LI 5 R.AR.Y OF THL UNIVERSITY or ILLINOIS 590.5 FI V. 37-38 BIOLOQt L161— O-1096 South American Marsh Rats^ Genus Holochilus, With a Summary of Sigmodont Rodents Philip Hershkovitz Associate Curator, Division of Mammals Marsh rats, genus Holochilus, are large cricetines specialized for palustrine or semi-aquatic life. Their natural abode— in cane brakes, grass-lined stream banks and swamps — has predisposed the rats for successful occupation of sugar-cane plantations in the same regions. The vernacular name for the animal in Brazil is rato de cana; in Span- ish-speaking countries it is rata de agua. The name used by the Gua- rani Indians is angudya pihta. The generic name Holochilus, meaning "whole lip" in Greek, was proposed by Brandt in 1835 because the upper lip in two dried skins he described appeared to be entire. It has since been shown that the upper lip of marsh rats is normally cleft in the midline, as in all other rodents. The marsh rats, Holochilus, the cotton rats, Sigmodon, the coney rat, Reithrodon, and the red-nosed vole, Neotomys, compose a natural, well-defined assemblage of cricetines to be known as the sigmodont group. Diagnostic characters of sigmodonts include reduced outer hind toes, webbed middle hind toes, spinous process of zygomatic plate, specialized posterior palatal region, simplified molars charac- terized by absence or obsolescence of a functional mesoloph (id) in, at least, myll, and the S-shaped enamel pattern of m^. Sigmodont rodents are inhabitants of open country, with a pre- dilection for cultivated fields, where they usually become excessively abundant. They are abroad day and night and breed continuously throughout the year. Their young do not pass through a well- marked, protracted juvenal nestling stage. Instead, they acquire adult type pelage and, except for some delay in Reithrodon, fully erupted functional third molars very shortly after birth. They be- come active foragers while still occupying the maternal nest. Sigmodont rodents are distributed from southern United States to Tierra del Fuego at the southern tip of South America (fig. 139). 639 640 FIELDIANA: ZOOLOGY, VOLUME 37 Cotton rats occur throughout the North American portion of the range and the grass- and scrublands of the tropical zones of northern South America. The range of marsh rats is nearly as extensive. It overlaps that of Sigmodon in Venezuela, the Guianas, and the north- eastern portion of the Amazonian basin. It continues southward along the eastern coast of Brazil and the swamps and grass-lined banks and tributaries of the Amazon and Parana rivers, and termi- nates in the Argentine pampas. The red-nosed Neotomys has a limited distribution in the high Andean grasslands of Peru, Bolivia, and extreme northern Patagonia. The little coney rat, Reithrodon, of the treeless plains and pampas of Uruguay, Argentina and Chile, occupies the southern portion of the range of sigmodont rodents. ABBREVIATIONS The following abbreviations are used for designating the institu- tions that house the specimens examined: CNHM = Chicago Natu- ral History Museum; MAHN = Museo Argentino de Historia Natural. CHARACTERS OF SIGMODONT RODENTS External characters. — Form variable, rat-like in Holochilus, cunic- uloid in Reithrodon, vole-like in Neotomys and Sigmodon (= Sig- momys); size moderately large to among the largest of American cricetines; pelage soft or harsh; tail shorter than combined head and body length, except in Holochilus magnus; ear small; hind foot small to extremely large, claws well developed, about one-half of length of corresponding digits, except in Reithrodon, where they are somewhat shorter; outer digits of hind foot reduced, first and fifth toes, less claw, not extending distad of base of adjacent toes, except in Holochilus, where fifth toe, less claw, reaches middle of first phalanx of fourth; mammae 8 or 10. Cranial characters (pis. 21-29). — Nasals truncate or rounded be- hind, and terminating on a line with, or slightly behind or in front of, fronto-maxillary suture, never with a long pointed process extending deeply between frontals; supraorbital edges square, ridged or beaded, never evenly rounded; zygomatic arches well expanded behind, con- vergent anteriorly; interparietal well developed, its width always greater than width across ridges at fronto-parietal sutures; infraorbi- tal foramen, seen from dorsal surface of skull, deeply excised, with more than one-half to entire width of zygomatic plate exposed to view; zygomatic plate high and broad, its width more than one-half of least interorbital breadth, the anterior border convex, the upper Fig. 139. Distribution of sigmodont rodents in North and South America. 641 60 50 40 30 20 50 Fig. 140. Distribution of Holochilus. Diagonal lines = outer limits of range; dots = type localities of H. magnus and nominal subspecies of H. brasiliensis. 642 HERSHKOVITZ: SOUTH AMERICAN MARSH RATS 643 comer produced as a short, sharp spine; incisive foramina narrow, pointed behind ; posterior border of palate situated sHghtly behind or in front of posterior plane of last molars; posterior portion of palate with two pitted fossae separated by a median ridge; parapterygoid fossa seen from ventral surface deep, the anterior corner usually un- dercut, the lateral walls well defined; sphenopalatine vacuities well opened or, sometimes, partially closed by a thin membranous bony tissue; bullae well inflated. Dental characters (see fig. 144 and pp. 650-1 for explanation of ter- minology) . — Upper incisors well developed, opisthodont or orthodont, their anterior face smooth or grooved; molar rows parallel-sided or slightly divergent posteriorly; molars large, crowns high, sometimes prismatic, with grinding surface flat or terraced ; mesoloph rudimen- tary or absent in first two molars, present or absent in third upper molar; mesostyle absent or extremely small, never functional or asso- ciated with mesoloph, if present; occlusal surface of uis with S-shaped enamel pattern; major and primary folds deeply penetrating, the second primary of upper (first primary of lower) molars always extending medially beyond midline of tooth and apex of major fold; greatest length of first primary fold of upper molars one-half or more of length of second primary fold; major and first primary folds of m^ often confluent and forming a lamina; second secondary fold (posterior cingulum) of m^"^ absent in adults, absent or extremely reduced in juvenals; internal folds absent; antero-internal fold some- times present in first molar; first minor folds of m^"^ absent in adult, sometimes present in juvenal. RELATIONSHIPS Evolution of American cricetines has paralleled that of the New World Perissodactyla. The primitive brachyodont, buno-mesolopho- dont cricetines have survived, like the tapir, in forested parts of the range. Like the horse, the progressive branch of cricetines, with mesoloph absent or vestigial, has become increasingly specialized for life in open country and a diet of grasses. The young have lost, or are in process of losing, the distinctive juvenal phase of pelage com- parable to the spotted coat of young tapirs and other woodland un- gulates. The molars have become higher, their crown surfaces flatter and tending toward lamination. The third molar of grazing cricetines is nearly always precociously functional. This condition provides the needed maximum grinding surface which, in young ungulates, is supplied by deciduous premolars. 644 FIELDIANA: ZOOLOGY, VOLUME 37 Sigmodont rodents are members of the relatively recently evolved, essentially grazing, New World mammalian fauna of open country. Morphologically, they merge into the phyllotine group {Phyllotis, Hesperomys, and others) by a recombination of characters. In both groups, the dentition is fundamentally similar. In sigmodonts, how- ever, specialization of the posterior palatal region and the maxillary process of the zygoma is extreme. On the basis of dentition alone, the neotomyine rodents also appear to be nearly related. Cranially, how- ever, wood rats conserve what may be regarded as the generalized cricetine posterior palatal region and zygomatic plate. A hypothet- ical common ancestor of sigmodont, phyllotine, and neotomyine rodents might have combined the cranial characters of a form like Neotomodon, and the bunodont but more brachyodont molars of a phyllotine like typical Hesperomys. The two basic types of posterior palatal regions are repeated, with more or less modifications, throughout all myomorphs. Among meso- lophodont cricetines, the short palate and undifferentiated pterygoids are found in true peromyscids, the long specialized palate in oryzo- myines. Specialization of the posterior palatal region in the latter, however, has not affected the pterygoids at all, compared with the condition in sigmodonts. THE MESOLOPH AND DEVELOPMENT OF M^ In cricetines a well-developed, functional mesoloph (id) is always associated with brachyodont, bunodont molars of primarily forest- dwelling species. In living forms (e.g., Oryzomys, true Peromyscus, Thomasomys, Aporodon) the mesoloph (id) is locked with the meso- style (id). A mesoloph (id) is vestigial or absent in specialized high flat-crowned molars of cricetines inhabiting open country, scrubland and, secondarily, forests (usually coniferous) . Regressive loss of the mesoloph is associated with the evolution of grazers and grain-eaters from browsers and fruit-eaters (Hershkovitz, in preparation). In some cases it is lost in the bunodont stage of molar structure, in others its degeneration takes place in evolutionary stages of molar planation (figs. 