Norntatesamerican MUSEUM PUBLISHED by the AMERICAN MUSEUM of NATURAL HISTORY CENTRAL PARK WEST at 79TH STREET, NEW YORK, N.Y

NorntatesAMERICAN MUSEUM PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 3050, 56 pp., 25 figures, 15 tables July 28, 1992

A Revision of the South American Species of Sigmodon (Mammalia: Muridae) with Notes on Their Natural History and Biogeography

ROBERT S. VOSS'

CONTENTS Abstract ...... 2 Resumen ...... 2 Introduction ...... 3 Acknowledgments ...... 4 Materials and Methods ...... 4 Specimens ...... 4 Measurements ...... 4 Age Criteria ...... 5 Karyotypes ...... 5 Multivariate Statistics ...... 5 Gazetteer ...... 5 The Genus ...... 11 Sigmodon Say and Ord ...... 11 Phylogenetic Relationships ...... 14 Variation Within and Among the South American Species ...... 15 Qualitative External Characters ...... 15 External Dimensions ...... 16 Qualitative Craniodental Characters ...... 16 Craniodental Dimensions ...... 21 Species Accounts ...... 26 Sigmodon hispidus Say and Ord ...... 26 Sigmodon alstoni (Thomas) ...... 30 Sigmodon peruanus J. A. Allen ...... 34 Sigmodon inopinatus Anthony ...... 36

Natural History ...... 37 SpeciesSpecies~~~~~~DescriptionsDsptos...... 373 Summary, Comparisons, and Discussion ...... 46 Biogeography ...... 47 Intercontinental Dispersal ...... 47 Nonforest Mammal Faunas and South American Paleohabitats ...... 49 References ...... 50 Appendix: List of Measured Specimens ...... 55

ABSTRACT Cotton rats, members of the muroid rodent ge- arterial supply. Known from only two localities in nus Sigmodon, range from Iowa and Nebraska to the high Andes of western Ecuador, inopinatus is northeastern Brazil and Peru. Seven valid species strikingly divergent from all other South American are currently recognized in North and Central species in rostral and interorbital morphology. America, but the 12 nominal species known from Available natural history information suggests South America have received no critical taxonom- that South American species of Sigmodon, like ic attention to date. This revision, based on mor- their North American congeners, inhabit nonfo- phological character data from almost 1300 mu- rest environments, are active both by day and at seum specimens, demonstrates the existence offour night, construct runways, and chiefly consume the diagnosable species in the South American fauna. green leaves and stems ofgrasses and forbs. Com- Specimens with ungrooved upper incisors from parisons with sympatric species of Zygodontomys Colombia and northwestern Venezuela closely re- and Holochilus suggest that cotton rats are the only semble typical hispidus in all qualitative and quan- strictly terrestrial (nonaquatic and nonarboreal) and titative comparisons, including karyotypes, and are predominantly herbivorous muroids in the trop- provisionally referred to that species. Specimens ical grassland ecosystems they inhabit. Apparent- ofalstoni, a smaller cotton rat with deeply grooved ly, other ecologically similar New World muroid upper incisors, distinctive craniodental propor- taxa occupy temperate latitudes. tions, and very divergent karyotypes, have been Since the genus Sigmodon is restricted to open found in sympatry with hispidus at several local- habitats, the distribution and relationships ofcot- ities; this species ranges from extreme northeast- ton rat species presumably reflect the Late Ce- ern Colombia throughout most ofVenezuela, and nozoic history of forested and unforested land- is also known from scattered localities in Guyana, scapes in North and South America. Some Surinam, and Brazil north of the Amazon River. historical-biogeographic implications of the dis- A third species, peruanus, is restricted to the Pa- tributional data reported here are considered brief- cific lowlands and foothills ofwestern Ecuador and ly, but phylogenetic information is required to dis- northwestern Peru; it differs from hispidus and criminate among alternative scenarios. The other congeners in numerous qualitative and sampling design ofcladistic studies to obtain such quantitative aspects of craniodental morphology information is discussed. including a uniquely primitive pattern ofstapedial

RESUMEN Los roedores muroides del genero Sigmodon al- cuantitativas (incluyendo cariotipos), y se refieren canzan desde Iowa y Nebraska (en los Estados provisionalmente a la misma especie. Especi- Unidos) hasta Peru y Surinam. Siete especies vil- menes de alstoni, una rata mas pequenia con surcos idas se reconocen actualmente en Am6rica del profundos en los incisivos superiores, propor- Norte y America Central, pero las doce especies ciones craniodentales distintivas, y cariotipos muy nominales conocidas de America del Sur no han divergentes, han sido encontrados simpiatrica- recibido atenci6n taxonomica critica hasta la fe- mente con hispidus en varias localidades; esta es- cha. Esta revision, basada en los caracteres mor- pecie alcanza desde el extremo noreste de Colom- fologicos de casi 1300 especimenes, demuestra que bia, por todas partes de Venezuela, hasta unas existen cuatro especies diagnosticables en la fauna localidades dispersas en Guyana, Surinam, y Bra- de America del Sur. Los especimenes sin surcos sil al norte del rio Amazonas. Una tercera especie, en los incisivos superiores, de Colombia y del no- peruanus, esta restringida al litoral Pacifico y las roeste de Venezuela, se asemejan mucho al tipico estribaciones Andinas adyacentes del Ecuador y el hispidus en todas las comparaciones cualitativas y noroeste de Peru; se puede distinguir peruanus de 1 992 VOSS: SOUTH AMERICAN SIGMODON 3 hispidus y de las otras especies congenericas por principalmente herbivoros en las sabanas tropi- muchos caracteres cualitativos y cuantitativos cales que ellos habitan. Aparentamente, otros mu- craniodentales, incluyendo un patr6n uinico de la roides del Nuevo Mundo semejantes a Sigmodon circulaci6n estapedial. Conocido solamente de dos en estos aspectos ecologicos se encuentran en lat- localidades en los altos Andes del Ecuador, ino- itudes templadas. pinatus es muy divergente de las otras especies en Como el genero Sigmodon se encuentra sola- la morfologia del rostro y de la regi6n interorbital. mente en los habitats abiertos, la distribucion y La informacion disponible sobre la historia nat- las relaciones filogeneticas de las especies del mis- ural sugiere que las especies suramericanas de Sig- mo por presuncion reflejan la historia cuaternaria modon, como las de America del Norte, se en- de los bosques y las sabanas en America del Norte cuentran en los habitats abiertos (no boscosos), y del Sur. Algunos significados historicos y bio- son activos de dia y de noche, construyen pistas geograficos de los datos distribucionales propor- o caminos, y consumen principalmente los tejidos cionados aqui se consideran brevemente, pero son verdes de gramineas y otras hierbas. Las compar- necesarias reconstrucciones filogeneticas para pro- aciones con dos taxa simpatricos, Zygodontomys bar los escenarios alternativos. El disefio de mues- y Holochilus, sugieren que las especies de Sig- treo para comparaciones morfologicas, citogene- modon son los 'unicos roedores muroides estric- ticas, o moleculares para obtener dichas tamente terrestres (no acuiticos y no boscosos) y reconstrucciones tambien se discute.

INTRODUCTION Species of the muroid rodent genus Sig- ica: 9 with ungrooved upper incisors were modon, commonly known as cotton rats, are referred to Sigmodon, and 3 with grooved abundant in much of the continental United teeth were classified as Sigmomys following States, especially in the prairies of the south- Thomas (1901, 1914). Hershkovitz (1955) ernmost plains states, in southwestern desert thought that all cotton rats (whether smooth- grasslands and mountain meadows, in the toothed or groove-toothed) might be conspe- open pine forests and fallow agricultural fields cific, but Cabrera (1961) only synonymized ofthe southeast, and generally wherever grass Sigmomys with Sigmodon while retaining is a significant component of the local veg- hispidus (including all the smooth-toothed etation. Further south, species of Sigmodon South American forms) and alstoni (includ- also occur in tropical lowland savannas and ing all the groove-toothed taxa) as valid spe- in the high-altitude Andean grasslands known cies. Although both Handley (1972) and Reig as paramos. Since Say and Ord's (1825) dis- (1986) recognized four South American spe- covery ofthe type species, S. hispidus, a total cies ofSigmodon (including Sigmomys), most of69 nominal taxa ofRecent Sigmodon have checklists (e.g., Corbet and Hill, 1980; Hon- been described. Modern concepts of species acki et al., 1982) have followed Cabrera limits and relationships among North Amer- (1961). Unfortunately, none ofthese diverse ican representatives of the genus (summa- classifications are accompanied by analyses rized by Hall, 1981) are largely the outgrowth or substantive discussions of character data, of seminal papers by Baker (1969) and Zim- and it is therefore impossible to evaluate the merman (1970) that provoked numerous cy- empirical basis for any taxonomic alternative tological studies ofpopulations in the United without extensive recourse to specimens. States and Mexico. Hoffmeister's (1986) syn- This report provides the first critical re- thesis of morphological, karyological, and vision of South American Sigmodon, sum- ecological information for the Arizona spe- marizes anatomical character variation with- cies is an exemplary result ofcareful system- in and among species, maps geographic atic research in an exceptionally well-col- ranges, and includes such natural history in- lected region, but elsewhere systematic formation as can be gleaned from fieldnotes treatments of Sigmodon have usually been and the scant published literature. Taken to- perfunctory. For the South American taxa, gether, these data reveal endemic ecomor- there is no revisionary literature at all. phological distributions that must be accom- The nomenclatural history of Neotropical modated in historical reconstructions of cotton rats was reviewed by Tate (1932), who cotton rat biogeography. Alternative biogeo- listed 12 nominal species from South Amer- graphic scenarios are briefly considered, and 4 AMERICAN MUSEUM NOVITATES NO. 3050 efforts to obtain the phylogenetic information Castellanos, Isa Torrealba, Carlos Gauveca, required to test them are encouraged. Carlos Lasso, and Pablo Morales. Collecting and export permits were kindly provided by ACKNOWLEDGMENTS the Venezuelan Ministerio del Ambiente y The research described herein is based on de los Recursos Naturales Renovables. data obtained from museum specimens. The splendid line drawings and cranial Those specimens, together with the labels and photographs that illustrate this report reflect fieldnotes that accompany them, represent the the artistic talents ofPatricia Wynne and Pe- efforts of numerous individuals and institu- ter Goldberg, respectively. Bill Barnett and tions committed to a worldwide inventory of Peiling Fong are responsible for the SEM im- biological diversity. Because the personal and ages of teeth, and Erika Bach provided pre- financial commitments that make such col- liminary analyses of karyotypes. lection resources available for research are Phil Myers, Bruce Patterson, and Nancy increasingly challenged by the demands of Simmons read the manuscript and provided other societal priorities, I gratefully acknowl- helpful criticisms; their generous contribu- edge the important contributions of the fol- tion oftime and effort to improve this report lowing institutions and their professional is sincerely appreciated. staffs: the American Museum ofNatural His- tory, New York (AMNH); the Natural His- MATERIALS AND METHODS tory Museum, London (BMNH); the Field Museum of Natural History, Chicago SPECIMENS (FMNH); the Museum of Comparative Zo- ology at Harvard University, Cambridge The 1273 specimens examined in the course (MCZ); the Museo de Historia Natural La of this research are deposited in the museum Salle, Caracas (MHNLS); the Museum Na- collections listed with their standard abbre- tional d'Histoire Naturelle, Paris (MNHN); viations in the Acknowledgments, above. In- the Museum ofVertebrate Zoology, Berkeley formation provided in the species accounts (MVZ); the Rijksmuseum van Natuurlijke that follow, under the heading Specimens Ex- Historie, Leiden (RMNH); the Royal Ontario amined, consists ofthe number ofspecimens Museum, Toronto (ROM); the Texas Co- from which data were recorded (listed by mu- operative Wildlife Collection, College Sta- seum collection) for each locality. Catalog tion (TCWC); the University of Michigan numbers are provided in the Appendix for Museum of Zoology, Ann Arbor (UMMZ); specimens whose measurements are sum- and the National Museum of Natural His- marized by statistics tabulated in the text. tory, Washington D.C. (USNM). Victoria L. Mayer, a former Research As- MEASUREMENTS sistant in the Department of Mammalogy, Total length (TL), length of tail (LT), and provided much valuable assistance at an ear- length of the hindfoot (HF) were recorded ly stage of this research: she reorganized and from specimen labels, but I remeasured dried recurated our Sigmodon collections, com- hindfeet (including the claws) whenever pos- piled locality information for the gazetteer sible; length of head-and-body (HBL) was from fieldnotes in our archives, and spent calculated by subtracting LT from TL. Tab- long hours in the library tracking down ob- ulated means of external measurements are scure references for the bibliography. With- from specimens collected by field workers who out her help this report would have been sub- are known or presumed to have followed the stantially longer in preparation. American measurement conventions de- It is a pleasure to thank the Museo de His- scribed by Hall (1981). toria Natural La Salle and its director, Giu- The following craniodental measurements seppe Colonnello, for logistical support while were determined with dial calipers as de- I was in Venezuela in 1986 and 1987. My scribed and illustrated by Voss (1991): con- fieldwork there would have been much less dylo-incisive length (CIL); length ofdiastema productive and enjoyable without the enthu- (LD); occlusal length ofmolars (LM); breadth siastic assistance of Helen Kafka, Hernan of Ml (BM 1); length of incisive foramen 1 992 VOSS: SOUTH AMERICAN SIGMODON 5

(LIF); breadth of rostrum (BR); breadth of were executed under the assumptions and fol- palatal bridge (BPB); breadth of zygomatic lowing the procedures explained by Voss plate (BZP); least interorbital breadth (LIB); (1991). General-size coefficients were esti- and depth of incisor (DI). Breadth of the in- mated as the elements ofthe first eigenvector cisor tips (BIT) was measured across the en- of the pooled within-group covariance ma- ameled bands of both upper teeth, and zy- trix; the normalized coefficients of general- gomatic breadth (ZB) across the squamosal size-invariant group differences were com- zygomatic roots, as illustrated by Voss (1988). puted as described by Voss and Marcus (in press). The interpretation of general-size co- AGE CRITERIA efficients in terms of ontogenetic allometry, Species of Sigmodon do not exhibit the and the growth-invariance of the residual marked ontogenetic transformations ofpost- within- and between-sample variation is dis- natal morphologythatpermit operational def- cussed extensively by Voss et al. (1990) and initions of age categories in other muroids: by Voss and Marcus (in press). the pelage of even the smallest juveniles closely resembles that of large adults, and GAZETTEER molar occlusal morphology is not much af- South American localities from which Sig- fected by wear. Unfortunately, the criteria for modon species have been collected are listed age determination in S. hispidus proposed by alphabetically below. Numbering is sequen- Chipman (1965) cannot be determined by tial only within countries; italic type identi- convenient, nondestructive methods from fies subordinate political units (departments, standard museum preparations. I reserve the states, provinces, etc.), and boldface identi- term "juvenile" for animals with incomplete- fies locality names as cited in the text of this ly erupted or unworn third molars, and "old report. Latitude (degrees and minutes N or adult" for specimens with molars worn below S) and longitude (degrees and minutes W) are the widest part ofthe crowns. All other spec- given in parentheses without punctuation imens, regardless of apparent age differences (e.g., 0230N/5600 for 2°30'N, 56°00'W). Most among them, are considered "adult" in the of the geographical coordinates provided present report. herein agree with those listed in gazetteers published by the Office of Geography, U.S. KARYOTYPES Department ofthe Interior (cited as USBGN by Paynter and his colleagues, e.g., Paynter Karyotypic information newly reported and Traylor, 1991), but I have not hesitated here was obtained from metaphase prepa- to use other sources that appeared more au- rations ofbone marrow cells. Specimens were thoritative for some areas (e.g., Handley, processsed in the field, initially by subcuta- verbatim neous injections of live yeast cultures, fol- 1976). Elevations are reported (in lowed after 12 to 24 hours by minor modi- feet or meters) from specimen labels, field- fications of standard in vivo colchicine/ notes, or cited secondary sources. Localities hypotonic citrate methods (Patton, 1967). Cell are mapped in figures 1 and 2. suspensions were fixed in Carnoy's solution BRAZIL and frozen in liquid nitrogen for transport to 1. Para, Serra do Tumucumaque, 12 km da the laboratory where they were flame-dried Guiana Holandeza (ca. 0230N/5600). on glass slides and stained with Giemsa. Only 2. Roraima, Frechal (= Frexal; 0350N/ modal diploid counts are provided from our 6032). incomplete analyses. 3. Roraima, Rio Cotinga, Limo (0356N/ 6030); elevation ca. 1500 ft (Tate, 1939). MULTIVARIATE STATISTICS 4. Roraima, Serra da Lua (0215N/6045); All multivariate statistics reported herein elevation ca. 300 m (Paynter and Tray- were computed from measurement values lor, 1991). This locality is presumed syn- transformed to their natural logarithms. onymous with the "Moon Mountains" Cluster analyses based on Mahalanobis dis- ofearly collectors who thought they were tances, and principal components analyses, in British Guiana (Guyana). 6 AMERICAN MUSEUM NOVITATES NO. 3050

Fig. 1. Collection localities in Colombia, Venezuela, the Guianas, and Brazil. Densely stippled land areas are those above 1500 m elevation.

COLOMBIA 4. Cesar, Pueblo Bello (1024N/7343); ele- 1. Boyaca, Muzo (0532N/7406); elevation vation 1067 m (Hershkovitz, 1960). 1240 m (Paynter and Traylor, 1981). 5. Cundinamarca, Bogota (0436N/7405), 2. Cesar, El Mamon (1037N/7333), Pueblo including "Plains of Bogota" and "Que- Viejo (?1032N/7325), and San Sebastian brada Seca"; elevation 2590 m (Paynter (1034N/7336) in the Sierra Nevada de and Traylor, 1981). Santa Marta; elevations recorded on W. 6. Cundinamarca, 10 km E Boqueron W. Brown's specimen tags are 8000-9000 (0416N/7433); elevation (at Boqueron) ft at El Mamon, 8000 ft at Pueblo Viejo, ca. 400 m (Paynter and Traylor, 1981). and 6600 ft at San Sebastian. 7. Cundinamarca, Caparrapi (0521 N/ 3. Cesar, El Salado (1022N/7329), NW of 7430), Los Volcanes; elevation ca. 1250 Valledupar (1029N/7315); elevation 430 m (from map). m (Hershkovitz, 1960). 8. Guajira, La Concepcion (ca. 1103N/ 1992 VOSS: SOUTH AMERICAN SIGMODON 7

7327); elevation ca. 800 m (Paynter and 12. Huila, Valle de Suaza (mouth of Rio Traylor, 1981). Suaza is at 0210N/7541, but most sub- 9. Guajira, Las Marimondas (1052N/7243), localities probably to SW; the only one in the Sierra de Perij'a; elevation 1000 m listed by USBGN is Marmato at 0155N/ (Hershkovitz, 1960). 7642), including C. del Aguila's sublo- 10. Guajira, San Antonio (1 103N/7326); el- calities Avispero, Belgica, Guapaton, evation 3450 ft (Paynter and Traylor, Mantagua, Marmato, Naranjal, Picuma, 1981). San Calixto, and San Isidro. 11. Guajira, Sierra Negra (1036N/7255), in 13. Huila, 17 km SE Villavieja (0313N/ the Sierra de Perija E ofVillanueva, and 7514); elevation 1600 ft. Villanueva (1037N/7259); elevation 274 14. Magdalena, Agua Dulce (ca. 1106N/ m at Villanueva and 1265 m at Sierra 7407), Bonda (1114N/7408), Cincinati Negra (Hershkovitz, 1960). (1 106N/7406), Mamatoco (= Mamato- 8 AMERICAN MUSEUM NOVITATES NO. 3050

Fig. 2. Collection localities in Ecuador and Peru. Densely stippled land areas are those above 1500 m elevation.

ca, 11 14N/74 10), Manzanares (11 14N/ 7329); elevation ca. 400 m (Paynter and 7413; but possibly a community and not Traylor, 1981). the river), Masinga and Masinga Vieja 17. Santander, Bucaramanga (0708N/7309); (both at ca. 1116N/7405), Min- elevations from 1050 to 1100 m. ca(l 109N/7407), Onaca (111 IN/7404), 18. Tolima, Mariquita (0512N/7454); ele- and "Santa Marta" (no coordinates; vation 3000 ft. probably a general reference to the whole sierra and not the coastal town); from ECUADOR near sea level on the Caribbean coastal 1. Azuay, Las Cajas National Park (0253S/ plain to about 3500 ft in the adjoining 7919), including "Torreadora" and northwestern foothills of the Sierra Ne- "Lake Luspa"; elevations on A. Bar- vada de Santa Marta. nett's specimen tags are 3750 and 3800 15. Meta, San Juan de Arama, Los Micos m. (0317N/7353); elevation 450 m (Paynter 2. Azuay, Tunguilla Valley (= Valle de and Traylor, 1981). Yunguilla; 0318S/7918); elevation 1500 16. Norte de Santander, Guamalito (0834N/ m. 1992 VOSS: SOUTH AMERICAN SIGMODON 9

