READINGS IN MAMMALOGY J. Knox Jones, Jr. Sydney Anderson Museum of Natural History The University of Kansas 19 70 HARVARD UNIVERSITY Library of the Museum of Comparative Zoology Gift of E. E. Williams Readings In Mammalogy Selected from the original literature and introduced with comments by J. Knox Jones, Jr. The University of Kansas, Lawrence AND Sydney Anderson The American Museum of Natural History, New York City /SCi Z. .- Ke ' .\oa Qt. :-v ' 1 (' nus I ■\' Monograph No. 2 Museum of Natural History The University of Kansas 1970 Monograph No. 2 Museum of Natural History, The University of Kansas ix-{-586 pp., 1970 Lawrence • Kansas MUS. COMP. ZOOL" LIBRARY MAR 12 1980 HARVARD , UNIVERSITY ' PRINTED BY THE XJNIVERSITY OF KANSAS PRINTING SERVICE PREFACE This anthology is intended as an introduction to the study of mammals, principally for those who already have some biological background and who want to know the general scope of the field of mammalogy. The subdisciplines or specialties of mammalog>% its relationship to other biological fields, and specific examples of the type of work done by mammalogists are here intro- duced by means of a selection of complete papers in their original form. We hope that these will help college students looking forward to graduate work in biology obtain a realistic general view of mammalogy as a possible specialty. Also, beginning graduate students in related disciplines such as ornithology, mammahan physiology, or ecology, or undergraduate majors in wildlife man- agement, may find their perspectives broadened by perusal of the present selection of papers and the introductory commentaries. The published literature on the scientific study of mammals, which, broadly speaking, comprises the field of mammalogy, includes about 90,000 separate papers, and new papers are now being published at the rate of 5000 to 6000 each year, the actual number depending on where one draws the borders of the discipline. Precise borders do not exist. Mammalogy, like other scientific fields, draws from and contributes to various areas of human knowledge. Our selection of the 64 papers here reproduced was influenced by : ( 1 ) our con- cept of the scope of mammalogy and of a reasonable and representative bal- ance of its parts at this time; (2) our desire to illustrate various ways in which a wide variety of information can be presented in published form; and (3) our awareness that most of our readers will be English-speaking Americans, which led us to use articles published in English and selected predominantly from American sources. Nevertheless, we judge that the broad sweep of concepts and methods portrayed is relevant to students of mammalogy in all parts of the world. We have selected short papers, in general less than 20 pages in length, in preference to either longer papers or excerpts therefrom — chiefly because of space, but also because we want the serious student, who may later contribute to the literature himself, to see each published work in its entirety as one tangible contribution to knowledge. He can then grasp its concept, its methodology, its organization, its presentation, its conclusions, and perhaps even its limitations. On the latter score we would note that, although we think the papers selected are worthy contributions and make the points we wish to emphasize, we do not pretend to have selected the finest papers ever published. We could have used a somewhat different selection to serve much the same purpose, and we are sure someone else would use a different selection to represent his views of mammalogy. Although many college and university libraries have some or most of the journals and other sources from which papers were selected, we decided an anthology was warranted for those who want an overview of the field, who may not know where to find the relevant literature, or who want the conve- nience of a collection of separate papers. Many undergraduate students are largely unaware of the existence of, or the nature of, the technical literature of science, although their textbooks are replete with terminal citations. Hope- Monograph No. 2 Museum of Natural History, The University of Kansas ix-\-586 pp., 1970 Lawrence • Kansas MUS. COMP. ZOOU LIBRARY MAR 12 1980 HARVARD UNIVERSITY ' PRINTED BY THE tJNIVERSITY OF KANSAS PRINTING SERVICE PREFACE This anthology is intended as an introduction to the study of mammals, principally for those who already have some biological background and who want to know the general scope of the field of mammalogy. The subdisciplines or specialties of mammalogy^ its relationship to other biological fields, and specific examples of the type of work done by mammalogists are here intro- duced by means of a selection of complete papers in their original form. We hope that these will help college students looking forward to graduate work in biology obtain a realistic general view of mammalogy as a possible specialty. Also, beginning graduate students in related disciplines such as ornithology, mammalian physiology, or ecology, or undergraduate majors in wildlife man- agement, may find their perspectives broadened by perusal of the present selection of papers and the introductory commentaries. The published literature on the scientific study of mammals, which, broadly speaking, comprises the field of mammalogy, includes about 90,000 separate papers, and new papers are now being published at the rate of 5000 to 6000 each year, the actual number depending on where one draws the borders of the discipline. Precise borders do not exist. Mammalogy, like other scientific fields, draws from and contributes to various areas of human knowledge. Our selection of the 64 papers here reproduced was influenced by: (1) our con- cept of the scope of mammalogy and of a reasonable and representative bal- ance of its parts at this time; (2) our desire to illustrate various ways in which a wide variety of information can be presented in published form; and (3) our awareness that most of our readers will be English-speaking Americans, which led us to use articles published in English and selected predominantly from American sources. Nevertheless, we judge that the broad sweep of concepts and methods portrayed is relevant to students of mammalogy in all parts of the world. We have selected short papers, in general less than 20 pages in length, in preference to either longer papers or excerpts therefrom — chiefly because of space, but also because we want the serious student, who may later contribute to the literature himself, to see each published work in its entirety as one tangible contribution to knowledge. He can then grasp its concept, its methodology, its organization, its presentation, its conclusions, and perhaps even its limitations. On the latter score we would note that, although we think the papers selected are worthy contributions and make the points we wish to emphasize, we do not pretend to have selected the finest papers ever published. We could have used a somewhat different selection to serve much the same pmpose, and we are sure someone else would use a different selection to represent his views of mammalogy. Although many college and university libraries have some or most of the journals and other sources from which papers were selected, we decided an anthology was warranted for those who want an overview of the field, who may not know where to find the relevant literature, or who want the conve- nience of a collection of separate papers. Many undergraduate students are largely unaware of the existence of, or the nature of, the technical literature of science, although their textbooks are replete with terminal citations. Hope- fully, many of our readers will be provoked to go to the library to seek out additional information on mammals, once they learn the interest and value of subjects treated in the pages of the Journal of Mammalogy and other scien- tific sources. We have formulated our ideas for this anthology over the past three years and we are grateful for suggestions received from many persons in that time. We are grateful also to editors, publishers, and living authors for permission to include their works in Readings in Mammalogy. J. Knox Jones, Jr. Sydney Anderson IV CONTENTS Introduction __ — -. 3 Section 1 — Systematics Grinnell, J. The museum conscience. Museum Work, 4:62-63, 1922 9 Thomas, O. Suggestions for the nomenclature of the cranial length measurements and of the cheek-teeth of mammals. Proc. Biol. Soc. Washington, 18:191-196, 1905 11 Hall, E. R. Criteria for vertebrate subspecies, species and genera: the mammals. Ann. New York Acad. Sci., 44:141-144, 1943 17 LiDiCKER, W. Z., Jr. The nature of subspecies boundaries in a desert rodent and its implications for subspecies taxonomy. Syst. Zool., 11:160-171, 1962 _. . 21 Hershkovitz, p. Generic names of the four-eyed pouch opossum and the woolly opossum (Didelphidae). Proc. Biol. Soc. Washington, 62: 11-12, 1949 33 Allen, J, A. Two important papers on North-American mammals. Amer. Nat., 35:221-224, 1901 35 Merriam, C. H. Descriptions of two new species and one new subspecies of grasshopper mouse, with a diagnosis of the genus Onychomys, and a synopsis of the species and subspecies. N. Amer. Fauna, 2:1-5, pi. 1, 1889 __.__.. 39 Handley, C. O., Jr. Descriptions of new bats {Choeroniscus and Rhinophylla) from Colombia. Proc. Biol. Soc. Washington, 79:83-88, 1966 46 Benson, S. B. The status of Reithrodontomys montanus (Baird). Jour. Mamm., 16:139-142, 1935 52 Hoffmeister, D. F., and L. de la Torre A revision of the woodrat Neotoma stephensi. Jour. Mamm., 41 :476-491, 1960 56 Lawrence, B., and W. H. Bossert Multiple character analysis of Canis lupus, latrans, and familiaris, with a discussion of the relationships of Canis niger. Amer. Zool., 7:223-232, 1967 72 Patton, J. L., and R. E. Dingman Chromosome studies of pocket gophers, genus Thomomys. I. The specific status of Thomomys umbrinus (Richardson) in Arizona. Jour. Mamm., 49:1-13, 1968 82 Nadler, C. F., and C. E, Hughes Serum protein electrophoresis in the taxonomy of some species of the ground squirrel subgenus Spermophilus. Comp. Biochem. Physiol, 18:639-651, 1966 95 Machado- Allison, C. E. The systematic position of the bats Desmodus and Chilonycteris, based on host-parasite relationships (Mammalia; Chiroptera). Proc. Biol. Soc. Washington, 80 : 223-226, 1967 108 Section 2 — Anatomy and Physiology Hill, J. E. The homology of the presemimembranosus muscle in some rodents. Anat. Rec, 59:311-314, 1934 115 Hooper, E. T. The glans penis in Sigmodon, Sigmomys, and Reithrodon (Rodentia, Cricetinae). Occas. Papers Mus. Zool., Univ. Michigan, 625:1-11, 1962 119 Hughes, R. L. Comparative morphology of spermatozoa from five marsupial families. Australian Jour. Zool., 13:533-543, pi. 1, 1965 130 MOSSMAN, H. W. The genital system and the fetal membranes as criteria for mammalian phylogeny and taxonomy. Jour. Mamm., 34:289-298, 1953 142 NOBACK, C. R. Morphology and phylogeny of hair. Ann. New York Acad. Sci., 53:476-492, 1951 152 Vaughan, T. a. Morphology and flight characteristics of molossid bats. Jour. Mamm., 47:249-260, 1966 169 Rabb, G. B. Toxic salivary glands in the primitive insectivore Solenodon. Nat. Hist. Misc., Chicago Acad. Sci., 170:1-3, 1959 181 Forbes, R. B. Some aspects of the water economics of two species of chipmunks. Jour. Mamm., 48:466-468, 1967 184 Pearson, O. P. The oxygen consumption and bioenergetics of harvest mice. Physiol. Zool., 33:152-160, 1960 187 Bartholomew, G. A., and R. E. MacMillen Oxygen consumption, estivation, and hibernation in the kangaroo mouse, Microdipodops pallidus. Physiol. Zool., 34:177-183, 1961 196 Scholander, p. F., and W. E. Schevill Counter-current vascular heat exchange in the fins of whales. Jour. Applied Physiol., 8:279-282, 1955 203 Section 3 — Reproduction and Development Hamilton, W. J., Jr. The reproductive rates of some small mammals. Jour. Mamm., 30:257-260, 1949 209 Spencer, A. W., and H. W. Steinhoff An explanation of geographic variation in litter size. Jour. Mamm., 49:281-286, 1968 --- 213 Sharmax, G. B. The effects of suckling on normal and delayed cycles of reproduc- tion in the red kangaroo. Z. Siiugetierk., 30:10-20, 1965 219 Wright, P. L., and M. W. Coulter Reproduction and growth in Maine fishers. Jour. Wildlife Mgt., 31:70-87, 1967 230 Jones, C. Growth, development, and wing loading in the evening bat, Nycticeiiis liwneralis (Rafinesque). Jour. Mamm., 48: 1-19, 1967 248 Butter worth, B. B. A comparative study of growth and development of the kangaroo rats, Dipodomys deserti Stephens and Dipodomys merriami Mearns. Growth, 25:127-138, 1961 267 Allen, J. A. Cranial variations in Neotoma micropus due to growth and individual differentiation. Bull. Amer. Mus. Nat. Hist., 6:233-246, pi. 4, 1894 279 LiNZEY, D. W., and A. V. Linzey Maturational and seasonal molts in the golden mouse, Ochrotomys nuttalli. Jour. Mamm., 48:236-241, 1967 294 Section 4 — Ecology and Behavior Caughley, G. Mortality patterns in mammals. Ecology, 47:906-918, 1966 303 Frank, F. The causality of microtine cycles in Germany. Jour. Wildlife Mgt., 21:113-121, 1957 316 Burt, W. H. Territoriality and home range concepts as applied to mammals. Jour. Mamm., 24:346-352, 1943 325 Miller, G. S., Jr. Migration of bats on Cape Cod, Massachusetts. Science, 5:541-543, 1897 -..- 332 Brown, L. N. Ecological distribution of six species of shrews and comparison of sampling methods in the central Rocky Mountains. Jour. Mamm., 48:617-623, 1967 335 Davis, D. E., and J. J. Christian Changes in Norway rat populations induced by introduction of rats. Jour. Wildlife Mgt., 20:378-383, 1956 342 Pearson, O. P. A traffic survey of Microtus-Reithrodontomys runways. Jour. Mamm., 40:169-180, 1959 ..-' 348 EsTES, R. D., and J. Goddard Prey selection and hunting behavior of the African wild dog. Jour. Wildlife Mgt., 31:52-70, 1967 360 Layne, J. N. Homing behavior of chipmunks in central New York. Jour. Mamm., 38:519-520, 1957 379 SUTHERS, R. A. Comparative echolocation by fishing bats. Jour. Mamm., 48:79-87, 1967 381 ElSENBERG, J. F. The intraspecific social behavior of some cricetine rodents of the genus Peromyscus. Amer. Midland Nat., 69:240-246, 1963 390 McCarley, H. Ethological isolation in the cenospecies Peromyscus leucopus. Evolution, 18:331-332, 1964 397 Lyman, C. P. Activity, food consumption and hoarding in hibernators. Jour. Mamm., 35:545-552, 1954 399 Real, R. O. Radio transmitter-collars for squirrels. Jour. Wildlife Mgt., 31:373-374, 1967 407 Section 5 — Paleontology and Evolution Reed, C. A. The generic allocation of the hominid species hahilis as a problem in systematics. South African Jour. Sci., 63:3-5, 1967 411 HiBBARD, C. W. Microtus pennsylvanicus (Ord) from the Hav Springs local fauna of Nebraska. Jour. Paleont., 30:1263-1266, 1956 414 Wilson, R. W. Tvpe localities of Cope's Cretaceous mammals. Proc. South Dakota Acad. Sci., 44:88-90, 1965 418 Radinsky, L. B. The adaptive radiation of the phenacodontid condylarths and the origin of the Perissodactyla. Evolution, 20:408-417, 1966 421 Miller, G. S., Jr., and J. W. Gidley Synopsis of the supergeneric groups of rodents. Jour. Washington Acad. Sci., 8:431-448, 1918 431 Wood, A. E. Grades and clades among rodents. Evolution, 19:115-130, 1965 449 Durrant, S. D., and R. M. Hansen Distribution patterns and phvlogeny of some western ground squirrels. Syst. Zool., 3:82-85, 1954 465 Guthrie, R. D. Variability in characters undergoing rapid evolution, an analysis of Microtus molars. Evolution, 19:214-233, 1965 469 Jansky, L. Evolutionarv adaptations of temperature regulation in mammals. Z. Saugetierk., 32:167-172, 1967 - -- 489 Section 6 — Zoogeography and Faunal Studies Dice, L. R. The Canadian Biotic Province with special reference to the mammals. Ecology, 19:503-514, 1938 497 GUILDAY, J. E. Pleistocene zoogeography of the lemming, Dicrostonyx. Evolution, 17:194-197, 1963 509 KooPMAN, K. F., and P. S. Martin Subfossil mammals from the Gomez Farias region and the tropical gradient of eastern Mexico. Jour. Mamm., 40:1-12, 1959 513 FiNDLEY, J. S., and S. Anderson Zoogeography of the montane mammals of Colorado. Jour. Mamm., 37:80-82, 1956 525 Jones, J. K., Jr., and T. E. Lawlor Mammals from Isla Cozumel, Mexico, with description of a new species of harvest mouse. Univ. Kansas Publ., Mus. Nat. Hist., 16:411-419, 1965 528 Davis, W. B. Relation of size of pocket gophers to soil and altitude. Jour. Mamm., 19:338-342, 1938 537 Davies, J. L. The Pinnipedia: an essav in zoogeography. Geographical Rev., 48:474-493, 1958 .' 542 Hagmeier, E. M. A numerical analysis of the distributional patterns of North American mammals. II. Re-evaluation of the provinces. Syst. Zool., 15:279-299, 1966 562 Literature Cited 583 READINGS IN MAMMALOGY INTRODUCTION The overall unity of the different fields of science and of other aspects of human experience, or at least their interdependence, is evident in both theory and practice. Nevertheless, this is an age of specialization. The sheer volume of information, the current rate of increase in knowledge, the changing and often elaborate techniques that must be learned, and human limitations all have contributed to the production of specialties. A definition of the specialty of mammalogy as "all scientific study of mam- mals" is too broad, for that definition encompasses, for example, all parts of animal physiology in which any mammal, such as a white rat in the laboratory, may happen to be used. It also would include much of medical practice, because humans are mammals. Generally speaking, those scientists who call themselves mammalogists are interested in the mammal as an animal — as an organism — not just as a specific case of some more general phenomenon, be it the nature of life or the nature of the nerve impulse. For example, a physiolo- gist who is interested in comparative studies between different mammals or in the function of a physiological process as an adaptive mechanism may regard himself as a mammalologist. A physiologist who studies one kind of laboratory animal and is interested in explaining a process in terms of progressively sim- pler mechanisms rarely will regard himself as a mammalogist. Both types of study, of course, contribute to biological knowledge. Life is best comprehended in terms of four basic concepts: first, that biology, as all of science, is monistic, assuming one universe in which the same natural laws apply to living and non-living things; second, that life is a dynamic and self-perpetuating process; third, that the patterns of life change with time; finally, that these factors together have resulted in a diversity of living forms. Different branches of biology tend to focus or concentrate on different concepts. Thus, the above four concepts are focal points, respectively, of ( 1 ) physiology and biochemistry, (2) ecology, (3) evolutionary biology, and (4) systematic biology. Mammalogy is the study of one systematic group or taxon, the Class Mammalia. Studies emphasizing different aspects of mammalian biology are evident in our section headings and in the specific papers repro- duced. Biology as a whole and mammalogy specifically may be best likened to a woven fabric rather than to a series of compartments. We feel that a unifying conceptual scheme for "mammalogy" lies in the realm of "systematic mammalogy." This scheme is unifying because it includes the basis for subsequent study and the only meaningful framework for the synthesis of existing knowledge of mammals. On this point, George Gaylord Simpson, in introducing his classic The Principles of Classification and a Classi- fication of Mammals (1945) wrote (Simpson at that time used the term "taxonomy" as we use "systematics" ) : "Taxonomy is at the same time the most elementary and the most inclusive part of zoology, most elementary because animals cannot be discussed or treated in a scientific way until some taxonomy has been achieved, and most inclusive because taxonomy in its various guises and branches eventually gathers together, utilizes, summarizes, and imple- ments everything that is known about animals, whether morphological, physiological, psychological, or ecological." Knowledge of the identity of any animal studied is essential so that the results may be compared with other knowledge about the same kind of animal and with the same kind of knowledge about different animals. We originally had hoped to develop the history of mammalogy along with our other objectives, but when the hard fact of page limitation was faced, some selections whose chief justification was historical were sacrificed. In the comments beginning each section, some historical information helps place the selections in an understandable framework. To attain variety we have included papers both of restricted and of general scope; for example, papers pertaining to local faunas and continental faunas, to higher classification and infraspecific variation, and to contemporary serum proteins and millions of years of evolution. Every serious student of mammalogy, whether amateur or professional, researcher or compiler, writer or reviewer, artist or teacher, must learn to use the literature. One does not learn all about mammals because that is impossi- ble. One learns what one can, where to look for further information, and, more important, how to evaluate what one finds. Most of the literature on mammals is in technical journals, a few of which are devoted exclusively to mammalogy: Journal of Mammalogy (USA), Mammallv (France), Zeitschrift fur Saugetierkunde (Germany), Sauge- TiERKLrxDLiCHE MiTTEiLUNGEN (Germany), LuTRA (Benelux countries). Lynx (Czechoslovakia), Acta Theriologica (Poland), The Journal of the Mam- MALOGICAL SOCIETY OF JaPAN, AuSTR.\LLVN MaMMAL SOCIETY BULLETIN, and Bulletin of the British Mammal Society. Also there are the specialized Folia Primatologica, an international journal of primatology, founded in 1963, and a number of serial publications such as Journal of Wildlife Man- agement, Bulletin of the Wildlife Disease Association, and journals issued by various game departments and conservation agencies that may deal in large part, but not exclusively, with mammals. However, much of the published information on mammals, as on most aspects of biology, is widely scattered. About 40 journals include 50 per cent of the current Hterature, but to cover 70 per cent, at least 150 journals must be consulted. Articles in the Journal of Mammalogy (now more than 800 pages each year) comprise only about three per cent of all current titles on mammals, for example. Some categories of literature other than journals are books, symposia, transactions of various meetings or groups such as the Transactions of the North American WildHfe and Natural Resources Conference (the 34th was issued in 1969), yearbooks such as the International Zoo Yearbook (the tenth was published in 1970), newsletters such as the Laboratory Primate Newslet- ter, Carnivore Genetics Newsletter, or Bat Research News, major revisions or compilations of special subjects, bibHographies, and abstracts. The chief bibliographic sources for mammalogists are the Journal of Mammalogy, through its lists of Recent Literature, Saugetierkut>jdliche Mitteilungen, through its "Schriftenschau" section, the Zoological Record, published by the Zoological Society of London, Biological Abstracts (quite incomplete for some branches of mammalogy), and the quarterly Wildlife Review that is issued by the U.S. Fish and WildUfe Service (along with the three collections of Wildlife Abstracts — a misnomer because only citations are included — compiled therefrom and published in 1954, 1957, and 1964); one especially useful bibliography to older papers on North American mammals is that com- piled by Gill and Coues (in Coues and Allen, 1877). Some individuals and institutions maintain records in the form of card files, or collections of sepa- rates, or both, over many years for special subjects, special geographic areas, or other more general purposes. It is important for the student to remember that large-scale faunal reports, catalogues, revisionary works, and the like often are valuable as bibliographic sources as well as sources of other informa- tion. Some of these reports are mentioned in the introductory remarks to several sections. An individual who delves into the literature on a particular subject usually begins with one or more pertinent recent works and proceeds backward in time by looking up publications cited in the later works or found in other bibliographic sources. An amazing amount of published information on a given subject frequently is available to the person willing to look for it. However, paradoxically, there is often no pubhshed record for what one might suppose to be nearly common knowledge. The questioning mind must return to nature when the Hterature holds no answer, exactly what the authors of papers reproduced in this anthol- ogy have done. A few decades ago only a small number of American colleges and universi- ties offered a formal course in mammalogy, and only since about 1950 have such courses been widely offered. It is not surprising, therefore, that only two textbooks, Cockrum's Introduction to Mammalogy (1962) and Principles in Mammalogy by Davis and Golley (1963) have been published in EngHsh. The formCT has a systematic orientation and the latter is predominantly eco- logical. Some instructors use general works like Recent Mammals of the World, A Synopsis of Families (edited by Anderson and Jones, 1967), Mam- mals of the World, a three- volume work by Walker et al. ( 1964), or Hamilton's ( 1939) American Mammals as texts or as references along with other suggested readings. Accounts of the mammals of certain states or regions also may be used as texts by persons in those places. Other general works of reference value are Bourhere's Natural History of Mammals ( 1954), Young's The Life of Mammals (1957), Crandall's Management of Wild Mammals in Captivity (1964), and the fascicles on mammals in the Traite de Zoologie (edited by Grasse, 1955 and later). Two classic general works less readily available are An Introduction to the Study of Mammals, Living and Extinct by Flower and Lydekker (1891) and Mammalia by Beddard (1902) in the Cambridge Natu- ral History series. Compact field guides to the mammals of a few parts of the world are avail- able, such as those of Burt and Grossenheider ( 1964 ) , Palmer ( 1954 ) , and Anthony (1928) for parts of North America, and Van den Brink (1967) for Europe. The dates in the tvvo preceding paragraphs suggest the recent expansion in the volume of work in mammalog)'. Another such measure is membership in The American Society- of Mammalogists, which grew from 252 in 1919 to more than 3200 in 1969, and half of the growth occurred after 1957. Persons inter- ested in mammalog)' are in\ited to appl>- for membership in this society', mem- bers of which receive the Journal of Mammalogy. Human medicine, veterinary- medicine, animal husbandry, and animal physiology (including much work with a comparatiNcly few species of mam- mals in the laboratory), all preceded mammalogy as separate disciplines deal- ing with mammals. Many of the first mammalogists (as defined here) trained themselves in one of these disciplines and some also practiced in fields other than mammalogy. C. Hart Merriam, who founded the U.S. Biological Survey, studied medicine, as did E. A. Mearns, who N\Tote on mammals of the Mexican boundary (1907). Harrison Allen %\Tote much of his first review of North American bats ( 1864 ) while on furloughs from duty as a surgeon in the Union army in the Ci\il War. Mammalogy continues to interact %\ith the above- mentioned fields to their mutual benefit. Another largely separate but partly overlapping field that flowered slightly later than mammalogy is genetics. We have included no papers on mammalian genetics as such, although the rele\ance of genetics is evident in some of our selections. A recent book on Comparative Genetics of Coat Color in Mammals by Searle (1968) contains about 800 references, including some fascinating works on species other than the oft-studied mouse (Mus musculus). Our six groupings of papers are somewhat arbitrary. Ecology is as closely allied to physiology or zoogeography as to behaxior, and anatomy could as well have been placed with development as with physiology. The present arrange- ment as to the sequence of sections and the contents of sections seems to be about as convenient and useful as any other, and that is the extent of our expec- tations. We imply no hierarchy of subdisciplines. In selecting works to be included here, we have, in addition to the consider- ations already noted, sought papers in which different kinds of mammals were compared, and in which different approaches, styles, and methods of presenta- tion were used. Individual papers often pertain to more than one area of study. In fact, we favored papers that illustrated the rele\ance of different disciplines and methods of study to each other. Perhaps the reader will be able to ap- preciate our moments of anguish as the final selections w-ere made for this anthology. Our introduction for each section is brief. We hope that our comments aid the reader in considering ( 1 ) some historical aspects that make the papers more meaningful, (2) the major areas of study and some major concepts that the papers illustrate, (3) the existence of related Hterature, to which we can only call attention by citing a few examples, and (4) the continuous transfer of ideas, methods, and results from one worker to another, from one field of science to another, and between science and other fields of human endeavor. SECTION 1— SYSTEMATICS A sound classification provides the necessary framework upon which other knowledge about mammals can be built. In order to classify organisms, it first is necessary to know their similarities and differences; in other words, structures and their functions need to be observed, described, and compared, and taxa need to be recognized and named before a useful and meaningful classification can be constructed. The field of study relating to classification frequently is called "taxonomy," although the broader term "systematics" is also widely used and is preferred by us. Prior to the first decade or so of the 20th century, the practice of mamma- Han taxonomy generally was based on a "hit-or-miss" typological approach, which, although it fostered considerable advancement in cataloguing the faunas of the world, was limited in perspective and potential. The develop- ment of evolutionary thought and the spectacular growth of genetics have led to the "new systematics," the biological species concept of today, as discussed in detail in such syntheses as Huxley (1943), Mayr et al. (1953), Simpson (1961), and Mayr (1963 and 1969), among others. Blackwelder's (1967) recent text in taxonomy also is deserving of mention here. Technological advances in the means of collecting, preparing, and storing specimens resulted in the accumulation of series of individuals of the same species (the invention of the break-back mouse trap might be mentioned here along with the relatively recent widespread use of mist nets for capturing bats ) and thus in turn allowed for assessment of variations within and between populations. Sophisticated studies of intergradation, hybridization, and the cenospecies concept in the past few decades are examples of results from tech- nological and conceptual advances in this area. To imply that all early taxonomic treatments of mammals were either inconsequential or poorly conceived would be a gross error. Pallas' (1778) revision of rodents, for example, was a monumental work far advanced for its day, as were many other outstanding contributions by 18th and 19th century mammalogists that could be mentioned. Nevertheless, one has only to compare the descriptions and accounts of Pallas with those found in papers reprinted here by Merriam, Handley, and Hoffmeister and de la Torre to appreciate the tremendous revolution in systematic practice. It is of interest to note that the 80-year-old paper by Merriam still is valid with reference to the specific status of grasshopper mice. The short contribution by J. A. Allen not only provides an example of a review, but deals in some detail with two substantial revision- ary works published at the turn of the century. Among the larger modern revisionary studies that might be recommended to the student are those of Osgood (1909), Jackson (1928), Hooper (1952), Pearson (1958), Lidicker (1960), and Packard (1960). Ellerman's (1940, 1941, 1948) well organized review of living rodents and Hill's ( 1953 and subsequent volumes ) somewhat more rambling and less critical coverage of the primates are also noteworthy as attempts to summarize selected bodies of knowledge of important groups of mammals. The goal of scientific nomenclature is to assure that each kind of organism has a unique name, and only one name. The International Code of Zoological Nomenclature ( latest edition, 1961 ) forms the legalistic framework for dealing with nomenclature, both past and present. The presently reprinted paper by Hershkovitz points up some of the kinds of nomenclatorial problems faced by the systematist. The Code is administered by the International Commission on Zoological Nomenclature but, as Blair (1968) pointed out, the Commission "has no way of enforcing its decisions, and the burden of holding names to conformity with the [Code] falls on the individual worker and on editors of scientific pubHcations." The emphasis in this section is mostly at the level of species and subspecies ( for example, the papers by Benson and by Lidicker ) . Higher categories are dealt with primarily in Section 5. Attempts over the years by taxonomists to standardize techniques and definitions are illustrated by the papers of Thomas and Hall. The short essay by Grinnell also bears on this point. The final four selections illustrate the application of new techniques and concepts to specific taxonomic problems, all of which were clarified in ways that might have been impossible otherwise. The treatment by Lawrence and Bossert of the problem of the relationships of the red wolf uses standard cranial measurements but analyzes them by discriminant analysis, a concep- tually simple but computationally difficult statistical technique that has become generally applicable only with the development of digital computers. The karyological studies reported by Patton and Dingman treat some taxonomically interesting and previously confusing gophers in Arizona and enabled the authors to extend and refine the tentative conclusions based on other evidence; this paper also provides a good example of the careful integration of a new systematic approach with ecological and distributional considerations. Sero- logical methods have been employed in taxonomic studies for many years, but new methods such as the serum protein electrophoresis used by Nadler and Hughes are increasing the useful means of approaching systematic problems at the biochemical or molecular level. Machado-AlHson's contribution on host- parasite relationships illustrates the relevance of taxonomic data from one group, in this case ectoparasites, to the taxonomy of a different (host) group, bats. Many papers in other sections of this anthology touch on systematics in one way or another, and the usefulness to the taxonomist of information from a variety of sources will be immediately evident to the reader. Two journals devoted to the concepts and practices of systematics and in which contributions in mammalogy regularly appear are Systematic Zoology and Zeitschrift fur ZOOLOGISCHE SySTEMATIK. The Museum Conscience THE scientific museum, the kind of museum with which my re- marks here have chiefly to do, is a storehouse of facts, arranged acces- sibly and supported by the written records and labeled specimens to which they pertain. The purpose of a scien- tific museum is realized whenever some group of its contained facts is drawn upon for studies leading to publication. The investment of human energy in the formation and maintenance of a re- search museum is justified only in proportion to the amount of real knowledge which is derived from its materials and given to the world. All this may seem to be innocuous platitude — but it is genuine gospel, never-the-!ess, worth pondering from time to time by each and every museum administrator. It serves now as a background for my further comments. For worthy investigation based upon museum materials it is absolutely es- sential that such materials have been handled with careful regard for ac- curacy and order. To secure accuracy and order must, then, once the mere safe preservation of the collections of which he is in charge have been attended to, be the immediate aim of the curator. Order is the key both to the accessi- bility of materials and to the apprecia- tion of such facts and inferences as these materials afford. An arrange- ment according to some definite plan of grouping has to do with whole col- lections, with categories of specimens within each collection, with specimens within each general category, with the card indexes, and even with the place- ment of the data on the label attached to each specimen. Simplicity and clearness are fundamental to any scheme of arrangement adopted. Noth- ing can be more disheartening to a research student, except absolute chaos, than a complicated "system," in the invidious sense of the word, carried out to the absurd limits reo nmended by some so-called "efiiciency expert." However, error in this direction is rare compared with the opposite extreme, namely, little or no order at all. To secure a really practicable scheme of arrangement takes the best thought and much experimentation on the part of the keenest museum curator. Once he has selected or devised his scheme, his work is not done, moreover, until this scheme is in operation throughout all the materials in his charge. Any fact, specimen, or record left out of order is lost. It had, perhaps, better not exist, for it is taking space some- where; and space is the chief cost initially and currently in any museum. The second essential in the care of scientific materials is accuracy. Every item on the label of each specimen, every item of the general record in The Ml'seum Conscience the accession catalog, must be precise as to fact. Many errors in published literature, now practically impossible to "head off," are traceable to mistakes on labels. Label-writing having to do with scientific materials is not a chore to be handed over casually to a "25- cent-an-hour" girl, or even to the ordinary clerk. To do this essential work correctly requires an exceptional genius plus training. The important habit of reading every item back to copy is a thing that has to be acquired through diligent attention to this very point. By no means any person that happens to be around is capable of doing such work with reliable results. Now it happens that there is scarcely an institution in the country bearing the name museum, even though its main purpose be the quite distinct func- tion of exhibition and popular educa- tion, that does not lay more or less claim to housing "scientific collections." Yet such a claim is false, unless an adequate effort has been expended both to label accurately and to arrange systematically all of the collections housed. Only when this has been done can the collections be called in truth scientific. My appeal is, then, to every museum director and to every curator responsi- ble for the proper use as well as the safe preservation of natural history specimens. Many species of vertebrate animals are disappearing; some are gone already. All that the investigator of the future will have, to indicate the nature of such then extinct species, will be the remains of these species preserved more or less faithfully, along with the data accompanying them, in the museums of the country. I have definite grounds for present- ing this appeal at this time and in this place. My visits to the various larger museums have left me with the un- pleasant and very distinct conviction that a large portion of the vertebrate collections in this country, perhaps 90 per cent of them, are in far from satis- factory condition with respect to the matters here emphasized. It is ad- mittedly somewhat difficult for the older museums to modify systems of installation adopted at an early period. But this is no valid argument against necessary modification, which should begin at once with all the means avail- able— the need for which should, in- deed, be emphasized above the making of new collections or the undertaking of new expeditions. The older materials are immensely valuable historically, often irreplaceable. Scientific interests at large demand special attention to these materials. The urgent need, right now, in every museum, is for that special type of cura- tor who has ingrained within him the instinct to devise and put into opera- tion the best arrangement of his materials — who will be alert to see and to hunt out errors and instantly make corrections — who has the museum con- science. March 29, 1921. 10 Vol. XVIII, pp. 191-196 September 2, 1905 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON SUGGESTIONS FOR THE NOMENCLATURE OF THE CRANIAL LENGTH INIEASUREMENTS AND OF THE CHEEK-TEETH OF MAMMALS. BY OLDFIELD THOMAS. Although various reasons prevent the general success of such a wholesale revolution in scientific terms as is described in Wilder and Gage's Anatomical Technology (1882), where the many arguments in favor of accurate nomenclature are admira- bly put forth, yet in various corners of science improvements can be suggested which, if the workers are willing and in touch with each other, may be a real help in reducing the inconvenience of the loose or clumsy terminology commonly in vogue. Two such suggestions, due largely to the instigation of Mr. Gerrit S. Miller, Jr., form the subject of the present paper. I. Length Measurements of the Skull and Palate. In giving the length measurement of the skull, not only do different authors at present use different measurements in de- scribing the skulls of similar or related animals, but in doing so they designate these measurements by terms of which it is often difficult or impossible to make out the exact meaning. Such a name as ' ' basal length ' ' has I believe been used by one person or another for almost every one of the measurements to be here- 34— Proc. Biol. Soc. Wash., Vol. XVIII, 1905. (191) 11 192 Thomas — Nomenclature of Measurements. after defined, and readers are expi'cted to know by heart every- thing that the user has ever written on the sul)ject, footnotes and all, in order to understand what is meant by the particular term employed. Such a state of things has many inconveniences, and it is hoped the present communication, if it meets with the approval of other workers on the subject, may do a little toward putting an end to the existing confusion. As long ago as 1894,* Ijy agreeing with Dr. Nehring for the definition of the terms ])asal and basilar in our own future writ- ings, I made a first step in this direction, and the present is an amplification of the principle then adopted. All the difficulty has arisen from the fact that both at the anterior and the posterior ends of the skull there are two meas- urement points, so that there are four different ways in which the l)asal length of the skull may l)e taken, and under that name some authors have adopted nearly every one of them. It is clear that if a definite name be given to each one of the four measurements, authors, by using these names, will be en- abled to give the measurements they fancy without causing con- fusion in the minds of their readers as to their exact meaning. *Ann. & Mag. Nat. Hist., Ser. 6, XIII, p. 203. 12 Thomas — Nomenclature of Measurements. 193 The different points are: Anteriorly : 1 . Tiik Gnatiiion, the most anterior point of the premaxilke, on or near the middle line. 2. The Henselion, the back of the alveolus of either of the median incisors, the point used and defined by Prof. ITensel in his cranio- logical work. Posteriorly: 3. The Basiox, a point in the middle line of the hinder edge of the basioccipital margin of the foramen magnum. 4. The Condylion, the most posterior point of the articular surface of either condyle. A fifth measuring point to })e referred to below is the Pala- TioN, the most anti'rior point of the hinder edge of the bony palate, whether in the middle line or on either side of a median spine. Now using these words for. the purposes of definition, I would propose>, as shown in the diagram, the following names for the four measurements that may be taken between the points above defined : — 1. Basal length, tlie distance from Basion to Gnathion. 2. Basilar len(;th, the distance from Basion to Henselion. 3. Condylo-basal lenoth, the distance from Condylion to Gnathion. 4. Condylo-basilar length, the distance from Condylion to Henselion. In addition there may be: 5. Gre.\test len(;th, to be taken not further divergent from the middle line than either condylion. A long diagonal to a projecting bulla or paroccipital process would thus be barred . If however the words ' ' between uprights ' ' be added the measurement would be between two ver- tical planes pressed respectively against the anterior and posterior ends of the skull at right angles to its middle line. 6. Upper length, from tip of nasals to hinder edge of occipi- tal ridge in middle line. The difference between the words basal and basilar, which at first seemed trivial and indistinctive, is founded on the use of 13 194 Thomas — Nomenrlnture of Measurements. the Englisli word ])asal by the older writers, such as Flower and others, who used tlie lueasurenient from tlie gnatliion; wliile basilai' is an adaptation of tlie German of Hensel and liis school, who used the ^' ha.^ilar-lcinge'" from the henselion. These names again, coml)ined with condylo-, readily express the points which are used hy those who like to adopt tlie condylion as a posterior measuring point. But further, the association of the ending " al " with a meas- urement from the gnathion, and " ilar " with one from the henselion, if once defined and fixed, may l)e utilized in a second case of similar character. The length of tlie l)ony palate is a measurement given hy all careful descrihers, hut tlu.' anterior measuring point used is again either the gnathion or henselion, doubt as to which is l)eing used often nullifying the valu(^ of the measurement altogether.* To avoid this doubt T would suggest, exactly as in the other case, that the name of the measurement from the gnathion should end in " al " and tliat from the henselion in "ilar." We should then liave: Palatal LE\(/rn, the distance from gnathion to palation. Palatilar LEX(iTii, the distance from henselion to palation. The indeterminate " palate length " would then be dropped altogether. II. The Names of the Cheek-teeth of Mammals. Although the cheek-teeth of mammals, the molars and pre- molars, have l)een studied and written about ever since the birth of zoology, no uniform system of naming them has been evolved and there is the greatest divergence between the usage of differ- ent workers on the subject. In old days all were called molars or grinders; tluMi the premolars were distinguished from the true molars (although French zoologists, Winge in Denmark, and Ameghino in Argentina, continued to use a continuous notation for the two sets of teeth combined) and the usual habit among zoologists in general was to speak of them individually as " second premolar," " third molar," and soon. Even here, however, an important difference cropped up owing to Hensel * I may explain that in my own descriptions the palate of any given animal has al- ways been measnred from tlie same anterior point, gnathion or henselion, as the skull itself, this latter being indicated by the use of the words basal or basilar. 14 Thomas — Nomenclature of Measurements. 195 and his school in lu'rnuiny numbering the premolars from be- hind forwards, while naturalists of other nations counted from before backwards, as with the incisors and molars, a difference often productive of fatal confusion. Of late years, however, partly owing to an increasing concensus of opinion that the seven cheek-teeth of Placentals, four pre- molars and three molars, are serially and individually homolo- gous with the seven of Marsupials, formerly reckoned as three premolars and four molars, many naturalists have again begun to think that a continuous numeration might be the best one. But the difficulties in the way of its adoption are very great, largely owing to the a])sence of any convenient and suitable word in English less clumsy than " cheek-tooth," to express a tooth of the combined premolar and molar series. To speak of the " first cheek-tooth " or of the " predecessor to the fourth cheek- tooth " would l)e so retrogressive a step that I am sure no one would adopt it. But if instead of trying to find a word for the series combined with a numeral to show the position, we were to have a name for each tooth, we should get some- thing of the immense convenience we have all realized in having definite names for the canine and the carnassial teeth, the latter name being found of value in spite of the fact that the upper and lower carnassials are not homologous with each other. Such names might be made from the positions of the teeth if their meanings were not so obtrusive as to confuse the minds of per- sons who do not readily understand how a tooth should be called ' ' the second " or " secundus ' ' when it is actually the most an- terior of the series. Now it fortunately happens that while the Latin terms ' ' pri- mus, " " secundus, ' ' etc, , express the serial positions too clearly for the convenience of weak minds. Latinized Greek terms have just about the right amount of unfamiliarity which would enable them to be used as names without their serial origin being too much insisted on, IVIoreover, their construction is similar to the process we all use in making generic names, and so far as I know they have never been previously utilized in zoology. Then, after Latinizing the Greek ordinal terms -ptoru^^ etc. for the cheek-teeth of the upper jaw, the same modification as is already used in cusp nomenclature might be adopted for those of the mandible. 15 196 Thomas — Nomenclature of Measurements. We should thus have, counting from before backwards: UPPER JAW. LOWER JAW Cheek-tooth 1 Protus Protid 2 Deuterus Deuterid 3 Tritus Tritid 4 Tetartus Tetartid 5 Pemptus Pemptid 6 Hectus Hectid 7 Hebdonius Heljdoniid To avoid any doubt, I would expressly allocate these names to the permanent teeth of placentals, leaving the names of the marsupial teeth to be settled in accordance with their placental homologies. For the milk teeth a further modification would l;)e available by prefixing the syllable Pro- to the names of the respective permanent teeth. We could thus for example in the case of a third lower milk premolar call it tlie protritid, and so use one word instead of foui'. Of course I have no supposition that this system would ever be frequently or generally used, l)ut I am convinced that in many special cases, and particularly in such descriptions and cata- logues of isolated teeth as paleontologists often liave to give, it might result in considerable convenience and saving of space. 16 CRITERIA FOR VERTEBRATE SUBSPECIES, SPECIES AND GENERA: THE MAMMALS By E. Raymond Hall University of California, Berkeley, California Mr. Chairman, members of the American Society of Ichthyologists and Herpetologists, members of the American Society of Mammalogists, and guests: We had expected as a speaker at this time one of the senior mammalogists who now is unable to attend. I am glad to appear as a substitute because the subject under discussion is one in which I am especially interested. In these extemporaneous remarks I propose: (1) to indicate some steps which I think useful to take in classifying mam- malian .specimens as to subspecies; (2) to express my personal views as to criteria for subspecies, species, and genera of mammals; (3) to illustrate how some of these criteria for subspecies and species may be applied to closely related insular kinds of mammals; and (4) to suggest a way in which subspecies may disappear without becoming extinct. When I undertake to classify mammalian specimens as to subspecies or species, or when I present a series to a beginning student for classifica- tion, I like to observe the following steps: (a) select for initial, intensive study a large series, 30 or more individuals, from one restricted locality; (b) segregate these by sex; (c) arrange specimens of each sex from oldest to youngest; (d) divide these into age-groups and within a given group, of one sex, from one locality, of what is judged to be one species, measure the amount of so-called individual variation; (e) with this measurement as a "yardstick," compare individuals, and if possible series, comparable as to sex and age (and seasons where characteristics of the pelage are involved) from this and other localities. The differences found are usually properly designated as geographic variations and form the basis for recognition of subspecies, which in turn comprise one of the tools used by some students of geographic variation. As to criteria for the recognition of genera, species and subspecies of mammals, it seems to me that if crossbreeding occurs freely in nature where the geographic ranges of two kinds of mammals meet, the two kinds should be treated as subspecies of one species. If at this and all other places where the ranges of the two kinds meet or overlap, no cross- breeding occurs, then the two kinds are to be regarded as two distinct, full species. The concept of a species, therefore, is relatively clear-cut (141) 17 142 ANNALS NEW YORK ACADEMY OF SCIENCES and precise; the species is a definite entity. Furthermore, if a zoologist knows the morphological characteristics diagnostic of the species, he has no difficulty in identifying a particular individual as of one species or another. In identification of subspecies, difficulty is frequently en- countered, especially with individuals which originate in an area of intergradation. The category next higher than the species, namely, the genus, is less definite and more subjective as regards its limits than is the species. As the species is the definite, clear-cut starting point for defining subspecies, the species is likewise the starting point for consideration of genera. Degree of difference is the criterion for a genus. The genus lies about midway between the species and the family. Because the limits of the family, like those of the genus, are subjective, it follows that the criterion for recognition of genera, although precise enough at the lower point of beginning, the species, is elastic at the upper end — namely, at the level of the family. In summary, the criterion for subspecies is intergradation, that for species is lack of intergradation, and that for genera is degree of difference. These ideas agree in general with the ideas expressed by the previous speakers. One of the situations in which it is difficult, or impossible, to apply these criteria to conditions actually existing in nature is comprised in some insular populations. Frequently the populations on two islands near each other differ enough to warrant subspecific or possibly specific distinction. A means of deciding on specific versus subspecific status for these populations is to find on the adjacent mainland a continuously dis- tributed, related kind of mammal which there breaks up into subspecies. Ascertain the degree of difference between each pair of mainland sub- species which intergrade directly. If the maximum degree of difference between the insular kinds is greater than the difference between the two subspecies on the mainland, which intergrade directly, and greater than that between either insular kind and the related population on the nearby mainland, the two insular kinds may properly be treated as full species. If the maximum degree of difference between the insular kinds is no greater than, or less than, the difference found on the mainland between pairs of subspecies which intergrade directly, the insular kinds may properly be treated as subspecies of one species. In fine, the criterion is degree of difference with the limitation of geographic adjacency, rather than intergradation or lack of it. Now to my fourth point, namely the suggestion that many subspecies disappear without becoming extinct. Permit me first to observe that 18 HALL: CRITERIA FOR MAMMALS 143 although species and subspecies seem to have the same kinds of dis- tinguishing characters, which appear to be inherited by means of essen- tially the same kinds of mechanisms in the germ plasm, there are two noteworthy differences between species and subspecies. One already implied is that, in a species which is continually distributee! over a given area, its characters at the boundaries of its range are sharp, definite, and precise. Some of its characters comprised in size, shape and color, at any one place are either those of one species or instead unequivocally those of some other, whereas the characters of a subspecies, particularly at or near the place where two subspecies meet, more often than not are various combinations of those of the two subspecies and in many indi- vidual characters there is blending. Second, through a given epoch of geological time while a species is in existence, one or more of its subspecies may disappear and one or several new subspecies may be formed. Subspecies, therefore, on the average are shorter-lived than species. Now the disappearance of subspecies is to be expected on a priori grounds if we suppose that new subspecies are formed in every geological epoch. There is reason to believe that in the Pleistocene, the epoch of time immediately preceding the Recent, there were even more species of mammals than there are now. In each of several successively corre- sponding periods of Tertiary time before the Pleistocene, probably there were as many species as now. Probably too, these species then were about as productive of subspecies as species are now. Had even half of these subspecies persisted, either as subspecies unchanged or in con- siderable part by becoming full species, there would now be an array of species and subspecies many times as numerous as actually does exist. It is obvious therefore that many disappeared. In accounting for this adjustment of numbers of kinds of mammals, I have spoken of the disappearance of subspecies rather than of their ex- tinction because I can imagine how a species, say, the pocket gopher Thomomys townsendii, in the middle Pleistocene with three subspecies (geographic races) could have come down to the present by means of each of the three subspecies having gradually changed its characters into those of one of the three subspecies existing today in the area of northern Nevada that I have in mind. In this way, disappearance of subspecies living in the Pleistocene has been accomplished, without their having become extinct in the sense that the subspecies left no living descendants. Of course this has to be true for some of the subspecies of each successively preceding epoch if any animals at all persist, but what I wish to empha- size is the strong probability that many, perhaps more than 50 per cent, 19 144 ANNALS NEW YORK ACADEMY OF SCIENCES disappeared thus without actually becoming extinct, when, for example, two successive stages of the Pleistocene, south of the ice sheet, are con- sidered. In this regard it is pertinent to recall that each of three Pleistocene kinds of pocket gophers, Thomomys (probably species tal- poides) gidleyi, Thomomys (probably species townsendii) vetus, and Thomomys (probably species hottae) scudderi, from a short distance over the northern boundary of Nevada, differs from living representatives cor- responding to it (several subspecies of one species) in greater width la- bially of the individual cheek teeth of the lower jaw. Significant for the thesis being defended is the point that each and all of these Thomomys in the Pleistocene differed, at least as regards the shape of the teeth, in the same way from the three living species which I feel confident are their descendants. Let us suppose that three hypothetical subspecies of Thomomys townsendii in middle Pleistocene time each gradually changed into three different subspecies inhabiting about the same areas in upper Pleistocene time, and that these in turn were the ancestors of the three subspecies living in those same general areas today. A total of nine kinds is thus accounted for. At any one time there was geographical intergradation, which has reference to horizontal direction. Also there was intergrada- tion up through time, which has reference for present purposes to a vertical direction. If I had before me all the material necessary to sub- stantiate this or a similar case, I would be inclined to recognize nine sub- species of one species. This hypothetical case emphasizes the import- ance of intergradation, the criterion for subspecies. In review : I have mentioned some preliminary steps useful for a person to take when he aims to analyze variation in mammals and to establish species and subspecies thereon; intergradation is the criterion for sub- species and degree of difference is the criterion for genera; degree of difference with the limitation of geographic adjacency may be used as the criterion for insular populations (the classification of which is doubt- ful as between subspecies and species); and, finally, I have sought to stress the importance of intergradation as a criterion for subspecies by showing how subspecies may disappear without becommg extinct. 20 The Nature of Subspecies Boundaries in a Desert Rodent and its Implications for Subspecies Taxonomy WILLIAM Z. LIDIGKER, JR. IT SEEMS to me that the wide diversity of opinion which exists concerning the usefulness of trinomial nomenclature re- volves in large measure on the more basic issue of whether or not it is possible to recognize infraspecific categories which reflect genetic relationships. As recently pointed out by Sneath (1961), taxonomic categories which are not based on rela- tionships are thereby rankless and cannot logically be included in a taxonomic hierarchy. Thus if the subspecies cate- gory is used merely as an instrument for describing geographic variation in a few characters or as a device for cataloging geographic variants (as is done by many taxonomists), artificial classifications of convenience are characteristically pro- duced. Such convenience classifications usually contain rankless groups (the "false taxa" of Sneath) which cannot be allocated in the taxonomic hierarchy. This is simply because categories based on a few arbitrary characters are themselves arbitrary, and lead to the objection of Brown and Wilson (1954) and others that trinomials based on one group of char- acters need not bear any relation to those based on different traits. Many of the same philosophical difficulties apply to systems such as that recently proposed by Edwards (1954) and Pimentel (1959) in which the subspecies would become a measure of isolation, by restricting its use to completely isolated and "obviously dif- ferent" populations. If on the other hand studies of infra- specific populations are focused on dis- covering evolutionary diversity or degrees of relationship between the various popu- lations, I see no philosophical objection to the use of the trinomen. The question then reduces to one of the feasibility and/ or desirability of searching for such rela- tionships, and of deciding what level of dissimilarity if any justifies use of the formal trinomen. It is primarily these two subsidiary questions which are ex- amined in this paper, with the frank hope that the subspecies can be rescued from the rankless limbo of the morph, ecotype, and form. If this rescue operation should prove successful, attention can then be directed to other problems of greater bio- logical interest, such as whether or not determinations of genetic relationships within a species, which are based on phenotj^es, can serve as a basis for specu- lations on phylogeny. Obviously geogra- phic relations would have to be con- sidered at this level, but, assuming that such information is taken into account, it would be highly informative to contrast phenetic and phylogenetic subspecies classifications. In any case, analyses of infraspecific relationships would very likely provide valuable clues concerning the environmental forces which have in- fluenced the development of the existing evolutionary diversity. In a previous paper (1960) I attempted to determine the genetic relationships among populations within a species of kangaroo rat (Dipodomys merriami Mearns, 1890) by a careful analysis of 20 morphological features. I concluded at that time that at least in well-known ter- restrial species an attempt to recognize relative relationships within a species was at least possible. And, at the same time it 21 SUBSPECIES BOUNDARIES 161 was apparent that (besides the philosophi- cal objections already pointed out) the subspecies category by itself was com- pletely inadequate for describing the complex geographic variation occurring in that species. It is the raw data from this former investigation that I have used here to test further the reliability of those tentative conclusions. The search for relationships among populations of the same species implies a search for total genetic differentiation (or at least its phenotypic manifestations), and hence of lineages with partially in- dependent evolutionary origins such that they have some internal homogeneity and their own adaptive tendencies. To detect this kind of differentiation it seems impor- tant to analyze, among other things, the populations occurring at the boundaries between differentiating groups, just as in the analysis of species relationships it is the boundaries between them, or areas of sympatry, where the most significant in- formation on relationships is to be found. This is not to say that information con- cerning the regions of greatest divergence or adaptive peaks (in this case peaklets) of infraspecific populations is not impor- tant, but only that such data should not be the only source material for taxonomic judgments. Thus it is the intent of this paper to focus attention on the previously all but ignored subspecies boundaries, and to examine the nature of these areas in Dipodomys merriami as I had pre- viously and without the benefit of this analysis defined them (Lidicker, 1960). Because the determination of these intra- specific units was guided in this case by a desire to find populations of comparable evolutionary relationship, careful scrutiny of the intergrading zones between them and surrounding areas should be of par- ticular interest. Comparisons will also be made with levels of differentiation in areas in which no subspecies boundary was recognized, as well as with one re- gion in which species level differentiation was postulated to have been reached by an island isolate. The second and related purpose of the paper is to describe a method which helps to accomplish the first objective by meas- uring total differentiation, or lack of similarity, in many diverse characters, and hence is proposed as a criterion of relationship. But at the same time the technique does not require the hard working taxonomist to have either access to a digital computer or facility with matrix algebra. The Method Most quantitative techniques available to the systematist, which concern them- selves with determining relationships, and hence with similarities as well as dif- ferences, either involve the analysis of qualitative or discontinuous characters and thus are most useful at the species or genus level (e.g., Michener and Sokal, 1957; Lysenko and Sneath, 1959), or in- volve calculations sufficiently complex (e.g., Williams and Lance, 1958) that they are avoided by most practicing systema- tists. What seems to be needed is an ad- ditional technique which is sufficiently adaptable to handle continuously variable, as well as discontinuous, characters of diverse types (and so is useful in infra- specific studies), and which at the same time is sufficiently practical that it will be widely useful. To this end the following proposed method is dedicated. It is not intended as a substitute for discriminant function analysis (Fisher, 1936; Jolicoeur, 1959) and related methods which attempt to discriminate between previously con- ceptualized populations by using combina- tions of variables. An analysis of relationship should ideally compare relative similarities and not differences. However, since the num- ber of similarities between populations within a species is very large, it is much easier to measure their differences and consider that the reciprocal of the amount of difference represents a measure of similarity. Thus as the amount of differ- ence approaches zero, the reciprocal ap- 22 162 SYSTEMATIC ZOOLOGY preaches infinity. The problem then be- comes one of summing the amount of difference in many diverse characters. To do this we must be able to express the differentiation for each trait by a pure number (no units). Cain and Harrison (1958), for example, accomplished this by dividing the differences which they observed between means by the maxi- mum value recorded for each character. The resulting ratios, which they called "reduced values," express the observed differences in terms of a fraction of the maximum size of each character. Al- though this permits the comparison of diversity among traits of different abso- lute size, it does not take into account either the possibility that various char- acters may have different variabilities, or the statistical significance of the ob- served mean differences. Furthermore, maximum size would seem to be a statistic of dubious biological importance in con- tinuously varying characters. On the other hand, all of these important vari- ables, the variance of each trait, character magnitude, as well as a consideration of whether or not mean differences have a high probability of representing real dif- ferences, are taken into account by ex- pressing differentiation as a proportion between the observed differences between samples and the maximum amount of dif- ference expected on the basis of chance sampling variation. Only mean differ- ences greater than that amount which may be due to chance would then be con- sidered as real differences. For our pur- poses the maximum chance variation ex- pected in any comparison can be equated to the minimum difference required for statistical significance (at any given con- fidence level). This minimum significant difference (msd) can be calculated in a number of ways. One possibility is to determine the standard error of the mean for each character for each sample. Then in comparing two samples for this char- acter, 2 SEj +2SF,^_^ = msd. This provides a conservative estimate of msd with con- fidence limits usuallv well in excess of 95%. For large studies, however, these calculations would be extremely laborious, as well as perhaps overly conservative, and a short-cut is suggested. If we can assume that each quantitative character in a given species exhibits a characteristic variability throughout its range, then calculations would be tre- mendously reduced if we were able to de- termine the expected or pooled standard deviation (5,,) and standard error (Sp_) of samples for which say r2>20. Very small samples would have to be grouped with adjacent samples whenever possible, or if necessary either ignored or have separate msd-values calculated for them. Under these conditions 4sp_represents our best estimate of Tusd. Unfortunately con- fidence limits cannot be calculated for its reliability, although again it is generally a conservative estimate. The statistic Sp can be conveniently determined by averag- ing the weighted variances for several samples of adequate size (Hald, 1952: 395). Note that as the estimate of s^ im- proves it approaches the population stand- ard deviation (a), and hence is applicable to a wider range of sample sizes. Better estimates of Sp require knowledge of the total number of individuals in each of the populations sampled (see Cochran, 1959: 72), an obvious impossibility in this type of problem. In the examples given in this paper 45 p. was estimated by using the standard deviation of one large sample collected near the center of the species' range, and by assuming n = 20. This ex- pediency seemed justified because of the close similarity in values of s calculated for a given trait among several samples, and because of the likelihood that s ap- proaches a under these circumstances. Still another method of deriving the statistic msd, but one not used in this re- port, involves a more laborious, but sta- tistically more precise, procedure. The confidence limits for the difference be- tween any pair of means can be calcu- lated (see Dixon and Massey, 1957:128) whether or not we assume that we know 23 SUBSPECIES BOUNDARIES 163 the variance characteristic of each trait (s^) or use only the pertinent sample variances (sj and s|). For a large study, the calculations are very much reduced if one can estimate Sp (see above), and per- haps even use only samples in which n>20. If these simplifications are pos- sible, a pair of confidence limits can be computed which will be characteristic for each trait studied. In either case, one con- fidence limit gives us our msd, since mean differences greater than this can with a known probability be considered real. We need not be concerned with the possibility that the mean differences are even larger than those observed. Consider then only those characters in which the differences in the mean values (Xi— X2) for a given pair of locations (samples) are greater than the minimum significant differences. Now, divide these significant differences in mean values by the minimum significant difference char- acteristic for that trait (or for that pair of samples). This procedure gives us our pure number which can be designated as di, d.,, dn for successive characters, each representing a measure of differen- tiation in one character between one pair of samples. Having defined the amount of differentiation in each character in terms of a pure number, we can now add these to arrive at an esti- mate of total differentiation in the char- acters studied (2c?i). In interpreting this statistic in any real situation, however, it seems apparent that the distance between the two samples compared should be taken into account. Obviously an amount of total differentiation exhibited between two samples which are close together geo- graphically would be more significant than the same amount of differentiation between samples geographically distant. To compensate for this effect of distance, I have divided the total differentiation by the distance (in miles) between the two samples. The resulting figure, which I have called D or the Index of Differen- tiation,^ represents the proportion of sig- nificant change that occurs between the two locations per mile. Then the re- ciprocal of D easily gives us our measure of similarity between populations. D need now only be further divided by the total number of characters studied, including those of course in which no differentia- tion occurred, to arrive at the mean char- acter differentiation per mile {MCD/mi.). The importance of considering distance between samples will depend in large measure on the specific problem under in- vestigation. Obviously air-line distance between samples does not always accu- rately reflect the real magnitude of the distance or barriers between them. I feel that this is not a serious difficulty, how- ever, since we are concerned with the abruptness of differentiation between ad- jacent populations and not with barriers per se. Moreover, in some ways D acts as a measure of restriction on gene flow, be- cause, if distance is kept constant, D will tend to increase as gene flow is reduced. Another potential difficulty with the dis- tance calculation is that it carries the assumption that if the two localities be- ing compared were actually closer to- gether, the amount of total differentiation shown would be less. This is not always true because not only are there sometimes large areas which exhibit very little geo- graphic variation, but also there exist un- avoidable gaps in specimen collections. For these reasons I felt that in the pres- ent analysis of D. merriami it was neces- sary to consider both 2c?i and B in assess- ing differentiation. One further complication seems worth considering. This concerns variation in the direction of change between different characters. It seemed to me more signifi- cant if one or two characters were found to change significantly in a direction op- posite to that of the other characters, than if they all changed in the same di- 1 Note that this is in no way similar to the "differentiation index" of Kurten (1958) which compares growth gradients. 24 1G4 SYSTEMATIC ZOOLOGY rection. Thus for each such direction change, I arbitrarily added one half the (Z-value for that specific character to 2d. This also serves to oppose any tendency to give too much weight to characters which may not be entirely independent in their variation, or to those varying allo- metrically. Otherwise no allowance has been made for differentially weighting characters which might be considered to have greater phylogenetic importance. Presumably this could be readily done, if there were some sound basis for making such judgments. Sneath ( 1961 ) , however, points out some of the dangers inherent in attempts to do this, and argues for considering each character equally. Table 1 summarizes the calculations for 2.40(>20) >.35(>l5)or>40«20) >.25{>IO)or>.35«l5) >.20(>IO)or>.25(.l5{>IO)or>.20{.IO(>IO)or).l5 (.05(>5)or>.IO(.05«5) .00 (0) Fig. 1. Observed diflferentiation of Dipodomys merriami in southeastern Arizona and adjacent Mexico and New Mexico. Numbers on the lines connecting the various sample localities represent the calculated values for D and in parenthesis Sd,. The key to the intensity of stippling is based on these same statistics. The scale associated with each map rep- resents a distance of 25 miles. See also the text for a more complete explanation of the figures. 26 !6G SYSTEMATIC ZOOLOGY "^S %_oo'.oQ^' Las Vegas \ Aguongo Fallon X Fig. 2. Observed differentiation of D. merriann in a) northern Nevada, b) southern Nevada and adjacent Mojave desert of California, and c) small area in extreme southern California. For a more complete explanation of the figures and a key to stippling intensity, see the text and Figure 1. ranging from essentially none to that judged to be at the species level. Figure 1 shows the boundary region between D. merriami merriami and D. m. olivaceus (nomenclature based on Lidicker, 1960) in southeastern Arizona and adjacent Mexico and New Mexico. Figure 2 illus- trates areas in northern Nevada (a), southern Nevada and the adjacent eastern Mojave desert of California (b), and fi- nally a small area in southern California at the boundary of D. m. collinus and D. m. arenivagus (c). Figure 3 represents the southern tip of the Baja California peninsula (a), and southern Sonora where the boundary between D. vi. mer- riami and D. m. mayensis is found (b). The first of these (3a) is of particular in- terest as it shows the entire range of D. m. melanurus and the adjacent island populations of D. m. margaritae and the presumed allopatric species D. insularis. Notice that the key takes into account both D and 2 ( skull large and broad melanophrys. with straiirht or .. . . ~ • i i i i Ifrl tl ' n } I J nosive foramina barely reach plane 8 ig y CO - "^ skull smaller I of first molar longicaudus. and nar- <; L rower I incisive foramina reach second cusp I of til vex edge L rower | inci ' tirst molar torridus. Cranial measurements of the known forma of the genua Onychomys Basilar length of Hensel (from foramen magnum to incisor) Zygomatic breadth Greatest parietal breadth Interorbital constriction Length of nasals Incisor to post-palatal notch Foramen magnum to incisive foramina Foramen magnum to palate Length of upper molar series (on alveolte) Length of incisive foramina Length of mandible Height of coronoid process from angle Katios to basilar length: Zygomatio breadth Parietal breadth Nasals Molar series (on alveolae) Incisive foramina Foramen magnum to incisive foramen Foramen magnum to palate O. leucogaster, ; vfplanonhrvo Fort Buford, -S^^rte"' Dakota. Kanab, Utah. 4418$ 4419d" 5393 cT 5894 cT 22 22 22.3 21.6 15 15.2 15.4 1.^5 12.9 12.7 12.8 12.5 4.5 4.5 5.2 4.8 10.8 11.6 10.7 10.7 12 12 11.7 11.5 14.7 14.6 15 14.5 9.7 10 10.2 9.9 4.5 4.2 4.6 4.8 5 5.7 5 5 15.5 15.8 15.7 15.3 6.5 7.3 6.8 6.8 68,1 69 69 71.7 58.9 57.7 57.3 57 49 52.7 47.9 49.5 20.4 ISJ 20.6 22 22.7 25.9 22.4 23.1 66 66.3 67.3 67 44 45.4 45.7 45.8 Longipes, Concho County, Texas, 23.3 15.5 12.2 4.4 12.5 12.4 15.7 10.6 4.4 5.3 16 7.2 66.6 52 52.3 20 22.7 67.3 45.4 Basilar length of Hensel (from foramen magnnm to incisor) Zygomatic breadth Greatest parietal breadth Interorbital constriction Length of nasals Incisor to post-palatal notch Foramen magnum to incisive foramina Foramen magnum to palate Length of upper molar series (on alveolae) Length of incisive foramina Length of mandible Height of coronoid process from angle Batios to basilar length : Zygomatic breadth Parietal breadth Nasals Molar series (on alveolae) Incisive foramina Foramen magnum to incisive foramen ., Foramen magnum to palate. Longicaudus. St. George, Utah. Torridus, Grant County, N. Mex. 5895? 19.3 13 11.2 4.7 10 10.5 13.5 8.8 3.8 4.3 13.4 6.2 67.3 58 51.8 19.6 22.2 68.3 45.5 5896,/ 5897 cf 2839 cT 19.3 19.4 13 18.1 11.5 11.2 4.7 4.8 9.5 9.7 10.5 10.4 13.4 13.3 8.7 8.7 3.8 3.8 4,3 4 4 13.5 13.2 6.3 6.2 67.3 68 59.5 57.7 49.2 50 19.6 19.5 22.2 22.6 69.4 68.5 45 44.8 18.5 12.5 11.4 4.2 9.6 10 12.5 8.5 3.5 5 13.2 5.8 67.5 61.6 51.8 18.9 27 67.5 45.8 43 PLATE I. Figs. 1, 2, 3, 4, and 5, Onychomys leucogaater, $ young. (Skull No. 4422. ) Fort Baford, Dakota. 1. Skull from above, and left under jaw from outside (X 2). 2. Crowns of left upper molars from below (x 10). 3. Crowns of left lower molars from above ( X 10). 4. Crowns of right upper molfirs from the side ( X 10). 5. Crowns of right lower molars from the side (X 10). Figs. 6 and 7, Onychomys leucogaster, $ ad. (No. 5012). Valentine, Nebraska. 6. Crowns of left upper molars from below ( X 10). 7. Crowns of left lower molars from above (X 10). Figs. 8 and 9, Onychomys longicaudus, $ ad. (No. 5896). St. George, Utah. 8. Crowns of left upper molars from below (X 10). 9. Crowns of left lower molars from above ( X 10). 38 44 North American Fauna, No. 2. Plate I. 1-5. Onychomys leucogaster, d j'oung. 6,7. Onychomys leucogaster, $ adult. 8, 9. Onychomys longicaudus, d axlult. 45 Vol. 79, pp. 83-88 23 May 1966 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON DESCRIPTIONS OF NEW BATS {CHOERONISCUS AND RHINOPHYLLA) FROM COLOMBIA By Charles O. Handley, Jr. U. S. National Museum, Washington, D. C. An imperfectly known endemic mammalian fauna is found on the Pacific coast and Andean foothills of northwestern Ecuador and Colombia and northward into Panama, where it crosses to the Caribbean slope and continues into Costa Rica and Nicaragua and in some instances even into Mexico. The relatives of its endemic species are mostly South American, but some are Mexican. Species characteristic of this fauna, snch as Carollia castanea, Vampyressa nymphaea, Heteromys australis, Oryzomys bombycinus, and Hoplomys gymnurus, were among the mammals collected in the course of virological studies of the Rockefeller Foundation on the Pacific coast of Colombia in 1962 and 1963. In addition there were striking new species of Choeroniscus and Rhinophylla. 1 am grateful to Wilmot A. Thornton, Center for Zoonoses Research, University of Illinois, Urbana (formerly at Universidad del Valle, Cali, Colombia) for the opportunity to study the Colombian material here reported. Richard G. Van Gelder, American Museum of Natural History (AMNH); Philip Hershkovitz and J. C. Moore, Chicago Natural History Museum ( CNHM ) ; Bernardo Villa-R, Instituto de Biologia, Mexico ( IB ) ; Barbara Lawrence, Museum of Comparative Zoology, Harvard University (MCZ); J. Knox Jones, Jr., Museum of Natural History, University of Kansas (KU); William H. Burt, Museum of Zoology, University of Michigan (UMMZ); A. Musso, Sociedad de Ciencias Naturales La Salle (LS); and Juhani Ojasti, Universidad Central de Venezuela (UCV) kindly permitted me to study comparative material. Specimens in the U. S. National Museum are designated by the abbreviation (USNM). Studies which led to the following descriptions were supported in part by National Science Foundation Grant G-19415. All measurements are in millimeters. For definition of cranial mea- surements see Handley (1959: 98-99). Capitalized color terms are From Ridgway (1912). 11— Proc. Biol. Soc. Wash., Vol. 79, 1966 (83) 46 84 Proceedings of the Biological Society of Washington Choeroniscus There are few specimens of the poorly known glossophagine genus Choeroniscus in collections. The limits of variation in the genus are incompletely known (Sanborn, 1954), and until now its separation from Choeronycteris has been questionable. A specimen of a new species of Choeroniscus from the west coast of Colombia greatly extends knowledge of the genus and strengthens its stature as a genus distinct from Choero- nycteris. Choeroniscus periosus, new species Holotype: USNM no. 344918, adult female, alcoholic and skull, collected 1 February 1963, by Wilmot A. Thornton, at the Rio Raposo, near sea level, 27 km south of Buenaventura, Departamento de Valle, Colombia, original niunber 592. Etymology: Greek periosus, immense. Distribution: Known only from the type-locality. Description: Body size large (forearm 41.2; greatest length of skull 30.3). Dorsal mass effect coloration (after three month's submersion in formalin) rich blackish-brown; basal three-fourths orange-brown in dorsal hairs; underparts but slightly paler than dorsum. Vibrissae abundant and conspicuous on snout and chin. Ears, chin, noseleaf, hps, membranes, legs, feet, and fingers blackish. Lancet of noseleaf relatively narrow, with three notches on each side near tip, and with prominent vertical median ridge on anterior face. Membranous "tongue-channel" on chin im- usually well developed, protruding 1.5 mm forward and 2.0 mm up from lower lip; dorsal and anterior edges scalloped. Ear short, tip rounded, antitragus well defined; tragus spatulate, 3.8 mm long, with margins entire ( except for prominent posterior notch opposite anterior base), and with anterior edge and posterior basal lobe thickened. Inter- femoral membrane broad, naked. Hind legs naked. Calcar shorter than foot, not lobed. Rostrum longer than braincase; cranium little elevated from basi- cranial plane; profiles of rostrum and cranium evenly tapered, without sharp angle in between; no orbital ridges or processes; zygoma absent; lambdoidal crest low; sagittal crest absent; maxillary toothrows sub- parallel; palate relatively broad anteriorly and narrow posteriorly; posterolateral margin of palate not notched; postpalatal extension parallel-sided, tubular, reaching posterior to level of mandibular fossae; mesopterygoid fossa reduced to a straight-sided, V-shaped notch; hamular processes greatly inflated and approaching, but not quite touching, auditory bullae; basial pits prominent, separated by broad median ridge. 2 12 3 Dentition weak. Dental formula -,-,—, — — 30. Upper incisors small, vmicuspid; inner upper incisors (P) separated by a space three to four times the width of the teeth; larger, outer upper incisor (P) sep- arated by somewhat less than its own width from P and from canine. 47 New Bats from Colombia 85 Upper canine with small posterobasal cusp. Upper premolars very nar- row; median cusp, particularly of anterior premolar, very little higher than weU-defined anterior and posterior cusps. Upper molars with cusps greatly reduced; M^ and M- similar in size and shape, M^ slightly shorter and broader. Upper premolars widely spaced; molars closer together, but not touching. Lower premolars narrow, with well-defined, subequal anterior, posterior, and median cusps. Metaconid cusps of lower molars enlarged and protoconid cusps reduced; paraconid cusps in line with protoconids, not inflected. Anterior lo\\"er premolar close behind, but not touching, canine; spaces between premolars great, but spaces between P^ and Mi and between other molars, much less. Measurements (All external dimensions taken from specimen in alco- hol): Total length 62, tail vertebrae 10, hind foot 12, ear from notch 15, forearm 41.2, tibia 13.3, calcar 7.9. Greatest length of skull 30.3. zygomatic breadth 11.0, postorbital breadth 4.7, braincase breadth 9.9, braincase depth 7.4, maxillary tooth row length 10.8, postpalatal length 7.0, palatal breadth at \P 5.2, pal- atal breadth at canines 4.6. Comparisons: C. periosus can be distinguished from all other species of Choeroniscus by its longer (longer than braincase), more robust rostrum; more inflated hamular process; and larger size {e.g., forearm 41.2 vs. 32.4-36.9; greatest length of skull 30.3 vs. 19.3-24.4; maxillary tooth row 10.8 vs. 6.5-9.2). It is allied with the Amazonian species C. minor, C. intermedius, and C. inca, and distinguished from the Central American and northern South American C. godmani, in having the posterolateral margin of the palate unnotched and the craniiun not so markedly elevated from the basicranial plane. Remarks: With the addition of C. periosus, the genus Choeroniscus includes five nominal species. C. periosus is much the largest species; C. inca Thomas, C. intermedius Allen and Chapman, and C. minor Peters are intermediate in size; and C. godmani Thomas is smallest. Choeroniscus is the most specialized of a group of nominal glossopha- gine genera which may be characterized briefly as follows: Teeth nearly normal pterygoids normal Lichonycteris Teeth shghtly reduced pterygoids? Scleronycteris Teeth reduced; PM high pterygoids shghtly Hylonycteris specialized Teeth reduced; PM high pterygoids specialized Choeronycteris Teeth greatly reduced; pterygoids greatly speciahzed Choeroniscus PM low Lichonycteris has 26 teeth and the other genera have 30. As here understood, the genus Choeronycteris includes Musonycteris harrisoni Schaldach and McLaughlin, which is distinguished from Choeronycteris mexicana Tschudi principally by its strikingly elongated rostrum and associated modifications in proportions. The disparity in 48 86 Proceedings of tJie Biological Society of Washington rostral proportions is much greater, however, between Choeroniscus godmani and Choeroniscus periosus than between Choeronycteris mexi- cana and Choeronycteris harrisoni. Thus, to distinguish C. harrisoni as representative of a separate genus tends to obscure relationships in this segment of the Glossophaginae. Muson-ycteris should be regarded as a synonym of Choeronycteris. Specimens examined: Choeroniscus godmani. COLOMBIA: Meta: Restrepo, 1 (MCZ). COSTA RICA: Vicinity of San Jose, 3000 ft, 5 (AMNH). HONDURAS: Cantoral, 1 (AMNH); La Flor Archaga, 2 (AMNH). MEXICO: Chiapas: Pijijiapan, 50 m, 1 (UMMZ); Guer- rero: 1 mi. SE San Andres de la Cruz, 700 m, 1 (UMMZ); Oaxaca: 16 km ENE Piedra Blanca, 1 (IB); Sinaloa: San Ignacio, 700 ft, 1 (KU). NICARAGUA: El Realejo, 1 (KU), 2 (USNM). VENEZUELA: Boli\'.\r: 38 km S El Dorado, 1 (UCV); Distrtto Federal: Caracas (Santa Monica), 900 m, 1 (LS); Chichiriviche, 1 (UCV). Choeroniscus inca. BRITISH GUIANA: Kamakusa, 1 (AMNH); Kartabo, 1 (AMNH). ECUADOR: Los Pozos, 2 (AMNH). VENEZUELA: Bolivar: Chi- manta-tepui, 1300 ft, 9 (CNHM). Choeroniscus intermedins: TRINI- DAD: Irois Forest, 1 (AMNH); Maracas, 1 (AMNH), Princesto\\Ti, 1 (holotype of C. intermedius, AMNH); Sangre Grande, 1 (AMNH). Choeroniscus minor. BR.\ZIL: P.ara: Belem, 3 (USNM). PERU: Pasco, San Juan, 900 ft, 1 (USNM); Puerto Melendez, above Maranon, 1 (AMNH). Choeroniscus periosus. COLOMBIA: Valle: Rio Raposo, 1 (holotype of C. periosus, USNM). Also, numerous specimens of Lichonycteris, Hylonycteris, and Choeronycteris (including C. harrisoni). Rhisophylla The carolliinine genus Rhinophylla has until now been kno\Mi only from the basin of the Rio Amazonas and the lowlands of northeastern South America (Husson, 1962: 152-153). The sole representative of the genus, R. pumilio Peters, has been regarded as closely related to, but more speciahzed than, the species of the abundant and widespread genus Carollia (Miller, 1907: 147). It is thus rather surprising to find in the collection of W. A. Thornton from the west coast of Colombia a number of specimens of a striking new species of Rhinophylla that is even more strongly differentiated from Carollia than is R. pumilio. Rhinophylla alethina, new species Holotype: USNM no. 324988, adult male, skin and skull, collected 13 July 1962, by Wilmot A. Thornton, at the Rio Raposo, near sea level, 27 km south of Buenaventura, Departamento de \'alle, Colombia, original number 172. Etymology: Greek, alethinos, genuine. Distribution: Known only from the t>3)e-locality. Description: Size large for genus (forearm 34.9-37.2 mm). Col- oration blackish, darkest anteriorly, paler posteriorly. In holotype, head and nape black, shading to Fuscous-Black on rump; underparts varying 49 New Bats from Colombia 87 from black on chin to Fuscous-Black on chest and to Natal Brown on abdomen. Another specimen (Univ. del Valle 220) slightly paler: Fus- cous-Black anteriorly and Natal Brown posteriorly on dorsum, and correspondingly paler on underparts. Hairs of dorsum and abdomen sharply tricolor: at mid-dorsum Slate-Black basally, with broad Benzo Brown median band; on sides, neck, and shoulders median band pales almost to Ecru-Drab and shows through to surface rather prominently. Noseleaf, lips, ears, tragis, fingers, forearms, legs, feet, and all mem- branes blackish. Fur soft, woolly; legs, feet, interfemoral membranes, and basal two-thirds of forearm hairy; interfemoral membrane fringed. Inter- femoral membrane narrow ( about 5 mm at base ) ; calcar short ( less than length of metatarsals); tibia and forearm stout; pinna with anterior margin convex, posterior margin concave, tip blunt, antitragus triangular; tragus usually blunt, with upper posterior margin entire or notched; lancet of noseleaf longer than broad, upper margins slightly concave; horseshoe of noseleaf with median half of base bound to lip; chin orna- ment composed of four parts — a central triangular element ( apex down ) , a pair of narrow, elongated lateral elements converging ventrally but not meeting (their outer margin more or less scalloped), and a small, circular median ventral element. Skull like that of Rhinophylla pumilio but rostrum shghtly heavier ( broader and deeper anteriorly ) , and a distinct low sagittal crest present. Dentition, with the exception of inner incisors, extremely weak and 2 12 3 reduced; formula -=32. Inner upper incisor (P) large, adz-shaped, with cutting edge entire; outer upper incisor (P) small, featureless. Canine simple, without cingulum or subsidiary cusps. Ante- rior upper premolar (P^) small and featureless; posterior upper premolar ( P* ) almost rectangular, longer than broad, with large median cusp and tiny posterior cusp. NP short and M" shorter, almost triangular in occlusal shape, each with a single prominent internal cusp (the metacone); protocone obliterated; paracone barely indicated in M\ obliterated in M^; parastyle and metastyle, particularly the latter, low and weakly devel- oped; M^ reduced to a tiny featureless spicule. Inner lower incisors (Ii) large, trilobed (occasionally bilobed); I2 small, unicuspid. Canine simple, without accessory cusps. Premolars simple, unicuspid; anterior premolar wider than any succeeding tooth. Molars very narrow, tricuspid; anterior and posterior cusps low on Mi and M2 and more or less obliterated on Ms. Measurements (Extremes in parentheses, preceded by means and followed by number of individuals ( only adults included ) . Measurements of the total length, ear, and weight were made by the collector in the field. All other measurements were made by me in the laboratory. ) : Total length $ 55, 58; hind foot $11 (11-11) 4, $ 11 (10-11) 6; ear from notch S 15, 16; forearm $ 36.4 (35.5-37.2) 4, $ 35.7 (34.9-36.6) 4; tibia 9 12.3 (11.2-12.9) 4, $ 12.0 (11.5-12.5) 4; calcar 2 3.1 (3.0- 3.5) 4, $ 3.4 (3.3-3.5) 5. Weight $ 12 gm, 16 gm. 50 88 Proceedings of the Biological Society of Washington Cranial measurements of male holotype: Greatest length of skull 19.5, zygomatic breadth 10.7+ ; postorbital breadth 5.3, braincase breadth 8.9, braincase depth 7.5, maxillary tooth row length 4.9, postpalatal length 7.2, palatal breadth at M" 6.4, palatal breadth at canines 5.1. Comparisons: Specimens of R. alethina are slightly larger than speci- mens of R. pumilio from the valley of the Rio Amazonas; have the inter- femoral membrane narrower; calcar shorter; hind legs stouter; legs, feet, and interfemoral membrane ( including posterior margin ) more hairy; fur more woolly in texture; and coloration, including that of lips, ears, and membranes, darker, more blackish. As noted in the description, the skulls of the two species are very similar. However, except for the inner incisors, the teeth of R. alethina are smaller and weaker, and the tooth rows are shorter than in R. pumilio. R. alethina has cutting edges of V and I2 entire rather than notched; P* shorter; cusps of upper molars more reduced; and I2, P\ M^, and lower molars notably smaller. Aside from its relative R. pumilio, R. alethina is likely to be confused only with the Glossophaginae and with Carollia castanea. Its non-exten- sible tongue and lack of rostral elongation are sufficient to distinguish it from the Glossophaginae. From Carollia castanea it can be distin- guished easily by its blacker coloration, narrow, fringed interfemoral membrane, hairy legs, simple chin ornament, and smaller, simplified teeth. In most of these characteristics R. alethina differs more from the species of Carollia than R. pumilio does. Specimens examined: Rhinophxjlla alethina. COLOMBIA: Valle Rio Raposo, 11 (including the holotype, USNM), 1 (Univ. del Valle). Rhinophylla pumilio. BRAZIL: ParA: Belem, 52 (USNM). ECUADOR: Boca de Rio Curaray, 2 (USNM). PERU: Pasco: San Juan, 900 ft, 4 (USNM). Literature Cited Handley, C. O., Jr. 1959. A revision of American bats of the genera Euderma and Plecotus. Proc. U.S. Nat. Mus., 110: 95-246, 27 figs. HussoN, A. M. 1962. The bats of Suriname. Zoologische Vemhande- lingen, Rijksmus. Nat. Hist., Leiden, 58: 1-282, 30 pis., 39 figs. Miller, G. S., Jr. 1907. The families and genera of bats. U.S. Nat. Mus. Bull. 57, xvii + 282 pp., 14 pis., 49 figs. RiDGWAY, R. 1912. Color standards and color nomenclature, iv + 44 pp., 53 pis. Sanborn, C. C. 1954. Bats from Chimanta-tepui, Venezuela, with re- marks on Choeroniscus. Fieldiana- Zoology, 34: 289-293. 51 BENSON — STATUS OF REITHKODONTOMYS MONTANUS 139 THE STATUS OF REITHKODONTOMYS MONTANUS (BAIRD) By Seth B. Benson The status and relationships of Reithrodontomys montanus have been uncer- tain ever since this harvest mouse was named and described by Baird in 1855. Study of the type specimen, and of specimens collected in the type locality of R. montanus, has revealed that all the specimens, except the type itself, are examples of the species Reithrodontomys megalotis (Baird). Confusion has arisen because these specimens have been mistakenly referred to R. montanus. The nomenclatural history is as follows. Baird (1855, p. 355) described Reithrodon montanus on the basis of a single specimen collected by a Mr. Kreutzfeldt [ = J. Creutzfeldt, botanist of Gunnison's expedition] at "Rocky Mountains, Lat. 38°." Later, Baird (1857, p. 450) gave the locality as "Rocky Mountains, 39°." Coues (1874, p. 186) listed Ochetodon montanus as a questionable species. This he also did later (1877, p. 130), stating "The single specimen is too imperfect to permit of final characterization, or to enable us to come to any positive conclusion; but if the size and coloration it presents are really permanent, we should judge it entitled to recognition as a vahd species. At present, however, we regard it with suspicion and are unwilhng to endorse its validity." This remained the status of the name until Allen (1893, p. 80), after exam- ining the type specimen, stated "I have therefore no hesitation in recognizing Reithrodontomys montanus (Baird) as a well-marked, valid species, which will probably be found to range from the eastern base of the Rocky Mountains eastward to middle Kansas." WTien Allen (1895) revised the harvest mice the type of montanus was still unique. In his treatment of the species (pp. 123-125) he determined the type locality to be the upper part of the San Luis Valley in Colorado. He stated that "Until this region has been thoroughly explored for 'topotypes' of R. montanus, it would be obviously improper to reject this species as unidentifi- able or to give the name precedence over R. megalotis for the form here recog- nized under that name." At this time the species currently recognized as albescens was not known, although Allen actually had specimens which he confused with the form now known as R. megalotis dychei (see Howell, 1914, p. 31). Subsequently Cary (1903, p. 53) described Reithrodontomys albescens from Nebraska, stating that the species required "no close comparison with any described Reithro- dontomys." Bailey (1905, p. 106) described Reithrodontomys griseus from Texas, and remarked that it probably graded into albescens. In 1907 Cary visited Medano Springs Ranch in search of topotypes of R. montanus. He collected twenty specimens, most of them immature, which he identified as montanus. Cary (1911, pp. 108-110), following a manuscript of A. H. Howell, regarded R. montanus as a species related to R. albescens and R. griseus. He placed albescens as a subspecies of montanus. &2 140 JOURNAL OF MAMMALOGY When Howell (1914) revised the harvest mice, the specimens from the type locality of montanus consisted of the type specimen and the specimens col- lected by Gary at Medano Springs Ranch. In this revision Howell altered his earlier opinions concerning the relationships of montanus. He wrote (p. 26) "The species, although combining in a remarkable degree the char- acters of the megalotis and albescens groups, seems not to be directly connected with either of them. It is perhaps best placed in the megalotis group, but seems not to intergrade with any member of it." He pointed out that the relationships of the species were yet not clear, since the type specimen did not agree with any of the "topotypes" collected by Gary, but instead resembled specimens of R. a. griseus from Texas. Because the color of the "topotypes" agreed with the original description of montanus, he decided to "consider the tj-pe skull aberrant, and to continue to use the name for the form represented Fig. 1. Drawings Made from Photographs of Skulls of Harvest Mice. Natural Size a. Reithrodontomys megalotis subsp., no. 61120, Mus. Vert. Zool., from Medano Ranch, 15 miles northeast of Mosca, Alamosa County, Colorado. b. Reithrodontomys montanus montanus, type specimen, no. 1036/441, U. S. Nat. Mus., from upper end of San Luis Valley, Colorado. c. Reithrodontomys montanus griseus, no. 58737, Mus. Vert. Zool., from 3 miles north of Socorro, Socorro County, New Mexico. by the modern series." It might be mentioned here that no specimens of the albescens group have as yet been taken near the type locality of montanus. In 1933, Miss Annie M. Alexander and Miss Louise Kellogg collected harvest mice from several localities in Golorado and New Mexico including the type localities of R. megalotis aztecus and R. montanus. Two of the three specimens from the Medano Ranch, 15 miles northeast of jMosca, Alamosa Gounty, Golorado, were adults similar to adult topotypes of aztecus. The third, a young individual, was so much smaller that at first I judged it was of a different species. I suspected then that Howell's treatment of montanus was the result of confusing two distinct species, one a small form like albescens, the other a larger one hke megalotis. The occurrence of two species at this locality seemed possible, since albescens occurs with dychei in Nebraska, griseus is known to occur with dychei in Kansas, and in 1933 I collected mega- lotis and griseus together in the bottom-land of the Rio Grande, three miles 53 BENSON — STATUS OF REITHRODONTOMYS MONTANUS 141 north of Socorro, New Mexico, which is in the same drainage system as San Luis Valley. At my request the Bureau of Biological Survey loaned me 16 of the speci- mens collected by Gary at Medano Ranch. Only one of these was fully adult (the one whose skull is figured as montanus in Howell's revision) . Among the younger specimens were some which matched the smallest of the three specimens collected by Miss Alexander and Miss Kellogg, and the rest formed a series approaching the largest specimens. The adult specimen collected by Gary is smaller than the other two adults, yet is similar to them in most characters. It was obvious that all belonged to a single species. I concluded that all the Medano Ranch specimens I had examined were of the species currently known as megalotis. It was also obvious that if the type of montanus were conspecific with the other San Luis Valley specimens, megalotis would become a synonym of montanus, since montanus has priority. Through the courtesy of Dr. Remington Kellogg and others in charge of the collection of mammals in the United States National Museum, I was granted the loan of the type specimens of R. megalotis and R. montanus. After study- ing these specimens I reached the following conclusions: (1) The Medano Ranch specimens are conspecific with the type of megalotis; (2) the type specimen of montanus is specifically distinct from megalotis, and is conspecific with albescens and griseus. Some characters in which the type of montanus and specimens of griseus (MVZ no. 41192, from Hemphill Go., Texas; no. 56220, from 44 miles northwest of Roswell, N. M.; and no. 58737, from 3 miles north of Socorro, N. M.) differ from megalotis are: smaller size; shorter, more depressed rostrum; narrower interorbital space; relatively shorter brain case. As a result, megalotis is not a synonym of montanus, and montanus becomes the specific name for the species currently known as albescens. Until addi- tional specimens of montanus from San Luis Valley are available to allow a more thorough appraisal of its characters, it seems best to regard albescens and griseus as valid races of montanus, although it is quite likely that griseus may become a synonym of montanus. The three races here recognized are : Reithrodontomys montanus montanus (Baird) Reithrodontomys montanus albescens Gary Reithrodontomys montanus griseus Bailey. It may be well to remark here that all the available information indicates that the species R. montanus is rarely abundant and that it prefers more arid, sandier ground than does its relative R. megalotis, although both species may be found together. The racial identity of the San Luis Valley megalotis has also presented some problems. At first I referred them to the race aztecus because some of them fell within the range of variation present in specimens from within the distributional area assigned to aztecus in Howell's revision. In addition, there was so much variation in size in the few adults available to me that I 54 142 JOURNAL OF MAMMALOGY felt it was possible they did not truly represent the population, and so could not serve as a satisfactory basis for the description of a new race. However, Mr. Howell, who has restudied the problem with the aid of a greater amount of material than was available to me, has concluded that the San Luis Valley megalotis represent an unnamed race. He will describe this race in another article. LITERATURE CITED Allen, J. A. 1893. List of mammals collected by Mr. Charles P. Rowley in the San Juan Region of Colorado, New Mexico and Utah, with descriptions of new spe- cies. Bull. Amer. Mus. Nat. Hist., vol. 5, pp. 69-84. April 28, 1893. 1895. On the species of the genus Reilhrodontomys. Bull. Amer. Mus. Nat. Hist., vol. 7, pp. 107-143. May 21, 1895. Bailey, V. 1905. Biological Survey of Texas. U. S. Dept. Agric, North Amer. Faima no. 25, 222 pp., 16 pis., 24 figs. October 24, 1905. Baird, S. F. 1855. Characteristics of some new species of North American Mammalia, collected chiefly in connection with U. S. Surveys of a Railroad route to the Pacific. Proc. Acad. Nat. Sci. Philadelphia, vol. 7, April, 1855, pp. 333-336. 1857. General report upon the mammals of the several Pacific Railroad routes. U. S. Pac. R. R. Expl. and Surv., vol. 8, pt. 1, xxxiv + 764 pp., 60 pis. Gary, M. 1903. A new Reithrodontomys from western Nebraska. Proc. Biol. Soc. Washington, vol. 16, pp. 53-54. May 6, 1903. 1911. A biological survey of Colorado. U. S. Dept. Agric, North Amer. Fauna no. 33, 256 pp., 12 pis., 39 figs. August 17, 1911. CouES, E. 1874. Synopsis of the Muridae of North America. Proc. Acad. Nat. Sci. Philadelphia, 1874, pp. 173-196. 1877. No. I. — Muridae, in Coues and Allen, Monog. North Amer. Rodentia (= U. S. Geol. Surv. Terr. [Hayden], vol. 9), pp. 481-542. Howell, A. H. 1914. Revision of the American harvest mice. U. S. Dept. Agric, North Amer. Fauna no. 36, 97 pp., 7 pis., 6 figs. June 5, 1914. Museum of Vertebrate Zoology, University of California, Berkeley, California. 55 A REVISION OF THE WOOD RAT NEOTOMA STEPHENSl By Donald F. Hoffmeister and Luis de la Torre Within the last few years, some field guides to mammals of the United States have appeared which do not include the wood rat Neotoma stephensi. The imphcation is that Neotoma stephensi is not a valid species. Some spe- cific characters of Neotoma stephensi were enumerated in the original descrip- tion by Goldman in 1905, and its specific distinctness (from Neotoma lepida) was reaffirmed by Goldman in 1932. Because of the previous confusion of Neotoma stephensi with Neotoma lepida, and even the subsequent confusion of the two species following Goldman's review (Jour. Mamm., 13: 59-67, 1932), it is somewhat understandable that some authors might question the validity of stephensi. Even the recently pubhshed work by Hall and Kelson (The Mammals of North America, 2: 690, 1959) fails to show the coexistence of N. lepida and N. stephensi in parts of their range, a fact which lends addi- tional evidence for the distinctness of the two species. With more adequate material, and using Goldman's work as a starting point, we have attempted to define more clearly the species N. stephensi. Many persons have made available, for study, specimens of Neotoma stephensi in their collections. We should especially like to thank Richard G. Van Gelder, Stanley P. Young, Viola Schantz, Laurence Huey, Seth B. Benson, William Z. Lidicker, and William H. Burt. Specimens from the following collections have been examined (the abbreviations in parentheses are used under Specimens examined): American Museum of Natiural History (AM); United States Biological Survey (BS); Grand Canyon National Park Museum (GC); San Diego Society of Natural History (SD); University of California, Museum of Vertebrate Zoology (UC); University of Illinois, Museum of Natural History (UI); University of Michigan, Museum of Zoology (UM). Capitalized color terms are from Ridgway ( Color Standards and Color Nomen- clature, 1912). All measurements are in millimeters. Under Additional records we have included localities from which the species is known but from which we have not examined specimens. The Graduate College of the University of Illinois has financially aided us in this study. Illustrations were prepared by Alice A. Boatright and Harry C. Henriksen of the University of Illinois. GENERAL CHARACTERS OF THE SPECIES Obvious characters of Neotoma stephensi, which in part are diagnostic, are the presence of a semi-bushy tail ( bushier than in all species except N. cinerea ) in a medium- to small-sized Neotoma; dusky coloration extending down the top of the foot one-fourth to one-third the distance below the ankle; skull resembling IV. lepida. A closer study indicates the following characters are diagnostic: Baculum. — Exclamation-mark-shaped or wedge-shaped (Fig. 1); small, being one-fifth or less than the length of the baculum in N. lepida; smaller than in 56 Nov., 1960 HOFFMEISTER AND DE LA TORRE— NEOTOMA 477 any species of Neotoma found in the United States; similar in size and shape to that of Neotoma phenax, except not indented along sides and thus not "violin-shaped." When everted from the prepuce, the distinctively small size of the baculum in N. stephensi is clearly noticeable, and in contrast to the condition in Neotoma lepida, N. albigula, and N. mexicana. In one specimen of N. s. relicta, the baculum is identical to that in N. s. stephensi. Skull. — Teeth: The pattern of M^ together with features of M^ serves to distinguish N. stephensi in most cases. In M^, the antero-medial fold is usually absent or, if present, is shallow, approaching albigula, never deeply re-entrant as in mexicana. In M', the postero-labial fold (Fig. 1) is directed postero- mediaUy and "terminates" posterior to the lingual fold. In N. lepida, the postero-labial fold (Fig. 1) is directed medially and less posteriorly, nearly meeting the lingual fold. In M^ of stephensi, the second loph is usually long and narrow, extending diagonally across the tooth (Fig. 1). In this regard, it is similar to N. mexicana and N. albigula, but differs from N. lepida in which the second loph is usually broad labially and nearly at right angles to the long axis of the toothrow (Fig. 2). In N. stephensi, the lingual fold of M3 is as deep as or deeper than the labial fold (Fig. 2), whereas in N. lepida and N. albigula, the lingual fold is shorter than the labial fold. In N. mexicana, both conditions seem to occur. Rostrum: The rostrum is narrower than in any other species within the range of N. stephensi ( Fig. 3 ) . Generally, the nasals in stephensi are truncate posteriorly, rarely sharply pointed as is common in lepida and albigula (Fig. 3). The posterior extensions of the premaxillaries seldom expand pos- teriorly as in albigula, but are more as in mexicana except that they are longer. Interorbital, region: The region between the orbits is broader and less depressed than in either lepida, albigula, or mexicana. In some instances, it is difficult to distinguish between stephensi and lepida on the basis of this character. The supraorbital ridges in stephensi tend to remain lateral as they continue forward toward the rostrum, whereas in the other species these ridges approach the midUne, ending almost in line with the posterior exten- sions of the premaxillaries (Fig. 3). second loph ^ » . ^ . stephensi postero-labial fold stephensi iepida lepida Fig. 1. — Third upper molar (x6.5) and baculum (x2.4) of Neotoma stephensi and Neotoma lepida. 57 478 JOURNAL OF MAMMALOGY Vol. 41, No. 4 Of perhaps lesser diagnostic significance when compared with N. lepida are the following skull characters: upper incisors strongly recurved; tubercle on outer face of mandible at base of lower incisor hardly noticeable, whereas more nearly knoblike in lepida. External features. — The bushiness of the tail, particularly the terminal third, approaches that of Neotoma cinerea. The tail is far more bushy than in any other species within the range of stephensi, and probably exceeds that of any wood rat except cinerea. In young stephensi, only a few weeks old, the tail already is bushy, and specific recognition is possible on this character. On the hind foot, a wedge of dusky-colored hair may extend onto the dorsal surface as much as one-third the distance to the base of the toes. In most specimens of N. lepida, particularly in those races which are not melanistic, the dorsal surface of the foot is whitish, with the dusky color stopping at the tarsal region. The underparts are suffused with an ochraceous or buffy wash in nearly all specimens. In N. stephensi stephensi the throat region also shows this same wash, but the region between the forelegs is whitish. In JV. stephensi relicta, the wash does not extend onto the throat and thus it is usually white. The coloration of the underparts in N. s. stephensi is similar to that in most speci- mens of the dark races of IV. lepida, such as monstrabilis and harteri, and is thus noticeably different than in the light-colored races of the latter species. Where N. lepida occurs within the range of N. stephensi, nearly all specimens of lepida have light-colored underparts and thus this character aids in dis- tinguishing the two. upper molars lower molars £mi stephensi lepida albigula stephensi lepida Fig. 2. — Occlusal view of right upper molars of Neotoma stephensi, N. lepida, N. albigula, and left lower molars of N. stephensi and N. lepida ( x6.5). 58 Not;., 1960 HOFFMEISTER AND DE LA TORRE— NEOTOMA 479 Stephens' tepida albigufa mexicona Fig. 3. — Dorsal view of anterior part of skull of Neotoma stephensi, N. lepida, N. albigula, and N. mexicana ( X 1 ) . VABIATION Age. — In N. stephensi, we can recognize four groups which include young animals, old animals, and two categories between these extremes. These groups are based on the degree of eruption and wear of the upper molar teeth. We attempted to corroborate this grouping by the use of other characters, such as closure of sutures, size and proportion of various parts of the skull, and molt, but these proved to be of httle help. Further discussion of the younger groups is given under Growth and Reproduction. Group 1. Immatures; M^ not erupted or in the process of erupting and occlusal surface of posterior loph isolated from that of more anterior lophs. Group 2. Young adults; females, at least, sexually mature. Folds of M^, as seen laterally, continuing down below alveolus; M^ fully erupted and oc- clusal pattern usually complete with posterior loph rarely isolated. Group 3. Adults; folds of M^ not extending to alveolus, but to or below an arbitrary midpoint between alveolus and occlusal surface. Group 4. Old adults; folds of M^ very short and not reaching midpoint between occlusal surface and alveolus. Folds may be entirely absent. Sexual. — Although there is a size difference between males and females, it is less marked in certain age groups. From our sample, we find that in Groups 1 and 2, the sexual difference is slight. In Group 3, males usually average larger than females in most measurements. For example, in specimens from the Hualpai Mountains (65 5, 109 9 ), the Grand Canyon (45 5, 49 2 ), and Wupatki (455, 499), the males are larger than the females by per- centages ranging from 1 to 13 in most measurements. In all three localities, however, the females average larger than the males in length of maxiUary toothrow as follows: Hualpai Mountains, 2.3; Grand Canyon, 1.2; Wupatki, 1.2 per cent. The larger average size of the toothrow in females is also indi- cated in specimens from other localities. Group 4 is represented by too few individuals to determine sexual differences. Aside from the quantitative dif- ferences pointed out, the skull of adult females is characteristically short and broad, in contrast to the long and narrower skull of the males. 59 480 JOURNAL OF MAMMALOGY Vol. 41, No. 4 No sexual difference in color was apparent. Color. — The color of an "average" specimen consists of a light buff wash extending from the chest to the inguinal region. The chest is white, with the throat a buffy color with the hair plumbeous basally and buffy-tipped. The white of the chest may extend anteriorly to the chin as a median narrow stripe. The inguinal area is whitish. The coloration of the underparts, including the buffy wash, throat, and chest color, in any one population, may be highly variable. For example, near Montezuma Well, Yavapai County, Arizona, one specimen has 98 per cent of the underparts heavily washed with a dark buff, one has little or no buff, and one has no white on the chest and throat. Still other specimens from here show varying stages of intermediacy. In other populations, indi- viduals can be found that have the throat entirely white, with the basal portions of the hairs not plumbeous or gray-colored. The whitish throat is "typical" of another, unrelated species — albigula. Such occurrences of "white- throatedness" are widely scattered throughout the range of the species, being found in populations from the Hualpai Mountains in the northwest to the Burro Mountains in the southeast. The extreme of this white-throatedness is found in N. stephensi relicta. In this race, nearly all individuals, and probably every adult, have white throats. Within the race stephensi, certain populations have most of the adults with white throats. This is true for the populations at Hilltop, west end of Grand Canyon National Park, but is not true for other populations within the Park or for most specimens from the Hualpais. The population from the Burro Mountains, New Mexico, has all of the specimens white-throated. In general, specimens with the entire throat region white have the buffy wash over the abdominal region greatly reduced or entirely lacking. Thus most specimens of relicta and specimens of stephensi from Hilltop and the Burro Mountains lack the buffy wash. Variation of the dorsal coloration within a population is less easily observed. The geographical variation of this color, within subspecies, is discussed beyond and illustrated in Plate I. GROWTH AND REPRODUCTIGN In the absence of growth data for IV. stephensi, we have been guided by previous work in other species of Neotoma, especially in N. albigula from Arizona (Richardson, Jour. Mamm., 24: 134, 1943). Although two species may well differ markedly in their rate of growth and reproductive patterns due to different heredity and different environmental conditions, we have used the N. albigula information to put forth tentative conclusions, leaving it to future work to prove or disprove the validity of our interpretations. We have, thus, assumed that the progress of growth is essentially the same in N. albigula from Arizona as it is in IV. stephensi from that area, and that the gestation period in N. stephensi is, as in other species, approximately 30 days 60 Nov., 1960 HOFFMEISTER AND DE LA TORRE— NEOTOMA 481 in length. From the growth curve of head and body size in N. albigula ( Rich- ardson, op. cit.), approximate ages and thus dates of birth were calculated for our sample of immature specimens ( Group 1 ) of N. stephensi. Table 1 indi- cates the chronological distribution of Group 1 and Group 2 specimens studied, and the dates of capture of pregnant and lactating females. The near disappearance of Group 1 individuals following the middle of August, and the clear increase of Group 2 animals from this time on, strongly suggest that Group 2 is a composite group. Group 2 specimens taken in June, July and August most likely are young born late in the preceding year, that is, in August, September and October. Specimens taken in September and October, however, probably represent young of the year, born in the early months following the earliest breeding period. ian«tK-iXfiiiHi''Wu*'M3SiaMM:A~'-^ %^M* PLATE I Variation in the intensity of the over-all color in 'Scotoma stephensi. The darkest- colored populations are represented by the darkest area; the lightest, by the lightest area. 61 482 JOURNAL OF MAMMALOGY Vol. 41, No. 4 Table L — Bimonthly distribution of Group 1 and Group 2 individuals, ivith dates of capture and age of pregnant and lactaiing females Date of Capture Group 1 Group 2 Pregnant ? ? Lactating ? 9 April 16-30 May 1-15 1 1 16-31 1 June 1-15 3 16-30 5 1 July 1-15 6 8 16-31 6 1 Aug. 1-15 6 5 16-30 3 Sept. 1-15 1 4 16-30 1 11 Oct. 1-15 11 Apr. 30 (Group 4) June 18 (Group 3) July 13 (Group 2) Late Sept. June 26 ( Group 3 ) July 15 (Group 2) Aug. 22 (Group 4) Using the growth rate of IV. albigula as a guide, the dates of birth of Group 1 individuals begin in early March, and thus mating must begin early in Feb- ruary. The dates of birth appear to form two groups — one including March and April, and the other from the middle of May to the middle of July. This suggests two litters but, if a litter can follow another in a period of 15 days, as has been reported for N. albigula, it is quite possible that IV. stephensi may have more than two litters. The usual litter seems to consist of two young, as indicated by four cases of pregnant females containing two embryos each. It is of interest that one of the pregnant animals taken on 13 July, and a lactating female taken 15 July, were clearly of Group 2. The date of capture, in addition to the age group of these animals, certainly seems to indicate that females reach sexual maturity the second season after birth. These females must have been born in the spring or summer of the preceding year. HABITAT Neotoma stephensi occupies an ecological niche quite distinctive from that of Neotoma lepida. In our collecting experience, IV. stephensi is found in rocky situations, usually where the rocks are in piles, and usually where there are piiions and junipers. Neotoma stephensi is not a cliff dweller, although it may be found in the general vicinity of cliffs, but is found where the rocks have rolled down and become stacked. However, even though suitable rocks may be present, N. stephensi most likely will not be found if piiions and junipers are absent. This wood rat is frequently found associated with the piiion mouse, Peromyscus truei, the brush mouse, Peromyscus boylii, or the cactus mouse, Peromyscus eremicus. Within Grand Canyon National Park, Neotoma stephensi is to be found in several ecological situations. In the majority of places, N. stephensi was found in rocky situations where there were piiion, juniper, scrub oak and cliffrose. In these situations, it was usually associated with Neotoma mexicana, but in 62 Nov., 1960 HOFFMEISTER AND DE LA TORRE— NEOTOMA 483 a few cases, with IV. alhigula. In Long Jim Canyon, N. stephensi was in a rocky place within the yellow pine forest. At Cedar Mountain, the species was at the lowest edge of the pinon-juniper belt. Here, on the rocky, desert slopes, there were many cacti and agave plants and only a few straggling piiions or junipers. N. stephensi widely overlapped the range of N. albigula here. At Hilltop, near the western boundary of the Park, N. stephensi was in rocky places decidedly below the piiion-juniper zone and in an area where there were sagebrush, Indian paintbrush and grasses. Here, Perog- nathus intermedins, Neotoma albigula, Eutamias dorsalis, and even Pero- myscus truei and P. hoylii were present. In the vicinity of Rimrock and Montezuma Well in eastern Yavapai County, Arizona, N. stephensi was found along the fractured limestone outcrop extending horizontally near the top of the low mesas. This outcrop was 5 to 8 feet in height and was fractured along its length into large blocks. N. stephensi occurred in the crevices among these blocks, along with many Peromyscus eremicus. Rat houses and piles of droppings were evident all through the outcrop. About 40 feet below the limestone stratum there oc- curred a short dark lava outcrop which was much more finely fractmred. Neotoma albigula was found here as well as on the valley floor where it lived among the cacti. No N. albigula was taken among the blocks of lime- stone where N. stephensi occurred. These field observations made by John S. Hall suggest an important difference in the habitat preference of N. stephensi and N. albigula, even in areas where these two species are very near to each other. At the western edge of the range of Neotoma stephensi, in the Hualpai Moimtains, it may also occur above the pinon-juniper zone in the scrub oak and yellow pine. However, it was taken in rocky situations and in association with Peromyscus boylii. Occasionally, N. stephensi will desert its preferred rocky habitat and move into or under man-made structures. At Pasture Wash, a female with half- grown young was living among bales of hay in the loft of an unused bam. Near McMillen Mine, individuals had homes under large sheets of building material which had been on the ground for a long period of time. In New Mexico, according to Bailey (N. Amer. Faima 53: 188, 1931), N. stephensi in the Burro Mountains live in "stick nests placed about logs or brush, and in places their burrows entered the ground about the bases of rocks with many sticks piled about the entrances." Bailey stated that they occupy the juniper-piiion plateau region, frequenting rocky places and even cliffs. Neotoma stephensi Goldman Range. — Central Arizona, from Hualpai Mountains on the west to western New Mexico on the east, and from McMillenville on the south to extreme south-central Utah (Navajo Mountain) on the north (Fig. 4). 63 484 JOURNAL OF MAMMALOGY Vol. 41, No. 4 Diagnosis. — A medium-sized species of Neotoma with a bushy tail (bushier than in any other species except IV. cinerea); dusky coloration on dorsal sur- face of foot extending from ankle nearly one-third distance to base of toes; small baculum, exclamation-mark-shaped (Fig. 1); upper molariform teeth with following combination of characters: M^ with antero-medial fold absent or poorly developed, and M^ with postero-labial fold and second loph directed obliquely (Figs. 1, 2); last lower molar with deep lingual fold as deep or deeper than labial fold (Fig. 2); skull broad and flat interorbitally; rostrum narrow; nasals truncate posteriorly (Fig. 3). Neotoma stephensi stephensi Goldman Neotoma stephensi Goldman, 1905, Proc. Biol. Soc. Wash., 18: 32, 2 Feb. Neotoma lepida stephensi, Goldman, 1910, N. Amer. Fauna 31: 80, 19 Oct. Neotoma stephensi stephensi, Goldman, 1932, Jour. Mamm., 13: 66, 9 Feb. Fig. 4. — Distribution of the subspecies of Neotoma stephensi. 64 Nov., 1960 HOFFMEISTER AND DE LA TORRE— NEOTOMA 485 Type.— Adult female, U.S. National Museum (Biol. Surv.), No. 117466, from Hualpai Mountains, 6,300 ft., actually at Horse Tank (see Remarks), Mohave County, Arizona, collected on 1 July 1902 by Frank Stephens; original no. 4192. Range. — From the Hualpai, Chemehuevis, and Harquahala mountains in western Arizona east along the MogoUon Plateau to the Burro and Gallinas mountains in southwestern New Mexico (see map, Fig. 4). Diagnosis. — A race showing considerable variation throughout its range but possessing the following characteristics: color of dorsum dark, being darkest in central Arizona and southwestern New Mexico (Plate I); sides above lateral line with narrow fulvous band near Pinkish Buff; nose, fore- head and cheeks grayish or plumbeous; hair of throat usually plumbeous basally; top of tail appears dark, almost black, because more black hairs and few or no white hairs interspersed or show through; size large, as indicated by weight in adult males of usually more than 180 grams; skull large, usually averaging more in greatest length than 41.2 mm. in adult males; upper tooth- row long, usually 8.4 mm. or more; skull with supraorbital ridges frequently beaded and heavy. Comparisons. — Since there is considerable variation within the subspecies stephensi, groups of populations of this race have been compared with the relatively homogeneous "population" of relicta. From relicta, the race stephensi from the Hualpai Mountains, Hackberry, Harquahala Mountains, and Grand Canyon National Park differs as follows: darker coloration (but less different than in other populations of stephensi); more grayish nose, forehead and cheeks; not "white-throated" except for specimens from HiUtop; markedly heavier in weight; broader skull, both actually and relatively to its length (especially marked in specimens from the Hualpai Mountains); longer toothrow and broader interorbitally (es- pecially in specimens from the Hualpais). Central Arizona populations (for discussion, see under Remarks) differ from N. s. relicta as follows: color markedly darker, both on the dorsum and sides; dorsal tail stripe more blackish, less gray; body-weight, length of tooth- row, and interorbital width greater, but not so pronounced as in topotypes of stephensi from the Hualpai Mountains; differ from relicta in other features the same way as do topotypes of stephensi. Populations from eastern Arizona and southwestern New Mexico differ from N. s. relicta as follows (for discussion, see under Remarks) : color darker, including facial region, sides and tail; throat tending to be partially or entirely white (in the latter case showing similarity to relicta); externally larger and heavier; skull with nasals broad, tending to be bell-shaped anteriorly, but constricted posteriorly; ascending arms of premaxillaries broader; differ from relicta in other features as do topotypes of stephensi. Measurements. — See Tables 2 and 3. Remarks. — As mentioned under Comparisons, there is considerable varia- 65 486 JOURNAL OF MAMMALOGY Vol 41, No. 4 Table 2. — External and cranial measurements of "Group 3" male specimens of Neotoma stephensi stephensi. Grouped localities are as follows: Western Arizona (Hackberry, Hualpai Mts., Lucky Star Mine); Northern Arizona (Grand Canyon); Central Arizona (Red Lake, Wupatki area); Southeastern Arizona (McMillenville area, Nantanes Plateau, Black River); Western New Mexico (Burley, Burro Mts., Glenwood) Western Northern Central Southeastern Western Arizona Arizona Arizona Arizona New Mexico Number 8 4 6 3 4 Total length 301.2 311.2 314.0* 308.3 298.3* 283.0-334.0 305.0-320.0 303.0-320.0 302.0-313.0 290.0-310.0 Head and body 173.6 175.8 183.2* 174.0 169.5" 161.0-201.0 172.0-178.0 171.0-203.0 165.0-182.0 165.0-177.0 TaU 127.6 135.0 132.7* 134.3 127.3" 120.0-135.0 127.0-144.0 125.0-138.0 131.0-137.0 120.0-133.0 Hind foot .— 32.3 31.5 32.7 31.7 31.6 31.0-34.0 29.0-34.0 32.0-34.0 31.0-32.0 30.0-33.0 Condylobasal length. 39.5' 39.7 40.0 38.6 37.3 37.6-41.5 39.3-40.1 37.7-42.5 37.7-39.3 36.5-39.2 Zygomatic breadth ... 22.r 21.2 21.3" 20.4" 21.0-23.0 20.6-21.8 20.6-21.9 21.0^ 20.^-20.8 Interorbital 5.7 5.4 5.5 5.5 5.3 5.6-6.0 5.1-5.7 5.3-6.0 5.4-5.6 5.0-5.6 Breadth of rostnmi . 6.5 6.4 6.1 6.5 6.1 5.9-6.9 6.3-6.5 5.9-6.4 6.2-7.0 5.6-6.5 Depth of rostrum ... 6.8 6.9 6.8 6.7 6.3 6.5-7.0 6.8-7.1 6.5-7.1 6.4-7.1 6.1-6.5 Nasal 15.5 16.0 16.3 16.0 15.1 15.0-16.2 15.4-16.8 15.7-17.5 15.8-16.3 14.1-16.8 Incisive foramen 9.1 9.0 9.5 8.7 8.4 8.8-9.5 8.8-9.4 9.0-10.2 8.2-9.2 8.0-9.0 Palatal bridge 7.6 7.5 7.5 8.1 7,7 7.4-7.9 7.3-7.6 7.0-7.7 7.9-8.4 7.5-7.9 Maxillary toothrow . 8.7 8.3 8.5 8.9 8.6 8.3-9.3 8.1-8.6 7.9-9.0 8.7-9.2 8.4-8.8 tion within the subspecies N. s. stephensi as here delimited. This variation can be discussed by grouping the populations into three segments, from (1) the Hualpai Mountains, including others from western Arizona and the Grand Canyon region, (2) central Arizona, and (3) eastern Arizona and western New Mexico. We prefer to describe these three "variants" but not to give them names. The first group of populations in western Arizona possesses the following characteristics: hght color (see Plate I); long toothrow, broad across zygo- matic arches, broad interorbitally, yet skull is only of average length; large foot, average length of body, and short tail. Within this group of populations, those specimens from the type locahty (Hualpai Mountains) have certain unique features such as relatively short but broad skulls, long toothrows, broad interorbital regions, greater weight, and slightly paler than "average" colora- tion for the subspecies. If only topotypes are used for comparison with other 66 Nov., 1960 HOFFMEISTER AND DE LA TORRE— NEOTOMA 487 populations of N. stephensi, the impression may be gained that the population is quite distinct. However, when the range of variation in all characters from throughout the geographic range of stephensi is taken into consideration, the differences are less impressive and not as significant. In many features, topotypes of stephensi are not "typical" or "average" for the race. In central Arizona, there is a series of populations of N. stephensi which possesses characteristics of an average difference distinguishing it from neigh- boring populations. This series, depicted by the dark area in Plate I, com- prises populations from Black Tank to the north, south through Deadman's Flat, Wupatki Monument, Winona, Walnut, Verde Valley, and west to Red Lake and Fort Whipple. Specimens from as far south as McMillenville and the Natanes Plateau in Gila County are not included. This complex of popula- tions possesses the following features, many of which are of an average sort: color dark; dorsal tail stripe often black; sides with the ochraceous color along the lateral line greatly restricted; body long, as long or longer, on the average, than in any population; body weight intermediate between that of relicta and Table 3. — External aand cranial measurements of "Group 3" female specimens of Neotoma stephensi stephensi. Grouped localities are as in Table 2 Western Northern Central Southeastern Arizona Arizona Arizona Arizona Number 12 5 5 3 Total length 274.2 298.3'^ 309.8^ 264.0-329.0 279.0-323.0 293.0-322.0 304.0^ Head and body 170.1 168.3" 175.0* 148.0-195.0 160.0-180.0 173.0-179.0 164.0^ Tail 129.0 130.0" 134.8^ 106.0-141.0 119.0-143.0 118.0-148.0 140.0^ Hind foot 32.0 31.3« 33.3* 30.0-34.0 30.0-32.0 33.0-34.0 32.5^ Condylobasal length -. 38.8" 36.6 38.6 38.8^ 36.8-40.7 35.9-37.5 38.0-39.7 38.2,39.3 Zygomatic breadth .... 21.8' 20.7 20.7 21.6 21.1-23.2 20.4-20.9 20.5-20.9 21.2-22.0 Interorbital 5.4 5.4 5.4 5.7 4.7-5.8 5.2-5.7 5.0-5.7 5.3-6.0 Rostrum breadth _ 6.3 5.9 6.3 6.4^ 6.0-6.6 5.6-6.0 6.1-6.5 6.3,6.5 Rostrum depth 6.6 6.1 6.5 6.7^ 6.2-7.0 5.5-6.2 6.4-6.7 6.5,6.8 Nasal 15.3 14.3 15.7 15.5= 13.9-16.6 13.7-14.6 15.3-16.3 14.9,16.1 Incisive foramen 8.5 7.9-8.8 8.4 8.1-8.8 9.1 8.5-9.5 8.7= 8.5,8.9 Palatal bridge 7.6 7.2 7.2 7.8 6.9-8.5 6.9-7.9 6.5-7.5 7.5-8.5 Maxillary toothrow ... 8.9 8.4 8.6 9.0 8.4-9.5 8.1-8.8 8.5-8.8 8.6-9.3 67 488 JOURNAL OF MAMMALOGY Vol. 41, No. 4 Specimens from the Hualpai Mountains; skull long and narrow, with the per- centage of zygomatic breadth to greatest length in adult males being between 48.4 and 50.4; for the Hualpais, 51.7 to 53.6; for the Grand Canyon, 49.7 to 51.4; similar percentages for adult females from the respective localities are between 49.2 and 53.8; 50.8 to 54.8; 51.6 to 53.5. The length of the upper toothrow and width of the interorbital region are intermediate between relicta and specimens from the Hualpai Mountains. Although this complex of populations may possess some average differences of color and size, we feel that it has not reached the subspecific stage of differentiation. In western New Mexico and eastern Arizona, there is a group of populations that has certain characters in common, and differs in an average way from N. s. stephensi in western Arizona. This group includes localities from 25 mi. N Springerville and from Springerville in Arizona southeast to the Burro Mountains, New Mexico, and northeast to Grants. Some features of this group, many of which are not diagnostic, include: nasals broad, almost bell- shaped anteriorly; posterior arms (ascending branches) of premaxillae on dorsum of skull broad; nasals, posteriorly, constricted more by premaxillae than in most other populations; color dark, as dark as that in central Arizona populations except for specimens from Burley. For remarks about the white- throatedness of specimens from the Burro Mountains, see page 480. One specimen from Grants, Catron County, New Mexico, in many ways exemplifies the extreme of these characters listed: nasals bell-shaped anteriorly, con- stricted posteriorly; posterior, ascending arms of premaxillae broad; inter- orbital region broad; color dark, with the addition of a heavy, rich fulvous wash over the entire underparts. This specimen lacks 90 per cent of the tail, and our first reaction was that this specimen was not a bushy-tailed Neotoma stephensi. The heavy wash of fulvous on the underparts is duplicated in one of three specimens of Neotoma mexicana from 5 mi. SE Grants. However, careful analysis, particularly of cranial features, indicates that the specimen is N. stephensi, not N. mexicana. This individual is either aberrant or reflects the extreme characters of this group of populations we have described above. We prefer the latter interpretation. The fulvous wash on the underparts may reflect the "influence" of the black lava near Grants. We do not think this fulvous wash represents a "dichromatic condition" that Goldman (N. Amer. Fauna 31: 81, 1910) alluded to, although more specimens may indicate that such is the case. A specimen from 4 mi. W McCartys, which is near Grants, is a very young animal and shows none of the features of the Grants' specimen. According to Laurence Huey, the type locality should be regarded as Horse Tank at the southern end of the Hualpai Mountains. Mr. Huey informs us that Frank Stephens provided him with this information. Specimens examined. — Total number, 163, from the following localities: Arizona — Mohave County: Hackberry, 7 (BS); Democrat Mine, 13 mi. ESE Kingman, Hualpai Mts., 1 (UI); Hualpai Mtn. Park, Hualpai Mts., 1 (UI); 1 mi. N Hualpai Peak, Hualpai Mtn. 68 Not;., 1960 HOFFMEISTER AND DE LA TORRE— NEOTOMA 489 Park, 7,000 ft., 1 (UI); Pine Lake, Hualpai Mtn. Park, 6,000 ft, 1 (UI); Hualpai Mts., 5,600 ft., 7,000 ft., 6 (BS), 1 (UI); Horse Tank, 5 (SD); Lucky Star Mine, Chemehuevis Mts., 3 (SD); Yuma County: Harquahala Mts., 5,000 ft., 3 (BS); Coconino County: Hilltop, S side GCNP [= Grand Canyon Nat. Park], 4 (UI); Lower end Prospect Valley, 5,200 ft., 2 (BS); Pasture Wash Ranger Station, 6,300 ft., GCNP, 3 (UI); Pasture Wash, Jet. rds. W9A & W9, GCNP, 1 (UI); 8 mi. N Pine Spg., Hualpai Indian Reservation, 3 (BS); E side Cedar Mt.. 6,400 ft., GCNP, 2 (UI); Yaki Bum, S rim Grand Canyon, 2 (GC); Yavapai Point Station, S rim Grand Canyon, 3 ( GC ) ; Grand Canyon Village, S rim Grand Canyon, 1 (GC); S rim Grand Canyon, 1 (GC); S boundary GCNP, nr. Rowes Well, GCNP, 1 (UI); School athletic grounds. Village, GCNP, 7 (UI); Shoshone Point, GCNP, 2 (UI); Wayside Museum, S rim Grand Canyon, 1 (GC); 1 mi. E, V2 mi. S Desert View Pt, GCNP, 1 (UI); Grandview Pt., GCNP, 1 (UI); Long Jim Canyon, 3 (UI); W side Zuni Pt., 7,200 ft., GCNP, 6 (UI); Rt. 64, SE Boundary GCNP, 1 (UI); Cataract Canyon, 12 mi. WSW Anita, 2 (BS); Red Butte, 2 (BS); Black Tank Lava Beds, 6,100 ft, 3 (UC); Lava Field, 12 mi. N Deadman's Flat, NE San Francisco Mt., 1 (UC); 2.6 mi. W Wupatki Ruins, 8 (UC); Wupatki Indian Ruins, 35 mi. NE Flagstaff, 5,100 ft, 1 (BS), 6 (UC); Deadman's Flat, NE San Francisco Mt., 6,400 ft., 2 (UC); 4 mi. NE Deadman Ranger Station, San Francisco Mt, 6,600 ft., 1 (BS); Red Lake, 5 (BS); 3 mi. NW Winona, 6,200 ft., 6,400 ft, 2 (BS); Aztec Tank, 5,800 ft., 2 (BS); Winona, 6,400 ft, 2 (UC); Wahiut, 1 (BS); Wabut Canyon, 5 mi. S Mt. Elden, 6,500 ft., 1 (BS); Anderson Mesa, Anderson Canyon, 30 mi. SE Flagstaff, 6,500 ft., 1 (BS); Yavapai County: Pine Flat, Jumper Mts., 20 mi. NW Simmons, 3 (BS); 5 mi. S, 1 mi. W Sedona, 1 (UI); 2 mi. N Montezimia WeU, 3 (UI); 2 mi. N Rimrock, 1 (UI); Montezuma WeU, 3,500 ft., 2 (BS); 3 mi. N Ft Whipple, 1 (BS); 6 mi. NW Camp Verde, 1 (UI); Mayer, 1 (BS); Gila County: 7 mi. N Payson, 4,500 ft., 1 (BS); Black River, 5 mi. above mouth White River, 4,600 ft., 1 (BS); nr. sawmill, 25 mi. NE Rice [nov/ San Carlos], Natanes Plateau, 5,800 ft., 3 (BS); McMillen Mine, 5 mi. E, 12y2 mi. N Globe, 4 (UI); Cazador Spring, S base Natanes Plateau, San Carlos Indian Reservation, 4,000 ft., 1 (BS); McMillenville, 4,300 ft., 3 (BS); Apache County: Zuni River, 3 (BS); 8 mi. S St Johns, 5,800 ft, 4 (BS); 25 mi. N Springerville, 2 (BS); Springerville, 7,000 ft., 1 (BS); Greenlee County: 3 mi. W Gosper Ranch, 6,000 ft, 1 (BS); Gosper Ranch, Blue River, 5,000 ft, 2 (BS); New Mexico— Valencia County: Grants, 1 (BS); 4 mi. W McCartys, 1 (UM); Socorro County: Burley [22 mi. N Augustine], 3 (BS); Catron County: Largo Canyon, 1 (UC); Whitewater Canyon, 5 mi. NE Glenwood, 5,450 ft., 1 (AM); Glenwood, San Francisco Valley, 5,000 ft, 1 (BS); Grant County: Burro Mts., 8 (BS). Additional records. — Arizona — Coconino County: S side Bass Camp, 8 mi. NW Grand Canyon (BS); 1 mi. N Bass Camp, 5,200 ft, (BS); 3 mi. S Bass Camp, 5,400 ft, (BS); Bass Camp (BS); nr. Bright Angel Trail, Grand Canyon, (BS); S Yaki Point, in yellow pine, Grand Canyon (BS); top of rim nr. Village, Grand Canyon, 6,800 ft., (BS); Grand View Point, Grand Canyon, (BS); Trash Tank, S rim Grand Canyon, (GC); Trash Wash, S rim Grand Canyon, ( GC ) ; N Red Butte, Main Road, Grand Canyon, ( BS ) ; Red Butte, Museum of Northern Arizona; Wupatki Nat'l Monument, 5,100 ft., Museum of Northern Arizona; Wupatki Indian Ruins, 35 mi. NE Flagstaff, 5,100 ft, (BS); Yavapai County: Fools Gulch, Weaver Mts., (BS). Neotoma stephensi relicta Goldman Neotoma stephensi relicta Goldman, 1932, Jour. Mamm., 13: 66, 9 Feb. Type.