Table Of Content

AMNHNOVITATES novi 00177 Mp 1 FridayAug18200010:31AM2000 Allen Press • DTPro System File # 01cc PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024 Number 3302, 20 pp., 6 figures, 4 tables October 16, 2000 Late Cenozoic Land Mammals from Grenada, Lesser Antilles Island-Arc R.D.E. MACPHEE,1 RONALD SINGER,2AND MICHAEL DIAMOND3 ABSTRACT We report on a small collection of late Cenozoic fossil vertebrates recovered from a lahar (mudflow) deposit at Locality 12(cid:1) North on the southern coast of Grenada.* 40K/40Ar–dated hornblende concentrate from the lahar deposit yielded age estimates of 2.6–3.7 Ma (Late Pliocene). Although these estimates date crystallization of the hornblende and not the lahar event, the latter is unlikely to be substantially younger. The contained fauna is here regarded as latest Pliocene or slightly younger. Dental specimens in the collection are readily referable to Hydrochaeridae (Rodentia, Ca- viida) and Megalonychidae (Xenarthra, Phyllophaga), groups heretofore unknown on this island. The capybara, Hydrochaeris gaylordi, new species, differs from extant Hydrochaeris hydrochaeris in the conformation of the maxillary second molar. The sloth teeth (two cani- niforms, one molariform) notably differ from one another in size, but whether they represent one species or two cannot be decided on this evidence. Because of the limitations of the material, attribution of the specimens to subfamily or tribe within Megalonychidae is also uncertain. Megalonychid sloths have never been found previously on any of the Lesser An- tilles, although they formed part of the terrestrial vertebrate fauna of most of the Greater Antilles. Curaçao is the only other island in the Caribbean Sea that has yielded sloth and capybara fossils. Sloths and capybaras might have reached that island as well as Grenada by short-distance over-water transport, perhaps during a time of lowered sea level. A late land connection with South America is perhapspossible, butthis would needtobeconfirmedwith suitable geological evidence. INTRODUCTION like that of the other Lesser Antilles (Allen, 1911). Its extant fauna comprises 19 species, ThelandmammalfaunaofGrenadaistyp- 11ofwhicharebats(table1).Theremainder ically viewed as being highly depauperate, consistsofspeciesthatwereprobablyorpos- *Contribution3intheseries‘‘OriginoftheAntilleanLandMammalFauna.’’ 1Curator, Division of Vertebrate Zoology ([email protected]); Adjunct Senior Research Scientist, CERC, Co- lumbiaUniversity. 2Professor of Anatomy and of Anthropology; Robert R. Bensley Professor of Biology and Medical Sciences, DepartmentofOrganismalBiologyandAnatomy,UniversityofChicago,1027E.57thSt.,ChicagoIL60637. 3310037thSt.,RockIslandIL61201-6532. Copyright (cid:2) American Museum of Natural History 2000 ISSN 0003-0082 / Price $2.50 AMNHNOVITATES novi 00177 Mp 2 FridayAug18200010:31AM2000 Allen Press • DTPro System File # 01cc 2 AMERICAN MUSEUM NOVITATES NO. 3302 sibly introduced by human agency (Varona, ternary contexts (Pregill et al., 1994), no ex- 1974; Eisenberg, 1989). At the specieslevel, tinct land mammals of any sort were known no Grenadian mammal is an exclusive en- from Grenada. demic,4 a fact that has been used (e.g., by This report adds two more entries to the Groome, 1970) to argue that this island was Grenadian faunal list: Hydrochaeris gaylor- never connected to South America (but see di,newspecies,andMegalonychidae,gen.