3. Chimborazo, Puente de Chimbo (021OS/ from Quevedo (0102S/7929) to Baba- 7910) near Bucay (0210S/7906); eleva- hoyo (0149S/7931). tion 1200 ft. 22. Los R(os, Vinces (0132S/7945), includ- 4. Chimborazo, Urbina (0130S/7844); el- ing nearby settlements Hacienda Pijigual evation 11,400 ft (Anthony, 1924). and Hacienda Santa Teresa (ca. 12 km 5. El Oro, Piedras (0338S/795 5); elevation NE Vinces); elevation 100 ft at Hacienda 300 m. Pijigual. 6. El Oro, Portovelo (0343S/7939); eleva- 23. Manabi, Chone (0041 S/8006); elevation tion 2000 ft. 20 m (Paynter and Traylor, 1977). 7. El Oro, Santa Rosa (0327S/7958); ele- 24. Manabl, Rio de Oro (ca. 0028S/7936). vation 100 ft. 8. Guayas, Balzar Mountains (= Balzar, GUYANA 0122S/7954); elevation 100 m (a mis- 1. Rupununi, Dadanawa (0250N/5930), nomer, see Paynter and Traylor, 1977). including Dadanawa Ranch and 15 mi 9. Guayas, Bucay (0210S/7906); elevations ENE Dadanawa; elevation ca. 150 m from 900 to 1000 ft. (Stephens and Traylor, 1985). 10. Guayas, Cerro Manglaralto (presumably 2. Rupununi, Kanuku Mountains (ca. near Manglaralto at 0150S/8044) and 0312N/5930), including J. J. Quelch's Hacienda El Refugio (5 km E Manglar- localities Kwaimattat (?= Kwaimatta), alto); elevation 1200 ft at Cerro Mang- Pirasa, Rupununi River, and Upocarit; laralto, but Manglaralto is at sea level elevation 240 ft. according to Paynter and Traylor, 1977). 3. Rupununi, Nappi Creek near Lethem 11. Guayas, Chongon (0214S/8004) and (0323N/5948); elevation 107 m at Leth- Chongoncito (east of Chongon); eleva- em (Stephens and Traylor, 1985). tion 60 m at Chongon, but Chongoncito is at sea level according to G. H. H. Tate's PERU fieldnotes (AMNH Department of 1. La Libertad, Pacasmayo (0724S/7934); Mammalogy archives). elevation 8 m (Stephens and Traylor, 12. Guayas, Guayaquil (0210S/7950), Dur- 1983. an (= Alfaro; 0212S/7950), and G and 2. La Libertad, Trujillo (0807S/7902); el- Q RR kilometer 8; near sea level. evation 200 ft. 13. Guayas, Puna and San Ramon, both on 3. Lambayeque, Eten (0654S/7952); ele- Isla Puna (0250S/8008); near sea level. vation 10-15 m. 14. Guayas, Pacaritambo, Canton de Balzar, 4. Piura, Hacienda Laredo, El Arenal 30 km S Quevedo (0102S/7929). (0455S/8102). 15. Loja, Casanga Valley near Guachanama 5. Piura, Huasimo (0419S/8012); elevation (0402S/7953) and Celica (0407S/7959); 750 m. elevation 2900 ft. 6. Piura, Jilili (0435S/7949); elevation be- 16. Loja, Malacatos (0414S/7915); eleva- tween 1000 and 2000 m (from map). tion 1600 m. 7. Piura, Lancones (0435S/8028). 17. Loja, Pindal (0409S/8010). 8. Tumbes, El Sauce (0403S/8034). 18. Los Rlos(?), Limon, on road from Bal- 9. Tumbes, Matapalo (0341S/8012); ele- sapamba (= Balzapamba; 0147S/7913) vation 54 m (Stephens and Traylor, to Babahoyo (0149S/7931); elevation 1983). 1200 ft. 10. Tumbes, Positos (0417S/8031); eleva- 19. Los Rios, Quevedo (0102S/7929); ele- tion 25 m. vation ca. 100 m (Paynter and Traylor, 11. Tumbes, Tumbes (0334S/8028); sea lev- 1977). el. 20. Los Rios, Rio Babahoyo near Pimocha (0152S/7935), including "Rio Baba- SURINAM hoyo," "Rio Babahoyo-Pimocha," and 1. Nickerie, Sipaliwini-savanne-vliegeveld "Rio Babahoyo/Pimocha-Sur." (0205N/5610); elevation ca. 358 m (Ste- 21. Los Rios, 1.5 km S San Carlos, on road phens and Traylor, 1985). 10 AMERICAN MUSEUM NOVITATES NO. 3050

2. Para, Zanderij (0527N/5512); elevation 14. Bolivar, San Ignacio Yuruani (ca. 0502N/ ca. 50 m (Stephens and Traylor, 1985). 6108); elevation 850 m. 15. Bolivar, Vetania, 46 km N Icabar(u VENEZUELA (0420N/6146); elevation 800 m (Han- 1. Amazonas, Paria Grande (25 km SSE dley, 1976). Puerto Ayacucho [0540N/6735]) and 16. Carabobo, El Trompillo (1004N/6746), Rueda (18 km SSE Puerto Ayacucho); near Lago Valencia; elevation 1300 ft elevation 119 m at Paria Grande and 125 (Thomas, 1914). m at Rueda. 17. Carabobo, La Florida (?0956N/6747), 2. Amazonas, San Juan Manapiare Serrania de Barbula; elevation 1050 m. (0518N/6613); elevation 155 m (Han- 18. Carabobo, near Montalban (1013N/ dley, 1976). 6820); elevations from 1055 to 1091 m. 3. Anzoategui, Rio Oritupano (ca. 0857N/ 19. Cojedes, "El Candelo," Tucuragua 6322); elevation 200 m. (0954N/6841), SW La Sierra; elevation 4. Apure, Entrada La Yaguita, Mantecal 640 m. (0733N/6909); elevation 12 m (ca. 75 m 20. Cojedes, Hato El Tirado (?09 1 ON/6823). according to Paynter, 1982). 21. Cojedes, Hato Itabana, 38 km SE Las 5. Apure, Hato Cariben (0633N/6713), 60 Vegas (0935N/6837); elevation 80 m. km NE Puerto P'aez; elevation 76 m. 22. Cojedes, Hato Los Caballos, on the road 6. Apure, Hato El Frio, 31 km (by road) W from Tinaco (0942N/6826) to El Baful El Saman de Apure (0755N/6844); ele- (0857N/6817); elevation 210 m. vation 60 m. 23. Cojedes, La Blanquera (ca. 0957N/6836), 7. Apure, 41 km NW Puerto Paez (0613N/ Cerro Azul; elevation 1100 m (Paynter, 6728 at Puerto Paez); elevation 76 m. 1982). 8. Aragua, Rancho Grande (1022N/6741), 24. Distrito Federal, Caracas (1030N/665 5), including El Limon (1018N/6738), and including Los Venados (5 km N Caracas) Guamita; Rancho Grande is situated at and "Phelps Hill"; elevation 950 m at ca. 900-1000 m just below the crest of Phelps Hill and 1485-1580 m at Los the coastal cordillera in the Lago Valen- Venados. cia catchment, while Guamita (at 700 m) 25. Distrito Federal, San Julian (ca. 1037N/ and El Limon (524 m) are in the foothills 6650); sea level (Paynter, 1982). a few km NW ofMaracay (101 5N/6736). 26. Falcon, Cerro Socopo (1028N/7048); el- 9. Aragua, Tiara (ca. 101 N/6709), includ- evation 1260-1265 m. ing adjacent localities in the Cordillera 27. Guarico, Calabozo (0856N/6726), Al- del Interior ESE of Maracay (101SN/ berto Ablan Farm; elevation ca. 100 m 6736), including Agua Amarilla, El Li- at Calabozo (Paynter, 1982). moncito, La Esperanza, and La Horque- 28. Guarico, Hato La Merrerefna, Las Mer- ta; and Camp Rangel (= Camp Rafael cedes (0907N/6624); elevation ca. 220 Rangel and Campamento Rangel). m at Las Mercedes (Paynter, 1982). 10. Barinas, near Altamira (0850N/7030); 29. Guarico, Parcela 200, ca. 20 km SW Cal- elevations from 600 to 1070 m (Han- abozo (0856N/6726); elevation less than dley, 1976). 500 m (from map). 11. Bolivar, El Llagual (= El Yagual at ca. 30. Guarico, San Juan de los Morros (0955N/ 0725N/6510); elevation pa; 100 m 6721). (Paynter, 1982). 31. Lara, Caserio Boro, 10 km N El Tocuyo 12. Bolfvar, Hato La Florida (ca. 0727N/ (0947N/6948); elevations from 518 to 6546); elevations from 43 to 68 m. 537 m. 13. Bolivar, Mount Roraima, including Ar- 32. Lara, vicinity of San Miguel (0953N/ abopo (also spelled "Arabupu"; 0506N/ 6931), including localities "Camp No. 6044) and Paulo (0508N/6049); eleva- 1," "La Fortuna," "Palenque," "San tion 4200 ft at Arabopo and 4000 ft at Geronimo," "San Rafael," and "east side Paulo. These localities were correctly ofTurbia valley" sampled by the Scarritt mapped by Tate (1930) in Venezuela, Venezuelan Expedition in 1938; eleva- but his specimen tags say Brazil. tions from 960 to 2500 m. 1 992 VOSS: SOUTH AMERICAN SIGMODON I1I

33. Merida, near Mesa Bolivar (0826N/ American species has revealed an unexpected 7134); elevation 690 m (Handley, 1976). range of variation even in traditional char- 34. Miranda, Curupao (103 1N/6638), 19 km acter complexes. This section summarizes E Caracas; elevation 1160 m. taxonomic, morphological, and geographical 35. Monagas, near Caripe (1012N/6329); information about the genus as a whole, and elevation 1180 m. evaluates prior conjectures concerning its 36. Monagas, Caripito (1008N/6306); ele- phylogenetic relationships. vation ca. 100 m (Paynter, 1982). 37. Monagas, Cocollar (IOION/6347), in- Sigmodon Say and Ord cluding Rio Cocoyal; elevations from Sigmodon Say and Ord, 1825: 354. 1575 to 2600 ft. Lasiomys Burmeister, 1854: 16. 38. Monagas, Hato Mata de Bejuco, 47-54 Deilemys Saussure, 1860: 98. km SE Maturin (0945N/6311 at Matu- Sigmomys Thomas, 1901: 150. rin); elevations from 18 to 36 m. 39. Monagas and Sucre, San Antonio Ma- TYPE SPECIES: Sigmodon hispidus Say and turin (1007N/6343) and Campo Alegre Ord (1825), based on three specimens col- (ca. I01ON/6345); elevations from 1500 lected in 1818 "in East Florida," probably to 1800 ft at San Antonio. from abandoned plantations along the St. Portuguesa, Los Mamones, Finca Santa Johns River where the animal was found to Rita (not located). be abundant (op. cit.: 355). The type series, 40. Sucre, 40 km NW Caripito (1008N/6306 originally deposited in the Philadelphia Mu- at Caripito). seum, is not extant2 but Say and Ord's illus- 41. Sucre, near Cumana (1028N/6410); sea tration (fig. 3) and description are sufficient level. to establish the name as pertaining to only 42. Sucre, Cumanacoa (1015N/6355); ele- one currently recognized muroid species in vation 700 ft. Florida. Bailey (1902) and Hall (1981) gave 43. Sucre, Finca Vuelta Larga, 9.7 km (by the type locality ofS. h. hispidus as "St. Johns road) SE Guaraiunos (1033N/6307); el- River, Florida," but Hall illogically included evation 10 m. the entire course of that river in the mapped 44. Sucre, Manacal (1037N/6301), 25 km E range of S. h. floridanus (op. cit.: 738). For Cartupano (1040N/6314); elevations comparisons with the South American taxa from 411 to 424 m. treated in this paper, a very large series of 45. Yaracuy, Finca El Jaguar, ca. 20 km NW specimens from the vicinity of Lake Placid Aroa (1026N/6854); elevation 690 m. in Highlands County, Florida is assumed to 46. Zulia, Empalado Savannas, ca. 30 mi E represent the typical morphology of hispidus El Panorama (ca. 1040N/7125); near sea (figs. 4, 5). level (Paynter, 1982). DISTRIBUTION: From about 41°N (in Iowa 47. Zulia, Novito (1002N/7243), 19 km and Nebraska) throughout most ofthe south- WSW Machiques; elevation 1135 m. ern, southeastern, and southwestern United 48. Zulia, Rio Aurare (= Rio Anaure), prob- States, Mexico, and Central America (see ably near El Panorama (ca. 1040N/7125); 2 The natural history collections of the Philadelphia near sea level (from map). Museum (also known as the Peale Museum) were sold 49. Zulia, Rio Cogollo (?1015N/7230); ele- in 1850 to P. T. Barnum and Moses Kimball. Barnum's vation 350 ft (Paynter, 1982). halfofthe collections was destroyed by fire in New York in 1865, but Kimball's was deposited in the Boston Mu- THE GENUS seum and later (in 1893 and 1899) given to the Boston Descriptions ofSigmodon provided by Say Society of Natural History. Some of the latter material and Ord (1825), Coues (1877), Gyldenstolpe was destroyed by the Society, and the original labels of (1932), and Ellerman (1941) are inadequate the remaining specimens were removed and lost. The surviving, unlabeled remnants ofthe Peale Museum were for the purposes ofthis report: the genus now finally obtained by the Museum ofComparative Zoology includes Sigmomys as a subjective junior at Harvard University in 1914, but the type series of synonym, many new characters have been Sigmodon hispidus was apparently not among them (Al- discovered by muroid morphologists in re- len, 1931). The sad history of the Peale Museum's im- cent decades, and close study of the South portant collections is chronicled by Barbour (1946). 12 AMERICAN MUSEUM NOVITATES NO. 3050

>14. .ij

'VI -Ifk .1 ki

......

Fig. 3. Say and Ord's (1825) illustration of Sigmodon hispidus.

Hall, 1981: maps 426-431), to northern South roots; stomach unilocular and hemiglandu- America. In South America, the genus occurs lar; gall bladder present; preputial glands in from Colombia southward to northwestern two pairs; glans penis complex, with tridigi- Peru, and eastward at least as far as Surinam. tate baculum. DIAGNOSIS: Small-eared, short-whiskered, DESCRIPTON: Pelage coarse, short, and close and short-tailed rodents of the muroid sub- in lowland forms, but deeper and softer in family Sigmodontinae (sensu Carleton and populations from higher elevations; always Musser, 1984) with coarsely grizzled, brown- coarsely grizzled-brownish dorsally, but ven- ish dorsal pelage; hindfoot unwebbed, with tral pigmentation variable within and among three middle digits (II, III, and IV) much species; ears small but always visible, sparse- longer than outer digits (I and V); mammae ly haired in lowland forms but furry in pop- ten (in five pairs) or eight (in four pairs); con- ulations and species from high elevations. vergent, beaded postorbital margins of skull Mystacial, superciliary, genal, submental, in- continuous with prominent temporal crests; terramal, and carpal vibrissae present; mys- zygomatic notches deeply excavated; inferior tacial vibrissae short, never extending be- zygomatic root broad, with spinous antero- yond pinnae when laid flat against cheek. dorsal process; parapterygoid fossae. deeply Manus and pes well-haired dorsally, but excavated; alisphenoid strut present (buccin- without conspicuous tufts of longer hairs at ator-masticatory foramen and foramen ovale bases ofclaws; entirely naked on plantar sur- accessorius separate); tegmen tympani over- faces (including heel of pes); manus with five laps posterior margin of squamosal antero- plantar tubercles (two carpal and three inter- dorsal to bulla; molars hypsodont and con- digital pads); pes with six plantar tubercles spicuously lophodont, with principal cusps (two metatarsal and four interdigital pads); and crests deployed in sigmoidal arrays; an- digits I and V of pes always much shorter terocone(id) of Mt/mt undivided; meso- than adjacent digits II and IV (claw ofI never loph(id)s absent; first upper molar with four extending to first interphalangeal joint of II, roots; first lower molar with three or four claw of V extending to but not beyond first 1 992 VOSS: SOUTH AMERICAN SIGMODON 13

Fig. 5. Left maxillary (A) and right mandibular (B) molar series of Sigmodon hispidus (AMNH 255409) from 8 mi S Lake Placid, Highlands County, Florida. Fig. 4. Cranium and mandible ofSigmodon hispidus (AMNH 255414) from 8 mi S Lake Placid, convergent and beaded; supraorbital beads Highlands County, Florida. continuous with prominent temporal crests along dorsolateral contours of braincase; zygomatic arches usually convergent (widest interphalangeal joint ofIV).3 Tail indistinctly across the squamosal roots); interparietal or sharply bicolored (dark above and paler usually large (but reduced and sometimes ab- below); always shorter than combined length sent in leucotis). Rostrum short and deep in ofhead-and-body; sparsely haired in lowland lateral view; zygomatic plate broad, with con- species (tails of highland forms are more cave anterior margin and spinous anterodor- densely haired but epidermal scales are never sal process. Incisive foramina long (extending completely concealed); tail tip without con- posteriorly to or between the molar alveoli) spicuous tuft or pencil oflonger hairs. Mam- or short (not so extending). Morphology of mae usually ten in inguinal, abdominal, tho- palatal bridge highly variable among species; racic, postaxial, and pectoral pairs (but some long or short, broad or narrow (sensu Hersh- specimens of hispidus have eight or twelve kovitz, 1962); with or without median keel, teats [Meyer and Meyer, 1944] and thoracic posterolateral sulci, and median posterior teats are apparently absent in most specimens spinous process. Anterior limit ofmesoptery- of inopinatus). goid fossa correspondingly diverse in posi- Skull distinctive in dorsal view, with stout tion and configuration, but roof of meso- rostrum flanked by deeply excavated zygo- pterygoid fossa always fenestrated by large matic notches; nasals long (concealing inci- sphenopalatine vacuities. Parapterygoid fos- sors from dorsal view) or short (incisors vis- sae deeply excavated above level of molar ible dorsally); posterior supraorbital margins alveoli. Stapedial foramen large or very small (entirely occluded in some specimens); other 3Hall's (1981: 736) illustration of S. hispidus erro- osteological traces ofcarotid circulation vari- neously shows the hindfoot with a very long fifth digit. able. Buccinator-masticatory and accessory 14 AMERICAN MUSEUM NOVITATES NO. 3050 oval foramina separated by vertical strut of digitate bacular cartilage and deep terminal alisphenoid. Auditory bullae large or small crater containing three bacular mounds, one but always globular, with short bony eusta- dorsal papilla, lateral papillae, and a bifurcate cian tubes. Braincase above bulla on each urethral flap (Hooper and Musser, 1964). side perforated by large postglenoid foramen CorNTs: Sixty-nine Recent species-group and subsquamosal fenestra separated by slen- taxa are referred to the genus; 57 of these, der hamular process of squamosal. Tegmen from North and Central America, may rep- tympani (petrosal roof of bullar capsule) resent only seven valid species (alleni, ari- overlaps posterior margin of squamosal be- zonae, fulviventer, hispidus, leucotis, mas- low postglenoid foramen. cotensis, and ochrognathus; see Hall, 1981), Mandible deep (from coronoid process to but the Central American "subspecies" of angle) in proportion to its length; coronoid hispidus have never been critically revised. process strong and falciform; angular process The twelve remaining nominal taxa, based extends posteriorly beyond condyle; lower on South American specimens, represent four incisor root contained in prominent capsular diagnosable species as shown below. Martin process posteroventral to base of coronoid. (1979) reviewed the fossil record of Sigmo- Upper incisors smooth or grooved, strong- don, which includes many nominal Pleisto- ly opisthodont or procumbent and almost or- cene forms. thodont, but enamel bands always deeply pigmented (yellow or orange, never whitish). Upper molar rows parallel or slightly con- PHYLOGENETIC RELATIONSHIPS vergent; upper molars hypsodont and lopho- Despite its extensive North American dis- dont, with principal cusps and crests contribution, Sigmodon has usually been com- nected in more-or-less continuous sigmoidal pared with several genera endemic to South series. Anterocone of Ml undivided; anter- America, Holochilus, Reithrodon, and Neo- oloph absent as a discrete structure on Ml, tomys, the four together constituting the "sig- present and well developed on M2 and M3; modont" group of Hershkovitz (1955). De- mesolophs absent on all upper teeth; poster- spite certain resemblances in zygomatic olophs absent (or fused with metacones) on architecture and molar occlusal morphology all upper teeth; Ml with four roots, M2 and noted by Thomas (1894), Ellerman (1941), M3 each with three roots. Anteroconid ofml Sanborn (1947), and Hershkovitz (1955), undivided; anterolophid and mesolophid ab- however, these genera diverge in other char- sent on all lower teeth; posterolophid well acters and their relationships are not well un- developed on ml and m2, absent (or fused derstood. In comparisons of penis morphol- with entoconid) on m3; ml with three or four ogy, Reithrodon and Holochilus differ roots, m2 and m3 each with three roots. strikingly from Sigmodon but resemble Axial skeletal elements exhibiting little members of other suprageneric groups of variation in number, with modal counts of Neotropical muroids: Hooper and Musser 12 ribs, 19 thoracicolumbar vertebrae, 4 sa- (1964) emphasized similarities between cral vertebrae, and 23 to 25 caudal vertebrae Reithrodon and some phyllotines, and be- in all available species samples. First rib ar- tween Holochilus and oryzomyines. The glans ticulates with seventh cervical and first tho- penis of Neotomys has not been described. racic vertebrae; greatly elongated neural spine The standard (unbanded) karyotypes of on second thoracic vertebra; entepicondylar Reithrodon and Neotomys resemble those of foramen of humerus absent. some phyllotines (Pearson and Patton, 1976) Stomach unilocular and hemiglandular but Gardner and Patton (1976: 31-32) also (Carleton, 1973); gall bladder present (Voss, noted chromosomal similarities with Sig- 1991: table 4); male accessory secretory or- modon and suggested that Neotomys ... gans consisting of one pair each of bulbour- could be annectant between the phyllotines ethral, anterior and dorsal prostate, ampul- and the sigmodonts." Some standard kary- lary and vesicular glands, and two pairs each otypes of Holochilus brasiliensis (sensu ofpreputials and ventral prostates (Voss and Hershkovitz, 1955) closely resemble those Linzey, 1981); glans penis complex, with tri- both of Sigmodon and Oryzomys species 1992 VOSS: SOUTH AMERICAN SIGMODON 15