— Adult female, U.S. National Museum (Biol. Surv.), No. 67780, from Keams Canyon, Navajo County, Arizona, collected on 22 July 1894 by A. K. Fisher; original no. 1649. 69 490 JOURNAL OF MAMMALOGY Vol. 41, No. 4 Table 4. — External and cranial measurements of "Group 3" specimens of Neotoma stephensi relicta. Localities represented are as follows: Utah — Navajo Mt., Rainbow Bridge; Arizona — Rainbow Lodge, Cedar Ridge, Keams Canyon; New Mexico — Wingate area, Long Canyon, Blanco Females Males Number Total length Head and body Tail Hind foot Condylobasal length Zygomatic breadth ... Interorbital Breadth of rostrum . Depth of rostrum . Nasal Incisive foramen Palatal bridge Maxillary toothrow 10 296.7' 273.O-320.0 163.0" 153.0-175.0 133.7" 120.0-145.0 30.4" 28.0-33.5 36.3« 33.7-38.0 20.0« 18.4-20.2 5.2 4.8-5.7 5.9 5.5-6.3 6.2 5.8-6.5 14.7' 13.5-15.8 8.4 7.8-9.0 7.2 6.7-7.8 8.1 7.7-8.3 10 304.7" 283.0-333.0 172.0" 159.0-187.0 130.9 123.0-144.0 31.2 29.0-33.0 37.8« 36.8-38.6 20.5" 20.0-21.1 5.3" 5.2-5.4 6.0" 5.5-eA 6.6 6.4-7.0 15.4 14.9-16.0 8.8 8.5-9.4 7.4 7.1-7.9 8.2 7.9-8.4 Range. — East of the Colorado River, from Navajo Mountain in extreme south-central Utah, south to the Little Colorado River, and eastward in New Mexico to the San Juan Basin at the north and the Zimi Mountains at the south (see map, Fig. 4). Diagnosis. — A race of Neotoma stephensi characterized by markedly pale buffy color, with much less blackish and brownish than in other populations of species; sides, above lateral line, with large extent of buffy color, which varies between Pinkish Buff and Light Ochraceous-Salmon; nose, forehead and cheeks buffy, less plumbeous tiian other subspecies; hairs on throat usually white throughout (base not plumbeous); top of tail appears grayish because few dark hairs present; underparts usually without fulvous wash; size small, being best indicated by Hght weight (adult males less than 117 gms.); skull small, greatest length averaging less than 41.2 mm. in adult males; upper toothrow short, usually less than 8.4 mm.; skull narrow, with interorbital width usually 5.3 mm. or less and narrow across zygomata, with males usually 70 Nov., 1960 HOFFMEISTER AND DE LA TORRE— NEOTOMA 491 averaging less than 21.0 mm.; skull delicate for the species, with supraorbital ridges not heavily beaded. Comparisons. — For a comparison with N. stephensi stephensi, and the geo- graphic variants within that subspecies, see the account of N. s. stephensi. Measurements. — See Table 4. Remarks. — N. s. relicta is a pale race inhabiting the mountainous areas of the elevated, short grass region of northeastern Arizona. The race does extend into New Mexico and Utah, but its principal center of distribution is in Arizona. N. s. relicta is quite distinctive from many populations of N. s. stephensi adjacent to its range. It is less distinctive from some other populations farther removed. However, with a combination of characters (principally on the basis of color), relicta and stephensi can be distinguished. As here delimited, relicta shows less morphological variation geographically than does stephensi. Specimens from the eastern part of Grand Canyon National Park, referred to stephensi, show considerable approach toward specimens from Cedar Ridge, referred to relicta. These two localities are less than 45 miles apart. The intermediacy is in color and, to some extent, size of the skull. Specimens from Wingate and Gallup are of a color intermediate between that of relicta and stephensi, but the skulls of the adult specimens seem more as in relicta, to which subspecies they are referred. Specimens examined. — Total number, 62, from the following localities: Utah — San Juan County: Rainbow Bridge, 4,000 ft., 1 (UC); nr. War God Spring, 8,500 ft., Navajo Mt, 1 (AM); Navajo Mt. Trading Post, SE side Navajo Mt., 1 (UC); Arizona — Coconino County: Rainbow Lodge, Navajo Mts., 6,400 ft., 2 (BS); 5 mi. S simmiit, Navajo Mt., 1 (UC); Cedar Ridge, 30 mi. N Tuba, 3 (BS); Cedar Ridge, 6,000 ft., 10 (UC); Tuba, 1 (BS); Navajo County: Long Canyon, 6,450 ft., 1 (AM); Keams Canyon, 80 mi. N Holbrook, 20 (BS); Keams Canyon, 6,200 ft., 6 (UC); Apache County: Ganado, 6,500 ft., 5 (BS); St. Michaels, 7,000 ft., 2 (BS); New Mexico— Sen Juan County: Blanco, 1 (BS); McKinley County: Gallup, 1 (BS); Wingate, 2 (BS); Fort Wingate, 3 (BS); 12 mi. S Gallup, 1 (BS). Museum of Natural History, Univ. of Illinois, Urbana. Received 18 August 1959. 71 Am. Zoologist, 7:223-232(1967). Multiple Character Analysis of Canis lupus, latrans, and familiaris, With a Discussion of the Relationships of Canis niger Barbara Lawrence, Museum of Comparative Zoology, Harvard University, and William H. Bossert, Department of Biology, Harvard University, Cambridge, Mass. Synopsis. A multiple character analysis was undertaken of a broadly representative sample of three species: Canis lupus (wolf), C. latrans (coyote), and C. familiaris (dog). These species are clearly and significantly distinguished by the technique of linear discrimination. The analy- sis provides a basis for the identification of skulls not obviously distinguishable by size or other diagnostic charactei^s. Early populations of Canis n. niger and C. n. gregoryi (red wolf) are compared with the three species above and are found to form a cluster with lupus and to be sharply distinct from the other two species. Additional comparisons show that while lupus lycaon and niger both overlap with lupus, they are distinct from each other. This entire cluster is quite distinct from latrans, with niger being the farthest removed. A sample population of C. ?!. gregoryi, from the edge of the extending range of C. latrans, was examined and found to show too great a range of vari- ation to be attributed to a single species. With the advent of white man in North a unit. When a particular subspecies is re- America and his consequent modification ferred to, a trinomial is used, as Canis lu- of the environment by lumbering and pus lycaon. Canis niger, the red wolf, is clearing for farming, coyotes have been ex- usually considered to include three sub- tending their range (Young and Jackson, species. Their status is uncertain, and Ca- 1951). As they have extended their range, 7us niger as used in the present work refers on the fringes of their newly acquired ter- to the typical form, C. n. niger, and to those ritories, animals which are difficult to iden- southeastern populations, presently called tify have frequently been captured. In the C. n. gregoryi, which show no evidence of South, as often as not, these are called red hybridization and which were collected wolves, in the Northeast, coydogs. In both from well outside the range of latrans. Ca- parts of the country these animals occur nis familiaris, the dog, presents no prob- where coyotes have moved into areas that lem because, in spite of its variability, it is formerly were inhabited by small races of monotypic. wolf. Coincident also with these shifts in The present study was undertaken be- distribution has been an upward revision cause attempts to identify skulls of the in the reported weights for coyotes. Young northeastern population of rather large- and Jackson (1951), eliminating a few out- sized members of the genus Canis bogged sized individuals, give a range of 18-30 down in a mass of overlapping characters, pounds for typical western coyotes, while It was then decided that before such fringe Burt (1946) gives a range of 23-50 pounds populations could be identified we needed for Michigan coyotes. The latter overlaps to know what, if any, combinations of with weights of a long series of wolves from characters reliably separated known Canis Algonquin Provincial Park (unpublished lupus, latrans, and familiaris, particularly data from the Ontario Department of if size were eliminated as a character. This Lands and Forests) and, as a result, size part of the work will be described in detail alone becomes a less useful criterion in dis- in section I. tinguishing between wolves and coyotes. While these three species are unquestion- In the following discussion, since Canis ably distinct, the red wolf, currently called lupus, the wolf, and Canis latrans, the coy- Canis niger, is a more problematical entity ote, are both composite species, these names and will be discussed in section II in the as used in the text refer to each species as light of our findings in section I. (223) 72 224 Barbara Lawrence and William H. Bossert SECTION I The purpose of this part of the study was to determine what, if any, combina- tions of characters separate the three spe- cies, C. lupus, C. latrans, and C. familiaris, and how widely they are separated. To do this, we have used a biased random selec- tion of 20 adult members, including males and females, of each species. In latrans, wide geographic distribution within the original range of the species was an impor- tant factor in choice of specimens. In lu- pus, only North American races were used and large individuals were avoided. In fa- miliaris, the selection was deliberately bi- ased to include the most wolf-like and coy- /t^ .#.- •«'> Fig. 4. — Karyotype of Thomomys umbrinus intermedins Mearns {S, UA 14987). Gardner Canyon, Santa Rita Mts., Santa Cruz Co., Arizona. 87 February 1968 PATTON AND DINGMAN— CHROMOSOMES OF THOMOMYS 7 X X ^^ Aft Aft ftO ^^ ^^ ^^ HA A^ Ail o^ A^ '"'ft ll#\ A^ -^^ ll Al* Aft ^SifN An ti ft A/4<»«* Fig. 6. — Karyotype of Thomomys timbrimis intermedins Mearns (6, UA 14990). Sycamore Canyon, Patagonia Mts., Santa Cruz Co., Arizona. the Huachuca Mountains to T. hottae. However, they incorrectly identified the material from Carr Peak as T. umhrinus intermedius Mearns, and on this basis, in part, assumed that only T. hottae occurred in the mountain range. Lange (1959) examined the type of T. u. intermedius and considered this name as valid and specifically distinct from the Carr Peak material examined by Hoffmeister and Goodpaster. Moreover, both Lange ( 1959 ) and Cockrum (1960) consider specimens from Brown Canyon and the vicinity of Panama Mine (near the west gate of Fort Huachuca) as T. umhrinus. Specimens examined by one of us (JLP) through the courtesy of Dr. Seth B. Benson from the Peterson Ranch, Sunnyside Canyon, definitely are referable to this species. Thomomys umhrinus is unquestionably present in the Huachuca Mountains, therefore, but additional specimens must be collected and karyo- typed to enable assessment of their ecological as well as genetical relation- ships to T. hottae in this same mountain range. Thomomys hottae is not known from the higher elevations of the Patagonia and Santa Rita mountains, although the latter supports well-developed pine forests similar to those occupied by T. hottae on more northern mountains in Arizona as well as the adjacent Huachuca Mountains. Gophers occurring in oak woodland habitats in other mountain ranges in southern Arizona, namely the Santa Catalina, Quinlan, and Chiricahua mountains, have karyotypes identical or similar to those of nearby popula- tions of T. hottae. These animals have been considered to be T. hottae and not T. umhrinus by most authors. The ecological separation of the two species is quite apparent in Sycamore and Italian canyons of the Patagonia Mountains. Generally, samples of T. hottae were obtained only at lower elevations in desert grassland or riparian 89 February 1968 PATTON AND DINGMAN— CHROMOSOMES OF THOMOMYS kit*' «**•••«• Fig. 7. — Karyotype of Fi hybrid between T. bottae modicus and T. umbrinus inter- medius (9, UA 15158). The proposed parental genomes are separated, with that of r. bottae on the left and that of T. umbrinus on the right. habitats (3600^800 ft), while T. umbrinus was found in open oak or juniper- oak woodlands (4500-6000 ft). Unlike the spatial separation of T. bottae and T. umbrinus at two localities in northwestern Chihuahua (Anderson, 1966), the two species are sympatric along a narrow zone in the flat ground bordering the stream beds of Sycamore and Italian canyons {ca. 4400-4850 ft; see Fig. 8). The floors of the canyons in this area consist of a mesquite- desert willow-grassland association with a few riparian elements {e.g., cotton- woods and sycamores) in Sycamore Canyon and a narrow desert riparian woodland community in Italian Canyon. Although both species were trapped on the canyon floors in such communities, T. umbrinus appears more limited to the open, rockier hillsides where Emory and Mexican blue oaks {Quercus emoryi and Q. oblongifoUa) predominate. Hijbridization. — At present, little more than speculative remarks can be made concerning the natural hybridization between T. bottae and T. umbrinus in Sycamore and Italian canyons of the Patagonia Mountains. The single Fi hybrid and the three individuals considered by their karyotypes to be back- cross hybrids all were trapped within the narrow zone of sympatry (see Fig. 8, note localities of hybrids), an indication that the main populations of both species are not affected to a great extent by the hybridization. Although the Fi hybrid was fertile and backcrossing to both parental species is, therefore, judged to occur, lack of extensive introgression coupled with the narrowness of the hybrid zone indicates restricted gene flow between T. bottae and T. umbrinus at this single locality. Historical considerations. — The present distributions of many species in the inland Southwest have been explained in part by the shifting climatic and vegetational events of the late Pleistocene (summarized by Martin and Mehringer, 1965). Some populations that were continuous during late glacial to post-glacial times became disjunct, and interconnections were formed between other previously disjunct populations. This is apparent in the cases 90 10 JOURNAL OF MAMMALOGY Vol. 49, No. 1 Miles Fig. 8. — Map of Sycamore and Italian canyons, Patagonia Mountains, Santa Cruz Co., Arizona, showing area of sympatric contact between T. hottae and T. umbrinus (shaded area) and locahties where gophers were trapped in relation to zone of contact. SoHd circles = T. hottae; solid triangles = T. umbrinus; solid square = Fi hybrid; half-filled circles = hybrids backcrossed to T. bottae; half-filled triangle = hybrid backcrossed to T. umbrinus. (Additional specir^ens of T. bottae were trapped from outside the area indicated on map.) of T. bottae and T. umbrinus. Their present distributional pattern with "islands" of T. umbrinus in a "sea" of T. bottae, and the increased contact between the two species in Recent times, have been accompanied by repro- ductive isolating mechanisms that were inadequate to prevent limited hybrid- ization between the two species in the restricted area of the Patagonia Mountains. Presumably this occurred as the two species came into contact following changes in the vegetation in the present area of sympatry. Present hybridization between other groups of vertebrates (for example, Cnemi- dophorus — Zweifel, 1962; Lowe and Wright, 1966) and plants (for example, oaks — Tucker, 1963) in the same general area has been attributed to these factors also. Much of the current vegetation between the Santa Rita, Huachuca, and Patagonia mountains is desert grassland with scattered oaks and mesquites. A more well-developed oak woodland is present between the Patagonia and Pajarito mountains, following elevational contours into and out of Sonora. If the oak woodland were lowered 400 m during the late Pleistocene ( Wiscon- sin glaciation), as suggested by Martin and Mehringer (1965), all of these ranges would have been connected by well-developed oak woodland. Such habitats appear ideal for T. umbrinus in these mountain ranges today. 91 February 1968 PATTON AND DIXGMAX— CHROMOSOMES OF THOMOMYS 11 Thomomys umbrinus is presently widespread on the Mexican Plateau, and was once undoubtedh' more widely spread in southern Arizona. The area that T. umbrinus now inhabits is known to contain the greatest degree of Mexican floral influence anywhere in Arizona ( Marshall, 1957; Martin, 1963; Lowe, 1964); this applies especially to the vegetation at the lower and intermediate elevations. The two main factors contributing to the presently isolated state of T. umbrinus populations in Arizona were, then, the withdrawal of the extensive woodlands of the late glacial period to their present positions during the past 10,000 years (Martin and Mehringer, 1965), and the spread of r. bottae through the lower elevations in recently invaded desert scrub and desert grassland communities. In areas where T. bottae is present in the higher pine forests and the lower valley floors (for example, in the Huachuca Mountains). T. umbrinus, so far as is known, is restricted to the intermediate elevations in the oak zones. However, in mountain ranges where no T. bottae are found in upper eleva- tions (for example, in the Santa Rita Mountains), T. umbrinus inhabits both the oak woodland and the pine forests. It is apparent, therefore, that the present restriction of T. umbrinus to the oak zone is due in part to dis- placement through inability to compete with T. bottae. In all cases where T. bottae and T. umbrinus approach or meet in Arizona, the former occupies the more friable soils of the valley floors and mountain tops, whereas T. umbrinus is restricted more to the harder soils of the somewhat steeply inclined middle elevations. Thomomys bottae probably does not compete with T. umbrinus for these somewhat marginal habitats, for even in mountain ranges where T. umbrinus is absent, T. bottae populations are scarce in the indurate soils of the oak zones. The ability of T. umbrinus to survive in these habitats appears to have resulted in the present spatial and ecological relationships of the two species of gophers in southern Arizona. Taxonomic conclusions. — A basic problem to systematics has arisen from the above discussion — that is, whether to consider T. bottae and T. umbrinus in Arizona as distinct species that infrequently hybridize, or to consider them subspecies that intergrade. At the present time, it would appear more im- portant not to overshadow the biological findings with nomenclatorial problems. In this respect, we judge that consideration of the two forms as distinct species is in greater accord with the biological inferences. This interpretation allows for a greater appreciation and understanding of the past historical events, present distributional and ecological discordance, and great chromosomal distinction between T. bottae and T. umbrinus. Acknowledgments We are considerably grateful to Drs. Sydney Anderson, ^^'illiam B. Heed, T. C. Hsu, and John W. Wright for critically evaluating the manuscript. Special appreciation is due Dr. Wright for his sound biological advice and for aid in the field, to Dr. Anderson for clarifying the confusion of names apphed to the gophers of the Huachuca Mountains, and to Dr. E. L. Cockrvim for providing equipment and encouragement throughout this 92 12 JOURNAL OF MAMMALOGY Vol. 49, No. 1 study. The field assistance of Robert J. Baker, Charles Drabek, and Oscar H. Soule is also ackno\%ledged. Specimens Examined Specimens prefixed by UA refer to those catalogued in the mammal collection, Depart- ment of Biological Sciences (Zoology), University of Arizona, Tucson. Other numbers refer to the personal field catalogue of one of us (JLP). Thomomys bottae catalinae {2$, 3 9 )• — Arizona. Pima Co.: Snow Bowl, Mt. Lemmon, Santa Catalina Mts. (JLP 737); Bear Wallow, Santa Catalina Mts. ( UA 15411-14). Thomomys bottae modicus (8$, 9 9). — Arizona. Pima Co.: Tucson (UA 1.5149-50, JLP 688); Molino Basin, Santa Catalina Mts. ( UA 1.5415-16). Santa Cruz Co.: 1.1 mi E Amado ( UA 15410); Verba Buena Ranch ( UA 14992, UA 15144, UA 15154); mouth of Italian Canyon, Patagonia Mts. ( UA 14988, UA 15409, UA 15943-44); Chamberlain Tank, Patagonia Mts. (UA 14993, UA 14906-07); Sycamore Canyon, Patagonia Mts., 9.3 mi E Jet Arizona 82 and Washington Camp Road (UA 15942). Thomomys bottae extenuatus (l $ , 19). — Arizona. Cochise Co.: 0.2 mi E Jet Arizona 181 and Turkey Creek Canyon Road, Sulfur Springs Valley ( UA 15145); 3.9 mi E Jet Arizona 181 and Turkey Creek Canyon Road, Sulfur Springs Valley (UA 15151). Thomomys bottae collinus {Z$, 5 9). — Arizona. Cochise Co.: El Coronado Ranch, West Turkey Creek Canyon, Chiricahua Mts. (UA 15154, UA 15147); 1.7 mi E El Coronado Ranch, West Turkey Creek Canyon, Chiricahua Mts. (UA 15152, UA 15148); Rucker Canyon, Chiricahua Mts., ca. 5600 ft (UA 15146, UA 15153); 1 mi below Rustlers Park, Chiricahua Mts. (UA 15406-07). Thomomys bottae proximus (1$, 3 9). — Arizona. Cochise Co.: Carr Canyon Ranch, Huachuca Mts. (UA 15404-05); 0.5 mi N Clark Spring, Carr Canyon, Huachuca Mts. (UA 1.5417-18). Thomomys bottae alienus (19). — Arizona. Cochise Co.: ca. 4.5 mi S St. David on US 80 (UA 14989). Thomomys bottae pusillus (1$, 19)- — Arizona. Pima Co.: ca. 1.5 mi below Kit Peak National Observatory, Quinlan Mts. (UA 15419-20). Thomomys umbrinus intermedius {\0 i , 12 9). — Arizona. Santa Cruz Co.: Madera Canyon, Santa Rita Mts. ( UA 15343, UA 15163, UA 15403, UA 15001); Gardner Canyon, Santa Rita Mts. (UA 14986-87); Sycamore Canyon, Patagonia Mts., ca. 9.9 mi E Jet Arizona 82 and Washington Camp Road ( UA 14985 ) and ca. 9.3 mi E Jet Arizona 82 and Washington Camp Road (UA 15937-41); ca. 1.2 mi E Crescent Spring (UA 15156, UA 15012); ca. 0.4 mi E Crescent Spring (UA 15160); ca. 0.6 mi E Crescent Spring (UA 14990); Crescent Spring (UA 15159); ca. 8.7 mi E Jet Arizona 82 and Washington Camp Road (UA 15408); mouth of Italian Canyon, Patagonia Mts. (UA 15157, UA 14983); Italian Canyon, Patagonia Mts. (UA 14984, UA 15936). Thomomys bottae x Thomom,ys umbrinus hybrids (15, 3 9). — Arizona. Santa Cruz Co.: Italian Canyon, Patagonia Mts. (UA 15158, UA 15424); Sycamore Canyon, Patagonia Mts., ca. 8.7 mi E Jet Arizona 82 and Washington Camp Road (UA 14991) and ca. 9.3 mi E Jet Arizona 82 and Washington Camp Road (UA 15935). Literature Cited Antjerson, S. 1966. Taxonomy of gophers, especially Thomomys, in Chihuahua, Mexico. Syst. Zool., 15: 189-198. Baker, R. H. 1953. The pocket gophers ( genus Thomomys ) of Coahuila, Mexico. Univ. Kansas Publ., Mus. Nat. Hist., 5: 499-514. CocKRUM, E. L. 1960. The Recent mammals of Arizona. Univ. Arizona Press, Tucson, vii -^- 276 pp. 93 February 1968 PATTON AND DINGMAN— CHROMOSOMES OF THOMOMYS 13 DuRRANT, S. D. 1946. The pocket gophers (genus Thomomys) of Utah. Univ. Kansas Publ, Mus. Nat. Hist., 1: 1-82. Goldman, E. A. 1947. The pocket gophers (genus Thomomys) of Arizona. N. Anier. Fauna, 59: 1-39. Hall, E. R., and W. B. Davis. 1935. Geographic distribution of pocket gophers (genus Thomomys) in Nevada. Univ. Cahfornia Publ. Zool., 40: 387-402. Hall, E. R., and K. R. Kelson. 1959. The mammals of North America. Ronald Press, New York, 1: xxx + 546 + 79. HoFFMEisTER, D. F. 1963. The yellow-nosed cotton rat, Sigmodon ochrognathus, in Arizona. Amer. Midland Nat., 70: 429-441. HoFFMEiSTER, D. F., AND W. W. GooDPASTER. 1954. The mammals of the Huachuca Mountains, southeastern Arizona. iHinois Biol. Monogr., 24: v -f 1-152. Howard, W. E. 1952. A live trap for pocket gophers. J. Mamm., 33: 61-65. Lane, J. D. 1965. Taxonomy of the pocket gopher, Thomomys haileyi. Ph.D. disserta- tion, Univ. Arizona, Tucson. 80 pp. Lange, K. I. 1957. Taxonomy and distribution of pocket gophers (genus Thomomys) in southeastern Arizona. M.S. thesis, Univ. Arizona, Tucson. 68 pp. . 1959. Taxonomy and nomenclature of some pocket gophers from southeastern Arizona. Proc. Biol. Soc. Washington, 72: 127-132. Lowe, C. H. 1964. The vertebrates of Arizona. Univ. Arizona Press, Tucson, vii + 259 pp. Lowe, C. H., and J. W. Wright. 1966. Evolution of parthenogenetic species of Cnemidophorus (whiptail lizards) in western North America. J. Arizona Acad. Sci., 4: 81-87. Marshall, J. T., Jr. 1957. Birds of the pine-oak woodland in southern Arizona and adjacent Mexico. Pacific Coast Avifauna, 32: 1-125. Martin, P. S. 1963. The last 10,000 years, a fossil pollen record of the American Southwest. Univ. Arizona Press, Tucson, 87 pp. Martin, P. S., and P. J. Mehringer, Jr. 1965. Pleistocene pollen analysis and bio- geography of the Southwest. Pp. 433-451, in The Quaternary of the United States (H. E. Wright and D. G. Frey, eds.). Princeton Univ. Press, New Jersey. Patton, J. L. 1967. Chromosome studies of certain pocket mice, genus Perognathiis (Rodentia: Heteromyidae ) . J. Mamm., 48: 27-37. Tucker, J. M. 1963. Studies in the Quercus undulatus complex. Ill, The contribution of Q. arizonica. Amer. J. Bot., 50: 699-708. Zweifel, R. G. 1962. Analysis of hybridization between two subspecies of the desert whiptail lizard, Cnemidophorus tigris. Copeia, 1962: 749-766. Department of Biological Sciences, University of Arizona, Tucson, 85721, and Depart- ment of Biology, University of San Diego, San Diego, California 92110. Accepted 30 October 1967. 94 Comp. Biochem. Physiol., 1966, Vol. 18, pp. 639 to 651. Pergamon Press Ltd. Printed in Great Britain SERUM PROTEIN ELECTROPHORESIS IN THE TAXONOMY OF SOME SPECIES OF THE GROUND SQUIRREL SUBGENUS SPERMOPHILUS* CHARLES F. NADLER and CHARLES E. HUGHES Department of Medicine, Northwestern University Medical School, Chicago, Illinois {Received IS January 1966) Abstract — 1. Serum protein patterns of Spermophilus undulatus, Spermophilus columbianiis and Spermophilus beldingi were analyzed by two-dimensional starch-gel electrophoresis. 2. The patterns, although generally similar, exhibited variation in nine fractions or groups of fractions and these fractions had taxonomic significance at the level of population, subspecies or species. Intraspecific variability of two unidentified protein fractions was observed in S. undulatus kennicotti and S. beldingi and a third protein polymorphism involving transferrin was also observed in S. undulatus kennicotti. Intraspecific divergence between Arctic and sub-Arctic subspecies of S. undulatus was found and proteins from the latter show a closer resemblance to S. columbianus . 3. The protein characters support present taxonomic concepts of the species of Spermophilus that indicate a close relationship between all three species and a closer relationship between S. undulatus and S. columbianus. 4. Protein characters, as observed in the genus Spermophilus, appear to offer great promise as a method for systematic investigation at intraspecific levels where gross morphologic characters are least definitive. INTRODUCTION The concept that protein synthesis is dependent on rigid genetic control and is therefore a reflection of the genotype has provided a sound theoretical basis for utilizing physico-chemical characteristics of proteins in taxonomic studies. Among the simpler and more reliable techniques for study of serum protein fractions are paper electrophoresis and starch-gel electrophoresis. The more recent use of starch-gel electrophoresis provides an increased resolving power that results in patterns containing as many as thirty protein fractions in some human sera (Smithies, 1959), in comparison to the usual five fractions observed with paper electrophoresis. Many different vertebrates including primates (Goodman, 1963) and Rodentia (Blumberg et al., 1960) of the class Mammalia, Reptilia and Amphibia (Dessauer et al., 1962) and fishes (Sanders, 1964) have been investigated with electrophoretic techniques and the data applied to the taxonomy of groups within the respective * This investigation was supported by National Science Foundation Grant GB-3251. 639 95 640 Charles F. Nadler and Charles E. Hughes classes. In their excellent review, Dessauer and Fox (1964) concluded that starch- gel electrophoresis showed greatest taxonomic promise at infraspecific and specific levels where the probability was high that proteins of identical mobility had identical structure. Certain proteins have been demonstrated to be polymorphic within a species and under genetic control. In the case of transferrin, which binds serum iron, gene frequencies can be calculated and successfully used to evaluate species and population relationships (Goodman et al., 1965). In the present investigation the serum proteins of Spermophilus undulatus, Spermophilus columbianus and Spermophilus beldingi were analyzed by two- dimensional starch-gel electrophoresis (Poulik & Smithies, 1958). These species constitute three of the eight species presently classified in the ground squirrel subgenus Spermophilus (Hall & Kelson, 1959) and they are of taxonomic interest for several reasons. First, Spermophilus is considered the most specialized of the ground squirrel subgenera (Bryant, 1945) and most recently evolved (Black, 1963) yet there is a paucity of gross morphological characters for convincing definition of interspecific relationships. Second, other lines of evidence have been applied to the latter problems that have yielded differing conclusions; zoogeographic evidence (Rand, 1954; MacPherson, 1965) and host-parasite observations (Holland, 1958) indicate a close affinity between S. undulatus and S. columbianus whereas chromosomal evidence (Nadler, 1963, 1966) suggests an equal degree of divergence between S. undulatus, S. columbianus and S. beldingi. Because introducing evidence from additional characters might resolve these interspecific problems, the present study was undertaken with the following aims: (1) to determine whether the number and mobilities of fractions comprising the total serum protein pattern can be analyzed to yield reliable taxonomic charac- ters; (2) to determine the appropriate taxonomic level at which the protein characters are applicable; and (3) to test the validity of these characters by correlat- ing and comparing them with known systematic data of Spermophilus. MATERIALS AND METHODS Serum was obtained from the following animals : Spermophilus undulatus kennicotti (Ross). Alaska: 30 miles E. of Anaktuvuk Pass, 9 males and 5 females; 6 miles E. of Anaktuvuk Pass, 15 males and 7 females. Spermophilus undulatus kodiacensis (Allen). Alaska: Kodiak Island, 1 male and 2 females. Spermophilus beldingi crebrus (Hall). Idaho: Twin Falls Co.; 10 miles N.W. of Buhl, 3 males and 3 females. Spermophilus beldingi oregonus (Merriam). Oregon: Harney Co.; Malheur Valley, 2 males and 1 female ; Burns, 6 males and 4 females. Spermophilus columbianus columbianus (Ord). Idaho: Adams Co.; Brundage Mountain, 4 males and 10 females. Blood was drawn from the heart using sterile equipment and the serum was stored at 4°C. Serum proteins were analyzed using a technique for horizontal, two- dimensional, starch-gel electrophoresis (Poulik & Smithies, 1958) combined with a 96 SERUM PROTEIN PATTERNS OF THE GROUND SQUIRREL 641 tris-discontinuous buffer system (Poulik, 1957). The procedure was modified as reported by Goodman (1963). Starch blocks were bisected, stained with Nigrosin for 2-5 min (1 g Nigrosin, 30 cc acetic acid, 135 cc methyl alcohol and 135 cc distilled water) and decolorized for 24 hr using the same solution without the dye. Two to three separations of each serum specimen were performed on different days and all runs were carefully compared before recording the protein pattern of an individual animal as a scale indian ink drawing. Fractions with staining density equal to albumin or transferrin were recorded in black, those with faint staining reaction were drawn with open lines and fractions of intermediate intensity were stippled. When question arose regarding the presence or absence of a given fraction, additional runs were made to clarify the problem. The densely staining fraction with fastest mobility on paper and starch is tentatively labeled albumin and the fraction with slowest migration rate is labeled gamma globulin. Transferrins {Tf) were identified in several animals of each species by Fe^^ radioautography (Smithies, 1959). Transferrin was investigated at the interspecific level by simultaneously running serum samples from each species in one-dimensional starch-gel separations. RESULTS The serum protein patterns of S. undulatus, S. columbianus and S. beldingi show an overall similarity in the number of protein fractions and their respective mobilities (Figs. 2-8). However, a careful comparison shows nine major differences, illustrated in Fig. 1, that are exhibited in one or more species or populations and Alb <£-3 *"" A Fig. 1. A hypothetical separation of ground squirrel serum proteins illustrating fractions with taxonomic significance. The inital separation on paper results in a horizontal separation from left to right, followed by starch-gel electrophoresis which produces a vertical separation. The labeled fractions correspond with the following characters: Albumin (Character I), Fraction Group A (II), Fraction B (IV), Fraction C (V), Transferrin or T/(III), Fraction D (VI), Fraction E (VII) and Fraction F (VIII). The number of arc-like fractions constitute Character IX. Gamma globulin, y, has the slowest mobility on both paper and starch. 22 97 642 Charles F. Nadler and Charles E. Hughes they may be evaluated as possible taxonomic characters. Other differences in the protein patterns were observed in some of the figures, but were inconsistent or difficult to evaluate because of faint staining properties. Because fractions other than transferrin were not characterized with regard to chemical structure or function, they are given a letter designation from A through F. It is recognized that similarity in mobility and configuration, as determined by a single technique, does not necessarily indicate identical chemical structure and it should be emphas- ized that no individual studied contained all nine characters. *is:i Fig. 2. A protein pattern from S. undulatus kemiicotti (30 mile population). The albumin is homogeneous. Fraction Group A contains four fractions; Fractions B and C are both present; two transferrins are present; Fractions D and E are absent; and Fraction Group F is not separated. One arc fraction is observed. For clarity, the nine protein characters observed in S. undulatus, S. columbianus and S. heldingi will be described individually and later comparisons made between taxa. The same data are presented in tabular form for each species in Table 1. Character I. This character consists of the presence or absence of a homogene- ous serum albumin fraction. S. undulatus (Figs. 2-6) and S. beldingi (Fig. 8) have a homogeneous serum albumin while S. columbianus (Fig. 7) showed an indication of two fractions migrating with nearly identical mobilities on starch. Character II. This character, designated Fraction Group A, is composed of three to four fractions that migrate slightly slower than albumin (Fig. 1). Both populations of S. undulatus kennicotti are characterized by Fraction A containing four individual fractions having an identical group configuration (Figs. 2, 3, 4, 5). In contrast, S. undulatus kodiacensis has only three fractions and these assumed a different configuration (Fig. 6). This pattern seen in S. undulatus kodiacensis is found to be identical to Fraction Group A in S. columbianus (Fig. 7). Fraction Group A of S. beldingi (Fig. 8) contains three fractions that exhibit a configuration different from the other taxa. Therefore, Character II shows both interspecific and intraspecific variability without individual variation. 98 SERUM PROTEIN PATTERNS OF THE GROUND SQUIRREL 643 .^ CO o < to a •2 >c 3 •■•^ O CO fT a s s CO u. o « M H O < X o Pd H O B CU b O z o IX < o U < I 00 c pa c o of trans rin mobi u a CO o t4H O c I- "-I o C s 1 c/5 3 2 Zh d. CO CO + oo" U Q c y c {/I > o c > w c o o w tM o u c Pi en 05 o o 2 fc c o c 2 o C 0} .2 'n en Oi C o C3 y Si 3 2; o 3 pa £ o ^ CO c o (U O > 3 u o ^ CO c o W) v u O oT c u 3 pa ^ CO 99 644 Charles F. Nadler and Charles E. Hughes Character ///consists of the number of transferrin fractions and their mobilities (Fig. 1). The population of S. undulatus kennicotti from 30 miles E. of Anaktuvuk Pass is polymorphic with respect to transferrin because 8 of 14 animals have two darkly staining bands with nearly identical mobility (Figs. 2, 3), both of which Z253. Tf ^ Fig. 3. A protein pattern from S. undulatus kennicotti (30 mile population). This animal differs from the animal in Fig. 2 by an absence of Fractions B and C. Two transferrins are present. Fraction Group F is continuous. \^^ <=> Ci> Tf ^*i!il^ Fig. 4. A protein pattern from S. undulatus kennicotti (30 mile population). This specimen lacks Fraction B, but Fraction C is present and only one transferrin {Tf) is observed. Note the absence of a separation within Fraction Group F. bind Fe^^. At present, the slower fraction is designated as a second molecular form of transferrin. The population of S. undulatus kennicotti from 6 miles E. of Anaktuvuk Pass is homogeneous with respect to transferrin, and only one fraction, migrating with a mobihty identical to the faster fraction from the 30 mile population, 100 SERUM PROTEIN PATTERNS OF THE GROUND SQUIRREL 645 is observed in each of the 22 animals (Fig. 5). On the basis of the transferrin polymorphism it is possible, when the populations are considered as a whole, to differentiate the two by this character. Fig. 5. A protein pattern from S. undulatus kennicotti (6 mile population). Fraction B is absent which is characteristic of all specimens from the 6 mile population. A single transferrin {Tf) is present. Fraction F is continuous. O Fig. 6. A protein pattern from S. undulatus kodiacensis. Fraction Group A contains three fractions with a configuration similar to S. columbianus (Fig. 7). Frac- tions B and C are present and a single transferrin (77) with a mobility identical to the faster fraction of 5. undulatus kennicotti is observed. A darkly staining Fraction D is present whereas Fraction E is absent. The two fractions comprising F are separated. All 3 specimens of S. undulatus kodiacensis (Fig. 6) have a single transferrin fraction that migrates at the same rate as the faster transferrin of both population samples of S. undulatus kennicotti, which suggests a common relationship between the three populations studied. 101 646 Charles F. Nadler and Charles E. Hughes S. columbianus (Fig. 7) and S. beldingi (Fig, 8) each have a single transferrin fraction and the transferrins of both species have an identical mobility that is faster than the mobility of S. undulatus transferrin. Character IV consists of a single fraction that migrates slightly more slowly, on both paper and starch-gel, than the fastest arc-like fraction, and it is designated Fraction B (Fig. 1). It stains with moderate or strong intensity. Fraction B was present in 4 of the 14 specimens of S. undulatus kennicotti (Fig. 2) from the 30 mile population and absent in all animals from the 6 mile population (Fig. 5). Fraction B was present in all 3 specimens of S. undulatus kodiacensis (Fig. 6), 8 of 19 specimens of S. beldingi (Fig. 8) and all 10 specimens of S. columbianus (Fig. 7). Fig. 7. A protein pattern from S. columbianus columbianus. The albumin is notched, indicating a heterogeneous fraction. Three fractions comprise Fraction Group A and their configuration resembles .S. undulatus kodiacensis (Fig. 6). Both Fractions B and C are present and a single transferrin {Tf) migrates more rapidly than Fraction C and the Tf of S. undulatus (Figs. 2-6). Fractions D and E are present and Fraction Group F is separated. Three arc fractions are observed, a characteristic feature of the species. Character V. Fraction C constitutes Character V, It migrates faster than transferrin, slower than Fraction B and is located diagonally between the two fractions on the two-dimensional separation (Fig. 1). Staining intensity varies between individual specimens from moderate to strong. Among specimens of S. undulatus kennicotti, from the 30 mile population Fraction C was observed in 8 of 14 specimens (Fig. 2, 4) and 15 of 22 specimens from the 6 mile population (Fig. 5). All specimens of S. undulatus kodiacensis (Fig. 6), S. columbianus (Fig. 7) and S. beldingi (Fig. 8) had patterns containing Fraction C, and no geographic variation was seen. Character VI is a strongly staining fraction, designated Fraction D, that migrates faster than gamma globulin on paper but slower in starch (Fig. 1). Frac- tion D is absent in both populations of S. undulatus kennicotti (Figs. 2, 3, 4 and 5) 102 SERUM PROTEIN PATTERNS OF THE GROUND SQUIRREL 647 and present in 5. undulatus kodiacensis (Fig. 6), S. coliimbianus (Fig. 7) and S. beldingi (Fig. 8). There is no intrapopulational variation. Character VII consists of a small lightly or moderately staining fraction, Fraction E, that migrates slightly faster than Fraction D on both paper and starch-gel (Fig. 1). Fraction D is absent in both populations of S. undulatus kennicotti (Figs. 2-5) and in S. undulatus kodiacensis (Fig. 6). It is present in S. columbianus (Fig. 7) and S. beldingi (Fig. 8). Fig. 8. A protein pattern from 5. beldingi crebriis. Fraction Group A contains three fractions with different configuration from S. undulatus and S. columbianus. Fractions B, C, D and E are present. A single transferrin {Tf) with mobility more rapid than Fraction C and similar to Tf oi S. columbianus is present. Fraction Group F is separated. Character VIII consists of two fractions that migrate more slowly than Fraction E on starch but at the same or faster rate on paper, and they are designated Fraction Group F (Fig. 1). The two fractions are separated by a gap in S. undulatus kodia- censis (Fig. 6), S. columbianus (Fig. 7) and S. beldingi (Fig. 8) that is produced by more rapid migration of one fraction during the initial paper run. Conversely, Fraction Group F is not separated by a gap in either population of S. undulatus kennicotti (Figs. 2, 3, 4, 5). Character IX. The number of arc-like fractions observed in the protein pattern varies. S. undulatus kennicotti, S. undulatus kodiacensis and S. beldingi patterns are characterized by one arc fraction whereas S. columbianus patterns (Fig. 7) contain three arc fractions. DISCUSSION The present study demonstrates that serum protein patterns from three related ground squirrel species can be analyzed to provide nine potential taxonomic characters. The theoretical basis for the valid use of proteins as characters rests upon the assumption that they are under genetic control and differences in genotype will be reflected by an alteration in chemical structure and behavior. However, 103 648 Charles F. Nadler and Charles E. Hughes before accepting these characters as rehable indicators of taxonomic relationships, it is important to attempt to exclude protein differences influenced by the stage of development or physiologic state of the animal (Dessauer & Fox, 1964). In this study, no juvenile or pregnant animals were studied and no differences in pattern could be attributed to the sex of the animal. With respect to seasonal influences, S. undulatus kodiacensis specimens were obtained in late April 1965, S. beldingi specimens were collected from 27 May-25 June 1965, and S. columbianus specimens were collected on 19 August 1965. The two populations of S. undulatus kennicotti that showed the greatest intraspecific variation in proteins were collected between 20-30 August 1965. Animals were not examined just before or after hibernation. These observations, we believe, indicate that the proposed protein characters are not due to non-genetic variation. Protein characters appear definitive at the species level where Characters I, II, III, VI, VII, VIII and IX may be used alone or in combination (Table 1) to distinguish individual species, and these characters appear fully as diagnostic as such gross morphological characters as pelage color, size, etc. (Howell, 1938). Two taxonomic conclusions at the species level are suggested by the protein characters derived from this investigation. First, S. undulatus, S. beldingi and S. columbianus all show a certain general similarity in their protein patterns that suggests a common ancestral relationship, although each exhibits a number of characteristic features. 5. beldingi is unique in its manifestation of Character II, by a combination of Characters I and IX it can be distinguished from S. columbian- us, and utilizing Characters III and VII it can be distinguished from S. undulatus. S. columbianus has two unique characters, I and IX, yet it exhibits a similarity and presumably close relationship to S. beldingi with respect to Characters III, VI, VII and VIII. Second, it is pertinent that S. undulatus kennicotti and S. undulatus kodiacensis, which share a number of characters that distinguish them from the other species (Table I), also differ with respect to Characters II, VI and VIII. In fact, these latter characters suggest a close relationship between S. undulatus kodiacensis and S. columbianus whereas S. undulatus kennicotti might be considered more divergent. The validity of these seemingly paradoxical observations receives support from zoogeographic and ectoparasite studies. Rand (1954) suggests that S. undulatus and S. columbianus originated from the same stock which became separated by continental glaciation. One part survived in the Beringia refugium during the Wisconsin phase of the Pleistocene and differentiated during the separation to become undulatus and the other survived in a refugium south of the ice and became what is now columbianus. Holland (1958) compared samples of fleas from S. undulatus of Western Alaska and Northern British Columbia with samples from S. columbianus and regarded the fleas only weakly differentiated at the subspecies level. These observations suggested the ranges of the two ground squirrels were at one time contiguous (Holland, 1963) and the possibility that they evolved from a single ancestral stock is also strongly suggested by protein data from the present study. MacPherson (1965) also accepts the thesis offered by Rand (1954) and 104 SERUM PROTEIN PATTERNS OF THE GROUND SQUIRREL 649 supported by Holland (1958). Therefore, among the three species we have examined, the kodiacensis population of S. iindulatus exhibits a closer degree of relationship to 5. columbiamis than that found between any other two species. Analysis of mitotic chromosomes from species of the genus Spermophilus has provided evidence for an equal degree of kar^otypic divergence between 5. beldingi with diploid number {In) of 30, S. columbianus 2n = 32 and S. undulatus with 2n = 34 (Nadler, 1966). Comparison of their chromosomes did not suggest a particularly close relationship between 5. undulatus and S. columbianus because several relatively uncommon types of rearrangements had to be postulated as the mechanisms responsible for their kar^'otypic divergence. The chromosome data might be interpreted to indicate that the two species diverged less recently than other lines of evidence suggest, but it is generally recognized that no one line of evi- dence invariably provides unequivocal evidence for satisfying taxonomic decisions. The subspecific divergence between S. undulatus kennicotti and S. undulatus kodiacensis, which is suggested by protein Characters II, VI and VIII, correlates with Holland's (1958, 1963) observations that Arctic and sub-Arctic populations of S. undulatus are parasitized by different species of fleas. He postulated a possible intraspecific divergence within these ground squirrels, although it was recognized that the differences could be explained by a dependence of the fleas upon ecologic factors other than the host. Perhaps these three lines of evidence, mammalian morpholog}', Siphonapteran morphology and protein analysis, may be interpreted as indicating differing rates of evolutionary divergence from what must have once been a common ancestral gene pool. The Arctic subspecies of S. undulatus may have diverged farther from the ancestral genotype than sub-Arctic subspecies which appear to share a greater number of common characteristics w^ith S. columbianus. It should, of course, be emphasized that the several evidences of divergence manifested by S. undulatus are of a low degree of magnitude and do not imply achievement of species status, although they do suggest the probability of incipient speciation within S. undulatus. The two subspecies of 5. beldingi that were studied could not be differentiated by trenchant protein characters. However, the frequency of Fraction B in two populations of S. beldingi oregonus was 1/10 and 1/3 and in S. beldingi crebrus, it was 6/6. This fraction, constituting Character IV, appears to be a genetically controlled protein that exists in a polymorphic state similar to, but distinct from, transferrin and haptoglobin. As such, its gene frequency might be determined in larger samples from additional populations and thereby serve as a means for distinguishing these tw'o subspecies of S. beldingi. Differences in protein pattern between populations of S. undulatus kennicotti are quite striking and they involve two apparently unrelated fractions (Table 1). First, 8 of 14 specimens of kennicotti from the 30 mile population exhibit two transferrin fractions as judged by the ability of both to bind Fe^^. In contrast, sera from the 22 animals obtained 6 miles from Anaktuvuk Pass contain only one transferrin. Thus, the two populations can be differentiated on the basis of the frequency of one versus two transferrins ; similar observations have been reported 105 650 Charles F. Nadler and Charles E. Hughes in primates (Goodman et ai, 1965) and reptiles (Dessauer et ai, 1962). A second population difference consists of the presence of a low frequency of Fraction B (4/14) in the 30 mile kennicotti population and a complete absence in the 6 mile population (0/22). The chemical identity and function of this fraction are not known: it exists in a polymorphic state in both S. undidatus kennicotti and S. beldingi, whereas it is present in all samples of ^S. undulatus kodiacensis and S. columbianus. As mentioned in the discussion of this fraction in S. beldingi, calcula- tion of its frequency in larger population samples might provide an additional reliable indicator of population composition that could be employed with transferrin in the study of kennicotti populations. A third protein, Fraction C, exhibits a nearly equal degree of variation in the two kennicotti populations and therefore this fraction does not aid in their differentiation, although the fact that all three speci- mens of 5. undulatus kodiacensis manifested this fraction suggests that Character V might be applicable at the subspecies level if larger samples were studied. Because of the differences shown to exist between the proteins of ground squirrels from the 30 and 6 mile localities, it is necessary here to consider the taxonomic status of specimens from the Anaktuvuk Pass region and the geographic distribtuion of animals examined in the present study. Concerning taxonomic status, Bee & Hall (1956) examined a large sample of S. undulatus from the entire Arctic slope of Alaska, including specimens from Anaktuvuk Pass and Tulugak Lake located 12 miles N. of Anaktuvuk Pass, and concluded that all were referable to a single subspecies, kennicotti. Specimens analyzed in the present study were taken from a locality on the Anaktuvuk River (Arctic slope Brooks Range) 6 miles E. of Anaktuvuk Pass and are probably also referable to kennicotti. The second population we studied, however, was taken 30 miles E. of Anaktuvuk Pass on Ernie Creek, which is a tributary of the Koyukuk-Yukon River Drainage (south slope Brooks Range). Since animals from the south slope of the Brooks Range have not yet been studied in detail, our sample from the 30 mile population can be only tentatively regarded as kennicotti and the possibility that they are referable to a different subspecies, perhaps osgoodi, is open to consideration. There are no obvious physical barriers at the divide that might separate the two populations of S. undulatus we studied, although we did observe that the terrain even in mid-August was wet and poorly drained for about 2 miles on either side and that area may be unsuitable for burrow construction. Future studies might be profitably directed toward analysis of proteins from larger samples and examination of additional north and south slope colonies in an attempt to further characterize and explain the mechanisms responsible for the divergence we observed within S. undulatus kennicotti. Acknoiuledgements — We thank Doctor Robert L. Rausch and Russell Pengelly for generously providing some of the specimens studied. Nancy W. Nadler, Bob Ahgook and Johnny Rulland rendered invaluable service to the senior author during field work in Alaska. Doctors Joseph Curtis Moore and Roy Patterson offered valuable suggestions and reviewed the manuscript. Doctor Morris Goodman offered encouragement and advice concerning methodology. 106 SERUM PROTEIN PATTERNS OF THE GROUND SQUIRREL 651 REFERENCES Bee J. W. & Hall E. R. (1956) Mammals of northern Alaska on the Arctic slope. Univ. Kans. Pubis Mus. nat. Hist. 8, 1-309. Black C. C. (1963) A review of the North American Tertiary Sciuridae. Bull. Mus. comp. Zool. Harv. 130, 109-248. Blumberg B. S., Allison A. C. & Garry B. (1960) The haptoglobins hemoglobins and serum proteins of the Alaskan fur seal, ground squirrel and marmot. J. cell. comp. Physiol. 55, 61-71. Bryant M. D. (1945) Phylogeny of Nearcac Sciuridae. Am. Midi. Nat. 33, 257-390. Dessauer H. C. & Fox W. (1964) Electrophoresis in taxonomic studies illustrated by analyses of blood proteins. In Taxonomic Biochemistry and Serology (Edited by Leone C. A.), pp. 625-647. Ronald Press, New York. Dessauer H. C, Fox W. & Hartwig Q. L. (1962) Comparative study of transferrins of Amphibia and Reptilia using starch-gel electrophoresis and autoradiography. Comp. Biochem. Physiol. 5, 17-29. Goodman M. (1963) Serological analysis of the systematics of recent Hominoids. Human Biol. 35, 377^36. Goodman M., Kulkarni A., Poulik E. & Reklys E. (1965) Species and geographic differ- ences in the transferrin polymorphism of macaques. Science, N.Y 147, 884-886. Hall E. R. & Kelson K. R. (1959) The Mammals of North America, Vol. 1, pp. 1-546. Ronald Press, New York. Holland G. P. (1958) Distribution patterns of northern fleas (Siphonaptera). Proc. \Oth int. Congr. Ent. 1, 645-658. Holland G. P. (1963) Faunal affinities of the fleas (Siphonaptera) of Alaska: With an annotated list of species. In Pacific Basin Biogeography (Edited by Gressitt J. L.), pp. 45-63. Bishop Museum Press, Honolulu. Howell A. H. (1938) Revision of the North American ground squirrels. A'^. Am. Fauna 56, l-25o. MacPherson a. H. (1965) The origin of diversity in mammals of the Canadian arctic tundra. Systematic Zool. 14, 153-173. Nadler C. F. (1963) The application of chromosomal analysis to taxonomy of some North American Sciuridae. Proc. XVI int. Congr. Zool. 4, 111-115. Nadler C. F. (1966) Chromosomes and systematics of the ground squirrel subgenus Spermophilus. To be published. Poulik M. D. (1957) Starch-gel electrophoresis in a discontinuous system of buffers. Nature, Lond. 180, 1477-1479. Poulik M. D. & Smithies O. (1958) Comparison and combination of the starch-gel and filter-paper electrophoretic methods applied to human sera: two-dimensional electro- phoresis. Biochem. J. 68, 636-643. Rand A. L. (1954) The ice age and mammal speciation in North America. Arctic 7, 31-35. Sanders B. G. (1964) Electrophoretic studies of serum proteins of three trout species and the resulting hybrids within the family Salmonidae. In Taxonomic Biochemistry and Serology (Edited by Leone C. A.), pp. 673-679. Ronald Press, New York. Smithies O. (1955) Zone electrophoresis in starch-gels: Group variations in the serum proteins of normal human adults. Biochem. J. 61, 629-641. Smithies O. (1959) Zone electrophoresis in starch-gels and its application to studies of serum proteins. Adv. Protein Chem. 14, 65-113. 107 \'oL SO, pp. 22-3-226 1 December 1967 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON THE SYSTEMATIC POSITION' OF THE BATS DESMODUS AND CHILOXYCTERIS. BASED OX HOST-PAK\SITE RELATIONSHIPS (MAMMALIA: CHIROPTEIL\)i By C. E. Machado-Allisox Instituto de Zoologia Tropical, Universidad Central de Venezuela Patterson (IGSe^i has pointed out that the fossils and ecto- parasites of bats pro\ide ver\- Little e\idence which can be used in clarif>"ing the problems of phylogem' in the order Chiroptera. Indeed, chiropteran fossils are scarce, and the majorit}- of chiropteran ectoparasites belong to groups that, ha\ing a Hfe histon- stage off the body of the host do not show notable specificit>". Another factor detracting from the use of ectoparasites is the intimate ecological association existing betvveen bats of different groups, particularly those found in caves, holes in trees, etc., where, occasionally, several species roost together. This beha\'ior favors, without doubt, polyhaematophag>", and there are striking cases of this such as the presence of fleas of the family IschnopsylHdae on bats of the distantly related famHies Molossidae (Tadarida Raf- inesque) and Xoctihonidae {ycctilio Linnaeus). However, host-parasite relationships may yet prove to be of value in shedding new hght on phylogenetic problems in Chiroptera. It must be realized that we still know httle about such relation- ships in the majorit>- of bats and that only in the last few years have careful well-documented collections of the ecto- parasites been made. In \iew of these facts, it becomes particularly important to study a group of ectoparasites, such as the Spintumicidae (Acarina. Mesostigmata) which apparently show great host -A eor-tribntion of the S~/.:':.-. -;in Venezuelan Project, supported by a • - DA-49-193-MD-27SS, of the Medical Research and Development _ ---r. i, Office cf the Siirgeon General, U. S. Army. 35— Pp.oc. Biol. Soc. W.\sh., Vol. SO, 1967 (223) 108 224 Proceedings of the Biological Society of Washington specif icit\' (Rudnick, 1960; Macbado-Alli'soD. 1965a), and also show peculiar modifications in their life c\'cle (Baer, 1952; Rudnick, op. cit.), for instance, ovo\i\~iparit>- and reduction in number of n)Tnphal stages. In the past few years I have been stud>ing the taxonomy of the Neotropical Spintumicidae, especially of the genus Perighschrus Kolenati ( Machado-.Allison, 1965b), which is intimately related to the bats of the familv PhvUostomidae. Comparing the arrangement of the genera and subfamilies of Phyllostomidae, based on the work of Miller (1907) and Simpson (1945), now acc-epted by most mammalogists, with certain data offered by the relationships of Spintumicidae and the bats, I find some significant disagreements which I want to point out. According to Simpson (op. cit.^, the superfamily Phyl- lostomoidea includes the families Phyllostomidae and Des- modidae. Simpson di\ided the family Phyllostomidae into seven subfamilies: Chilonycterinae, PhyUostominae, Glos- sophaginae, CaroUiinae, Stumirinae, Stenodermatinae, and PhyUonycterinae. Among these subfamilies, only one, Phyl- lon\cterinae, is not known to be parasitized by the Spin- tumicidae (there are no pubhshed data on the CaroUiinae, but I have recently found a new spintumicid on Rhinophylla pumilio Peters), The Chilon>"cterinae occupy a special position in the Phyl- lostomidae. The absence of a noseleaf and the lack of articula- tion of the trochiter with the scapula clearly differentiate these bats from those of the other subfanulies. These features led Winge (1923) to associate the Chilonycterinae \^-ith the Noctihonidae in a section of the Phyllostomidae that he called "Mormopini." Xo\ick (1963) found the orientation sounds and associated anatomical features of the Chilonycterinae to differ sharply from those of other phyUostomids. Spintumicidae have not been found on the Xoctilionidae. and the only South .American form that I have found on ChiJonycterls Gray presents morphological characteristics so peculiar that I have considered it to belong to a genus Camer- onieta Machado- Allison, distinct from Periglischrus (Machado- -AlHson. 1965a). The other subfamilies of PhvUostomidae are 109 Host-Parasite Relationships of Bats 225 Table 1. Host-parasite relationships of Phyllostomidae with Spinturnicidae. Spintumicid species Chiropteran genera Present subfamilial assignment Cameronieta ihomasi Chilonycteris Chilonycterinae Periglischrus acutisternus Periglischrus torrealbai Periglischrus parvus Phyllostomus Phyllostomus Microntjcteris Phyllostominae Periglischrus setosus Periglischrus squamosus Periglischrus hopkinsi Glossophaga Anoura Lionycteris Glossophaginae Periglischrus ojastii Sturnira Stumirinae Periglischrus iheringi Artiheus, Vampyrops, etc. Stenodennatinae Periglischrus sp. Rhinophylla Carolliinae Periglischrus herrerai Desmodus Desmodidae parasited by species of Periglischrus (three species on Glos- sophaginae, three on Phyllostominae, one on Carolliinae, one on Stumirinae, and one on Stenodermatinae; see Table 1). Desmodus rotundus E. Geoffrey, family Desmodidae, is the host of the species Periglischrus herrerai Machado-AlHson, which clearly belongs to the genus Periglischrus. In orienta- tion behavior Desmodus resembles phyllostomid genera ( Novick, op. cit. ) . The evidence presented here indicates that a reappraisal of the familial relationships of the Chilonycterinae and the Desmodidae is in order. I would suggest that rather than being a subfamily of the Phyllostomidae, the chilonycterines may form a distinct family. The desmodids, on the other hand, may be no more than a subfamily of the Phyllostomidae. Literature Cited Baer, J. 1952. Ecology of Animal Parasites. The University of Illinois Press, Urbana, 223 pp. Machado- Allison, C. E. 1965a. Notas Sobre Mesostigmata Neo- tropicales III. Cameronieta Thomasi: Nuevo Genero y Nueva Especie Parasita de Chiroptera (Acarina, Spin- turnicidae). Acta Biol. Ven., 4(10): 243-258, 15 Figs. 110 226 Proceedings of the Biological Society of Washington . 1965b. Las Especies Venezolanas del Genero Periglischrus Kolenati, 1857, (Acarina, Mesostigmata, Spintumicidae ) . Acta Biol. Ven., 4(11): 259-348, 46 Figs. Miller, G. S. 1907. The Families and Genera of Bats. Smithsonian Inst., U. S. Nat. Mus., Bull. no. 57, 282 pp., 14 pis. Nov^CK, A. 1963. Orientation in Neotropical bats. II Phyllostomatidae and Desmodontidae. Journ. Mamm., 44: 44-56. Patterson, B. 1956. Mammalian Phylogeny. ler. Symp. Specif. Parasit., Neufchatel, pp. 15-49. RuDNiCK, A. 1960. A Revision of the Mites of the Family Spin- tumicidae (Acarina). Univ. Calif. Publ. Entomol., 17(2): 157-284, pis. 18-48. Simpson, G. G. 1945. The principles of classification and a classifica- tion of mammals. Bull. Amer. Mus. Nat. Hist., vol. 85, 350 pp. Wince, H. 1923. Pattedyr-Slaegter. Kjobenhavn, H. Hagerups F. vol. 1, 360 pp. Ill SECTION 2— ANATOMY AND PHYSIOLOGY Form and function are intimately related. It is difficult to consider one for long or at all thoroughly without considering the other. In comments elsewhere we apply a concept of organizational levels. In taxonomy, classification begins with individuals and proceeds through local aggregates or populations, geographic variants, subspecies, and species, and on to groupings at the level of higher categories. In ecology, the individual organism is the basic unit, then progressively more inclusive and more com- plex levels are local populations of single species, local communities of many species, larger ecosystems, and finally the entire biosphere of life-supporting parts of the surface of the Earth. Similarly in anatomy and physiology there are organizational levels. However, in these fields the individual is the largest unit instead of the smallest, except as we may speak of the anatomical charac- ters of a species or other taxon. Form or function may be studied at the bio- chemical or molecular level, or at progressively higher levels through more complex molecules, mixtures and solutions, organelles, cells, tissues, organs, systems, and finally to the organism in its entirety. The study of anatomy began at the gross level and only after the invention of the microscope and development of special techniques of preparing mate- rials did histological and cytological studies become possible. The recent development of the electron microscope has added several orders of magnitude to the possibilities of studying fine structure. Physiology developed later than gross anatomy and in many ways paralleled chemistry and physics. Our selection of examples is a modest one, drawn from a rich field, and we will have to be content with the above mention of the broad scope of anatomy and physiology, inasmuch as none of our selections has electron photomicro- graphs or histochemical analyses. The selections do, nevertheless, serve to illustrate some fundamental biological concepts. The concept of homeostasis was conceived and broadly applied in physiol- ogy. The concept is relevant, at least by analogy, in ecology under the guise of the "balance of nature," recently expanded to include a sizable vocabulary of terms such as "feedback regulatory mechanism" and "damped cycles." We judge that homeostasis or the tendency of an organism to maintain internal conditions at a dynamic equilibrium is the most general concept of physiology, and that homology is the most general concept of anatomy. The short paper by Hill that begins this section presents one simple ana- tomical problem, and at the same time presents the concept of homology and the problems of interpreting it. The subsequent contributions by Hooper, Hughes, and Mossman are com- parative studies within one family ( Cricetidae ) , one order (Marsupialia), and one class (Mammalia), respectively. Each author studied a different part of the animals concerned and attempted to relate his observations to existing knowledge within the systematic framework. The next paper, by Noback, treats hair, one of the unique features of the Class Mammalia, and theorizes about its adaptive and phylogenetic impHca- 113 tions. This article is from a symposium that contains other interesting papers on hair. The two reprinted papers by Vaughan and Rabb treat form and function together, the former at the level of a taxonomic family, the latter in terms of one set of glands in one species. Among the classic works in mammalian anatomy is Weber's Die Sdugetiere (1927, 1928). English mammalogists dating back to Richard Owen and earlier have published many comparative papers on mammalian anatomy (see for example Pocock's The External Characters of the Pangolins, 1924 ) . One of the most productive American mammalian anatomists was A. B. Howell, whose Anatomy of the Wood Rat (1926) and Aquatic Mammals (1930) both have much to offer. Four good recent works of a comparative nature are Rinker's ( 1954 ) study of four cricetine genera, Vaughan's ( 1959 ) paper on three kinds of bats, Klingener's (1964) treatment of dipodoid rodents, and D. Dwight Davis' major work (1964) on the greater panda. Hildebrand's (1959) paper on locomotion in the cheetah and the horse should be consulted by any student interested in functional anatomy or locomotion. The Anatomical Record and Journal of Morphology are two of the more important serial pubHcations containing papers on anatomy. Among the environmental influences that are important to organisms, and whose effects within the organism must be mitigated, are water, oxygen and other gases, energy sources (food), ions, temperature, and radiation. Most of these factors are touched upon in one or more of the last four papers in this selection in ways that help clarify the adaptive nature of internal, behavioral, and ecological responses. In addition to these aspects of physiology, some areas of special mammalogical interest are hibernation, estivation, thermo- regulation, and sensory physiology. Examples appear in Section 4 (Ecology and Behavior ) as well as in this section. A recent paper by Brown ( 1968 ) , too long to include among our selections, is an excellent example of how physiological adaptations, related in this case to environmental temperature, can be studied comparatively. Other important contributions in mammalian physiology can be found in such journals as Com- parative Biochemistry and Physiology, Journal of Applied Physiology, Journal of Cell and Comparative Physiology, and Physiological Zoology. 114 THE HOMOLOGY OF THE PRESEMIMEMBRANOSUS MUSCLE IN SOME RODENTS JOHN ERIC HILL Museum of Vertebrate Zoology, University of California ONE FIGXTBE Appleton ('28) has distinguished between the caudofemo- ralis muscle and the presemimembranosus muscle (of Leche, 1883) on the basis of their respective relations to the 'nerve to the hamstring muscles.' At the same time he has empha- sized the importance of considering topographical relations in any discussion of the homologies of muscles. The caudo- femoralis, according to this author, crosses over the nerve to the hamstring muscles, dorsal and lateral to this, while the presemimembranosus is medial to the nerve. In many rodents (Parsons, 1894, 1896) the caudofemoralis arises from the caudal vertebrae and inserts on the medial epicondyle of the femur and the caudal surface of this bone. This is the condition in the white rat (Rattus norvegicus) where I found the muscle dorsal to the nerve. In this rodent, and in the others mentioned below, the caudofemoralis is sup- plied by the most cephalic branch of the nerve to the ham- string muscles. In the pocket gophers (Thomomys bulbivorus and Geomys bursarius), in the kangaroo rat (Dipodomys spectabilis), and in a specimen of the wood rat (Neotoma fuscipes), I found that the caudofemoralis arises from the ischial tuberosity and inserts on the medial epicondyle. Howell ('26) found that, in some specimens of the wood rat, the muscle originates from the caudal vertebrae, and in all these cases the caudofemoralis is dorsal (that is superficial) to the nerve to the hamstring muscles; so there is no doubt of its identity. In Dipodomys a few fibers of the muscle were medial to the nerve. 311 THE ANATOMICAL RECORD, VOL. 59, NO. 3 115 312 JOHN ERIC HILL However, in two specimens of a ground squirrel (Citellus richardsonii) and in a mountain beaver (Aplodontia rufa) I found a muscle which originated from the ischium medial (that is deep) to the nerve to the hamstrings, but which other- wise presented the same topographical relations as the caudo- femoralis in the other rodents. Like the latter muscle in these other forms, it was supplied by the most cephalic branch dorsal hjscia gluteus maximus (cut) aluleus medius iensc fasciae lahe uashis lalefalis qtuheus maximus ' (cuO Quadfatus femon's femorvcoccf^g e us (cut) Fascia la la- ienuissimua (cut) piriforinis nefi/e {o hamsifing mascJts ■femo''ococcu<^eiAS (cut) N. ^udendus Nfernoralis cuhneus posiethr I'endon of caudo^emofal/s caudal head of seniilendinasus c audofemorali s -ischial head of semiiendinosus adduclor ma^nus -N ischiadicus biceps femon's somimambrano Sus gracilis M Surae lateralis ienuissimus fculj vV peroneus communis biceps femoris Icuti fascia lafa Fig. 1 The muscles of the thigh of Sciurus griseus showing the relations of the caudofemoralis (presemimembranosus). of the nerve to the hamstrings. According to the views of Appleton ('28) and Leche (1883), however, this muscle would be a presemimembranosus and not a caudofemoralis. The condition of this muscle in the gray squirrel (Sciurus griseus) suggests a solution to the problem of its homology, in rodents at least. In the specimen dissected, the muscle arose by tendinous and fleshy fibers from the ischial bone and, by a slender tendon (fig. 1), from the transverse process 116 HOMOLOGY OF PRESEMIMEMBRANOSUS MUSCLE 313 of the first caudal vertebra. The tendon crossed superficial to the nerve to the hamstring muscles, while the part of the muscle arising from the ischium lay medial and deep to the nerve. Alezais ( '00) described the muscle in Sciurus vulgaris as the 'ischio-condylien.' In this animal it is apparently identical with the muscle described above in Sciurus griseus, but that author did not perceive the significance of the peculiar relation between the muscle and the nerve to the hamstrings. This relation may be considered a stage intermediate between the typical position of the caudofemoralis and that of the so- called presemimembranosus. It may be concluded that muscles do, in rare instances, change their topographical relations to nerves; and that the caudofemoralis and the presemimembranosus are homologous muscles. Consequently, when the muscle in question is con- tinuous with the semimembranosus, the condition should not be considered primitive or undifferentiated, but rather as a secondary fusion of the caudofemoralis and the semimem- branosus. Also, since it is generally accepted by workers in comparative myology (Alezias, '00; Parsons, 1892; Leche, '00) that the presimimembranosus forms part of the great ad- ductor in man, the identification of the former muscle as a modified caudofemoralis seems to fill a gap between the known morphological history of the sciatic part of the adductor magnus and the phylogeny of the caudofemoralis as traced by Appleton ('28). ' 117 314 JOHN EEIC HILL LITERATURE CITED Alezais, H. 1900 Contribution a la myologie des rongeurs. Theses presentees a la faculte des sciences de Paris. Appleton, a. B. 1928 The muscles and nerves of the post-axial region of the tetrapod thigh. J. Anat., vol. 62, pp. 364-438. Howell, A. B. 1926 Anatomy of the wood rat. Baltimore : Williams & Wilkins Company. Leche, W. 1883 Zur Anatomie der Beckenregion bei Insectivora. Kongl. Svensk. Vetensk.-Akad. Handl., vol. 20, pp. 1-112. 1900 Muskulatur. Saugethiere: Mammalia. In Bronn: Klassen und Ordnungen des Thier-Reichs, Bd. 6 : 5 : 1 : 2, S 649-919. Parsons, F. G. 1892 Some points in the myology of rodents. J. Anat. and Physiol., vol. 26, Proc Anat. Soc. Gt. Britain and Ireland. 1894 On the myology of the sciuromorphine and hystricomorphine rodents. Proc. Zool. Soc. London, pp. 251-296. ■ 1896 Myology of rodents. Part II. An account of the myology of the Myomorpha, together with a comparison of the muscles of the various suborders of rodents. Proc. Zool. Soc. London, pp. 159-192. 118 Number 625 May 10, 1962 OCCASIONAL PAPERS OF THE MUSEUM OF ZOOLOGY UNIVERSITY OF MICHIGAN Ann Arbor, Michigan THE GLANS PENIS IN SIGMODON, SIGMOMYS, AND REITHRODON (RODENTIA, CRICETINAE) By Emmet T. Hooper Cotton rats {Sigmodon and Sigmomys), marsh rats, (Holochilus), coney rats (Reithrodon), and red-nosed rats (Neotomys) compose an assemblage which Hershkovitz (1955) considers to be natural and which he designates as the "sigmodont group." This group contrasts with or)zomyine, ichthyomyine, phyllotine, akodont, and other supraspecific assemblages which various authors (e.g., Thomas, 1917; Gyldenstolpe, 1932; Hershkovitz, 1944, 1948, 1955, 1960; and Voront- sov, 1959) have recognized in analyzing the large cricetine fauna of South America. While all of these groups are tentative, at least in regard to total complement of species in each, nevertheless some are strongly characterized and probably natural; and all, whether natural or not, are useful in that they constitute conveniently assessable seg- ments of an unwieldly large South American cricetine fauna, now disposed in approximately 40 nominal genera. New information re- garding three of those genera is provided below. It is derived from fluid-preserved and partially cleared glandes (procedures described by Hooper, 1959) as follows: Reithrodon ciiniculoides: Argentina, Tierra del Fuego, 1 adult. Sigmodon alleni: Michoacan, Dos Aguas, 3 adults. 5. hispidus: Arizona, Pima Co., 1 subadult. Florida, Alachua and Osceola coun- ties, 3 adults. Michoacan, Lombardia, 2 adults. 5. minimus: New Mexico, Hidalgo Co., 1 juvenile. 5. ochrognathus: Texas, Brewster Co., 1 subadult. Sigjnomys alstoni: Venezuela, Aragua, 1 subadult. I am indebted to Elio Massoia for the specimen of Reithrodon and to Charles O. Handley, Jr., for the example of Sigmomys. Figures 1 and 2 were rendered by Suzanne Runyan, staff artist of the Museum of Zoology. The National Science Foundation provided financial aid. Listed below in sequence are representative measurements (in mm.) 1 119 2 Emmet T. Hooper Occ. Papers of Sigmodon hispidus (averages of five adults), Sigmomys ahtoni (one subadult), and Reithrodon cuniculoides (one adult). Length of hind foot: 34, 30, 33; greatest lengths of glans, 7.6, 6.6, 7.8; greatest diam- eter of glans, 6.2, 4.0, 5.0; length of main bone of baculum, 5.5, 4.9, 4.1; length of medial distal segment of baculum, 2.8, 2.0, 2.7; total length of baculum, 8.3, 6.9, 6.8. DESCRIPTION OF GLANDES Sigmodon hispidus.— In Sigmodon hispidus the glans is a spinous, stubby, contorted cylinder (Fig. 1), its length one-fourth to one-fifth that of the hind foot and its greatest diameter approximately three- fourths its length (see measurements). The spines which densely stud almost all of the epidemiis, except tJiat of the terminal crater, are short and thick-set; each is recessed in a rhombic or hexagonal pit. The glans is somewhat swayback and potbellied, yet in its basal one-half or tAvo-thirds it is essentially plain and cylindrical, without lobes or folds other than a short midventral frenum which, as an indistinct raphe, continues distad to the rim of the crater. The distal third or half of the glans is conspicuously hexalobate, the six lobes separated from each other by longitudinal troughs or grooves which increase in depth distad. The lobes are unequal in size and shape; the ventral pair is largest and the least convex, the lateral pair smallest, and the dorsal pair the most convex; the latter is a key item in the swayback appearance of the glans. These lobes converge distally, and tJieir crescentic lips form the scalloped, overhanging rim of the terminal crater. The largest structure in the crater is the mound which houses the medial distal segment of the baculum. Nestled between the lips of the ventral lobes, it projects outside the crater approximately to the limits of the dorsal lobes. The two smaller lateral mounds, housing the lateral processes of the baculum, are closely appressed to the medial mound, and the tip of each is distinctly pointed, rather than gently rounded like the medial mound. Immediately ventral to the medial mound is the meatus urinarius which is guarded ventrally by a ure- thral process. This process consists of a pair of rather thick arms each of which is out-curved and tapers to an obtuse tip (Fig. 1); in one specimen the ventral face of the process is studded with spines. Dorsal to the medial mound is the dorsal papilla, which is a single distensible cone of soft tissue dotted with spines both dorsally and laterally. Two additional pairs of crater conules, here termed "dorsolateral and lateral papillae," are particularly noteworthy because, insofar as known 120 No. 625 Glans Penis in Sigmodont Rodents 3 I I foot glans bac. bone Fig. 1. Views of glans penis of Sigmodon hispidus: a, dorsal; b, lateral; c, incised midventrally exposing urethra; d, epidermal spines, enlarged; e, urethral process, enlarged, ventral aspect; UMMZ 97270, Florida. 121 4 Emmet T. Hooper Occ. Papers in the New World cricetids studied to date, they are peculiar to Sigmodon and Sigmomys. All four of these are spine-studded, stubby, and smoothly rounded terminally. Each dorsolateral papilla is situated just below the crater rim at the junction of the dorsal and lateral lobes. Each lateral papilla is partly recessed in a pocket on the lower flank of the crater wall alongside a lateral bacular mound. There is no ventral shield (a large mass of tissue between the urethral process and the ventral lip of the crater) as seen in most microtines, and the bacular mounds are relatively free within the crater, there being no partitions connecting the lateral mounds with the crater walls; the urethra empties onto the crater floor, not into a partition-encircled secondary crater within the larger crater, an arrangement seen in some rodent species. Below the crater floor is a right and left pair of bilobed sacs (Fig. 1), each ovoid ventral lobe about 1.5 mm. in length, and each atten- uate dorsal lobe approximately a millimeter longer, its tip extending distad almost to the limits of the main bone of the baculum. These sacs or sinuses emerge from tissues situated beside the corpora cavernosa penis and they extend alongside the baculum and the corpus cavernosum urethra, but they apparently are not parts of either of those structures. Composed entirely of soft tissues and engorged with blood in some specimens, they appear to be continuous with the deep dorsal vein and, thus, they seem to be part of the vascular system. Similar sacs, as illustrated in Phyllotis by Pearson (1958:424) for ex- ample, occur in all of those New World cricetids studied to date that have a four-part baculum; they have not been observed in Peromyscus, Neotoma, or other cricetid groups which are characterized by a simple baculum and glans. The four-part baculum is at least as long as the glans and is one- fourth the hind foot in length (see measurements). The main bone, one-sixth the length of the hind foot, is angular and gross. The dorsal face of its wide and angular base is deeply concave between prominent lateral and proximal condyles to which the corpora cavernosa attach, while the ventral surface is almost flat except for a midventral keel of either cartilage or bone which, spanning approximately four-fifths the length of the bone, terminates at the cartilage of the digital junction. The shaft is oval in cross-section, the dorsoventral diameter exceeding the transverse one; as viewed laterally it is slightly bent and is con- stricted terminally, while in ventral view it is gently tapered distad before expanding to form a distinct terminal head. The three distal segments of the baculum are subequal in length. 122 No. 625 Glans Penis in Sigmodont Rodents 5 the lateral pair slightly shorter than the medial one. They differ con- siderably in shape and amount of ossification. In one breeding adult they are entirely cartilaginous, while in four other adults they con- tain various amounts of osseous tissue in addition to cartilage; proba- bly in very old animals they are entirely osseous. The medial segment, attached to the ventral sector of the main bone, projects distad and slightly ventrad, then it bends abruptly dorsad before terminating in a rounded tip. It is approximately oval in cross section in its distal three-fourths, but in its proximal fourth it is much wider than deep and is keeled ventrally; moreover, at the digital junction it bears a pair of lateral processes and a medial flange, the continuation of the midventral keel, w'hich extends over the ventral face of the head of the main bone. In all specimens at hand these three processes are cartilaginous; furthennore, the osseous tissue of the three distal seg- ments is restricted to, or concentrated in, the distal parts of each seg- ment, indicating that ossification apparently proceeds from the tip proximad in S. hispidus. The lateral segments, situated dorsolateral to the medial unit, attach onto the dorsal and lateral parts of the head of the main bone- dorsal to the flanges of the medial segment. Each is pointed and blade- shaped, the dorsoventral diameter exceeding the transverse one; and as viewed ventrally each curves gently distad and slightly laterad. Whether cartilaginous or osseous, they are situated in the lateral parts of each bacular mound, while the medial and distalmost parts of each mound consist entirely of soft tissue, a large part of which is vascular and appears to be instrumental in distention of the mounds. In some examples, the basal parts of the three distal segments of the baculum are more or less coalesced; this is particularly true of the two lateral units, and the two have been interpreted as a single horn-shaped structure (Hamilton, 1946). However, as indicated by Burt (1960) they are separate units (Fig. 1); their individual limits are clear in speci- mens at hand. Sigmodon minimus, S. ochrognathus, and 5. alleni.—l recognize no interspecific differences in the specimens of minimus and ochrognathus, both examples of which are young and rather unsatisfactory. Each closely resembles specimens of hispidus of like age in external size and shape, and in conformation of the six exterior lobes, dorsal papilla, dorsolateral papillae, lateral papillae, urethral process, crater mounds, and baculum. If there are interspecific differences, they are not clearly evident in the materal at hand. The three adults from Dos Aguas, Michoacan, which are labeled S. 123 6 Emmet T. Hooper Occ. Papers alleni, are also like adults of hispidus. The two series differ slightly in regard to size of glans and shape of baculum, but these are small differences and doubtfully interspecific. A few remarks regarding the identification of the specimens from Dos Aguas are needed. Until variation in Sigmodon is better under- stood, .S. alleni seems to be the most appropriate name to apply to these specimens and, as well, to others like them from the vicinity of Autlan, Jalisco, and Angahuan and Uruapan, Michoacan. Cranially and externally distinguishable from specimens of 5. hispidus and S. mclanotis from nearby localities in the same states, they appear to represent a species other than either hispidus or melanotis. They agree well with the description of alleni, but they have not been com- pared directly with the type specimen of that form. Sig7nomys alstoni.—The specimen of Sigmomys alstoni resembles examples of Sigmodon of comparable age in length (relative to hind foot), in external configuration (hexalobate, swaybacked and pot- bellied in lateral view, and covered with proximally directed, thickset, sharp, entrenched spines), shape of dorsal papilla (single, spine-stud- ded cone), appearance of urethral process (two outcurved arms with a longitudinal row of spines on the ventral face of each), shape of the bacular mounds (the medial one large and rounded, each lateral one smaller and rounded laterally but acute medially), position of digits of baculum with respect to the main bone, presence of ventral keel and lateral arms on the medial digit, and occurrence of a midventral keel on the main bone. The specimen differs from examples of Sigmodon in characters as follows: glans smaller in diameter (diam- eter-length ratio approximately 60 per cent, compared with 70-88 per cent in Sigmodon); the six external lobes, particularly the dorsal pair, less prominent; dorsolateral papillae smaller, scarcely more than the spine-studded infolding of the dorsal and lateral lobes; crater more extensively spinous (spines studding most of inner wall of each lateral lobe); medial digit of baculum projecting principally distad, its tip not sharply flexed dorsad; and the osseous proximal segment flatter and wider for a larger fraction of its length. The lateral papillae and baculum warrant additional comment. It is uncertain whether lateral papillae are present in the specimen. Two papillose vascular cores occur at sites where papillae are to be ex- pected, but in the present damaged specimen the overlying crater floor is not correspondingly papillose, although it is strongly spinous; the spiny area occupies most of the inner face of the lateral lobe and of the adjoining crater floor. On the left side of the specimen this 124 No. 625 Glans Penis in Sigmodont Rodents 7 roughly circular spiny area is plate-like, while on the right side it is buckled distad and, thus, resembles a large papilla. If, in undamaged specimens, these areas are papillose, then the lateral papillae in 5. alstoni are relatively larger than any yet seen in Sigmodon. In ventral view, the main bone of the baculum is shaped roughly like an isosceles triangle— wide basally and tapered rather evenly dis- tad (without pronounced incurve) almost to the slight constriction which subtends the small, round, terminal head. Its wide basal part is concave dorsally (between low lateral condyles) and almost fiat ventrally; but farther distad the bone is deeper than wide and, some- what triangular in cross section, it bears a slight midventral ridge to which a cartilaginous keel is attached. The distal segments are entirely cartilaginous. The medial one is deeper than wide in its distal half and blunt terminally; basally it bears a medial process and two lateral flanges. Each lateral segment, also deeper than wide and blunt termin- allv, is situated dorsolateral to the medial unit. Reithrodon cimiculoides— The glans of R. cuniciiloides (Fig. 2) is stubby (diameter-length ratio 64 per cent), subcylindrical, and indis- tinctly lobate, the lobes defined by four, shallow, longitudinal troughs. Two of these depressions, one situated middorsally and the other mid\entrally, extend approximately the full length of the glans and thereby divide the surface of the glans into right and left halves; the distal limit of each is a notch in the crater rim. The shorter third pair of troughs is situated dorsolaterally in the distal half of the glans, but each tenninates short of the rim. All of the epidermis as far distad as the crenate, membranous, overhanging rim of the crater is densely studded with small, conical, recessed tubercles. The three bacular mounds, together with the underlying baculum, resemble a fieur-de-lis in ventral aspect (Fig. 2); the erect medial part extends beyond the crater, while each of the truncate lateral pair sends off an attenuate lateral segment which curves laterad and then distad before terminating in an acute tip. These lateral processes con- tain no cartilage or bone; they consist entirely of soft tissues, a large part of which is vascular and apparently erectile. The spine-tipped dorsal papilla is unusually small and slender; it is a single cone, but a slight cleft near its tip suggests that the papilla may consist of two conules in other specimens. The urethral process is a bilobed flap with two attenuate and erect (not outcurved) arms; it bears two longi- tudinal rows, each of eight tubercles, on its ventral face. There are no lateral or dorsolateral papillae, and the crater walls and floor are smooth and non-spinous. 125 Emmet T. Hooper Occ. Papers I I I foot glans bac. bone Fig. 2. Views of glans penis of Reithrodon cuniculoides; UMMZ 109233, Argen- tina. For explanation see Fig. 1 and text. 126 No. 625 Glans Penis in Siemodont Rodents Q' The baculum is shorter than the glans (see measurements). Its prox- imal, osseous segment consists of a wide basal part and a slender shaft. The basal part, which bears large, proximally directed condyles (these separated medially by a deep notch), is broadly concave ventrally and narrowly and shallowly concave dorsally. The relatively straight shaft is slightly deeper (dorsoventrally) than wide and it bears a slight ventral keel; its terminal portion is slightly expanded laterad and slightly constricted dorsoventrally (Fig. 2). The three distal segments are cartilaginous. The long medial one (its length two-thirds that of the bone) is rod-like for much of its length, but it is enlarged basally and is tapered distally to a pointed tip. The lateral units are disc- shaped in cross section, the dorsoventral diameter of each much great- er than the transverse one. From its attachment on the head of the bone (the attachment dorsal and lateral to that of the medial unit) each lateral segment curves gently laterad and distad before it termin- ates at the base of the laterally projecting process of its lateral mound. DISCUSSION To judge from specimens at hand, the glandes of Sigmodon alleni, S. hispidus, S. minimus, and 5. ochrognathus are fundamentally alike, although they may differ interspecifically in details which can not be appraised in present samples. In each species the stubby, swayback, tubercle-invested glans bears six prominent exterior lobes which sur- round the terminal crater and divide its rim into six corresponding parts. Within the crater there are five spine-studded papillae consist- ing of dorsolateral and lateral pairs in addition to a single cone mid- dorsally. The urethral process bears two attenuate, outcurved arms. The bacular mounds are truncate except for a small, acute medial crest on each lateral mound, and the medial distal segment of the four-part baculum bears a medial keel and a pair of lateral processes on its base, while its tip is flexed sharply dorsad. These characters, together with others, distinguish Sigmodon from the other New World cricetid genera which have been studied to date, with the possible exception of Sigmomys. Sigmomys alstoni, the only species of Sig- momys about which there is information on the glans, appears to be closely similar to species of Sigmodon, but its characters are not yet adequately known. In contrast to the phalli of Sigmodon and Sigtnomys, the glans of Reithrodon cuniculoides is comparatively slim and simple. There are only four exterior lobes, and these are less prominent than the lobes of Sigmodon or Sigmomys. The membranous, crenate, and non-spiny 127 10 Emmet T. Hooper Occ. Papers crater rim is not divided into six distinct lobes. The crater, also smooth and spineless, has no dorsolateral or lateral papillae. The slender dorsal papilla bears spines only at its tip. Each lateral mound has an attenuate lateral process, and the entire configuration of the three crater mounds as well as of the underlying baculum is distinc- tive. The three, long, erect distal segments of the baculum, all car- tilaginous insofar as known, are essentially rod-like in form, without prominent keels or processes. These and other contrasting characters indicate that the glans of R. cuniculoides is morphologically quite different from that seen in Sigmodon and Sigmomys. Preliminary comparisons suggest that it may be more similar to glandes of phyllo- tine or other species which are not now included in the sigmodont group of rodents. 128 No. 625 Glans Penis in Sigmodont Rodents 11 LITERATURE CITED Burt, William H. 1960 Bacula of North American mammals. Miscl. Publ. Mus. Zool. Univ. Mich., 113:1-76,25 pis. Gyldenstolpe, Nils 1932 A manual of Neotropical sigmodont rodents. Kungl. Svenska Veten. Hand., Ser. 3, no. 3: 1-164, 18 pis. Hamilton, William J., Jr. 1946 A study of the baculum in some North American Microtinae. Jour. Mamm., 27:378-87, 1 pi., 3 figs. Hershkovttz^ Philip 1944 A systematic review of the neotropical water rats of the genus Nectornys (Cricctinae). Miscl. Publ. Mus. Zool. Univ. Mich., 58:1-88, 4 pis., 5 figs. 1948 Mammals of northern Colombia, preliminary report No. 3: water rats (genus Nectornys), with supplemental notes on related forms. Proc. U.S. Natl. Mus., 98:49-56. 1955 South American marsh rats, genus Holochilus, with a summary of sig- modont rodents. Fieldiana: Zoology, 37:639-73, 13 pis., 6 figs. 1960 Mammals of northern Colombia, preliminary report No. 8: arboreal rice rats, a systematic revision of the subgenus Oecomys, genus Oryzomys. Proc. U.S. Natl. Mus., 110:513-68, 12 pis., 6 figs. Hooper, Emmet T. 1959 The glans penis in five genera of cricetid rodents. Occ. Pap. Mus. Zool. Univ. Mich., 613:1-10, 5 pis. Pearson, Oliver P. 1958 A taxonomic revision of the rodent genus Phyllotis. Univ. Calif. Publ. Zool., 56:391-496, 8 pis., 21 figs. Thomas, Oldfield 1917 On the arrangement of the South American rats allied to Oryzomys and Rhipidomys. Ann. Mag. Nat. Hist., ser. 8, 20:192-8. Vorontsov, N. N. 1959 The system of hamster (Cricetinae) in the sphere of the world fauna and their phylogenetic relations. Bull. Mosk. Obsh. Ispyt. Prirody, Biol. Sec. (Bull. Moscow Soc. Naturalists), 64:134-7. Accepted for publication February 5, 1962 129 COMPARATIVE MORPHOLOGY OF SPERMATOZOA FROM FIVE MARSUPIAL FAMILIES By R. L. Hughes* [Manuscript received April 8, 1965] Summary The spermatozoa of 18 marsupial species derived from five families have been examined and of these only the spermatozoon of the bandicoot Perameles nasuta has previously been described adequately. The spermatozoon morphology within the families Macropodidae, Dasyuridae, Phascolarctidae, and Peramelidae was relatively homogeneous. A distinctive morphology occured between these families. Within the family Phalangeridae spermatozoa were morphologically diverse, however, as a group they were relatively separate from those of the other families studied. The spermatozoa of the Phascolarctidae (koala, Phascolarctos cinereus, and wombat, Phascolomis mitchelli) have a unique, somewhat rat-like morphology which clearly separates them from those of the other marsupial sperm studied. This finding is of considerable taxonomic interest as most authorities consider the koala to be more closely related to the phalangerid marsupials than to the wombat. I. Introduction Previous descriptions of marsupial spermatozoon morphology cover six of the major marsupial groups. A considerable proportion of these accounts is devoted to a study of the spermatozoon morphology of three species, each belonging to separate marsupial families. (1) Family Didelphidae: Didelphis [Selenka (1887), Fiirst (1887), Waldeyer (1902), KorflF (1902), Retzius (1909), Jordan (1911), Duesberg (1920), Wilson (1928), McCrady (1938), Biggers and Creed (1962)]; (2) family Phalangeridae: Phalangista vulpina ( = Trichosurus vulpecula) [Korff (1902), Benda (1897, 1906), Retzius (1906), Bishop and Walton (I960)]; (3) family Peramehdae: Perameles nasuta [Benda (1906), Cleland (1955, 1956, 1964), Cleland and Rothschild (1959), Bishop and Austin (1957), Bishop and Walton (I960)]. The spermatozoon morphology of two Dasyuridae, Phascogale albipes ( = Smin- thopsis murina) and Dasyurops maculatus, was studied by Fiirst (1887), Bishop and Austin (1957), and Bishop and Walton (1960). Benda's (1906) description of an epididymal sperm from the koala, Phascolarctos (family Phascolarctidae), is, as he admits, inadequate. Spermatozoon morphology studies on members of the family Macropodidae include those of an unknown Macropus sp. (Benda 1906), Macropus billardierii ( = Thylogale billardierii), Petrogale penicillata, Onychogale limata ( = Onychogalea lunata), Bettongia cimiculus (Retzius 1906), Macropus giganteus ( = Macropus canguru) (Binder 1927), and Potorous tridactylus (Hughes 1964). * Division of Wildlife Research, CSIRO, Canberra. Aust. J. ZooL, 1965, 13, 533-43 130 534 R. L. HUGHES The spermatozoa examined in the present study were obtained from members of the five Australasian marsupial families: Phalangeridae, Peramelidae, Dasyuridae, Phascolarctidae, and Macropodidae. The present series of observations has been viewed with reference to those of earlier workers and this has permitted at least an elementary discussion of the comparative aspects of spermatozoon morphology between the marsupial families examined. II. Material and Methods The testes together with the attached epididymis were removed from the scrotum soon after death and fixed in 10% neutral formalin or, more rarely, Bouin's fluid or Carnoy fixative. (i) Method for Adhering Spermatozoa to Microscope Slides The slides were labelled at one end with a diamond pencil and a 15-mm square was marked out at the opposite end. The entire surface of the slide was liberally smeared with Mayer's albumen. A small piece of epididymal tissue was placed in a drop of 10% neutral buifered formalin within the marked square and extensively teased with dissecting needles. Filter paper circles of 5-5 cm diam. were saturated with 10% formalin, drained, and placed over the specimen by a rolling action. Air bubbles were punctured with a needle. The filter paper was kept moist with 10% formalin for at least 30 min and then permitted to dry until free fluid between the slide and the filter paper had disappeared. The filter paper was then removed by a rolling action, excess tissue was removed with fine forceps, and the preparations rinsed and stored in water for staining. (ii) Staining of Spermatozoa (1) Heidenhains Iron Haematoxylin. — Slides containing adhering spermatozoa were transferred from water to a 5 % solution of iron alum and kept in a warm place for 2-3 days. They were then stained with Heidenhain's haematoxylin for a similar period. The area not containing the specimen was thoroughly cleaned with paper tissues during a 10-15 min rinsing period in running tap water. The preparations were then diff'erentiated in 5% iron alum under a staining microscope at 30 sec intervals. The preparation was washed in water and re-examined after each differentiation inter- val. Difl'erentiation times of between 30 sec and 5 min proved satisfactory to show the desired range of structures. The preparations were upgraded to absolute ethyl alcohol, placed in two changes of xylol, and mounted in euparal. (2) Periodic Acid-Schijf {with saliva controls). — Slides containing the mounted spermatozoa were removed from water and placed horizontally in two groups on a flat tray. One group was flooded with water and the other with saliva for 1 hr at a temperature of 37°C. The slides were then thoroughly rinsed in distilled water and stained by a method described by Carleton and Drury (1957, p. 143). The SchiflF's reagent used was de Tomasi (for preparation see Pearse 1961, p. 822). The prepara- tions were mounted in euparal. (3) Feulgen (with and without fast green counterstain). — Slides containing the adhering spermatozoa were removed from water and stained by a method described 131 MORPHOLOGY OF SPERMATOZOA 535 by Pearse (1961, p. 823). The Schiff' s solution used was de Tomasi. Half the Feulgen preparations were stained with fast green counterstain (0-5% solution in 70% ethyl alcohol) for 15-20 min. Both Feulgen and Feulgen-fast green preparations were quickly passed through three changes of 90% alcohol (dips only) to absolute ethyl alcohol and then cleared in xylol and mounted in euparal. Slides were stored until dry in an oven at a temperature of 37°C after mounting in euparal. Preparations were not permitted to dry out during any of the earlier stages in preparation. The drawings of spermatozoa shown in Figure 1 are based on camera lucida outlines using a Xl2 eyepiece in conjunction with a x 100 oil-immersion objective. The spermatozoon dimensions shown in Table 1 are means of 25 observations and were obtained with a special Leitz x 12 -5 screw micrometer eyepiece and a X 100 oil-immersion objective. The preparations used were fixed in 10% neutral formalin or, more rarely, Bouin's fluid or Carnoy and were stained with Heidenhain's iron haematoxylin. During the course of the observations on sperm it became apparent that the efferent ducts connecting the testis and epididymis were either multiple or single within each marsupial family. This was investigated further from frozen transverse sections stained with haematoxylin and eosin. The sections were prepared from the efferent duct or ducts at the point of their emergence from the testis and also approxim- ately midway between the testis and epididymis. The author follows Cleland and Rothschild (1959) in considering for the purpose of description that the flagellum is inserted into the ventral surface of the sperm head and the opposite surface is taken as dorsal. in. Results The mature epididymal spermatozoa of 1 8 marsupial species have been examined. The dimensions of 13 of these spermatozoa are shown in Table 1. The gross morphol- ogy of 14 of the spermatozoa is shown in Figure 1. Spermatozoa of each of the five marsupial families studied (Macropodidae, Phalangeridae, Dasyuridae, Peramelidae, Phascolarctidae*) exhibited sufficient homogeneity in morphology and dimensions of the head, flagellum, and fine structure to be of taxonomic value. The heads of all marsupial spermatozoa examined showed some dorsoventral flattening. This was most marked in the Dasyuridae and Peramelidae. It was least evident in the Phascolarctidae and the genus Pseudocheirus of the Phalangeridae. Macropod and the other phalangerid spermatozoa exhibited an intermediate con- dition. The distal extremity of the head of all species when viewed dorsally was relatively rounded while the shape of the lateral margins and proximal tip varied considerably. In the Dasyuridae the spermatozoon heads of up to 12-7 /a in length in Dasyuroides byrnei are among the longest known for mammals (Table 1). The * The author follows Sonntag (1923) in grouping the wombat and koala in the family Phascolarctidae. 132 536 R. L. HUGHES lateral head margins of dasyurid sperm are slightly convex in dorsal view and taper gradually to a proximal point. Macropod sperm heads are considerably shorter than those of the Dasyuridae and in dorsal outline are elongated ovoids bluntly pointed proximally. The sperm head of the macropod Megaleia rufa (Figs. 1^ and 1/?) is rapidly tapering, a condition typically found in the Phalangeridae. Phalangerid sperm, when viewed dorsally, exhibit considerable variability in the convexity of the lateral head margins. The proximal region of the head is typically semicircular, although sometimes bluntly pointed as in Pseudocheirus cupreus (Figs. \n and \o). Table 1 marsupial spermatozoon dimensions Mean ±SD (/a) Family and Species Head Middle -piece Flagellum Length Width Length Diameter Length Macropodidae Macropus canguru 7-3±016 2-2 + 011 10-7±0-24 1-5±014 111-6+ 3-60 Megaleia rufa* 5-l±0-2I 2-4±009 7-9±0-25 l-4i:012 1040± 4-74 Protemnodon rufogrisea 8-5±0-22 2-3±018 Il-7±0-34 l-6±0-14 115-4± 8-85 Protemnodon agilis] 7-l±0-38 1-8±012 110±0-28 1-4±013 — — Thylogale stiginatica* 7-2±009 2-2±0Il 10-9±0-22 1-5±012 1031± 4-43 Dasyuridae Dasyuroides byrnei 12-7±0-41 2-5±015 40-7 il -26 3- 1+0- 19 242- 1± 6-77 Sarcophilus harrisii llliO-45 2-2±0-I7 34-4:rO-84 2-6±013 207-4±1202 Phalangeridae Petaurus brevicepsX 5-9±019 2-5±018 8-3 :L0-27 1-4±011 101-3± 4-96 Pseudocheirus cupreus% 5-4±016 2-6±011 6-2^016 1-5±017 84-7± 2-47 Pseudocheirus peregrinus 5-9iO-38 3-8±018 6-9±0-21 21±0-22 106-9± 5-31 Phascolarctidae Phascolomis mitchelli 5-7±0-33 l-7x009 180±l-56 0-9±010 87-9± 8-23 Peramelidae Perameles nasuta 5-7±015 30±013 140±0-32 20±011 1941± 5-25 Isoodon macrourus 60±013 3-3±018 10-7±0-19 1-8±014 1651± 3-64 * Fixed in Bouin's fluid; t Carnoy fixative; % from New Guinea. Peramelid spermatozoon heads have concave lateral margins when seen in dorsal view and are relatively square proximally with a median cap. In phascolarctid sperm the proximal portion of the spermatozoon head of both the wombat Phascolomis mitchelli, and the koala, Phascolarctos cinereus, bears a strongly recurved hook. In all sperm, a positive Feulgen reaction for nuclear material (DNA) was given by almost the entire head mass. The DNA-negative areas that took up a fast green counterstain in Feulgen preparations were the acrosome (Fig. 1 ; AC) and basal granule complex which is located at the proximal tip of the flagellum. The acrosome 133 MORPHOLOGY OF SPERMATOZOA 537 20; Fig. 1. — Marsupial epididymal spermatozoa: the drawings are all at the same scale and are based on camera lucida outlines of formalin-fixed Heidenhain's iron haematoxylin preparations. A xl2 eyepiece was used in conjunction with a x 100 oil-immersion lens. Fam. Phascolarctidae: Phascolomis mitchelli, {a) lateral view; Phascolarctos cinereus, (b) lateral view of spermatozoon head with flagellum outline. Fam. Macropodidae: /'ro/f/?;Aio^o/7r///o^m£'fl, (c) ventral view, (^) lateral view; Protemnodon agilis*, (e) lateral view, (/) ventral view; Megaleia rufaj, (g) lateral view, (/;) ventral view; Macropus canguru, (/) lateral view; Thyiogale stiginaiica't, (/) lateral view; (A) ventral view. Fam. Dasyurinae: Dasyuroides byrnei, (I) dorsolateral view; Sarcophilius harrisii, (m) ventral view. Fam. Phalangeridae: Pseudocheirus cupreus, (n) dorsal view, io) lateral view; Pseudocheirus peregrinus, (p) lateral view, (q) dorsal view; Petaurus breviceps, (r) ventral view. Fam. Peramelidae: Isoodon macrourus, {s) ventral view; Peranieles nasuta, (t) ventral view. Key: AC, acrosome; AF, axial filament; CD, cytoplasmic droplet (middle-piece bead) ; CM, cortical helix of main-piece sheath ; MH, mitochondrial helix of middle-piece; A'G, neck granule; i?C, ringcentriole; FG, ventral groove. * Fixed in Carnoy fixative, t Fixed in Bouin's fluid. 134 538 R. L. HUGHES was also variably positive to periodic acid-Schiff (P.A.S.) between species and the basal granule complex was invariably strongly P.A.S. -positive. Neither acrosome nor basal granule complex exhibited any reduction in P.A.S. activity in saliva controls. A faint tinge of green over the entire head surface in Feulgen-fast green preparations pre- sumably represents a limiting membrane. A "nuclear rarefaction" of vacuole-like appearance results from a minute superficial nuclear indentation. The nuclear rarefaction was most conspicuous in the Dasyuridae and Peramelidac and least evident in the Macropodidae. This structure is located on the mid-median aspect of the ventral nuclear surface of all sperm with the exception of those of the Phascolarctidae, where its occurrence is also ventral and median but distal. In most of the marsupial sperm examined acrosomal material (Fig. 1 ; AC) was apparently confined to a relatively small surface area of the head. In the Macropodidae the acrosome is relatively small and is a discrete ovoid structure embedded super- ficially in the extreme proximal portion of the dorsal head surface. In some of the Phalangeridae it has a definite structure as in Pseudocheinis (Figs, \n-\q) where it occupies all but a marginal annular zone of the dorsal head surface and is rather deeply embedded. In other phalangerids, such as Petaurus breviceps (Fig. \r), the dorsal head surface in Feulgen-fast green preparations gives a diffuse acrosomal reaction and bears a shallow depression which extends to all but the margins. A similar diffuse acrosomal reaction of at least the proximal half of the dorsal head surface occurred in the Dasyuridae. The proximal dorsal tip of the dasyurid sperm has a concentration of acrosomal material situated in a minute groove. The acrosomal material in the Peramelidae was found in a small distally flanged proximal cap which covered a minute nuclear protuberance. In the Phascolarctidae the acrosome is a small "comma-shaped" structure. The body of the acrosomal "comma" is embedded superficially in about the middle of the dorsal head surface and the tail of the comma extends throughout the greater portion of the inner curvature of the head hook. In marsupial sperm the ventral surface of the head (by convention that bearing the flagellum) is typically grooved (Fig. 1 ; VG) or bears a shallow distal notch as in the case of the Phascolarctidae. At the distal extremity of the head the groove is broad and deep so that the head is here relatively broad and has the form of an extremely thin curved plate. The groove becomes shallow and narrow towards its proximal extremity; in the Macropodidae and Phalangeridae it terminates at about the mid- median portion of the ventral head surface. The groove is most extensive in the Peramelidae involving the whole of the ventral aspect of the nucleus, only the proximal acrosomal head cap is excepted. In the Dasyuridae it extends throughout the distal four-fifths of the head. Spermatozoa are immature when they enter the head of the epididymis and were characterized by the orientation of the long axis of the head at 90° to the flagellum which was directed towards the nuclear rarefaction. The ventral surface of the spermatozoon head was supported by a somewhat cone-shaped cytoplasmic droplet (Fig. 1 ; CD) of characteristic morphology for each species. Phascolarctid sperm from the head region of the epididymis differed from the other marsupial 135 MORPHOLOGY OF SPERMATOZOA 539 species examined in that tiie flagellum was most frequently observed not to meet the head at right angles and cytoplasmic droplets were small and often absent. On entering the epididymis the head hook of the phascolarctid spermatozoa were only slightly recurved or of an irregular spiral configuration. During the passage of spermatozoa through the epididymis the head hook became simple (vv'ithout spiral) and more tightly recurved. Maturation of spermatozoa is completed during their passage through the epididymis and is accompanied by shedding of the cytoplasmic droplet and rotation of the long axis of the head parallel to that of the flagellum. The neck of the flagellum of mature epididymal sperm in all species was inserted in the vicinity of the nuclear rarefaction. In the Dasyuridae the neck was inserted rather deeply into the proximal margin of the nuclear rarefaction. In the Peramelidae the proximal tip of the flagellum was also deeply inserted and extended from the proximal margin of the nuclear rarefaction to a point about midway between the anterior rim of the nuclear rarefaction and the most proximal extremity of the nucleus. The flagellum is traversed throughout its entirety by an axial filament (Fig. 3; AF). The size of the flagellum varies from species to species. The smallest flagellum was that of Pliascolomis mitchelli with a maximum diameter of 0-9 /^ and a minimum length of 87-9^ (Table 1). The giant flagella of dasyurid sperm are among the largest known for mammals. Dasyiiroides byrnei has a minimum flagellum length of 242 • 1 /Li and a maximum flagellum thickness of 3 • 1 /x. In an old museum specimen of the testes of the now possibly extinct dasyurid Thylacinus cynocephalus (Tasmanian wolf or tiger) the flagellum of epididymal sperm in wax sections had a maximum diameter of 3 0/x and comparable morphology to that of other dasyurids; the sperm heads, although degenerate, were in the form of a long narrow plate, dorso- ventrally flattened and with the flagellum inserted at about the mid-median ventral aspect. Peramelid sperm flagellae were also relatively large, having a maximum diameter of as much as 2 ;li and a minimum length of up to about 200 ^ (Table 1). Macropod and phalangerid sperm flagellae were of intermediate dimensions rarely varying from a maximum diameter of 1 -5 /m and a minimum length of a little over 100 /x. The basal granule complex located at the proximal end of the flagellum consists of at least fused proximal and distal components in the Dasyuridae and Peramelidae. The neck region of the flagellum is a slender proximally pointed cone with a smooth contour, and a small neck granule (Fig. 1 ; NG) is situated at approximately half its length. It was only possible to identify the neck granule with certainty in the Peramelidae, Dasyuridae, and Macropodidae. In the Peramelidae and Dasyuridae it seemed to be a more deeply stained, modified portion of the ground substance of the neck rather than the discrete granule found in the Macropodidae. The sperm of the dasyurids Dasyuroides byrnei, Sarcophilus harrisii, and Thylacinus cynocephalus had a neck length of about 3-5 /j. in comparison with 2-7 fi for the peramelids Isoodon macrourus and Perameles nasuta. Macropod sperm necks ranged in length from 1 • 8 /x in Thylogale stigmatica to 2 • 6 /x in Protemnodon rufogrisea. The neck lengthsof the Phalangeridae and Phascolarctidae were somewhat reduced in compari- son to those of other marsupial families. 136 540 R. L. HUGHES The proximal portion of the middle-piece in all species examined tapered gradu- ally to the diameter of the neck and was particularly firmly clasped by the lateral margins of the sperm head in the Peramelidae and Dasyuridae. The remainder of the middle-piece was relatively cylindrical. A mitochondrial helix (Fig. 1 ; MH) of spiral configuration gave the entire surface of the middle-piece sheath a slightly uneven contour. The mitochondrial helix is a relatively fine structure in the Dasyuridae and Peramelidae, of moderate thickness in the Macropodidae and Phascolarctidae and Petaiirus breviceps of the Phalangeridae. It was quite thick and granular with relatively few gyres in the genus Pseudocheirus of the Phalangeridae. The middle- piece is terminated distally by a ring centriole (Fig. 1 ; RC). The flagellum undergoes an abrupt reduction in diameter on the main-piece side of the ring centriole in both the Pseudocheirus species and to a moderate degree in Petaurus breviceps and the Macropodidae, but not to any appreciable extent in the Dasyuridae, Peramelidae, and Phascolarctidae. The main-piece of the flagellum tapers distally and in twisted specimens appears not to be circular in cross section in Pseudocheirus peregrinus, Peramelidae, Dasyuridae, and Macropodidae. Striations of the sheath of the main piece in all sperm indicate the presence of a fine spiral cortical helix (Fig. 1 ; CH). The tail sheath also gave a strong impression of two lateral thickenings in transverse axis in Macropus canguru, Protemnodon rufogrisea, Pseudocheirus peregrinus, Perameles nasuta, and Isoodon macrourus. The axial filament protruded beyond the terminal portion of the sheath of the main-piece in apparently complete sperm of all species but this cannot be positively taken to represent a true end-piece for in all preparations terminal breakage of the main-piece was prevalent. IV. Discussion Spermatozoa from three other Peramelidae, Perameles gunnii, Isoodon obesulus, and Echymipera rufescens have also been exam.ined superficially and it can be stated that they are comparable in morphology to other peramelid sperm. The spermatozoa of marsupial mice, Antechinus flavipes flavipes, A. f. leucogaster, A. swainsonii, A. stuartii, and Sminthopsis crassicaudata, have a morphology typical of other dasyurids (Woolley, personal communication). This similarity in morphology also extends to two other dasyurids, Phascogale albipes ( = Sminthopsis murina) (Fiirst 1887) and Dasyurops maculatus (Bishop and Austin 1957; Bishop and Walton 1960). The spermatozoon morphology of the macropod species examined in the present study varies in only minor details from that of six other macropods previously described by Benda (1906), Retzius (1906), and Hughes (1964). The phenomenon of conjugate spermatozoa (pairing of relatively numerous epididymal spermatozoa) redescribed and reviewed by Biggers and Creed (1962) for the American opossum, Didelphis, has not been observed in any of the sperm pre- parations from Australian marsupials; however, fresh unfixed material has been examined only for Potorous tridactylus (Hughes 1964) and Phascolomis mitchelli. 137 MORPHOLOGY OF SPERMATOZOA 541 Another feature worth mentioning is that the head of the epididymis was not fused with the testis in any marsupial examined, including Thylacinus cynocephalus and Dendrolagus lumholtzi. In the Dasyuridae and Peramelidae a relatively long single efferent duct together with associated blood vessels links the epididymis to one pole of the testis long axis by way of an extensive membrane, the mesorchium. A tract of relatively long multiple efferent ducts serves the same function in the Phalangeridae, Phascolarctidae, and Macropodidae. A ligament was inserted by way of the mesorchium into the opposite pole of the testis. In both the wombat and the koala the morphology of the sperm, particularly the head, differs strikingly from that of any marsupial sperm previously described. In both species the proximal portion of the spermatozoon head bears a strongly recurved hook not described for other marsupial sperm, and the flagellum is inserted into a notch on one side of the distal portion of the head (Plate 1, Fig. 1 ; and Figs. \a and \b). These features, although somewhat resembling those of certain murid sperm, are not strictly comparable (Plate 1, Fig. 2) (Friend 1936). The hook in the wombat sperm resembles that of Microtus hirtus, Lemmus lemmus, and several other members of the murid subfamily Microtinae in that the hook contains no supporting "rod" and its tip like that of Lemmus lemmus is typically extremely reflected so that it lies against the distal portion of the head (Friend 1936). The position of the hook in Phascolomis is not an artefact of fixation for it was observed in living spermatozoa from the epididymis of several specimens. In sperm from the head of the epididymis the curvature of the hook frequently approximated to that of rats and mice. It can be seen from Plate 1, Figure 1, and Figures \a and \b that the insertion notch of the flagellum of the wombat and koala sperm is located on the opposite side of the head to the hook, whereas in the hooked types of murid sperm both structures occur on the same side of the head (Plate 1, Fig. 2). A head hook is absent in at least the murine, Micromys mmutus and in the microtine Ondatra zibethica (Friend 1936). In Heidenhain's iron haematoxylin preparations the head length of the wombat sperm measured from the distal extremity to the most proximal point of the curvature of the hook (i.e. excluding the recurved portion of the hook) is about 5-7 /x in contrast to 8-0 /x and 11-7/x for mouse and rat, respectively (Friend 1936). Feulgen pre- parations (with or without fast green counterstain) of wombat and koala sperm have shown that nuclear material (DNA) extends to the tip of the hook and occupies all but a small comma-shaped acrosomal portion of the head. Herein lies the greatest departure of wombat sperm from the hooked varieties of murid sperm. In several microtine species the hook is formed entirely from a proximal extension of the nuclear cap (acrosome). In murine sperm a hooked portion of the nucleus bearing a rod extends into the hooked nuclear cap and follows its contour almost to its proximal extremity (Friend 1936). On the basis of skeletal and dental structure most workers consider the koala to be more closely related to the ringtail possums of the genus Pseudocheirus than the wombat (Wood Jones 1924; Simpson 1945). Comparisons of sperm morphology on which selection pressure would presumably be lower than that for external characters of an animal such as skeletal or dental characters, is therefore of considerable interest as a possible basis for taxonomic classification. 138 542 R. L. HUGHES Tt can be seen from the previous descriptions that the spermatozoon oi Pseudo- cheirus peregrinus is not intermediate in structure between the more typical marsupial types (Macropodidae and Dasyuridae) and those of the highly divergent wombat and koala. On the contrary, it deviates in quite a difTerent manner from the typical marsupial patterns. The head is broad (3-8/x) in comparison to its length (5-9/^), the anterior end lacks a hook and is semi-circular in dorsal view (Plate 1, Figs. 3 and 4). Other distinguishing features are the shape and position of the acrosome previously mentioned and a relatively short middle-piece (6-9 /x). The view that the koala is more closely related to the ringtail possum than the wombat is not supported by comparisons of sperm morphology. On the contrary, the findings reported here support the observations of Sonntag (1923) and Troughton (1957) who considered that the koala shares sufficient characters with the wombat for its classification along with the phalangers to be rejected. V. Acknowledgments The author wishes to express his sincere thanks to Dr. E. H. Hipsley, Director, Institute of Anatomy, Canberra; to J. T. Woods, Queensland Museum; to J. A. Thomson, Zoology Department, University of Melbourne; to Dr. M. E. GriflSths, W. E. Poole, K. Keith, M. G. Ridpath, Division of Wildlife Research, CSIRO, for material; to J. Sangiau and L. S. Hall for technical assistance; to E. Slater for photography; and to Professor K. W. Cleland and Dr. A. W. H. Braden who offered helpful criticism. VI. References Benda, C. (1897). — Neuere Mittheilungen iiber die Histiogenese der Saugethierspermatozoen. Verh. berl. physiol. Ges. 1897. [In Arch. Anat. Physiol. (Physiol. Abt.) 1897: 406-14.) Benda, C. (1906). — Die Spermiogenese der Marsupialier. Denkschr. med.-naturw. Ges. Jena6: 441-58. BiGGERS, J. D., and Creed, R. F. S. (1962). — Conjugate spermatozoa of the North American opossum. Nature, Lond. 196: 1112-3. Binder, S. (1927). — Spermatogenese von Macropus giganteus. Z. Zellforsch. 5: 293-346. Bishop, M. W. H., and Austin, C. R. (1957). — Mammalian spermatozoa. Endeavour 16: 137-50. Bishop, M. W. H., and Walton, A; (1960). — Spermatogenesis and the structure of mammalian spermatozoa. In "Marshall's Physiology of Reproduction". (Ed. A. S. Parkes.) 3rd Ed. Vol. 1, Pt. 2, pp. 1-129. (Longmans, Green and Co.: London.) Carleton, H., and Drury, R. A. B. (1957). — "Histological Technique." (Oxford University Press.) Cleland, K. W. (1955). — Structure of bandicoot sperm tail. Aust. J. Sci. 18: 96-7. Cleland, K. W. (1956). — Acrosome formation in bandicoot spermiogenesis. Nature, Lond. Ill: 387-8. Cleland, K. W. (1964). — History of the centrioles in bandicoot (Perameles) spermiogenesis. J. Anat. 98:487. Cleland, K. W., and Lord Rothschild (1959). — The bandicoot spermatozoon: an electron microscope study of the tail. Proc. R. Soc. B 150: 24-42. Duesberg, J. (1920). — Cytoplasmic structures in the seminal epithelium of the opossum. Carnegie Institute Contributions to Embryology. Vol. 9, pp. 47-84. Friend, G. F. (1936).— The sperms of the British Muridae. Quart. J. Micr. Sci. 78: 419-43. Furst, C. M. (1887). — Ueber die Entwicklung der Samenkorperchen bei den Beutelthieren. Arch. mikrosk. Anat. EntwMech. 30: 336-65. Hughes, R. L. (1964). — Sexual development and spermatozoon morphology in the male macropod marsupial Potorous tridactylus (Kerr). Aust. J. Zool. 12: 42-51. 139 MORPHOLOGY OF SPERMATOZOA 543 Jordan, H. E. (1911). — The spermatogenesis of the opossum {Didelphis virginiana) with special reference to the accessory chromosome and the chondriosomes. Arch. Zellforsch. 7: 41-86. KoRFF, K. VON (1902). — Zur Histogenese der Spermien von Phalangista vulpina. Arch, mikrosk. Anat. EntwMech. 60: 233-60. McCrady, E. (1938). — The embryology of the opossum. Am. Anat. Mem. 16: 1-233. Pearse, a. G. E. (1961).— "Histochemistry." (J. & A. Churchill: London.) Retzius, G. (1906). — Die Spermien der Marsupialier. Biol. Unlers. (N. F.) 13: 77-86. Retzius, G. (1909).— Die Spermien von Didelphis. Biol. Unfers. (N. F.) 14: 123-6. Selenka, E. (1887). — Studien ijber Entwickelungsgeschichte der Thiere. Das Opossum (Didelphis virginiana). Wiesbaden 1887, pp. 101-72. Simpson, G. G. (1945). — Principles of classification and a classification of mammals. Bull. Am. Mas. Nat. Hist. S5: 1-350. Sonntag, C. F. (1923). — On the myology and classification of the wombat, koala and phalangers. Proc. Zool. Soc. Lond. 1922: 683-895. Troughton, E. (1957). — "Furred Animals of Australia." 6th Ed. (Angus and Robertson: Sydney.) Waldeyer, W. (1902). — Die Geschlechtszellen. In "Handbuch der vergleichend und experimentellen Entwickelungsgeschichte der Wirbelthiere". Vol. 1, Pt. 1, pp. 86-476. Wilson, E. B. (1928). — "The Cell in Development and Heredity." (Macmillan: New York.) Wood Jones, F. (1924). — "The Mammals of South Australia." Pt. II. (Govt. Printer: Adelaide.) Explanation of Plate 1 Figures 1 and 2 are photographs of Heidenhain's iron-haematoxylin preparations from formalin-fixed epididymal material Fig. 1. — Phascolomis mitchelli, mature epididymal spermatozoon, lateral view. Fig. 2. — Rattus norvegicus, mature epididymal spermatozoon, lateral view. Fig. 3. — Pseudocheirus peregrinus, spermatozoon head, showing centrally placed acrosomal pit, dorsal view. Fig. 4. — Pseudocheirus peregrinus, epididymal spermatozoon, lateral view. 140 Hughes MORPHOLOGY OF SPERMATOZOA A Aust. J. ZooL, 1965, 13, 533-43 Plate 1 s 141 THE GENITAL SYSTEM AND THE FETAL MEMBRANES AS CRITERIA FOR MAMMALIAN PHYLOGENY AND TAXONOMY By H. W. Mobsman All systems of classification of natural phenomena are admittedly imperfect because these phenomena differ from one another in such infinitely variable de- grees that it is impossible to divide them into a system of groupings which do not frequently overlap or intergrade. It is also absolutely impossible consistently to select morphological criteria for a scheme of classification that will lend them- selves to the assignment of a series of ranks such as species, genera, and famiUes, in one group that ^vill be of parallel value to a similar series assigned to another related group. Yet in spite of these obviously insurmountable barriers to perfec- tion, convenience and orderliness in science demand that classifications be made, and that their categories be of as nearly parallel value as possible. Most biologists believe that the best basis for classification is phylogeny, that is, the evolutionary or genetic relationships within and of the group. This is so widely accepted that practically all other forms of classification are considered "artificial." Although artificial classification is often temporarily necessary, as in tentatively fitting a poorly understood entity into a general scheme, still a natural or phylogenetic classification should always be the ultimate aim. Criteria for classification of any group should therefore be characters of phylogenetic significance. Furthermore there should be some way to evaluate the relative significance of one set of criteria in relation to another set; for in- stance, dentition as compared to skull proportions; or pelage as compared to baculum. There should also be some method for determining whether a given set of characters is suitable for separating the lower categories such as species and genera, or the higher categories such as famihes and orders. The more conservative characters obviously will be of greater value in char- acterizing the higher categories, while the less conservative will only be of use in the lower categories. The presence of the fetal membranes (amnion, chorion, yolk sac, and allantois) is a highly conservative character appearing in all Am- niota. Yet certain aspects of the finer morphology of the placenta, such as whether it is villous, trabecular, or labyrinthine, are of use only in characterizing the families within a suborder, for example the Anthropoidea. This is true be- 289 142 290 JOURNAL OF MAMMALOGY Vol. 34, No. S cause parallel transitions in placental morphology are known to occur inde- pendently in other orders not closely related to the Anthropoidea or to one another. These same aspects of finer morphology are, however, too conservative to be of much use in separating genera or species, there being very slight or no differences in this character within single families or genera. The relative conservatism of biological characters is based on two entirely distinct and independent thhigs. The first is the human factor, or simply the question of generality versus particularity of concept. The greater the breadth or inclusiveness of a character concept the more conservative it is within its particular field. For example, the character "bony skeleton" is much broader and therefore much more conservative than the character "cranium", w'hile the latter is in turn much broader and more conservative than "zygoma." We can and do quite properly select the generalized and broader characters as criteria for separating the higher categories, and the more narrow and particular ones for the lower subdivisions. Superimposed on this, much more complex, and equally as important, is the second or biological factor in determining conservatism. This may be expressed as the degree to which a character has been subjected to natural selection. Assuming equal, intrinsic genetic factors, it is certainly true that the more intimately a character is related to the environment the more rapid and exten- sive will be its environmental adaptations during the course of evolution. The main thesis of this discussion rests on this assumption. It is maintained that the structural characters of the reproductive tracts and fetal membranes of mammals, because they have been largely independent of environmental selec- tion, show relatively little evolutionary divergence as compared to most other organ systems. This is not the place to discuss, statistically, results to be ex- pected in the evolution of a genetically variable system subjected to environ- mental selection in contrast to one subjected to very little and indirect en- vironmental action. However, it can be readily understood that without the selective effects of differing environments, variations would tend to be submerged because of continuous selection for one constant set of environmental conditions. A race under these conditions would become more and more stabilized and specialized, but no new divergent races would arise from it. But this is not the situation in regard to a genetically variable system, such as the fetal membranes that are practically isolated from any direct selective effects of external environ- ment. In such a system, in this case the fetal membranes, evolution is free to proceed along almost any line so long as it meets certain vital requirements: (1) the supplying of a means of maintaining an embryo within the mother's uterus until mature enough to be bom; (2) structural conditions which will allow the uterus to resume a normal nonpregnant state after delivery, so that another pregnancy may ensue. Certainly these fundamental requirements are far from simple in either a structural or physiological sense; they involve extremely complex mechanisms. But the point is that not only is the intrauterine en\'ironment relatively con- stant in mammals generally, but so are the requirements for maintenance of an 143 Aug., 195S MOBSMAN— TAXONOMIC CRITERIA 291 embryo and for restorability of the uterus to the condition for beginning another pregnancy. What then should one expect of the evolution of an organ system, such as the fetal membranes, largely isolated from the external environment and charged with a highly complex job to do, but one which is fundamentally the same in all members of the group, in this case the Subclass Eutheria? It seems that, regard- less of rate of genetic variation of the system, divergence would be relatively slow and narrow, and that intergradation would be the rule, there being slight environmental selection, and little race isolation due directly to adaptive varia- tions in this system itself. But while this type of evolution would be taking place in an environmentally independent organ system, the group of organisms in which the organ system existed could nevertheless be undergoing the usual evolutionary course in its other organ systems: diverging widely in adaptation to environmental niches old or new; losing species or whole major groups by extinction; in short, becoming the widely divergent, often aberrant or isolated groups that are characteristic of mammals at the present time. Such an environmentally isolated system must then be very conservative in its characters as compared to those organ systems related closely to the environ- ment and therefore subject to intense, adaptive evolution. The characteristics of such a system should vary relatively little between major groups, and almost none between closely related minor categories. This lack of divergence should make it possible to detect common characteristics among groups widely sepa- rated from one another in other characteristics, whether this be due to adapta- tions to divergent environments, or to extinction of intergrading groups. Con- versely, dissimilarity in characters of a conservative organ system should be good evidence of lack of close phylogenetic relationship. It is generally conceded that similarity of more conservative characters be- tween two forms is better evidence of close relationship than similarity of less conservative characters. Whether lack of conservatism is due to more intensive environmental selection or to a combination of this with such things as more intensive sexual selection, or greater innate potential of genetic variability, there is more chance that similarities in less conservative characteristics are due to evolutionary parallelism or convergence. In estimating phylogenetic relation- ship it is therefore extremely important to know which are the more conserva- tive organ systems and to give their characters the greater weight. It is also im- portant to evaluate the various characters of such an organ system to determine which are the more conservative within the system. The more conservative will serve to characterize the higher categories, the less only the lower ones. The foregoing discussion largely reiterates generally recognized principles which have been used for years by students of evolution and taxonomy. How- ever it seemed best to restate them here since the chief purpose of this paper is to point out that the reproductive tracts and fetal membranes of mammals are conservative systems which have never been adequately used in phylogenetic and taxonomic studies of this group. It will be shown that the relative conserva- tism of the various features of these organ systems can be estimated and the 144 292 JOURNAL OF MAMMALOGY Vol. S4, No. S characters used as criteria in proper relation to their significance. It is now best to proceed to specific examples. The first concerns the male genital tract of the Sciuridae. In 1923, Pocock pointed out the marked difference between the male external genitalia of Tamiasciurus and Sciurus and separated the two genera on this basis. In 1932, Mossman, Lawlah, and Bradley published a more detailed study of the tracts of Sciurus carolinensis, S. niger, Tamias striatus, Citellus tridecem- lineatus, Glaucomys volans, and Tamiasciurus hudsonicus. They showed that the typical sciurid male tract as seen in all these genera (except Tamiasciurus), and as also described by KroUing (1921) in Sciurus vulgaris, is characterized by a unique arrangement of the ducts of the bulbo-urethral glands. Tullberg (1899) gave enough information on several other genera to indicate the probability that they also have this same character, and Oudemans (1892) made it fairly clear that Petaurista petaurista is essentially like Sciurus. So, with the exception of Tamiasciurus, other sciurids so far as known have a pair of large bulbo- urethral glands drained by a pair of voluminous ducts which, upon entering the sheath of the corpus cavernosum urethrae in the bulb region, become highly modified to form another accessory organ, the bulbar gland. This whole com- plex is then drained by a single long glandular duct, the penile duct, w^hich lies ventral to the urethra throughout almost the entire length of the corpus cav- ernosum urethrae, finally entering the urethra at approximately the base of the glans. The author has examined some other genera, including Marmota and Heliosciurus and has found them also fundamentally like this. At the same time we showed that Tamiasciurus, the red squirrel or chickaree, possesses an entirely different type of bulbo-urethral gland and duct. It has no bulbar gland and no penile duct. It does have a urethral sinus in the bulb much like that of some Muridae. Later it was shown that the female tract of the chickarees is also unique, the vagina being extremely long and coiled while that of all other Sciuridae studied is short and broad (Mossman, 1940). Other peculiar features of the male tract of the chickarees, such as the long filiform penis and the absence of a baculum, were also pointed out at that time. (Layne (1952) has shown that Tamiasciurus does possess a minute baculum averaging only .26 mm, in length in adults. Pocock and the author both failed to note this structure.) A recent, and as yet unpublished manuscript by Mr. M. R. N. Prasad of Central College, Bangalore, India entitled, "Male genital tract of two genera of Indian squirrels," presents excellent descriptions of the palm squirrel, Funam- bulus palmarum palmarum Linn., and of the giant Malabar squirrel, Ratufa indica maxima Schreb. Ratufa has the typical sciurid tract, but F. palmarum is hke Tamiasciurus in having minute Cowper's glands and no bulbar gland or penile duct. Obviously this indicates close relationship between this species and the Tamiasciurinae, and poses a number of interesting questions in regard to sciurid phylogeny and geographic distribution, to say nothing of the doubt it throws on the present taxonomy of the whole group. Although the nature of the reproductive tracts of most of the genera of squir- rels is still unknown, enough knowledge is available to make it seem very prob- 145 Aug., 195S MOBSMAN— TAXONOMIC CRITERIA 293 able that instead of a wide range of types within the family, there are at least two fundamentally different types. This seems probable as it is definitely known that at least one genus of three of the six tribes that Simpson (1945) includes under the subfamily Sciurinae has the typical male sciurid tract; and that two genera of his second subfamily, Petauristinae, also are typical. One can raise the question then as to the logic in placing the chickarees and Funambulini as tribes of the subfamily Sciurinae when the reproductive tracts of Tamiasciurus and F. p. palmarum are so different from others of this group. Also, in view of Prasad's work, F. p. palmarum and Ratufa should not be in the same tribe with one another. Furthermore, it is illogical to place the chickarees and Funambulus in these subgroups and at the same time to put Petaurista and Glaucomys in a separate subfamily, although their genital tracts are almost identical to that of Sciurus. Obviously the flying squirrels (Petauristinae) have been separated on the basis of petagial characters; but these must be highly subject to environ- mental selection, hence nonconservative. Very similar petagial characters have been developed in certain marsupials (Acrohates and Petaurus), and in other Eutheria, namely in the Dermoptera, and in the Anomaluroidea among the ro- dents. Of course one could ask if similar reproductive tract characters may not also have developed in widely unrelated groups. There is no evidence that this is true, but it must be admitted that the data are insufficient. This in itself should be a challenge, to those who ask the question, to make an effort to gather the information. There may be those who will still fall back on the fundamental question of whether the genital tract characters are more conservative than others in the Sciuridae. If they will allow the ruling out of the genus Tamiasciurus and the species Funambulus p. palmarum as members of the family, then one can cite the fact that the genital tracts of Sciurus, Tamias, Citellus, Marmota, Ratufa, and Glaucomys are the same in fundamentals, and indeed very similar even in details, while there is great divergence between these various genera in body form, pelage, feet, ears, skeleton, skull, and teeth. If one is not willing to allow the ruling out of the chickaree group, and F. p. palmarum, then in the face of the apparent aberrance of these he is bound to withhold judgment until someone can show whether or not there are intergrading forms, so far as male genitals are concerned, between them and the typical sciurids. If there are, then perhaps the genitals are not as conservative as they now seem. There are a few minor characters of male genital tracts that are variable enough in some groups to be of use in separating species of a single genus. Howell (1938) pubhshed a plate showing the bacula of several species each from the genera Eutamias, Citellus, and Sciurus. There are fairly obvious intrageneric differences between these bones within each of the three genera, although oc- casionally in two species they are practically identical (Sciurus carolinensis and niger). It is probable that there are other features of the male genitals which would show intrageneric differences: certainly differences in the glans penis cor- related with those of the bacula would be expected. However, appreciable intra- generic, species differences in the internal genitaha of either the male or female 146 294 JOURNAL OF MAMMALOGY Vol. S4, No. S have never been noted by the author in the sciurids or other groups, but inter- generic differences are the rule, and are often rather extensive; for example between Tamias, Citellus, and Sdurus (Mossman, Lawlah and Bradley, 1932). To the author's knowledge, no attempt has ever been made to compare the reproductive tract at the level of the higher categories. It seems doubtful that such clearcut indications of relationship would be found as occur in the case of the fetal membranes, where, for example, considerable affinity is shown be- tween the Artiodactyla, Perissodactyla, and Cetacea; and between the Fis- sipedia and Pinnipedia. Before such a study would be sound, a thorough investi- gation of the various genera and families of several well-defined orders should be made, in order to evaluate the conservativeness of the various genital tract features. The female genital tract characters are probably about as equally conserva- tive as those of the male, but those of the female are in general more difficult to observe and define. A good example of this is the os clitoridis or baculum, which in recent years has been demonstrated in females of several species, leading one to expect that, where present in the male, it probably also occurs in the female. This female element may show intrageneric differences like its male homolog, but the fact that it is usually so small and so obviously rudimentary argues against its value as a character for taxonomic purposes. The other characters of the female external genitalia are also relatively indefinite and difficult to ob- serve. Female internal genitalia show a few very definite features which could be of considerable use, but they are almost always too conservative for intra- generic taxonomy, in fact in most cases they are of use only in separating groups higher than genera. For instance, the form of the uterus, oviduct, and ovary is almost identical in the Cervidae and Bovidae. The Mustelidae, Procyonidae, and Ursidae have a peculiar configuration of the oviduct in relation to the ovary and ovarian bursa that is highly characteristic and differs very little be- tween the three families. The Heteromyidae and Geomjddae likewise have a characteristic oviduct pattern which is practically identical in the two groups. By and large then, it may be said that the characters of the female internal genitalia are very conservative, but usually rather obscure and difficult to observe. Comparative studies of the microscopic anatomy of the ovaries have revealed no case where obvious intrageneric differences exist, except those due to differ- ences in body size of the species. However some intergeneric differences in microscopic structure do occur: for example; ovaries of mature females of Syl- vilagus and Lepus are rather easily distinguished from one another, but those of Lepus and Oryctolagus would be difficult; Citellus and Sdurus are separable, but Sdurus and Tamias are alike, except for size. When one reaches the higher categories such as families and orders, then microscopic features become char- acteristic for each group. All the mustelids examined have a very similar and tjrpical interstitial cell pattern: this includes Maries americana, Taxidea taxus, Mephitis mephitis, Spilogale interrupta, and Mustela vison, dcognani, frenata, and putorius. There are even examples of striking similarity in microscopic structure 147 Aug., 1953 MOBSMAN— TAXONOMIC CRITERIA 295 between related families, at least so far as the few genera of each studied are an indication. Examples of this are Ursidae and Procyonidae, Cervidae and Bovidae, and Erethizontidae and Dasyproctidae. Let us now turn our attention briefly to the fetal membrane characters of the Sciuridae (Mossman, 1937; Mossman and Weisfeldt, 1939). It is not necessary to enter into a detailed description of the sciurid membranes and placentation, as the main point to be made is that these are even more conservative characters than the male genital system of the group. So it is not surprising that in all the genera studied, including Tamiasciurus, the fetal membranes are practically the same, differing only in minor details. These include all the genera mentioned in connection with the male genitals, except Heliosciurus and Petaurisia, and two others in addition, Cynomys, Xerus (Rau, 1925). In fact the membranes and placenta of Aplodontia are distinctly sciuroid, although the male genitals lack the bulbar gland and penile duct. It appears then that the criteria of the male and female reproductive tracts and the fetal membranes could be applied to advantage in determining the phylogenetic relations of the Sciuridae, and consequently in estabhshing a more logical taxonomy of this group. The fetal membrane characters are even more conservative criteria than those of the male or female genital tracts. The application of the genital tract and fetal membrane criteria in the phy- togeny and taxonomy of the Sciuridae has been discussed first, since it is the only group in which the author has made reasonably comprehensive studies of both systems. Little has been said thus far of the evaluation of specific criteria, as the studies of the male tract in such a small group do not lend themselves to adequate analysis. The author has however made extensive comparative studies of mammalian fetal membranes and believes, that for them, the basis is broad enough to enable clearly reliable estimates to be made of the relative value of the different characters involved. Since this subject was presented fully in his 1937 monograph, it is not necessary to repeat the details here. Suffice it to point out that all the descriptions of the fetal membranes of specific species which have appeared in the literature since then, and all of the numerous additional observations of the author, bear out the thesis proposed at that time. The facts derived from these studies can be stated very simply. 