141, 142) or of hypsodonty. In a given species ulti- mate disappearance of the element may have taken place during phylogeny, during ontogeny, or during both geneses. There is no vestige of a mesoloph (id) in even the newly erupted molars of Sig- modon, Reithrodon, and Neotomys. In Holochilus magnus a well- defined vestige is present in each upper juvenal molar (pi. 17); in newly erupted mi_2, the vestige is smaller but still well marked HERSHKOVITZ: SOUTH AMERICAN MARSH RATS 645 (pi. 18) ; in nis it is obsolete. In all cases the mesoloph (id) is non- functional and is completely lost in the young adult tooth. The molars of H. brasiliensis (pis. 17, 18) are more specialized, compressed antero- posteriorly, their crowns flat, the cusps triangulate, and the mesoloph (id) absent in all but m^. Here the element is usually well developed and functional through juvenal and young adult stages (fig. 143). In some individuals, even juvenals, it may be obsolete or entirely lacking. Thus, in this species, the mesoloph may be lost either in phylogeny or during ontogeny. Combined with the regressive, or palingenetic, nature of the mesoloph in m^ is the caenogenetic character of lamination, a dental development for more efficient grinding of harsh grasses. In Holo- chitus brasiliensis, lamination of the anterior one-third of m^ by confluence of major and first primary folds appears in the juvenal and advances into the subadult molar (fig. 143, C-E). Lamination of the postcingulum in the same tooth of this species is also incipient and marked by the emergence of a new element, the second minor fold. This enamel fold is confluent with both second primary and second secondary folds in the newly erupted molar, confluent only with the second primary in the worn juvenal and in the adult molar, or is an enamel island. In the dentally more primitive Holo- chilus magnus, lamination of the anterior one-third of m^ appears only in the newly erupted tooth and there is no indication of a second minor fold or of lamination of the postcingulum (fig. 143, A, B). In Sigmodon, lamination of the anterior one-third of m^ has progressed through ontogeny into a permanent adult character (fig. 143, F). Lamination of the postcingulum is a juvenal character in Neotomys, but here the second minor fold has become established as a trenchant adult character (fig. 143, G). The extremely variable nature of the upper third molar stands in marked contrast with the comparatively stabilized lower third molar. The evolution of this tooth to the S-shaped pattern has been con- sistent in all sigmodont species (pi. 20). Eruption of the third molar is advanced, often prenatal, in most cricetines with mesoloph vestigial or absent in adult dentition. The tooth becomes functional before it is structurally mature. Its de- velopment at this point may be at the same level as that of the uncut or newly erupted molar of the more primitive mesolophodont crice- tines. 646 FIELDIANA: ZOOLOGY, VOLUME 37 DIAGNOSTIC CHARACTERS AND KEY TO THE GENERA OF SIGMODONT RODENTS 1. Hamular processes of pterygoids converging anteriorly to form a A-shaped meso- pterygoid fossa; incisive foramina extending posteriorly behind anterior plane of first molars; supraorbital region parallel-sided, the edges square, unridged; lateral outline of skull leporine, dorsal contour extremely arched; orbits large and high; interorbital vacuity extraordinarily large; condylar process of man- dible deeply excavated internally; opisthodont upper incisors with a deep me- dian groove; molars prismatic; second secondary fold of m^ well developed but confluent with second primary in adult; procingulum of unworn mi trilobate; second primary fold of m2 two-thirds or more of length of first primary fold; outer digits of hind foot extremely reduced, not reaching to base of adjoining digits; claws less than one-half of length of corresponding digits; sole hairy from back of heel to base of outer digits; ear comparatively large; mammae 2-2 = 8 Reithrodon Waterhouse, 1837 Synonyms: Ptyssophorus Ameghino, 1889; Proreithrodon Ameghino, 1908; ?Tretomys Ameghino, 1889. Species: Reithrodon physodes Olfers, 1818 (antedates auritus Desmarest, 1819). Synonyms: Reithrodon typicus Waterhouse, 1837; R. cuniculoides Water- house, 1837; R. caurinus Thomas, 1920; R. olivensis Rusconi, 1931 (Pleistocene); Ptyssophorus elegans Ameghino, 1889 (Pleistocene); P. rotundatus Rusconi, 1931 (Pleistocene); ?Tretomys atavus Ameghino, 1889 (Pleistocene); Proreithro- don chapalmalense Ameghino, 1908 (Pleistocene); P. incipiens Ameghino, 1908 (Pleistocene). For discussion of the Recent subspecies see Osgood, 1943 (Field Mus. Nat. Hist., 30: 220-224). (See pis. 19, 20, 25-29; text fig. 143.) Hamular processes of pterygoids parallel-sided or slightly divergent to form a H- or M-shaped meso pterygoid fossa; incisive foramina extending posteriorly in front of or behind anterior plane of first molar; sides of supraorbital region parallel, convergent at midline, or divergent, with or without ridges; lateral out- line of skull murine, dorsal contour of skull slightly or moderately convex; or- bits not markedly enlarged; interorbital vacuity normal in size; condylar process of mandible not concave internally; upper incisors opisthodont or proodont, grooved or ungrooved; molars more or less hypsodont, never prismatic; second secondary fold of m^ present or absent in adults; procingulum of mi semicircular, triangular, sometimes faintly trilobate; second primary fold of m2 more or less than one-half of length of first primary; outer hind digits, less claw, reaching at least to base of adjoining digits; claw more than one-half of length of correspond- ing digits; sole bare or hairy on heel only; ears comparatively small; mammae 8 or 10 2 2. Anterior portion of nasals with a pronounced expansion; sides of supraorbital region parallel or convex midfrontally, the edges square and raised, but not beaded; temporal ridges obsolete or absent; width of mesopterygoid fossa at base of hamular processes less than width of parapterygoid fossa at same plane; di- astema short, approximately equal to or less than alveolar length of molar row; posterior border of angle of mandible broadly rounded or truncate, not pointed, coronoid process weak; lower incisor short and broad, the cutting edge hardly extending above horizontal plane of molar crowns, tip of root not encased in pro- jecting capsule; upper incisors opisthodont, the anterior surface fiat with a pronounced inward slope when seen in cross section, lateral borders deeply grooved; postcingulum of m' well developed and always clearly defined in the adult by deeply penetrating second secondary and second minor folds; second primary fold of m2 about one-half as long as first primary; minor folds of m2_3 absent, ear and hind foot extremely small, heel covered with hair; snout con- trastingly colored reddish; mammae 2-2 = 8 Neotomys Thomas, 1894 Species and Subspecies: N. ebriosus ebriosus Thomas; N. e. vulturnus Thomas). (See pis. 19, 20, 25-29; text fig. 143.) HERSHKOVITZ: SOUTH AMERICAN MARSH RATS 647 Anterior portion of nasals with normally tapered expansion; sides of supraorbi- tal region parallel, convex midfrontally, or divergent, the edges square, ridged or beaded; temporal ridges well developed in adults; width of mesopterygoid fossa at base of hamular processes equal to or more than width of parapterygoid fossa at same plane; diastema always longer than alveolar length of molar row; posterior border of angle of mandible tapered to a blunt point, coronoid process well developed, as high as or higher than condylar process; lower incisor long and slender, cutting edge extending well above horizontal plane of molars, tip of root encased in projecting capsule; upper incisors opisthodont or orthodont, the an- terior face normally convex, grooves, if present, near midline of tooth, never at lateral border; postcingulum of m' weakly developed or absent in juvenal, ab- sent in adult; second primary fold of mo more or less than one-half of length of first primary; first minor fold of m2_3 present at least in juvenal; ear short or moderate in length; hind foot moderate to extremely large; heel bare, smooth; snout not contrastingly colored; mammae 3-2 = 10 3 3. Nasals truncate behind, not extending beyond fronto-maxillary suture; supra- orbital region broadly divergent, edges beaded; frontal sinuses usually well in- fiated, midfrontal region convex in transverse outline; fronto-parietal sutures straight and forming an obtuse angle at midline; width across fronto-parietal sutures more than alveolar length of molar row and approximately equal to or more than length of diastema; incisive foramina extending posteriorly to or be- hind anterior plane of first molars; incisors opisthodont, grooved or ungrooved; molar crowns high, plane, principal cusps lozenge-shaped; vestige of mesoloph absent; apices of major and first primary folds opposed and touching in m^"^, confluent in m'; second secondary folds (posterior cingula) absent in upper molars; first secondary folds absent in lower second and third molars; anterior internal folds absent in first molars; second primary fold of m2 short, less than one-half of length of first primary; external form vole-like; size moderately large; pelage coarse, often hispid, color pattern of upper parts and sides of body strongly agouti; tail shorter than combined head and body length; hind foot well developed, upper surface well haired, digital bristles usually as long as or longer than claws; first and fifth digits, less claw, not extending beyond base of corre- sponding adjacent toes Sigmodon Say and Ord, 1825 Synonym: Sigmomys Thomas, 1901 (based on individuals of South Ameri- can representatives of S. hispidus with grooved incisors). Species: Unrevised, but all named forms probably referable to S. hispidus. (See pis. 19, 20, 25-29; text fig. 143.) Genus Holochilus, as described below. Genus Holochilus Brandt f/oZoc/ii7iis Brandt, 1835, Mem. Soc. Imp. Sci., St. Petersb., (6), 1: 428 (subgenus of Mus; included species: leucogaster, type; anguya); Wagner, 1842, Arch. Naturg., (8), 1: 14 (generic rank, description, comparison with Sigmodon; species: leucogaster, anguya, sciureus); Wagner, 1843, Schreber's Saugth. Suppl., 3: 548 (species: brasiliensis, leucogaster, canellinus, sciureus, vul- pinus); Burmeister, 1854, Syst. Ubers. Thiere Bras., 1: 162 (part; sub- genus of Hesperomys; description; species: vulpinus, squamipes [nee Brants], physodes); Ameghino, 1889, Act. Acad. Cienc, 6: 116 (description; species: vulpinus, multannus); Trouessart, 1904, Cat. Mamm. Suppl., p. 411 (part; species: multannus, brasiliensis, vulpinus, darwini, sciureus, guianae, can- ellinus, nanus, physodes, russatus). Holochyse (sic) Lesson, 1842, Nouv. Tabl. Reg. Anim., p. 137 (misspelling, name only listed). Holochilomys (sic) Peters, 1861, Akad. Wiss. Berlin, 1860: 150, 151; Hensel, 1873, Abh. Akad. Wiss. Beriin, Phys. KL, 1872: 32 (in synonymy of Hes- peromys vulpinus; species: brasiliensis). 