 3.70 12 50 38.00 6.00 46.00 8.00 2.00 13 17 41.18 5.88 47.06 uae 5.88 A Bi7/ 97.30 Sls 2.70 —_ aes B 98 76.53 Wala 16.33 e Cc iLaly/ 94.02 lecal 3.42 0.85 =e D 69 92.75 1.45 4.35 ce. 1.45 E 23 82.61 4.35 13.04 oe at F 55 89.09 1.82 9.09 a we G 37 89.19 = 10.81 sie ETA H 43 86.05 = 13.95 an Es I 108 89.81 1.85 5.56 2.78 J 45 95.56 4,44 pase a: K 3D 94.28 we 2.86 2.86 I, 50 94.00 2.00 4.00 we ate M Al 73.24 8.45 18.31 ate = N 44 86.36 = 13.64 O 2 100.00 = —s fue a 12 16 87.50 6.25 Fe 6.25 Q 1 BZ x 100.00 ts R 9 DPA ON) 2222, 44.45 Wt AL et S 55 60.00 3.64 23.64 5.45 UU 70 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY Only one sample of N. micropus (N. m. planiceps, locality Q) demonstrates a higher percentage than some samples of N. floridana. Because planiceps is known only by the holotype, which has a small fork on the spine, this sample is not necessarily indicative of the condi- tion in the population. With the excep- tion of planiceps, however, percent morphological development of the fork in micropus is consistently below 15, and populations geographically adjacent to floridana show no apparent increase either in the frequency of occurrence of a fork or in size of the fork as compared to those not geographically adjacent to floridana. A fork classified “large” is seen only in one specimen of micropus (KU 69604), but five other specimens from the same _ locality (Comanche County, Kansas) do not have forked spines. Medium-sized forks are uncom- mon in micropus, but all samples repre- - > oS 7 108 96 | | ip ae iT a) | | eo . | ! = | \ lite. 4 } 2p | | © | es eS ts ee --- \ 432| 24-- | i eS a 108 ; a : AR Fic. 11. Geographic variation in morphol- ogy of anterior palatal spine in three species ot Neotoma. See figure 8 for geographic areas represented by each symbol and text for discus- sion of variation and calculations sented by more than 16 specimens have at least one individual with a bifurcated palatal spine. Grouping of samples was not conducive to demonstrating local populational variation; only one series of specimens from a single locality deviates noticeably in frequency of the palatal spine as compared to that of other speci- mens from the aggregate locality. Six of seven micropus (MHP 3377-81, 4634- 35) collected on the same date from 3 mi S and 14 mi W Johnson, Stanton Co., Kansas, have small terminal forks on the anterior palatal spine. Two of five ani- mals from other localities in Stanton County also have small forks. Therefore, although less than 10 percent of the other specimens from locality B have forked spines, 66.7 percent of those from Stanton County demonstrate the char- acter. Considering intraspecific variation in Neotoma_ floridana, samples of both campestris and attwateri from Colorado and Kansas are similar, ranging from near 65 percent to 80 percent develop- ment. Of the three samples from Okla- homa, specimens from locality 10 (south- eastern part of the state) have an aver- age similar to that of samples of floridana from Kansas. Frequency of the fork is noticeably lower in specimens from lo- cality 8, and even when present the forks tend to be relatively smaller. Many of the specimens in this sample originated from Blaine, Dewey, and Major counties, Oklahoma, all of which are near the known area of sympatry and_ natural hybridization of floridana with micropus. It should be reiterated that specimens from the locality of sympatry were not pooled with those from adjacent locali- ties, but were treated separately as a single sample (S). The intermediacy of this sample is evident in figure 11. Intermediacy of sample S was ex- pected. If considered alone, the ten- dency toward intermediacy in sample 8 might lead to the conclusion that hy- bridization of the two species in central Oklahoma is introgressive. This observa- tion must not be disregarded in evalua- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA ae tion of the specific integrity of floridana and micropus. The importance of the reduced frequency and size of forks on the anterior palatal spine of specimens from sample 8 may be even more sig- nificant when considered in view of the high frequency of the fork on specimens from adjacent sample 9. When only spec- imens from Oklahoman localities 8 and 9 and the two adjacent localities for micropus (G and H) are considered, those from locality § appear almost per- fectly intermediate. If these findings are interpreted as being the result of intro- gression, the flow of genetic materials then would appear to be eastward and not westward. However, Key (1968:19) has shown that in some instances hybrid suture zones are in a manner analogous to semipermeable membranes. If so, they probably allow flow of some genes in only one direction, others in the opposite direction, and still others to move in both directions. Lewontin and Birch (1966) have hypothesized, with support- ing evidence, that a species may actually become better fit to expand its distribu- tion as a result of having introgressed certain desirable genes from a closely related species. In neither case discussed did the original hybridizing taxa merge to form a single species. Samples of floridana from Texas dem- onstrate the lowest percent scores of sam- ples of that species. The sample of N. f. rubida (locality 13) shows least develop- ment of the spine and is not geograph- ically adjacent to a population of micro- pus. Localities 11 and 12, however, are geographically adjacent to the range of micropus and the reduction in frequency and development of the fork in these populations could be interpreted as dis- cussed above for specimens from locality 8. On the other hand, sample 8 and the Texas samples may represent no more than a geographic cline toward reduc- tion and loss of the fork. Further inter- pretation must await additional material from Oklahoma and Texas as well as study of this character in eastern popula- tions of floridana. Morphology of the anterior palatal spine in N. angustipalata is highly vari- able (see Hooper, 1953:9-10, for notes on apparent excessive variation within this species), but is more similar to flori- dana than micropus. The presence of a small fork and the deep reentrant angle of MI on the holotype of N. m. planiceps (both characters unlike typical micropus and resembling angustipalata) support my earlier suggestion that rats of these two nominal taxa may actually represent a single taxon. Additional specimens of angustipalata from localities in northern San Luis Potosi and more planiceps from near Rio Verde should elucidate this question. Because the anterior palatal spines of other species of Neotoma are not forked (Finley, 1958:252, reported the presence of a fork on one specimen of N. albigula), it seems logical to assume that the forked condition is derived, and that a pointed spine represents the original or “primi- tive’ grade. A solid vomer is also the exception rather than the rule for Neo- toma. Both characters apparently evolved together in the precursor of the angustipalata-floridana-micropus — com- plex. Subsequently, evolution either has favored the solid vomer equally in all three species or the character was al- ready “fixed” before speciation within the complex occurred. In the case of the forked spine, it appears that at least some members of the precursory species must have possessed the character but selec- tion has favored it more strongly in floridana and angustipalata than in mi- cropus. Alternatively, the frequency of occurrence may have been dispropor- tionately in favor of the direct ancestors of angustipalata and floridana at the time of isolation. Posterior Margin of Bony Palate.— Calculations of percent development of the posterior margin of the palate were conducted in a manner similar to those described for the palatal spine. Deter- mination of sequence for scoring varia- tion in the posterior margin of the palate was more difficult. On the palate, the 72 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY plain or rounded condition of the palatal margin probably is intermediate ( primi- tive?) with the deep notch being one extreme (advanced?) and the large con- vexity being the other (also advanced? ). The decision concerning which extreme to designate one and which to designate eight was arbitrary and does not imply evolutionary direction. Less arbitrary and more troublesome was the decision of scoring condition four (Fig. 10), in which two convexities form a median notch. Variation in morphology of the margin results from the manner in which the right and left halves of the develop- ing hard palate adjoin. When a single projection is present, approximately half of the bone involved was contributed by each side of the palate. Conversely, when an indentation is present, it is be- cause the two halves of the palate fuse anterior to the posterior margin. When two projections are present, each con- sists only of bone originating from a single half of the palate; the resultant “notch” between the projections clearly is homologous with the notch present in the absence of projections. Thus, two projections are homologous to the single projection characteristic of other animals. Because most (if not all) other species of woodrats tend to have either a rounded palatal margin or one with a convexity, is is most parsimonious to conclude that the notch is derived and that the double projection is an evolutionary grade in N. f. baileyi. It may be a condition derived from the single notch seen fre- quently in N. floridana. If the latter is the case, this condition probably should have been scored as one and placed at the left of the series. In any event, it apparently is more closely allied to the indentation than to the single convexity, and therefore, was scored as intermediate between the rounded margin and the smallest indentation. On the average, the sample of Neo- toma micropus canescens (locality E) from New Maxico exhibited greatest de- velopment of the posterior palatal con- vexity. Nearly 75 percent of specimens in that sample have convexities classified as medium or large. Therefore, this sam- ple was established as 100 percent de- velopment of the palatal convexity. The palatal margins of specimens of N. f. rubida (locality 13) show the least ten- dency toward a convexity, with 11 of 12 specimens having an indentation and none having a convexity. Thus, sample 13 is considered zero percent relative to sample E for this character (see Table 8, and Fig. 12). Geographic variation in the morphol- ogy of the posterior margin of the bony palate in micropus is slight, ranging from the established 100 percent in sample E to 75.9 percent for sample G. Intra- specific variation in floridana is greater than for micropus, ranging from the es- tablished zero percent at locality 13 to 53.7 percent at locality 5. In most sam- ples of floridana, percent development of a convexity is between 10 and 30, but in three samples (1, 5, and 11) it exceeds 40 percent. In sample 1 (baileyi), the high frequency (49 percent) of the double convex palatal margin discussed above accounts for the high index of de- velopment, but in samples 5 and 11, the frequency of this morph is not especially high. Both of these populations are char- acterized by a relatively high incidence of animals with rounded palatal margins. Rounded margins are common in labo- ratory hybrids having a floridana parent with an indentation and a micropus par- ent with a convexity. Locality 5, al- though adjacent in the broad sense to localities C and D, is situated geograph- ically so that woodrats from there are not currently in contact with any popula- tion of micropus. If the two species oc- curred together in the vicinity of the Arkansas River at some date in the past (prior to settlement of the area by Euro- pean man), it seems plausible that the intermediacy of population 5 could be the result of previous hybridization be- tween the two species. However, speci- mens from locality 5 demonstrate no morphological or other proclivities to- ward micropus in other characters, and re) 6ST 99'S ce Tl 6E'SP 99°S LS'8T IG El i €¢ S ie OSGI 09°69 00'SG aa =i ate ae 8 u x =? 00°00T aie i ag ne a I O gg’g CO CG 68°SE COSS Ta sails ae, chy ST d S Co SE ~ L9°99 7 “os cag age =x € O fe Sov LG’61 69°8S 6E LI “- = es pe OF N a 919 SO'TE IT8S SOP a = a as VL W = (one 67'SG 06’ 7S 69°ST eae — - ni Tg ‘ll = 98°G LS'8S SZ FS 6S FI ae = as a cE ¥ en 688 SL LE IPL CCGG ‘a is = i cP { EB 9¢°9 60°61 9C OV ST'8G ar . £6 H ES 0001 98°GI G8'GS 98°GS Vi Ra ot o. OL d ze OLE PEEL 89°6E BSCS = = = 961 6) O F6G LG'IG V6 GS LOG oY 86'0 — oa cOT da 7, 68°E1 Pr 6l IT 9€ 00'SG pt 8LG 8LG a 9¢ V a ny Pa LS €e'8 is €e'8S L991 LOOT al €1 Ay aK a COC €S0E pe PISS 0€'6 es eV ral = 2 S 9G IT GP €S' 01 PSOE 96ST —a 61 Il S =~ a eS ol €S'OT 96'S 68°LS 6L'ST iy 61 Or o) a 7% SS 68 Tl “= LE OL WO]tey cy OL's LG 6 Q << = ers 80'S 9G OL CT OP 8e°ST v; 6E 8 < , ae 80°€ EGG I€Gl T€Sg LLOT as g9 L 5 fs ind ak €6 CF 8o°L 88° LE 19 OL i 99 9 Z, ore a 6E°ST £669 69°L 69°L a el S a a “sa LVG Sr'eP re, SV'ePV L801 i.) OV v ww =. a as 66'0€ IV t ST’6S cvs a IL € n pie S; €9'G F686 vS'O1 €9°GS 96 = 8¢ G =z aur _ LL IT IP 66 60'67 VSL 96'T — TS I la (8) (L) (9) (g) (F) (€) (Z) (T) N jets) e) A}IXOAUOD A}IXOAUO9 A}IXOAUO0D pepunoy A}IXO9AUOO uolye}UepUr uoQe}USpUuL UOHe}USpUl A}[ROO'T oO asie'T wnIpayy yeus ul yes umnIpeyy dosoq = uoyeyUuapuy ‘AYI[BIOT pepoo yoRs uryyIM papnyo “Ul Svoie orydeis008 10f g oInSY pue solioge}eo [BoIso;oYdiour Jo SUOHRSNI[E 1OF QT 2INSY vag ‘pzDj)pd1ysnsZuv ‘NY pue ‘sndosonu * NY “puDpiLoyf DUW0J0aN jo sojduies podnois [¢g ul oyeyed Auoq oy} Jo ulsieUr 101I0}Ss0d oY} Jo sotoZe}eo [eorsofoydiour yysre jo 90uaLmMo00 Jo Aouonboiz JUSd19g “8 ATAVL ee 74 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY the sample is relatively small (13 indi- viduals). Probably it is best not to attribute the deviation seen in this pop- ulation to past or present hybridization, but rather to interpret it either as a local population phenomenon or an artifact of the small sample. Interpretation of palatal morphology in sample 11 is more difficult and may represent one result of hybridization of floridana and micropus in north-central Texas. Specimens of this sample also tend toward micropus with respect to the anterior palatal spine discussed above. As mentioned elsewhere, a sam- ple of floridana from southwest of Dallas, Texas, was nearly intermediate in color between the two species (see beyond under discussion of results of discrim- inant function analysis ). Specimens from locality 8, which show a marked _ ten- dency toward micropus in morphology 40 : Sa = : > \ &s oo ta ' J & A ©; hae ES eee \ | a . aera | 24\- { @: 1 | | o sO 150 Fic. 12. Geographic variation in morphol- ogy of the posterior margin of the bony palate in three species of Neotoma. See figure 8 for geographic areas represented by each symbol and text for discussion of variation and calcu- lations. of the anterior palatal spine, do not differ notiecably in morphology of the posterior palatal margin from specimens from lo- calities remote from the range of mi- cropus. Specimens from the locality of sympatry (sample S) are approximately intermediate between the adjacent pop- ulations of the two species in average morphology, but demonstrate a range of variation from medium indentations to large convexities. Increased variation in a hybrid population is, of course, to be expected. With respect to this character, specimens of the species angustipalata are more like micropus than floridana. Sphenopalatine Vacuities—Variation in form and size of the sphenopalatine vacuities was scored and treated in a manner similar to that explained pre- viously. The range of observed variation and values assigned to each category are | shown in figure 10; percent frequency of each category for each sample is shown in table 9, and the relative size of the vacuities for each population is shown geographically in figure 13. On the aver- age, vacuities in specimens from locality | G are largest. That sample was set as_ the standard of 100 percent development. Vacuities of specimens from locality 4 are smallest on the average, and were set at zero percent relative to the vacuities of animals from sample G. The average size of the sphenopala- tine vacuities is markedly different in samples of micropus as compared to that of floridana; samples of the former range from 60.3 percent (N. m. planiceps) to 100 percent (sample G), whereas those of floridana range from zero percent (lo- cality 4) to 50.6 percent for sample 1 (N. f. baileyi). Neotoma angustipalata is highly variable with respect to morphol- ogy of the vacuities; it is clearly flori- dana-like when averages are considered. Specimens from the locality of sympatry (sample S) are intermediate (58.6 per- cent), but on the average nearer micro- pus than floridana. Size of the sphenopalatine vacuities in specimens of baileyi is appreciably larger on the average than those of other WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 75 TABLE 9. Percent frequency of occurrence of six morphological categories of the spheno- palatine vacuities in 31 grouped samples of Neotoma floridana, N. micropus, and N. angusti- palata. See figure 10 for illustrations of morphological categories and figure 8 for geographic areas included within each coded locality. Very Locality Closed Minute Small Medium Large large code N @t) (2) (3) (4) (5) (6) 1 47 = am 44.68 38.30 17.02 2, 38 ae 23.68 44.74 23.68 7.90 3 63 =e 23.81 55.56 15.87 4.76 4 44 oe Tue 27.27 ee 5 12 —_ 2.5.00 66.67 8.33 6 59 1.70 22.03 67.80 8.47 7 56 1.79 41.07 51.78 5.36 8 27 = 18.52 59.26 DIDO, 9 13 = 15.38 84.62 = 10 1 = 8.33 75.00 16.67 11 16 sth 2.5.00 75.00 ae 12 4] am? 36.58 53.66 9.76 13 ify 8.33 33.33 41.67 16.67 = aut A 34 ae, aed 5.88 44.12 41.18 8.82 B 87 aL a2 2.30 29.88 66.67 PS C 78 — = 1.28 12.82 79.49 6.41 D 43 Be oS 2.33 41.86 43.48 230 E 14 a fat 14.29 50.00 28.57 7.14 F 16 ee a5 12.50 56.25 lb is G 14 ae eee ae 21.43 42.86 Soe H 30 _— ae: 10.00 23°35 63.34 3.33 I 2 =ae exh = 50.00 50.00 bea J 18 Ps mate 5.56 50.00 38.88 5.56 K 19 = a = 36.84 42.11 21.05 IG, 10 «<< ”? . . . . . Sl” and “S2” represent samples of micropus-like and floridana-like woodrats, respectively, from the same locality; “6 X D” is a sample of laboratory bred F; hybrids whose parents were from localities 6 and D. 78 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY The bacula of four adult F, hybrids between micropus from locality D and floridana from locality 6 also were mea- sured, averaged, and plotted in the dia- gram. This sample was plotted in a posi- tion nearly intermediate between the parental populations, but more like flori- dana if all localities are considered. The intermediacy in the bacula of hybrids demonstrates the assumed genetic basis for determination of the shape of the bone. Because woodrats from the locality of sympatry, 3 mi S Chester, Major Co., Oklahoma, were identified by charac- ters of the skin and skull, but not the baculum, the three specimens most like floridana (S1) were averaged indepen- dently from the five most like micropus (S2). When plotted as above, the S1 sample appears to be typical of floridana and gives no indication of effects of hy- bridization. Sample S2 plots in a posi- tion marginal to other samples of mi- cropus, but clearly more like micropus than floridana. Bacular measurements of woodrats from other localities do not indicate any obvious intraspecific or geo- graphic trends nor do they seem to be noticeably correlated to any of the other measurements and characters studied. In most instances the baculum of floridana can be distinguished from that of micro- pus or angustipalata, but no completely diagnostic differences exist in the mor- phology or size of the bone. Univariate Analyses of Mensural Characters The four external and ten cranial measurements described previously were analyzed by Powers’ UNIVAR Program in univariate assessment of geographic variation. This was done separately for each sex and each species, then sepa- rately for each sex with samples of both species treated simultaneously. These re- sults are discussed below, in a considera- tion of general trends of geographic vari- ation. A more detailed evaluation that includes discussion of each measurement and subset relationships of group-means as determined by SS-STP (as included exemplarily below for total length) may be found elsewhere (Birney, 1970). Neotoma floridana.—Standard statis- tics computed on measurements of spec- imens from the 13 pooled localities of Neotoma floridana are shown in table 10. Interlocality variation in the measure- ments of total length is significantly (P <0.01) different for both sexes. For males, localities are separated into two broadly overlapping, non-significant sub- sets (hereafter called subsets) indicating that animals from southeastern Texas (N. f. rubida, locality 13) and from south- eastern Kansas (locality 7) are signifi- cantly larger than those from localities 6 (northeastern Kansas) and 9 (north- eastern Oklahoma). Males from locality 12 (southern Texas), adjacent on the west to locality 13, also are large. Males from sample 7 are significantly larger than those from both adjacent localities (6 and 9), but not significantly larger than those from non-adjacent localities. In females, variation in total length is similar to that for males, but group- means separated into five subjects with females from locality 5 being. signifi- cantly larger than those from localities 8, 6, 10, and 9 (hereafter written as locality 5 > 8, 6, 10, 9), locality 13 > 6, 10, 9 and locality 12 >9. The only ad- jacent localities showing significant dif- ferences are 5 and 6, the two samples from north-central and northeastern Kansas. Hind foot length in both sexes is greatest in N. f. campestris and the sam- ple of N. f. attwateri (5) that is geo- graphically contiguous with campestris. Ear length is not significantly different for group-means in either sex. Coeffi- cients of variation for external measure- ments are high compared to those for cranial measurements. Although analy- ses of these characters are indicative of total size relationships, they probably are less reliable than analyses of cranial dimensions. Differences that may ac- tually exist in external size are difficult to document statistically because unre- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 79 TABLE 10. Geographic variation in 14 external and cranial measurements of Neotoma floridana from Nebraska, Colorado, Kansas, Oklahoma, and Texas. See figure 8 for geographic areas included within each coded locality. Locality Males Females Code N Mean + 2SE Range N Mean = 2SE Range Total length 1 7 381.3 10.04 (361.0-398.0 ) 9 3144 9.88 (350.0-393.0 ) 2. 9 379:6 11.04 (350.0-398.0 ) 8 365.8 9.40 (344,.0-382.0 ) 3 OR oSpel: 10.25 (341.0-408.0 ) 19 376.4 6b (349.0-409.0 ) 4 9 379.4 18.10 (345.0-434.0 ) IAS «63632 10.65 (340.0-402.0 ) 5 Ie 386.0) 21-00 (376.0-397.0 ) 2 39720) 9 26:00 (384.0-410.0) 6 Dee oped 160 ( 342..0-370.0 ) 8 354.0 14.83 (329.0-395.0 ) T 14 3943 14.22 (349.0-450.0 ) ee oles 9.41 (340.0-397.0 ) 8 15 380:0 11.99 (350.0-425.0 ) OD Sbrelt 10.87 (334.0-379.0) 9 20%) 359!5 Toth) (323.0-397.0 ) Mie S493 2205 (308.0-392.0 ) 10 One O 11.29 (334.0-400.0 ) ey 352 9.82 (320.0-374.0 ) 11 8 369.4 20.56 (328.0-420.0 ) 8 374.1 19.55 (322.0-340.0 ) 119? 7 388.0 25.61 (317.0-414.0) 8 384.6 11.96 (356.0-412.0 ) 13 4 407.0 18.20 (387.0-430.0 ) bre SOCOM 2ik3s (368.0-442.0 ) Length of tail vertebrae 1 7 159:7 1020 (138.0-176.0) 9 161.7 9.00 (136.0-180.0) Seo 15500 S716 (13001670) 8 1572 481 (146.0-168.0) 3 12 1613 894 (120.0-178.0) 19 1585 4.06 (144.0-175.0) eG Isai Got (130051740) 14) 1490) 55406 s6.0-1720) Be) 1505" 29:00! (14510-1740) 2 1645 17.00 (156.0-173.0) 6 5 1486 6.65 (139.0-156.0) 8 1508 616 (142.0-169.0) 7 914 1629. 3:74 (153.0-175.0) 12) 1615 439)) (148.0-17000) 8 15 ~=160.7 6.79 (149.0-185.0 ) 9 154.4 6.07 (137.0-165.0 ) 9 YAN) le3ayys: 5.39 (132.0-176.0 ) 16 = 150.3 5.02 (136.0-170.0 ) 10 122 162.2 5.30 (143.0-172.0) 13 151.0 4.59 (138.0-163.0 ) lat: 9 WES 7/ 13.65 (130.0-195.0) 8 162.9 9.01 (138.0-174.0) 12 1 168.4 9.08 (147.0-181.0) 8 See, 6.48 (156.0-185.0 ) 13 on LAO 21563 (159.0-195.0 ) 5 188.2 11.30 (175.0-207.0 ) Length of hind foot 8 39.8 0.60 (38.0-41.0) iE 39.1 0.51 (38.0-41.0) 10 41.8 1.26 (39.0-44.0 ) 8 39.5 0.65 (38.0-41.0) ) 19 39.8 0.77 (36.0-42.0 ) ) 15 38.7 0.67 (36.0-40.0 ) ) 2 42.0 0.00 ( 42.0-42.0) ) 10 36.8 1.36 (34.0-40.0 ) 0) fat 39.3 0.90 ( 37.0-42.0) 14 39.5 1.75 (36.0-49.0 ) 10 37.5 1.31 (35.0-42.0 ) 13 37.2 0.82 (35.0-40.0 ) 13 38.2 1.14 (35. ) 8 38.5 1.65 (35.0-42.0 ) ( ) ( 12 1 39.1 1.41 (37.0-42.0 11 39.3 1.01 36.0-42.0 13 4 38.8 1.50 (38.0-41.0 4 38.0 2.58 35.0-41.0 Length of ear 1 2 27.5 3.00 (26.0-29.0) 5 26.6 1.20 (25.0-28.0) 2 4 27.8 Ne7Al (26.0-30.0 ) 6 26.0 0.89 (25.0-28.0 ) 3 6 28.8 1.89 (26.0-33.0) 9 28.7 1.05 (27.0-32.0 ) 4 10 28.0 1.26 (25.0-31.0) 10 28.0 0.99 (25.0-30.0 ) 5 3 26.3 2.40 (24.0-28.0) 2, 29.0 0.00 (29.0-29.0 ) 6 6 26.7 1.33 (25.0-29.0) U 25.9 1.54 (23.0-28.0 ) ii Ili! 27.0 1.14 (25.0-30.0 ) 8 28.2 2.95 (25.0-38.0 ) 8 11 Da 0.66 (26.0-30.0 ) 4 26.2 2.06 (24.0-29.0) 9 16 27.9 0.97 (25.0-32.0) 11 27.4 1.91 (25.0-36.0 ) 10 10 27.2 1.07 (25.0-30.0 ) 1? 26.8 0.96 (25.0-29.0 ) 11 8 28.5 1.00 (27.0-30.0 ) 5 27.0 3.03 (21.0-29.0 ) MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 10.—Continued. Locality Males Females Code N Mean + 2SE Range N Mean + 2SE Range 12 3 lait 5.30 (25.0-33.0) 3 26.7 3.33 (25.0-30.0) 13 3 oe 2.91 (29.0-34.0) 3 30.3 2.91 (28.0-33.0) Greatest length’ of skull il 7 48.4 een (46.5-50.7 ) 11 48.8 0.44 (47.5-49.7 ) 2 7 49.6 2.06 ( 46.0-53.8 ) 6 48.9 0.87 (47.5-50.6 ) 3 12 Sle 0.81 (48.9-53.4 ) 18 49.7 0.72 (47.0-53.3 ) 4 9 51.4 1.41 ( 48.4-55.2 ) 1k 49.7 0.74 (48.3-52.1) 5 3 Bes 2.66 (49.1-53.7) 2 52 1.20 (50.6-51.8) 6 5 50.1 1.58 (48.0-51.7) 10 48.6 0.94 (47.0-52.2) i IN} 50.8 0.91 (47.4-53.5) 11 50.2 0.79 ( 48.6-52.1 ) 8 15 50.5 Pats (47.1-54.0) 8 48.6 1.00 ( 46.8-50.7 ) 9 16 50.2 0.82 (47.1-52.9) 14 48.2 0.84 (45.3-51.4) 10 13 50.2 0.69 (48.2-52.4) 13 48.3 1.01 (45.0-50.8 ) at 9 49.6 ers (44.4-53.0) 8 50.7 0.70 (49.5-51.9) 12 9 51.6 1.09 (49.8-54.0) 8 50.4 EOATE (47.9-53.2) 13 5 53.0 1.39 (50.7-54.5 ) 4 pA 1ST (51.0-54.1) Condylobasilar length 1 7 47.4 Lees, (45.7-49.8 ) 11 47.4 0.55 (46.0-48.9 ) 2; 9 48.6 1.71 (44.5-52.9 ) 1 47.6 0.77 (46.4-48.9 ) 3 ils} 49.5 0.89 (47.3-52.3 ) 14 47.8 0.73 (45.2-50.9 ) 4 9 50.6 1.47 (47.6-54.3) 12 48.6 0.81 (47.4-51.6) 5 3 49.3 3.27 (47.7-52.6) 2 50.1 1.20 (49.5-50.7) 6 5 49.3 1.61 (47.1-51.1) 10 47.2 0.99 (45.6-51.2) a 15 49.5 0.99 (46.1-52.2) 11 49.0 0.93 (47.3-51.7) 8 16 49.2 1.15 (45.2-53.2 ) 8 47.2 0.93 (44.9-48.8 ) 9 19 48.8 0.81 (45.5-52.4) 13 46.9 0.87 (50.0-54.0) 10 13 48.9 0.74 (47.1-51.0) 11 46.7 1.26 (42.9-49.3 ) ll 10 48.0 1.50 (43.2-50.9) 8 49.0 0.72 (47.6-50.5 ) 12 9 50.1 1.29 (48.1-52.4) 10 48.4 1.04 (46.3-51.2) 13 5 50.6 1.63 (47.4-52.0) 5 49.4 1.98 (46.4-52.6) Zy gomatic breadth il of 25.9 0.61 (24.8-27.1) 1131 26.1 0.27 (25.4-26.6) 2 8 26.9 0.72 (25.4-28.6) Ml 26.4 0.49 (25.2-27.0) 3 12 27.4 2.14 (26.2-28.3 ) 15 26.8 0.44 (25.1-28.0) 4 10 27.8 1.02 (25.5-29.9 ) 12 27.1 0.67 (25.9-29.6 ) 5 2 27.4 2.00 ( 26.4-28.4 ) 2 27.4 0.70 (27.1-27.8) 6 6 27.0 1.03 (25.7-28.9 ) 10 26.5 0.51 (25.6-28.3 ) a 13 27.9 0.53 (25.9-29.2 ) 12 27e3 0.49 (26:2-29.1 ) 8 16 26.9 0.68 (24.4-29.0) 6 26.6 0.91 (25.4-28.1) 9 18 26.9 0.51 (24.4-28.9) 15 26.1 0.58 (24.0-27.9) 10 3 Dil 0.55 (25.2-28.5) 14 26.1 0.51 (23.3-27.1) ll 9 Dial: 1.20 ( 22.8-29.3) 8 27.6 0.54 (26.1-28.3) 12 8 27.0 0.86 (25.5-28.7 ) 10 26.9 0.63 (25.5-29.2 ) 13 4 PATE 0.80 ( 26.6-28.5 ) 5 26.8 LS (25.5-28.4) Least interorbital constriction 1 9 6.9 1.70 (6.6-7.4 ) 11 6.6 0.15 (6.3-7.0) 2: 10 6.9 0.24 (6.4-7.7) 8 6.6 0.17 (6.2-6.9 ) 3 14 7.0 0.15 (6.4-7.5) 19 6.8 0.16 (6.1-7.5) 4 14 6.7 0.18 (6.4-7.0) 13 6.6 0.13 (6.2-6.9 ) 5 3 6.8 0.07 (6.8-6.9 ) 2 6.6 0.10 (6.5-6.6) 6 6 6.6 0.16 (6.4-7.0 ) 11 6.5 0.15 (6.2-6.9 ) of 16 6.7 0.22 (6.0-7.8 ) 12 6.6 0.21 (6.1-7.2) 8 18 6.8 0.18 (6.2-7.6 ) 10 6.7 0.24 (6.1-7.2) 9 22, 6.7 0.14 (5.8-7.2 ) 17 6.5 0.15 (6.1-7.3) 10 15 6.9 0.16 (6.5-7.5 ) 13 6.6 0.14 (6.2-7.2 ) WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA TABLE 10.—Continued. Locality Males Females Code N Mean + 2SE Range N Mean =+ 2SE Range 11 10 6.7 0.26 (6. 0-7 2) 9 6.9 0.16 (6.5-7.2') 12 9 6.8 0.23 ie 9-7.5) ial 7.0 0.24 (6.4-7.7) 13 5 7.0 0.19 (6.8-7.3 ) i 6.9 0.24 (6.7-7.3) Breadth at mastoids 1 8 19.0 0.51 (18.0-20.4 ) 11 19.0 0.27 (18.2-19.6) 2 8 19.2 0.83 (17.4-20.7 ) 7 18.9 0.37 (18.1-19.3) 3 13 19.7 0.30 (18.7-20.5 ) 15 19.2 0.30 (18.1-20.2 ) 4 11 20.3 0.63 (18.5-22.4) 14 19.7 0.34 (18.7-21.0) 5 2 19.8 0.70 (19.5-20.2 ) 2 2.0.0 0.50 (19.7-20.2 ) 6 5 19.5 0.76 (18.5-20.6 ) 11 19.0 0.44 (17.8-20.1 ) fl 15 20.0 0.55 (16.9-21.0) 12, 19.4 0.42 (18.2-20.4 ) 8 18 19.6 0.31 (18.2-20.7 ) 8 19.2 0.41 (18.5-20.2 ) 9 18 19.3 0.22 (18.7-20.2 ) 14 18.8 0.39 (16.5-19.6) 10 15 19.8 0.31 (19.0-20.9 ) 13 19.0 0.28 (18.1-19.6) iil 8 19.6 0.72 (17.4-20.6 ) 8 19.8 0.18 (19.4-20.1 ) 12 8 20.0 0.49 (19.0-21.0) 10 19.8 0.51 (18.8-21.6) 13 5 20.5 0.68 (19.5-21.6) 5 19.4 0.55 (18.5-20.0 ) Length of rostrum 1 8 18.9 0.43 (17.8-19.6) iL 18.8 0.32 (17.5-19.4) 2 8 19.5 0.91 (18.0-21.9 ) 8 18.9 0.65 (17.3-20.0) 3 14 19.9 0.41 (18.7-21.2) 19 19.3 0.39 (Give8-21) 4 12 20.3 0.69 (17.9-22.9) 14 19.4 0.28 (18.7-20.5) 5 3 20.1 1.10 (le) pee leil)) 9} 20.6 0.30 (20.4-20.7 ) 6 6 19.2 0.99 (17.7-20.5) 10 18.8 0.40 (18.1-20.0) ih 16 19.8 0.46 (18.1-21.5) 11 19.5 0.39 (18.6-20.6 ) 8 17 20.0 0.49 (18.4-21.6) 9 19.1 0.51 (18.2-20.4) 9 19 19.6 0.42 (18.2-21.1) 7 18.7 0.35 (17.3-19.9) 10 15 19.5 0.36 (18.2-20.4 ) 13 18.6 0.50 (16.4-19.5) 11 9 19.3 0.81 (17.0-21.0) fe) 19.9 0.43 (19.0-20.6 ) 12 fe) 20.3 0.58 (19.1-21.7) 8 19.7 0.50 (18.9-20.7 ) 13 5 ED 0.61 (20.1-21.9) 4 20.4 0.34 (19.9-20.7 ) Breadth of rostrum 1 9 7.9 0.17 (7.5-8.2 ) 18 7.9 0.13 (7.5-8.2) 2 9 8.2 0.58 (6.1-9.0) 8 8.1 0.16 (7.7-8.4) 3 15 8.7 0.15 (8.3-9.2 ) 17 8.4 0.18 (7.9-9.4) 4 12 8.5 0.36 (7.6-9.3 ) 15 8.5 0.14 (7.9-9.0) 5 3 8.3 0.35 (8.0-8.6 ) 2} 8.5 0.00 (8.5-8.5) 6 6 8.1 0.36 C6=87)) 10 8.0 0.15 (7.7-8.3 ) 7 15 8.4 0.22 (7.5-9.1) 11 8.2 0.14 (7.8-8.5) 8 18 8.2 0.21 (7.6-9.0) 10 8.0 0.20 (7.5-8.5) 9 22, 8.0 0.15 (TA-8.7 ) ef 7.8 0.14 (7.3-8.2) 10 15 7.9 0.17 (7.3-8.5) 13 7.8 0.17 (7.4-8.3) 11 10 8.0 0.24 (7.1-8.4) 9 8.2 0.21 (EES) 12 8 8.2 0.31 (7.5-8.7) 10 8.4 0.33 (E5292) 13 5 8.2 0.34 (7.8-8.6 ) 5 8.3 0.27 (7.9-8.7) Length of maxillary toothrow 1 9 9.5 0.20 (9.2-10.0) 11 9.4 0.17 (8.8-9.9) 2 10 9.3 0.23 (8.9-10.0) 8 9.4 0.38 (8.9-10.5) 3 15 9.9 0.18 (9.4-10.4) 19 9.7 0.12 (9.2-10.0) 4 12 9.6 0.19 (9.1-10.0) 15 9.7 0.16 (9.1-10.3) 5 3 9.6 0.64 (9.0-10.0) 2 9.8 0.40 (9.6-10.0) 6 6 9.7 0.28 (9.3-10.0) 11 9.3 0.21 (8.8-9.9 ) i 16 9.5 0.18 (9.0-10.2 ) 12, 9.4 0.23 (8.7-10.1) 8 19 9.6 0.15 (8.7-10.2 ) 10 9.4 0.26 (8.9-10.1) 9 22 9.3 0.15 (8.8-9.9 ) 17 9.2 0.14 (8.7-9.6 ) 81 82 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 10.