(Gardner and Patton, 1976), but Baker et al.'s will devolve to molecular comparisons. As- (1983) analysis of G-banded chromosomes pects of systematic sampling design for test- purported to demonstrate that H. brasiliensis ing biogeographically relevant hypotheses of is more closely related to Oryzomys than to relationships with Sigmodon are discussed in Sigmodon. the concluding section of this report. As a consequence of these and other anatomical and cytological comparisons, Reig VARIATION WITHIN AND AMONG (1984, 1986) and Olds and Anderson (1989) THE SOUTH AMERICAN SPECIES placed Reithrodon and Neotomys in the Phyl- The South American forms of Sigmodon lotini, but the latter authors cautioned that recognized as valid in this report, hispidus, tribal monophyly is not unambiguously sup- alstoni, peruanus, and inopinatus, can all be ported by character data. Reig (1984, 1986) identified by the morphological criteria de- classified Holochilus and Sigmodon as the scribed below. Although qualitative charac- only members ofthe (undiagnosed) tribe Sig- ters are sufficient for the simple purpose of modontini, but Voss (1991) pointed out that diagnosis, analyses of morphometric data the derived absence of a gall bladder only in provide additional evidence of phenotypic the former genus supports Hooper and Mus- divergence and reinforce the biological con- ser's suggestion of oryzomyine affinities. clusion that four species are represented These ambiguous and conflicting obser- among the specimens examined. Some char- vations about relationships are typical ofthe acters of traditional taxonomic interest but unsatisfactory current state of knowledge of little real value in the present context are concerning Neotropical muroid systematics also treated here for the sake of descriptive emphasized by Carleton and Musser (1989: completeness. 53-5 5). Acknowledging that ambiguity and controversy exist is frustrating but salutary: we should recognize the need for remedial QUALITATIVE EXTERNAL CHARACrERS research and admit that phylogenetic as- South American Sigmodon are remarkably sumptions current in the literature may be uniform in external appearance. No inter- wrong. Unfortunately, the widespread use of specific variation is apparent in the distri- formal nomenclature for suprageneric groups bution of vibrissae, the number of plantar of Neotropical muroids (e.g., Akodontini, tubercles, or any gross aspects of the ears, Ichthyomyini, Oryzomyini, Phyllotini, Scap- feet, and tail. Females of hispidus, alstoni, teromyini, Sigmodontini) conveys no such and peruanus consistently have ten teats in healthy skepticism. Despite Reig's (1980: 262) five pairs (inguinal, abdominal, thoracic, conviction that most South American mu- postaxial, and pectoral), but mammae num- roids ". . . show a clear-cut pattern of inter- ber may differ in inopinatus: one specimen generic affinity, as revealed by various organ- (AMNH 66308) has ten teats, but the tho- systems and by chromosomes," the mono- racic pair is apparently absent in six other phyly of only one ofthe above-named tribes mature females ofthis rare species. With this is unambiguously supported by shared, de- exception, qualitative variation in external rived attributes (Voss, 1988). features is limited to pelage density, texture, The phylogenetic relationships of Sigmo- and pigments. don are unknown, and the systematic context Pelage density and texture are correlated within which alternative hypotheses must be with altitude: specimens from lower eleva- evaluated is daunting. The lack ofconvincing tions have close, thin, coarse fur, whereas synapomorphies for any group ofNew World specimens from higher elevations have deep- muroids that includes the genus, even the er, denser, and softer pelts. This ecomor- subfamily Sigmodontinae (Carleton and phological correlation can be observed both Musser, 1984), necessitates sampling repre- within species (e.g., by comparing specimens sentative Old World taxa as outgroups for of alstoni from Hato El Frio, at 60 m in the any cladistic analysis that merits serious con- Llanos, with those from Arabopo, at 1300 m sideration. Quite possibly, significant future in the Gran Sabana), and between inopinatus progress at this level ofmuroid phylogenetics (endemic to the high pfiramos ofAndean Ec- 16 AMERICAN MUSEUM NOVITATES NO. 3050 uador) and other South American species found significant dimorphism in age-con- (which occur at much lower altitudes). Pop- trolled sex comparisons of head-and-body ulations with denser body pelage likewise have length for captive-bred S. hispidus, but sex more thickly furred ears and tails, an obser- differences in external dimensions are not ap- vation that can also be documented by com- parent in the samples treated here. parisons within species or between inopina- Ratios of tail and hindfoot to head-and- tus and its lowland congeners. body length (table 2) suggest that peruanus Pelage pigments vary within all of the and inopinatus have relatively shorter ap- widely ranging species of South American pendages than hispidus and alstoni, but sub- cotton rats: hispidus, alstoni, and peruanus. stantial variability in these proportions with- All species exhibit the coarsely grizzled, in species is evident and several samples are brownish dorsal pattern characteristic of the not large. Furthermore, since length oftail is genus, but populations differ in overall sat- subtracted from total length to obtain length uration (of reddish and yellow tones) and ofhead-and-body, minor methodological dif- darkness. Available ecological information ferences among collectors measuring tails suggests that dramatic contrasts in pelage col- could substantially affect the computed ratio or among geographic samples are correlated LT/HBL. An additional complexity is intro- with environmental humidity; an example is duced by ontogenetic variation: body length provided by the dark specimens ofperuanus increases steadily throughout life whereas the collected from secondary growth in the hu- tail and hindfeet complete most of their mid valley of the Rio Chimbo (Ecuadorean growth in the first few weeks after birth (Ji- locality 3) and the pale skins of the same menez, 1971); therefore, younger adults have species from arid landscapes in northwestern relatively longer appendages than older ani- Peru. mals, and sample discrepancies in age struc- The grizzled pattern ofthe dorsal body pel- ture may confound population comparisons age is relieved only by subtle facial markings of external proportions. In view of these in- common to all the South American species, terpretive difficulties, which cannot be ad- although varying in brightness and definition dressed with materials currently available, within and among populations: a patch of confident taxonomic inferences from the data clear reddish or yellow fur around the nose, in table 2 are impossible despite the impres- and rings ofpale (usually yellow) fur that en- sive magnitude of some apparent differences. circle the eyes. The tails of all South Amer- ican Sigmodon are darker above than below, QUALITATIVE CRANIODENTAL CHARACTERS a pattern that is indistinct in some samples and sharply defined in others of the same By contrast with the essentially undiffer- species. The long, silvery guard hairs on the entiated external appearance of South Amer- rump and flanks of inopinatus that Anthony ican cotton rats, many aspects ofcraniodental (1924) believed to be diagnostic are present morphology vary among the groups of pop- but less conspicuous in other South Ameri- ulations recognized as species in this report. can species as well. Qualitative and quantitative character variation in the head skeleton support the same EXTERNAL DIMENSIONS taxonomic assignments, but it is convenient to treat these sources of information sepa- South American species of Sigmodon do rately. The present section summarizes dis- not differ greatly in size, but external mea- crete variation, traits that can be scored as surements ofrepresentative samples (table 1) either present or absent, together with aspects reveal that alstoni is appreciably smaller in of shape variation that are more easily de- head-and-body length than hispidus, peru- scribed verbally than measured. anus, and inopinatus. The smaller size ofthe Nasal bones: The nasals are long in most former is also evident in hindfoot length, a species of Sigmodon, projecting anteriorly contrast of practical value for identification beyond the premaxillae to conceal the strong- where alstoni and hispidus are sympatric in ly opisthodont incisors from dorsal view (fig. Colombia and Venezuela. Chipman (1965) 6A, B). S. inopinatus, however, has short na- 1992 VOSS: SOUTH AMERICAN SIGMODON 17

TABLE 1 Summary Statisticsa for External Measurements (in millimeters) of Adult Specimens of South American Sigmodon Species Species Length of head-and-body Length of tail Length of hindfoot hispidusb 150 ± 14(125-174) 114 ± 11(92-135) 33 ± 1 (30-36) alstonic 137 ± 11(106-152) 91 ± 8 (72-102) 27 ± 1(24-29) peruanusd 148 ± 13(120-172) 93 ± 10(79-116) 31 ± 2(28-34) inopinatuse 153 ± 11(136-167) 88 ± 7 (78-99) 29 ± 1(28-30) a The sample mean plus or minus one standard deviation, and the observed range (in parentheses). b N = 24; sample from Montalban, Venezuelan locality 18. c N = 26; sample from Hato El Frio, Venezuelan locality 6. d N = 35; sample from Chongoncito, Ecuadorean locality 11. e N = 9; sample from Urbina, Ecuadorean locality 4. sals that do not project beyond the premax- or exceeds the length ofeither tooth; and long illae, and the procumbent incisors ofthis spe- because the mesopterygoid fossa does not ex- cies are consequently exposed when the skull tend between the third molars (i.e., the palate is viewed from a dorsal perspective (fig. 6C, is produced behind the molar rows). Ante- D). riorly, at the level ofthe first and second mo- Incisive foramina: The incisive foramina lars, the palatal bridge of hispidus is smooth, are prominent slots in the diastemal palate but a median keel is sometimes developed that accommodate the soft tissues of the na- posteriorly; when well developed, this keel is sopalatine apparatus and communicate the sometimes produced as a small median spine nasopalatine canals (Quay, 1954). Among on the posterior palatal margin. Shallow pos- muroids, these foramina display consider- terolateral sulci containing one or more small able taxonomic variation in size and shape, pits are found between the median keel and the functional significance of which is ob- the third molars in some specimens of his- scure; South American species of Sigmodon, pidus, but in others the posterior palate is however, only exhibit modest differences in smooth and almost devoid of relief. By con- foraminal length. In most specimens of his- trast, the palatal bridge ofperuanus (fig. 7B) pidus (fig. 7A), the foramina are long, ex- is narrow and short: the distance across the tending posteriorly to or between the molar palate between the first molars is less than alveoli (such that a line tangent to the anterior the length ofeither tooth, and the mesoptery- limits of both first molar sockets is also tan- goid fossa projects between the third molars. gent to or intersects one or both foramina); The palatal bridge ofperuanus is also marked in peruanus (fig. 7B), by contrast, the incisive by prominent grooves that extend on each foramina are usually short and do not extend side from the incisive foramen to a well-de- posteriorly to the level of the molar alveoli; fined posterolateral sulcus. The posterolat- alstoni and inopinatus exhibit greater vari- eral sulci are always separated by a conspic- ability in the posterior limits ofthese foram- uous median keel that is often produced as a ina (table 3). small spine on the posterior palatal margin. Palatal bridge: The palatal bridge (an an- The morphology of the palatal bridge in atomical term peculiar to muroid systemat- other South American species of Sigmodon ics, denoting the hard palate posterior to the is intermediate to the extremes described diastema) and the foramina it contains pro- above. In adult specimens ofalstoni, the pal- vide several important characters for dis- ate is broad (although generally not so broad criminating South American species of Sig- as in hispidus) but the mesopterygoid fossa modon. The palatal bridge of mature adult often extends between the third molars (not specimens of hispidus (fig. 7A) is broad and so deeply, however, as in peruanus); the pal- long in the sense defined by Hershkovitz atal bridge is usually smooth, lacking a well- (1962): broad because the distance across the defined median keel and deep posterolateral palate between the first upper molars equals sulci, but a small, blunt spine is usually pres- 18 AMERICAN MUSEUM NOVITATES NO. 3050

TABLE 2 TABLE 3 Mean External Proportions for Population Sam- Relative Frequencies of Alternative Morphologies ples of South American Sigmodon Species of the Incisive Foramina in South American Sigmodon Species Ratios Species samples LT/ HF/ Samplesa N HBL HBL inopi- hispidus alstoni peruanus natus hispidus (N= (N= (N= (N= Venezuela Morphologya 142) 67) 100) 11) 8. Rancho Grande 11 0.86 0.24 short 0.05 0.28 0.95 0.18 18. Montalban 24 0.76 0.22 intermediate 0.13 0.24 0.04 0.09 alstoni long 0.82 0.48 0.01 0.73 Venezuela a Key: short, incisive foramina do not extend to or 6. Hato El Frio 25 0.67 0.20 between molar alveoli; intermediate, incisive foramina 8. Rancho Grande 15 0.77 0.22 extend to but not between molar alveoli; long, incisive 39. San Antonio Maturin 12 0.74 0.23 foramina extend between molar alveoli. peruanus widely recognized as primitive for muroids Ecuador (e.g., by Bugge, 1970; Carleton, 1980; and 3. Puente de Chimbo 11 0.56 0.20 Voss, 1988) occurs only in peruanus. Every 11. Chongoncito 35 0.63 0.21 specimen ofthat species has a large stapedial 13. San Ram6n 17 0.69 0.21 foramen in the petrotympanic fissure on the inopinatus posteromedial surface of the bulla; the sta- Ecuador pedial artery enters the bullar capsule through 1. Las Cajas 3 0.55 0.19 this orifice, passes through the stapes, and 4. Urbina 9 0.58 0.19 exits the bulla through a cleft in the tegmen a Named and numbered as in the gazetteer. tympani. Inside the braincase, the stapedial artery divides into supraorbital and infraor- bital branches, and the former vessel leaves ent on the posterior palatal margin. Few spec- a prominent groove along the inner surfaces imens of inopinatus are available, but most ofthe squamosal and alisphenoid bones as it intact adult skulls have narrow, short palates passes rostrally to enter the orbit through a (though neither so narrow or as short as in large sphenofrontal foramen (see Carleton, peruanus); the palate lacks deep longitudinal 1980; Voss, 1988; and Carleton and Musser, grooves and well-defined posterolateral sulci, 1989 for illustrations of this plesiomorphic but a distinct median keel between the third condition). In all other species of Sigmodon, molars is always produced as a small, sharp the stapedial artery is extremely small and spine on the posterior palatal margin. probably supplies only the middle ear and Posterior palatal foramina: The posterior some of the meningeal circulation; the sta- palatal foramina are small perforations in the palatal bridge, one on each side, at about the TABLE 4 level where the second and third molars abut. Relative Frequencies of Alternative Morphologies In alstoni and peruanus, these foramina are of the Posterior Palatal Foramina in South Amer- always contained in the maxillary-palatine ican Sigmodon Species suture, and this is also the usual condition for specimens of inopinatus; in hispidus, Species samples however, the posterior palatal foramina are inopi- often enclosed completely by the palatine hispidus alstoni peruanus natus bones (table 4). (N= (N= (N= (N= Stapedial circulation: Two morphologies Morphologya 136) 69) 100) 11) of the stapedial arterial circulation are pres- P/M 0.24 1.00 1.00 0.82 ent among South American species of Sig- P 0.76 0.00 0.00 0.18 modon, and both can be recognized by os- a Key: P/M, posterior palatal foramina in the maxil- teological features. The stapedial pattern lary/palatine suture; P, foramina enclosed by palatines. 1 992 VOSS: SOUTH AMERICAN SIGMODON 19

Fig. 6. Dorsal and lateral views of the rostrum. A, B, Sigmodon peruanus (AMNH 67042); C, D, S. inopinatus (AMNH 66308). pedial foramen is then reduced to a minute alisphenoid groove and sphenofrontal fora- pore (which appears to be entirely occluded men are both absent. The facial (ophthalmic in some specimens), while the squamosal- and internal maxillary) circulation in this de- 20 AMERICAN MUSEUM NOVITATES NO. 3050 rived condition is instead supplied by a branch of the internal carotid (anastomosis A5 in Bugge's terminology; op. cit.) which leaves a diagonal, internal groove across the roof of the parapterygoid fossa that is sometimes visible externally as a streak oftranslucent bone (see fig. 21A in Carleton and Musser, 1989). Auditory bullae: The auditory bullae are small in hispidus (fig. 8A), alstoni, and inopinatus, but conspicuously larger and more inflated in peruanus (fig. 8B). Incisors: The upper incisors ofhispidus and peruanus are smooth and strongly opisthodont (not remarkably different from the common morphology ofthese teeth in many other Neotropical muroid genera) and differ only by their relative width, the incisors of hispidus being visibly narrower than those of peruanus when skulls of like size are com- pared. The upper incisors of inopinatus are also smooth, but are noticeably more procumbent than those ofhispidus and peruanus (fig. 6); in no specimen of inopinatus, however, are the incisors actually orthodont. The deeply grooved upper incisors of alstoni (fig. 9B) were the basis for the original assignment ofthat species to the genus Reith- rodon (by Thomas, 1880), and provide the only criterion for separating Sigmomys from Sigmodon. The grooves, which divide the en- ameled anterior surface of the tooth into un- equal medial and lateral moieties, are always distinct and in no population sample that I have examined is there any evidence that grooves are subject to polymorphic variation (as implied by Hershkovitz, 1955). Where groove-toothed and smooth-toothed Sig- modon have been collected together they differ by other characters as well and sympatry, not polymorphism, is the biologically appro- priate interpretation (see the species account for alstoni below). Fig. 7. Morphology of the posterior palate. A, Molar occlusal morphology: Differences in Sigmodon hispidus (USNM 442531); B, S. peru- molar occlusal design among Sigmodon spe- anus (AMNH 67028). cies are generally subtle, but the maxillary dentition ofperuanus diverges from the mor- bial cusps alternate down the long axis ofthe phology common to other South American toothrow, and their angular shapes (especial- congeners. In hispidus, alstoni, and inopi- ly when unworn) recall the prismatic divi- natus (fig. 1 OA, B, D) the protocone and par- sions ofmicrotine molars. Interspecific com- acone are opposite one another (or very near- parisons of mandibular toothrows (fig. 1 1) ly so), as are the hypocone and metacone, reveal similar if less conspicuous patterns. and the cusp apices are obtusely rounded. In Lower first molar roots: The anteriormost peruanus (fig. lOC), however, lingual and la- mandibular molar exhibits significant vari- 1 992 VOSS: SOUTH AMERICAN SIGMODON 21

Fig. 8. Relative size ofthe auditory bullae. A, Sigmodon hispidus (USNM 442531); B, S. peruanus (AMNH 67028). ation in root number among Sigmodon spe- tine that may be hard to see above the al- cies, and Martin (1979) used this character veolus (table 5). Because molar roots are of- as the primary basis for assigning Recent and ten difficult to observe in young specimens Pleistocene forms to species groups. A large whose teeth are deeply embedded in alveolar anterior root (under the anteroconid), a large bone, root numbers are best determined from posterior root (under the hypoconid-entocon- older individuals. id pair), and a smaller labial root (under the protoconid) are always present; the variable CRANIODENTAL DIMENsIoNs element is lingual. The lingual root is well Age-correlated growth and sexual dimor- developed (but always smaller than the an- phism are the most causes terior and posterior roots) in most examples important tangible and in and of morphometric variation within the field- ofhispidus inopinatus; alstoni pe- collected, mixed cross-sectional samples or- ruanus, however, the lingual root is either dinarily available for systematic studies of absent (the commonest condition), or it is mammals. The effects ofpostweaning growth very small and weak, a slender thread ofden- and sex on muroid craniodental measurement variation were recently analyzed by Voss TABLE 5 et al. (1990) and Voss and Marcus (in press) Relative Frequencies of Alternative Conditions of who discovered that the first principal com- the Lingual Root of ml in South American Sig- ponent ofthe within-sample covariance ma- modon Species trix (of log-transformed measurements) ap- proximates a general size factor with Species samples coefficients that reflect ontogenetic allometry. inopi- Within samples, specimen scores on general hispidus alstoni peruanus natus (N= (N= (N= (N= size are correlated with age (and sometimes Morphology 52) 50) 75) 11) also with sex), but variation residual to the first principal component is essentially age- well developed 0.73 0.00 0.04 0.91 and sex-invariant. An important implication small and weak 0.23 0.24 0.31 0.09 of these results is that estimates of Mahala- absent 0.04 0.76 0.65 0.00 nobis distances, and of sample mean differ- 22 AMERICAN MUSEUM NOVITATES NO. 3050

TABLE 6 Population Samples ofSigmodon Species Included in Morphometric Analyses ",l IS Samplesa N hispidus United States Lake Placid, Florida 37 Colombia 12. Valle de Suaza 37 14. Minca 28 Venezuela 8. Rancho Grande 14 18. Montalban 40 alstoni Guyana 2. Kanuku Mountains 14 Venezuela 6. Hato El Frio 16 8. Rancho Grande 26 I.8 39. San Antonio Maturin 12 peruanus Ecuador 3. Puente de Chimbo 12 11. Chongoncito 16 13. Isla Puna 34 inopinatus Ecuador 4. Urbina 8 Fig. 9. Upper incisors. A, Sigmodon hispidus (USNM 442508); B, S. alstoni (USNM 520676). a Named and numbered as in the gazetteer. ences adjusted for general size (see Materials and Methods: Multivariate Statistics), are in- trasts ofprimary interest here. All ofthe anal- sensitive to minor sample discrepancies in yses reported below are therefore of sample age and sex composition. data unadjusted for sex. Samples and sexual dimorphism: The sam- Mahalanobis distances: Although Mahal- ples employed~~Bin the following analyses (table anobis distances between some samples 6) consist of "adults" as defined in Materials judged to be conspecific by qualitative cri- and Methods, but a considerable ontogenetic teria are large, interspecific distances are con- range is obviously represented in each (the sistently greater (fig. 12). The Florida popu- development of cranial ridges and crests in- lation representing the typical morphology of creases with age and provides a subjective hispidus is the most divergent among the five basis for this judgment; see Chipman, 1965). samples referred to that species, but the clus- Summary univariate statistics for the sample ters identified as peruanus and alstoni are not of Sigmodon hispidus from Florida are tab- significantly more homogeneous despite the ulated by gender (table 7), but univariate sta- less extensive geographic ranges they repre- tistics for other samples (which appear in ta- sent. The type series of inopinatus (from Ur- bles accompanying the taxonomic accounts) bina) is clearly divergent from all other sam- are not. Males usually average larger than ples of South American Sigmodon for which females, but sex differences are always small sufficient measurement data are available for by comparison with the interspecific con- analysis. 1 992 VOSS: SOUTH AMERICAN SIGMODON 23

Fig. 10. Left maxillary molar series of South American Sigmodon species. A, S. hispidus (USNM 542133); B, S. alstoni (MHNLS 8218); C, S. peruanus (AMNH 63015); D, S. inopinatus (AMNH 66312). Although these results provide unambig- measurement data, and because morpho- uous evidence that taxa distinguishable by metric divergence cannot be assumed to pro- qualitative traits also differ in quantitative ceed at a constant rate in all lineages. The aspects of craniodental morphology, no ex- higher-order branching (among species), plicitly phylogenetic interpretation of the il- which occurs over a fairly narrow range of lustrated branching pattern is possible. This distance values, is especially unlikely to have limitation arises from the difficulty ofdistin- any historical significance. Some of the clus- guishing primitive from derived similarity in ters of populations identified as species in