648 FIELDIANA: ZOOLOGY, VOLUME 37 Holocheilus (sic) Coues, 1874, Proc. Acad. Nat. Sci. Philadelphia, 26: 177, foot- note (discussion). Sigmodon Winge, 1888, E. Mus. Lundii, 1, (3), pp. 12, 21 (comparisons; species: vulpinus); Merriam, 1895, Proc. Acad. Nat. Sci. Philadelphia, 1894: 226 and footnote 6 (species: vulpinus). Genotype. — Mus (Holochilus) leucogaster Brandt {=Holochilus brasiliensis leucogaster Brandt) and Mu^ (Holochilus) anguya Brandt (= Holochilus brasiliensis leucogaster Brandt); restricted to the first- named form by Miller and Rehn, 1901, Proc. Boston Soc. Nat. Hist., 30: 89. Included species. — Holochilus brasiliensis Desmarest; H. magnum sp. nov. The following, listed in some current works as probable species of Holochilus, are referable to other genera, as shown. Mus lutescens Gay, lMl=Rattu^ norvegicus Erxleben, 1771. Mws Simpsoni Philippi, 1900= Rattus norvegicus Erxleben, 1771. Mus agilis Philippi, 1900= Oryzomys longicaudatus Bennett, 1832. DISTRIBUTION Figure 140 Swamps, grasslands, and other moist, unforested situations from the coastal plains of Venezuela west into the Guianan lowlands, thence into suitable habitats in Brazil, Uruguay, Paraguay, the Ama- zonian basin in Peru and Bolivia, the Rio Paraguay basin in Bolivia, and in Argentina the Parana basin, the department of Buenos Aires, and possibly the department of Eva Peron; altitudinal range, sea level to approximately 2,000 meters above. TAXONOMIC HISTORY Holochilus was proposed by Brandt in 1835 as a subgenus of Mus. Improperly preserved dried skins gave Brandt the erroneous impres- sion that the upper lip was not parted in the midline, as is normal in rodents. The mistake was corrected by later authors but the truly distinctive characters of marsh rats were regarded as of sufficient magnitude to validate Holochilus as a genus. Similarities in dental structure between Holochilus and Sigmodon came to the attention of Wagner (1842), and systematists have since regarded the two genera as very nearly related. Winge (1888) went farther and treated marsh rats as a species of Sigmodon. Merriam (1895) held conflicting views. In his haste to criticize the arrangement of Argentine cricetines pre- plane Fig. 141. Molar planation: three evolutionary stages, from primitive crested to specialized plane grazing type. 1 = protocone Cid) ; 3 = paracone in upper, meta- conid in lower teeth. terraced plane Fig. 142. Unworn terraced molars (right upper) of juvenal Holochilus magnus and unworn plane molars of juvenal H. brdsiliensis. 649 ^ G H Fig. 143. Third right upper molar of A, Holochilus magnus (juvenal); B, H. magnus (adult); C, H. brasiliensis (juvenal with well-developed mesoloph); D, H. brasiliensis (juvenal with vestigial mesoloph); E, H. brasiliensis (adult with well-developed mesoloph); F, Sigmodon hispidus (adult); G, Neotomys ebriosus (adult); H, Reithrodon physodes (subadult). For explanation of symbols see below and page 651. NAMES OF ENAMEL FOLDS (ALL MOLARS) pf 1 , first primary fold nf 2, second minor fold pf 2, second primary fold amf, anterior median fold sf 1 , first secondary fold aif, anterior internal fold sf 2, second secondary fold asf, anterior secondary fold mf, major fold alf, anterior labial fold nf, first minor fold alif, anterior lingual fold 650 A B Fig. 144. Molars of Holochilus magnus. A, right upper molar row; B, left lower molar row. All dental elements present in sigmodont rodents are shown here. For explanation of symbols see below and page 650. NAMES OF CUSPS, CINGULA AND LOPHS Upper Molars 1, protocone 2, hypocone 3, paracone 4, metacone 5, anterolph 6, anterior cingulum 7, antero-labial conule 8, antero-lingual conule 9, mesoloph 10, posterior cingulum 11, posteroloph procingulum > postcinguli Lower Molars 1, protoconid 2, hypoconid 3, metaconid 4, entoconid 5, anterolophid 6, anterior cingulum 7, antero-lingual conule [ 8, antero-labial conule 9, mesolophid 10, posterior cingulum 11, posterolophid 661 652 FIELDIANA: ZOOLOGY, VOLUME 37 sented by Ameghino in 1889, Merriam pointed to the figures correctly determined as Holochilus by the former (Ameghino, 1889, atlas, figs. 11, 11, a, mandible and molars) and pronounced them Sigmodon. Paradoxically, he then decided that other specimens referred by Ame- ghino with equal justification to Holochilus were probably members of the over-inflated cricetine "subfamily" Neotominae Merriam! CHARACTERS External characters. — Form Rattus-like; size moderately large to among largest of crice tines; pelage soft, wool-hairs well developed; color of upper parts buffy, ochraceous or tawny, more or less mixed with blackish, especially mid-dorsally; sides distinctly paler, under parts varying from sharply defined white to uniformly ochraceous except for white or gray throat and, usually, inguinal region; basal portions of hairs of under parts white or gray; tail from more than three-fourths combined head and body length to nearly half again as long, coarsely scutulated, thinly haired, uniformly colored brown or slightly paler beneath than above; hind foot moderate to extremely large, thinly haired on upper surface, powerfully built and specialized for aquatic life; fifth toe, less claw, reaching to middle of first phalanx of fourth toe; first toe to base of second or slightly beyond; digital bristles sparse, shorter than claws; mammae (in H. hrasiliensis; un- known for H. magnus), 2 pairs pectoral, 2 pairs abdominal, 1 pair in- guinal. Cranial characters (pis. 21-29). — Nasals truncate or rounded be- hind, terminating on a line with, or slightly behind or in front of, fronto-maxillary suture; sides of supraorbital region maybe parallel, convex midfrontally, or slightly divergent, the edges square or ridged, not beaded; frontal sinuses hardly inflated, the midfrontal region concave in transverse outline; fronto-parietal sutures evenly curved or forming an angle at midline; width across fronto-parietal sutures more or less than alveolar length of molar row and less than length of diastema; incisive foramina not extending posteriorly behind anterior plane of first molars in adults; small jugal quite normal for cricetine rodents. Dental characters (pis. 17-20, figs. 141-144; for explanation of ter- minology see pp. 650-1). — Incisors opisthodont, or nearly orthodont, without grooves; molar crowns moderately high, plane or terraced (figs. 141, 142); cusps ovate or sub-triangular; a more or less de- veloped mesoloph often present in juvenal or subadult third upper molar; major and first primary folds of third upper molar opposed HERSHKOVITZ: SOUTH AMERICAN MARSH RATS 653 and touching in adult, nearly or quite confluent in juvenal (fig. 143) ; second secondary folds (posterior cingula) present in juvenal, obso- lete in adult upper molars; minor fold present in all lower molars; first secondary folds present at least in juvenal lower molars; anterior in- ternal folds present at least in juvenal mi; second primary fold of mo two- thirds or more of length of first primary fold. COMPARISONS Diagnostic and comparative characters of each genus of sigmodont rodent have been given above. The cotton rat Sigmodon is most nearly related to Holochilus and, in most respects, nearer H. brasili- ensis than the latter is to its larger congener, H. magnus. Sigmodon averages smaller but the proportions of its limbs and ears are about the same as in H. brasiliensis; the pelage, which is generally harsh, often hispid, in Sigmodon, is usually soft, rarely harsh, in Holochilus. Tawny individuals of Sigmodon may resemble Holochilus spp., but the agouti pattern of the former is approached in the latter only by a fine ticking on the dorsum. Hind toes of Sigmodon are not con- spicuously webbed like those of Holochilus. Cranial differences be- tween Sigmodon and Holochilus brasiliensis are slight and chiefly con- cerned with the broader, more inflated frontals and shorter rostrum of the former. Young individuals of H. brasiliensis, however, hardly differ in these respects. In H. magnus, the frontals are narrower and more depressed than in the smaller member of the genus. Mandibles of Sigmodon and Holochilus brasiliensis are similar and readily dis- tinguished from the elongated mandible of