1. In every group of mammals, high or low in category, in which the members can be clearly related to one another on the basis of total anatomical similarity, their fetal membranes are fundamentally similar, showing far less divergence than do other characters. These groups are perhaps best illustrated by the orders Lagomorpha, Rodentia, Carnivora, and Artiodactyla; by the suborders Micro- chiroptera, Lemuroidea, and Anthropoidea; by the families Tenrecidae, Sorici- dae, Talpidae, Cebidae, Cercopithecidae, Pongidae, Das5T)odidae, Sciuridae, Heteromyidae, Geomyidae, Muridae, Canidae, Mustelidae, Felidae, Cervidae, and Bovidae; and by numerous genera in these and some other families. All these groups consist of anatomically closely related forms and all have even closer fetal-membrane affinities. They are used as illustrations because enough is known about the fetal membranes of enough members of each group to make 148 296 JOURNAL OF MAM^IALOGY Vol. S4, No. S it reasonably certain that the above statement is true. There are other ana- tomically homogeneous groups such as the Megachiroptera, Cetacea, and Peris- sodactyla where the limited data available points in the same direction. This is particularly true of the data on the species of numerous genera of the groups already mentioned. 2. In many groups of mammals made up of subgroups widely divergent or of uncertain affinities there are fundamental differences in the fetal membranes. This is true of the Insectivora; the Chiroptera, where there is wide divergence between the membranes of the two suborders (Microchiroptera and Mega- chiroptera); the Primates, where the membranes of Lemuroidea and Tarsoidea differ widely from one another and from the Anthropoidea; and the Edentata, where the Bradipodidae and Myrmecophagidae are probably much alike, but differ markedly from the Dasypodidae. The Pholidota, often included in the order Edentata, differ fundamentally from all others of this order. 3. The membranes of certain groups commonly separated on anatomical grounds, but known to be somewhat related, are often so similar that the wide- ness of separation does not seem justified. This is true of the Pongidae and Hominidae, the Lagomorpha and Rodentia, and of the Perissodactyla and Artiodactyla. 4. There are fundamental similarities between certain aberrant groups and other groups to which they have not been supposed to be clearly related. Strik- ing examples of this are the strong resemblances between the membranes of the Brady podidae and the Anthropoidea; between the Lemuroidea, Pholidota, Ceta- cea, and Sirenia and those of the Perissodactyla and Artiodactyla; and between the Dasypodidae, Rodentia, Microchiroptera and Soricoidea. More data than are at present available are certainly necessary to warrant drawing more than very tentative conclusions as to the significance or non- significance of these facts, but it is interesting to note that there are many points of anatomical resemblance between lemurs and the hoofed animals, and between the sloths and anthropoids. Certainly taxonomists should not close their minds to the possibility that lemurs are an arboreal line derived from stock ancestral to the hoofed animals, and that their relatively slight resemblance to anthropoids is due to retention of primitive characters in both lines, and to convergence and parallelism effected by adaptation of both to an arboreal habitat. Nor should one close his mind to the possibility that the sloths repre- sent highly specialized edentulous forms derived from the same ancestral stock as the anthropoids. Now that the general method of application of the reproductive tract and fetal membrane criteria to the taxonomy and phylogeny of mammals has been described and illustrated, one must clear up the important point of the evalua- tion of the criteria used. The basis for this is the observation described as number 1 above ; namely, that in every group of mammals in which the members can be clearly related to one another on the basis of total anatomical similarity their fetal membranes are fundamentally similar, showing far less divergence than do other characters. This makes it possible to compare the variability of differ- 149 Aug., 19BS MOSSMAN— TAXONOMIC CRITERIA 297 ent features of the membranes within the group as a whole and within its various lesser taxonomic categories, as was done by the author some years ago (Moss- man, 1937). Those characters which are consistently constant throughout all the subgroups of a major category certainly are the more conservative ones of that group. If the same characters are also constant, but not necessarily alike, throughout several well established orders, then they are certainly characters conservative enough to be used for establishing phylogenetic relationships be- tween orders. If they are constant within single families, but vary between differ- ent families of an order, then they are only conservative enough to establish relationship between families. This general principle can be applied to all cate- gories. Obviously one must choose test groups in which the taxonomy is quite clear and definite, and must compare parallel, major groups in which the minor categories are also reasonably parallel in value. This is obviously arguing in a circle, and thus a dangerous practice if not tempered with good judgment and a reasonable scientific conservatism on the part of the person using it. However it is the best method available, and far superior to making no attempt at evalua- tion. One further argument in favor of the fetal membranes as criteria for phy- logeny must be stated. This point seems even more important than that of their conservatism. In fact, coupled with their conservatism, it makes them the most ideal of all anatomical criteria for recent forms. This is the fact that the history of the development of the fetal membranes of a species is the history of a com- plete, complex, and, structurally, highly independent organ system, from its inception during cleavage to its complete functional maturity, old age, and death at the time of birth of the young. We are therefore dealing with the complete life history of an organ system carried out in the relatively constant environment of the uterus, thus almost completely isolated from adaptational demands of the varying external environment. Other criteria conmionly used do not offer this overall picture of the individual. The fetal membranes and their develop- ment are comphcated, but far less so than the total history of all the organ sys- tems ordinarily used as criteria. Their conservatism, and the total develop- mental picture that they give, render them the most ideal of all organ systems for phylogenetic and taxonomic studies of recent mammals, SUMMARY When compared with the organ systems ordinarily employed as criteria for taxonomic and phylogenetic studies of manmials, the characters of the male and the female reproductive systems and of the fetal membranes are the more conservative. This is apparently due to the relatively minor role that adaptation to external environment has played in the evolution of these systems. Little effort has ever been made to apply male and female genital tract char- acters to such studies of mammals, but a limited consideration of them by the author indicates that they are less conservative than the fetal membranes. They furnish characters that are of use in the study of the interrelationships of genera and families, and, in some cases, even of species. Whether or not they would be 150 298 JOURNAL OF MAMMALOGY Vol. S4, No. S reliable in showing affinities between higher categories, such as orders and sub- orders, is unknown. On the contrary, the fetal membranes are so conservative that clear-cut inter- specific or even intergeneric differences seldom exist. Furthermore interorder and interfamily fetal-membrane similarities often demonstrate relationships between these major groups. This conservatism, plus the fact that the complete life cycle of the fetal membranes takes place during embryonic development and is therefore much more easily studied than that of any other organ-system ontogeny, makes this system the most ideal of all criteria for the study of phy- logenetic interrelationships of recent mammals. Acknowledgement. — This study was aided by grants from the Wisconsin Alumni Research Foundation. LITERATURE CITED Howell, A. H. 1938. Revision of the North American ground squirrels with a classifica- tion of the North American Sciuridae. U. S. D. A., N. Amer. Fauna, No. 56: 1-256. Kbolling, O. 1921. Die akzessorischen Geschlechtsdriisen und mannlichen Kopula- tionsorgane von Sciurus vulgaris. Zeitschr. f. Anat. u. Entwick., 61: 402-438. Layne, J. N. 1952. The os genitale of the red squirrel, Tamiasciurus. Jour. Manam., 33: 457-459. Mobsman, H. W. 1937. Comparative morphogenesis of the fetal membranes and ac- cessory uterine structures. Contribs. Embry., 158; Publ. 479, Carnegie Inst, Washington, 129-246. 1940. What is the red squirrel? Trans. Wis. Acad. Sci., Arts & Letters, 32: 123-134. Mobsman, H. W., J. W. Lawlah, and J. A. Bradley. 1932. The male reproductive tract of the Sciuridae. Amer. Jour. Anat., 51: 89-155. MossMAN, H. W., AND L. A. Weisfeldt. 1939. The fetal membranes of a primitive ro- dent, the thirteen-striped ground squirrel. Amer. Jour. Anat., 64: 59-109. Oudemans, J. T. 1892. Die accessorischen Geschlechtsdriisen der Saugetiere. Natuurkun- dije Verhandelingen van de Hollandsche Maatschappij der Wetenschappen. 3de Verz., Deel 5, 2de Stuk., 1887-1892. PococK, R. I. 1923. The classification of the Sciuridae. Proc. Zool. Soc. Lond., 1: 209- 246. Rau, a. S. 1925. Contributions to our knowledge of the structure of the placenta of Mustelidae, Ursidae, and Sciuridae. Proc. Zool. Soc. London, 1925: 1027-1070. Simpson, G. G. 1945. The principles of classification and a classification of mammals. Bui. Amer. Mus. Nat. Hist., 85: 350 p. Tullberg, T. 1899. Ueber das System der Nagetiere. Nova Acta Reg. Soc. Upsala, ser. 3, 18: 1-514. Department of Anatomy, University of Wisconsin, Madison. Received December 26, 1952. 151 MORPHOLOGY AND PHYLOGENY OF HAIR By Charles R. Noback* Department of Anatomy, College of Physicians and Surgeons, Columbia University, New York Hair is a structure found exclusively in mammals. With this in mind, Oken named the Mammalia, Trichozoa (hair animals), and Bonnet (1892) named them Pilifera (hair bearers). Of the many aspects of morphology and phylogeny of hair, only four will be discussed. These include (1) the principle of the arrangement of hairs in group patterns, (2) the types of hair and their relation to the principle of the group pattern, (3) a brief analysis of the structural elements of hair and their relation to the types of hair, and (4) the phylogeny of hair, with some remarks on (a) the relation of hair to the epidermal derivatives of other vertebrate classes and (b) aspects of the phylogeny of the hair and wool of sheep to illustrate that marked dififerences in hair coats exist between closely related animals. Hair is the subject of a voluminous literature. Toldt (1910, 1912, 1914, and 1935), Danforth (1925a), Pinkus (1927), Pax and Arndt (1929-1938), Trotter (1932), Lochte (1938), Smith and Glaister (1939), and Stoves (1943a) discuss the problem of mammalian hair in general. Wildman (1940), von Bergen and Krause (1942), and the American Society for Testing Materials (1948) discuss the problem of fiber identification as applied to textiles. Principle of the Group Pattern of Hairs In the only extensive survey of the grouping of hair in mammals, DeMei- jere (1894) documented the concept of the group pattern of hair (figures 1-6). Unfortunately, the few studies on this phase of the problem since that time have not fully exploited the implications of this concept. DeMei- jere concluded that hairs are mainly arranged in groups with the pattern of 3 hairs — with the largest hair in the middle — as the basic pattern. The concept of the basic trio as the primitive condition is accepted as an ade- quate working hypothesis by Wildman (1932), Galpin (1935), Hofer (1914), Gibbs (1938), Hardy (1946), and others. DeMeijere described 8 patterns: (1) 3 or less hairs behind each scale of the tail (as in the opossum, Didelphis marsupialis), (2) more than 3 hairs behind each scale of the tail (as in the rodent, Loncheres [Echimys] cristata), (3) 3 hairs (as in the back of the marmoset, Midas rosalia), (4) more than 3 hairs arranged in a regular pattern with some of greater diameter than others (as in the back hairs of Loncheres [Echimys] cristata in figure 3), (5) several hairs composed of a number of fine hairs and one coarse hair (as in the back of the dog, Canis familiaris, in figure 5D), (6) several hairs composed of a number of fine hairs and one isolated coarse hair (as in the back hairs of the mouse, Mus decumanus, in figure 6D), (7) scatterings of fine hairs with no apparent * The author wishes to thank Dr. Margaret Hardy, Division of Animal Health and Production, Sydney, Australia, for her valuable suggestions. 476 152 Noback: Morpholog}^ and Phylogeny 477 arrangement and a few intermingled coarse hairs (as in the back hairs of the cat, Felis domesticus in figure 4D), and (8) hairs in irregularly scattered groups (as in the back hair of the raccoon, Procyon cancrivorus) . Dawson (1930) does not completely agree with DeMeijere's pattern in the guinea pig. She found variations in the pattern and no correlation be- tween the size of hair and the arrangement of the hairs in each group. Histological study frequently shows follicle grouping which was not appar- ent to DeMeijere when he was examining only the skin surface, e.g., in Felis domesticus (see Hofer, 1914). This indicates that analyses of the group pattern of hairs are needed in both common laboratory mammals and mammals in general. In addition, DeMeijere analyzed the formation of the patterns by ex- amining the skins of animals during their development (figures 4-6). This phase of the problem has been extended to include a study of the ontog- eny of the arrangement of hair follicles in sheep (Wildman, 1932, Galpin, 1935, and Duerden, 1939), in the cat (Hofer, 1914), in marsupials (Gibbs, 1938, Stoves, 1944b, and Hardy, 1946), in the mouse (Calef, 1900, Dry, 1926, and Gibbs, 1941) in the rat (Frazer, 1928), and in a number of mam- mals (Duerden, 1939). The terminology used by these authors in this problem is summarized in table 1 (adapted from Wildman and Carter, 1939 and Carter, 1943). UtiHzing the terminology of Wildman and Carter, 1939, the following is a brief statement of the relation of the fiber generations. The first follicles to differentiate are the central trio follicles (figure 7). If these follicles appear at two different times as in the opossum (Gibbs, 1938), then the follicles are called "primary X" and "primary Y." The essential point is that each of these primary follicles will be the central follicle of different hair groups. Later in development, other follicles of the hair group differentiate in relation to these central trio follicles. The trio is formed when two follicles are differentiated lateral to the primary follicles (figure 8). The lateral follicles associated with primary X and primary Y are called re- spectively "primary x" and "primary y." If only one lateral follicle is formed adjacent to a primary follicle (X or Y), then a couplet follicle is formed. If no lateral follicles differentiate, a primary follicle (X or Y) is called a "solitary follicle." Later, another generation of follicles is differ- entiated— the secondary follicles. In the opossum (figure 9), these secondary follicles are located between the central trio follicle and the lateral trio follicles. The ontogenetic studies of follicle arrangement have added confirmatory evidence to DeMeijere's basic concept that in mammals there is a universal and regular grouping of hair follicles (Hardy, 1946). In general, the early differentiating follicles (central trio follicles) form the coarse overhair, while the late differentiating follicles (lateral trio follicles and secondary follicles) form the fine underhair. Lateral trio follicles sometimes at least produce overhair like that of the central fol- licles {e.g. in sheep) or intermediate types such as awns, which are classified by Danforth (1925a) as overhair. In Ornithorhynchus analinus (Spencer and Sweet, 1899) and many marsupials (Gibbs, 1938, Bolliger and Hardy, 153 478 Annals New York Academy of Sciences 1945, Hardy, 1946), however, the lateral trio fibers are indistinguishable from those of secondary follicles, so it is difficult to place them in either the "overhair" or the "underhair" category. Figures 1-9 {see facing page). Spencer and Sweet (1899) claimed that, in monotremes, each group of follicles was differentiated by budding from the central follicle. This has not been described in marsupials or in eutherians, in which the follicles arise independently as epidermal downgrowths. Monotremes and mar- 154 Noback: Morphology and Phylogeny 479 supials have in common the fact that a follicle group typically contains a large central follicle with a sudoriferous gland, and two or more clusters of smaller lateral follicles (Spencer and Sweet, 1899, Gibbs, 1938, Hardy, 1946). This arrangement is also found in some eutherians, such as the cat (Hofer, 1914) and dog (Claushen, 1933). In the cat and a few other eutherians, the first-formed lateral follicles (primary x and y of the classification of Wild- man and Carter, 1939) produce hairs intermediate in type between those of the central and the other lateral follicles. There are other eutherians in which the lateral primary x and y fibers are still more like the central pri- mary X and Y fibers, as in the pig (Hofliger, 1931) and the sheep (Carter, 1943). Except in the rodents, there is always a sudoriferous gland opening into the central primary X or Y follicle (Hardy, unpublished data). Many animals, such as the pig and sheep, also have a sudoriferous gland opening into each primary x and y follicle, but others do not (Duerden, 1939). Some of the eutherians have only primary follicles in their skin, each with a sudoriferous gland. Findlay and Yang (1948) showed that this is the arrangement in cattle, and the same is probably true in horses and in human head hair (Hardy, unpublished observations). Types of Hair DeMeijere's analysis leads to the classification of hair types by Toldt (1910 and 1935) and by Danforth (1925a). Many details of the hair types in many species of animals and the variations of the structure of these types are described, illustrated, and bibliographically annotated by Toldt (1935) and Lochte (1938). Types of Mammalian Hair (after Danforth, 1925a) 1. Hairs with specialized folhcles containing erectile tissue. Large, stiff hairs that are preeminently sensory. They have been variously designated as feelers, whiskers, FiGDRES 1-9 (see opposite page) . Figure 1. The trio hair group pattern on the back and tail of the marmoset, Midas rosalia (after DeMei- jere, 1896). .Ml hairs have similar diameters. Figure 2. The hair group pattern of more than 3 hairs with some fibers of greater diameter than other fibers on the back of the paca, Coelogenys paca (after DeMeijere, 1896). Figure 3. The hair group pattern of more than 3 hairs with some fibers of greater diameter than other fibers on the back of the rodent, Loncheres (Echimys) cristata (after DeMeijere, 1896). Figure 4. Ontogeny of a hair group on the back of the cat, Felis domesticus. A, from a newborn animal; B and C, from an older animal; and D, from an adult animal (after DeMeijere, 1896). Figure 5. Ontogeny of a hair group on the back of the dog, Canis Jamiliaris. A, from an embryo dog; B, from a newborn animal; C, from a young dog; and D, from an adult animal (after DeMeijere, 1896). Figure 6. Ontogeny of a hair group on the back of the mouse, Mus decumanus. A, from a 7 cm. long ani- mal; B, from a 9 cm. long animal; C, from a 12.5 cm. long animal; and D, from an adult animal. (Figures 4, 5, and 6 illustrate that the follicle of the first hair to erupt (A) will be the follicle of the coars- est hair of the hair group in the adult. The type of hair group pattern in the adult (D) in each figure is noted in the text. The X in the diagrams marks the location of erupting follicles.) Figure 7. The primary follicles X (the more differentiated follicles) and the primary follicles Y (the less differentiated follicles) in the transverse section of skin of a 12.5 cm. Australian opossum embryo [Tricliosurus vulpecula). Follicles are scattered irregularly. (After Gibbs, 1938.) Figure 8. Two new follicles (primary x or primary y) have become grouped with each previously differen- tiated follicle (primary X or primary Y) to form the typical trio arrangement. The trio would be either primary x, primary X, primary x or primary y, primary Y, primary y. Transverse section of skin of a 15.0 cm. Australian opossum embryo (Trichosurus vulpecula). (After Gibbs, 1938.) Figure 9. Two secondary follicles have added to each ffio group to form a 5 follicle group. The secondary follicles differentiate between the primary X (or Y) follicle and the primary x (or y) follicles. The five group would be either primary x, secondary follicle, primary X, secondary follicle, primary x or primary y, second- ary follicle, primary Y, secondary follicle, primary y. Transverse section of skin from 20.0 cm. Australian opossum emciryo (Trichosurus vulpecula). Note presence of a dermal capsule surrounding each 5 follicle group. (After Gibbs, 1938.) (In figures 7, 8, and 9, the terminology of Wildman and Carter (1939), noted in the text, is used.) 155 480 Annals New York Academy of Sciences n < < S S < > o o ►J o g s U i-l o o tt< (i w to Ji s ]o iy in ^ir 1 ]y 1911) {1914 1- 1^ a d B B 1 Toldl { ani Hofer 03 <5 iTl m s ^_^ >> .2 ~ (vj -Ci, >. c5 S: fVN JO ■^ m -c i> — . OJ So! s; 2- 0) :^ C.S Si^ :i>° ju _aj Qj 0; _U (J • S^ fi. u ^O ■S-5 1 11 "o "o o^^ ^^~ «*-• X>^ X >. __^ t/i ^ jj to _u <-i !2 "o c ' >> ^ f^» [/) rt C/i >, in Oj C o- b: i; — a> t- o U ,- o E o •S^ OJ ^ — •' c ~ c -^ rt IJ "S "o o "o o o 3 "O C.>*. (A. — *-* *-*-. CT t3 > 5 "cS iS a; « K G;^ (J JS^ J5 t« _a; C\ _o ^ 0\ _u .2 1 o rden {she 1^ 'u, *i .2 cS a; ::= lA o a (5 i ?j^ ^a « >> + + 1- o X>^ X >, y.> X> o! >. X >. >\ tt >k >^ 1- ^ i- i^ ^ ^ ■■ — ' rt 05 rt a a d rt a fe .E S E E EB B B :^ 'C 'j-i u ;- *C 'C U k. a C- c a c- r-s ^ p ^ C3 TJ -3 tn ' — ' tn t"! t/i >. t« • — ; S •— " TOO 3i _i. rt^ O Q sis Z .o C- o *_; o »^ •— • i;; >> S 11 cBS S ^ o -5 c o u Lh u 0-1 m a. o -o < 156 Noback: Morphology and Phylogeny 481 sensory hairs, sinus hairs, tactile hairs, vibrissae, etc. They occur in all mammals except man, and are grouped by Botezat (1914) (Pocock, 1914) essentially as follows: (1). Active tactile hairs — under voluntary control. (2) Passive tactile hairs — not under voluntary control. (a) Follicles characterized by a circular sinus. (b) Follicles without a circular sinus. 2. Hairs with follicles not containing erectile tissue. The remaining types of hair, most of which are more or less defensive or protective in function. In many cases, the follicles have a good nerve supply, endowing the hair with a passive sensory function as well. These hairs are grouped here according to their size and rigidity. (1). Coarser, more or less stiffened "overhair," guard hair, top hair. (a) Spines. Greatly enlarged and often modified defensive hairs, quills. (b) Bristles. Firm, usually subulate, deeply pigmented, and generally scattered hairs. "Transitional hairs" (Botezat, 1914), "Leithaare" (Toldt, 1910), "protective hair," "primary hair," "overhair." This group also includes mane hairs. (c) Awns. Hairs with a firm, generally mucronate lip but weaker and softer near the base. "Grannenhaare" (Toldt, 1910), "overhair," "pro- tective hair." (2). Fine, uniformly soft "underhair," "ground hair," "underwool." (a) Wool. Long, soft, usually curly hair. (b) Fur. Thick, fine, relatively short hair — "underhair," "wool hair." (c) Vellus. Finest and shortest hair — "down," "wool," "fuzz," "lanugo." (Danforth, 1939). The following comments supplement the above classification. The guard hairs are listed in a series from greater to lesser rigidity (in order: spines, bristles, and awns). There are many intergrade hairs between the typical bristle and the typical awn and between the typical awn and the typical fur hair (figures 10, 11, and 12). The tactile hairs have a rich nerv^e supply, while the roots of some are encircled by large circular sinuses containing erectile tissue. When the pressure in the circular sinus is increased the hair becomes a more efficient pressure receptor. The overhairs have a definite nerve supply, while the underhairs have no direct nerve supply. As a general but not absolute rule, the coarser hairs appear ontogenetically earlier than the finer hairs (Gibbs, 1938, Danforth, 1925a, Duerden, 1937 (reported by Wildman, 1937), Hofer, 1914, and Spencer and Sweet, 1899). The contour, diameter, and shape of a hair fiber changes from its root to its tip (Note awns, figures 16-18). The cross-sectional outline of hairs may vary from the thick rounded porcupine quill to the eccentric flattened hairs of seals. The former serves a protective function, while the latter is adapted to hug to the skin so as not to hinder aquatic locomotion. Many details of the anatomy of hair form are noted by Stoves (1942 and 1944a), Toldt (1935), and Lochte (1938). It is possible for a hair follicle to differentiate one type of hair at one stage and another type at another stage. The follicle of a bristle (kemp) of the Merino lamb may become the follicle of wool in the adult sheep (Duerden, 1937, reported by Wildman, 1937). A fine lanugo hair of the human fetus is associated with a follicle which will later be the follicle of a coarser hair. The theories of hair curling are reviewed by Herre and Wigger (1939). The curling of hair in primitive sheep is independent of the arrangement of hair, existence of hair whorls, or the cross section of the hair (Pfeifer, 1929). 157 482 Annals New York Academy of Sciences Wildman (1932) suggests that the shape of the foUicle, especially the curve in its basal portion, is a possible factor in hair curling. Reversal of the spiral in some wool fibers may be explained according to Wildman as due to a shift in the growing point of the follicle and inner root sheath. Spiral reversal occurs in human hair (Danforth, 1926). Pfeifer (1929) doubts that curHng is determined by a curve of the follicle alone and suggests that Tiinzer's (1926) contention that the follicle must be saber-shaped is im- 12 14 s. "D. c.T=is.ne Figures 10-15. Figure 10. The hair of the fox, Canis vulpes (after Toldt, 1935), illustrating intergrade hairs. From the left to the right, Toldt named the fibers Leithaar (bristles), Leit-Grannenhaar, thick Grannenhaar (awns), thin Grannenhaar, Grannen-WoUhaar, and WoUhaar (fur). Figure 11. The hair of the chinchilla. Chinchilla laniger (after Toldt, 1935) illustrating an animal hair coat with hairs of appro.ximately the same length. The 2 hairs on the left are awns, and the rest, either intergrade hairs or fur hairs. Figure 12. The hair of the wild pig, Sus scrofa (after Toldt, 1935) illustrating bristles on the left and underhair on the right with some intergrade hairs between them. Note the brushlike distal ends of the bristles. P'igure 13. The scale index (S. I.), according to Hausman (1930), is equal to the ratio of the free proximo- distal length of a scale (F) to the diameter of the hair shaft (E)). Figure 14. The thickness of the cuticle (C. T.), according to Rudall (1941), is equal to the length of a cuticular scale (1) times the sine of angle (sin 9) the scale makes with the cortex (X). Figure 15. Cross sections of several regions of a fur hair (left) and an awn (right) of the rabbit (after Toldt, 1935). The sections, at the top of the figure, are from the base of the hair and, at the bottom of the figure, from the tip of the hair. Illustrates general uniformity of the diameters of the fur haii and differences in diameters and contour of awn hairs throughout their lengths. portant. Waving of all compact wools is due at least in part to the flat- tening of the primary spiral and to the unequal lateral growth of the fiber (Duerden, 1927). The curling of hair in karakul sheep fetuses may be associated with the differences in the rates of growth in the various skin layers (Herre and Wigger, 1939). The factors responsible for curling and crimping of hair are as yet not completely known. 158 Noback : Morphology and Phylogeny 483 Structural Components of Hair The cuticle, cortex, and medulla are the three structural components in hair. They will be discussed in order. Cuticle. The cuticle consists of thin, unpigmented, transparent over- lapping scales, whose free margins are oriented toward the tip of the hair 20 Figures 16-21. KiGi'FE 16. Diagram of the fiber components of coat of a generalized non-wooled animal fafter Duerden, 1929). Note presence of bristles (coarse fibers), awns (fibers with fine basal segments and coarse distal segments) and fur fibers (fine fibers). Figure 17. Diagram of the fibers of the wild sheep (after Duerden, 1929). Note the presence of bristles (kemp), awns (heterotypes), and wool. Figure 18. Diagram of the fibers of British mountain breeds (after Duerden, 1929\ The fibers are mainly awns and wool. Few bristles are present. Figure 19. Diagram of the fibers of the British luster breeds (after Duerden, 1929). F'ibers on the left are wool fibers which are coarser than the wool fibers of wild sheep. The fibers on the right are modified awns with fine pro.ximal segments and slightly coarse distal segments. All fibers are elongated and spiraled. Figure 20. Diagram of fibers of adult Merino sheep (after Duerden, 1929). All fibers are wool. Note uniformity of all fibers as to size, length, and waviness. These wool fibers are coarser than wool fibers from wild sheep. Unlike the fibers of other breeds, the fibers of the adult Merino sheep grow from persistent germs and do not shed. Figure 21. Diagram of the fibers of the Merino lamb. Note the presence of bristles (kemp), awns (hetero- types), and wool. During later development, the bristles are shed and the distal coarse segments of the awns are lost. The adult coat is formed by the persistent growth of the wool fibers of the lamb, by the re- placement of wool in the follicles of the shed kemp, and by the persistence of the growth of the proximal seg- ments of the awns. (figure 22). Within the follicle, the free margins of the hair cuticular scales interlock with the inner root sheath cuticular scales, which are oriented in the opposite direction toward the papilla. This interlocking of scales helps to secure the hair in place (Danforth, 1925a). The cuticle functions as a capsule containing the longitudinally splitable corte.x (Rudall, 1941). This explains why the cortex of a hair frays at its severed end. In addition, the cuticle, with its oily layer, prevents the transfer of water (Rudallj 1041). 159 484 Annals New York Academy of Sciences The cuticular scales vary in thickness from 0.5 to 3 micra (Frolich, Spotel, and Tanzer, 1929). Since the scales overlap, the number of overlapping scales at any point on the hair surface determines the thickness of the cuticle. The cuticular thickness may be expressed as being equal to the length of the scales times the sine of the angle the scale makes with the cortical surface (figure 14, Rudall, 1941). The cuticular scales may be classified into two types: coronal scales and imbricate scales (Hausman, 1930). A coronal scale completely encircles the hair shaft. They are subdivided according to the contour of the free margins as: simple, serrate, or dentate (figure 23). Mliller (1939) con- tends that a coronal scale is in reality several scales whose lateral edges are HAIR SHAFT IN MICR0N5 ;oronal I ICORONAl I ACUMINATE EL0N6ATE OVATE CRENATE 1 FLATTENED Figure 22. Graph illustrating the relation of the diameter of the hair to the tjmes of cuticular scales. The finest hairs (with small diameters) have a high-scale index and coronal scales. The coarsest hairs (with large diameters) have a low-scale index and flattened scales. Diameters of hair shafts are plotted on the ordinate. General regions of the occurrence of scale forms are shown along the abscissa, the average scale indices along the curve. The figures of the scale types beneath the graph are not drawn to scale. (After Hausman, 1930.) fused. For example, a dentate coronal scale with 5 processes in its free border is the fused product of 5 elongated pointed scales. An imbricate scale does not completely surround the hair shaft. They are classified as ovate, acuminate, elongate, crenate, and flattened (figure 22, Hausman, 1930). Hausman (1930) devised a scale index to express the relation between the diameter of the hair shaft and the free proximo-distal dimension of the scales (figure 13). The free proximo-distal dimension is actually a means of expressing the type of scale. For example, coronal scales have a large proximo-distal dimension, while crenate scales have a small dimension (figure 22). An analysis of the scale indices indicates that a relation exists between the types of scales and the shaft diameters. In general, the finest 160 Noback: Morphology and Phylogeny 485 hairs have large scale indices and coronal scales, while the coarsest hairs have small scale indices and crenate or flattened scales. On the basis of the above, it is concluded that the types of cuticular scales present on hair are related not to the taxonomic status of the animal possessing the hair but rather to the diameter of the hair shaft (Hausman, 1930). In hairs with both thick and thin segments, the thick segments have the scale types of large diameter hairs while the thin segments have the scale types of small diameter hairs. A coarse guard hair has scales with free lips that are closely applied to the cortex and are scarcely raised. As a result, these hairs have a high luster (due to unbroken reflection of light from the hair surface) and do not inter- lock with other hairs. A fine underhair has scales with lips that have raised margins. As a result, these hairs are dull (due to broken reflection of light) and interlock with other fine hairs. Thus, mohair has a high luster but makes poor felt, while wool is dull but makes good textiles. C. Figure 23. Figures illustrating the types of coronal cuticular scales. A. simple scales, B. serrate scales, C. dentate scales, (after Hausman, 1930). Note raised nurgins on the free lips of scales. Many details of the cuticle, in many species of animals are presented and illustrated by Lochte (1938). Cortex. The cortex usually forms the main bulk of a hair. It is a column of fusiform keratinized cells which are coalesced into a rigid, almost homoge- neous, hyaline mass (Hausman, 1932). Damaged hairs tend to split length- wise because the elongated cortical cells are oriented longitudinally. The cortex has such a low refractive index — due to the degree of cornification^ that, in the absence of pigment, it is translucent. Since cortical scales have not been analyzed in such detail as cuticular scales, no statement can be made of a relation between cortical scale morphology and hair size. The form and distribution of the pigment in the cortex and the medulla is noted by Lochte (1938), Toldt (1935), and Hausman (1930). Hausman (1932 and 1944) analyzed the cortical air spaces known as cortical fusi — cortical in location and fusiform in shape — air vacuoles, air chambers, air vesicles, or vacuoles. As the irregular-shaped cortical cells located in the bulb rise to the follicular mouth, they carry between them cavities filled with tissue fluid. As the hair shaft dries out, the cavities lose 161 486 Annals New York Academy of Sciences the fluid, and air may fill the resulting spaces — the fusi. The shape of the fusi vary. They are largest, most numerous, and most prominent near the base of the hair, and they are fihform and thin or lost in the distal seg- ments of the hair. Seldom do they persist to the tip of a hair. As a rule, they are visible only under a microscope. Hausman implies that there is a relation between fusi and hair size. Presumably, the coarser a hair seg- ment is, the more numerous the fusi. Ringed hair results when the fusi appear in masses at regular intervals in the shaft. Fractured fusi result when hairs are damaged sufficiently to separate the cortical cells enough to allow air to collect between them. Fusi can be distinguished from pigment granules, for they are fusiform, whereas pigment granules have blunt ends. The presence of a thin membrane located between the cuticle and the cortex has been assumed by Lehmann (1944). Observations of pigment granules, cell nuclei, and submicroscopic fibrils are presented by Mercer (1942), Hausman (1930), and others. Medulla. The medulla (pith), when present, is composed of shrunken and variably shaped cornified remnants of epithelial cells connected by a fila- mentous network. In contrast to the cortex, the medulla is less dense and has fewer and larger cells, which are more loosely held together. In the medulla are air cells or chambers, which are filled by a gas, probably air. These air cells may be intracellular (deer) or intercellular (dog, weasel, and rat) (Lochte, 1938). The intercellular air cells are classifiable accord- ing to their coarseness and arrangement (Lochte, 1934 and 1938). Medullas are classified by Hausman (1930) as follows: absence of medulla, discontinuous medulla (air cells separate), intermediate medulla (several separate air cells of the discontinuous type arranged into regular groups), continuous medulla (air cells arranged to form a column), and fragmental medulla (air cells arranged into irregular groups). These types are illus- trated in FIGURE 24 and are arranged in the order of the sizes of hairs in which they are located. In the finest hairs (underfur), the medulla is either absent or of the discontinuous type. In the coarsest hairs, the medulla is either of the continuous or the fragmental type (figure 24). If a hair varies in thickness, its medulla will vary. For example, in the awns of sheep, the distal thickened segment has a medulla, while the fine proximal segment may have no medulla. The arrangement of the medullary air cells is related not to the taxonomic group of the animal possessing the hair nor the age of the hair, but rather to the diameter of the hair shaft (figure 24) (Hausman, 1930; Wynkoop, 1929; and Smith, 1933). The sheens and colors of hairs are largely determined by the light reflected from the medulla (Hausman, 1944). Although the cortex forms the bulk of the shaft in most hairs, the medulla assumes large proportions in some hairs. In rabbit hair (figure 15), the medulla is composed of large air cells separated by little more than a frame- work of cortex (Stoves, 1944c). The significance of the cuticle, cortex, and medulla in the commercial aspects of fur is presented by Bachrach (1946). Although many of the 162 Noback : Morphology and Phylogeny 487 details of the structural elements of hair cannot be detinitely utilized to identify an animal species (Hausman, 1944), it is possible that some morpho- logical features of hair can be used (Williams, 1938). Some chemical and physical aspects of the morphological elements of hair have been analyzed. Not only do the cuticle, cortex, and medulla exhibit different chemical and physical properties, but various segments of these structural elements may have different chemical and physical proper- FiGURE 24. Graph illustrating the relation of the diameter of the hair to the types of medullas. The finest hairs have no medulla, and the coarsest hairs have a fragmental medulla. Diameters in micra of the hair shafts are plotted on the ordinate. The figures of the types of medullas beneath the graph are not drawn to scale. (After Hausman, 1930.) ties (Rudall, 1944; Stoves, 19436, 1945; Lustig, Kondritzer, and Moore, 1945; Leblond, 1951; and Giroud and Leblond, 1951). Some Phylogenelic Aspects of Hair The relation of hair to the epidermal structures in non-mammalian ani- mals has been discussed by many authors and has been summarized by 163 488 Annals New York Academy of Sciences Botezat (1913 and 1914), Danforth (1925b), and Matkeiev (1932). No direct relation between hair and non-mammalian epidermal elements has been established. Hair is most probably an analog to these structures. Danforth (1925b) and others conclude that hair is probably a de novo morpho- logical entity in mammals. Broili (1927) reports that he identified hair and hair follicles in a fossil aquatic reptile, Rhamphorynchus. This animal is a specialized reptile, removed from those reptiles in the evolutionary line to mammals. If established, this observation would alter the concept that only mammals produce hair. The phylogeny of hair in related groups of animals has not been analyzed extensively. Because of the economic importance of wool, several studies of the hair types in the coat of a number of breeds of sheep have been made. One significant aspect of these studies is that they illustrate how the hair coat may vary in closely related forms. This statement is adapted primarily from Duerden (1927 and 1929). The generalized wild animal hair coat consists of an overcoat of bristles and awns and an undercoat of fur (figure 16). In the wild sheep and the black-headed Persian sheep, the hair coat is similar to that of the wild animal. These sheep have an overcoat of bristles (called kemp) and awns (called heterotypes) and an undercoat of wool (figure 17). The British mountain breeds have a hair coat consisting of awns and wool (figure 18). In these breeds, kemp formation is negligible. The coat of the British luster breeds have evolved in another direction. The awns retain their fine proximal segments. Their distal segments are still thicker than the proximal segments, but are thinner than the distal segments of the awns of primitive sheep. The wool undercoat fibers have thickened. Both fiber types are elongated and spiraled (figure 19). The adult Merino sheep, the most efficient wool-producing sheep, has a coat consisting of elongated, regularly crimped fibers of uniform diameters and lengths (figure 20). An analysis of the coat of the Merino lamb is essential for the identification of the types of fibers that form the coat of the adult sheep. The Merino lamb coat has bristle, awn, and wool fibers (figure 21). During ontogeny, the bristles are shed and then replaced by wool fibers. The awn fibers lose their distal thickened segments, but the thin proximal segments persist. The wool fibers are retained, but are coarser than the wool of primitive sheep. Hence, the adult Merino sheep coat consists of wool fibers differ- entiating from follicles which produced kemp in the lamb, of awns deprived of their distal segments, and of wool fibers differentiating from follicles which produced wool in the lamb. A major difference between the Merino sheep and other sheep is in the nature of the hair follicles. Whereas the coat of other breeds is shed periodically and then new hairs differentiate from the follicles, the fibers of the adult Merino sheep grow from persistent germs and are not shed. In the evolution of the sheep coat from primitive wild sheep to the various domestic breeds, several changes have occurred. As summarized by Duer- den (1927), the domestic wooled sheep has evolved in the direction of the 164 Noback : Morphology- and Phylogeny 489 loss of the protective coat both of bristles (kemp) and awns (heterotypes), the increase in length, density, and uniformity of the fibers, and the tend- ency of the retained bristles to become finer but still capable of being shed. In addition, the Merino sheep has developed persistently growing hair follicles. Important implications of the evolution of the sheep coat are that the types of hair in the hair coat may differ (1) in closely related animals and (2) at various stages of ontogeny with the same animal. Hence, data derived from a study of the coat of one animal species may not always apply to another animal species. Bibliography American Society for Testing Materials. 1948. Standards on textile materials. By the Society for Testing Materials. 560 pp. Philadelphia. Bachrach, M. 1946. Fur. 672 pp. Prentice-Hall. New York. BoLLiGER, A. & M. H. Hardy. 1945. The sternal integument of Tricliosurus vulpe- ciila. Proc. Roy. Soc. New South Wales 78: 122-133. Bonnet, R. 1892. Ueber Hypotrichosis congenita universalis. .\nat. Hefte 1: 233- 273. Botezat, E. 1913. Ueber die Phvlogenie der Saugetierhaare. Verhandl. d. Gesellsch. deutsch. Naturf. u. .\erzte. 85: 696-698. 1914. Phyiogenese des Haares der Saugetiere. Anat. Anz. 47: 1-44. Broili, F. 1927. Ein Rhamphorhynchus mit Spuren von Haarbedeckung. Sitzungs- ber. Math.-Naturw. Abt. Bayerische .\kad. VViss. Munchen 49-68. Calef, .\. 1900. Studio isologico e morfologica di un' appendice epitaliale del pelod nella pelle del Mus dectmanus var. albina e del Sus scrofa. .\nat. .A.nz. 17: 509-517. Carter, H. B. 1943. Studies in the biology of the skin and fleece of sheep. Council for Scientitic and Industrial Research, Commonwealth of .\ustralia. Bulletin No. 164: 1-59. Claushen, .\. 1933. Mikroskopische Untersuchungen iiber die Epidermatgebilde am Rumpfe des Hundes mit besonder Beriicksichtigung der Schweissdriisen. Anat. Anz. 77: 81-97. Danforth, C. 1925(Z. Hair, with special reference to hypertrichosis, .\merican Medi- cal -Association. Chicago, Illinois; .\rch. of Derm, and Svph. 11: 494-508, 637-653, 804-821. Daxforth, C. 19256. Hair in its relation to cjuestions of homology and phvlogeny. Am. J. Anat. 36: 47-68. Daxforth. C. 1926. The hair. Nat. Hist. 26: 75-79. Danforth, C. 1939. Physiology of human hair. Physiol. Rev. 19: 94-111. Dawson, H. L. 1930. .\ studv of hair growth in the guinea pig (Cavia cobava). Amer. J. Anat. 45: 461-484. DeMeijere, J. C. H. 1894. Cber die Haare der Siiugethiere besonders uber ihre .Anordung. Gegenbaur's Morphol. Jahrb. 21: 312-424. Dry, F. W. 1926. The coat of the mouse (.1/ms w;<5c;■• biceps brachii (glenoid head) •biceps brachii (coracoid head) Fig. 4. — Front views of the right shoulders of Antrozous pallidus (above), showing the lateral curvature of the coracoid process of the scapula typical of most vespertilionid bats, and of Eumops perotis (below), showing the medial curvature of the coracoid process typical of molossid bats. study differed widely (Fig. 2, Table 2), but certain differences between the molossid and vespertilionid wings are readily apparent and are indicative of functional contrasts. The ratios based on wing proportions illustrate that, com- pared to the vespertilionid wing, the molossid wing is narrow, has a long distal segment, a high aspect ratio (the ratio of length to width of a wing; for ir- regularly shaped wings it is considered to be the ratio of the wing span^/wing area), and a high wing loading (weight in pounds/wing area in feet^). For the molossid bats considered here the length of the chiropatagium averaged 154.1% of the length of the plagiopatagium, and the width of the wings averaged 36.8% of the length. In the vespertilionids measured the cor- responding figures were 138.5% and 48.7%. The aspect ratios reflect these differences and are higher for the molossids (8.60-9.98) than for the ves- pertilionids (5.99-6.74). In general, the higher the aspect ratio the more ef- ficient the wing because of the reduction of drag at the wing tip. The wing 176 May 1966 VAUGHAN— MORPHOLOGY AND FLIGHT OF MOLOSSIDS 257 Table 2. — Proportions of wings of bats. All figures are averages; the numbers of specimens measured are given in parentheses Species Area of chiropatagium Area of phigiopatagium Length of chiropatagium Length of plagiopatagium Greatest width of wing Length of wing Aspect ratio: Wing span^ Area of wings Myotis yumanensis (5) 0.95 L32 0.47 6.74 M. evotis (5) 0.80 L35 0.51 6.48 M. lucifugus (5) 0.69 L41 0.48 6.47 Plecotus townsendii (5) 0.86 L46 0.50 5.99 Tadarida brasUiensis (5) 0.93 1.58 0.38 8.60 T. molossa (3) 0.86 L47 0.36 9.71 Eumops perotis (5) 0.83 1.58 0.36 9.98 loadings of the molossid bats (0.325-0.546) are considerably higher than those of the vespertilionids (0.157-0.202). As a rule, the higher the wing loading the greater the speed necessary to produce adequate lift for sustained flight. Usually the smaller the bat the lower the wing loading because in small bats the ratio of mass to surface is small ( the volume and mass vary as the cube of the linear dimensions whereas the surface area varies as the square). Instructive comparisons can be made between habitat and wing form in both bats and birds and between the characteristics of the wings of certain bats and those of the wings of birds with well-known modes of flight. The plan form of the wing of Myotis evotis ( Fig. 2 ) is similar to that of many small passerine birds. This roughly elliptical wing is fairly efficient for low speed flight, and occurs in birds adapted to flight through brush or woods, or where numerous obstacles make long wings unmanageable (Savile, 1957). The short, broad wing of Myotis evotis is seemingly well suited to flight near the ground in the wooded or brushy areas the bat inhabits. A markedly different adaptation is Illustrated by the wings of molossid bats which are similar to the "high-speed" wings of many birds known to be strong enduring fliers or to feed on the wing ( falcons, plovers, sandpipers, swifts and swallows ) . Such birds characteristically fly in open places with few obstructions, a situation "allowing" the development of long, aerodynamically efficient wings. The wings of these birds have the following characteristics according to Savile ( 1957 ) : low camber; high aspect ratio; taper to a slender, elliptical tip; pro- nounced sweepback of the leading edge; and wing root fairing. These same features describe well the molossid wing. Thus, in wing design and foraging habits the molossids appear to be chiropteran counterparts of the swdfts and swallows, whereas the smaller vespertilionids seem to most nearly resemble the smaller flycatchers in wing design, but differ from the latter in flying con- tinuously while feeding. 177 258 JOURNAL OF MAMMALOGY Vol. 47, No. 2 Table 3. — Sizes, aerodynamic characteristics and computed minimum flight speeds of seven species of bats. All figures are averages; the numbers of specimens measured are given in parentheses Species Weight (g) Wing span ( cm ) Wing area (cm2) Wing loading lbs/ft2 g/cm2 Minimum flight speed (mph) Myotis yumanensis (5) 5.2 20.26 60.95 0.173 0.084 8.3 M. evotis (5) 6.2 22.80 80.20 0.157 0.077 7.9 M. lucifugus (5) 8.1 23.30 83.41 0.202 0.099 8.8 Plecotus townsendii (5) 9.1 24.52 100.41 0.184 0.090 8.5 Tadarida hrasiliensis (5) 12.2 25.08 73.14 0.339 0.165 11.6 T. molossa (3) 16.2 31.33 101.12 0.325 0.159 11.7 F.umops perotis (5) 53.5 44.58 199.22 0.546 0.266 14.7 From the weight of a bat and the area of its wing surfaces the speed it must fly to sustain level flight can be approximated by the equation (von Mises, 1945) where V is velocity (in feet per second); 2 gc is a unit- conversion constant; W is the total weight; A is the area of the wings; C^ is the coefficient of lift; and V is the density of air in pounds per cubic foot. The coefficient of lift is derived from the size, camber, aspect ratio, angle of attack and other characteristics of the wing, and for the present study was assumed to be 1.0, which probably approximates the actual values closely enough to cause little error. These calculations are based on the further as- sumption that each bat has its wings fully and rigidly outstretched. Although the calculated speeds may not correspond closely to the actual flight speeds of the bats, they probably reflect accurately the relative flight speeds. Except for the molossids, flight speeds arrived at by the above equation (Table 3) are fairly close to those found experimentally by Hayward and Davis ( 1964 ) . As these authors mention, the speeds shown for the molossids in their study are probably too low because the bats could not fly normally under their experimental conditions. The speeds calculated on the basis of the figures for total weight and wing area shown in Table 3 suggest that in order to maintain level flight the molossids must fly faster than do the vespertilionids. The speed for Myotis evotis (7.9 mph), for example, is roughly half that of Eumops perotis (14.7 mph). The calculated speeds probably approximate the relative speeds of the bats under study in level flight, but all of these bats seem to be capable of a wide range of flight speeds. Judging from my own observations, M. evotis can hover briefly and can fly at very low speeds; at the other extreme, some molossids are capable of rapid dives and of level flights at speeds far greater than those listed here. A complicating factor, but one of critical importance when considering the flight capabilities of a bat, is the animal's ability to 178 May 1966 VAUGHAN— MORPHOLOGY AND FLIGHT OF MOLOSSIDS 259 vary the camber, angle of attack and the areas of the membranes. A fm-ther complexit>' is the fact that the wings are in nearly constant movement dur- ing flight, and supply both the lift and the thrust necessary for flight. Various aerodynamic relationships are pertinent to the problem of relative flight speeds and differences in morphology in bats. For example, because drag increases in proportion to surface area and as the square of the speed, E. perotis is probably subject to about three times the drag faced by M. evotis. This explains why features which tend to minimize drag, such as low camber of the wing and short ears which present their most streamlined aspect to the airstream, are of vastly greater importance in the large E. perotis, and in most molossid bats, than in smaller, relatively slow-flying bats. Even the short, velvety fur of molossids may be an adaptation to reduce drag caused by the body during flight. Miscellaneous considerations. — The family Molossidae is unique in having developed the most rapid, enduring flight occurring in bats while retaining (or developing) the most accomplished terrestrial locomotion. Consequently, molossids offer many trenchant examples of a single morphological character serving diverse functional ends. One such character is the posterior flexion of the first phalanges of digits three and four. Because of this modification the long tip of the narrow wing is manageable when the bats are not flying, an important feature in a group including many species which take daytime refuge in narrow crevices. This unusual pattern of flexion may have developed prior to the lengthening of the wing tip and may have "allowed" the evolution of this typically molossid character. The part of the wing distal to the carpus folds into a bundle no longer than the radius, facilitating a lateral action of the forelimb during quadripedal locomotion which enables molossids to move remarkably rapidly and easily within the confines of narrow crevices. Thus, the pattern of phalangeal flexion in molossids has probably played a role in both terrestrial and aerial locomotion. The small uropatagium of these bats slips forward along the tail freeing the hind limbs to move in a wide arc when the animals run. Also, the re- duced drag during flight resulting from the small size of the uropatagium is probably aerodynamically important in furthering the cause of fast flight. The well-braced flight membranes of molossids may also serve two ends, for in addition to resisting effectively the force of the airstream during rapid flight, they are better able than are the delicate membranes of most bats to withstand the occasional rough treatment resulting from crawling between the irregular and abrasive surfaces of rock crevices. Perhaps the strong adductors and flexors of the hind limbs represent the best example of a dual-puipose molossid character. These limbs are re- markably robust and strongly muscled and account in large part for the ac- complished terrestrial locomotion typical of the group. Of equal importance, however, is their function in serving as a rigid anchor for the posterior portion 179 260 JOURNAL OF MAMMALOGY Vol. 47, No. 2 of the plagiopatagium. This function requires strong flexors and adductors of the shank to resist the powerful lateral and dorsal pull exerted by the flight membranes during the downstroke of the wing, and is of basic importance in maintaining the proper angle of attack and camber of the plagiopatagium. Considering the evolution of the family Molossidae, the sturdiness of the hind limb and the complete fibula suggest that the basal molossid stock may have diverged early from the rest of the Chiroptera, before the hind limbs had been greatly modified from their primitive structure and propor- tions. By contrast, in many members of the family Phyllostomidae the hind limbs have become so highly specialized that they no longer function effec- tively in quadripedal locomotion. But in molossids, probably because the primitive hind limbs suited the demands of both terrestrial and aerial locomo- tion, the hind limbs have remained basically primitive through a period of time that saw the evolution of the highly specialized molossid wing of today. Some vespertilionid bats and one rhinolophid bat have been shown to use the flight membranes in capturing insects (Webster and Griffin, 1962), In these species most modifications tending to reduce the dexterity of the phalanges were probably disadvantageous, and the evolution of the hand was probably influenced by its use in capturing insects, as well as by the de- mands of aerial and terrestrial locomotion. ACKNOWLEDGMENTS This study was supported in part by a grant from The Society of Sigma Xi and RESA Research Fund. For critically reading the manuscript and for helpful discussions of aerodynamics I am grateful to Dr. P. H. Baldwin and Dr. R. D. Haberstroh. LITERATURE CITED Davis, R. and E. L. Cockrum. 1964. Experimentally determined weight lifting capacity in individuals of five species of western bats. J. Mamm., 45: 643-644. Grinnell, H. W. 1918. A synopsis of the bats of CaHfomia. Univ. California Publ. Zool., 17: 223-404. Hall, E. R. and W. W. Dalquest. 1963. The mammals of Veracruz. Univ. Kansas Publ., Mus. Nat. Hist., 14: 165-362. Hayward, B. and R. Davis. 1964. Fhght speeds in western bats. J. Mamm., 45: 236-242. Howell, A. B. 1920. Contribution to the hfe history of the California mastiff bat. J. Mamm., 1: 111-117. Miller, G. S. 1907. The famihes and genera of bats. U. S. Nat. Mus. Bull., 57: xvii -I- 282 pp. Orr, R. T. 1954. Natural history of the pallid bat, Antrozous pallidus. Proc. CaUfomia Acad. Sci., 28: 165-246. Savtle, D. B. O. 1957. Adaptive evolution in the avian wing. Evolution, 11: 212-224. Struhsaker, T. T. 1961. Morphological factors regulating flight in bats. J. Mamm., 42: 152^159. Vaughan, T. a. 1959. Functional morphology of three bats: Eumops, Myotis, Macrotus. Univ. Kansas Publ, Mus. Nat. Hist., 12: 1-153. von Mises, R. 1945. Theory of flight. McGraw-Hill Book Co., New York. 629 pp. Webster, F. A. and D. R. Griffin. 1962. The role of the flight membranes in insect capture by bats. Animal Behavior, 10: 332-340. Colorado Agricultural Experiment Station, Colorado State University, Fort Collins. Accepted 23 September 1965. 180 NATURAL HISTORY MISCELLANEA Published by The Chicago Academy of Sciences Lincoln Park - 2001 N. Clark St., Chicago 14, Illinois No. 170 October 30, 1959 Toxic Salivary Glands in the Primitive Insectivore Solenodon George B. Rabb* In 1942 0. P. Pearson demonstrated the toxic property of the saliva of Blarina brevkauda, a common shrew of the eastern United States, and identified its principal source as the submaxillary gland. Compara- tive studies at that time and subsequently revealed that similar poison- ous factors were not present in the salivary glands of other soricid and talpid insectivores (Pearson, 1942, 1950, 1956). I had an unexpected opportunity to make a crude check on the salivary glands of Solenodon paradoxus, a remote relative of the shrews, when three of these animals died at the Chicago Zoological Park within two months after their arrival in 1958 from the Dominican Republic. Parts of the submaxillary and parotid glands of one animal that had died one to two hours beforehand were ground separately with sand, diluted to 10 per cent by weight solutions with 0.9 per cent XaCl solution, and filtered, following the procedure of Pearson (1942). These solutions were injected into a small series of male white mice that ranged in weight from 29 to 44 grams. All of the mice injected with extract from submaxillary gland showed some reaction — at least urination and irregular or rapid breathing for several minutes. Five that received intravenous doses of extract of .09 to .38 mg. submaxillary gland per gram of body weight did little more than this and recovered within 30 minutes. Five that received intravenous doses of .38 to .55 mg. per gram additionally ex- hibited protruding eyes, gasping, and convulsions before dying within two to six minutes. Two animals that had intraperitoneal injections of extract of .56 and .66 mg. per gram died in about 12 hours, and one injected at the level of 1.02 mg. per gram died in 13 minutes. Urination, cyanosis, and depression were observed in these animals. Three "con- trol" mice injected intravenously with extract of 1.02, 1.68, and 1.87 ♦Chicago Zoological Park, Brookfield, Illinois 181 No. 170 The Chicago Academy of Sciences, Natural History Miscellanea mg. of parotid gland per gram of body weight showed no distress except for initially very rapid breathing in the last case. In general these results are very like those described for Blarina extracts. It may be noted that the twentyfold lesser potency evident here of Solenodon extract as compared to that of Blarina may be due to postmortem inactivation of the toxic principle as reported by EiLs and Krayer (1955) for fresh Blarina material. Further tests with tne refined techniques of these authors using acetone treated glands will be necessary for a fairer assessment of the potency of Solenodon toxin. Sections were made of the submaxillary glands and stained w-th hematoxylin and eosin and also with a modification of Mallory's triple stain. These sections showed some large cells with coarse acidophilic granules and small nuclei in the secretory ducts. Pearson (1950) sus- pected that such cells in Blarina might be concerned in the production of the saliva's toxic principle, although somewhat similar cells are found in other soricids. The submaxillary glands of Solenodon are rather enormous and con- spicuous structures (see fig. 47 in Mohr, 1938). Each gland weighs three to four grams in adult animals. According to Allen (1910), the duct of the submaxillary gland ends at the base of the large deeply channeled second incisor tooth of the lower jaw (see fig. 19D in Mc- Dowell, 1958). Presumably toxic saliva would be conducted thereby into a wound. I could not induce Solenodon to bite live mice and there- fore have no direct evidence on this point. However, in 1877 Gundlach reported inflammatory effects of bites by Cuban Solenodon to himself and a mountaineer (although he dismissed the possibility of venomous action on the basis of authority!). Of his hand bite he said: "... I was bitten by the tame individual, which gave me four wounds cor- responding to the [large] incisors: those from the two upper incisors healed well, but those from the lower ones inflamed." Moreover, there are indications that Solenodon is not immune to its own venom. Autopsy of the third animal disclosed multiple bite wounds on the feet and no obvious internal evidence of other causes of death. Sections of the liver show considerable congestion in that organ. The snout, lips, limbs, and tail were very pale the afternoon preceding death. Mohr (1937, 1938) gave accounts of several cases in which death was the outcome of fighting with cage mates although only slight foot wounds were inflicted. Pearson (1950) reported that Blarina was rela- tively immune to its own venom, although the single test animal died and the interpretation was problematical. The utility of the venom for 182 Rabb: Toxic Salivary Glands of Solenodon 1959 Solenodon in its natural environment is unknown and is certainly not indicated by its insectivorous habits. The explanation may be phylo- genetic and historical rather than one of present-day function. I wish to acknowledge the help of the park's veterinarian, W. M. Williamson, and medical technician, Ruth M. Getty. Literature Cited Allen, Glover M. 1910. Solenodon paradoxus. Mem. Mus. Comp. Zool., 40: 1-54. Ellis, Sydney and Otto Krayer 1955. Properties of a toxin from the salivary gland of the shrew, Blarina brevicauda. Jour. Pharmacol, and Exptl. Therap., 114: 127-37. Gundlach, Juan 1877. Contribucion a la mamalogia Cubana. Havana, G. Monteil, 53 pp. McDowell, Samuel B., Jr. 1958. The Greater Antillean insectivores. Bull. American Mus. Nat. Hist., 115(3): 113-214. Mohr, Erna 1937. Biologische beobachtungen an Solenodon paradoxus Brandt in Gefangenschaft. HI. Zool. Anz., 117: 233-41. 1938. Biologische beobachtungen an Solenodon paradoxus Brandt in Gefangenschaft. IV. Ibid., 122: 132-43. Pearson, Oliver P. 1942. On the cause and nature of a poisonous action produced by the bite of a shrew {Blarina brevicauda). Jour. Mamm., 23: 159-66. 1950. The submaxillary glands of shrews. Anat. Record, 107; 161-69. 1956. A toxic substance from the salivary glands of a mammal (short-tailed shrew), pp. 55-58 in Venoms, ed. E. E. Buckley and N. Porges, American Assoc. Adv. Science Publ. No. 44, xii + 467 pp. 183 466 JOURNAL OF MAMMALOGY Vol. 48, No. 3 SOME ASPECTS OF THE WATER ECONOMICS OF TWO SPECIES OF CHIPMUNKS The water economics of chipmunks have not received much attention from physiological ecologists. Allen (New York State Mus. Bull., 314: 1-122, 1938) wrote of Tamias striatus: "Unlike many of the western ground squirrels, the Eastern chipmunk requires a great deal of water to drink." Panuska and Wade (J. Mamm., 38: 192-196, 1957) found that water consumption of captive T. striatus decreased from 33.4 ml per day just after capture to 29.2 ml per day after the animals had been confined for a time. Davis ( Murrelet, 15: 20-22, 1934) wrote that water was not a factor in determining the distribution of the cliff chip- munk, Eutamias dorsalis, in Nevada. Seton (Lives of game animals, 4: 184-215, 1929) observed that west and south of Manitoba the least chipmunk, E. minimus, is found in desert environments far from permanent water. ManviUe (Misc. Publ. Mus. Zool., Univ. Michigan, 73: 1-83, 1949) thought water to be of httle importance in the distribution of E. minimus in the Huron Mountains of Michigan. In the Itasca region of Minnesota the ranges of the gray eastern chipmunk, T. striatus griseus, and the least chipmunk, E. minimus neglectus, overlap. Since striatus and minimus apparently have markedly different water economics in the extremes of their ranges, I wondered if the two species, in the mesic, forested Itascan habitats, would differ from each other in their gross and weight-relative water consumption and in their responses to water deprivation. These studies were conducted in August and September 1963. Chipmunks of both species were captured in National Live Traps, 5^/4 X 5^2 X 16 inches, set within 3 miles of Itasca State Park, Hubbard and Clearwater counties, Minnesota. The chipmunks were transferred to an animal room in the zoology building at the University of Minnesota, Minneapohs. There were no provisions for regulating light, temperature, or humiclit; in the room. Nine striatus and nine minimus were confined individually in cages 18 X 18 X 12 inches with wood shavings provided for htter, and were fed only sunflower seeds. The seeds contained water amounting to about 12% of their weight. Tap water was provided ad libidum in 30 cc or 100 cc graduated drinking tubes. One tube of each size, hung on the rack of cages, permitted assessment of evaporative water loss from the tubes. For 36 days, daily records were kept of the change of water level in each tube. On 18 days, at least one of the striatus spilled water, indicated by wet litter below the tube. Least chipmunks were not known to spiU water. Records of water consumption for the 18 days on which no spillage was noted were used to calculate each animal's gross water consumption. Each animal's mean daily water consumption was calculated by dividing gross water consump- tion by 18. Each animal's water consumption per g of body weight was estimated by dividing gross water consumption by the mean value of the animal's body weight as recorded on the first and thirty-sixth days. The arithmetic mean, standard deviation, standard error of the mean, and coefficient of variation ( V ) were computed for each of the foregoing variables for each species. Following the studies of water consumption, seven individuals of each species were deprived of water for five consecutive days. Two individuals of each species served as controls and were allowed unrestricted access to drinking water. The chipmunks were weighed daily during the five days of water deprivation and for seven days after ad libidum access to water was restored. Each animal's daily weight was recorded as a percentage of its body weight at tlie outset of the experiment. Mean daily percentages were calculated for each species. Statistical procedures followed were those of Simpson, Roe, and Lewontin (Quantitative zoology, 1960). The level of significance used for tests of hypotheses was 95%. Data on water consumption are summarized in Table 1. Although the gross water requirement of minimus was about one-third that of striatus, there was no significant difference in the weight-relative water consuimptions of the two species. The coefficients 184 August 1967 GENERAL NOTES 467 Table 1. — Summary of data on water constimption of confined chipmunks. 1'. striatus E. minimus Variable Mean Range SE V Mean Range se V Weights of animals (g) Total HaO consumed ( ml ) Mean HaO consumed per day (ml) Mean H2O consumed per g body weight 115.0 102.8-132.8 296.1 208-471 16.4 11.6-26.2 2.52 1.98-<3.60 3.4 8.9 26.5 26.8 1.5 0.18 20.8 46.2 42.0-50.2 97.8 76-150 5.4 4.2-8.3 2.16 1.62r-2.96 0.8 7.6 0.4 0.17 5.1 23.2 23.6 of variation show that individual variation in water consumption was very large. There are individual and specific differences in the adjustments of the animals to captivity. As a group, eastern chipmunks were more sedentary in their cages than were least chipmunks, but activity among individual striatus was quite variable. The two species did not differ significantly from each other in their abilities to resist weight loss during water deprivation or to regain weight once access to water was restored (Table 2). When experimental animals were deprived of water, they first became more active than usual. Their activity decreased markedly during the last three days of water deprivation. Normally, a lively chase ensued before an animal could be caught by hand for weighing, but by the fifth day of water deprivation one could easily pick up a dehydrated chipmunk from its cage. Control animals remained quick and alert. Their weights varied only a few grams on either side of their pre-experimental weights during this study. The mean weights of rehydrating striatus are distorted by the weights of one individual that continued to lose weight even after access to water was restored. Ultimately, Table 2. — Percentages of pre-experimental weights of experimental chipmunks during dehydration and rehydration. Day n T. striatus Range Mean SD E. minimus number n Range Mean SD 1 7 89.9-95.5 93.0 1.8 7 89.5-95.4 93.4 1.8 2 7 85.5-91.6 88.2 2.2 7 85.6-91.6 88.6 1.9 3 7 80.0-86.7 83.6 2.6 7 79.2-S5.2 82.9 2.2 4 7 74.6-82.5 78.6 3.4 7 74.8-83.7 78.8 3.1 5 7 70.4-79.4 75.7 Access 4.0 to water 7 restored 69.4-79.2 74.9 3.7 6 7 76.6-85.4 82.5 2.9 7 80.7-92.0 84.7 4.5 7 6* 67.8-85.5 81.1 6.6 7 77.3-95.0 84.9 6.2 8 6 63.6-92.8 83.3 10.1 6* 79.1-96.5 88.1 6.3 9 6 62.8-94.3 85.7 11.6 6 80.7-100.0 89.1 7.0 10 6 59.4-97.4 87.0 14.1 6 79.9-100.0 89.1 7.1 12 6 57.9-100.0 89.1 16.3 6 82.7-100.0 90.6 6.0 * One experimental animal found dead. 185 468 JOURNAL OF MAMMALOGY Vol. 48, No. 3 this animal lost half of its pre-experimental weight, but the pre-experimental weight was eventually regained and siupassed. The literature suggests that striatus is somewhat more dependent upon a plentiful supply of drinking water than are minimus and its relatives. The present experimental evidence suggests that this is so. The gross water requirements of minimus are small. With some insects and fruit of high water content, and a morning supply of dew, Itascan minimus can probably keep themselves in good condition with no permanent source of drinking water. When raspberries ( Rubus minnesotanus ) are in fruit at Itasca, minimus is found in greatest abundance around raspberry thickets. Often, several chipmunks at a time can be seen eating the fruit, the seeds of which the animals carry away in their cheek pouches. While such a diet would supplement a marginal water supply, I think the chipmunks take the fruit as much for the seeds as for the moist, pulpy parts. It was surprising to me that m,inimus — a small, active species with high metabolic and breathing rates — did not require more water per g of body weight than did the larger, seemingly less active striatus whose metabohc and breathing rates are lower. Nor did comparison of rates of dehydration and rehydration suggest any significant difference between the water economics of the two species. At Itasca, as in other parts of its range, minimus is most common in exposed habitats such as the margins of slash piles and gravel pits. Exposure to wind and solar radiation is maximal in such situations; daytime tempera- tures, consequently, are often high and relative humidity is often low. In contrast, striatus remains beneath tree and shrub cover where, since insolation and wind are reduced, daytime temperatures are lower and relative humidity is higher than in open habitats. In view of the morphologic, physiologic, and behavioral differences between the two species, the similarities found in their water economics may represent the existence of physiologic adaptations in minimus to its somewhat more xeric Itascan microhabitats. A thorough analysis of the water economics of these and other chipmunks could, in addition to testing these results, provide information relevant to habitat preferences among the many species of Eutamias. In addition, Nadler (Amer. Midland Nat., 72: 298-312, 1964) has suggested that physiologic and ecologic study may shed light on phylogenetic problems involving Eutamias. I have found chipmunks to be difficult subjects for experi- ments of this sort. They are active and often hard to catch for weighing. Some individuals invariably shake water out of their drinking tubes; others are inclined to pack Htter into the tubes, but use of cedar tow as litter reduces this. Control of temperature, light, and hu- midity, and selection of experimental animals of about the same size and age, should reduce the variability in performance. Part of this work was done while I held an NSF Summer Fellowship for Teaching Assistants, awarded through the University of Minnesota. — Richard B. Forbes, Department of Biology, Portland State College, Portland, Oregon 97207. Accepted 30 January 1967. 186 THE OXYGEN CONSUMPTION AND BIOENER- GETICS OF HARVEST MICE OLIVER P. PEARSON Museum of Vertebrate Zoology, University of California, Berkeley R\TES of metabolism or of oxygen consumption have been reported k, for many species of small mam- mals, but little effort has been made to relate such measurements to the energy economy of small mammals in the wild. Such efifort has been avoided because the rate of metabolism varies so much with changes of the ambient temperature and with activity of the animal. I believe, however, that these variables can be handled with sufficient accuracy so that one can make meaningful estimates of the 24-hour metabolic budget of free-liv- ing mice in the wild. In this study I have measured the oxygen consumption of captive harvest mice under different con- ditions, and from these measurements I have estimated the daily metabolic ex- change of wild harvest mice living in Orinda, Contra Costa County, Califor- nia. The harvest mice used in the study {Reithrodontomys megalotis) are noctur- nal, seed-eating rodents living in grassy, weedy, and brushy places in the western half of the United States and in Mexico. In Orinda they encounter cool wet win- ters (nighttime temperatures frequently slightly below 0° C.) and warm dry sum- mers (daytime temperatures sometimes above 35° C, but nights always cool). They do nothibernate. MATERIAL AND METHODS Five adult harvest mice were caught on January 29 and 30, 1959, and were kept in two cages in an unheated room with open windows so that the air tem- perature would remain close to that out- side the building. They were fed a mix- ture of seeds known as "wild bird seed." Metabolic rates were tested between Jan- uary 29 and April 1 in a closed-circuit oxygen consumption apparatus similar to the one described by Morrison (1947) but without the automatic recording and re- filling features. All tests except the 24- hour runs were made during the daytime and without food. Since harvest mice are strongly nocturnal, several hours had usually elapsed between their last meal and the measuring of their oxygen con- sumption. When placed in the apparatus, the mice usually explored the metabolic chamber and groomed their fur for about half an hour and then went to sleep on the wire mesh floor of the chamber. One hour or more was allowed for the animals to become quiet and for the system to come to temperature equilibrium. The animals usually were left in the chamber until from five to ten determinations of oxygen consumption had been made, during which they had remained asleep or at least had made no gross movements. Each determination lasted between 9 and 24 minutes. The mice were weighed when they were removed from the apparatus. Oxygen consumptions are reported as volume of dry gas at 0° C. per gram of mouse. RESULTS SIZE X RATE OF METABOLISM Adult harvest mice weigh between 7 and 14 grams. Larger individuals con- sume oxygen at a lower rate per gram of 152 187 A'lETABOLISM OF HARVEST MICK 153 body weight (I'ig. 1). I'or example, at with restful surroundings, as in a nest, 12° (\ a 12-gram mouse would use only the animals probably relaxed their tcm- 1.17 times as much oxygen per hour as an perature control temporarily. This ex- 8-gram mouse, although it is 1.5 times as planation seems plausible in view of the heavy. The various points in the regres- known lability of the body temperature sion of body weight against rate of oxy- of some rodents such as Peromyscus gen consumption can be fitted ade- (Morrison and Ryser, 1959), Dipodomys quately with a straight line, and from the (Dawson, 1955), and Perognathus (Bar- slopes of such lines illustrating the re- tholomew and Cade, 1957) under similar gression at different ambient tempera- circumstances. Birds permit their body tures it may be seen (Fig. 1) that at cold temperature to drop about 2° C. when temperatures a variation of 1 gram in they sleep at night, and this is accom- body weight causes a greater change in panied by a drop of as much as 27 per metabolic rate than at 30° C. At 1°, 12°, cent in rate of metabolism (De Bont, and 24° a change of 1 gram in weight is 1945). The 40 per cent drop shown by associated with a change in oxygen con- some of the mice may have been accom- sumption of 0.98, 0.48, and 0.35 cc/g/hr, panied by a drop in body temperature of respectively. several degrees. At warm and moderate temperatures there was little variation in the measure- resting met.^bolism at different ments of each mouse during any one run temperatures (Fig. 1), but at 1° C. the variation was Since the weights of adult harvest mice sometimes enormous. Since each meas- vary so much, it is desirable to eliminate urement was made over a period while the the size variable by adjusting all rates of mouse was inactive, the variation must metabolism to a single average size (9 stem from a real difference in the resting grams) . This has been done by using the metabolism of each mouse at different series of regression lines in Figure 1. times. I believe that lability of body tem- Where each of these lines crosses the 9- perature is the cause. Harvest mice ex- gram ordinate, that value is taken as the posed to cold and hunger in box traps appropriate rate for a "standard" 9-gram sometimes are found to be torpid and harvest mouse and is used in Figure 2. with a cold body temperature. If they The middle curve in Figure 2 shows are tagged and released, they can be re- that the minimum rate of oxygen con- captured in good health at subsequent sumption of harvest mice (2.5 cc/g/hr trappings, demonstrating that harvest ^or a 9-gram mouse) is reached at the mice have a labile body temperature and relatively high ambient temperature of can recover trom profound hypothermia. ^^° or 34° C. and that there is almost no During the metaboUc tests at 1° C, espe- zone of thermal neutrality. Rate of me- cially those with the mouse in a nest, tabolism almost certainly begins to in- there was a tendency for most of the crease before 36° C. is reached so that the measurements to lie at one level; but zone of minimum metabolism could not there would be a few very low readings include more than 3°. The critical tem- and a few intermediate readings, pre- perature (33-34°) is remarkably close to sumably as the animal entered and the upper lethal temperature. The single emerged from the low-metaboUc condi- animal tested at 37° died after two hours tion (best shown by the U^-gram mouse at this temperature but provided several in Fig. 1). In response to cold coupled good measurements before entering the 188 12 II ^ 10 (T I o B 8 z 9 7 I- Q. 2 6 3 in §5 o z 4 UJ o ^ 3 o \ ^"^'f^ONEST •NONESr 8 9 10 WEIGHT IN GRAMS II 12 Fig. 1.— The relation beUveen body weight and rale of oxygen consumption under different conditions, showing also the variation in individual measurements. Each cluster or vertical array of points represents a series of values obtained from a single individual. 189 METABOLISM OF HARVEST MICE 155 final coma. Because of the large exposed surface of calcium chloride and soda lime in the metabolic chamber, relative hu- midity was probably low; heat death would probably occur at an even lower temperature under humid conditions in which cooling by evaporation would be limited. ercd body temperature. Inclusion of these low values causes the apparent de- crease of the slope of the two curves be- tween 12° and 1°. No body temperatures, however, dropped to the torpid level. Rcithrodontomys megalotis is able to main- tain its temperature well above the tor- pid level even when sleeping in cold sur- I I I I I T— r 10 _9 O ' z6 o Q. ^ z THREE MICE HUDDLED uj ^ >• X o r I I I I I - NOT HUDDLED r p •XX • Xf '■■■'■■'■ ■ ■ ' ■ X I I I BODY TEMR^< ■'''''■■*' 8 12 16 20 24 TEMPERATURE °C 28 32 36 Fig. 2. — The rate of oxygen consumption of resting harvest mice at different temperatures in a nest, without nest, and without fur. All three curves have been adjusted, on the basis of the regression Hnes shown in Fig. 1, to represent a 9-gram mouse. Triangles indicate rate of oxygen consumption of three mice huddled together without a nest compared with the expected rate for the same three mice singly (average weight 8.5 grams). I am grateful to Martin Murie for supplying the value for deej) body temperature, which was the average of manj^ determinations made during the day and night at ambient temperatures between 14° and 27° C. The increase in rate of metabolism at cool temperatures is almost linear be- tween ?>i° and 12°; each drop of 1° C. causes an increase in the rate of oxygen consumption of 0.27 cc/g/hr. This rate of change, possibly because of the small size of harvest mice, is greater than that of any of the rodents listed by Morrison and Ryser (1951) and by Dawson (1955). The averages used for the two points at 1° C. include several low values obtained while the animals probably had a slightly low- roundings. In this respect it differs from the pocket mouse {Perognathus longi- membris), a mouse with which it should be compared because of its similarly small size. When pocket mice are caged at cold temperatures with adequate food, they either drop into torpor or are continually awake and active. They may even be unable to maintain a high body temperature during a prolonged period of sleep at cool temperatures (Bartholo- mew and Cade, 1957). 190 156 OLRER P. PEARSON The only other report on the rate of oxygen consumption of harvest mice lists a rate of 3.8 cc/g/ hr at 24° C. for mice with an average weight of 9.6 grams (Pearson, 1948a). This rate is almost 10 per cent lower than the comparable rate obtained from I'lgure 1 and is below the range of variation obtained at this tem- perature. The difference may be ac- counted for by the fact that the mice used in the earlier study were acclimated to a warmer temperature (for discussion of the effect of acclimatization on me- tabolism see Hart, 1957). INSl'LATING EFFECTIVENESS OF FL'R Figure 2 shows also the metabolic ef- fect of removing all the fur (277 mg. in cent at intermediate temperatures and 24 per cent at 1° C. (lowest curve in Fig. 2). To obtain these measurements, individ- ual mice placed in the metabolic chamber were provided with a harvest mouse nest collected from the wild (shredded grass and down from Compositae), and this the mouse ciuickly rebuilt into an almost- complete hollow sphere about tliree inches in diameter. Metabolic rates were counted only when a mouse was resting Cjuietly deep in the nest. THERMAL ECONOMY OF IIIDDLINC, The metabolic economy of huddling was measured on one occasion with three mice at an environmental temperature of 1° C. without nesting material. The rate 10 o o 7 eh /\r~. \j -VA..AA J V ^V,__ _^ /-v.A/.-/\/v.7 v^- \/ \/"^v 12 I 2 3 4 5 6 7 e 9 10 II 12 I 2 3 4 5 6 7 8 9 10 II 12 PM AM Fig. 3. — Rale of oxygen consumption of a 9-gram harvest mouse for 24 hours at 12° C. the single 8.8-gram specimen used) with an electric clipper. When calculating the points for the curve in Figure 2, 0.28 grams was added to the naked weight and then this rate was adjusted to that for a 9-gram animal on the basis of the regression lines shown in I'lguie 1. The rate of metabolism of the naked mouse was about 35 per cent higher at each of the temperatures used, and the rate in- creased 0.38 cc/g/hr for each 1° C. drop in air temperature. INSULATING EFFECTIVENESS OF NESTS When normal, fully furred mice were given an opportunity to increase their in- sulation by constructing nests, their met- abolic rates were lowered about 17 per of metabolism per gram of huddled mice was 28 per cent less than it would have been for a single one of the mice (Fig. 2). The metabolic saving would probably be greater when more mice were huddled to- gether and less when only two mice were huddled, as is true for feral Mus (Pear- son, 1947) and laboratory mice (Prychod- ko, 1958). 24-HOUR OXYGEN CONSUMPTION IN CAPTIVITY Figure 3 illustrates the rate of oxygen consumption of a mouse kept in the ap- paratus at 12° C. without nesting mate- rial but with food and water for 24 hours. The mouse consumed 1,831 cc. of oxygen to give an average rate of 8.48 cc/g/hr. This is equal to a heat production of 191 METABOLISM OF HARVEST MICE 157 about 8.8 Calories per day. In agreement sumed during rest and during activity is with the fact that activity of harvest proportionately great at warm tempera- mice in the wild is greatest shortly after tures and small at cold temperatures, dusk (Pearson, 1960), the oxygen con- sumption was greatest at that time. The ^^^^^'^ °^- activity on metabolfsm prolonged low i)eriod lasting from aljout An athlete is able, for short periods, to 8:30 to 10:0:) p.m. was unexpected in raise his rate of metabolism to a level 15 this nocturnal animal. to 20 times his basal rate, but small Pearson (1947) used as an indicator of mammals do not match this effort. The the noctuvnality of different species the peak metabolic effort of mice running in ratio of the total amount of oxygen con- ^ wheel is only 6 to 8 times their basal sumed at ni.uht (6:00 p.m. to 6:00 A.m.) rate (Hart, 1950). At (f C. lemmings run- to that consumed in the daytime. I'or the ning in a wheel at a speed of 15 cm/sec harvest mouse described in Mgure 3, the increase their o.xygen consum[)tion less ratio is low- -1.02- but it should be than 35 per cent above the level of rest- pointed out that the record was made at ing lemmings (Hart and Heroux, 1955). 12° C, which is colder than the tempera- At cool ambient temperatures, such as ture used for the species in the earlier re- this, small mammals ex[-)end so much port. Temperature affects the night/day energy at rest that a considerable amount ratio of oxygen consumption because the of activity causes only a proi)ortionately ditterence in amount of oxygen con- small increase in oxygen consumption ; T.VHLK 1 'rill'; 24 iiouK ()XV(;i;\ coxsu.nu' rnix fix (( .) of a 9-(;r.\m harv i:si' Mousi'. nuKixc; DkckmuI'R axi) Junjc at (:)KIX|)A, Camiokxia Willi Wilh Willi- Undcr- With- Undct- out .1,'iounci out {^round Xc^l Xest Xesl Xest Noclur- nal haljit 4 hr. above grouiul at l°C.* 367 367 4 hr. above ground at 12°C.t 297 297 20 hr. under ground 1,548 1,296 20 hr. under ground 1,152 954 at ]0°C.t at 18°C.§ Activity correction |! + 119 2,034 + 119 1,782 cc. Activity correction | + 119 1,568 + 119 1 ,370 cc. (8.55 Ca].)# (6.58 Cal.),f i^iunial lialiit 20 hr. under ground at 10°C.| 1,.548 1,296 20 hr. under grounel at 18°C.§ 1,152 954 4 lir. above ground M3 3M 4 hr. above ground 155 155 at6°C.** at 25° C. ft Activity correctionjl + 119 2,000 + 119 Activit}' correction] + 119 1,426 + 119 1 , 748 cc. 1,228 cc. (8.39 Cal.)# (5.89 Cal.)# * ^fcaii tcniiicrature in runways at time of passage of harvest mite in December. t Mean temperature in runways at time of passage of harvest mice in June. X Underground temperature in December. § Underground temperature in June. II Add 40 per rent of the oxygen consumption on the surface at a temperature of 12° C // Assunii-d 4..S ( al. per liler of oxygen. *'■ .Mean half-hourly temperature in runways l>etwcen 6 a.m. and 6 P.M. in DeLtmbcr. tt Mean half-hoiiily temperature in runways between 6 a.m. and 6 p.m. in June. 192 158 OLIVER P. PEARSON and at cold temperatures the metabolic 24-hour metabolism in the field cost of keeping warm may be so high as The preceding observations indicate to leave little or no capacity tor exercise that ambient temperature is a much (Hart, 1953). During measurement of the more important variable than activity in resting metabolism of harvest mice, nu- the 24-hour energy budget of harvest meious measuring periods had to be dis- mice in the wild. By use of automatic de- carded because the mouse was moving vices that record the temperature in around in the metabolism chamber. Such mouse runway? whenever a mouse passes activity rarely raised the oxygen con- by, the temperature encountered by har- sumption more than 40 per cent above vest mice during their nightly periods of the level of a resting animal at the same activity are known (Pearson, 1960). I temperature. During the 24-hour run at have also recorded throughout the year 12° C, the highest metabolic rate oc- the temperature five inches below the curred during an 11-minute period when surface of the ground. This gives an ap- the average oxygen consumption was proximation of the temperature encoun- 10.36 cc/g/hr. This is only 40 per cent tered by the mice while they are in their greater than the lowest rate recorded for retreats during the daytime. Some of that mouse during any one measuring these surface and underground tempera- period. The maximum metabolic effort ture measurements have been used m the recorded for any harvest mouse was that calculations summarized m Table L of an 8.6-gram mouse at 1° C. This ani- ^ To complete the calculations m Table , • ^ 1 • • 1 • J 1, it has been necessary to estimate how mal persisted in gnawing, exploring, and , , , ^/^, , . , , , , r many hours of each 24 the mouse spends trying to escape from the chamber tor /, . r .lu i i u •^ ^, V T^ • 1^ • ori the surface of the ground and how more than two hours. During one lO-mm- ^^^^ ^^^^^ ^^^ ^^^^^^^ ^^ ^^^^ ^^^^ ute period its oxygen consumption aver- ^^^^^^ ^^ j ^^^^ ^^^^ ^^ estimate based aged 15.8 cc/g/hr, which is 50 per cent ^^ ^^^ behavior of captive animals and higher than the rate of a resting mouse at ^^ automatic recordings made at the exit the same air temperature and six times of ^n underground nest box being used by the minimum value for the species at ^ild harvest mice. Admittedly this esti- thermal neutrahty. This is probably not mate (4 hours on the surface each night) far from the peak metabolic effort of the could be wrong by 50 per cent or more, species. but it should be noted that an error of On several occasions I have watched two hours in this estimate would only undisturbed harvest mice carrying on alter the answer (the total 24-hour me- their normal activities in the wild, and I tabolism) by about 25 per cent. Assum- have been impressed by their leisurely ing that the rate of oxygen consumption approach to life. Hard physical labor and during above-ground activity is 40 per strenuous exercise must occur quite in- cent higher than the rate of a mouse rest- frequently. Most normal activities of ing at 12° C. (see above), the activity harvest mice are probably accomplished correction used in Table 1 can be calcu- without a rise in metabolic rate more lated. than 50 per cent above what it would be In 24 hours in December a harvest in a resting animal at the same air tem- mouse uses 8.55 Calories, and in June, perature. 6.58 Calories (Table 1), assuming that 193 METABOLISM OF HARVEST MICE 159 the mouse has the benefit of a nest. A nest reduces his daily energy budget by about 12 per cent. These estimates of daily metabolic demands seem reason- able when compared with the values ac- tually obtained by measuring the 24-hour oxygen consumption of captive animals, as reported above. The average metabol- ic impact, or daily degradation of en- ergy, by a single harvest mouse should be somewhere between that in December and that in June, perhaps 7.6 Calories. This is about the same as that of a hum- mingbird in the wild (Pearson, 1954)— less than half that of a much heavier English sparrow (Davis, 1955). BIOENERGETICS In seasons when harvest mice are abun- dant, there may be twelve of them per acre (Brant, 1953). At that population density, the species would be dissipating at the rate of 91 Calories per acre per day the solar energy captured by photosyn- thesis, or something like ^ of 1 per cent of the energy stored each day by the plants in good harvest-mouse habitat in the Orinda area. This percentage was cal- culated using a net productivity of 20,000 Cal/acre/day, which was esti- mated by assuming 4 Calories per gram of dry vegetation (based on data in Brody, 1945, pp. 35, 788) and an annual crop of 1,800 kg. of dry vegetation per acre (based on Bentley and Talbot, 1951). The harvest mice on this hypothetical acre are causing about the same caloric drain on the environment as all the small mammals in the acre of forest described by Pearson (19486). By using caloric units, direct compari- son can be made of the metabolic impact of different species, as in the example above. Similarly, the metabolic cost of different activities and different habits can be compared (Pearson, 1954). For example, harvest mice are strongly noc- turnal (Pearson, 1960), in spite of the fact that air temperatures are much colder at night and force mice to con- sume more oxygen and more food than if they were diurnal. Since evolution has permitted nocturnahty to persist, it seems logical to assume that the value of nocturnality to harvest mice is greater than the metabolic cost. I estimate that during a 24-hour period in December a 9-gram harvest mouse uses 0.16 more Calories by being nocturnal than it would if it were diurnal (Table 1). In summer, the difference is even greater, 0.69 Calories. The average is 0.42 Cal- ories, or about 3| grains of wheat. This is a rough estimate of the price each har- vest mouse pays for nocturnality. Some environmental pressure makes harvest mice remain nocturnal, and this pressure must be more than 0.42 Calories per mouse per day. If harvest-mouse noc- turnality evolved for one reason only— to avoid predation by hawks — -then we would have discovered a minimum esti- mate of the predation pressure of hawks on harvest mice. Surely the situation is not this simple; nevertheless, it is inter- esting to measure the pressure that makes harvest mice nocturnal even if the cause of the pressure is not known. SUMMARY Oxygen consumption of harvest mice reaches a minimum of 2.5 cc/g/hr at an ambient temperature of 33° C, and the zone of thermal neutrality is not more than 3°. Each drop of 1° in ambient tem- perature causes an increase in the rate of metabolism of 0.27 cc/g/hr. Removing the fur raises the rate of metabolism about 35 per cent, and use of a nest lowers it 17 to 24 per cent. Huddling by three mice at 1° reduces the rate 28 per cent. 194 160 OLIVER P. PEARSON Jvxcrcise at cool tcm{)eraturcs causes a relatively small increase in the rate o£ metabolism, whereas change of ambient temperature has a great effect! Making use of the temperatures that harvest mice are known to encounter in the wild, the 24-hour oxygen consumption of a wild harvest mouse was calculated to be 1,782 cc. in December and 1,370 cc. in June. The average (1,576 cc.) is equivalent to about 7.6 Calories per day. A dense p(Jp- ulation of harvest mice would dissipate about 91 Calories per day per acre, which is about ^ of 1 per cent of the energy stored by the plants each day. By being nocturnal, harvest mice en- counter cooler temperatures, and this habit increases the daily energy budget of each mouse by 0.42 Calories, or about 3| grains of wheat. LITERATURE CITED Bartholomew, G. A., and C.a.de, T. J. 1957. Tem- ])erature regulation, hibernation, and aestivation in the little pocket mouse, Perognatlnts longintem- bris. Jour. Mammal., 38:60-72. Bentley, J. R., and T.^lbot, M. W. 195L Efficient use of annual plants on cattle ranges in the California foothills. U.S. Dept. .Agriculture, Cir- cular No. 870, 52 pp. Brant, D. H. 1953. Small mammal populations near Berkeley, California: Reithrodontomys, Peromys- cus, Microtus. Doctoral thesis, University of Cali- fornia, Berkeley. Brody, S. B. 1945. Bioenergetics and growth. New- York: Reinhold Publishing Corp. D.xvis, E. A., Jr. 1955. Seasonal changes in the energy balance of the English Sparrow. Auk, 72:385-411. Dawson, VV. R. 1955. The relation of oxygen con- sumption to temperature in desert rodents. Jour. Mammal., 36:543-53. De Bont, A.-F. 1945. Metabolisme de repos de quelques es])eces d'oiseaux. .Ann. Soc. Roy. Zool. Belgique, 75 (1944) :75-80. Hart, J. S. 1950. Interrelations of daily metabolic cycle, activity, and environmental temperature of mice. Canadian Jour. Research, D, 28:293- 307. . 1953. The influence of thermal acclimation on limitation of running activity by cold in deer mice. Canadian Jour. Zool., 31:117-20. . 1957. Climatic and temperature induced changes in the energetics of homeotherms. Revue Canadienne de biol., 16:133-74. Hart, J. S., and Heroux, O. 1955. Exercise and temperature regulation in lemmings and rabbits. Canadian Jour. Biochem. & Physiol., 33:428-35. Morrison, P. R. 1947. An automatic api)aratus for the determination of oxygen consumption. Jour. Biol. Chem., 169:667-79. Morrison, P. R., and Ryser, F. .\. 1951. Tempera- ture and metabolism in some Wisconsin mam- mals. Federation Proc, 10:93-94. . 1959. Body temperature in the white-footed mouse, Peromyscus leiicopus noveboracensis. Phvs- iol. Zool., 32:90-103. Pearson, 0. P. 1947. The rate of metabolism of some small mammals. Ecology, 28:127-45. . 1948a. Metabolism of small mammals, with remarks on the lower limit of mammalian size. Science, 108:44. . 19486. Metabolism and bioenergetics. Sci- entific Monthly, 66:131-34. . 1954. The daily energy requirements of a wild Anna Hummingbird. Condor, 56:317-22. 1960. Habits of harvest mice revealed by automatic photographic recorders. Jour. Mam- mal, (in press). Prychodko, W. 1958. Effect of aggregation of labo- ratory mice {Mus musculus) on food intake at different temperatures. Ecology, 39:500-503. Reprinted for private circulation from PHYSIOLOGICAL ZOOLOGY Vol. XXXIII, No. 2, April 1960 Copyright 1960 by the University of Chicago 195 OXYGEN CONSUMPTION, ESTIVATION, AND HIBERNATION IN THE KANGAROO MOUSE, MICRODIPODOPS PALLIDUS^ GEORGE A. BARTHOLOMEW AND RICHARD E. MacMILLEN Departments of Zoology, University of California, Los Angeles, and Pomona College, Claremont, California THE pallid kangaroo mouse occurs material and methods only in the desert parts of western Experimental animals— The twenty- Nevada and extreme eastern Call- ^^iree kangaroo mice used were trapped fornia. Its habitat is restricted to areas of ^^ ^^^^ ^^^^^ f^^^ rniles south of Arle- fine sand which support some plant ^^^^ Ranch, Esmeralda County, Neva- growth. Like its relatives, the kangaroo ^^^ j^^ ^pj.-^^ ^959^ ^nd May, 1960. They rats (Dipodomys) and the pocket mice ^^^^ housed individually in small ter- {Perognathus) , it is nocturnal, fossorial, ^^^^^ p^^^^y ^^g^j ^ith ftne sand, kept in a and gramnivorous and can under some ^indowless room on a photoperiod of 12 circumstances live indefinitely on a dry j^^^j.^^ ^^^ f^^ ^^ ^ ^i^^ of j^ixed bird diet without drinking water. The genera] ^^^^ supplemented occasionally with life history (Hall and Linsdale, 1929) of ^^^-^^ p^g^g^ ^f cabbage. Survival was this kangaroo mouse and the details of excellent, and some of the animals were its distribution (Hall, 1946) are known, ^^^^ f^j. ^^gj. ^g^^ months, but virtually no quantitative data on its ^^^y temperatures— AW temperatures physiology are available. ^g^g measured with 30-gauge copper- The present study was undertaken to gonstantan thermocouples connected to compare the thermoregulation of Micro- ^ recording potentiometer. All body tem- dipodops with that of the better-known pg^atures were taken orally by inserting genera, Dipodomys and Perognathus. ^ thermocouple to a depth of at least These three genera belong to the family 2 cm Heteromyidae, which has been more sue- Ambient temperatures— The ambient cessful in occupying the and parts of tgn^pg^atures were monitored with ther- Tof'of mTmrnalT''''^ ^^^"^ ^""^ *'^^'' "^«^«^P^^^ ^^^ controlled by insulated ^ P ^ ■ chambers equipped with automatic heat- 1 This study was aided in part by a contract be- ing and cooling units, blowers, and tween the Office of Naval Research, Department of Uahts the Navy, and the University of California (Nonr ° ' • r\ 266[3ll). Oxygen consumption. — Oxygen con- 177 196 178 GEORGE A. BARTHOLOMEW AND RICHARD E. MacMILLEN sumption was measured by placing a mouse in an air-tight 500-cc. glass con- tainer equipped with a thermocouple and ports for the introduction and removal of air. The bottom of the container was covered to a depth of about 1 cm. with fine dry sand. The glass container with animal inside was placed in a tempera- ture-control chamber, and dry air was metered through the container at a rate the response of body temperature {Tb) to moderately low ambient temperatures {Ta), kangaroo mice were placed at Ta of 7°-9° C. for five days starting May 11, 1959, with food available in excess; measurements of Tb were made at 24- hour intervals. There were no apparent changes in Tb during the test period, nor was the mean Tb significantly different from that of animals maintained at room 42 o o q: 40 iij i 38 ffi 36 o o I o o I ^. OJ CO B o o in in" to I If) ^" ro D o o lO d I m ro 42 40 38 36 Fig. 1. — Body temperatures of M. pallidus at various ambient temperatures. A, 47 measurements on twelve animals; B, 38 measurements on thirteen animals; C, 7 measurements on four animals; D, 9 measure- ments on four animals (three other animals tested at this temperature died). The horizontal lines indicate the means {M). The rectangles inclose M ± a^- The vertical lines indicate the range. of 250 cc/min and then delivered to a Beckman paramagnetic oxygen analyzer which, used in conjunction with a re- cording potentiometer, gave a continu- ous record of oxygen consumption. All data used were from post-absorptive animals. RESULTS Body temperature during normal ac- tivity.— Normally active animals kept at room temperature (22.4°-25.4° C.) had body temperatures ranging from slightly less than 37° to as high as 41° C, with a mean of 38.8° C. (Fig. 1). To determine temperature. The animals appeared nor- mally active and unaffected by the change in environmental temperature. Animals were maintained at Ta of 37.5°-40.5° C. for 24 hours to test their response to moderately high environ- mental temperatures. They showed a conspicuous elevation in Tb with a mean almost 2° C. higher than that of animals at room temperature. Animals main- tained at Ta close to 35° C. also became hyperthermic and showed a mean Tb in- termediate between that of animals held at room temperatures and those held at 39° C. There was no mortality in animals 197 THERMOREGULATION IN THE KANGAROO MOUSE 179 held at 35° C, but exposure to 39° C. for more than a few hours killed three out of the seven animals tested. At a high Ta the kangaroo mice did not salivate or pant; they merely sprawled out fiat on the sand with legs extended and lower jaw and neck prone on the substrate. This prone posture alternated with brief bursts of intense activity characterized by repeated shifts in position and much digging and moving of sand. gm.) is 1.8 cc 02/gm/hr when the for- mula M = 3.811^-'' " jg ygg(j (^ggg Brody, 1945, and Morrison, Ryser, and Dawe, 1959). The observed basal metabolism of our kangaroo mice (mean, 1.3 ± 0.2 cc 02/gm/hr) was about three-fourths of the predicted value. This relatively low figure is consistent with the obser- vation on some other heteromyids (Daw- son, 1955). The only comparative data on the a: X o o -|4 •• • • • • • • • • • • • 10 20 30 40 Fig. 2. — The relation of oxygen consumption to ambient temperature. Data obtained from ten animals. Each point represents the minimum level of oxygen consumption maintained by an animal for half an hour. Oxygen volumes are corrected to 0° C. and 760 mm. (Hg.) pressure. Oxygen consumption. — The relation of oxygen consumption to Ta is summarized in Figure 2. There is no clearly defined zone of thermal neutrality, but oxygen consumption is minimal at about 35° C. The increase in oxygen consumption at temperatures above 35° C. is relatively more rapid than is the increase below this point of thermal neutrality. No dif- ferences in oxygen consumption were ap- parent between males and females. The calculated metabolism of Micro- dipodops paUidus (mean weight, 15.2 energy metabolism of Microdipodops is that of Pearson (1948) on M. megacepha- lus. Pearson's data, obtained at tempera- tures near 24° C. from animals that were not post-absorptive, gave oxygen con- sumptions of 3.4-3.7 cc 02/gm/hr. Pear- son's measurements, as might be expect- ed from the fact that he was not using post-absorptive animals, are higher than our determinations of 2.7 cc 02/gm/hr at 25° C. Hibernation and estivation. — No infor- 198 180 GEORGE A. BARTHOLOMEW AND RICHARD E. MacMILLEN mation on hibernation or estivation is available for Microdipodops. Hall (1946, p. 386) pointed out that kangaroo mice are often active above ground in temper- atures many degrees below freezing, and Ingles (1954, p. 214) suggested that kangaroo mice probably do not hiber- nate. Under laboratory conditions we found that kangaroo mice at any time of year 40 r there are no conspicuous physiological differences between arousal from spon- taneous dormancy and that from induced dormancy. Animals dormant at room tempera- tures (estivating) started to arouse im- mediately upon being handled. The rate of increase in Tb varied but usually fell between 0.5° and 0.8° C. per minute. Usually within 20 minutes of the onset -.40 10 20 30 40 50 MINUTES FROM START OF AROUSAL Fig. 3. — Increases in oral temperatures in nine kangaroo mice during arousal from torpor. All arousals took place in ambient temperatures between 23° and 26° C. Temperatures taken manually with thermo- couples. The five upper animals were dormant at room temperature (22°-25° C); the four lower animals were dormant at 5°-8° C. will spontaneously become dormant at ambient temperatures ranging at least from 5° to 26° C. and can readily be in- duced to hibernate (or estivate) over this range of temperatures by reduction of food for 24 hours or less. Body temperature and behavior dur- ing entry into torpor were not recorded, but the animals apparently entered tor- por in the crouching posture normally used in sleeping. Dormant animals had body temperatures l°-2° C. above am- bient. Judging from the course of body temperature during arousal from torpor, of arousal the animals attained their nor- mal operating temperature, and within as little as 12-15 minutes from the start of arousal they appeared to behave nor- mally, even though Tb approximated 30° C. Arousal from low temperatures was essentially the same as arousal from high temperatures (Fig. 3). However, animals arousing from low temperatures attained maximal body temperatures about 1° C. higher than did those arous- ing from room temperature. Incidental to the measurement of Tb the relations of various types of behavior 199 THERMOREGULATION IN THE KANGAROO MOUSE 181 to body temperature were noted during mals the ability to become dormant and nine arousals. Mice unsuccessfully at- to decrease body temperature and meta- tempted to right themselves when turned bolic activity may be more useful in the over at Tb between 16.1° and 18.2° C. summer than in the winter, and it may and successfully righted themselves at be as important for water conservation Tb between 19.0° and 22.0° C. The first as for energy conservation, vocalizations were given at Tb between Kangaroo mice are unique among 24.7° and 28.6° C. Grain was available heteromyids in having conspicuous de- to the animals during arousal, and seven posits of adipose tissue in the proximal of the nine animals ate during arousal, third of the tail, which is considerably The lowest Tb for eating was 25.5° C, larger than either its base or its distal and three animals ate at temperatures half. Hall (1946, p. 379) suggests that the between 25° and 29.4° C. The mean Tb fleshiness of the tail permits it to func- for onset of visible shivering for seven tion in balancing. However, since these animals was 25.5° C. Two of the nine mice hibernate but do not show conspicu- animals observed did not visibly shiver ous seasonal deposits of subcutaneous fat during arousal. Shivering usually stopped over the body as a whole, it seems reason- at a Tb of 34°-35° C . able to suggest that the fat tail serves as a reserve of energy for use during periods DISCUSSION q£ torpor. In the laboratory with food The general features of thermoregula- available in excess, many of the kangaroo tionin Microdipodops pallidus dirtsirmlzr mice showed a marked increase in tail to those of the related genus Perognathus diameter. in that both show well-developed pat- Our data (Fig. 1) show almost no in- terns of hibernation and estivation, es- dication of a discrete zone of thermal sentially normal behavior at Tb below neutrality for the kangaroo mouse. Its 35° C, obligate hyperthermia at Ta critical temperature is unusually high for above 35° C, and no apparent salivary an animal living in an area characterized response to elevated body temperature, by cold winters. For months on end kan- Microdipodops differs from the related garoo mice can be active only at tempera- genus Dipodomys in that the latter does tures below thermal neutrality. Presum- not readily become dormant at either ably, their energetic and thermal prob- high or low temperatures and does use lems are reduced in cold weather by pe- salivation as an emergency thermoregu- riodic episodes of torpor. It is of interest latory response (Schmidt-Nielsen and that we captured our animals on nights Schmidt-Nielsen, 1952). when environmental temperatures went In kangaroo mice, as in Perognathus below —10° C, and Hall (1946, p. 396) longimemhris (Bartholomew and Cade, reports that these animals are often "ac- 1957) and Citellus mohavensis (Bartholo- tive on nights when the temperature is so mew and Hudson, I960-), there appears to low as to freeze to a state of stiffness the be no sharp physiological differentiation bodies of mice caught in traps." Thus, al- between hibernation and estivation. This though they can hibernate, they are underscores the point that the faculta- also commonly active during subfreezing tive hypothermia shown by mammals weather. should not be thought of only as an adap- This species has remarkably shallow tive response to low environmental tem- burrows, often no deeper than 4 inches peratures; at least for small desert mam- (Hall, 1946, p. 396). Consequently, when 200 182 GEORGE A. BARTHOLOMEW AND RICHARD E. MacMILLEN high daytime temperatures occur, at in Microdipodops correlates nicely with least some members of the population its strong tendency toward hyperthermia may be exposed to temperatures near at high ambient temperatures. For ani- 35° C. It is possible, therefore, that the mals living in a desert environment high point of thermal neutrality of this where water is usually in short supply, species allows a significant metabolic hyperthermia is a more advantageous re- economy and a significant reduction in sponse to heat than is evaporative cool- pulmocutaneous water loss during the ing. severely hot desert summers. summary Extrapolation of the plot of metabo- Microdipodops pallidus occurs only on lism against ambient temperature below gp^^g^iy vegetated sand dunes in the thermal neutrality does not intersect the ^^^^^^ ^^^^^ ^^ western Nevada and abscissa within the usual range of body ^^^^^^^ California. In the absence of temperature (38°-39°C.) of kangaroo ^-gj^pei-ature stress body temperature, mice (Fig. 2). This means that, unlike j^ averages 38.8°C. There is no diminu- some of the species considered by Scho- ^j^^ ^^ j.^ ^^-^^^ decreasing ambient tem- lander et al. (1950), and unlike the perature, T^, at least to 8° C. However, masked shrew, Sorex cinereus (Morrison, hyperthermia is apparent at a Ta of Ryser, and Dawe, 1959), the kangaroo 350 ^ ^^^ ^^ 390 ^ j^^ averages 40.5° C. mouse does not follow Newton's empiri- E^pos^re for more than a few hours to cal law of cooling in a simple and direct 390 ^ j^ ^^^^^ lethal. At high ambient manner. The failure to follow the pattern temperatures kangaroo mice neither predicted by Newton's law of cooling p^^^ ^^^ ^^^^^ ^j^^y ^^^^ no clearly de- may be related to the fact that kangaroo ^^^^ ^^^^ ^f thermal neutrality; oxygen mice start to become hyperthermic as consumption is minimal at 35° C. and in- they approach their critical temperature ^^^^^^^ ^lore rapidly at temperatures (Fig. 1), and it suggests that the relation ^^^^^ ^^^^ ^^-^^^ ^^^^ ^^i^^ it. Basal between skin and ambient temperature ^letabolism is 25 per cent less than that in this species differs from the usual pat- pj-g^i^ted on the basis of body size. Kan- tern. It is of interest that Pearson's data ^^^^^ ^^^^ ^^^ capable of both estivation (1960) for Reithrodontomys show a situa- ^^^ hibernation. In the laboratory they tion similar to that reported here for ^^^^^ become dormant at ambient tem- Microc^z>(^o^5, that is, almost no zone of pg^atures ranging at least from 5° to thermal neutrality, a high critical tem- 25° c. The rate of temperature increase perature, and a failure of the curve of ^^^j^^g arousal at room temperature is metabolism against ambient temperature 0 5o__0.8°C. per minute. Terminal body to intersect the abscissa at the usual body temperatures after arousal from low tem- temperature. Although Pearson does not pej-^tures averaged about 1° C. higher comment on this point, it appears that in ^^iQ^n after arousal from room tempera- Reithrodontomys as in Microdipodops the ^^^.g gy ^^g time the Tb of arousing ani- curve of metabolism against ambient ^lals reaches 30° C, their behavior ap- temperature intersects the abscissa at a pears normal. The thermoregulatory re- point above the lethal temperature for sponses of kangaroo mice are compared the species. with those of other desert heteromyids. The apparent absence of a marked in- and the ecological significance of their crease in salivation at high temperatures physiological capacities is discussed. 201 THERMOREGULATION IN THE KANGAROO MOUSE 183 LITERATURE CITED Bartholomew, G. A., and Cade, T. J. 1957. Tem- perature regulation, hibernation, and aestivation in the Uttle pocket mouse, Perognatlms longimem- bris. Jour. Mamm., 38:60-72. Bartholomew, G. A., and Hudson, J. W. 1960. Aestivation in the Mohave ground squirrel, Citel- lus mohavensis. Bull. Mus. Comp. Zool., 124:193- 208. Brody, S. 1945. Bioenergetics and growth. New York: Reinhold Publishing Co. Dawson, W. R. 1955. The relation of oxygen con- sumption to temperature in desert rodents. Jour. Mamm., 36:543-53. Ingles, L. G. 1954. Mammals of California and its coastal waters. Stanford, Calif.: Stanford Univer- sity Press. Hall, E. R. 1946. Mammals of Nevada. Berkeley: University California Press. Hall, E. R., and Linsdale, J. M. 1929. Notes on the life history of the kangaroo mouse {Microdi- podops). Jour. Mamm., 10:298-305. Lyman, C. P., and Chatfield, P. 0. 1955. Physiol- ogy of hibernation in mammals. Physiol. Rev., 35:403-25. Morrison, P., Ryser, F. A., and Dawe, A. R. 1959. Studies on the physiology of the masked shrew Sorex cinereus. Physiol. Zool., 32:256-71. Pearson, 0. P. 1948. Metabolism of small mam- mals, with remarks on the lower limit of mam- malian size. Science, 108:44. . 1960. The oxygen consumption and bio- energetics of harvest mice. Physiol. Zool., 33: 152-60. Schmidt-Nielsen, K., and Schmidt-Nielsen, B. 1952. Water metabolism of desert mammals. Physiol. Rev., 32:135-66. Scholander, p. F., Hock, R., Walters, V., John- son, F., and Irving, L. 1950. Heat regulation in some arctic and tropical mammals and birds. Biol. Bull., 99:237-58. Reprinted for private circulation from PHYSIOLOGICAL ZOOLOGY Vol. XXXIV, No. 3, July 1961 Copyright 1961 by the University of Chicago PRINTED IN U.S.A. 202 Counter-Current Vascular Heat Exchange in the Fins of Whales^ p. F. SCHOLANDER and WILLIAM E. SCHEVILL. From the Woods Hole Oceanographic Instilution, Woods Hole, Afassaclmsetts IT MAY BE a source of wonder that whales swimming about in the icy waters of the polar seas can maintain a normal mam- malian body temperature. What prevents them from being chilled to death from heat loss through their large thin fins?- These are well enough vascularized to justify the question (fig. i). One may conjecture that a whale may be so well insulated by its blubber that it needs such large surfaces to dissipate its heat. On the other hand, if heat conservation is at a pre- mium, there must be some mechanism whereby the fins can be circulated without losing much heat to the water. One may point to two circulatory factors which would reduce the heat loss from the fin: a) slow rate of blood flow and, 6) precooling of the arterial blood by veins before it enters the fin. Bazett and his coworkers (i) found that in man the brachial artery could lose as much as 3°C/decimeter to the two venae comitantes. This simple counter-current exchange system is a mere rudiment compared to the multi- channelled arteriovenous blood vascular bun- dles which we find at the base of the extremities in a variety of aquatic and terrestrial mammals and birds. These long recognized structures have most recently been studied by Wislocki (2), Wislocki and Straus (3) and Fawcett (4). The function of these bundles has long been a mystery. No matter what else they do, they must exchange heat between the arteries and veins, and it has been pointed out that they very likely play an important role in the pres- ervation and regulation of the body heat of many mammals and birds (5). In the present study we describe a conspic- uous arteriovenous counter-current system in the fins and flukes of whales, which we interpret as organs for heat preservation. Received for publication July 21, 1955. ' Contribution Number 807 from the Woods Hole Oceanographic Institution. ^ In 'fin' we include the structures more specifically called flippers (pectoral fins), flukes (caudal fins) and dorsal fin. MATERIAL Two species of porpoises have been studied : Lageno- rliynclius aculits: dorsal fin, tail-fluke, and flipper of an adult female collected 50 miles east of Cape Cod; Tursiops tnmcatus: dorsal fin and tail-fluke of a 4- month-old calf from Florida, supplied through the courtesy of the Marineland Research Laboratory. Lagenorhynchiis is a genus of fairly high latitudes, the southern limit of L. acutns being about latitude 4i°N. on the New England coast and about 55°N. in the British Isles. It has been caught at least as far north as latitude 64°N. in west Greenland and Nor- wegian waters. Tursiops is found in lower latitudes, T. truncalus overlapping slightly with L. aciihts and occurring south into the tropics. DESCRIPTION Figure 2 illustrates the vascular supply at the base of the dorsal, pectoral and caudal fins. It may be seen that all major arteries are located centrally within a trabeculate venous channel. This results in two concentric con- duits, with the warm one inside. In addition to the circumarterial venous channels there are separate superficial veins, as seen in figure 2. The circumarterial venous channels are con- spicuously thin walled compared to the simple veins, as may be seen in figure 3. When an artery was perfused with saline, the solution returned through both of these venous systems. INTERPRETATION Based on the anatomical findings and on the perfusion experiments, we interpret the artery- within-vein arrangement as a heat-conserving counter-current system, as schematically pre- sented in fig. 4. In such an arrangement the warm arterial blood is cooled by the venous blood which has been chilled in the fin. The result is a steep proximodistal temperature drop from the body into the appendage. The heat of the arterial blood does not reach the fin, but is short-circuited back into the body in the venous system. Body heat is therefore conserved at the expense of keeping the appen- dage cold. There is reason to believe that the analogous blood vascular bundle in the proxi- 279 203 28o p. F. SCHOLANDER AND WILLIAM E. SCHEVILL Volume S Fig. I. Arterial supply to the flukes in the common porpoise {Plincoena phocnena), drawn from an x-ray l^icture by Braun (6). mal part of the extremities of sloths serves a similar function, inasmuch as these animals can barely keep warm even in their warm environment (5). Cold extremities have been described in many arctic mammals and birds as important factors for conservation of body heat (7), but to what extent arteriovenous counter-current structures are present in these animals is not known. The efficiency of heat exchange in a system like that diagrammed in figure 4 is related to the blood flow. The slower the flow, the more nearly identical will be the arterial and venous temperatures along the system, and the more efficient will be the heat conservation. At high rates of flow, warm blood will reach the periph- ery and cool venous blood will penetrate into the body.'' It was shown by perfusion experiments on the detached fins that the arterial blood may return via the concentric veins, and/or through the separate superficial veins. As pointed out above, the concentric vein channels are very thin walled and weak compared to the thick- walled superficial veins (fig. 3). One may inter- pret these anatomical facts in the following way. If the animal needs maximal heat con- servation, blood circulation through the fins should be slow, and the venous return should preferentially take place through the counter- current veins. But a slow rate of blood flow would need only weak venous walls, as actually found. If, on the other hand, the animal needed maximal cooling, as during exercise in rela- tively warm water, this would be most effec- tively accomplished by a high rate of blood flow through the fins, with venous return through the superficial veins and the least possible flow through the concentric veins. This would require the strong and thick walls of the superficial veins. One may even see the likelihood of a semiautomatic regulatory func- tion in the concentric system, for when the artery is swelled by increased blood flow, the concentric veins will be more or less obliterated, ^ The theory for a multichannel counter-current system has been elaborated in connection with the swimbladder in deep sea fishes (8). L. CAUDAL „I1„„IIU. .1 iv II,,.. |i»i DORSAL r— 1 >— I 1 1 O CEMTIMETERS .S TUR5I0P5 TRUhCflTUS L.PECTORflL mmumasssiiiiMsamm LAGEMORHYMCHU^ AOUTUS Fig. 2. Sections near base of fins and flukes of two species of jiorpoises. Each artery is sur- rounded by a multiple venous channel. Simple veins are near the skin (only the larger ones are indi- cated). 204 COUNTER-CURRENT HEAT EXCHANGE IN W HAEE FINS 281 Fig. 3. Sections from Tursiops Inincatus. (Courtesy of the Department of Anatomy. Harvard Medical School.) A. From tail-fluke. Upper: artery surrounded by thin-walled venous channels. Lower: superficial single thick- walled vein in the hypodermis. (X 9) B. From dorsal fin. Artery surrounded hy thin-walled venous channels (X 12) but will remain open when the diameter of the artery is reduced during decreased blood flow. Thus the anatomical findings fit logically into the simplest possible scheme of heat regulation in the fins. There are a few observations available in- dicative of heat regulation in the fins of por- poises. Tomilin (g) made some observations on an east Siberian 'white-sided dolphin' on deck, and found that the fins could vary be- tween 25° and 33. 5°C, while the body varied only 0.5°. Schevill observ^ed that the flukes in a Florida Tursiops out of water became about 10° warmer than the body surface itself. In both of these cases the animals were probably resisting overheating. On the other hand, Scholander (5) has recorded cold flippers in water-borne common porpoises (Phocoena). The concentric counter-current system of an artery within a vein appears to be a pecu- liarly cetacean arrangement, and we have seen it only in the fins, flippers and flukes of these animals."* This is an impressive example of bioengineering, which, together with other TRUNK 29" 39° h;^////>//,',7/]///////////^/////////^;//7777. ARTERY-«4-iO° 20° 30° 40° -- ■* The present material is from odontocetes, Ijut these structures have also been noted by Scholander in the tail flukes of a mysticete (fin whale). >:::^ TRUNK Fig. 4. Schematic diagram of hj^iothetical tempera- ture gradients in a concentric counter-current system. factors, adapts these homeotherms for a suc- cessful existence in a heat-hungry environment. SUMM.'^RY The vascular supply to the fins and flukes of two species of porpoises, Lagenorhynchus aciitus and Tursiops truncalus, is described. All major arteries entering the fins and flukes 205 282 1'. F. SCHOLANDER AND WILLIAM E. SCHEVILL Volintie 8 are surrounded by a trabeculate venous chan- nel. The arteries drain into these, but also into superlicial simple veins. The artery within the venous channel is interpreted as a heat-con- serving counter-current exchange system. The heat regulatory aspects of the two venous systems are discussed. We wish to express our appreciation to Dr. F. G. Wood, Jr., and the Marineland Research Laboratory, Marineland, Fla., for providing the material of Tur- siops tnmcatus, and to Dr. G. B. Wislocki of the Dept. of Anatomy, Harvard Medical School, Boston, Mass., for providing the histological sections and photographs. REFERENCES 1. B.\ZETT, H. C, L. Love, M. Newton, L. Eisen- BERG, R. Day and R. Forster II. J. Appl. Pliysiol. 1 : 3, 1948. 2. Wislocki, G. B. /. Morpliol. 46: 317, 1928. 3. Wislocki, G. B. and W. L. Straus, Jr. Ball. Mus. Comp. Zool. Harvard 74: i, 1933. 4. Fawcett, D. W. J Morphol. 71: 105, 1942. 5. ScHOLANDER, P. F. Evolution 9: 15, 1955. 6. Braun, M. Zool. Anz. 29: 145-149, 1905. 7. Irving, L. and J. Krog. J. Appl. Pliysiol. 7: 355, 1955- 8. Scholander, p. F. Biol. Bull. 107: 260, 1954. 9. ToMiLiN, A. G. Rybnoe Khozaistvo 26: 50, 1950. (In Russian.) 206 SECTION 3— REPRODUCTION AND DEVELOPMENT Just as animal structures must be adaptive, so must reproductive and devel- opmental patterns. In other words, the organism must be a functioning unit in its particular environment at all times. In our selections, Spencer and Steinhoff discuss the possible functional significance of geographic variation in litter size, and Sharman points out the adaptive value of some peculiarities of kan- garoo reproduction. The interrelationships of reproductive and developmental patterns in the fisher are evident in the account by Wright and Coulter. Super- fetation ( or the fertilization of new ova during gestation, known in kangaroos, rabbits, and some rodents), delayed implantation (which occurs in some mustelids), and delayed fertilization (through sperm storage, known in some bats) all are interesting variations of the reproductive theme, and all are adaptive in certain circumstances. A recent study of reproductive adaptations of the red tree mouse by Hamilton ( 1962, not reprinted here ) related small litters, long gestation, delayed implantation during lactation, and slow develop- ment of young with the limited amount of energy available in the food sources of the mice. The study here reproduced by Jones on the evening bat relates development to function, in this case flight, and also nicely illustrates some quantitative refinements that adequate data provide. The student of any field is well advised to learn what compilations, summa- ries, or collected works are available. Sometimes a summary is short, as is the paper by Hamilton included herein on reproductive rates of some small mam- mals. In the field of mammalian reproduction, the classic summary by Asdell (as revised in 1964), and recent collections of contributions edited by Enders ( 1963) on delayed implantation and by Rowlands ( 1966) on comparative biol- ogy of mammalian reproduction will repay study. Two classic books in the field of development in which relative growth rates were considered at length are On Growth and Form by D'Arcy Went- worth Thompson (1942) and Problems of Relative Growth by Julian Huxley ( 1932 ) . A comparative study of two related species ( Butterworth on Dipod- omys) is included here. A recent and detailed account (too long to include here) of one species in terms of relative growth and in comparison to several other species is the study by Lyne and Verhagen ( 1957 ) on Trichosurus vul- pecula, an Australian brush-tailed possum. A number of studies of single species may be found in the literature. Allen's paper (reprinted here) is of interest because it is one of the earliest to give serious consideration to variation as such and to possible relevance of variation to systematic and other problems. Hall (1926) described at greater length than could be included here and in detail uncommon at that time the changes during growth of the skull in the CaHfornia ground squirrel. Two among the many good recent studies of development of single species are by Layne ( 1960, 1966) on Ochrotomys nuttalli and Peromyscus floridanus. Although we have not included examples of methods of determining age other than the report of Wright and Coulter on the fisher, we must comment that age determination is important in many practical problems of wildlife management as well as in studies of population composition or of growth as 207 such. Managers of deer herds, for example, can examine the teeth of hunter- killed animals using standards developed by Severinghaus (1949) and later workers. The formation of annuli in dental cement in various kinds of mam- mals provides another method for determining age — see Adams and Watkins (1967) on its application to ground squirrels. Epiphyseal growth as observed in X-ray photographs and the use of lens weights are other means ( see Wight and Conaway, 1962, on aging cottontails ) . A short paper on maturational and seasonal molt in the golden mouse con- cludes our selection for this section. Studies of molt in furbearers are, of course, of special economic import, and knowledge of pelage differences related to age, sex, or season are of obvious use in most studies of mammalian popula- tions. 208 THE REPRODUCTIVE RATES OF SOME SMALL MAMMALS By W. J. Hamilton, Jr. Students of cyclic mammal populations realize the necessity of properly evaluat- ing the breeding rate. A proper assessment of the reproductive rate is frequently essential for a correct interpretation of population levels. Without a satisfactory' estimate of breeding rate, conclusions regarding cyclic populations may be in- valid. Unfortunately, it is difl&cult to secure accurate data on the breeding rate in feral species; analogous observations on captive species may not give a true picture of the breeding behavior in wild species. Many field investigators be- lieve that small mammals have successive litters, one following another in rapid succession during the height of the breeding season. On the other hand, there are those who argue that a post-partus oestrus in wild species seldom occurs. They insist that mating during the lactation period is rare under natural condi- tions; captive individuals alone exhibit this phenomenon due to crowding or other factors imposed by laboratory conditions. Since the subject is an important one in population studies, I present such data as are available to demonstrate that certain shrews, mice, and other small mammals are not only capable, but do mate successfully following partus. Actively lactating species are upon occasion gravid. 209 258 JOURNAL OF MAMMALOGY Vol. SO, No. S In an important paper on discontinuous development in mammals, Hamlett (1935) remarks that in a fairly extensive series of mice of various kinds, including free living house mice, Norway rats, and wild species of deer mice and field mice, pregnant suckling females were never found. He concludes that copulation im- mediately after parturition in mice is a response to domestication, and is rare or lacking in free living races. Pearson (1944) examined early pregnant females of Blarina in which placental scars were visible and the manmiary glands greatly developed, indicating that these individuals had been nursing young recently. He suggests the possibility that the young may have been lost or destroyed a short time before, and that mating took place after the loss of the young. Pear- son believes that shrews with advanced mammary development merely indicate that it is possible for Blarina to produce more than one litter, but his data do not prove that there is a true post-partum oestrus. He concludes that true post- partum mating and pregnancy during lactation rarely, if ever, occur in Blarina; remating depends upon the loss of a litter. Several investigators have demonstrated a post-partum oestrus in captive cricetid rodents. Bailey (1924) observed that captive meadow voles, Microtus p. pennsylvanicus, mated immediately following parturition, one female producing seventeen litters in a year. The gestation of this species is twenty-one days; lactation does not lengthen the period. Svihla (1932) reports several species of captive Pcromyscus mating shortly after parturition. Practically all females of the European wild rabbit (Oryctolagus cuniculus) under natural conditions become pregnant again at each post-partum oestrus during the height of the breeding season (Brambell, 1943). Elsewhere (Hamilton, 1940), I have shown that the cottontail, Sylvilagus floridanus mearnsii, presumably has a post-partus oestrus, for actively nursing females have contained embryos, suggesting that mating occurs shortly following parturition. Vorhies and Taylor (1940), in their study of the white-throated wood rat, Neotoma albigula, suggest that it is highly prob- able that this species produces successive litters with only very short intervals. In one den they found a female with two newly bom young hanging to the teats, with three half-grown young in the same den. This presumptive evidence of rapid succession in litters is nevertheless suggestive. Obviously with the species discussed above there is a post-partum oestrus at which time the female is receptive to the male. It appears unlikely that the im- position of captivity would modify the reproductive cycle in a short time. The absolute proof of such should be studied under natural conditions, though posi- tive data are rather difficult to secure. A feral Norway rat will cease lactating within forty-eight hours after the loss of a litter. Gentle traction on the teats fails to produce milk following this period. Meadow voles usually cease to lactate a day after the young are removed. The short-tailed shrew, Blarina hrevicavda, will not produce a flow of milk in a similar length of time if the nursing young be destroyed. Since such is known of these three species, we may adduce the probability of pregnant females nursing a litter if a flow of milk is possible. The mammary glands and teats swell noticeably in many mammals shortly before parturition. Interpretation of data may be faulty 210 Aug., 1949 HAMILTON— REPRODUCTIVE RATES 259 if such evidence is not considered with care. A colustral flow is evident in sev- eral cricetid rodents and some insectivores shortly before parturition. Inexperi- enced observers may mistake this secretion for milk and draw faulty conclusions. The following data lend support to the assumption that some small mammals, under natural conditions, successfully mate shortly after parturition and produce successive litters in rapid succession during part of the breeding season. Unless otherwise stated, the observations below relate to those of small mammals under conditions in the vicinity of Ithaca, New York. Blarlna brevicauda. Short-tailed shrew. — On July 23, 1947, E. W. Jameson, Jr. collected a nursing shrew. Under gentle pressure, milk could be drawn from the posterior teats. This shrew had four embryos, the uterine swellings measured 10 mm. in greatest width, sug- gestive of at least half-time pregnancy. A week later I trapped a nursing Blarina with five 11-mm. embrj'os. The teats, on traction, produced milk. A Blarina with six 9-nim. em- bryos was taken on August 10, 1948. The teats were drawn out and produced milk under gentle traction. In my field notes from 1926 to 1941 I have recorded data on many hundreds of these shrews, and noted on numerous occasions the occurrence of lactating gravid females. I have always considered this condition a perfectly natural one in shrews. Sorex f. fumeus. Smoky shrew. — Mating may follow parturition in this species (Hamil- tion, 1940). Actively nursing females have been taken which contained well-developed em- bryos. Specimens of pregnant Sorex cinereus have likewise been collected with prominent mammary glands containing an abundance of milk. Peromyscus leucopus noveboracensls. White-footed mouse. — Many hundreds of adult females have been examined during the breeding season over a twenty-year period. Many of these were actively nursing and contained embryos of various sizes. Pregnant nursing individuals appear more frequently in May and June collections. It is possible that more lactating individuals were gravid than my notes indicate, since early pregnancy cannot be determined by macroscopic examination. Oryzomys palustrls. Rice rat. — This is a prolific species. Field data obtained in Vir- ginia demonstrates that actively nursing females are occasionally pregnant ; there is evidence that a high fertility, resulting in vigorous females actually producing nine litters in a breed- ing season, is quite possible. The females are capable of breeding when seven weeks old. Clethrionomys gapperi. Red-backed vole. — On October 10, 1941, 1 took a 35-gram nursing female with four 3-mm. embryos and recent placental scars. The mammary glands, when dissected out, weighed 2.5 grams. On May 29, 1940, a nursing individual contained five 6-mm. embryos. Such data are admittedly fragmentary, but do suggest the rapidity of breeding in this species. Apparently many parous animals of the bank vole, Clethrionomys glareolus, are pregnant and lactating simultaneously. In such, pregnancies are prolonged by lactation causing a delay in implantation. This delay results in the blastocysts remaining in a resting state in the uterine lumen for a considerable period (Brambell and Rowlands, 1936). Microtus pennsylvanlcus. Meadow vole. — The reproductive behavior of this species in the wild is similar to that of captives. Adult females give birth to one litter after another in rapid succession. Data on this high fecundity are readily obtainable by live trapping. Females approaching full term are readily recognized ; repeated captures of specific individ- uals indicate that in some instances young are produced at three-week intervals over a period of several successive months. The capture of numerous wild pregnant females with nest- young is indubitable proof that post-parous mating occurs normally from early spring to fall. I have obtained many records of such. Ondatra zibethica. Muskrat. — Arthur C. Cook of the New York Conservation Depart- ment informs me (personal letter) that on June 20, 1941, he dug out a muskrat den at How- lands' Island, Cayuga County, New York. From the den he recovered a pregnant muskrat 211 260 JOURNAL OF MAMMALOGY Vol. SO, No. 3 with two litters of different size. In captivity, this gravid female suckled the smaller litter. This is presumptive evidence that the muskrat may mate following partus. Some polyoestrous rodents, in which two Htters a year are the rule, have the Htters widely spaced. Mating normally occurs in the late winter and midsum- mer. No post-partum oestrus occurs. Deanesly and Parkes (1933) indicate that there is no oestrus immediately after parturition or during lactation in the gray squirrel, Sciurus caroUnensis. A similar condition obtains with Tamiasdu- rus, Tamias, and Eutamias. Where populations of small mammals are sufficiently large, it appears prob- able that fruitful matings often occur following partus, at least during the height of the breeding season. If such a condition be generally true, it enables one to compute, in a measure, the probable reproductive rate. One may thus visualize the annual natural increment in a population during the breeding season. Such data are most useful to students of populations and animal behavior. LITERATURE CITED Bailey, Vernon. 