& Discussion and Conclusions). Until the re- sp. indet., material of which was recovered cent discovery of two unnamed species of by one of us (RS) from a lahar deposit on oryzomyin sigmodontines in very late Qua- the south end of the island. The capybara, represented by a partial maxillary dentition, 4An exclusive endemic is a species-level taxon that is similar but not identical to living Hydro- originated in,and whose primarynaturaldistributionis chaeris hydrochaeris from the South Amer- limited to, a single continuous area (usually small). In ican mainland. The sloth material is scanty: principle, origin could be allopatric or sympatric, al- little can besaid aboutitbeyondthefactthat thoughforobviousreasonsallopatricendemicsareeas- ier to document. Endemics can, of course, secondarily thespecimensaremegalonychid.Afewfrag- disperse,butbydefinitiontheyremainendemicintheir ments of bone belonging to smaller verte- areaoforiginuntilandunlessextinctionoccurstherein. brates (?lizards) were found in matrix sam- Islandendemicsareexclusiveonlywhensufficientlydif- ples,butnoneiscompleteenoughtoidentify. ferentiatedfrommainlandparentpopulations/species((cid:2) Whatever their limitations as specimens, allopatry). Until there is differentiation there is no ex- clusiveendemism,onlyrangeextension. these finds are important because they pro- AMNHNOVITATES novi 00177 Mp 3 FridayAug18200010:31AM2000 Allen Press • DTPro System File # 01cc 2000 MACPHEE ET AL.: GRENADIAN FOSSIL MAMMALS 3 vide faunal records that are essential for re- Other constructing the historical biogeography of the Caribbean region (cf. Iturralde-Vinent AMNH-VP American Museum of Natural and MacPhee, 1999). History, Division of Paleontology (Vertebrate Paleontology) bsl below sea level ACKNOWLEDGMENTS GIUA GeologicalInstituteoftheUniver- Pat and Joseph Gaylord, ownersofthe12(cid:1) sity of Amsterdam North Hotel, generously facilitated Singer’s 40K/40Ar potassium-argon (dating tech- work on their property in many ways, in- nique) cluding permitting him to hire two of their Ma millions of years (ago) staff,DavidAdolphusandNevilleJames,for UF Florida Museum of Natural His- part-time excavating.Mr. Gaylordkindlydo- tory, University of Florida natedallofthespecimensrecoveredbySing- er to the American Museum of Natural His- GEOGRAPHICAL AND GEOLOGICAL tory, to ensure their availability to scientific SETTING OF GRENADA workers.WethankSusanPhilip(AMNH)for assisting with this matter. Geographically and geologically, Grenada Singer thanks St. George’s University (12(cid:1)N, 61(cid:1)W; 310 km2) is associatedwiththe School of Medicine for invitations to teach Grenadines, the spray of (cid:4)40 small isletsly- anatomy at their campus on several occa- ing south of St. Vincent (figs. 1, 2). Grenada sions, which permitted the fieldwork de- and the Grenadines are the subaerial com- scribed in this paper. Singer’s fieldwork in ponentsoftheGrenadinesBank,alarge(180 Grenada was supported by a grant from the km (cid:3) 25 km) submarine structure extending Marion R. and Adolph J. Lichtstern Fund of NE-SW and having an average water depth the Department of Anthropology, University of 20–40 m. Regional slope of the top ofthe of Chicago.Financialsupportforpreparation bank is 1:300, indicating thatitisnearlyhor- of this paper was provided by a grant from izontally disposed (Dey and Smith, 1989). Mr. Robert Liberman (to RDEM). The bank drops off abruptly E and W into This paper benefitted from anonymous re- the Tobago and Grenada troughs. views and the advice of Brian J. Crother In the largerregionalpicture(fig.1),Gre- (Southeastern Louisiana University) on her- nada is situated 125 km from the closest petological matters. The authors additionally point on the South American mainland and thank Lorraine Meeker (specimen photogra- 115 km from Tobago, the closest non-Antil- phy), Patricia J. Wynne (maps), and Clare lean island.Measuringfromthe(cid:5)200miso- Flemming (editing) for their usual excellent bath, however, island and continent are even services. closer: only (cid:4)40 km separate the southern end of the Grenadines