H. magnus. Incisors of Sigmodon are more recurved and the molar rows less parallel-sided, more divergent posteriorly. In its flat molar crowns and lack of mesolophs (ids), Sigmodon is more nearly like H. brasiliensis. On the other hand, the enamel folds of the molars of Sigmodon and H. mag- nus are oblique and the cusps are laminate or ovate, whereas in H. brasiliensis the folds are more nearly transverse and the upper cusps triangular in shape. Of the three named forms, the molars of H. magnus are more primitive in the following respects: retention of a rudimentary mesoloph in each juvenal molar, ovate cusps, crested paracone, metacone, metaconid and entoconid, less simplified pro- cingulum of upper and lower first molars, and, in the juvenal molars, the presence of a vestige of the primitive enamel ridge originally sepa- rating primary and internal folds. The oryzomyine water rat, Nectomys squamipes, is superficially similar to Holochilus in size, proportions, and swimming modifica- 654 FIELDIANA: ZOOLOGY, VOLUME 37 tions. It differs, however, by well-marked juvenal pelage, more glossy, adpressed adult pelage, more gray under parts, less hairy ears, longer vibrissae, scaly heel, claws of hind feet a third shorter and more recurved, mammae with the oryzomyine formula, 2-2 = 8. Cranially, Nectomys differs in most of the more important diagnostic characters enumerated for the sigmodont group in general and Holo- chilus in particular; in addition, braincase more inflated, frontal sinuses comparatively well developed, posterior portion of frontals not depressed, palate always produced posteriorly beyond third molars, bullae from one-third to one-half size of those of comparably aged individuals of Holochilus. Dentally, Nectomys belongs to the primitive branch of cricetines characterized by bunodonty, presence of a functional mesolophostyle (id) complex in all molars, absence of lamination, and delayed eruption of the third molar. Its dentition has been described elsewhere (Hershkovitz, 1944, Misc. Publ. Univ. Mich- igan Mus. Zool. no. 58, pp. 14-17, 19, 23; figs. 4, 5). Resemblances between Nectomys and Holochilus are of the same order as those that occur in many small mammals with special adaptations for swimming. Actual relationship between the two genera is, of course, on the same level as relationships between the oryzomyine and sigmodont groups of cricetines. Another oryzomyine rodent with which Holochilus has been con- fused in bibliographic citations and synonymies is Oryzomys buccina- tus Olfers (antedates Oryzomys angouya Desmarest and authors). Resemblance between the two forms does not go beyond color, size, and characters common to cricetines in general. However, the first published descriptions gave no more information. This, plus the fact that the two species occur together in some regions, had created the illusion that anguyas and marsh rats were one and the same. HABITS AND HABITAT The larger species, Holochilus magnus, is more specialized for swimming than H. brasiliensis. Its molars, however, are transitional between the browsing-crushing and specialized grazing types. Both species are found together in some localities, but heretofore authors have not distinguished them. Possibly the mollusc-eating marsh rat mentioned below belongs to the new species, H. magnus. In his notes on the nest and habits of marsh rats, Burmeister (1879, Descr. Phys. Rep. Argentine, 3: 211-212) recorded: "The rat lives in swamps among tall reeds which grow to a height of 3 meters. HERSHKOVITZ: SOUTH AMERICAN MARSH RATS 655 It builds its nest in September and attaches it to stems of the stoutest reeds. The abode is oval-shaped and may have a diameter as great as 40 centimeters. The nest is set at right angles to the reeds and about 30 or 40 centimeters above the surface of the water. Short pieces of reeds woven firmly together are used in the construction. The lower half of the living quarters is solid because of the large quan- tity of reeds pressed down on the bottom; the upper half is more lightly built and provided with a side entrance. The floor of the chamber is strewn with dry, gnawed reeds that make a soft litter for the occupant. Should the rat be disturbed by a strange sound, it leaps from its nest, plunges into the water, and swims swiftly amidst the thick vegetation until out of reach of possible danger. It remains at a safe distance for a long time before venturing back. My aide spent two hours to no avail watching from cover with the hope of seeing the animal swim back home. The rat rests quietly in its nest during the day. It feeds at night on swamp plants and molluscs." (Translation.) Burmeister does not mention the locality where ob- servations were made. Presumably it is somewhere between the Parand and Paraguay rivers, in extreme northern Argentina. T. Ingalls, collector of the type specimen of Holochilus chacarius of the Paraguayan Chaco, reports (in Thomas, 1906, Ann. Mag. Nat. Hist., (7), 18: 447) that the rat "raises a nest on weeds [reeds?] about a foot above water." Nests of marsh rats found in northern Argen- tina by Llanos (1944, Rev. Argent. Zoogeogr., 4: 51-52) were built on the ground in sugar-cane fields. The living quarters are described as round and constructed of interlaced strips and scraps of cane. Means and extremes of weights and diameters of the nine nests meas- ured are, respectively, 50 gms. (25-95 gms.) and 17.5 cm. (12-27 cm.). According to Moojen (1943, Bol. Mus. Nac. Zool., 5: 11), who observed and collected marsh rats in Ceara, Brazil, from four to six young are produced in a litter. "ratadas" Under certain conditions and at certain times, marsh rats mul- tiply rapidly and become plagues. Such plagues are known as ratadas in Latin America. Devincenzi (1935, Ann. Mus. Hist. Nat., Monte- video, (2), 4: 71) reports Holochilus as the most abundant rat in Uruguay. Morrison-Scott (1937, Ann. Mag. Nat. Hist., (10), 20: 535) quotes Dr. L. D. Cleare, collector of the type series of Holochilus brasiliensis berbicensis, as follows: "This rat is, I believe, fairly com- mon on the coastland here [British Guiana] and is probably the more 656 FIELDIANA: ZOOLOGY, VOLUME 37 usual form attacking sugar cane. They have caused considerable damage to cane on one sugar plantation where, during the past four months, over 100,000 have been destroyed — the outbreak appears to be confined to the county of Berbice." A ratada of several species of cricetines, including Holochilus, was observed in southeastern Brazil by Giovannoni, Velloza, and Kubiak (1946, Arq. Biol. Tec. Inst. Biol, e Pesq., Curitiba, 1: 185-195, pis. 52-60). They correlated the phenomenon with the fruiting season of the taquara lisa (smooth bamboo; Merostichis sp.). The ratada was the second recorded for the State of Parana. The first occurred six or seven years before and coincided with the fruiting of the taquara lixa (rough bamboo; Merostichis fistulosa Doell). According to the investigators, the taquara lisa bore fruit in March and April, 1946. The seeds sufficed to nourish the rodents for two or three months. With depletion of this food supply, hordes of hungry mice turned to wheat, millet, potatoes, and other cultivated plants in fields and storehouses. Loss to farmers caused by the devastations of the ro- dents was considerable. During September a high mortality rate among the rats led to a rapid ebb of the ratada. A cause for the sud- den decrease in population was not determined by the investigators. They suggested that the change in diet from taquara seeds to culti- vated plants, or disease, might have contributed to the high death rate. Local destruction of the blossoming taquara was offered as the best remedy for prevention of ratadas. An eruption of marsh rats in sugar-cane plantations in the Rio San Francisco Valley, northern Argentina, was reported by Llanos (1944, Rev. Argent. Zoogeogr., 4: 51-57). Inordinate increase in the marsh rat population began in February, 1944, and reached a peak in April. This was followed by a sharp decline in the number of in- dividuals. Part of the drop was attributed to systematic destruction of rats by man. In July, when the ratada had run nearly its full course, Llanos set 367 traps in an area of 1,019.36 square meters and captured 96 rats. Of the catch, males outnumbered females nearly two to one. None of the captured females was gravid. This and other trapping statistics provided by Llanos are based on gross catch per gross number of traps set, rather than on the standard "trap- night" basis. After the period of maximum population. Llanos found many abandoned nests of marsh rats in the sugar-cane plantations. The rats were identified as Holochilus balnearum, said to be the resi- dent species, and H. chacariv^, an invader. No criteria were given for distinguishing one named form from the other, or for regarding HERSHKOVITZ: SOUTH AMERICAN MARSH RATS 657 one rat as native and the other foreign to the area. Whatever the number of species involved, data supplied by Llanos indicate only a natural cyclic increase of native populations. Oryzomys flavicans, the only other cricetine trapped in the region, also figured in the ratada. KEY TO THE SPECIES Tail of adult more than 210 mm., always longer than combined head and body length; hind foot, with claw, more than 48 mm.; incisive foramina extending posteriorly to, sometimes behind, anterior plane of first molars; m' as long as wide, shorter than m*; m' with three roots Holochilus magnus sp. nov. Tail of adult less than 210 mm., and, normally, approximately equal to or shorter than combined head and body length; hind foot, with