1924. Breeding, feeding, and other life habits of meadow mice (Micro- tus). Journ. Agr. Research, 27: 523-536. Bbambell, F. W. Rogers. 1943. The reproduction of the wild rabbit Oryctolagua cuni- culu8 (L.). Proc. Zool. Soc. London, 114: 1-45. Brambell, F. W. R., and I. W. Rowlands. 1936. Reproduction in the bank vole (Evo- tomys glareolus Schreber). I. The oestrous cycle of the female. Phil. Trans. Royal Soc. London, Series B., No. 531, 226: 71-97. Deanesly, Ruth, and A. S. Parkes. 1933. 3. The reproductive process of certain mammals. Part 4. The oestrous cycle of the grey squirrel {Sciurus caroUnensis) Phil. Trans. Royal Soc. London, B. 222: 47-78. Hamilton, W. J., Jr. 1940. The biology of the smoky shrew {Sorex Jumeus fumeuiMiWQr) . Zoologica, 25(4) : 473-492. 1940. Breeding habits of the cottontail rabbit in New York State. Jour. Mamm., 21: 8-11. Hamlett, G. W. D. 1935. Delayed implantation and discontinuous development in the mammals. Quart. Rev. Biol., 10: 432-447. Pearson, Oliver P. 1944. Reproduction in the shrew (Blarina brevicauda Say). Amer. Jour. Anat., 75: 39-93. SviHLA, Arthur. 1932. A comparative life history study of the mice of the genus Pero- myscus. Misc. Publ. Mus. Zool. Univ. Michigan, 24: 1-39. VoRHiES, Charles T., and Walter P. Taylor. 1940. Life history and ecology of the white-throated wood rat, Neotoma albigula albigula Hartley, in relation to grazing in Arizona. Univ. Ariz. Agr. Exp. Sta. Tech. Bull., 86: 456-629. Cornell University, Ithaca, New York. Received October 4, 1948. 212 AN EXPLANATION OF GEOGRAPHIC VARIATION IN LITTER SIZE Albekt W. SpexNCeh and Harold W. Steinhoff Absthact. — Our explanation of the latitudinal and altitudinal variation in litter sizes of small nianinials invokes the effect of length of season and parental mor- tality related to reproduction. It may be assumed that a portion of the maternal mortality rate varies directly as the size of litter produced. Short seasons limit the maximum number of times a female can reproduce in her lifetime and give an advantage to phenof\pes producing large litters. Long seasons favor producers of small litters. The contribution to the total rate of increase of the litters produced in the additional time afforded by long seasons is greater for producers of small litters because a larger proportion of parents of small litters survive to produce throughout the periods. The increment provided is sufficient to overcome the initial advantage of parents producing large numbers of young in their first litters. Variation in mean litter size related to latitude and altitude has been re- ported for several species of small mammals. Lord (1960) analyzed data on several species and concluded there was a regression of litter size on latitude in small nonhibernating prey species. Dunmire ( 1960 ) added the dimension of altitude when he reported an increase with elevation of mean litter sizes of deer mice from the White Mountains in California. Mean litter sizes in all cases were larger at northern locations and high elevations than at more southern or lower localities. Several explanations have been proposed to account for the observed varia- tion. Lord ( 1960 ) reviewed and rejected several hypotheses before proposing his own theory that higher mortality rates during northern winters required an increase in reproductive rate. We cannot accept this explanation because it contradicts established principles of population dynamics. The problem is part of the larger question of regulation of fecundity in all animals. Lack ( 1948, 1954 ) has explained the regulation of clutch size and litter size in terms of natural selection. The modal number in his view repre- sents the most consistently successful clutch or litter. Williams ( 1967 ) has presented a mathematical refinement of Lack's Principle. According to his concept, an allocation of parental energy between present and residual re- productive values that will maximize the total reproductive value is selectively advantageous. The cost of rearing a large or better nourished brood now is reduction of future reproduction. The litter size prevalent in a particular population represents the best investment possible for the particular situation of the population. Although Williams did not specifically consider intraspecific latitudinal variation in litter size, our own explanation, developed independently, sub- stantially parallels his general treatment. We believe the shorter seasons of more northern latitudes or higher altitudes limit the number of times an animal resident in those areas is able to reproduce in its lifetime compared to its relatives in lower or more southern regions. It therefore becomes ad- 281 213 282 JOURNAL OF MAMMALOGY Vol. 49, No. 2 vantageous for an animal to invest its energies in a few, large, early litters even though doing so reduces its life expectancy and total reproductive con- tribution below the maximum achievable by many small litters. This is so because short seasons make it impossible for the animal to realize the returns from the conservative approach within its life-span. The most productive strategy is the production of large litters. This idea is developed and illus- trated below. Results and Discussion Our interest in the problem began with the observations recorded in Table 1. Feromyscus maniculatus had been collected at several sites in Colorado along a transect extending from the plains of central Weld County to timber- line in western Larimer County. The increase in mean potential litter size parallels the experience of Dunmire ( 1960 ) in the White Mountains of Cali- fornia. The regression of mean litter size on elevation is of the same order of magnitude as that on latitude as determined by Lord (1960). The data thus illustrate the general nature of the variation. The major factors influencing the potential rate of increase of a species are the number of female young per female per parturition and the length of the period from birth to first reproduction. However in realistically appraising the number of young per female per litter it is necessary to consider the actual contribution to the next generation represented by that litter (Lack, 1948, 1954). If survival to maturity were a linear function of litter size and decreased 10% for each additional young, then only 2.5 (1.25 females) in a litter of five and only 2.4 ( 1.2 females ) in a litter of six would effectively be contributed on the average to the next generation. If these were the only factors influencing the rate of increase and time to maturation were uniform in the species, then a phenotype producing the greatest effective number (5) would have the highest potential rate of increase and thus the selective ad- vantage. Reproductive longevity and subsequent reproduction, however, also play a role in determining the rate of increase. Each successive litter contributes an increment to the total reproductive value. The value of each increment declines the later its production occurs in life, but the collective importance of the increments in influencing the rate of increase grows as the margin of difference between the effective numbers of different litter sizes diminish and as the length of the maturational period lengthens (Fisher, 1930; Birch, 1948). This portion of the reproductive value is the major element involved in the phenomenon of latitudinal and altitudinal variations in litter size. The magnitude of the contribution to the rate of increase by successive litters depends in part on the survival of the parents (Birch, 1948). Bearing and rearing young constitutes a risk for the parent. The presence of the young reduces concealment of parents from, and adds to their vulnerability to, predators. Increased foraging activity also exposes the parents to greater dangers of predation. Physiological stresses such as nutritional deficiencies 214 May 1968 SPENCER AND STEINHOFF"— LITTER SIZE 283 Table 1. — Fieciticncy distribution of embryo counts in Pcroinysciis nianiculatus from different elevations. Number Mean Mean of litter bod\' Site females 1 2 3 4 •■D 6 7 8 9 size length Plains (5100-5300 ft) Coal Creek 10 Pierce 17 Cobb Lake 29 Subtotal 56 5 3 1 2 9 4 2 17 9 1 9 29 14 1 4.0 95.5 -0.7 Foothills ( 5500-6500 ft ) Rist Canyon 22 Siiltzcr Culch 15 Subtotal 37 2 2 S 10 6 2 18 4.1 95.9 -0,9 Mountains (8000-11,000 ft; Buckhorn R. S. 7 1 3 1 I 1 Pennock Creek 4 1 3 Cirques 3 1 1 1 5.0 94.2 Subtotal 14 2 6 1 2 1 2 -2.2 Pingree Park 33 1 1 6 6 11 5 2 1 5.6 Mountain subtotal 47 1 O 12 7 13 6 4 1 5.4 and post partum diificulties impair the survival of the parents and damage their capabilities for reproduction in the future. It is reasonable to assume that the effects are proportional to the number of young produced ( Lack, 1954, 1948). Such risk is incurred at each successive reproductive event. The greater the risk and the more times repeated, the more the survival rate is lowered and reproduction in later life is curtailed. Less drastic effects than actual death of the parents are probable and perhaps even more important. However the end result is the same and we feel justified in regarding these lesser effects as a form of mortalitv. The natural longevity (maximum physical reproductive longevity) of the organism and extrinsic factors such as climate impose an upper limit on the maximum number of opportunities for reproduction (hereafter designated MOR). For example if the season permitted four litters annually and the maximum reproductive longevity of the species were 2 years the MOR would be eight. Only a fraction of any age class would attain the maximum limit MOR. The actual proportion of the population that achieved the maximum would depend upon the mortality rates discussed above, and would be in- versely proportional to litter size. It follows, then, that any reduction in the MOR would have a relatively greater effect on the rate of increase of pheno- types producing smaller number of young per litter and having higher sur- vival rate than on phenotypes producing large litters. 215 284 JOURNAL iW MAMMALOGY Vol. 49, No. 2 This is the Hne of reasoning that led to our conception of the connection between length of season and variation in litter size. Thus far we have ap- proached the problem strictly from the theoretical aspect by constructing models of the growth of idealized populations. We then studied the conse- quences for the potential intrinsic rate of increase of reducing the MOR and changing the length of the period of maturation. For simplicity, the relation- ships of survival of the young to maturity and survival of parents was as- sumed to be an inverse linear function of litter size. Consideration of avail- able data on guinea pigs (Lack, 1948), Peromysctis (McCabe and Blanchard, 1950) and birds and lizards (Lack, 1954) indicated that these may be accept- able approximations. All effects of fecundity on future parental reproduction were combined under the heading of parental mortality. All mortality not related to litter size was ignored. Litter size was assumed to be a constant characteristic throughout the population and independent of age. Only female births were considered. The unit of time was taken to be the minimum in- terval between successive parturitions and each mature female was assumed to reproduce at each interval. Net reproductive ratios and rates of increase were evaluated by a combination of numerical integration and graphic methods. Lotka (1925), Fisher (1930), and particularly Birch (1948) were the main sources of inspiration and methods used in the computations. Fig. 1 summarizes the findings of our study. Observe that the size of the litter with the greatest potential rate of increase declines as the MOR in- creases. The shift reflects the relatively greater increase in the total repro- ductive value of producers of small litters as the number of late reproductions is allowed to increase. Note also that the shift is rather abrupt and little affected by further increase in MOR. The range of the shift increases as the period of maturation lengthens. Changes in the parameters of survival of parents and young also affect the locus and magnitude of the shift but these effects have not yet been thoroughly explored. Approximations of the varia- tion shown in Table 1 were achieved by substitution into the calculations for curve C (Fig. 1) of the following coefficients. The coefficient of regression for survival of young to maturity on litter size used was -.10 and of survival of parents on litter size was -.016. It of course would be possible to obtain the same approximation by an almost limitless number of combinations, but the important point is that the magnitude of the coefficients required is of plausible order. The analysis offers an explanation of the observed distribution of latitudinal variation in litter size among mammals considered by Lord ( 1960 ) . The fossorial and hibernating species investigated (Spermophilus and Thomomys) usually have one litter annually. Their period of development from birth to first reproduction is about 1 year or 6 months when two litters are produced (Asdell, 1964). The effect of shortening the annual season would thus have little effect on the reproductive opportunities of these organisms. Their pat- terns of reproduction are approximated by that shown in curve A in Fig. 1. 216 May 1968 SPENCER AND STEINHOFF— LITTER SIZE 285 6- B 3 z 4 Maturity at one reproductive interval Maturity at two reproductive intervals Maturity at ttiree reproductive intervals P ►€ 2 4 8 16 Maiimum number of opportunities for reproduction Fig. 1. — Change in the fitness of Htter sizes as the maximum number of opportunities fur reproduction is increased. The Hnes connect the htter sizes liaving the greatest potential rate of increase under the hmitations imposed. Only drastic differences in length of season would produce noticeable effects. Even then the magnitude of variation would be relatively small. On the other hand, the species of mice and shrews that display latitudinal variation can produce several (up to 12) litters annually and yet have a period of develop- ment from birth to first reproduction equal to several reproductive intervals. Many shrews begin reproducing at the age of 1 year. The reproductive pat- terns of these animals correspond to curve C. Effects of significant magnitude could be produced with relatively small changes in the length of the season. In the Colorado data, for example, the growing seasons on the plains (5000 ft) and in subalpine areas (8000-10,000 ft) differ by almost a factor of two. An interesting aspect of the analysis is that the relative superiority of the favored phenotype is reduced directly as the MOR. Therefore, the genetic variance in the population would be correspondingly reduced. Attainment of maximum fitness in the population should be a long process; in two popula- tions, the one occupying the environment with the shorter season should have the greater variability in litter size after a given period. This would be 217 286 JOURNAL, OF MAMMALOGY Vol. 49, No. 2 particularly true if gene flow between them were occurring. This condition is seen in Table 1. Intraspecific variation may originate in many ways. There is, however, only one agency, natural selection, regularly producing the directed sort of variation we are considering. Lord (1960) has suggested that the increased size of litters may be a compensating response to mortality among the small mammals during winter. However, if a population of a species were able to increase its reproductive potential in this manner, the ability would be just as adaptive in any environment. Variation would quickly disappear. If the variation is to be described in terms of fitness then the particular phenotypes characteristic of each locality must have the greatest fitness in that situation. Our explanation offers an hypothesis explaining how the differential in fitness may originate. Literature Cited AsDELL, S. A. 1964. Patterns of mammalian reproduction. Cornell Univ. Pre.ss, Ithaca, 2ncl ed., viii + 670 pp. Bn^CH, L. C. 1948. The intrinsic rate of increase of an insect population. J. Anim. Ecol., 17: 15-26. DuxMiRE, W. W. 1960. An altitudinal survey of reproduction in Pcronnj.scus niaiiictt- latus. Ecology, 41: 174-182. Fisher, R. A. 1930. The genetical theory of natural selection. Clarendon Press, O.xford, 272 pp. Lack, D. 1948. The significance of litter-size. J. Anim. Ecol., 17: 45-50. . 1954. The natural regulation of animal numbers. Clarendon Press, Oxford, viii + .34.3 pp. Lord, R. D., Jr. 1960. Litter size and latitude in North American mannnals. Amer. Midland Nat., 64: 488^99. LoTKA, A. J. 1925. Elements of physical biology. Williams and Wilkins, Baltimore. McCabe, T. T., and B. D. Blaxchard. 1950. Three species of Peromyscus. Rood Associates, Santa Barbara, California, v -|- 136 pp. Williams, G. C. 1967. Natural selection, the costs of reproduction, and a refinement of Lack's principle. Amer. Nat., 100: 687-690. Division of Biological Science, Fort Lewis College, Durango, Colorado 81301, and Departmetit of Fishery and Wildlife Biology, Colorado State University, Fort Collins, 80521. Accepted 2 January 1968. 218 Sonderdruck aus Z. f. Saugetierkunde Bd. 30 (1965), H. 1, S. 10—20 A!le Rechte, audi die der Cbersetzung, des Nadidrudts und der photomedianischen Wiedergabe, vorbehalten. VERLAG PAUL PAREY ■ HAMBURG 1 • SPITALERSTRASSE 12 The effects of suckling on normal and delayed cycles of reproduction in the Red Kangaroo By G. B. Sharman Eingang des Ms. 23. 12. 1963 Introduction In non-lactating female marsupials the occurrence of fertilization, followed by imme- diate gestation of the embryo, does not delay the onset of the following oestrus. In those marsupials in which the gestation period is considerably shorter than the length of one oestrous cycle, such as Didelphis virginiana (Hartman, 1923) and Trichosurus vulpecula (Pilton and Sharman, 1962), oestrus recurs at the expected time if the young are removed at birth. In several species of Macropodidae, sudi as Setonix brachyurus (Sharman, 1955), Potorous tridactylus (Hughes, 1962) and the Red Kangaroo (Shar- man and Calaby, 1964), the gestation period occupies almost the length of one oestrous cycle and oestrus is imminent at the time of parturition. Oestrus thus recurs just after the young reach the pouch (post-partum oestrus) presumably because pro-oestrus chan- ges are initiated before the onset of the suckling stimulus. In all marsupials suckling of young in the pouch is accompanied by a lengthy period during which oestrus does not occur. This period is called the quiescent phase of lactation or, simply, the quies- cent phase. It differs from seasonal anoestrus in that the ovaries and other reproduc- tive organs respond to the removal of the suckling stimulus by resuming cyclic func- tions. Those marsupials in which post-partum oestrus occurs exhibit discontinuous embryonic development analogous to the delayed implantation which occurs in some eutherian mammals. If fertilization takes place at post-partum oestrus the resulting embryo assumes a dormant phase, at the blastocyst stage, and is retained as a dormant blastocyst during the quiescent phase. In these marsupials pregnancy (the interval between copulation at post-partum oestrus and parturition) is long and gestation of the embryo is interrupted by the dormant phase. In the Red Kangaroo, Megaleia rufa (Desm.), the oestrous cycle averages 34 to 35 days and the gestation period is 33 days in length (Sharman and Calaby, 1964). Post- partum oestrus occurs, usually less than 2 days afler the newborn young reaches the pouch, and a dormant blastocyst is found in the uterus of females, fertilized at post- partum oestrus, which are suckling young less than 200 days old in the pouch (Shar- man, 1963). If the young is removed from the pouch suckling ceases and the dormant blastocyst resumes development: the young derived from it being born about 32 days after removal of the pouch young (RPY). This birth is followed by another post- partum oestrus or, if the female was not carrying a blastocyst, by a normal oestrus. Oestrus recurs at the same number of days after RPY irrespective of whether a de- layed blastocyst was carried or not. The sequence of events from RPY to the next oestrus is called the delayed cycle of reproduction^ to distinguish it from the normal reproductive cycle which follows oestrus. The delayed reproductive cycle may be divided into delayed gestation and delayed oestrus cycle according to whether a dor- 1 The term "delayed cycle of reproduction" or "delayed (reproductive) cycle", was introduced by Tyndale-Biscoe (1963) to describe the resumption of ovarian activity, and the features associated with it, following removal of pouch young (RPY). 219 Normal and delayed cycles of reproduction in the Red Kangaroo 11 niant blastocyst does or does not complete development. If the young is retained in the pouch until it leaves in the normal course of events the delayed reproductive cycle occurs coincident with the latter stages of pouch life. The dormant phase of the blastocyst gives way to renewed development when the pouch young is a little over 200 days old and subsequent vacation of the pouch, at an average age of 235 days, is immediately followed by birth of another young (Sharman and Calaby, 1964). The young is suckled for another 130 days, that is until it is about a year old, after it leaves the pouch. During this period the normal reproductive cycle occurs if the pouch is not occupied. It is thus evident that, although the delayed reproductive cycle occurs after RPY and cessation af lactation, some factor other than the actual pro- duction of milk must be implicated for both delayed and normal cycles may also occur during lactation. The aim of the experiments reported below was to determine the effect of the suckling stimulus on both normal and delayed reproductive cycles. Additional suck- ling stimulus was provided by fostering an extra young on to females already suck- ling a young-at-foot. The experimental approach was suggested by chance obser- vations on a female Red Kangaroo which, while suckling her own young-at-foot, alternately fed the young of another female kept in the same enclosure. There are four teats in the pouch but the teat to which the young attaches after birth alone produces milk and its underlying mammary gland produces all the milk for the young from birth to weaning. The female's own young and the foster-young thus shared the products of a single mammary gland and used the same teat alternately. Some initial results, in so far as they were relevant to the theme of delayed implanta- tion, were reported earlier in a review of that subject (Sharman, 1963). Methods The results presented consist of observations on a minimum of five reproductive cyc- les in the female Red Kangaroo in each of the following categories: 1. Normal cycle of reproduction, suckling one young-at-foot. 2. Normal cycle of reproduction, suckling two young-at-foot. 3. Delayed cycle of reproduction, suckling one young-at-foot. 4. Delayed cycle of reproduction, suckling two young-at-foot. The results are compared with data on the normal and delayed cycles of reproduc- tion in non-lactating females most of which have been published elsewhere (Sharman, 1963; Sharman and Calaby, 1964; Sharman and Pilton, 1964). In most cases the experimental females were pregnant or carrying dormant blastocysts so that cycles of normal or delayed gestation with subsequent post-partum oestrus were studied. The gestation periods and cycles were regarded as having been significantly lengthened when they occupied a time greater by the length of two, or more, standard deviations than similar cycles in control, non-lactating, females Some difficulty was experienced in getting females to accept foster-young and only six females readily did so. The experiments were therefore done serially one female being used in two and two females in three experiments. The animals were watched from a hide overlooking the enclosures and observed with binoculars. An initial watch was always done to find whether females accepted their potential foster-young. Thereafter prolonged watches were kept on some females to determine the amount of time spent suckling the young-at-foot. Vaginal smears for the detection of oestrus and copulation were taken as reported previously (Sharman and Calaby, 1964). 220 12 G. B. Sharman Results Effects of suckling on the normal cycle of reproduction In thirteen non-lactating female Red Kangaroos forty-two intervals from oestrus to the succeeding oestrus averaged 34.64 days with a standard deviation of 2.22 days (34.64 ± 2.22 days). Twenty gestation periods in fourteen females lasted 33.00 ± 0.32 days (Fig. lA). In five females, each observed for a single reproductive cycle while suckling one young-at-foot, the intervals between two successive oestrous periods were not different from those in non-lactating females (Fig. IB). In another female (K32a) A 42 cycles K60 K53a Kic K30c K12c K32q K3lQ K31b K36c K30a Kdd K4e K35c 12 16 20 24 28 32 36 40 44 48 Intervals between successive oestrous periods Fig. 1. Intervals between successive oestrous periods in nonlactating (control) female Red Kangaroos (A), females sudiling one young-at-foot (B) and females suckling two young-at- foot (C). Black lines — continuous embryonic development, broken lines — approximate periods of dormant phase in embryo induced and maintained by suckling young-at-foot, open lines — no embryos present, bars inserted in A — standard deviations either side of mean. the gestation period was not significantly different from that of control females but oestrus did not occur until 5 days post-partum. This was the longest interval between parturition and post-partum oestrus recorded but it is not regarded as significant. Two cycles in female K31, one lasting 41 days and one 47 days, were abnormally long. The 41-day cycle is of special significance since the interval between copulation and birth was 40 days. This differs so much from the gestation period in the control, non-lacta- ting, females that it must be assumed that suckling of the single young-at-foot induced a short quiescent phase in the uterus accompanied by a dormant phase of about 7 days in the embryo. The 47-day cycle was over 12 days longer than the mean normal cycle length and 7 days longer than the maximum cycle length. The female copulated at oestrus but did not give birth so it is presumed that fertilization did not occur. In three females already suckling one young-at-foot, which had another young-at- foot fostered on to them at about the time of fertilization, the lengths of the repro- ductive cycles were not significantly different from those in control females. Two females had significantly longer cycles than in control females. One of these (K36) was used in three successive experiments while suckling the same two young-at-foot. In the first of these (K36a) the extra suckling stimulus had no significant effect on the length of the reproductive cycle. The second experiment concerned the delayed reproductive cycle and is reported below. During the third experiment (K36c) the young were being weaned but a highly significant result was obtained. The interval from copulation to 221 Normal and delayed cycles of reproduction in the Red Kangaroo 13 birth showed conclusively that a dormant phase had been induced and maintained in the embryo for about 14 days of the 47-day pregnancy. In the other female in which the cycle was prolonged (K4e) the embryo presumably had a dormant phase of about 6 days. Effects of suckling on the delayed cycle of reproduction In ten non-lactating females thirteen intervals from RPY to the succeeding oestrus were 34.46 ± 1.92 days. In seven of these females the delayed gestation period was 31.64 ± 0.65 days (Fig. 2A). There was no evidence that suckling one young-at-foot had any effect on the length of the delayed reproductive cycle (Fig. 2B). In one female (K12a) the interval from RPY to the following oestrus was 38 days but this falls short of the minimum interval accepted as significantly different. All six females suckling two young-at-foot (Fig. 2C) were carrying a dormant blastocyst in the uterus when the pouch young were removed. In five of these the Interval RPY to birth was significantly longer than in control females (Fig. 2C). The interval RPY to the next oestrus was longer than the mean for control non-lactating females in all six experimental females and in three of them (K4b, K30b, K36b) the difference from controls was highly significant. It must be concluded that the blasto- cysts of five of the above experimental females remained in the dormant phase for between 3 and 22 days longer afler RPY than did those of control non-lactating females and females suckling one young-at-foot. A 13 cycles B Kio K15 K38b K12b K12a K38a K63b K32b K36b K30b K4b 12 16 20 24 28 32 36 40 44 48 52 56 Intervals between RPY and oestrus Fig. 2. Intervals between removal of pouch young (RPY) and the next oestrus in non-lactating (control) female Red Kangaroos (A), females suckling one young-at-foot (B) and females suckling two young-at-foot (C). Black lines — continuous embryonic development, broken lines — approximate periods of continued dormant phase of embryo maintained by suckling young-at-foot, bars inserted in A — standard deviations either side of mean. The amount suckling in relation to occurrence of parturition and return to oestrus Observations on the habits of the pouch young suggested that the stimulus causing withholding of the mother's reproductive cycles might be tactile and received via the teat. The young during the early stages of pouch life, when reproductive cycles were withheld, were suckled continuously and could not regain the teat if removed before the age of 6 weeks. Later young were able to take the teat back into their mouths but were seldom found free of the teat before the age of about 5 months. On the other 222 14 G. B. Sharman o o « OX) c n u 3 Ui 4-1 t/) u O "T3 C CS C o c u 3 O" 3 C o o o 3 O >^ o ■w n rs u C o C ? o 0-x, — c -fi.2 c o z c s o es c 3 O o -r) -t3 C) 11 i-» 4-> t> 4) _C 4-* *-» _G ^ rt rt ^ C -T3 ^ 3 o K ■^ c *-> S g « u O -^ y ri a, 00 D. C X a V c 1) Ji 4-> ri rt rt _i -a -a rt die ein Junges bei Fufi saugten, ergab sich keinEinfluft des Saugens. Im ganzen 5 Cyclen. Bei 99> bei denen die Ruhezeit der Blastocyste resp. 3, 3, 6, 11 und 22 Tage betrug. In letzteren beiden setzte die Weiterentwicklung nicht ein, bevor nicht eines der Jungen weggenommen wurde. Im ganzen 6 Cyclen. Die Beobachtungen zeigten, dafi 9? ^^^ 2 Jungen bei Fufi ihre Jungen doppelt so lange saugen, wie sie ein einziges gesaugt haben wiirden. Daraus wurde geschlossen, dafi der Sauge- Stimulus, von einem oder zwei Jungen bei Fui5 ausgelost, sowohl die Ruhephase wahrend der Laktation, als auch die damit gleichlaufende Ruhephase des Embryos einleitet und erhalt. Die bisherige Erfahrung laEt annehmen, dafi der Stimulus, der den Beginn der Ruhephase bewirkt, tactil ist und iiber die Zitze empfangen wird, und dafi der verzogerte Cyclus der Fortpflanzung auftritt, oder der unterbrochene normale Cyclus wieder aufgenommen wird, wenn der Sauge- reiz sich vermindert. Einige von anderer Seite veroffentlichte Daten iiber Trachtigkeitsdauern von Beuteltieren enthalten offenbar Fehler, da die betreffenden Autoren die Bedeutung gleichlaufenden Saugens nicht beachteten. Mitgeteilte Falle, daf^ 9 Beuteltiere auch nach langer Isolierung vom S war- fen, kann ohne weiteres durch das Auftreten des verzogerten Fortpflanzungs-Cyclus erklart werden. Literature Carson, R. D. (1912): Retarded development in a red kangaroo; Proc. zool. Soc. Lond. 1912, 234-235. — Hartman, C. G. (1923): The oestrous cycle in the opossum; Am. J. Anat. 32, 353-421. — FiEDiGER, H. (1958): Verhalten der Beuteltiere (Marsupialia); Handbuch Zool. 8, 18 Lief 10(9), 1-28. — FiiLL, J. P., and O'Donoghue, C. H. (1913): The reproductive cycle in the marsupial Dasyurus viverrinus. Quart. J. micr. Sci. 59, 133-174. — Hughes, R. L. (1962): Reproduction in the macropod marsupial Potorous tridactylus (Kerr); Aust. J. Zool. 10, 193-224. — Morris, D., and Jarvis, C. (Eds.) (1959): The International Zoo Year Book, Vol. 1; London. — Owen, R. (1839-47): Marsupialia; In: The Cyclopaedia of Anatomy and Physiology, Vol. 3 (ed. R. B. Todd), London. — Pilton, P. E., and Sharman, G. B. (1962): Reproduction in the marsupial Trichosurus vulpecula. J. Endocrin. 25, 119-136. — Sharman, G. B. (1954): Reproduction in marsupials; Nature, Lond. 173, 302-303. — Sharman, G. B. (1955): Studies on marsupial reproduction. 3. Normal and delayed pregnancy in Setonix brachyurus; Aust. J. Zool. 3, 56-70. — Sharman, G. B. (1962): The initiation and maintenance 228 20 G. B. Sharman of lactation in the marsupial Trichosurus vulpecula; J. Endocrin. 25, 375-385. — Sharman, G. B. (1963): Delayed implantation in marsupials; In: Delayed implantation (ed. A. C. En- ders), Chicago. — Sharman, G. B., and Calaby, J. H. (1964): Reproductive behaviour in the red kangaroo in captivity; C. S. I. R. O. Wildl. Res. 9 (in press). — Sharman, G. B., and PiLTON, P. E. (1964): The life history and reproduction of the red kangaroo {Megaleia rufa). Proc. zool. Soc. Lond. 142 (in press). — Tyndale-Biscoe, C. H. (1963): The rule of the corpus luteum in the delayed implantation of marsupials; In: Delayed implantation (ed. A. C. Enders), Chicago. Author's address: Dr. G. B. Sharman, C. S. I. R. O. Division of Wildlife Research, Canberra, A. C. T., Australia 229 REPRODUCTION AND GROWTH IN AAAINE FISHERS' PHILIP L. WRIGHT, Montana Cooperafive Wildlife Research Unit and Departnnent of Zoology, University of Montana, Missoula MALCOLM W. COULTER, Maine Cooperative Wildlife Research Unit, University of Maine, Orono Abstract: New data concerning reproduction, aging techniques, and growth of fishers (Martes pennanti) were obtained from 204 specimens taken from October to April during 1950-64. All female fishers more than 1 year old were pregnant. The immature class consisted of juveniles in their first year. The period of delayed implantation lasted from early spring until mid- or late winter. Nine adult females taken in January, February, or March showed implanted embryos. Fishers in active pregnancy had corpora lutea 7 times the volume of those in the period of delay. Most litters are bom in March, but some as early as late February and some in early April. Counts of corpora lutea of 54 animals taken during the period of delay and during active pregnancy averaged 3.35 per female. The number of embryos, either un- implanted or implanted, corresponded exactly with the number of corpora lutea in 18 of 21 animals. Two recently impregnated 1-year-old females, recognizable from cranial characters, had tubal morulae, confirming that females breed for the first time when 1 year old. Also confirmed are previous findings of Eadie and Hamilton that juvenile females can be distinguished from adults by open sutures in the skull throughout their first year. Juvenile males can be recognized in early fall by open sutures in the skull, absence of sagittal crest, immature appearance and lighter weight of bacula, unfused epiphyses in the long bones, and small body size. The sagittal crest begins to develop in December and often is well developed by March. The baculum grows slowly during the early winter, but by February there was some overlap with weights of adult bacula. Male fishers showed active spermatogenesis at 1 year. Open sutures were found in juvenile male skulls throughout the first year. Pelvic girdles of juveniles were distinguished by an open pubo-ischiac symphysis; adults of both sexes showed the two innominates fused into a single bone resulting from at least a partial obliteration of the symphysis. Mean body weights of animals weighed whole in the laboratory were as follows: adult males, 10 lb 12 oz; juvenile males, 8 lb TV2 oz; adult females 5 lb 8 oz; juvenile females, 4 lb 11 oz. After reaching an all-time low during the early part of the century, the fisher has made a remarkable recovery during the past 25 or 30 years in Maine (Coulter 1960) and in New York State (Hamilton and Cook 1955). The increase in abundance of this high quality furbearer in New York to the point that it could be legally trapped al- lowed Hamilton and Cook ( 1955 ) and later Eadie and Hamilton ( 1958 ) to discover sig- nificant facts from studying carcasses ob- tained from trappers. ^ This study is a contribution from the Maine and the Montana Cooperative Wildlife Research Units, the University of Maine, the Maine De- partment of Inland Fisheries and Game, the Uni- versity of Montana, the Montana Fish and Game Department, the U. S. Bureau of Sport Fisheries and Wildlife, and the Wildlife Management In- stitute cooperating. The study was supported by Grant GB-3780 from the National Science Foun- dation. 230 Maine Fishers • Wright and Coulter 71 In Maine, the season was reopened in 1950, pennitting collection of data and ma- terial from fishers trapped there. The pur- pose of the present paper is to present new information about reproduction, age de- termination, and growth of fishers, derived from study of Maine animals obtained be- tween 1950 and 1964. More than a dozen biologists and many wardens of the Maine Department of In- land Fisheries and Game collected material from trappers. Special thanks are due to Myron Smart, Biology Aide, who assisted in numerous ways throughout the entire study, and to Maynard Marsh, Chief Warden, who made arrangements for confiscated speci- mens to be processed at the Maine Unit. Numerous graduate assistants at the Maine Unit helped with processing carcasses and the preparation of skulls and bacula. We are indebted to Howard L. Mendall for edi- torial assistance and to Virginia Vincent and Alden Wright who made the statistical calculations. Margaret H. Wright did the microtechnique work. Elsie H. Froeschner made the drawings. Some of these findings were summarized in an unpublished Ph.D. dissertation presented by Coulter at the State University College of Forestry at Syracuse University. FINDINGS OF PREVIOUS WORKERS Hall (1942:147) pubHshed data from fur farmers in British Columbia showing that the gestation period in captive fishers ranges from 338 to 358 days and that copu- lation normally takes place about a week after the young are born. Enders and Pear- son (1943) described the blastocyst of the fisher from sectioned uteri of trapper-caught animals and showed that the long gestation period is due to delayed implantation. It was assumed that the blastocysts remain in- active from spring until sometime during winter. De Vos ( 1952 ) studied fishers in Ontario and made preliminary attempts to establish an aging method based upon skulls of males and females and the bacula of males. Hamilton and Cook ( 1955 ) pub- lished information about the current status of fishers in New York State and described a technique for recovering the unimplanted blastocysts from fresh reproductive tracts by flushing them out with a syringe. Eadie and Hamilton (1958) provided additional data on the numbers of blastocysts in preg- nant tracts and described cranial differ- ences between adult and immature females. MATERIALS AND METHODS Coulter collected material in Maine from trapped fishers, starting in 1950 when the season was first reopened. The intensity of the collection varied over the years depend- ing upon the legal regulations in effect. Data are available from 204 animals. In addition to animals legally taken dur- ing the trapping season, Coulter obtained a number of animals both before and after the season, taken by trappers who were trapping other species, primarily bears and bobcats. Trappers who caught fishers ac- cidentally were required to turn them over to the Department of Inland Fisheries and Game which in turn brought or sent them to the Maine Unit at Orono where they were autopsied by Coulter. Unskinned fish- ers as well as skinned carcasses were sub- mitted to the laboratory. Whenever possi- ble, weights were taken immediately before and after skinning to obtain an index for converting the weights of carcasses re- ceived from trappers to whole weights. During the trapping season carcasses were collected at trappers' homes. Usually the material was submitted in fresh condition; often it was frozen or thoroughly chilled when received at the laboratory. Because of the interest of the cooperators, most of the material was accompanied by collection 231 72 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 dates, method of capture, locality, and other notes. These data together with measure- ments, weights, observations about the con- dition and completeness of the specimens, and a record of material saved for future study were entered on individual cards for each animal. A special effort was made from the fall of 1955 to the spring of 1958 to obtain com- plete skeletons, and 59 such specimens were obtained. Coulter trapped a series of espe- cially needed animals in late March and early April, 1957. Because of excellent co- operation by State Game Wardens and Re- gional Biologists, a good sample of speci- mens was available for study over a 6- month period from October to April. This series of fishers is an unusually valu- able one for discerning important aspects of the growth and the reproductive cycle of this mustelid. For example, nine females in active pregnancy were obtained, as well as several adult males in full spermatogenesis. Furthermore, the juvenile fishers were grow- ing and maturing rapidly during the col- lecting period, and this fairly large collec- tion has allowed us to reach significant con- clusions concerning the onset of sexual ma- turity and the distinction between the age classes with more assurance than de Vos ( 1952 ) was able to do with more limited material. The reproductive tracts of female fishers were removed in the laboratory and pre- served in 10-percent formalin, in AFA, or in special cases, Bouin's fluid. The bacula of all the males were air-dried, as were the skulls of both sexes. Testes from a few rep- resentative males were fixed in formalin also. Coulter solicited the cooperation of Wright in 1955 and all of the material then available was shipped to him for further analysis and for histological work. Most of the skeletal material was cleaned by der- mestid beetles in Montana. This study was carried out without the aid of known-age specimens. Since the study was completed, three known-age ani- mals have become available: an 18-month- old female in Maine which was in captivity for 1 year, and two females captured in central British Columbia, released in west- ern Montana, and recaptured 6 years later. Study of these three animals in no way af- fects the findings presented in this paper. Evidence is presented to indicate that young-of-the-year animals can be distin- guished from adults by studying either their skulls and skeletons or their reproduc- tive tracts. Animals judged by these criteria to be less than 1 year old are, for conve- nience, referred to as juveniles even though in a few cases they may be almost 1 year old. Except for one criterion for distinguish- ing yearUng females from older adult fe- males, described by Eadie and Hamilton (1958:79-81) and confirmed here, no method of determining the relative ages of adults was discovered. Wherever appropriate, standard devia- tions and standard errors have been calcu- lated, but generally such figures are not presented here. Wlien it is stated that a significant or highly significant difference exists, it is based upon the use of the t test. FINDINGS Female Reproductive Tracts The reproductive tract of the female fisher is similar to that of other mustelids. The ovaries are completely encapsulated with only a small ostium through which a small portion of the fimbria extends. The ovary must be cut free from the bursa under a dissecting scope with a pair of fine scissors. The oviduct encircles the ovary as in other mustelids. The oviducts were not highly enlarged in any animals studied, since no estrous stages were seen. The uterus has a common corpus uteri which allows embryos 232 Maine Fishers • Wright and Coulter 73 developing in one horn to migrate to the other horn. The uterine horns are 40-60 mm long in adult females in inactive pregnancy, and 2y2-4 mm in diameter. Immature fish- ers show smaller uteri with horns about 30- 40 mm long and IV2-2V2 mm in diameter. No search was made for an os clitoridis. The ovaries from each preserved tract were dissected from the fixed reproductive tract, blotted, and weighed. Each ovary from animals taken in fall or early winter was sliced macroscopically and the number of corpora lutea present determined by the use of a dissecting microscope. Of the 77 tracts handled in this way, 44 animals showed corpora and were thus judged to be adults. Thirty-three animals were without corpora and were judged to be immature. The average combined weights of the ovaries was 134.4 mg for adults and 76.5 mg for immatures. The average weight of the left ovaries ( I — 40.3 mg, A— 70.0 mg ) was greater than that of the right ovaries (1—36.2 mg, A— 64.4 mg) in both im- matures and adults, but no special signifi- cance is ascribed to this matter. The aver- age number of corpora lutea from this series of 44 adult females was 1.68 in the right ovaries and 1.60 in the left; the average was 3.28 per adult female. Eadie and Hamilton (1958) reported that the mean number of corpora lutea in 23 adult New York fishers was 2.72. The difference in the average number of corpora lutea between the Maine and New York samples is highly significant. The distribution of the corpora lutea from all of the Maine, pregnant animals is shown in Table 1. To determine the relationship between the number of corpora lutea in the ovaries and the number of blastocysts in the uteri, 11 tracts of adult females were studied in detail. After the ovaries were removed and sectioned by hand, uteri were selected that appeared to be the best preserved. These Table 1. Distribution of corpora lutea in ovaries of preg- nant Maine fishers. No. OF Cor- No. OF pora IN Corpora in No. OF Both Ova- No. of Single Cases ries of Females Ovaries Individual Females 4 3 2 1 0 2 14 42 43 8 Total 109* 5 4 3 2 1 21 30 2 54 * One case in which only one ovary available. entire uteri were dehydrated and cleared in wintergreen oil. Study of the entire cleared tract under a dissecting scope using transmitted light often revealed the loca- tion of blastocysts. Serial sections of each of these tracts were made to locate the blastocysts. As soon as all of the expected blastocysts were found, no further section- ing of that tract was done. In some cases the entire uterus was sectioned before all the blastocysts could be located, and in 2 of the 11 tracts, 1 potential blastocyst was not found. This represents a loss of only 6 percent, as there were 35 corpora in the ovaries of the 11 animals and 33 blastocysts were located. The technique of Hamilton and Cook (1955:30-31) of flushing the uteri for the blastocysts was not followed here since the tracts had been fixed in for- malin. The sectioned blastocysts were similar to those described by Enders and Pearson (1943:286). The extremely thick zona pel- lucida, 14.4 /a according to these authors, makes it possible to find the blastocysts in very poorly preserved material. None of the blastocysts studied was in better condi- tion than those seen by Enders and Pearson, and the relative numbers of nuclei in the trophoblast and the inner cell mass for this species is still not known. In order to ob- 233 January, 1965 February 2, 1961 118 77 3 1 February 7, 1956 February 21, 1964 Late February, 1959 March 3, 1959 182 110 98 179 98 73 108 138 3 2 1 3 0 1 3 1 March 11, 1965 2 1 March 13, 1956 March 20, 1957 92 121 92 137 2 1 1 2 74 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 Table 2. Findings in nine reproductive tracts of female fishers in active pregnancy. Weight of Distribution of Date Ovaries (mg) Corpora Lutea State of Killed Uterus Right Left Right Left 3 embryos, 18 mm CR (Crown-Rump) 4 embryos, 2R, 2L, 17-mm swelhngs, em- bryo 8 mm CR 3 embryos, IR, 2L, embryo 13 mm CR 3 embryos, 2R, IL, embryo 18 mm CR 4 embryos, 2R, 2L, embryo 8 mm CR 4 fetuses, 2R, 2L, 2 males, 2 females, fetuses 53, 54, 55, 57 mm CR 3 fetuses, 2R, IL, 3 males, fetuses 69, 71, 74 mm CR 3 early embryos, 2R, IL, 7-mm swellings 3 fetuses, 2R, IL, 3 females, fetuses 74, 80, 83 mm CR tain such material, adult tracts would have young before the end of February. Two re- to be preserved in a matter of minutes after cently captured females produced litters on the animal was killed. March 2 and on March 20 at the Maine Tracts of nine adult fishers in which Unit. The evidence indicates that the ma- there were implanted embryos were studied jority of Maine fishers produce their litters (Table 2). Studies of the marten and a during the month of March, but some do so weasel are of some value in estimating the as early as mid-February, and some as late times of parturition in these tracts. Jonkel as early April. and Weckwerth (1963:96-97) made a The ovaries of female fishers with im- series of laparotomies on late-winter adult planted embiyos were all serially sectioned, female marten ( Martes americana ) and de- The ovaries are much larger than those in termined that the interval between implan- inactive pregnancy, the average combined tation and parturition was less than 28 days, weight being 231.9 mg as compared with In the long-tailed weasel (Mustela frenata), 134.4 mg for the inactive group. The Wright (1948) showed that the postimplan- corpora lutea are markedly enlarged in tation period lasted about 23 or 24 days, active pregnancy as is generally known in In estimating the parturition dates from the mustelids with long periods of delayed im- pregnant fisher tracts it is assumed that the plantation (Wright 1963:87). The corpora period of active pregnancy is about 30 days, of three of these animals averaged 2,380, This seems reasonable on the basis of the 2,917, and 3,057 /x in diameter, whereas larger size of the fisher in comparison with corpora from two animals with unimplanted the marten and the weasel. blastocysts averaged 1,387 and 1,219 /x. Al- The female fisher with the largest fetuses, though these corpora in animals with im- taken on March 20, would probably have planted embryos are more than seven times borne young before April 1. The one with the volume of those with unimplanted em- the earliest stages was taken on March 13, bryos, the increased size of the ovaries is and it is estimated that her litter would not not due solely to the increase in corpus size, have been born until after April 1. The one In no case is the histological preservation with the 13-mm ( crown-rump ) embryos, of high quality, but the corpora lutea were taken on February 7, would have borne her readily seen and counted in all ovaries. 234 Maine Fishers • Wright and Coulter 75 There is a great deal of interstitial tissue in 11 tracts which were preserved during in- all of these ovaries, and in this they differ active pregnancy and which were sectioned from weasel ovaries (Deanesly 1935:484) in to locate all of the blastocysts, which the interstitial tissue is most active On a few occasions at the time of au- in late summer but by implantation time topsy, Coulter observed darkened areas in shows considerable degeneration. There the uteri which were apparently placental are also numerous small and medium-sized scars. After being fixed and cleared, most follicles in these fisher ovaries. In all cases of these areas were no longer visible, the cells of the corpora lutea are highly Wright (1966:29) found that in the badger vacuolated. Vacuolated cells in corpora are ( Taxidea taxus ) placental scars can readily common in many mustelids during the be found in cleared tracts of parous fe- period of inactive pregnancy. Eadie and males, provided the uteri were preserved at Hamilton (1958:78) noted that their fisher once after death. Placental scars are diffi- corpora in ovaries in inactive pregnancy were cult to find, even in lactating badgers, in highly vacuolated. Wright and Rausch material that is not freshly preserved. It (1955:348-350) describe vacuolated corpora seems likely that the general level of preser- in the wolverine {Gido gulo) in inactive preg- vation in these fisher tracts was not good nancy, but during active pregnancy the enough to preserve placental scars, vacuolation had disappeared. It appears then that vacuolated corpora lutea during Breeding Season active pregnancy is a condition not com- Earlier workers, Hall (1942:147), for monly seen in this group. We suppose that example, indicate that the female fisher the corpora lutea of active pregnancy are breeds soon after her litter is born; thus secreting progesterone, whereas during the the gestation period may be as long as 51 inactive period there may be no active weeks. Since no recently postparturient secretion of progesterone. This is suggested tracts were available for study, this particu by the urine analysis conducted in various lar point could not be confirmed from wild- stages of pregnancy by Ruffie et al. ( 1961 ) caught animals. However, among speci- on the European badger {Meles meles) mens collected in late March and early which has a similar reproductive cycle. April, 1957, two recently bred nulliparous The number of embryos or fetuses in females were obtained and the tracts pre- these eight animals averaged 3.38 and in served fresh. These two tracts are the bes*- each case the number of corpora lutea cor- preserved in the entire series, and tuba responded to the number of embryos; that embryos were found in each by serially sec- is, there was seen here no loss of potential tioning the oviducts. Each animal had embryos that may have occurred during three corpora lutea, 2 R, and 1 L, and 3 either the preimplantation period or the morulae were found in one and 2 in the postimplantation period. other. In the one taken on March 28, one There was evidence of migration of em- morula had about 228 nuclei (Fig. lA); the bryos from one uterine horn to the other in other embr\'os were of comparable devel- five of the eight animals. Migration of em- opment, but it was not possible to count bryos is well known in other mustelids. It the nuclei. apparently occurs largely during the process The animal taken on April 4 showed 2 of spacing just before implantation. Only morulae with 12 and 20 nuclei (Fig. IB). one example of migration was seen in all No evidence was found of the expected 235 76 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 -^ m \h '■f : 2200- ■ 2000- UJ 1600- 1400 1200- 1000- 800 600- 400 • • • • t o* o •• ^- • • o o o o o o OCT NOV T DEC JAN TT^rr MAR T APR Fig. 3. Baculum weights. Adults are shown in solid dots, juveniles with open circles. The continued growth of the juvenile baculum during the winter months is clearly shown as is the overlap in weights of February, March, and April juveniles and adults. as it probably does in all mustelids. This was demonstrated (Wright 1950) to be the case in the long-tailed weasel. Probably the fully adult type of baculum would develop by late spring in these year-old males, since Deanesly (1935:469) concluded that the adult baculum of the European stoat {Mustela erminea) develops to adult type within 1 month after the testes become active for the first time. Although Elder (1951:44) showed that bacula may continue to develop in succeeding years in sexually mature mink (Mustela vison), the lack of known-age fishers in this series does not make such conclusions possible here. The tentative conclusion reached by de Vos (1952), that bacula were not of value in distinguishing adult from juvenile fishers, resulted from failure to recognize changes 239 80 Journal of Wildlife Managemeni, Vol. 31, No. 1, January 1967 240 Maine Fishers • Wright and Coulter 81 in the rapidly maturing skulls of juvenile male fishers during the late winter. This will be discussed further in a later section. Skulls The specimens were placed in four groups (adult males, juvenile males, adult females, and juvenile females) on the basis of reproductive condition and skeletal analysis, and 12 measurements were taken of each skull (see Wright 1953:78-79). Means, standard errors, and coefficients of variation were calculated for each group. It is clear from study of these statistics that the skulls of tlie juvenile animals in both sexes have not reached maximum growth. In many cases the differences between the means is statistically significant, but, be- cause of overlap between the measure- ments in adults and juveniles, it is not pos- sible to develop aging criteria based on measurement of a single skull parameter, with one exception to be discussed later. The differences between the means of these measurements was generally much greater among males than among females. For example, the mean weight of adult male skulls was 70.6 g, whereas in juvenile males it was 53.9 g, a difference of some 20 percent. In female skulls, however, the adults average 32.1 g and the juveniles 31.1 g, a difference of only 3 percent. The postorbital constriction becomes somewhat smaller with increased age in both sexes of fishers, as it does in other mustehds. Another striking difference be- tween adult and juvenile skulls was seen in males where the zygomatic breadth averages 77.4 mm in adults and only 64.8 mm in juveniles. In spite of this 18 percent smaller measurement in juveniles, there is overlap. It is not possible to classify a male fisher as juvenile or adult solely on the basis of this measurement. Tire difference in zygomatic breadth would, in most cases, produce a broader appearing face on adult males than on juvenile males. The sutures in the skulls of fishers, like those of all other mustelids, tend to dis- appear at a relatively young age (Marshall 1951:278, Greer 1957:322^23) as com- pared to the Ursidae, for example, where they persist for many years (Rausch 1961: 86, Marks and Erickson 1966:398). Juvenile male fishers taken in early fall (Fig. 5A) show almost all of the sutures unfused, but on specimens during March or April (Fig. 5C) almost all have completely disappeared. Eadie and Hamilton (1958:77) showed, in New York fishers from which they had re- productive tracts, that "All breeding fe- males showed at least partial fusion of the temporal ridges ... to form a sagittal crest, and [that] the maxillary-palatine sutures were completely fused. Non-breeding fe- males showed the temporal ridges in various degrees of separation and had the maxillary- palatine sutures at least partly open. It is concluded that female fisher normally breed at the age of one year in the wild, and that these criteria will separate young- of-the-year from adults." Fig. 4. (A) Lateral view of skull of winter juvenile male, February, showing well developed sagittal crest and open zygo- matic-maxillary suture. (B) Lateral view of skull of fully adult male showing typical tremendously developed sagittal crest and disappearance of zygomatic-temporal suture. The heavily worn teeth shown ore not necessorily characteristic of adult fishers. (C to J) Bacula of male fishers showing progressive changes with age, distal end to the top, the youngest to the left and oldest to the right. C, D, and E ore from juveniles, C taken October 12, D taken December 3, E taken January 5. F and G are from late winter juveniles showing progressive changes toward the adult type with increased deposition of bone at the basal end. Both F and G were taken in February or early March. H, I, and J ore selected adult bacula showing the characteristic oblique ridge near the basal end and generally more massive appearance. H is from a smaller-than-average male (body weight, 9 lb, 5 oz), I and J from iarger-than-overoge males (I, carcass weight 10 lb; J, body weight, 14 lb 6 oz). 241 82 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 242 Maine Fishers • Wright and Coulter 83 Our findings from study of 66 female fishers from Maine, from which comparable data were available, confirm in detail the findings of Eadie and Hamilton ( 1958 ) . It is also clear that the maxillary-palatine su- ture is among the last, if not the last, to disappear. These authors also describe a frequency distribution in the length of the sagittal crests in adult females, and reference to their Fig. 3 shows that there are two peaks of sagittal crest lengths which they tenta- tively regarded as representing a group of iy2-year-old females and another group of older females. When we plotted our data in comparable fashion, the line exactly paralleled theirs; and there is thus further evidence that such separation into young adults and older adults is possible. The distribution of the lengths of the sagittal crests of the adult Maine female fishers, plotted in the same fashion as did Eadie and Hamilton, is as follows: 0-10, 1; 11-20, 11; 21-30, 3; 31-40, 8; 41-50, 19; 51-60, 1. The findings in the skulls and skeletons of the two recently bred nulliparous fe- males, whose reproductive tracts were des- cribed in an earlier section, also provide significant evidence that the onset of breed- ing in female fishers occurs when they are 1 year old. In each case there was no sagittal crest, and the maxillary-palatine suture was partially open. Eadie and Ham- ilton (1958:79) found this suture closed in all New York fishers judged to be adults. In their collection, adult fishers, taken en- tirely in fall and winter, would have been at least 20 months old, whereas our two animals were almost exactly 1 year of age. One of these animals shows the pubo- ischiac symphysis still open; the other shows it partly closed. Further, the fact that during the fall and winter there is only one type of skull to be found in fishers that have not bred makes it virtually certain that wild Maine fishers are regularly im- pregnated at the age of 1 year and thus produce their first litters at the age of 2 years. The sagittal crests of adult male fishers are extremely well developed as was men- tioned by Coues (1877:65), and the degree of sexual dimorphism in skulls of fishers is greater than in any other American muste- lid. All adult females develop sagittal crests, but even the most highly developed crests in females are almost vestigial com- pared with those of adult males. It is natu- ral to suspect that with this tremendous development in mature males the crest might begin to develop earlier in juvenile males than in females. This is exactly the case, and sagittal crests were first seen in one of two juvenile males taken in Decem- ber (Fig. 5B). By February, March, and April the crests of the juvenile class, now almost 1 year old, are well developed ( Fig. 5C), as much so as they ever become in adult females. In the female fishers it is clear that the sagittal crest develops first at the posterior end of the skull and grows progressively Fig. 5. Dorsal view of male fisher skulls showing characteristic changes associated with development. (A) Juvenile mole, October 12, showing narrow zygomatic breadth, all sutures in nasal region clearly open; the frontoparietal sutures are partly fused. The poorly developed temporal lines are wide apart and thus there is no sagittal crest. (B) Juvenile male, December 3, showing disappearance of fronto-parietal suture, less conspicuous sutures in nasal region, and characteristic early development of sagittal crest running throughout the middle and posterior portions of the cranium. (C) Juvenile male in late winter, February, in which these naso-maxillary and maxillary-frontal sutures are barely visible, but the zygomatic-temporal sutures are still very distinct and the sagittal crest is better developed. (Same skull as shown in Fig. 4A). (D) Skull of adult male in which the entire dorsal skull is ankylosed into a single unit; no suture visible except for faint remains of posterior internasal suture. The characteristic highly developed keel-like sagittal crest of all adult males is clearly shown. (Same skull as shown in Fig. 4B). 243 84 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 forward over a period of months or prob- as well as his "juvenile" class, and that only ably years. In the male fisher the temporal the animals he called "old adults" were lines move rapidly together during the win- adult males over 1 year of age. ter months; and as soon as the crest is It is concluded, therefore, that during formed, it mns essentially the entire length the early winter, adult males can be sepa- of the dorsal region from the postorbital rated from juvenile males by the occur- constriction to the inion, a distance of 50- rence of a well developed sagittal crest on 60 mm. The sagittal crest continues to adults; but by mid- or late winter only those develop in adult males, and they have the males with all of the skull sutures closed crest developed to the extent of forming a are adults, "thin, laminar ridge" (Coues 1877:65). It is difficult to measure the extent of this Body Weights ridge objectively; but since it extends Both de Vos (1952) and Hamilton and posteriorly in fully adult males, one can use Cook ( 1955 ) have provided body weights the method employed by Wright and Rausch of wild-caught fishers, and both studies ( 1955 ) on wolverines to subtract the con- show that males often weigh twice as much dylobasal length from the greatest length as females. The latter indicate an average of the sk-ull. This is one accurate method weight for males of 3,707 g (8 lb 3 oz) and of showing the posterior extension of this 2,057 g (4 lb 9 oz) for females. De Vos's crest. This indirect measurement shows no figures are roughly comparable. In both overlap whatever between males classed studies many of the body weights were esti- as adults and those classed as juveniles, mated from carcass weights by applying a The mean for the former group is 11.9 mm correction factor to skinned carcasses. (Most and for the latter, 3.9 mm (see Fig. 4, A fisher specimens coming to biologists are and B). Thus in male skulls if the differ- likely to be carcasses skinned by trappers.) ence between the greatest length of the Hamilton and Cook (1955:21-22) state skull and the condylobasal length is 6 or that the fresh carcasses average 80 percent more mm (may be as much as 15 mm), of the unskinned weight. In the present the animal is an adult; if it is less than 6 study many fishers were confiscated and mm, the animal is a juvenile. were available intact. Thus, it was pos- Another reason for assuming that skulls sible to obtain a sample of weights taken of males with immature bacula, but with directly from the entire unskinned carcasses, sagittal crests, are still in their first year of allowing consideration of differences be- life is provided by data on the closure of tween adult and juvenile classes in both sutures in the skull. The last sutures to sexes. close in males are the zygomatic-temporal, Data obtained from those fishers which the naso-maxillary, the internasal, and the were weighed entire in the laboratory are naso-frontal. In all of the skulls classed as shown in Table 4. The differences between adult, all of these sutures were closed, but the juveniles and adults in both sexes is in every male skull classed as juvenile, all highly significant although there is some four of these sutures were still open ( Figs, overlap in each case. Furthermore, juvenile 4 and 5 ) . males are significantly heavier than the On the basis of this evidence, it seems adult females. The available mean weights clear to us that males classed by de Vos of adults are probably more satisfactory (1952) as "adults" were in effect juveniles than those of the juveniles. Presumably, 244 Maine Fishers • Wright and Coulter 85 tlie adults were no longer growing, but the ]u\eniles were growing throughout the collection period from October to April. The sample is not large enough to allow a breakdown \\ithin the juvenile classes b\" month, but the smallest juveniles were taken in the fall. The fact that weights of the juvenile males are 21 percent less than those of the adult males, while the weights of juvenile females are onh" 15 percent less than those of the adult females, further indicates that juvenile female fishers are more nearly full grown during the first winter of life than are the juvenile males. In man\- cases, the fishers that were weighed whole were also weighed after skinning. This allowed detennination of a correction factor. Thirty-nine animals were weighed both before and after skin- ning: 14 adult males, 5 juvenile males, 8 adult females, and 12 juvenile females. The carcasses averaged 81.9 percent of the whole weight; or, stated conversely, one could multiply the carcass weight by 1.22 to obtain an estimate of the entire adult weight. This latter conversion factor was applied to those animals that were weighed only after being skinned. Esti- mated entire body weights obtained in this fashion were comparable for both adult and juvenile males, but weights of females were significantly below the weights of those females weighed entire. For this reason, it was obvious that in the interval between skinning and weighing, many of the female carcasses had lost significant weight. It was therefore necessary to aban- don any attempt to use the more numerous carcass weights for interpretation of pos- sible growth rates in the juveniles or other weight changes that might exist between months. Table 4. Body weights of Maine fishers weighed whole. -No. Mean Body sf. Class ok Weight in Max. Min. Ani- (0UNt:ES) oz MALS Adult S 23 172.1 ±6.30 14- 6 7-4 (10 lb 12 oz) Juv. 6 10 135.5 ±7.08 10- 8 6-8 (8-71/2) Adult 9 13 88.2 ±3.61 7-11 4- 8 (5-8) Juv. 9 17 75.0 ±2.35 6- 8 3-13 (4-11) DISCUSSION Tliis study indicates that in the fisher the adult class consists of all animals more than 1 year of age and that all animals of both sexes less than 1 year are sexually imma- ture. Females older than 1 year normally are carrying unimplanted blastocysts through- out the year except during active preg- nancy in late winter. The fisher, then, differs from all other American mustelids studied in this regard except the wolverine. The weasels, Mustela erminea and M. frenata, are similar in that the males reach sexual maturity in 1 year; but the females breed during their first summer and thus produce young at the age of 1 year (Wright 1963:83-84). In the marten, males also apparently reach sexual maturity in 1 year, but females may not breed until they are 2 years old, and thus two year-classes of immature females may be found in wild populations (Jonkel and Weckwerth 1963: 95-96). This has made further refinement of Marshall's ( 1951 ) original study of marten quite difficult. In the female otter it appears that sexual maturity is delayed another year beyond that in the fisher and that there are two age-classes of immature otters (Hamilton and Eadie 1964:245). In the badger the same type of situation prevails as in the fisher except that some females breed pre- cociously during their first summer and 245 86 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 such females would produce litters at the age of 1 year, whereas most badgers pro- duce their first litters at the age of 2 years (Wright 1966:42). Only in the wolverine {Gulo gulo) does it appear that a repro- ductive cycle like that of the fisher is found; but owing to a small sample of ani- mals of the former species, the matter of age at sexual maturity is somewhat in doubt. The recovery of the marten in Maine has been much slower than in the fisher (Coulter 1959) although both species orig- inally occurred sympatrically in much the same habitat. The present study indicates that the potential rate of reproduction in the fisher is higher than in the marten. A large sample of winter-caught marten is not available from Maine, but such material was obtained from Montana. Wright ( 1963: 79) indicates that corpora lutea counts averaged 3.02 in a sample of 44 trapper- caught marten. The present study showed 3.28 for the fisher. Perhaps of greater sig- nificance, though, is the fact that some female martens (in Glacier National Park) (Jonkel and Weckwerth 1963) do not produce litters for the first time until they are 3 years old. LITERATURE CITED Canivenc, R., and M. Bonnin-Laffargue. 1963. Inventory of problems raised by the delayed ova implantation in the European badger {Meles meles L.). Pp. 115-125. In A. C. Enders (Editor), Delayed implantation. Uni- versity of Chicago Press, Chicago, Illinois. 309pp. CouES, E. 1877. Fur-bearing animals: a mono- graph of North American mustelidae. Dept. Interior, Misc. Pub. 8, Washington, D. C. 348pp. Coulter, M. W. 1959. Some recent records of martens in Maine. Maine Field Naturahst 15(2):50-53. . 1960. The status and distribution of fisher in Maine. J. Mammal. 41(l):l-9. Deanesly, Ruth. 1935. The reproductive proc- esses of certain mammals. Part IX: Growth and reproduction in the stoat {Mustela erminea). Philos. Trans. Roy. Soc. London 225 (528): 459-492. de Vos, a. 1952. Ecology and management of fisher and marten in Ontario. Ontario Dept. Lands and Forests Tech. Bull. 90pp. Eadie, W. R., and W. J. Hamilton, Jr. 1958. Reproduction in the fisher in New York. New York Fish and Game J. 5(l):77-83. Elder, W. H. 1951. The baculum as an age criterion in mink. J. Mammal. 32(l):43-50. Enders, R. K., and O. P. Pearson. 1943. The blastocyst of the fisher. Anat. Rec. 85(3): 285-287. Greer, K. R. 1957. Some osteological charac- ters of known-age ranch minks. J. Mammal. 38(3):319-330. Hall, E. R. 1942. Gestation period in the fisher with recommendations for the animal's pro- tection in California. California Fish and Game 28(3) :143-147. Hamilton, W. J., Jr., and A. H. Cook. 1955. The biology and management of the fisher in New York. New York Fish and Game J. 2(l):13-35. -, and W. R. Eadie. 1964. Reproduction in the otter {Lutra canadensis). J. Mammal. 45(2):242-252. Jonkel, C. J., and R. P. Weckwerth. 1963. Sexual maturity and implantation of blasto- cysts in the wild pine marten. J. Wildl. Mgmt. 27(l):93-98. Marks, S. A., and A. W. Erickson. 1966. Age determination in the black bear. J. Wildl. Mgmt. 30(2):389-410. Marshall, W. H. 1951. An age determination method for the pine marten. J. Wildl. Mgmt. 15(3):276-283. Neal, E. G., and R. J. Harrison. 1958. Re- production in the European badger {Meles meles L.). Trans. Zool. Soc. London 29(2): 67-130. Rausch, R. L. 1961. Notes on the black bear, Ursus americanus Pallas, in Alaska, with par- ticular reference to dentition and growth, Z. Saugetier. 26(2):77-107. RuFFiE, A., M. Bonnin-Laffargue, and R. Canivenc. 1961. Les taux du pregnandiol urinaire au cours de la grossesse chez le Blaireau europeen. Meles meles L. Comptes rendus des seances de la Societe de Biologie 155(4):759-761. Taber, R. D. 1956. Characteristics of the pelvic girdle in relation to sex in black-tailed and white-tailed deer. California Fish and Game 42(1):15-21. 246 Maine Fishers • Wright and Coulter 87 Wright, P. L. 1948. Preimplantation stages in the long-tailed weasel ( Mustela frenata ) . Anat. Rec. 100(4):593-607. — . 1950. Development of the baculum of the long-tailed weasel. Proc. Soc. Expt. Biol, and Med. 75:820-822. . 1953. Intergradation between Martes americana and Martes caurina in western Montana. J. Mammal. 34(l):74-86. . 1963. Variations in reproductive cycles in North American mustelids. Pp. 77-97. In A. C. Enders (Editor), Delayed implantation. University of Chicago Press, Chicago, Illinois. 309pp. . 1966. Observations on the reproductive cycle of the American badger {Taxidea taxus). Pp. 27^5. In I. W. Rowlands, Editor, Com- parative biology of reproduction in mammals. Symposia Zool. Soc. London, No. 15. Aca- demic Press, London. 527pp. , AND R. Rausch. 1955. Reproduction in the wolverine, Gulo gulo. J. Mammal. 36(3): 346-355. Received for publication August 22, 1966. 247 GROWTH, DEVELOPMENT, AND WING LOADING IN THE EVENING BAT, NYCTICEIUS HUMERALIS (RAFINESQUE) Clyde Jones Abstract. — Selected aspects of growth and development of young evening bats are presented and summarized. In addition, information on wing loading and de- velopment of flight in known-age animals is given. Data regarding growth, de- velopment, and wing loading of Nycticeius humeralis and information that is available for some other species of bats are compared and discussed. Few observations have been made previously on growth and development of bats, and such information has been frequently incidental to other studies of natural history or reproduction. Ryberg (1947), while presenting data on parasites and natural history of bats, made some mention of growth and de- velopment of young. Considerable information on growth and development of two North American bats has been contributed by Pearson et al. ( 1952 ) and Orr ( 1954 ) . Some observations of young and early growth and develop- ment of Nycticeius humeralis were noted by Gates (1941). In general, re- production of bats has been summarized by Cockrum ( 1955 ) and Asdell (1964). To my knowledge, no information has been made available with regard to wing loading of Nycticeius humeralis. Vaughan ( 1959 ) , while presenting data on aerodynamic considerations of three species of bats, provided a survey of important earlier works. More recently, limited information on flight of some North American bats has been contributed by Struh&aker (1961), Davis and Cockrum (1964), Hayward and Davis (1964), Vaughan (1966), and others. The purpose of this report is to present information on growth and develop- ment and to discuss briefly some aspects of wing loading of Nycticeius humeralis. Materials and Methods This report is based upon observations of 28 young of 14 litters born in captivity to females netted at Clear Springs, Homochitta State Park, Franklin County, Mississippi. The adults were captured between 11:30 pm and 4:30 am on 9 and 10 May 1965. Pregnant bats were weighed periodically prior to parturition; following birth of the young, weights and measurements of all bats were taken regularly. The growth and development of three litters were followed in detail and the animals were measured daily. The remaining animals were measured each 3- or 5-day period. Because the bats were not anesthetized, it was not always possible to obtain relaxed individuals for measuring. 1 248 2 JOURNAL OF MAMMALOGY Vol. 48, No. 1 As a result of these methods, some discrepancies in the measurements of individuals from one date to another are evident in the data. Measurements that were taken include total length, length of tail, length of foot, length of ear from notch, length of forearm, and length of fifth finger. All measurements were taken with a Vernier caliper. At regular intervals of age of the animals, wings were outlined for the analysis of surface areas utilized in flight. For the purposes of measuring wing loading, I have followed the assumption of Vaughan (1959) that the wings extend through the body and I have computed the wing loadings in lb per sq ft. Pregnant females and females with young were housed in one-quart cardboard cans with screen tops. Following the weaning of the young, each female and her offspring were housed in one-half gallon cardboard cans with screen tops. Cardboard cans containing bats were stored on the sides. The animals were maintained at room temperatures and provided a daily diet of larvae of Tenebrio molitor and water with Theragran (Squibb Therapeutic Formula Vitamin) added. Feeding, watering, and handling of the bats were started at about the same time every day, usually between 3:30 and 5:30 pm. For purposes of identification, adult bats were banded. Young animals were toe clipped soon after birth, but were banded when adult size was attained. Some animals were sacrificed at various stages of development and then cleared and stained or were preserved either in fluid or as dried specimens. Results Birth of young. — The females gave birth to the young within 15 to 26 days after capture. The births recorded herein occurred on 25 May (two Htters), 26 May (one Htter), 29 May (two Htters), 30 May (five htters), 31 May (two htters), 2 June (one Htter), and 4 June (one Htter). Fourteen adult females gave birth to 12 females and 16 males. This ratio of sexes (0.75 to 1.00) was similar to the ratio (0.736 to 1.00) reported by Hooper (1939). AsdeH (1964) reported two young as the usual number per litter and stated that birth oc- curred in late May. All births observed were by breech presentation and the times recorded for births were from 3 to 114 minutes. With the exception of two litters that were bom at about 9:00 and 10:45 am, all litters were dropped between 1:00 and 4:00 pm. During parturition all females moved to the bottoms of the containers. Females ate the placentae and umbiHcal cords and licked the young very soon after birth. Placentae were eaten first, then umbilical cords were eaten to within 2 or 3 mm of the naval area. Some of the females hung head up on the screening of the containers while eating the placental materials, but then turned head down and licked the young thoroughly. Placentae and umbilical cords were eaten and the young were licked usually within 35 to 70 minutes after delivery. In every case observed, the young found and grasped the nipples of the adult within a short time, usually 5 to 8 min following birth. The young were aided often by the adult in climbing to the mammary glands. The young were oriented in the same direction as the adult. The plagiopatagium and uropatagium of the female enveloped the young. The young bats held firmly to the nipples of the females and had to be removed forcibly for measuring. 249 February 1967 100 -1 E E c o o 90 80 70 - 60 ■S 50 40 - 30 20 JONES— EVENING BAT (' \\. 1 1 1 1 1 1 1 1 1 1 0 10 20 30 40 50 60 70 80 90 100 Days Fig. 1. — Total lengths of young Nycticeius humeralis. Dots represent the arithmetic means and lines represent the ranges of measurements. 250 JOURNAL OF MAMMALOGY Vol. 48. No. 1 E E o o -I 42 38 34 30 o 26 22 18 - 14 10 ( ■ ■ . lip 1 — I 1 1 1 1 1 \ \ 1 1 0 10 20 30 40 50 60 70 80 90 100 Days Fig. 2. — Tail lengths of young evening bats. Dots represent the arithmetic means and lines represent the ranges of measurements. 251 February 1967 JONES— EVENING BAT Table 1. — Correlation of age and various aspects of growth and development of young evening bats. , Cfl *J 1 g 5 -o O >" .S E c o 0; a! "a _4; "H '^ "a "gs 1 •« c 1 c 5 .5* 'c S 1^ o o 1 « S II ^ a 1° 1^ E ii C Age in a * .So >> "0 C 2 2 -1 C3 O 2 S 4~> ^ 0 weeks < £ o o n ol it' "- ^ % « 0 1 X X 2 X X X X 3 X X X X X X X 4 X X X 5 X 6 X X 7 X X 8 X 9 X In addition, the feet and first fingers were utilized for clinging to the fur of the adults. '^kin and pelage. — Young bats examined within 1 hr after birth were pink with smooth, soft skin. Vicera were seen through the skin of the abdomen. There was slight dark pigmentation on the feet, membranes, tips of the pinnae, and lips. Only a few hairs were present on the feet and on the dorsum of the head and shoulders. Some vibrissae were evident on the swollen glandular areas of the lips. When about 6 hr old, the young had more pigmentation on the dorsal sides of the back and head than at birth. Pigmentation became evident on the venter at approximately 18 hr of age. Within 24 hr the dorsum was pigmented heavily and the venter was pigmented except for a small abdominal area that appeared rather opaque. The skin was very wrinkled and had the appearance of being hard and dry, but was soft and pliable to the touch. In the 2-day-old young, a few hairs appeared on the dorsum at the base of the uropatagium. The hairs and vibrissae on the feet and lips were noticeably stiffened. At 3 days of age a small patch of hair became noticeable on the dorsum over the scapulae. By the 4th day, pelage was seen on the dorsum over the scapulae, on the rump, and along the flanks. At this age, fur was first apparent on the venter at the base of the uropatagium and in the pectoral region. At 5 days of age, fur was present on the dorsum across the scapular region to the flanks and extended onto the rump. At this time the fur was short, soft, and gray in color. 252 6 JOURNAL OF MAMMALOGY Vol. 48, No. 1 E o a» 42 1 38 - 34 - E E 30 26 - ^ 22 18 - 14 - 10 _ 1 "T- 10 — T" 20 30 T 40 Days 50 — 1 1 1 1 1 60 70 80 90 100 Fig. 3. — Forearm lengths of young Nycticeius humeralis. Dots represent the arithmetic means and Hnes represent the ranges of measurements. 253 February 1967 JONES— EVENING BAT 7 On the 6th day, soft gray hairs covered most of the dorsum, and, except for a bare abdominal region, much of the venter. The feet were well furred. At 7 days of age, the hairs on the dorsum, in the scapular region, appeared longer and darker than the rest of the general pelage. At 8 to 9 days of age, the young were furred completely with grayish black hairs on the dorsum with long, dark fur over the scapulae. In contrast to the dorsum, the venter was grayish white in color. The aforementioned general appearance of the pelage remained until the young were approximately 30 days of age. At this time the hair became burnished slightly with brown at the tips, perhaps due to wearing of the ends of the hairs. Young bats were not observed to undergo molt during the course of this study. At the time the young bats reached about 80 to 95 days of age, the parent females molted; the pelages of young and adults were similar in appearance. Eyes, ears, and vocalization. — At birth the lids of the eyes were sealed, but the line of fusion was very evident. At 18 to 24 hr following birth, the eyes opened. At this age, the young would jump and scamper about in response to the flash of light from a photographic strobe. Young examined soon after birth had pinnae that were folded over. When the young were 24 to 36 hr old, the pinnae were unfolded and held erect. The young bats were vocal almost immediately after birth. The utterances of weak "squeaks" or "chirps" seemingly were continuous for about 10 days. After this time, bats made vocal sounds only when disturbed or handled. Dentition. — The complete number of deciduous teeth in young Nijcticeiiis humeralis is expressed by the formula i 2/3, c 1/1, p 2/2 = 22. Examination of newly-born young revealed that all of the deciduous teeth were erupted at birth. The deciduous teeth have two accessory cusps, one on either side of the main central cusp. In general, each cusp is in the shape of a hook and is curved backward and inward toward the mouth. The highest degree of development of curved, hook-shaped cusps is on the incisors. The cusps of the canines are hooked noticeably, but some premolars have relatively poorly developed accessory cusps that appear as small bumps rather than hooks. Hooked cusps are more highly developed in 2- and 7-day old young than in young 1 day of age. The complete number of permanent teeth is expressed by the formula i 1/3, c 1/1, p 1/2, m 3/3 = 30. In specimens of young 2 days old that were cleared and stained, the crowns of the permanent teeth are clearly visible in a position internal to the deciduous teeth. In specimens of young 7 days old, the permanent canines have penetrated through the gums and crowns of the other teeth are seen at the gum line just beneath the surface. In the order of appearance of permanent dentition, the canines become apparent first, fol- lowed in eruption by the incisors, premolars, and molars. At 4 weeks of age, the permanent teeth are generally in place; the third molars may not be in place fully, but the crowns are apparent well above the gum lines. 254 8 JOURNAL OF MAMMALOGY Vol. 48, No. 1 42 -1 38 34 - 30 - S 26 ^ 22 c -I 8 - 4 - O O 00 o o °o OcPo° o ° o o oo o o ° o o o o o • * • _*' o o ° .•• •.•^* •••^ o • , • • o • • • • • o -• •• • «•*•• . •• ^ o o^ 1 1 1 1 1 1 1 1 1 f 0 10 20 30 40 50 60 70 80 90 100 Days Fig. 4. — Forearm lengths of known-age evening bats. The open circles represent the arithmetic means of measurements for females; the dots represent the arithmetic means of measurements for males. 255 February 1967 JONES— EVENING BAT 9 Digits. — The digits of the hind limb were separated at birth and the claws were well developed and pigmented. Young bats at birth had feet nearly equal in size to the feet of adults. The first finger of the forelimb was developed very well at birth and thumbs of young were similar in size to thumbs of adults. Measurements of the length of the first finger of young and adults were 5.5 to 6.2 mm. The remaining digits of the forelimb were developed to a lesser degree in young than in adults. In the young less than 15 days of age, the distal portion of the forelimb was less developed than the proximal part of the wing; digits two to five were shorter in length than the forearm. In adults, the lengths of digits two to five were 18 to 61% greater than the length of the forearm. Flight of young. — In an attempt to determine the exact age when the young bats could fly, four young were thrown into the air each day when the adults were fed; four young and adults were housed in containers placed on a shelf 4.6 ft from the floor and the tops of the containers were taken off at feeding time; the remaining young and adults were cared for in the manner mentioned previously. When the young bats 10 to 14 days of age were thrown into the air, the wings were extended and fluttered, but the animals simply fell to the floor, sometimes without righting themselves. When the bats were 15 days of age, the fifth fingers were flexed and the bats would right themselves and then glide or "parachute" to the floor. At 18 days of age, bats thrown into the air flew 10 to 12 ft, but would hit a wall or the ceiling and flutter down to a flat surface such as the floor or a desk or would hang head up on a wall. At 19 days of age, one young bat housed on the aforementioned shelf emerged and flew across the room. When the young were 20 to 21 days old, they were observed to negotiate turns, land, and hang head down from walls and ceilings of the room. All of the young animals, including those with no practice previously, could fly short distances ( 10 to 12 ft) at 21 days of age, and all could fly well, turn, and land head down on the walls or ceilings at 23 days of age. All observations of flight were made in a room that measured 18 by 15 ft with a partial comer partition separating an area about 8 by 8 ft. The young bats could fly and avoid obstacles well in these spaces. Behavior of young and adults. — The young evening bats seemed weak and uncoordinated at birth. Although the babies attached themselves firmly to the parent, young 1 day of age seemed rather helpless when separated from the adult. At 1 day of age, the young could crawl about only feebly and were unable to right themselves when placed on the dorsum on a flat surface. By 3 days of age the young could crawl about very well and could right themselves quickly. In part, these abilities may be a reflection of the afore- mentioned unfolding of the pinnae at this time. For nearly the first 2 weeks of age the young were attached to the nipples of the adults almost constantly and remained enveloped by the membranes of the adults. With the exception of a few occasions when a young bat was 256 10 JOURNAL OF MAMMALOGY Vol. 48, No. 1 E E c 42 n 38 - 34 - 30 26 H- 22 - 18 - 14 - 10 \ 1 1 1 1 1 1 1 1 1 0 10 20 30 40 50 60 70 80 90 100 Days Fig. 5. — Fifth-finger lengths of young Nycticeius humeralis. Dots represent the arith- metic means and hnes represent the ranges of measurements. 257 February 1967 42 E E E E c o» o» c c 0> a> _l -I II II o X 38 34 E ;;: k. a o .ti ^ ^ 26 14 JONES— EVENING BAT 11 E 30 - XX ^x"** X X X X XX X ,X X xx v» X X X X X X X « "X 0 ^° 0, 0 00 0 0 0 X ° ° 0 <: X . 0 00 0 Cb%cPo%° _ O 0 0-, 00 0 ^0 00 0 X 0 22 - 8 - 0 o°o°o O^x 0 0 X 0 o x\ ,t X ox IX 1 1 1 1 1 1 1 1 1 1 0 10 20 30 40 50 60 70 80 90 100 Days Fig. 6. — Relationships of forearm lengths and fifth-finger lengths of Nycticeius humeralis. Open circles represent arithmetic means of measurements of forearm lengths; crosses represent arithmetic means of measurements of fifth-finger lengths. observed uncovered, the young did not leave the close association with the parent and move about in the containers until about 3 weeks of age. After this time, young scampered frequently about the cages, but hung adjacent to the female when at rest. 258 12 JOURNAL OF MAMMALOGY Vol. 48, No. 1 10 n 9 - 8 - 6 - E o o> c 5 4 - 3 - 2 - — I 1 1 1 1 \ 1 1 1 r 0 10 20 30 40 50 60 70 80 90 100 Days Fig. 7. — Weights of young bats. Dots represent the arithmetic means and lines represent the ranges of weights. 259 February 1967 JONES— EVENING BAT 13 Table 2. — The relationships of wing loading, proportions of the forearm and fifth finger, and total body weight of known-age Nycticeius humeralis. Age in Weight Wing loading Length of forearm/ Size of days in g in Ib/sq ft length of fifth finger sample 1 2.0 0.5365 1.00 11 2 2.2 0.4067 0.99 11 4 2.6 0.4523 1.02 7 5 2.8 0.3765 1.02 8 8 3.0 0.3300 1.04 11 12 3.4 0.2500 1.02 19 15 3.8 0.2500 0.99 23 18 4.2 0.2433 0.97 23 35 5.2 0.2375 0.93 23 43 5.6 0.2220 0.85 22 60 6.2 0.2079 0.85 19 73 7.1 0.1921 0.85 19 97 8.4 0.2261 0.83 5 As mentioned previously, the young were highly vocal for the first 10 days following birth, but then made vocal sounds only when disturbed or handled. During these observations, the adults emitted sounds only when disturbed, handled, or sometimes when offered food. Some definite specificities of adults for their young were noted. As long as the young bats were returned to the same nipples from which they were taken, no female refused to accept the young after they had been removed from the mother and measured or handled otherwise. On several occasions attempts were made to get adult females to accept nursing young from other females; all such efforts failed. The adults would bite and move away from the strange young. One young bat that was allowed to become attached to the nipple of a restrained female was attacked and thrown from her when the adult was released. This same adult accepted her own young a few minutes later. Litters and females could not be mixed successfully until nursing ceased. The refusal of adult females to accept other young may be a reflection of the manner in which the animals were maintained in relative isolation from other young and adults. Gates ( 1941 ) reported that he detected no specificity with regard to nursing young and adult females when the young and adults no longer remained together during periods of feeding. The young first showed an interest in food and water at approximately 3 weeks of age, when they appeared to smell and lick items of food ( portions of mealworm larvae) held before them. Early interests of young in water included considerable licking of the end of a water-filled dropper. At the age of 4 weeks, young bats were taking water from a dropper and eating small mealworms that were presented with forceps. Throughout the course of this study, the adults were given mealworms from forceps and water was administered from a dropper. In only two cases 260 14 JOURNAL OF MAMMALOGY Vol. 48, No. 1 14 -, 12 - 10 - i 8 • 6 - 4 - 2 - •• V*^* • •• •v.\\.. •Art-^" • % -• >v / T T T T T T T — \ r 1 n 20 10 0 10 20 30 40 50 60 70 80 90 100 Days Fig. 8. — Weights of Nycticeius humeralis. Open circles represent arithmetic means of weights of adults and dots represent the arithmetic means of weights of young. did individuals become accustomed to picking up and eating mealworms that were not presented by hand. In general, the bats made little effort to fly or move about while being fed and it was possible to feed four to six animals at one time. One adult crawled about almost continually while being fed. The young reared by this female behaved in similar fashion during the periods of feeding. Some animals began eating immediately when food was offered; others simply held a food item in the mouth for a short time. During this time the 261 February 1967 JONES— EVENING BAT 15 Flight Fig. 9. — Actual outlines of a wing of a young Nycticeius htimeralis of known age. Wing outlines were made when the bat was 1, 2, 4, 5, 8, 12, 15, 18, 35, 43, 60, 73, and 97 days of age. animals exhibited considerable shivering, presumably while the body tem- perature was increased. Growth. — Information relating to growth and development of young Nyc- ticeius humeralis given herein is based upon bats bom and reared in captivity. No malformations of any kind were noted and all animals seemed normal, but it is possible that under laboratory conditions the rate of growth and development may not have been normal. Data regarding growth and development of young bats are presented in Tables 1 and 2 and Figs. 1-9. Because the length of the first finger and length of foot of newborn bats were noted to increase little between birth and adulthood, those measurements are not depicted graphically. At 45 to 50 days of age, total length, length of tail, and length of fifth finger of young bats were of adult proportions and little growth occurred thereafter (Figs. 1, 2, and 5). Length of forearm, on the other hand, was of adult proportions when the bats were about 30 days of age (Figs. 3 and 4) and little growth occurred at later ages. Some sexual dimorphism of the length of the forearm was noted (Fig. 4). The smaller size of the forearm 262 16 JOURNAL OF MAMMALOGY Vol. 48, No. 1 of males became apparent when the animals were about 20 days of age. Pearson, Koford, and Pearson ( 1952 ) found a similar situation in their studies of growth of Plecofus townsendii. In young bats, the proportional relation- ships of the length of the forearm and the length of the fifth finger (Table 2 and Fig. 6) are correlated with flight and are discussed later with regard to flight. \\'eights of young bats were noted to increase rather constantly throughout the period of study (Figs. 7 and 8). Some activities of the bats, such as the development of the ability for flight and the acceptance of mealworms and water are reflected in the weights of the young. The gradual cessation of nursing and lactation seemingly is reflected more vividly in the weights of the adults ( Fig. 8 ) than in the weights of young during the same period of time. Growth, de\elopment, and relative surface area of the wings of bats of known ages are depicted in Fig. 9. The data presented indicate a gradual increase in surface area of the wing with the development of a wing of adult proportions at the age of 60 to 97 days; surface area is stabilized with the cessation of growth. At the age when young bats first began to fly, several changes in the wings were apparent for the first time. The length of the fifth finger was greater than the length of the forearm (Table 2 and Fig. 6) and the distal portion of the wing, that area from the apex of the wing to the first and fifth fingers, was nearly equal in surface area to the proximal portion of the wing, that region between the body and the first and fifth fingers (Fig. 9). In addition, there was a change in the ratio of body weight to surface area of the wing ( Table 2 ) . Discussion Data on various aspects of growth and development of some species of North American vespertilionids have been presented by Pearson et al. ( 1952 ) , Orr (1954), and others. The availibility of these data permits a general com- parison of the rates of growth and development of Nycticeius humeralis with those of Flecotus townsendii and Antrozous pallidus. The development of fur over the entire body of young evening bats was completed by 8 to 9 days of age. Short gray hair covered the bodies of Plecotus 4 days old and scanty fur was evident on the bodies of Antrozous at 10 days of age. The eyes of young Nycticeius were opened at the age of 18 to 24 hr, but eyes of young Plecotus and Antrozous were not opened until the age of 7 to 10 days. Pinnae of young Nycticeius were erected after 2 to 3 days of age; pinnae of the two other species were erected after 7 days of age. Vocalization was noted almost immediately following birth of Nycticeius and was evident within a few hours after birth of Plecotus. The complete set of deciduous teeth of Nycticeius was present at birth and these teeth were grown out fully by 7 days of age. Orr ( 1954 ) found the deciduous premolars lacking in newborn Antrozous, but noted that the 263 February 1967 JONES— EVENING BAT 17 deciduous teeth were grown out fully by the second week of age. In Nycticeius, all of the permanent teeth were erupted at an age of 4 weeks, and in Antrozous, permanent teeth were erupted at an age of 5 weeks. The forearm in young Nycticeius was of adult proportions at 30 days of age, but the forearm in young Plecotus reached adult size at 21 days of age. Comparisons of the rates of growth and development of the young of Nycticeius humeralis, Plecotus townsendii, and Antrozous pallidus, indicate that young Nycticeius exhibit more rapid growth and development and are perhaps more precocious than young of the other species. On the basis of the data presented in this report, it seems that young Nyc- ticeius humeralis, prior to 18 days of age, simply lack adequate surface areas of wing membranes to support the weight of the body in flight. For informa- tion with regard to surface areas of flight membranes and body weights of bats, see Vaughan (1959 and 1966) and Struhsaker (1961). In addition to the relationships of surface areas and weights, other factors of growth and development must have considerable bearing on the abilities of young bats to fly. For example, both lift and power for flight of young bats must be highly dependent upon the development of the ventral thoracic flight muscles as well as development of the musculature of the entire forehmb. For a discussion of surface areas of flight membranes and volumes of fhght muscles in relation to total volumes of the body, see Struhsaker (1961), and for descriptions and discussions of flight muscles, see Vaughan (1959 and 1966). It was noted that after the forearm of the bats observed during this study reached maximum length ( at about 30 days ) there was an increase in diameter of this portion of the forelimb. This increase in diameter of the forearm was noticed especially at the proximal portion and was due apparently to growth and development of the muscles that are located in this region. Sexual dimorphism in the surface area of wings was not detected, but some dimorphism in the length of the forearm was noted (Fig. 4). The slightly greater length of the forearm in females implies that perhaps females may be capable of supporting a slightly greater load in flight than males. At 18 to 21 days of age when young bats were capable of flight for the first time, the relationships of the surface areas of the wings to the total weights of the animals ( Table 2 ) perhaps were indicative of the optimal wing loading for flight in the species of bat considered herein. If this were true, an indication of maximum weight-carrying capacity of these bats could be obtained by comparing the wing loadings of bats capable of first flight with wing loadings of mature animals. It is of interest that none of the young bats were capable of flight unless the wing loading was less than 0.250 lb per sq ft (Table 2). Young bats may be capable of flight with the optimal wing loading for the species, but probably would lack the coordinations and skills of flight that were developed in the adults, thus more mature bats probably have the ability to carry extra weight at least for brief periods of time. The achievement of the relationship of surface area versus weight 264 18 JOURNAL OF MAMMALOGY Vol. 48, No. 1 was correlated closely with the relative growth rate of the forearm and fifth finger (Table 2 and Fig. 6). Wing loading of adult animals (73 and 97 days of age) varied from 0.1921 to 0.2261 lb per sq ft. On the basis of these afore- mentioned data, it can be suggested that, allowing for reasonable amounts of variation, adult animals with a weight of 9 to 11 g would have a wing loading not far removed from the postulated maximum of about 0.2500 lb per sq ft. According to Gates (1941), weight of two females prior to parturition was 11.6 g each. Some of the females observed in this study weighed as much as 14 g (the average was slightly more than 12 g) prior to parturition, but these animals were kept in confined situations. At the time of capture, 20 days before parturition, average weight of females was slightly more than 10 g (Fig. 8). Because weights of newborn bats recorded during this study were greater than weights of young bats given by Gates (1941), it is suggested that the young and females kept in captivity may have weighed more, due to overfeeding, than animals living in natural conditions. Bats have been reported to carry various weight loads in addition to the normal weight of the body (Davis and Cockrum, 1964). The weight carrying capacity of individuals of any given species of bats undoubtedly is important with regard to the relationships between adult females and the young, and may be reflected in the behavior of adults prior to and following parturition. As noted earlier in this report ( Table 1 ) , young Nycticeius humeralis remained associated closely with the parents for nearly 3 weeks. This relationship may be a reflection of the methods of housing the animals during these observa- tions. Gates ( 1941 ) suggested that the young remained with the adult for less than 10 days and he implied that this is a reflection of the weight carrying capacity of the adults. Hamilton (1943) mentioned that nursing female Nycticeius probably do not carry the young while foraging for food. The re- lationships of weights, surface areas of flight membranes, and related abilities of flight of the animals may be reflected in the selection of sites for roosting by adults both prior to and following birth of the young. Acknowledgments Sincere thanks are clue Dan Walton and Dr. Francis Rose for help in collecting the original material, Dr. Clyde Barbour for the preparation of photographs, and Glenn Clemmer for help in caring for the animals. This study could not have been conducted without the countless hours of help in feeding and caring for the animals that were con- tributed by Dr. Francis Rose and Charlene Jones. Dr. Andrew Arata photographed bats and made many helpful suggestions throughout this study. The study was supported in part by an American Cancer Society Grant to Tulane University (IN-24-G). Literature Cited AsDELL, S. 1964. Patterns of mammalian reproduction. Comstock Publishing Associates, New York, 2nd ed., 670 pp. Cockrum, E. 1955. Reproduction in North American bats. Trans. Kansas Acad. Sci., 58: 487-511. 265 February 1967 JONES— EVENING BAT 19 Davis, R., and E. Cockrum. 1964. Experimentally determined weight lifting capacity in individuals of five species of western bats. J. Mamm., 45: 643-644. Gates, W. 1941. A few notes on the evening bat, Nycticeius humeralis ( Raf inesque ) . J. Mamm., 22: 53-56. Hamilton, W. 1943. The mammals of eastern United States. Comstock Publishing Company, Inc., Ithaca, New York, 432 pp. Hayward, B., and R. Davis. 1964. Flight speeds in western bats. J. Mamm., 45: 236-242. Hooper, E. 1939. Notes on the sex ratio in Nycticeius humeralis. J. Mamm., 20: 369- 370. Orr, R. 1954. Natural history of the pallid bat, Antrozous pallidus (LeConte). Proc. California Acad. Sci., 28: 165-246. Pearson, O., M. Koford, and A. Pearson. 1952. Reproduction of the lump-nosed bat (Corynorhinus rafinesquei) in California. J. Mamm., 33: 273-320. Ryberg, O. 1947. Studies on bats and bat parasites. Svensk Natur, Stockholm, 319 pp. Struhsaker, T. 1961. Morphological factors regulating flight in bats. J. Mamm., 42: 152-159. Vaughan, T. 1959. Functional morphology of three bats: Eumops, Myotis, Macrotus. Univ. Kansas Publ, Mus. Nat. Hist., 12: 1-153. . 1966. Morphology and flight characteristics of molossid bats. J. Mamm., 47: 249-260. Department of Biology, Tulane University, New Orleans, Louisiana. Accepted 28 No- vember 1966. 266 Growth, 1961, 25, 127-139. A COMPARATIVE STUDY OF GROWTH AND DEVELOPMENT OF THE KANGAROO RATS, DIPODOMYS DESERTI STEPHENS AND DIPODOMYS MERRIAMI MEARNS Bernard B. Butterworth Department of Biology, University of Wichita, Wichita 8, Kansas During a recent study of sexual behavior and reproduction of the kangaroo rats, Dipodomys deserti and Dipodomys merriami, a com- parison of growth and development of closely related sympatric species reared under identical laboratory conditions was possible. Litters born in the laboratory were carefully examined and measured from birth to maturity. Although growth in other genera of mammals has been carefully studied, few instances of breeding in the genus Dipodomys have been recorded and limited growth data are available. The solitary nature of the animals prevents laboratory breeding under ordinary conditions. Animals confined in restricted space engage in fighting which usually results in the death of one or both of them. Chew and Butterworth (3) published an analysis of growth and development of Merriam's kanga- roo rat, D. merriami in which the senior author was successful in laboratory breedings of this species. The scattered literature on growth and development in the genus Dipodomys is summarized in their paper. Materials and Methods A total of 32 laboratory animals representing 8 D. deserti litters and 4 D. merriami litters that were the direct progeny of wild parents were used in this study. Seven of the 8 litters of D. deserti were the products of laboratory breedings, the first recorded instances for this particular species. The other litters were from pregnant females captured in the field and brought back to the laboratory for observation and which subsequently produced young. The parent animals were all obtained from one locality at the western edge of the Mojave Desert near the base of Alpine Butte located approximately 16 miles northeast of Palmdale in Los Angeles 127 267 128 GROWTH AND DEVELOPMENT OF KANGAROO RATS County, California. The altitude of the collecting area was approxi- mately 1000 meters (3261 feet). The captive animals were placed in two large breeding cages measuring 3 by 4 meters at the University of Southern California. A partition separated the two cages and sand and desert soil up to several inches in depth was spread on the floor. Animals were allowed to run in these cages without restriction. Nesting sites consisting of glass bottles, empty cardboard mailing tubes and cardboard boxes were pro- vided. One pair of each species was placed in each cage. Instances of breeding in D. dcscrti occurred in these cages. Animals were fed rolled oats and sunflower seeds. Lettuce was provided and water was also made available. An excess amount of food material was always available. All animals were measured from birth and the measurements were continued daily or at frequent intervals until adult sizes were attained. Standard measurements of total length, tail length, hind foot length, ear (from the notch) length, and body weight were taken. Measure- ments were analyzed as in Brody ( 2 ) and values were plotted on a semilogarithmic scale against age on the arithmetic scale. Linear seg- ments of such a plot indicate periods when growth increments are constant percentages of previous sizes. From these linear sections instantaneous growth rates were calculated as: T^ In m_> — In mi K = The value of K is the instantaneous percentage rate of growth for the unit of time in which tj and ti are expressed. Ln m^ and In mi are natural logarithms of the measurements made at ti and t2. Observations and Discussion General Development. The general pattern of development in D. deserti is similar to that described by Chew and Butterworth (3) for D. merriami. The desert kangaroo rat is born hairless and has a thin, pink, wrinkled, transparent integument. Viscera are apparent through the skin of the venter and sutures and blood vessels on the skull are visible. The snout area containing the vibrissa sheaths appear to be swollen and are richly vascular. The yellowish-brown vibrissae are about 6 millimeters long at birth, but lengthen to 13 mm. in 5 days and 268 BERNARD B. BUTTERWORTH 129 23 mm. by 15 days. The adult length of 72 mm. is attained by 90 days. Black pigmentation appears about 5 days after birth. The pigmenta- tion begins on the dorsum and top of the head. A faint dorsal tail stripe is present. The end of the tail is unpigmented for 10 mm. and then a black ring circles the entire tail for about 7 mm. The black area extends on the dorsal side of the tail for about 30 mm. and shades out to a pale gray color. The entire ventral surface of the tail is white except for the small black area near the tip. The feet and venter are white. The site of the dermal gland shows as a light depression just posterior to the scapulae. It darkens by 15 days, begins to lighten by 19 days and is covered with hair by 21 days. The head is sufficiently pigmented by 7 days so that the dorsal cranial sutures and blood vessels are no longer visible. By 21 days the mammae have become very distinct. The insides of the thighs and the venter, in general, are sparsely haired until about 15 days. By 21 days the animal is fully furred and the color pattern of the young is now similar to that of the adult. The young appear darker than the parents, however. The pelage of D. merriami is a darker yellowish-buff than that of D. deserti, the white tail stripe is wider than the dark tail stripe, and the terminal tuft is brown. Dark whisker patches are distinct in D. merriami but are absent in D. deserti. Juvenile D. merriami are com- pletely furred by 15 days. Developmental stages are shown in Figures 1 to 6. The pinnae of the ears, closed at birth, and only two mm. in length gradually open from 9 to 15 days. The length of time until opening varies with different litters. The ears are fully opened by 15 days in D. deserti and by 10 days in D. merriami. The nails are soft at birth and gradually become hard by 15 days in both species. The incisor teeth appear later in D. deserti than in D. merriami. They break through the gums at about 2 days in the latter ( Chew and Butterworth, 3 ) while they do not appear until about 9 days in the former. Teeth are white at first but gradually darken to yellow. By 25 days the teeth of D. deserti are strong enough to pierce the skin of man. Eye development is summarized in Figure 7 for various species of kangaroo rats. In my laboratory, individuals of D. deserti had their eyes open as follows: 3 on the 15th day; 3 on the 16th day and 6 on 269 130 GROWTH AND DEVELOPMENT OF KANGAROO RATS FIGURE 1 Dipodomys deserti, 3 days old. Note absence of hair and the relative sizes of the feet. Both eves and ears are tightly closed. FIGURE 2 Dipodoiuvs uien-iaini, a mother nursins her voung. FIGURE 3 Dipodomys deserti, 10 days old. A litter of 5. Note the white tail tip, one of the distinguishing features of this species. FIGURE 4 Dipodonivs uierriami, 11 davs old. A litter of 3. FIGURE 5 Dipodomvs deserti, 16 davs old. From the same litter as pictured in Figure 3. FIGURE 6 Dipodomys merriami, 16 days old 270 BERNARD B. BUTTERWORTH 131 the 17th day. D. merriami had their eyes open between the 11th and 15th days. Davs after birth Species that e\es open Reference Dipodomvs deserti 15-17 Butterworth D. deserti 16 Rush (8) D. merriami 11 Butterworth D. merriami 11-15 Chew and Butterworth (3) D. merriami 21 Do ran (5) D. merriami by third week Reynolds (7) D. heermanni 12-15 Fitch (6) D heermanni 14-16 Tappe (9) D. nit rat aides 13-14 Culbertson (4) D. spectabilis 14 Bailey (1) FIGURE 7 Eye development in various species of kangaroo rats. A comparison of general development in the two species in this study is shown in Figure 8. Eyes open Ears open Incisor teeth erupt Solid food eaten Solid feces first noted Nails harden Sand used for cleaning pelage Well haired Weaned Mammae first visible Drumming with feet first noted Days after birth D. deserti [ D. merriarni 11-17 11-15 9-15 8-10 7-10 2-8 15 13 15 17 12-15 12-15 17 13-15 11-15 14 15-25 17-22 21 11 35 FIGURE 8 Comparison of sequences of general development of Dipodomys deserti and Dipodomys merriami. Selected features based on 8 Utters of D. deserti and 4 litters of D. merriami. Growth Analyses: A comparison of weight increases in D. deserti and D. merriami is shown in Figures 9 and 10. Weights become constant at adult levels of approximately 145 grams in D. deserti and 40 grams in D. merriami. An analysis indicates that early growth is rapid in D. merriami and continues more slowly toward the maximum weight. At 10 days D. deserti had reached 16 per cent of its total adult weight while D. merriami had attained 26 per cent of its adult weight. At 30 days the two species had reached approximately half their maximum weight, 47 per cent in D. deserti and 53 per cent in D. merriami. At 271 132 GROWTH AND DEVELOPMENT OF KANGAROO RATS 15 days 20 days 30 days 50 days 90 days D.d. D.m. D.d. D.m. D.d. D.m. D.d. D.m. D.d. D.m. Total length 47 54 55 70 72 80 83 95 97 99 Tail length 39 47 50 61 73 81 91 97 93 99 Foot length 70 82 80 87 82 95 96 99 100 100 Ear length 63 63 67 82 78 90 81 95 88 99 Weight 27 35 33 43 47 53 75 64 91 78 FIGURE 9 Percentages of growth toward maturity completed at indicated intervals in D. deserti nd D. merriami. 50 days D. deserti had reached 75 per cent of its adult weight while D. merriami had only attained 64 per cent. At this point D. merriami lagged behind and then very slowly approached the maximum weight. At 90 days D. deserti had attained 91 per cent of its weight while in D. merriami only 78 per cent had been reached. In the later phases of growth D. merriami gained weight more slowly than D. deserti. Both attained full adult average weights by 150 to 180 days. In total length ( Figures 9 and 1 1 ) Z>. merriami grew more rapidly than D. deserti throughout the growth period. Maximum adult lengths are difficult to measure but approximate adult dimensions are attained by 90 days. The tail of D. merriami (Figures 9 and 11) grew faster than that of D. deserti during the developmental period. After 90 days D. merriami had little tail growth while D. deserti continued to increase its tail length by 7 per cent. The foot (Figures 9 and 12), which was already well developed at birth, grew fastest of all. At 15 days D. deserti had a foot 70 per cent of adult size while D. merriami had attained 82 per cent of the adult foot size. Both species attained maximum foot size between 50 and 90 days. The ear (Figures 9 and 12) grew rapidly in both species, that of D. merriami faster than that of D. deserti. The ear was approximately fully grown by 90 days in D. merriami but continued to grow slowly in D. deserti. Size of the litter (Figure 13) made little difference on increases in weight of D. deserti. Litters containing 2, 3, 4, and 5 individuals were compared and showed very little variation in weight changes. Apparently all individuals in the respective litters received sufficient food both by nursing and individually after weaning. Litters of D. merriami were not numerous enough to permit these observations. 272 (/I 1/1 3 3 O 0) cr C 4-^ o (-; o rrt ■■-1 ^ (/) C ^> •^^ £ (U cfl flJ 3 O rt > S ^ a. ■" .^^ (/) '«^ I- ."3 (u QJ O 2 = ^ > 1— I ^ ^1 TT — r— 1 — I — I — r o o o o o o 00 (o If) « SlAJVdO Nl 1H9I3M rt •C 13 S ^ 3 'S o .bi t« ej 1- »*■ cd O i> a j= ■*-» Cd QJ b£ C bc.ti c — 3 -a V. CM 3 — ■a .° j2 •" T3 C ^ , ^ u 'r '^ _C ^6 ■*~j flj 0^ j= So >- 6f rt « V. t« is 1=^^ Ct-I ■- &£• o — aj c °i$ o 0^ *k. 5^ rj a s <« E s ^ o !n-c u °-"? < (U o J= t- d M- 273 134 GROWTH AND DEVELOPMENT OF KANGAROO RATS 80-1 SO- SO 20 — O z 10 9 R I 1- 7 o 6 z LU 5 _l 4 3 — 2 — TOTAL LENGTH ^^ =.00436 ^j^ K =.00208 INCREASE IN LINEAR DIMENSIONS X P. DESERTI 0 D. MERRIAMI J I I L 1_J L_± 0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 AGE IN DAYS FIGURE 11 A composite graph showing increase in linear dimensions of total length and tail length in D. deserli and D. merriami, semilogarithmic plot. 274 BERNARD B. BUTTERWORTH 135 80-1 50 30- 20- X 10- I- 9 LiJ 7- _l 6. 5- 4- 3- 2- FOOT - K'. 00150 K = .00044 EAR - INCREASE IN LINEAR DIMENSIONS y p. DESERTI o D. MERRIAMI J_ _L _L _L _L _L 0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 AGE IN DAYS FIGURE 12 A composite graph showing increase in linear dimensions of foot length and ear length of D. deserti and D. merriami, semilogarithmic plot. 275 LU |sj fvj to * in Cr (T (T ct fK uj UJ UJ UJ , , t- I- »- »- LU K t- K H r- o o CVJ 1 o m 1 1 1 1 1 o o o o o O 0> 00 t- (0 — 1 — o in § I o ro O CM rO O CJ — -. O o t/1 ro c O £ iO 0) O C E * 3 CO C/) CO > § N < -4-' o (\j Q a h- ro.ii ^ -- e (0 05 -C o LU O o (0 O * < ID (t) ^ ^ CVJ * c (0 fO 8 CM CO -I £! 3 1/1 §: -4-) Cj o O SIAIVdO Nl 1H9I3M 276 BERNARD B. BUTTERWORTH 137 The differential rates of growth during development in the two species of kangaroo rats is significant. The smaller D. merriami matures more rapidly than the larger D. deserti. The development of various body parts in growing juveniles of D. merriami demonstrates more rapid growth and an earlier maturity than D. deserti. The precocious growth, especially during early development, brings D. merriami to an earlier seasonal reproductive potential than D. deserti and may allow for an earlier dispersion. Rapid development may be one contributing factor for the larger geographical range of Merriam's kangaroo rat. Summary and Conclusions The growth and development of two sympatric species of kangaroo rats is discussed, utilizing data from 8 litters of D. deserti and 4 litters of D. merriami reared under identical laboratory conditions. General development and sequences of hair acquisition are described. Animals of both species were completely furred by 15 days. Their pelages were darker than those of their parents. The ears, which were closed at birth, opened in 15 days in D. deserti and in 10 days in D. merriami. The nails hardened by 15 days in both species and the teeth erupted after 5 days in D. merriami and in 8 days in D. deserti. The teeth were white at first but gradually darkened to a shade of yellow. Eyes opened at about 16 days in D. deserti and from 11 to 15 days in D. merriami. Instantaneous growth rates were calculated for various measure- ments, such as weight, total length, tail length, foot length, and ear length. Both species of kangaroo rats attained average adult weights from 150 to 180 days. Each species reached approximately one-half of its adult weight by 30 days. Early growth was rapid in D. merriami but continued more slowly toward maximum weight. D. deserti at- tained maximum adult weight at a slightly earlier age than D. merriami. In total length, D. merriami grew more rapidly than D. deserti through- out the growth period. Both grew to approximately adult lengths by 90 days. The tail of D. merriami grew faster than that of D. deserti during development. After 90 days D. merriami had little tail growth while the tail of D. deserti continued to increase slowly in length. The foot showed the fastest development. By 15 days D. deserti had 70 per cent of its adult size while D. merriami had attained 82 per cent of its adult size. Both species attained maximum size between 50 and 277 138 GROWTH AND DEVELOPMENT OF KANGAROO RATS 90 days. The ear of D. merriami grew slightly faster than that of D. deserti. Ears were almost completely grown to adult size in D. merriami by 90 days. They continued to grow slowly in D. deserti. Size of litter made no appreciable difference in growth rates. The precocious early development of D. merriami may partly explain its larger geographical range. Acknowledgments Special appreciation is extended to the University of Southern California Graduate School which readily furnished equipment and facilities for carrying on the necessary laboratory work connected with this research and to the Committee for Research in Problems of Sex, National Academy of Sciences — National Research Council, which provided, in part, the financial help needed to complete the study. References 1. Bailey, V. 1931. Mammals of New Mexico. North American Fauna, 53, 412 pp. 2. Brody, S. 1945. Time relations of growth of individuals and populations. Chapter 16:487-574 in Bioenergetics and growth. New York: Reinhold. 1023 pp. 3. Chew, R. M., & Butterworth, B. B. 1959. Growth and development of Merriam's kangaroo rat, Dipodomys merriami. Growth, 23, 75-95. 4. CuLBERTSON, A. E. 1946. Observations on the natural history of the Fresno kangaroo rat. Jour. Mamm., 27, 189-203. 5. DoRAN, D. J. 1952, Observations on the young of the Merriam kangaroo rat. Jour. Mamm., 33, 494-495. 6. Fitch, H. S. 1948. Habits and economic relationships of the Tulare kangaroo rat. Jour. Mamm., 29, 5-35. 7. Reynolds, H. G. 1956. The ecology of the Merriam kangaroo rat {Dipodomys merriami Mearns) on the grazing lands of Southern Arizona. Ecol. Monographs, 28, 111-127. 8. Rush, W. M. 1945. Beau Brummel of the wild. Nat. Hist., New York, 54, 40-41. 9. Tappe, D. T. 1941. Natural history of the Tulare kangaroo rat. Jour. Mamm., 22, 117-148. 278 Article IX.— CRANIAL VARIATIONS IN NEOTOMA MICROPUS DUE TO GROWTH AND INDIVIDUAL DIFFERENTIATION. By J. A. Allen. Plate IV. In view of the stress naturally, and very properly, laid upon the importance of cranial characters in the discrimination of species in groups of closely-allied forms, it seems desirable to ascertain the character and amount of change in not only the general form of the skull but in the form of its separate bones due to growth, and also to determine the amount and kind of individual variation that may be expected to occur in skulls unquestionably of the same species. Having of late had occasion to examine a large amount of material relating to the genus Neotoma^ the subject has been forcibly brought to my attention, and some of the results of a careful examination of a large series of skulls pertaining to several species of this genus are here presented. No attempt is made to treat the subject exhaustively, only a few special points being here presented. As is well known to all experienced workers in mammalogy, tlie general contour of the brain-case, the relative size and form of individual bones, notably the interparietal, and the condition of the sujjraorbital and other ridges for muscular attachment, alter materially after the animal reaches sexual maturity ; the deposition of osseus matter, the closing of sutures, the building out of crests and rugosities continuing throughout life, so that a skull (){ a very old animal may differ notably from that of an indi- vidual of the same species in middle life, and this latter from one just reaching sexual maturity. The Museum has at present a large series of specimens of Neoionia niicropus Baird, including ages ranging from nursling young to very old adults. They are mainly from three localities in the eastern coast district of Texas, namely, Brownsville, Cor[)us Christi, and Rockport. In order to avoid any complications that [233] 279 2,34 Bulletin American Museum of Natural History. [Vol. VI, might arise through geographic variation, only the specimens from Rockjjort and Corpus Christi — localities less than twenty-five miles apart, and similar in physical conditions — are here consid- ered. There is not the slightest reason for questioning their con- specific relationshi|). The series selected to illustrate variations due to age are, with one exception, from Rockport ; those figured to show individual variation are all from Corpus Christi. Variations due to Age. General Contour. — The variation in the general form of the skull resulting from growth is due mainly to the lengthening of the several skull segments without a corresponding relative in- crease in the breadth of the skull. Hence in the young skull, in comparison with an adult skull of the same species, the brain- case is disproportionately large in comparison with the anteor- bital and basal portions of the skull. This is well shown in Plate IV, and in the subjoined table of measurements of three Measurements and Ratios showing Cranial Variations due to Age IN Neotoma tnicropus. Occipito-nasal length Length of nasals Length of frontais Length of parietals on median line. . . . (Ireatest length of parietals Length of interparietal Length of brain-case (Ireatest rostral breadth Least interorbital breadth . . , Breadth of brain-case Breadth of interparietal Greatest zygomatic breadth Depth of skull at middle of palate Depth of skull at front of basisphenoid. Length of tooth-row (crown surface). . . . Length of incisive foramina Width of incisive foramina Length of palatal floor No. 5834, 9 JUV. 1 Ratioi No. 4480, .', juv. 31 100 41 10 32.3 14.5 13 42 15 5 19.4 6 12 39 15 4.5 14.5 5.5 14 45.2 17 5.5 17.7 6.3 6 1 ;» . 4 6 16 51. H 19.5 11 35.5 10 20? ti4.B 23 8 2rt 11 11 35.5 12 8- 26.8 8 6 19.3 8.5 3 9.7 3 5 16.1 7 No. Ratio" 4478, ,' very old. 100 53 35.4 22 36.6 18 14.6 8 36.6 16 13.4 7 41.5 21 15.4 6.5 14.6 6 45 20 24.4 7.5 56.1 30 26 29 . 3 19.5 20.7 7.3 17 15 14 9 11. 3. 7 Ratioi 100 41.5 34 15 30.2 13.2 39. 12. 11 38 14.2 56.6 28.5 26.4 17 21.7 6.6 13.2 .6 .3 .3 ' Ratio to occipito-nasal length. 2 From No. 448?, 9 juv., in which the last molar has Just come into use. 280 1 894-] Allen, Cranial Variations in JVeo/oma micropus. 235 specimens of A', micropus from Rockport, Texas. No. 5834, $ juv., is a nursling so young that the last molar is still wholly enclosed in the jaw ;' No. 4480, ^, jnv., though not (]uite full- grown, would pass as a 'young adult'; No. 4478, r? ad., is a very old male, with the teeth well worn down, and the fangs visible at the alveolar border. Other specimens in the series furnish a complete series of gradations between the two extremes (Nos. 5834 and 447^)- In general contour (Figs, i-ii, PI. IV), the young skull, in comparison with adults, is much more convex in dorsal outline,' very broad posteriorly, and very narrow anteriorly. In compar- ing the relative length of the several skull segments the occipito- nasal length is taken as the basis, and the skulls will be referred to as A ( = No. 5834), B ( = No. 4480), and C( = No. 4478). Rostral Segment. — In A the ratio of the rostral segment to the total length is 32.3 per cent. ; in B, 35.4; in C, 41.5 — giving a rapid increase in the ratio with age. Frontal Segment. — In A the ratio of the frontal segment — /. e., the distance between the naso-frontal and fronto-parietal sutures — to the total length is 42 per cent. ; in B, 36.6 ; in C, 34 — a considerable decrease in the ratio with age. Parietal Segment. — In A the ratio of the parietal segment — /. (?., the distance from the latero-anteri. In each case the change in ratio is due to the disproportionate growth of the rostral portion of the skull. Thus in A the nasals have a length of only 10 mm. ; in B they have increased to 14.5 mm., and in C to 22 mm., while the total occipito-nasal length of ' The length of the tooth-row given in the table is taken from an older specimen (No. 4482, ^ juv.), in which the last molar has reached the level of the others and is just beginning to show traces of wear. '•^ In Figs. 10 and 11 it should be noted that the greater flatness of the skull interorbitally, as compared with Fig. 6, is masked by the raised supraorbital borders in the older skulls wiien viewed in profile. 281 236 Bulletin American Museum of Natural History. [Vol. VI, the skull has increased only from 31 mm. in A to 53 mm. in C. In other words, the nasal bones have increased in length 120 per cent., while the total length has increased only 77 per cent. Transverse Breadth. — In respect to the breadth of the skull the variations with growth are much less than in its length. Thus tlie greatest diameter of the rostrum varies only from 5.5 mm. in A to 6.5 in C — an increase of about 20 per cent, in the breadth of the rostrum, against an increase of 120 per cent, in its length. The interorbital breadth remains nearly constant, being 6 mm. in all three of the skulls here compared. The width of the brain-case shows an increase of 25 per cent, against an increase in the total length of the skull of 77 per cent. The zygomatic breadth shows an increase of about 50 per cent., due almost wholly to the thickening and increased convexity of the zygomatic arches. Vertical Depth. — In respect to the depth of the skull, the vari- ations with age prove especially interesting, although only such as would be expected from the facts already given. For present purposes the depth of the skull is taken at two points, namely, {a) at the middle of the palatal region, and {b) at the posterior border of the basisphenoid (basisphenoid-basioccipital suture). The palatal depth increases markedly with age, correlalively with the growth of the rostrum ; the basisphenoidal depth changes but slightly after the molars have attained to functional development. Thus in A the basisphenoidal depth is 11 mm. ; in /^, 12 mm. ; in C, 14 mm. — an increase of about 28 per cent. The palatal depth in ^4 is 8 mm. ; in B, 11 mm. ; in C\ 15 mm. — an increase of nearly 88 per cent. Tooth-row. — The length of the upper tooth-row varies about 12 per cent., due almost wholly to the wearing down of the teeth, the length of the crown surface being much less, in slightly worn teeth, than the length taken at the alveolar border. Interparietal. — The interparietal shows sur[jrising modilication with age, both as to size and form, but especially in respect to the latter. At early stages, as in ^, this bone is more or less crescentic in shape, with the transverse diameter more than twice 282 1894-] Allen, Cranial Variations in Neoto7na micropus. 237 the antero-posterior diameter. Thus in A the two diameters are respectively 11 and 4.5 mm. ; in B, 10 and 5.5 mm. ; in 6\ 7.5 and 7 mm. In other words, the short, broad, convex sub-cres- centic interparietal in A becomes transformed in C into a squarish, flat bone in which the two diameters are nearly equal, instead of the transverse being twice as great as the antero- posterior, as in A. This would be almost incredible were not the proof so abundantly furnished by the material in hand, where every stage of transition is shown. (Figs. i-S, PI. IV.) This change is coincident with the development of the raised supra- orbital borders and their prolongation backward as ridges to the parieto-occipital suture, and the flattening of the whole dorsal aspect of the post-rostral portion of the skull. In old age these ridges become confluent with the lateral edges of the interparietal which has now lost its postero-lateral moieties, partly apparently by absorption and partly by their being overgrown by the mediad posterior angle of the parietals. A sharp thin ridge for muscular attachment also extends back from the posterior base of the zygomatic arch. The interparietal at the same time develops a more or less prominent median angular projection at its posterior border, confluent with the median ridge of the supraoccipital. The contrast between these conditions, obtaining only in very old skulls, and their almost entire absence in skulls which have iust reached sexual maturity, is strikingly great. Supraoccipital. — The supraoccipital changes from a posteriorly convex, thin lamina of bone, in early life, to a thick, nearly ver- tical plate, with a strongly-developed median ridge produced into an angular spine at its superior border, and with a lateral ridge on either side about midway between the median line and its lateral borders ; these lateral ridges also each develop an angular rugosity or process about midway their length. The superior border is also produced into an incipient occipital crest. Basioccipital. — The basioccipital becomes greatly altered by growth, as in fact is the case with the whole postpalatal region. In comparing stages A and C it is found that the distance across the occipital condyles increases only about 15 per cent., while the breadth of the anterior border increases 100 per cent, and the length about 50 per cent. (Figs. 12-14, P^- ^^ ■) 283 230 Bulletin American Museum of Natural History. [Vol. VI, Basisphenoid. — The basisphenoid doubles in length, and its anterior third becomes differentiated into a narrow projecting neck. The presphenoid at stage A is nearly hidden by the palatal floor. (Figs. 12-14, PI. IV.) Postpalatal Region as a whole. — This doubles its length with an increase in breadth of only about 50 per cent. At stage A the postpalatal border terminates slightly behind the posterior edge of M.2 ; in stage 3 it holds very nearly the same position. The distance between the postpalatal border and the front border of the auditory bullae, compared with the total length of the skull, is as I to 9 in ^, and as i to 5 in C. In A the pterygoid hamuli reach the second fourth of the bullae ; in C they terminate slightly in advance of the bullae. The bullae themselves in A are more obliquely placed than in C, in relation to the axis of the skull, and are quite differently shaped. Also the form of the foramen magnum has undergone much change. These points are all well shown in Figs. 12-14 of ^l^e accompanying plate. Incisive Foramina. — Consequent upon the growth of the rostral portion of the skull, the incisive foramina undergo marked change in form, and somewhat in position, as regards both their anterior and posterior borders. In the stage designated as A they are short and broad, and extend relatively further both anteriorly and posteriorly than in stage B or C, their anterior border being nearer the base of the incisors, and their posterior border being carried back to or slightly behind the front border of the first molar. Thus in A the length of the incisive foramina is 6 mm., with a maximum breadth of 3 mm., while in C the dimensions are respectively 11.5 and 3.5 mm. — a great increase in length with only slight increase in breadth. At the same time the anterior border is considerably further from the base of the incisors, and the posterior border is slightly in advance, instead of slightly behind, the front border of the molars. Sphe7io-palatine Vacuities. — Va adults of N'eotoma micropus^ as in other species of the ' round-tailed ' section of the genus, there is a long, broad vacuity on each side of the presphenoid and ante- rior third of the basisphenoid, which Dr. Merriam has recently 284 1 894-] Allen, Cranial Vai'iations in Neotoma micropus. 239 named' the ^ spheno-palatine vacuilies,' and he has also called atten- tion to the fact that they are not present in some forms of the * bushy-tailed ' section of the genus. It is therefore of interest in the present connection to note that these vacuities are absent at stage A, and are only partially developed at later stages (Figs. 12-14, PI. IV). My attention was called to the matter by finding several nearly fully-grown skulls from Texas and northeastern Mexico with these vacuities either quite absent or represented by an exceedingly narrow slit, while I could find no differences in the skins or in other cranial characters that gave the slightest hint that the animals were not referable to JV. micropus. Further examination of young skulls of undoubted JV. micropus from Rockport and Corpus Christi, Texas, showed that the closed condition was in this species a feature of juvenility. It is thus of interest to find that a feature which proves to be merely a char- acter of immaturity (and quite inconstant as well) in IV. micropus is a permanent condition in N. cinerea occidentalis: In the development of these vacuities it appears that as the presphenoid increases in length it becomes reduced in width ; at the same time, as the skull broadens, the edges of the ascend- ing wings of the palatine bones become slightly incised. There is, however, much individual variation in this respect, as will be shown later. Molars. — When the molars first cut the gum they have nearly the entire crown-surface capped with enamel. Very soon, even before the tooth has attained its full height, the enamel begins to disappear from the centers of the enamel loops, the capping re- maining longer over the narrower loops than over the broader ones ; it quickly disappears from all as soon as the crown-surface becomes subject to wear. In stage ^, in which only M.i and M.2 have appeared, and are less than one-third grown, the enamel walls of the loops nearly meet over the dentinal areas — quite meeting over the narrower portions, especially in the case of the middle transverse loop of each tooth. Some time before the age represented by B is reached, the crown-surface is worn to an 1 Proc. Biol. Soc. Wash., VII I, p. 112, July, 1893. 2 Unfortunately the outline figures here given (Figs. 12-15, PI- IV, ) fail to show clearly the points at issue. 285 240 Bulletin American Museum of Natural History. [Vol. VI, even plane ; the tooth has reached its normal length, but the fluting of the sides still extends to the alveolar border. As attrition goes on, with the advance of the animal in age, the crown-surface wears down, and the neck of the tooth appears above the alveolar border, till, especially in the upper molars, the fluted terminal and the smooth basal portions are of nearly equal extent ; but in old age (as in C) the smooth basal portion is the longer and the division of the root into fangs is clearly shown. With this wearing down the tooth increases somewhat in both width and length, but the pattern of the enamel folds undergoes but slight change until nearly the whole crown is worn away, exce|:)t that the angles become gradually more rounded. Resume. — As already stated the change with age in the general form of the skull is due to the relatively disproportionate increase in length of the pre- over the post-orbital region, and the same disproportionate increase of the basal region as compared with the frontoparietal elements. In the first case the rostrum be- comes relatively greatly produced ; in the second the basiocci- pital and adjoining parts become so greatly enlarged as to change the entire aspect of the basal region of the skull. Thus the occipital condyles, which in A terminate slightly in advance of the most convex portion of the supraoccipital, and are crowded u]) very close to the bullae, form in C the most posterior part of the skull, with a considerable interval between them and the bulla;. (Figs. 12-14, PI- IV.) Individual Variation. In comparing a large series of skulls of the same species it quickly becomes apparent that no element of even the adult skull is constant, either as to form or relative size. There is also much variation in the size of skulls of the same sex and approxi- mately the same age. Variation in .5'/sr.