—Concluded. Locality Males Females Code N Mean 2SE Range N Mean =+ 2SE Range 10 15 9.2 0.20 (8.5-9.8 ) 14 9.4 0.14 (9.0-9.9 ) 11 8 9.4 0.26 (8.8-9.9 ) 9 9.4 0.24 (8.9-10.0) 1 9 9.0 0.28 (8.3-9.4) 11 9.1 0.22 (8.4-9.7 ) 13 5 9.6 0.19 (9.4-9.9) 5 9.4 0.43 (8.8-9.9) Length of palatal bridge 1 9 8.7 0.33 (8.2-9.6) 11 8.7 0.34 (7.3-9.2 ) 2 9 8.5 0.30 (7.6-9.0) 8 8.1 0.28 (7.5-8.7) 3 15 8.8 0.23 (7.9-9.6) 19 8.4 0.15 (7.8-9.0) 4 12 8.9 0.37 (7.9-9.8 ) 15 8.6 0.24 (8.0-9.4) 5 3 8.7 0.74 (8.2-9.4) 2, 8.8 0.40 (8.6-9.0) 6 6 8.5 0.35 (7.9-9.0) 11 8.4 0.27 (7.6-9.0) a 16 8.8 0.20 (8.1-9.6) 12 8.7 0.16 (8.3-9.3 ) 8 18 8.6 0.32 (7.2-9.6) 10 8.3 0.18 (7.8-8.7) 9 22, 8.5 0.23 (7.6-9.7 ) 16 8.2 0.21 (7.5-9.0) 10 15 8.3 0.12 (7.9-8.8 ) 14 8.5 0.21 (7.6-9.0) il 10 8.7 0.39 (7.8-9.5) 9 8.7 0.32 (8.1-9.5) 12 9 8.5 0.36 (7.7-9.4) 10 8.4 0.45 (7.3-9.8 ) lig 5 8.7 0.15 (8.6-9.0 ) 5 8.7 0.61 (7.9-9.7) Length of nasals i 8 18.7 0.43 (17.6-19.4) 11 18.7 0.33 (18.0-20.2 ) 2, 8 19.3 0.89 (17.7-20.9) 8 19.0 0.48 (18.0-20.3 ) 3 14 19.5 0.45 (18.3-20.9 19 19.0 0.37 (17.6-20.3) 4 12 20.3 0.71 (18.5-23.3 14 19.4 0.34 (18.5-20.4 ) 5 3 20.5 ESS (19.3-21.6 2 20.6 0.50 (20.3-20.8 ) 6 6 19.4 0.70 (18.5-20.7 9 18.7 0.40 (17.8-19.5) U 15 19.9 0.42 (18.0-21.6 11 19.5 0.47 (18.6-21.3) 8 17) 20.0 0.51 (17.7-21.5 9 19.3 0.41 (18.3-20.5 ) 9 19 19.8 0.49 (17.8-21.6 es 18.7 0.42 (16.7-20.2 ) 10 15 19.7 0.47 (18.2-20.9 14 18.8 0.55 (16.5-19.8) 11 9 19.4 0.99 (17.3-21.9 9 20.1 0.49 (18.9-21.1) 12 9 20.5 0.68 (19.2-22.1 8 19.9 0.47 (18.8-20.7 ) 13 5 Daler 0.94 (20.4-23.3 4 PAI 0.73 (20.2-22.0) lated factors contribute to within-group variation. Thus, trends in external char- acters are not shown graphically, al- though some of the observed trends were not evinced by studies of cranial di- mensions. Standard statistics and geographic variation in greatest length of skull are illustrated as a graphic example of vari- ation in a longitudinal cranial measure- ment (Fig. 15, males; Fig. 16, females). Results of SS-STP tests for this measure- ment for both sexes are shown in table 11. Table 12 illustrates observed results of samples of floridana and micropus tested together. Also included are com- parisons of subset relationships involving additional variance between groups, and results of SS-STP testing for greatest length of skull for both species treated simultaneously. Comparisons of males show that spec- imens from locality 11 are the smallest end of a gradual north to south cline in greatest length of skull. Females from locality 11 are noticeably larger than those from localities immediately to the north and more nearly equivalent in size to samples 12 and 13 from farther south and east in Texas. Otherwise, males and females are more or less similar with ex- pected minor shifts in sequence of means. Condylobasilar length was expected to differ little from greatest length of skull because both are measurements of the long axis of the skull. However, a remarkable amount of shifting in se- quence of means is evident between the WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 83 TABLE 11. Results of SS-STP tests comparing means of greatest length of skull for males and females separately from 13 localities of Neotoma floridana and 14 localities for N. micropus. See figure 8 for geographic areas included within each coded locality. | | Males Females | Maximal Maximal | Locality non-significant Locality non-significant code Mean subsets code Mean subsets Neotoma floridana 13 53.0 I 13 52.3 I 12 51.6 erat 5 lee, ae) 4 51.4 Tee 11 50.7 | i | 5 Dies Jia 1 50.4 J foveas) (gE 3 piel eel ra 50.2 1G) Be Cas ri 50.8 | Gare 4 49.7 Ganley cic} See 8 50.5 Tipe 3 49.7 lige ee eo Ee 9 50.2 eal 2, 48.9 |e Si I ee 10 50.2 1a | il 48.8 Ie oN Tey 6 50.1 ge 8 48.6 |e) Bae (iL 2 49.6 a 6 48.6 ) ieee) Ce 11 49.6 iran 10 48.3 el il 48.4 I ) 48.2 I Neotoma micropus Li 51.0 I D 49.7 If H 50.6 ees | © 48.8 | oe | D 50.5 |G 6 | B 48.8 or C 49.7 1d Wiles Cale | it 48.6 | gga hoes G 49.6 | gel Game (uses) igh i | I 48.5 | ued Bnd | A 49.4 To gk Serie G 48.3 | Fi ge i B 49.2 1 ee nal ge aa F 47.7 a te “se er K 48.7 1 ed A Dae | H 47.6 Ei Ts Sit I 48.3 [foe 2 a | K AT.1 1 AGI a aa F 48.1 ort ht M 46.7 1 a (Pf ij 47.9 | Ee) be i Le A 46.5 Lr N 46.8 ler J 46.2 Li P 46.6 It P 45.8 I M 46.6 I N 45.7 I itwo characters. Also, no significant dif- ferences were detected in group-means of males for condylobasilar length. No significant differences exist among group-means for zygomatic breadth of males. Although specimens of rubida (13) are larger, on the average, than in- dividuals from more northerly samples of the species in the measurements of longitudinal axis of the skull, they are narrower in zygomatic breadth than males from localities 7 and 4 (Fig. 17). However, the differences are not sig- nificant. Highly significant (P< 0.01) differences were detected in zygomatic breadth of females (Fig. 18). Least interorbital constriction is one of the least variable measurements con- sidered for N. floridana. Group-means are not significantly different for males, and sequence of means for this character (Fig. 19, males; Fig. 20, females) does not follow that seen in other measure- ments for either sex. The pattern of geographic variation in mastoidal breadth closely approximates that seen for zygomatic breadth. Woodrats from localities 3, 4, and 5 (the two samples of campestris from northwestern Kansas and the sample of attwateri from adjacent north-central Kansas) are generally larger in breadth of rostrum than are woodrats from south- em Texas (12 and 13). In measure- ments of length, specimens from locality 13 are consistently larger than specimens from northern Kansas. It is visually per- ceptible that the skulls of specimens of 84 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY campestris tend to be broader, more heavily structured and robust than those of rubida. To a lesser extent the same differences prevail between campestris and attwateri with the exception of spec- imens from locality 5. The latter are like other samples of attwateri in color, but in several aspects of overall size and general shape of the skull, they more closely re- semble campestris. Patterns of variation for alveolar length of the maxillary toothrow are similar to those of rostral breadth, but unique in certain aspects. Notably, spec- imens from locality 12 are smallest both for males and females, whereas those from localities 1 (baileyi) and 6 (north- eastern Kansas), which are among the smallest in most dimensions, are rela- tively much larger. Results of computa- tions for this measurement also are some- what unique in that more total variation exists between means of males than be- tween those of females. Of the dimensions analyzed, palatal bridge length demonstrates the least amount of geographic variation. Group- means are not significantly different for males and are significant only at the 0.05 level for females. Coefficients of variation for measurements of this char- acter are noticeably larger than for other cranial dimensions. The reduction in significant geographic variation in length of palatal bridge may be a result of high within-group variation. Geographic variation in nasal length exceeds that for other cranial dimen- sions. Geographic trends for this de- mension are similar to those for length of rostrum. These two measurements in- clude the same region of the skull, but TABLE 12. Results of SS-STP tests comparing means of greatest length of skull for males and females treated separately and localities of Neotoma floridana and N. micropus tested simul- taneously. See figure 8 for geographic areas included within each coded locality. Males Maximal non-significant subsets Locality code Mean 13 53.0 12 51.6 51.4 51.3 51.1 51.0 50.8 50.6 50.5 50.5 50.2 50.2 50.1 49.7 49.6 49.6 49.6 49.4 49.2 48.7 48.4 48.3 48.1 47.9 46.8 46.6 46.6 OGMarwourw — QDADMSOCH = OO me en Al cee cen OL eon fl ce Bl cern ce ee ee ee ee ee en fl seen fll cen OL een ce ee es ce Oe en ee ee ee ee OO Ln A ee cn Oe Oc ce cee ce eB ee en oe = eo Ae 5 ee oo ae We) ee OO Ln ll een eee I ces fl oe BE en ce en ce en cee ee Bl | re — Females Maximal non-significant subsets Locality code Mean 13 52.3 5 51.2 a 50.7 50.4 50.2 49.7 49.7 49.7 48.9 48.8 48.8 48.8 48.6 48.6 48.6 48.5 48.3 48.3 48.2 47.7 47.6 47.1 46.7 46.5 46.2 45.8 45.7 ZrmourZSartnoSoQnoaltortwOnwJarab Le A cee Hl seen Nl eee Hl eee en een eee ae ae Hl ee ee | me OS OO me OO Ln A ee A eee ec FN ee ee en cen BO en Od ce Bd ee en | OR me OO OR ee |e A cee cn ces ce Be Oe ce ce ce Bl | Le ee ee Oe Oe ee | WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 85 differ in that rostral length is in part dependent on shape of the anterior end of the zygomatic arch. When all measurements are con- sidered, there is general agreement be- tween the patterns of variation as indi- cated by separate analyses of males and females except for specimens from lo- cality 11 (northern Texas). Males from that locality are smaller (as determined by rank-order scoring of means for the 14 characters) than samples of males from all localities except northeastern Oklahoma (9). Rank-order scores of males from localities 9, 11, 6, and 1 were so close as to be indistinguishable by this method of analysis. In comparisons of females, however, specimens from lo- cality 11 were surpassed in total size only by those from samples 5 and 13. When the rank-order scores for males and females are summed, a general trend of geographic variation for floridana in the Central Great Plains can be seen. Neotoma floridana rubida is slightly larger than any of the samples of attwa- teri or campestris, and appreciably larger than baileyi. Specimens of campestris from localities in Kansas are larger than those from Colorado and larger than specimens of attwateri from all except adjacent localities in north-central Kan- sas. Within the subspecies attwateri, there is appreciable variation in size, but this variation does not follow expected trends. Specimens from north-central Kansas (5) are largest, followed in se- quence by those from southern Texas (12), southeastern Kansas (7), and west- ern Oklahoma (8). Specimens from east- ern Oklahoma (9 and 10) and northeast- ern Kansas (6) are the smallest examples of the subspecies. Neotoma floridana baileyi is slightly larger than the smaller representatives of attwateri, but smaller than the larger representatives of that subspecies. The only contiguous locali- ties from which samples of specimens frequently are significantly different in size are those in north-central Kansas (5) and northeastern Kansas (6). How- ever, on the basis of color, specimens from these localities are similar and indi- viduals in both samples are significantly darker than specimens of campestris. Neotoma micropus and N. angusti- palata—In univariate analysis of geo- graphic variation in Neotoma micropus and N. angustipalata, specimens from lo- calities O (White Sands National Monu- ment, New Mexico) and Q (Rio Verde, San Luis Potosi) were not included in UNIVAR computations because only single specimens were available. Sample R (N. angustipalata) was not included because only one adult specimen of each sex was available at the time UNIVAR analyses were conducted. Subsequently, two additional adult females were exam- ined and included with calculations of standard statistics shown in table 13. Sample E (N. m. canescens from New Mexico ) also was omitted from UNIVAR analyses because New Mexico was not included in the study area at the time computations were conducted. Standard statistics of this sample are shown in table 13. In all four instances, sequence of means were considered for rank-order analysis of trends in size variation, but significance levels are not known for these four samples. Considering the remaining 14 sam- ples of N. micropus, differences in group- means of total length are not significant for males, but they are significant (P< 0.05) for females. Specimens from lo- calities N and P (coastal Tamaulipas) are relatively large in this character, whereas those from adjacent Coahuila and Nuevo Leon are small. Specimens of both sexes from localities P and N (N. m. micropus) have the longest tails of any samples of the species. This is in marked disagreement with trends in cra- nial measurements, wherein specimens of N. m. micropus are among the smallest. Furthermore, they exhibit a marked ten- dency toward being unicolored, espe- cially in southern parts of Tamaulipas. Although somewhat variable and less reliable than cranial measurements, ex- ternal dimensions demonstrate certain overall trends. The two samples com- 86 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY _ o0o”MWO WON AO WH & WwW DN Localities zx onmooo p ir Om Om aa see) ao 42:5 43:5 44°55 45:5 46:5 «47:5 48.5 949'5 50:5 51/5° 52°55 953°5)54'5)) (5535 mm Fic. 15. Dice-grams illustrating geographic variation in greatest length of skull of male Neotoma angustipalata, N. floridana, and N. micropus. The upper point of the triangle is the arithmetic mean; the darkened bar is plus and minus two standard errors of the mean; the open bar is plus and minus one standard deviation of the mean, and the horizontal or vertical lines indicate the range. See figure 8 for geographic areas included within the coded localities indicated on the ordinate. WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 87 Localities zon i moo0onw »,p 0 ON OO WN RB W bw = > VZz Ss rFexKe | 43.0 440 450 460 470 48.0 49.0 50.0 510 52.0 53.0 54.0 mm Fic. 16. Dice-grams illustrating geographic variation in greatest length of skull of female Neo- toma angustipalata, N. floridana, and N. micropus. See figure 15 for explanation of symbols and figure 8 for geographic areas included within the coded localities indicated on the ordinate. 88 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY oO ON Ou fF WD = Localities zo Tem oC 1 = 7 oO vo 2 2 =x ol 230 23:5 240 24/55 250 255: 26:0 26:5. 270: 275 28:0 28:5 2910 (2915 mm Fic. 17. Dice-grams illustrating geographic variation in zygomatic breadth of male Neotoma angustipalata, N. floridana, and N. micropus. See figure 15 for explanations of symbols and figure 8 for geographic areas included within the coded localities indicated on the ordinate. Localities WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 89 Oo WO NOW fF W PD = 23:0) 23:5) 24:05) °24:5, 25.0, (25:55 «26:0 26:5 «270 27:5. 28:0 28:5 29:0 295 mm Fic. 18. Dice-grams illustrating geographic variation in zygomatic breadth of female Neotoma angustipalata, N. floridana, and N. micropus. See figure 15 for explanation of symbols and figure 8 for geographic areas included within the coded localities indicated on the ordinate. 90 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY Localities =- ronrnmooiow ep 0 ©) <7 OF Ul wh) ya OwveozezksrxK «= 5.5 5:7 59 6.1 6.3 6.5 6.7 69 TA 7.3 75 7.7 719 im m Fic. 19. Dice-grams illustrating geographic variation in least interorbital constriction of male Neotoma angustipalata, N. floridana, and N. micropus. See figure 15 for explanation of symbols and figure 8 for geographic areas included within the coded localities indicated on the ordinate. WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 91 — o OnN OD ON BW bw Localities 2B) Dwi 5.9 6.1 6.3 6.5 6.7 6.9 71 3 1) tll mm Fic. 20. Dice-grams illustrating geographic variation in least interorbital constriction of female Neotoma angustipalata, N. floridana, and N. micropus. See figure 15 for explanation of symbols and figure 8 for geographic areas included within the coded localities indicated on the ordinate. 92 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY posed of largest individuals (exclusive of the tail) tend to be L and D. Locality L is composed of specimens from along the Gulf Coast of Texas and D represents specimens from south-central Kansas. Specimens from localities N and P (N. m. micropus) have noticeably longer tails than even the samples of N. m. canescens characterized by large size. Specimens from southwestern Texas, in the general region of the Big Bend area (J), appear smallest in external characters. Geographic variation in greatest length of skull is shown graphically for Neotoma micropus in figures 15 (males) and 16 (females). Results of SS-STP tests illustrating maximally connected non-significant subsets for this dimension are given in table 11 for tests conducted only on the specimens of N. micropus from grouped localities, and in table 12 for tests made when specimens from grouped localities of N. floridana and N. micropus were tested simultaneously. Means of samples of the two species overlap broadly, but samples shown to be significantly different by the two methods are generally the same. The pattern of geographic variation seen in this measurement is typical of most cranial measurements. The sam- ples of largest woodrats are from locali- ties L (southern coastal Texas), C, D, and G (south-central Kansas and ad- jacent western Oklahoma inclusive of the panhandle). The smallest are N. m. mi- cropus (N and P) from Tamaulipas and N. m. canescens from samples J (south- west Texas) and M (Coahuila and Nuevo Leon). Specimens from other lo- calities tend to form gradual clines with the large and small samples listed above. The single exception is the marked change in size between specimens from coastal Texas (N. m. canescens, locality L) and northern Tamaulipas (N. m. micropus, locality N ). Specimens of both sexes from locali- ties in Colorado, Kansas, northern Texas, and Oklahoma generally are larger in condylobasilar length than those from southern localities, including sample I (northeastern part of the range in Texas just south of the Red River). The only_ breaks in this trend result from the large size of specimens from the Gulf Coast of Texas and the relatively small size of females (A) from Colorado. Geographic trends in zygomatic breadth are shown in figures 17 (males) and 18 (females). | Except the females from Colorado (A) appear larger and more nearly the size of females from other northern localities, these are comparable to those discussed for condylobasilar length. Trends in geographic variation of the least interorbital constriction are shown in figures 19 (males) and 20 (females). This dimension differs from other cranial characters of micropus in three respects. First, variation in group-means is not sig- nificant for females. Second, means of both sexes from locality L are relatively low, placing this population more nearly with others in the southern part of the distribution rather than with the larger northern populations. Thirdly, specimens from locality H, which generally are in- termediate in size (especially females), have the broadest average interorbital constriction. Interorbital constriction is consistently broader in floridana than micropus, and the possibility exists that the wide constriction in specimens of micropus from these localities has_re- sulted from introgression of floridana genes. The pattern of variation for breadth at the mastoids is typical of most cranial measurements save for the unusually large size of specimens from localities A and B. Analysis of rostral length, espe- cially for males, exemplifies the north to south and east to west clines in diminish- ing size. The pattern of geographic vari- ation in rostral breadth deviates from this general trend of variation in the large size of males from locality J (south- western Texas). In other characters, woodrats from locality J tend to be more like the smaller woodrats from western and southern parts of the range of the species. Variation in alveolar length of maxil- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA TABLE 13. Geographic variation in 14 external and cranial measurements of Neotoma micropus and N. angustipalata. See figure 8 for geographic areas included within each coded locality. Locality Males Females Code N Mean =+ 2SE Range N Mean + 2SE Range Locality A OeoGOses 1073) = ((S53.0-571.0) 10 348.5 12.74 (318.0-372.0) B 8 364.4 17.76 (340.0-410.0 ) 13) Gores 8.08 (323.0-381.0 ) C lowe Soul Wills} (334.0-411.0) 18 354.8 8.62 (310.0-382.0 ) D 8 3/48 12.45 (351.0-404.0 ) 14 362.3 10.52 (337.0-398.0 ) E 3} SPAILD) 16.04 (305.0-330.0 ) 5 329.0 18.25 (310.0-354.0 ) F 7 362.7 10.58 (340.0-380.0) 8 358.5—) 1246" ((33310-378.0) G A363.) 8.89 (355.0-376.0) 10 35a9 8.21 (328.0-373.0) H by Ba 12.42 (348.0-385.0 ) io SANE 11.14 (311.0-355.0) I Jae 3584 1021 (317.0-398.0) 30 345.3 9.32 (303.0-400.0) J 4 358.2 10.72 (350.0-374.0) 8 330.1 12:16 ~ (81010-352.0) kK 7 350.3 23.44 (302.0-380.0) 7 9351.3 22.63 (304.0-390.0) L Cee ono ON) 22572) (348.0-422.0 ) 8 365.9 18.16 (319.0-388.0 ) M 12 345.4 1G) Be? (302.0-368.0 ) 12 339) 9.40 (313.0-366.0 ) N 9 370.3 9.78 (349.0-390.0 ) 2 9361.5 23:00 (350.0-373.0 ) 0 i 3ls:0 eae | =) 0 = :- ( 1) 2 3.) 364:3 2.40 (362. 0-366. 0) 4 3545 24.84 (333. 0377. 0) Q ee Sok.0 ( z) 0 ae el 3) R 1 402:0 ( _) 3° 3903 25:36 (365. 0-404. 0) Length of tail vertebrae A 3) deere 481 (135.0-143.0) 10 += 139.6 429 (130.0-150.0) B Sml45.2 7.28 (131.0-160.0) ie | 14428 4.36 (130.0-157.0) C 15 156.6 5.99 (135.0-175.0 ) 18 148.8 5.51 (130.0-165.0) D 8 147.0 9.19 (120.0-164.0) 14 148.7 Delis (126.0-168.0) E Sy Ber foley (120-134) 0) Hee 3or6 7.60 (126.0-144.0) F 7 146.3 6.34 (133.0-156.0 ) 8 152.9 7.68 (138.0-170.0 ) G 4 149.0 3.74 (145.0-154.0) 10M 502 4.99 (136.0-162.0) H lara 1OO7 iC 13610-16510) 7 149.6 5.40 (138.0-159.0) I 22 146.0 4.98 (129.0-166.0) 30 §6144.8 5.39 (120.0-195.0) Vf 4 148.5 8.19 (140.0-156.0) 8 137.4 10.65 (110.0-153.0) K 7 149.7 10.62 (126.0-169.0) a 149.9 7.65 (133.0-161.0) ify Ga 15633 9.53 (150.0-180.0) See 57-5 6.29 (147.0-173.0) M 13 141.2 6.33 (113.0-154.0) 12 140.4 4.88 (121.0-153.0) N 9 1649 7.00 (147.0-177.0) 2 1645 P5007 2 C1502172.0) O i $120:0 Aes Bil 3) 0 ae eee Mega" feee ) iP, 3 169.7 4.06 ( 166.0-173. 0) AS NS Ovi ae w ( 155.0-193. 0) Q ey A167.0 ( z=) 0 = ae (ae ) R 1 200.0 ( 7s) 3 1907 — TTS: * ( 179.0-198. 0) Length of hind foot A 6 38.0 ESIC (35.0-40.0) 10 37.3 1.08 (34.0-39.0) B 9 38.7 1.76 (36.0-45.0) 13 38.5 0.89 (36.0-41.0) C 16 39.5 1455 (35.0-43.0) ef 38.4 0.68 (36.0-41.0) D 9 40.2 0.93 (38.0-42.0) 15 39.1 0.73 (37.0-41.0) E 3 36.3 0.67 (36.0-37.0) 6) 36.6 1.85 (35.0-40.0 ) F 6 OieD 235 (33.0-41.0) 8 35.9 0.80 (34.0-37.0) G 5 38.2 1233 (36.0-40.0 ) 9 38.2 1.14 (36.0-41.0 ) H i 36.6 WT (35.0-39.0) 8 37.0 1.20 (35.0-40.0 ) I PAL 3o1.3 1.54 (28.0-43.0) 31 36.7 0.79 (32.0-40.0) i 4 36.0 0.82 (35.0-37.0 ) 8 34.6 25 (33.0-38.0 ) K 9 39.1 0.62 (38.0-40.0 ) a 36.6 2.64 (30.5-40.0 ) 1 6 40.3 1.91 (37.0-43.0) 7 39.9 1.45 (37.0-43.0 ) M 13 Oiled. 1.49 (32.0-41.0) 1 37.9 0.87 (36.0-41.0) N 10 30.0 1.00 (35.0-40.0 ) 3 Silke 3s (36.0-38.0 ) O 1 36.0 = (i) 0 ah ms (en A) P 3 38.0 iets) (37.0-39.0 ) 4 36.5 1.29 (35.0-38.0 ) 93 94 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 13.—Continued. Locality Males Females Code N Mean =+ 2SE Range N Mean + 2SE Range Q 1 38.0 (ea rc 5§) 0 e a (oo SE a) R 1 42.0 (ak a) 3 38.7 2.91 (36.0-41.0) Length of ear A 6 28.3 0.42 ( 28.0-29.0) 9 27.0 0.94 (25.0-29.0) B 9 ipl 1.02 (25.0-29.0) 12 He les: 0.62 (25.0-29.0) C 7 27.0 1.07 (25.0-29.0) 12 26.8 0.95 (25.0-30.0 ) D 6 28.2 0.61 ( 27.0-29.0) sa 28.1 igi (25.0-32.0) E 3 26.3 1.76 (25.0-28.0) 5 28.4 1.85 ( 27.0-32.0) F 4 25.2 Soli (20.0-28.0 ) 6 28.2 0.61 (27.0-29.0) G 5 27.0 1.10 (25.0-28.0) 1 26.1 0.81 (25.0-28.0) H 6 26.3 0.42 (26.0-27.0) 6 26n 0.99 (25.0-28.0) I eT 25.9 eS (20.0-30.0 ) 24 25.9 0.79 ( 22.0-30.0) Ij 4 26.2 3.40 ( 22.0-29.0) 8 26.1 1.33 (23.0-28.0) K 6 28.8 2.39 (25.0-34.0) 2 23.5 7.00 (20.0-27.0) Je 6 28.2 1.50 (25.0-30.0 ) 6 29.1 1.42 ( 27.0-32.0) M 12 28.7 0.92 (26.0-31.0) 10 28.4 0.84 (26.0-30.0) N 10 29.4 0.90 (27.0-32.0) 3 29.0 1S (28.0-30.0) O 0 = a (Cae) 0 i ae (Cr Se P 2 28.0 2.00 (27.0-29.0) 4 28.2 2:22 (25.0-30.0 ) Q 0 ae ( =) 0 = —_ ( sz.) Pass R 1 36.0 ( )) 3 28.7 Ses: (27.0-32.0) Greatest length of skull A 6 49.4 leltzé (47.9-51.4) 9 46.5 1.20 ( 43.0-48.8 ) B 11 49.2 0.78 (47.6-51.6) 12 48.8 0.86 (46.1-51.3) G 14 49.7 0.95 (46.4-52.9) 15 48.8 1.09 (44.2-51.8) D 9 50.5 0.95 (48.8-53.4) its 49.7 0.56 (48.0-51.4) E 3 46.6 0.53 (46.2-47.1) 4 47.0 233 (43.8-49.2 ) F 6 48.1 0.61 (47.0-49.0) 8 AT.7 0.81 ( 46.0-49.2 ) G 5 49.6 1.65 (47.6-52.1) 8 48.3 0.55 (47.1-49.0) H u 50.6 1S (48.6-53.4) 8 47.6 1215 (45.2-50.5) I 19 48.3 0.86 (43.8-51.1) 28 48.5 0.65 (45.3-51.5) J 3 47.9 1.58 ( 46.3-48.8 ) 8 46.2 ila ye (44.0-48.8 ) K U 48.7 0.86 (47.5-50.5 ) 6 47.1 1.08 (45.9-49.5) iL 6 51.0 1.50 (47.9-53.0) il 48.6 127 (45.9-50.8 ) M Le 46.6 0.92 (42.5-48.5 ) 11 46.7 1.07 (43.9-49.3 ) N 8 46.8 1.58 (43.9-50.0 ) 2 45.7 1.80 (44.8-46.6 ) O 1 44.9 me (Ca eee at) 0 a £3; (Se) P 3 46.6 0.42 ( 46.2-46.9 ) 4 45.8 1.49 (43.9-47.1) Q 1 44.9 a ( =) 0 ase cell (eae) R 1 51.0 ( =) 2 49.8 0.10 (49.7-49.8 ) Condylobasilar length A 6 48.8 0.95 (47.7-50.3 ) 9 45.1 1.38 (41.1-47.5) B 1 48.2 0.75 (46.1-49.9 ) 13 47.0 0.68 (44.8-48.8 ) ( 13 48.3 1.06 ( 44.6-50.9 ) 16 47.1 0.92 ( 42.8-50.0) D 9 49.1 0.74 (47.2-51.3) 15 47.8 0.51 (45.7-49.2 ) E 3 44.8 1.05 ( 43.8-45.6 ) 5 45.1 1.60 ( 42.3-47.0) F 7 46.9 0.77 (45.4-48.3 ) 8 46.1 0.66 (44.7-47.1) G 5 AT.7 0.92 (46.6-49.2 ) 8 46.3 0.78 (45.0-47.8 ) H 8 48.4 0.76 ( 46.7-49.7 ) 8 45.9 1.08 (43.7-48.4) I 22, 46.9 0.79 (42.6-49.6) 30 46.4 0.58 (43.7-49.9 ) J 3 46.5 1.14 (45.4-47.2) 8 44.7 1.36 (41.7-47.5) K U 46.7 0.91 (45.0-48.8 ) 7 45.6 1.10 (43.8-48.1 ) L 6 49.4 1.48 (46.1-51.1) ¢ 46.8 1.21 (44.2-48.9) M 11 44.7 0.93 (41.3-46.2 ) 11 44.5 0.84 (41.9-46.7 ) N 8 45.2 1.40 (42.7-48.5 ) 2 44,4 0.80 (44.0-44.8 ) O 1 43.9 (ere Soc) 0 (ger ates) WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA TABLE 13.—Continued. Locality Code P Q R HOWOZS EAHA KH tesahoaePS DOWOZZrPAKH TODA DOePS rad halle ly |p )les| lea) te) (apy lool —_ ot Ee o| & bo RSPR WrOWaIWwWwoawtdwoiuwir Dd _— jd bo BPR WORE WRDOWRDWOUNWODF D — — He — Males Females Mean + 2SE Range N Mean + 2SE Range 44.2 0.75 (43.6-44.9 ) 4 43.2 1.44 (41.7-44.5) 44.0 ae (ee ee) 0 al a3 (Cee) Te) 49.6 (Ape eee!) 2 47.4 0.30 (47.3-47.6) Zygomatic breadth Dial: 0.90 (25.4-28.5) 8 26.3 0.99 (23.8-28.2 ) 26.5 0.55 (25.1-28.0) 13 26.7 0.70 (24.7-29.1 ) 26.8 0.48 (25.5-28.8 ) yf 26.5 0.45 (24.9-28.4 ) Niles’ 0.45 (26.7-28.9 ) 15 26.8 0.44 (25.5-28.4) 25S 0.64 (24.7-25.8 ) 5 25.2 0.99 (247322 7/al) 26.5 0.39 (26.0-27.4) i 26.1 0.70 (24.6-27.6) 26.5 0.75 (Q5:5=2725)) 10 26.1 0.60 (2405-2772) pl 0.67 (26.2-29.1 ) 8 25.8 0.51 (25.0-26.5 ) 26.1 0.41 (235722704) 32 26.1 0.36 (23.0-28.1 ) 25.3 0.20 (25.2-25.5) 8 25.1 0.40 (24.3-26.0) 25.8 0.81 (25.0-28.0 ) ih ess 0.58 (24.8-26.8 ) 27.6 0.88 (25.9-28.9 ) il 25.8 0.80 (24.6-27.3) 24.8 0.50 (23:9-25;7 ) fat 24.6 0.39 (23.6-25.3 ) 24.9 0.82 (23532264) 2; 24.9 1.00 (24.4-25.4) 25.4 Jus im on SANS 0 = = CP Ae Oe 25.0 giles (24.4-26.2 ) 4 24.0 0.38 (23.5-24.4) 235 (Gen =*) 0 = aie (amar SS) 25.8 ( =) 2 25.0 2.00 (24.0-26.0) Least interorbital constriction 6.4 0.23 (6.1-6.9 ) 10 6.3 0.21 (5.8-6.8 ) 6.2 0.16 (5.8-6.7 ) 14 6.3 0.16 (5.8-6.7) 6.4 0.13 (6.0-6.9 ) 18 6.3 0.15 G9=720)) 6.4 0.12 (6.1-6.7 ) 16 6.4 0.14 (5.9-7.0) 6.0 0.13 (5.9-6.1 ) 5 6.2 0.32 (5.9-6.8 ) 6.2 0.12 (6.0-6.4 ) 8 6.3 0.23 (5.8-6.7) 6.4 0.41 (5.9-6.9 ) 10 6.4 0.21 (5.8-6.9) 6.6 0.33 (6.1-7.7) 8 6.4 0.24 (5.9-6.9 ) 6.4 0.14 (WEA) 33 6.3 0.15 (5: 5-162) 6.3 0.07 (6.3-6.4 ) 8 5.9 0.18 (5.7-6.3) 6.1 0.19 (5.7-6.6 ) 7 6.1 0.16 (5.9-6.5) 6.2 0.21 (5.8-6.5) a 6.1 0.24 @557-6!5)) 6.2 0.18 (5.8-7.0) 12 6.2 0.13 (5.9-6.8 ) 6.1 0.18 (55-615) 3 6.0 0.31 (5.7-6.2;) 6.0 a (Gy Sea ) 0 =a my 0. (ee 6.0 0.29 (5.8-6.3) 4 6.3 0.36 (6.0-6.8 ) 5.6 (aia ) 0 “uae 25 nee (meres Dall (a ) 2 6.0 0.30 (5.8-6.1) Breadth at mastoids 19.9 0.57 (18.9-20.9) 8 18.9 0.50 (18.0-19.9 ) 19.1 0.40 (18.0-20.1 ) ill 19.3 0.35 (18.3-20.3 ) 19.5 0.38 (18.1-20.8 ) 16 18.9 0.27 (17.9-19.8 ) 19.6 0.25 (19.1-20.1 ) 14 19.0 0.23 (18.2-19.5) 18.4 0.76 (17.7-19.0) 5 18.3 0.53 (17.5-19.0) 18.9 0.59 (17.6-20.0 ) 8 18.8 0.29 (18.2-19.3) 19.2 0.49 (18.7-20.1 ) 8 18.9 0.42 (17-7196) 19.6 0.23 (19.0-20.0) 0 19.0 0.58 (18.2-20.1 ) 18.9 0.26 (18.0-20.0 ) 25 19.0 0.27 (76-2057) 18.8 0.46 (18.2-19.3) 8 18.5 0.32 (17.8-19.2 ) 19.0 0.40 (18.5-20.1 ) 7 18.8 0.38 (18.0-19.7 ) 19.7 0.42 (18.9-20.5 ) 8 19.4 0.50 (18.4-20.6) 19.0 0.47 (17.5-19.9) 10 18.6 0.23 (17.8-19.1) 18.8 0.71 (17.5-20.4 ) 2, Ie 0.20 (17.6-17.8 ) 95 96 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 13.—Continued. Locality Males Females Code N Mean = 2SE Range N Mean + 2SE Range O 1 18.1 gay (Cera, pars) 0 an te: (gee es [es 3 18.3 0.44 (18.0-18.7 ) 4 18.2 0.48 (17.6-18.6) Q 1 18.0 = ¢ me) 0 ae me ( =e R 1 19.5 cs ( = ) 2 18.9 0.20 (18.8-19.0) Length of rostrum A 6 19.5 0.69 (18.5-20.7 ) 9 i eferh 0.63 (18.8-19.0) B 11 19.2 0.40 (18.2-20.2 ) Ss 18.8 0.44 (AITETEZ.0 23) C 15 19.5 0.37 (17.8-20.7 ) 17 18.9 0.36 (17.2-20:1) D 9 19.9 0.67 (18.6-21.6) 16 19.9 0.34 (18.6-20.8 ) E 3 17.9 0.77 (17.5-18.7) 4 18.2 tals (16.7-18.6) F ra 18.7 0.20 (18.5-19.2) 8 18.5 0.35 (17.7-19.2) G 5 19.0 0.59 (18.2-20.0 ) 10 18.8 0.47 (17.6-20.1 ) H a 19.8 0.49 (18.7-20.7 ) 8 18.6 0.52 (17.6-19.7) I 21 18.9 0.41 (16.6-20.2 ) 30 18.8 0.31 (@iE3=20570) J 4 18.5 0.89 (17.2-19.2) 8 Let fart 0.69 (16.6-19.4) K 9 18.9 0.46 (17.9-20.0) fl 18.2 0.51 (17.0-19.2) IL. 6 19.8 0.94 (17.7-20.9 ) 8 18.6 0.74 (17.0-20.5 ) M 13 WPT 0.47 (15.7-18.6 ) 12, io 0.56 (16.0-18.9 N ll 16 0.63 (16.0-19.7 ) 3 2 0.98 (16.2-17.8) O i 17.4 as (eed =m) 0 = ae (aes EES) iP 3 17.4 0.47 (17.0-17.8) 4 17.6 0.91 (16.4-18.5) Q 1 yea (i SEES 3) 0 a pa (iz. © me R 1 20.7 ( =) 3 19.1 0.50 (18.8-19.6) Breadth of rostrum A 6 8.6 0.47 (8.1-9.3) 9 7.9 0.30 (7.0-8.5) B 11 8.3 0.26 (7.5-9.2 ) 14 8.2 0.23 (7.2-8.9) ¢ 16 8.4 0.16 (7.8-9.0) 18 8.3 0.18 (7.7-9.3) D 9 8.5 0.23 (8.1-9.1) 16 8.3 0.17 (7.7-9.1) E 3) 7.9 0.12 (7.8-8.0) 5 8.1 0.36 (8.2-8.4) F 7 8.2 0.18 (7.8-8.5) 8 7.9 0.11 (7.7-8.2) G 5 8.3 0.20 (8.0-8.6) 10 8.1 0.23 (7.6-8.6) H 8 8.2 0.23 (7.7-8.7) 8 7.9 0.24 (7.5-8.5) I 24 8.0 Onli, (7.4-8.9) 32 8.0 0.11 (7.4-8.8 ) J 3 8.4 0.18 (8.2-8.5) 8 7.8 0.25 (7.2-8.4) K 9 7.8 0.19 (7.5-8.4) 7 7.9 0.31 (7.4-8.5) ib 6 8.3 0.41 (7.8-8.9) if 8.0 0.39 (7.6-9.1 ) M 13 (ate 0.20 (7.1-8.0) 12 7.8 0.19 (7.3-8.5) N 11 7.9 0.26 (7.0-8.7 ) 3 7.4 0.35 Cleiea) O 1 8.0 a Cage ) 0 == at ( Ghee ) P 3 (65) 0.12 (7.4-7.6) 14 7.4 0.25 (7.2-7.8) Q 1 7.6 (aca ) 0 oe Sp. (a R 1 8.1 (ree ) 3 8.1 0.18 (7.9-8.2 ) Alveolar length of maxillary toothrow A 6 9.0 0.37 (8.5-9.6) 10 8.7 O27 (8.2-9.4) B 11 9.3 0.19 (8.7-9.8 ) 14 9.5 0.21 (8.8-10.0) Cc 16 9.4 0.17 (8.7-10.1) 18 9.3 0.20 (8.5-10.1) D 9 9.4 0.24 (8.6-9.8 ) 16 9.3 0.22 (8.6-10.1) E 3 9.0 0.58 (8.5-9.5) 5 8.6 0.36 (8.2-9.1) F 7 9.0 0.25 (8.5-9.4) 8 8.9 0.26 (8.1-9.2) G 5 9.1 0.45 (8.6-9.9) 10 9.2 0.26 (8.7-9.9) H 8 9.1 0.20 (8.7-9.6) 8 9.1 0.31 (8.4-9.7) I 22, 9.4 0.20 (8.6-10.3 ) 9 9.1 Ol7/ (8.0-10.1) J 4 9.1 0.19 (9.0-9.4) 8 8.9 0.22 (8.4-9.4) K 8 9.3 0.27 (8.8-9.8 ) 7 9.1 0.28 (8.6-9.5) IE- 6 9.7 0.34 (9.0-10.1 ) 8 9.4 0.34 (8.6-9.9) M 13 8.5 0.25 (8.0-9.5) 12 8.7 0.33 (7.9-10.0) WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 97 TABLE 13.