Bank from the edge ABBREVIATIONS of the continental shelf. However, seafloor depth in thisshortintervalreaches600mbsl Anatomical (fig. 1). BL buccolingual width Speed et al. (1993) have summarized a CA cross-sectional area (BL (cid:3) MD) wealth of data on the stratigraphy, magma- CF caniniform (i.e., anteriormost tism, depositional environments, and defor- tooth in megalonychid dentition) mation of the southern Lesser Antilles arc m1, m2, m3 mandibularfirst,second,thirdmo- platform, or SLAAP. According to these au- lar thors, the SLAAP, which forms the core of M1, M2, M3 maxillaryfirst,second,thirdmolar the Grenadines Bank, is a half-horst thatwas MD mesiodistal length uplifted early in the Miocene. Neogene vol- MF molariform (i.e., any postcanini- canism was widespread in this area, begin- form tooth in megalonychid dentition) ning about 12 Ma or shortly before, and it M (cid:3) W labioexternal to proximointernal continues into the present day. The only ac- angles (CF), greatest distance tivesubmarinevolcanointheregionisKick- AMNHNOVITATES novi 00177 Mp 4 FridayAug18200010:31AM2000 Allen Press • DTPro System File # 01cc 4 AMERICAN MUSEUM NOVITATES NO. 3302 Fig. 1. Physiographic diagram of Caribbean basin and contiguous part of Atlantic (adapted from maps by B. Heezen and M. Tharp); depths are in meters. The Grenadines Bank (encircled) terminates a comparatively short distance ((cid:4)40 km) north of the continental slope of South America, but the sea floor between bank and slope is (cid:6)600 m bsl. ’em-Jenny (Donnelly et al., 1990), located Middle Eocene onward, as suggested by the immediately north of Grenada (fig. 2). nature of deposits in the turbidite basins. The basement of Grenada is a volcanic However, neither the location nor the lon- unit of Middle Eocene age, succeeded by a gevity of these earlier islands can be esti- series of tuffs of distinctive composition in- mated fromthedataavailable.Extensivecar- terbedded with limestones and marlsranging bonate platform conditions, associated with in age from Middle Eocene (50 Ma) to Mid- uplift of the half-horst, came into being in dle Miocene (13 Ma) (Martin-Kaye, 1969; thelateEarly/MiddleMiocene.Thisisofim- Maury et al., 1990; Wadge, 1994). There is portance because it is the first indication of only one recognized substantial unconfor- extensive shallow marine and intertidal en- mity, provisionally dated to the interval be- vironments. Plio-Pleistocene reef limestones tween Late Oligocene and Middle Miocene occur patchily among the Neogene volcanics and possibly as much as 10 Ma long. This in various parts of the island, indicating the lacuna has so far been detected only on Car- continuing presence of land in more recent riacou in the Grenadines (Speed et al. 1993). epochs (Maury et al., 1990; Wadge, 1994). For the southern Lesser Antilles, Speed et al. (1993)identifiedthree‘‘generallysequen- LOCALITY 12(cid:1) NORTH tial’’ sedimentary depositional environments (pelagic basin, turbidite basin, platform) The area of interest is the toelike south- from Middle Eocene onward. Volcanic plat- western terminus of the island,whichiscon- forms with some subaerial exposure were in tinued for another 30 km as the shallowly existence in the area to the west from the submerged southern end of the Grenadines AMNHNOVITATES novi 00177 Mp 5 FridayAug18200010:31AM2000 Allen Press • DTPro System File # 01cc 2000 MACPHEE ET AL.: GRENADIAN FOSSIL MAMMALS 5 Fig. 2. Southern portion of Grenadines Bank, Lesser Antilles island-arc, showing the island of Grenada and major geographical features mentioned in text. Inset: Position of Locality 12(cid:1) North, west side of Lance aux E´pines peninsula on extreme southern coast. Bank (figs. 1, 2). Composed of a series of from which pebbles, shells