— Thus in Neotoma micropiis, from the same locality, there are dwarfs and giants. While the females average smaller than the males, size is by no means a safe criterion of sex. Thus two old females, not appreciably different in age, from Corpus Christi, Texas, vary as follows : No. 2948, total 286 1 894-] Alleti, Crania/ Variations in Ncotoma micropus. 24 1 length 51 mm., zygomatic breadth 26 mm. ; the corresponding dimensions in No. 2955 are 45 mm. and 24 mm. These are merely the extremes of a series of six specimens ; with a much larger series doubtless the difference would be considerably increased. A series of six old males, from the same locality and indistinguishable as to age, vary as follows: No. 2952, total length 50.5 mm., zygomatic breadth 27 mm. ; the corresponding dimensions in No. 2956 are 45 mm. and 25 mm. Nasals and ascending branches of the Premaxillce. — Ordinarily in N. micropus the nasals terminate in a gradually narrowed evenly rounded point, a little less than 2 mm. in front of the posterior termination of the ascending branches of the premaxills. The distance between the points of termination of the nasals and premaxillse, however, frequently varies between 1.5 and 2.5 mm. ; more rarely from i to 3 mm. These extremes each occur in the ratio of about 10 per cent, of the whole, while probably 60 per cent, would not vary much from the normal average of about 2 mm. (See Figs. 1-8 and t6, 17, PI. IV.) The nasals, as already said, usually terminate in an evenly rounded point, but in several of the 50 skulls of N. micropus before me their posterior border forms a double point, each nasal terminating in a distinctly rounded point ; in one or two the posterior border is squarely truncate ; in others it is irregularly uneven. The ascending branches of the premaxillae usually terminate in an obtusely V-shaped point, with a uniformly even outline, their breadth, however, being subject to variation ; in some specimens they terminate in a brush of irregular spiculae. (Figs. 1-8 and 16, 17, PI. IV.) Frontals. — The posterior border of the frontals is subject to great irregularity, varying from a nearly transverse line (rounded slightly at the outer corners) to a gentle, rather even convexity, and thence to an acute angle, involving the whole posterior border. It is difficult to decide what outline is the most frequent, though the tendency seems to be greatest toward a well-pro- nounced rather even convexity. Figures 1-8 and 18, 19, Plate V, well show the variation in the position and direction of the fronto-parietal suture. \_Scptenibcr , /Sq/.] 287 242 Bulletin American Museum of Natural History. [Vol. Vl, Parietals. — The anterior outline of the parietals of course con- forms to the posterior outline of the frontals, and must be equally variable. It hence follows that their length on the median line is also variable. Their posterior border is also subject to much variation in consequence of the great diversity in the form of the interparietal. Intej'parietal. — In middle-aged specimens the interparietal tends strongly to a quadrate form, varying from quadrate to diamond shape, through a more or less marked median angular extension of both its anterior and posterior borders, and occasionally of its lateral borders as well. Often it forms a quadrate figure, in which each of its four sides is slightly convex ; again the corners are so much rounded, and the lateral breadth so much in excess of the antero-posterior, as to give a lozenge-shaped figure. In other cases it is distinctly shield-shaped ; in others it is hexagonal. In size the variation is fully 50 per cent, of what may be regarded as the average dimensions. These remarks have strict reference to fully adult specimens, and as nearly as can be judged these variations are not at all due to differences of age, which, as already shown, has so great an influence upon the size and form of this exceedingly variable element of the skull.' (Figs. 20-23, PI. IV. Compare also the interparietal, as shown in Figs. 1-8.) Ventral aspect. — The ventral aspect of the skull presents numerous points of variability, only a few of which will be here mentioned. The palate varies more or less in breadth, and especially in the development of the anterior palatal spine, which is sometimes slight, and sometimes so strongly produced anteri- orly as to touch the vomer. The postpalatal border may be evenly concave, or present a slight median process. The pre- sphenoid is very variable in size, being often an exceedingly slender rod of bone, and at other times very stout, the variation in thickness being nearly or quite 100 per cent. The anterior third of the basisphenoid shares in the same variability. As the ' As regards variation with age in the form of the interparietal, Neotoma jnicropus is only an example of what doubtless prevails throughout the genus, and even in many other genera as well. Yet in adult animals the form of this bone seems, as a rule, to be sufficiently constant to be of more or less taxonomic value. Thus in the A'^. cinerea group it may be said to be nor- mally quadrate ; in the N .fuscipes group it is quite constantly shield-shaped. In N. floridana., however, and in the N. mexicana group, it seems to be nearly or quite as variable as in N. jiiicropus, both as to size and shape. 288 1 894-] Allen, Cranial Variations in Neotoma ?nicropus. 243 ascending borders of the palatals are also variable in respect to the extent of their development, it follows that there is, even among adults, a wide range of variation in the size of the spheno- palatine vacuities. Teeth. — Aside from differences due to age and attrition, the teeth vary in size to a considerable extent among individuals strictly comparable as to sex and age, some having a much heavier dental armature than others. But more particularly note- worthy in this connection is the variation in the color of the teeth, which seems strongly a matter of individuality. Although Dr. Merriam has recently placed JV. micropus in his " Neotoma leucodon group,'" which has, among other alleged characters, " color of teeth white or nearly white," the teeth in N. micropus average blacker than in any other species of the genus known to me. Were this all it might be considered that N. micropus was erroneously referred to X^^t' leucodon group'; but unfortunately the range of individual variation in the color of the teeth in the large series at hand covers also the whole range of variation for the genus. Thus in some instances the molar teeth are intensely black from base to crown, while the crown-surface itself is strongly blackish, even the enamel loops, as well as the enclosed dentine being tinged with blackish ; in other cases the teeth are merely slightly tinged with brownish near the base and at the bottom of the sulci. These extremes are connected by a series of very gradual intergradations. In other words, among hun- dreds of skulls of Neotoma, those with the blackest teeth occur in N. micropus, as well as those in which the teeth are practically white. In the suckling young the teeth are pure white ; before M.3 has come to wear, M.i and M.2 have become more or less blackened ; in young adults, and in middle aged specimens, the teeth are often intensely black ; in old specimens, with the teeth much worn, the teeth average lighter than in the younger indi- viduals. There is, however, a wide range of variation in the color of the teeth in specimens of corresponding age, whether old or young. The black coloring consists to a large extent of a » Proc. Biol. Soc. Wash., IX, p. ii8, July 2, 1894. 289 244 Bulletin American Museum of Natural History. [Vol. VI, superficial incrustation which tends to scale off in flakes in the prepared skull, and its absence apparently may be due sometimes to removal in the process of cleaning the skull for the cabinet. In other words, the blackness is to some extent an accidental or pathological condition, due probably more or less to the particu- lar character of the food or to the health of the animal. General Remarks. The bearing of what has been stated above respecting varia- tions in the form of the skull and of its principal elements due to age is of course obvious, the inference being that in animals which have reached sexual maturity variations due wholly to growth, in passing through adolescence to senility, may readily be mistaken, when working with very small series or with single specimens, for differences of subspecific or even specific import- ance. Not only do the individual bones vary in their outlines and proportions and in relative size, but the skull varies as a whole in its relative dimensions, including depth as well as length and breadth. There is beside this a wide range of purely indi- vidual variation, affecting every character that can be used in a diagnostic sense. Thus in a series of fifty skulls of Neotoma micropus it would be easy to select extremes, of even individual variation, that depart so widely from the average, in one or more characters, as to deceive even an expert, on considering these alone, into the belief that they must represent very distinct species ; yet in the present instance the proof that such is not the case is overwhelming. In N. micropus the coloration is re- markably constant, for a member of this genus, at all seasons and ages, so that the case is less complicated than it would be in many other species of the group, where the color of the pelage varies radically with season and age. Personal criticism is not the purpose of the present paper, and it was not my intention at the outset to refer specifically to the work of any of my confreres. Since its preparation was begun, however, its raison d'etre has perhaps been emphasized by the pub- lication of two brochures of ' preliminary descriptions ' of species and subspecies of the genus Neotoma.^ numbering altogether lo species and 8 subspecies, which added to the 22 species and sub- 290 1 894-] Allen, Cranial Variations in Neotonia nilcropus. 245 species previously standing practically unchallenged, makes, at the present writing, a total of 40 forms of the genus Neotoma. Of these no less than 26 have been described within the last nine months.' Without the material before me used by the original describers of these forms it would be presumptive to give an opinion respecting the merits of many of them. While the greater part may have some real basis, it is evident that others are almost unquestionably synonyms of previously-described forms, judging by ' topotypes ' in this Museum, the brief diagnoses accompanying the names affording in these cases no characters that are in the least degree distinctive. The genus Neotoma was chosen for treatment in this connec- tion in preference to some other almost solely by chance, as the facts of variation above presented are not at all exceptional. In fact the common muskrat {Fiber zlbethlcus) would have shown a still more striking case of variability, as would also various species of many other genera. Yet describers of new species are con- stantly laying stress upon cranial differences that have not neces- sarily the slightest s])ecific or even subspecific importance ; and, so far as can be judged from their descriptions, they are entirely unconscious that such can be the case. On the other hand, it is equally certain that such alleged characters may have the value assigned them ; since it is now a well known fact that the extremes of purely individual variation in any character, external or internal, may exceed in amount the average difTerences that serve to satisfactorily distinguish not only well-marked subspecies, but even forms that are unques- tionably specifically distinct. Hence it must often happen that the determination of the status of a species or subspecies origin- ally described from one or two specimens, in groups especially susceptible to variation, must depend upon the subsequent exam- ination of a large amount of material bearing upon this and its closely-related forms. ' For a list of the species and subspecies of Neotovta described prior to July 6, 1894, see Abstr. Free. Linn. Soc. New York, No. 6, pp. 34, 35, July, 1894. 291 246 Bulletin American Museum of Natural History. [Vol. VI,] EXPLANATION OF PLATE IV. Figures all Natural size. Neotoma micropus Baird. Showing cranial variations due to age and individualism. (Unless otherwise stated, the specimens are from Rockport, Texas.) Figs. 1-8. Dorsal aspect of skull, showing gradual change in form with age, and especially in the form and relative size of the interparietal. Fig. i. No. 5834, S juv. (suckling). Fig. 2, No. 2975, $ juv. (nearly sexually adult). Corpus Christi, Texas. Fig. 3, No. 5841, ? ad. Fig. 4, No. 4480, f, ad. Fig. 5, No. 2958, c? ad^., Corpus Christi. Fig. 6, No. 4479, c5 ad. Fig. 7, No. 4477, ? ad. Fig. 8, No. 4478, 6 ad. Figs. 9-1 1. vSkuU in profile, to show change of form with growth. Fig. 9, No. 5834, S juv. (nursling). Fig. 10, No. 44S0, _<; ad. (rather young). Fig. II, No. 4478, (5 ad. (very old). Figs. 12-15. Ventral aspect, showing variations in postpalatal region due to age. Fig. 12, No. 5834, ? juv. (nursling). Fig. 13, No. 5841, ? ad. (young- adult). Y\g. 14, No. 2958, Corpus Christi, (5 ad. (very old). Fig. 15, No. 1456, Neotoma cinerea occidentalis, (?ad.. Ducks, B. C. (for comparison with N. iincroptis). Figs. 16, 17. To show extremes of individual variation in relative posterior extension of nasals and ascending branches of premaxillie. Locality, Corpus Christi. Texas. Fig. 16, No. 2958, ,^ ad. Fig. 17, No. 2948, $ ad. Figs. 18, 19. To show extremes of individual variation in posterior border of frontals. Locality, Corpus Christi, Texas. Fig. 18, No. 2949, ,5 ad. Fig. 19, No. 2951, .^ ad. Figs. 20-23. To show individual variation in the size and form of the inter- parietal. Specimens all from Corpus Christi, Texas. Fig. 20, No. 2949, f, ad. Fig. 21, No. 2948, $ ad. Fig. 22, No. 2952, .5 ad. Fig. 23, No. 2945, S ad. Note. — If the Brownsville, Texas, series of specimens had also been included, the range of individual variation would have been considerably increased. 292 HiLi.. A. M. X. H. Vol.. VI., Pl. IV. Neotoma micropus. Fiarures nat. size. 293 MATURATIONAL AND SEASONAL MOLTS IN THE GOLDEN MOUSE, OCHROTOMYS NUTTALLI Donald W. Linzey and Alicia V. Linzey Abstract. — The adult pelage of the golden mouse {Ochrotomys mittalli) is attained by a single maturational molt. Data on the post-juvenile molt were obtained from 96 young golden mice. This molt began on the ventral surface and spread dorsally, meeting in the dorsal midline. It then proceeded anteriorly and posteriorly. The average age at which male golden mice began molting was 36 days, whereas that of females was 38 days. The average duration of molt for the sexes was 29 days and 25 days, respectively. Golden mice undergo two seasonal molts — spring and fall. Data were obtained from 36 mice. Tlie winter pelage was generally much darker than the summer pelage. Both spring and fall molts were more irregular than the post-juvenile molt, and the spring molt tended to be more irregular than the fall molt. Young golden mice born after 1 October and 8 April appeared to combine the post-juvenile and seasonal molt. Hair replace- ment was more irregular than during the normal post-juvenile molt. During the course of a study on the ecology and Hfe history of the golden mouse, Ochrotomys mittalli nuttaUi, in the Great Smoky Mountains National Park (Linzey, 1966), considerable data were obtained on pelage changes. The limited data presented by Layne (1960) have been the only published infomiation concerning molt in this species. Maturational Molt The adult pelage of the golden mouse is attained after a single matura- tional molt. Data on the post-juvenile molt were obtained from 96 young golden mice. Eighty-four of these mice were raised in captivity. Data from the remaining 12 individuals were obtained from field observations. The molt from the golden-brown juvenile pelage to the golden-orange adult pelage, although varying in details, followed a definite pattern ( Fig. 1 ) . The first indication of the beginning of the dorsal molt was the appearance of new golden fur along the line separating the golden-brown dorsal fur from the white fur of the ventral surface. The replacement of the juvenile pelage progressed dorsally on both sides and met on the dorsal midline forming a continuous band of new fur. The molt then proceeded anteriorly between the ears and onto the head, while posteriorly, it joined the molt proceeding dorsally near the thighs. By this time, new fur had appeared on the sides of the face and just anterior to the ears. The molt along the sides of the body had nearly been completed by this time. The last two areas in which the fur was replaced were the top of the head and the base of the tail. In some individuals, the new fur first appeared just in back of the front leg. It pro- ceeded both posteriorly and dorsally and formed a band of new fur just behind the ears. The molt proceeding posteriorly then covered the remainder of the body. This pattern of molt generally agrees with that described for Peromyscus 236 294 May 1967 LINZEY AND LINZEY— MOLT OF GOLDEN MOUSE 237 ^^^^.iirf^J^ Fig. L — Sequence of post-juvenile molt on the dorsum in Ochrotomys nuttalli. Shaded portions represent areas of active hair replacement. Stippled areas represent adult pelage. 295 238 JOURNAL OF MAMMALOCiY Vol. 48, No. 2 Taule 1. — Duration of post-juvenile molt and average age at beginning and ending of molt in 34 captive golden mice (Range of values in parentheses). Males (15) Females (19) Duration 29 days (14-45) 25 days (12-49) Beginning 36 days (33-42) 38 days (31-47) Ending 64 days (51-87) 63 days (51-84) tmei ( Hoffmeister, 1944), Feromyscus gossypinus (Pournelle, 1952) and Peromysctis boy lei (Brown, 1963). It differs from that reported for Pero- myscus leucopiis noveboracensis (Gottschang, 1956). Data on the begmning, ending, and duration of the post-juvenile molt on the dorsum in male and female golden mice are compared in Table 1. The average duration of molt for males was slightly longer than for females. The shortest time recorded was between 12 and 14 days, whereas the maximum time required was about 49 days. Approximately 3.5 weeks are required for most Peromysciis leucopus noveboracensis to attain their full adult coat according to Gottschang ( 1956 ) . He recorded a minimum duration of 12 days for captive individuals and 10 days for one wild mouse to undergo the com- plete molt; the maximum number of days required was about 36. In the field, animals undergoing various stages of maturational molt were recorded in June (1), July (1), August (2), and December (8). These mice were between 150 mm and 164 mm in total length (mean, 156 mm). In the captive population, male golden mice began molting when their total length was 149 mm, whereas females averaged 146 mm. At the completion of molt, their measurements averaged 163 mm and 160 mm, respectively. From these data, it appears that both wild and captive individuals molted at approxi- mately the same body size, although it is not known whether they were the same age. The youngest individuals in captivity to begin molting during the current study were 31 days of age. Layne (1960) recorded one young Ochrotomys molting at 31 days of age with the molt apparently being complete 10 days later. Molting was in progress in one four week old mouse, while in another of the same age, it had not yet begun (Layne, 1960). Collins (1918) reported that the transition from juvenile to post-juvenile pelage in Peromyscus usually began at 6 weeks and was completed about 8 weeks later. The earliest age at which Peromyscus leucopus noveboracensis began molting was 38 days (Gottschang, 1956). These were all males. The youngest female to begin molting was 40 days of age. Ninety-five per cent of his mice of both sexes started the pelage change between the ages of 40 and 50 days. Young Pero- myscus gossypinus began molting when they were between 34 and 40 days of age (Pournelle, 1952). Gottschang ( 1956 ) found that, in general, mice of the same sex in a single litter started molting simultaneously. However, in every case where a dif- ference did occur, he found that the males started to molt first. In the current 296 May 1967 LINZEY AND LINZEV— MOLT OF C;OLDE.\ MOUSE 239 study, the males in 13 out of 21 litters containing mice of both sexes began molting before the females, whereas the females began molting first in three litters. The initiation of molt was simultaneous in the remaining five litters. The progression of the ventral molt was studied in seven individuals (four males, three females). The white belly fur was dyed purple by the stain Nyanzol A (20 g per liter of water-hydrogen peroxide mixture in ratio of two to one) and replacement by new hairs was followed. The ventral molt began approximately 2-4 days before the dorsal molt. Hair replacement occurred first in the center of the belly and continued laterally, and then dorsally into the golden fur. Simultaneously, new hair appeared over the entire chest and abdomen. The last areas to acquire new pelage were the throat and the ventral bases of the hind limbs. The ventral molt was complete at about the time that the dorsal molt covered the entire back ( Fig. 1 ) , Seasonal Molt Mice of the genus Peromyscus are generally considered to undergo one annul adult molt in autumn (Collins, 1923). However, Osgood (1909) and Brown ( 1963 ) recorded two annual molts in Peromyscus melanotis and Pero- myscus boylei, respectively. Golden mice in the Great Smoky Mountains National Park apparently un- dergo two annual molts. These take place during the spring (April-June) and fall ( October-December ) . The difference between summer and winter pelage was clearly distinguishable with the unaided eye. The winter pelage was much darker than the usual summer pelage, especially on the mid-dorsum. Osgood (1909) noted that winter specimens of Peromyscus melanotis pos- sessed a paler colored pelage, whereas summer specimens were in a dark pelage. The fall molt of P. boylei was characterized by the replacement of a bright cinnamon-brown pelage by a more drab, brown winter pelage (Brown, 1963). Nineteen of 21 adult golden mice in captivity underwent a fall molt between October 20 and December 24. A total of 10 adult golden mice were observed in the wild between December 12-17. Six of these were molting; four already had the winter pelage. The fall molt appeared to be more irregular than the post-juvenile molt. In several animals, it began near the hind leg, covered the rump and then progressed anteriorly to the head. Replacement of the hair was completed first over the posterior half of the body. This separated the two remaining areas of molt — the base of the tail and the head. The replacement of fur at the base of the tail was completed shortly thereafter. The final area of molt was on the head between the ears, and this sometimes required several weeks for completion. This is in contrast to the post-juvenile molt, where the last area of molt in all of the animals was at the base of the tail. The spring molt must have occurred between 1 April and 15 June. All wild individuals observed between 26 March and 1 April 1964 still retained their 297 240 JOURNAL OF MAMMALOGY Vol. 48, No. 2 winter pelage. By 15 June, all adult golden mice had either already com- pleted their spring molt or were very near completion. Seventy-four per cent (23) of the adult individuals in the captive population molted during the spring. Of those molting, 83% (19) did so between 15 May and 30 June. As in the fall molt, the pattern was irregular. Hair replacement occurred in patches along the sides and across the shoulders, and a simultaneous molt of the entire dorsum took place in only five of 31 individuals ( 16% ) . In the cases where this molt was complete, it followed a more regular pattern, with hair replacement occurring last on the nape of the neck. Gottschang ( 1956 ) noted no difference in the onset, progress or length of time required for the pelage change between spring-, summer-, or fall-born litters of Pcromyscus leiicopus. During the current study, however, golden mice born after 1 October and 8 April appeared to combine the post-juvenile molt and seasonal molt. The process of hair replacement was more irregular than during the normal post-juvenile molt. The molt began at a point just behind the front legs, as in the regular post-juvenile molt. It then proceeded dorsally and posteriorly at approximately equal rates. During the combined fall molt (post- juvenile plus fall molt), the replacement of hair at the base of the tail was completed prior to the completion of molt on the head in all cases. In this respect, this combined molt was more similar to the regular seasonal molt than to the regular post-juvenile molt. Upon completion of this molt, the mice had acquired the typical dark winter pelage. However, during the combined spring molt (post-juvenile plus spring molt), hair replacement was completed last at either the tail or head regions. On the average, those animals born after 1 October began molt at a later age than did those animals bom earlier in the breeding season. Males in this group began molting at an average age of 37 days, whereas spring and summer-born males began at 35 days of age. Females born after 1 October began molting at an average age of 43 days, while females bom earlier in the season began molting at an average age of 37 days. Acknowledgments We thank Dr. W. Robert Eadie of Cornell University for his advice and criticism of the manuscript. We gratefully acknowledge the financial assistance provided by The Society of the Sigma Xi and the cooperation of the National Park Service. Literature Cited Brown, L. N. 1963. NLaturational and seasonal molts in Peromyscus boylei. Amer. Midland Nat., 70: 466-469. Collins, H. H. 1918. Studies of normal molt and of artificially induced regeneration of pelage in Peromyscus. J. Exp. Zool., 27: 73-99. . 1923. Studies of the pelage phases and nature of color variations in mice of the genus Peromyscus. J. Exp. Zool., 38: 45-107. Gottschang, J. L. 1956. Juvenile molt in Peromyscus leucopus noveboracensis. J. Mamm., 37: 516-520. HoFFMEiSTER, D. F. 1944. Phylogeny of the Nearctic cricetine rodents, with especial 298 May 1967 LINZEY AND LINZEY— MOLT OF GOLDEN MOUSE 241 attention to variation in Peromtjscus truei. Ph.D. thesis, Univ. Cahfornia, 406 pp. Layne, J. N. 1960. Tlu> growth and development of younp golden mice, Ochrotomijs mittalU. Quart. J. Fla. Acad. Sci., 23: 36-58. LI^fZEY, D. W. 1966. The life history, ecology and behavior of the golden mouse, Ochrotomijs n. nuttalli, in the Great Smoky Mountains National Park. Ph.D. thesis, Cornell Univ., 170 pp. Osgood, W. H. 1909. Revision of the mice of the American genus Peromyscus. N. Amer. Fauna, 28: 1-285. PouRNELLE, G. H. 1952. Reproduction and early post-natal development of the cotton mouse, Peromyscus gossypinus gossypinus. J. Mamm., 33: 1-20. Division of Biological Sciences, Cornell University, Ithaca, New York. Accepted 16 January 1967. 299 SECTION 4— ECOLOGY AND BEHAVIOR Ecology and behavior comprise amazingly varied, active, and expanding fields. Probably most current mammalogical publications relate to one or both of these disciplines. Ecology particularly is of special importance to man owing to his increasing awareness of, and concern for, his own environment and such problems as the need to regulate human populations and to reduce pollution of air and water. The papers selected here can suggest to the perspective reader some basic ecological principles that apply to man himself. A host of topics other than those we were able to include in our selection come to mind when the ecological literature is contemplated — topics such as food habits as learned from stomach contents or droppings, or small mammal populations as censused by various methods (one such method is the study of bones in pellets regurgitated by owls, which are very eflBcient "mouse traps" ) . Long term cycles in populations and daily cycles in activity have had their share of ecological work also, but lack of space precludes further discussion of these topics. A large and well-documented textbook on animal ecology is that by Allee et al. ( 1949 ) . Three books relating to animal populations and factors that may regulate them are by Lack (1954), Andrewartha and Birch (1954), and Wynne-Edwards (1962). Their views differ and are interesting; their exam- ples, however, are largely non-mammalian. The older term "natural history" is perhaps a broader concept than ecology, but the older naturalists were deeply committed to the types of studies that have come to be called ecological, as well as ethological (a word currently used for studies of behavior ) . In the latter context, Ewer's ( 1968 ) recent book entitled Ethology of Mammals is of note to the student interested in a general coverage of the field, and Maternal Behavior in Mammals (Rheingold, 1963) also is useful. Papers reproduced here illustrate concepts such as territoriality and home range (applied to mammals in the paper by Burt), relatively larger studies ( note the numbers of specimens mentioned in Frank's paper for example ) that provided a firm statistical base and sound quantitative results, and the applica- tion of experimental procedures (as in the manipulation of rats in city blocks reported by Davis and Christian or the tests run by McCarley in compart- mented cages). The application of newer techniques such as Pearson's traflBc counter for mouse runways, the squirrel radio described by Beal, and auto- matic recording equipment of various types, all have contributed to advances in ecology and ethology. The recent study by Estes and Goddard of the African wild dog will serve to remind the reader that careful observational methods such as were used so effectively by older field naturalists certainly have not been supplanted, but only expanded and supplemented. Recent field studies dealing with primates have relied heavily on good observational techniques. Schaller's (1963) book on the gorilla is a good example. Other recent workers have studied baboons, chimpanzees, langurs, and other primates in similar ways. A report by Struhsaker (1967), not here 301 reproduced, on vervet monkeys is a good example of a shorter paper on primate behavior in the field. The short paper, here included, by Miller, written more than 60 years ago, was based on limited data, but reflects a thoughtful and somehow modern way of looking at the problem of bat migration, about which, incidentally, little is known even today. Ecological problems may be approached at different levels of inclusiveness. For example, the relationships of all species of plants and animals in an entire community may be studied. Such a broad approach to entire ecosystems merges imperceptibly with problems concerning factors that limit distribu- tions, hence to ranges of species and faunal and zoogeographic problems. A short paper by L. R. Dice (1931), not included here, on the relation of mam- mahan distribution to vegetation types is a classic, for here he adopted the term "Biotic Province" for a major concept that he and others expanded in later American zoogeographic studies. Even an analysis of a few species such as Brown's study of six species of shrews has obvious zoogeographic relevance. At a less inclusive level the ecological relationships of a single species may be studied. This approach is called autecology as opposed to community or syn- ecological studies. If we restrict ourselves further to the environmental rela- tionships of individual animals, we find our studies, again by gradual stages, merge with those that are primarily physiological and behavioral. Physiologi- cal techniques also enter directly into the study of ecosystems when energy flow is considered, as often is the case in recent studies. Lyman's paper, repro- duced here, on hibernators relates to energy, its sources, and its dissipation. Two noteworthy contributions to the study of hibernation are by Lyman and Dawe (1960) and Kayser (1961); these and other studies are summarized and cited in the textbook by Davis and Golley ( 1963 ) . Some of the more important journals that regularly publish contributions relating to ecology and behavior, and of which the serious student should be aware, are Animal Behavior, Behavior, Ecological Monographs, Ecology, Journal of Animal Ecology, and Zeitschrift fur Tierpsychologie. 302 MORTALITY PATTERNS IN MAMMALS Graeme Caugiiley Forest Research Instilute, New Zealand Forest Service, Rotorua, and Zoology Department, Canterbury University, New Zealand (Accepted for publication December 8, 1965) Abstract. Methods of obtaining life table data are outlined and the assumptions implicit in such treatment are defined. Most treatments assume a stationary age distribution, but published methods of testing the stationary nature of a single distribution are invalid. Samples from natural populations tend to be biased in the young age classes and therefore, because it is least affected by bias, the mortality rate curve (q^) is the most efficient life table series for comparing the pattern of mortality with age in different populations. A life table and fecundity table are presented for females of the ungulate Heniitragiis jcmlahiens, based on a population sample that was first tested for bias. They give estimates of mean generation length as S.4 yr, annual mortality rate as 0.25, and mean life expectancy at birth as 3.5 yr. The life table for Hemitragus is compared with those of Ovis aries, O. dalli, man, Rattus norvegiciis, Microtns agrestis, and M. orcadensis to show that despite taxonomic and ecological differences the life tables have common characteristics. This suggests the hypotheses that most mammalian species have life tables of a common form, and that the pattern of age- specific mortality within species assumes an approximately constant form irrespective of the proximate causes of mortality. Introduction Most studies in population ecology include an attemin to determine mortality rates, and in many cases rates are given for each age class. This is no accident. Age-specific mortality rates are usually necessary for calculating reproductive values for each age class, the ages most susceptible to natural selection, the population's rate of in- crease, mean life expectancy at birth, mean gen- eration length, and the percentage of the popula- tion that dies each year. The importance of these statistics in the fields of game management, basic and applied ecology, and population genetics re- quires no elaboration. The pattern of changing mortality rates with age is best expressed in the form of a life table. These tables usually present the same information in a variety of ways : 1 ) Survivorship (/x) : this series gives the prob- ability at birth of an individual surviving to any age, X (/x as used here is identical with P^ of Leslie, Venables and Venables 1952). The ages 303 Autumn 1966 MORTALITY PATTERNS IN MAMMALS 907 are most conveniently spaced at regular intervals such that the values refer to survivorship at ages 0, 1, 2 etc. yr, months, or some other convenient interval. The probability at birth of living to birth is obviously unity, but this initial value in the series need not necessarily be set at 1 ; it is often convenient to multiply it by 1,000 and to increase proportionately the other values in the series. If this is done, survivorship can be redefined as the number of animals in a cohort of 1,000 (or any other number to which the initial value is raised ) that survived to each age x. In this way a ^/^ series is produced, where k is the constant by which all h values in the series are multiplied. 2) Mortality (c/x) : the fraction of a cohort that dies during the age interval x, x + 1 is designated dx- It can be defined in terms of the individual as the probability at birth of dying during the interval X, X + 1. As a means of eliminating decimal points the values are sometimes multiplied by a constant such that the sum of the d^ values equals 1,000. The values can be calculated from the h series by c'x = 'x — 'x + l 3) Mortality rate {q-s.) '■ the mortality rate q for the age interval x, x + 1 is termed ^x- It is calculated as the number of animals in the cohort that died during the interval x, x -f 1. divided by the number of animals alive at age x. This value is usually expressed as l,000(7x, the number of animals out of 1,000 alive at age x which died before x -f- 1- These are three ways of presenting age-specific mortality. Several other methods are available — e.g. survival rate {px), life expectancy (tx) and probability of death (Qx) — but these devices only present in a different way the information already contained in each of the three series previously defined. In this paper only the Ix, dx and qx series will be considered. Methods of Obtaining Mortality Data Life tables may be constructed from data col- lected in several ways. Direct methods : 1 ) Recording the ages at death of a large num- ber of animals born at the same time. The fre- quencies of ages at death form a kdx series. 2) Recording the number of animals in the original cohort still alive at various ages. The frequencies from a klx series. Approximate methods : 3) Recording the ages at death of animals marked at birth but whose births were not coeval. The frequencies form a kdx series. 4) Recording ages at death of a representative sample by ageing carcasses from a population that has assumed a stationary age distribution. Small fluctuations in density will not greatly affect the results if these fluctuations have an average wave length considerably shorter than the period over which the carcasses accumulated. The frequencies form a kdx series. 5 ) Recording a sample of ages at death from a population with a stationary age distribution, where the specimens were killed by a catastrophic event (avalanche, flood, etc.) that removed and fixed an unbiased sample of ages in a living popu- lation. In some circumstances (outlined later) the age frequencies can be treated as a klx series. 6) The census of ages in a living population, or a sample of it, where the population has assumed a stationary age distribution. Whether the speci- mens are obtained alive by trapping or are killed by unselective shooting, the resultant frequencies are a sample of ages in a living population and form a klx series in certain circumstances. Methods 1 to 3 are generally used in studies of small mammals while methods 4 to 6 are more commonly used for large mammals. Tests for Stationary Age Distribution Five methods have been suggested for deter- mining whether the age structure of a sample is consistent with its having been drawn from a stationary age distribution : a) Comparison of the "mean mortality rate," calculated from the age distribution of the sample, with the proportion represented by the first age class (Kurten 1953, p. 51). b) Comparison of the annual female fecundity of a female sample with the sample number multi- plied by the life expectancy at birth, the latter statistic being estimated from the age structure (Quick 1963, p. 210). c) Calculation of instantaneous birth rates and death rates, respectively, from a sample of the population's age distribution and a sample of ages at death (Hughes 1965). d ) Comparison of the age distribution with a prejudged notion of what a stationary age distri- bution should be like (Breakey 1963). e) Examination of the "/x" and "dx" series, calculated from the sampled age distribution, for evidence of a common trend (Quick 1963, p. 204). Methods a to c are tautological because they assume the sampled age distribution is either a klx or kdx series ; method d assumes the form of the life table, and e makes use of both assumptions. These ways of judging the stationary nature of 304 908 a population are invalid. But I intend something more general than the simple statement that these five methods do not test what they are supposed to test. Given no information other than a single age distribution, it is theoretically impossible to prove that the distribution is from a stationary population unless one begins from the assumption that the population's survival curve is of a par- ticular form. If such an assumption is made, the life table constructed from the age frequencies provides no more information than was contained in the original premise. Mortality Samples and Age Structure Samples Methods 4 to 6 for compiling life tables are valid only when the data are drawn from a sta- tionary age distribution. This distribution results when a population does not change in size and where the age structure of the population is con- stant with time. The concept has developed from demographic research on man and is useful for species which, like man, have no seasonally re- stricted period of births. Populations that have a restricted season of births present difficulties of treatment, some of which have been discussed by Leslie and Ranson (1940). Very few mammals breed at the same rate throughout the year, and the stationary age distribution must be redefined if it is to include seasonal breeders. For species with one restricted breeding season each year, a stationary population can be defined as one that does not vary either in numbers or age structure at successive points in time spaced at intervals of 1 yr. The stationary age distribution can then be defined for such popu- lations as the distribution of ages at a given time of the year. Thus there will be an infinite num- ber of different age distributions according to the time of census, other than in the exceptional case of a population having a constant rate of mor- tality throughout life. The distribution of ages in a stationary popula- tion forms a kl^; series only when all births for the year occur at an instant of time and the sample is taken at that instant. This is obviously impossible, but the situation is approximated when births occur over a small fraction of the year. If a popu- lation has a restricted season of births, the age structure can be sampled over this period and at the same time the number of live births produced by a hypothetical cohort can be calculated from the number of females either pregnant or suckling young. In this way a set of data closely approxi- mating a kl^ series can be obtained. GRAEME CAUGHLEY Ecology, Vol. 47, No. 6 If an age distribution is sampled halfway be- tween breeding seasons, it cannot be presented as a A;/x series with x represented as integral ages in years. With such a sample (making the usual assumptions of stability and lack of bias) neither /x nor dx can be established, but Qx values can be calculated for each age interval x -(- 3^, x + l^^. The age frequencies from a population with a con- tinuous rate of breeding are exactly analogous ; they do not form a kl^ series but can be treated as a series of the form ^(/x + /x + l)/2 This series does not allow calculation of l^ values from birth unless the mortality rate between birth and the midpoint of the first age interval is known. Because a sample consists of dead animals, its age frequencies do not necessarily form a mor- tality series. The kdx series is obtained only when the sample represents the frequencies of ages at death in a stationary population. Many published samples treated as if they formed a kdx series are not appropriate to this form of analysis. For instance, if the animals were obtained by shooting which was unselective with respect to age, the sample gives the age striicture of the living popu- lation at that time ; that the animals were killed to get these data is irrelevant. Hence unbiased shooting samples survivorship, not mortality, and an age structure so obtained can be treated as a klx series if all other necessary assumptions obtain. Similarly, groups of animals killed by avalanches, fires, or floods — catastrophic events that preserve a sample of the age frequencies of animals during life — do not provide information amenable to kdx treatment. A sample may include both Ix and dx compo- nents. For instance, it could consist of a number of dead animals, some of which have been unselec- tively shot, whereas the deaths of others are at- tributable to "natural" mortality. Or it could be formed by a herd of animals killed by an avalanche in an area where carcasses of animals that died "naturally" were also present. In both these cases dx and /x data are confounded and these hetero- geneous samples of ages at death can be treated neither as kdx nor klx series. Even if a sample of ages at death wc/e not heterogeneous in this sense, it might still give mis- leading information. If, for instance, carcasses attributable to "natural" mortality were collected only on the winter range of a population, the age frequencies of this sample would provide ages at death which reflected the mortality pattern during only part of the year. But the dx series gives the proportion of deaths over contiguous periods of 305 Autumn 1966 MORTALITY PATTERNS IN MAMMALS 909 the life span and must reflect all mortality during each of these periods. It has been stressed that the frequencies of ages in life or of ages at death provide Ufe-table in- formation only when they are drawn from a popu- lation with a stationary age distribution. This age distribution should not be confused with the stable distribution. When a population increases at a constant rate and where survivorship and fecun- dity rates are constant, the age distribution even- tually assumes a stable form (Lotka 1907 a, b; Sharpe and Lotka 1911 ). Slobodkin (1962, p. 49) gives a simple explanation as to why this is so. A stable age distribution does not form a kl^, series except when the rate of increase is zero, the season of births is restricted, and the sample is taken at this time. Hence the stationary age distribution is a special case of the stable age distribution. The Relative Usefulness of the /x, dx and gx series Most published life tables for wild mammals have been constructed either from age frequencies obtained by shooting to give a ^/x series, or by de- termining the ages at death of animals found dead, thereby producing a kd^ series. Unfortunately, both these methods are almost invariably subject to bias in that the frequency of the first-year class is not representative. Dead immature animals, especially those dying soon after birth, tend to decay faster than the adults, so that they are under- represented in the count of carcasses. The ratio of juveniles to adults in a shot sample is usually biased because the two age classes have different susceptibilities to hunting. With such a bias estab- lished or suspected, the life table is best presented in a form that minimizes this bias. An error in the frequency of the first age class results in dis- tortions of each Ix and dx value below it in the series, but q^ values are independent of frequencies in younger age classes. By definition, q is the ratio of those dying during an age interval to those alive at the beginning of the interval. At age y the value of q is given by Qy = dy/ly but dy ty I- y + 1 therefore Qy ^ (^y — ^y + i)/^y • Thus the value of ^y is not directly dependent on absolute values of /^ but on the differences between successive values. If the Ix series is calculated from age frequencies in which the initial frequency is inaccurate, each Ix value will be distorted. How- ever, the difference between any two, divided by the first, will remain constant irrespective of the magnitude of error above them in the series. Thus a qx value is independent of all but two survivor- ship age frequencies and can be calculated directly from these frequencies (fx) by qx = {fx — fx + i)/ix if the previously discussed conditions are met. The calculation of q from frequencies of ages at death is slightly more complex : by definition qy = dy/U- but therefore 00 y-i ly = 2dx — SOx x=o x=o 00 y-1 qy = dy/{i:dx — Zdx) x=o x=o = dy/Xdx: x=y but the frequencies of ages at death (fx) are them- 00 selves a kdx series and so Q'y=/'y/2/ x ■ x = y Thus the value of q at any age is independent of frequencies of the younger age classes. Although the calculated value of q for the first age class may be wrong, this error does not affect the qx values for the older age classes. The qx series has other advantages over the Ix and dx series for presenting the pattern of mor- tality with age. It shows rates of mortality di- rectly, whereas this rate is illustrated in a graph of the Ix series (the series most often used when comparing species) only by the slope of the curve. A Life Table for the Thar, Hemitragus jemlahicus The Himalayan thar is a hollow-horned ungu- late introduced into New Zealand in 1904 (Donne 1924) and which now occupies 2,000 miles^ of mountainous country in the South Island. Thar were liberated at Mount Cook and have since spread mostly north and south along the Southern Alps. They are still spreading at a rate of about 1.1 miles a year (Caughley 1%3) and so the popu- lations farthest from the point of liberation have been established only recently and have not yet had time to increase greatly in numbers. Closer to the site of liberation the density is higher (cor- related with the greater length of time that animals have been established ther^), and around the point of liberation itself there is evidence that the popu- lation has decreased (Anderson and Henderson 1961). 306 910 GRAEME CAUGHLEY Ecology, Vol. 47, No. 6 The growth rings on its horns are laid down in each winter of life other than the first (Caugh- ley 1965), thereby allowing the accurate ageing of specimens. An age structure was calculated from a sample of 623 females older than 1 yr shot in the Godley and Macaulay Valleys between Novem- ber 1963 and February 1964. Preliminary work on behavior indicates that there is very little dis- persal of females into or out of this region, both because the females have distinct home ranges and because there are few ice-free passes linking the valley heads. As these data illustrate problems presented by most mammals, and because the life table has not been published previously, the methods of treat- ment will be outlined in some detail. Is the population stationary f Although it is impossible to determine the sta- tionary nature of a population by examining the age structure of a single sample, even when rates of fecundity are known, in some circumstances a series of age structures will give the required in- formation. This fact is here utilized to investigate the stability of this population. The sample was taken about halfway between the point of liberation and the edge of the range. It is this region between increasing and decreasing populations where one would expect to find a stationary' population. The animals came into the Godley Valley from the southwest and presumably colonized this side of the valley before crossing the 2 miles of river bed to the northeast side. This pattern of establishment is deduced from that in the Rakaia Valley, at the present edge of the breeding range, where thar bred for at least 5 yr on the south side of the valley before colonizing the north side. Having colonized the northeast side of the Godley Valley, the thar would then cross the Sibald Range to enter the Macaulay Valley, which is a further 6 miles northeast. The sample can therefore be divided into three sub- samples corresponding to the different periods of time that the animals have been present in the three areas. A 10 X 3 contingency test for differ- ences between the three age distributions of fe- males 1 yr of age or older gave no indication 'that the three subpopulations differed in age structure (X=^ = 22.34; P = 0.2). This information can be interpreted in two ways : either the three subpopulations are neither increasing nor decreasing and hence are likely to have stationary age distributions, or the subpopu- lations could be increasing at the same rate, in which case they could have identical stable age distributions. The second alternative carries a Table I. Relative densities of thar in three zones Zone Number females autopsied Mean density inde.x" Standard error Godley Valley south Godley Valley north Macaulay Valley. 258 240 115 2.19 1.67 2.66 0.56 0.53 0 69 F2.56 for densities between valleys == 1.74, not significant 'Density indices were calculated as the number of females other than kids recorded a,s autopsied in a zone each day. divided by the number of shooters hunting in the zone on that day. corollary that the subpopulations would have dif- ferent densities because they have been increasing for differing periods of time. But an analysis of the three densities gives no indication that they differ (Table I). This result necessitates the rejection of the second alternative. The above evidence suggesting that the sample was drawn from a stationary age distribution is supported to some extent by observation. When I first passed through the area in 1957, I saw about as many thar per day as in 1963-64. J. A. Anderson, a man who has taken an interest in the thar of this region, writes that the numbers of thar in 1956 were about the same as in 1%4 (Anderson, pers. comm.). These are subjective evaluations and for that reason cannot by them- selves be given much weight, but they support in- dependent evidence that the population is station- ary or nearly so. Is the sample biased? A sample of the age structure of a population can be biased in several ways. The most obvious source of bias is behavioral or range differences between males and females. For instance, should males tend to occupy terrain which is more diffi- cult to hunt over than that used by females, they would be underrepresented in a sample obtained by hunting. During the summer thar range in three main kinds of groups : one consists of fe- males, juveniles and kids, a second consists of young males and the third of mature males. The task of sampling these three groupings in the same proportions as they occur throughout the area is complicated by their preferences for terrain that differs in slope, altitude and exposure. Conse- quently the attempt to take an unbiased sample of both males and females was abandoned and the hunting was directed towards sampling only the nanny-kid herds in an attempt to take a repre- sentative sample of females. The following analy- sis is restricted to females. Although bias attributable to differences in be- havior between sexes can be eliminated by the simple contrivance of ignoring one sex, some age 307 Autumn 1966 MORTALITY PATTERNS IN MAMMALS 911 classes of females may be more susceptible than no bias could be detected from a sample of this others to shooting. To test for such a difference, size, females other than kids were divided into two groups : those from herds in which some mem- bers were aware of the presence of the shooter before he fired, and those from herds which were undisturbed before shooting commenced. If any age group is particularly wary its members should occur more often in the "disturbed" category than is the case for other age groups. But a x" test (X^ = 7.28, df = 9, P = 0.6) revealed no signifi- cant difference between the age structures of the two categories. The sample was next divided into those females shot at ranges less than 200 yards and those shot out of this range. If animals in a given age class are more easily stalked than the others, they will tend to be shot at closer ranges. Alternatively, animals which present small targets may be under- represented in the sample of those shot at ranges over 200 yards. This is certainly true of kids, which are difficult to see, let alone to shoot, at ranges in excess of 200 yards. The kids have therefore not been included in the analysis be- cause their underrepresentation in the sample is an acknowledged fact, but for older females there is no difference between the age structures of the two groups divided by range which is not ex- plainable as sampling variation (x" = 9.68, df := 9. P^O.4). This is not to imply that no bias exists — the yearling class for instance could well be underrepresented beyond 200 yards — but that The taking of a completely representative sam- ple from a natural population of mammals is prob- ably a practical impossibility, and I make no claim that this sample of thar is free of bias, but as bias cannot be detected from the data, I assume it is slight. Construction of the life table The shooting yielded 623 females 1 yr old or older, aged by growth rings on the horns. As the sampling period spanned the season of births, a frequency for age 0 cannot be calculated directly from the number of kids shot because early in the period the majority had not been born. In any case, the percentage of kids in the sample is biased. The numbers of females at each age are shown in Table II, column 2. Although the ages are given only to integral years each class contains animals between ages x yr — "/^ month and x yr 4" 2j^ months. Variance owing to the spread of the kidding season is not included in this range, but the season has a standard deviation of only 15 days (Caughley 1965). Up to an age of 12 yr (beyond this age the values dropped below 5 and were not treated) the frequencies were smoothed according to the formula log y = 1.9673 -f 0.0246x — 0.01036 x^, where y is the frequency and x the age. The linear and quadratic terms significantly reduced Table II. Life table and fecundity table for the thar Hemitragus jemlahicus (females only) 1 Age in years X 2 Frequency in sample 3 Adjusted frequency 4 No. female live births per female at age x nil 5 1,000?. 6 1,000 d. 7 l,000gi 0 1 94 97 107 68 70 47 37 35 24 16 11 6 3 4 3 0 1 205" 95.83 94.43 88.69 79.41 67.81 55.20 42.85 31.71 22.37 15,04 9.64 5.90 0.000 0.005 0.135 0.440 0.420 0.465 0.425 0.460 0.485 0.500 0.500 1 ^0.470 J 0.350 1,000 467 461 433 387 331 269 209 155 109 73 47 29 533 6 28 46 56 62 60 54 46 36 26 18 533 13 2 61 3 106 4 145 5 187 6 223 7 258 8 297 9 330 10 356 11 382 12 13 14 15 16 17 ■Calculated from adjusted frequencies of females other than kids (column 3) and mx values (column 4). 308 Q12 GRAEME CAUGHLEY Ecology, Vol. 47, No. 6 AGE IN YEARS Fig. 1. Age frequencies, plotted on a logarithmic scale, of a sample of female thar, with a curve fitted to the values from ages 1 to 12 yr, and the mortality rate per 1,000 for each age interval of 1 yr (IfiOOq^^) plotted against the start of the interval. variance around the regression, but reduction by the addition of a cubic term was not significant at the 0.05 level. There are biological reasons for suspecting that the cubic term would have given a significant reduction of variance had the sample been larger, but for the purposes of this study its inclusion in the equation would add very little. The improved fit brought about by the quadratic term indicates that the rate of mortality increases with age. Whether the rate of this rate also increases, is left open. The computed curve closely fitted the observed data ( Fig. 1 ) and should greatly reduce the noise resulting from sampling variation, the differential effect on mortality of different seasons, and the minor heterogeneities which, although not detectable, are almost certain to be present. The equation is used to give ad- justed frequencies in Table II, column 3. The frequency of births can now be estimated from the observed mean number of female kids produced per female at each age. These are shown in column 4. They were calculated as the number of females at each age either carrying a foetus" or lactating, divided by the number of females of that age which were shot. These values were then halved because the sex ratio of late foetuses and kids did not differ significantly from 1:1 (93 S S : 97 9 9 ) . The method is open to a number of objections : it assumes that all kids were born alive, that all females neither pregnant nor lac- tating were barren for that season, and that twin- ning did not occur. The first assumption, if false, would give rise to a positive bias, and the second and third to a negative bias. However, the ratio of females older than 2 yr that were either preg- nant or lactating to those neither pregnant nor lactating did not differ significantly between the periods November to December and January to February (x' = 0.79, P = OA), suggesting that still births and mortality immediately after birth were not common enough to bias the calculation seriously. Errors are unlikely to be introduced by temporarily barren females suckling yearlings, because no female shot in November that was either barren (as judged by the state of the uterus) or pregnant was lactating. Errors resulting from the production of twins will be very small ; we found no evidence of twinning in this area. The products of each pair of values in columns 3 and 4 (Table II) were summed to give an esti- mate of the potential number of female kids pro- duced by the females in the sample. This value of 205 is entered at the head of column 3. The adjusted age frequencies in column 3 were each multiplied by 4.878 to give the 1,000/x survivor- ship values in column 5. The mortality series (column 6) and mortality-rate series (column 7) were calculated from these. Conclusions Figure 1 shows the mortality rate of females in this thar population up to an age of 12 yr. Had the sample been larger the graph could have been extended to an age of 17 yr or more, but this would have little practical value for the calcula- tion of population statistics because less than 3% of females in the population were older than 12 yr. The pattern of mortality with age can be di- vided into two parts — a juvenile phase charac- terized by a high rate of mortality, followed by a postjuvenile phase in which the rate of mortality is initially low but rises at an approximately con- stant rate with age. Table II gives both the k and m^ series, and these two sets of values provide most of the in- formation needed to describe the dynamics of the population. Assuming that these two series are accurate, the following statistics can be derived : generation length (i.e. mean lapse of time between a female's date of birth and the mean date of birth of her offspring), T: ^'^'"^^-^5.4yr; r = S/xWs mean rate of mortality for all age groups, g^: 'q^ — l/S/x = 0.25 per female per annum; life expectancy at birth, eo : ^0 = 2 /x — >^ = 3.5 yr. 309 Autumn 1966 MORTALITY PATTERNS IN MAMMALS 913 The last two statistics can also be expressed conveniently in terms of the mortality series by ^,= l/S(x+l)(f. and eo S (2x+l) d. The relationship of the two is given by ^, = 2/(2^0+1). Life Tables for Other Mammals The difficulty of comparing the mortality pat- terns of animals that differ greatly in life span can be readily appreciated. To solve this problem, Deevey (1947) proposed the percentage deviation from mean length of life as an appropriate scale, thereby allowing direct comparison of the life tables of, say, a mammal and an invertebrate. For such comparisons this scale is obviously useful, but for mammals where the greatest difference in mor- tality rates may be at the juvenile stage the scale often obscures similarities. By way of illustration, Figure 2 shows l.OOOgx curves for two model populations which differ only in the mortality rate of the first age class. When the values are graphed on a scale of per- centage deviation from mean length of life the close similarity of the two sets of data is no longer apparent. Thus the use of Deevey's scale for •/. DEVIATION FROM MEAN LENGTH OF LIFE eoo - X O 100 -50 0 .50 •100 ♦15 T ■ 1 » 1 1 ^ P - / / 1 1 / / 1 / - } / p - \ \'. *' P p - s. 1 1 1 1 f 1 AGE IN ^'-A^'; Fig. 2. The mortality rate per 1,000 for each year of life for two model populations that differ only in the de- gree of first-year mortality. These l.OOOq^ values are each graphed on two time scales : absolute age in years (continuous lines) and percentage deviation from mean life expectancy (broken lines). comparing mortality patterns in mammals might result in a loss rather than a gain of information. In this paper, absolute age has been retained as a scale in comparing life tables of different species, although this scale has its own limitations. Domestic sheep, Ovis aries. — Between 1954 and 1959, Hickey (19(30) recorded the ages at death of 83,113 females on selected farms in the North Island of New Zealand. He constructed a ^x table from age IJ^ yr by "dividing the number of deaths which have occurred in each year of age by the number 'exposed to risk' [of death] at the same age." An age interval of 1 yr was chosen and the age series V/z, 2]^, 3>4 etc. was used in preference to integral ages. The qx series conformed very closely to the regression : log g^ ^ 0.1 56x -f 0.24, enabling him in a subsequent paper (Hickey 1963) to present the interpolated q^ values at integral ages. He also calculated q for the first year of life from a knowledge of the number of lambs dying before 1 yr of age out of 85,309 (sexes pooled) born alive. These data probably provide the most accurate life table for any mammal. The 1,000^^ curve is graphed in Figure 3. 800 - I 1 T ■ I ■ y ■ ■ ? 600 - / ■ 400 - ^ ■ 200 \ 1 - 0 0 2 4 6 8 10 12 AGE IN YEARS Fig. 3. Domestic sheep : mortaHty rate per 1,000 for each age interval of 1 yr (l,000qj, plotted against the start of the interval. Data from Hickey (1963). Dall sheep, Ovis dalli. — During his study on the wolves of Mount McKinley National Park, Murie (1944) aged carcasses of dall sheep he found dead, their ages at death being established from the growth rings on the horns. This sample can be divided into those that died before 1937 and those that died between 1937 and 1941. The former sample was used by Deevey (1947) to construct the life table presented in his classic paper on mortality in natural populations. Kur- ten (1953) constructed a life table from the same 310 914 GRAEME CAUGHLEY Ecology, Vol. 47, Mo. 6 data, but corrected the iinderrepresentation of first-year animals resulting from the relatively greater perishability of their skulls by assuming that adult females produce 1 lamb per annum from about their second birthday. Taber and Dasmann (1957) constructed life tables for both males and females from the sample of animals dying between 1937 and 1941, and adjusted both the 0 to 1- and 1 to 2-year age frequencies on the assumption that a female produces her first lamb at about her third birthday and another lamb each year thereafter, that the sex ratio at birth is unity and that the loss of yearlings is not more than 10%. 1000 1 ■ ■ I I I 1 ■ aoo -■ / / eoo - // /. 400 A ?? - / / - 200 ^="=5=^=0- — "^ , 1 i^ XV \ T^^^^^dL^ .^^ -^ ^ ^^^■^^^^^^$^^$S^S^S^:^^$^^^^^^^ J, — ^y \\^o^$5oc^:SovN^?^^ ( ^ ^$$^^^^ \ { o ^^^ ^^^^^^ ""^N ^ XV ^ \ ;$$^ / \ X ^ M' • // X^^^^O /\_ . ^■~- '■'^ ^v '"~°^^^^ ) ° "" ) O f^^^^^^V^ \ ^^/""^ • O i ^"--— >^ V-^ ' y y HOME RANGE BOUNDARY 1^^:^^ NEUTRAL AREA TERRITORIAL BOUNDARY • NESTING SITE BLANK— UNOCCUPIED SPACE O REFUGE SITE Fig. 1. Theoretical quadrat with six occupants of the same species and sex, showing territory and home range concepts as presented in text. follows: "I have seen a large beaver house built in a small island, that had near a dozen houses under one roof; and, two or three of these only excepted, none of them had any communication with each other but by water. As there were beavers enough to inhabit each apartment, it is more than probable that each famOy knew its own, and always entered at their own door without having any 326 348 JOURNAL OF 1VL\MMAL0GY further connection with their neighbors than a friendly intercourse" (in Morgan, 1868, pp. 308-309). Morgan (op. cit., pp. 134-135), also writing of the beaver, made the following observation; "a beaver family consists of a male and female, and their offspring of the first and second years, or, more properly, under two years old. . . . When the first litter attains the age of two years, and in the third summer after their birth, they are sent out from the parent lodge." Mor- gan's observation was later confirmed by Bradt (1938). The works of Seton are replete wath instances in the lives of different animals that indicate territorial behavior. In the introduction to his "Lives" Seton (1909) states "In the idea of a home region is the germ of territorial rights." Heape (1931) devotes an entire chapter to "territory." Although he uses the term more loosely than I propose to, (he includes home ranges of individuals and feeding ranges of tribes or colonies of animals), he carries through his work the idea of defense of an area either by an individual or a group of individuals. Not only this, but he draws heavily on the literature in various fields to support his thesis. Al- though the evidence set forth by Seton, Heape, and other early naturalists is of a general nature, mostly garnered from reports by others, it cannot be brushed aside in a casual manner. The old time naturalists were good observers, and, even though their techniques were not as refined as those of present day biolo- gists, there is much truth in what they wrote. A few fairly recent published observations on specific mammals serve to strengthen many of the general statements made by earlier workers. In speak- ing of the red squirrel (Tamiasciurus), Klugh (1927, p. 28) writes; "The sense of ownership seems to be well developed. Both of the squirrels which have made the maple in my garden their headquarters apparently regarded this tree as their private property, and drove away other squirrels which came into it. It is quite likely that in this case it was not the tree, but the stores that were arranged about it, which they were defending." Clarke (1939) made similar observations on the same species. In raising wild mice of the genus Peromyscus in the laboratory, Dice (1929, p. 124) found that "when mice are placed together for mating or to conserve cage space it sometimes happens that fighting takes place, especially at first, and sometimes a mouse is killed. . . . Nearly always the mouse at home in the cage will attack the presumed intruder." Further on he states, "However, when the young are first born, the male, or any other female in the same cage, is driven out of the nest by the mother, who fiercely protects her young." Similarly, Grange (1932, pp. 4-5) noted that snowshoe hares (Lepus americanus)- in captivity "showed a definite partiality for certain spots and corners to which they became accustomed" and that "the female would not allow the male in her territory (cage) during late pregnancy and the males themselves were quarrelsome during the breeding season." Errington (1939) has found what he terms "intraspecific strife" in wild musk- rats (Ondatra). Much fighting takes place when marshes become overcrowded, especially in fall and winter during readjustment of populations. "But when invader meets resident in the tunnel system of one of [the] last lodges to be used in a dry marsh, confhct may be indeed savage." Gordon (1936) observed def- 327 BURT TERRITORIALITY AND HOME RANGE 349 inite territories in the western red squirrels (Tamiasciurus fremonti and T. douglasii) during their food gathering activities. He also performed a neat experiment with marked golden mantled squirrels {Citellus lateralis chysodeirus) by placing an abundance of food at the home of a female. This food supply- attracted others of the same species. To quote Gordon: "she did her best to drive away the others. Some of her sallies were only short, but others were long and tortuous. There were rather definite limits, usually not more than 100 feet from the pile, beyond which she would not extend her pursuit. In spite of the vigor and the number of her chases (one day she made nearly 60 in about 6 hours) she never succeeded in keeping the other animals away." This individual was overpowered by numbers, but, nevertheless, she was using all her strength to defend her own log pile. To my knowledge, this is the best observation to have been published on territorial behavior in mammals. I have observed a similar situation (Burt, 1940, p. 45) in the east- ern chipmunk (Tamias). An old female was watched fairly closely during two summers. Having marked her, I was certain of her identity. "Although other chipmunks often invaded her territory, she invariably drove them away [if she happened to be present at the time]. Her protected area was about fifty yards in radius ; beyond this fifty-yard limit around her nesting site she was not concerned. Her foraging range (i.e., home range) was considerably greater than the protected area (territory) and occasionally extended 100 or more yards from her nest site." From live trapping experiments, plotting the positions of capture of individuals on a map of the area covered, I in- terpreted (op. cit., p. 28) the results to mean that there was territorial be- havior in the white-footed mouse (Peromyscus leucopus), a nocturnal form. When the ranges of the various individuals were plotted on a map, I found that "the area of each of the breeding females is separate — that although areas sometimes adjoin one another, they seldom overlap." Carpenter (1942) writes thus: "The organized groups of every type of monkey or ape which has been adequately observed in its native habitat, have been found to possess territories and to defend these ranges from all other groups of the same species." In reporting on his work on the meadow vole {Microtus pennsylvanicus) , Blair (1940, pp. 154-155) made the statement "It seems evident that there is some factor that tends to make the females occupy ranges that are in part exclusive ; .... Possibly there is an antagonism between the females, particularly during the breeding season, but the available evidence does not indicate to me that they have definite territories which they defend against all trespassers. It seems highly probable that most mammalian females attempt to drive away intruders from the close vicinity of their nests containing young, hut this does not constitute territoriality in the sense that the term has been used by Howard (1920), Nice (1937), and others m reference to the breeding territories of birds." (Ital. mme.) To quote Howard (1920, pp. 192-193): "But the Guillemot is generally surrounded by other Guillemots, and the birds are often so densely packed along the ledges that there is scarcely standing room, so it seems, for all of them. Nevertheless the isolation of the individual is, in a sense, just as 328 350 JOURNAL OF MAMMALOGY complete as that of the individual Bunting, for each one is just as vigilant in resisting intrusion upon its few square feet as the Bunting is in guarding its many square yards, so that the evidence seems to show that that part of the inherited nature which is the basis of the territory is much the same in both species." Blair, in a later paper (1942, p. 31), writing of Peromyscus manicu- latus gracilis, states: "The calculated home ranges of all sex and age classes broadly overlapped one another. Thus there was no occupation of exclusive home ranges by breeding females. . . . That individual woodland deer-mice are highly tolerant of one another is indicated by the foregoing discussion of overlapping home ranges of all sex and age classes." Reporting on an extensive field study of the opossum, Lay (1942, p. 149) states that "The ranges of indi- vidual opossums overlapped so frequently that no discernible tendency towards establishment of individual territories could be detected. On the contrary, tracks rarely showed that two or more opossums traveled together." It seems quite evident that both Blair and Lay are considering the home range as syno- nymous with the territory when in fact they are two quite distinct concepts. Further, there is no concrete evidence in either of the above papers for or against territoriality in the species they studied. It is to be expected that the territory of each and every individual will be trespassed sooner or later regardless of how vigilant the occupant of that territory might be. It is not intended here to give a complete list of works on territorial behavior. The bibliographies in the works cited above lead to a great mass of literature on the subject. The point I wish to emphasice is that nearly all who have critically studied the behavior of w^ild mammals have found this behavioristic trait inherent in the species with which they worked. Also, it should be stressed, there are two fundamental types of territoriality hi mammals — one concerns breeding and rearing of young, the other food and shelter. These tw^o may be further subdivided to fit special cases. Mrs. Nice (1941) gives six major types of territories for birds. Our knowledge of territoriality in mammals is yet too limited, it seems to me, to build an elaborate classification of types. Some day we may catch up with the ornithologists. HOME RANGE The home range concept is, in my opinion, entirely different from, although associated with, the territoriality concept. The two terms have been used so loosely, as synonyms in many instances, that I propose to dwell briefly on them here. My latest Webster's dictionary (published in 1938), although satisfac- tory in most respects, does not list "home range," so I find no help there. Seton (1909) used the term extensively in his "Lives" where he explains it as follows: "No wild animal roams at random over the country: each has a home region, even if it has not an actual home. The size of this home region corresponds somewhat with the size of the animal. Flesh-eaters as a class have a larger home region than herb-eaters." I believe Seton was thinking of the adult animal when he wrote the above. We know that young adolescent animals often do a bit of wandering in search of a home region. During this time they do not have a home, nor, as I consider it, a home range. It is only after they 329 BURT TERRITORIALITY AND HOME RANGE 351 establish themselves, normally for the remainder of their lives, unless disturbed,, that one can rightfully speak of the home range. Even then I would restrict the home range to that area traversed by the individual in its normal activities of food gathering, mating, and caring for young. Occasional sallies outside the area, perhaps exploratory in nature, should not be considered as in part of the home range. The home range need not cover the same area during the life of the individual. Often animals will move from one area to another, thereby abandoning the old home range and setting up a new one. Migratory animals have different home ranges in summer and winter — the migratory route is not considered part of the home range of the animal. The size of the home range may vary with sex, possibly age, and season. Population density also may influence the size of the home range and cause it to coincide more closely with the size of the territory. Home ranges of different individuals may, and do, overlap. This area of overlap is neutral range and does not constitute part of the more restricted territory of animals possessing this attribute. Home ranges are rarely, if ever, in convenient geometric designs. Many home ranges prob- ably are somewhat ameboid in outline, and to connect the outlying points gives a false impression of the actual area covered. Not only that, it may indicate a larger range than really exists. A calculated home range based on trapping records, therefore, is no more than a convenient index to size. Overlapping of home ranges, based on these calculated areas, thus may at times be exaggerated. From trapping records alone, territory may be indicated, if concentrations of points of capture segregate out, but it cannot be demonstrated without question. If the occupant of an area is in a trap, it is not in a position to defend that area. It is only by direct observation that one can be absolutely certain of terri- toriality. Home range then is the area, usually around a home site, over which the animal normally travels in search of food. Territory is the protected part of the home range, be it the entire home range or only the nest. Every kind of mammal may be said to have a home range, stationary or shifting. Only those that protect some part of the home range, by fighting or agressive gestures, from others of their kind, during some phase of their lives, may be said to have territories. SIGNIFICANCE OF BEHAVIORISTIC STUDIES I think it will be evident that more critical studies in the behavior of wild ani- mals are needed. We are now spending thousands of dollars each year in an attempt to manage some of our wild creatures, especially game species. How can we manage any species until we know its fundamental behavior pattern? What good is there in releasing a thousand animals in an area large enough to support but fifty? Each animal must have so much living room in addition to other essentials of life. The amount of living room may vary somewhat, but for a given species it probably is within certain definable limits. This has all been said before by eminent students of wildlife, but many of us learn only by repetition. May this serve to drive the point home once more. 330 352 JOURNAL OF MAMMALOGY LITERATURE CITED Blair, W. F. 1940. Home ranges and populations of the meadow vole in southern Michi- gan. Jour. Wildlife Management, vol. 4, pp. 149-161, 1 fig. 1942. Size of home range and notes on the life history of the woodland deer- mouse and eastern chipmunk in northern Michigan. Jour. Mamm., vol. 23, pp. 27-36, 1 fig. Bradt, G. W. 1938. A study of beaver colonies in Michigan. Jour. Mamm., vol. 19, pp. 139-162. Burt, W. H. 1940. Territorial behavior and populations of some small mammals in southern Michigan. Miscl. Publ. Mus. Zool. Univ. Michigan, no. 45, pp. 1-58, 2 pis., 8 figs., 2 maps. Carpenter, C. R. 1942. Societies of monkeys and apes. Biological Symposia, Lan- caster: The Jaques Cattell Press, vol. 8, pp. 177-204. Clarke, C. H. D. 1939. Some notes on hoarding and territorial behavior of the red squirrel Sciurus hudsonicus (Erxleben). Canadian Field Nat., vol. 53, no. 3, pp. 42-43. Dice, L. R. 1929. A new laboratory cage for small mammals, with notes on methods of rearing Peromyscus. Jour. Mamm., vol. 10, pp. 116-124, 2 figs. Errington, p. L. 1939. Reactions of muskrat populations to drought. Ecology, vol. 20, pp. 168-186. Evans, L. T. 1938. Cuban field studies on territoriality of the lizard, Anolis sagrei. Jour. Comp. Psych., vol. 25, pp. 97-125, 10 figs. Gordon, K. 1936. Territorial behavior and social dominance among Sciuridae. Jour. Mamm., vol. 17, pp. 171-172. Grange, W. B. 1932. Observations on the snowshoe hare, Lepus americanus phaeonotus Allen. Jour. Mamm., vol. 13, pp. 1-19, 2 pis. Heape, W. 1931. Emigration, migration and nomadism. Cambridge: W. Heffer and Son Ltd., pp. xii + 369. Hearne, S. 1795. A journey from Prince of Wale's fort in Hudson's Bay, to the Northern Ocean. London: A. Strahan and T. Cadell, pp. xliv + 458, illustr. Howard, H.E. 1920. Territory in bird life. London: John Murray, pp. xii + 308, illustr. Klugh, a. B. 1927. Ecology of the red squirrel. Jour. Mamm., vol. 8, pp. 1-32, 5 pis. Lay, D. W. 1942. Ecology of the opossum in eastern Texas. Jour. Mamm., vol. 23, pp. 147-159, 3 figs. Morgan, L. H. 1868. The American beaver and his works. Philadelphia: J. B. Lippin- cott and Co., pp. xv + 330, illustr. Nice, M. M. 1941. The role of territory in bird life. Amer. Midi. Nat., vol. 26, pp. 441-487. Noble, G. K. 1939. The role of dominance in the life of birds. Auk, vol. 56, pp. 263-273. Seton, E. T. 1909. Life-histories of northern animals. An account of the mammals of Manitoba. New York City: Charles Scribner's Sons, vol. 1, pp. xxx + 673, illustr., vol. 2, pp. xii + 677-1267, illustr. 1929. Lives of game animals, Doubleday, Doran and Co., Inc., 4 vols., illustr. Museum of Zoology, Ann Arbor, Michigan. 331 [Rejninted from SCIENCE, N. S., Vol. V., No. 118, Pages 541-543, April 2, i597.] MIGRATION OF BATS ON CAPE COD, MASSA- CHUSETTS. Bat migration has received little atten- tion. Various writers have made vague reference to the fact that certain bats are found in winter at localities where they are not known to breed, but no detailed ac- count of the migratory movements of any species has yet been published. The only special paper on the subject that I have seen is by Dr. C. Hart Merriam,* who clearly establishes the fact that two North American bats migrate. The data on which this conclusion, rests are as follows : The hoary bat, one of the migratory species, is not known to breed south of the Cana- dian fauna. In the Adirondack region it appears about the middle of M'ay and dis- appears early in October. During the autumn and winter it has been taken in South Carolina (Georgetown, January 19th), Georgia (Savannah, February 6th), and on the Bermudasf (' autumn '). As the writer remarks, these facts may be fairly regarded as conclusive evidence of migration. The evidence of the migratory habits of the silver-haired bat rests chiefly on the ani- mal's periodical appearance in spring aud fall at the lighthouse on Mount Desert Rock, thirty miles off the coast of Maine. This species has also been observed on the Bermudas. In August and September, 1890 ajid 1891, I had the opportunity to watch the appear- *Trans. Royal Soc. Canada V (1887), Section V, p. 85, 1888. 1 1 may add that I have a bat of this species, killed at Brownsville, Texas, on October 22d. ance aud disappearance of three species of bats at a locality where none could be found during the breeding season. Highland Light, the place where my observations were made, is situated near the edge of one of the highest points in the series of steep blufts of glacial deposit which form the outer side of Cape Cod, Massachusetts. The lidit, which is less than ten miles from the northern extremity of the cape, is separated from the mainland toward the east and northeast by from twenty- five to fifty miles of water. The bluff on which it stands rises abruptly from the beach to a height of one hundred and fifty feet. I found the bats for the most part flying along the face of this bluff, where they fed on the myriads of insects blown there by the prevailing southwest winds. They chiefly frequented the middle and upper heights and seldom flew over the beach at the foot of the bluff or over the level ground about the light- house. I do not know where the animals spent the day, as careful search in old buildings, under the overhanging edge of the bluff, and in deserted bank swallow holes, failed to reveal their hiding places. It is possible that they found shelter in the dense, stunted, oak scrub with which the bluff is in many places crowned, but of this I have no evidence. I hope that the ob- servations given below may again call the attention of field naturalists to a subject which presents many difficult and interest- ing problems. Atalapha noveboracensis* (red bat) . Augtist 21, 1890. The first bats of the sea- son were seen this evening. There were *With bat nomenclature in its present unsettled state it is well to use the names adopted by Dr. 332 only two, and I could not positively identify them, but the}'^ were probablj- red bats. August 25, 1890. An adult male taken. August 28, 1890. Two seen. August 29, 1890. The evening was too chilly for many bats to be on the wing. A few A. noveboracends seen and two taken. Aiigust SO, 1890. Six or eight A. novehora- censis seen and three taken. The evening was warm and bats flew much more freely than on the 29th. Aiigust 31, 1890. A chill}' evening again, and onl}' two bats seen , both A. noveboracen- September 2, 1800. A few red bats seen and two taken. September 5, 1890. I was not at Highland Light this evening, but Mr. W. M. Small re- ported a heavy flight of bats. He shot five, all A. noveboracensis. September 8, 1890. Heavy fog, so that no bats could be seen, if any were moving along the face of the bluff. Three or four red bats flew about the light house tower dur- ing the first half of the night, feeding on insects attracted by the light. They flew mostly below the level of the deck which encircles the tower about six feet be- low the lantern and never approached the light itself. September 12, 1890. A single red bat shot this evening. After this date I watched for bats on sev- eral consecutive evenings. As I saw no more I concluded that the migration had ended. August 25, 1891. Fourteen Atalapha nove- boracensis, the first bats of the season, seen Harrison Allen in his latest Monograph of the Bats of North America (1893), although many of these will require revision. this evening. They were flying both north and south. Augiist 26, 1891. Evening very foggy. A red bat which flew about the lighthouse was the only one seen. August 27, 1891. Half a dozen red bats seen and one taken. August 28, 1891. Four red bats seen. All flew toward the south. August SO, 1891. A red bat caught in a house near the edge of the bluff. September 2, 1891. Eight or ten seen and three taken. The movement this evening was mostly, though not wholly, from north to south. September S, 1891. Six seen and three tak^n. September 5, 1891. Evening cold and misty. No bats moving. September 7 and 8, 1891. A few bats seen each evening, but none taken. All ap- peared to be this species. September 10, 1891. One red bat shot. September 11, 1891. One seen. September 12, 1891. One killed. About a dozen bats seen, but how many were of this species, and how manj^ Lasionycteris nodiva- gans T could not determine. September IS, 1891. About a dozen bats seen. Two of these were certainly red bats. After this date I watched for bats on con- secutive evenings for more than a week. As I saw none I finally gave up the search. ATALAPHA CINEREA (HOARYBAT). August 26, 1890. One Atalapha cinerea, the only bat seen, shot this evening. August 28, 1890. Two hoary bats taken, and at least two, and probably four, others seen. 333 August 30, 1890. Two taken and two others seen. September 2, 1890. Only two seen. Both taken. No more hoary bats seen during 1890. Augttst 25, 1891. A single Atalapha cinerea seen fljnng south along the face of the bluff this evening. September 2, 1891. One seen flying north. September 12, 1891. An adult male shot — the last of the season. At Highland Light I found the hoary bat less active and irregular in its movements than the red bat. Its large and compara- tively stead}^ flight made it easier to shoot than either of the two smaller species with which it was associated. It began to fly immediately after sunset. In the Adiron- dacks Dr. C. Hart Merriam found the hoary bat a late flyer, and an exceeding difficult animal to kill on account of its swift, ir- regular motions.* It is possible that while on Cape Cod the animal modifies its habits on account of the unusual surroundings in which it finds itself. The fatigue of a long * Trans. Linn. Soc New York, II, p. 78-83. 1884. migration might also have an appreciable effect on a bat's activity. lasionycteris noctivagans (silver-haired bat). September 1, 1890. One silver-haired bat taken. September 2, 1890. Four taken and per- haps a dozen others seen. The silver-haired bat was not seen again during 1890. September 10, 1891. Three shot and prob- ably half a dozen others seen. They were mostly flying north. September 11, 1891. Two shot and four or five more seen. September 12, 1891. About a dozen bats seen. Some were without doubt this species, but just what proportion I could not tell. While September 12th is' the latest date at which I have seen Lasionycteris nodivagans at Highland Light, I have a specimen killed there by Mr. W. M. Small on October 28, 1889. Gerrit S. Miller, Jr. 334 ECOLOGICAL DISTRIBUTION OF SIX SPECIES OF SHREWS AND COMPARISON OF SAMPLING METHODS IN THE CENTRAL ROCKY MOUNTAINS Larry N. Brown Abstract. — The ecological distribution of six species of shrews was studied using sunken cans in 14 montane and intermontane habitats in southern Wyoming. The vagrant shrew ( Sorex vagrans ) and masked shrew ( Sorex cinereiis ) were cos- mopolitan in distribution. Sorex cinereus was slightly more abundant in moist plant communities, whereas Sorex vagrans predominated in slightly drier communities. Merriam's shrew ( Sore.t merriami) occurred only in arid portions of the plains and foothills, and in short-grass prairie was the only shrew taken. The water shrew (Sorex palustris) occurred only along or near cold mountain streams and ponds. The dwarf shrew (Sorex iwmis) and pigmy shrew (Microsorex hotji) occupied re- stricted mountain habitats. The dwarf shrew was abundant in rocky locations in both alpine and subalpine plant communities; the pigmy shrew was taken only in peat-moss bogs in the spruce-fir zone. A comparison of snap traps and sunken cans as methods of collecting shrews revealed that snap traps failed to demonstrate the presence of Sorex nanus and Microsorex hoyi in areas where they were abun- dant. Also, densities of Sorex vagrans and Sorex cinereus indicated by snap traps were considerably below those indicated by sunken cans. The habitat preferences and ecological distribution of the six species of shrews found in the Central Rocky Mountains have not been extensively stud- ied. Only scattered references to the ecological distribution of shrews in the Rocky Mountains occur in the literature (Gary, 1911; Warren, 1942; Negas and Findley, 1959; and Spencer and Pettus, 1966). No thorough study dealing with Wyoming shrews has been reported. The species studied were the masked shrew (Sorex cinereus), the vagrant shrew (Sorex vagrans), the dwarf shrew (Sorex nanus), the Merriam's shrew (Sorex merriami), the water shrew (Sorex palustris), and the pigmy shrew (Microsorex hoyi). Information was collected on the ecological distribution of these species in southern Wyoming in terms of type of plant cover and prox- imity of water. Data on the indicated abundance of shrews using two different trapping methods were also compiled. Materials and Methods The Medicine Bow Mountains and Laramie Basin area of southern Wyoming have a wide range of plant communities, which occur at altitudes of from 7000 to 12,000 ft. The eight montane and intermontane plant communities selected for sampling were cottonwood-wil- low, short-grass prairie, sagebrush, mountain mahogany, aspen, lodgepole pine, spruce-fir and alpine tundra. Brief descriptions of the sampled areas in Albany County, Wyoming, are as follows: 1. Cottonwood-willotc. — 7160 ft; 10 miles SW Laramie, along Big Laramie River. Dom- inants: Populus angustifolia, Salix sp. Several grasses (Poa, Agropijron, Carex) present in understory. 617 335 618 JOURNAL OF MAMMALOGY Vol. 48, No. 4 2. Short-firass prairie. — 7180 ft; V2 mile E Laramie. Dominants: Boiiteloiia gracilis, Buchloe dactijloides. Numerous forbs {Eriogoniim, Gaitra, Phlox) also represented. 3. Sagebrush. — 7220 ft; 20 miles N Laramie. Dominants: Artemisia Iridentata, Purshia tridentaia. Several grasses ( Poa, Koeleria, and Agropijrorx ) abundant in open spaces be- tween shrubs. 4. Mountain mahogany. — 7250 ft; 1 mile E Laramie. Dominant: Cercocarpus montanus. Present: Symphoricarpos, Artemisia, Amelanchier, as well as several grasses and forbs. 5. Aspen. — 8205 ft; % mile NW Centennial. Dominant: Populus tremuloides. A dense understory of grasses (Poa, Agropyron) and scattered shrubs {Acer, Rosa, and Berberis). 6. Lodgepole pine. — 9300 ft; 5 miles VV Centennial. Dominant: Pinus contorta. Under- story of scattered forbs (Lupinus, Antennaria, etc.) and small trees (Abies, Juniperus, and Picea ) . 7. Spruce-fir. — 9630 ft; 7 miles W Centennial. Dominants: Picea engelmanni, Abies lasiocarpa. Several shrubs (Ribes, Rosa, Vaccinium, Berberis) common in understory. 8. Alpine tundra. — 10,470 ft; 5 miles W University of Wyoming Science Camp. Domi- nants: Artemisia scopulorum, Silene acaulis, Poa alpina, Trifolium sp., Salix sp. A more detailed description of most communities in the Front Range of the Rocky Moun- tains is found in Marr (1961). Moist bogs or marshes interrupted the uniformity of four of the above plant communities, specifically the aspen, lodgepole pine, spruce-fir, and alpine tundra. Therefore, to compare shrew populations in adjacent wet and dry situations, a bog or marsh near each original sampling station was trapped (sampling sites were 500-1000 ft apart). Brief descriptions of these wet plant communities are as follows: 1. Bog in alpine tundra. — 10,460 ft; 5 miles W University of Wyoming Summer Science Camp. Dominants: sedges (Carex .sp.), dwarfed willows (Salix sp.) around small, shallow pond. 2. Bog in spruce-fir. — 9620 ft; 7 miles W Centennial. Dominants: sedges (Carex sp.), horsetails (Equisetum sp.), willows (Salix sp.), sphagnum moss (Sphagnum sp. ). 3. Bog in lodgepole pine. — 9295 ft; 5 miles W Centennial. Dominants: willow (Salix sp.), alder (Alnus tenuifolia), sedges (Carex sp. ). Small pond of open water at edge of bog. 4. Bog in aspen. — 8200 ft; % mile W Centennial. Dominants: willow (Salix sp. ), alder (Alnus tenuifolia), aspen (Populus tremuloides) , sedges (Carex sp.), horsetails (Equisetum sp.). Two beaver ponds adjacent to the area. To check the preference of certain shrews for rocky areas, two e.xtensive rockslides were sampled at different elevations. One of these was a subalpine rockslide at 8480 ft elevation (3 miles NW Centennial) in an ecotone area of lodgepole pine, aspen, and sagebrush. The other trapping station was in a vast alpine rockslide above tree line at 10,600 ft elevation. Thus, 14 locations were sampled. Shrews were trapped in pit-fall traps made of one-gallon tin cans. These were buried in the ground with the mouth of each can just below the ground's surface. A grid with an interval of 30 ft was used in placing these traps. Each sampling plot consisted of 25 cans arranged in five rows having five cans each. In areas with a high water table, it was necessary to punch holes in the bottom or sides of each can. This allowed water to enter the can to equalize pressure and prevent the can from lieing forced up out of the ground. To catch shrews in such localities, it was desirable that the top of the water table ])e at least' an inch or so below ground surface. High water tables in several bogs required that cans be placed on a slight elevation or hummock nearest each grid intersect. The sinking of cans in rockslides was accomplished by removing rocks by hand until a pit into the interior of the slide was created. Wedged rocks and large lioulders usually limited the depth of each excavation to three to five feet. Each can was then placed at the deepest point and fist-sized rocks were used to rebuild the sulistrate almost up to the lip of 336 November 1967 BROWN— ECOLOGICAL DISTRIBUTION OF SHREWS 619 Fig. 1. — Diagrammatic representation of placement of pitfall traps (gallon cans) in study of rockslide habitats in mountain locations. the can. Then gravel was used to build a pavement to the lip of the can (Fig. 1). If only larger, fist-sized rocks were placed around each can, sufficient spaces remained adjacent to the mouth to constitute a barrier for access by the shrews. Between .30 May and .3 June 1966, 350 can traps were set out at the 14 locations. Cans were checked on alternate days for slightly more than 3 months. The total number of shrews by .species taken per habitat was determined for the total sampling period. In September 1966, a direct comparison of two methods of collecting shrews was made involving standard mouse-sized snap traps and the sunken cans. A snap trap was baited and set adjacent to each can in the sampling grid of the subalpine rockslide and the spruce- fir bog. Both sets of traps were checked once daily for 12 days. The bait for the snap traps consisted of a mi.xture of peanut butter, bacon grease, rolled oats, and chopped raisins. Results and Discussion Following the placement of cans in various plant communities no shrews were captured in any plot for a period of 10 to 14 days. After this period, shrews appeared in the cans with regularity throughout the remainder of the summer; this suggested that all shrews avoided the areas where cans were placed until they became accustomed to the change. Table 1 summarizes the relative abundance of five species of shrews at the various sampling stations. The masked shrew (Sorex cinereus) and vagrant shrew (Sorex va^rans) were the most cosmopolitan in distribution, being rep- resented in all the habitats sampled except short-grass prairie. These species were always taken together, but in varying densities that appeared to correlate with moisture conditions. Sorex cinereus was generally more abundant than S. va^rans in the moist bog localities regardless of altitude. Sorex va^ran.s had 337 620 JOURNAL OF MAMMALOGY Vol. 4H, No. 4 Tahle L — Ecological cllstrihuiion and relative abundance of five species of shrews in rela- tion to varions habitats in the Medicine Bow Mountains and Laramie Basin of Wyoming, summer 1966. Micro- Sorcx Sorcx Sorcx Sorcx sorex Total Habitat and ele\ ition cincrciis vagrans nanus mcrriami hoyi collected Cottoiiwood-willow (7160) 2 14 0 0 0 16 Short-grass prairie (7180) 0 0 0 3 0 3 Sagel:)nish ( 7220 ) 4 11 0 3 0 18 Mountain mahogany ( 7250 ) 2 6 0 2 0 10 Aspen ( 8205 ) 5 16 0 0 0 21 Bog in aspen ( 8200 ) 28 12 0 0 0 40 Snbalpine rockshde (8480) 11 20 25 0 0 56 Lodgepole pine ( 9300 ) 6 16 0 0 0 22 Bog in lodgepole ( 9295 ) 29 11 0 0 0 40 Spruce-fir ( 9630 ) 8 18 0 0 0 26 Bog in spruce-fir ( 9620 ) 32 15 0 0 6 53 Alpine tundra (10,470) 4 8 2 0 0 14 Alpine willow bog (10,460) 9 5 0 0 0 14 Alpine rockslide (10,600) 3 9 21 0 0 33 higher population densities in the mesic communities that were paired with the bogs (aspen, lodgepole pine, spruce-fir, and alpine tundra). The vagrant shrew was likewise slightly more abundant than S. cinereus in the other mesic situations sampled such as rockslides, sagebrush, mountain mahogany, and cot- tonwood-willow communities. These findings are in disagreement with those of Clothier (1955), who reported that S. vafirans was always more numerous in Montana than S. cinereus regardless of habitat. My results confirm the find- ings of Getz (1961) that S. cinereus is abundant in moist or standing-water situations, and the findings of Hoffmann and Taber (1960) that S. vagrans is present at high altitudes as well as low. The dwarf shrew, which is generally considered by mammalogists to be rare, was abundant in two restricted habitats and was present at a third. It was more numerous than S. vagrans by a slight margin (25 S. nanus as compared to 20 S. vagrans) in the subalpine rockslide and was by far the predominant shrew in the alpine rockslide (21 animals out of 33). Sorcx nanus was also rep- resented in the alpine tundra plot by two animals taken near a rock outcrop. This represents an altitudinal range extending from 8480 to 10,600 ft and in- cluding several types of montane plant communities. Since rockslides extend continuously from 10,600 to 12,000 ft in the Medicine Bow Range, there is little doubt the species reaches that altitude in this habitat. All dwarf shrews were captured at considerable distances from water, suggesting they may be some- what adapted to dry situations. The preference of S. nanus for rocky areas in the mountains was first sug- gested by Hoffmann and Taber (1960), who collected several in polygonal rock fields on the Beartooth Plateau. They have been recorded in a variety of other 338 Novetnber 1967 BROWN— ECOLOCIICAL DISTRIBUTION OF SHREWS 621 montane habitats by Schellbach (1948), Clothier (1957), and Bradshaw (1961). Spencer and Pettus (1966) reported S. nanus was abundant in an open clear- cut area of spruce-fir forest. They did not mention the presence or absence of rocks. Sorcx merriami was trapped in three plant communities that were repre- sented only at lower elevations in southern Wyoming. These habitats were short-grass prairie, sagebrush, and mountain mahogany. None of these habi- tats occurred higher than the mountain foothills ( 7500 ft ) . These three plots produced onl\ eight Merriam's shrews for the whole summer, indicating that population levels in the areas sampled were not high. In the short-grass prairie plot, S. merriami was the only species of shrew present; in the low foothills, where sagebrush and mountain mahogany communities were sampled, they were present in low numbers with S. vagrans and S. cinercus. The only Mer- riam's shrew previously taken in southeastern Wyoming was reported by Mickey and Steele (1947), from short-grass prairie near Laramie. In W^ash- ington, Johnson and Clanton ( 1954 ) found that this species preferred the sage- brush-bunch grass community, where individuals were taken in the tunnels of the sagebrush vole (Laounis curtatus). Hoffmann (1955) likewise collected S. merriami in sagebrush in California. One of the few records of occurrence of the species in mountain mahogany was that of Hoffmeister ( 1956 ) in Owl Creek Canyon in northeastern Colorado, about 40 miles south of the present study area. In southeastern Wyoming, S. merriami occurred in the driest habitats and generally at lower elevations than did other species. Merriam's shrew was taken with Lagurus curtatus in the sagebrush and short-grass prairie area, but this vole was absent from the mountain mahogany community. The pigmy shrew (Microsorex hoiji) was not known to occur in Wyoming until it was taken in 1963 in the Medicine Bow Range ( Brown, 1966 ) . A dis- junct population occurs in southern Wyoming and adjacent northern Colorado and is more than 500 miles south of the nearest locality in Montana where the species has been reported (Hall and Kelson, 1959). In the course of the pres- ent study, pigmy shrews were captured at one sampling station, in the spruce- fir bog at 9620 ft elevation. Here Microsorex was encountered only around the periphery of the bog in an area dominated by a deep, spongy mat of sphagnum moss. In the strip of sphagnum, they were taken in equal numbers with S. cinereiis and in greater numbers than S. vagrans. Spencer and Pettus (1966), in their study of a single bog area west of Fort Collins, Colorado, reported that the pigmy shrew was most abundant in the transition area between the bog and the surrounding spruce-fir forest. This is in agreement with my findings. The habitat preferences of Microsorex in the other parts of its range seem to be rather broad, including heavy woods, clearings, and pastures in both wet and dry situations (Burt, 1957). However, Jackson (1961) and Buchner (1966) noted that in Wisconsin and Canada the species occasionally is found in cold sphagnum or tamarack bogs. In the MacDonald Range of northwestern Mon- 339 (y22 journal of MAMMALOGY Vol. 48, No. 4 Taisle 2. — Comparison of trap])ii\<^ results for .shrews ii.sinfi .snap traps and .sunken cans for 12 days at two location.s in the Medicine Bow Mountains, Wtjoniin^. Subalpine rockslide Spruce-fir bog Species Snap tr aps Sunken cans Snap traps Sunken cans Horex vapran.s 2 4 2 3 Sorex cinereu.s 1 2 4 9 Sorcx nanus 0 6 0 0 Microsorex hoiji 0 0 0 2 Totals 3 12 6 14 tana, Conaway (personal communication) captured Microsorex in dry areas of clearcut forest having a dense ground cover. The relict population located in the central Rockies may have a broader habitat specificity than indicated by the present study. The water shrew {Sorcx palustris) is too large to be retained in gallon cans unless there is some water present. Water shrews were taken in partially flooded cans in habitats ranging from the alpine willow bog at 10,470 ft eleva- tion down to the willow-alder bog in the aspen community at 8200 ft elevation. The species was never trapped at a distance greater than 100 ft from a moun- tain stream or pond. The total samples of shrews at each location were compared, and two types of habitats were found to support especially high densities of shrews. The most productive habitats were rockslide areas and marshy or boggy areas, both of which have high invertebrate populations that can serve as a readily available food supply. The least productive trapping plot for shrews was located in short- grass prairie. This may have been due to the scarcity of suitable cover or to a relative scarcity of invertebrates. A comparison of the number and species of shrews collected utilizing equal numbers of snap traps and cans is presented in Table 2. In 300 trap-nights us- ing snap traps, three shrews of two species were collected at the subalpine rock slide. The sunken cans produced four times as many shrews during the same period and three species were represented. Six S. tmnus were taken in cans while none appeared in adjacent snap traps, suggesting that the species readily avoids traps and therefore gives the appearance of being rare. Twenty-five sunken cans produced slightly more than twice as many shrews ( 14 animals) than did 25 snap traps (six animals) in the bog in the spruce-fir community during the 12-day period. Again, S. vagrans and S. cinereus were taken in both snap traps and cans, but Microsorex was captured only in cans. Even for the two "common" species, higher numbers were recorded for the can traps. These and other trapping results indicate that S. nanus, S. merriami, and Microsorex hoiji are seldom captured in snap traps even when all are abun- dant. Sorex vagrans, S. cinereus, and S. palustris were readily captured in snap traps, but data provided by pitfall traps suggest that densities calculated for 340 Xoinnhcr /.967 BHOWN— ECOI .OCICAL DISTRIBl'TION OK SUKEWS 62:3 tlit'se species on the basis ot snap-trap catches ma\ l)e consistently too low. MacLeod and Lethiecq (1963) presented similar data when comparing these trapping methods for S. cinereus in Newfoundland. ACKNOWLEDCMENTS I wisli to thank tlie following students who helped with colleetiiiK the shrews: James Bradley, Robert CiiKj-Mars, Ronald Lynde, Paul Lussow, Mar\in Maxell, and Richard Mc- (luire. Literature Cited Bradshaw, C. \. R. 1961. New Arizona localit\ for the dwarf shrew. J. Manini., 42: 96. Rhown. L. X. 1966. First record of the pigmy shrew in Wyoming and description of a new subspecies (Mammalia: InsectiNora ). Proc. Biol. Soc. Washington, 79: 49-51. BucKNEH, C. H. 1966. Populations and ecological relationship.s of shrews in tamarack Iiogs of southeastern Manitoba. J. Mamm., 47: 181-194. BiHi, W. 11. 1957. Mammals of the Clreat Lakes Region. Uni\. Michigan Press, .Ann .Arbor, \\ + 246 pp. Gary, .M. 1911. A biological surve\ of Colorado. \. Amer. P'auna, 33: 1-256. Clothieh, R. R. 1955. Contribution to the life history of So/c.r vagrans in Montana. J. Mamm., 36: 214-221. . 1957. .\ second dwarf shrew from \ew Mexico. J. Mamm., 38: 256. C.ETZ, L. L. 1961. Factors influencing the local distribution of shrews. Amer. Midland Nat., 65: 67-88. Hall, E. R., and K. R. Kelson. 1959. The mammals of North America. The Ronald Press, New York, 1: xxx + 546 + 79. Hoffmann, R. S. 1955. Merriam shrew in California, j. Mamm., 36: 561. Hoffmann, R. S., and R. D. Taueh. 1960. Notes on Suicx in the Northern RocIcn Moun- tain alpine zone. J. Mamm., 41: 230-234. HoFFMEisTER, D. F. 1956. A record of ^orex merriami from northeastern Colorado. J. Mamm., 37: 276. Jackson, H. H. T. 1961. Mammals of Wisconsin. Uni\-. Wisconsin Press, Madison, xii + 504 pp. Johnson, M. L., and C. W. Clanton. 1954. Natural history of Sorex merriami in Wash- ington State. Murrelet, 35: 1—1. MacLeod, C. F., and J. L. Lethiecq. 1963. A comparison of two trapping procedures for Sorex cinereus. J. Mamm., 44: 277-278. NLahk, J. 1961. Ecosystems of the east slope of the Front Range in Colorado. L'niv. Colo- rado Stud. Biol., 8: 1-134. Mickey, A. B., and C. N. Steele, Jh. 1947. A record of Sorex merriami merriami for .southeastern Wyoming. J. Mamm., 28: 293. Negus, iX. C, and J. S. Flndley. 1959. Mammals of Jackson Hole, Wyoming. J. Mamm., 40: .371-381. ScHELLB.\CH, L., HI. 1948. A record of the shrew Sorex luiiius for Arizona. J. Mamm., 29: 295. Spenceh, A. W., AND D. Pettus. 1966. Habitat preferences of fi\e sympatric species of long-tailed shrews. Ecology, 47: 677-683. Warhen, E. R. 1942. The mammals of Colorado. . . . l'ni\ . Oklahoma Press, x\ iii + 330 pp. Department of Zoology and Physiology, University of Wyoming, Laramie {present ad- dress: Department of Zoology, University of Sonthern Florida. Tampa). Accepted 1 Sep- tember 1967. 341 378 Journal of Wildlife Management, Vol. 20, No. 4, October 1956 CHANGES IN NORWAY RAT POPULATIONS INDUCED BY INTRODUCTION OF RATS David E. Davis and John J. Christian Division of Vertebrate Ecology, Johns Hopkins School of Hygiene and Pubhc Health, Baltimore 5; Naval Medical Research Institute, Bethesda, Maryland The introduction of aliens into an existing population of mammals may be followed by unexpected effects that relate to social structure and population composition. These effects were studied by introducing alien rats into stationary and increasing popula- tions of rats in city blocks. This work is part of a continuing study of the mechanisms of change in vertebrate populations using Nor- way rats ( Rattus norvegicus ) in residential areas in Baltimore as experimental animals (Davis, 1953). These rats inhabit back yards, basements, and garages and feed on garbage. The human sanitary conditions in general are poor and remain unchanged for months at a time, so that the food supply of the rats has only slight seasonal variations. Other environmental conditions are similarly sub- ject to little change for many months at a time. The constancy of these factors permits Methods and Procedures The procedures followed to study the ef- fects of introducing strange rats into a popu- lation consisted of taking some rats from a stationary or increasing population in one block and introducing them into a compara- ble population in another block and observ- ing the resulting changes in the second pop- ulation. The status (stationary or increasing) of the population was determined by esti- mates at bimonthly intervals for more than a year. Blocks that appeared to be either stationary or increasing were selected in October and monthly estimates made. From this group 4 stationary and 4 increasing pop- ulations were chosen. To get a base line for adrenal weights, six rats of one sex, weighing over 200 grams each, were removed from each block during the first experimental week ( week 1 ) . Alien rats were then intro- experiments on populations in a relatively duced into each block during the third week, stable environment. Finally, the population and at the same time native rats were re- of rats in each block is essentially discrete moved from the increasing populations. The and isolated, as rats rarely travel from one details of these removals and introductions block to another ( see Davis, 1953, for refer- are contained in tables 2 and 3. Estimates ences). were made during the sixth and eighth 342 ' Changes in Rat Populations — Davis and Christian 379 weeks, each followed by the removal of a small sample of rats for adrenal weights. The adrenal weights were expressed for each sample as the mean per cent of standard reference values (Christian and Davis, 1955). The rats to be introduced into a population were indi\'idually marked by toe-clipping prior to introduction, whereas the native rats were not marked. The details of the history of each population were complicated by the impossibility of introducing exactly the same number of rats into each block on exactly the same days, and the numerical popula- tion size also differed in each block. A discussion of the likelihood of error is desirable when it is claimed that two popu- lations differ in number, since the detection of changes is fundamental to the conclusions derived from these introductions. Some as- pects of the census method were discussed by Brown, et al. ( 1955 ) . However, the basic problem is that, even witli trapping, the true number of rats in a block is not known. Nevertheless, a check on the validity of esti- mation can be made by comparing estimates before and after a trapping program. Sup- pose that an estimate of Nj rats is first ob- tained, subsequently T rats are removed and a second estimate of No rats is made. Obviously N^ should equal T + No. A figure for percentage of error can be given as Ni-(T + N,). Table 1. - - Distribution OF Differences Among Estimates r" Blocks Per cent Erro Positive Negative Total 0-9 13 6 19 10-19 6 11 17 20-29 4 4 8 30-39 2 3 5 40-49 1 0 1 Totals 26 24 50 » Ni - (T + No) N, For example, if the estimate for a block is 151 rats and then 81 are re- moved by trapping and an estimate of 63 is made, then ^ti — 4.6 per cent. Other procedures could be used such as ^-(^j-N') or N- -^y_-/- The first procedure is preferred because it bases the calculations on N^, which is the estimate that was used to determine the status of the block. A total of 50 populations was avail- able to determine the extent of error. Each had been trapped during the past 6 years, and an estimate had been made before trap- ping and another within a month after ces- sation of trapping. Naturally, some changes can occur during the intervening month, but for practical reasons it is usually not possible to make an estimate promptly after the ces- sation of trapping. These blocks contained 3,707 rats by the estimates (NJ and 1,502 were trapped. The number of rats per block Ni varied from 15 to 182. The distribution of errors is given in Table 1. The percentage of error was independent of the number of rats in the population. From these differ- ences the standard error of the difference can be calculated to be 10.7 per cent. This value can be used as an indication of the errors to be expected in estimates of popula- tion changes in blocks. For example, from Table 2 it is seen that the estimate (block 150128) before introduction was 116 and after was 89. The percentage difference is 23.3 which when divided by 10.7 gives a ratio of 2.2. This difference appears to be statistically significant. Results and Discussion The histories of the populations are given by blocks in tables 2 and 3 and figures 1, 2, and 3. The terms "replacement" and "sup- plement" require clarification for this dis- cussion. We mean by replacement that ap- proximately the same number of rats was introduced as was removed. Supplement means that many more alien rats were intro- duced than were removed. A quantitative percentage might have been used to dis- tinguish these two terms, but it would have been rather meaningless because ( 1 ) the size of the individual rats varies consider- ably, and (2) immediate mortality is prob- ably high. Therefore, we really do not know the actual number of rats that produced the results. Another factor is that births and deaths are normally high in any population of rats. The average monthly death rate is about 20 per cent for stationary rat popula- tions; therefore, their birth rate is also about 20 per cent. Comparable mortality and birth rates for increasing populations are prob- 343 380 Journal of Wildlife Management, Vol. 20, No. 4, October 1956 Table 2. — Results of Introduction of Rats into Stationary Populations Block number Zero week is 140338 Dec. 16 140344 Dec. 16 140118 Feb. 9 150128 Feb. 9 Week Rats w R w R w R Population -20 62 -20 30 -19 105 -18 122 Population -13 42 -13 32 -11 98 -10 118 Population - 5 40 - 5 38 - 6 87 - 6 120 Population 0 49 0 35 0 100 0 116 Rats removed 1 6M 1 6F 1 6F 1 6M Rats introduced 3 lOM 3 8F 2-6 23F 2-6 27M Rats removed — — — — — — — — Population 6 45 6 44 6 76 7 89 Rats removed 6 6M 6 6F 6 6F 7 6M Population 8 56 8 23 11 63 10 86 Rats removed 8 22 8 12 11 27 10 36 Week Indexi w I w 1 w 1 1 83.0 1 91.8 1 93.4 1 79.2 Adrenal size 6 84.1 6 85.3 6 90.4 7 71.4 8 88.0 8 102.4 11 68.4 10 73.6 1 Mean of the individual per cent of appropriate reference value for the sex and size of the rat. ablv about 15 per c ent and 25 per cent ] Der STATIONARY month respectively. It is unwise, under these circumstances, to attempt a precise measure- ment of numerical differences in the num- bers of rats used. The population estimates (Table 2) in the two replacement blocks 6 weeks after the introduction of rats showed ( Fig. 1 ) that block 140338 was not significantly different from the previous estimate, while block 140344 had apparently increased ( P is about .04). While the apparent difference in re- sults in these 2 blocks might be due to sex ( the females less disturbing ) or to numbers (fewer introduced into 140344), no inter- pretation will be attempted for the reasons cited above. The two supplemented blocks declined significantly (P is about .04 for each) (Fig. 1). We recognize that the replacement pair of populations was done in December 1953, and the supplemented pair in February 1954, and that differences might be due to some seasonal aspect. However, the only known seasonal change, an increase in breeding from December to February, would produce the opposite result. Population growth ceased in all four in- creasing populations following the replace- ment of native with alien rats (figs. 2, 3,). In no block was the difference statistically significant. It apparently made no differ- ence whether the sex of the introduced rats Fig. 1. The changes in four stationary populations for 20 weeks before introduction of rats ( at 0 time ) and about 10 weeks after. The number added is indicated by a plus sign, the number removed by a minus sign. 344 Changes in Rat Populations — Davis and Christian Table 3. — Results of Introduction of Rats into Increasing Populations 381 1 Block number 140111 140134 140201 140222 Zero week is . . Dec 16 Dec . 16 Feb 9 Feb 9 Week Rats W R W R w R Population -20 110 -20 80 -25 86 -20 57 Population -13 118 -13 88 -16 100 -14 62 Population - 5 133 - 4 115 - 8 105 - 5 95 Population 0 150 0 140 0 135 0 90 Rats removed 1 6F 1 6F 1 6M 1 6M Rats introduced 3 22F 3 28F 3 18M 3 20M Rats removed 3 28F 3 22F 3 20M 3 ISM Population 6 167 6 130 6 130 6 85 Rats removed 7 6F 7 6F 7 6M 7 6M Population 10 152 10 130 10 130 10 70 Rats removed 10 48 10 73 10 57 10 24 Week Indexi W I w I w I 1 92.1 1 91.5 1 104.4 1 93.8 Adrenal size 7 93.8 102.8 7 85.3 7 99.1 10 101.5 10 90.4 10 91.1 10 84.6 1 Mean of the individual per cent of appropriate reference value for the sex and size of the rat. was male or female. The population from block 140222 (Fig. 2) may have become stationary just prior to the introduction of aliens, but the high rate of reproduction ( 4/6 mature females were pregnant) suggests that the population was increasing. The popula- tion in block 140111 increased numerically after the introduction, but the difference be- tween the two estimates is within the error of estimate and does not indicate a change in population. It appears that introducing a number of alien rats may halt the growth of increasing populations. The reader may have noticed that the total number of rats removed from the four in- creasing blocks was about 5 per cent greater than the number introduced, so that the rats in these populations were not replaced in the strict arithmetic sense of the word. However, considering the previously mentioned birth and mortality factors, it is not desirable to be more precise. All aspects considered, it is likely that the four increasing populations were somewhat reduced following replace- ment procedures. The four blocks (taken together) increased by 173 rats in the 20 weeks preceding replacement, so that they might have been expected to have had 600 rats 10 weeks after replacement instead of the observed 482, although the rate of in- crease would decline as the population in- creased. On several occasions episodes have been noted that appear to be explainable on the Q. O Q. < tr. 150 125 100- 75 50 25 0 MALES _L -30 •20 -t — • -10 WEEKS 10 Fig. 2. The changes in two increasing blocks before and after the introduction of males (symbols as in Fig. 1.) basis of introduction or actual immigration. In January 1946, about 60 rats were released in a block in one night as part of an experi- 345 382 Journal of Wildlife Management, Vol. 20, No. 4, October 1956 I75r 150 125 100 z o !5 ^75 Q. o Ql < DC 50 25 -6 +22 -28 / \ / \ / \ + 28 22 FEMALES ^x ^X -30 -20 -10 WEEKS 10 Fig. 3. The changes in two increasing blocks before and after the introduction of females ( symbols as in Fig. 1.) ment on "homing" ability in rats. The block originally contained about 100 rats but with- in 3 weeks there were so few rats left in the block that the project was stopped. Calhoun ( 1948 ) noticed the same result when he in- troduced rats into blocks. These episodes, as well as miscellaneous observations, stim- ulated a test in 1947 of the idea that the in- troduction of rats into a population would result in its decline. Accordingly, rats were introduced over a period of 4 months into two populations that had just reached a level judged to be stationary (Davis, 1949). The introduction of 90 rats in one block and 101 in the other was accompanied by declines of about 25 per cent and of 40 per cent respec- tively. The populations increased after the introductions ended. The present experiments suggest that the introduction of large numbers of rats into a population disrupts the population mech- anisms in some way that causes the popula- tions either to decline in numbers or stop growing. The decrease is due in part to a decline in reproduction. Data are not available for the period immediately after introduction, as it is not feasible to follow the population changes and simultaneously to collect a number of rats for reproductive data. However, the large sample of rats collected from the blocks 8 to 11 weeks after introduction had a high reproductive rate ( Table 4 ) and a low lacta- tion rate. One would conclude from these data that the number of pregnancies was low immediately after the introduction. Only about 25 per cent of the females were lacta- ting at 10 weeks, whereas normally about 40 per cent of the females of these rats are lactating (Davis, 1953). The high preva- lence of pregnancy presumably resulted from their more or less simultaneous re- covery from the effects of introduction. The decreases in rat populations obviously may have been due largely to mortality and move- ment, but data on this aspect are impossible to obtain under these conditions. Table 4. — Reproductive Records of Local Rats Captured 8-11 Weeks after Artificial Immi- gration of Rats into Blocks Number Mean Population Mature Per cent Number Per cent status Females Pregnant Embryos Lactating Increasing Stationary 85 66 33.0 31.8 10.38 9.63 20.0 28.8 Previous experiments have shown that the weight of the adrenal glands in rats responds to changes in population. An increase in adrenal weight parallels increases in popu- lation; the artificial reduction of a popula- tion also results in a decrease in adrenal weight (Christian, 1954; Christian and Davis, 1955). The adrenal responses of the two sexes are parallel ( ibid. ) . Experiments have indicated that changes in adrenal weight in response to changes in population result primarily from changes in cortical mass ( Christian, 1955a, 1955b, 1956 ) . To examine these problems, the adrenals of the rats from each block were removed and weighed. The observed adrenal weight for each rat was compared with a standard reference weight for the appropriate sex and size (length of head and body ) of rat ( Christian and Davis, 346 Changes in Rat Populations — Davis and Christian 383 1955), and expressed as a per cent of the reference value. These percentages for the rats from each sample were averaged and the means are recorded at the bottom of tables 2 and 3. We have used the mean value of a given sample as the unit of measurement for comparing adrenal weight with popula- tion size (Christian, 1954; Christian and Davis, 1955). The results indicate that, in the replace- ment stationary blocks (140338 and 140344), there was a small increase in adrenal weight after 8 weeks, while the populations appar- ently remained practically unchanged (ta- bles 2 and 3, Fig. 1 ) . A mean decline in population size, par- alleled by a decrease in adrenal weight in at least one of the two blocks, followed the addition of a large number of alien rats to stationary populations (blocks 140118 and 150128 ) . The adrenal weights probably re- flect largely the final results of population manipulation rather than the immediate ef- fects, as the adrenal samples were obtained several weeks after the introductions or esti- mates. Therefore, the changes in adrenal weight probably reflect overall population changes rather than any immediate social strife resulting from the introductions. An experiment to collect samples a few days after the introductions is in progress and may show an increase in adrenal weight. The adrenal glands of rats from the in- creasing blocks showed no consistent change, although population growth terminated ( Ta- ble 3, figures 2 and 3 ) . The replacement of rats in increasing populations had little ef- fect on the adrenal weights of rats examined 10 weeks later. The results reported here may be applica- ble to certain stocking programs. A routine part of many game-management programs has been the introduction of a number of animals into an area with the expressed hope of increasing the population either directly or eventually by reproduction. Indeed, such stocking has often been considered a pana- cea for all hunting problems. The present results, using rats as experimental animals, show that the disruption of a population following an introduction may actually pro- duce a decline under certain conditions. Evi- dently the introduction of a number of ani- mals may have disastrous results when a population is above the halfway point on a growth curve. Summary Wild Norway rats {Rattus norvegicus) were introduced from one city block to an- other to simulate immigration. The pop- ulation changes were determined by fre- quent estimates for about 20 weeks before introduction and 8 to 11 weeks thereafter. From two blocks with stationary rat popula- tions, some rats were removed and then replaced by aliens. The populations re- mained stationary. In two blocks about four times as many rats were introduced as were removed. The populations declined about 25 per cent. In four blocks with increasing populations about one-fourth of each pop- ulation was removed and replaced by alien rats from other blocks. The increase halted. The reproductive rate 8 to 11 weeks after the introduction was normal for an increas- ing population, but the lactation rate was low, indicating that the decline in popula- tion growth was due in part to a decreased reproductive rate, and that the population was back to normal pregnancy rate in two months. The adrenal weights were also es- sentially normal for the population level two months after introduction. References Brown, R. Z., W. Sallow, David E. Davis, and W. G. Cochran. 1955. The rat population of Baltimore 1952. Amer. J. Hyg., 61( 1 ) : 89-102. Calhoun, J. B. 1948. Mortality and movement of brown rats {Rattus norvegicus) in artificially super-saturated populations. J. Wildl. Mgmt., 12(2):167-172. Christl^n, J. J. 1954. The relation of adrenal size to population numbers of house mice. Sc. D. dissertation, Johns Hopkins Univ., Baltimore. . 1955a. Effect of population size on the adre- nal glands and reproduction organs of male mice in populations of fixed size. Amer. J. Physiol., 182(2):292-300. . 1955b. Effect of population size on the weights of the reproductive organs of white mice. Amer. J. Physiol., 181(3) :477-480. . 1956. Adrenal and reproductive responses to population size in mice from freely growing populations. Ecology, 37(2):258-273. AND D. E. Davis, 1955. Reduction of adrenal weight in rodents by reducing population size. Trans. N. Amer. Wildl. Conf., 20:177-189. Davis, D. E. 1949. The role of intraspecific compe- tition in game management. Trans. N. Amer. Wildl. Conf., 14:225-231. . 1953. The characteristics of rat populations. Quart. Rev. Biol, 28(4):373-401. Received for publication October 24, 1955. 347 A TRAFFIC SURVEY OF MICROTUS-REITHRODONTOMYS RUNWAYS By Oliver P. Pearson Patient observation of the comings and goings of individual birds has long been one of the most rewarding activities of ornithologists. The development in recent years of inexpensive electronic flash photographic equipment has made it possible and practical for mammalogists to make similar studies on this aspect of the natural history of secretive small mammals. The report that follows is based on photographic recordings of the vertebrate traffic in mouse runways over a period of 19 months. Species, direction of travel, time, temperature and relative humidity were recorded for each passage. In addi- tion, many animals in the area were live-trapped and marked to make it possible to recognize individuals using the runways. THE APPARATUS Two recorders were used. Each consisted of an instrument shelter and a camera shelter. Each instrument shelter was a glass-fronted, white box con- taining an electric clock with a sweep second hand, a ruler for measuring the size of photographed individuals, a dial thermometer and a Serdex membrane hygrometer. The ends of the box were louvered to provide circulation of air as in a standard weather station. This box was placed along one side of the runway, across from the camera shelter, so that the instruments were visible in each photograph ( Plate I ) . The camera shelter was a glass-fronted, weather- proof box containing a 16-mm. motion picture camera synchronized to an electronic flash unit. In one of the recorders the camera was actuated by a counterweighted treadle placed in the mouse runway immediately in front of the instrument shelter (Plate I, bottom). An animal passing along the runway depressed the treadle, thereby closing an electrical circuit through a mercury-dip switch. This activated a solenoid that pulled a shutter-release pin so arranged that the camera made a single exposure. The electronic flash fired while the shutter was open. This synchronization was easily ac- complished by having the film-advance claw close the flash contact. The camera would repeat exposures as rapidly as the treadle could be depressed, but at night about three seconds were required for the flash unit to recharge sufficiently to give adequate light for the next exposure. 169 348 170 JOURNAL OF MAMMALOGY Vol.40,No.2 The other recorder was actuated by a photoelectric cell instead of by a treadle. A beam of deep red light shone from the camera shelter across the runway and was reflected back from a small mirror in the instrument shelter to a photoelectric unit in the camera shelter. When an animal interrupted the light beam, the photoelectric unit activated a solenoid that caused the camera to make a single exposure, as in the other recorder. To avoid the possibility of frightening the animals it would be desirable to use infra-red-sensitive film and infra-red light, but standard electronic flash tubes emit so little energy in the infra-red that this is not practical. Instead, I used 18 layers of red cellophane over the flash tube and reflector to give a deep red flash of light. Wild mice, like many laboratory rodents, are probably insensitive to deep red light. I found no evidence that the flash, which lasts for only 1/lOOOth of a second, frightened the mice. A muffled clunk made by the mechanism also seemed not to alarm the mice unduly. When the camera diaphragm was set to give the proper exposure at night, daytime pictures were overexposed, since the shutter speed was considerably slower than l/30th of a second. To reduce the daytime exposure, a red filter was put on the camera lens. The filter did not affect night exposures because red light from the flash passed the red filter with little loss. In addition, on one of the cameras the opening in the rotary shutter was reduced to give a shorter exposure. Both recorders function on 110- volt alternating current. The treadle-actuated one could be adapted to operate from batteries. The units continue to record until the motion picture camera runs down or runs out of film. One winding serves for several hundred pictures. The film record can be studied directly by projecting the film strip without making prints. The camera shelter and instrument shelter had overhanging eaves to prevent condensation of frost and dew on the windows. A small blackened light bulb was also kept burning in the camera shelter to raise the temperature enough to retard fogging on the glass. Animals were encouraged to stay in their usual runway by a picket fence made of twigs or slender wires. No bait was used. A few individual animals could be recognized in the pictures by scars or molt patterns, but most had to be live-trapped and marked. Using eartags and fur-clipping I was able to mark distinctively (Plate I, bottom) all of the mice captured at any one station. The clipping remained visible for days or months depending upon the time of the next molt. The apparatus produces photographic records such as those shown in the lower pictures in Plate I. These can be transposed into some form as Fig. 2. THE STUDY AREA The study centered around a grassy-weedy patch surrounding a brush pile in Orinda, Contra Costa County, California (Plate I). The runways wound through a 20 X 20-foot patch of tall weeds ( Artemisia vulgaris, Hemizonia sp. and Rumex crispus) and under the brush pile. The weeds were surrounded 349 May, 1959 PEARSON— TRAFFIC SURVEY 171 by and somewhat intermixed with annual grasses. Oaks and other trees, as well as a house and planting, were 50 feet away. Summer climate in this region is warm and sunny with official mean daily maximum temperatures rising above 80 °F. in late summer. Official temper- atures occasionally reach 100°, and temperatures in the small instrument shelters used in this study sometimes exceeded this. Nights in summer are usually clear and with the mean daily minimum temperature below 52° in each month. About 27 inches of rain fall in the winter and there is frost on most clear nights. The mean daily maximum temperature in January, the coldest month, is 54°, and the mean daily minimum 31°. PROCEDURE I placed the first recorder in operation on January 29, 1956, and the second on October 19, 1956. Except for occasional periods of malfunction and a few periods when I was away they continued to record until the end of the study on September 10, 1957. Approximately 778 recorder-days or 111 recorder- weeks of information were thus obtained. The monthly distribution of records was as follows: January, 54 days; February, 70; March, 90; April, 80; May, 84; June, 67; July, 52; August, 88; September, 48; October, 33; November, 52; and December, 60. The recorders were placed at what appeared to be frequently used Microtus runways, usually situated on opposite sides of the weedy patch 20 to 30 feet apart. For one period of four months one of the recorders was placed at a similar weedy patch 70 yards away. Early in the study it was discovered that a neighbor's Siamese cat sometimes crouched on the camera shelter waiting for mice to pass along the exposed runway in front of the instrument shelter. Consequently, a 2iA-foot fence of 2-inch-mesh wire netting was set up enclosing most of the weedy patch. This prevented further predation by cats at the center of the study area, although cats continued to hunt outside of the fence a few yards away from the recorders. The only other tampering with predation was the removal of two garter snakes on April 11, 1957. RESULTS Traffic in individual runways. — The recorders were operated at eighteen different stations. At seven of these apparently busy runways a traffic volume higher than a few passages per day never developed, and so the recorders were moved within two weeks. Perhaps the mice originally using these runways had abandoned them or had been killed shortly before a recorder was moved to their runway, or perhaps the disturbance of placing a recorder caused the mice to divert their activities to other runways. At the other eleven stations a satisfactory volume of traffic was maintained for three to more than twenty weeks. A station was abandoned and the recorder moved when the traffic had decreased to a few passages per day. Subsequently, I found that even this little activity does not indicate that the mice are going to abandon the runway, 350 172 JOURNAL OF MAMMALOGY Vol. 40, No. 2 for on several occasions traffic in a runway dropped this low and then climbed again to high levels. At one recorder the total number of passages in consecutive weeks was 183, 84, 26, 75 and 203. The runway represented in Fig. 1 was one of those used most consistently, but even it shows marked daily and weekly fluctuations. It is probable that after a few weeks of disuse during the season when grass and weeds are growing rapidly, a runway would not be reopened, but during the rest of the year an abandoned runway remains more or less passable and probably more attractive to mice than the surrounding terrain. Figure 1 summarizes the traffic in one of the busiest runways. On the first night there were an unusual number of records of harvest mice whose curiosity may have been aroused by the apparatus. Obviously they were not frightened away. After a short time traffic increased to a high level and remained high until the middle of November, when passages by Microtus decreased sharply. During the week before the decrease, seven marked individuals provided most of the Microtus traffic. One of these individuals, an infrequent passerby, dis- appeared at the time of the decrease, but the other six remained nearby for at least another week and continued to pass occasionally. Those Microtus that disappeared later were replaced by others so that even the infrequent passages in late November and early December were being provided by seven marked individuals. The decrease of Microtus traffic was caused, therefore, not by deaths but by a change in runway preference. Several of these same individuals were using another runway 20 feet away in mid-January, February and March. Three to six marked Reithrodontomys, depending upon the date, were pro- viding most of the harvest-mouse traffic in the runway represented in Fig. 1. The average number of passages per day of animals of all kinds was eighteen. In the ten other most successful runways, the average number of passages per day ranged from two to nineteen. Figure 2 gives a detailed accounting of the traffic at a single recording station for six days. One can judge from this figure the kind of information (excluding n 'OTHER ^ = MEADOW MICE f{ ■HARVEST MICE OCTOBER NOVEMBER DECEMBER Fig. 1. — Traffic volume along one runway for 16 weeks. Meaning of symbols under the base line: T= live-trapping carried out for part of this day; 0= full moon; £= total eclipse of the moon; R= rain. Columns surmounted by a vertical line represent days for which the recording was incomplete; the heights of the various segments of these columns should be considered minimum values. 351 May, 1959 PEARSON— TRAFFIC SURVEY 173 temperatures and humidities) obtained with the recorders and can at the same time catch a reveahng ghmpse of an aspect of the biology of small mammals that has heretofore been revealed inadequately by trapping and other techniques. It may be seen that the mouse traffic was provided by one female and two male harvest mice and by three male, three female, and one or more unidentified meadow mice; together they gave between 15 and 24 passages each day. No individual passed more than eight times in one day. One hai-vest mouse ( R2 ) seemed to spend the day to the left and to make a single excursion ^:^:M^k^- PLATE I Top: Camera shelter (foreground) and instrument shelter in position at a mouse runway on the study area. Bottom: The kind of records obtained with the recorder; left — a meadow mouse marked by cHpping two strips of fur on the hips; riglit — a marked harvest mouse crossing the treadle. 