claw, less than 48 mm.; incisive foramina not extending posteriorly to anterior plane of first molars; m' longer than wide, longer than m^; m' with four roots. Holochilus brasiliensis Desmarest Holochilus magnus sp. nov. Plates 17-21, 23, 25-29; text figs. 142-144. Holochilus vulpinus Sanborn (nee Brants) 1929, Field Mus. Nat. Hist., Zool. Ser., 17: 158— part, Uruguay (Treinta y Tres; Rfo Cebollati). (l)Holochilus vulpinus Devincenzi, 1935, An. Mus. Hist. Nat., Montevideo, (2), 4, no. 10, p. 71— Uruguay. {l)Hesperomys vulpinus Hensel, 1873, Abh. Akad. Wiss. Berlin, 1872: 34 — part, Brazil (Porto Alegre, Rio Grande do Sul; one specimen in alcohol). {1)Sigmodon vulpinus Winge, 1888, E. Mus. Lundii, 1, (3), p. 23 — part, Brazil (Lag6a Santa, Minas Geraes; one specimen in alcohol). Type. — Adult male, skin and skull, Chicago Natural History Mu- seum no. 29258; collected November 23, 1926, by Colin C. Sanborn. Type locality. — Paso de Averias, Rio Cebollati, about 40 kilome- ters south of Treinta y Tres, eastern Uruguay. Distribution (fig. 140). — Known only from eastern Uruguay, but may range north as far as Minas Geraes in Brazil, and west into the Rio Parana basin of northern Argentina. Characters. — Largest sigmodont rodent; hairs of under parts gray or ochraceous terminally, plumbeous basally; tail longer than com- bined head and body length; hind foot, with claw, from 28 to 35 per cent of combined head and body length in adults and subadults (longer in juvenals); sides of supraorbital region of skull narrower than in brasiliensis and evenly convergent midfron tally; fronto- parietal sutures angular, not crescentic as in brasiliensis; incisive foramina extending posteriorly to, in juvenals behind, anterior plane of molars; ramus of mandible elongated, angle between base of body and ramus nearly straight, masseteric ridge not extending to anterior plane of first molar, coronoid process oblique, capsule of lower incisor 658 FIELDIANA: ZOOLOGY, VOLUME 37 not projecting posteriorly beyond vertical plane of coronoid process; crowns of molars terraced (fig. 142), cusps and cingula ovate in out- line and separated by comparatively broad folds; m^ shorter than m^ and approximately as long as wide; m^ with outer middle root ab- sent or extremely rudimentary; rudimentary mesoloph present in all Juvenal molars, obsolete or absent in adult molars; anterior median fold present in all but extremely worn m^ absent in mi; anterior in- ternal fold present in juvenal m^ and in juvenal and adult mi ; long axes of major and primary folds oblique; apices of first primary and major folds opposed and touching in upper adult molars, confluent in juvenal m^; second minor fold never present in m^. Measurements. — See Table 1. Remarks. — Direct comparisons of H. magnus with H. brasiliensis and other sigmodont rodents have been made under the generic heading. The long tail and proportionately larger hind feet indicate that magnus is more highly specialized than brasiliensis for swim- ming. In molar structure magnus is least specialized. The first upper molar in this species, however, is three-rooted; in all other sigmodonts it is four-rooted. Two specimens preserved in spirits and recorded as vulpinus, one by Winge (1888, supra cit.), the other by Hensel (1873, supra cit.), are, judged by published measurements (Table 1), referable to H. magnus. Neither authority suspected that two species of Holochilus live side by side. They described external characters of the genus from specimens of both magnum and brasiliensis preserved in spirits, and cranial and dental characters from skulls and skeletons of brasili- ensis only. It is surprising that of 18 names previously proposed for marsh rats, none applies to the present species. In view of the facility with which Thomas described a half dozen "species" of Holochilus, it is unlikely that he ever examined a specimen of magnus. Specimens examined. — 6. Uruguay: Paso de Averias, Rio Cebol- lati, Minas, 3 (CNHM); Treinta y Tres, Treinta y Tres, 3 (CNHM). Holochilus brasiliensis Desmarest. Plates 17-20, 22, 24-29; text figs. 142, 143. Distribution (fig. 140). — Same as for the genus. The species has been recorded also from Pleistocene cave deposits in Minas Geraes, Brazil, and from the Pleistocene pampean formations of Buenos Aires, Argentina. HERSHKOVITZ: SOUTH AMERICAN MARSH RATS 659 Characters. — Larger than any sigmodont rodent except Holochilus magnus; hairs of throat and inguinal region usually white or buffy ter- minally, white or gray basally; tail normally less than combined head and body length (may measure longer in dry, contracted skin) ; hind foot, with claw, from 20 to 25 per cent of combined head and body length in adults and subadults (longer in juvenals); sides of supra- orbital region of skull divergent; fronto-parietal sutures evenly curved, not forming an angle at midline; incisive foramina not ex- tending posteriorly behind anterior plane of first molars; ramus of mandible subquadrate, angle between base of body and ramus pro- nounced, masseteric ridge extending anteriorly beyond first molar to border of mental foramen, coronoid process nearly vertical, capsule of lower incisor extending behind to emargination between coronoid and condylar processes; crowns of molars plane (fig. 142), cusps and cingula compressed antero-posteriorly, subtriangular in unworn tooth; m^ longer than m^ and longer than wide; m^ with outer middle root normally present; mesoloph often present, and usually well de- veloped in m^, absent in remaining molars; anterior median fold not present in upper and lower first molars; anterior internal fold present in mi, absent in m^; long axes of major and first primary folds of upper molars transverse; major fold penetrating between first and second primary folds in m^~^; major and first primary opposed in adult m^ confluent in juvenal; second minor fold present, open to the margin in juvenals, an enamel island in adults. Taxonomic history. — After recording and describing marsh rats under various names for more than three decades, Thomas (1928, Ann. Mag. Nat. Hist., (10), 2: 260) concluded that he was "more and more convinced of the essential identity of all the Holochilus water- rats of the whole of the Amazonian drainage area, from Pemambuco to Peru, Guiana to Bolivia, and equally that of the Rio San Fran- cisco." Accordingly, Thomas assigned all eastern Peruvian speci- mens to Holochilus sciureus Wagner, of the Rio Sao Francisco, Minas Geraes, eastern Brazil. Concerning incarum, from Cuzco, Peru, Thomas (op. cit.) thought it might "survive as a local highland sub- species [of H. sciureus]." As for amazonicus and guianae, Thomas had already (1920, Proc. U. S. Nat. Mus., 58: 227) confessed to "con- siderable doubt as to whether [they] ought to have been separated from H. sciureus." Another named form not specifically mentioned by Thomas but included, by his sweeping statement, in the synony- my of sciureus, is Holochilus nanus. This animal from the island of Marajo in the mouth of the Amazon was described as "scarcely half 660 FIELDIANA: ZOOLOGY, VOLUME 37 the bulk" of sciureus. It proves to be nothing more than a juvenal of the common Brazihan marsh rat. If there is but one species of Holochilus within the geographic limits fixed by Thomas, the earliest valid name that applies is not H. sciureus Wagner, 1842, but Mus hrasiliensis Desmarest, 1819. The name was based on a specimen collected in Minas Geraes by Auguste St.-Hilaire. Its tail is said to be a little longer than the body, the under parts gray. Included with the description of the type is a reference to another specimen in the Paris Museum labeled "Rat du Br^sil." In 1820, Desmarest decided that his hrasiliensis was the same as the white-bellied, long-tailed species he had described as Mus angouya (=Oryzomys buccinatus Olfers, 1818). This was a mistake. Waterhouse examined the type and identified it as a Holo- chilus. The description of the dentition given by Gervais in 1849, though general, can apply only to Holochilus when external charac- ters are taken into account. Subsequent authorities have sustained these opinions. Mus vulpinus A. Brants, 1827, is the next available name for marsh rats. According to Lichtenstein (1830, Darstellung. . . . Heft 15, pi. 33, fig. 2 and text) the type is from Rio Uruguay, southeastern Brazil. This authority gives body length as 9"6'" (=248 mm,), tail 6"6'" (=170 mm.), hind foot with claw 2"3'" (=59 mm.). The measurements exceed those of any type or other recorded or avail- able specimen of the common Brazilian marsh rat. Nevertheless, the color, comparatively short tail and short hind foot (23,8 per cent of head and body), combined with the characters of the molars as de- scribed by Burmeister (1854, Syst, tJbers, Thiere Bras,, 1: 163), leave no doubt of the identity of vulpinus as a large representative of H. hrasiliensis. Mus physodes A. Brants, 1827, appears to be another short- tailed member of the common species. Its type locality is given by Lichten- stein as Sao Paulo, Brazil, The name might be retained as a sub- species of hrasiliensis, were it not preoccupied by M. physodes Olfers, 1818, a Reithrodon. Mus (Holochilus) leucogaster and Mus (Holochilus) anguya were described in 1835 by J, F, Brandt, The original data (including color plates) and present knowledge of variation in Brazilian marsh rats, indicate that the two named forms represent the same species. The types of both were secured by Baron von Langsdorff, then Russian consul general in Brazil. No definite locality was given for either, but authors have identified marsh rats of the