and other inclu- weathered tuffs (Tomblin, 1970), the prom- sions (other than teeth and bones) are virtu- ontory is strikingly indented by a series of ally absent. On initial inspection, RS noted narrow embayments typical of a subsiding the sloth molariform (AMNH-VP 132714) coastline. The most westerly subaerial pro- and larger caniniform (AMNH-VP 132715) jection is Point Salines; Lance aux E´pines, 5 described below, as well as several uniden- km to the east, is the most southerly projec- tifiable fragments of bone (fig. 3B). These tion. Locality 12(cid:1) North is positioned on the elements were removed in a single block of latter’s west side, facing into Prickly Bay. matrix and prepared in Chicago. Further ef- In 1982, Mr. Joseph Gaylord, co-owner of forts to secure fossils were made in subse- the 12(cid:1) North Hotel, pointed out to RS the quent years by exploration of the cliff face. existenceofsmallbonyfragmentsinthetuff- This effort was rewarded in 1989 by the dis- aceous cliff face below the resort’s garden. covery of another sloth caniniform (AMNH- The cliff is entirely composed of a massive, VP132716)andsomewhatlater(in1991)by homogenous yellow-brown clay (fig. 3), recovery of a partial maxillary dentition of a AMNHNOVITATES novi 00177 Mp 6 FridayAug18200010:31AM2000 Allen Press • DTPro System File # 01cc 6 AMERICAN MUSEUM NOVITATES NO. 3302 Fig. 3. View of Locality 12(cid:1) North, southern coast of Grenada, from a photo taken in 1989 by R. Singer. Measuring rod held by man is approximately 120 cm long; approximately 5 m of cliff face is exposed at low tide. The deposit,composedofmassive,homogeneousyellow-brownclaywithvirtually no inclusions derives from a lahar event that probably occurred in late Pliocene or early Pleistocene. Foot of measuring rod rests on spot where initial fossil discoveries—the sloth molariform and larger caniniform—were made (circled area in inset shows teeth and bone fragments in situ before their removal). Capybara maxilla was found two years later at approximately the same level as the other teeth, but some 15 m further to the right (off margin of photograph). capybara (AMNH-VP 132713). These last similar observations by Tomblin, 1970). The finds were located about 15 m from the po- simplestscenariofortheoriginofthedeposit sition of the 1982 finds, suggesting that the at Locality 12(cid:1) North is that it derives from areaaroundLocality12(cid:1)Northoughttoyield a lahar, possibly triggered by an eruptive many more specimens if properly excavated. event. It seems probable that the lahar swept Dr. Richard Hazlett (Joint Sciences De- down valley from its point of origin, picking partment, Claremont Colleges) has kindly up older, weathered material as it went and examined samples of the matrixinwhichthe rapidly burying pumice generated by the fossils were found. He reported that the ma- eruption. The ash may have been reworked trix is mostly composed of smectite, a clay to a minor extent after deposition. In any derived from the weathering of volcanicash. case, the few skeletal elements recovered Small,diffuselyborderedwhitelumpswithin were incomplete at the time of discovery, the matrix are pumice; these pieces showlit- and no associated bones were encountered. tle sign of weathering and are therefore sug- DATING gestive of rapid burial. Also present in some abundance are lapilli and small crystals of Here we report two 40K/40Ar age estimates gabbro, pyroxene, olivine, and augite (see for Locality 12(cid:1) North, based on hornblende AMNHNOVITATES novi 00177 Mp 7 FridayAug18200010:31AM2000 Allen Press • DTPro System File # 01cc 2000 MACPHEE ET AL.: GRENADIAN FOSSIL MAMMALS 7 concentrates extracted from fresh matrixcol- of much of the island’s geomorphology is lected at fossiliferous