352 174 JOURNAL OF MAMMALOGY Vol. 40, No. 2 to the light each night. Harvest mice first appeared in the evening between 6:37 and 7:22 and none passed after 6:26 in the morning. Five or six Microtus passed within a few hours (February 24), and there was nightly near-coinci- dence of Reithrodontomijs and Microtus. Traffic in all runways combined. — During the 111 recorder-weeks, the follow- ing passages of animals were photographed: Meadow mouse, Microtus calif ornicus 6,077 Harvest mouse, Rcithrodontomys megalotis 1,753 Bird (see following account) 382 Brush rabbit, Sylvilagus hachmani 94 Shrew, Sorex ornatus 56 Peromyscus (see following account) 39 Fence lizard, Sceloporus occidentalis 33 Garter snake, Thamnophis sp. 17 Salamander (see following account) 11 Alligator lizard, Gerrhonotus sp. 10 House cat, Felis domesticus 6 Newt, Taricha sp. 5 Pocket gopher, Thomomys hottae 3 Gopher snake, Pituophis catenifer 3 Mole cricket, Stenopelniatus sp. 2 Ground squirrel, Citellus beecheyi 1 Weasel, Mustela frenata 1 King snake, Lampropeltis getulus 1 Racer, Coluber constrictor 1 Total 8,495 On the basis of trapping results in this and in similar habitat nearby, large numbers of meadow mice and harvest mice were expected. The recording of at least 26 other species in the runways came as a pleasant surprise. Whereas all of these species would be expected to record their presence eventually, some of them are rarely seen or trapped near this location. After living five years on the study area, after doing considerable field work nearby, and after checking the recorders twice each day during the study, I have not yet seen a weasel or a ground squirrel within at least a mile of the study area. Weasels could easily escape detection, but large, diurnal ground squirrels must be very rare. The single individual recorded on August 31 may have been a young squirrel emigrating from some distant colony. Noteworthy absences were those of wood rats (Neotoma fuscipes), moles {Scapanus latimanus), and probably California mice (Peromyscus calif ornicus) , all of which were common within 100 feet of the recorders. An opossum (Didelphis marsupicilis) was seen a few feet from one of the recorders but did not appear on the films. No house mice [Mtis musculus) were detected in the photographs, although 353 May, 1959 PEARSON— TRAFFIC SURVEY 175 it is possible that some passages of Mtis were listed as of Reithrodontomys. House mice were caught occasionally in houses nearby and in a field near a poultry house 200 yards away, but none was caught during frequent live- trapping near the recorders. The total of 8,495 passages of animals gives an average of 11 passages per day in each runway. A patient, non-selective predator waiting for a single catch at runways such as these could expect, theoretically, a reward each 2.2 hours. The mean weight of animal per passage was about 31 grams, which would yield approximately 40 calories of food. This much each 2.2 hours would be more than enough to support an active mammal the size of a fox. Meadow mouse. — The 6,077 Microtus passages were distributed throughout the day and night as shown in Fig. 3 (above). The hours of above-ground activity, however, were quite different in winter than in summer, so Fig. 3 is only a year-around average somewhat biased by the fact that more Microtus were recorded in the spring than in the other seasons. A more detailed analysis of the Microtus data will be given in a later report. By marking as many of the mice as possible, it was found that usually four or more individual Microtus were using each runway but rarely more than ten. On some occasions more than 60 Microtus passages were recorded at a single point in 24 hours. Harvest mouse. — Harvest mice were almost entirely nocturnal (Fig. 3, center). They not only used the Microtus runways, but their passages were frequently intermixed with those of Microtus ( Fig. 2 ) . On fourteen occasions the two species passed within 60 seconds of each other, and on one occasion -■ r-i ' 1 ' — I • — ^1 — 1 ■ r— 1 1 ni- o MB M M MM M MM MM R M M MM MM MM RMRR 2 566 55 765BJ59 '' ^ %l ^ K IIV, 9 a i a (J(J 9cJc;9(J- O z UJ §5 - MEADOW MICE ALL MONTHS N = 6077 - UJ U. 6 7 8 9 10 II 12 I NOON 234567 89 lO II 12 I 23456 NIGHT Sic >- o z UJ cr u. HARVEST MICE ALL MONTHS N = 1753 6 7 8 9 10 II 12 I 2 3 4 5 6 7 8 9 10 II 12 I NOON NIGHT 2 3 4 5 6 20 > o I 10 o or BRUSH RABBIT N= 94 m 6 7 8 9 10 II 12 I 2 3 4 5 6 7 8 9 10 II 12 I NOON NIGHT 2 3 4 5 6 Fig. 3. — Distribution by hours of 6,077 passages of meadow mice (above); 1,753 passages of harvest mice (center); and 94 passages of brush rabbits (below). 355 May, 1959 PEARSON— TRAFFIC SURVEY 177 in the runway and it began to look unused. By the end of April almost all traffic had ceased. The Reithrodontomys data will be analyzed in a later report. Birds. — Of the 382 bird records, at least 255 were of sparrows ( at least 122 song sparrow; the remainder mostly fox sparrow, white-crowned sparrow and ■golden-crowned sparrow ) . Other birds recognized were wren-tit, wren, brown towhee and thrush. On several occasions birds, especially song sparrows, battled their reflections in the window of the instrument shelter. This caused long series of exposures. Each series was counted as a single passage. If the bird stopped for a minute or more and then returned to the battle, this was counted as another passage. All bird records were during daylight hours. On three occasions a sparrow and an adult Microtus appeared in the same photograph. On one of these occurrences a song sparrow was battling its reflection when an adult, lactating Microtus came along the runway. The sparrow retreated about 12 inches toward the camera shelter and, as soon as the mouse had passed, returned to the runway. Brush rabbit. — All except four of the records of brush rabbits were in June and July of 1957, a season when these animals, especially young ones, were abundant. Figure 3 (below) shows that they were most active in the early morning. -^20- o ^ 10 ■=> a tu CL jh 30 >20 o z UJ 3 S 10 cr 6 8 10 12 2 4 6 NOON 8 10 12 2 4 NIGHT JFMAMJJASOND Fig. 4. — Distribution by hours of 56 passages of shrews (left) and distribution by months of 56 passages of shrews (right). Shrew. — The dry, weedy habitat chosen was not favorable for shrews, and they were near the minimum weight necessary to depress the treadle of one of the recorders, so that some may have passed along the runway without making a record. The shrews were highly nocturnal (Fig. 4, left) and avoided the surface runways during the dry summer months (Fig. 4, right). Since captive specimens of Sorex are rarely inactive for more than one hour ( Morrison, Amer. Midi. Nat., 57: 493, 1957), the scarcity of records in the daytime probably means only that the shrews were not moving above ground at this time. They may have been foraging along gopher, mole and Microtus tunnels during the daytime. 356 178 JOURNAL OF MAMMALOGY Vol. 40, No. 2 A shrew was marked on March 4, a few inches from one of the recorders. It was captured 15 feet away on May 30 and 5 feet farther away on June 23. It passed along the study runway five times in the 16-week interval between first and last capture: on March 13, 27, 31, and April 17, and possibly on April 10 (markings obscured). Another shrew was recorded on March 27. Unless baited traps attract shrews from a considerable distance, or the recorder repels them, a trapper setting traps in this runway for a few nights would have had small chance of recording the presence of this individual which apparently was nearby for at least 16 weeks. Not a single shrew was recorded during the dry summer months of June, July and August. Nevertheless, on July 8 when I was checking the photo- electric recorder at 5:55 am, a shrew emerged completely from a small hole in 100 90 80 70 60 1 50 Q i X UJ 40 > UJ a: 30 20 10 • : SHREWS O = LIZARDS 50 60 TEMPERATURE CF) Fig. 5. — A comparison of the temperatures and humidities encountered by shrews and fence Hzards in tlie runways. The larger circles show the position of the mean for each species. The large polygon encloses the range of temperatures and humidities available to the animals during the study. 357 May, 1959 PEARSON— TRAFFIC SURVEY 179 the ground a few inches from the instrument shelter, twitched his nose rapidly for a few seconds, and retreated down the same hole. The air temperature was 54° and the relative humidity 78 per cent — normal for this season. Obviously shrews were present on the study area during some or all of the summer months but were not frequenting the surface runways. Figure 5 shows the temperatures and relative humidities encountered above ground by the shrews on the study area compared with the total range of temperatures and humidities recorded throughout the study. By their nocturnal, winter-time activity shrews encountered the coldest, most humid conditions available in the region. In contrast, the similarly small, insectivorous fence lizards existing in the same habitat managed by their own behavioral patterns to encounter a totally different climate (Fig. 5). The mean of the temperatures recorded at the times of lizard passages was 39° warmer than that recorded for shrew passages, and relative humidity was 36 per cent lower. This activity pattern of shrews differs from that reported by Clothier (Jour. Mamm., 36: 214-226, 1955) for Sorex vagrans in Montana. He found shrews there to be active "both day and night and throughout the year, even during extremely bad weather." It is important to understand, however, that he collected in damp areas near water, where the shrews may not have had to modify their activity to avoid desiccation. Extremely bad weather, for a shrew, is hot dry weather. Peromyscus. — Peromyscus truei was abundant in brushy places and in houses nearby; P. maniculatus was scarce. Some of the Peromyscus records were clearly of truei and some may have been of maniculatus, but many could not be identified with certainty. No adult P. californicus was recognized although a few young ones may have passed and been listed as truei. All passages of Peromyscus were at night. Salamander. — The record includes passages by both Etisatirui escholtzii and Aneides luguhris. They were recorded in October, November, March and April. By being nocturnal and by avoiding the dry season, they encountered in these autumn and spring months about the same microclimate as shrews, but were recorded neither in the winter months nor at temperatures below 39°. A third species, Batrachoccps attenuatus, was common in the study area but is so small that it could not be expected to actuate either of the recorders. One Batrachoccps electrocuted itself underneath the treadle but has not been included in the records. Comparison of traps and recorders. — The combination of live-trapping and photographing revealed a failure of small mammals to move between runways only a few feet apart. On several occasions meadow mice and harvest mice were live-trapped a few feet from one of the recorders, were released at the same place, and were recaptured a week or more later not more than a few feet away, yet during the intervening time they failed to pass the recorder. Conversely, some individual mice repeatedly recorded themselves on the films yet could never be induced to enter any of a large number of live traps placed 358 180 JOURNAL OF MAMMALOGY Vol. 40, No. 2 in the same runway and in nearby runways. It is obvious that all mice present do not use all of the active runways close to their home, and it is also obvious that neither the recorders nor traps give a complete accounting of the mice present. SUMMARY A motion-picture camera synchronized to an electronic flash unit was used to record the passage of animals along meadow-mouse runways and to record the temperature, relative humidity and time at which they passed. More than 26 species used the runways during 111 weeks of recording. Meadow mice, harvest mice, sparrows, brush rabbits and shrews passed most frequently. The average traffic per day in each runway was 11 passages; on some days there were more than 60 passages. Rarely more than ten meadow mice or six harvest mice used a runway in any one period. Meadow mice and harvest mice used the same runways simultaneously. Traffic by harvest mice alone did not keep the run- ways open. Meadow mice were active during the day and night; harvest mice were strongly nocturnal. Brush rabbits were active primarily early in the morning. Almost all shrews were recorded at night and in the winter months. Consequently, they encountered the coldest, most humid conditions available to them. In contrast, the similarly small, insectivorous fence lizards encountered a microclimate that was 39° warmer and 36 per cent less humid. Neither traps nor recorders accounted for all the individuals living nearby. Museum of Vertebrate Zoology, Berkeley, California. Received October 29, 1957. 359 PREY SELECTION AND HUNTING BEHAVIOR OF THE AFRICAN WILD DOG' RICHARD D. ESTES, Division of Biological Sciences, Cornell University, Ithaca, Nevv York JOHN GODDARD, Game Biologist, Ngorongoro Conservation Area, Tanzania Abstract: African wild dog (Lijcaon picttis) predation was observed in Ngorongoro Crater, Tanzania, between September, 1964, and July, 1965, when packs were in residence. The original pack of 21 dogs remained only 4 months, but 7 and then 6 members of the group reappeared in the Crater at irregular intervals. The ratio of males: females was disproportionately high, and the single bitch in the small pack had a litter of 9 in which there was only one female. The pack functions primarily as a hunting unit, cooperating closely in kilhng and mutual defense, subordinating individual to group activity, with strong discipline during the chase and unusually amicable relations between members. A regular leader se- lected and ran down the prey, but there was no other sign of a rank hierarchy. Fights are very rare. A Greeting ceremony based on infantile begging functions to promote pack harmony, and appeasement behavior substitutes for aggression when dogs are competing over meat. Wild dogs hunt primarily by sight and by daylight. The pack often approaches herds of prey within several hundred yards, but the particular quarry is selected only after the chase begins. They do not run in relays as commonly sup- posed. The leader can overtake the fleetest game usually within 2 miles. While the others lag behind, one or two dogs maintain intervals of 100 yards or more behind the leader, in positions to intercept the quarry if it circles or begins to dodge. As soon as small prey is caught, the pack pulls it apart; large game is worried from the rear until it falls from exhaustion and shock. Of 50 kills observed, Thomson's gazelles (Gazella thomsonii) made up 54 percent, newborn and juvenile wildebeest {Connochaetes taiirmus) 36 percent. Grant's gazelles (Gazella granti) 8 percent, and kongoni (Alcelaphus buselaphus cokei) 2 percent. The dogs hunted regularly in early morning and late afternoon, with a success rate per chase of over 85 percent and a mean time of only 25 minutes between starting an activity cycle to capturing prey. Both large and small packs generally killed in each hunting cycle, so large packs make more efficient use of their prey resource. Reactions of prey species depend on the behavior of the wild dogs, and disturbance to game was far less than has been represented. Adult wildebeest and zebra (Equus burchelli) showed little fear of the dogs. Territorial male Thomson's gazelles, which made up 67 percent of the kills of this species, and females with concealed fawns, were most vulnerable. The spotted hyena (Crocuta crocuta) is a serious competitor capable of driving small packs from their kills. A minimum of 4-6 dogs is needed to function effectively as a pack. It is concluded that the wild dog is not the most wantonly destructive and disruptive African predator, that it is an interesting, valuable species now possibly endangered, and should be strictly protected, particularly where the small and medium-sized antelopes have increased at an alarming rate. The habits of the African wild dog or Cape hunting dog (Lijcaon pictus) have been described, sometimes luridly, in most books about African wildlife. Accounts by such famous hunters and naturalists as Selous (1881), Vaughan-Kirby (1899), and Percival (1924), repeated and embellished by other authors, have created the popular image of a wanton killer, more destructive and disruptive to game than any other African predator. ^ Field work supported by the National Geo- graphic Society; also by grants from the New York Explorers Club and the Tanzania Ministry of Agriculture, Forests and Wild Life. 52 Because of its bad reputation, the wild dog was relentlessly destroyed in African parks and game reserves for many years. In Kruger National Park, for instance, it was shot on sight from early in the present century up until 1930 as part of an overall policy to keep predators down. In Rhode- sia's Wankie National Park some 300 wild dogs were killed by gun and poison be- tween 1930 and 1958. Acceptance of modem concepts of wild- life management has finally brought an end to the indiscriminate destruction of wild dogs and other predators in most, if not all, African national parks. There is now 360 The African Wild Dog • Estes and Goddard 53 a general awareness among game wardens in sanctuaries, paid very little attention to of the predator's role in regulating popula- cars and could be watched undisturbed tions, which perhaps began with Stevenson- from within 30 yards or less. It was also Hamilton ( 1947 ) , Warden of Kruger Park feasible to keep pace with the pack during for almost 30 years, who related the alarm- chases over the central Crater floor, either ing increase of impala {Aepijceros me- driving parallel to the leader at a distance lampus) to the disappearance of the park's of 100-200 yards or following behind and formerly large wild dog packs. to one side so as not to get in the way of While the wild dog has benefited from other pack members. To locate the pack more enlightened concepts of game man- initially, we often drove to an observation agement, its reputation, still based on pop- point on a hill and scanned the Crater with ular writings and myth, remains unchanged, binoculars and a 20-power binocular tele- I But recent scientific investigations indicate scope. When the pack was moving it could that a new and less-prejudiced evaluation often be spotted at a distance of over 5 of this species is long overdue. Kiihme miles, and a number of chases and kills ( 1965 ) has studied the social behavior and were clearly observed from a hilltop family life at the den of a pack with young through the telescope. whelps on the Serengeti Plains, Tanzania. We have observed prey selection and hunt- RESULTS ing behavior in a free-ranging pack of _ , _ J ,^ J . 1 • V XT Pack Composition adults and juveniles m nearby Ngorongoro ^ Crater, a caldera with a floor area of 104 The pack that first entered Ngorongoro square miles that supports a resident pop- Crater in September, 1964, contained 21 ulation of around 25,000 common plains animals, including 8 adult males, 4 adult herbivores. The two studies together throw females, and 9 juveniles. They remained quite a different light on the habits, char- more or less continually in residence through acter, and predator-prey relationships of December, then disappeared and were pre- this highly interesting species. sumed to have left the Crater. One juvenile We are indebted to Dr. B. Foster of the female had died of unknown causes. Dur- Royal College, Nairobi, and to G. C. ing January, 1965, seven members of the Roberts of the Crater Lodge for reporting same pack, 4 males, 1 female, and 2 juvenile four kills and one kill respectively, that males, reappeared; after a lapse of 5 they witnessed in the Crater; also to Pro- months, apparently the same animals, minus fessors W. C. Dilger, O. H. Hewitt, and one male, again took up residence, and H. E. Evans of Cornell University for criti- have been observed off and on up to the cal readings of the manuscript. Nomencla- present writing. In March, 1966, the female ture follows Haltenorth (1963) for artio- whelped but died 5 weeks later, leaving dactyls, and Mackworth-Praed and Grant 8 male and 1 female pups. They were (1957) for birds. brought up by the 5 males, who fed them by regurgitation until they were old enough MtTHODS j-Q j.yj^ \Ni\h the pack. However, the female Most observations were made from a and 4 male pups died, leaving an all-male vehicle; we each had a Land Rover and pack of 9 in August, 1966. usually operated independently. The wild While an all-male pack must be excep- dogs, like many other African predators tional, there is other evidence to suggest 361 54 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 that a high proportion of males may be observed sexual behavior on only two oc- common in this species. The pack Kiihme casions, when one male mounted another studied consisted of 6 adult males and 2 repeatedly as the latter was feeding at a adult females, which had 11 and 4 pups kill. Kiihme also saw very little sexual be- respectively, sex unreported. During 2-3 havior. When two animals were competing years of shooting in Kruger National Park, for the same piece of meat, each would try the ratio of males was 6 : 4, despite an to burrow beneath the other, its forequarters attempt to select females (Stevenson-Hamil- and head flat to the ground and hind- ton 1947). We have no explanation to offer quarters raised, tail arched and sometimes for the discrepancy, but if it is real and wagging. The ears were flattened to the not normal, it might help explain the head and the lips drawn back in a "grin," reported decline of wild dogs during recent while each gave excited twittering calls. As years in many parts of Africa. Kiihme observed (p. 516), the dogs "tried to outdo each other in submissiveness." Social Organization In this way juveniles and even subadults Leadership and Rank Hierarchy. — In the manage to monopolize kills in competition full pack of 21 and in the pack of 7, the with adults. The young thus enjoy a priv- same adult male was consistently the ileged position in the pack. Pups at the leader; he usually led the pack on the hunt, den successfully solicit any adult to regur- selected the prey, and ran it down. In the gitate food by poking their noses into the pack of 6, from which the above male was corner of the adult's mouth, sometimes lick- absent, the adult female was the leader, ing and even biting at the lips. Since all One of the males filled the position after pack members contribute to feeding and her death. protection of the young, the mother is not Apart from the position of leader, we saw essential to their survival after the first few no indication of a rank order. Kiihme con- weeks. eluded there was no hierarchy in the pack Greeting Ceremony. — Whenever the pack he observed, nor even a leader. The equality became active after a rest period, and of pack members may partly explain the particularly if two parts of the pack were singularly amicable relations typical of the reunited after being separated, the mem- species. On the other hand, competition for bers engaged in a Greeting ceremony ( Fig. food and females could easily lead to ag- 1), in which face-licking and poking the gression; yet neither Kiihme nor we ever nose into the corner of the mouth played a saw a fight. prominent part. The ceremony thus ap- Food Solicitation and Appeasement Be- f)ears to be ritualized food solicitation; the havior. — Overt aggression and fighting are fact that Kiihme actually saw regurgitation minimized through ritualized appeasement elicited by begging adults supports this in- behavior derived from infantile food beg- terpretation. The Greeting ceremony in ging. Begging and appeasement appear in the wolf (Canis lupus), in which one takes almost every contact between individuals, another's face in its jaws, may have the and particularly in situations where aggres- same derivation. sion would be most likely to occur — for As a prelude to greeting, dogs typically instance, when animals are competing over adopted the Stalking attitude (Fig. 2), with a kill. However, we cannot comment on the head and neck held horizontally, shoul- sexual competition, having seen none; we ders and back hunched, and the tail usually 362 The African Wild Dog • Estes and Goddard 55 Fig. 1. Greeting ceremony. hanging. Kiihme (p. 512) interprets this posture as inhibited aggression; the same attitude is adopted when approaching po- tential prey and competitors of other species. The Stalking posture changed to greeting when dogs got close. In greeting- solicitation, as they licked each other's lips and poked the nose into the corner of the mouth, one or both crouched low, with head, rump, and tail raised stiffly ( Fig. 1 ) . Except for the raised head, this resembles the submissive posture displayed when two dogs are competing over food. The Greet- ing ceremony was also frequently per- formed while two dogs trotted or ran side by side. Vocal Communication. — Although Perci- val (1924), Stevenson-Hamilton (1947), Ma- berly (1962), Kiihme (1965), and others have given good descriptions of wild dog calls, the function of the calls has often been misinterpreted. This applies partic- ularly to two of the three most frequently heard calls (Nos. 1 and 3): 1. Contact call — a repeated, bell-like "hoo." Often called the Hunting call, it has nothing to do with hunting as such, but is given only when members of a pack are separated. Though a soft and musical sound, it carries well for 2 or more miles. When members of the Ngorongoro pack- were missing, an imitation of the Contact call would bring the rest to their feet, whereas there was at best onlv a mild reac- Fig. 2. The Stalking attitude, here displayed by the pack leader while approaching a herd of gazelles. tion to imitations when the full pack was assembled. 2. Alarm hark — a deep, gruff bark, often combined with growling, given when star- tled or frightened. A good imitation near a resting pack elicited an immediate star- tled reaction. 3. Twittering — a high-pitched, birdlike twitter or chatter. The most characteristic and unusual vocalization, it expresses a high level of excitement. It is given in the prelude to the hunt, while making a kill, in mobbing hyenas or a pack member, and by dogs competing over food. Its primary function is evidently to stimulate and con- cert pack action. Kiihme described this call (Schnattern) only in the context of the Greeting ceremony (p. 513). Besides these vocalizations, whining may be heard during appeasement behavior and when pups are begging, and members of the pack sometimes yelp like hounds when close on the heels of their prey. Kiihme (p. 500) further distinguishes an Enticing call (Locken) given by adults calling the young out of the den, and a Lamenting call (Klage) given by pups when deserted. Olfactory and Visual Communication. — Wild dogs hunt primarily by sight and by daylight. We never saw them track prey by scent. Though they evidently have a good nose and may well use it for tracking in bush country, olfaction in this species seems to have a primarily intraspecific significance. 363 ^cn 56 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 \^ regular hunting cycle is the likeUest ex- planation; it was more usual, however, for the pack then to wait until the following regular period. Wild dogs will also hunt on moonlight nights, as Stevenson-Hamilton noted. When the Crater dogs had not killed before dusk, the hunt was sometimes pro- longed. The latest kill we recorded was at 7:32 PM, when it was fully dark. Since they are capable of functioning as a pack and of hunting successfully after dark, the fact that wild dogs are so strongly diurnal may seem puzzling. But it may be Fig. 3. Time distribution of 50 wild dog kills. explained by the fact that they hunt mainly 1 1 CO CO _j 1 1 ^ _i U- o L o 1 ■ .0. 1 I ■ Ji J 6 li- 7'8 ' 12 ' 1 ' 2 ' 3 ' 4 5 PM^ TIME by sight; it would be much more difficult Wild dogs are renowned for their peculiarly to locate prey and single out a quarry at strong, and to many humans disgusting, night. As to the regularity and brevity of odor, which may emanate from anal glands their hunting cycles in early morning and but seems to come from the whole body, late afternoon, this is partly a measure of Sniffing under the tail, responsive urination their hunting efficiency, discussed below, and defecation are socially important activ- Also, of course, these are the times in the ities. But the main role of the strong body day when diurnal animals, particularly odor may be to permit high-speed tracking herbivores, are most active and most ap- of the pack by members that have lost proachable. visual contact. Lagging members seen run- Apart from a certain amount of play and ning on the track taken by the rest of the other social activities shortly before starting pack sometimes appeared to be using their to hunt and immediately after feeding, noses. Similarly, the white tail tip probably pack members were usually active only helps maintain visual contact in bush coun- while actually hunting. At other times they try, high grass, and under crepuscular con- could often be found resting near or in the ditions; in a species notable for every pos- same place where they had settled after the sible color variation, a white-tipped tail is morning or evening kill. When resting, the most constant and conspicuous mark. pack members customarily lay touching in close groups (Fig. 4). Generally speaking, Daily Activity Pattern ^^xe pack became active between 5:30 and The Ngorongoro pack had two well-de- 6:15 pm, and in the morning within Mj fined hunting periods each day (Fig. 3). hour of dawn, remaining active for 1-2 That this periodicity is characteristic of the hoiu-s. But where game is less plentiful species may be inferred from Kiihme's ob- than it is in Ngorongoro, and a pack must servations (p. 511), and from Stevenson- range more widely (Stevenson-Hamilton Hamilton's (1947) observations in Kruger gives a range of at least 1,500 square miles National Park. In nine recorded instances, for a Transvaal pack whose movements though, the Ngorongoro dogs killed between were reported over a period of years ) , a 8:30 AM and 3:30 pm, well outside the nor- good deal of time between and during mal periods. Failure to kill during the hunts must be spent in travel. 364 The African Wild Dog • Estes and Goddard 57 Fig. 4. Part of a resting pack, lying typically close together. Hunting Behavior Prelude to the Hunt. — Periods of activity were initiated by the actions of one or a few dogs apparently more restless than the others; rarely did the whole pack arise spontaneously at the start of an activity- cycle. Typically, one dog would get up and run to a nearby group, nose the others and tumble among them until they re- sponded. Within a few minutes the whole pack would usually become active. But if, as sometimes happened, the majority failed to respond to the urging of a few, then all would settle down to rest again. Sometimes, after a brief bout of general activity, the whole pack would lie down once more, even if it was past the usual time of hunt- ing. During the first 5 or 10 minutes after rousing, the pack members sniffed, urinated, defecated, greeted, and romped together. Play and chasing tended to become pro- gressively wilder and reached a climax when the whole pack milled together in a circle and gave the twittering call in unison. As soon as this melee broke up, the pack usually set off on the hunt. Kiihme (p. 522) interprets this performance (specifically the Greeting ceremony) as "a daily re- peated final rehearsal for the behavior at the kill," wherein mutual dependence and friendliness are reinforced by symbolic beg- ging, thus enabling the dogs to share the kill amicably, ^^'hile this may be one func- tion, the progressive buildup of excitement before hunting looked to us like nothing so much as a "pep rally," that served to bring the whole pack to hunting pitch. The be- havior of domestic dogs urging one another to set off on a chase is somewhat similar. The Mobbing Response. — During the mill- ing preparatory to hunting, we sometimes saw what appeared to be incipient mob- bing action toward a pack member, when up to half a dozen dogs would gang up on one, tumble and roll it but without actually biting it. Intensive play between two or three animals usually preceded and seemed to trigger a mobbing reaction in other mem- bers, who signaled their intentions by ap- proaching in the Stalking posture. Percival (1924:48) reports seeing a pack mob and kill a wild dog he had wounded. The oc- currence of "play" mobbing suggests that it could indeed become serious when an animal is maimed. On the other hand, sick and crippled pack members are often not molested: one very sick-looking old male in the large pack trailed behind the others for over a month before recovering, and though he kept usually a little apart, was tolerated at kills. It is significant that basically the same mobbing behavior, at high intensity, is dis- played when wild dogs kill large prey and when they harass spotted hyenas, their most serious competitor. It seems very likely, in fact, that mobbing is an innate re- sponse which governs pack action in hunt- ing, killing, and mutual defense. It is per- haps the key to pack behavior in all animals that display it. That mobbing appears in 365 58 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 play and can be released by a conspecific more than 300 yards was generally not which is wounded or otherwise transformed pursued. As far as we could tell, the prey from its normal self, supports the hypothesis animal was never singled out until after that it is an innate response. It is also note- the pack, or at any rate the leader(s), had worthy that a wild dog removed from its broken into a run. pack apparently makes little effort to defend In the pack of 21, juveniles and some itself against attack. Selous ( in Bryden adults usually lagged far behind, and often 1936:24) reported that a wild dog caught caught up 5-10 minutes after the kill was by a pack of hounds shammed death and made. In the small pack, however, corn- then escaped when he was about to skin it. monly all kept together and spread out on Hunting Technique. — Sometimes the pack a front during the stalking phase. When all would set off on the hunt at a run and started running on a front, sometimes more chase the first suitable prey that was than one dog picked out a quarry from the sighted. More often, there was an inter- fleeing herd, whereupon the pack might val of 10-20 minutes during which the dogs split, some following one dog, the rest trotted along, played together, and engaged another. Kiihme ( p. 527 ) considered this in individual exploratory activity, stopping the normal pattern and noted that often to sniff at a hole or a tuft of grass, then each animal acted for itself in selecting a running to catch up with the rest. At this quarry before all combined on a common stage, when the hunt had started but before goal. In this way the slowest prey tended any common objective had been deter- to be selected. Selection by this method mined, individuals might forage for them- was exceptional for the Ngorongoro pack, selves. The observer would suddenly notice which had a definite leader; as a rule the that a dog was carrying part of a gazelle lead dog made the choice and the rest of fawn or a young hare (Lepus capensis), the pack fell in behind him. Nor did it that must have been simply grabbed as it appear that any effort was made to single lay in concealment. Once during a moon- out the slowest prey, although that would light hunt by a small pack that visited the be difficult to observe clearly. Crater in 1963, individual dogs were seen Again as a general rule, no attempt was to pick up at least two gazelle fawns and made to carry out a concealed stalk, which one springhare {Pedetes surdaster), a would in any case be practically impossible strictly nocturnal rodent, within V2 hour, by daylight on the short-grass steppe. But Concealed small game such as this is ap- on one occasion the pack of six made use parently not hunted by the pack in concert, of a tall stand of grass to get near a group Preparatory to the chase, there was fre- of Thomson's gazelles. On another hunt the quently a preliminary stalking phase dur- pack apparently took advantage of a slight ing which the pack approached herds of elevation in the expectation of surprising game at a deliberate walk, in the Stalking any game that might be out of sight on the attitude (Fig. 2). The dogs appeared to be far side. They moved deliberately up the attempting to get as close as possible with- slope, then broke into a run and swept at out alarming the game, and certainly the full speed over the crest on a broad front flight distances were much less than when —but without finding any quarry that time, the pack appeared running. The chase was When the leader had selected one of a launched the moment the game broke into fleeing herd, he immediately set out to run flight. But game that began running at it down, usually backed up by one or two 366 The African Wild Dog • Estes and Goddard 59 other adults who maintained intervals of 100 yards or more behind him, but might be left much further behind in a long chase. The rest of the pack lagged up to a mile in the rear. Discipline during the chase was so remarkable among all pack members that even gazelles which bounded right be- tween them and the quarry were generally ignored. The average chase lasted 3-5 minutes and covered 1-2 miles. At top speed a wild dog can perhaps exceed 35 mph, and can sustain a pace of about 30 mph for several miles. Once when a chase had begun but no single quarry had yet been selected, a male in the pack of 21 broke away and proceeded to make a 5- mile circular sweep quite by itself, turning on bursts of speed when gazelles bounded off before him, but without ever singling one out. His average speed, as determined by pacing him in a vehicle, was approxi- mately 20 mph. In descriptions of wild dog hunting methods, much has been made of their intelligent cooperation in "cutting comers" on their prey, and particularly of their relay running, with fresh dogs taking the place of tired leaders. We concur that there is a basis for the first idea, but we saw no evi- dence whatever to support the contention that wild dogs run in relays. The truth is that wild dogs have no need to hunt in relays. The lead dog has ample endurance, if not the speed, to overtake probably any antelope, of which gazelles are among the fleetest. The fact that other members of the pack are able to cut comers on the prey is at least partly accounted for by the prey's tendency to circle instead of fleeing in a straight line. As explained later, some prey animals have a greater tendency than others of their species to do this. Of course, once overtaken, even a quarry that has been running straight is forced to start dodging if it is to avoid being caught straightaway. Thus a dog running not too far behind the leader is well placed to cut corners when the quarry changes course, and it frequently happened that one of the followers made the capture. Most game, after a hard chase of a mile or two, was too exhausted by the time it began dodging to have any real chance of evading its pursuers. Killing and Eating. — Wild dogs killed small game hke Thomson's gazelles with amazing dispatch. Once overtaken, a gazelle was either thrown to the ground or simply bowled over, whereupon all nearby dogs fell on it instantly. Grabbing it from all sides and pulling against one another so strongly that the body was suspended between them, they then literally tore it apart ( Fig. 5 ) . It happened so quickly that it was never possible to come up to a kill before the prey had been dismembered. If it didn't go down at once, dogs began tear- ing out chunks while it was still struggling on its feet. We once saw a three-quarter term fetus torn from a Thomson's gazelle within seconds of the time it was overtaken and before it went down. As Kiihme ob- served, there is no specific killing bite as in felids (Leyhausen 1965). When dealing with larger prey such as juvenile wildebeest and notably a female kongoni, the dogs slashed and tore at the hind legs, flanks, and belly — always from the rear and never from in front — until the animal fell from sheer exhaustion and shock. They then very often began eating it ahve while it was still sitting up (Fig. 6). Self-defense on the part of a prey was never once observed; the kongoni, for example, did little more than stand with head high while the dogs cut it to ribbons, looking less the victim than the witness of its own execution. In eating, the dogs began in the stomach cavity, after first opening up the belly, and proceeded from inside out. Entrance was 367 60 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 Fig. 5. The pack tearing apart a young gnu calf. also effected through the anus by animals unable to win a place in the stomach cavity. While several dogs forced their heads in- side and ripped out the internal organs, others quickly enlarged the opening in struggling for position. This resulted in skinning out the carcass, leaving the skin still attached to the head, which was sel- dom touched. Apart from these, the back- bone and the leg bones, very little of a Thomson's gazelle would remain at the end of 10 minutes. In the pack of 21, if only part had managed to eat their fill, sometimes the rest went off to hunt again before the carcass was cleaned. They pro- ceeded to chase and pull down another gazelle within as little as 5 minutes from the time of the previous kill, to be joined shortly by the other dogs. As each animal became satisfied it withdrew a little from the kill and joined others to rest, play, or gnaw at a bone it had taken along. Some- times the pack stayed at the scene until the next hunting period; more often it with- drew to a nearby stream or waterhole and settled down there. Kiihme never saw wild dogs drink. The Ngorongoro pack drank, though irregularly, before hunting and after eating. Selection of Prey and Frequency of Kills Table 1 summarizes prey selection by species, sex, and age in 50 recorded kills. The 11 wildebeest calves were all taken in January during the peak calving season, when the pack of seven dogs apparently specialized on them; only kills of calves were seen by us or reported by Crater visi- tors in this month. Thus the percentage of calves in the total gives a biased picture of prey selection during the rest of the year. With new calves excluded, the adjusted percentages, based on 39 kills, are as fol- lows: Thomson's gazelles 69 percent Juvenile wildebeest 18 percent Grant's gazelles 10 percent Kongoni one kill Wright (1960:9) records a similar pre- ponderance of Thomson's gazelles in 10 Fig. 6. Dogs begin eating a yearling-class gnu while it is still alive, but evidently in a state of deep shock. 368 The African Wild Dog • Estes and Goddard 61 Table 1. Prey selection by species, sex, and age in 50 kills of the African wild dog. Prey Species Total No. Adult Males Adclt Females JUVENILE- SUBADULT Young* Percent of Total Kills Thomson's gazelle 27 18 6 2 1 54 Wildebeest 18 0 0 7 11 36 Grant's gazelle 4 1 1 2 0 8 Kongoni 1 0 1 0 0 2 * Less than 6 months old. kills on the Serengeti Plains (7 Thomson's gazelles, 1 wildebeest, 1 impala, and 1 reed- buck [Redunca redwica]), and notes that it is the staple diet of wild dogs in the Serengeti. Kiihme also observed that wild dogs prey mainly on Gazella thomsonii and G. granti, and young wildebeest in the Serengeti. In terms of actual preference, informa- tion from the Serengeti, where the Thom- son's gazelle is by far the most numerous herbivore, is far less revealing than the fig- ures from the Crater, where this species oc- curs in relatively small numbers. The status of the principal ungulates in Ngorongoro, based on an aerial count by Turner and Watson ( 1964 ) , on two ground counts of the gazelles by the authors in collaboration with the Mweka College of Wildlife Man- agement, and on our ground counts of the less numerous species, is as follows: Wildebeest 14,000 Zebra 5,000 Thomson's gazelle 3,500 Grant's gazelle 1,500 Eland {Taurotragus oryx) 350 Waterbuck {Kobus defassa) 150 Kongoni 100 Reedbuck 100 (?) The evidence suggests, then, that Thom- son's gazelle is the preferred prey of the wild dog in East African steppe-savanna. In the miambo woodland (Brown 1965) that extends from mid-Tanzania into South Africa, where gazelles are not found, the main prey may be impala, followed by other medium- to small-sized antelopes and the young of large antelopes. In Kruger Park, for example, of 88 identified wild dog kills, 85 percent were impala (Bourliere 1963). Stevenson-Hamilton Hsted other prey as reedbuck, bushbuck ( Tragelaphus scrip- tiis), duiker (Sijlvicapra grimmia and Cephal- ophus spp.), and steinbok (Raphicerus campestris), also female waterbuck and kudu {Tragelaphus strepsiceros) when pressed by hunger. In Wankie Park, war- dens' reports indicate a considerable toll of young kudu, eland, sable (Hippotragus niger), and tsessebe (Damaliscus lunatus). Instances where adult female and even adult male kudu were pulled down by wild dogs are also cited. Bourliere states (1963:21) that "Carnivores actually only prey upon herbivores of about the same size and weight." While this gen- eralization is open to dispute, it applies well enough to East African wild dogs preying on Thomson's gazelles. Where the main prey is impala, reedbuck, etc. that weigh in the 100-150 lb class, weight and size may be double or triple that of the wild dog. But the wild dogs of the East African steppe-savanna are smaller (also darker, with more black and less tan and white) than their counterparts in Central and South African woodland (Fig. 7). The average weight of the animals we have seen in East Africa would not exceed 40 lb; the members of a pack seen in Wankie Park, by comparison, looked to be a good 3 inches taller and 20 lb heavier. This consistent geographic size variation may be adapted to size of the principal prey species; specifi- 369 62 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 Fig. 7. Two specimens of the larger, lighter-colored wild dog of southern Africa, photographed in Wankie National Park. cally, wild dogs of the East African plains may be smaller as the result of specializa- tion on Thomson's gazelle. Kill Frequency. — Because of the difficulty of locating and relocating a free-ranging pack, our data for consecutive hunting periods are inadequate for defining the aver- age kill frequency and average food intake per animal per day. Even when the pack was observed during the two daily hunting periods, it was rarely certain that it had not killed before, after, or between these periods. Nonetheless, because this type of information is badly needed, data covering consecutive hunting periods are presented in Table 2 as a rough average of kill fre- quency and meat available per animal per day. The average frequency of two kills per day derived from the data for consecutive hunting periods agrees with our general ob- servation that the pack usually killed dur- ing each period. To demonstrate this, on 28 hunts the pack performed as follows: Chases 29 Kills 25 Failures 4 Did not chase 5 This indicates a success rate per chase of over 85 percent. As a further indication of efficiency in locating and running down prey where game is plentiful, on eight oc- casions when the dogs were watched from the moment they left their resting place to the moment they killed, the mean time was only 25 minutes, with a range of 15-45 min- utes. On five other occasions the pack failed to hunt seriously during the normal period; this was offset in the above figures by five periods during which the dogs chased and killed twice. The possibility that hunting activity and success might be reduced after having killed larger or more than one of the usual prey is not borne out by the six instances when the pack was ob- served during the next hunting period: in four cases the pack killed again. There are some grounds for asserting, then, that wild dogs kill twice daily regardless of what their prey may be. Certainly they do not feed more than once from the same kill, at least not in Ngorongoro Crater, where the numerous scavengers dispose of all left- overs in very short order. Meat Available per Animal per Day. — The amount of meat available per wild dog per day works out at roughly 6 lb, assuming that 40 percent of the prey animal consists of inedible or unpalatable bone, skin, and stomach contents. Wright's ( 1960 ) calcula- tion of 0.15 lb of food per day per pound 370 The African Wild Dog • Estes and Goddard 63 Table 2. Kill frequency and meat available per dog per day, based on observations of consecutive hunting cycles. Avail- Est. Wt. No. IN able* Meat/ Date Prey (LB) Pack Meat/ Dog Dog/Day 1964 Sept. 30 Oct. 1 Nov. 11 Nov. 12 Nov. 27 Nov. 28 Dec. 5 (pm) to Dec. 7 (aai) 1965 Jan. 17 (pm) to Jan. 19 (am) July 16 Juvenile wildebeest Thomson's gazelle (adult M) M ., (adult F) Thomson's gazelle (adult F, including fetus) 2 Thomson's gazelles (adult F) Thomson's gazelle (adult M) Kongoni (adult F) 2 Thomson's gazelles ( adult M ) Thomson's gazelle (subadult M) Grant's gazelle (subadult F) Thomson's gazelle ( adult M ) 2 Thomson's gazelles ( adult M ) Thomson's gazelle (juv. M) 4 wildebeest calves 2 Thomson's gazelles ( adult M ) 125 21 3.6 60 21 1.7 40 21 1.1 50 21 1.4 80 21 2.2 60 21 1.7 250 21 7.4 120 21 3.4 50 21 1.4 90 21 2.6 60 21 1.7 120 12 6.0 40 180 120 21 6 1.1 15.5 12.0 3.6 2.8 3.6 9.1 4.8 4.3 3.5 7.8 12.0 Kill frequency = 2 kills/day. Meat available per dog per day: combined average = 6 lb; for pack of 21 = 4.5 lb; for pack of 7-6 9 lb * A\ailable meat is based on 60 percent of carcass weight. of dog also works out to 6 lb per day if the average weight of a dog is taken as 40 lb, but his figures are based on the total weight of the prey. In either case, two to three times as much food per day is avail- able to wild dogs as is given to domesti- cated dogs of the same size. However, the number of dogs in the pack is an important factor. When there were 21 dogs, the amount of meat available per day was less than 5 lb per animal; in the pack of 7 and 6, each animal had approximately twice as much available meat. Since the small packs killed at the same rate, large packs are un- doubtedly less wasteful. Reactions of Prey Species The reactions of game depended on the behavior of the wild dog pack. When the pack was at rest, all game would graze un- concernedly within 150 yards. When the dogs were walking or trotting, potential prey would stand until approached within 350-250 yards, or less if the pack was not headed directly toward them. When stalked, gazelles often stood watching until the pack came within 3(X)-2(X) yards. But when the pack was running, gazelles, and wildebeest herds containing young, often acted alarmed at a distance of 500 yards, although again, individual animals not directly in the ap- proach line might let the pack go by as close as 150 yards. Gazelles. — The moment a running wild dog pack appeared on the plain, both ga- zelle species immediately reacted by per- forming the stiff-legged bounding display, with tail raised and white rump patch flash- ing, called Stotting or Pronking. Un- doubtedly a warning signal, it spread wave- like in advance of the pack. Apparently in response to the Stotting, practically every 371 64 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 gazelle in sight fled the immediate vicinity, cially territorial bulls, show little fear of Adaptive as the warning display may wild dogs, which is a good indication that seem, it nonetheless appears to have its they have little reason to fear them under drawbacks; for even after being singled out normal circumstances (Fig. 8). While even by the pack, every gazelle began the run territorial males will get out of the way of for its life by Stotting, and appeared to lose a running pack, they rarely leave their precious ground in the process. Many have grounds, but merely trot to one side and argued that the Stotting gait is nearly or turn to stare as the pack goes by. Bulls not quite as fast as a gallop, at any rate decep- infrequentiy act aggressively toward walk- tively slow. But time and again we have ing or trotting dogs, and may even make a watched the lead dog closing the gap until short charge if the dogs give ground. In the quarry settled to its full running gait, Rhodesia we have seen a pack of the larger when it was capable of making slightly bet- variety of wild dogs chased by females and ter speed than its pursuer for the first half yearlings of the blue wildebeest (C. t. mile or so. It is therefore hard to see any taurinus) which is also larger and perhaps advantage to the individual in Stotting when generally more aggressive than the Western chased, since individuals that made no dis- wliite-bearded gnu. But like zebras, all play at all might be thought to have a bet- wildebeest will on occasion follow behind ter chance of surviving and reproducing, walking or trotting dogs, apparently moti- On theoretical grounds, then, it has to be vated by curiosity, just as they will gather assumed that the Stotting display offers an to stare at and follow lions ( Panthera leo ) . individual selective advantage which simply In hunting wildebeest, wild dogs are ob- remains to be determined. Nor is this type viously highly selective. Having walked in of display confined to the gazelles: during the Stalking attitude to within several hun- the aforementioned kongoni chase, all six dred yards or less and then run into the members of the herd began Stotting when midst of a large concentration, the pack the wild dog pack first headed in their di- sphts up and works through it, approaching rection, and the victim continued to Stot for one gnu after another only to turn away if some time after being singled out. it proves adult. Meanwhile the wildebeest Table 1 shows that 67 percent of the mill and run in all directions, without ever Thomson's gazelles killed were adult males, making any effort to form a defensive ring This is evidently the result of territorial be- — even when young calves are present. A havior. Because of attachment to territory, defensive ring has been reported in some probably coupled with inhibition about of the wild dog literature. Kiihme (p. 528) trespassing on the grounds of neighboring observed something of the sort in large rivals, territorial males tend to be the last Serengeti concentrations, though they did to flee from danger. Moreover they show not form any regular ring but simply a greater tendency to circle back toward crowded together in a milling mass. Indi- home, and these two traits together make vidual females, on the other hand, defend them more vulnerable to wild dog preda- their calves after being overtaken in flight, tion than other members of the population. Against a pack, however, one wildebeest The same tendencies are displayed by fe- cannot put up any effective defense; while males with young, concealed fawns, making it confronts one or two, the rest go around them also more vulnerable. and seize the calf. Wildebeest. — Adult wildebeest, espe- Zebra. — The only other herbivore whose 372 The African Wild Dog • Estes and Goddard 65 Fig. 8. Adults, and even a yearling gnu (4th from left) show little fear of running wild dogs, though they ran out of the way immediately after the picture was taken. The quarry is a young calf, visible as a light spot in the upper left. reactions to wild dogs we observed in de- j tail, zebras are the least concerned about them, and do not hesitate to attack dogs that come too close. Wild dogs on their part rarely stand up to them. Since the members of a harem would probably co- operate with the herd stallion to defend the foals, it would appear that wild-dog preda- tion on zebra is quite rare. Relations with Other Predators and Scavengers Vultures. — Since wild dogs customarily kill in early morning and late afternoon, the larger vultures, the white-backed (Pseu- dogyps africanus) , Riippells griffon (Gyps riippelJi), and lappet-faced (Torgos trache- liotus), whose activities are largely regu- lated by the presence or absence of thermal updrafts, benefit rather little from their pre- dation. Large vultures were more likely to appear at afternoon than morning kills. But the two smallest species, the hooded and Egyptian vultures (Necrosyrtes motwchus and Neophron perenopterus) , were regu- larly to be found at wild-dog kills, a good hour before other scavengers were even air- borne. In addition to these vultures, other regularly encountered scavengers included the tawny eagle (Aquila rapax) and the kite [MUxjus migrans), while the uncom- mon white-headed vulture (Trigonoceps occipitalis), the bateleur eagle (Terathopius ecaudatus), and Cape rook (Corvus capen- sis) showed up infrequently. On several occasions hooded vultures were seen following a chase and landing before the prey had even been pulled down, shortly after full daylight. Aside from glean- ing bits and pieces around the kill, vultures had to wait until the dogs left before they could feed on the carcass. But the kite suc- cessfully stole small pieces from the dogs by swooping, grabbing, and mounting again to eat on the wing. Although young ani- mals sometimes stalked and ran at vultures 373 66 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 that approached close to the kill, the dogs stools they found, and coming dangerously were generally tolerant toward avian scav- close, to stand staring with their short tails engers. twitching — a sign of nervousness. Kiihme Jackals. — The Asiatic jackal (Canis (p. 534) reports an instance in which a aureus) was seen more frequently at kills hyena even touched a resting wild dog's than the black-backed jackal (C. meso- face, meanwhile "whining friendly." Such melas). Since the latter seemed to predomi- boldness, particularly near the time when nate at nocturnal kills by lions or hyenas, the pack was becoming active, often trig- it may be that one is more nocturnal and gered the Mobbing response, one more diurnal in its habits. Also, the Hyenas, which weigh up to 150 lb, would Asiatic jackal tended to behave more boldly be more than a match for wild dogs if they and aggressively at kills. It would move had the same pack (mobbing) instinct, closer to a feeding pack of dogs and take Lacking it, they are nearly defenseless advantage of any opportunity to steal meat, against a wild dog pack. With three to a When threatened by a dog, a little 15-lb dozen dogs worrying its hindquarters, the jackal, coat fluffed, head down, and snarl- best a hyena can do is to squat down and ing, would stand its ground and snap fero- snap ineffectively over its shoulder, while ciously if the dog continued to advance, voicing loud roars and growls. On rare oc- Although it was pure bluff that quickly casions a hard-pressed one would simply lie ended in flight if a dog attacked in earnest, down and give up; a hyena we once saw it proved a surprisingly effective intimida- crowded by a persistently curious group of tion display in most encounters. But on the juvenile wildebeest did the same thing. The whole, wild dogs behaved almost as toler- spotted hyena seems on the whole to be antly toward jackals as toward vultures. notably timid by nature, as may be judged Spotted Hyenas. — Spotted hyenas, on the from the fact that mothers will often not other hand, seriously compete with the even defend their offspring. Yet they are dogs for their kills, attempting to play a driven by hunger to take incredible and commensal role against active resistance, sometimes fatal risks. In a place like the Crater, with an excep- Often, as under the above circumstances, tionally large hyena population for such a they provoked attack by their own rash- small area, numbering some 420 adults ness. But in other cases the dogs seemed (Kruuk 1966:1258), it is probably safe to to go out of their way to harry hyenas en- say that wild dogs hunting singly or in twos countered during the early stages of a hunt, and threes would very frequently lose their Those unwary enough to let the pack get kills to hyenas, since this happened oc- close could still usually get off entirely by casionally even to the pack of 21. cowering down and lying still. But those Hyenas actually stayed near the resting that stayed until the pack was close and pack for hours at a time, evidently waiting then ran away were inviting pursuit and a for a hunt to begin. It was not unusual to good mauling. At the same time, hyenas see one or more of them slowly approach a following behind the pack were generally group of dogs, then crawl to within a few ignored. On one notable occasion, the pack yards and lie gazing at them intently, as of 21 took it in turns to mob the hyenas it though urging them to get started. Often happened upon in a denning area inhabited several would wander between resting by more than 30 adults and cubs, many groups, sniffing the ground, consuming any of which were foregathered as usual prior 374 The African Wild Dog • Estes and Goddard 67 to the evening foraging. What was most surprising was that none, on this or any other occasion, attempted to take refuge underground. When hard-pressed, even half-grown pups bolted into nearby dense streamside vegetation, where the dogs did not follow. But presumably young pups were hidden in the dens, since one lactating female was reluctant to quit the immediate vicinity. She was repeatedly mobbed. Set upon by five or six dogs at a time, she would maintain a squatting defense as long as she could bear it, then break free to race for the nearest hole. Instead of going down it or backing into it, she threw herself into cup-shaped depressions next to the holes, which may or may not have been excavated by the hyenas themselves ( territorial wilde- beest also dig these depressions by pawing and horning the earth ) . In these she lay flat and tried to defend herself from the dogs, to whom only her back and head were ex- posed, while keeping up a steady volume of roars, growls, and staccato chuckles. Eventually she also took refuge in the bushes. Neither this hyena nor the next, which the dogs turned on its back and mauled for 2 minutes, bore any visible wounds. In fact, we have never known the dogs to kill or even seriously injure one. Either hyenas have exceedingly tough hides or else wild dogs are less in earnest about mobbing them than might appear. Yet the degree to which hyenas are able to capitalize on wild-dog predation for their own benefit would justify a deep antago- nism. They frequently drive away the last dogs on a kill unless the rest of the pack remains close by, and are quite capable of taking meat away from one or two dogs only a few yards removed from a kill where the rest are feeding. A more extraordinary example of this exploitation is the way hyenas take advantage of the wild dog's hunting technique: in the final moments of Fig. 9. Hyenas appropriate a wild dog prey and begin eat- ing it alive, while one of two dogs that caught it looks on, panting heavily from the chase. Hyena in foreground is half-grown. As shown in Fig. 6, the dogs reclaimed their prey when the rest of the pack arrived. a chase, when only one or a few dogs are close to the quarry, hyenas have an oppor- tunity to appropriate it before the rest of the pack arrives (Fig. 9). They attempted this with considerable regularity in the Crater, and we succeeded in recording one instance on film. In some cases it was a matter of chance that hyenas were near enough the scene of the capture to dash in at the decisive moment; in others up to three or four actually took part in the chase from the beginning. Tliough not as fast as the dogs, they were able to be in a position to intercept the quarry if it doubled back, or to grab it away from the dog(s) as soon as it was caught. When only two or three of them were on hand, the dogs hesitated to launch an immediate counterattack, par- ticularly if more than one hyena was in- volved. But usually other pack members quickly appeared, joined together to mob the hyenas, and forced them to surrender the kill. But sometimes the dogs were defeated by sheer numbers. Once when the leader of the pack of 21 had pulled down a juve- nile wildebeest in a hyena denning area, 375 68 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 some 40 hyenas closed in on the kill before 5. The provision of food for infants at the others could gather. Apparently intimi- the den and the adults that remain with dated by so many competitors, the dogs re- them when the pack is hunting, and for venged themselves by mobbing stragglers, juveniles and sick or old adults unable to punishing them savagely. Twenty minutes kill for themselves, later, while they were ranging for new prey, the hyenas pulled down an adult fe- Effective Pack Size male wildebeest on their own, quite near We have presented evidence, though ad- the first kill. Their clamor drew the dogs mittedly tentative, that large packs utilize back to the scene. But they did nothing prey resources more efficiently than small this time but look on — there were now 60 packs, with less waste. Competition from hyenas! hyenas, where they are numerous, must exert a strong selective pressure in favor of CONCLUSIONS large packs as well as for close cooperation n , n X- at kills. While the observed tendency for Pack Function „ , , , i . i small packs to keep closer together m the Hunting is undoubtedly the primary func- ^j^^^^ ^^^ ^^ ^^^^ ^^^j^ ^^^^ ^^ compen- tion of the free-ranging pack. Wild-dog be- ^^^^ somewhat for low numbers, there must havior is highly specialized and adapted for ^^ ^ minimum below which competition pack life by dint of the equal and excep- f^.^^ j^y^^^^^ ^^^ reduced hunting and kill- tionally friendly relations between individ- .^^ capability, would become a serious uals, subordination of individual to group i.^^dicap. From our observations of both activity, disciphne during the chase, and j^^.^^ ^^^ ^^^jj p^^j,^^ ^^^^ ^^ ^-^ ^^^1^ close cooperation in killing prey and mutual ^^^^ ^j^^^ ^^ ^^^ minimum effective unit, defense. It may, in fact, be seriously ques- ^^ helieve that wherever wild dogs are tioned whether a single wild dog could sur- ^^^^^^^ ^o such small packs, their ability Vive for long on its own. As demonstrated ^^ ^^^^.^^ ^^^ reproduce may be endan- by the successful rearing of a litter after the ^^^^^ j^.^ -^ ^^^ ^^^-^^ -^^^ account the mother died, feeding and protection of the possibility of a differential birth or mor- young is another important pack function. ^^^^^ ^^^^ ^^at results in a low ratio of fe- The main selective advantages of the pack ^^^^j^^ j^ -^ represents a pathological con- hunting unit may be summarized as follows: ^i^i^^^ ^bis alone could mean that the 1. Increased probability of success species is in serious trouble; a prompt in- through cooperation, hence better oppor- vestigation of reproduction and neonatal tunity to eat regularly at less cost in indi- mortality is called for to find out to what vidual effort extent an abnormally low percentage of fe- 2. More efficient utilization of food re- males may be responsible for the apparent sources decline of the species in many parts of 3. Less disturbance of prey populations Africa, than would result if each animal hunted individually ^'^V Relations 4. Mutual protection against competitors It seems clear that wild dogs are highly ( spotted hyenas ) and possible predators selective in the species they prey upon, spe- ( hyena, leopard [Panthera pardus], and ciahzing in East African steppe-savanna on hon) Thomson's gazelles, and on wildebeest 376 The African Wild Dog • Estes and Goddard 69 calves during the gnu calving season. Con- sidering their selectivity, their rate of killing, and the observed reactions of herbivores to them, it can only be concluded that wild dogs are by no means so wantonly destruc- tive or disruptive to game as is commonly supposed. Kiihme reached the same con- clusion (p. 528). Indeed, until one comes to realize that plains game simply has no place to hide and no sanctuary where predators cannot follow, it is a recurrent surprise to note how short-lived and local- ized are disturbances due to predation. In a prey population as small as that of Thomson's gazelles in Ngorongoro, if one assumes an average annual recruitment rate of roughly 10 percent, predation at the rate of only one a day obviously would re- duce the population if maintained over a long period. There was, however, no evi- dence that the Thomson's gazelle popula- tion declined after wild dogs became resi- dent in the Crater; our gazelle censuses in October, 1964, and May, 1965, showed no reduction that could not be accounted for by simple counting errors. Even an actual reduction would have no relevance to the overall situation, as Thomson's gazelles are the most numerous herbivores in their cen- ters of distribution (the steppe-savanna from central Kenya to north-central Tan- zania). On the Serengeti Plains, where the gazelle population is estimated at 800,000 and there are probably fewer than 500 wild dogs, predation by this species could have no appreciable effect. Indeed wild dogs are only one of nine predators on the Thom- son's gazelle (Wright 1960), and not the most important one at that; jackals, which are numerous and specialize in catching new fawns, are probably the main predators. Since wild dogs are nowhere numerous and everywhere apparently specialize on the most abundant small to medium-sized antelopes, it can be argued that more, not fewer, of them are needed. The population explosion of impala in Kruger National Park and many other places where wild dog numbers have declined offers convinc- ing evidence. The high percentage of ter- ritorial males in wild dog kills of Thomson's gazelles offers a more subtle example of how predation may benefit a prey species: in "probably every gregarious, territorial antelope species, there is always a surplus of fit, adult and young-adult males which cannot reproduce for want of enough suit- able territories, so that the removal of ter- ritorial males by predation is of perhaps major importance in opening up territories for younger and sexually more vigorous males. In our judgment, the wild dog is an in- teresting, valuable predator whose con- tinued survival may be endangered. We feel it should be strictly protected by law in all African states where it occurs, and that it should be actively encouraged, if this is possible, in every park and game re- serve. LITERATURE CITED BouRLiERE, C. F. 1963. Specific feeding habits of African carnivores. African Wild!. 17(1): 21-27. Brown, L. 1965. Africa, a natural history. Ran- dom House, New York. 299pp. Bryden, H. a. 1936. Wild life in South Africa. George G. Harrap Co. Ltd., London. 282pp. Haltenorth, T. 1963. Klassifikation der Sauge- tiere: Artiodactyla. Handbuch Zool. Bd. 8. 167pp. Kruuk, H. 1966. Clan-system and feeding habits of spotted hvaenas {Crocuta crocttta Erxleben). Nature 209 ( 5029) : 1257-1258. KiJHME, W. 1965. Freilandstudien ziir Soziologie des Hyanenhundes. Zeit. Tierpsych. 22(5): 495-541. Leyhausen, p. 1965. tjber die Funktion der rela- tiven Stimmungshierarchie. Zeit. Tierpsych. 22(4):395-^12. Maberly, C. T. Astley. 1962. Animals of East Africa. Howard Timmins, Cape Town. 221pp. Mackworth-Praed, C. W., and C. H. B. Grant. 1957. Birds of eastern and north eastern Africa. Series I, Vol. I. African handbook 377 70 Journal of Wildlife Management, Vol. 31, No. 1, January 1967 of birds. Longmans, Green and Co., London. of game in Ngorongoro Crater. E. African 806[+40]pp. Wildl. J. 2:165-168. Percival, a. B. 1924. A game ranger's note- Vaughan-Kirby, F. 1899. The hunting dog. Pp. book. Nisbet & Co., London. 374pp. 602-606. In H. A. Bryden (Editor), Great Selous, F. C. 1881. A hunter's wanderings in and small game of Africa. R. Ward Ltd., Africa. R. Bentley & Son, London. 455pp. London. 612pp. Stevenson-Hamilton, J. 1947. Wild life in Wright, B. S. 1960. Predation on big game in South Africa. Cassell & Co., Ltd., London. East Africa. J. Wildl. Mgmt. 24(1):1-15. 343pp. Turner, M., and M. Watson. 1964. A census Received for publication March 21, 1966. 378 Novejnber, 1957 GENERAL NOTES 519 HOMING BEHAVIOR OF CHIPMUNKS IN CENTRAL NEW YORK Homing movements ranging from about 150 to 700 yards have been recorded for Tamias by Seton (life histories of northern animals, vol. 1: 341, 1909), Allen (BuU. N. Y. State Mus., 314: 87, 1938), Burt (Misc. Publ. Mus. Zool. Univ. Midi., 45: 45, 1940), and Hamilton ( American mammals, p. 283, 1939 ) . While engaged in other studies during the summer of 1952, I had the opportunity of making additional observations on the homing behavior of the eastern chipmunk, Tamias striatus lysieri (Richardson), on the campus of Cornell University at Ithaca, Tompkins County, New York. Live-trapping was conducted from July 26 to August 3 in a tract of approximately 3 acres of hemlock and mixed hardwood forest bordering a small artificial lake. A maximum of 12 traps was employed. The chipmunks taken were sexed, aged (subadult or adult), marked by cUpping patches of fur on various parts of the body, and transported in a cloth bag to one of six release points. The latter were situated in similar continuous habitat or in an area of campus buildings, lawns, shrubbery, and widely spaced trees adjacent to the woodland. An individual was considered as having homed when it was retaken within 115 feet of the original point of capture. Those chipmunks that returned and were recaptured were im- mediately released again in a different direction and usually at a greater distance. First releases averaged 675 feet (310-1,160) and second ones, 1,015 feet (500-1,570). Two animals that returned after second removals were liberated for a third time at distances of 1,130 and 2,180 feet. All distances given are calculated from the station where the animal was originally trapped. Since the mean home range size of chipmunks in this vicinity has been calculated as about .28 acre (Yerger, Jour. Mamm., 34: 448-458, 1953), it is assumed 379 520 JOURNAL OF MAMMALOGY Vol. 38, No. 4 that in most, if not all, instances the removal distances involved were great enough to place the animal in unfamiliar territory beyond the boundaries of its normal range of movements. A total of 18 individuals, consisting of five adult males, four adult females, two subadult males and seven subadult females, were marked and released a total of 29 times through July 30. Aniinals handled after tliis date are not included in tlie treatment of the data, since it is felt that there was insufficient opportunity for them to be retaken following their release. Seven of the chipmunks returned to the vicinity of original capture a total of ten times over distances varying from 430 to 1,200 and averaging 650 feet. In six of the ten returns the animals were retaken in the same trap in which they were initially caught. In two instances individuals were retrapped at stations 20 and 40 feet removed from the one where first taken, and in two others tlie individuals were recovered at a distance of 115 feet from the original site of captuie. Two adult males returned from 490 and 540 feet in two and three days, respectively, but were not recovered after second removals to 1,060 and 1,150 feet. Two other mature males trapped following their release at 310 and 750 feet had moved in a direction other tlian that of their original capture. An adult female was found in a trap a day after having been released at 775 feet. She returned a second time from 600 feet in two days. Anodier adult female was retrapped at the original trap station seven days following her release only 430 feet away. A single subadult male was recaptured in his original location tiie next day after his initial removal to 750 feet and tv/o days after a second liberation at 1,200 feet. He was not retrapped subsequent to a third relocation 2,180 feeL distant. Another young male was captured at a point 380 feet closer to its home area six days after being released 940 feet away. Three subadult females homed successfully. One returned [rom 580 feet the day aiter release, another from 650 feet in two days' time, and the third from 490 feet after an interval of five days. None were retaken foliowing second liberations ranging from 940 to 1,570 feet. Two other subadult females were captured 100 and 120 feet closer to their original capture sites the day after having been released at 460 and 450 feet, respectively. These limited data suggest that homing ability was restricted to rather short distances, only one individual being known to have returned from a point more than 775 feet away. The extent of these movements may be somewhat less than several reported by authors pre- viously mentioned. However, because of tlie smaller home range size of chipmunks in this area as compared to odier habitats in which the animals homed from more distant points, the actual distances mov^;.t over strange territory may be fairly comparable. The present results indicate no obvious differences in the proportion of adults and subaduits homing nor in the average distance over which individuals in each of tliese age classes returned. The intervals of one to seven days between releases and recoveries, the relatively short distances involved, and the rather low proportion of returns ( 38.8 per cent) suggest that the animals may have returned to their home areas through random movements until familiar terrain was encountered. It should be mentioned, however, that tlie small number of traps employed might have been a factor in the low rate of recovery, since a chipmunk returning to its home region had a lower probabihty of being recaptured than would have been the case had more traps been present. This might also have tended to increase the apparent time taken to reach the home area following release. On tlie otlier hand, the use of a limited number of traps may have been advantageous in that there was less interference by traps with the normal activities and movements of the animals. — James N. Layne, Dept. of Biology, Univ. of Florida, Gainesville. Received December 1, 1956. 380 COMPARATIVE ECHOLOCATIOX BY FISHING BATS RODEHICK A. SUTHERS Abstract. — The acoustic orientation of two species of fish-catching hats was studied as they negotiated a row of strings or fine wires extending across their fhght path. Orientation sounds of Pizomjx vivesi consisted of a steep descending FM sweep lasting about 3 msec. Noctilio leporimis used 8 to 10 msec pulses composed of an initial nearly constant frequency portion followed by a descending frequency modulation. The echolocation of small wires by N. leporimis differed from that of surface fish in that during wire avoidance no nearly constant fre- quency or entirely FM pulses were emitted, nor was the pulse duration markedly shortened as the barrier was approached. There was extensive temporal overlap at the animal's ear of returning echoes with the emitted cries when the bat was near the barrier — a strong contrast to the apparent careful minimization of such overlap during feeding maneuvers. Soctiliu increased its average pulse duration about 2 msec when confronted with a barrier of 0.21 mm, as opposed to 0.51 mm, diameter wires. Fizonijx detected these wires well before pulse-echo overlap began, but at a shorter range than did N. leporimis, suggesting the latter species may have a longer effective range of echolocation. At least two species of Neotropical bats have independently evolved an ability to capture marine or aquatic organisms. A comparison of the acoustic orientation of these animals is of particular interest in view of their convergent feeding habits yet strikingly different orientation sounds. Noctilio leporimis Linnaeus ( Noctilionidae ) catches fish by occasionally dipping its dispro- portionately large feet into the water as it flies low over the surface. Very small surface disturbances can be echolocated and play an important role in determining the locations of the dips (Suthers, 1965). Fi.sh caught in this way are transferred to the mouth and eaten. Pizomjx vivesi Menegaux (Vesper- tilionidae) also possesses disproportionately large feet. Much less is known concerning the feeding behavior of this species, though it is reasonable to assume that it uses its feet in a manner similar to N. leporimis (but see Reeder and Norris, 1954). Extensive attempts to induce captive P. vivesi to catch pieces of shrimp from the surface of a large pool were unsuccessful. The fol- lowing comparison is therefore based on the ability of these bats to detect small obstacles. Methods The experimental animals consisted of two P. vicesi, selected as the l)est flyers of several collected in the Gulf of California, and one N. leporimis captured in Trinidad. The research was conducted at the William Beebe Memorial Tropical Research Station of the New York Zoological Society in Trinidad. The bats were flown in a 4 X 15 m outdoor cage described elsewhere (Suthers, 1965). The test obstacles consisted of a row of strings or wires 2.5 m long which were hung at 55 cm intervals across the middle of the cage. Four sets of obstacles were used: 2 mm diameter strings, 0.51 mm, 0.21 mm, and 0.10 nun diameter wires, respectively. The bats were forced to pass through this barrier in order to fl\' the length of the cage. Each 79 381 80 JOURNAL OF MAMMALOGY Vol. 48, No. 1 ihght was scored as a hit or a miss according to whether any part of the animal touched the test obstacles. Movement of the larger wires was easily visible following even gentle contact, but lateral illumination of the barrier was necessary in order to score flights through the row of 0.10 nun wires. The wires were occasionally shifted laterally about 20 cm across the width of the cage in order to reduce the possibility that the bats might learn their location. An attempt was made to test each animal on two or more sets of obstacles per night, though this was not always possible. Cases in which the bat was making unusually frequent landings or was particularly reluctant to fly are omitted. Also excluded are flights on which the barrier was approached very near the upper ends of the wires, along either wall of the cage, or at an angle to the row of obstacles which was decidedly smaller than 90°. Experiments with P. vivesi no. 1 were terminated by its sudden death, which oc- curred before the 0.10 mm diameter wire was available. Tests with this size wire were therefore performed with a second healthy Pizonyx. A series of flights through the 0.51 mm and 0.21 mm diameter wires was photographed with a 16 mm sound motion picture camera, while simultaneous two-channel tape re- cordings of the orientation sounds were obtained from microphones placed on opposite sides of the barrier. The position of the flying bat relative to the barrier was calculated by comparing the arrival time of each orientation sound at either microphone and also by matching the image of the bat in each frame of the film with rectified orientation sounds on the optical sound track. Details of these methods and the instrumentation are de- scribed elsewhere ( Suthers, 1965). The overall frequency response of the recording sys- tem was approximately uniform between 15 and 100 kc/sec. A total of 45 flights by P. vivesi and N. leporinns was tape recorded and photographed. Sixteen of these were discarded for reasons listed above, or because the bat did not fly on a straight path between the microphones, or because of a poor signal-to-noise ratio on one of the channels. The remaining 29 flights were analyzed and pulse intervals ( the silent period from the end of one pulse to the beginning of the next) were plotted against the distance of the bat from the barrier (see Fig. 2). The animal's position was deter- mined to within an accuracy of about ± 10-15 cm at a distance of two meters from the wires and ± 5-10 cm in the immediate region of the wires. Results and Conclusions Obstacle avoidance scores are given in Table 1. The greater success of P. vivesi in avoiding 0.10 mm wires may reflect its shorter maximum wingspan of 40 cm, compared to 50 cm for N. leporinus. Audio monitoring of the recti- fied orientation sounds emitted by these species during their flights indicated Table 1. — Percent of flights through harrier on which hat missed obstacles spaced at 55 cm intervals. Total number of flights in parentheses. Maximum wingspan of P. vivesi is about 40 cm; that of N. leporinus is about 50 cm. OBSTACLE DIAMETER (MM) Bat 2 0.51 0.21 0.10 Pizxmijx vivesi (1) 94% 83% 51% (163) (416) (232) Pizonyx vivesi (2) 71% (151) 37% (74) Noctilio leporinus 91% 76%> 60% 20% (207) (203) (217) (55) 382 February 1967 SUTHERS— FISHING BATS 81 TIME (MSEC) Fig. 1. — Sound spectrographs of orientation sounds emitted by Pizonyx vivesi (a) and Noctilio leporinus (b) when approaching wire obstacles. A pair of consecutive pulses, reproduced at two different filter settings of the sound spectrograph, is shown for each species. The narrow band filter setting (top) best indicates the frequency spectrum of the cries, whereas pulse duration and temporal relationships are more accurately shown using a wide band filter (bottom). that approaches to the three larger diameter obstacles were accompanied by increases in the pulse repetition rate, whereas no such increase was noted during approaches to the 0.10 mm wires. This suggests that these latter wires were too small to be detected at an appreciable distance and that tests using them may indicate chance scores. Tape recordings of flights between 0.51 and 0.21 mm diameter wires showed that these two species used distinctly different kinds of orientation sounds in detecting the obstacles ( Fig. 1 ) . When approaching the barrier at a dis- tance of about 2 m, P. vivesi emitted ultrasonic pulses with a duration of about 3 msec at a mean repetition rate of 10 to 20 per sec. Each of these was frequency modulated ( FM ) , sweeping downward from about 45 kc/sec to 20 kc/sec and accompanied by a second harmonic. The shghtly lower starting frequency ( 36 kc/sec ) reported by Griffin ( 1958 ) may be due to the lower sensitivity to high frequencies of microphones available at that time. At a similar distance from the barrier N. leporinus produced pulses with a duration of about 8 to 10 msec at comparable repetition rates. These sounds, however, were composed of an initial portion at a nearly constant frequency of about 60 kc/sec followed by an FM sweep down to 30 kc/sec. Neither species made any pronounced change in the frequency structure of its pulses as it approached and negotiated 383 82 JOURNAL OF MAMMALOGY Vol. 48, No. 1 2 10 2 1 DISTANCE FROM WIRES (METERS) Fig. 2. — Examples of changes in orientation pulse intervals during flights by fishing bats through a barrier of fine wires spaced at 55 cm intervals across their flight path. Each dot represents one orientation sound: (a) Noctilio leporinus flying between 0.51 mm diameter wires; (b) Pizonyx vivesi flying between 0.51 mm diameter wires; (c) N. leporinus flying between 0.21 mm diameter wires; (d) P. vivesi flying between 0.21 mm diameter wires. On flights a, b, and d, the bat did not touch the wires. On the flight shown in c the wires were hit by the bat. Vertical dashed line indicates position of the wires. the barrier. The pulse repetition rate was increased, however, to about 30 or 35 per sec. The use of a single pulse type by IV. leporinus contrasts with its echolocation during normal cruising and feeding when constant frequency and entirely FM pulses are also employed (Suthers, 1965). Fig. 2 gives examples of alterations in pulse intervals during one flight by each species through a barrier of 0.51 mm and of 0.21 mm diameter wires. The possible significance of the tendency to alternate long and short pulse intervals during the approach to the barrier'is not known. It was not possible to reliably distinguish hits from misses on the basis of these graphs. The minimum average distance of detection was estimated by calculating the point at which the bat began to shorten the pulse intervals. Pizonyx vivesi and N. leporinus must have detected the 0.51 mm wires at an average distance from the barrier of at least 110 and 150 cm, respectively, and the 0.21 mm 384 February 1961 SUTHERS— FISHINC; BATS 83 CO ^ 100- GO CO CO DISTANCE FROM WIRES (METERS) Fig. 3. — Mean pulse intervals (solid line) and mean pulse duration (broken line) of Noctilio leporinus (a) and Pizomjx vivesi (b) during approaches to the 0.51 mm diameter wires spaced across the flight path at 55 cm intervals. The bat is flying from left to right. Vertical dashed line indicates the position of the v/ires. Arrows indicate estimated minimum mean distance of detection as judged by progressive shortening of the pulse intervals. Dotted diagonal line shows the distance at which pulse-echo overlap will first occur for any given pulse duration. Echoes from the wires of pulses whose mean duration lies above this line will overlap with the emitted pulse by an average amount equal to their vertical distance above the line. Each point represents tlie mean interval or duration of pulses emitted in the adjacent ± 10 cm. Intervals for N. leporinus are averages of five flights: P. vivesi intervals, of seven flights. All pulse durations are averages of three flights. 385 84 JOURNAL OF MAMMALOGY Vol. 48, No. 1 lOOh 80 60 40 20 CO ^ 100 CO 80 60 40 20- 10 8 6 4 2 CO II CO 2 1 0 DISTANCE PROM WIRES (METERS) Fig. 4. — Mean pulse intervals (solid line) and mean pulse duration (broken line) of Noctilio leporinus (a) and Pizonyx vivesi (b) during approaches to the 0.21 mm diameter wires spaced across the flight path at 55 cm intervals. For explanation see legend of Fig. 3. Possible alternate interpretations of the point at which a progressive decrease in pulse intervals first appears are indicated by small arrows. The more conservative estimates denoted by the large arrows have been used in the text. Pulse durations and intervals of N. leporinus are averages of 10 flights; those of P. vivesi are averages of seven flights. 386 February 1967 SUTHERS— FISHING BATS 85 wires at an average of at least 70 and 130 cm, respectively (Figs. 3 and 4). Pulse durations did not markedly shorten as the barrier was approached. Thus at close ranges the echoes returning from the wires must have overlapped extensively with the emitted pulse. In the case of N. leporinus this overlap may have begun on the average when the bat was still 130 and 170 cm from the 0.51 and 0.21 mm wires, respectively (Figs. 3 and 4). The data do not exclude the possibility that the start of pulse-echo overlap and the start of a progressive reduction in the pulse intervals by N. leporinus may occur simultaneously or be closely synchronized. The pulses of P. vivesi must have overlapped with their echoes during the last 40 and 50 cm of the approach to the 0.5] and 0.21 mm wires, respectively (Figs. 3 and 4). It seems clear that P. vivesi began to decrease its pulse intervals well before the first pulse-echo overlap occurred. Noctilio leporinus regularly emitted longer pulses when approaching the 0.21 mm wires than when approaching the 0.51 mm wires. The significance of this difference is not known, although it is possible that the earlier initiation of pulse-echo overlap, or the increased duration of overlap at a given dis- tance, when longer pulses are used, in some way facilitated detection of the finer wires. If this is true, however, why is overlap minimized with such apparent care during the detection of small cubes of fish muscle tissue pro- jecting above the water surface (see below)? Since the difference in pulse duration as a function of wire diameter was already present when the bat was two meters from the barrier, either leporinus must have determined some- thing about the wire diameter at a distance of more than two meters, or it must have remembered what kind of wires it had to detect and adopted a suitable pulse duration prior to their detection. Details of the echolocation of P. vivesi during feeding are not known. Pulse- echo overlap during wire avoidance by N. leporinus, however, contrasts strongly with its apparent careful avoidance during catches of stationary 1 cm- cubes of fish muscle tissue projecting above the surface of the water. Pulse lengths under these conditions were progressively shortened as if to avoid pulse-echo overlap until the bat was 30 cm or less from the food (Suthers, 1965). Thus in the case of N. leporinus, at least, information concerning the position and nature of small wire obstacles is probably received in the presence of overlap, whereas most of this information regarding potential food must be obtained without such overlap. It has yet to be determined whether or not pulse-echo overlap is actually utilized by the bat. It has been suggested (Pye, I960; Kay, 1961) that possible nonlinearities in the ear may allow bats to utilize beat notes arising from pulse-echo overlap as a means of determining distance. Three species of chilonycterine bats have subsequently been found to maintain an overlap during the pursuit and catch- ing of Drosophila (Novick, 1963, 1965; Novick and Vaisnys, 1964). Myotis lucifugus (Cahlander et al., 1964) and N. leporinus, on the other hand, appear to minimize overlap when catching tossed mealworms or fish, respectively. 387 86 JOURNAL OF MAMMALOGY Vol. 48, No. 1 Should pulse-echo overlap be utilized by N. leporinus in determining its dis- tance from the wires, then some basically different method, such as the temporal delay of the returning echo (Hartridge, 1945), must be employed in determining the range of potential food. Since the constant frequency portion of the Doppler-shifted echo would at first overlap with the FM portion of the call, and later with part of both the constant frequency and FM portions of the call, any resulting beat note would have a complexly varying frequency structure from which it would be difficult for the bat to determine its distance from the barrier. One would like to know if there is a significant difference in the range of echolocation for these fishing bats. Noctilio leporinus emits very loud pulses with a peak-to-peak sound pressure of up to 60 dynes/cm- at a distance of 50 cm from the mouth (Griffin and Novick, 1955). The intensity of sounds emitted by P. vivesi has not been measured, although the shorter range at which they can be detected on an ultrasonic receiver suggests they are less intense than those of Noctilio. M. hicifugus, a vespertihonid closely related to Pizonyx, can detect 0.46 mm diameter wires at 120 cm and 0.18 mm wires at 90 cm (Grinnell and Griffin, 1958), thus comparing favorably with fishing bats in this respect. Peak-to-peak sound pressures of this species have been measured at 12 dynes/cm- at 50 cm (Griffin, 1950). Sound intensity of the emitted pulse, however, is but one of a number of physical and physiological factors which must play important roles in determining the range of such a system for acoustic orientation. Acknowledgments I wish to thank Prof. Donald R. Griffin, Drs. H. Markl, N. Suga, and D. Dunning for helpful criticism and assistance. Drs. R. E. Carpenter, G. W. Cox, and A. Starrett gave valuable assistance in obtaining live P. vivesi. Appreciation is also expressed to the San Diego Society of Natural History for use of the Vermillion Sea Station at Bahia de los Angeles, Baja California, and to the New York Zoological Society for the use of the William Beebe Memorial Tropical Research Station in Trinidad. The coopera- tion of the Mexican, Trinidadian, and United States governments in the transit of bats is gratefully acknowledged. This work was supported by grants from N.LH., The Society of the Sigma Xi Research Fund, and the Milton Fund of Harvard University. Literature Cited Ghiffix, D. R. 1950. Measurements of the ultrasonic cries of bats. J. Acoust. Soc. Amer., 22: 247-255. . 1958. Listening in the dark. Yale Univ. Press, New Haven, 413 pp. Griffin, D. R., and A. Novick. 1955. Acoustic orientation of neotropical bats. J. Exp. Zool., 1.30: 251-300. Grinnell, A. D., and D. R. Griffin. 1958. The .sensitivity of echolocation in bats. Biol. Bull., 114: 10-22. Hartridge, H. 1945. Acoustic control in the flight of bats. Nature, 156: 490-494. Kay, L. 1961. Perception of distance in animal echolocation. Nature, 190: 361-.362. Novick, A. 1963. Pulse duration in the echolocation of insects by the bat, Pteronotus. Ergebnisse Biol, 26: 21-26. 388 Fehruary 1967 SUTHERS— FISHING BATS 87 . 1965. Echolocation of flying insects by the Ixit, Chilonijcteris psilotis. Biol. Bvill., 128: 297-314. NoviCK, A., AND J. R. V.\isNYs. 1964. Echolocation of flying insects by the bat, Chilonijcteris parnelli. Biol. Bull., 127: 478-488. Pye, J. D. 1960. A theory of echolocation by bats. J. Laryng. Otol., 74; 718-729. Reeder, W. G., -AND K. S. NoRRis. 1954. Distribution, habits, and type locality of the fish-eating bat, Pizonyx vivesi. J. Mamm., 35: 81-87. SuTHEBs, R. a. 1965. Acoustic orientation by fishing bats. J. E.xp. Zool, 158: 319-348. The Biological Laboratories, Harvard University, Cambridge, Massachusetts (present address: Department of Aimtomy and Phijsiology, Indiana University, Bloomington, Indiana 47401). Accepted 11 November 1966. 389 The Intraspecific Social Behavior of Some Cricetine Rodents of the Genus Peromyscus JOHN F. EISENBERG Department of Zoology, University of California, Berkeley Abstract: A three-compartment territorial cage was employed in studying the intraspecific social behavior of Perorjiyscus californicus, P. erernicus, P. crinitus, and P. maniculatus. Two pairs of mice of the same species were allowed to interact for one week following a period of separa- tion from each other in the end compartments. All species showed similar forms of agonistic behavior patterns. The closely related P. erernicus and P. californicus employed a modified fighting technique involving an attack leap, scuffle, and avoidance leap. The females of P. crinitus and P. californicus showed a pronounced nest site attachment and defense. Agonistic behavior with the exception of nest defense was confined almost entirely to the males. The aggressivity of P. erernicus was the least pro- nounced. Peromyscus crinitus males showed the most agonistic behavior, and P. californicus and P. maniculatus were intermediate. Pair associa- tions in the territorial boxes were prolonged for P. californicus and, to a lesser extent, for P. crinitus. The other two species showed a weak pair bonding. Introduction In a previous paper (Eisenberg, 1962), the social behavior of Peromyscus californicus parasiticus and P. maniculatus gambelii was described and contrasted. In the present study the methods used were extended to P. crinitus stephensi and P. eremicus eremicus. Enough data now exist to present a comparison of all four species. Peromyscus californicus and P. eremicus are closely related (Hooper, 1958) and are included in the Eremicus division of the genus Peromyscus. Peromyscus crinitus is considered by Hooper to be intermediate between P. boylei and P. maniculatus, all of which are included within the Maniculatus division. The individuals of P. californicus and P. maniculatus used in this study were trapped in the vicinity of Berkeley, Alameda Co., Califor- nia. They inhabit <:ympatrically the chaparral areas in west central California, and their ecology has been intensively studied by McCabe and Blanchard (1950). ' Peromyscus eremicus and P. crinitus inhabit the arid Sonoran life zones of western North America. In general, P. crinitus occupies the Upper Sonoran high deserts while P. eremicus ranges in the Lower Sonoran zone. The individuals of P. crinitus used in this study were trapped in the vicinity of Cottonwood Springs, San Bernardino Co., California. The individiials of P. eremicus were all collected around 15 miles west of Borrego Springs in San Diego Co.. California. These two species are sympatric over part of their ranges. In general, P. crinitus is a rock dweller while P. erernicus inhabits the brushy, flat areas of the desert floor. Both P. californicus and P. eremicus have smaller litters when compared with P. maniculatus, while P. crinitus 240 390 1963 Eisenberg: Social Behavior of Cricetines 241 has an intermediate average litter size. McCabe and Blanchard report average litter sizes of 1.91 and 5 for P. calijornicus and P. maniculatus ^ambelii, respectively. Hall gives an average litter size of 4 for P. crinitus, while Asdell records an average litter size of 3.7 for P. eremicus. The maturation and growth of P. calijornicus and P. mani- culatus (^ambelii are analyzed in detail by McCabe and Blanchard (1950). Acknowledgments. — I wish to express my gratitude to Drs. Peter Marler and Seth B. Benson for their valuable criticism of this work. Part of the research was conducted during the tenure of a National Science Foundation Prc-doctoral Fellowship. Methods Alter the individual behavior patterns had been described for the various species, a series of encounters between two individuals of the sames species was run in order to obtain descriptions of the various postures and activities employed in the .social context. Following these preliminary experiments a series of territorial encounters was arranged using two pairs of mice of the same species for each encounter. The territorial boxes were 9 x 26 x 75 inches in size with glass tops, sides, and fronts; the back was of quarter-inch hardware cloth, and the bottom of wood. Each box was divided into three equal compartments by wood partitions. A small opening, 2 x 1 inches, cut in each partition served as a door, and was closed with screen except at the time of experimentation. A pair of animals was placed in each end compart- ment with food, water, and cotton nesting material, and left for two weeks. The doors between the compartments were then opened and the subsequent behavior of the four individuals was noted for one hour. These observations were made at night using a red light for illumination. The doors were then left open for one week and daily observations at set intervals allowed me to ascertain in which compart- ment and with whom the animals were dwelling. The specimens were liH-clipped for identification. These experiments were run for all species during the spring, summer, and fall. Five experiments were run with P. maniculatus and P. crinitus, and six with P. calijornicus and P. eremicus. Each experiment was run with a difTerent pair of animals. Agonistic Behavior Patterns A complete description of the adult behaxior patterns for P. calijor- nicus and P. maniculatus is included in a previous publication (Eisen- berg, 1962). The adult behavior of P. eremicus and P. crinitus is very similar. In the present paper only the agonistic patterns which ap- peared in the encounters will be discussed. Fightini>. — Two animals approach and one or both rush. The animals lock together with their ventral surfaces in contact and roll about, gripping one another very lightly with their fore and hind feet. This is called the locked fii^htino posture. It generally ends either by breaking apart followed by a chase-flight sequence, or when one 391 242 The American Midland Naturalist 69(1) Table I. — Bouts f)f the different ftirnis of agonistic beha\i()r during six territorial encounters (P. californicus) Type of Fighting NVst Average;' eiKoiiMlrr Modified Locked Clia>e Upright defense ^ Encounter Male to Male 4 17 38 9 0 68 11.3 Male to Female 1 0 4 0 0 5 .8 Female to Male 0 I 5 3 10 19 3.2 Female to Female 0 0 1 1 0 2 .3 V 5 18 48 13 10 94 15.6 animal is forced on its back in a defeat posture. Peromyscus califor- nicus and P. erernicus often employ a modified fightiriii technique where the animals spring at one another, scuffle briefly, and then jump away (cf. Jumping avoidance technique, Eisenberg, 1962). Chasitii^. — This is usually confined to the floor of the cage, but small leaps may be employed. Durin" chasing the pursuing animal may bite at the rump or tail of the subordinate. Upright posturing. — Upright posturing was scored when an animal neither attacked nor fled, but raised its forepaws off the ground and with its body at about 45-90 degrees off the ground either darted its head at or e.xtended its forepaws toward the aggressor. This posture can be broken down into several sub-patterns. Nt'st defense. — The animal assumes a crouched upright in the nest and by wardin": with the forepaws or darting the head repulses the intruder. In P. maniculatus and P. californicus this movement is often accompanied by an explosive squeak or ''chit.'' Peromyscus crinitus has a slightly different nest defense sound in that chits are uttered in bursts of 2 to 20 at a rate of 5-6 per second. Species Comparisons Tables I through IV summarize the number of bouts of the five major agonistic patterns for each class of interaction (male to male; male to female; female to male; and female to female) for all four species. In general, after the doors were opened the males would Table II. — Bouts of the different forms of agonistic behavior during six territorial encounters (P. eremicus) Type of Fighting Nest Average/ encounter Modified Locked Chase Upright defense 2 Encounter Male to Male 12 2 31 1 0 46 7.7 Male to Female 1 0 9 0 0 10 1.7 Female to Male 1 0 4 2 1 8 1.3 Female to Female 2 16 1 3 4 48 0 3 0 1 7 71 1.2 2 11.9 392 1963 Eisenbf.ro: Social Behamor of Cricetines 243 Table III. — Bouts of the different forms of agonistic behavior during fi\ e territorial encounters (P. maniculatus) Type of Fighting Nest Average encounter Modified Locked Chase L'pright defense 2i Encounter Male to Male 0 20 24 11 2 57 11.4 Male to Female 0 1 6 3 0 10 2.0 Female to Male 0 n 0 3 2 5 1.0 Female to Female n 0 1 22 0 30 0 17 0 4 1 73 .2 V 14.6 encounter one another and fit^ht. After a variable number of fights the superior male would invade the nest chamber of the subordinate pair. The females were less prone to become involved in fighting, but females of P. calijornicus and P. crinitus engaged in a significant amount of nest defense. To a marked extent, males of P. calijornicus and P. eremicus employed the modified fighting technique of an attack leap, scuffle, and jumping away. Pfroniyscus eremicus males employed this [pattern almost exclusively. Perornyscus eremicus showed the lowest number of male-male aggressive bouts per encounter [1 .1 \ : P. calijornicus and P. maniculatus were intermediate with 11.3 and 11.4 bouts per male-male encounter, respectively. Perornyscus crinitus was the highest with an average of 17.0 bouts per encountei. In all sjjecies. bouts of chasing comprised the greatest percentage of agonistic behavior. Agonistic behavior involving females was low except for P. crinitus and P. calijornicus where nest defense was shown. By taking notes on the distribution of the animals in the encounter cages on the week following the opening of the doors, it was possible to determine how long a pair remained together and separate from tfie other pair. Peromyscus calijornicus' has a strong tendency to re- main paired and separate. PcrotJiyscus crinitus exhibits this to a lesser extent and the other two species show a weak pair association. It appears that the pair association results in part from the strong nest defense tendencies by the females ot P. crinitus and P. calijornicus. The male, whether the winner or loser of the first night's battles, is able to remain with his female while her nest defense mitigates against Table I\'. — Bouts of the different forms of agonistic behavior during five territorial encounters (P. crinitus) Type of Fighting .Nest Average encounter Modified ^ocked Chase Upright defense ^' Encounter Male to Male Male to Female Female to Male Female to Female 2 2 30 70 11 40 153 30.6 2 27 50 5 1 85 17.0 0 3 18 0 1 22 4.4 0 1 6 36 43 8.6 0 -■ 1 0 2 3 .6 393 244 The American Midland Naturalist 69(1 ) Table V. — Length of time the pairs remained together and separate from the other mice in the territorial experiments Number of ex- Average number Species periments of days* P. californicus 6 5.5 p. eremicus 6 1.0 p. crinitus 5 4.4 p. maniculatus 5 .8 * Maximum possible number of days — 7. the integration of the group. In the experiments with P. eremicus and P. maniculatus, the winning male and both females generally nested together on the following day. The losing male was gradually assim- ilated after two or three days. Table VI portrays the average number of days a pair or a single individual remained in its original nest during the week following the opening of the doors. Peromyscus californicus is again appreciably higher. Peromyscus crinitus does not demonstrate the same trend even though this species tend.s to preserve its pair structure. This is probably caused by the extreme aggressiveness of the superior male who displaces the inferior male and his mate. That P. californicus can maintain its original nest site may be in part a result of its ability to control the expression of aggressive behavior (Eisenberg, 1962). Although the male P. californicus fight readily, they do not persist with overt as;?ression but soon settle down to a dominance situation in the cage. The males never wound each other seriously by tail and rump bites, and P. californicus employs a special mewing cry to inhibit aggressive rushes by a conspecific. The other species seem not to be able to control their agonistic behavior so well. Since P. eremicus males fight so little, wounding during the subsequent week was minimal. In only one experiment did a male lose his tail from tail biting. However, P. maniculatus and P. crinitus males persisted throughout the week in their aggressive chasing and rump and tail biting. In three experiments with P. crinitus and two experiments Table VI. — Consecutive days that a separate or paired animal remained in its original compartment Number of ex- Average number Species periments of days* P. californicus 6 5.5 p. eremicus 6 1.3 p. crinitus 5 2.8 p. maniculatus 5 2.0 Maximum possible number of days — 7. 394 1963 Eisenberg: Social Behavior of Cricetines 245 with P. maniculatus, the inferior males lost up to one-half their tails from persistent biting. In the breeding cages where fajnilies of these mice were allowed to reproduce and live together, one could see a reflection of these aggressive trends. Peromyscus crinitus is quarrelsome and the females are intolerant during parturition. Serious wounds on the rumps or ta:ils were of common occurrence with tfiis species. Peromyscus maniculatus females are more tolerant during parturition and may allow the male to remain with them. However, male-male fights in the breeding cages often resulted in serious rump and tail wounds. Peromyscus eremicus females are tolerant and the males fight little among them- selves, but at high densities in the breeding cages rump and tail wounds occasionally resulted from male-male and male-female antag- onism. Although P. calijornicus males will fight and the females defend their nests vig^orouslv, the anim.als in the breeding cashes formed stable adult groups with a minimum of wounding. A female would readily permit the male and mature litters to remain with her during par- turition. Discussion The four species seem to exemplify three types of social organ- ization. Peromyscus crinitus seems to be a form having a dispersed social organization with a high male-male antagonism and with separate nesting by females with young. Adult social groupings are probablv confined to brief pairings by males and females during the breeding season. This seems quite comparable to the situation e.x- emplified by P. leucopus described by Nicholson (1941). Peromyscus eremicus and P. maniculatus gambelii seem to have a loose type of .social structure. Pairing appears to be transient, but males, females, and litters may remain associated in nature for longer periods. This may be comparable to the situation described by Howard (1948) for P. maniculatus bairdii. Peromyscus californicus with its low reproductive potential builds and defends a complex nest and occupies a given area for rather prolonged periods of time (McCabe and Blanchard, 1950). This species may have a prolonged pair bond and if the male and female do not actually nest together through parturition, a male remains in the vicinity of a given female and the litter for several months. This species appears to be developing a social system based on small semi- permanent family groups. The ecology of P. eremicus and P. crinitus is too little known to attempt a correlation between the mode of environmental exploitation and the form of their social organization. Peromyscus californicus has probably evolved its tolerance abilities and family fonnation as a concomitant of its lower reproductive potential and its rather restricted mode of habitat exploitation. 395 246 The American Midland Naturalist 69(1 ) References AsDELL, S. A. 1946. Patterns of mammalian reproduction. Comstock Publ. Co., Ithaca, x -}- 437 p. EisENBERG, J. F. 1962. Studies on the behavior of Peromyscus maniculatus gambelii and Peromyscus calif ornicus parasiticus. Behaviour, 19: 177-207. Hall, E. R. 1946. Mammals of Nevada. Univ. of Calif. Press, Berkeley. xi + 710 p. Hooper, E. T. 1958. The male phallus in mice of the genus Peromyscus. Misc. Pub. Mus. Zool. Univ. Mich., 99: 1-59. Howard, W. E. 1948. Dispersal, amount of inbreeding, and longevity in a local population of prairie deer mice on the George Reserve, southern Mich- igan. Cont. Lab. Vertebr. Biol. Univ. Mich., 43:1-50. McCabe, T. T. and B. D. Blanchard. 1950. Three species of Peromyscus. Rood Associates, Santa Barbara, v -j- 136 p. Nicholson, A. J. 1941. The homes and social habits of the woodmouse Pero- myscus leucopus novaboracensis in southern Michigan. Am. Midi. Nat., 25:196-223. 396 ETHOLOGICAL ISOLATION IN THE CENOSPECIES PEROMYSCUS LEUCOPUS Howard McCarley Department of Biology, Austin College, Sherman, Texas Accepted December 30, 1964 Peromyscus leucopus and P. gossypinus, con- stituting the cenospecies Peromyscus leucopus, have diverged genetically .so that they have differ- ent morphological and adaptive norms. Genetic isolation, however, is apparently not complete hecau.se interspecific hybridization may occur (Dice, 19,17; McCarley, 1954a). The present paper is a report of an ethological mechanism that helps maintain the genetic distinctness of the two species. Previous studies by Dice (1940), Calhoun (1941), and McCarley (19.54b, 1963) showed that leucopus and gossypinus were generally ecologically separated in areas of sympatric distribution: leucopus in upland woods and gossypinus in lowland woods. Overlapping frequently occurs, however, during the winter and spring reproduc- tive sea.sons (McCarley, 1963). Consequently, ecological separation alone would not be adequate to account for the few recorded examples of natural interspecific hybrids in this cenospecies (Howell, 1921; McCarley, 1954a). Work done by McCarley (1953) and Bradshaw (1957) using the procedures of Blair and Howard (1944) suggested that continued species separa- tion of leucopus and gossypinus may, in part, depend on ethological, or species discrimination mechanisms. Experiments were begun in 1959 using techniques modified from the procedures of Blair and Howard (1944). These tests utilized three individuals, one male and two females or one female and two males. If males were to be tested, a male was placed in one of the two middle compartments of a four-compartmented cage and was free to move between these two compartments. A leucopus female was confined to one end compartment and a gossypinus female to the other end compartment. A reciprocal arrangement of mice was used when females were tested. Each combination of three mice was observed daily, usually early in the morning, for not less than 5 nor more than 1 1 days. Ob- servations were discontinued randomly. If the mouse being tested was observed nesting next to the mouse of its own .species, it was recorded as a positive observation, otherwise as a negative observation. Only mice in breeding condition were used. Sympatric mice were from Leon County, Texas. Allopatric leucopus were from Bryan and Tillman counties, Oklahoma; allo- patric gossypinus were from Nacogdoches County, Texas. The results of these association experiments arc summarized in table 1. Sympatric leucopus fe- males and gossypinus males and females demon- strated a significant positive association with members of their own species of the opposite sex. Sympatric leucopus males associated with females of their own species more frequently than with gossypinus females but the deviation from the expected was insufficient to produce a significant x- value. Table 1 also presents the results of tests utilizing allopatric stocks of leuco- pus and gossypinus. Allopatric mice, in this instance, did not associate with members of their own species significantly more often than with members of the other species. (In the case of allopatric leucopus females, five of the six tested showed a preference for individuals of the op- posite species.) This suggests that existing iso- lating mechanisms are being reinforced (Koopman, 1950) in sympatric areas. McCarley (1963) pointed out that in areas where leucopus and gossypinus are sympatric, the general restriction of leucopus to upland habitats (as opposed to the situation in allopatric Table 1. Results of discrimination tests using three mice in a four-compartmented cage No. of tests No. of individuals tested Positive observations Negative ob.senations X- values Sympatric Sympatric Sympatric Sympatric Allopatric Allopatric Allopatric Allopatric leucopus males leucopus females gossypinus males gossypinus females leucopus males leucopus females gossypinus males gossypinus females 36 19 113 78 3.010 34 16 330 65 88.000 54 18 233 126 15.605 20 12 126 46 18.604 14 7 69 36 4.8,W 9 6 25 69 10.297 20 10 81 52 2.925 12 S 54 64 0.423 331 397 332 NOTES AND COMMENTS areas where leucopus occupies both uplands and lowlands) was the result of the presence of gossypinus in lowlands. The presence of an ethological mechanism in the form of species discrimination would support this hypothesis. This study was supported by Grants No. G- 8019 and G-19387 from the National Science Foundation. In addition to Austin College, facili- ties at the University of Oklahoma Biological Station and Southeastern Oklahoma State College were provided while I was in residence at these institutions. Literature Cited Blair, \V. F., .and W. E. Howard. 1944. Ex- perimental evidence of sexual isolation between three forms of mice of the cenospecies Peromyscus maniculatus . Contrib. Lab. Vert. Biol., ijniv. Michigan, No. 26: 1-19. Bradshaw, VV. N. 1957. Reproductive isolation in the Peromyscus leucopus group of mice. M.A. Thesis, Univ. of Texas, Austin. Calhoun, J. B. 1941. Distribution and food habits of mammals in the vicinity of the Reelfoot Lake Biological Station. Proc. Tenn. Acad. Sci., 6: 207-225. Dice, Lee R. 1937. Fertility relations in the Peromyscus leucopus group of mice. Contrib. Lab. Vert. Gen., Univ. Michigan, No. 4: 1-3. . 1940. Relations between the wood mouse and the cotton-mouse in eastern Virginia. J. Mammal., 21: 14-23. Howell, A. H. 1921. A biological survey of Alabama. U. S. Dept. of Agric. Bur. Biol. Surv., North Amer. Fauna, No. 45. KooPMAN, K. F. 1950. Natural selection for reproductive isolation between Drosophila pseudoobscura and Drosophila persimilis. Evolution, 4: 135-148. McCarley, Howard. 1953. Biological relation- ships of the Peromyscus leucopus species group of mice. Ph.D. Thesis, Univ. of Texas, Austin. . 1954a. Natural hybridization in the Peromyscus leucopus species group of mice. Evolution, 8: 314-323. . 1954b. The ecological distribution of the Peromyscus leucopus species group in eastern Texas. Ecology, 35: 375-379. . 1963. The distributional relationships of sympatric populations of Peromyscus leucopus and P. gossypinus. Ecology, 44: 784-788. 398 ACTIVITY, FOOD CONSUMPTION AND HOARDING IN HIBERNATORS By Charles P. Lyman In previous papers it has been emphasized that the phenomenon of hiberna- tion is not precisely the same among all mammals and differs even among various famihes of rodents. For example, the electroencephalograms of "le sper- mophile" (probably Citellus citellus) and the woodchuck (Marmota monax) differ markedly in hibernation and during the process of arousal from that of the golden hamster (Mesocricetus auraius) (Kayser, Rohmer and Hiebel, 1951; Chatfield, Lyman and Purpura, 1951; Lyman and Chatfield, 1953). The rela- tively active electroencephalograms of the deeply hibernating ground squirrel and woodchuck reflect the behavior of these species, for they respond to strong stimuli with uncoordinated muscular movements. The deeply hibernating ham- ster, on the other hand, is totally inert until the process of arousal is well under way. Another important difference concerns the nutritional requirements during the hibernating period. Most rodents that hibernate, including the woodchuck and the ground squirrel, become extremely fat before the period of dormancy and apparently Uve on this fat during hibernation (Kayser, 1950). In contrast, the golden hamster, at least in the laboratory, loses weight when exposed to cold and actually enters hibernation when quite lean (Lyman, 1948). Although Citellus 399 546 JOURNAL OF MAMMALOGY Vol. S5 No. 4 tridecemlineatus (Howell, 1938) and many other hibernators store some food, storage appears to be of paramount importance in the hamster, for it eats during its periodic arousals from the hibernating state and apparently cannot live throughout the hibernating period if there is no food available (Lyman and Leduc, 1953). The experiments described below were designed to clarify these differences. The food and water intake and the activity of a series of hamsters and ground squirrels were measured when the animals were kept in a warm environment and compared with the same measurements when the animals were exposed to cold and when they hibernated. During the course of the experiments it became apparent that hamsters which were denied food for storage did not hibernate as soon as control animals. There- fore, a second experiment was designed to elucidate this point. MATERIALS AND METHODS Experiment 1 . — A series of nine male hamsters (Mesocricetus auratus) between 15 and 20 weeks of age was housed in activity cages equipped with exercise wheels. The rotating wheel of each cage was 36 cm. in diameter and 11.5 cm. wide, and the resting cage measured 38 cm. long by 23 cm. deep by 23 cm. high. The animals were fed ground Purina chow from non-spillable food cups, and water was supplied from non-spillable water containers. Food and water intake and the number of revolutions of the activity wheel were usually measured each day. If very small amounts of water were consumed daily, allowance was made for evaporation from the water container. The animals were maintained under these conditions for five to nine weeks in the fall of the year in an animal stock room. The environmental temperature was 24 ± 2°C. and the animals were exposed to between eight and nine hours of hght daily. At the end of this period the hamsters were moved into a cold room main- tained at 5° ± 2°C., which was illuminated daily for eight hours. The measure- ments were continued under these conditions. A second series of two female hamsters of 15 weeks of age and two female ground squirrels (Citellus tridecemlineatus) over three years of age was housed in individual cages of the same size as the previous experiment. The cages were suspended from springs, and balanced from below by a tambour at each corner. Any major movement of the animal in the cage depressed one or more of the tambours. This was recorded pneumatically by means of a tambour and a stylus on a slowly revolving smoked chart. Food and water intake were measured as in the first experiment, and the animals were exposed to the same environmental conditions. In all experiments, when an animal was observed to be hibernating, a small amount of fine shavings was placed on its back so that, if it moved enough to displace the shavings, this could be noted at the next observation. Experiment 2. — In order to test the effect of hoarding on hibernation a series of six male and six female hamsters, 15 weeks old, was housed in individual cages supplied with ample shavings in the cold room (5° ± 2°C.). They were fed water ad libitum and ground Purina Chow in non-spillable food containers. 400 Nov., 1954 LYMAN— ACTIVITY AND FOOD IN HIBERNATORS 547 Under these conditions the animals were unable to store food, as is their in- variable custom if supplied with pellets of compressed food. These animals were observed daily and the day on w^hich they first hiberated was noted. To provide a controlled comparison with the experimental animals, the rec- ords of 373 animals that had been obser\^ed during the last few years in this laboratory w^ere used. The use of this large number of animals was Employed because it sheds some light on the variations encountered in a large group of animals. These hamsters were housed under conditions identical with the ex- perimental animals in the cold except that they were fed Purina laboratory chow checkers and hence could store their food. RESULTS The nine hamsters in the wheel-type activity cages consumed, in the warm room, an average of 7.8 gms. (Standard Deviation 1.7) of Purina Chow each day. Omitting one animal that spilled water in spite of all precautions, the water consumption averaged 11.1 cc. (S.D. 1.2) per day. The animals averaged 7335 revolutions (S.D. 1834) of the wheel each 24 hours. It was apparent that most of the activity took place at night, whether the animals had an exercise wheel or whether the motion of the cage was recorded. There was no clear evi- dence of regular periodic activity other than the diurnal cycle. When moved to the cold room, the food intake of the hamsters invariably rose and averaged 12.9 gms. (S.D. 2.3). This rise is statistically highly significant according to the "t" test, for P is less than .01. The water intake rose to 16.3 cc. (S.D. 4.7) per day. This rise is also highly significant in spite of the fact that one animal showed a slight decrease in water intake (from 11.3 to 10.5 cc./day). The hamsters were no more active in the cold than in the warm room as meas- ured by the activity w^heels, averaging 7141 revolutions (S.D. 3029) per day. Furthermore, there was no evidence in the tambour records that the animals moved about in their cages to a greater extent when exposed to cold. Although most of the hamsters provided with activity wheels were kept in the cold room for over a year, only one animal hibernated on and off for 54 days after being in the cold for 44 days. During this period it was observed to be awake on 21 different days and the longest period of continuous hibernation was three days. In contrast, the average period of hibernation, punctuated by brief waking periods, for 19 typical controls was 95.1 days (S.D. 21.7). Furthermore, most hamsters remain continually in the hibernating state for protracted periods, the longest recorded in this laboratory being 21 days. There was no diminution of exercise or food and water intake in the days just prior to entering the hibernating state. On each awakening the animal ate, drank and ran on the exercise wheel. Considering only the 21 days that the animal was awake, it averaged 4207 revolutions of the wheel, 12.7 gms. of food, and 25 cc. of water per day. This animal averaged 11,646 revolutions of the wheel, 14.8 gms. of food and 21.9 cc. of water per day in the cold room before entering hiber- nation. One of the hamsters maintained in the tambour-recording cages hibernated 401 548 JOURNAL OF MAMMALOGY Vol. 35, No. 4 after only 11 days exposure to cold, but the total hibernating period lasted only 18 days during ten days of which the animal was observed to be awake. The period in the cold room before hibernation took place was too short to give a reliable figure, but the food intake increased and the water intake decreased during the 11 days. There was no evidence that food and water intake or exercise decreased in the last few days just prior to hibernation. In the ten days the animal was awake during the hibernating period it averaged 8.4 gms. of food and 10 cc. of water per day. This compares with 8.1 gms. of food and 5.1 cc. of water per day in the eleven days before hibernation. The hamster in the other tambour-recording cage adopted a peculiar storing habit which spoiled the records of food intake and caused the abandonment of records on this animal, but also led to Experiment 2 of this series. This hamster was able to obtain food from the unspillable food cup by forcing its muzzle into the ground dog chow. Upon lifting its head from the food cup, it removed the crumbs clinging to its vibrissae with its fore paws and transferred them to its cheek pouches. After it had obtained a satisfactory amount of food in this way, it carried it to a far corner of the cage and stored it in the manner observed in hamsters fed sohd pellets of food. This animal hibernated on the 72nd day after exposure to cold. Ground Squirrels. — In the 62 days of observation prior to exposure to cold, ground squirrel no. 1 averaged 7.0 gms. of food per day and 12.2 cc. of water. During this period its weight dropped from 266 gms. to 237 gms. Ground squirrel no. 2 averaged 10.9 gms. of food per day and 22.5 cc. of water, and its weight dropped only from 275 to 267 gms. Both animals were very fat at the start of the experiment. In sharp contrast to the hamsters, the two ground squirrels hibernated within 24 hours after being exposed to cold. Ground squirrel no. 1 hibernated over a period of 62 days, at the end of which time the animal was moved to the warm room because of its emaciated condition. Of the 62 days, it was observed to have moved enough to displace the shavings on its back on 14 days and was active enough to cause movement of the cage for a total period of 171 hours, or 12 per cent of the time. During the time it was awake it ate a total of 16 gms. of food and drank 87 cc. of water. It lost 111 gms. body weight during the 62 days of the sojourn in the cold. Ground squirrel no. 2 hibernated over a period of 120 days after which it was moved to the warm room. During this period it was observed to have lost the shavings from its back on 23 days and was active enough to move the cage for a total of 225 hours or 8 per cent of the time. During the time it was awake it ate a total of 6 gms. of food and drank no water. It lost 130 gms. body weight during the 120 days. Storing Experiment. — The twelve hamsters that were moved to the cold room but not allowed to store food hibernated after an average period of 97.6 days (S.D. 20.4). The earliest hibernation occurred on the 54th day and the latest on the 114th day. The average time before entering hibernation was approximately the same for both sexes. Once in the hibernating state, the animals remained 402 Nov., 1954 LY^^IAN— ACTIVITY AND FOOD IN HIBERNATORS 549 in this condition, with the usual periodic awakening, for a period of about three months. Thus the length of the hibernating period, once started, was not cur- tailed. Of the 252 control animals that entered hibernation the average time before the hibernating state occurred was 56.6 days (S.D. 33.2). The shortest period before hibernation occurred after moving to the cold room was three days and the longest was 218 days. On the other hand, 121 animals of the control groups died before entering the hibernating state. The average time when death oc- curred after moving to the cold room was 71.5 days (S.D. 67.2). In a statistical comparison between the time of onset of hibernation in the 12 hamsters deprived of storing and the 252 controls that hibernated, P is less than .01. This highly significant difference shows that the abihty to store food has a profound effect on the occurrence of hibernation. On the other hand, the data on the other 121 control animals show^ that in a large population of hamsters in the laboratory, some animals will Uve in the cold and eventually die T\athout enter- ing hibernation. Therefore it may be more accurate to include the animals that died in the cold with the other control animals. If the total span of life in the cold of these animals is averaged with the total time in the cold before hiberna- tion in other controls, one obtains a figure of 61.4 days (S.D. 47.4). When this figure is compared with the 97.6 days for the experimental animals the difference between the two figures is still found to be highly significant according to the Fisher "t" test (P <.01). In other w^ords, the possibility that the marked delay in the onset of hibernation in the non-storing hamsters is due to chance is less than one in 100. DISCUSSION The increased food and water intake of hamsters when exposed to cold is tj''pical of mammals that do not hibernate, and is to be expected mth the in- crease of metabolic rate. That the metabolic rate does increase in hamsters when exposed to cold can be seen by comparing the figure 1014 cc. 02/K/hr. for golden hamsters at 29.6° (Kayser, 1940) or 930 cc. Os/K/hr. at 30-34°C. (Adolph and Lawrow, 1951) with 2877 cc. 02/K/hr. at 5°C. (Lyman, 1948). These reactions are in sharp contrast to the almost immediate hibernation and cessation of eating and drinking that took place w^hen the two ground squirrels were exposed to cold. Although ground squirrels do not invariably hibernate within 24 hours after exposure to cold, as Johnson (1930) has amply demon- strated, still the onset of hibernation is usualty quite rapid. On the other hand, of the more than 1000 hamsters exposed to cold in this laboratory, the shortest recorded time before hibernation occurred was three days, with most animals far exceeding this figure. The hibernating phase of the two hamsters recorded here cannot be regarded as typical, because the period was curtailed and the animals awoke more often than is normal. However all indications are that golden hamsters keep up high food and water intake and remain active up to the time they hibernate. Further- more, the data emphasize that the hamsters will consume considerable nourish- 403 550 JOURNAL OF MAMMALOGY Vol. 35, No. 4 ment each time they wake from hibernation. Although observations in the wil'd on the golden hamster are lacking, the European hamster is known to be an inveterate hoarder of food, so much so that the German word "hamster" means to hoard or store in anticipation of need. It is highly probable that hamsters do not have access to water when they are holed up for the winter, and it is reasonable to suspect that metabolic water and water from the stored food are their only sources. The data indicate, however, that the animals will drink a considerable quantity if water is available. The second experiment demonstrates clearly that the ability to store food has a profound effect on the time of onset of hibernation. Waddell (1951) has shown that noxious conditions such as illuminating the food bin will increase the stor- age of food by hamsters, and McCleary and Morgan (1946) have demonstrated similar reactions in the rat when it is exposed to cold. The denial of the ability to hoard when stimulated to do so by cold must upset the behavioral pattern of the animal so that the onset of hibernation is delayed, or, in the case of one animal in the cage on tambours, the period of hibernation abnormally shortened. This takes place in spite of the fact that there is obviously ample food obtainable at all times from the food containers. Again in contrast to the hamsters, the ground squirrels hardly touched food or water while they were exposed to cold and the fast is reflected in the great loss of body weight. This loss of weight bears out the recent work of Kayser (1952) with the European ground squirrel (Citellus citellus). He was able to show that the large weight losses observed in animals during the hibernating period were not due to loss of weight during hibernation, but were directly correlated with the amount of time the animals were awake during that period, for the process of arousal and the awake condition consumed a great deal of energy. Thus in ground squirrel no. 1, which was awake a much larger proportion of the time, the weight loss was much faster (1.8 gms./day) than in ground squirrel no. 2 (1.1 gms./day). Although laboratory experiments can not completely clarify the conditions that occur in the wild, it seems justifiable to draw from these results a sharp distinction l^etween the preparation for hibernation in animals such as the ground squirrel on the one hand, and the hamster on the other. Under stimuli which have yet to be elucidated, the former animals grow extremely fat in the period previous to hibernation. Exposure to cold is then apt to cause almost immediate hiberation. If hibernation does not occur at once, denial of food will hasten it (Johnson, 1930, confirmed in these laboratories). Although the ground squirrel stores food (Howell, 1938), it is apparent that it can survive repeated awakenings with practically no nourishment by utilizing its stored fat. From all indications the woodchuck (Marmota monax) is very similar to the ground squir- rel, for it becomes extremely obese by fall, and does not even store food in its burrow (Merriam, 1884). The hamster, on the other hand, apparently makes no physiological prepara- tion for hibernation until actually exposed to cold. Under the stimulus of cold exposure, it stores food and loses weight. If food storage is denied, hibernation 404 Nov., 1954 LYMAN— ACTIVITY AND FOOD IN HIBERNATORS 551 is delayed. Food storage is essential for the maintenance of the hamster during the hibernating period, for the energy necessary for the periodic awakening is soon exhausted if the stored food is removed and the animal cannot eat during the periods of activity (Lyman and Leduc, 1953). Unpubhshed experiments in these laboratories have conclusively shown that, in contrast to the ground squirrel, denial of food or water will never cause hibernation in the hamster, which is precisely what might be expected in view of the importance of stored food. Thus the golden hamster is provided with a check against hibernation before food has been stored, which, under natural conditions, must protect the animal against starvation during the hibernating period. How the lack of food hoarding can delay hibernation is a problem in itself. Indications are that it is not the only psychic factor which can influence the onset of hibernation, for of the nine animals with activity wheels only one hiber- nated for a very short period, though all were exposed to cold for over a year. Possibly exercise on activity wheels itself may halve an effect on the onset of hibernation. In any event, it is evident that a behavior pattern such as storing can have a profound effect on the onset of hibernation in the hamster. That behavior must be considered along with all the physiological factors when at- tempting to resolve the cause of hibernation certainly compounds the com- plexity of the problem. SUMMARY In a series of experiments with golden hamsters (Mesocricetus auratus) and thirteen-lined ground squirrels {Citellus tridecemlineatus) it was found that the former animals did not hibernate at once, but increased their food and water intake when moved to a cold environment, while the amount of exercise taken remained unchanged. In the two hamsters that eventually hibernated there was no decrease of eating, drinking or exercise up to the time of hibernation. The hamsters ate and drank considerable quantities and exercised to some extent on each awakening from hibernation. Hamsters that were prevented from hoard- ing food showed a marked delay in entering hibernation. In contrast, the ground squirrels hibernated within 24 hours after exposure to cold. In their periodic wakenings the ground squirrels ate and drank almost nothing and in consequence lost much of their stored fat. It is concluded that the ground squirrel is stimulated in some unknown way to prepare for hibernation by becoming extremely obese. Exposure to cold brings on hibernation quickly and the animal lives almost exclusively on his stored fat during the hibernating period. The golden hamster, on the other hand, does not fatten prior to hibernation, but is stimulated by cold to store food. This food is used during the periodic arousals in the hibernating period, and without the store the animal would perish. Denial of the ability to hoard causes a delay in the onset of hibernation and thus must provide the animal with a check against premature hibernation that would result in starvation. Acknowledgement. — This research was supported by a grant under U. S. Air Force Contract AF 33 (038)-18133. 405 552 JOURNAL OF MAMMALOGY Vol. 35, No. 4 LITERATURE CITED Adolph, E. F. and J. W. Lawrow. 1951. Acclimatization to cold air; hypothermia and heat production in the golden hamster. Am. Jour. Physiol., 166: 62-74. Chatfield, p. O., C. p. Lyman, and D. P. Purpura. 1951. The effects of temperature on the spontaneous and induced electrical activity in the cerebral cortex of the golden hamster. Electroencephalography and Clinical Neurophysiol., 3: 225- 230. Howell, A. H. 1938. Revision of the North American ground squirrels. N. A. Fauna No. 56. Johnson, G. E. 1930. Hibernation of the thirteen-lined ground squirrel Citellus tri- decemlineatus (Mitchill). V. Food, light, confined air, precooling, castration and fatness in relation to production of hibernation. Biol. Bui., 59: 114-127. Kayser, C. 1940. Les ^changes respiratoires des hibernants. Theses, Fac. d. sci. d. rUniversit^ de Strasbourg, 1-364. Also in Annales de Physiol, et Physicochimie biologique, 15 and 16, 1939, 1940. . 1950. Le sommeil hibernal. Biol. Rev., 25: 255-282. . 1952. La d^pense d'^nergie des mammif^res hibernants pendant toute la dur^e de I'hibernation. Arch. d. Sci. Physiol., 6: 193-212. Kayser, C, F. Rohmer and G. Hiebel. 1951. L'E.E.G. de I'hibernant. L6thargie et r^viel spontan^ du spermophile. Essai de reproduction de I'E.E.G. chez le sper- mophile r^veill6 et le rat blanc. Rev. Neurologique, 84: 570-578. Lyman, C. P. 1948. The oxygen consumption and temperature regulation of hibernating hamsters. Jour. Exp. Zool., 109: 55-78. Lyman, C. P., and P. O. Chatfield. 1953. Hibernation and cortical electrical activity in the woodchuck {Marmota monax). Science, 117: 533-534. Lyman, C. P. and E. H. Leduc. 1953. Changes in blood sugar and tissue glycogen in the hamster during arousal from hibernation. Jour. Cell, and Comp. Physiol., 41 : 471-492. McCleary, R. a. and C. T. Morgan. 1946. Food hoarding in rats as a function of en- vironmental temperature. Jour. Comp. Psychol., 39: 371-378. Merriam, C. H. 1884. The vertebrates of the Adirondack Region. Trans. Linn. Soc. N. Y., 2:9-214. Waddell, D. 1951. Hoarding behavior in the golden hamster. Jour, of Comp. and Phys- iol. Psychol., 44:383-388. Department of Anatomy, Harvard Medical School and Museum of Comparative Zoology, Harvard University. Received December IS, 1953. 406 RADIO TRANSMITTER-COLLARS FOR SQUIRRELS' ROGER O. BEAL, Waterloo Wildlife Experiment Station, Ohio Division of Wildlife, New Marshfield Abstract: An adjustable plastic transmitter-collar suitable for the gray squirrel {Sciurus carolinensis) and fox squirrel {Sciurus niger) was developed. It can be attached at the trap site within 5 min. Placed on 23 squirrels, the transmitter-collar permitted 35-45 days of radio-tracking data to be ob- tained from each squirrel. Transmitter-collars have been used suc- cessfully on rabbits, hares, and raccoons ( Mech et al. 1965 ) . This paper describes a transmitter-collar which was found satis- factory for use on the gray squirrel and fox squirrel. The author believes that this same type of apparatus can be attached to other mammals. The electronic components for a squirrel transmitter-collar were mounted on a self- locking, 11-inch adjustable plastic hospital ID bracelet (purchased from Aloe, 1831 Olive St., St. Louis, Missouri). The com- ponents were then waterproofed with syn- thetic latex rubber and wrapped with a layer of plastic tape ( Fig. 1 ) . The antenna loop was made the same size on each transmitter-collar; its circum- ference was measured for a snug fit over the head of the largest anticipated squirrel. The plastic bracelet, not the antenna loop, is the adjustable component of the' collar (Fig. 1). Average weight of the completed trans- mitter-collars was 0.75 oz. The transmitter battery life averaged 35-45 days, with a ^ A contribution from Ohio P.