—Concluded. Locality Males Females Code N Mean = 2QSE Range N Mean =+ 2SE Range N 10 8.7 0.27 (8.1-9.4) 3 8.5 0.75 (7.8-9.1 ) O il 9.2 ws (Gees ) 0 se paces US Coe Se P 3 9.1 0.41 (8.8-9.5 ) 4 9.0 0.34 (8.6-9.4 ) Q 1 9.4 a (paki ate ) 0 le oe) ee R 1 9.9 (Ca ae ) 3 9.8 0.58 (9.3-10.3 ) Length of palatal bridge A 6 8.6 0.25 (8.3-9.1 ) 10 7.9 0.24 (7.1-8.4) B 11 8.2 0.33 (6.8-8.9 ) 13} 7.9 0.23 (7.1-8.3 ) C 15 8.0 0.19 (7.3-8.5) 18 8.0 0.28 (7.1-9.5) D 9 8.3 0.35 (7.7-9.0 ) 16 8.0 0.15 (7.4-8.5 ) E 3 7.9 0.23 (7.7-8.1) 5 es 0.20 (7.6-8.1 ) F 1 7.8 0.43 (6.9-8.4 ) 8 8.0 0.20 (7.5-8.4 ) G 5 8.2 0.60 (7.0-8.7 ) 10 8.0 0.34 (7.1-8.9 ) H Uf 7.9 0.19 (7.7-8.4) 8 7.9 0.32 (7.4-8.4) I 24 8.1 0.17 (7.4-9.0 ) 33 8.0 0.14 (7.1-8.9 ) J 3 7.9 0.70 (7.2-8.3 ) 8 7.4 0.60 (7.0-7.9) K 8 7.8 0.23 (7.2-8.2) i 7.8 0.35 (7.3-8.5 ) IL 6 8.4 0.51 (7.3-9.0) U 8.2 0.32 (7.8-9.1) M 13 7.6 0.19 (7.0-8.1 ) 13 Uae 0.22 (6.9-8.1) N 11 8.1 0.26 (7.2-8.9 ) 3 7.6 0.76 (7.0-8.3 ) O 1 8.6 ASO A A ) 0 fee. Ge = ) 12 3 Wao 0.81 (6.8-8.2 ) 4 7.8 0.42 (7.5-8.4) Q i 8.6 a (Hat ) 0 | ey (ee ) R 1 9.3 (eyes ) 3 8.7 0.31 (8.5-9.0) Length of nasals A 6 19.9 0.64 (18.8-20.8 ) 9 18.4 0.61 (16.9-19.4 ) B Bl 19.7 0.46 (18.0-20.7 ) He} 19.2 0.55 (AES=ZO 7p) C 15 19.9 0.45 (625-2715) ILy/ 19.3 0.48 (HIGHEZTED)) D 9 20.1 0.55 (18.9-21.4) 16 20.3 0.30 (19.3-21.6) E 3 18.3 PAIL (17.6-19.5) 4 18.7 I foyll (16.9-20.4 ) F 7 19.6 0.43 (18.5-20.2 ) 8 18.8 0.41 (18.0-19.7 ) G 5 19.7 0.77 (18.8-21.1) 10 19.2 0.54 (17.5-20.8 ) H 7 20.5 0.68 (19.5-22.0) 8 19.1 0.65 (17.6-20.6 ) I PALL 19.3 0.46 (16.6-20.8 ) 30 19.2 0.35 (17.8-21.0) J 4 19.0 Tak) (17.4-20.0) 8 18.0 0.80 (16.4-19.8 ) K 9 19.0 0.44 (17.8-19.7) i 18.4 0.55 (17.6-19.4) L 6 20.5 1.00 (18.6-22.0) 8 19.0 0.54 (17.8-19.9) M 13 18.1 0.51 (16.3-19.2 ) 12, ie9) 0.60 (16.2-19.3) N 1a 18.0 0.76 (16.1-20.3 ) 3 We 0.19 Goals) O 1 17.9 (Se Gaae) 0 ee ae ( Gee) P 3 17.9 0.35 (17.6-18.2 ) 4 7A 0.81 (16.3-18.2 ) Q 1 17.9 (2 ) 0 aa = (ae es) R i 19.6 (ae ) 3 19.0 1.39 (17.9-20.3 ) lary toothrow deviates noticeably from the anticipated trends. For example, sam- ples such as L and D, are large and M and N are smaller. Specimens from other localities, such as P and K, have propor- tionately longer toothrows, whereas those from locality A are atypically short. The patterns of variation in palatal bridge length shown by sequence of means are similar to those of toothrow length ex- cept that specimens from locality P are among the smallest. The pattern of vari- ation in nasal length is similar to that of rostral length. The rank-order system of scoring means described above was employed to search for trends in size variation in N. micropus. Specimens from localities D (south-central Kansas) and L (coastal Texas ) were thus calculated to be larger 98 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY than other specimens in the range of the species. Moving away from locality D, size decreases gradually but consistently in specimens to the south (localities G, H, and I) and west (localities C, B, and A). Specimens from localities southwest (localities F, E, O, and J) of the above- listed seven localities decrease in size clinally. The large area in central Texas from which specimens of N. micropus are absent (see Fig. 6 and discussion of distribution for N. m. canescens) lies be- tween locality I and the two in southern Texas (K and L). It is possible that the only route for gene flow between locality L and I is through western Texas; or in other words, animals from locality L may be effectively isolated from the northern populations of woodrats that they re- semble in size. Specimens from locality K are considerably smaller than those from L, slightly smaller than those from I and slightly larger than those from J. Specimens from Coahuila (locality M) are probably smallest of N. m. canescens, but if so, they are only slightly smaller than those from New Mexico and the Big Bend area (J) of Texas. Specimens of N. m. micropus from coastal Tamauli- pas have proportionately longer tails, and thus have high means for total length and tail length. Cranially they are as small or possibly smaller than those from adjacent locality M. The single break in the gradual cline in size is across the lower Rio Grande River. Locality L, having among the largest specimens of the species, is contiguous with locality N, which supports small woodrats having relatively long and often unicolored tails (N. m. micropus). The steepness of this “step” in the cline decreases to the north- west along the Rio Grande. For exam- ple, specimens from locality K are con- siderably smaller than those from L, but considerably larger than those from M. Farther west, specimens from Chihuahua are indistinguishable from those in south- west Texas. Size relationships between specimens of N. m. micropus from northern Tamau- lipas and southern Tamaulipas (pre- viously N. m. littoralis) demonstrate that rats from the two localities are similar, | especially with respect to external and. cranial dimensions and, to a lesser extent, color. I detect no real basis for con-. tinuing to recognize littoralis as a sub- species. Because the only available “adult” specimens from Rio Verde, San Luis Potosi (planiceps, locality Q) and from White Sands, New Mexico (canescens, sample O, previously known as _ leu- cophea) are of age-group V, comments on size of these groups must be tentative. However, woodrats from White Sands appear to be nearly the same size as other specimens from New Mexico. The holotype of planiceps is similar in size to other specimens of N. micropus from Mexico. On the basis of the four adult speci- mens (one male, three females) of N. angustipalata examined, it appears that individuals of this species are much larger than those of N. micropus from all localities in Mexico. The only areas from which specimens of N. micropus compare favorably in size to individuals of angustipalata are localities D and L. The external measurements of angusti- palata far exceed those of all samples of micropus, and measurements of lengths of all or parts of the skull are generally larger. In measurements of breadth, however, several samples of micropus are larger than angustipalata (see table 13 for comparative measurements ). Comparisons of sequence of means and subset relationships from SS-STP computations including the 13 samples of N. floridana and 14 of N. micropus together, reveal that floridana generally is the larger (see Table 12 and Figs. 15-20). Samples of floridana from north- western (campestris) and north-central (attwateri) Kansas and those from south- eastern Texas (rubida) are larger on the average in most dimensions than samples of micropus. Samples of attwateri from southeastern Kansas and from both lo- calities in Texas usually are larger than samples of micropus; however, in some WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 99 dimensions, samples of micropus from ‘south-central Kansas and coastal south- ern Texas are larger than one or more of the floridana samples. Specimens of micropus from the above two localities and from localities in western Kansas and northern Oklahoma are usually larger than samples of baileyi, campestris from Colorado, and attwateri from north- eastern Kansas and eastern Oklahoma. Specimens of micropus from Colorado, extreme southwestern Kansas, southwest- ern Oklahoma, and northern Texas are nearly equal in size to specimens of the last-mentioned samples of floridana. The micropus from southern and western lo- calities are consistently smaller than woodrats of either species from other localities. The single exception is tail length in N. m. micropus, which exceeds that of all other samples of N. micropus and most samples of N. floridana. Wood- rats from the panhandle of Texas, south- western Texas, and non-coastal southern Texas are intermediate in size between specimens from farther to the south and west and those from localities to the north and east. The general trends noted above are less well marked in certain dimensions than others; for example, least interor- bital constriction is broader in all sam- ples of floridana females than in any sample of micropus females. The only samples of micropus that occasionally average larger than N. angustipalata are those from south-central Kansas and southern coastal Texas. However, cer- tain samples of floridana (rubida, cam- pestris from Kansas, and attwateri from north-central Kansas) are larger than angustipalata in many dimensions. In conclusion, univariate analyses in- dicate that the range of size variation within the species micropus exceeds that in the populations of floridana studied. Also, variation in micropus tends to be clinal and relatively consistent geograph- ically. The single major exception is the large size of specimens from coastal southern Texas. Specimens of micropus from western and southern localities gen- erally are smaller than those from north- ern and eastern localities, but those from coastal Texas are larger on the average than all samples of micropus from con- tiguous localities and larger than most samples from northern parts of the range of the species. This phenomenon could have resulted from one or more of three distinct possibilities. The samples from this locality consisted of less than 10 in- dividuals of each sex, and the apparent large size of these woodrats might be the result of sampling error. I consider this possibility relatively unlikely, how- ever, because large size was evinced by specimens of both sexes. The proba- bility of sampling error involving two samples from the same locality is much lower than when only one sample is involved. Secondly, specimens of flori- dana from southern Texas are large; pos- sibly the large size of micropus from contiguous localities is the result of in- trogression of genetic material from flori- dana. Thus, the robustness of animals from coastal Texas perhaps should be interpreted as evidence for hybridiza- tion. Thirdly, there is the possibility that whatever selective forces have resulted in large individuals of floridana in south- ern Texas also are operating on adjacent populations of micropus, resulting in the unexpected large size. Evaluation of the latter two possibilities is difficult and must remain speculative until additional data are available. If hybridization is the answer, it might be expected that specimens of floridana from southern Texas would be slightly smaller than, instead of slightly larger than, specimens from northern Texas and southern Okla- homa. However, it also must be remem- bered that certain qualitative cranial characters discussed previously indicated that specimens of floridana from locali- ties in southern Texas were relatively more micropus-like than are those in most other populations of floridana. Multivariate Analyses of Mensural Characters Moss (1968) and Sokal and Michener 100 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY (1967) have shown that standardization of character states to equally weight characters greatly influences the cluster- ing of OTU’s in correlation phenograms and further tends to reduce the degree of isolation of “aberrant” OTU’s. The dis- tance phenogram usually is less influ- enced by standardization of character states. Comparisons by correlation and by distance augment one another, and must be considered simultaneously in the interpretation of phenograms. Correla- tions are relatively independent of size (totally independent with unstandard- ized character states) and cluster OTU’s primarily on the basis of relative propor- tions. On the other hand, distances are less dependent on relative proportions, and cluster primarily on the basis of ab- solute differences between the numerical values of characters. For example, when Moss (1968:38) multiplied continuous characters by a factor of two to create hypothetical “giant? OTU’s and com- pared original and “giant” OTU’s using unstandardized data, the normal and “giant” OTU’s clustered in the correla- tion phenogram at the 1.0 level. How- ever, in the distance phenogram based on unstandardized data, there was com- plete separation of normal and “giant” OTU’s with comparable clustering within each major cluster. Standardization of the data resulted in a high frequency of group clustering (a cluster of normal OTU’s joined its respective cluster of giant OTU’s) in the correlation pheno- gram, but the distance phenogram was altered very little by standardization. As previously mentioned, character states were standardized for all computa- tions by CLSNT. In most cases I have illustrated distance phenograms, and in all cases the clustering relationships of correlation phenograms are discussed to emphasize proportional relationships and absolute differences among the woodrats. When woodrats from grouped localities were compared using all of the available characters, both correlation and _ dis- tance phenograms are presented. A certain amount of distortion results in the clustering process from a multidimen- sional correlation or distance matrix to a two-dimensional phenogram. The coef- ficients of cophenetic correlation, calcu- lated to express correlation between the original matrices and the resultant phe- nograms, are given beyond for all phe- nograms. This coefficient is normally about 10 percent higher for distance phenograms than for correlation pheno- grams. Rohlf (1968) discussed various relationships between results of cluster analyses and those of principal compo- nents. He recommended (1968:254) that “numerical taxonomic studies should use both cluster analyses and 3-D models in order to extract and present as much in- formation as is possible from the raw data.” In all cases, I have analyzed prin- cipal components in conjunction with cluster analyses. Additionally, for each projection of OTU’s onto the three prin- cipal components, a minimally intercon- nected network (Cavalli-Sforza and Ed- wards, 1967) was computed from the among-OTU distance matrix. When the number of OTU’s was low, these have been included on the 3-D drawings. Models are presented as _ perspective drawings; thus, with respect to the left rear corner of a square platform, the viewer is 0.3 units (one unit is equal to the length of one side of the platform) to the right, 3.0 units toward the front, and at a height (the third principal com- ponent) even with the OTU projected farthest from the platform. Neotoma_ floridana—A _ correlation phenogram was computed from among- OTU correlations for Neotoma floridana females using means of the four external and 10 cranial dimensions as character states; the coefficient of cophenetic cor- relation is 0.714. The 13 OTU’s (corre- sponding to the 13 grouped localities shown in figure 8) cluster into five major groups, which are separated by correla- tions of zero or negative values. Clusters grouped females as follow: 1) localities 1 (N. f. baileyi) and 10 (N. f. attwateri from southeastern Oklahoma); 2) local- ities 2 and 3 (N. f. campestris); 3) local- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 101 ities 4 (campestris), 5, 6, and 7 (attwa- teri from Kansas ); 4) localities 8, 11, and 12 (attwateri from western Oklahoma and Texas); and 5) localities 9 (attwa- teri from northeastern Oklahoma) and 13 (N. f. rubida from southeastern Texas). The coefficient of cophenetic correlation between the distance phenogram and the ) respective matrix is 0.831 for the 13 sam- ples of floridana females. As shown in figure 21, four major clusters emerged; the least distance between two major clusters is 1.35. The proportional relationships of campestris females indicate that speci- mens from Colorado and Nebraska are most like other campestris from western Kansas; but because of differences in size, they are placed with attwateri in the dis- tance phenogram. The large size of fe- males from north-central Kansas (local- ity 5) results in the separation of these rats from other attwateri females in the distance phenogram. The correlation in size between sample means of attwateri females from northeastern Oklahoma and owonns - 12 13 a Oe 1.89 i209 0.69 those of rubida from southeastern Texas was unexpected. The correlation phenogram for flori- dana males has a coefficient of cophenetic correlation of only 0.699. Four major clusters are separated from each other by correlations of —0.02 or less. N. f. campestris from localities 2 and 3 are in the same cluster with baileyi. The sec- ond cluster includes attwateri from Kan- sas, campestris from locality 4, and attwateri from the western sample in Oklahoma. The two samples of attwateri from eastern Oklahoma (9 and 10) are placed in the third cluster with samples 12 (southern locality of attwateri in Texas) and 13 (rubida). The sample of attwateri from locality 11 is alone in the fourth group, but anastomoses with clus- ter three before these join the first two clusters. The distance phenogram, which has a coefficient of cophenetic correlation of 0.816, has four clusters separated by a distance of 1.25 or more (Fig. 21). The most noteworthy differences in the two phenograms for floridana males 1 B 2 6 9 il 10 3 4 5 8 vi 12 13 2.01 1,41 0.89 Fic. 21. Phenograms of UPGMA cluster analyses based on distance coefficients comparing stand- ardized means of 14 mensural characters for 13 grouped localities of Neotoma floridana: A—females; B—males. See text for coefficients of cophenetic correlation and see figure 8 for geographic areas included within the coded localities. 102 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY involves the placement of rats from local- ities 3, 11, and 13. Based on correlations, sample 3 is most like samples 1 and 2, sample 11 is most distinctive, and sample 13 is similar to sample 12. When the emphasis is on differences (distance), sample 3 is most like the larger attwateri, sample 11 is similar to sample 9 and other small attwateri; sample 13 is unlike any other sample. The first three principal components account for 84.4 percent and 78.1 percent of the total variation for females and males, respectively. Percent variation in each component for females and males, respectively, is 63.4 and 48.6 in the first, 14.5 and 15.8 in the second, and 6.5 and 15.7 in the third. The 3-D perspective drawings of projections of OTU’s (the 13 samples) onto the first three principal components are shown in figure 22 for both sexes. The 3-D projection for females is reminiscent of the cluster relationships seen in the distance phenogram; the same four basic groups obtain. Samples 5 and 13 are relatively isolated. Large campes- tris (3 and 4) and large attwateri (7, 11, and 12) are situated relatively close, and baileyi and smaller campestris are near the samples of small attwateri. The mini- mally connected network shows the three samples of campestris (2-4) intercon- nected and connecting to the cluster of smaller-sized females through sample 2 and the cluster of larger-sized females through sample 4. Within the cluster of “small” rats, baileyi is most distinct and is connected to the rest of the samples in the cluster through sample 10. The 3-D projection of floridana males also bears a strong resemblance to the respective distance phenogram. In the projection, sample 1 (baileyi) connects through sample 2 to the cluster of large attwateri and campestris. The group of smaller attwateri connects to the “large” group through sample 8. N. f. rubida is farthest separated from other OTU’s and connects to the group of larger rats through sample 12, which is connected to sample 8. Thus, sample 8 appears to be more or less intermediate, serving to interconnect the various clusters. On the basis of mensural characters, sample 13 (rubida) is the only sample that clearly and consistently is distinct. The females in sample 5 (north-central Kansas) evince distinct differences from either attwateri or campestris, and ap- pear to be nearly as distinct as rubida; this relationship is not seen in the com- parison of males. The three samples of campestris do not form a closely allied group. In attwateri, a tendency exists toward one cluster of smaller woodrats and a second of larger woodrats. Geo- graphically, however, localities from which specimens of the two “groups” originated are such that populations of “small” and “large” rats are interspersed. Animals from locality 12 (previously the only locality from which specimens were assigned attwateri) are much like mem- bers of the “large” group of other attwa- teri (previously osagensis) and do not resemble rubida. Neotoma micropus and Neotoma an- gustipalata—Neotoma angustipalata is treated with samples of N. micropus in all CLSNT analyses. These two taxa were combined because, on geographic grounds, it appeared that angustipalata might be a subspecies of micropus (Hooper, 1953:10; Alvarez, 1963:453) and that N. angustipalata and N. m. planiceps might best be considered as a single taxon. The coefficient of cophenetic correla- tion between the correlation phenogram and matrix for the 15 samples of N. mi- cropus and one sample of N. angusti- palata females is 0.835. Two major clus- ters are formed. The first consists of one subcluster of samples A (Colorado and Cimarron County, Oklahoma), F (Texas panhandle) and E (New Mexico) and a second subcluster that joins the first at a correlation of 0.05. This subcluster in- cludes samples J (Big Bend area of Texas), K (non-coastal southern Texas ), H (southwestern Oklahoma), I (north- eastern Texas), B (southwestern Kansas and adjacent Oklahoma panhandle), D WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 103 I Fic. 22. Three-dimensional perspective drawings of the projections of 13 samples (OTU’s) of Neotoma floridana onto the first three principal components based on correlation among 14 mensural characters: A—females; B—males. Dashed lines between OTU’s illustrate the minimally intercon- nected networks computed from the respective among-OTU distance matrices. See text for per- centages of variation in each component and figure 8 for geographic areas included within the coded localities. (south-central Kansas), C (south-central Kansas just west of locality D and ad- jacent Oklahoma panhandle), and G (northwestern Oklahoma south of local- ity D). The second major cluster is com- posed of samples L (southern coastal Texas), R (N. angustipalata), M (Coah- uila and Nuevo Leén), N (N. m. mic- ropus from northern Tamaulipas) and P (N. m. micropus from southern Tamauli- 104 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY pas). In essence, this phenogram places all samples of N. m. canescens except two (M and L) in a single major cluster that is not highly correlated internally. The other major cluster includes an array of taxa including N. angustipalata, both samples of N. m. micropus, and two sam- ples (one of large woodrats and the other of small woodrats) of N. m. canescens. The distance phenogram (Fig. 23) for micropus and angustipalata females has a correlation of 0.820 with the dis- tance matrix. If a distance of 1.25 or greater is considered to separate major clusters, four were computed. The pat- tern of clustering seen in the distance phenogram corresponds quite well to the taxonomic arrangement of these two spe- cies. That is, angustipalata appears most distinct and is recognized at the specific Oonoaonwnxzke- MS DP > vz20rF tf = SSS SS OO ——————————eeeeeeeeeee 2.065 1.365 0.665 level, and the two samples of N. m. mi- cropus are distinct from those of other micropus, but closer to them than to angustipalata. The most variable sub- species, canescens, consists of two groups that correspond, with a few exceptions, to the previously recognized boundaries of N. m. canescens and N. m. micropus. Nevertheless, both clusters include sam- ples of woodrats previously assigned to the two different names. The coefficient of cophenetic correla- tion for the correlation phenogram of 17 samples of micropus males and one sam- ple of angustipalata males is relatively high (0.849) for a correlation pheno- gram. Of the three major clusters (sepa- rated by a correlation of —0.05 or less), one contains the two samples from New Mexico (E and O); the second, joins —-—n" ono wrroo09 PF rmanoonmrsktvye2zx iu eee ee eee eee 2.20 1.40 0.60 Fic. 23. Phenograms of UPGMA cluster analyses based on distance coefficients comparing stand- ardized means of 14 mensural characters for 16 (females) and 18 (males) grouped localities of Neo- toma micropus (A-Q) and N. angustipalata (R): females; B—males. See text for coefficients of cophenetic correlation and figure 8 for geographic areas included within the coded localities. WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 105 the first before these two join the third, and includes all samples of canescens from localities in the United States ex- cept E, O, and K. The third major cluster contains two distinct subgroups that join at a correlation of only 0.12. In the first subgroup are samples K and M (both N. m. canescens). In the other, samples N and P (both N. m. micropus ) join at a correlation of 0.62, and Q (N. m. planiceps) joins R (N. angustipalata) at 0.800. These two couplets anastomose at a correlation of 0.45. This phenogram illustrates the high proportional simi- larities between angustipalata and plani- ceps, the affinities between the two sam- ples of the subspecies micropus, and the complete intermixing of samples of N. micropus from the northern part of the range where two subspecies pre- viously were recognized. The distance phenogram for this series of male samples is shown in figure 23. The coefficient of cophenetic corre- lation between the phenogram and _ its matrix is 0.840. There is a marked ten- dency in both phenograms of males for most samples of N. m. canescens ( excep- tions were samples E, O, and M, espe- cially) to cluster together; the two sam- ples of N. m. micropus are always more similar to one another than either is to any other sample, but woodrats in sam- ple M appear to be as near micropus as canescens. The correlation between N. angustipalata and N. m. planiceps is re- markably high, indicating proportional similarity. Although available material indicates that angustipalata is much larger than planiceps, it must be remem- bered that the only known specimen of planiceps is a young adult. The simi- larities of samples O and E from New Mexico are noteworthy because speci- mens in sample O previously were recog- nized as a distinct subspecies, N. m. leucophea. Principal components analysis for fe- males extracted a total of 86.2 percent of the variation (components one to three composed of 55.1, 24.3, and 6.8 percent, respectively). The 3-D drawing of OTU’s projected onto the first three com- ponents is shown in figure 24. Sample R, as expected, is most distinct and iso- lated from other OTU’s, especially on the second component. Sample M serves as the intermediate through which sam- ples of N. m. micropus connect with samples of N. m. canescens. The latter are separated on the first component (which is highly correlated with size), but were similarly placed on the second component. The first three components for males contain 56.1, 27.3, and 7.6 percent of the total variation, respectively, for a total extraction of 91.0 percent. The 3-D pro- jection (Fig. 24) of OTU’s onto these components placed angustipalata as the most distinct OTU, connected to mi- cropus through sample L. Neotoma mi- cropus planiceps is the only other rela- tively distinct OTU, but it is separated much less on component two than is angustipalata; planiceps connects to other samples of micropus through sam- ple P. In the projection, as in the dis- tance phenogram, sample M appears to share as many affinities with the subspe- cies micropus as with canescens. The 3-D projections of both sexes elucidate the distinctiveness of angusti- palata, and to a lesser degree, the dis- tinctiveness of planiceps. The affinities of sample L with the northern samples of large N. micropus are evident. Like- wise, the similarity of the two samples of N. m. micropus from Tamaulipas, and the intermediacy of woodrats from Coahuila and Nuevo Leén between mi- cropus and canescens are shown. Finally, the projections document the absence of any distinct steps in the clinal variation in samples of N. m. canescens. Simultaneous Treatment of Three Species——When all samples of N. flori- dana, N. micropus and N. angustipalata were treated simultaneously in a multi- variate analyses of 14 character states, the results are surprising. The correla- tion phenogram for the 29 samples of fe- males has a coefficient of cophenetic cor- relation of only 0.686, and is divided 106 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY dD . aE II | / : AY 2 I ‘eu ” | A : - I =C ; oe C y ‘| | Or mS Sse s | Oe LA 6): | ae T rel | | | D ail | | | ey = a at | Bayne = y Y | | | ; | | | | loadeeal | wis joie ll Je em Ty / | | | 3 | | | | oO. | eu Fic. 24. Three-dimensional perspective drawings of the projections of 16 (females) and 18 (males) samples of Neotoma micropus (A-Q) and N. angustipalata (R) onto the first three princi- pal components based on correlations among 14 mensural characters: A—females; B—males. Dashed lines between OTU’s illustrate the minimally interconnected networks computed from the respective among-OTU distance matrices. See text for percentages of variation in each component, and fig- ure 8 for geographic areas included within the coded localities. into two primary clusters separated by a micropus with those of floridana and the negative correlation (-—0.25). Most sur- marked alteration of the clustering rela- prising is the intermixing of samples of tionships seen previously for samples of WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 107 floridana. Although clustering of sam- ples of angustipalata and micropus is not exactly as discussed above, it is notice- ably less altered. The first major cluster contains N. f. rubida, N. angustipalata, the samples of small N. micropus, and sample L. The two samples of N. m. micropus form a highly correlated coup- let (0.78) that anastomoses first with the sample of angustipalata. The other ma- jor cluster contains the remaining sam- ples of both species arranged into four subclusters. One of these consists only of sample 7 and another is made up of the three samples of smallest floridana ( baileyi and attwateri samples 9 and 10). The two remaining subclusters include samples of floridana and micropus ar- ranged such that neither species appears to be distinct. Some geographically contiguous sam- ples tend to cluster together and there is a general tendency of samples of each species to cluster in the same subgroups. However, the deviations from these ten- dencies are so great as to cause this phe- nogram to bear little resemblance to the classification used herein or to the gen- eral conclusions based on univariate analyses. The placement of OTU’s in the dis- tance phenogram for all samples of fe- males (Fig. 25) is more nearly congru- ent with previously discussed pheno- grams, 3-D projections, and the classifi- cation that I have proposed. Although, some samples of floridana and micropus appear together in one of the four major clusters, in only one instance is a sample of one species placed more closely to a sample of the other than to a conspecific sample. This exception was sample D. The coefficient of cophenetic correlation between phenogram and matrix is 0.736; the first bifurcation is at a distance of 1.86 and each cluster thereby formed is composed of two major subclusters. The correlation phenogram for the 31 samples of males closely resembles the two correlation phenograms for males discussed above. It has a coefficient of cophenetic correlation of 0.720 to the correlation matrix. Three clusters are separated from each other by negative correlations. The most distinct of these contains N. angustipalata in a couplet with N. m. planiceps, N. f. baileyi in a couplet with N. f. campestris from local- ity 2, a couplet with the two samples of N. m. micropus, and lastly a couplet with two geographically contiguous samples of N. m. canescens (K and M). The re- maining two clusters are more highly correlated than either is to the first. Sam- ples from the five most southern localities of N. floridana (9, 10, 11, 12 and 13) are included in one cluster with sample H (N. m. canescens from southwestern Oklahoma). In the third, sample 4 (campestris) forms a couplet with sam- ple A (canescens) in a subcluster includ- ing sample L and northern samples of canescens. Samples E, J, and O (canes- cens from New Mexico and adjacent southwestern Texas) form another sub- cluster with sample 3 (campestris). In the remaining subcluster, four samples of attwateri (5, 6, 7, and 8) are placed near a sample of canescens (1). Those OTU’s that form highly corre- lated couplets or clusters when treated in the restricted samples usually tend to cluster when the three species are treated simultaneously. Similarly, samples that appear highly distinctive when treated in conspecific groupings usually retain at least partial distinctness when all sam- ples are treated together. However, those samples that are neither especially distinct nor highly correlated when treated conspecifically, usually are un- predictable in their clustering relation- ships when the number of samples and total amount of variation are increased. The distance phenogram for 31 sam- ples of males, which has a coefficient of cophenetic correlation of 0.765, is shown in figure 25. Three major clusters sepa- rated by a distance of 1.5 or more are evident. The two phenograms for all samples of males of both species indicate the distinctness of angustipalata as com- pared to micropus and floridana. The size relationships of micropus and _flori- 108 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY - A A B B A G F Cc vi 6 E 9 J 11 K H N D P L B 3 Cc 4 G 5 H 8 I 7 6 10 8 12 9 1 10 2 1 13 74 F L I D J 3 K 4 N 7 P 11 M 12 E R oO 5 Q 13 R 1.925 1225 0.525 2.16 1.36 0.56 — Fic. 25. Phenograms of UPGMA cluster analyses based on distance coefficients comparing stand- ardized means of 14 mensural characters for 29 (females) and 31 (males) grouped localities of Neo- toma floridana (1-13), N. micropus (A-Q), and N. angustipalata (R): A—females; B—males. See text for coefficients of cophenetic correlation, and figure 8 for geographic areas included within the coded localities. WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 109 dana overlap sufficiently that mensural characters alone do not segregate them into separate major clusters. The larger micropus tend to form conspecific group- clusters which are near group-clusters of larger floridana; smaller micropus tend to cluster separately from samples of smaller floridana. On some occasions samples A, F, and K, which are from localities geographically intermediate be- tween large micropus and small micro- pus, clustered with the large micropus and smaller floridana; on other occasions, they clustered with the small micropus. The 3-D projections of all samples of females and of all samples of males are shown separately in figure 26. The mini- mally connected networks were com- puted but have been omitted from the drawings to enhance determination of relative positioning of the OTU’s. Pro- jection of 29 OTU’s on the three com- ponents for females results in a more nearly complete separation of samples of micropus and floridana than is seen in the two phenograms for females. Be- cause each OTU is necessarily connected by the network directly to at least one other OTU and all OTU’s are intercon- nected, it is necessary that at least one connection exist between a sample of micropus and one of floridana. In this instance two such connections exist. Sample 9 (northeastern Oklahoma flori- dana) connects to sample H (southwest- ern Oklahoma micropus), and sample D (south-central Kansas micropus) con- nects to sample 7 (southeastern Kansas floridana). All samples of N. micropus are directly or indirectly interconnected without involving a sample of floridana, but floridana samples 1, 2, 6, 8, 9, and 10 are connected to other samples of floridana through a series of samples of micropus (9 to H to I to G to C to D to 7). N. angustipalata connects only to N. m. canescens sample L. In the 3-D projection of males, only the minimal number (two) of inter- specific connections were computed. Sample 9 is connected to sample H to join micropus and floridana. Neotoma angustipalata joins only to sample 13, N. f. rubida. In the unconnected draw- ing (Fig. 26), it can be seen that several samples of micropus overlapped samples of floridana on the first component. The connections of these micropus samples were G to A, and B to C to D to L. In the 3-D projections for females there are two distinct groups of floridana samples. A tendency exists toward a similar sepa- ration on the projections for males, but it is less distinctly defined. Multivariate Analyses of Size, Color, and Qualitative Cranial Characters Following multivariate analyses of the 14 mensural characters discussed above, similar analyses were conducted using the same 14 mensural characters together with four color reflectance scores and three, scored, qualitative cra- nial characters. Thus a total of 21 char- acters was available for each sex of each sample. Instead of analyzing sexes sepa- rately as above, the 21 characters for each were pooled and used as 42 char- acters for each sample. For the two sam- ples composed only of a single male spec- imen each (O and Q), the 21 characters were treated twice, once with those for males and once with those for females. Measurements of bacula or other charac- ters were not included because data for one or more samples were not available. The results obtained depict the rela- tionships of the various OTU’s in a more comprehensive way than do interpreta- tions based on separate analyses (by sex) of mensural characters alone. However, an understanding of the relationships based only on mensural characters is necessary to understand overall geo- graphic trends in size and to determine relative distinctness or indistinctness of various taxa based solely on dimensions and relative proportions. Neotoma floridana—CLSNT based on 42 characters as described above us- ing the 13 samples of Neotoma floridana yielded results congruent with the pres- ent classification. The coefficient of co- phenetic correlation is 0.727 between the 110 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY / / ——— Fic. 26. Three-dimensional perspective drawings of the projections of 29 (females) and 31 (males) samples of Neotoma floridana (1-13), N. micropus (A-Q), and N. angustipalata (R) onto the first three principal components based on correlations among 14 mensural characters: A—fe- males; B—males. See text for percentages of variation in each component and figure 8 for geo- graphic areas included within the coded localities. correlation phenogram (Fig. 27) and the matrix from which the phenogram was computed. Two major clusters are sepa- rated by a correlation of —0.24. This phenogram, which was computed from data on males and females, is remarkably unlike the correlation phenogram dis- cussed above for floridana females. Despite several minor alterations, it re- sembles the correlation phenogram of floridana males. The relatively high cor- relation between N. f. attwateri and N. f. rubida is indicative of proportional similarity of the two, and possibly re- flects intergradation between the two in southeastern Texas. The two samples of N. f. campestris from western localities (2 and 3) are also relatively highly cor- related. Characteristics of campestris from locality 4 are more highly correlated with those of attwateri than with other samples of campestris. In this pheno- gram, balieyi appears more like cam- pestris than attwateri. The distance phenogram (Fig. 27) for floridana is characterized by several major shifts in the positioning of OTU’s relative to that seen in the correlation phenogram. These shifts affect primarily the non-attwateri samples; samples 6, 8, 9, and 10 remain relatively close to- gether and samples 7 and 11 remain to- gether. Sample 12, which is highly corre- lated to 13, is placed with two other WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 111 samples of attwateri, 7 and 11, in the distance phenogram. Samples 4 and 5 are removed from a correlation cluster with the “large” attwateri samples and placed with sample 3. Sample 3 is clus- tered with samples 1 and 2 in the correla- tion phenogram. These two samples (1 and 2) remain together, but are sepa- rated by an appreciable distance (1.15). Sample 13 (rubida) is the most distinct sample in the distance phenogram. The coefficient of cophenetic correlation for this phenogram is 0.769. The first five principal components were extracted in computations involv- ing 42 characters. The percent variation in these is 36.6, 22.6, 12.2, 9.8, and 5.1 for a total of 86.3 percent; 71.4 percent of the variation is in the first three com- ponents. The 3-D projection of the 13 OTU’s on the first three components (Fig. 28) and the minimally connected network (not figured) indicate that sam- ple 7 is intermediate between other sam- ples in many respects; five independent subgroups interconnect by direct attach- ment to sample 7. One subgroup consists 1 A 2 3 6 8 9 10 11 12 13 -0.30 0.10 0.50 only of sample 5 (sample 5 and 7 are the two most distant directly-connected samples of attwateri) and another only of sample 11. A third subgroup includes sample 12 relatively close and sample 13 at a considerable distance. Another includes the four samples of small-sized attwateri and the last includes the three samples of campestris and one of baileyi. The latter “lineage” is especially interest- ing because neither the correlation nor the distance phenogram placed the three samples of campestris together. More- over, the sequence of the connections is 4 to 3 to 2 to 1; geographically this cor- responds east-west for the three samples of campestris. The significance of the apparent relationship between _ baileyi and campestris will be discussed below with respect to zoogeographic history. Neotoma micropus and Neotoma angustipalata—tThe correlation pheno- gram for the 18 samples of N. micropus and N. angustipalata (Fig. 29) has a coefficient of cophenetic correlation of 0.815 and consists of three major clusters. Placement of the five samples of small 13 [a 1.85 1.35 0.85 Fic. 27. Phenograms of UPGMA cluster analyses based on correlation (A) and distance (B) coefficients comparing standardized means of 42 mensural, color, and scored cranial characters for 13 geographic localities of Neotoma floridana. See text for coefficients of cophenetic correlation, and figure 8 for geographic areas included within the coded localities. 112 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY 11 II I Fic. 28. Three-dimensional perspective drawings of the projections of 13 samples (OTU’s) of Neotoma floridana (A) and 18 samples (OTU’s) of N. angustipalata and N. micropus (B, com- pare with Fig. 30) onto the first three principal components based on correlations among 42 mensural, color, and scored cranial characters. Dashed lines between OTU’s illustrate the minimally intercon- nected networks computed from the respective among-OTU distance matrices. See text for percent- ages of variation in each component, and figure 8 for geographic areas included within the coded localities. WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 113 and pallid woodrats from New Mexico, western Texas, Coahuila, and Nuevo Leon in a single cluster seems reasonable on an a priori visual basis. The inter- mediacy of specimens from locality Kk is indicated by the low correlation (0.16) between that sample and other samples from eastern and northern localities. As in other phenograms reflecting both cor- relation and distance, samples N and P (N. m. micropus) cluster separately from samples of N. m. canescens. Although specimens of canescens from locality L and specimens of angustipalata do not resemble each other in general appear- ance, these two samples are similar pro- portionally. However, as previously dis- cussed, they do not cluster together when distances are emphasized. The distance phenogram (Fig. 29) > ox p Q©H w~zeareemnzsomrereeon oOo eee Ee ———— ee -0.24 0.16 0.56 for these 18 samples of woodrats is some- what unique in that there are few small group-clusters anastomosing to form larger clusters; instead, there is a high incidence of individual OTU’s sequen- tially joining clusters composed of less distinct OTU’s. The coefficient of co- phenetic correlation between the pheno- gram and the distance matrix is 0.891. The distinctness of angustipalata is again substantiated by the distance phe- nogram, and N. m. planiceps appears more distinct than indicated by results of other analyses. It must be remem- bered, however, that the latter “sample” consists of a single young adult male. This consideration is especially important because characters of that one specimen have been used for both males and fe- males. Nevertheless, the results indicate Oo oO OD Xt PP ao vzsi=e 1 OM S| = oc = ———————— SS SSS SS 2.135 1.435 0.735 Fic. 29. Phenograms of UPGMA cluster analyses based on correlation (A) and distance (B) co- efficients comparing standardized means of 42 mensural, color, and scored cranial characters for 18 geographic localities of Neotoma micropus (A-Q) and N. angustipalata (R). See text for coeffi- cients of cophenetic correlation, and figure 8 for geographic areas included within the coded localities. 114 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY that planiceps is not especially similar to samples of either N. m. micropus, or N. m. canescens and should be recog- nized as a distinct subspecies until more specimens are available. Also, planiceps did not cluster with angustipalata as has been seen in some other analyses. Clustering relationships indicate that there is slight morphological basis for recognizing two subspecies (as has been done in the past) for the woodrats that I have considered collectively under the name N. m. canescens. Of the six sam- ples most closely clustered, three pre- viously were assigned to one subspecies and three to another. Furthermore, placement of samples F, J, and K in the various phenograms and 3-D projections together with results of univariate analy- ses clearly indicate that any subspecies boundary merely would divide the wood- rats into two groups from some arbitrarily selected place within a series of partially discordant clines. Principal components analysis of these 18 samples considered 87.2 percent of the total variation when five compo- nents were extracted. Percents of varia- tion in the first five components con- sidered sequentially are 37.2, 30.3, 9.5, 5.7, and 4.5. When projected onto the first three components (Fig. 28), which contain 77 percent of the total variation, the results are similar to those seen in the distance phenogram. The impression of close relationship among the six sam- ples of N. m. canescens from northern and eastern localities in the range is maintained; samples D, L, and K con- nect directly to this “cluster” but do not connect directly to each other. Sample R (angustipalata) constitutes the most distinctive OTU and connects only to sample L. Sample M, which serves as an “intermediate”, connects to sample K, then serves to connect samples E and O on one “lineage”, F and J on a second, and N on a third. Sample N connects first to sample P and at a much greater distance to sample Q (planiceps). The placement of OTU’s and the con- nections shown by the minimally con- nected network of the 3-D projection for N. micropus is congruent with the no- menclatorial arrangement of these wood- rats proposed here. Figure 30 illustrates the placement of OTU’s on the first and second components better than can be seen in the 3-D drawing (Fig. 28). This figure represents a two-dimensional scat- ter-diagram of the OTU’s on these two components; the three-dimensional mini- mally interconnected network has been added together with the distance coef- ficients from the original distance matrix for all directly connected OTU’s. In viewing this figure it should be noted that discrepancies between apparent and computed distances separating OTU’s are accounted for in the third dimension, which is illustrated in figure 28. Neotoma angustipalata, recognized as a distinct species, is well separated to the left and to the front of the plotting sur- face on figure 30. Furthermore, angusti- palata is not connected to any of the samples of micropus from geographically contiguous localities. N. m. planiceps is well separated from all other OTU’s, especially on the second component, and connects only to a sample of N. m. mi- cropus from adjacent Tamaulipas. The two samples of N. m. micropus are closely placed on the first and second principal components and are situated in a position nearly intermediate between the nearest OTU’s, representing canescens and planiceps. They connect directly to planiceps and to the sample of canescens from adjacent Nuevo Leén and Coahuila. The remaining OTU’s all representing samples of canescens, are placed in a single “cluster”, showing the trend _ to- ward smaller individuals in the south- western parts of the range (samples E, O, J, and M) and the intermediacy of samples F and K. Simultaneous Treatment of Three Species——The correlation phenogram that was computed when the 42 charac- ter states of the 31 samples (all three species ) of woodrats were treated simul- taneously is shown in figure 3l. This phenogram has a coefficient of cophe- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 115 De... as C 0.42 = 0.56 »° & 0.57 & fe H Bee 0 48° ; je 0.74 % E s . Gl” A E1EOO ee ae o e N Saas 0 © al : ¢ ‘ ‘ oi e ; y @? ue \ 0.66 : a wa xs ‘ ef i \ 0.99 97% a eX Wa ya wy) ‘ OR ! 1 ie SG OEE (dorsal. i Te, H Z \ Oo af . ees Ce) / poss é 0.81 ; : aS / 1.58 / / / At u¢ Pe Ui 5 i II v a Ge Hi 107, / ae GA ¢: We yy ise Be we N/ 0.66 OR veP / ri / ! t / ' ! ! ‘ / 1,31 ' L ‘ i J I 1 4 ' ' ' <2 I Fic. 30. Two-dimensional drawing of the projections of 18 samples (OTU’s) of Neotoma micropus (A-Q) and N. angustipalata (R) onto the first two principal components based on correlations among 42 mensural, color, and scored cranial characters. Dashed lines between OTU’s illustrate the min- imally interconnected networks computed from the among-OTU distance matrices. Distance coeffi- See text cients from the distance matrix are given for each pair of directly connected localities. for percentage of variation in each component, and figure 8 for geographic areas included within the coded localities (nominal taxa are shown with distinctive symbols). This figure should be compared with figure 28B. cies, with the obvious exception of the netic correlation of 0.863 with the corre- lation matrix. The major separation at a placement of N. f. rubida. correlation of —0.325 separated all sam- The distance phenogram (Fig. 31) ples of N. floridana into one cluster and for computations on samples of the three all samples of N. micropus with the species simultaneously has a coefficient single sample of N. angustipalata into of cophenetic correlation of only 0.714 the other. This phenogram corresponds to the distance matrix. This is unusually well with results seen in 3-D projections low compared to the coefficient (0.863) for the correlation phenogram and ma- and in distance phenograms for both spe- 116 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY R R A L B 1 N 6 P 8 Q 9 A 10 H 7 8 11 Cc 12 G 2 i 3 K 4 E 5 ry) 13 M A F B J Cc D G 1 t 2 H 3 D 4 K 5 L 7 E 11 M 12 tr) 13 F 6 J 8 N 9 P 10 Q -0.325 0.175 0.675 1.74 1.14 0.54 Fic. 31. Phenograms of UPGMA cluster analyses based on correlation (A) and distance (B) co- efficients comparing standardized means of 42 mensural, color, and scored cranial characters for 31 geographic localities of Neotoma floridana (1-13), N. micropus (A-Q), and N. angustipalata (R). See text for coefficients of cophenetic correlation, and figure 8 for geographic areas included within the coded localities. WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 117 trix; normally the highest coefficient is for distance. Four major clusters are seen in the distance phenogram. At the first bifurcation, all samples of micropus are placed together in two clusters. Neo- toma angustipalata forms a third “clus- ter” by itself that connects at a distance of 1.53 to the fourth major cluster (com- posed of the 13 samples of N. floridana). Within the floridana cluster, two OTU’s (5 and 13) are relatively distinct; this was expected for sample 13 (rubida) but I anticipated that sample 5 (attwateri from north-central Kansas ) would cluster either with other samples of attwateri or with samples of campestris. The three samples of campestris form a distinct cluster, and baileyi appears relatively dis- tinct, but joins the samples of smaller attwateri before they are joined by the samples of larger attwateri. Of the two major clusters of N. mi- cropus, one consists of the 14 samples of N. m. canescens and the other of the two samples of N. m. micropus along with N. m. planiceps. The two samples of N. m. micropus join at a distance of 0.63, and planiceps joins that couplet at a dis- tance of 1.02. In the distance phenogram shown in figure 29, canescens from local- ity M appears more like samples of N. m. micropus than like other samples of canescens; in the large phenogram, sam- ple M is placed with other samples of canescens. In both phenograms, the small, pallid woodrats from New Mexico, western Texas, and adjacent Mexico tend to form a relatively homogeneous and distinct subgroup. This relationship is seen also when OTU’s are projected onto principal components. I considered the possibility (see introductory remarks in the account of N. micropus above) of applying the available name N. m. leu- cophea to those woodrats from localities E, F, J, M, and O. However, there are no indications of well marked “steps” in clines of variation, and no apparent past or present geographic or physiographic barriers. Because there seems to be no way to designate a meaningful boundary between the two potential taxa, this ar- rangement was rejected. Principal components analysis on the character correlation matrix extracted 50.1, 15.9, 7.8, 6.6, and 3.9 percent of the variance for the first five principal com- ponents, respectively. Of this, 73.8 per- cent is in the first three components. A three-dimensional drawing of the 31 OTU’s projected onto the first three com- ponents is shown in figure 32. The mini- mally connected network has been omit- ted from the 3-D projection but is given in figure 33, which shows two two-dimen- sional scatter diagrams wherein the 31 OTU’s are projected onto components one and two (33A) and one and three (33B). The three drawings considered simultaneously show undistorted spatial relationships of the OTU’s on the prin- cipal components and further elucidate the congruency between the results of these analyses and the nomenclatorial ar- rangement I have applied to the wood- rats. Samples of micropus and floridana are completely separated on the first component, although micropus sample L nearly overlaps floridana samples 1 and 2. Also on the first component, angustipalata is placed with floridana and widely separated from micropus. Neotoma floridana rubida is placed to the left of other samples of floridana on that component, and the samples of small southwestern N. m. canescens are placed to the right of the samples of larger northern and eastern canescens. On the second component, samples of N. f. campestris are separated from other samples of floridana. Sample 5 (attwa- teri from north-central Kansas) is be- tween samples of campestris and those of other attwateri, but on a tangent rela- tive to the first component. This sample of attwateri (5) and the sample of baileyi (1) both are connected to other samples of attwateri, but not to samples of cam- pestris or to each other. In part, this pro- jection implies that baileyi might best be considered as the same taxon as attwateri ( baileyi is the oldest available name and would be the valid name for all if this 118 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY <) Ons @) 6 ) On 8 o @ 7 12 Y 2 4 © ¢ ) 3 O) on 5 O Qt OF II Fic. 32. Three-dimensional perspective drawing of the projections of 31 samples (OTU’s) of Neotoma floridana (1-13), N. micropus (A-Q), and N. angustipalata (R) onto the first three princi- pal components based on correlations among 42 mensural, color, and scored cranial characters. See text for percentages of variation in each component, and figure 8 for geographic areas included within the coded localities. This figure should be compared with figure 33. was done). Neotoma angustipalata is distinctly separated from samples of flori- dana on the second component. When samples of N. micropus are considered, it can be seen that the second component separates samples P and N (N. m. mi- cropus) and sample Q (planiceps) in one direction and sample E (canescens from the Texas panhandle) in the other; re- maining samples are similarly placed on these two components. The third component further sepa- rates angustipalata from all samples of floridana and clearly demonstrates the distinctiveness of micropus samples D and L from floridana samples 1, 2, 6, and 9. On both the first and second compo- nents, these six samples are placed in relatively close proximity. From the dis- tance matrix, it can be seen that the separation between D and 2 is 1.15, that between L and 8 is 1.04, that between L and 6 is 1.05, and that between L and 2 (which appear especially close on the 3-D projection) is 1.16. In the minimally connected network, only two intercon- necting lines (the minimal number) con- nect OTU’s of different species. Neotoma angustipalata is connected to N. f. attwa- teri sample 7 at a distance of 1.58, and N. m. canescens sample D is connected to N. f. attwateri sample § at a distance of 0.94. With the exceptions of a relatively high degree of distinctiveness in attwa- teri sample 5, the apparent affinities of baileyi with attwateri, and the relative distinctiveness of samples of canescens from southwestern localities, the classifi- cation I have employed is in accord with the results of this principal components analysis. It clearly shows that angusti- palata is phenetically distinct and that floridana and micropus are morphologi- cally distinct and should not be consid- ered conspecific. The distinctiveness of 119 WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA ‘SE VINSY YWA\ poreduioo oq plnoys sinsy sty, *(sjoquiAs SATIOUTSIP YA UMOYS 1B BXB} [eUIUION ) SaTzI[VOOT popoo oy} UIYWA popnypour svore orydes8003 10y g dINSy pue ‘UoUOdUIOD YoRo UT UOTPLIvA jo osvzusoIed OF 7X9} 99g ‘saTPTBOOT peyoouuo9 AQooirp fo ued yoro 10F (A[TUO VW) UOAIS O1e XTI]VBUT DOUBISIP 9Y} WOIF SJUSTOFI0O 9ouRysICy “XEQeUL souRysIp ,LO-Ssuoure oY} WoIZ poynduiod syIOMyoU pos}DIUUOdI}UI AT[euTUTUE oY} 97eIYSNTE Y UO $.4,LO UeeMjeq soul] poysep SY], “stojovIvYyo [eLUPIO paioos pue “1ojoo ‘Ternsuoul ZF Suoure suoNLpe1109 uO paseq syuauodtuoo [edriourd (gq) pAryy pue ysiy pue ‘(y) puooss pure ysiy 94} OWUO (yY) pyojod -1jsn3up *N pure ‘(Q-y) sndosonu ‘yy ‘(ET-1) vuDpuoyf DUOJOaN JO (S.ALO) sedures TE Jo suonoeford ay} Fo SSUIMPIP [BUOISUBUTTIP-OMT, “EE “OTT it I ull é; i? VaaN C) v) ea £6 oO, (@ nists eS, \, 73'0 z6'0, r cd N Mn aN £10; \ 980 ) e i e 0 6x0} “! a L \ 79 or 4@ eo” III w@ ee ! See lar " sO J \ 060 y® eH no st \ 5@ \ v® e y®@ 0 \ a© : e \ f , eo H® 90 \ e »? \ ) rad) 1O \ ° E re) i fe) Ne 8 ' {0 nO 790 | a O ‘ 2 4 TOT ‘ 01 120 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY N. f. campestris and N. f. rubida as com- pared to contiguous samples of N. f. at- twateri can be seen also. Similarly, N. m. micropus and N. m. planiceps are shown to be clearly distinct from N. m. canes- cens. There appears to be some basis for recognition of another name (N. m. leu- cophea) for specimens from New Mexico, western Texas, and adjacent Mexico; the question that arises is basically a matter of one’s concept of the limits of a sub- species. Were Neotoma floridana baileyi and N. f. attwateri geographically contiguous, I would consider them a single taxon. However, baileyi clearly is isolated, and has been shown to share nearly equal affinities with campestris. Furthermore, it is relatively unique with respect to qualitative cranial characters. Therefore, I think it best to continue to recog- nize baileyi as subspecifically distinct. To consider those rats from locality 5 (north-central Kansas) as a new subspe- cies on the basis of their uniqueness in the distance phenogram and 3-D projec- tion just discussed would be, in my esti- mation, a gross error. The sample from that area is small and there is no reason to believe that the population of wood- rats there represents a truly distinctive evolutionary unit. It is possible that the large size and apparent distinctiveness of woodrats from the narrow zone of sec- ondary intergradation between campes- tris and attwateri reflect some effect ( pos- sibly heterosis) of recent hybridization. There is some basis for naming as a distinct subspecies those large and rela- tively distinct populations (L) of mi- cropus from southern coastal Texas. They are different from woodrats in adjacent Tamaulipas (and herein are assigned to different subspecies), but study and sta- tistical analysis of all available specimens from southern Texas indicate that the pattern of variation to the north and west from coastal Texas is clinal and that woodrats from locality K (non-coastal southern Texas ) show varying degrees of intermediacy between those from locality L and those from localities J, I, and M. Discriminant Function Analysis Discriminant function analysis has been employed by Lawrence and Bossert (1969) to distinguish dog-coyote hy- brids; these authors (1967) also were able to identify skulls of wild canids with this relatively sophisticated technique. Anderson (1969:44) conducted a pre- liminary discriminant analysis to dis- tinguish specimens of Neotoma micropus and N. albigula; he found that members of the two species could be separated better in this way than by factor analysis. I have used discriminant function analysis (MULDIS) to compare indi- vidual specimens of woodrat taxa as fol- lows: Neotoma floridana baileyi with N. f. attwateri; N. f. campestris with N. f. attwateri; N. f. campestris with N. m. canescens; and N. f. attwateri with N. m. canescens. In some cases, specimens from geographically intermediate sam- ples, suspected hybrids, or laboratory- bred hybrids were included as a third group for comparison with reference samples. In one instance, only the ref- erence samples were compared. These analyses were conducted to determine: 1) if the nominal taxa are sufficiently and consistently distinctive at the level of the individual in the 10 cranial dimensions, four color reflectance scores, and three scored qualitative characters so that dis- criminant analysis could distinguish members of the different taxa by differ- entially weighting characters to accentu- ate existing differences; 2) if known hy- brids between floridana and micropus could be distinguished by use of the dis- criminant technique; 3) if discriminant scores of suspected hybrid individuals would be similar to those of known hy- brids; and 4) if a series