TABLE 7 Summary Statisticsa for Craniodental Measurements (in millimeters) of Adult Sigmodon hispidus from 8 mi S Lake Placid, Florida Females Males (N = 20) (N = 17) Whole sample CIL 29.7 ± 2.1 (26.4-34.1) 30.5 ± 3.2 (26.4-35.5) 30.1 ± 2.7 LD 8.1 ± 0.8 (6.9-9.7) 8.5 ± 1.3 (6.8-10.7) 8.3 ± 1.1 LM 6.4 ± 0.2 (6.0-6.8) 6.4 ± 0.3 (6.0-6.0) 6.4 ± 0.2 BM1 2.1 ± 0.1 (1.9-2.2) 2.1 ± 0.1 (2.0-2.3) 2.1 ± 0.1 LIF 6.9 ± 0.6 (6.0-8.2) 7.1 ± 0.9 (5.7-8.6) 7.0 ± 0.8 BR 5.2 ± 0.3 (4.6-5.8) 5.4 ± 0.6 (4.4-6.7) 5.3 ± 0.5 BPB 2.5 ± 0.3 (1.9-3.0) 2.5 ± 0.4 (2.0-3.3) 2.5 ± 0.4 BZP 3.4 ± 0.3 (2.7-4.0) 3.4 ± 0.5 (2.7-4.4) 3.4 ± 0.4 LIB 4.8 ± 0.3 (4.4-5.4) 4.8 ± 0.3 (4.4-5.4) 4.8 ± 0.3 DI 1.7 ± 0.2 (1.4-2.0) 1.8 ± 0.2 (1.4-2.0) 1.7 ± 0.2 BIT 2.4 ± 0.3 (1.9-2.8) 2.4 ± 0.3 (1.8-3.0) 2.4 ± 0.3 ZB 18.3 ± 1.0 (16.4-20.2) 18.6 ± 1.5 (16.4-21.0) 18.5 ± 1.3 a The sample mean plus or minus one standard deviation, and the observed range (in parentheses). 24 AMERICAN MUSEUM NOVITATES NO. 3050

Fig. 1 1. Right mandibular molar series ofSouth American Sigmodon species. Species and specimens are the same as those in fig. 10. figure 12 may be monophyletic, but this anal- eral size coefficients estimated from the larg- ysis by itself does not provide a sufficient est South American sample of each species basis for inferring monophyly. (table 8) reveal strikingly similar patterns of General size and species differences: Gen- multivariate allometric growth. Isometry for

Rancho Grande J Montalb6n Minca hispidus Valle de Suaza Florida J Chongoncito ) Isla Puna' peruanus Puente de Chimbo Urbina inopinatus Hato El Fri'o San Antonio Maturin alstoni

-1- I I 9 8 7 6 5 4 3 2 1 0 Mahalanobis D Fig. 12. UPGMA clustering of Mahalanobis distances among 13 population samples of Sigmodon species. The units ofMahalanobis distance, along the horizontal, are within-sample standard deviations adjusted for measurement covariance. Sample sizes and gazetteer references are provided in table 6. 1992 VOSS: SOUTH AMERICAN SIGMODON 25

TABLE 8 TABLE 9 General Size Coefficients of Log-Transformed Interspecific Differences Adjusted for General Sizea Measurements for Four Samples of South Amer- in Twelve Log-Transformed Craniodental Mea- ican Sigmodon Species surements hispidusa alstonib peruanusc inopinatusd Between Sigmodon hispidus and (N= 40) (N= 26) (N= 34) (N= 8) S. S. S. CIL 0.24 0.24 0.27 0.23 alstoni peruanus inopinatus LD 0.34 0.33 0.35 0.36 CIL -0.08 0.14 0.07 LM 0.00 0.10 0.03 0.02 LD 0.11 0.04 0.09 BM1 0.02 0.12 0.02 0.06 LM -0.72 0.27 0.19 LIF 0.31 0.29 0.30 0.28 BM1 -0.32 0.18 0.23 BR 0.25 0.20 0.28 0.20 LIF -0.13 -0.33 -0.14 BPB 0.54 0.41 0.42 0.59 BR -0.30 0.15 0.15 BZP 0.25 0.40 0.45 0.30 BPB 0.12 -0.63 -0.41 LIB 0.11 0.18 0.11 0.10 BZP 0.06 0.31 0.27 DI 0.36 0.29 0.27 0.30 LIB -0.37 -0.10 -0.67 BIT 0.38 0.45 0.36 0.36 DI 0.02 0.28 0.02 ZB 0.17 0.20 0.21 0.16 BIT 0.30 0.39 0.37 a Sample from Montalbin, Venezuelan locality 18. ZB -0.13 0.08 0.18 b Sample from Rancho Grande, Venezuelan locality 8. a Tabulated values are mean differences among the c Sample from Isla Punfa, Ecuadorean locality 13. d species samples listed in the footnotes to table 8, adjusted Sample from Urbina, Ecuadorean locality 4. for general size in separate analyses of covariance, and normalized (so that the sum of the squared coefficients these data is indicated by normalized coef- in each column equals unity). Coefficients are signed to ficients about equal to 0.29 (i.e., the inverse indicate divergence from Sigmodon hispidus. Specimen square-root of the number of variables, see scores on these factors are plotted along the vertical axes Jolicoeur, 1963); measurement variables with of figures 13, 14, and 15. coefficients much smaller than this value exhibit negative allometry, whereas variables sometimes been synonymized with it. At with much larger coefficients exhibit positive constant general size, alstoni differs most allometry. In conformity with results previ- conspicuously by its smaller molar dimen- ously reported for other muroids (Voss, 1988; sions (LM, BM1) and narrower interorbit Voss et al., 1990; Voss and Marcus, in press), (LIB); ofthe few positive coefficients for this molar measurements (LM, BMI), interorbi- comparison, only the value for BIT is note- tal breadth (LIB), and measurements that span worthy, indicating the relatively broader in- the braincase (ZB in this study), have con- cisors of this generally smaller species. The sistently smaller coefficients than diastemal largest general-size-adjusted coefficients con- length (LD), palatal breadth (BPB), and trasting peruanus and hispidus describe the breadth across the incisor tips (BIT). There- much narrower palate (BPB), shorter incisive fore, small (and usually young) specimens foramina (LIF), and broader incisors (BIT) have disproportionately larger molars and of the former. Adjusted for general size, in- neurocrania, but smaller rostra, and narrower opinatus has a much narrower interorbital palates and incisors by comparison with larg- region (LIB), narrower palate (BPB), and er (and usually older) specimens in the same broader incisors (BIT) than hispidus. These sample. Comparisons between samples must inferences from statistical computations can be adjusted for these ontogenetic allometries be visually confirmed by inspection of the in order to identify age-invariant aspects of cranial photographs that accompany the spe- interspecific shape differences. cies accounts below. Mean sample differences adjusted for gen- General size and general-size-adjusted dif- eral size in separate analyses of covariance ference factors provide biologically inter- (table 9) reveal diverse morphometric con- pretable axes for ordinating specimens. For trasts between Sigmodon hispidus and the each ofthe pairwise species comparisons dis- other South American species that have cussed in the preceding paragraph, specimen 26 AMERICAN MUSEUM NOVITATES NO. 3050

-2.5 0 0 00 0 0 8 -2.61 0 0 0 00 00 0 00 a) 0 0 00 0 C 0 00 . U1) -2.7 F -oi:5 a) A U) N -2.8 1- A

A A A A A I AA& A AL -2.9p- A

4.2 4.6 5.0 5.4 5.6 General Size Fig. 13. Specimen scores of Sigmodon hispidus (open triangles, from Venezuelan locality 18) and S. alstoni (open circles, from Venezuelan locality 8) on factors representing general size and general-size- adjusted differences. Overlapping observations (conspecific specimens with the same factor scores) are not plotted separately. scores illustrate broadly overlapping general- cies are defined in the preceding section of size distributions, but divergence in general- this report. Only South American taxa are size-adjusted (age-invariant) aspects ofshape listed in the synonymy for Sigmodon hispi- variation provides a compelling basis for dis- dus; North American and Central American crimination (figs. 13, 14, 15). Although other synonyms are given by Hall (1981). Holo- pairwise ordinations (among samples of al- types ofevery South American species-group stoni, peruanus, and inopinatus) are not doc- taxon except hirsutus were examined in the umented here, the Mahalanobis distances course of this study; these and other specimens from which I recorded data are summarized earlier (in fig. 12) are sufficient listed at the to support the conclusion that their multi- conclusion of each account. variate dispersions are likewise discrete. Sigmodon hispidus Say and Ord (Figures 3-5, 7-11, 16, 17) SPECIES ACCOUNTS Sigmodon hispidus Say and Ord, 1825: 354. The following accounts provide synony- Lasiomys hirsutus Burmeister, 1854: 16. mies, information about type specimens, Lasiuromys hirsutus Giebel, 1857: caption to plate summaries of geographic distributions, lists 5, figure 1 (an inadvertant misuse ofLasiuromys of Deville for Burmeister's species according to diagnostic characters, congeneric compar- Winge, 1941: 153) isons, descriptions of geographic variation, Isothrix hirsutus Trouessart, 1880: 178. and comments on taxonomic problems that Sigmodon bogotensis J. A. Allen, 1 897b: 121. remain to be solved. Character states listed Sigmodon sanctaemartae Bangs, 1898: 189. in the Morphological Diagnosis for each spe- Sigmodon hirsutus Thomas, 1914: 413. l1992 VOSS: SOUTH AMERICAN SIGMODON 27

1.6 0 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0) 1.4 1- 0 u 0 00 0 a) 0 0 0 0 0 0 0

. _ 0

.0N 1.2 A A A A A6 A A A A AA& A A A An A A& As A A A A 1.0 H A A A

4.6 4.8 5.0 5.2 5.4 5.6 General Size Fig. 14. Specimen scores of Sigmodon hispidus (open triangles, from Venezuelan locality 18) and S. peruanus (open circles, from Ecuadorean locality 13) on factors representing general size and general- size-adjusted differences. Overlapping observations (conspecific specimens with the same factor scores) are not plotted separately.

Sigmodon hispidus bogotensis Cabrera, 1961: 508. zuelan states of Barinas, Cojedes, Carabobo, Sigmodon hispidus hirsutus Cabrera, 1961: 508. and Gu'arico. The species is unknown from the Recent fauna of any Caribbean island, TYPE: Not extant, as explained earlier for but Hooijer (1967) reported Pleistocene re- the genus. mains of "Sigmodon cf. hispidus" from Aru- DISTRIBUTION AND SYMPATRY: From about ba. 41°N in the United States, southward through The geographic ranges of hispidus and al- Mexico and Central America (see Hall, 1981) stoni overlap from extreme to South In northeastern Co- America. South America, hispi- lombia (in the western foothills of the Ser- dus occurs in natural or non- anthropogenic rania de Perija) eastward to the low mountains forest habitats below about 3000 m elevation surrounding Lago Valencia in Venezuela, and from the Caribbean coast ofColombia south- southward into the northwestern ward throughout the entire catchment of the Llanos, perhaps as far as the Rio Magdalena,4 and eastward in Rio Apure. The two species coastal low- have been collected sympatrically at El Li- lands and sierras at least as far as Curupao mon and Guamita (both near Venezuelan lo- (66°38'W) in Estado Miranda, Venezuela; cality 8), at Camp south ofthe coastal sierras, hispidus is known Rangel (Venezuelan lo- from the northwestern Llanos in cality 9), at El Trompillo (Venezuelan locality the Vene- 16), and at Hato Itabana (Venezuelan locality 4Cuervo Diaz et al. (1986) also reported hispidus from 21). the valleys of the Rio Cauca and the Rio Patia in Co- MoRPHoLOGIcAL DiAGNosIs: A large spe- lombia, but I have not seen specimens from either region cies of Sigmodon distinguishable from other and none are cited elsewhere in the literature. South American congeners by its long nasal 28 AMERICAN MUSEUM NOVITATES NO. 3050

0 0 0.9 1 0 0 0 0 0 0 0.8 - vV

0.71- -a

a A NN) 0.6 F- A A A A AA6 AA AA6 A A &aAAZA~~~~AA A A a As A A A I- a A A 0.5 Zs

4.5 4.7 4.9 5.1 5.3 5.5 General Size Fig. 15. Specimen scores of Sigmodon hispidus (open triangles, from Venezuelan locality 18) and S. inopinatus (open circles, from Ecuadorean locality 4) on factors representing general size and general- size-adjusted differences. bones; broad interorbital region; long incisive resemble those of typical hispidus in gross foramina; broad and long palatal bridge with- morphology. Karyotypes from Colombian out conspicuous longitudinal grooves; pos- populations are unknown. terior palatal foramina usually enclosed by COMPARISONS: Comparisons between his- the palatine bones; derived stapedial circu- pidus and other South American species are lation; small auditory bullae; narrow, strong- provided in subsequent accounts. ly opisthodont, ungrooved upper incisors; GEOGRAPHIC VARLATION: Evidence for sig- upper molars with opposite, obtusely round- nificant geographic patterns ofmorphological ed cusps; and first mandibular molars usually variation is not apparent among available with four well-developed roots. samples of South American hispidus. Minor These characters are not sufficient to dis- aspects ofpelage pigmentation or fur texture tinguish hispidus from some of its North distinguish a few populations, and specimens American congeners, which instead differ by from the upper Rio Magdalena valley (e.g., other traits, especially karyotypes. Valle de Suaza in table 10) have slightly larger KARYOTYPES: Karyotypes of South Amer- molars and narrower palates than samples ican specimens of hispidus reported by Kib- from the Caribbean coastal plains and sierras, linsky (1969) from Rancho Grande (Vene- but I have been unable to discover any well- zuelan locality 8) and Barinitas (Estado marked regional differences worthy of sub- Barinas, Venezuela, at 84°5'N, 70°25'W) have specific recognition. the same diploid number (2n = 52) and au- REMARKS: South American populations tosomal morphology as karyotypes of puta- with the diagnostic attributes cited above are tive conspecifics from North and Central provisionally referred to hispidus in the ab- America (differences in Y-chromosome mor- sence of any characters to distinguish them phology, however, were remarked by Zim- unambiguously from the typical North merman, 1970). Two hitherto unpublished American form. Although average differ- specimens from Finca El Jaguar (Venezuelan ences in craniodental morphology between locality 45) also have 52 chromosomes that South American and Florida specimens are 1992 VOSS: SOUTH AMERICAN SIGMODON 29

TABLE 10 Summary Statisticsa for Craniodental Measurements (in millimeters) of Adult Sigmodon hispidus from South America Colombia Venezuela Valle de Suazab Mincac Montalb'nd Rancho Grandee (248,132) (158, 112, 2 unk.) (218, 192) (78, 72) CIL 32.4 ± 2.5 (26.2-37.3) 32.2 ± 1.8 (28.9-36.3) 32.6 ± 2.3 (28.4-37.3) 31.7 ± 1.8 (28.1-34.7) LD 9.3 ± 1.1 (6.9-11.6) 9.4 ± 0.8 (8.1-11.0) 9.5 ± 0.9 (7.9-11.4) 9.1 ± 0.7 (7.8-10.4) LM 6.5 ± 0.2 (6.2-6.9) 6.2 ± 0.2 (5.9-6.5) 6.3 ± 0.1 (6.0-6.7) 6.2 ± 0.2 (6.0-6.5) BMI 2.1 ± 0.1 (2.0-2.2) 2.0 ± 0.1 (1.9-2.2) 2.0 ± 0.1 (2.0-2.2) 2.1 ± 0.1 (2.0-2.2) LIF 7.8 ± 0.7 (5.9-9.3) 7.7 ± 0.6 (6.9-8.9) 7.8 ± 0.7 (6.5-9.2) 7.6 ± 0.6 (6.4-8.2) BR 6.1 ± 0.4 (5.0-6.9) 5.8 ± 0.3 (5.2-6.6) 5.9 ± 0.4 (5.1-7.0) 5.8 ± 0.4 (5.1-6.3) BPB 2.8 ± 0.4 (2.0-4.0) 3.2 ± 0.4 (2.7-4.2) 3.3 ± 0.5 (2.4-4.4) 3.2 ± 0.4 (2.4-3.9) BZP 4.1 ± 0.4 (3.2-5.0) 4.0 ± 0.3 (3.4-4.7) 3.8 ± 0.3 (3.2-4.4) 3.9 ± 0.3 (3.4-4.3) LIB 5.4 ± 0.3 (4.9-6.0) 5.6 ± 0.3 (5.1-6.1) 5.4 ± 0.3 (5.0-6.1) 5.3 ± 0.3 (4.9-5.6) DI 2.0 ± 0.2 (1.5-2.2) 1.9 ± 0.2 (1.6-2.2) 1.9 ± 0.2 (1.6-2.3) 1.9 ± 0.1 (1.6-2.0) BIT 2.7 ± 0.3 (2.1-3.3) 2.5 ± 0.2 (2.0-3.0) 2.6 ± 0.3 (2.0-3.1) 2.6 ± 0.2 (2.1-2.8) ZB 19.7 ± 1.2(16.6-22.0) 19.2 ± 0.9 (17.8-21.4) 19.4 ± 1.0 (17.1-21.5) 19.3 ± 1.0 (17.2-21.1) a The sample mean plus or minus one standard deviation, and the observed range (in parentheses). b Colombian locality 12; all measured specimens are from sublocality Naranjal. CColombian locality 14; measured specimens are from sublocalities Minca (N = 15), Onaca (5), Masinga Vieja (5), Bonda (2), and Masinga (1). d Venezuelan locality 18. e Venezuelan locality 8; measured specimens are from sublocalities El Lim6n and Guamita. not wanting, none appears to be sufficient for merits special attention as the oldest epithet diagnosis. Since the South American samples for a South American cotton rat, and as the are morphometrically similar to typical his- second-oldest name in the entire genus. I have pidus (fig. 12), and because the available kar- not seen the type, but Burmeister's (1854) yotypes from North and South America are description and Giebel's illustration (repro- alike, if not identical, in chromosome num- duced in fig. 18), were enough for Thomas ber and morphology, I see no reasonable tax- (1914) to infer that hirsutus is a species of onomic alternative to the arrangement Sigmodon with ungrooved incisors, a deci- adopted herein. sion with which Tate (1932) concurred as I Nevertheless, the hypothesis that a repro- do also. The type (from Maracaibo) is in the ductively continuous series of populations zoological museum of Halle (Mohr, 1941), extends from Colombia to Nebraska will and should be reexamined if the name is re- doubtless provoke future researchers to apply quired by some future taxonomic contingen- more sophisticated molecular and cytological cy. methods than have hitherto been brought to SPECimENS EXAMINED: Colombia -Boyaca, bear on the problem. Such efforts should be Muzo (FMNH 4); Cesar, El Mamon (BMNH encouraged in the expectation that hispidus, 1, MCZ 3), Pueblo Viejo (MCZ 2, including as currently recognized, may prove to consist the topotype of sanctaemartae), San Sebas- ofseveral morphologically cryptic but genet- tian (FMNH 24, MCZ 9, USNM 1), El Salado ically distinct geographical units. The Central (USNM 6), Pueblo Bello (FMNH 2, USNM American populations, in particular, have re- 22); Cundinamarca, Bogota (AMNH 6, ceived no revisionary attention and merit BMNH 3 [including the type of bogotensis]), close scrutiny by any means available. 10 km E Boqueron (AMNH 1), Caparrapi In the event that new characters are dis- (USNM 1); Guajira, La Concepcion (MCZ 1, covered that require recognition of one or USNM 1), Las Marimondas (USNM 18), San more different species for the South Ameri- Antonio (MCZ 1), Sierra Negra (USNM 1); can populations here assigned to hispidus, all Huila, Valle de Suaza (USNM 193), 17.5 km of the species-group names listed in the syn- SE Villavieja (MVZ 1); Magdalena, Agua onymy above are available. Ofthese, hirsutus Dulce (FMNH 1), Bonda (AMNH 13), Cin- 30 AMERICAN MUSEUM NOVITATES NO. 3050

W-,

/V, l! II

A B C D Fig. 16. Dorsal and ventral views of crania (x 1.5) of South American Sigmodon species. A, S. hispidus (USNM 442531); B, S. alstoni (USNM 520676); C, S. peruanus (AMNH 67028); D, S. inopinatus (AMNH 66308). cinati (FMNH 1), Mamatoca (FMNH 4), Caserio Boro (USNM 30), San Miguel Manzanares (AMNH 3), Masinga (FMNH 3, (AMNH 25); Merida, Mesa Bolivar (USNM USNM 1), Masinga Vieja (AMNH 6, USNM 1); Miranda, Curupao (USNM 1); Yaracuy, 1), Minca (AMNH 49, USNM 5), Onaca Finca El Jaguar (AMNH 1, MHNLS 2); Zu- (AMNH 12, BMNH 4), Santa Marta (BMNH lia, Novito (USNM 2). Total = 683. 1); Norte de Santander, Guamalito (USNM 5); Santander, Bucaramanga (MSU 9); To- lima, Mariquita (BMNH 4). Venezuela-Ar- Sigmodon alstoni (Thomas) agua, Rancho Grande (AMNH 9, UMMZ 1, (Figures 9-11, 16, 17) USNM 29), Tiara (MHNLS 18), Camp Ran- Reithrodon alstoni Thomas, 1880: 691. gel (USNM 27); Barinas, Altamira (USNM Sigmomys alstoni Thomas, 1901: 150. 9); Carabobo, El Trompillo (BMNH 10), La Sigmomys savannarum Thomas, 1901: 150. Florida (MHNLS 2), Montalban (USNM 66); Sigmomys venester Thomas, 1914: 412. Cojedes, Tucuragua (MHNLS 1), Hato Ita- Sigmodon alstoni alstoni Cabrera, 1961: 508. bana (MHNLS 8), La Blanquera (MHNLS Sigmodon alstoni savannarum Cabrera, 1961: 508. 1); Distrito Federal, Caracas (AMNH 3, Sigmodon alstoni venester Cabrera, 1961: 508. BMNH 2, USNM 8), San Julian (USNM 1); TYPE: A British Museum specimen (BMNH Falc6n, Cerro Socopo (USNM 2); Guarico, 47.6.23.7), consisting of the fluid-preserved San Juan de los Morros (USNM 1); Lara, carcass, skull, and mandibles of a very old 1 992 VOSS: SOUTH AMERICAN SIGMODON 31

male collected in Venezuela by D. Dyson sometime prior to 1880. The skull is missing the tip ofthe left nasal, both zygomatic arches, the right auditory bulla, and most of the basicranium; the heavily worn molars retain A no details of occlusal morphology. Thomas (1901) restricted the type locality to Cumana (Venezuelan locality 41). DISTRIBUTION AND SYMPATRY: Apparently rare in the valley ofthe Rio Cesar in extreme northeastern Colombia and in the Lake Ma- racaibo basin of Venezuela from which regions only a few specimens are known, but common throughout unforested and deforested habitats below about 1300 m elevation east of the Andes and south of the coastal cordilleras, including all ofthe Llanos,5 some B isolated savannas in the Venezuelan state of Amazonas, deforested regions in the north- :f eastern Venezuelan highlands, the Gran Sa- bana in southeastern Venezuela, the savan- e A A' -- nas of the upper Rio Branco in Brazil, the Rupununi savannas of Guyana, the coastal and interior savannas of Surinam, and con- tiguous savannas in the upper Rio Paru drainage of the Brazilian state of Para. No specimens have yet been collected in French Guiana, but alstoni should occur in the coastal savannas there and in the neighboring Bra- C zilian state of Amapfi. The species is known from cave fossils presumed to be oflate Pleis- tocene age from Tobago (Eshelman and Mor- gan, 1985) but is apparently not part of any Recent insular fauna. Geographic overlap and sympatry between alstoni and hispidus have already been described in the account for the latter species. MORPHOLOGICAL DIAGNosIs: A small species of Sigmodon distinguishable from other congeners by its broad, strongly opisthodont, and deeply grooved upper incisors. Addi- D 5 The distribution ofalstoni in the Llanos ofColombia is more extensive than that suggested by the single locality plotted in fig. 1. Specimens in the collections of the Universidad del Valle (in Cali) document the presence of this species at San Juan de Arama and Puerto Gaitan (in Departamento Meta), and from the vicinity Fig. 17. Lateral views ofcrania (x 1.5) ofSouth of Puerto Carreiio (near the confluence of the Rio Meta American Sigmodon species. Species and and the Orinoco in Comisaria Vichada) according to M. speci- Alberico (in litt.). Groove-toothed cotton rats probably mens are the same as those in fig. 16. range throughout the eastern Colombian savannas, which extend from the Venezuelan border south to the Rio Guaviare. 32 AMERICAN MUSEUM NOVITATES NO. 3050