State of Sao Paulo, where HERSHKOVITZ: SOUTH AMERICAN MARSH RATS 661 Langsdorff explored extensively, with H. leucogaster. Accordingly, the name leucogaster Brandt may replace the preoccupied physodes Brants. The name anguya Brandt is discarded not only as a syno- nym of leucogaster, but also as a homonym of Mus angouya Desma- rest, an Oryzomys. The substitute name canellinus proposed by Wag- ner in 1843, together with his russatus, also from Sao Paulo are, there- fore, synonyms of leucogaster. Thomas (1897, Ann. Mag. Nat. Hist., (6), 19: 496) suspected that leucogaster and sciureus might prove to be the same as H. brasiliensis Desmarest, Nevertheless, he was averse to conserving the name and proposed that the large marsh rat in the British Museum identified by Waterhouse in 1840 as Mus hraziliensis (sic) be called H. darwini in- stead. The renaming cannot be justified. Waterhouse had compared his specimen with the type of brasiliensis in the Paris Museum and found the two identical. In addition, the description, color plate, and figure of skull and dentition, prove that his specimen is the same as the sciureus of Thomas, or H. brasiliensis of this paper. The remaining named Recent forms are Holochilus berbicensis Morrison-Scott, Holochilus venezuelae J. A. Allen, H. balnearum Thomas, and H. chacarius Thomas. All show the diagnostic charac- ters of H. brasiliensis. The first three are given here provisional sub- specific rank. The last is indistinguishable from vulpinus of the same general region. ''H[olochilus] chrysogaster Waterhouse" is a misnomer published by Cabrera in his catalogue of specimens in the Natural History Museum of Madrid. No description or locality data accompanied the name and the specimen it designated. Possibly chrysogaster is a lapsus for leucogaster. Holochilus multannus Ameghino, 1889, from Pleistocene forma- tions of Buenos Aires appears to be a member of the common species and identical with H. brasiliensis vulpinus. The subspecies. — Material at hand is insufficient for a compre- hensive study of subspeciation in Holochilus brasiliensis. Present recognition of subspecies is based entirely on the availability of tech- nical names for marsh rats in geographic areas where subspeciation might occur. There is absolutely nothing in the original descriptions to support the named forms as subspecific categories. The type specimens themselves are so imperfect as to be useless for compara- tive purposes. The distribution given here for each subspecies is for the convenience of systematists who may wish to identify marsh rats by trinomials. 662 FIELDIANA: ZOOLOGY, VOLUME 37 Holochilus brasiliensis brasiliensis Desmarest Mus brasiliensis Desmarest, 1819, Nouv. Diet. Hist. Nat., 29: 62; 1826, Diet. Sei. Nat., 44: 483 (redescription of type); Lesson, 1823, Manual Mammal., p. 270; P. Gervais, 1849, Diet. Univ. Hist. Nat., 10: 731 (description of dentition); Osgood, 1943, Field Mus. Nat. Hist., Zool. Ser., 30: 235 (Mus brasiliensis a Holochilus). H[olochilus] brasiliensis Thomas, 1897, Ann. Mag. Nat. Hist., (6), 19: 496 (pos- sibly applies to "form I have provisionally called H. sciureus"). Hol[ochilus] sciureus Wagner, 1842, Arch. Naturg., (8), 1: 16, 17 — type locality, Rio Sao Francisco, Minas Geraes, Brazil (Spix, collector); 1843, Sehreber's Saugth., Suppl., 3: 553 (known from type only). H[olochilus] sciureus Thomas, 1897, Ann. Mag. Nat. Hist., (6), 19: 495 (com- pared with nanus Thomas). Holochilus sciureus Thomas, 1901, Ann. Mag. Nat. Hist., (7), 8: 149 (Bahia); 1920, op. cit., (9), 6: 276— Par4; Moojen, 1943, Bol. Mus. Nac. Rio de Janeiro, Zool., 5: 11 — Crato, Ceara. Sigmodon vulpinus Winge (nee Brants) 1888, E. Mus. Lundii, 1, (3), p. 21, pi. 1, fig. 3 (head), fig. 4 (hind foot), pi. 2, fig. 5 (skull), fig. 5, a (molar) — part, Lagoa Santa, Minas Geraes (Pleistocene and Recent). "H[olochilu3] chrysogaster Waterhouse," Cabrera, 1912, Trab. Mus. Cienc. Nat., Madrid, no. 11, p. 102 — "un ejemplar [eat. no.] 170, sin localidad comprado a la casa Verreaux de Paris" (a nomen vanum, probably a lapsus for H. leucogaster) . Mus angouya Desmarest (nee Desmarest), 1820, Mammalogie, 1: 305 — part (Mus brasiliensis in synonymy only). Holochilomy (sic) brasiliensis Hensel, 1873, Abh. Akad. Wiss. Berlin, Math.- Phys. Kl., 1872: 32 — part (in synonymy of Hesperomys vulpinus). Type. — Adult, sex unknown, Museum National d'Histoire Na- turelle, Paris; collected by Auguste St.-Hilaire between December, 1816, and March, 1818. Type locality. — "Br^sil"; according to collector's itinerary, Minas Geraes, Brazil. Here restricted to Lagoa Santa, Minas Geraes. Distribution. — Coast of Brazil, from the State of Minas Geraes northward to the mouth of the Amazon. Measurements. — Those of the type and its synonyms are given in Table 2. Remarks. — Under parts of the original H. brasiliensis are described as gray, those of the type of sciureus, white. Holochilus brasiliensis vulpinus Brants M[us] vulpinus Brants, 1827, Het Geslacht der Muizen, Berlin, p. 137 — "Bra- zil." Mus vulpinus Lichtenstein, 1830, Darstellung neuer oder wenig bekannter Saugeth., Heft 15, tab. 33, fig. 2 (col.) and text — type locality, Rio Uru- guay. HERSHKOVITZ: SOUTH AMERICAN MARSH RATS 663 H[olochiliis] vulpinus Wagner, 1843, Schreber's Saugth. Suppl., 3: 554; Thom- as, 1897, Ann. Mag. Nat. Hist., (6), 19: 496 — Parang and Uruguay river systems to La Plata; Goya, Corrientes. Holochilus vulpinus Ameghino, 1889, Act. Acad. Nac. Cienc. Rep. Argent., 6: 116, pi. 4, figs. 10, 11 (skull, molars) — Argentina (Recent and Pleistocene); Thomas, 1917, Ann. Mag. Nat. Hist., (8), 20: 96— Argentina (Isla Ella, Delta del Parana) ; Allen, 1916, Bull. Amer. Mus. Nat. Hist., 35: 571— Bra- zil (Urucum, Mato Grosso); Marelli, 1924, Elenco Sistematico Fauna Bue- nos Aires, Min. Obras Publ., Buenos Aires, p. 670 — Argentina (Corrientes; Entre Rios; Buenos Aires); Yepes, 1935, Rev. Inst. Bacteriol. Buenos Aires, 7: 234, pi. 7 (animal); Gyldenstolpe, 1932, K. Sv. Vet. Akad. Handl., 2: 63, pi. 6, fig. 2 (skull)— Argentina (Goya, Corrientes); Sanborn, 1929, Field Mus. Nat. Hist., Zool. Ser., 17: 158 — part, Uruguay (Trienta y Tres, one specimen only); Yepes, 1938, Rev. Centro Estud. Doc. Cienc. Nat., 2: 15 — Buenos Aires; 1938, An. Soc. Arg. Estud. Geogr., 6: 52 — Brazil and Argentina; Dennler, 1939, Physis, 16: 231 (Guarani iia.me = Angudya pihta). Hesperomys (Holochilus) vulpinus Burmeister, 1854, Syst. tJbers. Thiere Bras., 1: 163 (redescription of type; Mus braziliensis Waterhouse in synonymy); Burmeister, 1879, Descr. Phys. Rep. Argentine, 3: 210 (description). Hesperomys vulpinus Hensel, 1873, Abh. Akad. Wiss. Berlin, Math.-Phys. Kl., 1872: 32, figs. 13, 23 (dentition)— Brazil (Porto Alegre, Rio Grande do Sul). Holochilus multannus Ameghino, Act. Acad. Nac. Cienc. Rep. Argent., 6: 117, pi. 4, fig. 12 (mandible and lower molars) — type locality, "Piso lujanense (pampeano lacustre) de la formacion pampeana," Buenos Aires, Argentina. Mu£ braziliensis (sic) Waterhouse, 1839 (nee Desmarest), Zool., Voy. "Beagle," Mamm., p. 58, pi. 19 (animal), pi. 33, fig. 3 (skull and dentition) — Argen- tina (Bahia Blanca, Buenos Aires). H[olochilus] brasiliensis Wagner (nee Desmarest), 1843, Schreber's Saugth. Suppl., 3: 551 (redescription of Mus braziliensis Waterhouse). Holochilus brasiliensis Gray (nee Desmarest), 1843, List Mamm. Brit. Mus., p. 114 — Argentina (Bahia [Blanca]). H[olochilus] darwini Thomas, 1897, Ann. Mag. Nat. Hist., (6), 19: 496 (new name for Mus braziliensis Waterhouse). Holochilus darwini Thomas, 1910, Ann. Mag. Nat. Hist., (8), 5: 242 — Argen- tina (Los Ynglesas, Buenos Aires; type of darwini a female). Holochilus chacarius Thomas, 1906, Ann. Mag. Nat. Hist., (7), 18: 446 — type locality, one league northwest of Concepcion, Chaco, Paraguay; Yepes, 1938, Rev. Centro Estud. Doc. Cienc. Nat., 2: 15 — Paraguay (Concepcion; Puerto Guarani), Argentina (Tabacal, Salta; Chaco?; Formosa?). Type. — Presumably in Zoologisches Museum, Berlin; collected by Herr Sello. Type locality. — Rio Uruguay, southeastern Brazil. Brants states, "het vaterland van deze soort ist Brasilien." Lichtenstein specifies, "diese Art is von Herr Sello am Uruguay zuerst gefunden." This has 664 FIELDIANA: ZOOLOGY, VOLUME 37 been transformed by some authors into Darwin's collecting locality, Maldonado, Uruguay! Distribution. — Southeastern Brazil, in Mato Grosso and Parana, southward into Uruguay, Paraguay, and northeastern Argentina, in the states of Misiones, Corrientes, Entre Rios, Buenos Aires, and possibly adjoining parts of Eva Peron. Measurements. — See Tables 2 and 3. Remarks. — Brants' and Lichtenstein's descriptions of vulpinus are of a large, short-tailed, white-bellied marsh rat. No cranial and dental characters are mentioned. However, Burmeister (1854, supra cit.) supplied a detailed description of the size, shape, and enamel pattern of the molars of the type. This removes any reasonable doubt regarding the identity of vulpinus with brasiliensis. Hensel (1873, supra cit.) also described and figured the molars of marsh rats from the type region under the name Hesperomys vulpinus. It is almost certain that his specimens were compared with the type of vulpinus in the Berlin Museum. In any case, authors (Winge, Thomas) have accepted the identity of Hensel' s rats from Porto Alegre, Rio Grande do Sul, as representative. The "Mus braziliensis" of Waterhouse, from Bahia Blanca, is certainly the Brazilian species, but if it is sub- specifically separable the name vulpinus applies. Holochilus darwini, proposed by Thomas for the Waterhouse specimen, is superfluous. A