locationsapproximate- certainly consistent with theyoung datesand ly 15 m apart. The dates—2.7 (cid:7) 0.3 and 3.6 light weathering of fossiliferous matrix at (cid:7) 0.4 Ma (table 2)–are reasonably concor- Locality 12(cid:1) North. dant and overlap at two standard errors. However, some caution is necessary in inter- A NEW HYDROCHAERINE RODENT preting very young 40K/40Ar dates (Geyhand FROM GRENADA Schleicher, 1990), especiallywhenthepotas- Maxillary M1–M3 (AMNH-VP 132713): sium content of the dated material is low, as This specimen is unquestionably a hydro- it is in the present case. chaerid caviidan (fig. 4).5 It is a partial max- The simplest interpretation of the age es- illary dentition (M1–M3) of an adult animal, timates for Locality 12(cid:1) North is that they held together by a few remnant patches of date the timeof crystallizationofhornblende maxillary bone (fig. 4; table 3). Hypsodont, within the original tephra matrix. The lahar multilamellar tooth organization of the kind is not directly dated thereby; it was certainly seen in thisspecimenisdiagnosticofHydro- somewhat later, because there is no evidence chaeridae (Kraglievich, 1930; Woods, 1989) that the fossils were exposed to high tem- and requires no special comment. The M1 peratures. On the other hand, the lahar event andM2aresubstantiallyintactexceptformi- isunlikelytohavebeenverymuchlater,giv- nor abrasions and cracks. However, the M3 en the appearance of the matrix. We provi- is clearly incomplete: distally the tooth ends sionally conclude that the mudflow and its abruptly, in the middle of an enamel-dentine contained mammal fossils were depositedno lamella (or ‘‘prism’’). earlier than 2–3 Ma, most probably in the Three subfamilies of Hydrochaeridae are terminal Pliocene, or perhaps as late as the generally recognized: Cardiatheriinae, Pro- Early Pleistocene. tohydrochoerinae, and Hydrochaerinae, with This conclusion is in good agreementwith the last two being the more closely related Tomblin’s (1970) weathering-profile evi- (Mones, 1984; but see McKenna and Bell, dence that the tuffaceous sediments forming 1997). According to Mones(1984),cardiath- Point Salines are ‘‘not older than Pliocene,’’ eriines have fewer lamellae (6–10) on max- as well as the very young 40Ar/39Ar dates re- illary M3s than do hydrochaerines (10–17) cently published by Speed et al. (1993) for or protohydrochoerines (12–18). As noted, the Mt. Craven volcanic center at the north the Grenadian specimen is broken and its endoftheisland.Thesedates,basedonmag- true count is therefore not known (but was matic hornblende, cluster between 1.4 and surely higher than the nine that remain).The 1.7 Ma and suggest that the present volca- Grenadian specimen is trivially similar to nogenic surface of Grenada may be very re- Cardiatheriinae in that the major lamellaeon cent indeed. (Older work suggested that the maxillary M2 are united buccally (fig. 4D, Mt. Craven center was in excess of 21 Ma, F). However, there are no other features that which now appears to be invalid.) Although can be viewed as specifically cardiatheriine. the Mt. Craven event was evidently somewhat later than lahar deposition at the far 5For spelling of names based on hydrochaer-, see southern end of the island, the relativeyouth Woods(1989). AMNHNOVITATES novi 00177 Mp 8 FridayAug18200010:31AM2000 Allen Press • DTPro System File # 01cc 8 AMERICAN MUSEUM NOVITATES NO. 3302 Fig. 4. Hydrochaeris gaylordi, new species, AMNH-VP 132713 (holotype, partial right maxillary dentition) from Locality 12(cid:1) North, Grenada: A, lingual aspect; B, mesial aspect (of M1); C, buccal aspect; D, drawing of buccal aspect of M1 and M2 (asterisk on M2 identifies buccally conjoined lamellae); E, occlusal aspect; F, drawing of occlusal