TABLE 11 Diploid Chromosome Counts of Venezuelan Specimens of Sigmodon alstoni Diploid Locality Field number Museum number Sex Na countb Hato El Frioc RSV 1197 USNM 448698 C 19 82 HatoElFrioc RSV 1198 MHNLS 7913 9 15 78 Hato El Frioc RSV 1199 MHNLS 7914 Q 10 78 Hato El Frioc RSV 1210 USNM 448639 2 10 80 Hato El Frioc RSV 1211 MHNLS 7919 d 12 78 Hato El Frioc RSV 1217 USNM 448640 d11 80 San Ignacio Yuruanid RSV 1018 USNM 448632 2 12 78 San Ignacio Yuruanid RSV 1062b MHNLS 7900 2 8 80 San Ignacio Yuruanid RSV 1234 AMNH 257341 2 10 80 Finca Vuelta Largae RSV 1378 AMNH 257342 2 8 82 Finca Vuelta Largae RSV 1379 MHNLS 8151 2 12 82 a Number of cells analyzed. b Modal values. c Venezuelan locality 6. d Venezuelan locality 14. e Venezuelan locality 43. tionally, alstoni differs from other South locality 36) karyotyped by J. W. Bickham and American species by its long nasal bones; un- M. D. Engstrom had a modal count of 80 constricted interorbital region; incisive fo- chromosomes (Bickham, in litt.) ramina that usually extend to or between the COMPAIuSONS: Hershkovitz (1955) sug- molar alveoli; unconstricted palatal bridge; gested that the genus Sigmomys was based posterior palatal foramina always bordered on specimens of South American Sigmodon by both maxillary and palatine bones; de- hispidus with grooved incisors, but the con- rived stapedial circulation; small auditory specificity of alstoni and hispidus can be re- bullae; upper molars with opposite, obtusely jected unequivocally on several grounds. rounded cusps; and first mandibular molars Firstly, the upper incisor grooves do not have with only three well-developed roots. the aspect of a population polymorphism: KARYOTYPES: Diploid counts of 80 to 82 there are no intermediate morphologies, chromosomes were reported for alstoni by specimens from the broad geographic zone Reig et al. (1990) who cited Reig (1987) as of overlap having either deep, well-defined the original source ofthis observation. I found grooves or none at all. Secondly, other mor- no reference to Sigmodon karyotypes in Reig phological characters, especially measure- (1987), but Reig (1986) reported a diploid ments, are correlated with the presence of number of 82 for alstoni, citing unpublished grooved incisors; this has already been dem- data; the geographic origin ofthe karyotyped onstrated for the multivariate data (figs. 12, specimen(s) was not stated. 13), but is also obvious from univariate com- Eleven specimens that my colleagues and parisons (e.g., of hindfoot length in table 1; I karyotyped in the field at three Venezuelan and of molar length, LM, in tables 10 and localities have modal counts of78 to 82 chro- 12). The two incisor phenotypes also differ mosomes (table 11). The modes for most significantly in frequencies of alternative specimens are well defined and appear to be morphologies ofthe posterior palatal foram- inversely correlated with the number oflarge ina (table 4) and of the lingual root of ml metacentric (or submetacentric) elements (table 5). Finally, conspicuous differences in present; Robertsonian polymorphism is a karyotypes (2n = 52 for smooth-toothed rats, plausible explanation for this variability, but and 2n = 78 to 82 for groove-toothed spec- karyotypes with such high counts are difficult imens) would appear to preclude any effective to analyze and this interpretation is conjec- genetic exchange between the two forms where tural. they occur sympatrically. A single male from Caripito (Venezuelan No close comparisons are required with 1 992 VOSS: SOUTH AMERICAN SIGMODON 33

TABLE 12 Summary Statisticsa for Craniodental Measurements (in millimeters) of Adult Sigmodon alstoni Rancho Grandeb Hato El Frioc San Antonio Maturind Kanuku Mountainse (176, 92) (96, 72) (46, 82) (76, 72) CIL 28.4 ± 1.7 (25.0-31.2) 29.8 ± 1.4 (27.2-31.4) 28.5 ± 2.1 (25.8-31.2) 28.5 ± 1.5 (25.8-30.6) LD 8.3 ± 0.7 (6.9-9.4) 8.8 ± 0.6 (7.8-9.6) 7.9 ± 0.9 (6.8-9.2) 8.1 ± 0.6 (7.0-9.2) LM 5.2 ± 0.2 (4.8-5.6) 5.5 ± 0.2 (5.2-5.8) 5.7 ± 0.1 (5.6-5.9) 5.4 ± 0.1 (5.2-5.6) BM1 1.8 ± 0.1 (1.6-2.0) 2.0 ± 0.1 (1.9-2.1) 2.0 ± 0.1 (1.9-2.2) 2.0 ± 0.1 (1.9-2.1) LIF 6.6 ± 0.5 (5.4-7.5) 7.1 ± 0.4 (6.4-7.8) 6.5 ± 0.7 (5.6-7.6) 6.7 ± 0.4 (6.0-7.2) BR 4.9 ± 0.2 (4.2-5.4) 5.0 ± 0.4 (4.4-5.5) 4.7 ± 0.3 (4.2-5.2) 5.0 ± 0.3 (4.3-5.6) BPB 2.6 ± 0.3 (2.2-3.2) 2.9 ± 0.4 (2.2-3.6) 2.5 ± 0.4 (2.0-3.2) 2.7 ± 0.3 (2.1-3.4) BZP 3.3 ± 0.3 (2.5-3.9) 3.4 ± 0.3 (2.6-3.7) 3.2 ± 0.4 (2.8-3.8) 3.4 ± 0.2 (3.0-3.6) LIB 4.6 ± 0.2 (4.0-5.0) 4.8 ± 0.1 (4.5-5.0) 4.6 ± 0.2 (4.2-5.1) 4.5 ± 0.2 (4.2-5.0) DI 1.6 ± 0.1 (1.4-1.9) 1.8 ± 0.1 (1.4-2.0) 1.7 ± 0.2 (1.4-2.0) 1.8 ± 0.1 (1.6-1.9) BIT 2.3 ± 0.3 (1.7-2.8) 2.6 ± 0.2 (1.9-2.8) 2.3 ± 0.3 (1.9-2.9) 2.7 ± 0.2 (2.2-3.0) ZB 17.2 ± 0.9 (15.1-19.0) 17.8 ± 0.8 (16.2-18.7) 17.3 ± 1.0 (15.9-19.0) 17.2 ± 0.7 (16.0-18.6) a The sample mean plus or minus one standard deviation, and the observed range (in parentheses). b Venezuelan locality 8; measured specimens are from sublocality El Lim6n. c Venezuelan locality 6. d Venezuelan locality 39. e Guyana locality 2; includes the type of Sigmomys savannarum. peruanus and inopinatus, geographically re- tention of venester and savannarum as sub- mote smooth-toothed species differing from species serves no purpose at the present time. alstoni in many other attributes that are listed Both names are available, however, should in their respective diagnoses below. significant geographic patterns of molecular GEOGRAPHIC VARIATION: Available popu- or cytological divergence be discovered in fu- lation samples exhibit modest variation in ture studies. pelage color and texture. Lowland popula- REMARKS: Thomas (1901) erected Sigmo- tions (e.g., those from the Rupununi savan- mys in recognition of the many morpholog- nas of Guyana, from the upper Rio Branco ical characters that distinguish alstoni from drainage of Brazil, and from the lower Lla- Reithrodon, to which genus the species was nos) have short, close, coarse fur, whereas first referred. "Altogether," he remarked, "it specimens collected at higher elevations (e.g., seems probable that this form is rather a those from the Gran Sabana and the eastern groove-toothed Sigmodon than any close re- Caribbean highlands of Venezuela) have lation to Reithrodon" (op. cit.: 150). Hersh- softer, longer, and denser pelage. Skins from kovitz (1955, 1966b) and Cabrera (1961) Guyana and Brazil are also paler than other concluded that alstoni is only what Thomas specimens, a contrast that perhaps is corre- suggested, a species ofSigmodon with grooved lated with soil color. incisors, and treated Sigmomys as a subjec- Some average differences in craniodental tive junior synonym. Persistent opinions to measurements among available population the contrary, that Sigmomys should be re- samples of alstoni are noteworthy (e.g., the tained as a valid taxon (Husson, 1978; Wil- contrast in molar toothrow length between liams et al., 1983), are based on the supposed Rancho Grande and San Antonio Maturin, "importance" of incisor grooves for muroid table 12) but do not appear to represent any classification. Special recognition for favored consistent geographic pattern. Since the mag- characters, however, has no place in rational nitude of overall morphometric divergence systematic practice: grooved incisors are among these samples is similar to that ob- merely an autapomorphy of the species al- served within other South American Sig- stoni and no tangible biological purpose is modon species (fig. 12), and because popu- served by retaining a monotypic higher taxon lation differences in qualitative aspects of to reflect this solitary distinction. cranial morphology are very slight, the re- SPECIMENS EXAMINED: Brazil -Pard, Serra 34 AMERICAN MUSEUM NOVITATES NO. 3050 do Tumucumaque (USNM 1); Roraima, Fre- Sigmodon peruanus peruanus Reig, 1986: 407. chal (AMNH 1), Limo (AMNH 18), Serra da Sigmodon peruanus simonsi Reig, 1986: 407. Lua (BMNH 14, FMNH 10, MCZ 1). Colom- Sigmodon peruanus chonensis Reig, 1986: 407. bia-Guajira, Villanueva (USNM 1); Meta, Sigmodon peruanus puna Reig, 1986: 407. Los Micos (FMNH 8). Guyana-Rupununi, TYPE: A specimen in the American Mu- Dadanawa (ROM 11), Kanuku Mountains seum of Natural History consisting of the (AMNH 2, BMNH 12 [including the type of separately cataloged skin (AMNH 11818) and savannarum]), Nappi Creek (ROM 10). Su- skull with mandibles (AMNH 101 10) of an rinam -Nickerie, Sipaliwini-savanne-vlie- old adult female collected by 0. T. Baron on geveld (RMNH 2); Para, Zanderij (RMNH 25 May 1895 at Trujillo, Departamento La 2). Venezuela -no other data (BMNH 1, the Libertad, Peru. The skin is intact and un- type of alstoni); Amazonas, Paria Grande damaged but the skull lacks both auditory (USNM 2), Rueda (USNM 2), San Juan bullae, the basioccipital, and much ofthe lat- Manapiare (MHNLS 3, USNM 4); Anzoa- eral wall of the braincase below the squa- tegui, Rio Oritupano (MHNLS 1); Apure, En- mosal root of the left zygomatic arch; the trada La Yaguita (MHNLS 1), Hato Cariben pterygoid region is largely destroyed as well, (USNM 2), Hato El Frio (AMNH 4, MHNLS and the angles ofboth mandibles are broken. 15, USNM 8), 41 km NW Puerto Paez The molar dentition is worn almost to the (USNM 7); Aragua, Rancho Grande (USNM roots and retains only traces of the occlusal 28), Camp Rangel (USNM 1); Bolivar, El Lla- pattern. gual (AMNH 1), Hato La Florida (USNM 4), DISTIuBUTION AND SYMPATRY: In xeric, Mount Roraima (AMNH 19), San Ignacio semiarid, and deforested landscapes of the Yuruani (AMNH 1, MHNLS 1, USNM 2), Pacific coastal plain and adjoining Andean Vetania (USNM 3); Carabobo, El Trompillo foothills below about 1600 m elevation in (BMNH 1, the type ofvenester); Cojedes, Hato western Ecuador and northwestern Peru. The El Tirado (MHNLS 2), Hato Itabana known range extends in latitude from about (MHNLS 8), Hato Los Caballos (MHNLS 1); 8°S almost to the equator. Guarico, Calabozo (USNM 1), Hato La Mer- Sigmodon peruanus is not known to be rereiia (MHNLS 3), Parcela 200 (MHNLS sympatric with any congeneric species. 12); Monagas, Caripe (USNM 4), Caripito MORPHOLOGICAL DIAGNOSIS: A large spe- (AMNH 1, TCWC 1), Cocollar (AMNH 5, cies ofSigmodon distinguished from all other BMNH 1), Hato Mata de Bejuco (USNM 16), congeners by its complete, primitive stape- San Antonio Maturin (AMNH 23), Campo dial circulation. From other South American Alegre (BMNH 1); Portuguesa, Los Ma- species, peruanus can also be distinguished mones (USNM 1); Sucre, 40 km NW Cari- by its long nasal bones; unconstricted inter- pito (TCWC 1), Cumana (USNM 5), Cu- orbital region; short incisive foramina (sel- manacoa (AMNH 13), Finca Vuelta Larga dom extending to or between the molar al- (AMNH 1, MHNLS 2), Manacal (USNM 5); veoli); very narrow, short, and grooved palatal Zulia, Empalado Savannas (FMNH 3), Rio bridge with deep posterolateral sulci; poste- Aurare (FMNH 4), Rio Cogollo (FMNH 7, rior palatal foramina always bordered by both MCZ 1). Total = 326. maxillary and palatine bones; large auditory bullae; broad, strongly opisthodont, un- Sigmodon peruanus J. A. Allen grooved upper incisors; upper molars with (Figures 6-8, 10, 11, 16, 17) alternate, acutely angled cusps; and first mandibular molars usually with only three well- Sigmodon peruanus J. A. Allen, 1897a: 118. developed roots. Sigmodon simonsi J. A. Allen 1901: 40. Sigmodon puna J. A. Allen, 1903: 99. KARYOTYPES: Unknown. Sigmodon chonensis J. A. Allen, 1913: 479. COMPARIONS: Considered conspecific with Sigmodon lonnbergi Thomas, 1921: 448. hispidus by Cabrera (1961), peruanus nev- Sigmodon hispidus chonensis Cabrera, 1961: 508. ertheless differs from that taxon in many re- Sigmodon hispidus peruanus Cabrera, 1961: 509. spects. In addition to possessing a pattern of Sigmodon hispidus puna Cabrera, 1961: 509. stapedial arterial supply unique in the genus, Sigmodon hispidus simonsi Cabrera, 1961: 509. peruanus can be distinguished from hispidus 1 992 VOSS: SOUTH AMERICAN SIGMODON 35

TABLE 13 Summary Statisticsa for Craniodental Measurements (in millimeters) of Adult Sigmodon peruanus Isla PunkP Chongoncitoc Puente de Chimbod (176, 169, 1 unk.) (106, 62) (76, 59) CIL 31.8 ± 1.5 (28.5-35.2) 31.4 ± 1.9 (28.7-34.2) 30.9 ± 1.5 (28.3-33.2) LD 8.7 ± 0.6 (7.6-9.9) 8.8 ± 0.8 (7.5-10.0) 8.8 ± 0.7 (7.6-10.1) LM 7.0 ± 0.2 (6.6-7.3) 6.8 ± 0.3 (6.3-7.2) 6.7 ± 0.2 (6.3-7.0) BM1 2.2 ± 0.1 (2.0-2.3) 2.1 ± 0.1 (2.0-2.3) 2.1 ± 0.1 (2.0-2.2) LIF 6.3 ± 0.5 (5.3-7.3) 6.4 ± 0.7 (5.4-7.6) 6.6 ± 0.5 (5.7-7.4) BR 5.7 ± 0.3 (5.3-6.7) 5.7 ± 0.4 (5.1-6.4) 5.6 ± 0.3 (5.1-6.1) BPB 2.2 ± 0.2 (1.9-2.9) 2.2 ± 0.3 (1.7-2.8) 2.5 ± 0.4 (1.9-3.0) BZP 3.9 ± 0.4 (3.2-4.5) 3.9 ± 0.3 (3.5-4.4) 4.0 ± 0.3 (3.5-4.5) LIB 5.0 ± 0.2 (4.6-5.6) 5.0 ± 0.3 (4.5-5.4) 5.0 ± 0.3 (4.6-5.4) DI 1.9 ± 0.1 (1.7-2.2) 1.8 ± 0.1 (1.6-2.0) 1.9 ± 0.1 (1.6-2.0) BIT 2.7 ± 0.2 (2.3-3.2) 2.7 ± 0.2 (2.3-3.0) 2.9 ± 0.1 (2.7-3.2) ZB 18.9 ± 0.8 (16.6-20.5) 18.5 ± 1.0 (17.2-20.4) 18.4 ± 0.6 (17.1-19.4) a The sample mean plus or minus one standard deviation, and the observed range (in parentheses). b Ecuadorean locality 13; measured specimens are from both sublocalities. c Ecuadorean locality 11. d Ecuadorean locality 3. by craniodental proportions that include ditory bullae, more strongly opisthodont up- shorter incisive foramina, a conspicuously per incisors, and first maxillary molars usually narrower and shorter palate, broader zygo- with three (versus four) well-developed roots. matic plates, larger auditory bullae, and GEOGRAPHIC VARIATION: Skins from the broader incisors; these contrasts are repre- island of Pun'a and from northwestern Peru sented in the morphometric analyses de- are paler than most specimens from the Ecu- scribed earlier (except for palatal length and adorean mainland, but there is no evidence bullar size, which were not measured), and that this color variation is anything more than can be confirmed visually from cranial pho- concealing pigmentation that may be ex- tographs (figs. 16, 17). Also, the unworn up- pected to vary clinally with local soils. None per molars of peruanus have alternating, ofthe cranial dimensions cited by Allen (1901, sharp-angled cusps, whereas those ofhispidus 1903, 1913) and Thomas (1921) in their de- and other Sigmodon species have opposite, scriptions of nominal species here synony- obtusely rounded cusps. Other comparisons mized with peruanus exceed the normal range involve qualitative trait frequencies that are ofindividual and ontogenetic variation with- less satisfactory for diagnosis but which in such large samples as those from Chon- nonetheless document substantial differ- goncito, Puente de Chimbo, and Isla Puna ences. The posterior palatal foramina of his- (table 13). Close study of the available ma- pidus are usually contained only by the pal- terial suggests that, with the exception ofpel- atine bones, but those ofperuanus are always age color, this is a phenotypically homoge- bordered both by maxillaries and palatines neous species with no natural subdivisions (table 4). Also, the first maxillary molars of worthy of trinomial recognition. peruanus usually have only three well-devel- SPECIMENS EXAMnNED: Ecuador -Azuay, oped roots, by contrast with the four well- Tunguilla Valley (FMNH 6); Chimborazo, developed roots that usually anchor this tooth Puente de Chimbo (AMNH 40); El Oro, Pie- in hispidus (table 5). dras (FMNH 1), Portovelo (AMNH 1), Santa From geographically adjacent populations Rosa (AMNH 5, BMNH 1); Guayas, Balzar of inopinatus, peruanus is easily distin- Mountains (BMNH 1), Bucay (AMNH 4), guished by its longer nasal bones, much Cerro Manglaralto (AMNH 6), Hacienda El broader interorbital region, shorter incisive Refugio (USNM 1), Chongon (BMNH 1), foramina, narrower palate, primitive stape- Chongoncito (AMNH 48), Guayaquil dial circulation (versus derived), larger au- (AMNH 3, BMNH 1, USNM 1), Duran 36 AMERICAN MUSEUM NOVITATES NO. 3050

(AMNH 5), G and Q RR kilometer 8 (AMNH 4), Puna (AMNH 1 1, BMNH 6 [including the type ofpuna]), San Ramon (AMNH 24), Pacaritambo (MNHN 1); Loja, Casanga Val- ley (AMNH 7), Malacatos (FMNH 1), Pindal (USNM 1); Los Rios, Limon (AMNH 6), Quevedo (BMNH 1 [the type of lonnbergi]), Rio Babahoyo (MNHN 6), San Carlos (MSU 2), Vinces (AMNH 13, USNM 2); Manabi, Chone (AMNH 10 [including the type ofcho- nensis]), Rio de Oro (AMNH 1). Peru-La Libertad, Pacasmayo (FMNH 3), Trujillo (AMNH 1 [the type ofperuanus], USNM 4); Lambayeque, Eten (AMNH 3, BMNH 2 [including the type ofsimonsi]); Piura, El Arenal (USNM 2), Huasimo (FMNH 1), Jilili (MCZ 1), Lancones (USNM 1); Tumbes, El Sauce (USNM 3), Matapalo (FMNH 1), Positos (FMNH 5), Tumbes (USNM 1). Total = 248. Sigmodon inopinatus Anthony