skull in the British Museum from Goya, Corrientes, Argentina, figured by Gyldenstolpe (1932, supra cit.) under the name Holochilus vulpinus shows clearly the diagnostic cranial and dental characters of H. brasiliensis. Holochilus chacarius Thomas from Paraguay is an- other form of brasiliensis and is quite like vulpinus on a subspecific level. Ameghino (1889, supra cit.) recorded an individual of Holo- chilus vulpinus from the Pleistocene of Buenos Aires. In the same formation he discovered the mandible of a smaller specimen and de- scribed it as Holochilus multannus. The published figure of the mo- lars of the latter is poor. Whether the faulty representation is owing to the artist's liberties or to the imperfect condition of the fossil, it remains that multannus cannot be identified with any sigmodont ro- dent other than H. brasiliensis. The textual characterization of multannus, based on comparisons with fossil and living vulpinus, ex- poses only the variable dental characters of the species H. brasiliensis. A very old individual at hand from Uruguay agrees almost exactly with Burmeister's description of the type; it is nearly as large, with upper parts ochraceous-tawny mixed with dark brown, the darker color disappearing on the sides; under parts white with ochraceous HERSHKOVITZ: SOUTH AMERICAN MARSH RATS 665 wash on chest and belly, hairs slaty at the roots except on throat. Five specimens from the Paraguayan Chaco are younger, conse- quently smaller, their under parts more clearly defined, with less ochraceous, the hairs pale gray basally; in one specimen under parts nearly wholly white. Two juvenals from Mato Grosso are similar to the Paraguayan marsh rats. Specimens examined. — 8. Brazil: Urucum de Corumba, 2 (CNHM). Paraguay: Rio Pilcomayo, 15 miles above mouth, Chaco, 3 (CNHM); 30 km. northeast of Villa Militar, Chaco, 2 (CNHM). Uruguay: Treinta y Tres, 1 (CNHM). Holochilus brasiliensis balnearum Thomas Holochilus balnearum Thomas, 1906, Ann. Mag. Nat. Hist., (7), 18: 447; 1920, op. cit., (9), 5: 190 — Villa Carolina, Jujuy (specific distinction [from brasi- liensis] doubtful); Yepes, 1938, Rev. Centro Estud. Doc. Cienc. Nat., 2: 15 — Jujuy and Tucuman; Llanos, 1944, Rev. Argent. Zoogeogr., 4: 53 — Rio San Francisco Valley, Jujuy; Gyldenstolpe, 1932, K. Sv. Vet. Akad. Handl., 2: 64, pi. 17, fig. 16 (upper molars of type). Holochilus chacarius Llanos (nee Thomas), 1944, Rev. Argent. Zoogeogr., 4: 53 — Rio San Francisco Valley, Jujuy. Type. — Female, British Museum (Natural History) no. 4.10.2.5; collected June 18, 1904, by L. Dinelli. Type locality. — Baiiado de San Felipe, Tucuman, northwestern Argentina; altitude, 435 meters above sea level. Distribution. — Northern Argentina, in departments of Tucuman, Jujuy, and Salta, to altitudes not exceeding 1,000 meters above sea level. Measurements. — See Tables 2 and 4. Remarks. — The original description is based on one specimen, a Juvenal, hence the characterization of balnearum as "small." The oblique angle formed by the long axes of the molar rows, as described for the type, is not reproduced in any other known specimen. Molars of the type figured by Gyldenstolpe (1932, supra cit.) are those of the common species. In 1920, Thomas (supra cit.) recorded a large series of marsh rats from Villa Carolina, Jujuy, under what he now regarded as the doubtfully valid name balnearum. He also expressed remorse for the naming of this and other "species" of Holochilus. Holochilus brasiliensis leucogaster Brandt M[us] physodes Brants, 1827, Het Geslacht der Muizen, p. 139 — type locality, Brazil (name preoccupied by M. physodes Olfers, 1818, in Eschwege, Neue Bibliothek Reisenbreschr., Weimar, 15: 209, a Reithrodon). 666 FIELDIANA: ZOOLOGY, VOLUME 37 Mus physodes Lichtenstein, 1830, Darstellung neuer oder wenig bekannter Saugeth., Heft 15, tab. 34, fig. 1 and text — type locality restricted to Sao Paulo, Brazil. Hesperomys [Holochilus] physodes Burmeister, 1854, Syst. tJbers. Thiere Bras., 1: 167 — Sao Paulo (rtissatus Wagner, a synonym). Hesperomys physodes Pelzeln, 1883, K. K. Zool.-bot. Gesellsch., Wien, 33, Bei- heft, p. 71 — Ypanema, Sao Paulo. Holochilus physodes Leuderwaldt, 1929, Rev. Mus. Paulista, 16: 27 — Sao Paulo. Mus (Holochilus) leucogaster Brandt, 1835, Mem. Soc. Imp. Sci., St. Petersb., (6), 1: 92 (author's separate), pi. 12 — type locality, "Brazil." Holochilus leucogaster Giovannoni, Vellozo, Kubiak, 1946, Arq. Biol. Tec, Curitiba, 1: 188, pi. 58, figs. 10, 11 — Piraquara, Parana. Mus (Holochilus) anguya Brandt, 1835, Mem. Soc. Imp. Sci., St. Petersb., (6), 1: 94 (author's separate), pi. 13 — type locality, "Brazil." Hesperomys russatus Wagner, 1848, Abh. Akad. Wiss., Miinchen, Math.-Phys. Kl., 1850, 5: 312— type locality, Ypanema, Sao Paulo; Pelzeln, 1883, K. K. Zool.-bot. Gesellsch., Wien, 33, Beiheft, p. 71 — Ypanema, Sao Paulo. H[olochilus] canellinus Wagner, 1843, Schreber's Saugth. Suppl., 3: 552 (new name for Mus anguya Brandt, preoccupied by Mus angouya Desmarest, an Oryzomys). Hesperomys (Holochilus) brasiliensis von Ihering, 1894, Os mammiferos de Sao Paulo, Catalogo, Sao Paulo, p. 19 — Sao Paulo (also listed, H. physodes, leu- cogaster, sciureus). Holochilus brasiliensis Pelzeln (nee Desmarest), 1883, K. K. Zool.-bot. Ge- sellsch., Wien, 33, Beiheft, p. 73 (part, excl. syn.) — Ypanema, Sao Paulo. Holochilus sciureus Thomas (nee Wagner), 1910, Ann. Mag. Nat. Hist., (8), 6: 500 — Serra de Ibiapaba, Sao Paulo. Hesperomys squamipes Burmeister (nee Brants), 1854, Syst. tJbers. Thiere Bras., 1: 165 — Sao Paulo (synonyms: H. sciureus Wagner, anguya Brants, canellinus Wagner). Type. — Adult, sex unknown, presumably in Leningrad Zoological Museum; collected by Baron von Langsdorff. Type locality. — Brazil, here restricted to State of Sao Paulo. Distribution. — Known only from the coast and coastal mountain range of the states of Rio de Janeiro, Sao Paulo, and Parana. Measurements. — See Table 2. Remarks. — The under parts of all described forms referred to leucogaster are white, some with an ochraceous wash. Subspecific status of leucogaster is maintained pending comparisons with typical brasiliensis and vulpinus. Holochilus brasiliensis nanus Thomas Holochilus nanus Thomas, 1897, Ann. Mag. Nat. Hist., (6), 19: 495; Goeldi and Hagmann, 1904, Bol. Mus. Goeldi, Mus. Paraense, 4, 1, pp. 76, 83 (Portu- guese translation of original description). HERSHKOVITZ: SOUTH AMERICAN MARSH RATS 667 Type. — Juvenal female, skin and skull from specimen originally preserved in spirits, British Museum (Natural History) no. 97.4.1.2; presented by E. A. Goeldi. Type locality. — Soure (misspelled "Source" in original descrip- tion), Ilha de Marajo, mouth of Amazonas, Pard, Brazil. Distribution. — Known from type locality only. Measurements. — See Table 2. Remarks. — Characterized as a small species, the type proves to be a Juvenal of the common Brazilian marsh rat. Presence of adult type of pelage and functional third molars in all but suckling young of most sigmodont rodents, misled a number of systematists into describing juvenals as distinct from adults of the same species. Pending com- parisons with true hrasiliensis, the name nanus must be retained as the oldest available for marsh rats of the Amazonian and Guianan regions. Holochilus hrasiliensis amazonicus Osgood Mus (Holochilus) hrasiliensis Gervais, 1855, in Castelnau, Exped. Amer. du Sud, 7, ZooL, pt. 1, p. Ill — Bolivia (Santa Ana, Santa Cruz). Holochilus sp. Osgood, 1914, Field Mus. Nat. Hist., Zool. Ser., 10: 167— Peru (Tambo Yacu, San Martin). Holochilus amazonicus Osgood, 1915, Field Mus. Nat. Hist., Zool. Ser., 10: 188; Allen, 1916, Bull. Amer. Mus. Nat. Hist., 35: 571— Brazil (lower Rio Soli- moes). H[olochilus] sciureus Thomas, 1920, Proc. U. S. Nat. Mus., 58: 227— Peru (Rio Pachitea; guianae and amazonicus doubtfully distinct). Holochilus sciureus Thomas, 1927, Ann. Mag. Nat. Hist., (9), 19: 369 (Peruvian and lower Amazonian forms conspecific); op. cit., 1928, (10), 2: 260 — Peru (Cumeria; San Jeronimo); Tate, 1939, Bull. Amer. Mus. Nat. Hist., 76: 192 — Venezuela or Brazil (Rio Negro); Sanborn, 1949, Journ. Mamm., 30: 285 — Peru (Yarinacocha). Type. — Adult male, Chicago Natural History Museum no. 20136; collected May 11, 1913, by Robert H. Becker. Type locality. — Itacoatiara, north bank of Rio Amazonas, below mouth of Rio Madeira, Amazonas, central Brazil. Distribution. — Amazonian basin of Brazil, Bolivia, Peru. The species is unknown from eastern Ecuador and Colombia. Measurements. — See Tables 2 and 5. Remarks. — The type series is represented by skins and skulls of four adults. In these, chest and belly are gray overlaid with ochra- ceous; throat, inguinal region and inner sides of limbs dirty white 668 FIELDIANA: ZOOLOGY, VOLUME 37 faintly washed with buff. Osgood believed his amazonicus was larger than nanus and true brasiliensis as represented by Winge's Sigmodon vulpinus from Minas Geraes, It appears, however, that the type of nanus is a half-grown juvenal and the skull figured by Winge, basis for Osgood's comparison, is that of a subadult. Winge described other specimens with dimensions that exceed those of amazonicus. Valid characters for distinguishing amazonicus may exist, but they cannot be determined with certainty in present material. Whatever the eventual disposition of the name, Amazonian marsh rats from Peru and Bolivia agree completely with the type series of amazonicus. Specimens examined. — 21. Brazil: Itacoatiara, Amazonas, 9, including type and 5 newborn young in alcohol (CNHM). Peru: Moyobamba, San Martin, 1 (CNHM); Yarinacocha, Rio Ucayali, 8 (CNHM). Bolivia: Buenavista, Santa Cruz, 1 (CNHM); Mar- ban Province, Rio Mamor^, Beni, 2 (MAHN). Holochilus brasiliensis incarum Thomas Holochilus incarum Thomas, 1920, Proc. U. S. Nat. Mus., 58: 226, pi. 14, fig. 1 (skull of type). Type. — Immature female. United States National Museum no. 194195; collected December 22, 1914, by E. C. Erdis. Type locality. — Santa Ana, upper Rio Urubamba Valley, Cuzco, Peru; altitude, 1,061 meters above sea level. Distribution. — Known only from type locality. Measurements. — See Table 2. Remarks. — Of three specimens in the type series, the holotype is a juvenal, the others represented by skins only. Thomas (supra cit.) summed up the remarkably distinctive specific characters of incarum as "less buffy on flanks" than sciureus {= brasiliensis). He also ob- served that "it is not improbable that from Peru, at a comparatively low altitude, right down the Amazon to Para and Pernambuco, only one species of the genus is found." Provisional subspecific status for incarum rests on nothing more than present indecision regarding the systematic position of amazonicus. Holochilus brasiliensis guianae Thomas Holochilus guianae Thomas, 1901, Ann. Mag. Nat. Hist., (7), 8: 149. H[olochilus] guianae Thomas, 1920, Proc. U. S. Nat. Mus., 58: 227 (doubtfully distinct from sciureus [= brasiliensis]); Morrison-Scott, 1937, Ann. Mag. Nat. Hist., (10), 20: 536 (comparison; discussion). HERSHKOVITZ: SOUTH AMERICAN MARSH RATS 669 Holochilus sciureus Thomas, 1928, Ann. Mag. Nat. Hist., (10), 2: 260 (Holo- chilus monotypic). Type. — Adult male, British Museum (Natural History) no. 1.6.4.87; collected November 28, 1900, by J. J. Quelch. Type locality. — Kanuko Mountains, southern British Guiana; al- titude, 152 meters above sea level. Distribution. — Known from type locality only. Measurements. — See Table 2. Remarks. — Morrison-Scott (supra cit., p. 537) states that Thomas wrote on the label of the type of guianae " = sciureus." It remains to be decided whether or not guianae is as much as subspecifically dis- tinct from either typical brasiliensis or nanus. The status of ama- zonicus Osgood also depends on this decision. Holochilus brasiliensis berbicensis Morrison-Scott Holochilus sciureus berbicensis Morrison-Scott, 1937, Ann. Mag. Nat. Hist., (10), 20: 535. Type. — Adult male, British Museum (Natural History) no. 1937.6.24.2; collected April 10, 1937, by Dr. L. D. Cleare. Type locality. — Blairmont Plantation, Berbice, coastal British Guiana. Measurements. — See Table 6. Remarks. — Two specimens from Nonpareil Plantation on the De- merara coast, just west of the type locality of berbicensis, are slightly darker throughout than the type series of amazonicus Osgood from the middle Rio Amazonas. On this basis, the difference between ber- bicensis and the geographically adjacent guianae must be insignifi- cant, if at all real. In fact, Morrison-Scott refers a specimen from the Demerara River to sciureus {= brasiliensis). If berbicensis is at all valid, its range should include all the Demerara coast. Morrison- Scott did note that berbicensis was matched in color by a specimen of venezuelae J. A. Allen, from Carabobo, on the Venezuelan coast. However, the probability that venezuelae and berbicensis might be the same was discounted because of the larger feet and longer tail of the single Venezuelan specimen. Our near topotypes of berbicensis, however, have even larger feet and longer tails than Morrison-Scott's specimen from Carabobo. The oldest available name for marsh rats of the Guianan region is Holochilus brasiliensis nanus Thomas, from the island of Marajo, Para, Brazil. The fact that the type and only specimen of nanus is 670 FIELDIANA: ZOOLOGY, VOLUME 37 represented by a skin and skull removed from a juvenal originally preserved in spirits seems to have encouraged authors to describe better-prepared specimens as new species. Specimens examined. — 2. British Guiana : Nonpareil Plantation, Demerara, 2 (CNHM). Holochilus brasiliensis venezuelae J. A. Allen Holochilus venezuelae J. A. Allen, 1904, Bull. Amer. Mus. Nat. Hist., 20: 330, 341— part (type only); Goodwin, 1953, Bull. Amer. Mus. Nat. Hist., 102: 324 (type data; measurements). H[olochiliis] venezuelae Morrison-Scott, 1937, Ann. Mag. Nat. Hist., (10), 20: 537 — Carabobo, Venezuela. Type. — Immature female, American Museum of Natural History no. 16973; collected March 20, 1901, by S. M. Klages. Type locality. — El Llagual (Yagual), lower Rio Caura, Bolivar, Venezuela. Measurements. — See Table 2. Remarks. — The original description of venezuelae is based partly on a "half grown female" and, chiefly, on a "very old male." The younger specimen, however, was designated as type because the un- worn teeth "showed it to be a Holochilus and not a Nectomys." Al- len's misgivings regarding the true identity of the adult prove to have been well founded. The designated cotype is shown by Goodwin to be a Nectomys. The description of venezuelae, like that of nanus Thomas, is, therefore, zoologically worthless. The relationship of the Venezuelan marsh rat to other described forms of northern South America remains to be determined. Table 1. — Measurements of Holochilus magnus sp. nov. Head Greatest Zygo- Molar row and Hind length of matic Alveolar Locality body Tail foot skull breadth length Rio Cebollati' ... 210 230 62 44.9 25.6 8.5 Rio Cebollati .... 230 280 66 48.3 27.0 9.0 Treinta y Tres2 . . . 160 230 61 40.3 8.5 Porto Alegre' 175 210 47^ Lagoa Santas ... 185 208 50.5^ » Of type. " Juvenal. ' Of juvenal in spirits, from Hensel (1873, Abh. Akad. Wiss. Berlin, 1872: 34). * Without claw; add 5 mm. for shrinkage and claw. * Of juvenal in spirits, from Winge (1888, E. Mus. Lundii, 1, (3), p. 23). Sh *-■ '^ o s IM SS •OOJU3 t< CO bO o c o 1=5 t; CO o S|)^ : : : ; : :;Si2 ■^ t~ t- «D t- ••^ g — +J OS .2 m o^ 4S --S £3 0> ," i£> U (-, 'T'co gu3 T3 +j t- C O :^ o 05 oi 05 00 -^ CO w "3 O) w • 1— I 00 U3 U3 t- t- • lo CO CO CO CO CO M S 'S'O^ CO ;© «0 00 OS lO lO §5 t- -^ CO ->* .-I U5 CT3 > > > l-S ^ l-S l-S 1-5 0) 3 .3 M (-1 ooo ; ; ; ! ; lo'bo'^o —H CD T3 ■& eS ,£3 T3 bO C3 C 3 !5 '^ a _ ^^ So O CD 0^ ^ « "5 > '^ ^E-i j bfi C cc • ^ us 5^-S fe s ^es §Si3.^S£ . 53 C [2 "^ rt^ a S M jjso^S . Jj-^ Q >» II c 1ij g'-lrH00^>g s ^ « « ^ s c 3 C ^ o rt o j:3 a»: 671 672 FIELDIANA: ZOOLOGY, VOLUME 37 Table 3. — Measurements of Holochilus brasiliensis vulpinus Brants Head Greatest Zygo- Molar row and Hind length of matic Alveolar Locality body Tail foot skull breadth length BRAZIL Rio Uruguay' 246 167 59 URUGUAY TreintayTres ... 55« 45.5 26.0 7.8 PARAGUAY Villa Militar 170 160 .... 21.4 7.9 RfoPilcomayo 182 164 41 39.9 22.3 7.5 Rio Pilcomayo 178 160 39 38.3 21.0 7.0 Concepcion' 175 164 38 36.7 19.5 6.9 ARGENTINA Goya^ ... 42.2 24.6 7.8 ' Of type of vulpinus Brants, from original description. 2 From dry skin. ' Of type of chacarius Thomas, "a young adult" (probably subadult), from original description. * Of specimen in British Museum, from Gyldenstolpe (1932, K. Sv. Vet. Akad. Handl., 2: 142). Table 4. — Measurements of Holochilus brasiliensis balnearum Thomas Locality San Felipe' . Calilegua . Head and body 132 187 181 168 173 168 Tail 1332 153 168 160 156 158 Hind foot 35.5 42 40 43 41 41 Greatest length of skull 35.0 39.3 38.2 37.9 37.6 37.5 ' Of type, a juvenal, from original description. 2 Skin "contracted." Zygo- Molar row matic Alveolar breadth length 20.0 21.3 20.4 20.5 21.0 20.5 7.5 7.6 7.8 8.0 7.6 7.7 HERSHKOVITZ: SOUTH AMERICAN MARSH RATS 673 Table 5. — Measurements of Holochilus brasiliensis amazonicus Osgood Head Greatest Zygo- Molar row and Hind length of . matic Alveolar Locality body . Tail foot skull breadth length BRAZIL f 193 182 45 41.2 7.6 Itacoatiara' . . . 201 188 45 7.5 ' 194 180 42 40!2 22!o 7.6 190 158 42 40.6 7.7 PERU r 185 143[+] 40 40.8 22.3 7.6 Yarinacocha . . . 170 165 36 39.8 39.0 21.0 20.7 7.7 7.4 162 140 37.7 19.6 7.4 1 First measurements are those of type. Table 6. — Measurements of Holochilus brasiliensis berbicensis Morrison-Scott Greatest Con- Head length dylo- Zygo- Mo- and Hind of basal matic lar Locality body Tail foot Ear skull length breadth row f 178 144 39 16 36.6 21.5 7.19 190 145 35 17 36.3 21.3 7.29 153 132 35 17 33.8 20.4 7.17 Berbice' . . . , 140 146 142 189 170 117 108 139 138 150 35 35 35 36 37 16 16 16 17 15 31.9 19.2 7.22 Nonpareil. . ' 185 190 170 442 45'' 40 .1 37.9 22.2 7.5 7.2 » First measurements are those of type; the remainder, those of paratypes. All measurements from original descriptions. 2 Dry, with claw. Fieldiana: Zoology, Volume 37 Plate 17 Upper molars: 1 and 2, juvenal and adult (type) of Holochilus magnus; 3 and 4, Juvenal and adult of H. brasiliensis. Approximately X 8. Fieldiana: Zoology, Volume 37 Plate 18 Lower molars: 1 and 2, juvenal and adult (type) of Holochilus magnus; 3 and 4, Juvenal and adult of H. brasiliensis. Approximately X 8. s I .00 S O o S CSS ' CO (M 00 A^ X +J, -IJ e8 s g s ■R c>> o u s s <1 'ro -« M > be o s > bo o 43X > s > o 3 T3 CIS X a: I CO tr: ax * CO o, o o ^ OS a^ Fieldiana: Zoology, Volume 37 Plate 27 Outer surface of mandibles: 1, Holochilus magniis (type); 2, H. brasiliensis; 3, Sigmodon hispidus; 4, Neotomys ehriosus; 5, Reithrodon physodes. Approximately X9/5. Fieldiana: Zoology, Volume 37 Plate 28 Inner surface of mandibles: 1, Holochilus magnus (type); 2, H. brasiliensis; 3, Sigmodon hispidus; 4, Neotomys ebriosus; 5, Reithrodon physodes. Approximately X9/5. Fieldiana: Zoology, Volume 37 Plate 29 Lateral surface of skulls: 1, Holochilus magnus (type); 2, H. brasiliensis; 3, Sigmodon his-pidus; 4, Neotomys ebriosus; 5, Reithrodon physodes. Approximately X4/3.