-R. Project W- 105-R. range of % to VA miles, as determined by the Drake 2-B receiver and two Hy-Gain 5-element beams. The behavior of a penned gray squirrel with a transmitter-collar attached during a 2-month observation period appeared nor- mal. The squirrel ate regularly and moved without hindrance. At the end of the test period, the squirrel was in good physical condition and the neck hair was not worn. In the wild, squirrels were livetrapped in a wooden box trap, and removed from the trap with heavy leather gloves. The activated transmitter-collar was slipped over the head of the squirrel, adjusted to the correct fitting, permanently snap-locked, and the loose end of the collar cut off. The squirrel was ear-tagged and then released. This entire procedure was performed at the trap site by two workers, and required 5 minutes or less. Between February, 1964, and August, 1966, transmitter-collars were attached to 23 squirrels (6 adult gray males, 11 adult gray females, 4 adult fox males, 1 adult fox female, and 1 juvenile fox male). In 15 cases squirrels were recovered with the transmitter-collar attached. In all but one 407 374 Journal of Wildlife Management, Vol. 31, No. 2, April 1967 Fig. 1. Construction sequence of transmitter on adjustable plastic hospital ID bracelet. adult male fox squirrel which was re- trapped after 1 month of radio-tracking. Its neck had a deep cut, possibly caused by the edge of the collar. The transmitter- collar was removed and the squirrel re- leased. The animal was retrapped 4 months later and a transmitter-collar was attached after examination showed that the neck had completely healed. There is evidence that the plastic brace- let will break and free the transmitter after the duration of batter>' life. A gray squirrel was trapped in May, 1965, tagged and re- leased. In August, 1965, this animal was retrapped without the transmitter-collar. Examination of several recovered collars showed that they were brittle after being worn for a month or more. In three instances collars were chewed, presumably by another squirrel. However, this did not appear to be a serious problem. The transmitter-collar described meets the requirements for continued study by radio telemetry of squirrel behavior in southeastern Ohio forests. LITERATURE CITED Mech, L. D., V. B. KuECHLE, D. W. Warner, AND J. R. Tester. 1965. A coUar for at- taching radio transmitters to rabbits, hares, and raccoons. J. Wildl. Mgmt. 29(4) :898- 902. case, the squirrels appeared in good physi- cal condition. The one exception was an Received for publication September 14, 1966. 408 SECTION 5— PALEONTOLOGY AND EVOLUTION If there be one unifying principle that pervades all of biology, it is that of evolution. Not only is this evident in consideration of the papers here repro- duced (which range from one that deals in part with intrapopulational varia- tion up to those concerned with higher taxonomic categories), but it also is evident in the contents of virtually all other papers chosen for inclusion in this antholog)'. The few selections in this section, then, provide but a glance at some aspects of mammalian evolution. Linked inseparably with the e\olutionary process is the fossil record, which is unusually good for some groups of mammals and provides much of the raw data for phylogenetic considerations. For papers relating to paleontolog}% we have chosen one on a local fauna ( Hibbard ) , one ( Wilson ) that alludes to the importance of sound geographic and stratigraphic data and that ties in with the historic record, one (Radinsky) that deals with evolution and early radia- tion of perissodact>'ls, and two on rodents, one a classic early paper by Miller and Gidle>^ in which the major groups of that extremely complex order are outlined, and the other a modern treatment of the same problem by Wood ( see also Wood, 1959). The study by Guthrie compares evolutionary change in mo- lar teeth, using both fossil and Recent species oi'Microtiis, and thus stresses the on-going evolutionary process. The short paper by Reed clearly presents an interesting problem arising from attempts to classify some early relatives of man. The essay by Durrant and Hansen places biogeography in the evolutionary framework. Some of the species of ground squirrels mentioned by them are treated also in the serum protein analysis of Nadler and Hughes included in Section 1. The paper by Jansky is interesting because it provides an excellent example of evolutionary trends in features other than those directly related to "hard anatomy." The literature of mammalian evolutionary and paleontological studies is wddely scattered. Aside from the journals and bibliographic sources mentioned in the Introduction, the interested student should consult Evolution, a quar- terly journal published by the Society for the Study of Evolution, and the Journal of Paleontology. He should also be aware of the Bibliography of Fossil Vertebrates, 1959-1963, compiled by Camp et at. (as well as earlier vol- umes in the same series) and the News Bulletin of the Societ>' of Vertebrate Paleontolog}'. Romer's textbook, Vertebrate Paleontology (1966) and Simpson's (1945) The Principles of Classification and a Classification of Mammals are especially recommended as sources of considerable information on the fossil history and evolution of mammals, and we would be remiss not to mention also Zittel's 1891-93) classic Handbuch der Palaeontologie (volume 4, Mammalia). Three substantial longer papers on systematics and evolution of special groups are Shotwell's (1958) study of aplodontid and mylagaulid rodents, Dawson's (1958) review of Tertiary leporids, and Black's (1963) report on the Tertiary sciurids of North America. Extensive paleofaunal studies of note are many; 409 those by Hibbard (1950) on the Rexroad Formation from Kansas and by Wilson (1960) on Miocene mammals from northeastern Colorado serve as excellent examples. 410 South African Journal of Science Suid-Afrikaanse Tydsl.?> microns. The measurements were quite repeatable. The exterior edge of the enamel was used in all measurements. Pictures of both left and right sides were taken of M. pennsylvanicus. The side with the picture of highest con- trast was used, and, if there was any ques- tion, measurements were taken on both sides. Rarely was there a break or crack on both sides so that no measurement could be taken. Measurements were made as illustrated in Fig. 1. The measurements on the whole were well defined. The only possible ex- ceptions were the anterior part of Mi and the posterior part of M^. However, this is a function of their variability in form. Several measurements were used on the anterior part of Mi and posterior part of M'', but no one expresses adequately the vari- ation in shape. The width measurements for each tooth are numbered serially from the loop. Con- sequently, the uppers are numbered from anterior to posterior and the lowers from posterior to anterior. The total length is designated by L and the anterior and pos- terior lengths by a and p, respectively. Forty-two measurements were taken on each individual, 20 measurements on the uppers and 22 on the lowers. Discussion of Molar Variations The variation in Mi, M-, and M^* is represented in Fig. 2. The teeth viewed from left to right depict the nature and ex- tent of the shape variations present in these samples. In reality, this variation does not fall into discrete classes as portrayed in Fig. 2; rather, each tooth in the figure represents a point along the variation con- tinuum. The most variable portions are cross-hatched to facilitate the comparisons. Notice that in Mi the rounded primordium on the lower part, actually the anterior part of the tooth, is utilized to construct new- salient angles by the penetration of re- entrant angles into its lateral margins. In the M- a new salient angle is formed by the budding off of the extreme posterior part of the crown, and varies in these sam- ples all the way from absence to almost the size of the other salient angles. M. paro- perarius has only a slight suggestion of this bud in some individuals, with most not having it at all. In M. pennsylvanicus this rudimentary stage is present only at a low frequency, most of the individuals having a well-developed salient angle. The cross-hatched area in the posterior portion of the M-^ behaves differently than the cross-hatched area in the Mi. M'' in- creases its number of salient angles phylo- genetically by dropping a bud posteriorly 474 220 R. D. GUTHRIE c.v. 12 3 12 3 12 3 4b P dp dp L L L C.V. 12 3 12 3 I 2 3 4 S a p a p a p L L L C.V. 12 3 12 3 12 3 4 5 a p a p d p L L L C.V. 2 3 12 3 12 3 4b a p a p a p L L L Fig. 3. Coefficients of variation (C.V.) of the upper molars of M. paroperarius (sample 1) and M. pennsylvanicus (samples 2-4) ; samples are identified in text. The tongue inserts are equal to two standard errors in each direction. The measurements at the base of each histo- gram correspond to those in Fig. 1. 475 VARIABILITY IN MICROTUS MOLARS 221 c.v. 1 2 3 4 S 6 ? 12 3 12 3 a p a p 3 p L L L 12 11 . 10 ■ J ■ M, 1 C.V. ' ■ - J_l ll 1 nWJ 1 UH 1 ■1 23.29 Sample 2 12 3 4 5 6 7 12 3 12 3 C.V. 12 3 4 5 6 7 12 3 12 3 1,2,3 a p J p a p L L L Sample 4 C.V. M 1^1 2 3 iAA 12 3 4 5 6 7 12 3 12 3 a p a p a p Fig. 4. Coefficients of variation (C.V.) of the lower molars of M. paroperarms (sample 1) and M. pennsylvanicus (samples 2-4) ; samples are identified in text. The tongue inserts are equal to two standard errors in each direction. The measurements at the base of each histo- gram correspond to those in Fig. \. 476 222 R- D. GUTHRIE and enlarging it lingually. However, on the of variation than any of the other width labial side, the penetration of the re-entrant measurements of either M^ or M^. This is angles and the outgrowth of the salient the incipient angle which is predominantly angles act in a manner much the same as present in M. pcnnsylvanicus and expressed in the Mi. There is very little difference in in some individuals of M. paroperarius as a principle in the mode of addition of salient rudimentary bump. angles in any of these teeth, only slight In every case in the upper molars the variations in detail. anterior length is less variable than the These cross-hatched areas are the ones posterior length. Fig. 3, (a) and (p) re- that vary most between species. For exam- spectively. In the case of M^ in samples pie, the M^ tooth pattern at the extreme three and four, which represent males and right in Fig. 2 is present in only one indi- females from one series, the difference be- vidual in the samples of M. pennsylvanicus, tween (a) and (p) is not outstanding. The but is the most common tooth pattern in difference between the coefficients of vari- M. chrotorrhinus. Komarek (1932) reports ation of the anterior and posterior length is a specimen of M. chrotorrhinus which has greatest in M'', which has no overlap at two one less angle in the M^ than usual. This standard errors in either direction. The en- specimen would correspond to the most tire length measurements (L) of M^ and common M. pennsylvanicus pattern. In ad- M- appear to have about the same magni- dition to M. pennsylvanicus, several other tude of variation. The length measurement species of Microtus have hints of the pos- of M^ has a larger variation in all cases teriolingual bud on the M-, and in M. cali- than either M^ or M^. It will be remem- jornicus it is of creditable magnitude bered that the upper molars add to the (Hooper and Hart, 1962). A further dis- tooth complexity from the posterior mar- cussion of the intrageneric variations in gins. From the findings here it may also Microtus is given by Hooper and Hart in be stated that these phylogenetically varia- the preceding reference. ble posterior areas of the uppers have the There is some overlap in shape between greater intrapopulational variability, the fossil M. paroperarius and the recent The uniformity of the four samples M. pennsylvanicus. Referring to Fig. 2, in would seem to increase with the order in Ml the third pattern from the left, in M- which they are listed, as there are progres- the second, and in M^ the fourth are com- sively fewer collecting restrictions imposed, mon to both species. However, it must be The fossil M. paroperarius sample was kept in mind that the discrete patterns taken from several localities and with some illustrated here are only chosen points along temporal variation involved. The second a continuum. sample, of M. pennsylvanicus, was taken The 42 different measurements are repre- over a wider territory than samples three sented by histograms in Figs. 3 and 4. The and four, which were collected near a small most striking pattern is the high variation city. Since there is a high interpopulational in the width measurements in the anterior variation in M. pennsylvanicus, even within part of Ml and the posterior part of M'^. the same subspecies (Snyder, 1954), the Although this varies slightly in magnitude difference in uniformity of the collecting between samples, the general pattern is restrictions might be thought to affect the much the same. The width measurements relative amount of within-sample variation, of M^ have relatively low coefficients of With but one or two exceptions, the mea- variation, all under six. The M^ width surements did not show this expected vari- measurements also have relatively low co- ational gradient between samples. There efficients of variation. The width measure- also proved to be no pattern differences ment number three of M^, which includes of appreciable magnitude between the two the incipient angle, has a larger coefficient sexes of M. pennsylvanicus. 477 VARIABILITY IN MICROTUS MOLARS 223 In the uppers the measurements of M. paropcrarhis tend to be more variable than the samples of M. pcnnsylvaniciis, especially the posterior part of M^ where the co- efficient of variation is about double, at least in the width measurements. In the width measurements of the phylogenetically more stable teeth IVP and M- there is no notable difference in magnitude between M. paroperarius and M. pennsylvanicus. The M- widths have a relatively low to moderate variation, with a coefficient of variation of about six or less, and no out- standing pattern within the tooth. M3 width measurements tend to be more vari- able than those of the M2 with the anterior width measurements having the greater variation. The coefficients of variation are very large in the anterior part of Mi (note width measurements five, six, and seven). Another peculiarity of Mi in M. pennsyl- vanicus is that the width measurements in the midsection of this tooth are less vari- able than either the anterior or posterior ones. Some of the other teeth show this to a minor degree (note M^ and M-). In the lowers the anterior length measurements (a) are more variable than the posterior length (p) in every case except the Mo of sample four. Unlike the uppers, the lowers add on to the anterior margins of the teeth, and we may conclude from the coefficients of variation in Fig. 4 that these anterior areas of the lower molars also have the greatest variation. In both the posterior lengths (p) and the whole lengths (L) there is a trend toward greater variation in an anterior to posterior direction in both the uppers and lowers. This is not so well marked in the anterior length (a) measurements. Of the measurements of the entire tooth length, the length (L) of M3 is the most variable in M. pennsylvanicus while the length (L) of M^ is the most variable in M. paroperarius. This is a case where the patterns produced by the length variations (L) are somewhat misleading. In M. penn- sylvanicus M^ is the upper tooth with the most variation, which both the width and the length measurements suggest. Mi, on the other hand, is the most variable tooth in form among the lowers. This is evident in the width measurements but does not show up in the length (L) measurements of M. pennsylvanicus. Although Mi is the most variable lower tooth it has developed a long stable posterior area which dampens the variations occurring at the anterior part of the tooth, thereby producing a decep- tively low coefficient of variation for the entire tooth length. This effect is not present to the same degree in the Mi of M. paroperarius (see Fig. 2). At this early phylogenetic stage the tooth has a rela- tively smaller stable posterior section. M3 has a relatively higher variability than the other phylogenetically more stable teeth M^ M-, and M2. It is the one tooth that crosses over the incisor root and has a limited role in adding to the crown com- plexity of the tooth-row, and may even be in a state of reduction in this particular genus. In some other genera of microtines, Dicrostonyx for example, the incisor root does not cross over in this fashion and the M3 has developed a more complex crown pattern. Also, it is not reduced in size laterally as it is in Microtus. These facts suggest that the peculiar relationship of M3 to the incisor places some limitations on its potential for increased complexity. In many of the cricetines both the upper and lower third molars have undergone considerable reduction; this is not the case in Microtus. Some individuals of M. penn- sylvanicus have a longer M^ than M^. In summary then, a quantification in these two species of the molar variability reveals an overall pattern of higher vari- ation in the posterior parts of the upper molars and the anterior parts of the lowers. The greatest amount of variation is present in the anterior end of Mi and the posterior end of M^. A direct positive correspon- dence exists between those areas of the teeth which are changing phylogenetically and those which exhibit a greater magni- tude of variation. 478 224 R. D. GUTHRIE Supporting Evidence The significance of a positive association between the rapidly evolving tooth charac- ters and a relatively high variability in Microtus is dependent upon its general ap- plicability. This may be either a special case or an expression of a more general phenomenon. The following is a presenta- tion of evidence supporting its more general nature. In the microtines this association is not limited to the M. paroperarius-pennsyl- vaniciis line, but rather it is a common feature of the whole group. Dicrostonyx has the most complex crown pattern of the sub- family. D. torqiiatus, the species repre- sented in the second phase of the last glaci- ation (Zeuner, 1958), has a variable ex- pression of new salient angles on the pos- terior margin of M^ and M- and the an- terior margin of Mo and M3. These salient angles are highly variable in their occur- rence, grading to complete absence in some individuals. The characteristic species of the last glaciation, phase one (early Wis- consin), was D. henseli, which did not possess the salient angle or bud as did D. totquatus. This bud seems to be a nascent character developing through the last glacial age. D. groenlandicus, a recent species, has this character present in all individuals. D. hudsonius, a species with a distribution presently limited to the Hudson Peninsula, is a living relict representative of the D. henseli tooth pattern of the early part of the last glaciation. D. torquatus exists as the modern Old World collared lemming. Thus there is a chronological and geographi- cal representation of the stages of develop- ment of this salient angle. The fossil D. henseli and the recent D. hudsonius do not have the salient angle. D. torquatus, both modern and fossil, has a varied expression of the salient angle from absent to fully present (Hinton, 1926). In populations of D. groenlandicus all individuals have it. Some taxonomists give these forms only subspecific status; however, the principle dealt with here remains valid. Kurten (1959) suggested that the aver- age rate of mammalian evolution during the Pleistocene was relatively higher than during the Tertiary. His analysis of the variability in several rapidly evolving groups, widely separated taxonomically, re- vealed an increase in the coefficient of variation in more lines than a decrease. Although his study did not deal in detail with the specific characters which are changing (he used an average of several measurements), it did serve to illustrate that rapidly evolving populations do not all tend toward morphological uniformity. On the contrary, it suggested the opposite. Wright, in the discussion at the end of Kurten's paper, proposed that recombina- tion is responsible for this ampHfication of potential variability. Skinner and Kaisen (1947) noted that while there are few diagnostic patterns in the evolution of Bison cheek teeth, there is a general trend toward the molarization of P4. The metastylid and median labial root of the P4 increase in frequency through time. In early fossil Bison these characters are virtually absent and in modern ones almost universally present. The increases in the complexity of P4 seem to have oc- curred over a relatively short period of time during the late Pleistocene. Since these evolving areas range from absent to fully developed in some populations during this period of incipiency, the variability is greater than that of the analogous areas of neighboring teeth. Simpson (1937) discussed a sample of ?)2> Eocene notoungulates, Henricosbornia lophodonta, which he considered to be from one population, since their variation is normally distributed and they are from the same horizon and locality. These were originally described by Ameghino as be- longing to 17 species, seven genera, and three families, principally on the basis of the variation present in the upper third molar. The variations present within this primitive form are characteristic of later species, genera, and families with which Ameghino was familiar. Here is an exam- ple of a considerable amount of variation 479 VARIABILITY IN MICROTUS MOLARS 225 in one population, the elements of which are later characteristic of higher taxa. It would be consistent with the evidence to assume that the tooth is undergoing evolu- tionary change in a manner which contrib- utes to the types characteristic of later higher taxa. Hooper's (1957) study of the dentition of Peromyscus gives supporting evidence to the main thesis proposed here of rapid evo- lution being accompanied by high pheno- typic variation. A series of P. mankulatus from Distrito Federal, Mexico, for example, has highly variable molars. The mesoloph and mesostyle patterns found in this one series resemble the common patterns of the other 17 species of Peromyscus studied. In other words, the mesostyle and mesoloph patterns observed in 1 7 species of Peromys- cus are also seen in this single series. P. mankulatus is first known from the Wis- consin age and has expanded its distribu- tion over a considerable part of North America. It is considered to be one of the "younger" species of Peromyscus (King, 1961), and therefore has recently under- gone evolutionary change at the species level. The occurrence of the crochet in horse teeth is another example of an incipient character that is highly variable in the same population (Simpson, 1953; Stirton, 1940). The acquisition of this plication is one of the first features in a general trend toward increased tooth complexity. The crochet, an anastomosing ridge between metaloph and protoloph, shows up in the Miohippus-Parahippus line. It is also pres- ent in some species of Archeohippus and sometimes in the milk teeth of Hypohippus (Stirton, 1940). The incipient crochet juts out as a peninsula or pier from the meta- conular part of the metaloph toward the protoloph. The degree of its development is extremely variable, from absence to a small spur extending halfway across, to a complete connection between the two lophs. The crochet varies both in frequency and extent between populations and within them, occurring in its various stages of representation in individuals of the same species at one locality. Butler (1952), speaking of the molariza- tion of premolars in Eocene horses, stated that the metaconule evolving in the pre- molars is most variable at the intermedi- ate stages of molarization. Wood's (1962) discussion of the tooth cusp variations in the early paramyid ro- dents showed that the hypocone is added to the tooth by two basically different means. In some forms it is derived from an enlargement of the posterointernal cin- gulum; in others it originates as a division of the protocone. Wood attributed these two distinctly different means of achieving fundamentally the same end product to a general selection toward the development of a posterointernal cusp irrespective of the nature of its origin. The addition of the fourth cusp, hypocone, is a common phe- nomenon in many lines during this part of the Tertiary, and seems to be correlated with the exploitation of more demanding food substances. Wood stated, "There is no question but that all of these variants may occur within a single genus and some- times within a single species." Here again, when a directional selection pressure is being applied, more phenotypic variation is exhibited in the incipient than in the non- incipient cusps. The lower third premolar is used to char- acterize various genera of fossil rabbits. Hibbard (1963) observed much variation within a primitive rabbit genus, Nekrolagus, and found at a low frequency a pattern of the P3 that is characteristic of modern genera. The common tooth pattern of Nekrolagus is also found at a very low fre- quency in some modern genera. This com- parative study documents a chronological frequency change in which the early fossil populations have the incipient characters represented at a low frequency and the modern populations at a high frequency. Here is another case in which there is a high variation associated with incipient characters, and the axis of this variation is parallel to phylogenetic change. 480 226 R. D. GUTHRIE Another opportunity to try the hypothe- sis is on the results of artificial selection experiments. If the hypothesis does ap- proximate the real condition, the character that is artificially selected for or against should behave in a manner similar to the evolving characters that have just been dis- cussed. That is, characters undergoing arti- ficial selection could be expected not to ex- perience a decrease in their phenotypic variation, but to maintain or even increase the variation. MacArthur (1949) selected for large and small size in mice using the weight at 60 days as a measure of size. In the un- selected control the coefficient of variation was 11.1. However, in the strain selected for large size it was 12.8, and in the small line 14.3. Falconer (1955) also selected for large and small size in mice using the sixth week weight as a measure of size. He stated, ''The phenotypic variability, also, does not reflect the expected decline of genetic vari- ance, and in addition reveals a striking and unexpected change in the small line." He further reported that the large line showed a slight increase in variation over the whole course of the experiment, although it re- mained relatively low compared to the vari- ation of the small line. The coefficient of variation in the small line increased to about double the original value between the seventh and ninth generations and re- mained at this high level. The realized heritability remained substantially constant up to the point at which response ceased. This phenomenon, he suggested, was due to the release of genetic variation through recombination. In their selection experiments for wing length in Drosophila, Reeve and Robertson (1953) found that the coefficients of vari- ation at the twentieth to seventy-ninth gen- erations were all below two in the unselected strain and all two or above in the selected strain. The strain selected for long wings showed an increase of about 50 per cent in total variance. They attributed this en- tirely to an increase in additive genetic variance, which rose about two and one- half times, while the absolute amount of other genetic variance remained about the same. This led them to suppose that selec- tion for long wing length would be far more effective in the selected than in the un- selected stock. Clayton and Robertson (1957), selecting for low and high bristle number in Drosoph- ila, concluded that "Selection had by no means led to uniformity, but in some cases even magnified the total variation." Robertson (1955) selected for thorax length in three stocks of Drosophila with about the same initial amount of variation. The coefficient of variation in the small lines increased immediately in the first gen- erations and was higher than the control in all three lines, although there were be- tween-strain differences in the pattern of increase in variation. In the large lines the variation fluctuated around that of the con- trol stock. Thus, in the large strains the changes in response to selection occurred without appreciable change in the coeffi- cient of variation, while the variation of the small line increased. Although the changes in variation ac- companying selection response in these ex- periments do not behave in a completely uniform manner, they do maintain and usually increase the initial magnitude of variation. Thus, evidence supporting the association between directional selection and a constant or increased variation is found both in rapidly evolving groups and in artificial selection experiments in which the degree of variational change has been recorded. The Theory and Model The most frequently employed explana- tion for an inordinate amount of variation is vestigiality. In such a case the charac- ters under consideration are not becoming more complex phylogenetically but are de- creasing in pattern complexity. Morpho- logical characters which are in the process of reduction or elimination exhibit more variation than do their more functional 481 VARIABILITY IN MICROTUS MOLARS 227 homologues. This high correlation between vestigial and highly variable characters no doubt influenced Hinton (1926) to believe the microtines to be, in tooth form, degen- erate descendants of the multituberculates and consequently undergoing reduction in tooth complexity. However, there is a time gap in the fossil record of some 35 million years between the multituberculates and microtines. The concept of the vestigial nature of microtine teeth has been perpetu- ated by some mammalogists (Goin, 1943; Hall and Kelson, 1959). But the position that microtines did not arise from a crice- tine stock and have not undergone a gen- eral increase in tooth complexity is untena- ble. Not only does the fossil record support an increase in microtine tooth complexity, but there is an almost complete continuum of recent intermediate forms between the Microtinae and Cricetinae. Vestigiality can be discounted as an explanation of the variation differential in the other examples as well, as these characters are also in- creasing in complexity. Lately, much attention has been given to the loss of buffering capacity against en- vironmental stress as the genome tends toward homozygosity (Lerner, 1954). Since directional selection reduces the amount of individual heterozygosity, the loss of buf- fering would result in a greater magnitude of individual deviation from the mean, in- creasing the phenotypic variation of that population. This process may be the cardi- nal factor involved in an explanation of the phenomenon of an increase in variation ac- companying directional selection. However, there is some discouraging evidence against an explanation of this nature. (1) The in- crease of phenotypic variation becomes evi- dent early in artificial selection (Robertson, 1955) before an appreciable amount of genetic variance could have been lost by selection. (2) A character in which selec- tion has considerably altered the mean can often be returned with little difficulty to the original mean by reversed selection. This reversal could not take place if the population had reached a relatively homozy- gous level for that particular character. (3) A correlate of the latter is that often a substantial heritable component is still present after the mean has been considera- bly altered by selection (Lerner, 1958). (4) Bader (1962) showed that, in tooth form, inbred mice exhibit slightly less phenotypic variation than wild popula- tions; and the outcrossed heterozygote is less variable than either. (5) If the tooth variation discussed here in Microtus is non- genetic, it is difficult to explain the phylo- genetic increase in tooth complexity, since the most important cause of evolutionary change is selection acting upon heritable variation. From some preliminary crosses of microtines (Steven, 1953; Zimmermann, 1952), it does seem that these variations are heritable. In at least one species of Microtus [M. arvalis) there is also a geo- graphic cline in the frequency of tooth complexity. The variations were classed into two discrete types (simple or com- plex) ; the frequency of ''complex" ranges from five per cent to 95 per cent in the cline from northern to southern Europe (Zimmermann, 1935). The accumulated evidence from breeding experiments suggests, contrary to the "wild type" or normality concept, that there is considerable heterozygosity underlying the relatively coherent facade of the phenotype. The variation expressed in the phenotype is only a fraction of the total possible varia- tion present (Mather, 1956). There is a diversity of opinion as to the mechanisms involved in the maintenance of this large amount of potential variability. The posi- tion that the balanced additive factors maintain the stored variabihty has much evidence in its favor in terms of its general applicabiUty to evolution at the intrapopu- lational level. Stated in more detail, this position asserts that there exist balanced systems of linked heterozygous polygenes structurally associated and maintained by selection and perhaps also by decreased recombination. Delayed responses to selection are best accounted for on the basis of genetic link- 482 228 R. D. GUTHRIE age. A rather common phenomenon in ex- perimental breeding is for a selected strain to reach a plateau of response only to have it resume progress after a period of relaxa- tion of the selection pressures. The most plausible explanation of this phenomenon is linkage disassociation; the various ele- ments are unable to segregate out im- mediately because of linkage restrictions (Mather, 1949). The ineffectiveness of ex- periments to reduce the variation by selec- tion for intermediates (Lerner's type II se- lection, Lerner, 1958; Falconer, 1957), and the ineffectiveness of selection for the ex- tremes to alter the variation (type III se- lection. Falconer and Robertson, 1956) both suggest that the additive genetic ma- terial resides in balanced linkage groups. Structural change, which often inhibits crossing over, may establish an isolation of segments of the chromosome where crossing over is likely to occur only with configura- tions of that same type; however, the gen- eral importance of this mechanism is still not clear. As well as promoting these de- vices that inhibit recombination, selection can operate directly to maintain these blocks intact (Lerner, 1958) and this is probably the most important mechanism. Carson (1959) reports that most natural inversions are heterotic when removed from nature to the laboratory culture, and that strains derived from a single pair of wild flies retain with extreme tenacity most of their initial inversion variability. The advantages of a system of balanced linkage groups are multiple. The popula- tion can maintain a high degree of hetero- zygosity in many individuals without the rigorous selection required if these elements were segregating at random. The close linkage association also serves as a buffer against random fluctuations away from the optimum. And perhaps most important, it holds genetic material in reserve, thereby maintaining an evolutionary plasticity. There is evidence that integrated chro- mosome segments are important in the as- sociation or correlation of continuously dis- tributed characters, and that they behave in a manner similar to single independent genes acting pleiotropically. To resolve or disassociate the correlation of two charac- ters by selection would produce strong evi- dence for linkage. Such disassociation has been accomplished (Mather and Harrison, 1949; Mather, 1956). Correlation due to pleiotropy is, of course, more resistent to evolutionary change than the more labile system of linkage groups. Linkage groups can originate or be disposed of by the selec- tion for various recombination and struc- tural patterns. It would be a slow process for the population to await a new mutation at one locus which acted upon the desired characters in exactly the right magnitude. Selection can maintain a frequency of bal- anced genetic material within each chromo- somal block or "internally" at levels that insure a considerable proportion of "rela- tionally" balanced, or heterozygous, indi- viduals in the population. As long as this block remains intact it will carry reserves of variability which may be released and made available for segregation by crossing over. With selection against the cross- overs, this residual genetic variability can be maintained (Lerner, 1958). In order to maintain the internally balanced linked groups a selection intensity would be re- quired equivalent to the frequency of cross- overs which deviate from the balanced con- figuration (Falconer, 1960). The increase in variation of evolving characters may be further enhanced when an interbreeding population experiences the stress of two selective optima. This condi- tion would occur in most evolutionary changes when the group is partially exploit- ing two adaptive zones. Thus, a character in transition may be expected to experience some reduction in stabilizing selection along its axis of change. The high variation usually associated with vestigiality can also be accounted for in the context of this theory. A vestigial character is in essence an evolving charac- ter, as reduction plays a great part in evo- lutionary change. According to the ex- planation given for the greater amount of 483 VARIABILITY L\ MICROTUS MOLARS 229 variation in evolving characters, the stored variability is maintained in a linked system by stabilizing selection. When this balance is altered by directional selection, the vari- ability is released. Due to its decreasing functional role, the variation of a vestigial character would also be compounded by a decrease in stabilizing selection. Carson (1955), in his discussion of the genetic composition of marginal popula- tions, surmised that, since the marginal populations contain fewer inversions than central populations, the more stringent selection on the periphery is against the heterotic groups which predominate in the central population. These findings are in agreement with the idea expounded here, that directional selection away from the mean is selection for the breakdown of present linkage configurations. Carson fur- ther reported that when strong artificial selection was applied to both marginal and central population lines, the marginal lines showed the greater initial response. This difference would exist if the genetic ma- terial has been made available for segrega- tion in the marginal populations by the breakdown of the linkage groups. Reeve and Robertson (1953), selecting for wing length in Drosophila, found that the selected strain showed an increase in additive genetic variance of 250 per cent, all other genetic variance remaining about the same. They further suggest that selec- tion for long wing length would be more effective in the selected than in the un- selected stock. Robertson (1955) states: "Selection generally leads to an increase in variance which appears to be largely due to the increased effects of genetic segregation and this constitutes an aid to selection progress." This release of additive genetic variation provides a mechanism whereby directional selection, in continuously distributed poly- genic systems, increases its own resolving power. Selection against the mean and its present balance situation is selecting against the present linkage configurations, which results in a breakdown of these integrated units. The genetic components are then re- leased and made available for novel segre- gants hitherto unavailable. The conse- quence of this is an increase in phenotypic variation, which is heritable in an additive fashion. As the amount of variation is a determining factor of the effectiveness of selection, in conjunction with selection in- tensity and heritability, further selection gains are facilitated. To set up a simple visual model of this theory let us suppose, as is expressed in Fig. 5, that there is a series of loci with alleles acting in an additive fashion either to the left or right of the mean. Loci a, b, c, and d control the size of character X and e, f, g, and h control Y. The contribution of each allele is specified. Further, suppose these are balanced "internally" and "rela- tionally," with an equal frequency of each linkage group. The mean will be consid- ered as zero with the deviations from it in both positive and negative directions. A stabilizing selection for the mean would cull out deviants, the crossovers, from this configuration. The genetic material present is potentially able to produce an individual representative of any point in the figure, but this particular linkage configuration limits the phenotypes to a coherent cluster around the mean. The broken circle repre- sents a variation of two standard deviations from the mean, if each locus were acting individually with an equal frequency of each allele. The linked configuration, how- ever, would produce a population with a lower variation, expressed here at two stan- dard deviations by the solid inner circle. If a new adaptive optimum (A'-) were created with a consequent directional selec- tion of moderate magnitude exerted on the distribution, the linkage groups would be selected against by a selection for the cross- overs in the direction of the new adaptive optimum, resulting in a partial breakdown of the coherent phenotype. A structural association of the loci con- trolling the two characters (Fig. 6) would result in their correlation. The points all fall along the diagonal axis between +2 484 230 R. D. GUTHRIE 6 5 4 C+) 3 2 1 Y 0 1 2 (-) 3 4 • y N y / a, b2 32 b, \ \ \ \ \ c, d2 e C2 d, e a=0.5 b = 1.0 e f ■ g h ' o \ c = 0.5 d= 1.0 e = 0.5 f =1.0 g = 0.5 h = 1.0 ' \ "^ y 1 1 / y • sub1=(— ) sub2=(+) a , b, c, d 2 '1 3 (-) 0 X 3 (+) gl ^2 %2 hi Fig. S. An elementary model of the non-correlated case of two characters X and Y, where low phenotypic variation is maintained by selection for the linkage configuration represented in the upper right. With equal frequencies of each linkage group the variation of the population, at two standard deviations, would be circumscribed by the solid circle. The dashed Hne represents the same loci with no linkage. Selection for A'o would increase the variation as the linkage configuration would be selected against. and -2 units, as shown by the ellipse. If one were to think in terms of the major axis of variation as size, this provides a rela- tively constant individual shape throughout a population in which the individuals are varying in size. As in Fig. 5 it will be noted that an imposed directional selection will produce an increase in variation. Even directional selection parallel with the main axis of size will increase the variation. The greatest increase in variation, however, would be produced by a selection pressure at right angles to the principal axis of vari- ation, toward X2, which would be selection for shape changes. The long-term effect of this type of selection would be twofold: ( 1 ) an increase in phenotypic variation, and (2) a decrease in the correlation of char- acters X and F. Unlike the situation in Fig. 5, if a selection pressure is exerted on only one character (perpendicular to the scale of the other), the second character is also initially affected. However, it is an inherent mechanism of the model that the linkage which provides the correlation of the characters will be selected against when only one character is subjected to direc- tional selection. This system then would further contribute to evolutionary plas- ticity. 485 VARIABILITY IN MICROTUS MOLARS 231 6 5 4 (+)3 2 1 Y 0 1 2 (— ) 3 4 a^ ^2 ej f2 c^ dj g^ hj 32 b, ejf^ C2 d^ g2h^ e f g h b. c .d ■bi 3 C— ) 0 X 3 a = 0 b= 1 c = 0.5 d=l e = 0 f = 1 g=0.5 h=1.0 sub 1 = (— ) sub 2 = (-I-) Fig. 6. Same as Fig. S except that characters A' and Y are now correlated due to the linkage asso- ciation. This new linkage configuration maintains a coherence differentially on the size and shape axes. Selection toward A^o would considerably increase the variation along the shape axis. I do not wish to imply that the theory expressed here accounts for all the various behavior exhibited by residual genetic varia- tion. Rather, I have investigated one aspect, the association of directional selection and the maintenance or increase of the initial phenotypic variation, and have hopefully offered a plausible explanation, which will be further explored soon by breeding experi- ments with Microtus. Summary This is a study of the intrapopulational variability present in the dentition of two species of Microtus, and the more general questions arising from it. The central thesis is that quantitative characters undergoing rapid evolution do not show the decline in phenotypic variation predicted by our pres- ent evolutionary concepts. On the contrary, the variation is maintained and usually in- creased. Of the two species used, the fossil species is thought to be ancestral to the modern meadow vole ; thus the study materials com- prise an evolutionary line with two grades of tooth complexity represented. In the molar crowns of both species, the areas which are changing phylogenetically are those which vary most within the popula- tion. Evidence from other sources in which characters are undergoing directional selec- 486 in R. D. GUTHRIE tion, both evolutionary and artificial, sug- gests that a greater variation in characters undergoing directional selection is a general condition. A theory to account for association be- tween rapidly evolving characters and a relatively higher amount of phenotypic variation is that the coherence of the popu- lation around the mean is due to balanced heterozygous linkage groups and that with the application of directional selection this organization is partially broken down. The genetic variation is then released and made available for recombination. The relatively high variability associated with vestigial characters is also fitted into the context of the theory. The theory suggests that direc- tional selection on continuously distributed characters increases its own effectiveness. Acknowledgments I wish to express my appreciation to Dr. E. C. Olson, University of Chicago, the chairman of my graduate committee, for his encouragement and assistance with the presentation. Thanks are also due to Dr. Vernon Harms and Dr. Brina Kessel, Uni- versity of Alaska, for their helpful criti- cisms of the manuscript. My deepest grati- tude goes to Dr. R. S. Bader, University of Illinois, for the many stimulating discus- sions which were to form the nucleus of my interests in evolutionary mechanisms. I wish to thank also those in charge of collec- tions at the University of Kansas Museum of Natural History, University of Michigan Museum of Zoology, Carnegie Museum, and Chicago Museum of Natural History for the use of the specimens. Literature Cited Bader, R. S. 1955. Variability and evolutionary rate in oreodonts. Evolution, 9: 119-140. . MS. Phenotypic and genotypic variation in odontometric traits of the house mouse. (Manu- script submitted for publication.) Butler, P. M. 1952. Molarization of premolars in the Perissodactyla. Proc. Zool. Soc. London, 121: 819-843. Carson, H. L. 19SS. The genetic characteristics of marginal populations of Drosophila. Cold Spring Harbor Symp. Quant. Biol., 20: 276- 285. . 1959. Genetic conditions which promote or retard the formation of species. Cold Spring Harbor Symp. Quant. Biol., 24: 87-105. Clayton, G. A., and A. Robertson. 1957. An experimental check on quantitative genetical theory H. Long term effects on selection. J. Genetics, 55: 152-180. Falconer, D. S. 1955. Patterns of response in selection experiments. Cold Spring Harbor Symp. Quant. Biol., 20: 178-196. . 1957. Selection for phenotypic intermedi- ates in Drosophila. J. Genetics, 55: 551-561. . 1960. Introduction to quantitative genetics. Ronald Press, New York. Falconer, D. S., and A. Robertson. 1956. Selec- tion for environmental variabihty of body size in mice. Z. indukt. Abstamm.-u. Vererblehre, 87: 385-391. GoiN, O. B. 1943. A study of individual vari- ation in Microtus pennsylvanicus pennsylvani- cus. J. Mamm., 24: 212-223. Guthrie, R. D. ms. A technique for detailed biometrical analysis of two-dimensional crowns. (Manuscript submitted for publication.) Hall, E. R., and K. R. Kelson. 1959. The mammals of North America. Ronald Press, New York. Hibbard, C. W. 1944. Stratigraphy and verte- brate paleontology of Pleistocene deposits of Southwestern Kansas. Geol. Soc. Amer. Bull., 55: 707-754. . 1959. Late Cenozoic microtine rodents from Wyoming and Idaho. Papers Michigan Acad. Sci., Arts, and Letters, 44: 3-40. -. 1963. The origin of the P3 pattern of Syl- vilagus, Caprolagus, and Lepus. J. Mamm., 44: 1-iS. HiNTON, M. A. C. 1926. Monograph of the voles and lemmings (Microtinae) living and extinct. Richard Clay, Suffolk. Hooper, E. T. 1957. Dental patterns in mice of the genus Peromyscus. Univ. Michigan Mus. Zool., Misc. Publ., No. 99. Hooper, E. T., and B. S. Hart. 1962. A synopsis of recent North American microtine rodents. Univ. Michigan Mus. Zool., Misc. Publ. No. 120. King, J. H. 1961. Development and behaviorial evolution in Peromyscus. Pp. 122-147 in W. F. Blair (ed.), Vertebrate Speciation. Univ. of Texas Press, Austin. Komarek, R. V. 1932. Distribution of Microtus chrotorrhinus, with description of a new sub- species. J. Mamm., 13: 155-158. Kurten, B. 1959. Rates of evolution in fossil mammals. Cold Spring Harbor Symp. Quant. Biol., 24: 205-215. Lerner, I. M. 1954. Genetic homeostasis. John Wiley, New York. . 1955. Concluding survey. Cold Spring Harbor Symp. Quant. Biol., 20: 334-340. 487 VARIABILITY IN MICROTUS MOLARS 233 . 1958. The genetic basis of selection. John Wiley, New York. MacArthur, J. W. 1949. Selection for small and large body size in the house mouse. Genetics, 34: 194-209. Mather, K. 1949. Biometrical genetics. Me- thuen, London. . 1956. Polygenetic mutation and variation in populations. Proc. Royal Soc. London, 145: 293-297. Mather, K., and B. J. Harrison. 1949. The manifold effect of selection. Heredity, 3: 1- 52, 131-162. Paulson, R. G. 1961. The mammals of the Cudahy fauna. Papers Michigan Acad. Sci., Arts, and Letters, 46: 127-153. Reeve, E. C. R., and F. W. Robertson. 1953. Studies in quantitative inheritance II. Analysis of a strain of Drosophila melanogaster selected for long wings. J. Genetics, 51: 276-316. Robertson, F. W. 1955. Selection response and the properties of genetic variation. Cold Spring Harbor Symp. Quant. Biol., 20: 166-177. Simpson, G. G. 1937. Supra-specific variation in nature and in classification. Amer. Nat., 71: 236-276. . 1953. The major features of evolution. Columbia Univ. Press, New York. Skinner, M. F., and O. C. Kaisen. 1947. The fossil Bison of Alaska and preliminary revision of the genus. Bull. Amer. Mus. Nat. Hist., 89: 127-256. Snyder, D. P. 1954. Skull variation in the meadow vole (Microtus pennsylvanicus) in Pennsylvania. Ann. Carnegie Mus., 33: 201- 234. Steven, D. M. 1953. Recent evolution in the genus Cleithrionomys. Symp. Soc. Exp. Biol., 7: 310-319. Stirton, R. a. 1940. Phylogeny of the North American Equidae. Univ. CaUfornia Publ., Bull. Dept. Geol. Sci., 25: 165-198. Zeuner, F. E. 1958. Dating the past. Methuen, London. Zimmermann, K. 1935. Zur Rassenanalyse der mitteleuropaischen Feldmause. Arch. Natur- gesch., N. F. 5. . 1952. Die simplex-Zahnform der Feld- maus, Microtus arvalis. Pallas. Verb. Deut. Zool. Ges., Freiburg. 488 Sonderdruck aus Z. f. Saugetierkunde Bd. 32 (1967) H. 3, S. 167—172 Alle Rcdite, audi die der Obersetzung, des Nachdrucks und dcr photomechanischcn Wiedergabe, vorbehalten. VERLAG PAUL PAREY ■ HAMBURG 1 • SPITALERSTRASSE 12 © 1967 Verlag Paul Parey, Hamburg und Berlin Evolutionary adaptations of temperature regulation in mammals^ By L. Jansky Eingang des Ms. 25. 10. 1966 Generally speaking, adaptations may take place either during individual life of animals (acclimations and acclimatizations), or they may be specific to certain species (evolu- tionary adaptations) (Hart 1963b). They may be realized by different mechanisms with different degree of efficiency, however the aim of all adaptations is essentially the same — to reduce the dependence of animals on environmental conditions and thus to increase their ecological emancipation. The study of physiological mechanisms of adaptations is therefore of great ecological importance since it helps us to elucidate physiological processes influencing limits of distribution of different species and having a profound effect on the quality or density of animal populations. The comparison of individual and evolutionary adaptations permits us to trace the evolutionary progres- sive physiological processes and to contribute to the problems of phylogeny. In lowered temperatures mammels tend to lose heat. Theoretically, they can prevent hypothermia either by increasing heat production in the body or by reducing heat loss from the body to the environment. Heat production is realized by shivering; heat conservation may be manifested by reducing the body surface, by improving its insu- lation qualities and by decreasing the body— air temperature gradient according to formula: „ „ •y np H = K ~ (1) (Hart, 1963b) I :2 O o Q o t— Lu 3: Uj CK >- Q O QQ ^OLD ADAPTED WARM ADAPTED nonshivering thermogenesis lower critical temperature Fig. 1. Scheme of heat production of rats adapted to warm (30° C) and cold (5° C) environments. According to Hart & Jansky, 1963 ^ Presented at the 40th meeting of the German Mammalogical Society in Amsterdam. 489 168 L. Jansky INSULATION °C/CAL/mVhR CD o D o -^ — ^ o iC^ en o en r®-r— , 1 1 , , , . 1 , . 1 CD r- !>• 1— -o DEER MOUSE >C • 0 Co X3 1 i • ^^o LEMMING o 2 - o ft^tSe^ RED SQUIRREL rn rn - m^ooS) MUSKRAT HARE • «Oti^qd fssss^'ss^'sssa RED FOX -^ o •«• •• o %o oo WOLVERINE • • •• aoo #0 (K^ss^ir^K^^ u/r;/ /- o • •• • • 00 BLACK BEAR •••• o^^^lt^^ POLAR BEAR o o o - Fig. 2. Seasonal changes in fur insulation in various mammals (Hart, 1956) (H = heat production, K = a constant representing the body surface area, Tr = body temperature, Ta = air temperature, I = insulation qualities of the body surface.) Similary, the adaptations of temperature regulation to cold can be realized either by increasing the capacity of heat production or by mechanisms leading to reduction of heat loss from the body. The adaptation to cold appears as a shift of the lowest temperature limit animals can survive (lower critical temperature). In our earlier work we have shown that the individual adaptations are manifested predominantly by an increased capacity of heat production owing to the development of a new thermogenetic mechanism — called nonshiveringthermogenesis (Hart, Jansky 1963). Physiological background of this phenomenon consists in an acquired sensitivity of muscular tissue to thermogenetic action of noradrenaline liberated from sympathetic nervous endings (Hsieh, Carlson 1957). Nonshivering thermogenesis potentiates heat production from shivering and in rats shifts the lower critical temperature for about 20° C (from -18° C down to - 37° C; Fig. 1). Mechanisms controlling heat loss by changes in body surface area or by changes in body-air temperature gradient are not common in individual adaptations. On the other hand it is well known, that certain species can improve body insulation in winter 490 Evolutionary adaptations of temperature regulation in mammals 169 season. However, this phenomenon becomes functionally justified only in animals of greater size (size of fox and larger; Fig. 2. Hart 1956). The individual adjustments with the aid of nonshivering thermogenesis are encoun- tered both in acclimations under laboratory conditions and in seasonal acclimati- zations induced in the same species under natural conditions. They are undoubtelly very efficient and biologically important. On the other hand, from the ecological point of view, they have also their negative side. The increased heat production results in higher demands for energy restitution in the body, which is attained in cold adapted animals by an increased food consumption. As a result, individuals adjusted this way become more dependent on the quantity and availability of food and they are forced to use more effort to provide it. The reduced dependence of animals on temperature factors is thus substituted by increased dependence on food factors. Contrary to individual adaptations, in evolutionary adaptations mechanisms lea- ding to the reduction of the heat loss are greatly emphasized. Their importance consists in the fact that they save energy for the organism and have lower demands to its restitution in the body. This fact is obviously evolutionary very important — in the processes of phylogeny there occurs natural selection of those individuals that are less impeded by the lack of food, often occuring in nature. Evolutionary adaptations are realized in the first place by an increased insulation of the body cover (fur. Fig. 3). This adjustment, typical for arctic animals, can reduce the heat loss so efficiently, that even considerably reduced ambient temperatures (down to — 50^ C) do not result in an increased heat production in larger animals. (Fig. 4; ScHOLANDER et al. 1950a, b). The same role plays a thick layer of subcutaneous fat which appears in some mammals, such as seal and swine. The insulation qualities of this fat layer can be increased by an active restriction of the blood flow to this area. This results in superficial hypothermia, which also efficiently prevents the heat loss (Irving 1956). Animals endowed with superficial hypothermia have normal thermogenetic abilities. However, compared to the species from tropical regions with little insulation and to arctic species with great surface insulation they show a reduced sensitivity of afferent sensory input INSULATION to temperature stimuli (Fig. 5). A tendency to reduce heat loss by reduction of the body surface area may be considered as another type of evolutio- nary adaptations. This phenomenon occurs in animals living perma- nently in cold climate, which are generally lar- ger and have shorter body appendages than animals from tropical zone(BERG- mann's and Allen's ru- les). Both the validity and the physiological signifi- cance of these rules have 0 10 20 30 40 50 60 70 been recently questionend THICKNESS IN MM by several workers, how- pj^ j Insulation in relation to winter fur thidtness in arctic ever. and tropical mammals (Scholander et all., 1950 b) P COTTON 40 r 30 - 491 170 L. Jansky 400 300 Co d 200 S ^ 100 MAMMALS ARCTIC TROPICAL BASAL METABOLIC RAIE = 100 ■ OBSERVED EXTRAPOLATED -70 ^LOWEST -50 TEMP ON EARTH -30 -10 10 AIR TEMPERATURE IN "C ^ 30 / BODY TEMP Fig. 4. The effect of environmental temperature on metabolism of arctic and tropical mammals (ScHOLANDER et all., 1950 a) 300 -J 5 o 200 5: O Uj SWINE INFANT CARIBOUo\° THICK- FURRED o' ANIMALS, MAN WHITE RAT MUSKRAT The reduction of heat loss by changing the body-air temperature gradient can be realized either by active choice of higher environmental temperature or by consi- derable lowering of body temperature. It is generally recognized that the active choice of the environmental temperature occurs by seasonal migrations and by changes in patterns of daily activity. It was found that different species of voles and shrews transfer the peak of daily activity to war- mer part of the day in a cold weather (Jansky & Hanak 1959). The mechanisms leading to reduction of body-air tempera- ture gradient by lowering of body temperature are especially developed in hibernators. Ac- cording to the latest view hiber- nation is not considered as a lack of temperature regulation rather as a special adaptation of thermogenetic processes. There are two reasons for that: first, hibernators have the same capa- city of heat production as other hemeotherms of similar size (see Jansky, 1965) and second, the entering, the arousal and the deep hibernation are under remarkably precise physiological control (see Lyman, 1963). This indicates a leading role of central nervous system in controlling hibernation, which is adapted to hypothermal con- 100 0 HARBOR ( SEAL ^•. HARPO SEAL " GO 0 10 SKIN 20 TEMP 30 Fig. 3. Heat production as a function of skin tempera- ture under fur of the back for a series of mammals (Hart, 1963 a) 492 nfter hexameihonium 2/.0 nun. Evolutionary adaptations of temperature regulation in mammals 171 ditions and it is functional at all levels of body temperature. This adaption has certainly its metabolic background, however only little is known about this phenomenon so far. The control of entering into hibernation is realized by the active inhibition of shivering heat production by signals from subcortical centres of the brain. Simultane- ously with the decrease in shivering an active inhibition of the activity of the sym- pathetic nervous system also takes place, which is manifested by the reduction of heart rate and by vasodilatation. These changes facilitate the lowering of body temperature of animals which is realized successively in the form of "undulating" cooling so the organism can slowly prepare to hypothermia (Fig. 6). Nervous control of hibernation persists in deep hypo- thermia as evident from ''c the sensitivity to thermal and other stimuli. The arousal from hibernation is equally an active pro- cess, very efficiently con- trolled, so that organism can produce a great amount of heat in mini- "^i^-i ^ 1 * '"''"^' mum of time. The coordi- nation of thermogenetic processes depends also on the activity of nervous centres. Characteristic or fig, (,_ Changes in body temperature of the bat Myotis myotis awakening is the prepon- during entering hibernation (Jansky, Hajek, 1961) derence of sympathetic nervous system, leading to vasoconstriction and to an increase in heart rate. The main source of heat in awakening is again constituted by shivering. However, nonshivering heat production was also found during arousal and also the rapidly beating heart, working against a high pressure, may contribute a certain amount of heat. Summary On the basis of all mentioned data we conclude that the adaptations of temperature regula- tion to cold may be realized either by an increased ability to produce heat or by reducing the heat loss. While the individual adaptations are manifested chiefly metabolically as evident from an Increased capacity of heat production, the Inherited adaptations are realized mainly by mechanisms leading to the heat loss reduction (e. g. Increased Insulation by fur or by superficial hypothermia, reduction of body surface area, active choice of environmental tem- perature and lowering the body temperature). The control of the mentioned adjustments con- sists In the changes In function of the central and sympathetic nervous systems Inducing changes In intensity of the energy metabolism (individual adaptations), changes in the plasti- city of vasomotor medianlsms and In heat production of hibernators during entering into and awakening from hibernation (evolutionary adaptations). Morphologically based adjust- ments (Improvement of insulation by fur) appearing in both evolutionary and individual adaptations forms the connecting link between both types of adaptations. Zusammenfassung Aus alien erwahnten Daten folgern wir, dafi die Adaptatlonen der Temperaturregullerung bel Kalte entweder durch die erhohte Warmeproduktion oder durch die Verringerung des Warmeverlustes errelcht werden. Wiihrend die Indlviduellen Adaptatlonen hauptsachlich meta- bollscher Art sind, was durch die erhohte Kapazltat der Warmeproduktion In Erschelnung tritt, findet man erbllche Adaptatlonen zumelst in Form von Mechanlsmen, die eine Verringerung des Warmeverlustes bewlrken (z. B. erhohte Isollerung durch das Fell oder durch oberflachllche 493 172 L. Jansky Hypothermic, Verringerung der Korperoberflache, aktive Wahl der Umgebungstemperatur und Abslnken der Korpertemperatur). Die Steuerung der erwahnten Anpassungen beruht auf Veranderungen in der Funktion des zentralen und des sympathischen Nervensystems, welche Veriinderungen in der Intensitat des Energiestoffwechsels (individuelle Adaptationen) hervor- rufen, weiterhin Veranderungen in der Plastizitat der vasomotorischen Mechanismen und in der Warmeproduktion von Winterschlafern beim Einritt in den Winterschlaf und beim Erwachen (evolutive Adaptationen). Morphologische Adaptationen (Verbesserung der Isolierung durch das Fell), die sowohl als evolutive und auch als individuelle Adaptationen vorkommen, stellen die Verbindung zwischen beiden Typen der Adaptation her. Literature Hart, J. R. (1956): Seasonal changes in insulation of the fur. Can. J. Zool. 34: 53 — 57. — (1963a): Surface cooling versus metabolic response to cold. Fed. Proc. 22: 940 — 943. — (1963b): Physiological responses to cold in nonhibernating homeotherms. Temperature — Its Measurements and Control in Science and Industry 3: 373 — 406. Hart, J. S., and Jansky, L. (1963): Thermogenesis due to exercise and cold in warm and cold acclimated rats. Can. J. Biochem. Physiol. 41: 629 — 634. HsiEH, A. C. L., and Carlson, L. D. (1957): Role of adrenaline and noradrenaline in chemical regulation of heat production. Amer. J. Physiol. 190: 243 — 246. Irving, L. (1956): Physiological insulation of swine as bare-skinned mammals. J. Appl. Phy- siol. 9: 414— 420. Jansky, L. (1965): Adaptability of heat production mechanisms in homeotherms. Acta Univ. Carol.-Biol. 1—91. Jansky, L., and Hajek, I. (1961): Thermogenesis of the bat Myotis myotis Borkh. Physiol. Bohemoslov. 10: 283—289. Jansky, L., and Hanak, V. (1959): Studien iiber Kleinsaugerpopulationen in Siidbohmen. II. Aktivitat der Spitzmause unter natUrlichen Bedingungen. Saugetierkundliche MItteilungen 8: 55—63. Lyman, C. P. (1963): Homeostasis in Hibernation. Temperature — Its Measurement and Control in Science and Industry 3: 453 — 457. Scholander, p. F., Hock, R., Walters, V., Johnson, P., and Ikving, L. (1950a): Heat regu- lation in some arctic and tropical mammals and birds. Biol. Bull. 99: 237 — 271. Scholander, P. P., Walters, V., Hock, R., and Irving, L. (1950b): Body insulation of some arctic and tropical mammals and birds. Biol. Bull. 99: 225 — 236. Author's address: L. Jansky, Ph. D., Department of Comparative Physiology, Charles Uni- versity, Prague 2, VInicna 7, CSSR 494 SECTION 6— ZOOGEOGRAPHY AND FAUNAE STUDIES Studies of faunas, both of local areas and of broad regions, have contributed substantially to the literature in mammalogy. From the earliest contributions to the present, papers and books dealing with faunistics have included much information on systematics, ecology, distribution, ethology, and reproduction, among other topics. The sobriquet "natural historian" implied an interest in all these fields and more. Darlington's (1957) book Zoogeography and Udvardy's (1969) Dynamic Zoogeography are the best single sources of general information on the sub- ject; Hesse et al. ( 1937 ) is a substantial and still useful earlier reference. Insular biogeography was aptly dealt with by Carlquist ( 1965 ) and in a more mathematically-oriented way by MacArthur and Wilson (1967). Matthew's (1939) Climate and Evolution and Dice's (1952) Natural Communities are but two of the other general treatises that should be called to the attention of the beginning student. Among the major f aunal catalogues are Allen ( 1939 ) for Africa, Ellerman and Morrison-Scott (1951) for the Palearctic Region, Miller and Kellogg (1955) and Hall and Kelson (1959) for North America, Cabrera (1958, 1961) for South America, and Troughton (1965) for Australia. At the regional or provincial level, Kuroda's (1940) treatment of Japanese mammals, Laurie and Hill (1954) on New Guinea and the Celebes, and Peterson's (1966) The Mam- mals of Eastern Canada are good examples as are many of the state lists pub- lished for North America (e.g., DeKay, 1842; Miller, 1899; Hall, 1946; Jackson, 1961; Baker and Greer, 1962; Jones, 1964), of which Hall's Mammals of Nevada stands out in completeness of coverage from most points of view. In terms of smaller geographic areas, Harper ( 1927 ) on the Okefinokee Swamp, Johnson et al. (1948) on the Providence Mountains of California, Anderson (1961) on the Mesa Verde of Colorado, and Foster's ( 1965) study of the Queen Charlotte Islands illustrate that substantial information can be gleaned from the study of a geographically restricted fauna. These papers as well as several reproduced here certainly indicate that the serious student of faunistics must be as broadly trained as any student in the discipline of mammalogy. Because of the sustained interest in faunal studies over the years, it was inevitable that certain "rules," "laws," and "systems" — directed at overall explanations for natural phenomena associated with distribution and variation — would emerge. These have been of two basic sorts, various "ecological rules" such as those proposed by Allen, Bergmann, and Gloger, and the biogeographic systems proposed on a world-wide scale by Wallace and others and ap- plied more specifically to North America by Merriam (Life-zones), Shelford ( Biomes ) , and Dice ( Biotic Provinces ) . Space does not permit the reproduc- tion of the lengthy papers dealing with these subjects, but a short contribution by Dice, which is included, serves to introduce the reader to this aspect of mammalian zoogeography. Other selections for this section deal with zoogeographic problems related to Pleistocene, sub-fossil, and Recent faunas (Guilday, Koopman and Martin, and Findley and Anderson). One paper concerns a local fauna (Jones and 495 Lawlor ) but also includes information relevant to other sections of this anthol- ogy. The essay by Davies covers an entire order (or suborder according to many other authors ) of mammals, whereas the short paper by Davis deals with the relationship of soil types and altitude to the distribution of a single species in a restricted area. The final paper is a statistical treatment by Hagmeier of distributional patterns on a continental basis. This analysis is based on data compiled in one of the faunal catalogues (Hall and Kelson) cited above. 496 THE CANADIAN BIOTIC PROVINCE WITH SPECIAL REFERENCE TO THE MAMMALS Lee R. Dice University of Michigan In eastern North America many zoogeographers recognize six life zones, all assumed to be transcontinental in extent and named, respectively, Arctic (or Arctic-alpine). Hudsonian, Canadian, Transition, Upper Austral, and Lower Austral (Merriam, '98: 18-53). The AUeghanian, Carolinian, and Austroriparian faunas are the eastern portions, respectively, of the Transi- tion, Upper Austral, and Lower Austral life zones. In addition, some zoogeographers recognize a Tropical region which covers the southern part of Florida. Ecologists, on the contrary, divide eastern North America somewhat differently. Shelford, Jones, and Dice ('26: 60-73) recognized here the Arctic Tundra, Northern Coniferous Forest, Mixed Coniferous and De- ciduous Forest, Deciduous Forest, and Southeastern Coniferous Forest biotic areas. Weaver and Clements ('29, frontis) divide the eastern part of the continent among the Tundra, Boreal Forest, Lake Forest, Deciduous Forest, and Tropical climaxes. There is rather general agreement among biogeographers on the im- portance of the Tundra (Arctic) and Boreal Forest (Hudsonian) divisions. Also most students of distribution would accept the Southeastern Coniferous Forest (Austroriparian) as at least a minor unit. On the other hand, there is little agreement on the biogeographical division of the remaining middle portion of eastern North America. In order to examine the several biotic provinces of northeastern North America I drove by automobile in the summer of 1936 through considerable parts of eastern Canada and of the northeastern United States. The observa- tions made on this trip and a subsequent study of the available descriptions of the vegetation and faunas of the area have convinced me that the so-called Canadian and AUeghanian faunas are only different aspects of the same eco- logical complex. The name Canadian is more descriptive of this complex than is the name AUeghanian, and, therefore, the term Canadian biotic prov- ince is here adopted for that part of northeastern North America in which 503 497 LEE R. DICE Ecology, Vol. 19, No. 4 Fig. 1. Map of eastern North America showing the distribution of the several biotic provinces. 498 October, 1938 CANADIAN BIOTIC PROVINCE 505 hardwoods form the climax and conifers of several kinds form several types of subclimaxes. A biotic province is, according to my definition, a major biogeographic division of a continent, characterized by the biotic communi- ties which compose it. The Canadian biotic province as here recognized (fig. 1) covers much of southern Quebec, including the Gaspe Peninsula, and all of New Bruns- wick, Nova Scotia, and the adjacent islands. It extends southward to include much of New England, most of New York state, and the mountainous parts of Pennsylvania. It covers Michigan and Wisconsin, except their southern parts; the northeastern half of Minnesota; and all of southern Ontario, except along Lake Erie. The province is in my interpretation not trans- continental, although there are some similarities between its biota and that of the western mountains. An excellent description of the vegetation of the Canadian biotic province has been given by Nichols ('35: 403-422) under the designation Hemlock — white pine — northern hardwood region. Nichols, however, includes the southern Appalachian Mountains in his region, while I consider it better to include these mountains in the Carolinian biotic province, by which they are surrounded. The western end of the northern boundary of the Canadian province should in my opinion be placed a little north of the position given by Nichols ('35, fig. 4), so as to include all the important stands of sugar maple, white pine, and Norway (red) pine (see Howe and Dymond, '26: 288-291). An unpublished study of the plant ecology of Isle Royale by Clair A. Brown shows that there are considerable stands of sugar maple forest on this island, which lies in the northern part of Lake Superior. The northernmost im- portant stands of sugar maple and white pine observed by me along the road between North Bay and Cobalt, Ontario, were about 50 miles north of North Bay, on the rocky ridges of the Timagami Provincial Forest. A second-growth forest of white spruce and pine was studied at James Lake, 11 miles south of Latchford, or about 66 miles by road north of North Bay. This forest was growing on moderately high rolling rocky hills, and the soil, while often thin, contained much humus. White birch was the most common tree, and balsam fir, black ash, and aspen were numerous. There were a few small white and Norway pines, and a few of a species of poplar with long leaves. A maple (Acer spicatum) was the most common shrub. Signs of snowshoe hare were numerous, and red squirrels and chipmunks (Taniias) were seen. Ten red-backed voles {Clethrionomys gafferi gafferi and 1 deer-mouse {Peromyscus maniculatus gracilis) were trapped on a short trap-line. This situation obviously is transitional between the Hud- sonian and Canadian provinces. The transition between the Canadian and Hudsonian provinces is in some places rather abrupt. On the northern margins of their ranges the sugar maple, yellow birch, and white pine occur only in the most favorable habitats 499 506 LEE R. DICE Ecology, Vol. 19, No. 4 and a slight change in topography or soil may make it impossible for these Canadian province species to exist. As an example of a fairly abrupt transition I submit herewith two logs of the vegetation observed in the summer of 1936 along the highway in two parts of northern Ontario. The distance each type of forest was traversed by the highway was determined by readings of the automobile mileage meter at each change in vegetation. The figures given cannot be assumed to be a dependable measure of the proportionate occurrence of the several vegetation types in the two regions, because the highway undoubtedly avoids the steeper rocky slopes and at least the more widespread of the bogs. Nevertheless, the figures do give a general indication of the relative abundance of the several ecologic types in the two situations. The stretch of highway logged in the Canadian province extends from 10 miles north of North Bay, Ontario, to a point 10 miles further north. The road here crosses a number of rocky ridges, and the general exposure is to the south. In the 10 miles of the log 72 per cent is dominated by forests of sugar maple, yellow birch, and white pine. Bogs, in which black spruce was most conspicuous, cover 22 per cent of the distance. In these bogs there occur also some balsam fir and some tamarack. Burns and badly mixed vegetation cov^ 6 per cent of the recorded distance. Much of the maple forest along t1ie highway has been logged, and part of both the upland forest and of the bogs has been burned. For contrast, there is available the log of the dominant vegetation along the highway from Cochrane south nearly to Swastika, a distance of 94.7 miles, all in the Hudsonian province. The black spruce type of vegetation formed the original cover for 49 per cent of this distance. Much of this black spruce occurs in lowland bogs called muskegs, and these muskegs are especially extensive on the nearly flat plain near Cochrane. A few balsam firs occur with the black spruce at least as far north as Cochrane, and in the better drained situations the white spruce also occurs. Sphagnum forms a heavy mat under the spruces and Labrador tea is a characteristic low shrub. Near Cochrane the black spruce type of bog vegetation is not restricted to low and undrained situations, but extends also over the lower hills. Another important type of vegetation between Cochrane and Swastika is the jack pine forest, which covers 11 per cent of the log. This forest type is restricted to sandy areas. Associated with the jack pine are fre- quently the aspen, the black and the white spruces, and less commonly the white birch. Blueberries are a characteristic low shrub. Fires have been frequent in this region and 26 per cent of the distance between the two cities is dominated by aspen. Mixed types of vegetation of various sorts, includ- ing arbor vitae, black spruce, alder, willow, aspen, balsam fir, white spruce, jack pine, and white birch made up 11 per cent of the vegetation. Clearings in which the original type of vegetation could not be determined from the road cover 3 per cent of the distance. 500 October, 1938 CANADIAN BIOTIC PROVINCE 507 Although the sections of highway described above in the Canadian and Hudsonian provinces, respectively, are only about 150 miles apart at their nearest approach, there are very obvious differences in their vegetation. In the Hudsonian province near Cochrane the sugar maple, yellow birch, and white pine type of forest characteristic of the Canadian province is absent, while the spruce bog is the most extensive community. Climatic differences related to latitude are probably in large part responsible for these dissimi- larities in vegetation. However, the soil characters of the two regions are very different and these variations in soil greatly affect the vegetation. The rocky ridges 10 to 20 miles north of North Bay are probably at nearly the northern limit of the forest of sugar maple, yellow birch, and white pine, and it is doubtful if this type of forest could exist at Cochrane even on rocky slopes. On the other hand, the general lack of jack pine along the highway near North Bay is obviously due to the absence of sandy soil in this area, for jack pine forest is an important vegetation type on sandy soils much further south. North of Quebec sugar maples occur on the southern slopes of the mountains of the Laurentides National Park, but, so far as could be seen on a hasty drive through this park, all the higher parts of the mountains are dominated by spruce forest (see also Fuller and Marie- Victorin, '26: 295- 296). On the northern slopes of these mountains, toward Lake St. John, sugar maples reappear, and numbers of these trees were noted near Abbey- ville. The position of the boundary between the Hudsonian and Canadian in this region must therefore be drawn somewhat arbitrarily. The tip of the Gaspe Peninsula has been excluded by Nichols from the Hemlock region (Canadian province). It is true that the high interior parts of the peninsula would be expected to be dominated by spruce forests. How- ever, along the northeastern Gaspe coast, on crossing a ridge several hundred feet high between the villages of Chloridorme and Gaspe, I noted the occur- rence of a few white pines and sugar maples, along with fir. spruce, arbor- vitae. aspen, white birch, and mountain ash. Several kinds of mammals reach their northern limits in the Gaspe peninsula, and probably occur over most of the peninsula. It therefore is much simpler for the mammalogist if all the peninsula is included in the same province. I have for this reason drawn the northern boundary of the Canadian province down the middle of the Bay of St. Lawrence. The southern boundary of the Canadian province in the Appalachian Mountains is difficult to place, because isolated areas of northern type forests and fauna recur on the Appalachian Mountains south as far as North Caro- lina and Tennessee. I have here arbitrarily drawn the southern boundary of the Canadian province at the southern border of the state of Pennsylvania. It is, however, quite possible that a more natural division between the north- ern and southern Appalachian Mountains may occur somewhere in West Virginia. 501 508 LEE R. DICE Ecology, Vol. 19, No. 4 The climax vegetation of the Canadian province is a hardwood forest, in which the sugar maple (Acer sacchanim) and yellow birch (Betula lutea) are the most characteristic trees. The eastern white pine (Pinus strobus) and the eastern hemlock (Tsuga canadensis) occur frequently. The beech {Fagus grandifolia) is also characteristic, except that it does not occur in the extreme western part of the province. Numerous other trees and many shrubs are characteristic. The Canadian province therefore corresponds closely to the "northern hardwood forest" of some foresters (Frothingham, '15: 1-7). The pines {Pinus strobus, P. resinosa, and P. divaricata) , often growing in a pure stand of one species, are in many places an important subclimax stage in the Canadian province. The pine forests nearly always occupy sandy or gravelly soil. On some very poor sandy soils the pine forest may persist indefinitely as an edaphic climax, for in such situations insufficient humus may accumulate ever to make the site suitable for a hardwood forest. On good soil the pines are usually quickly succeeded and shaded out by hard- woods, although the white pine often persists in the climax forest. Some of the white pine trees in the climax forest may be relicts of an earlier pine stage, persisting by reason of their height, but some probably spring up from time to time in openings in the forest produced by the fall of large trees. In poorly drained situations in the Canadian province, many areas are covered by bogs and swamps in which there occur varying mixtures of balsam fir {Abies balsamea), black spruce {Picea mariana), white spruce {Picea canadensis) , northern white cedar {Thuja occidentalis) , and tamarack {Larix laricina), with occasional hardwoods such as black ash and red maple. As pointed out by Nichols ('35: 411-412) these bogs and swamps "closely resemble the climatic climax of the northern conifer forest region," which here is called the Hudsonian province. Following fires in the forest of the Canadian province there often springs up a first forest of aspen {Populus tremuloides and P. grandidentata) or of white birch {Be tula papyrifera). The aspens and birches are usually fol- lowed by pines. In northern Michigan, if hardwood forest on clay soil is destroyed by fire, there may be no complete succession through aspens and pines, but at least in some situations the hardwood forest regenerates directly after a brushy stage, and the aspen and pines stages are omitted. Sandy soils and also poorly drained situations occur over nearly the whole area of the Canadian province, alternating irregularly with good soils and with well drained situations. In some districts hundreds or thousands of square miles may be covered mostly with sandy soil, with accompanying pine forests. In other places equal areas of poorly drained soil may support forests of spruces and firs. Nevertheless, on the better soils throughout this whole area the northern hardwood type of forest natively formed the characteristic vegetation, and it certainly is the ultimate climax for the climate. The firs and pines which have been assumed to distinguish the 502 October, 1938 CANADIAN BIOTIC PROVINCE 509 Alleghanian biota from the Canadian biota are actually successional stages which mostly are characteristic only of certain soil types. These soil types and their accompanying coniferous vegetation recur throughout most of the area which has been called Alleghanian as well as over the area previously assigned to the Canadian. Spruce forests occur on the upper parts of mountains and in lowland bogs over most of the Canadian province and also even in parts of the Carolinian province. The fauna and flora of these isolated spruce habitats resemble in part the fauna and flora of the Hudsonian province. Neverthe- less, it is futile to attempt to mark on the map every local spruce habitat. In my opinion it is better to treat these isolated habitats as minor communi- ties of the province in which they occur. Actually these isolated spruce communities are never exactly like the major communities of the Hudsonian province, which they resemble superficially, for most or all of the larger Hudsonian mammals are missing. No attempt will here be made to consider all of the kinds of animals which are characteristic of the Canadian biotic province. In this report I shall confine my attention entirely to the mammals, the group of animals with which I am most familiar. However, it is known that some other classes of land vertebrates and some kinds of invertebrates also are restricted in distribution by some of the same ecological barriers as are the mammals. In making up the lists of mammals I have followed in general the nomen- clature of Miller ('24). For data on the distribution of the several species I have used such revisions as are available, chiefly in the valuable North American Fauna series prepared by the United States Biological Survey. Unfortunately, many species and genera have not been recently revised and the distribution of some species has never been adequately mapped. For records of the distribution of many of the ungulates and larger carnivores I have depended chiefly upon the maps presented by Seton ('29). The following species of mammals range over most or all of the Canadian province as I have defined it: Condylura cristata (Star-nosed mole), Sorex cinereiis (Masked shrew), Sorcx palustris (Water shrew), Microsorex Jioyi (Pigmy shrew), Blarina hrevicauda (Short-tailed shrew), Myotis lucifngus (Little brown bat), Myotis kecnii (Little brown bat), Lasionycteris nncti- vagans (Silver-haired bat), Eptesicus fuscus (Large brown bat), Lasiurus borealis (Red bat), Ursiis americanus (Black bear), Procyon lotor (Rac- coon), Maries americana (Marten), Maries pennanii (Fisher), Musiela vison (Mink), Musiela cicognanii (Short-tailed weasel), Musiela frenata (Long- tailed weasel), Gulo luscus (Wolverine), Luira canadensis (Otter), Mephiiis mephitis (Striped skunk), Vulpes fulva (Red fox), Canis lycaon (Timber wolf), Lynx canadensis (Canada lynx), Marniota monax (Woodchuck), Taniias striaius (Chipmunk), Sciurus hudsonicus (Chickaree), Glaucomys sabrinus (Flying-squirrel), Castor canadensis (Beaver), Peromyscus nianicu- laius (Deer-mouse), Synaptomys cooperi (Bog-lemming), Clethrionomys 503 510 LEE R. DICE Ecology, Vol. 19, No. 4 gapperi (Red-backed vole), Microtus pennsylvanicus (Meadow-vole), On- datra sibctJiica (Muskrat), Zapus Jindsonius (Meadow jumping-mouse), Napaeozapus insignis (Woodland jumping-mouse), Ercthizon dorsatum (Porcupine), Lcpus americanus (Snowshoe hare), OdocoUcus virginianus (White-tailed deer), Alces amcricana (Moose). Of these species the masked-shrew, silver-haired bat, black-bear, fisher, mink, otter, red-fox. wolf, chickaree, beaver, deer-mouse, meadow-vole, muskrat, and meadow jumping-mouse (Zapus) ranged natively over most or all of the Hudsonian of eastern Canada as well as over the Canadian, and all of them also extended into biotic provinces south of the Canadian. These then are wide-ranging forms which extend beyond the Canadian province on both sides. A few species which are mainly of more northern distribution ranged south natively into the northern part of the Canadian. These are the Arctic shrew {Sorcx arcticus), wolverine, northern lemming-vole (Syuaptomys horcalis), heather-vole (Pliciiacoinys uiigava), and caribou (Rangifer cari- bou). Most of these forms are rare in the Canadian and for the larger forms the original southward distribution in eastern North America is not well known. However, probably none of them originally extended as far south as the southern boundary of the Canadian province. Species which ranged natively over most of the Hudsonian of eastern Canada and which in eastern North America reached their southern limits near the southern border of the Canadian as here mapped are the marten, short-tailed weasel, Canada lynx, northern flying-squirrel (Glaucomys sa- brinus), red-backed vole, porcupine, and snowshoe hare. The moose seems to have been the only form which was limited in its southward distribution in eastern North America by the southern border of the Canadian province as here mapped, and which also ranged into the southern part only of the Hudsonian province. The pigmy-shrew (Microsorex) is rare and its distribution not well known. Although it seems to occur mostly in the Hudsonian and Canadian provinces it also occurs in other areas to the southward and westward. The hoary bat (Lasiunis ciucrca) occurs in the Canadian province, and is at least in part a migrant, breeding farther north as well as farther south. Species which range from districts south of the Canadian province north across the whole Canadian and into the southern part of the Hudsonian province are the little brown bat (Myotis lucifugus), striped skunk, and woodchuck. Species of more southern distribution which reach their northern limits at or close to the northern boundary of the Canadian province are the short- tailed shrew, little brown bat (Myotis keenii), large brown bat, red-bat, raccoon, long-tailed weasel, chipmunk (Tatnias), Lemming-vole (Synaptomys cooperi), and white-tailed deer. The deer is now extending its range further northward. 504 October, 1938 CANADIAN BIOTIC PROVINCE 511 Species wliich mainly arc of more southern distribution but which natively ranged into some of the southern parts of the Canadian province are the mole (Scalopus aquaticiis), pipistrelle {Pipistrellus suhflavus), gray-fox (Urocyon cinercoargoitcus), cougar (Fclis couguar), bob-cat (Lynx rufus), gray- squirrel (Sciurus caroli)ieusis), fox-squirrel (Sciitrus nigcr), flying-squirrel (Glaucomys volans), wood-mouse (Peromyscus Icucopus), American elk {Cervus canadensis), and bison (Bison bison). In southern Michigan the mole, fox-squirrel, and wood-mouse were probably originally mostly absent from the Canadian province, but with the clearing of the forests they have extended their ranges some distance into the province. The bob-cat has in Michigan in historic time greatly extended its range northward. The star-nosed mole and woodland jumping-mouse (Napaeosapus) cover most of the area of the Canadian province and both extend south in suitable habitats along the Appalachian Mountains. In addition both also extend slightly beyond the limits of the province in other directions. Several species of mammals occupy part only of the Canadian province and also range south into the Carolinian province along the Appalachian Mountains. These are the hairy-tailed mole (Parascaiops breweri), smoky- shrew (Sorcx fuiiieus), another shrew (Sorcx dispar), woodrat (Neotoma pennsylvanica) , rock-vole (Microtus chrotorrhinus) , and New England cot- tontail (Sylvilagus transitionalis) . Several species which are mostly more western in distribution range into the western parts of the Canadian province. These are the least-weasel (Mustcla rixosa), badger (Taxidea taxus), coyote (Canis lafrans), western chipmunk (Eutamias minimus), and cottontail (Sylvilagus floridanus) . Of these the coyote has within recent time considerably extended its range east- ward. The cottontail seems not to have originally occurred in the Canadian province, but in recent years it has spread northward in Michigan and east- ward in Wisconsin well into the province. The Gaspe shrew (Sorex gaspensis) seems to be the only species of mammal restricted to the Canadian province, but this species occurs only in the Gaspe region, so that it is not characteristic of all the province. Certain subspecies of mammals are largely limited to the Canadian prov- ince, but no one subspecies which is restricted to the Canadian completely covers the whole province. In Sorcx palustris the two subspecies albibarbis and gloveralleni are largely confined to the northeastern part of the Canadian province and the subspecies hydrobadistes largely to the western part of the province. In Peromyscus maniculatus the subspecies abietorum covers largely the eastern part of the province, gracilis the western part, while the subspecies nubiterrae extends from the south somewhat into the province in Pennsyl- vania and possibly in southern New York. In Synaptomys cooperi the sub- species cooperi covers practically the Canadian area, but it extends in places slightly south of the boundaries of the province. It is shown by the above discussion that several of the mammals which 505 512 LEE R. DICE Ecology, Vol. 19, No. 4 occur in the Canadian province are wide-ranging species which occupy biotic provinces both to the north and to the south of the Canadian. A number of the mammals of the Canadian biotic province are species which are char- acteristic of the eastern deciduous forests, and some of these species reach their northern Hmits at the northern edge of the province where the deciduous forests terminate. A few others extend still further north into the Hud- sonian province or completely across it. Several species of mammals found in the Canadian province are characteristic of the northern spruce forests (Hudsonian province) and a few of these species reach their southern limits at or near the southern border of the Canadian province. No species of mammal which ranges over all or over most of the Canadian province seems to be restricted to that province. However, a few subspecies or groups of related subspecies are restricted fairly closely to the limits of the province. In its mammalian fauna the Canadian biotic province is characterized by the intermingling of species which are abundant in the eastern deciduous forest with those which are abundant in the northern spruce forests. This is what would be expected from the fact that the important mammalian habitats of the Canadian province are predominantly of two types, hardwood forest and conifer swamp and bog. It is true that in the Canadian province there is a considerable occurrence of pine forest, which is largely absent from the Hudsonian province, and which, at least in the interior, is mostly absent from the Carolinian province. However, the pine forests have a very sparse population of mammals, and no form of mammal seems to be restricted to the northern pine forest habitat. The assemblage of mammalian species ranging over most of the Canadian province is different from the assemblages of species living in either of the adjacent provinces. This is true in spite of the fact that no species or sub- species of mammal which ranges over all the province is restricted to it. It is believed, therefore, that the Canadian biotic province as described above forms a natural biogeographical unit for the mammals as well as for the vegetation. If a faunal area is to be recognized between the Hudsonian and the Can- adian provinces as here described it will have to be cut off from the southern part of the Hudsonian, from the northern part of the Canadian, or made up of parts of both provinces. All of the mammals listed above as occupying only the southern part of the Hudsonian of eastern Canada range far to the southward and all of them extend into the Carolinian province, except the fisher, flying-squirrel (Glau- comys subrinus), and moose, which stop at the northern boundary of the Carolinian. None of these forms, therefore, could be made the basis for a subdivision of the Hudsonian of eastern Canada into two longitudinal belts, the southern of which might be called the Canadian. Of the forms of more northern distribution which range south into the northern part of the Canadian as here delimited, the Arctic-shrew is rare and 506 October, 1938 CANADIAN BIOTIC PROVINCE 513 has a discontinuous distribution in this area; the original range of the wol- verine in this area is questionable and the species is now extirpated ; the northern lemming-vole occurs only in the most eastern part of the area; the heather-vole extends into only the extreme northern edge of the Canadian; and the caribou was probably nomadic and erratic in occurrence as it is now in other parts of its range. It is evident that no division of the Canadian to form also an Alleghanian fauna can be made on the basis of the distribution of these forms. A small number of forms of southern distribution extend their range into the southern part of the Canadian as I have mapped it. Of these it is ques- tionable if the eastern mole {Scalopus aquaticus), pipistrelle, southern flying- squirrel (Glaucomys volans), mole-mouse, or bison ever natively extended far into the province. The fox-squirrel is absent from the eastern part of the province. The cougar, elk, and bison are now absent from the region, and the gray-fox has been nearly or completely extirpated in Michigan. On the other hand the eastern-mole, bob-cat, fox-squirrel, wood-mouse, prairie deer-mouse and Mearns cottontail are extending their ranges northward in Michigan until they now occupy a considerably larger part of the Canadian than they did at the time the region was first settled. It would therefore seem inadvisable to base any major faunal area on the distribution of these species. The Canadian province, as I have described it, is not especially well marked by its mammalian species, and to establish two faunal divisions in this area instead of one, would mean that each of the two would not only have no characteristic species of mammals, but that only a few forms would reach their limits at or near the presumed faunal boundaries. Two biotic districts, an eastern and a western, can be recognized as sub- divisions of the Canadian biotic province. The red spruce (Picea rubra) which occurs over the eastern part of the Canadian and which also extends southward along the Appalachian Mountains, is absent from the western part of the province (Nichols, '35, fig. 5F). The beech is more widespread in the province than the red spruce, but it is absent from western Michigan, western Wisconsin, and Minnesota (Transeau, '35, fig. 12). The Gaspe shrew which occurs locally in the eastern part of the Canadian province, is absent from the western part. Several species of prairie mammals, as al- ready mentioned, invade the western part of the Canadian, but do not occur in the eastern part. Further, the water-shrew, woodchuck, chipmunk, deer- mouse, snowshoe hare, and perhaps other species of mammals are repre- sented by different subspecies in the eastern and western parts, respectively, of the province. New Brunswick does not greatly differ, however, from northern Michigan, either in its vegetation or in its mammalian fauna. There is no sharp transi- tion at any place between an eastern and western district of the Canadian 507 514 LEE R. DICE Ecology, Vol. 19, No. 4 province, and it is impossible to suggest, with the present lack of detailed in- formation, where the boundary between the two districts should be placed. Summary A detailed consideration of the vegetation and of mammalian distribution in eastern North America shows that the so-called Canadian and Alleghanian faunas are only different aspects of the same major ecologic complex. They are therefore here combined under the name Canadian biotic province. The Canadian biotic province is characterized by a hardwood climatic climax in which the hemlock and white pine frequently occur. Pines of several species constitute an important subclimax, or edaphic climax, on sandy and gravelly soils. In poorly drained situations spruces of several species, balsam fir, arbor vitae, and tamarack form another subclimax or group of subclimaxes. No species or subspecies of mammal which occurs all over the Canadian province is limited to the province. Nevertheless, the assemblage of species living in the province is different from the assemblages living in the adjacent provinces. Furthermore, a number of species of mammals reach their dis- tributional limits at or near the northern or southern boundaries of the province. Literature Cited Frothingham, E. H. 1915. The northern hardwood forest: its composition, growth, and management. U. S. Dcpt. Agric, Bull. 285, 79 pp. Fuller, Geo. D., and Brother Marie- Victorin. 1926. The Province of Quebec. In Naturalist's Guide to the Americas, pp. 293-299. Baltimore: Williams and IVilkins Co. Howe, C. D., and J. R. Dymond. 1926. Ontario. In Naturalist's Guide to the Americas, pp. 288-293. Baltimore: Williams and Wilkins Co. Merriam, C. Hart. 1898. Life zones and crop zones of the United States. U. S. Dept. Agric., Div. Biol. Surv., Bull. 10, 79 pp. Miller, Gerrit S., Jr. 1924. List of North American Recent mammals, 1923. U. S. Nat. Mus., Bidl. US. 16 + 673 pp. Nichols, G. E. 1935. The hemlock — white pine — northern hardwood region of eastern North America. Ecology 16: 403-422. Seton, Ernest T. 1929. Lives of game animals. Garden City, N. Y.: Doubleday, Doran, and Co. Shelford, V. E., L. Jones, and L. R. Dice. 1926. Descriptive list of North Ameri- can biota (south to central Mexico). In Naturalist's Guide to the Americas, pp. 60-74. Balti)nore: Williams and Wilkins Co. Transeau, Edgar N. 1935. The prairie peninsula. Ecology 16: 423-437. Weaver, John E. and Frederic E. Clements. 1929. Plant Ecology. Nczv York: McGraii'-Hill Book Co. xx -f 520 pp. 508 PLEISTOCENE ZOOGEOGRAPHY OF THE LEMMING, DICROSTONYX^ John E. Guilday Carnegie Museum, Pittsburgh, Pa. Received August 30, 1962 The collared lemmings, genus Dicros- tonyx Gloger, are currently divided into two subgenera. Misothermus Hensel con- tains a single species, D. hudsonius Pallas, isolated on the tundra of northern and coastal L^ngava from all other Dicrostonyx (see fig. 1). The subgenus Dicrostonyx Gloger contains the remaining species of the genus, D. torquatus Pallas of the palae- arctic, D. groenlandicus (Traill) of the nearctic, and D. exsul G. M. Allen confined to St. Lawrence Island in the Bering Straits. The torquatus-groenlandicus-exsid species group may be conspecific as inti- mated by Ellerman and Morrison-Scott (1951, p. 653). It is clear that they are more closely related to one another than to the isolated D. hudsonius. In the absence of any fossil record and interpreting on the basis of modern geo- graphical distribution alone, one might argue that the differentiation of D. hudson- ius dated from the Wisconsin glaciation; that as the ice front and its presumed periglacial tundra belt shrank to the north, the eastern segment of the retreating lem- ming population was cut off by Hudson Bay. The bay eventually cut the Canadian tundra into an eastern and a western com- ponent, each with its distinctive form of collared lemming. This does not appear to be the case, however. The inaccuracy of this interpretation is shown by the fossil record. The one record from the North American Pleistocene, frag- mentary skulls and mandibles of at least four individuals from Sinkhole no. 4, New Paris, Pennsylvania (Guilday and Doutt, 1961), is that of typical Misothermus (for characters, see Miller, 1898; Hinton, 1926; ^ Research conducted under National Science Foundation Grant no. G-20868. Evolution 17: 194-197. June, 1963 194 Hall and Kelson, 1959), indistinguishable from the modern D. hudsonius. Carbon particles taken from a position five feet higher in the sinkhole matrix were dated at 11,300 ± 1,000 years (Yale Univ. lab. no. 727). The age of the lemming remains is somewhat in excess of this. The Miso- thermus dental pattern was fixed prior to the Wisconsin recession and the formation of present Hudson Bay. There have been two species of collared lemmings described from the palaearctic Pleistocene. Dicrostonyx guliclmi Sandford based upon cranial material from Hutton Cave, Somersetshire, England, is a late Pleistocene form of the living D. torquatus, and may be conspecific with it (see Kowal- ski, 1959, p. 229). It is a common Eur- asian Pleistocene fossil. The second Old World fossil form, D. henseli Hinton, described from cranial material from a fissure deposit at Ightham, Kent, England, appears to be a typical Misothermus (see the description by Hin- ton, 1926, p. 163). D. {Misothermus) henseli has been recorded from the Pleis- tocene of England, Ireland, Jersey, France, and Germany (Hinton, 1926; Brunner, various papers) ; D. {Dicrostonyx) tor- quatus (or gulielmi), from the Pleistocene of England, Ireland, France, Poland, Czech- oslovakia (Hinton, 1926; Kowalski, 1959; Fejfar, 1961). This by no means exhausts the list of Old World Pleistocene Dicro- stonyx localities. But enough has been cited to indicate a geographical and pos- sibly a chronological overlap between the two species. Both forms were recovered by Hinton from Merlin's Cave, Wye Valley, Herefordshire, England (22 D. gulielmi skulls, 4 D. henseli) and at Langwith Cave, Derbeyshire (1 D. gulielmi, 2 or 3 D. hen- seli). Brunner recorded D. henseli from 509 ZOOGEOGRAPHY OF DICROSTONYX 195 Fig. 1. Outline map of North America, showing approximate limit of continental glaciation. A. Mainland modern distribution of the subgenus Dicrostonyx in America. B. Modern distribution of the subgenus Misothermus. thirteen Bavarian cave deposits. In one, the Markgrabenhohle, Brunner (1952c, p. 465) reported that 25% of the mandibles resembled D. gulielmi in possessing a small anteroexternal vestigial angle on M3, "eine deutliche aiissere Schmeltzfalte." Many uncorrelated fissure, cave, and terrace de- posits are involved, however. And while they are all middle to late Pleistocene in age, their sequential position within that time span has not been established with any degree of confidence. The facts at hand seem to indicate that two forms of the genus Dicrostonyx inhab- ited Eurasia and perhaps North America during the Pleistocene, and that one of them survives as a postglacial relict, isolated in the tundra of Ungava (fig. 2). The pre- Wisconsin origin of the Misothermus dental pattern is demonstrated by fossil forms in both continents. If we assume, as does Hinton (1926), that Misothermus is not a true phylogenetic category but that D. henseli and D. hud- 510 196 JOHN E. GUILDAY Fig. 2. Postulated distribution of Dicrostonyx (A) and Misotherntus (B) in North America at Wisconsin glacial maximum. Note site of Sinkhole no. 4, New Paris, Bedford County, Pennsylvania. Hachures indicate approximate limit of continental glaciation. sonius are of independent origin from Dicrostonyx proper, we are faced with the apparent coincidence that a form (henseli) was replaced by modern Dicrostonyx in the Old World while its morphological equiva- lent, D. hudsonius, which ranged as far south as central Pennsylvania during late Wisconsin times, survives today only where it is completely isolated from all contact with the Eurasian-Western Nearctic Dicro- stonyx. Both D. henseli and D. hudsonius appear to have been completely or partially re- placed by true Dicrostonyx. Is it possible that modern D. hudsonius (or some form of Misothermus) at one time ranged through- out the holarctic, and that it was replaced during late Pleistocene times by lemmings of the subgenus Dicrostonyx, first in the Old World, later in the New; this latter replacement occurring sometime after the post-Wisconsin formation of Hudson Bay and the division of the mainland North 511 ZOOGEOGRAPHY OF DICROSTONYX 197 American tundra into an eastern and a western component? I wish to thank Dr. J. Kenneth Doutt, Curator of Mammals, Miss Caroline A. Heppenstall, Assistant Curator of Mam- mals, and Dr. Craig C. Black, Gulf Associ- ate Curator of Vertebrate Fossils, Carnegie Museum, for their helpful criticism. Summary The modern distribution of Dicrostonyx hudsonius Pallas (confined to the tundra of Ungava) is believed, on the basis of the fossil record, to be a relict of a former holarctic pre-Wisconsin distribution. References Cited Brunner, Georg. 1949. Das Gaisloch bei Miiri- zinghof (Mfr.) mit Faunen aus dem Altdiluvium und aus jiingeren Epochen. Neuen Jahrbuch Mineral., etc. Abhandlungen, 91(B): 1-34. . 1951. Eine Faunenfolge von Wiirm III Glazial bis in das Spat-Postglazial aus der "Quellkammer" bei Pottenstein (Ofr.). Geol. Bl. NO-Bayern, 1: 14-28. . 19S2a. Das Dohlenloch bei Pottenstein (Obfr.). Eine Fundstelle aus dem Wiirm II Glazial. Abhandlungen Naturhist. Ges. Niirn- berg, 27(3): 49-60. . 19S2b. Der "Distlerkeller" in Pottenstein Ofr. Eine Faunenfolge des Wiirm I-III inter- stadial. Geol. Bl. NO-Bayern, 2: 95-105. . 1952c. Die Markgrabenhohle bei Potten- stein (Oberfranken). Eine Fauna des Altdi- luviums mit Talpa episcopalis Kormos u. a. Neues Jahrbuch Geol. Palaontol. Mh., 10: 457-471. . 1953a. Die Heinrichgrotte bei Burggaillen- reuth (Oberfranken). Eine Faunenfolge von Wiirm I-Glazial bis Interstadial. Neues Jahr- buch Geol. Palaontol. Mh., 6: 251-275. . 19S3b. Die Abri "Wasserstein" bei Betzen- stein (Ofr.). Eine subfossile Fauna mit Sorex tniniitissimus H. de Balsac. Geol. Bl. NO- Bayern, 3: 94-105. — . 1955. Die Hohle am Butzmannsacker bei Auerbach (Opf.). Geol. Bl. NO-Bayern, 5: 109-120. — . 1956. Nachtrag zur Kleinen Teufelshohle bei Pottenstein (Oberfranken). Eine Ubergang von der letzten Riss-Wiirm-Warm-fauna zur Wiirm I-Kaltfauna. Neues Jb. Geol. Palaontol. Mh., 2: 75-100. — . 1957a. Die Breitenberghohle bei Gosswein- stein/Ofr. Eine Mindel-Riss-und eine post- glaziale Mediterran-Fauna. Neues Jb. Geol. Palaontol. Mb., 7-9: 352-378, 385-403. — . 1957b. Die Caciliengrotte bei Hirschbach (Opf.) und ihre fossile Fauna. Geol. Bl. NO- Bayern, 7: 155-166. — . 1958. Das Guckerloch bei Michelfeld (Opf.). Geol. Bl. NO-Bayern, 8: 158-172. 1959. Das Schmeidberg-Abri bei Hirsch- bach (Oberpfalz). Palaont. Z., 33: 152-165. Ellerman, J. R., AND T. C. S. Morrison-Scott. 1951. Checklist of Palaeoarctic and Indian mammals. British Museum (Natural History) . London. 810 p. Fejfar, Oldrich. 1961. Review of Quaternary Vertebrata in Czechoslovakia. Instytut Geo- logiczny. Odbitka z Tomu 34 Prac. Czwar- torzed Europy Srodkowej I Wschodniej, p. 109-118. Guilday, J. E., and J. K. Doutt. 1961. The Col- lared Lemming (Dicrostonyx) from the Penn- sylvania Pleistocene. Proc. Biol. Soc. Wash- ington, 74: 249-250. Hall, E. R., and R. K. Kelson. 1959. The mam- mals of North America. Vol. II, p. 547-1083. Hinton, M. a. C. 1926. Monograph of the voles and lemmings (Microtinae) living and ex- tinct. Vol. I. Richard Clay and Sons, Suffolk. 488 p. KowALSKi, K. 1959. Katalog Ssakow Plejstocenu Polski. Polska Akademia Nauk. Inst. Zool. Oddzial W Krakowie. 267 p. Miller, G. S., Jr. 1896. Genera and subgenera of voles and lemmings. North American Fauna no. 12, 80 p., U. S. Dept. Agriculture. 512 SUBFOSSIL MAMMALS FROM THE GOMEZ FARIAS REGION AND THE TROPIGAL GRADIENT OF EASTERN MEXICO By Karl F. Koopman and Paul S. Martin In the spring of 1953 Byron E. Harrell and P. S. Martin collected superficial animal remains at three cave localities in the Sierra Madre Oriental of south- western Tamaulipas. Locally, the mountains are sufficiently humid to support small, isolated patches of Cloud Forest and Tropical Evergreen Forest ( Martin, 1958). At this latitude, 23° north, these two tropical plant foraiations appear to reach their limit. In recent years four faunal papers on mammals of southern Tamaulipas have appeared, each reporting certain tropical species unknown at higher latitudes (Baker, 1951; Goodwin, 1954; Hooper, 1953; de la Torre, 1954). Southern Tamaulipas appears to be of primary significance in the Gulf lowlands faimal gradient, a system extending from southern Veracruz to southern Texas. In the lowlands of eastern Mexico many of the dominant tropical American taxa reach their range limits. We seek to define the north- eastern section of this gradient with regard to tropical mammal faunas, to relate it to shifts in vegetation and to indicate the relative importance of the Gomez Farias region as a faunal terminus. The Gomez Farias region is defined as the area from 22°48' to 23°30' north latitude and 99° to 99°30' west longitude, or approximately the rectangle enclosed by the towns of Llera, Jaumave, Ocampo and Limon. In establishing the presence of eight species known in the Gomez Farias region only from the skeletal remains, and in extending the altitudinal or ecological ranges of others, the present collection supplements previous reports. It indicates the value of using owl peUet deposits as a cross check on standard trapping technique. DESCRIPTION OF DEPOSITS Cretaceous limestones comprising the precipitous east slope of the Sierra Madre Oriental near the village of Gomez Farias are severely folded. Under torrential summer rains they have eroded into very rough karst terrain with virtually no surface drainage. Caves and sink holes, a characteristic of karst, are a common feature. Most do not appear to be inhabited by bats or owls. Of those that are, the constant high humidity and frequent flushing from 513 JOURNAL OF MAMMALOGY Vol 40, No. 1 percolating rainwater apparently prevent accumulation of deep bone or guano deposits. Large bat colonies numbering thousands of individuals occur in Tropical Deciduous Forest in the Canyon of the Rio Boquilla, 8 km. southwest of Chamal, and in a guano cavern at El Abra, 8 km. northeast of Antiguo Morelos. However, no colonies of similar size have been found in the more himiid portions of the G6mez Farias region. Of approximately thirty caves and sink holes explored near Gomez Farias, material from three is represented in our collection. All occur in the mountains west of the village of G6mez Farias, latitude 29°03' north, longitude 99°09' west, and within 16 km. of each other ( see Table 1 ) . The following information will serve to characterize them: 1. Paraiso. Aserradero del Paraiso is the name of a small sawmill located in Tropical Evergreen Forest 13 km. north-northwest of Chamal. A narrow ravine about 1 km. south of the sawmill harbors several caves, including a deep, wet grotto with permanent water. On the sloping floor of a small Table 1. — Mammals from cave deposits in the Gdmez Farias region (numbers refer to anterior skull parts identified) Distance from G6mez Farias: Elevation in meters: Vegetation type: PABAISO SW 13 km. 420 Tropical Ever- green Forest RAJ^CHO DEL CIELO 6 km. NW 1050 Cloud Forest INFERNO 7 km. W 1320 Cloud Forest Didelphis marsupudis — Marmosa mexicana Cryptotis pergracilis Cryptotis mexicana Chilonycteris parnellii Enchisthenes harti Artibeus cinereus Centurio senex Eptesicus fuscus Lasiurus cinereus Antrozous paUldus Sylvilagus floridanus Glaucomys volans Liomys irroratus Reithrodontomys mexicanus Reithrodontomys megalotis . Baiomys taylori Peromyscus boylei Peromyscus pectorclis Peromyscus ochraventer Oryzomys alfaroi Sigmodon hispidus Neotoma angustapalata Total identifications 3 1 1 1 1 1 3 29 9 51 1 3 1 2 1 9 1 18 10 1 1 1 12 5 1 17 2 7 5 17 98 514 Feb., 1959 KOOPMAN AND MARTIN— MAMMALS OF EASTERN MEXICO 3 dry cave on the west wall of this ravine many scattered (apparently water- transported) small bones were found. 2. Rancho del Cielo. A small sink and cave several hundred meters east of this important Cloud Forest collecting locality contained a few bat remains and fresh owl pellets. 3. Inferno (also designated Infemillo). A rather large cave adjacent to an abandoned sawmill of this name is located in upper Cloud Forest roughly 2 km. south of tlie mill settlement called La Gloria. Abundant pellet remains were found on a dislodged boulder just inside the cave mouth. All the caves are surrounded by heavy forest, much of it recently lumbered. Jagged, shrub-covered or almost bare karst ridges and pinnacles confine the areas of tall forest to valleys, pockets and gentle slopes. A more complete description of the Gomez Farias region is in preparation (Martin, 1958). Although the bam owl, Tyto alba, occurs in a very large cavern at El Abra north of Antiguo Morelos, the only raptor definitely known to roost in caves in the Gomez Farias region is the wood owl, Ciccaba virgata. While it is possible that certain non-volant mammals died from other causes, all skeletal remains of these, and perhaps even some of the bats, can be ascribed to Ciccaba. Cave records do not make ideal locality data, especially when information on ecological or altitudinal distribution is sought. We assume that the forest types and the elevation in the immediate vicinity of each cave represent the habitat in which the owls fed, but this is by no means certain. The hunting range and nocturnal movements of Ciccaba are unknown. There is no stratigraphic evidence for assuming that any material is older than very recent. Limestone deposit on certain bones may represent the concretion of a single rainy season. SYSTEMATIC TREATMENT OF THE MAMMALS We emphasize first that only a fraction of the mammalian and none of the bird bones have been identified. Very little attempt has been made to identify any of the post-cranial elements, and in the case of the cricetine rodents only the best of cranial material could be identified witli any confidence. Particular difficulty was encountered with lower jaws of cricetines, even when teeth were present. As a result, determinations in this group are considered less reliable than in the others. In particular this applies to the Feromyscus- Baiomys-Reithrodontomys group of genera, in which not only specific but also good generic characteristics were hard to find in most of the material. Size, molar form, palatal morphology and shape of the zygomatic plate were found to provide the most useful taxonomic characters in this subfamily. In many cases a closely adhering limestone drip deposit was present and was sometimes difficult to remove without damaging tlie underlying bone. No attempt has been made to identify subspecies. In our opinion the theoretical requirements of population sampling and measurement, as employed in sub- 515 4 JOURNAL OF MAMMALOGY Vol. 40, No. 1 specific identification, are not met in any but perhaps the best fossil material. For the following species determinations Koopman alone is responsible. Didelphis marsupialis. — Inferno: one rostral fragment of a young individual. Previously recorded from Gomez Farias (Hooper, 1953). Marniosa mexicana. — Inferno: three rostral fragments and 15 mandibles. On the basis of dentition, these are clearly opossums, but of the Middle Amer- ican genera of Didelphidae, all but Marmosa are much too large. Of the four species of this genus occurring north of Panama, all but M. mexicana differ from the subfossil material either in larger size or greater development of a precondylar crest on the outer side of the mandible. The species has not previously been recorded north of Jalapa, Veracruz. Cryptotis per gracilis. — Paraiso: one rostnmi and two mandibles. Restriction to the family Soricidae and the genus Cryptotis may be made on the basis of dentition. In eastern Mexico north of the Isthmus of Tehuantepec (states of Tamaulipas, San Luis Potosi, Hidalgo, Puebla and Veracruz) the following six species of Cryptotis are known: C. parva, C. pergracilis, C. obscura, C. micrura, C. mexicana and C. nelsoni. All except C. parva and C. pergracilis may be ruled out on the basis of larger size. While clear-cut cranial differences between the latter two species appear to be absent, the eastern race of per- gracilis, C. p. pueblensis, has a somewhat deeper nasal emargination of the rostrum than the southwestern race of parva, C. p. berlandieri. Though slightly broken anteriorly, the subfossil rostrum appears to have a nasal emargination somewhat closer to that of C. pergracilis pueblensis than to that of the two Tamaulipan specimens of C. parva berlandieri recorded by Goodwin (1954). However, no direct comparison of the Paraiso specimens with C. pergracilis pueblensis was made, but only with a sketch. Unfortunately our material seems inadequate to solve the problem of specific status, i.e., are C. parva and C. pergracilis sympatric in southern Tamaulipas, do they integrade through a narrow hybrid zone, or is there a more complex arrangement? The northernmost previous record of C. pergracilis is Platanito in San Luis Potosi. Cryptotis mexicana. — Inferno: six rostra, four mandibles, and three humeri; Rancho del Cielo: one rostrum. Specimens were trapped at the latter locality (Goodwin, 1954). Chilonycteris parnellii. — Paraiso: one mandible. Goodwin (1954) records a series from El Pachon. As Koopman (1955) has pointed out, the mainland C. rubiginosa and the West Indian C. parnellii are almost certainly conspecific. Since Koopman believed that C. parnellii Gray had several months priority over C. rubiginosa Wagner, the combined species was called C. parnellii. De la Torre (1955) has shown that this is not the case and, finding no way of determining which name was published first, recommended: "In the absence of conclusive evidence, the better known and more widely used name rubiginosa should be retained." Unfortunately, if there is no clear priority, the law of the first reviser would 516 Feb., 1959 KOOPMAN AND MARTIN— MAMMALS OF EASTERN MEXICO 5 seem to hold, in this case tlie first to use one of the two names to include both forms, i.e., Koopman (1955). It is not legally possible to withdraw from this position. Therefore it appears that C. parnellii must stand as the name assigned to both mainland and West Indian large Chilonycteris. The question of nomenclature should not obscure the more significant taxo- nomic conclusion that a single species is involved. Artibeus cinereus. — Rancho del Cielo: two partial skulls, one lower jaw. De la Torre has also recorded this species from Rancho del Cielo. Enchisthenes harti. — Inferno: one partial skull. The short broad rostrum and characteristic molar pattern rule out all American bats outside the Steno- derminae. Of the Middle American stenodermines, only Uroderma bilobatum, Vampyrops helleri and Enchisthenes harti agree with the Inferno skull in size and dental formula (i- o^ p- m^). Both Uroderma and Vampyrops, however, have rostra considerably longer than that of the subfossil skull. On the other hand, there is close resemblance to skulls of Enchisthenes from Honduras and Ecuador. De la Torre (1955) has recently summarized the knovsoi records, specimens from Ciudad Guzman in Jalisco being the closest to Tamaulipas geographically. Four other localities extend the distribution south to Trinidad and Ecuador. Centurio senex. — Inferno: one rostrum. Of all the North and Middle Amer- ican bats, only Centurio senex agrees with the subfossil skull in dental formula (i^ c^ p^ m^) and in the palate being more than twice as wide as it is long. The Inferno rostnim resembles a skull of Centurio senex in all important respects. De la Torre (1954) recorded a single specimen from Pano Ayuctle. Eptesicus fuscus. — Paraiso: one mandible. Several characters of this bone inmiediately narrow the field considerably. These are mandibular length, tooth size, dental formula (is Ci p2 ma), molar pattern and height of the coronoid process. This leaves us with only two North and Middle American species, Eptesicus fuscus and Dasypterus intermedius. Of these, Dasypterus may be ruled out on the basis of its more robust mandibular ramus. Comparison of the Paraiso mandible with Eptesicus fuscus reveals no important differences. I have been able to find no other records of this bat in Tamaulipas, the nearest localities being Rio Ramos in Nuevo Leon to the west (Davis, 1944) and Caiiada Grande in San Luis Potosi to the southwest (Dalquest, 1953). Good series from Tamaulipas, if they could be obtained, should show integradation between E. f. fuscus and £. /. miradorensis. Lasiurus cinereus. — Inferno: one nearly complete skull. All other species of North and Middle American bats may easily be excluded from consideration on the basis of size, rostral shape and dental formula ( i^ c^ p^ m^ ) . It matches L. cinereus closely. Since this bat is migratory, it is impossible to say whether this individual belonged to a resident population or was merely a winter visitor. The nearest previous records are Matamoros in northern Tamaulipas (Miller, 1897) and EI Salto in eastern San Luis Potosi (Dalquest, 1953). Antrozous pallidus. — Paraiso: one partial lower jaw. Mandible and tooth 517 6 JOURNAL OF MAMMALOGY Vol. 40, No. 1 size, molar pattern and dental formula (12 Ci p2 ma) rule out all North and Middle American bats except Promops centralis and Antrozous. Promops may be excluded by the quite different appearance of the labial surface of the coronoid region. Of the two species of Antrozous recognized by Orr ( 1954 ) , A. bunkeri is distinctly larger than the Paraiso mandible. A. pallidus resembles it in all respects. The University of Michigan Museum of Zoology has six specimens from Tula, which were mentioned by Orr (1954). Sylvilagus floridanus. — Paraiso: one maxillary fragment of a young indi- vidual. Goodwin (1954) records the species from Gomez Farias, Pano Ayuctle and Chamal. Glaucomys volans. — Inferno: two palatal fragments, ten mandibles; Rancho del Cielo: one mandible; Paraiso: one mandible. From the dental formula 0 2 ^ (ii Co Pi ms) these specimens are clearly referable to the Sciuridae, of which all northeastern Mexican species except Eutamms hulleri, E. dorsalis and Glau- comys volans are clearly too large. In Eutamias, however, the mandible is much less deep than in the Tamauhpas material. The latter bears a convincing resemblance to G. volans. The nearest locahty from which the species had previously been obtained is Santa Barbarita in San Luis Potosi ( Dalquest, 1953 ) . Liomys irroratus. — Paraiso: one maxillary fragment. Goodwin (1954) re- cords a series from Pano Ayuctle. Oryzomys alfaroi. — Inferno: four maxillary fragments, one partial skull; Paraiso: three mandibles. Goodwin (1954) and Hooper (1953) have each recorded the species from Rancho del Cielo. R. (Reithrodontomys) megalotis. — Inferno: one partial skull. Identification of this fragment is tentative. The species has already been recorded from Rancho del Cielo by both Goodwin (1954) and Hooper (1953). R. (Aporodon) mexicanus. — Inferno: five partial skulls; Rancho del Cielo: two partial skulls. The species has been recorded previously from Rancho del Cielo by Goodwin (1954) and Hooper (1953). Peromyscus boylei. — Inferno: two partial skulls, 15 maxillaries; Rancho del Cielo: one mandible. The species is recorded by Goodwin (1954) from both Rancho del Cielo and Rancho Viejo. Peromyscus pectoralis. — Inferno: one partial skull, one maxillary. It has been recorded by Goodwin (1954) from both La Joya de Salas and 2 km. west of El Carrizo. Peromyscus ochraventer. — Inferno: one partial skull, six maxillaries. The species has been recorded from Rancho del Cielo by both Goodwin (1954) and Hooper (1953). Baiomys taylori. — Paraiso: one mandible. Goodwin (1954) and Hooper (1953) have recorded it from Pano Ayuctle. Sigmodon hispidus. — Paraiso: one rostral half, ten maxillaries, two pre- maxillaries, three braincase elements, 16 mandibles. The species has been recorded from Pano Ayuctle by Goodwin (1954) and Hooper (1953). Neotoma angustapalata. — Inferno: four partial rostra, five maxillaries, eight 518 Feb., 1959 KOOPMAN AND MARTIN— MAMMALS OF EASTERN MEXICO 7 mandibles; Rancho del Cielo: one maxillary; Paraiso: one maxillary fragment, two premaxillaries, six mandibles. There is also a great deal of additional Neotoma material from Inferno that probably belongs here, but which has not been specifically identified. Both Goodwin (1954) and Hooper (1953) list specimens from Rancho del Cielo and El Pachon although, as Hooper points out, the precise status of N. angnstapalata and its various southern Tamaulipas populations is far from clear. At the present time this name appears to be something of a "catch-aU." It is felt, however, that a revision should be based on entire specimens rather than on skeletal fragments. FAUNAL COMPARISONS AND THE STATUS OF GLAUCOMYS The identifications summarized in Table 1 represent total number of an- terior skull elements and not total number of individuals, which may be some- what less. Although such data do not lend themselves to close quantitative inspection, we feel that the faunas sampled near Inferno and Paraiso reveal important differences. Two quite different habitats. Cloud Forest and Tropical Evergreen Forest, are represented. It would be surprising if the faunas were qualitatively similar. Peromyscus, the dominant genus comprising 27 per cent of the Inferno deposits, is unrepresented at Paraiso. In turn, Sigmodon, which comprises 57 per cent of the material obtained at Paraiso, is absent from Inferno. Such a discrepancy may reflect an ecological shift in the dominant cricetine form. Other rather common species which appeared only in the Cloud Forest caves include: Marmosa mexicana (18%), Cryptotis mexicana (10%) and Reithrodontomys mexicanus (5%). One genus, Neotoma, occurs at both localities with a relatively constant frequency, 17-18 per cent. Within the G6mez Farias region the following are known only from their skeletal remains: Marmosa mexicana, Cryptotis per gracilis, Chilonycteris parnellii, Enchisthenes harti, Eptesicus fuscus, Lasiurus cinereus, Antrozous pallidus and Glaucomys volans. Presence of the latter is perhaps of greatest interest. In Middle America flying squirrels are very poorly known, presumably the result of their nocturnal and arboreal habits rather than an inherent scarcity. We are aware of ten other locaHty records between Chihuahua and Honduras, each represented by one or two specimens. In the Gomez Farias region remains of Glaucomys appeared at each cave locality, suggesting a general range through humid forest, both Cloud Forest and Tropical Evergreen Forest, between 420 and 1,320 meters. To our knowledge Glaucomys has not previously been collected in lowland tropical forests (below 1,000 meters). THE LOWLAND TROPICAL GRADIENT IN EASTERN MEXICO We might imagine a lowland tropical fauna to decline with increasing lati- tude at a rather regular rate. However, in reality the environment and fauna undergo a series of discrete changes, stepwise ( see Fig. 1 ) . In eastern Mexico four major tropical lowland vegetation types are represented. From south to north they terminate in the following sequence: Rainforest, Tropical Ever- 519 8 JOURNAL OF MAMMALOGY Vol 40. No. 1 green Forest, Tropical Deciduous Forest and Thorn Forest (Leopold, 1950). Each terminus is marked by a steepening of the faunal gradient. In defining the lowland tropical fauna we have excluded those genera with distributional centers in temperate montane habitats or of limited lowland tropi- cal ranges, e.g., Idionycteris, Neotoma, Sciurus, Sigmodon and Baiomys. On the other hand, some of the species selected may range into montane habitats or reach temperate latitudes, such as Marmosa mexicana and Didelphis mar- supialis. In general the species hsted in Table 2 are of wide distribution in lowland tropical America. Although they occupy tropical environments of the Mexican escarpment and coastal plain, we exclude Baiomys and Sigmodon because their phylogeny indicates a north temperate origin and in the case of Baiomys because it does not range extensively into Central America. Clearly the question of "tropicality" can be vexing. As one might expect the bats have much more extensive ranges than the small terrestrial mammals. The faunas of Trinidad and southern Tamaulipas share 18 species of bats, but none of rodents. NORTHERN LIMIT OF TROPICAL MAMMALS 10 20 30 40 VEGETATION NUMBER OF TROPICAL SPECIES TYPE Fig. 1. — Relationship between lowland tropical mammals and latitude in northeastern Mexico. Known range limits for 43 species listed in Table 2 are shown on the map. Major vegetation types are indicated at the right. Tropical Evergreen Forest is abbreviated to TEF. 520 Feb., 1959 KOOPMAN AND MARTIN— MAMMALS OF EASTERN MEXICO 9 The general relationship between vegetation, latitude and tropical fauna is shown on Fig. 1. In southern Tamaulipas a rapid "tliinning" of tropical forms is evident. Between 23° and 24° north latitude the following 17 genera find their range limits: Philander, Marmosa, Chilonycteris, Pteronotus, Micro- nycteris, Macrotus, Glossophaga, Sturnira, Artibeus, Enchisthenes, Centurio, Natalus, Wwgeesa, Molossus, Heterogeomys, Eira and Mazama. At this latitude (24° north) the Tropical Deciduous Forest, well developed and widespread east of the Sierra de Tamaulipas and the Sierra Madre Oriental, disappears. Small, perhaps relict, stands of Cloud Forest and Tropical Evergreen Forest in the Gomez Farias region also enrich the environmental opportunity for tropical mammals. However, these habitats, especially the Tropical Evergreen Forest, are more extensive in southeastern San Luis Potosi and northern Veracruz. In this region seven genera terminate: Carollia, Tamandua, Coendou, Cuni- culus, Potos, Galictis and Ateles. Although the vegetation near Xilitla, San Luis Potosi, has been designated as Rainforest, it seems preferable to reserve that term for the more luxuriant forests of southern Veracruz with their short dry season. Lowland tropical forests near Xilitla appear taller and richer than those of southern Tamaulipas; however, the area has suffered a long history of intensive Huastecan agriculture. Almost certainly the primeval fauna of southeastern San Luis Potosi included a larger number of tropical genera at their northern limit. While we have not attempted to represent tropical distributions and plant formations south of San Luis Potosi, the following genera or subgenera approach their northern limit in southern Veracruz: Caluromys, Vampyrum, Rhynchiscus, Centronycteris, Mimon, Chrotopterus, Hylonycteris, Chiroderma, Alouatta, Tylomys, Dasyprocta, Tayassu, Tapirella and Jentinkia. Is there a relationship between these and the northern limit of Rainforest (see Leopold, 1950)? Extending from central Tamaulipas northward to Nuevo Leon and southern Texas is a rather barren Thorn Forest and Thorn Scrub. Gradually these arid habitats lose their tropical character as the Rio Grande Valley is approached. By comparison with plant formations to the south, this environment is poor in tropical fauna. The shift from tropical to temperate thorn scrub involves no sharp faunal boundary among the mammals. Oryzomys couesi and Liomys irroTOtus are among the forms reaching southern Texas. At this latitude the herpetological fauna includes such tropical genera as Coniophanes, Drymobius, Leptodeira, Smilisca and Hypopacus. However, among the reptiles and am- phibians, as well as the mammals, the greatest reduction in tropical fauna is found in southern Tamaulipas (Martin, 1958). DISCUSSION Although our analysis in confined to eastern Mexico, some interesting com- parisons can be made with the tropical biota of the Pacific Coast. Arid tropical vegetation and the genera Macrotus, Balantiopteryx, Chilonycteris, Pteronotus, Mormoops, Glossophaga, Desmodus, Natalus, Rhogeesa and Nasua extend far- 521 10 JOURNAL OF MAMMALOGY Vol. 40, No. 1 Table 2. — Northern limits of neotropical mammals LOCALITY AND APPROX. ELEV. REFERENCE SPECIES PRESENT San Luis Potosi: A. Tamazunchale, 120 m. B. Xilitla and vicinity 630-1350 m. Dalquest, 1953 Dalquest, 1953 1. Tamandua tetradactyla 2. Sturnira ludovici 3. Coendou mexicanum 4. Cuniculus paca 5. Potos flavus 6. Galictis canaster C. Rio Verde, 990 m. Dalquest, 1953 7. Molosstis major D. VaUes, 75 m. Koopman, 1956 8. Balantiopteryx plicata E. EI Salto, 660 m. Dalquest, 1953 9. Carollia perspicillata Tamaulipas: F, Rancho del Cielo Goodwin, 1954 10. Marmosa mexicana and vicinity, Hooper, 1953 11. Enchisthenes harti 1000-1320 m. present report 12. Oryzomys alfaroi 13. Reithrodontomys (Aporodon) mexicanus 14. Mazama americana Pano Ayuctle, 100 m. Goodwin, 1954 15. Micronycteris megalotis Hooper, 1953 16. Sturnira lilium de la Torre, 1954 17. Artibeus jamaicensis 18. A. lituratus 19. A. cinereus 20. Eira barbara G. 2 km. W of El Carrizo, Baker, 1951 21. Philander opossum 800 m. 22. Heterogeomys hispidus H. Jaumave, 730 m. UMMZ specimens 23. Macrotus mexicanus I. 10-16 mi. WSW Anderson, 1956 24. Chilonycteris parnellii Piedra, 400 m. 25. Glossophaga soricina 26. Centurio senex 27. Natalus mexicanus J. 2 mi. S Victoria, Davis, 1951 28. Molossus rufus 400 m. K. 30 km. NW Victoria, Malaga-Alba, 1954 29. Diphylla ecaudata 1000 m. L. La Pesca, 10 m. Anderson, 1956 30. Rhogeesa tumida M. Rancho Santa Rosa, Anderson, 1956 31. Pteronotus davyi 260 m. N. 8 mi. SW Padilla, Lawrence, 1947 32. Oryzomys melanotis 100 m. Nuevo Ledn: O. 25 km. SW Linares, Malaga-Alba, 1954 32. Desmodus rotundus 700 m. P. 20 mi. NW General Hooper, 1947 33. Oryzomys fulvescens Teran, 300 m. Tera.^: Q. Hidalgo Co., 60 m. Blair, 1952b 34. Oryzomys couesi R. Raymondville, 30 m. Blair, 19526 35. Liomys irroratus 522 Feb., 1959 KOOPMAN AND MARTIN— MAMMALS OF EASTERN MEXICO H Table 2. — Continued LOCALITY AND APPROX. ELEV. REFERENCE SPECIES PRESENT Coahuila and West Texas: Edwards Plateau, Texas: Eastern United States: Baker, 1956 Miller and KeUogg, 1955 Blair, 1952a Blair, 1952& Miller and KeUogg, 1955 36. Choeronycteris mexicana 37. Leptonycteris nivalis 38. Mormoops megalophylla 39. Nasua narica 40. Dasypus novemcinctus 41. Didelphis marsupialis ther north on the western side. On the other hand the humid tropical fauna and plant formations, e.g., Cloud Forest, Tropical Evergreen Forest and Rainforest, are absent or poorly represented on the Pacific slope north of Chiapas. As a general rule for those species or vicariant species occurring on both sides of Mexico, the arid tropical forms range farther north on the west, and the humid tropical forms farther north on the eastern side. This pattern is evident also in the distribution of lowland tropical birds, reptiles, insects, etc. Other than noting a rather close "fit," it is beyond our purpose to explore the causal relationship between formation and fauna. In brief our conclusions may be summarized as follows: 1. The decline of the tropical fauna in eastern Mexico corresponds with the vegetation gradient. 2. Where the vegetation gradient steepens and a plant formation is lost, one finds a variety of tropical animals at their range limits. 3. In southern Tamauhpas the northern limit of many tropical mammals corresponds roughly to the boundary of Tropical Deciduous Forest. Contribu- ting to the rich tropical fauna of the Gomez Farias region are relict outposts of Tropical Evergreen Forest and Cloud Forest. 4. The problem of establishing a Nearctic-Neotropical faunal boundary in eastern Mexico can be approached reaUstically in terms of steps in an environ- mental gradient. ACKNOWLEDGMENTS First we wish to thank Dr. Byron E. Harrell for assistance in the field work. We are indebted to the Mammal Department of the American Museum of Natirral History for use of their facilities including collections of comparative material. Mr. Sydney Anderson of the Museum of Natural History, University of Kansas, and Dr. William H. Burt, Museirai of Zoology, University of Michigan, kindly advised us concerning specimens in their care. LITERATURE CITED Anderson, Sydney. 1956. Extensions of known ranges of Mexican bats. Univ. Kans. Publ. Mus. Nat. Hist., 9: 349-351. Baker, Rollin H. 1951. Mammals from Tamauhpas, Mexico. Univ. Kans. Publ. Mus. Nat Hist., 5: 207-218. 523 12 JOURNAL OF MAMMALOGY Vol. 40, No. 1 . 1956. Mammals of Coahuila, Mexico. Univ. Kans. Publ. Mus. Nat. Hist., 9: 125-335. Blair, W. Frank. 1952a. Bats of the Edwards Plateau in Central Texas. Tex. Jour. Sci., 4: 95-98. . 19526. Mammals of the Tamaulipan Biotic Province in Texas. Tex. Jour. Sci., 4: 230-250. Dalquest, Walter W. 1953. Mammals of the Mexican State of San Luis Potosi. La. State Univ. Studies, No. 1: 1-112. Davis, William B. 1944. Notes on Mexican mammals. Jour. Mamm., 25: 370—403. . 1951. Bat, Molossus nigricans, eaten by the rat snake, Elaphe laeta. Jour. Mamm., 32: 219. GooDWTN, George G. 1954. Mammals from Mexico collected by Marian Martin for the American Museum of Natural History. Amer. Mus. Novit., No. 1689: 1-16. Hooper, Emmet T. 1947. Notes on Mexican mammals. Jour. Mamm., 28: 40-57. . 1953. Notes on mammals of Tamauhpas, Mexico. Occ. Pap. Mus. Zool., Univ. Mich., No. 544: 1-12. Koopman, Karl F. 1955. A new subspecies of Chilonycteris from the West Indies and a discussion of the mammals of La Gonave. Jour. Mamm., 36: 109-113. . 1956. Bats from San Luis Potosi with a new record for Balantiopteryx plicata. Jour. Mamm., 37: 547-548. Lawrence, Barbara. 1947. A new race of Oryzomys from Tamauhpas. Proc. New England Zool. Club, 24: 101-103. Leopold, A. Starker. 1950. Vegetation zones of Mexico. Ecology, 31: 507-518. Malaga-Alba, Aurelio. 1954. Vampire bat as carrier of rabies. Amer. Jour. Pubhc Health, 44: 909-918. Martin, Paul S. 1958. Herpetology and biogeography of the G6mez Farias region, Mexico. Misc. Publ. Mus. Zool., Univ. Mich. In press. Miller, Gerrit S., Jr. 1897. Revision of the North American bats of the family Vespertilionidae. N. Amer. Fauna 13: 1-155. and Remustgton Kellogg. 1955. List of North American Recent mammals. Bull. U.S. Nat. Mus., 205: 1-954. Orr, Robert T. 1954. Natural history of the paUid bat, Antrozom pallidus (Le Conte). Proc. Calif. Acad. Sci., 28: 165-246. Torre, Lins de la. 1954. Bats from southern Tamaulipas, Mexico. Jotu. Mamm., 35: 113-116. . 1955. Bats from Guerrero, Jalisco, and Oaxaca, Mexico. Fieldiana: Zool., 37: 695-703. Dept. of Biology, Queens College, Flushing, New York and Ijistitut de Biologie, Universite de Montreal, Montreal, Canada. Received August 6, 1957. 524 ZOOGEOGRAPHY OF THE MONTANE MAMMALS OF COLORADO By James S. Findley and Sydney Anderson In Colorado the distribution of montane or boreal habitats is at present closely- associated with the local chmate produced by the mountains. Peculiarities of this habitat are high precipitation, both in winter and summer, cool temperatures, a continuous water supply and a coniferous forest. Certain mammals are more or less restricted in geographic range, in this part of the continent, to the mountains. In Pleistocene time the distribution of boreal habitat and hence of boreal mammals has undoubtedly fluctuated widely with the advances and retreats of continental and alpine glaciers. The contemporary pattern of distribution is, at least in part, a result of the most recent major glacial advance and retreat which took place in the Wisconsinan Age. It seems probable to us that most contemporary subspecies have differentiated in late Pleistocene time; otherwise the frequent correspondence of their ranges with current topographical and ecological features, which stem from late Pleistocene events in many cases, seems inexplicable. In the western United States the Boreal Zone is found at higher and higher elevations as one proceeds southward until it is scattered on isolated mountain peaks. The presence of isolated populations of boreal mammals on some of these mountains is evidence of a former displacement southward and downward in altitude of the Boreal Zone in a glacial age, presumably the Wis- consin, and subsequent elevation of the Boreal Zone in altitude and latitude in an ensuing interglacial interval, presumably the Recent. These southern, marginal populations would have been the first to become isolated with the retreat of the ice. The separation of boreal habitat in the mountains of Colorado from boreal habitat in the Uinta and Wasatch Mountains of Utah and the mountains of northwestern Wyoming is probably of later origin than is the isolation of the southern boreal "islands." We have studied the boreal mammals of Colorado in their relation to those of Utah and northwestern Wyoming. These mammals may be grouped according to the pattern of their variation and distribution as follows: Group I. — Rare, extinct, or insufficiently known to use in this study: Alces americana, Ovis canadensis, Lepus americanus, Sylvilagus nuttallii, Phenacomys intermedius, Mustela erminea, and Gulo luscus. Group II. — Occurring only north and west of the barrier formed by the Wyo- ming Basin and the Green River (Fig. 1) : Eutamias amoenus, Glaucomys sahrinus, Microtus richardsoni, and Martes pennanti. Group III. — Occurring only southeast of the above mentioned barrier: Sciurus aberti. Group IV.