(Figures 6, 10, 11, 16, 17) -~~~~ ) cZ Sigmodon inopinatus Anthony, 1924: 3. Sigmodon hispidus inopinatus Cabrera, 1961: 509. TYPE: A specimen in the American Mu- seum of Natural History (AMNH 66310) consisting of the skin, skull, and mandibles Fig. 18. Figure 1 from plate V ofGiebel (1857), ofan adult female collected by G. H. H. Tate illustrating the skull and mandibles of "Lasiuro- on 26 October 1923 at Urbina, Provincia mys" hirsutus (= Lasiomys hirsutus Burmeister). Chimborazo, Ecuador. The skin and skeletal materials are undamaged. ungrooved upper incisors; upper molars with DISTRIBUTION AND SYMPATRY: Known opposite, obtusely rounded cusps; and first from only two localities at high elevations in mandibular molars usually with four well- the Ecuadorean Andes. The type series was developed roots. collected at 11,400 ft (3508 m) near Chim- KARYOTYPES: Unknown. borazo in the province of that name, and COMPARIsONs: Sigmodon inopinatus dif- several additional examples have recently fers from hispidus by its shorter nasals, very been collected between 3750 and 3800 m near constricted (versus broad) interorbital region, Cuenca in Provincia Azuay. narrower and shorter palate, posterior palatal Sigmodon inopinatus is not known to be foramina usually bordered by both maxillary sympatric with any congeneric species. and palatine bones (versus usually enclosed MORPHOLoGICAL DIAGNOsIs: A large spe- in the palatines), and broader, more procum- cies of Sigmodon distinguished from other bent incisors. congeners by its short nasal bones (exposing Comparisons with the geographically ad- the incisors to dorsal view); very narrow in- jacent species peruanus were provided in the terorbital region; long incisive foramina (usu- preceding account. ally, but not always extending between the GEOGRAPHIC VARIATION: The material molar alveoli); unconstricted palatal bridge; available from Las Cajas conforms closely posterior palatal foramina usually bordered with the type series in measurements and both by maxillaries and palatines; derived qualitative characters. The only discrepancy stapedial circulation; small auditory bullae; worthy of note is the interorbital measure- procumbent (but still weakly opisthodont), ment (LIB) ofthe single adult Las Cajas skull, 1 992 VOSS: SOUTH AMERICAN SIGMODON 37

TABLE 14 ecological information are those collected by Summary Statisticsa for Craniodental Measure- the Smithsonian Venezuelan Project from ments (in millimeters) of Adult Sigmodon inopi- 1965 to 1968; Handley (1976: 53) summa- natus rized data from 326 specimens: Las Cajasc Found ... near streams, irrigation ditches, and other Urbinab (BMNH in sites (28, 69) 82.814 d) moist areas (91 percent) and only rarely dry (9 percent); usually in grass, but rarely in weeds, bush- CIL 32.6 (29.4-34.8) 34.2 es, and houses; in croplands (44 percent), savannas LD 9.5 (8.0-10.6) 10.5 and pastures (31 percent), yards and orchards (16 per- LM 6.7 (6.6-6.8) 6.8 cent), and in grassy openings in forest (9 percent). BM1 2.2 (2.1-2.4) 2.2 LIF 7.3 (6.4-8.1) 7.8 Noteworthy in this laconic synopsis, as in BR 6.0(5.5-6.5) 6.6 numerous ecological descriptions of North BPB 2.7 (2.3-3.4) 3.0 and Central American populations (e.g., Al- BZP 4.0 (3.4-4.6) 3.9 drich and Bole, 1937; Bailey, 1905; Goertz, LIB 4.3 (4.0-4.4) 4.8 1964; Goertz and Long, 1973; Goldman, DI 1.9 (1.6-2.0) 2.0 1920; Kaufman and Fleharty, 1974; Odum, BIT 2.8 (2.3-3.1) 3.3 ZB 20.2 (18.7-21.4) 21.6 1955), is an affinity for natural or anthropogenic nonforest habitats that are usually a The mean and the observed range only are provided dominated by grass. All South American col- for the Urbina sample; too few specimens are available lection localities for this species are either in to provide a meaningful estimate of the standard devi- where predom- ation. regions nonforest vegetation b Ecuadorean locality 4. inates for edaphic or climatic reasons (such c Ecuadorean locality 1. as the northwestern Llanos, and the semiarid upper Rio Magdalena valley; see Voss, 1991) or regions where extensive deforestation has which somewhat exceeds the range of vari- accompanied human settlements; no speci- ation among the Urbina specimens (table 14). mens are known to have been collected in SPECIMENS EXAMINED: Ecuador -Azuay, undisturbed rain forest. Fifty South Ameri- Las Cajas National Park (BMNH 4); Chim- can collection localities range in elevation borazo, Urbina (AMNH 12, including the from near sea level to about 2600 m, but type). Total = 16. hispidus is most often taken between 500 and 1500 m (fig. 19). NATURAL HISTORY At El Limon and Guamita (both near Ven- ezuelan locality 8), C. 0. Handley, Jr. and D. South American species ofSigmodon have I. Rhyme collected hispidus and alstoni to- not been the subjects of any comprehensive gether, in traplines set in fallow agricultural natural history studies, and such information fields; no microhabitat differences were ap- as exists must be gleaned from a few papers from their field observations (Han- on special ecological topics, some general fau- parent nal reports that include sparse observations dley, personal commun.). Elsewhere, the cir- on habitats, and the unpublished fieldnotes cumstances ofsympatry remain undescribed, ofcollectors. The latter, quoted or abstracted and future collectors should be alert to eco- below, are preserved in the archives of the logical nuances that might distinguish the two Department ofMammalogy at the American species where they occur together. and the All of the 326 specimens of hispidus ob- Museum ofNatural History Division tained by the Smithsonian Venezuelan pro- ofMammals at the Field Museum ofNatural ject (Handley, 1976) were taken on the History. ground, as were three specimens that I collected at Finca El Jaguar (Venezuelan locality SPECIES DEscRIPrIoNs 45). In North and Central American habitats Sigmodon hispidus: Despite abundant the species may be active both in the daytime South American material of this species in and at night (Aldrich and Bole, 1937; Cam- museums, the only series accompanied by eron et al., 1979; Harper, 1927; Hoffmeister, 38 AMERICAN MUSEUM NOVITATES NO. 3050

_ hispidus (N = 50)

*-- alstoni (N = 42)

---- peruanus (N = 26)

inopinatus (N = 2) mm

0 5 3100 0 500 1000 1500 2000 2500 3000 3500 4000 _~ = 5 records Elevation (meters) Fig. 19. Frequency histograms of elevation records for South American species of Sigmodon.

1986; Lowery, 1974), and is said to construct hispidus have been recovered from the stom- runways (Aldrich and Bole, 1937; Bailey, achs of the fox Cerdocyon thous and the oc- 1905; Goldman, 1920; Harper, 1927), but no elot Felis pardalis (Bisbal, 1986; Bisbal and published observations are available to con- Ojasti, 1980), but the species is probably im- firm these behavioral traits for South Amer- portant in the diets of many other sympatric ican populations. The diet of hispidus in the vertebrate carnivores. Llanos consists principally of the green parts Arthropod ectoparasites of Sigmodon his- of grasses and herbaceous dicots (Vivas and pidus collected by the Smithsonian Venezu- Calero, 1988), a result consistent with dietary elan Project (Handley, 1976) were described analyses ofNorth American populations (e.g., by Furman (1972), Jones et al. (1972), John- Fleharty and Olson, 1969; Kincaid and Cam- son (1972), Tipton and Machado-Allison eron, 1982). (1972), Brennan and Reed (1975), and Saun- Information from a few field studies ofde- ders (1975). mography and reproduction in Venezuela was Sigmodon alstoni: The groove-toothed cot- summarized by Vivas and Calero (1985, ton rat is usually encountered in open, grassy 1988): reproductive activity in local popu- habitats; examples typically cited in the lit- lations of hispidus is apparently continuous erature include savannas, airfields, pastures, throughout the annual cycle of flooding and overgrown garden plots, and weedy roadsides drought that characterizes the lower Llanos, (Handley, 1976; Husson, 1978; Ib'afiez and but may be most intense in the rainy season Moreno, 1982; O'Connell, 1981; Osgood, when population densities also seem highest. 1912; Tate, 1939; Vivas, 1986; Williams et In South America, remains of Sigmodon al., 1983). Descriptions oftwo localities where 1 992 VOSS: SOUTH AMERICAN SIGMODON 39

Fig. 20. The lower Llanos near Hato El Frio (Venezuelan locality 6). During the wet season (when this photograph was taken) Sigmodon alstoni inhabits unflooded grassy bancos and the weedy margins of matas (islands of low forest in the background); during the dry season, however, S. alstoni occurs throughout the savanna. Photographed in September 1987. my colleagues and I collected alstoni in Ven- under dense grass, in unflooded parts ofthe savanna. ezuela serve to illustrate the diversity of nat- Another muroid species, Zygodontomys brevicauda, ural environments that the species occupies. used the same runways and was always caught in greater numbers [abstracted from my fieldnotes; also HATO EL FRIO: This large cattle ranch at 60 m el- see Voss, 1991, for references and a more detailed evation near the Rio Apure (fig. 1, Venezuelan locality site description]. 6) is representative of much of the lower Llanos in SAN IGNACIO YURUANI: The town of San Ig- climate and vegetation. The annual average temper- nacio Yuruani (fig. 1, Venezuelan locality 14) is sit- ature is 27°C and virtually all ofthe total annual pre- uated at 850 m elevation in the Gran Sabana, a rolling cipitation (probably about 1500 mm on average) oc- landscape of savannas and forests drained by the up- curs from April to November; in the dry season, per Rio Caroni (fig. 21). The average annual precip- December through March, there is scarcely any rain- itation at San Ignacio is about 2100-2200 mm with fall (see Voss, 1991, for references). The landscape some monthly variation but no extremely dry season; consists of seasonally inundated savannas (fig. 20) the average annual temperature is 21-23°C. The town dissected by rivers and streams with wooded banks was our base camp from 9-27 June 1986 and from and scattered with islands of mata vegetation (a sem- 20 June to 7 July 1987, during which time we collected ideciduous woodland oflow stature) growing on slightly only six specimens of Sigmodon alstoni. All were elevated soils. Most of the savanna is flooded during trapped on the ground in dense grass, either in open the rainy season, when only the matas and raised, savanna or in arbustales, extensive thickets of mixed grassy ridges known as bancos provide natural refugia evergreen shrubs and small trees interspersed with for nonaquatic organisms. We collected 17 specimens grassy openings. Two other muroid species, Akodon of Sigmodon alstoni on two brief visits to Hato El cf. urichi and Zygodontomys brevicauda, occurred Frio, from 2-8 August 1986 and from 3-7 September syntopically with S. alstoni and were captured much 1987. All were taken on the ground, along runways more frequently: the former in both savanna and ar- 40 AMERICAN MUSEUM NOVITATES NO. 3050

Fig. 21. The Gran Sabana near San Ignacio Yuruani (Venezuelan locality 14). Open savannas dominated by the grass Trachypogon plumosus and scattered groves of moriche palms (Mauritia flexuosa) are characteristic of the region, as are the numerous low termite mounds in the background. Here Sigmodon alstoni was collected in meadows of dense grass in valley bottoms with rich organic soils, and in grassy openings in thickets of evergreen shrubs. Photographed in June 1986.

bustales, the latter only in arbustales [abstracted from 1912; Tate, 1939; Vivas et al., 1986) and are my fieldnotes; also see Voss, 1991, for references and said to construct runways (Tate, 1939; more detailed habitat accounts]. O'Connell, 1982). Ib-anez and Moreno (1982) concluded that this species is "tipicamente Sigmodon alstoni is a lowland species with herbivora" based on their examination of a known elevational range from sea level to stomach contents, and captives offered a va- only 1300 m (fig. 19). Tate (1939) remarked riety offoods collected from natural habitats that alstoni was found in savannas at the base exhibit a strong preference for grass (O'Con- of Mt. Roraima but not on the upper slopes nell, 1981). where grasslands extend to at least 1600 m. Autopsies of 156 specimens kill-trapped in All of the 140 specimens of alstoni ob- the lower Llanos ofEstado Apure, Venezuela, tained by the Smithsonian Venezuelan pro- documented pregnancies throughout the an- ject (Handley, 1976) were collected on the nual cycle of flooding and drought that char- ground. Strictly terrestrial habits are also sug- acterizes this extremely seasonal environ- gested by my field observations at the local- ment, but reproductive activity may be least ities described above, by O'Connell's (1981) in the dry season; the number of embryos trapping results, and by the absence of any counted in utero ranged from two to eight, reference to other substrates in the literature with a mean of 4.4 (Iba-nez and Moreno, cited earlier. Groove-toothed cotton rats are 1982). The demographic studies of O'Con- active in broad daylight as well as nocturnally nell (1981) and Vivas (1986) did not include (Ib'ainez, 1980; O'Connell, 1981; Osgood, reproductive autopsies, but their data also 1 992 VOSS: SOUTH AMERICAN SIGMODON 41 suggest continuous reproductive activity in omys [caliginosus], are quite generally distributed. Llanoan populations. Sigmodon [peruanus] on the other hand, I have only Remains of Sigmodon alstoni have been found in the second-growth land close to the hacienda, recovered from the stomachs or scats of sev- and in the younger part ofthe platanal [a garden plot] across the river, i.e., that part recently cleared, in eral mammalian carnivores including foxes which the logs have not decomposed [Tate's notes for (Cerdocyon thous), ocelots (Felis pardalis), 29 August]. jaguarundis (Felis yagouaroundi), and gri- sons (Galictis vittata)(Bisbal, 1986; Bisbal and At another camp on the 1922 expedition, Ojasti, 1980; Sunquist et al., 1989). White- near Chongoncito (Ecuadorean locality 11), tailed kites (Elanus leucurus) eat groove- Tate and Wickenheiser collected 48 speci- toothed cotton rats (Husson, 1978) and most mens ofperuanus between 8 and 12 October: other large savanna raptors are probably im- We are actually about one half hour east of Chon- portant predators as well. goncito, encamped on an "estero" [a tidal estuary]. It Arthropod ectoparasites of Sigmodon al- is only salt for a short way above camp . . . It gets stoni collected by the Smithsonian Venezu- very hot [here] through the noon hours, and is of elan Project (Handley, 1976) were described course intensely dry. The country is practically flat by Furman (1972), Jones et al. (1972), John- "Ceiba country" -thick brush interspersed with great son (1972), Tipton and Machado-Allison Ceibas [Tate's notes for 6 October]. (1972), Brennan and Reed (1975), and Saun- It [the landscape] is intensely dry now, but shows ders (1975). evidence of being the very opposite during the rains. Sigmodon peruanus: There are no pub- It is not in the least comparable with [the] Santa Elena lished accounts of the environments occu- [peninsula]. That place was purely desert, not even producing cactus... .One thing I fail to understand is pied by this species in western Ecuador and that all this part is scrub forest averaging 15-20 feet northwestern Peru, and the following de- high. Now, the people state that they take excellent scriptions are therefore abstracted verbatim crops out year after year; therefore, this smallness of from the manuscript journals in -Aiich G. H. the trees cannot be due to poverty of soil. The ap- H. Tate and other collectors working for the parent reason is the long and intensely dry summer American Museum of Natural History re- season. . .[Tate's notes for 11 October]. corded their field observations. Tate and H. E. Wickenheiser collected 40 From 22 October to 5 November, Tate and specimens of Sigmodon peruanus at Puente Wickenheiser were based at Hacienda Piji- de Chimbo (Ecuadorean locality 3) between gual near Vinces (Ecuadorean locality 22), 3 August and 26 September 1922. Camped where they collected 13 specimens of peru- beside the Rio Chimbo at 1134 ft (348 m) anus. The camp at Pijigual was located in elevation, the collectors encountered a di- ...slightly rolling [country]... .broken here and there versity of natural and manmade habitats: by very small watercourses running NE and now dry... .vegetation scanty, [with] scattered Ceibas and, There are several distinct types of collecting country in the ravines, other trees. Nearing the river this coun- here.... First, near the house is second-growth land try is at first interspersed with, and eventually gives and portrero [pastures]. Next, the loma [a vegetation way almost entirely to, manglars [mangroves], now formation peculiar to the region], forested, moder- dry, but in the rains anything from 6-12 feet under ately open and dry, quite flat. Thirdly, the playas [river water [Tate's notes for 26 October]. beaches or floodplains] and riversides, and lastly the very dense, wet montaiia [forest] across the bridge The specimens ofperuanus collected near from the hacienda ... From a collector's point ofview Hacienda Pijigual were trapped under dense this place is excellent, since besides this variety of grass along the dry watercourses mentioned country close by, the fauna is rich and varied [Tate's in the passage quoted above. notes for 12 August]. On the next American Museum expedition The weather is generally true to the following pro- to Ecuador, Tate visited Hacienda San Ra- gram. At noon a light shower, clear p.m. and sunset; early in the evening the garuia [a dense nocturnal fog] mon on Isla Puna (Ecuadorean locality 13), sets in, often continuing until 8 o'clock the next morn- where 24 specimens of peruanus were col- ing. Morning fair till mid-day shower. In spite ofthis lected between 23 and 28 December 1923. garua, the whole second growth is arid ... The two Tate's notes for 22 December include the ob- commonest rats, Oryzomys [talamancae] and Melan- servation that the arid habitats on Isla Puna 42 AMERICAN MUSEUM NOVITATES NO. 3050

Fig. 22. Semiarid hillsides with green ceiba trees (one is in the middle foreground) in the valley of the Rio Casanga (Ecuadorean locality 15). Owl pellets containing skull fragments ofSigmodon peruanus were collected at this locality by H. E. Anthony and G. K. Cherrie. Photographed in September 1920 by H. E. Anthony.

appeared almost identical with those de- or cactus. A beautiful grove of green-barked Ceibas scribed above at Chongoncito. extending some four or five miles down the valley in The only photographs suitable for repro- scattered groups is the one redeeming floral fea- duction of landscapes inhabited by peruanus ture... .The only spots where rodents can exist is where (figs. 22, 23) were taken by H. E. Anthony in fields are fenced to keep the goats out. September 1920 when he camped in the val- A feature apparently common to all of the ley of the Rio Casanga (Ecuadorean locality habitats in which peruanus was collected by 15) with G. K. Cherrie. There they only col- AMNH expeditions in western Ecuador is the lected owl pellets containing fragments of dramatic seasonality of rainfall. The local seven cotton rat skulls, but Anthony's vivid rainy "winter" or invierno ends in April or rendition of the setting is consistent with May, and the dry "summer" or verano lasts Tate's prosaic but more analytical descrip- from June to October or even December tions ofperuanus habitats elsewhere: (Johnson, 1976). Between 100 and 800 m The Casanga valley is an open, extremely arid and elevation, condensation from nocturnal fog desolate ravine, everywhere fearfully eroded. .. Al- (the garua; see Tate's account for Puente de though a desert for most of the year, evidently the Chimbo, above; and Chapman, 1926: 706- rains when they do come are exceedingly heavy. Great 707) provides sufficient moisture during the stretches ofthe floor and slopes ofthis valley are either seasonal drought to support a perennially barren or else sparsely clothed with thomy mesquite green loma vegetation, but elsewhere in the 1 992 VOSS: SOUTH AMERICAN SIGMODON 43

Fig. 23. Green ceibas (Ceiba sp., Bombacaceae), characteristic and conspicuous trees with bright greenish bark in the semiarid habitats where Sigmodon peruanus is abundant in western Ecuador. Photographed in the valley ofthe Rio Casanga (Ecuadorean locality 15) by H. E. Anthony in September 1920.