aspect of M1 and M2, showing effect on occlusal patternofbuccallyconjoinedlamellae(presentinM2,leftarrow;absentinM1,rightarrow).Bycontrast, in extant H. hydrochaeris, lamellae are buccally separate in both M1 and M2. AMNHNOVITATES novi 00177 Mp 9 FridayAug18200010:31AM2000 Allen Press • DTPro System File # 01cc 2000 MACPHEE ET AL.: GRENADIAN FOSSIL MAMMALS 9 m3, the angles formed by lamellae as seen medially or laterally, and other minor char- acters. Dental measurements of AMNH-VP 132713 are within the size range of mid- sized extant Hydrochaeris hydrochaeris (Hooijer, 1959; Ojasti, 1973). The only ob- vious difference between the Grenadian specimen and extant capybaras is the condi- tion of the M2. In extant capybaras, both the M1 and the M2 consist of two independent lamellae with well-marked infolds that are completely separated by intervening cementum. In the Grenadian specimen,bycontrast, this description applies only to M1; in M2, the lamellae are united buccally by a narrow band of enamel and dentine (see asteriskand arrows in fig. 4C–F). In the M3 of H. hydrochaeris the first and last lamellae are bilobate; the rest (usually 10) are entire. In this specimen the bilobate first lamella is followed by eight others, which suggeststhatthe lastthreearemissing (assuming same total count). Discussion: Today, capybaras are com- monlyfoundinmanypartsofnorthernSouth America and, for that matter, throughout the Importantly, buccal union of all M3 lamellae continent north of the southern cone (Mares (when viewed in occlusal aspect), a feature and Ojeda, 1982). However,untilthediscov- distinctive of Cardiatheriumitself(Kragliev- eryreportedhere,CuraçaowastheonlyWest ich, 1930), is absent in the Grenadian spec- Indianislandwheretheywereknowntohave imen. existed during the Quaternary (on the basis It is equally easy to distinguish the Gren- of fossil evidence only). Hooijer (1959) de- adian specimen from the truly gigantic taxa scribed a series of teeth, apparentlyreferable comprising subfamily Protohydrochoerinae to a juvenile member of the living species, (cf. Paula Couto, 1979). Deep longitudinal recovered from a hard oolithic phosphate incisures (infolds) on the buccal aspects of matrix in the eastern part of the island. The the middle lamellae of M3 are a distinguish- origin of these deposits, found on several ing feature of Protohydrochoerus (Kragliev- WestIndianislands,iscontroversialbutthere ich, 1930). This feature is lacking in other is no reason to suspect that they are of any subfamilies and in the Grenadian specimen. great antiquity (Robert Halley, pers. oral M1 and M2 of Protohydrochoerus are simi- commun.). Indeed, Hooijer (1959) even lar in occlusal pattern to their homologs in raised the question whether the capybara extant Hydrochaeris, but the lingual borders from eastern Curaçao represented an animal of their lamellae are less sharply angledthan introduced by humans. He seems to have in living capybaras. been driven to this rather unlikelypossibility In general aspect and in the shape of in- because this island has a markedly xeric as- dividual lamellae making up each represent- pect at present, and was therefore not, in his ed tooth, the Grenadian specimen is over- estimation, fit habitat for water-loving H.hy- whelmingly similartomembersofsubfamily drochaeris. Hydrochaerinae. Hydrochaerine species are Kraglievich (1940a, 1940b) noted that, in principally distinguished by small differenc- extant Hydrochaeris, individual teeth vary es in thenumberoflamellaecomprisingM3/ ontogenetically in the degree to which la- AMNHNOVITATES novi 00177 Mp 10 FridayAug18200010:31AM2000 Allen Press • DTPro System File # 01cc 10 AMERICAN MUSEUM NOVITATES NO. 3302 mellaeareangled,separated,orjoined.Inthe TYPE LOCALITY AND AGE: Locality 12(cid:1) unerupted or little-worn cheek teeth of neo- North, Lance aux E´pines, Grenada (Lesser nates, lamellae are frequently joined buccal- Antilles); probably Late Pliocene, less likely ly. As wear progresses, such connections are Early Pleistocene (40K/40Ar age estimate on quickly lost, yielding occlusal patterns typi- hornblende in matrix, 2.7–3.6 Ma). cal of adults. This would seem to provide a ETYMOLOGY: For Mr. Joseph Gaylord, in simple basis for the existence of conjoined recognition of his many kindnesses to RS lamellae in adult dentitions, and Kraglievich over many years. documented a number of examplesofpersis- DISTRIBUTION: Known only from Grenada tent connections between lamellae in lower in the southern Lesser Antilles. teeth. However, he also specifically noted DIAGNOSIS:Asmallhydrochaerinethatcan that maxillary P4–M2 show relatively few bedistinguishedfromallothernamedhydro- variations as compared to M3. In fact, M1 chaerines, including extant Hydrochaerishy- and M2 appearto beessentiallyinvariantex- drochaeris, by its unique occlusal pattern, in cept for the presence of slight compression whichbuccalconnectionofmaxillaryM2la- or smoother reentrant angles (as seen on oc- mellae is never lost during ontogeny. clusal surfaces). DISCUSSION: Although in unerupted cheek It is therefore of some systematic impor- teeth of Hydrochaeris hydrochaeristhesum- tance that Kraglievich (1940a) did not notice mits of individual lamellae are typically the occurrence, in extant Hydrochaeris, of joined, reflecting the earliest stage of crown buccal union of M2 lamellae, the one clearly ontogeny, these connections are lost with distinctive feature of the Grenadian speci- wear. By the young juvenile stage, the no- men. (As no M2s of the Curaçao capybara tably hyposdontcheek teethofcapybarasap- have been found, morphologies cannot be pear to be made up of separate enamel-den- compared.) Indeed, the only group in which tinelamellaeofcomplicatedshape,unitedby this conformation isregularlyseenisthecar- plaques of cementum. Although M3 may be diatheres, and AMNH-VP 132713 from Lo- somewhat variable in that someinterlamellar cality 12(cid:1) North differs in numerous details connections may be preserved into later life, fromanyrecognizedmemberofthissubfam- observationalevidenceindicatesthatM1and ily. Although this is a fairly narrow basis on M2 do not vary in this regard. Thus, pres- which to found a new species, the unusual ervation of a buccal connection between M2 nature of the M2 warrants it. lamellae in the Grenadian specimen appears Systematic Paleontology: We name, diag- tobeaderived(albeitneotenic)featurewith- nose,andclassifythenewspeciesasfollows: in Hydrochaeris, justifying recognition of H. gaylordi.Thisfeatureisabsentorunrecorded MAGNORDEREPITHERIAMCKENNA,1975 in the other species of Hydrochaeris recog- ORDERRODENTIABOWDICH,1821 nized by Mones (1984): Lujanian H. balles- SUBORDERHYSTRICOGNATHAWOODS,1976 terensis (Rusconi, 1934) and extant H. isth- INFRAORDERHYSTRICOGNATHI mius (Goldman, 1912), sometimes regarded TULLBERG,1899 as a valid species (cf. Mares and Ojeda, PARVORDERCAVIIDA 1982). BRYANTANDMCKENNA,1995 SUPERFAMILYCAVIOIDEA MEGALONYCHID FOSSILS FROM FISCHERDEWALDHEIM,1817 GRENADA FAMILYHYDROCHAERIDAEGRAY,1825 SUBFAMILYHYDROCHAERINAEGRAY,1825 As noted in greater detail in subsequent paragraphs, the three teeth described in this Hydrochaeris gaylordi, new species sectionarereferabletoMegalonychidae,gen. HOLOTYPE: AMNH-VP 132713 (fig. 4), & sp. indet. The homologs of phyllophagan partial maxilla retaining M1-M3. teeth in other placentals are not certain, and DISCOVERER AND DATE OF DISCOVERY: weshall followtheconventionofidentifying Ronald Singer and party, 1991. the highly trenchant first tooth as the cani-