of discriminant multipliers could be calculated from identified reference samples so that fu- ture material could be identified by mul- tiplying the values of the same 17 char- acters by the discriminant multipliers and then summed to compare discrim- inant scores. The characters used in these analyses have been described previously. WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 121 Males and females of Groups VI-VIII were treated together. In addition to the calculations of values previously mentioned, MULDIS also computed the “best” placement of each individual in the original reference samples and indi- cated the number, if any, that would “best” have been included in the other reference sample. In comparisons of N. f. baileyi and N. f. attwateri, 18 specimens of baileyi from several localities within the range of the subspecies were compared with 36 specimens of attwateri from localities 6-10. Because baileyi is geographically isolated, there are no suspected natural hybrids or geographic intermediates and discriminant scores for a “test” sample of laboratory hybrids were not computed. Although there is no overlap between the discriminant scores of the two sub- species (Fig. 34), one specimen of attwa- teri is more like the reference specimens of baileyi than like other specimens of attwateri. The mean and range (in parentheses ) of discriminant scores from attwateri and baileyi, respectively, are 14.20 (12.32-16.85) and 18.87 (17.47- 20.92). When the single “wrongly placed” specimen of attwateri was re- moved, the upper extreme of that sam- ple was reduced to 15.82. From the list of discriminant multipliers computed for these two taxa (Table 14), it can be seen that measurements of interorbital con- striction, breadth of the rostrum, length of the nasals, and morphology of the sphenopalatine vacuities best serve to distinguish baileyi from attwateri. Only condylobasilar length and length of max- illary toothrow were weighted at espe- cially low levels. These results indicate that baileyi and attwateri are generally distinguishable at the level of the individual, but that a few individuals of one taxon may closely resemble members of the other morphol- ogically. As previously discussed, most skulls of baileyi can be identified by the three scored cranial characters included in the discriminant function analysis. Of these, only one (sphenopalatine vacui- ties) was weighted relatively high. Re- flectance values were computed by anal- ysis of variance and SS-STP to be signifi- cantly different between baileyi and at- twateri, but for some reason, probably because of the high within-group vari- ance, these scores are not weighted no- ticeably higher than cranial dimensions. A frequency histogram showing sepa- ration of the reference samples of N. f. campestris and N. f. attwateri together with the projection of four specimens from locality 5 (all from Ellsworth County, Kansas) is shown in figure 35. No overlap between the two samples is observed and all specimens are “best” considered in the sample with which they were originally placed. The mean and extreme (in parentheses) discriminant scores for the 36 specimens of attwateri (localities 6-10) and 27 specimens of campestris (localities 2-4), respectively, are 13.64 (11.05-15.32) and 18.66 (16.35- 20.66). The discriminant multipliers cal- culated for the comparisons are shown in table 14. That for rostral breadth is especially high and those for reflectance of blue and green are relatively high. Color reflectance was expected to be heavily weighted, considering the differ- ences in color between members of the two subspecies. Several characters are weighted relatively low, especially con- dylobasilar length. Within the attwateri reference sam- ple, there is a slight but noticeable ten- dency for the discriminant scores of spec- imens from localities farthest from the range of campestris to be least like those of campestris. Furthermore, within the campestris reference sample there exists an obvious tendency for specimens from locality 4 (adjacent to the range of attwateri) to cluster toward the attwateri reference sample. The specimen of campestris (KU 119700) with the second lowest discriminant score (17.11) is from a locality in Russell County, Kansas, only one mile west of the Russell-Ellsworth County boundary, which I have con- sidered the general line of demarcation between the two races. Discriminant MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY fo] rS6'0 Ice'T L88'T S6L'°0 VI9O— £660 Soe’o 8910 OFS O— PST O LESS 0LO'O0— 0) 869: 0— c60°0— 0100 80¢'0— SUBISIUDI sndoswouu ‘"N 149JDNY}D puppiJoyy “"N 000°0 660'0 GOs EO) IL8'0 69 0— 901 0— ¢s0'0 Pr6'0 OCT ia L69°0 L8o'0— LIGO— OL8'0 CIC O— 8sc0— €86'0— 61¢'0 666 0 099'0— 9LT T 190'0— C8PF'0 OOT'O 610 0— 680'0 S96 0— g98'T 16¢°0 L89°0— 867 0— S910 10€°0 cs¢0'0— suaasaup9 sndowou * Nj pue pue 1ajDN}WD DuDpwoy. *N pue suysadups Duppuoy *"N — stqsadwipa Duppi.oyf *N sioydy[nyy yUeUTWULIOSIG, rayovieyy Ska -— = 2 tone [2}0} douRPITOY Clea. 2 ta cee UdIIB JO VOURO]OY Grohe ae ee anyq Jo sourpoyoy OPSiOL8 ” .2 ~~ Sea oe por Jo vourpIHoy 6960 Teta oe tel ee saqmova sunejedousyds fo Asopoydioyy ISeOr 9 9 |. west ayeyed jo ulsieul Iote}sod yo ABopoydiojy FOCOle Oe 8 aaa aurds jeyeed IoLiajue Jo Asofoydioyy OSS Ss PSs SS ee sjeseu Jo WysuUeT SL Om Ee ps espuq je}eed jo ysueT [LLO0° = =i“ =p eee A\OIY}OO} AIB][IXVUL JO YYBUI| Iv[OoATy VOCs « MI ie sa war eee win.ijso1 Jo Yypeeig PAC el eet aa oe ae wM.ijso1 JO YysUI'T LeSos SE ee Sploysvul }e Ypeolg 686'0 ~~ WOQOLYSUOD [P}IGIOLO}JUT JSBI"T OG Oe IPpesiq oMVUlosAz DQ Ysue] Iepiseqo[Apuor) GCOi() nn ane [[2AS JO Wsue] ysoyvoig 119]DN}WD DuDpLwoyf *N pue thajwq vuppi.oy *N ‘poyBoIpU! BX} }VIPOOM Jo sited IM0F 9y} yO sojdures souaseyor Jo sasAfeuR uoHouNg yuRUTULIOSIp Aq suopoRieYyo LT JO Yors 10F poyndiwoo sioydyynur yuvurposiq “pl ATAVL WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 123 Frequency 12.5 13.5 14.5 15.5 16.5 17.5 18.5 19.5 20.5 21.5 Discriminant Score Fic. 34. Frequency histogram of discriminant scores computed by discriminant function analysis comparing individuals of Neotoma floridana attwateri (6-10) and N. f. baileyi (1). See figure 8 for geographic area of origin, indicated by numerals on the histogram, of each specimen. scores of two of four specimens from just east of this line are intermediate between the extremes of the two reference sam- ples. Although the scores of the other two are within the extremes for attwateri, they are near the campestris side of the histogram. As determined by these analyses, N. f. campestris and N. f. attwateri are con- sistently distinct at the level of the indi- vidual. The area of presumed secondary intergradation apparently includes both Russell and Ellsworth counties, Kansas, because specimens from this area tend to have convergent, albeit distinct, dis- criminant scores. Additional specimens from the relatively narrow zone of con- tact between the two taxa might reveal Frequency Discriminant Score Fic. 35. Frequency histogram of discriminant scores computed by discriminant function analysis comparing Neotoma floridana attwateri (5-10) and N. f. campestris (2-4). Solid lines enclose in- dividuals that were included in the two reference samples, whereas dashed lines enclose individuals of attwateri (5) from a geographically intermediate area whose scores were computed on the basis of discriminant multipliers computed during comparison of the reference samples. See figure 8 for geographic area of origin, indicated by numerals on the histogram for each specimen. 124 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY overlap of discriminant scores on the histogram between the two_ reference samples. The histogram showing frequency of discriminant scores for 41 Neotoma mi- cropus canescens, 27 Neotoma floridana campestris, and 11 laboratory-bred hy- brids is shown in figure 36. Mean and extreme (in parentheses) scores for canescens and campestris are, respec- tively, -12.33 (-14.20 --9.31) and -5.53 (—7.82 - 3.80). In each reference sam- ple, one individual has a discriminant score that approaches the scores of the other species, but in both instances the specimens are “best” included with the conspecific reference sample. With these two individuals included, the reference samples are separated by 1.49 units, whereas without them the separation would have been 3.71 units. As noted above, specimens of campestris from lo- cality 4 tend to have scores that are more like those of attwateri than are the scores of specimens from localities 2 and 3. A similar relationship is not observed when campestris is computed against canes- Frequency cens. Discriminant multipliers computed for the 17 characters are given in table 14. One cranial dimension, interorbital constriction, is weighted especially heav- ily. This undoubtedly results from the previously discussed (see account of qua- litative cranial characters) differences in the morphology of the interorbital region. All three scored cranial characters are weighted relatively heavily, but only morphology of the sphenopalatine vacui- ties is given an absolute multiplier value greater than unity, indicating that differ- ences in the vacuities between the two races are more consistent (less variance ) than differences in the anterior spine and posterior margin of the palate. Reflec- tance of red and green and the summa- tion of all reflectance readings are weighted low, but the reflectance value for blue is computed an above average discriminant multiplier. Discriminant scores were computed for 11 laboratory-bred hybrids and pro- jected onto the histogram. Of five indi- viduals of the first filial generation (F1), two have scores in the same frequency Discriminant Score Fic. 36. Frequency histogram of discriminant scores computed by discriminant function analysis comparing Neotoma floridana campestris (2-4) and N. micropus canescens (B-D). See figure 35 for significance of solid and dashed lines. Fl and F2 indicate laboratory-bred hybrids between the two taxa of the first and second filial generations, respectively. F3 denotes a back-cross individual whose non-hybrid parent was an N. f. campestris. See figure 8 for geographic area of origin, indi- cated by numerals and letters on the histogram, for other specimens. WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 125 class as the “aberrant” individual of campestris discussed above. The other three have scores between the extremes of the canescens reference sample. These results are somewhat surprising, because I had expected F1 hybrids would fall in the 3.7 unit zone between the two refer- ence samples. Considering the results of discriminant analysis of Fl hybrids, the scores of specimens from the second filial generation (F2) are about as expected. Again there is a tendency for the scores to be more like those of canescens than those of campestris. The range of varia- tion is slightly greater than that of F1’s, and the range of variation is less than that seen for the two reference samples. Only one adult (of 30 weeks of age or more) back-cross individual was avail- able. This woodrat (F3), the progeny of an F1 hybrid mated to a campestris, has a discriminant score of —7.52, which placed it within the campestris range on the histogram. The separation between N. f. cam- pestris and N. m. canescens is adequate to demonstrate that the two can be dis- tinguished by discriminant analysis. In the absence of the two extreme indi- viduals, the separation would have been relatively great considering the small number of characters used and the close relationship of the two species. How- ever, larger samples probably would show that the two specimens discussed are not “aberrant,” but rather near the extremes of discriminant scores of the two taxa compared. Reference samples of 36 N. f. attwa- teri from localities 6-10 and 41 N. m. canescens from localities B-D were used to compute discriminant multipliers in comparisons of these two taxa. Mean and extreme (in parentheses) dicrim- inant scores for individuals of the refer- ence samples were 12.77 (10.29-14.98) for attwateri and 19.54 (17.34-21.56) for canescens. All individuals of both sam- ples were computed to be in the “cor- rect’ sample; when discriminant scores were plotted on a frequency histogram (Fig. 37), none of the members of either reference sample had deviate scores re- sulting in placement in a class disjunct from other classes of the species. As shown in table 14, rostral breadth again is weighted relatively heavily. Other characters that appear to differ consistently between the two taxa are interorbital constriction, sphenopalatine vacuities, and color reflectance of blue. The summated reflectance score (total) was given a multiplier value of 0.0, and thus was of no use in distinguishing in- dividuals of N. f. attwateri and N. m. canescens. Mastoid breadth also was shown to be of little value in distinguish- ing members of the two taxa. Discrim- inant scores of 40 woodrats from a variety of sources were computed and projected onto the frequency histogram (Fig. 37) with scores of specimens in the two ref- erence samples. This test group is com- posed of the following specimens: 1) 12 laboratory hybrids of the first filial (F1) generation; 2) six hybrids of the second filial (F2) generation; 3) two back-cross hybrids resulting from the mating of F1 hybrids with micropus (M3); 4) eight specimens that I had previously identified as micropus (S1) from the locality of sympatry of the two reference taxa (3 mi S Chester, Major Co., Oklahoma); 5) four specimens that I had previously identified as floridana (S2) from the lo- cality of sympatry; 6) five specimens previously identified as natural hybrids from that locality (S3); 7) and three specimens from 8.9 mi S Aledo (14 mi SW Fort Worth), Parker Co., Texas (T), that did not appear to be typical repre- sentatives of floridana. Six of the Fl hybrids have discrim- inant scores between the extremes of the scores of the parental species, as was expected for all. Five of the other six have scores that were near the lower end of the range of N. m. canescens, but the score of the sixth is like that of typical canescens. Most F1 hybrids apparently tend to be intermediate in the 17 char- acters employed, but somewhat more like micropus than floridana. Second generation hybrids (F2) are, on the aver- 126 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY Frequency 15.2 Discriminant Score Fic. 37. Frequency histogram of discriminant scores computed by discriminant function analysis comparing Neotoma floridana attwateri (6-10) and N. micropus canescens (B-D). See figure 35 for significance of solid and dashed lines and see figure 8 for geographic area of origin, indicated by symbols on the histogram, for most specimens. Those not included in figure 8 and their identifying symbols (in parentheses) are as follows: laboratory-bred hybrids between the two taxa of the first and second filial generations (Fl and F2, respectively); laboratory-bred back-cross hybrids whose non-hybrid parent was N. m. canescens (M3); specimens from 3 mi S Chester, Major Co., Okla- homa, that had been identified previously as N. micropus, N. floridana, or hybrids (S1, $2, and $3, respectively ); and specimens from 8.9 mi S Aledo, Parker Co., Texas, that were suspiciously atypical in color (T). age, slightly more micropus-like than are Fl hybrids, and also are more variable. The score of one approaches the upper extreme of micropus scores; that of an- other approaches the most micropus-like floridana scores. None overlapped the scores of floridana. One of the two back- cross specimens (M3) demonstrates no perceptible affinities for floridana, but the score of the other is similar to those of F1 hybrids. Considering animals from natural pop- ulations, the three suspiciously-colored specimens from Parker County, Texas, all have discriminant scores typical of floridana. The scores of two are above the mode for the floridana reference sam- ple and one is near the upper extreme; none could be identified as “hybrid” or “intergrade” on the basis of discriminant scores. The scores of seven of the spec- imens from the area of sympatry that had been identified as micropus (S1) are typical of the scores of reference mi- cropus. The score of the eighth (17.68), however, is larger than that of only one individual in the micropus reference sam- ple, and thus similar to scores of labora- tory-bred Fl hybrids. All four of the specimens identified as N. floridana (S2) have scores within the extremes of ref- erence floridana. One, in fact, is near the lower limit for floridana, but two others have scores slightly above the mode. Of greatest interest is the placement on the histogram of scores of the five speci- mens identified as natural hybrids; all have scores between the extremes of the WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 127 two reference samples and are in fre- quency classes with known FI and F2 hybrids. These results show that laboratory reared hybrids between N. floridana and N. micropus tend to have discriminant scores intermediate between those of non-hybrids, but when not intermediate they are more like micropus than flori- dana. Second generation hybrids again appear to be more variable and also more micropus-like than F1_ hybrids. Backcross hybrids, as expected, tend to be either intermediate or more like their non-hybrid parent. Woodrats from the known area of sympatry in northern Oklahoma may be similar to members of either species, or they may be inter- mediate. Results of discriminant func- tion analysis further substantiate conclu- sions by Spencer (1968) and by me that the two species interbreed at the locality of sympatry. In addition, it can be seen that identifications based on visual anal- ysis of pelage and cranial characteristics are highly reliable, as indicated by re- sults of discriminant function analysis. Although the series of specimens from Parker County, Texas, near the western edge of the range of the species in north- ern Texas, are suspiciously grayish, spec- imens from that locality are either typical floridana or at least much more like flori- dana than micropus. It appears that discriminant function analysis is a sophisticated statistical tool that has tremendous potential in studies of geographic variation, especially in lo- cating zones of marked morphological change (see Rees, 1970, for comparable results in studies of white-tailed deer), and for distinguishing natural hybrids. However, the observed tendency for floridana-micropus F1 and F2 hybrids to be more like micropus than floridana in- dicates that hybrids of known ancestry (laboratory-bred) should be used as “controls” whenever possible in discrim- inant analyses to determine where na- tural hybrids likely will score relative to the scores of the parental species. NON-MORPHOLOGICAL CHARACTERS COMPARATIVE REPRODUCTION In view of the emphasis placed on “reproductive isolation” in the biological species concept (Mayr, 1965:19), studies of reproductive patterns and habits con- stitute a major aspect of investigations of closely related taxa. It has been sug- gested (erroneously I think) that evi- dence of hybridization among animals having internal fertilization implies con- specificity. As presently understood, laboratory experiments in hybridization merely provide an indication of the pres- ence, absence, or efficiency of isolating mechanisms, especially postmating mech- anisms (see Mayr, 1963:92). Brand and Ryckman (1969) demonstrated a clear understanding of this concept in their interpretations of studies on Peromyscus. Hybridization in natural habitats is more often indicative of conspecificity, but as stated by Mayr (1969:195) “allopatric forms that hybridize only occasionally in the zone of contact are full species.” In fact, populations that have diverged mor- phologically during isolation often hy- bridize when geographic contact is re- established until such time as selection has established isolating mechanisms or reinforced incipient mechanisms (Key, 1968). Premating isolating mechanisms, such as habitat isolation, clearly are tenuous forms of isolation, but they may be sufficiently effective to prevent loss of species integrity as a result of hybridiza- tion. The emphasis on reproductive iso- lation would seem better placed on the abilities of taxa to maintain their respec- tive specific integrities rather than on the production of hybrids. Nevertheless, studies of hybridization both in the labo- ratory and the field often serve as im- mensely useful indicators of animal rela- tionships. Breeding Cages——Two cages were 128 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY constructed especially for breeding woodrats in the laboratory. Each was 60 by 60 by 18 inches with a #-inch plywood floor and hinged top. The sides and top were of }-inch hardware cloth secured on the inside of a wooden frame. Two other cages constructed for other pur- poses were found to be excellent breed- ing cages. These were 60 by 18 by 18 inches with }-inch plywood floors and sides, and with hinged tops of hardware cloth on wood frames. Internally, these cages were constructed so that remov- able hardware cloth partitions on wood frames could be inserted at one-foot in- tervals. The partitions had sliding metal doors allowing woodrats to be separated or penned together easily and without handling. When these four cages were in use, an upright rack of four metal cages, 48 by 24 by 18 inches, frequently was converted into two “two. story” breeding cages by replacing the metal trays that served as the floor of the upper and third (from the top) cages with trays 24 by 24 inches. Attempts to breed woodrats in unconverted cages, 48 by 24 by 18 inches, and the smaller 24 by 24 by 18 inch cages never were successful and often resulted in death of one of the rats. Recognition of Breeding Readiness. —External indications of breeding con- dition for both sexes were described by Rainey (1956:605-609) for Neotoma floridana, and by Raun (1966:14-17) for N. micropus. In breeding males the testes become noticeably enlarged and scrotal; the swollen convoluted cauda epididymis forms a conspicuous bulge (Linsdale and Tevis, 1951:354), and the skin of the scrotum appears thinner, more darkly pigmented, and less haired than in nonbreeding males. In both the field and the laboratory it was noted that these characteristics are somewhat more easily discernible in breeding floridana males than in breeding micropus males. Also in floridana, the testes often descend farther into the scrotal sac than do the testes of micropus males. Healthy adult laboratory males of both species had ap- parently viable sperm in the epididy- mides and testes throughout the year, even when the testes were abdominal and reduced in size. In sexually inactive females, the vulva is imperforate and cornified, the clitoris is small and white or pinkish, and the teats are small. Breeding females are easily distinguished by a turgid, perforate vagina and an en- larged, vascularized clitoris. Age at Sexual Maturity—Both male and female woodrats born early in spring usually appear to be in breeding condi- tion by late summer. The testes of young males are smaller than those of adults and the cauda epididymis protrudes less; both the epididymides and testes contain sperm. Several attempts were made to place first-year males of micropus with adult breeding females, but each attempt was necessarily terminated to prevent the male from being killed. A first-year floridana male was placed with an adult female floridana from 8 August until 4 September 1967; although the two were compatible, they were never observed to display sexual interest nor to copulate; no litter resulted. The same male sired several litters during the 1968 and 1969 breeding seasons. In late summer of 1968, two first-year females of each species were placed with adult males of their own species for more than two weeks each, but no young was born to either. One of three subadult female N. f. baileyi (Table 15), obtained in late August of 1968, was nursing three young judged to be less than two weeks of age; the other two females were neither pregnant nor lactating. At least occasional females bear young in the natural environment late in the season of the year in which they are born. Brown (1969:538) found that female Neotoma mexicana born in the first lit- ters of spring (April and May) normally produce litters in June and July, often while still at least partially in juvenile pelage. This clearly is not the case in either N. floridana or N. micropus. Young of both sexes born as late as August were consistently successful breeders in the WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 129 laboratory by March of the following year. Breeding Seasons.—As indicated by age composition of samples of N. micro- pus and as reported by Raun (1966:14), breeding takes place throughout the year in southern populations of the species, with only a slight tendency for season- ality. Females in northern populations of micropus (Spencer, 1968:45, and Fin- ley, 1958:486; Tables 15 and 16) appar- ently begin breeding in December and January, produce at least two and prob- ably three litters before July, and some females have an additional litter and possibly two between early August and the end of October. Neotoma floridana in Kansas (Tables 15 and 16; Rainey 1956:609-613) begin breeding in Febru- ary and females bear their first litters in March or April and their second in May or June. Most females are appreciatively less active reproductively in July, then have an additional litter in August or September with occasional litters being born as late as October. As indicated in tables 15 and 16, the first litter of female N. f. baileyi is born in April, a second in June, and a third in July or August. I doubt that baileyi produces litters after mid-September, but available data do not preclude the possibility. Late sum- mer and autumn breeding in both species may involve mostly young females born earlier that year. Estrous Cycle —Techniques described by Chapman (1951:269) and Zarrow et al. (1964:36-37) for determination of estrus by examination of cells lining the vagina were attempted early in my TABLE 15. Reproductive status of adult and subadult Neotoma floridana and N. micropus females captured in Nebraska, Colorado, Kansas, and Oklahoma from September, 1966 to April, 1969. Number progeny Number collected Date progeny born _ with female Locality Date litter ( county ) captured Age bom ottot eae) oS Oe Remarks Neotoma floridana baileyi Cherry 31 Mar. Adult 13 Apr. 2 2 Cherry 31 Mar. Adult 9 Apr. 2 2 = Cherry 31 Mar. Adult f =: . No litter born Cherry 31 Mar. Adult 26 Apr. 2 2 Cherry 31 Mar. Adult 6 Apr. DD 1 iz Cherry 31 Mar. Adult Died 17 Apr.; had 3 resorbing embryos Cherry 31 Mar. Adult 16 Apr. 1 3 z Cherry 24 Aug. Subadult No litter born Cherry 24 Aug. Adult Killed 29 Aug.; had 4 embryos < 45 mm. Cherry 24 Aug. Subadult Killed 29 Aug.; not pregnant Rock 21 Aug. Subadult 1 2 Killed 29 Aug.; not pregnant Rock 22, Aug. Subadult Killed 29 Aug.; not pregnant Neotoma floridana campestris Logan 29 Aug. Adults (2) No litters born Logan 29 Aug. Subadult Died 8 Sept.; not pregnant Ness 4 Sept. Subadult Killed 4 Sept.; not pregnant 130 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 15.—Continued. Number progeny Number collected Date progeny bom with female Locality Date litter ( county ) captured Age born od ee) Jd 2° Remarks Ness 4 Sept. Subadult Killed 13 Sept.; not pregnant Finney 5 Sept. Subadult Killed 5 Sept.; not pregnant Finney 5 Sept. Adults (2) Killed 5 Sept.; neither pregnant Finney 5 Sept. Subadults (5) a é- Killed 13 Sept.; none pregnant Finney 5 Sept. Adults (4) as = Killed 13 Sept.; none pregnant Hodgeman 8 Sept. Adult Killed 13 Sept.; not pregnant Ellis 18 Dec. Adults (9) No litters born Ellis 18 Dec. Adults (2) Killed 18 Dec.; neither pregnant Ellis 19 Dec. Adults (2) Killed 21 Dec.; neither pregnant Russell 21 Dec. Adult Killed 21 Dec.; not pregnant Russell 21 Dec. Adults (4) Killed 13 Jan.; none pregnant Russell 21 Dec. Adult No litter born Neotoma floridana attwateri Major 31 Jan. Adult 10 Feb. 2 0 = is Major 3 Jan. Adult es a = = No litter born Douglas 3 Mar Adult 10 Mar. 2 1 Douglas 10 Mar. Adult 6 Apr. 1 3 = Douglas 10 Mar. Subadult re No litter bor Douglas 10 Apr. Adult 0 2 Killed 11 Apr.; not pregnant Major 7 June Adult Killed 7 June; had 4 embryos a a 7 Tague Number Number Number Mean Mode per matings successful Percent progeny litter litter attempted Females attempted matings success of size size mating N. f. baileyi X N. f. attwateri 1 100.0 il 2 3.0 3 3.00 TOTAL 3 2 66.7 3 4 3.5 3-4 233 N. f. campestris X N. f. attwateri males N. f. campestris 2) 1 50.0 2 1 3.0 3 1.50 N. f. campestris X N. f. attwateri I 0) 0.0 = fe? =e se 0.00 TOTAL 3 1 Soo 2) 1 3.0 3 1.00 N. f. baileyi X N. m. canescens males N. f. baileyi 1 1 100.0 2; @ 4.0 4 4.00 N. m. canescens (2) 1 0 0.0 gst ins! — A 0.00 TOTAL 2; 1 50.0 2) 2 4.0 4 2.00 N. f. campestris X N. m. canescens males N. f. campestris X N. m. canescens 6 2 BiB 8) 2 2 2.0 oD, 0.67 N. f. attwateri X N. m. canescens 1 1 100.0 0) 2 2.0 2 2.00 TOTAL o 3 42.9 2 4 2.0 2, 0.86 N. f. attwateri X N. m. canescens F,; males N. f. campestris 1 0 0.0 a mh an 0.00 N. m. canescens (2) 4 2 50.0 2 1 15 1-2 0.75 N. f. campestris X N. m. canescens 3} 2 66.7 2 3 QS 2-3 1.67 N. f. attwateri X N. m. canescens F, 6 2 Bie Eee) 3 2 2-5 2-3 0.83 LO MATE 14 6 42.9 0 6 22, 9) 0.93 N. f. attwateri X N. m. canescens F» males N. f. attwateri Il 1 100.0 1 2; 3.0 3 3.00 N. f. attwateri X N. m. canescens F2 0 0.0 _ ae eats we: 0.00 TOTAL 2 il 50.0 1 2 3.0 3 1.50 N. m. canescens X (N. f. attwateri X N. m. canescens ) male N. f. campestris X N. m. canescens 1 0 0.0 0.00 All Neotoma floridana males N. floridana 63 31 49.2 49 Dik Sip 3 1.59 N. micropus DD 10 45.5 1L9/ WG 3.4 4 IES Species-hybrids 3 2, 66.7 2 3 QD 2-3 1.67 TOA 88 43 48.9 68 fal 3.2 3 1.58 All Neotoma micropus males N. floridana 31 u 22.6 10 9 2.7 2-3-4 0.61 N. micropus 38 9 Sh 14 10 2a 3 0.63 Species-hybrids 2, 1 50.0 2 1 3.0 3 1.50 TOTAL rial W7f 23:9 26 20 ell 3 0.65 All species-hybrid males N. floridana 3 2 66.7 3 4 3.5 3-4 2.33 N. micropus 5 2 40.0 2 1 1.5 1-2 0.60 Species-hybrids 18 i 38.9 ih 9 23 2 0.89 TOTAL 26 11 42.3 12 14 2.4 } 1.00 142 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 18.—Concluded. Progeny Number Number Number Mean Mode per matings successful Percent progeny litter litter attempted Females attempted matings success loftos conte) size size mating All non-hybrid males Same species 101 40 39.6 63 61 3.1 3 1323 Other species 53 17 32/1 27 26 3.1 4 1.00 Species-hybrids 5 3 60.0 4 4 OM | 3 1.60 TOTAL: 159 60 Sia 94 91 Bip 3 1.16 All males All females 185 Hl 38.4 106 105 3.0 3 1.14 a male of the same race. Minima and maxima frequently were 33 to 36 days. No clearcut differences were observed in gestation periods of the two species either by me or by Spencer (1968). Ap- parently the gestation period normally fluctuates from 32 to 38 days with a modal duration of 35 days. Post-partum estrus and prolonged gestation appar- ently occur in both species, but the fre- quency of this phenomenon is not well known and the physiology associated therewith has not been investigated. Size of Litters. —Literature pertaining to litter size of Neotoma floridana has been summarized by Rainey (1956:613). In most populations that have been studied, females regularly produced lit- ters of one to four; occasional litters of five have been reported. Modal litter size for the species is three and the mean usually is near three. Litter size of N. micropus (Asdell, 1964:279) is appar- ently slightly smaller. Feldman (1935: 301, 302) studied members of this species from Carlsbad, New Mexico; each of 11 litters consisted of two young and in each litter both progeny were of the same Sex. Size of litters born to females of the northern N. f. baileyi in the laboratory was slightly greater than those of any of the other woodrats, as evinced both by the mean (3.5) and the mode (four). Litter size of the other two subspecies of N. floridana and that of N. m. canes- cens are comparable, having means near 3.0 and modes of three. ‘Two litters born to N. f. baileyi X N. f. campestris “hy- brid” females had only two progeny each. This may indicate some type of partial sterility, but more observations would be necessary to draw meaningful conclu- sions. The single litter born to an N. f. baileyi X N. f. attwateri “hybrid” female was the same as the modal litter size (three) of floridana females; a male of this cross sired a litter of four when mated to a baileyi female. Litter size of species-hybrids was no- ticeably lower than that of either of the parental species. Considering all mat- ings involving at least one species-hy- brid, modal litter size both of males and of females was only two and the mean for both was 2.4. Because hybrid males sired only four litters with non-hybrid fe- males and hybrid females produced only five litters from matings with non-hybrid males, it was not possible to determine unequivocally from the data whether lit- ters of hybrids are smaller because of partial sterility in both sexes or only in one. However, it can be seen in tables 17 and 18 that no hybrid female pro- duced a litter of more than three young, whereas one hybrid male sired a litter of four when mated to a floridana female. Of the 71 litters born as a result of labo- ratory matings, only two consisted of a single progeny; one of these matings was between a hybrid male and a micropus female. Mean litter size of non-hybrid females bred to hybrid males was 2.5 whereas that of the reciprocal was 2.7. Matings involving two hybrid individuals WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 143 averaged only 2.3 progeny per litter, pos- sibly indicating that hybrids of both sexes are less fertile than non-hybrids. Table 15 provides additional informa- tion on litter size and seasonal patterns of reproduction. Data presented in table 16 give further information on litter sizes and the reproductive season over a wide geographic area. Litter sizes of floridana and micropus may vary geographically, with more northerly populations having larger litters. In N. f. baileyi the mean was more than 3.5 and the mode was four progeny per litter; these data are based on successful matings in the labo- ratory, parturiation of previously con- ceived litters in the laboratory, and em- bryo counts recorded on specimen labels. All parameters studied indicate that N. f. campestris, N. f. attwateri, and north- ern populations of N. m. canescens most frequently have litters of three young each and that the mean usually is near or only slightly greater than three. South- ern populations of canescens most fre- quently have litters of two progeny as seen in table 16 and reported by Raun (1966:17) and Baker (1956:286). The correlation between latitude and litter size of mammals has been observed previously (Lord, 1960, and others) and probably has been best explained by Spencer and Steinhoff (1968), who ex- panded the theory originally put forth by Lack (1948, 1954). Individuals of northern populations have shorter breed- ing seasons and can place more progeny in subsequent generations by exerting more “energy” per litter on large litters, whereas those in southern populations are most successful by conserving “energy expended per litter and produc- ing more litters with each containing fewer individuals. In the woodrats studied, at least, information discussed previously regarding breeding seasons further substantiates the hypothesis. Sex Ratios at Birth—Sex ratios of progeny of all rats studied appear to be the typical one to one relationship. The only sample that deviates significantly (P <0.05, tested by Chi-square) is the 26 males to 12 females born to N. micro- pus females that were pregnant when captured; four of 11 litters consisted only of male offspring. However, among young micropus collected with adult fe- males there were more females than males and when the two samples are considered together the number of males only slightly exceeds the number of fe- males. Reproduction in Neotoma angusti- palata—Information on the reproductive habits of N. angustipalata is limited. Hooper (1953:9) reported two nursing young collected with a female on 19 May, and an adult female (Table 16) contained a single embryo when ob- tained in July. Two juveniles (UNAM 2166 and 2167) were collected on 6 October. Discussion and Conclusions—Repro- ductive habits of N. floridana and N. micropus vary both intra- and_ inter- specifically. In the primary study area, the first litters of micropus are born two or three weeks prior to the first litters of floridana, but the breeding seasons of the two species are otherwise approximately the same. Laboratory and field observa- tions conducted by me, and Spencer (1968) show that the two species do hybridize, and that hybrid progeny are somatically and reproductively viable. However, hybridization results in partial hybrid sterility, as evinced by reduced litter size. Reproductive isolation is at best incomplete between the two species and they probably will hybridize at all localities of sympatry, at least until selec- tion has had sufficient time to establish and reinforce a mechanism of reproduc- tive isolation. Only a single area of sym- patry is presently known and the re- sultant hybrid zone (see account of N. m. canescens above) has not increased in size during the five years that it has been under observation. Furthermore, no conslusive evidence exists to indicate that the hybridization in north-central Oklahoma is introgressive or that intro- gression has occurred elsewhere along the potential zone of contact between 144 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY the two species. Reproductive and distri- butional evidence indicate that the two species may be in an allopatric phase of speciation as described by Key (1968), wherein dynamic tension zones, such as the mixed population in north-central Oklahoma, are formed prior to establish- ment of full reproductive isolation. COMPARATIVE SEROLOGY The dependency of protein synthesis on genetic control indicates that physico- chemical and immunological characteris- tics of protein macromolecules are phe- notypic expressions of the genotype. Comparative study of proteins, therefore, serves as an important means of studying the relationships of animals (Florkin, 1964; Boyden, 1964). Evolutionary rates of serological characters undoubtedly are stochastic (Kirsch, 1969), as are those of most characters. Starch Gel Electrophoresis of Hemoglobins Electrophoretically demonstrable var- iation in mammalian hemoglobins has been the subject of much study recently. In the Carnivora, hemoglobins appear to be relatively stable even at the ordinal level (Seal, 1969), whereas in the Chir- optera variation has been reported at the familial and generic levels ( Mitchell, 1966; Valdivieso et al., 1969). Hemo- globin ionographs of primates (Neel, 1961; Ingram, 1963; Hill and Buettner- Janusch, 1964; Sullivan and Nute, 1968) and rodents (Johnson, 1968; Foreman, 1960; and others) have been shown to vary intraspecifically in many taxa. Ad- ditionally, Foreman (1964) discovered by tryptic hydrolysis that in the genus Peromyscus some _ electrophoretically identical hemoglobins are chemically distinct. Birney and Perez (1971) reported multiple hemoglobins in Neotoma flori- dana, N. micropus, and laboratory-bred hybrids of the two species. They ob- served major bands of four migration rates, designated (from slowest to fastest) 1’, 1, 2, and 3, and several minor bands that were not studied. Based on the number and position of major elec- trophoretic bands, seven distinctive he- moglobin patterns or phenotypes were described and labeled A through G as follows (Fig. 39): “A” occurred only in micropus and consisted of band 1 with a leading diffuse zone that terminated in a minor band at position 3; “B” was ob- served in both species and in hybrids and consisted of bands 1 and 2 and a leading diffuse zone; “C”, observed only in flori- dana and hybrids, was composed of bands 1, 2, and 3; “D” consisted of bands 2 and 3 with a trailing diffuse zone and also was limited to floridana and hybrids; “E” was seen only in micropus and hy- brids and was composed of bands 1’ and 2 with a long leading diffuse zone and a terminal minor band; “F” was observed only in floridana and consisted of a heavy band, considered to be band 1, preceded by a short diffuse zone; “G” was observed only in laboratory-bred hybrids and con- sisted either of bands 1’, 2, and 3 or of all four major bands. Results of electrophoresing precipi- tated globins by the urea—veronal method of Chernoff and Pettit (1964) in- dicated to Birney and Perez (1970) that the electrophoretically demonstrable var- iation resided in the beta (8) chains of ORIGIN 1 aa ea wo a | aw 3 & | 2 me = = — a a3) —— a a r=] 4 ace a, B c D E F G Phenotypes Fic. 39. Diagrammatic representation of electrophoretic patterns (phenotypes) of hemo- globins of Neotoma floridana and N. micropus. Major bands are represented by solid rectangles and minor bands by horizontal lines. The cath- ode is indicated by a minus sign and the anode by a plus sign. See Bimey and Perez (1971) for photographs of starch gels showing electro- phoresed woodrat hemoglobins. WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 145 the hemoglobin molecules. They further proposed tentative models that explained the inheritance patterns observed and demonstrated a possible sequence for the evolution of multiple beta loci in the spe- cies studied. According to the model, multiple beta loci have arisen by gene duplication so that at least three such loci (probably closely linked) now are present in floridana and at least two exist in micropus. Neotoma micropus may have either a third locus for production of the beta peptides of molecules forming band I’, or genes producing these pep- tides may be allelic with those producing the beta peptides seen in band 1. “Alleles” that produce no peptide (termed 8° al- leles) apparently are present at all beta loci. These non-functional alleles may be either deleted areas on the chromosomes or they may be areas that are physically present but under control of modifier genes, or for other reasons do not con- tribute a peptide chain. If modifier genes are involved, it is likely that the minor bands result from limited produc- tion of the same peptides that form major bands when the £ locus involved is fully active. Materials and Methods.—Studies of hemoglobin samples discussed here were conducted simultaneously with those re- ported by Birney and Perez (1971). De- tailed methods were outlined in that re- port and are only summarized here. Samples of whole blood were suspended in a trisodium citrate anticoagulate, washed three times in phosphate buf- fered saline, and lysed in distilled water. Hemoglobin phenotypes were deter- mined by horizontal starch-gel electro- phoresis in sodium borate buffer. Gels were sliced and stained in a solution of amido black in water-methanol-glacial acetic acid. The iodoacetimide method described by Riggs (1965) was em- ployed to determine that none of the observed bands resulted from polymer- ization. Laboratory-bred woodrats are not included in the discussion of hemo- globin variation presented here because they would tend to bias the frequency of various phenotypes in favor of the phenotype(s) of their parents. Results and Discussion.—F requencies of the seven hemoglobin phenotypes ob- served in natural populations of Neotoma floridana and N. micropus are shown in table 19. At the time ancestral popula- tions of floridana and micropus consti- tuted a single species, it would appear that only hemoglobin bands 1 and 2 were present. These two bands are common to both species and when they occur to- gether to form phenotype B, the patterns of the two species are essentially indis- tinguishable. When band 1 occurs with- out band 2 (phenotypes A and F), it is heavier; the band migrates slightly slower in floridana, and has a longer leading diffuse zone in micropus. The major band probably is formed by the same, or only slightly modified, peptides. The single banded situation may be the primitive hemoglobin phenotype for the two species, or phenotypes A and F both may have been secondarily derived after the two species were isolated. In any event, it appears that 6! and 2 both were present at the time of isolation be- cause both were expressed by some indi- viduals of the two species from every locality from which I have a sample of three or more individuals. The B' and £° alleles apparently originated in populations of micropus and floridana, respectively, after the two incipient species were geographically iso- lated. Although both alleles are present in the sample of woodrats from 3 mi S Chester, Major Co., Oklahoma, they have not been observed together elsewhere; B' is not known for floridana nor is £3 known for micropus. If the 8" allele ever was present in floridana, or if B? ever was present in micropus, they have been lost secondarily or exist in those species in extremely low frequency. It is not known whether the genes controlling production of the beta chains of molecules forming bands | and I’ are alleles of a single locus or if they occur at separate loci. Similarly, it is not known with certainty what phenotype results 146 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY TABLE 19. Frequency, expressed in percent, of seven hemoglobin phenotypes at selected localities of Neotoma floridana and N. micropus. Hemoglobin Phenotype Origin of samples N A B C D E F G Neotoma floridana baileyi Cherry Co., Nebraska 17, 5.9 52.9 41.2 Rock Co., Nebraska 2 MP 50.0 50.0 All localities 19 5s3 52.6 42.1 Neotoma floridana campestris Logan Co., Kansas 3 66.7 33.3 ae Finney Co., Kansas 6 66.7 33.3 Ness Co., Kansas 3 100.0 aes. Hodgeman Co., Kansas 2 oe 100.0 Ellis Co., Kansas 4 100.0 as = Russell Co., Kansas 1G 5.9 35.3 58.8 ei All localities 35 40.0 22.9 ob By Neotoma floridana attwateri Ellsworth Co., Kansas 4 75.0 _ 25.0 Douglas Co., Kansas 13 76.9 23h bea All localities 1 76.5 17.6 5.9 Neotoma floridana All localities (0) 39.4 29.6 28.2 2.8 Neotoma micropus canescens Baca Co., Colorado 29 10.3 20.7 69.0 Hamilton Co., Kansas 9) 50.0 50.0 Haskell and Stevens cos., Kansas 22 18.2 40.9 40.9 Meade Co., Kansas 6 = 50.0 50.0 Barber Co., Kansas 13 = 46.2 53.8 All localities 72 ia Jon nae 55.6 Neotoma from 3 mi S Chester, Major Co., Oklahoma Specimens morphologically like N. micropus 40.0 40.0 20.0 Specimens morphologically like hybrids 8 12S 25.0 12.5 50.0 All specimens 13 Ae a 15.4 23.1 38.4 when both genes are present (Birney and Perez, 1971). Therefore, it is not possible to calculate their frequency ac- curately. Moreover, no micropus has been observed that lacked both bands 1 and J’, but breeding data presented by Birney and Perez indicate that the ° allele also occurs in low frequency at the 1-81 locus(i) in that species, as it does in floridana. A crude estimate of the frequency of 8! and £"’ can be calculated by the Hardy-Weinberg formula, if it is assumed that 8! and £"’ are allelic, that when both are present 8” acts as a dom- inant, and that the frequency of the B° allele associated with that locus is sufficiently low to be ignored. By using this formula and following the genetic scheme proposed by Birney and Perez (1971), estimates of the frequency of all other alleles for both species can be calculated as accurately as size of available samples permits (Table 20). Changes in hemoglobin allele fre- quency of Mus musculus on the Jutland Peninsula appeared to Selander, Hunt, and Yang (1969:384) to be directional from west to east, but in the United States variation apparently is north to south (Selander, Yang, and Hunt, 1969: 285). The frequency of the £1’ allele in N. micropus fluctuates geographically, but no clear picture of the nature of this variation emerges from examination of WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 147 the three populations (Table 20) which have sufficiently large samples to warrant calculation of gene frequencies. Some individuals at all localities sampled pos- sess the allele and it probably is wide- spread in the population. However, it was not found to be fixed at any locality. The 8? locus is polymorphic at the four western localities sampled, but the 7 allele may be fixed in eastern populations of the species as indicated by the absence of phenotype A in samples from Meade and Barber counties and in the “hybrid” sample from Oklahoma. The £7 locus was polymorphic in N. floridana at only one locality (Finney County, Kansas), but additional sampling might have yielded animals of phenotype F from other localities. Two populations of N. floridana (campestris from Russell County and baileyi) appear to have slightly higher frequencies of the £' allele and consider- ably higher frequencies of the f° allele than do other populations of the species (Table 20). Of the 17 woodrats from Russell County, 16 were from a single juniper windbreak and all of these had the £° allele; hemoglobin of the only animal obtained from another windbreak several miles distant lacked band 3. None of the 13 individuals of N. f. attwateri from Douglas County lacked band 1, but hemoglobin of one of four animals from Ellsworth County (near the attwateri-campestris subspecies boun- dary ) did not form the band (Table 19). Frequency of 8° appears to be low in the sampled populations of N. f. attwateri, but the relatively high frequency of this allele in the “hybrid” population from Major County, Oklahoma, indicates that in adjacent populations of N. f. attwateri the frequency of this allele is either rela- tively high or that phenotype G conveys a strong selective advantage. Presence of hemoglobin phenotype G in several individuals from 3 mi S Ches- ter, Major Co., Oklahoma, clearly indi- cates that the two species have hybrid- ized at this locality. The phenotype has been observed previously in known hy- brids (Birney and Perez, 1971), but not in woodrats of either species from local- ities of allopatry. Functional relationships of the dif- TABLE 20. Frequency of hemoglobin alleles at selected localities of Neotoma floridana and N. micropus. Localities with samples of less than 10 individuals are not shown separately but are included in totals. Sample N Bl’ Bl Bo? [spas faio) Bom 60D Neotoma floridana baileyi Cherry County, Nebraska ILy/ 0.00 0.36 0.64 1.00 0.00 0.76 0.24 All localities 19 0.00 0.35 0.65 1.00 0.00 Ont O23 Neotoma floridana campestris Russell County, Kansas 17, 0.00 0.23 0.77 1.00 0.00 0.76 0.24 All localities 35 0.00 0.44 0.56 0.76 0.24 0.37 0.63 Neotoma floridana attwateri Douglas County, Kansas 13 0.00 1.00 0.00 1.00 0.00 0.12 0.88 All localities 17 0.00 0.76 0.24 1.00 0.00 0.13 0.87 Neotoma floridana All localities 71 0.00 0.47 0.53 0:63 ‘O”'7 0.35 0.65 Neotoma micropus canescens Baca County, Colorado 29 0.44 0.56 0.00 0.68 0.32 Haskell and Stevens counties, Kansas 22 0.23 .0:77% 0:00 0.57 0.43 Barber County, Kansas 13 0.32 0.68 0.00 1.00 0.00 All localities 74 0.33 0.67 0.00 O67, 0:33 Neotoma sp. Major County, Oklahoma 13 iS 1.00 0.00 0.32 0.68 “See text for assumptions made to calculate frequencies of these alleles for N. micropus. > There presently is no evidence that this locus occurs in N. micropus. ° Frequency of these alleles in this population cannot be calculated. 148 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY ferent hemoglobin phenotypes are un- known. The data appear to indicate a poorly defined tendency for animals from arid habitats to have fewer electropho- retic bands (thus fewer kinds of beta polypeptides) than animals from more mesic habitats. Manwell et al. (1963) found that hemoglobins of hybrids may have selective advantages over those of either parental species in some birds, but no data are available on this subject for woodrats. Immunoelectrophoresis of Esterases It has been shown that injections of whole serum alone stimulate production of a relatively specific antiserum of low antibody titer (Durand and Schneider, 1963), but that whole serum emulsified with complete Freund’s adjuvant results in an antiserum of relatively high anti- body titer and low specificity (Anthony, 1965). Because adjuvant was used in this study, antisera were of the latter type. A pilot study of precipitin tests (Boyden, 1964, and elsewhere ) indicated that either the antisera were not suf- ficiently specific to distinguish minor dif- ferences in woodrat proteins, or that pro- teins of the closely related woodrats un- der study had not diverged perceptibly at the sites of antigen-antibody reaction. Anthony (1965) found that precipitin tests were inefficient for comparing closely related races and species of the genus Canis. Micro-immunoelectrophoresis (Schnei- degger, 1955) has a distinct advantage over precipitin testing because individual arcs of precipitate are formed. However, Gerber (1968) observed that neither to- tal counts nor weighted scores based on intensity of general protein arcs were reliably indicative of systematic relation- ships of bats. To reduce the number of arcs to be considered, Anthony (1965) conducted immunoelectrophoresis and differentially stained only for esterases. Because intra- and interspecific varia- tion in esterases is well known (see Au- gustinsson, 1961) and because it was desirable to determine if the immune reaction might further elucidate informa- tion on the relationships of woodrats, this technique was employed. Materials and Methods.—Rats were bled by cardiac puncture. Sera were sep- arated by centrifugation and preserved by freezing at -15°C. Antisera against pooled samples of whole sera were pre- pared in rabbits following the procedure outlined by Gerber and Birney (1968: 413). Antisera obtained following the second series of immunizing injections were used in all reactions. Woodrats from which sera were used for produc- tion of antiserum are as follows: Neo- toma floridana baileyi—15, from Cherry County, Nebraska (A); N. f. campestris —S&, from Logan and Finney counties, Kansas (B); N. f. attwateri—l4, from Douglas County, Kansas (F); N. f. magister—4, from Giles County, Virginia (G); N. m. canescens—16, from Haskell County, Kansas (1); N. m. canescens— 13, from Barber County, Kansas (K). Serum of each of the above-listed wood- rats also was used in reactions with anti- sera and, in addition, sera of woodrats listed below were reacted against anti- sera but not used in production of the latter: N. f. campestris—l4, from Ellis County, Kansas (C); N. f. campestris— 4, from Russell County, Kansas (D); N. f. attwateri—1, from Ellsworth County, Kansas (E); N. m. canescens—11, from Baca County, Colorado (H); N. m. canescens—7, from Meade County, Kan- sas (J); Neotoma sp.—3, from area of sympatry in Major County, Oklahoma (L). Letters following localities indicate geographic origin of specimens in figure 42 and table 21. All animals used were adults and had been in captivity at least two weeks. An attempt was made to in- clude an equal number of animals of both sexes in each sample, but smaller samples did not always have equal sex ratios. The sample of Neotoma_ floridana magister was included to serve as a standard for other comparisons and be- cause the taxonomic status of this taxon is unclear; magister may be a species dis- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 149 tinct from other populations of floridana. Concentration of protein-nitrogen for immunoelectrophoresis and for injecting rabbits was determined with an Aloe- Hitachi hand protein-refractometer. Se- rum was diluted to contain one gram per cent protein. Slides for immunoelectro- phoresis were prepared by layering three ml of a two percent Ionager solution on each microscope slide. The arrangement of antigen wells and antibody troughs used is shown in figure 40. Electropho- resis was conducted in Michalis’ buffer, ionic strength 0.05, pH 8.7, for 35 min- utes at 40 volts. Ten lambda of unpooled serum from individual woodrats were placed in each antigen well immediately prior to electrophoresis. Each slide was prepared in duplicate. Following elec- trophoresis, gel in the precut trough was removed and the trough filled with anti- serum. Reactants were allowed to inter- act in a humidity chamber for 24 hours. Unbound protein was removed by wash- ing the agar slides for two days in three washes of borate-buffered saline, pH 8.6. Salts were removed similarly in three rinses of distilled water. The agar then was dried to a thin film and stained for esterase activity. The staining solution consisted of 40 ml of 0.2 M Tris-maleate and 0.2 M sodium hydroxide adjusted to a pH of 7.0, 1 ml of one per cent «- naphthyl acetate (in acetone), and 20 mg of Fast Blue RR diazonium salt. Reagents were mixed immediately be- fore use and gels incubated in the solu- tion for 20 minutes at room temperature. Stained slides were soaked for 15 min- utes in a two per cent glycerol solution to prevent cracking and peeling of the agar. The size and intensity of the major esterase band (Fig. 41) formed by the antigen of each woodrat against each antiserum was assigned a value on a scale of zero to eight. The minor band was scored similarly on a zero to two scale. Exemplary slides were selected with bands of each value to standardize scoring. The two values for each indi- vidual were added together and sub- Antigen well Antibody trough ee) O Anode a no) ° <= ~ © oO Fic. 40. Diagrammatic representation of microscope slide with Ionagar gel as used for immunoelectrophoresis. jected to a Y + 1 conversion to eliminate zero scores. These values then were averaged for the woodrats from each lo- cality. Because six antisera were used and two bands were considered for each antiserum, a total of 12 values was calcu- lated for specimens from each locality. These values were used as characters and the sample from each locality was treated as an OTU in the CLSNT subroutine dis- cussed previously. Results and Discussion —Mean values of scores for size and intensity of esterase bands are shown in table 21. It is im- mediately apparent that band scores for the population of N. f. campestris from Ellis County are highest in every case. Normally, in immunological tests, the homologous reaction is expected to ex- ceed all cross reactions, and cross reac- tions are considered in terms of percent immunological correspondence to the homologous reaction, which is set at 100. This technique is clearly not applicable in evaluation of the data at hand; even when the population from Ellis County is disregarded, the homologous reaction invariably is exceeded by at least one other cross reaction. Anthony (1965) found that the im- munological response to esterase antigens in cross reactions often surpassed that in reference reactions in studies of dogs. Several variables apparently interact to result in this phenomenon. The number of evolutionarily based changes in ester- ase molecules at antibody-antigen reac- tion sites between closely related popula- tions undoubtedly is low, antiserum pro- duced against an emulsion of whole serum and adjuvant is not highly specific, individual subjects have varying concen- 150 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY Fic. 41. Examples of esterase bands formed by differentially staining antigen-antibody pre- cipitate in dried Ionagar on microscope slides. The antiserum used was prepared against the serum of Neotoma floridana campestris. Sepa- rate samples of serum from two woodrats were placed in the wells (left and right) of each slide as follows: A—N. f. campestris from lo- cality C; B—N. f. baileyi from locality A; C— N. f. attwateri from locality F; D—N. f. magis- ter from locality G; E—N. micropus canescens from locality I; F—N. m. canescens from lo- cality K. See accompanying text for explanation of locality codes. trations of an enzyme in the serum de- pending on a variety of both intrinsic and extrinsic factors (see, for example, Jones and Bunde, 1970), and the cataly- tic action of enzymes is often reduced by reaction with corresponding antibodies, especially in antibody excess (Cinader, 1957:373). If the antigen of a given sam- ple contained a sufficient quantity of esterase molecules to stimulate produc- tion of a high titer of anti-esterase anti- bodies of relatively low specificity, and if a related sample contained the same or a similar esterase in greater quantity, the cross reaction would be expected to exceed the reference reaction when the two samples were tested. Therefore, the immune reaction involving single en- zymes has severe limitations when com- paring closely related taxa. Nevertheless, Anthony (1965) found that correlations of these kinds of data frequently corresponded well with gen- erally accepted theories of relationships of various breeds of dogs. The correla- tion and distance phenograms as calcu- lated by CLSNT for the woodrats studied are shown in figure 42. Values for N. f. campestris (C) from Ellis County are sufficiently greater than values for other samples that the distance phenogram separates that sample from all others at a distance (2.36) nearly double that (1.24) of the next major separation. Elsewhere in the distance phenogram, the two samples of N. f. attwateri appear as a subgroup closely allied to the hybrid sample from Oklahoma, the three sam- ples of N. m. canescens from adjacent lo- calities in Kansas form a single subgroup, but samples from the remaining localities do not correspond well with other data concerning relationships. In the correlation phenogram, which should reflect the relative degree of re- activity (enzyme similarity?) rather than the magnitude of reactions, the Ellis County population is closely coupled with the sample of N. f. attwateri from nearby Ellsworth County. The sample of N. f. attwateri from Douglas County is next to join that subgroup, followed by the sample from the locality of sym- patry. It was expected that the popula- tion of N. f. campestris (D) from Russell County, which is geographically and morphologically intermediate between those from Ellis and Ellsworth counties, would also be in that subgroup. How- ever, that sample formed a relatively distinct subgroup with the sample of N. f. magister, to which it certainly is not closely allied either geographically or morphologically. The sample of N. f. baileyi, and to a lesser extent, the sample of N. f. cam- pestris (B) from Logan and Finney counties, Kansas, appear in both pheno- grams to have esterases more like those of N. micropus than like those of other populations of N. floridana. This rela- tionship may somehow correspond more 151 WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 00OG ees 8ST 889 6c 00°9 cot 86S eve |l SSSA OOT es eet (V) taping f{ 'N louly = 1oleyy loury sloleyy loulyy toleyy loulyy = loleyy suos que spuergd jo VIIMLOS (>) (A) (d) (V¥) LOISIGDUL *f * NJ 1aqonyyw “f{°N — Siuqsaduind *f *N thajwd “f£°N vilosquy ‘S}BIPOOM JO UISLIO JO SOQI[eOOT JORXO IOF }XO} 99g ‘RAOSUL XIS YPIM poyover ues sndo/onw “NY puke DUDpWoY DWOJOaN JO So[duuvs ZT JO SpuRq esv19}So FO So1OOS UBIIY “1G HTAVL ihe) MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY N.f. baileyi A N.m. canescens H A N. m. canescens I N.f. campestris B N.m. canescens K N.m. canescens J N.f. campestris D N.f. magister G N.f. campestris (e N.f. attwateri E N.f. attwateri F Neotoma sp. L N.f. baileyi A N.f.campestris B B N.m. canescens H N. m. canescens K N.m. canescens ] N.m. canescens J N.f. campestris D N.f.magister G N.f. attwateri E N.f. attwateri F Neotoma sp. L N.f. campestris C a gg sp “gs 2.36 1.96 1.56 1.16 0.76 0.36 Fic. 42. Correlation (A) and distance (B) phenograms generated from mean scores of size and intensity of precipitated esterase-antibody bands. The coefficient of cophentic correlation of A is 0.792, whereas that of B is 0.895. WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 153 to adaptation to arid, relatively harsh en- vironments than to phylogeny. For whatever reason, both phenograms also indicate that, with respect to esterases, the sample of suspected hybrids from Major County, Oklahoma, more closely resembles floridana from eastern Kansas than it does western populations of mi- cropus. This result is not consistent with characteristics of pelage and skulls of the specimens involved. The fact that rather well marked dif- ferences in the antibody-antigen esterase bands were observed between woodrats of the two species from different locali- ties clearly shows that esterases in wood- rats differ qualitatively, quantitatively, or both. However, these data must be interpreted with respect to all other data relating to the supposed relationships of the various populations studied. Addi- tional studies on individual variation of esterases, quantitative variation of the esterases of an individual under various environmental conditions, and antibody specificity to enzymes will be necessary before data such as these can be fully and reliably interpreted with respect to the relationships of mammals. COMPARATIVE KARYOLOGY The karyotype of Neotoma floridana first was described by Cross (1931), who reported the diploid number as 52. Mat- they (1953) verified the diploid number and described two large submetacentric and two large subtelocentric chromo- somes in the complement. Mitotic chro- mosomes of Neotoma micropus were described by Hsu and_ Benirschke (1968); the diploid number was shown to be 52 and the karotype illustrated re- sembled that reported for floridana. Baker and Mascarello (1960) de- scribed the chromosomes of several spe- cies of Neotoma, and redescribed the karyotypes of both floridana and micro- pus. They reported that the number of large biarmed elements varies from one to four in micropus, but that females of floridana have four biarms and males have three. They concluded that the Y is a medium-sized subtelocentric chro- mosome in both species. The chromo- somal polymorphism in micropus was discussed by Baker et al. (1970) and shown to be a widespread phenomenon geographically, involving the X chromo- somes and one pair of large autosomes. Materials and Methods.—Prepara- tions of chromosomes were made from cells in bone marrow using a modifica- tion of the blaze-dry techniques de- scribed by Patton (1967) and Lee (1969). Chromosomes were stained in a saturated solution of crystal violet. Only specimens collected from natural populations are reported. Results dis- cussed below are based on study of at least 10 chromosome spreads from each of 58 woodrats, including specimens of Neotoma floridana baileyi, N. f. campes- tris, N. f. attwateri, and Neotoma micro- pus canescens. Included also are six specimens collected from the area of sympatry between floridana and micro- pus—3 mi S Chester, Major Co., Okla- homa. The maximum number of chromo- somes counted in any cell was 52. Some cells had less than 52 chromosomes, but the difference undoubtedly resulted from a loss of chromosomes during prepara- tion. Those having less than 52 chromo- somes were not studied or included in the 10 counts. In complete cells, no in- traindividual variation beyond that at- tributable to differential contraction of chromosomes was observed. Results and Discussion.—A consistent but relatively subtle (and heretofore un- noted) difference exists between the karyotypes of Neotoma floridana and Neotoma micropus, irrespective of the number of large biarmed elements. In the karyotype of both there is a graded series of 22 pairs of so-called acrocentrics. However, only rarely can chromatin be seen beyond the centromere opposite the arm in metaphase preparations of the acrocentric chromosomes of micropus, whereas in floridana there is invariably a visable amount of chromatin beyond the 154 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY centromeres of at least the larger acro- centrics. These tiny “arms” are evident in the karyotype of an N. f. attwateri illustrated by Baker and Mascarello (1969:189) and also were present in the karyotypes of floridana studied by me. If considered to be chromosome arms, these bits of chromatin would increase the Fundamental Number (FN) of the karyotype of N. floridana. Only in re- laxed spreads of N. f. campestris, how- ever, are these “arms” sufficiently large to cause difficulty in determining the number of biarmed chromosomes. Until more information concerning — these “arms is available, I consider it best not to include them in calculations of the FN and have attempted to distinguish such chromosomes from the subtelocen- trics that are involved in the polymorphic system. Although it may be misleading to refer to these chromosomes as acro- centrics, I will do so in an attempt to preserve the terminology used by Baker and Mascarello (1969). In the chromosomal complement of each of the nine female N. f. baileyi ex- amined, only two large and distinctly biarmed elements were seen (Fig. 43); in each instance the two biarms were of nearly equal size. Two distinctly bi- armed elements (one noticeably larger than the other) also were seen among the chromosomes of each of the four male baileyi karyotyped (Fig. 43). There is no chromosome in baileyi males that closely resembles the chromosome Baker and Mascarello (1969:189) designated as the Y of N. f. attwateri. Therefore, I consider the smaller of these two sub- metacentric elements as the Y chromo- some and the larger as the X. In three male N. f. attwateri from Douglas County, Kansas, four biarmed elements were observed (Fig. 44). Three had arms of unequal length and were dis- tinctly larger than the fourth in each in- stance. The fourth was indistinguishable from the chromosome thought to be the Y in baileyi. Each of six N. f. attwateri females (two from Ellsworth County and four from Douglas County, Kansas) had four, large, biarmed chromosomes that were indistinguishable from the three larger biarms described for males. The karyotype of these females was like that described by Baker and Mascarello (loc. cit.) for five females from Payne County, Oklahoma. Apparently the Y chromo- some varies in attwateri, but because I examined only three males from a single locality, and Baker and = Mascarello studied only two from another locality, it is not possible to determine if the vari- ation is geographic or if it is polymorphic at some localities. Available evidence in- dicates that the number of large biarmed chromosomes exclusive of the Y is con- stant in attwateri at four in females and three in males. A small submetacentric chromosome that is indistinguishable from the Y in baileyi and in attwateri from Douglas County was seen in the karyotypes of three of four N. f. campestris males. These animals were from Logan, Finney, and Russell counties, Kansas. The fourth male, from Ness County, Kansas (Fig. 44), lacked such a chromosome, but several “acrocentrics” in the preparation had suf- ficiently large “arms” beyond the centro- mere as to be considered subtelocentrics; one of these probably is the Y. Addition- ally in the Ness County male, there is one large submetacentric, one large sub- telocentric, and another large chromo- some that may be a subtelocentric. In the karyotype of each of the other three animals there is one large submetacen- tric chromosome, one distinctly subtelo- centric element, and a third large chro- mosome that probably is a subtelocentric but may be an “acrocentric.” The karyo- types of nine female campestris all were characterized by two large submetacen- trics and either one or two large subtelo- centrics. One female from Finney County had two subtelocentrics in all except one of the cells examined; chro- mosomes in this cell were severely con- tracted and the two subtelocentrics were indistinguishable from acrocentrics of similar size. The presence of two large submetacentrics in females and one in WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 155 ax 2am & Jo Vv i an 02 pf AN 6K Ak 46 tA TY a ee, er, oe | ee aw 6 aa aan ae ~~ ae B2 aa af (8) nae LANA NGQ RRKHA RZ ANAM nh AR ALR AB Ar ON AH AA ~a— NA AR AR Ae ee MR Fic. 43. Karyotypes of a female (A) and a male (B) Neotoma floridana baileyi from Rock and Cherry counties, Nebraska, respectively. The scale applies to both karyotypes. males indicates that these are the X chromosomes. The presence of one sub- telocentric in some females and two in others may represent a polymorphic sys- tem the same as, or similar to, that dis- cussed by Baker et al. (1970) for N. micropus. On the other hand, these observations may be a result of the dif- ficulty in distinguishing large subtelo- centrics from the so-called acrocentrics. A medium-sized subtelocentric chro- mosome was present in each of the karyotypes of five N. m. canescens males (three from Barber County and two from Haskell County, Kansas). This element was not seen in any of the karyotypes of 11 females and undoubtedly is the Y chromosome as shown by Baker and Mascarello (1969) and by Baker et al. (1970). Both males from Haskell County had one large submetacentric and one large subtelocentric in addition to the Y chromosome. The three males from Barber County each had one large sub- metacentric and two large subtelocen- trics. Three of four female N. m. canescens from Baca County, Colorado, one of two from Haskell County, Kansas, and four XY ah 62 fh fe 0b fe tt oe Fae Wee ee) | PY eeeey Qe ana Om 6 Pe ee “y oa Gf Mh ft Ak te he te ai eo nb oa 46 86 08 4a PY ee Y ee Y ee Fic. 44. Karyotypes of a male Neotoma floridana attwateri (A) from Douglas County, Kansas, and a male N. f. campestris (B) from Ness County, Kansas. The scale applies to both karyotypes. of five from Barber County, Kansas, each had two large submetacentrics and two large subtelocentrics. The karyotypes of these animals were indistinguishable from that illustrated by Hsu _ and Benirschke (1968) for a micropus fe- male. The other three females each had two large submetacentrics, but only a single subtelocentric. Two males, both identified as hy- brids, from the area of sympatry between floridana and micropus (3 mi S Chester, Major Co., Oklahoma) had karyotypes indistinguishable from those of the three male N. micropus from Barber County, Kansas. Four females identified as mi- cropus from that locality each had four large biarmed elements. The karyotypes of three of these were indistinguishable from that of micropus females having four biarms, but the fourth had two acro- centrics with tiny “arms” as described for floridana. The presence of floridana- 156 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY like acrocentrics in the karyotype of this animal undoubtedly resulted from hy- bridization of the two species. A female identified as a hybrid had three distinctly biarmed elements, but spreads were not sufficient to determine the presence or absence of “arms” beyond the centro- mere of large acrocentrics. The polymorphic system involving the number of large biarmed chromo- somes in N. micropus reported by Baker and Mascarello (1968) and discussed by Baker et al. (1970) probably will be found to exist throughout the range of the species. Baker et al. discussed vari- ation in number of biarms in animals from three widely separated localities in Texas and one locality in Oklahoma. The karyotype is now known to be variable at one locality in Colorado and two in Kansas. The only locality from which specimens of micropus have been karyo- typed and not reported to be polymor- phic is 16 km N Ciudad Victoria, Tamau- lipas (Hsu and Benirschke, 1968). These animals were of the subspecies N. m. mi- cropus, whereas the polymorphic popula- tions all are N. m. canescens. However, Hsu and Benirschke (1968) did not indi- cate if chromosomes of more than two rats (the karyotypes illustrated) from Tamaulipas were examined. Baker et al. (1970) discussed the possible origin of polymorphism in micropus; apparently it is the result of neither a Robertsonian change (as frequently is seen in mam- mals) nor of a single pericentric inver- sion. These authors suggested that both inversions and translocations may be in- volved in the origin of this system. The discovery of a submetacentric Y chromosome in some populations of N. floridana is of interest. No other species of the genus has been reported to have a submetacentric Y chromosome and, although the karyotype of N. f. baileyi was not found to vary, these findings in- dicate that a previously undescribed polymorphism exists in the Y chromo- some of N. f. attwateri and N. f. campes- tris. The number of large biarmed chro- mosomes (exclusive of the Y) has not been found to vary from three in males and four in females for N. f. attwateri, but only two biarms were seen in the karyotype of the nine baileyi females and only one (exclusive of the Y) in the four baileyi males examined. In N. f. campes- tris, the number of large submetacentric chromosomes seems to be constant at two in females and one in males. Although it is difficult to distinguish large subtelo- centrics from some “acrocentrics” in rats of this subspecies, apparently the num- ber fluctuates from zero to two. The large biarms in baileyi and the large sub- metacentrics in campestris probably are the X chromosomes, but in attwateri the X chromosomes cannot be certainly identified. As shown by Baker et al. (1970), the X chromosomes in micropus cannot be distinguished with certainty, because polymorphism is involved and also be- cause the relative lengths of the arms in the large, biarmed chromosomes is vari- able. In the micropus examined by me, at least one large submetacentric always was found in males and two such ele- ments were present in females; in the absence of other information these would appear to be the X chromosomes. How- ever, Baker et al. (1970) reported two females that had only one large, biarmed chromosome. All males examined by them had at least one biarmed chromo- some in addition to the smaller subtelo- centric Y. This chromosome probably is the X, but as seen in the two females having only a single biarmed element, at least one X can be an acrocentric in females. Because of the limited number of animals examined and the chromo- somal complexity of this group, a discus- sion of the evolution of chromosomes in N. floridana and N. micropys is some- what premature. However, in N. flori- dana previously reported by Cross (1931), Matthey (1953), and Baker and Mascarello (1969), and in N. f. attwateri from northeastern Kansas, the number of biarmed chromosomes was always four in females and three plus the Y in WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 157 males. This is also the number of biarms reported for N. m. micropus from Tamaulipas and the most common karyo- type seen in N. m. canescens. It is likely, then, that this is the karyotype from which others have evolved. The number of large biarms is unstable in micropus and possibly fluctuates in N. f. campes- tris, but in N. f. baileyi the two biarmed autosomes have been replaced by two acrocentrics. Because the Y chromosome is a me- dium-sized subtelocentric in all popula- tions of N. micropus examined and in all floridana that have been examined ex- cepting baileyi, some campestris, and the attwateri from northeastern Kansas, the subtelocentric undoubtedly is the primi- tive Y chromosome and the submetacen- tric is derived. If it is eventually found that the lengthened arm of the submeta- centric Y was acquired by a transloca- tion of an arm of one of the original biarmed autosomes, then the two poly- morphic systems may have a common origin. In any event, the submetacentric Y chromosome evidently was present in at least some floridana males prior to the time baileyi and campestris dispersed to the geographic areas they now occupy. The apparent fixation of this element in baileyi is not surprising considering that the subspecies is isolated in a relatively small geographic area. The submetacen- tric Y probably is commoner than the subtelocentric in campestris. Both forms of the Y chromosome are known in at- twateri; but the relative status of the two is not known. SUMMARY AND ZOOGEOGRAPHIC CONSIDERATIONS Neotoma angustipalata, N. floridana, and N. micropus form a closely related complex of almost entirely allopatric taxa. The distributions of floridana and micropus are most appropriately termed stasipatric. Key (1968:22) discussed stasipatry as follows: “We could perhaps distinguish a condition of ‘stasipatry’ as a special case of parapatry in which the zone of overlap is limited by an impair- ment of the fecundity of freely produced hybrids rather than by ecological fac- tors... The only known locality where the two species occur together (3 mi S Chester, Major Co., Oklahoma) is char- acterized by the presence of hybrids, identification of which was based on a variety of comparisons with hybrids reared in the laboratory. For example, the electrophoretic pattern of hemoglo- bins of some animals from the locality of sympatry was otherwise observed only in known hybrids; discriminant function analysis based on 17 characters of speci- mens of the two groups indicated they were generally intermediate between non-hybrid specimens of the two species; and the karyotype of one animal from the locality in question almost certainly con- tained chromosomes derived from both species. There is no reason to believe that floridana and micropus ever have oc- curred together without producing na- tural hybrids or that they will do so in the near future. Although floridana gen- erally is an inhabitant of relatively mesic woodland habitats (N. f. campestris be- ing a notable exception) and micropus generally is associated with arid grass- lands, either species probably could ex- pand its range (at least slightly) in the absence of the other (although the exis- tence of a relatively broad hiatus be- tween the distributions of the two species throughout much of the region of poten- tial contact might argue against the latter point). In any event, the two apparently hybridize when in contact and results of laboratory breeding studies strongly sug- gest some hybrid inviability. By defini- tion, then, “stasipatry” best explains the distributional relationship of these two woodrats. The distribution of N. floridana is not contiguous with that of N. angusti- palata. Neotoma micropus and N. an- gustipalata occupy adjacent geographic 158 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY areas, but sympatry is unknown and no evidence of natural hybridization has been uncovered (but, see Alvarez, 1963: 452). To predict how micropus and angustipalata would act if they were sympatric clearly is speculative; repre- sentatives of angustipalata are larger than are those of micropus from adjacent localities in México and I doubt that the two would hybridize. However, size alone apparently is a relatively inefficient isolating mechanism among woodrats. Morphologically, angustipalata is approx- imately as distinct from both floridana and micropus as these two species are from each other. Considering the fact that floridana and micropus hybridize in the laboratory and also in nature, some systematists might argue that the two are conspecific. However, micropus and floridana have maintained a high level of specific in- tegrity in the past, apparently are doing so at present, and I predict, after exten- sive field and laboratory study, that they will continue to do so. Although there obviously is at least some genetic com- patibility between floridana and micro- pus, the process of speciation between the two is essentially complete, and I regard it as having reached an irrever- sible point in time. Furthermore, I do not believe that our understanding of the evolutionary history and systematic relationship of these rats would be en- hanced by formally placing micropus in the specific synonymy of floridana. In fact, such an arrangement would suggest that the two intergrade broadly and are more closely related than is the case. If considered as a single species, individuals or populations that should be studied separately might eventually be treated together in research by non-taxonom- ically oriented biologists, whose research design is partially dependent on deci- sions by taxonomists. Despite my convictions that floridana and micropus should be considered sepa- rate species, certain data indicate that hybridization is, or recently has been, introgressive. Analyses of frequency and size of the fork on the anterior palatal spine and of the morphology of the pos- terior margin of the bony palate indicate that in some instances one or more pop- ulations of one species from localities geographically contiguous with popula- tions of the other may have acquired selected genetic material introgressively. This was suggested most strongly by pop- ulations of floridana in Oklahoma and Texas. Also, specimens of micropus from localities adjacent to the range of flori- dana in south-central Kansas and coastal Texas are larger (thus somewhat like floridana) than specimens of micropus from localities not geographically contig- uous with populations of floridana. It is possible that each species is selectively acquiring a limited amount of genetic material from the other (Key, 1968; Lewontin and Birch, 1966). However, certain other characters, such as electro- phoretic patterns of hemoglobins and analyses of karyotypes, do not indicate introgression. Introgression is extremely difficult to “prove” or “disprove,” and in the case of floridana and micropus elucidation of this phenomenon must await additional data. The micropus species-group estab- lished by Burt and Barkalow (1942) is meaningless. The three species studied (angustipalata, floridana, and micropus ) share a common ancestor in the not too distant past and represent a single spe- cies-group, the floridana-group. Ander- son (1969) and Finley (1958) have shown that N. albigula is closely allied to N. micropus (Burt, 1960; Hooper, 1960), and Anderson (1969) indicated the possibility that floridana, micropus, and albigula eventually may best be con- sidered a single species. Thus, it would seem that albigula and related species (palatina, nelsoni, and varia—see Hall and Genoways, 1970) also should be in- cluded in the floridana-group. When the nomenclatorial arrange- ments of floridana and micropus are con- sidered below the level of the species, some conclusions are relatively clearcut, whereas others are somewhat arbitrary. WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 159 Among populations of floridana studied, total geographic variation was less than that found in micropus. Neotoma flori- dana baileyi has certain unique features of the skull, and is relatively less variable than other samples as evinced by the absence of intraspecific variation in chro- mosomal complement and by lower co- efficients of variation in most mensural characters; baileyi also has evolved a relatively distinctive pattern of repro- duction. In certain aspects, some cranial dimensions and color for example, baileyi appears to have its affinities as much with campestris as with attwateri, but in final multivariate analysis, baileyi ap- peared more like attwateri than like campestris. As discussed beyond, the probable evolutionary history of these woodrats also suggests that the affinities of baileyi are with attwateri. Neotoma floridana campestris is the most distinctive of the western subspe- cies of floridana with respect to color. Statistical analysis of mensural data indi- cated that in a few instances attwateri is larger than campestris. Specimens in a sample of campestris from Colorado and Nebraska were especially small rela- tive to those in two samples of attwateri from localities in Kansas. A tendency was observed for woodrats from the zone of intergradation between campestris and attwateri to be larger than indi- viduals in adjacent populations of either subspecies and for rats from localities west of this zone to become clinally smaller. Apparently, campestris exists in relatively small and semi-isolated popu- lations that occupy discontinuous areas of suitable habitat. One acquires this impression in field study of campestris and additional evidence of localized stocks includes: 1) high variability of mensural characters in samples of this subspecies; 2) two of six animals from a semi-isolated population in Finney County, Kansas, had unique hemoglobin; and 3) three males each from different populations had a submetacentric Y chro- mosome, but a male from a fourth pop- ulation had a subtelocentric Y chromo- some. Neotoma floridana attwateri, as here recognized, includes those animals pre- viously assigned to attwateri and to the subspecies osagensis. As indicated above, specimens assignable to attwateri from near the range of campestris are espe- cially large. Those from southeastern Kansas and southern Texas were next largest among the samples studied, and those from eastern Oklahoma and north- eastern Kansas were smallest. The sub- species has not been found to be poly- morphic for the number of large biarmed chromosomes, but it is polymorphic in the morphology of the Y chromosome. Three males from northeastern Kansas had a distinct submetacentric Y element, whereas two from near Stillwater, Okla- homa (Baker and Mascarello, 1969), had the more common subtelocentric Y. Variation in Neotoma micropus is more easily definable geographically than that in N. floridana, but more dif- ficult to resolve nomenclatorially at the subspecific level. N. m. planiceps is known only by the holotype; thus varia- tion within the subspecies is unknown. The holotype is a small woodrat similar in size to other Mexican representatives of the species. Multivariate analyses indicated that planiceps is relatively dis- tinct morphologically from both canes- cens and micropus. Possibly N. m. plani- ceps and N. angustipalata represent a single taxon. The holotype of the former is a young adult and conclusions regard- ing the affinities of planiceps must be regarded as tentative. The name N. m. micropus has been restricted to the brownish, long-tailed woodrats that occur on the coastal plain and Sierra de Tamaulipas in the state of Tamaulipas. An appreciable amount of geographic variation exists even within this restricted area. Woodrats become progressively less brownish and more grayish from south to north in Tamauli- pas. Variation in size especially and that in color to a lesser extent forms a sharp step-cline across the lower Rio Grande. 160 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY However, the zone of contact between micropus and canescens in western Tamaulipas and eastern Nuevo Leon is relatively broad; the type locality of N. m. micropus (Charco Escondido, Tamau- lipas) is in this zone of intergradation, but woodrats from that locality resemble more closely those from coastal Tamauli- pas than rats from most of the range of N. m. canescens. Neotoma micropus canescens is the most variable subspecies studied. Speci- mens from northern and eastern parts of the range are larger than those from southern and western localities, with the exception that specimens from coastal southern Texas are among the largest of the species. The darkest individuals oc- cur in the northeastern parts of the range, but members of the subspecies become progressively paler from east to west. Distributional records of available spec- imens indicate that a large area in cen- tral Texas is not inhabited by N. micro- pus. If true, populations of large wood- rats from coastal southern Texas are only circuitously connected geographically with populations of large woodrats to the north. Routes of gene flow between northern and southern populations of large woodrats thus would include pop- ulations of smaller western rats. The sub- species canescens could be subdivided into five subspecies with some merit as discussed previously. Another logical ar- rangement might recognize the small pallid woodrats of New Mexico, south- western Texas, and adjacent México (ex- clusive of coastal Tamaulipas) as one subspecies (leucophea) and restrict the name canescens to the large woodrats from the northern and eastern parts of the range of the subspecies as here recog- nized. However, such an arrangement would not account for intermediacy in size and color of woodrats from south- sastern Colorado, the panhandle of Texas, and non-coastal southern Texas. It might also result in an arrangement whereby populations of one subspecies are separated by populations of another. Neotoma angustipalata is known by too few specimens to permit a meaning- ful analysis of intraspecific variation. Hooper (1953) commented on the ex- treme variability in this species and I observed the same phenomenon. More specimens of this enigmatic species are needed. SUGGESTIONS FOR ADDITIONAL RESEARCH Many aspects of the biology, and in particular the systematics, of woodrats of the floridana species-group (as here defined) need additional study. I have attempted throughout the preceding dis- cussions to indicate these needs, and summarize them here. My study and others (Anderson, 1969; Finley, 1958) demonstrated the importance of con- tinued field and laboratory work on the distributional and systematic relation- ships of N. albigula with both N. flori- dana and N. micropus. Especially criti- cal geographic areas include southeastern Colorado, New Mexico, western Texas, the Edwards Plateau, southern Chihua- hua, and montane areas of Coahuila. Additional field work to study the exact distributional relationships of flori- dana and micropus is needed in south- eastern Colorado, and throughout the area of general contact in Oklahoma and Texas. Sustained search for areas of sympatry and continued study of the one such area now known should result in elucidation of the distinctiveness of the two species and the presence or absence of introgression. Collecting efforts in that area of central Texas not now known to be inhabited by either Neotoma floridana or Neotoma micropus will elucidate the distributional status of woodrats and po- tential routes of gene flow in that state. The acquisition of additional specimens of N. m. micropus and N. m. canescens from Tamaulipas will help to clarify the relationships of these two taxa. Speci- mens from southern Tamaulipas and San Luis Potosi are needed to assess the sys- tematic status of N. angustipalata and N. m. planiceps and to better understand the distributional relationship of these WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 161 woodrats with adjacent populations of N. m. micropus and N. m. canescens. Certain problems requiring a com- bination of field and laboratory research have been studied, but most still lack definitive solutions. Birney and Perez (1971) presented hypotheses concerning the nature of variation and the mode of inheritance of woodrat hemoglobin, but these hypotheses need to be tested and refined. The chromosomal polymorphism in the number of large biarmed chromo- somes first reported for N. micropus by Baker and Mascarello (1969), discussed by Baker et al. (1970), and observed in N. floridana by me needs to be analyzed more intensively from the standpoint of evolution, geographic distribution, and function. Baker and Mascarello (1969: 195) stated that “our results demand .. . introduction of individuals to wild pop- ulations different in chromosomal consti- tutions.” I do not believe that such a means of study is necessary or advisable and am strongly opposed to research that might alter natural patterns of evolu- tionary phenomena in animals. How- ever, studies wherein mating of captive individuals of “different chromosomal constitution” could be followed with analysis of meiosis in progeny would be enlightening. The polymorphism involv- ing the Y chromosome in N. floridana should be studied to discern its distribu- tion and origin. Boice (1969) observed behavioral differences between N. albigula and N. micropus and Birney and Twomey (1970) reported evidence for physiolog- ical divergence of N. floridana and N. micropus. These areas of research hold promise in their own right and in terms of clarifying the overall systematics of the woodrats of the floridana species- group. ZOOGEOGRAPHIC COMMENTS Hibbard (1967:128) suggested that “the stock that gave rise to Neotoma must have separated off from a general- ized cricetine in the Upper Miocene.” He considered the extinct genus Plio- tomodon, named by Hoffmeister (1945) from Pliocene deposits in California, as a specialized side branch related to Neo- toma, but not in the direct lineage of Recent woodrats. The specimen from the Cumberland Cave Fauna ( Pleisto- cene) named as a distinct genus, Para- hodomys, by Gidley and Gazin (1933: 356) may represent another such off- shoot, but also best may be considered as a member of the genus Neotoma. The three species of Neotoma re- ferred to the subgenus Paraneotoma by Hibbard (1967) from the Upper Plio- cene and Middle Pleistocene of Kansas are more like N. (Hodomys) alleni than Recent members of the subgenus Neo- toma. It is not possible to determine whether Paraneotoma is ancestral to Re- cent Neotoma or if the subgenus repre- sents a once widely distributed group of species related to N. alleni. Alvarez (1966:9) named N. magnodonta from the Middle or Upper Pleistocene of Méx- ico (state of México) as a member of the subgenus Hodomys; thus it clearly is not in the lineage of the floridana species- group. The only other fossil named as a distinct species that might relate sig- nificantly to the evolutionary history of the floridana species-group is N. ozark- ensis, which was described by Brown (1909:196) from Middle to Late Pleisto- cene (Conard Fissure) deposits of northern Arkansas. This woodrat may prove to be no more than a subspecies of floridana; however, if the specimens are from pre-Wisconsin deposits it is likely that they predate all but the ear- liest processes of divergence of floridana and micropus. These fossil records indicate that the genus Neotoma originated in the late Miocene or early Pliocene and evolved during the Pliocene to the extent that presently recognized subgenera were dis- tinct by the beginning of the Pleistocene. There is no evidence that woodrats of the floridana-group inhabited the Central Plains during the Yarmouth. Neotoma (Paraneotoma) taylori occurred in at least parts of the Great Plains at that time 162 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY (Hibbard, 1967; 1970). Hibbard (1963: 209) reported a specimen that he con- sidered more like floridana than micropus from a late Illinoian fauna in Kansas, and Hibbard and Taylor (1960:175) re- ported N. micropus from the Sangamon of Kansas. These specimens were tenta- tively identified on the basis of the shape of the posterior triangle of the anterior loop of Ml. Semken (1966:151) and Dalquest et al. (1969:249) have indi- cated, and I concur, that this is not a diagnostic character to distinguish the two species. Semken (loc. cit.) reported additional material from the late Ili- noian that compared favorably with mi- cropus, but he elected not to assign the material to either species. Dalquest et al. (1969) reported N. floridana, N. micropus, and N. albigula from deposits considered to be 11,000 to 8000 years BP. Although I have not examined this material, neither the mea- surement they used for specific identifica- tion (breadth of molar rows) nor any other single measurement taken by me will serve to distinguish Recent speci- mens of the three species. Furthermore, identification of Recent specimens of the three based on fragmentary skulls and lower jaws would be difficult, especially distinguishing between floridana and mi- cropus. Possibly albigula occurred sym- patrically with either floridana or micro- pus on the Edwards Plateau in the late Pleistocene, but I question whether mi- cropus and floridana were in sympatry that early and I cannot conceive of all three species having occurred there simultaneously. As I interpret these findings they indi- cate that woodrats of the floridana-group occurred on the Great Plains by late Ilinoian. They may have diverged from related groups as late as the Illinoian. Neotoma albigula and related species also could have diverged from a floridana- like stock during the Illinoian, because results of most studies (e.g. Sprague, 1941; Burt and Barkalow, 1942; Burt, 1960; Hooper, 1960) indicate that micro- pus and floridana are more alike mor- phologically than either resembles al- bigula. With the advance of Wisconsin ice, the basal stock of floridana probably re- treated southward. Neotoma albigula might have been restricted to the Mexi- can Plateau or to the region of southern California, Arizona, and New Mexico (or both), micropus to the lowlands of coastal southern Texas and Tamaulipas, and floridana to the southeastern United States, possibly to peninsular Florida (see Sherman, 1952; Blair, 1958). Guil- day et al. (1964:158) suggested that N. f. magister survived Wisconsin glaciation in the southern Appalachian Mountains. In view of the striking morphological and ecological distinctness of magister as compared with all other subspecies of floridana, I agree with Guilday et al. and further suggest that magister and flori- dana eventually will be found to repre- sent biological species at least as distinct as floridana and micropus. However, the status of magister is beyond the scope of the present paper and the relationship of this taxon must await detailed field and laboratory study. Speculation on the distribution in Wisconsin time of Neotoma angustipa- lata and Neotoma palatina also is of in- terest. The affinities of angustipalata as shown herein are clearly with micropus and floridana, but proclivities toward one, more than the other, are lacking. Neotoma palatina is thought to be most closely related to N. albigula. Neotoma angustipalata has a relatively restricted range in southern Tamaulipas and San Luis Potosi, whereas palatina is restricted to the barranca of Rio Balanos, associated tributaries, and adjacent uplands. Be- sides micropus and floridana, only angus- tipalata and palatina are characterized by the absence of a maxillovomerine notch. Possibly this characteristic has evolved twice, but the solid vomerine septum of these four species may be in- dicative of a pre-Wisconsin common an- cestor. Neotoma palatina apparently was sufficiently isolated from adjacent popu- lations of albigula to effect speciation. WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 163 Neotoma angustipalata probably was iso- lated in montane habitats of the Sierra Madre Oriental during the Wisconsin, whereas micropus occurred on the coastal lowlands. Both angustipalata and _ pala- tina represent peripherally-distributed species that have managed to avoid ex- tinction, but for some reason (possibly the presence of adjacent populations of micropus and albigula, respectively) have been unable to significantly expand their ranges since the recession of Wisconsin LCE! Recent advances in paleobiology, ge- ology, and meteorology have improved understanding of late Pleistocene and Holocene climatic and vegetational pat- terns on the Great Plains. These data were summarized by Hoffmann and Jones (1970) according to post-Pleisto- cene chronology and terminology pro- posed by Bryson et al. (1970). During Full-glacial (to approximately 13,000 BP), floridana and micropus probably were completely isolated in their respec- tive refugia. Following initial isolation, the two incipient species evolved dis- tinctive hemoglobins from the original double-banded phenotype. Differences in chromosomal complements, morphol- ogy, and color also have evolved under differential selective pressures since that time. Blair (1958) included floridana and micropus in a list of mammals and reptiles that previously were isolated into eastern and western populations, but that since have reestablished contact in the forest-grassland ecotone. The barrier that isolated micropus and_ floridana along the Gulf Coast of the southeastern United States may have been the Missis- sippi Embayment, but as discussed by Blair (1958) most species separated by this barrier remain disjunct. During the more equable climate of Late-glacial (13,000 to 10,500 BP), and with north- eastward retreat of continental ice (de- spite minor phases of retreat and read- vance), both isolated populations began a northward dispersal. During the more continental climates of the Pre-boreal (10,500 to 9140 BP), Boreal (9140 to 8450 BP), and Atlantic (8450 to 4680 BP), micropus and floridana reached their present limits of distribution and floridana at least occurred somewhat farther north and west of the present range (see Jones, 1964). It probably was during this time that woodrats ad- vanced northward along the Missouri and westward along the Niobrara rivers. The Smoky Hill, Saline, Republican, Ar- kansas, and possibly other rivers and tributaries served as corridors for flori- dana to disperse across western Kansas into eastern Colorado and southwestern Nebraska. Several late Pleistocene-early Recent records of floridana from locali- ties north of the present range are avyail- able (e.g., Parmalee and Jacobson, 1959; Bader and Hall, 1960; Parmalee et al. 1961). Morphologically, N. f. baileyi somewhat resembles N. f. campestris, especially those specimens from popula- tions in Colorado and southwestern Ne- braska. Possibly there was appreciable north-south gene flow through eastern and central Nebraska during this period. However, I doubt that the Sand Hills region of Nebraska ever was inhabited by floridana and the similarities between baileyi and campestris more likely repre- sent convergence. When all characters are considered, baileyi clearly is more closely related to attwateri from north- eastern Kansas than to campestris. During and following Full-glacial, the Sierra de Tamaulipas may have served as a barrier for the small, brownish, long- tailed woodrats that presently occur there. When populations of micropus subsequently dispersed northward, these rats reestablished contact and_ inter- graded with other populations. The re- sult of this intergradation and_ subse- quent selection is the coastal subspecies to which the name N. m. micropus is re- stricted. The evolutionary history of N. m. planiceps cannot be understood until the relationships and status of this nominal subspecies are better known. However, planiceps apparently is iso- lated from other populations of micropus on the Mexican Plateau at present, but 164 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY probably has not been isolated since the Full-glacial. The Sub-boreal (4680 to 2690 BP) probably was the coolest post-glacial pe- riod on the Northern Great Plains. This period was characterized by a southward shift in both the northern and southern limits of the boreal forest. It probably was during this cooler period that flori- dana retreated slightly southward and eastward, leaving the isolated N. f. baileyi in the sheltered canyons of the Niobrara River and_ associated tribu- taries. Jones (1964) and Hoffman and Jones (1970) suggested that baileyi be- came isolated during a hot, dry post- glacial period. Although that hypothesis remains plausible, it is now thought that the earliest period characterized by hot, dry climate was the Scandic (1690 to 1100 BP); the degree of distinctness ex- hibited by baileyi suggests a longer pe- riod of isolation. Rainey (1956) and Jones (1964) discussed the apparent in- ability of N. f. attwateri to extend its range into superficially suitable habitat in southeastern Nebraska; attwateri pos- sibly is unable to move farther north be- cause of the severity of the climate, espe- cially in winter. Riparian habitats that appear suitable for habitation by baileyi are present in southern South Dakota, but for some reason (possibly the ab- sence of enough shelter during winter) woodrats have not been found there. Perhaps baileyi has been able to persist in north-central Nebraska only in spe- cially sheltered areas, the most important element furnished by the canyons of the Niobrara being protection from severe winters rather than a relatively cool en- vironment during hot, dry summers. The Sub-boreal also may have iso- lated N. f. campestris from N. f. attwa- teri, and the two probably were disjunct for an extensive period. Contact may have been reestablished in the relatively warm moist Sub-Atlantic, broken again during the dryer Scandic, and not rees- tablished until European man fostered the spread of riparian and other wood- land habitats in north-central Kansas. Al- though the two subspecies appear to have been separated for a lengthy pe- riod, they are presently in contact. Dur- ing the dry Scandic and possibly during other periods as well, it appears that campestris was distributed in an un- known number of small isolated popula- tions in disjunct and probably marginal habitats. Since the time of initial isolation, the submetacentric Y chromosome appar- ently became fixed in baileyi, but the translocation obviously had occurred prior to isolation as it is seen also in campestris and in attwateri from north- eastern Kansas. Origin of the 6? hemo- globin locus also occurred prior to the time of isolation, because the £* allele is seen in all three of the western sub- species of floridana. During the warm, wet Neo-Atlantic and since that time, campestris probably has dispersed some- what from the isolated, relict populations of the Scandic, but continues to occur in relatively disjunct, semi-isolated pop- ulations. With the possible exception of N. m. micropus and N. m. planiceps, 1 doubt that any of the populations of micropus have evolved long in isolation from other members of the species. Variation in size and color both are clinal, the chro- mosomal polymorphism involving num- ber of large biarmed elements has been observed at localities from which signifi- cant numbers of woodrats have been karyotyped (the only exception is N. m. micropus from north of Ciudad Victoria, Tamaulipas), and no marked changes in hemoglobin phenotypes were observed in northern populations of the species. Neotoma micropus and N. floridana apparently evolved in the classical man- ner during the Pleistocene as a result of isolation during glacial advance. Al- though two species in my _ estimation, they are closely related, recently evolved, and retain limited genetic compatibility. It is impossible to define precisely at what stage two evolving phena should be called species, but as I understand floridana and micropus the process of WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 165 evolution almost certainly has reached a point in time whereby irreversible dif- ferentiation has taken place. Although micropus and floridana apparently did not evolve exactly according to the stasi- patric model (White et al., 1967; Key, 1968), they are stasipatrically distributed today and apparently are continuing the process of speciation in a manner similar to that seen in morabine grasshoppers, for which this model was proposed. That is, the two species are in contact and form a tension zone wherein hybrids with reduced viability are produced. This tension zone undoubtedly shifts slightly within the deciduous forest- grassland ecotone in response to environ- mental changes. Introgression of a few advantageous genes may be occurring in one or both species as a result of hy- bridization within the tension zone, but clearly introgression (if it has occurred at all) has been limited and mostly pre- vented by rigid selection. 166 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY LITERATURE CITED ALLEN, J. A. 1891. Notes on new or little-known North American mammals based on recent additions to the Collection of Mam- mals in the American Museum of Natural History. Bull. Amer. Mus. Nat. Hist., 3:263-310. 1894a. Cranial variations in Neotoma micro- pus due to growth and _ individual differentiation. Ibid., 6:233-246. 1894b. Descriptions of ten new North Amer- ican mammals, and remarks on oth- ers. Ibid., 6:317-332. ALVAREZ, T. 1963. The Recent mammals of Tamaulipas, México. Univ. Kansas Publ., Mus. Nat. Hist., 14:363-473. 1966. Roedores fosiles del Pleistoceno de Tequesquinahua, estado de México, México. Acta Zool. Mexicana, 8(3):1- 16. ANDERSEN, K. W., AND E. D. FLEHARTY 1967. Mammalian distribution within biotic communities in northeastemn Jewell County, Kansas. Fort Hays Studies, Sci. Ser., 6:v + 1-46. ANDERSON, S. 1969. Taxonomic status of the woodrat, Neotoma albigula, in southem Chi- huahua, Mexico. Misc. Publ., Univ. Kansas Mus. Nat. Hist., 51:25-50. ANTHONY, R. L. 1965. Serological, immunoenzymatic and biochemical comparisons among reg- istered breeds of dogs. Unpublished Ph.D. dissertation, Univ. Kansas, 82 pp. ARMSTRONG, D. M. 1972. Distribution of mammals in Colorado. Monog. Mus. Nat. Hist., Univ. Kan- sas, 3:x + 1-415. ASDELL, S. A. 1964. Patterns of mammalian reproduction. Comstock Publ. Assoc., Cornell Univ. Press, Ithaca, New York, 670 pp. AUGUSTINSSON, K. 1961. Multiple forms of esterase in verte- brate blood plasma. Ann. New York Acad. Sci., 94:844-860. Baber, R. S., AND J. S. Hau 1960. Mammalian remains from an Indiana cave. Jour. Mamm., 41:111-112. BaILey, V. 1905. Biological survey of Texas. ... N. Amer. Fauna, 25:1-222. 1932. Mammals of New Mexico. Ibid., 5331-412. BairD, S. F. 1855. Characteristics of some new species of Mammalia, collected by the U.S. and Mexican Boundary Survey, Major W. H. Emory, U.S.A. Commissioner. Proc. Acad. Nat. Sci. Philadelphia, 7:331-333. 1858. Mammals, in Reports of explorations and surveys .. . from the Mississippi River to the Pacific Ocean... . 8(1): xxi-xlviii + 1-757. Baker, R. H. 1942. Notes on small mammals of eastern Texas. Jour. Mamm., 23:343. 1951. Mammals from Tamaulipas, Mexico. Univ. Kansas Publ. Mus. Nat. Hist., 5:207-218. 1953. Mammals from owl pellets taken in Coahuila, Mexico. Trans. Kansas Acad. Sci., 56:253-254. 1956. Mammals of Coahuila, México. Univ. Kansas Publ. Mus. Nat. Hist., 9:125- BoD: BAKER, R. J., AND J. T. MASCARELLO 1969. Karyotypic analyses of the genus Neotoma (Cricetidae, Rodentia). Cytogenetics, 8:187-198. Baker, R. J., J. T. MASCARELLO, AND R. G. JORDAN 1970. Polymorphism in the somantic chro- mosomes of Neotoma micropus Baird, the plains woodrat. Experientia, 26:426-428. BANGs, O. 1898. The land mammals of peninsular Florida and the coast region of Georgia. Proc. Soc. Nat. Hist., 28: 158-235. Berry, R. J., AND A. G. SEARLE 1963. Epigenetic polymorphism of the ro- dent skeleton. Proc. Zool. Soc. Lon- don, 140:577-615. Brrcu, L. C., AnD P. R. EHRLICH 1967. Evolutionary history and taxonomy. Syst. Zool., 16:282-285. BrmneEy, E. C. 1970. Systematics of three species of wood- rats. . .. Unpublished Ph.D. Disser- tation, Univ. Kansas, Lawrence. Birney, E. C., AND J. E. PEREZ 1971. Inheritance of multiple hemoglobins in two species of woodrats, genus Neotoma (Rodentia: Cricetidae). Univ. Kansas Sci. Bull., 49:345-356. Birney, EF. C., AND S. L. TWOMEY 1970. Effects of sodium chloride on water consumption, weight, and survival in the woodrats, Neotoma micropus and WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 167 Neotoma_ floridana. Jour. Mamm., 51:372-375. Buarr, W. F. 1938. Ecological relationships of the mam- mals of the Bird Crest Region, north- eastern Oklahoma. Amer. Midl. Nat., 20:473-526. 1939a. New mammals from Texas and Okla- homa, with remarks on the status of Thomomys texensis Bailey. Occas. Papers Mus. Zool. Univ. Michigan, 403:1-7. 1939b. Faunal relationships and geographic distribution of mammals in Okla- homa. Amer. Midl. Nat., 22:85-133. 1940. A contribution to the ecology and faunal relationships of the mammals of the Davis Mountain Region, south- western Texas. Misc. Publ. Mus. Zool. Univ. Michigan, 46:1-39. 1943a. Ecological distribution of mammals in the Tularosa Basin, New Mexico. Contrib. Lab. Vert. Biol. Univ. Mich- igan, 20: 1-24. 1943b. Populations of the deer-mouse and associated small animals in the mes- quite association of southem New Mexico. Ibid., 21:1-40. 1952. Mammals of the Tamaulipan biotic province in Texas. Texas Jour. Sci., 4:230-250. 1954. Mammals of the mesquite plains bi- otic district in Texas and Oklahoma, and speciation in the central grass- lands. Ibid., 6:235-264. 1958. Distributional patterns of vertebrates in the southern United States in re- lation to past and present environ- ments. Pp. 433-468 in Zoogeography (C. L. Hubbs, ed.). Amer. Assoc. Advance. Sci., 51:x + 1-509. Buair, W. F., AND C. E. MILLER, JR. 1949. The mammals of the Sierra Vieja re- gion, southwestern Texas, with re- marks on the biogeographic position of the region. Texas Jour. Sci., 1:67- 92. Botce, R. 1969. Water intake as a function of ease of access in Neotoma. Jour. Mamm., 50:605-607. Bootu, E. S. 1957. Mammals collected in Mexico from 1951 to 1956 by the Walla Walla College Museum of Natural History. Walla Walla College Publ., 20:1-19. Boynpen, A. A. 1964. Perspectives in systematic serology. Pp. 75-99 in Taxonomic biochemistry and serology (C. A. Leone, ed.). The Ronald Press, New York, x + 728 pp. BRAND, L. R., AND R. KE. RyCKMAN 1969. Biosystematics of Peromyscus eremi- cus, P. guardia, and P. interparietalis. Jour. Mamm., 50:501-513. Brown, B. 1909. The Conard Fissure, a Pleistocene bone deposit in northern Arkansas: with descriptions of two new genera and twenty new species of mammals. Mem. Amer. Mus. Nat. Hist., 9:155- 208. Brown, J. H. 1968. Adaptation to environmental temper- ature in two species of woodrats, Neotoma cinerea and N. albigula. Misc. Publ. Mus. Zool. Univ. Mich- igan, 135:1-48. Brown, J. H., AND A. K. LEE 1969. Bergmann’s rule and climatic adapta- tion in woodrats (Neotoma). Evolu- tion, 23:329-338. Brown, L. N. 1963. Maturational and seasonal molts in Peromyscus boylii. Amer. Midl. Nat., 70:466-469. 1969. Reproductive characteristics of the Mexican woodrat at the northern limit of its range in Colorado. Jour. Mamm., 50:536-541. Bryson, R. A., D. A. BAERREIS, AND W. M. WENDLAND 1970. The character of late-glacial and post- glacial climatic changes. Pp. 55-74 in Pleistocene and Recent environ- ments of the Central Great Plains (W. Dort, Jr., and J. K. Jones, Jr., eds.). Special Publ., Univ. Kansas Dept. Geol., 3:xii +- 1-433. Burt, W. H. 1954. The subspecies category in mammals. Syst. Zool., 3:99-104. 1960. Bacula of North American mammals. Misc. Publ. Mus. Zool. Univ. Mich- igan, 113:1-76. Burt, W. H., ANnp F. S. BARKALOW, JR. 1942. A comparative study of the bacula of wood rats (Subfamily Neotominae). Jour. Mamm., 23:287-297. Cary, M. 1911. A biological survey of Colorado. N. Amer. Fauna, 33:1-256. CAVALLI-SFORZA, L. L., AnD A. W. F. Epwarps 1967. Phylogenetic analysis: models and estimation procedures. Evolution, 21: 550-570. CHapMan, A. O. 1951. The estrous cycle in the woodrat, Neotoma floridana. Univ. Kansas Sci. Bull., 34:267-299. 168 MISCELLANEOUS PUBLICATION MUSEUM OF NATURAL HISTORY CHERNOFF, A. I., AND N. M. PETTIT, JR. 1964. The amino acid composition of hemo- globin. III. A qualitative method for identifying abnormalities of the poly- peptide chains of hemoglobin. Blood, 24:750-756. CuHoateE, J. R. 1970. Systematics and zoogeography of Middle American shrews of the genus Cryptotis. Univ. Kansas Publ. Mus. Nat. Hist., 19:195-317. CHOATE, J. R., AND H. H. GENoways 1967. Notes on some mammals from Ne- braska. Trans. Kansas Acad. Sci., 69: 238-241. CINADER, B. 1957. Antibodies against enzymes. Ann. Rev. Microbiol., 11:371-390. Cocxrum, E. L. 1952. Mammals of Kansas. Univ. Kansas Publ. Mus. Nat. Hist., 7:1-303. Cougs, E. 1877. Monographs of North American Ro- dentia. No. I—Muridae. Pp. x + 1-264, in Monographs of North American Rodentia (E. Coues and J. A. Allen, eds.). Bull. U.S. Geol. Surv. Territories, 11:xii + 1-1091. Cross, J. C. 1931. A comparative study of the chromo- somes of rodents. Jour. Morphol., 52:373-401. Crowson, R. A. 1965. Classification, statistics and phylog- eny. Syst. Zool., 14:144-148. DataQuest, W. W. 1951. Six new mammals from the state of San Luis Potosi, México. Jour. Wash- ington Acad. Sci., 41:361-364. 1953. Mammals of the Mexican state of San Luis Potosi. Louisiana State Univ. Stud. Biol. Sci. Ser., 1:1-229. 1968. Mammals of north-central Texas. Southwestern Nat., 13:13-21. Dauguest, W. W., E. Rory, AND F. Jupp 1969. The mammal fauna of Schulze Cave, Edwards County, Texas. Bull. Flor- ida State Mus., 13:205-276. Davis, W. B. 1960. The mammals of Texas. Bull. Texas Game Fish Comm., 27:1-252. IDi Cero Irs IRE 1937. Mammals of the San Carlos Moun- tains and vicinity. Univ. Michigan Studies, Sci. Ser., 12:245-268. Dice, L. R., AND H. J. Leraas 1936. A graphic method for comparing sev- eral sets of measurements. Contrib. Lab. Vert. Genetics Univ. Michigan, Selle} Duranp, M., AND R. SCHNEIDER 1963. Etude immunoelectrophorétique du sérum de chien et ses relations anti- géniques avec les diverses especes de mammiféres. Arch. Inst. Pasteur, 40:91-95. EcoscugE, H. J. 1957. The desert woodrat: a laboratory colony. Jour. Mamm., 38:472-481. 1962. The bushy-tailed wood rat: a labora- tory colony. Jour. Mamm., 43:328- 337. Enruicnu, P. R. 1961. Has the biological species concept outlived its usefulness. Syst. Zool., 10:167-176. Euruicu, P. R., AnD R. W. Hom 1963. The process of evolution. McGraw- Hill Book Co., Inc., New York, xvi + 347 pp. Enruicu, P. R., AnD P. H. RAVEN 1969. Differentiation of populations. Sci- ence, 165: 1228-1232. EvuiorT, D. G. 1899. Description of apparently new species and sub-species of mammals from Oklahoma Territory. Field Colum- bian Mus. Publ. 37, Zool. Ser., 1:279- 282. 1901. A synopsis of the mammals of North America and the adjacent seas. Ibid., Dixy EAT A. FELDMAN, H. W. 1935. Notes on two species of wood rats in captivity. Jour. Mamm., 16:300-303. FINLEY, R. B., Jr. 1958. The wood rats of Colorado: distribu- tion and ecology. Univ. Kansas Publ. Mus. Nat. Hist., 10:213-552. Fircu, H. S., AND D. G. RAINEY 1956. Ecological observations on the wood- rat, Neotoma floridana. Ibid., 8:499- DOG: FLorkIn, M. 1964. Perspectives in comparative biochem- istry. Pp. 51-74 in Taxonomic bio- chemistry and serology (C. A. Leone, ed.). The Ronald Press Co., New York, x + 728 pp. FOREMAN, C. W. 1960. Electromigration properties of mam- malian hemoglobins as taxonomic cri- teria. Amer. Midl. Nat., 64:177-186. 1964. Tryptic peptide pattems of some mammalian hemoglobins. Jour. Cell. Comp. Physiol., 63:1-6. GABRIEL, K. R. 1964. A procedure for testing the homo- geneity of all sets of means in analy- WOODRATS (GENUS NEOTOMA) IN CENTRAL NORTH AMERICA 169 sis of variance. Biometrics, 20:459- AT7. GABRIEL, K. R., AND R. R. SOKAL 1969. A new statistical approach to geo- graphic variation analysis. Syst. Zool., 18:259-278. Genoways, H. H. 1971. Systematics and evolution of the spiny pocket mice of the genus Lio- mys. Unpublished Ph.D. Dissertation, Univ. Kansas, Lawrence. GERBER, J. D. 1968. Electrophoretic and immunologic comparisons of the serum proteins of bats. Unpublished Ph.D. Dissertation, Univ. Kansas, Lawrence. GERBER, J. D., AND E. C. BIRNEY 1968. Immunological comparisons of four subgenera of ground squirrels. Syst. Zool., 17:413-416. GIDLEy, J. W., AND C. L. GaziIn 1933. New Mammalia in the Pliestocene fauna from Cumberland Cave. Jour. Mamm., 14:343-357. Guass, B. P. 1949. Report on the mammals of the Black Mesa region, Oklahoma. Proc. Okla- homa Acad. Sci., 30:26-30. GoERTz, J. W. 1970. An ecological study of Neotoma flori- dana in Oklahoma. Jour. Mamm., 51:94-104. GoLpMAN, E. A. 1905. Twelve new wood rats of the genus Neotoma. Proc. Biol. Soc. Washing- ton, 18:27-33. 1910. Revision of the wood rats of the ge- nus Neotoma. N. Amer. Fauna, 31:1- 124. 1933. New mammals from Arizona, New Mexico, and Colorado. Jour. Wash- ington Acad. Sci., 23:463-473. Goopw1, G. G. 1954. Mammals from Mexico collected by Marian Martin for The American Mu- seum of Natural History. Amer. Mus. Novitates, 1689:1-16. Guitpay, J. E., P. S. MarTIn, AND A. D. McCrapy 1964. New Paris No. 4: a late Pleistocene cave deposit in Bedford County, Pennsylvania. Bull. Natl. Speleol. Soc., 26: 121-194. EVATS es ee 1926. Changes during growth in the skull of the rodent Otospermophilus gram- murus beecheyi. Univ. California Publ. Zool., 21:355-404. 1955. A new subspecies of wood rat from Nayarit, Mexico, with new name- combinations for the Neotoma mexi- cana group. Jour. Washington Acad. Sci., 45:328-332. Hatt, E. R., anp W. W. DALQuEsT 1963. The mammals of Veracruz. Univ. Kansas Publ. Mus. Nat. Hist., 14:165- 362. Hau, E. R., Anp H. H. GENoways 1970. Taxonomy of the Neotoma albigula- group of woodrats in central Mexico. Jour. Mamm., 51:504-516. Hatt, E. R., AND K. R. KELSON 1959. The mammals of North America. The Ronald Press Co., 2: viii -+- 547-1083 JE 7S): Hamiton, W. J., Jr. 1953. Reproduction and young of the Flor- ida wood rat, Neotoma f. floridana (Ord). Jour. Mamm., 34:180-189. Hays, H. A. 1958. The distribution and movement of small mammals in central Oklahoma. Ibid., 39:235-244. Hences, S. R. 1969. Epigenetic polymorphism in popula- tions of Apodemus sylvaticus and Apodemus flavicollis (Rodentia, Mu- ridae). Jour. Zool. Soc. London, 159:425-442, Hipparp, C. W. 1963.