Ecuadorean coastal lowlands south ofManta, peninsula, and in coastal Peru, these semiarid deciduous woodlands, thomscrub, and dry formations are replaced by real deserts. With- grasslands predominate. On the Santa Elena in this climatic region, peruanus appears re- 44 AMERICAN MUSEUM NOVITATES NO. 3050

Fig. 24. Paramo grasslands on the eastern slopes of Chimborazo above Urbina (Ecuadorean locality 4). The summit ofChimborazo is hidden by clouds and a siding ofthe railroad from Guayaquil to Quito runs across the middle foreground. H. E. Anthony and G. H. H. Tate camped near the railroad when they collected the type series of Sigmodon inopinatus at this locality. Photographed in May 1916 by F. M. Chapman. stricted to such open habitats as grass, low high elevations in the Ecuadorean Andes: near scrub, garden plots, and pastures. The known Urbina (Ecuadorean locality 4), where G. H. altitudinal range of the species extends from H. Tate and H. E. Anthony trapped twelve sea level to about 1600 m (fig. 19). specimens from 24 to 26 October 1923 at Tate summarized his observations con- about 3500 m on the eastern slopes ofChim- cerning the habits ofmammalian species col- borazo; and in Parque Nacional Las Cajas lected in Ecuador on typewritten cards that (Ecuadorean locality 1) where A. Barnett col- are still filed in the AMNH Department of lected four specimens at 3700-3800 m in Mammalogy archives. On one such card he 1981, 1983, and 1984. Both collection lo- noted that peruanus makes distinct runways calities lie within the paramo, an alpine life and ". . . is the only rat I have found that is zone characterized by cold, humid, grass- diurnal. It comes out from two pm onwards covered hillsides and wet valley bottoms with [and]... stays out till morning." W. H. Os- ponds, bogs, marshes, and dense, spongy good's fieldnotes (in the FMNH Division of hummocks of moss. Although paramos are Mammals archives) from 30 March to 2 April not without woody plants, trees and shrubs 1912 at Pacasmayo (Peruvian locality 1) also are minor elements in the vegetation and usu- imply diurnal habits and suggest that the spe- ally occur in discontinuous patches on shel- cies leaves ". . . green cuttings like those made tered hillsides and in well-drained valleys (see by Microtis," but no other natural history papers in Vuilleumier and Monasterio, 1986, information worthy of record is available. for numerous references to the extensive lit- Sigmodon inopinatus: This species has been erature on paramo ecology). collected at just two localities, both at very Anthony (1924: 4) reported only that spec- 1 992 VOSS: SOUTH AMERICAN SIGMODON 45

Fig. 25. Bunch-grass and a mossy hummock, characteristic features ofpdramo vegetation; the habitat of Sigmodon inopinatus on the eastern slopes of Chimborazo above Urbina (Ecuadorean locality 4). Photographed in May 1916 by F. M. Chapman. imens of inopinatus were collected near Ur- the atmosphere was too smoky for successful bina ". . . in the tufts of coarse grass which photography because the local people were covered the paramo," and his fieldnotes are burning the grass. Several years earlier, how- no more illuminating except to explain that ever, F. M. Chapman encountered more fa- 46 AMERICAN MUSEUM NOVITATES NO. 3050 vorable atmospheric conditions, and his pho- strictly terrestrial, and reproduces continu- tographs include a general view ofthe paramo ously throughout the year, even in highly sea- above Urbina (fig. 24), and a close-up of the sonal environments; Zygodontomys is also vegetation (fig. 25). comparable to Sigmodon in body size (table Labels attached to the four specimens that 15) but differs by its predominantly or exclu- A. Barnett collected in Las Cajas National sively nocturnal activity and apparently om- Park all refer to grassy situations or to grasses nivorous diet (see review by Voss, 1991). Ac- mixed with shrubby Compositae, in or near cording to O'Connell (1982), Zygodontomys marshy areas of the paramo. also differs from sympatric S. alstoni in the No other data concerning the habitats or Llanos by not constructing runways. habits of this species have been recorded. Holochilus brasiliensis (sensu Hershkovitz, 1955) is found throughout the unforested SUMMARY, CoMPARisoNs, AND DISCUSSION lowlands of Colombia, Venezuela, and the Guianas, where it often occurs sympatrically Available information concerning the nat- with Sigmodon hispidus or S. alstoni. Like ural history of South American Sigmodon cotton rats, H. brasiliensis is predominantly species suggests close resemblance with their herbivorous, subsisting chiefly on the green better-known North American congeners in leaves and stems of grasses (including such most important respects. As a reasonable cultivated species as sugar cane and rice), and generalization, species of Sigmodon can be reproduces throughout the year even where characterized as strictly terrestrial rats that precipitation is seasonal (Twigg, 1962, 1965; inhabit open country, cleared areas in for- Martino and Aguilera, 1989). However, H. ested regions, or wherever grass is an impor- brasiliensis is much larger than any species tant component ofthe local vegetation. Some, of Sigmodon (table 15), and is semiaquatic; and perhaps all, cotton rats are diurnal as well its webbed hindfeet presumably facilitate as nocturnal, construct runways, and subsist aquatic locomotion in such typical habitats chiefly on the green leaves and stems ofgrass- as seasonally inundated marshes, flooded rice es and forbs. In those tropical populations fields, grassy riversides, and wet canebrakes studied to date, reproductive activity is ap- (Hershkovitz, 1955; Twigg, 1962, 1965; Hus- parently continuous throughout the rainy and son, 1978). Annecdotal observations suggest dry seasons. Within the nonforest environ- that H. brasiliensis is nocturnal (Twigg, 1962). ments they occupy, species of Sigmodon are Little ecological information is available for probably one of the more important lower other muroids with which South American links in the food chains that sustain verte- Sigmodon species may often be sympatric: brate secondary consumers. Akodon urichi, Calomys hummelincki, and Ecological comparisons with other muroid Oligoryzomysfulvescens (with S. alstoni and/ taxa encountered in the same Neotropical or S. hispidus in Colombia and Venezuela); habitats reveal differences that suggest a dis- Oryzomys xantheolus (with S. peruanus in tinctive trophic role for cotton rats. Of the the western lowlands of Ecuador and Peru); 91 identifiable South American localities from and Akodon mollis and Phyllotis haggardi which Sigmodon alstoni and/or S. hispidus (with S. inopinatus in the Ecuadorean An- are known, species of Zygodontomys have des). Nevertheless, none of these species also been collected at 44, or about 48 percent closely resembles Sigmodon in features ofthe (see Voss, 1991, for a gazetteer of Zygodon- craniofacial skeleton and dentition that would tomys collection localities). Sigmodon and indicate similar diets. From this observation Zygodontomys are often taken in the same and the preceding comparisons, it may be traplines or trapping grids and in apparently hypothesized that species of Sigmodon are identical microhabitats, but where they occur the only strictly terrestrial and predomi- together Zygodontomys usually appears to be nantly herbivorous muroids in the tropical the commoner taxon, or at least the more ecosystems they inhabit; apparently, their easily trapped (Aldrich and Bole, 1937; Iba- closest ecological counterparts in the New niez and Moreno, 1982; O'Connell, 1981; Vi- World muroid fauna occupy higher latitudes. vas, 1984; Voss, 199 1). Like Sigmodon, Zyg- Species of Reithrodon, for example, inhabit odontomys is restricted to open habitats, is the temperate grasslands of southern South 1 992 VOSS: SOUTH AMERICAN SIGMODON 47

TABLE 15 Adult2 Weights (in grams) of Sigmodonb and Other South American Muroids Discussed in the Text Weights Species N Mean ± 1 SD Observed range Sigmodonalstonic 33 67.1 ± 15.4 37-93 Sigmodon hispidusc 43 107.5 ± 34.8 42-184 Zygodontomys brevicaudac 46 56.7 ± 19.4 25-105 Zygodontomys brunneusd 14 75.6 ± 10.6 56-96 Holochilus brasiliensise 26 155.4 ± 58.2 59-250 a Specimens with completely erupted molar rows are considered adults. b Weight data are unavailable for inopinatus and peruanus. c Specimens from various Venezuelan localities. d Specimens from Colombian locality 12. e Specimens from Departamento Beni, Bolivia (no adequate series are available from Colombia or Venezuela).

America where they subsist almost entirely evolving North American fauna of muroids on green vegetation (Pearson, 1988; Scaglia with simpler genitalia. In this scenario, living et al., 1982), while voles of the genus Micro- and fossil species of Sigmodon in North tus are the dominant small mammalian her- America are hypothesized to be descendants bivores in many north-temperate grasslands of a South American ancestor that dispersed (Rose and Birney, 1985). Comparisons among back across the Bolivar seaway sometime pri- these and other herbivorous, grassland- or to a direct land connection in the late Plio- dwelling rodents have provoked speculations cene or early Pleistocene: about convergence in behav- evolutionary The North American range of Sigmodon is greater ioral, physiological, and demographic traits than that of any other complex-penis-type cricetine. (Baker, 1971; Fleming, 1975; O'Connell, Its present and past distribution in North America 1982), but such lie outside the scope of this points to the cotton rat as one of the earliest over- report. water migrants from South America ... Successive invasions from South America, or reinvasions ofSouth America from Middle America, may account for the BIOGEOGRAPHY continuous distribution of living Sigmodon hispidus Species ofSigmodon are biogeographically in Middle and northern South America [Hershkovitz, interesting because of their close association 1966a: 736]. with open vegetation and because the genus Reig's (1986: 429-430) more elaborate re- ranges considerably north and south of the construction is also based on the assumption Rio Atrato lowlands in northwestern Colom- that Sigmodon evolved in South America: bia, site of the last marine barrier to intercontinental dispersal ofAmerican land mam- Even though the Sigmodontini [an undiagnosed tribe comprising only Sigmodon and Holochilus] are now mals in the Late Tertiary (Haffer, 1970). in South America a predominantly non-Andean group Whether the genus evolved in North America . . . there are strong indications that they first evolved and invaded South America, or vice versa, in the North-Central Andes from a phyllotine ances- and when, are richly speculative topics in the tor. Neotomys is intermediate between phyllotines and literature. sigmodont rodents ... and its presence in the living fauna of the northern puna is suggestive of an origin of the genus Sigmodon in the pre-Pliocene Andes of INTERCONTINENTAL DISPERSAL northern Peru. The occurrence of S. peruanus ... in According to Hershkovitz (1 966a), muroid the lowlands and mountain slopes west of the Andes rodents with a complex glans penis invaded of Ecuador and north-central Peru, and of S. inopinatus in the high Andes of Ecuador, is in good agree- South America from Central America by ov- ment with this hypothesis. Thus, Sigmodon is likely erwater dispersal, perhaps as early as the to have originated from a Neotomys-like ancestor in Miocene, and thereafter underwent an exten- north-central Peru, and to have further dispersed sive adaptive radiation in isolation from the westward to the western coastal lowland region of... 48 AMERICAN MUSEUM NOVITATES NO. 3050

north-central Peru and Ecuador, and northward and Lindsay's (1981: 428) conclusions about through the valleys ofthe Andes ofEcuador up to the the historical biogeography of cotton rats: proto-Magdalena Valley ofColombia, and eventually to have spread from the latter to the lowlands of Co- Prosigmodon is closely related to Calomys (Benson- lombia and Venezuela... Sigmodon inopinatus, found omys) on the one hand, and Sigmodon on the other. isolated on the paramo of Chimborazo ... is likely It is more advanced than Calomys (Bensonomys) while to have arrived there by the uplifting of the paramos being more primitive than Sigmodon in height of during Plio-Pleistocene times. As a consequence of crown, strength of connections between cusps, ro- its northward spread, Sigmodon most probably en- bustness, and development ofaccessory rootlets. The tered Central America by overland dispersal once the age of Prosigmodon is late Hemphillian and early Panamanian land bridge was established, some 5.5 Blancan. It is morphologically and temporally suited million years ago .. . and differentiated in the arid to represent a descendant of Calomys (Bensonomys). and mesic biomes ofNorth America into several spe- The more lingual connection of the anterocone and cies ... The very primitive, 2n = 82 karyotype ... protocone in Prosigmodon relative to Calomys and of S. alstoni suggests that its carrier is a remnant of Sigmodon precludes Prosigmodon (at least P. oroscol) the early Sigmodon differentiation, but ... knowledge from direct ancestry of Sigmodon. However, overall ofthe chromosomes ofS. peruanus and S. inopinatus similarity ofProsigmodon and Sigmodon suggests that would be critical for this evaluation, and their kary- Prosigmodon is near the ancestral stock of Sigmodon otypes are unknown so far ... several living North and supports the derivation ofSigmodon from North American species of Sigmodon have derived, low- American cricetids,6 rather than from Old World cri- number karyotypes... cetodontines ... or from South American cricetids ... To observe that these historical-biogeo- Unfortunately, virtually all North Ameri- graphic hypotheses are underdetermined by can fossil muroid taxa are known only from data is perhaps unnecessary. As Patterson and isolated molars, some complete toothrows, Pascual (1972) and many subsequent stu- and mandibular fragments. This material is dents of mammalian paleontology have ob- better than nothing, but the few comparisons served, the earliest known South American possible with such anatomical scraps do not muroid fossils are from Montehermosan encourage confidence in phylogenetic conclu- (Pliocene) deposits, dated at ca. 3.5 m.y. sions that have important biogeographic im- (Marshall et al., 1979). Thus, there is no di- plications. Students of Recent muroid sys- rect evidence that muroids had an extensive tematics routinely survey many more pre-Pliocene history in South America. A characters, and scepticism about the signifi- second weakness in these scenarios is the lack cance ofthe very limited morphological data of compelling cladistic evidence that Sig- presently available for most fossil taxa is not modon is closely related to any South Amer- inappropriate. ican muroids, including other "sigmodonts" Analyses of phylogenetic relationships (see Phylogenetic Relationships, above). Fi- among extant taxa could provide important nally, Reig's assumptions concerning primi- data for evaluating the alternative biogeo- tive and derived diploid numbers in Sig- graphic scenarios described above. For ex- modon have no basis in phylogenetically ample, if cladistic analyses yield compelling defensible analyses of chromosomal evolu- evidence that Sigmodon is not closely related tion (and see Baker et al., 1983, for alterna- to the endemic South American "sigmodont" tive views concerning plesiomorphic Neo- genera Holochilus, Neotomys, and Reithro- tropical muroid karyotypes). don, then much ofthe argument for southern Paleontologists, by contrast, have usually ancestry evaporates. Morphological, cytolog- argued that Sigmodon evolved in North ical, or molecular studies to critically test America. Fossil evidence cited to support this "sigmodont" monophyly should minimally view is the presence of Sigmodon and Pro- include representatives of all of these genera sigmodon in late Miocene (Hemphillian) or together with early Pliocene (Blancan) sediments from the representative oryzomyines (a southwestern United States and northern 6 Baskin's (1978) hypothesis that Calomys (Bensono- Mexico (Martin, 1979; Jacobs and Lindsay, mys) is descended from Copemys, a North American 1981; Czaplewski, 1987). The phylogenetic genus also believed to be ancestral to Peromyscus, is an position of Prosigmodon is crucial for Jacob implicit assumption of this argument. 1 992 VOSS: SOUTH AMERICAN SIGMODON 49 group to which Holochilus may belong), phyl- tained by studies like those outlined above lotines (the group in which Neotomys and will be relevant for reconstructions of vege- Reithrodon are often placed), North Ameri- tational history as a causal factor in Neo- can genera with simple genitalia, and Old tropical faunal evolution. Webb (1978, 199 1) World taxa (as putative outgroups; Calo- and Marshall (1979) have postulated that sa- myscus and Mystromys would be good vanna corridors across or around the Ama- choices). Unfortunately, current morpholog- zon basin connected the Llanos and other ical evidence of relationships (especially the northern grasslands with grasslands on the absence ofsynapomorphies for the subfamily Brazilian Shield and in Patagonia during gla- Sigmodontinae sensu either Carleton and cial episodes in the Late Cenozoic. As noted Musser [1984] or Reig [1980]) permits no less by Voss (1991), however, savanna muroid extensive sampling design. faunas north and south ofAmazonia provide Different relationships among the species little evidence of former habitat continuity. ofSigmodon are predicted by alternative bio- If Sigmodon is not closely related to Pata- geographic hypotheses. Ifthe genus had a long gonian or central-Andean "sigmodonts" evolutionary history in North America prior (Reithrodon, and Neotomys), then its current to invading South America across the newly distribution only in the northern nonforest emerged landbridge, and if South America regions of South America suggests that the was invaded only once, then the earliest cla- Late Cenozoic dispersal of small, grassland- dogenetic events in cladistic reconstructions inhabiting mammals was not unimpeded, ei- should be among Recent North American ther via Webb's "high road" through the An- species, and the South American species des or his "low road" through eastern Ama- should constitute a monophyletic group. Al- zonia. If, however, Reithrodon and/or ternatively, if the genus evolved and spe- Neotomys are sister taxa of Sigmodon, then ciated in South America prior to invading the presently disjunct distribution of these North America, and if the invasion was pastoral sigmodonts would lend additional unique, then the earliest branches in cladistic support (to the herpetological data cited by reconstructions should be South American Webb, 1978) for ancient corridors of grass- lineages, and the North American species land habitats. should constitute a monophyletic group. Of Two species ofSigmodon, hispidus and al- course, these are the simplest alternatives and stoni, are members of a distinctive lowland other phylogenetic results could suggest more fauna of nonforest mammals in northern complex biogeographic scenarios. Morpho- South America that also includes Lutreolina logical, cytological, or molecular data sets to crassicaudata, Dasypus sabanicola, Calomys evaluate relationships in this context should hummelincki, two species of Zygodontomys, include all of the currently recognized valid Cavia aperea, and Sylvilagusfloridanus. This species of Sigmodon: alleni, alstoni, arizon- fauna is most diverse in the Llanos but also ae, fulviventer, hispidus, inopinatus, leucotis, occurs in adjacent but disjunct nonforest low- mascotensis, ochrognathus, andperuanus. As land regions of Colombia, Venezuela, the emphasized earlier in this report, the genetic Guianas, and extreme northern Brazil. The integrity of hispidus cannot be assumed, and occurrence of these small, nonvolant mam- the prudent systematist will include several mals in grasslands now separated by exten- samples each from North, Central, and South sive tracts ofrain forest has been interpreted America. Outgroups should ideally be chosen as evidence for vicariance of pre-Recent sa- on the basis ofa prior analysis ofintergeneric vanna ecosystems that were probably most relationships, such as that described in the extensive during the last glacial maximum preceding paragraph. (Voss, 1991). Neither Zygodontomys nor any other NONFOREST MAMMAL member of this northern savanna fauna, FAUNAS AND SOUTH however, is known to occur sympatrically AMERICAN PALEOHABITATS with Sigmodon peruanus in the nonforest Because cotton rats are restricted to open lowland habitats of western Ecuador and habitats, the phylogenetic information ob- northwestern Peru. This observation suggests 50 AMERICAN MUSEUM NOVITATES NO. 3050 the historical absence of any grassland cor- 1901. Descriptions oftwo new species ofSouth ridor around the northern Andes and down American Muridae. Bull. Am. Mus. Nat. the west coast of South America. Presum- Hist. 14: 39-40. ably, the Choco rain forests of westem Co- 1903. Description of a new species of Sig- lombia and northwestern Ecuador modon from Ecuador. Bull. Am. Mus. effectively Nat. Hist. 19: 99-100. isolated the lowland nonforest faunas of 1913. New mammals from Colombia and Ec- northern South America from those of the uador. Bull. Am. Mus. Nat. Hist. 32: Pacific littoral during arid climatic phases of 469-484. the Quaternary. Montane grasslands, how- Anthony, H. E. ever, perhaps resembling those now inhab- 1924. Preliminary report on Ecuadorean ited by S. inopinatus in the Ecuadorean An- mammals No. 4. Am. Mus. Novitates des, might have provided ephemeral avenues 114: 6 pp. for dispersal of cold-resistant taxa between Bailey, V. the Pacific lowlands and the unforested inter- 1902. Synopsis ofthe North American species Andean valleys of Colombia. ofSigmodon. Proc. Biol. Soc. Washing- Because these patterns of nonforest mam- ton 15: 101-116. malian distributions broadly resemble those 1905. Biological survey of Texas. North previously American Fauna 25: 221 pp. described for nonforest birds by Baker, R. H. Haffer (1967), I conclude this essay with a 1969. Cotton rats of the Sigmodonfulviventer prediction that assumes a common cause for group. Misc. Publ. Univ. Kansas Mus. such zoogeographic congruence. The arid in- Nat. Hist. 51: 177-232. termontane valley ofthe upper Rio Marai-non 1971. Nutritional strategies of myomorph ro- in the Amazonian catchment ofnorthern Peru dents in North American grasslands. J. (fig. 2) is inhabited by a nonforest avifauna Mammal. 52: 800-805. essentially the same as that ofthe dry Pacific Baker, R. J., B. F. Koop, and M. W. Haiduk lowlands (Chapman, 1926; Dorst, 1957; Haf- 1983. Resolving systematic relationships with fer, 1967). If this fact is explained by vicar- G-bands: A study offive genera ofSouth iance rather than dispersal, then populations American cricetine rodents. Syst. Zool. of 32: 403-416. nonforest lowland mammals closely relat- Bangs, 0. ed to Pacific littoral taxa should be expected 1898. A new Sigmodon from the Santa Marta along the upper Marafion (some ofthese may region of Colombia. Proc. Biol. Soc. be members ofPearson's [1982] "grupo peri- Washington 12: 189-190. andino"). Future collectors should be alert to Barbour, T. the possible presence of cotton rats in the 1946. A Naturalist's Scrapbook. Cambridge: Mara-non Valley, either Sigmodon peruanus Harvard Univ. Press. or a new sister species. Baskin, J. A. 1978. Bensonomys, Calomys, and the origin REFERENCES of the phyllotine group of Neotropical cricetines (Rodentia: Cricetidae). J. Aldrich, J. W., and B. P. Bole, Jr. Mammal. 59: 125-135. 1937. The birds and mammals ofthe western Bisbal, F. J. slope ofthe Azuero peninsula (Republic 1986. Food habits of some Neotropical car- of Panama). Sci. Publ. Cleveland Mus. nivores in Venezuela. Mammalia 50: Nat. Hist. 7:196 pp. 329-339. Allen, G. M. Bisbal, F. J., and J. Ojasti 1931. Type specimens ofmammals in the Mu- 1980. Nicho trofico del zorro Cerdocyon thous seum of Comparative Zoology. Bull. (Mammalia, Carnivora). Acta Biol. Ve- Mus. Comp. Zool. 71: 229-289. nezuelica 10: 469-496. Allen, J. A. Brennan, J. M., and J. T. Reed 1 897a. On a small collection ofmammals from 1975. A list of Venezuelan chiggers, particu- Peru, with descriptions of new species. larly of small mammalian hosts (Aca- Bull. Am. Mus. Nat. Hist. 9: 115-122. rina: Trombiculidae). Brigham Young 1897b. Description of a new species of Sig- Univ. Sci. Bull., Biol. Ser. 20(1): 45-75. modon from Bogota, Colombia. Bull. Bugge, J. Am. Mus. Nat. Hist. 9: 121-122. 1970. The contribution ofthe stapedial artery 1992 VOSS: SOUTH AMERICAN SIGMODON 51

to the cephalic arterial supply in muroid Czaplewski, N. J. rodents. Acta Anat. 76: 313-336. 1987. Sigmodont rodents (Mammalia; Mu- Burmeister, H. roidea; Sigmodontinae) from the Plio- 1854. [Ueber eine neue Ratte, Lasiomys hir- cene (early Blancan) Verde Formation, suta, von Maracaibo.] Abhand. Natur- Arizona. J. Vertebr. Paleontol. 7: 183- forsch. Ges. Halle 2(1), Sitzungsber- 199. ichte: 15-17. Dorst, J. Cabrera, A. 1957. Contribution a 1'etude ecologique des 1961. Catalogo de los mamiferos de America oiseaux du haut Marafion (Perou Sep- del Sur. Rev. Mus. Argentina Cienc. Nat. tentrional). L'Oiseaux et la Revue Fran- "Bernardino Rivadavia" 4(2): 309-732. 9aise d'Ornithologie 27: 235-269. Cameron, G. N., W. B. Kincaid, and B. A. Carnes Ellerman, J. R. 1979. Experimental species removal: tempo- 1941. The families and genera of living ro- ral activity patterns of Sigmodon his- dents, vol. 2, Family Muridae. London: pidus and Reithrodontomys fulvescens. British Museum (Natural History). J. Mammal. 60: 195-197. Eshelman, R. E., and G. S. Morgan Carleton, M. D. 1985. Tobagan Recent mammals, fossil ver- 1973. A survey ofgross stomach morphology tebrates, and their zoogeographic im- in New World Cricetinae (Rodentia, plications. Natl. Geogr. Res. Rep. 21: Muroidea), with comments on func- 137-143. tional interpretations. Misc. Publ. Mus. Fleharty, E. D., and L. E. Olson Zool. Univ. Michigan 146: 43 pp. 1969. Summer food habits of Microtus och- 1980. Phylogenetic relationships in neotom- rogaster and Sigmodon hispidus. J. ine-peromyscine rodents (Muroidea) Mammal. 50: 475-486. and a reappraisal of the dichotomy Fleming, T. H. within New World Cricetinae. Misc. 1975. The role of small mammals in tropical Publ. Mus. Zool. Univ. Michigan 157: ecosystems. In F. B. Golley et al. (eds.), 146 pp. Small mammals: their productivity and Carleton, M. D., and G. G. Musser population dynamics, pp. 269-298. 1984. Muroid rodents. In S. Anderson and J. London: Cambridge Univ. Press. K. Jones (eds.), Orders and families of Furman, D. P. Recent mammals ofthe world, pp. 289- 1972. Laelapid mites (Laelapidae: Laelapinae) 379. New York: Wiley and Sons. ofVenezuela. Brigham Young Univ. Sci. 1989. Systematic studies of oryzomyine ro- Bull., Biol. Ser. 17(3): 58 pp. dents (Muridae, Sigmodontinae): A Gardner, A. L., and J. L. Patton synopsis of Microryzomys. Bull. Am. 1976. Karyotypic variation in oryzomyine ro- Mus. Nat. Hist. 191: 83 pp. dents (Cricetinae) with comments on Chapman, F. M. chromosomal variation in the Neotrop- 1926. The distribution ofbird-life in Ecuador. ical cricetine complex. Occas. Pap. Mus. Bull. Am. Mus. Nat. Hist. 55: i-xiv, 1- Zool. Louisiana State Univ. 49: 48 pp. 784, + folding map. Giebel, C. G. Chipman, R. K. 1857. Beitrage zur Osteologie der Nagetiere. 1965. Age determination of the cotton rat Abhandl. Naturw. Vereines fur Sachsen (Sigmodon hispidus). Tulane Stud. Zool. und Thuringen in Halle 1: 191-261 + 12: 19-38. 5 pls. Corbet, G. B., and J. E. Hill Goertz, J. W. 1980. A world list of mammalian species. 1964. The influence of habitat quality upon London: British Museum (Natural His- density of cotton rat populations. Ecol. tory). Monogr. 34: 359- 381. Coues, E. Goertz, J. W., and R. C. Long 1877. Monographs of North American Ro- 1973. Habitats of five species of rat in Loui- dentia, No. 1. Muridae. Rep. U.S. Geol. siana. Am. Midl. Nat. 90: 460-465. Survey Terr. 11: i-x, 1-264, + 5 pls. Goldman, E. A. Cuervo-Diaz, A., J. Hernaindez-Camacho, and A. 1920. Mammals of Panama. Smithson. Misc. Cadena G. Coll. 69(5): 309 pp. 1986. Lista actualizada de los mamiferos de Gyldenstolpe, N. Colombia anotaciones sobre su distri- 1932. A manual ofNeotropical sigmodont ro- buci6n. Caldasia 15: 472-501. dents. Kungl. Svenska Vetenskapsakad. 52 AMERICAN MUSEUM NOVITATES NO. 3050