— Occurring in the mountains of northwestern Wyoming and the mountains of Colorado as a single subspecies; this group includes eight of fifteen species that occur on both sides of the barrier shown in Figure 1 : Sorex cinereus, Sorex vagrans, Sorex palustris, Clethrionomys gapperi, Microtus montanus, Micro- tus longicaitdus, Zapus princeps, and Erethizon dorsatum. Group V. — Occurring in the mountains north of the Wyoming Basin and the mountains southeast of the basin, but as different subspecies: Martes americana, 525 Feb., 1956 FINDLEY AND ANDERSON— MONTANE MAMMALS 81 105 Fig. 1. — The distribution of the Boreal Zone (diagonally lined) in Wyoming, Colorado, and Utah. The major barrier (consisting of the Wyoming Basin and the Green River) sepa- rating the boreal habitat in Colorado from the mountains of Utah and northwestern Wyo- ming is shown. Marmota flaviventris, Citellus lateralis, Eutamias vmbrinus, Tamiasciurus hud- sonicus, Microtus pennsylvanicus, and Ochotona princeps. In Figure 1 we have mapped the distribution of boreal habitat and the barrier discussed. We note that the arboreal species, namely, the two tree squirrels, the flying squirrel, the marten, and the fisher, occur either on one side of the barrier only or else have distinct northern and southeastern subspecies. The other species that occur only on one side of the barrier or that have separate sub- species on the north side and on the southeast side of the Wyoming Basin are: Marmota flaviventris, Ochotona princeps, Microtus richardsoni, Microtus pennsyl- vanicus, Citellus lateralis, and two species of Eutamias. The species named im- mediately above and the arboreal species (in comparison to the next group of species to be discussed) seem to be relatively restricted in the range of habitats that they utilize. The pattern in the chipmunks is complicated by other species of less montane chipmunks whose presence may act as a biological barrier. The 526 82 JOURNAL OF MAMMALOGY Vol. S7,No. 1 red squirrel, the marten, and the golden-mantled ground squirrel have popula- tions in Colorado and northern Utah that are alike and differ from corresponding populations on the northern side of the Wyoming Basin. The species that have a single subspecies occurring on both the north side and the southeast side of the Wj^oming Basin are as follows: three species of shrews, three microtines, the jumping mouse, and the porcupine. The porcupine is a ubiquitous creature, prone to wander. The other seven species are small mammals which may migrate by way of narrow avenues found along stream- courses where the water draining from montane areas supports growths of brush, scrub willow, and grasses and sedges. Furthermore, these species do not seem to be dependent upon forests or forest-edge communities. If the montane mammals here dealt with are arranged in order of their decreasing dependence upon mon- tane conditions it is seen that those kinds appearing early in the list are those that occur only on one side of the barrier shown in Figure 1 or those that have distinct northern and southeastern subspecies (that is, belong in Group V). Those that appear later in such a list are in general those that are subspecifically the same north and southeast of the Wyoming Basin (Group IV). It might be concluded that montane meadow and streamside habitats connected the southern and central Rockies across the Wyoming Basin long after they ceased to be con- nected by continuous forests. The red-backed mouse, Clethrionomys, is restricted to the forested areas of the mountains more than the other species. Investigation of the most recent work on Clethrionomys (Cockrum and Fitch, Univ. Kansas Publ., ]\Ius. Nat. Hist., 5: 283, 1952), reveals that these authors regarded north- western Wyoming and the Bighorn Mountains as centers of incipient subspecies. Judging by their comments on C. gapperi galei and on C. g. uintaensis in Utah we feel that the four populations, (1) in the Uinta Mountains, (2) in north- western Wyoming, (3) in the Bighorn Mountains, and (4) in northern Colorado, are dilTerentiated from one another and might be regarded equally well as one or as four subspecies. The latter supposition would place them in Group V and would obviate the seeming inconsistency. On the basis of the information presented above it seems that: (1) The ranges of montane species are correlated with their dependence upon special habitats; the more dependent species are more restricted in range, both locally and regionally. (2) The more dependent, and therefore relatively restricted, species show more differentiation on opposite sides of the Wyoming Basin than the species that are less restricted. (3) The closest affinities of the boreal mammals in the Rocky Mountains of Colorado are with the boreal mammals of the Uinta Mountains across the Green River Canyon rather than with those of the central Rocky Mountains to the north of the Wyoming Basin. (4) The discontinuity in the boreal forest produced by the erosion of the Green River Canyon has become important as a barrier to montane mammals later than the discontinuity caused by the desiccation of the Wyoming Basin. Department of Zoology, University of New Mexico, Albuquerque, and Museum of Natural History, University of Kansas, Lawrence. Received February 6, 1955. 527 Mammals from Isla Cozumel, Mexico, With Description of a New Species of Harvest Mouse BY J. KNOX JONES, JE., AND TIMOTHY E. LAWLOR Isla Cozumel, or Cozumel Island, lies in the Gulf of Mexico ap- proximately 16 kilometers ofiF the east coast of the Yucatan Penin- sula. Administratively, the island is attached to the Mexican Ter- ritory of Quintana Roo. The strait that separates Cozumel from the mainland reaches a depth of more than 300 meters, and the current in the strait is swift. The island itself is approximately 45 kilometers long (northeast-southwest) and averages about 14 kilo- meters wide. "It is composed of limestone and its greatest eleva- tion is about 10 meters above the sea" (Paynter, 1955:8). Vegeta- tionally, Cozumel supports mostly scrubby deciduous forest and mangrove swamps. From August 7 to 11, 1962, a field party from the Museum of Nat- ural History of The University of Kansas collected vertebrate animals in the vicinity of San Miguel on die west coast of Cozumel. The present report concerns the mammals obtained or observed by the party, among which are several species previously unreported from the island. One of these is a new harvest mouse of the genus Reithrodontomijs that is named and described beyond. Mention is made also of species previously reported from Cozumel, especially by Hall and Kelson (1959), Koopman (1959), Merriam (1901), and Thomas (1888). Field operations on Cozumel were supported by funds made available through a contract ( DA-49-193-MD-2215) between the U. S. Aniiy's Medical Research and Development Command and The University of Kansas. Didelphis tnarsupialis cozumelae Merriam, 1901. — Ten specimens (91428- 37), including six pouch young, were taken 3/2 km. N San Miguel, where opossums were seen nightly at a garbage dump. The female that carried the si.\ young was obtained on August 8; the young weighed an a\erage of 18.4 (17^6-19.5) grams. We tentatively retain the subspecific name cozumelae for the insular opos- sums. Comparison of our material with specimens of D. in. i/ticatancnsis from the adjacent mainland fails to support Merriam's (1901:102) contention that the two differ in certain cranial features or that cozumelae is the larger in size of body. The tail does, however, average shorter in relation to length of (411) 528 412 Unu'ersity of Kansas Publs., Mus. Nat. Hist. body than in specimens from the mainland, and the white tip on the tail is noticeably shorter (one-half to two-thirds as long). Selected measurements of an adult male and the largest axailable female (the one with young) are, respectively: total length, 770, 633; length of tail, 319, 300; length of hind foot, 60, 53; length of ear, 54, 52; greatest length of skull, 115.1, 90.7; zygomatic breadth, 63.5, 45.2; palatal length, 65.7, 54.8; length of M1-M4, 19.5, 17.9. Micronycteris megalotin mexicana Miller, 1898. — Our only specimen (91539), a female in juvenal pelage and with unfused phalangeal epiphyses, was taken in a mist net stretched between two palm trees adjacent to the cottage in which we stayed. Goldman (1951:443) earlier listed this species from Cozumel under the name Macroius pijgmaeus. Artibeus jamaicensis yucatanicus J. A. Allen, 1904. — Judging from our experience, this species is the commonest of the bats occurring on Cozumel. Eighteen individuals were collected as follows: 4 km. N San Miguel, 6 (91724-29); 3)^ km. N San Miguel, 12 (91730-40, 91781). All specimens taken were netted, either along small roads through the scrubby forest or among coconut palms adjacent to residences near the beach. Five of 11 females ob- tained were lactating; the testes of one male measured 10 mm. Several authors previously have reported this bat from the island. Artibeus lituratus palmarum Allen and Chapman, 1897. — One specimen (91780), a male having testes measuring 6 mm., was netted along with several individuals of A. jainaicensis among coconut palms SM km. N San Miguel. This species has not been reported previously from Cozumel. Artibeus phaeotis phaeotis (Miller, 1902). — A male and two females of this small fruit-eating bat were trapped in mist nets stretched across a narrow road in the forest 4 km. N (91790) and 3)i km. N (91791-92) San Miguel. Each of the females carried a single embryo ( 23 and 25 mm. crown-rump ) . Although this species long has been known from the Yucatan Peninsula, it was not fomierly known from Cozumel. Previous authors ( Hershkovitz, 1949:449, Dalquest, 1953:64, and Davis, 1958:164, among others) have regarded A. p. phaeotis (type locality, Chichen- Itza, Yucatan) as a subspecies of Artibeus cinereus. Apparently none of the authors who thus treated plmeotis examined the holotype, which actually is identical with the species later described by Andersen (1906:422) as Artibeus turpis (type locality, Teapa, Tabasco). Therefore, A. p. phaeotis replaces A. t. iurpis as the correct name for the bat of the Caribbean lowlands of southern Mexico and adjacent areas that is characterized by its small size, relatively broad and naked uropatagium, and short, up-turned rostrum. The slightK' smaller subspecies of Pacific coastal areas (see Davis, 1958:163) henceforth should bear the name Artibeus phaeotis nanus. We are grateful to Dr. C. O. Handley, Jr., of the U. S. National Museum, who currently is revising the genus Artibeus, for allowing one of us (Jones) to examine the holotype of phaeotis. Our attention first was drawn to this mat- ter when we discovered that all individuals of small Artibeus in our collection from the Yucatan Peninsula resembled "turpis," which was not reported from there, rather than "cinereus," which was said to occur there. Measurements of the male and two females are, respectively: total length, 57, 54, 58; length of hind foot, 12, 10, 11; length of ear, 14, 16, 17; length of 529 Mammals from Isla Cozumel, Mexico 413 forearm, 38.2, 38.3, 40.8; greatest length of skull, 19.6, 19.1, 19.3; zygomatic breadth, 11.5, 11.7, 11.7; length of maxillary tooth-row, 6.0, 5.8, 5.9. Natalus stramineus saturatus Dalquest and Hall, 1949. — This species, previ- ously unreported from the island, is represented in the U. S. National Museum by 32 specimens in alcohol from San Miguel. Oryzomys palustris cozumelae Merriam, 1901. — Rice rats were abundant in tangled, second-growth brush and vines. Thirty-six specimens were col- lected from 3 km. N (92185-86) and 3)i km. N (92168-84, 92187-203) San Miguel. A female obtained on August 8 carried three embryos that measured 15mm. (crown-rump) and our sample contains many two-thirds to three- fourths grown young. Up to now, O. p. cozumelae has been regarded as a distinct species, al- though its close relationship with O. palustris of the adjacent mainland has been recognized (see Goldman, 1918:43). None of the specimens among our material are as large as the holotype of cozumelae, but a number fall within the range of variation cited for adults by Goldman ( loc. cit. ) . When our specimens were compared with individuals of O. p. couesi from the Yucatan Peninsula, we found that cozumelae differed noticeably only in being larger externally; cranially, couesi and cozumelae differ only in minor details (for example, the skull of cozumelae averages slightly larger, is less arched over the orbits, and has heavier teeth and larger nasals), and the latter averages only slightly darker than mainland specimens. Furthermore, adults of co- zumelae do not exceed in external size individuals from several of the named populations of O. palustris. For all these reasons, and because cozumelae long has been recognized as only an insular relative of palustris, we employ the name Oryzomys palustris cozumelae for it. We feel the relationships of the insular population are best reflected by such usage. Reithrodontomys spectabilis new species Holotype. — Adult male, skin and skull, no. 92294 Museum of Natural His- tory, The University of Kansas, from 2/2 km. N San Miguel, Isla Cozumel, Quintana Roo; obtained by Ticul Alvarez on August 8, 1962 (original no. 848). Distribution. — Known only from Cozumel Island. Diagnosis. — Size large both externally and cranially (see measurements); tail long in relation to head and body (134-148 per cent in adults), scantily haired; pelage short and relatively sparse; upper parts brownish ochraceous over-all, brighter ochraceous on sides; underparts grayish white, the individual hairs wlaite terminally and plumbeous basally; pinkish buff pectoral spot some- times present; tail dark brown above, only slightly paler below; braincase relatively flattened and uninflated; zygomatic arches broad and strong; rostrum relatively short and broad; mesopterygoid fossa broad; auditory bullae large but only moderately inflated; incisive foramina rarely reaching level of Ml; teeth huge; first and second molars typical of the subgenus Aporodon in having well developed mesolophs(ids) and mesostyles(ids); third lower molar essentially a smaller replica of first two; baculum long ( 9.5 and 10.9 mm. in two adult males), slender, curved dorsally at the distal end, broadly arrow- shaped basally (width of base 1.1 and 1.2 mm. in the two adult bacula studied), possibly largest among members of genus. The skull and teeth are illustrated in Figure 1. 530 414 University of Kansas Publs., Mus. Nat. Hist. Fig. 1. Skull and teeth of Reithrodontomys spectahilis. Right upper (A) and left lower (B) molars of KU 92293 (x 15), and dorsal (C) and ventral (D) views of skull of holotype, KU 92294 (x3). T. H. Swearingen made the drawings from photographs by J. F. Downhower. Measurements. — E.\ternal and cranial measurements of the holotype followed by average and (in parentheses) extreme measurements of eight specimens (four males and four females, including the tvpe) are: total length, 216, 213.8 (205-221); length of tail, 124, 125.7 (121-132); length of hind foot, 22, 21.3 (20-22); weight (in grams), 20.6, 20.2 (18.1-21.4); greatest length of .skull, 24.7, 25.2 (24.6-26.2); zygomatic breadth, 12.2, 12.3 (11.8-12.7); interorbital breadth, 3.7, 3.7 (3.5-3.9); breadth of braincase, 11.2, 11.2 (11.0-11.5); depth of skull, 8.5, 9.0 (8.5-9.4); length of rostrum, 8.8, 9.0 (8.7-9.8); breadth of rostnnn, 4.1, 4.2 (3.9-4.5); length of incisive foramen, 4.4, 4.5 (4.4-4.8); breadth of mesoptervgoid fossa, 1.7, 1.7 (1.5-1.8); length of palatal bridge, 3.9, 4.0 (3.8-4.3); alveolar length of maxillary tooth-row, 3.8, 3.8 (3.7-3.9); alveolar length of mandibular tooth-row, 3.4, 3.5 (3.4-3.7). 531 Mammals from Isla Cozumel, Mexico 415 Ccmipamans. — The new species is a member of the subgenus Aporodon and is allied to Reithrodontotnys mexicarms and R. gracilis of the R. mexicanus species group. It is the largest member of the mexicanus group as defined by Hooper ( 1952 ) and among the largest species of the genus Reithwdontomys. Of the two kinds to which it appears most closely related, the new species resembles R. gracilis of the adjacent mainland of the Yucatan Peninsula to a somewhat greater degree than R. mexicanus, known nearest Cozumel Island from the highlands of Guatemala and Honduras. In comparison with gracilis, R. spectabilis is immediately recognized by its much larger size (total length averaging 213.8 in adult spectabilis but only 175.7 in six adult R. g. gracilis from the Yucatan Peninsula, length of hind foot 21.3 and 17.8, greatest length of skull 24.7 and 21.9, zygomatic breadth 12.2 and 10.8), generally darker coloration, and in having a massive skull with broader, heavier zygomatic arches. R. spectabilis resembles R. gracilis (in contrast to R. mexicanus) in that the dark tarsal stripe does not extend onto the hind foot and in having a flattened and relatively uninflated braincase, incisive foramina that rarely reach the level of Ml, and in other general features of the cranium. The breadth and depth of the braincase are even less, relative to length of the skull, than in gracilis — the breadth averages 44.6 per cent of the greatest lengdi of skull (47.5 in gracilis studied), and the depth of skull averages 35.9 in rela- tion to length (36.9 in gracilis). R. spectabilis resembles R. mexicanus more than R. gracilis in size (measure- ments of mexicanus studied — subspecies hmcelU and orintis — are intermediate between those of spectabilis and gracilis) and to some extent in general colora- tion. Cranially, aside from averaging smaller, mexicanus can be distinguished most easily from spectabilis by its proportionately broader and deeper braincase. Because of its resemblance in many features to the smaller R. gracilis, we assume that the precursors of R. spectabilis reached Cozumel from the adjacent mainland of the Yucatan Peninsula. The magnitude of the differences be- tween the two species suggests, to us at least, that they have been separated for a relatively long time, since at least late Pleistocene. Remarks. — Some of the harvest mice from Cozumel Island were trapped in tangled, second-growth vines and brush adjacent to (beachward from) scrub forest; Oryzomys palti.^ris coziimelae was abundant in this same habitat. Other individuals were taken in forest in traps set at the bases of trees and along a stone wall. One specimen was caught by hand at night as it climbed in the branches of a small tree, indicating that the Cozumel harvest mouse is at least partly arboreal in habits. Our sample contains several juvenal and subadult specimens. One female, trapped on August 9, had been recently lactating, but no other females evidenced gross reproductive activity. The testes of two adult males measured 13 and 14 mm. Specimens examined, 16, as follows: 2^2 km. N San Miguel, 3 (92294-96): 33^ km. N San Miguel, 13 (92281-93). Peromyscus Icucopus cozumelae Merriam, 1901. — Six white-footed mice were trapped along trails in scrub forest or in places marginal between forest 532 416 University of Kansas Publs., Mus. Nat. Hist. and second-growth bnisli. Our specimens are from 3/2 km. N (92417-21) and 2)2 km. N (92422) San Miguel. A female obtained on August 11 was lactating. P. I. cozumehie differs from the subspecies of the mainland of the Yucatan Peninsula (P. /. castcincus) in being larger, both externally and cranially, and in having heavier teeth. The two kinds closely resemble each other in color. Dasyprocta punctata yucatanica Goldman, 1913. — According to Merriam (1901:100), D. punctata was introduced on Cozumel "shortly before" the visit of Nelson and Goldman to the island in 1901. Goldman actually observed an individual in the forest near San Miguel. Natives reported to us that agoutis still occur on the island. Agouti paca subsp. — On the morning of August 11, William E. Duellman observed a paca along a trail through the forest appro.ximately 4 km. N San Miguel. We queried local residents concerning the status o*^ this large rodent on the island and, while aware of its presence, they had no knowledge of whether or not it had been introduced. Urocynn cinereoargenteus subsp. — We did not obtain specimens of the gray fox, but local residents reported its occurrence to us. Earlier, Merriam (1901:99) wrote of this species on Cozumel as follows: "The only mammal heard of [by Nelson and Goldman] which was not secured is a small Gray Fox ( Urocyon ) reported by natives as rather rare, but more common on the eastern and southern parts of the island. From the accounts it agrees with the Rac- coon, Nasua, and Peccary in being much smaller than the mainland species." Procyon pygmaeus Merriam, 1901. — A subadult male raccoon (92565) was shot on August 8 from a coconut palm situated along the beach 3/2 km. N San Miguel. Two other individuals were seen in the same tree and the three may have been from the same family group. Our specimen diflFers in cranial features from raccoons of the Yucatan Peninsula ( P. lotor shufeldti ) in ways described by Goldman (1950:76-77), and we follow Goldman in preserving specific recognition for pygmaeus. It is perhaps worthy of note that our speci- men has a distinct "interparietal" bone approximately 13 mm. long by 8 mm. wide, at the juncture of the parietal and frontal bones. Nasua nelsoni Merriam, 1901. — According to local residents, coatis are com- mon in the vicinity of San Miguel. Several were seen at night and in early morning by our party. One (92570), an adult female with well-developed teats (probably recently lactating), was obtained 3/2 km. N San Miguel. We retain N. nelsoni as a full species because it differs so strikingly in size from the coati (Nasua narica yucatanica) of the adjacent mainland. Measure- ments of our specimen, followed in parentheses by measurements of an adult female of yucatanica from 7 km. N and 51 km. E Escarcega, Campeche, are as follows: total length, 741 (990); length of tail, 332 (485); length of hind foot, 76 (99); length of ear, 35 (40); greatest length of skull, 103.4 (118.6); zygomatic breadth, 50.3 (58.3); interorbital constriction, 20.4 (24.5); palatal length, 62.7 (72.9); breadth of braincase, 38.3 (42.0); alveolar length Ml- M3, 16.6 (19.9). Aside from its over-all smaller size, the skull of nehoni is notable for its more delicate construction and distinctly smaller bullae when compared with N. n. yucatanica. Trichechus manatus manatus Linnaeus, 1758. — Local residents reported that manatees were observed occasionally along the west coast of the island and that they were common in the bays and lagoons on the adjacent coast of Quintana Roo. 533 Mammals from Isla Cozumel, Mexico 417 Tayassu tajacu nanus Merriam, 1901. — The collared peccary of Cozumel was described as a subspecies distinct from that on the adjacent mainland of Yucatan ( T. t. angulatus) on the basis of smaller size and blacker nose and chin. Subsequenth-, Hershkovitz (1951:567) noted that the species had been introduced on the island from the adjacent mainland (see also de Vos et al, 1956:176) and suggested that the small size claimed for nanus resulted from heavy hunting pressure, which did not allow animals to attain adult size. Ad- ditional specimens are needed before Hershkovitz's interesting hypothesis can be tested. Natives on Cozumel reported the collared peccary as common and intensively hunted. Mammals Reportedly Collected on Cozumel by G. F. Gaumer George F. Gaumer, well-known naturalist who lived for many years on the Yucatan Peninsula, reported himself, or sent to others, a number of mammals alleged to have come from the island of Cozumel. Some of these probabK originated from the island but others seemingly did not, as discussed below. Oldfield Thomas (1888), in the first technical paper dealing with mam- mals from Cozumel, reported five species that were collected by Gaumer and communicated to Thomas by Salvin and Godman. These five were Didelphis marsupialis, Pteronotus parnellii, Artiheus jamaicensis, Tadarida laticaudaia, and "Nasua nasica." The opossum, Jamaican fruit-eating bat, and coati (in the fonn of the small Nasua nelsoni) subsequently have been found to be common on the island. No other records of the two remaining bats, Pteronotus and Tadarida, have been forthcoming, but each is widespread on the adjacent Yucatan Peninsula and we do not doubt that each occurs on Cozumel. In his "Monografia de los mamiferos de Yucatan," Gamner (1917:117) re- ported the presence of the Yucatan deer mouse, Peromyscus yucatanicus, on the island. He did not, however, record Peromyscus leucopus from Cozumel and Koopman (1959:237) concluded that Gaumer confused the two species. We are inclined to agree with Koopman, because insofar as we know P. yucatanicus does not occur on the island. Gaumer (op. cit. :6S) also recorded the white-lipped peccary (as "Dicotyles labiatus") from Cozumel but this, too, seems to be in error. By far the most perplexing collection of mammals relating to Cozumel is a lot of specimens acquired from Gaumer by the Museum of Natural History in the early 1900's. Many of these specimens were cited by Hall and Kelson (1959) and all species represented were listed by Koopman (1959). The holotype and paratypes of Mimon cozumehe Goldman (1914) were among the specimens in this collection. Aside from M. cozumelae, species represented (and their catalogue numbers) are: Micronycteris megalotis mexicana (1659- 60); Glossophaga soricina leachii (1655-58); Artiheus jamaicensis yucatanicus (1641-42); Centurio senex (1669-70); Lasiurus horealis teliotis (1655); Pleco- tus (possibly mexicanus) (1658); Molossus ater nigricans (1663-64); Ateles geoffroyi yucatai^ensis ( 885 ) ; Tamandua tetradactyla mexicana ( 869-872, 880 ) ; Heterogeomys hispidus yucatanensis (1370); Nasua narica yucatanica (876-77, 1599-1600); Galictis allamandi canaster (873); and Tayassu sp. (875 — said to be T. t. nanus, probably solely on supposed geographic grounds, but speci- men not now to be found ) . Most of the labels that now accompany the above-listed specimens are not the original labels of Gaumer, and, in any event, bear no additional informa- tion that could support or refute the contention that the specimens actualK 534 418 University of Kansas Publs., Mus. Nat. Hist. came from Cozumcl. Many other specimens received at the same time arc labeled simply as from "Yucatan." Some of the specimens said to be from Cozumel obviously did not come from there as shown below and there is a strong possibility that others did not. Perhaps few or none of the specimens actually originated on the island. Of the eight bats, only two (A. iamaicensis and M. megalotis) have been obtained on Cozumel by other collectors. One, Plecotus, seems likely not to occur there (see also Koopman, 1959:237). The remaining five (Mimon, Glossophaga, Ccnturio, Lasiurus, and Molossus) are widespread in Middle America and each is known from the adjacent mainland. We took G. soricinn on the much smaller Isla Mujeres that lies to the north of Cozumel. There is a strong possibility that all five species actually occur on Cozumel and that additional collecting will establish their presence on the island. We are especially hopeful of this development with reference to Mimon cozumelae. With reference to the six strictly terrestrial kinds, we doubt that any save Tayassii occurs on the island. The major habitat, sciub forest, doubtfully would support tamanduas or monkeys (although the latter might have been intro- duced) and neither was reported to us by local residents (nor was the grison mentioned). We especially questioned natives about the occurrence of pocket gophers but they assured us that "tuzas" were not present. All four coatis, formerly labeled "Nasiia r^elsoni" are unquestionably the much larger N. narica yucatonica that occurs only on the adjacent mainland. Zoogeog,raphy As Koopman (1959) pointed out, the zoogeographic relationships of Cozumel, at least with respect to mammals, are undoubtedly with the adjacent mainland, not with the Antilles. Even though the strait that separates Cozumel from the mainland of Quintana Roo is deep, and the current in it strong, we suppose that most of the mammals that occur on the island reached it by "rafting" across the strait or possibly from the mainland to the south of Cozumel. At a time of ma.ximal glaciation, because of a lowering of the sea level, the strait would have been reduced to appro.ximately half its present width, theoretically making "rafting" much easier than now, es- pecially if favorable winds prevailed. At least three of the mammals that currently inhabit the island are known or suspected to have been introduced by man. 535 Mammals from Isla Cozumel, Mexico 419 LITERATURE CITED Andersen, K. 1906. Brief diagnoses of a new genus and ten new forms of stenodermatous bats. Ann. Mag. Nat. Hist., 18:419-423, December. Dalquest, W. W. 1953. Mexican bats of the genus Artibeus. Proc. Biol. Soc. Washington, 66:61-66, August 10. Davis, W. B. 1958. Review of Mexican bats of the Artibeus "cinereus" complex. Proc. Biol. Soc. Washington, 71:163-166, 1 fig., December 31. de Vos, a., Manville, R. H., and Van Gelder, R. G. 1956. Introduced mammals and their influence on native biota. Zoologica, 41:163-194, 1 fig., December 31. Gaumeb, G. F. 1917. Monografia de los mamiferos de Yucatan. Dept. de Talleres Graficos de la Secretaria de Fomento, Mexico, xii -f 331 pp., 57 pis., 2 photographs, 1 map. Goldman, E. A. 1914. A new bat of the genus Mimon from Mexico. Proc. Biol. Soc. Washington, 27:75-76, May 11. 1918. The rice rats of North America (genus Oryzomys). N. Amer. Fauna, 43:1-100, 6 pis., 11 figs., September 23. 1950. Raccoons of North and Middle America. N. Amer. Fauna, 60: vi -f- 1-153, 22 pis., 2 figs., November 7. 1951. Biological investigations in Mexico. Smiths. Misc. Coll., 115: xiii + 1-476, frontispiece, 71 pis., 1 map, July 31. Hall, E. R., and Kelson, K. R. 1959. The mammals of North America. 2 vols, (xxx-f 1-546 -j- 79 and viii + 547-1083 -f 79), illustrated, March 31. Hershkovitz, p. 1949. Mammals of northern Columbia. Preliminary report no. 5: bats (Chiroptera). Proc. U. S. Nat. Mus., 99:429-454, May 10. 1951. Mammals from British Honduras, Mexico, Jamaica and Haiti. Fieldiana-Zool., Chicago Mus. Nat. Hist., 31:547-569, July 10. Hooper, E. T. 1952. A systematic review of the harvest mice (genus Reithrodontomys) of Latin America. Misc. Publ. Mus. ZooL, Univ. Michigan, 77: 1-255, 24 figs., 12 maps, January 16. Koopman, K. F. 1959. The zoogeographical hmits of the West Indies. Jour. Mamm. 40: 236-240, May 21. Merriam, C. H. 1901. Six new mammals from Cozumel Island, Yucatan. Proc. Biol. Soc. Washington, 14:99-104, July 19, Paynter, R. a., Jr. 1955. The omithogeography of the Yucatan Peninsula. Bull. Peabody Mus. Nat. Hist., 9:1-347, 4 pis., 2 maps. Thomas, O. 1888. List of mammals obtained by Mr. G. F. Gaumer on Cozimiel and Ruatan Islands, Gulf of Honduras. Proc. Zool. Soc. London d 129, June. ^ Transmitted July 7, 1964. 536 RELATION OF SIZE OF POCKET GOPHERS TO SOIL AND ALTITUDE By William B. Davis While working out the distribution and taxonomy of pocket gophers in southern Idaho, I became interested in the problem of correlation of size with soil conditions and altitude, I do not claim to have settled a problem; in fact, I intend merely to point out one that, to me, warrants further study. The specimens and field notes on which this discussion is based are con- tained in the Museum of Vertebrate Zoology, Berkeley, California. I wish to thank the officials of that museum for the many privileges extended me; also, to acknowledge the generous assistance of Messrs. David Johnson and D. Tillotson in supplying additional specimens and information. Pocket gophers, especially the males, tend to increase in size with age. The ultimate size attained appears to be correlated directly with the type of soil inhabited, and indirectly with altitude. At high elevations where the soil usually is shallow and rocky, or at lower elevations where the same general environmental conditions prevail, races and individuals of the same species tend to be small. In places of this kind the skulls of males and fe- males often are indistinguishable; they are smooth and lack ridges. They are juvenile in character and in many respects appear not to have developed beyond the subadult state of forms living under better enviroimiental con- ditions. If one compare individuals from poorer soils (for pocket gophers) with others from progressively better and deeper ones, the size of both sexes is found generally to increase, males more so than females. Under optimum conditions the actual weight of the skulls may average two or even three times that of individuals that live amid adverse conditions. In the deeper soils sexual dimorphism is evident and the skulls of both sexes are angular in outline and have well developed sagittal, lambdoidal, and temporal ridges. This general reduction in size at higher altitude is illustrated by specimens from near Pocatello, Idaho. There the species Thomomys quadratus occurs altitudinally from 4400 feet on the floor of Portneuf Valley to over 7000 feet in the Bannock Mountains. Individuals taken from the valley are considerably larger than those from higher altitudes. The skulls are massive, prognathous, angular, and ridged, the males much larger than the females. Individuals taken at progressively higher altitudes are smaller, the skulls tend to be less angular and ridged, as well as less prognathous, and the degree of sexual dimorphism is reduced. Taking the product of three dimensions of the skull, basilar length, zygomatic breadth, and palatofrontal depth, as an index of size, the following results were obtained : 537 DAVIS — POCKET GOPHERS 339 Males Females Difference 4500 feet 1295 (3)* 1035 (3) 260 5000 feet 1360 (4) 930 (3) 430 5800 feet 1060 (2) 880 (2) 180 6300 feet 920 (6) 840 (6) 80 7000 feet 920 (2) 850 (1) 70 * Number of specimens averaged. Certain discrepancies are evident in the comparisons. Males from 5000 feet are larger than those from 4500 feet; the female from 7000 feet is larger than those from 6300 feet. These exceptions do not invalidate the general principle because age differences in the males and too few specimens in the females probably account for the deviations from the expected size. A reduction in length of body, length of hind foot, and length of tail accom- panies a reduction in "volume" of the skull. Similar results were obtained in a study of another species in another locality. Thomomys bottae occurs on the floor of Monitor Valley, Nevada, at an altitude of 6900 feet, and also on the adjacent Toquima Mountains, which rise to over 10,000 feet. Individuals taken at altitudes ranging from 9000 to 10,000 feet on the mountain are considerably smaller than those from the valley; specimens from Meadow Creek Canyon, at 8000 feet on the east side of the mountain, are intermediate in size. Sexual dimorphism is pronounced in specimens from the valley; it is slight in specimens from above 9000 feet. These facts are evident from the following tabulation. Males Females Difference 6900 feet 1775 (6) 1250 (6) 525 8000 feet 1320 (2) 1060 (2) 260 9000 to 10000 feet 1060 (5) 920 (2) 140 Similarly, specimens of Thomomys bottae taken at different elevations on Mt. Moriah in eastern Nevada and western Utah exhibit the same trend. Skulls of specimens from 5000 feet elevation are much larger than those from high on the mountain. Between the two extremes the skulls are intermedi- ate, grading, as evidenced below, from large, with pronounced sexual di- morphism, at the bottom, to small at the top, with little difference between males and females. Males 5000 feet 1664 (3) 5400 feet 1268 (2) 6000 feet 1177 (2) 6700 feet — 9100 feet 1139 (3) 9800 feet 1086 (1) 11400 feet — Females Difference 1221 (5) 443 1085 (3) 183 1001 (2) — 994 (6) 45 1009 (1) 77 971 (2) — 538 340 JOURNAL OF MAMMALOGY Again, certain discrepancies appear, but the gradation is clearly evident. Aside from consistently smaller size, specimens of both quadratus and hoitae from the higher altitudes are but little different from those at low levels. The most significant difference is found in the relatively shorter rostrum in specimens from high altitudes. This I interpret as an expression of arrested development. Numerous studies have shown that the rostrum in subadult pocket gophers is consistently relatively, as well as actually, shorter than in adults. Gradation from large to small size is found in every area studied in the Great Basin where one species occupies both the valley floor and the adjacent mountain. Because of this, it becomes increasingly difficult to reconcile present practices in taxonomy with the situation as it actually exists. The tendency of certain students to assign all the populations of pocket gophers occurring on different, isolated mountain tops to one subspecies and those occurring in the lowlands to another defeats the purpose of systematics. To me, it is illogical to assume that the several alpine populations are closely related inter se and genetically different, as a unit, from the populations of the same species occurring in the valleys. Nor does it seem logical to assume that the populations of a species in two valleys, separated by a high mountain range on which small individuals of the same species occur, are genetically related inter se and at the same time genetically distinct from the smaller alpine individuals. Various authors have referred the alpine populations of pocket gophers in southern Idaho to Thomomys uinta or Thomomys quadratus uinta and those in the lowlands to Thomomys bridgeri or Thomomys quadratus bridgeri. In doing this the topography and geologic history of the area probably were not considered. In this region the mountains certainly are older than the genus Thomomys and consequently we cannot assume that populations now restricted to alpine areas once occupied a continuous range that subsequently was disrupted by geologic changes. Nor can we assume that the population of large pocket gophers moved in and usurped the lowland portions of a range once occupied by the smaller animals. Such an assumption would neces- sitate a divergence of the two at a time earlier than that suggested by cranial characters. Furthermore, it would necessitate a migration of the large in- dividuals over high passes in order to explain their present distribution in southern Idaho. In interpreting the past history of Thomomys quadratus in southeastern Idaho I have assumed that during the Pleistocene most of the mountains were glaciated and the lower valleys under water, so that intermediate alti- tudes alone were available to pocket gophers. It is well known that the yearly increase of a successful species is greater than the carrying capacity of the area occupied, and that as a result of population pressure every avail- able niche is sought out and occupied. As the glaciers receded and the lakes decreased in extent, additional territory became available both above and 539 DAVIS — POCKET GOPHERS 341 below the former range. This new territory was occupied by the surplus of the yearly increase. Those gophers that moved down hill encountered deeper and richer soils and consequently could grow larger without handicap. Those that moved uphill found conditions progressively more adverse; the soil was shallower and rockier and plant food less abundant, although suffi- cient to maintain life. Only individuals with small bodies could survive amid such conditions. Whether size per se be heritable or due to ontogenetic processes is not of prime importance to the question at issue because the end result — size — in either instance is the limiting factor in shallow soil. Ul- timately, from a population of pocket gophers of medium size, both the mountains and the lowlands became populated, the former by small indi- viduals, the latter by large ones. The parent stock probably remained at intermediate levels. Such an interpretation explains many facts of distribution. Each high mountain in southern Idaho harbors a population of dwarfed individuals, in each valley the gophers are large, and at middle altitudes, of intermediate size. The alpine populations, even on isolated mountain tops, resemble one another more than they do those in the valleys, yet it is difficult to conceive of them as being subspecifically related inter se and at the same time sub- specifically distinct from those at lower elevations. The similar cranial characters of the two extremes, the gradation from large size to small, and the topography and geologic history of the region, point toward the conclu- sion that the alpine and lowland forms are of the same subspecies. This applies also to the two localities in Nevada whence specimens were examined. The only other logical method I see of treating these alpine populations is to name each one of them, a practice that would lead to confusion and ridi- cule, yet one that is followed by some students. We should be more cautious in our designation of new races. All too little attention has been given to the study of the effect of environment upon the size of animals. If environmental conditions produce a condition of arrested development, should animals exhibiting this condition be recognized by name? Whether pocket gophers from alpine areas would increase in size if transplanted to better soil at lower elevations, or whether they would retain their identity, is not known. An experiment designed to test this point would be well worth while. SUMMARY 1. In each of three localities in southern Idaho and the Great Basin where the same species of pocket gopher inhabits both the lowlands and the adjacent mountain, a gradation from large, sexually dimorphic individuals, at lower elevations where the soil is deeper, to small individuals with little, if any, sexual dimorphism, at higher elevations where the soil is shallower, is exhibited. 2. Since the same trend occurs at each of three widely separated locaUties, 540 342 JOURNAL OF MAMMALOGY it seems logical to conclude that the gradient is a true one ; also that the popu- lations at the two extremes of such a gradient are not worthy of recognition by separate names. 3. The possibiUty of this gradient occurring in other locaUties should merit careful consideration in systematic work with Thomomys so that the formal naming of populations not worthy of recognition by name will be avoided. Department of Wild Game, Agricultural and Mechanical College of Texas, College Station, Texas. 541 THE PINNIPEDIA: AN ESSAY IN ZOOGEOGRAPHY* J. L. DAVIES STUDIES that seek to explain the present distribution of animal forms must draw on a wide and varied field of evidence. Some of the evidence, notably that provided by the paleontologists, is direct and generally capable of reliable interpretation; but rarely, except for some groups of land mammals, is the fossil evidence sufficient. More commonly, paleontology can provide only the framework, perhaps one or two major clues, or even just a tiny piece of the whole picture. The most successful ventures into the field of historical zoogeography have been made by paleomammalogists who have studied particularly the groups living in relatively large numbers in habitats where sedimentation is most rapid. Thus the historical geography of such plains-dwelling groups as the horses and the elephants is comparatively well known, but forms that inhabit mountains, forests, or seas are represented only fragmentarily in the fossil record. For these latter groups other evidence must be invoked, and it is provided by the taxonomy, physiology, ecology, and distribution of the existing forms. This evidence, although easier to accumulate, is not capable of such reliable interpretation as that provided by paleontology, but it must always be explained and often provides the only clues available. A third, more nebulous, category of evidence derives from our knowledge of past climates and past distributions of land and sea. At present, however, this knowledge is so uncertain that it is clearly dangerous to place too much reliance on it; at the same time, it should not be disregarded, since it can often indicate possibilities and probabilities and provide a set of limits within which the correct solution to any given problem may be found. Finally, it is necessary to keep in mind present trends of thought in the fields of genetics and evolution, because, although there is anything but unanimity of opinion within these fields, the most generally accepted concepts will have an important bearing on possible zoogeographical conclusions. Individual bits of evidence culled from all four sources — paleontological, neozoological, paleogeographical, and evolutionary — may often appear *The writer is indebted to Professor Peter Scott, head of the Department of Geography at the Uni- versity of Tasmania, for reading and commenting on the manuscript. > Mr. Davies is Senior Lecturer in the Department of Geography, University of Tasmania, Hobart. 542 THE PINNIPEDIA 475 tenuous and circumstantial by themselves, but if they are placed together to make a unified picture, each one supporting others, the probability of their correctness, and the correctness of the whole picture, is much increased. The present essay examines an order of marine carnivorous mammals, the Pinnipedia, which is divided by taxonomists into three families: the seals (Phocidae), the sea lions (Otariidae), and the walruses (Odobenidac).' The order is comparatively poorly represented in the fossil record, and paleon- tologists have hesitated to discuss its distributional history. Published dis- cussions of pinniped geography have been by neozoologists, who based their findings on the taxonomy and distribution of existing forms, and such a hmitation of the field of evidence has led to some conclusions that are doubt- ful and even demonstrably wrong. For instance. Von Boetticher" followed Sclater^ in postulating a southern origin for the sea lions and a spread from south to north, though the fossil evidence alone makes this highly unlikely. The present study attempts to use evidence from as many fields as possible, and, although it lacks the firm basis of paleontological data that would be desirable, it does seem to provide the only theory of pinniped origin and spread that is fully tenable in the light of the facts as they are known at present. The evidence of existing distribution is summarized on a series of maps that, together with a series published previously,'^ represent an attempt to map the distribution o£ all pinniped forms. ^ The fossil evidence was gathered together in 1 922 by Kellogg,^ but his work is now out of date in some respects and must be supplemented from a rather scattered literature. ' The fossil Semantor was placed in a fourth family by G. G. Simpson: The Principles of Classification and a Classification of Mammals, Bull. Amer. Museum of \at. Hist., Vol. 85, 1945, pp. 1-350, but this animal was almost certainly a mustelid (see E. Thenius: Uber die systematische und phylogenetische Stellung der Genera I'roineles und Semantor, Sitziingsber. Oster. Akad. der IViss. in IVien, Vol. 158, 1949, PP- 323-336). ^ H. von Boetticher: Die geographische Vcrbreitung der Robben, Zeitschr. fur SHugetierkunde , Vol. 9. 1934. pp. 359-368. ■'P. L. Sclater: On the Distribution of Marine Mammals, Proc. Zool. Soc. of London, 1897, pp. 349-359- •*J. L. Davies: Pleistocene Geography and the Distribution of Northern Pinnipeds, Ecology, Vol. 39. '•95«, PP- 97-113- ■' The scientific nomenclature used here corresponds with that of Simpson {op. cit. [see footnote 1 above]) except that Sivertsen (Erling Sivertsen: A Survey of the Eared Seals (Family Otariidae) with Remarks on the Antarctic Seals CoUerted by M/K "Norvegia" in 1928-1929, Scientific Results of the S'orwegian .Antarctic Expeditions \0.36, Det Norske Videnskaps-Akademi i Oslo, 1954) has been followed in separating the Australian sea lion as Xeophoca and including in that genus Hooker's sea lion, formerly known as Phocarctos, which Simpson makes a synonym o( Otaria '' Remington Kellogg: Pinnipeds from Miocene and Pleistocene Deposits of California, Univ. of California Pubis., Bull. Dept. ofCeol. Sciences, Vol. 13. No. 4, 1922. 543 476 THE GEOGRAPHICAL REVIEW A Working Hypothesis The general argument that will be advanced depends largely on one work- ing hypothesis: The pinnipeds are, and always have been, generally tied to a cold-water environment. There can be little disagreement with the statement that present-day pinnipeds are cold-water animals and, with a few exceptions, are found where sea temperature does not exceed 20° C. at any time of the year (Fig. 1). The over 27.5° C. Fig. 1 — Pinnipeds and sea temperatures. Distribution of pinnipeds is shown in black; shaded areas indicate warm-month mean sea-surface isotherms. The exceptional genus Moiiachiis has been omitted from this map for the salce of clarity, but see Figure 8. greatest concentrations both of species and of numbers are found in the sub- arctic North Atlantic and North Pacific and around the fringe of Antarctica. The one major oceanic region from which they are entirely absent is the Indo-Malayan-West Pacific, which is also the only warm-water region with a continuous history as such since the beginning of the Tertiary. In addition, all fossil forms have been recovered from regions where sea temperatures when the deposits were formed lay within prevailing existing tolerances. The distribution of pinnipeds is, and possibly always has been, generally com- plementary to that of the reef-forming corals, and there is no evidence that they have ever inhabited tropical waters. That pinnipeds are physiologically adapted to life in a cold-water environ- ment needs little elaboration. In particular, they have progressed far in the development of insulating hair and blubber and of highly efficient circulatory systems. The upper limit of exterior temperature at which internal tempera- 544 THE PINNIPEDIA 477 ture regulation becomes difficult is presumably not high, and most, if not all, species experience obvious difficulty in regulating body temperatures while on land.^ Physiological adaptation provides evidence that pinnipeds have long fre- quented cold waters, but it does not provide a reason. The reasons are proba- bly many and include the abundance of available food in the colder high- latitude waters and the virtual absence of competitors. The large carnivorous reptiles and fishes have been and are warm- water forms: it has remained for the homoiothermal seals and whales to exploit the cold-water environment. Origin of the Pinnipedia Pinniped ancestry has long been debated. Separate creodont origin, as suggested by Wortman,^ now seems most unlikely, and the general consensus would probably be that the pinnipeds are derived from the ancestral dog-bear stock.^ Recent serological work^° has indicated a close relationship with the bears, a relationship forecast on morphological grounds by Weber" among others. The present study is concerned only with the time and place of pinniped origin, and in order to arrive at reasonable estimates it is not neces- sary to discuss these arguments at length. The latest possible time of origin is determined by the earliest fossil pinnipeds, which are sea Hons from the lower Miocene of California, and, as these are fairly advanced and diversified forms, the latest probable date must surely be sometime about the middle Oligocene. The earliest possible time of origin is determined by the earliest occurrence of the ancestral group. If the early dog-bear stock is accepted as ancestral, then this date is upper Eocene; if it is necessary to look to the miacid creodonts as ancestors, the date is pushed back into the lower Eocene, and even possibly into the upper Paleocene. Derivation from the "inadaptive" creodonts would push it still farther back in the Paleocene, but this likelihood does not seem to be en- visaged today and need not be considered here. The possible extremes are thus upper Paleocene to lower Miocene, and the probable extremes middle ^ See, for instance, G. A. Bartholomew and F. Wilke: Body Temperature in the Northern Fur Seal, Callorhitms ursinus, Journ. of Mammalogy , Vol. 37, 1956, pp. 327-337. *J. L. Wortman: Osteology o( Patriofelis , a Middle Eocene Creodont, Bull. Amer. Museum of Nat. Hist., Vol. 6, 1894, pp. 129-164. 9 See, for instance, W. D. Matthew: The Carnivora and Insectivora of the Bridger Basin, Middle Eocene, Memoirs Amer. Museum of Nat. Hist., Vol. 9, 1909, pp. 289-567, references on pp. 413-417; A. S. Romer: Vertebrate Paleontology (Chicago, 1933); and Simpson, op. cit. [see footnote 1 above]. '° C. A. Leone and A. L. Wiens: Comparative Serology of Carnivores, JoMr«. of Mammalogy, Vol. 37, 1956, pp. 11-23. " M. C. W. Weber: Die S^ugetiere (Jena, 1909). 545 478 THE GEOGRAPHICAL REVIEW Eocene to middle Oligocene. A late Eocene or early Oligocene date is sug- gested by Simpson/^ and because of the highly specialized nature of the lower Miocene pinnipeds, Kellogg'-' concluded that the pinniped stock originated no later than the Eocene. Most considerations point, therefore, to the upper Eocene as the likely time of origin. The place of origin is limited by the distribution of the ancestral group, and whether the ancestral group is canoid, miacid, or creodont, an area some- where in the Holarctic is inevitable. The Carnivora did not reach Africa until the Oligocene and South America until the Pliocene, and only the dingo has arrived in Australia. To judge by this evidence and by the success of the relatively ancient penguins, the Carnivora have never reached Antarctica. But, although an origin somewhere within the Holarctic seems certain on paleontological grounds, further limitation must be attempted by reference to other considerations. Matthew'"^ suggested the Arctic Basin as the most likely place, and the arguments in its favor appear overwhelming. It is centrally placed to provide the fossil and living sea lions of the North Pacific and the fossil and living seals and walruses of the North Atlantic. Its shores have an abundance of shallow waters and wide, long estuaries that would provide the right conditions for a first venture into the marine environment. The Pacific shores are steep and surf-battered, with mountain folds parallel to the coast. This difference between Arctic and Pacific coasts must have con- tinued throughout Cenozoic time and stems from fundamental structural and tectonic differences. The third major argument in favor of the Arctic Basin as the birthplace of the Pinnipedia hinges on the postulate that they have always been cold- water animals. If the group originated as a cold-water form, it can have originated only in the Arctic, which was the only northern marine region where sea temperatures in the Eocene were comparable with temperatures tolerated by pinnipeds today and by those fossil forms which have so far been discovered. The most satisfactory conclusion, therefore, is that the pinnipeds originated from an ancestral dog-bear stock in the middle or late Eocene Arctic Basin. From the Arctic there was a spread southward during succeeding periods, which was influenced in rate and extent by certain geo- morphic and climatic barriers and avenues. " Op. cit. [see footnote i above]. '■5 Op. cit. [see footnote 6 above]. '^ W. D. Matthew: Climate and Evolution (2nd edit.), Special Pubis. New York Acad, of Set., Vol. 1, 1939- 546 the pinnipedia 479 Geomorphic Barriers and Avenues Much has been written on problems connected with land mammals and past distributions of land and sea. In a consideration of the pinnipeds the same problems must be faced, but from the opposite direction. Here the primary concern is with the existence or nonexistence of seaways between the Arctic and the North Pacific, between the Arctic and the North Atlantic, and be- tween the Caribbean and the East Pacific. Studies in the geography of land animals have been concerned with the existence of land bridges across these seaways and obviously have an important bearing on the question. There is general agreement that the two Americas were separated almost continuously from the middle Paleocene to the late Pliocene, and during this time there must have been virtually no obstacle to migration between the Caribbean and the East Pacific. Simpson'^ has examined in detail paleomammalogical evidence for the existence of a land bridge between North America and Asia and between North America and Europe during the Cenozoic. A transatlantic land bridge may have existed in the Eocene, and possibly even in the Oligocene, but Simpson could find no supporting evidence. The problem remains open, but it seems unlikely that there was any continuous barrier between the Arctic Sea and the North Atlantic at these times. Considerable marine deposition took place throughout Europe during the early Tertiary, and there must have been seas separating Western Europe from the main Eurasian land mass and joining the northern seas with the Tethyan Mediterranean. Ekman^^ mentions a probable connection (Obik) betv^een the Tethys Sea and the Arctic Sea in the Eocene. Regarding the North American-Asian bridge across present-day Bering Strait, Simpson reaches fairly firm conclusions from the degree of interchange of land-mammal faunas between the two continents. The evidence suggests that there was a land bridge continually, if not continuously, throughout the Tertiary except about the middle Eocene and the middle to late Oligocene. There may also have been a shorter break during the early Pliocene. From this it may be deduced that interchange of sea faunas would have been pro- nounced at these three times, but it would be wrong to infer that it did not take place at other times. It seems fairly clear that in the million years or so 's G. G. Simpson: Holarctic Mammalian Faunas and Continental Relationships during the Cenozoic, Bull. Geol. Soc. of America, Vol. 58, 1947, pp. 613-687. '* Sven Ekman: Zoogeography of the Sea (translated from the Swedish by Elizabeth Palmer; London, 1953), p. 96. 547 480 THE GEOGRAPHICAL REVIEW of the Pleistocene there was interchange of seal populations through Bering Strait, even though important movements of land mammals took place across it. It is reasonably certain that there v;^as a series of sea-level fluctuations, though Simpson believes the land connection must have been constant in the early Pleistocene, w^hen most of the land movements took place. To a great extent the Pleistocene epoch w^as exceptional. The rapid fluctuations in sea level that took place then were related to the expansion and contraction of glacier ice, and no reason is known why there should have been similar rela- tively rapid fluctuations in earlier epochs. The most that can be assumed is that the middle Eocene, the middle and late Oligocene, and perhaps the early Pliocene were times of exceptionally easy movement between the Arctic and the North Pacific. Movements at other times would not have been precluded, but there would have been much less time for them to occur, and they are therefore less likely. Climatic Barriers and Avenues Climatic barriers, though less tangible and potentially less difficult to overcome, are none the less real. If the hypothesis that the pinnipeds have always been cold-water forms is allowed, it follows that the major climatic barrier to their spread would be high sea temperatures. Evidence regarding Cenozoic sea-surface temperatures in the North Pacific has been summarized by Durham. ^^ There seems to be no such convenient and documented sum- mary for the North Atlantic, but the general history of its marine fauna during the Cenozoic is outlined by Ekman.'^ Durham's summary shows that in the Eocene the cold-water environment in the North Pacific was of small extent and the February surface isotherm of 20° C. lay somewhere between 50° and 60° N. The August 20° C. isotherm would have been even farther north. During succeeding epochs there was a gradual southward movement of isotherms, so that the cold-water environment expanded progressively until the early Pliocene. There was then a small northward movement, followed by increasing cooling in the Pleistocene. A similar progressive cooling of the seas took place in the North Atlantic. The tropical faunas of the Eocene gave way to subtropical faunas in the Oligocene, and these in turn were replaced in a mass invasion of northern forms at the beginning of the Miocene. The climatic deterioration extended ''J. W. Durham: Cenozoic Marine Climates of the Pacific Coast, Bull. Geol. Soc. of America, Vol. 61, 1950, pp. 1243-1263. '* Op. cit. [see footnote 16 above]. 548 THE PINNIPEDIA 48 1 to the West Indian region, where, according to Ekman,'^ Caribbean tempera- ture "seems to have sunk from 26-27° C. to 19-20°." During the Phocene, temperatures recovered and tropical faunas reappeared in the West Indies. A similar deterioration and recovery occurred in the Mediterranean region but not to so marked a degree. Much less is known about fluctuations in sea temperatures in the Southern Hemisphere. The most significant feature here is the corridor formed by the Peru Current, which brings cold water almost to the equator. There is no reason for believing that it was much less effective through the latter half of the Cenozoic at least, and there is clear evidence in the distribution of existing genera and species that the current has served as a route for pinniped migra- tion on at least two occasions. The conclusions to be drawn from this summary of climatic barriers and avenues are as follows. The cold-water environment, defined for present purposes as that where surface temperature does not exceed 20° C, was of small extent in the Eocene and was virtually limited in the north to the Arctic Basin. In the Oligocene and Miocene it spread considerably, so that by the end of the Miocene it extended south to California and to the Caribbean. Since the Central American isthmus did not exist at this time, there was no land barrier between the cold Caribbean and the cold west-coast waters of South America. There were thus two cold-water routes available, one from the North Atlantic via the Caribbean to Peruvian waters, the other from the North Pacific along the west coasts of the two Americas. Toward the end of the Pliocene, the route from the Atlantic disappeared because of the rein- vasion of the Caribbean by tropical waters and the closing of the isthmus. However, the availability of the Pacific route depended only on sea tempera- tures, and it became usable again on several occasions during the Pleistocene. Family Beginnings Despite suggestions to the contrary, it seems unlikely that the pinniped families are of separate origin. There is general agreement that seals diverged from sea lions and walruses at a very early stage and that the split into the latter two families occurred later. The walruses are, in fact, little more than highly specialized sea lions. In all three families the limbs are used for swim- ming and the tail is vestigial. This suggests that they are descended from an ancestral form in which the tail was too short to develop into an organ of '^ Ibid., p. 71. 549 482 THE GEOGRAPHICAL REVIEW propulsion, whereas in all other aquatic mammals it is used, m a varyhig degree, as propeller or rudder or both. The physiological researches of HowelP° strongly suggest that, doghke, the original pinnipeds used both fore and hind hmbs in swimming but that subsequently they split into two groups, one ancestral to the seals, in which use of the hind limbs came to predominate, the other ancestral to the sea lions and walruses, in which the fore limbs were used more and more. All the evidence of present and past distributions points to the Bering land barrier as the place where this split occurred; for no sea lions, either fossil or existing, are knowm from the Arctic and the North Atlantic, and the only seals reliably recorded from the North Pacific are living forms derived during the Pleistocene from the Arctic and the North Atlantic. If the original split did take place at the Bering barrier, it would have been necessary for the ancestral population to spread from the Arctic to the North Pacific at a time when the barrier was nonexistent, for the barrier to reappear, splitting the population into two, and for it to remain in existence long enough to prevent further contact and interbreeding between the two groups. The split between sea-lion and walrus ancestors followed after an interval long enough for the Pacific group to acquire the considerable number of characters common to both families. This time, population movement was reversed, and a more northern, bottom-feeding group took advantage of another sinking of the Bering bridge to spread back into the Arctic. A subse- quent re-emergence of the land or increasing ecological specialization then cut this group off for an indefinite period, so that it evolved independently into the later walruses, which before the Pleistocene are known only from the Arctic and the North Atlantic. It IS immediately evident that this picture of family origins fits readily into the history of the Bering bridge as deduced by Simpson. The break in the middle Eocene could have been that required for the spread of the original population between Arctic and Pacific. It was followed, according to Simp- son, by a strong land connection in the late Eocene and the early Oligocene, which would have allowed time for characters common to sea lions and walruses to develop in the Pacific group, and then by a break in the middle Oligocene, which would have permitted walrus ancestors to move back into the Arctic. However, the bridge did not close strongly again until the early, or perhaps even the middle, Miocene, so that continued separation of ancestral " A. B. Howell: Contribution to the Comparative Anatomy of the Eared and Earless Seals (Genera Zaloplnis and Phoca), Proc. U. S. Natl. Museum, Vol. 73, 1928, pp. 1-142. 550 THE PINNIPEDIA 483 walruses would be attributable to their increasing adaptation to life in shallow, northern waters. Prorosmarus , the earliest known walrus, appears in the upper Miocene of Atlantic North America. Such a timetable would necessitate a slightly earlier origin for the pinni- peds than might be favored by most, but this seems no insuperable obstacle. A long interval need not have occurred between the first entry into the marine environment and the spread from Arctic to Pacific; an early canoid fissipcd ancestry is not incompatible with the sequence of events postulated above. The Sea Lions (Otariidae) The earliest sea lions were found in the lower Miocene of Oregon and California. According to the hypothesis of generally continuous association between pinnipeds and cold water, this would be expected, since it was in the Miocene that the requisite low sea temperatures reached California. That pinnipeds are absent from Paleogene formations in California, Oregon, and Washington is not surprising; for if the hypothesis is correct, we can expect to find sea-lion forms older than the extinct Allodesmus and Desinatophoca, not in the United States, but in Eocene and Oligocene marine deposits of Canada and Alaska. Unfortunately, such deposits are rare and have been little worked. The cold-water environment spread southward until by the end of the Miocene or the early Pliocene it had reached its greatest extent. By this time a large number of sea-lion types had arisen in the North Pacific; in addition to Allodesmus and Desmatophoca, the genera Neotherium , Pithaiwtaria, Diisigna- thus, PontoUs, Atopotarus, and Pliopedia have been named. All are now extinct. By this time too the East Pacific cold-water route to the Southern Hemisphere was available. The population groups that spread along this avenue were ancestral to the present-day southern sea lion, Otaria, the Australian sea lion, Neophoca, and the southern fur seal, Arctocephalus; and the earliest known southern member of the family comes from the Pliocene beds of the Parana in Argentina.^' It is possible that Otaria is derived from the same stock as the northern sea lion, Eumetopias , and that Neophoca is related to the California sea lion, Zalophus; Arctocephalus is usually linked with the northern fur seal, Callorhinus. The probable lines of spread of the ancestral forms are well shown by the distribution of the present-day sea lions illustrated in Figure 2. All fossil sea lions have been found within the regions inhabited by existing forms. " These deposits are listed as Miocene in Kellogg, op. cit. [see footnote 6 above], but are now believed to be of Pliocene age and are listed as such in Simpson, The Principles of Classification . . . of Mammals [see footnote 1 above]. 551 484 THE GEOGRAPHICAL REVIEW Fig. 2 — Distribution of the sea lions (family Otariidae). Fig. 3 — Distribution of the sea-lion genera Zalophus and Neophoca. When distributions are examined at the generic instead of the family level, it becomes clear that the East Pacific cold-water route has been used on several occasions; for the species that today occupy sections of this route are not derived from the original populations that spread southw^ard but from newer groups that spread both southward and northward along the route on later occasions. These later occasions can almost certainly be equated with the glacial ages of the Pleistocene. Renewed spread from the north is suggested by the genus Zalophus, represented by the California sea lion, which is found as far south as the Tres Marias Islands off the Mexican coast, and by its close relative Zalophus woUehaeki of the Galapagos (Fig. 3). Renewed spread from the south is suggested by the distribution of the southern sea lion, Otaria, which has spread northward as far as northern Peru (Fig. 4). The distribution of the southern fur seal, Arctocephalus , is particularly 552 THE PINNIPEDIA 485 Fig. 4 — Distribution of the sea-lion genera Eumctopias and Otaria. Fig. 5 — Distribution of the southern fur seals, Arctocephaltis. interesting, since it indicates at least three transequatorial population move- ments (Fig. 5). The genus itself is derived by evolution in the Southern Hemisphere from an ancestral stock that spread southward at the time of general movement of the sea-lion family, which is here assigned to the late Miocene or the early Pliocene. It has spread successfully throughout the temperate Southern Hemisphere and has given rise to a series of allopatric species that are a bone of contention among taxonomists. But according to a recent study by King" the East Pacific cold-water corridor is inhabited by two species, the ranges of which overlap considerably. They are both re- corded from Juan Fernandez, off Chile. There seems little doubt that these two sympatric species, A. phiUppi and A. attstralis, are the result of two "J. E. King: The Otariid Seals of the Pacific Coast of America, Bull. Brit. Museum (.V.if. Hist.): Zoology, Vol. 2, 1954, pp. 311-337- 553 486 THE GEOGRAPHICAL REVIEW separate movements along the cold-water route. During a glacial age of the Pleistocene an Arctocephalus population spread northward to California; during the subsequent interglacial it became cut off and had an opportunity to evolve independently into the A. philippi group. In a subsequent glacial it spread south again, while at the same time the main A. australis population of South America once more moved north. The two now overlap and pre- sumably have diverged sufficiently to enable them to remain distinct species. The Walruses (Odobenidae) Early walruses, Prorosmarus, Alachtherium , and Trichecodon, were ap- parently restricted to the North Atlantic, but the fact that they came of an ancestry shared by the sea lions implies that the walruses as a whole originated in the North Pacific or at least in the vicinity of Bering Strait. They first appear in upper Miocene beds of eastern North America, and their gravita- tion across the Arctic and into the North Atlantic was part of a general move- ment of marine animals in this direction in mid-Tertiary time that is cor- related with a southward expansion of the cold-water habitat. According to Ekman,^^ a considerable part of the North Atlantic Boreal fauna was derived from the North Pacific about this time. The present walrus, Odohenus, is also essentially Arctic-North Atlantic in distribution, and its entry into the Bering Sea is almost certainly of Pleistocene date. The distribution of Odohenus has been discussed and mapped in an earlier paper. ^"^ The Seals (Phocidae) The Phocidae are the most diverse and well distributed of the pinniped famihes (Fig. 6), and the members are best adapted to marine hfe. Ancestral seals were separated from the sea lion-walrus stock at a very early stage, and the initial development of the family undoubtedly took place in the Arctic-North Atlantic region. By the middle Miocene, when the fossil record begins, at least five genera had made their appearance, and two of the present subfamilies, the Phocinae and the Monachinae, were distinguishable. The existing genus Phoca was also recognizable. Here, as in the North Pacific, the coincidence is found between the southward extension of the cold-water environment and the sudden appearance of pinnipeds in fossil beds, and here the negative fossil record can be quoted with greater confidence; for the marine Paleogene beds of Europe are much more extensive and better worked ^■5 Op. cit. [see footnote 16 above], p. 159. "• Da vies, op. cit. [see footnote 4 above]. 554 THE PINNIPEDIA 487 ^ Vh / fM ^^^C\ ' 1 \ ^ Distribution of Phocidae J \ \_- 1 L^i Fig. 6 — Distribution of the seals (family Phocidae). Fig. 7 — Distribution of the northern seals (subfamily Phocinae). than those of North America. There is even more reason, then, for the con- clusion that both seals and walruses were confined to the Arctic until the Miocene, when they moved south into the North Atlantic. This southward movement to occupy an expanding environment would have been accom- panied by a marked multiplication of forms and of numbers, which would explain the comparative variety of genera and species named from the Miocene and Pliocene of Europe. By the end of the Miocene, three major groups, which today are named as subfamilies, had formed; a fourth sub- family is named from the Pliocene. THE NORTHERN SEALS (pHOCINAe) The subfamily Phocinae must always have been the northernmost seal 555 488 THE GEOGRAPHICAL REVIEW group, Since it alone has succeeded in invading the North Pacific across the Arctic and, unhke the other groups, it has no Southern Hemisphere repre- sentative. The fossil genera, Aliophoca, Leptophoca, PropJwca, Callopfwca, Gryphoca, Platyphoca, and Phocanclla, have been described from North Atlantic shores, and the existing genera, Phoca, Erignathus, and Halichocrus, are North Atlantic and Arctic in distribution, though the first tw^o have recently separated representatives in the North Pacific (Fig. 7). THE MONK SEALS (mONACHINAe) The monk seals v^ere v^ell represented in Miocene and Pliocene European waters by species of such now extinct genera as Monothcriwii , Paleophoca , and Pristiplwca. The group seems to have been in the van of phocid expansion southward and later in its history to have been particularly characteristic of the Tethyan Mediterranean. Only the ancestors of the present-day Monachus survived the constriction and rewarming of the Mediterranean toward the end of the Pliocene, a feat they doubtless accomplished by adapting them- selves to the warmer environment. Thus arose the one major exception to the rule of association between pinnipeds and cold water. Indeed, Monachus was so successful that, at a time when the Mediterranean was open to the west, it was able to spread across the Atlantic to the Caribbean and enter the Pacific shortly before the Central American isthmus closed toward the end of the Pliocene. Subsequently, the Pacific population was able to spread to the Hawaiian Islands (Fig. 8). There is some possibihty that the spread from Caribbean to Pacific took place in the Pleistocene rather than the Pliocene. King,^^ in a recent discussion of the monk seals, points to many characters in which the Caribbean tropicalis population is much closer to the Hawaiian schaninslandi population than to the Mediterranean monachus population. In fact, schaninslandi and tropicalis appear to be so close anatomically that by analogy with other species it seems doubtful whether they could have been separated since the Pliocene. To meet this objection, King suggests an over- land migration across Panama in the Pleistocene, but this seems even more doubtful, and if the transisthmian spread did take place at that time, it is more probably related to a high sea-level stage during an interglacial. How- ever, there seems to be no other evidence to substantiate such a possibility. The Monachus westward spread seems to have been an isothermal one, taking place along the northern fringe of tropical waters. ^5 J. E. King: The Monk Seals (Genus Monachus), Dull. Brit. Museum (iVar. Hist.): Zoohj^y, Vol. 3, 19>6, pp. 203-256. 556 THE PINNIPEDIA 489 Fig. 8 — Distribution of the monk seals and Antarctic seals (subfamilies Monachinae and Lobo- dontinae). Fig. 9 — Distribution of the bladdemosed seals (subfamily Cystophorinae). THE ANTARCTIC SEALS (lOBODONTINAe) The Antarctic seals (Fig. 8) comprise four species, each of which has been placed in a separate genus, though they are probably derived by adaptive radiation from one group. The Weddell seal, Leptoiiychotcs weddelli, lives farther south than any other mammal; it keeps breathing holes in the ice and feeds principally on fish and squid. The crabeater seal, Lohodon carcinophaga , lives along the edge of the pack ice and feeds on pelagic crustaceans. The leopard seal, Hydrurga leptonyx, ranges from the ice edge northward to the subantarctic islands and feeds on fish, penguins, and young seals. The Ross seal, Ommatophoca rossi, is little known but seems to be a deep diver, feeding mainly on squid. The four species are thus to some extent separated geo- graphically, and they occupy distinct ecological niches. Their evolutionary 557 490 THE GEOGRAPHICAL REVIEW divergence has an analogy in the rapid development o£ the many forms of Australian marsupials from a possible common ancestor that found itself in a continent unpopulated by other mammals. In the case of the Antarctic seals, the nature of this common ancestor can only be inferred from a consideration of anatomical relationships. In her study of the monk seals, King points to many relationships between the Antarctic seals and the monk seals. There is little doubt that the two groups are much more closely related than either is to any other group. In fact, a few taxonomists include them both in the subfamily Monachinae. The inference to be drawn from this is that the Antarctic seals were derived from a group of ancestral monk seals that spread southward to the Caribbean at the time of sea-water cooling in the Miocene and took advantage of the absence of the isthmian barrier to enter the East Pacific cold-water route to the south. Unfortunately, only one fossil find can be called in evidence on this question of Lobodontinae antecedents. A mandible and some teeth found in late Miocene or early Pliocene beds in Argentina were given the name of Prionodt'lphis rovereti and were later found to belong to a seal. Of these, Kellogg^^ says: "The ornamentation and general configuration of the crowns of these teeth are not unlike those of corresponding teeth of the Recent West Indian Seal, Monachus tropicalis. Teeth of this type conceivably might also represent a stage ancestral to that of the Recent Antarctic Weddell seal, Leptonychotes weddelli.'' THE BLADDERNOSED SEALS (cYSTOPHORINAe) The bladdernosed seals form much the smallest seal subfamily, being represented only by Mesotaria from the Pliocene of Europe and by the living hooded seal, Cystophora, of the North Atlantic and the elephant seal, Mi- rounga,o£ the East Pacific and Subantarctic (Fig. 9). Their derivation from the main seal stock implies origin in the North Atlantic, a conclusion sup- ported by the presence o£ Cystophora and Mesotaria in this region. Invasion of the Southern Hemisphere would then have been roughly contemporary with that by the ancestors of the Antarctic seals and would have followed the same Caribbean-East Pacific route. The elephant seals arose subsequently in the Southern Hemisphere, where the existing species, leonina, has succeeded in colonizing most of the anti-Boreal zone. It has also reinvaded the Northern ^* Remington Kellogg: Tertiary, Quaternary, and Recent Marine Mammals of South America and the West Indies, Proc. Eighth Amer. Sci. Congr., Washington, 1940, Vol. 3, Washington, 1942, pp. 445- 473; reference on p. 453. 558 THE PINNIPED I A 491 Hemisphere by retracing the ancestral route of dispersal along the west coast of South America. But this later spread must have taken place in a Pleistocene glacial age, and by this time the door into the Caribbean was closed, with the result that the product of the expansion, the group known as Miroiinga angnstirostris , is found in the Lower California region, having been cut off from the main elephant-seal population by the rewarming of the seas. Rates of Evolution The postulation that pinnipeds did not enter the Southern Hemisphere until the late Miocene or the early Pliocene will probably meet with the objection that this provides too little time for southern forms to have de- veloped to the extent that they have. But uneven rates of evolution are as marked in the pinnipeds as in other groups of mammals. The existence of clearly defmed seals, sea lions, and walruses by the Miocene has long inhibited zoologists from placing the date of pinniped origin as late as might seem desirable on other grounds. It is even more notable that seals of the modern genus Phoca, and only slightly distinguishable from the modern species Phoca vititliua, are recorded from the Miocene. The evidence, although small in bulk, clearly shows that there was comparatively rapid evolution up to the Miocene but that since then rates have been extremely slow, at least in the Northern Hemisphere. The general hypothesis of pinniped origin and dispersal advanced here seems to provide a satisfactory explanation for this change in the evolutionary tempo. From what is known or inferred about the evolutionary processes, rates would have been rapid in the early stages when the ancestral pinnipeds were first entering the new marine environment. Equally, the expansion of the cold-water environment from the late Eocene through the Oligocene and Miocene would also have provided conditions for rapid evolution, bringing, as it must have done, great increases in population numbers and in varieties of habitats. But by the end of the Miocene the expansion had come to a halt: all available regions had presumably been occupied and most readily utilized habitats exploited. The one major exception was the Southern Hemisphere. The seal and sea-lion groups that entered southern waters discovered a whole new cold-water environment in which they were able to multiply enor- mously and to occupy a variety of habitats. A second golden age of pinniped evolution dawned, but one that has lasted a much shorter time than that of the Paleogene in the Northern Hemisphere. As a result, differentiation in the south has reached the generic stage only. 559 492 THE GEOGRAPHICAL REVIEW Regional Geography of the Pinnipeds Sclater^' proposed marine zoogeographical regions based on the distribu- tion of marine mammals, and Von Boetticher^^ re-examined them with reference to the distribution of the pinnipeds. The usefulness of these regional defmitions may be doubted, but it is worth while to review them here in order to summarize some of the conclusions reached in the present study. Figure lO shows Sclater's regions somewhat modified. Notopelagia Fig. 10 — Pinniped regions: I, Arctatlantis; II, Arctirenia; III, Mesatlantis; IV, Mesirenia; V, Noto- pelagia. The arrows indicate the two major dispersal routes between the Northern and Southern Hemi- spheres. I. Arctatlantis extends from Bering Strait across the Arctic to about the August sea-surface isotherm of 20° C. It is the region of development of the seals and walruses, and there is no evidence that the sea lions have ever penetrated it. It is characterized today by the presence of Halichoerm and Cystophora, which are endemic, and by Phoca, Erignathus, and Odohcnus, which were probably endemic until the Pleistocene. II. Arctirenia extends from Bering Strait south to about the August sea- surface isotherm of 20° C. Here the sea lions developed and still have their headquarters. The walruses must have originated in the extreme north of Arctirenia but early moved into Arctatlantis. Both seals and walruses spread into the northern section of Arctirenia during the Pleistocene; earlier in- ^^ Op. cit. [see footnote 3 above]. ^' Op. cit. [see footnote 2 above]. 560 THE PINNIPEDIA 493 cursions may have occurred but left no trace. The region is characterized by the presence of Callorhinus and Ewnetopias, which are endemic, and by Zalophus, which would be endemic if it had not spread to the Galapagos. Comparatively recently, Phoca, Eri^nathus, and Odohemis entered from the Arctic and Arctocephalus and Aliroim^a spread into the extreme southeast off Lower California. III. Mcsatlantis comprises the Mediterranean and Caribbean Seas. IV. Mesirenia consists of the seas around the Hawaiian Islands. Regions III and IV are the home of the relic genus Monachus, the only surviving monk seal, which is endemic and the only form that occurs here. V. Notopelagia extends from Antarctica northward to about the February sea-surface isotherm of 20° C. and thus includes the Southern Ocean and the coastal waters of Chile and Peru. All forms — Lohodon, Omniatophoca, Leptoriycfwtes , Hydrur<^a, Mirounga, Otaria, Neophoca, and Arctocephalus — are endemic except that Mirounga and Arctocephalus have partly invaded Arcti- renia. All have evolved in the Southern Hemisphere from ancestral groups derived from the north. 561 A Numerical Analysis or tne Distributional Patterns or Nortn American Mammals. 11. Re -evaluation or tne Provinces EDWIN M. HAGMEIER Abstract In an earlier paper, numerical techniques were developed and used to analyze distribu- tion patterns of the native terrestrial mammals of North America. An error in method is here corrected, indicating that 35 provinces, 13 superprovinces, four subregions, and one region may be recognized. The methods used are relatively objective, quantitative, and suited to computerization. Introduction In an earlier paper ( Hagmeier and Stults, 1964, hereafter referred to as H & S ) , quan- titative and relatively objective methods were used to demonstrate that ( 1 ) the range limits of North American terrestrial mammals are grouped, (2) that as a result it was possible to delimit geographic re- gions of faunistic homogeneity which were termed mammal provinces, and (3) that such provinces could be useful in the analy- sis of other zoogeographic phenomena. This paper is concerned with the recal- culation of some of the data of the second item above. It was assumed in our earher paper (H & S) that several of the large provinces of the northern half of the con- tinent required further analysis. On initia- tion of this analysis, it became apparent that an error in method had been made which required correction. As a consequence of the correction, the number of North Ameri- can mammal provinces is here increased from 22 to 35, two of which are of uncertain status. The general philosophy, methodology, and conclusions reached in our earlier paper ( H & S ) do not differ from those arrived at here, and the earlier paper should be referred to for accounts of these. The ma- terial given here, since it is essentially re- visionary, is presented in as brief a form as possible. Because of the changes reported here, the analysis of mammal areas given in H & S (p. 141-146 and Figs. 6c-8d) needs revision, and this revision will form the sub- ject of a future paper. Since submission of our earlier paper (H & S), Simpson ( 1964 ) has considered variation in abun- dance of species of North American mam- mals in a superior manner, and his work should be referred to for a treatment of the subject. Derivation of Provinces In our earlier paper ( H & S ) , the ranges of all 242 species of native terrestrial North American mammals were converted into a model, first by separately computing the percentage of species and genera whose ranges ended within blocks 50 miles by 50 miles throughout the continent (each such value was called Index of Faunistic Change, or IFC), second by plotting species and genus IFCs on maps of North America, and third by fitting isarithms. The species IFC map resulting was given as H & S, Figure 1. Low IFC values indicated faunistic homogeneity, and regions characterized by such values were termed primary areas. 279 562 280 SYSTEMATIC ZOOLOGY Fig. 1. Eighty-six primary areas derived through examination of IFC Map ( H & S, Fig. 1 ). Twenty-four of these were identified through examination of IFC maps of both species and genus and species checklists of each were prepared (H & S, Table 2). The per- centage of species common to all combina- tions of pairs of provinces (Coefficients of Community, or CCs, Jaccard, 1902) were then computed. CCs were then subjected to cluster anah sis using the weighted pair- group method, and simple averages (Sokal and Sneath, 1963:180-184, 304-312; H & S: 132, 137), and drawn up in the form of a dendrogram (H & S, Fig. 5). Primary areas pooled at a mean CC level lower than 62.5% were termed mammal provinces, those pool- ing at below 39% were teraied mammal superprovinces, those below 22.5%, mam- mal subregions, those below 8% as mammal regions. The basis of our error lay in the fact that we attempted to identify only the 24 North American mammal areas corresponding to those described by Kendeigh ( 1961 ) in his modification of Dice's (1943) scheme of biotic provinces. This error became appar- ent only when subsequent analysis of parts of certain of these provinces showed that the parts in some cases merited full prov- ince status. The correction made and reported here 565 DISTRIBUTION OF NORTH AMERICAN MAMMALS 281 u l-l n < uBiuipjBiuag UBS ueiiqBiQ UBIlUOjrlB^ uBiABqopv UBajBjv UBg UBIJUIfl UBJUOqBAB^ UBtUltdBJ^ UBiutnb8^\ UBJOUOg UBiqBqiE^j nBisaiuajJV nBtquinio3 UBiUBismc-i UBX3X UBdjinBuiBX UBIUOOIBg UBiUB/v\aqo4B>]SBS UBSUBJJ nBiuijojB3 UBIOUini uBiuBqgaiiv uBjjais uBppioquinH UBiuoSajQ UBUSAnOOUB^ UBpBJOlO^ UBIUBJUOJ^ UBipBUBQ -yVX uBipBUB3 -g uBniospnjj 'g UBiuospnjj -^ UBlU03(nj^ uBi^najY nBijsBjv UBOuii>(sg -jw TiBOUii5{S3 -g UBABgUf^ in CO X X n n X CO i-H CO X o CO X X X 05 IN X 00 * X X CO X X O ■ ^ • o ■ =o i 2 : .O <3 3 « "5 •S -2 CQ •S2 h " Q " 3 »r <» CJ CO 2 'Xs <5j 2 c» : -a _ o -- a -^ 2 S- a o „ o S g-t -^ O < 12; c/2 CA3 c/2 o ci CO b a 3 ,cs u a. 35 ^ "-^ ^-^ o 35 ^ "13 3> O C ^ 3 3 ^J ■h 3 CO -« 'a 3 o 3 y c Co 3 3 - '^ i 3s -^ « O 3 3 3 ^ ~ V g Q 30^/<>'_:3i C^c/^UQOOc/ovDCoooco&ococ/:) 564 282 SYSTEMATIC ZOOLOGY c o a -] a < uBiuipjEiuag UBS J2 ueiiuojnsD ^ ueiABqojv ^ m;aiBi\ ucs j:^ UBtUOqBABl^ § UBllUldBJ^ * UBiuinbB^\ (Nj UEJOUOS cq w uBiqBqiB]^ cq UBisaiuajjy c^ UBiqiunjOQ M UBiuEisinoq cq UBdi[nBiuBX § UBtuoajBg 2 UBIUBMaqDJBJJSBS rt ubsub;S ^ CO UBtUipjB3 ^ ID UBIOUIJIJ -H uBfUBqSajiv 21 ueuajs 2 uBijpjoqiunjj 2 uBiaogajQ ^ UBU3AnOOUBj\ 2 UBpBI0[O3 uBiaB4uoj\ 00 UBipBUB3 -AV ^ w UBIPBUB3 '3 p; UBiuospnji '3 <3 UEiuospnjj "w ^ uBiuojfn^^ in uBi}najY Tr ue^SBiv « UBOUII31S3 AV g UBOui:>(S3 3 ^ UBAESUfJ --I X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X to s c a ^ =0 c £ = ^ s V. o ^ . . ►J ►J ►J X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X XXX X XX X X X X X X X X X X X X X X X X X X .2 C •2-S 05 3 ■S "s- 9 CO 5 "2 >-j >-i kj -i.i •2 Eisj bi UJ bj I^i 5 ^ i^ s ^1 "o 2 = .1 3 a . . tq bj hj fai hj tj ti tq tij :§ 15 ;§ *. 'o •2 a fc X X X X X X X X X •2 a >« Co 3 o g o UBIIUOJIIBO ^ UBiAuqojv ^ UB<)1BPV UBS n UBIUOqBABfJ gj UBIlUldB^ ^ UBiumbB^ g^ UBJOUOS S UBiqBqiBX M uBisaiuanv ^ uB!quini03 ?5 UBiuBtsmoq ci uBxax ^ UBtJjinBlUBJ, (N UBlUOOJBg r-( UBlUB/V\9q04BJfSBS * U8SUBX i^ UBtUlJOJB^ 2 UBiouqii 12 uBiuBqSaiiv r^ HBJI3IS 2 UBi4pioqiunH 2 uBjuoSajQ ^ UBuaAnoouB^ S UBpBJ0[03 C5 UBIUB^UOpl^ '^ UBIPBUB3 -jw > UBIPBUB3 3 W UBtuospnH -a § UBiuospnH -M I UBtuo5(n^ in uBijnaiv "* UB5(SBIY CO UBOUIt3(S3 a ^ UBABgUf^ ^ X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X S S c H -2 c - - cat: CO .2 » X X X X X X X X X X X X X X X X X X c/3 c/o 00 tyi c/:i CO c/i c/i CJ CO CO c 3 -2 c X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X •a X X X X X X X X •2 2 Co Co 3 ■* r< . a 3 K u 6 u c^ 50 c ?r> y '^ ^ i ^ I •- >. .iO •" w. »~ Cic o c a a 3 § 3> 3 •S £ 3 CO 3) ■>3 O b S tic 2 3 C-o 2 o 3 a — 5 -t3 Q CO o . Q • ~ • -s: ty:jocc/2COCA;E~iE~ •* :rt z^ s SI 2 ^ CO &2 o .s Co O 2 8 _■ ^r^ c ^ E-c ^1 o 2 ^J U CO 5a ?=■»-« ,- - . CJ U U (^ 0^ c^ a^ 566 284 SYSTEMATIC ZOOLOGY s e s o n < uBiuipjetuag UBg "^ UBiiuoj!ie3 n UEIABqOIV' ^ ub3;bi\ UBS n UBIUOqBABfiJ ^ UBIUIldBp^ ^ UBtutnbBj^ c5 UBJOUOS * UBiqcqiBX c5 UBisaius^iy (Ji uB!qiun|03 ^ IN uBiuEismoq cm UBX3X M uBdjinBuiBX § uBiuoaiBg 2 uBtnBAvaqojBJ^SBS * UBSUB^ ^ UBIU!|0JB3 2 UBIOUini ^ UBiuBqSaiiv 3 uBjjats 2 UBppioquinjj 2 UBinogaJO '^ uBuaAnoouB^ o UBpBIOlOQ 05 UBIU6}UOI>J oQ UBipBUBQ "^ ^ UBipBUBQ -g W UBruospriH 3 § UBinospnjj -yy^ ^ UBtuo>fn^ lO nBJfSBjv w UBOUiiijsa -A^ > UBOui!i[sa -a § UBABgUfl 'H X X X X X XXX X X X XXX XXX X X XX X X X X X X X X X .2 X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X XXX X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X (n^ lO uBijiiaiy rr UB51SB[V ^ UBOuiinsa ■A\ g nBABguj^ ^ X X X X X X X X X X X X X X X X X X X X XXX Co X X X X X X X X X X X X X X X X X X X 00 3 X X X X X X X X X X X X O 5u X X X X X X X X X X X X X X X X X X V X X X X X X X X X X X X X X XX X X X X X X X X X X X X X X X X X X X o o 3 X X X X •S a.-~ ::i 3)-^ CO -r CO 3 c o X X X X X X X X XXX X X X X X X X X X X X X X X X CO CO 3 3 "ig o X X X X X X X X •^r' = 2 ^ 35 c -g "e -s 3§s^§^5c 5 tUD o o 00 3 O ■Si ^ s. §■ X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X S; 3 3> .2 c CO — y a c^ 3 •^ ■£• Si K U 1^ O c ■2 3 o tie o CO »-<» ■to 0 0. 3 "^ CO 3 3 3 - ■ 3 O c O 568 286 SYSTEMATIC ZOOLOGY UBiuipjcaiag ub<; w uBiajoj!ii!3 « UBiABqoiv ^ ueaiBJV "«§ ^ UBIUOHBABfiJ ^ UBlUIldBJV * nBtninbB_^ c5 nBiouos S uBiqBqtB}^ ^ uBisaiuajJV m O UB!qiunj03 oq cq UBlUBISinO'l c^ UBxax J3 UBdjinBuiBX § UBJUOOIBQ 2 HBIUBA\aq04B31SBS 2 UBSUB^J ^ UBIUipjBQ 52 UBIOUIIJI J2 UBiuBqSaUV 2 uBjjajs w uBiipioqiunfj 2 UBiuoSaiQ '"' UBIiaAnOOUBA o UBpBJOlO^ c; UBIUB^nOI^ 00 UBipEUBQ -AV ^ UBIPBUB3 3 W uBiuospriH 3 g CO UBIUOJpi^ lO UBiaospnjj ■ w nBjinaiv "* nBijsBiv ^ uBouiijfsg Av g UBOUIT3(S3 -3 " UBABgUfJ ^ X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X c C CO 5 = S2 — X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X ■ X X X X X X X X X X X X X X 7x- «<5 S =0 CO -C 3 i -r -^ -« § S ^ = ^ §-^:i C K 5J ». k'lsc ^ CO »~ tS 'S -^ cc 5 c^ V. — i _rs .= -« § h 2 -5 ^ 2 O o o y ^ N *' s .2 sa a -■ r^ Q t •- o a. & be 1 2 ^ I c CO 5 c c/D c/2 Q N N N •2 So a c N O P *^ ' 12: w l>5 a CO », «9 3 ca H = = a. -2 ■s c o 5i) to 3 c a 2 <=> uo<:>>;:3;::);:i 569 DISTRIBUTION OF NORTH AMERICAN MAMMALS 287 UBiaipjBUjag UBg UBIiqBIQ UBIIUOJIJB3 UBlABqOJA^ 55 ub3;bj\[ UBS n UBJUOqCAB^ ^ UBIUIldBJ^ ^ UBiuinbBj^ ^ UBJOUOg ^ uBiqBqiB}j S UBis3Uia4JV ^ UBiquinjoQ O UBiuBismoq ^ UBdiinBiuBX O UBIUOOJEg 05 UBlUBMaqO}B)(SBS CO ubsub;X t^ UBIUIJOaBQ CO UBIOUTUI W UBiuBqSaiiv ^ UEijats 2 uBppioquinfj 2 uBiuoSajQ '"' uBuaAnoouE^ o UBpBiOI03 05 UBIUB4UOJ^ 00 UBipBUBO 'M ^ UBIPBUB3 a W UBHiospnH a g uBiuospnH'AV ^ UBiuoijn^ w uBpnajY •* UBJJSBIV w UBOlUlJlSa "M ^ uBouii^isa a ^ X X X X X X X X X X X XXX XXX XXX X X X X X X X XXX X X X X X X X X X X X X X X XXX XXX XXX XXX XXX XXX XXX XXX XXX X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X XXX X X X XXX X X XXX XXX X X X X XXX XXX X X X X X X X X X X X X X X X X X X X X X X X X X X X X X XXX XXX XXX X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X 3 •c .y cc c 3-5 XXX X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X XX XX X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X XXX X X X X X X XXX X X X X X XXX X X X X X XXX XXX X X X X X XX X X X XXX X X X X X X X X X X X XX XXX X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X •^ s •2 CO « & a c 3 a ca y s c . 3 o c :§;§:§suESt^ ■C S - a -^ OS 3 3 S Is O 3 o c o •2 a ^ .2 1 § 3 ? § ^ CO-* 3 Q CO 3 c a ■S CO CiD » ^ c S.2 a C §••2 ■■^ s e -2 CO 3 c ^ 2 '^ N .2 2 CO O "^ §'^ ■2't3 V. a » . oooo 570 288 SYSTEMATIC ZOOLOGY £6 - /oo 1 ^ ¥3- i.S' Fig. 2. Final trellis or matrix giving Coefficients of Community ( percentage of species common to pairs of provinces). Ordering of provinces is that resulting from cluster analysis. Heavy lines outline superprovinces and subregions. The classes of shading shown in the mirror image are determined by the critical mean CC values used to obtain higher categories of areas. 571 DISTRIBUTION OF NORTH AMERICAN MAMMALS 289 consisted of laying a transparent overlay of primary areas with CCs higher than 65% over the species IFC map ( H & S, Fig. 1 ) (as determined by actual calculation or by and drawing hues through all regions of averaging during cluster analysis) were high IFC value, dehmiting ultimately a total pooled to create secondary areas. The of 86 (rather than the original 24) primary whole process was then repeated. A new areas. The distribution of these is given as species checklist was prepared for each area, Figure 1. The genus IFC map was not used CCs were computed and subjected to in the corrected analysis. cluster analysis, and a new dendrogram Subsequent procedure was that of our prepared. Pooling of areas was again done earlier paper. Species checkHsts of each of where CC values were higher than 65%. the primary areas were prepared, and CCs In all, four such sets of sequential opera- were computed and subjected to cluster tions were carried out. In the final opera- analysis. Because of the large number of tion, the total number of primary areas had CCs involved (86! =3,741), calculation of been reduced from 86 to 38 secondary areas, CCs in this and subsequent operations was all but three of which had CCs lower than done by computer. 65% (2E and 2W, 6E and 6W, and 7E and In our earlier paper (H & S: 137-138), 7W; see Figs. 2 and 3). These three sets a mammal province was defined as an area of secondary areas were not pooled because with a mean CC of 62.4%) or less, when com- they occur over large geographic areas and pared to other areas by cluster analysis, because I was concerned with their detailed This decision was based on the work of analysis. The 35 secondary areas with CCs Preston (1962), who found that analysis of less than 65% constitute mammal provinces faunas by means of a "Resemblance Equa- by the standards used here and are so tion" (RE) indicated that values oi z (as treated, although two pairs of these fall derived from the RE ) of about 0.27 repre- within the questionable range 60-65% ( 15 sented the break between faunistic homo- and 16, 34 and 35, see Fig. 4 ) . Figure 2 is geneity and heterogeneity. In our earher the matrix resulting from cluster analysis, paper we converted z to S (Similarity), showing ordering of provinces and Coeffici- where S = 100 (l-z), and calculated both ents of Community between pairs. Figure S and CC for all items in the matrix. These 3 is a map showing geographic distribution were compared by regression, giving a slope of the provinces, and Figure 4 is a dendro- b = 1.17 ± 0.02. Conversion of Preston's gram delineating the faunistic relationships critical z value to S gave an S of 73%, and existing between provinces, as determined conversion of the critical value of S to CC by cluster analysis. A species checklist for equaled 73/1.17 = 62.4%, and hence our use each of the provinces, as it was used in the of this value. We did not, however, allow final operation, is given as Table 1. for the effects of statistical error. If this is In our earher paper (H & S), mammal incorporated, in the form of plus and minus provinces were grouped into the higher cat- two standard errors ( providing limits at the egories of superprovinces, subregions, and 95% level of probabihty), the critical CC regions. The method used in deriving these value falls within the range 60.30-64.60%. was to draw lines across the dendrograms at As a result, in this and subsequent papers I suitable CC levels, the choice of CC level propose the use of a CC value of 65% as being arbitrary but providing what ap- critical for the determination of mammal peared to be a useful classification (H & S: provinces. This is a conservative standard, 139-140, 149). I have tried here to make and all values lying between 60 and 65% as little change from the original scheme as should be considered suspect, and are re- possible; however, a small number of minor ported. adjustments have been necessary. The results of cluster analysis were evalu- The 0-8% CC range of the dendrogram ated according to this new standard. Pairs still stands as a level useful for the category 572 290 SYSTEMATIC ZOOLOGY ALASKAN SIERRAN MOHAVIAN SAN BERNARDINIAN KAIBAIIAI SONORAN YAOUINIAN Fig. 3. Final grouping of primary areas into mammal provinces. Broken lines indicate subdivisions of provinces. The approximate relationships of island faunas are also shown. of region. At this level, one region, the Nearctic is isolated. Subregions in our ear- lier paper stood between the 20-25% CC range; this is changed here to the 22-27% CC range, and still encompasses four subre- gions, following Wallace ( 1876 ) . The cate- gory of superprovince was, in the earlier paper, set at a mean CC level of about 39%. The selection of this value was based on conclusions reached by Savage (1960), de- tails of which may be obtained from H & S, p. 139-140. The 39% level would, in the case of the dendrogram used here ( Fig. 4 ) give 11 superprovinces. Several cluster at a level very httle higher than this, and I have arbitrarily moved the limit up to about 42.5%, so as to encompass these, giving a total of 13 superprovinces. These decisions Fig. 4. Final dendrogram showing relationship between provinces. Ordering of provinces and mean Coefficients of Community ( CC ) are the results of cluster analysis. Solid vertical lines show mean CC levels at which regions, subregions, superprovinces, and provinces segregate. The three vertical lines for provinces represent the mean critical value plus and minus two standard errors. Per cent similarity is mean Coefficient of Community ( CC ) . The "diamonds" of provinces 2, 6, and 7 represent the mean CCs at which the subdivisions of these provinces pool on cluster analysis. 573 DISTRIBUTION OF NORTH AMERICAN MAMMALS 291 — r- ¥0 re INIAN bl^ 292 SYSTEMATIC ZOOLOGY continue to fill the desirable requirements outlined in our earlier paper. Good single values for each of these hier- archic levels would be: provinces 62.5%, superprovinces 42.5%, subregions 25%, and regions 5%. By this scheme the 35 mammal provinces of North America are grouped into 13 superprovinces, four subregions, and one region. These are named and their dis- tributions mapped on Figures 5a and 5b. They are also blocked out in the matrix (Fig. 2) and marked by lines drawn at ap- propriate CC levels on the dendrogram (Fig. 4). The problem of nomenclature of mammal areas is discussed subsequently. The nearest approach to the biotic prov- inces of Dice ( 1943), and Kendeigh ( 1961 ) through the analysis carried out here, is ob- tained by fitting a line at about the 54% CC level of the dendrogram. The more in- tuitive decisions of these workers implies a degree of segregation about 10% lower than the one used here. Nomenclature and Status of Areas No changes in the names or status of regions or subregions over those of our earlier paper have resulted from the reanal- ysis. Because of the increase in numbers of provinces however, minor adjustments in these and in other categories have been obligatory. As few name changes as possible have been made. Where new names have been appHed, an attempt has been made to take them from the literature and to apply them on the basis of priority. Where there has been need to coin names, I have tried to follow the spirit of earher workers. As an aid to recognition, names of provinces are in the form of adjectives, names of super- provinces in the form of nouns. Those cases in which provinces segregate within the doubtful 60-65% CC range, or in which segregation occurs at a level only slightly higher than 657©, are here de- scribed. More detailed analyses will doubt- less result in some changes in status within these groups. The following provinces are new and have been named by me: no. 1, Ungavan; no. 12, Humboldtian; no. 25, Kaibabian; no. 30, Uintian; no. 31, San Matean; no. 34, DiabHan and no. 35, San Bernardinian. Those provinces that are new but that have been given an older name are, together with the source of the name: no. 3, Alaskan (Allen, 1892); no. 5, Yukonian (Cooper, 1859); no. 10, Vancouverian (Van Dyke, 1939); no. 13, Sierran and no. 23, Colum- bian (Miller, 1951). The names Saskatche- wan and Mapimi are here converted to the adjectival forms, Saskatchewanian, ai?d Mapimian. The Hudsonian (no. 6) is a new province. The name was formerly applied to the prov- ince here termed Canadian (no. 7), and the Canadian of earlier workers is here termed Alleghenian (no. 14), following the precedence set by Cooper (1859), and Allen (1892), after Kendeigh (1954). The Carolinian ( 16 ) is a new province. The name was applied to what is here largely represented by the Louisianian province (22), in our earlier paper. Its present ap- plication is the correct one, however, by the standards of older workers. The Louisian- ian should properly be called the Austrori- parian, following Dice (1943), Kendeigh (1961), and H & S. Because it has super- province as well as province status, and re- quires the nounal form of the same as well as an adjectival one, I have applied Allen's (1892) terminology to it. Not all provinces segregate clearly. The northern limit of no. 3, the Alaskan, was dif- ficult to locate, it being a region of broad transition. Its mapped limit is relatively arbitrary. The Sitkan province of H & S is here included in the Yukonian (no. 5), and the Vancouverian (no. 10), clustering with the former with a CC of 76%. That part of the Yukonian province made up of the Brooks Range very nearly segregates with a CC of 66%. The Eskimoan, Hudsonian, and Canadian provinces (nos. 2, 6, and 7) have, for rea- sons given elsewhere, each been split into eastern and western components. Of these, 575 DISTRIBUTION OF NORTH AMERICAN MAMMALS 293 OREGON COLUMBIA CALIFORNIA MOHAVE MAPIMI NAVAHO a. SUBREGIONS b. SUPERPROVINCES Fig. 5, a and b. The distribution of mammal subregions and mammal superpiovinces, as determined from the dendrogram. the Eskimoan components cluster with a CC of 67%, and the Canadian with a CC of 66%. These are nearly critical values, indi- cating that the components ahnost merit province status. The Oregonian (no. 11) almost segre- gates into western coastal and eastern Cas- cadian provinces, pooHng with a CC of 66%. The Alleghenian (no. 14) segregates from the eastern component of the Canadian (no. 7) with a CC of only 64%, a critical value. It clusters with the CaroHna superprovince on analysis, however, and does so with prov- ince rating. The Carolinian (no. 16) is not a clearly defined province and probably should have been pooled with the Illinoian (no. 15), the two clustering with a CC of 64%. The Car- olinian is made up of three distinctive geo- graphic components; that east of the Appa- lachian Mountains clusters with the rest of the province with a CC of 67%, and that of the Ozark Mountains clusters at 66%. Two of these components are distinct enough from the Illinoian that I have provisionally kept the Carolinian as a full province. The Balconian (no. 19) was incorrectly identified by H & S as part of what is here called the Tamaulipan (no. 20). The Bal- conian in its present sense stands as a full province. The Louisianian (no. 22), under the name Austroriparian, was in part identified as the Carolinian by H & S. Its distribution as determined by reanalysis is the more realistic one. The larger part of the Columbian prov- ince (no. 23) was named Artemesian by H & S. The latter tenn is here applied to a restricted portion of the Columbian as prov- ince no. 24. The Palusian of H & S is here pooled with the Columbian, with a CC of 70%. The Mohavian province (no, 32) presents something of a puzzle. It was not recog- nized through examination of the IFC map but appeared through scrutiny of individ- ual species maps. Once recognized, how- ever, cluster analysis caused it to segregate out to the extent of meriting superprovince status, and I have accepted it as this. Its geographic limits, however, have been de- 576 294 SYSTEMATIC ZOOLOGY termined subjectively, and they should be considered as suspect. The Diablian (no. 34) is also of uncer- tain status, as it clusters with the San Ber- nardinian (no. 35) with a CC of 617o, a critical value. Because not all of the latter occurs in the area studied, its analysis is incomplete, and for this reason the distinc- tion is provisionally accepted here. Of superprovinces, the Texas, Columbia, Mapimi, and Mohave are new, and the names Hudson and Austroriparian of H & S are replaced by the names Canada and Louisiana, for reasons given elsewhere. Insular Faunas Sixt\'-four species (27%) of the total mammal fauna occur on the larger islands adjacent to the continent and on the islands of the Great Lakes. Insular faunas were in each case compared with the faunas of several of the nearest mainland provinces by means of the Coefficient of Community and Simpson's Coefficient ( SC ) . The latter is a measure of the percentage of species occurring on an island that also occur in in the mainland province (Simpson, 1943; H & S: 131). Results are given in Table 2. Most island faunas give a CC much lower than Preston's critical 65% value when com- pared with the faunas of adjacent mainland provinces (Table 2). Most islands would therefore merit full province status, if this standard were to be applied. The generally low CC obtained, however, is the result of the small size of insular faunas, a bias being introduced as a consequence of it. Simp- son's Coefficient is, in these circumstances, a more reliable measure, and I attribute greater significance to it. No critical value of SC is available, however. Because of this, and because there is more interest in similarities than dissimilarities, island faunas have in all cases been named as part of the fauna of the adjacent province to which they show nearest relationship as determined primarily by Simpson's Coef- ficient. Table 2. Island No. of species Adjacent provinces CC SC Long Island 29 14 Alleghanian 16 Carolinian 51 52 93 61 Cape Breton 31 14 Alleghanian 7E E. Canadian 55 64 94 87 Prince Edward Island 29 7E E. Canadian 14 Alleghanian 68 48 93 90 Anticosti 5 6E E. Hudsonian 7E E. Canadian 14 Alleghanian 17 13 10 100 100 100 Newfoundland 12 6E E. Hudsonian 7E E. Canadian 14 Alleghanian 40 28 19 100 92 83 Belcher 3 1 Ungavan 6E E. Hudsonian 7E E. Canadian 25 10 3 100 100 33 Manitoulin 25 14 Alleghanian 7E E. Canadian 49 62 100 96 Isle Royale 9 7E E. Canadian 14 Alleghanian 24 18 100 100 Arctic Archi- pelago Kodiak 10 12 2E E. Eskinioan 1 Ungavan 5 Yukonian 4 Aleutian 3 Alaskan 67 57 33 57 39 100 80 100 92 85 Alexander Archi pelago - 21 10 Vancouverian 5 Yukonian 61 44 95 90 Queen Charlottes 12 10 Vancouverian 5 Yukonian 36 24 92 83 Vancouver 22 11 Oregonian 10 Vancouverian 33 54 83 79 Long Island shows closest relationship to the Alleghenian province (no. 14), not the Carolinian (no. 16), as might have been expected. Cape Breton is nearest to the Alleghenian by Simpson's Coefficient; I have placed it there though it shows a very high CC with the eastern Canadian (64%)r Prince Edward Island lies nearest to the eastern Canadian province (no. 7E), which is surprising in the light of its geographic proximity to the Alleghenian. Anticosti is grouped with the eastern Hudsonian (no. 6E ), on the basis of its high CC, as is New- foundland, the latter on the basis of both coefficients. Belcher Island is, because of its CC only, treated as being most closely related to the Ungavan (no. 1). Manitoulin has highest SC with the Alleghenian, high- 577 DISTRIBUTION OF NORTH AMERICAN MAMMALS 295 est CC with the eastern Canadian, but be- cause greater weight is given to Simpson's Coefficient, I have grouped it with the Alleghenian. Isle Royale shows closest af- finity with the eastern Canadian province. Of the islands of the west coast, the Alex- ander Archipelago and the Queen Charlotte Islands show closest relationship to the Vancouvarian (no. 10). Vancouver Island, on the basis of its SC only, is closest to the Oregonian (no. 11). Kodiak Island has highest CC with the Aleutian (no. 4), but highest SC with the Yukonian (no. 5), and following the policy set previously is considered most closely re- lated to the latter. Of the Arctic archipelago and Greenland, the following groups of islands have identi- cal faunas: group 1, Baffin, Southampton, and Coats islands; group 2, Somerset Island; group 3, Banks Island; group 4, Greenland, Sverdrup Islands, Borden and Prince Pat- rick islands; group 5, Victoria, Prince of Wales, Melville, Bathurst, Comwalhs, Devon, and Ellesmere islands. The faunas of these groups of islands, together with those of the Ungavan and eastern Eskimoan provinces (nos. 1 and 2E) were analyzed by first computing Coefficients of Com- munity between them, then subjecting these to cluster analysis, using the methods outlined previously. All of the island groups cluster at a mean level of 61% or higher, falling within the critical range or better. The groups taken together segregate from the eastern Eskimoan with a mean CC of 54% and from the Ungavan with a mean CC of 44%. The equivalent mean SCs are 1007o and 85% respectively. As a result I have treated all of the Arctic archipelago and Greenland as part of the eastern com- ponent of the Eskimoan province. A generalized mapping of these relation- ships is given in Figure 3. It should be noted that the affinities of Cape Breton, Prince Edward and Long islands are indi- cated incorrectly here. Discussion The general conclusions reached as a result of this re-evaluation differ in no way from those obtained through our earlier analysis (H & S; 147-151), and they are not treated further. It is important that the subjectivity of the methods used here be kept in mind however. The sources and attempted controls of these have been dis- cussed in our earlier paper (H & S: 148- 149, 151 ) and include taxonomic errors, dis- tributional errors, choice of point or block for sample; size of sample block, fitting of isarithms, selection of primary areas, choice of coefficient of association, choice of clus- tering method, and others. The methods used here are ideally suited to computer techniques. This reanalysis could not in fact have been completed within reasonable time had such techniques not been available. Miller, Parsons, and Kof sky ( 1960 ) have described the use of so-called successive scanning mode micro- densitometers, which automatically map the densities of films and other kinds of trans- parencies. Such devices are sold by Beck- man and Whitley of San Carlos, California, under the registered trade name of Iso- densitracer. The use of such a device on a transparent map showing the distribution of all North American mammals drawn in inks or paints which gave progressively less translucency as additional layers were added, would be ideal in the development of more refined IFC maps. The IFC map used in this work ( H & S, Fig. 1) was based on the computation of the percentage of species whose ranges ended within blocks 50 miles to a side. The abso- lute value of an IFC is a function of size of block (H & S: 148). I suggest that any future use of IFCs incorporate as subscript to values given, a statement of the area of the block in kilometers. Converting size of block used here to square kilometers gives an area per block of approximately 6,500 square kilometers, and the IFCs used here are symbolized as IFCe.Goo- Subscripts made up of a statement of length of side of a block rather than area would be less cum- bersome. I suspect however that circles 578 296 SYSTEMATIC ZOOLOGY may prove more useful than blocks as sam- pling units, especially if microdensitometers are used, which make the use of area necessary. Since preparation of our earlier study, a number of similar papers have been drawn to my attention or have been published. Munroe ( 1956 ) gave a fine account of the ecologic and zoogeographic features of Can- ada and an analysis of the insect faunas of the continent. Udvardy ( 1963 ) provided an excellent analysis of the bird faunas of North America. His methods differed from ours in that, rather than treating all species simultaneously, he grouped them by type of distribution pattern, and then prepared maps showing numbers of species geo- graphically, by type of pattern. By this method he was able to recognize the pres- ence of 17 primary faunas and 25 secondary ones. The methods used, while different in basic respects from those used here, could easily prove to be more useful. The following should be added to our earlier summary of coefficients of associa- tion (H & S: 131-132). Smith (1960) used the term "Faunistic Relation Factor" ( FRF ) for the Coefficient of Community, and Huheey ( 1965) called it a "Divergence Fac- tor" ( D ) , when subtracted from 100. Fager (1965) has devised a new coefficient in the form of 100 C/yJniUo - V2\/n2, where rii is less than rin. Long ( 1963 ) gives a review of coefficients and suggests use of an "average resemblance formula" first used by Kulczn- ski in 1927, and hsted in H & S, p. 132. In our earlier study (H & S: 128-129, 148) we pointed out that our work has been based on Webb's ( 1950 ) analysis of the mammals and herpetofauna of Texas and Oklahoma. The work of Ryan ( 1963 ) who improved on Webb's technique in analyz- ing the mammal faunas of Central America was not known to us at the time. Subse- quently, Huheey ( 1965 ) published an ac- count of further modifications of the tech- nique in the study of the herpetofauna of Illinois. Since the methods used in all of these are related, and because they are sim- ilar in principle, their comparison may be of interest, and I have attempted to do this briefly in the account following. Webb's (1950) method was to lay a grid of sample points at 100-mile intervals on a map of the area to be studied, to prepare a species checklist for each sample point, to compute Coefficients of Community between sample points and then plot these, to draw lines connecting CCs of equal value, providing a form of "contour map," and to consider "valleys" with CCs of 75% or more as "bio- geographic regions." A key point underly- ing Webb's analysis lies in the fact that he found CCs computed in a north-south plane to differ statistically from those com- puted in an east-west plane. Since he found that the east-west data gave most significant results, he accepted these in the preparation of his final map and rejected the north-south data. Webb deserves com- mendation for being first, to my knowledge, to devise a numerical technique for biogeo- graphic analyses in two dimensions. Ryan's ( 1963 ) analysis used a methodol- ogy only slightly different from that of Webb. Because of the unusual shape of the area studied, the grid of one portion of it was made up of points 100 kilometers ( about 62 miles ) to a side, of a second por- tion of it, 50 kilometers ( about 31 miles ) to a side. CCs, called "Similarity Values" by both Webb and Ryan, were calculated for both the north-south and east-west planes, and both sets of data were used in prepara- tion of the final contour maps, so far as I can determine. Ryan called the contour lines "isobiots." It is not clear whether Webb's 75% rule for biogeographic regions was used. Huheey's ( 1965 ) method differed to some degree from the preceding. It was to lay a grid of 20 miles to a side onto the area to be studied. Within each block of the grid a species checklist was prepared. For each of the four sides of all blocks, a Divergence Factor ( D ) was computed, where D = 100-CC; thus D is the complement of the Coefficient of Community. Huheey refers to the CC by Smith's ( 1960) term, "Faunis- tic Relation Factor," or FRF. The average 579 DISTRIBUTION OF NORTH AMERICAN MAMMALS 297 of the four Ds for each block was computed, this being the mean D for that block. Fin- ally, contour lines called "isometabases" were drawn around mean Ds of equal value. From the contour map herpetofaunal re- gions were described. Webb's and Ryan's methods, it will be observed, are essentially the same, differing only in distance between sample points and planes in which CCs are computed. Hu- heey's method, and the method used in this and in our earlier study differ considerably, though they seek identical ends through development of contour maps depicting faunistic change. I have been led to under- stand that still other techniques and refine- ments of those discussed here are in prepa- ration. For example, Valentine (1965) re- ported a study of the distribution of north- eastern Pacific molluscan distributions us- ing methods similar to those employed in this and in our earlier paper. It is apparent that there is need for a comparative testing of the several methods of analysis presently at hand, using the same basic materials in each. I plan to attempt such a study. Earlier (H & S: 129), we mentioned a partial testing of Webb's original method on the mammal fauna of North America. In view of the preceding, a brief account of the testing follows: Webb's method was followed exactly, except that the grid of sample points was placed on a northeast- southwest plane, giving better coverage of certain coastal areas. A number of varia- tions in the planes in which CCs were com- puted were attempted. These variations included: (1) computing and plotting CCs in the northeast-southwest plane only; (2) doing the same in the northwest-southeast plane only; (3) averaging adjacent CCs taken in both planes and plotting these; (4) plotting highest CCs only of pairs computed in both planes. We did not try Ryan's device of plotting all CCs taken in both planes. However, of the variants tested, none gave results that appeared anywhere near reasonable in terms of what we knew generally of the distribution of biogeo- graphic and ecologic zones. The variants used by Ryan and Huheey, however, ap- pear to work well on the basis of their evidence, and my conclusions apply in no way to their results. No attempt has been made to take into account the effects of altitude on mammal distributions. Dice, in his original study of biotic provinces (1943) described such ef- fects in terms of "life belts" and named a number of these. Kendeigh ( 1954 ) on the other hand did not see altitude as a con- founding factor in delimitation of biotic provinces. He wrote: "A mountain range may have several life zones represented on it, but only a single biotic province, pro- vided there is a similar tendency for specific or subspecific distinctiveness of the fauna in all the zones. The two concepts there- fore have quite different objectives." I have not been able to decide which of the two views applies in studies of the sort carried out here. It should be realized, how- ever, that the methods employed here are capable of analyzing the effects of altitude on distribution, and segregating altitudinal provinces, if they exist, given distribution maps of sufficient accuracy in the first place. The maps used here failed to show details of vertical distribution, and as a consequence this aspect of the problem has not proven solvable. An attempt was made to analyze altitud- inal distribution in a different way. Each species of mammal was given its life zone distribution, this infonnation being collated from a large number of sources, chiefly certain of the North American Fauna Series. The faunas of each of the life zones within provinces occurring in generally mountain- ous parts of the continent were treated as primary areas. Coefficients of Community were computed, and the results subjected to cluster analysis. The initial results were un- satisfactory, however, and because of this and because of the circularity of reasoning involved, the method was abandoned. No attempt has been made to relate the distribution of mammal areas to the distri- bution of other natural units, either physio- graphic, climatic, or vegetational, although 580 298 SYSTEMATIC ZOOLOGY a comparison of Figure 3 to maps showing the distribution of such features (e.g., Lobeck, 1948; Thomethwaite, 1948; Rowe, 1959; Shantz and Zon, 1923), shows that the relationship is very close. Summary 1. An earlier study demonstrated that range limits of North American terrestrial mammals were grouped, and that regions of faunistic homogeneity could as a conse- quence be identified. 2. The method used to identify such regions was to compute percentage of spe- cies whose ranges ended in blocks fifty miles on a side (IFCs), and to then fit isarithms. Topographic "valleys" in the map represented regions of faunistic homoge- neity, or "primary areas," and for 24 of these, species checklists were prepared. Tlie percentage of species common to pairs of primary areas (CCs) were computed, and the results subjected to cluster analysis, us- ing the method of weighted pair-groups with simple averages. This resulted in a matrix and dendrogram showing relation- ships and ordering of primary areas. Using a conversion of Preston's Resemblance Equation, a CC of 62.5% was considered critical. Primary areas with a CC lower than this were considered "mammal provinces." By this criterion, 22 mammal provinces grouped into nine superprovinces, four sub- regions, and one region were recognized. 3. Many more primary areas should have been derived from the IFC map. Starting with 86 primary areas and carrying out four sequential sets of cluster analyses leads to the conclusion that a minimum of from 33 to 35 mammal provinces occur in the con- tinent. These are mapped, named, and briefly described. For statistical reasons, the upper limit of Preston's critical value is raised to a CC of 65 7o. 4. Higher categories of mammal areas are derived by grouping the provinces on the matrix and dendrogram at appropriate mean CC levels. A mean CC of 42.5% gives 13 superprovinces; a mean CC of 25%, four subregions; a mean CC of 5%, one region (the Nearctic). These are mapped, named, and briefly described. In general, provinces are named as adjectives derived from geo- graphic place-names, superprovinces as nouns. Where possible the names of these and of regions and subregions are taken from the literature on a priority basis. 5. Approximately one-quarter of the mammal species of the continent also occur on nearby continental islands. Island faunas are always smaller than those of the adja- cent mainland and always show closest faunistic similarity to nearby provinces. 6. The methods used have the advantage of being relatively objective, repeatable, and well-suited to computer operations. The use of a successive scanning mode micro- densitometer may prove useful in the prepa- ration of accurate IFC maps. 7. Accounts of several techniques in bio- geographic analysis similar in aim and method to those used here have recently appeared. These are briefly compared. There is need for a critical testing and evaluation of these to determine which provides the best basis for further refine- ment. 8. While the methods used here are suit- able for analyzing the effects of altitudinal zonation on distribution, lack of requisite detail in the distribution maps now avail- able makes such analyses impractical. 9. There appears to be a high degree of correlation between the distribution of mammal areas and other kinds of natural areas. REFERENCES Allen, J. A. 1892. The geographical distribu- tion of North American mammals. Bull. Amer. Mus. Nat. Hist. 4:199-243. Cooper, J. G. 1859. On the distribution of the forests and trees of North America .... An- nual Report Board of Regents, Smithsonian Inst. : 246-280. Dice, L. R. 1943. The biotic provinces of North America. Univ. Michigan Press, Ann Arbor. Eager, E. W. 1963. Communities of Organisms. In Hill, M. N., The sea, vol. 2, p. 415-437. Hagmeier, E. M., and C. D. Stxjlts. 1964. A 581 DISTRIBUTION OF NORTH AMERICAN MAMMALS 299 numerical analysis of the distributional patterns of North American mammals. Systematic Zool. 13:125-155. HuHEEY, J. E. 1965. A mathematical method of analyzing biogeographical data. 1. Herpeto- fauna of Illinois. Amer. Midi. Nat. 73:490-500. Jaccard, D. 1902. Gezetze der Pflanzenverthei- lung in der Alpinen Region. Flora 90:349-377. Kendeigh, S. C. 1954. History and evaluation of various concepts of plant and animal communi- ties in North America. Ecology 35:152-171. 1961. Animal ecology. Prentice-Hall, Engle- wood Cliffs. KuLCZYNSKi, S. 1937. Die pflanzenassoziationen der Pieninen. Bull. Internat. Acad. Polish Sci. Lett. 1927 (2):57-203. LoBECK, A. K. 1948. Physiographic provinces of North America. Geographical Press, New York. Long, C. A. 1963. Mathematical fonnulas ex- pressing faunal resemblance. Trans. Kansas Acad. Sci. 66:138-140. Miller, A. H. 1951. An analysis of the distribu- tion of the birds of Cahfomia. Univ. Calif. Publ. Zool. 50:531-644. Miller, C. S., F. G. Parsons, I. L. Kofsky. 1964. Simplified two-dimensional microdensitometry. Nature 202:1196-1200. MuNROE, E. 1956. Canada as an environment for insect life. Canadian Entomologist 88:372- 476. Preston, R. W. 1962. The canonical distribu- tion of commonness and rarity. Ecology 43:185- 215, 410-432. RowE, J. S. Forest regions of Canada. Bull. Ca- nadian Dept. Northern Affairs and Natural Re- sources 123, p. 1-71. Ryan, R. M. 1963. The biotic provinces of Cen- tral America as indicated by mammalian distri- bution. Acta Zoologica Mexicana 6:1-55. Savage, J. M. 1960. Evolution of a peninsular herpetofauna. Systematic Zool. 9:184-212. Shantz, H. L., and R. Zon. 1924. Atlas of American agriculture. Pt. 1, Sect. E, Natural Vegetation. U.S.D.A., Bureau Agric. Economics, p. 1-29. Simpson, G. G. 1943. Mammals and the nature of continents. Amer. J. Sci. 241:1-31. 1964. Species density of North American Re- cent mammals. Systematic Zool. 13:57-73. Smith, H. M. 1960. An evaluation of the biotic province concept. Systematic Zool. 9:41-44. SoKAL, R. R., and P. H. E. Sneath. 1963. Principles of numerical taxonomy. Freeman, San Francisco. Thornethwaite, C. W. 1948. An approach to a rational classification of climate. Geogr. Rev. 38:55-94. Udvardy, M. D. F. 1963. Bird faunas of North America. Proc. 13th Internat. Ornithol. Con- gress: 1147-1167. Valentine, J. 1965. Numerical analysis of north-eastern Pacific molluscan distributions. Paper read to the Pacific Division of A.A.A.S., Riverside, California, June 24. Van Dyke, E. C. 1939. The origin and distribu- tion of the coleopterous insect fauna of North America. Proc. 6th Pacific Science Congress, 4:255-268. Wallace, A. R. 1876. The geographical distri- bution of animals. 2 vols. Macmillan, London. Webb, W. L. 1950. Biogeographic regions of Texas and Oklahoma. Ecology 31:426-433. EDWIN M. HAGMEIER is a member of the Department of Biology at the University of Vic- toria, Victoria, B. C, Canada. The work was sup- ported by the University and by the National Research Council. Mr. Peter Darling of the Uni- versity Computer Center has played an important part in the analyses reported here. The illustrations were prepared by Miss Elizabeth Swemle. The writer extends his thanks to both, and dedicates the study to S.E.H. 582 LITERATURE CITED Adams, L., and S. G. Watkins 1967. Anniili in tooth cementum indicate age in California ground squirrels. Jour. Wildlife Mgt., 31:836-839. Allee, W. C, a. E. Emerson, O. Park, T. Park, and K. P. Schmidt 1949. Principles of animal ecology. W. B. Saunders, xii+837 pp. Allex. G. M. 1939. A checklist of African mammals. Bull. Mus. Comp. Zool., 83:1-763. Allex, H. 1864. Monograph of the bats of North America. Smithsonian Misc. Coll., 7(165): xxiii-f-1-85. Anderson, S. 1961. Mammals of Mesa Verde National Park, Colorado. Univ. Kansas Publ., Mus. Nat. Hist., 14:29-67. Anderson, S., and J. K. Jones, Jr. (eds.) 1967. Recent mammals of the world — a synopsis of families. Ronald Press, viii+453 pp. Andrewartha, H. G., and L. C. Birch 1954. The distribution and abundance of animals. Univ. Chicago Press, xv+782 pp. Anthony, H. E. 1928. Field book of North American mammals. G. P. Putnam's Sons, xxvi+674 pp. Asdell, S. a. 1964. Patterns of mammalian reproduction. Comstock Publishing Associates, 2nd ed., viii-f670 pp. Baker, R. H., and J. K. Greer 1962. Mammals of the Mexican state of Durango. Publ. Mus. Michigan State Univ., Biol. Sen, 2:25-154. Beddard, F. E. 1902. Mammalia. Macmillan, xiii-t-605 pp. Black, C. C. 1963. A review of the North American Tertiary Sciuridae. Bull. Mus. Comp. Zool., 130:109-248. Blackvvelder, R. E. 1967. Taxonomy — a text and reference book. John Wiley and Sons, xiv+698 pp. Blair, W. F. 1968. Introduction. Pp. 1-20, in Vertebrates of the United States (W. F. Blair et al), McGraw-Hill, ix-t-616 pp. BOURLIERE, F. 1954. The natural history of mammals. Alfred A. Knopf, xxi+363+xi pp. Brown, J. H. 1968. Adaptation to environmental temperature in two species of woodrats, Neotoma cinerea and N. albigula. Misc. Publ. Mus. Zool., Univ. Michigan, 135:1-48. Burt, W. H., and R. P. Grossenheider 1964. A field guide to the mammals. Houghton-Mifflin, xxiii+284 pp. Cabrera, A. 1958. Catalogo de los mamiferos de America del Sur [I. Metatheria, Unguiculata, Carnivora]. Revista Mus. Argentino Cien. Nat. "Bernardino Rivadavia," Cien. Zool., 4(1) :iv+l-307. 1961. Catalogo de los mamiferos de America del Sur [II. Sirenia, Perissodactyla, Artiodactyla, Lagomorpha, Rodentia, Cetacea]. Revista Mus. Argentino Cien. Nat. "Bernardino Rivadavia," Cien. Zool., 4(2):.xxii+309-732. Carlqihst, S. 1965. Island Life. . . . Natural History Press, viii4-451 pp. COCKRUM, E. L. 1962. Introduction to mammalogy. Ronald Press, viii-f 455 pp. Crant)all, L. S. 1964. The management of wild mammals in captivity. Univ. Chicago Press, xv+761 pp. Darlington, P. J., Jr. 1957. Zoogeography. John Wiley and Sons, xi+675 pp. Davis, D. D. 1964. The giant panda — a morphological study of evolutionary mechanisms. Fieldiana- ZooL, Mem., 3:1-339. Davis, D. E., and F. B. Golley 1963. Principles in mammalogy. Reinhold Publishing Co., .xiii+335 pp. Dawson, M. R. 1958. Later Tertiary Leporidae of North America. Univ. Kansas Publ, Paleo. Contrib. (Vertebrata), 6:1-75. 583 DeKay, J. E. 1842. Zoology of New York . . . [Part 1. Mammalia]. Albany, xiii4-146 pp. Dice, L. R. 1931. The relation of mammalian distribution to vegetation types. Sci. Monthly, 33: 312-317. 1952. Natural communities. Univ. Michigan Press, x-f 547 pp. Ellerman, J. R. 1940. The families and genera of living rodents. Volume 1. Rodents other than Muridae. British Mus., x.\vi4-689 pp. 1941. The families and genera of living rodents. Volume 2. Family Muridae. British Mus., xii-(-690 pp. 1948. The families and genera of living rodents. Volume 3 [additions and corrections]. British Mus., v+210 pp. Ellerman. 1. R.. and T. C. S. Morrison-Scott 1951. Checklist of Palaearctic and Indian mammals 1758 to 1946. British Mus., 810 pp. Enders, a. C. (ed.) 1963. Delayed implantation. Rice Univ. and Univ. Chicago Press, x+318 pp. Ewer, R. F. 1968. Ethology of mammals. Plenum Press, xiv+418 pp. Flower, W. H., and R. Lydekker 1891. An introduction to the study of mammals living and extinct. Adam and Charles Black, xvi+763 pp. Foster, J. B. 1965. The evolution of the mammals of the Queen Charlotte Islands, British Columbia. Occas. Papers British Columbia Prov. Mus., 14:1-130. Gill, T., and E. Coues 1877. Material for a bibliography of North American mammals. Pp. 951-1081, in Coues and Allen, U.S. Geol. Surv. Territories, ll:x-]-l-1091. Grasse, P.-P. (ed.) 1955. Mammiferes: les ordres, anatomic, ethologie, systematique. Traite de zoologie, 17:1-2300. Hall, E. R. 1926. Changes during growth in the skull of the rodent Otospermophilus grammurus beecheyi. Univ. California Publ. Zool, 21:355-404. 1946. Mammals of Nevada. Univ. California Press, .xi-}-710 pp. Hall, E. R., and K. R. Kelson 1959. The mammals of North America. Ronald Press, l:xxx+ 1-546+ 79 and 2:viii +547-1083+79. Hamilton, W. J., Jr. 1939. American mammals. . . . McGraw-Hill, xii+434 pp. Hamilton, W. J., Ill 1962. Reproductive adaptations of the red tree mouse. Jour. Mamm., 43:486-504. Harper, F. 1927. The mammals of the Okefinokee Swamp region of Georgia. Proc. Boston Soc. Nat. Hist., 38:191-396, pis. 4-7. Hesse, R., W. C. Allee, and K. P. Schmidt 1937. Ecological animal geography. John Wiley and Sons, xiv+597 pp. HiBBARD, C. W. 1950. Mammals of the Rexroad Formation from Fox Canyon, Kansas. Contrib. Mus. Paleo., Univ. Michigan, 8:113-192. HiLDEBRAND, M. 1959. Motions of the running cheetah and horse. Jour. Mamm., 40:481-495. Hill, W. C. O. 1953. Primates — comparative anatomy and taxonomy. I — Strepsirhini. Univ. Press, Edinburgh, xxiii+798 pp. Hooper, E. T. 1952. A systematic review of the harvest mice (genus Reithrodontomys) of Latin America. Misc. Publ. Mus. Zool., Univ. Michigan, 77:1-255. Howell, A. B. 1926. Anatomy of the woodrat. Monogr. Amer. Soc. Mamm., l:x+ 1-225. 1930. Aquatic mammals. . . . Charles C. Thomas, xii+338 pp. Huxley, J. 1932. Problems of relative growth. Dial Press, xix+276 pp. 1943. Evolution — the modern synthesis. Harper and Brothers, 645 pp. Jackson, H. H. T. 1928. A taxonomic revision of the American long-tailed shrews ( genera Sorex and Microsorex) . N. Amer. Fauna, 51:vi+l-238. 1961. Mammals of Wisconsin. Univ. Wisconsin Press, xii+504 pp. Johnson, D. H., M. D. Bryant, and A. H. Miller 1948. Vertebrate animals of the Providence Mountains area of California. Univ. Cali- fornia Publ. Zool., 48:221-376. 584 JoxEs, J. K., Jr. 1964. Distribution and taxonomy of mammals of Nebraska. Univ. Kansas Publ., Mus. Nat. Hist., 16:1-356. Kayser, C. 1961. The physiology of natural hibernation. Pergamon Press, vii-f 325 pp. Klingener, D. 1964. The comparative myology of four dipodoid rodents (genera Zapus, Napaeozapus, Sicista, and Jaculus). Misc. Publ. Mus. Zool., Univ. Michigan, 124:1-100. KURODA, N. 1940. A monograph of the Japanese mammals. Sanseido Co., Ltd., 311 pp. Lack, D. 1954. The natural regulation of animal numbers. Clarendon Press, viii-|-343 pp. Latjrie, E. M. O., and T. E. Hill 1954. List of land mammals of New Guinea, Celebes, and adjacent islands 1758-1952. British Mus., 175 pp. Layxe, J. H. 1960. The growth and development of young golden mice, Ochrotomijs nuttalli. Quart. Jour. Florida Acad. Sci., 23:36-58. 1966. Postnatal development and growth of Peromyscus floridanus. Growth, 30:23-45. LiDicKER, W. Z., Jr. 1960. An analysis of intraspecific variation in the kangaroo rat Dipodomys merriami. Univ. California Publ. Zool., 67:125-218. Lymax, C. p., and A. R. Da we (eds. ) 1960. Mammalian hibernation. Bull. Mus. Comp. Zool., 124:1-549. Lyxe, a. G., and A. M. W. Verhagex 1957. Growth of the marsupial Trichosiirus vulpecula, and a comparison with some higher mammals. Growth, 21 : 167-195. Mac Arthur, R. H., and E. O. Wilsox 1967. The theory of island biogeography. Monogr. Pop. Biol., Princeton Univ. Press, l:xi+ 1-203. Matthew, W. D. 1939. Climate and evolution. Spec. Publ. New York Acad. Sci., l:xii+ 1-223. Mayb, E. 1963. Animal species and evolution. Harvard Univ. Press, xiv+797 pp. 1969. Principles of systematic zoology. McGraw-Hill, xi+428 pp. Mayr, E., E. G. Linsley, and R. L. Usixger 1953. Methods and principles of systematic zoology. McGraw-Hill, ix-|-336 pp. Me.\rxs, E. a. 1907. Mammals of the Mexican boundary of the United States. Bull. U.S. Nat. Mus., 56 :.xv+ 1-530. Miller, G. S., Jr. 1899. Preliminary list of the mammals of New York State. Bull. New York State Mus., 6:273-390. Miller, G. S., Jr., and R. Kellogg 1955. List of North American Recent mammals. Bull. U.S. Nat. Mus., 205 :xii-f 1-954. Osgood, W. H. 1909. Revision of the mice of the American genus Peromyscus. N. Amer. Fauna, 28: 1-285. Packard, R. L. 1960. Speciation and evolution of the pygmy mice, genus Baiomys. Univ. Kansas Publ., Mus. Nat. Hist., 9:579-670. Pallas, P. S. 1778. Novae species quadrupedum e glirium ordine. . . . Erlangae, 388 pp. Palmer, R. S. 1954. The mammal guide. Doubleday and Co., 384 pp. Pearsox, O. p. 1958. A taxonomic revision of the rodent genus Phyllotis. Univ. California Publ. Zool., 56:391-496. Petersox, R. L. 1966. The mammals of eastern Canada. Oxford Univ. Press, xxxii+465 pp. POCOCK, R. I. 1924. The external characters of the pangolins (Manidae). Proc. Zool. Soc. London, pp. 707-723. Rheixgold, H. L. (ed. ) 1963. Maternal behavior in mammals. John Wiley and Sons, viii+349 pp. RlXKER, G. C. 1954. The comparative myology of the mammalian genera Sigmodon, Oryzomys, Neotoma, and Peromyscus (Cricetinae), with remarks on their intergeneric relationships. Misc. Publ. Mus. Zool., Univ. Michigan, 83:1-124. ROMER, A. S. 1966. Vertebrate paleontology. Univ. Chicago Press, 3rd ed., ix-f-468 pp. 585 Rowlands, I. W. (ed.) 1966. Comparative biology of reproduction in mammals. Symp. Zool. Soc. London, 15 :xxi+ 1-559. SCHALLER, G. B. 1963. The mountain gorilla. . . . Univ. Chicago Press, xvii4-431 pp. Searle, a. G. 1968. Comparative genetics of coat colour in mammals. Logos Press, xii+308 pp. Severinghaus, C. W. 1949. Tooth development and wear as criteria of age in white-tailed deer. Jour. Wild- life Mgt., 13:195-216. Shotwell, J. a. 1958. Evolution and biogeography of the aplodontid and mylagaulid rodents. Evolu- tion, 12:451-484. Simpson, G. G. 1945. The principles of classification and a classification of the mammals. Bull. Amer. Mus. Nat. Hist., 85 :xvi-j- 1-350. 1961. Principles of animal taxonomy. Columbia Univ. Press, xii+247 pp. Struhsaker, T. T. 1967. Behavior of vervet monkeys ( Cercopithecus aethiops). Univ. California Publ. Zool, 82:1-74. Thompson, D'A. W. 1942. On growth and form. Cambridge Univ. Press, 2nd ed., 1116 pp. TroughtOiN, E. 1965. Furred animals of Australia. Halstead Press, 8th ed., xxxii+376 pp. Udvardy, M. D. F. 1969. Dynamic zoogeography. . . . Van Nostrand Reinhold Co., xviii+445 pp. Van den Brink, F. H. 1967. A field guide to the mammals of Britain and Europe. Collins, 221 pp. Vaughan, T. a. 1959. Functional morphology of three bats: Eumops, Myotis, Macrotus. Univ. Kan- sas Publ., Mus. Nat. Hist., 12:1-153. Walker, E. P., and associates 1964. Mammals of the world. Johns Hopkins Press, 3 vols. Weber, M. 1927-28. Die Saugetiere. . . . Fischer, 2nd ed., 2 vols. (l:xv+l-444, 2:xxiv-i-l-898). Wight, H. M., and C. H. Conaway 1962. A comparison of methods for determining age of cottontails. Jour. Wildlife Mgt., 26:160-163. Wilson, R. W. 1960. Early Miocene rodents and insectivores from northeastern Colorado. Univ. Kan- sas Publ., Paleo. Contrib. ( Vertebrata ) , 7:1-92. Wood, A. E. 1959. Are there rodent suborders? Syst. Zool., 7:169-173. Wynne-Edwards, V. C. 1962. Animal dispersion in relation to social behavior. Oliver and Boyd, xi+653 pp. Young, J. Z. 1957. The life of mammals. Oxford Univ. Press, xv-f 820 pp. Zittel, K. a. 1891-93. Handbuch der Palaeontologie. Band 4, Vertebrata (Mammalia). R. Olden- bourg, xi-}-799 pp. 586 Harvard MCZ Lib/af; 3 2044 062 537 139 1