Handl., Ser. 3, Band 11(3): 164 pp + tions: the search for a cause, pp. 399- 18 pl. 406. New Haven: Yale Univ. Press. Haffer, J. Hooper, E. T. 1967. Zoogeographical notes on the "nonfo- 1962. The glans penis in Sigmodon, Sigmo- rest" lowland bird faunas ofnorthwest- mys, and Reithrodon (Rodentia, Crice- ern South America. El Hornero 10: 315- tidae). Occas. Pap. Mus. Zool. Univ. 333. Michigan 625: 11 pp. 1970. Geologic-climatic history and zoogeo- Hooper, E. T., and G. G. Musser graphic significance ofthe Urub'a region 1964. The glans penis in Neotropical cicetines in northwestern Colombia. Caldasia 10: (Muridae) with comments on classifi- 603-636. cation of muroid rodents. Misc. Publ. Hall, E. R. Mus. Zool. Univ. Michigan 123: 57 pp. 1981. Mammals of North America (second Husson, A. M. edition), vol. 2. New York: Wiley. 1978. The mammals of Suriname. Leiden: E. Handley, C. 0. H., Jr. J. Brill. 1972. Mammalogy in Panama. Bull. Biol. Soc. Ibafiez, C. Washington 2: 217-228. 1980. Ritmo de actividad de algunos ratones 1976. Mammals of the Smithsonian Venezu- de los Llanos de Apure. Doiiana Acta elan Project. Brigham Young Univ. Sci. Vert. 7: 117-120. Bull., Biol. Ser. 20(5): 89 pp. + map. Harper, F. Ib'aniez, C., and S. Moreno 1982. Ciclo reproductor de algunos cricetidos 1927. The mammals ofthe Okefinokee swamp region ofGeorgia. Proc. Boston Soc. Nat. (Rodentia, Mammalia) de los Llanos de Hist. 38: 191-396, + 4 pls. Apure (Venezuela). Actas VIII Congr. Hershkovitz, P. Latinoamer. Zool. Tomo 1: 471-480. 1955. South American marsh rats, genus Hol- Jacobs, L. L., and E. H. Lindsay ochilus, with a summary of sigmodont 1981. Prosigmodon oroscoi, a new sigmodont rodents. Fieldiana: Zoology 37: 639-673 rodent from the late Tertiary ofMexico. + pls. 17-29. J. Paleontol. 55: 425-430. 1960. Mammals of northern Colombia, pre- Jimenez, J. liminary report No. 8: Arboreal rice rats, 1971. Comparative post-natal growth in five a systematic revision of the subgenus species of the genus Sigmodon. I. Ex- Oecomys, genus Oryzomys. Proc. U.S. ternal morphological character relation- Natl. Mus. 110: 513-568. ships. Rev. Biol. Trop. 19: 133-148. 1962. Evolution of Neotropical cricetine ro- Johnson, A. M. dents (Muridae) with special reference 1976. The climate of Peru, Bolivia, and Ec- to the phyllotine group. Fieldiana Zool. uador. In W. Schwerdtfeger (ed.), Cli- 46: 524 pp. mates of Central and South America 1966a. Mice, land bridges, and Latin American (World Survey ofClimatology, vol. 12), faunal interchange. In R. L. Wenzel and pp. 147-218. Amsterdam: Elsevier. V. J. Tipton (eds.), Ectoparasites ofPan- Johnson, P. T. ama, pp. 725-751. Chicago: Field Mus. 1972. Sucking lice ofVenezuelan rodents, with Nat. Hist. remarks on related species (Anoplura). 1966b. South American swamp and fossorial Brigham Young Univ. Sci. Bull., Biol. rats of the scapteromyine group (Cri- Ser. 17(5): 62 pp. cetinae, Muridae) with comments on the Jolicoeur, P. glans penis in muroid taxonomy. Z. 1963. The multivariate generalization of the Saugetierkd. 31: 81-149. allometry equation. Biometrics 19: 497- Hoffmeister, D. F. 499. 1986. Mammals of Arizona. Tucson: Univ. Jones, E. K., C. M. Clifford, J. E. Keirans, and G. Arizona Press. M. Kohls Honacki, J. H., K. E. Kinman, and J. W. Koeppl 1972. The ticks of Venezuela (Acarina: Ixo- 1982. Mammal species of the world. Law- doidea) with a key to the species ofAm- rence: Allen Press. blyomma in the western hemisphere. Hooijer, D. A. Brigham Young Univ. Sci. Bull., Biol. 1967. Pleistocene vertebrates of the Nether- Ser. 17(4): 40 pp. lands Antilles. In P. S. Martin and H. Kaufman, D. W., and E. D. Fleharty E. Wright (eds.), Pleistocene extinc- 1974. Habitat selection by nine species of ro- 1992 VOSS: SOUTH AMERICAN SIGMODON 53

dents in north-central Kansas. South- 1912. Mammals from western Venezuela and western Nat. 18: 443-452. eastern Colombia. Field Mus. Nat. Hist. Kiblinsky, P. Zool. Ser. 10: 33-66, + 2 pls. 1969. The chromosomes of the hispid cotton Patterson, B., and R. Pascual rat, Sigmodon hispidus, from two lo- 1972. The fossil mammal fauna of South calities in Venezuela. J. Mammal. 50: America. In A. Keast, F. C. Erk, and B. 810-811. Glass (eds.), Evolution, mammals, and Kincaid, W. B., and G. N. Cameron southern continents, pp. 247-309. Al- 1982. Dietary variation in three sympatric ro- bany: SUNY Press. dents on the Texas coastal prairie. J. Patton, J. L. Mammal. 63: 668-672. 1967. Chromosome studies of certain pocket Lowery, G. H., Jr. mice, genus Perognathus. J. Mammal. 1974. The mammals of Louisiana and its ad- 48: 27-37. jacent waters. Baton Rouge: Louisiana Paynter, R. A., Jr. State Univ. Press. 1982. Ornithological gazetteer of Venezuela. Marshall, L. G. Cambridge, MA: Museum of Compar- 1979. A model for paleobiogeography ofSouth ative Zoology. American cricetine rodents. Paleobiol- ogy 5: 126-132. Paynter, R. A., Jr., and M. A. Traylor, Jr. Marshall, L. G., R. F. Butler, R. E. Drake, G. H. 1977. Ornithological gazetteer of Ecuador. Curtis, and R. H. Tedford Cambridge, MA: Museum of Compar- 1979. Calibration of the great American in- ative Zoology. terchange. Science 204: 272-279. 1981. Ornithological gazetteer of Colombia. Martin, R. A. Cambridge, MA: Museum of Compar- 1979. Fossil history of the rodent genus Sig- ative Zoology. modon. Evol. Monogr. 2: 36 pp. 1991. Ornithological gazetteer ofBrazil, 2 vols. Martino, A. M. G., and M. Aguilera M. Cambridge, MA: Museum of Compar- 1989. Food habits ofHolochilus venezuelae in ative Zoology. rice fields. Mammalia 53: 545-561. Pearson, 0. P. Meyer, B. J., and R. K. Meyer 1982. Distribucion de pequefios mamiferos en 1944. Growth and reproduction of the cotton el altiplano y los desiertos del Peru. Ac- rat, Sigmodon hispidus hispidus, under tas VIII Congr. Latinoamer. Zool. Tomo laboratory conditions. J. Mammal. 25: 1: 263-284. 107-129. 1988. Biology and feeding dynamics ofa South Mohr, E. American herbivorous rodent, Reith- 1941. Saugetiertypen im Zoologischen Muse- rodon. Stud. Neotrop. Fauna Envir. 23: um Halle. Z. Naturw. (Halle) 94: 215- 25-39. 226. Pearson, 0. P., and J. L. Patton O'Connell, M. A. 1976. Relationships among South American 1981. Population ecology of small mammals phyllotine rodents based on chromo- from northern Venezuela. Ph.D. disser- some analysis. J. Mammal. 57: 339-350. tation, Texas Tech University. Quay, W. B. 1982. Population biology ofNorth and South 1954. The anatomy of the diastemal palate in American grassland rodents: a compar- microtine rodents. Misc. Publ. Mus. ative review. In M. A. Mares and H. H. Zool. Univ. Michigan 86: 41 pp. + 4 Genoways (eds.), Mammalian biology pls. in South America. Pymatuning Symp. Reig, 0. A. Ecol. 6: 167-185. 1980. A new fossil genus of South American Odum, E. P. cricetid rodents allied to Wiedomys, with 1955. An eleven year history of a Sigmodon an assessment of the Sigmodontinae. J. population. J. Mammal. 36: 368-378. Zool. London 192: 257-281. Olds, N., and S. Anderson 1984. Distribuico geografica e historia evo- 1989. A diagnosis of the tribe Phyllotini (Ro- lutiva dos roedores muroideos sula- dentia, Muridae). In K. H. Redford and mericanos (Cricetidae: Sigmodontinae). J. F. Eisenberg (eds.), Advances in Neo- Rev. Brasileira Genet. 7: 333-365. tropical mammalogy, pp. 55-73. 1986. Diversity patterns and differentiation of Gainesville: Sandhill Crane Press. high Andean rodents. In F. Vuilleumier Osgood, W. H. and M. Monasterio (eds.), High altitude 54 AMERICAN MUSEUM NOVITATES NO. 3050

tropical biogeography, pp. 404-439. 1932. The taxonomic history ofcertain South New York: Oxford Univ. Press. and Central American cricetid Roden- 1987. An assessment of the systematics and tia: Neotomys, with comments on its re- evolution of the Akodontini, with the lationships; the cotton rats (Sigmodon, description ofnew fossil species ofAko- and Sigmomys); and the "fish-eating" don (Cricetidae: Sigmodontinae). Field- rats (Ichthyomys, Anotomys, Rheomys, iana: Zool. (n. ser.) 39: 347-399. Neusticomys, and Daptomys). Am. Mus. Reig, 0. A., M. Aguilera, and A. Perez-Zapata Novitates 583: 10 pp. 1990. Cytogenetics and karyosystematics of 1939. Mammals of the Guiana region. Bull. South American oryzomyine rodents Am. Mus. Nat. Hist. 76: 151-229. (Cricetidae: Sigmodontinae) II. High Thomas, 0. numbered karyotypes and chromosom- 1880. Description of a new species of Reith- al heterogeneity in Venezuelan Zygo- rodon, with remarks on the other spe- dontomys. Z. Siiugetierkd. 55: 361-370. cies of the genus. Proc. Zool. Soc. Lon- Rose, R. K., and E. C. Birney don 1880: 691-696. 1985. Community ecology. In R. H. Tamarin 1894. Descriptions of some new Neotropical (ed.), Biology of New World Microtis, Muridae. Ann. Mag. Nat. Hist. 6(14): pp. 310-339. Spec. Publ. Am. Soc. 346-366. Mammal. No. 8. 1901. On a collection of mammals from the Sanborn, C. C. Kanuku Mountains, British Guiana. 1947. The South American rodents of the ge- Ann. Mag. Nat. Hist. 7(8): 139-154. nus Neotomys. Fieldiana Zool. 31: 51- 1914. Four new small mammals from Vene- 57. zuela. Ann. Mag. Nat. Hist. 8(14): 410- Saunders, R. C. 414. 1975. Venezuelan Macronyssidae (Acarina: 1921. New Sigmodon, Oryzomys, and Echi- Mesostigmata). Brigham Young Univ. mys from Ecuador. Ann. Mag. Nat. Hist. Sci. Bull., Biol. Ser. 20(2): 73-90. 9(7): 448-450. Saussure, M. H. de Tipton, V. J., and C. E. Machado-Allison 1860. Note sur quelques mammiferes du Me- 1972. Fleas of Venezuela. Brigham Young xique. Rev. et Mag. de Zoologie Ser. 2, Univ. Sci. Bull., Biol. Ser. 17(6): 115 12: 97-110. PP. Say, T., and G. Ord Trouessart, E.-L. 1825. Description of a new species of Mam- 1880. Catalogue des mammiferes vivants et malia, whereon a new genus is proposed fossiles. Bull. Soc. Etudes Scient. Angers to be founded. J. Acad. Nat. Sci. Phil- 1880: 58-212. adelphia 4: 352-355. Twigg, G. I. Scaglia, 0. A., C. A. Velasquez, and M. A. Cau- 1962. Notes on Holochilus sciureus in British hepe Guiana. J. Mammal. 43: 369-374. 1982. Plant composition ofconey rat's (Reith- 1965. Studies on Holochilus sciureus berbicen- rodon auritus) diet. Acta Theriol. 27: sis, a cricetine rodent from the coastal 350-353. region ofBritish Guiana. Proc. Zool. Soc. Stephens, L., and M. A. Traylor, Jr. London 145: 263-283. 1983. Ornithological gazetteer of Peru. Cam- Vivas, A. M. bridge, MA: Museum of Comparative 1984. Ecologia de poblaciones de los roedores Zoology. de los Llanos de Guarico. Trabajo de 1985. Ornithological gazetteer ofthe Guianas. Ascenso, Universidad Simon Bolivar. Cambridge, MA: Museum of Compar- 1986. Population biology ofSigmodon alstoni ative Zoology. (Rodentia, Cricetidae) in the Venezue- Sunquist, M. E., F. Sunquist, and D. E. Daneke lan Llanos. Rev. Chilena Hist. Nat. 59: 1989. Ecological separation in a Venezuelan 179-191. Llanos carnivore community. In K. H. Vivas, A. M., and A. C. Calero Redford and J. F. Eisenberg (eds.), Ad- 1985. Algunos aspectos de la ecologia pobla- vances in Neotropical mammalogy, pp. cional de los pequefios mamiferos en la 197-232. Gainesville: Sandhill Crane Estacion Biologica de los Llanos. Bol. Press. Soc. Venezolana Cienc. Nat. 143: 79- Tate, G. H. H. 99. 1930. Notes on the Mount Roraima region. 1988. Patterns in the diet of Sigmodon his- Geogr. Rev. 20: 53-68 + folding map. pidus (Rodentia: Cricetidae) in relation 1992 VOSS: SOUTH AMERICAN SIGMODON 55

to available resources in a tropical sa- Voss, R. S., L. F. Marcus, and P. Escalante P. vanna. Ecotropicos 1: 82-91. 1990. Morphological evolution in muroid ro- Vivas, A. M., R. Roca, E. Weir, K. Gil, and P. dents I. Conservative patterns of cra- Gutierrez niometric covariance and their onto- 1986. Ritmo de actividad nocturna de Zygo- genetic basis in the Neotropical genus dontomys microtinus, Sigmodon al- Zygodontomys. Evolution 44: 1568- stoni, y Marmosa robinsoni en Masa- 1587. guaral, Estado Guarico. Acta Cient. Vuilleumier, F., and M. Monasterio (eds.) Venezolana 37: 456-458. 1986. High altitude tropical biogeography. Voss, R. S. New York: Oxford Univ. Press. 1988. Systematics and ecology of ichthy- Webb, S. D. omyine rodents (Muroidea): patterns of 1978. A history of savanna vertebrates in the morphological evolution in a small New World. Part II: South America and adaptive radiation. Bull. Am. Mus. Nat. the great interchange. Ann. Rev. Ecol. Hist. 188: 259-493. Syst. 9: 393-426. 1991. An introduction to the Neotropical mu- 1991. Ecogeography and the great American roid rodent genus Zygodontomys. Bull. interchange. Paleobiology 17: 266-280. Am. Mus. Nat. Hist. 210: 113 pp. Williams, S. L., H. H. Genoways, and J. A. Groen Voss, R. S., and A. V. Linzey 1983. Results of the Alcoa Foundation Suri- 1981. Comparative gross morphology ofmale name Expeditions 7. Records of mam- accessory glands among Neotropical mals from central and southern Suri- Muridae (Mammalia: Rodentia) with name. Ann. Carnegie Mus. 52: 329-336. comments on systematic implications. Winge, H. Misc. Publ. Mus. Zool. Univ. Michigan 1941. The interrelationships of the mamma- 159: 41 pp. lian genera, vol. 2. Copenhagen: C. A. Voss, R. S., and L. F. Marcus Reitzels Forlag. 1992. Morphological evolution in muroid ro- Zimmerman, E. G. dents II. Craniometric factor divergence 1970. Karyology, systematics, and chromo- in seven Neotropical genera, with ex- somal evolution in the rodent genus perimental results from Zygodontomys. Sigmodon. Publ. Mus. Michigan State Evolution (in press). Univ., Biol. Ser. 4: 385-454.

APPENDIX List of Measured Specimens TABLE 7 8 mi S Lake Placid: AMNH 255329, 255359, 442517-442520, 442523, 442525-442528, 255377, 255378, 255381-255389, 255391, 442530, 442531, 442535, 442536, 442544- 255394, 255396-255398, 255401, 255402, 442546, 442549, 442550, 442552. Rancho Gran- 255404-255406, 255408-255410, 255412- de: USNM 517632, 517639, 517640, 517642, 255416, 255418-255422, 255424. 517643, 517646, 517648, 517651-517653, TABLE 10 517655, 517657-517659. Valle de Suaza: USNM 541969, 542093-542104, TABLE 12 542106-542109, 542111-542116, 542119- Rancho Grande: USNM 517660,517661,517663, 542122, 542124, 542125, 542127-542134. Min- 517666, 517667, 517669, 517670, 517672- ca: AMNH 15249-15251, 15253-15255, 15259, 517674, 517677-517682, 517684-517688, 15260, 15263, 15264, 15270, 15273, 15278, 517690, 517695-517698. Hato El Frio: AMNH 15281, 15283, 15284, 15286, 15287, 15489, 257337-257340; MHNLS 8218,8220,8222,8231; 23227, 23238, 23240, 23243, 23245, 23248, USNM 448633-448640. San Antonio Maturin: 23251; USNM 123511, 123512. Montalban: AMNH 69757, 69762-69765, 69768, 69770, USNM 442453,442454,442456,442457,442481, 69773-69775,69778,69971. Kanuku Mountains: 442484, 442487, 442490-442492, 442495, AMNH 18978, 18979; BMNH 1.6.4.098, 442496, 442498, 442499, 442501, 442507, 1.6.4.100, 1.6.4.102-1.6.4.107, 1.6.4.174- 442508, 442510, 442511, 442514, 442515, 1.6.4.177. 56 AMERICAN MUSEUM NOVITATES NO. 3050

TABLE 13 62978, 62981, 62984-62986, 62989, 62991, Isla Puna: AMNH 19584-19586, 19588-19590, 62996, 63000, 63003, 63014. Puente de Chimbo: 19592, 19593, 19596, 67022, 67024, 67025, AMNH 62169, 62171, 62176, 62178, 62181, 67028-67030, 67032-67036, 67038-67043; 62183, 62188, 62195, 62197, 63027-63029. BMNH 99.8.1.11-99.8.1.13,99.8.1.16,99.8.1.17, TABLE 14 99.8.1.61; USNM 120236, 120237. Chongoncito: Urbina: AMNH 66303, 66304, 66306-66308, AMNH 62967, 62968, 62971, 62972, 62974, 66310-66312.

Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates, Bulletin, and Anthropological Papers published during the last five years are available free of charge. Address orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, N.Y. 10024.

THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER.